|Home | About | Journals | Submit | Contact Us | Français|
Here, we report the 1.53-Å crystal structure of the enzyme 7-cyano-7-deazaguanine reductase (QueF) from Vibrio cholerae, which is responsible for the complete reduction of a nitrile (CN) bond to a primary amine (H2C–NH2). At present, this is the only example of a biological pathway that includes reduction of a nitrile bond, establishing QueF as particularly noteworthy. The structure of the QueF monomer resembles two connected ferrodoxin-like domains that assemble into dimers. Ligands identified in the crystal structure suggest the likely binding conformation of the native substrates NADPH and 7-cyano-7-deazaguanine. We also report on a series of numerical simulations that have shed light on the mechanism by which this enzyme affects the transfer of four protons (and electrons) to the 7-cyano-7-deazaguanine substrate. In particular, the simulations suggest that the initial step of the catalytic process is the formation of a covalent adduct with the residue Cys194, in agreement with previous studies. The crystal structure also suggests that two conserved residues (His233 and Asp102) play an important role in the delivery of a fourth proton to the substrate.
Transfer RNAs (tRNAs) undergo significant post-transcriptional processing, resulting in a variety of structural changes1 to the canonical nucleosides that provide structural stabilization and improved translational fidelity.2,3 One of the heavily modified nucleosides is queuosine,4 a guanosine derivative synthesized by bacteria, which is found in the third (wobble) position of the anticodon in tRNAAsn, tRNAAsp, tRNAHis, and tRNATyr, where it presumably improves the fidelity of translation. Queuosine is utilized but not synthesized in eukaryotic cells, making the queuosine biosynthetic pathway the subject of continued interest: disruption of the pathway in Shigella flexneri results in loss of pathogenicity,5 for example. Development of a better understanding of the biochemistry involved could well lead to new antibiotic agents.
A key component of the queuosine synthetic pathway shown in Fig. 1 is the enzyme 7-cyano-7-deazaguanine reductase (QueF), which converts the intermediate 7-cyano-7-deazaguanine (preQ0) to 7-aminomethyl-7-deazaguanine (preQ1). There are two related sequence subfamilies of QueF enzymes: YqcD (two-domain proteins) and YkvM (single-domain proteins).6 The YqcD subfamily, to which the enzyme from Vibrio cholerae belongs, is highly conserved across species. BLAST analysis of the protein sequence identified several hundred homologs, of which a representative set of proteins from both subfamilies is aligned in Fig. S1. These proteins (YqcD and YkvM) show sequence similarity to the family of GTP cyclohydrolases I (PF01227) but recent biochemical studies of the orthologous proteins from the YqcD subfamily from Escherichia coli (with 63% sequence identity) and YkvM subfamily from Bacillus subtilis (33% sequence identity) demonstrated that QueF proteins are NADPH-dependent 7-cyano-7-deazaguanine oxidoreductases.7,8 The YqcD subfamily of QueF enzymes is composed of two conserved domains: the N-terminal sequence is categorized as a member of the COG2904 superfamily that is composed of proteins of unknown function and the C-terminal sequence, which encompasses the catalytic residues, is classified as a tunneling fold (T-fold) superfamily. The T-fold is a common structural motif (composed of four β-strands and two α-helices) that oligomerizes to form a β2nαn-type β-barrel (somewhat resembling a TIM barrel unit of βnαn) and is present in a number of different enzymes acting on flat substrates of the purine or pterin families.9 The amino acid sequence of the YkvM subfamily maps into the T-fold C-terminal region of the YqcD subfamily. A recent study elucidates details of the enzymatic synthesis of the preQ0 intermediate; the initial step from GTP to preQ0, as illustrated in Fig. 1, actually requires four separate enzymatic steps.10 The enzymes involved, QueC, QueD, and QueE, are part of the same operon as QueF.6
The conversion of preQ0 to preQ1 catalyzed by QueF requires two molecules of NADPH for each molecule of preQ1 synthesized and a potential chemical pathway has been proposed (Fig. 2).7 As part of the Protein Structure Initiative effort, we have determined a crystal structure of the enzyme QueF from V. cholerae via X-ray crystallography, which provides a number of insights into the structure of the enzyme and catalytic mechanism. Additionally, we have conducted a number of numerical simulations of the enzyme, utilizing molecular dynamics and hybrid, quantum/classical (QM/MM) calculations that further shed light on the catalytic mechanism.
