Molecular dynamics simulations of ribonuclease T1: comparison of the free enzyme and the 2' GMP-enzyme complex

Molecular dynamics simulations were performed on free RNase T1 and the 2'GMP-RNase T1 complex in vacuum and with water in the active site along with crystallographically identified waters, allowing analysis of both active site and overall structural and dynamics changes due to the presence of 2'GMP. Differences in the active site include a closing in the presence of 2'GMP, which is accompanied by a decrease in mobility of active site residues. The functional relevance of the active site fluctuations is discussed. 2'GMP alters the motion of Tyr-45, suggesting a role for that residue in providing a hydrophobic environment for the protein-nucleic acid interactions responsible for the specificity of RNase T1. The presence of 2'GMP causes a structural change of the C-terminus of the alpha-helix, indicating the transmission of structural changes from the active site through the protein matrix. Overall fluctuations of both the free and 2'GMP enzyme forms are in good agreement with X-ray temperature factors. The motion of Trp-59 is influenced by 2'GMP, indicating differences in enzyme dynamics away from the active site, with the calculated changes following those previously seen in time-resolved fluorescence experiments.     

Structural and dynamic properties of cytochrome P450 BM-3 in pure water and in a dimethylsulfoxide/water mixture

Solvent molecules play an important role for the structural and dynamical properties of proteins. A major focus of current protein engineering is the development of enzymes that are catalytically active in the presence of organic solvents. The monooxygenase P450 BM-3 is one of the best-studied enzymes and promising for industrial applications but with limited activity in the presence of organic solvents or cosolvents. To gain insights into the structural and dynamical properties of the heme domain of this enzyme in solution, molecular dynamics simulations in pure water and in a 14% DMSO/water mixture were performed. The results of the simulations show overall similar structural fluctuations in both solvent systems, with no indication of partial or global unfolding. In 14% DMSO, the regions comprising the helices E, F, and the EF loop (implicated in controlling the entry to the active site channel) undergo a large shift. Significant changes were also observed near the active site access channel at the residue R47. During the simulation, no DMSO molecule penetrated the active site. However, a significant accumulation of DMSO molecules close to the substrate-binding site and to the Flavin Mononucleotide (FMN) reductase domain interface was observed.     

Solvent effects on protein motion and protein effects on solvent motion. Dynamics of the active site region of lysozyme

The stochastic boundary molecular dynamics methodology is applied to the active site of the enzyme lysozyme. A comparison is made of in vacuo dynamics results from the stochastic boundary method and a full conventional molecular dynamics simulation of lysozyme. Excellent agreement between the two approaches is obtained. The influence of solvent on the residues in the active site region is explored and it is shown that both the structure and dynamics are affected. Of particular importance for the structure of the protein is the solvation of polar residues and the stabilization of like-charged ion pairs. The magnitude of the fluctuations is only slightly altered by the solvent; the overall increase in the root-mean-square fluctuations, relative to the vacuum run, is 11%. The solvent effect on dynamical properties is found not to be simply related to the solvent viscosity. Both the solvent exposure and dynamic aspects of protein-solvent interactions, including the relative time scales of the motions, are shown to play a role. The effects of the protein on solvent dynamics and structure are also observed to be significant. The solvent molecules around atoms in charged, polar and apolar side-chains show markedly different diffusion coefficients as well as exhibiting different solvation structures. One key example is the water around apolar groups, which is much less mobile than bulk water, or water solvating polar groups.     

Structural determinants of ligand imprinting: a molecular dynamics simulation study of subtilisin in aqueous and apolar solvents

The phenomenon known as "ligand imprinting" or "ligand-induced enzyme memory" was first reported in 1988, when Russell and Klibanov observed that lyophilizing subtilisin in the presence of competitive inhibitors (that were subsequently removed) could significantly enhance its activity in an apolar solvent. (Russell and Klibanov, J Biol Chem 1988;263:11624-11626). They further observed that this enhancement did not occur when similar assays were carried out in water. Herein, we shed light on the molecular determinants of ligand imprinting using a molecular dynamics (MD) approach. To simulate the effect of placing an enzyme in the presence of a ligand before its lyophilization, an inhibitor was docked in the active site of subtilisin and 20 ns MD simulations in water were performed. The ligand was then removed and the resulting structure was used for subsequent MD runs using hexane and water as solvents. As a control, the same simulation setup was applied using the structure of subtilisin in the absence of the inhibitor. We observed that the ligand maintains the active site in an open conformation and that this configuration is retained after the removal of the inhibitor, when the simulations are carried out in hexane. In agreement with experimental findings, the structural configuration induced by the ligand is lost when the simulations take place in water. Our analysis of fluctuations indicates that this behavior is a result of the decreased flexibility displayed by enzymes in an apolar solvent, relatively to the aqueous situation.     

