quantitative)functional)analysis)of ...quantitative)functional)analysis)of...

QUANTITATIVE FUNCTIONAL ANALYSIS OF MEMBRANES AND MEMBRANE PROTEINS USING MOLECULAR DYNAMICS SIMULATION Project Number：hp140157 Suyong Re、Takaharu Mori、Yuji Sugita、Yasuaki Komuro、Pai-‐Chi Li、Wataru Nishima、Raimondas Galvelis、Isseki Yu、Po-‐Hung Wang

一般財団法人高度情報科学技術研究機構

第２回「京」を中核とするHPCIシステム利用研究課題成果報告会　（2015年10月26日、日本科学未来館）

ConnecKng “Snapshots” from high-‐resoluKon crystal structures of membrane proteins

SimulaKng funcKonal moKon of membrane proteins by integraKng methods ranged from quantum to coarse-‐grained simulaKons in future.

OBJECTIVES

Ligand binding/extrusion, Chemical reacKon, Structural change Lipid-‐protein interacKon, Membrane dynamics

Conforma=onal and chemical dynamics in terms of free energy landscape

第２回成果報告会（1/11）

GENESIS (Release1) Development Team

Y. Sugita J. Jung T. Mori C. Kobayashi Y. Matsunaga

New Features of GENESIS •  Inverse Lookup Table Scheme for Nonbonded InteracKon.

•  J. Jung et al. J.Comp.Chem. 2013, 34, 2412-‐2420. •  Midpoint Cell Method for Hybrid ParallelizaKon.

•  J. Jung et al. J.Comp.Chem. 2014, 35, 300-‐308. •  New Replica-‐exchange molecular dynamics method (Surface-‐tension REMD).

•  T. Mori et al. J.Chem.Theor.Comp. (2013) 9, 5629-‐5640. •  3D-‐FFT based on the volumetric decomposiKons.

•  J. Jung et al. submiaed. •  SoSware Focus on GENESIS

•  J. Jung, T. Mori, et al. WIREs Comput Mol Sci 2015, 5:310–323. doi: 10.1002/wcms.1220

N. Takase T. Ando K. Yagi T. Ando

New MD soSware from RIKEN AICS GENESIS (Generalized-‐Ensemble Simula=on System)


OUTCOMES

CHARMM Force-Fields with Modified Polyphosphate ParametersAllow Stable Simulation of the ATP-Bound Structure of Ca2+-ATPaseYasuaki Komuro,†,‡,§ Suyong Re,‡ Chigusa Kobayashi,§ Eiro Muneyuki,† and Yuji Sugita*,‡,§,∥,⊥

†Graduate School of Science and Engineering, Chuo University, 1-13-27, Kasuga, Bunkyo-ku, Tokyo 112-8551, Japan‡RIKEN Theoretical Molecular Science Laboratory, 2-1, Hirosawa, Wako-shi, Saitama 351-0198, Japan§RIKEN Advanced Institute for Computational Science, International Medical Device Alliance (IMDA) 6F, 1-6-5minatojima-minamimachi, Chuo-ku, Kobe, Hyogo 650-0047, Japan∥RIKEN Quantitative Biology Center, International Medical Device Alliance (IMDA) 6F, 1-6-5 minatojima-minamimachi, Chuo-ku,Kobe, Hyogo 650-0047, Japan⊥RIKEN iTHES, 2-1, Hirosawa, Wako-shi, Saitama 351-0198, Japan

