最小自由エネルギー経路探索法による  多剤排出トランスポーターの薬剤排出機構の解明

1  理研・次世代計算科学  2  横浜市立大・生命ナノ  

3  東京大・農学生命  4日本医科大・物理  

5  理研・AICS  

木寺詔紀1,2    寺田透1,3    藤崎弘士1,4    森次圭1    松永康佑1,5    池口満徳1,2  

hp120027

平成24年度 「京」を中核とするHPCIシステム利用研究課題中間報告会  平成25年3月14日〜15日 イイノカンファレンスセンター

1/11

内容

•  モチベーション  – 多剤排出トランスポーター(AcrB)の紹介  

•  計算手法  – 有限温度ストリング法  

•  計算の進捗状況

2/11

多剤排出トランスポーター(AcrB)

of PN1 and PN2 (subdomains within the porter domain) from thesubstrate binding pocket in the binding protomer is blockedwith thisinclined a-helix (Fig. 1d and Fig. 2). In contrast, the exit from thevacant binding site of the protomer contributing this inclined helix isopen towards the exit funnel. We suggest that this protomer rep-resents the situation just after extrusion of the substrates, and namedit the ‘extrusion’ protomer (Fig. 1 and Fig. 3a, red). The thirdprotomer has a vacant binding site but an upright central helix,representing an intermediate state between extrusion and binding;thus, it is named the “access” protomer, waiting for the next substratebinding event (Fig. 1, green).

Drug binding siteA substrate binding pocket exists in the porter domain of eachprotomer. The porter domain consists of four subdomains—PN1,PN2, PC1 and PC213—each of which consists of two b–a–b sand-wiches. These four subdomains are packed with their b-sheetssurrounding a voluminous pocket (Fig. 1d). Bound minocyclineand doxorubicin are observed in one of the three protomers at the

distal end of the pocket, between theb-sheets of PN2 andPC1 (Fig. 1dand Fig. 2). The bromine atom of the 9-bromo-minocycline wasclearly identified in the anomalous difference Fourier map (6.2j;0.02417 electrons A23) at the expected position for the structure ofminocycline (Fig. 2a). The substrate binding pocket is rich inaromatic amino-acid residues: Phe 136 and Phe 178 (PN2), andPhe 610, Phe 615, Phe 617 and Phe 628 (PC1). These aromatic sidechains may interact with the drug molecule by hydrophobic oraromatic–aromatic interactions (Fig. 1d). There are some polarresidues in this area, such as Asn 274 and Gln 176, possibly forminghydrogen bonds with the drug molecule.Figure 2b, c shows details of the binding site with substrates

bound. The methyl moiety of the 7-dimethylamino group and theC-ring of minocycline interact with Phe 178 and Phe 615, respect-ively. Two oxygen atoms of the 1-oxo and 2-amido groups interactwith Asn 274 (see also Supplementary Fig. S2). Doxorubicin interactswith Phe 615, similar to minocycline, but interacts with Gln 176 andPhe 617 instead of Asn 274 and Phe 178, respectively. Therefore,different sets of residues are used for binding of the different kinds

Figure 1 | Structure of the AcrB–minocycline complex based on theasymmetric crystal. The three AcrB protomers are individually coloured(blue, red and green).a, Ribbonrepresentationviewed fromthe sideparallel tothe membrane plane. The minocycline molecule is shown in CPK (Corey–Pauling–Koltun) representation, with the position of the C, N and O atomsindicated by yellow, blue and red balls, respectively. The extra-membrane(periplasmic) headpiece is at the top and the transmembrane region is at thebottom. b, Top view of the ribbon representation. c, Close-up view of the

central helix in theporter domain.Key residues for the central helix inclinationare shown in the ball-and-stick representation. Hydrogen bonds are indicatedbydotted lines.d, Cut viewof the porter domain from the distal side of the cell.Phenylalanine residues, including those in the substrate binding pocket, areshown in the ball-and-stick representation. The difference Fourier map(F (drug) 2 F (native)) of bound minocycline is depicted by an orange-colouredcage contour at 3j (0.3058 electrons A23). Arrows indicate the relativemovements of the subdomains in comparison with those in the former state.

