HIV I protease cleaves the Gag and Gag-Pol polyproteins into the mature viral proteins . It has a highly conserved pair of catalytic triads to fold into an active form for facilitating the hydrolysis of peptide bonds  at a specific active site (Asp25, Thr26, Gly27) . The catalytic triads are located at the periphery of the substrate-binding pocket and opposite the flap region, which is believed to open and close for enabling the entry and binding of the substrates . Investigation of the dynamics of the opening and closing events facilitates more comprehensive understanding of the binding of substrates or inhibitors, and of the enzyme-catalyzed cleavage process. Also, it facilitates the design of more effective inhibitors for the treatment of HIV infection.
Although the open conformation of HIV-1 protease has to be observed, a semi-open conformation has been elucidated by X-ray diffraction studies for the ligand-free enzyme [4–6], which is readily reversed into the closed conformation upon ligand binding [6,7]. Fluorescence and NMR studies have demonstrated that the flaps are flexible in the ligand-free enzyme and the Gly-rich flap region moves into the opening conformation at a sub-nanosecond time scale [8–12]. An NMR spectroscope study of the ligand-free enzyme in solution has also indicated the existence of a semi-open conformation that is dominant at 298 K, but nevertheless does not rule out the possibility of the co-existence of open and closed conformers that constitute a significantly lower percentage of the conformer population than the semi-open conformers .
Molecular dynamics simulation has been applied to characterize the open and semi-open conformations and the opening dynamics of the ligand-free enzyme [13,14], as well as the study of the rare flap opening events, dimerization processes and inhibitor or substrate binding dynamics [14–23]. Some studies have suggested that the opening process involves a transition from the closed state to a semi-open state in which the flaps in the semi-open conformation are loosely packed onto each other to form a significantly less compact structure compared with that of the crystal structure of the inhibitor-bound protease . Moreover, some simulation results have indicated that the predominant flap conformations of the dimmer is likely an ensemble of semi-open conformations with closed and fully open structures being a minor component of the overall ensemble, although the simulated conformations may be strongly influenced by the approximations in solvent modeling or force field selection [16,18,25,26].
The opening timescale presented in molecular dynamics simulations roughly follows the observed scale. For instance, simulation in the gas phase has shown that the flap opens a 25 Å gap within 200 ps . Another simulation in explicit water has shown irreversible flap opening after 3 ns . Based on these simulation results, a reliable metric has been suggested to characterize the conformation changes in molecular dynamics simulation of HIV-1 protease , such that the semi-open state is reached when the root-mean-square deviation (RMSD) of the flap residue α-carbons exceeds 3.3 Å from those of the closed state, which is to be definitively confirmed experimentally . Another proposed metric for the transition from the closed conformation to semi-open one uses the distance between the flap tips (Ile-50Cα-Ile-50'Cα), such that the distance is about 4.3 Å for the closed state and about 5.4 Å for the semi-open one .
Molecular dynamics simulations of the opening processes have indicated the existence of more than one opening path. Scott and Schiffer  have proposed that the opening is triggered by a flap curling motion, based on their observation of a hydrophobic cluster involving Ile-50 at the flap tip and the nearby Val-32, Pro-79, Thr-80 and Pro-81. Hornak  has suggested that, although some flap movements are accompanied by formation of this cluster, the more significant opening events may be associated with greater departure of flap tips away from the binding site. It is likely that different opening paths may be exposed by different simulation studies that use different approximations, such as the treatment of solvent setting and the different force fields. It has been shown that the dramatic flap arrangements do not appear in solution , although such arrangements do in simulations without solution . Meagher  proposed the vacuum cavities, which produced by the molecular movement of water, influence the flap arrangement.