*S Supporting Information

ABSTRACT: Adenosine triphosphate (ATP) is an indispen-sable energy source in cells. In a wide variety of biologicalphenomena like glycolysis, muscle contraction/relaxation, andactive ion transport, chemical energy released from ATPhydrolysis is converted to mechanical forces to bring aboutlarge-scale conformational changes in proteins. Investigation ofstructure−function relationships in these proteins by moleculardynamics (MD) simulations requires modeling of ATP insolution and ATP bound to proteins with accurate force-fieldparameters. In this study, we derived new force-field parametersfor the triphosphate moiety of ATP based on the high-precisionquantum calculations of methyl triphosphate. We tested our newparameters on membrane-embedded sarcoplasmic reticulumCa2+-ATPase and four soluble proteins. The ATP-boundstructure of Ca2+-ATPase remains stable during MD simulations, contrary to the outcome in shorter simulations using originalparameters. Similar results were obtained with the four ATP-bound soluble proteins. The new force-field parameters were alsotested by investigating the range of conformations sampled during replica-exchange MD simulations of ATP in explicit water.Modified parameters allowed a much wider range of conformational sampling compared with the bias toward extended formswith original parameters. A diverse range of structures agrees with the broad distribution of ATP conformations in proteinsdeposited in the Protein Data Bank. These simulations suggest that the modified parameters will be useful in studies of ATP insolution and of the many ATP-utilizing proteins.

■ INTRODUCTIONProteins in cells function as molecular machines under theinfluence of significant disturbance from thermal noise. Motorproteins and membrane transporting proteins commonly utilizeadenosine triphosphate (ATP) as a substrate to carry out theirfunctions in such conditions. They convert chemical energyreleased from ATP hydrolysis to mechanical forces andundergo large-scale conformational changes to effect theirspecific function. Molecular mechanisms underlying the energyconversion have been investigated extensively both exper-imentally and theoretically. X-ray crystal structures of ATP-bound proteins are, in particular, useful for understandingfunction at an atomic level. Approximately 1300 structures ofATP-bound proteins are in the Protein Data Bank (http://www.pdb.org/).Ca2+-ATPase of skeletal muscle sarcoplasmic reticulum

(SERCA1a) is, structurally and functionally, one of the best-

studied proteins utilizing ATP. The ATPase is an integralmembrane protein that transports two Ca2+ from the cytoplasminto the lumen of the sarcoplasmic reticulum (SR) against a104-fold concentration gradient.1 According to classical E1/E2theory, the transmembrane Ca2+-binding sites have high affinityfor Ca2+ and face the cytoplasm in the E1 state, whereas theyhave low affinity and face the lumen of the SR in the E2state.2−4 Biochemical studies established that ATP binds on thecytoplasmic side, and X-ray crystallography placed the bindingsite at P and N domains and provided atomic coordinates of thebound ATP conformation.5,6 There are still unsolved questionsconcerning how ATP contributes to ATPase function.7−12 Howis the chemical energy from ATP hydrolysis converted intomechanical forces that transfer bound Ca2+ to the lumen,

Received: May 13, 2014Published: August 21, 2014

Article

pubs.acs.org/JCTC

© 2014 American Chemical Society 4133 dx.doi.org/10.1021/ct5004143 | J. Chem. Theory Comput. 2014, 10, 4133−4142

Accurate modeling of ATP bound state Y. Komuro et al., J. Chem. Theory Comput. (2014)

Proton transport of mulK-‐drug transporter W. Nishima et al., submi2ed

Replica state exchange metadynamics R. Galvelis & Y. Sugita, J. Comput. Chem. (2015)

Modeling of OM protein (S. Re)

HydraKon structure at lipid/water interface S. Re et al., J. Phys. Chem. Le2. (2014)

Sphingomyelin cluster in membrane (P. Li)


COMPUTATIONAL MODELING OF OUTER MEMBRANE PROTEINS

CollaboraKon with Prof. W. Im in The University of Kansas 第２回成果報告会（4/11）

Challenge for an=bio=c transport through outer membrane Bacteria cell wall

Lipopolysaccharide C. Erridge et al. Micr. Infect. (2002) Outer&core�Inner&core� Lipid&A�O0an2gen&polysaccharide&chain�

repea2ng&unit�1�

1�

2�

3�

4�

5�

6�

7�

8�

9�

10�

11�

12�14�15�13�

16�

GlcN�

Kdo�Hep�

Glc�

Gal�GlcNAc�Man�GalNAc�

Glc�

PO42–�

CO2–�

Highly&variable,&Primary&target&for&an2body�

Very&highly&conserved�

Highly&conserved&

Common&sugars�

Over&160&iden2fied&O0serotypes&for&E0coli,&

only&five&core&structures&(R10R4&and&K12)�

hap://semoneapbiofinalexamreview.wikispaces.com

OUTER MEMBRANE PROTEIN F (OMPF) 第２回成果報告会（5/11）

Proteins in the outer membrane of Escherichia coli reside in anasymmetric bilayer in which the outer leaflet is made of theglycolipid lipopolysaccharide and the inner leaflet is composedof phospholipids with either the phosphatidylglycerol (PG) orphosphatidylethanolamine (PE) head group (16, 17). Thisasymmetry can be recreated in vitro with planar bilayers, butsuch bilayers do not remain stable on the time scale of OMPfolding experiments (18) and therefore are not amenable tofolding studies. Instead, vesicles composed entirely of phospho-lipids have been used to study OMP folding (11–15). Despitethe asymmetry of outer membranes, it is likely that phospho-lipids are the most appropriate model for folding studies,because OMPs first encounter the inner leaflet (composedentirely of phospholipids) as they fold into the outermembranein vivo (10).Themost extensively studied!-barrel foldingmodel in phos-

pholipid vesicles is outermembrane proteinA (OmpA). Studiesof OmpAhave revealed how insertion occurs (19–23) and havemeasured the stability of the native structure in different lipidenvironments (24–27). The principles garnered from thesestudies allow conclusions to be drawn for how OmpA behaves,but they cannot reliably be applied to all OMPs until the behav-iors of other proteins are observed in the same environment.Moreover, folding studies have been performed on otherOMPs(12–15), but no comprehensive folding screen exists that facil-itates comparison between proteins. To directly compare thefolding propensities ofmembrane proteins, we probed the fold-ing conditions of nine different !-barrel OMPs: OmpX,OmpW, OmpA, the crcA gene product (PagP), OmpT, outermembrane phospholipase A (OmpLa), the fadl gene product(FadL), the yaet gene product (Omp85), and OmpF.The OMPs we chose all reside in the outer membrane of

E. coli. Despite inhabiting the same bilayer environment in vivoand sharing a common TM motif, the primary sequences ofthese nine OMPs could not be aligned altogether or in pairwiseBLAST queries (data not shown). Furthermore, the structure ofeach OMP varies from the next (Fig. 1). These model OMPshave barrel sizes ranging from eight!-strands (OmpX,OmpW,OmpA, and PagP) to 16 !-strands (OmpF). Their extramem-brane structures also vary.OmpAandOmp85have periplasmic

domains as large as their TMdomains, whereas FadL and OmpThave significant amounts of struc-ture extending from their barrelstoward the extracellular side of thebiological membrane.The study of these nine OMPs

constitutes the largest set of OMPsevaluated in tandem to date. In thiswork, we have established OmpW,OmpT, OmpLa, FadL, and Omp85as novel models for folding studies.For purposes of direct comparison,we included OMPs that have previ-ously been shown to fold into phos-pholipid bilayers in vitro: OmpA(11), OmpF (12), OmpX (28), andPagP (15). Our data set doubles the

number of E. coli OMPs for which folding has been systemati-cally examined in lipid bilayers. Further, because these OMPsare sequentially and structurally diverse yet fold into the sameenvironment in vivo, our results allowed the deduction of broadrules that define folding similarities and differences.

EXPERIMENTAL PROCEDURES

Vesicle Preparation—Lipids dissolved in chloroform (AvantiPolar Lipids) were dried to a thin film in glass vials under agentle stream of nitrogen gas. Lipid films were evacuated for atleast 3 h to remove residual solvent and stored at !20 °C untiluse. Lipid films were reconstituted in buffer containing 2 mMEDTA (Fluka) and 10 mM borate (Sigma), pH 10. Vesicles usedin pH studies were brought up in the same concentration ofappropriate buffers at various pH values. To make small unila-mellar vesicles (SUVs), lipids reconstituted in buffer were son-icated on ice for 50 min with a 50% duty cycle with a BransonSonifier as described previously (25). Large unilamellar vesicles(LUVs) were made by extruding reconstituted lipids 11 timesthrougha0.1"Mfilter using amini-extruder (Avanti PolarLipids).Cloning and Expression of OMPs—Primers were designed to