ARTICLES NATURE|Vol 443|14 September 2006

174

多剤排出トランスポーターの結晶構造 S.  Murakami,  et  al.,  Nature  443,  173  (2006).

side  view top  view

ホモ三量体、非対称構造 3/11

抗生物質耐性をもつ細菌(多剤耐性菌)の細胞膜に存在するタンパク質。細菌に作用するはずの薬剤を能動的に細胞外へ排出してしまう。

膜貫通  領域

ペリプ  ラズム

細胞内

薬剤

薬剤排出メカニズムの仮説

これまでの研究から提出された仮説    1.  3つの異なる状態が回転(機能回

転)することで薬剤が能動輸送される  

2.  膜貫通領域におけるプロトン化/脱プロトン化が機能回転の原動力。それがペリプラズム側のドメインへ伝わって構造変化が起こる。  

薬剤を排出過程の全原子シミュレーションを行いこれらの仮説を検証  機能回転仮説の検証、プロトン化状態によるエナジェティクスの変化、薬剤排出のボトルネックとなりうる相互作用を調べる

Binding Access

Extrusion

Extrusion Binding

Access 機能回転

4/11

有限温度ストリング法による  最小自由エネルギー経路探索

レプリカ間の弱い相互作用

1本のパス上に配置された多数のレプリカを並列に動かすことでサンプリング

経路探索空間 レプリカ間の弱い相互作用

1本のパス上に配置された多数のレプリカを並列に動かすことでサンプリング

経路探索空間

“ふつう”にシミュレーションするのではなく、始状態と終状態を繋ぐ

構造変化経路上に少しずつ構造の異なるコピー系を配置し、それら

をまとめて動かすことで効率良くサンプリングする

高並列マシン(スパコン)があって初めて可能な手法

5/11

有限温度ストリング法アルゴリズムの概要 E  W.,  Ren  W.,  Vanden-­‐Eijnden  E.  (2005)  J.  Phys.  Chem.  B  109:  6688-­‐6693.  Maragliano  L.,  Fischer  A.,  Vanden-­‐Eijnden  E.,  CiccoS  G.  (2006)  J.  Chem.  Phys.  125:  24106.  Maragliano  L.,  Vanden-­‐Eijnden  E.  (2007)  Chem.  Phys.  Le2.  446:  182-­‐190.

Algorithm  1.  経路探索空間(粗視化空間)を定

義する  2.  経路をimagesと呼ばれる点で離

散化する  3.  Images  を 自由エネルギー地形上

でsteepest  descentで動かす  4.  imagesの間の相互作用をequi-­‐

distance  parametrizaWonで近似  5.  3.  4.  を繰り返し、最小自由エネル

ギー経路に収束  

apo-AK� holo-AK�

apo-AK� holo-AK�open� close� open� close�

initial path� initial path�

converged path�

TS?�

substate?�

Conformational Transition Pathways of Adenylate Kinase Explored by the String Method�Yasuhiro Matsunaga1,* , Hiroshi Fujisaki2,3, Tohru Terada3,4, and Akinori Kidera3,5 * ymatsunaga@riken.jp

1 RIKEN AICS, Kobe, Japan, 2 Nippon Medical School, Kawasaki, Japan, 3 RIKEN, Wako, Japan, 4 The University of Tokyo, Tokyo, Japan, 5 Yokohama City University, Yokohama, Japan.

Large-scale conformational changes in proteins involve barrier-crossing transitions on the complex free energy surfaces of high-dimensional space. Here we show that, by combining the on-the-fly string method and the multi-state Bennett acceptance ratio (MBAR) method, the free energy profile of a conformational transition pathway in Escherichia coli adenylate kinase can be accurately determined. The minimum free energy paths (MFEPs) of adenylate kinase were searched by the on-the-fly string method in 20-dimensional space spanned by the 20 largest-amplitude principal modes, and the free energy and various kinds of average physical quantities along the MFEPs were successfully evaluated by the MBAR method. The influence of ligand binding on the MFEPs was characterized in terms of rigid-body motions of the LID and AMPbd domains. It was found that the LID domain was able to partially close without the ligand, while the closure of the AMPbd domain required the ligand binding. The transition state ensemble of the ligand bound form was identified as those structures characterized by highly specific binding of the ligand to the AMPbd domain, and was validated by unrestrained MD simulations. It was also found that complete closure of the LID domain required the dehydration of solvents around the P-loop. These findings suggest that the interplay of the two different types of domain motion is an feature in the conformational transition of the enzyme.�

1.   The combination of the finite temperature string method and MBAR works well in finding reaction coordinates and the evaluation of free energy along a path

2.   Finding and characterizing a MFEP gives us an in-depth mechanism of the conformational transition of protein

3.   The transition state ensemble was identified as the highly specific binding of the AMP moiety to the AMP-binding domain. This has been validated by a set of non-biased MD simulations launched from the region of the transition state.

Ut(f)S(f)U(f)= diagf 1;:::; 3N ( 0)g

Conformational transitions of proteins •  microseconds to seconds •  To characterize such slow events, the free energy profile or the potential of mean

force (PMF) along a reaction coordinate play a central role •  In particular, to evaluate the barrier-height and reproduce the correct kinetics, the

identification of transition state is indispensable

Minimum Free Energy Path (MFEP) •  Defined as the transition path of maximum likelihood in some collective variables •  Good candidate for a reaction coordinate

Finite-temperature String Method •  Designed for finding a MFEP from the high-dimensional free energy landscape

E W., Ren W., Vanden-Eijnden E. (2005) J. Phys. Chem. B 109: 6688-6693. Maragliano L., Fischer A., Vanden-Eijnden E., Ciccotti G. (2006) J. Chem. Phys. 125: 24106. Maragliano L., Vanden-Eijnden E. (2007) Chem. Phys. Lett. 446: 182-190.�

1.  Consider collective variable space 2.  Discretize the initial path by beads (images) 3.  Steepest descent for images 4.  Equi-distance parameterization 5.  Repeat 3. and 4., and images converge to

a MFEP

Test case: alanine-dipeptide

Adenylate kinase from E. coli •  An enzyme which catalyzes the phoshoryl transfer reaction: AMP + ATP !" 2ADP •  Crystal structures suggest that, upon ligand binding, the enzyme undergoes open-

to-close transition to form the catalytically competent closed structure. •  NMR and single-molecule FRET studies have shown that the conformational

transition is the rate-limiting step of the enzyme activity�

Models •  Two Systems: Adenylate kinase without

ligand and with ligand •  Ligand: Ap5A (analog inhibitor) •  Software: mu2lib (in-house MD library) •  Force field: Amber ff03 and GAFF •  Explicit solvent (TIP3P) •  Particle mesh Ewald for electrostatic

interactions

Crystal structures of adenylate kinase�

Questions and purposes: •  What kinds of interactions

determine the transition state of conformational change ?

•  How does the presence of ligand change the conformational transition pathway ?

Without ligand (apo-AK)�

With ligand (holo-AK)�

Simulation Protocol •  # of images: 65 •  As the collective variables, we chose twenty principal components which

are large enough to describe the domain motions of adenylate kinase. •  12 ns (on-the-fly) string method calculation for converged string •  600 ps or 9 ns umbrella sampling to obtain statistical averages (e.g., PMF)

!"#"$%&'(#)*+','$-'+.#&/)-&0)1#&,"(0#%2)#3&%24#5)#6"(7+6%)7#+.$-),,&#1&.3,'(0#1'.+,&8"(19#�

!'.)#1%)3#":#+.$-),,&#1&.3,'(0�

!'.)#1%)3#":##1%-'(0#.)%2"7�

!"#$%&'()"*+,�

-(.#)//0'10(2/$%&�

;#(1�

<#(1�

=#(1�

><#(1#?6"(/)-0)7@�

A;;#31#-)1%-&'()7#BC#:"-#)&62#'.&0)#?'.&0)1#&-)#DE)7#7+-'(0#%2)#1&.3,'(0@�

A;;#31#-)1%-&'()7#BC#:"-#)&62#'.&0)�

A;;#31#-)1%-&'()7#BC#:"-#)&62#'.&0)�

F#(1#-)1%-&'()7#BC#:"-#)&62#'.&0)�

PMF estimation along the curved pathways in 20-dimensional space •  We applied the multi-state Bennett

acceptance ratio (MBAR) method in combination with the kernel density estimator method.