Molecular dynamics simulations of arachidonic acid complexes with COX-1 and COX-2: insights into equilibrium behavior

The cyclooxygenase (COX) enzymes are responsible for the committed step in prostaglandin biosynthesis, the generation of prostaglandin H(2). As a result, these enzymes are pharmacologically important targets for nonsteroidal antiinflammatory drugs, such as aspirin and newer COX-2 selective inhibitors. The cyclooxygenases are functional homodimers, and each subunit contains both a cyclooxygenase and a peroxidase active site. These enzymes are quite interesting mechanistically, as the conversion of arachidonic acid to prostaglandin H(2) requires two oxygenation and two cyclization reactions, resulting in the formation of five new chiral centers with nearly absolute regio- and stereochemical fidelity. We have used molecular dynamics (MD) simulations to investigate the equilibrium behavior of both COX-1 and COX-2 enzyme isoforms with bound arachidonate. These simulations were compared with reference simulations of arachidonate in solution to explore the effect of enzyme on substrate conformation and positioning in the active site. The simulations suggest that the substrate has greater conformational freedom in the COX-2 active site, consistent with the larger COX-2 active site volume observed in X-ray crystal structures. The simulations reveal different conformational behavior for arachidonate in each subunit over the course of extended equilibrium MD simulations. The simulations also provide detailed information for several protein channels that might be important for oxygen and water transport to or from active sites or for intermediate trafficking between the cyclooxygenase and peroxidase active sites. The detailed comparisons for COX-1 versus COX-2 active site structural fluctuations may also provide useful information for design of new isozyme-selective inhibitors.     

A 175-psec molecular dynamics simulation of camphor-bound cytochrome P-450cam.

The structure and internal motions of cytochrome P-450cam, a monooxygenase heme enzyme with 414 amino acid residues, with camphor bound at the active site have been evaluated on the basis of a 175-psec molecular dynamics simulation carried out at 300 K. All hydrogen atoms were explicitly modeled, and 204 crystallographic waters were included in the simulation. Based on an analysis of the time course of the trajectory versus potential energy, root mean square deviation, radius of gyration, and hydrogen bonding, the simulation was judged to be stable and representative of the average experimental structure. The averaged structural properties of the enzyme were evaluated from the final 135 psec of the simulation. The average atomic displacement from the X-ray structure was 1.39 A for all heavy atoms and 1.17 A for just C-alpha atoms. The average root-mean-square (rms) fluctuations of all heavy atoms and backbone atoms were 0.42 and 0.37 A, respectively. The computed rms fluctuations were in reasonable agreement with the experimentally determined temperature factors. All 13 segments of alpha-helix and 5 segments of beta-sheet were well preserved with the exception of the N-terminal half of helix F which alternated between an alpha-helix and a 310-helix. In addition there were in general only small variations in the relative orientation of adjacent alpha-helices. The rms fluctuations of the backbone dihedral angles in the secondary structure elements were almost uniformly smaller, with the fluctuation in alpha-helices and beta-sheets, 31 and 10% less, respectively, than those in nonsecondary structure regions. The reported crystal structure contains kinks in both helices C and I. In the simulation, both of these regions showed high mobility and large deviations from their starting positions. Since the kink in the I helix is at the oxygen binding site, these motions may have mechanistic implications.     

500-picosecond molecular dynamics in water of the Man alpha 1----2Man alpha glycosidic linkage present in Asn-linked oligomannose-type structures on glycoproteins

Molecular dynamics simulations of the Man alpha 1----2Man alpha glycosidic linkage found in the N-linked glycans of glycoproteins were performed in vacuo and in the presence of water. In the latter case significant dampening of the molecular fluctuations was found when compared to the in vacuo simulation. A 500-ps dynamics simulation in water showed only occasional short-lived deviations from the minimum-energy conformation, more consistent with carbohydrate "breathing" than flexibility. These studies add further evidence that oligosaccharides can maintain "fixed" geometries with relatively long lifetimes and are in agreement with experimental NMR-derived parameters for the same linkage in oligomannose structures.     

