CHARMM Research Articles

Influence of Polarization on Carbohydrate Hydration: A Comparative Study Using Additive and Polarizable Force Fields

Carbohydrates are known to closely modulate their surrounding solvent structures and influence solvation dynamics. Spectroscopic investigations studying far-IR regions (below 1000 cm(-1)) have observed spectral shifts in the libration band (around 600 cm(-1)) of water in the presence of monosaccharides and polysaccharides. In this paper, we use molecular dynamics simulations to gain atomistic insight into carbohydrate-water interactions and to specifically highlight the differences between additive (nonpolarizable) and polarizable simulations. A total of six monosaccharide systems, α and β anomers of glucose, galactose, and mannose, were studied using additive and polarizable Chemistry at HARvard Macromolecular Mechanics (CHARMM) carbohydrate force fields. Solvents were modeled using three additive water models TIP3P, TIP4P, and TIP5P in additive simulations and polarizable water model SWM4 in polarizable simulations. The presence of carbohydrate has a significant effect on the microscopic water structure, with the effects being pronounced for proximal water molecules. Notably, disruption of the tetrahedral arrangement of proximal water molecules was observed due to the formation of strong carbohydrate-water hydrogen bonds in both additive and polarizable simulations. However, the inclusion of polarization resulted in significant water-bridge occupancies, improved ordered water structures (tetrahedral order parameter), and longer carbohydrate-water H-bond correlations as compared to those for additive simulations. Additionally, polarizable simulations also allowed the calculation of power spectra from the dipole-dipole autocorrelation function, which corresponds to the IR spectra. From the power spectra, we could identify spectral signatures differentiating the proximal and bulk water structures, which could not be captured from additive 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.

Probing the accuracy of reactive and non-reactive force fields to describe physical and chemical properties of graphene-oxide

Graphene-oxide (GO) has been shown to be a promising material for numerous technological applications. Extensive computational work has been devoted to elucidate the structure, stability and properties of different GO species. As some applications require large GO sizes, classical molecular dynamics simulations are necessary to fully investigate GO properties and behavior. There are few classical force fields parameterized to examine oxidized hydrocarbon compounds, and not all of them are able to simulate GO structures. Here, we present the first comparative study of some GO properties between three classical force fields: Reactive Empirical Bond Order for carbon, hydrogen and oxygen (REBO-CHO), third generation of the Charge Optimized Many Body (COMB3) and Chemistry at HARvard Macromolecular Mechanics (CHARMM) force field. The chemical and physical properties tested include key binding energies, carbon–oxygen bond distances and elastic modulus. When compared to density functional theory calculations or experimental data, the COMB3 reactive force field is shown to provide the best overall results, while REBO-CHO and CHARMM provide good results for certain properties of most GO-systems.

Modeling, molecular dynamics, and docking assessment of transcription factor rho: a potential drug target in Brucella melitensis 16M.

The zoonotic disease brucellosis, a chronic condition in humans affecting renal and cardiac systems and causing osteoarthritis, is caused by Brucella, a genus of Gram-negative, facultative, intracellular pathogens. The mode of transmission and the virulence of the pathogens are still enigmatic. Transcription regulatory elements, such as rho proteins, play an important role in the termination of transcription and/or the selection of genes in Brucella. Adverse effects of the transcription inhibitors play a key role in the non-successive transcription challenges faced by the pathogens. In the investigation presented here, we computationally predicted the transcription termination factor rho (TtFRho) inhibitors against Brucella melitensis 16M via a structure-based method. In view the unknown nature of its crystal structure, we constructed a robust three-dimensional homology model of TtFRho’s structure by comparative modeling with the crystal structure of the Escherichia coli TtFRho (Protein Data Bank ID: 1PVO) as a template in MODELLER (v 9.10). The modeled structure was optimized by applying a molecular dynamics simulation for 2 ns with the CHARMM (Chemistry at HARvard Macromolecular Mechanics) 27 force field in NAMD (NAnoscale Molecular Dynamics program; v 2.9) and then evaluated by calculating the stereochemical quality of the protein. The flexible docking for the interaction phenomenon of the template consists of ligand-related inhibitor molecules from the ZINC (ZINC Is Not Commercial) database using a structure-based virtual screening strategy against minimized TtFRho. Docking simulations revealed two inhibitors compounds – ZINC24934545 and ZINC72319544 – that showed high binding affinity among 2,829 drug analogs that bind with key active-site residues; these residues are considered for protein-ligand binding and unbinding pathways via steered molecular dynamics simulations. Arg215 in the model plays an important role in the stability of the protein-ligand complex via a hydrogen bonding interaction by aromatic-π contacts, and the ADMET (absorption, distribution, metabolism, and excretion) analysis of best leads indicate nontoxic in nature with good potential for drug development.