•  While the WHAM requires an large storage space for grid points, the MBAR requires no grid point, and hence applicable to large dimensional data sets.

PMF along the curved pathways in 20-dimensional space

Pathways projected onto the inter-domain distances apo-AK: LID-first closure is preferred holo-AK: AMPbd-first closure is preferred

LID domain closure follows the dehydration around the P-loop

Highly specific binding of AMP determines the transition state

Takashi Imai (RIKEN), Yuji Sugita (RIKEN), Motoyuki Shiga (JAEA), Michael Feig (Michigan State Univ.), Ryuhei Harada (RIKEN), Jaewoon Jung (RIKEN), Sotaro Fuchigami (Yokohama City Univ.), and Mitsunori Ikeguchi (Yokohama City Univ.)

Shirts M.R., Chodera J.D. (2008) J. Chem. Phys. 129: 124105.�

Algorithm

Ap5A�

Validation of TS by a set of unbiased MD simulations�

initial distribution�

distribution after 10ns�

open� close�

6/11

最小自由エネルギー経路の特徴

2つの性質を持つ    •  (overdampedを仮定すれば)多くの実際のトラジェ

クトリが最小自由エネルギー経路付近を通る  →構造変化空間を縮約できる    

•  isocommiZor  surfaceに直交する  →遷移状態を同定できる  

Maragliano,  L.,  Fischer,  A.,  Vanden-­‐Eijnden,  E.,  &  CiccoS,  G.  (2006).  J  Chem  Phys,  125(2),  24106.   最小自由エネル

ギー経路

rareなパス

最小自由エネルギー経路は、経路探索空間(粗視化空間)において

尤も確率の高い構造変化パスになっている

最小自由エネルギー経路は、構造変化を記述するための“良い”反応座標に成りうる

ほとんどのパス

7/11

応用例:  アデニル酸キナーゼの構造変化

Y.  Matsunaga,  et  al.,  PLoS  Comput.  Biol.  8,  e1002555  (2012).

and calculated the PMF and the averages of various physicalquantities using the MBAR method. Our analysis elucidates an in-depth mechanism of the conformational transition of AK.

Results

Minimum Free Energy Path and the CorrespondingDomain Motion

The MFEPs for apo and holo-AKs, and their PMFs, wereobtained from the string method and the MBAR method,respectively (see Videos S1 and S2). The MFEPs were calculatedusing the same 20 principal components selected for the collectivevariables. The holo-AK calculations were undertaken with thebisubstrate analog inhibitor (Ap5A) as the bound ligand withoutimposing any restraint on the ligand. Figures 2A and 2B show theMBAR estimates of the PMFs along the images of the MFEP (theconverged string at 12 ns in Figures 2A and 2B) for apo and holo-AK, respectively. Here, the images on the string are numberedfrom the open (a~1; PDBid: 4ake [28]) to the closed conforma-tion (a~65; PDBid: 1ake [29]). These terminal images were fixedduring the simulations to enable sampling of the conformationsaround the crystal structures. In the figures, the convergence of thePMF in the string method process is clearly seen in both systems.Convergence was also confirmed by the error estimates (FigureS1), and by the root-mean-square displacement (RMSD) of thestring from its initial path (Figure S2). The PMF along the MFEPreveals a broad potential well on the open-side conformations ofapo-AK, suggesting that the open form of AK is highly flexible[20]. This broad well is divided into two regions, the fully open(a*4) and partially closed states (a~20,:::,35, encircled) by asmall PMF barrier. In holo-AK (Figure 2B), the MFEP exhibits asingle substantial free energy barrier (a*33) between the openand closed states, which does not appear in the initial path. Thisbarrier will be identified as the transition state below. It is shown inthe PMF along the MFEP that the closed form (tightly binding theligand) is much more stable than the open form with loose bindingto the ligand (large fluctuations of the ligand will be shown later).

