Crystal Structure of Helicobacter pylori Formamidase AmiF Reveals a Cysteine-Glutamate-Lysine Catalytic Triad*

Helicobacter pylori AmiF formamidase that hydrolyzes formamide to produce formic acid and ammonia belongs to a member of the nitrilase superfamily. The crystal structure of AmiF was solved to 1.75Å resolution using single-wavelength anomalous dispersion methods. The structure consists of a homohexamer related by 3-fold symmetry in which each subunit has an α-β-β-α four-layer architecture characteristic of the nitrilase superfamily. One exterior α layer faces the solvent, whereas the other one associates with that of the neighbor subunit, forming a tight α-β-β-α-α-β-β-α dimer. The apo and liganded crystal structures of an inactive mutant C166S were also determined to 2.50 and 2.30Å, respectively. These structures reveal a small formamide-binding pocket that includes Cys166, Glu60, and Lys133 catalytic residues, in which Cys166 acts as a nucleophile. Analysis of the liganded AmiF and N-carbamoyl d-amino acid amidohydrolase binding pockets reveals a common Cys-Glu-Lys triad, another conserved glutamate, and different subsets of ligand-binding residues. Molecular dynamic simulations show that the conserved triad has minimal fluctuations, catalyzing the hydrolysis of a specific nitrile or amide in the nitrilase superfamily efficiently.

Utilization of nitrogen from nonpeptide carbon-nitrogen compounds usually involves nitrilase or nitrile hydratase coupled with amidase to produce ammonia for further assimilation or transfer (1). Based on multiple sequence analysis, nitrilases, cyanide hydratases, and aliphatic amidases, along with ␤-alanine synthases that reduce organic nitrogen compounds are found to possess several conserved sequence features, including an invariant cysteine active residue (2).
The first determined structure of the nitrilase (Nit) domain is from worm NitFhit that shows a tetramer, and each Nit domain has a four-layer ␣-␤-␤-␣ sandwich fold (3). Additionally, a novel Cys-Glu-Lys (CEK) triad that contains the highly conserved cysteine is seen in a solvent-accessible pocket. Crystal structures of N-carbamoyl D-amino acid amidohydrolase reveal a homologous fold, in which the CEK triad is also found (4,5). Structural and site-directed mutagenesis results suggest the roles of the CEK residues; the active cysteine acts as the nucleophile, glutamate mediates the proton transfer, and lysine stabilizes a tetrahedral transition state (6,7). Such a novel catalytic triad had not been observed in any other hydrolytic enzymes.
In 2001, Pace and Brenner (8) defined these proteins having the strictly conserved CEK site as the nitrilase superfamily. On the basis of sequence similarity and the presence of additional domains, 13 branches can be classified, nine of which have known or deduced specificity for specific nitrile or amide hydrolysis or amide condensation reactions, including nitrilase, aliphatic amidase, N-terminal amidase, biotinidase, ␤-ureidopropionase, carbamoylase, glutaminase domain of glutaminedependent NAD synthetase, and apolipoprotein N-acyltransferase. Subsequently, structures from two members of the nitrilase superfamily (a yeast hypothetical protein with sequence homologous to CN hydralase from yeast (9) and a hypothetical protein from Pyrococcus horikoshii (10)) show the characteristic ␣-␤-␤-␣ sandwich architecture.
Given the functional diversity and the wide evolutionary distribution of the nitrilase superfamily members, we sought to investigate the structure-function relationship of Helicobacter pylori formamidase AmiF that shows restricted substrate specificity. H. pylori is a human gastric pathogen that persistently colonizes the mucus layer overlaying the epithelium of the stomach. AmiF and the other paralogue AmiE, along with urease, play a major role in producing ammonia to protect against gastric acidity or as a nitrogen source or a cytotoxic molecule, enabling H. pylori to adapt into such an exclusive environment (11,12). In response to growth at mild acidic conditions, genes encoding these proteins are up-regulated (13,14). Both AmiF and AmiE belong to members of aliphatic amidases. The presumed invariant active cysteine is identified as Cys 165 for AmiE and Cys 166 for AmiF from site-directed mutagenesis studies (12). Additionally, AmiF is noted to demonstrate restricted substrate specificity in which formamide is its only known substrate. Here we report the 1.75 Å crystal structure of the native AmiF. Crystal structures of an inactive mutant AmiF C166S were also determined in its apo and liganded forms. Structural analysis reveals a small binding pocket that has the key CEK catalytic residues to reduce the organic nitrogen compounds efficiently.
