This page intentionally left blank
Jens Dietrich Catherine Venien-bryan University of Oxford, UK
Imperial College Press
Published by Imperial College Press 57 Shelton Street Covent Garden London WC2H 9HE Distributed by World Scientific Publishing Co. Pte. Ltd. 5 Toh Tuck Link, Singapore 596224 USA office: 27 Warren Street, Suite 401-402, Hackensack, NJ 07601 UK office: 57 Shelton Street, Covent Garden, London WC2H 9HE
British Library Cataloguing-in-Publication Data A catalogue record for this book is available from the British Library.
STRATEGIES FOR TWO-DIMENSIONAL CRYSTALLIZATION OF PROTEINS USING LIPID MONOLAYERS Copyright © 2005 by Imperial College Press All rights reserved. This book, or parts thereof, may not be reproduced in any form or by any means, electronic or mechanical, including photocopying, recording or any information storage and retrieval system now known or to be invented, without written permission from the Publisher.
For photocopying of material in this volume, please pay a copying fee through the Copyright Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923, USA. In this case permission to photocopy is not required from the publisher.
ISBN 1-86094-428-0
Typeset by Stallion Press Email:
[email protected]
Printed in Singapore.
“… mare … omne oleo tranquillari, et ob id urinantes ore spargere, quoniam mitiget naturam asperam lucemque deportet.” Caius Plinii Secundi Naturalis historia, Liber II CVI. Mirabilia fontium et fluminum
“All sea water is made smooth by oil, and so divers sprinkle oil from their mouth because it calms the rough element and carries light down with them.” Pliny the Elder Roman Naturalist AD 23–79, Natural History,Book 2 Chapter CVI
This page intentionally left blank
Acknowledgments
We would like to thank the following people: • Christine Ziegler and Peter Bond for help with some figures; • Elizabeth Hewat, Kevin Leonard, Richard K. Bryan and Michael Troke for critically reading the manuscript. • The Wellcome Trust and Human Frontier Science Programme for financial support Jens Dietrich Catherine Vénien-Bryan July 2004
vii
This page intentionally left blank
Introduction
After sequencing the human genome and the genome of many other organisms, the age of “structural genomics” has started. It is now possible to take the genetic sequence and infer the amino acid sequence of all proteins from it. Unfortunately this does not necessarily tell us something about the function of these proteins. Even for the organism that is genetically best understood, Escherischia coli, the function of ca. 40% of its protein is not known. Under the structural genomics initiative, many groups set out to determine as many protein folds as possible. A protein fold is a stretch of amino acids that fold into a defined secondary structure motif. These structural prototypes adopt similar structures, but do not necessarily have detectable sequence homology. At the moment, about 1,000 different folds are known and it is estimated that a few thousand different folds exist. Often X-ray crystallography of 3D protein crystals and nuclear magnetic resonance (NMR) studies of concentrated protein solutions are the only source of structural information at atomic level, and most of the work in structural genomics is limited to watersoluble proteins. Membrane proteins and larger protein complexes are mostly excluded from these efforts because they are less amenable to this “high-throughput” approach. Membrane proteins constitute 25% of all proteins, therefore it is crucial to improve our understandings of these proteins. The “Protein Data Base” (PDB) stores ca. 28,000 structures of water-soluble proteins (about 18,000 of these are unrelated), but only 87 structures of membrane proteins are available (of which 51 are unrelated) and only 93 structures originate from proteins with a molecular weight ⬎250kDa. ix
x
Introduction
The handling of membrane proteins and large protein complexes is much more demanding than for water-soluble proteins. The reasons for this are: • • •
Membrane proteins are usually difficult to express in large quantity for structural analysis; Monodispersity and stability of the purified protein or large protein complexes are often difficult to control; The localization of membrane proteins in the lipid bilayer requires that they exhibit hydrophilic and hydrophobic surfaces.
Therefore solubilization and purification of membrane proteins necessitates the use of detergent for masking the hydrophobic area. As a consequence of this dual property — hydrophobic and hydrophilic — the total hydrophilic surfaces available to provide a good crystal contact necessary for 3D crystallization are very limited. The hydrophobic domain, which is masked by detergent micelles, does not play a major role in the crystal contact. The growth of 3D crystal of membrane proteins is therefore a complicated task. In this context, electron microscopy of single particles (for protein with a molecular weight >250 kDa) and 2D crystals is a powerful technique for which, in some cases, no other alternative approach is possible. Thanks to advances in electron microscopy instrumentation, specimen preparation and image processing, this technology is beginning to satisfy the demand for structures and allow learning about mechanisms at atomic level. The atomic structures of light harvesting complex, bacteriorhodopsin, and tubulin have now been solved, crucial elements of secondary structure have been revealed in several membrane proteins (aquaporin, rhodopsin, gap junctions and Ca2⫹ and H⫹-ATPase) and a novel viral fold of the hepatitis B core protein has been determined through the application of this technique. Nevertheless the resolution obtained by this technique is often in the medium range and the ability to combine structures of macromolecular complexes derived by electron microscopy with X-ray or NMR structures of their components allows the reconstruction of molecular machines and large multi-protein complexes in considerable detail. Two-dimensional crystals have been most successfully used to obtain high and medium resolution structural information by electron microscopy.
Introduction xi
In this book we offer an overview of the technique of crystallization of proteins on lipid monolayers. The structural information is then obtained by imaging the 2D crystals using electron microscopy and image processing. This method allows any soluble or membrane protein from very small molecular weight to large complexes such as viruses to be crystallized in two dimensions. Strategies to adsorb, concentrate, orient and organize proteins or macromolecules on supports suitable for electron microscopic observation and with fluidity properties similar to biological membranes will be presented and discussed. Biophysical techniques to monitor and improve the process of crystallization will be detailed.
This page intentionally left blank
Contents
Introduction 1
2
3
ix
Two-dimensional Crystallization on Lipid Monolayers 1.1. Overview 1.2. Non-specific Adsorption through Electrostatic Interactions 1.3 Specific Adsorption via Ligands Two-dimensional Crystallization of Membrane Proteins 2.1 Overview 2.2. Naturally Occurring and Induced 2D Crystals 2.3. 2D Crystallization by Reconstitution into a Lipid Bilayer 2.4 Surface Crystallization of Membrane Proteins Methods to Monitor Surface Crystallization Introduction to the Biophysical Studies of Interfaces 3.1. Fluidity of the Monolayers 3.2. Ellipsometry 3.3. Rigidity Measurements 3.4. Brewster Angle Microscopy xiii
1 1 6 10
33 33 34 35 39 45 45 46 53 59 62
xiv Contents
3.5. X-Ray Reflectivity and X-Ray Grazing Incidence Diffraction 3.6. Atomic Force Microscopy 4
5
6
Electron Microscopic Observations and Image Analysis 4.1. Transfer of the Monolayer onto an Electron Microscope Grid 4.2. Preparation of the Specimen for Electron Microscopic Observation 4.3. The Electron Microscope 4.4. Image Processing
66 72
75 75 80 82 86
Practical Considerations 5.1. General Practical Considerations for Soluble Proteins or Multi-Protein Complexes 5.2. Membrane Protein Crystallization 5.3. Single Particle Observation
89
Other New Methods 6.1. DNA Scaffolds 6.2. Nanotubes
95 95 96
89 92 93
Appendix
99
Glossary
113
Bibliography
117
Index
137
Chapter 1
Two-dimensional Crystallization on Lipid Monolayers
1.1. Overview In 1971 it was first shown by Fromherz (1971) that an ordered arrangement of protein can be generated underneath a lipid monolayer. Ferritin molecules were observed in a regular arrangement after adsorption to a lipid film. Limited lateral mobility and heterogeneity in the lipid monolayer resulted in only moderately ordered protein in these early experiments. About 10 years later, for the first time Kornberg’s group was capable of obtaining 2D crystals of an antibody using a monolayer of lipid hapten (Uzgiris and Kornberg, 1983). Since then, this 2D crystallization technique has been successfully employed for a variety of different proteins (see Appendix). These are mainly water-soluble proteins, but recent studies have shown that this technique is also applicable to membrane proteins. This may provide a much needed extension in the repertoire of membrane protein crystallization, thus improving our structural knowledge of this protein class. The crystallization on lipid layers is an elegant method because it is possible to work with very dilute protein solutions and still generate a locally high concentration of protein constrained in 2D. Nonetheless, the proteins retain sufficient mobility to allow for organization into crystalline 2D arrays by lateral diffusion. Lipid monolayers can be spread (driven by surface tension) over the whole air/water interface of a drop to form a flat, one molecule thick film. This provides a substrate for protein binding, leading to a layer of closely packed proteins at the interface which can be organized into a 2D crystal suitable for structure determination by electron crystallography. 1
2 Strategies for Two-dimensional Crystallization of Proteins
The first step in the crystallization process is the adsorption of the protein to a lipid monolayer (Fig. 1.1). This limits the protein to a few orientations relative to the lipid plane which facilitates crystallization. The hydrophilic headgroup of the lipid is responsible for this function and operates as a recognition element for the protein in one of two ways. The first involves electrostatic interactions of a charged lipid layer with the protein allowing non-specific binding to a surface layer. The second occurs by specific binding of protein to a surface monolayer. This is achieved by interaction of the protein with a ligand attached to the polar headgroup of the lipid. Both types of interactions lead to a densely packed protein layer at the lipid/water interface. It is possible to obtain a locally high concentration of protein in the order of 500–1,000mg/mL from a very dilute solution (10–100g/mL) (Kornberg and Darst, 1991). Lipids provide a substrate for protein binding and therefore a functional basis of the crystallization technique. In general, lipids consist firstly of a hydrophilic headgroup which can carry a charge or a functionalized ligand group, for example a Ni2-NTA group, that can be utilized for protein binding via a His-tag. The second part of a lipid is a long hydrophobic tail which usually consists of two acyl chains. It is necessary for the
Fig. 1.1
The method of surface crystallization for soluble proteins.
Two-dimensional Crystallization on Lipid Monolayers 3
molecule to accommodate its hydrophobic and hydrophilic part in a suitable environment. Therefore, amphiphilic molecules like lipids have a tendency to self-organize. In solution, they form lipid bilayers in the form of vesicles or tubes; at the air/water interface they form lipid monolayers. The hydrophobic tails are sheltered from the water by burying within the bilayer or by pointing into the apolar air. In this way interaction of each part with the opposite phase is minimized. This is favorable because the acyl chains are not able to form hydrogen bonds with water. The organization of the monolayer is driven by the hydrophobic effect associated with the lipid tail groups of the lipid molecules which also determines the stability and fluidity of the monolayer. The structure of this water insoluble moiety is responsible for the physical properties of the lipid layer, because the long hydrophobic chains interact extensively with each other. The length and saturation of the acyl chain as well as branches in the chain influence the fluidity of the lipid at a given temperature. Fluidity of a lipid monolayer is given by intramolecular contacts between the hydrophobic lipid chains. This interaction is dependent on the shape of the hydrophobic tails of the lipids. A cis-double-bond introduces a kink in the acyl chain which sterically hinders packing of the hydrocarbon chains in the same fashion as a branched chain and both result in an increased fluidity of the lipid (see Sec. 3.1). It is clear from the literature that the lipid monolayer being in the fluid state is the most favorable condition for the growth of 2D crystals. Although closely packed protein molecules have been found on monolayers in the solid state, 2D crystals were not observed under these conditions (Darst et al., 1991a; Mosser and Brisson, 1991). The physical properties of the monolayer system are determined mainly by the chemical composition of the lipids, the temperature and the composition of the underlying buffer. In order to allow protein crystallization these parameters have to permit the lateral diffusion of the protein molecules attached to the monolayer. To achieve a favorable physical state of the lipid layer it has often proven useful or even essential to use mixtures of different lipids. As these additional lipids usually do not carry a functional group, they are generally referred to as diluting lipids. A reason for the dependence on diluting lipids might be the difference in surface covered by proteins and the much smaller lipids. A phospholipid occupies about 50 times less area than a 100kDa globular protein.
4 Strategies for Two-dimensional Crystallization of Proteins
Therefore, one protein molecule can interact with many lipid molecules. Thus the composition of the lipid layer is of high importance for crystallization, especially in the case of non-specific interactions, as it determines the local electrostatic milieu. Furthermore, it is possible to adjust the fluidity properties for a given monolayer by mixing functionalized lipids with diluting lipids of different structure. Application of the surface crystallization technique for the creation of 2D crystals always bears the possibility of producing hexagonally closely packed protein, which might be mistaken for protein crystals at low resolution. Some proteins crystals like cholera toxin show a true hexagonal space group (Kornberg and Ribi, 1987), other denser areas of the same protein on the grid were found to exhibit a hexagonal diffraction pattern (Ludwig et al., 1986). Further investigation indicated that the diffraction pattern was due to the presence of hexagonally closely packed protein molecules, and not due to a crystal. Surface crystallization trials generally lead to a closely packed layer of protein and are therefore prone to produce this kind of artifact, which rarely lead to 2D crystals. Figure 1.2 shows different organizations of protein which are common for surface crystallization trials. The observation of growing 2D crystals on lipid monolayers can also give insight into the fundamental processes during crystallization (Ku et al., 1993). Due to reduced dimensionality, aspects of the transition between disordered and ordered states can be studied more easily. 2D crystals have also been used to promote epitaxial growth of 3D crystals from 2D crystals from lipid layers. In this case, 2D crystals serve as nuclei for the formation of 3D crystals. Furthermore, it has been established that the lipid layers alone can trigger epitaxial crystal growth (Hemming et al., 1995). In conclusion, there are some general requirements for 2D crystallization of proteins: • • • •
limiting the protein diffusion to a plane; a high concentration of the protein in the plane; orientation of the protein; providing mobility of the protein within the plane to allow sampling of various interaction arrangements.
The physical properties of the lipids spread at the air/water interface, the protein adsorption and binding capacity at the interface, the
Two-dimensional Crystallization on Lipid Monolayers 5
Fig. 1.2 Different stages of protein organization on lipid films: (A) aggregates without order, (B) closely packed hexagonal arrays, (C) 2D crystal.
lipid reorganization induced by protein binding and the crystallization process of the proteins have been extensively studied using numerous biophysical techniques. Some of these techniques include film balance measurements, ellipsometry, light scattering microscopy, epifluorescence microscopy, fluorescence spectroscopy, fluorescence microscopy, phase contrast microscopy, Brewster angle microscopy (BAM), electron microscopy (EM), atomic force microscopy (AFM). More recently, other methods have been developed such as X-ray reflectivity, X-ray grazing incidence diffraction, neutron reflectivity, shear modulus measurements and scanning near field optical microscopy (SNOM). Some of these techniques will be presented in chapter 3.
1.2. Non-specific Adsorption through Electrostatic Interactions Lipids can contain headgroups with positive, negative, or neutral charges (Fig. 1.3). The attraction of opposite electrical charges provides the basis for electrostatic interactions. Charged lipids can be used to create a charged surface that can interact with the surface potential of a protein. This type of interaction can be compared to processes involved in ion-exchange chromatography: Proteins carry positive and negative charges, according to the acidic or basic side chains of single amino acids. At acidic pH values the side chain lysine, arginine and histidine are
6 Strategies for Two-dimensional Crystallization of Proteins
2
3
4
5
6
7
8
9
10
11
Relative change
Relative change
Phosphatidylcholine pka = 1.0 1 0,75 0,5 0,25 0 1 -0,25 -0,5 -0,75
1 0,75 0,5 0,25 0 -0,25 1 -0,5 -0,75
Phosphatidylserine R1-HPO4, pka = 2.5 R-COOH, pka = 5.5 R-NH3+ = 11.5
2
3
4
5
8
9
10
11
8
9
10
11
-1,25 pH
1 0,75 0,5 0,25 0 1 -0,25 -0,5 -0,75
pH
Phosphatidylinositol pka = 2.5
Phosphatidylethanolamine R2-HPO4, pka = 1.7 R-NH3+, pka = 9,8
1 0,75
2
3
4
5
6
7
8
9
10
11
Relative change
Relative change
7
-1
-1 -1,25
-1 -1,25
6
0,5 0,25 0
1
2
3
4
5
6
7
-0,25 -0,5 -0,75 -1
pH
-1,25
pH
Fig. 1.3 Chemical structure, surface representation and relative charge as a function of pH of some common phospholipids.
protonated and a protein shows a cationic behavior. In contrast, at basic pH the negative charges of aspartic acid and glutamic acid render the protein anionic. Therefore, the net charge of a protein is dependent on the pH of the surrounding buffer and the number of exposed charged amino acids on the surface of the protein. The overall charge of a protein is described by the pI value. If the pH of the buffer is at the pI, the net charge of the protein is zero and therefore the capacity of the protein for electrostatic interactions is low, unless there is an unequal distribution of charged amino acids on the surface of the protein, which leads to regions with a positive or negative charge.
This seems to be the case for -actinin which forms 2D crystals on a positively charged lipid layer at the isoelectric point of the protein
Two-dimensional Crystallization on Lipid Monolayers 7
(Taylor and Taylor, 1993). The properties of the surface potential of the protein and the charge of the lipid layer can be controlled by varying the pH. For lipid layers, mixtures of differently charged lipids can provide different charge patterns on the surface facing the crystallization solution. If these conditions are chosen correctly, electrostatic attraction is sufficient to establish a stable interaction between a protein and the lipid layer. The extent of electrostatic interaction can be mediated not only by changing the pH, but also by altering the ionic strength of the aqueous phase. Increasing ionic strength weakens the electrostatic interaction and ultimately prevents protein binding, usually at a salt concentration of above 400mM. The first protein that was crystallized on lipid monolayers exploiting electrostatic interactions was the RNA polymerase from Escherichia coli (Darst et al., 1988). The attraction between the acidic surface of the protein and the positively charged lipid layer was sufficient to generate 2D crystals at the interface. RNA polymerase II is a very good example for the potential of surface crystallization when it comes to large complexes. The catalytically competent core consists of five different subunits and has a molecular weight of about 400kDa. The complex binds to positively charged lipid layers via electrostatic interaction and forms 2D crystals useful for structure determination (Darst et al., 1988). Another large complex, the 50S ribosomal subunit, was crystallized on negatively charged lipids (Avila-Sakar et al., 1994). A group of proteins that shows a natural affinity for negatively charged lipids are annexins. The self-assembly capability of some annexins on cell membranes (Kaetzel et al., 2001) seems to be important for membrane-related processes including exocytosis, endocytosis and vesicle trafficking (Gerke and Moss, 1997; Konig et al., 1998; Raynal and Pollard, 1994). The tendency of annexin to naturally form 2D arrays has been used to grow crystals on negatively charged lipid membranes (Mosser et al., 1991; Newman et al., 1989; Newman et al., 1991). 2D crystals of annexin reproducibly form on lipid monolayers containing at least 5% of the negatively charged phosphatidylserine and are strictly Ca2⫹dependent (Andree et al., 1990). Annexin V (consisting of four 40kDa subunits) exhibits a high affinity to negatively charged lipids, such as phosphatidylserine (KD ⬍ 0.1nM). The structure of lipid bound annexin V
8 Strategies for Two-dimensional Crystallization of Proteins
was presented in 1991 at 20Å (Brisson et al., 1991). The atomic structure of annexin V had simultaneously been solved by R. Huber (Huber et al., 1990). The protein in membrane bound form and its soluble equivalent from X-ray crystallographic work was then compared and the precise orientation of the protein with respect to the membrane was determined (Brisson et al., 1991). Later, a 3D model at 17Å (Voges et al., 1994) and a projection map at 8Å resolution (Olofsson et al., 1994), and then at 6.5Å resolution (Oling et al., 2000) allowed an analysis of the conformational change from soluble to membrane bound form. Another protein with a natural affinity for negatively charged lipids is brush border myosin-I. As with the annexins, this property is linked to the suggested function of mediating vesicle transport (Celia et al., 1996). To date, three different human blood coagulation factors have been crystallized, utilizing their natural affinity to negatively charged lipids (Stoylova et al., 1994; Stoylova et al., 1998; Stoylova et al., 1999). This affinity has a functional importance as the assembly of different factors into active complexes of the blood clotting cascade requires the preceding binding of these factors to negatively charged phospholipids. Using these data, a hypothetical model of membrane-bound factor VIII associated with the classic hemophilia was proposed. A pore-forming protein, ␣-toxin from Staphylococcus aureus, is known to interact with lipids. The interaction leads to the oligomerization of the protein, insertion into the membrane and pore formation. A negatively charged lipid monolayer was used to crystallize the protein (Ellis et al., 1997; Olofsson et al., 1990). Reconstruction of a 3D model of an oligomeric pore provided the first glimpse of the process by which the toxin inserts into the membrane. As the electrostatic interactions are non-specific, the purity of the protein preparation is extremely important to prevent the accumulation of contaminants at the surface which could inhibit the crystallization process. Furthermore, the non-specific nature of the interaction limits the probability that a protein is pre-oriented at the interface which is responsible in facilitating crystallization. Nevertheless, Lebeau et al. (1996) observed that the RNA polymerase, which has an acidic surface, can bind in an orientated way upon interaction with a positively charged lipid. In the same study, the RNA polymerase was observed to adsorb (to a lower extent) to polar but uncharged lipids, implying that
Two-dimensional Crystallization on Lipid Monolayers 9
hydrogen bonds are also involved in the protein-lipid interaction. On the other hand, if only charged lipids were used, without the addition of neutral dilution lipids, no crystals were obtained, probably due to interference of the protein-protein interactions necessary for crystallization by the strongly charged surface. It was shown for the chaperonin TF55 that differently charged lipids resulted in crystals with the protein in different orientations (Ellis et al., 1998). Neutral lipids lead to an interaction of the protein in a sideon manner, whereas negatively charged lipids resulted in an end-on attachment to the lipid monolayer. Although only the latter crystals were ordered well enough for structure determination, it is interesting to note that a lipid-dependent orientation of a protein could be helpful for structure determination by electron crystallography. The missing information due to a limited tilt range (referred to as the missing cone problem) could be filled in by using crystals with a different orientation towards the crystal plane (Bischler et al., 1998). It is possible to form 2D protein crystals at a phase interface other than the lipid/water interface of a monolayer. Aoyama et al. (1995) conducted trials using a thin layer of the organic liquid dehydroabietylamine (DHAA) which carries a positive charge on its amino group (Fig. 1.4). Since it is insoluble in water, it will not denature protein. It forms a sharp interface with the aqueous protein solution. Its positive charge attracted negatively charged proteins to the interface and 2D crystals of various proteins, namely ferritin, catalase, chaperonin and the 50S ribosome were obtained at resolutions of 20–28Å (Aoyama et al., 1995). Crystal sizes for the first three proteins were between 5m and 0.5m whereas the 50S ribosome yielded only very small crystals which could still be used to calculate a 3D map to approximately 28Å. A drawback of using a thin 3D liquid is an elevated noise level due to the greater thickness of
Fig. 1.4 DHAA (dehydroabietylamine).
10
Strategies for Two-dimensional Crystallization of Proteins
the layer compared to a lipid monolayer. A similar organic compound carrying a negative charge has not yet been found. Another alternative to lipid monolayers is the use of a synthetic polypeptide poly(1-benzyl-L-histidine). This polypeptide was spread at the air/water interface and its positively charged imidazole groups were able to attract catalase, ferritin or streptavidin from which 2D crystals grew (Furuno and Sasabe, 1993; Sato et al., 1993). A higher temperature was required for crystallization (up to 40⬚C for 30min) to compensate for the lower fluidity compared to using unsaturated phospholipids. Other proteins have been crystallized using electrostatic interaction, but it seems not to be a general method for every protein. Although we know from ion chromatography that it is theoretically possible to find conditions for an electrostatic interaction for every protein dependent on its surface charge pattern, the lack of specificity in binding and pre-orientation may cause difficulties in succeeding to find crystallization conditions for a particular protein. A more straight forward way is the use of specific adsorption of proteins via ligands, which is described in the next section.
1.3 Specific Adsorption via Ligands The crystallization of protein on lipid monolayer using natural lipid headgroups will be presented first. This approach was used for a small number of proteins. The ability to synthesize lipids carrying various synthetic groups on lipid headgroups has been exploited and was a major breakthrough for the development of the 2D crystallization on lipid monolayers. Functionalized lipids carrying small molecules or metal ions will be subsequently described.
1.3.1. Molecular recognition between protein and lipid headgroup Cholera toxin B/monosialoganglioside Cholera toxin is one of the first proteins that have been crystallized by specific anchoring to planar lipid films (Ludwig et al., 1986). This protein
Two-dimensional Crystallization on Lipid Monolayers 11 OH H HO
CH 2OH O H H NH
HO CH 2OH O H H O OH H H
O H O
H 3C O
H HOOC
H
O H
HO H HO CH 2OH
CH 2OH O H O H OH H
H
C16 H33
H
H
O H
HO
H
H
H HO CH 2OH
NH
H
OH O
C13H27
C H2 H
OH
H OH
H
H O
NH CH3
Fig. 1.5 GM1, monosialoganglioside.
is secreted by the bacterium Vibrio cholerae and is composed of two different subunits A (MW ⫽ 27kDa) and B (MW ⫽ 11.6kDa) assembled with the stoichiometry AB5. The B5 assembly is responsible for binding of the protein to its cellular receptor, monosialoganglioside GM1 (Fig. 1.5). The A1 fragment of the A subunit catalyzes the ADP-ribosylation of the Gs␣ regulatory component of the adenylate cyclase complex (van Heyningen, 1983). The ultimate result of cholera toxin action is an extensive fluid loss characteristic of the disease. A 3D model of cholera toxin B subunit has been calculated at medium resolution (Ribi et al., 1988). The 2D crystals of cholera toxinGM1 complexes are extremely well ordered once transferred onto an electron microscope grid. An electron diffractogram exhibiting electron diffraction peaks extending to a resolution better than 0.35nm has been produced (Mosser et al., 1992) and demonstrates the potential for high resolution electron crystallographic analysis of this type of monolayer crystals.
1.3.2. Functionalized lipids Various synthetic lipids made of a protein ligand coupled to a lipid molecule have been synthetized. These protein ligands are often small molecules: biotin, ATP or ions: Ni2⫹, Cu2⫹; but they can also be larger such as steroid-derivative lipids. Lipids can also exhibit multifunctional groups. Firstly, moiety-carrying lipids (where molecules are attached to lipids) will be described (Sec. 1.3.2.1), then metal chelating lipids and multi functionalized lipids will be presented (Secs. 1.3.2.2 and 1.3.2.3).
12
Strategies for Two-dimensional Crystallization of Proteins
Imagination and creativity helped to originate numerous functionalized lipids. This approach was not only motivated by the hope of producing a general method for 2D crystallization, but also by the need to create biofunctionalized interfaces. The lipid monolayer spread at the air/water interface was then used as a model system to investigate protein functions such as molecular and cellular recognition processes at the lipid interfaces. Biosensing technologies using biofunctionalized interfaces are another intensively studied field of research. It is important to know about the existence of these specially synthesized lipids as they can also be used for setting up 2D crystals. A short presentation of these designed lipids initially developped to create biofunctionalized surfaces will be given in Sec. 1.3.2.4. 1.3.2.1. Moiety-carrying lipids Biotin-lipid streptavidin system The high affinity of streptavidin for biotin (KD ⫽ 10⫺15 M) which is only one order of magnitude lower than for a covalent bond (Green, 1975), is at the origin of many biotechnological applications. The complex has been used for anchoring streptavidin onto biotinylated planar lipid surfaces. Streptavidin is tetrameric with four high affinity binding sites for the small molecule biotin (vitamin H). It has been used extensively in the past for specific adsorption, localization and other analysis of biotin conjugates. When streptavidin, introduced into the subphase under the layer of biotinylated lipids (Fig. 1.6), was studied for the first time by epifluorescence microscopy, the result was striking. Large protein domains up to 200m in diameter were visualized (Blankenburg et al., 1989). This structural study was pursued with image analysis of electron micrographs and led to a 3D model of the streptavidin tetramer at 14Å resolution
O H3C H3C
O O O
O O P O
S
H N
OH
O
HN
NH O
Fig. 1.6 Biotinylated-lipid: biotin-LC-DPPE.