The structure of QueF from V. cholerae was determined at 1.53 Å, and its monomer and dimer architecture and active sites are shown in Fig. 3. The protein shares 63% sequence identity with the experimentally characterized QueF protein from E. coli, and therefore, it is likely to possess the same enzymatic activity and serves as a good model for structural studies of the family. The protein crystallizes as a tetramer, a dimer of dimers (Fig. S2). Solution data (Fig. S3) suggest that the protein is a dimer and it may exist in equilibrium with monomer. Size-exclusion chromatography showed two peaks and a lower molecular weight shoulder. The two peaks correspond to molecular masses of 67.9 kDa (dimer) and 43.7 kDa (1.33×monomer) and a shoulder of ~33 kDa (monomer), suggesting dimer/monomer equilibrium in solution.
Each monomer is composed of a seven-stranded, antiparallel β-sheet, eight α-helices, and several loops. Seven helices are on the concave side of the β-sheet and one 1-turn helix is on the convex side. The monomer seems to be formed by duplication of a simpler ferredoxin-like structural module. One unit is composed of a three-stranded β-sheet and two α-helices (residues 57–136, part of the COG2904 superfamily sequence) and the other is composed of a four-stranded β-sheet with two α-helices (residues 176–271, similar to a T-fold) (Fig. 3).
Therefore, the QueF monomer is made up of two ferredoxin-like domains aligned together with their β-sheets that have additional embellishments. An example of such embellishment is a 30-residue loop (residues 138–168) that connects two ferredoxin-like sub-domains and crosses over to the convex side of the β-sheet. These loops contribute to the dimer interface as well as to the formation of the tetramer in the crystal (Fig. S2). Each loop makes contact with the loops from the other three monomers. QueF was co-crystallized with GTP, but only guanine, phosphate, and pyrophosphate molecules were observed in the crystal structure. These are depicted as space-filling representations in Fig. 3. One guanine molecule is bound to each of the four monomers in the crystal asymmetric unit (for clarity, only two are shown in Fig. 3), in the cavity formed near the dimer interface on the concave side of each β-sheet between two sub-domains. The two pyrophosphates bind on the interface between two monomers in the dimer, at equal distance from two guanines and are presumably positioned close to where the diphosphate moiety of NADPH binds. This arrangement suggests that all four sites in a tetramer can be occupied by a substrate, but only one molecule of NADPH cofactor can bind to a dimer (two orientations are possible, see below) and can serve only one of the two sites occupied by substrate. Therefore, as is consistent with QueF solution studies and previous reports by Van Lanen et al.,6 the dimer is a minimal biological unit and QueF should exhibit half-site reactivity.