Site-directed mutagenesis was performed essentially using the method of overlap extension PCR with plasmid pQE30-AmiF as the template (16,17). All mutations were confirmed by sequencing of the whole ligated PCR fragment.
Expression of wild-type AmiF or the C166S mutant in Escherichia coli JM109 cells transformed with pQE30-AmiF or pQE30-C166S was induced at 37°C. Bacterial pellets were fractionated, and soluble proteins in cytosolic fractions were collected. The expressed protein with a His 6 tag was purified by immobilized nickel-ion chromatography, followed by MonoQ ion exchange chromatography (GE Healthcare), and then analyzed by SDS-PAGE to verify the purity. The protein concentration was assayed according to the Bradford method with bovine serum albumin as a standard (18).
Crystallization-All crystallization was performed by the hanging drop vapor diffusion method at 20°C. Initial crystallization conditions were screened using Crystal Screen I and II kits (Hampton Research) and Clear Strategy Screen I and II kits (Molecular Dimension). One l of protein solution (10 mg/ml) in 50 mM Tris-HCl, pH 8.0, was mixed with an equal volume of the well precipitant. The best diffracting AmiF crystals were found in a modified solution containing 0.1 M sodium cacodylate (pH 6.5), 0.15 M potassium thiocynate, and 20% polyethylene glycol (PEG) 550 MME. The crys-tal grew as rods and reached a maximum size of about 0.6 ϫ 0.1 ϫ 0.1 mm within 2 days at 20°C. The native crystal was characterized as belonging to space group R3, with the unit cell dimensions a ϭ b ϭ 147.21, c ϭ 72.27 Å. There are two molecules per asymmetric unit.
Crystals of the Se-Met-AmiF protein were obtained in a condition (0.1 M sodium cacodylate (pH 6.0), 0.8 M sodium formate, 8% PEG 20,000, and PEG 550 MME) using Se-Met-AmiF (20 mg/ml). Rod-shaped crystals with a maximum size of about 0.6 ϫ 0.1 ϫ 0.1 mm grew within 4 days at 20°C. The Se-Met-AmiF crystal also belongs to space group R3 with unit cell dimensions of a ϭ b ϭ 147.80 Å and c ϭ 72.78 Å.
Crystals of C166S were obtained in a solution containing 0.1 M sodium acetate (pH 5.0), 0.2 M lithium sulfate, and 14% PEG 2000 MME. Of two crystal forms (thin plates or rods), only the thin plate crystal diffracted beyond 3 Å using the in-house x-ray source. The thin plate crystal was characterized as space group C2, with the unit cell dimensions of a ϭ 117.72, b ϭ 130.53, c ϭ 144.59 Å, and ␤ ϭ 99.44°. There are six molecules per asymmetric unit.
Attempts to obtain liganded C166S crystals using the soaking method failed. After extensive trials, the C166S complex crystal was finally obtained by the co-crystallization method in a solution containing 0.1 M sodium acetate (pH 5.0), 0.2 M lithium sulfate, 6% PEG 1500, and 3 mM formamide. The complex crystal grew as a rod and was characterized as the space group P2 1 with cell dimensions of a ϭ 83.09, b ϭ 151.80, c ϭ 89.08 Å, and ␤ ϭ 114.99°. There are six molecules per asymmetric unit.
Data Collection-Prior to data collection, crystals were dipped into Fomblin cryoprotectant oil for several seconds and then flash-frozen in a liquid nitrogen stream. Diffraction data were collected using a MSC X-Stream cryosystem and an R-AXIS IVϩϩ image plate system with double mirror-focused CuK␣ x-ray radiation generated from a Rigaku RU-300 rotating anode generator at the Macromolecular X-ray Crystallographic Laboratory of National Tsing Hua University, Hsinchu, Taiwan. The 1.75 Å native data set, the 2.27 Å Se-Met-AmiF data set, the 2.50 Å C166S data set, and the 2.30 Å C166S complex data set were collected on the BL12B2 Taiwan beamline at SPring-8, Japan, using an ADSC Quantum 4R CCD detector. All data sets were collected at Ϫ165°C and processed with the HKL/HKL2000 suite (21). Data collection statistics are shown in Table 1.