encompass the mature forms of the OMPs and add NdeI (5")and BamHI (3") sites. Primers are listed in supplemental TableS1. OMP genes were amplified using ExTaq polymerase(Takara) from an overnight growth of E. coliK12MG1655 (29).The PCR products were cut with restriction enzymes andligated into a pET11a vector. The resulting plasmids weretransformed into a laboratory supply of electrocompetentDH5# cells. The sequenceswere confirmed by double-strandedDNA sequencing using the T7 promoter and T7 terminatorprimers for all clones. Additional primers were designed andused for Omp85 to cover the length of the insert. The expres-sion products were confirmed by matrix-assisted laser desorp-tion ionization (MALDI) mass spectrometry at the Johns Hop-kins Medical Institute.Plasmids were transformed into BL21(DE3) StarTM cells

(Invitrogen). Transformed cells were grown in 500 ml of LBmedium to an optical density of 0.6 at 600 nmbefore expressionwas induced by the addition of 1 mM !-D-1-thiogalactopyrano-side. Cells continued to incubate for 3–4 h at 37 °C and were

FIGURE 1. Structures of OMPs used in this study. Structures for each OMP are shown in their relative orien-tation in the outer membrane (gray bar). Across the top of the figure are listed the names and numbers of!-strands that constitute the transmembrane barrel of each protein. Omp85 is postulated to have either 12strands (46) or 16 strands (44), so both numbers are shown. Images were made in PyMOL with the followingfiles from the Protein Data Bank: 1QJ8 (OmpX), 2F1V (OmpW), 1BXW (OmpA), 1THQ (PagP), 1I78 (OmpT), 1QD5(OmpLa), 1T16 (FadL), 2QDF (Omp85 POTRA domains), and 2OMF (OmpF). Protein domains of unknown struc-ture are represented as geometric shapes: a purple oval for the periplasmic region of OmpA, a green oval for oneof the five periplasmic domains of Omp85, and a green square for the transmembrane domain of Omp85.Furthermore, OmpF most often occurs as a trimer, but only a single monomer is shown.

OMP Folding Profile

SEPTEMBER 26, 2008 • VOLUME 283 • NUMBER 39 JOURNAL OF BIOLOGICAL CHEMISTRY 26749

N. K. Burgess et al., J. Biol. Chem. (2008)

OmpX OmpW OmpA PagP

OmpT OmpLa FadL

Omp85

OmpF

β2

(Figure 11(c)), there is an abrupt increase in the ion-accessible area and the maximum density of Cl2 isfound between Lys16 and Arg270. Notably, theposition of the density maximum is now rotatedin a counter-clockwise direction relative to the pos-ition in the constriction zone. This rotation isapparently caused by stabilizing interaction withGlu62 which is located right below the argininecluster. In the periplasmic vestibule (Figure 11(d)),Cl2 tend to stay near the barrel wall close to thebasic residue (Lys10, Lys305, and others) near the3-fold axis whereas Kþ tend to stay close to theacidic residues (Asp92, Asp149, Asp266, andothers) near the outer rim of the pore. In the extra-cellular vestibule (Figure 11(a)), the highest densityof Kþ is observed at the outer rim of the b-barrel(near Asp113, Glu117, and Asp121 on L3, andGlu29 on L1) and the monomeric interface (nearAsp37 and Asp74). The region of high Cl2 density,well separated from that of Kþ, is found aboveArg167 and Arg168. No particular pattern isobserved in the ion distribution in the bulk sol-ution although Kþ has a slightly higher propensitythan Cl2 to be located along the trimeric axis.