Molecular Dynamics Simulations of Sphingomyelin-Cholesterol Bilayers

Sphingolipids are one of the major components in animal cell membranes. One type of sphingolipids is sphingomyelins which are abundant in myelin sheath surrounding nerve cell axons. The two common forms of sphingomyelin targeted in this research are palmitoylsphingomyelin (PSM) and stearoylsphingomyelin (SSM). Another bilayer component included in our membranes that has high affinity to sphingomyelins is cholesterol, which aids in maintaining membrane fluidity and structural integrity. Sphingomyelins and cholesterol make up a great part of highly dynamic membrane domains, termed lipid rafts. With that being said, this work tests the accuracy of molecular dynamics (MD) simulations with the CHARMM36 (C36) lipid force field for PSM and SSM, each SM bilayer has varying concentrations of cholesterol. Properties such as surface area per lipid, x-ray and neutron form factors, and chain deuterium order parameters are to be compared to those yielded from past experiments to prove C36's ability to correctly simulate the mixed SM-cholesterol lipid bilayers. Based on past studies on C36 lipid force field, we expect to see excellent agreement between experimental and MD simulation data for more complex mixtures of membranes that contain SM lipids.

Investigating Lipid Phase Changes from Liquid Crystalline to Ripple to Gel Phases with All-Atom Molecular Dynamics Simulations

Lipid mixed with water can exist in many different phases that depends on various factors, such as hydration, temperature and pressure (J. Chem. Theory Comput.,6 (8): 2488 (2010)). The liquid crystalline (chain-disordered state) is a well-studied phase for single lipid bilayers both experimentally and computationally. At low enough temperatures or hydration levels, this chain-disordered state can change to a gel state with high chain order and a certain chain tilt with respect to the membrane normal. Depending on the lipid, an intermediate phase can exist between the gel and liquid crystalline phase, which is known as the ripple phase. The main objective of this study was probing the accuracy of the CHARMM36 (C36) force field (FF) (J. Phys. Chem. B.,114 (23): 7830 (2010)) in predicting phase changes in lipid bilayers. Pure and mixed lipid bilayers at different compositions of 1,2-dimyristoyl-sn-glycero-3-phosphocholine (DMPC) and dipalmitoylphosphatidylcholine (DPPC) were studied using molecular dynamics (MD) simulations. Based on the pure gel-phase transition temperature of each lipid, a range of temperatures was selected (from 200C to 400C) (Biochem. 18: 3280 (1979)). MD simulations were able to capture ripple phase in %25 DMPC and %75 DPPC mixture and pure DMPC at 250C. Simulations were run for 200-300 ns. MD simulations also show either a gel state or gel-like state depending on if the system gets trapped (gel-like is a state without proper leaflet alignment). Phase diagram was compared to that obtained from experimental data such as NMR. Based on the phase diagram and our transition temperatures, C36 FF accurately predicts the phase transitions of this fully saturated PC lipids. Therefore, studies on phase coexistence of liquid ordered and liquid disordered domains are likely to be accurate using the C36 FF.

Web-based computational chemistry education with CHARMMing II: Coarse-grained protein folding.

A lesson utilizing a coarse-grained (CG) G-like model has been implemented into the CHARMM INterface and Graphics (CHARMMing) web portal (www.charmming.org) to the Chemistry at HARvard Macromolecular Mechanics (CHARMM) molecular simulation package. While widely used to model various biophysical processes, such as protein folding and aggregation, CG models can also serve as an educational tool because they can provide qualitative descriptions of complex biophysical phenomena for a relatively cheap computational cost. As a proof of concept, this lesson demonstrates the construction of a CG model of a small globular protein, its simulation via Langevin dynamics, and the analysis of the resulting data. This lesson makes connections between modern molecular simulation techniques and topics commonly presented in an advanced undergraduate lecture on physical chemistry. It culminates in a straightforward analysis of a short dynamics trajectory of a small fast folding globular protein; we briefly describe the thermodynamic properties that can be calculated from this analysis. The assumptions inherent in the model and the data analysis are laid out in a clear, concise manner, and the techniques used are consistent with those employed by specialists in the field of CG modeling. One of the major tasks in building the G-like model is determining the relative strength of the nonbonded interactions between coarse-grained sites. New functionality has been added to CHARMMing to facilitate this process. The implementation of these features into CHARMMing helps automate many of the tedious aspects of constructing a CG G model. The CG model builder and its accompanying lesson should be a valuable tool to chemistry students, teachers, and modelers in the field.