Figure 4 displays one of the guanine molecules in the active site where it mimics the binding of a substrate: preQ0. The guanine is bound into a groove between two ferredoxin-like modules and is sandwiched between the two N-termini helices α2 and α5. It is precisely oriented by a series of hydrogen bonds and a ring-stacking interaction with a phenylalanine residue (Phe232). The guanine N1 and N2 nitrogen atoms are coordinated by the carboxyl moiety of Glu234, which is strictly conserved (Fig. S1). From the chain A monomer, we observe the distances to be d(N1–Oε1)=2.81 Å and d(N2–Oε2)=2.80 Å, with similar distances (± 0.03 Å) for other monomers. The backbone carbonyl oxygen atom of Ile93, which is sometimes substituted by valine, also coordinates the guanine N2 nitrogen atom, with d(N2–O)=2.74 Å. The guanine N3 nitrogen atom is coordinated by the amide nitrogen atom of the Ser95 residue d(N3–N) = 2.85 Å and the guanine N9 nitrogen atom is coordinated by the carboxylate moiety of Glu94 d(N9–Oε2)=2.74 Å; both residues are strictly conserved (Fig. S1). The hydroxyl moiety of Ser95, found in two conformations in the crystal structure, also interacts with the N9 nitrogen atom of guanine: d(N9–Oγ)=3.03 Å. Finally, the guanine O6 oxygen atom is coordinated by the backbone amide nitrogen of the strictly conserved His233 residue with d(O6–N)=2.77 Å. We observe the His233 residue in close proximity to the strictly conserved Asp201. The hairpin loop near the active site containing the catalytic6 Cys194 is disordered in all four monomers.
We note that, as depicted in Fig. 4, the ligand observed in our X-ray structure is guanine and not GTP: there is no electron density corresponding to the ribophosphate moiety of GTP, suggesting that QueF retains some residual nucleosidase activity. Indeed, the QueF protein shows sequence similarity to GTP cyclohydrolases and was initially categorized as such.6 Van Lanen et al. reported that the QueF proteins from B. subtilis and E. coli do not have cyclohydrolase activity, which includes hydrolysis of the N9–C1′ bond, but are oxidoreductases.6 A Dali search11 of the Protein Data Bank (PDB) for proteins structurally similar to QueF finds several GTP cyclohydrolase I proteins with Z scores above 3. The closest is GTP cyclohydrolase I from E. coli (PDB ID: 1A9C; Z=6.7, RMSD of 11.2 Å between Cα atoms by SSM). The structure of the dimer of the cyclohydrolase with bound GTP is depicted in Fig. S3. This enzyme monomer also has one ferredoxin-like domain composed of a four-stranded β-sheet with two α-helices and assembles to a pentamer. The GTP binds in a groove between two ferredoxin-like modules coming from two adjacent subunits (Fig. S4). The guanine moiety of the GTP rests in a pocket between two α-helices in a mode reminiscent of guanine binding to QueF, but, unlike what we observed in the QueF structure, each of the α-helices is provided by a separate monomer. Moreover, the cyclohydrolase active-site pocket has a very different amino acid composition; for example, two of the three cyclohydrolase His residues are not present in QueF. The catalytic mechanism of N-glycosyl bond cleavage by the GTP cyclohydrolase I enzymes utilizes a Zn ion to facilitate the reaction.12 We find no such evidence for ordered metal ion binding in our structure.
Nevertheless, the crystal structure of QueF suggests potential activity of the N-glycosylase reaction. From Fig. 4, the conserved residue His233 is in an appropriate position to donate a proton to the N7 nitrogen of the guanine moiety of GTP. Protonation of the nucleotide will result in cleavage of the N9–C1′ bond.13 This reaction can potentially be facilitated by the conserved residue Arg262; this residue is not oriented appropriately in the crystal structure but molecular dynamics simulations indicate that motion of the Arg262 side chain can lead to catalytically competent conformations of these key residues. Both His233 and Arg262 are contributed by the C-terminal catalytic domain. This work is still ongoing and will be reported in a subsequent communication. Recently, we have solved the structure of QueF with ATP (data not shown), and in the structure, we observe adenine and pyrophosphate molecules, similar to the structure reported here. Therefore, it is possible that the apparent QueF nucleosidase activity may be the residual reaction carried out in this unique QueF catalytic site. This activity is not essential for the primary function of preQ0 reduction, and because of the unique nature of the product and its benefits to bacteria, it is being tolerated. Nevertheless, we cannot exclude, at present, other possible explanations for the N-glycosylase activity, such as (i) the presence of a contaminating nucleosidase activity in the QueF preparation or (ii) the presence of guanine and pyrophosphate in the commercial GTP preparation. We note that the substrate NADPH also has an N-glycosyl bond, which potentially could impact its binding and integrity during reaction (see discussion below).
