Probing the role of Asp-120(81) of metallo-beta-lactamase (IMP-1) by site-directed mutagenesis, kinetic studies, and X-ray crystallography.

Metallo-beta-lactamase IMP-1 is a di-Zn(II) metalloenzyme that efficiently hydrolyzes beta-lactam antibiotics. Wild-type (WT) IMP-1 has a conserved Asp-120(81) in the active site, which plays an important role in catalysis. To probe the catalytic role of Asp-120(81) in IMP-1, the IMP-1 mutants, D120(81)A and D120(81)E, were prepared by site-directed mutagenesis, and various kinetics studies were conducted. The IMP-1 mutants exhibited 10(2)-10(4)-fold drops in k(cat) values compared with WT despite the fact that they contained two Zn(II) ions in the active site. To evaluate the acid-base characteristics of Asp-120(81), the pH dependence for hydrolysis was examined by stopped-flow studies. No observable pK(a) values between pH 5 and 9 were found for WT and D120(81)A. The rapid mixing of equimolar amounts of nitrocefin and all enzymes failed to result in the detection of an anion intermediate of nitrocefin at 650 nm. These results suggest that Asp-120(81) of IMP-1 is not a factor in decreasing the pK(a) for the water bridging two Zn(II) ions and is not a proton donor to the anionic intermediate. In the case of D120(81)E, the nitrocefin hydrolysis product, which shows a maximum absorption at 460 nm, was bound to D120(81)E in the protonated form. The three-dimensional structures of D120(81)A and D120(81)E were also determined at 2.0 and 3.0 A resolutions, respectively. In the case of D120(81)E, the Zn-Zn distance was increased by 0.3 A compared with WT, due to the change in the coordination mode of Glu-120(81)OE1 and the positional shift in the conserved His-263(197) at the active site.

␤-Lactam antibiotics are effective chemotherapeutic agents for the treatment of infectious diseases caused by bacteria. The mechanism of the antibacterial activity of ␤-lactams involves the inhibition of the biosynthesis of the bacterial cell wall peptidoglycan. Since the introduction of ␤-lactams into clinical use, bacteria continue to evolve and to defend themselves from antibiotics by means of several strategies. One of these adaptations is the production of ␤-lactamases, which hydrolyze ␤-lactams to biologically inactive compounds. As of this writing, more than 300 ␤-lactamases have been identified (1).
A subdivision of class B ␤-lactamases based on primary amino acid sequences has been proposed (20). Those belonging to subclass B1 have the amino acid sequence HXHXD, including ␤-lactamase II (BcII) from Bacillus cereus (21), CcrA from Bacteroides fragilis (22), IMP-1 found in some clinical isolates of Serratia marcescens (4) and Pseudomonas aeruginosa (23), and VIM-2 found in some clinical isolates of P. aeruginosa (5). Those belonging to subclass B2 have the amino acid sequence NXHXD, including CphA and ImiS from Aeromonas (24,25). Finally, those belonging to subclass B3 have the amino acid sequence H(Q)XHXDH, including L1 from Stenotrophomonas maltophilia (3) and THIN-B from Janthinobacterium lividum (26). Therefore, the identification of common structural and mechanistic features of metallo-␤-lactamases constitutes a clinically important issue.
IMP-1, first isolated from S. marcescens in Japan, is a single chain peptide composed of 246 amino acids (4). The IMP-1 gene is mainly encoded in a plasmid and is generally thought to be the most dangerous enzyme. Several bacteria have been reported to produce this enzyme (27). The three-dimensional structure of IMP-1 has been reported by Concha et al. (14), and IMP-1 was found to possess two Zn(II) ions in the active site for activity, as has been reported for CcrA.
Mechanistic studies of di-Zn(II) MBLs have been carried out with a CcrA belonging to subclass B1 (28 -33, 47). In the threedimensional structure of CcrA, reported in 1996, Concha et al. (28) concluded that Zn1 (the two Zn(II) ions are termed Zn1 and Zn2, respectively) is coordinated by three His(s) and a bridging OH 2 or OH Ϫ , whereas Zn2 is coordinated by Asp, His, Cys, a bridging OH 2 or OH Ϫ , and another OH 2 (Fig. 1). OH 2 or OH Ϫ bridging between the two Zn(II) ions in the active site is proposed to act as the deprotonated nucleophile to attack the carbonyl carbon of the ␤-lactam ring with the assistance of Asp-120(103) (following BBL numbering), 2 which is conserved in all MBLs (20,28). Wang et al. (29,30,47) observed a transient increase in the intermediate in the hydrolysis of nitrocefin by CcrA in a pre-steady-state study and reported that the decay of this intermediate, a ring-opened form of nitrocefin with an anionic nitrogen, is rate-limiting. Yanchak et al. (31) reported that Asp-120(103) in the active site of CcrA plays a role in generating an intermediate but does not donate a proton to the intermediate during the rate-limiting step in the catalysis. In a quantum chemical study, Díaz et al. (32) predicted that the Zn2-bound Asp-120 in CcrA could participate as a proton shuttle or proton donor. Moreover, Suá rez et al. (33) proposed that the unprotonated Asp-120 results in the rapid formation of a rigid Zn1-OH-Zn2 linkage, whereas the neutral Asp-120 is compatible with a fluctuating Zn1-Zn2 distance through the breaking and/or formation of the Zn1-OH-Zn2 bridge.
To investigate the role of Asp-120(81) in catalysis by IMP-1, we prepared two mutants in which this amino acid is replaced with Ala and Glu, and we carried out a steady-state kinetics study with four ␤-lactams and pre-steady-state experiments with nitrocefin as a reporter substrate. Moreover, x-ray crystallographic studies of the mutants were performed to elucidate the correlation between the reactivity with respect to the hydrolysis of ␤-lactams and the coordination structure at the active site.

