MARTINI Coarse-grained Force Field Research Articles

Composition Fluctuations in Lipid Bilayers

Cell membranes contain multiple lipid and protein components having heterogeneous in-plane (lateral) distribution. Nanoscale rafts are believed to play an important functional role, but their phase state—domains of coexisting phases or composition fluctuations—is unknown. As a step toward understanding lateral organization of cell membranes, we investigate the difference between nanoscale domains of coexisting phases and composition fluctuations in lipid bilayers. We simulate model lipid bilayers with the MARTINI coarse-grained force field on length scales of tens of nanometers and timescales of tens of microseconds. We use a binary and a ternary mixture: a saturated and an unsaturated lipid, or a saturated lipid, an unsaturated lipid, and cholesterol, respectively. In these mixtures, the phase behavior can be tuned from a mixed state to a coexistence of a liquid-crystalline and a gel, or a liquid-ordered and a liquid-disordered phase. Transition from a two-phase to a one-phase state is achieved by raising the temperature and adding a hybrid lipid (with a saturated and an unsaturated chain). We analyze the evolution of bilayer properties along this transition: domains of two phases transform to fluctuations with local ordering and compositional demixing. Nanoscale domains and fluctuations differ in several properties, including interleaflet overlap and boundary length. Hybrid lipids show no enrichment at the boundary, but decrease the difference between the coexisting phases by ordering the disordered phase, which could explain their role in cell membranes.

Enhanced Sampling of Phase Transitions in Coarse-Grained Lipid Bilayers.

Freezing and melting of dipalmitoylphosphatidylcholine (DPPC) bilayers are simulated in both the explicit (Wet) and implicit solvent (Dry) coarse-grained MARTINI force fields with enhanced sampling, via the isobaric, molecular dynamics version of the generalized replica exchange method (gREM). Phase transitions are described with the entropic viewpoint, based upon the statistical temperature as a function of enthalpy, TS(H) = 1/(dS(H)/dH), where S is the configurational entropy. Bilayer thickness, area per lipid, and the second-rank order parameter (P2) are calculated vs temperature in the transition range. In a 32-lipid Wet MARTINI system, transitions in the lipid and water subsystems are strongly coupled, giving rise to considerable structure in TS(H) and the need to specify the state of the water when reporting a lipid transition temperature. For gel lipid + liquid water → fluid lipid + liquid water, we find 292.4 K. The small system is influenced by finite-size effects, but it is argued that the entropic approach is well suited to revealing them, which will be particularly relevant for studies of finite nanosystems where there is no thermodynamic limit. In a 390-lipid Dry MARTINI system, two-dimensional analogues of the topographies of coexisting states ("subphases") seen in pure fluids are found. They are not seen in the 32-lipid Wet or Dry system, but the Dry lipids show a new type of state with gel in one leaflet and tilted gel in the other. Dry bilayer transition temperatures are 333.3 K (390 lipids) and 338 K (32 lipids), indicating that the 32-lipid system is not too small for a qualitative study of the transition. Physical arguments are given for Dry lipid system size dependence and for the difference between Wet and Dry systems.

Au Nanoparticles in Lipid Bilayers: A Comparison between Atomistic and Coarse-Grained Models

The computational study of the interaction between charged, ligand-protected metal nanoparticles and model lipid membranes has been recently addressed at both atomistic and coarse-grained levels. Here we compare the performance of three versions of the coarse-grained Martini force field at describing the nanoparticle–membrane interaction. The three coarse-grained models differ in terms of treatment of long-range electrostatic interactions and water polarizability. The NP–membrane interaction consists of the transition from a metastable NP–membrane complex, in which the NP is only partially embedded in the membrane, to a configuration in which the NP is anchored to both membrane leaflets. All three of the coarse-grained models provide a description of the metastable NP–membrane complex that is consistent with that obtained using an atomistic force field. As for the anchoring transition, the polarizable-water Martini correctly describes the molecular mechanisms and the energetics of the transition. The standard version of the Martini model, instead, underestimates the free-energy barriers for anchoring and does not completely capture the membrane deformations involved in the transition process.