To visualize the average preferred pathway ofKþ and Cl2, a superimposition of 100 instan-taneous configurations of the ions in all threepores was constructed. The result is shown fromthree different points of view in Figure 12. It isobserved that the Kþ and Cl2 pathways are verywell separated and span over nearly 40 A alongthe axis of the aqueous pore, roughly fromZ ¼ 220 A to 22 A. The charge separation existsthroughout the pore, though it is more pronouncedin the extracellular vestibule and the constrictionzone than in the periplasmic vestibule (see alsoFigure 11). In the central region of the pore, Kþ

and Cl2 follow two left-handed screw-like path-ways, undergoing a counter-clockwise rotation of1808 from the extracellular vestibule to the pore

periplasmic side. Overall, there appears to be agood correspondence of high densities of Kþ andCl2 with the positions of the charged residues inthe OmpF structure, though it should be empha-sized that the average pathways could not havebeen deduced easily from a simple inspection ofthe X-ray structure. The resulting average ion dis-tribution arises not only from the strong transverseelectric field in the pore, but also from the electro-static interactions with a large number of residues.A similar transverse field was noted by Karshikoffet al.27 on the basis of continuum electrostatic calcu-lations. On the basis of this observation, theysuggested that its role could be to facilitate the per-meation of dipolar solutes. Nonetheless, the bio-logical significance of this feature of OmpF isunclear. Perhaps the left-handed twist in the direc-tion of the transverse field, shown here to extendnearly over 20 A, adds to the stability of the right-handed b-barrel structure.

On the basis of this analysis, one can envisionthe average journey of Kþ and Cl2 from the extra-cellular side to the periplasmic side. Kþ preferablyenter the external vestibule (mouth of pore) nearthe monomeric interface due to Asp37 and Asp74or above the tip of loop L3 due to Glu29, andAsp121 and Glu117 on loop L3. In contrast, Cl2

preferably enter the mouth of pore following theelectrostatic field generated by Arg167 and Arg168(see Figure 11(a)). As they progress towards theintracellular side along the pore, the position ofboth ions is shifted in an anti-clockwise directionfollowing the charged residues (see Figure 11(a)and (b)). The Kþ move down close to Asp113 onloop L3, while the Cl2 move down close to thearginine cluster along the barrel wall close to the3-fold axis. After passing the constriction zone, theCl2 continue down along the barrel wall towardLys16 and Arg270 and away from Glu62, resultingagain in an anti-clockwise rotation of more than

Figure 12. A superimposition of 100 snapshots of the ions in all three pores (the Kþ are magenta and the Cl2 aregreen) Each snapshot was extracted every 50 ps from the 5 ns trajectory and all the ions in monomers M2 and M3were superimposed into monomer M1 by rotations. (left) View from the 3-fold symmetric axis. (middle) left viewrotated by 1208. (right) left view rotated by 2408. The Figure was produced with DINO (http://www.dino3d.org).

1190 MD Simulation of OmpF Porin

e.g. W. Im & B. Roux, J. Mol. Biol. (2002), B. Dhakshnamoorthy et al., J. Am. Chem. Soc. (2013), P. R. Singh et al. J. Phys. Chem. B (2012)

Model DMPC-‐OmpF System

Several outer membrane proteins

OmpF Passive transport of ion/molecule (~ 600Da), Weak selecKvity for caKons

“ConstricKon Zone” R42, R82, R132

D113, E117, D121

The largest and best studied system

BEYOND THE MODEL SYSTEM 第２回成果報告会（6/11）

What is the impact of different LPS environments?

O-‐anKgen

R1 core

Lipid A

PL

K12

lps5-‐ompf k12-‐ompf

lipa-‐ompf lps0-‐ompf

lipa-‐lps0-‐lps5-‐ompf lipa-‐lps5-‐ompf lps0-‐lps5-‐ompf

7 systems (~300K atoms), 0.5μs/each 6 ns / day with 1,024 cores (64 nodes)

TRAJECTORY MOVIE OF LPS5-‐OMPF 第２回成果報告会（7/11）

Lps5 (flexible)

LipidA (rigid)

Inner/Outer cores (very rigid)

Phospholipid (flexible)

Ca2+ (very rigid)

Flexible LPSs cover the channel entrance

Channel entrance region

OmpF

slow ions fast ions

K+ (purple)/Cl–(green)