It seems that the two opening paths revealed by Hornak  and Scott  are not sufficient to explain all of the experimentally detected opening behaviors. For instance, NMR or EPR (Electron Paramagnetic Resonance) experiments of the opening process have detected opening motions drastically different from a large amplitude curling motion [25,27]. Scott and Schiffer  pointed out that the crystal contact stabilizes the semi-open of HIV-1 protease and when the interaction decreases, the semi-open conformation will be destabilized and a curling conformation will trigger the opening process. Hornak  suggested that there are widely downward motions out of the flap area and the flap upward motion causes the opening process. Therefore, the opening path indicated by the NMR and EPR experiments might be different from those mentioned before [13,16]. In the present study, we tested whether additional opening paths exist, by applying the molecular dynamics simulation to the early opening process of the HIV-1 protease from the closed conformation. CHARMM 34b1 package with an explicit solvent model in a larger water tank was adopted here [28,29].
The X-ray diffraction structure of HIV-1 protease, PDB ID: 1D4L, reported by Tyndall , was used as the template of this study. In order to properly compare the simulation results with the experimental data, the mutant residues in 1D4L were restored to the original residues by the model building method using the standard geometry with the residue operation command of CHARMM. The protein structure was placed at the center of a cubic box (116 × 116 × 116 Å3) filled with TIP3P water molecules  and sodium chloride molecules. More sodium ions were added to neutralize the charges, so as to form a 0.15 mol/l sodium chloride solution with pH 7.0 under the standard conditions. Molecular dynamics simulation and energy minimization were performed using CHARMM c34b1 package [28,29], with the CHARMM force field and the periodic boundary conditions. In all cases, the cut-off radius was set at 12 Å. The Particle-Mesh Ewald method  was used to compute the nonbonded potential energy and forces between two grid points generated by the sixth order B-spline interpolation. To improve the simulation efficiency, all bonds linking hydrogen atoms were fixed with the SHAKE algorithm . The time step of the molecular dynamics simulation was set at 0.5 fs. The pressure (1 atm) and the temperature (300 K) of the systems were controlled [33–35]. The molecular dynamics method (VV2, a velocity-Verlet method in CHARMM) is based on the Operator-splitting technique [36,37].
Prior to each simulation run, the system was energy-minimized for 500 steps by using the steepest descend algorithm followed by another minimization run of 2000 steps by using the Basis Newton–Raphson algorithm, which ensures the protein to be in a sufficiently minimized conformation that is optimally adjusted to the surrounding solvent before being subjected to molecular dynamics simulations in the NVT ensemble with periodic boundary condition. Moreover, the systems were further equilibrated for 10 ps by using the VV2 method prior to the start of each simulation run in the NVT ensemble for 2.5 ns. For each molecular dynamics simulation run in the NPT ensemble, the values of RMSDs of the trajectories were recorded at every 1 ps interval with respect to the equilibrium structure. All the snapshot pictures of the molecules were generated by the VMD package .
Results and discussion
An alternative opening path from the closed to the semi-open state
Our simulation showed that the closed form of the inhibitor-free HIV-1 protease structure spontaneously converts to a semi-open conformation. This conclusion is consistent with several experimental findings that the flaps spontaneously rearrange from the closed to the semi-open conformations and the flaps exhibit significantly higher flexibility [11,12]. However, the opening motion to the semi-open conformation is in a manner differing from both the mode detected by Hornak  and the large amplitude flap curling motion reported by Scott and Schiffer . Therefore, simulated opening route by us may represent an alternative opening route to those by some of the earlier studies [13,16,26].
