Role of the D-Loops in Allosteric Control of ATP Hydrolysis in an ABC

Feb 27, 2012 - Peter M. Jones. † and Anthony M. George*. School of Medical and Molecular Biosciences, and iThree Institute, University of Technology...
0 downloads 0 Views 3MB Size
Download PDF
Article pubs.acs.org/JPCA

Role of the D-Loops in Allosteric Control of ATP Hydrolysis in an ABC Transporter Peter M. Jones† and Anthony M. George* School of Medical and Molecular Biosciences, and iThree Institute, University of Technology Sydney, P.O. Box 123, Broadway, NSW 2007, Australia S Supporting Information *

ABSTRACT: ABC transporters couple ATP hydrolysis to movement of substrates across cell membranes. They comprise two transmembrane domains and two cytosolic nucleotide-binding domains forming two active sites that hydrolyze ATP cooperatively. The mechanism of ATP hydrolysis is controversial and the structural dynamic basis of its allosteric control unknown. Here we report molecular dynamics simulations of the ATP/apo and ATP/ADP states of the bacterial ABC exporter Sav1866, in which the cytoplasmic region of the protein was simulated in explicit water for 150 ns. In the simulation of the ATP/apo state, we observed, for the first time, conformers of the active site with the canonical geometry for an in-line nucleophilic attack on the ATP γ-phosphate. The conserved glutamate immediately downstream of the Walker B motif is the catalytic base, forming a dyad with the Hloop histidine, whereas the Q-loop glutamine has an organizing role. Each D-loop provides a coordinating residue of the attacking water, and comparison with the simulation of the ATP/ADP state suggests that via their flexibility, the D-loops modulate formation of the hydrolysis-competent state. A global switch involving a coupling helix delineates the signal transmission route by which allosteric control of ATP hydrolysis in ABC transporters is mediated.

■

The NBDs form a head-to-tail “sandwich” dimer in which each C-motif is opposed to the Walker A and B of the opposite monomer, forming two composite ATP-binding sites.13,14 The D-loops (comprising the consensus sequence, EATSALD) run antiparallel between the active sites at the dimer interface; the N-terminal glutamate engages one active site whereas the Cterminal aspartate engages the P-loop of the opposite active site. The D-loops are proposed to mediate communication between the active sites.9,13,16,18 The H-loops contain a conserved active site histidine suggested to organize the geometry of the prehydrolytic state.19 The Q-loops contain a conserved glutamine, observed in crystal structures to coordinate the catalytic metal and the ATP-phosphate and proposed to coordinate the nucleophilic water of the hydrolysis reaction.14 The downstream Q-loop is thought to couple rotation of the HSD,10 and/or changes in the NBD:TMD interface,15 to nucleotide binding and hydrolysis. The TMDs form the transmembrane pore and substrate binding sites. Three distinct folds have been observed in structures of ABC transporter TMDs.20 A common feature is a TMD intracellular loop (ICL), which includes an α-helix situated in a groove between the core and helical subdomains of the NBD. This ICL α-helix, the coupling helix (CH),21 is proposed to couple ATP binding and hydrolysis to substrate

INTRODUCTION ABC transporters couple the hydrolysis of ATP to movement of substrates across cellular membranes. They form a large and ubiquitous protein superfamily that transports a vast array of compounds and includes, atypically, gated channels and receptors.1 ABC transporters are significant biomedically due to their involvement in multidrug resistance (MDR) exhibited by many cancers, pathogenic microbes, parasites and plants, and in a range of human genetic diseases, including cystic fibrosis.2−6 The canonical architecture of an ABC transporter is a dimer of dimers: two transmembrane domains (TMDs) and two cytosolic nucleotide-binding domains (NBDs). The NBDs contain a number of conserved sequence motifs: the Walker A (P-loop) and B consensus motifs for nucleotide binding,7 the LSGGQ linker peptide, signature sequence or C-motif,8 and the D-, Q-, and H-loops.9−11 The NBDs are structurally conserved, comprising a RecA-like subdomain that contains the Walker A and B motifs, an ABC subdomain unique to the ABC ATPase,12 and a helical subdomain (HSD) that contains the LSGGQ, also involved in ATP binding, but across the dimer interface from the opposite monomer.13,14 The RecA-like and ABC subdomains are structurally integrated into a core subdomain (CSD), to which the HSD is flexibly attached10,15−17 by the Q-loop at its N-terminus and by the Pro-loop at its C-terminus, which joins it to the Walker B motif (Figure 1). © 2012 American Chemical Society

