Advances in Planar Lipid Bilayers and Liposomes, Volume 7 (Advances in Planar Lipid Bilayers and Liposomes)

  • 22 143 4
  • Like this paper and download? You can publish your own PDF file online for free in a few minutes! Sign Up

Advances in Planar Lipid Bilayers and Liposomes, Volume 7 (Advances in Planar Lipid Bilayers and Liposomes)

V O LU M E S E V E N ADVANCES IN PLANAR LIPID BILAYERS AND LIPOSOMES EDITORIAL BOARD Professor Professor Professor P

867 142 5MB

Pages 283 Page size 468 x 684 pts Year 2010

Report DMCA / Copyright

DOWNLOAD FILE

Recommend Papers

File loading please wait...
Citation preview

V O LU M E

S E V E N

ADVANCES IN PLANAR LIPID BILAYERS AND LIPOSOMES

EDITORIAL BOARD Professor Professor Professor Professor Professor Professor Professor Professor Professor Professor Professor Professor Professor Professor Professor Professor Professor Professor Professor Professor

Dr. Roland Benz (Wuerzburg, Germany) Dr. Hans G.L. Coster (Sydney, Australia) Dr. Herve Duclohier (Rennes, France) Dr. Yury A. Ermakov (Moscow, Russia) Dr. Alessandra Gliozzi (Genova, Italy) Dr. Alesˇ Iglicˇ (Ljubljana, Slovenia) Dr. Bruce L. Kagan (Los Angeles, USA) Dr. Wolfgang Knoll (Mainz, Germany) Dr. Reinhard Lipowsky (Potsdam, Germany) Dr. Yoshinori Muto (Gifu, Japan) Dr. Ian R. Peterson (Coventry, UK) Dr. Alexander G. Petrov (Sofia, Bulgaria) Dr. Jean-Marie Ruysschaert (Bruxelles, Belgium) Dr. Bernhard Schuster (Vienna, Austria) Dr. Masao Sugawara (Tokyo, Japan) Dr. Yoshio Umezawa (Tokyo, Japan) Dr. Erkang Wang (Changchun, China) Dr. Philip J. White (Wellesbourne, UK) Dr. Mathias Winterhalter (Bremen, Germany) Dr. Dixon J. Woodbury (Provo, USA)

V O LU M E

S E V E N

ADVANCES IN PLANAR LIPID BILAYERS AND LIPOSOMES Editor

PROFESSOR DR. A. LEITMANNOVA LIU Department of Physiology, Michigan State University, East Lansing, Michigan, USA and Centre for Interface Sciences, Microelectronics Department, Faculty of Electrical Engineering & Information Technology, Slovak Technical University, Bratislava, Slovak Republic Founding Editor

PROFESSOR DR. H.T. TIEN Department of Physiology, Michigan State University, East Lansing, Michigan, USA

Amsterdam Boston Heidelberg London New York Oxford Paris San Diego San Francisco Singapore Sydney Tokyo Academic Press is an Imprint of Elsevier

ACADEMIC PRESS

Academic Press is an imprint of Elsevier 84 Theobald’s Road, London WC1X 8RR, UK Radarweg 29, PO Box 211, 1000 AE Amsterdam, The Netherlands Linacre House, Jordan Hill, Oxford OX2 8DP, UK 30 Corporate Drive, Suite 400, Burlington, MA 01803, USA 525 B Street, Suite 1900, San Diego, CA 92101-4495, USA First edition 2008 Copyright # 2008 Elsevier Inc. All rights reserved No part of this publication may be reproduced, stored in a retrieval system or transmitted in any form or by any means electronic, mechanical, photocopying, recording or otherwise without the prior written permission of the publisher Permissions may be sought directly from Elsevier’s Science & Technology Rights Department in Oxford, UK: phone (+44) (0) 1865 843830; fax (+44) (0) 1865 853333; email: [email protected]. Alternatively you can submit your request online by visiting the Elsevier web site at http://www.elsevier.com/locate/permissions, and selecting Obtaining permission to use Elsevier material Notice No responsibility is assumed by the publisher for any injury and/or damage to persons or property as a matter of products liability, negligence or otherwise, or from any use or operation of any methods, products, instructions or ideas contained in the material herein. Because of rapid advances in the medical sciences, in particular, independent verification of diagnoses and drug dosages should be made ISBN-13: 978-0-12-374308-4 ISSN: 1554-4516 For information on all Academic Press publications visit our website at books.elsevier.com

Printed and bound in USA 08 09 10 11 12 10 9 8 7 6 5 4 3 2 1

CONTENTS

Preface Contributors

1. Random Processes in the Appearance and Dynamics of an Electropore in a Lipid Membrane

xi xiii

1

Malgorzata Kotulska and Krystian Kubica 1. Introduction 2. Lipid Membrane—Structure and Dynamics 2.1. Lipid Conformation 2.2. Random Movements of Lipids Molecules 2.3. MC Methods in Modeling Membranes 3. Randomness in Electroporation 3.1. Induction of a Pre-Pore 3.2. Appearance of a Hydrophilic Pore 3.3. Ionic Concentration and Membrane Susceptibility to Electroporation 4. Randomness in the Dynamics of Long-Lived Electropores 4.1. Creating a Single Long-Lived Electro-Nanopore 4.2. Analytical Methods for Stochastic Processes with a Memory 4.3. Stochastic Characteristics of the Electropore Fluctuations 4.4. Modeling Stochastic Processes of a Power-Law Spectrum 5. Summary References

2. Functionalized Liposomes

2 4 4 6 7 12 12 16 17 19 19 21 26 30 31 32

39

Raghavendra Palankar, Yannic Ramaye, Didier Fournier, and Mathias Winterhalter 1. Introduction 2. Liposome Formation 3. Stabilization of Liposomes 3.1. PEGylated Liposomes 3.2. Template Polymerization 3.3. Coating with Polyelectrolytes 3.4. DNA Coating 4. Functionalization 4.1. Encapsulation of Water-Soluble Molecules 4.2. Controlled Permeability by Polyelectrolyte Coating 4.3. Reconstitution of Membrane Channels into Liposomes

40 41 42 42 43 45 46 46 46 49 49

v

vi

Contents

5. Possible Application of Functionalized Capsules in Diagnostics 5.1. Capsule Array 5.2. Manipulation of Capsules with External Fields: Donnan Potential 5.3. Functionalized Liposome Capsules as Intracellular Sensors and Delivery Vehicles 6. Conclusions Acknowledgments References

3. Pore-Suspending Membranes on Highly Ordered Porous Alumina and Porous Silicon Substrates: Preparation, Characterization, and Application

50 50 51 53 55 56 56

59

Claudia Steinem 1. Introduction 2. Porous Substrates 2.1. Highly Ordered Porous Alumina 2.2. Highly Ordered Porous Silicon 3. Formation of Micro- and Nano-BLMs 4. Characterization of Nano- and Micro-BLMs 4.1. Electrical Characterization of Nano-BLMs 4.2. Optical Characterization of the Rupturing Process of Micro-BLMs 4.3. Lateral Diffusion of Lipids in Micro-BLMs 5. Applications of Nano-BLMs 5.1. Monitoring the Activity of Bacteriorhodopsin 5.2. Monitoring Single Channel Events of Peptides and Proteins 6. Solvent-Free Pore-Suspending Membranes 7. Conclusions Acknowledgments References

4. Antiphospholipid Syndrome: Mechanisms Revealed in Erythrocyte and Liposome Studies

60 61 61 62 63 64 64 67 68 69 69 71 73 75 75 75

79

Snezˇna Sodin-Sˇemrl, Blazˇ Rozman, Alesˇ Iglicˇ, and Veronika Kralj-Iglicˇ 1. Introduction 2. Biochemical Aspects of Mechanisms Involved in APS 2.1. Exposure of Negatively Charged Membrane Surfaces due to Loss of Membrane Asymmetry and Its Role in APS 2.2. Cardiolipin 2.3. Phosphatidylserine 2.4. Protein Cofactors and Their Antibodies 3. Cellular Microexovesicles (Microparticles) in Thrombosis and Hemostasis with Special Emphasis on APS 3.1. Mechanisms Leading to Microvesiculation 3.2. Effect of b2GPI and aPL on Budding of Phospholipid Vesicles

80 81 81 82 83 84 88 89 97

Contents

4. Coalescence of Membranes Caused by b2GPI and aPL 4.1. Experimental Evidence on the Effect of b2GPI and aPL on Coalescence of Phospholipid Vesicles 4.2. Theoretical Description of the Coalescence of Membranes: Interaction Between Charged Surfaces Mediated by Ions with Dimeric Distribution of Charge 5. The Effects of ANXA5, b2GPI and aPL on Phospholipid Membranes 6. Conclusions Acknowledgments References

vii

99 100

101 111 113 113 114

5. The Single GUV Method to Reveal Elementary Processes of Leakage of Internal Contents from Liposomes Induced by Antimicrobial Substances 121 Masahito Yamazaki 1. Introduction 2. Experimental Methods of the Single GUV Method 2.1. Preparation of GUVs 2.2. The Method to Induce the Interaction of Substance Solution with a Single GUV 3. Effect of Antimicrobial Peptide, Magainin 2, on Membrane Permeability and Membrane Structure 4. Effect of Antibacterial Tea Catechin, EGCg, on Membrane Permeability and Membrane Structure 5. Conclusion and Advantage of the Single GUV Method Acknowledgments References

6. Flexible Membrane Inclusions and Membrane Inclusions Induced by Rigid Globular Proteins

122 125 125 126 127 134 139 141 141

143

Miha Fosˇnaricˇ, Alesˇ Iglicˇ, Tomazˇ Slivnik, and Veronika Kralj-Iglicˇ 1. Introduction 2. Flexible Anisotropic Membrane Inclusions 3. Membrane Inclusions Induced by the Rigid Membrane-Embedded Protein 3.1. Perturbation of Lipid Molecules Around Rigid Membrane-Embedded Proteins 3.2. Energy of Membrane Inclusion Induced by a Single Rigid Membrane Protein 4. Estimation of the Model Parameters 4.1. Basic Model 4.2. Advanced Model 5. Free Energy of Bilayer Membrane with Membrane-Embedded Inclusions 6. Conclusions Acknowledgments References

144 144 149 149 151 155 155 157 163 165 166 166

viii

Contents

7. A New Class of Peptide-Forming Channel: Calcitonins

169

Silvia Micelli, Daniela Meleleo, H. Ti Tien, Angelica Leitmannova Liu, Vittorio Picciarelli, and Enrico Gallucci 1. Introduction 2. Materials and Methods 2.1. The Single-Channel Measurements 2.2. Chemicals 2.3. The Data Analysis 3. Evidence of hCt Channels Formation 3.1. hCt Can Form Voltage-Dependent Channels with Weak Anion Selectivity 4. Role of Detergents on the Characteristics of hCt Single Channels 4.1. hCt Channel Activity Can Be Increased by SDS 4.2. There Is an Optimal Molecular Ratio Between hCt:SDS for hCt Channel Activity 4.3. The Addition of SDS to hCt Does Not Change the Voltage Dependence or the Lifetime, While Its Selectivity Shifts to Cations 4.4. Different Detergents Produce Different Effects on hCt Channel Activity 5. Effects of pH Variations on Insertion and Channel Formation of hCt 5.1. pH Bathing Condition Has Different Effects on the Channel Characteristics Depending on the Nature of PLM 6. Investigations on sCt and eCt 6.1. sCt and eCt Channels Form Voltage-Dependent Channels Having Almost Equal Selectivity for Anions/Cations 6.2. eCt Glycosylation Modifies the Activation Voltage to Form Channels and Shifts Ion Selectivity to Anion 7. Concluding Remarks 7.1. The Results 7.2. Relationships with Other Research 7.3. Tentative Interpretation of Physiological Relevance References

8. Lipid-based Strategies in Inorganic Nano-materials and Biomineralization Study

171 173 173 174 175 176 176 178 180 181 183 183 184 184 189 189 190 193 194 194 198 198

203

Xiaohua Liu, Haixin Bai, Lixue Zhang, and Erkang Wang 1. Introduction 2. Synthesis of Various Nanocrystals Based on Lipid Strategies 2.1. Nanocrystals Protected by Lipid Membrane 2.2. Other Synthesis Methods of Nanocrystals Based on Lipid Strategies 3. Nanocrystal–Lipid Hybrid Materials 3.1. The Encapsulation of Various Nanocrystals in Lipid Membrane 3.2. Assembly of Lipid Membrane on Nanocrystals

204 204 204 209 210 210 211

Contents

ix

4. The Application of Lipid-based Nanocrystals 4.1. Interaction of Nanocrystals with Lipid Membrane 4.2. Examples of the Applications of Lipid-based Nanocrystals 5. The Application of Lipid in Biomineralization 6. Conclusion and Perspectives Acknowledgments References

212 212 213 213 215 215 216

9. Self-Reproduction of Micelles, Reverse Micelles, and Vesicles: Compartments Disclose a General Transformation Pattern

221

Pasquale Stano and Pier Luigi Luisi 1. Introduction 2. Theoretical Backgrounds 2.1. Autopoiesis 2.2. General Concepts in the Self-Reproduction of Micelles 3. Self-Reproduction of Micelles and Reverse Micelles 4. Self-Reproduction of Vesicles 4.1. ‘‘Small’’ Vesicles 4.2. The ‘‘Matrix’’ Effect 4.3. Theoretical Considerations and Modeling of Self-Reproduction 4.4. Homeostatic Systems 4.5. Giant Vesicles 5. Vesicle-based Semisynthetic Cells and Their Self-Reproduction 5.1. The Minimal RNA Cell 6. Final Remarks Acknowledgments References Subject Index

222 223 223 226 230 235 236 240 246 247 248 250 254 256 258 258 265

This page intentionally left blank

PREFACE

Volume 7 covers the results of recent research on both planar lipid bilayers and spherical liposomes from many scientists working in this exciting field. Many of the contributors still remember their past, long lasting collaboration with the late Prof. H. Ti Tien, the founding editor of this book series. In this volume, chapters are included from the younger generation of scientists who follow this interdisciplinary field of research. The reason is because there are many new venues to follow and we are convinced that this interdisciplinary field will continue to flourish for many years to come. The Late Prof. H. Ti Tien wanted to see these ideas growing further in many areas of interface sciences and he also started to develop many useful practical applications in the field of biosensors, new molecular electronic devices, etc. As is evidenced by the six already published volumes, as well as this volume, the research based on planar lipid bilayers and spherical vesicles has developed into a great interdisciplinary field. A conventional bilayer lipid membrane (BLM) ~5–6 nm thick is interposed between two aqueous solutions. It is, together with spherical vesicles (liposomes), the most widely used model of biomembranes. The liquidcrystalline BLM has served over many decades of the experimental work as a barrier, a conduit for transport, a reactor for energy conversion, a transducer for signal processing, a bipolar electrode for redox reactions, and a site for molecular recognition and other application purposes such as the very advanced disease research. A biomembrane is a selectively permeable barrier, which is capable of material transport. The process may be accomplished in many ways: by passive, simple diffusion or by facilitated diffusion or by active transport. Biomembranes are described by the so-called ‘‘dynamic membrane hypothesis,’’ which helps to explain more clearly the basic membrane function. The self-assembled lipid bilayer is in a dynamic and liquid-crystalline state. A functional biomembrane should be considered from a molecular point of view as well as from an electronic point of view. It can support both ion and electron transport, and it is also the site of cellular activities because it functions as a ‘‘device’’ for either energy conversion or signal transduction. Such a system, as we can assume, must act as some kind of transducer capable of gathering information, processing it, and then delivering a response based on this information. Volume 7 of the Advances series continues to include invited chapters from many contributors all over the world on many different topics of both these reconstituted systems, namely planar lipid bilayers and spherical liposomes. We continue to invite newcomers from a broad spectrum of this interdisciplinary field to write a chapter for the Advances series, and at the same time we provide enough space for wellestablished and experienced researchers. We keep in mind the far-reaching goal for both systems, namely planar lipid bilayers and spherical liposomes, for the future development of this interdisciplinary research field world wide. All chapters xi

xii

Preface

presented in Volume 7 have one feature in common: they are dedicated to further experimental and/or theoretical exploration of the planar lipid bilayer system and spherical liposomes. We wish to thank all contributors for their work in preparing and writing their respective chapters and in that way honoring the legacy of the late Prof. H. Ti Tien. Their newly achieved results and their interpretation are highly appreciated by all members of the scientific community in this interdisciplinary research area. We, the editor and the editorial board of this Advances series, would like to thank every author whose chapter appears in this volume. We will continue to do our best to keep this Advances series alive in both fields covering the planar lipid bilayers and spherical liposomes. In future volumes we will dedicate each volume to both topics, planar lipid bilayers and spherical liposomes, because progress is advancing very fast in both research areas. We plan to continue our challenging and interesting work of this Advances series. In this way we pay tribute to the late Prof. H. Ti Tien, who called himself a scientific missionary and who continued to spread all over the world the latest results of his experimental work on BLMs in the last four decades. Angelica Leitmannova Liu (Editor)

CONTRIBUTORS

Haixin Bai State Key Laboratory of Electroanalytical Chemistry, Changchun Institute of Applied Chemistry, Chinese Academy of Sciences, Changchun, Jilin 130022, P.R. China College of Chemistry and Chemical Engineering, Henan Institute of Science and Technology, Xinxiang 453003, P.R. China Miha Fosˇnaricˇ Laboratory of Physics, Faculty of Electrical Engineering, University of Ljubljana, Trzˇasˇka 25, SI-1000 Ljubljana, Slovenia Didier Fournier Institut de Pharmacologie et Biologie Structurale, 205 rte de Narbonne, Universite´ Paul Sabatier, UMR5089, Toulouse F-31077, France Enrico Gallucci Department of Farmaco-Biologico, Universita` degli Studi di Bari, 70126 Bari, Italy Alesˇ Iglicˇ Laboratory of Physics, Faculty of Electrical Engineering, University of Ljubljana, Trzˇasˇka 25, SI-1000 Ljubljana, Slovenia Malgorzata Kotulska Institute of Biomedical Engineering and Instrumentation, Wroclaw University of Technology, 50-370 Wroclaw, Poland Veronika Kralj-Iglicˇ Laboratory of Clinical Biophysics, Faculty of Medicine, University of Ljubljana, Lipicˇeva 2, SI-1000 Ljubljana, Slovenia Krystian Kubica Institute of Biomedical Engineering and Instrumentation, Wroclaw University of Technology, 50-370 Wroclaw, Poland Angelica Leitmannova Liu Membrane Biophysics Laboratory, Biomedical and Physical Sciences Building, Department of Physiology, Michigan State University, East Lansing, MI 48824, USA Department of Microelectronics, Faculty of Electrical Engineering and Information Technology (FEI STU), Slovak Technical University, 812 19 Bratislava, Slovak Republic Xiaohua Liu State Key Laboratory of Electroanalytical Chemistry, Changchun Institute of Applied Chemistry, Chinese Academy of Sciences, Changchun, Jilin 130022, P.R. China

xiii

xiv

Contributors

College of Chemistry and Chemical Engineering, Henan Institute of Science and Technology, Xinxiang 453003, P.R. China Pier Luigi Luisi Biology Department, University of Rome ‘‘RomaTre,’’ Viale G. Marconi 446, 00146 Rome, Italy Daniela Meleleo Department of Farmaco-Biologico, Universita` degli Studi di Bari, 70126 Bari, Italy Silvia Micelli Department of Farmaco-Biologico, Universita` degli Studi di Bari, 70126 Bari, Italy Raghavendra Palankar Jacobs University gGmbH, Campus Ring 1, D-27725 Bremen, Germany Vittorio Picciarelli Department of Interateneo di Fisica, Universita` degli Studi di Bari, 70126 Bari, Italy Yannic Ramaye Jacobs University gGmbH, Campus Ring 1, D-27725 Bremen, Germany Blazˇ Rozman Department of Rheumatology, University Medical Center, Vodnikova 62, SI-1000 Ljubljana, Slovenia Tomazˇ Slivnik Laboratory of Physics, Faculty of Electrical Engineering, University of Ljubljana, Trzˇasˇka 25, SI-1000 Ljubljana, Slovenia Snezˇna Sodin-Sˇemrl Department of Rheumatology, University Medical Center, Vodnikova 62, SI-1000 Ljubljana, Slovenia Pasquale Stano ‘‘Enrico Fermi’’ Study and Research Centre, Compendio del Viminale 00184 Rome, Italy and Biology Department, University of Rome ‘‘RomaTre,’’ Viale G. Marconi 446, 00146 Rome, Italy Claudia Steinem Institute for Organic and Biomolecular Chemistry, Georg-August University, Tammannstr. 2, 37077 Go¨ttingen, Germany H. Ti Tien Membrane Biophysics Laboratory, Biomedical and Physical Sciences Building, Department of Physiology, Michigan State University, East Lansing, MI 48824, USA Erkang Wang State Key Laboratory of Electroanalytical Chemistry, Changchun Institute of Applied Chemistry, Chinese Academy of Sciences, Changchun, Jilin 130022, P.R. China

Contributors

xv

Mathias Winterhalter Institut de Pharmacologie et Biologie Structurale, 205 rte de Narbonne, Universite´ Paul Sabatier, UMR5089, Toulouse F-31077, France Masahito Yamazaki Integrated Bioscience Section, Graduate School of Science and Engineering, Shizuoka University, 836 Oya, Suruga-ku, Shizuoka 422–8529, Japan Department of Physics, Faculty of Science, Shizuoka University, Shizuoka 422–8529, Japan Innovative Joint Research Center, Shizuoka University, Hamamatsu 432–8011, Japan Lixue Zhang State Key Laboratory of Electroanalytical Chemistry, Changchun Institute of Applied Chemistry, Chinese Academy of Sciences, Changchun, Jilin 130022, P.R. China

This page intentionally left blank

C H A P T E R

O N E

Random Processes in the Appearance and Dynamics of an Electropore in a Lipid Membrane Malgorzata Kotulska1,* and Krystian Kubica1 Contents 1. Introduction 2. Lipid Membrane—Structure and Dynamics 2.1. Lipid Conformation 2.2. Random Movements of Lipids Molecules 2.3. MC Methods in Modeling Membranes 3. Randomness in Electroporation 3.1. Induction of a Pre-Pore 3.2. Appearance of a Hydrophilic Pore 3.3. Ionic Concentration and Membrane Susceptibility to Electroporation 4. Randomness in the Dynamics of Long-Lived Electropores 4.1. Creating a Single Long-Lived Electro-Nanopore 4.2. Analytical Methods for Stochastic Processes with a Memory 4.3. Stochastic Characteristics of the Electropore Fluctuations 4.4. Modeling Stochastic Processes of a Power-Law Spectrum 5. Summary References

2 4 4 6 7 12 12 16 17 19 19 21 26 30 31 32

Abstract The chapter presents basic principles of the randomness in the dynamics of the electropores. It describes the experimental techniques that permit to create an electro-nanopore and observe its dynamics. Consequently, the analysis of the results leads us to infer some knowledge on the processes in the membrane before and after electroporation. The pore appearance and its dynamics is a random process related to the membrane structure and dynamics. It concerns the pre-pore induction and its conversion into the hydrophilic nanopore, as shown by Monte Carlo (MC) modeling referring to the randomness of the membrane processes. Similarly, long correlations and self-similarity in the electropore dynamics may be well associated with the nature of lipids’ self-diffusion, sensitive to the experimental conditions. * Corresponding author. Tel.: þ48-71-3203974; Fax: þ48-71-3277727; E-mail address: [email protected] (M. Kotulska). 1

Institute of Biomedical Engineering and Instrumentation, Wroclaw University of Technology, 50-370 Wroclaw, Poland

Advances in Planar Lipid Bilayers and Liposomes, Volume 7 ISSN 1554-4516, DOI: 10.1016/S1554-4516(08)00001-X

#

2008 Elsevier Inc. All rights reserved.

1

2

M. Kotulska and K. Kubica

1. Introduction Cells and sub-cellular organelles of living organisms are encapsulated and separated from the environment by the plasma membrane, which is composed mainly of lipids. The biological membrane includes organized domains of several lipids, mainly from the group of phospholipids. Because of the hydrophobic layer, the membrane is almost impermeable to inorganic ions, such as Kþ, Naþ, Cl, Ca2þ, and so on. Only some very small compounds or uncharged polar molecules have a chance to penetrate the membrane. Membrane proteins, forming ionic channels, carriers, and pumps, mediate a controlled transport of biologically relevant molecules. Lipid membranes are sensitive to the physicochemical conditions, such as biochemical environment, temperature, mechanical stress, and electromagnetic field. It was discovered that some of these factors can induce pores in the membrane. This phenomenon was observed in the electric field of high intensity and labeled electroporation (reviewed in [1–4]). First experimental report on electroporation appeared in 1958 from Sta¨mpfli [5]; regular study started in the 1970s [6, 7]. Pores also appear at high temperature [8, 9] and under mechanical stress [10]. It has not been fully recognized if the mechanisms underlying these phenomena are similar. The expansion of the electropores may result from the mechanical pressure exerted on the membrane by the electric field (Maxwell effect), which affects membrane molecular organization and its geometry [11]. However, the Maxwell effect may not be the only mechanism underlying the electroporation. Electroporation experiments, Monte Carlo (MC) and molecular dynamics (MD) computer simulations of this phenomenon show how an electropore may evolve. First, pre-pores appear, which are converted into hydrophilic electropores. Hydrophilic electropores are surrounded by lipid molecules exposing their heads to the interior of the pore, which permits free flow of ions (Fig. 1). The pore appearance and its dynamics are statistical phenomena and it seems that they are closely related to the statistical features of the membrane structure and dynamics, dependent on the physicochemical environment.

Figure 1 An electropore in the lipid membrane. Adopted from [21] with permission from Biophysical Society.

Random Processes in the Appearance and Dynamics of an Electropore

3

Electroporation of a cell membrane was initially recognized by researchers as a harmful side effect of an electric field, which may put a cell life into jeopardy. Disruption of the membrane continuity, following application of a strong electric field, results in flow of ions and various biomolecules into and out of the cell, changing the biochemical contents of the cytoplasm. Under prolonged application of a high electric field, the electropores grow beyond the safe dimensions and the plasma membrane breaks up, resulting in the cell death. Pores significantly increase the membrane conductance and open a transport pathway for the molecules. Opening the new pathway through the plasma membrane has severe consequences for the cell, which is no longer separated from its environment. Various molecules, normally kept outside the cell, can get into the cytoplasm, avoiding the usual strict control. Although it is undesirable in the physiological conditions, it can be useful in a therapy. Because of the potential applications in medicine and technology, electroporation attracts a great interest. Genetic material and drug molecules can be more easily transported into the cell across the impermeable barrier of the plasma membrane. A very important application of electroporation is gene therapy or gene vaccination, where it has become a standard method for delivering DNA or RNA into cells since the 1980s [12]. In the late 1990s, the electroporation laid foundation for the electrochemotherapy (ECT), which improves the efficiency of cancer treatment [13]. Spectacular successes were achieved especially in the therapy of melanoma. Other types of cancer are considered as a subject for ECT. Moreover, it was recently shown that ultrashort nanosecond pulses alone are capable of eradicating melanoma cells with no use of cytostatic drugs [14]. The studies indicate the possibility of the electroporation-induced apoptosis in the tumor cells subjected to nanosecond pulses. Electroporation has also been observed as a side effect of the heart defibrillation. Although it can be lethal to a certain number of cardiac cells, there are reports that the electroporation accompanying the defibrillation is a necessary factor to terminate the reentrant spiral waves in cardiac arrhythmia [15]. However, the mechanism of electroporation is still not sufficiently explored and not fully understood. Because of the small size, combined with very high dynamics, the electropore cannot be visually observed. Currently, only indirect experimental methods supported with models and computer simulations are employed for the study. The main objectives of the studies on the electropores include better understanding of the mechanisms leading to the electroporation and its progress, sensitivity of this process to the physicochemical conditions of the environment, and the development of methods for a better control of the electropore size and opening time. Especially methods for obtaining big and fairly stable electropores that would enhance a delivery of big impermeant molecules are of interest. A major problem with the electroporation is the uncontrollable growth of the electropore when the potential is clamped at the original value above the breaking potential. On the other hand, too low a value of the potential leads to the resealing of the electropore. To prevent an irreversible rupture of the membrane, several methods have been proposed. The most commonly used method is a pulsating electric field, where the potential drops before the electropore grows excessively. Typically, short electric pulses are utilized to electroporate the membrane. The use of ultrashort electric pulses [14, 16, 17] and various periodic signals is also under

4

M. Kotulska and K. Kubica

study [18]. The pulsating electric field is used in the medical applications of the electroporation for delivering molecules into the cells. Although this method protects the membrane, it does not permit regulating the electropore size and lifetime. Also observation of the natural evolution of the pore and its dynamics is not possible in the pulsating field. Another method is electroporation by a current-controlled method [19, 20]. This technique introduces a negative feedback that opposes an excessive growth of the electropore and produces a single long-lived pore with a diameter of the order of nanometers. The data collected from a current-controlled experiment by the chronopotentiometry (ChP) permits studying the stochastic properties of such a long-lived electropore. The only possibility of studying a long-lived electropore without any feedback is Chronoamperometry After Current Clamp (CACC) electroporation—a novel experimental method, which combines advantages of the current-clamp and voltage-clamp approaches [21]. The study by this method showed that the membrane is exposed to the danger of the irreversible break mainly during the process of the electropore creation. The membrane can tolerate well-adjusted voltage-clamp conditions after the pore has been created and the pore edge fully formed. This method creates a single electro-nanopore, which naturally maintains quite stable under voltageclamp conditions and its diameter can be regulated by the current value at the first stage of the experiment. An electropore, studied under constant potential by CACC, shows naturally fluctuating dynamics that is sensitive to the environment and exhibits very interesting stochastic properties. Thorough understanding of the electroporation phenomenon is necessary for its full application in medicine and technology. Therefore, new experimental and modeling methods are sought after. These methods may help to control the electroporation by the membrane composition or physicochemical conditions of the environment leading to generation of an electro-nanopore with a designed diameter and lifetime, limited to cells of a specific location or type. Stochastic characteristics of the electropore may bring about new insight into the mechanisms of electroporation. This chapter reviews current state of the art concerning the randomness of electropores and its possible association with the membrane structure and dynamics.

2. Lipid Membrane—Structure and Dynamics 2.1. Lipid Conformation Lipids present 30–80% dry mass of biological membranes [22]. Therefore, the research on biological membranes is often focused on purely lipid membranes. Such simplified experimental models can address many questions related to biological membranes. Properties of a lipid membrane (bilayer) are strictly related to the structure of a lipid molecule. Detailed analysis of some pure lipids in crystal form has been done by Hauser and coworkers [23, 24]. They defined membrane properties often used in the membrane research, such as torsion angles, layer interface, thickness of the polar region, head group, chain tilt, and chain cross-section. Theoretical results indicate four regions in the lipid molecule: (1) perturbated water, where water structure is

Random Processes in the Appearance and Dynamics of an Electropore

5

affected by the membrane surface; (2) interface; (3) high tail density and low free volume; and (4) bilayer center [25]. The thickness of the fully hydrated polar part of the membrane is valued to 1 nm [25, 26] and the chains—hydrophobic part of a bilayer—to 4 nm [25]. Experiments performed on lipids in crystalline form have shown that the first two carbon atoms in beta chain lie in membrane plane. As a result, the effective length of both lipid chains can be different even for chains with the same amount of carbon atoms. It may play a role in some membrane transitions such as a fusion or pore formation. Most lipid chains in animal cells are built from 16 to 24 carbon atoms. The beta chain is usually saturated, whereas the gamma chain usually has 1–3 unsaturated bonds [26]. Papahadjopoulos and Kimelberg [27] have shown that the lipid bilayer permeability grows with the degree of unsaturation. Bilayer formation results from amphiphilic properties of lipid molecules. The environmental conditions of experimental work are related to polar head structure (size, charge, and hydration) and chain characteristics (length and degree of saturation). For example, lysolecithin with only one chain does not form bilayers, and phosphatidylethanolamine only creates bilayer in pH > 11 in a limited range of the temperature. Experiments with some nonbiological amphiphilic compounds, synthesized in a laboratory, show also their ability to form bilayers [28]. These examples demonstrate that bilayer existence depends on hydrophobic effects resulting from chain and head properties. Publishing in 1972, the fluid mosaic membrane model by Singer and Nicolson [29] forced the scientific world to change the views on the lipid part of biological membranes. According to this model, a biological membrane is composed of mobile lipids and proteins. Proteins are immersed in the lipid bilayer in such a way that their polar parts are oriented toward water and the hydrophobic parts build the inner membrane component. Lipid bilayer X-ray studies have shown how important are the lipid chain rotational isomer analyses to understand the structure and membrane function [30]. A large number of lipid chain rotational isomers are the result of the rotational freedom around C–C bonds. The lowest rotational energy, equal to 0, is attributed to trans conformation (anti-periplanar) [31]. The gauche (anti-clinal) rotation around C–C bond of an angle 112 is only 0.6 kcal/mol CH2 higher. When the energy values of these conformations (trans and gauche) are compared to cis (syn-periplanar) rotational energy (5.64 kcal/mol CH2), the trans and gauche conformations are almost equally stable. If a given hydrocarbon chain has gauche bonds and there is at least one trans bond between them then such a configuration is labeled as kink [32]. Each kink shortens the effective chain length of a 0.127-nm and increases the volume occupied by such a chain of 0.25–0.50 nm3. Larger number of kinks cause proportional changes in the effective length and volume. Small energy differences between gauche and trans bonds permit to treat the appearing kinks as instable structures able to ‘‘move’’ along the chain. Many membrane phenomena have been explained based on the trans–gauche transitions [32–35]. The idea of a kinkblock appearance, due to cooperative interactions, has been also discussed [36–38]. The chain packing density in kink-blocks has been estimated at 0.22–0.25 nm2 [31] and when the upper value is reached the kink-blocks are transformed into loosely packed gauche-blocks (0.33–0.4 nm2). Such structures enable incorporation of some new membrane components [39]. The kink–gauche block transitions can

6

M. Kotulska and K. Kubica

influence the membrane transport, energy transformation, transfer of information, membrane surface enzymatic reactions, contact breaking, or immunological response [31]. These transitions can be evoked by changes of the ionic strength or by interaction with membrane proteins. The chain length is also a salient feature of a lipid [40, 41]. According to Nagle [33], bilayers formed from the lipids whose chains have different length undergo the transition mentioned above at lower temperatures. NMR studies have shown greater mobility of terminal CH3 groups. This observation allowed for simplification of some theoretical studies on bilayers to only one layer analysis. The gel–fluid main phase transition observed in lipid bilayers has been obtained by Scott [34] on 21-chain state theoretical model. Although the theoretical temperature of the transition significantly exceeded the temperature observed in experimental techniques, Scott proved that such studies are useful. He showed the role of van der Waals interactions in this process. The van der Waals energy was estimated as dependent on chain packing density and its negative value indicates membrane stabilization. Differences in chain transitions, observed in lipid monolayers [42] and bilayers, can result from polar head mobility. Stabilizing van der Waals energy should balance destabilizing polar head interaction energy. It suggests that the lipid polar head can influence the hydrophobic part of the membrane [43–45]. Some defects in the hydrophobic membrane part, evoked by differences in chain and polar head geometry, that is, triangular for the chains and square for the heads, may also appear [42]. It seems well justified that almost all membrane processes are based on pure statistical changes but discovery of the membrane asymmetry requires revision of some views on the lipid role in biological membranes. The asymmetric lipid molecule distribution results from flipases selectivity or a controlled flip-flop diffusion. Other mechanisms responsible for the membrane asymmetry are related to the membrane and lipid vesicle fusion, where the flip-flop—the statistical phenomenon— may be the first step in it. The asymmetric lipid distribution in both membrane layers plays a role in lipid domain formation that participate in regulation of some biochemical reactions; some lipids are responsible for protein anchoring in membranes. Others, such as the phosphatidylinositol, play an important role in a signal transduction [46].

2.2. Random Movements of Lipids Molecules Lipid molecules building membranes move in the membrane plane (lateral diffusion), rotate around the axe parallel to the membrane normal and can be spontaneously shifted between lipid layers (flip-flop). Both processes show statistical features dependent on the membrane structure and physicochemical environment. The experiments by nanovid microscopy have shown that lipids in the membrane undergo random Brownian motion [47]. The diffusion of the molecules is random, however, rather short ranged in the physiological conditions (e.g., [48]). The flip-flop process was studied for single lipids as well as for biological membranes, where it is particularly important regarding membrane asymmetry. The asymmetry of the biological membrane composition is mainly a result of some protein activity (flipase, translocase, and

Random Processes in the Appearance and Dynamics of an Electropore

7

scramblase) but the role of random flip-flop should be also considered in some membrane processes. Experiments on small liposomes, performed almost 40 years ago, showed that half of lipid molecules, with regard to the liposome composition, move between lipid layers within 6 h [49] to 80 days [50]. A long period of time needed for flip-flop observation leads to a conclusion that this is a slow process. The halftime of flip-flop process depends on many environmental conditions and can last shorter than several seconds [51, 52]. It has been shown that the more unsaturated the lipid acyl chain, the faster is the flip-flop process [53]. The chain length [54, 55] and the amount of chains in one lipid molecule are also important. A decrease in alkyl chain length increases the rate of flip-flop [56]. For the fatty acid, the rate constant decreases exponentially with the increasing chain length [57]. The molecule translocation is faster for the derivatives with two acyl chains when compared with a single acyl chain [58]. The temperature studies have shown that the flip-flop rate constant also increases with the temperature [57]. The influence of the temperature on the transbilayer movement can be reduced by addition of cholesterol, which is particularly visible at the phase transition temperature [59]. The curvature of lipid membrane also plays a role [60]. Flip-flop is faster for small unilamellar vesicles (SUV) and slower for large (LUV) and giant (GUV) unilamellar vesicles [57]. The influence of ultraviolet irradiation on lipid translocation has also been studied [61]. Some compounds modify the membrane properties, for example, anionic amphiphilic polymers due to electrostatic binding accelerate the flip-flop process of anionic lipids [62]. The presence of caveolin-1 and cholesterol molecules influences the lateral movements and the flip-flop kinetics by changing the membrane organization [63]. The flip-flop process has also been studied with reference to drug uptake [64, 65], fatty acids uptake [66], and cholesterol transfer between lipid vesicles [56]. Incorporation of GALA-30, amino acid synthetic peptide that forms a transmembrane pore in lipid membrane, also accelerates the flip-flop [67]. The influence of the electric field on the flip-flop in erythrocytes [38] allows assuming that the process is important in pore formation [3, 17]. Therefore, changing its rate by the experimental conditions may significantly impact the rate of a pore appearance.

2.3. MC Methods in Modeling Membranes In the MC simulation of the simplest form, a large number of random trials are run to explore the search space. The MC simulations are very useful to solve problems from statistical mechanics, and to model systems with a large number of interacting molecules, which take into account stochastic aspects of the studied phenomena. In case of large numbers of the particles, deterministic theoretical models do not reflect the reality of such complex systems. In the simulations, each microstate appears with a certain probability. A microstate is given by a set of mechanical variables ! Oi ¼ ðOi1 ; . . . ; OiN Þ, representing all possible degrees of freedom of N molecules ! ! that constitute the system. Each element Oi that belongs to the phase space fO g of all possible microstates is represented by the values of the microstate variables, for example, spatial coordinates, conformation, charges, and so on. The mechanical vari! ables are associated into the Hamiltonian HfO g, whose exact formula depends on the

8

M. Kotulska and K. Kubica

system and the applied model, reflecting the energetical state of the system. In the canonical description of lipid membranes, the probability of a microstate is given by the Boltzmann distribution, dependent on the Hamiltonian value (e.g., [68]): !

eHðO Þ=kT ; pðO Þ ¼ Z !

ð1Þ

where Z is the molecular partition function obtained by integrating the phase space over all possible microstates. For a discrete system, Z can be defined in the following way:



X

!

eHðO Þ=kT

ð2Þ

!

fOg

The macroscopic state of the system is!represented by a weighted average of all possible microstates. By calculating pfOg, the thermodynamic value of any ! macroscopic physical quantity f fOg can be obtained according to:

hfi ¼

X

!

!

f ðO ÞpðO Þ

ð3Þ

!

fO g

However, in the development of the classical Boltzmann distribution, it was assumed that a large number of possible different microstates are known. Practically, it is difficult to obtain the analytical calculation of the partition function Z, which is possible only in very simplified situations. In 1953, Metropolis et al. [69] proposed a new sampling procedure where this problem was solved by replacing the set of all ! microstates by a representative subset of M possible microstates fOi gM i¼1 , found in the simulation. In the limit, M ! 1, each microstate occurs with the frequency described by the Boltzmann probability distribution function (PDF). This approach has correct limiting properties if the transition probabilities fulfill the balance condition: !

!

!

!

!

!

pfOj gpfOj ! Ok g ¼ pfOk gpfOk ! Oj g;

ð4Þ

and the steady-state condition: !

pfOj g ¼

M X

!

!

!

pfOi gpfOi ! Oj g

ð5Þ

i¼1

The chosen probability of the state passage has to satisfy the ergodic rule; that is, each state is reached from any other state by a limited number of steps. The process in the phase space is a Markov process. Therefore, the process has no memory; that is, the probability of assuming any microstate depends only on one preceding state and it does not depend on the history of the previous transitions.

Random Processes in the Appearance and Dynamics of an Electropore

9

From Eq. (3) a value of any physical quantity can be obtained. The partition ! function can be calculated from the subset fOi gM i¼1 , where the sum is over the states available to the system:

h fi ¼

X

!

!

f ðOi ÞpðOi Þ

ð6Þ

! M fOi gi¼1

As a result, each quantity can be approximated as a simple average:

PM h fi 

!

f ðOi Þ M

i¼1

ð7Þ

This approximation becomes exact as the number M of MC trials becomes infinite. The sampling procedure of the Metropolis MC simulation has the following algorithm: !

Choose an arbitrary initial configuration, O1 . ! Choose randomly a trial state O20 according to the transition probability p12. Calculate the energy change DH. ! ! If DH  0, the transition to the trial state is accepted, that is, O2 ¼ O20 . Else (DH > 0), draw a random number n 2 [0,1] generated by a uniform probability distribution. If eH=kT > n, the transition to the trial state is accepted, ! !0 ! ! that is, O2 ¼ O2 , else the transition is rejected. O2 ¼!O1. 6. Return to step (2) by randomly choosing a trial state O30 , and so on.

1. 2. 3. 4. 5.

In the limit of a large number of accepted transitions (MC steps), a generated sequence of microstates is distributed according to the Boltzmann distribution. The MC simulations of the lipid membrane assume the existence of lipids aggregated into a membrane. Typically, a triangular lattice of lipid molecules represents a lipid layer, where each vortex of a triangle is occupied by a single acyl chain of the lipid molecule (Fig. 2). In the simulation, periodic conditions are imposed on the boundaries of the whole lattice. Although the lattice approximation affects the model accuracy, the MC methods permit modeling long-scale phenomena, which is not possible by numerically more exact MD simulations. The model of microscopic interactions, formulated for the simulations, accounts for the forces between constituents of the system in terms of conformational states. The MC simulations of lipid monolayer and bilayers (reviewed in [70]) significantly contributed to the knowledge on the membrane phase transitions. In 1980, Pink’s model was published [71] and has become the base of much research on membranes. The theory of chain conformations has been used to predict the intensity of Raman line obtained from a dipalmitoylphosphatidylcholine. Good agreement with the experimental data has encouraged further research. According to the model assumption, each lipid chain occupies one node in triangular lattice and can take 1 of the 10 defined conformational states. Each state characterizes internal

10

M. Kotulska and K. Kubica

M

Figure 2 Representation of a lipid bilayer by the triangular lattice. Each chain occupies a site denoted by a dot; the arrow represents a lipid head.

energy, energy degeneracy, area per chain, and chain length in C–C bond units. Among the 10 states, two have limiting characteristics: the all-trans state with zero internal energy and the longest effective length, which has the degeneracy equal to 1. This state is characteristic for the gel phase. State characteristic for the fluid phase is cold melted. It has the greatest value of degeneracy and the internal energy. It has also the shortest effective length and occupies the maximal area. The remaining eight states are less accurately defined; some of them may have the same effective lengths, internal energies, degeneracy but these values may not be identical simultaneously. In Pink’s model, the contribution to the Hamiltonian comes from three components: conformational states of single chains Hconf, van der Waals interactions between nonpolar chains HvdW, and interactions between polar heads Hs:

H ¼ HvdW þ Hconf þ Hs

ð8Þ

van der Waals interactions are calculated according to [72]:

HvdW ¼ 

N X 10 J0M X f ðrnm ÞSn Sm Lni Lmj ; 2 i; j¼1 n;m¼1

ð9Þ

where J0M denotes the interaction energy between two parallel chains in all-trans conformation. Lattice coordinates are represented by i (site index ranging from 1 to N ) and j (index of six sites neighboring with site i). Indexes of chain conformational states, n and m, range from 1 (all-trans) to 10 (fluid). The distance rij between two chains at sites

Random Processes in the Appearance and Dynamics of an Electropore

11

i and j depends on the chains conformational states. The state operator Lni of the chain located in site i takes 1 if the i-th chain is in the conformation n, and 0 otherwise. The order parameter Sn for acyl chain in conformation n is determined by the relation:

P Sn ¼ P

p Snp

ð10Þ

p S1p

The bond is characterized by the angle ynp between normals to the bilayer and the plane spanned by the p-th CH2 group of the chain. Snp order parameter of the p-th C–C bond can be calculated as:

1 Snp ¼ ð3 cos2 ynp  1Þ 2

ð11Þ

van der Waals interactions between chains in the m-th and n-th conformations depend on inter-chain distance rnm according to:

 f ðrnm Þ ¼ wn

r12 rn rm

5=2 ;

ð12Þ

where rn denotes the radius of the space occupied by an average chain in the n-th conformation. A weakening factor wn was introduced into the basic Pink model [72] to improve the agreement between the model and experimental data. The factor is related to the large number of different fluid conformations, hence it decreases the calculated interactions between fluid chains: w10 ¼ 0.4 if n ¼ 10 (chains are in fluid state) and wn ¼ 1 if n 6¼ 10. The conformational energy Hconf is defined in relation to the internal energies En of the chains in conformations occurring in the studied system (Lni ¼ 1), tabularized in [71]:

Hconf ¼

N X 10 X

En Lni

ð13Þ

i¼1 n¼1

For over 20 years, the Pink’s model has been intensively used. Based on this model, many membrane-related problems have been studied: gel–fluid phase transition [72–74], acyl chain density fluctuations [75], the effects of cholesterol [76, 77], the influence of foreign molecules on membrane properties [78], density fluctuations and partitioning of foreign molecules into bilayers [79], the problem of membrane heterogeneity [80, 81], lateral distribution of proteins in membranes [82], lipid– protein interaction [82–84], dynamical order in membranes [85], drug research problems [86], dynamics of phase separation in membranes [87], biological activity of ionic modifiers [88], ripple phase modeling [44], the role of lipid polar head

12

M. Kotulska and K. Kubica

[45, 89], membrane influence on fluorescein-PE fluorescence intensity [90], and finally an electropore appearance [91]. Another membrane model studied by MC method, presented by Sugar et al., takes into consideration only the conformational interactions [92]. The head–head interactions are omitted because their contribution to the monolayer energy is constant. The energy of the membrane consists of the intra- and inter-chain energies. The conformational energy of the system is expressed by the cooperativity parameter. Based on such assumptions, the behavior of the two-component lipid system is discussed regarding the role of weak and strong repulsion between molecules and the critical concentration as well. There are also other attempts to model lipid membranes by MC methods, for example those discussed in the next section, which are of interest with regard to modeling a pore appearance (see Section 3.1).

3. Randomness in Electroporation 3.1. Induction of a Pre-Pore Pre-pore excitation develops as a first stage of the electroporation process when the lipid membrane is exposed to an electric field. At this stage, the membrane evolves into an anomalous excited state, which can be termed as a ‘‘stress state.’’ The stress state can remain for tens of minutes after removing the electric field. Although no hydrophilic pores appear at this stage of electroporation, the membrane displays an increased conductance and current fluctuations. These effects can be observed even at a quite low voltage (120 mV). The first step of the pre-pore excitation, called ‘‘induction step’’ [93], results from randomly localized loose sites in the membrane structure. Such sites may also occur in the lipid membrane at the physiological conditions, which can be explained by the thermal motion of the lipid molecules. It seems very probable that a number of these sites increase when the electric field is applied, facilitating the onset of electroporation. A molecular mechanism leading to these defects is still not sufficiently explored. MD simulations show [94–97] that in the first stage of the electroporation, water molecules penetrate the hydrophobic core of the bilayer, forming highly irregular wire-like structures (pre-pores) that can join each other. When the pre-pores increase their size to the nanometer scale, the lipid headgroups start to translocate to the interior of the bilayer. The translocation process is much slower than the reorientation of the solvent molecules at the bilayer– water interface [94] and it may be sensitive to the ions in the solvent. Subsequently, nonregular hydrophilic pores appear, which surround and stabilize the water columns. The pore shape is not cylindrical as usually postulated in modeling, although it becomes more regular as the pores expand. The oldest models explaining the initiation of the electroporation are based on variations of the electrocompression theory. Because of this theory, free charges on the water–membrane interface, driven by the electric field, exert a mechanical pressure on the membrane. The pressure induces a continuous deformation of lipid molecules. The hydrocarbon chains of the molecules tend to assume anomalous

Random Processes in the Appearance and Dynamics of an Electropore

13

squeezed conformations, different from the gel or fluid conformations [98]. The overall decrease in the membrane thickness would be a result of such a transition. However, the electrocompression theory does not fully agree with experimental facts, hence it is still debatable. First, it does not explain randomness of the pore formation, which is a salient feature of this phenomenon. Secondly, the total deformation of the membrane, predicted by the electrocompression theory, should result in significant decrease in the membrane capacitance [3]. Experiments show that in the stress state, the membrane capacitance, Cm, representing mostly the hydrophobic layer of the membrane, is usually constant to at least 0.5–2% of Cm [3]. Moreover, this weak increase in the capacitance is also related to changes in the membrane geometry, which is sensitive to intermolecular distances. The anomalous conformations of the molecules are not necessary to account for this effect. It is possible that the main result of the mechanical pressure exerted by the electric field is maintaining and extending the existing pores according to the energy balance rather than inducing anomalous conformations of acyl chains. Another mechanism that may be involved in electroporation can be related to electrostatic interactions between polar groups and the electric field, which would result in a randomly appearing increase of the distances between lipid molecules. Such loose sites may develop into electropores. The pore appearance can be studied by MC modeling. For example, Schillcock and Seifert [99] examined the behavior of multiple holes in a model fluid membrane. Their membrane model is based on three parameters: the barrier height hampering the hole creation, the linear and lateral tensions. The electric field affects the membrane by modulating the energy barrier. Hydrophilic pores that were observed resulted from the growth of the hydrophobic membrane defects in two steps: thermal fluctuations overcome an energy barrier creating the minimal-sized holes that evolve to the pores whose size is controlled by a constant line tension. The appearing pores take shapes different from circular. The conditions for the pore dynamics, such as their growth, shrinking, coalescence, fragmentation and disappearance were studied by this model. Majority of models studied by MC techniques are two-dimensional or quasitridimensional. An example of a tri-dimensional MC study of the semi-grand canonical ensemble is Muller and Schick model [100]. In this model, a pore appears as a result of the composition fluctuations. The examined model membrane is in equilibrium with the surroundings and the bilayer is held by interactions at its boundaries. The model is spanned on a cubic lattice, where hydrophilic and hydrophobic segments interact via a square well potential, extended over the nearest 54 lattice nodes. There is an attraction between the same kind of segments and repulsion between different segments. The absolute values of the interaction energies are equal. Although the model does not take into consideration the detailed molecular structure, the approximation is sufficient to observe the process of the pore formation. We explored the idea of the electropore induction by MC simulations based on the Pink model [91]. In the original Pink model, lipid heads are not represented explicitly by dipoles and the term Hs is based on the averaged lateral pressure, corresponding to Coulomb forces. This approach hampers modeling some

14

M. Kotulska and K. Kubica

membrane phenomena, for example, interactions between lipid molecules and the electric field. Therefore, in modeling the electropermeabilization, a model with explicit representation of the dipoles, which includes the energy of interactions between dipoles Hdip and the electric field He, is a more appropriate choice. In such a model, the Hamiltonian H involves four terms:

H ¼ HvdW þ Hconf þ Hdip þ Heq

ð14Þ

Contribution of the polar part Hdip, which replaced Hs in the classical model, can be approximated as:

Hdip ¼

N X 14 1X 2 i¼1 j¼1

X aQai bQbi expðkraibj Þ : 4peeraibj a;b¼1;1

ð15Þ

Here e is the electrolyte dielectric constant, e0 represents the permittivity constant, Qai (or Qbi) is the effective polar head charge with Qai ¼ q/2, where q is the actual dipole charge equal to the elementary charge [44], a denotes the charge sign: a ¼ 1 for a positive charge and a ¼ 1 for a negative one. The distance between charges a and b of the dipoles at sites i and j was represented by raibj. The inverse of Debye length, k, defines the range of electrostatic interactions with screening. The energy He, which reflects interactions between the polar parts of the molecules and the electric field, is defined in the following way:

He ¼

N X

Qi dEð1  cosðyÞÞ

ð16Þ

i¼1

where Qi is the dipole charge, d is the dipole length, y is the angle between the directions of the dipole and the electric field E, which can be represented by two extreme tilts toward the membrane surface normal 12 (standing) and 78 (lying). The MC simulations of the modified Pink model, where chains could only assume typical conformations and no additional mechanical pressure is considered, showed how electrostatic interactions alone can facilitate the appearance of the pore [91]. First, the tilt of the heads is sensitive to the external electric field (Fig. 3). The essential mechanism of the reorientation is based on electrostatic interactions between the head charges and the electric field. The polar heads barely reorient in the weak field, below U0  80 mV. However, when the potential exceeds the critical value Uc  250 mV, a massive collective reorientation of the polar heads can be observed. For some molecules, the change of the tilt affects their electrostatic interactions with adjacent lipids and alters inter-molecular distances and conformation of the chains. As a result, the average number of chains in the gel and fluid conformations changes, especially in the negative layer—introducing less compact structure (Fig. 4). The other layer becomes slightly more packed. This asymmetry may start the whole sequence of events, as described in the next section. The conformational effects are much less significant than in the phase transition, indicating that the sites

15

Random Processes in the Appearance and Dynamics of an Electropore

100

Standing dipoles [%]

80

60

40

20

0

100

102

104

106

108

1010

E [V/m]

Figure 3 The reorientation rate of the polar heads. When the potential is greater than Uc  250 mV, a massive collective reorientation of the polar heads can be observed in the negative membrane layer. 1.2

Fluid [%]

1.15

1.1

1.05

1

100

102

104

106

108

1010

E [V/m]

Figure 4 The electric field whose value is above the breaking potential UB, known from electroporation experiments, increases the rate of the lipid molecules in fluid conformation in the negative layer.

with less compact structure do not appear massively but sparsely and randomly. Statistically, only a very limited number of sites are able to give rise to pre-pores that may develop into the electropores. The existence of the threshold U0, below which dipoles show completely no reaction to the electric field, suggests that it is energetically favorable for the membrane lipids to maintain the same conformation when the membrane potential is within physiological limits occurring in cells. This observation is especially relevant for plasma membranes of excitable cells, indicating that lipids do not significantly change the characteristics throughout most part of the action potential.

16

M. Kotulska and K. Kubica

On the other hand, the critical value Uc above which the significant reaction of the molecules can be observed is very close to the experimentally determined electric field that results in the electroporation of planar membranes. It seems that the increased number of spots with loose fluid conformation, which are likely to appear for energetical reasons and asymmetry between both layers, may promote the appearance of pre-pores and facilitate subsequent electro-permeabilization.

3.2. Appearance of a Hydrophilic Pore Our MC study showed asymmetry in the layers subjected to the electric field: certain compression in the positive layer and a relaxation in the negative layer (Fig. 5). Changes in the hydrocarbon chain packing density evoke hard to predict changes in the polar head ordering. This imbalance in the packing density of the layers is very likely to encourage facilitated flip-flop process—a chain (only one) from the denser layer would move to the more relaxed layer. Statistically, for 200 chains, only one lipid chain would proceed from one layer to the other. Consequently, the polar head of this lipid molecule is forced into the interior of the hydrophobic membrane part (Fig. 6). Because it happens upon the influence of the electric field, we should also expect the reorientation of the lipid polar head at the positive side. Such a process lowers the energy of the system of about 4  1021 J per one reoriented head, at the electric field of intensity 108 V/m. It is comparable with kT value, which proves the statistical mechanism of the pore appearance. If this is a next step to the pore formation, it should also explain the thermally induced pores [99]. The head reorientation, as a step in a hydrophilic pore formation, would be significant because the lipid polar part is hydrated [101]. On average, 4.2 water molecules hydrate choline methyl group; 10.2 water molecules hydrate the N(CH3)3 group; 4.0 water molecules hydrate the phosphate, and 1.0 water molecules hydrate a carbonyl group.

E Figure 5 The first step of the pre-pore state. Electric field of the intensity above the threshold value changes orientation of the polar heads; it compresses and relaxes the lipid chain packing in the positive and negative layers, respectively.

17

Random Processes in the Appearance and Dynamics of an Electropore

E Figure 6 The second step of the pre-pore state. Flip-flop of one lipid chain from the positive to the negative layer can balance the chain packing densities in both layers. The chain movement accompanies the hydrated polar head reorientation. Adopted from [103] with permission from Elsevier.

Because of this, hydrated or water molecules can enter the created pore. In the next step, other polar heads would reorient—finally creating a hydrophilic pore. If the electroporation would proceed according to the presented scheme, there would be no place for a hydrophobic pore [102]. The subsequent electroporation would lead to the hydrophilic pore extension. To model such a process by a MC model based on the triangular lattice, for the geometrical reasons, it is necessary to remove one lipid from the triangular network. Although removal of the molecule as a first step in the pore creation model may be debatable, it is essential to preserve the triangular geometry of the system. It permits to connect two parallel triangular networks. The triangular geometry of such a structure allows creating pores of different sizes. The model always starts from the same molecular arrangement, preserves the total amount of lipid molecules in the membrane, and keeps the constant distance between layers. Although such a structure reminds a cylindrical shape with sharp edges, the pore walls would be curved to avoid exceeding the maximal chain packing. Since the distance between adjacent nodes, occupied by lipid chains, is not stable and depends on the lipid conformations, the described model of the pore may be useful in further studies [103].

3.3. Ionic Concentration and Membrane Susceptibility to Electroporation Electroporation experiments on lipid membranes confirm that there is an observable effect of ionic concentration on the membrane properties. Although the values of the breaking potential UB in the reversible electroporation do not seem sensitive to the ionic concentration of the solute [104, 105], membranes are less susceptible to irreversible rupture when surrounded with highly concentrated buffer [105, 106]. The quantitative change was observed during electroporation under current-clamp conditions. The membrane in such an experiment has much longer lifetime, and

18

M. Kotulska and K. Kubica

electropore fluctuations are more stationary when ionic concentration is high, which was shown in [106]. Other experiments confirm that the ionic strength has also an effect on the liposome stability [107]. The increased durability of the membrane may be related to stronger interactions between molecules, which would result in decreased self-diffusion of lipids. Indeed, experiments show that in highly concentrated monovalent salts, diffusion coefficients of the membrane molecules decrease [108]. Yet, a reason for the relation between ionic concentration and lipid dynamics is not clear for researchers. One possibility is some kind of adsorption of ions on the membrane. Nevertheless, some authors claim that interactions between monovalent ions and uncharged or zwitterionic lipids are too weak to account for the decreased self-diffusion [109]. To address this problem, the influence of ionic concentration on lipid membrane was studied by MD simulations [108, 110, 111]. The MD model showed that when the ionic concentration in the electrolyte is increased, cations from the buffer bind to the carbonyl oxygen in the polar part of the lipids, forming charged complexes of greater size and lower mobility [108]. The MD model shows fatty acyl chains changing their conformation and increasing order parameter, which is consistent with experimental results [30]. Lipid molecules tend to occupy smaller area, so the membrane is more tightly packed. The lipid headgroups also change their configuration by altering their tilt into a more standing position. The MC method permits studying a possible influence of the solute concentration on the membrane sensitivity to the electric field from a perspective different to the MD simulations. At the given ionic concentration, the most energetically stable conformation of lipid membrane can be examined, separating a possible influence of cations bound to the lipid membrane. The simulations based on the modified Pink model [112] showed that the rate of fluid molecules above the critical value of the electric field depends on the ionic concentration (Fig. 7). Hence, the membrane 1.2

Fluid [%]

1.15

1.1

1.05

1

100

102

104 106 E [V/m]

108

1010

Figure 7 Above the critical potential, the rate of fluid molecules depends on the buffer ionic concentration (dots at 10 mM, triangles at 1 M). The breaking potential is not sensitive to ionic concentration.

Random Processes in the Appearance and Dynamics of an Electropore

19

becomes more stable at higher ionic concentration, which may contribute to the longer lifetime of the electroporated membrane under current-clamp conditions [105]. On the other hand, the critical value Ec does not change with the ionic concentration. This would suggest that the breaking potential UB is independent of the concentration, such as indicated by experiments.

4. Randomness in the Dynamics of Long-Lived Electropores 4.1. Creating a Single Long-Lived Electro-Nanopore 4.1.1. Chronopotentiometry in current-controlled conditions The main problem in experiments on electroporated membranes is their huge sensitivity to the electric field, which can easily lead to an irreversible rupture of the membrane. This sensitivity is related to the energetical balance of the electroporated membrane and the difficulty to reach the equilibrium. If the electric potential after the electroporation is not reduced and kept constant above the breaking potential UB, required for creating a hydrophilic electropore, the electropore is likely to continue a rapid uncontrollable growth. As a result, the membrane is usually destroyed within less than a millisecond. The only possibility to protect the membrane against the rupture is decreasing the potential as soon as the electropore is created. However, even in such a case, obtaining a pore of a stable size is very difficult. The electropore either keeps growing or starts resealing. To prevent the irreversible rupture of the membrane, two methods have been developed. The most commonly used method is electroporation by short or ultrashort electric pulses. In this method the electropore size oscillates according to the changes of the electric field, closing between consequent electric pulses. Although it may be an acceptable behavior for transporting small molecules through the membrane, it does not permit regulating the electropore size and lifetime. It is not satisfactory if a large electropore is needed or the opening time should be long. Moreover, the oscillating field hampers studying natural properties of the electropore, such as the pore evolution, its natural dynamics, and sensitivity to the membrane composition and environment. The observation that the electropore size is sensitive to the potential level rather than the current flowing through the electropore laid the foundation for the idea of electroporating membranes under current-controlled conditions [19, 20, 113–115]. Chronopotentiometric (ChP) measurements on planar membranes under currentclamp conditions show that the electropore lifetime can be prolonged to an hour or even above. In the current-controlled experiment, the membrane stability is maintained by a negative feedback, which keeps the electropore within a safe dimensions limit. The mechanism of the feedback is based on the sequence of the following events. As the electropore grows, the transmembrane potential drops. Consequently, the electropore whose size is potential dependent, shrinks and the conductivity of the electropore is reduced. Then the transmembrane potential grows, so the electropore is likely to increase its diameter again. This sequence produces stochastic fluctuations of

20

M. Kotulska and K. Kubica

the electropore. The stochastic behavior of lipid membranes is related to their probabilistic nature—typical in biological systems [116]. An exemplary ChP curve is shown in Fig. 8. The time course begins with an exponential rise of the membrane potential due to the membrane capacitance. It is followed by a sudden decrease of the potential, which corresponds to a pore formation. The pore conductance can be estimated from the voltage fluctuations based on certain simplifying assumptions concerning the membrane electrical parameters and the pore shape. The problem with a nanopore conductance is not trivial. Nanochannels show nonohmic conductance, strongly dependent on the pore size, profile, and a concentration gradient. The models of I–V characteristics are usually based on the Poisson–Nernst–Planck (PNP) model or Brownian dynamics models [117]. However, they are computationally very demanding. As shown in [118], improvements on the classical PNP algorithm may significantly enhance the computational process although the accuracy is still limited. In case of electro-nanopores, the situation is even more complicated, since the pore structure is voltage dependent and random. From the size estimation and comparison to the energy models predicting the lowest possible dimensions of a hydrophilic electropore, it seems that during electroporation under constant current only one nanopore is created. The reason lies in the negative feedback, since the appearance of the first hydrophilic electropore immediately reduces the transmembrane potential, especially in the area adjacent to the electropore. The potential drops below the value of the breaking potential UB and hampers consequent electroporation. The mean electropore size can be approximately controlled by the current value. The spectral analysis of the fluctuations shows a strong dominance of very low frequencies, yet no characteristic frequency component can be observed. Although current-controlled conditions permit maintaining an open electropore for a long time, it was suspected that the strong feedback affects the results and may 300

Potential [mV]

250

200

150

100

50

0

50

100

150

Time [s]

Figure 8 Chronopotentiometric (ChP) curve under constant-current conditions—electroporation at 280 mV.

21

Random Processes in the Appearance and Dynamics of an Electropore

completely distort the electropore dynamics. It was not even certain if the fluctuations observed from ChP do not originate from the feedback only, and if they are related to the inherent dynamics of the electropore. Moreover, the fluctuating potential across the membrane increases the capacitance current resulting from the charging and discharging of the membrane. An exact value of this current is difficult to estimate and depends on the membrane capacitance, which varies during the experiment. Therefore, the value of any qualitative analysis of the ChP data was debatable, although such an analysis would be very desirable, as ChP provides an extremely good method for keeping the electropores stable for very long time. The observation that only the very moment of creating a hydrophilic electropore is critical for the membrane stability [21] allowed introducing a new experimental method—CACC electroporation, combining advantages of the current-clamp and voltage-clamp approaches. Until now, it has been the only experimental method creating a long-lived electropore with natural dynamics. 4.1.2. CACC electroporation The CACC electroporation [21] includes two stages of the experiment (Fig. 9). At the first stage, a hydrophilic pore is created and stabilized. This stage is critical for the membrane stability. As a result the experimenter obtains an electropore with a fairly stable conductance, close to the conductance achieved at the current-clamp stage of the experiment. The electropore usually survives for a few minutes, which is sufficient to test its dynamical properties. The CACC electroporation eliminates the feedback, preserving the long-lasting hydrophilic nanopore whose natural properties are not blurred. The current recorded at the voltage-clamp stage of CACC shows stochastic fluctuations (Fig. 10). The fluctuations are sensitive to the experimental conditions as confirmed by their rigorous stochastic analysis and their periodograms show a power-law dependence, which indicates a stochastic process of self-scaling properties [21].

4.2. Analytical Methods for Stochastic Processes with a Memory Both experimental methods on a long-lived electropore, ChP and CACC, provide time series of the electropore fluctuations. It can be easily observed that these fluctuations are stochastic and sensitive to the physicochemical conditions of the experiment. Therefore, their rigorous quantitative analysis may offer a new insight into the phenomenon of electroporation, the electropore dynamics and its sensitivity to the membrane composition and environment. Quantitative analysis should be based on appropriately chosen mathematical methods used for analyzing stochastic processes of

Clamping current

Figure 9

Stabilization of the electropore edge under constant current

Clamping voltage

Chronoamperometry After Current Clamp (CACC) electroporation.

22

M. Kotulska and K. Kubica

Current [nA]

1 0.5 0 −0.5 0

20

40

60

80

100

120

140

Time [s]

Figure 10 Current fluctuations in the second stage of Chronoamperometry After Current Clamp (CACC) (voltage clamp). Adopted from [21] with permission from Biophysical Society.

this type. The preliminary analysis of the time series from ChP and CACC showed that the stochastic process is nonstationary with long-range correlations. There are specific analytical methods for stochastic processes of this type. This section presents a review of the methods used for the analysis of the electropore fluctuations. Standard spectral and stochastic analysis usually requires the stationarity of the signal (Wiener–Khinchin theorem), which in general means that statistical properties of a stochastic process are constant in time. There are two types of stationarity. Strict stationarity means that the joint probability of distribution function of time series {X1, X2, . . . , Xn} is identical with the joint probability distribution of the time series {X1þh, X2þh, . . . , Xnþh}; weak stationarity means that only joint moments up to order 2 of the above probability distributions exist and are identical. In most physical systems, only the weak stationarity is tested. An important property of a weakly stationary time series is time-invariant value of the expected value and variance. The autocovariance function does not depend on the time but the time lag only. Typical methods to test the stationarity of the stochastic process, which were used for the study on the electropore fluctuations, detect if there is a trend in the mean and variance of the process. This idea was extended into a quantile line method [119], which can detect nonstationarity and indicates its source. A quantile of order e, 0  e  1, is such a value ke(t) for which the probability of assuming, by the time series at time t, a value less than ke(t), equals e:

PfXt  ke ðtÞg ¼ e

ð17Þ

For each quantile order, a function of time (quantile line) is determined. For stationary processes, quantile lines of various quantile orders are parallel to the time axis. Otherwise, the process is nonstationary. Lines parallel to each other but not to the time axis indicate a process with a constant variance and a variable mean (if the process has a finite mean and variance). Lines that are not parallel to each other reveal a variable variance (or a scale parameter for processes without finite variance). Analyzing nonstationary data, such as the electropore fluctuations, can be rather challenging. The most popular methods to deal with a nonstationary process are the classical decomposition method, segmentation and analysis of the stationary segments

Random Processes in the Appearance and Dynamics of an Electropore

23

only, and the wavelet analysis. The classical decomposition method, employed for the electropore fluctuations, is applicable when a nonstationary process Xt can be decomposed into a stationary random noise Yt and the remaining part, which may include a slowly varying trend mt, and a seasonal component st:

Xt ¼ m t þ s t þ Y t

ð18Þ

The differencing transform often eliminates nonstationarity by extracting the stationary component. It is especially useful when the trend is also stochastic. The difference operator r is defined by the following relation [120]:

rj Xt ¼ Xt  Xtj

ð19Þ

where Xt is the original process, for example, obtained from an experiment, and Xtj is the same process although shifted by the step j. Generally, a differenced process of increments rjXt is deprived of the low frequency component (trend) and shows the dynamics of the fluctuations. However, in some situations, the analysis of the increments rjXt can also provide information on the original process Xt, as in the case of linear processes with a short-range memory and self-similar processes with a long-range memory. Processes with a slowly decaying autocorrelation function, such as the electropore fluctuations, are called processes with a memory. It means that there is a certain relationship between past, present, and future values of the process. Processes with a memory can be divided into processes with short-range correlations (short-range memory) and long-range correlations (long-range memory). Processes with a longrange memory also include processes whose memory is called ‘‘short,’’ which means that each increase in the process value is more likely to be followed by a decrease, and a decrease by a subsequent increase. Information on the process memory is essential for statistical analysis, especially if the methods assuming data independence are to be employed. On the other hand, weak dependence between observations is negligible for practical purposes. The decay of the autocorrelation function of the electropore fluctuation may help to determine if the process has a short-range memory, which is more difficult in case of an antipersistent process, such as the electropore fluctuations. Stochastic processes with a long-range memory may have scaling properties. Similarly as the process of the electropore fluctuates, a power spectrum density (PSD) function S( f ) of such a scaling process has a power-law form [121]:

Sð f Þ 

1 ; B 6¼ 0; fB

ð20Þ

Scaling properties mean that if the frequency of a stochastic process is multiplied by a certain factor, the PSD function diminishes by the same fraction, regardless of the frequency value. The value of the exponent B in the PSD function gives information on the process stationarity and memory. If B > 1 such a process is nonstationary and the process of its increments, obtained by the differencing transform (1.19), is stationary with a power-law S( f ) function and the exponent Bdiff,

24

M. Kotulska and K. Kubica

Bdiff < 1. The relation between the exponent B of the original process and the exponent Bdiff of the differenced process is given by [122]:

Bdiff ¼ B  2

ð21Þ

A nonstationary process with scaling properties and a long-range memory is called a self-similar (fractal) stochastic process [121, 123–128]. A self-similar stochastic process is scale invariant, that is, when rescaled by a certain factor, it is statistically indistinguishable from the original process: d

XðatÞ ¼ aH XðtÞ; a > 0;

ð22Þ

d

H is called a self-similarity index, H > 0, and ¼ denotes equality of all finitedimensional distributions of the process. A self-similar process has several mathematically interesting properties and can be quantitatively characterized on the basis of the value of its self-similarity index H. If the amplitude of a self-similar process is rescaled by aH, the rescaled process X(at) looks like the original process X(t). If a self-similar process has a finite variance, the variance increases with the time, Var(X(t))  t 2H. The memory of a process can be quantitatively expressed by a number d proportional to the self-similarity index, and depending on the PDF of the process (discussed below). The best known self-similar process is the fractional Brownian motion (f Bm), which is a Gaussian self-similar process with stationary increments [127]. It can be obtained as a sum of fractional Gaussian noises (fGn), which are Gaussian processes with powerlaw spectrum density function (21), where 0 < B < 1. The autocovariance of fBm yields: 1 2

Rðt1 ; t2 Þ ¼ fjt1 j2H þ jt2 j2H þ jt1  t2 j2H g

VarðXð1ÞÞ

ð23Þ

The well-known Brownian motion is a special case of the fBm with H ¼ 0.5. Introduction of a more general class of stable self-similar processes extends the idea of fractional motions to non-Gaussian stochastic processes. In the fractional Le´vy stable motion (FLSM), which is another class of stable self-similar processes complementary to fBm, the distribution is Le´vy-stable, as discussed below. For the Gaussian distribution, FLSM becomes fBm. An FLSM process fZaH ðtÞgt2R is nonstationary and it can be represented as [128]: 1 ð

ZaH ðtÞ

½ðt  uÞH1=a  ðuÞH1=a dZa ðuÞ; þ þ

¼

ð24Þ

1

where Za(u) is a symmetric Le´vy a-stable motion, and a is the stability index of a-stable distribution. The increment process of FLSM is stationary and it is called a fractional stable noise (FSN). Stationarity of the increments, expected in FLSM, yields: d

XðatÞ  XðasÞ ¼ aH ðXðtÞ  XðsÞÞ;

a>0

ð25Þ

Random Processes in the Appearance and Dynamics of an Electropore

25

A class of a-stable distributions was discovered by Paul Le´vy as a result of his study on the sums of random processes. The a-stable distributions have scaling properties—a sum of independent and identically distributed random variables maintains the same shape of the distribution. This property is common with the Gaussian distribution, which also belongs to the class of stable distributions. Only a few a-stable distributions have direct formulas for their probability density function. Usually only the characteristic function is given. The distinctive properties of a-stable distributions are their long tails, infinite variance and, in some cases, infinite mean value. Nevertheless, not all distributions with long tails are a-stable, and so the classification needs appropriate statistical methods, such as maximum likelihood estimation (MLE) or less accurate methods based on the sample characteristic function or quantiles [129, 130]. An a-stable distribution is characterized by four parameters: a is the stability index, a 2 (0,2), a < 2 if the distribution has no finite variance, a < 1 if there is no finite expected value (a ¼ 2 is assumed to denote a Gaussian distribution); b represents skewness, b 2 [1,1]; g is a scaling parameter, g 2 (0,1); d is a location parameter, d e (1,1). The self-similarity index H characterizes fractal properties of self-scaling process and autocovariance or codifference function (for processes with infinite variance) [128]. The index H gives quantitative information on the process. Its value can be estimated by several methods. The most popular methods, which were also applied to the analysis of the electropore fluctuations, include rescaled range analysis (the Hurst method) [131], Detrended Fluctuation Analysis (DFA) [132], and an exponent-based (EB) method [122]. The estimation by the rescaled range analysis (R/S), which compares the correlations measured at different time scales, is a heuristic method for analyzing processes with a long memory. The self-similarity index H for a nonstationary process with stationary increments is calculated from the process of increments (19). The series x of the length N is divided into smaller nonoverlapping fragments of the length n, n > 2 and two statistics, S(n) and R(n), are calculated in the following way [131]: n 1X hxin ¼ xi n i¼1 " #1=2 n X 1 SðnÞ ¼ ðxi  hxin Þ2 n i¼1 i X Xði; nÞ ¼ xu  hxin u¼1

RðnÞ ¼ max Xði; nÞ  min Xði; nÞ 1in 1in ! R RðnÞ / ðnÞH ¼ S SðnÞ n ! R log / H logðnÞ S n

ð26Þ

26

M. Kotulska and K. Kubica

The self-similarity index H can be calculated as a slope of the statistics (R/S )n plotted on the double-logarithmic scale (1.26). Similarly to the rescaled range analysis, the detrended fluctuation analysis requires dividing the series of a length N into fragments of a length n. A local trend is determined for each of the fragments and then removed; the sequence of detrended fragments yn(i ) is obtained. The mean variance of the detrended process yn(i), where the mean is calculated for all the fragments, provides the statistics Fd(n) [132]: n 1 XX ¼ y2n ðiÞ N n¼1 i¼1 N =n

Fd2 ðnÞ

ð27Þ

The self-similarity index H can be estimated from the power-law dependence:

Fd2 ðnÞ / n2H

ð28Þ

The simplest and very commonly used technique is based on the value of the exponent B obtained from the periodogram of the original self-similar process. The self-similarity index by the EB method yields [122]:



B1 2

ð29Þ

All presented statistics may give different values of the self-similarity index, depending on the regularity of the data and its statistical properties. Therefore, the optimal technique should be selected individually. The range of dependence in the time series is expressed by the memory index d of the process, which is given by the relation of self-similarity index H to the stability index a [125]:

d¼H

1 a

ð30Þ

The relation (1.30) also applies to Gaussian processes, with a ¼ 2. If d > 0, a process has a long memory (evaluated by the codifference function), d ¼ 0 indicates a random process without a memory, and d < 0 shows a process with a short memory (antipersistent) [128]. Visually, the shorter the memory of the process, the rougher is the time series. Classification and quantitative characterization of the random fluctuations observed in a single electropore can base on the presented statistical methods.

4.3. Stochastic Characteristics of the Electropore Fluctuations 4.3.1. Fluctuations in chronopotentiometry The analysis of the stationarity of the fluctuation process obtained from ChP showed that the process is nonstationary with a varying variance, and the classical decomposition method significantly improves the process stationarity [106]. Voltage

Random Processes in the Appearance and Dynamics of an Electropore

27

fluctuations of the electroporated membrane proved appropriate for the classical decomposition method, which efficiently removes their nonstationary component. The study of the process memory, testing if there is a possibility of representing the differenced fluctuations by a linear Autoregressive Moving Average (ARMA) process, confirms that the process has a memory. Nevertheless, it was observed that the electropore voltage fluctuations have a long-range memory and they cannot be modeled by means of a linear Autoregressive Integrated Moving Average (AIRMA) process of a reasonably low order. The spectrum of the voltage and conductance fluctuations from ChP always has a power-law form, regardless of the ionic content or the solute concentration [106, 133]. Moreover, the stochastic fluctuations of the electropore conductance are sensitive to the environment. This result means that the fluctuations with the feedback have long-range correlations, whose extent depends on the physicochemical conditions. The sensitivity of the stochastic process of the voltage increments to the experimental conditions includes the value of the exponent B. At higher currents, when the electropore increases its size, the exponent B of the voltage fluctuations assumes a lower value. The observed increase of B is related to a higher rate of the low frequency component. By means of the statistical tests, a hypothesis on the Gaussian distribution of the differenced voltage fluctuations was rejected. The extreme data form left and right tails of the PDF, whose shape changes with the current value. The right tail (DU > 0) is related to the process of charging the membrane, by the capacitance current ic, which occurs when the pore is closing. However, it is much more interesting to analyze the time course of the conductance than the voltage fluctuations. The stochastic analysis of the approximated membrane conductance gave more insight into the electropore dynamics, and the contribution from the charging/discharging process of the membrane capacitance was significantly reduced. The conductance analysis [106] showed scaling properties of the conductance fluctuations with a PSD of 1/f B. The exponent B depends on the membrane environment and assumes values closer to 1 than that of the voltage fluctuations. A more accurate evaluation of the exponent B shows that B 1 [21], so the stochastic process is a self-similar motion. The exponent B is sensitive to the chemical environment and to the electropore size in a repetitive manner, giving quantitative information on the electropore state. The influence of a solute on the electropore dynamics depends on the ionic concentration. At higher ionic concentration, the electropore is smaller and the membrane is more durable. Although the mechanism of this effect is not exactly known, it may be related to the structural changes of the lipid membrane and the decreased mobility of lipid molecules, as shown for monovalent salts by fluorescence correlation spectroscopy confirmed by MD simulations [108] and MC simulations [112]. Similar to the voltage fluctuations, the PDF is not Gaussian—it has tails. Statistical analysis [21] shows that the probability distribution of the fluctuations can be approximated by a-stable distribution, hence the FLSM process can be used for modeling the conductance fluctuations. In all cases, the memory evaluation shows an antipersistent process with long correlations. With regard to the feedback, inherent to the experiments under current-clamp conditions, this result could be expected. It is because the negative

28

M. Kotulska and K. Kubica

feedback opposes each change of the conductance, adding to the effect of the line tension as the electropore is expanding or cooperating with the surface tension when the electropore is shrinking. Importantly, sensitivity of the memory index to the experimental conditions show that the feedback alone does not determine the process memory, and that the fluctuations are also affected by the membrane state. The memory value depends on the electropore environment and diameter. For example, fluctuations of small electropores or fluctuations in the environment hampering the membrane dynamics are more antipersistent. However, the chronopotentiometry cannot answer the question what is the contribution from the feedback and whether the feedback is dominant over all other processes. 4.3.2. Natural random fluctuations of an electropore, based on CACC The spectral analysis showed that self-scaling properties of the conductance fluctuations observed in ChP also appear in natural fluctuations of a single electropore obtained from CACC electroporation (Fig. 11) [21]. The power-law dependence, 1/f B, is maintained in CACC, and the power spectral density is sensitive to the electrolyte composition. Qualitatively, in terms of B and the power spectrum amplitude, the results from CACC are correlated to the results obtained from ChP. Moreover, the values of B are more sensitive to the environment and more distinctive when obtained from CACC electroporation. This correlation proves that the classical ChP is able to reflect natural dynamics of the electropore, although less distinctively. Regarding the fact that an electropore studied by ChP lives longer, the correspondence between natural fluctuations from CACC and those from ChP supported by the feedback, is a very desirable feature. The probability density function of the conductance transitions also shows longtailed a-stable probability density function (Fig. 12) [21]. Therefore, the process of

Sf [nS2/Hz]

101

100

10−1

10−2 100

101

Frequency [Hz]

Figure 11 Power-law periodogram of the conductance fluctuations in Chronoamperometry After Current Clamp (CACC) electroporation.

29

Random Processes in the Appearance and Dynamics of an Electropore

4

PDF

3

2

1

0 −1.5

−1

−0.5 0 0.5 Differenced conductance [nS]

1

1.5

Figure 12 Statistical methods show that the probability density function of the conductance dynamics can be approximated as a long-tailed a-stable distribution. Adopted from [21] with permission from Biophysical Society.

the natural fluctuations of an electropore can be classified as a FLSM. The long tail of the PDF function disappears as the electropore increases its size. It indicates that the conductance fluctuations tend from FLSM to the fBm when the electropore grows. The transition from long tails to Gaussian-like distribution with finite variance may reflect a different shape of the electropore edge. There is a possibility that lipid headgroups translocated into the membrane interior manage to form a more regular edge of the electropore when the pore has a greater size. An electropore of such a regular shape may be less susceptible to abrupt significant changes of its edge and effective surface by incorporation of adjoining water wires and other defects. Such abrupt changes are responsible for the long tails in the probability density function of a small electropore. The quantitative memory study shows that the process is antipersistent (short memory) and the memory index is sensitive to the membrane environment and the electropore diameter. This result quantitatively shows the competition between opening and resealing processes, resulting from the surface and line tensions, and their sensitivity to the membrane environment. Importantly, all the characteristic quantities of the stochastic process from CACC are correlated with the ChP results, although the memory of the ChP fluctuations is shorter than the memory of CACC fluctuations. This result is very well explicable. Fluctuations observed in chronopotentiometry arise from the overlap of the feedback with the processes related to the line and surface tensions. The negative feedback opposes each change of the conductance, adding to the effect of the line tension when the electropore is expanding, or cooperating with the surface tension when the electropore is shrinking. Therefore, the process becomes more antipersistent than in CACC. It confirms that for very unstable membranes, ChP may be a good alternative to CACC.

30

M. Kotulska and K. Kubica

Summarizing the most important features of an electropore:

An electropore has naturally fluctuating dynamics. These fluctuations can be quantitatively evaluated by their stochastic characteristics, giving information about the membrane dynamics and its sensitivity to the membrane environment. Memory of the process is short (antipersistent), which may show competitive processes, for example, line and surface tensions of the pore. The feedback (present under current-clamp conditions in ChP) affects fluctuations of an electropore, shortening the process memory. However, the memory indexes obtained from ChP and CACC are correlated, which proves that the feedback does not completely cover the natural dynamics of the nanopore. Dependence between the nanopore diameter and its stochastic characteristics may be related to a different edge structure for small and large nanopores.