In our simulation, the gap between the flap tips in semi-open conformation is 15 Å wider than that in the closed conformation (a1 and b1 of Fig. 1). When the flap tips are separated widely, the tips show no further ‘opening’ (a2 and b2 Fig. 1). In contrast to the large amplitude hydrophobic cluster curling motions reported by Scott and Schiffer (Ile50 at the flap tip and the nearby Val32, Pro79, Thr80 and Pro81) , our simulation reveals only a swift and minor curling movement (with very small curling amplitude) prior to the large separation motions (Fig. 2a). The minor curling movement occurred for only about 5 ps with a small amplitude of 0.2–0.5 Å. Specifically, the time-dependent variations of the Ile50-Thr80 and the Ile50-Pro81 distances showed minor curling motions, but the variations of the Ile50-Pro79, Ile50A-Ile50B and Ile50-Val32 distances showed no identifiable curling motions (Fig. 2b). On the contrary, the opening motions discovered by Hornak  showed that the flaps move upward without curled conformation. Therefore, it is of interest to raise the possibility that the opening route presented by our simulation may correspond to those observed by some NMR or EPR experiments that failed in identifying a large amplitude curling motion [25,27]. It is noticed that the large amplitude curling motions were found in our simulation of a monomer of HIV-1 protease, which has a different molecular environment than that of the dimmer.
It seems that the opening paths are sensitive to the conditions such as solvent . Hornak  simulated the opening path with continuum solvation. Collins  simulated the process in the gas phase, whereas Scott and Schiffer  did the study with explicit solvents. In our simulation, not only the explicit solvents but also the ions are all considered. Such a treatment should be regarded closer to the real environment.
Structural features of flap opening
Previous studies have suggested that, before their opening, the flaps separate widely up to 30 Å [19,27]. Despite the discrepancy of the exact extent of opening, both these reported studies [19,27] and our study suggested that the flaps may separate wide enough to allow the entry and subsequent binding of the substrates or inhibitors to the catalytic center. In the flaps of the inhibitor-free HIV-1 protease, the flexible region is mostly localized in the tips, especially in the early stage of opening process (Fig. 3). In the first 2.0 ns, the RSMD of the backbone Cα of the HIV-1 protease and that of the tips (residues from 43 to 58) are about 2.0 Å; whereas after 2.0 ns, that of the β-hairpins increases to 4.0 Å sharply. This is in agreement with the previous study that has shown the β-hairpins are present in solution and are well ordered except at the tips [13,16]. We found that the flap β-hairpins each moves as a rigid body with highly flexible tips, with the flap elbows acting as ‘hinges’, which is consistent with the results of Hornak et al. . However, the ‘rigid body’ also has some soft characters and has some ‘fluctuating-like’ movements during the simulation. But our simulation showed that when the β-hairpins have separated up to about 10 Å, the HIV-1 protease is still at semi-open form by the metric suggested by Hornak . Other regions of the HIV-1 protease are less structurally variable. For instance, the conformations of the C-termini and N-termini ends were largely constrained by the intact hydrogen bonds during the simulation.
In the simulation of HIV-1 protease at different temperatures by Toth et al. , no conventional hydrogen bond has been found between the two flaps in the semi-open conformation, and some weakly polar interactions have been identified. Our simulation also showed no hydrogen bond between the two flaps, but there are multiple hydrogen bonds outside the flap regions between the two monomers, and these interactions play key roles in keeping the two flaps together by forming a substantial energy barrier to block the opening motions.
Our simulation study reveals the existence of an alternative opening path to the semi-open conformation of HIV-1 protease with respect to those exposed by earlier simulation studies. Our results are consistent with the opening behavior detected by some experimental studies that is otherwise un-explainable by the opening paths of the earlier simulation studies . Via this alternative path, the highly flexible flap tips are ready to open to such an extent that they allow the entrance and binding of substrates and inhibitors. Our study combined with earlier studies suggests the existence of multiple opening paths and molecular dynamics simulation is capable of identifying these paths and in a broader perspective the multiple transition paths between different states of proteins.
The present study is supported by the National Natural Science Foundation of China (No. 20873087).
1. Darke PL, Nutt RF, Brady SF, Garsky VM, Ciccarone TM, Leu CT, et al
. HIV-1 protease specificity of peptide cleavage is sufficient for processing of gag and pol polyproteins. Biochem Biophys Res Commun 1988; 156:297–303.
