Chemistry and Physics of Lipids 141 (2006) 2–29 Review Modelling of proteins in membranes Maria Maddalena Sperotto a , Sylvio May b , Artur Baumgaertner c,∗ a Biocentrum, The Technical University of Denmark, Lyngby, Denmark Department of Physics, North Dakota State University, Fargo, USA Department of Solid State Research, Research Centre Jülich, Germany b c Received 20 December 2005; accepted 20 February 2006 Available online 27 March 2006 Abstract This review describes some recent theories and simulations of mesoscopic and microscopic models of lipid membranes with embedded or attached proteins. We summarize results supporting our understanding of phenomena for which the activities of proteins in membranes are expected to be significantly affected by the lipid environment. Theoretical predictions are pointed out, and compared to experimental findings, if available. Among others, the following phenomena are discussed: interactions of interfacially adsorbed peptides, pore-forming amphipathic peptides, adsorption of charged proteins onto oppositely charged lipid membranes, lipid-induced tilting of proteins embedded in lipid bilayers, protein-induced bilayer deformations, protein insertion and assembly, and lipid-controlled functioning of membrane proteins. © 2006 Elsevier Ireland Ltd. All rights reserved. Keywords: Dissipative particle dynamics; Coarse-grain model; Mesoscopic model; Molecular dynamics; Monte Carlo; Hydrophobic mismatch; Tilting; Phase transition; Cooperative behavior; Protein insertion; Ion channel; Poisson–Boltzmann; Lipid–protein interaction Contents 1. 2. 3. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Statistical mean-field theories . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1. Chain-packing theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2. Membrane elasticity theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.3. Poisson–Boltzmann theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Mesoscopic modelling and DPD simulations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1. Mesoscopic models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2. Dissipative particle dynamics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3. Protein-induced bilayer perturbations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.4. Lipid-induced protein tilting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 6 6 7 8 10 10 12 12 15 Abbreviations: DPD, dissipative particle dynamics; MC, Monte Carlo; MD, molecular dynamics; CG, coarse grain (or mesoscopic); DMPC, dimyristoylphosphatidylcholine; DPPC, dipalmitoylphosphatidylcholine; POPC, palmitoyloleoylphosphatidylcholine; POPE, palmitoyloleoylphosphatidylethanolamine; PB, Poisson–Boltzmann; TM, transmembrane ∗ Corresponding author. Tel.: +49 2461 614074; fax: +49 2461 612893. E-mail address: [email protected] (A. Baumgaertner). 0009-3084/$ – see front matter © 2006 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.chemphyslip.2006.02.024 M.M. Sperotto et al. / Chemistry and Physics of Lipids 141 (2006) 2–29 4. 5. Simulations of membrane processes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.1. Protein translocation into membranes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2. Protein assemblies in membranes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2.1. Protein assemblies in membranes: single peptides . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2.2. Protein assemblies in membranes: interacting peptides . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.3. Lipid-controlled function of membrane proteins . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Concluding remark . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1. Introduction Biological membranes are heterogeneous, highly cooperative and very complex systems. Because of the intricate interplay between biomembrane functioning, membrane chemistry, and physical properties of the lipid bilayer (Sackmann, 1995), to relate the physical properties of biomembranes to their biological function (Sackmann, 1995) – the ultimate goal of biomembrane science – it is not only common but usually necessary to study model (or reconstituted) membranes: lipid bilayers composed of few lipid species with embedded proteins, or natural or artificial peptides, or other biologically relevant molecules. Therefore reconstituted membranes have become the subject of an enormous number of interdisciplinary experimental, as well as theoretical, investigations, which have and are helping to understand membrane organization and biofunctioning. Interestingly, experimental findings have initially motivated the development of lipid–protein interaction models, which have then, in turn, started to help designing useful experiments. This review focuses on some selected lipid–protein modelling approaches that have been used to investigate phenomena occurring in biomembrane mimetic system, such as the formation of domains, lipid-induced structural changes (and activities) of membrane proteins, protein-induced changes in lateral membrane organization, the role of charges for the binding of proteins to membranes, or the insertion of proteins into membranes. Theoretical studies have also been used to proof conceptual hypothesis which in the recent years have constituted a matter of debate among scientists. One example is the role played for some biomembrane phenomena by the hydrophobia matching (between the lipidbilayer hydrophobic thickness and the protein hydrophobic length). There are roughly three different classes of modelling approaches, each with their own advantages and limitations: statistical theories, mesoscopic models, and all-atom models. The three main sections of this review focus each on one of these classes. This 3 17 17 19 19 21 22 24 24 25 review focus on the results obtained by theoretical studies (and their possible interplay with experimental investigations), rather than with the theoretical or computational methods used to study the models. Details of the methodologies can be found in the references cited in the various sections of the review. During the last decades, it has become evident that the behavior of biomembranes is governed by some basic physical principles, which manifest themselves in some detectable properties. Among them, membrane elasticity seems to play a role in the budding, fusion, fission, and pore formation processes. Also, the shape of the lipid molecules is important for determining the stability of the three-dimensional structure of the lipid aggregates. A hypothesis that has early been proposed is that of hydrophobic matching (Mouritsen and Blom, 1984; Sackmann, 1984; Mouritsen and Sperotto, 1993; Gil et al., 1998; Killian, 1998; Dumas et al., 1999) and related theories (see references in Abney and Owicki, 1985). According to this hypothesis, hydrophobic mismatch may affect membrane organization and biological functions: it is now known that hydrophobic matching is involved, among others, in the secretory pathway in the Golgi (Munro, 1995, 1998; Bretscher and Munro, 1993; Pelham and Munro, 1993). Hydrophobic matching also seems to play a role in sequestering proteins with long transmembrane regions (McIntosh et al., 2003) into sphingolipids–cholesterol biomembrane domains denoted as ‘rafts’ (Simons and Ikonen, 1997; Binder et al., 2003; Mukherijee and Maxfield, 2004), which are known to be involved in numerous diseases (Fantini et al., 2002). The ways that biological membranes may use to compensate for hydrophobic mismatch (de Planque and Killian, 2003) imply changes of the membrane structure and dynamics on a microscopic as well as on a macroscopic scale (Killian, 1992; Epand, 1998; Mouritsen, 1998; Gil et al., 1998; Dumas et al., 1997; Morein et al., 2002; Fahsel et al., 2002; Mall et al., 2001; Fernandes et al., 2003; Harroun et al., 1999a,b)—and may therefore affect biological functions (Montecucco et al., 1982; Johansson et al., 1981; 4 M.M. Sperotto et al. / Chemistry and Physics of Lipids 141 (2006) 2–29 In’t Veld et al., 1991; Lee, 1998, 2003). Regarding the mismatch-induced membrane changes happening on the microscopic scale, there are a number of experimental evidences indicating that, on the one end, proteininduced bilayer deformations may arise in the vicinity of the protein–lipid interface (Jost et al., 1973; Hesketh et al., 1976; Jost and Hayes Griffith, 1980; Rehorek et al., 1985; Piknová et al., 1993; Harroun et al., 1999a,b; Bryl and Yoshihara, 2001). Also, the protein may prefer, on a statistical basis, to be associated with the type of lipids that best matches its hydrophobic surface (Dumas et al., 1997; Lehtonen and Kinnunen, 1997; Fahsel et al., 2002; Fernandes et al., 2003). Phenomena like the one described above may induce the formation of bilayer domains (Binder et al., 2003), the possible precursors of the rafts, whose functional properties differ from those of the bulk, i.e. the unperturbed bilayer (Tocanne, 1992; Tocanne et al., 1994; Thomson et al., 1995). On the other end, there may be mismatch effects which are lipidmediated, such as the tilting (or even bending) of a whole protein/peptide to adapt to a too-thin bilayer (Glaubitz et al., 2000; Harzer and Bechinger, 2000; Killian, 1998; Sharpe et al., 2002; van der Wei et al., 2002; Koehorst et al., 2004; Strandberg et al., 2004; Özdirekcan et al., 2005; Ramakrishnan et al., 2005), or even tilting of the individual helices which form a protein; there is indeed some experimental evidence that the latter phenomenon may occur in channel proteins (Lee, 2003), and that a change of the tilt angle of the individual helices could be the cause of a change in protein activity. The issues about the protein-induced bilayers deformations and the lipidinduced proteins tilting are discussed below in Sections 2.2, 3.3 and 3.4. Because of the many degrees of freedom involved, biomembrane processes occur over a wide range of time and length scales (König and Sackmann, 1996). There are fast processes such as the trans-gauche isomerization time of the lipid chains (10−10 s), or the time of a lipid to diffuse within the membrane a distance that corresponds to its own size (10−8 s). Among the slow processes are the flip-flop motion of a single lipid from one leaflet of a bilayer to the other (103 s), or cooperative phenomena such as a lateral phase separation of two immiscible lipid species. Similarly for length scales, a process might depend only on the local membrane environment, as is the case when a lipid chain changes its conformation. On the other hand, cooperative phenomena typically involve correlated fluctuations over large length scales, such as in the fluid–gel phase transition. To model membranes it is necessary to decide, a priori, on the level of description of the system (i.e. to deliberately neglect those details unimportant to the process one wants to investigate). Often, this necessity follows the fact that some theoretical methods are limited in their applicability by the long computational times needed to calculate statistical quantities. Generally, the level of description needs to match the time and length scale of the process under consideration. That is, processes occurring on short time scales can be integrated out whereas all slow processes can be assumed to be infinitely slow. In fact, these general considerations provide the basis for the application of equilibrium-thermodynamics methods. In the past, mean-field theories, thermodynamical models, phenomenological and statistical microscopic models have been developed, often in conjunction with experimental investigations to study macroscopic, as well as mesoscopic, behavior of lipid–protein model systems (Gil et al., 1998). For example, lattice models have been useful to investigate the cooperative behavior of lipid–protein mixtures, and how the presence of protein-like impurities affects the phase behavior of the lipid-bilayer system (Sperotto and Mouritsen, 1991; Sperotto, 1997; Dumas et al., 1997; Morein et al., 2002). However, lattice models have limitations as the discrete lattice structure, its connectivity and topology does not adequately reflect the fluid-like and self-assembled nature of a lipid membrane. Still, these models turn out to be useful in connection with their ability to predict mechanisms for membrane domain formation (Hinderliter et al., 2001). One advantage of lattice models is, that they can be analyzed on a mean-field level (May, 2000), allowing to obtain simple expressions for the system’s free energies that can be combined with other theoretical approaches. With regard to modelling lipid–protein interaction there are a number of other phenomenological approaches ranging from elasticity to Poisson–Boltzmann theory, some of which will be discussed in Section 2. All these methods involve parameters that reflect the energetics of the system on a microscopic scale. They appear, for example, as elastic moduli or as interaction strength in a MC simulation. Clearly, the determination of these parameters is outside the scope of phenomenological approaches but requires alternatives such as experimental determination, microscopic-level methods, or atomistic simulations. Despite the development of powerful experimental techniques such as X-ray crystallography, electron microscopy, variants of nuclear magnetic resonance (NMR), fluorescence spectroscopy, electron-spinresonance (ESR), and others, the characterization of lipid–protein systems on the molecular level remains difficult because of their complexity. At the present time, even qualitative information gained by perform- M.M. Sperotto et al. / Chemistry and Physics of Lipids 141 (2006) 2–29 ing detailed computer simulations of protein–membrane complexes can be valuable, because only scarce information is available from experiments about the structure and dynamics of these systems. Therefore, the investigation of proteins solvated in water and in lipids employing molecular dynamics (MD) methods is becoming increasingly important. The length and time scales covered by MD simulations have reached considerably beyond the nanosecond and nanometer ranges where important structural and dynamical phenomena are expected. In particular, in Section 4.1 we discuss MD studies which reveal molecular details of the insertion mechanism of peptides into membranes. The individual and collective behavior of membrane peptides solvated in an explicit environment of fully hydrated membrane models are discussed in Section 4.2. Among others, the characteristics of single melittin, alamethicin, dynorphin, and SP-C are presented, as well as some results on the oligomerization of transmembrane ␣-helices. Very recently a few MD studies have addressed the significance of lipids for the control and active regulation of the function of large membrane proteins: the lipid-mediated gating of mechanosensitive ion channels. This is summarized in Section 4.3. Beside pure atomistic models used for simulations, there are notable also hybrid- and multi-scale approaches (Chang et al., 2005). For example, Biggin and Sansom (2003) used atomistic MD simulations to model the behavior of a protein channel in the membrane whereas the lipid bilayer was described by a mean-field potential. Because of the extremely long time required for the simulations, the information that can be obtained by all-atom simulations are limited to phenomena that occur at the nanoscopic level and on a nanosecond time-scale (up to 100 nm). Therefore, hybrid- and multi-scale approaches which use implicit models of lipids and/or water are in some cases unavoidable to study biologically relevant membrane processes which occur much beyond the nanosecond time range (König and Sackmann, 1996), for example in-plane phase transitions, phase separations, membrane fusion, or the self-assembly of peptides in a membrane. To bridge the gap between the informations that can be obtained by using phenomenological modelling and those from all-atom modelling, a number of solvent free membrane models (Goetz and Lipowsky, 1998; Cooke et al., 2005; Brannigan et al., 2005) and coarse-grain (or mesoscopic) models have been developed, which have been studied by MD simulations (Shelley et al., 2001a,b; van der Eerden et al., 2002; Nielsen et al., 2004; Nielsen et al., 2005a,b), and off-lattice MC simulations (Sintes and Baumgärtner, 1997; Gompper and 5 Kroll, 1997). The mesoscopic approach is based on the idea of modelling the system by an ensemble of effective particles, or ‘beads’. Each particle represents a lump of atoms (or even lipids) whose atomistic details are not relevant to the process under consideration. The internal degrees of freedom of each ‘bead’ are integrated out, contributing merely to a set of interaction parameters that defines bead–bead interactions. The main advantage of this approach is that it allows to access longer time and length scales compared to what is permitted by MD simulations all-atom models. The obvious limitation is the loss of atomistic information. There are biomembrane processes that ultimately have to be studied using both approaches, the mesoscopic and the atomistic. For example, the fusion event between two apposed membranes involves a high activation energy and thus is a rare event (happening at times longer that milliseconds) that becomes only accessible using CG methods. However, once it happens, its progression depends on molecular details (specified by the fusion protein machinery). To capture both levels is one of the major future challenges. Despite the advantages that arise by minimal modelling in connection with simulation methods like MD and MC, the possibility to study processes that involve the cooperative nature of biomembranes is still limited. To try to overcome this limitation, the use of a faster simulation technique, dissipative particle dynamics (DPD), on CG models has thus been considered. In fact, the DPD simulations allow for a timestep that is at least three orders of magnitude longer than what is used in atomistic MD simulations, which is typically of the order of a few femtoseconds (Groot, 2000; Groot and Rabone, 2001). The DPD method was first used to study mesoscopic models for pure lipid-bilayer systems (Venturoli and Smit, 1999; Shillcock and Lipowsky, 2002), and then lipid bilayers containing impurities such as alcohols (Kranenburg and Smit, 2004; Kranenburg et al., 2004b). The results from the simulation studies demonstrated that with the DPD–CG approach one was able to reproduce structural and thermodynamic properties resulting from the cooperative behavior of the investigated system (Kranenburg et al., 2003a,b). Based on the mesoscopic model for pure phospholipid bilayer (Venturoli and Smit, 1999), Venturoli et al. (2005) have developed a coarse-grain model for proteins of different sizes and have then used the DPD simulation method to study the behavior of a mesoscopic model for lipid bilayers with embedded proteins. These authors have correlated in a systematic way the extent of the protein-induced bilayer perturbation and the lipid-induced protein tilting with the 6 M.M. Sperotto et al. / Chemistry and Physics of Lipids 141 (2006) 2–29 lipid–protein hydrophobic mismatch. Comparison with experimental data indicates that some details of the protein structures are not relevant in relation to how much proteins tilt in a mismatched lipid bilayer. Brief descriptions of the lipid–protein systems studied with CG models by the MD or MC simulation methods, of the CG + DPD modelling approach, and of the membrane issues investigated by this approach are found in Section 3. 