Two-dimensional Crystallization on Lipid Monolayers 13
(Darst et al., 1991a) which was compared to the known X-ray structure (Hendrickson et al., 1989; Weber et al., 1989). Extension of the investigation and development of cryo-specimen preparation techniques allowed the first high resolution projection map (3Å) for an interfacially crystallized soluble protein to be calculated (Avila-Sakar and Chiu, 1996). A comparison of this projection map with the X-ray structure of streptavidin allowed assignment of -sheet structures in the projection map. The excellent agreement of the projected density with the atomic model provides strong evidence that this type of monolayer crystal has the potential for high resolution electron crystallographic analysis. Comparison of the electron and X-ray structures confirmed that biotinbinding sites of streptavidin are available and appropriately orientated to bind additional biotinylated macromolecules in solution. Streptavidin has four binding sites for the biotin, two biotin binding sites are interacting with the biotinylated lipids and the two other binding sites are on the opposite side, facing the solution. This first layer made of a 2D crystal of streptavidin could provide a scaffold for the formation of another layer of protein molecules. Specific binding of biotinylated ferritin to lipid-bound streptavidin was demonstrated by electron microscopy (Darst et al., 1991a). Nonetheless crystallization of streptavidin bound protein using these preformed 2D crystals have not been reported so far despite various attempts. Recently, immobilization of biotinylated DNA on 2D streptavidin crystals has been reported by Crucifix et al. (2004). DNA was endlabeled with a biotin moiety in order to decorate 2D crystals of streptavidin which formed in contact with a biotinylated lipid layer. RNA polymerase was then able to interact with the DNA, and was immobilized by that strategy in various orientations. The immobilized nucleoprotein complexes were then suitable for 3D reconstruction from isolated molecular views, a method which requires usually a homogeneous angular distribution. The use of immobilized DNA while keeping it accessible for interacting partners opens the possibility to drive the macromolecular assemblies into a different functional state by simply changing the incubation buffer. The streptavidin/biotin system provides a convenient model for studying the molecular mechanisms underlying protein crystallization since the crystallization conditions are well known. The relationship
14
Strategies for Two-dimensional Crystallization of Proteins
between lipid-protein molecular recognition, crystallization solution conditions, and the crystal properties such as coherence, space group and morphology is an important issue. A better understanding of these relationships will aid in the design of rational strategies for promoting high-quality protein crystallization and for controlling protein assembly at interfaces in the biomaterial and nanotechnology fields. The combined macroscopic and microscopic observations of the 2D crystals of streptavidin in various conditions shed light on the intermolecular interactions taking place between neighboring streptavidin proteins during the process of 2D crystallization. Depending on the conditions of crystallization, streptavidin crystals can form a variety of macroscopic morphologies. These macroscopic patterns can be observed using a brewster angle microscope (BAM) (Frey et al., 1996) or by epifluorescence microscopy and were compared to the associated microscopic crystal structures observed using electron microscopy (Gast et al., 1999). For instance, lattices with P1 symmetry appear for 1.5 pH 5, P1 and P2 for 5 pH 6, and C222 symmetry for 7 pH 11. P1 crystals nucleate rapidly and form a particular pattern of thin needle-shaped crystals. C222 crystals grow more isotropically and exhibit H- and X-shapes (or butterfly-like shapes) as observed with BAM (Fig. 1.7). The transition from C222 to P1 or P2 crystals can be accomplished in minutes by lowering the pH. Dynamics of 2D protein crystallization growth have been investigated, introducing avidin in the subphase as a non crystallization protein contaminant. The macro-pattern of the 2D crystals changed according to the concentration of avidin in the solution and shifted from dendritic growth to amorphous finger and needle like structure (Ku et al., 1992). The molecular mechanisms underlying asymmetrical growth in 2D streptavidin crystals have been investigated in more detail. Comparison
Fig. 1.7 H- and X-shaped streptavidin crystals observed with a Brewster angle microscope.
Two-dimensional Crystallization on Lipid Monolayers 15
of the macroscopic morphology of 2D crystals of mutated streptavidin with 2D crystals of wild type streptavidin has been presented by Edwards et al. (2002): At neutral pH, streptavidin forms crystals with C222 symmetry and X-shaped morphology that arises from asymmetric growth rates. The molecular mechanism coupling biotin binding and growth asymmetry has been clearly shown using a mutated streptavidin (threonine 20 replaced by alanine). It appears that Tyr22-Thr20 hydrogen bond interaction across the protein-protein contact interface is the crystallization contact that is altered by biotin binding to give growth asymmetry in wild-type crystals. Using the same strategy, it has been shown that the mutant protein crystallizes with a different morphology than the wild type protein when lysine 132 is replaced by leucine (Edwards et al., 1998). Lysine 132 seems also essential for the crystal morphology. The side chains of lysine 132 interact with each other across the dyad-related crystal contacts. These studies and others (Ratanabanankoon and Gast, 2003; Wang et al., 1999; Yatcilla et al., 1998) help to understand the complicated effects of intermolecular interactions on 2D protein crystal growth. This understanding is essential for the production of large and well ordered crystals. Lipid-hapten Early experiments using a lipid monolayer for 2D crystallization of proteins utilized a lipid hapten (DNP-PE) monolayer that had an affinity to a monoclonal antibody (anti-DNP IgG) (Uzgiris and Kornberg, 1983). A monolayer of this lipid hapten was used to form an ordered array of antibodies. After that, the C1q component of the complement was bound to the Fc tail of the antibody. The decorated antibodies also formed an ordered array and allowed the calculation of a projection map of the entire antigen-antibody-complement complex. ATP-lipid The universality of the ATP ligand in nature makes the use of ATP-lipid very attractive: any ATP-binding protein should bind to the ATP/lipid interface, the first step of the 2D protein crystallization process. The use of an effector nucleotide linked to lipids allowed the formation of 2D crystals of the B1 subunit of ribonucleotide reductase.
16
Strategies for Two-dimensional Crystallization of Proteins
The projection map revealed structural features of the subunit up to 18Å resolution (Ribi et al., 1987). The effector-lipid consisted of dATP coupled through the ␥-phosphatyl group and a -aminocaproyl linker to phosphatidylethanolamine (Fig. 1.8A). Another class of hydrolyzable and non-hydrolyzable ATP-lipids where the nucleotides are covalently attached via the C8- or N6-position of the adenine ring to a synthetic lipid has been synthesized (Fig. 1.8B and Fig. 1.8C) (Schmitt and Tampe, 1996). These ATP-lipids can be applied not only to anchor, orient and crystallize ATP-binding proteins at a lipid interface, but they also have the ability to act as an energy source in the 2D plane. These ATP-lipids were characterized by various enzyme assays in
(A)
NH 2 N
N
N
N
O
O HO P
H 3C H 3C
O
O
O
O P O
H N
O
OH
O
O
O OH HO P O
HO
O
N H
O
P O OH
(B) NH 2
H 3C O H 3C
N O
N H
(
)n
H N
N
N
N
O
HO
N
O
O
O P
O P
O X
OH
OH
O
O
P
OH
OH
OH
(C) H 3C H 3C
O N O
N H
(
)n
H N N
N N
N
O
O P O P X OH HO
OH
O P OH OH
OH
Fig. 1.8 ATP-lipids: (A) dATP-aminocaproyl-PE. (B) DODA-HM-C8-AMPPCP; (C) DODAHM-N6-AMPPCP; nonhydrolyzable (X ⫽ CH2) and hydrolyzable (X ⫽ O) lipid. The spacer length can vary from ethylene (n ⫽ 1) to hexamethylene (n ⫽ 3).
Two-dimensional Crystallization on Lipid Monolayers 17
micellar solution, resulting in ATPase and competition activities that are comparable to their free counterparts. Actin was used as a model for an ATP-binding protein attaching to an ATP-lipid. Following binding to ATP-lipid, drastic changes in the viscoelastic properties and shape transitions of vesicles were observed by phase contrast microscopy (Schmitt and Tampe, 1996). Novobiocin-lipid 2D crystals of the Escherichia coli DNA gyrase B subunit were obtained upon specific interaction with novobiocin linked phospholipid films. DNA gyrase is a heterotetramer with an A2B2 stoichiometry. The B subunit has a MW of 90kDa and was shown to bind coumarins such as the antibiotic novobiocin in the nanomolar concentration range. Therefore, novobiocin was an obvious natural ligand to attach to lipids in order to initiate structural studies of the B subunit. The specially designed lipid consists of the antibiotic group, novobiocin, modified and anchored to the lipid at the coumarin region, which preserved its affinity for the protein (Fig. 1.9). A polyethylene oxide hydrophilic spacer was placed between the dioleoylphosphatidic acid and the antibiotic in order to improve accessibility (Lebeau et al., 1996). A projection map (Lebeau et al., 1990) and later a 3D model (Celia et al., 1994) of the gyrase B subunit were produced. Dichlorophenyl-lipids In the pioneering work of Hirata and Miyaka (1994), functionalized lipids (quinonylphospholipid) were designed to bind the bacterial photosynthetic reaction center onto monolayers. Later, new functionalized
O H3C H3C
O
O
O
O P(
O
)n
H O N O
OH
O
O
H2 N
O
O
O
H N O
OH O
O OH O
Fig. 1.9 Novobiocin-lipid. The spacer length can vary from n ⫽ 1 to n ⫽ 4.
18
Strategies for Two-dimensional Crystallization of Proteins
lipids were synthesized (Trudel et al., 2001). These dichlorophenyl-lipids (DCPU-lipids, Fig. 1.10) bound very efficiently to the Qb site of photosystem II core complex, but no crystals were reported. Other specific lipid ligands Phospholipids linked to steroid-hormones derivatives (Fig. 1.11) have been synthetized for the study of the crystallization of progesterone and estradiol receptors (Lebeau et al., 1991). O H 3C
O O
H 3C
(O
N H
)3 O
N NH
Cl Cl
Fig. 1.10 DCPU lipid. O H 3C
O
O
O
O P
H 3C
(O
H O N
)n
O
OH
O
OH
OH O H 3C
H 3C
O
O
O
O P
(O
)n
O O
OH
O
O
H N
O O H 3C
O O
H 3C O
O O
P OH
(O
)n
H N
R O R
Fig. 1.11 Steroid-hormone derivatives lipids designed for two-dimensional crystallization experiments with hormone-binding receptors R ⫽ OH or R ⫽ H, n ⫽ 1–4.
Two-dimensional Crystallization on Lipid Monolayers 19
Glycerolipids linked to hydroxamate derivatives were designed for crystallization of mammalian aminopeptidase M (Altenburger et al., 1992).
1.3.2.2. Metal-chelating lipids (Histidine-tag/Ni2⫹ lipid headgroup) Although the lipid-layer crystallization approach has been applied successfully to a variety of ligands for specific protein-binding, appropriate lipid-based ligands are sometimes difficult to obtain. An alternative strategy is to devise a general adaptor molecule that will link a wide variety of macromolecules to lipid layers through specific binding to a common lipid-based ligand. In this strategy, strong interactions between a Ni2⫹ ion carried by specially synthesized lipids and a His-tag from expressed protein have been successfully used. It was shown initially by electron spin resonance (ESR) that native proteins are targeted to Cu2⫹-IDA lipid assemblies through coordination by surface histidine amino acids (Shnek et al., 1994). The copper lipids have now been almost entirely replaced by the nickel chelates for routine protein binding. A short historical reference to the copper lipid is given at the end of the Sec. 1.3.2.4. NTA-Nickel lipid headgroup design An optimal design was found to be the quadridentate nitrilotriacetic acid (NTA, also called N,N bis[carboxymethyl]glycine (Fig. 1.12A), which binds to Ni2⫹ with three carboxyl groups and one nitrogen (Hochuli et al., 1987). The remaining two ligand positions in the octahedral coordination sphere of Ni2⫹ are available for tight but reversible selective protein interactions. A string of six histidines was found to produce a strong and specific bonding to Ni2⫹ when it was held by NTA, replacing the two water molecules on the Ni2⫹ with histidine side chains. NTA represents an optimal compromise between the tridentate iminodiacetic acid (IDA, Fig. 1.12B) derivatives which bind Ni2⫹ weakly through three coordination positions and the pentadentate tris(carboxymethyl) ethylenediamine, which occupies five coordination positions on Ni2⫹, leaving only one free for protein binding.
20
Strategies for Two-dimensional Crystallization of Proteins (A)
O
O
O
O
O
O N O O
N
O
Ni OH2
OH2
Ni
O
O O HN
N
N N H
protein (B)
O
O O N
O
Ni OH2
OH2 OH2
Fig. 1.12 Ni2⫹-NTA (nitrilotriacetic acid) and Ni2⫹-IDA (iminodiacetic acid) headgroup design. (A) Ni2⫹-NTA; (B) Ni2⫹-IDA.
Histidine-tag Histidine is a relatively rare amino acid, accounting for only 2% of the amino acids in globular proteins. Only half are exposed on the protein surface. Therefore, on average, a 100-residue protein will only have one surface histidine. Consequently design of a protein expression system with His-tag became very attractive. If the high affinity for chelated metals is conferred on a specific protein, it becomes unique and thereby the adsorption of the protein on the lipid monolayer will be unidirectional, a prerequisite for the formation of 2D crystals. Binding properties of Ni2⫹-Histidine The dissociation constant (KD) of 6 His-tag proteins to Ni2⫹-NTA has been measured to be 10⫺13M at pH 8 (Schmitt et al., 1993). This binding affinity is stronger than those of most antibodies which typically range from 10⫺6 to 10⫺12M (Harlow and Lane, 1988). Only the avidin/biotin system shows a stronger binding with a KD of 10⫺15M (Green, 1975). A segment made of 10 histidines binds even tighter
Two-dimensional Crystallization on Lipid Monolayers 21
(Hoffmann and Roeder, 1991) but seems to be unnecessarily long, so six histidines are commonly used. The pKa of the histidine is close to 7.0. The proper interaction between histidine and a Ni2⫹-complex occurs in the pH range 7.5–8.5. The interaction between Ni2⫹ and nitrogen in the imidazole ring is ionic and can be abolished by protonating these nitrogens, lowering the pH below the pKa value. The typical elution conditions for stripping a His-tag protein off a Ni2⫹ column is to lower the pH to a value ⬍5 (Schmitt et al., 1993; Schmitt et al., 1994), to use a pH gradient from 6.0 to 4.0, or to add a chelating agent (up to 100mM EDTA) (Paborsky et al., 1996). Milder conditions are nevertheless required to avoid denaturation of proteins at low pH (Hoffmann and Roeder, 1991). It was found that imidazole could be used instead of change of pH, competing with the imidazole groups of the histidine residues for the Ni2⫹. Typically, with a standard buffer and pH adjusted to 8, 50 to 200mM imidazole is necessary for a purification or elution step. Effect of imidazole In the crystallization trough, no adsorption of His-tag recombinant HIV1 reverse transcriptase on a Ni2⫹-lipid monolayer was observed when imidazole (100mM) or EDTA (1mM) was present in the protein buffer (Kubalek et al., 1994). Imidazole displaces the protein by competing for the Ni2⫹-NTA group. In the case of the 2D crystallization of yeast RNA polymerase I on Nickel-chelating lipids, a minimum imidazole concentration was required to form crystalline areas (40–90mM) (Bischler et al., 1998) whereas at a concentration higher than 200mM imidazole, the protein no longer bound. The low concentration of imidazole required for the enzyme to crystallize on these lipids reflects the behavior of His-tag proteins in immobilized metal ion affinity chromatography. Small amounts of imidazole are believed to prevent non-specific binding (mediated, for instance, by surface histidines) and favor the specific interaction with the engineered 6 His-tag. Interestingly, the injection of imidazole (200mM final) into the crystallization trough, after crystals had formed, did not provoke any major desorption of the 2D crystals of His-HupR (a transcription factor from
22
Strategies for Two-dimensional Crystallization of Proteins
Rhodobacter capsulatus) as shown using ellipsometric measurements (Courty et al., 2002). However, the observation of the electron microscope grids after transfer of the monolayer showed that the crystals were smaller and contained multiple defects when compared with crystals before the addition of imidazole. A partial complexation of the Ni2⫹ ion with imidazole occurs as the interaction with imidazole is weaker than that with hexa-histidine (KD ⫽ 10⫺8M compared with 10⫺13M at pH 8.0 (Schmitt et al., 1993)). In addition, the cohesive force within a crystal resulting from protein-protein interactions contributes to prevent a massive desorption of the protein. In a separate experiment, 200mM imidazole initially present in the protein solution prevented any adsorption of proteins to the monolayer. Effect of EDTA, DTT and various ions A concentration of 10mM EDTA prevented crystal formation of RNA polymerase I (Bischler et al., 1998), EDTA disrupts binding of protein by fully complexing the Ni2⫹ ions. The addition of divalent or trivalent cations at a final concentration of 5mM had different effects depending on the nature of the ion. Poor transfer of the RNA polymerase I crystals onto the electron microscope grid was observed in the presence of Fe3⫹, Zn2⫹, or Cu2⫹, whereas crystals could be observed in the presence of Ca2⫹ and Mg2⫹. Although all multivalent cations should be able to outcompete the Ni2⫹ ions, this distinct behavior probably reflects the variable affinity of the divalent cations for the NTA moiety and the 6 His-tag. The effect of additional molecules interfering with the 2D crystallization of His-MoCa — a viral capsid protein — has been reported by Barklis et al. (1997). A solution containing 5mM NiCl2, 10mM EDTA, 5mM DTT, and 5mM -mercaptoethanol prevented the adsorption of the His-protein to the Ni2⫹-lipid monolayer. In the case of a membrane protein, the proton-ATPase from plant plasma membrane, the addition of 1mM EDTA and 0.25mM DTT did not interfere with the formation of crystalline monolayers. These conditions were similar to those used routinely for column chromatography of the His-tag ATPase on Ni2⫹-chelating columns. Control experiments in the presence of 10mM EDTA, 150M NiSO4 or 250mM imidazole which would release the His-tag protein from the NTA column did not yield any protein crystalline sheets (Lebeau et al., 2001).
Two-dimensional Crystallization on Lipid Monolayers 23
Addition of EDTA in the trough after crystallization of streptavidin on Cu2⫹ lipid monolayers (adsorption through a naturally exposed histidine) lead to vacancies on the layer observed with light scattering microscopy (Schief et al., 2000). Effect of pH For the crystallization of His-tag proteins on Ni-lipid monolayers, a pH range of 7.5–8.5 is usually used for the crystallization buffer. Upon injection of concentrated HCl into the subphase to lower pH to 6.5, a rapid and complete desorption of 2D crystals of His-HupR from the lipid layer is observed using ellipsometric measurements (Courty et al., 2002). Effect of high salt, detergent, glycerol and other buffer conditions Another interesting property of the interaction of 6 His-tag with Ni2⫹NTA is that it is virtually unaffected by high salt (up to 1M), or by various detergents such as 10mM Triton X-100 or up to 1% Tween 20, 10mM dodecyl-maltopyranoside (Lévy et al., 1999), 0.1% (w/w) dodecyl maltoside, (Lebeau et al., 2001), organic solvents, ethanol or glycerol up to 30%, 8M urea or 6M guanidine HCl. A variety of buffer conditions are thus possible making this method especially suitable for membrane protein crystallization. Crystallization of His-tag proteins In this paragraph, examples of various proteins crystallized into 2D crystals on a Ni-lipid monolayer will be presented. A small number of Nilipids are available. They differ by the structure of the headgroup (ethanolamine, stearilamine), and the aliphatic chains (the length, saturated or unsaturated), they can be phospholipids or just lipids. The lipid monolayer is usually made of a mixture of lipids: the binding lipid and diluting lipids which will give the necessary fluidity of the lipid bilayer. The diluting lipids are often DOPC, DOPE or DOPS. It is useful to try various diluting lipids such as those cited above and others such as E. coli lipids. The molar ratio binding lipid:diluting lipid conducive to 2D crystals is usually between 1:1 and 1:15. The use of the metal chelating phospholipid Ni-NTA-DOPE (Fig. 1.13A) forming the monolayer was successful for the crystallization
24
Strategies for Two-dimensional Crystallization of Proteins
(A) O O
O
O
O
H 3C
O
O O
H 3C
P
O
N H
OH
O
H N
N
O
Ni
O
O
O
OH 2
OH 2
O
(B)
O O H 3C H 3C
OH
O O
O
O
H N
N
O
Ni
O O
OH 2
OH2
(C)
O O H 3C
O O
H 3C
O
H N
O
N
Ni
O
O
O
O
OH 2
OH 2
O
(D)
O O H 3C H 3C
O
H N
N
N
Ni
O
O
O
O
OH 2
OH 2
(E)
O O O
H 3C
O O
O O
H3C
O
H N
N
Ni
O O
O
OH 2
O OH 2
O
Fig. 1.13 Ni-lipids used for 2D crystallization of protein. (A) Ni-NTA-DOPE; (B) Ni-DHGN; (C) Ni-NTA-DOGA; (D) Ni-NTA-DODA; (E) Ni-NTA-DOGS.
of the His-tag recombinant HIV-1 reverse transcriptase (Kubalek et al., 1994). 2D crystals preserved in negative stain diffracted strongly to ~21Å. Other successful crystallizations were achieved using Ni-DHGN, a saturated metal chelating lipid (Fig. 1.13B). His-tag Moloney murine leukaemia virus capsid protein (His-MoCa) formed extensive 2D protein crystals with reflections out to 9.5Å resolution (Barklis et al., 1997).
Two-dimensional Crystallization on Lipid Monolayers 25
The His-tag MoCa arrays show some of the retrovirus core (Gag) protein interactions which occur in assembling virus particles. These 2D crystals were of good quality and a 3D structure of this tagged protein has been determined (McDermott et al., 2000). Another functionalized lipid Ni-NTA-DOGA (Fig. 1.13C) was successfully used to grow 2D crystals of His-tag yeast RNA polymerase I (Bischler et al., 1998). More interestingly, RNA polymerase I which was 6 His-tagged on two different subunits yielded two different crystal forms, the orientation of the enzyme in both crystal forms being different. This is a significant advantage of the technique of 2D crystallization on lipid monolayer as specific His-tags, (or any other peptide tag) can be inserted into the protein of interest at multiple sites using recombinant DNA technology. This may allow the protein to be bound to a lipid monolayer in a variety of different orientations, thus overcoming the “missing cone” problem in electron crystallographic analysis. Combining data from protein crystals in which the molecules are oriented at different angles with respect to the crystal plane, would allow structural data to be collected effectively at all angles. A variation of the Ni-NTA-DOGA, the saturated Ni-NTA-DSA was also tested with the RNA polymerase I but was not conducive to crystallization. A transcriptional factor from the photosynthetic bacterium Rhodobacter capsulatus, His-tag HupR has also been crystallized in 2D on a monolayer of Ni-NTA-DOGA, the crystals diffracted up to 9Å resolution in ice (Vénien-Bryan et al., 1997; Vénien-Bryan et al., 2000). HupR belongs to the nitrogen regulatory protein (NtrC) subfamily of transcriptional regulator. To date, none of the proteins belonging to the NtrC family have been crystallized in 3D form. These proteins are difficult to crystallize due to their propensity to aggregate and also the presence of large flexible domains inside the proteins, which does prevent crystallization in 3D but accommodate in the 2D geometry. An elegant illustration of the potential of a protein-engineered His-tag to immobilize macromolecules in a predictable orientation at metal-chelating lipid interfaces is found in the work of Thess et al. (2002). The recombinant 20S proteasome was His-tagged in various positions, the lipid monolayer was made of the chelating lipid: Ni-NTADODA (Fig. 1.13D). Proteasomes His-tagged at their sides displayed exclusively side-on views, as seen on the electron micrographs, while
26
Strategies for Two-dimensional Crystallization of Proteins
proteasomes His-tag at their ends displayed exclusively end-views. This oriented immobilization of His-tag proteins at lipid interfaces can assist structural studies, not only 2D crystal but also single molecule analysis. Isotopic structural informations is achievable when viewing of the proteins in all possible orientation is possible. Often single particles have a preferred orientation on the electron microscope grid. The use of a lipid monolayer can help to force the protein in another orientation on the grid and gives access to structural information hidden otherwise. Ni-NTA-DOGS (Fig. 1.13E) was essential for the crystallization of soluble proteins and membrane proteins. The structure of a membrane-bound murine molecule (MHC) and its relationship with the membrane was studied after crystallization on a Ni-NTA-DOGS lipid monolayer (Celia et al., 1999). This molecule is expressed at the surface of cells and plays the role of presenting peptides to T cells. The helper component proteinase (HC-Pro), a key protein encoded by plant viruses of the genus Potyvirus was crystallized with the same Ni-lipid, image analysis of cryo micrographs gave some structural information up to 9Å resolution (Plisson et al., 2003). 2D arrays of a recombinant fragment of human vascular endothelium (VE) cadherin were produced on a lipid monolayer of Ni2⫹-NTADOGS (Al-Kurdi et al., 2004). The 2D crystals preserved in uranyl acetate diffracted to 17Å. VE cadherin crystallized in 2D forms compact dimers in contact with the lipid monolayer. It was suggested that this compact cis-dimeric state may occur on isolated cells and that this compact form may serve to protect the molecule from degradation. The system of crystallization on a monolayer of lipid is able, as illustrated in this example, to mimic the behavior of biological molecules at the cell surface. Ni-NTA-DOGS were used for the crystallization of two membrane proteins, FhuA, a high-affinity receptor from the outer membrane of E. coli, and F0F1-ATP-synthase from thermophilic Bacillus PS3 (Lévy et al., 1999). Conditions were found to avoid solubilization of the lipid layer. Image analysis of these negatively stained crystals showed structural information up to 15Å for FhuA and 30Å for F0F1ATP-synthase. Partially fluorinated Ni 2⫹ -chelating lipids (Fig. 1.14) that formed stable monolayers at the air/water interface even in the
Two-dimensional Crystallization on Lipid Monolayers 27
Fig. 1.14
Chemical structure of different fluorinated lipids.
28
Strategies for Two-dimensional Crystallization of Proteins
presence of detergents were used to prepare 2D crystals of the plasma membrane proton-ATPase from Arabidopsis thaliana which diffracted up to 8Å resolution (Lebeau et al., 2001), details are given in Sec. 2.4. 1.3.2.3. Multifunctionalized lipids The structures of lipids are very versatile and allow not only the design of monofunctional groups, but also multifunctional groups. A headgroup containing both biotin and NTA chelator moieties have been synthesized in order to bind streptavidin- and polyhistidine-tagged proteins (Drakopoulou et al., 2000). This bifunctionalized lipid 1,2-dipalmitoyl-sn-glycero-3-phosphoethanolamine, (Fig. 1.15) is believed to form a helical tube rather than a monolayer structure. This work was based on previous results where it was shown that biotinylated dioctadecylamine (DODA) molecules were capable of forming tubular structures. Addition of streptavidin to the tubes produced ordered helical streptavidin arrays. Although it is difficult to predict whether the addition of an NTA-moiety close to the biotin group could interfere with the formation of tubes, it has been shown in the case of galactosylceramide tubes that the addition of NTA-lipids does not prevent their formation (Wilson-Kubalek et al., 1998; Wilson-Kubalek, 2000). See also Sec. 6.2.
O H 3C H 3C
O O O
OH
O
O
O O P
O
N H
N H
S
H N O HN
NH
O
NH O
O
O O N
Ni
O O
OH2
O OH2
Fig. 1.15 Bifunctionalized lipid containing both biotin and NTA chelator moieties.
Two-dimensional Crystallization on Lipid Monolayers 29
1.3.2.4. Immobilization of proteins on biofunctionalized surfaces using synthetic lipids A variety of lipids and biophysical methods have been developed along with the study of immobilization of proteins on biofunctionalized surfaces. These designed lipids and techniques have greatly benefited the field of 2D crystallization. Functionalized lipids which have been used for coating surfaces or forming monolayers at the air/water interface in order to immobilize the biomolecules in a defined orientation at the interface is presented in this chapter. Biophysical experiments for the observation of the binding of proteins are briefly described. Moiety-carrying lipids Lipid hapten synthesis has been described (Balakrishnan et al., 1982) and these lipids have been used to form a monolayer, coating quartz microscope slides to serve as specific antibody-dependent targets for rat basophil leukaemia cells (Weis et al., 1982). It also has been shown that the use of lipid-hapten and specific anti-hapten IgG antibody was required for the binding of guinea pig peritoneal macrophages to planar lipid monolayers on alkylated glass (Hafeman et al., 1981). Proteins such as antibody fragments have been used as recognition moieties and extensively developed for biotechnological application. Immuno-liposomes have been exploited as vehicles for targeted drug delivery and for gene therapy (Bendas, 2001; Maruyama, 2000). Liposomes endowed with specific binding functions have been employed as simplified model systems to study ligand-membrane receptors interactions (Egger et al., 1992, Lee et al., 1993). The progress in the bacterial expression of functional antibodies such as Fab fragments and as single-chain molecules motivated the use of genetic engineering to convert antibodies into membrane-bound molecules for immuno-liposomes applications (Laukkanen et al., 1994; Martin et al., 1981). Lipids have also been modified using sugars (Haensler and Schuber, 1988; Spevak et al., 1993), and NAD⫹ (Salord et al., 1986), these synthetic lipids are good candidates for 2D crystallization providing an anchor for protein interacting with these groups.
30
Strategies for Two-dimensional Crystallization of Proteins
Metal-chelating lipid Coordination of Cu2⫹-IDA lipid by surface histidine amino acid has been first described using ESR (Shnek et al., 1994). The interest for Cu2⫹-lipid then developed as it became a tool for understanding the process of binding and immobilization of proteins on a metal-chelated monolayer of lipids. Among various proteins studied, streptavidin has been chosen most often because the adsorption and crystallization conditions for this test protein are well characterized. The organization of streptavidin was monitored using fluorescence microscopy and Brewster angle microscopy (Frey et al., 1996; Pack et al., 1997a; Schief et al., 2000; Vogel et al., 1997). Binding of native streptavidin to the Cu-DOIDA lipids (Fig. 1.16A) spread at the air/water interface occurred via a His-87 located on the protein surface near the biotin binding pocket. BAM reveals 2D streptavidin crystals with a microscopic shape that differed from crystals that formed beneath biotinylated lipids. DOIDA was modified with a fluorescent pyrene moiety to produce Cu-PSIDA (Fig. 1.16B). These chelating lipids served as reporters of membrane reorganization induced by binding of ligands as shown using fluorescence spectroscopy and fluorescence microscopy (Ng et al., 1995; Pack et al., 1997b).
(A) O
O H3C
H3C
O O
O O
O
Cu OH 2
OH 2
(B)
O
O H 3C
O
N
O
O O
O O
O
O
N
O
Cu OH 2 OH 2
Fig. 1.16 Cu-lipids used for adsorption of proteins on solid surface (A) Cu-DOIDA; (B) Cu-PSIDA.
Two-dimensional Crystallization on Lipid Monolayers 31
Ni-NTA-DODA and Ni-NTA-DPPE were also designed to achieve a defined deposition and immobilization of His-tag biomolecules (heat shock factor HSF24 and peptides containing oligohistidine residues) (Dietrich et al., 1995; Dietrich et al., 1996; Schmitt et al., 1994). The NTA chelator was coupled either to a phospholipid, DPPE, or to a synthetic lipid, DODA. Epifluorescence microscopy was used to follow the binding process.