The role of short-range Cys171-Cys178 disulfide bond in maintaining cutinase active site integrity: A molecular dynamics simulation

Understanding structural determinants in enzyme active site integrity can provide a good knowledge to design efficient novel catalytic machineries. Fusarium solani pisi cutinase with classic triad Ser-His-Asp is a promising enzyme to scrutinize these structural determinants. We performed two MD simulations: one, with the native structure, and the other with the broken Cys171-Cys178 disulfide bond. This disulfide bond stabilizes a turn in active site on which catalytic Asp175 is located. Functionally important H-bonds and atomic fluctuations in catalytic pocket have been changed. We proposed that this disulfide bond within active site can be considered as an important determinant of cutinase active site structural integrity.     

A molecular dynamics simulation of bacteriophage T4 lysozyme.

An analysis of a 400 ps molecular dynamics simulation of the 164 amino acid enzyme T4 lysozyme is presented. The simulation was carried out with all hydrogen atoms modeled explicitly, the inclusion of all 152 crystallographic waters and at a temperature of 300 K. Temporal analysis of the trajectory versus energy, hydrogen bond stability, r.m.s. deviation from the starting crystal structure and radius of gyration, demonstrates that the simulation was both stable and representative of the average experimental structure. Average structural properties were calculated from the enzyme trajectory and compared with the crystal structure. The mean value of the C alpha displacements of the average simulated structure from the X-ray structure was 1.1 +/- 0.1 A; differences of the backbone phi and psi angles between the average simulated structure and the crystal structure were also examined. Thermal-B factors were calculated from the simulation for heavy and backbone atoms and both were in good agreement with experimental values. Relationships between protein secondary structure elements and internal motions were studied by examining the positional fluctuations of individual helix, sheet and turn structures. The structural integrity in the secondary structure units was preserved throughout the simulation; however, the A helix did show some unusually high atomic fluctuations. The largest backbone atom r.m.s. fluctuations were found in non-secondary structure regions; similar results were observed for r.m.s. fluctuations of non-secondary structure phi and psi angles. In general, the calculated values of r.m.s. fluctuations were quite small for the secondary structure elements. In contrast, surface loops and turns exhibited much larger values, being able to sample larger regions of conformational space. The C alpha difference distance matrix and super-positioning analyses comparing the X-ray structure with the average dynamics structure suggest that a 'hinge-bending' motion occurs between the N- and C-terminal domains.     

Molecular dynamics of HIV-1 protease.

Molecular dynamics simulations have been carried out based on the GROMOS force field on the aspartyl protease (PR) of the human immunodeficiency virus HIV-1. The principal simulation treats the HIV-1 PR dimer and 6990 water molecules in a hexagonal prism cell under periodic boundary conditions and was carried out for a trajectory of 100 psec. Corresponding in vacuo simulations, i.e., treating the isolated protein without solvent, were carried out to study the influence of solvent on the simulation. The results indicate that including waters explicitly in the simulation results in a model considerably closer to the crystal structure than when solvent is neglected. Detailed conformational and helicoidal analysis was performed on the solvated form to determine the exact nature of the dynamical model and the exact points of agreement and disagreement with the crystal structure. The calculated dynamical model was further elucidated by means of studies of the time evolution of the cross-correlation coefficients for atomic displacements of the atoms comprising the protein backbone. The cross-correlation analysis revealed significant aspects of structure originating uniquely in the dynamical motions of the molecule. In particular, an unanticipated through-space, domain-domain correlation was found between the mobile flap region covering the active site and a remote regions of the structure, which collectively act somewhat like a molecular cantilever. The significance of these results is discussed with respect to the inactivation of the protease by site-specific mutagenesis, and in the design of inhibitors.     

Molecular dynamics studies of a DNA-binding protein: 2. An evaluation of implicit and explicit solvent models for the molecular dynamics simulation of the Escherichia coli trp repressor.

