Session Q39: Physical Mechanisms of Membrane Remodeling

How some proteins tubulate membranes

Endocytosis, exocytosis, membrane transport between intracellular compartments, virus or toxin entry or exit out of the cell, all imply to deform membrane. Membrane deformation mechanisms of cell membranes by proteins are currently actively studied. Giant vesicles (GUV) are interesting model membrane systems because they are composed of a very limited number of components compared to cellular membranes. The deformations induced by the interaction with a specific protein or any other additional components to the system, can then be directly monitored and the deformation mechanism eventually understood. In this talk, we will focus on different tubular structures induced by proteins. We will show that the B-subunits of Shiga toxin or Cholera Toxin, binding to their lipid receptors, Gb3 or GM1 respectively, incorporated in GUV membrane, induce negative membrane curvature and form tubular invaginations, in absence of any other cellular machinery. Tubular structures can also be obtained when molecular motors walking along microtubules exert a pulling force on the membrane of GUV. The helicoidal assembly of dynamin, a protein involved \textit{in vivo} in membrane fission can also produce tubular structures. This assembly has been reconstituted around membrane nanotubes of controlled diameter; we will show that the initial tube diameter strongly influences dynamin polymerisation. In each case, a physical framework for understanding deformation mechanism will be presented   

Membrane Disruption Effects of antimicrobial Peptide Protegrin-1

Molecular dynamics simulations have been performed to better understand membrane disruption induced by antimicrobial peptide Protegrin-1 (PG-1). Two distinct setups were adopted for atomistic simulations for DMPC/PG-1 systems. One started from bilayered ribbons either with or without the PG-1 peptides embedded and another one started from random lipid mixtures in the presence of the PG-1 peptides. Line tensions deduced from the ribbon simulations were generally lower with the PG-1 peptides embedded in ribbon edge, which supports edge-active role of the peptides. The random mixtures self-assembled into various structures. Extended simulations are being carried out to investigate the relation between concentration of the PG-1 peptides and the resultant structures. Furthermore, coarse-grained models have been used to simulate larger DMPC bilayers with the PG-1 peptides embedded. The PG-1 peptides were found to self-assemble into clusters. However, pore formation was not observed within our simulation period up to 3 microseconds. (DMPC: 1,2-Dimyristoyl-sn-Glycero-3-Phosphocholine)   

Activation Dependent Organization of T Cell Membranes

We investigate the role of lipid anchor motifs in the micro-organization of T-cell plasma membranes. To that end we generated a combinatorial library of protein constructs by fusing different lipid-modification sites of lipid anchored proteins with one of two fluorescent proteins (EGFP and Cherry). Two of these constructs that encode either for myristoylation, palmitoylation, geranylation or glycosylphosphatidylinositol (GPI) elaboration were co-expressed and dual color fluorescence cross-correlation spectroscopy (FCCS) was used to exploit co{\-}movement as a signature of co-localization. We find that in living T cells most anchors only co-localize with themselves, while different anchors move independently from each other. This suggests that a variety of domains with different chemical compositions exist in the plasma membrane and that the lipid anchor structure plays a key role in domain- specific recruitment of proteins. Furthermore, the degree of aggregation is found to depend on the activation state of the T cells. Cholesterol depletion and actin-drug experiments indicate that both are involved in the dynamic organization of the T cell plasma membrane.   

Blebs in Model Lipid Membranes

It is now widely recognized that biomembranes exhibit complex lateral heterogeneities. Among these are blebs, which are localized balloon-like membrane protrusions observed during cell apoptosis, necrosis, and cytokinesis. Despite the poorly understood mechanism of bleb formation and their physiological role, they recently received a renewed attention. In order to investigate the physical mechanism leading to bleb formation, we developed a model based on an implicit-solvent lipid membrane model with soft interactions recently proposed by us [J. Chem. Phys. {\bf 128}, 035102 (2008)]. The model also incorporates an explicit fluctuating polymer meshwork simulating a cytoskeleton, which is anchored to the membrane. Using systematic large-scale simulations of membranes with varying values of the lipid density, cytoskeleton grafting-sites density and cytoskeleton tension, we found that localized blebs are formed on the membrane exoplasmic side in the presence of mismatch between tensions of the bare membrane and cytoskeleton. The blebs are pinned by the cytoskeleton anchors, reminiscent to those observed in apoptotic cells. The distance between neighboring anchors determines the neck of a bleb. The remaining membrane surrounding the blebs stiffens to accommodate the tensed cytoskeleton.   

Membrane Fusion Proteins as Nanomachines

Membrane fusion is key to fertilization, virus infection, and neurotransmission. Specific proteins work like nanomachines to stitch together fluid, yet highly ordered lipid bilayers. The energy gained from large exothermic conformational changes of these proteins is utilized to fuse lipid bilayers that do not fuse spontaneously. Structural studies using x-ray crystallography and NMR spectroscopy have yielded detailed information about architecture and inner workings of these molecular machines. The question now is: how is mechanical energy gained from such protein transformations harnessed to transform membrane topology? To answer this question, we have determined that a boomerang-shaped structure of the influenza fusion peptide is critical to generate a high-energy binding intermediate in the target membrane and to return the ``boomerang'' to its place of release near the viral membrane for completion of the fusion cycle. In presynaptic exocytosis, receptor and acceptor SNAREs are zippered to form a helical bundle that is arrested shortly before the membrane. Ca binding to interlocked synaptotagmin releases the fusion block. Structural NMR and single molecule fluorescence data are combined to arrive at and further refine this picture.   