This page intentionally left blank
Chapter 2
Two-dimensional Crystallization of Membrane Proteins
2.1 Overview A simple view of biological membranes is that they consist of a continuous lipid bilayer in which membrane proteins are embedded. The functions of membranes, on the other hand, are very diverse and are determined by the types of lipids and membrane proteins within the bilayer. This is reflected in the quantity of proteins within the membrane that can vary from 25% in myelin membranes, which insulate the nerve cell axons to 75% in thylacoid membranes of chloroplasts. The general function of membranes is to separate and isolate cells from each other and from the extracellular medium; the membrane serves as the interface between the machinary within the cell and the extracellular fluid. In Eukaryotic cells, elaborate systems of internal membranes create various membrane-enclosed compartments within the cells. Membrane proteins are required to provide a means of exchange across the membrane as the membrane is impermeable to many molecules. Hydrophobic molecules such as hydrocarbons or oxygen can cross the membrane because they can dissolve in the lipid bilayer. Small polar, uncharged molecules like water and carbon dioxide are also able to diffuse across the bilayer. Large polar, uncharged molecules like sugars can not cross the membrane. The same is true for charged ions (Na⫹, Cl⫺, etc.). Carrier proteins, energy dependent transport systems and pores allow a controlled exchange of these compounds and all metabolites necessary for cell function. Membrane proteins are associated with important enzymatic reactions. Here, enzymatic reactions of non-polar substrates are possible, 33
34
Strategies for Two-dimensional Crystallization of Proteins
like the biosynthesis of lipids in the endoplasmatic reticulum. In the specialized membranes of mitochondria and chloroplasts, the core reactions of energy conversion, oxidative phosphorylation and photosynthesis take place. Membrane proteins are not only important for enzymatic processes, but also play a crucial role in cell and organelle morphology. Furthermore, membrane proteins control key aspects in signal transduction of the neural system in higher organisms. Synapses are highly specialized contact points between nerve cells. Neurotransmitters cross the synaptic cleft to trigger membrane bound receptors in the next nerve cell, transmitting signals from cell to cell. Signals which are directed into the cell are generally transmitted via G-protein coupled receptors, which make up about 1% of all proteins encoded in the genome of higher organisms (Herz et al., 1997). In view of the diverse functions and importance of membrane proteins, it is not surprising that they are target for many drugs. Structural information is required to understand many fundamental biological processes and also to find new starting points for the development of drugs. About one quarter of all gene products in higher eukaryotes are membrane proteins (Frishman and Mewes, 1997). Despite their fundamental importance, structural information is rare and lacks far behind the structural knowledge for water-soluble proteins. Most membrane proteins that have been crystallized to date show a high natural abundance. For this reason, the available structural information is probably not representative for this class of proteins. More than half of the membrane proteins for which a structure is known, belong to -sheet proteins, even though the majority of membrane proteins are likely to contain ␣-helical bundles in the transmembrane region. In the group of membrane proteins with a ␣-helical transmembrane region, available structures mostly contain a prosthetic group, again something that is not typical for this class of proteins ( Jones et al., 1991). Until overexpression of membrane proteins is a more standard procedure and crystallization techniques improve, we will only have a limited insight into the structure and function of this important class of proteins. Compared to the structure determination for soluble proteins, where Xray crystallography and NMR predominate, electron crystallography of 2D crystals has proven to be a valuable technique to attain structural information for membrane proteins.
Two-dimensional Crystallization of Membrane Proteins 35
2.2. Naturally Occurring and Induced 2D Crystals 2D crystals of membrane proteins occur naturally in the cell membrane of some bacteria. Amongst these are porins (Kessel et al., 1988) and fumarate reductase from E. coli (Cole et al., 1985). In the plasma membrane of the archea Halobacterium halobium, bacteriorhodopsin forms large 2D crystals. Henderson and Unwin used these crystals as the starting point for the development of the method to determine protein structure from 2D crystals. The result was the first 3D model of a membrane protein (bacteriorhodopsin) obtained by electron crystallography (Henderson and Unwin, 1975; Unwin and Henderson, 1975). Another way to generate 2D crystals in situ is to over-produce transmembrane proteins. If the protein is integrated into the membrane in high amounts, there is a chance that the protein will crystallize in the membrane. Even if it is only highly enriched in the membrane, other method of generating 2D crystals can be used. Phospholipase A2 treatment is a common method of reducing the amount of lipid in the membrane, and improving the crystal quality. The treatment with detergent can have a similar effect. The detergent might also remove some unwanted classes of membrane proteins present in the native membrane. The crystalline order can often be improved by altering the temperature during incubation of the crystallization experiment. However, most membrane proteins occur naturally only in rather small amounts. Also over-expression of membrane proteins are often produced in an amount that is not sufficient for in situ crystallization. This limits the application of this approach.
2.3. 2D Crystallization by Reconstitution into a Lipid Bilayer The localization of membrane proteins within the lipid bilayer requires that they possess both, hydrophilic and hydrophobic surfaces. This renders them insoluble in aqueous solution and makes the use of detergents, which mask the hydrophobic areas necessary for manipulating membrane proteins (Fig. 2.1). As the detergent micelle is not the natural environment for a membrane protein, it can cause a variety of problems during handling and crystallization of these proteins. This may include loss of function, conformational changes and precipitation.
36
Strategies for Two-dimensional Crystallization of Proteins
Solubilization, Purification
Addition of lipids Detergent removal: Dialysis, BioBeads® ...
Fig. 2.1 Solubilization, purification and reconstitution of membrane proteins.
Another problem is that over-produced membrane proteins tend to aggregate into insoluble inclusion bodies. To recover the protein from inclusion bodies, purification under denaturizing conditions is necessary, but refolding of membrane proteins is rarely successful. 2D crystals are grown during a reconstitution of the solubilized protein into membranes by removal of the detergent using micro dialysis or by adsorption of the detergent to small polystyrol beads (BioBeads®). Reconstitution has the advantage that it is possible to choose from a wide range of natural and synthetic lipids, which increases the chances of crystallization. There have been reports of obtaining crystals by mixing all components in the right ratio without detergent removal. This method was successfully used for the light harvesting complex LHC II (Wang and Kühlbrandt, 1991) and bacteriorhodopsin (Michel et al., 1980), but it is not a general method. The recent development of a machine for controlled dilution for membrane protein reconstitution and crystallization provides an interesting alternative to standard detergent removal techniques (Remigy et al., 2003). Utilizing this method, slow dilution of a ternary mixture of protein, lipid and detergent under the critical micellar concentration of the detergent succeeded in a formation of vesicles containing ordered arrays of protein. Progress of the experiment is recorded by light scattering which allows detection of the
Two-dimensional Crystallization of Membrane Proteins 37
formation of proteoliposomes. The validity of this approach was shown with different proteins. This technique allows determination of one of the most crucial points in 2D crystallization by reconstitution, which is the point when protein is re-incorporated into the lipid bilayer. This is an important difference to other techniques like dialysis where this step is not controlled, but happens during removal of detergent more or less rapidly — determined by the dialysis surface and the temperature. The possibility to control the onset of reconstitution, in terms of time and temperature, will help to find crystallization conditions more quickly. A typical crystallization experiment uses 50–100L of a solution containing the purified protein in a detergent solution at a concentration of 0.5–2mg/mL and lipids at a protein-to-lipid ratio ranging from 0.1–10(w/w). The reconstitution occurs by removal of the detergent. Dialysis is a common method to achieve this. There are several different dialysis devices available (Fig. 2.2). The simplest one is dialysis tubing. It is available in various volumes and it is inexpensive. The dialysis surface is large, but it is difficult to collect test samples during the dialysis. The so called “hockey sticks” are bent glass tubes which are sealed by a dialysis membrane on one end (Kühlbrandt, 1992). The sampling is easy, but the dialysis surface is small. Dialysis slides (Slide-A-Lyser®) have a large surface area, provide for easy sampling, but are expensive for extended screens. Dialysis buttons are available in various volumes and they are easy to handle. The dialysis surface is sufficient, but sampling during the experiment is difficult. An alternative to dialysis is the use of BioBeads®. These are small hydrophobic polystyrol beads which bind detergent very efficiently. Detergent removal occurs very quickly. This might not be ideal for every
(A)
(B)
(C)
(D)
Fig. 2.2 Different dialysis devices: A — tubing, B — dialysis button, C — “hockey stick”, D — dialysis cassette (“Slide-A-Lyser®”).
38
Strategies for Two-dimensional Crystallization of Proteins
system, but BioBeads® are very easy to use and it is always worth to try them. Important parameters at the beginning of 2D crystallization experiments are the type of lipid and the lipid to protein ratio. The quality of the lipid preparation is of great importance. It is possible to use lipid extracts from natural sources as well as synthetic lipids, but it can not be generalized as to which is more successful, depending on the specific protein. Very important is the detergent that is used to keep the membrane protein soluble. If the critical micellar concentration (CMC) of the detergent is very low, detergent removal can take a long time. It often seems advantageous to solubilize the lipid in a different detergent that has a higher CMC (Williams, 2000). Of course the composition of the buffer (pH) and its additives like salt and precipitants (glycerol, etc.) are important, too. The temperature is another factor which should be controlled carefully. The hydrophobic effects within the membrane become more prominent with rising temperature. Sometimes it helps crystallization to induce a lipid phase transition by altering the temperature. The phase transition temperature of the lipid used should be kept in mind when choosing the temperature at which the experiment is carried out. When the first crystals have been generated, the most important parameters for the improvement of the crystal quality are temperature cycles. Freeze-thaw cycles have been especially successful for the cytochrome b6f-complex (Mosser et al., 1997). The readdition of detergent sometimes leads to an enlargement of the crystalline vesicles and better ordered crystals (Chami et al., 2001; Zhuang et al., 1999). Here again, digestion with phospholipase can improve the crystal order. If reconstitution into a lipid bilayer is successful and 2D crystals form, they can have different morphologies (Fig. 2.3). They can form extended sheets as well as vesicles. Because the vesicles collapse on the EM grid, there are often two lattices present — from both faces of the vesicle. In practice this is not a problem, because most of the times one lattice is much more prominent than the other. Even if both lattices are visible, they can be separated during image analysis. A third possible crystal morphology is a tubular arrangement. If the tubes are wide and resemble elongated vesicles they can be treated in the same way as the vesicles. If they have a diameter of only a few tens of nanometers, they do not collapse and must be analyzed by helical reconstruction
Two-dimensional Crystallization of Membrane Proteins 39
Fig. 2.3 Vesicle morphology: Electron microscopic image of negatively stained samples representing different crystal forms of the cytochrome b6f-complex form spinach. (a) Tubular crystals projecting from a vesicle, (b) vesicle with mosaic areas of ordered protein, (c) multilayered crystal, the bar represents 0.1m.
(Brisson and Unwin, 1984; DeRosier and Moore, 1970b; Toyoshima and Unwin, 1988).
2.4 Surface Crystallization of Membrane Proteins Application of surface crystallization on lipid monolayers to membrane proteins is complicated by the tendency of detergents to solubilize monolayers of regular lipids. Recently it was reported that two histidine-tagged membrane proteins could be crystallized on a monolayer of conventional lipids derivatized with a Ni2-chelating headgroup, NiNTA-DOGS (Fig. 1.13) (Lévy et al., 1999). Under a preformed layer of functionalized lipids, a mixture of protein, detergent and reconstituting
40
Strategies for Two-dimensional Crystallization of Proteins
lipid was injected. A fast adsorbance of the protein to the preformed lipidic surface layer via ligand interaction prevented the detergent from solubilizing the surface lipid film. Subsequent detergent removal led to a reconstitution of the protein into a lipid bilayer which allowed the organization into 2D crystals. The experiments depended on the use of a low CMC detergent (CMCs: TritonX100 ⫽ 0.23mM, n-dodecyl--Dmaltopyranoside ⫽ 0.17mM) and protein with a very high binding affinity for the functionalized lipid in order to stabilize the interface. After injection of the protein solution, there is a competition between the protein and the detergent for the lipid layer. If, (due to its high binding affinity) the protein is able to adsorb to the monolayer before it is solubilized by the detergent, the protein layer will eventually protect the lipid layer from being dissolved. For this reason, low CMC detergents are favorable for this kind of experiments because they are active in keeping the protein soluble at low absolute detergent concentrations. Yet, the dependence on a low CMC detergent is a restriction which may limit the general application of this approach. To avoid solubilization of the lipid monolayer by the detergent, L. Lebeau and C. Moskowski in Strasbourg developed a new class of partially fluorinated lipids (Fig. 1.14). These lipids are not only hydrophobic, but also lipophobic. Therefore, an aqueous mixture of fluorinated lipids and traditional lipids will result in a three phase system composed of the buffer, the fluorinated lipid and the traditional lipid. To assess the suitability of this new class of lipids for surface crystallization, monolayer properties of two of the fluorinated lipids were investigated by mechanical and optical techniques. Isothermal surface pressure versus area curves were determined using a Langmuir trough equipped with an electrobalance (Lebeau et al., 2001). Profiles indicate that the two fluorinated lipids form stable monolayers at the air/water interface. As discussed earlier for surface crystallization with soluble proteins, it is important that the lipid matrix is spread in the fluid liquid-expanded phase to allow in-plane movements of the lipid-protein complex in order to facilitate lateral reorganization into 2D arrays. In this case, necessary fluidity is provided by branched aliphatic chains, which prevent tight packing of the lipids (Fig. 1.14). To establish that the fluorinated monolayers have a sufficient stability in the presence of detergents, ellipsometric measurements were carried out (Lebeau et al., 2001). This non-invasive method is sensitive to
Two-dimensional Crystallization of Membrane Proteins 41
the density and thickness of the interface layer and therefore especially useful for investigation of surface monolayer behavior. For a detailed description of this technique see Sec. 3.2. The ellipsometric angle ␦ is taken to be zero for a pure air/water interface. Changes in the ellipsometric angle are correlated with material adsorbed to the interface. In this case, spreading the fluorinated lipid at the interface to form a monolayer produced a shift in ␦ of 7(⫾1)⬚. Addition of the detergent Emulgen 911 into the aqueous subphase resulted in a transient perturbation at the interface, which disappeared after a few minutes and the initial value was restored. The monolayer was stable after repeated injection of detergent, even at 5-fold CMC over a period of days. Similar results were obtained with a variety of detergents with an extended range of CMC values (CMCs: dodecyl--D-maltoside, 1.8mM; CHAPS, 8mM; sodium dodecyl sulfate, 2.6mM and octyl -D glucopyrannoside, 19mM) and showed that the partially fluorinated lipids are resistant to solubilization when spread at the air/water interface. The stability of the monolayer in the presence of detergent is a prerequisite for the work with membrane proteins. It is also possible to functionalize the headgroup of the fluorinated lipids with a Ni2⫹-NTA group, which provides a strong affinity to a His-tag that can be engineered on any overexpressed membrane protein. This opens a wide field of membrane proteins available for experiments using surface crystallization. During the experiments, fluorinated lipids with linkers of different length between the hydrophobic part and the functional Ni2⫹-NTA-group were used (Fig. 1.14). A longer linker may provide more freedom for rearrangement of the protein to facilitate the formation of ordered arrays. As with surface crystallization for soluble proteins, it is important to find a suitable mixture of functional Ni2⫹-NTA-lipids and diluting lipids to provide the optimal fluidity properties for crystallization. Differently charged diluting lipids may be important to enhance favorable electrostatic interactions necessary for crystallization. The H⫹-ATPase from the plant Arabidopsis thaliana which was expressed in yeast with a C-terminal His-tag (Jahn et al., 2001) served as a test case for the crystallization of a membrane protein using lipid monolayers. It became the first membrane protein to be crystallized with the new method using fluorinated lipidic monolayers (Lebeau et al., 2001).
42
Strategies for Two-dimensional Crystallization of Proteins
Fig. 2.4 The method of surface crystallization for membrane proteins using fluorinated lipids.
The method works as follows (Fig. 2.4): Detergent solubilized protein is mixed with solubilized lipid and put into a Teflon trough (4mm in diameter, 1–2 mm depth). On top of this drop a monolayer of fluorinated lipids is spread. The lipid is prepared in a mixture of chloroform/hexane (1:1, v/v). The protein adsorbs via its His-tag to the Ni2-NTAheadgroup of the F-lipid. From a dilute solution (150g/mL) a locally high protein concentration is created at the interface
Two-dimensional Crystallization of Membrane Proteins 43
which can be observed as large non-crystalline sheets if the surface is picked up by a hydrophobic electron microscopy grid. At this stage, the protein has attached to the lipid layer, probably as tightly clustered and orientated single molecules or oligomers while still being surrounded by detergent. After an incubation time of 24 hours, BioBeads® are introduced through a channel into the Teflon trough to avoid disturbance of the surface in order to remove the detergent. That leads to a reconstitution of the membrane protein into a lipid bilayer underneath the surface monolayer. After another 24 hours, 2D crystals could be observed by touching the surface with a hydrophobic electron microscopy grid and examined in the electron microscope.
Crystals only formed after reconstitution of the protein into a lipid bilayer. Control experiments showed that 250mM imidazole or 150M NiSO4 inhibited crystallization or even the assembly of sheets of disordered protein under otherwise identical conditions. Both chemicals work as inhibitors of protein binding to the Ni2⫹-headgroup. Imidazole is routinely used in Ni2⫹-chelating chromatography to release bound protein from the columns. Low concentrations of NiSO4 were sufficient to saturate the Ni2⫹-chelating histidine tags on the protein so that it was no longer able to attach to the derivatized lipid. These observations indicate that attachment of the protein and crystallization is dependent on the interaction between the His-tag of the protein and the Ni2⫹-NTA-derivatized lipid. Large membrane sheets, often measuring more than 10m across, were frequently found and consisted of either a single crystal lattice, or of a few lattices fused in plane, as revealed by EM of negatively stained crystals (Fig. 2.5). It is interesting to note that it can be concluded from the p22121 symmetry of the 2D crystals that only one half of the molecules make specific contacts through their His-tag with the Ni2⫹-NTA moiety of the lipid headgroups. The other half packs into the crystal in an upside-down orientation, resulting in a head-to-tail arrangement of pairs of ATPase molecules related by the 2-fold screw axis in the plane of the membrane. This arrangement allows a much denser packing of the highly asymmetric ATPase molecules, which seems to be more favorable for crystallization. The study resulted in a projection map of the H⫹-ATPase in a frozen-hydrated state at 9Å resolution.
44
Strategies for Two-dimensional Crystallization of Proteins
Fig. 2.5 Different stages of crystal formation of the H-ATPase using fluorinated lipids. (a) 3h and 24h (b,c) after detergent removal. (d) Enlargement of (c).
There are several advantages in the use of fluorinated lipids for the 2D crystallization of membrane proteins. The technique requires only minimal amounts of protein and the protein solution can be very low in concentration. This is an advantage especially for membrane proteins which — even if over-expressed — are often of limited availability. The crystallization is quick and should give large single layers, offering the potential of electron diffraction as a straightforward test of crystalline order and state of preservation. This will help to optimize crystallization conditions. The approach may be applied to any membrane protein that can be expressed with a His-tag, and thus offers the prospect of a much needed rationale, conceptually simple and potentially general method for the 2D crystallization of membrane proteins.
Chapter 3
Methods to Monitor Surface Crystallization
Introduction to the Biophysical Studies of Interfaces Organic and biological monolayers at the air/water interface have been extensively studied over a long period of time. Thirty years ago, the main tool used to investigate a new molecule spread at the water surface as a monolayer was the well-known Langmuir balance, which gives pressurearea isotherms (Gaines, 1966). Since that time, several other physical techniques have been developed in order to investigate the various phenomena at the interface in more detail. Here we consider surface pressure measurements, ellipsometry, fluorescence imaging, Brewster angle microscopy, X-ray reflectivity, grazing-incidence X-ray diffraction, and measurements of monolayer rigidity. Although ellipsometry has been described for the first time in 1887 by Drude (Drude, 1887), this technique started to be extensively used in the 1970s. In brief, an ellipsometric measurement consists of measuring the difference of the two components of the polarization of light, before and after reflection from a surface. From this the averaged surface density of the monolayer can be calculated and hence the thickness and refractive index of the monolayer (Azzam and Bashara, 1977). The application of this technique to the lipid monolayer system was a major breakthrough: the observation of the air/water interface was becoming fascinating; it was then possible to monitor in real time the variation of the surface density of a specimen. In 1983 and 1984, fluorescence microscopy was developed and enabled observation of the growth of dense domains (Losche et al., 1983; McConnell et al., 1984; Weis et al., 1982). The major drawback of this 45
46
Strategies for Two-dimensional Crystallization of Proteins
technique comes from the use of exogenous fluorescent molecules which can be a source of perturbation in a biological system. The data can also be biased by phenomena such as quenching of fluorescence or transfer of energy between molecules, which makes the interpretation uncertain. A few years later the Brewster angle microscope (BAM) was developed (Hénon and Meunier, 1991; Hönig and Möbius, 1991). This method is non-invasive which is a big advantage compared to the fluorescence microscopy. The BAM is now widely used in systematic explorations of new systems. L.G. Parratt described in 1954 the study of surfaces by X-ray reflectivity (Parratt, 1954). This technique gives access to the vertical electron density profile of the monolayer and was developed later for the study of biological monolayers (Als-Nielsen, 1991). Grazing incidence X-ray diffraction of 2D solid systems (thin layers of aluminium gallium arsenide, GaAsAl) on a solid support had been reported for the first time in 1979 (Marra et al., 1979). Then in 1987, two teams (Dutta et al., 1987; Kjaer et al., 1987) had shown the possibility of the observation of X-ray diffraction from organic monolayers deposited on a water surface. This technique requires the intense radiation from a synchrotron source. Its application to organized biological molecules was described by Als-Nielsen in 1994 (Als-Nielsen et al., 1994).
3.1. Fluidity of the Monolayers Crystallization depends to a considerable extent on the fluidity properties of the lipid matrix. This is acknowledged throughout the literature that is reporting 2D crystallization of proteins involving interaction with the lipid films. In all cases, protein crystals are only obtained on lipid films in a fluid state.
3.1.1. Measurement of the fluidity of lipids At different temperatures, lipids can exist in different phases, and transitions from one phase to another can be detected by physical techniques as the temperature changes. The phase transition, which is most consistently observed, occurs during temperature increase when the membrane passes from a tightly ordered “gel-” or “solid-” phase to a “liquid-crystal”
Methods to Monitor Surface Crystallization 47
or “fluid” phase. The elevated temperature leads to a higher freedom of movement of individual molecules. This greater mobility can be rotational but can also be a displacement of the molecules in the plane. The most widely used method for determining the phase transition temperature (Tc) is differential scanning calorimetry. The influence of hydrocarbon chain length and saturation as well as the properties of the head group on the value of Tc is substantial. In general, increasing the chain length, or increasing the saturation of the chains increases the transition temperature. Membranes made from egg yolk lecithin have a transition temperature from ⫺15⬚C to ⫺7⬚C, compared with membranes from mammalian sources which are usually in the range 0⬚C to 40⬚C. In the case of 2D crystallization using lipid monolayers, isotherm surface pressure measurements can be performed in order to find out if a monolayer is in a fluid state. The isothermal pressure versus molecular area curve is determined using a Langmuir trough equipped with an electrobalance (Ulman, 1991). The lipid monolayer must collapse in the fluid state. The collapse pressure (c) is the critical pressure for which the lipid monolayer is becoming too compact and unstable. The monolayer is then collapsing. A mono-molecular thick lipid film in a solid phase is characterized by a high lipophilic cohesion between the fatty chains and a low translational diffusion coefficient D (~10⫺7cm2s⫺1). If the temperature increases, the lipid film will exhibit a fluid phase in which D increases by at least an order of magnitude. If the temperature raises further, the value of D does not increase more than twice, while the lipid is in the fluid phase.
3.1.2. Fluidity of the lipid monolayer and crystallization Once a protein is concentrated on a lipid layer, rotation and displacement of the lipid-protein complexes in the plane is necessary to obtain optimal contacts between proteins prior to the formation of crystals. These motions imposed onto the film result from a balance between the driving force that originates from protein-protein interactions and a counter-force due to lipid-lipid interactions. That counter force is an unusually strong attraction arising between the non-polar moieties of the amphiphilic molecules and has been termed the “hydrophobic interaction” (Blokzjil and Engberts, 1993). Interactions resulting from the hydrophobic effect
48
Strategies for Two-dimensional Crystallization of Proteins
can be partly decreased or enhanced through electrostatic interactions between the polar heads of the lipids and between these polar heads and the bulk solvent. The balance between these forces can be regarded in terms of fluidity of the lipid layer. However, when the protein-lipid interactions are weak (typically Coulombic interactions), the fluidity of the lipid film is likely to be less crucial because such a weak interaction is easily broken and the protein can then recreate another one with a neighboring lipid. The resulting dynamic equilibrium allows the protein to move laterally in the plane and eventually compensates for the lack of fluidity of the lipid layer. In the case of the strong interaction existing between a Histagged protein and a Ni2⫹-chelating lipid (pH8, KD ⫽ 10⫺13M) or between any receptor-ligand couple, the lipid-protein complex once formed is unlikely to break. The only way for the anchored protein to move at the interface is by an in-plane diffusion process of the lipid-protein complex, which requires the use of a fluid-lipid monolayer. Consequently, lipids spread in the solid phase at the air/water interface are producing at best a close packing of proteins, but in no case 2D crystals.
3.1.3. How to modify the fluidity of a lipid monolayer? Fluidity variations can be induced in various ways: Modification of the surface tension imposed by the lipid at the interface This can be realized by compressing the monolayer using a Langmuir trough or film balance techniques. However, that methodology requires the use of a large trough (several mL) containing large amount of proteins. This is often not possible due to the scarcity of some expressed proteins. Usually the crystallization process takes place in small Teflon wells (4mm diameter, 2mm depth) which does not allow direct monitoring of the lipid film physical state by surface tension measurements. In these small troughs, the lipids, when deposited at the interface, do not spread into a homogeneous monolayer but form a lipid bulk reservoir at the edges of the trough. This is called the “border effect” and is easily visualized when handling chromophoric amphiphiles. The density of the lipid film at the central region of the interface is likely to be far lower than at areas close to the edges. So, if the theoretical amount of lipid needed to cover the interface with a fluid condensed (or liquid) monolayer is actually
Methods to Monitor Surface Crystallization 49
spread on a very small trough, there is in fact a lack of lipid when moving off the edge and a formation of a fluid expanded (or gaseous) lipid phase towards the center. Binding of a water-soluble macromolecule onto such a film results in progressive solubilization of the lipid into the aqueous phase and subsequent concentration and crystallization of the protein can not be achieved (Lebeau et al., 1996). This is the reason why all the 2D crystallization experiments are systematically performed with an excess of lipid (5 to 10 times the theoretical amount needed to form one layer on the small crystallization trough). This surface pressure is called the pseudo-equilibrium spreading pressure (Lebeau et al., 1990) or maximal spreading pressure (Mosser and Brisson, 1991). That particular value of the surface pressure can be measured on larger troughs and appears to be slightly below the collapse pressure (c). Consequently, it is absolutely necessary to perform the crystallization experiments at the temperature above the main phase transition temperature in order to have a lipid film in a fluid phase and if the crystallization takes place in small wells the amount of lipid must be in excess. Effect of temperature An augmentation of temperature increases the diffusion of the lipids in the monolayer. Most of the proteins rapidly loose their functions when handled at 20⬚C and a fortiori faster at higher temperature. Therefore, usually during the crystallization process the temperature is kept constant at 20⬚C. The parameters which can be varied more easily for varying the fluidity property of the monolayer is the structure of lipids. Variation of the molecular structure of lipids Due to the development of supramolecular chemistry and biophysical approaches, it is becoming easier to design new amphiphilic molecules and to test their mechanical properties experimentally when spread at the air/water interface as well as their ability to bind macromolecules. A series of 18 amphiphilic compounds exhibiting different fluidity properties at the same temperature have been designed and synthesized (Nuss et al., 1999). These original amphiphilic molecules possessed either unsaturated chains (a cis double bond in the middle of an alkyl chain introducing a kink in the linear structure), branched chains (preventing
50
Strategies for Two-dimensional Crystallization of Proteins
O
R
H O N
O
O
H N
(A) O
H 2N
O
O
O
O
OH
O
O OH O
(B)
H 3C
CH3
O
(C)
(D)
O
O
O O
O
O
O OH
O OH P
OH
O O
CH2
(E)
(F) F
F F F
F F F F F
F F F F F F
F F F
O
O OH
OH
O
O
OH
OH
Fig. 3.1 Design of novobiocin-lipids used for testing the kinetics of 2D crystallization of the B subunit of DNA gyrase. A: novobiocin; where R is the reference lipid matrix (B); R ⫽ branched lipids (C and D); R ⫽ polyhydroxylated headgroup (E); R ⫽ partially fluorinated alkylchains (F).