2. Davies DR. The structure and function of the aspartic proteinases. Annu Rev Biophys Biophys Chem 1990; 19:189–215.
3. Tyndall JD, Reid RC, Tyssen DP, Jardine DK, Todd B, Passmore M, et al
. Synthesis, stability, antiviral activity, and protease-bound structures of substrate-mimicking constrained macrocyclic inhibitors of HIV-1 protease. J Med Chem 2000; 43:3495–3504.
4. Lapatto R, Blundell T, Hemmings A, Overington J, Wilderspin A, Wood S, et al
. X-ray analysis of HIV-1 proteinase at 2.7 A resolution confirms structural homology among retroviral enzymes. Nature 1989; 342:299–302.
5. Ringhofer S, Kallen J, Dutzler R, Billich A, Visser AJ, Scholz D, et al
. X-ray structure and conformational dynamics of the HIV-1 protease in complex with the inhibitor SDZ283-910: agreement of time-resolved spectroscopy and molecular dynamics simulations. J Mol Biol 1999; 286:1147–1159.
6. Wlodawer A, Miller M, Jaskolski M, Sathyanarayana BK, Baldwin E, Weber IT, et al
. Conserved folding in retroviral proteases: crystal structure of a synthetic HIV-1 protease. Science 1989; 245:616–621.
7. Miller M, Schneider J, Sathyanarayana BK, Toth MV, Marshall GR, Clawson L, et al
. Structure of complex of synthetic HIV-1 protease with a substrate-based inhibitor at 2.3 A resolution. Science 1989; 246:1149–1152.
8. Furfine ES, D'Souza E, Ingold KJ, Leban JJ, Spector T, Porter DJ. Two-step binding mechanism for HIV protease inhibitors. Biochemistry 1992; 31:7886–7891.
9. Rodriguez EJ, Debouck C, Deckman IC, Abu-Soud H, Raushel FM, Meek TD. Inhibitor binding to the Phe53Trp mutant of HIV-1 protease promotes conformational changes detectable by spectrofluorometry. Biochemistry 1993; 32:3557–3563.
10. Ishima R, Freedberg DI, Wang YX, Louis JM, Torchia DA. Flap opening and dimer-interface flexibility in the free and inhibitor-bound HIV protease, and their implications for function. Structure 1999; 7:1047–1055.
11. Freedberg DI, Ishima R, Jacob J, Wang YX, Kustanovich I, Louis JM, et al
. Rapid structural fluctuations of the free HIV protease flaps in solution: relationship to crystal structures and comparison with predictions of dynamics calculations. Protein Sci 2002; 11:221–232.
12. Nicholson LK, Yamazaki T, Torchia DA, Grzesiek S, Bax A, Stahl SJ, et al
. Flexibility and function in HIV-1 protease. Nat Struct Biol 1995; 2:274–280.
13. Scott WR, Schiffer CA. Curling of flap tips in HIV-1 protease as a mechanism for substrate entry and tolerance of drug resistance. Structure 2000; 8:1259–1265.
14. Collins JR, Burt SK, Erickson JW. Flap opening in HIV-1 protease simulated by ‘activated’ molecular dynamics. Nat Struct Biol 1995; 2:334–338.
15. York DM, Darden TA, Pedersen LG, Anderson MW. Molecular dynamics simulation of HIV-1 protease in a crystalline environment and in solution. Biochemistry 1993; 32:1443–1453.
16. Hornak V, Okur A, Rizzo RC, Simmerling C. HIV-1 protease flaps spontaneously open and reclose in molecular dynamics simulations. Proc Natl Acad Sci U S A 2006; 103:915–920.
17. Hornak V, Okur A, Rizzo RC, Simmerling C. HIV-1 protease flaps spontaneously close to the correct structure in simulations following manual placement of an inhibitor into the open state. J Am Chem Soc 2006; 128:2812–2813.
18. Perryman AL, Lin JH, McCammon JA. Restrained molecular dynamics simulations of HIV-1 protease: the first step in validating a new target for drug design. Biopolymers 2006; 82:272–284.