To characterize the MFEP in terms of the domain motions, theMFEP was projected onto a space defined by two distances fromthe CORE domain, the distance to the LID domain and thedistance to the AMPbd domain (the distance between the masscenters of Ca atoms for the two domains; Figures 2C and 2D). ThePMF was also projected onto this space. The comparison of thetwo figures shows that ligand binding changes the energylandscape of AK, suggestive that this is not a simple population-shift mechanism. In apo-AK, the motions of the LID and AMPbddomains are weakly correlated, reflecting the zipper-like interac-tions on the LID-AMPbd interface [19]. The MFEP clearlyindicates that the fully closed conformation (a~65) involves theclosure of the LID domain followed by the closure of the AMPbddomain. The higher flexibility of the LID domain has beenreported in previous studies [17,19,20]. In holo-AK, the pathwaycan be described by two successive scenarios, that is, the LID-first-closing followed by the AMPbd-first-closing. In the open state(a~1,:::,32), the MFEP is similar to that of apo-AK, revealing thatLID closure occurs first. In the closed state (a~34,:::,65), however,the AMPbd closure precedes the LID closure. This series of thedomain movements was also identified by the domain motionanalysis program DynDom [24] (Figure S3).

It is known in the string method that the convergence of thepathway is dependent on the initial path. In order to checkwhether the MFEP obtained here is dependent on the initial pathor not, we conducted another set of the calculations for apo-AK byusing a different initial path, which has an AMPbd-first-closing

pathway, opposed to the LID-first-closing pathway shown above.If the LID and AMPbd domains move independently of eachother, it is expected that LID-first-closing and AMPbd-first-closingpathways are equally stable. Despite this initial setup, however, ourcalculation again showed the convergence toward the LID-first-closing pathway (see Figure S4). As described above, this tendencyof the pathways would be due to the reflection of the highly flexiblenature of the LID domain.

Furthermore, in order to check whether the samples around theMFEP are consistent with the experiments, we compared the PMFas a function of the distance between the Ca atoms of Lys145 andIle52 with the results of the single-molecule FRET experiment byKern et al. [16] (see Figure S5). The PMF was calculated by usingthe samples obtained by the umbrella samplings around theMFEP. In the figure, the stable regions of the PMF for holo-AKare highly skewed toward the closed form, and some populationtoward the partially closed form was also observed even for apo-AK, which is consistent with the histogram of the FRET efficiency[16].

‘‘Cracking’’ in the Partially Closed State in apo-AKTo more clearly illustrate the energetics along the MFEP in

terms of the domain motions, we separately plot the PMF as afunction of the two inter-domain distances defined above(Figures 3A and 3B). We observe that the PMF of apo-AK has adouble-well profile for the LID-CORE distance (indicated by theblue line in Figure 3A), whereas the PMF in terms of the AMPbd-CORE distance is characterized by a single-well (Figure 3B). Thesingle-molecule FRET experiments monitoring the distancesbetween specific residue pairs involving the LID domain (LID-

Figure 2. PMF along the strings and the corresponding domainmotions. (Top) PMFs along the snapshots of the strings for (A) apo and(B) holo-AK at t = 0 (initial path), 2, 4, 6, 8, 10, 12 (MFEP) ns. The indicesof string images are numbered from the open (PDBid: 4ake) to closedcrystal structures (PDBid: 1ake). Partially-closed state of apo-AK and asubstantial PMF barrier of holo-AK are encircled. (Bottom) Projections ofMFEP onto the space defined by the distances of Ca mass centersbetween the LID-CORE and AMPbd-CORE domains, for (C) apo and (D)holo-AK. The black curves indicate the initial paths (t = 0), and the redcurves are the MFEPs (t = 12 ns). The PMF is visualized by the coloredcontour lines (delineating regions of low energy (blue) to regions ofhigh energy (red)).doi:10.1371/journal.pcbi.1002555.g002

Minimum Free Energy Path of Adenylate Kinase

PLoS Computational Biology | www.ploscompbiol.org 3 June 2012 | Volume 8 | Issue 6 | e1002555

and calculated the PMF and the averages of various physicalquantities using the MBAR method. Our analysis elucidates an in-depth mechanism of the conformational transition of AK.

