5829
2006, 110, 5829-5833 Published on Web 03/03/2006
Peptide Plane Can Flip in Two Opposite Directions: Implication in Amyloid Formation of Transthyretin Mingfeng Yang,†,| Ming Lei,†,‡,| Boyan Yordanov,†,§ and Shuanghong Huo*,† Gustaf H. Carlson School of Chemistry and Biochemistry, Clark UniVersity, 950 Main Street, Worcester, Massachusetts 01610, and Department of Chemistry, School of Science, Beijing UniVersity of Chemical Technology, Beijing, 100029, P. R. China ReceiVed: December 3, 2005; In Final Form: February 15, 2006
Transthyretin (TTR) is one of the known 20 or so human proteins that form fibrils in vivo, which is a hallmark of amyloid diseases. Recently, molecular dynamics simulations using ENCAD force field have revealed that under low pH conditions, the peptide planes of several amyloidogenic proteins can flip in one direction to form an R-pleated structure which may be a common conformational transition in the fibril formation. We performed molecular dynamics simulations with AMBER force fields on a recently engineered double mutant TTR, which was shown experimentally to form amyloid fibrils even under close to physiological conditions. Our simulations have demonstrated that peptide-plane flipping can occur even under neutral pH and room temperature for this amyloidogenic TTR variant. Unlike previously reported peptide-plane flipping of TTR using ENCAD force field, we have found two-way flipping using AMBER force field. We propose a new mechanism of amyloid formation based on the two-way flipping, which gives a better explanation of various experimental and computational results. In principle, the residual dipolar and hydrogen-bond scalar coupling techniques can be applied to the wild-type TTR and the variant to study the peptide-plane flipping of amyloidogenic proteins.
Introduction Transthyretin (TTR) is one of the known 20 or so human proteins that form fibrils in vivo, which is a hallmark of amyloid diseases.1 TTR is a plasma protein responsible for carrying thyroid hormones in plasma and cerebrospinal fluid as well as indirect transport of vitamin A (for a recent review see ref 2). In its native state, TTR is a homotetramer with each monomer consisting of eight β-strands (A-H) organized into the inner (DAGH) sheet and outer (CBEF) sheet (Figure 1).3 The monomer-monomer interface is formed by the F and F′ strands (from the adjacent monomers) as well as H and H′ strands (from the same adjacent monomers). The dimer-dimer interface involves the contacts between loop regions. The TTR amyloid formation can be accelerated either by low pH environment4 or fatal point mutations.5 The pathway of TTR amyloid formation has been proposed based on lines of experimental evidence: TTR tetramers first dissociate into native-like monomers, which is shown to be a rate-limiting step in the formation of fibrils.6-8 Subsequently, the monomeric species partially unfold to form the aggregation intermediates. Once such intermediates are formed, the following self-assembly process is a downhill polymerization.9 Based on pieces of information from various * Corresponding author: Phone: 508-793-7533, Fax: 508-793-8861, E-mail:
[email protected]. † Clark University. ‡ Beijing University of Chemical Technology. § Current address: Biomedical engineering graduate program, Boston University, Boston, Massachusetts 02215. | These two authors equally contributed to the work.
10.1021/jp0570420 CCC: $33.50
Figure 1. Monomeric structure of TTR. The inner sheet (DAGH) is in cyan, while the outer sheet (CBEF) is in magenta. The V14N and V16E, which are shown explicitly using the ball-and-stick model, are virtually mutated from the wild-type monomer structure (PDB entry: 1DVQ chain A). The atoms of N14 and E16 are colored with the following scheme: C is in gray, N is in blue, H is in white, and O is in red.
