Protein-membrane Interface Research Articles

Recalibration of MARTINI-3 Parameters for Improved Interactions between Peripheral Proteins and Lipid Bilayers.

The MARTINI force field is one of the most used coarse-grained models for biomolecular simulations. Many limitations of the model including the protein-protein overaggregation have been improved in its latest version, MARTINI-3. In this study, we investigate the efficacy of the MARTINI-3 parameters for capturing the interactions of peripheral proteins with model plasma membranes. Particularly, we consider two classes of proteins, namely, annexin and epsin, which are known to generate negative and positive membrane curvatures, respectively. We find that current MARTINI-3 parameters are not able to correctly describe the protein-membrane interface and the protein-induced membrane curvatures for any of these proteins. The problem arises due to the lack of proper hydrophobic interactions between the protein residues and lipid tails. Making systematic adjustments, we show that a combination of reduction in the protein-water interactions and enhancement of protein-lipid hydrophobic interactions is essential for accurate prediction of the interfacial structure including the protein-induced membrane curvature. Next, we apply our model to a couple of other peripheral proteins, namely, Snf7, a core component of the ESCRT-III complex, and the PH domain of evectin-2. We find that our model captures the protein-membrane interfacial structure much more accurately than the MARTINI-3 model for all of the peripheral proteins considered in this study. However, the strategy described in this study may not be suitable for oligomeric transmembrane proteins where protein-protein hydrophobic interactions should be increased instead of protein-lipid hydrophobic interactions.

Membrane curvature sensing and symmetry breaking of the M2 proton channel from Influenza A.

The M2 proton channel aids in the exit of mature influenza viral particles from the host plasma membrane through its ability to stabilize regions of high negative Gaussian curvature (NGC) that occur at the neck of budding virions. The channels are homo-tetramers that contain a cytoplasm-facing amphipathic helix (AH) that is necessary and sufficient for NGC generation; however, constructs containing the transmembrane spanning helix, which facilitates tetramerization, exhibit enhanced curvature generation. Here, we used all-atom molecular dynamics (MD) simulations to explore the conformational dynamics of M2 channels in lipid bilayers revealing that the AH is dynamic, quickly breaking the fourfold symmetry observed in most structures. Next, we carried out MD simulations with the protein restrained in four- and twofold symmetric conformations to determine the impact on the membrane shape. While each pattern was distinct, all configurations induced pronounced curvature in the outer leaflet, while conversely, the inner leaflets showed minimal curvature and significant lipid tilt around the AHs. The MD-generated profiles at the protein-membrane interface were then extracted and used as boundary conditions in a continuum elastic membrane model to calculate the membrane-bending energy of each conformation embedded in different membrane surfaces characteristic of a budding virus. The calculations show that all three M2 conformations are stabilized in inward-budding, concave spherical caps and destabilized in outward-budding, convex spherical caps, the latter reminiscent of a budding virus. One of the C2-broken symmetry conformations is stabilized by 4 kT in NGC surfaces with the minimum energy conformation occurring at a curvature corresponding to 33 nm radii. In total, our work provides atomistic insight into the curvature sensing capabilities of M2 channels and how enrichment in the nascent viral particle depends on protein shape and membrane geometry.

Ligandability at the Membrane Interface of GPx4 Revealed through a Reverse Micelle Fragment Screening Platform.