2. Statistical mean-field theories 2.1. Chain-packing theory Membrane-inserted proteins perturb the host membrane. One reason for this perturbation is the rigidity of the protein facing the fluid-like lipid environment. The lipid chains close to the protein surface are no longer able to explore their full conformational space as they cannot penetrate into the protein’s interior. This implies a penalty in conformational entropy which rises the lipid–protein interaction energy. A fairly simple molecular-level approach, able to predict the corresponding energies, is the chain-packing theory developed by Ben-Shaul and coworkers (Ben-Shaul, 1995). This theory was originally applied to homogeneous lipid aggregates and has later been generalized to nonhomogeneous membranes, including membranes that contain rigid inclusions. The basis of the chain-packing theory forms the constraint of uniform segment density within the entire membrane’s hydrocarbon core. This constraint is approximatively fulfilled in any lipid aggregate. As discussed recently (Siegel, 1999), there are no regions within the membrane that are void of chain segments. On the other hand, X-ray scattering experiments and MD simulations both show that the assumption of a constant and homogeneous segment density is not strictly valid (Wiener and White, 1992; Tieleman et al., 1997). The generalized version of the chain-packing approach is concerned with two functional degrees of freedom. One is the density σ(r) of lipid headgroups on the aggregate interface A. The other one is the conditional probability P(α|r) to find a lipid chain in a given conformation α if the corresponding headgroup of this chain is located at position r ∈ A. The overall free energy of the inclusion-containing membrane F[σ(r), P(α|r)] is then written on a mean-field level and minimized with respect to σ(r) and P(α|r) under the constraint of uniform chain segment density everywhere (May and BenShaul, 2000). Fattal and Ben-Shaul (1993) have used the chain-packing approach to investigate the perturbation of a lipid bilayer induced by the presence of a single transmembrane protein. The protein was modelled as a straight wall, impenetrable to the hydrocarbon chains of the lipids and imposing a certain degree of hydrophobic mismatch between protein and membrane. While their study was mainly aimed to get insight into the energetics of hydrophobic mismatch (discussed below) there was another notable result: protein insertion into the host bilayer involves an energetic penalty that results from conformational restrictions of the perturbed lipid chains. It may be speculated that this penalty contributes to driving protein aggregation in membranes. To further pursue the mechanism, the interaction between two parallel, membrane-inserted rigid walls was calculated using the chain-packing approach (May and Ben-Shaul, 2000). The result indicated that the energy gain upon aggregation between the two walls is preceded by an energetic barrier. Such a barrier was also predicted by other theoretical approaches, including membrane elasticity theory (Dan et al., 1993) and MC simulations (Sintes and Baumgaertner, 1998). The barrier generally locates at distances between the inclusions that correspond to the size of at most a few lipid molecules. It is notable that a simple “toy” model (May and Ben-Shaul, 2000) recovers some of the predictions made by chain-packing theory, including the non-monotonic membrane-mediated interaction potential between two walls. This so-called director model approximates the conformational space of a lipid hydrocarbon chain by the different orientations of a single (unit) vector. Thereby it is assumed that all accessible orientations of this vector occur with the same probability. Not accessible are those orientations for which the vector leaves the hydrocarbon core of the membrane. The presence of rigid membrane inclusions, such as one or more rigid walls, further restricts the orientational space, rendering all inclusion-penetrating vector orientations inaccessible. Calculating the loss of orientational entropy imposed by a rigid wall (and similarly for any given inclusion shape) involves summation over the individual contributions of all affected chains. It turns out that the corresponding free energy measured per unit length of a membrane-inserted wall is predicted to be 0.24 kB T/Å, in reasonable agreement with the more detailed chain-packing calculations for which 0.37 kB T/Å were obtained for (CH2 )13 CH3 chains. More important, qualitative agreement is also observed for several other observables: for the wall-induced tilt of the lipid chains, for the non-monotonic interaction potential between two walls (May and Ben-Shaul, 2000), and for the calculation of the so-called tilt modulus (May et al., 2004), discussed further below. It is this agreement which makes the simple director model a candidate to use for more complex geometries (Kessel et al., 2001) M.M. Sperotto et al. / Chemistry and Physics of Lipids 141 (2006) 2–29 (those that are numerically difficult to access using the chain-packing theory) or to interpret the results of chainpacking calculations. Another recent application of the chain-packing approach concerns the interaction of interfacially adsorbed and pore-forming amphipathic peptides with a lipid membrane. Amphipathic peptides, such as mellitin, magainin or alamethicin, are typically alpha-helical with a hydrophobic and a polar (typically cationic) face. They often possess antimicrobial activity which makes them naturally occurring antibiotics (Tossi et al., 2000). That is, upon interaction with a lipid membrane they are able to exert lytic activity. Among the suggested mechanisms is the formation of pores (Epand et al., 1995). In this case, the peptides are initially adsorbed – driven mostly by electrostatic and hydrophobic interaction – to the membrane in interfacial orientation, inserting their hydrophobic face into the host bilayer. Peptide aggregation and self-assembly then can lead to an orientational change upon which the peptides rearrange into a membrane-inserted, pore-forming structure (Huang, 2000). Yet, in order for the peptides to self-assemble there must be an energetic driving force that compensates for the loss of their in-plane translational and orientational entropy. A recent model study (Zemel et al., 2005) offers an interesting and new explanation for the origin of this driving force. In this study, the mean-field chainpacking approach was applied to a membrane-adsorbed cylinder-like particle and a membrane-spanning wall as models for amphipathic peptides in their isolated and pore-forming states, respectively. The calculations showed that the perturbation of the hydrocarbon chain region induced by an isolated cylinder-like particle is quite substantial and results from the need of the lipids to fill out and pack with uniform segment density the region just underneath the cylinder. This need is predicted to result in local membrane thinning and in a decrease of the average segmental order parameter of the lipid hydrocarbon chains (Zemel et al., 2004), both being in agreement with experimental data (Ludtke et al., 1995; Koenig et al., 1999). Further analysis shows that the lipid chains in the two apposed lipid monolayers are perturbed differently: those in the cylinder-containing monolayer curl on average toward the region underneath the adsorbed cylinder whereas those in the opposite monolayer stretch toward the cylinder. The behavior for the membrane-adsorbed cylinder is in notable contrast to a simple membranespanning wall for which the conformational confinement of the lipid chains (discussed above in terms of the director model) provides the main contribution to the membrane perturbation free energy. As a consequence, the membrane-spanning wall is energetically preferred over 7 the adsorbed cylinder; the chain-packing calculations predict about 4–5 kB T per peptide which would provide the energy to account for experimentally observed peptide self-assembly (and pore formation). It should be mentioned that various interactions not accounted for by the chain-packing approach are likely to modify the peptide’s propensity to self-assemble. Most notable among these are interactions between acidic lipid headgroups and basic peptide residues (Zemel et al., 2003). Still, the perturbation of the membrane chain region would provide a plausible non-specific driving force that complements with alternative mechanisms (Huang et al., 2004) and could underlie the biological function of amphipathic pore-forming peptides. 2.2. Membrane elasticity theory Transmembrane proteins or peptides are rigid bodies when compared to the fluid-like host membrane. It is therefore common to model these molecules as rigid inclusions that reside in the membrane. An important class of transmembrane proteins/peptides can be represented as inclusions with up-down symmetry. In this case, the mid-surface of the inclusion-containing membrane (that is the surface that divides between the two membrane leaflets) remains planar. At the same time, the thickness of the membrane-inserted hydrophobic part of an inclusion need generally not match that of the host membrane, a case which is referred to as hydrophobic mismatch (Mouritsen and Blom, 1984). In fact, hydrophobic mismatch can be positive or negative depending on whether the thickness of the inclusion’s hydrophobic part is larger or smaller than that of the host membrane. It is notable that the thickness mismatch between transmembrane proteins (or peptides) and lipid membranes can be altered experimentally (say, by using lipids of different chain length) and can thus be studied systematically. Among the implications of hydrophobic mismatch are protein aggregation and conformational changes, lipid sorting, peptide tilt, and structural phase transitions of membranes (Killian, 1998). All these are the consequence of an unfavorably high energy penalty associated with a too large hydrophobic mismatch. From a theoretical perspective it is desirable to estimate the corresponding free energy cost: one convenient and frequently used method is the application of membrane elasticity theory. Here, a number of different deformation modes such as splay, saddle splay, compression, and tilt contribute to the free energy of an inclusioncontaining membrane. The former two modes are well known as they are energetically equivalent to a curvature deformation of a lipid layer, described by the bending 8 M.M. Sperotto et al. / Chemistry and Physics of Lipids 141 (2006) 2–29 rigidity κ and the Gaussian modulus κ̄ (Helfrich, 1973). The compression mode is related (through the volume conservation constraint of the membrane) to a lateral extension. Here too, the corresponding elastic modulus – the lateral area compressibility modulus K – is well known from experiments. The remaining deformation mode is associated with tilt of the lipid chains. We note that for a fluid-like lipid layer tilt represents the average over the (conveniently defined) tilt of many different chain conformations. Unlike for the other deformation modes, the corresponding modulus of a tilt deformation – the tilt modulus κt – is not known from experiment but has been estimated based on several different theoretical approaches including the above-mentioned mean-field chain-packing approach (May et al., 2004). To apply membrane elasticity theory, the energies associated with the different deformation modes are expressed in terms of conveniently defined functional order parameters such as the relative change in membrane thickness and/or the (average) orientation of the lipid chains. Minimization of the overall free energy with respect to the order parameters results in differential equations whose boundary conditions are determined by the inclusion geometry. The membrane perturbation induced by a single symmetric inclusion exhibits a typical damped oscillating behavior. The wave-length ξ C of the oscillating part and the characteristic length ξ P of the exponential decay are given by −1/2 dL0 1 0 −1/2 ξC/P = dL K ∓ (1) κt (κK)1/2 where dL0 is the equilibrium thickness of the membrane (May, 2002). Typically, lipid membranes are characterized by both a large stretching and tilt modulus and a comparatively small bending stiffness. This leads to both ξ P and ξ C being small, on the order of 1 nm, indicating that the elastic perturbation of the membrane decays fast – within a few lipids – to its equilibrium value dL0 . Even though this raises concerns about the appropriateness of elasticity theory in the first place, this approach has frequently been applied to interpret experimental results (Harroun et al., 1999a,b; Lundbæk and Andersen, 1999). Moreover, as will be discussed in Section 3.3, the typical overshooting effect (Dan et al., 1993) is also observed in computer simulations such as DPD (Venturoli et al., 2005) or self-consistent field theory (Kik et al., 2005) that take into account the discrete nature of the lipids. (In Fig. 8, we shall compare the predictions of membrane elasticity theory with results from a DPD simulation.) The general reason for the non- monotonic membrane relaxation lies in the competition between the stretching and bending modes of deformation. This has an interesting implication concerning the membrane-mediated elastic interaction between two (or more) inclusions which, too, is non-monotonic. That is, there is an energy barrier that separates an attractive from a repulsive region, a prediction similar to that derived within the chain-packing approach and within the director model as discussed above. Under the assumption that the elastic energy of the membrane and the inclusion-induced conformational restrictions are the main energetic contributions to lipid–protein interaction (that is, neglecting all specific and particularly all lipid headgroup–protein interactions), an approach has recently been made (May, 2002; Bohinc et al., 2003) to combine elasticity theory with the director model. Among the predictions is that even a matching transmembrane inclusion induces a small but notable membrane thickening and that the optimal “mismatch” is a negative one. Similar predictions are made by self-consistent field theory (Kik et al., 2005) but still await a systematic experimental inspection. 