Although aqueous simulations with periodic boundary conditions more accurately describe protein dynamics than in vacuo simulations, these are computationally intensive for most proteins. Trp repressor dynamic simulations with a small water shell surrounding the starting model yield protein trajectories that are markedly improved over gas phase, yet computationally efficient. Explicit water in molecular dynamics simulations maintains surface exposure of protein hydrophilic atoms and burial of hydrophobic atoms by opposing the otherwise asymmetric protein-protein forces. This properly orients protein surface side chains, reduces protein fluctuations, and lowers the overall root mean square deviation from the crystal structure. For simulations with crystallographic waters only, a linear or sigmoidal distance-dependent dielectric yields a much better trajectory than does a constant dielectric model. As more water is added to the starting model, the differences between using distance-dependent and constant dielectric models becomes smaller, although the linear distance-dependent dielectric yields an average structure closer to the crystal structure than does a constant dielectric model. Multiplicative constants greater than one, for the linear distance-dependent dielectric simulations, produced trajectories that are progressively worse in describing trp repressor dynamics. Simulations of bovine pancreatic trypsin were used to ensure that the trp repressor results were not protein dependent and to explore the effect of the nonbonded cutoff on the distance-dependent and constant dielectric simulation models. The nonbonded cutoff markedly affected the constant but not distance-dependent dielectric bovine pancreatic trypsin inhibitor simulations. As with trp repressor, the distance-dependent dielectric model with a shell of water surrounding the protein produced a trajectory in better agreement with the crystal structure than a constant dielectric model, and the physical properties of the trajectory average structure, both with and without a nonbonded cutoff, were comparable.     

Structure and energetics of ligand binding to proteins: Escherichia coli dihydrofolate reductase-trimethoprim, a drug-receptor system

A study of the binding of the antibacterial agent trimethoprim to Escherichia coli dihydrofolate reductase was carried out using energy minimization techniques with both a full, all-atom valence force field and a united atom force field. Convergence criteria ensured that no significant structural or energetic changes would occur with further minimization. Root-mean-square (RMS) deviations of both minimized structures with the experimental structure were calculated for selected regions of the protein. In the active site, the all-atom minimized structure fit the experimental structure much better than did the united atom structure. To ascertain what constitutes a good fit, the RMS deviations between crystal structures of the same enzyme either from different species or in different crystal environments were compared. The differences between the active site of the all-atom minimized structure and the experimental structure are similar to differences observed between crystal structures of the same protein. Finally, the energetics of ligand binding were analyzed for the all-atom minimized coordinates. Strain energy induced in the ligand, the corresponding entropy loss due to shifts in harmonic frequencies, and the role of specific residues in ligand binding were examined. Water molecules, even those not in direct contact with the ligand, were found to have significant interaction energies with the ligand. Thus, the inclusion of at least one shell of waters may be vital for accurate simulations of enzyme complexes.     

Vasopressin conformational fluctuations: a molecular dynamics study

Molecular dynamics simulations have been used to study the conformational fluctuations of the oligopeptide hormone vasopressin. Starting coordinates for these simulations were built upon the crystal structure of pressinoic acid, the cyclic ring moiety of vasopressin, recently determined by x-ray diffraction. Coordinates for the additional tripeptide “tail” of vasopressin were selected by arbitrary positioning of this segment using interactive computer graphics. Two such starting configurations were minimized to relax strains, and long dynamics simulations (20 and 40 ps) in vacuo were then conducted following extensive heating and equilibration sequences (36 ps). In these studies, vasopressin was found to undergo few substantial conformational changes at 300 K on the time scale simulated, in contrast to the results of a shorter previous simulation, but comparable structural transitions were observed during the equilibration periods. The pressinoic acid structure was found to be a reasonably stable possible conformation for vasopressin in vacuum on this time scale.     

Carbon monoxide binding to the heme group at the dimeric interface modulates structure and copper accessibility in the Cu,Zn superoxide dismutase from Haemophilus ducreyi: in silico and in vitro evidences