It has been proposed that the initial step of the reduction of PreQ0 involves formation of a covalent adduct through the nucleophilic attack of a conserved cysteine residue (Cys194 in V. cholerae) on the cyano group of the substrate (Fig. 2).6 This covalent bond can hold the substrate in place while two NADPH molecules are recruited to the active site to donate hydrides to the substrate. The catalytic Cys194 residue resides on the hairpin loop between strands β7 and β8 of the protein (Fig. S1), which is disordered in all four monomers in the crystal structure. To investigate the formation of the covalent adduct, we have conducted a number of numerical simulations.
Simulation models were constructed from the atomic coordinates of the crystal structure and solvated in a large box of water. The box dimensions were chosen to provide a diffuse limit, in which protein atoms do not interact directly with protein atoms in the periodic images. Sodium and chlorine ions were added to the simulation model to provide overall charge neutrality and an ionic strength of 0.15 mM. Molecular dynamics simulations were conducted with the CHARMM 27 force field, using the program NAMD. The tetramer, as observed in the crystal structure, was used in the simulations, providing four nominally equivalent sites and two biological units. Simulations included four preQ0 molecules in addition to the protein to investigate formation of the covalent adduct. Missing residues were built into the structure with the PSFGEN module of VMD. Not surprisingly, and consistent with the crystal structure, the simulations demonstrated significant mobility of the hairpin loop. This is illustrated in Fig. S5, in which a single time step from a 2-ns simulation is depicted. The resulting four models of monomers have been stacked by aligning their protein backbones away from the loop segment.
We identified one conformation from these simulations in which the Cys194 was oriented in an appropriate attack conformation as the starting point for a series of QM/MM calculations using the program NWChem. Molecular mechanics atoms utilized the AMBER force field, and quantum atoms were treated via the density function method B3LYP, using an Ahlrichs pVDZ basis set. A reactant state was defined by optimizing the structure from the molecular dynamics trajectory, keeping atoms more than 12 Å from the C10 carbon atom of the substrate fixed. A product state was defined by optimizing the structure with constraints on the sulfur–carbon (Sγ–C10) distance to form the bond and then relaxing the constraints and re-optimizing. The product-state conformation is shown in Fig. 5. A minimum-energy pathway between reactant and product states was defined by the nudged elastic band (NEB) method, using 10 replicas (beads) of the system, resulting in the pathway shown in Fig. 6. We observe a large barrier between reactant and product states that we believe is due to a relatively poor choice for the reactant state. In this conformation, there are no other residues in proximity to the catalytic cysteine to aid the reaction. It is possible that additional dynamics simulations will identify a better conformation of the mobile loop and reduce the observed barrier. We observe, however, that the relative barrier to the serine-catalyzed reaction is significantly higher than the cysteine reaction (~80 kJ/mol). This is consistent with the observation in B. subtilis that the equivalent C55S mutation resulted in loss of catalytic function.7
In the reduction of preQ0, two hydrides are transferred to the substrate sequentially from two molecules of NADPH. We have investigated binding of the NADPH molecules via molecular dynamics simulations. From Fig. 3, there are two equivalent sites; each is occupied by a guanine molecule and potentially each could form a covalent adduct. We have conducted a series of simulations with preQ0 and adenine molecules in the active sites, initially stacking the adenine molecules on the side opposite the Phe232 residues. The results of these simulations indicate that there is insufficient space in the active sites to accommodate the adenine moiety of NADPH when preQ0 also occupies the site.