2.3. Poisson–Boltzmann theory Charged water soluble proteins (and other macroions such as DNA) interact with oppositely charged lipid membranes predominantly through electrostatic interactions. The corresponding energetics can be described on the mean-field level by Poisson–Boltzmann (PB) theory. The crucial quantities in this theory are the local concentrations n+ and n− of the positively and negatively charged mobile ions, respectively, near the protein and membrane. On the mean-field level, they are given by the Boltzmann distributions n± = n0 exp(∓Ψ ) where n0 is the corresponding bulk concentration and Ψ is the (reduced) electrostatic potential. Combination with Poisson’s law of electrostatics leads to the PB equation 2 Ψ = sinh Ψ where denotes the Laplacian and l lD D the Debye screening length. We note that in the presence of di- and higher valent mobile ions, the PB approach becomes increasingly inappropriate as ionic correlations start affecting (and eventually dominate) the free energy (Grosberg et al., 2002). Hence, most applications of PB theory are restricted to deal with monovalent ions. Yet, within this restriction the PB approach is powerful as it can be used to derive membrane–protein interaction energies on an atomistic level (Wang et al., 2004). When well separated, both the membrane and the protein individually immobilize corresponding counterions by forming a diffuse double layer. This layer reflects a compromise between the loss in translational entropy of M.M. Sperotto et al. / Chemistry and Physics of Lipids 141 (2006) 2–29 9 Fig. 1. Schematic illustration of a spherical protein adsorbed onto a two-component membrane. In the fluid-like state, the lipids are mobile in lateral direction, implying the possibility of local compositional changes. The diagram shows predictions of PB theory for the adsorption of a single sphere (of radius R = 10 Å with uniform charge density corresponding to seven positive charges) onto a mixed membrane that contains 20% negatively charged lipids as a function of the scaled protein–membrane distance h/lD where lD = 10 Å is the Debye length (reproduced from May et al. (2000), with permission). The three adsorption free energies (in units of kB T) correspond to fixed surface charge density (Fφ ), mobile lipids (F), and constant membrane surface potential (FΨ ). The inset shows the local membrane composition, η, for the three cases at h/lD = 0.3. the counterions and electrostatic energy. Upon adsorption of the protein onto the membrane, some of the previously immobile counterions can be released into the aqueous solution. As is well known, the release of counterions generally constitutes the driving force for the association of oppositely charged macroions. Yet, for the adsorption of proteins onto lipid membranes a peculiarity arises in the case of a mixed lipid membrane: mixed lipid membranes, consisting of a charged and an uncharged lipid species, are two-dimensional fluids, able to adjust their local composition (and thus, charge density). Hence, upon adsorption, proteins should in principle be able to sequester oppositely charged lipids as is schematically illustrated in Fig. 1. This is a physically interesting situation because the release of counterions is driven by an energetically similar process, namely the immobilization of charged lipids in the vicinity of the protein. In other words, the diffuse ionic layer is replaced by a diffuse in-plane layer of charged lipids. The additional demixing degree of freedom can be taken into account within Poisson–Boltzmann theory by a special boundary condition that was first derived in a study to model cationic DNA–lipid complexes (Harries et al., 1998). The boundary condition accounts for the possibility of lipid migration within the membrane but assigns an (ideal) demixing free energy penalty to variations in the local composition. As a result, the boundary condition describes a case intermediate between the two thermodynamic limits: fixed surface charge density (that is, suppressed demixing) and constant electrostatic potential at the membrane surface. Indeed, it was shown that also the adsorption free energy of a spherical model protein is intermediate between that of the two thermodynamic limits (May et al., 2000). Lipid sequestration could be a biologically important phenomenon as is suggested for the enrichment of phosphatidylinositol 4,5-bisphosphate (PIP2 ) induced by myristoylated alanine-rich C kinase substrate (Gambhir et al., 2004). Yet, we also note that – in contrast to the prediction of PB theory – there is currently no experimental indication that charged peptides are also able to sequester monovalently charged lipids (Golebiewska et al., 2005). Generally, two-component lipid membranes exhibit non-ideal mixing behavior, often characterized by an effective attraction between lipids of the same species (Garidel and Blume, 2000). If this attraction is sufficiently large it will drive lateral phase separation between the two membrane components. The electrostatic repulsion between the like-charged lipid headgroups is expected (at least on the mean-field level) to stabilize the mixed membrane. That is even, if a hypothetically uncharged membrane would phase separate, the charged one need not. Within a combination of a mean-field lattice gas description and PB theory, the influence of the electrostatic interactions on the stability of the membrane can be calculated (Gelbart and Bruinsma, 1997; May et al., 2002). It is then interesting to speculate that the adsorption of oppositely charged proteins tends to effectively neutralize the membrane and could thus re-introduce an instability of the membrane. However, a recently proposed two-state model suggests a different scenario: protein-adsorption generates compositional gradients within the membrane plane that are associated with an unfavorable energy due to the non-ideal demixing properties. This energy acts as a 10 M.M. Sperotto et al. / Chemistry and Physics of Lipids 141 (2006) 2–29 line tension around the circumference of each proteinadsorption region, giving rise to an effective membranemediated protein–protein attraction that can render the membrane unstable (May et al., 2002). The emergence of a line tension induced by the adsorption of proteins and the subsequent membrane destabilization is also corroborated by calculations on the PB level beyond the two-state model. Mbamala et al. (2005) have recently calculated the stability of mixed membranes decorated by (cylindrically symmetric) model proteins of various disk-like and sphere-like shapes. Notably, due to the non-ideal lipid demixing, the boundary condition for the PB equation at the membrane surface becomes itself a differential equation. It was found that the protein’s ability to induce phase separation depends sensitively on its charge distribution and shape. The most potent candidate for inducing phase separation would be a large protein with a cluster of negative charges on a flat face in immediate vicinity to the lipid headgroups. No additional charges should be located at the side face of that protein as these would give rise to direct electrostatic protein–protein repulsion. Experimental evidence corroborates the notable influence of non-ideal lipid demixing on membrane domain formation. Hinderliter et al. (2001, 2004) have observed that small changes in the chemical structure of the lipids – for example by changing the chain length of the uncharged lipid component – affect the ability of certain proteins (such as the C2A and C2B domains of synaptotagmin) to induce membrane domain formation. A discussion of this ability in connection with cholesterolcontaining membranes was recently provided by Epand (2004). 3. Mesoscopic modelling and DPD simulations 3.1. Mesoscopic models Thanks to the development of mesoscopic models – also referred to as coarse-grain (CG) models – it has been possible to investigate a number of processes related to biomembrane physics, which would have been difficult to study by MD simulation methods on all-atom models: the self-assembly of phospholipids into various phases, both in the absence and in the presence of biologically relevant molecules such as anaesthetics, and alkanes (Shelley et al., 2001a,b); the lipid-mediated range of attraction between proteins embedded in a lipid bilayer (Sintes and Baumgärtner, 1997); the formation of the striped (or striated) phases which were detected in supported phospholipid bilayers with embedded synthetic ␣-helical peptides (van der Eerden et al., 2002). Very recently, Nielsen et al. (2005a,b) have carried out MD simulations on a mesoscopic model to analyze the lipid-bilayer perturbation around a transmembrane hydrophobic nanotube. The simulation results are in qualitative agreement with those obtained previously by MC simulations on a lattice model (Sperotto and Mouritsen, 1991), and all-atom MD simulations (Jensen et al., 2001; Jensen and Mouritsen, 2004). The same CG + MD approach has been used to study the process of insertion of a nanotube (simulating a protein channel or a pore) into a membrane, and how the insertion may depend on the presence of hydrophilic sites at the heads of a hydrophobic nanotube (Lopez et al., 2005). Nielsen et al. (2004) studied, by MD on a CG model, the lipid-sorting mechanism which occurs when a protein is embedded in a bilayer formed by a mixture of phospholipids having very different chain lengths. The simulation results confirms what was observed in the past by an investigation which combined fluorescence spectroscopy experiments with MC simulations on a lattice model (Dumas et al., 1997): the protein selects in its vicinity the lipid type which better matches its hydrophobic surface. Nielsen et al. (2004) suggest that the lipid-sorting mechanism can explain the onset of the fusion process; this occurs via the formation of a meniscus in the vicinity of the protein, which is then the triggering factor for the transition from the bilayer to a non-bilayer phase. The use of a relatively new simulation method on CG models, the dissipative particle dynamics (DPD) method, opens the possibility to investigate thermodynamic processes, such as phase transitions, that involve the cooperative nature of biomembrane systems (Venturoli and Smit, 1999), and that are outside the timeand length-scale range of all-atom simulations methods. The DPD method has been adopted to study mesoscopic models for pure lipid-bilayer systems (Venturoli and Smit, 1999; Shillcock and Lipowsky, 2002), for bilayers containing co-surfactants (Kranenburg and Smit, 2004; Kranenburg et al., 2004b). Very recently, Venturoli et al. (2005) have developed a CG model (and studied by the DPD method) for a DMPC lipid bilayer with embedded proteins of different sizes and hydrophobic lengths. The aim of the model study was to understand whether, due to hydrophobic mismatch and via the cooperative nature of the system, proteins may prefer to tilt (with respect to the normal direction of the bilayer plane) or even to bend, rather than to induce a bilayer deformation without a tilting. To illustrate the application of the DPD method to lipid membranes, some details of the mesoscopic model of Venturoli et al. (2005) are discussed below. M.M. Sperotto et al. / Chemistry and Physics of Lipids 141 (2006) 2–29 11 Fig. 2. Atomistic and mesoscopic representation of DMPC. Hydrophilic head-beads are indicated in shading and hydrophobic tail-beads in open representation. To build a mesoscopic model of membrane, one starts by coarse-graining each molecule of the system (or groups of molecules) by a set of beads, which – depending on whether they refer to water molecules, hydrophobic or hydrophilic part of the lipids or the proteins – interact differently with the surrounding beads. An atomistic representation of a DMPC lipid and its corresponding coarse-grain model is shown in Fig. 2, where the hydrophilic beads are indicated in shading and the acyl-chain beads in open representation. In Venturoli et al. (2005), the phospholipid is modelled by three headgroup beads and five beads in each chain (Kranenburg et al., 2004a). The model of for a transmembrane protein is built by first connecting a varying number of hydrophobic-like beads into a chain, to the ends of which are attached three headgroup-like beads; these are then linked together into a bundle of NP amphipathic beadchains. Three typical model-protein sizes were considered, consisting of NP = 4, 7 and 43 chains, respectively. These sizes represent typical protein/peptide sizes: the hydrophobic section of single-spanning membrane proteins like glycophorin (MacKenzie et al., 1997), and the M13 major coat protein from phage (Stopar et al., 2003; Bechinger, 1997) or ␣-helical synthetic peptides (Morein et al., 2002) may be modelled by a skinny NP = 4 type. -Helix proteins like gramicidin A (Killian, 1992) may be modelled by a NP = 7 type. Larger proteins consisting of transmembrane ␣-helical peptides that associate in bundles, or -barrel proteins (von Heijne and Manoil, 1990) may be modelled by a NP = 43 type: bacteriorhodopsin (Henderson and Unwin, 1975), lactose permease (Foster et al., 1983), the photosynthetic reaction center (Deisenhofer et al., 1985), cytochrome c oxidase (Iwata et al., 1995), or aquaglyceroporin (Fu et al., 2000). By varying the protein hydrophobic sections, i.e., the number of hydrophobic chain-beads, one can sample different hydrophobic mismatch conditions. Fig. 3a and b shows a cartoon of a model lipid and a protein of size NP = 43, respectively. Fig. 3c shows the snapshots of typical configurations of the assembled bilayer Fig. 3. Schematic representation of a model-lipid (a), and a model protein (NP = 43) (b). In the snapshots in (c) are shown typical configurations (as results from the simulations) of the assembled bilayer with embedded model proteins with NP = 43, 7 and 4, respectively. 12 M.M. Sperotto et al. / Chemistry and Physics of Lipids 141 (2006) 2–29 with embedded proteins with three different sizes, corresponding to NP = 43, 7 and 4, respectively. 3.2. Dissipative particle dynamics The (DPD) simulation method (Hoogerbrugge and Koelman, 1992; Warren, 1998; Jury et al., 1999) was originally based on the idea of simulating the fluid hydrodynamics of systems composed of beads (where in the case of a fluid the ‘bead’ is a small droplet of the fluid), in analogy with the way the Navier–Stokes equations reproduce the motion of a real fluid. Each bead moves according to Newton’s equation of motion, and interacts according to simplified force laws. The beads interact with each other via conservative, random, and dissipative forces of the pairwise-additive type. The combined effect of the dissipative and the random forces, acts as a thermostat, which conserves the (angular) momentum, and thus provides the correct hydrodynamics to the system. The total force acting on each bead i, is thus expressed as a sum over all other beads, j, which are within a certain cutoff radius Rc from bead i. The force of conservative type is related to the interactions between beads which may be bound or not bound together. In the case of beads not bound together (i.e., not belonging to the same mesoscopic molecule) the interaction potential is softrepulsive, having an extension range defined by a cutoff radius. The conservative force may also have an elastic contribution, which derives from the harmonic force used to tie two consecutive beads in the chains of either the lipid or the protein. A detailed description of the DPD-simulation potentials, and other physical parameters, can be found in Venturoli et al. (2005). A physical quantity which still constitute a matter of debate is the value of the surface tension to use in the simulations of a model bilayer. It has been suggested (Jähnig, 1996) that unconstrained, self-assembled bilayers are at their free energy minimum, characterized by having a zero value of the surface tension. Nevertheless, it is still a matter of debate which value of the surface tension should be used in molecular simulations (Feller and Pastor, 1996, 1999; Marrink and Mark, 2001; Goetz et al., 1998). Venturoli et al. (2005) have adopted an approach in which they mimic the experimental condition by simulating a system in which they impose a value of the surface tension. This is done by using a hybrid scheme based on both the DPD and the Monte Carlo (MC) simulation method. The DPD method was used to evolve the positions of the beads, and the MC method was used to impose a given surface tension on the bilayer. This hybrid method ensures the total volume of the system to remain constant (Venturoli and Smit, 1999). The next two sections focus on the issues concerning the range of perturbation induced by a protein on the nearby mismatched lipid bilayer, and its dependence on protein size, and on the simultaneous occurrence of protein tilting (or even bending) to adjust for hydrophobic mismatch. 