experimental results and simulations, an emerging picture about the TTR amyloidogenic intermediate state is that it adopts an altered tertiary structure with the C-strand-loop-D-strand region dislocated from the core of the protein.4,10,11 The fibril state depicted according to the results of site-directed spin labeling and NMR combined with DH-exchange is that the A, B, E, F, G, and H strands retain their native-like conformation and constitute the fibril core with the C and D strands dislocated © 2006 American Chemical Society
5830 J. Phys. Chem. B, Vol. 110, No. 12, 2006 from their native region.12,13 It has been suggested that the subunit interface in the fibril state is formed between strands A and A′, B and B′, F and F′, and H and H′ with possible shift in register between the two F strands as well as between the two H strands.12,13 However, the structural details of the aggregation intermediates and how they associate with each other remain elusive. The amyloid fibrils originating from different proteins or peptides share a common cross-β motif,14 even though multiple distinct morphologies are found.15 An hypothesis on the formation of common β-pleated sheets from nonhomologous proteins is that, while these proteins are not related in their native folded states, their amyloidogenic intermediate conformations share common features upon partial denaturation, which facilitates self-assembly into amyloid fibrils.16,17 Supporting evidence of this hypothesis emerges from the recent discovery of an oligomer-specific antibody which binds to amyloid-soluble oligomers, regardless of sequence, but binds neither their precursor proteins nor the fibrils.18 Further direct evidence is from the molecular dynamics simulations at low pH that have suggested the formation of R-pleated sheet structure may be a common conformational transition in the fibril formation.19-21 The unique R-pleated sheet structure explains well why the antibody’s oligomer-specificity is sequence independent. In this paper, we performed molecular dynamics simulations on a recently engineered double mutant of TTR(V14N/V16E) (TTR-A) which was shown experimentally to form amyloid fibrils even under close-to-physiological conditions,22 while the wild-type is nonamyloidogenic under the same conditions. To eliminate possible artifacts caused by high temperature or extremely low pH simulations, we chose to run the simulations at room temperature and neutral pH. The analysis was focused on the R-sheet transition and a new mechanism of TTR fibril formation was proposed. Materials and Methods All molecular dynamics simulations were performed at neutral pH and constant temperature, 300 K. The AMBER simulation package23 and the all-atomic force field, parm94,24 were employed. The chain A of the X-ray TTR tetrameric form (PDB entry: 1DVQ25) was used as the starting point of the simulation. The double mutant (TTR-A) was generated by substituting Val14 and Val16 in the wild-type with Asn and Glu, respectively. To keep the whole system neutral, four Na+ counterions were added for the wild-type, and five for TTR-A. The TIP3P three-site rigid water model26 was used to solvate the protein and counterions. The system was constructed using the periodic boundary conditions consisting of the protein, counterions, and water molecules that totaled up to 26730 atoms. The MD simulations were performed in the N, P, and T ensemble. The temperature and pressure of the system were regulated using the Berendsen coupling algorithm27 with a coupling constant of 1.0 ps. The particle mesh Ewald summation method28 was used to treat the long-range electrostatic interactions. The shortrange nonbonded interactions were truncated with a 9 Å cutoff. All bond lengths involving H atoms were constrained with the SHAKE algorithm.29 The time step was 1.5 fs and the nonbonded pair list was updated every 25 steps. The system was minimized 1000 steps to relax the bad contacts. After that, the system was gradually heated to 300 K and then equilibrated for 30 ps. In the production run, the snapshots were saved every 1.5 ps. With this protocol, six independent trajectories were collected, among which four were 4.8 ns (3200 conformations) and the other two were 20.4 ns (13600 conformations) for both
Letters
Figure 2. Schematic representation of a β-sheet and an R-sheet. Only the backbone atoms are presented. The R-pleated sheet proposed by Pauling and Corey30 consists of residues that adopt alternating RR(φ ) -87(30°, φ ) -49(30°) and RL(φ ) 45(30°, φ ) 92(30°)20 conformations with the main-chain NH group aligned on one side of the sheet and the CO group on the other side. As a result, each NH/ CO group forms two hydrogen bonds with two other CO/NH groups from a neighboring strand. Therefore, each residue in the R-pleated sheet structure forms more hydrogen bonds than that in the β-sheet.