4.4. Modeling Stochastic Processes of a Power-Law Spectrum Fluctuations with power spectra of 1/f B, 0 < B < 2, are widely found in nature in a large variety of physical, biological, geophysical, traffic, financial, and other complex systems. They are considered ubiquitous in systems that are distributed and loosely coupled. The phenomenon is well known for macro systems, but recently it has also been found in biological protein channels [134–136], synthetic membranes with a fluctuating nanopore [135], and electro-nanopores in lipid bilayers [21, 106]. The experiments on electro-nanopores in lipid membranes and nanopores in synthetic membranes showed that complexity of highly regulated biological channels is not a necessary condition. Such a widespread occurrence of signals exhibiting this behavior may suggest that a generic mathematical explanation of 1/f spectrum might exist. However, there is no generally accepted rigorous theory explaining the mechanism of this phenomenon. With regard to conducting nanochannels, a hypothesis was proposed that random transitions between various conductivity states of the channel might underlie such a spectrum [134, 135]. There is also a question to which extent the noise characteristics depend on transport properties of the channel or complexity of its structure. It has been observed that random fluctuations typically produce two types of spectrum: the Lorentzian spectrum S( f ) / f2, typical of stochastic processes with no memory, and a power-law spectral density S( f ) / fB, which indicates a fractional noise (0 < B  1) or a self-similar motion (B > 1, B 6¼ 2). A process with the Lorentzian spectrum appears in a dichotomous process without memory. It is applicable when a pore assumes a few discrete states between which energy barriers are discrete and constant, and transition probabilities between these states are independent of the channel history. In most nanopores, however, the potential fluctuations have a power-law spectrum density. This type of response is more difficult to explain in terms of a Markovian process [137]. There has been no generally accepted model, although there are several theories from which a sum of Lorentzians is the most common [136, 138–140]. Another model of a stochastic process with the power-law spectral density is SelfOrganized Criticality (SOC). The SOC-based models apply to dynamical systems

Random Processes in the Appearance and Dynamics of an Electropore

31

with minimal stability and spatial scaling. In such a system, each perturbation activates a cascade of energy dissipation on all length scales and propagates through the self-similar clusters. Through the domino-resembling effect, the minimally stable states of the clusters are disturbed, which leads to a power-law in temporal fluctuations [141]. In these systems, self-similarity occurs both in time and space. A good example of such a phenomenon is turbulence. Recently, the SOC theory was also proposed with regard to the electroporation phenomenon [142]. Both presented models have severe drawbacks. The approach based on the summation of Lorentzians does not apply to non-Markovian processes with a memory. Secondly, no physical interpretation for the summation with an exponential distribution of relaxation times has been found. On the other hand, SOC-based models require spatial self-similarity of the system and dynamical minimal stability. In the case of nanopores, there is no experimental evidence that these assumptions are met, although their appearance cannot be completely excluded. Hence, up to date the power-law spectrum of the fluctuating nanopore still has no convincing explanation that could be validated experimentally.

5. Summary Disruption of the membrane continuity, following application of a strong electric field, results in flow of ions and various molecules into and out of the cell, changing the biochemical contents of the cytoplasm. Under prolonged application of a high electric field, the electropores grow beyond the safe dimensions and the plasma membrane breaks up, resulting in the cell death. It was discovered, however, that controlled electroporation provides a useful method for delivering of biologically active molecules, such as nucleic acids and drugs, into the cells. Currently, the electroporation is commonly used for genetic modifications of cells as the cleanest available method. ECT—electroporation supported treatment of cancerous cells— extensively increases antitumor effects. In some cases, abandoning of the surgery is even possible. However, the development of the methods for controlling the electroporation process remains the main problem. The progress is still hampered by poor understanding of the phenomenon. The pore appearance is a random process and it has been observed not only under electric field but also at high temperature and under mechanical stress, which may imply a similar mechanism. Electroporation experiments and MD computer simulations of this phenomenon show appearance of the pre-pores first, gradually penetrated by solvent molecules, which expand into regular hydrophilic electropores. The pore appearance may be related to the increased probability of local conformational transitions of the membrane molecules, dependent on the physicochemical conditions. This hypothesis has been supported by MC simulations. Experiments show randomness in the dynamics of a single long-lived electropore. The stochastic analysis of natural electropore fluctuations from CACC experiments reveal characteristic features of this randomness, also dependent on the physicochemical

32

M. Kotulska and K. Kubica

conditions of the membrane. The fluctuations are a self-similar stochastic process with a short memory (antipersistent), which can be modeled as a FLSM for small nanoelectropores, and tends to a fBm when the electropore increases its diameter. Till date, there is no generally accepted model explaining self-similar properties of such a non-Markovian stochastic process with long correlations, although a concept based on a sum of Markovian processes with appropriate distribution of their relaxation time or the theory of a SOC has been considered. Statistical features of the electroporation phenomenon are closely related to the membrane structure and its dynamics. The environmental conditions affecting lateral diffusion and flip-flop processes can significantly influence the pore appearance, stability, and dynamics. For example, lowering the rate of the flip-flop process by the environmental changes would hamper the pore appearance. On the other hand, increasing the rate of lateral diffusion by physicochemical conditions would increase the electropore dynamics, which may result in its instability and will affect its stochastic characteristics. The membrane composition and environmental changes affect random processes in the lipids of intact membrane and an electropore in a correlated manner. Thorough understanding of the electroporation phenomenon and electropore characteristics is fundamental for applications in medicine and technology. The research aims to obtain a better control over the electroporation process through changes in the membrane composition or physicochemical conditions of the environment. Randomness is an inherent feature of the electroporation process, related to the nature of lipid molecules and their diffusion, hence it cannot be disregarded.

REFERENCES [1] D.C. Chang, B.M. Chassy, J.A. Saunders, A.E. Sowers (Eds.), Guide to Electroporation and Electrofusion, (1992) Academic Press, San Diego. [2] E. Neumann, S. Kakorin, K. Toensing, Fundamentals of electroporative delivery of drugs and genes, Bioelectrochem. Bioenerg. 48 (1999) 3–16. [3] J.C. Weaver, Yu.A. Chizmadzhev, Theory of electroporation: A review, Bioelectrochem. Bioenerg. 41 (1996) 135–160. [4] M. Kotulska, Electro-Nanopores in the Lipid Membrane and Their Application in Medicine, IBIB, Warsaw, (2007) (habilitation thesis). [5] R. Sta¨mpfli, Reversible electrical breakdown of the excitable membrane of a Ranvier node, Ann. Acad. Bras. Cienc. 30 (1958) 57–63. [6] E. Neumann, K. Rosenheck, Permeability changes induced by electric impulses in vesicular membranes, J. Membr. Biol. 10 (1972) 279–290. [7] R. Benz, F. Beckers, U. Zimmermann, Reversible electrical breakdown of lipid bilayer membranes: A charge-pulse relaxation study, J. Membr. Biol. 4 (1979) 181–204. [8] E.M. el-Mashak, T.Y. Tsong, Ion selectivity of temperature-induced and electric field induced pores in dipalmitoylphosphatidylcholine vesicles, Biochemistry 24 (1985) 2884–2888. [9] L. Movileanu, D. Popescu, S. Ion, A.I. Popescu, Transbilayer pores induced by thickness fluctuations, Bull. Math. Biol. 68 (2006) 1231–1255. [10] D.V. Zhelev, D. Needham, Tension-stabilized pores in giant vesicles: Determination of pore size and pore line tension, Biochim. Biophys. Acta 1147 (1993) 89–104. [11] M. Winterhalter, W. Helfrich, Effect of voltage on pores in membranes, Phys. Rev. A 36 (1987) 5874–5876.

Random Processes in the Appearance and Dynamics of an Electropore

33

[12] E. Neumann, M. Schaefer-Ridder, Y. Wang, P.H. Hofschneider, Gene transfer into mouse lyoma cells by electroporation in high electric fields, EMBO J. 1 (1982) 841–845. [13] L.M. Mir, S. Orlowski, Mechanisms of electrochemotherapy, Adv. Drug Deliv. Rev. 35 (1999) 107–118. [14] S.J. Beebe, P.M. Fox, L.J. Rec, E. Lauren, K. Willis, K.H. Schoenbach, Nanosecond, high intensity pulsed electric fields induce apoptosis in human cells, FASEB 17 (2003) 1493–1495. [15] B.J. Roth, W. Krassowska, The induction of reentry in cardiac tissue. The missing link: how electric fields alter transmembrane potential, Chaos 8 (1998) 204–220. [16] W. Frey, J.A. White, R.O. Price, P.F. Blackmore, R.P. Joshi, R. Nuccitelli, S.J. Beebe, K.H. Schoenbach, J.F. Kolb, Plasma membrane voltage changes during nanosecond pulsed electric field exposure, Biophys. J. 90 (2006) 3608–3615. [17] T.P. Vernier, Y. Sun, M.A. Gundersen, Nanoelectropulse driven membrane perturbation and small molecule permeabilization, BMC Cell Biol. 7 (2006) 37–53. [18] T. Kotnik, G. Pucihar, M. Rebersek, D. Miklavcic, L.M. Mir, Role of pulse shape in cell membrane electropermeabilization, Biochim. Biophys. Acta 1614 (2003) 193–200. [19] S. Kalinowski, G. Ibron, K. Bryl, Z. Figaszewski, Chronopotentiometric studies of electroporation of bilayer lipid membranes, Biochim. Biophys. Acta 1369 (1998) 204–212. [20] M. Robello, A. Gliozzi, Conductance transition induced by an electric field in lipid bilayers, Biochim. Biophys. Acta 982 (1989) 173–176. [21] M. Kotulska, Natural fluctuations of an electropore show fractional Le´vy stable motion, Biophys. J. 92 (2007) 2412–2421. [22] K.M. Jain, R.C. Wagner, Introduction to Biological Membranes, John Wiley and Sons, New York (1980) pp. 1–382. [23] H. Hauser, Conformation of phospholipids crystal structure of a lyso-phosphatidylcholine analogue, J. Mol. Biol. 137 (1980) 249–264. [24] H. Hauser, I. Pascher, R.H. Pearson, S. Sundell, Preferred conformation and molecular packing of phosphatidylethanolamine and phosphatidylcholine, Biochim. Biophys. Acta 650 (1981) 21–51. [25] D.P. Tieleman, S.J. Marrink, H.J.C. Berendsen, A computer perspective of membranes: Molecular dynamics studies of lipid bilayer systems, Biochim. Biophys. Acta 1331 (1997) 235–270. [26] M. Edidin, Rotational and lateral diffusion of membrane proteins and lipids: Phenomena and function, in: R.D. Klausner, Ch. Kempf, J. van Renswoude (Eds.), Vol. 29. (1987) Academic Press. [27] D. Papahadjopoulos, H.K. Kimelberg, Phospholipid vesicles (liposomes) as models for biological membranes: Their properties and interactions with cholesterol and proteins, in: S.G. Davson (Ed.), Pergamon Press Oxford (1973), pp. 141–232. [28] T. Kunitake, Syntheses aggregate morphologies and applications of membrane-forming amphiphiles, J. Am. Chem. Soc. 99 (1977) 3860–3871. [29] S.J. Singer, G.L. Nicolson, The fluid mosaic model of the structure of cell membranes, Science 175 (1972) 720–731. [30] J. Seelig, A. Seelig, Lipid conformation in model membranes and biological membranes, Q. Rev. Biophys. 13 (1980) 19–61. [31] G. Lagaly, Kink-block and gauche-block structures of bimolecular films, Angew. Chem. Int. Ed. Engl. 15 (1976) 576–586. [32] J. Seelig, W. Niederberg, Two pictures of a lipid bilayer. A comparison between deuterium label and spin-label experiments, Biochemistry 13 (1974) 1585–1588. [33] J.F. Nagle, Theory of biomembrane phase transitions, J. Chem. Phys. 58(1) (1973) 252–264. [34] H.L. Scott, A theoretical model for lipid monolayer transitions, Biochim. Biophys. Acta (BBA) 406 (1975) 329–346. [35] J.A. McCammon, J.M. Deutch, Semiempirical models for biomembrane phase transitions and phase separations, J. Am. Chem. Sci. 97 (1975) 6675–6681. [36] W. Pechhold, S. Blasenberg, Molekulbewegung in polymeren, Kolloid Z. Z. Polym. 241 (1970) 955–976. [37] S. Blasenbrey, W. Pechhold, Theorie der phasenumwandlung in polymeren, Ber. Bunsenges Phys. Chem. 74 (1970) 784–796.

34

M. Kotulska and K. Kubica

[38] M. Baumann, Early stage shape change of human erythrocytes after application of electric field pulses, Mol. Memb. Biol. 18 (2001) 153–160. [39] P.B. Hitchcock, R. Mason, K.M. Thomas, G.G. Shipley, Structural chemistry of 1,2 dilauroyl-dlphosphatidylethanolamine: Molecular conformation and intermolecular packing of phospholipids, Proc. Natl. Acad. Sci. USA 71(8) (1974) 3036–3040. [40] T.W. Healy, J. Ralston, Specific cation effects under monolayers of octadecanol, Nature 220 (1968) 1026–1027. [41] J. Kuczera, T. Janas, S. Przestalski, S. Witek, M. Owicimska, Influence of some amphiphilic derivatives of glycine esters on sulphate ion transport across lecithin liposome membrane, Studia Biophys. 105 (1985) 167–176. [42] E. Sackman, A. Fischer, W. Frey, Polymorphism of monolayers of monomeric and macromolecular lipids: On the defect structure of crystalline phases and the possibility of hexatic order formation in physics of amphiphilic layers, in: J. Meunier, D. Langevin, V. Boccara (Eds.), Physics of amphiphilic layers, Springer, Berlin (1987). [43] M. Langner, H. Pruchnik, K. Kubica, The effect of the lipid bilayer state on fluorescence intensity of fluorescein-PE in a saturated lipid bilayer, Z. Naturforsch. 55 (2000) 418–424. [44] K. Kubica, Monte Carlo simulation towards ripple phase modeling, Comput. Chem. 25 (2001) 245–250. [45] K. Kubica, Computer simulation studies on significance of lipid polar head orientation, Comput. Chem. 26 (2002) 351–356. [46] M. Langner, K. Kubica, The electrostatics of lipid surfaces, Chem. Phys. Lipids 101 (1999) 3–35. [47] G.M. Lee, A. Ishihara, K.A. Jacobson, Direct observation of Brownian motion of lipids in a membrane, Proc. Natl. Acad. Sci. USA 88(14) (1991) 6274–6278. [48] I.P. Sugar, R.L. Biltonen, Lateral diffusion of molecules in two-component lipid bilayer: A Monte Carlo simulation study, J. Phys. Chem. B 109(15) (2005) 7373–7386. [49] R.D. Kornberg, A. McConnell, Inside-outside transitions of phospholipids in vesicle membranes, Biochemistry 10 (1971) 1111–1120. [50] M. Roseman, B.J. Litman, T.E. Thompson, Transbilayer exchange of phosphatidylethanolamine in single-walled bilayer vesicles, Biochemistry 14 (1975) 4826–4830. [51] J.A. Hamilton, Fatty acid transport: Difficult or easy? J. Lipid Res. 39 (1998) 467–481. [52] F. Kamp, W. Guo, R. Souto, P.F. Pilch, B.E. Corkey, J.A. Hamilton, Rapid flip-flop of oleic acid across the plasma membrane of adipocytes, J. Biol. Chem. 278 (2003) 7988–7995. [53] V.T. Armstrong, M.R. Brzustowicz, S.R. Wassall, L.J. Jenski, W. Stillwell, Rapid flip-flop in polyunsaturated (docosahexaenoate) phospholipid membranes, Arch. Biochem. Biophys. 414 (2003) 74–82. [54] A.N. Dickey, R. Faller, How alcohol chain-length and concentration modulate hydrogen bound formation in a lipid bilayer, Biophys. J. 92 (2007) 2366–2376. [55] F. Kamp, J.A. Hamilton, How fatty acids of different chain length enter and leave cells by free diffusion, Prostaglandins Leukot. Essent. Fatty Acids 75 (2006) 149–159. [56] R. Leventis, J.R. Silvius, Use of cyclodextrins to monitor transbilayer movement and differential lipid affinities of cholesterol, Biophys. J. 81 (2001) 2257–2267. [57] J.P. Kampf, D. Cupp, A.M. Kleinfeld, Different mechanisms of free acid flip-flop and dissociation revealed by temperature and molecular species dependence of transport across lipid vesicles, J. Biol. Chem. 281 (2006) 21566–21574. [58] M.J. Moreno, L.M.B.B. Estronca, W.L.C. Vaz, Translocation of phospholipids and dithionite permeability in liquid-ordered and liquid-disordered membranes, Biophys. J. 91 (2006) 873–881. [59] K. John, S. Schreiber, J. Kubelt, A. Herrmann, P. Muller, Transbilayer movement of phospholipids at the main phase transition of lipid membranes: Implications for rapid flip-flop in biological membranes, Biophys. J. 83(6) (2002) 3315–3323. [60] L. Biavas, M. Winterhalter, P. Meleard, P. Bothorel, Elasticity of bilayer containing PEG lipids, Europhys. Lett. 41 (1998) 261–266. [61] Y. Ibuki, A. Suzuki, R. Goto, UVA irradiation induces energy-independent phospholipid-flip in mammalian plasma membrane, Photochem. Photobiol. 73 (2001) 513–517. [62] A.A. Yaroslavov, N.S. Melik-Nubarov, F.M. Menger, Polymer-induced flip-flop in biomembranes, Acc. Chem. Res. 39 (2006) 702–710.

Random Processes in the Appearance and Dynamics of an Electropore

35

[63] T. Meshulam, J.R. Simard, J. Wharton, J.A. Hamilton, P.F. Pilch, Role of caveolin-1 and cholesterol in transmembrane fatty acid movement, Biochemistry 45 (2006) 2882–2893. [64] A. Bootsveld, R. Degenhardt, D. Kamp, C.W.M. Haest, On the mechanism of drug-induced acceleration of phospholipid translocation in the human erythrocyte membrane, Mol. Memb. Biol. 21 (2004) 315–322. [65] A. Fahr, P. van Hoogevest, S. May, N. Bergstrand, M.L.S. Leigh, Transfer of lipophilic drugs between liposomal membranes and biological interfaces: Consequences for drug delivery, Eur. J. Pharm. Sci. 26 (2005) 251–265. [66] W. Stremmel, J. Pohl, A. Ring, T. Herrmann, A new concept of cellular uptake and intercellular trafficking of long-chain fatty acids, Lipids 36 (2001) 981–989. [67] W.J. Li, F. Nicol, F.C. Szoka, GALA: A designed synthetic pH-responsive amphipathic peptide with applications in drug and gene delivery, Adv. Drug Deliv. Rev. 56 (2004) 967–985. [68] F. Reif, Statistical Physics. Berkeley Physics Course Vol. 5. (1967) McGraw-Hill Book Company, New York. [69] N. Metropolis, A.W. Rosenbluth, M.N. Rosenbluth, A.H. Teller, E. Teller, Equation of state calculations by fast computing machines, J. Chem. Phys. 21 (1953) 1087–1092. [70] O.G. Mouritsen, Computers simulations of cooperative phenomena in lipid membranes, in: R. Brasseur (Ed.), Molecular Description of Biological Membranes by Computer Aided Conformational Analysis, CRC Press, Boston, 1990. [71] D.A. Pink, T.J. Green, D. Chapman, Raman scattering in bilayers of saturated phosphatidylcholines. Experiment and theory, Biochemistry 19 (1980) 349–356. [72] O.G. Mouritsen, A. Boothroyd, R. Harris, N. Jan, T. Lookman, L. MacDonald, D.A. Pink, M.J. Zuckermann, Computer simulation of the main gel-fluid phase transition of lipid bilayers, J. Chem. Phys. 79 (1983) 2027–2041. [73] K. Kubica, Pink’s model and lipid membranes, Cell. Mol. Biol. Lett. 2 (1997) 257–263. [74] K. Kubica, The effect of amphiphilic counterions on the gel-fluid phase transition of the lipid bilayer, Appl. Math. Comp. 87 (1997) 261–270. [75] J. Ipsen, K. Jorgensen, O. Mouristen, Density fluctuations in saturated phospholipid bilayers increase as the acyl-chain decreases, Biophys. J. 58 (1990) 1099–1107. [76] J. Ipsen, O. Mouristen, M. Zuckermann, Theory of thermal anomalies in the specific heat of lipid bilayers containing cholesterol, Biophys. J. 56 (1989) 661–667. [77] J. Ipsen, O. Mouristen, M. Bloom, Relationships between lipid membrane area, hydrophobic thickness, and acyl-chain orientational order, Biophys. J. 57 (1990) 405–412. [78] K. Jorgensen, J. Ipsen, O. Mouritsen, D. Bennett, M. Zuckermann, The effect of density fluctuations on the partitioning of foreign molecules into lipid bilayers: Application to anaesthetics and insecticides, Biochi. Biophys. Acta 1067 (1991) 241–253. [79] K. Jorgensen, J. Ipsen, O. Mouritsen, D. Bennett, M. Zuckermann, A general model for the interaction of foreign molecules with lipid membranes: Drugs and anaesthetics, Biochim. Biophys. Acta 1062 (1991) 227–238. [80] O. Mouritsen, K. Jorgensen, Dynamic lipid-bilayer heterogeneity: A mesoscopic vehicle for membrane function? BioEssays 14 (1992) 129–136. [81] O. Mouritsen, K. Jorgensen, Dynamical order and disorder in lipid bilayers, Chem. Phys. Lipids 73 (1994) 3–25. [82] M.M. Sperotto, O.G. Mouritsen, Monte Carlo simulation studies of lipid order parameter profiles near integral membrane proteins, Biophys. J. 59 (1991) 261–270. [83] M. Sperotto, O. Mouritsen, Lipid enrichment and selectivity of integral membrane proteins in tow-component lipid bilayers, Eur. Biophys. J. 22 (1993) 323–328. [84] Z. Zhang, M. Sperotto, M. Zuckermann, O. Mouritsen, A microscopic model for lipid/protein bilayers with critical mixing, Biochim. Biophys. Acta 1147 (1993) 154–160. [85] O. Mouritsen, K. Jorgensen, Micro- nano- and meso-scale heterogeneity of lipid bilayers and its influence on macroscopic membrane properties, Mol. Membr. Biol. 12 (1995) 15–20. [86] O. Mouritsen, K. Jorgensen, A new look at lipid-membrane structure in relation to drug research, Pharm. Res. 15 (1998) 1507–1519. [87] K. Jorgensen, O.G. Mouritsen, Phase separation dynamics and lateral organization of tow-component lipid membranes, Biophys. J. 95 (1995) 942–954.

36

M. Kotulska and K. Kubica

[88] K. Kubica, Polar head charge of membrane modifiers and their biological activity: The Monte Carlo simulation studies, Task Quarterly 2 (1998) 601–609. [89] K. Kubica, Computer simulation studies on the significance of lipid polar head charge, Cell. Mol. Biol. Lett. 7 (2002) 971–982. [90] K. Kubica, M. Langner, J. Gabrielska, The dependence of fluorescein-PE fluorescence intensity on lipid bilayer state. Evaluating the interaction between the probe and lipid molecules, Cell. Mol. Biol. Lett. 8 (2003) 943–954. [91] M. Kotulska, K. Kubica, S. Koronkiewicz, S. Kalinowski, Modeling the induction of lipid membrane electropermeabilization, Bioelectrochemistry 70 (2007) 64–70. [92] I.P. Sugar, D. Tang, Parkson Lee-Gau Chong, Monte Carlo simulation of lateral distribution of molecules in a two-component lipid membrane. Effect of long-range repulsive interactions, J. Phys. Chem. 98 (1994) 7201–7210. [93] J. Teissie, M. Golzio, M.P. Rols, Mechanisms of cell membrane electro-permeabilization: A minireview of our present (lack of ?) knowledge, Biochim. Biophys. Acta 1724 (2005) 270–280. [94] D.P. Tieleman, The molecular basis of electroporation, BMC Biochem. 19 (2004) 5–10. [95] H. Leontiadou, A.E. Mark, S.J. Marrink, Molecular dynamics simulations of hydrophilic pores in lipid bilayers, Biophys. J. 86 (2004) 2156–2164. [96] C. Loison, M. Mareschal, F. Schmid, Pores in bilayer membranes of amphiphilic molecules: Coarse-grained molecular dynamics simulations compared with simple mesoscopic models, J. Chem. Phys. 121 (2004) 1890–1900. [97] M. Tarek, Membrane electroporation: A molecular dynamics simulation, Biophys. J. 88 (2005) 4045–4053. [98] I.P. Sugar, A theory of the electric field-induced phase transition of phospholipid bilayers, Biochim. Biophys. Acta 556 (1979) 72–85. [99] J.C. Schillcock, U. Seifert, Thermally induced proliferation of pores in a model fluid membrane, Biophys. J. 74 (1998) 1754–1766. [100] M. Muller, M. Schick, Structure and nucleation of pores in polymeric bilayers: A Monte Carlo simulation, J. Chem. Phys. 105 (1996) 8282–8292. [101] S.J. Marrink, H.J.C. Berendsen, Simulation of water transport through a lipid membrane, J. Phys. Chem. 98 (1994) 4155–4168. [102] R.W. Glaser, S.L. Leikin, L.V. Chernomordik, V.F. Pastushenko, A.I. Sokirko, Reversible electrical breakdown of lipid bilayers: Formation and evolution of pores, Biochim. Biophys. Acta 940 (1988) 275–287. [103] K. Kubica, A pore creation in a triangular network model membrane, Comput. Biol. Chem. (2008), in press. [104] A. Diederich, B. Gu¨nther, M. Winterhalter, Influence of surface charges on the rupture of black lipid membranes, Phys. Rev. E 58 (1998) 4883–4889. [105] S. Koronkiewicz, Badanie procesu elektroporacji i odtwarzania ciaglej struktury dwuwarstwowych membran lipidowych. Ph.D. thesis, University of Warmia and Mazury, Olsztyn. [106] M. Kotulska, S. Koronkiewicz, S. Kalinowski, Self-similar processes and flicker noise from a fluctuating nanopore in a lipid membrane, Phys. Rev. E 69 (2004) 031920–031930. [107] F.J. Carrion, A. De La Maza, J.L. Parra, The influence of ionic strength and lipid bilayer charge on the stability of liposomes, J. Coll. Interf. Sci. 164 (1994) 78–87. [108] R.A. Bo¨ckman, A. Hac, T. Heimburg, H. Grubmu¨ller, Effect of sodium chloride on a lipid bilayer, Biophys. J. 85 (2003) 1–9. [109] S.A. Tatulian, Ionization and ion binding, in: G. Cevc (Ed.), Phospholipids Handbook, Marcel Dekker, New York, (1993) pp. 511–552. [110] A.A. Gurtovenko, Asymmetry of lipid bilayers induced by monovalent salt: Atomistic molecular-dynamics study, J. Chem. Phys. 122 (2005) 244902. [111] J.N. Sachs, H. Nanda, H.I. Petrache, T.B. Woolf, Changes in phosphatidylcholine headgroup tilt and water order induced by monovalent salts: molecular dynamics simulations, Biophys J. 86 (2004) 3772–3782. [112] M. Kotulska, K. Kubica, Structural and energetic model of the mechanisms for reduced selfdiffusion in a lipid bilayer with increasing ionic strength, Phys. Rev. E 72 (2005) 061903–061909.

Random Processes in the Appearance and Dynamics of an Electropore

37

[113] S. Koronkiewicz, S. Kalinowski, K. Bryl, Changes of structural and dynamic properties of model lipid membranes induced by alpha-tocopherol: Implication to the membrane stabilization under external electric field, Biochim. Biophys. Acta 1510 (2001) 300–306. [114] S. Koronkiewicz, S. Kalinowski, K. Bryl, Programmable chrono-potentiometry as a tool for the study of electroporation and resealing of pores in bilayer lipid membranes, Biochim. Biophys. Acta 1561 (2002) 222–229. [115] S. Koronkiewicz, S. Kalinowski, Influence of cholesterol on electroporation of bilayer lipid membranes: Chronopotentiometric studies, Biochim. Biophys. Acta 1661 (2004) 196–203. [116] M. Kotulska, A comparison between deterministic and probabilistic approaches to the phenomenon of anomalous diffusion in tissue, Cell. Mol. Biol. Lett. 4 (1999) 105–116. [117] B. Corry, S. Kuyucak, S.-H. Chung, Tests of continuum theories as models of ion channels. II. Poisson–Nernst–Planck theory versus Brownian dynamics, Biophys. J. 78 (2000) 2364–2381. [118] W. Dyrka, A.T. Augousti, M. Kotulska, Ion flux through membrane channels—an enhanced algorithm for Poisson-Nernst–Planck model, J. Comp. Chem. (2008), in press. [119] S. Mercik, K. Weron, Stochastic origins of the long-range correlations of ionic current fluctuations in membrane channels, Phys. Rev. E 63 (2001) 051910–051919. [120] P.J. Brockwell, R.A. Davis, Introduction to Time Series and Forecasting. Springer, Berlin, (1987). [121] J. Beran, Statistics for Long-Memory Processes. Chapman and Hall, (1994) New York. [122] A. Eke, P. Herman, J.B. Bassingthwaighte, G.M. Raymond, D.B. Percival, M. Cannon, I. Balla, C. Ikrenyi, Physiological time series: Distinguishing fractal noises from motions, Pflugers Arch. Eur. J. Physiol. 439 (2000) 403–415. [123] S. Mercik, K. Weron, K. Burnecki, A. Weron, Enigma of self-similarity of fractional Levy stable motions, Acta Phys. Pol. B 34 (2003) 1–19. [124] K. Burnecki, J. Rosinski, A. Weron, Spectral representation and structure of self-similar processes, in: I. Karatzas, B. Rajput, M. Taqqu (Eds.), Stochastic Processes & Related Topics, Birha¨user, (1998) pp. 1–14. [125] J.W. Lamperti, Semi-stable stochastic processes, Trans. Am. Math. Soc. 104 (1962) 62–78. [126] M. Maejima, Self-similar processes and limit theorems, Sugaku Expositions 2 (1989) 102–123. [127] B. Mandelbrot, J.W. Van Ness, Fractional Brownian Motions, Fractional Noises and Applications, SIAM Rev. 10 (1968) 422–437. [128] A. Weron, K. Burnecki, S. Mercik, K. Weron, Complete description of all self-similar models driven by Levy stable noise, Phys. Rev. E 71 (2005) 016113–016123. [129] J.P. Nolan, Maximum likelihood estimation of stable parameters, in: O.E. Barndorff-Nielsen, T. Mikosch, S.I. Resnick (Eds.), (2001) pp. 379–400. Birkha¨user, Boston. [130] J.P. Nolan, Fitting data and assessing goodness-of-fit with stable distributions, in: J.P. Nolan, A. Swami (Eds.), (1999) . [131] H.E. Hurst, Long-term storage capacity of reservoirs, Trans. Am. Soc. Civ. Eng. 116 (1951) 770–808. [132] C.K. Peng, S.V. Buldyrev, S. Havlin, H.E. Stanley, A.L. Goldberger, Mosaic organization of DNA nucleotides, Phys. Rev. E 49 (1994) 1685–1689. [133] A. Ridi, E. Scalas, A. Gliozzi, Noise measurements in bilayer lipid membranes during electroporation, Eur. Phys. J. E 2 (2000) 161–168. [134] S.M. Bezrukov, M. Winterhalter, Examining noise sources at the single-molecule level: 1/f noise of an open maltoporin channel, Phys. Rev. Lett. 85 (2000) 202–205. [135] Z. Siwy, A. Fulinski, Origin of 1/f a noise in membrane channel currents, Phys. Rev. Lett. 89 (2002) 15810–15814. [136] F. Wohnsland, R. Benz, 1/f noise of open bacterial porin channel, J. Membr. Biol. 158 (1997) 77–85. [137] A. Fulinski, Non-Markovian noise, Phys. Rev. E 50 (1994) 2668–2681. [138] P. Dutta, P.M. Horn, Low-frequency fluctuations in solids: 1/f noise, Rev. Mod. Phys. (1981) 497–515. [139] B. Kaulakys, V. Gontis, M. Alaburd, Point process model of 1/f noise vs a sum of Lorentzians, Phys. Rev. E 71 (2005) 051105.

38

M. Kotulska and K. Kubica

[140] M.B. Weissman, Models for 1/f noise in nerve membranes, Biophys J. 16 (1976) 1105–1108. [141] P. Bak, C. Tang, K. Wiesenfeld, Self-organized criticality: An explanation of the 1/f noise, Phys. Rev. Lett. 59 (1987) 381–384. [142] A.R. Brazhe, G.V. Maksimov, Self-organized critical gating of ion channels: On the origin of long-term memory in dwell time series, Chaos 16 (2006) 033129–37.

C H A P T E R

T W O

Functionalized Liposomes Raghavendra Palankar,1 Yannic Ramaye,1 Didier Fournier,2 and Mathias Winterhalter2,* Contents 40 41 42 42 43 45 46 46 46 49 49 50 50 51

1. Introduction 2. Liposome Formation 3. Stabilization of Liposomes 3.1. PEGylated Liposomes 3.2. Template Polymerization 3.3. Coating with Polyelectrolytes 3.4. DNA Coating 4. Functionalization 4.1. Encapsulation of Water-Soluble Molecules 4.2. Controlled Permeability by Polyelectrolyte Coating 4.3. Reconstitution of Membrane Channels into Liposomes 5. Possible Application of Functionalized Capsules in Diagnostics 5.1. Capsule Array 5.2. Manipulation of Capsules with External Fields: Donnan Potential 5.3. Functionalized Liposome Capsules as Intracellular Sensors and Delivery Vehicles 6. Conclusions Acknowledgments References

53 55 56 56

Abstract Functionalized liposomes are currently used in the area of nanobiotechnology. For example, the interior of a liposome provides a controlled nano-environment with a defined pH, protects molecules against dilution and from hostile environment, for example, from proteases. The substrate entry and the enzymatic reaction product exit (or prodrug digest for therapeutic application) are controlled through membrane channels. In order to benefit from liposomes as building blocks in a larger context, one needs to achieve stability in shape and structure. In biological membranes, the extracellular matrix Corresponding author. Tel.: þ49 421 200-3248; Fax: þ49 421 200-3249; E-mail address: [email protected] (M. Winterhalter).

* 1 2

a

Jacobs University gGmbH, Campus Ring 1, D-27725 Bremen, Germanya Institut de Pharmacologie et Biologie Structurale, 205 rte de Narbonne, Universite´ Paul Sabatier, UMR5089, Toulouse F-31077, France Formerly International University Bremen

Advances in Planar Lipid Bilayers and Liposomes, Volume 7 ISSN 1554-4516, DOI: 10.1016/S1554-4516(08)00002-1

#

2008 Elsevier Inc. All rights reserved.

39

R. Palankar et al.

40

and the cytoskeleton offer a very good stabilization of the lipid membrane. This stratified layer solution was adopted to stabilize liposomes. Various polymers were used to coat liposomes on the outside or inside of the phospholipid bilayer. Furthermore, a few examples are demonstrated.

1. Introduction A current research goal in nanotechnology is to gain control over functional complexes on nanometer scales. Although, achieving this in the laboratory is quite a challenge, evolution developed optimized solutions for this task since the origin of life. For example, in nature, bacteria require for survival a continuous adoption to the environment and a selective exchange of material across the cell wall. Here we show how liposomes can be used to mimic properties of biological cells. The term liposome describes self-assembled, closed structures made of lipids. Knowledge about liposomes evolved with the available tools. For example, first, cell-like structures formed of lipids were observed through microscope by Lehmann a century ago [1]. Later lipids found their interest as colloidal structures and their effect on phase diagrams, osmotic pressure, conductance, or light scattering were investigated [2]. About 40 years ago, the group of Alec D. Bangham studied intensively liposomes as a closed membrane model system with specific permeabilities and surface properties [3]. Since then, many recipes for reproducible formation of liposomes were developed and tools for characterization of structures below micrometer scale improved the field of liposome research. Liposomes are very attractive as biocompatible tools for drug delivery. However, only recently some of the unforeseeable problems like instability in the blood stream and subsequent rapid removal could be partially solved [4]. Modification of the lipid composition provided a larger mechanical stability but the lifetime in the blood stream could only be significantly enhanced using PEGylated lipids based on steric stabilization [4, 5]. Nowadays, hollow nanometer-sized containers are of particular interest in nanotechnology [6–8]. As containers they can protect peptides, proteins, enzymes, or drugs from hostile surroundings and provide an optimal microenvironment different from the bulk medium [7]. Such nanocontainers may be used in drug delivery, in medical diagnostics or as intracellular reporters [8]. Encapsulated molecules can be hidden from the outside; protected against physical, chemical, and biological degradation; targeted to specific cells in defined tissues; and released in a controlled manner. However, liposomes are not suited for many material applications, because of their sensitivity toward environmental changes such as pH, osmotic stress, lipases, and the presence of detergents. Many different approaches have been devised to stabilize liposomes. Notable examples are the use of polymerizable surfactants, incorporation of polymers during the formation of vesicles, and surface grafting with water-soluble polymers [9, 10]. Here we show some of the techniques and a few possible applications.

Functionalized Liposomes

41

2. Liposome Formation Nowadays, many reliable recipes for making liposomes exist [10, 11]. Generally, hydration of dried lipids leads to the formation of multilamellar vesicles (MLVs) and multivesicular vesicles (MVVs). Intensive sonication of such a crude mixture of liposomes forms small unilamellar vesicles (SUV) with diameters of about 30–100 nm. However, SUVs are highly curved leading to instability with the tendency to aggregate and fuse slowly to larger structures. Large unilamellar vesicles (LUV) in the range of 100 nm–1 mm are of particular interest in medicine for drug delivery, diagnostics, food technology, and as sub-micron sized biochemical reactors [2]. A typical recipe for the preparation of LUV is to dissolve 10 mg of lipids, for example, egg-PC, in chloroform in a 10-ml glass tube and then drying the lipids by rotating the tube under gentle stream of nitrogen (N2). During this process, the lipid condensates homogeneously on the wall of the tube and forms a thin film. The latter is subsequently vacuum dried for 3–6 h to remove the residual solvent. Hydration of thus formed lipid film in 1 ml buffer under continued vortexing results in a crude liposome suspension. This suspension is then repeatedly frozen in liquid N2 and thawed in a 25  C waterbath for 10 times, which results in unilamellar vesicles. For some lipid compositions, freeze–thaw cycling alone provides already a homogeneous size distribution. In order to achieve a homogeneous size distribution, the liposome suspension is then extruded 10 times through a filter of desired cut-off diameter. Giant unilamellar vesicles (GUVs) having diameter in micrometer ranges are typically produced by the process of electroformation or electroswelling [12–15]. Because of their size, giant liposomes are extremely sensitive to small changes in osmotic pressure (Laplace law), small osmotic changes, for example, due to water evaporation may cause deflation and the vesicle starts to undulate. Lowering the osmotic pressure outside will cause leakage or rupture. Giant liposomes with diameters above 1 mm can be readily formed from a film of lipids dried on one or both the electrodes that are separated by an aqueous solution and this process is popularly known as the electroswelling. Briefly, lipid dissolved in a solvent is coated onto aluminosilicate glass slides coated with indium tin oxide (ITO) (Delta Technologies, Stillwater http://www.delta-technologies.com). For example, 20 ml of solvated lipid stock of 2.5 mg/ml of egg-PC, dissolved in chloroform can be distributed on the ITO coated glass slide and complete removal of the solvent from the ITO glass slide can be achieved by drying it under vacuum overnight that results in a thin lipid film corresponding to a thickness of 25 bilayers [12, 14]. These slides can be assembled into a flow chamber (Fig. 1) [12]. The lipid film is then hydrated with 300 mM D-sucrose solution and then application of a succession of voltages of a 20-min interval of 10 Hz sinusoidal AC voltage, VRMS ¼ 100 mV, a 3-h interval of 10 Hz sinusoidal AC voltage of VRMS ¼ 1.5 V, and finally a 20-min interval of 0.1 Hz sinusoidal AC voltage of VRMS ¼ 1.5 V. It was found that an initial lower voltage phase results in better yield assisting liposome nucleation [12], and a final very low frequency signal effectively separates the liposomes from the glass electrodes and speeds up their dispersion in the cell [14].

R. Palankar et al.

42

Fabrication

A

Silicone tubing

PDMS Petri dish

Side view B 24 mm

Scotch tape

6.7 mm Top view

Flow chamber

C

Glass plates with ITO

2.1 mm AC

Figure 1 Schematic view as suggested by M. Mayer of a simple set-up for giant liposome formation [12]. It allows to harvest giant liposomes after some time by aspiration on one side of the tubing. (A) In a plastic petri dish, an ITO glass slide is placed on the bottom. (B) A polydimethylsiloxane (PDMS) spacer as shown is tightly placed on top of the ITO glass. (C) Prior to covering the chamber with the second ITO glass slide, the lipids have to be spread and dried.

Another method of liposome preparation, called reverse phase, involves introduction of an aqueous buffer into a mixture of phospholipids dissolved in organic solvent and upon subsequent removal of the organic solvent under reduced pressure [10, 11]. This is an effective method to prepare giant liposomes (diameter of 1–10 mm) and also efficient for encapsulation of substances in the liposomes.

3. Stabilization of Liposomes 3.1. PEGylated Liposomes A well-known technique for colloid stabilization is the effect of steric stabilization. Covalently grafted, flexible water-soluble polymers on the surface of particles prevent their coagulation and precipitation. This strategy was successfully applied

Functionalized Liposomes

43

to stabilize liposomes [2, 4, 16–19]. This type of liposomes is also called ‘‘stealth liposomes’’ due to the steric stabilization by incorporating a few mol% of the polymer polyethylene glycol (PEG) moieties (‘‘PEGylation’’) attached to the lipids head group. The PEG stabilizing effect results from local surface concentration of highly flexible groups that sterically inhibit both hydrophobic and electrostatic interactions of a variety of blood components at the liposome surface [17, 18]. PEGylation of liposomes creates an entropic repulsion reducing protein binding, vesicle aggregation, and fusion [19]. Under in vivo conditions, PEGylation results in reduced uptake by the reticulo-endothelial system and significant prolongation of liposome residence time in the blood which makes them suitable for drug delivery [18]. The PEGylated lipids can be easily incorporated into liposomes by mixing them initially with other lipids in organic solvent and following one of the previously described procedures. Usually the presence of a few mol% PEGylated lipids favors the spontaneous formation of unilamellar vesicles with the PEGylated lipids inserted on both sides of the liposome bilayer. In the context of steric stabilization, it is interesting to note that recently attempts have been made to replace lipids by self-assembling amphiphilic block copolymers, in order to tailor size, form, and stability specifically for individual applications [20, 21]. Notably, AB, ABA, and ABC block copolymers were shown to form micelles, vesicles, or ‘‘sponge’’-like aggregates. Aggregation is avoided by long water-soluble moieties. In contrast to double chain lipids, block copolymers have one hydrophobic chain, which results typically in a finite solubility in water. Depending on the hydrophobic thickness, the mechanical stability may be greatly enhanced [21].

3.2. Template Polymerization A versatile method in polymer chemistry is template polymerizations in surfactant phases: transforming a self-organized molecular assembly into a mechanically and chemically stable supramolecular material. Liposome can be used as template for loading monomers on their surface as well as in the lipid bilayer, which can be subsequently polymerized for obtaining a stabilizing element inside the liposome. Hydrophobic mono- and bifunctional methacrylate monomers were incorporated within the hydrophobic part of the bilayer and their radical polymerization was induced by UV light, leading to the formation of a polymer network within the lipid bilayer [22–22d]. The use of a mono- and bifunctional monomer leads to the formation of a polymer network, which maintains size and shape after removing the lipids. The methacrylate monomers used were butylmethacrylate (BMA), hydroxylethylmethacrylate (HEMA), and ethyleneglycol dimethacrylate (EGDMA) as cross-linker. The main advantage of using UV radiation to initiate the chain reaction is in the very high polymerization rates that can be achieved under intense irradiation, so that the monomer to polymer transition takes place within a fraction of a second. Longer time intervals would promote phase separation as only a homogeneous distribution of mono- and bifunctional monomers within the lipid bilayer can lead to a polymer hollow sphere. The yield of fully polymerized vesicles depends very much on the experimental boundary condition which will render possible upscale difficult. The size distribution was determined before and after UV polymerization

44

R. Palankar et al.

as well as after removal of lipids and confirmed by cryo-transmission electron microscopy (TEM) (Fig. 2). The potential net-like polymer scaffold inside the vesicle bilayer seems not to restrict the lateral mobility of the lipids and the trans-bilayer diffusion of low molecular weight substances, which is important for investigating the interaction of pharmaceutically active substances with biological membranes.

Figure 2 Cryo-transmission electron microscopy (TEM) micrographs of egg-PC/BMA/ EGDMA/Irgacure vesicles at a molar ratio of 2/1/10/20. (A) Control liposomes only composed of egg-PC. (B) Mixed liposome/polymer capsules after UV polymerization. With permission from [22].

45

Functionalized Liposomes

3.3. Coating with Polyelectrolytes Layer-by-Layer (LbL) assembly of charged polymers is a simple but robust and versatile technique for deposition and adsorption of layered films of oppositely charged polyelectrolytes on template surfaces of any given geometry typically inorganic templates that are used for LbL hollow capsule formation [23–25]. Template particles are immersed into a first polyelectrolyte solution centrifuged and washed in polymer-free solution. This step is repeated for each coating. However, this procedure cannot be directly applied to small liposomes as they have negligible density difference to the solution, their separation using centrifugation technique is inefficient. The LbL coating can be scaled up by automatic slow injection of polyelectrolytes followed by a dialysis step [24]. An effective coating requires homogeneity of the polyelectrolyte network, which should increase with the bilayer number. It was shown that more than 15 and 25 polyelectrolyte layers are needed to obtain a homogeneous coating preserving the shape of the capsule following the lipid bilayer removal. Another technique is the incorporation of magnetic nanoparticles, which allows for separation of polyelectrolyte-coated liposomes from the non-coated ones (Fig. 3). Most commonly used polymer for the LbL nanoassembly is the polyanionic polymer polystyrene sulfonate (PSS) and the polycationic polymer polyallylamine hydrochloride (PAH). In this special case, the polyelectrolyte layer can be crosslinked to enhance the stability [24]. The stratified shell could be considered as an artificial analogue of biological cell membranes, the polyelectrolyte component being compared to the cytoskeleton that stabilizes the lipid bilayer. Earlier attempts have been made to coat polyelectrolyte hollow particles with a lipid bilayer [26–28]. However, likely the fuzzy structure of the external polyelectrolyte causes defects in the lipid membrane with significant permeability for ions [29].

PSS coating

PAH coating

Application of magnetic field to separate polyelectrolyte (PE) coated magneto-liposomes from free PE

Figure 3 Schematics showing magnetic separation of polyelectrolyte-coated liposomes encapsulating superparamagnetic nanoparticles.

R. Palankar et al.

46

3.4. DNA Coating Recently, an interesting variation of polyelectrolyte coating using natural polymers has been demonstrated [30]. Covalent grafting of oligonucleotides functionalizes the surface of liposomes and was achieved with maleimid functionalized lipids. Subsequent addition of terminal deoxynucleotidyl transferase elongates the single-stranded DNA. The elongated DNA hybridizes creating a random network. The short segments of double-stranded DNA provide a substrate for the Klenow fragment of E. coli DNA polymerase which synthesizes a double-stranded DNA reinforcing the network. Alternate action of both enzymes leads to a three-dimensional network anchored on the liposome surface (Fig. 4).

4. Functionalization 4.1. Encapsulation of Water-Soluble Molecules Encapsulation of compounds in closed and protective structures is a wide field of research. Liposomes can provide controlled environment for encapsulated molecules and also protect them from (or reverse in the case of toxic compounds) hostile environments. Furthermore, liposome may act as a carrier easier to manipulate than an individual molecule and also as a reservoir with controlled permeability [7, 14, 30].

Figure 4 (A) The combined action of two polymerases creates a DNA shell on a liposome. Singlestrand oligonucleotide steps were grafted on lipid heads. (B) Elongation of single-stranded DNA is catalyzed by terminal deoxynucleotidyl transferase. Depending on the available nucleotides for elongation, the single strands hybridize. (C) Subsequent addition of Klenow DNA polymerase recognizes double-stranded fragments and complements the strands. With permission from [30].

47

Functionalized Liposomes

Hydrophobic molecules may accumulate in the hydrophobic core and have a relatively fast exchange with the external solution depending on the hydrophobic moiety. In contrast to the neutral ones, charged molecules may be entrapped because of the large barrier (bond energy) for ions to cross the hydrocarbon layer of low dielectric constant. Numerous techniques and recipes were suggested to achieve high encapsulation yields within liposomal environment. For example, the so-called pH gradient method is applied to loaded liposomes with small charged particles: the uncharged form of the molecule may easily permeate the lipid membrane. If the inner pH is chosen to be in the range causing ionization, the protonated form will be trapped inside [5]. More complex is the encapsulation of larger molecules. For example, the quantity of encapsulated protein depends on the chosen method and is not fully understood. For example, acetylcholinesterase (AChE) is an interesting enzyme for biosensing and an important target for encapsulation [31–35]. The established thought suggests that the encapsulation occurs with the liposome formation during hydration of the lipid film. However, it appeared that only negligible amounts of protein were encapsulated during the film hydration [32]. In contrast, freeze–thaw seemed to be the essential treatment to load the liposomes with the proteins to achieve the maximum of encapsulation efficiency. This strongly suggested that encapsulation does not occurs during the vesicle budding-off but proteins are loaded in liposomes during freeze–thaw cycles. To test this hypothesis, we prepared liposome suspensions without the enzyme simply by mixing 50 mg of crude lipids with 1 ml 25 mM MOPS buffer. The solution was then submitted to 10 freeze– thaws to obtain monolamellar vesicles. Then, the enzyme was added to the liposomes suspension and the freeze–thaw was applied to load the empty liposomes with the enzyme [35]. Figure 5 clearly demonstrates that encapsulation occurs during the freeze–thaws and its efficiency increases according to the number of cycles.

% Encapsulation

40

30

20

10

0 0

25 50 75 Freeze-thaw cycles number

100

Figure 5 Encapsulation of acetylcholinesterase into liposomes. Here a number of freeze–thaw cycles in the presence of enzymes were applied and the encapsulated activity was measured. The encapsulation was normalized to the initial activity [35].

48

R. Palankar et al.

The plateau reached after 50 cycles represents the maximum of encapsulation efficiency and corresponds to the calculated volume ratio of inner volume out of the total volume. In experiment shown in Fig. 5, the inner volume corresponded to 30% of the total volume in the cuvette; after 50 freeze–thaw cycles, 30% of the initial protein in the cuvette was encapsulated in the liposomes. How does the enzyme gain entry into the liposomes during the freeze–thaw cycles? Our hypothesis was that encapsulation is mostly due to the large expansion of the water at the freezing point which will cause rupture of the lipid vesicles. During the freezing cycle, the encapsulated water changes the volume by 10% leading to a burst in the lipid shell. It is tempting to hypothesize that during the first thawing, the enzyme enters into the liposome through the unclosed fracture. Concentration of the enzyme inside and outside does not equilibrate within a single freeze–thaw cycle either because closing of the membrane is too rapid during the thawing or because the probability to generate a fracture with a sufficient size during the freezing is weak. After concentration of enzyme inside and outside the liposomes equilibrates, the encapsulation efficiency reaches a plateau corresponding to the maximum encapsulation efficiency. The defect size occurring during a freeze– thaw cycle can be estimated using the results from elasto-mechanical studies on liposomes. Micropipette stretching of giant liposomes revealed a critical area increase of 2–4% above which the liposomes undergo irreversible rupture. Assuming a 200-nm diameter liposome with the membrane surface area of 1.2  105 nm2, the volume increase by water freezing requires an additional 6–7% or 0.8  105 nm2 in extra lipid area. As typically 2–3% lipid area stretching cause rupture, we expect defect formation. In the rigid frozen membrane of the liposome, this surface increase generates at least one fracture among the lipids. This fracture’s area corresponds to 80 times the size exclusion of the enzyme (10 nm2). To test this explanation, we made the enzyme exit the liposomes. Enzyme containing liposomes were subjected to a number of freeze–thaw cycles and the enzymatic activity outside was revealed. In this case, the concentration driving force obliges the enzyme to diffuse outside the liposomes [35]. Repetitive freeze–thaw is usually detrimental for protein. During thermal changes, buffer acid or basic species have different precipitation behavior in solution, locally creating dramatic pH changes that denature proteins. Fast freezing in liquid nitrogen (195  C) and thawing in water-bath (37  C) avoided this differential precipitation and preserved the protein during the encapsulation procedure. Furthermore, high concentration of the enzyme to be encapsulated and high concentration of liposomes protect the protein from denaturation by freezing. In all experiments, denaturation due to the freeze–thaw cycles was always below 10% [35]. Furthermore, it should be noted that any methods in which detergent monomers appear in larger concentration will lead to denaturation. For example, our experience with triblock copolymer has shown that the high monomer concentration leads to rapid denaturation of the enzyme. Also, we tested the template polymerization method for stabilizing our acetylesterase. The presence of hydrophobic monomers denatured strongly the function of the enzyme.

Functionalized Liposomes

49

4.2. Controlled Permeability by Polyelectrolyte Coating Intensive research in the area of the LbL technique now allows nanometer-scale control over a range of chemical and physical properties [23]. During the LbL process, environment-responsive fluorescent probes, drugs, peptides, small proteins, nucleic acids and nanoparticles can be entrapped in the polymer nanoassembly. Cross-linking of polyelectrolytes increases further the stability of hollow capsules, reduces their permeability and consequently allows a sustained slow released of the encapsulated material [24].

4.3. Reconstitution of Membrane Channels into Liposomes Many reconstitution protocols for various membrane proteins have been developed [36]. Most proteins are fragile and the handling during reconstitution is crucial for the function. With respect to functionalization of the liposome, we report the most relevant. Organic solvent-mediated reconstitution is one of the technique to incorporate protein in liposomes. The limit of this technique is the denaturation of amphiphilic membrane protein by solvent exposure [37]. Bacteriorhodopsin can be reconstituted in liposomes by a low pH sonication process. The incorporation of bacteriorhodopsin in these proteoliposomes is predominantly in the same direction as in vivo and the direction of proton pumping is from inside to outside the liposomes [38]. The advantage of sonication technique is the rapidity of the technique and that no detergent is needed. The limit of the sonication technique is the inactivation of proteins by long-time sonication. The most successful and frequently used strategy for proteoliposome preparation is that involving the use of detergents, because most membrane proteins are isolated and purified in the presence of detergents. In the standard procedure, these proteins are first cosolubilized with phospholipids in the appropriate detergent in order to form an isotropic solution of lipid–protein– detergent and lipid–detergent micelles. Next, the detergent is removed resulting in the progressive formation of bilayer vesicles with incorporated protein. Solubilized protein in detergent can also incubated with preformed liposomes with a concentration of detergent above the critical micellar concentration. Upon lowering the detergent concentration by adsorption onto Bio-Beads, the protein either reincorporates directly into the liposomes saturated in detergent or associates with the lipids upon removal of the detergent to form proteoliposomes [39]. Very simple and robust is the reconstitution of OmpF, an outer membrane protein from E. coli. The protein (1 mg/ml) itself is stocked in 1% detergent Octyl-Polyethyleneoxide (OPOE) solution. Small amounts of prediluted stock solution (a few microliters into a milliliter of liposomes) are added to the liposomes and vortexed. Within a few minutes, the protein inserts into the liposome. Alternatively, the stock solution can also be added to the dry lipid film. A few microliters of the stock solution is spread over the film. The vacuum is applied for a few minutes and the film is resuspended and vigorously vortexed. It is interesting to note that membrane proteins can also be reconstituted in a functional state into polymeric membranes [21]. At least two features of the polymer membrane facilitated protein miscibility: (i) polymer chains can be compressed

R. Palankar et al.

50

considerably due to the high flexibility of the hydrophobic polydimethylsiloxane (PDMS) block; and (ii) polydispersity of the polymer allows small chains to segregate around a membrane protein.

5. Possible Application of Functionalized Capsules in Diagnostics 5.1. Capsule Array Grafting oligonucleotides to the surfaces allows the specific attachment of the capsules on designed places. Fixing capsules of about a few 100 nm in size requires a 30-mers to deliver enough energy for selective and stable hybridization. A possible approach to functionalize the oligonucleotide is at the 50 end using a thiol group. On the other hand, the lipid headgroup may be linked to the thiol group. We choose the maleimido phenyl butyrate (MBP) group which has the advance to form a covalent bond R–C–S–R0 under biologically favorable condition (Hepes at neutral pH). A straight solution is to use the following commercially available lipids (Avanti polar lipids): DSPE-PEG(2000) maleimide (1,2-distearoyl-sn-glycero-3-phosphoethanolamineN-[maleimide(polyethylene-glycol)2000]) (ammonium salt) and 18:1 MPB-PE (1,2-dioleoyl-sn-glycero-3-phosphatidylethanolminde-N-[4-( p-maleimidiphenyl) butyramide]) (sodium salt). The PEG spacer (45 units) will provide extra flexibility. The oligonucleotides will be fixed to the lipid prior to vesicle formation. However, in this case, half of the functionalized lipids will be found inside the capsule. As the grafting protocol is compatible with the capsule stability, we may link the oligonucleotides after capsule formation [40]. Conversion of a DNA chip to a nanocapsule array was performed by grafting oligonucleotides on a liposome. The template used is complementary to an oligonucleotide bound to the array. Each liposome may be loaded by a soluble molecule or may present a hydrophobic or amphiphilic molecule inserted in its wall. To detect liposomes on the chip, we used fluorescent dyes encapsulated in the liposome internal volume or fluorescent lipids. We observed that oligonucleotide-grafted liposomes containing a defined dye specifically accumulated on the area where its complementary oligonucleotide has been spotted on the array. The virtually unlimited amount of addresses allows the specific binding of large amount of liposomes in one single batch. Oligonucleotide-grafted liposomes are very exciting building blocks to create networks and other super structures. DNA provides a virtual addressing without limits. Earlier systems used competitive binding of oligonucleotide to DNA covered liposomes to probe for the presence of specific DNA fragments. However, the liposomes serve only as an amplifier due to their size. Others created a larger network and crosslinking was observed via turbidity measurements. The liposomes appear to be excellent carriers for molecules to design of a new kind of chip. The use of oligonucleotides as a mean of anchoring shows a very high specificity and simplifies the localization of molecules on the array. Then, molecules can be directed to their assigned place on the array in a single incubation step. However, the possibility to detect chemical enzymatic reactions is still limited by the finding of a unique reaction substrate for all the spotted

51

Functionalized Liposomes

proteins, and a system to avoid the spreading of the detectable product on the surface of the array. Here again, the use of containers could be an advantage to maintain proteins in a specific reaction environment and to trap the enzymatic reaction products.

5.2. Manipulation of Capsules with External Fields: Donnan Potential Liposomes are exceptionally large barriers for charged molecules. Encapsulation of polyelectrolytes inside liposomes will also include the necessary counterions to maintain the inside neutral. However, adding channels will create a semipermeable membrane allowing ions to equilibrate whereas the channel is impermeable for the polyelectrolyte. Channel forming proteins are tailored by evolution to incorporate into lipid membranes. Adding a porin preparation, solubilized in a detergent shell to a solution containing liposomes, results in semipermeable nanocapsules. This will give rise to a Donnan potential. This Donnan potential may be used for external manipulation of nanocontainers via coupling of the capsule to an external electric field, or for the selective uptake of small charged molecules into the capsule [14]. A possible application would be to use the Donnan potential for external manipulation and molecular sieving. A membrane, permeable only for small cations and anions, separates aqueous solutions of the same salt concentration, c 0 . On one side of the barrier, a polyelectrolyte of charge zp and concentration cp is added which is not able to permeate the membrane and the following Donnan potential is expected:

E¼

RT c 00 ln 0 c F

ð1Þ

In the limit of low salt concentrations, c 0  |zpcp| and with z > 0, c 00  zpcp results. The number of counterions of the polyelectrolyte inside the liposome as in Fig. 6 just equals the number of charges of the enclosed polyelectrolyte. To maintain the balance of the chemical potential of the permeating species, less ions of the same charge as the polyelectrolyte can be found within the liposome. The concentration gradient of the counterion to the polyelectrolyte across the liposome membrane causes the formation of a transmembrane potential. Because of the difficulty to directly measure transmembrane potentials of liposomes, we used a Malvern Instruments Zetasizer Nano ZS to obtain the related zeta potential. The instrument calculates the zeta potential from determining the electrophoretic mobility of the particles by performing an electrophoresis experiment on the sample and measuring the velocity of the particles using Laser Doppler Velocimetry. A Donnan potential allows to couple otherwise uncharged liposomes to an external electric field. Furthermore, the Donnan potential gives us a tool to investigate the permeability of lipid membranes. With the expression of a Donnan potential, we may study the diffusion of ions across lipid membranes through channel proteins and the permeability of porins for macromolecules. As long as there is a concentration gradient for the encapsulated polyelectrolyte between inside and outside of the capsule, the liposome acts like a charged particle and moves within

R. Palankar et al.

52

Transmembrane voltage V ϕ⬙

ϕ⬘

Na+ (C⬙+) CI− (C⬙−) pzp (Cp)

Na+ (C⬘+) CI− (C⬘−)

Figure 6 Schema of the building of a Donan potential through encapsulation of a charged polyelectrolyte. Adapted from [14].

an external electric field. If the concentration gradient vanishes, no force is acting on the liposome. With carefully chosen different species of liposomes—specific polyelectrolyte content and specific fluorescence labels—it should be possible to map the interior of cells revealing different concentrations of charged molecules as RNA. Furthermore, we demonstrated that liposomes might be manipulated by external electric fields, as used for the electrophoretic zeta potential measurements. In these liposomes, we may encapsulate different enzymes for later use in biosensing. In combination with channel forming proteins, in our case porins, we may control the permeation of substrates through the lipid membrane. Currently, we are working on improving the selectivity of the capsule permeability. The availability of structural information for several porins to a few Angstrom resolution permits the design of more efficient or more specific mutants of channel forming proteins by molecular engineering. It might be possible to create affinity sites necessary for specific sensors. Additionally, we investigate the controlled opening and closing of specific porins using bacterial phages recognizing a site on the porin [42]. Nanoreactors also represent a new type of delivery devices for application in pharmacology and diagnostics. It is especially important that such systems provide a constant release of substances over an extended period of time. It has to be emphasized that the system introduced in this study is a step into a novel brand of

Functionalized Liposomes

53

nanometer-sized bioreactors. In fact, nature provides many specific, unspecific, or ligand-gated channels, which can be reconstituted in the same way, providing a unique tool to control permeation across the nanoreactor shells. We believe that the principle of using the protective feature of such nanocontainers in combination with controlled permeability either by natural or genetically modified channels or pumps will have many future applications.

5.3. Functionalized Liposome Capsules as Intracellular Sensors and Delivery Vehicles Currently, we investigate inasmuch functionalized liposome capsules could be introduced into mammalian cells in culture to probe for intracellular substrates or for controlled delivery of encapsulated contents. Analysis of intracellular analytes of living cells or sub-cellular compartments there-in involves sample volumes and spatial dimensions that are below microliter and micrometer regimes, respectively. Traditionally, this has been achieved though miniaturized optodes and electrodes, which are bulky and impractical for routine measurements inside the cells, since they occupy a considerable amount of cellular space and cause undesirable cellular perturbations [43]. In contrast, individual molecular probes (free dyes) which are powerful tools for probing intracellular environment are also being widely used [44]. However, these probes suffer from chemical interference between probe and cellular compartments, organellar sequestration, toxicity, and leakage from the cell which frequently complicate the measurements and data interpretation. This complex environment can be probed only with embedded, stand-alone sensors which can help in detection of ions, molecular oxygen, pH, and range of other substrates involved in biochemical processes in a specific cellular compartment. To achieve data analysis in real time the sensor must have a narrow size distribution, a homogeneous amplification of the signal, minimal interference and perturbation of the cellular environment. For effective introduction of functionalized liposome capsules into intracellular environment of mammalian cells in vitro, electroporation and microinjection can be used. Electroporation or electropermeabilization is a well-known method to transfect mammalians cells with extracellular molecules, gene transfer, and for cell hybridization [45]. Depending on the electric field strength, we observed a stimulated uptake of polyelectrolyte coated liposome capsules from 200 nm up to a few micrometers in diameter (Fig. 7). Smaller capsules can be mechanically inserted by capillary pressure microinjection (CPM) using a microinjector. CPM is a mechanical and effective technique for delivery of small volumes of samples into adherent or suspended cells. The technique employs a thin glass capillary which is injected into the cell through the cell membrane [46]. With the help of precision pumps, the desired substance can then be delivered to the cell upon a pressure pulse. The size of the capsule that can be injected through the capillary depends on the inner diameter of the fine tip of the capillary. We tested the delivery of 200 nm sized liposomes labeled with RhodamineB (RhoB) and coated with four layers of polyelectrolytes and injected

54

Figure 7 (Continued )

R. Palankar et al.

Functionalized Liposomes

55

Figure 7 Confocal laser scanning microscopy (CLSM) images of uptake of coated liposome capsule by electropermeabilized Vero cells. A and B show fluorescence signals emitted from RhodamineB (RhoB, red channel, indicated with the arrow) labeled lipids incorporated in the liposome bilayer and the PAH-fluorescein isothiocyanate (FITC, green channel, indicated with the arrow) coated on the liposome, respectively, and C is a overlay of A and B confirming the colocalization of the two fluorescence signals (yellow signal, indicated by the arrow) coming from the liposome capsule. Scale bar 10 mm.

them into adherent Vero cells using an Eppendorf Femto Jet microinjection system connected to a micromanipulator Eppendorf InjectMan NI 2. After microinjection, fluorescence of the injected particles was detected with a confocal laser scanning microscopy (CLSM). Analysis of the images reveals successful intracellular delivery of the nanocapsules.

6. Conclusions We presented a number of new approaches to create nanometer sized capsules with enhanced stability and multifunctionality. Stabilization of these size tunable nanocompartments can be achieved by a coating of polymers at the surface or by in situ polymerization of monomers in the lipids bilayer or at the surface. This stabilization provides a better chemical and thermal resistance. Incorporation of molecules or protein-like enzyme in these liposomes and the control of the permeability made these vesicles an ideal tool for intracellular reporter or for delivery of encapsulated contents.

56

R. Palankar et al.

ACKNOWLEDGMENTS This research was supported by the Volkswagenstiftung through Nanoengineered Polymer capsules (I/80 052–54) and by AC Nanosciences-Nanotechnologies (NN082).

REFERENCES [1] O. Lehmann, Die flu¨ssigen Kristalle, Akademische Verlagsanstalt, Leipzig, 1911. [2] D.D. Lasic, Y. Barenholz, Non Medical Application of Liposomes, Vol. I. CRC Press, Boca Raton, 1995. [3] A.D. Bangham, M.M. Standish, J.C. Watkins, Diffusion of univalent ions across the lamellae of swollen phospholipids, J. Mol. Biol. 13 (1965) 238–252. [4] A. Gabizon, D. Papahadjopoulos, Liposome formulations with prolonged circulation time in blood and enhanced uptake by tumors, Proc. Natl. Acad. Sci. USA 85 (1988) 6949–6953. [5] B. Ceˇh, M. Winterhalter, P.M. Frederik, J.I. Vallner, D.D. Lasic, StealthÒ liposomes: from theory to product, Adv. Drug Deliv. Rev. 24 (1997) 165–177. [6] E. Donath, G.B. Sukhorukov, F. Caruso, S.A. Davis, H. Mo¨hwald, Novel hollow polymer shells: Fabrication, characterization and potential applications, Angew. Chem. Int. Ed. 37 (1998) 2202–2205. [7] M. Nasseau, Y. Boublik, W. Meier, M. Winterhalter, D. Fournier, Substrate-permeable encapsulation of enzymes maintains effective activity, stabilizes against denaturation, and protects against proteolytic degradation, Biotechnol. Bioeng. 75 (2001) 615–618. [7a] J.P. Colletier, B. Chaize, M. Winterhalter, D. Fournier, Fournier, Protein encapsulatin in liposomes. Efficiency depends on interactions between protein and phospholipid bilayer, BMC-Biotechnol. 2 (2002) 9–17. [8] G.B. Sukhorukov, A.L. Rogach, M. Garstka, S. Springer, W.J. Parak, A. Munoz-Javier, O. Kreft, A.G. Skirtach, A.S. Susha, Y. Ramaye, R. Palankar, M. Winterhalter, Multifunctionalized polymer microcapsules: Novel tools for biological and pharmacological applications, Small 3 (2007) 944–955. [9] H. Ringsdorf, B. Schlarb, J. Venzmer, Molecular architecture and function of polymeric oriented systems: Models for the study of organization, surface recognition, and dynamics of biomembranes, Angew. Chem. Int. Ed. Engl. 27 (1988) 113–158. [10] F. Szoka, D. Papahadjopoulos, Comparative properties and methods of preparation of lipid vesicles (liposomes), Annu. Rev. Biophys. Bioeng. 9 (1980) 467–508. [11] M.R. Mozafari, Liposomes: An overview of manufacturing techniques, Cell. Mol. Biol. Lett. 10 (2005) 711–719. [12] D.J. Estes, M. Mayer, Giant liposomes in physiological buffer using electroformation in a flow chamber, Biochim. Biophys. Acta 1712 (2005) 152–160. [12a] M. Mayer, S. Terrettaz, L. Giovangrandi, H. Vogel, Functional analysis of ion channels: Planar patch clamp and impedance spectroscopy of tethered lipid membranes, in: J.M. Cooper, A.E.G. Cass (Eds.), Biosensors: A Practical Approach (2nd ed.), Oxford University Press, Oxford, 2003, p. 153. [12b] A. Moscho, O. Orwar, D.T. Chiu, B.P. Modi, R.N. Zare, Rapid preparation of giant unilamellar vesicles, Proc. Natl. Acad. Sci. USA 93 (1996) 11443–11447. [13] M.I. Angelova, D.S. Dimitrov, Liposome electroformation, Faraday Discuss. 81 (1986) 303–311. [14] M. Lindemann, M. Winterhalter, Membrane channels as a tool to control nanoreactors, IEEProc. Nanobiotechnol. 153 (2006) 107–111. [15] A commercial instrument is now available called Vesicle Prep Pro from Nanion Technologies GmbH, Mu¨nchen, Germany. [16] V.P. Torchilin, V.G. Omelyanenko, M.I. Papisov, A.A. Bogdanov, V.S. Trubetsloy, J.N. Herron, C.A. Gentry, Poly(ethylene glycol) on the liposome surface: On the mechanism of polymer-coated liposome longevity, Biochem. Biophys. Acta 1195 (1994) 11–20.

Functionalized Liposomes

57

[17] M.C. Woodle, D.D. Lasic, Sterically stabilized liposomes, Biochim. Biophys. Acta 1113 (1992) 171–199. [18] D.D. Lasic, F.J. Martin, A. Gabizon, S.K. Huang, D. Papahadjopoulos, Sterically stabilized liposomes: A hypothesis on the molecular origin of the extended circulation times, Biochim. Biophys. Acta 1070 (1991) 187–192. [18a] M.C. Woodle, Sterically stabilized liposomes, Adv. Drug Deliv. Rev. 32 (1998) 139–152. [19] M. Ka¨sbauer, D.D. Lasic, M. Winterhalter, Polymer induced fusion and leakage of small unilamellar vesicles, Chem. Phys. Lipids 86 (1997) 153. [20] S. Nardin, M. Thoeni, M. Winterhalter, W. Meier, Nanoreactors based on polymerized ABAtriblock copolymer vesicles, Chem. Commun. (2000) 1433–1434. [21] T. Ruysschaert, A.F. Sonnen, T. Haefele, W. Meier, M. Winterhalter, D. Fournier, Hybrid nanocapsules: Interactions of ABA block copolymers with liposomes, J. Am. Chem. Soc. 127 (2005) 6242–6247. [22] J. Gomes, A.F. Sonnen, A. Kronenberger, J. Fritz, M. Coelho, D. Fournier, C. Fournier-No¨el, M. Mauzac, M. Winterhalter, Stable polymetacrylate nanocapsules from ultraviolet light induced template radical polymerization of unilamellar liposomes, Langmuir 22 (2006) 7755–7759. [22a] N. Poulain, E. Nakache, A. Pina, G. Levesque, Nanoparticles from vesicle polymerization: Characterization and kinetic study, J. Polymer Sci. Part A 34 (1996) 729–737. [22b] A. Graff, M. Winterhalter, W. Meier, Nanoreactors from polymer-stabilized liposomes, Langmuir 17 (2001) 919–923. [22c] M. Jung, D.H.W. Hubert, P.H.H. Bomans, P.M. Frederik, J. Heuldisk, A.M. van Herck, H. Fisher, A.L. German, A topology map for novel vesicle  polymer hybrid architectures, Adv. Mater. 12 (2000) 210–213. [22d] J.K. Murtagh, J. Thomas, Mobility and reactivity in colloidal aggregates with motion restricted by polymerization, Faraday Discuss. Chem. Soc. 81 (1986) 127–136. [23] A.A. Antipov, G.B. Sukhorukov, Polyelectrolyte multilayer capsules as vehicles with tunable permeability, Adv. Colloid Interf. Sci. 111 (2004) 49–61. [24] M. Germain, S. Grube, V. Carriere, H. Richard-Foy, M. Winterhalter, D. Fournier, Nanocontainer with composite wall: Lipid vesicles coated with several layers of crosslinked polyelectrolytes, Adv. Mater. 21 (2006) 2868–2871. [25] K.C. Wood, H.F. Chuang, R.D. Batten, D.M. Lynn, P.T. Hammond, Controlling interlayer diffusion to achieve sustained, multiagent delivery from layer-by-layer thin films, Proc. Natl. Acad. Sci. USA 103 (2006) 10207–10212. [26] O.P. Tiourina, I. Radtchenko, G.B. Sukhorukov, H. Mohwald, Artificial cell based on lipid hollow polyelectrolyte microcapsules: Channel reconstruction and membrane potential measurement, J. Membr. Biol. 190 (2002) 9–16. [27] L. Ge, H. Mo¨hwald, J. Li, Mechanical property of lipid-coated polyelectrolyte microcapsules, J. Nanosci. Nanotechnol. 6 (2006) 2489–2493. [28] G. Krishna, T. Shutava, Y. Lvov, Lipid modified polyelectrolyte microcapsules with controlled diffusion, Chem. Commun. (2005) 2796–2798. [29] R. Georgieva, S.E. Moya, H. Ba¨umler, H. Mo¨hwald, E. Donath, Controlling ionic conductivity in lipid polyelectrolyte composite capsules by cholesterol, J. Phys. Chem. 109B (2005) 18025–18030. [30] T. Ruysschaert, L. Paquereau, M. Winterhalter, D. Fournier, Stabilization of liposomes through enzymatic polymerisation of DNA, Nano Lett. 6 (2006) 2755–2757. [31] B. Chaize, D. Fournier, Sorting out molecules reacting with acetylcholinesterase by enzyme encapsulation in liposome, Biosens. Bioelectron. 15 (2004) 49–55. [32] B. Chaize, J.B. Colletier, M. Winterhalter, D. Fournier, Encapsulation of enzymes in liposomes: High encapsulation efficiency and control of substrate permeability, Artif. Cells Blood Substit. Biotechnol. 32 (2004) 67–75. [33] P. Walde, S. Ichiwaka, Enzymes inside lipid vesicles: preparation, reactivity and application, Biomol. Eng. 18 (2001) 143–177. [34] S. Sotiropoulou, D. Fournier, N. Chaniotakis, Genetically engineered acetylcholinesterase-based biosensor for attomolar for detection of dichlorvos, Biosens. Bioelectron. 20 (2005) 2347–2352.

58

R. Palankar et al.

[35] B. Chaize, M. Winterhalter, D. Fournier, Encapsulation of acetylcholinesterase in preformed liposomes, Biotechniques 34 (2003) 1158–1162. [36] M. Winterhalter, C. Hilty, S.M. Bezrukov, C. Nardin, W. Meier, D. Fournier, Controlling membrane permeability with bacterial porins: Application to encapsulated enzymes, Talanta 55 (2001) 965–971. [36a] M. Yoshimoto, S. Wang, K. Fukunaga, D. Fournier, P. Walde, R. Kuboi, K. Nakao, A novel immobilized liposomal glucose oxidase system using the channel protein OmpF and catalase, Biotechnol. Bioeng. 90 (2005) 231–238. [37] J.L. Rigaud, B. Pitard, D. Levy, Reconstitution of membrane proteins into liposomes: Application to energy-transducing membrane proteins, Biochim. Biophys. Acta 1231 (1995) 223–246. [38] A.J.M. Driessen, K.J. Hellingwerf, W.N. Konings, Light-induced generation of a proton motive force and Ca[2þ]-transport in membrane vesicles of Streptococcus cremoris fused with bacteriorhodopsin proteoliposomes, Biochim. Biophys. Acta 808 (1985) 1–12. [39] L. Plancon, M. Chami, L. Letellier, Reconstitution of FhuA, an Escherichia coli outer membrane protein into liposomes, J. Biol. Chem. 272 (1997) 16868–16872. [40] B. Chaize, M. Nguyen, T. Ruysschaert, V. Le Berre, E. Tre´visiol, A.-M. Caminade, J.P. Majoral, G. Pratviel, B. Meunier, M. Winterhalter, D. Fournier, Microstructured liposome array, Bioconjug. Chem. 17 (2006) 245–247. [41] C.R. Cantor, P.R. Schimmel, Biophysical Chemistry, Vol. 3. Freeman and Company, New York, 1980, p. 1336. [42] P.A. Gurnev, A.B. Oppenheim, M. Winterhalter, S.M. Bezrukov, Docking of a single phage lambda to its membrane receptor maltoporin as a time-resolved event, J. Mol. Biol. 359 (2006) 1447–1455. [43] S.M. Bucka, H. Xua, M. Brasuel, M.A. Philbert, R. Kopelman, Nanoscale probes encapsulated by biologically localized embedding (PEBBLEs) for ion sensing and imaging in live cells, Talanta 63 (2004) 41–59. [44] I. Johnson, Fluorescent probes for living cells, Histochemical 30 (1998) 123–140. [45] V. Kiermer (Ed.), Classic protocol, Transfection of mammalian cells by electroporation, Nat. Methods 3 (2006) 67–68. [46] O. Seksek, J. Biwersi, A.S. Verkman, Evidence against defective trans-golgi acidification in cystic fibrosis, J. Biol. Chem. 271 (1996) 15542–15548.

C H A P T E R

T H R E E

Pore-Suspending Membranes on Highly Ordered Porous Alumina and Porous Silicon Substrates: Preparation, Characterization, and Application Claudia Steinem1,* Contents 60 61 61 62 63 64 64 67 68 69 69 71 73 75 75 75

1. Introduction 2. Porous Substrates 2.1. Highly Ordered Porous Alumina 2.2. Highly Ordered Porous Silicon 3. Formation of Micro- and Nano-BLMs 4. Characterization of Nano- and Micro-BLMs 4.1. Electrical Characterization of Nano-BLMs 4.2. Optical Characterization of the Rupturing Process of Micro-BLMs 4.3. Lateral Diffusion of Lipids in Micro-BLMs 5. Applications of Nano-BLMs 5.1. Monitoring the Activity of Bacteriorhodopsin 5.2. Monitoring Single Channel Events of Peptides and Proteins 6. Solvent-Free Pore-Suspending Membranes 7. Conclusions Acknowledgments References

Abstract In the past years, we have developed a planar chip-based arrangement of poresuspending lipid bilayers that combine the merits of freestanding and solid supported bilayers. Here, I will discuss the preparation and characterization of pore-suspending membranes on highly ordered porous alumina and porous silicon substrates starting either from lipids dissolved in an organic solution [nano- and micro-black lipid membranes (BLMs)] or from vesicles. Impedance spectroscopy results reveal that single lipid bilayers are formed, which are extraordinary stable. By a combination of * Corresponding author. Tel.: þ49-551-393294; Fax: þ49-551-393228; E-mail address: [email protected] (C. Steinem). 1

Institute for Organic and Biomolecular Chemistry, Georg-August University, Tammannstr. 2, 37077 Go¨ttingen, Germany

Advances in Planar Lipid Bilayers and Liposomes, Volume 7 ISSN 1554-4516, DOI: 10.1016/S1554-4516(08)00003-3

#

2008 Elsevier Inc. All rights reserved.

59

60

C. Steinem

fluorescence and scanning force microscopy, we were able to visualize and manipulate these pore-suspending membranes. Fluorescence recovery after photobleaching experiments has proven that the lateral mobility of the lipids within the plane of the membrane is high. Applications of nano-BLMs range from the investigation of the light-activated proton pump bacteriorhodopsin to various ion channel forming peptides and proteins. Owing to the very high membrane resistance obtained for nano- and micro-BLMs prepared from organic lipid solution, channel activities can be monitored down to the single channel level.

1. Introduction About one third of all proteins are membrane proteins and, besides G-protein coupled receptors, ion channels and transporter proteins are of particular importance for the pharmaceutical industry. They represent opportunities for drug intervention owing to the fact that they are remarkably sensitive to chemical modulation. Currently, there exist ion channel therapeutics for anesthesia, anxiety, epilepsy, hypertension, insomnia, and pain, and excellent opportunities for ion channel therapeutic modulation in, for example, affective disorders, allergic disorders, autoimmune diseases, contraception, incontinence, and stroke [1]. Motivated by the importance to understand the functioning and possible modulations of ion channels, several systems have been developed with the aim to be able to monitor an ion channel-mediated selective ion flow across a membrane. Besides techniques that are based on cellular systems, such as patch clamp and whole cell recordings [2], artificial membranes have been established such as freestanding and solid supported membranes. In particular, since the pioneering work of Mu¨ller and Rudin [3], lipid bilayers suspending a small aperture have become an invaluable tool to investigate ion channels and pumps. However, these membranes, known as black lipid membranes (BLMs), generally lack mechanical and long-term stability and they cannot be integrated, automatized, and parallelized, which are prerequisites for the development of a chip-based sensor technology. To overcome these problems, micromachined supports for bilayers have been developed in recent years. For example, small apertures are fabricated in silicon by standard lithography [4–6]. Evans and coworkers [7–9] constructed a micromachined support by conventional photolithography consisting of a planar silver/silver chloride electrode on glass, upon which a small compartment was formed by the photoresist SU8. Fertig et al. [10, 11] used apertures that were manufactured in glass substrates by the single ion track etching method with diameters between 1 and 10 mm. By spreading diphytanoylphosphatidylcholine in n-decane across the hole, bilayers were established with typical membrane capacitances and resistances between 1 and 100 GO that were stable for hours. Another approach to stabilize the very fragile BLMs, while still providing two aqueous compartments to ensure full functional insertion of transmembrane proteins and the electrical readout down to a single ion channel, is to use porous supports that are suspended by lipid bilayers. For example, Hemmler et al. [12] prepared lipid

Pore-Suspending Membranes on Porous Alumina and Porous Silicon Substrates

61

membranes by directly spreading lipids from organic solution on the surfaces of laser structured arrays of pores with pore diameters of 30–70 mm and track etched polycarbonate membranes with pore diameters of 5 and 0.8 mm. Mixed hybrid bilayer membranes composed of octadecanethiol and phosphatidylcholine on functionalized polycarbonate membranes with pore diameters of 1 mm and a pore density of 105–107 pores/cm2 were described by Favero et al. [12–14]. We have been able to develop lipid bilayers suspending the pores of highly ordered porous substrates such as porous alumina and porous silicon. The preparation of the porous materials, the formation of pore-suspending membranes as well as their applications for the detection of ion transport mediated by ion channels and pumps will be discussed in more detail in this chapter.

2. Porous Substrates For the formation of pore-suspending membranes, we aimed at a porous material that fulfills certain requirements: (i) the pore size of the material is defined and well adjustable in the nanometer regime; (ii) the substrate is mechanically, chemically, and electrically stable; and (iii) the pore array is highly ordered with a narrow pore size distribution. Electrochemically grown porous alumina substrates turned out to fulfill these requirements. Even though electrochemically grown pores in metals have been studied for about 50 years [15], only in the last 10 years, intense research efforts have led to ordered pore arrays with a homogeneous morphology of parallel pores, which grow perpendicular to the surface with a narrow distribution of pore diameters and interpore spacings [16].

2.1. Highly Ordered Porous Alumina Starting from neat aluminum disks, which are already covered by a thin aluminum oxide layer, they are first electropolished to smoothen the surface followed by anodization in acidic electrolyte, which is described by the following set of equations:

Anode:

2Al þ 3H2 O ! Al2 O3 þ 6Hþ þ 6e

Cathode:

6Hþ þ 6e ! 3H2 :

Only if the formed aluminum oxide is weakly dissolved in the electrolyte solution, pores are generated during anodization. The thickness of the porous layer can be adjusted by the anodization time, while the pore diameter and interpore distance are determined by the applied voltage [17]. The pore depth depends linearly on the anodization time. If the anodization is performed in oxalic acid at a voltage of 40 V, pore growth is reported to be 1–2 mm/h, while in phosphoric acid at a voltage of 160 V, the pore growth is 4–5 mm/h [17, 18]. Both, pore diameter and interpore distance are proportional to the applied potential, with a proportionality constant for the interpore distance of 2.5 nm/V [19]. Low temperature favors the ordering of the pore array.

62

C. Steinem

Although the pores start growing at the upper surface exposed to the electrolyte at almost random positions, virtually hexagonally ordered pores are found at the bottom of the oxide layer. In 1995, Masuda and Fukuda [20] rationalized this selforganized arrangement of neighboring pores in ordered hexagonal arrays by a moderate repulsive interaction between the pores during growth, which favors energetically the closest pore packing. Mechanical stress due to volume expansion during aluminum oxidation has been proposed by Jessensky et al. [16] and Parkhutik and Shershulsky [21] to be the origin of these repulsive forces. To obtain hexagonally pores throughout the entire substrate, a two-step anodization process is used. Porous alumina formed in the first anodization step is selectively removed by immersion in a 60  C hot aqueous solution of phosphoric and chromic acid [15]. The remaining pattern of the aluminum surface acts as a mask [22–24]. In order to obtain a sieve-like porous alumina structure, the remaining aluminum is first removed from the oxide layer by incubation in saturated HgCl2 solution:

2Al þ 3HgCl2 ! 2AlCl3 þ 3Hg; followed by a removal of the pore bottom barrier oxide layer. This is achieved by selective chemical dissolution of the alumina substrate in phosphoric acid solution at 30  C. The process of pore bottom opening can be followed time-resolved by impedance spectroscopy [25]. Figure 1A shows a characteristic scanning electron microscopy (SEM) image of a porous alumina substrate with a surface porosity of 34% obtained from anodization in 0.3 M oxalic acid at 40 V. Since the pores are perfect cylinders, this is also the bulk porosity of the material. The pores are typically round or slightly hexagonally shaped as shown in the SEM image. The pore size distribution is rather narrow with a mean pore diameter of (68  3) nm (Fig. 1B).