Molecular dynamics simulations of palmitoyloleoylphosphatidylglycerol bilayers

Phosphatidylglycerol (PG) is one of the important components of biological membranes, but there is a paucity of experimental data to test the accuracy of molecular dynamics (MD) simulations. This work consists of testing the accuracy of the CHARMM36 (C36) lipid force field on 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphoglycerol (POPG) lipid bilayers. MD simulations of POPG lipid bilayers are compared to recently available X-ray and neutron scattering and deuterium NMR measurements. Overall, the C36 lipid force field accurately represents the X-ray and neutron form factors, bilayer and hydrocarbon thicknesses and chain deuterium order parameters. The surface area per lipid from MD simulations with C36 (67.7 ± 0.2 Å2) is in excellent agreement with the experimentally determined value of 66.0 ± 1.3 Å2. C36 outperforms the lipid force field developed by Berger et al. [15] and suggests that past studies with this force field may result in lateral areas that are too small. Moreover, our studies give some insight into the structural model used in experiments and suggest that the functional form for the head group may not be Gaussian-like. Based on our simulations, the POPG lipid in the C36 lipid force field is well parameterised and can be used for other PG lipids and membrane models with mixed lipids.

A systematic molecular dynamics simulation study of temperature dependent bilayer structural properties

Although lipid force fields (FFs) used in molecular dynamics (MD) simulations have proved to be accurate, there has not been a systematic study on their accuracy over a range of temperatures. Motivated by the X-ray and neutron scattering measurements of common phosphatidylcholine (PC) bilayers (Kučerka et al. BBA. 1808: 2761, 2011), the CHARMM36 (C36) FF accuracy is tested in this work with MD simulations of six common PC lipid bilayers over a wide range of temperatures. The calculated scattering form factors and deuterium order parameters from the C36 MD simulations agree well with the X-ray, neutron, and NMR experimental data. There is excellent agreement between MD simulations and experimental estimates for the surface area per lipid, bilayer thickness (DB), hydrophobic thickness (DC), and lipid volume (VL). The only minor discrepancy between simulation and experiment is a measure of (DB−DHH)/2 where DHH is the distance between the maxima in the electron density profile along the bilayer normal. Additional MD simulations with pure water and heptane over a range of temperatures provide explanations of possible reasons causing the minor deviation. Overall, the C36 FF is accurate for use with liquid crystalline PC bilayers of varying chain types and over biologically relevant temperatures.

Application of molecular simulation to investigate chrome(III)-crosslinked collagen problems

Molecular dynamics simulation with a modified CHARMM (Chemistry at Harvard Macromolecular Mechanics) force field was carried out to investigate the properties of chrome-tanned collagen in comparison with chrome-free collagen under hydrated and dehydrated conditions. An attempt has been made to explain the microcosmic origins of the various properties of the chromium(III)-crosslinked collagen. The present simulation describes the clear crosslinking topology of polychromiums to peptide chains, identifies the linking site and the capacity of the linkage, explains why the efficiency is not 100% in a practical tanning process and provides a new viewpoint on the crosslinking of the polychromium with the side chains of the collagen.