wild-type TTR (WT-TTR) and TTR-A. Totally, 120 ns MD simulations were carried out. Results and Discussion Definition of r-Sheet Structure. The R-pleated sheet proposed by Pauling and Corey30 consists of residues that adopt alternating RR(φ ) -87(30°, φ ) -49(30°) and RL(φ ) 45(30°, φ ) 92(30°)20 conformations with the main-chain NH group aligned on one side of the sheet and the CO group on the other side. As a result, each NH/CO group forms two hydrogen bonds with two other CO/NH groups from a neighboring strand. Therefore, each residue in the R-pleated sheet structure forms more hydrogen bonds than that in the β-sheet (Figure 2). However, according to the recent MP2/cc-pVTZ// MP2/6-31G** study of alanine dipeptide, the RL conformation is in the energetically less favorable region of the Ramachandran plot.31 Obviously, the high energy cost of RL is compensated by the bifurcated hydrogen bonding in the R-sheet. The transition from β-sheet to R-sheet requires the peptide-plane to flip. In our analysis, residue i was considered to adopt an R-sheet conformation if it satisfied the following two criteria simultaneously: (1) it formed an RR and RL pair with either residue i-1 or i+1; (2) it took part in the bifurcated hydrogen bonding with its neighboring strand(s). The hydrogen bonding analysis was carried out with DSSP program,32 and the energy cutoff was -0.4 kcal/mol. We have found that the dihedral angle falling into the RR and RL region did not guarantee that bifurcated hydrogen bonds can be formed. For example, in various loop regions, we found residues in the alternating RR and RL regions; however, none of them formed the bifurcated hydrogen bonds. The bifurcated hydrogen bonding in turn did not warrant that the dihedral angle fell into the RR and RL region. The bifurcated hydrogen bonds can be formed due to local distortions, rather than the peptide plane flipping. Therefore, we listed the residues which met both requirements simultaneously in a trajectory with the occurrence frequency higher than 0.5% in Table 1. Peptide Plane Can Flip in Two Opposite Directions. We first performed six independent runs for both TTR-A and the wild-type for 4.8 ns. The root-mean-square deviation as a function of time shown in the supplementary data indicates that the statistical convergence has been attained in 4.8 ns (Figure S1, Supporting Information). We have found continuous bifur-
Letters
J. Phys. Chem. B, Vol. 110, No. 12, 2006 5831
TABLE 1: Bifurcated Hydrogen Bonding Networka # of snapshots Percentage
Hydrogen bond pair TTR
G
H
TTR-A
Thr106 CO Ile107 CO Thr106 CO Ile107 CO Ile107 CO Ala108 CO Ile107 CO Ala108 CO Ala109 CO Leu110 CO Ser112 CO
Ala120 NH
Leu110 NH WT
Thr106 NH
T3
T4
1313 9.7% 131 1.0% 2542 18.7% 179 1.3%
Val121 NH Thr119 NH Ala120 NH Ser117 NH Ser115 NH Tyr116 NH Tyr116 CO Ser117 CO Ala120 CO Val121 CO
T5
136 4.3%
428 13.4%
51 1.6%
83 0.6% 279 2.1% 153 4.8%
a We listed the residues which satisfied R-sheet requirements in a trajectory with the occurrence frequency higher than 0.5%. Both the occurrence frequency and the number of snapshots in the trajectory that retained the flipped peptide plane were presented. T3, T4, T5 denote trajectories 3, 4, and 5, respectively. The NH or CO group involved in the peptide plane flipping is in bold. For example, the first two lines indicate that Ala120(NH) forms bifurcated hydrogen bonds with Thr106(CO) and Ile107(CO), among which the peptide planes of Ile107-Ala108 and Thr119-Ala120 flipped. In this table, only T5 of TTR-A is 20.4 ns. Others are 4.8 ns.