2.2. Highly Ordered Porous Silicon Maximum pore sizes that can be achieved by the anodization of aluminum are 300 nm. For a detailed optical inspection of pore-suspending membranes, however, pore diameters in the range of the resolution of an optical microscope are required. In these cases, we utilize highly ordered macroporous silicon substrates. Macroporous silicon has been pioneered in the early 1990s by Lehmann and Fo¨ll [26]. Its formation in silicon is anisotropic and occurs preferentially along the (100), (010), and (001) direction of single crystalline silicon [27]. Very regular pore arrays with pore diameters of 0.5–7 mm with high aspect ratios can be obtained in the n-Si/HF system after photolithographic prepatterning. In detail, the start pits for the growth of the pores are etched into the n-type silicon surface using anisotropic KOH etching. Pore growth is performed in hydrofluoric acid at the anode under backside illumination to produce holes [28, 29], which are required to form SiO2 according to the following equations:

Anode:

 Si þ 2hþ þ 6F ! SiF2 6 þ 2e

Cathode:

6Hþ þ 2e ! H2 þ 4Hþ :

Pore-Suspending Membranes on Porous Alumina and Porous Silicon Substrates

63

A

100 nm

B

0.12

Frequency

0.10 0.08 0.06 0.04 0.02 0.00 40

60 80 Pore diameter/nm

100

Figure 1 (A) Scanning electron microscopy image of a porous alumina substrate. The porosity is determined to be 34% from pixel analysis. (B) Histogram analysis of the pore size distribution. The solid line is the result of a Gaussian function fit to the data with a mean pore diameter of (68  3) nm.

The photocurrent is the decisive parameter for the pore size, while the pore depth is a linear function of the anodization time. Even for differences of one order of magnitude in voltage, current or doping, the average velocity is found to be 0.5 mm/min. To obtain a membrane with through-pores, the bulk silicon on the backside of the wafer is removed by KOH etching. To guarantee chemical stability of the substrates, they are thermally oxidized in an oxygen atmosphere.

3. Formation of Micro- and Nano-BLMs In general, micro-BLMs on porous silicon and nano-BLMs on porous alumina are formed in the same way. The procedure is sketched in Fig. 2. First, the porous substrate is covered by a thin gold layer, which enables one to modify the

64

C. Steinem

Gold evaporation

Lipid-solvent droplet

Chemisorption

Thinning

Figure 2 Schematic illustration of the procedure to prepare nano- and micro-BLMs on porous substrates.

upper surface chemically by adsorbing thiol components. Alkanethiols as well as the phospholipid 1,2-dipalmitoyl-sn-glycero-3-phosphothioethanol (DPPTE) have been proven to be well suited to render the surface hydrophobic. This is a prerequisite for the establishment of micro- and nano-BLMs, which are prepared similar to the painting technique, established by Mu¨ller and Rudin [30] for the formation of freestanding BLMs and adapted by Florin and Gaub [31] for the formation of painted supported lipid membranes. In principle, the porous substrates are mounted horizontally or vertically in a Teflon cell and the cis and trans compartments are filled with electrolyte or buffer solution. The application of a small droplet of lipid dissolved in n-decane results in the formation of an insulating layer that was identified as a pore-suspending membrane, which is termed nano- and micro-BLM dependent on the size of the underlying pores.

4. Characterization of Nano- and Micro-BLMs 4.1. Electrical Characterization of Nano-BLMs Porous alumina substrates with an average pore size of 60 and 280 nm are routinely used for the preparation of nano-BLMs. By means of impedance spectroscopy, it is possible to obtain the electrical characteristics of the pore-spanning membranes. Impedance spectra obtained from the porous substrate separating the two aqueous compartments, only show the Ohmic resistance of the electrolyte. Covering the porous substrate with gold and a self-assembled monolayer on the pore rims does not alter the impedance behavior indicating that the current solely flows through the pores [25]. Any changes on top of the alumina pore columns cannot be detected. After application of the lipid-solvent droplet and the thinning process, the impedance spectrum changes considerably (Fig. 3). Control experiments, in which the porous alumina substrate was not functionalized with a hydrophobic monolayer, demonstrated that it is required for the formation of an insulating layer, as none is formed in the absence of the self-assembled monolayer. To extract membranespecific parameters from the impedance data, equivalent circuit (C ) shown in Fig. 4 is used, composed of a parallel RC element (Rm and Cm) representing the

65

1010

−80

108

−60

106

−40

104

−20

f/°

|Z|/Ω

Pore-Suspending Membranes on Porous Alumina and Porous Silicon Substrates

102

10−1

0 101

103

105

f/Hz

Figure 3 Impedance spectrum of a porous alumina substrate with a pore size of 280 nm after functionalization with DPPTE and the formation of a nano-BLM composed of DPhPC in 20 mM HEPES, 100 mM KCl, pH 7.4. The symbol (○) shows the magnitude of the impedance, while () shows the phase angle. The solid lines are the results of the fitting procedure using the equivalent circuit depicted in Fig. 4C with Cm ¼ 0.66 mF/cm2.

electrical behavior of a lipid bilayer in series to the Ohmic resistance Rel representing the electrolyte solution in the pores and bulk. For membrane resistances larger than 3 GO, Rm was not detected in the observed frequency range and thus neglected in the fitting routine. For reasons of comparison, it is widely accepted to give capacitance values in units of mF/cm2. Hence, the monitored membrane covered area needs to be known. This is not a simple task for porous substrates. Two scenarios are in principle conceivable that are represented by two equivalent circuits shown in Fig. 4. In Fig. 4A, an equivalent circuit is depicted, in which the membrane covered pore walls and pores are separated in two branches, which are not electrically connected. For this circuit, the admittance reads:

Ytotal ¼ Y1 þ Y2

ð1Þ

1 and 1=ðGm s þ ioCm s Þ þ 1=ðGox þ ioCox Þ 1 Y2 ¼ 1=ðGm p þ ioCm p Þ þ 1=Gp

ð2Þ

with

Y1 ¼

Since Gm s þ ioCm s  Gox þ ioCox [32] and Gp  Gm p þ ioCm p , it can be written:

Y1 ¼ Gox þ ioCox and Y2 ¼ Gm p þ ioCm

p

ð3Þ

66

C. Steinem

A

Gm_s

Gox Cox

Cm_s Gm_p

Gp Cm_p

Membrane B

Gm_s

Porous substrate

Gox Cox

Cm_s Gm_p

Gp Cm_p

C

Rm

Rel

Cm

Figure 4 Equivalent circuits describing the electrical properties of membranes spanning the pores of a porous substrate if (A) there is no and (B) there is electrical connection underneath the membrane. The circuit (C) represents the simplified equivalent circuit used to model the impedance data.

Moreover, since Y1  Y2, the measured admittance reduces to: A A Ytotal ¼ ðGm p þ ioCm p ÞAp

ð4Þ

A A with Gm ¼ Gm p =Ap the specific membrane conductance and Cm ¼ Cm p =Ap the specific membrane capacitance. Hence, the specific capacitance and conductance is related to the porous area. If there is electrical contact between the two branches (Fig. 4B), the situation changes leading to the following admittance:

Ytotal ¼

1 1=Y1 þ 1=Y2

ð5Þ

67

Pore-Suspending Membranes on Porous Alumina and Porous Silicon Substrates

with

Y1 ¼ Gm p þ ioCm p þ Gm s þ ioCm s and Y2 ¼ Gox þ ioCox þ Gp Gp

ð6Þ as Gox þ ioCox  Gp : Since Y1  Gp, it follows:

Ytotal ¼ Gm p þ ioCm p þ Gm s þ ioCm Ytotal ¼

s

ð7Þ

ðGm p þ ioCm p ÞAp ðGm s þ ioCm s ÞAs A A þ ¼ ðGm þ ioCm ÞðAp þ As Þ Ap As ð8Þ

A with Gm ¼ Gm p =Ap ¼ Gm s =As the specific membrane conductance and A Cm ¼ Cm p =Ap ¼ Cm s =As the specific membrane capacitance. In this case, the specific capacitance and conductance is related to the total area (Ap þ As). These two scenarios imply that either no electrical contact underneath the membranes occurs or an infinite one. Reality might be in between those two limits. However, we assume that the functionalization of the porous substrates with alkanethiols and DPPTE, respectively, results in almost no conductance underneath the membrane and hence, we relate the area-specific values to the porous area, which can be calculated from SEM images. For the impedance spectrum depicted in Fig. 2, a specific membrane capacitance of 0.66 mF/cm2 is calculated, while the membrane resistance is too high to be detected within the observed frequency range. An average capacitance of (0.65  0.14) mF/cm2 was obtained from at least 10 independent measurements. From this capacitance value, it can be assumed that single lipid bilayers have been formed across the pores [33, 34]. The membrane resistance is a measure for the stability of the bilayer system. Thus, the electrical characteristics of nano-BLMs were investigated time-resolved by means of impedance spectroscopy. It turned out that the membrane resistance decreases more or less continuously over time and differs significantly from the expected behavior of a single membrane spanning an aperture, which typically ruptures in one single step [25]. We proposed that each membrane suspending a pore might rupture individually, which implies that each membrane is decoupled from the others. To prove this hypothesis, we made use of micro-BLMs spanning the pores of a porous silicon substrate with pore sizes that allow us to observe each individual pore by an optical microscope.

4.2. Optical Characterization of the Rupturing Process of Micro-BLMs By means of fluorescence microscopy, micro-BLMs doped with Texas Red DHPE on DPPTE-functionalized porous silicon substrates can be visualized (Fig. 5) [35]. While the membranes spanning the pores fluoresce, membranes on the goldcovered pore rims appear black as the fluorescence is quenched. This quenching

68

C. Steinem

A

t=0h

B

t = 0.5 h

C

t = 1.0 h

F

t = 3.0 h

20 μm

D

t = 1.5 h

E

t = 2.5 h

G

t = 3.5 h

H

t = 4.0 h

Figure 5 Fluorescence micrographs of micro-BLMs composed of DPhPC doped with 0.1 mol% Texas Red DHPE on functionalized porous silicon immersed in 0.5 M KCl over a time period of 4 h. Membrane covered pores appear bright, while non-covered pores are black (reprinted with permission from ACS [35]).

on the pore rims enables one to monitor every single membrane covering an individual pore. Directly after the micro-BLMs have been formed, almost all pores are covered with lipid membranes (Fig. 5A). Over time, dark spots occur as observed in the left lower corner of Fig. 5B indicative of the rupturing of fluorescently labeled individual micro-BLMs, which grow in number as time progresses. Very similar results were obtained for micro-BLMs prepared on porous silicon substrates functionalized with an octadecanethiol monolayer instead of DPPTE stressing the point that the functionalization of the porous silicon substrate does not significantly influence the rupturing process of micro-BLMs. This result corroborates the notion that the continuous decrease in membrane resistance is a result of the individual rupturing of membranes spanning the pores.

4.3. Lateral Diffusion of Lipids in Micro-BLMs A prerequisite for a fully functional lipid bilayer is that the lipids diffuse laterally within the plane of the membrane. In general, artificial membranes provide an excellent opportunity to study the lateral diffusion at the molecular level. Over the past 30 years, a number of techniques have been developed and applied to determine diffusion constants of lipids in membranes, such as nuclear magnetic and electron spin resonance, fluorescence correlation spectroscopy, and fluorescence recovery after photobleaching (FRAP). To elucidate the lateral mobility of microBLMs, we performed FRAP experiments. Bleaching was accomplished by a 3-s circular laser pulse. From the recovery curve (Fig. 6), the diffusion coefficient can be extracted using the theory developed by Axelrod et al. [36] for a Gaussian intensity profile. From 10 independent measurements, an average effective diffusion

69

Normalized fluorescence intensity

Pore-Suspending Membranes on Porous Alumina and Porous Silicon Substrates

1.0 0.8 0.6 0.4 0.2 0.0 0

20

40

60

80

100

t /s

Figure 6 Normalized fluorescence recovery curve extracted from the fluorescence images after a bleaching pulse. The solid line is the result of the fitting routine according to Axelrod et al. [36] resulting in an effective diffusion coefficient of Deff ¼ (10.9  0.7) mm2/s and an immobile fraction of about 6% (reprinted with permission from ACS [35]).

coefficient of Deff ¼ (14  1) mm2/s and an immobile fraction of (15  5)% was obtained [35]. The diffusion coefficient is very similar to what has been obtained for classical BLMs [37, 38] and larger than those obtained for membranes immobilized on glass supports [39]. We refer to the determined diffusion coefficient as an effective one, since it is expected that there are basically two different diffusion coefficients on the surface, one of the freely moving lipids in the pore-suspending region and one of the lipids that are on the pore rims and experience enhanced friction due to the interaction with the functionalized solid support.

5. Applications of Nano-BLMs With respect to their electrical characteristics, nano-BLMs are very similar to classical BLMs but their stability is greatly enhanced due to the underlying porous support. Several examples will be given to demonstrate the applicability of nanoBLMs such as the investigation of the light-activated proton pump bacteriorhodopsin as well as several ion channels.

5.1. Monitoring the Activity of Bacteriorhodopsin Bacteriorhodopsin (bR) is an integral membrane protein that uses light to translocate protons across membranes and is a robust model for transporter proteins. It is found in highly ordered two-dimensional hexagonal arrays in Halobacterium salinarium, termed purple membranes [40]. Already in the 1970s, adsorption of purple membranes on black lipid membranes (BLMs) has been used to study the light-induced charge translocation in a time-resolved manner [41–43]. Even though the low specific conductance and high specific capacitance of the BLMs lead to transient

70

C. Steinem

50 nA/cm2

capacitive currents, and also stationary current if the BLM is permeabilized for protons, after activation of the ion pump, these membrane sandwiches are rather unstable. To overcome this problem, a similar membrane sandwich was produced on gold surfaces [44] with a much larger stability. However, owing to the direct coupling of the membrane to the polarizable gold electrode, the underlying capacitance prevents the establishment of a stationary current. Since nano-BLMs combine freestanding and solid supported membranes, a membrane sandwich composed of purple membrane fragments, which are attached to a nano-BLM should result in an extraordinary high long-term stable membrane system that allows the sensitive detection of light-induced proton currents of bacteriorhodopsin. Nano-BLMs composed of DPhPC doped with the positively charged detergent octadecylamine immobilized on DPPTE sub-monolayers were prepared and purple membrane fragments were electrostatically attached to them. They were illuminated through a glass window in the cis compartment, while recording the current via platinized platinum wires with a current amplifier. Light of a 250-W halogen lamp was used that was filtered with a 515-nm cut-off filter. Figure 7 shows three different current traces. The current signal shown in trace (2) is characterized by two transient currents upon switching the light on and off, respectively [45]. First, the current density rises to a maximum value Jmax-on and then decays with a characteristic decay time to a stationary current density of a few nA/cm2. A positive current means a net transport of positive charges from the cis to the trans compartment. The observed transient current is explained in terms of the light-induced net proton translocation by bR charging the underlying capacitance of the nano-BLM. The stationary current density of a few nA/cm2 indicates that even

1s

(3) (2) (1)

Figure 7 Photocurrent before (1) and 40 min after (2) addition of purple membranes to a nano-BLM. The first transient corresponds to switching the light on, the second transient to switching the light off. (3) shows the photocurrent that is detected 15 min after the addition of 40 mM of the protonophor CCCP. The arrows indicate the time point, where the light has been switched on (arrow up) and off (arrow down).

Pore-Suspending Membranes on Porous Alumina and Porous Silicon Substrates

71

without a proton ionophore there is presumably a small proton conductivity of the nano-BLM. Upon switching the light off, a second transient current density Jmax-off is observed, which decays to zero, which is the result of back diffusion of charges. To ensure that the observed current traces are not a result of a photo artifact, nanoBLMs were illuminated prior to the addition of purple membrane fragments [trace (1)] showing no significant current upon illumination. Different to what has been observed for solid supported membranes without any aqueous compartments underneath the lipid bilayer, significant stationary currents produced by bR can be observed if the proton conductance of the nano-BLM is increased by adding a membranesoluble proton ionophore such as carbonyl cyanide m-chlorophenylhydrazone (CCCP). CCCP facilitates the diffusion of protons pumped from bacteriorhodopsin toward the nano-BLM into the pores of the underlying alumina substrate resulting in a significant stationary current [trace (3)] that is stable for at least 5 min.

5.2. Monitoring Single Channel Events of Peptides and Proteins If nano-BLMs have a sufficiently high membrane resistance in the GO regime, they should be ideally suited for low noise electrical recordings of transmembrane ion currents. To demonstrate this, single channel recordings were performed on nanoand micro-BLMs [46] composed of DPhPC immobilized on DPPTE-sub-monolayers under voltage-clamp conditions and symmetrical buffer conditions. As it is well established that gramicidin ion channels can only be observed in single lipid bilayers, it is a good tool for proving the functionality of nano-BLMs. When gramicidin from Bacillus brevis was added to both, the cis and trans compartment (0.5 M KCl) of nano-BLMs on 280 nm pores, single conductance states and multiples of those could be observed [25]. A mean conductance of 60 pS applying a holding potential of 70 mV was found, which is in good accordance to what has been reported by others [47, 48]. Alamethicin from Trichoderma viride is known as a peptide that self-integrates into bilayers and forms voltage-gated ion channels of defined conductances by oligomerization. It was chosen to demonstrate that the lateral diffusion of the peptides in nano-BLMs is sufficient to form conducting helix bundles. Alamethicin monomers were successfully inserted into nano-BLMs bathed on either side in 0.5 M KCl while applying a holding potential of U ¼ þ70 mV after peptide addition to the cis side [25]. Voltage-dependent activation of single alamethicin channels with up to five distinct conductance states was observed, which are defined by the number of monomers making up the pore forming aggregate. The question arose, if not only small peptides but also large proteins can be functionally inserted into nano-BLMs. To investigate this, the outer membrane protein F (OmpF) from E. coli was added to nano-BLMs on porous alumina substrates with pore sizes of 280 and 60 nm, respectively, and the resulting current was recorded. OmpF was chosen as it is well characterized in terms of structure [49, 50] and channel activity. It is composed of 16 antiparallel aligned b-sheets (b-barrel) that are connected by loops and turns building up a water-filled pore. Three of these monomeric units with a molecular weight of 37.1 kDa [49] and a length of 5 nm [50] are arranged around a threefold molecular axis. A constriction

72

C. Steinem

zone in the center of each pore is assumed to be the decisive factor for the observed conductivity and ion selectivity [51, 52]. The insertion of the trimeric protein in the membrane gives rise to a three-step increase and decrease in current due to the stepwise opening and closing of each subunit pore [53]. Two characteristic current traces after addition of the protein in the presence of 1 M KCl, 1 mM CaCl2, pH 6.0 at a holding potential of U ¼ 100 mV is depicted in Fig. 8 indicating the insertion of a single channel. While in Fig. 8 (upper part), a current trace is shown from an experiment, in which a nano-BLM on 280 nm pores was used, Fig. 8 (lower part) shows the result obtained from nano-BLMs on 60 nm pores. Both graphs indicate the current of the closed state (C) and of the three opening levels (O1, O2, O3). Point histogram analysis of the current traces allowed for the determination of the three different conductance levels with G1 ¼ (1400  100) pS, G2 ¼ (2800  200) pS, and G3 ¼ (4350  250) pS for OmpF inserted into nano-BLMs on 280 nm pores. Slightly larger conductance values were obtained for OmpF inserted into nano-BLMs on 60 nm pores with G1 ¼ (1700  80) pS, G2 ¼ (3360  80) pS, and G3 ¼ (5060  50) pS. The results indicate that the porous substrate underneath the membrane does not prevent the full functionality of a large transmembrane protein, even though one membrane, spanning a 60-nm pore of the porous alumina substrate, is composed of only roughly 4000 lipids.

200 pA

2s O3 O2 O1 C

200 pA

2s O3

O2

O1

C

Figure 8 Characteristic channel activity of one OmpF trimer reconstituted in nano-BLMs immobilized on a porous substrate with (upper part) 280 nm large pores and (lower part) 60 nm large pores at a holding potential of U ¼ 100 mV. The current levels of the different opening states (O1, O2, O3) and of the closed state (C) are indicated.

Pore-Suspending Membranes on Porous Alumina and Porous Silicon Substrates

73

6. Solvent-Free Pore-Suspending Membranes Nano- and micro-BLMs are well suited for monitoring ion channel activities on the single channel level. However, these membranes permit protein insertion only after the bilayer has been established. If high protein densities are required, this reconstitution technique is not well suited. Another disadvantage of nano- and micro-BLMs arises from the fact that they are prepared from an organic solvent solution resulting in membranes that, even after the thinning process, where most of the solvent is removed from the bilayer, contain a residual amount of solvent. Several membrane proteins lose activity in the presence of organic solvents and hence bilayers prepared without addition of solvent are highly desirable. Already some years ago, we followed a strategy to form pore-suspending bilayers starting from unilamellar vesicles [54–56]. In this approach, the upper part of the porous substrate was covered with gold followed by chemisorption of a thiol-component to chemically distinguish between the upper surface and the inner pore walls, which should prevent the fusion process occurring within the pores (Fig. 9). When

Gold evaporation

Chemisorption

Unilamellar vesicles A

B

Figure 9 Schematic representation of the preparation procedure to obtain solvent-free poresuspending membranes from vesicle spreading. (A) Positively charged unilamellar vesicles spread and fuse to a negatively charged surface forming electrostatically attached pore-suspending bilayers. (B) Vesicles spread and fuse to form a pore-suspending membrane on a porous substrate, which is functionalized with a cholesterol derivative providing a spacer between bilayer and support.

74

C. Steinem

mercaptopropionic acid was used to render the surface negatively charged, it is possible to spread and fuse unilamellar vesicles composed of N,N-dimethyldioctadecylammonium bromide (DODAB) leading to pore-suspending membranes as observed by scanning force microscopy [55, 56]. Dependent on the load force, the membranes are more or less indented into the pores (Fig. 10) [57]. However, these membranes are rather leaky and do not form an insulating continuous layer. In a different approach, the porous substrate was functionalized with a cholesterylpolyethylenoxy thiol (CPEO3), a cholesterol derivative with a hydrophilic linker terminated with a thiol group, rendering the upper surface hydrophobic [58]. Unilamellar vesicles composed of DPhPC/1,2-dioleoyl-sn-glycero-3-phosphocholine (DOPC) (60:40) spread and fuse onto this surface leading to a pore-suspending membrane, whose electrical characteristics could be monitored by impedance spectroscopy. A characteristic impedance spectrum is depicted in Fig. 11. To extract the electrical parameters of the membrane system, again equivalent circuit (C ) depicted in Fig. 4 was used. A good accordance between model and experimental data is found with a membrane capacitance of Cm ¼ 4 109 F and a membrane resistance Rm ¼ 8 107 O. Taking the porous area into account, the absolute capacitance value translates into a mean specific capacitance of 0.4 mF/cm2, which is A

0 nm

B

100 nm

100 nm

25 nm

Figure 10 Scanning force microscopy images obtained in contact mode of pore-spanning lipid bilayers composed of DODAB at (A) low load force and (B) three times larger load force (adapted from ref. [57]). 108

−80 −60

106 −40

105

f/°

|Z|/Ω

107

−20

104

0

10−3

10−1

101

103

105

f/Hz

Figure 11 Impedance spectrum of a CPEO3-functionalized porous alumina substrate after vesicle spreading and fusion leading to an insulating pore-suspending lipid bilayer in 0.1 M NaCl. The solid lines are the results of a fitting routine using equivalent circuit (C) shown in Fig. 4 with Cm ¼ 4 109 F and Rm ¼ 8 107 O (reproduced by permission of the Royal Society of Chemistry [58]).

Pore-Suspending Membranes on Porous Alumina and Porous Silicon Substrates

75

a reasonable value for the supposed pore-suspending membranes. The membrane resistance is still lower than those obtained for nano- and micro-BLMs. However, it is already sufficiently high to monitor channel activities in an integral manner [58]. We anticipate that this membrane system will lead us to a solvent-free membrane system with the possibility to establish high protein densities, which will pave the way to electrically investigate protein pumps in the near future.

7. Conclusions Insulating pore-suspending membranes on highly ordered porous substrates bridge the gap between solid supported membranes and black lipid membranes. From the body of evidence that has been gathered over the last couple of years, it has become clear that these membranes combine the merits of solid supported membranes—that is, mechanical and long-term stability as well as the possibility to implement the system into a chip-based technology—and freestanding bilayers, which allow monitoring ion channels down to the single channel level and pumps, which cannot be achieved by solid supported membranes. Together with poresuspending membranes prepared from vesicles to allow for the integration of high protein densities, these systems will open up a new way to develop chip-based screening assays for drugs modulating ion channel and transporter protein activities.

ACKNOWLEDGMENTS I am very grateful to all, who have contributed to this work, namely J. Drexler, O. Gassmann, C. Hennesthal, C. Horn, A. Janshoff, W. Ro¨mer, S. Steltenkamp, E.K. Schmitt, and D. Weiskopf. I would like to thank the DFG, BMBF, and the Fonds der Chemischen Industrie for financial support.

REFERENCES [1] D.J. Triggle, M. Gopalakrishnan, D. Rampe, W. Zheng, Voltage-Gated Ion Channels as Drug Targets. Wiley-VCH, Weinheim, 2006. [2] B. Hille, Ion Channels of Excitable Membranes. 3rd ed., Sinauer Associates, Sunderland, MA, 2001. [3] P. Mu¨ller, D.O. Rudin, Action potentials induced in biomolecular lipid membranes, Nature 217 (1968) 713–719. [4] M. Goryll, S.J. Wilk, G.M. Laws, T.J. Thornton, S.M. Goodnick, M. Saranti, J. Tang, R.S. Eisenberg, Silicon-based ion channel sensor, Superlatt. Microstruct. 34 (2003) 451–457. [5] S.J. Wilk, M. Goryll, G.M. Laws, S.M. Goodnick, T.J. Thornton, M. Saranti, J. Tang, R.S. Eisenberg, Teflon-coated silicon apertures for supported lipid bilayer membranes, Appl. Phys. Lett. 85 (2004) 3307–3309. [6] C. Schmidt, M. Mayer, H. Vogel, A chip-based biosensor for the functional analysis of single ion channels, Angew. Chem. Int. Ed. 39 (2000) 3137–3140. [7] Y. Cheng, R.J. Bushby, S.D. Evans, P.F. Knowles, R.E. Miles, S.D. Ogier, Single ion channel sensitivity in suspended bilayers on micromachined supports, Langmuir 17 (2001) 1240–1242. [8] Y. Cheng, S.D. Ogier, R.J. Bushby, S.D. Evans, Discrete membrane arrays, Mol. Biotechnol. 74 (2000) 159–174.

76

C. Steinem

[9] S.D. Ogier, R.J. Bushby, Y. Cheng, S.D. Evans, S.W. Evans, T.A. Jenkings, P.F. Knowles, R.E. Miles, Suspended planar phospholipid bilayers on micromachined supports, Langmuir 16 (2000) 5696–5701. [10] N. Fertig, M. Klau, M. George, R.H. Blick, J.C. Behrends, Activity of single ion channel proteins detected with a planar microstructure, Appl. Phys. Lett. 81 (2002) 4865–4867. [11] N. Fertig, C. Meyer, R.H. Blick, C. Trautmann, J.C. Behrends, Microstructured glass chip for ion-channel electrophysiology, Phys. Rev. E 64 (2001) 040901(R). [12] R. Hemmler, G. Bo¨se, R. Wagner, R. Peters, Nanopore unitary permeability measure by electrochemical and optical single transporter recording, Biophys. J. 88 (2005) 4000–4007. [13] G. Favero, L. Capanella, A. D’Annibale, R. Santucci, T. Ferri, Mixed hybrid bilayer lipid membrane incorporating valinomycin: Improvements in preparation and function, Microchem. J. 74 (2003) 141–148. [14] G. Favero, A. D’Annibale, L. Capanella, R. Santucci, T. Ferri, Membrane supported bilayer lipid membrane array: Preparation, stability and ion-channel insertion, Anal. Chim. Acta 460 (2002) 23–34. [15] F. Keller, M.S. Hunter, D.L. Robinson, Structural features of oxide coatings on aluminum, J. Electrochem. Soc. 100 (1953) 411–419. [16] O. Jessensky, F. Mu¨ller, U. Go¨sele, Self-organized formation of hexagonal pore arrays in anodic alumina, Appl. Phys. Lett. 72 (1998) 1173–1175. [17] A.P. Li, F. Mu¨ller, A. Birner, K. Nielsch, U. Go¨sele, Hexagonal pore arrays with a 50–420 nm interpore distance formed by self-organization in anodic alumina, J. Appl. Phys. 84 (1998) 6023–6026. [18] O. Jessensky, F. Mu¨ller, U. Go¨sele, Self-organized formation of hexagonal pore structures in anodic alumina, J. Electrochem. Soc. 145 (1998) 3735–3740. [19] K. Nielsch, J. Choi, K. Schwirn, R.B. Wehrspohn, U. Go¨sele, Self-ordering regimes of porous alumina: The 10% porosity rule, Nano Lett. 2 (2002) 677–680. [20] H. Masuda, K. Fukuda, Ordered metal nanohole arrays made by a two-step replication of honeycomb structures of anodic alumina, Science 268 (1995) 1466–1468. [21] V.P. Parkhutik, V.I. Shershulsky, Theoretical modeling of porous oxide growth on aluminium, J. Phys. D: Appl. Phys. 25 (1992) 1258–1263. [22] A.P. Li, F. Mu¨ller, A. Birner, K. Nielsch, U. Go¨sele, Polycrystalline nanopore arrays with hexagonal ordering on aluminium, J. Vac. Sci. Technol. A 17 (1999) 1428–1431. [23] A.P. Li, F. Mu¨ller, A. Birner, K. Nielsch, U. Go¨sele, Fabrication and microstructuring of hexagonally ordered two-dimensional nanopore arrays in anodic alumina, Adv. Mater. 11 (1999) 483–487. [24] H. Masuda, F. Hasegwa, S. Ono, Self-ordering of cell arrangement of anodic porous alumina formed in sulfuric acid solution, J. Electrochem. Soc. 144 (1997) L127–L130. [25] W. Ro¨mer, C. Steinem, Impedance analysis and single-channel recordings on nano-black lipid membranes based on porous alumina, Biophys. J. 86 (2004) 955–965. [26] V. Lehmann, H. Fo¨ll, Formation mechanism and properties of electrochemically etched trenches in n-type silicon, J. Electrochem. Soc. 137 (1990) 653–658. [27] S. Licht, Semiconductor electrodes and photoelectrochemistry, in: A.J. Bard, M. Stratmann (Eds.), Encyclopedia of Electrochemistry, Wiley-VCH, Weinheim, 2002. [28] M.H. Klu¨hr, A. Sauermann, C.A. Elsner, K.H. Thein, S.K. Dertinger, Partially oxidized macroporous silicon: A three-dimensional photonic matrix for microarray applications, Adv. Mater. 23 (2006) 3135–3139. [29] V. Lehmann, The physics of macropore formation in low doped n-type silicon, J. Electrochem. Soc. 140 (1993) 2836–2843. [30] P. Mu¨ller, D.O. Rudin, H.T. Tien, W.C. Wescott, Methods for the formation of single bimolecular lipid membranes in aqueous solutions, J. Phys. Chem. 67 (1963) 534–535. [31] E.L. Florin, H.E. Gaub, Painted supported lipid membranes, Biophys. J. 64 (1993) 375–383. [32] T.P. Hoar, G.C. Wood, The sealing of porous anodic oxide films on aluminium, Electrochim. Acta 7 (1962) 333–353. [33] R. Benz, O. Fro¨hlich, P. La¨uger, M. Montal, Electrical capacity of black lipid films and of lipid bilayers made from monolayers, Biochim. Biophys. Acta 394 (1975) 323–334.

Pore-Suspending Membranes on Porous Alumina and Porous Silicon Substrates

77

[34] H.T. Tien, A.L. Ottova, Membrane Biophysics: Planar Lipid Bilayers and Spherical Liposomes. Elsevier, Amsterdam and New York, 2000. [35] D. Weiskopf, E.K. Schmitt, M.H. Klu¨hr, S.K. Dertinger, C. Steinem, Micro-BLMs on highly ordered porous silicon substrates: rupture process and lateral mobility, Langmuir 23 (2007) 9134–9139. [36] D. Axelrod, D.E. Koppel, J. Schlessinger, E. Elston, W.W. Webb, Mobility measurement by analysis of fluorescence photobleaching recovery kinetics, Biophys. J. 16 (1976) 1055–1069. [37] S. Ladha, A.R. Mackie, L.J. Harvey, D.C. Clark, E. J. A. Lea, M. Brullemans, H. Duclohier, Lateral diffusion in planar lipid bilayers. A FRAP investigation of its modulation by lipid composition, cholesterol or alamethicin content and divalent cations, Biophys. J. 71 (1996) 1364–1373. [38] Z.I. Lalchev, A.R. Mackie, Molecular lateral diffusion in model membrane systems, Coll. Surf. B 15 (1999) 147–160. [39] J.T. Groves, S.G. Boxer, Electric field-induced concentration gradients in planar supported bilayers, Biophys. J. 69 (1995) 1972–1975. [40] D. Oesterhelt, W. Stoeckenius, Functions of a new photoreceptor membrane, Proc. Nat. Acad. Sci. USA 70 (1973) 2853–2857. [41] E. Bamberg, H.-J. Apell, N.A. Dencher, W. Sperling, H. Stieve, P. La¨uger, Photocurrents generated by bacteriorhodopsin on planar bilayer membranes, Biophys. Struct. Mech. 5 (1979) 277–292. [42] Z. Danzshazy, B. Karvaly, Incorporation of bacteriorhodopsin into a bilayer lipid membrane: A photoelectric-spectroscopic study, FEBS Lett. 72 (1976) 136–138. [43] L.A. Drachev, A.A. Jasaitis, A.D. Kaulen, A.A. Kondrashin, E.A. Liberman, I.B. Nemecek, S.A. Ostroumov, A.Y. Semenov, V.P. Skulachev, Direct measurement of electric current generation by cytochrome oxidase, Hþ-ATPase and bacteriorhodopsin, Nature 249 (1974) 321–324. [44] K. Seifert, K. Fendler, E. Bamberg, Charge transport by ion translocating membrane proteins on solid supported membranes, Biophys. J. 64 (1993) 384–391. [45] C. Horn, C. Steinem, Photocurrents generated by bacteriorhodopsin adsorbed on nano-black lipid membranes, Biophys. J. 89 (2005) 1046–1054. [46] W. Ro¨mer, Y.H. Lam, D. Fischer, A. Watts, W.B. Fischer, P. Go¨ring, R.B. Wehrspohn, U. Go¨sele, C. Steinem, Channel activity of a viral transmembrane peptide in micro-BLMs: Vpu1–32 from HIV-1, J. Am. Chem. Soc. 126 (2004) 16267–16274. [47] O.S. Andersen, Ion movement through gramicidin A channels, Biophys. J. 41 (1983) 119–133. [48] E. Bamberg, K. Noda, E. Gross, P. La¨uger, Single-channel parameters of gramicidin A, B and C, Biochim. Biophys. Acta 419 (1976) 223–228. [49] L.K. Buehler, S. Kusumoto, H. Zhang, J.P. Rosenbusch, Plasticity of Escherichia coli porin channels. Dependence of their conductance on strain and lipid environment, J. Biol. Chem. 266 (1991) 24446–24450. [50] S.W. Cowan, T. Schirmer, G. Rummel, M. Steiert, R. Ghosh, R.A. Pauptit, J.N. Jansonius, J.P. Rosenbusch, Crystal structures explain functional properties of two E. coli porins, Nature 358 (1992) 727–733. [51] J. Bredin, N. Saint, M. Mallea, E. De, G. Molle, J.M. Pages, V. Simonet, Alteration of pore properties of Escherichia coli OmpF induced by mutation of key residues in anti-loop 3 region, Biochem. J. 363 (2002) 521–528. [52] H. Miedema, A. Meter-Arkema, J. Wierenga, J. Tang, B. Eisenberg, W. Nonner, H. Hektor, D. Gillespie, W. Meijberg, Permeation properties of an engineered bacterial OmpF porin containing the EEEE-locus of Ca2þ channels, Biophys. J. 87 (2004) 3137–3147. [53] E.K. Schmitt, C. Steinem, Channel activity of OmpF monitored in nano-BLMs, Biophys. J. 91 (2006) 2163–2171. [54] J. Drexler, C. Steinem, Pore-suspending lipid bilayers on porous alumina investigated by electrical impedance spectroscopy, J. Phys. Chem. B 107 (2003) 11245–11254. [55] C. Hennesthal, J. Drexler, C. Steinem, Membrane-suspended nanocompartments based on ordered pores in alumina, Chem. Phys. Chem. 3 (2002) 885–889. [56] C. Hennesthal, C. Steinem, Pore-spanning lipid bilayers visualized by scanning force microscopy, J. Am. Chem. Soc. 122 (2000) 8085–8086.

78

C. Steinem

[57] S. Steltenkamp, M. Mu¨ller, M. Deserno, C. Hennesthal, C. Steinem, A. Janshoff, Mechanical properties of pore-suspending membranes probed by scanning force microscopy, Biophys. J. 91 (2006) 217–226. [58] E.K. Schmitt, M. Nurnabi, R.J. Bushby, C. Steinem, Insulating pore-suspending membranes on highly ordered porous substrates obtained from vesicle spreading. Soft Matter 4 (2008) 250–253.

C H A P T E R

F O U R

Antiphospholipid Syndrome: Mechanisms Revealed in Erythrocyte and Liposome Studies Snezˇna Sodin-Sˇemrl,1,* Blazˇ Rozman,1 Alesˇ Iglicˇ,2 and Veronika Kralj-Iglicˇ3 Contents 1. Introduction 2. Biochemical Aspects of Mechanisms Involved in APS 2.1. Exposure of Negatively Charged Membrane Surfaces due to Loss of Membrane Asymmetry and Its Role in APS 2.2. Cardiolipin 2.3. Phosphatidylserine 2.4. Protein Cofactors and Their Antibodies 3. Cellular Microexovesicles (Microparticles) in Thrombosis and Hemostasis with Special Emphasis on APS 3.1. Mechanisms Leading to Microvesiculation 3.2. Effect of b2GPI and aPL on Budding of Phospholipid Vesicles 4. Coalescence of Membranes Caused by b2GPI and aPL 4.1. Experimental Evidence on the Effect of b2GPI and aPL on Coalescence of Phospholipid Vesicles 4.2. Theoretical Description of the Coalescence of Membranes: Interaction Between Charged Surfaces Mediated by Ions with Dimeric Distribution of Charge 5. The Effects of ANXA5, b2GPI and aPL on Phospholipid Membranes 6. Conclusions Acknowledgments References

80 81 81 82 83 84 88 89 97 99 100

101 111 113 113 114

Abstract The loss of membrane asymmetry is an important feature in mechanisms involved in maintaining immunity. Negatively charged phospholipids are present only in small amounts in the outer leaflet of the membranes of viable cells. In some pathological conditions and in apoptosis, negatively charged phospholipids become exposed on the Corresponding author. Tel.: þ386 1 5225596; Fax: þ386 1 5225598; E-mail address: [email protected] (S. Sodin-Semrl).

* 1 2 3

Department of Rheumatology, University Medical Center, Vodnikova 62, SI-1000 Ljubljana, Slovenia Laboratory of Physics, Faculty of Electrical Engineering, University of Ljubljana, Trzˇasˇka 25, SI-1000 Ljubljana, Slovenia Laboratory of Clinical Biophysics, Faculty of Medicine, University of Ljubljana, Lipicˇeva 2, SI-1000 Ljubljana, Slovenia

Advances in Planar Lipid Bilayers and Liposomes, Volume 7 ISSN 1554-4516, DOI: 10.1016/S1554-4516(08)00004-5

#

2008 Elsevier Inc. All rights reserved.

79

S. Sodin-Sˇemrl et al.

80

outer membrane surface facing among other surroundings, the blood plasma. There, they can contribute to the catalytic potential for chemical reactions leading to blood clot formation. Also, they serve, in connection with protein cofactors, as antigens in the immune system leading to autoantibody formation and autoimmune diseases, among them of special emphasis, antiphospholipid syndrome (APS). APS is manifested clinically by venous and arterial thromboses and fetal loss. Laboratory analysis reveals a presence of antiphospholipid antibodies (aPL) in the patients’ blood and research analysis indicates an increased amount of cell-derived microvesicles (MVs). MVs often have an impaired membrane asymmetry and can be procoagulant. Elevation of MV quantity in patient sera from different cells has been shown to be associated with thromboses. It is indicated that MVs can also be antigenic, depending on the properties and content of the blood plasma, that is, the presence in plasma of certain proteins called protein cofactors. Although numerous methods have been used in attempt to understand the antigen–phospholipid–antibody interactions and their role in APS, knowledge of the underlying mechanisms remains fragmentary. In this chapter, we study complex interactions between protein cofactors, phospholipid membranes, and aPL in giant phospholipid vesicles and in erythrocytes. We consider the process of budding and vesiculation in giant phospholipid vesicles and in erythrocytes, plasma protein-mediated attraction of negatively charged phospholipid membranes, and effects of annexin A5 (ANXA5) on interaction of protein cofactors and aPL with the phospholipid membranes. Our results indicate that phospholipid vesicles and erythrocytes can be useful not only in elucidating basic mechanisms involved in APS but can serve as models of antigen–phospholipid–antibody interactions with potential for improving clinical diagnosis and/or prognosis of this debilitating and sometimes lethal autoimmune disorder. Further studies in this direction should be undertaken.

1. Introduction Antiphospholipid syndrome (APS) is a rather new clinical entity known since 1983 when definite recognition of APS started with the introduction of sensitive solid phase assays of anticardiolipin antibodies (aCL) [1]. Clinical criteria encompass vascular thrombosis and pregnancy morbidity. Laboratory criteria include the presence of lupus anticoagulants (LA) in plasma and antiphospholipid antibodies (aPL) in serum and plasma. Early studies of the mechanisms involved in APS focused on the interactions between phospholipids and antibodies, until the 1990s, when the first protein cofactor, that is, b2-glycoprotein I (b2GPI) was determined as antigenic [2–4]. It is now acknowledged that the antigenic targets of aPL in ELISA tests are not the phospholipids themselves, but plasma protein cofactors, such as b2GPI, prothrombin (PT), and annexin A5 (ANXA5), interacting with phospholipid surfaces [5–7]. One of the processes underlying both thrombosis and the autoimmune response is microexovesiculation of the cell membrane which is the final event in the process of membrane budding. Microexovesicles (also called microparticles) enter the circulation and may convey material and information to distant cells. They have been recognized to play an important role in thrombosis, inflammation, and promotion of cancer [8–11]. In particular, increased levels of MVs were found in patients with APS [12].

Antiphospholipid Syndrome

81

In spite of numerous biochemical studies, a decisive answer concerning the role of protein cofactors and antiphosphoplipid antibodies in the pathogenesis of APS has not yet been given. The study of physical mechanisms should be undertaken to complement the study of biochemical aspects. In this regard, microvesiculation represents a promising subject worthy to be further explored. It is of particular interest to study the role of phospholipids, protein cofactors, and antibodies in the process of membrane vesiculation and propose possible mechanisms of their role in related clinical manifestations. There is a continuous search for appropriate in vitro models of biological cells. Erythrocytes and giant phospholipid vesicles have been proven useful due to lack of internal structure and sufficient size for observation under the light microscope. In these systems, the budding and vesiculation of the membrane can be induced by different trigger mechanisms in the presence or absence of the protein cofactors and antibodies in the surrounding solution. We believe that elucidating these processes would help in understanding protein–protein and protein–lipid interactions underlying the clinical manifestations of thrombosis and/or fetal loss in APS patients.

2. Biochemical Aspects of Mechanisms Involved in APS In this section, we describe the key factors involved in APS and their possible roles in corresponding biochemical mechanisms, in particular in interaction with membranes including negatively charged phospholipids. The most thoroughly studied molecules involved in APS are phospholipids [cardiolipin (CL) and phosphatidylserine (PS)], protein cofactors (b2GPI, PT, and ANXA5), and their antibodies.

2.1. Exposure of Negatively Charged Membrane Surfaces due to Loss of Membrane Asymmetry and Its Role in APS One of the major mechanisms implicated in the development of immunogenicity against phospholipids such as CL, PS, and their protein cofactors is the loss of membrane phospholipid asymmetry. The outer membrane leaflet of eukaryotic cells is formed predominantly with cholinephospholipids (such as phosphatidylcholine, sphingomyelin, and glycosphingolipids) whereas the inner (cytoplasmic) leaflet is composed predominantly of aminophospholipids such as phosphatidylethanolamine [13]. Other phospholipids of minor abundance, such as phosphatidic acid and phosphatidylinositol are located in the inner leaflet. Human erythrocytes normally comprise of less than 4% PS in the outer leaflet of the membrane [14], whereas human platelets have a slightly higher PS content [14, 15]. Virtually every cell in the body restricts PS to the inner leaflet of the plasma membrane by an energydependent transport from the outer to the inner leaflet of the bilayer [16]. The loss of transmembrane phospholipid asymmetry, with consequent exposure of PS in the external monolayer occurs both in normal and pathologic conditions. PS distributes to the outer leaflet during cell activation and apoptosis involving flippase,

82

S. Sodin-Sˇemrl et al.

floppase, and scramblase enzymes [17]. The flippase enzyme activity is highly selective for PS and functions to keep this lipid sequestered from the cell surface [18]. Loss of this asymmetry is associated with many physiological states, in addition to apoptosis, such as cell activation, clearance, and senescence, as well as with pathological conditions such as thrombosis and tumorigenesis [19, 20]. The translocation of CL, which is normally located on the inner mitochondrial membrane, to plasma membranes (as is the case with PS) also appears to occur in apoptosis. While the process of PS exposure can occur within seconds, the exposure of CL is slower as it must traverse multiple membranes.

2.2. Cardiolipin CL is a phospholipid located predominantly in energy transducing membranes such as mammalian inner mitochondrial membranes where it plays a role in many multimeric complexes associated with these membranes. CL differs from all other phospholipids in that it has four acyl chains and a dimeric headgroup which carries at physiologic pH two negative charges. CL rapid remodeling into highly unsaturated species (tetra-linoleyl-CL) most commonly found in adult tissues, involves relocation to other membranes (outer mitochondrial membrane and extra-mitochondrial membranes) [21]. CL and its metabolites, such as monolyso-CL (MCL), dilyso-CL (DCL), and hydro-CL move from mitochondria to other cellular membranes during death receptor-mediated apoptosis where CL exhibits segregation. The chemistry of CL acyl chains is important for both the binding to b2GPI as well as for the intrinsic immunogenicity of the CL molecule [21]. The CL degradation process occurs in both, membranes of the mitochondria as part of remodeling mature lipids, and in lysosomes as well. CL is one of the most resistant phospholipids (against phospholipase A2 hydrolysis) in membranes of healthy cells. Some MCL exist also in healthy tissue; however, their abundance increases significantly in response to apoptotic stimuli. b2GPI shows a differential binding to CL derivatives (with highest binding to MCL > CL > DCL). The degree of unsaturation of acyl chains in CL influences the binding of b2GPI to CL and (hydro)peroxidation of CL is essential for enhancing the binding of aPL [22]. CL hydrolysis of one fatty acid chain could yield MCL to enhance CL antigenicity. MCL has hybrid properties between a diacyl-lipid (bilayer forming) and a lyso-lipid (micelle forming) form, which also may facilitate autoantibody reactivity [21]. The loss of an acyl chain in DCL and an unusually large and hydrophilic polar head could destroy the binding affinity of this CL metabolite for b2GPI. It is therefore indicated that probable links exist between apoptotic changes, oxidative states, CL metabolism, generation of CL derivatives, and the recognition and production of aCL antibodies. Leakage of CL and its exposure on cell surfaces precede DNA fragmentation, so CL localization to the plasma membrane represents an early event of apoptosis and may be a trigger of aCL development [23]. At physiologic Caþþ cytoplasmic concentrations, both the translocase (flippase) and floppase are active generating and maintaining phospholipid asymmetry. The lipid scramblase activity, however, can lead to the collapse of the membrane asymmetry, especially in conditions of high Caþþ when the scramblase is activated

Antiphospholipid Syndrome

83

and the actions of translocase (flippase) and floppase are blocked, a condition leading to loss of asymmetry and having the consequence of cells being primed for pathological events, such as autoimmune antibody development. Subang et al. [24] confirmed this hypothesis by showing that b2GPI bound to anionic phospholipids (that may be expressed on the surface of apoptotic cells), was immunogenic. The group immunized BALB/C mice with b2GPI in the presence of CL or PS vesicles. CL vesicles induced the highest levels of anti-b2GPI and aCL IgG antibodies. Alessandri et al. [21] demonstrated that a critical number of acyl chains in CL derivates is important for binding of aPL and that MCL is one of the antigenic targets with immunoreactivity comparable to CL in APS and systemic lupus erythematosus. IgG fractions from APS patients were used to analyze the distribution pattern of CL and its derivatives on the surface of apoptotic endothelial cells (HUVEC). Stronger binding of IgG fractions were observed to apoptotic HUVEC than to nonapoptotic, untreated cells. b2GPI has a strong affinity for PS, phosphatidic acid, and CL and has been previously reported not to bind to neutral phospholipids, such as PC, phosphatidylethanolamine, and sphingomyelin [21]. 2.2.1. Antibody characterization It is generally agreed that the term aCL, if not stated otherwise, defines the antibodies against CL detected by the classical aCL ELISA as both, b2GPI independent as well as b2GPI dependent. Even though this method is standardized [25], there is variation between the tests. 2.2.2. Clinical significance aCL as detected by the classical ELISA are not only believed to be markers of thrombosis occurring in APS and other autoimmune diseases, but also occasionally show an association with other diseases, especially infections. Infectious aCL usually disappear from patient sera in less than three months.

2.3. Phosphatidylserine A critical role for PS in thrombosis was suggested by Bevers et al. [26, 27] who reported that the asymmetric orientation of phospholipids in blood platelets was rapidly lost during their activation following an influx of Caþþ. This points to the involvement of the enzyme lipid scramblase, which requires a continuous presence of cytoplasmic calcium, in moving all classes of lipids bidirectionally. Without the action of lipid scramblase, lipids of the platelet plasma membrane would flip-flop resulting in a loss of membrane asymmetry within minutes. When proteins fractionated from such platelet membranes were reconstituted into artificial lipid vesicles, they exhibited Caþþ-dependent lipid scrambling activity that was pronase-, heat-, and sulfhydryl sensitive [28]. Other consequences of PS presence in the outer leaflet of membranes include among others: formation of a procoagulant surface following platelet activation [14], early detection of apoptosis [16], and high adherence of damaged erythrocytes (sickled cells or ghosts) to endothelial cells. There was an accelerated clearance of damaged cells by monocytes and macrophages. This mechanism of recognition could be responsible for the elimination of aged erythrocytes

84

S. Sodin-Sˇemrl et al.

from the bloodstream. Failure to efficiently remove apoptotic cells may contribute to inflammatory responses and autoimmune diseases resulting from chronic exposure of PS. It is interesting that non-apoptotic PS externalization is induced by several activation stimuli, including engagement of immunoreceptors [29]. Unregulated loss of PS asymmetry may contribute significantly to heart disease and stroke [16]. 2.3.1. Antibody characterization Anti-PS antibodies (anti-PS) have been mainly studied in connection with different protein cofactors, such as PT and b2GPI. 2.3.2. Clinical significance In autoimmune diseases, anti-PS can sometimes be detected independently from aCL [30]. It has also been indicated that anti-PS destroy human trophoblasts, halt human chorionic gonadotropin production, and limit trophoblast invasion [31].

2.4. Protein Cofactors and Their Antibodies 2.4.1. b2-Glycoprotein I b2GPI is a 50-kDa protein as estimated by denaturing polyacrylamide gel electrophoresis and represents a major autoantigen for the production of aPL shown to be involved in autoimmune diseases, such as APS. The human b2GPI 326 amino acid sequence was determined from purified protein [32] and the entire single chain polypeptide has extensive internal homology within its five consecutive 60 amino acid homologous sushi domains. The crystal structure of b2GPI has been resolved and reveals a J-shaped structure with four aligned segments and a fifth structurally different segment with an additional hydrophobic lysine rich loop thought to insert into the phospholipid membrane [33]. The lysine rich loop, located between Cys 281-Cys 288 of the fifth b2GPI domain [34], is held responsible for electrostatic interactions between the positively charged patch in domain five and anionic phospholipid headgroups [35]. Insertion of this lysine rich loop into the phospholipid membrane was found to be necessary for vesicles and cell membrane fragments to be cleared from the circulation [36]. The three-dimensional structure of a typical domain has a hydrophobic core containing conserved residues bordered on either side by small anti-parallel b-pleated sheets. Each domain has two disulfide bonds, with the exception of the fifth which has three. b2GPI is highly glycosylated, with five attached glucosamine-containing oligosaccharides, located in domains III and IV, the variability of which is thought to account for at least five isoforms as judged by isoelectric focusing [37]. Glycosylation influences the conformation and antibody binding, but does not affect the b2GPI binding capacity to lipid membranes [38, 39]. b2GPI is presumed to undergo the sorting and targeting transit through the Golgi apparatus after initial synthesis in the rough endoplasmic reticulum. There are no studies to date describing b2GPI mutation effects on localization, or degradation. It has been shown that b2GPI also plays a role in the modulation of triacylglycerolrich lipoprotein metabolism [40], atherogenesis [41], and atherosclerosis [42, 43]. In vitro studies have shown b2GPI to be an anti-coagulant factor involved in mechanisms such as inhibition of the coagulation pathway [44], platelet prothrombinase

Antiphospholipid Syndrome

85

activity [45], and platelet aggregation [46], all of which propose a protective role in the pathogenesis of thrombosis. There have been reports, however, describing the role of b2GPI in potentially pro-coagulant activities such as inhibition of activated protein C anti-coagulant activity [47] and its ability to bind monocyte surface and promote tissue factor expression [48]. This latter binding required the presence of aPL. Putative receptors for b2GPI have been identified in HUVEC [49, 50], renal epithelial cells [51], and platelets [52]. When macrophages fail to maintain their membrane lipid asymmetry, b2GPI has been shown to bind PS [53]. The binding of b2GPI to HUVEC is of high affinity (Kd ¼ 18 nM) and has been shown to be specifically mediated by annexin 2 [49]. Endothelial cell activation by anti-b2GPI antibodies has been indicated to be due to binding with b2GPI likely associated with one or more toll-like receptors [50] causing proadhesive and proinflammatory responses. In the kidneys, b2GPI has been suggested to be a cell recognition mediator for PS exposing particles. An efficient and physiologically relevant receptor for b2GPI has been identified in renal epithelial cells to be megalin, which following interaction with b2GPI, mediates endocytosis and renal clearance. b2GPI bound to PS and CL shows an increased affinity for megalin [51]. Recently, an additional receptor has been identified as apolipoprotein E receptor 20 which has been shown to be involved in activation of platelets by dimeric b2GPI [52]. 2.4.1.1. Antibody characterization b2GPI is known to bind not only to negatively charged phospholipids in artificial systems but also to surface membranes of cells directly involved in the pathogenic mechanism of APS, such as activated or apoptosing cells (platelets, endothelial cells). There is a lack of information which would unite crucial data necessary in order to determine whether negatively charged phospholipids are essential for both activation and apoptotic signaling to occur. The proposed mechanism of platelet activation is dependent on the dimerization of b2GPI by aPL. This binding stabilizes the complex on the surface of platelets leading to a stronger adhesion of b2GPI-aPL to phospholipids [52]. b2GPI binding to HUVEC cells lead to exposed epitopes which can be recognized by circulating anti-b2GPI antibodies, which stimulate induction of cell activation, triggering prothrombotic events, contributing to the hypercoagulable state of APS patients [54]. The proposed function and consequence of aPL binding to phospholipid bound b2GPI on either monocytes or endothelial cells elicits p38 MAP kinase phosphorylation, leading to translocalization of NF-kB into the nucleus and transcription of procoagulant factors, such as tissue factor, plasminogen activator inhibitor-1, tumor necrosis factor alpha, and endothelin-1 [55, 56]. b2GPI also binds to sulfatides (acidic glycosphingolipids with sulfate esters on oligosaccharide chains) at physiological concentrations and these complexes are recognized by antibodies from patients with APS [57]. Anti-b2GPI either recognize conformational and cryptic epitopes on b2GPI after binding to phospholipid membranes or bivalently bind to b2GPI molecules in close proximity [5]. Anti-b2GPI are generally believed to be of low avidity and bind only to high density of b2GPI molecules on phospholipid membranes. Recently it was shown that neither high density of the antigen nor high avidity of the antibodies (or Fab fragments) alone was sufficient for the binding of anti-b2GPI to its antigen.

86

S. Sodin-Sˇemrl et al.

Some conformational modifications and consequently, exposed neoepitopes are required for the recognition of b2GPI by polyclonal anti-b2GPI [58]. 2.4.1.2. Clinical significance Anti-b2GPI antibodies are an independent risk factor for thrombosis [59, 60] and pregnancy complications [61, 62], even though some studies deny these associations mainly because of methodological differences and lack of standardization [63]. In meta-analyses of 20 studies, anti-b2GPI antibodies seemed to be more often associated with venous thrombosis as compared to arterial events [59].

2.4.2. Prothrombin PT is an approximately 72 kDa glycoprotein synthesized in the liver, secreted into circulation and present at 100 mg/ml concentrations in normal plasma. Posttranslationally PT undergoes carboxylation of its glutamic residues, known as the GLA domain, which is essential for the Caþþ-dependent binding of phospholipid to PT. The GLA domain is followed by the two Kringle domains responsible for interaction with substrate–cofactor receptors [64, 65]. PT is unique among the activated coagulation proteinases in that it completely loses the domains important for initial recognition interactions when it is activated to its serine proteinase derivative, thrombin [66]. PT is activated at sites of vascular injury to yield its active form thrombin. Thrombin is generated by cleavage to yield a F1 þ 2 fragment containing GLA and two Kringle domains, which originally function to localize PT on membrane surfaces as part of the activation complex, that includes factor Va and factor Xa. Because it loses its F1 þ 2 fragment, thrombin’s determinants for substrate recognition remain on its catalytic domain. Thrombin is one of the most studied major enzymes involved in many stages of blood coagulation and platelet aggregation, fibrinogen cleavage, and fibrin clot formation. Antibodies against PT are postulated to dysregulate some of these processes [66]. 2.4.2.1. Antibody characterization There are two types of anti-PT antibodies able to be detected. anti-PT can be evaluated by utilizing either human PT directly coated onto ELISA plates or complexed with PS, neither of which is standardized. anti-PT bind to PT coated on irradiated or high-activated polyvinylchloride, but not on plain polystyrene ELISA plates. IgG and/or IgM antibodies to human PT in solid phase have been reported in approximately half of the patients with aPL. PT is recognized more efficiently when the protein is bound to PS-coated ELISA plates using calcium ions. Multiple explanations could account for this. The first one considers that PT complexed to PS is not likely to be restricted in its lateral movements; this would allow clustering and proper orientation, offering better binding conditions for the antibodies. The second alternative considers that the circulating PT-anti-PT immune complexes present in some samples may be captured on ELISA with PS. The third alternative is that anti-PT might react with neoepitopes that PT makes available only when bound to PS through calcium ions [67]. Recent evidence from Rauch et al. [68] indicates that anti-PT recognize PT also when bound to hexagonal phase phosphatidylethanolamine and that the plasma LA activity is

Antiphospholipid Syndrome

87

specifically neutralized by the prothrombin/hexagonal phase phosphatidylethanolamine complex [68]. 2.4.2.2. Clinical significance Galli et al. presented eight studies with multivariate analyses: two of those indicated that anti-PT were independent risk factors for thrombosis and three other similar studies showed that anti-PT added to the risk borne by LA or aCL. In this widespread analysis report, 37% of anti-PT and thrombosis associations were significant: only 3 out of 11 associations were significantly correlated to arterial thrombosis and 7 out of 18 with venous thrombosis. In conclusion, no clear association with thrombosis was found for anti-PT, irrespective of isotype, site, and type of event [59].

2.4.3. Annexin A5 ANXA5 is an anionic phospholipid-binding protein with an apparent molecular weight of 36 kDa. Based on its activity to interact specifically with acidic phospholipids, ANXA5 is used widely as a reagent to detect PS on the surface of apoptotic cells. As a radiolabeled probe, it is thus employed to detect apoptotic changes in cells or tissues of live animals. Its physiological roles in cells include inhibition of phospholipase A2, acting as a calcium ion channel, binding collagens and potent anti-coagulant activity by protecting vascular endothelium from aPL [69]. ANXA5 requires rather high calcium concentrations (in the 10-mM range) for phospholipid binding [70]. Regardless of the site of action, that is, the cytoplasmic or exoplasmic leaflets of cell membranes, the mechanistic basis of ANXA5 activity is mostly linked to its ability to form two-dimensional crystals on planar lipid bilayers [71]. ANXA5 forms this antithrombotic shield around procoagulant anionic phospholipids, which blocks their participation in phospholipid-dependent coagulation reactions. Without the shield, there is a net increase in the quantity of anionic phospholipids on cell membranes that are available to accelerate coagulation reactions. The formation of this protective shield is Caþþ dependent. Calcium has also been shown to promote the binding of ANXA5 to wells coated with a mixture of phospholipids including PS or PC. It was found that PS promoted binding of ANXA5 to the wells in presence of Caþþ [72]. It has been recently demonstrated that such a crystalline protein network on cell membranes opens a new portal for entry into the cell [73]. This network, specific for ANXA5, elicits budding, endocytotic vesicle formation, and cytoskeleton-dependent trafficking of the endocytotic vesicle. The novel, ANXA5-mediated internalization is independent of membrane ruffling and actin polymerization and can mediate the uptake of tissue factor in a macrophage cell model [74]. Downregulation of tissue factor through extracellular ANXA5 has also been observed in a mouse carotid artery injury model (same study). Such regulation of extracellular tissue factor levels through ANXA5-mediated cell entry might be part of a more general mechanism to control cell surface receptors under physiological stress conditions [71]. 2.4.3.1. Antibody characterization IgG fractions from APS patients reduce the levels of ANXA5 on cultured trophoblasts and endothelial cells, and accelerate coagulation of plasma exposed to these cells [69]. The IgG fractions also reduce

88

S. Sodin-Sˇemrl et al.

the level of ANXA5 bound to noncellular phospholipid bilayers and this reduction depends on the presence of b2GPI and results in acceleration of coagulation. The hypothesis stating that disruption of the two-dimensional ANXA5 shield by aPL (with cofactor b2GPI) enabling the coagulation factors to bind and activate prothrombotic pathways has been shown by tissue immunohistochemistry, cell culture studies, coagulation assays with noncellular phospholipids, and competition experiments on artificial phospholipid bilayers [69]. Specifically, the single and combined effects of antibodies against ANXA5 (anti-ANXA5) and anti-b2GPI antibodies on binding of ANXA5 to negatively charged phosholipid membranes were examined on giant vesicles and were found to differentially effect the binding. The results indicate competition between ANXA5 and complexes of b2GPI-anti-b2GPI for the same binding sites and support the shield disruption hypothesis on procoagulant membranes [75]. 2.4.3.2. Clinical significance Clinical studies have primarily focused on two groups of patients: having either pregnancy complications or systemic autoimmune diseases (including rheumatoid arthritis), with or without thrombotic events. The majority of studies indicate anti-ANXA5 as a strong risk factor for fetal loss. AntiANXA5 have been detected in some systemic autoimmune diseases, particularly in SLE with thrombotic events, and APS. However, there are some reports of the presence of anti-ANXA5 in patients with arterial thrombosis lacking criteria for APS or evident autoimmune disease (reviewed in [76]). ANXA5 is abundant in late stage atherosclerotic lesions and may play a role in cardiovascular disease. Sera from SLE patients with a history of cardiovascular disease inhibited ANXA5 binding to endothelium caused by IgG antibodies [77].

3. Cellular Microexovesicles (Microparticles) in Thrombosis and Hemostasis with Special Emphasis on APS MVs can be defined as 0.1–1 mm cell-derived vesicular structures that are shed off cells into the circulation and therefore represent an important system of transport of matter and information within the body [8–11]. It was found that MVs can be released from cells (with failure of membrane asymmetry) that have been exposed to a procoagulative and/or inflammatory environment [78]. They can be released by all cell types upon activation or apoptosis [78]; moreover, they were suggested to play a role in vascular disease [79], inflammation [80], and communication of tumor cells with macrophages [81]. Generally, MVs are enriched in various bioactive molecules and may directly stimulate the release of pro- as well as anti-inflammatory mediators, transfer membrane receptors (such as chemokine receptors), proteins, mRNA and organelles (e.g., mitochondria) between cells, promote leukocyte rolling, deliver infectious agents into cells (for instance, human immunodeficiency virus, prions) [11], and induce immune cell apoptosis. Figure 1 shows an example of an erythrocyte membrane bud and a MV from the same sample.

89

Antiphospholipid Syndrome

A

500 nm

B

200 nm

Figure 1 Transmission electron microscopy micrographs of freeze–fracture replicas showing the tubular bud on top of the echinocyte spicule (A) and the free tubular microvesicle (MV) released from the membrane (B). The budding/vesiculation was induced by adding 40 mM of dodecylmaltoside to the erythrocyte suspension. Adapted from [82].

MVs are also a potential diagnostic tool as the analysis of their abundance and composition can convey information on processes that are taking place in a particular organism. The abundance of MVs can be measured by flow cytometry while staining by different cell surface markers can point to their possible origin and give information on processes in mother cells. For example, MVs contain varying amounts of surface-exposed PS which can be detected by ANXA5. The function of MVs vary according to cellular origins and inducers of vesiculation. Hence, MVs can be pro-coagulant or anti-coagulant, but when the balance is disrupted in favor of the former population, this reflects an increased thrombotic risk. It was found that patients with APS have an increased level of MVs in blood [12]. Pinching off a MV from the cell membrane (called microvesiculation) is the final event in a complex process of membrane budding, a process which may be induced by various mechanisms reflected in changes of the local membrane curvature. These changes are strongly coupled to the lateral composition of the membrane which is—in turn—determined by the intrinsic properties of the molecules that constitute the structural domains, and nonlocal effects deriving from maximization of entropy, constraints upon the membrane area and enclosed volume, and interaction of each membrane leaflet with the surrounding solution. While the interdependence between the intrinsic shape of the molecules that constitute the membrane and the local membrane shape is an obvious feature in budding, also the above-mentioned nonlocal effects are important, especially in the initial phases, since they may drive the membrane to a point where the budding would more likely take place [83–87].

3.1. Mechanisms Leading to Microvesiculation 3.1.1. Initiation of the budding process by intercalation of the exogeneous molecules into the outer layer of the membrane Within the bilayer couple hypothesis [88–90], the two membrane leaflets are strongly coupled due to the hydrophobic effect; therefore, changes in the difference between the areas of the two membrane leaflets may bend the membrane inward or

S. Sodin-Sˇemrl et al.

90

outward, (depending on the relative change of the two areas) thereby initiating the process of budding. If the area of the outer leaflet increased with respect to the area of the inner leaflet, the membrane would bend outward while if the area of the inner leaflet increased with respect to the area of the outer leaflet, the membrane would bend inward. Additionally, if the intrinsic shape of the part of the molecule intercalated into the membrane leaflet, it would also contribute to the change of the local membrane curvature [91]. The relative change in the leaflet areas can thus be induced by intercalating of exogeneous molecules into one of the leaflets. In early experiments, it was shown that the normal discoid shape of erythrocytes changed into echinocyte shape by addition of substances to the suspension which intercalated preferentially into the outer membrane leaflet [88]. Theoretical studies have shown that the parameter that drives the outward or inward folding of the membrane is indeed the difference between the outer and the inner membrane leaflet areas [92]. A method has been developed to calculate the equilibrium shape of the membrane by minimization of the membrane bending energy considering the membrane as a thin isotropic elastic shell [91, 93],

kc Wb ¼ 2

ð ð2HÞ2 dA;

ð1Þ

A

where kc is the membrane bending constant, H ¼ (C1 þ C2)/2 is the local mean curvature, C1 and C2 are the two principal membrane curvatures, dA is the area element, and A is the membrane area. The relevant geometrical constraints requiring fixed membrane area A

ð dA ¼ A

ð2Þ

A

and fixed difference between the two membrane leaflet areas DA [92]

ð

d

2HdA ¼ DA;

ð3Þ

A

where DA is the prescribed area difference and d is the distance between the two neutral areas of the two membrane leaflets, are taken into account. The above variational problem can be expressed by a system of Euler–Lagrange differential equations which are solved numerically. Figure 2 shows the calculated shapes of a confined membrane segment with prescribed area A and increasing difference between the two membrane leaflet areas DA which may develop by continuous intercalation of molecules into the outer membrane layer. As the process of the budding proceeds, redistribution of membrane proteins as well as changes in cytoskeleton attachment occur; therefore, the composition of the MV that is eventually pinched off the membrane is a reflection of the sorting of membrane constituents that took place prior to the detachment of the MV from the membrane.

91

Antiphospholipid Syndrome

Δa = 0.2740

Δa = 0.3677

Δa = 0.4336

Δa = 0.5089

Δa = 0.6791

Δa = 0.9451

Δa = 1.1784

Δa = 1.3796

Δa = 1.5877

Δa = 1.8986

Δa = 2.0264

Δa = 2.5298

Figure 2 Calculated shapes of a confined membrane segment for increasingp normalized differffiffiffiffiffiffiffiffiffi ence between the two membrane leaflet areas Da ¼ DA/pdRs, where Rs ¼ A=p. The shapes correspond to the respective minima of the membrane bending energy at prescribed area of the segment A and prescribed normalized difference between the areas of the two membrane leaflets DA. Adapted from [85].

The membrane constituents are more or less free to move laterally over the membrane surface, so they would accumulate in regions of local curvature which are favorable for them—that is, which enable them to be in their lowest possible energetical state, while regions of unfavorable curvature would be depleted of these constituents. The direct interactions between the membrane constituents may promote separation of membrane constituents to form membrane rafts— mobile sphingolipid and cholesterol-based microdomains in the membrane. Rafts vary in composition and size (6–50 nm in diameter) [94–97], host-specific proteins, and can coalesce into larger, functional domains [98, 99]. The local clustering of raft elements, with a preference for high spherical curvature, and the concomitant local shape changes may be strongly coupled to form a part of the driving mechanism of the vesicle budding [100]. 3.1.2. Curvature sorting of membrane constituents Membrane constituents may be single molecules or small complexes of molecules (membrane nanodomains) (Fig. 3). We assume that the membrane constituent of the i-th species as a result of its structure and local interactions energetically prefers a local geometry that is described by the two intrinsic principal curvatures (C1m,i and C2m,i). If C1m,i ¼ C2m,i, the in-plane orientation (Fig. 4) of the constituent is immaterial. Such a constituent is called isotropic (Fig. 3B). If C1m,i ¼ 6 C2m,i, the constituent is called anisotropic

S. Sodin-Sˇemrl et al.

92

A

90⬚ C1m > 0, C2m = 0

B 90⬚ C1m = 0, C2m = 0

C 90⬚ C1m > 0, C2m = 0

D 90⬚ C1m > 0, C2m ≈ 0

Figure 3 Schematic presentation of different membrane constituents. The constituents may be single molecules (A, B) and complexes of molecules (C) intercalated into both membrane leaflets (A–C) or into one leaflet only (D). The constituent (B) is isotropic, others are anisotropic.

(Fig. 3A,C,D). The orientation of anisotropic membrane constituent with respect to the local coordinate system of the membrane is important for its energy (Fig. 4). It is assumed that anisotropic constituent will on the average spend more time in the orientation that is energetically more favorable than in any other orientation. The intrinsic principal curvatures of molecules are in general different (C1m,i 6¼ C2m,i) (Fig. 3). Also, small complexes of molecules that form a membrane constituent may have in general different intrinsic curvatures C1m,i and C2m,i from the intrinsic curvatures of the molecules which compose the constituent (Fig. 3). The free energy of a single membrane constituent of the i-th kind can be written in the form [101]:

   xi xi þ xi ðxi þ xi ÞDm;i D 2 2 2 ðD þ Dm;i Þ  kT ln Io ; Fi ¼ ðH  Hm;i Þ þ 2 4 2kT ð4Þ

93

Antiphospholipid Syndrome

A

B w

w=0

w≠0

Figure 4 Schematic presentation of an anisotropic membrane constituent. The intrinsic principal directions are rotated for an angle o with respect to the local principal directions of the membrane.

where xi and xpi ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi are the interaction constants, H ¼ (C1 þ C2)/2 and D ¼ jC1  C2 j=2 ¼ H 2  C1 C2 are the mean curvature and curvature deviator, and Hm,i and Dm,i are the intrinsic mean curvature and intrinsic curvature deviator describing the intrinsic shape of the membrane constituent of the i-th species, kT is the thermal energy, and I0 is the modified Bessel function. Assume a single (sufficiently large) membrane patch of lateral area A and local mean and deviatoric curvatures H and D, respectively. Let the membrane contain three constituent species, N1 constituents of type i ¼ 1, N2 constituents of type i ¼ 2, and N3 constituents of type i ¼ 3. All three molecular species can in general be anisotropic (by having a nonzero curvature deviator Dm,i). From the conservation of the overall number of constituents in the lipid layer (N ), we express the number of molecules of first type N1 as N1 ¼ N  N2  N3. For the sake of simplicity, we assume that all three species occupy the same lateral cross-sectional area per constituent in the membrane layer a ¼ A/N. The nature of the lipid bilayer allows its constituents to laterally redistribute. That is, in a nonhomogeneously curved lipid layer, the molecules of all species are assumed to migrate toward their energetically preferred membrane regions so as to minimize the overall free energy. We describe this degree of freedom by the local compositions of the species, namely m2,m3, and m1 ¼ 1  m2  m3, where mi ¼ Ni/N, are equal to the local fractions of the membrane area covered by the molecules of type 1, 2, and 3, respectively. We denote the average compositions m  i as:

ð 1 m i ¼ mi dA: A

ð5Þ

A

The free energy per molecule F of the membrane layer includes the contribution of the constituents’ ( fi ¼ Fi/kT ), the contribution of the configurational entropies

S. Sodin-Sˇemrl et al.

94

[101, 102], and the contribution of the direct interactions between molecules of the second type (i ¼ 2), considered within the Bragg–Williams approximation [103],

ð F N ¼ kT A A

(

)   3  X mi wz2 2 m dA ð6Þ þ mi fi þ kT mi ln  mi þ m i m i 2 2 i¼1

where w is the constant of direct interactions between constituents of type i ¼ 2 (in units of kT ), and z2 is the corresponding coordination number. We can construct a Lagrangian of the form:

   3  X mi wz2 2 ~ m ð7Þ L¼ mi fi þ kT mi ln  ðmi  m  i Þ þ li ðmi  m iÞ þ m i 2 2 i¼1 where ~ li are the Lagrangian multipliers. Inserting into the Lagrangian the relations m1 ¼ 1  m2  m3 and m 1 ¼ 1  m 2  m  3 , we can eliminate one of the Lagrangian multipliers by defining l2 ¼ ~ l2  ~ l1 and l3 ¼ ~ l3  ~l1 . Using the Euler–Lagrange equations qL/qm2 ¼ 0 and qL/qm3 ¼ 0, we can derive an expression for composition m3 in the form:

m3 ¼ m 3

m2 e f2 f3 þwz2 m2 Ð ; 1=A m2 e f2 f3 þwz2 m2 dA

ð8Þ

while m1 can be obtained from m1 ¼ 1  m2  m3. The equation for m2 is in general an integral equation. However, for small direct interactions (w 1), we can apply the expansion up to the first relevant term in w,

m2 ¼ m02 ð1  wz2 m02 ð1  m02 ÞÞ

ð9Þ

where

m02 ¼

m  2 ef2 l2

m  2 ef2 l2 þm  3 ef3 l3 þ m  1 ef1

ð10Þ

is the value of m2 for w ¼ 0. The values of the Lagrangian multipliers l2 and l3 are obtained from the constraint (4,5). For illustration, a case is considered where the membrane is assumed to be composed of lipid molecules (type i ¼ 1) which for the sake of simplicity are taken to be isotropic (i.e., Dm,1 ¼ 0)), anisotropic constituents (i ¼ 2) (Dm,2 6¼ 0), and isotropic conical constituents with a preference for a high spherical curvature (i ¼ 3) (Dm,3 ¼ 0, Hm,3 6¼ 0). The constituents of type 2 and 3 are taken to be distributed only in the outer layer of the membrane bilayer. The closed membrane shape given in Fig. 5 was calculated by minimization of the Helfrich bending energy [Eq. (1)] as

95

Antiphospholipid Syndrome

explained elsewhere [104]. It was considered that while affecting the energy of the vesicle shape, the presence of the anisotropic constituents in the membrane does not substantially affect the shape itself [104]. It can be seen in Fig. 5 that the anisotropic saddle-like membrane constituents (m2) are predominantly distributed in the region of the membrane neck, while the isotropic constituents with a preference for a highly isotropically curved membrane (m3) are predominantly accumulated on the spherical daughter vesicle/bud. Increase of m2 with increasing interaction constant w (not shown) indicates that the direct interactions between the constituents may play an important role in the energetics of the budding process and clustering of membrane constituents (i.e., raft formation). It can be concluded that the budding process is promoted by the accumulation of anisotropic membrane constituents in the membrane necks and by accumulation of isotropic constituents that energetically prefer large positive curvature on the bud. Below we present experimental evidence on the curvature sorting of membrane constituents. A

B 100 nm m1

0.75

0.5

0.25

Spherical bud

Neck m2 m3

−0.25

−0.5

w = 0.5

Echinocyte spicule

Figure 5 (A) Equilibrium lateral distribution of membrane constituents in the budding region of the bilayer membrane for a 3-component membrane (lipids: 1; anisotropic constituents: 2; isotropic conical constituents: 3). The axisymmetric closed membrane shape has a relative average mean curvature of 1.075 and a relative volume 0.99. The values are calculated relative pffiffiffiffiffiffiffiffiffiffiffi to the sphere with radius Rs ¼ A=4p. The values of the model parameters are: x1 ¼ 16 kT nm2, Hm,1 ¼ Dm,1 ¼ 0; x2 ¼ 320 kT nm2, Hm,2 ¼ 0, Dm,2 ¼ 0.3 nm1, m 2 ¼ 0:01, z2 ¼ 6; x3 ¼ 240 kT  3 ¼ 0:05; xi ¼ 0. (B) Spherical bud on the tip of the nm2, Hm,3 ¼ 0.3 nm1, Dm,3 ¼ 0, m echinocyte spicule. Budding was induced by adding 40 mM of dodecylzwittergent to the suspension of erythrocytes. Adapted from [108] and [103].

96

S. Sodin-Sˇemrl et al.

Methods have been developed to induce budding, microexovesiculation, and endovesiculation of the erythrocyte membrane, for example, by exogeneously adding amphiphilic molecules to the erythrocyte suspension [105] or increasing cytosolic calcium level and to analyze the composition of the MVs [106]. It was found that composition of some membrane constituents in the MVs differs from the composition of the mother membrane and that the membrane cytoskeleton detaches prior to microexovesiculation [107]. Recently, the distribution of raft markers in curved membrane exvaginations and invaginations, induced in human erythrocytes by amphiphile treatment or increased cytosolic calcium level, was studied by fluorescence microscopy; cholera toxin subunit B and antibodies were used to detect raft components [100]. Figure 6A shows the exvaginated shape of erythrocytes treated with ionophore A23187 plus calcium. The spiculae are distributed around the spheroid main body of the cell. Membrane buds, apparently in the process of being released, can be seen on the tips of spiculae. A marked enrichment of stomatin, sorcin, and synexin was detected in calcium-induced erythrocyte spiculae (Fig. 6B1, C1, and D1). Clustering of rafts and membrane proteins in highly curved membrane regions (invaginations) and MVs has also been observed previously [107, 109]. In order to study the differential segregation of membrane proteins during budding by a complementary method, we isolated MVs shed by the erythrocytes upon the respective treatments and compared the relative amounts of several membrane marker proteins. As most of the MVs probably originate from the tips

Figure 6 (A) Scanning electron micrograph showing exvaginated human erythrocytes following incubation with ionophore A23187 (1 mM, 10 min, 37  C) in the presence of 3.8 mM calcium. (B1) Immunocytochemical detection of stomatin in erythrocytes treated with A23187 plus calcium. (C1) Immunocytochemical detection of sorcin in erythrocytes treated with A23187 plus calcium. (D1) Immunocytochemical detection of synexin in erythrocytes treated with A23187 plus calcium. Notably, the magnification is larger in A than in other micrographs. Adapted from [100].

Antiphospholipid Syndrome

97

of membrane protrusions [82], the membrane protein content of the released vesicle reflects the local membrane protein composition of the protrusions’ tips. In accordance with previous results [107, 110, 111], the glycosylphosphatidylinositolanchored acetylcholinesterase, which is a raft marker located at the exoplasmic side of the membrane, is found to be highly enriched in all types of highly curved MVs. The vesicular aliquots were normalized to acetylcholinesterase activity to show the differences in the depletion of the respective proteins relative to this raft marker. The MVs are strongly depleted in the peripheral membrane protein band 6 and the membrane skeletal components spectrin, band 4.1, band 4.2, and actin. Band 3 and the major integral raft components, stomatin and flotillin-2, are present in the MVs. It was indicated that two complementary mechanisms may take place in the budding of heterogeneous membranes containing isotropic and anisotropic constituents: accumulation of saddle-preferring membrane constituents in the neck connecting the bud and the parent membrane [103] and accumulation of strongly spherically curved membrane-preferring constituents in the spherical region of the daughter vesicle/bud [100, 112–114]. On the basis of theoretical and experimental studies, it was concluded that membrane skeleton-detached, laterally mobile raft elements may sort into curved or flat membrane regions, dependent on their intrinsic molecular shape and/or direct interactions between the raft elements. Curvature sorting of membrane constituents explains enrichment or depletion of buds and MVs with respect to certain membrane constituents and/or membrane rafts.

3.2. Effect of b2GPI and aPL on Budding of Phospholipid Vesicles In previous studies [115–117], giant phospholipid vesicles (GPVs) have been proven to be useful systems for determination of the effects of b2GPI and aPL on phospholipid membranes. Interest in these models was recently renewed due to their diagnostic and prognostic clinical potential. Below we present the results of our observations. 3.2.1. Methodology b2GPI was purified from pooled human sera by affinity column chromatography. The monoclonal anti-b2GPI Cof-22 (generously provided by Prof. Takao Koike) obtained from BALB/c mice immunized with human b2GPI and recognizing domain III of b2GPI was used in a concentration of 1 mg/mL. IgG fractions were isolated from the sera of one patient with APS and from two children with atopic dermatitis containing high titers of IgG anti-b2GPI. Monoclonal HCAL anti-b2GPI antibodies (chimeric mouse monoclonal antibodies containing the human Fc region), obtained from BALB/c mice immunized with human b2GPI [118] and recognizing domain V of b2GPI, were dialyzed in PBS. In all experiments, the final concentration of HCAL anti-b2GPI was 1 mg/l. GPVs were prepared at room temperature (23  C) by the electroformation method [119] with small modifications as described in detail in [120, 121]. Briefly, 10 ml of lipid mixture was applied to platinum electrodes. The solvent was allowed

98

S. Sodin-Sˇemrl et al.

to evaporate in a low vacuum for 2 h. The coated electrodes were placed in the electroformation chamber which was then filled with 0.2 M sucrose solution. An AC electric current with an amplitude of 5 V and a frequency of 10 Hz was applied to the electrodes for 2 h, which was followed by 2.5 V and 5 Hz for 15 min, 2.5 V and 2.5 Hz for 15 min, and finally 1 V and 1 Hz for 15 min. The content was rinsed out of the electroformation chamber with 0.2 M glucose solution and stored in a plastic test tube. Before placing the vesicles into the observation chamber, the sample was gently mixed. Unilamellar GPVs were prepared out of 10 mol% 1-pamitoyl-2-oleoyl-phosphatidylserine (POPS) and 90 mol% 1-pamitoyl-2oleoyl-phosphatidylcholine (POPC). An individual GPV of medium size (diameter 40–60 mm) and without visible irregularities on the surface was aspirated by a glass micropipette from the compartment containing GPVs, and then transferred to the compartment containing b2GPI (100 mg/ml) and/or anti-b2GPI or purified normal human IgG in the control experiments. 3.2.2. Experimental results The budding of GVPs was observed for the period of 1 h in solution containing b2GPI (Fig. 7). Budding and release of daughter vesicles from GPV were demonstrated following transfer of GPV into the compartment containing b2GPI and Cof-22 (Fig. 8). The effects were similar, but less pronounced, when using an IgG fraction from an APS patient. There were no changes observed using GPV and Cof-22 without b2GPI. When children’s (atopic dermatitis) or normal human serum were used, no changes on the GPV were found. The results of this study imply that b2GPI alone and in combination with antib2GPI promote the budding of phospholipid membrane. The budding of GPVs could be initiated by the insertion of the C-terminal loop of b2GPI into the outer layer of a phospholipid membrane. As the C-terminal loop of b2GPI is relatively short, it does not expand through the whole width of the outer layer of the membrane. Therefore, the insertion of the C-terminal loop in an external portion of the outer layer of the membrane causes an increase of the area of the outer layer of the phospholipid bilayer with respect to the area of the inner layer as well as an increase of the local curvature of the membrane. Consequently, the membrane is bent outward and curvature sorting of membrane constituents further promotes the formation of the bud and the neck. The permeabilization and consequent fading of GPVs were observed, mostly in more spherical GPVs [120]. The volume of these GPVs was very close to the maximal volume that could be reached with a finite membrane area. Theoretically, as the mean shape of a vesicle approaches a sphere, the thermal shape fluctuations become smaller, since, for a sphere as a stationary shape, there is no space available for fluctuations at the constant volume. In the case of a nearly spherical vesicle, when additional molecules of b2GPI were inserted into the outer monolayer, the relative difference in the surface between the monolayers caused an increased tension in the inner monolayer, which could trigger a transient pore formation and consequent leaking of the membrane [122].