Structure Determination and Refinement-The AmiF structure was solved by single wavelength anomalous dispersion methods. Twelve selenium sites were located with the program SOLVE (22). Phase determination, density modification, and automated model building with SOLVE/RESOLVE (22,23) led to an interpretable density map and initial model. The atomic model was further built using O (24). Crystallographic refinement was carried out using the maximum likelihood target function embedded in program REFMAC5 (25,26). The native AmiF model was obtained using MOLREP (27) and refined using REFMAC5 (25) coupled to ARP/wARP (28), which was used to add water molecules automatically. The protein model was assessed using the program PROCHECK (29). The apo and liganded C166S structures were determined by the molecular replacement methods using the native AmiF structure as a search model by MOLREP (27) and refined by REFMAC5 (25) coupled to ARP/wARP (28).
Structural Comparisons-Structure comparisons with N-carbamoyl-D-amino acid amidohydrolase (Protein Data Bank entry codes 1ERZ and 1FO6) (4,5), the Nit domain of NitFhit (Protein Data Bank entry code 1EMS) (3), and the putative CN hydrolase from yeast (Protein Data Bank entry code 1F89) (10), were carried out using the program LSQMAN in O (24) to superimpose C␣ atoms.
Combined sequence and secondary structure alignments and figure preparation ( Molecular Dynamic Simulations and Gaussian Network Model-Dynamic simulations were performed with the GROMACS 3.3 program running on a Fedora Linux system essentially as previously described (17,33). All simulations were carried out for 2 ns at 300 K. The first 1-ns run was used to attain equilibrium, whereas the trajectories of the last 1 ns were used for analysis. The low frequency motions were calculated based on the Gaussian network model (GNM) (34,35).

RESULTS AND DISCUSSION
Structure Description of the Native AmiF-The 1.75 Å resolution electron density map of AmiF revealed two molecules (AB) per asymmetric unit. Except for the N-terminal segment (residues 1-12) that contained weak or negative density and could not be built into the model, the main-chain and sidechain atoms from 13 to 334 residues are well defined in A and B subunits. Only one residue, Cys 166 , lies in the disallowed region of the Ramachandran plot, despite its low B factor. The unusual geometry is stabilized by three hydrogen bonds around this peptide unit: Cys 166 O⅐⅐⅐N Asp 168 , Cys 166 O⅐⅐⅐N Gly 169 , and Cys 166 O⅐⅐⅐N2 Arg 188 . The final model was refined to an R value of 17.4% (R free ϭ 21.5%; Table 1). The average B-factor for all polypeptide atoms is 23.6 Å 2 .
In the crystal structure, AmiF is a hexamer (Fig. 1): Subunit A and Subunit B associate into an AB dimer related by a noncrystallographic 2-fold axis (Fig. 1B). Along the 3-fold axis, three AB dimers assemble into a hexamer (A1B1A2B2A3B3) (Fig. 1A). The overall shape of each trimer (A1A2A3 or B1B2B3) is approximately an equilateral triangular prism (ϳ95 Å along the prism axis).
Each monomer folds into a four-layer ␣-␤-␤-␣ structure, characteristic of members in the nitrilase superfamily (Fig. 2, A and B). One exterior ␣ layer (␣1 and ␣2) is solvent-accessible to shield the inner ␤ sheets from the solvent, whereas the other ␣ layer (␣3-␣7) associates with the neighboring Subunit B. One side of the four-layer sandwich contains short turns or loops to connect the secondary elements, whereas the other side has several long loops, particularly the ␤2-1-␣2, ␤5-␤6, and ␤9 -␤10 segments. Buried residues at the A-B intersubunit interface are largely from residues of the ␤5-␤6 loop, near or in those of ␣3-␣7 helices and the C-terminal region of each monomer. The symmetric assembly (␣3 versus ␣3, ␣4 versus ␣4, and ␣7 versus ␣7) forms a tightly associated ␣-␤-␤-␣-␣-␤-␤-␣ dimer (Fig. 1B). In total, there are 456 interactions within 3.8 Å, including 36 hydrogen bonds and 13 salt bridges. The total solvent-accessible surface of the AB dimer is ϳ29,700 Å 2 , calculated by protein interfaces, surfaces, and assemblies service PISA at the European Bioinformatics Institute (available on the World Wide Web at www.ebi.ac.uk/msd-srv/prot_int/pistart.html). Of the subunit surface area, 28.8% is used for the formation of the dimer.
Crystal Structures of C166S-To obtain liganded structures, crystallization trials using the co-crystallization or soaking methods in the presence of various concentrations of formamide were attempted. Of various structures, only one showed additional nonpeptide electron density near Cys 166 (data not shown). Given the relatively large peak, either formamide or a solvent cacodylate molecule could be docked into this density in either subunit. Nonetheless, a piece of residual density that could not be ignored is present in either model, indicating a possibility that formamide and cacodylate co-exist in this density. However, we were not able to model both properly into this density, possibly due to the disordered formamide/cacodylate displacement.