Received: November 18, 2011 Revised: February 19, 2012 Published: February 27, 2012 3004

dx.doi.org/10.1021/jp211139s | J. Phys. Chem. A 2012, 116, 3004−3013

The Journal of Physical Chemistry A

Article

Figure 1. ABC NBD Dimer. The Sav1866 NBD dimer, monomer A bottom, monomer B top. Stereopair of cartoon with CSD gray, HSD green, Dloop red, D-helix orange, H-loop blue, P-loop purple, Q-loop yellow, C-terminal helix cyan. ATP and residues Q422, E503, and H534 are in stick form, with carbon yellow, nitrogen blue, and oxygen red.

routes and mechanisms by which ATP hydrolysis is allosterically coupled to events in the TMDs, we performed a 150 ns MD simulation of the ATP/apo state of the truncated Sav1866 complex described above. In this simulation we observed conformers of the active site consistent with the hydrolysis scenario proposed,14 and with the suggested roles for the conserved histidine and glutamine outlined above. This occurred following a concerted global switching of the NBD dimer, which resulted in an altered deployment of the conserved active site histidine and glutamate within the ATPbound active site. The global switching appeared to be an allosteric response to the removal of the nucleotide from one site and involved diverse regions of the protein including the Q-, H-, and D-loops. We find that the D-loops have twin roles in both coordinating and activating the nucleophilic water. We compared the ATP/apo state with that observed in a 150 ns simulation of the ATP/ADP state. This indicated that the Dloops are conformationally flexible, thereby suggesting a structural dynamic basis for allosteric control of ATP hydrolysis.

transport. Crystal structures indicate that the coupling helix (CH) that is common to both ABC importers and exporters is the second of two CHs in ABC exporters. In contrast to importers, this second CH interacts with the NBD from the opposite half of the transporter. The mechanism of ATP hydrolysis in ABC transporters has not been solved.4,22 The conserved glutamate immediately downstream of the Walker B motif has been proposed as a catalytic base, abstracting a proton from a water molecule oriented for an in-line nucleophilic attack on the ATPphosphorus atom.23,24 In crystal structures of isolated ABC ATPase dimers, a putative nucleophilic water is hydrogen bonded to the backbone carbonyl oxygen of the conserved alanine residue of the D-loop (EATSALD) of the opposite NBD.9,14,18,19,25 Smith et al.14 proposed a scenario for the hydrolysis mechanism in which the attacking water forms Hbonds to the conserved glutamate and the aforementioned Dloop oxygen atom. This orientation directs the nucleophile lone pair electrons toward the ATP-phosphorus atom, thus constituting the canonical reactant geometry facilitating formation of the transition state. Here we report atomistic molecular dynamics (MD) simulations of a truncated structure of the bacterial ABC exporter Sav1866 in explicit water. The simulated complex comprised the predicted cytoplasmic region of the transporter, which includes the NBDs and the intracellular regions of the TMDs, the ICLs, which are expected to mediate coupling and coordination between the TMDs and NBDs. Previously, using this same system in MD analysis, we found that the ATP/apo complex evolved within 100−120 ns to a conformation consistent with an experimentally observed occluded state,26 in which one active site binds nucleotide tightly while the other can exchange nucleotide freely. Experimentally, the occluded site has been observed to bind ATP, nonhydrolyzable ATP analogues, or ligands expected to mimic the transition state of the ATP hydrolysis reaction, such as ADP and vanadate. This suggests that the tightly bound site is, at some point, competent for ATP hydrolysis. To investigate this possibility further, and attempt to shed new light on the ATP hydrolysis mechanism and the signal