It is possible that the NADPH can displace one of the preQ0 molecules from its binding site and that the adenine moiety will occupy the same pocket as preQ0. This is illustrated in Fig. 7 where we plot the relative orientation of adenine in the binding pocket. This conformation is taken from one frame of a dynamics simulation and is representative of the conformations observed. Note that the strictly conserved Lys96 residue is now coordinating the phosphate group from the NADPH and the Glu94 residue has adopted an alternative conformation. The N6 nitrogen atom of the adenine is coordinated by the carboxylate of Glu234, in the same fashion as the N2 nitrogen atom of preQ0. Indeed, the recently solved structure of QueF with ATP confirms that adenine can bind in the active site and binding of adenine is consistent with numerical simulations (data not shown).
A model of the overall binding of NADPH to the enzyme is depicted in Fig. 8, where we use a surface representation of the protein to emphasize the binding cleft. The nicotinamide moiety of NADPH is positioned to deliver a hydride to the C10 carbon of the substrate. In this simulation, the preQ0 was not modeled as covalently bound to the Cys194 side chain. The diphosphate moiety of the NADPH occupies a position that is comparable to the position of the pyrophosphate shown in Fig. 3. The orientation of the nicotinamide reaffirms our earlier observation that the adenine moiety will not fit into the active site in a conformation in which it is stacked against preQ0. The nicotinamide fits into the narrow cleft in a conformation where the nicotinamide ring is transverse to the preQ0 ring plane.
We propose that the fourth proton delivered to the substrate is provided by the strictly conserved His233 residue. This residue is found in close proximity to the strictly conserved Asp201 residue in the crystal structure and throughout our simulations, as is indicated in Figs. 3–5. Formation of the covalent adduct, as shown in Fig. 5, modifies the geometry of the C10 carbon atom of the substrate from linear to planar, pushing the N11 nitrogen atom towards the His233. The transfer of the final proton to the N11 nitrogen atom will be facilitated by the carboxyl moiety of Asp201.
This final step completes the process of the reaction mechanism at work in QueF (illustrated in Fig. 2) and accounts for the four protons and electrons delivered to the substrate. The initial proton is provided by Cys194 as a consequence of the formation of the covalent adduct. Two hydride transfers from NADPH account for the four electrons and two additional protons. The final proton is obtained from His233. Use of the crystal structure and numerical simulations to fill in missing details has enabled us to provide a more complete view of the catalytic process at work in this enzyme system.
The proposed mechanism suggests that the preQ0 binds first and forms a covalent adduct with the enzyme (Fig. 2). The Cys194 attacks the C10 atom of the preQ0 cyano group, donating one proton to the N11 nitrogen atom of the substrate (Fig. 5). NADPH binds subsequently and, if necessary, displaces one of the preQ0 molecules and positions itself for hydride transfer. This commits the reaction to one substrate-binding site. In the next step, a hydride is transferred from NADPH to the C10 carbon atom of the substrate and a third proton is provided by His233. During all of these steps, the substrate remains covalently attached to the enzyme, assuring productive completion of the reaction. Finally, a second molecule of NADPH provides a second hydride to the C10 carbon atom, breaking the adduct and allowing the product (preQ1) to leave the active site. Additional simulations are planned to study the detailed energetics along the proposed reaction pathway.
The high-resolution structure of the QueF protein reported here provides a structural framework for mechanistic studies and supports previous biochemical characterizations of the enzyme.7,8 Consistent with previous reports and this work, the QueF dimer is a minimal catalytic unit and the enzyme should exhibit half-site reactivity. QueF is unique and facilitates the transfer of four electrons and four protons to its PreQ0 substrate using NADPH as a cofactor, reducing the nitrile bond to a primary amine. Utilizing the structure as a starting state, we conducted a series of numerical simulations to provide structural information about the catalytic cysteine residue that is disordered and is not observed in the electron density and propose the catalytic role of several residues in the active site. Combining these calculations with the experimental X-ray structure that contains bound guanine and pyrophosphate has enabled us to begin to characterize each of the four steps of the reaction (two proton and two hydride transfers) and the roles of the active-site residues in binding and catalysis. We also found unexpected nucleosidase activity for the QueF enzyme and propose a potential mechanism, although further experimental work will be required to validate that model.