Crystal Structure of C166S⅐Formamide Complex-After extensive trials, the complex crystal was also obtained by the co-crystallization method in a solution containing 3 mM formamide. The structure was then solved by the molecular replacement methods, showing six subunits. Residues that could not be defined include the following: N-terminal segment (residues 1-12), Ala 194 , Phe 195 , and a loop region (residues 223-227).
As shown in Fig. 3, the electron density maps show a nonpeptide electron density nearby the catalytic triad of the C166S complex structure as compared with a water molecule in the apo structure (Fig. 3). This density is defined as a formamide molecule, in which two possible directions of formamide are considered (model 1 and model 2; Fig. 3, B and C). In either model, formamide is modeled in three subunits (ACF) and is omitted in the other subunits due to the weak density.
Given the opposite orientation in two models, formamide interacts with nearby polar groups distinctively ( Table 2). It is noted that the oxygen atom of formamide is in close proximity to the main-chain nitrogen atom of His 167 (3.02 Å in Subunit A) and the Lys 133 N atom (3.15 Å in Subunit A) in model 2. These interactions may together stabilize the transition state intermediate that bears a negative charge (discussed under "Proposed Mechanism"). Model 1, in contrast, lacks such contacts. Additionally, superposition between liganded C166S and N-carbamoyl D-amino acid amidohydrolase (6) structures reveals that formamide of model 2 superimposes relatively well with the reactive atoms of the other substrate, N-carbamoyl-D-p-hydroxyphenylglycine (discussed under "Conserved Catalytic CEK and Glutamate Residues and Distinct Ligand-binding Residues between AmiF and N-Carbamoyl D-Amino Acid Amidohydrolase"). These results together suggest that model 2 has a correct assignment for formamide. Subsequent analysis is thus conducted based on this model. Like the native AmiF, the C166S crystal structure shows a tight dimer (AB) related by a noncrystallographic 2-fold axis. Along a noncrystallographic 3-fold axis, three AB dimers pack as a hexamer. Similarly, the C166S-formamide structure reveals a tight AB dimer and a loosely associated hexamer (AB) 3 .
Superposition of the C␣ atoms of the CEK triad between apo and liganded C166S structures shows virtually identical C␣ conformation in the binding site region (Glu 60 , Lys 133 , Trp 137 , Glu 141 , C166S, His 167 , and Tyr 192 ) with or without the ligand (Fig. 4B). Moreover, the side chains of Glu 60 , Lys 133 , and C166S superimpose relatively well between apo and liganded forms of C166S. It is noted that a water molecule is positioned between Glu 60 and C166S (Glu 60 (⑀1)-Wat 515 (O), 2.84 Å; Glu 60 (⑀2)-Wat 515 (O), 3.53 Å; C166S (O␥)-Wat 515 (O), 3.24 Å in Subunit A) in the apo form, which is replaced by formamide in the bound form (Fig. 3). The wild-type apo structure also includes a water molecule at the same site. These results suggest that Glu 60 , Lys 133 , Trp 137 , Glu 141 , C166S, His 167 , and Tyr 192 are the ligand-binding residues. Upon binding to a formamide molecule, the reactive atom (the S␥ atom of Cys 166 in AmiF) may then react toward the carbon atom of formamide for the presumed nucleophilic attack.