Molecular Dynamics Study on Lipid Wrapped Carbon Nanotube as an Artificial Membrane Channel

Carbon nanotube (CNT) is an ideal membrane channel because of its hydrophobic inner pore and prominent transport properties like typical biological channels. These CNT porins can transport water, small ions, and DNA. CNT-membrane interaction simulation shows that CNTs having a length similar to the thickness of lipid bilayer can be inserted into membrane and sustain angle almost perpendicular to membrane. Much longer ones can also be inserted into membrane but they are buried inside hydrophobic region of membrane (1). However, recent experimental research elucidated that longer CNTs wrapped by lipids can be inserted into lipid-bilayer and maintain their angle within 15 degrees (2).In this study, we performed molecular dynamics simulation for investigating interaction between lipids wrapped CNT and lipid-bilayer. Martini coarse-grained force field was used to reduce computational cost significantly. Also, wrapping process was analyzed by using the coarse-grained carbon nanotube (CG-CNT) and the 1,2-dioleoyl-sn-glycero-3-phosphocholine (DOPC) in various CNT length and surface adsorption density, respectively. Furthermore, we performed insertion simulation to see how these properties affect insertion process and finally compared CG-CNT's insertion angles with experimental data. As a result, CNT insertion takes longer time in general as the density of wrapped lipid increases. Also, we realized that CNT can maintain its insertion angle almost perpendicular to lipid-bilayer at a certain critical density.Finally, its insertion mechanism is fully discussed with respect to interaction energy between the wrapped CG-CNT and lipid membrane.(1) Lelimousin, M., and M. S. P. Sansom. 2013. Small 9:3639-3646.(2) Geng, J., K. Kim, J. F. et al. 2014. Nature 514:612-+.

Software for Direct Comparison of Membrane Scattering Experiments Data to Molecular Dynamics Simulations

Detailed information on lipid bilayers structure is very important for understanding the processes in cellular membranes. Nowadays X-ray and neutron scattering experiments are widely used as a source of lipid bilayers structural parameters (area per lipid, thickness, etc.), but in contrast to scattering patterns from crystal structures, lipid membranes produce very broad peaks in scattering intensity profiles due to their fluid nature. This makes the process of retrieving structural details indirect and introduces approximations to convert experimental structure factors to electron or neutron scattering profiles. On the other hand, molecular dynamics simulations are able to provide direct sub-atomic level information on lipid bilayers structure, but can produce unreliable data because of inaccuracy of the force fields. Simulation to Experiment (SIMtoEXP) software is designed to facilitate direct, model free, comparison of MD simulations to experimentally measured structure factors for lipid bilayers. We present SIMtoEXP version 2.0 - an updated version of the original software. The new version of this program is written in C++ using the Qt GUI library, instead of the original C/Tcl implementation. New features, including a MD trajectory reading module, were added to simplify the workflow for users. In future releases SIMtoEXP software will support structure factor calculations from simulations with the coarse-grained Martini force field. The Martini force field provides significant computational speedup while retaining significant chemical detail of the lipids. This allows simulating large lipid-protein aggregates on longer timescales and investigate the structure and lateral organization of cell membranes. Incorporation of Martini support in SIMtoEXP will allow direct comparison of coarse-grained simulations to experiments, which will benefit both Martini force-field parametrization and the interpretation of experimental results.

Structural effects and translocation of spontaneous membrane-translocating peptides with POPC bilayer

Martini coarse-grained force field simulations have been carried out to estimate the free energy profiles of the spontaneous membrane-translocating peptide TP2 and one negative control peptide ONEG with POPC as the model bilayer. The results show that the free energy minimum of TP2 is [Formula: see text]20[Formula: see text]kJ/mol lower than that of ONEG. In addition, the minimum of TP2 shifts slightly to the bilayer center compared with ONEG. The translocation barrier height for TP2 and ONEG are 119.0[Formula: see text]kJ/mol and 155.7[Formula: see text]kJ/mol, respectively. The lower central energy barrier of TP2 facilitates the transition between two leaflets of POPC. Both translocating peptides induce the formation of funnel-shaped structures at the bilayer center, but TP2 has a more compact structure and brings less perturbation compared with ONEG. Subsequently all atom molecular simulations testify the findings. It is indicated that compared with its negative control ONEG, TP2 binds better with lipid and penetrates deeper into bilayer with less perturbation to the bilayer structure. Our findings may shed light on the design and virtual screening of spontaneous membrane-translocating peptides.

A refined polarizable water model for the coarse-grained MARTINI force field with long-range electrostatic interactions.