99

Antiphospholipid Syndrome

0.5 min

1.5 min

3 min

5 min

10 min

60 min

61 min

65 min

74 min

140 min

Figure 7 Budding of GPVs in solution containing 100 mg/ml of b2GPI. Two buds (marked by black arrows) were formed on the GPV about 90 s after transfer of the GPV into the solution containing b2GPI. No significant changes were seen during further observation for up to 60 min. Only one bud is visible 5 min after transfer as the other one was out of focus. Bar ¼ 20 mm. Adapted from [120].

30 sec

45 sec

60 sec

70 sec

80 sec

100 sec

120 sec

180 sec

250 sec

500 sec

Figure 8 Budding of GPVs transferred into solution containing 100 mg/ml b2GPI and Cof-22. Intense budding of the GPV started quickly after the transfer. Small daughter vesicles (marked with black arrow) separated and moved away from the GPV. Steady state of the GPV was reached after 3 min. Bar ¼ 20 mm. Adapted from [120].

4. Coalescence of Membranes Caused by b2GPI and aPL Collective interactions between membraneous structures are of great interest, especially as they can be mediated by protein cofactors and aPL involved in APS. Close approach and coalescence of membranes enable the onset of biochemical processes involving molecules of interacting membranes. Below we give experimental evidence on the effect of b2GPI and aPL on the coalescence of GPVs and suggest a possible mechanism for explanation of the mediating effect of the aPL.

100

S. Sodin-Sˇemrl et al.

4.1. Experimental Evidence on the Effect of b2GPI and aPL on Coalescence of Phospholipid Vesicles 4.1.1. Methodology b2GPI (Hyphen BioMed, France) was resuspended, aliquoted, and stored at 70  C. In all experiments, the final concentration of b2GPI in PBS was 100 mg/l, which is approximately half the concentration of physiological b2GPI in normal human plasma (about 200 mg/l) [123, 124]. An IgG fraction was isolated from the serum of a patient with APS. The IgG fraction contained high titers of anti-b2GPI as determined by affinity purification on a 2-ml protein-G column (Pierce, Rockford, USA), using the protocol recommended by the manufacturer. The IgG fraction was equilibrated against PBS (pH 7.4) in a desalting column. This IgG fraction gave comparable results to whole serum in an anti-b2GPI ELISA performed as described previously [125]. The levels of anti-b2GPI in the IgG preparations were considered as medium range, amounting to about a half of the anti-b2GPI levels in serum. The synthetic lipids CL (1,10 2,20 -tetraoleoyl cardiolipin), POPC (1-palmitoyl-2oleoyl-sn-glycero-3-phosphocholine), and cholesterol were purchased from Avanti Polar Lipids, Inc. Appropriate volumes of POPC, CL, and cholesterol, all dissolved in a 2:1 chloroform/methanol mixture, were combined in a glass jar and thoroughly mixed. For charged CL vesicles, POPC, cholesterol, and CL were mixed in the proportion of 2:2:1. For neutral POPC GPVs, POPC and cholesterol were mixed in the proportion of 4:1. Cholesterol was added to POPC to increase the longevity of GPVs. The GPVs were left for sedimentation under gravity for one day at 4  C. 200–400 ml of the sediment was collected from the bottom of the tube and used for a series of experiments. GPVs were observed by a Zeiss Axiovert 200 inverted microscope with phase contrast optics (objective magnification 100) and recorded by a Sony XC-77CE video camera. The solution containing GPVs was placed in an observation chamber made from cover glasses sealed with grease. The larger (bottom) cover glass was covered by two smaller (18  18 mm glasses), each having a small semicircular part removed at one side. Covering the bottom glass with two opposing smaller glasses thus formed a circular opening in the middle of the observation chamber. The circular opening enabled subsequent addition of PBSdissolved protein. In all experiments, the solution of GPVs (45 ml) was placed in the observation chamber. The solution containing the substance under investigation (5 ml) was added through the circular opening in the middle of the observation chamber. The osmolarity of the sample containing GPVs was 205 mosm/l (measured by a Knauf Semiosmometer). The observation chamber was mounted on a temperature-regulated microscope stage. In observing the coalescence of the population of GPVs, the temperature was kept at 40  C while the budding of the GPVs was induced by increasing the temperature above the room temperature [126]. 4.1.2. Experimental results The solution of GPVs contained a heterogeneous population of shapes. Most of the GPVs were flaccid. Thermal fluctuations of shapes were notable. Few minutes after the addition of the solution containing dissolved b2GPI, thermal fluctuations of

Antiphospholipid Syndrome

101

GPVs diminished, protrusions disintegrated into spherical fragments while GPVs attained nearly spherical shapes. This phase of the morphological changes was observed also after addition of PBS alone. With PBS, the sample ultimately contained nearly spherical fluctuating GPVs. With b2GPI however, nearly spherical fragments joined into two or multi-compartment structures composed of spherical parts and flat walls (Fig. 9A and B). Furthermore, GPVs adhered to the bottom of the glass slide and ceased to fluctuate. Large surfaces of contact between spherical compartments could be observed. The sample ultimately contained motionless aggregates of GPVs which adhered to the bottom of the glass slide. GPVs in the sample retained their shape until the end of observation (one hour). When the IgG fraction of a patient with APS was added to the solution with GPVs, a similar time course was observed as when b2GPI was added to GPVs (Fig. 9C and D). However, in addition to the suppression of fluctuations and adhesion of vesicles, we observed lateral segregation within the membrane in the time scale of minutes (Fig. 9F). In the next step, we added b2GPI to the solution of GPVs one hour following addition of the patients’ IgG fraction. We have observed additional transformations of GPVs into ghosts and also sudden bursts of some GPVs. The sample was eventually formed from adhered GPVs and ghosts, bi- and multicompartment GPV and ghost formations, some exhibiting lateral segregation and regions of larger curvature (Fig. 9E). The reversed order of addition of b2GPI and the patients’ IgG fraction in the same time intervals resulted in similar observations; however, we did not directly observe bursts of the membrane.

4.2. Theoretical Description of the Coalescence of Membranes: Interaction Between Charged Surfaces Mediated by Ions with Dimeric Distribution of Charge The interaction between like-charged membranes is described theoretically by two interacting electric double layers consisting of two charged flat surfaces, each of the area A, separated by a distance D (Fig. 10). It is assumed that each surface bears uniformly distributed charge with surface charge density s0  0 describing negative charge of CL headgroups. The space between the charged surfaces is filled with an electrolyte solution, one of the ion species representing antibodies. The antibodies are composed of four chains, two heavy ones and two light ones, organized into a dimeric structure. In this work, for simplicity, it is assumed that an antibody molecule is a globular multivalent ion with a simple distribution of charge where two equal effective charges e are separated by a distance l (Fig. 10). The antibodies are taken to be equal and indistinguishable. The ions in the solution distribute according to the electrostatic and entropic effects. Because of the electrostatic effects, it could be expected that counterions are attracted by the charged surfaces while coions are repelled by these surfaces so that the electric field is shielded by the distribution of ions. In equilibrium, an equilibrium configuration is attained which yields consistently related distribution functions of ions, electric field, and free energy of the system. The dependence of the free energy on the distance between the surfaces gives the interaction between the surfaces. If the free energy decreases with increasing distance, the interaction is

102

S. Sodin-Sˇemrl et al.

Figure 9 Effect of b2GPI and IgG fraction of a patient with APS syndrome on GPVs containing cardiolipin (CL) and cholesterol; sticky formations induced by addition of b2GPI (A and C) and by addition of IgG fraction of a patient with APS (B and D). A flat wall dividing the vesicular compartments (area of contact of the two vesicles) is indicated as the two solutions are of different

103

Antiphospholipid Syndrome

s0

s0 + +

+ +

+

+ l

+ + +

+

+ + x d

0 y e l/2

l/2

r1

t e

r

r2 z

x

Figure 10 Schematic illustration of two like-charged flat surfaces with surface charge density s0 and area A, separated by a distance d (upper). The space between the surfaces is filled with electrolyte solution. The globular ions with internal distribution of charge represent the antibodies with dimeric structure. Within an antibody, the charge distribution is represented by two point charges separated by a distance l (lower). contents (D). The areas of contact seem larger after addition of b2GPI (A and C). Regions of the lateral separation can be observed after the addition of IgG fraction of a patient with APS (B and D). CL-containing GPVs after addition of IgG fraction of a patient with APS (E). Lateral separation within the membrane is indicated (darker regions within the membrane). Some GPVs transformed into ghosts (light vesicles) indicating formation of membrane pores. The sample with CL-containing GPVs (F). Bars denote 10 mm. Adapted from [127].

S. Sodin-Sˇemrl et al.

104

repulsive while it is attractive if the free energy increases with increasing distance. If the global minimum of the equilibrium free energy with respect to the distance between the charged surfaces exists, the system attains the stable shape at the corresponding distance between the surfaces. Here, we focus on the role of the spatial distribution of charge within ions. We consider the simplest case of a symmetric electrolyte, where the antibodies have the role of counterions. As the assumed distribution of charge is not spherically symmetric, different orientations of the ion with respect to the electric field are not energetically equivalent. It can be expected that the ion will spend on the average more time in the orientation that is energetically more favorable. The orientation represents an additional degree of freedom within the system that could at certain conditions in the system give rise to an attractive interaction between the likecharged membranes. Below we present the theory taking into account these effects and show that orientational effects of the particular spatial distribution of charges within ions for certain model parameters cause an attractive interaction between membranes. 4.2.1. Electric field created by a single species of dimeric ions and a charged surface When the internal distribution of charge within the ion is taken into account, the microscopic density of charge is not uniform so that the macroscopic and the microscopic densities are in general different. The center of the mass of the dimeric ion is given by the radius vector r. The two charges are then located at r1 ¼ r þ (l/2)t and r2 ¼ r  (l/2)t where t ¼ ðsin y cos f; sin y sin f; cos yÞ is the director defining the orientation of the dimeric ion (Fig. 10). The microscopic density of charge is given by [128]:

ðr0 Þ ¼ edðr0  r1 Þ þ edðr0  r2 Þ;

ð11Þ

where d(r) is the Dirac delta function. At a given distribution of ions, the average microscopic density of charge of a chosen ion hi is obtained by [128]:

ð

hi ¼ d3 r0 mðr0 Þðr0 Þ ¼ emðr1 Þ þ emðr2 Þ;

ð12Þ

where m is the number of ions per volume and h. . .i stands for spatial averaging. For small enough distances between the charges within the ion l, the functions m(r1) and m(r2) can be expanded into Taylor series up to the terms of the second order with respect to the distance l,

  l l X qmðrÞ l2 X q2 mðrÞ m r t ¼ mðrÞ ti þ ti tj ; 8 ij 2 2 i qri qri qrj

ð13Þ

105

Antiphospholipid Syndrome

where ti(i ¼ x,y,z) are the components of the vector t. The expansions (13) are inserted into Eq. (12) to yield:

hi ¼ 2em þ

el 2 X q2 mðrÞ ti tj ; 4 ij qri qrj

ð14Þ

where the first term in the above equation (2em) corresponds to the macroscopic density of charge r while the second term represents the quadrupolar contribution to hi. The linear terms representing the density of the dipoles cancel out. The tensor T ¼ ti tj can be written in the matrix form [129]:

0

sin2 y cos2 f T ¼ @ sin2 y cos f sin f ― sin y cos y cos f

sin2 y sin f cos f sin2 y sin2 f sin y cos y sin f

1 sin y cos y cos f sin y cos y sin f A cos2 y

ð15Þ

In flat electric double layer, the density of charge varies only in the direction perpendicular to the charged surfaces x, therefore m(r) m(x), so that

hi ¼ 2em þ

el 2 DmðxÞ cos2 y 4

ð16Þ

where DmðxÞ ¼ d2 m=dx2 . In calculating an effective density of charge reff, all possible orientations of a dimeric ion are taken into account,

reff

1 ¼ 4p

ð 2p

ðp df

0

hisin y dy;

ð17Þ

0

where dO ¼ sin y dy df. Performing the necessary integrations yields [130]

reff ¼ 2em þ

el 2 Dm 12

ð18Þ

The effective density of charge reff enters the basic law of the electromagnetism expressed by the Maxwell equation [128]

ee0 r E ¼ reff ;

ð19Þ

where e is the permittivity of the solution and e0 is the permittivity of the free space. Inserting Eq. (18) into the above Maxwell equation yields:

ee0 r E ¼ 2em þ

el 2 Dm 12

ð20Þ

S. Sodin-Sˇemrl et al.

106

while considering that E ¼ rF we obtain the Poisson equation

ee0 DF ¼ reff :

ð21Þ

Eq. (20) is rewritten into



el 2 r ee0 E  rm 12

 ¼ 2em;

ð22Þ

wherefrom we can deduce the electric field density D,

D ¼ ee0 E 

el 2 rm 12

ð23Þ

The boundary condition is stated by considering that the normal component of the density of the electric field is connected to the surface density of charge

D n ¼ s0 ;

ð24Þ

where n is the vector in the direction of the surface normal and s0 is the area density of charge at the charged surface. It was taken that there is no electric field outside the electrolyte solution. This can be justified by low dielectric constant of the phospholipid tails. It follows from Eqs. (23) and (24) and from the definition of the derivative with respect to the direction of the normal to the surface En ¼ qF/qn that [130]

ee0

qF ¼ seff ; qn

ð25Þ

where

seff ¼ s0 þ

el 2 qm 12 qn

ð26Þ

is the effective area density of charge at the charged surface.

4.2.2. Electrostatic energy of the electric double layer composed of a charged surface and electrolyte solution containing dimeric counterions and coions We consider a charged surface with area density of charge s0 in contact with electrolyte solution containing dimeric counterions and coions. To make the derivation more simple, we include in the model also dimeric coions which turn

107

Antiphospholipid Syndrome

out to be strongly depleted from the system for relevant surface densities of charge but affect the results only negligibly [130]. The local concentrations of counterions and coions are denoted by mþ and m, respectively. Because of the spatial separation of the charges along the dimeric ion, the effective volume charge density reff, receives contributions from the local concentrations of both species and also from their second derivatives representing the internal quadrupolar distribution of charge within an individual ion [Eq. (18)] [130],

reff ¼ 2eðmþ  m Þ þ

el 2 ðDmþ  Dm Þ 12

ð27Þ

Similarly, the effective area density of charge at the charged surface is subject both to the charges on the surface and to the gradient of the concentrations,

el 2 ¼ s0 þ 12

seff



qmþ qn





qm  qn A

  :

ð28Þ

A

The electrostatic free energy can be written as [128]:

ð ð ð ee0 1 1 2 Fel ¼ ðrFÞ dV ¼ Freff dV þ FA seff dA; 2 2 2 V

V

ð29Þ

A

where dV is the volume element, dA is the area element, and FA is the potential at the charged surface. Using the expressions for the effective densities of charge [Eqs. (27) and (28)] and Green’s theorem, we reexpress Fel as:

1 Fel ¼ 2

ð



  ð  el 2 1 el 2 qF ðmþ  m Þ 2eF þ DF dV þ s0 FA þ ðmþ  m Þ 12 12 2 A qn V

ð30Þ To derive the differential equation for the electric potential, we consider in the free energy also the contribution of the distributional entropy of ions Fentr. In the entropy, ions of the same species are taken to be dimensionless, equal, and indistinguishable. Thus, the free energy of the system F is:

F ¼ Fel þ Fentr

ð31Þ

where [131]

ð  Fentr ¼ V

 mþ m mþ ln þ n ln  ðmþ þ m  2m0 Þ dV m0 m0

ð32Þ

S. Sodin-Sˇemrl et al.

108

where m0 denotes the bulk concentration of the dimeric ions. We chose the reference state where Fðmþ m m0 Þ ¼ 0. In thermal equilibrium, the free energy F must be minimal with respect to the distributions of ions mþ and m. To find the corresponding equilibrium distributions, we perform the first variation of F(mþ, m), resulting in

! " # ð ð el 2 qF el 2 mþ dF ¼ ðdmþ  dm ÞdA þ dmþ 2eF þ DF þ ln dV 12 A qn 12 m0 V " A # ð el 2 m þ dm 2eF  DF þ ln dV ð33Þ 12 m0 V Vanishing of dF for arbitrary dm and dmþ gives rise to both the Boltzmann distributions [130]:

   el 2 m ¼ m0 exp 2eF þ DF 12

ð34Þ

and the boundary condition at the charged surfaces



qF qn

 ¼0

ð35Þ

A

Note that Eq. (35) implies seff ¼ 0, indicating the tendency of the dimeric ions to fully neutralize the bare charges on the charged surfaces. Complete neutralization can occur only for l > 0 without a prohibitively large entropic penalty of immobilizing the ions onto the charged surfaces. Inserting m from Eq. (34) into the Poisson equation yields the differential equation which we express in terms of the dimensionless electrostatic potential C ¼ 2eF=kT . In the following we shall also use dimensionless spatial coordinates, x ¼ x=lD and so on, scaled by the Debye length lD ¼ 1=k with k2 ¼ 4  8plB lm0 where lB ¼ e2 =4pee0 ¼ 0:7 nm is the Bjerrum length in water. We then obtain a fourth order, nonlinear, partial differential equation [130]

DC ¼ sinhðC þ x2 DCÞ þ x2 D sinhðC þ x2 DCÞ

ð36Þ

The above equation depends on the dimensionless parameter:

kl x ¼ pffiffiffiffiffi 24

ð37Þ

109

Antiphospholipid Syndrome

that expresses the effective distance between the charges within the dimeric ion (Fig. 10). For x ¼ 0, the differential equation reduces to DC ¼ sinh C which is the familiar equation used for a symmetric salt solution of structureless, point-like ions. Solving the differential equation [Eq. (36)] requires the specification of two boundary conditions at the charged surfaces. The first one is given by (qC/qn)A ¼ 0; see Eq. (35). To formulate the second boundary condition, it is convenient to express the (local) surface charge density of the charged surfaces, s0 ¼ pe=8plB lD , in terms of the dimensionless charge density, p ¼ 8plB lD =a, where a is the (local) area per surface charge (the sign of the charged surface is determined by the sign of p). The second boundary condition seff ¼ 0, can be written as [130]:

p q 2 DC 4 ¼ coshðC þ x DCÞ qn x

ð38Þ

Upon insertion of the equilibrium distribution for m into F, we can show that the free energy can be calculated by the charging process [132]:

ð

sð0

F ¼ da A

0

ð ðp 1 Fðs0 Þ d s0 ¼ da Cð p0 Þdp0 16plB lD 0

0

A

ð39Þ

0

¨ckel limit 4.2.3. The Debye–Hu In the Debye–Hu¨ckel (DH) regime, the electrostatic potential is small everywhere (C  1) and linearization of the differential equation [Eq. (36)] yields

x4 r4 C þ ð2x2  1ÞDC þ C ¼ 0

ð40Þ

where r4 is the biharmonic operator. The boundary condition, Eq. (38) yields [130] p=x4 ¼ qðDCÞ=qn. Consider two large, like-charged, planar surfaces, located at (dimensionless) positions x ¼ 0 and x ¼ d ¼ d=lD , each having area A/2 and bare surface charge density s0 (with corresponding scaled surface charge density p). The electrostatic potential depends only on the x -direction, and we must solve the equation  ¼0 x4 C0000 þð2x2  1ÞC00 þ C ¼ 0 with the boundary conditions C0 ð0Þ ¼ C0 ðdÞ 4 000 000  and C ð0Þ ¼ C ð dÞ ¼ p=x . The solution can be written as Cð xÞ ¼ P4 oi x B e with o ¼ o , o ¼ o , and [130] 3 1 4 2 i¼1 i

ð1Þi p Bi ¼ 4  i ; x oi ðo21  o22 Þð1  edo Þ

o1;2 ¼

1

pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 1  4x2 2x2

ð41Þ

S. Sodin-Sˇemrl et al.

110

Most notably, for x > 1=2, the potential C exhibits damped oscillations whereas for x > 1=2 it decays monotonically. For the potential at the surfaces, we find  ¼ pCðx; dÞ  where the function Cðx; dÞ  is given by [130]: Cð0Þ ¼ CðdÞ

 ¼ Cðx; dÞ

  2  1 do do 1 1 1  coth coth 2 2 o1 x4 ðo21  o22 Þ o2

ð42Þ

 Upon insertion of C(0) into Eq. (39), we obtain the free energy F ¼ pNCðx; dÞ=2 where N ¼ ðs0 =eÞðA=2Þ is the number of fixed charges on each of the two flat  determines the nature of the interaction between the surfaces. The function Cðx; dÞ like-charged macroionic surfaces. For two isolated surfaces (d ! 1; the left diagram of Fig. 11 shows the potential for some selected cases), we obtain that Cðx; d ! 1Þ ¼ 1 and the surface potential C(0) ¼ p as well as the free energy F ¼ pN/2 are independent of x. In fact, this is the familiar Debye-Huckel result for point-like divalent salt ions [132]. The interaction between two like-charged planar surfaces represented by the normalized free energy C ¼ 2F/Np in dependence on their mutual distance d for x ¼ 0, 0.5, and 0.7 is presented in Fig. 11 (right diagram). For x < ½, the interaction 2  ¼ cothd þ d=sinh  is always repulsive. For x ¼ 1/2, we obtain Cðx ¼ 1=2; dÞ d  which still is a monotonously decaying function of d, implying repulsion between the two surfaces. However, for x > ½, the interaction turns attractive above a sufficiently  large separation d ¼ d between the two surfaces.   Figure 12 shows position d at which Cðx > 1=2; d Þ adopts a minimum, in dependence on the parameter x expressing the effective separation between the two charges within the dimeric ion [Eq. (37)]. A border of the phase diagram is shown separating the repulsive from the attractive regime. For example, x ¼ 0.7

1.0 2.0

0.8

1.6 C

y(x ) p

0.6 0.4

(b)

(c)

(c)

0.0

0.8 0

1

(a)

1.2

(a)

0.2

2

3 x

4

5

6

1

2

(b) 4

3

5

6

d/lD

Figure 11 Results for the Debye-Huckel regime. Left diagram: The normalized potential,C/p, of an isolated planar surface as a function of the scaled distance, x ¼ x=lD to that surface. Right diagram: The normalized free energy C ¼ 2F/Np as a function of the distance d ¼ d=lD between two charged surfaces. Both diagrams show the cases x ¼ x=lD for x ¼ 0 (a), x ¼ 0.5 (b), and x¼ 0.7 (c). Adapted from [130].

111

Antiphospholipid Syndrome

 (corresponding to l ¼ 3.4lD) results in an energetic minimum at d ¼ 3:2 (corresponding to d ¼ 3.2lD). Hence, the dimeric ions just match the optimal separation between the charged surfaces, indicating that a bridging mechanism is responsible for the attractive interactions.

5. The Effects of ANXA5, b2GPI and aPL on Phospholipid Membranes In order to study additional possible mechanisms preventing the exposure of procoagulant surfaces, GPVs with fluorescence-labeled ANXA5 were studied. As it was given in Section 2.4.3, ANXA5 is believed to form a protective shield by binding to negatively charged phospholipids thereby interfering with mechanisms involved in vascular thrombosis and pregnancy complications. The disruption of the ANXA5 shield by aPL [69] previously proposed the mechanism of pro-thrombotic action of aPL and indicated their possible involvement in fetal loss in APS. Antibodies in the presence of b2GPI compete with ANXA5 for the same binding sites on phospholipids causing the disruption of the protective ANXA5 shield on the procoagulant surfaces enhancing the activation of coagulation factors. In order to study the influence of b2GPI on the binding of ANXA5 to the negatively charged phospholipids in the presence of anti-b2GPI and anti-ANXA5, the GPVs were prepared out of 15 mol% POPS and 85 mol% POPC as described in Section 3.2.1. Using fluorescent microscopy, the accumulation of ANXA5 conjugated with ALEXA-fluor 488 on the GPV surfaces in the presence of purified IgG containing anti-b2GPI or anti-ANXA5 (with addition of b2GPI and 3 mM CaCl2) was measured. Purified IgG fractions were derived from four different

8

6 Attraction d

*

4

2

0

Repulsion

0.5

0.7

0.9

x

Figure 12 Results for the Debye-Huckel regime. The equilibrium distance between the charged  surfaces d in dependence on the parameter x expressing the effective separation between the two charges within the dimeric ion kl. The repulsive and attractive regions are marked. Adapted from [130].

S. Sodin-Sˇemrl et al.

112

human sera (APS anti-b2GPI, systemic sclerosis anti-ANXA5, anti-ANXA5 blood donor, aPL negative control). The ANXA5 fluorescent signal was stronger and its development was significantly faster in the presence of anti-ANXA5 sera as compared to the case when antib2GPI and control sera were present in the solution. A clear dependency of time and dose reduction of fluorescent ANXA5 emission on GPVs by anti-b2GPI is shown in Fig. 13. A distinct difference in the influence of anti-b2GPI compared with anti-ANXA5 on the ANXA5 binding to the negatively charged phospholipids exists on GPVs, which could suggest different mechanisms of their pathogenic action: anti-b2GPI acting as a procoagulant and anti-ANXA5 interfering with ANXA5 binding.

A

B

C

3 min

2 min

4 min

6 min

8 min

10 min

Emission (arbitrary units)

240

13 min

16 min

0 mg/l ab2GPI 30 mg/l ab2GPI

180

60 mg/l ab2GPI 1440 mg/l ab2GPI

120 60 0 0

5

10 Time (min)

15

20

Figure 13 (Upper panel) Micrograph images of GPV transferred into the test solution containing fluorescent ANXA5, Caþþ, and b2GPI in the final concentration of 105 mg/l and (A) no antib2GPI; (B) 60 mg/l anti-b2GPI; and (C) 1440 mg/l anti-b2GPI. Time after the transfer is indicated below images. (Lower panel) The influence of anti-b2GPI at different concentrations on the binding of fluorescent ANXA5 in the presence of b2GPI (105 mg/l) and Caþþ. Each line represents the average growth of emission at a particular time for three GPVs; bars indicate the maximum and minimum values. a.u., arbitrary units). Reprinted from [75], with permission from Oxford University Press.

Antiphospholipid Syndrome

113

These results support the proposed destructive role of anti-b2GPI on the ANXA5protective shield on the negatively charged phospholipids.

6. Conclusions Studies of physical mechanisms involved in budding of cell membranes and interactions between membraneous structures can importantly contribute to the knowledge of the physiological roles of aPL antibodies and their protein cofactors involved in APS. We have observed that b2GPI caused budding of PS-containing GPVs while the effect was enhanced when both b2GPI and Cof-22 antibodies were present in the solution. The Cof-22 antibodies alone had no effect, while the effect of the IgG fraction of the patient with APS was weak. b2GPI and the IgG fraction of the patient caused coalescence of the CL-containing GPVs. We observed permeabilization of the GPV membrane due to the presence of Cof-22 antibodies and IgG fraction of the patient with APS. It was shown that anti-ANXA5 and anti-b2GPI antibodies have different effects on the ANXA5 protective shield in GPVs. The mechanisms involved in budding are based on the increase of the difference between the two membrane leaflet areas and on the curvature sorting of the membrane constituents. There seem to be multiple mechanisms underlying the coalescence of membranes. We have shown that orientational ordering of dimeric ions (representing the antibodies) can lead to an attractive interaction between the like-charged membranes—as observed in experiments. GPVs and erythrocytes have proven to be a convenient system for the study of the effects of protein cofactors and aPL on the membraneous structures. In turn, experiments with plasma proteins have contributed to an advance in the electric double layer theory. Orientational ordering of the molecules (both, in-plane ordering of membrane constituents as well as of ions with spatial distribution of charge in solution) represent a unifying mechanism underlying the budding process as well as the coalescence of membranes. The orientational ordering represents degrees of freedom within the system which can lower the free energy and promote different processes. We hope that experiments and constructed theoretical models presented in this work will be followed by many studies that will reveal new and useful knowledge on the APS and related disorders.

ACKNOWLEDGMENTS The authors are grateful to coauthors of the published works ([30, 58, 75, 76, 82, 85, 101–104, 108, 120, 121, 125, 127, 129–131]) for their collaboration. The work of S. Sodin-Sˇemrl was supported, during the writing of this chapter, by the MC-IRG #028414. The figures were reproduced with permission from the publishers.

114

S. Sodin-Sˇemrl et al.

REFERENCES [1] G.R. Hughes, Thrombosis, abortion, cerebral disease, and the lupus anticoagulant, Br. Med. J. (Clin. Res. Ed.) 287 (1983) 1088–1089. [2] M. Galli, D. Luciani, G. Bertolini, T. Barbui, Anti-beta 2-glycoprotein I, antiprothrombin antibodies, and the risk of thrombosis in the antiphospholipid syndrome, Blood 102 (2003) 2717–2723. [3] E. Matsuura, Y. Igarashi, M. Fujimoto, K. Ichikawa, T. Koike, Anticardiolipin cofactor(s) and differential diagnosis of autoimmune disease, Lancet 336 (1990) 177–178. [4] P. McNeil, R.J. Simpson, C.N. Chesterman, S.A. Krilis, Anti-phospholipid antibodies are directed against a complex antigen that includes a lipid-binding inhibitor of coagulation: b2-glycoprotein I (apolipoprotein H), Proc. Natl. Acad. Sci. USA 87 (1990) 4120–4124. [5] P.G. de Groot, M. van Lummel, M. Pennings, R. Urbanus, H. Bas de Laat, P.J. Lenting, R.H. Derksen, Beta2-glycoprotein I and LDL-receptor family members, Thromb. Res. 114 (2004) 455–459. [6] P.G. de Groot, R.H. Derksen, Pathophysiology of antiphospholipid antibodies, Neth. J. Med. 62 (2004) 267–272. [7] W. Lim, M.A. Crowther, J.W. Eikelboom, Management of antiphospholipid antibody syndrome: a systematic review, JAMA 295 (2006) 1050–1057. [8] T.J. Greenwalt, The how and why of exocytic vesicles, Transfusion 46 (2006) 143–152. [9] D.D. Taylor, P.H. Black, Neoplastic and developmental importance of plasma membrane vesicles, Am. Zool. 26 (1987) 411–415. [10] T.L. Whiteside, Tumour-derived exosomes or microvesicles: Another mechanism of tumour escape from the host immune system? Br. J. Cancer 92 (2005) 209–211. [11] J. Ratajczak, M. Wysoczynski, F. Hayek, A. Janowska-Wieczorek, M.Z. Ratajczak, Membranederived microvesicles: important and underappreciated mediators of cell to cell communication, Leukemia 20 (2006) 1487–1495. [12] F. Dignat-George, L. Camoin-Jau, F. Sabatier, D. Arnoux, F. Anfosso, N. Bardin, V. Veit, V. Combes, S. Gentile, V. Moal, M. Sanmarco, J. Sampol, Endothelial microparticles: A potential contribution to the thrombotic complications of the antiphospholipid syndrome, Thromb. Haemost. 91 (2004) 667–673. [13] R.F. Zwaal, A.J. Schroit, Pathophysiologic implications of membrane phospholipid asymmetry in blood cells, Blood 89 (1997) 1121–1132. [14] A. Zachowski, Phospholipids in animal eukaryotic membranes: transverse asymmetry and movement, Biochem. J. 294 (1993) 1–14. [15] B. Perret, H.J. Chap, L. Douste-Blazy, Asymmetric distribution of arachidonic acid in the plasma membrane of human platelets. A determination using purified phospholipases and a rapid method for membrane isolation, Biochim. Biophys. Acta 556 (1979) 434–446. [16] R.A. Schlegel, P. Williamson, Phosphatidylserine, a death knell, Cell Death Differ. 8 (2001) 551–563. [17] V.A. Fadok, P.M. Henson, Apoptosis: giving phosphatidylserine recognition an assist-with a twist, Curr. Biol. 13 (2003) R655–R657. [18] D. Dolis, C. Moreau, A. Zachowski, P.F. Devaux, Aminophospholipid translocase and proteins involved in transmembrane phospholipid traffic, Biophys. Chem. 68 (1997) 221–231. [19] A. Zachowski, Phospholipids in animal eukaryotic membranes: Transverse asymmetry and movement, Biochem. J. 294 (1993) 1–14. [20] M. Ikeda, A. Kihara, Y. Igarashi, Lipid asymmetry of the eukaryotic plasma membrane: Functions and related enzymes, Biol. Pharm. Bull. 29 (2006) 1542–1546. [21] C. Alessandri, M. Sorice, M. Bombardieri, P. Conigliaro, A. Longo, T. Garofalo, V. Manganelli, F. Conti, M.D. Esposti, G. Valesini, Antiphospholipid reactivity against cardiolipin metabolites occurring during endothelial cell apoptosis, Arthritis Res. Ther. 8 (2006) R180–R191. [22] S. Horkko, E. Miller, E. Dudl, P. Reaven, L.K. Curtiss, N.J. Zvaifler, R. Terkeltaub, S.S. Pierangeli, D.W. Branch, W. Palinski, J.L. Witztum, Antiphospholipid antibodies are

Antiphospholipid Syndrome

[23] [24] [25] [26] [27] [28] [29] [30] [31] [32] [33] [34] [35] [36] [37] [38] [39] [40] [41]

115

directed against epitopes of oxidized phospholipids. Recognition of cardiolipin by monoclonal antibodies to epitopes of oxidized low density lipoprotein, J. Clin. Invest. 98 (1996) 815–825. M. Sorice, A. Circella, R. Misasi, V. Pittoni, T. Garofalo, A. Cirelli, A. Pavan, G.M. Pontieri, G. Valesini, Cardiolipin on the surface of apoptotic cells as a possible trigger for antiphospholipids antibodies, Clin. Exp. Immunol. 122 (2000) 277–284. R. Subang, J.S. Levine, A.S. Janoff, S.M. Davidson, T.F. Taraschi, T. Koike, S.R. Minchey, M. Whiteside, M. Tannenbaum, J. Rauch, Phospholipid-bound beta 2-glycoprotein I induces the production of anti-phospholipid antibodies, J. Autoimmun. 15 (2000) 21–32. A. Tincani, G. Balestrieri, F. Allegri, M. Cinquini, M. Vianelli, M. Taglietti, M. Sanmarco, K. Ichikawa, T. Koike, P. Meroni, M.C. Boffa, Overview on anticardiolipin ELISA standardization, J. Autoimmun. 15 (2000) 195–197. E.M. Bevers, P. Comfurius, J.L. van Rijn, H.C. Hemker, R.F. Zwaal, Generation of prothrombin-converting activity and the exposure of phosphatidylserine at the outer surface of platelets, Eur. J. Biochem. 122 (1982) 429–436. E.M. Bevers, P. Comfurius, R.F. Zwaal, Changes in membrane phospholipid distribution during platelet activation, Biochim. Biophys. Acta 736 (1983) 57–66. P. Comfurius, P. Williamson, E.F. Smeets, R.A. Schlegel, E.M. Bevers, R.F. Zwaal, Reconstitution of phospholipid scramblase activity from human blood platelets, Biochemistry 35 (1996) 7631–7634. D. Smrz, L. Draberova, P. Draber, Non-apoptotic phosphatidylserine externalization induced by engagement of glycosylphosphatidylinositol-anchored proteins, J. Biol. Chem. 282 (2007) 10487–10497. B. Bozˇicˇ, T. Kveder, M. Stegnar, E. Morosini-Berus, M. Kos-Golja, P. Peternel, B. Rozman, Influence of degraded phosphatidylserine on binding of antiphospholipid antibodies, Int. Arch. Allergy Immunol. 112 (1997) 19–26. J.A. McIntyre, Antiphospholipid antibodies in implantation failures, Am. J. Reprod. Immunol. 49 (2003) 221–229. J. Lozier, N. Takahashi, F.W. Putnam, Complete amino acid sequence of human plasma beta 2-glycoprotein I, Proc. Natl. Acad. Sci. USA 81 (1984) 3640–3644. R. Schwarzenbacher, K. Zeth, K. Diederichs, A. Gries, G.M. Kostner, P. Laggner, R. Prassl, Crystal structure of human beta2-glycoprotein I: Implications for phospholipid binding and the antiphospholipid syndrome, EMBO J. 18 (1999) 6228–6239. J. Hunt, S. Krilis, The fifth domain of beta 2-glycoprotein I contains a phospholipid binding site (Cys281-Cys288) and a region recognized by anticardiolipin antibodies, J. Immunol. 152 (1994) 653–659. B. Bouma, P.G. de Groot, J.M. van den Elsen, R.B. Ravelli, A. Schouten, M.J. Simmelink, R.H. Derksen, J. Kroon, P. Gros, Adhesion mechanism of human beta(2)-glycoprotein I to phospholipids based on its crystal structure, EMBO J. 18 (1999) 5166–5174. A. Chonn, S.C. Semple, P.R. Cullis, Association of blood proteins with large unilamellar liposomes in vivo. Relation to circulation lifetimes, J. Biol. Chem. 267 (1992) 18759–18765. A. Gries, J. Nimpf, H. Wurm, G.M. Kostner, T. Kenner, Characterization of isoelectric subspecies of asialo-beta 2-glycoprotein I, Biochem. J. 260 (1989) 531–534. G.M. Willems, M.P. Janssen, M.M. Pelsers, P. Comfurius, M. Galli, R.F. Zwaal, E.M. Bevers, Role of divalency in the high-affinity binding of anticardiolipin antibody-beta 2-glycoprotein I complexes to lipid membranes, Biochemistry 35 (1996) 13833–13842. S. Sodin-Sˇemrl, B. Rozman, b-2-glycoprotein I and its clinical significance: From gene sequence to protein levels, Autoimmun. Rev. 6 (2007) 547–552. H. Wurm, E. Beubler, E. Polz, A. Holasek, G. Kostner, Studies on the possible function of beta 2-glycoprotein-I: influence in the triglyceride metabolism in the rat, Metabolism 31 (1982) 484–486. K. Kobayashi, M. Kishi, T. Atsumi, M.L. Bertolaccini, H. Makino, N. Sakairi, I. Yamamoto, T. Yasuda, M.A. Khamashta, G.R. Hughes, T. Koike, D.R. Voelker, E. Matsuura, Circulating oxidized LDL forms complexes with beta2-glycoprotein I: implication as an atherogenic autoantigen, J. Lipid Res. 44 (2003) 716–726.

116

S. Sodin-Sˇemrl et al.

[42] D. Harats, J. George, Beta2-glycoprotein I and atherosclerosis, Curr. Opin. Lipidol. 12 (2001) 543–546. [43] J. George, N. Yacov, E. Breitbart, L. Bangio, A. Shaish, B. Gilburd, Y. Shoenfeld, D. Harats, Suppression of early atherosclerosis in LDL-receptor deficient mice by oral tolerance with beta 2-glycoprotein I, Cardiovasc. Res. 62 (2004) 603–609. [44] I. Schousboe, Beta 2-glycoprotein I: A plasma inhibitor of the contact activation of the intrinsic blood coagulation pathway, Blood 66 (1985) 1086–1091. [45] J. Nimpf, E.M. Bevers, P.H. Bomans, U. Till, H. Wurm, G.M. Kostner, R.F. Zwaal, Prothrombinase activity of human platelets is inhibited by beta 2-glycoprotein-I, Biochim. Biophys. Acta 884 (1986) 142–149. [46] J. Nimpf, H. Wurm, G.M. Kostner, Beta 2-glycoprotein-I (apo-H) inhibits the release reaction of human platelets during ADP-induced aggregation, Atherosclerosis 63 (1987) 109–114. [47] T. Mori, H. Takeya, J. Nishioka, E.C. Gabazza, K. Suzuki, Beta 2-glycoprotein I modulates the anticoagulant activity of activated protein C on the phospholipid surface, Thromb. Haemost. 75 (1996) 49–55. [48] F. Conti, M. Sorice, A. Circella, C. Alessandri, V. Pittoni, B. Caronti, C. Calderaro, T. Griggi, R. Misasi, G. Valesini, Beta-2-glycoprotein I expression on monocytes is increased in anti-phospholipid antibody syndrome and correlates with tissue factor expression, Clin. Exp. Immunol. 132 (2003) 509–516. [49] K. Ma, R. Simantov, J.C. Zhang, R. Silverstein, K.A. Hajjar, K.R. McCrae, High affinity binding of beta 2-glycoprotein I to human endothelial cells is mediated by annexin II, J. Biol. Chem. 275 (2000) 15541–15548. [50] E. Raschi, C. Testoni, D. Bosisio, M.O. Borghi, T. Koike, A. Mantovani, P.L. Meroni, Role of the MyD88 transduction signaling pathway in endothelial activation by antiphospholipid antibodies, Blood 101 (2003) 3495–3500. [51] S.K. Moestrup, I. Schousboe, C. Jacobsen, J.R. Leheste, E.I. Christensen, T.E. Willnow, Beta2-glycoprotein-I (apolipoprotein H) and beta2-glycoprotein-I-phospholipid complex harbor a recognition site for the endocytic receptor megalin, J. Clin. Invest. 102 (1998) 902–909. [52] M. van Lummel, M.T. Pennings, R.H. Derksen, R.T. Urbanus, B.C. Lutters, N. Kaldenhoven, P.G. de Groot, The binding site in beta2-glycoprotein I for ApoER20 on platelets is located in domain V, J. Biol. Chem. 280 (2005) 36729–36736. [53] K. Balasubramanian, A.J. Schroit, Characterization of phosphatidylserine-dependent beta2glycoprotein I macrophage interactions. Implications for apoptotic cell clearance by phagocytes, J. Biol. Chem. 273 (1998) 29272–29277. [54] N. Del Papa, Y.H. Sheng, E. Raschi, D.A. Kandiah, A. Tincani, M.A. Khamashta, T. Atsumi, G.R. Hughes, K. Ichikawa, T. Koike, G. Balestrieri, S.A. Krilis, P.L. Meroni, Human beta 2-glycoprotein I binds to endothelial cells through a cluster of lysine residues that are critical for anionic phospholipid binding and offers epitopes for anti-beta 2-glycoprotein I antibodies, J. Immunol. 160 (1998) 5572–5578. [55] M. Vega-Ostertag, K. Casper, R. Swerlick, D. Ferrara, E.N. Harris, S.S. Pierangeli, Involvement of p38 MAPK in the up-regulation of tissue factor on endothelial cells by antiphospholipid antibodies, Arthritis Rheum. 52 (2005) 1545–1554. [56] T. Atsumi, O. Amengual, S. Yasuda, E. Matsuura, T. Koike, Research around beta2-glycoprotein I: A major target for antiphospholipid antibodies, Autoimmunity 38 (2005) 377–381. [57] M. Merten, S. Motamedy, S. Ramamurthy, F.C. Arnett, P. Thiagarajan, Sulfatides: targets for anti-phospholipid antibodies, Circulation 108 (2003) 2082–2087. ˇ ucˇnik, T. Kveder, B. Rozman, B. Bozˇicˇ, Binding of high-avidity anti-beta2-glycoprotein I [58] S. C antibodies, Rheumatology (Oxford) 43 (2004) 1353–1356. [59] M. Galli, D. Luciani, G. Bertolini, T. Barbui, Lupus anticoagulants are stronger risk factors for thrombosis than anticardiolipin antibodies in the antiphospholipid syndrome: A systematic review of the literature, Blood 101 (2003) 1827–1832. [60] G. Reber, P. de Moerloose, Anti-beta2-glycoprotein I antibodies—when and how should they be measured? Thromb Res. 114 (2004) 527–531. [61] D. Faden, A. Tincani, P. Tanzi, L. Spatola, A. Lojacono, M. Tarantini, G. Balestrieri, Anti-beta 2 glycoprotein I antibodies in a general obstetric population: Peliminary results on the

Antiphospholipid Syndrome

[62] [63]

[64] [65] [66] [67] [68] [69] [70] [71] [72] [73] [74]

[75] [76] [77] [78] [79] [80] [81]

117

prevalence and correlation with pregnancy outcome. Anti-beta2-glycoprotein I antibodies are associated with some obstetrical complications, mainly preeclampsia-eclampsia, Eur. J. Obstet. Gynecol. Reprod. Biol. 73 (1997) 37–42. R.M. Lee, W. Emlen, J.R. Scott, D.W. Branch, R.M. Silver, Anti-beta2-glycoprotein I antibodies in women with recurrent spontaneous abortion, unexplained fetal death, and antiphospholipid syndrome, Am. J. Obstet. Gynecol. 181 (1999) 642–648. S. Miyakis, M.D. Lockshin, T. Atsumi, D.W. Branch, R.L. Brey, R. Cervera, R.H. Derksen, P.G. DE Groot, T. Koike, P.L. Meroni, G. Reber, Y. Shoenfeld, A. Tincani, P.G. Vlachoyiannopoulos, S.A. Krilis, International consensus statement on an update of the classification criteria for definite antiphospholipid syndrome (APS), J. Thromb. Haemost. 4 (2006) 295–306. D.A. Lane, H. Philippou, J.A. Huntington, Directing thrombin, Blood 106 (2005) 2605–2612. E.W. Davie, J.D. Kulman, An overview of the structure and function of thrombin, Semin. Thromb. Hemost. 32 (2006) 3–15. J.A. Huntington, Molecular recognition mechanisms of thrombin, J. Thromb. Haemost. 3 (2005) 1861–1872. M. Galli, T. Barbui, Antiprothrombin antibodies: Detection and clinical significance in the antiphospholipid syndrome, Blood 93 (1999) 2149–2157. J. Rauch, M. Tannenbaum, A.S. Janoff, Distinguishing plasma lupus anticoagulants from anti-factor antibodies using hexagonal (II) phase phospholipids, Thromb. Haemost. 62 (1989) 892–896. J.H. Rand, X.X. Wu, H.A. Andree, J.B. Ross, E. Rusinova, M.G. Gascon-Lema, C. Calandri, P.C. Harpel, Antiphospholipid antibodies accelerate plasma coagulation by inhibiting annexin-V binding to phospholipids: A ‘‘lupus procoagulant’’ phenomenon, Blood 92 (1998) 1652–1660. P. Raynal, H.B. Pollard, Annexins: The problem of assessing the biological role for a gene family of multifunctional calcium- and phospholipid-binding proteins, Biochim. Biophys. Acta 1197 (1994) 63–93. A. Markoff, V. Gerke, Expression and functions of annexins in the kidney, Am. J. Physiol. Renal Physiol. 289 (2005) 949–956. K.G. Maffey, L.B. Keil, V.A. DeBari, The influence of lipid composition and divalent cations on annexin V binding to phospholipid mixtures, Ann. Clin. Lab. Sci. 31 (2001) 85–90. H. Kenis, H. van Genderen, A. Bennaghmouch, H.A. Rinia, P. Frederik, J. Narula, L. Hofstra, C.P. Reutelingsperger, Cell surface-expressed phosphatidylserine and annexin A5 open a novel portal of cell entry, J. Biol. Chem. 279 (2004) 52623–52629. S. Ravassa, A. Bennaghmouch, H. Kenis, T. Lindhout, T. Hackeng, J. Narula, L. Hofstra, C. Reutelingsperger, Annexin A5 down-regulates surface expression of tissue factor: A novel mechanism of regulating the membrane receptor repertoire, J. Biol. Chem. 280 (2005) 6028–6035. N. Gasˇpersˇicˇ, A. Ambrozˇicˇ, B. Bozˇicˇ, J. Majhenc, S. Svetina, B. Rozman, Annexin A5 binding to giant phospholipid vesicles is differentially affected by anti-beta2-glycoprotein I and antiannexin A5 antibodies, Rheumatology (Oxford) 46 (2007) 81–86. B. Bozˇicˇ, S. Irman, N. Gasˇpersˇicˇ, T. Kveder, B. Rozman, Antibodies against annexin A5: Detection pitfalls and clinical associations, Autoimmunity 38 (2005) 425–430. A. Cederholm, J. Frostegard, Annexin A5 in cardiovascular disease and systemic lupus erythematosus, Immunobiology 210 (2005) 761–768. J.H.W. Distler, L.C. Huber, A.J. Hueber, C.F. Reich III, S. Gay, O. Distler, D.S. Pisetsky, The release of microparticles by apoptotic cells and their effects on macrophages, Apoptosis 10 (2005) 731–741. M. Diamant, M.E. Tushuizen, A. Sturk, A.R. Nieuwland, Cellular microparticles: New players in the field of vascular disease? Eur. J. Clin. Invest. 34 (2004) 392–401. J.H.W. Distler, D.S. Pisetsky, L.C. Huber, J.R. Kalden, S. Gay, O. Distler, Microparticles as regulators of imflammation, Arthritis Rheum. 52 (2005) 3337–3348. M. Baj-Krzyworzeka, R. Szatanek, K. Weglarczyk, J. Baran, B. Urbanowicz, P. Branski, M.Z. Ratajczak, M. Zembala, Tumour-derived microvesicles carry several surface determinants and mRNA of tumour cells and transfer some of these determinants to monocytes, Cancer Immunol. Immunother. 55 (2005) 808–818.

118

S. Sodin-Sˇemrl et al.

[82] V. Kralj-Iglicˇ, H. Ha¨gerstrand, P. Veranicˇ, K. Jezernik, B. Babnik, D.R. Gauger, A. Iglicˇ, Amphiphile-induced tubular budding of the bilayer membrane, Eur. Biophys. J. 34 (2005) 1066–1070. [83] W. Wiese, W. Harbich, W. Helfrich, Budding of lipid bilayer vesicles and flat membranes, J. Phys. Condens. Matter 4 (1992) 1647–1657. [84] R. Lipowsky, Budding of membranes induced by intramembrane domains, J. Phys. II (France) 2 (1992) 1825–1840. [85] A. Iglicˇ, H. Ha¨gerstrand, Amphiphile-induced spherical microexovesicles correponds to an extreme local area difference between two monolayers of the membrane bilayer, Med. Biol. Eng. Comput. 37 (1999) 125–129. [86] Z. Yamashita, S.M. Masum, T. Tanaka, Y. Tamba, M. Yamazaki, Shape changes of giant unilamellar vesicles of phosphatidiylcholine induced by a de novo designed peptide interacting with their membrane interface, Langmuir 18 (2002) 9638–9641. [87] M.M. Hanczyc, J.W. Szostak, Replicating vesicles as models of primitive cell growth and division, Curr. Opin. Chem. Biol. 8 (2004) 660–664. [88] M.P. Sheetz, S.J. Singer, Biological membranes as bilayer couples. A molecular mechanism of drug-erythrocyte interactions, Proc. Natl. Acad. Sci. USA 71 (1974) 4457–4461. [89] W. Helfrich, Blocked lipid exchange in bilayers and its possible influence on the shape of vesicles, Z. Naturforsch. 29c (1974) 510–515. [90] E. Evans, Bending resistance and chemically induced moments in membrane bilayers, Biophys. J. 14 (1974) 923–931. [91] W. Helfrich, Elastic properties of lipid bilayers—theory and possible experiments, Z. Naturforsch. 28c (1973) 693–703. [92] S. Svetina, B. Zˇeksˇ, Membrane bending energy and shape determination of phospholipid vesicles and red blood cells, Eur. Biophys. J. 17 (1989) 101–111. [93] P.B. Canham, The minimum energy of bending as a possible explanation of the biconcave shape of the human red blood cell, J. Theor. Biol. 26 (1970) 61–81. [94] A. Pralle, P. Keller, E.L. Florin, K. Simons, J.K. Horber, Sphingolipid-cholesterol rafts diffuse as small entities in the plasma membrane of mammalian cells, J. Cell Biol. 148 (2000) 997–1008. [95] I.A. Prior, C. Muncke, R.G. Parton, J.F. Hancock, Direct visualization of Ras proteins in spatially distinct cell surface microdomains, J. Cell. Biol. 160 (2003) 165–170. [96] P. Sharma, R. Varma, R.C. Sarasij, Ira, K. Gousset, G. Krishnamoorthy, M. Rao, S. Mayor, Nanoscale organization of multiple GPI-anchored proteins in living cell membranes, Cell 116 (2004) 577–589. [97] S.J. Plowman, C. Muncke, R.G. Parton, J.F. Hancock, H-ras, K-ras, and inner plasma membrane raft proteins operate in nanoclusters with differential dependence on the actin cytoskeleton, Proc. Natl. Acad. Sci. USA 102 (2005) 15500–15505. [98] K. Simons, D. Toomre, Lipid rafts and signal transduction, Nat. Rev. Mol. Cell. Biol. 1 (2000) 31–39. [99] K. Simons, W.L. Vaz, Model systems, lipid rafts, and cell membranes, Annu. Rev. Biophys. Biomol. Struct. 33 (2004) 269–295. [100] H. Ha¨gerstrand, L. Mrowczynska, U. Salzer, R. Prohaska, A.K. Michelsenn, V. Kralj-Iglicˇ, A. Iglicˇ, Curvature dependent lateral distribution of raft markers in the human erythrocyte membrane, Mol. Membr. Biol. 23 (2006) 277–288. [101] V. Kralj-Iglicˇ, V. Heinrich, S. Svetina, B. Zˇeks, Free energy of closed membrane with anisotropic inclusions, Eur. Phys. J. B. 10 (1999) 5–8. [102] V. Kralj-Iglicˇ, A. Iglicˇ, G. Gomisˇcˇek, F. Sevsˇek, V. Arrigler, H. Ha¨gerstrand, Microtubes and nanotubes of a phospholipid bilayer membrane, J. Phys. A: Math. Gen. 35 (2002) 1533–1549. [103] A. Iglicˇ, B. Babnik, K. Bohinc, M. Fosˇnaricˇ, H. Ha¨gerstrand, V. Kralj-Iglicˇ, On the role of anisotropy of membrane constituents in formation of a membrane neck during budding of a multicomponent membrane, J. Biomech. 40 (2007) 579–585. [104] V. Kralj-Iglicˇ, B. Babnik, D.R. Gauger, S. May, A. Iglicˇ, Quadrupolar ordering of phospholipid molecules in narrow necks of phospholipid vesicles, J. Stat. Phys. 125 (2006) 727–752. [105] B. Isomaa, H. Ha¨gerstrand, Shape transformations induced by amphiphiles in erythrocytes, Biochim. Biophys. Acta 899 (1987) 93–103.

Antiphospholipid Syndrome

119

[106] H. Ha¨gerstrand, B. Isomaa, Morphological characterization of exovesicles and endovesicles released in human erythrocytes following treatment with amphiphiles, Biochim. Biophys. Acta 1109 (1992) 117–126. [107] H. Ha¨gerstrand, B. Isomaa, Lipid and protein composition of exovesicles released from human erythrocytes following treatment with amphiphiles, Biochim. Biophys. Acta 1190 (1994) 409–415. [108] V. Kralj-Iglicˇ, A. Iglicˇ, H. Ha¨gerstrand, P. Peterlin, Stable tubular microexovesicles of the erythrocyte membrane induced by dimeric amphiphiles, Phys. Rev. E 61 (2000) 4230–4234. [109] T. Harder, K. Simons, Caveolae, DIGs, and the dynamics of sphingolipid-cholesterol microdomains, Curr. Opin. Cell Biol. 9 (1997) 534–542. [110] D. Allan, P. Thomas, A.R. Limbrick, The isolation and characterization of 60 nm vesicles (‘nanovesicles’) produced during ionophore A23187-induced budding of human erythrocytes, Biochem. J. 188 (1980) 881–887. [111] U. Salzer, P. Hinterdorfer, U. Hunger, C. Borken, R. Prohaska, Caþþ-dependent vesicle release from erythrocytes involves stomatin-specific lipid rafts, synexin (annexin VII), and sorcin, Blood 99 (2002) 2569–2577. [112] J.M. Holopainen, M.I. Angelova, P.K.J. Kinnunen, Vectorial budding of vesicles by asymmetrical enzymatic formation of ceramide in giant liposomes, Biophys. J. 78 (2000) 830–838. [113] H. Sprong, P. van der Sluijs, G. van Meer, How proteins move lipids and lipids move proteins, Nat. Rev. Mol. Cell Biol. 2 (2001) 504–513. [114] J.C. Holthius, G. van Meer, K. Huitema, Lipid microdomains, lipid translocation and the organization of intracellular membrane transport (review), Mol. Membr. Biol. 20 (2003) 231–241. [115] D. Tang, D. Borchman, N. Harris, S. Pierangeli, Lipid interactions with human antiphospholipid antibody, beta 2-glycoprotein 1, and normal human IgG using the fluorescent probes NBD-PE and DPH, Biochim. Biophys. Acta 1372 (1998) 45–54. [116] S.X. Wang, G. Cai, S. Sui, Intrinsic fluorescence study of the interaction of human apolipoprotein H with phospholipid vesicles, Biochemistry 38 (1999) 9477–9484. [117] H. Wurm, Beta 2-lycoprotein-I (apolipoprotein H) interactions with phospholipid vesicles, Int. J. Biochem. 16 (1984) 511–515. [118] M. Igarashi, E. Matsuura, Y. Igarashi, H. Nagae, K. Ichikawa, D.A. Triplett, T. Koike, Human beta2-glycoprotein I as an anticardiolipin cofactor determined using mutants expressed by a baculovirus system, Blood 87 (1996) 3262–3270. [119] M.I. Angelova, S. Soleau, P. Meleard, J.F. Faucon, P. Bothorel, Preparation of giant vesicles by external AC electric fields. Kinetics and applications, Prog. Colloid Polym. Sci. 89 (1992) 127–131. [120] A. Ambrozˇicˇ, B. Bozˇicˇ, T. Kveder, J. Majhenc, V. Arrigler, S. Svetina, B. Rozman, Budding, vesiculation and permeabilization of phospholipid membranes-evidence for a feasible physiologic role of beta2-glycoprotein I and pathogenic actions of anti-beta2-glycoprotein I antibodies, Biochim. Biophys. Acta 1740 (2005) 38–44. [121] N. Tomsˇicˇ, B. Babnik, D. Lombardo, B. Mavcˇicˇ, M. Kandusˇer, A. Iglicˇ, V. Kralj-Iglicˇ, Shape and size of giant unilamellar phospholipid vesicles containing cardiolipin, J. Chem. Inf. Model. 45 (2005) 1676–1679. [122] M. Mally, J. Majhenc, S. Svetina, B. Zˇeksˇ, Mechanisms of equinatoxin II-induced transport through the membrane of a giant phospholipid vesicle, Biophys. J. 83 (2002) 944–953. [123] E. Polz, G.M. Kostner, The binding of beta 2-glycoprotein-I to human serum lipoproteins: Distribution among density fractions, FEBS Lett. 102 (1979) 183–186. [124] T. McNally, I.J. Mackie, D.A. Isenberg, S.J. Machin, Immunoelectrophoresis and ELISA techniques for assay of plasma beta 2-glycoprotein-I and the influence of plasma lipids, Thromb. Res. 72 (1993) 275–286. [125] A. Ambrozˇicˇ, T. Avcˇin, K. Ichikawa, T. Kveder, E. Matsuura, M. Hojnik, T. Atsumi, B. Rozman, T. Koike, Anti-beta(2)-glycoprotein I antibodies in children with atopic dermatitis, Int. Immunol. 14 (2002) 823–830. [126] R. Lipowsky, The conformation of membranes, Nature 349 (1991) 475–481.

120

S. Sodin-Sˇemrl et al.

ˇ ucˇnik, N. Tomsˇicˇ, J. Urbanija, M. Lokar, B. Babnik, B. Rozman, A. Iglicˇ, [127] A. Ambrozˇicˇ, S. C V. Kralj-Iglicˇ, Interaction of giant phospholipid vesicles containing cardiolipin and cholesterol with b2-glycoprotein-I and anti-b2-glycoprotein-I antibodies, Autoimmun. Rev. 6 (2006) 10–15. [128] J.D. Jackson, Classical Electrodynamics (3rd ed.), John Wiley & Sons, 1998. [129] K. Bohinc, V. Kralj-Iglicˇ, T. Slivnik, S. May, A. Iglicˇ, Spatial distribution of divalent rod-like ions in contact with a charged surface, Electrotech. Rev. 73 (2006) 137–142. [130] K. Bohinc, A. Iglicˇ, S. May, Interaction between macroions mediated by divalent rodlike ions, Europhys. Lett. 68 (2004) 494–500. [131] V. Kralj-Iglicˇ, A. Iglicˇ, A simple statistical mechanical approach to the free energy of the electric double layer including the excluded volume effect, J. Phys. II (France) 6 (1996) 477–491. [132] D.F. Evans, H. Wennerstro¨m, The Colloidal Domain, Where Physics, Chemistry, and Biology Meet (2nd ed.), VCH Publishers, 1994.

C H A P T E R

F I V E

The Single GUV Method to Reveal Elementary Processes of Leakage of Internal Contents from Liposomes Induced by Antimicrobial Substances Masahito Yamazaki1,2,3,* Contents 1. Introduction 2. Experimental Methods of the Single GUV Method 2.1. Preparation of GUVs 2.2. The Method to Induce the Interaction of Substance Solution with a Single GUV 3. Effect of Antimicrobial Peptide, Magainin 2, on Membrane Permeability and Membrane Structure 4. Effect of Antibacterial Tea Catechin, EGCg, on Membrane Permeability and Membrane Structure 5. Conclusion and Advantage of the Single GUV Method Acknowledgments References

122 125 125 126 127 134 139 141 141

Abstract So far almost all studies of biomembrane function have been done using a suspension of many small-size vesicles such as large unilamellar vesicles (LUVs) (the conventional LUV suspension method). In these studies, the average values of the physical parameters of vesicles have been obtained from a large number of vesicles, and thereby much information has been lost. Recently we have proposed a novel method, the single GUV method. In this method, we observe and measure physical properties of single giant unilamellar vesicles (GUVs) with a diameter of 10 mm, and analyze these results over many GUVs statistically, which will provide a great deal of new information that cannot be obtained by the conventional LUV suspension method. In this review, we describe an application of * Corresponding author. Tel.:/Fax: +81 54 238 4741; E-mail address: [email protected] (M. Yamazaki). 1

2 3

Integrated Bioscience Section, Graduate School of Science and Engineering, Shizuoka University, 836 Oya, Suruga-ku, Shizuoka 422–8529, Japan Department of Physics, Faculty of Science, Shizuoka University, Shizuoka 422–8529, Japan Innovative Joint Research Center, Shizuoka University, Hamamatsu 432–8011, Japan

Advances in Planar Lipid Bilayers and Liposomes, Volume 7 ISSN 1554-4516, DOI: 10.1016/S1554-4516(08)00005-7

#

2008 Elsevier Inc. All rights reserved.

121

122

M. Yamazaki

the single GUV method to investigations of leakage of internal contents such as small fluorescent probes from liposomes. Such investigations using the conventional LUV suspension method have been extensively used to obtain information on interactions of various kinds of substances (e.g., antimicrobial substances, peptides/proteins, and drugs) with biomembranes/lipid membranes. However, this method could not reveal main cause of the leakage and its elementary steps. Here, we show two examples of the single GUV method studies on the leakage of internal contents from liposomes induced by antimicrobial peptide, magainin 2, and antibacterial substance, tea catechin. We have succeeded, for the first time, in observing detailed elementary processes of the substance-induced leakage and also obtaining a direct evidence for the cause of the leakage. The statistical analysis of individual events in single GUVs over many single GUVs gave us important information on the rate constants of elementary processes, such as the rate constant of the magainin 2-induced pore formation and that of the EGCginduced bursting of GUVs. These results clearly show that the single GUV method is much more useful than the conventional LUV method in studies of leakage of internal contents from liposomes. We also discuss the advantage of the single GUV method in biomembrane research.

1. Introduction Liposomes or vesicles of lipid membranes have been extensively used as model biomembranes and as drug delivery systems (DDS) and biosensors. Conventionally, unilamellar vesicles of lipid membranes are classified by their size; small unilamellar vesicles (SUVs) with a diameter of 50 nm, large unilamellar vesicles (LUVs) with a diameter of 100 nm–10 mm, and giant liposomes or giant unilamellar vesicles (GUVs) with a diameter of 10 mm. So far almost all studies of structure and function of biomembranes/lipid membranes have been conducted using a suspension of many small liposomes such as LUVs or SUVs and also multilamellar vesicles (MLVs) and various biophysical techniques such as static and dynamic light scattering, fluorescence spectroscopy, electron spin resonance, and X-ray scattering (from now on, we call this method as the conventional LUV suspension method). In these studies, the average values of the physical parameters of liposomes such as size, shape, and fluorescence intensity have been obtained from a large number of liposomes, and thereby much information has been lost (Fig. 1). For example, for measurement using fluorescence spectroscopy, which is one of the most sensitive methods, we need a suspension (1 ml) of 50 mM lipid concentration, corresponding 1011 LUVs with a diameter of 200 nm. Thereby, from such measurement we obtain an ensemble average of fluorescence properties (such as intensity and spectrum) of each LUV over 1011 LUVs. When we measure static structures (size and shape) of vesicles, we can obtain the average size and the average shapes of many liposomes. Interactions of peptides/proteins and various substances with lipid membranes induce many kinds of changes in physical properties of size, shapes, and fluorescence intensity, and also various events such as membrane fusion and vesicle fission. Generally, such kinds of events do not occur simultaneously in all the liposomes, and thereby, we measure an ensemble average of various kinds of

Single GUV Method to Investigate Leakage of Internal Contents from Liposomes

Figure 1

123

A schematic diagram of the principle of the conventional LUV suspension method.

stages of these events. For example, in membrane fusion, we measure physical properties such as fluorescence intensity of the mixture of vesicles at various stages such as after fusion, during fusion, at association state before the membrane fusion, and in original dispersed state. From these measurements, we cannot obtain information on their elementary processes directly. On the other hand, GUVs of lipid membranes have been used for investigations of the physical properties of vesicle membranes such as elasticity [1–4]. The structure of a single GUV and its physical properties in water can be measured in real time, and hence GUVs have a great advantage over smaller liposomes such as LUVs and SUVs in investigating physical properties and structural changes of liposomes. Recently, we have proposed a novel method, ‘‘the single GUV method’’, to probe functions and dynamics of biomembranes/lipid membranes [5–8]. In the single GUV method, we observe structures and physical properties of single GUVs as a function of time and spatial coordinates using various optical microscopes (e.g., phase contrast microscope, fluorescence microscope, differential interference contrast microscope), and make a statistical analysis of the physical parameters of a single GUV over many single GUVs (Fig. 2). Thereby, the single GUV method has great possibilities to obtain new information on structure and function of biomembranes/lipid membranes and also on interactions of substances with biomembranes, which cannot be obtained by the conventional LUV suspension method. To investigate interactions of substances such as peptides/proteins and low molecular weight antimicrobial substances with single GUVs, we add various concentrations of substance solution (in the same buffer as that outside the GUVs) slowly in the vicinity of a single GUV through a 20 mm diameter glass micropipet whose position is precisely controlled by a micromanipulator (Fig. 3). In this case, the substance concentration in the surrounding solution of a single GUV becomes almost the same concentration as that in the micropipet. We have succeeded in revealing new aspects of interactions of substances with lipid membranes and also of biomembrane dynamics such as membrane fusion and vesicle fission. For example, using the single GUV method, we revealed interactions of peptides and ions with the membrane interface of electrically

124

M. Yamazaki

Figure 2

A schematic diagram of the principle of the single GUV method.

Figure 3 A scheme of the method to induce interactions of substances with a single GUV. Various concentrations of substance solution (in the same buffer as that outside the GUVs) are added slowly in the vicinity of a single GUV through a 20-mm diameter glass micropipet whose position is precisely controlled by a micromanipulator.

neutral phospholipid membranes [9, 10]. We succeeded in observing a detailed process of membrane fusion between two GUVs for the first time, and on the basis of this result we have proposed a new mechanism of the membrane fusion [6]. We found that low concentrations of single long-chain amphiphiles such as lysophosphatidylcholine (lyso-PC) and lysophosphatidic acid (lyso-PA) induced vesicle fission of single GUVs of lipid membranes in the liquid-ordered phase, and analyzed its mechanism [7, 11]. In our previous review [5], we summarized these results. In this review, we describe an application of the single GUV method to investigations of leakage of internal contents such as a small fluorescent probe from liposomes. Such investigations using the conventional LUV suspension method have been extensively used to obtain information on interactions of various

Single GUV Method to Investigate Leakage of Internal Contents from Liposomes

125

kinds of substances [e.g., antimicrobial substances, peptides/proteins, drugs, and fusogens (which induce membrane fusion of vesicles)] with biomembranes/lipid membranes [e.g., 12–16]. A large amount of leakage indicates that the substance strongly interacts with lipid membranes, inducing instability in structures of vesicles and lipid membranes. However, in the conventional LUV suspension method, we cannot determine the main cause of the leakage, because there are many possible factors that can be involved in the leakage; for example, instability of membrane structure due to large deformation of vesicles or membrane fusion, formation of a narrow pore or ion channels, and disruption of vesicles. Recently substances having antibacterial or antimicrobial activities have attracted much attention in basic bioscience and also in biotechnology. It is well recognized that lipid membranes are one of the targets of the antibacterial activity of these substances. Many researchers have reported that these substances induced leakage of small fluorescent probes such as calcein and carboxyfluorescein from the inside of LUVs and SUVs, indicating that they caused damage to lipid membranes. However, the mechanism of their bactericidal activity is controversial and the details of the interaction between these substances and lipid membranes remain unclear. Here we show two examples of the single GUV method studies on effects of antimicrobial substances on structure and physical property (such as membrane permeability) of lipid membranes; one is antimicrobial peptide, magainin 2, and the other is antibacterial substance, tea catechin.

2. Experimental Methods of the Single GUV Method 2.1. Preparation of GUVs We prepared GUVs by the natural swelling of a dry lipid film using the prehydration [9, 10]. GUVs of electrically neutral lipid membranes such as phosphatidylcholine (PC) membrane can be formed easily in water, but cannot be formed in buffers containing salts. However, using the PEG lipid method that we previously developed [17], we can easily prepare these GUVs in buffers containing high concentrations (up to 2.0 M) of salts such as NaCl and CaCl2. In the PEG lipid method, hydrophilic polymers are attached on the surface of lipid membranes by including a small amount (0.5–2.0 mol%) of poly-(ethylene glycol) [PEG]-grafted phospholipid such as PEG2K-DOPE (dioleoylphosphatidylethanolamine) in PC membrane. The hydrophilic polymers attached on the membrane surface increase the intermembrane distance in multilayers of lipid membranes when GUVs are formed, which makes it possible to form GUVs in high ionic strength. For experiments of the interaction between catechin and single PC-GUVs, we prepared GUVs of egg PC membrane containing 1 mol% PEG2K-DOPE in buffer H (10 mM HEPES, pH 7.4, 150 mM NaCl) containing 0.1 M sucrose. On the other hand, we can prepare GUVs of lipid membranes containing high concentrations of negatively charged lipid such as dioleoylphosphatidylglycerol (DOPG) in a buffer containing 150 mM NaCl without using the PEG lipid method. This is probably due to the electrostatic repulsive interaction between neighboring

126

M. Yamazaki

membranes in the multilayers of lipid membranes. For experiments of the interaction between magainin 2 and single GUVs, we prepared 50%-DOPC/50%-DOPGGUVs in buffer A (10 mM PIPES, pH 7.0, 150 mM NaCl, and 1 mM EGTA) containing 0.1 M sucrose. It is difficult to observe GUVs of lipid membranes by the phase contrast microscopy due to their low contrast. Thereby, to increase the contrast, we added 0.1 M sucrose in solution inside the GUV and 0.1 M glucose solution outside the GUV. For this purpose, in most cases, we diluted a GUV suspension prepared in a buffer containing 0.1 M sucrose into the same buffer containing 0.1 M glucose. For example, a 10-ml aliquot of a GUV suspension in a buffer containing 0.1 M sucrose (internal solution) was diluted into 290 ml of the same buffer containing 0.1 M glucose (external solution). As GUVs, we selected vesicles whose membranes had a low contrast and a high undulation motion. The more reliable method to select a GUV is that a small amount of a fluorescent lipid probe is included in GUV membranes and we select GUVs whose membrane’s fluorescence intensity is the lowest in fluorescence microscopic images. The best method to select a GUV is that we measure the isothermal area expansion modulus KA by the micropipet aspiration technique [1] and select GUVs with KA values corresponding to a unilamellar vesicle. For the leakage experiment, we prepared GUVs containing a fluorescence probe such as calcein as follows [8, 18]. At first, we prepared GUVs in a buffer containing 1 mM calcein. Then, the GUV suspension was centrifuged at 14,000  g for 10 min at 20  C to remove MLVs. The supernatant was collected and passed through a Sephadex G-75 column with the same buffer containing 0.1 M glucose, and fractions containing GUVs were collected.

2.2. The Method to Induce the Interaction of Substance Solution with a Single GUV A GUV suspension (300 ml) was transferred into a hand-made microchamber (1 cm  1 cm wide, 3 mm high, internal volume is 0.3 ml) which had been formed on a glass slide by inserting a U-shaped silicone-rubber spacer between a cover slip and the glass slide (Fig. 4) [9, 10]. To prevent strong interaction between the glass surface (of the cover slip and the glass slide) and GUVs, in most cases, the glass slide and the cover slip inside the microchamber were coated with 0.1% (w/v) BSA in the same buffer as that outside of the GUVs. We observed GUVs by an inverted fluorescence, phase contrast microscope (IX-70, Olympus, Tokyo, Japan). The images of GUVs were recorded through a CCD camera on a video recorder. The fluorescence intensity inside of GUVs was measured using Image J (Ver.1.33u; National Institute of Health, Bethesda, MD), and the average intensity per GUV was estimated. Various concentrations of substance solution (in the same buffer as that outside the GUVs) were continuously added slowly into the vicinity of a single GUV via a glass micropipet (diameter, 10–20 mm) whose position was controlled precisely by a micromanipulator (e.g., MMW-23, Narishige, Tokyo, Japan). At first, we selected a GUV under the microscope, and approached the tip of the micropipet near the

Single GUV Method to Investigate Leakage of Internal Contents from Liposomes

127

Differential pressure transducer

Pressure amplifier

Digital multimeter

Micropipet

Coverslip

Slide glass Silicone-rubber spacer

Figure 4 The apparatus to induce the interaction of substance solution with a single GUV in the single GUV method.

GUV using the micromanipulator. To apply a small pressure on the inside of the micropipet to add the substance solution in the vicinity of a GUV, we used a handmade apparatus for the micropipet aspiration technique (Fig. 4). We controlled the injection pressure by changing the height of vertical column of water in a U-shaped glass tube to which the micropipet was hydraulically connected (Fig. 4). The glass micropipet was prepared by pulling 1.0-mm glass tubing (G-1, Narishige, Tokyo, Japan) to a needlepoint using a puller (PP-83, Narishige, Tokyo, Japan), and then breaking it by quick fracture to a desired tip diameter. Then, the micropipet was filled with various concentrations of substance solution (in the same buffer as that outside the GUVs) by aspiration using a vacuum pump.

3. Effect of Antimicrobial Peptide, Magainin 2, on Membrane Permeability and Membrane Structure Antimicrobial peptides with bactericidal and fungicidal activity have been found and isolated from a wide variety of organisms, including amphibians, invertebrates, plants, and mammals [19]. Among these antimicrobial peptides, magainin 2, which was at first isolated from the African clawed frog Xenopus laevis [20], has been extensively investigated. The sequence of magainin 2 (23-mer) is GIGKFLHSAKKFGKAFVGEI-MNS, and it has an amide-blocked C terminus, and thereby it is a positively charged peptide. Magainin 2 binds with negatively charged membrane of bacteria due to electrostatic attraction, and forms an a-helix in the lipid membranes interface parallel to the membrane surface [21]. Figure 5 shows a 3-D

128

M. Yamazaki

structure of magainin 2 in dodecylphospocholine micelles, which is one of the 10 lowest energy structures calculated from the NMR data [22]. According to Fig. 5, four lysine (Lys) residues face almost the same side, and many hydrophobic amino acid residues including three phenylalanine (Phe) residues face the opposite side, indicating that magainin 2 forms an amphipathic a-helix in the lipid membrane interface. So far many researchers have investigated interaction of magainin 2 with lipid membranes using the conventional LUV suspension method. They investigated interaction of magainin 2 with LUVs containing internal contents such as a small fluorescence probe (e.g., calcein), and measured the magainin 2-induced leakage of the internal contents from many LUVs in a suspension [13, 14]. Figure 6 shows a typical data of magainin 2-induced leakage of calcein from a suspension of 50 mol%DOPG/50 mol%-DOPC-LUV (shortly 50%-DOPG/DOPC-LUV) (diameter 200 nm) [8]. Magainin 2 induced gradual leakages over a 10–20 min period, and the rate of leakage increased with an increase in magainin 2 concentration. So far this data has been interpreted as that the gradual leakage occurred from each LUV over the long time. However, the detail characteristics of the membrane permeability remained unclear.

Figure 5 A 3-D structure of magainin 2 in dodecylphosphocholine micelles, which is one of the 10 lowest energy structures calculated from the NMR data. (Drawn from 2MAG.pdb. Ref. [22].) 10 μM

100

7

Leakage (%)

80

5

60 40

4

20

3

0

0 0

5

10 15 Time (min)

20

Figure 6 Time course of magainin 2-induced leakage of calcein from a suspension of 50% DOPG/DOPC-LUVs after the addition of various concentrations of magainin 2. ( ) 0 mM, (□) 3 mM, () 4 mM, (D) 5 mM, (▲) 7 mM, and (○) 10 mM magainin 2. This figure is reprinted from Ref. [8] with permission from the American Chemical Society.



Single GUV Method to Investigate Leakage of Internal Contents from Liposomes

129

To reveal the effect of magainin 2 on membrane permeability, we investigated the interaction of magainin 2 with single DOPG/DOPC-GUVs containing calcein using the single GUV method. We added various concentrations of magainin 2 in buffer A (10 mM PIPES, pH 7.0, 150 mM NaCl, and 1 mM EGTA) containing 0.1 M glucose at room temperature (20  2  C) in the vicinity of a single 50%DOPG/DOPC-GUV containing 1 mM calcein, and observed the fluorescence intensity inside the GUV due to calcein and the structure of the GUV during the interaction of magainin 2 with the GUV. At first, we investigate the effect of 7 mM magainin 2 on single 50%-DOPG/ DOPC-GUVs. Before the addition, the GUV had a high contrast in a phase contrast microscopic image [Fig. 7(1)] due to the difference in concentrations of sucrose and glucose between the inside and the outside of the GUV [i.e., 0.1 M sucrose (inside the GUV) and 0.1 M glucose (outside the GUV)]. A fluorescence microscopic image of the same GUV [Fig. 7(2)] shows that there was a high concentration of calcein inside the GUV at this time. At 42 s after starting the addition of the 7 mM magainin 2 near the GUV, the fluorescence intensity began to decrease rapidly. After 47 s, fluorescence was not detected inside the GUV, but a phase contrast image of the same GUV [Fig. 7(3)] shows that the GUV was not broken and still existed. These results clearly show that the leakage of calcein from the inside to the outside of the GUV did not occur as a result of the disruption of the GUV. During the leakage of calcein, the GUV did not associate with another GUV and the shape of the GUV did not change (i.e., it was always spherical), indicating that the leakage of calcein did not occur as a result of instability of the membrane structure due to large deformation of the associated vesicles [12]. Thereby, we can reasonably consider that calcein leaked from the GUV because of the formation of a pore (or pores) in the membrane by magainin 2. These results, therefore, provide a direct evidence for the pore formation in the membrane by magainin 2 as proposed by Matsuzaki et al. based on their extensive studies using the conventional LUV suspension method [13, 23]. Comparison of the phase contrast images in Fig. 7(1) and (3) also shows that there was a substantial loss in the contrast of the GUV, indicating that, during the leakage of the calcein, sucrose and glucose passed through the membrane, apparently through the same pore (or pores) that allowed the calcein leakage.

Figure 7 Leakage of calcein from single 50%DOPG/DOPC-GUVs induced by 7 mM magainin 2. Fluorescence images (2) show that the calcein concentration inside the GUV progressively decreased after the addition of magainin 2. The numbers above each image show the time after the addition. Also shown are phase contrast images of the GUV at 0 (1) and at 47 s (3). The bar corresponds to 10 mm. This figure is reprinted from Ref. [8] with permission from the American Chemical Society.

130

M. Yamazaki

Next, we investigate the effect of lower concentration (4 mM) of magainin 2 on single 50%-DOPG/DOPC-GUVs. During the addition of 4 mM magainin 2, the fluorescence intensity inside the GUV gradually decreased for the first 135 s, after which there was a rapid decrease between 135 and 170 s [Fig. 8A(2) and B(○)]. After 200 s, there was no fluorescence intensity inside the GUV, although the phase contrast image [Fig. 8A(3)] showed that the GUV was not broken and not deformed. In contrast, in some GUVs under the same condition, there was only a continual gradual decrease in the fluorescence intensity during the addition of 4 mM magainin 2 [Fig. 8B()]. Other data shows that this gradual decrease in the fluorescence intensity was due to the photobleaching of calcein. The phase contrast image of the GUV [Fig. 8A(3)] shows a large decrease in the contrast as in Fig. 8A(1), indicating that, during the leakage of the calcein, sucrose and glucose passed through the pore from which calcein leaked.

A

C

Fluorescence intensity

Fluorescence intensity

B

135 s

170 s

0

100 200 Time (s)

300

0

50 100 150 200 250 300 350 Time (s)

Figure 8 Leakage of calcein from single 50%DOPG/DOPC-GUVs induced by 4 mM magainin 2. (A) Fluorescence images (2) show that the calcein concentration inside the GUV progressively decreased after the addition of magainin 2. The numbers above each image show the time after the addition. Also shown are phase contrast images of the GUV at 0 (1) and at 200 s (3). (B) Time course of the change of the fluorescence intensity of the GUV shown in (A). (C) Other examples of change in the fluorescence intensity of the single GUVs with time under the same condition. The bar corresponds to 20 mm. These figures (A, B, C) are reprinted from Ref. [8] with permission from the American Chemical Society.

131

Single GUV Method to Investigate Leakage of Internal Contents from Liposomes

Figure 8C shows the changes in fluorescence intensity of several single GUVs under the same conditions as in Fig. 8B. After some lag time, the rapid decrease in the fluorescence intensity started stochastically. Thereby, this result indicates that the rapid leakage of the fluorescent probe from a single GUV started stochastically, and once it began, the complete leakage occurred rapidly within 30 s. This result indicates that the pore formation is the rate-determining step, rather than the leakage through the pore, and also that the pore formation by magainin 2 occurred stochastically. When we examined 16 single GUVs of the same membrane by the method described above, we found a similar rapid complete leakage of calcein from the GUVs within 5 min in 11 out of 16 GUVs. Thus, the probability that a GUV did not contain calcein at 5 min under the same conditions was 0.69. In the estimation of the leakage of the internal contents of single liposomes, an important factor is the probability, PL(t), that a given liposome does not contain calcein molecules at any given time, t. Such empty liposomes are produced by the complete leakage of calcein from the liposome at any time within t. Thereby, we call PL(t) the fraction of completely leaked GUV at t among all the examined single GUVs; more concisely, we call it the fraction of leaked GUV. The fraction of leaked GUV, PL, increased with time (Fig. 9A). These findings suggest that the increase in the fraction of completely leaked LUVs with time, that is, the increase in the number of leaked LUVs with time, is responsible for the gradual increase in the leakage from the suspension of many LUVs over time (Fig. 6). Figure 9B(○) shows the dependence of PL(t) of 50%DOPG/DOPC-GUVs on the magainin 2 concentration. PL(t) increased as the magainin 2 concentration increased; in the case of t ¼ 5 min, at and less than 2 mM, magainin 2 did not induce the leakage [i.e., PL(5 min) ¼ 0], and PL(5 min) at 3 and 4 mM were 0.10  0.06 and 0.57  0.11, respectively, and at 7 mM, PL(5 min) was 1.0. Thereby, this result suggests that the pore formation of magainin 2 in the membrane greatly depended on the magainin 2 concentration in the membrane interface. B

Fraction of leaked GUV

1.0

7 μM 5 μM

10 μM

0.8 4 μM

0.6 0.4 0.2 0.0

3 μM 0 μM

10 mM 0

1

2 3 4 Time (min)

5

6

Fraction of leaked GUV

A

1.0

5 min

0.8

1 min

0.6 0.7 min

0.4 0.2 0.0 0

2 4 6 8 10 Magainin 2 conc. (μM)

12

Figure 9 (A) Time course of the fraction of completely leaked GUV, PL(t), among single 50% DOPG/DOPC-GUVs after the addition of various concentrations of magainin 2. ( ) 0 mM, (□) 3 mM, () 4 mM, (D) 5 mM, (▲) 7 mM, and (○) 10 mM magainin 2. (B) Dependence of PL(t) on magainin 2 concentration. PL(t) at t ¼ 5 min (○), t ¼ 1 min (▼), and t ¼ 0.7 min (▲) after the addition of magainin 2.