■

EXPERIMENTAL METHODS System Setup. The starting coordinates for MD simulations were taken from the X-ray structure of AMP-PNPbound ABC exporter Sav186627 (PDB 2ONJ). A truncated structure comprising the predicted cytosolic regions was prepared using coordinates for residues 82−138 (ICL1), 183−246 (ICL2), and 300−578 (ICL3-NBD) from each monomer. The two Sav1866 protomers are referred to as A and B, according to the chain identifiers in the Protein Data Bank entry 2ONJ. This structure includes a short “buffer zone” comprising the 4 residues at the truncated ends of the ICLs; these extend beyond the predicted cytosolic region of the protein, and allow for “unravelling” of the termini, to avoid the truncation affecting the structure of the cytosolic zone. The Nterminal residues of the ICLs were acetylated and the Cterminal residues of ICLs 1 and 2 N-methylamidated. To place ATP in the consensus orientation in the active site, the P-loop and ATP molecule of ATP-bound MJ0796 (PDB 1L2T; residues 32−49)14 was root-mean-square (rms) fitted to 3005

dx.doi.org/10.1021/jp211139s | J. Phys. Chem. A 2012, 116, 3004−3013

The Journal of Physical Chemistry A

Article

Analysis. Principal component analysis of the simulation Cα atom trajectory from the ATP/apo simulation was performed using the GROMACS package,30 with frames taken at 100 ps intervals over the entire 150 ns trajectory and aligned to the starting coordinates. The program Hingefinder was used to analyze domain rotations using Cα coordinates, the slow partitioning algorithm and a tolerance of 80%.34 VMD,35 Xplor-NIH,36 and Simulaid37 were used to prepare the system and analyze MD trajectories. All structural figures were prepared using PyMol (http://www.pymol.org/ pymol).

Sav1866 using the Cα coordinates of P-loop residues (368−385 in Sav1866). The resultant coordinates of the pyrophosphate moiety were used in the starting structure. Two systems were prepared: ATP/apo and ADP/ATP. Each complex was solvated in a truncated octahedral periodic cell with a minimum of 20 Å between periodic images of the protein and neutralized with a 0.2 M NaCl solution. All histidine residues were neutral and protonated at the ε nitrogen, with the exception of active site histidine 534, which was ionized; all other ionizable residues were in the default ionization state. Simulation Parameters. MD simulations were performed using NAMD version 2.7b328 with the CHARMM27 force field,29 including φ/ψ cross-term map corrections,30 and the TIP3P model for water.31 The SHAKE and SETTLE algorithms were used to constrain the bonds containing hydrogens to equilibrium length.32 A cutoff of 11 Å (switching function starting at 9.5 Å) for van der Waals and real space electrostatic interactions was used. The particle-mesh Ewald method33 was used to compute long-range electrostatic forces with a grid density of approximately 1/Å3. An integration time step of 2 fs was used with a multiple timestepping algorithm; interactions involving covalent bonds and short-range nonbonded interactions were computed every time step, whereas long-range electrostatic forces were computed every two time steps. Langevin dynamics was utilized to maintain a constant temperature of 310 K with a friction coefficient of 5 ps−1 on all non-hydrogen atoms. A Langevin piston was used to control pressure with a target of 1 atm, a decay period of 100 fs and a damping time scale of 50 fs. Equilibration. To ensure the canonical coordination sphere of the catalytic divalent cation in ATP- and ADP-bound active sites, during the equilibration, harmonic forces with a force constant of 10 kcal/(mol Å2) were applied to the coordinating atoms of the Mg2+ ion, these being two water molecules, a βphosphate oxygen, a γ-phosphate oxygen in the case of ATP or a water molecule oxygen in the case of ADP, the amido side chain oxygen of conserved glutamine 422, and the side chain hydroxyl oxygen of Walker A serine 381, constraining them to an equilibrium distance of 2.2 Å from the Mg2+. The solvated starting structure was minimized using conjugate gradient minimization to a 0.5 kcal/(mol Å) rms gradient with all protein and Mg2+ATP/ADP heavy atoms fixed. Water molecules, NaCl ions, and hydrogens were then further minimized during a 50 ps MD run at 310 K, in which all protein and Mg2+ATP/ADP heavy atoms were again fixed. This starting model was then minimized with harmonic positional constraints on the NCαCO backbone. A 100 kcal/(mol Å2) force constant was used to minimize the system to a 0.5 kcal/ (mol Å) rms gradient. The constraints were gradually removed by subsequent minimizations to a 0.1 kcal/(mol Å) rms gradient, scaling the initial force constants by factors of 0.5, 0.15, 0.05, and 0. The minimized structure was then heated from 50 to 310 K in steps of 25 K using velocity reassignment during a 30 ps MD run. This system was then simulated for 500 ps using Langevin temperature and pressure controls with the harmonic restraints on the coordinating atoms of the catalytic Mg2+ remaining. Production Runs. The equilibrated systems were used for the production runs without restraints. Two systems using the truncated structure were simulated each for 150 ns: one of the ATP/apo state and the other of the ADP/ATP state. All simulations remained stable to completion. For analysis, the coordinates were saved every 1 ps.