Methods to Monitor Surface Crystallization 51
a tight packing of the chains), hydroxylated chains (polyhydroxylated headgroups laterally expand the amphiphilic structures at the air/water interface), or fluorinated chains (Fig. 3.1). Asymmetrical structures were synthesized in order to avoid a regular packing of the chains inside the monolayer. Introduction of partially fluorinated alkyl chains aimed to benefit from the non-ideality of hydrocarbon-fluorocarbon mixtures (Mukerjee and Yang, 1976) in order to lower the hydrophobic interaction. Some of these amphiphiles have been derivatized with novobiocin for crystallization experiments with the B subunit of DNA gyrase (Lebeau et al., 1999). The crystallization kinetics depend to a considerable extent on the fluidity properties of the lipid matrix. Whereas optimum incubation time with the reference lipid B (Fig. 3.1B) is 24–28 hours, it is only 5–6 hours with the compounds C and D (Figs. 3.1C and 3.1D) and 10–12 hours for compounds E and F (Figs. 3.1E and 3.1F). Clearly, modifying the fluidity of the lipid film, the kinetics of 2D crystallization of DNA gyrase B subunit can be increased 4-fold. This is especially interesting as a number of proteins rapidly lose their functionality when incubated at 20⬚C. When analyzing the diffraction quality of the crystals obtained at 20oC, these five lipid matrices gave the same resolution close to 1.5–1.7nm. This is probably the maximal resolution one can expect with negatively-stained specimens. In another crystallization system using a His-tag protein (His-HupR) bound to Ni-lipid, the influence of the structure of the lipids has been tested on the kinetics of crystallization (Courty et al., 2002). The classical Ni-lipid, Ni-NTA-DOGA is shown in Fig. 3.2A. Subsequently Ni-NTA-BB (Fig. 3.2B) and Ni-NTA-BF (Fig. 3.2C) were designed so they spread into stable but more fluid monolayers at the air/water interface than the classical Ni-NTA-DOGA. Ni-NTA-BB is made of the same amphiphilic molecule as compound C (Fig. 3.1C) but derivatized with NiNTA for crystallization experiments with His-HupR. Note that Ni-lipids are always used with diluting lipids such as DOPC (ratio 1:1 to 1:15). Ni-lipid used alone has never been successful for producing 2D crystals of His-HupR. When Ni-NTA-BB is used to form the monolayer, it accelerates the process of protein adsorption and the protein crystallization is three times faster than when Ni-NTA-DOGA is used. Making the crystallization faster is of special importance for fragile and easily degradable proteins at room temperature. The quality of the crystals produced
52
Strategies for Two-dimensional Crystallization of Proteins
F F F F F F F F O
O
O
O
O
O
O
O
O O
F F F F F F F F F
O NH
NH
O 4
O 4
O HN
O OO N Ni O H O O O HO H H
O OO N Ni O H O O O HO H H
O OO N Ni O H O O O HO H H
(A)
(B)
(C)
NI-NTA-DOGA
NI-NTA-BB
NI-NTA-BF
Fig. 3.2 Design of three Ni-lipids used for testing the kinetics of 2D crystallization of His-HupR. (A) Ni-NTA-DOGA, classical Ni-lipid; (B) Ni-NTA-BB, with two branched alkyl; (C) Ni-NTA-BF, with one branched and one fluorinated chain.
with these fluidity-enhanced Ni-lipids was the same as these obtained with the classic Ni-lipid Ni-NTA-DOGA. It is interesting to note that the value of the shear modulus, which gives access to the rigidity of the crystallized monolayer, is 58mN/m whatever lipid is used. Although the quality of the crystals obtained with Ni-NTA-BF are equally good, the process of crystallization is slower compared with Ni-NTA-DOGA. A possible segregation of the fluorinated chains interacting with each
Methods to Monitor Surface Crystallization 53
other and preventing a proper mixing with DOPC could explain this phenomenon. The meticulous and systematic study of the 2D crystallization process using various lipids exhibiting a variety of fluidity characteristics is especially important to improve our understanding of the different steps involved in the self-organization of a protein adsorbed on a lipid monolayer.
3.2. Ellipsometry Ellipsometry is a sensitive optical method that employs polarized light which is reflected at an interface to determine the thickness and refractive index of thin films (Azzam and Bashara, 1977; de Feijter et al., 1978). The resolution which can be achieved is suitable for the measurement of films, which are only one molecule-layer thick. It can be used at solid/liquid interfaces as well as air/water interfaces. It has been useful to understand basic mechanisms of adsorption and the surface dynamics of proteins. Since it is a non-destructive method, it is possible to follow experiments by in situ analysis. There are different practical approaches to ellipsometric measurements, but they all follow the same principle experimental setup. Ellipsometers are generally arranged in the following way: Light source (white light, laser light, synchrotron radiation) → polarizer (P) → compensator (C) (retarder, quarter wave plate) → sample (S) → analyzer (which is another polarizer) (A) → detector (D).
The simplest case of ellipsometry is null-ellipsometry, which can be used to determine the ellipsometric angles to high accuracy. It uses the fact that a distinct state of polarization of the incident beam will lead to linearly polarized light upon reflection. Figure 3.3 shows the setup of a null-ellipsometer: Laser light passes through a polarizer (P) and will be linearly polarized. By passing through a quarter wave plate (C), the light will be elliptically polarized. The angular positions of P and C have to be changed until linear polarization of the reflected beam is achieved. In this case, the analyzer (A) can be set to a position of 90⬚ in respect to the linear polarization axis, which extinguishes the reflected beam and thus the photo detector records a minimum of signal. These angular positions
54
Strategies for Two-dimensional Crystallization of Proteins
Laser
Photodiode Polarizer (P)
Analyzer (A)
Quarter Wave Plate (C)
Sample
Fig. 3.3 Principle of a null-ellipsometer. See text for details.
depend on the properties of the interface, and are therefore a measure that is sensitive to changes at this interface. Adjusting the polarization optics to a position where the reflected beam is cancelled, allows the calculation of the ellipsometric angles from their recorded angular positions. The ellipsometric parameters (psi) and ⌬ (delta) describe the change in polarization after reflection for the ratio of amplitudes and the phase-shift, respectively. Ellipsometry uses the fact that light can be described as an electromagnetic wave. Its electrical field E has a much stronger interaction with matter than its magnetic field. Monochromatic light can be described by two harmonic oscillations of the electric field vector in p (parallel to the plane of incidence) and s (perpendicular to the plane of incidence). These oscillations have the same frequency, but can differ in amplitude and phase and express the state of polarization of light. The most general state of polarization is elliptical. The electrical field vector travels along an ellipse, if observed at a fixed point in space. If the phases of p and s are identical, the ellipse changes into a straight line. If the phase difference is 90⬚, the ellipse becomes a circle. These two special cases lead to linear and circular polarization, respectively. In 1883, the German physicist Paul Drude observed a time-dependent change in phase shift between s- and p-component of polarized light after reflection from freshly cleaved Sb2S3 (antimony-sulfide (stibnite)) which was exposed to air. Upon cleavage, the surface became oxidized by the surrounding air and an oxide layer (Sb2O4) built up. This was the first ellipsometric experiment. Although Drude worked out the theoretical basis of ellipsometry a few years later (Drude, 1887; Drude, 1888), the
Methods to Monitor Surface Crystallization 55
first technical applications appeared only at the end of the 1960s when the computer industry required the exact measurement of thin layers on silicon wafers. is an angle whose tangent represents the ratio of amplitude changes for the p and s components. Thus, |Rp| tan ⫽ ᎏᎏ; |Rs| where |Rp| and |Rs| are the ratios of the outgoing wave amplitude for the parallel and perpendicular components, respectively. The value of can be between 0⬚ and 90⬚. ⌬ stands for the relative phase shift of the p and s component upon reflection and can be between 0⬚ and 360⬚. ⌬ ⫽ ␦1⫺␦2; ␦1 and ␦2 are the phase differences between the parallel component and the perpendicular component of the incoming (␦1) and outgoing (␦2) wave, respectively. This relationship is given by the basic equation of ellipsometry: rp tan ᐉi⌬ ⫽ᎏrᎏ. s
The two ellipsometric angles and ⌬ can be used to determine physical quantities of the sample under examination, for example, thickness of a film on a substrate or the refractive index of a film. However, as ellipsometry is an indirect method determining the angles of and ⌬, an optical model is necessary for quantitative evaluation of parameters like film thickness or film refractive index. For measurements at a solid/ liquid interface the calculations are more straightforward than at an air/water interface. It has to be pointed out that, especially at the air/water interface, changes in ⌬ are linearly correlated with film thickness whereas changes in are often very small (Reiter et al., 1992; Reiter et al., 1993). For that reason, it is not always possible to determine the layer’s refractive index and its thickness simultaneously in these cases. Therefore, the film refractive index is often assumed as a value between 1.4–1.6. Alternatively, it can be approximated by X-ray reflectometry which gives a value in that very range (Helm et al., 1987). This leaves the film thickness to be the only free parameter and makes modeling of this parameter easier. The simplification of the optical model is sometimes questioned, but using radio labeled films for comparative calibration measurements has shown good correlations with ellipsometric measurements (Jonsson et al., 1985). For that reason, ellipsometric
56
Strategies for Two-dimensional Crystallization of Proteins
data for pure compounds is generally regarded as generating highly accurate data. In contrast, the measurements of mixtures of compounds by ellipsometry are not quantitative. In the case of protein crystallization using lipid monolayers, ellipsometry can be of help to follow different events during protein crystallization. Often the quantitative thickness of a film is not of interest, but rather the stability of lipid films or the dynamics of protein adsorption to a lipid monolayer. This is expressed in a relative change in the ellipsometric angle. The ellipsometric angle correlates with the amount of lipid and detergent adsorbed at the interface and is taken to be zero for a pure air/water interface. Lebeau and colleagues used ellipsometry to show that a lipid monolayer of fluorinated lipids is stable in the presence of high concentrations of various detergents (Lebeau et al., 2001). This is important for the application of surface crystallization to membrane proteins. These proteins have to be in a detergent solution to keep them soluble. However, the detergent will solubilize a lipid monolayer rapidly. Using a detergent-resistant lipid monolayer composed of fluorinated lipid, it was possible to crystallize a plasma membrane H⫹-ATPase with crystals ordered to a resolution of 9Å (Lebeau et al., 2001). Ellipsometric studies were used to systematically analyze the conditions favorable for the crystallization process of the transcriptional regulator HupR (Courty et al., 2002). Three different lipids were used for 2D crystallization. Only a dense lipid monolayer allowed a reproducible crystallization of the protein. The necessary amount of lipid to obtain a lipid monolayer at maximum density and surface pressure could be determined by using this technique. The adsorption of protein to the lipid monolayer could also be monitored by ellipsometry. This gave important insights into the kinetics of crystallization of this protein. It showed that the first step in crystallization is the formation of patches of packed protein, which could be visualized by BAM (Courty et al., 2002). During this phase the ellipsometric angle fluctuated. After the adsorption of protein was complete, the value of the ellipsometric angle stabilized and large 2D crystals could be observed after transfer onto an electron microscope grid. Ellipsometry was used to demonstrate that the adsorption of vascular anticoagulant ␣ (annexin V) to phospholipid bilayers was strictly calcium dependent (Andree et al., 1990). It was elegantly shown that
Methods to Monitor Surface Crystallization 57
repeated adsorption and desorption of annexin V depended on the addition of calcium or its removal by EDTA, respectively. The important role of the phase state of the lipid monolayer for protein binding was analyzed by ellipsometric studies of streptavidin binding to a biotinylated phospholipid monolayer at the air/water interface (Reiter et al., 1993). Only the fluid phases allowed for maximal protein adsorption whereas protein binding was blocked at highly condensed states. Ellipsometric measurements require a larger surface to be analyzed than is the case with experiments for electron microscopy. Compared to wells with a diameter of 3mm and a volume of about 40L, which are used for preparing specimens for electron microscopy, the size of a trough to do ellipsometric measurements is in the order of 5 ⫻ 5cm and has a volume of 20mL. On the other hand, these measurements can generally be performed at a low protein concentration of 3–15g/mL, compared to a concentration of about 50–100g/mL for electron microscopy. The formation of a dense lipid monolayer and the adsorption of protein onto a lipid monolayer are crucial steps preceding the formation of 2D crystals under lipid monolayers. Ellipsometry can be used to follow both of these processes. This might be of special interest when tests for binding of soluble or membrane proteins are performed. Insight into the pace of adsorption and monitoring the accumulating protein density over time can speed up the determination of favorable crystallization conditions considerably. This is particularly useful when membrane proteins are used. The adsorption of membrane protein to the lipid monolayer can be monitored by ellipsometry. When the layer is well stabilized, BioBeads® can be added for 2D reconstitution of the membrane protein into lipid bilayer (see also Sec. 2.4.). Because ellipsometry is a noninvasive technique, the adsorption (which is the first step of the crystallization process) can be monitored continuously. This is more conclusive and more convenient than taking a time course by interrupting the experiment at different points and preparing an electron microscope grid.
3.3. Rigidity Measurements Mechanical measurements were performed in order to have some indication on the lateral crystalline order of the adsorbed protein.
58
Strategies for Two-dimensional Crystallization of Proteins
When an electron microscope grid is deposited on the top of a crystallization drop in order to transfer the crystal for electron microscope observation, the grid will either move freely on the surface of the drop or it will stay still. This qualitative test of rigidity of the layer can be a hint for the presence of crystals at the interface. If the layer presents a liquid phase state, it will not resist the drive of the grid toward the edge of the trough. In opposite, if the layer is solid, the displacement is frozen. The measure of the modulus of the shear elastic constant allows quantification of this observation. The principle of the measurements is simple, and consists of applying a shear strain to a monolayer and to measure the associated deformation. The analysis of the curves of deformation versus strain gives access to the shear elastic constant. The layer is sheared using the rotation of a float containing a small magnet by application of a magnetic field. The shear elastic constant is defined by the formula: ␦xy ⫽ 2uxy, where ␦ is the strain tensor and u is the deformation tensor (Landau and Lifchitz, 1969). The experiment allowing a direct measurement of the lateral rigidity of the monolayer is described in detail elsewhere (Zakri, 1997). A schematic view of the home-made device is shown in Fig. 3.4. Briefly, a Teflon trough — about 50mm in diameter — containing 8mL of the solution covered by the monolayer under study is placed in the middle of the setup (shaded area). A small piece of paraffin-coated aluminium disk (about 10mm diameter) is floating at the center of the trough and is surrounded by the monolayer whose rigidity is measured. The float contains a small magnet (black rectangle inside the boat), and a small vertical mirror; the total weight is about 32mg. A magnetic field applies a torque on the float, inducing its rotation around the vertical axis. We then measure directly the resistance that the monolayer opposes to the rotation of the float. The amplitude of rotation of the float, measured with a laser beam reflected on the small mirror is directly related to the rigidity of the monolayer: the motion of the float is restricted by a monolayer behaving like a solid. The small solenoid just above the center of the trough applies an inhomogeneous magnetic field parallel to the horizontal.
Methods to Monitor Surface Crystallization 59
Fig. 3.4 Schematic representation of the set up built for the measurement of the lateral rigidity of the monolayers under study. The large coils at the exterior of the experiment serve to apply a torque to the float. In response to this torque, the float rotates around the vertical axis. The amplitude of rotation is dependent on the restoring forces applied by the monolayer under shear deformation. Figure taken from (Lenne, 1998).
The experimental procedure was usually as follows: The lipid layer containing the ligand lipids was spread first. This layer, in a liquid phase did not offer any detectable resistance to shear flow and the response was indeed indistinguishable from the pure subphase response. We then recorded the amplitude and phase of the mechanical response at a fixed frequency, usually 5Hz, as a function of time. After complete stabilization, the protein was injected into the subphase. The homogeneity of the solution was completed using a peristaltic pump for a few minutes. Once the protein was injected into the subphase, we monitored the kinetics of rigidification up to stabilization (Vénien-Bryan et al., 1998).
60
Strategies for Two-dimensional Crystallization of Proteins
Figure 3.5 shows the changes of the ellipsometric angle as well as rigidity of streptavidin on a lipid layer of biotin(LC)-DPPE/DOPC 1:4 during a short period of time (Lenne, 1998). The appearance of rigidity is only observed after a delay of about 25 minutes (Fig. 3.5). This delay corresponds to the period of adsorption of the protein to the layer. The rigidity is expected to appear after crystallites have grown sufficiently to touch each other. Non-zero shear rigidity then corresponds to the percolation of 2D crystals over the whole area between the float and the edge of the trough. After an hour, the elastic constant still increases, but at a slower rate (Fig. 3.5B), and continues to increase for more than 5 hours. This corresponds to another regime, where annealing of some defects occurs. This could be coalescence of neighboring crystallites or thinning of the grain boundary zone between individual crystals (Fig. 3.6). There is a direct relation between the value for the shear elastic constant and the resolution obtained by X-ray grazing incidence diffraction (Lenne et al., 2000). The higher the value of the shear elastic constant is, the better the crystals will diffract in situ. Ellipsometry and rigidity measurements have been also performed to monitor the crystallization process of cholera toxin B (CTB) subunit and annexin V (Vénien-Bryan et al., 1998). If a lipid monolayer of DOPC (dioleoyl-phosphatidylcholine) and GM1 (monosialoganglioside) at a ratio of 4:1 (mole/mole) was applied on top of a water surface, it resulted in a relative change in the ellipsometric angle compared to a pure water surface. After injecting protein into the subphase, the ellipsometric angle ⌬ was found to be proportional to the amount of protein adsorbed to the lipid monolayer. Maximal adsorption of protein was observed after 1 hour and led to an increase of the ellipsometric angle. Only after this time was an increase of the shear rigidity of the monolayer detected. This indicated that the crystalline units had grown sufficiently to touch each other. The experiments with annexin V gave similar results. Interestingly, during these experiments 2D crystals of CTB observed by electron microscopy were often multilayered, especially at higher protein concentration. However, ellipsometric measurements using increasing protein concentrations showed that the protein adsorbed to the monolayer was only a single layer. This raised the point
Methods to Monitor Surface Crystallization 61
(A)
(B)
Fig. 3.5 Streptavidin (15g/mL), biotin-(LC)-DPPE/DOPC 1:4 (mole/mole). (A) Ellipsometric angle (empty rectangle) and shear elastic constant (black lozenge). The measurements are done simultaneously in different trough; (B) Elastic constant during the whole experiment. The origin of the times corresponds to the injection of the protein in the subphase.
62
Strategies for Two-dimensional Crystallization of Proteins
Percolation
Float
Annealing of crystals
Coalescence
Crystal growth
Fig. 3.6 The rigidity increases after crystallites have grown sufficiently to touch each other: there is percolation of 2D crystals over the whole area between the float and the edge of the trough. Then annealing of the defects occurs, either through coalescence of neighboring crystallites or growth of the crystallites.
that the transfer of the crystals onto electron microscope grids is a critical stage during the experiment, and might influence the appearance and quality of the crystals. This issue will be further discussed in Sec. 4.1.
3.4. Brewster Angle Microscopy BAM uses the phenomenon that if a beam of p-polarized light is incident at the Brewster angle, the reflectivity coefficient disappears and consequently no light is reflected from the surface under investigation (Fig. 3.7). (Brewster’s law: tan ␣ ⫽ n2/n1 with n1 and n2 being the refractive indices. The angle between the reflected and the refracted beam is 90⬚.) For pure water, the Brewster angle is 53.12⬚, but even the presence
Methods to Monitor Surface Crystallization 63 1
R RS RP 0
0˚
90˚
Brewster Angle
θB θB
air monolayer water
Fig. 3.7 Principle of a Brewster angle microscope. See text for details.
of a one molecule thick monolayer changes the conditions sufficiently to restore some reflectivity and provides contrast (Hönig and Möbius, 1991). This allows imaging the finest changes occurring at the surface layer. Using a BAM, it was possible to visualize regions of different protein density of streptavidin underneath a lipid monolayer and to follow the crystallization process of the protein (Frey et al., 1996; Schief et al., 1999; Courty, 2001). After injection of streptavidin underneath a biotinylated lipid monolayer, the initial protein adsorption to the surface layer could be monitored as an increase in grey-level of the BAM image. After a short time, the typical H-shaped 2D streptavidin crystals appeared as bright areas because of their higher protein density compared to the surrounding protein layer (Fig. 3.8). The crystal growth could be observed as an increase in crystal size and number over time. Moreover, it was possible to determine quantitatively the surface density according to the grey value of the surface. Crystallization only started after exceeding a defined protein density which could not be increased by lateral compression of the monolayer. In other experiments, it was
64
Strategies for Two-dimensional Crystallization of Proteins
Fig. 3.8 Kinetics of growth of the 2D crystals of streptavidin observed by a BAM. The monolayer of lipids is made of biotin-LC-DPPE/DOPC 1:3 (mol/mol). The size of the image is 768 850m. Figure taken from (Courty, 2001).
also possible to discriminate areas of the lipid-protein monolayer that were crystalline or not according to their grey value (Courty et al., 2002). 2D crystals of HupR, a soluble transcriptional regulator from Rhodobacter capsulatus, appeared denser and provided additional contrast compared to the surrounding non-crystalline regions (Fig. 3.9). Crystal growth and an increase in the number of crystals could be observed over time, leading to a merging of crystalline patches. The growth of dense domains can be observed as well in situ by fluorescence microscopy (Brockman, 1999; Darst et al., 1991a; Maloney et al., 1999). Unlike BAM, fluorescence microscopy requires the use of
Methods to Monitor Surface Crystallization 65
Fig. 3.9 Growth of the 2D crystals of HupR observed by a BAM. The lipid monolayer was made of Ni-NTA-DOGA/DOPC 1:3 (mol/mol). The size of the image is 768 850m.
fluorescent probes which can interfere with the biological system (Hénon and Meunier, 1991). Observation of crystallization by BAM does not perturb the biological phenomenon which makes this technique very valuable. In addition, it may be possible to discriminate crystalline from non-crystalline areas. This can help to optimize crystallization conditions by assessing the size of crystals at the air/water interface in situ. A pre-selection will then speed up the screening process, because successful crystallization events can be determined prior to transfer onto an electron microscope grid and screening by electron microscopy. These insights into the dynamics of the crystallization process and the
66
Strategies for Two-dimensional Crystallization of Proteins
mechanical properties of the lipid-protein film will assist in the complicated task of determining conditions for 2D crystallization of proteins using lipid monolayers. This is particularly true for membrane proteins because the elimination of detergent by adding BioBeads® must be done when the protein has completely adsorbed to the lipid monolayer.
3.5. X-Ray Reflectivity and X-Ray Grazing Incidence Diffraction The physical techniques, X-ray reflectivity and X-ray grazing incidence diffraction, were initially developed for the study of the behavior of solid surfaces and interfaces and helped to understand the corrosion of a metal or the resistance of glass to scratches or the property of a mirror. They have been extended to the study of soft interfaces such as lipidprotein monolayers at the air/water interface or on a solid support. X-rays reflections at very low incidence can be used in two ways: “reflectivity” and “surface diffraction”. Reflectivity is the measure of the beam intensity reflected by the surface. Surface diffraction appears when organized structures such as crystalline planes at the surface tend to reflect the light in specific directions which are not identical with the usual specular reflections. The observation of this effect requires X-rays of stronger intensity than for reflectivity.
3.5.1. X-Ray reflectivity The X-ray reflectivity is similar to the optical phenomenon of total reflection. Total reflection occurs when light or other electromagnetic radiation passes from a medium of a higher refractive index to one of a lower index, provided that the incoming light direction makes a sufficiently small angle with the plane of the interface between the two media (Fig. 3.10). The critical angle below which total reflection occurs is, from Snell Law, ⫽ cos⫺1(nlow/nhigh), where is the angle of incidence between the water surface and the incident beam and nlow and nhigh are the indices of refraction. For X-rays, the index of refraction of solids and liquids is slightly less than 1, so an X-ray beam passing from air or vacuum to a solid or liquid will be totally reflected if its angle of
Methods to Monitor Surface Crystallization 67
incident beam
α
reflected beam
α
Fig. 3.10 Principle of an X-ray reflectivity experiment.
incidence is at or below the critical angle; all solids or liquids behave as a mirror for X-rays. Total reflection is easy to observe with visible light between water (index higher than 1) and air. If a diver looks at the water surface above him at a small angle, he will see the surface as a mirror and can not see through the surface into the air. This same phenomenon, but with the roles of air and material medium reversed, is used with X-rays for studying surfaces at low incidence. Since the indices of refraction are both very close to 1, the critical incidence angles for which total reflection occurs are very small. For example, with air and water the critical angle for total reflection is 2.7mrad (about 0.15⬚) with X-rays of wavelength 1.54Å. For a heavy material such as gold, this critical angle rises to 10mrad (0.6⬚). In practice, the angular domain explored is between 1 and 70mrad (i.e. 0.05⬚ and 4⬚). The principle of this technique is simple: an X-ray beam is directed to the surface, a thin film or interface, at a small incident angle. A detector measures the reflected intensity as a function of the incident angle. The reflectivity is then defined as the ratio of the reflected intensity to the incident intensity. The result is mainly the variation of the electron density profile of material which indicates superficial defects, modification of chemical composition or the composition of a monolayer of lipid-protein at the air/water interface. Proteins crystallized in 2D on a lipid monolayer have been investigated with this technique. Crystalline bacterial S-layer is a protein crystal sheet that forms the outermost cell envelope component for a large number of prokaryotic
68
Strategies for Two-dimensional Crystallization of Proteins
organisms; this system has received a lot of attention particularly in the field of biomimetic surface modification. The capability of binding and recrystallization of S-protein isolated from B. sphaericus at lipid surfaces has been monitored using X-ray reflectivity and neutron reflectivity (Weygand et al., 1999; Weygand et al., 2002). This study revealed alterations of the molecular-level organization of the lipid headgroups upon protein binding and recrystallization. The authors proposed a model in which protein interpenetrated the phospholipids headgroups at least up to the level where the phosphates are located. The protein then induced a slight reorganization of the headgroup. X-ray reflectivity gives access to the precise distribution of lipids and protein along a vertical axis. Vertical electron density profiles of lipid-protein layer made of a 2D crystals of His-HupR, a transcriptional regulator from the photosynthetic bacterium Rhodobacter capsulatus bound to a Ni2⫹-chelated lipid monolayer have been investigated (Courty et al., 2002). A model has been proposed in which the thickness of the protein HupR crystallized on the lipid monolayer is 85Å. The thickness corresponding to the head and the hydrophilic tail of the lipid are 6Å and 15Å, respectively.
3.5.2. X-Ray grazing incidence diffraction on monolayers at the surface of water X-ray diffraction is the prime means for determining the 3D structure of crystallized molecules: small molecules or biological components such as protein, DNA or even bigger complex macromolecules such as viruses. It is therefore very interesting to study the same compounds in 2D at the water surface, in situ, without transferring the 2D crystals onto a grid for electron microscope observation. Grazing incidence Bragg diffraction can be used to study 2D crystals at a water surface. In the 1980s X-ray grazing incidence was primarily developed to investigate thin films of amorphous materials. X-ray grazing incidence was then used to investigate directly the structures and textures of molecular films at the surface of water. The study of soft condensed matter at liquid interfaces gives information on the dynamics of a film at a liquid interface. Unlike small molecule specimens, the unit cell dimensions of 2D protein crystals are large and therefore a high angular resolution is
Methods to Monitor Surface Crystallization 69
needed to separate the numerous Bragg reflections. This geometry is obtained from third-generation synchrotron beams. For structural analysis, the incidence angle is slightly lower than the critical angle so that the X-ray beam is totally reflected at the air/water interface. It is the evanescent wave which propagates parallel to the surface and penetrates to a depth of several tens of Å which is diffracted. The depth of penetration of the evanescent wave typically varies from 40 to 1000Å depending on the incidence angle; the higher this angle, the deeper the wave penetrates. If there is a crystalline plane close to the surface, Bragg diffraction of the evanescent wave will occur. A detector records the intensity of diffraction (Fig. 3.11). The X-ray beam has to be well
Top View Diffracted beams
Collimator
Detector
2θ X-ray beam
Perspective view (11) Specular reflection α
X-ray beam
(20)
2θ[20]
2θ[11]
Crystallites with various horizontal orientation Fig. 3.11 Grazing incidence diffraction set-up. Top view and perspective view: the monolayer is composed of 2D crystalline patches with various azimuthal orientations. Brag rods are elongated along the vertical direction, due to the 2D periodicity. Only a few are in the correct orientation for diffracting a given reflection.