132

M. Yamazaki

We performed several experiments to clarify the mechanism of the pore formation by magainin 2. Analysis of shape changes of GUVs is a highly sensitive method for detecting the interaction of substances with lipid membrane [9, 10]. Low concentration of magainin 2 (e.g., 0.5 mM) transformed a prolate shape of 50%DOPG/DOPC-GUV into a dumbbell shape and then into two spheres connected by a neck. The shape of a GUV of a lipid membrane is determined by the minimization of the elastic energy of the closed membrane. It is considered that the bilayer-couple model and also generalized bilayer-couple model (i.e., the area– difference–elasticity model: ADE model) effectively explain shape changes of a single GUV [24–26]. The analysis of the magainin 2-induced shape change based on these models indicates that the binding of magainin 2 to the external monolayer of the GUV increased its membrane area. The increase in the area can be explained as follows. Magainin 2 has many Phe and leucine (Leu) residues, which have high interfacial hydrophobicity [27]; therefore, after the binding of magainin 2 to the negatively charged membrane interface due to the electrostatic attraction, magainin 2 can then be partitioned deeply into the membrane interface owing to its high interfacial hydrophobicity (see Fig. 5). It induces an increase in the steric repulsion between the peptide and the hydrophilic segments of lipids in the membrane interface, which increases the area of the external monolayer. In our previous reports, we clearly showed that a de novo designed peptide (WLFLLKKK) containing a segment with high interfacial hydrophobicity (WLFLL) can bind to the electrically neutral membrane interface of DOPC membranes and monoolein membranes, increasing the area of monolayer membrane [10, 28]. The increase in the area of the external monolayer due to the deep insertion of magainin 2 into the membrane interface is greater than the decrease in its area due to the electric neutralization at the binding of positively charged magainin 2 to negatively charged lipid membrane. Such binding of magainin 2 to the external monolayer of the GUV also raises its surface pressure. On the other hand, the addition of lyso-PC into the external monolayer of GUVs increased PL at all concentrations of magainin 2, suggesting that the increase in the surface pressure in the external monolayer increases the rate of the pore formation. On the basis of these results, we have proposed a hypothesis of the two-state transition model to describe the magainin 2-induced pore formation in the membrane. The first state (Bex state) is the binding state of magainin 2 to the external monolayer membrane of the liposomes, where magainin 2 forms the a-helix parallel to the membrane surface, and the second state (P state) is the pore state in the membrane. The rate constant of transition from the Bex state to the P state, kP, depends on its activation energy. As shown in Fig. 8, individual events of the two-state transition occurred stochastically. To estimate the rate constant kP, we have to obtain the fraction of the Bex state of GUVs among all the examined GUVs. If we assume that the pore formation occurred by the irreversible two-state transition from the Bex state to the P state, the fraction of the Bex state can be determined by the fraction of intact GUV which calcein did not leak (shortly the fraction of intact GUV), Pintact(t). Experimentally we obtained the time at which the rapid decrease in the fluorescence intensity of a GUV started (Fig. 8), and we define the state of intact GUV before this time as the Bex state. We plotted the time course of

Single GUV Method to Investigate Leakage of Internal Contents from Liposomes

0 μM 3 μM

1.0 Fraction of intact GUV

133

0.8 0.6 4 μM

0.4 0.2 0.0

5 μM 7 μM

10 μM

0

1

2 3 4 Time (min)

5

6

Figure 10 Time course of the fraction of intact GUV, Pintact(t), after the addition of various concentrations of magainin 2. ( ) 0 mM, (□) 3 mM, () 4 mM, (D) 5 mM, (▲) 7 mM, and (○) 10 mM magainin 2. Solid lines represent the best fitted curves of Eq. (1). This figure is reprinted from Ref. [8] with permission from the American Chemical Society.



the fraction of intact GUV in the presence of various concentrations of magainin 2 (Fig. 10). The fraction of the Bex state can be expressed using the rate constant, kP, as follows:

Pintact ðtÞ ¼ expf kP ðt teq Þg

ð1Þ

where teq is an adjustable parameter. At present, we do not know the exact time required for the equilibration of the magainin 2 concentration in the vicinity of a GUV and the time required for the binding equilibrium of magainin 2 from aqueous solution to the membrane interface of the GUV. We can reasonably consider that teq is the time when the binding equilibrium of magainin 2 is attained, and thereby the state of the GUV becomes the Bex state at teq. As shown in Fig. 10, all the curves of the time course of Pintact(t) were well fitted by a single exponential decay curve defined by Eq. (1). The rate constant increased with an increase in magainin 2 concentration, and kP for 10 mM magainin 2 (1.6 min 1) was about 10 times larger than kP of 4 mM magainin 2 (0.18 min 1). These results suggest that the pore formation induced by magainin 2 occurred by the two-state transition. At present, we have a hypothesis of the mechanism for the magainin 2-induced pore formation in the lipid membrane as follows. With an increase in the amount of the magainin 2 bound to the external monolayer of the GUV, the surface pressure of the external monolayer increases, destabilizing the Bex state. Hence, it increases the free energy of the Bex state, causing a decrease in the activation energy Ep. As a result, the rate constant of the transition from the Bex state to the P state increases as the magainin 2 concentration in the membrane interface of the external monolayer is raised. To reveal the mechanism for the magainin 2-induced pore formation clearly, we need additional data and analysis.

134

M. Yamazaki

4. Effect of Antibacterial Tea Catechin, EGCg, on Membrane Permeability and Membrane Structure Tea catechins are flavonoids, and are the main components of green tea extracts (Fig. 11). Some tea catechins have been found to have antibacterial activity (e.g., against Streptococcus mutans and MRSA), and antioxidant activity [29, 30]. (–)-Epigallocatechin gallate (EGCg) [Fig. 11(D)] is the most abundant catechin in tea extract, and has the strongest antibacterial activity among tea catechins [16]. Several studies indicate that catechins bind to lipid membranes, and that lipid membranes are one of the targets of the antibacterial activity of catechins. Many researchers reported that catechins induce gradual leakage of small fluorescent probes such as calcein from the inside of LUVs of PC membranes, indicating that catechins cause damage to lipid membranes [15, 16]. Figure 12 shows a typical data of EGCg-induced leakage of calcein from a suspension of egg PC-LUV (diameter 200 nm) [31]. EGCg induced gradual leakages over a 10–15 min period, and the rate of leakage increased with an increase in EGCg concentration. So far this data has been interpreted as that the gradual leakage occurred from each LUV over the long time. However, there have been no detailed studies of the interaction of catechins with lipid membranes, and the mechanism of catechin-induced leakage of the internal contents of vesicles remained unclear. To reveal the effect of catechins on structure and permeability of lipid membranes, we investigated the interaction of EGCg with egg PC-GUVs using the single GUV method [31]. We first examined the effect of 100 mM EGCg on single egg PC-GUVs containing 1 mM calcein. Before the addition of the EGCg solution, the GUV had a high contrast in a phase contrast microscopic image [Fig. 13A(1)]. A fluorescence A

OH

B

OH OH

OH HO

HO

O

O

OH

OH

OH

OH

OH

C

OH

OH

D OH

HO

OH HO

O O OH O

O

OH

OH OH OH

O OH O

OH OH OH

Figure 11 Chemical structures of various tea catechins. (A) (–)-Epicatechin (EC), (B) (–)-epigallocatechin (EGC), (C) (–)-epicatechin gallate (ECg), and (D) (–)-epigallocatechin gallate (EGCg).

Single GUV Method to Investigate Leakage of Internal Contents from Liposomes

135

Leakage (%)

100 80

100 μM

300 μM

60 40

50 μM

20

0 μM

0 0

200

400 600 Time (s)

800

Figure 12 Time course of EGCg-induced leakage of calcein from a suspension of egg PC-LUVs after the addition of various concentrations of EGCg. This figure is reprinted from Ref. [31] with permission from the American Biophysical Society.

A

C

B

1.0 Fluorescence intensity

Fluorescence intensity

1.0 0.8 0.6 0.4 0.2

0.8 0.6 0.4 0.2 0.0

0.0 0

10

20 30 Time (s)

40

0

10 20 30 40 50 60 Time (s)

Figure 13 Leakage of calcein from single egg PC-GUVs induced by 100 mM EGCg. (A) Fluorescence microscope images (2) show that the calcein concentration inside a single GUV progressively decreased after the addition of EGCg. The numbers above each image show the time after the addition of EGCg. Also shown are phase contrast images of the GUV at 0 (1) and at 39 s (3). (B) Time course of the change of the fluorescence intensity of the GUV. () for the data shown in (A), and (◊) for the data of another GUV without the addition of EGCg. (C) Other examples of change in the fluorescence intensity of single GUVs with time under the same conditions. Each symbol shows the time course of the change of the fluorescence intensity of each single GUV. The bar corresponds to 10 mm. This figure is reprinted from Ref. [31] with permission from the American Biophysical Society.

microscopic image of the same GUV [Fig. 13A(2)] shows that there is a high concentration of calcein inside the GUV at this time. From 18.77 to 22.20 s after starting the addition of 100 mM EGCg near the GUV, the fluorescence intensity decreased rapidly [Fig. 13A(2) and B()]. After 22.20 s, fluorescence was not detected

136

M. Yamazaki

inside the GUV, and a phase contrast image of the GUV [Fig. 13A(3)] showed that the GUV had changed into a small lump of lipid membranes. When we observed single GUVs without the addition of EGCg as a control experiment, we observed only a continual gradual decrease in the fluorescence intensity [Fig. 13B(◊)]. It suggests that this gradual decrease in the fluorescence intensity was due to the photobleaching of calcein. When we examined the interaction of 100 mM EGCg solution with other single egg PC-GUVs using the method described above, there was a similar rapid complete leakage of calcein from most of GUVs (Fig. 13C). To elucidate the mechanism of EGCg-induced leakage of calcein from a single GUV, we investigated the process of the transformation of a single GUV into a small lump of lipid membranes. Figure 14A shows a typical example of the 100 mM EGCg-induced transformation of a GUV into a small lump. At 27.20 s, a large hole was suddenly produced in the GUV membrane and sucrose inside the GUV diffused rapidly to the outside of the GUV through the hole. We can call this event the bursting of the GUV. Then, the diameter of the GUV rapidly decreased with time, and finally the GUV changed into a small lump of lipid membranes. We also

Figure 14 Structural change of single egg PC-GUVs induced by EGCg. Phase contrast images of single GUVs in the interaction of (A) 100 mM EGCg and (B) 300 mM EGCg. The numbers above each image show the time after the addition of EGCg. The bar in (A) corresponds to 20 mm, and the bars in (B) correspond to 40 mm. This figure is reprinted from Ref. [31] with permission from the American Biophysical Society.

137

Single GUV Method to Investigate Leakage of Internal Contents from Liposomes

observed this process at a higher concentration of EGCg (300 mM) (Fig. 14B). Figure 14B clearly shows a large hole in the GUV membrane, and also shows small high-contrast particles at the edge of the hole. After the large hole appeared, the diameter of the GUV rapidly decreased with time. The present results suggest that the leakage of calcein from the inside to the outside of the GUV occurred as a result of bursting of the GUV. To confirm the correlation between the leakage of calcein and the bursting of GUVs, we investigated the effect of EGCg concentration on fraction of leaked GUV and the fraction of burst GUV (Fig. 15A). As in the magainin 2-induced leakage from single GUVs, in the EGCg-induced leakage from egg PC-GUVs, the fraction of completely leaked GUV, PL(t), is an important factor. PL(t) increased as the EGCg concentration increased [Fig. 15A(□)]. On the other hand, in the estimation of the bursting of a single liposome, an important factor is the probability, PB(t), that a given liposome (in this case, a GUV) is burst at any given time, t. Such burst liposomes are produced by the complete bursting of the liposome at any time within t. Thereby, we call PB(t) the fraction of burst GUV at t. The fraction of burst GUV at 5 min increased with increasing EGCg concentration [Fig. 15()]. The dependence of PB(t) on EGCg concentration was almost the same as the dependence of PL(t) on EGCg concentration. These results strongly indicate that the leakage of calcein from the inside to the outside of the GUV occurred as a result of bursting of the GUV. We investigated time course of the EGCg-induced burst of GUVs using various concentrations of EGCg (Fig. 15B). The fraction of burst GUV increased with time, which is reasonably considered almost the same as the time course of the fraction of leaked GUV, on the basis of the strong correlation between PB(t) and PL(t). This result suggests that the gradual increase in the leakage from a suspension of LUVs

1.0

0.8

0.8

0.6

0.6

0.4

0.4

0.2

0.2

0.0

0.0 0

100 200 EGCg conc. (μM)

300

Fraction of burst GUV

B 1.0

Fraction of burst GUV

Fraction of leaked GUV

A

1.0

100 μM

0.8

80 μM

0.6

60 μM

0.4 0.2 20 μM

0.0 0

1

2

3 4 Time (min)

5

6

Figure 15 (A) Dose–response of EGCg for the fraction of completely leaked GUV, PL(t), (□), and for the fraction of burst GUV, PB(t), () among all the examined single egg PC-GUVs at t ¼ 5 min after the addition of EGCg. The bars show the standard deviation. The average values of PL(t) and PB(t) and their standard deviations were obtained by three independent experiments. (B) Time course of the fraction of burst GUV among all the examined single GUVs after the addition of various concentrations of EGCg. (r) 20 mM, (□) 60 mM, (▲) 80 mM, and (○) 100 mM EGCg. This figure is reprinted from Ref. [31] with permission from the American Biophysical Society.

138

M. Yamazaki

with time shown in Fig. 12 is due to an increase in PL(t), and is therefore due to an increase in the number of individual LUVs that had completely lost their internal contents (i.e., calcein). The threshold EGCg concentration of the leakage from the LUV suspension (50 mM) was almost the same as that of leakage from the single GUVs (70 mM). However, there is one large difference between the curve of leakage from the LUV suspension (Fig. 12) and the curve of the fraction of burst GUV (Fig. 15B); the maximum leakage from the LUV suspension was much smaller than the maximum leakage from single GUVs (e.g., at 300 mM EGCg, the former was 61% at 15 min and the latter was 100% at 1 min). We consider this discrepancy as follows (see the details in Ref. [31]). Low concentration of EGCg induced association of PC-LUVs or PC-GUVs. To elucidate the effect of the association of vesicles on the efficiency of the leakage, after many GUVs had associated in the presence of 10 mM EGCg, we added 100 mM EGCg into the vicinity of the associated GUVs. All the GUVs at the edge of the associated GUVs burst, but many GUVs inside the associated GUVs did not burst and also no leakage of the internal contents of these GUVs occurred after 9 min of exposure to the EGCg solution [31]. We conclude that the small lumps of lipid membranes that had formed at the edge of the associated GUVs due to the EGCginduced bursting of GUVs became a kind of wall at the edge, which prevented the EGCg solution from entering into the internal region of the associated GUVs. This is one of the main reasons for the low maximum value of the leakage from the LUV suspension [31]. We can analyze the kinetics of the EGCg-induced bursting of a single GUV. First, EGCg molecules bind to the membrane interface of the external monolayer of a single GUV (Bex state). At the critical concentration of EGCg, bursting of the GUV occurs, and then the membrane of the GUV transforms into a small lump. To estimate the rate constant of transition from the Bex state to the bursting of the GUV, kP, we obtained the fraction of GUVs that are in the Bex state, from the fraction of intact GUVs from which calcein did not leak (i.e., the fraction of intact GUVs), Pintact(t). Figure 16 shows the time course of the fraction of intact GUV in the presence of various concentration of EGCg. All the curves of the time course of the fraction of the Bex state were well fitted with the single exponential decay curves defined in Eq. (1). The rate constant increased with increasing EGCg concentration: kP for 100 mM EGCg, 2.5  0.5 min 1; 80 mM EGCg, 1.2  0.6 min 1; 60 mM EGCg, 0.35  0.04 min 1. These results indicate that the EGCg-induced bursting of the GUV followed the first-order reaction. To clarify the mechanism of the EGCg-induced bursting, we investigated effect of EGCg on shapes of egg PC-GUVs. The shape changes in the GUVs such as the prolate to the two spheres connected by a narrow neck were induced at very low concentrations of EGCg, which could not induce bursting of GUVs. The analysis of these shape changes indicates that the area of the external monolayer of the GUV increases with increasing amount of the EGCg bound to the external monolayer of the GUV, inducing the increase in the surface pressure of the external monolayer. As the amount of the bound EGCg in the external monolayer increases, the free energy of the Bex state increases, causing a decrease in the activation energy, Ep. As a result, the rate constant of the bursting of the GUV increases as the EGCg

Fraction of intact GUV

Single GUV Method to Investigate Leakage of Internal Contents from Liposomes

1.0

139

20 μΜ

0.8 0.6 0.4

60 μΜ

0.2

80 μΜ

0.0 0

1

2 3 4 Time (min)

100 μΜ 5 6

Figure 16 Time course of the fraction of intact GUV, Pintact(t), after the addition of various concentrations of EGCg. (r) 20 mM, (□) 60 mM, (▲) 80 mM, and (○) 100 mM EGCg. Solid lines represent the best fitted curves of Eq. (1). This figure is reprinted from Ref. [31] with permission from the American Biophysical Society.

concentration is raised. This hypothesis for the mechanism is consistent with the presently observed dependence of the rate constant of the bursting of the GUV on the EGCg concentration. We also investigated the mechanism of the formation of a small lump after the EGCg-induced rupture of a PC-GUV. Small-angle X-ray scattering experiments indicated that the spacing of DOPC-MLVs greatly decreased to 5.0 nm at EGCg concentrations above the threshold, indicating that neighboring membranes in the MLV came in close contact with each other. Using the partition coefficient of EGCg from aqueous solution to lipid membrane, we can calculate the free EGCg concentration in aqueous solution [31]. The free EGCg concentration when the neighboring membranes in DOPC-MLV came in close contact with each other was less than the threshold EGCg concentration for the induction of bursting of egg PC-GUVs [31]. These findings suggest that after bursting of GUVs, the membranes associate with each other via the same mechanism involved in EGCg-induced association of the neighboring membranes in DOPC-MLVs, which induces the formation of the small lump.

5. Conclusion and Advantage of the Single GUV Method It has previously been reported that magainin 2 and EGCg induce gradual leakage of small fluorescent probes from suspensions of many LUVs or SUVs over 10–20 min period, indicating that they cause damage to lipid membranes. However, the mechanism of the leakage was not clear. Our investigation of their interactions with lipid membranes using the single GUV method directly showed that magainin 2 formed the pore in the membrane and EGCg induced the bursting of the GUV. In both cases, the rapid leakage of fluorescent probe (calcein) from a GUV started stochastically, and once it began, the complete leakage occurred rapidly within

140

M. Yamazaki

1 min and 20 s for magainin 2 and EGCg, respectively. These results indicate that the pore formation or the bursting is the rate-determining step in the total leakage process. Experimental results indicate that the fraction of completely leaked liposome, PL(t), increased with time. These results suggest that the gradual increase in the leakage from many LUVs with time is due to an increase in PL(t), and is therefore due to an increase in the number of individual LUVs that had completely lost their internal contents. Furthermore, as the magainin 2 or EGCg concentration increased, PL(t) increased. On the basis of our results, we have proposed the two-state transition model for the magainin 2-induced pore formation and the mechanism for the EGCg-induced bursting. We succeeded in estimating the rate constants of the magainin 2-induced pore formation and also that of the EGCg-induced bursting from the time course of the fraction of intact GUVs, Pintact(t). These rate constants increased with an increase in magainin 2 (or EGCg) concentration. In summary, using the single GUV method we have succeeded in revealing completely new aspects of the interaction of magainin 2 (or EGCg) on the lipid membranes and also of the leakage induced by them. Here, as a concluding message, we summarize the advantage of the single GUV method. Using the single GUV method, we can observe individual events in single GUVs, such as the pore formation induced by antimicrobial peptide, magainin 2 [8]; the bursting of GUVs induced by antibacterial substance, EGCg [31]; the membrane fusion of two GUVs [6]; and vesicle fission of single GUVs induced by single longchain amphiphiles such as lyso-PC and lyso-PA [7, 11]. The single GUV method enabled us, for the first time, to investigate the detailed elementary processes of these events, which have never been revealed by the conventional LUV suspension method. Our results strongly indicated that, in the measurement of leakage of internal contents induced by the interaction of substances with lipid membranes, the single GUV method will give us a direct evidence for the cause of the leakage. The statistical analysis of individual events in single GUVs over many single GUVs gave us important information on the rate constants of elementary processes, such as the rate constant of the magainin 2-induced pore formation and the EGCg-induced bursting of GUVs. To increase the accuracy of the rate constants, we have to investigate effect of substances on single GUVs using much more single GUVs and make the statistical analysis of the physical parameters of a single GUV over these single GUVs. One more advantage of the single GUV method is the precise control of concentration of substances in a buffer near a single GUV. In the case of the LUV suspension method, there is large area of membranes due to many LUVs in the solution, and thereby, the free (equilibrium) concentration of substances in the bulk solution is smaller than that in the absence of LUVs and decreases with an increase in lipid concentration as a result of the binding of the substances to the LUV membranes. On the other hand, in the single GUV method, the substance concentration near a single GUV is constant, the same as that of the substance concentration introduced by the micropipet, because a given concentration of substance solution is continued to add into the vicinity of the single GUV from the micropipet. However, the present method of the addition of a substance solution through a micropipet has a technical drawback; small dilution of the substance solution occurs due to the diffusion of the substance in a buffer, and thereby the substance

Single GUV Method to Investigate Leakage of Internal Contents from Liposomes

141

concentration near the single GUV is a little lower than that of the solution in the micropipet. Development of a new technical method to add a substance solution near a single GUV will solve this drawback near future. For further analysis, we have to improve the experimental methods and the data analysis of the single GUV method. We believe that the single GUV method will provide a great deal of novel information on the mechanism for interactions of substances and proteins with biomembranes and also for biomembrane dynamics.

ACKNOWLEDGMENTS This work was supported in part by a Grant-in-Aid for General Scientific Research (B) (No. 17310071) and a Grant-in-Aid for Scientific Research in Priority Areas (System Cell Engineering by Multi-scale Manipulation) (No. 18048020) and also a Grant-in-Aid for Scientific Research in Priority Areas (Soft Matter Physics) (No. 19031011) from the Ministry of Education, Culture, Sports, Science, and Technology (MEXT) of Japan to M. Yamazaki. We thank Dr. Yukihiro Tamba for drawing Figs. 5 and 11. We thank Dr. Vasily V. Kuvichkin for his critical reading and comments.

REFERENCES [1] E. Evans, W. Rawicz, Entropy-driven tension and bending elasticity in condensed-fluid membranes, Phys. Rev. Lett. 64 (1990) 2094–2097. [2] E. Farge, P.F. Devaux, Shape changes of giant liposomes induced by an asymmetric transmembrane distribution of phospholipids, Biophys. J. 61 (1992) 347–357. [3] A. Saitoh, K. Takiguchi, Y. Tanaka, H. Hotani, Opening-up of liposomal membranes by talin, Proc. Natl. Acad. Sci. USA 95 (1998) 1026–1031. [4] O. Sandre, L. Moreaux, F. Brochard-Wyart, Dynamics of transient pores in stretched vesicles, Proc. Natl. Acad. Sci. USA 96 (1999) 10591–10596. [5] M. Yamazaki, Y. Tamba, The single GUV method for probing biomembrane structure and function, e-J. Surf. Sci. Nanotech. 3 (2005) 218–227. [6] T. Tanaka, M. Yamazaki, Membrane fusion of giant unilamellar vesicles of neutral phospholipid membranes induced by La3þ, Langmuir 20 (2004) 5160–5164. [7] T. Tanaka, R. Sano, Y. Yamashita, M. Yamazaki, Shape changes and vesicle fission of giant unilamellar vesicles of liquid-ordered phase membrane induced by lysophosphatidylcholine, Langmuir 20 (2004) 9526–9534. [8] Y. Tamba, M. Yamazaki, Single giant unilamellar vesicle method reveals effect of antimicrobial peptide magainin 2 on membrane permeability, Biochemistry 44 (2005) 15823–15833. [9] T. Tanaka, Y. Tamba, S.M. Masum, Y. Yamashita, M. Yamazaki, La3þ and Gd3þ induce shape change of giant unilamellar vesicles of phosphatidylcholine, Biochim. Biophys. Acta 1564 (2002) 173–182. [10] Y. Yamashita, S.M. Masum, T. Tanaka, M. Yamazaki, Shape changes of giant unilamellar vesicles of phosphatidylcholine induced by a de novo designed peptide interacting with their membrane interface, Langmuir 18 (2002) 9638–9641. [11] Y. Inaoka, M. Yamazaki, Vesicle fission of giant unilamellar vesicles of liquid-ordered-phase membranes by amphiphiles with a single long hydrocarbon chain, Langmuir 23 (2007) 720–728. [12] M. Yamazaki, T. Ito, Deformation and instability in membrane structure of phospholipid vesicles caused by osmophobic association: Mechanical stress model for the mechanism of poly (ethylene glycol)-induced membrane fusion, Biochemistry 29 (1990) 1309–1314. [13] K. Matsuzaki, O. Murase, N. Fujii, K. Miyajima, Translocation of a channel-forming antimicrobial peptide, magainin 2, across lipid bilayers by forming a pore, Biochemistry 34 (1995) 6521–6526.

142

M. Yamazaki

[14] J.M. Boggs, J. Euijung, I.V. Polozov, R.F. Epand, G.M. Anantharamaiah, J. Blazyk, R.M. Epand, Effect of magainin, class L, and class A amphipathic peptides on fatty acid spin labels in lipid bilayers, Biochim. Biophys. Acta 1511 (2001) 28–41. [15] H. Ikigai, T. Nakae, Y. Hara, T. Shimamura, Bactericidal catechins damage in lipid bilayer, Biochim. Biophys. Acta 1147 (1993) 132–136. [16] K. Kajiya, S. Kumazawa, T. Nakayama, Steric effects on interaction of tea catechins with lipid bilayers, Biosci. Biotechnol. Biochem. 65 (2001) 2638–2643. [17] Y. Yamashita, M. Oka, T. Tanaka, M. Yamazaki, A mew method for preparation of giant liposomes in high salt concentrations and growth of protein microcrystals in them, Biochim. Biophys. Acta 1561 (2002) 129–134. [18] Y. Tamba, T. Tanaka, T. Yahagi, Y. Yamashita, M. Yamazaki, Stability of giant unilamellar vesicles and large unilamellar vesicles of liquid-ordered phase membranes in the presence of Triton X-100, Biochim. Biophys. Acta 1667 (2004) 1–6. [19] M. Zasloff, Antimicrobial peptides of multicellular organisms, Nature 415 (2002) 389–395. [20] M. Zasloff, Magainins, a class of antimicrobial peptides from Xenopus skin: Isolation, characterization of two active forms, and partial cDNA sequence of a precursor, Proc. Natl. Acad. Sci. USA 84 (1987) 5449–5453. [21] B. Bechinger, M. Zasloff, S.J. Opella, Structure and orientation of the antibiotic peptide magainin in membranes by solid-state nuclear magnetic resonance spectroscopy, Protein Sci. 2 (1993) 2077–2084. [22] J. Gesell, M. Zasloff, S.J. Opellar, Two-dimensional 1H NMR experiments show that the 23-residue magainin antibiotic peptide is an a-helix in dodecylphosphocholine micelles, sodium dodecylsulfate micelles, and trifluoroethanol/water solution, J. Biomol. NMR 9 (1997) 127–135. [23] K. Matsuzaki, O. Murase, N. Fujii, K. Miyajima, Translocation of a channel-forming antimicrobial peptide, magainin 2, across lipid bilayers by forming a pore, Biochemistry 34 (1995) 6521–6526. [24] U. Seifert, K. Berndl, R. Lipowsky, Shape transformations of vesicles: Phase diagram for spontaneous curvature and bilayer-coupling models, Phys. Rev. A 44 (1991) 1182–1202. [25] A. Iglic, V. Kralj-Iglic, J. Majhenc, Cylindrical shapes of closed lipid bilayer structures correspond to an extreme area difference between the two monolayers of the bilayer, J. Biomech. 32 (1999) 1343–1347. [26] L. Miao, U. Seifert, M. Wortis, H.-G. Do¨bereiner, Budding transitions of fluid-bilayer vesicles: The effect of area-difference elasticity, Phys. Rev. E. 49 (1994) 5389–5407. [27] W.C. Wimley, S.H. White, Experimentally determined hydrophobicity scale for proteins at membrane interface, Nat. Struct. Biol. 3 (1996) 842–848. [28] S.M. Masum, S.J. Li, Y. Tamba, Y. Yamashita, T. Tanaka, M. Yamazaki, Effect of de novo designed peptides interacting with the lipid-membrane interface on the stability of the cubic phases of the monoolein membrane, Langmuir 19 (2003) 4745–4753. [29] H. Wang, G.J. Provan, K. Helliwell, Tea flavonoids: Their functions, utilization and analysis, Trends Food Sci. Technol. 11 (2000) 152–160. [30] K. Kajiya, H. Hojo, M. Suzuki, F. Nanjo, S. Kumazawa, T. Nakayama, Relationship between antibacterial activity of (þ)-catechin derivatives and their interaction with a model membrane, J. Agric. Food. Chem. 52 (2004) 1514–1519. [31] Y. Tamba, S. Ohba, M. Kubota, H. Yoshioka, H. Yoshioka, M. Yamazaki, Single GUV method reveals interaction of tea catechin (–)-epigallocatechin gallate with lipid membranes, Biophys. J. 92 (2007) 3178–3194.

C H A P T E R

S I X

Flexible Membrane Inclusions and Membrane Inclusions Induced by Rigid Globular Proteins Miha Fosˇnaricˇ,1 Alesˇ Iglicˇ,1 Tomazˇ Slivnik,1 and Veronika Kralj-Iglicˇ2,* Contents 1. Introduction 2. Flexible Anisotropic Membrane Inclusions 3. Membrane Inclusions Induced by the Rigid Membrane-Embedded Protein 3.1. Perturbation of Lipid Molecules Around Rigid Membrane-Embedded Proteins 3.2. Energy of Membrane Inclusion Induced by a Single Rigid Membrane Protein 4. Estimation of the Model Parameters 4.1. Basic Model 4.2. Advanced Model 5. Free Energy of Bilayer Membrane with Membrane-Embedded Inclusions 6. Conclusions Acknowledgments References

144 144 149 149 151 155 155 157 163 165 166 166

Abstract We present a theoretical approach to the study of flexible membrane inclusions and membrane inclusions induced by rigid membrane-embedded proteins. We derive the contribution to the free energy of the membrane bilayer for both kinds of inclusions. For flexible membrane inclusions, the phenomenological interaction constants that appear in the free energy expression depend on the physical and geometrical properties of the molecules that constitute the inclusion. The cases of constrained and unconstrained local shape perturbations of the membrane around a rigid membrane inclusion are discussed. The total free energy of membrane bilayer with membrane-embedded inclusions (membrane nanodomains) is derived.

Corresponding author. Tel.: þ386 41 720766; Fax: þ386 1 4768850; E-mail address: [email protected] (V. Kralj-Iglic).

* 1 2

Laboratory of Physics, Faculty of Electrical Engineering, University of Ljubljana, Trzˇasˇka 25, SI-1000 Ljubljana, Slovenia Laboratory of Clinical Biophysics, Faculty of Medicine, University of Ljubljana, Lipicˇeva 2, SI-1000 Ljubljana, Slovenia

Advances in Planar Lipid Bilayers and Liposomes, Volume 7 ISSN 1554-4516, DOI: 10.1016/S1554-4516(08)00006-9

#

2008 Elsevier Inc. All rights reserved.

143

M. Fosˇnaricˇ et al.

144

1. Introduction Membrane inclusions are important functional building blocks of biological membranes. As an addition to the lipid bilayer(s), they can significantly increase the complexity and alter the physical properties of biological membranes. In this work, we divide membrane inclusions into two groups. In the first are flexible membrane inclusions, which are small complexes composed of proteins and lipids where the proteins are often chain-like biopolymers that cross the membrane bilayer a few times (Fig. 1A) [1]. Membrane nanodomains and raft elements of biological membranes usually fall into this category. The second group are membrane inclusions (membrane nanodomains) induced by a single rigid globular membrane protein, which can be described in the first approximation as a rigid object of a simple geometrical shape (Fig. 1B) [2]. Some of the membrane-embedded peptides may induce such inclusions (nanodomains). The scope of this contribution is to derive a single-inclusion energy for both kinds of biological inclusions (i.e., membrane nanodomains).

2. Flexible Anisotropic Membrane Inclusions Thin surface of the membrane is in general anisotropic with respect to the curvature of the normal cuts [3–5] and can attain various equilibrium shapes that are not flat or spherical [6].

A

B

Figure 1 Schematic illustration of membrane inclusions (shaded area): a flexible membrane inclusion (A) and a membrane inclusion induced by membrane-embedded rigid protein (B).

145

Flexible Membrane Inclusions and Membrane Inclusions

The local shape of the membrane surface is described by two principal curvatures C1 and C2 (Fig. 7). The flexible membrane inclusion is treated as a small two-dimensional flexible plate with area a0. The inclusion is in general anisotropic; therefore, its intrinsic shape can be described by the two intrinsic principal curvatures, C1m and C2m (Fig. 2) and by the in-plane orientation of the inclusion in the membrane (Fig. 3). Accordingly, we define the elastic energy of a small plate-like membrane inclusion (1) with area a0 as the energy of the mismatch between the actual local curvature of the membrane and the intrinsic (spontaneous) curvature of the inclusion. Therefore, we define the tensor [5] M ¼ R C R1   C , where the tensor  C   m    describes the actual local curvature, the tensor  C describes the intrinsic curvature of m  the protein (Fig. 2), and



cos o R ¼ ― sin o

 sin o cos o

 ð1Þ

is the rotation matrix (see also Fig. 3). In the respective principal systems, the matrices that represent curvature tensors include only the diagonal elements:

C1m > 0 C2m = 0

C1m = 0 C2m = 0

C1m > 0 C2m < 0

Figure 2 Schematic illustration of the most favorable shapes of flexible membrane inclusions having different values of their intrinsic (spontaneous) curvatures C1m and C2m .

A

B w

w =0

w≠0

Figure 3 Schematic illustration of different orientations of a flexible membrane inclusion with intrinsic principal curvatures C1m > 0 and C2m ¼ 0 (see also Fig. 2). The shape of the membrane is cylindrical (C1 > 0 and C2 ¼ 0).

M. Fosˇnaricˇ et al.

146



C1 C ¼ ― 0

 0 ; C2



C1m Cm ¼ 0

 0 : C2m

ð2Þ

The principal systems of these two tensors are in general rotated in the tangent plane of the membrane surface by an angle o with respect to each other (Fig. 3). The elastic energy of the inclusion per unit area ðwÞ should be a scalar quantity. Therefore, each term in the expression for w must also be a scalar [7], that is, invariant with respect to all transformations of the local coordinate system. In this work, the elastic energy density w is approximated by an expansion in powers of all independent invariants of the tensor M up to the second order in the components of M . The trace    and the determinant of the tensor are taken as the set of invariants [5, 8]:

w ¼ m0 þ

K1 ðTr M Þ2 þ K2 Det M ; ― ― 2

ð3Þ

where m0 is the minimal possible value of w, while K1 and K2 are constants. For the sake of simplicity, m0  0. Taking into account the definition of the tensor M , it  follows from Eqs. (2) and (3) that the elastic energy of the flexible membrane inclusion can be written as:

E ¼ a0 ð2K1 þ K2 ÞðH  Hm Þ2  a0 K2 ðD2  2DDm cos2o þ Dm2Þ;

ð4Þ

where

1 H ¼ ðC1 þ C2 Þ; 2

ð5Þ

is the membrane mean curvature,

1 D ¼ jC1  C2 j; 2

ð6Þ

is the membrane curvature deviator, Hm ¼ ðC1m þ C2m Þ=2 is the intrinsic (spontaneous) mean curvature, and Dm ¼ jC1m  C2m j=2 is the intrinsic (spontaneous) curvature deviator. It can be seen from Eq. (4) that the material properties of an anisotropic flexible membrane inclusion can be expressed in a simple way by only two intrinsic curvatures C1m and C2m and constants K1 and K2 . Figure 2 shows a scheme of a cylindrical, flat, and saddle-like intrinsic (spontaneous) shapes of the flexible membrane inclusions. The values of the membrane mean curvature H ¼ ðC1 þ C2 Þ=2, the curvature deviator D ¼ jC1  C2 j=2, and the orientation angle of the inclusion o that correspond to the minimum of the function E for given values of Hm ¼ ðC1m þ C2m Þ=2

147

Flexible Membrane Inclusions and Membrane Inclusions

and Dm ¼ jC1m  C2m j=2, can be calculated from the necessary conditions for the extremum of the function E [8]:

@E ¼ 2a0 ð2K1 þ K2 ÞðH  Hm Þ ¼ 0; @H

ð7Þ

@E ¼ K2 a0 ð2D  2Dm cos2oÞ ¼ 0; @D

ð8Þ

@E ¼ 4a0 K2 DDm sin2o ¼ 0; @o

ð9Þ

and the sufficient conditions for the minimum of E [9]:

@2E ¼ 2a0 ð2K1 þ K2 Þ > 0; @H 2  2   2   2 2 @ E @ E @ E  ¼ 4K2 a20 ð2K1 þ K2 Þ > 0; 2 2 @H @D @H@D "  2   2 2 # @2E @2E @ E @ E  2 2 2 @H @D @o @D@o ¼ 16K22 a30

@ E ðDDm cos2o  D2m sin2 2oÞ > 0; 2 @H

ð10Þ ð11Þ

ð12Þ

2

where it was taken into account that @ 2 E=@H@D ¼ 0 and @ 2 E=@H@o ¼ 0. Considering only positive values of o, it follows from Eqs. (7) to (9) and (12) that at the minima of E:

H ¼ Hm ;

D ¼ Dm ;

o ¼ 0; p;

ð13Þ

and [8]

K1 > K2 =2;

K2 < 0:

ð14Þ

If flexible membrane inclusions have C1m > 0 and C2m ¼ 0 (see Fig. 2), the energetically favorable membrane shapes would be tubular or collapsed tubular (in the form of a twisted strip—helix A, see Fig. 4). For C1m > 0 and C2m < 0 (see Fig. 2), the favorable membrane shape would be saddle like (constituting the neck connecting the daughter vesicle and the parent cell) or the collapsed tubular, twisted in the form of a helix B strip (see Fig. 4 and [5]). The flexible membrane inclusion adapts its shape in order to fit its curvature to the actual membrane curvature (which is also influenced by inclusions). Since all

M. Fosˇnaricˇ et al.

148

Helix A

Figure 4

Helix B

Schematic presentation of a helical (A and B) configuration.

orientations of the single flexible inclusion do not have the same energy [see Eq. (4)], the partition function of a single inclusion can be written in the form:

1 Q¼ o0

ð 2p 0

  EðoÞ exp  do; kT

ð15Þ

with o0 as an arbitrary angle quantum. The free energy of the flexible membrane inclusion is then obtained by the expression fi ¼ kT lnQ. Combining Eqs. (4) and (15) allows us to write the free energy of a single flexible membrane inclusion up to the constant as:

fi ¼ ð2K1 þ K2 ÞðH  Hm Þ2 a0  K2 ðD2 þ D2m Þa0    2K2 DDm a0 : kT ln I0 kT

ð16Þ

Flexible Membrane Inclusions and Membrane Inclusions

149

By knowing the equilibrium density distribution of the membrane inclusions over the membrane [10], the contribution of the inclusions to the overall membrane’s free energy can be attained by integration of Eq. (16) over the whole membrane surface. This possibility makes the above described approach an efficient theoretical tool to study equilibrium (closed) shapes of membranes with (in general anisotropic) membrane inclusions [11–13].

3. Membrane Inclusions Induced by the Rigid Membrane-Embedded Protein 3.1. Perturbation of Lipid Molecules Around Rigid Membrane-Embedded Proteins A rigid protein, intercalated in the lipid bilayer, perturbs the structure of the surrounding lipids. Therefore, we can define the membrane inclusion as the embedded rigid protein and the surrounding lipids that are significantly distorted due to the presence of the embedded rigid protein [11]. The energy of such membrane inclusion induced by an embedded rigid protein is therefore mainly attributed to the change of the energy of the surrounding lipids. The energy of lipid molecule depends on the particular sequence of trans, gaucheþ, and gauche orientations along the lipid chain, the van der Waals interactions of lipid chain with its neighbors, steric repulsion between hard cores of each atom of neighboring lipid chains, and ionic interactions between polar lipid headgroups [14, 15]. The change in the ordering of lipids that surround the rigid protein leads to an indirect lipid-mediated interaction between two rigid proteins when they approach each other [15]. If the two proteins are close enough, the total lipid perturbation decreases, which may result in a net attractive force between the membrane-embedded rigid proteins and therefore in their aggregation [15]. Cone-like rigid proteins [2] are characterized by a cone-angle onto which the membrane shape has to adapt. The mesoscopic-level description of the membrane identifies the rigid protein’s cone-shape with a local discontinuity in the membrane curvature field. On a more microscopic level, another degree of freedom of the membrane becomes significant, namely the tilt of the lipid molecules [16, 17]. Helfrich and Prost [3] have shown that symmetric lipid bilayer may exhibit an intrinsic bending force if the lipid molecules are collectively tilted. However, membrane perturbations that involve lipid tilt are often short-ranged, with a characteristic length extending over a few lipids. Lipid tilt is thus important for processes where the local membrane geometry changes over short distances such as for non-bilayer lipid phases [18, 19], or for the periodic ‘‘ripple’’ phase [20–22]. In the theoretical works cited above, the membrane-embedded rigid proteins exhibit cylindrical symmetry about their axis normal to the membrane, that is, they are isotropic. More general, if cylindrical symmetry of rigid membrane protein is absent (Fig. 5), the membrane inclusion free energy depends on the protein’s inplane orientation within the membrane. The intrinsic shape of the rigid protein is then characterized by two intrinsic principal curvatures, C1m and C2m . The lateral

M. Fosˇnaricˇ et al.

150

90⬚

Figure 5 Schematic illustration of an anisotropic membrane-embedded rigid protein.

A

B

Rs

Rs C= 1 Rs

Figure 6 Schematic illustration of the lipid bilayer of prescribed spherical curvature ðC ¼ C1 ¼ C2 ¼ H ¼ 1=Rs Þ defined at mesoscopic scale level. The intercalated rigid protein has conical shape. In the case A the local membrane shape does not differ from the mesoscopic spherical curvature of the membrane ðcÞ, while in the case B also the local microscopic (nano-scale) membrane shape perturbation of the spherical surface with curvature c is induced due to the presence of the rigid protein. In the case B lipids accommodate to the intrinsic shape of the intercalated rigid protein through the curvature deformation and via changes in lipid tilt, while in the case A lipids accommodate to the protein intrinsic shape via changes in lipid tilt (adapted from [28]).

organization of anisotropic proteins can be quite complex, ranging from chain-like assembly [23], saddle-like membrane regions [11] to periodic pattern formation [24]. Within the standard theory of elasticity of lipid bilayer, its elastic energy is decomposed into contributions due to area stretching, tilt of the lipid molecules, local bending, and non-local bending [16, 25, 26]. On a mesoscopic-scale level, the local and non-local bending energies can be described in terms of its two local principal membrane curvatures C1 and C2 [25, 26]. The question arises, how the elastic behavior of a membrane bilayer is affected by membrane-embedded rigid proteins, if the local microscopic membrane shape perturbation (at the nano-scale level) due to each individual protein is taken into account (Fig. 6). In general, the theoretical description of local microscopic perturbations of lipid molecules around the intercalated rigid protein falls in between the two limiting cases. In the first case, the membrane intercalated rigid proteins are distributed over the whole membrane surface or at least over a large portion of it (see also [1]). Therefore, possible local microscopic perturbations of the membrane shape around each of the rigid proteins (as schematically shown in Fig. 6B) would greatly increase the nonlocal bending energy of the bilayer membrane. This energy contribution, also called

Flexible Membrane Inclusions and Membrane Inclusions

151

the relative stretching energy (since it originates from different stretching of both monolayers during bending of the bilayer at constant average membrane area) [25–27], can be written as

Wn ¼ kn AðhHi  H0 Þ2 ;

ð17Þ

Ð where hHi ¼ A1 HdA is the average mean curvature, H ¼ ðC1 þ C2 Þ=2, H0 is the spontaneous mean curvature [12], kn is the non-local bending rigidity [27], A is the membrane area, and dA is the membrane area element. For a closed, nearly flat bilayer membrane (where hHi  0; H0  0), with N homogeneously distributed intercalated rigid proteins, the membrane’s non-local bending energy Wn can be approximately written as [28]

Wn ffi kn AðN hHip  NH0 p Þ2 / N 2 ;

ð18Þ

where H0p and hHip refer to the disturbed membrane patch around single membrane-intercalated rigid protein (Fig. 6B). Since the energy Wn increases quadratically with the total number of membrane-embedded rigid proteins, the local microscopic perturbation of the membrane shape around each of the intercalated rigid proteins (Fig. 6B) would be energetically less favorable for large enough N than the locally unperturbed membrane shape where the lipids accommodate to intrinsic shape of rigid protein predominantly via changes in the lipid tilt (Fig. 6A). In the opposite limit, the membrane region with intercalated rigid proteins is spatially confined (i.e., small) and in contact with a reservoir of relaxed lipid bilayer. Therefore, the lipids surrounding the intercalated rigid protein are free to adjust their conformation also by perturbation of the local membrane shape, as schematically shown in Fig. 6B. In biological membranes, the majority of the membrane proteins are laterally distributed over the whole membrane area. In addition, the number of the membrane proteins ðN Þ is very large. Therefore, the first scenario, that is, the case of constrained microscopic deviations of the membrane shape around the intercalated rigid inclusions (Fig. 6A), seems to be more relevant [see also Eq. (18)].

3.2. Energy of Membrane Inclusion Induced by a Single Rigid Membrane Protein Coupling between non-homogeneous lateral distribution of membrane-embedded rigid proteins and membrane shapes may be a general mechanism of generation and stabilization of highly curved membrane structures (spherical buds, membrane necks, thin tubular membrane protrusions) [11, 12, 29–32]. On the phenomenological level, membrane bending may couple energetically to the local density of membrane-embedded rigid proteins by introducing the composition-dependent local bending constant and spontaneous curvature. The underlying model (including also the direct interactions between rigid protein and

M. Fosˇnaricˇ et al.

152

configurational entropy of rigid proteins) was suggested by Markin [29] and used in subsequent applications [33]. Leibler [34] proposed a similar thermodynamical model. Another theoretical approach starts from a phenomenological expression for the energy of a single membrane inclusion induced by intercalation of the rigid protein [10, 11] where the term inclusion is used for an entity consisting of the embedded rigid protein and lipids that are significantly distorted due to the presence of the embedded rigid protein [11] (see also Fig. 1B). It is proposed that the energy of such inclusion derives from the mismatch between the local shape of the membrane and the intrinsic shape of the membrane-embedded rigid protein. The local curvature of the membrane is represented by curvatures of all possible normal cuts of the surface through the site of the inserted rigid protein. The energy of a single inclusion induced by intercalation of single rigid protein is then given by a phenomenological expression consisting of two terms [11],

x E¼ 4p

ð 2p 0

x ðC  Cm Þ dc þ 16p 2

ð 2p  0

2 d ðC  Cm Þ dc; dc

ð19Þ

where x and x are positive interaction constants, C is the curvature of the membrane normal cut that is for an angle c rotated in the principal axes system of the membrane surface, Cm is the curvature of the normal cut corresponding to the protein intrinsic shape in the same direction. The first contribution takes into account the differences of the curvatures of the normal cuts of the two systems while the second contribution takes into account the coupling between the neighboring curvatures of the normal cuts of the two systems. The orientation of the membrane-embedded rigid protein is described by considering that the principal directions of the membrane surface are in general different from the principal directions of the protein intrinsic shape. The mutual orientation of the two systems is determined by the angle o. We consider the Euler equations for the curvatures of the respective normal cuts of the continuum

C ¼ C1 cos2 c þ C2 sin2 c

ð20Þ

Cm ¼ C1m cos2 ðc þ oÞ þ C2m sin2 ðc þ oÞ;

ð21Þ

and

where C1 and C2 are the principal curvatures describing the local shape of the surface (Fig. 7), and C1m and C2m are the principal curvatures describing the intrinsic shape of the membrane-embedded rigid protein. By performing the integration in Eq. (19), we get

x x þ x 2 2 ðD  2DDm cos2o þ D2m Þ; E ¼ mm þ ðH  Hm Þ þ 4 2

ð22Þ

153

Flexible Membrane Inclusions and Membrane Inclusions

R2

n

R1

C1 = 1 R1 C2 = 1 R2

Figure 7

Schematic illustration of the two principal curvatures of membrane surface.

where mm is the constant, H ¼ ðC1 þ C2 Þ=2 is the mean curvature, D ¼ jC1  C2 j=2 is the curvature deviator, while Hm ¼ ðC1m þ C2m Þ=2 and Dm ¼ jC1m  C2m j=2 are the intrinsic mean and deviatoric curvatures that reflect the preferred local macroscopic membrane curvature of the membrane-embedded rigid protein. The membrane inserted protein is called isotropic if C1m ¼ C2m , while it is called anisotropic if C1m 6¼ C2m . Figure 8 gives a schematic presentation of different intrinsic shapes of inserted rigid proteins. At this point, let us stress that the energy of a single membrane inclusion induced by membrane-embedded rigid protein (Eq. 22) is mathematically equivalent to the energy of a single flexible membrane inclusion (Eq. 4). Combining both equations yields relations between the interaction constants, x ¼ 2a0 ð2K1 þ K2 Þ and x ¼ 2a0 ð2K1 þ 3K2 Þ. However, the origin of the interaction constants can be

M. Fosˇnaricˇ et al.

154

Isotropic constituents 90⬚ C1m = C2m

C1m = C2m > 0

90⬚ C1m = C2m

C1m = C2m < 0

90⬚ C1m = C2m

C1m = C2m = 0

Anisotropic constituents 90⬚ C1m ≠ C2m

C1m > 0, C2m = 0

90⬚ C1m = 0, C2m < 0

C1m ≠ C2m

90⬚ C1m ≠ C2m

C1m > 0, C2m < 0

Figure 8 Schematic illustration of different isotropic and anisotropic shapes of the membraneembedded constituents (rigid proteins). The intrinsic shape of the protein is characterized by two intrinsic principal curvatures C1m and C2m .

different in each case. Namely, in the case of a membrane inclusion induced by the membrane-embedded protein, the interaction constant originates in the deformation of the lipids surrounding the rigid protein, while in the case of a flexible membrane inclusion, the biopolymer(s) itself is (are) also deformed. The maximum and the minimum of Ei ðoÞ are for protein orientation angle o ¼ 0 and o ¼ p=2, respectively. The single inclusion energy (Eq. 22) comprises the contribution due to deformation of the lipids that surround the intercalated protein (Fig. 6) [10, 11, 35]. The possible microscopic (nano-scale) perturbations of the membrane shape around the intercalated rigid protein (Fig. 6B) are not explicitly taken into account in Eq. (22), but rather hidden in the phenomenological constants mm , x, x , C1m , and

Flexible Membrane Inclusions and Membrane Inclusions

155

C2m (or Hm and Dm ) (Fig. 8), where it is assumed that the distorted regions of lipids of the neighboring proteins do not overlap. The concept of the single inclusion energy was taken as a base for a self-consistent description of equilibrium shapes of a closed bilayer vesicle and the related lateral distribution of intercalated inclusions [10, 11, 36]. In accordance with previous results [29], clustering and lateral phase separation of the inclusion has been predicted [1]. Within the above described phenomenological (mean-field) approach, the influence of membrane-embedded rigid proteins on the elastic properties of the lipid bilayer can be calculated in terms of the properties of the host membrane and the properties (geometry) of the intercalated rigid proteins. The non-homogeneous lateral distribution of the isotropic rigid proteins are an internal degree of freedom that lowers the equilibrium free energy of the membrane and in this way contributes to the decrease of the local bending modulus kc [10, 34, 36]. The change in the membrane elasticity depends linearly on the density of membrane-embedded rigid proteins. In the case of anisotropic rigid proteins, their rotational ordering is another internal degree of freedom, which additionally decreases the membrane local bending constant [11, 35].

4. Estimation of the Model Parameters 4.1. Basic Model In the previous section, we derived the expression for the energy of the membrane inclusion (membrane nanodomain) induced by the membrane-embedded rigid protein. In this subsection, the phenomenological parameters describing the single inclusion energy Hm , Dm , and x [see Eq. (22)] are estimated using a simple theoretical model describing the elasticity of lipid bilayer [13]. In this analysis, we assume that the local microscopic shape deformations of the membrane around the membrane-embedded rigid protein are constrained (Fig. 6A) and the lipids accommodate to the intrinsic shape of rigid protein only via changes in the lipid tilt. This corresponds to the biologically relevant case of the membrane proteins that are distributed all over the cell membrane. Let us consider a single cone-like rigid protein. To render the induced inclusion anisotropic, we introduce a dependency of the cone angle y ¼ yðoÞ on the azimuthal angle o (Fig. 9). For small variations of y, we can write

yðoÞ ¼ y þ Dy cosð2oÞ;

ð23Þ

 

where y is the average ‘‘cone-ness’’ of the protein and Dy is the corresponding deviator. The rigid protein is embedded in a lipid bilayer of mean and deviatoric curvatures H and D, respectively. Hence, according to the lemma of Euler, the curvature measured in the radial direction of the inclusion, at azimuthal angle o, is

CðoÞ ¼ H þ D cosð2oÞ

ð24Þ

M. Fosˇnaricˇ et al.

156

Membrane

w

q (w)

−cR

r

h(r)

Protein

r0 R

Figure 9 Schematic illustration of a protein in the membrane for the model of constrained (dark gray—in front) local shape perturbation from Section 5.2, and unconstrained (light gray—in back) local shape perturbations from Section 5.1. For anisotropic inclusions, the cone angle y depends on the azimuthal angle o.

Formally, the protein-induced perturbation free energy of the lipid bilayer can ~ be expressed as an integration of the free energy density EðoÞ per unit length of the circumference of the inclusion’s core, L ¼ 2pr , where r is the radius of the 0 0 Ð Ð ~ ¼ ðL=2pÞ EðoÞdo ~ (see Fig. 9). For inclusion’s core (i.e., rigid protein): E ¼ L EdL ~ yðoÞ depends only parametsufficiently large radius r0 , we expect that E~ ¼ E½CðoÞ; rically on o, namely via the relations CðoÞ and yðoÞ. More generally, E~ should also depend on the derivatives of CðoÞ and yðoÞ with respect to o. This additional dependence should become relevant if the radius r0 were smaller than the characteristic decay length z of membrane perturbations. Using membrane elasticity theory, the characteristic decay length z has recently been calculated [37] for a planar (C ¼ 0) lipid layer in contact with a wall tilted by an angle y; it depends on the thickness of the lipid bilayer, the lateral stretching modulus, and the tilt modulus (kt ). Typical values for a lipid monolayer [13] yield z ¼ 0:9 nm. Hence, assuming that r0  z, we can write 2ðp E 1 ~ ¼ E½CðoÞ; yðoÞdo L 2p

ð25Þ

0

In this case, E~ can be calculated using a one-dimensional model for the elastic interaction of a lipid layer with an infinitely extensive, rigid wall. Such model has frequently been suggested in previous works [37, 38] and can be generalized to a bent lipid layer of curvature C [13],

k0 ~ EðC; yÞ ¼ ðy  Cr0 Þ2 þ ðC0  CÞðy  Cr0 Þ; 2z

ð26Þ

Flexible Membrane Inclusions and Membrane Inclusions

157

where k0 is the bending stiffness of the lipid monolayer and C0 is the spontaneous curvature. After inserting yðoÞ from Eq. (23) and CðoÞ from Eq. (24) into Eq. (26), the comparison of the obtained expression with Eq. (22) yields [13]:

Hm;1

    y r0 þ z zC0 Dy r0 þ z þ ; Dm;1 ¼ ¼ r0 þ 2z r0 r0 þ 2z r0 r0 þ 2z   r0 2 þ 2 ; x ¼ 0 x ¼ 2pr0 k0 z

ð27Þ ð28Þ

This confirms the expectation that the shape of the inclusion’s core (i.e., the shape of membrane-embedded rigid protein) is incorporated in the expressions for the spontaneous mean curvature and the spontaneous curvature deviator so that Hm;1 ¼ y=r0 and Dm;1 ¼ Dy=r0 , respectively. Note the strong dependence of the interaction constant x r03 on the protein radius (for r0 z); this is a consequence of both the rigidity of the protein (contributing r02 ) and the linear increase of the circumference with r0 . Dependence of x on the protein radius (r0 ) is plotted in Fig. 12 for the characteristic decay length z ¼ 0:9 nm. Note also that the last relation in Eq. (28), x ¼ 0, follows from our assumption that the rigid protein has a sufficiently large radius that E~ does not depend on the derivatives of CðoÞ and yðoÞ with respect to o.

4.2. Advanced Model In this subsection, we introduce more advanced theoretical model in order to estimate the constants Hm and x, where now the tilt deformation is explicitly taken into account [28]. In the model from Section 5.1, the tilt degree of freedom enters the model only through the characteristic decay length z. In the model [28] we consider a lipid membrane that consists of two opposed monolayers, an external (E) and an internal (I) one. Both monolayers are described by a height profile, hE and hI , and by their local directors (unit vectors), tE and tI , that describe the average orientation of the lipid chains (see Fig. 10). The elastic free energy per unit area, f^E , of the external monolayer can be written up to quadratic order in hE and tE as

ks kt B kh f^E ¼ ðr tE Þ2 þ ðtE  rhE Þ2 þ ðhE  hÞ2 þ ðDhE Þ2 2 2 2 2 K þ ðr tE Þ2 þ k det hE;ij 2

ð29Þ

The first term in Eq. (29) characterizes the splay energy of the lipid chains with ks being the corresponding splay modulus. The second term accounts for the energy cost of tilting the director tE away from its orientation normal to the surface hE ; the

M. Fosˇnaricˇ et al.

158

z

hE(r)

tE

h(r)

tI

h1(r)

y x

{x,y}

Figure 10 Illustration of a perturbed lipid bilayer with indicated local directors tE and tI and height profiles hE and hI of the external and internal leaflet, respectively. The average height of the bilayer is h ¼ ðhE þ hI Þ=2. Two lipid molecules are shown schematically (adapted from [28]).

prefactor kt is the tilt modulus [42]. Thickness changes of the monolayer are accounted for by the third term where B is the compression modulus and h is the height of the reference surface with respect to which the compression/expansion of the monolayer is measured. It is reasonable to assume that for given membrane thickness hE  hI , the thickness of each monolayer is allowed to relax; this specifies h ¼ ðhE þ hI Þ=2 to be the average height profile of the bilayer. The fourth term in Eq. (29) expresses the bare bending energy of the external monolayer with corresponding modulus kh . Note that this term is distinct from the splay energy; only for kt ! 1 splay and bare bending refer to the same deformation. While the splay energy mainly accounts for the splay deformation of the lipid chains, the bending term originates predominantly in the headgroup region of the monolayer. For example, the electrostatic contribution to the bending modulus contributes entirely to kh . One might therefore refer to the modulus kh as the head group contribution to the bending stiffness. The last two terms in Eq. (29) describe the energetic contribution of a twist deformation of the chains (with corresponding modulus K) and of a saddle deformation of hE (with the modulus k). Starting from f^E , we obtain the elastic free energy of the internal leaflet, f^I , by replacing hE ! hI and tE ! tI (the minus sign in the latter reflecting the opposite orientation of the two opposed monolayers). Hence,

ks kt B kh f^I ¼ ðr tI Þ2 þ ðtI þ rhI Þ2 þ ðhI  hÞ2 þ ðDhI Þ2 2 2 2 2 K þ ðr tI Þ2 þ k det hI; ij 2

ð30Þ

Flexible Membrane Inclusions and Membrane Inclusions

159

The elastic free energy of the lipid bilayer per unit area f^bl is then

f^bl ¼ f^E þ f^I :

ð31Þ

At this point it is convenient to switch to a new set of variables, namely to the average shape h and thickness dilation u, defined through hE ¼ h þ u and hI ¼ h  u (see also Fig. 10). Similarly, we define the average director t and the difference director d via the relations tE ¼ t þ d and tI ¼ t  d. This allows us to express f^bl ¼ f^tu þ f^dh as the sum of the two independent contributions [17]

f^tu ¼ ks ðr tÞ2 þ kt ðt  ruÞ2 þ Bu2 þ kh ðDuÞ2 þ Kðr tÞ2 þ 2 k det uij

ð32Þ

and

f^dh ¼ ks ðr dÞ2 þ kt ðd  rhÞ2 þ kh ðDhÞ2 þ Kðr dÞ2 þ 2 k det hij

ð33Þ

The two contributions can be treated separately. The first one depends on the tilt difference t and thickness dilation u which is relevant for proteins with up-down symmetry including the case of hydrophobic mismatch. The corresponding rigid protein-induced deformation is short-ranged and has been studied intensively in the past [39, 40]. In the present chapter, we focus our interest entirely on the second contribution (namely Eq. 33). In other words, we consider membrane deformations due to isotropic, cone-like rigid proteins with no hydrophobic mismatch (implying Ð f^tu ¼ 0). We thus seek to minimize the overall elastic free energy Fdh ¼ f^dh da where da ¼ dxdy½1 þ ðrhÞ2 1=2 denotes the area element of the lipid bilayer. The corresponding Euler-Lagrange equations pertaining to Fdh are

kt ðd  rhÞ  ks rðr dÞ þ Kr ðr dÞ ¼ 0 kh r4 h þ kt ðr d  DhÞ ¼ 0

ð34Þ

To make the model tractable analytically, we assume a cylindrical symmetry around a rigid protein. In other words, we are in this subsection only interested in inclusions (i.e., membrane nanodomains) induced by isotropic membraneembedded rigid proteins. Also, we adopt a cell model, that is, we assume that the inclusions are homogeneously distributed over a membrane segment of prescribed spherical curvature (c1 ¼ c2 ¼ c), defined at the mesoscopic level. The cell model starts from a hexagonal arrangement of spatially fixed cone-like inclusions of (average) radius r0 [28] (see Fig. 11). The radius R of the unit cell (see Fig. 9) then defines the (uniform) area fraction m ¼ r02 =R2 of rigid proteins in the membrane segment. Our aim is to characterize—at the mesoscopic scale—the bending stiffness of a rigid

M. Fosˇnaricˇ et al.

160

Top view:

Protein

Cross-section: Membrane

Unit cell

Figure 11 Top view: Schematic illustration of a hexagonal array of laterally fixed isotropic conelike membrane-embedded rigid proteins (shaded circles). The unit cell around each protein is approximated by a circle. The membrane shape in the cross-section is also given. The shaded cones represent cross-sections through the inclusions (adapted from [28]).

protein-containing membrane patch with prescribed sphere-like membrane curvature. Hence, the membrane curvatures at the boundaries of each unit cell are fixed to be c1 ¼ c2 ¼ c, where c is the sphere-like (mesoscopic level) membrane curvature. The fact that the curvatures at the cell boundaries are all equal is a consequence of both the symmetry of the deformation and the isotropy of the protein. The local, microscopic, membrane shape perturbation within the unit cell is allowed to minimize the membrane free energy (see also Fig. 11). Equation (34) can be solved analytically for cylindrical symmetry and the corresponding free energy Fdh can be calculated. This derivation is explained in detail elsewhere [28], here we discuss only the dependencies of the constants Hm and x [see Eq. (22)] on the parameters of the microscopic model. In the model presented in Section 2, the single inclusion energy [Eq. (22)] induced by an isotropic rigid protein (Dm ¼ 0) in a spherical membrane curvature field (H ¼ c ¼ const: and D ¼ 0) simplifies to:

x E ¼ mm þ ðH  Hm Þ2 ; 2

ð35Þ

where curvature c is defined at the mesoscopic level. In other words, the possible local microscopic curvature deformation around the rigid protein (Fig. 6B) is not shown directly in c, instead, it is hidden in the phenomenological parameters x and Hm . By comparing Eq. (35) with the free energy Fdh from the model described above, we can obtain the relations [28]

Flexible Membrane Inclusions and Membrane Inclusions

Hm ’

ð1 þ krel Þ c0 ; krel

x ’ pR2 k0 krel :

161

ð36Þ ð37Þ

Here k0 is the (local) bending stiffness of the (rigid protein-free) lipid bilayer, R is the radius of the cylindrically symmetric unit cell (Fig. 11), c0 is the spontaneous curvature of the rigid protein-containing membrane, and krel is the relative change of the bending stiffness k due to the presence of the rigid proteins in the membrane bilayer; namely krel ¼ k=k0  1. The expressions for c0 and krel can be derived analytically [28]. In the compact form they can be written in terms of the relative cell size r ¼ ðR=r0 Þ2  1 and the quantities

kh ; ks   ð1 þ 2 Þ ks þ 1; ¼ ~ ¼ kh 2 2 ¼

k ; ½2ðks þ kh Þ  1=2 ~z ¼ kt þ kt ks kh a¼

ð38Þ ð39Þ ð40Þ ð41Þ

and



2~z I1 ðR=~zÞK1 ðr0 =~zÞ  I1 ðr0 =~zÞK1 ðR=~zÞ r0 I1 ðR=~zÞK0 ðr0 =~zÞ þ I0 ðr0 =~zÞK1 ðR=~zÞ

ð42Þ

where In and Kn give the modified Bessel functions of the first and second kind, respectively. We find that

R 2 c0 ð1  P a~ Þð1 þ rÞ ¼ 2 yr0 1 þ rð1 þ aÞ þ Pa f1  ~2 ½1 þ rð1 þ aÞg

ð43Þ

and

krel

1 1  P2 ð1 þ a~ 2 Þ ¼ 1 þ a r þ P2 ð1  a~ 2 rÞ

ð44Þ

M. Fosˇnaricˇ et al.

162

Local stability condition implies k0 > k=2 > 0 [12, 41] (where k0 and k are local bending (splay) modulus and saddle-splay (Gaussian) modulus, respectively); therefore, 0:5 < a < 0. The estimated values of kt [42, 43] yield ~z 0:2nm. The above expressions contain the microscopic membrane shape perturbations around a rigid protein through curvature deformation and through changes in lipid tilt (Fig. 6B). However, the model described in the present subsection can also be used for a biologically important case of restricted local shape perturbations (Fig. 6A). Relations for Hm and x [Eqs. (36) and (37)] remain the same, but the expression for the relative bending stiffness becomes

krel

  1 1 1 ; ¼ 1 þ r Pð1 þ 2 Þð1 þ aÞ

ð45Þ

where the function P is the same as in Eq. (42), with ~z now being replaced with ~zc ¼ ~zðkh ! 1Þ:

~zc ¼

 1=2 kt : ks

ð46Þ

Relations (22), (36), and (37) are valid only as long as local deformations around the membrane-embedded neighboring rigid proteins do not overlap. Otherwise the interaction constant x [Eq. (22)] would depend on the area fraction of proteins (m ¼ r02 =R2 ) in the considered membrane patch. For the case of unconstrained local shape perturbations around the rigid proteins (Fig. 6B) the above relations are valid up to a certain value of the area fraction of the proteins. For most of the relevant cases, the actual area fraction of rigid proteins (m) is well below this value. In the case of restricted local (microscopic) shape perturbations around the rigid protein (Fig. 6A), the decay of lipid (tilt) deformation around the protein is exponential (i.e., short-ranged). Therefore, the overlapping of the short-ranged lipid deformations around neighboring proteins becomes important only if the proteins are very close. Consequently, the interaction constants (x, x , Hm , and Dm ) depend on the local density of the inclusions only for very large m. For small values of m, we can expand the expression for x [Eq. (37)]. For unconstrained local membrane shape relaxation (Fig. 6B), we get

" # pr02 k0 22~zð1 þ a~ 2 Þ K1 ðr0 =~zÞ 1 ; x¼ 1a r0 K0 ðr0 =~zÞ

ð47Þ

whereas the case of constrained local membrane shape relaxation (Fig. 6A) yields

"

# ~ r0 K0 ðr0 =zc Þ 1 : x ¼ pr02 k0 2 ~ 2zc ð1 þ  Þð1 þ aÞ K1 ðr0 =~zc Þ

ð48Þ

163

Flexible Membrane Inclusions and Membrane Inclusions

100

x/K0 [nm2]

80 60 40 20 0 0

1

2

3 r0[nm]

4

5

6

Figure 12 Interaction constant x [Eq. (22)] as a function of the average radius of the rigid protein (r0 ) in the model of constrained local membrane microscopic shape perturbation (Fig. 6A) calculated from Eq. (28) for z ¼ 0:9 nm (gray full curve) and from Eq. (48) (dashed curve) for 2 ¼ 1, a ¼ 0:2, and ~zc ¼ 0:2 nm. The figure also shows the dependency of x on r0 for unconstrained local membrane microscopic shape perturbation (Fig. 6B), as calculated from Eq. (47) for ~z ¼ 0:2 nm and same values of 2 and a (black solid curve).

It can be seen in Eqs. (47) and (48) that the interaction constant x adopts negative values for r0 < 2~z2 ð1 þ a~ 2 ÞK1 ðr0 =~zÞ=K0 ðr0 =~zÞ (unconstrained case) and r0 < 2~zc ð1 þ 2 Þð1 þ aÞK1 ðr0 =~zc Þ=K0 ðr0 =~zc Þ (constrained case). Therefore, for large enough a and ~z (or ~zc ), and for small enough radius of the protein, rigid inclusions could locally soften the membrane [28]. This could not be predicted within the theory presented in Section 4.1, where the tilt degree of freedom is not explicitly taken into account and enters the model only through the characteristic decay length z. In Fig. 12, the dependence of x on the average radius of the membraneembedded rigid protein r0 is shown for different presented models. The case of constrained local membrane shape perturbations (Fig. 6A) is shown in gray curve for the model from Section 4.1 and in dashed curve for the above described model [Eq. (48)]. The case of unconstrained local shape perturbations (Fig. 6A) is shown in black solid curve [see Eq. (47)].

5. Free Energy of Bilayer Membrane with Membrane-Embedded Inclusions As already mentioned, the expression for the energy of membrane inclusion induced by single rigid membrane protein [Eq. (22)]:

x x þ x 2 2 ðD  2DDm cos2o þ D2m Þ; EðoÞ ¼ ðH  Hm Þ þ 4 2

ð49Þ

M. Fosˇnaricˇ et al.

164

is mathematically equivalent to the expression for the energy of a single flexible membrane inclusion [see Eq. (4)], where

x ¼ 2a0 ð2K1 þ K2 Þ;

x ¼ 2a0 ð2K1 þ 3K2 Þ:

ð50Þ

Therefore, in the following, only the expression (49) is used to describe the energy of a single inclusion. It can be seen from Eq. (49) that the energy of a single inclusion attains a minimum when cos ð2oÞ ¼ 1, while the single inclusion energy attains a maximum when cos ð2oÞ ¼ 1. In the first case, the energy of a single inclusion is

x x þ x 2 x þ x ðD þ D2m Þ  DDm ; Emin ¼ ðH  Hm Þ2 þ 4 2 2

ð51Þ

whereas in the second case it is

Emax

x x þ x 2 x þ x 2 2 ðD þ Dm Þ þ DDm : ¼ ðH  Hm Þ þ 4 2 2

ð52Þ

The states o ¼ 0; p and o ¼ p=2; 3p=2, respectively, are degenerate. Since all orientations of the inclusion do not have the same energy, the partition function [44] of a single inclusion can be described within the four-state model (considering only orientations o ¼ 0; p=2; 3p=2; pÞ as:



   Emin Emax þ 2 exp : Q ¼ 2 exp kT kT

ð53Þ

The free energy of a single membrane inclusion can then be obtained by the expression

x x þ x 2 2 ðD þ D2m Þ fi ¼  kT lnQ ¼ ðH  Hm Þ þ 2  4 DD m ;  kT ln cosh ðx þ x Þ 2kT

ð54Þ

where we omitted the constant terms that can be neglected for the case of constant total number of inclusions in the membrane. In the following, we derive the free energy of a bilayer membrane with membrane-embedded inclusion. The excluded volume principle, that is, the finite volume of the membrane inclusions, is taken into account by applying the lattice statistics [44]. Therefore, the membrane is divided into small patches, which still contain a large number of molecules so that the methods of statistical physics can be used. The mesoscopic membrane curvature (see also Fig. 6) is taken to be constant over the patch. In a single patch, a lattice with M sites is imagined. There are

Flexible Membrane Inclusions and Membrane Inclusions

165

N inclusions in a given patch, each contributing the free energy fi . The direct interactions between the inclusions are not taken into account. The canonical partition function of the patch is therefore:

QP ¼

QN M! ; N !ðM  N Þ!

ð55Þ

where the partition function of the single inclusion Q is defined by Eq. (53). The Helmholz free energy of the patch is F P ¼ kT lnQP :

  N N ; F  NkT lnQ þ kTN ln þ kT ðM  N Þln 1  M M P

ð56Þ

where we applied the Stirling approximation ln x! ffi x ln x  x. The free energy of all inclusions in the bilayer membrane can be obtained by summing the contributions of all patches in the membrane:

ð

ð Fi ¼

nfi m0 dA þ kTm0 A

ðn ln n þ ð1  nÞ ln ð1  nÞÞdA;

ð57Þ

A

where n ¼ N =M is the local membrane area fraction occupied by the membrane inclusions, dA is the membrane area element (area of the patch), and m0 ¼ M=dA ¼ 1=a0 , where a0 is the area of the single inclusion. The total free energy of the bilayer membrane with the embedded inclusions (F ) can thus then be written as [1, 29]

ð ð F a0 k0 2 ¼ ð1  nÞ ð2HÞ da þ n fi da m0 A A 2 A ð þ kT ðn ln n þ ð1  nÞ ln ð1  nÞÞda;

ð58Þ

A

where A is the membrane area, da ¼ dA=A, and k0 is the (local) bending constant of the bilayer membrane. The first term accounts for the bending energy of the lipid bilayer, while the last term accounts for the configurational entropy of the inclusions [1, 29].

6. Conclusions Theoretical approaches to study the coupling of non-homogeneous lateral distribution of membrane inclusions and membrane shapes [10, 29] were described for flexible membrane inclusions and for inclusions induced by rigid membrane proteins [1, 17]. As it is shown in this work, both cases yield a mathematically

166

M. Fosˇnaricˇ et al.

equivalent result for the energy of a single membrane inclusion [see Eqs. (4) and (22)]. In the first case of flexible inclusions, the phenomenological constants K1 and K2 originate in the deformation of the small membrane complex (nanodomain) composed of lipids and proteins (Fig. 1A) and depend on the elastic and geometric properties of this complex (membrane nanodomain); while in the second case (where inclusion is induced by the rigid globular protein), the globular proteins are treated as rigid bodies (Fig. 1B) and the whole contribution to the interaction constants x and x [Eq. (22)] originate in the deformation of the surrounding lipid molecules and the geometry of the rigid protein. The value of the interaction constant x grows with the average radius of the rigid protein (r0 ) (Fig. 12). In biological membranes, the majority of the membrane rigid proteins are laterally distributed over the whole membrane area. In addition, the number of the membrane proteins ðN Þ is very large. For such systems, the case of constrained microscopic perturbation of the membrane shape around the intercalated rigid protein (Fig. 6A) seems to be biologically more relevant than the case of unconstrained microscopic perturbation of the membrane shape around the intercalated rigid protein (Fig. 6B). As for example in erythrocytes, the budding takes place over the whole cell membrane surface (although it is located predominately at the top of membrane spicules) [1, 12]. Because of that, local microscopic membrane shape perturbations around the membrane-embedded proteins are strongly restricted (see Eq. 18). As a consequence, the lipid deformation around the membrane-embedded proteins is predominantly a consequence of the change of tilt of lipid molecules around the proteins (Fig. 6A), and not of the microscopic membrane shape perturbation (Fig. 6B). Coupling between non-homogeneous lateral distribution of membraneembedded rigid proteins and specific membrane shapes may be an important mechanism of generation and stabilization of highly curved membrane structures. Therefore, the theoretical models of membrane inclusions described in this chapter provide a tool to study processes in membranes that involve membrane regions with high curvatures, like spherical buds [1], membrane necks [45, 46], or thin tubular membrane protrusions [12, 47].

ACKNOWLEDGMENTS The authors appreciate the collaboration with S. May and H. Ha¨gerstrand and are grateful to Blazˇ Babnik for technical assistance.

REFERENCES [1] H. Ha¨gerstrand, L. Mrowczynska, U. Salzer, R. Prohaska, K.A. Michelsen, V. Kralj-Iglicˇ, A. Iglicˇ, Curvature dependent lateral distribution of raft markers in the human erythrocyte membrane, Mol. Membr. Biol. 23 (2006) 277–288. [2] H. Gruler, Chemoelastic effect of membranes, Z. Naturforsch. 30c (1975) 608–614.

Flexible Membrane Inclusions and Membrane Inclusions

167

[3] W. Helfrich, J. Prost, Intrinsic bending force in anisotropic membranes made of chiral molecules, Phys. Rev. A 38 (1988) 3065–3068. [4] R. Oda, I. Huc, M. Schmutz, S.J. Candau, F.C. MacKintosh, Tuning bilayer twist using chiral counterions, Nature 399 (1999) 566–569. [5] A. Iglicˇ, B. Babnik, U. Gimsa, V. Kralj-Iglicˇ, On the role of membrane anisotropy in the beading transition of undulated tubular membrane structures, J. Phys. A: Math Gen. 38 (2005) 8527–8536. [6] A. Iglicˇ, M. Tzaphlidou, M. Remsˇkar, B. Babnik, M. Daniel, V. Kralj-Iglicˇ, Stable shapes of thin anisotropic nano-strips, Fullerenes Nanotubes Carbon Nanostruct. 13 (2005) 183–192. [7] L.D. Landau, E.M. Lifshitz, Theory of Elasticity (3rd ed.), Butterworth-Heinemann, Oxford, 1997. [8] A. Iglicˇ, T. Slivnik, V. Kralj-Iglicˇ, Elastic properties of biological membranes influenced by attached proteins, J. Biomech. 40 (2007) 2492–2500. [9] D.V. Widder, Advanced Calculus, Prentice-Hall, Inc., New York, 1947. [10] V. Kralj-Iglicˇ, S. Svetina, B. Zeks, Shapes of bilayer vesicles with membrane embedded molecules, Eur. Biophys. J. 24 (1996) 311–321. [11] V. Kralj-Iglicˇ, V. Heinrich, S. Svetina, B. Zˇeksˇ, Free energy of closed membrane with anisotropic inclusions, Eur. Phys. J. B 10 (1999) 5–8. [12] V. Kralj-Iglicˇ, H. Ha¨gerstrand, P. Veranicˇ, K. Jezernik, A. Iglicˇ, Amphiphile-induced tubular budding of the bilayer membrane, Eur. Biophys. J. 34 (2005) 1066–1070. [13] M. Fosnaricˇ, K. Bohinc, D.R. Gauger, A. Iglicˇ, V. Kralj-Iglicˇ, S. May, The influence of anisotropic membrane inclusions on curvature elastic properties of lipid membranes, J. Chem. Inf. Mod. 45 (2005) 1652–1661. [14] S. Marcˇelja, Chain ordering in liquid crystal II. Structure of bilayer membranes, Biochim. Biophys. Acta 367 (1974) 165–176. [15] S. Marcˇelja, Lipid-mediated protein interaction in membranes, Biochim. Biophys. Acta 455 (1976) 1–7. [16] W. Helfrich, Elastic properties of lipid bilayers: Theory and possible experiments, Z. Naturforsch. 28 (1973) 693–703. [17] J.B. Fournier, Microscopic membrane elasticity and interactions among membrane inclusions: Interplay between the shape, dilation, tilt and tilt-difference modes, Eur. Phys. J. B 11 (1999) 261–272. [18] M. Rappolt, A. Hickel, F. Bringezu, K. Lohner, Mechanism of the lamellar/inverse hexagonal phase transition examined by high resolution X-ray diffraction, Biophys. J. 84 (2003) 3111–3122. [19] S. May, A. Ben-Shaul, Molecular theory of lipid-protein interaction and the La–Hp transition, Biophys. J. 76 (1999) 751–767. [20] T.C. Lubensky, F.C. MacKintosh, Theory of ‘‘ripple’’ phases of lipid bilayers, Phys. Rev. Lett. 71 (1993) 1565–1568. [21] J.B. Fournier, Coupling between membrane tilt-difference and dilation: A new ‘‘ripple’’ instability and multiple crystalline inclusions phases, Europhys. Lett. 43 (1998) 725–730. [22] U. Seifert, J. Shillcock, P. Nelson, Role of bilayer tilt difference in equilibrium membrane shapes, Phys. Rev. Lett. 77 (1996) 5237–5240. [23] P.G. Dommersnes, J.B. Fournier, N-body study of anisotropic membrane inclusions: membrane mediated interactions and ordered aggregation, Eur. Phys. J. B 12 (1999) 9–12. [24] P.G. Dommersnes, J.B. Fournier, The many-body problem for anisotropic membrane inclusions and the self-assembly of ‘‘Saddle’’ defects into an ‘‘Egg Carton’’, Biophys. J. 83 (2002) 2898–2905. [25] W. Helfrich, Blocked lipid exchange in bilayers and its possible influence on the shape of vesicles, Z. Naturforsch 29c (1974) 510–515. [26] E.A. Evans, R. Skalak, Mechanics and Thermodynamics of Biomembranes, CRC Press, Boca Raton, Florida, 1980. [27] W.C. Hwang, R.E. Waugh, Energy of dissociation of lipid bilayers from the membrane skeleton of red cells, Biophys. J. 72 (1997) 2669–2678. [28] M. Fosˇnaricˇ, A. Iglicˇ, S. May, Influence of rigid inclusions on the bending elasticity of a lipid membrane, Phys. Rev. E 174 (2006) 051503/1–051503/12. [29] V.S. Markin, Lateral organization of membranes and cell shapes, Biophys. J. 36 (1981) 1–19.

168

M. Fosˇnaricˇ et al.

[30] U. Seifert, Configurations of fluid membranes and vesicles, Adv. In Phys. 46 (1997) 13–137. [31] M. Laradji, P.B.S. Kumar, Dynamics of domain growth in self-assembled fluid vesicles, Phys. Rev. Lett. 93 (2004) 198105/1–4. [32] J.M. Allain, M. Ben Amar, Biphasic vesicle: Instability induced by adsorption of proteins, Physica A 337 (2004) 531–545. [33] U. Seifert, Curvature-induced lateral phase segregation in two-component vesicles, Phys. Rev. Lett. 70 (1993) 1335–1338. [34] S. Leibler, Curvature instability in membranes, J. Phys. (France) 47 (1986) 507–516. [35] L.B. Fournier, Nontopological saddle-splay and curvature instabilities from anisotropic membrane inclusions, Phys. Rev. Lett. 76 (1996) 4436–4439. [36] B. Bozˇicˇ, V. Kralj-Iglicˇ, S. Svetina, Coupling between vesicle shape and lateral distribution of mobile membrane inclusions, Phys. Rev. E 73 (2006) 041915/1–11. [37] S. May, Membrane perturbations induced by integral proteins: Role of conformational restrictions of the lipid chains, Langmuir 18 (2002) 6356–6364. [38] N. Dan, S.A. Safran, Effect of lipid characteristics on the structure of transmembrane proteins, Biophys. J. 75 (1998) 1410–1414. [39] N. Dan, P. Pincus, S.A. Safran, Membrane-induced interactions between inclusions, Langmuir 9 (1993) 2768–2771. [40] C. Nielsen, M. Goulian, O.S. Andersen, Energetics of inclusion-induced bilayer deformations, Biophys. J. 74 (1998) 1966–1983. [41] A. Ben-Shaul, Molecular theory of chain packing, elasticity and lipid-protein interactions in lipid bilayers, in: R. Lipowsky, E. Sackmann (Eds.), Structure and Dynamics of Membranes, Elsevier, Amsterdam, 1995, p. 382. [42] W. Helfrich, Elastic properties of lipid bilayers: Theory and possible experiments, Z. Naturforsch. 28c (1973) 693–703. [43] S. May, Y. Kozlovsky, A. Ben-Shaul, M.M. Kozlov, Tilt modulus of lipid monolayer, Eur. Phys. J. E 14 (2004) 299–308. [44] T.L. Hill, An Introduction to Statistical Thermodynamics, General Publishing Company, Toronto, 1986, pp. 209–211. [45] A. Iglicˇ, B. Babnik, K. Bohinc, M. Fosˇnaricˇ, H. Ha¨gerstrand, V. Kralj-Iglicˇ, On the role of anisotropy of membrane constituents in formation of a membrane neck during budding of a multicomponent membrane, J. Biomech. 40 (2007) 579–585. [46] V. Kralj-Iglicˇ, P. Veranicˇ, Curvature-induced sorting of bilayer membrane constituents and formation of membrane rafts, in: A. Leitmannova Liu (Ed.), Advances in Planar Lipid Bilayers and Their Liposomes, Elsevier, Amsterdam, London, 2007, pp. 129–149. [47] A. Iglicˇ, H. Ha¨gerstrand, P. Veranicˇ, A. Plemenitasˇ, V. Kralj-Iglicˇ, Curvature induced accumulation of anisotropic membrane components and raft formation in cylindrical membrane protrusions, J. Theor. Biol. 240 (2006) 368–373.