Results

Minimum Free Energy Path and the CorrespondingDomain Motion

The MFEPs for apo and holo-AKs, and their PMFs, wereobtained from the string method and the MBAR method,respectively (see Videos S1 and S2). The MFEPs were calculatedusing the same 20 principal components selected for the collectivevariables. The holo-AK calculations were undertaken with thebisubstrate analog inhibitor (Ap5A) as the bound ligand withoutimposing any restraint on the ligand. Figures 2A and 2B show theMBAR estimates of the PMFs along the images of the MFEP (theconverged string at 12 ns in Figures 2A and 2B) for apo and holo-AK, respectively. Here, the images on the string are numberedfrom the open (a~1; PDBid: 4ake [28]) to the closed conforma-tion (a~65; PDBid: 1ake [29]). These terminal images were fixedduring the simulations to enable sampling of the conformationsaround the crystal structures. In the figures, the convergence of thePMF in the string method process is clearly seen in both systems.Convergence was also confirmed by the error estimates (FigureS1), and by the root-mean-square displacement (RMSD) of thestring from its initial path (Figure S2). The PMF along the MFEPreveals a broad potential well on the open-side conformations ofapo-AK, suggesting that the open form of AK is highly flexible[20]. This broad well is divided into two regions, the fully open(a*4) and partially closed states (a~20,:::,35, encircled) by asmall PMF barrier. In holo-AK (Figure 2B), the MFEP exhibits asingle substantial free energy barrier (a*33) between the openand closed states, which does not appear in the initial path. Thisbarrier will be identified as the transition state below. It is shown inthe PMF along the MFEP that the closed form (tightly binding theligand) is much more stable than the open form with loose bindingto the ligand (large fluctuations of the ligand will be shown later).

To characterize the MFEP in terms of the domain motions, theMFEP was projected onto a space defined by two distances fromthe CORE domain, the distance to the LID domain and thedistance to the AMPbd domain (the distance between the masscenters of Ca atoms for the two domains; Figures 2C and 2D). ThePMF was also projected onto this space. The comparison of thetwo figures shows that ligand binding changes the energylandscape of AK, suggestive that this is not a simple population-shift mechanism. In apo-AK, the motions of the LID and AMPbddomains are weakly correlated, reflecting the zipper-like interac-tions on the LID-AMPbd interface [19]. The MFEP clearlyindicates that the fully closed conformation (a~65) involves theclosure of the LID domain followed by the closure of the AMPbddomain. The higher flexibility of the LID domain has beenreported in previous studies [17,19,20]. In holo-AK, the pathwaycan be described by two successive scenarios, that is, the LID-first-closing followed by the AMPbd-first-closing. In the open state(a~1,:::,32), the MFEP is similar to that of apo-AK, revealing thatLID closure occurs first. In the closed state (a~34,:::,65), however,the AMPbd closure precedes the LID closure. This series of thedomain movements was also identified by the domain motionanalysis program DynDom [24] (Figure S3).

It is known in the string method that the convergence of thepathway is dependent on the initial path. In order to checkwhether the MFEP obtained here is dependent on the initial pathor not, we conducted another set of the calculations for apo-AK byusing a different initial path, which has an AMPbd-first-closing

pathway, opposed to the LID-first-closing pathway shown above.If the LID and AMPbd domains move independently of eachother, it is expected that LID-first-closing and AMPbd-first-closingpathways are equally stable. Despite this initial setup, however, ourcalculation again showed the convergence toward the LID-first-closing pathway (see Figure S4). As described above, this tendencyof the pathways would be due to the reflection of the highly flexiblenature of the LID domain.