Comparison with Homologous Structures-Based on analysis with DALI (34), all members of the nitrilase superfamily exhibit structural homology, including a hypothetical protein from P. horikoshii that shares 24% sequence identity (Protein Data  Bank code 1F89, Z score ϭ 29.1, r.m.s. deviation ϭ 2.8 Å for 251 residues) (10), NitFhit from Caenorhabditis elegans that shares 20% identity (Protein Data Bank code 1EMS, Z score ϭ 27.9, r.m.s. deviation ϭ 2.9 Å for 245 residues) (3), N-carbamoyl D-amino acid amidohydrolase that shares 29% identity (Protein Data Bank code 1ERZ, Z score ϭ 25.4, r.m.s. deviation ϭ 2.8 Å for 239 residues) (4) (Fig. 5A). All of these proteins have the same ␣-␤-␤-␣ fold (Fig. 5B). Significant structural conservation is seen in the CEK triad among nitrilase members in which each catalytic residue is located in or adjacent to the C-terminal anchoring point of a ␤ strand (Fig. 5B) (Glu 60 , ␤2; Lys 133 , ␤5; Cys 166 , ␤7). Moreover, they exist as a dimer or a higher symmetric form in the crystal. These dimers make intersubunit contacts largely from an exterior ␣ layer and the C-terminal region, forming an ␣-␤-␤-␣-␣-␤-␤-␣ architecture (Fig. 5B).  (6). These interactions together enable the active nucleophile to polarize so as to attack the susceptible C-N bond of formamide or the carbamoyl moiety efficiently for the hydrolytic reaction. Apart from the triad, Glu 141 of AmiF superimposes relatively well with Glu 146 of N-carbamoyl Damino acid amidohydrolase (Fig. 6), although their sequence is at a nonconserved segment between ␤5 and ␤6 in the nitrilase superfamily (Fig.  5A). For Glu 141 of AmiF, it sits just next to a ␤ turn, and its carboxyl group of the side chain contacts with the polar groups of Glu 60 and Lys 133 . It is noted that Glu 141 also contacts with formamide (Table 2). Similarly, for Glu 146 of N-carbamoyl D-amino acid amidohydrolase, it is just next to a 3 10 helix and makes hydrogen bonds with side chains of Glu 47 and Lys 127 and the ligand. These results suggest that this conserved glutamate is crucial to maintain the side-chain geometry of the catalytic CEK triad as well as to facilitate the docking of a substrate  residues 191-197)) and a short 3 10 turn between ␤7 and ␣3 (residues 166 -170) enclosing formamide are colored beige. Glu 60 , Lys 133 , and C166S are shown as heavy gray sticks, whereas other ligand-binding residues are shown as green sticks. B, superposition between apo-C166S (green) and C166S⅐formamide (orange) subunits. C166S-Glu 60 -Lys 133 residues are shown as heavy sticks. Formamide (FRA) in A and B is drawn as ball-and-stick models. The carbon, oxygen, and nitrogen atoms are colored white, red, and blue, respectively.  APRIL 20, 2007 • VOLUME 282 • NUMBER 16 Apart from the CEK and Glu 146 , N-carbamoyl D-amino acid amidohydrolase has a subset of residues contacting with the carboxyl group of the ligand (Asn 173 , Arg 175 , Arg 176 , Asn 197 , and Thr 198 ) (Fig. 6, A and B,  right). This enzyme also possesses a relatively large space for a bulky side-chain substrate (D-phenylglycine, D-methionine, and D-leucine) (37,38).

Structure of H. pylori AmiF
In AmiF, residues corresponding to bind to the carboxyl moiety in N-carbamoyl D-amino acid amidohydrolase (His 167 , Gly 169 , Met 170 , Gly 191 , and Tyr 192 ) are not conserved (Fig. 6A). Such a structural disparity may explain the distinct substrate specificity between N-carbamoyl-Damino acid amidohydrolase and AmiF. Moreover, Trp 137 and Tyr 192 form an exterior wall in the other side (Fig. 6B, left), forming a much smaller room of the active site. As a result, AmiF can only accommodate a small ligand like formamide.
Proposed Mechanism-Given the common CEK residues that point toward the presumed susceptible C-N bond of their ligands between AmiF and N-carbamoyl D-amino acid amidohydrolase, these catalytic residues are proposed to share the same catalytic roles: Cys 166 is the nucleophile; the nearby Lys 133 plays an essential role in stabilizing the transition state intermediate; and Glu 60 mediates the proton transfer. A water molecule is noted to sit between Glu 60 and Cys 166 , which may be coupled with Glu 60 for the proton transfer given the low pK a of the glutamate. N-carbamoyl Damino acid amidohydrolase also possesses a water molecule at a similar site.