cated hydrogen bonding only in trajectory 5 (T5) of TTR-A. To probe the stability of the initially formed R-sheet located in the AGH strands, we continued to run T5 up to 20.4 ns. It turned out that the frequency of R-sheet formation during the course of the simulation is about the same for the first 4.8 ns and the whole 20.4 ns. Considering the computational cost, we randomly picked another trajectory as control to continue to run up to 20.4 ns, instead of running all of the trajectories for 20.4 ns. For the wild-type, since no extended R-sheet hydrogen bonding was found, we randomly picked two trajectories to continue to run for 20.4 ns. By the residual dipolar and hydrogen-bond scalar coupling techniques, the tendency of peptide-plane flipping has been confirmed to be natural for β-sheet conformation and the motion occurs in the nanosecond to millisecond time scale.33 However, the formation of an extended R-sheet under neutral pH and 300 K may be beyond the millisecond time scale. As seen in Table 1, although the peptide plane flipping was observed in simulations of both WT-TTR and TTR-A, the continuous bifurcated hydrogen bonding was seen in T5 of TTRA, on the contrary, only a pair of residues, Ala120 and Val121, adopted alternating RR and RL with high occurrence frequency (>0.5%) in the simulations of WT-TTR. The residues involved in the R-sheet formation were exclusively located in the AGH strands (in the inner sheet), whereas the CBEF strands (in the outer sheet) were stable during the simulations and none of their residues was found to participate in the peptide plane flipping. In the simulations of TTR-A, Lys15 and Glu16 in the A strand, Ile107, Ala108, Ala109, and Leu110 in the G strand, and Tyr116, Ser117, Thr118, Thr119, Ala120, and Val121 in the H strand can form an RR/RL pair with its neighboring residue at least once. Interestingly, the majority of these residues (9 out of 12) plus three more residues in the GH strands (Tyr105, Thr106, and Val122) also were involved in the RR/RL pair in the simulations of the wild-type, despite no continuous RR/RL pairs. The formed R-sheet was very dynamic in that the peptideplane flipped back and forth and, meanwhile, the bifurcated hydrogen bonds underwent breaking and reforming along the
course of simulation. The longest time interval when bifurcated hydrogen bonds maintained in our simulations was 531 ps between Ile107(CO) and Ala108(CO) and Thr119 (NH) in TTRA. Even though the peptide plane flipping in our simulations (under neutral pH and 300 K) occurred relatively infrequently and the formed R-sheet was small compared to the simulations at high temperature and lower pH,19,20 we did see the peptide plane flipping and R-sheet. Interestingly, in our simulations the peptide plane can flip in two opposite directions (Table 1), while Daggett and co-workers observed only one-way flipping.19,20 For example, in T5(TTRA), after the peptide planes of Ile107-Ala108 and Thr119Ala120 flipped, the main-chain carbonyl groups of Ile107 and Thr119 pointed to the H strand and solvent, respectively; as a result, a relatively stable R-sheet hydrogen bonding network was formed, consisting of residues 106, 107, 108, 119, 120, and 121 (Figure 3, middle panel). We call this flipping “COoutward flipping”, which was not observed in the simulations of Daggett and co-workers.19,20 If all of the peptide planes flip in the same direction, the main-chain CO group of the H strand (which is the edge strand) will point toward the solvent while the main-chain NH of the same strand will point inward. Meanwhile, in the same trajectory, we have observed flipping toward the opposite direction. The