The open reading frame of the queF gene from V. cholerae O1 biovar El Tor str. N16961 (GenBank accession no. AAF94064.1) was amplified from genomic DNA with KOD DNA polymerase using conditions and reagents provided by Novagen (Madison, WI). The gene was cloned into pMCSG7 vector using a modified LIC protocol. This process generated an expression clone producing a fusion protein with an N-terminal His6 tag and a tobacco etch virus protease recognition site. A selenomethionine derivative of the expressed protein was prepared as described by Walsh et al.14 and purified using standard procedures on an AKTAxpress automated purification system (GE/Amersham) as described by Kim et al.15 The concentration of the purified protein was determined utilizing an ND-1000 Spectrophotometer System (NanoDrop Technologies). The fusion tag was then removed by adding recombinant tobacco etch virus protease at a ratio of approximately 1:75 (mg) and incubated for 48 h at 4 °C. The cleaved protein was then separated on a nickel-nitrilotriacetic acid agarose nickel charged resin column (Qiagen Inc.). The purified protein solution was dialyzed in a crystallization buffer (20 mM Hepes, pH 8.0, 250 mM NaCl, and 2 mM DTT) for 24 h and concentrated using a Centricon Plus-20 concentrator with a nominal molecular weight limit of 5000 (Millipore Corp.).
Size-exclusion chromatography was performed on a Superdex-200 10/300GL column using AKTAExplorer (GE Biosciences). The column was pre-equilibrated with crystallization buffer (20 mM Hepes, pH 8.0, 250 mM NaCl, and 2 mM DTT) and calibrated with premixed protein standards, including ovalbumin (43 kDa), con-albumin (75 kDa), aldolase (158 kDa), and Blue Dextran (2000 kDa). A 100-μl protein sample at 29 mg/ml was injected into the column. The chromatography was carried out at room temperature at a flow rate of 0.3 ml/min. The calibration curve of Kav versus log molecular weight was prepared using the equation Kav =(Ve − Vo)/(Vt − Vo,), where Ve is the elution volume for the protein, Vo is the column void volume, and Vt is the total bed volume. Size-exclusion chromatography indicates a split peak of protein monomer and dimer (Fig. S3).
The protein was crystallized using hanging drop vapor diffusion at 289 K in a CrystalQuick® standard profile—LBR round bottom plate (Greiner Bio-One North America, Inc.). A 400-nl droplet of protein (199 mg/ml) was mixed with a 400-nl droplet of crystallization reagent and allowed to equilibrate over 135 μl of crystallization reagent. Nanopipetting was performed using the Mosquito® nanoliter liquid handling system (TTP LabTech). The finished plate was then incubated at 16 °C within a RoboIncubator® automated plate storage system (Rigaku). Automated crystal visualization was utilized in locating several crystals [Minstrel III® (Rigaku)]. These crystals were cryoprotected and flash frozen in liquid nitrogen. The protein crystallized in space group P1 with cell dimensions of a = 71.52 Å, b = 72.58 Å, c = 71.51 Å, α = 119.25°, β=110.18°, and γ = 99.58°. The crystallization buffer included 0.04 M sodium dihydrogen phosphate, 0.96 M di-potassium hydrogen phosphate, and 10 mM GTP.
Diffraction data were collected at 100 K at the 19BM beamline of the Structural Biology Center at the Advanced Photon Source, Argonne National Laboratory. The single-wavelength anomalous dispersion (SAD) data at 0.9793 Å (near selenium absorption white line, 12.6605 keV) up to 1.53 Å were collected from a single (0.1 mm×0.04 mm×0.05 mm) selenomethionine-labeled protein crystal. The crystal was exposed for 3 s per 1.0° rotation of ω with a crystal-to-detector distance of 280 mm. The data were recorded on a CCD detector scanning a full 360° on ω.