We present a refined version of the polarizable Martini water model - coined refPOL - designed specifically for the use with long-range electrostatics. The refPOL model improves the agreement with the experimentally measured dielectric constant and the mass density of water at room temperature compared to the original polarizable Martini water force field when particle mesh Ewald electrostatics are employed. Our study reveals that the model remains applicable with various commonly used settings for the non-bonded interactions, including reaction field electrostatics. The oil/water partitioning behavior of uncharged Martini bead types is thoroughly investigated: Lennard-Jones interactions between the refPOL model and the remaining Martini beads are adjusted to reproduce the hydration free energies obtained with the original polarizable water model, while free energies of solvation in apolar media remain unchanged. The cross-interactions with charged bead types are parameterized to agree with the experimentally observed area per lipid of a fully solvated dipalmitoylphosphatidylcholine bilayer. We additionally verify the model by analyzing the potentials of mean force between different sample pairs in refPOL water and comparing the results to reference data obtained using the original polarizable Martini water model as well as fully atomistic simulations. Based on the results, we suggest to replace the original polarizable Martini water model with the new refPOL model for future applications.

Lipid and Peptide Diffusion in Bilayers: The Saffman-Delbrück Model and Periodic Boundary Conditions.

The periodic Saffman-Delbrück (PSD) model, an extension of the Saffman-Delbrück model developed to describe the effects of periodic boundary conditions on the diffusion constants of lipids and proteins obtained from simulation, is tested using the coarse-grained Martini and all-atom CHARMM36 (C36) force fields. Simulations of pure Martini dipalmitoylphosphatidylcholine (DPPC) bilayers and those with one embedded gramicidin A (gA) dimer or one gA monomer with sizes ranging from 512 to 2048 lipids support the PSD model. Underestimates of D∞ (the value of the diffusion constant for an infinite system) from the 512-lipid system are 35% for DPPC, 45% for the gA monomer, and 70% for the gA dimer. Simulations of all-atom DPPC and dioleoylphosphatidylcholine (DOPC) bilayers yield diffusion constants not far from experiment. However, the PSD model predicts that diffusion constants at the sizes of the simulation should underestimate experiment by approximately a factor of 3 for DPPC and 2 for DOPC. This likely implies a deficiency in the C36 force field. A Bayesian method for extrapolating diffusion constants of lipids and proteins in membranes obtained from simulation to infinite system size is provided.

Modeling the Hydrophobicity of Nanoparticles and Their Interaction with Lipids and Proteins.

We present a method of modeling nanoparticle (NP) hydrophobicity using coarse-grained molecular dynamics (CG MD) simulations, and apply this to the interaction of lipids with nanoparticles. To model at a coarse-grained level the wettability or hydrophobicity of a given material, we choose the MARTINI coarse-grained force field, and determine through simulation the contact angles of MARTINI water droplets residing on flat regular surfaces composed of various MARTINI bead types (C1, C2, etc.). Each surface is composed of a single bead type in each of three crystallographic symmetries (FCC, BCC, and HCP). While this method lumps together several atoms (for example, one cerium and two oxygens of CeO2) into a single CG bead, we can still capture the overall hydrophobicity of the actual material by choosing the MARTINI bead type that gives the best fit of the contact angle to that of the actual material, as determined by either experimental or all-atom simulations. For different MARTINI bead types, the macroscopic contact angle is obtained by extrapolating the microscopic contact angles of droplets of eight different sizes (containing Nw = 3224-22978 water molecules) to infinite droplet size. For each droplet, the contact angle was computed from a best fit of a circular curve to the droplet interface extrapolated to the first layer of the surface. We then examine how small nanoparticles of differing wettability interact with MARTINI dipalmitoylphosphotidylcholine (DPPC) lipids and SP-C peptides (a component of lung surfactant). The DPPC shows a transition from tails coating the nanoparticle to a hemimicelle coating the water-wet NP, as the contact angle of a water droplet on the surface is lowered below ∼60°. The results are relevant to developing a taxonomy describing the potential nanotoxicity of nanoparticle interactions with components in the lung.

A coarse-grained model for PCL: conformation, self-assembly of MePEG-b-PCL amphiphilic diblock copolymers

Poly-ε-caprolactone (PCL) is a biodegradable hydrophobic polyester that has been widely used in medical devices, tissue engineering and nanoparticle-based drug delivery. Coarse-grained molecular dynamics (CGMD) has been employed to study and gain insights into the conformational, structural and self-assembly behaviour of polymers, lipids and amphiphilic macromolecules. In this work, we developed a model for PCL within the framework of the MARTINI coarse-grained force field. The non-bonded interactions were based on the existing MARTINI bead types, while the bonded interactions were mapped onto a PCL rendition obtained from atomistic simulations. The model accurately reproduces the structural and dynamic properties of the PCL homopolymer and shows very reasonable temperature and solvent transferability. We also studied self-assembly of MePEG-b-PCL linear diblock copolymers using an existing MARTINI model for MePEG (Methoxy Polyethylene glycol), by analysing the critical micelle concentration (CMC), as well as the shape, size and morphology of the nano-polymeric micelles. We obtained excellent agreement of the CMC, while the size was under-predicted compared to experimental data. This robust model paves the way for CGMD modelling of PCL and serves as a starting point for future designs of PCL-related polymeric systems .