70
Strategies for Two-dimensional Crystallization of Proteins
collimated and intense and therefore requires the use of a synchrotron. It is possible to get vertical as well as in-plane structural information. Whereas for a perfect 3D crystal, the diffraction pattern consists of a 3D array of sharp spots; for a perfect 2D crystal, the diffraction pattern is an array of lines perpendicular to the surface. Grazing-incidence Bragg diffraction looks at structure in the plane of the surface and hence sees a 2D array of sharp spots formed by the intersection of a plane parallel to the surface through the array of lines. If there is more than one layer, the lines are no longer uniform, and the diffraction intensity becomes more concentrated around the 3D Bragg reflections as the number of layers grows. Hence by measuring the variation in intensity along the lines one can obtain depth information. For more detailed discussion, see Als-Nielsen et al., 1994; Baruchel et al., 1993; Berge et al., 1998. In situ X-ray diffraction of a 2D protein array at the lipid/water interface has been investigated for the first time by Haas in 1995 (Haas et al., 1995). This pioneering study with streptavidin 2D crystals bound to biotinylated lipid layers demonstrated that grazing incidence diffraction is a useful new method for studying protein crystallization and interactions between proteins. From the width of the peaks, the authors deduced that crystals consisting of as few as 100 monomers contributed to the diffraction. This in situ method is free from artifacts by staining or sample preparation which is needed for observation using a microscope. Another group was successful in developing this technique to study a single layer of purple membrane (Verclas et al., 1999). Diffraction from hexagonal crystals of bacteriorhodopsin which form naturally in the purple membrane was observed to a resolution of about 9Å. This diffraction was measured from samples with only 1013 protein molecules in the beam. More recently, four protein-ligand systems known to produce 2D crystals have been investigated (Lenne et al., 2000) at the ESRF (European Synchrotron Radiation Facility) Troika beam line, in Grenoble, France. The four proteins were: Streptavidin, which crystallizes under a biotinylated lipid monolayer; Annexin V, under negatively charged phospholipids; HupR, an expressed transcription factor whose His-tag interacts with Ni2⫹ chelating lipid; and CTB, under monogangliosides (GM1). For the first three systems, Bragg’s peaks were reproducibly observed. A direct correlation between the shear elastic constant (described by (Vénien-Bryan et al., 1998)), which gives a macroscopic
Methods to Monitor Surface Crystallization 71
determination of the rigidity of the crystalline layers, and the resolution of the Bragg diffraction has been clearly established. The higher the modulus of the shear elastic constant, the better the resolution obtained. In the case of streptavidin, the angular range of the observed diffraction corresponds to a resolution of 10Å in plane and 14Å normal to the plane (Fig. 3.12). However, the diffraction data were not yet sufficient for structural determination at atomic resolution. Dynamic disorder such as internal molecular motions and thermally excited capillary waves are probably limiting the resolution. The disorder can be reduced by stiffening the crystals with protein cross-linkers. Protein monolayers at solid/water interfaces might be more suitable for studying than at air/water interfaces as this dynamic disorder might be less. No electron diffraction was observed when CTB was investigated. This result was extremely surprising as diffraction from CTB crystals was systematically observed by electron microscopy after transfer of the layer to an electron microscope grid. One possible reason could be that the CTB crystals were not present in situ at the air/water interface but were appearing during the transfer process. However, this hypothesis is rather unlikely because the CTB crystals observed with electron microscopy are very large and the transfer is comparatively fast. Another possibility is that the 2D crystals do indeed exist in situ but are not directly in contact with the lipid monolayer: a first layer in contact
Fig. 3.12 Grazing incidence diffraction map of intensity reconstructed from scans performed during a high-resolution experiment at the ESRF Troika beam-line. Bragg rods are visible as vertical lines. Intensity of the 2 last peaks has been multiplied by 5 to be visible. Figure taken from (Lenne, 1998).
72
Strategies for Two-dimensional Crystallization of Proteins
with the lipid could be disorganized and a second layer could form 2D arrays (see also the discussion in Section 4.1.). In conclusion, grazing incidence X-ray diffraction at the air/ water interface is a good complement to electron microscopy. Medium resolution structural information can be obtained for molecules that form 2D crystals but can not form 3D crystals, and conformational movements upon changes of pH or salt composition or concentration can also be studied in situ. We must keep in mind that these experiments must take place with a synchrotron source and therefore can not be performed routinely.
3.6. Atomic Force Microscopy AFM provides a method to image biological macromolecules in their native environment. It measures the surface topology by raster scanning the sample below a sharp tip that is attached to a flexible cantilever. The topographs are recorded without sample deformation and yield a lateral resolution of better than 1nm and a vertical resolution of about 1Å. An outstanding signal-to-noise ratio allows the recording of single molecules as well as sampling conformational changes and energy landscapes of protein surfaces. The measurement can be done in physiological buffers. This is an advantage over electron microscopy which relies on sample preparation and protection against the vacuum. AFM measurements can be done in different modes. In deflection mode, the bending of the cantilever is detected by a positional change of the reflection of a laser beam (Fig. 3.13A). In constant force mode the force is kept constant by adjusting the vertical position of the sample (Fig. 3.13B). In practice that means that the position of the reflected laser beam is kept constant. It is possible to attach a single protein molecule to the probe and establish unfolding pathways by retracting the tip. The AFM also allows the monitoring of the movement of single polypeptide loops. Conformational changes can be visualized because the same area can be scanned repeatedly under different buffer conditions. Besides measuring the topography of a sample, the tip can be functionalized to obtain information of conductivity, electrostatic interaction or hydrophobicity at high spatial resolution.
Methods to Monitor Surface Crystallization 73
Laser
Photodiode
Laser
Photodiode
(A)
(B)
(C)
Fig. 3.13 Atomic force microscope. (A) Deflection mode; (B) constant force mode; (C) tapping mode.
New developments allow imaging by rapid oscillations of the cantilever and measuring the dampening when the tip touches the sample (Fig. 3.13C). This mode is called tapping mode. It minimizes friction and prevents capillary forces and allows studying even molecules that are only weakly adsorbed to the substrate. A comparative study of ordered arrays of retrovirus gag proteins assembled on lipid bilayers using EM imaging and AFM gave similar, yet slightly different results (Zuber and Barklis, 2000). The fact that EM images represent 2D projections of electron densities whereas AFM images show surface topologies accounts for this discrepancy. Using
74
Strategies for Two-dimensional Crystallization of Proteins
both types of information led to a new model explaining the arrangement of the monomers in the final structure. As AFM requires a support, supported planar bilayers can be used as a lipid matrix with affinity to the respective protein instead of a lipid monolayer floating on top of a drop (Reviakine et al., 1998). These supported planar bilayers can easily be generated by spreading small unilamellar vesicles onto a cleaved mica surface. Streptavidin and annexin V were crystallized on planar supported bilayers as in the case for a lipid monolayer. The application of AFM to proteins crystallized on lipid monolayers was shown for annexin V (Reviakine et al., 1998). It was established that annexin V 2D crystals covering single lipid bilayers existed as the p6 symmetry crystals observed by EM whereas the p3 crystal form was not detected and might only emerge after transfer to an electron microscopy grid (see Sec. 4.1). AFM is very successful at imaging membrane proteins reconstituted into a lipid bilayer. For example, it was possible to reveal the native molecular organization of the photosynthetic core complex of the photosynthetic membrane from the bacterium Rhodopseudomonas viridis (Fotiadis et al., 2003). In another study, it was possible to visualize Ca2⫹-induced closure of gap junction hemi channels (Müller et al., 2002). It was possible to image both connexion surfaces and their conformational changes. Evidence for two different gating mechanisms at the extracellular and the cytoplasmatic surface were given suggesting two separate gates at the ends of the pore. Furthermore the conformational states of a different membrane protein surface have been sampled. Domain movements were estimated from a large number of single molecule topographs and an energy landscape was calculated (Scheuring et al., 2002). AFM is a valuable addition to the investigation of proteins and protein crystals by electron microscopy because it allows to image the molecules under native conditions without interfering sample preparation. The resolution is sufficient to allow detailed visualization of conformational changes in functional enzymes. For a review of this fast changing technical developments see (Frederix et al., 2003).
Chapter 4
Electron Microscopic Observations and Image Analysis
Electron microscopy can be used to probe the interaction between the macromolecules and the lipid monolayer, the degree of organization of the complexes, and their structures at the molecular and sub-molecular level. This instrument is well adapted as it covers a large range of resolutions from a few tens of microns to the sub-nanometric level. Under well-defined imaging conditions, specimen thickness, and irradiation dose, the electron microscope produces a useful projection of the atoms constituting the specimen. The intrinsic resolution limit of about 0.1nm for modern electron microscopes has so far not been attained even for best preserved biological specimens, due to radiation damage, object drift, specimen flatness, and local charging. Nevertheless, structure determinations around 20Å are ubiquitous, those in the range 6–9Å are rapidly multiplying and those near 3Å resolution are becoming more common.
4.1. Transfer of the Monolayer onto an Electron Microscope Grid A great technical difficulty in the use of 2D crystals grown on lipid layers for structural determination by electron crystallography is the transfer of the crystals from the air/water interface of the crystallization drop to the surface of an electron microscope grid. This is because the crystals are exposed to strong mechanical forces during the transfer which can lead to their distortion or fragmentation. The most common technique to get surface crystals onto an electron microscope grid is to put 75
76
Strategies for Two-dimensional Crystallization of Proteins
a grid covered with carbon film facing the lipid tails on top of the drop and withdrawing it gently. This method is known as the LangmuirSchaefer transfer method (Langmuir and Schaefer, 1938). The carbon film interacts with the hydrophobic alkyl chains of the lipid and the crystals are transferred and can be observed by electron microscopy. Two main types of carbon films deposited on electron microscope grids can be used for transferring protein-lipid films: • •
Continuous carbon film and Perforated carbon films, presenting holes and commonly named holey carbon films.
For more details for the preparations of these films, see Brisson et al. (1999b). It is worthwhile trying different batches of carbon film which are either hydrophilic or hydrophobic. Carbon film can be made hydrophilic by glow discharging. Hydrophobic carbon film is especially efficient at interacting with the lipid layer. Rendering the carbon film hydrophobic by chemical alkylation can be useful (Kubalek et al., 1991) or simply baking the grids for 150⬚C for one hour before the transfer proved to be very efficient in our laboratory. The success rate of using plain carbon film is different for different crystals. Sometimes the transfer is very efficient and the electron microscope grid is covered with numerous and large crystals, but often small, broken crystals are seen on the grid. 2D crystals of fluoresceinated streptavidin on a biotinylated-lipid monolayer have been observed in parallel in situ using fluorescence microscopy and also with an electron microscope after transfer on continuous carbon film grids (Darst et al., 1991a). These observations provide an ideal approach to evaluating the effect of crystal transfer onto an electron microscope grid. The ordered domains observed by fluorescent microscopy were about 10 times larger in diameter than the crystals transferred onto carbon-coated grids. This discrepancy has also been confirmed with 2D crystals of streptavidin when X-ray grazing incident measurements and electron microscope observation of crystals after transfer have been done (Lenne et al., 2000). Kubalek and colleagues (Kubalek et al., 1991) were the first to emphasize the importance of the type of film used for transfer, showing in the case of streptavidin, that the efficiency of transfer and the crystalline preservation were improved
Electron Microscopic Observations and Image Analysis 77
when transfer was done on holey grids. It is true that holey carbon grids ensure a better preservation of material pre-existing at the water surface. This can be illustrated by two other examples: 2D crystals of annexin V and of cholera toxin. 2D crystals of annexin V have been observed in situ at the air/water interface using grazing incidence synchrotron X-ray diffraction (Lenne et al., 2000) and on solid-supported lipid bilayers using AFM (Reviakine et al., 1998). In both cases, only the p6 crystal form of annexin V has been found. When transfer onto a grid is performed, the p6 form is indeed found over the holes of the holey grids, but transfer onto continuous carbon film leads to the most striking structural reorganization. The native p6 crystalline assembly is transformed into another, more ordered crystal form with p3 symmetry (Brisson et al., 1999a). The interpretation is that the p6 crystals involve rather weak interactions between the annexin V trimers, and that the stress induced by transfer onto a carbon support allows the system to stabilize into a structure of higher crystalline order, formed by the p3 assembly of annexin V trimers (Oling et al., 2000). No diffraction was obtained when X-ray grazing incidence was performed on cholera toxin, although the conditions used where systematically conducive to the observation of highly ordered 2D crystals when the monolayer was transferred onto plain carbon electron microscope grids (Lenne et al., 2000). The same is true for experiments using Brewster angle microscopy: no features indicating crystalline areas could be observed (Fig. 4.1A/B). Careful observation of holey grids showed that a basal layer of CTB was covering the holes. This layer was not crystalline but rather disordered. Then, on top of this basal layer were highly ordered crystalline domains. These crystalline domains were observed with both holey films and plain carbon films. The nature and mechanism of growth of these domains of CTB observed on top of a protein monolayer is not yet clear. The first disordered protein layer might form an adequate substrate for the adsorption and crystal growth of an additional layer. The growth of 3D crystals would then be expected. This indeed has been found: up to 4–5 layers of CTB domains have been observed (Fig. 4.2). This explanation would imply that specific binding to a lipidic ligand is not directly involved in the formation of such crystals. However, Brisson (Brisson et al., 1999a) pointed out that some protein-lipid complexes seen on the grids could help the
78
Strategies for Two-dimensional Crystallization of Proteins
(A)
(B)
Fig. 4.1 Images obtained with BAM on layers of CTB on monolayers of lipids GM1/DOPC 1:3 (mol/mol). (A): no dense domains has been observed after several hours of incubation. Only an optical in-homogeneity is observed (B) after 24h — probably due to the denaturation of the protein. Figure taken from Courty et al., 2002.
formation of these crystalline domains on top of the basal protein layer. It is interesting to note that for both cases, CTB and annexin V, the rigidity of the monolayer has been reported to be very low: 5mN/m and 3mN/m, respectively (Lenne et al., 2000). In both cases, we know that either there is no diffraction detected by X-ray grazing incidence (CTB), or if diffraction exists (Annexin V), the interaction of proteins inside the crystals is very weak. In the case of streptavidin or HupR, where the 2D crystals observed on EM grids do really pre-exist at the interface before transfer, the shear rigidity value is 90mN/m and 58mN/m, respectively (Courty et al., 2002; Lenne et al., 2000). The shear rigidity value is an excellent indication of the order of the crystals. Various approaches have been tried to overcome this problem of transfer •
In a variation of the Langmuir-Schaefer technique, a wire loop can be used to pick up the surface layer in a very gentle way (Asturias and Kornberg, 1995). A wire loop, which should be slightly larger than an electron microscope grid, can be lowered onto the surface of a drop in a way that the entire loop is in contact with the surface. Care must be taken that the loop does not pass through the water/lipid interface into the subphase. After that an electron microscope grid is used to pick up the film by bringing it in parallel
Electron Microscopic Observations and Image Analysis 79
Fig. 4.2 This figure shows the accumulation of layers of 2D crystals of CTB. The lipid monolayer is made of GM1/DOPC 1:3 (mol/mol). Scale bar 0.1m.
•
contact with the loop. With this technique it is possible to use continuous as well as holey carbon film. This technique was more successful for RNA polymerase II than using holey carbon film alone (Asturias and Kornberg, 1995). Otherwise, this method, which has so far only been applied in a limited number of studies, did not present advantages for the proteins annexin V, streptavidin and cholera toxin (Brisson et al., 1999a), nor for the His-HupR protein which crystallizes on a Ni-lipid monolayer (S. Courty and C. Vénien-Bryan, data not published). Glutaraldehyde was used to stabilize a fragile crystal form of horse liver apoferrin (Scheybani et al., 1996). The square lattice
80
Strategies for Two-dimensional Crystallization of Proteins
type crystals were not previously observed and probably became disordered during transfer onto electron microscope grids when not stabilized by chemical cross-linking. Introducing glutaraldehyde into the subphase of the crystallization trough improved the transfer efficiency of streptavidin dramatically (Ku et al., 1993). The shear rigidity of the crystals generally increases after injection of glutharaldehyde into the subphase (0.5% final (v/v)) (Lenne et al., 2000), for instance, shear rigidity is increased from 90mN/m to 200mN/m in the case of 2D crystals of streptavidin and from 58mN/m to 70mN/m in the case of HupR. The overall usefulness of chemical cross-linkers is questionable when it comes to high resolution data because the effect of chemical cross-linking on the protein structure is difficult to predict. Taken together, it is clear that the transfer of surface crystals onto electron microscope grids is a crucial point during sample preparation. In many cases, holey carbon film provides the best preservation of the crystals. In cases where no crystals have been observed using holey grids, it is important to try plain carbon film. It has been reported that for factor Va, the crystals are observed only on carbon plain film (Stoylova et al., 1994). The investigator must decide which method works best in each particular case and this important aspect of sample preparation deserves further investigation. To have a variety of transfer methods available will increase the chances of finding a suitable method for each particular case.
4.2. Preparation of the Specimen for Observation with an Electron Microscope 4.2.1. Negative staining Biological specimens are sensitive to electron radiation. They are relatively thin to allow penetration of the strongly interacting electrons and consist of light elements (C, H, N, O, that poorly scatter electrons). This limits the contrast. Since the electron microscope is necessarily under high vacuum to allow passage of the electrons, the specimens must be preserved or protected from damage, deformation and collapsing that would result from dehydration. A simple method to protect the specimen
Electron Microscopic Observations and Image Analysis 81
and enhance contrast is to replace the solvent by heavy metal salts such as uranyl acetate, ammonium molybdate or phosphotungstic acid (Bremer et al., 1992). The sample is deposited on an electron microscope grid covered with a thin (5–10nm) carbon film. Carbon film is used to provide a mechanically stable support, which is translucent to electrons and which conducts electrons. The most commonly used stain is uranyl acetate at a concentration of 2% w/v. Uranyl acetate envelopes the protein molecules and provides some fixation of the sample. Residual stain is blotted off and the sample is air-dried. Upon drying, the specimen is cast in a mold of stain that strongly scatters electrons and is less sensitive to electron radiation. In effect the contrasting agent surrounding the protein is reproduced in the image rather than the protein itself. Therefore the name negative staining. This technique is ideal for screening crystallization samples, because the high contrast makes detection of a crystal lattice easy. The negative staining method, however, shows drastic limitations when it comes to probing specimen details smaller than 1.5–2nm. The contrast arises from the stain and only marginally from the specimen. Consequently, the surface of the protein is revealed and not its internal structure. Moreover, the stain may interact preferentially with some residues, locally penetrate the protein, or be excluded by part of the surface and therefore the consistency between the actual volume of the protein and the stain excluding volume is fulfilled only to a limited resolution.
4.2.2. Cryo-electron microscopy The goal in preparing specimens for cryo-microscopy is to keep the biological sample as close as possible to its native state in order to preserve the structure to atomic or near atomic resolution in the microscope and during microscopy. The methods by which numerous types of macromolecules and macromolecular complexes have been prepared for cryo electron microscopy study are now well established (Dubochet et al., 1988). This specimen preparation involves cooling samples at a rate of about 10,000⬚C/sec, fast enough to permit vitrification (to a solid glass-like state) rather than crystallization of the bulk water. To do this, the sample, often less than 5L, at 0.2–5mg/mL is applied to a carbon or holey carbon support film, and blotted with a piece of filter paper. The grid is then rapidly frozen in a cryogen, usually liquid ethane (kept near its freezing
82
Strategies for Two-dimensional Crystallization of Proteins
point by a reservoir of liquid nitrogen). The ability to freeze samples on a timescale of milliseconds affords cryo electron microscopy one of its unique and, as yet, perhaps most under utilized advantages: capturing and visualizing dynamic structural events that occur over time periods of a few milliseconds or longer. For instance, spraying acetylcholine onto its receptor just before freezing the specimen for observation causes the receptor to open (Unwin, 1995); lowering the pH of an enveloped virus sample initiates early events of viral fusion (Fuller et al., 1995). Crystalline 2D samples can often be prepared for cryo EM by means of an alternative, simpler procedure, and vitrification of the bulk water is not always essential to achieve success. This alternative procedure consists of washing the specimen applied to the EM grid with 0.5–4% solutions of solutes like glucose, trehalose or tannic acid as cryo protectants in the buffer to prevent the formation of ice-crystals. After blotting with filter paper to remove excess solution the grid is frozen in liquid nitrogen. The blotted grid can also be loaded at room temperature into a cold holder, inserted into the microscope, and, finally cooled to liquid-nitrogen temperature. All subsequent steps, up to and including the recording of images under low dose conditions in the microscope are carried out in a manner that maintains the sample below ~⫺170⬚C to avoid devitrification which occurs at ~⫺140⬚C and leads to recrystallization of the bulk water to form cubic ice. Specimen preparation for cryo electron microscopy is easier to describe than perform. Success or failure depends critically on many factors such as: sample concentration (usually much higher than that needed for negative staining) wetting properties of the support film, need for hydrophobic or hydrophilic grids, time of sample adsorption to the film, humidity near the sample, extent of blotting and time elapsed before freeze-plunging, and concentrations and types of solutes present in the aqueous sample or the need to remove them.
4.3. The Electron Microscope 4.3.1. Formation of the image When an electron interacts with a free atom, it is simultaneously attracted to the nucleus because of the nuclear positive charge and
Electron Microscopic Observations and Image Analysis 83
repelled by the surrounding electron cloud of the atom. An electron scattering event is a composite of these forces. An electron “sees” the potential function of the atom, which can be approximated as a “screened Coulomb potential function”. This function is often referred to as a mass-density function and is analogous to the electron density function in the case of an X-ray photon, which is scattered only by the electrons of an atom. Because of the strong interactions between an electron and an atom, the scattering cross section of an atom is about 105 times greater for electrons than it is for X-rays, so significant scattering using electrons is obtained for crystals or other specimens that are 1–10nm thick, whereas scattering or adsorption of a similar fraction of an illuminating X-ray beam requires crystals that are 100–500m thick. The other main difference between the use of electrons and X-rays to probe structure is that electrons are much more easily focused than X-rays since they are charged particles that can be deflected by electric or magnetic fields. As a result, electron lenses are much superior to X-ray lenses and can be used to produce a magnified image of an object as easily as a diffraction pattern. This allows the electron microscope to be switched back and forth instantly between imaging and diffraction modes. By contrast, X-ray microscopy has been much less valuable than X-ray diffraction, but may be useful for imaging at the cellular level. Inside the microscope, the incident electron beam passes through the specimen and individual electrons are either unscattered or scattered by the atoms of the specimen. This scattering occurs either elastically or inelastically. The coherent, elastically scattered electrons contain all the highresolution information describing the structure of the specimen. The amplitudes and phases of the scattered electron beams are directly related to the amplitudes and phases of the Fourier components of the atomic distribution in the specimen. When the scattered beams are recombined with the unscattered beam in the image, they create an interference pattern (the image) which, for thin specimens, is related approximately linearly to the density variations in the specimen. Compared to the un scattered electrons, the scattered electrons have the same amplitude, but the phase is shifted by ≈ /2. Since biological specimens consist mainly of light elements these thin samples scatter the electrons
84
Strategies for Two-dimensional Crystallization of Proteins
only weakly. They behave as a weak phase object. As in light microscopy, phase contrast can be used to generate an image. Whereas in light microscopy, a phase plate is used to allow interference of the scattered and the unscattered beam to create an amplitude modulation, in the electron microscope, the objective lens focus setting allows the creation of a phase shift that is sufficient to give phase contrast. The contrast transfer function which modulates the image contrast depends on the focus setting. In effect different spatial frequencies, i.e. different resolution bands are transmitted differently. The phase contrast switches from positive to negative contrast passing via zero for alternate resolution bands. Therefore, image analysis is essential to correct for the effects of the contrast transfer function. This correction has to be made for each image individually. For cryo electron microscopy of unstained specimens the under-focus normally used is from 0.5–4.0m. During an inelastic scattering event, energy is transmitted from the electron to the object. This is initially transferred to secondary electrons which have an average energy (20eV) that is 5–10 times greater than the valence bond energies. These secondary electrons interact with other components of the specimen and produce numerous reactive chemical species including free radicals. In turn, these react with the embedded macromolecules and create a great variety of radiation products such as modified side-chains, cleaved polypeptide backbones and a host of molecular fragments. The result is a serious damage of the sample, and the consequence is that the resolution that can be achieved is decreased. Inelastically scattered electrons do not add to useful information on the image.
4.3.2. Instrumental advances A way to avoid radiation damage is to work at low temperature. For 100kV electrons the dose that can be used to produce an image in which the starting structure at high resolution is still recognizable is about 1 electron/Å2 for organic or biological materials at room temperature, 5 electron/Å2 for a specimen near liquid-nitrogen temperature (⫺170⬚C) and 10 electron/Å2 for a specimen near liquid-helium temperature (4–8⬚K) (Stark et al., 1996). Microscopes incorporating liquid heliumcooled stages require considerably more sophisticated instrumentation
Electron Microscopic Observations and Image Analysis 85
and there are additional practical difficulties associated with the poor conductivity of ice and carbon at very low temperatures. But it seems that very low temperature microscopy could play an increasingly significant role in the future. There are different types of electron emitters, including tungsten filaments, LaB6 crystals and field emission guns, all of which use different mechanisms to generate the electrons. The field emission source is another technological development that has become widely implemented over the last few years. It produces the brightest, most monochromatic beam. The improved coherence of the field emission source (both spatial and temporal) leads to substantial improvements in contrast transfer at high resolution, particularly when the image is strongly defocused. A field emission source is the best choice for high-resolution data collection. The recording medium of an electron microscope can be a photographic film or a slow-scan charge-coupled device (CCD) camera. In recent years, a more extensive use has been made of the CCD camera because of its large dynamic range and digital output. CCD cameras have greatly enhanced the speed, accuracy of collection and the processing of electron diffraction data. If the crystals are large enough, — the size required depends on the unit cell dimensions, but is usually in the range of one to a few m — electron diffraction data can be recorded. Electron diffraction does not contain any phase information, but the acquired amplitude information is less noisy and thus more precise than the amplitude information that can be retrieved from images. Diffraction patterns are easier to record than high-resolution images, because they are less sensitive to drift. However, for high-resolution image recording, when the recorded area, the pixel resolution, the signalto-noise ratio and the dynamic range must be considered, photographic film is the optimal choice. Theoretically, electron microscopy may be used for the direct experimental determination of atomic charge states in macromolecules. This comes from the fact that there is a substantial difference in the electron scattering factors of neutral and ionized atoms at resolution below 5Å. This phenomenon has been observed with small organic molecules, and results with biological 2D crystals, plant light-harvesting complex (Kühlbrandt et al., 1994) and bacteriorhodopsin (Kimura et al., 1997),
86
Strategies for Two-dimensional Crystallization of Proteins
suggest that the charge state of atoms may have a significant effect on the observed Coulombic potential map.
4.4. Image Processing To determine the 3D structure of an object by cryo electron microscopy methods, the specimen must be frozen, vitrified as a thin film, transferred to the electron microscope and photographed by means of lowdose selection and focusing procedures. The resulting image must be digitized and then processed using a series of sophisticated computer programs that allow different views of the specimen to be combined into a 3D reconstruction. These image processing programs used to this end have been developed over many years by the groups of the Medical Research Council (MRC), Cambridge, UK, (Crowther et al., 1996). They are currently used by most groups in the field. Other programs have been developed; they are summarized in a special issue of the Journal of Structural Biology (Special Issue, 1996). Important steps of the processing are described below.
4.4.1. Averaging imaging to improve the resolution In the case of unstained frozen hydrated specimens, the poor contrast is hidden by strong noise arising from the statistical interaction of electrons with matter, under the low irradiation conditions required for structural preservation. The signal-to-noise ratio increases with the square root of the number of averaged images. A considerable number of images are necessary to get high resolution information. For instance, to determine the 3D structure of the light harvesting complex II at about 3Å resolution, Kühlbrandt et al. used around 5 million molecules (Kühlbrandt et al., 1994). The most favorable case in the averaging process is when all macromolecules are ordered in a 2D crystal. All macromolecular projections are then identical except for the noise and are spatially related by constant translational vectors. However, the images are generally imperfect because of bending of the crystal. A computational procedure called “unbending” has been devised, using crosscorrelation techniques to identify the precise position of each unit cell
Electron Microscopic Observations and Image Analysis 87
(Henderson et al., 1986). Amplitudes and phases of the Fourier components of this unbent crystal are obtained. In addition, there are various technical corrections that must be made to the image data to allow an unbiased model of the structure to be obtained. These include correction of the amplitudes and phases from each image for the phase contrast function CTF, and for the effects of beam tilt. The data from many other crystals are merged by scaling and origin refinement taking into account the proper symmetry of the 2D space group of the crystal. For 2D crystals, it is as well possible to correct for loss of high-resolution contrast for any reason by “sharpening” the data by application of a negative temperature factor (Havelka et al., 1995).
4.4.2. Three-dimensional information The observation of a single macromolecular projection is not sufficient to describe its 3D shape unless multiple projections arising from different observation directions are combined. The mathematics of 3D reconstructions has been established a long time ago (Radon, 1917) but the earliest successful application of the idea of combining projections to reconstruct the 3D structure of a biological assembly was made by DeRosier and Klug (1968). The idea is that after Fourier transformation each 2D projection corresponds to a central section of the 3D transform of the assembly. If enough independent projections are obtained, then the 3D transform will have been fully sampled and the structure can then be obtained by back-transformation of the averaged, interpolated and smoothed 3D transform. In practice, a series of micrographs of single 2D crystals are recorded at different tilt angles, with random azimuthal orientations. Each crystal is unbent and corrected as described above then finally the whole data set is fitted by least squares to constrained amplitudes and phases along the lattice lines (Agard, 1983) prior to calculating a map of the structure. The absolute hand of the structure is automatically correct since the 3D structure is calculated from images with known tilt axes and tilt angle. Nevertheless, care must be taken not to make any number of trivial mistakes that would invert the hand. While images of untilted samples kept at liquid nitrogen or liquid helium temperature have routinely given resolutions in a range of 3–4Å, tilted samples often show a reduced resolution for
88
Strategies for Two-dimensional Crystallization of Proteins
diffraction spots that are far from the tilt axis. The reasons are mainly the sample charging and the crystal flatness. Spot-scan illumination appears to be one approach to reduce specimen charging effects (Downing, 1991). Improvements in carbon film flatness have been achieved by the use of special molybdenum grids (Booy and Pawley, 1993; Fujiyoshi, 1998).
4.4.3. Combining modelling and interpretation of results One area that is providing more and more insights into biological systems is work in which electron microscopic maps of large macromolecular complexes, obtained at medium resolution, are combined with atomic structures of their constituting parts, thereby putting the atomic structures into their functional context. X-ray crystallography and NMR studies still provide structural data of proteins to the highest resolution. But it is often difficult to crystallize large protein complexes and membrane proteins. Structural information of the entire complex at an intermediate resolution range of 10–30Å obtained by electron crystallography or single particle analysis can allow to position and orient the individual components that are known to high resolution within these maps and to identify the interaction surfaces. This procedure will usually allow specification of their relative positions and orientations to within a few angstroms, thus identifying the interaction surfaces. This crucial information about interactions can not be inferred from the structures of the individual components, even at very high resolution. An example for which this approach promises to be enlightening is the understanding of the molecular basis of the movement of motor proteins on microtubules (Rice et al., 1999). In another case, mechanistic information about the mode of action could be obtained by fitting structures of two distantly related members of the P-type ATPase family (Kühlbrandt et al., 2002).