X-ray absorption near-edge structure (XANES) spectroscopy and molecular dynamics (MD) simulations have been jointly applied to the study of the Cu,Zn superoxide dismutase from Haemophilus ducreyi (HdSOD) in interaction with the carbon monoxide molecule. The configurational flexibility of the Fe(II)-heme group, intercalated between the two subunits, has been sampled by MD simulations and included in the XANES data analysis without optimization in the structural parameter space. Our results provide an interpretation of the observed discrepancy in the Fe-heme distances as detected by extended X-ray absorption fine structure (EXAFS) spectroscopy and the classical XANES analysis, in which the structural parameters are optimized in a unique structure. Moreover, binding of the CO molecule to the heme induces a long range effect on the Cu,Zn active site, as evidenced by both MD simulations and in vitro experiments. MD simulation of the CO bound system, in fact, highlighted a structural rearrangement of the protein-protein hydrogen bond network in the region of the Cu,Zn active site, correlated with an increase in water accessibility at short distance from the copper atom. In line, in vitro experiments evidenced an increase of copper accessibility to a chelating agent when the CO molecule binds to the heme group, as compared to a heme deprived HdSOD. Altogether, our results support the hypothesis that the HdSOD is a heme-sensor protein, in which binding to small gaseous molecules modulates the enzyme superoxide activity as an adaptive response to the bacterial environment.     

Molecular dynamics simulations of lignin peroxidase in solution

The dynamical and structural properties of lignin peroxidase and its Trp171Ala mutant have been investigated in aqueous solution using molecular dynamics (MD) simulations. In both cases, the enzyme retained its overall backbone structure and all its noncovalent interactions in the course of the MD simulations. Very interestingly, the analysis of the MD trajectories showed the presence of large fluctuations in correspondence of the residues forming the heme access channel; these movements enlarge the opening and facilitate the access of substrates to the enzyme active site. Moreover, steered molecular dynamics docking simulations have shown that lignin peroxidase natural substrate (veratryl alcohol) can easily approach the heme edge through the access channel.     

Computational design of novel fullerene analogues as potential HIV-1 PR inhibitors: Analysis of the binding interactions between fullerene inhibitors and HIV-1 PR residues using 3D QSAR, molecular docking and molecular dynamics simulations

A series of experimentally reported as well as computationally designed monoadducts and bisadducts of [60]fullerene analogues have been used in order to analyze the binding interactions between fullerene based inhibitors and HIV-1 PR employing docking studies. MD simulations of ligand-free and the inhibitor bound HIV-1 PR systems complemented the above studies and provided proper input structure of HIV-1 PR in docking simulations. The obtained results revealed a different orientation of the beta-hairpin flaps at these two systems. In inhibitor bound system, the flaps of the enzyme are pulled in toward the bottom of the active site (the closed form) while, in ligand-free system flaps shifted away from the dual Asp25 catalytic site and this system adopts a semi-open form. The structural analysis of these systems at catalytic and flexible flap regions of the HIV-1 PR through the simulation, assisted in understanding the structural preferences of these regions, as well as, the adopted orientations of fullerene derivatives within the active site of the enzyme. Five different combinations of steroelectronic fields of 3D QSAR/CoMSIA models were obtained from the set of biologically evaluated and computationally designed fullerene derivatives (training set=43, test set=6) in order to predict novel compounds with improved inhibition effect. The best 3D QSAR/CoMSIA model yielded a cross validated r(2) value of 0.739 and a non-cross validated r(2) value of 0.993. The derived model indicated the importance of steric (42.6%), electrostatic (12.7%), H-bond donor (16.7%) and H-bond acceptor (28.0%) contributions. The derived contour plots together with de novo drug design were then used as pilot models for proposing the novel analogues with enhanced binding affinities. Such structures may trigger the interest of medicinal chemists for novel HIV-1 PR inhibitors possessing higher bioactivity.     

Molecular dynamic simulations of the N-terminal receiver domain of NtrC reveal intrinsic conformational flexibility in the inactive state

The N-terminal receiver domain of NtrC is the molecular switch in the two-component signal transduction. It is the first protein where structures of both the active (phosphyroylated) and inactive (unphosphyroylated) states are determined experimentally. Phosphorylation of the NtrC at the active site induces large structural change. NMR experiments suggested that the wild type unphosphorylated NtrC adopts both the active and the inactive conformations and the phosphorylation stabilizes the active conformations. We applied free (unconstrained) molecular dynamic (MD) simulation to examine the intrinsic flexibilities and stabilities of the NtrC receiver domain in both the active and inactive conformations. Molecular dynamic simulations showed that the inactive state of NtrC receiver domain is more flexible than the active state. There were large movements in helix 4 and loop beta3-alpha3 which coincide with major structural differences between the inactive and active states. We observed large root-mean-square deviations from the initial starting structure and the large root-mean-square fluctuations during MD simulation for the inactive state. We then investigated the activation pathway with Targeted MD simulation. We show that the intrinsic flexibility in the loop beta3-alpha3 plays an important role in triggering the conformational change. Phosphorylation at the active site may serve to stabilize the conformational change. These results together suggest that the unphosphorylated NtrC receiver domain could be involved in a conformational equilibrium between two different states.     