Steered molecular dynamics analysis of the role of cofilin in increasing the flexibility of actin filaments

Cofilin is one of the most essential regulatory proteins and participates in the process of disassembling actin filaments. Cofilin induces conformational changes to actin filaments, and both the bending and torsional rigidity of the filament. In this study, we investigate the effects of cofilin on the mechanical properties of actin filaments using computational methods. Three models defined by their number of bound cofilins are constructed using coarse-grained MARTINI force field, and they are then extended with steered molecular dynamics simulation. After obtaining the stress–strain curves of the models, we calculate their Young's moduli and other mechanical properties that have not yet been determined for actin filaments. We analyze the cause of the different behaviors of the three models based on their atomistic geometrical differences. Finally, it is demonstrated that cofilin binding causes changes in the distances, angles, and stabilities of the residues in actin filaments.

Exploring the Impact of Tail Polarity on the Phase Behavior of Single Component and Mixed Lipid Monolayers Using a MARTINI Coarse-Grained Force Field.

Coarse-grained molecular dynamics simulations have been used to investigate the effect of dipalmitoylphosphatidylcholine (DPPC) tail group polarity on the structural and phase behavior of both single component and binary mixed monolayers using the MARTINI force field. Surface pressure-area isotherms of single component systems indicate that DPPC monolayers become more expanded as a function of increasing tail group polarity, as observed in experimental measurements in the literature. A combination of radial distribution functions and tilt angle measurements indicate that increasing tail group polarity results in the formation of increasingly disordered monolayers. For the mixed monolayer systems, the time dependence of the radial distribution function as well as average cluster size measurements indicate that phase separation takes place between components of different tail group polarity when the monolayers undergo phase transition into disordered configurations.

Adaptive Resolution Simulation of Supramolecular Water: The Concurrent Making, Breaking, and Remaking of Water Bundles.

The adaptive resolution scheme (AdResS) is a multiscale molecular dynamics simulation approach that can concurrently couple atomistic (AT) and coarse-grained (CG) resolution regions, i.e., the molecules can freely adapt their resolution according to their current position in the system. Coupling to supramolecular CG models, where several molecules are represented as a single CG bead, is challenging, but it provides higher computational gains and connection to the established MARTINI CG force field. Difficulties that arise from such coupling have been so far bypassed with bundled AT water models, where additional harmonic bonds between oxygen atoms within a given supramolecular water bundle are introduced. While these models simplify the supramolecular coupling, they also cause in certain situations spurious artifacts, such as partial unfolding of biomolecules. In this work, we present a new clustering algorithm SWINGER that can concurrently make, break, and remake water bundles and in conjunction with the AdResS permits the use of original AT water models. We apply our approach to simulate a hybrid SPC/MARTINI water system and show that the essential properties of water are correctly reproduced with respect to the standard monoscale simulations. The developed hybrid water model can be used in biomolecular simulations, where a significant speed up can be obtained without compromising the accuracy of the AT water model.

Necessity of high-resolution for coarse-grained modeling of flexible proteins.

The popular MARTINI coarse-grained (CG) force field requires the protein structure to be fixed, and is unsuitable for simulating dynamic processes such as protein folding. Here, we examine the feasibility of developing a flexible protein model within the MARTINI framework. The results demonstrate that the MARTINI CG scheme does not properly describe the volume and packing of protein backbone and side chains and leads to excessive collapse without structural restraints in explicit CG water. Combining atomistic protein representation with the MARTINI CG solvent, such as in the PACE model, dramatically improves description of flexible protein conformations. Yet, the CG solvent is insufficient to capture the conformational dependence of protein-solvent interactions, and PACE is unable to properly model context dependent conformational transitions. Taken together, high physical resolution at or near the atomistic level is likely necessary for flexible protein models with explicit, microscopic solvents, and the coarse-graining needs to focus on possible simplification in interaction potentials. © 2016 Wiley Periodicals, Inc.

