Catalytic Mechanism of Nitrile Hydratase Proposed by Time-resolved X-ray Crystallography Using a Novel Substrate, tert-Butylisonitrile*

Nitrile hydratases (NHases) have an unusual iron or cobalt catalytic center with two oxidized cysteine ligands, cysteine-sulfinic acid and cysteine-sulfenic acid, catalyzing the hydration of nitriles to amides. Recently, we found that the NHase of Rhodococcus erythropolis N771 exhibited an additional catalytic activity, converting tert-butylisonitrile (tBuNC) to tert-butylamine. Taking advantage of the slow reactivity of tBuNC and the photoreactivity of nitrosylated NHase, we present the first structural evidence for the catalytic mechanism of NHase with time-resolved x-ray crystallography. By monitoring the reaction with attenuated total reflectance-Fourier transform infrared spectroscopy, the product from the isonitrile carbon was identified as a CO molecule. Crystals of nitrosylated inactive NHase were soaked with tBuNC. The catalytic reaction was initiated by photo-induced denitrosylation and stopped by flash cooling. tBuNC was first trapped at the hydrophobic pocket above the iron center and then coordinated to the iron ion at 120 min. At 440 min, the electron density of tBuNC was significantly altered, and a new electron density was observed near the isonitrile carbon as well as the sulfenate oxygen of αCys114. These results demonstrate that the substrate was coordinated to the iron and then attacked by a solvent molecule activated by αCys114-SOH.

Nitrile hydratases (NHases) 2 catalyze the hydration of nitriles to the corresponding amides and are used as catalysts in the production of acrylamide, making them one of the most important industrial enzymes (1,2). NHases contain a nonheme Fe 3ϩ or non-corrin Co 3ϩ catalytic center. Iron-type NHases show unique photoreactivity; the enzyme is inactivated by nitrosylation in the dark and immediately reactivated by photo-induced denitrosylation (3)(4)(5). The protein structure is highly conserved among all known NHases (6 -9) as well as a related enzyme, thiocyanate hydrolase (10). The metal site is also conserved, with a distorted octahedral geometry. All ligand residues are involved in a strictly conserved motif of the ␣ subunit, Cys 1 -Xaa-Leu-Cys 2 -Ser-Cys 3 , where two amide nitrogens of Ser and Cys 3 and three Cys sulfurs are coordinated to the metal (6). Cys 2 and Cys 3 are post-translationally modified to cysteine-sulfinic acid and cysteine-sulfenic acid, respectively (7), which probably take deprotonated forms at the metal site (11). The sixth ligand site is occupied by a solvent molecule (8) or by a NO molecule in nitrosylated iron-type NHase (7).
Several reaction mechanisms have been proposed based on the protein structures (1,6). First, nitriles directly bind to the metal to facilitate the nucleophilic attack of a water molecule on the nitrile carbon. In the other mechanisms, a water molecule activated by the metal directly or indirectly attacks nitriles trapped near the metal. In all cases, the metal is suspected to function as a Lewis acid. By reconstituting iron-type NHase from recombinant unmodified subunits, we demonstrated that the post-translational modifications of its cysteine ligands are essential for its catalytic activity (12). We also found that specific oxidation of the cysteine sulfenic acid ligand to cysteine sulfinic acid resulted in irreversible inactivation (13). Kovacs and co-workers (14) studied the ligand exchange reaction in the low spin Co 3ϩ -containing NHase model complexes and concluded that the trans-thiolate sulfur played an important role in promoting the ligand exchange at the sixth site. Later, by using a sulfenate-ligated iron complex, they showed that protonation/deprotonation states of the sulfenate oxygen were modulated by the unmodified Cys thiolate ligand (15) (16). Interestingly, the hydration activity was enhanced by the mono-oxygenation of one of two sulfur ligands (17). Heinrich et al. (18) demonstrated that Na[Co(L-N 2 SOSO)(tBuNC) 2 ] exhibited the nitrile hydration activity but that (Me 4 N)[Co(L-N 2 SO 2 SO 2 )(tBuNC) 2 ] did not. These results indicate that the oxidized cysteine ligands, especially the cysteine sulfenic acid ligand, play an important role in the catalysis. Recently, theoretical calculations, including density functional calculations (19,20) as well as molecular dynamics simulations (21), have been applied to the mechanisms described above. However, the detailed mechanism remains unclear because of a lack of direct information on the reaction intermediates.