While they account for a large portion of drug targets, membrane proteins present a unique challenge for drug discovery. Peripheral membrane proteins (PMPs), a class of water-soluble proteins that bind to membranes, are also difficult targets, particularly those that function only when bound to membranes. The protein-membrane interface in PMPs is often where functional interactions and catalysis occur, making it a logical target for inhibition. However, protein-membrane interfaces are underexplored spaces in inhibitor design, and there is a need for enhanced methods for small-molecule ligand discovery. In an effort to better initiate drug discovery efforts for PMPs, this study presents a screening methodology using membrane-mimicking reverse micelles (mmRM) and NMR-based fragment screening to assess ligandability at the protein-membrane interface. The proof-of-principle target, glutathione peroxidase 4 (GPx4), is a lipid hydroperoxidase that is essential for the oxidative protection of membranes and thereby the prevention of ferroptosis. GPx4 inhibition is promising for therapy-resistant cancer therapy, but current inhibitors are generally covalent ligands with limited clinical utility. Presented here is the discovery of noncovalent small-molecule ligands for membrane-bound GPx4 revealed through the mmRM fragment screening methodology. The fragments were tested against GPx4 under bulk aqueous conditions and displayed little to no binding to the protein without embedment into the membrane. The 9 hits had varying affinities and partitioning coefficients and revealed properties of fragments that bind within the protein-membrane interface. Additionally, a secondary screen confirmed the potential to progress the fragments by enhancing the affinity from >200 to ∼15 μM with the addition of certain hydrophobic groups. This study presents an advancement of screening capabilities for membrane-associated proteins, reveals ligandability within the GPx4 protein-membrane interface, and may serve as a starting point for developing noncovalent inhibitors of GPx4.

Binding equations for the lipid composition dependence of peripheral membrane-binding proteins

The specific recognition of peripheral membrane-binding proteins for their target membranes is mediated by a complex constellation of various lipid contacts. Despite the inherent complexities of the heterogeneous protein-membrane interface, the binding dependence of such proteins is, surprisingly, often reliably described by simple models such as the Langmuir Adsorption Isotherm or the Hill equation. However, these models were not developed to describe associations with two-dimensional, highly concentrated heterogeneous ligands such as lipid membranes. In particular, these models fail to capture the dependence on the lipid composition, a significant determinant of binding that distinguishes target from non-target membranes. In this work, we present a model that describes the dependence of peripheral proteins on lipid composition through an analytic expression for their association. The resulting membrane-binding equation retains the features of these simple models but completely describes the binding dependence on multiple relevant variables in addition to the lipid composition, such as protein and vesicle concentration. Implicit in this lipid composition dependence is a new form of membrane-based cooperativity that significantly differs from traditional solution-based cooperativity. We introduce the Membrane-Hill number as a measure of this cooperativity and describe its unique properties. We illustrate the utility and interpretational power of our model by analyzing previously published data on two peripheral proteins that associate with phosphatidylserine-containing membranes: The transmembrane immunoglobulin and mucin domain-containing protein 3 (TIM3) that employs calcium in its association, and milk fat globulin epidermal growth factor VIII (MFG-E8) which is completely insensitive to calcium. We also provide binding equations for systems that exhibit more complexity in their membrane-binding.

Using deep learning and large protein language models to predict protein-membrane interfaces of peripheral membrane proteins.

Characterizing interactions at the protein-membrane interface is crucial as abnormal peripheral protein-membrane attachment is involved in the onset of many diseases. However, a limiting factor in studying and understanding protein-membrane interactions is that the membrane-binding domains of peripheral membrane proteins (PMPs) are typically unknown. By applying artificial intelligence techniques in the context of natural language processing (NLP), the accuracy and prediction time for protein-membrane interface analysis can be significantly improved compared to existing methods. Here, we assess whether NLP and protein language models (pLMs) can be used to predict membrane-interacting amino acids for PMPs. We utilize available experimental data and generate protein embeddings from two pLMs (ProtTrans and ESM) to train classifier models. Overall, the results demonstrate the first proof of concept study and the promising potential of using deep learning and pLMs to predict protein-membrane interfaces for PMPs faster, with similar accuracy, and without the need for 3D structural data compared to existing tools. The code is available at https://github.com/zoecournia/pLM-PMI. All data are available in the Supplementary material.

Phase-Separated Lipid-Based Nanoparticles: Selective Behavior at the Nano-Bio Interface.