Effects of H2A Histone Variants on DNA Sequence and Nucleosome Structure using Coarse Grain Simulations

Coarse grain molecular dynamics (CG-MD) simulations are seeing a rising tendency in a wide range of applications because their potential to enhance the sampling of all-atomistic simulations (AA-MD) and the capacity to make near-atomistic millisecond-timescale simulations practical, putting the second threshold on the horizon. In this field, the coarse-grained Martini force field has a prominent position and has recently extended its already range of applications to DNA-containing biomolecules [1].Eukaryotic genomic DNA exists as highly compacted nucleosome arrays called chromatin. Each nucleosome contains a 147-base-pair (bp) stretch of DNA, which is sharply bent and tightly wrapped around a histone protein octamer [2]. Several mechanisms regulate DNA accessibility, including replacement of canonical histones with specialized histone variants. Among the core histone variants, the H2A family is the biggest one -to date 5 histones variants are documented- showing also the highest sequence divergence among histone families. H2A variants show striking differences in three sites that are critical in intra- and inter-nucleosome interactions: the docking domain, which is close to the DNA entry/exit region, the L1 loop and the acidic patch, which is involved in nucleosome-nucleosome interactions. We have studied the structural and dynamic differences in those regions among H2A histone variants-containing nucleosome structures.Furthermore, it has been shown that DNA sequence effects vary between nucleosomes resulting in a significant factor in their stability. For this reason, we have also analyzed the differences in the associated DNA flexibility along the wrapped DNA sequence derived from MD simulations.

PlexinA3 Trans- and Juxtamembrane Dimer Helix Association

Plexins are a group of protein receptors that have a single transmembrane helix that associates with another plexin to form an active dimer. These receptors are involved in a wide range tissue development (neurons, vascular and lymphatic system) and implicated cancer metastasis (lung, pancreatic, and breast). The objective of this study is to understand the transmembrane (TM) and juxtamembrane (JM) association of the Danio rerio PlexinA3 helical TM and JM domain (1241-1311). The structure of this peptide was built through homology modeling of a known crystal structure of a similar plexin. Molecular dynamics (MD) simulations of helical dimers were run using the MARTINI coarse-grained (CG) force field for 10 μs with triplicate runs. The two peptides had their TM region of the protein placed in a 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine (POPC) membrane so association is not biased to an initial structure. The wildtype and four mutants were simulated to investigate how potential destructive and associative mutants influence helix dimerization. Our MD simulations show that JM destructive mutants result in a reduced dimerization in this region but enhances TM association. The mutants that showed enhanced dimerization with an AraTM assay showed an increase in TM association. For a select set of CG simulations, all-atom CHARMM36 (C36) simulations were run for 300ns each to probe structural changes after association with the more flexible and accurate C36 force field.

Multiscale Molecular Dynamics Simulations of Model Hydrophobically Modified Ethylene Oxide Urethane Micelles.

The flower-like micelles of various aggregation numbers of a model hydrophobically modified ethylene oxide urethane (HEUR) molecule, C16E45C16, and their corresponding starlike micelles, containing the surfactants C16E22 and C16E23, were studied by atomistic and coarse-grained molecular dynamic (MD) simulations. We used free energies from umbrella sampling to calculate the size distribution of micelle sizes and the average time for escape of a hydrophobic group from the micelle. Using the coarse-grained MARTINI force field, the most probable size of the model HEUR molecule was thereby determined to be about 80 hydrophobes per micelle and the average hydrophobe escape time to be about 0.1 s, both of which are consistent with previous experimental studies. Atomistic simulations reveal that hydrogen bond formation and the mean lifetime of hydration waters of the poly(ethylene oxide) (or PEO) groups are location-dependent in the HEUR micelle, with PEO groups immediately adjacent to the C16 groups forming the fewest hydrogen bonds with water and having hydration waters with longer lifetimes than those of the PEO groups located further away from the C16 groups.

Reversible and Irreversible Adsorption Energetics of Poly(ethylene glycol) and Sorbitan Poly(ethoxylate) at a Water/Alkane Interface.