We recently found that an iron-type NHase from Rhodococcus erythropolis N771 (ReNHase) catalyzes the conversion of isonitriles to the corresponding amines (22). Although the other product derived from the isonitrile carbon was not identified, the kinetic analyses revealed that the K m for tBuNC was comparable with that for methacrylonitrile, whereas k cat (1.8 ϫ 10 Ϫ2 s Ϫ1 ) was 1.8 ϫ 10 5 times smaller. In this study, taking advantage of the slow reactivity of tBuNC as well as the photoreactivity of nitrosylated inactive ReNHase (3,4), we obtained structural evidence on the reaction mechanism by studying the time course of the tBuNC catalysis with x-ray crystallography. Based on the results, we propose a reaction mechanism in which the sulfenate group of ␣Cys 114 -SO Ϫ plays a key role in the catalysis.

EXPERIMENTAL PROCEDURES
Materials-Nitrile hydratase from R. erythropolis N771 (ReNHase) was inactivated by endogenous NO molecules in living cells in the dark (4,23). ReNHase was purified in the nitrosylated form in the dark as described previously (23). The purified nitrosylated ReNHase was stored in 50 mM Tris-HCl, pH 7.5, at Ϫ80°C in the dark at a concentration of 20 mg/ml. The concentration of the nitrosylated NHase was determined by measuring the absorbance at 280 nm (⑀ 280 ϭ 1.7 ml mg Ϫ1 cm Ϫ1 ). tBuNC was purchased from Tokyo Chemical Industry Co., Ltd. (Tokyo, Japan). All the other reagents used in this study were of the highest grade available.
ATR-FTIR Measurements-The nitrosylated ReNHase (70 mg/ml) in 50 mM sodium phosphate, pH 7.5, was loaded on the surface of a three-reflection silicon prism (3 mm in diameter) in the ATR accessory (DuraSamplIR II, Smiths Detection, Danbury, CT) and dried under nitrogen gas flow. Subsequently, 1.5 l of water (H 2 16 O or H 2 18 O) was added to the sample. The sample space was sealed with a CaF 2 plate and a greased Teflon spacer (0.7 mm in thickness). The substrate, 0.5 l of neat tBuNC, was also enclosed in this sealed space as a drop on the CaF 2 plate, to be supplied to the ReNHase solution as a vapor. The sample was stabilized at room temperature in the dark for 4 h.
FTIR spectra were measured on a Bruker IFS-66/S spectrophotometer equipped with an MCT detector (D313-L). All of the spectra were recorded at 4 cm Ϫ1 resolution. A single-beam spectrum was recorded for 100 s before illumination, and ten spectra (100 s scans) were successively recorded after 10 s of illumination by continuous white light from a halogen lamp (Hoya-Schott HL150; ϳ60 milliwatt cm Ϫ2 at the sample).
Light-induced difference spectra were calculated by subtracting the dark spectra from each spectrum after illumination. The base-line distortion was corrected by subtracting the corresponding spectra measured in the same manner but without illumination.
For CO detection, 6 l of hemoglobin (50 mg/ml) in 50 mM Tris-HCl, pH 7.5, was lightly dried on a silicon ATR prism, and 6 l of the nitrosylated ReNHase sample (70 mg/ml) in Tris-HCl, pH 7.5, was placed beside the hemoglobin. The sample space was sealed with a CaF 2 plate, on which 0.5 l of tBuNC was placed, and a greased Teflon spacer (Fig. 1). The ReNHase sample was photoactivated by white light illumination for 1 min. FTIR spectra with 10-s scans were recorded at 0, 20, 30, and 60 min after illumination.