C H A P T E R

S E V E N

A New Class of Peptide-Forming Channel: Calcitonins Silvia Micelli,1,* Daniela Meleleo,1 H. Ti Tien,2 Angelica Leitmannova Liu,2,3 Vittorio Picciarelli,4 and Enrico Gallucci1 Contents 1. Introduction 2. Materials and Methods 2.1. The Single-Channel Measurements 2.2. Chemicals 2.3. The Data Analysis 3. Evidence of hCt Channels Formation 3.1. hCt Can Form Voltage-Dependent Channels with Weak Anion Selectivity 4. Role of Detergents on the Characteristics of hCt Single Channels 4.1. hCt Channel Activity Can Be Increased by SDS 4.2. There Is an Optimal Molecular Ratio Between hCt:SDS for hCt Channel Activity 4.3. The Addition of SDS to hCt Does Not Change the Voltage Dependence or the Lifetime, While Its Selectivity Shifts to Cations 4.4. Different Detergents Produce Different Effects on hCt Channel Activity 5. Effects of pH Variations on Insertion and Channel Formation of hCt 5.1. pH Bathing Condition Has Different Effects on the Channel Characteristics Depending on the Nature of PLM 6. Investigations on sCt and eCt 6.1. sCt and eCt Channels Form Voltage-Dependent Channels Having Almost Equal Selectivity for Anions/Cations 6.2. eCt Glycosylation Modifies the Activation Voltage to Form Channels and Shifts Ion Selectivity to Anion 7. Concluding Remarks 7.1. The Results

171 173 173 174 175 176 176 178 180 181 183 183 184 184 189 189 190 193 194

* Corresponding author. Tel.: þ39 0805442794; Fax: þ39 0805442796; E-mail address: [email protected] (S. Micelli). 1 2

3

4

Department of Farmaco-Biologico, Universita` degli Studi di Bari, 70126 Bari, Italy Membrane Biophysics Laboratory, Biomedical and Physical Sciences Building, Department of Physiology, Michigan State University, East Lansing, MI 48824, USA Department of Microelectronics, Faculty of Electrical Engineering and Information Technology (FEI STU), Slovak Technical University, 812 19 Bratislava, Slovak Republic Department of Interateneo di Fisica, Universita` degli Studi di Bari, 70126 Bari, Italy

Advances in Planar Lipid Bilayers and Liposomes, Volume 7 ISSN 1554-4516, DOI: 10.1016/S1554-4516(08)00007-0

#

2008 Elsevier Inc. All rights reserved.

169

S. Micelli et al.

170

7.2. Relationships with Other Research 7.3. Tentative Interpretation of Physiological Relevance References

194 198 198

Abstract Calcitonin is a small peptide hormone present in all vertebrates with an important effect on Ca2þ metabolism. It exerts its main action by inhibiting osteoclast-mediated bone resorption and inducing calcium regulation through an interplay of clearance through the kidney and intestine. This action is responsible for its widespread clinical use in the treatment of bone disorders, including Paget’s disease, osteoporosis, hypercalcemia of malignancy, and musculoskeletal pain. Unfortunately, this peptide (at least the human calcitonin, or hCt) shows a high tendency to aggregate, resulting in fibrillation that impairs its use as a drug for the above pathologies. This could explain why other calcitonins with a lower tendency to fibrillate, in particular salmon and eel calcitonin (sCt, eCt), are currently in use. The mechanism of fibrillation is common to other peptides such as insulin, prion protein, Alzheimer Ab, and the a-synuclein responsible for Parkinson’s disease, is far from being well understood despite the many investigations which have been undertaken. Thus, if the mechanism underlying the fibrillation of one peptide could be resolved, this may be extrapolated to the other fibrillating peptides, thus shedding light on the ‘‘channel pathologies’’ which are presumably linked by a common denominator. The technique used to study the fibrillation mechanism is performed both in solution and on the membrane surface. In fact, it has been found that membrane phospholipids can play an important role either inducing or counteracting aggregation by incorporating peptides into the membrane. In the past few years, we have been performing systematic studies using model Planar Lipid Membranes (PLM) in order to:  gather evidence of calcitonin incorporation into the membrane and channel formation under different experimental conditions (i.e., membrane applied voltage, bathing conditions);  test the influence on hCt channel properties of varying the PLM composition, or the pH of the medium, and of inducing and stabilizing a-helices by means of an anionic detergent;  understand the role of glycosylation at a different position in eCt during peptide incorporation and channel formation. The results obtained can be summarized as follows:  on the ability to form channels: hCt and sCt are able to form channels when incorporated into DOPC/DOPG (molar ratio 85:15) model membranes. Moreover, the average single-channel conductance decreases as a function of the membrane applied voltage (from 0.2/0.5 nS at 10 mV for hCt and sCt, respectively, to 0.014 nS at 150 mV for both) and has different values under different bathing conditions (higher for NaCl, CaCl2, KCH3COO than for KCl). The permeability for cations/anions is almost identical. Finally eel, porcine, and carbocalcitonin also form channels, with a mean conductance higher than that of hCt and sCt.  on the way in which the acid-base conditions in the bathing medium affect the hCt channels: At pH 7, hCt is able to interact and forms channels with negatively charged dioleoyl-phosphatidyl-glycerol (DOPG) bilayers and with zwitterionic palmitoyl-oleoyl-

A New Class of Peptide-Forming Channel

171

phosphatidylcholine (POPC) bilayers containing 15% negatively charged DOPG, but not with POPC bilayers. At low pH (4.4 and 3.8), the conformational variation of the peptide enables it to insert into POPC and POPC:DOPG but not into DOPG bilayers.  on the effect of detergent on the hCt channels: Sodium dodecyl sulfate (SDS) an anionic detergent able to induce and stabilize a-helices of polypeptides, at concentrations ranging between 0.26 mM and 5 pM [all concentrations are below the critical micelle concentration (CMC)] increases the rate and number of hCt channel formation in PLM at both high and low hCt concentration, with a maximum increase at a molecular hCt/SDS ratio of 1000:1. The action of SDS could be attributable to the strength of the sulfate negative charge and the hydrophobic chain; in fact a similar effect was obtained with lauryl sarcosine and not with a neutral detergent such as n-dodecyl-b-D-maltoside.  on the glycosylation at different positions in eCt: eCt glycosylation at different positions (Ct3-GlcNAc, Ct14-GlcNAc, Ct20-GlcNAc, and Ct26-GlcNAc) preserves the molecular structure and slightly changes the energy of incorporation and channel formation into POPC: DOPG (85:15 w/w). The voltage needed to form channels decreases as the attached carbohydrate moves toward the C-terminal (eCt ¼ Ct3-GlcNAc > Ct14-GlcNAc ¼ Ct20-GlcNAc > Ct26-GlcNAc). Interestingly, all the Cts tested maintain the characteristic voltage–conductance dependence found for other Cts; the only channel property modified being ion selectivity that shifts toward anion selectivity (eCt ¼ 0.97, Ct3-GlcNAc ¼ 0.49, Ct14-GlcNAc ¼ 0.41, Ct20-GlcNAc ¼ 0.36, and Ct26-GlcNAc ¼ 0.47).

1. Introduction Human calcitonin (hCt) is a 32-amino acid amphipathic peptide synthesized and secreted by the C-cells of the thyroid gland and involved in calcium metabolism and bone regulation. In fact, it inhibits osteoclast-mediated bone resorption and induces calcium uptake from body fluid. This action has suggested a widespread clinical use of hCt for the treatment of bone disorders, including Paget’s disease, osteoporosis, hypercalcemia of malignancy and musculoskeletal pain [1]. Unfortunately, hCt shows an extremely high tendency to self-associate to form amyloid fibrils that can deposit in vivo in patients with medullar carcinoma of the thyroid and in vitro in preparations designed for patient administration, thus causing serious problems for production, storage, and administration. Moreover, the aggregation causes a significant decrease in its activity as a drug. In fact, if hCt is manipulated through means that inhibit molecular aggregation, not only does its activity improve, but it becomes more effective than salmon Ct (sCt) which has a lower propensity to aggregate [2]. Electron microscopic studies have shown that the fibrils consist of fibers of 8 nm in diameter and that they often associate with one another [3]. The mechanism of fibril formation is under intensive investigation; however, it is critically influenced by solid–liquid interface and is also time, pH, and concentration dependent [4–7]. In particular, it has been shown that in acidic aqueous solutions [8], an amphiphilic a-helical structure is formed in the central region of hCt. On the other hand, local conformational transitions from a-helix to b-sheet at the C-terminus region were simultaneously induced during fibril formation in acetic acid solution (pH 3.3) [6].

172

S. Micelli et al.

The mechanism of fibril formation seems to consist in a first step in which the nucleus of fibrils forms by a homogeneous process, followed by a second step in which the fibrils mature by a heterogeneous autocatalytic process [6]. Besides, fibril formation depends on pH; in fact, at neutral pH, fibril formation is much faster than in an acidic medium because the hCt monomer is more stable. The fibrils formed at pH 7.5 are composed of antiparallel b-sheets, whereas the mixture of antiparallel and parallel b-sheet structures is formed at pH 3.3. A model has been put forward to explain hCt fibrillation, which takes into account the protonation of Lys18 and the deprotonation of Asp15 at different pH values [6]. Recent work indicates a critical role of residues 18 and 19 for the oligomerization state and bioactivity of hCt. The ability of the short fragment (15–19) of hCt to form fibrils was investigated, demonstrating that this pentapeptide has remarkable amyloidogenic potential [9]. Because of its lower propensity to aggregate, in spite of its low sequence identity (about 50%) with hCt, sCt is a good candidate as a therapeutic alternative. Furthermore, unlike hCt, sCt prevents or retards bone loss and increases bone strength in rat and other species. A similar anabolic effect is also found for cartilage formation, bone matrix synthetic activity, and bone growth [1]. However, there are various side effects such as immune responses, resulting in resistance or allergic reactions on the part of recipients [10–12], and increases in drug-induced cytotoxicity [13, 14] that limit its therapeutic use and suggest the need to look for a better bioavailability. Concerning the mechanism promoting the interaction between Ct and cells, Segre and Goldring [15] have proposed that Ct associates specifically with cells through a membrane-bound receptor of a superfamily of seven transmembranespanning helices G protein-coupled receptors that act via adenylyl cyclase and/or phosphoinositide-specific phospholipase C pathways. However, G-protein stimulation by bovine Ct and b-amyloid (Ab) has also been shown on cells where the receptors had been removed [16]. Ct structure is consistent with an N-terminal portion, specific for receptor activation, and a more variable central helical segment [17]. It is worth mentioning that one of the sites at which Ct interacts with the receptor is located deep inside the membrane [17]. On the other hand, when Ct is reconstituted into PLMs made up of POPC:DOPG (85:15), its molecules aggregate to form ion channels [18–21]. In the past few years, we have been involved in various investigations stimulated by the data collected and the analyses performed by other researchers, aiming to look for new approaches that could help increase hCt channel formation activity and manipulate their channel characteristics in order to fit therapeutic requirements counteracting hCt’s fibrillating property, by systematically comparing the properties of channels due to other calcitonin peptides like sCt, eel calcitonin (eCt), or their modified forms. Figure 1 shows the difference in the 32 amino acid composition existing among the calcitonins used in our investigations. In this paper, after discussing the experimental method used in our investigations and the data analysis performed (in Section 2), we first of all report the evidence collected on hCt channel formation and on channel characteristics (Section 3), the results obtained regarding the role of detergents (Section 4), and the effects of pH

A New Class of Peptide-Forming Channel

173

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 Human Ct

C G N L S T C M L G T Y T Q D F N K F H T F P Q T A I G V G A P

Salmon Ct

C S N L S T C V L G K L S Q E L H K L Q T Y P R T N T G S G T P

Eel Ct

C S N L S T C V L G K L S Q E L H K L Q T Y P R T D V G A G T P

Figure 1

Amino acid compositions of human (hCt), salmon (sCt), and eel (eCt) calcitonin.

variations on hCt insertion and channel formation (Section 5), an analysis of the data obtained with other calcitonins—in particular with sCt and eCt. For the latter, we also discuss the effects of glycosylation on channel characteristics (Section 6). A tentative interpretation of the mechanisms that could give rise to the properties observed and full discussions of the implications for further research and therapeutic applications are reported in our concluding section (Section 7).

2. Materials and Methods In this section, we briefly discuss the measurement approach, the chemicals used, and the data analysis performed to gather information on various channel characteristics (mean conductance, voltage dependence and gating charge, calcitonin activity and channel occurrence, mean channel lifetime, and channel permeability).

2.1. The Single-Channel Measurements The Mueller–Rudin or painted technique [22–24] was used to form PLM with lipids solubilized in n-decane (1% w/v). Briefly, a small volume of 0.5–1.0 ml of lipid solution is applied through a micropipette directly onto the hole of the Teflon set; a PLM forms in 10 min after draining the excess solvent into the aqueous bathing solution. The experimental set-up used to study the interactions of Cts with PLM is reported in Fig. 2. A Teflon chamber is used that has two aqueous compartments connected by a small circular hole with a diameter of 0.2 mm. The membrane current is monitored with an oscilloscope and recorded on a chart recorder for data analysis by hand. The cis and trans chambers are connected to the amplifier head stage by Ag/AgCl electrodes in series with a voltage source and a highly sensitive current amplifier. The single-channel instrumentation has a time resolution of 1–10 ms depending on the magnitude of the single-channel conductance. The polarity of the voltage is defined according to the side where the Ct is added (the cis-side). A trans-negative potential (indicated by a minus sign) means that a negative potential was applied to the trans-side, the compartment opposite the one where the Ct is added. In some experiments, Ct is added to both aqueous phases of black bilayer membranes without affecting the results. These experiments are considered to be a sensitive tool for the study of channelforming substances [25, 26]. For these measurements, it is important that the substances

S. Micelli et al.

174

ac Vdc

5 mVfixe Output impedence adapter Absolute signal rectifier

Rm t

Amplifier and level adjust

Cm dc gen. ac gen.

Sum Vs − +

OPA 129 Vl

− +

Active 2 stage pass band filter 10 kHz

Vlh To recorder CH 1

dc amplifier

Vdc Amplifier and level adjust

Salt solution General amplifier

To recorder CH 2

Clear RC filter

Output impedence adapter

Figure 2

Experimental set-up used during investigations into Ct single-channel incorporation.

form defined channels and do not act as detergents inducing lipid perturbation. The observation of non-random discrete step-like increases in conductance could be considered to conform to these criteria. Nevertheless, various experiments are performed preliminarily in order to test for channel formation. For example first of all, we test the conductance and capacitance of each membrane by applying a voltage of 200 mV for 10–15 min under stirring, to ensure that the membrane is stable. Upon calcitonin addition, current jumps compatible with channel-type openings and closures with different conductance levels and lifetimes are observed. On the other hand, channel formation is inhibited when protease is added to the medium before calcitonin addition. Control experiments with protease present in the medium facing the membrane fail to give discrete current fluctuations for several hours.

2.2. Chemicals Salts and other basic chemicals were bought from Merck (Darmstadt, FRG, analytical grade) and biochemicals from Sigma (Mu¨nchen, FRG). The phospholipids used in this study were purchased from Avanti Polar Lipids, (Alabaster, AL). hCt was from Novartis Pharma AG (Basel, Switzerland), porcine and eel calcitonin were purchased from Calbiochem (USA), Asn26-eCt was purchased from Sigma, and glycosylated eCts from K. Toma (The Noguchi Institute, Tokyo, Japan).

A New Class of Peptide-Forming Channel

175

2.3. The Data Analysis The data obtained in our experiments were analyzed in order to estimate the mean conductance and its voltage dependence, various parameters related to calcitonin– PLM interactions (calcitonin activity, channel occurrence, and activation time), channel lifetime, and permeability ratio. 2.3.1. Calcitonin activity and channel occurrence The current through each channel is proportional to the applied voltage; the ratio between the channel current and the applied voltage gives the conductance of the channel. Depending on the polarity of the applied voltage, the increase in current, open channel, can be seen as deflections, either upward (for positive voltages) or downward (for negative voltages). Closure of the channel is represented by a deflection on the opposite site. Often, the openings are much more frequent than the closures. Calcitonin activity was measured by the total number of current jumps (i.e., events) observed (N ), while the occurrence represents the mean number of events in a period of 60 s, that is, the N observed divided by the observation time. 2.3.2. Mean conductance for single channel In general, the single-channel data were obtained from at least two/four experiments performed on different days. All single-channel events (more than 100 singlechannel events for each experiment) were used to calculate the event conductance irrespective of duration. A histogram of conductance distribution for each experiment was constructed and fitted by a Gaussian distribution function (GraphPad PrismTM version 3.0; GraphPad Software, Inc., http://www.graphpad.com). Results are expressed as mean  S.E. 2.3.3. Voltage dependence and gating charge We investigated the voltage dependence of calcitonin channels in all PLMs used by performing experiments at various voltages (in the range 10–200 mV) and measuring the amplitude of channel events, from which the mean conductance Lc was estimated. The data were parameterized, in general, with the exponential form:

Lc ¼ AeðKVm Þ þ p

ð1Þ

where A is the difference between the conductance at Vm ¼ 0 and at Vm ¼ membrane black ( p) and K is the constant correlated with the gating charge n (n ¼ KRT/F ). R, T, and F have their usual meanings. 2.3.4. Lifetime of the channels Single-channel recordings with a conspicuous number of channels (opening–closing) were analyzed to obtain cumulative open-state lifetime distributions.

S. Micelli et al.

176

The average lifetime of the conductance unit was estimated by the formula:

N ¼ A1 eðt=t1 Þ þ A2 eðt=t2 Þ

ð2Þ

where N is the number of channels that remain open for a time equal to or greater than a certain time t, A1 and A2 are the zero time amplitudes, and t1 and t2 are related to the fast and slow components of the time constant. The single-exponential distribution is included in the formula (A2 ¼ 0). In order to choose between the two models, we performed an appropriate statistical test (F-test Graphpad Prism 3). 2.3.5. The permeability ratio To identify the charge on the ion carrying the current, we measured the shift in the reversal potential induced by a change from a symmetrical to an asymmetrical KCl solution system. When the membrane conductance reached a virtually stable value following Ct insertion, the salt concentration on the cis-side of the membrane was raised by the addition of concentrated salt solution. Generally, an increase in salt concentration of between two- and ten-fold was used. In the absence of applied voltage, the ions for which the channel is selective will pass from the cis to the transside, through the channel, creating a current across the membrane. The generated membrane potential or reversal potential was determined by changing the holding potential step by step by 2 mV and measuring the mean reversal potential Vr, which was used to solve the Goldman–Hodgkin–Katz equation:

 Vr ¼

RT F



  PK ½Kt þ PCl ½Clc ln PK ½Kc þ PCl ½Clt

ð3Þ

where [X]t and [X]c are the concentrations of the ion species X in the trans and cis compartments, respectively; R, T, and F have their usual meanings; and to determine the ratio of permeability of cations to anions.

3. Evidence of hCt Channels Formation When incorporated into PLMs made up of DOPG, POPC:DOPG (85:15, w:w) hCt forms channels, whereas it does not incorporate into PLMs made up of POPC. Our investigation aimed to characterize the main channel properties.

3.1. hCt Can Form Voltage-Dependent Channels with Weak Anion Selectivity We have performed experiments with membranes formed from a 1% (w/v) solution of pure POPC and of a mixture of POPC and DOPG (molar ratio 85:15) and various hCt concentrations (5 nM, 24.5 nM, 49 nM, 85 nM, and 125 nM) and variable voltage (from a minimum of 10 mV to a maximum of 200 mV) applied to the Ag/AgCl electrodes.

177

A New Class of Peptide-Forming Channel

Moreover, in order to investigate the influence of the bathing conditions, the membranes were bathed in different solutions (1 M KCl, 1 M NaCl, 1 M CaCl2). The main results can be summarized as follows:   





at high concentrations, hCt shows a lower channel activity that can be overcome by decreasing its concentration or by increasing the applied voltage up to 150 mV. no channel activity was observed in PLMs made up of POPC. channels were observed (see Fig. 3 for a sample recording of the transmembrane current) in PLMs made up of a mixture of POPC and DOPG (molar ratio 85:15) and of pure DOPG. a typical exponential dependence of the mean channel conductance Lc on the membrane applied voltage was found for all hCt concentrations. An example, in the POPC/DOPG mixture, at a mean concentration value of 49 hCt, is shown in Fig. 4(□). a considerable variation in hCt single-channel conductance was observed with different salt solutions (see Table 1) by using DOPC:DOPG membranes (hCt 49 nM). 0.033 nS 5.025 pA POPC

POPC:DOPG

100 s

0.033 nS 5.025 pA 5s

DOPG

0.033 nS 5.025 pA 2s

Figure 3 Recording of transmembrane current through PLMs containing hCt channels in each PLM used. Each trace represents a fragment of the recording of the activity obtained in individual experiments with different PLMs: (A) POPC, (B) POPC:DOPG, and (C) DOPG. Experimental conditions: KCl 1 M, hCt 125 nM was present on the cis-side of the membrane, pH 7, and 150 mV membrane applied voltage.

S. Micelli et al.

178

0.3

Λc

0.2

0.1

0.0 0.00

0.05

0.10 0.15 Vm/V

0.20

0.25

Figure 4 Conductance–voltage relationship for hCt channels in the absence and in the presence of sodium dodecyl sulfate (SDS) (0.26 mM). Experimental conditions: KCl 1 M, hCt (49 nM) þ SDS (0.26 mM) was present on the cis-sides of the POPC:DOPG (85:15) membrane. The curves superimposed on the data are the results of the fit with the model: Lc ¼ AeðKVm Þ þ p, where A is the difference between the conductance at Vm ¼ 0 and at Vm ¼ membrane black (p); K is the constant correlated with the gating charge n (n ¼ KRT/F). (▪) A ¼ 0.234  0.004 (nS), p ¼ 0.0067 (nS), K ¼ 37.13  0.59 (V1), R2 ¼ 0.988; (□) A ¼ 0.29  0.002 (nS), p ¼ 0.0067 (nS), K ¼ 38.3  0.48 (V1), R2 ¼ 0.99.

the permeability ratio, PKþ =PCl , measured by the KCl concentration gradient method described above, was 0.71 for POPC:DOPG membranes, indicating a poorly anion-selective channel. Approximately the same result was obtained with the I–V curve, where the measured amplitude of the channel events at each potential was used.  the lifetime of the single channel was used to further characterize the channels. For this analysis, no less than 100 individual channels (opening and closing) were utilized. 

Independent of the calcitonin concentration or voltage used, the distribution of the open times was found to follow a two-exponential function, except for the calcitonin concentration of 85 nM in which the statistical test does not distinguish between a one- or two-exponential function. t1 ranges between 0.1 and 1 s, where the lowest value is observed for low calcitonin concentrations, whereas t2 ranges between 3.3 and 7.5 s.

4. Role of Detergents on the Characteristics of hCt Single Channels As previously discussed, at higher concentrations, hCt requires very high membrane potential (150 mV) to induce channel formation, due to its fibrillation properties. Since SDS is an a-helical inductor and a membrane mimicking agent, we tested whether SDS could change the properties of hCt channels, such as mean

Table 1

The effective conductance of hCt for different salt solutions KCl (1 M)

V (mV)

150 80 50 10

Lc  S.D. (nS) 5

0.014  8.5E 0.026  4.3E4 0.038  1.5E3 0.160  1.8E3

Occur.  S.D.

0.61  0.01 5.37  0.01 4.96  0.02 1.85  0.04

NaCl (1 M) Lc  S.D. (nS) 3

0.014  5.5E 0.029  1.3E2 0.047  1.7E2 0.210  6.9E2

Occur.  S.D.

11.25  0.06 25.76  0.10 16.99  0.09 8.58  0.05

CaCl2 (1 M) Lc  S.D. (nS) 3

0.019  7.4E 0.039  1.3E2 0.068  2.6E2 0.185  5.2E3

Occur.  S.D.

2.00  0.01 4.29  0.02 6.77  0.03 6.30  0.04

The membranes were formed of POPC/DOPG (85:15), the hCt concentration was 49 nM, pH 7, and T ¼ 22  C. The number of events (Nt) considered for each series of experiments was 128 < Nt < 1135.

S. Micelli et al.

180

conductance, voltage dependence, ion selectivity, onset and occurrence of channel formation, and the voltage at which the channel activity is observed.

4.1. hCt Channel Activity Can Be Increased by SDS First of all, we excluded any nonspecific and destabilizing effect of SDS per se on PLMs by leaving SDS, at all concentrations used, in the medium facing the membrane for up to 24 h. SDS caused no variations in membrane conductance and capacitance. Experiments were then performed by adding hCt to a medium containing SDS at a monomeric concentration of 0.26 mM. After addition of hCt, there was an early onset of step-like activity, indicative of channel formation, characterized by a continuous channel activity, with rare bursts, and with more frequent multiple levels of conductance (Fig. 5B and D). Furthermore, these patterns were more evident A

0.0335 nS 5 pA 20 s

B

20 s C

10 s

D

5s E 5s F 5s

Figure 5 Single-channel features of hCt in the absence and in the presence of sodium dodecyl sulfate (SDS). (A) hCt 125 nM, (B) hCt 125 nM þ SDS (0.26 mM), (C) hCt 24.5 nM, (D) hCt 24.5 nM þ SDS (0.26 mM), (E) hCt 5 nM, (F) hCt 5 nM þ SDS (5 pM). Experiments were performed on a POPC:DOPG (85:15) membrane in the presence of hCt and of hCt þ SDS added to the cis-side; the voltage was set to þ150 mV, the aqueous phase contained 1 M KCl (pH 7) and T ¼ 22  C. Note the increase in channel occurrence (channels/min) when SDS was present in the medium.

181

A New Class of Peptide-Forming Channel

when lower hCt concentrations were used (Fig. 5F and E). Yet at all concentrations of hCt used, the channel occurrence frequency (i.e., the mean number of openings in a period of 60 s) increased and the activation time (i.e., when the first event appeared) decreased compared with the experiments in which only hCt was present (Table 2). However, a lower applied voltage (10 mV) is required to activate channel formation. Furthermore, a t-test showed that, in the presence of SDS, the central channel conductance is statistically modified at each concentration of hCt used except for 49 nM. Another experimental protocol used was to mix peptide and SDS together and after 20 min the mixture was added to the medium facing membrane. The results obtained in this case were comparable to those obtained with the previous experimental protocol. The same results on central conductance, voltage dependence, onset and occurrence of the channels were obtained if the experimental procedure was changed by adding hCt to the membrane before SDS, indicating that the aggregation phenomenon can be reversed. On the other hand, if hCt was present on the cis-side while SDS was added to the trans-side of the membrane, SDS failed to produce the above-mentioned effects.

4.2. There Is an Optimal Molecular Ratio Between hCt:SDS for hCt Channel Activity In order to determine the lower limit of SDS concentration at which the onset of incorporation and the rate of channel formation were modified, we performed an experiment in which hCt was held constant at 125 nM and SDS was changed in the range of 0.0125–12.5 nM (Fig. 6 and Table 3). Table 2 Single-channel parameters for hCt channels in the absence or in the presence of SDS (0.26 mM) and at different hCt concentrations

Mode

Lc  S.D. (nS)

Occur.  S.D.

Activation time (min)

[hCt] ¼ 125 nM [hCt þ SDS] ¼ 125 nM þ 0.26 mM [hCt] ¼ 85 nM [hCt þ SDS] ¼ 85 nM þ 0.26 mM [hCt] ¼ 49 nM [hCt þ SDS] ¼ 49 nM þ 0.26 mM [hCt] ¼ 24.5 nM [hCt þ SDS] ¼ 24.5 nM þ 0.26 mM [hCt] ¼ 5 nM [hCt þ SDS] ¼ 5 nM þ 5 pM

0.014  0.006 0.013  0.005 0.015  0.006 0.014  0.008 0.014  0.006 0.015  0.008 0.013  0.005 0.016  0.009 0.007  0.005 0.007  0.008

2.73  0.09 3.01  0.11 4.26  0.13 6.38  0.24 1.25  0.10 2.49  0.11 1.10  0.06 3.39  0.17 1.26  0.15 3.49  0.18

50 16 16 6 31 19 19 5 10 1

The membranes were formed of POPC/DOPG (85:15), the voltage was set to 150 mV, the aqueous phase contained 1 M KCl (pH 7), and T ¼ 22  C. The number of events (Nt) considered for each series of experiments was 171 < Nt < 1015.

S. Micelli et al.

182

A

0.0335nS 5 pA 20 s

B 5s C 5s D 5s E 10 s F 100 s G 100 s

Figure 6 Single-channel features of hCt in the absence and in the presence of sodium dodecyl sulfate (SDS) at different concentrations. (A) hCt 125 nM, (B) hCt 125 nM þ SDS (12.5 nM), (C) hCt 125 nM þ SDS (1.25 nM), (D) hCt 125 nM þ SDS (0.125 nM), (E) hCt 125 nM þ SDS (0.0125 nM), (F) SDS (0.26 mM), (G) SDS (0.125 nM). Experiments were performed on a POPC: DOPG (85:15) membrane in the presence of hCt and of hCt þ SDS added to the cis-side; the voltage was set to þ150 mV, the aqueous phase contained 1 M KCl (pH 7) and T ¼ 22  C. Table 3 Single-channel parameters for hCt channels in the absence or in the presence of SDS at different concentrations and at fixed hCt concentration

Mode

Lc  S.D. (nS)

Occur.  S.D.

[hCt] ¼ 125 nM [hCt þ SDS] ¼ 125 nM þ 0.26 mM [hCt þ SDS] ¼ 125 nM þ 12.5 nM [hCt þ SDS] ¼ 125 nM þ 1.25 nM [hCt þ SDS] ¼ 125 nM þ 0.125 nM [hCt þ SDS] ¼ 125 nM þ 0.0125 nM

0.014  0.006 2.73  0.09 0.013  0.005 3.01  0.11 0.014  0.006 7.49  0.32 0.014  0.008 10.0  0.54 0.013  0.007 11.68  1.07 0.012  0.004 5.82  0.31

Activation time (min)

50 16 12 6 5 12

The membranes were formed of POPC:DOPG (85:15), the voltage was set at 150 mV, the aqueous phase contained KCl 1 M (pH 7), and T ¼ 22  C. The number of events (Nt) considered for each series of experiments w as 119 < Nt < 836.

183

A New Class of Peptide-Forming Channel

Figure 6F and G as a control also report the results obtained with the maximum (0.26 mM) and the minimum (0.0125 nM) SDS concentration, respectively, on POPC:DOPG membranes. It is interesting to note that there seems to be an optimum molecular ratio between hCt and SDS for obtaining the maximum increase in the occurrence of channel formation and the minimum time of the first channel appearance. The remarkable activity at a very low hCt–SDS molar ratio (1000:1 and 10000:1) could suggest a catalytic action of the detergent on the peptide. Finally, experiments were performed in which hCt channel activity at a low peptide concentration (5 nM) maintained the optimal molecular ratio between hCt and SDS (1000:1). In this case, the SDS-increased channel occurrence is remarkable, indicating that even at non-fibrillating hCt concentrations, quite near the physiological hormone concentration, the detergent is able to exert its action of a-helix induction.

4.3. The Addition of SDS to hCt Does Not Change the Voltage Dependence or the Lifetime, While Its Selectivity Shifts to Cations The voltage dependence and selectivity of hCt channels in the presence of SDS is reported in Fig. 4(▪) for an hCt concentration of 49 nM and SDS concentration of 0.26 nM. As can be seen, the conductance of the hCt channel remains practically unchanged. The same behavior occurs in the presence of high and low SDS concentrations, indicating that the detergent does not modify the voltage dependence of the hCt channel. Furthermore, in the presence of SDS, the channel occurrence increases on decreasing the applied voltage. However, in the presence of SDS (1000:1), the PKþ =PCl was 3.25, indicating that the presence of SDS shifts the selectivity of the channel toward cations, while neither of the lifetime constants t1 and t2 are modified by the SDS.

4.4. Different Detergents Produce Different Effects on hCt Channel Activity In order to test the role of the detergent charge in this phenomenon, two more detergents, namely lauryl sarcosine and n-dodecyl-b-D-maltoside, were used at the same molecular ratio as the optimum activity found for SDS (1000:1). Table 4 Table 4 Single-channel parameters for hCt channels in the presence of different detergents Mode

Lc  S.D. (nS)

Occur.  S.D.

[hCt] ¼ 125 nM [hCt þ SDS] ¼ 125 nM þ 0.125 nM [hCt þ N-lauryl sarcosine] ¼ 125 nM þ 0.125 nM [hCt þ n-dodecyl b-D-maltoside] ¼ 125 nM þ 0.125 nM

0.013  0.007 0.013  0.007 0.013  0.008 0.013  0.010

3.03  0.09 11.68  1.07 11.86  0.52 6.66  0.23

The membranes were formed from POPC/DOPG (85:15), the voltage was set to 150 mV, the aqueous phase contained 1 M KCl (pH 7), and T ¼ 22  C. The number of events (Nt) considered for each series of experiments was 107 < Nt < 140.

184

S. Micelli et al.

reports the central channel conductance, their occurrence and the onset time as compared to the basal value of hCt alone. It can be noted that the occurrence frequency is very low for n-dodecyl-b-Dmaltoside compared with SDS or lauryl sarcosine, indicating that a major role is played by the strength of the charged moiety as anchor to the peptide in facilitating the a-helix formation.

5. Effects of pH Variations on Insertion and Channel Formation of hCt A set of experiments was performed in order to study the effects of pH variations on hCt insertion and channel formation. The basic strategy consists in overcoming the amyloidal formation of hCt by modulating a pH-induced proto-deprotonation of amino acids responsible for the molecule aggregation. With this aim, we concentrated on the role of the pH-induced charge variation of hCt and its ability to interact with a membrane bilayer composed of zwitterionic, negatively charged, and mixed phospholipid membranes. We considered channel activity as being an indication of hCt fraction in an inserted state, even at a quantity too small for detection by less sensitive biochemical methods.

5.1. pH Bathing Condition Has Different Effects on the Channel Characteristics Depending on the Nature of PLM In a first series of experiments, the hCt conductance was studied at a pH value of 7 at which the fibrillation process seems to take place easily [6]. The pH value was then lowered to 4.5 and/or 3.8 (by adding a small quantity of concentrated HCl solution and controlling the pH during and at the end of the experiment). Figure 7 shows a typical example of single-channel recordings for POPC, POPC:DOPG, and DOPG bilayers, respectively, when the medium was KCl 1 M, the initial pH value 7, and the applied voltage 150 mV. It can be observed that single-channel activity appears when 15% of the negatively charged phospholipid in the bilayers is added to POPC and a further increase of activity is observed when pure DOPG membranes are used. We observed alternating periods of ‘‘paroxystic’’ channel activity followed by quiescent periods, open times interrupted by closures, and conductance steps that were twice that of the central conductance, a clear indication that two channels were simultaneously incorporated. On the other hand, when the pH value changes to 3.8, hCt forms single channels in zwitterionic POPC or in POPC:DOPG (85:15) bilayers, but fails to form channels in DOPG PLMs (Fig. 7). In this set of experiment, we observed that the periods of ‘‘paroxystic’’ channel activity are more frequent than at pH 7 in POPC:DOPG PLMs. These periods are followed by quiescent periods and the open time interrupted by briefer closures than at pH 7. Moreover, conductance steps were two to three times that of the central conductance, for all PLMs. Finally, single-channel activity at a pH value of 3.8 is more frequent than at pH 7 (see also Table 4).

185

A New Class of Peptide-Forming Channel

pH = 3.8

pH = 7.0 POPC

POPC:DOPG

DOPG

0.033 nS 5.025 pA 100 s

0.033 nS 5.025 pA 5s

0.033 nS 5.025 pA 2s

POPC

POPC:DOPG

DOPG

0.033 nS 5.025 pA 5s

0.033 nS 5.025 pA 5s

0.033 nS 5.025 pA 100 s

Figure 7 Recordings of transmembrane current of PLMs containing hCt channel in each PLM used at different medium pH values (pH 7 or 3.8, respectively) and 150 mV applied voltage. Channel openings and closings are represented by upward and downward deflections, respectively. Each trace represents a fragment of the recording of the activity obtained in individual experiments. Experimental conditions: KCl 1 M, hCt 125 nM was present on the cis-side of the membrane.

In a second series of experiment, the channel recordings were collected under different pH conditions at different applied voltages. Figure 8 reports examples of single-channel traces for hCt in POPC bilayer at different pHs and applied voltages. At pH 7, it can be seen that although a high voltage is applied, no ion channels were observed; in contrast, by lowering the pH to 4.5 and subsequently to 3.8, the formation of well-defined ion channels at all applied voltages takes place. Note that at pH 3.8, the single-channel activity (i.e., the number of events) is more frequent than at pH 4.5. In this context, the open probability (Po) increases at low pH for POPC:DOPG at all voltages examined, except at 10 mV (data not shown). Table 5 summarizes the central conductance (Lc  S.D.), channel occurrence frequency, hCt single-channel lifetime, t1 and t2, and the activation time at pH 7 and 3.8 for the responsive membranes. For the lifetime, no less than 300 individual channels (opening and closing) were utilized. Independent of the pH values, the distribution of the open times was found to follow a two-exponential function. t1 ranges between 0.13 and 1.40 s, whereas t2 ranges between 2.42 and 7.73 s, except for POPC at pH 3.8 where the open time follows a one-exponential function. At pH 7, hCt forms single channels with similar mean activation times and values to t1 and t2 in negatively charged DOPG and in mixed POPC:DOPG PLMs; on the other hand, Lc values in DOPG are smaller than in POPC:DOPG PLMs. Occurrence is 7.6 times higher in DOPG than in POPC:DOPG PLMs. At pH 3.8, the Lc values in POPC and POPC:DOPG PLMs are not significantly different. The high values of occurrence frequency (higher in POPC than in

S. Micelli et al.

186

pH 7 0.03 nS Vs = 150 mV

pH 4.5 Vs = 10 mV

pH 3.8 0.50 nS 2s

100 s Vs = 30 mV

0.17 nS

Vs = 50 mV

0.10 nS

Vs = 80 mV 0.06 nS

Vs = 150 mV 0.03 nS

Figure 8 Examples of chart recordings of hCt channel formation in POPC PLM at different medium pH values and various holding potential (indicated at the top of the tracings). Channel openings and closings are represented by upward and downward deflections, respectively. Each trace represents a fragment of the recording of the activity obtained in individual experiments at different times. Experimental conditions: KCl 1 M, hCt 125 nM was present on the cis-side of the membrane.

POPC:DOPG PLMs) agree with a good incorporation of peptides through the membranes and their ability to form ion channels at low pH. Besides, hCt presents two lifetimes in POPC:DOPG and one lifetime in POPC PLMs. By comparing hCt channel parameters at two different pH values for POPC: DOPG PLMs, it is worth noting that the Lc values are not different and the occurrence frequency increases three times at low pH. This increase in occurrence frequency may be due to two factors: (a) higher quantities of hCt a-helices, and (b) titration pH-induced membrane surface. Moreover, the activation time is halved compared with pH 7, indicating that both peptide and membrane structural changes play an important role in hCt channel formation. The voltage dependence of hCt channels in all PLMs used was studied by performing experiments at various voltages or by varying the voltage during the experiment (in the range 10–150 mV) and measuring the amplitude of channel events. The data are reported in Fig. 9. hCt channel conductance remains inversely correlated with membrane potential independently of the pH value or membrane composition. The results with POPC: DOPG PLMs corroborate the previous findings [18, 19].

Table 5 The mean conductance fitted by Gaussian distribution (Lc ), the occurrence (channels/min), the fitted lifetimes (see text) of the singlechannel events (s), the activation time and the total number (N) of channels in different PLMs at different pH

PLM

pH

Lc  SD (nS)

POPC

7.0 3.8 7.0 3.8 7.0 3.8

– 0.015  0.008 0.014  0.008 0.014  0.001 0.010  0.008 –

POPC:DOPG DOPG

P

– – 0.0001*

Occur.  S.D.

t1 (s)

t2 (s)

Mean activation time (min)

Number of events

– 14.81  0.41 3.30  0.09 10.16  0.47 25.28  0.41 –

– 1.40 1.36 0.13 0.98 –

– – 7.73 2.42 6.35 –

– 35 49 21 43 –

– 1315 1234 461 3402 –

S. Micelli et al.

188

Λc ± SE/nS

0.3

POPC:DOPG, pH = 3.8 POPC:DOPG, pH = 7.0

0.2 0.1 0.0 0.00

0.05

0.10 Vs/V

0.15

Λc ± SE/nS

0.3

POPC, pH = 3.8

0.2 0.1 0.0 0.00

0.05

0.10 Vs/V

0.15

0.3 Λc ± SE/nS

0.20

0.20

DOPG, pH = 7.0

0.2 0.1 0.0 0.00

0.05

0.10 Vs/V

0.15

0.20

Figure 9 Conductance–voltage relationship for hCt channels in each PLM used at different medium pH values of 7 and/or 3.8, respectively. The data points were obtained from current histograms; at least 300 events were analyzed for each point. The curves superimposed on the data are the results of the fit with the model: Lc ¼ AeðKVm Þ þ p, where A is the difference between the conductance at Vm ¼ 0 and at Vm ¼ membrane black (p), K is the constant correlated with the gating charge n (n ¼ KRT/F). For POPC:DOPG membrane at pH 7.0/3.8: A ¼ 0.33  0.05/0.31  0.04 (nS), p ¼ 0.0067 (nS), K ¼ 57.7  10.0 (V1), n ¼ 1.48/1.20, R2 ¼ 0.97/0.97. For POPC membrane at pH 7.0: A ¼ 0.34  0.02 (nS), p ¼ 0.0067 (nS), K ¼ 47.4  4.2 (V1), n ¼ 1.22, R2 ¼ 0.99. For DOPG membrane at pH 3.8: A ¼ 0.31  0.05 (nS), p ¼ 0.0067 (nS), K ¼ 54.3  10.5 (V1), n ¼ 1.39, R2 ¼ 0.97 Experimental conditions: KCl 1 M, hCt 125 nM was present on the cis-side of the membrane.

To gain information about the ion selectivity of channels at different pHs, experiments in asymmetrical conditions, that is, in the presence of a salt gradient, have been performed on POPC:DOPG PLMs. Both hCt channels at pH 7 and 3.8 in POPC:DOPG membranes show poor anion selectivity in asymmetrical KCl conditions. For hCt channels at pH 7/3.8, the reversal potential was 2.1/3.88 mV in a 0.9/0.5 M ratio of KCl concentration giving a calculated Kþ/Cl permeability ratio of 0.75/0.58, respectively.

A New Class of Peptide-Forming Channel

189

6. Investigations on sCt and eCt Calcitonins are divided into two classes: mammalian and ultimobranchial. As is well known, sCt and eCt, whose structures, as is evident from Fig. 1, differ by only three amino acids, have been thought to have a more potent hypocalcemic action than the mammalian type [27, 28]. In fact, they are widely used for the treatment of bone disorders such as Paget’s disease and hypercalcemia of malignancy. In order to understand if this effect is related to a difference in incorporation and channel formation characteristics into PLMs, we performed systematic investigations on these two types of Cts. Moreover, we have recently also studied the effects of glycosylating different portions of the eCt molecule on channel formation and channel properties in PLMs compared to the channel activity of wild-type eCt. In fact, glycosylation, the most common posttranslational modification of proteins, is a mechanism by which the surface properties of proteins are varied in order to suit the extracellular medium and protect the protein against enzymatic or nonspecific attacks. It has been suggested that glycosylation of calcitonin could enhance peptide receptor affinity, thus varying the biological activity of the peptide [29]. The model employed is eCt [29], because naturally occurring eCt is not glycosylated, so the results found can be easily imputable to glycosylation.

6.1. sCt and eCt Channels Form Voltage-Dependent Channels Having Almost Equal Selectivity for Anions/Cations Let us first of all present the evidence we have collected on channel formation and their main characteristics. Figure 10 reports representative current traces illustrating the single-channel activity for sCt and eCt at 125 and 49 nM concentrations in DOPC/DOPG (molar ratio 85:15) and a membrane applied voltage of 50 mV and 1 M KCl bathing. The single-channel recordings shown demonstrate that discrete conductance steps were observed in the presence of sCt and eCt. The conductance fluctuations were not uniform in size and most of the steps were directed upwards, while downward steps (closing channel) were more rarely observed under these conditions. The mean single-channel conductance as a function of the membrane applied voltages is reported in Fig. 11 for both calcitonins at a concentration of 49 nM. Both sCt and eCt show similar voltage dependence and the data can be fitted by the usual model Lm ¼ Ae(kV ) þ p. In Table 6, we report the main channel characteristics (conductance, occurrence, average lifetime, and selectivity values) obtained for sCt and eCt, respectively, at the same membrane applied voltage þ50 mV and at two concentrations—125 and 49 nM. It can be observed from the results that the main properties of the channels are similar, indicating that the minimal difference in amino acid composition does not play any role. However, it must be remarked that the occurrence of eCt is higher at

190

S. Micelli et al.

A

0.1 nS 5s

B

5s C

5s

D

2s

Figure 10 Example of chart recordings of sCt and eCt channel formation in POPC/DOPG (molar ratio 85:15). The experimental conditions are: KCl 1 M, sCt and eCt 125 nM (A and B) and 49 nM (C and D), and a membrane applied voltage of 50 mV.

0.3

sCt

Λc ± SE

eCt 0.2

0.1

0.0 0.00

0.05

0.10 Vs (V)

0.15

0.20

Figure 11 Conductance–voltage relationship for sCt and eCt channels. The experimental conditions are: KCl 1 M, sCt and eCt 49 nM (▪ and ▲).

the two concentrations tested as compared to sCt, and both eCt and sCt maintain a high rate of channel formation at the lower concentration. This pattern is reminiscent of that found for hCt.

6.2. eCt Glycosylation Modifies the Activation Voltage to Form Channels and Shifts Ion Selectivity to Anion From the amino acid sequence of eCt reported in Fig. 1, it was found that sequence 3–5 constitutes a natural consensus for N-glycosylation, while positions 14 and 20 are Gln instead of Asn and to obtain N-glycosylation at position 26 the Asp was first substituted with Asn (Asn26-eCt).

191

A New Class of Peptide-Forming Channel

Table 6 The main channel characteristics (occurrence, average lifetime, and selectivity values) obtained for sCt and eCt, respectively, at the same membrane applied voltage þ50 mV and two concentrations of 125 nM and 49 nM Mode

Lc  S.D. (nS)

Occur.  S.D.

t1 (s)

t2 (s)

PKþ =PCl

sCt [125 nM] eCt [125 nM] sCt [49 nM] eCt [49 nM]

0.039  0.013 0.036  0.012 0.040  0.033 0.040  0.022

1.17  0.07 1.44  0.09 6.57  0.60 10.93  0.52

– – – 0.47

– – – 3.06

1.03 – – 0.97

The membranes were formed from POPC/DOPG (85:15), the voltage was set to 50 mV, the aqueous phase contained 1 M KCl (pH 7), and T ¼ 22  C. The number of events (Nt) considered for each series of experiments was 117 < Nt < 611.

In our experiments, we investigated the channel-forming activity for Ct3-GlcNAc, Ct14-GlcNAc, Ct20-GlcNAc, and Ct26-GlcNAc and compared the results obtained to those from the wild-type sequence. For Ct26-GlcNAc, a comparative study is reported with Asn26-eCt. Single ion-channel currents were measured in POPC:DOPG (85:15) bilayers using native eCt, four glycosylated eCts, and Asn26-eCt, as reported above. After Ct addition, we usually applied 20 mV and if no channel had appeared after 30 min, the voltage was increased stepwise (each 20/10 mV step lasting 30 min), and the voltage and time at which the first channel appeared (referred to as the activation time) was recorded. As no channels were observed in the 20–120 mV range when eCt or Ct3-GlcNAc was present, in subsequent experiments after Ct addition we started with an applied voltage of 120 mV. The sequence of voltage dependence incorporation is the following: eCt (150 mV) ¼ Ct3-GlcNAc (150 mV) < Ct14-GlcNAc (80 mV) ¼ Ct20-GlcNAc (80 mV) < Ct26-GlcNAc (20 mV). The mean activation time ranges between 70 and 100 min, except for Ct26-GlcNAc, which is 26 min. The analoguesubstituted Asn26-eCt behaves like Ct26-GlcNAc and shows a reduced mean activation time (13 min after addition). This lag time is presumably due to the small area of the hole. In fact, in macroscopic interaction, the lag time is no longer than 8 min. To reduce the lag time of incorporation, in some experiments the peptide was added before membrane formation. The conductance fluctuations appear as alternating periods of paroxystic channel activity followed by quiescent periods, and open times interrupted by brief closures, in which channel transitions between different conductivity states could be observed. Figure 12 shows typical examples of current fluctuations for eCt, glycosylated eCts, and Asn26-eCt, at different applied voltages. One interesting aspect is that all Cts used show strong voltage dependence. In Fig. 13, we report the data on the dependence of conductance on the Vm. By using the fitting procedure previously discussed, we obtained an n value at both positive and negative voltage Vm for each Ct used, and there was no significant difference between them (t-test). The mean value of n was n ¼ 0.97  0.08/1.07  0.12 at Vm > 0/Vm < 0, respectively.

192

S. Micelli et al.

A 0.03 nS 2s B 0.06 nS 2s C 5s 0.25 nS

D 0.10 nS 5s E 5s 0.10 nS F 5s 0.25 n

Figure 12 Representative current traces illustrating the single-channel activity of eCt (A), Ct3GlcNAc (B), Ct14-GlcNAc (C), Ct20-GlcNAc (D), Ct26-GlcNAc (E), and substituted eCt with Asn in position 26 (Asn26-eCt) (F ) recorded from a voltage-clamped POPC:DOPG (85:15) bilayer in symmetrical KCl (1 M cis/trans) with associated histograms of the conductance fluctuations. The voltage was set to 150 mV(A), 80 mV(B), 20 (C, F ), 50 mV (D), 50 mV (E), pH 7, and T ¼ 22  2  C.

Although the majority of peptide channels have the same characteristics, some of them exhibit distinctive variations in both occurrence and lifetime. Table 7 reports the mean values of occurrence and lifetime for the Cts considered in this study, with the highest/lowest mean occurrence/lifetime being shown by Asn26-eCt. To ascertain whether glycosylation could modify channel selectivity, we measured the reversal potentials and calculated the I–V curves in the presence of a concentration gradient of KCl for each Ct used. In all experimental series, we found a very good correspondence of reversal potential with both of the above methods. It can be seen that the eCt channel is almost unselective (PKþ =PCl ¼ 0:97), as is sCt (1.03). The substitution of Asp in position 26 with Asn does not modify the channel selectivity (PKþ =PCl ¼ 0:94), while glycosylation in this position determines a slight shift toward anion selectivity (PKþ =PCl ¼ 0:47). On the other hand, a tendency toward anion selectivity can be seen in the series: Ct3-GlcNAc < Ct14-GlcNAc < Ct20GlcNAc (PKþ =PCl ¼ 0:49, PKþ =PCl ¼ 0:41, PKþ =PCl ¼ 0:36).

193

A New Class of Peptide-Forming Channel

0.3 eCt Ct3–GlcNAc Ct14–GlcNAc Ct20–GlcNAc Ct26–GlcNAc Asn26–eCt

0.0 0.00

0.04

0.08

0.12

0.16

Figure 13 Conductance–voltage relationship for eCt channels. The curves superimposed on the data are the results of a fit with the model: Lc ¼ AeðKVm Þ þ p, where A is the difference between the conductance at Vm ¼ 0 and at Vm ¼ black membrane (p); K is the constant correlated with the gating charge n (n ¼ KRT/F). For Vm > 0/Vm < 0 and p ¼ constant: A ¼ 0.23  0.02/0.28  0.05 (nS), K ¼ 36.1  5.5/30.0  6.8 (V1), R2 ¼ 0.981/0.950 (□); A ¼ 0.23  0.02/0.35  0.03 (nS), K ¼ 37.9  7.7/31.9  4.0 (V1), R2 ¼ 0.970/0.990 (△); A ¼ 0.29  0.01/0.38  0.11 (nS), K ¼ 34.7  3.9/39.6  12.6 (V1), R2 ¼ 0.990/0.920 (r); A ¼ 0.21  0.03/0.42  0.11 (nS), K ¼ 49.4  14.8/52.3  12.8 (V1), R2 ¼ 0.940/0.960 (r); A ¼ 0.22  0.02/0.58  0.18 (nS), K ¼ 42.0  7.2/59.6  15.9 (V1), R2 ¼ 0.980/0.960 (○); A ¼ 0.15  0.01/0.31  0.06 (nS), K ¼ 26.6  5.9/35.3  8.0 (V1), R2 ¼ 0.950/0.960 (*). Table 7 The channel occurrence and lifetime of wild-type and modified eCts

Calcitonin

Occurrence (number of channels/min) [mean  S.E., P ¼ 0.003]

t1 (s) [mean  S.E., P ¼ 0.096]

t2 (s) [mean  S.E., P ¼ 0.033]

eCt Ct3-GlcNAc Ct14-GlcNAc Ct20-GlcNAc Ct26-GlcNAc Asn26-eCt

13.19  3.03 11.76  2.72 14.05  1.33 7.19  2.08 14.49  3.45 25.3  4.17

1.06  0.26 0.90  0.27 0.97  0.17 0.48  0.14 1.05  0.1 1.26  0.09

2.67  1.06 2.68  0.43 3.30  0.93 2.16  0.10 5.83  1.17 –

Values represent the mean values of occurrence and of lifetime obtained at the different voltages.

7. Concluding Remarks Let us now make few final remarks on what we have learned from the reported investigations. We distinguish the results and their relationships with other related research from the possible interpretations and implications. While the first are well established, for the second we can only argue and seek possible reasonable interpretations.

194

S. Micelli et al.

7.1. The Results In our investigations, we have found that at pH 7: 





 

hCt does not show channel activity with PLMs made of POPC, while channels were observed with a mixture of POPC:DOPG (85:15 molar ratio) and DOPG membranes. the mean channel conductance was related to the hCt concentration. At the highest concentration (125 nM), there was practically no channel activity up to 150 mV of membrane applied voltage. independent of the hCt concentrations and voltages used, the distributions of the single-channel lifetime follow a two-exponential behavior, with the time constants t1 and t2 in the range of 0.1–1 s and 3.3–7.5 s, respectively. the channel activity can be increased by SDS with an optimal molar ratio of hCt: SDS (1000:1) at a concentration of 5 nM hCt. the addition of SDS to hCt does not change the mean conductance voltage dependence and the lifetime characteristics while the ion selectivity shifts to cation.

When the bathing conditions were lowered to pH 3.8: 

hCt forms channels in zwitterionic POPC or in POPC:DOPG (85:15) bilayers, but fails to form channels in DOPG;  by comparing the hCt channel characteristics at two different pH values (i.e., 7 and 3.8) for the POPC:DOPG membrane, the mean conductance values remain unchanged and their voltage dependences are similar, the ion selectivity remains poorly anionic being PKþ =PCl ¼ 0:75 (0.58) at pH 7 (3.8), while the occurrence frequency increases three times at the lower pH. Concerning the other calcitonins used, we found that: 

sCt and eCt form channels with POPC:DOPG membranes at pH 7, in contrast to hCt, at the highest concentration (125 nM) and low activation membrane potential.  the glycosylation of eCt at different amino acid positions (namely 3, 14, 16, 26) modifies the activation potential required to form channels in the following sequence: eCt (150 mV) ¼ Ct3-GlcNAc (150 mV) < Ct14-GlcNAc (80 mV) ¼ Ct20-GlcNAc (80 mV) < Ct26-GlcNAc (20 mV) and shifts the ionic permeability from almost unselective (for eCt is PKþ =PCl ¼ 0:97) to slightly anion selective in the range 0.49–0.36.

7.2. Relationships with Other Research The interest in the study of Cts and in particular in hCt has been dictated by the necessity to understand the protein misfolding that many naturally occurrent peptides have in human disease. hCt, for which the kinetics of fibril formation have been ascertained and studied, has been shown to be quite similar to the b-amyloid, responsible for Alzheimer’s disease, with which it shares a propensity to channel formation. It has been found

A New Class of Peptide-Forming Channel

195

that at nanomolar concentrations, SDS seems to inhibit fibril formation in this peptide, thus making hCt of dual interest, both biophysical and technical. In order to accomplish its physiological activity, Ct must interact with receptors, and for this to take place an a-helical conformation is required [30, 31]. Moreover, an a-helix is required to penetrate the bilayer [31–35]. There is plenty of evidence accounting for the folding of membrane-active peptides induced by phospholipid membranes [36]. Most of these peptides require negatively charged phospholipids [18, 37–41]; however, it has been found that some peptides do not require anionic lipids to fold or to induce channels [42–44]. Lipids can either retard or enhance the process, depending on their structure. It is worth mentioning that one of the sites at which Ct interacts with the receptor is located deep in the membrane [17]. As previously stated, we found that at pH 7, Cts interact with DOPG containing POPC membranes and assemble to form channels [18], but not with POPC membranes. This failure has been tentatively explained as being due to the fibrillation (at least for hCt) that occurs on the membrane surface, although this can be counteracted by applying a high potential to the membrane or by adding nanomolar amounts of SDS [18, 19]. Fragment 15–19 is considered the most crucial for fibril formation [9]. It is worth remarking that a similar group is present in AbP(1–40), a peptide that undergoes fibrillation and—similarly to hCt—does not interact with POPC membranes [45]. hCt fibrillation has been studied by various groups of researchers [6, 46] who found that pH can play an important role in the assembly of hCt fibrils. In fact, at low pH (pH 3.8), the side chains of Lys18, His20, and of the NH3 terminus are positively charged, whereas the negatively charged Asp15 is protonated. These variations in the molecular organization of hCt slow down the formation of fibrils, which are composed of a mixture of antiparallel and parallel b-sheets up to the furthest C-terminal region [47]. Moreover, the molecular conformation of both POPC membranes and hCt could be responsible for the lack of interaction at pH 7. This could be the result of POPC’s inability to form negative curvature in the presence of hCt at pH 7. In support of this is the finding that a small amount of DOPG (15%) is needed to catalyze hCt incorporation and molecular assembly in the POPC membrane at pH 7, or a reduction in the pH (3.8) of the medium that protonates the His20 group of the peptide and adds one positive charge in the area of the peptide which is known to contain the fibrillating sequences [9]. An increase in the number of positive charges in the peptide would cause a repulsive force between congener molecules, thus not allowing the peptides to interact with each other to form the seed of nucleation preceding the fibrillation. Furthermore, in organic solvents or SDS micelles at low pH, the protonation of Asp15 allows hCt to form one more helix turn involving the amino acid segment from 15/16 to 20/21 [9]. At low pH, the positive lipid headgroup charge dominates the equilibrium and although the peptide also shows a positive charge, it could be the protonated Asp group that is responsible for incorporation. This finding has been found for other proteins and peptides such as diphtheria toxin, annexin V and VI, vesicular stomatitis virus (VSV), and colicin E1 that show maximum efficiency of incorporation at low pH [48–53]. Similarly to hCt, these proteins possess His groups that are charged at low pH. Our results lend

196

S. Micelli et al.

support to the notion that the bilayer interface determines the deprotonation of His which acts as an anchor for further protonation of negatively charged Asp [54]. Finally, differential scanning calorimetry studies on flanking His-hydrophobic peptide incorporation in the DPPC lipid at pH 5 have shown a decrease in relative enthalpy, indicative of an interaction between a positively charged peptide and a hydrophilic lipid headgroup [55], as well as a hydrophobic interaction. Moreover, it has also been found that positively charged peptides induce more disturbance on DPPC bilayers than on uncharged ones [56], and that this disturbance can facilitate peptide incorporation. By contrast, at low pH, in membranes made up of DOPG, hCt does not form channels. At this low pH, the positively charged hCt electrostatically interacts with negatively charged lipids, forming an anchor that renders translocation into the hydrophobic medium difficult. Based on the results shown here, we propose that, depending on the membrane surface and on the protonation/ deprotonation of His and Asp residues, through a balance between electrostatic and hydrophobic interaction, hCt more or less easily forms the amphipathic a-helices required for insertion. This provides information on the lipid–peptide interaction involved in many functions of the cell such as protein/peptide translocation and membrane–receptor binding, and evidentiates that a very low variation in electrostatic charge may have widespread effects. An important molecular feature of hCt is its high flexibility along the whole polypeptide chain [56]; this structural feature may account for the different degrees of its interaction with membranes (particularly when high concentrations of hCt are used) where the hydrophobic side of the helix can trigger intermolecular interaction, culminating in fibrillation, as indicated by the lower channel frequency at higher hCt concentrations. In fact, single channels are formed more easily at low hCt concentrations and the channel central conductance is a linear function of the hCt concentration up to 85 nM, while the conductance decreases as the hCt concentration increases. These results, albeit indirectly, can be appropriately correlated with the well-known fibrillating property of hCt [4]. In aqueous solution, Ct presents a random coil structure, and when SDS at critical micelle concentration (CMC) is added, an a-helical structure is promoted [8,17], although in comparison to sCt, hCt acquires the a-helix conformation more slowly [4] and shows a shorter helix (9–16) in SDS [57] than in a trifluoroethanol– water mixture [58]. The increased rate of channel formation using as low as pico–nano molar SDS concentrations (and to a lesser extent lauryl-sarcosine) is presumably due to the induction of a helical conformation in hCt that counteracts the transition from a monomeric to a fibrillar state by facilitating incorporation. In fact, SDS increases both the number and the frequency of hCt channel formation at all hCt concentrations used. Although the exact modality of molecular interaction between peptide and SDS remains to be elucidated, it could be assumed that a bivalent contact takes place between the two molecules, with the cationic group of lysine and hydrophobic residues, similarly to that reported for lysozyme and for PF4(56–70) [59, 60]. In particular, by enwrapping the hydrophobic part of the peptide (specifically the

A New Class of Peptide-Forming Channel

197

aromatic phenylalanines responsible for the self-assembly process) with its long acyl chain, SDS may induce both a-helix formation and the transmembrane position of the peptide that prevents interaction with other helices, thus not allowing the bundle to organize. In fact, due to its propensity to form b-sheet structures, the sequence from D15 to F19 of hCt has been implicated in the fibrillation process. Moreover, it has been found that the minimum length of polypeptide that will undergo fibrillation is four amino acids with a propensity for b-strand conformation [9, 61]. This property has also been found for human transthyretin, the major component of amyloid in senile heart. In fact, by using 2% SDS, Altland and Winter [62] found that dimer dissociation leads monomers to associate in a complex with SDS, thus losing their capacity to dimerize. It has been suggested that the first step in the nucleation process is due to a hydrophobic interaction between hCt monomers in the aggregating phenomenon to form bundles of fibrils [46]. On the other hand, it has been found that at pH 7, the positive charge of lysine and the negative charge of aspartate play a further role in the fibrillation process [6]; by competing with the hCt molecules, SDS may disrupt the process by interacting with positively charged lysine, leaving the peptide negatively charged. In our study, the important aspect of SDS action is the high hCt:SDS (1000:1) molar ratio, at all hCt concentrations used, sufficient to trigger the phenomenon observed. This effect could be tentatively explained by a catalyst action exerted through a favorable electrostatic and hydrophobic interaction that is worthy of further investigation. Thus, the rapid induction of a-helices by SDS could be beneficial for lipid penetration and receptor interaction, as has been found for the ACTH hormone [36]. We could expect that a competition can take place between hCt receptor and bilayer, depending on the receptor concentration and lipid structure of the membrane. It is worth mentioning that many naturally occurring chaperones control protein/peptide folding. For example, Hsp70 chaperone has been shown in a Drosophila melanogaster model to suppress the toxicity of a-synuclein, the main protein implicated in the pathogenesis of Parkinson’s disease in man [63]; glycerol and methylamine induce the random coil to form a b-sheet structure in Ab [63]; cholesterol determines the conformational transition from random coil to a structure designed to incorporate voltage-dependent anion channels into PLMs [44, 64, 65]. It can be speculated that SDS could act as a chaperone, shielding the exposure of the hydrophobic groups of Ct and preventing their interaction with both polar and hydrophobic groups of congener molecules to form a bundle of helices. Although it has not yet been demonstrated that hCt forms channels in the target tissue (though this angle is currently being studied), the action of SDS is of great interest because SDS could facilitate hCt–receptor interaction sites in the membrane and counteract fibril formation, especially when therapeutically high concentrations of Cts are used. It would be useful, apart from the relevant biophysical interest, to explore whether such a simple molecule, or other similar molecules, could have a therapeutic interest besides technological applications, such as transmembrane transport of proteins or other compounds attached to the peptide–SDS complex.

198

S. Micelli et al.

7.3. Tentative Interpretation of Physiological Relevance As proposed in our previous papers [18, 19], at least four inserted hCt molecules are assembled to form a transmembrane channel with a hydrophilic central pore. In other words, the increased rate of hCt incorporation and channel formation could be a valid mechanism for destabilizing the reaction between congener peptide molecules which form the seed of nucleation before fibril formation. The therapeutically marked activity of sCt and eCt may be the result of an ancestor channel activity owing to the need that these fish species have to fine-tune ion regulation during migration. In fact, both sCt and eCt show conspicuous channel activity in vitro. However, hCt has also shown vestiges of channel activity which could be exhumed by environmental conditions such as high potential, nanomolar concentrations of SDS [18, 19] and—as found in our study—by low pH. From the physiological point of view, osteoclasts have shown sensitivity to Cts both in vivo and in vitro [66]. Ct inhibits bone resorption by inhibiting osteoclast activity. Bone resorption is accomplished by osteoclasts, whose membrane ruffles are the initial events in the process that transports acidifying vesicles along the microtubules and inserts them into the plasma membrane [67, 68]. In these contact areas, acidification of the environment by Hþ-ATPase initiates bone demineralization [69, 70]; consequently, high levels of extracellular Ca2þ are sensed by a receptor that will constrain the detachment of the osteoclasts from the bone surface [71–74]. Preliminary results on PLMs show Ca2þ permeability through hCt channels at low pH. One tempting possibility could be that when hCt senses low environmental pH, it undergoes a structural conformation promoting its incorporation and assembly into POPC-rich plasma membranes, to form channels that are sensitive and permeable to Ca2þ; this aspect is under investigation. These channels could be considered parallel pathways, complementing the known Ca2þ-sensing receptor [75, 76], and potentiating osteoclast detachment. Channel formation by Ct may thus be considered an additional mechanism to push Ca2þ across the membrane into the cytosol for signal transduction. Besides this aspect, pharmacological approaches in gene therapy are stimulating basic study into the uptake of peptides, oligonucleotides, and DNA through plasma membranes, and physiological peptides could be promising tools.

REFERENCES [1] S. Wallach, G. Rousseau, L. Martin, M. Azria, Effects of calcitonin on animal and in vitro models of skeletal metabolism, Bone 25 (1999) 509–516. [2] A. Cudd, T. Arvinte, R.E. Gaines Das, C. Chinni, I. McIntyre, Enhanced potency of human calcitonin when fibrillation is avoided, J. Pharm. Sci. 84 (1995) 717–719. [3] P. Sieber, B. Riniker, M. Brugger, B. Kamber, W. Rittel, Human calcitonin. VI. Synthesis of calcitonin, M. Helv. Chim. Acta 53 (1970) 2135–2150. [4] T. Arvinte, A. Cudd, A.F. Drake, The structure and mechanism of formation of human calcitonin fibrils, J. Biol. Chem. 268 (1993) 6415–6422. [5] H.H. Bauer, M. Muller, J. Goette, H.P. Merkle, U.P. Fringeli, Interfacial adsorption and aggregation associated changes in secondary structure of human calcitonin monitored by ATRFTIR spectroscopy, Biochemistry 33 (1994) 12276–12282.

A New Class of Peptide-Forming Channel

199

[6] M. Kamihira, A. Naito, S. Tuzi, A. Nosaka, H. Saito, Conformational transitions and fibrillation mechanism of human calcitonin as studied by high-resolution solid-state 13C NMR, Protein Sci. 9 (2000) 867–877. [7] D. Meleleo, S. Micelli, V. Capozzi, G. Perna, E. Mezzenga, V. Picciarelli, E. Gallucci, From b-sheet to a pH-induced a-helical structure in human calcitonin: An integrated investigation by Circular Dicroism, Raman Spectroscopy and channel formation in planar lipid membranes, FEBS J. 272(s1) (2005) C4–029P. [8] A. Motta, G. Andreotti, P. Amodeo, G. Strazzullo, M.A. Castiglione Morelli, Solution structure of human calcitonin in membrane-mimetic environment: The role of the amphipathic helix, Proteins 32 (1998) 314–323. [9] M. Reches, Y. Porat, E. Gazit, Amyloid fibril formation by pentapeptide and tetrapeptide fragments of human calcitonin, J. Biol. Chem. 38 (2002) 35475–35480. [10] F. Levy, R. Muff, S. Dotti-Sigrist, M.A. Dambacher, J.A. Fischer, Formation of neutralizing antibodies during intranasal synthetic salmon calcitonin treatment of Paget’s disease, J. Clin. Endocrinol. Metab. 67 (1988) 541–545. [11] R. Muff, M.A. Dambacher, J.A. Fischer, Formation of neutralizing antibodies during intranasal synthetic salmon calcitonin treatment of postmenopausal osteoporosis, Osteoporos. Int. 1 (1991) 72–75. [12] A. Grauer, R. Ziegler, F. Raue, Clinical significance of antibodies against calcitonin, Exp. Clin. Endocrinol. Diabetes 103 (1995) 345–351. [13] T.J. Martin, Calcitonin, an update, Bone 24 (1999) 63S–65S. [14] A. Cranney, V. Welch, J.D. Adachi, J. Homik, B. Shea, M.E. Suarez-Almazor, P. Tugwell, G. Wells, Calcitonin for the treatment and prevention of corticosteroid-induced osteoporosis, Cochrane Database Sys. Rev. (2) (2000) CD001983. [15] G.V. Segre, S.R. Goldring, Receptors for secretin, calcitonin, parathyroid hormone (PTH/PTH related peptide), vasoactive intestinal peptide, glucagon-like peptide 1, growth hormonereleasing hormone, and glucagon belong to a newly discovered G-protein-linked receptor family, Trends Endocrinol. Metab. 4 (1993) 309–314. [16] D.L. Rymer, T.A. Good, The role of G protein activation in the toxicity of amyloidogenic Abeta-(1–40), Abeta-(25–35), and bovine calcitonin, J. Biol. Chem. 26 (2001) 2523–2530. [17] P.M. Sexton, D.M. Findlay, T.J. Martin, Calcitonin, Curr. Med. Chem. 6 (1999) 1067–1093. [18] V. Stipani, E. Gallucci, S. Micelli, V. Picciarelli, R. Benz, Channel formation by salmon and human calcitonin in black lipid membranes, Biophys. J. 8 (2001) 3332–3338. [19] S. Micelli, D. Meleleo, V. Picciarelli, M.G. Stoico, E. Gallucci, Effect of nanomolar concentrations of sodium dodecyl sulfate, a catalytic inductor of alpha-helices, on human calcitonin incorporation and channel formation in planar lipid membranes, Biophys. J. 87 (2004) 1065–1075. [20] D. Meleleo, S. Micelli, K. Toma, K. Haneda, E. Gallucci, Effect of eel calcitonin glycosylation on incorporation and channel formation in planar phospholipid membranes, Peptides 27 (2006) 805–811. [21] S. Micelli, D. Meleleo, V. Picciarelli, E. Gallucci, Effect of pH-variation on insertion and ion channel formation of human calcitonin into planar lipid bilayers, Front. Biosci. 11 (2006) 2035–2044. [22] P. Muller, D.O. Rudin, H.T. Tien, W.C. Wescott, Reconstitution of cell membrane structure in vitro and its transformation into an excitable system, Nature 194 (1962) 979–980. [23] H.T. Tien (Ed.), Bilayer Lipid Membranes: Theory and Practice, Marcel Dekker, Inc., New York (1974) Chapters 4 and 5. [24] H.T. Tien, J.D. Mountz, A.N. Martinosi (Ed.), Protein-lipid interaction in bilayer lipid membranes (BLM), in: The Enzyme of Biological Membranes. NY Plenum, I, (1977) pp. 139–170. [25] H.T. Tien, A.L. Ottova, Membrane Biophysics: As Viewed from Experimental Bilayer Lipid Membranes (Planar Lipid Bilayers and Spherical Liposomes), Elsevier, Amsterdam (2000) p. 648. [26] H.T. Tien, A.L. Ottova, The lipid bilayer concept and its experimental realization: From soap bubbles, kitchen sink, to bilayer lipid membranes, J. Membr. Sci. 189 (2001) 83–117. [27] M. Azria, The Calcitonins: Physiology and Pharmacology, Basel, Karger, 1989. [28] A. Kapurniotu, J.W. Taylor, Structural and conformational requirements for human calcitonin activity: Design, synthesis, and study of lactam-bridged analogues, J. Med. Chem. 38 (1995) 836–847.

200

S. Micelli et al.

[29] T. Mizuka, I. Hideki, T. Kazunori, Three-dimensional structure and biological activity of glycosylated calcitonin, Trends Glycosci. Glycotechnol. 13 (2001) 373–383. [30] R.C. Keller, J.A. Killian, B. de Kruijff, Anionic phospholipids are essential for a-helix formation of the signal peptide of prePhoE upon interaction with phospholipid vesicles, Biochemistry 31 (1992) 1672–1677. [31] R.M. Epand, R.F. Epand, R.C. Orlowski, R.J. Schlueter, L.T. Boni, S.W. Hui, Amphipathic helix and its relationship to the interaction of calcitonin with phospholipids, Biochemistry 22 (1983) 5074–5084. [32] D.M. Engelman, T.A. Steitz, The spontaneous insertion of proteins into and across membranes: The helical hairpin hypothesis, Cell 23 (1981) 411–422. [33] R.E. Jacobs, S.H. White, The nature of the hydrophobic binding of small peptides at the bilayer interface: Implications for the insertion of transbilayer helices, Biochemistry 28 (1989) 3421–3437. [34] M. Milik, J. Skolmick, Insertion of peptide chains into lipid membranes: An off-lattice Monte Carlo dynamics model, Proteins 15 (1993) 10–25. [35] S.H. White, W.C. Wimley, Membrane protein folding and stability: Physical principles, Annu. Rev. Biophys. Biomol. Struct. 28 (1999) 319–365. [36] D.F. Sargent, J.W. Bean, R. Schwyzer, Conformation and orientation of regulatory peptides on lipid membranes. Key to the molecular mechanism of receptor selection, Biophys. Chem. 31 (1988) 183–193. [37] W. Jordi, B. De Kruijff, D. Marsh, Specificity of the interaction of amino- and carboxy-terminal fragments of the mitochondrial precursor protein apocytochrome c with negatively charged phospholipids. A spin-label electron spin resonance study, Biochemistry 28 (1989) 8998–9005. [38] E. Terzi, G. Holzemann, J. Seelig, Interaction of Alzheimer beta-amyloid peptide(1–40) with lipid membranes, Biochemistry 36 (1997) 14845–14852. [39] J.P. Bradshaw, Phosphatidylglycerol promotes bilayer insertion of salmon calcitonin, Biophys. J. 72 (1997) 2180–2186. [40] J. McLaurin, D.S. Yang, C.M. Yip, P.E. Fraser, Review: Modulating factors in amyloid-beta fibril formation, J. Struct. Biol. 130 (2000) 259–270. [41] C.M. Yip, J. McLaurin, Amyloid-beta peptide assembly: A critical step in fibrillogenesis and membrane disruption, Biophys. J. 80 (2001) 1359–1371. [42] M. Martinez-Senac, S. Gorbalan-Garcia, J.C. Gomez-Fernandez, The structure of the C-terminal domain of the pro-apoptotic protein Bak and its interaction with model membranes, Biophys. J. 82 (2002) 233–243. [43] E. Gallucci, D. Meleleo, S. Micelli, V. Picciarelli, Magainin 2 channel formation in planar lipid membranes: The role of lipid polar groups and ergosterol, Eur. Biophys. J. 32 (2003) 22–32. [44] S. Micelli, E. Gallucci, V. Picciarelli, Studies of mitochondrial porin incorporation parameters and voltage-gated mechanism with different black lipid membranes, Bioelectrochemistry 52 (2000) 63–75. [45] S. Micelli, D. Meleleo, V. Picciarelli, E. Gallucci, Effect of sterols on beta-amyloid peptide (AbetaP 1–40) channel formation and their properties in planar lipid membranes, Biophys. J. 86 (2004) 2231–2237. [46] K. Kanaori, A.Y. Nasaka, Study of human calcitonin fibrillation by proton nuclear magnetic resonance spectroscopy, Biochemistry 34 (1995) 12138–12143. [47] M. Kamihira, Y. Oshiro, S. Tuzi, A.Y. Nosaka, H. Saito, A. Naito, Effect of electrostatic interaction on fibril formation of human calcitonin as studied by high resolution solid state 13C NMR, J. Biol. Chem. 278 (2003) 2859–2865. [48] K.J. Oh, L. Senzel, R.J. Collier, A. Finkelstein, Translocation of the catalytic domain of diphtheria toxin across planar phospholipid bilayers by its own T domain, Proc. Natl. Acad. Sci. USA 96 (1999) 8467–8470. [49] G. Kohler, U. Hering, O. Zschornig, K. Arnold, Annexin V interaction with phosphatidylserinecontaining vesicles at low and neutral pH, Biochemistry 36 (1997) 8189–8194. [50] M. Golczak, A. Kicinska, J. Bandorowicz-Pikula, R. Buchet, A. Szewczyk, S. Pikula, Acidic pHinduced folding of annexin VI is a prerequisite for its insertion into lipid bilayers and formation of ion channels by the protein molecules, FASEB J. 15 (2001b) 1083–1085.

A New Class of Peptide-Forming Channel

201

[51] M. Golczak, A. Kirilenko, J. Bandorowicz-Pikula, S. Pikula, Conformational states of annexin VI in solution induced by acidic pH, FEBS Lett. 496 (2001a) 49–54. [52] F.A. Carneiro, F. Stauffer, C.S. Lima, M.A. Juliano, L. Juliano, A.T. Da Poian, Membrane fusion induced by vesicular stomatitis virus depends on histidine protonation, J. Biol. Chem. 278 (2003) 13789–13794. [53] S.D. Zakharov, M. Lindeberg, Y. Griko, Z. Salamon, G. Tollin, F.G. Prendergast, W.A. Cramer, Membrane-bound state of the colicin E1 channel domain as an extended two-dimensional helical array, Proc. Natl. Acad. Sci. USA 95 (1998) 4282–4287. [54] A.S. Ladokhin, S.H. White, Interfacial folding and membrane insertion of a designed helical peptide, Biochemistry 43 (2004) 5782–5791. [55] H.A. Rinia, J.-W.P. Boots, D.T.S Rijkers, R.A. Kik, M.E. Snel, R.A. Demel, J.A. Killian, J. P. J. M. van der Eerden, B. de Kruijff, Domain formation in phosphatidylcholine bilayers containing transmembrane peptides: Specific effects of flanking residues, Biochemistry 41 (2002) 2814–2824. [56] P. Amodeo, A. Motta, G. Strazzullo, M.A. Castiglione Morelli, Conformational flexibility in calcitonin: The dynamic properties of human and salmon calcitonin in solution, J. Biomol. NMR 13 (1999) 161–174. [57] A. Motta, A. Pastore, N.A. Goud, M.A. Castiglione Morelli, Solution conformation of salmon calcitonin in sodium dodecyl sulfate micelles as determined by two-dimensional NMR and distance geometry calculations, Biochemistry 30 (1991) 10444–10450. [58] M. Doi, Y. Kobayashi, Y. Kyogoku, M. Takimoto, K. Goda, Structure study of human calcitonin, in: J.E. Rivier, R. Marshall (Eds.), Peptides: Chemistry, Structure and Biology, Eleventh American Peptide Symposium, ESCOM, Leiden, The Netherlands (1990) pp. 165–167. [59] A. Yonath, A. Podjarny, B. Honig, A. Sielecki, W. Traub, Crystallographic studies of protein denaturation and renaturation. 2. Sodium dodecyl sulfate induced structural changes in triclinic lysozyme, Biochemistry 16 (1977) 1418–1424. [60] R. Montserret, M. McLeich, A. Bockmann, C. Geourjon, F. Penin, Involvement of electrostatic interactions in the mechanism of peptide folding induced by sodium dodecyl sulfate binding, Biochemistry 39 (2000) 8362–8373. [61] L. Tjernberg, W. Hosia, N. Bark, J. Thyberg, J. Johansson, Charge attraction and beta propensity are necessary for amyloid fibril formation from tetrapeptides, J. Biol. Chem. 277 (2002) 43243–43246. [62] K. Altland, P. Winter, Polyacrylamide gel electrophoresis followed by sodium dodecyl sulfate gradient polyacrylamide gel electrophoresis for the study of the dimer to monomer transition of human transthyretin, Electrophoresis 24 (2003) 2265–2271. [63] D.S. Yang, C.M. Yip, T.H.J. Huang, A. Chakrabartty, P.E. Fraser, Manipulating the amyloidbeta aggregation pathway with chemical chaperones, J. Biol. Chem. 274 (1999) 32970–32974. [64] B. Popp, A. Schmid, R. Benz, Role of sterols in the functional reconstitution of water-soluble mitochondrial porins from different organisms, Biochemistry 34 (1995) 3352–3361. [65] E. Gallucci, S. Micelli, G. Monticelli, Pore formation in lipid bilayer membranes made of phosphatidylinositol and oxidized cholesterol followed by means of alternating current, Biophys. J. 71 (1996) 824–831. [66] M. Zaidi, K. Fuller, P.J. Bevis, R.E. GainesDas, T.J. Chambers, I. MacIntyre, Calcitonin generelated peptide inhibits osteoclastic bone resorption: A comparative study, Calcif. Tissue Int. 40 (1987) 149–154. [67] Y.Y. Kong, U. Feige, I. Sarosi, B. Bolon, A. Tafuri, S. Morony, C. Capparelli, J. Li, R. Elliott, S. McCabe, T. Wong, G. Campagnuolo, et al., Activated T cells regulate bone loss and joint destruction in adjuvant arthritis through osteoprotegerin ligand, Nature 402 (1999) 304–309. [68] S. Kotake, N. Udagawa, N. Takahashi, K. Matsuzaki, K. Itoh, S. Ishiyama, S. Saito, K. Inoue, N. Kamatani, M.T. Gillespie, T.J. Martin, T. Suda, IL-17 in synovial fluids from patients with rheumatoid arthritis is a potent stimulator of osteoclastogenesis, J. Clin. Invest. 103 (1999) 1345–1352. [69] R. Baron, L. Neff, D. Louvard, P.J. Courtoy, Cell-mediated extracellular acidification and bone resorption: Evidence for a low pH in resorbing lacunae and localization of a 100-kD lysosomal membrane protein at the osteoclast ruffled border, J. Cell Biol. 101 (1985) 2210–2222. [70] H.K. Vaananen, E.K. Karhukorpi, K. Sundquist, B. Wallmark, I. Roininen, T. Hentunen, J. Tuukkanen, P. Lakkakorpi, Evidence for the presence of a proton pump of the vacuolar H(þ)-ATPase type in the ruffled borders of osteoclasts, J. Cell Biol. 111 (1990) 1305–1311.

202

S. Micelli et al.

[71] M. Zaidi, H.K. Datta, A. Patchell, B. Moonga, I. MacIntyre, Calcium-activated intracellular calcium elevation: A novel mechanism of osteoclast regulation, Biochem. Biophys. Res. Commun. 163 (1989) 1461–1465. [72] A. Malgaroli, J. Meldolesi, A.Z. Zallone, A. Teti, Control of cytosolic free calcium in rat and chicken osteoclasts. The role of extracellular calcium and calcitonin, J. Biol. Chem. 264 (1989) 14342–14347. [73] B.S. Moonga, D.W. Moss, A. Patchell, M. Zaidi, Intracellular regulation of enzyme secretion from rat osteoclasts and evidence for a functional role in bone resorption, J. Physiol. 429 (1990) 29–45. [74] M. Zaidi, A.S. Alam, C.L. Huang, M. Pazianas, C.M. Bax, B.E. Bax, B.S. Moonga, P.J. Bevis, V.S. Shankar, Extracellular Ca2þ sensing by the osteoclast, (1993) 271–277. [75] C.M.R. Bax, V.S. Shankar, B.S. Moonga, C.L.-H. Huang, M. Zaidi, Is the osteoclast calcium ‘‘receptor’’ a receptor-operated calcium channel? Biochem. Biophys. Res. Commun. 183 (1992) 619–625. [76] O.A. Adelabanjo, V.S. Shankar, M. Pazianas, A. Zaidi, B. Simon, C.L.-H. Huang, M. Zaidi, Modulation of the osteoclast Ca2þ receptor by extracellular protons: Possible linkage between Ca2þ sensing and extracellular acidification, Biochem. Biophys. Res. Commun. 199 (1994) 742–747.

C H A P T E R

E I G H T

Lipid-based Strategies in Inorganic Nano-materials and Biomineralization Study Xiaohua Liu,1,2 Haixin Bai,1,2 Lixue Zhang,1 and Erkang Wang1,* Contents 1. Introduction 2. Synthesis of Various Nanocrystals Based on Lipid Strategies 2.1. Nanocrystals Protected by Lipid Membrane 2.2. Other Synthesis Methods of Nanocrystals Based on Lipid Strategies 3. Nanocrystal–Lipid Hybrid Materials 3.1. The Encapsulation of Various Nanocrystals in Lipid Membrane 3.2. Assembly of Lipid Membrane on Nanocrystals 4. The Application of Lipid-based Nanocrystals 4.1. Interaction of Nanocrystals with Lipid Membrane 4.2. Examples of the Applications of Lipid-based Nanocrystals 5. The Application of Lipid in Biomineralization 6. Conclusion and Perspectives Acknowledgments References

204 204 204 209 210 210 211 212 212 213 213 215 215 216

Abstract This chapter will mainly discuss the application of lipid in inorganic nano-materials and biomineralization studies, including the synthesis of various inorganic nanocrystals based on lipid strategies, the encapsulation of various nanocrystals in lipid membrane, assembly of lipid membrane on nanocrystals, application of lipid-based nanocrystals, the effects of lipid membrane in biomineralization, and so on. There are several excellent features such as good stability, low toxicity, easy functionalization, and some potential applications when inorganic materials are composite with lipid. We will combine the literatures published with our recent researches in this field to discuss this topic.