peptide plane of Tyr116Ser117 (in the H strand) flipped, causing the main-chain CO group of Tyr116 to point to the G strand and the bifurcated hydrogen bonds to be formed between Leu110(NH) and Tyr116(CO) and Ser117(CO)(Figure 3, middle panel). This CO-inward flipping was reported previously,19,20 causing the main-chain CO of the H strand to point inward the core of the protein. Our simulations demonstrated that the edge strand H can have aligned either carbonyl groups or amide groups toward the solvent, instead of the latter alignment only.19,20 Recent experiments have suggested that the TTR fibrillar structure may retain a native-like conformation in which the fibril core consists of A, B, E, F, G, and H strands while the C-strand-loop-D-strand is dislocated and a native-like interface is formed between F-F′ and H-H′ strands with another subunit interface contributed by A-A′ and B-B′ strands.12,13 It was discovered that residues Val122 and Asn124 in the H strand and Asn98 in the F strand were protected from DH exchange in the fibril state; however, these residues were completely exposed to the solvent in its native structure.13,34 A possible explanation for the change in protection factor is that there is a shift in register between the H and H′ strands and F and F′ strands in the fibril state. Coincidently, we found a shift between the G and H strands within a monomer in our simulation. As seen in Table 1, Thr106(CO) and Ile107(CO) can form a bifurcated hydrogen bond with either residue Ala120(NH) or Val121(NH). The hydrogen bonding depicted in Figure 3 illustrates that Thr106(CO) and Ile107(CO) form bifurcated hydrogen bonds with Ala120(NH) without a shift in the H strand (Figure 3, middle panel), while the hydrogen bonding among Thr106(CO), Ile107(CO), and Val121(NH) involves a shift toward the N-terminus of the H strand (Figure 3, right panel). An extended R-sheet may require half or one residue shift in register of the whole H strand. If the H strand of all TTR monomers shifts half or one residue toward its N-terminus, since the H′ and H strands are antiparallel, the H-H′ interface will shift in register by 1 to 2 residues in aggregation, which is in line with the high protection factor of Val122 and Asn124 in the fibril state.13 Based on our simulations and other evidence,12,13 we propose a new mechanism of TTR aggregation, which can well explain
5832 J. Phys. Chem. B, Vol. 110, No. 12, 2006
Letters
Figure 3. Structures of the DAGH (inner sheet) strands in T5 of TTR-A. The coloring scheme is that C is in gray, N is in blue, and O is in red. The residue number is labeled near the backbone oxygen of each residue, and the backbone N of the same residue can be traced upstream. Hydrogen atoms are not shown. Left panel: The initial structure of the simulation. Middle panel: The snapshot at 12.3 ns. The peptide plane of Ile107Ala108 and Thr119-Ala120 underwent the CO-outward flipping, while the peptide plane of Tyr116-Ser117 was involved in the CO-inward flipping. Thr106 (CO) and Ile107 (CO) form bifurcated hydrogen bonds with Ala120(NH) without a shift in the H strand. Right panel: The snapshot at 5.98 ns. The H strand shifted toward its N-terminus. For example, originally, the backbone N of Val121 was below the backbone O of Thr106 (t ) 0 ns). At t ) 5.98 ns, the N atom (of Val121) is slightly above the O atom of Thr106. As a result, Thr106 (CO) and Ile107 (CO) form bifurcated hydrogen bonds with Val121(NH).
Figure 4. Two possible subunit interfaces formed by complementary partially charged strands. The H-H′ association was suggested experimentally.12,13 The H-A′ association could be prevented by other mechanism(s).