We simulate poly(ethylene glycol) (PEG) oligomers and model Tween 80 (polyoxyethylene sorbitan monooleate) molecules at water/alkane interfaces. Using the weighted histogram analysis method (WHAM), including an extension of WHAM to two reaction coordinates to remove hysteresis, we calculate interfacial potentials of mean force (PMFs) for PEG and Tween 80 using three force fields: the atomistic GROMOS 53a6OXY+D and two coarse-grained (CG) MARTINI force fields. Because the force fields have not yet been validated for PEO adsorption to hydrophobic interfaces, we calculate PMFs for alcohol ethoxylates C12E2 and C12E8 and find that they agree with semiempirical results of Mulqueen and Blankschtein [Langmuir 2002, 18 (2), 365-376] for the GROMOS 53a6OXY+D force field, whereas for both MARTINI force fields, PEO adsorbs too weakly to a clean hydrophobic interface. One MARTINI force field incorrectly shows depletion rather than adsorption to a clean hydrophobic interface. We find that the adsorption free energy for PEG oligomers at a clean, planar water/alkane interface is around 1.3 kBT per monomer for the atomistic force field but is less than half of this for the two CG force fields. With the newly validated GROMOS 53a6OXY+D force field, we bracket the dilute adsorption free energy for a model Tween 80 molecule at the clean water/squalane interface. We also calculate the pressure-area isotherm. We exploit these data with the Nikas-Mulqueen-Blankschtein (NMB) theory and a simple transport model to demonstrate a transition from irreversible to reversible adsorption with increasing surface coverage, consistent with experimental results of Reichert and Walker [Langmuir 2013, 29 (6), 1857-1867].

Spontaneous adsorption of coiled-coil model peptides K and E to a mixed lipid bilayer.

A molecular description of the lipid-protein interactions underlying the adsorption of proteins to membranes is crucial for understanding, for example, the specificity of adsorption or the binding strength of a protein to a bilayer, or for characterizing protein-induced changes of membrane properties. In this paper, we extend an automated in silico assay (DAFT) for binding studies and apply it to characterize the adsorption of the model fusion peptides E and K to a mixed phospholipid/cholesterol membrane using coarse-grained molecular dynamics simulations. In addition, we couple the coarse-grained protocol to reverse transformation to atomistic resolution, thereby allowing to study molecular interactions with high detail. The experimentally observed differential binding of the peptides E and K to membranes, as well as the increased binding affinity of helical over unstructered peptides, could be well reproduced using the polarizable Martini coarse-grained (CG) force field. Binding to neutral membranes is shown to be dominated by initial binding of the positively charged N-terminus to the phospholipid headgroup region, followed by membrane surface-aligned insertion of the peptide at the interface between the hydrophobic core of the membrane and its polar headgroup region. Both coarse-grained and atomistic simulations confirm a before hypothesized snorkeling of lysine side chains for the membrane-bound state of the peptide K. Cholesterol was found to be enriched in peptide vicinity, which is probably of importance for the mechanism of membrane fusion. The applied sequential multiscale method, using coarse-grained simulations for the slow adsorption process of peptides to membranes followed by backward transformation to atomistic detail and subsequent atomistic simulations of the preformed peptide-lipid complexes, is shown to be a versatile approach to study the interactions of peptides or proteins with biomembranes.

Molecular dynamics simulations of the bacterial UraA H+-uracil symporter in lipid bilayers reveal a closed state and a selective interaction with cardiolipin.

The Escherichia coli UraA H+-uracil symporter is a member of the nucleobase/ascorbate transporter (NAT) family of proteins, and is responsible for the proton-driven uptake of uracil. Multiscale molecular dynamics simulations of the UraA symporter in phospholipid bilayers consisting of: 1) 1-palmitoyl 2-oleoyl-phosphatidylcholine (POPC); 2) 1-palmitoyl 2-oleoyl-phosphatidylethanolamine (POPE); and 3) a mixture of 75% POPE, 20% 1-palmitoyl 2-oleoyl-phosphatidylglycerol (POPG); and 5% 1-palmitoyl 2-oleoyl-diphosphatidylglycerol/cardiolipin (CL) to mimic the lipid composition of the bacterial inner membrane, were performed using the MARTINI coarse-grained force field to self-assemble lipids around the crystal structure of this membrane transport protein, followed by atomistic simulations. The overall fold of the protein in lipid bilayers remained similar to the crystal structure in detergent on the timescale of our simulations. Simulations were performed in the absence of uracil, and resulted in a closed state of the transporter, due to relative movement of the gate and core domains. Anionic lipids, including POPG and especially CL, were found to associate with UraA, involving interactions between specific basic residues in loop regions and phosphate oxygens of the CL head group. In particular, three CL binding sites were identified on UraA: two in the inner leaflet and a single site in the outer leaflet. Mutation of basic residues in the binding sites resulted in the loss of CL binding in the simulations. CL may play a role as a “proton trap” that channels protons to and from this transporter within CL-enriched areas of the inner bacterial membrane.