The first stage to hydrolyze formamide by AmiF is an acylation reaction (Fig. 7). (i) When a formamide molecule diffuses into the pocket, the substrate would bind onto the CEK triad of AmiF, positioning the formamide in a right orientation via interacting with ligand-binding residues. Glu 60 may then mediate the pro-FIGURE 6. A, superposition of the binding sites between AmiF (C166S)⅐formamide (orange) and N-carbamoyl D-amino acid amidohydrolase (C172S)⅐N-carbamoyl-D-p-hydroxyphenylglycine (HPG) (green) complex structures. Catalytic CEK residues are shown as heavy sticks, whereas other ligand-binding residues are shown as thin sticks. Formamide (FRA) and hydroxyphenylglycine are drawn as ball-and-stick models. The carbon, oxygen, nitrogen, and sulfur atoms are colored white, red, blue, and yellow, respectively. B, molecular surfaces of AmiF (C166S)⅐FRA (left) and N-carbamoyl D-amino acid amidohydrolase (C172S)⅐HPG (right) complexes. Surfaces are colored by the type of residues: glutamate (red), cysteine (orange), lysine (blue), histidine (cyan), arginine (dark blue); serine (yellow orange), asparagine (green), threonine (cyan), tryptophan (brown), and tyrosine (yellow). FRA and HPG are drawn as ball-and-stick models. The carbon, oxygen, and nitrogen atoms are colored white, red, and blue, respectively. ton transfer from the -SH group of C166, yielding a strongly nucleophilic -S Ϫ ion to initiate the first attack to the carbon atom of formamide. (ii) A tetrahedral transition state negatively charged intermediate is formed, which is surrounded with polar atoms (Table 2). Notably, the carbonyl oxygen makes a hydrogen bond with the peptide H atom of His 167 , which may stabilize the transition state intermediate in a manner similar to the oxyanion hole in the serine proteases (39). The Lys 133 N atom is also nearby. Hence, these interactions together reduce the activation energy barrier. (iii) Instability of the negative charge on the substrate carbonyl oxygen is likely to result in collapse of this intermediate within a short period of time, which leads to the production of an acyl-enzyme intermediate, breaking the C-N bond, and the release of an NH 3 molecule.
The second phase is to deacylate the acyl-enzyme intermediate. An incoming water molecule is deprotonated by Glu 60 , generating a nucleophilic hydroxide ion to start the second nucleophilic attack. Like the acylation stage, a tetrahedral intermediate is formed and may be stabilized by Lys 133 and His 167 . The subsequent collapse from the unstable intermediate yields a formic acid molecule. After release from the active site, a free enzyme is regenerated.
CEK Residues Have Significantly Lower Translational Fluctuations and Are Positioned in or near the Key Mechanical Sites-We previously found that the CEK residues coincide with or are near to the minima in the fluctuation profile computed by molecular dynamic simulation (17). In this investigation, the AmiF dimer (apo) shows fluctuation profiles comparable with those of N-carbamoyl-D-amino acid amidohydrolase (data not shown). Moreover, structurally conserved CEK residues (Glu 60 , Lys 133 , and Cys 166 ) display nearly minimal translational fluctuations (Glu 60 , 0.048; Lys 133 , 0.050; Cys 166 , 0.048). For ligand-binding residues of AmiF, there is larger degree of distribution in the fluctuation profiles (Trp 137 , 0.049; Glu 141 , 0.061; His 167 , 0.046; Tyr 192 , 0.079). Our results thus support the view of Yang and Bahar that most catalytic residues are located at or near the so-called global hinge regions and have lower translational fluctuations, whereas some subsets of residues may include more flexible sites for more efficient recognition of a diffusing substrate (40).
Analysis of the low frequency motions based on GNM model (34) reveals that the first or the slowest mode of the dimeric AmiF moves symmetrically with respect to extended C-terminal regions between the two monomers (data not shown), indicating a global hinge region lying at the interface. The second mode of AmiF describes the global motion within each monomer. This mode restrains the CEK catalytic residues to the key mechanical sites running through one side of the inner ␤ sheets, as evidenced by their low mobility scores (data not shown). These results thus suggest a close coupling of the CEK residues to a mechanically key site for an efficient catalytic hydrolysis, since small reorganization energy will reduce activation free energy and, hence, accelerate chemical reactions (41).
In conclusion, we have determined the 1.75 Å structure of a novel formamidase from H. pylori. The four-layer ␣-␤-␤-␣ topology falls into the category of the nitrilase superfamily. We also determined the C166S apo and complex structures, enabling us identify CEK catalytic residues as well as other ligand-binding residues. In parallel with its restricted substrate specificity, AmiF possesses a relatively small binding pocket. An acyl-enzyme intermediate-mediated catalytic mechanism that involves the CEK triad is also proposed here. Structural comparison between N-carbamoyl D-amino acid amidohydrolase and AmiF reveals a common CEK triad but distinctive subsets of ligand-binding residues. Furthermore, molecular simulation and GNM analysis results demonstrate minimal fluctuations of the triad that sits in a mechanically key region for both N-carbamoyl D-amino acid amidohydrolase and AmiF. These results together support the catalytic roles of the conserved CEK triad in the nitrilase superfamily.