3.3. Protein-induced bilayer perturbations The spatial fluctuations which may occur in biomembranes give rise to inhomogeneities in the lateral distribution of membrane components. Lateral inhomogeneities can be induced, as well as harvested, by the presence of proteins. In the past, the following quantities have been determined by computer simulations on lattice models: the extent of the lipid-bilayer perturbation induced by proteins (Sperotto and Mouritsen, 1991), and its dependence on factors such as the degree of hydrophobic mismatch and the size of the protein (i.e. the curvature of the protein hydrophobic surface in contact with the lipid hydrocarbon chains). It was found that, away from the protein, the perturbation decays in a exponential manner, and can therefore by characterized by a decay length, ξ P . ξ P is a measure of the size of small-scale inhomogeneities (i.e. domains) experienced by proteins when embedded in the lipid bilayer. In a sense, ξ P is also a measure of the extension of the range over which the lipid-mediated interaction between proteins may operate. Data from MD simulations on all-atom models have confirmed too that, within the time scale of the simulations, under mismatch conditions, a protein can induce a deformation of the lipid-bilayer structure (Chiu et al., 1999; Petrache et al., 2000, 2002; Jensen et al., 2001), and that the deformation is of the exponential type (Jensen and Mouritsen, 2004). The same type of studies have also shown that tilting may also occur for membrane peptides (Belohorcová et al., 1997; Shen et al., 1997); however, to reduce a possible hydrophobic mismatch synthetic peptides might instead prefer to deform the lipid bilayer, or even bend, rather than undergo tilting (Petrache et al., 2002). To determine the structural changes of the bilayer due to the presence of the protein, Venturoli et al. (2005) calculated first the hydrophobic thickness, dL0 , of the pure lipid bilayer, i.e., far a way from the protein surface (see Fig. 4). The effect of the protein, on the surrounding bilayer structure, was then determined by calculating the lipid-bilayer hydrophobic thickness, dL (r), as function of the radial distance r from the protein hydrophobic surface, namely at the interface with the lipid hydrocarbon chains, as schematically illustrated in Fig. 4; while the bilayer-induced effects on the protein M.M. Sperotto et al. / Chemistry and Physics of Lipids 141 (2006) 2–29 Fig. 4. Schematic illustration which shows the statistical quantities calculated from the simulations: the pure lipid-bilayer hydrophobic thickness, dL0 , the perturbed lipid-bilayer hydrophobic thickness, dL (r), the protein hydrophobic length, dP , the tilted-protein hydrophobic length, dPeff , and the tilt angle, φtilt . were quantify in terms of the protein tilt angle φtilt with respect to the bilayer normal, as shown in Fig. 4, where dP is the protein hydrophobic length, and dPeff is defined as the projection onto the normal of the bilayer plane of the protein hydrophobic length, dPeff = dP cos(φtilt ). The protein tilt issue is discussed in the following section. The method of calculation of dL (r), dPeff , dP , and φtilt are discussed in details in Venturoli et al. (2005). The behavior of dL (r) allows to access the extension of the protein-mediated perturbation on the bilayer. Based on previous theoretical findings (Sperotto and Mouritsen, 1991; Fattal and Ben-Shaul, 1993), one can first assume that the perturbation induced by the protein on the surrounding lipids is of an exponential type. One can then verify this assumption later by analyzing the deviation of the functional form of the calculated dL (r) from the assumed one. If the behavior of dL (r) is exponential, the protein-induced perturbation can be expressed in terms of a typical coherence length, i.e. the decay length ξ P : dL (r) = dL0 + (dP − dL0 ) e−r/ξP (2) where dL0 is the mean hydrophobic thickness of the unperturbed pure lipid bilayer. The above equation expresses the fact that away from the protein surface, and at distances at least of the order of ξ P , the perturbed dL (r) decays to the bulk value dL0 , namely the value corresponding to that of the pure lipid system at a chosen temperature. By knowing dL (r), dP , and dL0 , and by using Eq. (2), one can estimate ξ P . Since the proteins may be subjected to tilt, the input parameter for dP that one uses to fit Eq. (2) is not the actual hydrophobic length of the model protein, but instead the effective length, dPeff (see Fig. 4). Venturoli et al. (2005) have calculated the lipidbilayer hydrophobic thickness profile, dL (r), as a function of the distance r from the protein surface, and for a number of systems, containing proteins having one of the three different sizes, NP = 4, 7, and 43, and with different hydrophobic lengths dP , i.e., hence subjected to different hydrophobic mismatch, d. It was found that, 13 when subjected to hydrophobic mismatch, the protein induces a perturbation of the lipid bilayer in its vicinity, and that the perturbation decays in a manner that depends on the mismatch, and on protein size. For a given protein size, NP , if d < 0 the correlation length of the perturbation, ξ P , increases with decreasing mismatch (absolute value), while for positive mismatch the opposite happens, and the correlation length increases with increasing mismatch. Also, in the case of d < 0 the decay length increases by increasing the protein size. Instead there is no detectable ξ P dependence on d in the case of d > 0, at least at the considered temperature, around 60 ◦ C, well above the melting temperature of the pure system. Fig. 5 shows the thickness profiles for two values of mismatch, d < 0 (left column), and d < 0 (right column), and for the three considered protein sizes. In the case of d < 0, the lipids around the protein shrink to match the protein hydrophobic surface, while in the case of d > 0 the lipids in the vicinity of the protein stretch and become more gel-like than the bulk lipids far away from the protein. Also, at negative mismatch, the orientation of the protein is perpendicular to the bilayer plane (see Fig. 5a, c and e, where dP = dPeff ), while at positive mismatch, the protein tilts to decrease its effective hydrophobic length, the more skinny the protein is the more pronounced the effect becomes (see Fig. 5b, d and f, where dP = dPeff ). The tilt issue is discussed in details in the following section. MD simulations (Jensen and Mouritsen, 2004) on allatom model of bilayers of fluid POPE and POPC with embedded the membrane channel aquaglycerolporin – which would correspond to the mesoscopic-protein size NP = 43, and to a negative mismatch d ∼ −4 Å – predict an exponential protein-induced bilayer deformation with a decay length ξ P ∼ 10 Å, in good agreement with the results from the DPD simulations. Also, MD simulations on POPC bilayers with embedded the membrane channel gramicidin A – corresponding to the mesoscopicprotein size NP = 7 – show an exponential deformation with a coherence length smaller than that obtained for aquaglycerolporin (M.Ø. Jensen, private communication). The few experimental attempts to estimate the extent of the protein-induced bilayer perturbation confirm a mismatch dependence of the extent of the perturbation, too. These experiments refers to pure lipid bilayers with embedded Bacteriorhodopsin (Rehorek et al., 1985; Bryl and Yoshihara, 2001), lactose permease (Lehtonen and Kinnunen, 1997), and the synthetic ␣helical peptides (Ridder et al., 2004; Weiss et al., 2003). Experimental studies on protein-induced lipid flip-flop (Kol et al., 2003) indicate that the larger the protein size the more reduced is its ability to induce flip-flop. This 14 M.M. Sperotto et al. / Chemistry and Physics of Lipids 141 (2006) 2–29 Fig. 5. Calculated values of dL (r) (open circles) and fitted ones using Eq. (2) (solid line) as function of the distance r from the protein surface. For convenience, in each plot is also shown the level value of the pure lipid-bilayer thickness (dashed line), the measured protein hydrophobic length dP (shaded area), and the effective protein hydrophobic length dPeff (open area) which is defined as the projection of dP onto the normal to the bilayer plane. The data refer to the three protein sizes: NP = 4 (a and b), NP = 7 (c and d), and NP = 43 (e and f), and to a negative (see plots in 1st column of page), and to a positive mismatch (see plots in 2nd column of page). The systems are studied at a temperature around 60 ◦ C. fact is an indirect confirmation of the protein-size dependence of the extent of the perturbation induced by the protein (Ridder et al., 2004), as suggested by the MD, as well as, the DPD simulation data. An interesting prediction which arises from the DPD simulations is that the larger the protein, the more the behavior of dL (r) deviate from an exponential one (see Fig. 5e and f). The DPD simulation data show an ‘over- M.M. Sperotto et al. / Chemistry and Physics of Lipids 141 (2006) 2–29 Fig. 6. Re-plot of the DPD simulation data (open circles) from Fig. 5e and f (). The solid lines are obtained from membrane elasticity theory (May, 2002) through fitting of the data points to Eq. (3). The fitting parameters are dP , dL0 , ξ P , and ξ C . shooting’ (at d < 0) or ‘undershooting’ (d > 0) effect. The ‘undershooting’ (‘overshooting’) phenomenon is a consequence of the constraint of uniform density in the bilayer core, and occurs if the protein is large enough that tilting is unfavorable, and if the mismatch is high (low) enough that even the ordered, gel-like (disordered, fluidlike) lipids closest to the protein are not able to match the protein hydrophobic surface. One can imagine the undershooting region as forming a sector of an inverted micelle, and the overshooting region as forming a sector of a micelle. The occurrence of ‘undershooting’ and ‘overshooting’ is also a signature for the damped oscillatory behavior predicted by membrane elasticity theory (Dan et al., 1993; Nielsen et al., 1998; May, 2002)—that is discussed in details in Section 2.2. It is particularly pronounced for large proteins and adds to the characteristic decay length, ξ P , of the membrane thickness relaxation a second characteristic length, ξ C , corresponding the wave length of the oscillatory part. The two lengths ξ P and ξ C depend on the elastic membrane properties; see Eq. (1). According to the elasticity theory the reason for the overshooting and undershooting effect is the competition between stretching and bending modes of the membrane. Fig. 6 shows again the DPD data sets corresponding to the large proteins displayed in Fig. 5e and f. The solid lines are fits to the data sets from membrane elasticity theory (May, 2002) according to dL (r) = dL0 + (dP + dL0 ) e−r/ξP ξC2 − 3ξP r r + 2 × cos sin ξC ξP ξP − 3ξC (3) In both cases it was found a value of ξ P ≈ ξ C ≈ 1 nm in agreement with the expectation from membrane elasticity theory: see Eq. (1). This is remarkable because 15 the DPD simulation operates with discrete lipid models whereas elasticity theory is based on a continuum approach. It suggests that the membrane elasticity approach is reasonable even though only a few lipid shells around the protein are perturbed. However, while it is easy to include the finite radius of the protein into Eq. (3), no statistical model is yet available that accounts for the ability of the protein to tilt. The advantage of the mesoscopic approach is that it can account for protein tilting, as discussed in the next section. Incidentally, Nielsen et al. (1998), using a phenomenological elastic lipid–protein model predicted an overshooting behavior of dL (r) similar to what is shown in Fig. 6. However, despite this similarity, the non-monotonic behavior of dL (r) observed by Nielsen et al. (1998) is probably due to the some specific boundary conditions imposed a priori on the system (like, for example, the contact slope of the lipids nearest to the protein), rather than to the competition between the stretching and bending modes of the bilayer, or – said in mesoscopic terms – to constant density requirements. The perturbations of the lipid structure around a protein may have biological implications. In particular, they could promote lipid sorting and, hence, favor bilayer fusion and affect lipid-mediated protein–protein interactions, as recently suggested by Nielsen et al. (2004). Such curved structures might also affect passive permeability in the vicinity of the protein. Also, if the amount of lipids involved in the ‘overshooting’ (‘undershooting’) phenomenon is sufficiently high to be detected experimentally, the value of the lipid order parameters measured by spectroscopic techniques could be affected in such a way that the derived values of the lipid-bilayer hydrophobic thickness may be underestimated (overestimate). 3.4. Lipid-induced protein tilting To adapt to a too-thin bilayer, and to minimize the exposure of their hydrophobic moieties to the water environment, proteins may tilt in a manner that is mismatch- and protein-size dependent (i.e. the larger the protein, the less pronounced the tilting). Very recently, Ramakrishnan et al. (2005) have confirmed the tilt dependence on protein size. By infrared spectroscopy they measured the tilt angle of -barrel proteins, OmpA and PhuA, characterized by the same hydrophobic length, but having two very different sizes; they found that the larger protein tilt systematically less than the smaller, at equal mismatch. Despite the limited timescale sampled by MD simulations, the possibility that different type of skinny synthetic peptides may tilt (or even bend) when subjected to positive mismatch condi- 16 M.M. Sperotto et al. / Chemistry and Physics of Lipids 141 (2006) 2–29 Fig. 7. The protein tilt angle, φtilt , dependence on mismatch, d = d̃P − dL0 (i.e. different protein hydrophobic lengths). The results refer to a temperature around 60 ◦ C – hence a fixed value of mismatch – and to the three protein sizes NP = 4, 7 and 43. The dashed lines are only a guideline for the eye. On the right are shown the snapshots of typical configurations of the systems with embedded proteins of different sizes. tions has also been confirmed by MD simulations on allatom models (Shen et al., 1997; Belohorcová et al., 1997; Petrache et al., 2002)—although the degree of tilting varies from system to system. This might suggest that the degree of tilting depend on local features of the peptides, for example their amino acid sequence. Recent experimental investigations, which were performed to systematically correlate peptide-tilting with mismatch might confirm this suggestion. In fact the data from solid state NMR spectroscopy (Strandberg et al., 2004; Özdirekcan et al., 2005) referring to hydrophobic synthetic ␣-helical peptides, WALP23 and KALP23, flanked by either tryptophan or lysines residues (Strandberg et al., 2004; Özdirekcan et al., 2005), and from fluorescence spectroscopy referring to the M13 major coat protein peptides (Koehorst et al., 2004), indicates that – although in all three cases it was found that the tilt angle systematically increases by increasing hydrophobic mismatch – the actual values of the tilt angles is different even if the different types of peptides are subjected to approximately the same hydrophobic positive mismatch. Turning now to the simulation data shown in Fig. 5a, c and e, the difference between the values of dP and dPeff suggest too that proteins may tilt when subjected to positive mismatch. Fig. 7 shows the calculated protein tilt angle, φtilt (see schematic representation in Fig. 4, where dPeff is defined as the projection onto the normal of the bilayer plane of the protein hydrophobic length, dPeff = dP cos(φtilt )), with respect to the bilayer normal as function of d, and for the three protein sizes, NP = 4, 7, and 43. As the dP increases (and d becomes positive), the protein may undergo a significant tilting: depending on NP , the more skinny the protein is, the more pronounced the tilt becomes, consistently with the experimental data of Ramakrishnan et al. (2005)—as illustrated by the snapshots on the right of Fig. 7, which refers to the three chosen protein sizes, and to the highest (positive) value of mismatch. Also, when the skinny model protein (NP = 4) experiences a high positive mismatch, bending of the protein may take place (in addition to tilting), as shown in the snap-shot in Fig. 7(top, right), and as predicted by MD simulations on all-atom models (Shen et al., 1997; Belohorcová et al., 1997). In an attempt to see if the functional dependence of the tilt angle on mismatch is similar for the three types of ␣-helical peptides, mentioned above, WALP23, KALP23, and M13, and if the dependence bears some resemblance to the one predicted by the DPD simulations, in Fig. 8 is plotted φtilt as a function of mismatch, for the three different type of peptides, and for the model peptide with NP = 4. The coloured areas are defined by the error bars. The plotted φtilt values show a remarkable resemblance of the functional dependence on d between the considered experimental systems and the simulated one. The predicted functional dependence of the tilt on mismatch for the case of the smaller model proteins, NP (see Fig. 8) might suggest that, while the actual values of the protein tilt angle might depend on the specific peptide sequence, the functional dependence on mismatch might have a somehow general character—although more experiments are needed to confirm this idea. Model studies on tilt angle like the one presented in this section may help to understand whether the tilting of helices belonging to bundles is due to an intrinsic M.M. Sperotto et al. / Chemistry and Physics of Lipids 141 (2006) 2–29 Fig. 8. The protein tilt angle, φtilt , dependence on mismatch, d = d̃P − dL0 , at temperatures above the melting temperature of the pure lipid bilayer. The simulation data refer to a model protein corresponding to NP = 4 (in black), and the experimental data refer to three different types of ␣-helical peptides: M13 coat protein peptide (in red), KALP23 synthetic peptide (in blue), and the WALP23 synthetic peptide (in green). The dashed lines are a guideline for the eye. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of the article.) property of the helices or is due, instead, to hydrophobic matching. The results shown in Fig. 8 and the proteinsize dependence shown in Fig. 7 might help to predict how the tilt angle can change by changing the number of synthetic ␣-helical peptides of which a proteinbundle may be made. In fact, knowing the tilt angle of individual peptides, and the relative change of tilt angle (as shown in Fig. 7) upon protein-size changes, one could suggest what would be the tilt angle of the bundle. 