The membrane-protein interface on lipid-based nanoparticles influences their in vivo behavior. Better understanding may evolve current drug delivery methods toward effective targeted nanomedicine. Previously, the cell-selective accumulation of a liposome formulation in vivo is demonstrated, through the recognition of lipid phase-separation by triglyceride lipases. This exemplified how liposome morphology and composition can determine nanoparticle-protein interactions. Here, the lipase-induced compositional and morphological changes of phase-separated liposomes-which bear a lipid droplet in their bilayer- are investigated, and the mechanism upon whichlipases recognize and bind to the particles is unravelled. The selective lipolytic degradation of the phase-separated lipid droplet is observed, while nanoparticle integrity remains intact. Next, the Tryptophan-rich loop of the lipase is identified as the region with which the enzymes bind to the particles. This preferential binding is due to lipid packing defects induced on the liposome surface by phase separation. In parallel, the existing knowledge that phase separation leads to in vivo selectivity, is utilized to generate phase-separated mRNA-LNPs that target cell-subsets in zebrafish embryos, with subsequent mRNA delivery and protein expression. Together, these findings can expand the current knowledge on selective nanoparticle-protein communications and in vivo behavior, aspects that will assist to gain control of lipid-based nanoparticles.

Docosahexaenoic acid promotes vesicle clustering mediated by alpha-Synuclein via electrostatic interaction.

α-Synuclein (α-Syn) is an intrinsically disordered protein whose aggregation is associated with Parkinson's disease, dementia, and other neurodegenerative diseases known as synucleinopathies. However, the functional role of α-Syn is still unclear, although it has been shown to be involved in the regulation of neurotransmitter release via the interaction with synaptic vesicles (SVs), vesicle clustering, and SNARE complex assembly. Fatty acids have significant occupancy in synaptic vesicles; and recent studies suggest the interaction of fatty acids with α-Syn affect the formation of (pathological) aggregates, but it is less clear how fatty acids affects the functional role of α-Syn including α-Syn-membrane interactions, in particular with (SV-like) vesicles. Here, we report the concentration dependent effect of docosahexaenoic acid (DHA) in synaptic-like vesicle clustering via α-Syn interaction. Through molecular dynamics simulation, we revealed that DHA promoted vesicle clustering is due to the electrostatic interaction between DHA in the membrane and the N-terminal region of α-Syn. Moreover, this increased electrostatic interaction arises from a change in the macroscopic properties of the protein-membrane interface induced by (preferential solvation of) DHA. Our results provide insight as to how DHA regulates vesicle clustering mediated by α-Syn and may further be useful to understand its physiological as well as pathological role. Description: In physiological environments, α-Synuclein (α-Syn) localizes at nerve termini and synaptic vesicles and interacts with anionic phospholipid membranes to promote vesicle clustering. Docosahexaenoic acid (DHA) increases binding affinity between α-Syn and lipid membranes by increasing electrostatic interaction energy through modulating the local and global membrane environment and conformational properties of α-Syn.

Dynamic lipid interactions in the plasma membrane Na+,K+-ATPase

The function of ion-transporting Na+,K+-ATPases depends on the surrounding lipid environment in biological membranes. Two established lipid-interaction sites A and B within the transmembrane domain have been observed to induce protein activation and stabilization, respectively. In addition, lipid-mediated inhibition has been assigned to a site C, but with the exact location not experimentally confirmed. Also, possible effects on lipid interactions by disease mutants dwelling in the membrane-protein interface remain relatively uncharacterized. We simulated human Na+,K+-ATPase α1β1FXYD homology models in E1 and E2 states in an asymmetric, multicomponent plasma membrane to determine both wild-type and disease mutant lipid-protein interactions. The simulated wild-type lipid interactions at the established sites A and B were in agreement with experimental results thereby confirming the membrane-protein model system. The less well-characterized, proposed inhibitory site C was dominated by lipids lacking inhibitory properties. Instead, two sites hosting inhibitory lipids were identified at the extracellular side and also a cytoplasmic CHL-binding site that provide putative alternative locations of Na+,K+-ATPase inhibition. Three disease mutations, Leu302Arg, Glu840Arg and Met859Arg resided in the lipid-protein interface and caused drastic changes in the lipid interactions. The simulation results show that lipid interactions to the human Na+,K+-ATPase α1β1FXYD protein in the plasma membrane are highly state-dependent and can be disturbed by disease mutations located in the lipid interface, which can open up for new venues to understand genetic disorders.