CHARMM36 United Atom Chain Model for Lipids and Surfactants

Molecular simulations of lipids and surfactants require accurate parameters to reproduce and predict experimental properties. Previously, a united atom (UA) chain model was developed for the CHARMM27/27r lipids (Hénin, J., et al. J. Phys. Chem. B. 2008, 112, 7008-7015) but suffers from the flaw that bilayer simulations using the model require an imposed surface area ensemble, which limits its use to pure bilayer systems. A UA-chain model has been developed based on the CHARMM36 (C36) all-atom lipid parameters, termed C36-UA, and agreed well with bulk, lipid membrane, and micelle formation of a surfactant. Molecular dynamics (MD) simulations of alkanes (heptane and pentadecane) were used to test the validity of C36-UA on density, heat of vaporization, and liquid self-diffusion constants. Then, simulations using C36-UA resulted in accurate properties (surface area per lipid, X-ray and neutron form factors, and chain order parameters) of various saturated- and unsaturated-chain bilayers. When mixed with the all-atom cholesterol model and tested with a series of 1,2-dimyristoyl-sn-glycero-3-phosphocholine (DMPC)/cholesterol mixtures, the C36-UA model performed well. Simulations of self-assembly of a surfactant (dodecylphosphocholine, DPC) using C36-UA suggest an aggregation number of 53 ± 11 DPC molecules at 0.45 M of DPC, which agrees well with experimental estimates. Therefore, the C36-UA force field offers a useful alternative to the all-atom C36 lipid force field by requiring less computational cost while still maintaining the same level of accuracy, which may prove useful for large systems with proteins.

Mutation induced structural variation in membrane proteins

Point mutations on membrane proteins may lead to small structural variations. Prediction of such structural variations can help to further understand the related bio-activities of membrane proteins. We constructed fifteen hybrid energy functions on the basis of Chemistry at Harvard Macromolecular Mechanics(CHARMM) force field, hydrogen bonding potential and distance-scaled, finite ideal-gas reference(DFIRE)-like statistical energies, and evaluated their performance on a representative dataset of homologous membrane proteins via a newly developed all-atom replica exchange Monte Carlo algorithm. The energy function composed of CHARMM and hydrogen bonding potential has the best performance, and the original DFIRE potential shows much better performance than the DFIRE-like potentials constructed from membrane proteins. We can conclude that more membrane protein structures with high resolution are necessary for the construction of robust prediction method of mutation induced membrane protein structure variations.

CHARMM36 all-atom additive protein force field: Validation based on comparison to NMR data

Protein structure and dynamics can be characterized on the atomistic level with both nuclear magnetic resonance (NMR) experiments and molecular dynamics (MD) simulations. Here, we quantify the ability of the recently presented CHARMM36 (C36) force field (FF) to reproduce various NMR observables using MD simulations. The studied NMR properties include backbone scalar couplings across hydrogen bonds, residual dipolar couplings (RDCs) and relaxation order parameter, as well as scalar couplings, RDCs, and order parameters for side-chain amino- and methyl-containing groups. It is shown that the C36 FF leads to better correlation with experimental data compared to the CHARMM22/CMAP FF and suggest using C36 in protein simulations. Although both CHARMM FFs contains the same nonbond parameters, our results show how the changes in the internal parameters associated with the peptide backbone via CMAP and the χ1 and χ2 dihedral parameters leads to improved treatment of the analyzed nonbond interactions. This highlights the importance of proper treatment of the internal covalent components in modeling nonbond interactions with molecular mechanics FFs.

Developing CHARMM-Compatible Lipid Parameters for Ceramides and United Atom Chains

Biological membranes form a barrier to protect the cell from its environment and consist of a wide variety of lipids. In molecular simulations, accurate force field (FF) parameters are crucial for representing biological membranes. Although all-atom FFs, such as CHARMM36 (C36), can now accurately represent a variety of lipid chains (saturated, unsaturated, and branched) and head groups (PC, PE, PG, and PS), parameters for ceramide lipids are lacking. These lipids are common in biology (neuronal and ocular lens membranes) and are involved in lipid domain formation with cholesterol. Quantum mechanical calculations are used to develop torsional profiles and partial charges of the head group. Molecular dynamics (MD) simulations of palmitoylsphingomyelin (PSM) are used to test new parameters in comparison with available deuterium order parameters. The goal of this new lipid parameter set is to obtain the proper phase of PSM as a function of temperature, where previous models have failed. The development of a CHARMM-compatible united atom (UA) chain model will also be presented, referred to here as C36-UA. This is an extension of previous work (Hénin et al. JPC, 112: 7008 (2008)) to the C36 FF. MD simulations of pure and mixed bilayers with cholesterol demonstrate a UA-chain model that is accurate and compatible with the CHARMM family of force fields. MD simulations of micelle formation of short-chain lipids and single-chained detergents are used to demonstrate the ability of C36-UA to self-assemble lipids. MD simulations with C36-UA allow for timescales not approachable with all-atom models without a loss of accuracy, which may have applications to lipid organization and studies with membrane-associated proteins.