■

RESULTS The MD simulations used a system comprising the cytosolic regions in the crystal structure of the AMPPNP-bound Sav1866 ABC transporter from S. aureus,27 as described previously.26 Sav1866 is a homodimer of two half-transporters whose primary sequence and biochemical characteristics identify it as an MDR export ABC transporter. In ABC transporters, the ATP hydrolysis cycle is allosterically coupled to the cycle of substrate translocation in the TMDs; however, the exact correlation between the two sets of events is unknown. Global changes in the protein that represent transitions between stages of the transport cycle occur on the microsecond−millisecond time scale and are likely to be beyond the reach of current MD simulations of the full membrane-embedded transporter. In the truncated complex used in the present simulations, conformational changes in the ICLs and NBDs can occur uncoupled from substrate-dependent or other contingent conformational changes in the TMDs. Thus, simulations of the truncated complex, in enabling relatively long simulation times, and in freeing the ICLs and NBDs from restraints imposed by the TMDs, can provide insights into the molecular mechanism that may not be available in current simulations of the whole transporter. Initially, a 150 ns unrestrained simulation of the ATP/apo system was performed with ATP bound to the core subdomain of protomer A. This used the identical starting structure to that described in our previous report 26 but was equilibrated and run using different random seeds for the initial atom velocity assignments. Figure 2 shows the rms deviation of all Cα atoms from the starting structure. This shows that the complex is stable, with overall deviation from starting structure significantly below that reported in simulations of the membraneembedded whole Sav1866 transporter, including measures from

Figure 2. rmsd of Cα atoms in the ATP/apo simulation. rms deviation of Cα atoms from the starting structure in the ATP/apo simulation. All Cα atoms are black; Cα atoms of ICLs 1 and 2, excluding the 4 residues at each truncated terminus (residues 86−134 ICL1, 187−242 ICL2) are red. 3006

dx.doi.org/10.1021/jp211139s | J. Phys. Chem. A 2012, 116, 3004−3013

The Journal of Physical Chemistry A

Article

these studies for the NBDs alone.38−40 Figure 2 also shows the rms deviation of Cα atoms of ICLs 1 and 2 from the starting structure, not including the 4 buffer residues at the truncated termini. This shows that the regions of the ICLs that interact with the NBDs remain very close to the starting structure throughout the 150 ns simulation, indicating that the truncations do not compromise their stability. Principal component analysis (PCA) of the Cα atom trajectory was used to identify and characterize the global concerted motions in the simulation. PCA defines a set of eigenvectors (PCs) derived from the matrix of pairwise correlated motion of atoms. PCA can distinguish concerted global motions from random thermal fluctuations, and PCs are ranked according to the amplitude of the protein motions they describe. In general, the first 1−4 PCs account for the most concerted 50% or more of protein fluctuations. PC1 represented 27.1% of all Cα atom fluctuations, whereas PC2 and PC3 represented 13.5% and 11.6%, respectively. Figure 3