* Corresponding author. Tel.:/Fax: þ86 431 85689711; E-mail address: [email protected] (E. Wang). 1

2

State Key Laboratory of Electroanalytical Chemistry, Changchun Institute of Applied Chemistry, Chinese Academy of Sciences, Changchun, Jilin 130022, P.R. China College of Chemistry and Chemical Engineering, Henan Institute of Science and Technology, Xinxiang 453003, P.R. China

Advances in Planar Lipid Bilayers and Liposomes, Volume 7 ISSN 1554-4516, DOI: 10.1016/S1554-4516(08)00008-2

#

2008 Elsevier Inc. All rights reserved.

203

X. Liu et al.

204

1. Introduction In recent years, there is a considerable interest in synthesizing inorganic materials composed of nanocrystals due to their unique optical, electronic, magnetic, thermal, mechanical, and other properties being different from their bulk counterparts and potential applications such as sensors, catalysis, paints, optoelectronics, and so on [1, 2]. Among the methods of synthesizing inorganic nano-materials, the solution chemistry approach has been widely developed. Generally, surfactant and polymers were often used as protective agents to prevent the nano-materials agglomeration in the solution. Lipid, with an amphipathic character and having a nonpolar end and a polar end, can possess more than one kind of organized structures after forming a hydration shell around the headgroup. The particular form predominately depends on such parameters as the lipid concentration, temperature, pressure, ionic strength, and pH [3]. The major organized forms of lipid in aqueous solution consist of micelle, hexagonal phase, cubic phase, and bilayer phase [4, 5]. Thus, lipid has also been considered as an ideal and effective protective agent or template for the synthesis of nanocrystals. At the same time, the nano-materials composite with lipid are provided with good stability, low toxicity, easy functionalization and many potential applications as bio- and chemical materials. Biomineralization, defining as the mineral formation process occurring in organisms, can generate materials of very highly controlled size, texture, composition, and structure [6]. Lipid, especially phospholipid, often in the form of bilayer vesicles has been commonly involved in natural biomineralization processes [7]. Therefore, lipid can be used as an ideal model to study biomimetic mineralization and provides not only a confined, organized microenvironment but also organic matrix for biomimetic mineralization [6, 8–11]. During crystal growth, the complex crystal shapes and textures can be produced through altering the shape of the lipid matrix [6]. In this chapter, we will introduce the lipid-based synthesis of various nanocrystals, including noble metal (Au, Ag, Pt) nanocrystals, magnetic nanocrystals, quantum dots, and so on. The other topics related to lipid and nanocrystals will also be explained, such as encapsulation of nanocrystals in lipid bilayer, assembly of lipid membrane on nanocrystals, applications of lipid-based nanocrystals, and so on. In addition, this chapter will also deal with some aspects about recent applications of lipid in biomineralization.

2. Synthesis of Various Nanocrystals Based on Lipid Strategies 2.1. Nanocrystals Protected by Lipid Membrane The nanocrystals coated by lipid membrane can be divided into two catalogs, monolayer-protected nanocrystals, which cannot dissolve in the aqueous solution and bilayer-protected nanocrystals with good solubility in aqueous solution.

Lipid-based Strategies in Inorganic Nano-materials and Biomineralization Study

205

Since Brust et al. [12] synthesized the very stable alkane thiols protected Au nanoparticles in a liquid–liquid phase system, materials prepared in this way are particularly attractive owing to their stability, tunable solubility, and relative ease of characterization [13]. Many papers about monolayer-protected clusters were published [14–16]. Because of the stability similar to thiols-protected nanoparticles, lipid monolayer-coated nanoparticles also draw attentions. Generally, they are prepared in organic solvent, such as toluene, chloroform, and so on. In our group, we synthesized a kind of gold nanoparticle protected by a synthetic lipid (didodecyldimethylammonium bromide, DDAB) [17]. Our synthesis method is similar to Brust et al.’s [12]: 0.5 ml 10 mM HAuCl4 solution was added into a 1-ml chloroform solution containing 2 mg DDAB. After vigorous stirring, HAuCl4 was transferred to organic phase by the extraction of DDAB. The water layer changed its color from yellow to clarity, while the organic layer turned to orange. Freshly prepared aqueous solution of sodium borohydride (NaBH4, 0.2 ml, 0.4 M) was slowly added to the solution stirred vigorously. The solution turned to amaranth soon. After further stirring for 2 h, the organic layer was separated. Among the processes as mentioned above, the DDAB acted as both phase-transfer reagent and nanoparticle-protective reagent. As-prepared Au nanoparticles can keep stability for at least 8 months. Furthermore, they have a distinctive feature, namely, they can facilitate the direct electron transfer (DET) from protein to electrode because of its good biocompatibility. Figure 1 shows that the typical DDAB monolayer-protected nanoparticles were 6.42 nm in diameter with a standard deviation of 1.1 nm, which is roughly consistent with the result reported by Lin and Sorensen [18]. They exhibit welldefined roundness and they uniformly distribute. After glassy carbon electrode (GCE) was modified with DDAB monolayer-protected Au nanoparticles and hemoglobin, the modified GCE was then transferred to deoxygenated acetate buffer solution (pH ¼ 5.5), and measured by cyclic voltammetry. The cyclic voltammogram of the electrode done as mentioned above is shown in Fig. 2C. A pair of distinct redox peaks emerges. However, when only particles (Fig. 2A) or hemoglobin (Fig. 2B) was dropped onto the electrode, redox peaks do not exist. It means that the lipid-protected Au nanoparticles play an important role in DET from hemoglobin to electrode. It was the first time to realize the direct electrochemistry of hemoglobin with the help of lipid-protected nanoparticles. Some other literatures also reported the synthesis of lipid monolayer-coated nanoparticles. Li and his coworkers demonstrated that phospholipids (l-a-dipalmitoylphosphatidylcholine, DPPC)-capped gold nanoparticles could be synthesized with one step by reduction of HAuCl4 in aqueous solution using sodium citrate and simultaneously accompanying with the phase transition across the water/toluene interface into the organic phase [19]. According to Bahadur, phosphatidylcholine (PC)-stabilized magnetite particles were prepared based on the strong affinity between magnetite and phosphate groups and the effects of PC on the growth of magnetite superparamagnetic particles have been investigated. These coated particles gave stable suspension in chloroform and were used for magnetoliposome preparation [20]. During the past few years, bilayer-protected nanostructures have attracted increasing research efforts due to their water solubility, biocompatibility, and potential applications in many fields. After Sastry and coworkers demonstrated that

X. Liu et al.

206

20

15

10

5

0

4

5

6 7 Diameter (nm)

8

9

Figure 1 Transmission electron microscopic (TEM) image and histogram of lipid-protected gold nanoparticles. The scale bar is 10 nm.

nanoscale curvature enables the formation of interdigitated bilayers of fatty acid on a colloidal silver particle surface [21], a wealth of papers appeared on the successful preparation of water-soluble, bilayer-protected nanostructures, including noble metal nanoparticles [22–28], metal nanorods [29, 30], magnetic nanoparticles [31], and semiconductor quantum dots [32]. For example, Urban described a simple twostep approach of modification of 1 nm diameter Au nanoparticles using an aqueous solution of (1,2-dipalmitoyl-sn-glycero-3-phosphothio-ethanol) phospholipids [33].

207

Ι / μΑ

Lipid-based Strategies in Inorganic Nano-materials and Biomineralization Study

7 6 5 4 3 2 1 0 −1 −2 −3 −4 −5 −6 −7 −8 −600

a b c

−400

−200 0 E/mV vs Ag/AgCl

200

400

Figure 2 Cyclic voltammogram of the glassy carbon electrode (GCE) coated with (A) 10 ml lipid-protected nanoparticles, (B) 10 ml hemoglobin, (C) 10 ml lipid-protected gold nanoparticles, and 10 ml hemoglobin, in acetate buffer solution (pH ¼ 5.5). Scan rate is 100 mV/s.

In these previous studies, the formation of bilayer-protected nanoparticles usually involved multiple steps. In our laboratory, we developed a facile one-step method for the preparation of very stable, DDAB lipid bilayer-protected Au nanoparticles in an aqueous medium by in situ chemical reduction of HAuCl4 with NaBH4 in the presence of DDAB [34]. The DDAB lipid bilayer-protected gold nanoparticles were prepared according to the following procedure: First, 375 ml of 0.048 M HAuCl4 aqueous solution and 60 ml of 1.4 mM DDAB aqueous solution were mixed together under vigorous stirring. Then, 200 ml of freshly prepared 0.4 M NaBH4 aqueous solution was added to the former stirred solution drop by drop. The as-formed wine-colored solution was stored at room temperature. As-prepared nanoparticles were characterized by UV–vis spectra, transmission electron microscopy, dynamic light scattering analysis, and X-ray photoelectron spectroscopy. All these data supported the formation of Au nanoparticles. Figure 3 shows a transmission electron microscopic (TEM) image of gold nanoparticles. It is clearly seen that most of the gold nanoparticles are spherical in shape and well separated from each other. Fourier transform infrared spectroscopy (FTIR) and differential thermal analysis/thermogravimetric analysis data revealed that DDAB existed in a bilayer structure formed on the particle surface, resulting in a positively charged particle surface. We also demonstrated that the as-prepared Au nanoparticles can be fabricated into (PSS/Au-nanoparticles)n multilayers on a cationic poly(ethylenimine) (PEI)-coated indium tin oxide (ITO) substrate via the layer-by-layer (LBL) selfassembly technique [35], which should be due to electrostatic interactions between such Au nanoparticles and anionic polyelectrolyte poly(sodium 4-styrenesulfonate) (PSS). As shown in Fig. 4, we can see that the surface morphology of the ITO/PEI/ (PSS/Au-nanoparticles)4 film is dominantly comprised of individual, uniformly distributed nanoparticles and some aggregated clusters of Au nanoparticles. The Au nanoparticles solution we prepared can be stored for at least 5 months without

X. Liu et al.

208

Figure 3 Typical transmission electron microscopic (TEM) image of the DDAB-protected gold nanoparticles. Reprinted with permission from ref. [34]. Copyright (2006) American Chemical Society. Flatten 2.00

50.0 nm

25.0 nm

1.00

0.0 nm

Digital Instruments NanoScope Scan size 2.000 μm Scan rate 1.507 Hz Number of samples 256

0

1.00

0 2.00 μm

Figure 4 Atomic force microscopic (AFM) image of ITO/PEI/(PSS/AuNPs)4 multilayers. Reprinted with permission from ref. [34]. Copyright (2006) American Chemical Society.

observable aggregation under air conditions, which indicated as-prepared Au nanoparticles have long-term stability. The as-prepared DDAB-protected Au nanoparticles can be used as pseudostationary phase in capillary electrophoresis for the analysis of acidic and basic proteins, leading to greater separation efficiency and high reproducibility [36]. We can also obtain dimyristoylphosphatidylglycerol (DMPG)-protected Au nanoparticles using the same method. Hybrid bilayer membrane, which usually consists of a lipid monolayer and a hydrophobic thiol layer on planar substrate, has been widely investigated. In the

Lipid-based Strategies in Inorganic Nano-materials and Biomineralization Study

209

same way, hybrid bilayer formed at the surface of nanoparticles is another way to prepare nanocrystals in aqueous media. Fan et al. incorporated monosized, hydrophobic gold nanocrystals into the hydrophobic interiors of surfactant micelles, resulting in a hybrid bilayer shell on gold nanocrystals [28]. This method has been extended to the synthesis of water-soluble, lipid-coated nanoparticles. After nanoparticles functionalized with hydrophobic reagent, lipid monolayer can selfassemble on the hydrophobic nanoparticles, increasing the solubility of nanoparticles in aqueous solution. Fan also synthesized water-soluble and biocompatible fluorescent quantum dot micelles by encapsulation of monodisperse, hydrophobic quantum dots within surfactant/lipid micelles [32]. Vogel and his coworkers prepared water-soluble semiconducting nanocrystals through assembling single lipid monolayer on nanocrystals. First, trioctylphosphine oxide stabilized CdSe nanocrystals were prepared and then cosolubilized with lipids in organic solvent and transferred in one-step to an aqueous solution. Thus, the nanocrystals coated with a monolayer of lipid and their surface properties can be tuned by varying the polar head groups of the lipid [37]. In the same way, the addition of functional phospholipid to the hydrophobic iron oxide nanoparticles can not only make nanoparticles transform into hydrophilic ones but also functionalize conveniently with biotin, –COOH, –SH, and –NH2, which facilitating the nanoparticles link to biomolecules for biomedical applications [38].

2.2. Other Synthesis Methods of Nanocrystals Based on Lipid Strategies Lipid can have other effects in the synthesis of nanocrystals besides it acts as protective agent. On the one side, due to liposome can encapsulate various reactants and serve as a selective barrier to the input and output of reacting species. Thus, it can become a ‘‘reaction controller’’ or ‘‘micro-reactor’’ for the formation of nanocrystals. Schelly and his coworkers developed electroporation method of vesicles to prepared ultrasmall, uncapped AgBr [39], CdS [40], PbS [41], and gold nanoclusters [42]. Electric field was used to induce transient pore in the bilayer membrane of synthetic large unilamellar vesicles, which had encapsulated one reactant inside. During opening of the pores, a fraction of the vesicle’s entrapped content was ejected into the bulk solution, contenting another reactant. Therein, the nanoclusters were produced through the reaction of two reactants. In this method, no stabilizing ligand was added, which allowed direct observation of slow cluster growth in distinct steps in the molecular size regime. Monbouquette and his coworkers demonstrated that detergent dialysis PC vesicles were impermeable to Cd2þ and nearly uniform in size. Therein, PC vesicles could serve as ‘‘nanoreactors’’ for the growth of crystalline size-monodisperse nanocrystals of a predetermined size [43–45]. Nanoscale Au particles were synthesized by the reduction of gold(III) chloride involving phospholipid liposome with bound Pd catalysts using an electroless metallization procedure and the effect of vesicle membrane structure and charge on the synthesis of gold nanoparticles was investigated [46, 47]. On the other side, lipid can form different organized structures, therefore, it can be considered as an effective template for the synthesis of nanocrystals. Some

X. Liu et al.

210

literatures have contributed on this aspect. Price et al. exploited spiral seam features of self-assembled diacetylenic phospholipid microtubules as templates to prepare nanoscale metal coils using an electroless metallization approach. They declared that asprepared metal spiral structures are potentially useful as springs or inductors for ‘‘microelectromechanical systems’’ devices and interlocking reinforcements for composite materials [48]. Shelnut et al. reported a new method for making hollow metallic nanospheres using templating liposomes containing photocatalyst molecules. They obtained novel porous platinum nanocages with 2 nm shell thickness and diameters up to 200 nm [49], which may be suitable for many applications, such as catalysts, electrocatalysts, and so on. In addition, they declared that this method could be applied to prepare other metals and alloys nanocages. New spherical nanostructures of titania could be synthesized through formation of liposome–TiO2 nanocomposites by using egg lecithin lipid as a template [50]. Gold nanoparticles could assemble in situ within the patterned lipid matrix, which showed immense potential for extension to assemblies of nanoparticles in more intricate patterns [51]. In situ formation of silver nanoparticles on the surface of the DNA templating molecules in a closed-packed structure within the lipid matrix was also reported [52]. Self-assembled DNA–lipid (DOTAP) membrane complexes could be considered as templates for the growth of CdS nanorod. DNA is highly anionic and condensed the Cd2þ ions, while the cationic lipid membrane modulates the concentration of condensed Cd2þ ions to control the final CdS nanorod dimensions [53]. Gold nanoplates were prepared by the photoreduction of NaAuCl4 in the presence of DMPG under irradiation of highpressure mercury. When the DMPG concentration is appropriate, hexagonal nanoplates can be formed [54]. Bicontinuous cubic phase of lipid (glycerol monooleate, sodium dioctyl sulfosuccinate) also can provide a viable matrix for the synthesis of palladium and PbS nanocrystal [55, 56]. Very recently, a seed-growth method was applied to synthesize the gold nanoparticles with DMPG and hexamethylene-1,6-bis(dodecyldimethylammonium bromide) (12–6–12) as capping agents. With the addition of CuSO4 with high concentration, small Cu nanoparticle appeared which arranged themselves around large Au nanoparticle in a typical pseudo-core-shell type arrangement, which was achieved by the fusion of lipid bilayers of lipid-capped Au and Cu nanoparticle [57].

3. Nanocrystal–Lipid Hybrid Materials 3.1. The Encapsulation of Various Nanocrystals in Lipid Membrane Nanocrystal-based organic–inorganic functional hybrid materials yielding novel, exceptional properties have attracted considerable attentions because of their potential applications in nanobiotechnology. Lipid, especially phospholipid, can provide a biocompatible environment for biomolecules. Therefore, nanocrystal–lipid hybrid materials have been explored recently. The encapsulation of various nanocrystals in lipid membrane and the assembly of lipid in nanocrystals result in the formation of hybrid nano-bio-materials, which afford some special properties and functions, such as good biocompatible, unique recognition, high catalysis, and so on.

Lipid-based Strategies in Inorganic Nano-materials and Biomineralization Study

211

Hydrophobic quantum dots can be readily incorporated into the bilayer membrane of lipid vesicles. Fluorescent resonance energy transfer (FRET) suggests that most of the quantum dots are located deep within the lipid vesicles [58]. Such lipid/ quantum dots hybrid vesicles are capable of fusing with live cells, thereby staining a cell’s plasma membrane selectively with fluorescent quantum dots and transferring the vesicle’s cargo into the cell [59]. Moreover, CdSe quantum dots within liposome basically retain optical properties of free nanocrystals, which can be used as bright fluorescent labels in biological applications [60]. The encapsulation of quantum dots in liposome and the average number of nanoparticles inside each liposome can be measured by fluorescence correlation spectroscopy [61]. English also studied the incorporation of organic-capped silicon nanoparticles into the hydrophobic interior of phospholipid vesicle bilayers and the accessibility of membrane-embedded nanoparticles to aqueous and lipid-bound quenchers was determined using photoluminescence quenching [62]. A sensitive and easily regenerated nano-optical sensor can be prepared based on immobilization of avidincoated colloidal gold particles on a biotin-modified planar lipid bilayer supported on the walls of a quartz cuvette, which is proven sensitive enough to follow the hybridization kinetics of 15-mer fully complementary DNA strands without the introduction of labels or secondary signal amplification [63]. Mayer and his coworkers found that multilayers of various phospholipids on silicon substrates could induce spontaneous embedding of FePt nanoparticles deposited from the gas phase. A lipid monolayer could be formed around individual nanoparticles when lipid was in the intermediate phase and liquid crystalline phase, and the molecular mobility of the multilayers allows for self-assembly of the particles in regular two-dimensional arrangements at the same time [64, 65]. Phospholipid liposomes can be stabilized using charged PS nanoparticles because of the formation of a nanoparticle–liposome hybrid structure under the driving of charge–dipole attraction [66]. In addition, liposome can be immobilized on the surface of solid substrate using this method and retain stable over a period of days [67].

3.2. Assembly of Lipid Membrane on Nanocrystals The creation of a lipid bilayer onto the solid surface has become a widely used approach to render a surface biocompatible. Recently, the assembling of lipid membranes on carbon nanotubes (CNT) have been widely reported, including supported lipid bilayers over single-walled carbon nanotube transistors [68], onedimensional lipid bilayers on carbon nanotube [69], and lipid bilayer at multiwalled CNT [70]. Single bilayer membranes of 1-palmitoyl-2-oleoyl-sn-glycero-3phosphocholine (POPC) can be formed on micron thin-films of hydrophilized CNT by fusion of small unilamellar vesicles. This platform lends support to homogeneous and continuous bilayer membranes that have promising applications in the fields of bio-materials, biosensors, and biophysics [71]. Some reports on the formation of lipid membranes on nanocrystals have been introduced in the text above. In addition, the formation of supported lipid bilayers on silica nanoparticles has been revealed by cryoelectron microscopy [72].

X. Liu et al.

212

In our laboratory, we copied the preparative procedure [73–77] of supported hybrid phospholipid/alkanethiol bilayers on a planar gold macroelectrode to the asprepared gold colloid electrodes [78]. The whole preparative process of anchoring hybrid alkanethiolate/phospholipid bilayers to the as-prepared gold colloid electrodes is outlined in Scheme 1. Immersion of the as-prepared gold colloid electrodes into dilute octadecanethiol ethanol solution results in selective anchoring of octadecanethiol onto the gold nanoparticles by a strong gold–sulfur bond. The formation of the subsequent phospholipid layer was done by immersing into a 7-mg/ml DPPC chloroform solution and subsequent incubating in 0.1 M KCl according to the well-developed preparative procedure for supported hybrid bilayers on a planar macroelectrode [73–77]. The electrochemical and spectral results show that the bilayers on colloid electrodes are interdigitated, which are different from their two-dimensional counterparts on a planar macroelectrode.

4. The Application of Lipid-based Nanocrystals 4.1. Interaction of Nanocrystals with Lipid Membrane With the development of nanotechnology, the applications of various nanocrystals have become the main research subject gradually. In particular, the applications in biochemical and life science attract more interest of scientists. As the first step to determine the practical condition of nanocrystals in biology, the interaction of nanocrystals with biomembrane must be investigated. Besides, nanoparticles have the similar size to biomacromolecules and can be used as the model of biomacromolecules to study the interaction of biomacromolecules with biomembrane. Various lipid membranes, including monolayer, bilayer lipid membrane, supported bilayer lipid membrane, liposome, and so on, are useful models in the studies of

Gold

APTMS ITO

Au colloid ITO

ITO

Gold

Gold

Gold di-palmitoyl-PC

Octadecanethiol ITO

ITO

Scheme 1 Schematic of preparative process of interdigited l-a-dipalmitoylphosphatidylcholine (DPPC)/octadecanethiol bilayers on the APTMS-tethered gold nanoparticle films.

Lipid-based Strategies in Inorganic Nano-materials and Biomineralization Study

213

biomembranes. Thus, the studies on interaction between nanocrystals and biomimetic membranes provide important information for the application of nanocrystals in biochemical and life science research. The effects of various nanocrystals on lipid membranes have been probed. Gold/silver nanoparticles can be loaded in hydrophobic region of the bilayer of DPPC liposomes, named as gold/silver-loaded liposomes. The membrane fluidities of DPPC bilayer increased after loading the nanoparticles [79, 80]. The system demonstrates that gold/silver nanoparticles provide an effective means for designing thermally sensitive liposome. Bhattacharya and Srivastava investigated the interactions between gold nanoparticles bearing cationic single-chain, double-chain, and cholesterol and DPPC vesicular membranes. They found that increased doping of single-chain nanoparticles in DPPC resulted in the phases that melt at higher temperatures; inclusion of an incremental amount of double-chain nanoparticles lowered the melting temperature of DPPC and the cationic cholesteryl nanoparticle effect on DPPC in a manner somewhat analogous to that of cholesterol itself and broadened the DPPC melting transition [81]. Interaction of quantum dots coated with an amphiphilic macromolecule (amphipol polymers) with giant lipid vesicles was analyzed. It was concluded that it was important of pH as a regulating parameter of the interaction and that in common biological buffers at neutral pH, no interactions occurred between quantum dots and neutral or positively charged lipid vesicles [82]. With nanoparticles as simple models of proteins or colloids, the interaction of bilayer vesicles and adhesive nanoparticles was studied using a Brownian dynamics simulation. The results showed that the adhering nanoparticles induced the morphological change of the vesicle: budding, formation of two vesicles, and promotion of fission [83]. Blick and coworkers reported that quantum dots initiated current bursts in lipid bilayer membranes upon application of a bias voltage. The phenomenon they observed resembled those produced by the peptaibol class of antibiotics such as alamethicin and trichorzins [84].

4.2. Examples of the Applications of Lipid-based Nanocrystals It has been point out that the application of nanocrystals was given more and more attention after the various synthesis methods were developed. The biocompatibility of lipid-based nanocrystals has been emphasized because the use of nanocrystals in biological system always poses concerns about potential cytotoxicity. It has been reported lipid-modified nanocrystals were less toxic or nontoxic [59, 85, 86]. Subsequently, we summarize some applications of lipid-based nanocrystals as in Table 1 [17, 36, 37, 59, 86–98].

5. The Application of Lipid in Biomineralization Several excellent reviews of lipid-based biomineralization have been published [6, 99, 100]. Herein, we deal with the recent progress of biomineralization of Ca2þ based on lipid membrane, which consists of the crystallization of calcium phosphate, calcium carbonate, calcium oxalate, and so on.

X. Liu et al.

214 Table 1

Some applications of lipid-based nanocrystals

Number

Nanocrystals

Applications

References

1 2

Pt nanoparticles in s-BLM DDAB monolayerprotected Au nanoparticles DDAB bilayer-capped gold nanoparticles Lipid-coated CdSe nanocrystals

Biosensor Realizing DET of protein

[87] [17]

Pseudostationary phase in capillary electrophoresis Multifunctionalized luminescent scaffolds for supramolecular biological assemblies Controlled targeting of live cells, bioimaging Protein separations

[36]

Targeted gene delivery, targeted drug delivery, in vivo imaging, and treatment of carcinoma Multimodal diagnostic and therapeutic applications

[90–97]

3 4

5 6 7

8

Lipid/CdSe nanocrystals hybrid nanocontainers DMPG-coated magnetic particles Magnetic liposomes (liposomes with incorporated magnetite particles) Liposome–nanoparticle hybrids

[37]

[59, 86, 88] [89]

[98]

DMPG, dimyristoylphosphatidylglycerol; DDAB, didodecyldimethylammonium bromide; DET, direct electron transfer.

In order to control the mineralization process under well-controlled conditions in confined spaces, Ball and coworkers produced giant liposomes containing calcium ions as active ions in the mineralization process, spermine as activator of crystal growth, and alkaline phosphatase as a catalyst to convert phosphate esters into phosphates and subsequently to calcium phosphate crystals. This opens the route to control the calcium phosphate particle size in biomimetic systems [101]. They also investigated the immobilization of liposome in polyelectrolyte multilayers as submicronic reactors for mineralization using the same method [102]. For the sake of clarifying the role of lipids in urinary stones, Talham’s group and other groups studied the calcium oxalate precipitation at phospholipid Langmuir monolayers, since calcium oxalate is the principal mineral component of most urinary stones and lipid membranes possibly provide sites for the initial nucleation event [103–106]. These studies have important significance to the problem of urinary stone formation. The effects of several factors such as multifunctional sodium carboxylates, the concentration of lecithin, calcium as well as oxalate on the phase composition of calcium oxalate crystals grown in lecithin–water liposomes were investigated [107–109]. The mineralization of calcium phosphates under lipid Langmuir monolayer is also explored [110]. In our group, we found the interesting finding that crystallization of calcium carbonate in the presence of DMPG vesicles by a simple gas diffusion method

Lipid-based Strategies in Inorganic Nano-materials and Biomineralization Study

Figure 5

215

SEM image of a CaCO3 spherule.

resulted in the formation of unusual microscopic calcium carbonate (CaCO3) spherules, which take on complex macroporous network structure (Fig. 5). Our experiments showed that DMPG vesicles might play an important role in the process of CaCO3 crystallization [111].

6. Conclusion and Perspectives In conclusion, lipid has been applied widely in the research of inorganic nanomaterials and biomineralization. It can be used as protective agent, template, and reaction compartment owing to its amphipathic character and several different phases. The synthesis methods of inorganic material have been investigated extensively and their applications are still a challenge for scientists in materials filed. People have focused on the organic–inorganic hybrid materials due to their special characters. Lipid–inorganic hybrid materials are in possession of good stability, biocompatibility, easy functionalization, which have been applied in biochemical, medicinal, and life science researches by some pioneering work and are very promising for future biomolecular manipulations and applications, such as labeling, detection, separation, transfer of drugs, and genetic delivery. There still are many unknown and significant aspects needed to be studied and explored.

ACKNOWLEDGMENTS This work was supported by the National Natural Science Foundation of China via Grant Nos. 20335040 and 20275036 and National Key Basic Research Development Project 2001 CB5102.

216

X. Liu et al.

REFERENCES [1] J.H. Fendler, Nanoparticles and Nanostructured Films: Preparation, Characterization and Applications, Wiley-VCH, Weinheim, 1998. [2] G. Schmid, Large clusters and colloids, metals in the embryonic state, Chem. Rev. 92 (1992) 1709–1727. [3] R.B. Gennis, Biomembranes: Molecular Structure and Function, Springer-Verlag, New York, 1989, pp. 40–41. [4] J.M. Seddon, Structure of the inverted hexagonal (HII) phase, and non-lamellar phase transitions of lipids, Biochim. Biophys. Acta 1031 (1990) 1–69. [5] D.P. Siegel, Inverted micellar intermediates and the transitions between lamellar, cubic, and inverted hexagonal lipid phases. II. Implications for membrane-membrane interactions and membrane fusion, Biophys. J. 49 (1986) 1171–1183. [6] J.H. Collier, P.B. Messersmith, Biomimetic mineralization, mesoporous structures, in: K.H. Buschow, R.W. Cahn, M.C. Flemings, B. Ilschner, E.J. Kramer, S. Mahajan, P. Veyssiere (Eds.), Encyclopedia of Materials: Science and Technology, Elsevier Science Ltd., Amsterdam, 2001, p. 602. [7] H.A. Lowenstam, S. Weiner, On Biomineralization, Oxford University Press, New York, 1989. [8] H.T. Schmidt, A.E. Ostafin, Liposome directed growth of calcium phosphate nanoshells, Adv. Mater. 14 (2002) 532–535. [9] J.M. Ouyang, L. Duan, B. Tieke, Effects of carboxylic acids on the crystal growth of calcium oxalate nanoparticles in lecithin-water liposome systems, Langmuir 19 (2003) 8980–8985. [10] Q.L. Feng, Q.H. Chen, H. Wang, F.Z. Cui, Influence of concentration of calcium ion on controlled precipitation of calcium phosphate within unilamellar lipid vesicles, J. Cryst. Growth 186 (1998) 245–250. [11] S. Mann, J.P. Hannington, R.J.P. Williams, Phospholipid vesicles as a model system for biomineralization, Nature 324 (1986) 565–567. [12] M. Brust, M. Walker, D. Bethell, D.J. Schffrin, R. Whyman, Synthesis of thiol-derivatised gold nanoparticles in two-phase liquid-liquid system, J. Chem. Soc. Chem. Commun. 7 (1994) 801–802. [13] M.J. Hostetler, J.E. Wingate, C.J. Zhong, J.E. Harris, R.W. Vachet, M.R. Clark, J.D. Londono, S.J. Green, J.J. Stokes, G.D. Wignall, G.L. Glish, M.D. Porter, N.D. Evans, R.W. Murray, Alkanethiolate gold cluster molecules with core diameters from 1.5 to 5.2 nm: Core and monolayer properties as a function of core size, Langmuir 14 (1998) 17–30. [14] S.W. Chen, A.C. Templeton, R.W. Murray, Monolayer-protected cluster growth dynamics, Langmuir 16 (2000) 3543–3548. [15] D.E. Cliffel, F.P. Zamborini, S.M. Gross, R.W. Murray, Mercaptoammonium-monolayerprotected, water-soluble gold, silver, and palladium clusters, Langmuir 16 (2000) 9699–9702. [16] W.P. Wuelfing, F.P. Zamborini, A.C. Templeton, X. Wen, H. Yoon, R.W. Murray, Monolayer-protected clusters: Molecular precursors to metal films, Chem. Mater. 13 (2001) 87–95. [17] X.J. Han, W.L. Cheng, Z.L. Zhang, S.J. Dong, E.K. Wang, Direct electron transfer between hemoglobin and a glassy carbon electrode facilitated by lipid-protected gold nanoparticles, Biochim. Biophys. Acta 1556 (2002) 273–277. [18] X.M. Lin, C.M. Sorensen, Ligand-induced gold nanocrystal superlattice formation in colloidal solution, Chem. Mater. 11 (1999) 198–202. [19] H.F. Zhu, C. Tao, S.P. Zheng, J.B. Li, One step synthesis and phase transition of phospholipidmodified Au particles into toluene, Colloids Surf. A: Physicochem., Eng. Aspects 257–258 (2005) 411–414. [20] J. Giria, S.G. Thakurtaa, J. Bellareb, A.K. Nigamc, D. Bahadur, Preparation and characterization of phospholipid stabilized uniform sized magnetite nanoparticles, J. Magn. Magn. Mater. 293 (2005) 62–68. [21] V. Patil, K.S. Mayya, S.D. Pradhan, M. Sastry, Evidence for novel interdigitated bilayer formation of fatty acids during three-dimensional self-assembly on silver colloidal particles, J. Am. Chem. Soc. 119 (1997) 9281–9282.

Lipid-based Strategies in Inorganic Nano-materials and Biomineralization Study

217

[22] V. Patil, M. Sastry, Surface derivatization of colloidal silver particles using interdigitated bilayers: A novel strategy for electrostatic immobilization of colloidal particles in thermally evaporated fatty acid/fatty amine films, Langmuir 14 (1998) 2707–2711. [23] M. Sastry, K.S. Mayya, V. Patil, Facile surface modification of colloidal particles using bilayer surfactant assemblies: A new strategy for electrostatic complexation in Langmuir-Blodgett films, Langmuir 14 (1998) 5921–5928. [24] N. Lala, A.G. Chittiboyina, S.P. Chavan, M. Sastry, Biotinylation of colloidal gold particles using interdigitated bilayers: A UV–visible spectroscopy and TEM study of the biotin–avidin molecular recognition process, Colloids Surf. A 205 (2002) 15–20. [25] A. Swami, A. Kumar, M. Sastry, Formation of water-dispersible gold nanoparticles using a technique based on surface-bound interdigitated bilayers, Langmuir 19 (2003) 1168–1172. [26] W.L. Cheng, S.J. Dong, E.K. Wang, Studies of electrochemical quantized capacitance charging of surface ensembles of silver nanoparticles, Electrochem. Commun. 4 (2002) 412–416. [27] S. Mandal, S.K. Arumugam, S.D. Adyanthaya, R. Pasricha, M. Sastry, Use of aqueous foams for the synthesis of gold nanoparticles of variable morphology, J. Mater. Chem. 14 (2004) 43–47. [28] H.Y. Fan, K. Yang, D.M. Boye, T. Sigmon, K.J. Malloy, H.F. Xu, G.P. Lopez, C.J. Brinker, Self-assembly of ordered, robust, three-dimensional gold nanocrystal/silica array, Science 304 (2004) 567–571. [29] B. Nikoobakht, M.A. El-sayed, Evidence for bilayer assembly of cationic surfactants on the surface of gold nanorods, Langmuir 17 (2001) 6368–6374. [30] A. Gole, C.J. Orendorff, C.J. Murphy, Immobilization of gold nanorods onto acid-terminated self-assembled monolayers via electrostatic interactions, Langmuir 20 (2004) 7117–7122. [31] L. Shen, P.E. Laibinis, T.A. Hatton, Bilayer surfactant stabilized magnetic fluids: Synthesis and interactions at interfaces, Langmuir 15 (1999) 447–453. [32] H.Y. Fan, E.W. Leve, C. Scullin, J. Gabaldon, D. Tallant, S. Bunge, T. Boyle, M.C. Wilson, C.J. Brinker, Surfactant-assisted synthesis of water-soluble and biocompatible semiconductor quantum dot micelles, Nano Lett. 5 (2005) 645–648. [33] P. He, M.W. Urban, Phospholipid-stabilized Au-nanoparticles, Biomacromolecules 6 (2005) 1224–1225. [34] L.X. Zhang, X.P. Sun, Y.H. Song, X.E. Jiang, S.J. Dong, E.K. Wang, Didodecyldimethylammonium bromide lipid bilayer-protected gold nanoparticles: Synthesis, characterization, and self-assembly, Langmuir 22 (2006) 2838–2843. [35] G. Decher, J.B. Schlenoff (Eds.), Multilayer Thin Films, Wiley-VCH, New York, 2002. [36] C.J. Yu, C.L. Su, W.L. Tseng, Separation of acidic and basic proteins by nanoparticle-filled capillary electrophoresis, Anal. Chem. 78 (2006) 8004–8010. [37] I. Geissbuehler, R. Hovius, K.L. Martinez, M. Adrian, K.R. Thampi, H. Vogel, Lipid-coated nanocrystals as multifunctionalized luminescent scaffolds for supramolecular biological assemblies, Angew. Chem. Int. Ed. 44 (2005) 1388–1392. [38] J. Xie, S. Peng, N. Brower, N. Pourmand, S.X. Wang, S.H. Sun, One-pot synthesis of monodisperse iron oxide nanoparticles for potential biomedical applications, Pure Appl. Chem. 78 (2006) 1003–1014. [39] N.M. Correa, H.G. Zhang, Z.A. Schelly, Preparation of AgBr quantum dots via electroporation of vesicles, J. Am. Chem. Soc. 122 (2000) 6432–6434. [40] H.X. Zeng, D.S. Marynich, Z.A. Schelly, CdS quantum dots prepared via electroporation of vesicles: Experimental and computational results, in: 225th ACS National Meeting, New Orleans, . [41] S.X. Wu, H.X. Zeng, Z.A. Schelly, Preparation of ultrasmall, uncapped PbS quantum dots via electroporation of vesicles, Langmuir 21 (2005) 686–691. [42] S.X. Wu, H.X. Zeng, Z.A. Schelly, Growth of uncapped, subnanometer size gold clusters prepared via electroporation of vesicles, J. Phys. Chem. B 109 (2005) 18715–18718. [43] B.A. Korgel, H.G. Monbouquette, Synthesis of size-monodisperse CdS nanocrystals using phosphatidylcholine vesicles as true reaction compartments, J. Phys. Chem. 100 (1996) 346–351. [44] M.T. Kennedy, B.A. Korgel, H.G. Monbouquette, J.A. Zasadzinski, Cryo-transmission electron microscopy confirms controlled synthesis of cadmium sulfide nanocrystals within lecithin vesicles, Chem. Mater. 10 (1998) 2116–2119.

218

X. Liu et al.

[45] B.A. Korgel, H.G. Monbouquette, Controlled synthesis of mixed core and layered (Zn, Cd)S and (Hg, Cd)S nanocrystals within phosphatidylcholine vesicles, Langmuir 16 (2000) 3588–3594. [46] G.M. Chow, M.A. Markowitz, R. Rayne, D.N. Dunn, A. Singh, Phospholipid mediated synthesis and characterization of gold nanoparticles, J. Colloid Interf. Sci. 183 (1996) 135–142. [47] M.A. Markowitz, D.N. Dunn, G.M. Chow, J. Zhang, The effect of membrane charge on gold nanoparticle synthesis via surfactant membranes, J. Colloid Interf. Sci. 210 (1999) 73–85. [48] R.R. Price, W.J. Dressick, A. Singh, Fabrication of nanoscale metallic spirals using phospholipid microtubule organizational templates, J. Am. Chem. Soc. 125 (2003) 11259–11263. [49] Y.J. Song, R.M. Garcia, R.M. Dorin, H.R. Wang, Y. Qiu, J.A. Shelnut, Synthesis of platinum nanocages by using liposomes containing photocatalyst molecules, Angew. Chem. Int. Ed. 45 (2006) 8126–8130. [50] M.J. Yoon, M.J. Seo, C.J. Jeong, J.H. Jang, K.S. Jeon, Synthesis of liposome-templated titania nanodisks: Optical properties and photocatalytic activities, Chem. Mater. 17 (2005) 6069–6079. [51] S. Mandal, S.R. Sainkar, M. Sastry, Electrostatic entrapment of chloroaurate ions in patterned lipid films and the in situ formation of gold nanoparticles, Nanotechnology 12 (2001) 358–362. [52] A. Kumar, V. Ramakrishnan, R. Gonnade, K.N. Ganesh, M. Sastry, Electrostatically entrapped DNA molecules in lipid thin films as templates for the in situ growth of silver nanoparticles, Nanotechnology 13 (2002) 597–600. [53] H.J. Liang, T.E. Angelini, P.V. Braun, G.C.L. Wong, Roles of anionic and cationic template components in biomineralization of CdS nanorods using self-assembled DNA-membrane complexes, J. Am. Chem. Soc. 126 (2004) 14157–14165. [54] D. Ibano, Y. Yokotay, T. Tominaga, Preparation of gold nanoplates protected by an anionic phospholipid, Chem. Lett. 32 (2003) 574–575. [55] S. Puvvada, S. Baral, G.M. Chow, S.B. Qadi, B.R. Ratna, Synthesis of palladium metal nanoparticles in the bicontinuous cubic phase of glycerol monooleate, J. Am. Chem. Soc. 116 (1994) 2135–2136. [56] J.P. Yang, S.B. Qadri, B.R. Ratna, Structural and morphological characterization of PbS nanocrystallites synthesized in the bicontinuous cubic phase of a lipid, J. Phys. Chem. 100 (1996) 17255–17259. [57] M.S. Bakshi, F. Possmayer, N.O. Petersen, Simultaneous synthesis of Au and Cu nanoparticles in pseudo-core-shell type arrangement facilitated by DMPG and 12-6-12 capping agents, Chem. Mater. 19 (2007) 1257–1266. [58] J.A. Kloepfer, N. Cohen, J.L. Nadeau, FRET between CdSe quantum dots in lipid vesicles and water- and lipid-soluble dyes, J. Phys. Chem. B 108 (2004) 17042–17049. [59] G. Gopalakrishnan, K. Danelon, P. Izewska, M. Prummer, P.Y. Bolinger, I. Geissbuhler, D. Demurtas, J. Dubochet, H. Vogel, Multifunctional lipid/quantum-dot hybrid nanocontainers for controlled targeting live cells, Angew. Chem. Int. Ed. 45 (2006) 5478–5483. [60] L.Y. Feng, X.G. Kong, K.F. Chao, Y.J. Sun, Q.H. Zeng, Y.L. Zhang, Efficient phase transfer of hydrophobic CdSe quantum dots: From nonpolar organic solvent to biocompatible water buffer, Mater. Chem. Phys. 93 (2005) 310–313. [61] C.S. Chen, J. Yao, R.A. Durst, Liposome encapsulation of fluorescent nanoparticles: Quantum dots and silica nanoparticles, J. Nanopart. Res. 8 (2006) 1033–1038. [62] H. Jang, L.E. Pell, B.A. Korgel, D.S. English, Photoluminescence quenching of silicon nanoparticles in phospholipid vesicle bilayers, J. Photochem. Photobiol. A 158 (2003) 111–117. [63] L. Olofsson, T. Rindzeviclus, I. Pfeiffer, M. Kall, F. Hook, Surface-based gold-nanoparticle sensor for specific and quantitative DNA hybridization detection, Langmuir 19 (2003) 10414–10419. [64] A. Terheiden, C. Mayer, K. Moh, B. Stahlmecke, S. Stappert, M. Acet, B. Rellinghaus, Postdeposition organic coating and self-assembly of gas phase prepared FePt nanoparticles on lipid reservoir films, Appl. Phys. Lett. 84 (2004) 3891–3893. [65] A. Terheiden, B. Rellinghaus, S. Stappert, M. Acet, C. Mayer, Embedding and selforganization of nanoparticles in phospholipid multilayers, J. Chem. Phys. 121 (2004) 510–516. [66] L.F. Zhang, S. Granick, How to stabilize phospholipid liposomes (using nanoparticles), Nano Lett. 6 (2006) 694–698. [67] L.F. Zhang, L. Hong, Y. Yu, S.C. Bae, S. Granick, Nanoparticle-assisted surface immobilization of phospholipid liposomes, J. Am. Chem. Soc. 128 (2006) 9026–9027.

Lipid-based Strategies in Inorganic Nano-materials and Biomineralization Study

219

[68] X.J. Zhou, J.M. Moran-Mirabal, H.G. Craighead, P.L. McEuen, Supported lipid bilayer/ carbon nanotube hybrids, Nat. Nanotechnol. 2 (2007) 185–190. [69] A.B. Artyukhin, A. Shestakov, J. Harper, O. Bakajin, P. Stroeve, A. Noy, Functional onedimensional lipid bilayers on carbon nanotube templates, J. Am. Chem. Soc. 127 (2005) 7538–7542. [70] J.S. Ye, H.F. Cui, Y. Wen, W.D. Zhang, A. Ottova, H.T. Tien, G.Q. Xu, F.S. Sheu, selfassembly of bilayer lipid membrane at multiwalled carbon nanotubes towards the development of photo-switched functional device, Electrochem. Commun. 7 (2005) 81–86. [71] J. Gagner, H. Johnson, E. Watkins, Q.W. Li, M. Terrones, J. Majewski, Carbon nanotube supported single phospholipid bilayer, Langmuir 22 (2006) 10909–10911. [72] S. Mornet, O. Lambert, E. Duguet, A. Brisson, The formation of supported lipid bilayers on silica nanoparticles revealed by cryoelectron microscopy, Nano Lett. 5 (2005) 281–285. [73] E.L. Florin, H.E. Gaub, Painted supported lipid membranes, Biophys. J. 64 (1993) 375–379. [74] J. Li, L. Ding, E. Wang, S. Dong, The ion selectivity of monensin incorporated phospholipid/ alkanethiol bilayers, J. Electroanal. Chem. 414 (1996) 17–21. [75] A.L. Plant, Self-assembled phospholipid/alkanethiol biomimetic bilayers on gold, Langmuir 9 (1993) 2764–2767. [76] C.W. Meuse, G. Niaura, M.L. Lewis, A.L. Plant, Assessing the molecular structure of alkanethiol monolayers in hybrid bilayer membranes with vibrational spectroscopies, Langmuir 14 (1998) 1604–1611. [77] E. Kalb, S. Frey, L.K. Tamm, Formation of supported planar bilayers by fusion of vesicles to supported phospholipid monolayers, Biochim. Biophys. Acta 1103 (1992) 307–316. [78] W.L. Cheng, X.J. Han, E.K. Wang, S.J. Dong, Interdigited phospholipid/alkanethiol bilayers assembled on APTMS-supported gold colloid electrodes, Electroanalysis 16 (2004) 127–131. [79] S.H. Park, S.G. Oh, J.Y. Mun, S.S. Han, Loading of gold nanoparticles inside the DPPC bilayers of liposome and their effects on membrane fluidities, Colloids Surf. B Biointerf. 48 (2006) 112–118. [80] S.H. Park, S.G. Oh, J.Y. Mun, S.S. Han, Effects of silver nanoparticles on the fluidity of bilayer in phospholipid liposome, Colloids Surf. B Biointerf. 44 (2005) 117–122. [81] S. Bhattacharya, A. Srivastava, Synthesis and characterization of novel cationic lipid and cholesterol-coated gold nanoparticles and their interactions with dipalmitoylphosphatidylcholine membranes, Langmuir 19 (2003) 4439–4447. [82] C. Luccardini, C. Tribet, F. Vial, V. Marchi-Artzner, M. Dahan, Size, charge, and interactions with giant lipid vesicles of quantum dots coated with an amphiphilic macromolecule, Langmuir 22 (2006) 2304–2310. [83] H. Noguchi, M. Takasu, Adhesion of nanoparticles to vesicles: A Brownian dynamics simulation, Biophys. J. 83 (2002) 299–308. [84] S. Ramachandran, G.L. Kumar, R.H. Blick, D.W. van der Weide, Current bursts in lipid bilayers initiated by colloidal quantum dots, Appl. Phys. Lett. 86 083901, (2005) 1–3. [85] H. Takahashi, Y. Niidome, T. Niidome, K. Kaneko, H. Kawasaki, S. Yamada, Modification of gold nanorods using phosphatidylcholine to reduce cytotoxicity, Langmuir 22 (2006) 2–5. [86] B. Dubertret, P. Skourides, D.J. Norris, V. Noireaux, A.H. Brivanlou, A. Libchaber, In vivo imaging of quantum dots encapsulated in phospholipid micelles, Science 298 (2002) 1759–1762. [87] J.S. Ye, A. Ottova, H.T. Tien, F.S. Sheua, Nanostructured platinum-lipid bilayer composite as biosensor, Bioelectrochemistry 59 (2003) 65–72. [88] S.W. Liu, C.M. Lee, S.N. Wang, D.R. Lu, A new bioimaging carrier for fluorescent quantum dots: Phospholipid nanoemulsion mimicking natural lipoprotein core, Drug Deliv. 13 (2006) 159–164. [89] S. Bucak, D.A. Jones, P.E. Laibinis, T.A. Hatton, Protein separations using colloidal magnetic nanoparticles, Biotechnol. Prog. 19 (2003) 477–484. [90] K. Hirao, T. Sugita, T. Kubo, K. Igarashi, K.J. Tanimoto, T. Murakami, Y.J. Yasunaga, M. Ochi, Targeted gene delivery to human osteosarcoma cells with magnetic cationic liposomes under a magnetic field, Int. J. Oncol. 22 (2003) 1065–1071. [91] A.A. Kuznetsov, V.I. Filippov, R.N. Alyautdin, N.L. Torshina, O.A. Kuznetsov, Application of magnetic liposomes for magnetically guided transport of muscle relaxants and anti-cancer photodynamic drugs, J. Magn. Magn. Mater. 225 (2001) 95–100.

220

X. Liu et al.

[92] M.S. Martina, J.P. Fortin, C. Menager, O. Clement, G. Barratt, C. Grabielle-Madelmont, F. Gazeau, V. Cabuil, S. Lesieur, Generation of superparamagnetic liposomes revealed as highly efficient MRI contrast agents for in vivo imaging, J. Am. Chem. Soc. 127 (2005) 10676–10685. [93] M. Babincova, P. Sourivong, D. Chorvat, P. Babinec, Laser triggered drug release from magnetoliposomes, J. Magn. Magn. Mater. 194 (1999) 163–166. [94] M. Shinkai, B. Le, H. Honda, K. Yoshikawa, K. Shimizu, S. Saga, T. Wakabayashi, J. Yoshida, T. Kobayashi, Targeting hyperthermia for renal cell carcinoma using human MN antigenspecific magnetoliposomes, Jpn. J. Cancer Res. 92 (2001) 1138–1145. [95] M. Shinkai, M. Yanase, H. Honda, T. Wakabayashi, J. Yoshida, T. Kobayashi, Intracellular hyperthermia for cancer using magnetite cationic liposomes: In vitro study, Jpn. J. Cancer Res. 87 (1996) 1179–1183. [96] M. Yanase, M. Shinkai, H. Honda, T. Wakabayashi, J. Yoshida, T. Kobayashi, Intracellular hyperthermia for cancer using magnetite cationic liposomes: An in vivo study, Jpn. J. Cancer Res. 89 (1998) 463–469. [97] M. Yanase, M. Shinkai, H. Honda, T. Wakabayashi, J. Yoshida, T. Kobayashi, Antitumor immunity induction by intracellular hyperthermia using magnetite cationic liposomes, Jpn. J. Cancer Res. 89 (1998) 775–782. [98] Wafa’ T. Al-Jamal, K. Kostarelos, Liposome-nanoparticle hybrids for multimodal diagnostic and therapeutic applications, Nanomedicine 2 (2007) 85–98. [99] J.H. Collier, P.B. Messersmith, Phospholipid strategies in biomineralization and biomaterials research, Annu. Rev. Mater. Res. 31 (2001) 237–263. [100] J.M. OuYang, N. Zhou, Research progress of biomineralization in liposome systems, J. Synth. Cryst. 33 (2004) 898–908. [101] M. Michel, M. Winterhalter, L. Darbois, J. Hemmerle, J.C. Voegel, P. Schaaf, V. Ball, Giant liposome microreactors for controlled production of calcium phosphate crystals, Langmuir 20 (2004) 6127–6133. [102] M. Michel, Y. Arntz, G. Fleith, J. Togquant, Y. Haikel, J.C. Voegel, P. Schaaf, V. Ball, Layerby-layer self-assembled polyelectrolyte multilayers with embedded liposomes: Immobilized submicronic reactors for minerlization, Langmuir 22 (2006) 2358–2364. [103] D.R. Talham, R. Backov, I.O. Benitez, D.M. Sharbaugh, S. Whipps, S.R. Khan, Role of lipids in urinary stones: Studies of calcium oxalate precipitation at phospholipid Langmuir monolayers, Langmuir 22 (2006) 2450–2456. [104] I.O. Benitez, D.R. Talham, Calcium oxalate monohydrate precipitation at membrane lipid rafts, J. Am. Chem. Soc. 127 (2005) 2814–2815. [105] S.P. Deng, J.M. Ouyang, Effects of dipalmitoyl phosphatidylcholine monolayers to the crystallization of calcium oxalate monohydrate from the solution containing chondroitin sulfate C, Colloids Surf. A: Physicochem. Eng. Aspects 47–50 (2005) 257–258. [106] J.M. Ouyang, S.P. Deng, J.P. Zhong, Bernd Tieke, S.H. Yu, Crystallization of calcium oxalate monohydrate at dipalmitoylphosphatidylcholine monolayers in the presence of chondroitin sulfate A, J. Cryst. Growth 270 (2004) 646–654. [107] J.M. Ouyang, N. Zhou, L. Duan, B. Tieke, Ability of multifunctional sodium carboxylates to favor crystal growth of calcium oxalate dihydrate and trihydrate in lecithin-water liposome systems, Colloids Surf. A: Physicochem. Eng. Aspects 245 (2004) 153–162. [108] J.M. Ouyang, F. Deng, L. Duan, Effect of concentrations of lecithin, calcium and oxalate on crystal growth of calcium oxalate in vesicles, Colloids Surf. A: Physicochem. Eng. Aspects 257–258 (2005) 215–220. [109] J.M. Ouyang, S.P. Deng, N. Zhou, B. Tieke, Effect of tartrates with various counterions on the precipitation of calcium oxalate in vesicle solutions, Colloids Surf. A: Physicochem. Eng. Aspects 256 (2005) 21–27. [110] L.J. Zhang, H.G. Liu, X.S. Feng, R.J. Zhang, L. Zhang, Y.D. Mu, J.C. Hao, D.J. Qian, Y.F. Lou, Mineralization mechanism of calcium phosphates under three kinds of Langmuir monolayers, Langmuir 20 (2004) 2243–2249. [111] X.H. Liu, L.X. Zhang, Y.L. Wang, C.L. Guo, E.K. Wang, Biomimetic crystallization of unusual macroporous calcium carbonate spherules in the presence of phosphatidylglycerol vesicles, Cryst. Growth Des. 8 (2008) 759–762.

C H A P T E R

N I N E

Self-Reproduction of Micelles, Reverse Micelles, and Vesicles: Compartments Disclose a General Transformation Pattern Pasquale Stano1,2 and Pier Luigi Luisi2,* Contents 222 223 223 226 230 235 236 240 246 247 248 250 254 256 258 258

1. Introduction 2. Theoretical Backgrounds 2.1. Autopoiesis 2.2. General Concepts in the Self-Reproduction of Micelles 3. Self-Reproduction of Micelles and Reverse Micelles 4. Self-Reproduction of Vesicles 4.1. ‘‘Small’’ Vesicles 4.2. The ‘‘Matrix’’ Effect 4.3. Theoretical Considerations and Modeling of Self-Reproduction 4.4. Homeostatic Systems 4.5. Giant Vesicles 5. Vesicle-based Semisynthetic Cells and Their Self-Reproduction 5.1. The Minimal RNA Cell 6. Final Remarks Acknowledgments References

Abstract The discovery of self-reproduction of micelles and vesicles about 15 years ago opened a new page in the field of the supramolecular chemistry of surfactant aggregation phenomena, both for the importance of dynamic aspects of complexity of these systems and for its biological meaning. In fact, the self-reproduction of vesicles has suggested that the growth and the population increase of structures resembling the cells may take place solely because of physical and chemical forces. An increasing number of reports demonstrate that reverse micelles, micelles, sub-micrometric as well as giant vesicles readily Corresponding author. Tel.: þ39 06 57336329; Fax: þ39 06 55176329; E-mail address: [email protected] (P.L. Luisi).

* 1 2

‘‘Enrico Fermi’’ Study and Research Centre, Compendio del Viminale 00184 Rome, Italy Biology Department, University of Rome ‘‘RomaTre,’’ Viale G. Marconi 446, 00146 Rome, Italy

Advances in Planar Lipid Bilayers and Liposomes, Volume 7 ISSN 1554-4516, DOI: 10.1016/S1554-4516(08)00009-4

#

2008 Elsevier Inc. All rights reserved.

221

222

P. Stano and P. L. Luisi

undergo self-reproduction, generating new particles from a suitable precursor. The process follows an autocatalytic pattern, namely, the progressive increase in particle number is a nonlinear time course. In this chapter, we will review the most significant studies on the self-reproduction of different compartments, by following a combined historical and classifying approach that spans from the pioneering work on reverse micelles, to the case of normal (aqueous) micelles, to the studies on vesicles, giant vesicles, and water-in-oil emulsion droplets. Similarities and differences in reactive patterns are highlighted, indicating at the same time the unanswered questions. Some of the theoretical models, which have been proposed in the literature to interpret or model self-reproduction of micelles and vesicles, will be illustrated. We will also discuss whether and to what extent such processes comply with the theory of autopoiesis—from which they have been in fact generated, from the historical as well as strategic viewpoint. Finally, we will also shortly discuss the relevance of the self-reproduction of vesicles for emerging avenues of research, in particular for the field of minimal cells, meant as the compartments having the minimal and sufficient complexity to be defined as living.

1. Introduction Synthetic compartments such as micelles, reverse micelles, and vesicles have been often used as a model of biological compartments. For example, micelles are often used in catalysis or to reconstitute membrane proteins [1, 2], whereas reverse micelles have been used in biotechnology to carry out enzymatic reactions in apolar media [3], and also recently as a model of nuclear DNA supercondensation [4]. Vesicles, on the contrary, are the most consistent model of biological compartments, since their properties closely resemble those of cellular membranes. Thanks to this characteristic, vesicles and more particularly lipid vesicles (liposomes) have been widely studied in the last 40 years with respect to basic biophysics and biochemistry [5, 6], to investigate membrane structure, dynamics, binding, and permeability, and have been used for hydrophobic protein reconstitution. Also, a large amount of research has been devoted to the use of vesicles as drug- and gene-delivery systems [7]. In many of these studies, synthetic compartments are used as a material, and relatively little explicit attention has been paid to the aspects of their structural reactivity, that is, the reactivity of the compartment in itself. This is particularly true in the field of vesicles, where—if one excludes the classical studies on vesicle aggregation and fusion [8–10]—extensive investigations on vesicle reactivity are missing. The discovery of self-reproduction of micelle, reverse micelles, and vesicles, on the contrary, has contributed in broadening the current knowledge on compartment reactivity. The existence of a self-reproduction pathway has a special relevance for those scientists involved in the origins of life studies, and for those who attempt to construct cell-like particles (artificial, synthetic, or semisynthetic cells). Since its discovery in the early 1990s, not enough attention has been given to these studies in the biochemistry literature for several years, probably due to the overemphasis given to nucleic acids and to their replication as the basic mechanism of life processes. Things have been changing in the last few years, due to a more systemic view of biological processes—following the coming of system biology. This has also

Self-Reproduction of Micelles, Reverse Micelles, and Vesicles

223

caused a renewed interest in autopoiesis, something we will see later on in this chapter. Today, there is a tendency to consider self-reproduction of synthetic compartments as an important phenomenon that effectively affects the viewpoint of prebiotic chemists and biologists, since one of the most fundamental process of living organisms, that is, self-reproduction, can occur thanks to the relatively simple action of physicochemical forces. Of course, it has been taken as a model of prebiotic self-reproduction of protocells, or of similar and very simplified particles, that could have had a role in the origin of life on the earth. In this chapter, we will describe the most significant aspects of self-reproduction of synthetic compartments, as micelles, reverse micelles, and vesicles. All of them are formed by self-assembly of amphiphiles. We will skip the introduction to the physical-chemistry of self-assembly, assuming that most of the readers are familiar with it. When, however, assemblies of fatty acids will be described, a short introduction to the behavior of such surfactants in aqueous phase will be provided. The interested reader can find good general introduction and reference texts on fatty acids self-assembly in the work of Small [11], Cistola et al. [12], and Walde et al. [13]. It is important to notice that the studies on self-reproduction actually started by using reverse micelles, and soon after micelles. In particular, these studies represented the first attempt to experimentally implement the concept of autopoiesis, developed by the Chilean neurobiologist Humberto R. Maturana and Francisco J. Varela in the 1970s [14]. Therefore, before going on with the more technical sections on compartments self-reproduction, a short introduction to the autopoietic theory will be given, followed by a review on the self-reproduction research in chronological order, that is, reverse micelles and micelles first, then vesicles. In particular, vesicle self-reproduction will be discussed in detail, since vesicles are very important cell models and also because one of the most attractive and sharp-edged research in current biology deals with the construction of semisynthetic cells [15]. Some other aspects of self-reproduction, like computer simulations or theoretical treatments, will not be discussed here. Again, the interested reader can refer to the original papers [16–18]. Finally, a note on semantics. We will use the term ‘‘self-reproduction’’ dealing with micelles and vesicles, and not self-replication. In fact, as noted earlier [19–22], selfreplication refers to an exact replica of the original, as in the case of DNA structure. Self-reproduction is a more statistical process, in which the same kind of structure, but not necessarily an identical one, is being formed. Molecules self-replicate, cells selfreproduce. This discrimination may be important also from the historical point of view, as according to Dyson [22], self-reproduction, a statistical process, being simpler, preceded self-replication processes, which necessitate a more precise mechanism.

2. Theoretical Backgrounds 2.1. Autopoiesis The starting point for the concept of autopoiesis (from Greek auto ¼ self, poiesis ¼ production) is cellular life. Developed by Maturana and Varela, autopoiesis deals with the question ‘‘what is life?’’ by analyzing living systems as they are—‘‘here and

224

P. Stano and P. L. Luisi

now’’—that is, without referring to the question on how they originate. In this sense, autopoiesis is a descriptive theory based on phenomenology, and starts from the recognition that the main activity of the cell is the maintenance of its own individuality: and this is so despite the large number of transformations taking place inside its boundary. Self-maintenance is possible since the cell regenerates from the inside all the components that are being transformed and/or disposed of (boundary molecules included!). Cell life is recognized as a series of processes that produce all components that in turn originate the processes that produce such components and so on (Fig. 1A). This is possible since the cell—an autopoietic unit—is capable of taking up energy (e.g., in the form of chemical compounds) from the environment, and uses such energy to carry out internal processes that lead to its own existence. Thermodynamically speaking, an autopoietic unit is an open system that exchanges matter and energy with its environment, but is operationally closed, since its selforganization does not change and does not depend on outside sources, that is, it is an intrinsic property of such a system. Thanks to this theoretical framework, we can define life at the cellular level. In fact, the property of being alive belongs to the cell as a whole and not to single parts. In this respect, autopoiesis is clearly related to the general systems theory and explicitly states that investigation on life must be focused to self-bounded, A

The cyclic logic of autopoiesis

A minimal autopoietic system

B P

… the processes bringing to the production of molecules that constitute…

S

S

S

S

S S

S

S

S S

P

kP

S

S

S

S S

S S

S S S S S

S kD W

… the boundary that allows…

d[S] dt

= nP – nD = kP[P] – kD[S]

nP – nD > 0 Growth, self-reproduction nP – nD = 0 Homeostasis nP – nD < 0 Death

Figure 1 Autopoiesis and minimal autopoietic systems. Autopoietic organization follows a cyclic logic (A), as overemphasized by the self-pointing arrows, which link processes and structural elements in an endless reciprocal causation. A minimal autopoietic system (B) is constituted by a self-bounded system, which can uptake a precursor P from the environment, transform it by a (net of ) reaction into the boundary element S, which can also undergo a degradative process to W. Three possible cases can be envisaged: growth, homeostasis, and death/collapse.

Self-Reproduction of Micelles, Reverse Micelles, and Vesicles

225

self-maintaining, and self-generating systems. According to autopoiesis, the last three properties are characteristic of the definition of life. Associated to this cell-oriented view of autopoiesis, the contribution of Maturana and Varela has been developed to social autopoiesis, second- and third-order autopoiesis, and cognition. The reader can find a basic introduction to autopoiesis in recent reviews [21, 23]. Autopoiesis is not restricted to be a merely descriptive statement of the living. It might be used as a theoretical framework to realize complex chemical and biochemical systems that obey, at different degree, the foundations of autopoietic theory. The question is: is it possible to design and construct autopoieitic (molecular) systems? In Fig. 1B, a cartoon represents a minimal autopoietic system. Although this scheme is very general, we will make explicit reference to molecular systems. The autopoietic system sustains itself by transforming external components (here indicated as P) into elements (S) of the autopoietic system, which self-organize to give rise to the autopoietic unit. The transformation, here indicated by a simple process (P ! S), occurs within the system, that is, within the boundary that separates and distinguishes the system from the environment. Together with the constructive step (P ! S), there is a destructive step (S ! W) that transforms the elements of the autopoietic system into a different form. In molecular terms, molecule P is the precursor of molecule S, which is finally converted into W. Thus, an autopoietic system, working out of equilibrium, continuously uptakes and releases components in the environment with the only result being the maintenance of its own organization and structure. More precisely, structural components are continuously renewed by the two concurrent anabolic and catabolic processes; however, despite this continual turnover of components, the whole organization of the autopoietic unit does not change. A simple phenomenological kinetic analysis is also shown in Fig. 1B. The two processes (P ! S and S ! W) proceed with rates vP and vD, here indicated in their pseudo first-order form. Three possible relations between the rates originate three different states of autopoietic unit. In the first case, the rate of S production exceeds the decay rate (vP  vD > 0) and the autopoietic unit grows; we will see soon that this ‘‘growth’’ is related to the autopoietic self-reproduction. In the second case, vP  vD ¼ 0, the two rates being equal, and representing a homeostatic stationary state, with no net changes despite the two concurrent reactions. In the third case, vP  vD < 0 and the production rate is overwhelmed by the decay rate, so that the autopoietic unit collapses (or dies) by gradual consumption of its components. As will be shown in the following paragraphs, a homeostatic system based on fatty acid vesicles has been already implemented [24]. Within this chapter, the discussion will mainly focus on the first case only, when the rate of S production d[S]/dt> 0 and the system shows a tendency to grow. It is now clear, however, that growth alone brings about a physical enlargement of the autopoietic system, without self-reproduction. Self-reproduction implies growth and division, since two or more systems must originate from the initial one. This is possible only if a critical state is reached after growth, and if such a state allows a spontaneous rearrangement, generating new systems structurally similar to the initial one. Needless to say, the new systems must also maintain the autopoietic

226

P. Stano and P. L. Luisi

organization in order to perpetuate autopoiesis. In other words, self-reproduction is seen as a kinetic mode of the autopoietic unit, dictated solely by the internal activity as a result of the relative value of rate processes: in this sense, it can be defined as an autopoietic self-reproduction process. Within this framework, it really represents the simplest possible membrane self-reproduction process, and as such, a suitable model for the prebiotic life processes. In Section 2.2, the general principles of self-reproduction will be discussed, illustrating how the autopoietic theory actually inspired the experimental studies on compartment self-reproduction.

2.2. General Concepts in the Self-Reproduction of Micelles Micelles, reverse micelles, and vesicles are supramolecular structures formed by selfassembled amphiphilic molecules (Fig. 2A). These particles, which differ both in size and in structure, form under different conditions. Micelles and vesicles form in aqueous solvents, where the amphiphilic building blocks prefer to segregate their hydrophobic tail in the form of a micelle core or in the bilayer that forms the vesicle A

S =

Micelle

Reverse micelle

B SS S P S S S S S S S S S S

P

S S S S

Vesicle

S S S SS S

S

S S S

S S

SS S S S S S S SS S S

S

S SS S S S S S S S

SS S S S S S

S

Figure 2 Surfactant-based supramolecular structures can self-reproduce by a common pathway. The structure of micelles, reverse micelles, and unilamellar vesicles (A); drawn not to scale. Micelles and vesicles form in aqueous phase, whereas reverse micelles in apolar solvent. Micelles and reverse micelles have typical size of few nanometers, whereas vesicles’ size spans from 30 nm to more than 50 mm. General mechanism for micelles, reverse micelles, and vesicle selfreproduction (B). A suitable precursor is taken up by a particle, which converts it—within its boundary—into the boundary-forming compound S, so that the consequent enlargement (to an unknown intermediate) brings about destabilization, and division into two new particles.

Self-Reproduction of Micelles, Reverse Micelles, and Vesicles

227

membrane. Reverse micelles, on the contrary, form when certain surfactants are solubilized in an apolar solvent (e.g., hexane) in the presence of very small amount of water. Despite the quite different structure of micelles and reverse micelles, for the viewpoint of this review, it is useful to discuss these two systems together, since their reactive pattern differs substantially from that of vesicles. In addition to the obvious difference in size (micelles in the range 5–15 nm, vesicle from 30 nm to >50 mm), there is an even more fundamental difference between these aggregates. In fact, because of the highly dynamic nature of micellar aggregates, they are often considered as structures at the thermodynamic equilibrium, that is, the amphiphiles reversibly self-assemble in the most stable form. On the contrary, vesicles, especially those formed by long-chain amphiphiles, are classically described as kinetically trapped structures, that is, structures which form according to a specific pathway, and do not convert each other (or only with the greatest difficulty)—which also implies that a given structure may exist without being the most stable one and which, in turn, means that there is an extremely slow dynamics of vesicle amphiphiles within the aggregate and between the aggregate and the soluble monomeric form. The clearest evidence of this fact is the experimental observation of broad vesicle size distributions after vesicle preparation, indicating that several vesicle sizes coexist without collapsing into a single-species vesicle. Studies on self-reproduction have shown, however, that such different compartments under certain conditions may disclose a common reactive pattern, undertaking the growth–division path and therefore creating new particles starting from initial particles. Very likely, the detailed molecular and supramolecular mechanism of self-reproduction differs in the two cases (micelles/reverse micelles vs vesicles). In the last 17 years, supramolecular aggregates have been used to implement the so-called autopoietic self-reproduction, with consequent rise of interest in the possible use of such systems to build minimal autopoietic units, and consequently to model some important aspects of living systems. In many cases, this has been achieved using a very simple chemical setup, and therefore revealing how selfreproduction can actually be an accessible path for surfactant-based self-assembled particles. The general mechanism for self-reproduction of supramolecular structures as micelles, reverse micelles, and vesicles is shown in Fig. 2B. The scheme, closely resembling the one indicating the minimal autopoietic unit (Fig. 1B), involves the uptake of a precursor P and its transformation into the boundary-forming molecule S. The transformation may occur in the internal core of the structure or at its boundary. In any case, S, once formed, will spontaneously become part of the boundary, increasing the particle surface. This growing process brings the structure to an unstable state, which collapses to give rise to two or more new structures, having the same static and dynamic organization as the parent one. It is important to remark here that the event of division is a consequence of physicochemical restrains due to the nature of self-assembled structures, and does not derive from autopoietic theory. Most of the reported studies on autopoietic self-reproduction closely follow this scheme. They can differ in the nature and the form of the precursor P, and

228

P. Stano and P. L. Luisi

on the precise locus of its transformation into the boundary-forming molecule S (P ! S reaction may occur in the core or at the boundary). Also, with a notable exception [24], in all cases the catabolic process that leads to the consumption of S does not take place in the conditions of experiments. There are some consequences that stem from the general mechanism of Fig. 2B. The first is about the general stoichiometry of such process, which involves a nonlinear growth of the particle number. In fact, the increase in particle number follows a geometrical progression (1, 2, 4, 8, . . . 2N after N division), as in cellular division. Second, the process can be seen as an autocatalytic one, where a certain structure or particle ultimately catalyzes its own formation, according to the following general scheme (P has been substituted with S, for the sake of clarity):

M þ nS ! M0 þ M00

ð1Þ

RM þ nS ! RM0 þ RM00

ð2Þ

V þ nS ! V0 þ V00

ð3Þ

where M, RM, or V indicate micelle, reverse micelle, or vesicle, respectively, and the superscripts have been added to emphasize that in principle the products of a selfreproduction step are similar to, but may differ from, the parent structure. The above-mentioned equations simply state that new structures derive from preexisting ones, by addition of building blocks. In some instances, as we will see later, it has been shown that V  V0  V00 , that is, the daughter vesicles are similar to each other and similar to the vesicle ‘‘mother.’’ We can speak of autocatalysis since it has been also shown, for certain systems, that the spontaneous reaction ‘‘nP ! nS ! product’’ is significantly slower than the process where preformed structures are involved. In the following paragraphs, we will provide several experimental examples that can be understood using the above formalism. For detailed modeling, the reader should refer to more specific papers [16–18, 25–30]. In Table 1, the experimental strategies to accomplish self-reproduction are classified according to the type of supramolecular structure and the nature of precursors P. In all cases, if we discard the recent report from Sugawara and coworkers [31], fatty acids have been used as amphiphiles, since they readily selfassemble in different forms, depending on the solvent and on the pH of the solution. In fact, as will be shown later in detail, fatty acids form micelles or vesicles in aqueous phase, at high and low pH, respectively, and also depending on the chain length; furthermore, they form reverse micelles in apolar solvents. Thus, fatty acids, the building blocks of the autopoietic structure of Fig. 2B, have been formed starting with the corresponding salts by a process of protonation (Table 1, line 1), or by alkaline hydrolysis of fatty acid esters and anhydrides (lines 2 and 3), or through the oxidation of the corresponding long-chain primary alcohol (line 4). Sugawara et al., on the contrary, introduced a non-fatty acid selfreproducing system, where the building block S is an ad hoc designed bola amphiphile (line 5), formed by a condensation reaction between an aromatic aldehyde and an amine, to form a Schiff’s base [31].

229

Self-Reproduction of Micelles, Reverse Micelles, and Vesicles

Table 1

List of precursors used for the self-reproduction of supramolecular structures

No.

Precursor

Micellesa

Reverse micellesb

1

Fatty acid carboxylate in the form of micelles Fatty acid methyl or ethyl esters Fatty acid anhydrides Long-chain primary alcohol Ad hoc designed bola amphiphile





p

p

— p

— p





2 3 4 5

Vesiclesc,d

p

p p — pe

a

Formed by fatty acids in aqueous phase at high pH. Formed in apolar phase. Formed by fatty acids at intermediate pH. d Inclusive of giant vesicles (GVs). e pNon-fatty acid vesicles. : done. b c

From a theoretical viewpoint, a couple of specific comments are needed. As noted before, micelles, reverse micelles, and vesicles grow as a consequence of the relation vP  vD > 0; however, this relation does not explain self-reproduction. In a certain sense, self-reproduction can be seen as an emergent property of the aggregate that can be explained only when a new, hierarchical level is considered. Self-reproduction occurs through a growth–division mechanism, that is proper of the supramolecular aggregate (higher hierarchical level) and not of the surfactant molecules (lower hierarchical level). The two levels are of course functionally and dynamically connected, since changes in the concentration of assembled surfactant molecules brings about changes in the structure (and stability) of the aggregate (consider, e.g., the case of reverse micelles, see below). The detailed explanation of self-reproduction mechanisms has not been provided yet by the experimental studies that will be described in the following sections. The self-reproduction of micelles and reverse micelles, due to their nature of equilibrium systems, might be understood according to a simpler model; the self-reproduction of vesicles—in contrary—might need more complex rationalization. In fact, it is well known that micelle structural stability is determined by many factors such as the surfactant, the ionic strength, and the temperature. The size of a micelle, indicated by its aggregation number (N ), is determined by the physical parameters of the system. If a change in shape is not possible because of structural restrain, a micelle cannot increase its N without a departure of thermodynamic stability. Therefore, the fission of grown micelles, to produce more smaller micelles, is a way to restore the thermodynamically favored state. Similar arguments can be valid for reverse micelles, where the ratio w0 ¼ [water]/[surfactant] rules the stability profile of these particles. Since in a reverse micelle solution [water] ¼ constant, and a process of surfactant production will decrease the value of w0, a reverse micelle rearrangement is a way to restore the structural stability. Vesicle stability is generally explained in terms of the natural curvature of the membrane, and therefore in this case also one expects to have the most stable vesicle structure, maybe defined in terms of radius of curvature, and

230

P. Stano and P. L. Luisi

consequently in terms of aggregation number. However, experimental observations clearly indicate that spontaneously formed vesicles always have a relatively broad size distribution, and therefore different species coexist without a measurable rate of conversion to the most stable specie. It has been argued that all vesicles are thermodynamically unstable structures, the (almost) planar membrane being the low-energy state. The picture that emerges from these facts is that vesicles are often described as kinetically trapped structures. Consequently, it is expected that a process of growth brings the vesicle to a new state—the swelled vesicle, but no rearrangement is foreseeable. On the contrary, self-reproduction of vesicles does occur and therefore a more complex and detailed model must be considered to explain the experimental data. According to the most recent views, factors such as nonspherical growth and budding [31, 32], shell duplication [33], and hydraulic [32] as well as osmotic forces [17] are involved in vesicle self-reproduction. There is—however—no general agreement on these aspects, and the mechanism of vesicle self-reproduction is still considered unknown, also because it is strongly dependent on the chemical nature of the surfactant used. In the next paragraphs, we will give an overview on the published reports dealing with the self-reproduction of micelles and reverse micelles, whereas a detailed account will be provided for the corresponding vesicles’ work, introducing the issue of minimal cell construction.

3. Self-Reproduction of Micelles and Reverse Micelles The first report on the self-reproduction of supramolecular aggregates dates back to 1990 [34], where the autopoietic approach was applied for the first time to chemical systems. The theoretical framework was already sketched in a programmatic paper from Luisi and Varela [35], who, for the first time, proposed supramolecular aggregates like reverse micelles as candidates for implementing autopoietic self-reproduction. Reverse micelles are supramolecular structures that form spontaneously when certain surfactants are dispersed in an apolar phase (the ‘‘oil’’), in the presence of small amount of aqueous phase (e.g., 0.1–0.5% v/v). As shown in Fig. 2A, their architecture is characterized by a generally spherical shell of surfactant molecules, which expose the hydrophobic tail to the oil, whereas the hydrophilic head groups are faced toward the inner water-filled cavity of the reverse micelle. An important feature of reverse micelles is their monodispersity in terms of size, which means, in other terms, that they are stable only in one given dimension. This implies that a process of growth will bring about a thermodynamic stabilization, which then will favor a reassessment in the pristine size. Reverse micelles represent a facile system to achieve complete compartmentation. In fact, a reverse micelle solution is a microheterogeneous system where water-soluble substance will be compartmentalized in the so-called water-pool (the microdispersed aqueous phase), whereas hydrophobic compounds are solubilized in the oil phase.

231

Self-Reproduction of Micelles, Reverse Micelles, and Vesicles

Amphipatic molecules, in addition to the micelle-forming surfactants, might be at the oil–water interface. Reverse micelles meet several criteria for being potential autopoietic particles. A precursor P can be solubilized in oil, and the micelle can uptake it, transform it into the boundary-forming molecule S by means of a reaction catalyzed by internalized compounds. Three different approaches were used to produce the reverse micelles’ forming surfactant, which was in all cases octanoic acid (Fig. 3). In the first one [36], octanoic acid octyl ester was hydrolyzed to octanoate and octyl alcohol by LiOH, solubilized within reverse micelles (Fig. 3A). Using triglycerides as precursors of fatty acids, Luisi and coworkers [37] have also shown that lipase-containing reverse micelles can grow and divide by means of an enzymatic hydrolysis (Fig. 3B). Finally, the permanganate-based oxidation of 1-octanol to octanoic acid [37] also provides a route to autopoietic self-reproduction (Fig. 3C). Notice that in all cases, the reaction scheme closely follows the general autopoietic network indicated in Fig. 1B, where a precursor (P ¼ octanoic acid octyl ester, trioctylglycerol, and 1-octanol) is taken up by a self-bounded system (the reverse micelle) and transformed, within the boundary of the particle into the micelle-forming building block (S ¼ octanoic acid); a consequent enlargement of the surface follows, with destabilization and division. A second important point is that the reactants needed to accomplish the P ! S transformation are water-soluble compounds, so that the reaction must occur within the reverse micelles. No external reaction is possible, and the system is forced to behave autopoietically. In addition, since the total volume of water phase is fixed, a decrease in size is expected (and observed).

P S P S

A

S S S S

X

Y

S

Water S Oil S S S General scheme S

B

R C R⬘

O LiOH O −O

O

C R

HO R⬘

C

R C R⬘⬘

MnO4−

O Lipase O O

−O

C R

R CH2OH

O −O

C R

Figure 3 Approaches to the self-reproduction of reverse micelles. The general scheme (top-left) shows that a precursor P reaches the oil–water interface, enters into a reactive step that transforms it into the boundary-forming compound S, at the expenses of internalized water-soluble reactant X. In real examples (A–C), the transformation of P into S is (A) alkaline hydrolysis of carboxylic esters; (B) enzyme-catalyzed hydrolysis of carboxylic esters; and (C) permanganate oxidation of primary alcohols. In all cases, the free carboxylate is released, which increases the surface of the reverse micelle.

232

P. Stano and P. L. Luisi

These pioneristic studies were realized by the group of Luisi at the ETH in Zurich, who was specialized in studies on enzymes-containing reverse micelles. The know-how on micellar systems, joined by the interest on autopoiesis, was effectively employed to realize then the first example of chemical autopoiesis, and to open the way to more sophisticated studies on self-reproducing systems, emphasizing their relevance for the origins of life, as well as for basic research on selfassembly structures and their general reactivity. In a way, such studies are at the roots of current chemical- and molecular biology-oriented approaches to realize the first minimal autopoietic living system. These new aspects will be discussed in Paragraph 5. Also ‘‘normal’’ aqueous micelles have been studied within the context of autopoietic self-reproduction. Normal micelles, contrary to reverse micelles, form in aqueous media, and can be considered self-bounded hydrophobic compartments (Fig. 2A). Initial studies [37] aimed to achieve self-reproduction by using the permanganate oxidation described above. Octanoic acid micelles were loaded with 1-octanol, so that the precursor P in this case is embedded within the micelle. Permanganate, on the contrary, is water soluble, so that the two partners of the oxidation reaction (1-octanol and permanganate) are in two different phases. The reaction occurs at the micellar interface and the precursor is transformed into the micelle-forming compound S (octanoate). This approach was successful, and the occurrence of self-reproduction was demonstrated by measuring concentration of micelles by a fluorescence assay. After the total conversion of the precursor, the micelle concentration increases by a factor of 1.4. Next studies on micelle self-reproduction employed a different approach, with the aim of realizing the closest modeling of autopoietic growth (Fig. 2B). In fact, the setup of the previous case [37] involved the use of micelles already loaded with the surfactant precursor. On the contrary, a hydrophobic precursor could be taken up by micelles if it would be present in the surrounding. A new experimental setup was therefore designed as following: ethyl caprylate, a water-insoluble ester, is stratified over an alkaline solution, either in the absence or in the presence of preformed caprylate micelles. Let us start by illustrating the first case (absence of preformed micelles), since this approach represents the spontaneous formation of autocatalytic self-reproducing systems from scratch, basing only on physicochemical forces [38]. When ethyl caprylate is overlaid on an NaOH solution (Fig. 4), the hydrolysis to form caprylate and ethanol is very sluggish, taking place at the macroscopic interface between the ‘‘oil’’ and the aqueous solution. At the beginning, the concentration of caprylate in water increases slowly with time, and its value is below the caprylate critical aggregation concentration (c.a.c.) (in these conditions, c.a.c. 100 mM). Caprylate molecules accumulate in water phase without a significant macroscopic change, till the moment the c.a.c. is reached, and the first micelles form. This path represents the onset of micelles that would act as autopoietic units. Once micelles form, they can uptake ethyl caprylate molecules, and solubilize them in their hydrophobic core. Hydrolysis, in this case, occurs at the micellar interface, and new caprylate molecules form, bringing forth the micelle growth and division. Since more micelles form, more ethyl caprylate is taken up and hydrolyzed and so on, establishing an autocatalytic autopoietic self-reproduction. The caprylate concentration versus time profile clearly shows a sharply sigmoidal shape (Fig. 4), indicating

233

Self-Reproduction of Micelles, Reverse Micelles, and Vesicles

Ester

Monomer

Monomer

NaOH

Caprylate (M)

1.5

1.0

0.5

0.0

0

10

20

30 Time (h)

40

50

60

Figure 4 Autocatalytic self-reproduction of micelles. Caprylate ester, stratified on alkaline solution, is slowly hydrolyzed at the water–ester interface, producing caprylate. In the initial phase, the caprylate concentration is low, below its critical micelle concentration (c.m.c.), so that it is solubilized in aqueous phase as a monomer. As soon the c.m.c. is reached, the first micelles form. Additional caprylate ester can be solubilized into micelles, hydrolyzed at the micelle–water interface, forming a swelled micelle, which divides into new micelles. In turn, new micelles can solubilize more ester, grow and divide, and so on, in a geometrical progression. On the bottom, the time course of caprylate concentration in aqueous phase is shown, revealing a strong sigmoidal shape, which suggests the autocatalytic mechanism. c.m.c. is reached after 32 h. Adapted from Bachmann et al. [38].

an autocatalytic mechanism. In particular, in the first 30 h, caprylate concentration increases slowly, then suddenly increases to give rise to a sort of step function, deriving from a geometrical progression of micelle concentration (1 ! 2 ! 4 ! . . .) with a typical ‘‘explosive’’ pattern. The overall rate of ethyl caprylate hydrolysis increases by a factor of 900. Once all ethyl caprylate is converted into caprylate, the process stops, and the system becomes homogeneous. If micelles are present from the beginning, the lag phase decreases or disappears. This experiment has a great relevance in the field of self-reproduction of supramolecular aggregates. In fact, not only it is a demonstration that normal micelles might act according to the autopoietic scheme of Fig. 2B but it also suggests how an autopoietic unit can emerge from simpler components, driven by physicochemical forces only (in this case self-assembly). In a sense, this is a model for a stepwise increase in complexity (molecules, aggregates, dynamical process within aggregates, population of aggregates, dynamical processes within an aggregate population) that could be suggested as a typical out-of-equilibrium self-organization pattern. Self-assembled structures interact with their environment, establish relations

234

P. Stano and P. L. Luisi

of chemical transformations, and produce themselves, all in agreement with an autopoietic organization (Fig. 1B). An interesting, and still debated, aspect involved in this experiment is the discussion of the role of hydroxide ions for the establishment of the autopoietic mechanism depicted above. In fact, hydroxide ions (and their sodium counterparts) are needed to accomplish the conversion of the precursor P into the micelle-forming component S at the interface between the autopoietic micelle and the environment. Although it is true that NaOH is a component of the environment, the micelle-associated ions (in particular the hydroxide ions) are reasonably the true reactants for this reaction. According to this view, the physical (and chemical) space of the autopoietic unit—the micelle—although not well defined, represents the locus of the P ! S transformation. We will see that some modern approaches to autopoietic systems try to avoid this vagueness, which stems from our unawareness of molecular dynamics, and however is pertinent to the theoretical framework of autopoiesis, not limiting at all the concrete aspects of self-reproduction processes. A short remark must be made on the word ‘‘catalysis’’ when applied to these systems. In fact, it has been argued [39, 40] that this term cannot be used properly in such context, since it generally refers to the decrease in activation energy of key reactions. In this context, the ‘‘catalytic’’ effect comes from the tremendous increase in microscopic interface, which is a sort of ‘‘physical catalysis.’’ The use of micelles (and reverse micelles) as model autopoietic systems was soon overcome by the interest in vesicle self-reproduction. Quite recently, however, micelle-based systems have again attracted the interest of researches in Los Alamos (USA) and in Europe, who aim at the development of the so-called ‘‘Los Alamos Bug’’ [41, 42], a chemical system embedded in micelles which should perform a series of reactions leading to the self-reproduction of the whole structure. Moreover, new studies [43] have been recently published on the self-reproduction of waterin-oil microdroplets, with the aim of investigating the division of such reverse micelles-like structure. Fluorescence microscopy allows real-time observation of the droplets’ division, also in the presence of internalized components, engaged in complex biochemical reactions, as protein expression. To summarize, the studies on self-reproduction of supramolecular aggregates started with reverse micelles and soon after with micelles, as indicated in Table 2. These systems are characterized by a fast molecular dynamics and are generally considered equilibrium systems; as a consequence, micellar growth is followed by a spontaneous rearrangement (i.e., formation of new micelles) since large micelles are not stable. Complete compartmentation of water-soluble compounds is achieved by using reverse micelles, which are, however, suspended in an apolar medium. By contrast, micelles, which form in water, do not have aqueous compartments, and are therefore not valuable as cell-like structures. Reverse micelles and micelles, thus, are relatively poor cellular models when compared with vesicles, and further research was aimed at extending these early results to vesicles. It should be added, however, that although aqueous micelles themselves are not good cell models, early studies on self-reproduction have also revealed how fatty acid micelles (in particular oleate micelles) can be seen and used as precursors of vesicles [44]. In fact, if oleic anhydride is hydrolyzed in alkaline unbuffered solution,

235

Self-Reproduction of Micelles, Reverse Micelles, and Vesicles

Table 2 List of experimental articles on micelles and reverse micelles self-reproduction No.