IMPACT OF LPS ON ANTIOBDY ACTIVITY Ac=vity increases by trimming LPS


A. T. Bentley & P. E. Klebba, J. Bacteriol. (1988）

1

1

2

3 4

5 6

7 9

10

8

11 12

GlcN Kdo Hep

Glc

Gal

GlcNAc

PO42–

CO2–

Hep

1066 BENTLEY AND KLEBBA

LPS CHEMOTYPESE co/i

08:K27wtd Rc RdI

-IEF]0 00

DUE-

L-I IEUI

LII FIIF-

DLI*-DELI

DIILIID -IDI I'DDE]*-D D_

I0 0FO El-EEl O-IFO *0

5. typhimuriAmwt' Re Rb2 Rb3

DD DDD DDFD DF EDD DFDD IDD DDF

DDD I IDD El I ID DF EDDolII

DD I Io ID IIoD Do o

DD LIIooIDDDoII

DoJo FD Do oD Do o

FDDDEIL LIID o L1oD Do o

FIG. 4. Summary of cytofluorometric reactivity of anti-E. coli B/r porin MAbs with E. coli and S. typhimurium LPS mutants by flowcytometry. Symbols: Empty squares, no activity; striped squares, low activity (less than 3 times normal mouse serum); stippled squares, highactivity (more than 3 times normal mouse serum). wt; Wild type. Notes: a, E. coli B/r strain CM6; b, E. coli K-12 strain BN1071; c, E. coliK-12 strain X1715; d, E. coli 08:K27 strain D280; e, S. typhimurium LT2 strain SL3840.

adsorbed 15 to 17 MAbs. Two antibodies showed variablereactivities with these mutants. MAb 13 was marginallypositive on the rfaB, rfaG, and rfaD mutants (althoughfluorescence was above that of the background, it wassignificant only at P = 0.1 to 0.15) but was strongly positivefor the rfaP mutant, which has a structure similar to that ofrfaB but is not phosphorylated on the heptose. MAb 19 wasreactive with the rfaB, rfaP, and rfaD strains, but not withthe rfaG strain. In general, decreased LPS core lengthresulted in an increased number of different MAbs that wereable to bind the intact cell.

Wild-type E. coli 08:K27, which has an intact 0 antigen,did not bind any of the antibodies within the panel (Fig. 4),but its rfa derivatives (which have shorter cores than E. coliB/r) adsorbed 15 MAbs. MAb 20 was unique in the panel inthat it reacted with an epitope that was exposed on thesurface of E. coli 08:K27 but that was not exposed on eitherE. coli B/r or K-12.MAb binding to the S. typhimurium cell surface. None of

the MAbs adsorbed to S. typhimurium strains with wild-type, rfaL (which has a complete core but lacks the 0

antigen), or rfaJ LPS (Fig. 4). Three MAbs ultimatelyrecognized rough mutants of S. typhimurium. MAb 69 beganreactivity at the rfaI (Rb2 chemotype) level, and both MAbs19 and 20 recognized rfaF (Rd2 chemotype) or rfaE (Re

chemotype) strains. Again, as the LPS core became shorter,more porin surface epitopes became available for MAbbinding. Furthermore, for any particular MAb, as the LPScore structure became rougher, the fluorescence intensityincreased (data not shown). This general trend was observedmost clearly in cytofluorimetry of S. typhimurium strains,but was also seen for several of the E. coli strains.

DISCUSSIONAntibodies to denatured E. coli B/r porin presumably

recognize consecutive amino acids in its polypeptide se-quence, while antibodies to the native globular protein (as inouter membrane fragments) may bind conformational deter-minants that contain folded and juxtaposed peptide chainswhich are susceptible to denaturation (39). Antibodies of theformer type had the ability to differentiate among E. coli andS. typhimurium strains. Immunoblots showed high cross-reactivities among the E. coli strains, reflecting the extensiveamino acid homology between B/r and K-12 OmpF porins,which differ by only 1% (15), and suggesting less, but stillhigh, homology for a porin of 08:K27. On the other hand,reactivity of anti-OmpF MAb for denatured OmpC wasminimal (15%), in spite of the 61% sequence homologybetween these two porins (23). These data, which are