4. Simulations of membrane processes 4.1. Protein translocation into membranes For membrane proteins to perform certain biological functions as channels, transporters, signal transducers and other devices (Baumgaertner, 2006) they need to be inserted into compartmental walls made of lipid bilayers. The majority of membrane proteins, so-called constitutive membrane proteins (White and Wimley, 1994), are inserted into a lipid bilayer at the same time as they are synthesized. This class of peptides and proteins usually accomplish insertion into or translocation across membranes assisted by a complex proteinacious machinery, “translocon” (Schnell and Hebert, 2003; White and 17 von Heijne, 2004), or through fusion of membranes. The translocon, binds to the ribosome synthesizing the membrane protein and threads the protein into a lipid bilayer through an internal channel. In contrast, the non-constitutive membrane proteins, such as toxins and antimicrobial peptides, insert themselves into a lipid bilayer without assistance. Since the membrane itself keeps the function of a permeability barrier between the compartments involved, and since proteins are macromolecules, their passage through or insertion into the membrane creates a non-trivial problem. The insertion mechanism of such proteins is poorly understood. The spontaneous insertion is based on physicochemical processes, for which two-stage or four-stage models have been proposed (Engelman and Steitz, 1981; Popot and Engelman, 2000; White and Wimley, 1994, 1999). According to the models, the peptides first get adsorbed to the membrane surface where they change their conformation into an ␣-helix, and then they start insertion, whose molecular mechanism is not well understood, followed by aggregation into an organized structure. Wellaccepted driving force of insertion is the hydrophobic nature of the peptide segments, and in fact peptide insertions are usually discussed by use of the hydrophobicity scale of the component amino acids. The field-driven insertion process of the antimicrobial peptide alamethicin in lipid–water and octane–water environments have been studied by MD simulations (Tieleman et al., 2001). An external electric field was used to mimic the membrane potential. They found during MD simulation of 10 ns that alamethicin did not insert into a phospholipid bilayer which was attributed to the slow dynamics of the peptide and lipids. However, in octane N-terminal insertion occurs at sufficiently high field strengths. Insertion of alamethicin occurred in two steps, corresponding to desolvation of the Gln7 side chain, and the backbone of Aib10 and Gly11. Surface binding and subsequent penetration of the bilayer was observed during MD simulations (Shepherd et al., 2003) for the hydrophobic ally oriented peptides, while the charge-oriented peptides demonstrated at most partial surface binding. The peptides were initially placed in an ␣-helical conformation on either side of a zwitterionic lipid bilayer about 10 Å from the interface. Insertion of the peptides into the bilayer caused a dramatic increase in the lateral area per lipid and decrease in the bilayer thickness, resulting in substantial disordering of the lipid chains. The work of Lopez et al. (2004) demonstrates how a direct insertion of a typical non-constitutive protein into a lipid membrane could take place. Modelling a membrane protein as a hydrophobic tube with hydrophilic sites at 18 M.M. Sperotto et al. / Chemistry and Physics of Lipids 141 (2006) 2–29 the tube’s ends, they observed a spontaneous insertion of a generic nanosyringe into a lipid bilayer. One of the main findings was that lipids assist one of the ends of the nanosyringe to cross the hydrophobic core of the membrane. In the simulated model, small groups of atoms were represented by sites interacting with each other by bounded and non-bounded forces, like atoms in conventional MD simulations. In a subsequent work (Lopez et al., 2005) the authors considered the insertion of a model pore into a membrane using a mesoscopic approach. The papers of Lopez et al. (2004, 2005) as well as the mesoscopic modelling approach are discussed in more detail in Section 3. Im and Brooks (2005) have explored membrane insertion and interfacial folding for the WALP and TMX series of peptides by using an implicit membrane generalized Born model and replica exchange molecular dynamics. All WALP and TMX peptides showed spontaneous N-terminal-led insertion through the formation of a continuous ␣-helix arising from thermal fluctuations of ␣-helical hairpin conformations formed at the interface or in the membrane. A straight ␣-helix was not observed as a stable interface-bound conformation but existed as a transient conformation before membrane insertion as a TM helix. The authors suggest that the formation of such a straight ␣-helix is a rate-determining step in insertion. The conformation associated during the MD simulations suggest that a revision in traditional membrane insertion/association free energy calculations is necessary to include the influence the conformational changes that occur between peptide–membrane association and inser- tion, i.e., the presence of helical hairpins at the interface. However, it must be noted that the implicit membrane generalized Born model may not include correctly the proper interfacial characteristics of the lipid–water interface, which may have a significant effect on the transient protein structure while crossing the interface. The interactions of a model peptide (WALP-16) with an explicitly represented DPPC membrane bilayer was simulated (Nymeyer et al., 2005) using the replica exchange molecular dynamics algorithm (Swendsen and Wang, 1986; Sugita and Okamoto, 1999; Nymeyer et al., 2004). A spontaneous, unbiased insertion of WALP-16 into the DPPC bilayer and its folding into an ␣-helix with a transbilayer orientation was observed. From calculations of the free energy surface it is concluded that insertion of the peptide into the DPPC bilayer precedes secondary structure formation. This latter observation disagrees with the dominant conceptual model (Popot and Engelman, 2000; White and Wimley, 1999) which is that a surface-bound helix is an obligatory intermediate for the insertion of ␣-helical peptides into lipid bilayers. The observed translocation mechanism is favored because of a large (>100 kcal/mol) increase in system entropy that occurs when the unstructured WALP-16 peptide enters the lipid-bilayer interior. The insertion/folding pathway that is lowest in free energy depends sensitively on the near cancellation of large enthalpic and entropic terms. This suggests the possibility that intrinsic membrane peptides may have a diversity of insertion/folding behaviors depending on the exact system of peptide and lipid under consideration. Fig. 9. (a) Snapshot of the extracting melittin pore at time t = 100 ps. (b) Stretching z(t) of the C␣ atoms of the charged residues Lys21-Arg22 (red), Arg22-Lys23 (green), Lys23-Arg24 (blue), Arg24-N27 (black) as function of time t. The vertical broken lines approximately indicates the transition zone. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of the article.) M.M. Sperotto et al. / Chemistry and Physics of Lipids 141 (2006) 2–29 The importance of the lipid–water interface for the kinetics of the insertion process was concluded from MD simulation of the “inverse” insertion process, that is, the extraction of membrane peptides from a lipid-bilayer membrane (Lin and Baumgaertner, 2005). During the extraction of a tetrameric ␣-helical melittin pore from a POPC bilayer (Fig. 9a), it was observed that the force F(t) increases almost linearly with time until a certain maximum at time tc1 ≈ 100 ps. Accordingly, the intermolecular distances between the successive residues, as shown in Fig. 9b, exhibit a characteristic transition regime. Since the pulling force was exerted on the N-terminus, N27, the distance to its nearest neighbor residue R24, z(t) = z[R24](t) − z[N27](t), exhibits a significant stretching until R24 undergoes a rupture at tc2 out of the lipid–water interface. The first (tc1 ) and the second (tc2 ≈ 200 ps) vertical broken lines in Fig. 9b indicate approximately the transition regimes where the first (R24) and the last (K21) charged residue of the melittin signature KRKR have followed the extraction process and have left the lipid-head–water interface. A more detailed analysis of the results from this computer experiment indicates that Coulomb interactions between the charged residues and the lipid–water interface plays a significant role and may represent a barrier against spontaneous insertion. This suggestion is corroborated by the experimental observations that in many cases, in particular for melittin, an aggregate of peptides is necessary in order to induce a significant perturbation of the lipid–water interface to promote a subsequent insertion event. 4.2. Protein assemblies in membranes Membrane protein domains are often organized as assemblies of polypeptide segments interacting with the lipid bilayer and constituting a functionally active and finely regulated biological machine involved in ion and molecular transport across the membrane, cell communication, signaling, etc. Studies of membranebound segments are thus essential for understanding structure–function relationships of membrane proteins. Studies of proteins in simplified membrane models were described by Roux and Karplus (1994), Baumgaertner (1996), and others. Often, the hydrophobic core of a membrane is modelled by Lennard–Jones hydrocarbon-like particles, a polarizable cubic lattice with low dielectric permeability (reviewed in Roux and Karplus, 1994), or by a monolayer of hard parallel cylinders representing the lipid chains (Baumgaertner, 1996). Many other mesoscopic models have been proposed which are discussed in Section 3. Among other 19 properties, these models permit investigation of orientational order and lateral density fluctuation of the lipid matrix, which are important for partitioning and ␣-helix formation of TM peptides. In a number of studies the membrane was approximated by introducing an additional solvation term into the potential energy function to represent interaction of a protein with its environment. Usually such potentials are taken dependent on hydrophobic properties of residues and their positions relative to the bilayer (Edholm and Jähnig, 1988; Milik and Skolnick, 1993, 1995; Baumgaertner, 1996). The results obtained provide interesting insights into peptides’ behavior in the membrane environment. However, such methodology seems to be somewhat oversimplified because amino acid residues are treated as point “hydrophobic sites” without taking into account the conformation and hydrophobic nature of atoms and/or atomic groups. 4.2.1. Protein assemblies in membranes: single peptides Specific lipid–protein interactions involved in the anchoring and stabilization of membrane-bound proteins are of central importance in a large number of fundamental processes occurring at the surface of the cell. However, despite the development of powerful techniques such as X-ray crystallography, electron microscopy, and nuclear magnetic resonance (NMR), the characterization of lipid–protein interactions remains difficult because of the complexity of the bilayer environment. At the present time, even qualitative information gained by performing detailed computer simulations of protein–membrane complexes can be valuable, because only scarce information is available from experiments about the structure and dynamics of these systems. One of the first all-atom simulations of an ␣-helical peptide in an explicit bilayer membrane with explicit water molecules was conducted by Tieleman et al. (1999a). They found that alamethicin underwent hingebending motion about its central Glyf-X-X-Pro sequence motif, because the polar C-terminal side chains provided an “anchor” to the bilayer/water interface via formation of multiple H-bonds. This explains why the preferred mode of helix insertion of alamethicin into the bilayer is N-terminal. The interaction of melittin with a fully hydrated DMPC bilayer was examined by molecular dynamics simulations (Bernéche et al., 1998). The initial configuration of the system was constructed with melittin in an ␣-helical conformation bound parallel to the membrane–solution interface. Melittin perturbs the bilayer significantly such that the order of the lipid acyl 20 M.M. Sperotto et al. / Chemistry and Physics of Lipids 141 (2006) 2–29 chains is smaller in the upper layer, whereas it is larger for those in the lower layer. The perturbation of the bilayer results in a local curvature, a reduction of the thickness of the membrane, and the hydrophobic core of the membrane is reduced by 30%. However, the acyl chains of the lipids adopt particular conformations to avoid leaving a large cavity under the amphipathic helix. Recently, an implicit solvation model for membranebound proteins was proposed (Efremov et al., 1999a). In this model solvation parameters take into account the hydrocarbon core of a membrane, water, and weak polar solvent (octanol). An optimal number of solvation parameters was chosen based on analysis of atomic hydrophobicities and fitting experimental free energies of gas–cyclohexane, gas–water, and octanol–water transfer for amino acids. This solvation model was used to assess membrane-promoting ␣-helix formation. To accomplish this, all-atom models of 20-residue homopolypeptides – poly-Leu, poly-Val, poly-Ile, and poly-Gly in initial random coil conformation – were subjected to non-restrained Monte Carlo conformational search in vacuo and with the solvation terms mimicking the water and hydrophobic parts of the bilayer. All the peptides demonstrated their largest helix-forming tendencies in a non-polar environment. In a subsequent work Efremov et al. (1999b) have employed Monte Carlo simulations of the implicit solvation model to explore conformational space of several membrane-binding peptides in environments of different polarity and in vacuum. The solvent effects were treated using an atomic solvation parameters (ASP). The simulations were done for all-atom models of membrane-bound peptides, such as transmembrane segments A and B of bacteriorhodopsin, the hydrophobic segment of surfactant lipoprotein, SPC, and magainin2. The results emphasize that the ␣helical conformation is promoted by non-polar solvent and exists in a wide energy range. Conformational properties of SP-C and magainin2 in the membrane-like environment were also found to be in accord with available experimental data. The results for SP-C do not confirm preference of all-helical structure for SP-C in water. Unlike the membrane-spanning proteins, atomicscale structural information about peripheral membrane proteins is scarce. Functions of some of these proteins require them to be folded in an aqueous environment and also be capable of inserting themselves into membranes. Studies of such “ubiquitous” molecules provide an opportunity to examine the determinants of an insertion event. Unfortunately, the experimental analysis is seriously hampered by difficulties in preparation of suitable samples containing these proteins in the membranebound state. Their high-resolution structures obtained so far reveal the only binding motif: an amphiphilic ␣-helix either lying on the bilayer surface or partly immersed into the hydrophobic core (White and Wimley, 1999; Shai, 1999). Based on these structural data, a number of successful molecular modelling studies of interactions between ␣-helices and membrane interface have been reported (Forrest and Samson, 2000; Tieleman et al., 2001). One question of interest is whether peripheral membrane proteins possess also other types of binding motifs. This has been investigated recently (Efremov et al., 2002) by incorporation of -sheet proteins into membrane employing Monte Carlo simulations with an implicit membrane model (Efremov et al., 1999a). Cardiotoxins, are found to retain the overall “three-finger” fold interacting with membrane core and lipid–water interface by the tips of the “fingers” (loops). The implementing of the peptide at certain places of the membrane, critically depends upon the structure, hydrophobicity, and electrostatics of certain regions. The simulations reveal apparently distinct binding modes for cardiotoxins via the first loop or through all three loops, respectively. This computational study may be used to study “partitioning” of proteins with diverse folds into lipid bilayers. The structural properties of the endogenous opioid peptide dynorphin A (1–17) (DynA), a potential analgesic, were studied with MD simulations in DMPC bilayers (Sankararamakrishnnan and Weinstein, 2000). Starting with the known NMR structure of the peptide, the N-terminal helical segment of DynA was initially inserted in the bilayer in a perpendicular orientation with respect to the membrane plane. Parallel simulations were carried out from two starting structures, that differ by 4 Å in the vertical positioning of the peptide helix. The simulation of the system (dynorphin + 86 lipids + 5300 waters) revealed that the orientation of the helical segment of DynA had undergone a transition from parallel to tilted with respect to the bilayer normal in both systems. Analysis shows that the tilted orientation of 50◦ adopted by the N-terminal helix is due to specific interactions of residues in the DynA sequence with phospholipid headgroups, water, and the hydrocarbon chains. Key elements are the “snorkel model”-type interactions of arginine side chains, the stabilization of the N-terminal hydrophobic sequence in the lipid environment, and the specific interactions of the first residue, Tyr. Water penetration within the bilayer is facilitated