various observations (Figure 4). Due to the two-way flipping, the edge strand can have either CO or NH group aligned toward the solvent, leading to two complementary partially charged interfaces. When the two complementary partially charged H strands associate, the R-strand H-H′ interface will be formed in the fibril state (Figure 4) with shift in register by 1-2 residues from the native-state interface. Since the CBEF strands were not found to form an R-sheet in both our simulations and others,19,20 we propose that the F-F′ interface is formed by β-strands. The shift in β-sheet register of the F strands in the fibril state could be due to the changes in the F strand distortions.11 Each TTR protofilament could be double layered, in which one layer is an R-sheet formed by AGH strands and the other layer is a β-sheet consisting of BEF strands. When the protofilaments wrap around each other to form the fibril, the R-sheet layer can be buried inside while the β-sheet layer is exposed to the outside. Thereby, the R-sheet conformation, which is formed in the amyloidogenic intermediate state, does not have to flip back to form a β-sheet in the fibril state as proposed previously.20 Our new model provides a possible explanation for the fact that the oligomer-specific antibody inhibits the soluble intermediates but not the mature fibril, because it recognizes the R-sheet exposed to the solvent in the oligomeric state, but has no way to access to the R-sheet that is buried in the fibril state. However, our model does not exclude the possibility that the two complementary partially charged H
and A strands form a non-native-like subunit interface. There could be other mechanism(s) to prevent this association. Conclusions Our simulation has demonstrated that peptide plane flipping can occur even under neutral pH and room temperature conditions for an amyloidogenic TTR variant. Unlike previously reported peptide plane flipping of TTR19,20 using ENCAD35 force field, we have found two-way flipping using parm94 force field.24 We propose a new mechanism of amyloid formation based on the two-way flipping, which gives a better explanation of various experimental and computational observations. In principle, the residual dipolar and hydrogen-bond scalar coupling techniques33 can be applied to the wild-type TTR and the variant to study the peptide-plane flipping of amyloidogenic proteins. Acknowledgment. This work was supported by National Institutes of Health (1R15 AG025023-01) and National Science Foundation Major Research Instrumentation (DBI-0320875) and partially supported by Petroleum Research Fund (PRF No. 39205-G4) and Beijing Novel Fund (2005B17). The molecular graphics images were generated using the Chimera36 package from the Computer Graphics Laboratory, University of California, San Francisco, CA (supported by National Institutes of Health P41 RR-01081). We thank the National Center of Supercomputing Applications and the Scientific Computing and
Letters Visualization group at Boston University for providing part of the computational resources. Supporting Information Available: Figure S.1. Root-meansquare deviation as a function of time for the four 20.4 ns simulations. This material is available free of charge via the Internet at http://pubs.acs.org. References and Notes (1) Sacchettini, J. C.; Kelly, J. W. Nat. ReV. Drug DiscoVery 2002, 1, 267. (2) Hamilton, J. A.; Benson, M. D. Cell Mol. Life Sci. 2001, 58, 1491. (3) Blake, C. C.; Geisow, M. J.; Oatley, S. J.; Rerat, B.; Rerat, C. J. Mol. Biol. 1978, 121, 339. (4) Lai, Z.; Colon, W.; Kelley, J. W. Biochemistry 1996, 35, 6470. (5) Lashuel, H. A.; Wurth, C.; Woo, L.; Kelly, J. W. Biochemistry 1999, 38, 13560. (6) Hammarstrom, P.; Schneider, F.; Kelly, J. W. Science 2001, 293, 2459. (7) Schneider, F.; Hammarstrom, P.; Kelly, J. W. Protein Sci. 2001, 10, 1606. (8) Hammarstrom, P.; Jiang, X.; Hurshman, A. R.; Powers, E. T.; Kelly, J. W. Proc. Natl. Acad. Sci. U.S.A. 2002, 99 Suppl 4, 16427. (9) Hurshman, A. R.; White, J. T.; Powers, E. T.; Kelly, J. W. Biochemistry 2004, 43, 7365. (10) Goldsteins, G.; Persson, H.; Andersson, K.; Olofsson, A.; Dacklin, I.; Edvinsson, A.; Saraiva, M. J.; Lundgren, E. Proc. Natl. Acad. Sci. U.S.A. 1999, 96, 3108. (11) Yang, M.; Lei, M.; Bruschweiler, R.; Huo, S. Biophys. J. 2005, 89, 433. (12) Serag, A. A.; Altenbach, C.; Gingery, M.; Hubbell, W. L.; Yeates, T. O. Nat. Struct. Biol. 2002, 9, 734. (13) Olofsson, A.; Ippel, J. H.; Wijmenga, S. S.; Lundgren, E.; Ohman, A. J. Biol. Chem. 2004, 279, 5699. (14) Sunde, M.; Blake, C. AdV. Protein Chem. 1997, 50, 123. (15) Petkova, A. T.; Leapman, R. D.; Guo, Z.; Yau, W. M.; Mattson, M. P.; Tycko, R. Science 2005, 307, 262.