Chapter 5
Practical Considerations
The technique of 2D crystallization using lipid monolayers is dependent on an interaction between the protein and the head group of the lipid film at the air/water interface. The lipid monolayer is formed by spreading the lipid (usually a mixture of different lipids) which is dissolved in organic solvent on the surface of the liquid. The protein of interest which will be present in the sub-phase binds to the lipid monolayer either unspecifically via electrostatic interaction or by specific interaction via a specialized headgroup. This process will lead to a high concentration of the protein at the air/water interface. The lipid monolayer provides enough fluidity to permit diffusion of the protein-lipid complex in the plane. This leads to a self-organization of the protein into a 2D crystal.
5.1. General Practical Considerations for Soluble Proteins or Multi-Protein Complexes Experiments for 2D crystallization of proteins using lipid monolayers take between hours and several days. They are usually performed at a temperature around 20⬚C. This means that the protein has to be sufficiently stable for the course of the experiment. Requirements for buffer conditions are usually flexible enough to be favorable for the stability of the particular protein. The experiments do not require a high protein concentration. The usual range is between 10–150g/mL. Crystallization trials are carried out in Teflon wells. Because electron microscopy is still the most common way to screen for crystallization 89
90
Strategies for Two-dimensional Crystallization of Proteins
conditions, the Teflon well should have a diameter of 4mm (with the electron microscope grids having a diameter of 3mm). With a depth of 1–2mm that makes about 30–40L per crystallization trial. The Teflon should be thoroughly cleaned with a detergent, boiled for 10 minutes or sonicated in chloroform for 5 minutes, followed by an extensive rinse with tap water and then deionized water. The Teflon should be hydrophobic after this treatment. The protein solution is then placed into the wells. A Hamilton syringe is used to deposit the lipid solution onto the drop. Lipid solutions are prone to oxidation and must be stored under Argon in an air-tight glass container equipped with a Teflon cap. Lipid solutions are stored in organic solvents at ⫺20⬚C and handled with glass syringes. The lipid should be dissolved in a mixture of chloroform/hexan (1:1, v/v) at a concentration of 500M. 0.5L of the lipid solution is placed on top of the crystallization drop. The lipid spreads and leaves a monolayer at the air/water interface while the organic solvents evaporate. The amount of lipid is more than would be required for a single layer, usually 5–10 times surplus. The excess lipid forms a reservoir at the edge of the Teflon well. The incubation time lasts usually from 2 to 24 hours in a humid chamber made of a Petri dish, were some water saturated paper is placed at the bottom of the dish. The layer can then be transferred to electron microscope grids. The transfer often works best with grids covered with a hydrophobic carbon film. The film can be rendered hydrophobic by baking it for one hour at 150⬚C. If the transfer is not satisfactory it is worth trying hydrophilic carbon film. This kind of film can be produced by glow discharge in an evacuated chamber. Observation of the transferred monolayer by electron microscopy will show if there is any protein binding. Bound protein can be recognized as extended grey areas (see Fig. 2.5). If there is a layer of protein, higher magnification will show if it is arranged in an ordered way. With smaller proteins it is necessary to take images and check them for diffraction at an optical bench. From here two routes can be taken to advance the experiment as depicted in Fig. 5.1. If there is no or not sufficient binding of protein to the monolayer, there are four main parameters to be modified. The first one is the protein concentration. Although in many cases protein binding to the monolayer can be observed at concentrations as low as 15g/mL, it might be necessary to increase the protein concentration to up to 200g/mL. Another
Add protein to Teflon well Modify protein concentra -tion Change ratio: diluting lipid / ligand lipid
Modify temperature
Spread lipid Binding of protein to monolayer?
Modify incubation time
Ordered arrangement of protein?
Modify fluidity of lipid monolayer using various diluting lipids
Transfer Modify incubation time
Modify buffer
Image analysis
Fig. 5.1 Practical considerations for protein crystallization on lipid monolayers.
Practical Considerations 91
Electron microscopic observation
92
Strategies for Two-dimensional Crystallization of Proteins
important factor is the ratio of ligand lipid to diluting lipid. It is worth trying a range of 1:1 to 1:12. This ratio can have a great influence on the density at which the protein binds to the monolayer. Dependent on the affinity of the protein to the lipid ligand, and dependent on the concentration of the protein, the incubation time may vary considerably. Allow 2 to 24 hours for the binding of the protein to the monolayer. In some cases changing the buffer conditions can help the binding of the protein. When using a His-tag protein in combination with a Ni-NTA-lipid, a pH around 8.0 is suitable. EDTA and DTT as well as imidazole should be avoided or be present at low concentration. Modifying these parameters should lead to a dense protein layer underneath the lipid monolayer. If this is achieved the protein has to be in a favorable environment to allow an ordered arrangement of the protein. The most important parameter is the fluidity of the lipid monolayer. This can be modified by using various diluting lipids. The lipid layer should be in the liquid expanded state or “fluid” phase. This physical property depends on the temperature, concentration, surface pressure and the buffer in the subphase and reflects the degree of interaction between the lipid molecules (Roberts, 1990). Only if the fluidity is high enough, the protein can move freely in the plane to arrange itself into an ordered array. Also the ratio between the ligand lipid and the diluting lipid is important (see above). Other important parameters are again temperature and incubation time. After it has been achieved that the protein forms crystalline sheets it is possible to move on to image analysis and structure determination.
5.2.1. Membrane protein crystallization Although the general advices given above are applicable to membrane proteins, a few more considerations need to be detailed. When crystallization of membrane proteins is performed, the lipids (same amount as for the soluble proteins) are first spread at the surface of the drop (which consists of buffer). Usually Ni-functionalized lipids are used. The Ni-fluorinated lipids need several hours (10 or more) to be perfectly stabilized. The solution containing the concentrated protein and lipid in detergent is then injected at the bottom of the well using an orifice specially designed for that purpose. A minuscule magnetic stirring bar previously placed at the bottom of the well is rotated to ensure
Practical Considerations 93
a good homogeneity of the solution. The final concentration of the membrane protein in the crystallization well is the same as for the soluble protein (i.e. 10–150 g/mL) and the lipid concentration should be such as the lipid protein ratio is in the range 0.2-1 (w/w). The lipids used for the reconstitution are usually DOPC, or E. coli lipids. The reconstitution step of the membrane protein into the lipid bilayer is done by elimination of detergent using BioBeads® which are added to the bottom of the well using a little orifice. The addition of these BioBeads® should be performed when the binding of the membrane protein to the lipid monolayer has been completed. This is easily checked by transferring the layer to an electron microscope grid. At low magnification, some large protein-containing planar domains should clearly be seen on the grids. If not, the advices given for soluble proteins, to increase the adsorption of protein onto the lipid monolayer, are applicable. Then, usually 24 hours after the addition of BioBeads®, the protein is reconstituted into the lipid bilayer and hopefully forms crystals. The speed of elimination of detergent during the reconstitution step is directly controlled by the amount of BioBeads® added into the trough (from 2mg to 5mg are usually added, about 10–30 beads). This parameter is important and often the quality of 2D crystals depends on it.
5.3. Single Particle Observation Immobilization of soluble membrane proteins or large complexes on lipid monolayers in a particular orientation is sometimes needed for structure determination using single particle analysis. In that case, the concentration of the particle on the surface should be limited as crystals are not needed and observation of individual, well separated particles is necessary. This can be achieved in several ways: •
• •
lowering the concentration of the protein; this is not always a good solution, as some proteins do not resist high dilution, the concentration of the protein in the small trough should not be less than 10g/mL. decreasing the ratio of ligand lipid to diluting lipid or decreasing the time of incubation, several minutes should be enough to collect the few particles needed for image analysis.
This page intentionally left blank
Chapter 6
Other New Methods
6.1. DNA Scaffolds A new approach aims to design DNA crystals with protein binding sites, or “DNA scaffolds” (Seeman, 2003; Turberfield, 2003). Such scaffolds would enable the construction of protein crystals in which the order is not solely dependent on protein-protein interactions, but is instead provided by the underlying DNA structure. Individual strands are designed to form four-arm (Holliday) junctions with short sticky ends (Fig. 6.1A) (Holliday, 1964). These sticky ends hybridize causing the Holliday junctions to assemble into the final extended structure (Fig. 6.1B). The 2D DNA arrays are assembled by gradual annealing. The nature of this final structure is dependent upon the design and configuration of the Holliday junction used. The sequence and length of the constituent DNA strands can be chosen to vary the size, periodicity and symmetry of the DNA crystal and its binding sites to suit and accommodate the desired protein. Cross-stacked junctions assemble to form a honeycomb-like lattice. The first protein used to bind to the DNA crystal was the bacterial homologous recombination protein, RuvA (Malo et al., 2005). A very different lattice forms from the same individual strands when the Holliday junction has been unfolded into a square-planar configuration by this protein. RuvA has an architectural role — it facilitates branch migration by unfolding the Holliday junction into a square planar. By changing the configuration of the Holliday junctions, RuvA completely changes the structure of the crystals that assemble when the four oligonucleotides are annealed. For this reason, the two crystals have different symmetries, and can not be interconverted without 95
96
Strategies for Two-dimensional Crystallization of Proteins
(A)
(B)
Fig. 6.1 DNA scaffolds.
disassembly. The perspective is to extend the application of the DNAscaffolding method to produce artificial crystals of other proteins, including membrane proteins. These proteins could be bound to ligands attached to covalently modified oligonucleotides in a self-assembled DNA array. These could include a general linker like and Ni2⫹-NTA group to bind histidine-tagged proteins or an avidin/biotin system as well as small-molecule receptor substrates. This method of structure determination may be useful for proteins or protein complexes not amenable to more conventional crystallization techniques.
6.2. Nanotubes The use of lipid nanotubes, cylindrical tubes of lipid, provides a scaffold for helical crystallization. The added advantage over interfacial crystallization techniques is that tilt series data collection is no longer needed for 3D reconstruction, as all necessary views of the protein are present in images of the helical protein arrangement. Therefore, the “missing cone” associated with the limits to which a periodic array can be tilted is no longer a problem, and a 3D map with isotropic resolution can be calculated (DeRosier and Moore, 1970a). In some approaches lipid nanotubes have been used as a substrate to bind proteins in an ordered
Other New Methods 97
way (Ringler et al., 1997; Wilson-Kubalek et al., 2000). Functionalized as well as various derivatized or charged unilamellar lipid tubes were used for binding and helical crystallization of several test proteins. Among these were His-tag proteins interacting with Ni2⫹-NTA-lipids, streptavidin interacting with biotinylated lipids and charged lipids interacting with surface charges of proteins (Wilson-Kubalek et al., 1998). The test proteins ranged in size from 35 kDa (annexin V) to 450 kDa (RNA polymerase). It was possible to work at very low protein concentrations, because adsorption is facilitated by electrostatic or specific high affinity interaction. Carbon nanotubes can also be used to bind protein in an ordered way. These tubes are several micrometers long with a diameter of 2–30 nm and can easily be prepared by the arc-discharge method (Ebbesen, 1997). They exhibit perfect straightness, cylindrical shape and strong rigidity. It has been shown that streptavidin as well as the DNA-binding protein HupR can bind to carbon nanotubes and acquire a helical organization (Balavoine et al., 1999). Furthermore, it was possible to functionalize the surface of the carbon nanotubes by binding a lipidic molecule with a NTA moiety to it to enable binding of any His-tagged protein (Richard et al., 2003). Both findings are the first steps to develop carbon nanotubes into a support for protein crystallization suitable for structure determination by helical reconstruction using electron microscopy.
This page intentionally left blank
Appendix
List of Proteins Crystallized on Planar Lipid Films
99
Lipid
Resolution (Å)
Projection Map or 3D Model
Neg. stain/cryo
Interaction
Reference
20S proteasomes (recombinant)
Ni-NTA-DODA/SOPC; Ni-NTA-DODA/DSPC
25
Projection Map
Neg. stain
Specific binding via His-tag
(Thess et al., 2002)
50S ribosomal DOPS/DOPC subunits from Bacillus stearothermophilus
20
Projection Map
Neg. stain
Electrostatic interaction
(Avila-Sakar et al., 1994)
␣-actinin from chicken
DDMA / DLPC
25
Projection Map
Neg. stain
Interaction with pos. charged lipids
(Taylor and Taylor, 1993; Taylor and Taylor, 1999)
␣-actinin from rabbit skeletal muscle
DLPC-chlorine/ DDMA-bromide or, DLPC/DDMA
15
3D Model
Cryo
Binding to pos. charged lipids
(Tang et al., 2001)
␣-toxin from Staphylococcus aureus
Platelet lipids
25
3D Model
Neg. stain
Interaction with platelet lipids
(Olofsson et al., 1990)
␣-toxin from Staphylococcus aureus
Platelet lipids
21
Projection Map
Neg. stain
Binding to neg. charged phospholipids
(Ward and Leonard, 1992)
100 Strategies for Two-dimensional Crystallization of Proteins
Protein
Lipid
Resolution (Å)
Projection Map or 3D Model
Neg. stain/cryo
Interaction
Reference
␣-toxin from Staphylococcus aureus
DMPS
17
3D Model
Neg. stain
Binding to neg. charged phospholipids
(Ellis et al., 1997)
Annexin IV
DMPE
35
Projection Map
Neg. stain
Binding to phospholipids
(Newman et al., 1991)
Annexin V
DOPC/DOPS
8
Projection Map
Neg. stain
Binding to neg. charged phospholipids
(Olofsson et al., 1994)
Annexin V
DOPS/DOPC
10
Projection Map
Cryo
Binding to neg. charged phospholipids
(Oling et al., 2001)
Annexin V
DOPS
20
3D Model
Neg. stain
Binding to neg. charged phospholipids
(Voges et al., 1994)
Annexin V
DOPC/DOPS
20
Projection Map
Neg. stain
Binding to neg. charged phospholipids
(Reviakine et al., 2001)
Annexin V
DOPC/DOPS
24
Projection Map
Neg. stain
Binding to neg. charged phospholipids
(Mosser et al., 1991)
Appendix: List of Proteins Crystallized on Planar Lipid Films 101
Protein
Lipid
Resolution (Å)
Projection Map or 3D Model
Neg. stain/cryo
Interaction
Reference
Annexin VI
DMPC/DMPS/DOPE
22
3D Model
Neg. stain
Binding to neg. charged phospholipids
(Avila-Sakar et al., 2000)
Annexin VI
DMPE
50
Projection Map
Neg. stain
Binding to phospholipids
(Newman et al., 1989)
Antibody (anti DNP IgG)
DNP-PE/PC
60
Projection Map
Neg. stain
Binding to hapten
(Uzgiris and Kornberg, 1983)
Antibody IgE
Phospholipid with hapten 2,6-dinitrophenyl groups
—
Projection Map
Neg. stain
Binding to hapten
(Uzgiris, 1987)
Antibodies from Phospholipid mouse: IgE, IgG2a and IgG2b; from rabbit: IgG
—
Projection Map
Neg. stain
Binding to hapten
(Uzgiris, 1990)
Antigen-antibody- DNP-PE/PC complement complex (C1q component)
—
Projection Map
Neg. stain
Binding to hapten and antibody
(Uzgiris and Kornberg, 1983)
102 Strategies for Two-dimensional Crystallization of Proteins
Protein
Lipid
Resolution (Å)
Projection Map or 3D Model
Neg. stain/cryo
Interaction
Reference
ArsA ATPase
DMPS/DOPC
24
Projection Map
Neg. stain
Binding to neg. charged phospholipids
(Wang et al., 2000)
Avidin
Biotinylated DPPE/DOPC
27
Projection Map
Neg. stain
Specific interaction
(Qin et al., 1995)
Cadherin
Ni-NTA-DOGS/DOPE
17
Projection Map
Neg. stain
Specific binding via His-tag
(Al-Kurdi et al., 2004)
Catalase
Poly(1-benzyl-L-histidine) (synthetic peptide)
25
Projection Map
Neg. stain
Electrostatic interaction
(Sato et al., 1993)
Chaperonin TF55 from Sulfolobus solfataricus
DOPG
18
Projection Map
Neg. stain
Binding to neg. charged phospholipids
(Ellis et al., 1998)
Cholera toxin
Egg PC/GM1
15
3D Model
Neg. stain
Specific interaction
(Ribi et al., 1988)
Cholera toxin
Egg PC/GM1
30
Projection Map
Neg. stain
Specific interaction
(Reed et al., 1987)
Cholera toxin, -subunit
Egg PC/GM1, DOPE/GM1
9
Projection Map
Cryo
Specific interaction
(Mosser et al., 1992)
Appendix: List of Proteins Crystallized on Planar Lipid Films 103
Protein
Lipid
Resolution (Å)
Projection Map or 3D Model
Neg. stain/cryo
Interaction
Reference
Cholera toxin, -subunit
Egg PC/GM1, DOPE/GM1
15
Projection Map
Neg. stain
Specific interaction
(Ludwig et al., 1986)
Coagulation factor Va
PS/PC
20
Projection Map
Neg. stain
Binding to neg. charged phospholipids
(Stoylova et al., 1994)
Coagulation factor VIII
DOPC/DOPS
15
Projection Map
Neg. stain
(Stoylova et al., 1999)
Coagulation factor IX
DOPC/DOPS
30
Projection Map
Neg. stain
Binding to neg. charged phospholipids Binding to neg. charged phospholipids
C-reactive protein (human)
Egg-PC/ lyso-PC
20
Projection Map
Neg. stain
Specific interaction
(Wang and Sui, 2001a)
C-reactive protein (rabbit)
Egg-PC/lyso-PC/DMPS
20
Projection Map
Neg. stain
electrostatic interaction
(Wu et al., 2003)
C-reactive protein (rabbit)
DS8PE/DOPC
22
Projection Map
Neg. stain
Specific interaction
(Sui et al., 1996)
C-reactive protein DS8PE/DOPC (rabbit)
26
Projection Map
Neg. stain
Specific interaction
(Wang and Sui, 1999)
(Stoylova et al., 1998)
104 Strategies for Two-dimensional Crystallization of Proteins
Protein
Lipid
Resolution (Å)
Projection Map or 3D Model
Neg. stain/cryo
Interaction
Reference
C-reactive proteins (pentameric, rabbit)
Egg-PC/ lyso-PC
22
Projection Map
Neg. stain
Specific interaction
(Wang and Sui, 2001b)
DNA gyrase -subunit
Novobiocin-DOPA
10
Projection Map
Cryo
Specific binding
(Celia et al., 1994)
DNA gyrase -subunit
Novobiocin-DOPA
27
Projection Map
Neg. stain
Specific binding
(Lebeau et al., 1990)
DNA gyrase -subunit
Novobiocin-DOPA
27
3D Model
Neg. stain
Specific binding
(Celia et al., 1994)
Ferritin
Stearic acid methyl ester eicosic-trimethyl ammonium bromide
—
—
—
—
(Fromherz, 1971)
Ferritin (Catalase, chaperonin 50S ribosomal subunit)
DHAA (organic liquid)
25 (Catalase) 20 (Chaperonin)
Projection Neg. stain Map (Catalase, chaperonin and 50S ribosomal subunit) 3D Model (50S ribosomal subunit)
Interaction with pos. charged organic liquid
(Aoyama et al., 1995)
Appendix: List of Proteins Crystallized on Planar Lipid Films 105
Protein
Lipid
Resolution (Å)
Projection Map or 3D Model
Neg. stain/cryo
Interaction
Reference
FhuA (porin)
Ni-NTA-DOGS/ E. coli lipids
15
Projection Map
Neg. stain
Specific binding via His-tag
(Lévy et al., 1999)
9
Projection Map
Cryo
Specific binding via His-tag
(Plisson et al., 2003)
Helper component Ni-NTA-DOGS/DOPC proteinase (HC-Pro) HIV-1 capsid protein
Ni-DHGN
24
Projection Map
Cryo
Specific binding via His-tag
(Barklis et al., 1998)
HSP16.3 (small heat shock protein)
Stearylamine/DOPC
22
3D Model
Neg. stain
Interaction with pos. charged lipids
(Chen et al., 2003)
HupR, response regulator from Rhodobacter capsutulas
Ni-NTA-DOGA/DOPC
9
Projection Map
Cryo
Specific binding via His-tag
(Vénien-Bryan et al., 2000)
HupR, response regulator from Rhodobacter capsutulas
Ni-NTA-DOGA/DOPC
18
Projection Map
Neg. stain
Specific binding via His-tag
(Vénien-Bryan et al., 1997)
106 Strategies for Two-dimensional Crystallization of Proteins
Protein
Lipid
Resolution (Å)
Projection Map or 3D Model
Neg. stain/cryo
Interaction
Reference
Leukemia virus capsid protein (moloney murine)
Ni-NTA-DOGS/PC
15
3D Model
Neg. stain
Specific binding via His-tag
(Ganser et al., 2003)
Leukemia virus capsid protein (moloney murine)
Ni-DHGN/PC or Ni-DOGS/PC
22
3D Model
Neg. stain
Specific binding via His-tag
(McDermott et al., 2000)
Leukemia virus capsid protein (moloney murine)
Ni-DHGN/PC or Ni-DOGS/PC
10
Projection Map
Neg. stain
Specific binding via His-tag
(Barklis et al., 1997)
MHC, class 1, murine
Ni-NTA-DOGS/DOPC
14
3D Model
Neg. stain
Specific binding via His-tag
(Celia et al., 1999)
Myelin basic protein (MBP) (recombinant; murine)
Ni-NTA-DOGS/PI or Ni-NTA-DOGS/PI(4)P
—
—
Neg. stain
Specific binding via His-tag
(Hill et al., 2002)
Myosin I (brush border)
DOPS, DOPS/DLPC
20
Projection Map
Neg. stain
Binding to neg. charged phospholipids
(Celia et al., 1996)
Appendix: List of Proteins Crystallized on Planar Lipid Films 107
Protein
Lipid
Resolution (Å)
Projection Map or 3D Model
Neg. stain/cryo
Interaction
Reference
Myosin I (brush border)
DOPS
20
3D Model
Neg. stain
Binding to neg. charged phospholipids
(Jontes and Milligan, 1997)
Neurotoxin complex (botulinum )
LC/POPC
15
Projection Map
Neg. stain
Specific binding via gangliosidelipids
(Burkard et al., 1997)
P-glycoprotein Ni-NTA-DOGS/eggPC (Pgp) from mouse
22
Projection Map
Neg. stain
Specific binding via His-tag
(Lee et al., 2002)
Protein kinase C (PKC) delta
10
Projection Map
Neg. stain
Affinity for neg. charged lipids
(Solodukhin et al., 2002)
9
Projection Map
Cryo
Specific binding via His-tag
(Lebeau et al., 2001)
DOPC/DOPS/DO
Proton ATPase Nickel functionalized from plant plasma fluorinated lipids membranes Reverse transcriptase (HIV I)
Ni-NTA-DOPE/egg-PE
21
Projection Map
Neg. stain
Specific binding via His-tag
(Kubalek et al., 1994)
Ribonucleotide reductase
dATP-PE/egg-PC
18
Projection Map
Neg. stain
Specific binding
(Ribi et al., 1987)
108 Strategies for Two-dimensional Crystallization of Proteins
Protein
Lipid
Resolution (Å)
Projection Map or 3D Model
Neg. stain/cryo
Interaction
Reference
RNA polymerase holoenzyme from E. coli
Egg PC/octadecylamine
27
3D Model
Neg. stain
Binding to pos. charged lipids
(Darst et al., 1989)
RNA polymerase holoenzyme from E. coli
Egg PC/octadecylamine
30
Projection Map
Neg. stain
Binding to pos. charged lipids
(Darst et al., 1988)
RNA polymerase I Egg PC/octadecylamine (yeast)
12
Projection Map
Cryo
Binding to pos. charged lipids
(Asturias et al., 1998)
RNA polymerase I Ni-NTA-DOGA (yeast)
25
Projection Map
Neg. stain
Specific binding via surface histidines
(Bischler et al., 1998)
RNA polymerase I PC/PS (yeast)
35
Projection Map
Neg. stain
Binding to pos. charged lipids
(Schultz et al., 1990)
RNA polymerase II DMPC/octadecylamine (yeast)
30
Projection Map
Neg. stain
Binding to pos. charged lipids
(Edwards et al., 1990)
Appendix: List of Proteins Crystallized on Planar Lipid Films 109
Protein
Projection Map or 3D Model
Neg. stain/cryo
Interaction
Reference
RNA polymerase II Octadecylamine/egg-PC (yeast)
16
3D Model
Neg. stain
Binding to pos. charged lipids
(Darst et al., 1991b)
PC/PS; cetyl-N⫹(CH3)3 / oleyl alcohol
30
3D Model
Neg. stain
Binding to pos. charged lipids
(Schultz et al., 1993)
RNA polymerase II PC/ stearylamine (yeast)
20
Projection Map
Neg. stain
Binding to pos. charged lipids
(Poglitsch et al., 1999)
Sticholysin II
Bovine brain sphingomyelin/egg-PC and egg-PC:cholesterol
15
3D Model
Neg. stain
Nonspecific interaction
(Martin-Benito et al., 2000)
Streptavidin
Different biotinylated lipids
—
—
—
Specific binding
(Blankenburg et al., 1989)
Streptavidin
Biotin-X-DPPE/egg-PC
—
—
Neg. stain
Specific binding
(Kubalek et al., 1991)
Streptavidin
DOPC/biotin-X-DPPE
Projection Map
Cryo
Specific binding
(Avila-Sakar and Chiu, 1996)
RNA polymerase II,I (yeast)
Lipid
3
110 Strategies for Two-dimensional Crystallization of Proteins
Resolution (Å)
Protein
Neg. stain/cryo
Interaction
Reference
5
Projection Map
n.a.