by the immersed DynA, in particular surrounding the arginine side chains. A mechanism of receptor interaction is proposed for DynA, involving the tilted orientation observed from these simulations of the peptide in the lipid bilayer. Kandasamy and Larson (2004) have performed molecular dynamics simulations of the interactions of M.M. Sperotto et al. / Chemistry and Physics of Lipids 141 (2006) 2–29 two ␣-helical antimicrobial peptides, magainin2 and its synthetic analog of MSI-78 embedded in an explicit POPC lipid bilayers. They used different initial conditions of the peptide, including a surface-bound state parallel to the interface, a trans-membrane state, and a partially inserted state. The simulations showed that both magainin2 and MSI-78 were most stable in the lipid environment, with the peptide destabilized to different extents in both aqueous and lipid–water interfacial environments. From all the simulations, they concluded that the hydrogen bonding interactions between the lysines of the peptides and the oxygens of the polar lipid headgroups are the strongest and determine the destabilization of the bilayer environment, as observed by the increase in lipid tail disorder and the induction of local curvature on the lipid headgroups by the peptides. The orientation and motion of a model lysineterminated transmembrane polypeptide, acetyl-KK(LA) 11-KK-amide, implemented in a POPC bilayer were investigated by MD simulation by Goodyear et al. (2005). In one simulation, initiated with the peptide oriented along the bilayer normal, in a second simulation the initial peptide orientation was chosen to match a set of experimentally observed alanine methyl deuteron quadrupole splittings. Simulated alanine methyl group orientations were found to be inequivalent, a result that is consistent with 2H NMR observations of specifically labeled polypeptides in POPC bilayers. Helix tilt varied substantially over the durations of both simulations. In the first simulation, the peptide tended toward an orientation about the helix axis similar to that suggested by experiment. In the second simulation, orientation about the helix axis tended to return to this value after an excursion. These results showed that interactions at the bilayer surface can constrain reorientation about the helix axis while accommodating large changes in helix tilt. Although large excursions in helix tilt may be accommodated by extension of the terminal lysine side chains, a situation often referred to as “snorkeling”, the localization of lysine side-chain charges at the bilayer surface appears to contribute to the adoption of a preferred orientation of the tilted helix about its axis. The results are relevant to understanding 2H NMR observations of alanine methyl deuterons on lysine-terminated transmembrane polypeptides. 4.2.2. Protein assemblies in membranes: interacting peptides Insertion and formation of membrane proteins involves the interaction of protein helices with one another in lipid environments. It has been postulated (Popot and Engelman, 2000) that individual helices are 21 stable separately as domains in a lipid bilayer. Their stability as domains is a consequence of the hydrophobic effect and main-chain hydrogen bonding. Other interactions then drive side-to-side helix association, resulting in a functional protein. Specific folding energy is provided mainly by packing of the preformed helices with each other, by loop structures, and by interactions with prosthetic groups. Additionally, ion pairs and hydrogen bonds between helices are sometimes found, and general contributions are made by interactions with the lipid environment MD simulations were performed on M13 coat protein (Sanders et al., 1991). The lipid bilayer was represented by a hydrophobic potential. The -sheet was more flexible than the ␣-helix. A comparison of the energies after 100 ps MD simulation showed that of the monomers, the ␣-helix has the lowest energy. The energy difference between ␣- and -structures decreases from 266 to 148 kJ/mol, when going from monomers to dimers. It was suggested that this difference will decrease with higher aggregation numbers. The non-specific lipid-mediated attraction between two proteins embedded in a bilayer membrane have been investigated for a model system using Monte Carlo simulations (Sintes and Baumgärtner, 1997). Two types of attraction with different regimes were identified: a depletion-induced attraction at short distances and a fluctuation-induced long range attraction, which originates from the gradients of density and orientational fluctuations of the lipids around each protein. The first MD simulation of an ion channel formed by a bundle of ␣-helices in a full lipid bilayer was conducted by Tieleman et al. (1999b). They investigated the effect of bundle stability of the ionization state of the ring of Glu18 side chains. If all of the Glu18 side chains were ionized, the bundle was unstable; if none of the Glu18 side chains were ionized, the bundle was stable. pKa calculations suggested that either zero or one ionized Glu18 is present at neutral pH, correlating with the stable form of the helix bundle. The dipole moments of water molecules within the pore were aligned antiparallel to the helix dipoles which contribute to the stability of the helix bundle. Lin and Baumgaertner (2000) have investigated the configuration and the stability of a single membrane pore bound by four melittin molecules and embedded in a fully hydrated POPC bilayer employing MD simulations. It was found that the initial tetrameric configuration decays with increasing time into a stable trimer and one monomer. This continuous transformation was accompanied by a lateral expansion of the aqueous pore exhibiting a final size comparable to experimental findings. The transformation lead to a “hydrophilic pore” 22 M.M. Sperotto et al. / Chemistry and Physics of Lipids 141 (2006) 2–29 where some lipid heads had translocated from the rim to the central part of the interface. It was hypothesized that pore growth, and hence cell lysis, is induced by a melittin-mediated line tension of the pore. MD simulations of glycophorin A (GpA) transmembrane helices embedded in sodium dodecyl sulfate (SDS) micelles have been performed by Braun et al. (2004) in order to identify contacts significant for helix dimerization. The simulation shows the formation of a complete micelle around wild-type GpA from an initially random placement of SDS molecules in an aqueous environment. The assembly and oligomerization of transmembrane ␣-helices is known to play a critical role, e.g., in immune system activity and cancers, genome sequencing studies reveal that 20–30% of open reading frames encode membrane proteins, indicating their biological significance. However, our understanding of assembly and oligomerization of membrane proteins is not as advanced as aggregation processes in soluble proteins. In particular, there have been a series of papers recently on MD simulations of the aggregation of prion proteins (e.g., Sekijima et al., 2003) and the formation of amyloid fibrils (e.g. Urbanc et al., 2004). Amyloid fibril structures formed by small peptides are amenable to both experimental studies and computer simulations and therefore they are particularly suitable for the investigation of the determinants of protein aggregation. Fig. 10 shows a peptide fragment from transthyretin, TTR(105–115). The native form of transthyretin, one of the 20 or so proteins Fig. 10. X-ray structure (PDB code 1BMZ) (Peterson et al., 1998) of the amyloid TTR dimer. The segment Y105-S115, whose sequence is YTIAALLSPYS, of each monomer is shown in magenta and red. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of the article.) that have been linked to amyloid diseases, is a homotetramer composed of 127 residue, mainly  subunits The TTR(105–115) peptide adopts a -strand conformation both in the native structure (in red and purple) and, with minor structural changes, in the amyloid fibril. A major challenge for molecular dynamics simulation in the future will be, following the recent progress for soluble proteins, to study the aggregation and the formation of membrane proteins. 4.3. Lipid-controlled function of membrane proteins Integral membrane proteins interact with lipid molecules in cellular membranes through their hydrophobic transmembrane regions. The interactions between membrane proteins and lipids can be categorized as either general or specific (Popot and Engelman, 2000). By general interactions, it is meant those resulting from the multiphase (membrane-aqueous) environment created by lipids in water, necessary for the stability of most integral membrane proteins. Specific interactions refer to the close association of certain lipids that may bind to the membrane protein, akin to a cofactor, to confer structural stability or to affect the protein’s function. Biochemical studies have demonstrated specific lipid binding to certain integral membrane proteins (Lee, 1998, 2003; Marsh and Horvath, 1998) and high-resolution crystal structures, now available for a few membrane proteins, reveal the presence of associated lipid molecules (Fyfe et al., 2001; McAuley et al., 1999; Iwata et al., 1995). The precise role of these specific lipid interactions is yet to be defined, but their importance is revealed by functional assays demonstrating that a number of membrane proteins such as electron transfer complexes I and III, cytochrome c oxidase (Fry and Green, 1981), ion channels (Valiyaveetil et al., 2002) and a number of transporter proteins require specific lipids for optimal activity (Lin et al., 1990; Vemuri and Philipson, 1989). The influence of lipid-bilayer deformations on the function of specific ion channels, ‘mechanosensitive (MS) channels’, has been investigated by means of computer simulations very recently by three groups (Gullingsrud et al., 2001; Elmore and Dougherty, 2001; Gullingsrud and Schulten, 2003; Biggin and Sansom, 2003; Sotomayor and Schulten, 2004). Mechanical forces act on living organisms from all directions throughout the environment, making mechanosensory transduction one of the fundamental sensory transduction processes in the biological world. MS channels are a class of ubiquitous membrane proteins gated by mechanical strain in the cellular membrane. As M.M. Sperotto et al. / Chemistry and Physics of Lipids 141 (2006) 2–29 molecular switches, MS ion channels convert mechanical forces exerted on cellular membranes into electrical or biochemical signals in physiological processes ranging from cellular turgor control in bacteria to touch and hearing in mammals. Some toxic peptides are selective inhibitor of MS channels. The mechanism of inhibition remains unknown, but it is known (Suchyna et al., 2004) that they modify the gating, thus violating a trademark of the traditional lock-and-key model of ligand–protein interactions. Suspecting a bilayer-dependent mechanism, the effect of toxins on gramicidin A (gA) channel gating have been examined experimentally (Suchyna et al., 2004). It was shown that the inhibition increases with the degree of hydrophobic mismatch between bilayer thickness and channel length, meaning that the toxic peptide decreases the energy required to deform the boundary lipids adjacent to the channel. These results suggest that modulation of membrane proteins by amphipathic peptides (mechanopharmacology) involves not only the protein itself but also the surrounding lipids. This has important therapeutic implications. MscL, a bacterial mechanosensitive channel of large conductance, is the first structurally characterized mechanosensor protein (Sukharov et al., 2001; Perozo and Rees, 2003). The protein is a pentamer (Fig. 11), approximately 50 Å wide in the plane of the membrane and 85 Å tall. Each 151-residue subunit consists of two transmembrane helices, labeled TM1 and TM2, and a cytoplasmic helix that extends some 35 A below the membrane. The TM1 helices are arranged so as to block diffusion through the channel at their N-terminal ends. Excision of the cytoplasmic domains has been found to have little effect on the gating properties of the channel. In general, it is believed that the gating of MS channels is induced by changes in the intra-bilayer pressure profiles which originate from bilayer deformation. In order to change the membrane tension it has been suggested (Perozo et al., 2002) that different hydrophobic mismatches at the protein–lipid interface induced by different types of lipids (mixtures) may cause an asymmetry of tension across the bilayer membrane and hence lead to a spontaneous curvature which controls the open and the closed state. Recently, MD simulations (Gullingsrud et al., 2001; Elmore and Dougherty, 2001) have indicated that the least mobile part of the protein could be identified as the gate, on the same location suggested by experimental findings. This part comprises the first five residues of the TM1 helices, which are shown to be pinched together to form a nonleaky occlusion. In particular, steered MD simulation (Gullingsrud et al., 2001) of the bare protein without membrane and without water, but under the application 23 Fig. 11. Ribbon representations of the structures of MscL. The top part (a) depicts the view from the plane of the membrane (gray area), whereas the transmembrane region viewed down the membrane normal is illustrated at the bottom (b). Individual subunits are represented in different colours (PDB code 1MSL). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of the article.) of a constant surface tension on the protein have shown that the transmembrane helices tilted considerably as the pore opened. The protein refolded into an open conformation, where the transmembrane helices flattened as the pore widened, with a minimal loss of secondary structure. The results indicate that membrane thinning and hydrophobic mismatch within the transmembrane helices my indeed drive gating. More recently steered 24 M.M. Sperotto et al. / Chemistry and Physics of Lipids 141 (2006) 2–29 MD simulations (Gullingsrud and Schulten, 2003) have revealed the mechanism for transducing membrane forces into channel opening. The channel was most easily opened when force was applied predominantly on the cytoplasmic side of MscL. The expanded state agrees well with proposed models of MscL gating (Perozo and Rees, 2003), in that it entails an irislike expansion of the pore accompanied by tilting of the transmembrane helices. MD simulations of the Mycobacterium tuberculosis homolog of the bacterial mechanosensitive channel of large conductance (Tb-MscL) have been performed recently (Elmore and Dougherty, 2001). Channel mutations led to observable changes in the trajectories, such as an alteration of intersubunit interactions in one of the mutants. In addition, interesting patterns of protein–lipid interactions, such as hydrogen bonding, arose in the simulations. MscS, the mechanosensitive channel of small conductance, is found in the inner membrane of Escherichia coli and its crystallographic structure in an open form has been recently solved (Bass et al., 2002; Edwards et al., 2005). Much of what we know about the molecular mechanisms of gating in MscS channel is derived from its crystal structure. Only recently have MD simulations (Sotomayor and Schulten, 2004) and experimental studies (Edwards et al., 2005) shed additional light on the structural changes that occur upon MscS gating. The crystal structure of E. coli MscS solved at a resolution of 3.9 Å reveals that the channel folds as a homoheptamer and has a large cytoplasmic region. Each subunit contains three TM domains. The precise conformation of MscS is controversial at present. Edwards et al. (2005) proposed a new structural model of MscS gating that involves rotation and tilt of pore-lining transmembrane helices. Rotation of the transmembrane domains in MscS closely resembles a current model for gating of MscL (Perozo and Rees, 2003). Recent study using molecular dynamics simulations (Sotomayor and Schulten, 2004) implied that water and ions cannot pass through the channel pore, suggesting that the crystal structure may reflect an inactive or desensitized state rather than the open state. When surface tension was applied, this led to channel widening. 