J. Phys. Chem. B, Vol. 110, No. 12, 2006 5833 (16) Kelly, J. W. Curr. Opin. Struct. Biol. 1996, 6, 11. (17) Kelly, J. W. Curr. Opin. Struct. Biol. 1998, 8, 101. (18) Kayed, R.; Head, E.; Thompson, J. L.; McIntire, T. M.; Milton, S. C.; Cotman, C. W.; Glabe, C. G. Science 2003, 300, 486. (19) Armen, R. S.; DeMarco, M. L.; Alonso, D. O.; Daggett, V. Proc. Natl. Acad. Sci. U.S.A. 2004, 101, 11622. (20) Armen, R. S.; Alonso, D. O.; Daggett, V. Structure (Camb) 2004, 12, 1847. (21) Armen, R. S.; Bernard, B. M.; Day, R.; Alonso, D. O.; Daggett, V. Proc. Natl. Acad. Sci. U.S.A. 2005, 102, 13433. (22) Olofsson, A.; Ippel, H. J.; Baranov, V.; Horstedt, P.; Wijmenga, S.; Lundgren, E. J. Biol. Chem. 2001, 276, 39592. (23) Pearlman, D.; Case, D. A.; Caldwell, J. W.; Ross, W. S.; Cheatham, T. E.; DeBolt, S.; Ferguson, D.; Seibel, G.; Kollman, P. A. Comput. Phys. Commun. 1995, 91, 1. (24) Cornell, W. D.; Cieplak, P.; Bayly, C. I.; Gould, I. R.; Merz, J. K. M.; Ferguson, D. M.; Spellmeyer, D. C.; Fox, T.; Caldwell, J. W.; Kollman, P. A. J. Am. Chem. Soc. 1995, 117, 5179. (25) Klabunde, T.; Petrassi, H. M.; Oza, V. B.; Raman, P.; Kelly, J. W.; Sacchettini, J. C. Nature Struct. Biol. 2000, 7, 312. (26) Jorgensen, W. L. J. Chem. Phys. 1982, 77, 4156. (27) Berendsen, H. J. C.; Postma, J. P. M.; van Gunsteren, W. F.; DiNola, A.; Haak, J. R. J. Chem. Phys. 1984, 81, 3684. (28) Darden, T.; York, D.; Pedersen, L. J. Chem. Phys. 1993, 98, 10089. (29) Ryckaert, J. P.; Ciccotti, G.; Berendsen, H. J. C. J. Comput. Phys. 1977, 23, 327. (30) Pauling, L.; Corey, R. B. Proc. Natl. Acad. Sci. U.S.A. 1951, 37, 251. (31) Wang, Z. X.; Duan, Y. J. Comput. Chem. 2004, 25, 1699. (32) Kabsch, W.; Sander, C. Biopolymers 1983, 22, 2577. (33) Bouvignies, G.; Bernado, P.; Meier, S.; Cho, K.; Grzesiek, S.; Bruschweiler, R.; Blackledge, M. Proc. Natl. Acad. Sci. U.S.A. 2005, 102, 13885. (34) Liu, K.; Cho, H. S.; Hoyt, D. W.; Nguyen, T. N.; Olds, P.; Kelly, J. W.; Wemmer, D. E. J. Mol. Biol. 2000, 303, 555. (35) Levitt, M.; Hirshberg, M.; Sharon, R.; Daggett, V. Comput. Phys. Commun. 1995, 91, 215. (36) Pettersen, E. F.; Goddard, T. D.; Huang, C. C.; Couch, G. S.; Greenblatt, D. M.; Meng, E. C.; Ferrin, T. E. J. Comput. Chem. 2004, 25, 1605.