The structure was determined by SAD phasing using HKL-3000,16 SHELX,17 MLPHARE,18 and SOLVE/RESOLVE19 and refined to 1.53 Å using REFMAC20 in CCP4.19 The initial model was completed by using ARP/wARP21 and manual tailoring using Coot.22 The final R was 14.3% with an Rfree of 18.3% with all data (Table 1).
The stereochemistry of the structure was checked with PROCHECK23 and a Ramachandran plot. The main-chain torsion angles for all residues are in allowed regions.
Molecular dynamics simulations were performed with the code NAMD,24 developed by the Theoretical and Computational Biophysics Group in the Beckman Institute for Advanced Science and Technology at the University of Illinois at Urbana-Champaign. The CHARMM 27 force field25 parameters were used. Analysis of dynamics trajectories was conducted with the program VMD.26 Initial coordinates for protein atoms were obtained from the crystal structures. Missing atom coordinates (hydrogens) were defined with the PSFGEN module of VMD, which was also used to solvate the protein and neutralize the total charge of the simulation model. The solvation box was extended for 10 Å beyond the protein. With a 12-Å cutoff for electrostatic interactions, the box dimensions ensured a “diffuse” simulation model, in which no protein atoms interacted directly with protein atoms in the periodic images. Electrostatic interactions were computed using a smooth particle-mesh Ewald method,27 with a grid size of approximately 1 Å spacing. Simulations began with a small amount of minimization (~1000 steps) and then utilized typically 100,000 steps of NVT dynamics and 100,000 steps of NPT dynamics run with 1-fs time steps to equilibrate the system. Constant temperature was maintained by a Langevin method,28 and constant pressure conditions were enforced through a modified version of the Langevin piston29 and Hoover30,31 methods. Typical production runs were of 1-ns duration and were conducted using NPT dynamics with 2-fs time steps, recording coordinate information at 1-ps intervals.
Simulations were conducted for the tetrameric conformation of the protein, as observed in the crystal structure. There were approximately 100,000 atoms in the simulations, in a box of dimension 84 Å×97 Å×132 Å.
To study energetics along the proposed reaction pathway, we performed QM/MM calculations with the program NWChem.32 We utilized a NEB method33 implemented recently. The NEB method requires the definition of reactant and product states. Initial coordinates for reactant states were taken from snapshots of the dynamics trajectories in which the attacking water molecule was positioned in what we would describe as a near attack conformation.34 The quantum partitions for all simulations included the side chain of residue 194 and the 10 preQ0 substrate. Atoms beyond 15 Å from the target C carbon atom of the substrate were frozen in place and only atoms within that spherical region were allowed to move. The reactant state was defined by a process in which the original model was optimized using the density functional method B3LYP35 and a 3–21 g* basis set to speed convergence to a coarse estimate of the state geometry. Subsequent optimizations and dynamics used the Ahlrichs pVDZ basis set.36 An initial estimate of the product-state geometry was obtained by constraining the Sg–C10 distance between the attacking cysteine and the substrate to 1.8 Å and the Hg–N11 distance to 1 Å and by repeating the optimizations at the B3LYP/Ahlrichs pVDZ level of theory. All atom constraints were then relaxed and the model geometry was re-optimized to define the product state. The NEB simulations were conducted using 10 replicas (beads) to define the reaction pathway.
The authors would like to thank the members of the Midwest Center for Structural Genomics and Structural Biology Center for their support and Marat Valiev at Pacific Northwest National Laboratory for his help with NWChem. This research has been funded in part by a grant from the National Institutes of Health (GM074942) and by the U.S. Department of Energy, Office of Biological and Environmental Research, under Contract DE-AC02-06CH11357. One of us (M.A.C.) has also received support through the National Science Foundation’s FaST program (HRD-0703584), administered by the Department of Educational Programs at Argonne National Laboratory. The authors acknowledge the Texas Advanced Computing Center at The University of Texas at Austin for providing HPC resources that have contributed to the research results reported within this article†.