19. Toth G, Borics A. Flap opening mechanism of HIV-1 protease. J Mol Graph Model 2006; 24:465–474.
20. Hou T, Yu R. Molecular dynamics and free energy studies on the wild-type and double mutant HIV-1 protease complexed with amprenavir and two amprenavir-related inhibitors: mechanism for binding and drug resistance. J Med Chem 2007; 50:1177–1188.
21. Lauria A, Ippolito M, Almerico AM. Molecular dynamics studies on HIV-1 protease: a comparison of the flap motions between wild type protease and the M46I/G51D double mutant. J Mol Model 2007; 13:1151–1156.
22. Tozzini V, Trylska J, Chang CE, McCammon JA. Flap opening dynamics in HIV-1 protease explored with a coarse-grained model. J Struct Biol 2007; 157:606–615.
23. Sadiq SK, Wan S, Coveney PV. Insights into a mutation-assisted lateral drug escape mechanism from the HIV-1 protease active site. Biochemistry 2007; 46:14865–14877.
24. Louis JM, Dyda F, Nashed NT, Kimmel AR, Davies DR. Hydrophilic peptides derived from the transframe region of Gag-Pol inhibit the HIV-1 protease. Biochemistry 1998; 37:2105–2110.
25. Ishima R, Louis JM. A diverse view of protein dynamics from NMR studies of HIV-1 protease flaps. Proteins 2008; 70:1408–1415.
26. Meagher KL, Carlson HA. Solvation influences flap collapse in HIV-1 protease. Proteins 2005; 58:119–125.
27. Ding F, Layten M, Simmerling C. Solution structure of HIV-1 protease flaps probed by comparison of molecular dynamics simulation ensembles and EPR experiments. J Am Chem Soc 2008; 130:7184–7185.
28. Brooks BR, Bruccoleri RE, Olafson DJ, States DJ, Swaminathan S, Karplus M. CHARMM: a Program for Macromolecular Energy, Minimization, and Dynamics Calculations. J Comput Chem 1983; 4:187–217.
29. MacKerel Jr AD, Brooks Iii CL, Nilsson L, Roux B, Won Y, Karplus M. CHARMM: the energy function and its parameterization with an overview of the program
. Chichester: John Wiley & Sons; 1998. pp. 271–277.
30. Jorgensen WL, Chandrasekhar J, Madura JD, Impey RW, Klein ML. Comparison of simple potential functions for simulating liquid water. J Chem Phys 1983; 79:926–935.
31. Tom D, Darrin Y, Lee P. Particle mesh Ewald: an N [center-dot] log(N) method for Ewald sums in large systems. J Chem Phys 1993; 98:10089–10092.
32. Ryckaert J-P, Ciccotti G, Berendsen HJC. Numerical integration of the cartesian equations of motion of a system with constraints: molecular dynamics of n-alkanes. J Comput Phys 1977; 23:327–341.
33. Parrinello M, Rahman A. Crystal-structure and pair potentials: a Molecular-Dynamics study. Phys Rev Lett 1980; 45:1196–1199.
34. Hoover WG. Canonical dynamics: equilibrium phase-space distributions. Phys Rev A 1985; 31:1695–1697.
35. Parrinello M, Rahman A. Polymorphic transitions in single-crystals: a new molecular-dynamics method. J Appl Phys 1981; 52:7182–7190.
36. Martyna GJ, Tuckerman ME, Tobias DJ, Klein ML. Explicit reversible integrators for extended systems dynamics. Mol Phys 1996; 87:1117–1157.
37. Lamoureux G, MacKerell AD, Roux B. A simple polarizable model of water based on classical Drude oscillators. J Chem Phys 2003; 119:5185–5197.
38. Humphrey W, Dalke A, Schulten K. ‘VMD - Visual Molecular Dynamics’. J Mol Graph 1996; 14:33–38.