Improving the CHARMM Force Field for Polyunsaturated Fatty Acid Chains

CHARMM36 (C36) is the most up-to-date pairwise additive all-atom lipid force field and is able to accurately represent bilayer properties of saturated and monounsaturated lipid molecules in the natural constant particle, pressure, and temperature (NPT) ensemble. However, molecular dynamics (MD) simulations on 1-stearoyl-2-docosahexaenoyl-sn-glycerco-3-phosphocholine (SDPC) bilayers of the polyunsaturated fatty acid (PUFA) chains result in inaccuracies of the surface area per lipid (SA), deuterium order parameters (S(CD)), and X-ray form factors. Therefore, in this study, high-level quantum mechanical calculations are used to improve the dihedral potential of neighboring double bonds, and the corresponding force field is referred to as C36p. The SA for SDPC at 303 K increases from 63.2 ± 0.2 (C36) to 70.8 ± 0.2 (C36p) Å(2) and agrees favorably with X-ray diffraction results at 297 K. The resulting S(CD) are in excellent agreement with experimental values of both the sn-1 and sn-2 chains. Calculated NMR (13)C relaxation times and X-ray form factors from MD simulations of SDPC bilayers also agree with experiments. MD simulations of 1,2-diarachidonyl-phosphatidylcholine (DAPC) bilayers are used to further validate our force field parameters on a lipid with both chains containing PUFAs. As expected, the thickness of DAPC bilayers is reduced, and the SA is increased compared to the SDPC bilayers. This update in the PUFA force field should allow for accurate MD simulations of PUFA-containing bilayers in the NPT ensemble.

In-silico designing of acyclic nucleoside phosphonates and their anti-HIV potential

Method The object of our study is to develop newer nucleoside phosphonate analogs bearing unsaturation and modifications in heterogenous bases and prediction of their antiHIV potential. Designing is done keeping the Lipinski’s Rule of Five in focus. The diphosphates of compounds have been docked into the active site of wild type HIVRT (PDB: ID 2B6A). The forcefield of the Chemistry at Harvard Macromolecular mechanics (CHARMm) was applied to 3D models of PD HIV RT-nevirapine complex and synthesized ligands. The energy function is based on separable internal coordinate terms and pair wise non-bond interaction terms.

Structural and Energetic Details of the DNA-Binding and Cleavage Core of Saccharomyces Cerevisiae Topoisomerases II Linked to DNA through its Active Site Tyrosine

DNA topoisomerases are enzymes that can change the topological state of DNA without changing its primary structure by changing the linking number of the molecule. Topoisomerases are able to solve all the topological problems of DNA that arise during such processes as replication, transcription, recombination or chromosome separation. In our study, S. cerevisiae topoII structure is used to investigate structural and energetic changes of the protein from initial state to final state ( our final state is the DNA-TopoII complex when the protein binds to DNA and the stage just before the DNA cleavage by TopoII). The second goal is to investigate conformational changes of cleavage dependent DNA-gate and C-gate mechanisms of the protein after the protein captures the DNA. Lastly, we aim to get the transition rate constants from free energy profile by employing umbrella sampling method after defining the reaction pathway. In order to investigate the conformational changes during DNA-gate and C-gate opening and/or closing processes, small perturbations are introduced by using HQBMD (Half Quadratic Biased MD) and TMD (Targeted MD) methods by performing a version of CHARMM (Chemistry at HARvard Macromolecular Mechanics).

In Silico Model Escherichia Coli Membranes: Simulating a Lipid with a Cyclopropane Ring