5. Concluding remark Theoretical approaches like those previously described may be adopted, either to investigate the biophysical behavior of specific biomembrane systems (see Section 4), or to understand what are the basic physical principles that govern the behavior of biomembranes (see Sections 2 and 3). The choice of one approach instead of another may depend on how detailed one needs to describe a system. In general, the more details are needed the smaller are the system sizes that one can consider, and the shorter the time scale that can be sampled. For example, the CG model approach is more advantageous than the all-atom model approach when one wants to investigate biomembrane processes which involve the cooperative rearrangements of large number of molecules, and which are characterized by long time scales. The drawback of this approach is due to the use of ‘beads’ to effectively describe the system: by modelling a system with a set of beads, one can describe its overall three-dimensional structure, but not its microscopic details. However, it is worth stressing that, to model a systems does not mean to try to reproduce all the possible details (i.e., consider all possible degrees of freedom involved), but rather to focus one some of them and approximate others, depending on the phenomena that one wants to understand. This a priori choice of approximation will naturally imply that a model study will, alone, not be able provide a full understanding of the biophysical behavior of a chosen investigated system. Nevertheless, this limitation can be overcome, if one or more theoretical approaches are used together with experimental investigations. If used in this intermethodological way, results from theoretical studies can become really useful for providing a framework of interpretation for the experimental data and for revealing information not otherwise accessible; also, they may constitute a source of inspiration for future experiments. Acknowledgements MMS is grateful to Maddalena Venturoli (FR), and to Berend Smit (NL, FR), Marcus A. Hemminga (NL), John H. Ipsen (DK) and J. Antoinette Killian (NL), for numerous stimulating discussions; and wishes to thank the Biochemistry and Nutrition Group and the Center for Biological Sequence analysis, both at Biocentrum (The Technical University of Denmark, Kgs. Lyngby, Denmark), and CECAM (Lyon, France) for hospitality. SM is supported by ND EPSCoR through grant #EPS0132289 and acknowledges discussions with Avinoam Ben-Shaul, Ales Iglic, Anne Hinderliter, including their research group members. The authors thank the European Science Foundation program COST D22 (“Lipid–protein Interactions”) for support. They would like to express their gratitude to the chairman of D22, John Findlay, and to the organizer of the Dubrovnic conference, Greta Pifat. M.M. Sperotto et al. / Chemistry and Physics of Lipids 141 (2006) 2–29 References Abney, J.R., Owicki, J.C., 1985. Theories of protein–lipid and protein–protein interactions in membranes. In: Watts, A., De Pont, J. (Eds.), Progress in Protein–Lipid Interactions. Elsevier Science Publishers, New York. Bass, R.B., Strop, P., Barclay, M., Rees, D.C., 2002. Crystal structure of Escherichia coli MscS, a voltage-modulated and mechanosensitive channel. Science 298, 1582–1587. Baumgaertner, A., 1996. Insertion and hairpin formation of membrane proteins. Biophys. J. 71, 1248–1255. Baumgaertner, A., 2006. Biomolecular machines. In: Rieth, M., Schommers, W. (Eds.), Handbook of Theoretical and Computational Nanotechnology. American Scientific Publishers. Bechinger, B., 1997. Structure and dynamics of the M13 coat signal sequence in membranes by multidimensional high-resolution and solid-state NMR spectroscopy. Proteins: Struct. Funct. Genet. 27, 481–492. Belohorcová, K., Davis, J.H., Woolf, T.B., Roux, B., 1997. Structure and dynamics of an amphiphilic peptide in a bilayer: a molecular dynamics study. Biophys. J. 73, 3039–3055. Ben-Shaul, A., 1995. Molecular theory of chain packing, elasticity and lipid protein interaction in lipid bilayers. In: Lipowsky, R., Sackmann, E. (Eds.), Structure and Dynamics of Membranes. Elsevier, Amsterdam. Bernéche, S., Nina, N., Roux, B., 1998. Molecular dynamics simulation of melittin in a POPC bilayer membrane. Biophys. J. 75, 1603–1618. Biggin, P.C., Sansom, M.S.P., 2003. Mechanosensitive channels: stress relief. Curr. Biol. 13, 183–185. Binder, W.H., Barragan, V., Menger, F.M., 2003. Domains and rafts in lipid membranes. Angew. Chem. Int. Ed. 42, 5802–5827. Bohinc, K., Kralj-Iglič, V., May, S., 2003. Interaction between two cylindrical inclusions in a symmetric lipid-bilayer. J. Chem. Phys. 119, 7435–7444. Brannigan, C., Philips, P.F., Brown, F.L.H., 2005. Flexible lipid bilayers in implicit solvent. Phys. Rev. E 72, 011915. Braun, R., Engelman, D.M., Schulten, K., 2004. Molecular dynamics simulations of micelle formation around dimeric glycophorin A transmembrane helices. Biophys. J. 87, 754–763. Bretscher, M.S., Munro, S., 1993. Cholesterol and the Golgi apparatus. Science 261, 1280–1281. Bryl, K., Yoshihara, K., 2001. The role of retinal in the long-range protein–lipid interactions in bacteriorhodopsin– phosphatidylcholine vesicles. Eur. Biophys. J. 29, 628–640. Chang, R., Ayton, G.S., Voth, G.A., 2005. Multiscale coupling of mesoscopic- and atomistic-level lipid bilayer simulations. J. Chem. Phys. 122, 244716. Chiu, S.-W., Subramaniam, S., Jakobsson, E., 1999. Simulation study of a gramicidin/lipid bilayer system in excess water and lipid. I. Structure of the molecular complex. Biophys. J. 76, 1929–1938. Cooke, I.-R., Kremer, K., Deserno, M., 2005. Tunable generic model for fluid bilayer membranes. Phys. Rev. E 72, 1–4. Dan, N., Pincus, P., Safran, S.A., 1993. Membrane-induced interactions between inclusions. Langmuir 9, 2768–2771. de Planque, M.R.R., Killian, J.A., 2003. Protein–lipid interactions studied with designed transmembrane peptides: role of hydrophobic matching and interfacial anchoring. Mol. Membrane Biol. 20, 271–284. Deisenhofer, J., Epp, O., Miki, K., Huber, R., Michael, H., 1985. Structure of the protein subunits in the photosynthetic reaction center of Rhodopseudomonas viridis at 3 Å resolution. Nature 318, 618–624. 25 Dumas, F., Sperotto, M.M., Lebrum, M.C., Tocanne, J.-F., Mouritsen, O.G., 1997. Molecular sorting of lipids by bacteriorhodopsin in dilauroylphosphatidylcholine/distearoyl-phosphatidylcholine lipid bilayers. Biophys. J. 73, 1940–1953. Dumas, F., Lebrum, M.C., Tocanne, J.-F., 1999. Is the protein/lipid hydrophobic matching principle relevant to membrane organization and functions? FEBS Lett. 458, 271–277. Edholm, O., Jähnig, F., 1988. The structure of membrane-spanning polypeptide studied by molecular dynamics. Biophys. Chem. 30, 279–292. Edwards, M.D., Li, Y., Kim, S., Miller, S., Bartlett, W., Black, S., Dennison, S., Iscla, I., Blount, B., Bowie, J.U., Booth, I.R., 2005. Pivotal role of the glycine-rich TM3 helix in gating the MscS mechanosensitive channel. Nat. Struct. Mol. Biol. 12, 113– 119. Efremov, R.G., Nolde, D.E., Vergoten, G., Arseniev, A.S., 1999a. A solvent model for simulations of peptides in bilayers. I. Membranepromoting alpha-helix formation. Biophys. J. 76, 2448–2459. Efremov, R.G., Nolde, D.E., Vergoten, G., Arseniev, A.S., 1999b. A solvent model for simulations of peptides in bilayers. II. Membrane-spanning alpha-helices. Biophys. J. 76, 2460–2471. Efremov, R.G., Volynsky, P.E., Nolde, D.E., Dubovskii, P.V., Arseniev, A.S., 2002. Interaction of cardiotoxins with membranes: a molecular modeling study. Biophys. J. 82, 144–153. Elmore, D.E., Dougherty, D.A., 2001. Molecular dynamics simulations of wild-type and mutant forms of the Mycobacterium tuberculosis MscL channel. Biophys. J. 81, 1345–1359. Engelman, D.M., Steitz, T.A., 1981. The spontaneous insertion of proteins into and across membranes: the helical hairpin hypothesis. Cell 23, 411–422. Epand, R.M., Shai, Y., Segrest, J.P., Anantharamaiah, G.M., 1995. Mechanisms for the modulation of membrane bilayer properties by amphipathic helical peptides. Biopolymers (Peptide Sci.) 37, 319–338. Epand, R.M., 1998. Lipid polymorphism and lipid–protein interactions. Biochim. Biophys. Acta 1376, 353–368. Epand, R.M., 2004. Do proteins facilitate the formation of cholesterolrich domains? Biochim. Biophys. Acta: Biomembranes 1666, 227–238. Fahsel, S., Pospiech, E.-M., Zein, M., Hazlet, T.L., Gratton, E., Winter, R., 2002. Modulation of concentration fluctuations in phaseseparated lipid membranes by polypeptide insertion. Biophys. J. 83, 334–344. Fantini, J., Garmy, N., Mahfoud, R., Yahi, N., 2002. Lipid rafts: structure, function and role in HIV, Alzheimer’s and prion diseases. In: Expert Reviews in Molecular Medicine. Cambridge University Press, UK, pp. 1–22. Fattal, D.R., Ben-Shaul, A., 1993. A molecular model for lipid–protein interactions in membranes: the role of hydrophobic mismatch. Biophys. J. 65, 1795–1809. Feller, S.E., Pastor, R.W., 1996. On simulating lipid bilayers with an applied surface tension: periodic boundary conditions and undulations. Biophys. J. 71, 1350–1355. Feller, S.E., Pastor, R.W., 1999. Constant surface tension simulations of lipid bilayers: the sensitivity of surface areas and compressibilities. J. Chem. Phys. 111, 1281–1287. Fernandes, F., Loura, L.M.S., Prieto, M., Koehorst, R., Spruijt, R.B., Hemminga, M.A., 2003. Dependence of M13 major coat protein oligomerization and lateral segregation on bilayer composition. Biophys. J. 85, 2430–2441. Forrest, L.R., Samson, M.S.P., 2000. Membrane simulations: bigger and better? Curr. Opin. Struct. Biol. 10, 174–181. 26 M.M. Sperotto et al. / Chemistry and Physics of Lipids 141 (2006) 2–29 Foster, D.L., Boublik, M., Kaback, H.R., 1983. Structure of the lac carrier protein of Escherichia coli. J. Biol. Chem. 258, 31–34. Fry, M., Green, D.E., 1981. Cardiolipin requirement for electron transfer in complexes I and III of the mitochondrial respiratory chain. J. Biol. Chem. 256, 1874–1880. Fu, D., Libson, A., Miercke, L.J.W., Weitzman, C., Nollert, P., Krucinski, J., Stround, R.M., 2000. Structure of a glycerol-conducting channel and the basis for its selectivity. Science 290, 481–486. Fyfe, P.K., McAuley, K.E., Roszak, A.W., Isaacs, N.W., Cogdell, R.J., Jones, M.R., 2001. Probing the interface between membrane proteins and membrane lipids by X-ray crystallography. Trends Biochem. Sci. 26, 106–112. Gambhir, A., Hangyás-Mihályné, G., Zaitseva, I., Cafiso, D.S., Wang, J.Y., Murray, D., Pentyala, S.N., Smith, S.O., McLaughlin, S., 2004. Electrostatic sequestration of PIP2 on phospholipid membranes by basic/aromatic regions of proteins. Biophys. J. 86, 2188–2207. Garidel, P., Blume, A., 2000. Miscibility of phosphatidylethanolamine–phosphatidylglycerol mixtures as a function of pH and acyl chain length. Eur. Biophys. J. 28, 629–638. Gelbart, W.M., Bruinsma, R., 1997. Compositional–mechanical instability of interacting mixed lipid membranes. Phys. Rev. E 55, 831–835. Gil, T., Ipsen, J.H., Mouritsen, O.G., Sabra, M.C., Sperotto, M.M., Zuckermann, M.J., 1998. Theoretical analysis of protein organization in lipid membranes. Biochim. Biophys. Acta 1376, 245–266. Glaubitz, C., Grobner, G., Watts, A., 2000. Structural and orientational information of the membrane embedded M13 coat protein by 13 CMAS NMR spectroscopy. Biochim. Biophys. Acta 1463, 151–161. Goetz, R., Lipowsky, R., 1998. Computer simulations of bilayer membranes: self-assembly and interfacial tension. J. Chem. Phys. 108, 7397–7409. Goetz, R., Gompper, G., Lipowsky, R., 1998. Mobility and elasticity of self-assembled membranes. Phys. Rev. Lett. 82, 221–224. Golebiewska, U., Gambhir, A., Hangyás-Mihályné, G., Zaitseva, I., Radler, J., McLaughlin, S., 2005. Membrane-bound basic peptides sequester the multivalent lipid PIP2 but not the monovalent lipid PS, preprint. Gompper, G., Kroll, D.M., 1997. Network models of fluid, hexatic and polymerized membranes. J. Phys. Condens. Matter 9, 8795–8834. Goodyear, D.J., Sharpe, S., Grant, W.M., 2005. Molecular dynamics simulation of trans-membrane polypeptide orientational fluctuations. Biophys. J. 88, 105–117. Groot, R.D., 2000. Mesoscopic simulation of polymer–surfactant aggregation. Langmuir 16, 7493–7502. Groot, R.D., Rabone, K.L., 2001. Mesoscopic simulation of cell membrane damage, morphology change and rupture by nonionic surfactant. Biophys. J. 81, 725–736. Grosberg, A.Y., Nguyen, T.T., Shklovskii, B.I., 2002. Colloquium: the physics of charge inversion in chemical and biological systems. Mod. Rev. Phys. 74, 329–345. Gullingsrud, J., Kosztin, D., Schulten, K., 2001. Structural determination of MscL gating studied by MD simulations. Biophys. J. 80, 2074–2081. Gullingsrud, J., Schulten, K., 2003. Gating of MscL studied by steered molecular dynamics. Biophys. J. 85, 2087–2099. Harroun, T.A., Heller, W.T., Wiess, T.M., Yang, L., Huang, H.W., 1999a. Experimental evidence for hydrophobic matching and membrane-mediated interactions in lipid bilayers containing gramicidin. Biophys. J. 76, 937–945. Harzer, U., Bechinger, B., 2000. Alignment of lysine-anchored membrane peptides under conditions of hydrophobic mismatch: a CD, 15 N and 31 P solid-state NMR spectroscopy investigation. Biochem- istry 39, 13106–13114. Henderson, R., Unwin, P.N.T., 1975. Three-dimensional model of purple membrane obtained by electron microscopy. Nature 257, 28–32. Harries, D., May, S., Gelbart, W.M., Ben-Shaul, A., 1998. Structure, stability and thermodynamics of lamellar DNA–lipid complexes. Biophys. J. 75, 159–173. Harroun, T.A., Heller, W.T., Weiss, T.M., Yang, L., Huang, H.W., 1999b. Theoretical analysis of hydrophobic matching and membrane-mediated interactions in lipid bilayers containing gramicidin. Biophys. J. 76, 3176–3185. Helfrich, W., 1973. Elastic properties of lipid bilayers: theory and possible experiments. Z. Naturforsch. 28, 693–703. Hesketh, T.R., Smith, G.A., Houslay, M.D., McGill, K.A., Birdsall, N.J.M., Metcalfe, J.C., Warren, G.B., 1976. Annular lipids determine the ATPase activity of a calcium transport protein complexed with dipalmitoyllecithin. Biochemistry 15, 4145–4151. Hinderliter, A., Biltonen, R.L., Almeida, P.F., 2004. Lipid modulation of protein-induced membrane domains as a mechanism for controlling signal transduction. Biochemistry 43, 7102–7110. Hinderliter, A., Almeida, P.F., Creutz, C.E., Biltonen, R.L., 2001. Domain formation in a fluid mixed lipid bilayer modulated through binding of the C2 protein motif. Biochemistry 40, 4181–4191. Hoogerbrugge, P.J., Koelman, J.M.V.A., 1992. Simulating microscopic hydrodynamics phenomena with dissipative particle dynamics. Europhys. Lett. 19, 155–160. Huang, H.W., 2000. Action of antimicrobial peptides: two-state model. Biochemistry 39, 8347–8352. Huang, H.W., Chen, F.Y., Lee, M.T., 2004. Molecular mechanism of peptide-induced pores in membranes. Phys. Rev. Lett. 92, 198304. Im, W., Brooks, C.L., 2005. Chemical theory and computation special feature: interfacial folding and membrane insertion of designed peptides studied by molecular dynamics simulations. Proc. Natl. Acad. Sci. U.S.A. 102, 6771–6776. In’t Veld, G., Driessen, A.J.M., Op den Kamp, J.A.F., Konings, W.N., 1991. Hydrophobic membrane thickness and lipid–protein interactions of the leucine transport system of Lactococcus lactis. Biochim. Biophys. Acta 1065, 203–212. Iwata, S., Ostermeier, C., Ludwig, B., Michel, H., 1995. Structure at 2.8 Å resolution of cytochrome c oxidase from Paracoccus denitrificans. Nature 376, 660–669. Jähnig, F., 1996. What is the surface tension of a lipid membrane? Biophys. J. 71, 1348–1349. Jensen, M.Ø., Mouritsen, O.G., 2004. Lipids do influence protein function—the hydrophobic matching hypothesis revised. Biochim. Biophys. Acta Biomembranes 1666, 205–226. Jensen, M.Ø., Tajkhorshid, E., Schulten, K., 2001. The mechanism of glycerol conduction in aquaglyceroporins. Structure 9, 1083–1093. Johansson, A., Smith, G.A., Metcalfe, J., 1981. The effect of bilayer thickness on the activity of (Na+ –K+ )-ATPase. Biochim. Biophys. Acta 641, 416–421. Jost, P.C., Hayes Griffith, O., 1980. The lipid–protein interface in biological membranes. Ann. NY Acad. Sci. U.S.A. 348, 391– 405. Jost, P.C., Hayes Griffith, O., Capaldi, R.A., Vanderkooi, G., 1973. Evidence for boundary lipids in membranes. Proc. Natl. Acad. Sci. U.S.A. 70, 480–484. Jury, S., Bladon, P., Cates, M., Krishna, S., Hagen, M., Ruddock, N., Warren, P., 1999. Simulation of amphiphilic mesophases using dissipative particle dynamics. Phys. Chem. Chem. Phys. 1, 2051–2056. M.M. Sperotto et al. / Chemistry and Physics of Lipids 141 (2006) 2–29 Kandasamy, S.K., Larson, R.G., 2004. Binding and insertion of ␣helical anti-microbial peptides in POPC bilayers studied by molecular dynamics simulations. Chem. Phys. Lipids 132, 113–132. Kessel, A., Ben-Tal, N., May, S., 2001. Interactions of cholesterol with lipid bilayers: the preferred configuration and fluctuations. Biophys. J. 81, 643–658. Kik, R.A., Leermakers, F.A.M., Kleijn, J.M., 2005. Molecular modeling of lipid bilayers and the effect of protein-like inclusions. Phys. Chem. Chem. Phys. 7, 1996–2005. Killian, J.A., 1992. Gramicidin and gramicidin–lipid interactions. Biochim. Biophys. Acta 1113, 391–425. Killian, J.A., 1998. Hydrophobic mismatch between proteins and lipids in membranes. Biochim. Biophys. Acta 1376, 401–416. Koehorst, R.B.M., Spruijt, R.B., Vergeldt, F.J., Hemminga, M.A., 2004. Lipid bilayer topology of the transmembrane ␣-helix of M13 major coat protein and bilayer polarity profile by site-directed fluorescence spectroscopy. Biophys. J. 87, 1445–1455. Koenig, B.W., Ferretti, J.A., Gawrisch, K., 1999. Site-specific deuterium order parameters and membrane-bound behavior of a peptide fragment from the intracellular domain of HIV-1 gp41. Biochemistry 38, 6327–6334. Kol, M.A., van Delen, A., de Kroon, A.I.P.M., de Kruijff, B., 2003. Translocation of phospholipids is facilitated by a subset of membrane-spanning proteins of the bacterial cytoplasmic membrane. J. Biol. Chem. 278, 24586–24593. König, S., Sackmann, E., 1996. Molecular and collective dynamics of lipid bilayers. Curr. Opin. Colloid Interf. Sci. 1, 78–82. Kranenburg, M., Venturoli, M., Smit, B., 2003a. Molecular simulations of mesoscopic bilayer phases. Phys. Rev. E 67, 060901(R). Kranenburg, M., Venturoli, M., Smit, B., 2003b. Phase behaviour and induced interdigitation in bilayer studied with dissipative particle dynamics. J. Phys. Chem. B 107, 11491–11501. Kranenburg, M., Smit, B., 2004. Simulating the effects of alcohol on the structure of a membrane. FEBS Lett. 568, 15–18. Kranenburg, M., Nicolas, J.P., Smit, B., 2004a. Comparison of mesoscopic phospholipid–water models. Phys. Chem. Chem. Phys. 6, 4142–4151. Kranenburg, M., Vlaar, M., Smit, B., 2004b. Simulating induced interdigitation in membranes. Biophys. J. 87, 1596–1605. Lee, A.G., 1998. How lipids interact with an intrinsic membrane protein: the case of the calcium pump. Biochim. Biophys. Acta 1376, 381–390. Lee, A.G., 2003. Lipid–protein interactions in biological membranes: a structural perspective. Biochim. Biophys. Acta 1612, 1–40. Lehtonen, J.Y.A., Kinnunen, P.K.J., 1997. Evidence for phospholipid microdomain formation in liquid crystalline liposomes reconstituted with Escherichia coli lactose permease. Biophys. J. 72, 1247–1257. Lin, J.H., Baumgaertner, A., 2000. Stability of a melittin pore in a lipid bilayer: a molecular dynamics study. Biophys. J. 78, 1714– 1724. Lin, G.R., McCormick, J.I., Dhe-Paganon, S., Silvius, J.R., Johnstone, R.M., 1990. Role of specific acidic lipids on the reconstitution of sodium-dependent amino acid transport in proteoliposomes derived from Ehrlich cell plasma membranes. Biochemistry 29, 4575–4581. Lin, J.H., Baumgaertner, A., 2005. Extraction of a melittin pore from a lipid bilayer: a molecular dynamics study. J. Am. Chem. Soc., submitted for publication. Lopez, C.F., Nielsen, S.O., Moore, P.B., Klein, M.L., 2004. Understanding nature’s design for a nanosyringe. Proc. Natl. Acad. Sci. U.S.A. 101, 4431–4434. 