Nonspecific binding
(Furuno and Sasabe, 1993)
N-biotinyl(S-[1,2-bis (octadecyloxycarbonyl) ethyl]cysteine)
12
3D Model
Neg. stain
Specific binding
(Darst et al., 1991a)
Streptavidin
Cu-DOIDA
15
Projection Map
Neg. stain
Specific binding via surface histidines
(Frey et al., 1998; Pack et al., 1997a)
Tetanus toxin
Egg PC/GM1
14
3D Model
Neg. stain
(Robinson et al., 1988)
TF0F1 ATP synthase
Ni-NTA-DOGS/ E. coli lipids
30
Projection Map
Neg. stain
Specific ligand interaction Specific binding via His-tag
Lipid
Streptavidin
Poly(1-benzyl-histidine)
Streptavidin
Resolution (Å)
(Lévy et al., 1999)
Appendix: List of Proteins Crystallized on Planar Lipid Films 111
Projection Map or 3D Model
Protein
This page intentionally left blank
Glossary
g L 2D 3D Å AFM AMPPCP ATP BAM biotin-X-DPPE CMC CTB Cu-DOIDA DCPU-lipids DDMA DHAA DHGN
DLPC DMPC DNP-PE DO
microgram (10⫺6 gram) microliter (10⫺6 liter) two-dimensional three-dimensional Ångstrom (1Å ⫽ 0.1nm) atomic force microscope non-hydrolyzable analog of ATP adenosine triphosphate Brewster angle microscope N-((6-(biotinoyl)amino)heanoyl) dipalmitoyl-L-␣-phosphatidylethanolamine critical micellar concentration cholera toxin B 1,2-dioleyl-rac-glycero-3-(8-3,6-dioxy) octyl-1-amino-N,N-diacetic acid dichlorophenyl-lipids didodecyldimethylammonium dehydroabietylamine 1,2-di-O-hexadecyl-sn glycero-3-(19-299-Rhydroxy-39-N-(5-amino-1-carboxypentyl) iminodiacetic acid)propyl ether dilaurylphosphatidylcholine dimiristoylphosphatidylcholine N-dinitrophenyl phosphatidylethanolamine diolein 113
114 Strategies for Two-dimensional Crystallization of Proteins
DODA DOGA DOGS DOIDA DOPA DOPC DOPE DPPE DS8PE DSA DSIDA DSPC DTT EDTA EM ESR GM1 His-tag IDA KD kDa LC LHC M mg mL mM MoCa MW Ni2⫹
N␣,N␣-bis(carboxymethyl)-NC⫺[(dioctadecylamino)succinyl]-L-lysine 1,3-di-O-oleylglyceroxy-acetyl-amino hexanoic acid 1,2-dioleoyl-sn-glycero-3-succinate 1,1’-[[9-[2,3-bis[(Z)-octadec-9-enyloxy]propyl]3,6,9-trioxanonyl]imino] ⫻ diacetic acid 1,2-dioleoyl-sn-glycero-3-phosphatidic acid 1,2-dioleoyl-sn-glycero-3- phosphatidylcholine 1,2-dioleolyl-sn-glycero-3phosphatidylethanolamine 1,2-dipalmitoyl-sn-glycero-3phosphatidylethanolamine dioctdecanyl N-[N⬘-(aminoethyl phosphatoethyl)succinamido-N-yl-]aspartate 6-[2-1,3-di-O-stearyl-glyceroxy-acetyl-amino] hexanoic acid distearoyl-glycerylether-iminodiacetic acid 1,2-distearoyl-sn-glycero-3-phosphatidylcoline dithio treitol ethylene diamino tetra-acetic acid electron microscope/electron microscopy electron spin resonance monosialoganglioside histidine tag, used for interaction with Ni2⫹-NTA group iminodiacetate dissociation constant kilo Dalton, molecular weight lactosyl ceramide light harvesting complex molar milligram (10⫺3 gram) milliliter (10⫺3 liter) millimolar Moloney murine leukemia virus capsid protein molecular weight nickel ion
Glossary 115
nm NMR NTA PC PI PI4P POPC PSIDA
SNOM SOPC v/v w/w
nanometer (10⫺9 meter) nuclear magnetic resonance nitrilotriacetic acid, also called N,N-bis [carboxymethyl]glycine phosphatidylcholine phosphatidylinositol phosphatidylinositol-4-phosphate palmitoyl oleoyl phosphatidylcholine 1,1%-[[2-octadecyloxy,3-(9pyrenylnonyloxy)propyl]-3,6,9-trioxanonyl] ⫻ imino diacetic acid scanning near field optical microscopy 1-stearoyl-2-oleoyl-sn-glycero-3phosphatidylcholine volume by volume weight by weight
This page intentionally left blank
Bibliography
Agard, D.A. 1983, A least-squares method for determining structure factors in three-dimensional tilted-view reconstructions. Journal of Molecular Biology 167:849–852. Al-Kurdi, R., D. Gulino-Debrac, L. Martel, J.F. Legrand, A. Renault, E.A. Hewat and C. Vénien-Bryan. 2004, A soluble VE-cadherin fragment forms 2D arrays of dimers upon binding to a lipid monolayer. Journal of Molecular Biology 337:881–892. Als-Nielsen, J. 1991, X-ray reflectivity studies of liquid surfaces, in Handbook on Synchrotron Radiation, 3:471–503. Brown, G. and D.E. Moncton, editors. Elsevier North-Holland, Amsterdam, Netherlands. Als-Nielsen, J., D. Jacquemain, K. Kjaer, F. Leveiller, M. Lahav and L. Leiserowitz. 1994, Principles and applications of grazing incidence X-ray and neutron scattering from ordered molecular monolayers at the air-water interface. Physics Reports-Review Section of Physics Letters 246:251–313. Altenburger, J.M., L. Lebeau, C. Mioskowski and D. Schirlin. 1992, Synthesis of New Glycerolipids Linked to Hydroxamate Derivatives Designed for 2-Dimensional Crystallization of Aminopeptidase-M. Helvetica Chimica Acta 75:2538–2544. Andree, H.A., C.P. Reutelingsperger, R. Hauptmann, H.C. Hemker, W.T. Hermens and G.M. Willems. 1990, Binding of vascular anticoagulant alpha (VAC alpha) to planar phospholipid bilayers. Journal of Biological Chemistry 265:4923–4928. Aoyama, K., K. Ogawa, Y. Kimura and Y. Fujiyoshi. 1995, A method for 2D crystallization of soluble proteins at liquid-liquid interface. Ultramicroscopy 57:345–354. 117
118 Strategies for Two-dimensional Crystallization of Proteins
Asturias, F.J. and R.D. Kornberg. 1995, A novel method for transfer of two-dimensional crystals from the air/water interface to specimen grids. EM sample preparation/lipid-layer crystallization. Journal of Structural Biology 114:60–66. Asturias, F.J., W. Chang, Y. Li and R.D. Komberg. 1998, Electron crystallography of yeast RNA polymerase II preserved in vitreous ice. Ultramicroscopy 70:133–143. Avila-Sakar, A.J., T.L. Guan, T. Arad, M.F. Schmid, T.W. Loke, A. Yonath, J. Piefke, F. Franceschi and W. Chiu. 1994, Electron cryomicroscopy of Bacillus stearothermophilus 50S ribosomal subunits crystallized on phospholipid monolayers. Journal of Molecular Biology 239:689–697. Avila-Sakar, A.J. and W. Chiu. 1996, Visualization of beta-sheets and side-chain clusters in two-dimensional periodic arrays of streptavidin on phospholipid monolayers by electron crystallography. Biophysical Journal 70:57–68. Avila-Sakar, A.J., R.H. Kretsinger and C.E. Creutz. 2000, Membranebound 3D structures reveal the intrinsic flexibility of annexin VI. Journal of Structural Biology 130:54–62. Azzam, R.M.A. and N.M. Bashara. 1977, Ellipsometry and Polarized Light, North Holland, Amsterdam, The Netherlands. Balakrishnan, K., F.J. Hsu, A.D. Cooper and H.M. McConnell. 1982, Lipid hapten containing membrane targets can trigger specific immunoglobulin-E-dependent de-granulation of rat basophil leukemia-cells. Journal of Biological Chemistry 257:6427–6433. Balavoine, F., P. Schultz, C. Richard, V. Mallouh, T.W. Ebbesen and C. Mioskowski. 1999, Helical crystallization of proteins on carbon nanotubes: A first step towards the development of new biosensors. Angewandte Chemie-International Edition 38:1912–1915. Barklis, E., J. McDermott, S. Wilkens, E. Schabtach, M.F. Schmid, S. Fuller, S. Karanjia, Z. Love, R. Jones, Y. Rui, X. Zhao and D. Thompson. 1997, Structural analysis of membrane-bound retrovirus capsid proteins. EMBO Journal 16:1199–1213. Barklis, E., J. McDermott, S. Wilkens, S. Fuller and D. Thompson. 1998, Organization of HIV-1 capsid proteins on a lipid monolayer. Journal of Biological Chemistry 273:7177–7180. Baruchel, J., J.L. Hodeau, M.S. Lehmann, J.R. Regnard and C. Schlenker. 1993, Neutron and synchrotron radiation for condensed matter studies, in Les éditions de physique: Les Ulis. Springer-Verlag, France. Bendas, G. 2001, Immunoliposomes — A promising approach to targeting cancer therapy. Biodrugs 15:215–224.
Bibliography 119
Berge, B., P.F. Lenne and A. Renault. 1998, X-ray grazing incidence diffraction on monolayers at the surface of water, Current Opinion in Colloids & Interface Science 3:312–326. Bischler, N., F. Balavoine, P. Milkereit, H. Tschochner, C. Mioskowski and P. Schultz. 1998, Specific interaction and two-dimensional crystallization of histidine tagged yeast RNA polymerase I on nickel-chelating lipids. Biophysical Journal 74:1522–1532. Blankenburg, R., P.H. Meller, H. Ringsdorf and C. Salesse. 1989, Interaction between biotin lipids and streptavidin in monolayers: Formation of oriented two-dimensional protein domains induced by surface recognition. Biochemistry 28:8214–8221. Blokzjil, W. and B.F.N. Engberts. 1993, Hydrophobic effects. Opinion and facts. Angewandte Chemie-International Edition 32:1545–1579. Booy, F.P. and J.B. Pawley. 1993, Cryo-crinkling — What happens to carbon-films on copper grids at low-temperature. Ultramicroscopy 48:273–280. Bremer, A., C. Henn, A. Engel, W. Baumeister and U. Aebi. 1992, Has negative staining still a place in biomacromolecular electron microscopy. Ultramicroscopy 46:85–111. Brisson, A. and P.N. Unwin. 1984, Tubular crystals of acetylcholine receptor. Journal of Cell Biology 99:1202–1211. Brisson, A., G. Mosser and R. Huber. 1991, Structure of soluble and membrane-bound human annexin V. Journal of Molecular Biology 220:199–204. Brisson, A., W. Bergsma-Schutter, F. Oling, O. Lambert and I. Reviakine. 1999a, Two-dimensional crystallization of proteins on lipid monolayers at the air-water interface and transfer to an electron microscopy grid. Journal of Crystal Growth 196:456–470. Brisson, A., O. Lambert and W. Bergsma-Schutter. 1999b, Twodimensional crystallization of soluble proteins on planar lipid films, in Crystallization of Nucleic Acids and Proteins: A Practical Approach. Ducruix, A. and R. Giegé, editors. Oxford University Press, Oxford, UK. Brockman, H. 1999, Lipid monolayers: why use half a membrane to characterize protein-membrane interactions? Current Opinion in Structural Biology 9:438–443. Burkard, F., F. Chen, G.M. Kuziemko and R.C. Stevens. 1997, Electron density projection map of the botulinum neurotoxin 900-kilodalton complex by electron crystallography. Journal of Structural Biology 120:78–84.
120 Strategies for Two-dimensional Crystallization of Proteins
Celia, H., L. Hoermann, P. Schultz, L. Lebeau, V. Mallouh, D.B. Wigley, J.C. Wang, C. Mioskowski and P. Oudet. 1994, 3-Dimensional model of Escherichia coli gyrase-B subunit crystallized in 2-Dimensions on novobiocin-linked phospholipid films. Journal of Molecular Biology 236:616–628. Celia, H., J.D. Jontes, M. Whittaker and R.A. Milligan. 1996, Twodimensional crystallization of brush border myosin I. Journal of Structural Biology 117:236–241. Celia, H., E. Wilson-Kubalek, R.A. Milligan and L. Teyton. 1999, Structure and function of a membrane-bound murine MHC class I molecule. Proceedings of the National Academy of Sciences of the United States of America 96:5634–5639. Chami, M., G. Pehau-Arnaudet, O. Lambert, J.L. Ranck, D. Levy and J.L. Rigaud. 2001, Use of octyl beta-thioglucopyranoside in twodimensional crystallization of membrane proteins. Journal of Structural Biology 133:64–74. Chen, Y., Y.J. Lu, H.W. Wang, S. Quan, Z.Y. Chang and S.F. Sui. 2003, Two-dimensional crystallization of a small heat shock protein HSP16.3 on lipid layer. Biochemical and Biophysical Research Communications 310:360–366. Cole, S.T., C. Condon, B.D. Lemire and J.H. Weiner. 1985, Molecular biology, biochemistry and bioenergetics of fumarate reductase, a complex membrane-bound iron-sulfur flavoenzyme of Escherichia coli. Biochimica et Biophysica Acta 811:381–403. Courty, S. 2001, Solides bidimensionnels à l’interface air-eau: mesures mécaniques et optiques. Thesis, University Joseph Fourier-Grenoble I, France. Courty, S., L. Lebeau, L. Martel, P.F. Lenné, F. Balavoine, W. Dischert, O. Konovalov, C. Mioskowski, J.F. Legrand and C. Vénien-Bryan. 2002, Two-dimensional crystallization of a histidine-tagged protein on monolayers of fluidity-enhanced Ni2⫹-chelating lipids. Langmuir 18:9502–9512. Crowther, R.A., R. Henderson and J.M. Smith. 1996, MRC image processing programs. Journal of Structural Biology 116:9–16. Crucifix, C., M. Uhring and P. Schultz. 2004, Immobilization of biotinylated DNA on 2-D streptavidin crystals. Journal of Structural Biology 146:441–451. Darst, S.A., H.O. Ribi, D.W. Pierce and R.D. Kornberg. 1988, Twodimensional crystals of Escherichia coli RNA polymerase holoenzyme on positively charged lipid layers. Journal of Molecular Biology 203:269–274.
Bibliography 121
Darst, S.A., E.W. Kubalek and R.D. Kornberg. 1989, Three-dimensional structure of Escherichia coli RNA polymerase holoenzyme determined by electron crystallography. Nature 340:730–732. Darst, S.A., M. Ahlers, P.H. Meller, E.W. Kubalek, R. Blankenburg, H.O. Ribi, H. Ringsdorf and R.D. Kornberg. 1991a, Two-dimensional crystals of streptavidin on biotinylated lipid layers and their interactions with biotinylated macromolecules. Biophysical Journal 59:387–396. Darst, S.A., A.M. Edwards, E.W. Kubalek and R.D. Kornberg. 1991b, Three-dimensional structure of yeast RNA polymerase II at 16Å resolution. Cell 66:121–128. de Feijter, J.A., J. Benjamins and F.A. Veer. 1978, Ellipsometry as a tool to study the adsorption behavior of synthetic and biopolymers at the air-water interface. Biopolymers 17:1759–1772. DeRosier, D.J. and A. Klug. 1968, Reconstruction of three dimension structures from electron microsgraphs. Nature 217:130–134. DeRosier, D.J. and P.B. Moore. 1970a, Three-dimensional reconstruction of F-actin, thin filaments and decorated thin filaments. Journal of Molecular Biology 52:355–369. DeRosier, D.J. and P.B. Moore. 1970b, Reconstruction of three-dimensional images from electron micrographs of structures with helical symmetry. Journal of Molecular Biology 52:355–369. Dietrich, C., L. Schmitt and R. Tampe. 1995, Molecular-organization of histidine-tagged biomolecules at self-assembled lipid interfaces using a novel class of chelator lipids. Proceedings of the National Academy of Sciences of the United States of America 92:9014–9018. Dietrich, C., O. Boscheinen, K.D. Scharf, L. Schmitt and R. Tampe. 1996, Functional immobilization of a DNA-binding protein at a membrane interface via histidine tag and synthetic chelator lipids. Biochemistry 35:1100–1105. Downing, K.H. 1991, Spot-scan imaging in transmission electron microscopy. Science 251:53–59. Drakopoulou, E., G.M. Tsivgoulis, A. Mukhopadhyay and A. Brisson. 2000, Design and synthesis of multifunctional phospholipids. Tetrahedron Letters 41:4131–4134. Drude, P. 1887, Annalen der Physik 32:584. Drude, P. 1888, Annalen der Physik 34:489. Dubochet, J., M. Adrian, J.J. Chang, J.C. Homo, J. Lepault, A.W. McDowall and P. Schultz. 1988, Cryoelectron microscopy of vitrified specimens. Quarterly Reviews of Biophysics 21:129–228.
122 Strategies for Two-dimensional Crystallization of Proteins
Dutta, P., J. Beng, B. Lin, J. Ketterson, M. Prakash, P. Georgopoulos and S. Ehrlich. 1987, X-ray diffraction studies of organic monolayers on the surface of water. Physical Review Letters 58:2228–2231. Ebbesen, T.W. 1997, Carbon Nanotubes — Preparation and Properties, CRC, Boca Raton, FL. Edwards, A.M., S.A. Darst, W.J. Feaver, N.E. Thompson, R.R. Burgess and R.D. Kornberg. 1990, Purification and lipid-layer crystallization of yeast RNA polymerase II. Proceedings of the National Academy of Sciences of the United States of America 87:2122–2126. Edwards, T.C., S. Koppenol, W. Frey, W.R. Schief, V. Vogel, R.E. Stenkamp and P.S. Stayton. 1998, Molecular basis for ionic strength dependence and crystal morphology in two-dimensional streptavidin crystallization. Langmuir 14:4683–4687. Edwards, T.C., N. Malmstadt, S. Koppenol, M. Hara, V. Vogel and P.S. Stayton. 2002, Molecular basis for asymmetrical growth in twodimensional streptavidin crystals. Langmuir 18:7447–7451. Egger, M., S.P. Heyn and H.E. Gaub. 1992, Synthetic lipid-anchored receptors based on the binding-site of a monoclonal-antibody. Biochimica et Biophysica Acta 1104:45–54. Ellis, M.J., H. Hebert and M. Thelestam. 1997, Staphylococcus aureus alpha-toxin: characterization of protein/lipid interactions, 2D crystallization on lipid monolayers, and 3D structure. Journal of Structural Biology 118:178–188. Ellis, M.J., S. Knapp, P.J. Koeck, Z. Fakoor-Biniaz, R. Ladenstein and H. Hebert. 1998, Two-dimensional crystallization of the chaperonin TF55 from the hyperthermophilic archaeon Sulfolobus solfataricus. Journal of Structural Biology 123:30–36. Fotiadis, D., Y. Liang, S. Filipek, D.A. Saperstein, A. Engel and K. Palczewski. 2003, Atomic-force microscopy: rhodopsin dimers in native disc membranes. Nature 421:127–128. Frederix, P.L.T.M., T. Akiyamay, U. Staufery, C. Gerberz, D. Fotiadis, D. Müller and A. Engel. 2003, Atomic force bio-analytics. Current Opinion in Chemical Biology 7:641–647. Frey, W., W.R. Schief and V. Vogel. 1996, Two-dimensional crystallization of streptavidin studied by quantitative Brewster angle microscopy. Langmuir 12:1312–1320. Frey, W., J. Brink, W.R. Schief, W. Chiu and V. Vogel. 1998, Electron crystallographic analysis of two-dimensional streptavidin crystals coordinated to metal-chelated lipid monolayers. Biophysical Journal 74:2674–2679.
Bibliography 123
Frishman, D. and H.W. Mewes. 1997, Protein structural classes in five complete genomes. Nature Structural Biology 4:626–628. Fromherz, P. 1971, Electron microscopic studies of lipid protein films. Nature 231:267–268. Fujiyoshi, Y. 1998, The structural study of membrane proteins by electron crystallography. Advances in Biophysics 35:25–80. Fuller, S.D., J.A. Berriman, S.J. Butcher and B.E. Gowen. 1995, Low pH induces swiveling of the glycoprotein heterodimers in the Semliki Forest virus spike complex. Cell 81:651–653. Furuno, T. and H. Sasabe. 1993, Two-dimensional crystallization of streptavidin by nonspecific binding to a surface film: study with a scanning electron microscope. Biophysical Journal 65:1714–1717. Gaines, G.L. 1966, Insoluble Monolayers at Liquid-Gas Interfaces, Interscience Publishers, New-York, USA. Ganser, B.K., A.C. Cheng, W.I. Sundquist and M. Yeager. 2003, Threedimensional structure of the M-MuLV CA protein on a lipid monolayer: a general model for retroviral capsid assembly. EMBO Journal 22:2886–2892. Gast, A.P., C.R. Robertson, S.W. Wang and M.T. Yatcilla. 1999, Twodimensional streptavidin crystals: macropatterns and microorganization. Biomolecular Engineering 16:1–4. Gerke, V. and S.E. Moss. 1997, Annexins and membrane dynamics. Biochimica et Biophysica Acta — Molecular Cell Research 1357:129–154. Green, N.M. 1975, Avidin. Advances in Protein Chemistry 29:85–133. Haas, H., G. Brezesinski and H. Möhwald. 1995, X-ray diffraction of a protein crystal anchored at the air/water interface. Biophysical Journal 68:312–314. Haensler, J. and F. Schuber. 1988, Preparation of neo-galactosylated liposomes and their interaction with mouse peritoneal-macrophages. Biochimica et Biophysica Acta 946:95–105. Hafeman, D.G., V. Vontscharner and H.M. McConnell. 1981, Specific antibody-dependent interactions between macrophages and lipid haptens in planar lipid monolayers. Proceedings of the National Academy of Sciences of the United States of America 78:4552–4556. Harlow, E. and D. Lane. 1988, Antibodies: A Laboratory Manual 27–28. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY, USA. Havelka, W.A., R. Henderson and D. Oesterhelt. 1995, Three-dimensional structure of halorhodopsin at 7Å resolution. Journal of Molecular Biology 247:726–738.
124 Strategies for Two-dimensional Crystallization of Proteins
Helm, C.A., H. Möhwald, K. Kjaer and J. Alsnielsen. 1987, Phospholipid monolayer density distribution perpendicular to the watersurface — a synchrotron X-ray reflectivity study. Europhysics Letters 4:697–703. Hemming, S.A., A. Bochkarev, S.A. Darst, R.D. Kornberg, P. Ala, D.S.C. Yang and A.M. Edwards. 1995, The mechanism of protein crystal-growth from lipid layers. Journal of Molecular Biology 246: 308–316. Henderson, R. and P.N. Unwin. 1975, Three-dimensional model of purple membrane obtained by electron microscopy. Nature 257:28–32. Henderson, R., J.M. Baldwin, K.H. Downing, J. Lepault and F. Zemlin. 1986, Structure of purple membrane from Halobacterium halobium: recording, measurement and evaluation of electron micrographs at 3.5Å resolution. Ultramicroscopy 19:147–178. Hendrickson, W.A., A. Pahler, J.L. Smith, Y. Satow, E.A. Merritt and R.P. Phizackerley. 1989, Crystal-structure of core streptavidin determined from multiwavelength anomalous diffraction of synchrotron radiation. Proceedings of the National Academy of Sciences of the United States of America 86:2190–2194. Hénon, S. and J. Meunier. 1991, Microscope at the Brewster-angle — direct observation of 1st-order phase-transitions in monolayers. Review of Scientific Instruments 62:936–939. Herz, J.M., W.J. Thomsen and G.G. Yarbrough. 1997, Molecular approaches to receptors as targets for drug discovery. Journal of Receptor & Signal Transduction Research 17:671–776. Hill, C.M., I.R. Bates, G.F. White, F.R. Hallett and G. Harauz. 2002, Effects of the osmolyte trimethylamine-N-oxide on conformation, self-association, and two-dimensional crystallization of myelin basic protein. Journal of Structural Biology 139:13–26. Hirata, Y. and J. Miyake. 1994, Molecular construction of a photosynthetic reaction center at the interface by its affinity with quinonylphospholipid. Thin Solid Films 244:865–868. Hochuli, E., H. Dobeli and A. Schacher. 1987, New metal chelate adsorbent selective for proteins and peptides containing neighboring histidine-residues. Journal of Chromatography 411:177–184. Hoffmann, A. and R.G. Roeder. 1991, Purification of his-tagged proteins in non-denaturing conditions suggests a convenient method for protein interaction studies. Nucleic Acids Research 19:6337–6338. Holliday, R. 1964, A mechanism for gene conversion in fungi. Genetics Research 5:282–304.
Bibliography 125
Hönig, D. and D. Möbius. 1991, Direct visualization of monolayers at the air-water-interface by Brewster-angle microscopy. Journal of Physical Chemistry 95:4590–4592. Huber, R., J. Romisch and E.P. Paques. 1990, The crystal and molecularstructure of human annexin-V, an anticoagulant protein that binds to calcium and membranes. EMBO Journal 9:3867–3874. Jahn, T., J. Dietrich, B. Andersen, B. Leidvik, C. Otter, C. Briving, W. Kühlbrandt and M.G. Palmgren. 2001, Large scale expression, purification and 2D crystallization of recombinant plant plasma membrane H⫹-ATPase. Journal of Molecular Biology 309:465–476. Jones, T.A., J.Y. Zou, S.W. Cowan and A. Kjeldgaard. 1991, Improved methods for binding protein models in electron density maps and the location of errors in these models. Acta Crystallographica — Section A — Foundations of Crystallography 47:110–119. Jonsson, U., M. Malmqvist and I. Ronnberg. 1985, Adsorption of immunoglobulin-G, protein-a, and fibronectin in the submonolayer region evaluated by a combined study of ellipsometry and radiotracer techniques. Journal of Colloid and Interface Science 103:360–372. Jontes, J.D. and R.A. Milligan. 1997, Three-dimensional structure of Brush Border Myosin-I at approximately 20 Å resolution by electron microscopy and image analysis. Journal of Molecular Biology 266:331–342. Kaetzel, M.A., Y.D. Mo, T.R. Mealy, B. Campos, W. Bergsma-Schutter, A. Brisson, J.R. Dedman and B.A. Seaton. 2001, Phosphorylation mutants elucidate the mechanism of annexin IV-mediated membrane aggregation. Biochemistry 40:4192–4199. Kessel, M., M.J. Brennan, B.L. Trus, M.E. Bisher and A.C. Steven. 1988, Naturally crystalline porin in the outer membrane of Bordetella pertussis. Journal of Molecular Biology 203:275–278. Kimura, Y., D.G. Vassylyev, A. Miyazawa, A. Kidera, M. Matsushima, K. Mitsuoka, K. Murata, T. Hirai and Y. Fujiyoshi. 1997, Surface of bacteriorhodopsin revealed by high-resolution electron crystallography. Nature 389:206–211. Kjaer, K., J. Als-Nielsen, C. Helm, L.A. Laxhuber and H. Möhwald. 1987, Ordering in lipid monolayers studied by synchrotron X-ray diffraction and fluorescence microscopy. Physics Review Letters 58:2224–2227. Konig, J., J. Prenen, B. Nilius and V. Gerke. 1998, The annexin Ii-P11 complex is involved in regulated exocytosis in bovine pulmonary artery endothelial cells. Journal of Biological Chemistry 273: 19679– 19684.
126 Strategies for Two-dimensional Crystallization of Proteins
Kornberg, R.D. and H.O. Ribi. 1987, Formation of two-dimensional crystals of proteins on lipid layers, in Protein Stuctrue, Folding and Design 2: 175–186. Oxender, D. L., editor. Alan R. Liss, New York, USA. Kornberg, R.D. and S.A. Darst. 1991, Two-dimensional crystals of proteins on lipid layers. Current Opinion in Structural Biology 1:642–646. Ku, A.C., S.A. Darst, R.D. Kornberg, C.R. Robertson and A.P. Gast. 1992, Dendritic growth of 2-dimensional protein crystals. Langmuir 8:2357–2360. Ku, A.C., S.A. Darst, C.R. Robertson, A.P. Gast and R.D. Kornberg. 1993, Molecular analysis of 2-dimensional protein crystallization. Journal of Physical Chemistry 97:3013–3016. Kubalek, E.W., R.D. Kornberg and S.A. Darst. 1991, Improved transfer of 2-dimensional crystals from the air-water-interface to specimen support grids for high-resolution analysis by electron-microscopy. Ultramicroscopy 35:295–304. Kubalek, E.W., S.F. Le Grice and P.O. Brown. 1994, Two-dimensional crystallization of histidine-tagged, HIV-1 reverse transcriptase promoted by a novel nickel-chelating lipid. Journal of Structural Biology 113:117–123. Kühlbrandt, W. 1992, Two-dimensional crystallization of membrane proteins. Quarterly Reviews of Biophysics 25:1–49. Kühlbrandt, W., D.N. Wang and Y. Fujiyoshi. 1994, Atomic model of plant light harvesting complex by electron crystallography. Nature 367:614–621. Kühlbrandt, W., J. Zeelen and J. Dietrich. 2002, Structure, mechanism, and regulation of the neurospora plasma membrane H⫹-ATPase. Science 297:1692–1696. Landau, L.D. and E.M. Lifchitz. 1969, Elasticity, Mir, Moscow, Russia. Langmuir, I. and V.J. Schaefer. 1938, Activities of urease and pepsin monolayers. Journal of the American Chemical Society 60:1351–1360. Laukkanen, M.L., K. Alfthan and K. Keinanen. 1994, Functional immunoliposomes harboring a biosynthetically lipid-tagged singlechain antibody. Biochemistry 33:11664–11670. Lebeau, L., E. Regnier, P. Schultz, J.C. Wang, C. Mioskowski and P. Oudet. 1990, 2-dimensional crystallization of DNA gyrase-B subunit on specifically designed lipid monolayers. FEBS Letters 267: 38–42. Lebeau, L., P. Oudet and C. Mioskowski. 1991, Synthesis of new phospholipids linked to steroid-hormone derivatives designed for two-dimensional crystallization of proteins. Helvetica Chimica Acta 74:1697–1706.
Bibliography 127
Lebeau, L., P. Schultz, H. Célia, P. Mésini, S. Nuss, C. Klinger, S. Olland, P. Oudet and C. Mioskowski. 1996, Specifically designed lipid assemblies as tools for two-dimensional crystallization of soluble biological macromolecules, in Handbook of Non-Medical Applications of Liposomes, Vol. II: 153–186. Barenholz, Yechezkel and Danilo D. Lasic, editors. CRC Press, London, UK. Lebeau, L., S. Nuss, P. Schultz, P. Oudet and C. Mioskowski. 1999, Selfassembly of soluble proteins on functionalized lipid layers: a tentative correlation between the fluidity properties of the lipid film and protein ordering. Chemistry and Physics of Lipids 103:37–46. Lebeau, L., F. Lach, C. Vénien-Bryan, A. Renault, J. Dietrich, T. Jahn, M.G. Palmgren, W. Kühlbrandt and C. Mioskowski. 2001, Two-dimensional crystallization of a membrane protein on a detergentresistant lipid monolayer. Journal of Molecular Biology 308:639–647. Lee, J.Y., I.L. Urbatsch, A.E. Senior and S. Wilkens. 2002, Projection structure of P-glycoprotein by electron microscopy — evidence for a closed conformation of the nucleotide binding domains. Journal of Biological Chemistry 277:40125–40131. Lee, K.D., A.B. Kantor, S. Nir and J.C. Owicki. 1993, Aggregation of hapten-bearing liposomes mediated by specific antibodies. Biophysical Journal 64(3):905–918 Lenne, P.-F. 1998, Deux exemples de cristallisation à la surface de l’eau: -cristaux 2-D de protéines; -monocouches d’alcools en présence d’amphiphiles solubles. Thesis, University Joseph Fourier-Grenoble I, France. Lenne, P.-F., B. Berge, A. Renault, C. Zakri, C. Vénien-Bryan, S. Courty, F. Balavoine, W. Bergsma-Schutter, A. Brisson, G. Grubel, N. Boudet, O. Konovalov and J.-F. Legrand. 2000, Synchrotron radiation diffraction from two-dimensional protein crystals at the air/water interface. Biophysical Journal 79:496–500. Lévy, D., G. Mosser, O. Lambert, G.S. Moeck, D. Bald and J.L. Rigaud. 1999, Two-dimensional crystallization on lipid layer: A successful approach for membrane proteins. Journal of Structural Biology 127:44–52. Lösche, M., E. Sackmann and H. Möhwald. 1983, A fluorescence microscopy study concerning the phase diagram of phospholipids. Berichte der Bunsen Gesellschaft für physikalische Chemie — An international Journal of Physical Chemistry 87:848–852. Ludwig, D.S., H.O. Ribi, G.K. Schoolnik and R.D. Kornberg. 1986, Two-dimensional crystals of cholera-toxin B-subunit receptor complexes — projected structure at 17Å resolution. Proceedings of
128 Strategies for Two-dimensional Crystallization of Proteins
the National Academy of Sciences of the United States of America 83:8585–8588. Malo, J., J.C., Mitchell, C. Vénien-Bryan, J.R. Harris, H. Wille, D.J. Sherratt and A.J. Turberfield 2005, Engineering a 2D Protein-DNA Crystal. Angewandte Chemie International Edition in press.(Online publication, Wiley-Vch Verlag GmbH & Co. KGaA, Weinham, April 2005). Maloney, K.M., W.R. Schief, D.W. Pack, W. Frey, F.H. Arnold and V. Vogel. 1999, Binding and two-dimensional crystallization of streptavidin at the air/water interface via engineered Cu-IDA chelator lipids. Coordination Chemistry Reviews 183:3–18. Marra, W.C., P. Eisenberger and A.V. Cho. 1979, Title for this reference: X-ray total-external-reflection-Bragg diffraction: A structural study of the GaAs-Al interface. Journal of Applied Physics 50:6927. Martin, F.J., W.L. Hubbell and D. Papahadjopoulos. 1981, Immunospecific targeting of liposomes to cells: a novel and efficient method for covalent attachment of Fab fragments via disulfide bonds. Biochemistry 20:4229–4238. Martin-Benito, J., F. Gavilanes, V. de los Rios, J.M. Mancheno, J.J. Fernandez and J.G. Gavilanes. 2000, Two-dimensional crystallization on lipid monolayers and three-dimensional structure of sticholysin II, a cytolysin from the sea anemone Stichodactyla helianthus. Biophysical Journal 78:3186–3194. Maruyama, K. 2000, In vitro targeting by liposomes. Biological & Pharmaceutical Bulletin 23:791–799. McConnell, H.M., L.K. Tamm and R.M. Weis. 1984, Periodic structures in lipid monolayer phase transitions. Proceedings of the National Academy of Sciences of the United States of America 81:3249–3253. McDermott, J., K. Mayo and E. Barklis. 2000, Three-dimensional organization of retroviral capsid proteins on a lipid monolayer. Journal of Molecular Biology 302:121–133. Michel, H., D. Oesterhelt and R. Henderson. 1980, Orthorhombic twodimensional crystal form of purple membrane. Proceedings of the National Academy of Sciences of the United States of America 77: 338–342. Mosser, G. and A. Brisson. 1991, Conditions of two-dimensional crystallization of cholera toxin B-subunit on lipid films containing ganglioside GM1. Journal of Structural Biology 106:191–198. Mosser, G., C. Ravanat, J.M. Freyssinet and A. Brisson. 1991, Subdomain structure of lipid-bound annexin V resolved by electron image analysis. Journal of Molecular Biology 217:241–246.