Molecular dynamics simulations of human glutathione transferase P1-1: conformational fluctuations of the apo-structure.

We have investigated by molecular dynamics simulations the conformational fluctuations of the monomer of human apo-glutathione transferase P1-1. After attainment of steady-state dynamics, the structural fluctuations involve mainly the protein segments that participate also in the holo-apo transition discussed in the accompanying article (Stella et al., 1999:37:1-9.). The most mobile region is the C-terminal segment of helix 2. In contrast, helices 1, 6, 7, and 8 constitute a relatively rigid protein core. An "essential dynamics" analysis of the simulation shows that the largest fluctuations involve specific regions of glutathione transferases. In such regions, atomic motions are correlated. Motions of helix 2 are accounted for by the second most prominent principal component, which reveals a fluctuation between two distinct conformations. The residues that constitute the H-site undergo a breathing motion, possibly relevant during the binding of hydrophobic cosubstrates. Based on our simulation, several experimental findings can be rationalized, including the viscosity-dependent reactivity of Cys 47 and Cys 101 as well as the selective proteolysis of the peptide bond between Lys 44 and Ala 45. We have also modeled the structural changes that lead to the formation of an intrachain disulfide bridge between cysteines 47 and 101 and to the inactivation of the enzyme. The resulting structure maintains essentially the native fold except for helix 2, which closes the G-site. Proteins 1999;37:10-19.     

Molecular dynamics simulations of glycosyltransferase LgtC

Molecular dynamics simulations have been performed on fully solvated alpha-(1-->4)-galactosyltransferase LgtC from Neisseria meningitidis with and without the donor substrate UDP-Gal and in the presence of the manganese ion. The analysis of the trajectories revealed a limited movement in the loop X (residues 75-80) and a larger conformational change in the loop Y (residues 246-251) in the simulation, when UDP-Gal was not present. In this case, the loops X and Y open by almost 10A, exposing the active site to the solvent. The 'hinge region' responsible for the opening is composed of residues 246-247. We have also analyzed the behavior of the manganese ion in the simulations. The coordination number is 6 when UDP-Gal is present and it increases to 7 when it is absent. In the latter case, three water molecules become coordinated to the ion. In both cases, the coordination is very stable implying that the manganese ion is tightly bound in the active site of the enzyme even if UDP-Gal is not present. Further analysis of the structural water molecules location confirmed that the mobility of water molecules in the active site and the accessibility of this site for solvent are higher in the absence of the substrate.     

Molecular dynamics simulations of Hydrogenobacter thermophilus cytochrome c552: comparisons of the wild-type protein, a b-type variant, and the apo state

Molecular dynamic simulations have been performed for wild-type Hydrogenobacter thermophilus cytochrome c(552), a b-type variant of the protein, and the apo state with the heme prosthetic group removed. In the b-type variant, Cys 10 and Cys 13 were mutated to alanine residues, and so the heme group was no longer covalently bound to the protein. Two 8-ns simulations have been performed for each system at 298 and 360 K. The simulations of the wild-type protein at 298 K show a very close agreement with experimental NMR data. A fluxional process involving the side chain of Met 59, which coordinates to the heme iron, is observed in accord with proposals from NMR studies. Overall, the structure and dynamical behavior of the protein during the simulations of the b-type variant is closely similar to that of the wild-type protein. However, side chains in the heme-binding site show larger fluctuations in the b-type variant simulation at 360 K. In addition, structural changes are seen for a number of residues close to the heme group, particularly Gly 22 and Ser 51. The simulations of the apo state show significant conformational changes for residues 50-59. These residues form a loop region, which packs over the heme group in the wild-type protein and hydrogen bonds to the heme propionate groups. In the absence of heme, in the apo state simulations, these residues form short but persistent regions of beta-sheet secondary structure. These could provide nucleation sites for the conversion to amyloid fibrils.