that the NBD subdomains that comprise the composite ATPbound active site, CSD(A) and HSD(B), and CH1 and CH2 from monomer A that engage principally with these subdomains, remain relatively static. Conversely, the HSD of NBD(A) and the CSD of NBD(B) and CH2 of monomer B underwent correlated motions (Figure 4). The spatial distribution of the changes that occurred in this transition is shown by the Hingefinder analysis, which identifies rotating domains between the maximum and minimum projection structures of PC1 (Figure 5A). The HSD of NBD(A) rotated in concert with residues within (and at) the N- and C-termini of CH2(B). The radius of this rotation was long relative to the distance moved and can thus be described approximately as a translation; this occurred along a vector parallel to the membrane, away from the CSD of the opposite monomer (Figure 5A,B). In addition, most of the CSD of NBD(B), with the exception of the D-loop and downstream αhelix, rotated outward from the central axis of the complex (Figure 5A,C). In our previous simulations of this system,26 we found dynamic opening and closing of the equivalent core subdomain on the 100 ns time scale and showed that this is consistent with experimental data. In view of this, we would expect that the near reversal of the rotation observed at the end of the present simulation (Figure 5C) is likely to be transitory, and part of a cyclical process. Notably, the rotation remains above a baseline of between 3 and 6 degrees relative to the starting structure. Anticorrelated movements between these two rotating regions (magenta and cyan in Figure 5A) resulted in partial opening of the empty active site. Conformational Changes in the D-Loops. In the Hingefinder analysis of rotating domains in PC1, motions of the D-loops are detected as the third and fourth largest amplitude motions, respectively, after the motions described above. In NBD(B), the rotating domain includes most of the NBD(B) D-loop (EATSALD) and the downstream α-helix (Dhelix; residues 510−524) and residue M375 of the NBD(A) Ploop (Figure S1A, Supporting Information). The D-loop aspartate side chain forms a H-bond to the backbone of Walker A residue 3 in the opposite monomer (S376 in Sav1866), and this is a conserved feature of the NBD:NBD interface. In PC1, the NBD(B) D-helix moved along its long axis toward the opposite (ATP-bound) P-loop (Figure S1A,B, Supporting Information), with the interaction between S395 and D509 (Figure S1C, Supporting Information) coupling the motion of M375 with those of the opposite D-loop and D-helix. In the ATP-bound NBD(A), the pattern of changes in the Dloop and D-helix differ markedly from that in NBD(B) (Figure 6). Four domains or residues undergoing distinct motions were

Figure 3. Global switching in the ATP/apo complex. Projection of PCA mode 1 onto the Cα atom trajectory.

shows the time series of the projection of the first PCA mode (PC1); the projection illustrates the extent to which the complex has moved along the transition described by the mode. Figure 3 thus depicts an equilibration of the protein from one conformation to another, with a relatively abrupt transition occurring at around t = 75 ns. PC1 was chosen for further analysis because it delineated and well characterized the most significant global change undergone by the protein during the simulation. Opening of the Apo Active Site. Figure 4 shows the per residue fluctuations due to PC1; high values identify regions that undergo correlated (same direction) or anticorrelated (opposite direction) motions in the global mode. This shows

Figure 4. Domains involved in the global switch. Per residue fluctuations due to PCA1; protomer A is in black, and protomer B, red. Regions and subdomains discussed in the text are delineated with overbars. Loops are labeled by the corresponding letter. “CT” refers to the C-terminal helix, and “α4”, to the second α-helix of the HSD. 3007

dx.doi.org/10.1021/jp211139s | J. Phys. Chem. A 2012, 116, 3004−3013

The Journal of Physical Chemistry A

Article

Figure 5. continued which comprises residues 82−138 in ICLs 1A and 1B, residues 183− 246 in ICL 2A, residues 300−422, 495−506, 525−557 in NBD(A), and residues 300−329, 408−498 in NBD(B), as defined by the Hingefinder analysis of PC1. Hingefinder was used to compare trajectory frames with the starting structure and calculate the angle change between these two subdomains.