Molecular dynamics simulation of phosphatidylcholine membrane in low ionic strengths of sodium chloride

The one-microsecond molecular dynamics simulations of a membrane-protein complex investigate the influence of the aqueous sodium chloride solutions on the structure and dynamics of a palmitoyl-oleoyl-phosphatidylcholine bilayer membrane. The simulations were performed on five different concentrations (40, 150, 200, 300, and 400 mM) in addition to a salt-free system by using the charmm36 force field for all atoms. Four biophysical parameters, (membrane thicknesses of annular and bulk lipids, and the area per lipid of both leaflets), were computed separately. Nevertheless, the area per lipid was expressed by using the Voronoi algorithm. All time-independent analyses were carried out for the last 400 ns trajectories. Different concentrations revealed dissimilar membrane dynamics before equilibration. The biophysical properties of the membrane (thickness, area-per-lipid, and order parameter) have non-significant changes with increasing ionic strength, however, the 150 mM system had exceptional behavior. Sodium cations were dynamically penetrating the membrane forming weak coordinate bonds with single or multiple lipids. Nevertheless, the binding constant was unaffected by the cation concentration. The electrostatic and Van der Waals energies of lipid-lipid interactions were influenced by the ionic strength. On the other hand, the Fast Fourier Transform was performed to figure out the dynamics at the membrane-protein interface. The nonbonding energies of membrane-protein interactions and order parameters explained the differences in the synchronization pattern. All results were consensus with experimental and theoretical works.Communicated by Ramaswamy H. Sarma

Membrane composition drives sidechain ionization and assembly of transmembrane protein domains.

Ionization states of amino acid residues play significant roles in membrane protein assembly and function; however, they are difficult to decipher experimentally. The analysis is further complicated by the dearth of data on gradients in polarity, electric potentials, and hydration at the protein-membrane interface. Here we examine how electrostatic interactions, suggested to be essential for T-cell receptor (TCR) membrane assembly, could be manipulated by modifying the membrane composition. Novel pH-sensitive ionizable EPR labels were employed to profile heterogeneous dielectric environment along the transmembrane protein-lipid interfaces. Transmembrane α-helical WALP peptide was used for obtaining the profile of effective pK(a) as a function of membrane depth. It was found that effective pK(a) of membrane-buried sidechains can be significantly shifted by varying the membrane surface charge density. A peptide corresponding to the transmembrane domain of TCRα was labeled with pH-sensitive nitroxide and incorporated into liposomes. EPR of this label reported on the sidechain protonation state, membrane insertion, and association of the helices within the membrane. It was found that an increase in negative change density at the membrane surface alters the protonation state of membrane-buried ionizable sidechain in the transmembrane domain of TCRα. Hyperfine Correlation Spectroscopy (HYSCORE) was used to assess effects of sidechain ionization state on water penetration at membrane-peptide interface while Double Electron-Electron Resonance (DEER) quantified the domain agglomeration. Turning the charge of the sidechain “off” increases tendency of the transmembrane domain to agglomerate - a phenomenon that could be critical for the TCR assembly and its degradation. Overall, EPR of ionizable probes is informative for uncovering mechanisms as to how the changes in local environment along the protein-lipid interface, can result in a “tunable” pKa of the ionizable sidechains and drive the structural changes in membrane proteins.

Novel approaches to understand function and explore inhibition within protein-membrane interfaces.