Miscibility Critical Points in Plasma Membranes

Lipid bilayers surround all cells and are home to a host of proteins and lipids that mediate interactions between the cell and its environment. Recent experimental work has shown that simple membranes composed of three lipid components show complex phase behavior at temperatures in the physiological range. For example, two liquid phases and a gel or solid phase are seen, and a second order phase transition with Ising critical behavior can be reached at a boundary of the liquid-liquid coexistence region [1]. Surprisingly, membrane vesicles isolated from living cells can be tuned with a single parameter (temperature) to criticality [1]. This suggests that cell membranes in vivo sit near miscibility critical points, and may help explain some of the paradoxes associated with putative lipid rafts proposed in other experiments. Here we report on work mapping phase diagrams for the simple membranes utilizing NMR and microscopy data. In addition, we use canonical models of phase transitions to understand the qualitative features of the membranes. Finally we explore ideas from information theory and self organized criticality to understand how and why real cells maintain a membrane near criticality. [1] Honerkamp-Smith, Veatch, and Keller, Biochim Biophys Acta. 2008 (in press)   

Calcium-induced gel domains in bilayer -- more elusive than thought

As a highly bioactive divalent cation, calcium can in principle crossbridge anionic groups and induce domain formation and rigidification, but past reports with lipid systems appear conflicted. We mix, as a robust model system, anionic and neutral polymer amphiphiles within vesicle and cylinder micelle morphologies and add calcium. Based on micro- measurements, calcium forms definitive crossbridges between the anionic amphiphiles, rigidifying the charged membranes across a fluid-gel transition and also leading to lateral phase separation without disrupting the assemblies. A systematic phase diagram shows that long-lived domains occur in a narrow region near the polyanion's p$K$'s. The phase behavior appears well described by a relatively simple model in which -- among electrostatic and entropic contributions -- counterion entropy outcompetes attractive crossbridging to drive remixing of the highly charged polyacid at high pH. Initial observations extend from polymers to a polyanionic lipid involved in cell signaling [phosphatidylinositol (4,5)-bisphosphate], highlighting both the generality and limits of calcium effects.   

Nonequilibrium instabilities of membranes with multiple-state active proteins

We present a theoretical model for the dynamics of membranes containing active proteins that have several conformational states. The proteinss are active because their conformational transition rates depend on the strength of external energy source that drives the system out of equilibrium. We show that there exist several types of nonequilibrium phase transitions for a membrane with proteins that have typical transition rates and in-plane diffusion constant.   

Physical Foundations of PTEN/Phosphoinositide Interaction

Phosphoinositides act as signaling molecules by recruiting critical effectors to specific subcellular membranes to regulate cell proliferation, apoptosis and cytoskeletal reorganization, which requires a tight regulation of phosphoinositide generation and turnover as well as a high degree of compartmentalization. PTEN is a phosphatase specific for the 3 position of the phosophoinositide ring that is deleted or mutated in many different disease states. PTEN association with membranes requires the interaction of its C2 domain with phosphatidylserine and the interaction of its N-terminal end with phosphatidylinositol-4,5-bisphophate (PI(4,5)P$_{2})$. We have investigated PTEN/PI(4,5)P$_{2}$ interaction and found that Lys13 is crucial for the observed binding. We also found that the presence of cholesterol enhances PTEN binding to mixed PI(4,5)P$_{2}$/POPC vesicles. Fluorescence microscopy experiments utilizing GUVs yielded results consistent with enhanced phosphoinositide domain formation in the presence of cholesterol. These experiments were accompanied by zeta potential measurements and solid state MAS $^{31}$P-NMR experiments aimed at investigating the ionization behavior of phosphoinositides.   

Multiscale modeling of phospholipid bilayers: from explicit-solvent all-atom to solvent-free coarse-grained simulations

To advance the efficiency of phospholipid bilayer simulations and permit the treatment of challenging phenomena on mesoscopic length- and time-scales, several solvent-free Coarse-Grained (CG) phospholipid bilayer models have been presented in the past ten years. Most of them were derived in a top-down scheme, aiming to qualitatively reproduce phase diagrams, bending and stretching rigidity, area per lipid, and the thickness of a generic bilayer in experiments. Here, we derive a bottom-up CG model of an implicit-solvent lipid bilayer by matching its structural and mechanical properties with that of a membrane in all-atom resolution. Besides preserving chemical specificity and quantitative accuracy, we expect to gain a more fundamental understanding of the relationship between the elastic, mechanical, and diffusive properties of implicit solvent bilayers and their underlying CG interaction potentials, specifically bonded and non-bonded forces, as well as the effective interactions replacing the solvent.   

Effects of cholesterol and PIP2 on membrane domain formation

Lipid head group size, acyl chain saturation, the relative amounts of cholesterol, phospholipids and sphingolipids, and electrostatic effects due to highly charged anionic lipids such as phosphatidylinositol bisphosphate (PIP2) all contribute to the force balance that determines the conditions at which domains form as well as their size, shape and stability. Giant plasma membrane vesicles derived from intact cells reveal lipid phase separation in a system with appropriate biological complexity. Formation of liquid ordered domains large enough to visualize by light microscopy form under physiologically realistic conditions in cell-derived vesicles, and their dependence on cholesterol content and temperature are consistent with studies of purified lipids. Compared to the effects of cholesterol, PIP2 has a smaller but still significant effect on liquid ordered / liquid disordered domain formation, but compared to other lipids, PIP2 is much more strongly segregated in the liquid disordered domains, away from those enriched in cholesterol. These results suggest physical mechanisms by which the cell can rapidly alter local PIP2 concentration to trigger cellular signals.