27 Lopez, C.F., Nielsen, S.O., Ensing, B., Moore, P.B., Klein, M.L., 2005. Structure and dynamics of model pore insertion into a membrane. Biophys. J. 88, 3083–3094. Ludtke, S., He, K., Huang, H., 1995. Membrane thinning caused by magainin 2. Biochemistry 34, 16764–16769. Lundbæk, J.A., Andersen, O.S., 1999. Spring constants for channel induced lipid bilayer deformations. Estimates using gramicidin channels. Biophys. J. 76, 889–895. MacKenzie, K.R., Prestegard, J.H., Engelmann, D.M., 1997. A transmembrane helix dimer: structure and implications. Science 276, 131–133. Mall, S., Broadbridge, R., Sharma, R.P., East, J.M., Lee, A.G., 2001. Self-association of model transmembrane helix is modulated by lipid structure. Biochemistry 40, 12379–12386. Marrink, S.J., Mark, A.E., 2001. Effect of undulations on surface tension in simulated bilayers. J. Phys. Chem. B. 105, 6122– 6127. Marsh, D., Horvath, L.I., 1998. Structure, dynamics and composition of the lipid–protein interface. Perspectives from spin-labelling. Biochim. Biophys. Acta 1376, 267–296. May, S., 2000. Theories on structural perturbations of lipid bilayers. Curr. Opin. Colloid Interf. Sci. 5, 244–249. May, S., Ben-Shaul, A., 2000. A molecular model for lipid-mediated interaction between proteins in membranes. Phys. Chem. Chem. Phys. 2, 4494–4502. May, S., Harries, D., Ben-Shaul, A., 2000. Lipid demixing and protein–protein interactions in the adsorption of charged proteins on mixed membranes. Biophys. J. 79, 1747–1760. May, S., Harries, D., Ben-Shaul, A., 2002. Macroion-induced compositional instability of binary fluid membranes. Phys. Rev. Lett. 89, 268102–268105. May, S., 2002. Membrane perturbations induced by integral proteins: role of conformational restrictions of the lipid chains. Langmuir 18, 6356–6364. May, S., Kozlovsky, Y., Ben-Shaul, A., Kozlov, M.M., 2004. Tilt modulus of a lipid monolayer. Eur. Phys. J. E 14, 299–308. Mbamala, E.C., Ben-Shaul, A., May, S., 2005. Domain formation induced by the adsorption of charged proteins on mixed lipid membranes. Biophys. J. 88, 1702–1714. McAuley, K.E., Fyfe, P.K., Ridge, J.P., Isaacs, N.W., Cogdell, R.J., Jones, M.R., 1999. Structural details of an interaction between cardiolipin and an integral membrane protein. Proc. Natl. Acad. Sci. U.S.A. 96, 14706–14711. McIntosh, T.J., Vidal, A., Simon, S.A., 2003. Sorting of lipids and transmembrane peptides between detergent-soluble bilayers and detergent-resistant rafts. Biophys. J. 85, 1656–1666. Milik, M., Skolnick, J., 1995. A Monte Carlo model of fd Pf1 proteins in lipid membranes. Biophys. J. 69, 1382–1386. Montecucco, C., Smith, G.A., Dabbeni-Sala, F., Johansson, A., Galante, Y.M., Bisson, R., 1982. Bilayer thickness and enzymatic activity in the mitochondrial cytochrome c oxidase and ATPase complex. FEBS Lett. 144, 145–148. Morein, S., Killian, J.A., Sperotto, M.M., 2002. Characterization of the thermotropic behaviour and lateral organization of lipid–peptide mixtures by a combined experimental and theoretical approach: effects of hydrophobic mismatch and role of flanking residues. Biophys. J. 82, 1405–1417. Mouritsen, M.M., 1998. Self-assembly and organization of lipid–protein membranes. Curr. Opin. Colloid Interf. Sci. 3, 78–87. Mouritsen, O.G., Blom, M., 1984. Mattress model of lipid–protein interactions in membranes. Biophys. J. 46, 141–153. 28 M.M. Sperotto et al. / Chemistry and Physics of Lipids 141 (2006) 2–29 Mouritsen, O.G., Sperotto, M.M., 1993. Thermodynamics of lipid–protein interactions in lipid membranes. In: Jackson, M. (Ed.), Thermodynamics of Membrane Receptors and Channels. CRC Press, Inc, Boca Raton, FL, pp. 127–181. Mukherijee, S., Maxfield, F.R., 2004. Membrane domains. Ann. Rev. Cell Dev. Biol. 20, 839–866. Munro, S., 1995. An investigation of the role of transmembrane domain in Golgi protein retention. EMBO J. 14, 4695–4704. Munro, S., 1998. Localization of proteins to the Golgi apparatus. Trends Cell Biol. 8, 11–15. Nielsen, S.O., Ensing, B., Ortiz, V., Moore, P.B., Klein, M.L., 2005a. Lipid bilayer perturbations around a transmembrane nanotube: a coarse grain molecular dynamics study. Biophys. J. 88, 3822– 3828. Nielsen, S.O., Lopez, C.F., Ivanov, I., Moore, P.B., Shelley, J.C., Klein, M.L., 2004. Transmembrane peptide-induced lipid sorting and mechanism of La -to-inverted phase transition using coarse-grain molecular dynamics. Biophys. J. 87, 2107–2115. Nielsen, C., Goulian, M., Andersen, O.S., 1998. Energetics of inclusion-induced bilayer deformations. Biophys. J. 74, 1966–1983. Nielsen, S.O., Ensing, B., Ortiz, V., Moore, P.M., Klein, M.L., 2005b. Lipid bilayer perturbations around a transmembrane nanotube: a coarse grain molecular dynamics study. Biophys. J. 88, 3822–3828. Nymeyer, H., Gnanakaran, S., Garcia, A.E., 2004. Atomic simulations of protein folding, using the replica exchange algorithm. Meth. Enzymol. 383, 119–149. Nymeyer, H., Woolf, T.B., Garcia, A.E., 2005. Folding is not required for bilayer insertion: replica exchange simulations of an ␣-helical peptide with an explicit lipid bilayer. Proteins 59, 783–790. Özdirekcan, S., Rijkers, D.T.S., Liskamp, R.M.J., Killian, J.A., 2005. Influence of flanking residues on tilt and rotation angles of transmembrane peptides in lipid bilayers. A solid-state 2 H NMR study. Biochemistry 44, 1004–1012. Pelham, H.R.B., Munro, S., 1993. Sorting of membrane–proteins in the secretory pathway. Cell 75, 603–605. Perozo, E., Rees, D.C., 2003. Structure and mechanism in prokaryotic mechanosensitive channels. Curr. Opin. Struct. Biol. 13, 432–442. Perozo, E., Kloda, A., Cortes, D., Martinac, B., 2002. Physical principles underlying the transduction of bilayer deformation forces during mechanosensitive channel gating. Nat. Struct. Biol. 9, 696–703. Peterson, S.A., Klabunde, T., Lashuel, H.A., Purkey, H., Sacchettini, J.C., Kelly, J.W., 1998. Inhibiting transthyretin conformational changes that lead to amyloid fibril formation. Proc. Natl. Acad. Sci. U.S.A. 95, 12956–12960. Petrache, H.I., Grossfield, A., MacKenzie, K.R., Engelman, D.M., Woolf, T., 2000. Modulation of glycophorin A transmembrane helix interactions by lipid bilayers: molecular dynamics calculations. J. Mol. Biol. 302, 727–746. Petrache, H.I., Zuckerman, D.M., Sachs, J.N., Killian, J.A., Koeppe II, R.E., Woolf, T., 2002. Hydrophobic matching by molecular dynamics simulations. Langmuir 18, 1340–1351. Piknová, B., Pérochon, E., Tocanne, J.-F., 1993. Hydrophobic mismatch and long-range protein/lipid interactions in bacteriorhodopsin/phosphatidylcholine vesicles. Eur. J. Biochem. 218, 385–396. Popot, J.-L., Engelman, D.M., 2000. Helical membrane protein folding, stability, and evolution. Ann. Rev. Biochem. 69, 881–922. Ramakrishnan, M., Qu, J., Pocanschi, C.L., Kleinschmidt, J.H., Marsh, D., 2005. Orientation of -barrel protein OmpA and PhuA in lipid membranes. Chain length dependence from infrared dichroism. Biochemistry 44, 3515–3523. Rehorek, M., Dencher, N.A., Heyn, M.P., 1985. Long-range lipid– protein interactions. Evidence from time-resolved fluorescence depolarization and energy-transfer experiments with bacteriorhodopsin-dimyristoylphosphatidylcholine vesicles. Biochemistry 24, 5980–5988. Ridder, A.N.J.A., Spelbrink, E.R.J., Demmers, J.A.A., Rijkers, D.T.S., Liskamp, R.M.J., Brunner, J., Heck, A.J.R., de Kruijff, B., Killian, J.A., 2004. Photo-crosslinking analysis of preferential interactions between a transmembrane peptide and matching lipids. Biochemistry 43, 4482–4489. Roux, B., Karplus, M., 1994. Molecular dynamics simulations of the gramicidin channel. Annu. Rev. Biophys. Biomol. Struct. 23, 731–761. Sackmann, E., 1984. Physical basis of trigger processes and membrane structure. In: Chapman, D. (Ed.), Biological Membranes, vol. 5. Academic Press, London, pp. 105–143. Sackmann, E., 1995. Biological membranes architecture and function. In: Lipowsky, R., Sackmann, E. (Eds.), Structure and Dynamics of Membranes. Elsevier, Amsterdam, pp. 1–65. Sanders, J.C., van Nuland, N.A., Edholm, O., Hemminga, M.A., 1991. Conformation and aggregation of M13 coat protein studied by molecular dynamics. Biophys. Chem. 41, 193–202. Sankararamakrishnnan, R., Weinstein, H., 2000. MD simulations predict a tilted orientation for the helical region of dynorphon A(1–17) in DPPC bilayers. Biophys. J. 79, 2331–2344. Schnell, D.J., Hebert, D.N., 2003. Protein translocons: multifunctional mediators of protein translocation across membranes. Cell 112, 491–505. Sekijima, M., Satoshi Yamasaki, S., Kaneko, K., Akiyama, Y., 2003. Molecular dynamics simulation of dimeric and monomeric forms of human prion protein: insight into dynamics and properties. Biophys. J. 85, 1176–1185. Shai, Y., 1999. Mechanism of the binding, insertion and destabilization of phospholipid bilayer membranes by alpha-helical antimicrobial and cell non-selective membrane-lytic peptides. Biochim. Biophys. Acta 1462, 55–70. Sharpe, S., Barber, K.R., Grant, C.W.M., Goodyear, D., Morrow, M.R., 2002. Organization of model helical peptides in lipid bilayers: insight into the behaviour of single-span protein transmembrane domains. Biophys. J. 83, 345–358. Shen, L., Bassolino, D., Stouch, T., 1997. Transmembrane helix structure, dynamics, and interactions: multi-nanoseconds molecular dynamics simulations. Biophys. J. 73, 3–20. Shelley, J.C., Shelley, M.Y., Reeder, R.C., Badyopadhyay, S., Klein, M.L., 2001a. A coarse grain model for phospholipid simulations. J. Phys. Chem. B 105, 4464–4470. Shelley, J.C., Shelley, M.Y., Reeder, R.C., Badyopadhyay, S., Moore, P.B., Klein, M.L., 2001b. Simulations of phospholipids using a coarse grain model. J. Phys. Chem. B 105, 9785– 9792. Shepherd, C.M., Vogel, H.J., Tieleman, D.P., 2003. Interaction of the designed antimicrobial peptide MB21 and truncated dermaseptin S3 with lipid bilayers: molecular dynamics simulations. Biochem. J. 370, 233–243. Shillcock, J.C., Lipowsky, R., 2002. Equilibrium structure and lateral stress distribution of amphiphilic bilayers from dissipative particle dynamics. J. Phys. Chem. 117, 5048–5061. Siegel, D.P., 1999. The modified stalk mechanism of lamellar/inverted phase transitions and its implication for membrane fusion. Biophys. J. 76, 291–313. Simons, K., Ikonen, E., 1997. Functional rafts in cell membranes. Nature 387, 569–572. M.M. Sperotto et al. / Chemistry and Physics of Lipids 141 (2006) 2–29 Sintes, T., Baumgärtner, A., 1997. Protein attraction in membranes induced by lipid fluctuations. Biophys. J. 73, 2251–2259. Sintes, T., Baumgaertner, A., 1998. Membrane mediated protein attraction. A Monte-Carlo study. Physica A 249, 571–575. Sotomayor, M., Schulten, K., 2004. Molecular dynamics study of gating in the mechanosensitive channel of small conductance MscS. Biophys. J. 87, 3050–3060. Sperotto, M.M., 1997. A theoretical model for the association of amphiphilic transmembrane peptides in lipid bilayers. Eur. Biophys. J. 26, 405–416. Sperotto, M.M., Mouritsen, 1991. Monte Carlo simulation studies of lipid order parameter profiles near integral membrane proteins. Biophys. J. 59, 261–270. Stopar, D., Spruijt, R.B., Wolfs, C.J.A.M., Hemminga, M.A., 2003. Protein–lipid interactions of bacteriophage M13 major coat protein. Biochim. Biophys. Acta 1611, 5–15. Strandberg, E., Özdirekcan, S., Rijkers, D.T.S., van der Wei, P.C.A., Koeppe Jr., R.E., Liskamp, R.M.J., Killian, J.A., 2004. Tilt angles of transmembrane model peptides in oriented and non-oriented lipid bilayers as determined by 2 H solid state NMR. Biophys. J. 86, 3709–3721. Suchyna, T.M., Tape, S.E., Koeppe, R.E., Andersen, O.S., Sachs, F., Gottlieb, P.A., 2004. Bilayer-dependent inhibition of mechanosensitive channels by neuroactive peptide enantiomers. Nature 430, 235–240. Sugita, Y., Okamoto, Y., 1999. Replica-exchange molecular dynamics method for protein folding. Chem. Phys. Lett. 314, 141–151. Sukharov, S., Betanzos, M., Chiang, C.S., Guy, H.R., 2001. Gating mechanism of the large mechanosensitive channel MscL. Nature 409, 720–724. Swendsen, R., Wang, J., 1986. Replica Monte Carlo simulation of spin-glasses. Phys. Rev. Lett. 57, 2607–2609. Thomson, T.E., Sankaram, M.B., Biltonen, R.B., Marsh, D., Vaz, W.L.C., 1995. Effect of domain structure on in-plane reactions and interactions. Mol. Membrane Biol. 12, 157–162. Tieleman, D.P., Marrink, S.J., Berendsen, H.J.C., 1997. A computer perspective of membranes: molecular dynamics studies of lipid bilayer systems. Biophys. Biochim. Acta 1331, 235–270. Tieleman, D.P., Sansom, M.S.P., Berendsen, H.J.C., 1999a. Alamethicin helices in a bilayer and in solution: molecular dynamics simulations. Biophys. J. 76, 40–49. Tieleman, D.P., Berendsen, H.J.C., Sansom, M.S.P., 1999b. An alamethicin channel in a lipid bilayer: molecular dynamics simulations. Biophys. J. 76, 1757–1769. Tieleman, D.P., Berendsen, H.J.C., Sansom, M.S.P., 2001. Voltagedependent insertion of alamethicin at phospholipid/water and octane/water interfaces. Biophys. J. 80, 331–346. Tocanne, J.-F., 1992. Detection of lipid domains in biological membranes. Comm. Mol. Cell. Biophys. 8, 53–72. Tocanne, J.-F., Cézanne, L., Lopaz, A., Piknová, B., Schram, V., Turnier, J.F., Welby, M., 1994. Lipid domains and lipid/protein interactions in biological membranes. Chem. Phys. Lipids 73, 139–159. Tossi, A., Sandri, L., Giangaspero, A., 2000. Amphipathic, alphahelical antimicrobial peptides. Biopolymers 55, 4–30. 29 Urbanc, B., Cruz, L., Ding, F., Sammond, D., Khare, S., Buldyrev, S.V., Stanley, H.E., Dokholyan, N.V., 2004. Molecular dynamics simulation of amyloid  dimer formation. Biophys. J. 87, 2310– 2321. Valiyaveetil, F.L., Zhou, Y., MacKinnon, R., 2002. Lipids in the structure, folding, and function of the KcsA K+ channel. Biochemistry 41, 10771–10777. van der Wei, P.C.A., Strandberg, E., Killian, J.A., Koeppe II, R.E., 2002. Geometry and intrinsic tilt of a tryptophan-anchored transmembrane ␣-helix determined by 2 H NMR. Biophys. J. 83, 1479–1488. van der Eerden, J.P.J.M., Snel, M.M.E., Makkinje, J., van Dijk, A.D.J., Rinia, H., 2002. Striped phases in thin layers: simulation and observation. J. Crystal Growth 237, 111–115. Vemuri, R., Philipson, K.D., 1989. Influence of sterols and phospholipids on sarcolemmal and sarcoplasmic reticular cation transporters. J. Biol. Chem. 264, 8680–8685. Venturoli, M., Smit, B., 1999. Simulating the self-assembly of model membranes. Phys. Chem. Comm. 10. Venturoli, M., Smit, B., Sperotto, M.M., 2005. Simulation studies of protein-induced bilayer deformations, and lipid-induced protein tilting, on a mesoscopic model for lipid bilayers with embedded proteins. Biophys. J. 88, 1778–1798. von Heijne, G., Manoil, M., 1990. Membrane proteins: from sequence to structure. Protein Eng. 4, 109–112. Wang, J.Y., Gambhir, A., McLaughlin, S., Murray, D., 2004. A computational model for the electrostatic sequestration of PI(4,5)P2 by membrane-adsorbed basic peptides. Biophys. J. 86, 1969– 1986. Warren, P.B., 1998. Dissipative particle dynamics. Curr. Opin. Colloid Interf. Sci. 3, 620–624. White, S.H., Wimley, W.C., 1994. Peptides in lipid bilayers: structural and thermodynamic basis for partitioning and folding. Curr. Opin. Struct. Biol. 4, 79–86. White, S.H., Wimley, W.C., 1999. Membrane protein folding and stability: physical principles. Ann. Rev. Biophys. Biomol. Struct. 28, 319–365. White, S.H., von Heijne, G., 2004. The machinery of membrane protein assembly. Curr. Opin. Struct. Biol. 14, 397–404. Wiener, C.M., White, S.H., 1992. Structure of a fluid dioleoylphosphatidylcholine bilayer determined by joint refinement of X-ray and neutron diffraction data. III. Complete structure. Biophys. J. 61, 437–447. Weiss, T.M., van der Wei, P.C.A., Killian, J.A., Koeppe II, R.E., Huang, H.W., 2003. Hydrophobic mismatch between helices and lipid bilayers. Biophys. J. 84, 379–385. Zemel, A., Fattal, D.R., Ben-Shaul, A., 2003. Energetics and selfassembly of amphipathic peptide pores in lipid membranes. Biophys. J. 84, 2242–2255. Zemel, A., Ben-Shaul, A., May, S., 2004. Membrane perturbation induced by interfacially adsorbed peptides. Biophys. J. 86, 3607–3619. Zemel, A., Ben-Shaul, A., May, S., 2005. Perturbation of a lipid membrane by amphipathic peptides and its role in pore formation. Eur. Biophys. J. 34, 230–243.
© Copyright 2026 Paperzz