Year

Description

Reference

1

1990

[34]

2

1991

3

1991

4

1991

5

1991

6

1992

Octanoate reverse micelle self-reproduction by LiOH-catalyzed octanoate ester hydrolysis Octanoate reverse micelle self-reproduction by LiOH-catalyzed octanoate ester hydrolysis Octanoate reverse micelle self-reproduction via the lipase-catalyzed trioctylglycerol hydrolysis Octanoate reverse micelle self-reproduction by permanganate oxidation of 1-octanol Octanoate micelles (in water) self-reproduction via permanganate oxidation of 1-octanol solubilized in the micelle Autocatalytic caprylate micelle self-reproduction by alkaline hydrolysis of ethyl caprilate

[36] [37] [37] [37]

[38]

a pH drop accompanies the hydrolytic step, so that the product of the reaction (fatty acid) initially self-assembles as micelles. When the pH of the aqueous phase reaches lower values (between 7 and 9.5), the oleate micelles spontaneously transform into oleate vesicles (see Paragraph 4). Both kind of aggregates (micelles and vesicles) self-reproduce in their own range of existence, but the system evolves spontaneously to the vesicle state. In such way, the primary chemical act (anhydride hydrolysis) induces first the formation of micelles and their self-reproduction, and then the consequent pH-dependent transformation of micelles into vesicles, and their self-reproduction.

4. Self-Reproduction of Vesicles The first studies on self-reproduction of vesicles were carried out immediately after the results achieved with reverse micelles and later on with aqueous micelles. In this section, we will discuss first the work done on ‘‘small’’ vesicles (50–200 nm), and then on the so-called giant vesicles (>1 mm), always referring to the general selfreproduction principles indicated in Fig. 2B, that is, the addition of a molecular precursor to vesicles, its transformation into the bilayer-forming compound, and the enlargement of the vesicle, with possible division. It must be emphasized that— in contrast to micelles and reverse micelles—it seems more reasonable to invoke a complex mechanism for the self-reproduction of vesicles, since these systems are generally described as kinetically trapped ones, that is, a fast equilibrium with the most stable vesicle state is not attained. Our actual lack of knowledge on the selfreproduction mechanism of vesicles is witnessed by the not-unified view of the process, and by the absence of unifying theoretical models. With the recent rising

236

P. Stano and P. L. Luisi

interest in such kind of phenomena, more detailed explanations will certainly appear in the next few years.

4.1. ‘‘Small’’ Vesicles In this chapter, we will refer to vesicles with size below 200 nm as ‘‘small’’ vesicles, diverging a bit from classical nomenclature, where the terms ‘‘small unilamellar vesicles (SUV)’’ define quite precisely a class of vesicles having size below 50 nm. In our case, ‘‘small’’ is used in contrast to ‘‘giant,’’ since we can classify the work on self-reproduction according to the size of vesicles. Table 3 summarizes the experimental papers on the self-reproduction of vesicles. With the exception of a study done on cationic systems [45], all works on the reactivity of small vesicles have been carried out with anionic vesicles, and more in particular with fatty acid vesicles and mixed vesicles (phospholipids/fatty acids). There are at least two reasons for this choice. The first is related to the properties of fatty acid vesicles. These are rather dynamical systems when compared with phospholipids. Fatty acid vesicles have a relatively high c.a.c. and differ in many aspects (permeability, stability, interaction with other lipids) from classical phospholipid vesicles. These facts, per se, would indicate fatty acids as convenient models for Table 3

List of experimental articles on vesicles and giant vesicles self-reproduction

No.

Year

Description

Reference

1

1991

[110]

2

1993

3

1994

4

1994

5

1994

6

1994

7

1995

8

1998

9

1999

Reconstitution of four enzymes in lipid vesicles with the aim of synthesizing lipids from within Attempts to self-reproduce oleate vesicles by the lipasecatalyzed hydrolysis of ethyl oleate Attempts to self-reproduce oleate vesicles by the lipasecatalyzed hydrolysis of ethyl oleate Self-reproduction of oleate vesicles by alkaline hydrolysis of oleic anhydride (and simultaneous internalized polymerization of ADP to poly(A)) Self-reproduction of oleate vesicles by alkaline hydrolysis of oleic anhydride (and simultaneous RNA replication inside) Self-reproduction of oleate vesicles by alkaline hydrolysis of oleic anhydride Self-reproduction of oleate vesicles by alkaline hydrolysis of oleic anhydride (and simultaneous RNA replication inside) Self-reproduction of oleate vesicles by oleate micelle-tovesicle conversion Self-reproduction of POPC and oleate vesicles by oleate micelle-to-vesicle conversion

ADP, adenosine diphosphate; POPC, 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphatidylcholine.

[59] [60] [64]

[61]

[62] [63]

[65] [66]

237

Self-Reproduction of Micelles, Reverse Micelles, and Vesicles

studying vesicle reactivity. In addition to this aspect, there is a second—and maybe more relevant—feature that capitalizes the researcher attention on these systems. In fact, fatty acid vesicles are, to date, the most plausible candidate to play the role of prebiotic compartments. Pioneering works of David Deamer first [46, 47], and our group later, together with modern works by several scientists, share the same interest in protocell origin, stability, and dynamics, exploring the possible role of vesicles in the origin of cellular life. Fatty acid vesicles were described for the first time by Gebicki and Hicks [48], and it was soon recognized that such structures could have had a central role in the origin of cellular life. Monocarboxylic acids (C2-C12) have been isolated from the Murchison meteorite [49], pointing out their possible synthesis in abiotic conditions. In addition, it has been shown that fatty acids can be synthesized in allegedly prebiotic conditions [50–52]. The interested reader can find more details in recently published reviews, where the roles of fatty acids in the origin of protocells, as well as other prebiotic surfactants, are discussed [53–55]. To understand the chemistry of fatty acid assemblies, their ionization pattern must be examined first. This interpretative model is currently used to describe the behavior of fatty acids in aqueous solution. However, it must be recognized that this accounts only partially for the physicochemical properties of such structures. The effect of counterions, cosolvents, cosurfactants, temperature, and concentration must be considered in order to describe the system completely. With these limitations in mind, let us proceed with the simplified description of fatty acid systems according to Fig. 5. One of the main parameters governing fatty acid association in aqueous system is the ionization degree. It has been shown experimentally [48] that fatty acid bilayers can be formed only at the intermediate pH, where the carboxylate and the acid form coexist. Moving toward the two extremes, one finds at low pH the formation of oil droplets composed by completely protonated fatty acids, and at high pH the micellar form of fatty acid salts (often indicated as soaps).

A

High pH (> 10) O O

Conical shape

B

Intermediate pH (7.5

9.5) O O

H O

Cylindrical shape

O

Figure 5 Simplified geometrical models of oleic acid molecules at different pH. At high pH (A), oleate molecules form micelles, as expected from conical-shaped surfactants; at intermediate pH (B), a dimeric structure is formed that can form bilayers and therefore vesicles. At lower pH, oleic acid separates from the solution as oil droplets (not shown).

238

P. Stano and P. L. Luisi

As shown in Fig. 5, carboxylate anions can bind the carboxylic form through hydrogen bonding, forming a sort of dimer, which avoids head group repulsion and facilitates the formation of a cylinder-like unit. According to the classical treatment of surfactant parameter [56], a bilayer architecture becomes possible for cylindrical-shaped surfactants. It is interesting to remark that the simultaneous and almost equimolar presence of the acid and basic form of carboxylic acids is possible at pH  pKa, which means that the pKa of self-assembled fatty acids (from 7 to 9.6) is significantly higher than that of water-soluble (and monomeric) acids, for example, acetic acid (4.7). This behavior has been explained by considering that polyanionic surfaces such as fatty acid layers have a local pH well below the bulk pH (3 units) [57]. This electrostatic effect explains why fatty acids begin to be protonated at pH < 9, and why they are fully protonated at pH < 7. At high pH, ionization is complete and fatty acid salts aggregate in the form of micelles. This attitude has been rationalized also in terms of surfactant parameter, as indicated by the conical shape overemphasized in Fig. 5. In the first approximation, therefore, one can simply state that fatty acids form two different kinds of ordered aggregates, depending on the pH of the solution. At intermediate pH (generally between 7 and 9), a lamellar state is built, whereas at high pH values, micelles form. Recent reports, however, tend to reevaluate this simple relation between pH-morphology, by measuring—via electron spin resonance—the coexistence of mixed systems, for example, vesicles and other micelle-like forms, at intermediate pH [58]. Despite these complications, which are currently investigated, many of the basic properties of fatty acid vesicles, especially with respect to their reactivity in general, and self-reproduction in particular, can be explained in terms of the simplified view shown above. Fatty acid vesicles have a primary role in the field of self-reproduction of supramolecular structures. In fact, in addition to their significance in the origin of life studies, they offer noteworthy advantages from the practical viewpoint. The experiments that we will describe in the following paragraphs were possible since a facile way of forming fatty acids is practicable. Anhydride and ester hydrolysis has been implemented in the beginning, achieving the self-reproduction of fatty acid vesicles in heterogeneous systems, like in the case of micelles shown above [38]. A second way to form fatty acid vesicles involves the micelles-to-vesicles transformation, driven by a jump of pH; in fact, fatty acid micelles have been used as precursors of vesicles. Water-insoluble precursors, such as fatty acid ethyl and methyl esters, as well as fatty acid anhydrides, have been successfully employed in the first studies, dated in the early 1990s. For example, initial efforts were devoted to the lipase-catalyzed hydrolysis of ethyl oleate (the precursor), to give oleate (the boundary-forming molecule), which binds to and enlarges the surface of oleate vesicles, thus bringing about vesicle growth and division [59, 60]. Since lipase was entrapped inside preformed vesicles, the experimental setup matched very closely the ideal model of Fig. 2B. In fact, the mechanism proposed for the system dynamics involves the initial uptake of ethyl oleate by oleate vesicles, driven by chemical affinity of the water-insoluble ester for the hydrophobic oleate membrane. The process was rather

Self-Reproduction of Micelles, Reverse Micelles, and Vesicles

239

slow, being complete in 200 h, and a clear demonstration of growth–division steps was not provided yet. The shift from fatty acid esters to fatty acid anhydrides was realized soon after [61–64], with some advantages, as judged by the extensive studies focused on this system. The transformation of the precursor was not any longer enzymatic, as in the previously reported examples, but chemical, accomplished by carrying out the reaction in alkaline media (mainly at pH 8.5). This shift has an important consequence: the reactive event of the P ! S conversion occurs at the interface between the membrane and the solution (see Figs. 1B and 2B), therefore, may be still considered within the vesicle. However, one specie present in the solution (i.e., the hydroxide ion) is also involved, and therefore the transformation of precursors is accomplished by a component that is dynamically bound to the surface (the hydroxide ion). The considerations discussed in Paragraph 3 hold also in this case. It is clear that the close adherence of above-mentioned cases (reverse micelles, enzymes inside vesicles) to the general scheme of Fig. 2B is partially reduced, since the agent that transforms the precursor P into the membrane-forming compound S is also present in the environment. About this point, it must be said that although direct hydrolysis of the sparingly soluble anhydride might in principle take place, membrane processes are dominant, for the ‘‘catalytic’’ reasons mentioned above. In the work referred above [61–63], the formation of oleic acid vesicles from oleic anhydride follows a different path depending on whether preformed vesicles are available or not, confirming that the process is autocatalytic, in the sense that vesicles ‘‘catalyze’’ their own formation. This evidence was obtained by electron microscopy and chemical analysis of the reactive mixture. The next stage of investigation involved a new strategy to self-reproduction, where the heterogeneous system, based on oleic anhydride/oleate vesicles, is replaced by a homogeneous one, based on the spontaneous micelles-to-vesicles transition, typical of fatty acids. In fact, the anhydride-based system may hinder real-time spectroscopic analysis such as spectroturbidimetry, fluorescence measurements, and light scattering. Conversely, the micelles-to-vesicles transition occurs in single (microheterogeneous) aqueous phase, and offers some practical advantages. In this approach, a small aliquot of concentrated oleate micelles (pH > 10) is added to a buffered solution (pH 8.5), containing—or not—preformed oleate vesicles. It should be noticed that in this case the transformation P ! S (see Fig. 2B) does not involve the breaking of strong covalent bonds, as in the case of ester or anhydride hydrolysis. The process consists of the reassessment of the protonation state of oleate molecules, with the formation of protonated carboxylic group and the consequent rise of H-bonded structures as in Fig. 5. These steps might take place in solution, independently from the preformed vesicles, since the ‘‘agent’’ which causes the transformation does not reside inside (or at the boundary of ) vesicles. On the contrary, the experiments show with little doubts that the presence of preformed vesicles does indeed affect the micelles-to-vesicles transition, and that, according to the current interpretations, preformed vesicles exert their function of directing the self-assembly of new vesicles by taking up the added surfactant and concur to its transformation.

240

P. Stano and P. L. Luisi

4.2. The ‘‘Matrix’’ Effect The use of oleate micelles instead of oleic anhydride, as surfactant precursors, was introduced in 1998 by Bloechliger et al. [65], discovering that preformed vesicles accelerate the rate of formation of new vesicles. A sigmoidal turbidity-time profile suggests that an autocatalytic mechanism operates in the system, and electron microscopy analysis indicates—surprisingly—that an unexpected homogeneous size distribution is obtained. This is in contrast to the spontaneous micelles-tovesicles transformation of oleate micelles, when carried out in the absence of preformed vesicles: a broad size distribution is in fact generally obtained (Fig. 6). Turbidimetric kinetic analysis, a very sensitive technique that is important to follow vesicle reactivity, made possible the identification of the autocatalytic mechanism. Dynamic light scattering (DLS) and electron microscopy analysis were also carried out, to structurally analyze the vesicles. The authors called the simultaneous occurrence of (1) the enhancement of vesicle formation rate by the preformed vesicle and

pH 8.5

A Oleate micelle (pH > 10)

pH 8.5 Preformed oleate vesicles V*

Oleate vesicles

1

Oleate vesicles

2

B C

1 Number distribution, %

Micelles or anhydride

V*

2 Vesicles 0.0

0.1

0.2

0.3

0.4 0.5 0.6 Diameter, µm

0.7

0.8

0.9

1.0

Figure 6 Self-reproduction of oleate vesicles and the matrix effect. Experimental setup (A) oleate micelles are added to buffer solution (1) or to a suspension of preformed oleate vesicles V* (2). Electron microscopy analysis (B) shows that whereas in the first case, micelles spontaneously transform into vesicles, producing a population with a broad size distribution (1), in the second case, when preformed vesicles having a narrow size distribution are present, micelles form vesicles having size distribution very close to the size distribution of preformed ones (2). This has been called ‘‘a matrix effect’’; self-reproduction of vesicle occurs with approximately constant size; therefore, an induction mechanism must be present (C). Adapted from Bloechliger et al. [65].

241

Self-Reproduction of Micelles, Reverse Micelles, and Vesicles

(2) the fact that newly formed vesicles have the same size of preformed ones, a ‘‘matrix effect,’’ that is, a sort of size-controlling effect exerted by the preexisting vesicles on the formation of new vesicles. In order to explain this behavior, vesicles should enter in the process of their own assembly, as emphasized by the dashed arrow of Fig. 6C. As a consequence of this striking discovery, a large amount of work has been done in the following years, by the group of Luisi and by others, as will be shown later. Lonchin et al. [66] extended the previous observation of oleate micelles/oleate vesicles system to a mixed system, where oleate micelles were added to preformed phospholipid vesicles (made of POPC ¼ 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphatidylcholine), demonstrating that a ‘‘matrix effect’’ is even present at 1:1000 POPC/oleate molar ratio, suggesting a very strong interaction between preformed vesicles and freshly added oleate. This very important outcome is still far to be understood; although—as will be shown below—many mechanistic studies focus on the equimolar addition, or on small excess addition, not much work has been dedicated to explain the ‘‘1:1000 effect.’’ In Fig. 7, we report some of the most significant findings of such study. Turbidity versus time profiles and DLS analysis both suggest the existence of a mechanism A

B

Relative turbidity

DLS mean radius, nm

200

0

5

10

20 50 Time, min

80

110 140

150

100

50

0 Init

1:1

2:1

4:1 Control

C

Oleate micelle

SLOW Oleate vesicles

Preformed oleate vesicles

FAST

An intermediate

Figure 7 Self-reproduction of vesicles and the matrix effect. Further studies on micelles-tovesicles transformation indicate that preformed vesicles increase the rate of vesicle formation (A). Size of newly formed vesicles resembles the size of preformed (initial) vesicles, also when oleate micelles are added in excess (e.g., micelles/vesicles 4/1); control experiments (addition of micelles to buffer solution) show that oleate micelles form vesicles with larger sizes (B). The mechanism shown in (C) has been proposed: oleate micelles slowly transform spontaneously into oleate vesicles, whereas when preformed oleate vesicles are present, a fast pathway is accessible (oleate uptake, growth, and division). Adapted from Lonchin et al. [66].

242

P. Stano and P. L. Luisi

where vesicles uptake oleate (in the form of monomer or of micelles) and give rise to an intermediate (a swelled vesicle?) that eventually divides to give rise to new vesicles. The latter have a mean size which coincides with the size distribution of initial (preformed) vesicles. According to this view, preformed vesicles are involved in the rearrangement of oleate micelles to vesicles, actively incorporating the ‘‘precursor’’ and so increasing their surface, and also divide after the initial growth step so that the size is conserved among generation of vesicles. In order to investigate this process, in recent years, there have been several mechanistic studies, based on different techniques like electron microscopy, DLS, turbidity, fluorescence, stopped-flow, and chromatographic separations, as well as theoretical modeling (see Table 4). The mathematical modeling of selfreproducing vesicles has attracted, very recently, an increasing number of scientists,

Table 4

Further studies of vesicle self-reproduction mechanism

No.

Year

Description

Reference

1

2001

[71]

2

2001

3

2003

4

2003

5

2004

6

2004

7

2004

8

2004

9

2005

10

2005

11

2006

12

2006

Cryo-TEM study of ferritin-containing POPC and oleate vesicles self-reproduction after addition of fresh oleate in the form of micelles and anhydride, respectively Cryo-TEM study of the ‘‘matrix effect’’ in POPC/oleate vesicles self-reproduction DLS and turbidity study on oleate vesicle self-reproduction Turbidimetric study on size-dependent self-reproduction rate in oleate vesicles DLS and turbidity study on oleate vesicle self-reproduction Demonstration of vesicle number duplication via DLS size distribution analysis Vesicle self-reproduction study via vesicle SEC separation and analysis Stopped-flow fluorescence study of oleate vesicle self-reproduction Vesicle self-reproduction study via vesicle SEC separation and analysis Vesicle self-reproduction study via vesicle SEC separation and analysis Stopped-flow turbidimetric study of vesicle self-reproduction ffEM study of oleate vesicle self-reproduction

[72] [73] [75] [74] [76] [77] [33] [78] [79] [82] [83]

cryo-TEM, cryo-transmission electron microscopy; POPC, 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphatidylcholine; DLS, dynamic light scattering; SEC, size exclusion chromatography (gel filtration chromatography); ffEM, freeze–fracture electron microscopy.

Self-Reproduction of Micelles, Reverse Micelles, and Vesicles

243

mainly from artificial life and complexity fields, and it will not be discussed here (if interested, see references [17, 18, 67–70]). The classical approach to investigate mechanisms involves the study of the relation between initial and final states, and the identification of the intermediate states. These observations, together with the knowledge of surfactant behavior, might allow a deeper understanding of vesicle self-reproduction. In a couple of 2001 papers, again from the group of Luisi [71, 72], the strategy was to label initial vesicles with water-soluble compounds, and to follow the distribution of such probes after the self-reproduction. Vesicles from POPC, as well as oleate, were loaded with ferritin, an iron-rich protein (more than 4000 Fe/ ferritin molecule) that can be easily visualized by electron microscopy. Oleic anhydride or oleate micelles were added to preformed ferritin-containing vesicles, so that the vesicle could grow and divide. The initial state (preformed vesicles) and the final one (product of the reaction) were carefully analyzed by means of cryotransmission electron microscopy (cryo-TEM), so that vesicles were characterized by their size and ferritin content. The aim of the experiment was to follow the size distribution of vesicles, and the relation of the latter with the ferritin numerical distribution (i.e., average number of ferritin molecules per vesicle). In fact, with the hypothesis that ferritin does not escape vesicles during the self-reproduction, the above-mentioned measurements provide a powerful method to indirectly follow the mechanism of self-reproduction. Two limit processes can be depicted theoretically (Fig. 8): (a) the formation of new vesicles follows an independent path, without interaction between vesicle-forming molecules (oleate coming from oleic anhydride hydrolysis or oleate micelle pH-induced rearrangement); (b) there is an interaction, with possible uptake of membranogenic precursor by preexisting vesicles which produce their surface growth. In order for the self-reproduction to take place, a division step must follow the growth step. The average number of trapped ferritin molecules does not change only in case (a) where the new vesicles do not contain ferritin molecules, in contrast to preformed vesicles, which keep their ferritin inside. In case (b), the average number of trapped ferritin is maintained only if a division step does not follow the growth. On the contrary, if vesicles grow, the size growth can be easily visualized by cryo-TEM. Therefore, in a true self-reproduction process, a decreased number of trapped ferritin per vesicle is expected; if matrix effect operates, the average size of vesicles is conserved. When oleic anhydride is added to oleate vesicles (5/1), vesicle growth and de novo formation of vesicles are the most abundant processes. In contrary, when oleate micelles are added to POPC vesicles (25/1), some small empty vesicles are found, together with ferritin-containing vesicles, and there is a clear reduction in the average number of ferritin molecules per vesicle, indicating that vesicles divide and ferritin is distributed among daughter vesicles. This study confirmed the previously DLS-based result of constancy of size (the newly formed vesicles have similar size to initial ones). The turbidity and DLS were used again some years later, when Rasi et al. [73, 74] demonstrated that the matrix effect described above is also size independent, in the sense that preformed vesicles induce the formation of new vesicles (from oleate micelles) of size similar to their one, so that the self-reproduction of 50 nm vesicles

244

P. Stano and P. L. Luisi

A

Precursor (oleate micelles or oleic anhydride)

Preformed vesicles

New vesicles formed independently (de novo)

B

Precursor (oleate micelles or oleic anhydride)

Preformed vesicles

New vesicles formed after interaction with pre-formed ones

Figure 8 Ferritin-containing oleate vesicles reveal details of self-reproduction mechanism. When oleate micelles are added to oleate vesicles, two possible paths can be undertaken. In (A), the independent vesicle formation by micelles is shown. In this case, new vesicles do not contain ferritin molecules, and have a broad size distribution. The average number of ferritin molecules per vesicle does not change. In (B), the added micelles interact with preformed vesicles, the first step being the uptake of oleate into preformed vesicles. If the swelled vesicle divides into more new vesicles, the average number of ferritin molecules per vesicle must decrease, and if ‘‘empty’’ vesicles are found, their formation can derive from asymmetric division or from spontaneous vesicle formation. The size distribution of ‘‘filled’’ vesicles allows the distinction between simple growth and growth–division processes. Adapted from Berclaz et al. [71, 72].

produces vesicles of 50 nm, whereas 100 nm vesicles self-reproduce to give vesicles of 100 nm. In this work, Rasi et al. also introduced a new treatment of DLS data, which involves the manipulation of average size and polydispersity index (two parameters obtained by the so-called cumulant analysis). By means of this treatment, the authors were able to show that 50 nm oleate vesicles have a higher tendency to grow than do larger ones (100 nm). If oleate micelles, on the contrary, are added to preformed POPC vesicles (50 and 100 nm), the tendency to growth–division is higher. These results were successively expanded by the report of Cheng and Luisi [75], who have showed that 135 nm vesicles undergo the process of self-reproduction faster than do smaller ones (65 nm), by a rate factor of 3. DLS was again the preferred analytical method [76], to confirm that selfreproduction mainly involves simple division of grown intermediate. In fact, it has been introduced as a method for extracting number-weighted size distribution data from raw DLS results, within the so-called Rayleigh-Gans-Debye approximation. It was demonstrated that during a classical self-reproduction experiment—when matrix effect is operative—the number of vesicles doubles when an equimolar amount of fresh surfactant is added to preformed vesicles (1:1 addition).

Self-Reproduction of Micelles, Reverse Micelles, and Vesicles

245

In a series of papers, Ueno and coworkers investigated the effect of preformed lecithin vesicles on the vesicle formation from oleate micelles [77–79]. The results, achieved by separating vesicles of different size by size exclusion chromatography (SEC), indicate that new small vesicles form in addition to vesicles with the same size as that of the initial ones. SEC profiles on samples obtained by using large (100 nm) preformed vesicles show that inferences achieved by analyses done on small number of vesicles (the cryo-TEM study of Berclaz et al. [71, 72]) are reasonable. When small preformed vesicles were used, the final SEC-determined size distribution overlaps with the initial one. The authors were also able to determine the chemical composition of vesicle classes, confirming that final vesicles are actually mixed oleate/ lecithin vesicles. This was somehow expected on the basis of previous data on fatty acids uptake by lecithin vesicles [80, 81]. From this study, the authors concluded that a second mechanism, in addition to growth–division, is operative when oleate micelles are added to preformed lecithin vesicles. In this second mechanism, preformed vesicles are solubilized by oleate micelles, which act as a detergent (to give mixed micelles). However, the final state reached by this second process is again the vesicular one. Kinetic studies on the mechanism of growth–division of vesicles have also been recently carried out. In their detailed investigation, Chen and Szostak, [33] as well as Walde and Robinson [82], have employed stopped-flow techniques to shed light on the very first steps involved in the oleate micelles-to-vesicles transformation, in the absence and in the presence of preformed vesicles. The report of Chen and Szostak is based on a stopped-flow fluorescence study of fatty acid vesicle growth after micelles addition [33]. Fluorescent dyes are incorporated into preformed oleate vesicles, and the decrease in FRET signal is followed. The dilution of dyes is ascribed to vesicle growth, which follows a single exponential profile for low micelles-to-vesicles ratio ( 0,’’ that is, growth. The reasons for this choice are related to the need of understanding

248

P. Stano and P. L. Luisi

how compartments can grow in size and eventually divide, so that the number of particles increases, very much like biological cellular growth. There is, however, a second aspect that deserves modeling, since it shares with self-reproduction a key role in living beings: this is homeostasis. In contrast with self-reproduction, there is only an isolated report on homeostasis, in which fatty acid vesicles were used again as model systems. The work of Zepik et al. [24] differs from all other investigation because—for the first time—the rates of surfactant production (vP) and surfactant destruction (vD) have been varied, whereas in other cases the relation ‘‘vP  vD > 0’’ was always true, in particular vP > 0 and vD ¼ 0. By combining two competitive reactions, that is, formation of surfactant and its degradation, it has been shown that vesicles can sustain homeostasis. Moreover, the general behaviors depicted in Fig. 1B have been implemented by fine-tuning of the two reaction rates vP and vD, so that the three patterns (growth, homeostasis, collapse) can be experimentally observed. This has been made possible by feeding a vesicle population with the reactants for the anabolic and the catabolic reactions at different rates. The two reactions of choice were the oleic anhydride alkaline hydrolysis that brings about the formation of oleic acid (the anabolic process), and the osmium tetroxide/potassium ferricyanide oxidation of oleic acid that brings about the formation of 9,10-dihydroxyoleic acid (the catabolic process).

4.5. Giant Vesicles The term ‘‘giant’’ vesicles (GVs) applies to large vesicles, sometimes unilamellar, with size well above the submicrometric scale, and typically in the range 10–200 mm. Thanks to their large dimension, GVs are easily visualized by optical microscopy, and their formation, stability, and reactivity can be followed in real time by direct observation. This advantage is partially counterbalanced by their preparation methods, which generally require a fine control of experimental conditions and by the heterogeneous nature of many specimens. GVs require special methods of preparation, for example, electroformation [85], controlled swelling [86, 87], and others [87–89]. The number of studies on GVs self-reproduction is limited when compared with conventional vesicles. The first attempt to self-reproduce GVs was reported in 1994–1995 by Wick et al. [90, 91], who implemented the classical approach (oleate GV þ oleic anhydride). Contrary to small vesicle dynamics, which remains largely elusive, the experiments reported by Wick et al. evidence possible pathways of the self-reproduction of vesicles. It is fair to use the adjective ‘‘possible,’’ since such mechanisms, operating on GVs, can differ from those of small vesicles. In fact, small and giant vesicles differ in several aspects, one of the principal being membrane curvature and the consequent different membrane behavior. When oleic anhydride is added to oleate GVs in basic solution, oleate molecules, produced by the hydrolytic step, will insert themselves in the GV membrane, increasing its surface. Optical microscopy analysis reveals that—although with difficult reproducibility—two processes may take place in GVs systems. They consist, respectively, of (1) asymmetric budding followed by division into two vesicles; (2) formation of inclusion vesicles, that is, a vesicle inside another, followed by vesicle translocation. This second mechanism is analogous to what is described by Menger [92].

249

Self-Reproduction of Micelles, Reverse Micelles, and Vesicles

There is only one more study on GVs self-reproduction, reported recently by Sugawara and coworkers [31, 93]. In this new approach, fatty acids are substituted by another surfactant that is specially designed to carry out transformations at molecular and colloidal levels. The membrane is composed by a surfactant S, whose precursors are two building blocks A and B (Fig. 11). This is the first divergence from all the above-mentioned approaches, where the P ! S transformation involves bond breakage and not bond formation. The precursor A is added in a protected A0 form, so that it can react only inside GVs, where a catalyst C is present in order to accomplish the A0 ! A transformation. On the contrary, B is already present inside GVs, in the form of microscopic oil droplets. According to the data presented by Sugawara [31], A0 enters into vesicles, where it is transformed into the active form A, which, in turn, reacts with B (present inside GVs), to give S. As a consequence, the GV grows because of the increase in membrane surface. The actual observation indicates the formation of internalized vesicles (inclusion vesicle) that are expelled outside. The system designed by Sugawara is a successful example in the area of complex supramolecular systems, where two different levels (microscopic and macroscopic) are dynamically connected, and give rise to system properties such as self-reproduction. Notice that A0 , the ‘‘nutrient’’ of the vesicles, is metabolized inside the compartment by using internal components (B, C) and concur to the formation of new vesicles. The surfactant S is a bola amphiphile, formed by two single-chain surfactants connected tail-to-tail by a Schiff base functionality (A is an aldehyde, B an amine).

A⬘ A⬘ C C S

B S

A C

S S

B C

S S

S

S

S S

S S

Figure 11 Giant vesicles composed by bola amphiphiles may undergo complex transformation with formation of new vesicles. The membrane-forming molecule S is created by reaction of two precursors A and B (present in the preformed giant vesicle as oil droplets). The precursor A is administrated in protected form A0 , which—to be converted into the reactive A form—must react with a catalyst C, embedded in the hydrophobic pool (vesicle membrane, oil droplets) of the giant vesicle. Once formed, S self-assembles to form a new internalized vesicle that can be translocated to originate a novel vesicle. Adapted from Takakura et al. [31].

250

P. Stano and P. L. Luisi

5. Vesicle-based Semisynthetic Cells and Their Self-Reproduction Vesicles self-reproduction, and more in general the self-reproduction of synthetic compartments, although not yet completely clarified in molecular details, is today recognized and acknowledged by researchers in the field of origins of life, since it provides an experimental evidence for the plausibility of protocell scenarios, in which primitive cell-like systems can undergo growth and self-reproduction without sophisticated enzyme-based machineries, typical of modern cells. From a slightly different viewpoint, the discovery of vesicle self-reproduction has prompted to consider an experimentally accessible goal the realization of vesiclebased cell-like systems, with the final aim of creating life in the laboratory. These constructs, which are called protocells, artificial cells, synthetic cell, and semisynthetic cells, represent a flourishing research area where chemistry and biology converge, attracting the attention of several international groups, last but not the least the synthetic biology community. In this short paragraph, our viewpoint on this subject will be shortly illustrated, inviting the interested reader to refer to specific reviews for a more detailed treatment [15, 21, 94, 95]. Within possible approaches to minimal living systems, a research line that currently seems more promising is the semisynthetic approach [15], introduced by our group some years ago [63, 94–96] (Fig. 12). It consists of the design, realization, and study of vesicle-based assemblies composed by the minimal number of

DNA

rRNA mRNA Ribosomes

Enzymes proteins

Ribosomes

Enzymes

DNA

Lipids

Figure 12 A cartoon depicting a hypothetical semisynthetic minimal cell. Extant molecules, such as DNA, enzymes, and structures like ribosomes are inserted into lipid vesicles, to form celllike structures, having internal organization. To date, research is focused to express functional proteins inside vesicles. Low molecular weight compounds are either provided from the beginning or added from outside.

Self-Reproduction of Micelles, Reverse Micelles, and Vesicles

251

components. Such components are extant enzymes, genes, and RNAs, which are put together in synthetic compartments (vesicles). The rationale for understanding the philosophy of minimal cells states that cellular life originated by a gradual increase in molecular and supramolecular complexity (atoms, molecules, biomonomers, macromolecules, polymer complexes, metabolic networks, cells). At a certain stage of this progression, early living cells appeared, even if ‘‘limping’’ [15], meaning simpler and less efficient than modern ones. The biological evolution of these early cells might have originated full-fledged living cells—similar to the cell we know today. Semisynthetic minimal cells can be considered concrete and experimentally achieved models of early—maybe limping—minimal cellular structures, since they implement a minimal number of functions. Semisynthetic cells are built from modern enzymes and genes, and therefore they cannot be considered models of ancient cells from the molecular viewpoint. What they model is the organized ensemble of functional components that interact with each other according to their properties and spatial organization. It is important to remark that—despite the lack of numerous functions— minimal cells still will have a ‘‘basic’’ organization that allows us to call them ‘‘living,’’ and such an organization finds its roots in the theory of autopoiesis (the relation between ‘‘autopoietic’’ and ‘‘living’’ has been recently discussed in [21, 23, 97]). A full discussion of the theoretical and experimental aspects of semisynthetic minimal cells lies outside the scope of this review; however, some features will be shortly described below, especially those related to the self-reproduction of vesicles. Current approaches to semisynthetic minimal cells involve the entrapment of proteins and genes inside vesicles in order to achieve some structural or enzymatic functions. These first steps are important since they improve our ability to construct minimal living systems. There are several reports on complex enzymatic reactions inside vesicles (see Table 5), ranging from polymerization of ADP to form poly(A), [64, 98], RNA replication [63], polymerase chain reaction [99], mRNA translation [100], and DNA transcription [101]. Some of these reactions have been already carried out simultaneously to vesicle self-reproduction [63, 64] (see asterisk-labeled entries in Table 5). All these initial studies paved the way to more advanced reactions, as protein expression [102–109]. Once it is possible to express functional proteins inside vesicles, cell-like systems can in principle be realized, for example, implementing some interesting functions inside vesicles, such as nutrient uptake, replication of components, cell growth, or lipid synthesis. Then, the question is: is it possible to achieve self-reproduction of semisynthetic minimal cells? In order to accomplish complete self-reproduction, minimal cells must undergo the so-called core and shell self-reproduction (Fig. 13). By ‘‘core’’ self-reproduction, we mean the reproduction—from within—of all components present in the minimal cell (enzymes, genes, etc.). By ‘‘shell’’ self-reproduction, we mean a process where the vesicle membrane (that encloses the core components and physically defines the minimal cell) uptakes membrane-forming molecules, with a consequent growth and division. The ‘‘shell’’ self-reproduction may occur in two ways: (1) the minimal cell itself produces—from within—the molecular components of its membrane, very much like living cells; (2) the minimal cell uptakes surfactant precursors

252 Table 5

P. Stano and P. L. Luisi

Compartmentalized reactions carried out in vesicles

No.

Year

Description

Reference

1 2 3 4 5 6 7 8 9 10 11

1994 1994 1994 1995 1995 1999 2001 2001 2002 2003 2004

[98] [64] [61] [63] [99] [100] [101] [102] [103] [104] [105]

12

2004

13

2006

14

2007

15

2007

ADP polymerization to poly(A), catalyzed by PNPase ADP polymerization to poly(A), catalyzed by PNPase(*) RNA replication in small vesicles(*) RNA replication in small vesicles(*) Polymerase chain reaction (PCR) in small vesicles mRNA translation in small vesicles DNA transcription in large and giant vesicles Protein (GFP) expression in large and giant vesicles Protein (GFP) expression in small vesicles Protein (GFP) expression in giant vesicles Cascade proteins (T7 RNA polymerase and GFP) expression in large and giant vesicles Proteins (hemolysin and GFP) expression in giant vesicles Protein (GFP) expression with purified enzymes (PURESYSTEMÒ ) in large and giant vesicles Protein (GFP) expression with PURESYSTEMÒ in submicrometric vesicles Membrane proteins (GPAT and LPAAT) expression in submicrometric vesicles by using PURESYSTEMÒ

[106] [107] [108] [109]

(*) Works where internalized reactions occurred together with vesicle (shell) self-reproduction. ADP, adenosine diphosphate; PNPase, polynucleotide phosphorilase; GFP, green fluorescent protein; T7, a kind of promoter sequence; PURESYSTEMÒ, a kit of 36 purified enzymes, tRNAs, and ribosomes of known composition; GPAT, glycerol-3-phosphate 1-acyltransferase, EC [2.3.1.15]; LPAAT, lysophosphatidic acid 2-acyltransferase, EC [2.3.1.51].

from the environment, and by a mechanism very similar to what was described in Paragraph 4 may self-reproduce. It is clear that a higher degree of sophistication—and adherence to the ideal autopoietic model of Fig. 1B—foresees the coupling of these two processes, that is, the core-and-shell–coupled self-reproduction, where at least a process links the selfreproduction of compartmentalized components and self-reproduction of the shell. Notice that the twofold nature of self-reproduction process (of the core elements and of the container) brings with it a further important issue, namely, the redistribution of the components among the two, or more, new compartments that arise from the division of the parent one. In particular, since in principle this redistribution is not regulated by some sophisticated mechanism, a stochastic process is expected, with the possible occurrence of new compartments that lack one or more key components. In this case, if the new system cannot accomplish all needed functions—and in particular those functions that are germane to being alive—a new phenomenon emerges, the so-called ‘‘death by dilution,’’ since components (and processes), initially present in the first minimal cell generation, are diluted among further generations. If the self-reproduction of the core components does not occur,

253

Self-Reproduction of Micelles, Reverse Micelles, and Vesicles

A

B

S

C

S

X, Y

X, Y S

X, Y S

S

S R

S S

S

S

X, Y X, Y

X, Y S

S

R S

S

X, Y X, Y

S

X, Y S

S

S

S

S S

S

S

S S

S

S

S

X, Y S

S

S X, Y S S S

Figure 13 Schematic illustration of shell, core, and core-and-shell self-reproduction in minimal cell systems. In shell self-reproduction (A), the membrane grows by uptaking a membrane component S (or a membrane precursor P, not shown, which is used by the system to endogenously produce S). Following this uptake (or internal production), the compartment grows and eventually divides into two or more new compartments. Since the internal components have not been reproduced, the statistics of redistribution of such components into the new compartments may bring about incomplete systems, that is, systems that lack one or more key internal components. If core self-reproduction is considered (B), the internal components self-reproduce (possibly by replication) by uptaking precursors R from the environment. Osmotic stresses can affect this reactive pathway. In core-and-shell self-reproduction, the two above-mentioned reactions (A, B) occur simultaneously, producing new compartments with no dilution of core components. The redistribution of core components, however, is not necessarily symmetric. When coupled core-and-shell reproduction is considered, the two self-reproduction processes are functionally coupled, for example, by means of a common intermediate (e.g., a chemical network that produces a catalyst, which, in turn, allows the formation of S). In (A–C), the number of components has been reduced for the sake of simplicity. Spherical symmetry is only hypothetical.

whereas the shell self-reproduction is operative, the fate of the cell population is this kind of death. To date, this is a rather far and difficult goal, but some groups are attempting to carry out studies by dividing this complex task into simpler subtasks, so that the final results might be achieved by stepwise increasing our knowledge and technical ability. The very first attempt to accomplish a sort of shell self-reproduction was carried out in 1991 by Schmidli et al. [110], who describe the reconstitution—in lecithin liposomes—of the whole four-enzyme system to carry out the biosynthesis of lecithin itself starting from water-soluble precursors. This system was clearly a way to achieve self-reproduction of vesicles by enzymatic transformation of precursors P into the membrane-forming compound S (a lecithin). Clearly, the catalysts of such transformation (the four enzymes) were not reproduced during the process (i.e., the components of the core were not reproducing), so that—even if successful, this approach would bring about death by dilution, after a certain number of selfreproduction cycles. In the work of Schmidli et al., carried out at ETH in Zurich,

254

P. Stano and P. L. Luisi

the low overall yield did not allow the imitation of the general scheme indicated in Fig. 2B; however, by a careful choice of the building blocks, it was possible to show a morphological transformation that followed the synthesis of lecithin molecules. In fact, it has been demonstrated that the four enzymes that convert glycerol-3phosphate into phosphatidylcholine (PC) can be functionally reconstituted in lecithin vesicles, producing long-chain PC in ca. 10% yield. In this case, no changes in physical parameters of the vesicles are observed. On the contrary, when short-chain (C8) PC was produced, a decrease in average vesicle size is observed, indicating that the in situ production of lipid can induce vesicle destabilization and rearrangement. A first possible advance to realize core-and-shell self-reproduction based on the system of Schmidli et al. [110] is to insert inside lipid vesicles the machinery to produce in situ the enzymes required for the synthesis of lipids. In this way, enzymes are produced, which, in turn, produce lipids, and core-and-shell self-reproduction can be sustained—at least partially. In fact, the core self-reproduction is only partial, since the machinery that produces the enzymes does not reproduce itself. Even if this approach is not completely satisfactory from the theoretical viewpoint, it represents an advancement to the realization of minimal cell. The biochemical machinery that synthesizes the enzymes needed for lipid production is the full set of genes, ribosomes, enzymes, and tRNAs required to perform in vitro transcription and translation processes. Therefore, the starting components are (a) the genes codifying for the enzyme responsible for this transformation; (b) the whole set of macromolecular components required to synthesize these enzymes from the corresponding genes; (c) water-soluble precursors (Fig. 14A). With this aim, a new research line has been recently introduced, which intends to reconstruct the first steps of lipid biosynthesis inside lipid vesicles. In order to further understand the activity of the enzymes involved in the lipidsalvage pathway, a detailed molecular study was carried out by Luci [111]. Recently, Schimdli’s approach (see Fig. 14B) has been investigated by our group [109], limited however to the first two enzymes, which convert glycerol-3-phosphate to phosphatidic acid. Two genes, codifying for the first two enzymes of the lipid-salvage pathway, are introduced inside lipid vesicles, together with PURESYSTEMÒ , a set of purified enzymes [112], ribosomes, and low molecular weight compounds, in order to produce the two enzymes within vesicles. The enzymes are then probed for the transformation of water-soluble substrates (glycerol-3-phosphate, palmitoylCoA, and oleoyl-CoA) into 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphatidic acid, a membrane-forming component. This work is currently in progress and the efficiency of such transformation is not known yet. If successful, this approach would represent a significant step toward the experimental realization of minimal living cells.

5.1. The Minimal RNA Cell DNA/protein-based minimal cells have been introduced in the previous paragraph, taking for granted that no other biopolymers could have genetic information and processing functions. The ‘‘RNA world’’ hypothesis, on the contrary, puts ribozymes under the spotlight of the primitive scene, since they jointly support the two above-

255

Self-Reproduction of Micelles, Reverse Micelles, and Vesicles

A Gene Enzyme P

S

S

Membrane growth and division

P B Glycerol-3-(P)

Lecithin

[2.3.1.15]

[2.7.8.2]

1-Acylglycerol-3-(P)

Diacylglycerol

[2.3.1.51]

[3.1.3.4] Phosphatidic acid

Figure 14 A cell-like structure that produces its own membrane. The strategy (A) implemented by researchers is to entrap into vesicles the genes which codify the enzyme(s) that can convert a precursor P into S, the membrane-forming compound S. Vesicles will grow and eventually divide. To date this system has not been realized yet. Current studies (B) focus on the reconstitution of the salvage pathway (from glycerol 3-phosphate and proper co-precursors) inside vesicles.

mentioned functions. Although the catalytic properties of known ribozymes cannot be compared with those of enzymes, it is interesting to devise a conceptual model of a minimal autopoietic living system, based on RNA instead of DNA and proteins. This hypothetical construct, which represents a nice example of combining the RNA world and compartment world, has been called the ‘‘minimal RNA cell’’ [113]. Let us imagine two ribozymes, called Rib-1 and Rib-2, which have the following properties: (1) Rib-1 is a replicase that can replicate itself and Rib-2; (2) Rib-2 is a synthase that can transform a precursor P into a membrane-forming compound S. If a certain number of Rib-1 and Rib-2 ribozyme molecules are now enclosed within a vesicle formed by S surfactant, and if the precursor P is available, as well as the activated nucleotides needed to replicate the ribozymes, a potentially autopoietic unit may arise (Fig. 15). In fact, the replication of internalized components and the production of membrane-forming compounds not only occur simultaneously but are also functionally coupled, since a relation exists between the two phenomena. In this RNA-based minimal cell, only two ribozymes suffice to accomplish core-and-shell self-reproduction, so that the minimal number of functions needed to be implemented in a synthetic construct is two. The system works thanks to the semipermeable character of the membrane, and to self-assembly rules. Unfortunately, the minimal RNA cell cannot be built in the laboratory, since two ribozymes with the above-mentioned catalytic (and structural) properties do not exist, or have been not discovered yet. Nevertheless, it is important to consider the minimal RNA cell as an interesting object that can reveal significant features of

256

P. Stano and P. L. Luisi

S Rib-1 Rib-1 NTPs

S Rib-2

NTPs S S

Rib-1 S

S

Rib-1 Rib-1

Rib-1 Rib-1

Rib-2

Rib-2 Rib-2

S

S

Rib-2

Rib-2

Rib-1 S

P

S

Rib-1

P

Rib-1 S

Rib-2 Rib-2 S

Figure 15 The minimal RNA cell. Two ribozymes Rib-1 (a replicase) and Rib-2 (a lipid synthase) are entrapped inside a vesicle composed by the membrane-forming compound S. If properly fed by activated nucleotides and by the precursor P, the system can give rise to autopoietic growth since Rib-1 makes a copy of itself and of Rib-2 (as well as complementary copies). Rib-2 catalyzes the P ! S transformation, so that the core-and-shell reproduction can be achieved, with consequent cell division. Notice that the minimal RNA cell contains only two ‘‘genes.’’ Adapted from Szostak et al. [113].

compartmentalized systems, since it is easy to model, so that numerical simulations may disclose unexpected collective/coordinated behavior.

6. Final Remarks In this chapter, we have resumed the main results on the self-reproduction of vesicle, micelles, and reverse micelles, discussing the experimental results within the framework of autopoiesis. We have seen how reverse micelles were successfully used to create the first chemical example of autopoietic theory, as foreseen by the programmatic paper with Francisco Varela in 1990 [35]. Aqueous micelles, on the contrary, may self-reproduce in certain conditions by an autocatalytic mechanism that does not need preformed micelles, since the first micelles form spontaneously by following a self-assembly path [38]. This result is particularly noteworthy since it demonstrates not only that autopoietic cycles exist but also that autopoietic cycles may start spontaneously according to physical laws. The reader will realize that the process is possible since the system lies in a far-from equilibrium state, and a continuous consumption (and processing) of compounds present in the environment occurs together with the onset of autopoietic micelle self-reproduction.

Self-Reproduction of Micelles, Reverse Micelles, and Vesicles

257

Vesicle self-reproduction represents a breakthrough in the field of supramolecular chemistry for at least three important reasons. The first is related to the discovery of a new reactive path in vesicle systems that enriches the list of known vesicle processes (fusion, aggregation, fission, swelling and shrinking, interaction with polymers, biopolymers, etc.), so that a deeper understanding of vesicles has been achieved. The second is instead linked to the field of origin of life, due to the observation that fatty acid vesicles are—to date—the most plausible candidate for prebiotic compartments, and therefore the discovery of their self-reproduction further supports origins-of-life scenarios, and strengthens the so-called ‘‘compartments-first approach,’’ where the role of compartmentalized and compartmentalizing systems is emphasized. The third reason that accounts for the great relevance of vesicles self-reproduction is somehow related to origin of life studies, but in a slightly different perspective. This view focuses on the realization of minimal living systems in the laboratory, with the aim of demonstrating that life can be reconstructed starting from a finite and known number of nonliving components. As other emergent properties, life becomes a property of a molecular system when a certain degree of complexity is reached, accompanied by a well-defined and specific selforganization pattern. Autopoiesis (Fig. 1) has been described as the blueprint of cellular life, and therefore autopoietic organization can be seen as the universal selforganization pattern that characterizes life. The theme of constructing minimal living cells is a Grand Challenge of modern science, and vesicle self-reproduction stays as the foundation of such perspective. Recent investigations on vesicle self-reproduction, and the initial studies on micelles, have revealed a common unified view of supramolecular assemblies selfreproduction, even though a detailed view on the mechanism of these reactions has not been established yet. Some unclear points persist, and the scientific debate on them is still lively. Fatty acid systems dominate the experimental landscape of published work, and only recently other systems have been investigated [31, 93]. Although the reasons for using fatty acids are related to their role in prebiotic compartment formation, as well as to their easy availability, an extension of self-reproduction studies to other surfactants (anionic, neutral, and cationic [45]) most probably would favor a better understanding. From a mechanistic side, it is possible that vesicles follow a complex route, especially if molecular details are concerned. The introduction of oleate micelles as the source of surfactant facilitated the use of photometric methods, as well as recent studies based on electron spin resonance [58], but—on the contrary—enriched the possible interaction between the components, in their different aggregation forms (monomers, micelles, vesicles), making the analysis complex. In addition to these physicochemical aspects, the theoretical approach still lacks a coherent view on the thermodynamics and kinetics of vesicle systems, and on the interplay of these two aspects as well. Because of their nature of ‘‘kinetic traps,’’ many of the (equilibrium) analyses cannot be applied successfully in these systems. Further studies may be focused on some factors that might affect self-reproduction capacity or mechanism, such as the membrane rigidity, vesicle size and composition, internal content, electroosmotic gradients, and phase transition temperature. Another unclear aspect (but very important) concerns the fate of water-soluble compounds entrapped in the parent vesicles, following the act of

258

P. Stano and P. L. Luisi

growing and dividing. Will trapped compound be evenly distributed among daughter vesicles? This and other complex aspects of vesicle self-reproduction are of course related to the construction of minimal living cells. More ambitious objectives refer in fact to this new research area: can semisynthetic minimal living cells be synthesized in the laboratory? What are the minimal functions that must be implemented in a semisynthetic construct to achieve this goal? Will a division produce two cells that can further grow and reproduce, or will stochastic processes lead to a sort of death due to the missing of key components? How many generations of minimal cells can be generated? In conclusion, thanks to the discovery of micelles and vesicle self-reproduction, a new stimulating chapter in the surfactant science appeared, noticeable for the manifold relevance of such behavior in supramolecular and prebiotic chemistry, origins of life studies, and self-organizing complex chemical systems. The concept of minimal cells, closely related to vesicle self-reproduction, and more in general to autopoietic organization, has been recently recognized as part of the chemical edge of synthetic biology [114, 115]—a young and emerging research area where biology and engineering fuse together, attracting several scientists with diverse backgrounds. Technical as well as theoretical aspects of further advancements are both very important and difficult to achieve; however, the interest of scientific community in the issue of compartments self-reproduction, and in the minimal cell counterpart, witnesses a sort of shared confidence that these are experimentally accessible targets. It might happen that the experimental study of molecular self-organization and autopoiesis will bring us to understand the very nature of cellular life, its origin, and development.

ACKNOWLEDGMENTS Pasquale Stano greatly acknowledges the ‘‘Enrico Fermi’’ Study and Research Centre for financial support. This work is part of the EU-funded project SYNTHCELLS (Contract Nr. 043359).

REFERENCES [1] I.V. Berezin, K. Martinek, A.K. Yatsimir, Physicochemical principles of micellar catalysis, Usp. Khim. 42 (1973) 1729–1756. [2] C.R. Sanders, G.C. Landis, Reconstitution of membrane-proteins into lipid-rich bilayered mixed micelles for NMR-studies, Biochemistry 34 (1995) 4030–4040. [3] P.L. Luisi, B.E. Straub (Eds.), Reverse Micelles: Biological and Technological Relevance of Amphiphilic Structures in Apolar Media (1st ed.), Plenum Press, N.Y. and London, 1984. [4] S. Osfouri, P. Stano, P.L. Luisi, Condensed DNA in lipid microcompartments, J. Phys. Chem. B 109 (2005) 19929–19935. [5] D.D. Lasic, Liposomes: From Physics to Applications (1st ed.), Elsevier, Amsterdam, 1993. [6] M. Rosoff (Ed.), Vesicles (Surfactant Science Series), Vol. 62. Marcel Dekker, New York, 1996. [7] D.D. Lasic, D. Papahadjopoulos (Eds.), Medical Applications of Liposomes, Elsevier, Amsterdam, 1998. [8] J. Wilshut, N. Duzgunes, R. Fraley, D. Papahadjopoulos, Studies on the mechanism of membrane fusion: Kinetics of calcium ion induced fusion of phosphatidylserine vesicles followed by a new assay for mixing of aqueous vesicle contents, Biochemistry 19 (1980) 6011–6021.

Self-Reproduction of Micelles, Reverse Micelles, and Vesicles

259

[9] G. Cevc, H. Richardsen, Lipid vesicles and membrane fusion, Adv. Drug Deliv. Rev. 38 (1999) 207–232. [10] D.P. Pantazatos, S.P. Pantazatos, R.C. MacDonald, Bilayer mixing, fusion, and lysis following the interaction of populations of cationic and anionic phospholipid bilayer vesicles, J. Membr. Biol. 194 (2003) 129–139. [11] D.M. Small (Ed.), Handbook of Lipid Research 4: The Physical Chemistry of Lipids. From Alkanes to Phospholipids, Plenum Press, New York, 1986. [12] D.P. Cistola, J.A. Hamilton, D. Jackson, D.M. Small, Ionization and phase behavior of fatty acids in water: Application of the Gibbs phase rule, Biochemistry 27 (1988) 1881–1888. [13] P. Walde, T. Namani, K. Morigaki, H. Hauser, Formation and properties of fatty acid vesicles (liposomes), in: G. Gregoriadis (Ed.), Liposome Technology (3rd ed.), Vol. 1: Informa Healthcare, New York, 2006, pp. 1–19. [14] F.J. Varela, H.R. Maturana, Autopoiesis: The organization of living systems, its characterization and a model, Biosystems 5 (1974) 187–196. [15] P.L. Luisi, F. Ferri, P. Stano, Approaches to semi-synthetic minimal cells: A review, Naturwissenschaften 93 (2006) 1–13. [16] H. Fellermann, R.V. Sole´, Minimal model of self-replicating nanocells: A physically embodied information-free scenario, Philos. Trans. R. Soc. B 362 (2007) 1803–1811. [17] F. Mavelli, K. Ruiz-Mirazo, Stochastic simulations of minimal self-reproducing cellular systems, Philos. Trans. R. Soc. B 362 (2007) 1789–1802. [18] R. Serra, T. Carletti, I. Poli, Synchronization phenomena in surface-reaction models of protocells, Artifi. Life 13 (2007) 123–138. [19] G.R. Fleischaker, A few precautionary words concerning terminology, in: G.R. Fleischaker, S. Colonna, P.L. Luisi (Eds.), Self-Production of Supramolecular Structures. From Synthetic Structures to Models of Minimal Living Systems, NATO ASI Series, Vol. 446: Kluver Academic Publishers, Dordrecht, The Nederlands, 1994, pp. 33–41. [20] P.L. Luisi, Appendix, in: G.R. Fleischaker, S. Colonna, P.L. Luisi (Eds.), Self-Production of Supramolecular Structures. From Synthetic Structures to Models of Minimal Living Systems, NATO ASI Series, Vol. 446: Kluver Academic Publishers, Dordrecht, The Nederlands, 1994, pp. 196–197. [21] P.L. Luisi, The Emergence of Life: From Chemical Origin to Synthetic Biology, Cambridge University Press, Cambridge, 2006. [22] F.J. Dyson, The Origins of Life, Cambridge University Press, Cambridge, 1985. [23] P.L. Luisi, Autopoiesis: A review and a reappraisal, Naturwissenschaften 90 (2003) 49–59. [24] H.H. Zepik, E. Bloechliger, P.L. Luisi, A chemical model of homeostasis, Angew. Chem. Int. Ed. 40 (2001) 199–202. [25] Y. Chizmadzhew, M. Maestro, F. Mavelli, A simplified kinetic model for an autopoietic synthesis of micelles, Chem. Phys. Lett. 22 (1994) 656–662. [26] F. Mavelli, P.L. Luisi, Autopoietic self-reproducing vesicles: A simplified kinetic model, J. Phys. Chem. 100 (1996) 16600–16607. [27] D. Segre´, D. Lancet, O. Kedem, Y. Pilpel, Graded autocatalysis replication domain (GARD): Kinetic analysis of self-replication in mutually catalytic sets, Orig. Life Evol. Biosph. 28 (1998) 501–514. [28] C.D. Bolton, J.A.D. Wattis, The size-templating matrix effect in vesicle formation 1: A microscopic model and analysis, J. Phys. Chem. B 107 (2003) 7126–7134. [29] C.D. Bolton, J.A.D. Wattis, Size-templating matrix effect in vesicle formation 2: Analysis of a macroscopic model, J. Phys. Chem. B 107 (2003) 14306–14318. [30] R.V. Sole´, A. Munteanu, C. Rodriguez-Caso, J. Macia, Synthetic protocell biology: From reproduction to computation, Philos. Trans. R. Soc. B 362 (2007) 1727–1739. [31] K. Takakura, T. Toyota, T. Sugawara, A novel system of self-reproducing giant vesicles, J. Am. Chem. Soc. 125 (2003) 8134–8140. [32] B. Bozic, S. Svetina, A relationship between membrane properties forms the basis of a selectivity mechanism for vesicle self-reproduction, Eur. Biophys. J. 33 (2004) 565–571. [33] I.A. Chen, J.W. Szostak, A kinetic study of the growth of fatty acid vesicles, Biophys. J. 87 (2004) 988–998.

260

P. Stano and P. L. Luisi

[34] P.A. Bachmann, P. Walde, P.L. Luisi, J. Lang, Self-replicating reverse micelles and chemical autopoiesis, J. Am. Chem. Soc. 112 (1990) 8200–8201. [35] P.L. Luisi, F.J. Varela, Self-replicating micelles—A chemical version of minimal autopoietic systems, Orig. Life Evol. Biosph. 19 (1990) 633–643. [36] P.A. Bachmann, P.L. Luisi, J. Lang, Self-replication of reverse micelles, Chimia 45 (1991) 266–268. [37] P.A. Bachmann, P. Walde, P.L. Luisi, J. Lang, Self-replicating micelles: Aqueous micelles and enzymatically driven reactions in reverse micelles, J. Am. Chem. Soc. 113 (1991) 8204–8209. [38] P.A. Bachmann, P.L. Luisi, J. Lang, Autocatalytic self-replicating micelles as models for prebiotic structures, Nature 357 (1992) 57–59. [39] T. Buhse, R. Nagarajan, D. Lavabre, J.C. Micheau, Phase-transfer model for the dynamics of ‘‘micellar autocatalysis,’’ J. Phys. Chem. A 101 (1997) 3910–3917. [40] T. Buhse, D. Lavare, R. Nagarajan, J.C. Micheau, Origin of autocatalysis in the biphasic alkaline hydrolysis of C-4 to C-8 ethyl alkanoates, J. Phys. Chem., A 102 (1998) 10552–10559. [41] S. Rasmussen, L. Chen, M. Nilsson, S. Abe, Bridging nonliving and living matter, Artif. Life 9 (2003) 269–316. [42] S. Rasmussen, L. Chen, D. Deamer, D.C. Krakauer, N.H. Packard, P.F. Stadler, M.A. Bedau, Transitions from nonliving to living matter, Science 303 (2004) 963–965. [43] D. Fiordemondo, P. Stano, Lecithin-based water-in-oil compartments as dividing bioreactors, ChemBioChem (2007) DOI: 10.1002/cbic.200700112. [44] P. Walde, R. Wick, M. Fresta, A. Mangone, P.L. Luisi, Autopoietic self-reproduction of fatty acid vesicles, J. Am. Chem. Soc. 116 (1994) 11649–11654. [45] C.F. Thomas, P.L. Luisi, Novel properties of DDAB: Matrix effect and interaction with oleate, J. Phys. Chem. B 108 (2004) 11285–11290. [46] W.R. Hargreaves, S.J. Mulvhill, D.W. Deamer, Synthesis of phospholipids and membranes in prebiotic conditions, Nature 266 (1977) 78–80. [47] W.R. Hargreaves, D.W. Deamer, Liposomes from ionic, single-chain amphiphiles, Biochemistry 17 (1978) 3759–3768. [48] J.M. Gebicki, M. Hicks, Ufasomes are stable particles surrounded by unsaturated fatty acid membranes, Nature 243 (1973) 232–234. [49] J.G. Lawless, G.U. Yuen, Quantification of monocarboxylic acids in the Murchison carbonaceous meteorite, Nature 282 (1979) 431–454. [50] D.W. Deamer, J. Oro, Role of lipids in prebiotic structures, Biosystems 12 (1980) 167–175. [51] T.M. McCollom, G. Ritter, B.R.T. Simoneit, Lipid synthesis under hydrothermal conditions by Fischer-Tropsch-type reactions, Orig. Life Evol. Biosph. 29 (1999) 153–166. [52] A.I. Rushdi, B.R.T. Simoneit, Lipid formation by aqueous Fischer-Tropsch-type synthesis over a temperature range of 100 to 400  C, Orig. Life Evol. Biosph. 31 (2001) 103–118. [53] D.W. Deamer, J.P. Dworkin, Chemistry and physics of primitive membranes, Top. Curr. Chem. 259 (2005) 1–27. [54] P. Walde, Surfactant assemblies and their various possible roles for the origin(s) of life, Orig. Life Evol. Biosph. 36 (2006) 109–150. [55] K. Morigaki, P. Walde, Fatty acid vesicles, Curr. Opin. Colloid. Interface Sci. 12 (2007) 75–80. [56] J.N. Israelachvili, D.J. Mitchell, B.W. Ninham, Theory of self-assembly of lipid bilayers and vesicles, Biochim. Biophys. Acta 470 (1977) 185–201. [57] T.H. Haines, Anionic lipid headgroups as a proton-conducting pathway along the surface of membranes: A hypothesis, Proc. Natl. Acad. Sci. USA 80 (1983) 160–164. [58] H. Fukuda, A. Goto, H. Yoshioka, R. Goto, K. Morigaki, P. Walde, Electron spin resonance study of the pH-induced transformation of micelles to vesicles in an aqueous oleic acid/oleate system, Langmuir 17 (2001) 4223–4231. [59] P.L. Luisi, P. Vonmont-Bachmann, M. Fresta, Self-reproduction of micelles and lipsomes and the transition to life, J. Liposome Res. 3 (1993) 63l–638. [60] P. Vonmont-Bachmann, P. Walde, P.L. Luisi, Lipase-catalyzed reactions in vesicles as an approach to vesicle self-reproduction, J. Liposome Res. 4 (1994) 1135–1158.

Self-Reproduction of Micelles, Reverse Micelles, and Vesicles

261

[61] P.L. Luisi, P. Walde, T. Oberholzer, Enzymatic RNA synthesis in self-reproducing vesicles: An approach to the construction of a minimal synthetic cell, Ber. Bunsen-Ges. Phys. Chem. 98 (1994) 1160–1165. [62] P. Walde, R. Wick, M. Fresta, A. Mangone, P.L. Luisi, Autopoietic self-reproduction of fatty acid vesicles, J. Am. Chem. Soc. 116 (1994) 11649–11654. [63] T. Oberholzer, R. Wick, P.L. Luisi, C.K. Biebricher, Enzymatic RNA replication in selfreproducing vesicles: An approach to a minimal cell, Biochem. Biophys. Res. Commun. 207 (1995) 250–257. [64] P. Walde, A. Goto, P.A. Monnard, M. Wessicken, P.L. Luisi, Oparin’s reactions revisited: Enzymatic synthesis of poly(adenylic acid) in micelles and self-reproducing vesicles, J. Am. Chem. Soc. 116 (1994) 7541–7544. [65] E. Bloechliger, M. Blocher, P. Walde, P.L. Luisi, Matrix effect in the size distribution of fatty acid vesicles, J. Phys. Chem. 102 (1998) 10383–10390. [66] S. Lonchin, P.L. Luisi, P. Walde, B.H. Robinson, A matrix effect in mixed phospholipid/fatty acid vesicle formation, J. Phys. Chem. B 103 (1999) 10910–10916. [67] N. Ono, Computational studies on conditions of the emergence of autopoietic protocells, BioSystems 81 (2005) 223–233. [68] H. Fellermann, S. Rasmussen, H.-J. Ziock, R.V. Sole´, Life cycle of a minimal protocell—A dissipative particle dynamics study, Artif. Life 13 (2007) 319–345. [69] J. Macia, R.V. Sole´, Protocell self-reproduction in a spatially extended metabolism–vesicle system, J. Theor. Biol. 245 (2007) 400–410. [70] K. Kaneko, Question 8: From a set of chemical reactions to reproducing cells, Orig. Life Evol. Biosph. 37 (2007) 449–453. [71] N. Berclaz, M. Mueller, P. Walde, P.L. Luisi, Growth and transformation of vesicles studied by ferritin labeling and cryotransmission electron microscopy, J. Phys. Chem. B 105 (2001) 1056–1064. [72] N. Berclaz, E. Bloechliger, M. Mueller, P.L. Luisi, Matrix effect of vesicle formation as investigated by cryotransmission electron microscopy, J. Phys. Chem. B 105 (2001) 1065–1071. [73] S. Rasi, F. Mavelli, P.L. Luisi, Cooperative micelle binding and matrix effect in oleate vesicle formation, J. Phys. Chem. B 107 (2003) 14068–14076. [74] S. Rasi, F. Mavelli, P.L. Luisi, Matrix effect in oleate micelles-vesicles transformation, Orig. Life Evol. Biosph. 34 (2004) 215–224. [75] Z. Cheng, P.L. Luisi, Coexistence and mutual competition of vesicles with different size distributions, J. Phys. Chem. B 107 (2003) 10940–10945. [76] P.L. Luisi, P. Stano, S. Rasi, F. Mavelli, A possible route to prebiotic vesicle reproduction, Artif. Life 10 (2004) 297–308. [77] S. Chungcharoenwattana, M. Ueno, Size control of mixed egg yolk phosphatidylcholine (EggPC)/oleate vesicles, Chem. Pharm. Bull. 52 (2004) 1058–1062. [78] S. Chungcharoenwattana, M. Ueno, New vesicle formation upon oleate addition to preformed vesicles, Chem. Pharm. Bull. 53 (2005) 260–262. [79] S. Chungcharoenwattana, H. Kashiwagi, M. Ueno, Effect of preformed egg phosphatidylcholine vesicles on spontaneous vesiculation of oleate micelles, Colloid Polym. Sci. 283 (2005) 1180–1189. [80] F. Kamp, J.A. Hamilton, pH Gradients across phospholipid-membranes caused by fast flip-flop of unionized fatty-acids, Proc. Natl. Acad. Sci. USA 89 (1992) 11367–11370. [81] F. Kamp, D. Zakim, F. Zhang, N. Noy, J.A. Hamilton, Fatty-acid flip-flop in phospholipidbilayers is extremely fast, Biochemistry 34 (1995) 11928–11937. [82] M.L. Rogerson, B.H. Robinson, S. Bucak, P. Walde, Kinetic studies of the interaction of fatty acids with phosphatidylcholine vesicles (liposomes), Colloids Surf. B: Biointerfaces 48 (2006) 24–34. [83] P. Stano, E. Wehrli, P.L. Luisi, Insights on the oleate vesicles self-reproduction, J. Phys.: Condens. Matter 18 (2006) S2231–S2238. [84] K. Ruiz-Mirazo, F. Mavelli, On the way towards ‘basic autonomous agents’: Stochastic simulations of minimal lipid-peptide cells, BioSystems (2007) doi: 10.1016/j.biosystems.2007.05.013.

262

P. Stano and P. L. Luisi

[85] M.I. Angelova, D.D. Dimitrov, Liposome electroformation, Faraday Discuss. Chem. Soc. 81 (1986) 303–311. [86] K. Akashi, H. Miyata, H. Itoh, K. Kinoshita Jr., Preparation of giant liposomes in physiological saline and characterization under an optical microscope, Biophys. J. 71 (1996) 3242–3250. [87] N. Magome, T. Takemura, K. Yoshikawa, Spontaneous formation of giant liposomes from neutral phospholipids, Chem. Lett. 3 (1997) 205–206. [88] P.L. Luisi, P. Walde (Eds.), Giant Vesicles, Perspectives in Supramolecular Chemistry, John Wiley & Sons, Chichester, 2000. [89] P. Walde, Preparation of vesicles (liposomes), in: H.S. Nalwa (Ed.), Encyclopedia of Nanoscience and Nanotechnology, Vol. 9: American Scientific Publishers, Los Angeles, 2004, pp. 43–79. [90] R. Wick, P. Walde, P.L. Luisi, Autocatalytic self-reproduction of giant vesicles, J. Am. Chem. Soc. 117 (1995) 1435–1436. [91] R. Wick, P. Walde, P.L. Luisi, Giant vesicles, in: G.R. Fleischaker, S. Colonna, P.L. Luisi (Eds.), Self-Production of Supramolecular Structures. From Synthetic Structures to Models of Minimal Living Systems, NATO ASI Series, Vol. 446: Kluver Academic Publishers, Dordrecht, The Nederlands, 1994, pp. 196–197. [92] F.M. Menger, K. Gabrielson, Chemically-induced birthing and foraging in vesicle systems, J. Am. Chem. Soc. 116 (1994) 1567–1568. [93] K. Takakura, T. Sugawara, Membrane dynamics of a myelin-like giant multilamellar vesicle applicable to a self-reproducing system, Langmuir 20 (2004) 3832–3834. [94] P.L. Luisi, Toward the engineering of minimal living cells, Anat. Rec. 268 (2002) 208–214. [95] S. Islas, A. Becerra, P.L. Luisi, A. Lazcano, Comparative genomics and the gene complement of a minimal cell, Orig. Life Evol. Biosph. 34 (2004) 243–256. [96] P.L. Luisi, T. Oberholzer, A. Lazcano, The notion of a DNA minimal cell: A general discourse and some guidelines for an experimental approach, Helv. Chim. Acta 85 (2002) 1759–1777. [97] M. Bitbol, P.L. Luisi, Autopoiesis with or without cognition: Defining life at its edge, J. R. Soc. Interface 1 (2004) 99–107. [98] A.C. Chakrabarti, R.R. Breaker, G.F. Joyce, D.W. Deamer, Production of RNA by a polymerase protein encapsulated within phospholipid vesicles, J. Mol. Evol. 39 (1994) 555–559. [99] T. Oberholzer, M. Albrizio, P.L. Luisi, Polymerase chain reaction in liposomes, Chem. Biol. 2 (1995) 677–682. [100] T. Oberholzer, K.H. Nierhaus, P.L. Luisi, Protein expression in liposomes, Biochem. Biophys. Res. Commun. 261 (1999) 238–241. [101] K. Tsumoto, S.M. Nomura, Y. Nakatani, K. Yoshikawa, Giant liposome as a biochemical reactor: Transcription of DNA and transportation by laser tweezers, Langmuir 17 (2001) 7225–7228. [102] W. Yu, K. Sato, M. Wakabayashi, T. Nakatshi, E.P. Ko-Mitamura, Y. Shima, I. Urabe, T. Yomo, Synthesis of functional protein in liposome, J. Biosci. Bioeng. 92 (2001) 590–593. [103] T. Oberholzer, P.L. Luisi, The use of liposomes for constructing cell models, J. Biol. Phys. 28 (2002) 733–744. [104] S.M. Nomura, K. Tsumoto, T. Hamada, K. Akiyoshi, Y. Nakatani, K. Yoshikawa, Gene expression within cell-sized lipid vesicles, ChemBioChem 4 (2003) 1172–1175. [105] K. Ishikawa, K. Sato, Y. Shima, I. Urabe, T. Yomo, Expression of a cascading genetic network within liposomes, FEBS Lett. 576 (2004) 387–390. [106] V. Noireaux, A. Libchaber, A vesicle bioreactor as a step toward an artificial cell assembly, Proc. Natl. Acad. Sci. USA 101 (2004) 17669–17674. [107] T. Sunami, K. Sato, T. Matsuura, K. Tsukada, I. Urabe, T. Yomo, Femtoliter compartment in liposomes for in vitro selection of proteins, Anal. Biochem. 357 (2006) 128–136. [108] G. Murtas, Y. Kuruma, P. Bianchini, A. Diaspro, P.L. Luisi, Protein synthesis in liposomes with a minimal set of enzymes, Biochem. Biophys. Res. Comm. 363 (2007) 12–17. [109] Y. Kuruma, Question 7: Biosynthesis of phosphatidic acid in liposome compartments—toward the self-reproduction of minimal cells, Orig. Life Evol Biosph. 37 (2007) 409–413.

Self-Reproduction of Micelles, Reverse Micelles, and Vesicles

263

[110] P.K. Schmidli, P. Schurtenberger, P.L. Luisi, Liposome-mediated enzymatic synthesis of phosphatidylcholine as an approach to self-replicating liposomes, J. Am. Chem. Soc. 113 (1991) 8127–8130. [111] P. Luci, Gene cloning expression and purification of membrane proteins, ETH-Z Dissertation Nr. 15108 Zurich, 2003. [112] Y. Shimizu, A. Inoue, Y. Tomari, T. Suzuki, T. Yokogawa, K. Nishikawa, T. Ueda, Cell-free translation reconstituted with purified components, Nat. Biotechnol. 19 (2001) 751–755. [113] J.W. Szostak, D.P. Bartel, P.L. Luisi, Synthesizing life, Nature 409 (2001) 387–390. [114] P.L. Luisi, C. Chiarabelli, P. Stano, From the never born proteins to the minimal living cell: Two projects in synthetic biology, Orig. Life Evol. Biosph. 36 (2006) 605–616. [115] P.L. Luisi, Chemical aspects of synthetic biology, Chem. Biodiv. 4 (2007) 603–621.

This page intentionally left blank

SUBJECT INDEX Aggregation number 229, 230 Anionic and neutral detergents 170, 171 Anisotropic membrane inclusion 144, 149 Antibacterial substances 125, 140 Antigenicity 82 Antimicrobial peptides 127 Antiphospholipid syndrome 80 Application 203, 204, 205, 209, 210, 211, 212, 213, 214, 215 Assembly 203, 204, 210, 211 Autoimmunity 116, 117 Autopoiesis 222, 223, 224, 225, 226, 232, 234, 251, 256, 257, 258

Flexible membrane inclusion 143, 144, 145, 146, 147, 148, 153, 154, 164, 165 Fractional Brownian motion (fBm) 24, 29, 32 Fractional Le´vy stable 24

Bending energy 150, 151, 158, 165 Bilayer 226, 235, 237, 238 composition 184 Biomineralization 203, 204, 213, 215 Boundary 224, 225, 226, 227, 228, 231, 238, 239 Bursting of GUV 122, 137, 138, 139, 140

hCt fibrillation 172, 195 a-Helix inductor 178 Homeostasis 224, 248

Calcitonins 170, 172, 173, 189, 194 Caprylate micelles 232 Carboxylic acid 238 Catechins 134 Cell extract 244 CMC 171, 196 Conformational variation 171 Controlled release 40 Core-and-shell self-reproduction 253, 254, 255 Cryo-TEM 242, 243, 245 Death 224, 258 Death by dilution 252, 253 Diagnostic 40, 41, 50, 52 Director membrane nanodomains 143, 144, 159 Division 225, 226, 227, 228, 231, 232, 234, 235, 238, 241, 243, 244, 245, 246, 247, 248, 251, 252, 255, 256, 258 Dynamic light scattering (DLS) 240, 241, 242, 243, 244 eCt glycosylation 171, 190, 191, 192, 193 Electropermeabilization 53 Encapsulation 42, 46, 47, 48, 51, 52, 203, 204, 209, 210, 211 Epigallocatechin gallate (EGCg) 134, 135, 136, 137, 138, 139, 140 Epitopes 85 ETH 232, 253 Fatty acid 223, 225, 228, 229, 231, 234, 235, 236, 237, 238, 239, 245, 248, 249, 257 Ferritin 242, 243, 244

Giant vesicle 221, 222, 229, 235, 236, 248, 249, 252 b2-Glycoprotein I 80, 84, 85, 86 Green fluorescent protein 252 Growth 221, 224, 225, 227, 228, 229, 230, 232, 234, 238, 239, 241, 242, 243, 244, 245, 246, 247, 248, 250, 251, 256

Impedance spectroscopy 59, 62, 64, 67, 74 Inorganic nano-materials 203, 204 Intrinsic curvatures 146 Intrinsic shape of molecule 149, 150, 151, 152, 154, 155 Ion channel 60, 61, 69, 71, 73, 75, 172, 185, 186, 191, 197 Kinetic trap 257 Leakage of internal contents 122, 124, 140 Lecithin 245, 253, 254, 255 Limping cell 251 Lipase 231, 235, 236, 238 Lipid 203, 204, 205, 206, 207, 208, 209, 210, 211, 212, 213, 214, 215, 222, 236, 250, 251, 254, 256 bilayer 144, 149, 150, 151, 155, 156, 158, 159, 161, 165 precursor 222, 225, 227, 228, 229, 231, 232, 234, 235 salvage pathway 254 Liposome 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 55, 222, 253 Living system 223, 227, 232, 250, 251, 255, 257 Local shape perturbation 143, 156, 162, 163 LUV suspension method 121, 122, 123, 124, 125, 128, 129, 140 Magainin 2 122, 125, 126, 127, 128, 129, 130, 131, 132, 133, 137, 139, 140 Matrix effect 240, 241, 242, 243, 244, 246 Membrane 222, 226, 227, 229, 230, 238, 239, 246, 248, 249, 251, 252, 253, 254, 255, 256, 257 Membrane-binding 82, 83, 84, 85 Membrane free energy 160 Meteorite 237

265

266 Micelle 221, 222, 223, 226, 227, 228, 229, 230, 231, 232, 233, 234, 235, 236, 237, 238, 239, 240, 241, 242, 243, 244, 245, 246, 247, 256, 257, 258 Minimal cell 222, 230, 250, 251, 252, 253, 254, 255, 258 Minimal living cell 254, 257, 258 Minimal RNA cell 254, 255, 256 Murchison meteorite 237 Nanobiotechnology 39 Nano-black lipid membrane 76, 77 Nanocapsules 51, 55 Nanopore 1, 4, 19, 20, 21, 30, 31 Oleate 234, 235, 236, 237, 238, 239, 240, 241, 242, 243, 244, 245, 246, 247, 248, 257 Oleate vesicles 235, 236, 238, 239, 240, 241, 242, 243, 244, 245, 246, 247 Oleic acid 237, 239, 248 Orientational ordering 149 Permeability 222, 236, 246 pH 170, 171, 172, 177, 179, 180, 181, 182, 183, 184, 185, 186, 187, 188, 192, 194, 195, 196, 197, 198 Phospholipid 236, 241 Phospholipid-binding 85, 87 PLM 170, 171, 172, 173, 175, 176, 177, 180, 184, 185, 186, 187, 188, 189, 194, 197, 198 Polymer network 43 POPC 236, 241, 242, 243, 244 Pore formation in lipid membranes 133 Porous substrate 61, 63, 64, 65, 66, 67, 72, 73, 74, 75 Precursor 222, 224, 225, 226, 227, 228, 229, 231, 232, 234, 235, 238, 239, 240, 242, 243, 244, 247, 249, 251, 253, 254, 255, 256 Pre-formed vesicle 238, 239, 240, 241, 242, 243, 244, 245, 246, 247 Proton pump 60, 69 PURESYSTEM 252, 254 Release 225, 231, 245 Replicase 255, 256 Reverse micelle 221, 222, 223, 226, 227, 228, 229, 230, 231, 232, 234, 235, 239, 256

Subject Index

Ribozyme 254, 255, 256 Rigid membrane protein 149, 151, 163, 165 RNA world 254, 255 Schiff base 249 Self-replication 223 Self-reproduction 221, 222, 223, 224, 225, 226, 227, 228, 229, 230, 231, 232, 233, 234, 235, 236, 238, 239, 240, 241, 242, 243, 244, 246, 247, 248, 249, 250, 251, 252, 253, 254, 255, 256, 257, 258 Semisynthetic minimal cell 250, 251 Single GUV method 121, 122, 123, 124, 125, 127, 129, 134, 139, 140, 141 Size distribution 227, 230, 240, 242, 243, 244, 245 Size exclusion chromatography 242, 245 Stochastic 4, 7, 19, 20, 21, 22, 23, 24, 26, 27, 29, 30, 31, 32 Stopped-flow 242, 245 Supramolecular assemblies 257 Surfactant 221, 223, 226, 227, 229, 230, 231, 232, 237, 238, 239, 240, 243, 244, 246, 248, 249, 251, 255, 257, 258 precursor 232, 240, 251 Synthase 255, 256 Synthesis 203, 204, 205, 209, 210, 213, 215 Template polymerization 43, 44, 48 Translocation 248 Turbidity 240, 241, 242, 243 Twin vesicles 246

Vesicle 40, 41, 43, 44, 47, 48, 49, 50, 55, 221, 222, 223, 225, 226, 227, 228, 229, 230, 234, 235, 236, 237, 238, 239, 240, 241, 242, 243, 244, 245, 246, 247, 248, 249, 250, 251, 252, 253, 254, 255,256, 257, 258 division 225, 226, 227, 228, 229, 234, 235, 238, 243, 244, 245, 246, 247, 248, 251, 252, 255, 256, 258 growth 238, 243, 245 size 227, 254, 257 Water-pool 230