Despite their importance in biology and disease, peripheral membrane proteins (PMPs) have presented a significant challenge. This class of proteins reversibly binds to membranes to perform function. High-resolution functional and structural studies of PMPs are often limited to their water-solubilized state due to technical limitations. Large blind spots in functional interactions within membranes are therefore common and require methodological and technological advancement to overcome. Additionally, inhibitor development for PMPs in their functionally relevant, membrane embedded state is difficult, often impossible, using current methods. We address these barriers by utilizing newly developed membrane mimicking reverse micelles (mmRMs), which house PMPs within a spherical, nanoscale assembly of lipids. Studies using mmRMs have been applied to several PMPs, including glutathione peroxidase 4 (GPx4). Use of mmRMs to investigate PMPs has proven advantageous over other membrane models for various NMR-based experimental approaches. Observations of protein-lipid interactions and lipid specificity determinations are greatly enhanced. Applying mmRMs to PMP structural analysis promises to reveal greater detail of the effects of membrane binding. Finally, mmRMs are helping to overcome challenges associated with inhibitor discovery and design for membrane embedded PMPs. mmRM-based screening for fragments in the protein-membrane interface reveals building blocks that may be advanced to inhibitors. Furthermore, discovery of small-molecule binders within membrane interfaces promises to reveal fundamental properties of this largely unexplored chemical space. Together, these approaches promise to advance our knowledge and add to the tool belt for the important and elusive peripheral membrane category of proteins.

Theater in the Self-Cleaning Cell: Intrinsically Disordered Proteins or Protein Regions Acting with Membranes in Autophagy.

Intrinsically disordered proteins and protein regions (IDPs/IDPRs) are mainly involved in signaling pathways, where fast regulation, temporal interactions, promiscuous interactions, and assemblies of structurally diverse components including membranes are essential. The autophagy pathway builds, de novo, a membrane organelle, the autophagosome, using carefully orchestrated interactions between proteins and lipid bilayers. Here, we discuss molecular mechanisms related to the protein disorder-based interactions of the autophagy machinery with membranes. We describe not only membrane binding phenomenon, but also examples of membrane remodeling processes including membrane tethering, bending, curvature sensing, and/or fragmentation of membrane organelles such as the endoplasmic reticulum, which is an important membrane source as well as cargo for autophagy. Summary of the current state of knowledge presented here will hopefully inspire new studies. A profound understanding of the autophagic protein–membrane interface is essential for advancements in therapeutic interventions against major human diseases, in which autophagy is involved including neurodegeneration, cancer as well as cardiovascular, metabolic, infectious, musculoskeletal, and other disorders.

Predicting protein-membrane interfaces of peripheral membrane proteins using ensemble machine learning.

Abnormal protein–membrane attachment is involved in deregulated cellular pathways and in disease. Therefore, the possibility to modulate protein–membrane interactions represents a new promising therapeutic strategy for peripheral membrane proteins that have been considered so far undruggable. A major obstacle in this drug design strategy is that the membrane-binding domains of peripheral membrane proteins are usually unknown. The development of fast and efficient algorithms predicting the protein–membrane interface would shed light into the accessibility of membrane–protein interfaces by drug-like molecules. Herein, we describe an ensemble machine learning methodology and algorithm for predicting membrane-penetrating amino acids. We utilize available experimental data from the literature for training 21 machine learning classifiers and meta-classifiers. Evaluation of the best ensemble classifier model accuracy yields a macro-averaged F1 score = 0.92 and a Matthews correlation coefficient = 0.84 for predicting correctly membrane-penetrating amino acids on unknown proteins of a validation set. The python code for predicting protein–membrane interfaces of peripheral membrane proteins is available at https://github.com/zoecournia/DREAMM.