Bibliography 129
Mosser, G., V. Mallouh and A. Brisson. 1992, A 9 Å two-dimensional projected structure of cholera toxin B-subunit-GM1 complexes determined by electron crystallography. Journal of Molecular Biology 226:23–28. Mosser, G., C. Breyton, A. Olofsson, J.L. Popot and J.L. Rigaud. 1997, Projection map of cytochrome b(6)f complex at 8 angstrom resolution. Journal of Biological Chemistry 272:20263–20268. Mukerjee, P. and A.Y.S. Yang. 1976, Nonideality of mixing of micelles of fluorocarbon and hydrocarbon surfactants and evidence of partial miscibility from differential conductance data. Journal of Physical Chemistry 80:1388–1390. Müller, D.J., G.M. Hand, A. Engel and G.E. Sosinsky. 2002, Conformational changes in surface structures of isolated connexin 26 gap junctions. EMBO Journal 21:3598–3607. Newman, R., A. Tucker, C. Ferguson, D. Tsernoglou, K. Leonard and M.J. Crumpton. 1989, Crystallization of P68 on lipid monolayers and as three-dimensional single crystals. Journal of Molecular Biology 206:213–220. Newman, R.H., K. Leonard and M.J. Crumpton. 1991, 2D crystal forms of annexin IV on lipid monolayers. FEBS Letters 279:21–24. Ng, K., D.W. Pack, D.Y. Sasaki and F.H. Arnold. 1995, Engineering protein-lipid interactions — targeting of histidine-tagged proteins to metal-chelating lipid monolayers. Langmuir 11:4048–4055. Nuss, S., C. Mioskowski and L. Lebeau. 1999, Synthesis of new fluidityenhanced amphiphilic compounds for soluble protein two-dimensional crystallization purpose. Chemistry and Physics of Lipids 103:21–35. Oling, F., J. Santos, N. Govorukhina, C. Mazeres-Dubut, W. BergsmaSchutter, G. Oostergetel, W. Keegstra, O. Lambert, A. Lewit-Bentley and A. Brisson. 2000, Structure of membrane-bound annexin A5 trimers: a hybrid Cryo-EM-X-ray crystallography study. Journal of Molecular Biology 304:561–573. Oling, F., W. Bergsma-Schutter and A. Brisson. 2001, Trimers, dimers of trimers, and trimers of trimers are common building blocks of annexin A5 two-dimensional crystals. Journal of Structural Biology 133:55–63. Olofsson, A., U. Kaveus, I. Hacksell, M. Thelestam and H. Hebert. 1990, Crystalline layers and three-dimensional structure of Staphylococcus aureus alpha-toxin. Journal of Molecular Biology 214:299–306. Olofsson, A., V. Mallouh and A. Brisson. 1994, Two-dimensional structure of membrane-bound annexin V at 8Å resolution. Journal of Structural Biology 113:199–205.
130 Strategies for Two-dimensional Crystallization of Proteins
Paborsky, L.R., K.E. Dunn, C.S. Gibbs and J.P. Dougherty. 1996, A nickel chelate microtiter plate assay for six histidine-containing proteins. Analytical Biochemistry 234:60–65. Pack, D.W., G.H. Chen, K.M. Maloney, C.T. Chen and F.H. Arnold. 1997a, A metal-chelating lipid for 2D protein crystallization via coordination of surface histidines. Journal of the American Chemical Society 119:2479–2487. Pack, D.W., K. Ng, K.M. Maloney and F.H. Arnold. 1997b, Ligandinduced reorganization and assembly in synthetic lipid membranes. Supramolecular Science 4:3–10. Parratt, L.G. 1954, Surface studies of solids by total reflection of X-rays. Physical Review 95:359–369. Plisson, C., M. Drucker, S. Blanc, S. German-Retana, O. Le Gall, D. Thomas and P. Bron. 2003, Structural characterization of HC-Pro, a plant virus multifunctional protein. Journal of Biological Chemistry 278:23753–23761. Poglitsch, C.L., G.D. Meredith, A.L. Gnatt, G.J. Jensen, W.H. Chang, J. Fu and R.D. Kornberg. 1999, Electron crystal structure of an RNA polymerase II transcription elongation complex. Cell 98: 791–798. Qin, H., Z. Liu and S.F. Sui. 1995, Two-dimensional crystallization of avidin on biotinylated lipid monolayers. Biophysical Journal 68: 2493–2496. Radon, J. 1917, Über die Bestimmung von Funktionen durch ihre Integralwerte längs gewisser Mannigfaltigkeiten. Berichte über die Verhandlungen der Königlich Sächsischen Gesellschaft der Wissenschaften zu Leipzig. Mathematisch-Physikalische Klasse 69:262. Ratanabanankoon, P. and A.P. Gast. 2003, Effect of ionic strength on twodimensional streptavidin crystallization. Langmuir 19:1794–1801. Raynal, P. and H.B. Pollard. 1994, Annexins: the problem of assessing the biological role for a gene family of multifunctional calcium- and phospholipid-binding proteins. Biochimica et Biophysica Acta 1197:63–93. Reed, R.A., J. Mattai and G.G. Shipley. 1987, Interaction of cholera toxin with ganglioside GM1 receptors in supported lipid monolayers. Biochemistry 26:824–832. Reiter, R., H. Motschmann, H. Orendi, A. Nemetz and W. Knoll. 1992, Ellipsometric microscopy — imaging monomolecular surfactant layers at the air-water-interface. Langmuir 8:1784–1788. Reiter, R., H. Motschmann and W. Knoll. 1993, Ellipsometric characterization of streptavidin binding to biotin-functionalized lipid monolayers at the water air interface. Langmuir 9:2430–2435.
Bibliography 131
Remigy, H.W., D. Caujolle-Bert, K. Suda, A. Schenk, M. Chami and A. Engel. 2003, Membrane protein reconstitution and crystallization by controlled dilution. FEBS Letters 555:160–169. Reviakine, I., W. Bergsma-Schutter and A. Brisson. 1998, Growth of protein 2D crystals on supported planar lipid bilayers imaged in situ by AFM. Journal of Structural Biology 121:356–362. Reviakine, I., W. Bergsma-Schutter, A.N. Morozov and A. Brisson. 2001, Two-dimensional crystallization of annexin A5 on phospholipid bilayers and monolayers: a solid-solid phase transition between crystal forms. Langmuir 17:1680–1686. Ribi, H.O., P. Reichard and R.D. Kornberg. 1987, Two-dimensional crystals of enzyme effector complexes-ribonucleotide reductase at 18Å resolution. Biochemistry 26:7974–7979. Ribi, H.O., D.S. Ludwig, K.L. Mercer, G.K. Schoolnik and R. Kornberg. 1988, Three-dimensional structure of cholera toxin penetrating a lipid membrane. Science 239:1272–1276. Rice, S., A.W. Lin, D. Safer, C.L. Hart , N. Naber, B.O. Carragher, S.M. Cain, E. Pechatnikova, E.M. Wilson-Kubalek, M. Whittaker, E. Pate, R. Cooke, E.W. Taylor, R.A. Milligan and R.D. Vale. 1999, A structural change in the kinesin motor protein that drives motility. Nature 402:778–782. Richard, C., F. Balavoine, P. Schultz, T.W. Ebbesen and C. Mioskowski. 2003, Supramolecular self-assembly of lipid derivatives on carbon nanotubes. Science 300:775–778. Ringler, P., W. Müller, H. Ringsdorf and A. Brisson. 1997, Functionalized lipid tubules as tools for helical crystallization of proteins. Chemistry-a European Journal 3:620–625. Roberts, G. 1990, Langmiur-Blodgett Films, Plenum Press, New York. Robinson, J.P., M.F. Schmid, D.G. Morgan and W. Chiu. 1988, Threedimensional structural analysis of tetanus toxin by electron crystallography. Journal of Molecular Biology 200:367–375. Salord, J., C. Tarnus, C.D. Muller and F. Schuber. 1986, Targeting of liposomes by covalent coupling with Ecto-Nad⫹-glycohydrolase ligands. Biochimica Et Biophysica Acta 886:64–75. Sato, A., T. Furuno, C. Toyoshima and H. Sasabe. 1993, Two-dimensional crystallization of catalase on a monolayer film of poly(1-benzyl-Lhistidine) spread at the air/water interface. Biochimica et Biophysica Acta 1162:54–60. Scheuring, S., D.J. Müller, H. Stahlberg, H.A. Engel and A. Engel. 2002, Sampling the conformational space of membrane protein surfaces
132 Strategies for Two-dimensional Crystallization of Proteins
with the AFM. European Biophysics Journal with Biophysics Letters 31:172–178. Scheybani, T., H. Yoshimura, W. Baumeister and K. Nagayama. 1996, Stabilization of a fragile two-dimensional protein crystal at the waterair interface: the square lattice of apoferritin. Langmuir 12:431–435. Schief, W.R., T. Edwards, W. Frey, S. Koppenol, P.S. Stayton and V. Vogel. 1999, Two-dimensional crystallization of streptavidin: in pursuit of the molecular origins of structure, morphology, and thermodynamics. Biomolecular Engineering 16:29–38. Schief, W.R., S.R. Dennis, W. Frey and V. Vogel. 2000, Light scattering microscopy from monolayers and nanoparticles at the air/water interface. Colloids and Surfaces a-Physicochemical and Engineering Aspects 171:75–86. Schmitt, J., H. Hess and H.G. Stunnenberg. 1993, Affinity purification of histidine-tagged proteins. Molecular Biology Reports 18:223–230. Schmitt, L., C. Dietrich and R. Tampe. 1994, Synthesis and characterization of chelator-lipids for reversible immobilization of engineered proteins at self-assembled lipid interfaces. Journal of the American Chemical Society 116:8485–8491. Schmitt, L. and R. Tampe. 1996, ATP-lipids — protein anchor and energy source in two dimensions. Journal of the American Chemical Society 118:5532–5543. Schultz, P., H. Celia, M. Riva, S.A. Darst, P. Colin, R.D. Kornberg, A. Sentenac and P. Oudet. 1990, Structural study of the yeast RNA polymerase A. Electron microscopy of lipid-bound molecules and two-dimensional crystals. Journal of Molecular Biology 216:353–362. Schultz, P., H. Celia, M. Riva, A. Sentenac and P. Oudet. 1993, Threedimensional model of yeast RNA polymerase I determined by electron microscopy of two-dimensional crystals. EMBO Journal 12: 2601–2607. Seeman, N.C. 2003, DNA in a material world. Nature 421:427–431. Shnek, D.R., D.W. Pack, D.Y. Sasaki and F.H. Arnold. 1994, Specific protein attachment to artificial membranes via coordination to lipidbound copper(II). Langmuir 10:2382–2388. Solodukhin, A.S., H.L. Caldwell, J.J. Sando and R.H. Kretsinger. 2002, Two-dimensional crystal structures of protein kinase C-delta, its regulatory domain, and the enzyme complexed with myelin basic protein. Biophysical Journal 82:2700–2708. Special issue (guest editors: Carragher, B. and P.R. Smith). 1996, Advances in Computational Image Processing for Microscopy. Journal of Structural Biology 116.
Bibliography 133
Spevak, W., J.O. Nagy, D.H. Charych, M.E. Schaefer, J.H. Gilbert and M.D. Bednarski. 1993, Polymerized liposomes containing C-glycosides of sialic-acid — potent inhibitors of influenza-virus in vitro infectivity. Journal of the American Chemical Society 115:1146–1147. Stark, H., F. Zemlin and B. Boettcher. 1996, Electron radiation damage to protein crystals of bacteriorhodopsin at different temperatures. Ultramicroscopy 63:75–79. Stoylova, S., K.G. Mann and A. Brisson. 1994, Structure of membranebound human factor Va. FEBS Letters 351:330–334. Stoylova, S., E. Gray, T.W. Barrowcliffe, G. Kemball-Cook and A. Holzenburg. 1998, Structural determination of lipid-bound human blood coagulation factor IX. Biochimica et Biophysica Acta 1383:175–178. Stoylova, S.S., P.J. Lenting, G. Kemball-Cook and A. Holzenburg. 1999, Electron crystallography of human blood coagulation factor VIII bound to phospholipid monolayers. Journal of Biological Chemistry 274:36573–36578. Sui, S.F., Z. Liu, W. Li, C.D. Xiao, S.X. Wang, Q.F. Gao and Q.Z. Zhou. 1996, Two-dimensional crystallization of rabbit C-reactive protein on lipid monolayers. FEBS Letters 388:103–111. Tang, J.H., D.W. Taylor and K.A. Taylor. 2001, The three-dimensional structure of alpha-actinin obtained by cryoelectron microscopy suggests a model for Ca2⫹-dependent actin binding. Journal of Molecular Biology 310:845–858. Taylor, K.A. and D.W. Taylor. 1993, Projection image of smooth muscle alpha-actinin from two-dimensional crystals formed on positively charged lipid layers. Journal of Molecular Biology 230:196–205. Taylor, K.A. and D.W. Taylor. 1999, Structural studies of cytoskeletal protein arrays formed on lipid monolayers. Journal of Structural Biology 128:75–81. Thess, A., S. Hutschenreiter, M. Hofmann, R. Tampe, W. Baumeister and R. Guckenberger. 2002, Specific orientation and two-dimensional crystallization of the proteasome at metal-chelating lipid interfaces. Journal of Biological Chemistry 277:36321–36328. Toyoshima, C. and P.N. Unwin. 1988, Ion channel of acetylcholine receptor reconstructred from images of postsynaptic membranes. Nature 336:247–250. Trudel, E., J. Gallant, S. Mons, C. Mioskowski, L. Lebeau, K. Jeuris, P. Foubert, F. De Schryver and C. Salesse. 2001, Design of functionalized lipids and evidence for their binding to photosystem II core complex by oxygen evolution measurements, atomic force
134 Strategies for Two-dimensional Crystallization of Proteins
microscopy, and scanning near-field optical microscopy. Biophysical Journal 81:563–571. Turberfield, A. 2003, DNA as an engineering material. Physics World 16:43–46. Ulman, A. 1991, An Introduction to Ultrathin Organic Films-From LangmuirBlodgett to Self-Assembly, Academic Press, San Diego, USA. Unwin, N. 1995, Acetylcholine receptor channel imaged in the open state. Nature 373:37–43. Unwin, P.N. and R. Henderson. 1975, Molecular structure determination by electron microscopy of unstained crystalline specimens. Journal of Molecular Biology 94:425–440. Uzgiris, E.E. and R.D. Kornberg. 1983, 2-Dimensional crystallization technique for imaging macro molecules with application to antigen antibody complement complexes. Nature 301:125–129. Uzgiris, E.E. 1987, Self-organization of IgE immunoglobulins on phospholipid films. Biochemical Journal 242:293–296. Uzgiris, E.E. 1990, Antibody organization on lipid films. Biochemical Journal 272:45–49. van Heyningen, S. 1983, The interaction of cholera-toxin with gangliosides and the cell-membrane. Current Topics in Membranes and Transport 18:445–471. Vénien-Bryan, C., F. Balavoine, B. Toussaint, C. Mioskowski, E.A. Hewat, B. Helme and P.M. Vignais. 1997, Structural study of the response regulator HupR from Rhodobacter capsulatus. Electron microscopy of two-dimensional crystals on a nickel-chelating lipid. Journal of Molecular Biology 274:687–692. Vénien-Bryan, C., P.F. Lenne, C. Zakri, A. Renault, A. Brisson, J.F. Legrand and B. Berge. 1998, Characterization of the growth of 2D protein crystals on a lipid monolayer by ellipsometry and rigidity measurements coupled to electron microscopy. Biophysical Journal 74:2649–2657. Vénien-Bryan, C., G.F.X. Schertler, E. Thouvenin and C. Courty. 2000, Projection structure of a transcriptional receptor, HupR, determined by electron cryo-microscopy. Journal of Molecular Biology 296:863–871. Verclas, S., P. Howes, K. Kjaer, A. Wurlitzer, M. Weygand, G. Büldt, N. Dencher and M. Lösche. 1999, X-ray diffraction from a single layer of purple membrane at the air/water interface. Journal of Molecular Biology 287:837–843. Vogel, V., W.R. Schief and W. Frey. 1997, Dynamics of two-dimensional protein crystallization at the air/water interface: streptavidin targetted to
Bibliography 135
surfaces via high-affinity binding or metal coordination. Supramolecular Science 4:163–171. Voges, D., R. Berendes, A. Burger, P. Demange, W. Baumeister and R. Huber. 1994, 3-Dimensional structure of membrane-bound annexin-V — a correlative electron microscopy-X-ray crystallography study. Journal of Molecular Biology 238:199–213. Wang, D.N. and W. Kühlbrandt. 1991, High-resolution electron crystallography of light-harvesting chlorophyll a/b-protein complex in three different media. Journal of Molecular Biology 217:691–699. Wang, H.W. and S.F. Sui. 1999, Pentameric two-dimensional crystallization of rabbit C-reactive protein on lipid monolayers. Journal of Structural Biology 127:283–286. Wang, H.W., Y.J. Lu, L.J. Li, S.S. Liu, D.N. Wang and S.F. Sui. 2000, Trimeric ring-like structure of ArsA ATPase. FEBS Letters 469: 105–110. Wang, H.W. and S.F. Sui. 2001a, Dissociation and subunit rearrangement of membrane-bound human C-reactive proteins. Biochemical and Biophysical Research Communications 288:75–79. Wang, H.W. and S.F. Sui. 2001b, Two-dimensional assembly of pentameric rabbit C-reactive proteins on lipid monolayers. Journal of Structural Biology 134:46–55. Wang, S.W., C.R. Robertson and A.P. Gast. 1999, Molecular arrangement in two-dimensional streptavidin crystals. Langmuir 15:1541–1548. Ward, R.J. and K. Leonard. 1992, The Staphylococcus aureus alpha-toxin channel complex and the effect of Ca2⫹ ions on its interaction with lipid layers. Journal of Structural Biology 109:129–141. Weber, P.C., D.H. Ohlendorf, J.J. Wendoloski and F.R. Salemme. 1989, Structural origins of high-affinity biotin binding to streptavidin. Science 243:85–88. Weis, R.M., K. Balakrishnan, B.A. Smith and H.M. McConnell. 1982, Stimulation of fluorescence in a small contact region between rat basophil leukemia cells and planar lipid membrane targets by coherent evanescent radiation. Journal of Biological Chemistry 257:6640–6645. Weygand, M., B. Wetzer, D. Pum, U.B. Sleytr, N. Cuvillier, K. Kjaer, P.B. Hower and M. Lösche. 1999, Bacterial S-layer protein coupling to lipids: X-ray reflectivity and grazing incidence diffraction studies. Biophysical Journal 76:458–468. Weygand, M., K. Kjaer, P. Howes, B. Werzer, D. Pum, U.B. Sleytr and M. Lösche. 2002, Structural reorganization of phospholipid
136 Strategies for Two-dimensional Crystallization of Proteins
headgroups upon recrystallization of an S-layer lattice. Journal of Physical Chemistry 106:5793–5799. Williams, K.A. 2000, Three-dimensional structure of the ion-coupled transport protein NhaA. Nature 403:112–115. Wilson-Kubalek, E.M., R.E. Brown, H. Celia and R.A. Milligan. 1998, Lipid nanotubes as substrates for helical crystallization of macromolecules. Proceedings of the National Academy of Sciences of the United States of America 95:8040–8045. Wilson-Kubalek, E.M. 2000, Preparation of functionalized lipid tubules for electron crystallography of macromolecules, in Sphingolipid Metabolism and Cell Signaling, Pt B, 312: 515–519. Series: Methods in Enzymology. Wu, Y., H.W. Wang, S.R. Ji and S. Sui. 2003, Two-dimensional crystallization of rabbit C-reactive protein monomeric subunits. Acta Crystallographica Section D-Biological Crystallography 59:922–926. Yatcilla, M.T., C.R. Robertson and A.P. Gast. 1998, Influence of pH on two-dimensional streptavidin crystals. Langmuir 14:497–503. Zakri, C. 1997. Etude de la fusion-cristallisation de monocouches de l-alcools a la surface de l’eau: mesures d’élasticité latérale par diffraction de rayons X et par une méthode mécanique. Ph.D. Thesis. University of Grenoble I, France. Zhuang, J.P., G.G. Prive, G.E. Werner, P. Ringler, H.R. Kaback and A. Engel. 1999, Two-dimensional crystallization of Escherichia coli lactose permease. Journal of Structural Biology 125:63–75. Zuber, G. and E. Barklis. 2000, Atomic force microscopy and electron microscopy analysis of retrovirus gag proteins assembled in vitro on lipid bilayers. Biophysical Journal 78:373–384.
Index
50S ribosomal subunit 7 50S ribosome 9 ␣-toxin 8 Atomic force microscopy 5, 72, 73, 74, 77 Annexin V 7, 56, 60, 70, 74, 77, 78, 79, 97 Antibody 1, 15, 20, 29, 32 Avidin 14, 20, 96 Bacteriorhodopsin 35, 36, 70, 85 Brewster angle microscopy 5, 14, 30, 45, 46, 56, 62, 63, 64, 65, 77, 78 BioBeads® 37, 43, 57, 66, 93 Biotin 11, 12, 13, 15, 20, 28, 30, 60, 61, 64, 96 Brewster angle microscope 63 Brush border myosin-I 8 Carbon film 76, 77, 79, 80, 81, 88, 90 Catalase 9, 10 Chaperonin TF55 9 Charged lipids 2, 5, 6, 7, 8, 9, 97 Cholera toxin 4, 10, 11, 60, 77, 79
Critical micellar concentration (CMC) 36, 38, 40, 41 Conformational changes 8, 35, 72, 74 Controlled dilution 36 Crystal morphology 15, 38 Crystallization techniques 2, 4, 34, 96 Cholera toxin B 60, 70, 71, 77, 78, 79 Dehydroabietylamine 9 Detergent 23, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 56, 66, 90, 92, 93 Detergent removal 36, 37, 38, 40, 44 Dehydroabietylamine (DHAA) 9 Dialysis 36, 37 Dialysis devices 37 Dialysis buttons 37 Dialysis slides 37 Dialysis tubing 37 Diluting lipids 3, 4, 23, 41, 51, 92, 93 DNA scaffolds 95, 96 Domain movements 74 Drugs 34 137
138 Index
Electron crystallography 1, 9, 34, 35, 75, 88 Electron microscope 11, 22, 26, 43, 57, 58, 62, 68, 71, 75, 76, 77, 78, 80, 81, 83, 84, 85, 86, 90, 93 Electron microscopy 5, 13, 14, 38, 43, 56, 57, 60, 65, 71, 72, 73, 74, 75, 76, 78, 81, 82, 84, 85, 86, 89, 90, 97 Electrostatic interaction 2, 5, 6, 7, 8, 10, 41, 48, 72, 89 Ellipsometry 5, 45, 53, 54, 55, 56, 57, 60 Epifluorescence 5, 12, 14, 31 F0F1-ATP-synthase 27 Factor VIII 8 Ferritin 1, 9, 10, 13 FhuA 27 Film balance 5, 48 Fluidity 3, 4, 10, 23, 40, 41, 46, 47, 48, 49, 51, 52, 53, 89, 92 Fluorescence microscopy 5, 30, 31, 45, 46, 64, 76 Fluorescence spectroscopy 5, 31 Freeze-thaw cycles 38 Fumarate reductase 35 Functionalized lipids 4, 10, 11, 12, 17, 25, 28, 29, 39, 40, 92 Gag-protein 73 Gap junctions 74 Glutaraldehyde 79, 80 Helical reconstruction 38, 97 His-tag 2, 19, 20, 21, 22, 23, 24, 25, 26, 31, 41, 42, 43, 44, 51, 70, 92, 97 HIV-1 reverse transcriptase 21, 24 Hockey sticks 37
Holey carbon film 76, 79, 80 Holliday junction 95 HupR 25, 56, 64, 65, 68, 70, 78, 80, 97 Hydrophobic effect 3, 38, 47 Image processing 86 in situ crystallization 35 Inclusion bodies 35 Light harvesting complex 36, 86 Light scattering 5, 23, 36 Lipid bilayer 23, 33, 35, 36, 38, 40, 43, 57, 74, 93 Lipid to protein ratio 37 Major histocompatibility factor (MHC) 26 Membrane proteins 1, 23, 24, 26, 28, 30, 33, 34, 35, 36, 38, 39, 41, 42, 43, 44, 56, 57, 66, 74, 88, 92, 93, 96 Metal chelating lipids 11, 24 Micelle 35 Missing cone problem 9, 25, 96 Moloney murine leukaemia virus capsid protein 24 Monosialogangliodide (GM1) 11, 60, 70 Morphology 14, 15, 34, 38, 39 Nanotubes 96, 97 Natural lipid 10 Neutron reflectivity 5, 68 Ni2⫹-NTA group 2, 21, 41, 96 NMR 34, 88 Non-specific binding 2, 21 Non-specific interaction 4 Over-expression 34, 35
Index 139
Phase contrast 5, 17, 84, 87 Phase contrast microscopy 5, 17 Phase transition 38, 46, 47, 49 Phospholipase 35, 38 Photosynthetic core complex 74 Porins 34 Protein refolding 35 Proton-ATPase 22, 28 Radiation damage 75, 84 Reconstitution 35, 36, 37, 38, 40, 43, 57, 93 RNA polymerase 7, 8, 13, 21, 22, 25, 79, 97 RuvA 95 Scanning near field optical microscopy 5 Shear modulus measurements 5 Single particle analysis 26, 88, 93 Specific binding 2, 13, 19, 29, 77 Streptavidin 10, 12, 13, 14, 15, 23, 28, 30, 57, 61, 63, 64, 70, 71, 74, 76, 78, 79, 80, 97 Structure determination 1, 7, 9, 34, 92, 93, 96, 97 Surface crystallization 2, 4, 7, 39, 40, 41, 42, 56 Synthetic lipids 11, 16, 29, 30, 31, 36, 38
Temperature 3, 10, 35, 37, 38, 46, 47, 49, 51, 82, 84, 85, 87, 89, 92 Three-dimensional crystals 4, 70, 72, 77 Tubular 28, 38, 39 Tubular crystals 39 Two-dimensional crystals 1, 3, 4, 5, 6, 7, 9, 10, 11, 12, 13, 14, 15, 16, 17, 21, 22, 23, 24, 26, 27, 34, 35, 36, 38, 40, 43, 48, 51, 56, 57, 60, 62, 64, 65, 68, 70, 71, 72, 74, 75, 76, 77, 78, 79, 80, 85, 86, 87, 89, 93 Two-dimensional crystallization 1, 4, 10, 12, 14, 15, 22, 23, 24, 25, 29, 30, 35, 36, 37, 38, 44, 46, 47, 48, 49, 50, 51, 52, 53, 56, 66, 89 Uranyl acetate 26, 81 Vesicle 3, 7, 8, 17, 36, 38, 39, 74 Vitrification 81, 82 Water-soluble proteins 1, 34 X-ray crystallography 34, 88 X-ray grazing incidence 5, 60, 66, 68, 77, 78 X-ray grazing incidence diffraction 5, 60, 66 X-ray reflectivity 5, 45, 46, 66, 67, 68