delineated: the conserved glutamate E503, D-loop residues 504−508 (EATSALD), D-loop residue 509 together with Dhelix residues 510−522, and residues 523 and 524 at the Cterminus of the D-helix. Notably, the Hingefinder algorithm groups the changes in NBD(A) D-loop residues 504−508 (EATSALD) with NBD(A) C-motif residues 477−484 (KLSGGQQQ), indicating that motions of these residues are directly coupled. In addition, the D-helix, which is positioned between the H-loop and the HSD, pivoted about residues L523 and S524 at its C-terminus (Figure S2A, Supporting Information) and moved together with HSD(A) away from its neighboring H-loop (Figure S2B, Supporting Information). In broad terms, this shows that the D-helix pivots at its Cterminus, undergoing large changes at its N-terminus, in concert with movement of the HSD and D-loop, including the conserved and putative catalytic glutamate. Absence of nucleotide from site B resulted in significant alterations in the complex, including within the opposite ATPbound active site. In PC1 (Figure 4), rotation of the CSD of NBD(B) involved movement of the P-loop, the H-loop, and the C-terminal α-helix. All three of these regions interact directly with the D-loop and/or D-helix of NBD(A), and their movement occurred in concert with changes in the conformation of the D-loop of NBD(A), including the conserved glutamate. Thus, the direct interactions of the CSD of NBD(B) with the D-loop and/or D-helix of NBD(A) couple changes in the apo NBD(B) active site to alterations in the NBD(A) active site via changes in the NBD(A) D-loop. Thus, a cooperative allostery is engaged across both NBD protomers via the D-loops/D-helices of the opposite protomers. Coupling of CH2 and the Helical Subdomain. The Hingefinder analysis of PC1 revealed coupled motions of NBD(A) HSD and CH2(B) (Figure 5) that is illustrated more generally by the covariance matrix of Cα atom fluctuations (Figure S3A, Supporting Information). Closer examination reveals that the coupling between these domains is mediated principally by Q-loop F427, residues L437 and G438 at the end of the N-terminal HSD α-helix, residues R490 and N494 at the C-terminus of the C-motif helix, and in CH2(B) by G209, V213, F216, and I218 (Figure S3B, Supporting Information). The coupling between NBD(A) HSD and CH2(B) is thus mediated largely by direct hydrophobic and van der Waals interactions between residues at their mutual interface (Figure S3B and S3C, Supporting Information). Although not directly in contact with CH2(B), motions of N434 of the ENI motif are also highly correlated with those of CH2(B) (Figure S3B, Supporting Information), as recently observed in simulations of the maltose permease system.41 N434 appears to stabilize the coupling interface, and in particular to anchor coupling residues F427 and L437 to the HSD via the interaction of adjacent I435 with the HSD hydrophobic core, and through bracing H-bonds of its side

Figure 5. Opening of the empty active site. (A) Structure of the simulated complex showing rotating domains and their hinge axes as defined by the Hingefinder analysis of the maximum and minimum projection structures of PC1. Axes are longer red or blue lines, with the two shorter same-colored lines connecting the center of mass of the rotating domain to the pivot point on the axis. Cartoon image of the protein viewed toward the cytoplasmic side, with the ICLs in the foreground (ring of α-helices) and NBDs behind; NBD(A) left and NBD(B) right. Rotating CSD of NBD(B) is cyan and its hinge axis is red (long red line roughly normal to the plane of the membrane). Rotating domain including HSD of NBD(A) and CH2(B) (residues 203−223, 426−470, 472, 474, 476, 485−494) is magenta and its hinge axis is blue. The P-loop of the empty active site is indicated. (B) Movement of HSD(A) and CH2(B). Time series of the distance moved by the geometric center of the Cα atoms of HSD(A) and CH2(B) (residues as in A) from their position in the starting structure, relative to the static domain. (C) Rotation of the NBD(B) core subdomain. Time series of the rotation of the geometric center of the Cα atoms of CSD(B) (residues 336−417, 498−499, 529−577) from their position in the starting structure, relative to the static domain, 3008

dx.doi.org/10.1021/jp211139s | J. Phys. Chem. A 2012, 116, 3004−3013

The Journal of Physical Chemistry A

Article

Figure 6. Motions of the D-loop in the ATP-bound NBD. Stereopair illustrating motions of the D-helix and D-loop in NBD(A) in PCA1. Cartoon of NBD(A), with the CSD gray, D-helix orange, D-loop and C-motif cyan, remainder of the HSD yellow, Q-loop red, H-loop blue, CH2(B) crimson, and residues 523−524 magenta. ATP is in stick form. Hinge axes are as described in Figure 5 with that of the D-helix rotation green and that of the coupled D-loop and C-motif rotation blue.

reaction with a degree of associative character,42 which is expected to be the case for the reaction in P-loop proteins.43 The trajectory was scanned for water molecules meeting the criteria of the two H-bonds and an approach distance