Furthermore, in order to check whether the samples around theMFEP are consistent with the experiments, we compared the PMFas a function of the distance between the Ca atoms of Lys145 andIle52 with the results of the single-molecule FRET experiment byKern et al. [16] (see Figure S5). The PMF was calculated by usingthe samples obtained by the umbrella samplings around theMFEP. In the figure, the stable regions of the PMF for holo-AKare highly skewed toward the closed form, and some populationtoward the partially closed form was also observed even for apo-AK, which is consistent with the histogram of the FRET efficiency[16].

‘‘Cracking’’ in the Partially Closed State in apo-AKTo more clearly illustrate the energetics along the MFEP in

terms of the domain motions, we separately plot the PMF as afunction of the two inter-domain distances defined above(Figures 3A and 3B). We observe that the PMF of apo-AK has adouble-well profile for the LID-CORE distance (indicated by theblue line in Figure 3A), whereas the PMF in terms of the AMPbd-CORE distance is characterized by a single-well (Figure 3B). Thesingle-molecule FRET experiments monitoring the distancesbetween specific residue pairs involving the LID domain (LID-

Figure 2. PMF along the strings and the corresponding domainmotions. (Top) PMFs along the snapshots of the strings for (A) apo and(B) holo-AK at t = 0 (initial path), 2, 4, 6, 8, 10, 12 (MFEP) ns. The indicesof string images are numbered from the open (PDBid: 4ake) to closedcrystal structures (PDBid: 1ake). Partially-closed state of apo-AK and asubstantial PMF barrier of holo-AK are encircled. (Bottom) Projections ofMFEP onto the space defined by the distances of Ca mass centersbetween the LID-CORE and AMPbd-CORE domains, for (C) apo and (D)holo-AK. The black curves indicate the initial paths (t = 0), and the redcurves are the MFEPs (t = 12 ns). The PMF is visualized by the coloredcontour lines (delineating regions of low energy (blue) to regions ofhigh energy (red)).doi:10.1371/journal.pcbi.1002555.g002

Minimum Free Energy Path of Adenylate Kinase

PLoS Computational Biology | www.ploscompbiol.org 3 June 2012 | Volume 8 | Issue 6 | e1002555

open closed

open closed

基質なし

基質あり

構造変化パス上のエネルギー変化

8/11

最小自由  エネルギー経路

最小自由  エネルギー経路

「京」を使用した多剤排出トランスポーターの進捗状況

計算プロトコルと進捗状況  (済)  各100  nsの平衡サンプリング  (MARBLEを使用)  (済)  平衡サンプリングデータから集団運動モードを経路探索空間として抽出  (済)    10  nsのTargeted  MDにより機能回転と薬剤排出の初期経路を作成(MARBLEを使用)  (済)  初期パス上の初期パス(100個のレプリカ)を1.4  ns  かけて平衡化(mu2libを使用)  (計算中)  ストリング法による経路探索と最小自由エネルギー経路上でのUmbrella  sampling  (mu2libを使用)  

目的：AcrBにおける機能回転モデルの妥当性を検証し、薬剤排出機構を原子レベルの相互作用に基づいて解明する  •  膜貫通領域にいけるプロトン化/脱プロトン化が

どのように機能回転につながるか？  •  薬剤排出のボトルネックとなりうる相互作用は

何か？  

集団座標上の平衡サンプリングデータと初期パス

Targeted  MDの軌道

結晶構造(BEA)

結晶構造(EAB)

有限温度ストリング法により機能回転・薬剤排出の最小自由エネルギー経路を探索

Binding Access

Extrusion

Extrusion Binding

Access 機能回転経路

2つの異なるプロトン化した系でエナジェティクスなどを比較

9/11

途中結果のスナップショットムービー

10/11

まとめ

•  グランドチャレンジで開発してきた  MARBLE  と  mu2libの互いの長所を生かして京でAcrBの機能回転シミュレーションを実行している  

•  機能回転前後の始・終状態まわりでの各100  ns平衡サンプリングから経路探索空間を抽出  

•  始・終状態間での10  ns  Targeted  MDにより初期経路を作成  

•  現在有限温度ストリング法により経路探索中

11/11