Crystallization of ReNHase-Crystals of the nitrosylated ReNHase were grown using the vapor diffusion hanging drop method at 20°C. Two microliters of the nitrosylated ReNHase (20 mg/ml protein in 50 mM Tris-HCl, pH 7.5) was mixed with an equal amount of the precipitant solution (20% polyethylene ZnSe Silicon ATR prism CaF 2 plate greased Teflon spacer NHase ATR stage FIGURE 1. The sample unit of the ATR-FTIR measurement for CO detection using hemoglobin. Hemoglobin was loaded on a silicon ATR prism, and an NHase solution and tBuNC were separately placed in a sealed space. tBuNC was supplied to the NHase solution as a vapor, and the reaction was initiated by photoactivation of NHase. solutions including NHase and tBuNC were recorded before and 100 (purple), 300 (blue), 500 (cyan), 700 (green), and 900 (red) s after illumination of ReNHase, and difference spectra were calculated relative to before illumination values. C, the spectrum of tBuNH 2 in an aqueous solution (in a protonated tBuNH 3 ϩ form) after subtraction of water absorption is presented for comparison.

Catalytic Mechanism of Nitrile Hydratase
glycol 8000, 0.10 M Tris-HCl, pH 7.5, 0.30 M MgCl 2 ) and equilibrated against 0.40 ml of precipitant solution. Crystals with dimensions of approximately 0.4 ϫ 0.3 ϫ 0.3 mm 3 grew within a day in the dark at 20°C. When crystals of the nitrosylated NHase were dissolved in 50 mM Tris-HCl, pH 7.5, the enzyme solution exhibited trace amounts of methacrylonitrile hydration activity in the dark, but it had a specific activity of 7.3 ϫ 10 2 units/mg after light-induced denitrosylation (10,000 lϫ) with a cold light illumination system (LG-PS2; Olympus, Tokyo, Japan) for 15 min.
Preparation of the ReNHase Crystals without or with tBuNC-Crystals of the nitrosylated ReNHase were first vapor-soaked with cryoprotectant solution (30% polyethylene glycol 8000, 0.10 M Tris-HCl, pH 7.5, 0.60 M MgCl 2 ) for 1 day by being swapped in mother liquor. They were then vapor-soaked for a day with mother liquor solution containing tBuNC at a final concentration of 0.10 M. After being mounted, ReNHases in the crystals were activated by light-induced denitrosylation (10,000 lϫ) with a cold light illumination system (LG-PS2; Olympus), and the reaction proceeded for 18, 120, 340, and 440 min at 20°C. At each elapsed time, the reaction was terminated by flash cooling with N 2 gas at 95 K.
X-ray Data Collections, Structure Determinations, and Refinements-Diffraction data were collected using a Quantum 315 CCD detector (Area Detector Systems Corporation, Poway, CA) at the beamline BL-5A ( ϭ 1.000 Å) of the Photon Factory (Tsukuba, Japan) at 95 K. Each data set was indexed, merged, and scaled with the HKL2000 program suite (24 2 18 O (B) buffer and tBuNC were separately placed in a sealed space (Fig. 2). The spectra at 0 (blue), 20 (cyan), 30 (green), and 60 (red) min after photoactivation of NHase were recorded. ReNHase crystals belonged to the C2 space group. One heterodimer of ␣ and ␤ subunits populated the asymmetric unit. Molecular replacement was performed with MOLREP (25) in the CCP4 program suite (26) using the structure of the nitrosylated ReNHase in the P2 1 2 1 2 space group (Protein Data Bank code 2ahj) (7) as the initial coordinates. The obtained models were improved by iterative cycles of crystallographic refinement using REFMAC5 (27) and manual model rebuilding using Coot (28). The models were cross-validated by the SigmaAweighted electron density maps (29) calculated with both 2mF obs Ϫ DF calc and mF obs Ϫ DF calc coefficients. The refinements were performed using a maximum likelihood target with bulk solvent corrections. During the structure refinement, ϳ5% of the amplitude data were set aside to monitor the progress of refinement using the R free factor. Solvent water molecules were gradually introduced if the peaks that were contoured at more than 4.0 in the mF obs Ϫ DF calc electron density were in the range of a hydrogen bond. tert-Butyl groups of tBuNC were fit on the resultant difference electron density map by handling, and their coordinate data were then refined using REFMAC5 (27). All of the structural figures were generated using PyMol.