Protonation of model ionizable sidechains in transmembrane protein domains

The ionization states of individual amino acid residues of membrane proteins in lipid membrane environment are difficult to decipher or assign directly. The effective pK(a) values of protein groups are determined by a complex interplay between local polarity, Coulomb interactions, and structural reorganizations. The analysis is further complicated by the dearth of information about gradients in polarity, electric potentials, and hydration at the protein-membrane interface. In this work we report on a spin-labeling EPR method for assessing effects of membrane surface potential, local environment at the protein-membrane interface, and water penetration along this interface on effective pK(a) of membrane-buried ionizable groups.

The conformational cycle of prestin underlies outer-hair cell electromotility.

The voltage-dependent motor protein prestin (also known as SLC26A5) is responsible for the electromotive behaviour of outer-hair cells and underlies the cochlear amplifier1. Knockout or impairment of prestin causes severe hearing loss2-5. Despite the key role of prestin in hearing, the mechanism by which mammalian prestin senses voltage and transduces it into cellular-scale movements (electromotility) is poorly understood. Here we determined the structure of dolphin prestin in six distinct states using single-particle cryo-electron microscopy. Our structural and functional data suggest that prestin adopts a unique and complex set of states, tunable by the identity of bound anions (Cl- or SO42-). Salicylate, a drug that can cause reversible hearing loss, competes for the anion-binding site of prestin, and inhibits its function by immobilizing prestin in a new conformation. Our data suggest that the bound anion together with its coordinating charged residues and helical dipole act as a dynamic voltage sensor. An analysis of all of the anion-dependent conformations reveals how structural rearrangements in the voltage sensor are coupled to conformational transitions at the protein-membrane interface, suggesting a previously undescribed mechanism of area expansion. Visualization of the electromotility cycle of prestin distinguishes the protein from the closely related SLC26 anion transporters, highlighting the basis for evolutionary specialization of the mammalian cochlear amplifier at a high resolution.

Protein-induced membrane curvature in coarse-grained simulations

Using the endosomal sorting complex required for transport (ESCRT)-III membrane remodeling complex as an example, we analyze three popular coarse-grained models (the regular MARTINI, polarizable MARTINI (POL-MARTINI), and big multipole water MARTINI (BMW-MARTINI)) for the description of membrane curvature sensing and generation activities of peripheral proteins. Although the three variants of the MARTINI model provide consistent descriptions for the protein-protein interface in a linear filament model of ESCRT-III, they differ considerably in terms of protein-membrane interface and therefore membrane curvature sensing and generation behaviors. In particular, BMW-MARTINI provides the most consistent description of the protein-membrane interface as compared to all-atom simulations, whereas the regular MARTINI is most consistent with atomistic simulations in terms of the qualitative sign of membrane curvature sensing and generation. With POL-MARTINI, the ESCRT-III model interacts weakly with the membrane and therefore does not exhibit any curvature-sensitive activities. Analysis suggests that the incorrect membrane curvature activities predicted by BMW-MARTINI are due to overestimated insertion depth of an amphipathic helix and incorrect sign for the spontaneous curvature of anionic lipids. These results not only point to ways that coarse-grained models can be improved but also explicitly highlight local lipid composition and insertion depth of protein motifs as essential regulatory factors for membrane curvature sensing and generation.

Estimating the Accuracy of the Martini Model Towards the Investigation of Peripheral Protein - Membrane Interactions

Peripheral membrane proteins play a major role in numerous biological processes by transiently associating with cellular membranes, often with extreme membrane specificity. Because of the short-lived nature of these interactions, molecular dynamics (MD) simulations have emerged as an appealing tool to characterize at the structural level the molecular details of the protein-membrane interface. Transferable coarse-grained (CG) MD simulations, in particular, offer the possibility to investigate the spontaneous association of peripheral proteins to lipid bilayers of different compositions at limited computational cost, but they are hampered by the lack of a reliable a priori estimation of their accuracy and thus typically require a posteriori experimental validation.