mAbNlS

10

12

14

15

6

81 1

16

4

7

so5

9

13

19

6944

49

20

B/rclass wt I

G il

G2A U

GlilG2B 0G2B U

G2B U

G3 KG2A U

G3

G1G3

GlilGlIG3G2A U

ND I

GlIGil-

M II

Wtb wt

D DSiim

-i

LID DD DD DD DD DDF

D DD DD D

Rb2

DF.IIUIUIMM

FIDDDDDDDDDDD

Rb3 Rc

iiI.Em0

1: 0I,Fo NF] 0M F-

ii0D-IFD-I -

DIFDIF

RcP

D

IIII

-

I

IDDDDFoFo

Rd l Re

EU

= _

F -IF-11

F1 F"

U.

Ii

I.U.I.I0E0

Rd1

L]LIDDDDDDDDDDDDDDDDILIDLI

Rd2

D

L-IDDDDDDDDDDDDDIFI

_oIF-F-F]F.

Re

D

L-IL-IL-IL-I

L-I

L-IFILI

00

LI

IFom

J. BACTERIOL.

on March 6, 2014 by RIKAG

AKU KENKYUJO (RIKEN)

http://jb.asm.org/

Downloaded from

Ec K12 core structure

①

③

④

① ② ③

②

④

K12

Cytofluorometric reacKvity

LPS REDUCES ACCESSIBILITY TO EPITOPES

1�

1�

2�

3�4�

5�6�

7�9�

10�

8�

11�12�

GlcN�Kdo�Hep�

Glc�

Gal�

GlcNAc�

PO42–�

CO2–�

Hep�


Surface epitope S3c (L1) Exposed to water solvent

“RealisKc modeling” (i) well explains the experimental anKbody reacKvity, and (ii) has the potenKal to reveal binding sites accompany with protein-‐LPS interacKons.

Surface epitope S1 (L4) Interact with neighboring protein Exposed to water solvent

Surface epitope S2 (L5) Various interacKons with inner-‐, outer-‐core sugars of LPS

Ec K12

SUMMARY AND PERSPECTIVE Summary: •  Atomis=c simula=ons have revealed molecular details of structure and dynamics

of membranes and membrane proteins relevant for biological func=ons, including the ATP-‐bound state Ca2+-‐pump, the proton transport in MATE, the lipid-‐protein interacKons in OM as well as the hydraKon and lipid moKons of membranes.

•  An efficient free-‐energy calcula=on method based on the combined replica-‐exchange and metadynamics frameworks, and a protocol for hierarchical computa=ons were build for an integrated funcKonal analysis of membrane proteins.

Perspec=ve: Further extension of the study, including the integraKon of a path-‐sampling method with mulK-‐scale simulaKon techniques for example, could elaborate on the quan=ta=ve free-‐energy changes through out the func=onal mo=ons of membrane proteins.


ACKNOWLEDGEMENTS Members (RIKEN Wako) Dr. Takaharu Mori

Dr. Yasuaki Komuro (Eisai Co., Ltc.)

Dr. .Pai-‐Chi Li

Dr. Wataru Nishima

Dr. Raimondas Galvelis

Dr. Isseki Yu,

Dr. Po-‐Hung Wang

Dr. Yuji Sugita

Computer resource HPCI computaKonal resources

FX10 (Univ. Tokyo)

3,743,728 node-‐hours

Total usage : 85.2%

Collaborators RIKEN

Dr. Tahei Tahara (RIKEN)

Dr. Toshihide Kobayashi (RIKEN)

Dr. Koichiro Shirota (RIKEN)

Dr. Chigusa Kobayashi (RIKEN AICS)

Dr. Kiyoshi Yagi (RIKEN)

Chuo Univ.

Prof. Eiro Muneyuki (Chuo Univ.)

0.0

25.0

50.0

75.0

100.0

0 2 4 6 8 10 12

Usage (FX10, Oakleaf)

(month)

(usage, %

)

Univ. Tokyo Prof. Ryuichiro Ishitani Prof. Osamu Nureki Univ. Kansas (USA) Prof. Wonpil Im Dr. Dhilon S. Patel


quantitative)functional)analysis)of ...quantitative)functional)analysis)of...

Documents