Identification of the Product from the Isonitrile Carbon by ATR-FTIR Measurements-
To identify all products except for the amine, the reaction was monitored using ATR-FTIR. tBuNC was added as a vapor to nitrosylated NHase, and the enzyme was activated by light-induced denitrosylation. Several prominent positive peaks, all arising from tert-butylamine (tBuNH 2 ), increased their intensities as the reaction proceeded, whereas no signals from other origins were detected (Fig. 2). Supposing that the other product possessing a carbon atom was CO, which escaped from the solution as a gas, we attempted CO detection by trapping using hemoglobin. Hemoglobin was located on a silicon ATR crystal, and nitrosylated ReNHase solution and tBuNC were separately placed in a sealed space (Fig. 1). A CO molecule-bound hemoglobin was monitored by ATR-FTIR after light activation of ReNHase (Fig. 3). The CO stretching signal of CO-hemoglobin was observed at 1953 cm Ϫ1 , and its intensity increased with the reaction time. The CO peak appeared at a downshifted frequency of 1908 cm Ϫ1 when the ReNHase reaction was performed in an H 2 18 O buffer, confirming that CO was produced by the ReNHase-consuming water. Thus, we concluded that ReNHase hydrolyzed tBuNC to produce tBuNH 2 and CO (tBuNC ϩ H 2 O 3 tBuNH 2 ϩ CO).
Time-resolved X-ray Crystallography of the Reaction of ReNHase with tBuNC-Crystals of nitrosylated ReNHase were soaked with tBuNC, and the reaction was started by light-induced denitrosylation at 293 K. At 18, 120, 340, and 440 min, the reaction was stopped by flash cooling at 95 K, at which point the crystal structure was determined. Details of data collection and refinement statistics are summarized in Table 1. Unfor-tunately, we could not collect data from the crystals that were incubated longer because those crystals were damaged. The overall structure at each elapsed time was essentially unchanged except for the pocket above the Fe 3ϩ center (Fig.  4). ␣Cys 112 -SO 2 Ϫ (␣CSD112) and ␣Cys 114 -SO Ϫ (␣CSO114) modifications were clearly observed in all of the structures determined.
Before soaking with tBuNC, an NO molecule was observed at a distance of 2.1 Å from the Fe 3ϩ (Fig. 4A). The Fe-N(NO) distance is 0.6 Å longer than observed in the previous structure (Protein Data Bank code 2ahj) (7). In the previous structure, the NO was likely to be pushed toward Fe 3ϩ by 1,4-dioxane, the co-precipitant used (7). After soaking with tBuNC, the electron density of tBuNC was clearly observed in the pocket (Fig. 4B) with the tert-butyl group facing the NO molecule coordinated to the Fe 3ϩ . Because of the limited space in the hydrophobic pocket, the bulky tert-butyl group must face the iron in its nitrosylated state. In addition to the original conformation (conformer A), S␥ of ␤Met 40 took another conformation (conformer B) with occupancies of A:B ϭ 0.25:0.75. Movement of S␥ of ␤Met 40 to conformer B is likely due to the occupation of the hydrophobic pocket by tBuNC. We hypothesize that conformer B is less stable because of steric hindrance between S␥ and the amide oxygen of ␤Met 40 (Fig. 5).