Wetting Transitions in the ATP Synthase C-Subunit Ring, a Large-Conductance Ion Channel

Apart from its known role in proton transport, the mammalian ATP synthase c-subunit ring (Fo domain) has been observed in recent electrophysiology experiments to form a high-conductance voltage-gated ion channel (∼1.5 nS). Yet, the ring's hydrophobic central lumen has been assumed to be dry, precluding a physical explanation for the observed conductance. We present an atomistic molecular dynamics simulation study of lumen wetting for mammalian and yeast c-subunit rings. Consistent with prior work, we find that the ∼12 Å diameter yeast ring remains dry on the several-hundred-nanosecond timescale of the simulations. In contrast, the narrower, ∼8 Å diameter, mammalian ring undergoes rapid (∼50 ns) liquid-vapor transitions, forming a transient but continuous water column across the lumen. Modifying the charge on the glutamate residues of the proton-transport pathway at the membrane-protein interface affects the duration and extent of lumen wetting, as do in silico mutations at various sites. The wetting variations are qualitatively consistent with electrophysiology experimental findings for wild-type and mutant forms of the mammalian c-subunit ring. We are encouraged to think that the simulations will serve as a valuable “computational microscope” for understanding how charge conductance can occur in the ATP synthase c-subunit ring. To determine the mechanism of wetting and its modulation, we are pursuing analysis of the surface hydrophobicity of the lumen lining and its response to the mutations. The simulation finding of reversible lumen wetting is supported by published work indicating c-subunit ring wetting in the intact ATP synthase. New experiments and published data additionally indicate that the Fo domain may exist at least partly dissociated from its F1 counterpart. ATP synthase c-subunit ring wetting, therefore, has implications with respect to osmotic swelling and the enigmatic mitochondrial permeability transition.

Estimating the accuracy of the MARTINI model towards the investigation of peripheral protein-membrane interactions.

Peripheral membrane proteins play a major role in numerous biological processes by transiently associating with cellular membranes, often with extreme membrane specificity. Because of the short-lived nature of these interactions, molecular dynamics (MD) simulations have emerged as an appealing tool to characterize at the structural level the molecular details of the protein-membrane interface. Transferable coarse-grained (CG) MD simulations, in particular, offer the possibility to investigate the spontaneous association of peripheral proteins with lipid bilayers of different compositions at limited computational cost, but they are hampered by the lack of a reliable a priori estimation of their accuracy and thus typically require a posteriori experimental validation. In this article, we investigate the ability of the MARTINI CG force field, specifically the 3 open-beta version, to reproduce known experimental observations regarding the membrane binding behavior of 12 peripheral membrane proteins and peptides. Based on observations of multiple binding and unbinding events in several independent replicas, we found that, despite the presence of false positives and false negatives, this model is mostly able to correctly characterize the membrane binding behavior of peripheral proteins, and to identify key residues found to disrupt membrane binding in mutagenesis experiments. While preliminary, our investigations suggest that transferable chemical-specific CG force fields have enormous potential in the characterization of the membrane binding process by peripheral proteins, and that the identification of negative results could help drive future force field development efforts.

A computational model of protein induced membrane morphology with geodesic curvature driven protein-membrane interface

Continuum or hybrid modeling of bilayer membrane morphological dynamics induced by embedded proteins necessitates the identification of protein-membrane interfaces and coupling of deformations of two surfaces. In this article we developed (i) a minimal total geodesic curvature model to describe these interfaces, and (ii) a numerical one-to-one mapping between two surface through a conformal mapping of each surface to the common middle annulus. Our work provides the first computational tractable approach for determining the interfaces between bilayer and embedded proteins. The one-to-one mapping allows a convenient coupling of the morphology of two surfaces. We integrated these two new developments into the energetic model of protein-membrane interactions, and developed the full set of numerical methods for the coupled system. Numerical examples are presented to demonstrate (1) the efficiency and robustness of our methods in locating the curves with minimal total geodesic curvature on highly complicated protein surfaces, (2) the usefulness of these interfaces as interior boundaries for membrane deformation, and (3) the rich morphology of bilayer surfaces for different protein-membrane interfaces.