Bacterial membranes are composed mostly of lipids with phosphoethanolamine (PE) and phosphoglycerol (PG) head groups and a variety of fatty acid chains. A defining characteristic of bacterial membranes is the presence of lipids with a cyclic-containing chain. To our knowledge, a phospholipid with such a chain has never been studied in a molecular dynamics simulation. Thus, novel force field parameters to describe a cyclopropane moiety were developed using parameters from the CHARMM36 (C36) general force field as a basis and high-level quantum mechanical energies. Two simple membranes and one complex membrane for use as realistic model Escherichia coli (E. coli) cytoplasmic membranes were designed for future work with secondary active transporters expressed in E. coli. Compositions of the membranes were based on several different experimental methods using an E. coli K12 strain grown on Luria broth. One simple membrane consisted of 85 % PE 18:1/16:0 (POPE) and 15 % PG 18:1/16:0(POPG), and the other of only PE cy17:0/16:0 (PMPE). The complex membrane consisted of six different phospholipids, the most prevalent being the cyclic-containing lipid, PMPE. NPT simulations were carried out at 310 K for 50 ns using the C36 lipid force field and the developed cyclopropane moiety force field for each membrane. NMR deuterium order parameters (SCD), density profiles, and diffusion constants are compared with experiment. The model membranes will provide a basis to study membrane bound proteins or other bound molecules expressed in E. coli with realistic molecular dynamic simulations.

Refining and Testing CHARMM Lipid Parameters for Biologically Important Membranes

Biological membranes form a barrier to protect the cell from its environment and consist of a wide variety of lipids. A recent modification to the CHARMM lipid force filed, CHARMM36 (C36), resulted in significant improvements in deuterium order parameters (SCD), water hydration, and area compressiblities. Moreover, C36 simulations resulted in excellent surface areas per lipid compared to experiment with NPT molecular dynamics (MD). We have extended this force field to aliphatic chain branched lipids, which exists in many bacterial membranes. MD simulations of DPhPC agree with experimental form factors and suggest that current lipid force field parameters are valid for these branched chain lipids. Branching leads to an increase in average surface areas per lipid, area elastic moduli and lipid axial relaxation times. However, MD simulations of mixed bilayers with cholesterol and polyunsaturated lipids suggest that the atomistic force field describing these lipids requires modification (improper sterol orientation and head-to-head group spacings). MD simulations on decalin resulted in a heat of vaporization that is 10 kJ/mol lower than experiment, but modifications to the Lennard-Jones parameters of cholesterol yielded near perfect agreement with experiment. Simulations with these parameters were in excellent agreement with experimental SCDs for the DMPC acyl chain and SCDs for ring and chain positions on cholesterol. Moreover, these simulations agree favorably with experimental form factors of DMPC-cholesterol bilayers. However, simulations with a polyunsaturated lipid, DAPC, and cholesterol still resulted in the abovementioned inaccuracies. Since the thickness of these polyunsaturated bilayers was too large by ∼8 A, high-level quantum mechanical (QM) calculations were used for dihedrals adjacent to the double bonds. It was found that the previous force field restricted the conformational flexibility and MD simulations are underway to test the QM-based potentials accuracy compared to experimental diffraction studies.

A Modified Lipid Force Field for Charmm: Development and Application to Single-Celled Organism Membranes

Biological membranes form a barrier to protect the cell from its environment and selectively control the entrance/exit of small molecules. Molecular simulations of these biological membranes require an accurate lipid force field (a major component of the membrane). Previously, extensive ab initio quantum mechanical (QM) calculations have been used to improve the aliphatic portion of the CHARMM27 lipid force field. Although this was a significant improvement, the lipid head group required additional modifications to agree with experimental lipid bilayer deuterium order parameters (SCD) and solvation free energies. Therefore, we modified the atomic charges in the carbonyl-glycerol region and fit dihedral energy terms to high-level QM calculations and/or experiment. Molecular dynamics (MD) simulations with this new force field, referred to as CHARMM36 (C36), resulted in a significant improvement to the SCD's and water hydration for DPPC lipid bilayers. The calculated electrostatic profile and lipid bilayer surface tension decreased significantly. Consequently, the C36 force field resulted in excellent surface areas per lipid (and other properties) with NPT simulations, which is a significant improvement from the C27r force field that required constant area simulations (NPAT) to prevent some bilayers from laterally condensing. MD simulations of other pure lipid bilayers and monolayers also agreed favorably with experimental densities, monolayer surface tensions, and SCD's. The success of the C36 force field allowed for the study of complex lipid membranes in single-celled organisms. Model membranes were developed and simulated for yeast (six phospholipids, cholesterol, and 25-hydroxysterol) and Chlamydia (five unbranched lipids, a branced lipid, and cholesterol). These membranes are currently being used to study intracellular sterol transport and a porin protein that induces an immune response.