At 18 min, electron densities of NO and tBuNC, especially that of the isonitrile group, were attenuated (Fig. 4C). S␥ of ␤Met 40 remained disordered, but the occupancy of conformer A increased to 0.55. At 120 min, the NO disappeared, and a tBuNC molecule was coordinated to Fe 3ϩ with an Fe-C(-NC) Ϫ and ␣Cys 114 -SO Ϫ , respectively. Yellow, blue, red, green, and brown spheres represent carbon, nitrogen, oxygen, sulfur, and iron atoms, respectively.  Catalytic Mechanism of Nitrile Hydratase DECEMBER 26, 2008 • VOLUME 283 • NUMBER 52 length of 2.1 Å (Fig. 4D). ␤Met 40 took conformer A again. The rotation of the tBuNC molecule could be driven by the recovery of ␤Met 40 to conformer A. The F o Ϫ F c electron density at 340 (supplemental Fig. S1) and 440 min (Fig. 4E) were very similar to one another but distinct from those observed at 120 min. In both structures, the F o Ϫ F c electron density corresponding to the tert-butyl group was moved ϳ1.0 Å away from the iron, and an extra electron density was observed near the isonitrile carbon as well as the sulfenate oxygen of ␣Cys 114 . When the products, tBuNH 2 and CO, were included in the calculation of the electron density at 440 min, the refined model of tBuNH 2 was well fit on the 2F o Ϫ F c electron density, but that of CO was not (supplemental Fig.  S2). In addition, two positive electron densities were observed near the CO molecule in the F o Ϫ F c electron density. Alternatively, we calculated the electron density at 440 min by assuming the presence of only tBuNH 2 . As shown in Fig. 4F, tBuNH 2 was well fit on the 2F o Ϫ F c electron density, and two positive electron densities were observed above the iron ion and near O␦ of the sulfenate group, in the F o Ϫ F c electron density. We assigned the positive densities as the carbon of the isonitrile group and the solvent water molecule (named as H 2 Oa), respectively (  Table 2). The O(H 2 Oa)-O(-SO) distance cannot be explained. These atoms may be disordered because the occupancies of H 2 Oa and O␦ of ␣Cys 114 -SO Ϫ converged on 0.50. Interestingly, a positive difference density was observed below S␥ of ␣Cys 114 -SO Ϫ in the 2F o Ϫ F c electron density map after coordination of tBuNC (Fig. 4, D-F). The distance between the density and S␥ of ␣Cys 114 -SO Ϫ is 1.4 Å, and the angle O␦(␣Cys 114 -SO Ϫ ) Ϫ S␥(␣Cys 114 -SO Ϫ ) Ϫ the density was 133°. The positive density may represent an alternative position of O␦ of ␣Cys 114 -SO Ϫ .
Proposed Catalytic Mechanisms of NHase-Based on the results, we propose the following catalytic mechanism: the tBuNC substrate binds the metal directly, and then a water molecule, activated by O␦ of ␣Cys 114 -SO Ϫ , makes a nucleophilic attack on the isonitrile carbon to produce tBuNH 2 and CO (Fig. 6A). Considering the similarity between isonitriles and nitriles, nitrile hydration is likely to proceed in a similar manner (Fig. 6B). When a nitrile coordinates to the metal, the nitrile carbon is attacked by a water molecule, activated by O␦ of ␣Cys 114 -SO Ϫ . The low k cat value for isonitrile may be due to limited accessibility of the activated water molecule because of steric hindrance by O␦ of ␣Cys 114 -SO Ϫ . Nitrile coordination to the Fe 3ϩ was suggested by electron spin resonance measurements (30). Involvement of ␣Cys 114 -SO Ϫ in the catalytic reaction had been suggested by our previous studies using the inhibitor, 2-cyano-2-propyl hydroperoxide (13), specifically oxidizing ␣Cys 114 -SO Ϫ to Cys-SO 2 Ϫ , and the site-directed mutant NHases (31). Yano et al. (32) have extensively studied the N 2 S 2 (tBuNC) 2 -type Co 3ϩ model complexes with different sulfur oxidation states and concluded that sulfur oxidations promoted the Lewis acidity of the Co 3ϩ center and that only the sulfenyl oxygen exhibited a nucleophilic character. Theoretical calculation studies have indicated that O␦ of ␣Cys 114 -SO Ϫ could be a catalytic base when nitrile coordination was assumed (19). These antecedent studies support the mechanism of the substrate coordinated to the iron being attacked by water activated by ␣Cys 114 -SO Ϫ . Recently, involvement of the Ser ligand (␣Ser 112 ; corresponding to ␣Ser 113 of ReNHase) and of the vicinal Tyr and Trp residues (␤Tyr 68 and ␤Trp 72 ; corresponding to ␤Tyr 72 and ␤Tyr 76 of ReNHase) in the catalytic mechanism was suggested by temperature-and pH-dependent kinetic studies of the Co-type NHase from Pseudonocardia thermophila JCM 3095 (33). However, the corresponding residues of ReNHase were unchanged during our investigations (Fig. 4). Our findings represent the first structural evidence of reaction intermediates in NHase catalysis. The present results demonstrate a reaction mechanism in which the sulfenate group of ␣Cys 114 -SO Ϫ plays a key role in the catalysis. Cysteine oxidation has been found to play important roles in various proteins (34). The present work reveals a novel role of cysteine sulfenic acid as a catalytic base.