EXPERIMENTAL PROCEDURES
Site-directed Mutagenesis-The bla IMP gene cloned from strain S. marcescens TN9160 was amplified using two primers, IMP-PCR-F (5Ј-CCgATACAAgAACAACAg-3Ј) and IMP-PCR-R (5Ј-gTAgTggCgT-gCTgCAAC-3Ј) (4). The PCR products of the 1.2-kb DNA fragment were digested with SmaI and subcloned into a vector pKF19k that had been digested with SmaI in advance. The ligated pKF19k was named pKF19k/IMP. Site-directed mutagenesis was performed with the help of a Takara Mutan TM -Super Express Kit (Tokyo, Japan) by PCR. The PCRs were performed with the selection primer attached to the kit and mutagenic primers for D120(81)A (5Ј-CCgTgCTggCgCTAT-3Ј) and D120(81)E (5Ј-ATTCCgCCCgTCCTgTCgCTATg-3Ј). After the PCR products were transformed into Escherichia coli MV1184 cells, the base sequences of the two IMP-1 mutants were confirmed by DNA sequencing by using two primers IMP for (5Ј-CggTCTgTAggCTgTAATgC-3Ј) and IMP rev (5Ј-TgTggAATTgTgAgCgg-3Ј).
Enzyme Preparation-WT was purified as described previously (4,34). The mutagenic plasmids were transformed into E. coli JM109. The IMP-1 mutants were also purified according to the procedure of WT with the modification of the conditions as follows. 1) The antibiotic for selection was changed to 50 g/ml kanamycin. 2) The elution buffers for the SP-Sepharose Fast Flow column and Sephadex G-75 column were Tris-HCl buffer (50 mM, pH 7.4, 10 M Zn(NO 3 ) 2 for D120(81)A or 100 M Zn(NO 3 ) 2 for D120(81)E).
Circular Dichroism-The conservation of secondary structures was observed by CD spectroscopy. The proteins were diluted to a concentration of 1.5 M in phosphate buffer (10 mM, pH 7.0). The concentrations of WT and the IMP-1 mutants were determined by measuring the absorbance at 280 nm using an extinction coefficient of ⑀ 280 ϭ 49,000 M Ϫ1 cm Ϫ1 (34). CD spectra were recorded on a Jasco J-720 circular dichrograph (Tokyo, Japan) at 25°C over 200 -250 nm.
Metal Contents-Before the analyses of metal content, the buffer used for the protein samples was changed from phosphate buffer (50 mM, pH 7.0) to Tris-HCl buffer (50 mM, pH 7.4, 0.5 M NaCl), and the solution was applied to an Amersham Biosciences PD-10 column (inner diameter, 15 ϫ 50 mm) and eluted with metal-free Tris-HCl buffer (50 mM, pH 7.4). Each of the IMP-1 mutants was diluted to give the predetermined protein concentrations (2.0 and 4.0 M) with 0.2% HNO 3 . The determination of Zn(II), by atomic absorption, was conducted on a Hitachi Z8000 atomic absorption spectrophotometer (Tokyo, Japan) in the flame mode.
Zn(II) Ion Dissociation from the IMP-1 Mutants-The rate of hydrolysis of nitrocefin by the IMP-1 mutants was monitored in varying concentrations of Zn(NO 3 ) 2 (0 -6000 M) in acetic acid/sodium acetate buffers (0.2 and 0.5 M NaCl) adjusted to pH 4.4 -5.6 as described previously (34). Buffer solutions were prepared using ultrapure water from Wako (Osaka, Japan). The saturated activity, k max , and the dissociation constant of Zn(II) ion from the IMP-1 mutants (Equation 1) were obtained by fitting v init to Equation 2 by using a nonlinear regression analysis with KaleidaGraph, where v init is the initial velocity; [E] T is the total concentration of enzyme, and v 0 is the velocity without the addition of Zn(II) ion.

The pH Dependence of Hydrolysis under Conditions of Excess Zn(II)-
The pH dependence studies were carried out with MTEN buffer (50 mM Mes, 25 mM Tris, 25 mM 25 mM 2-aminoethanol, 100 mM NaCl) adjusted to pH 5.6 -9.0, and with acetic acid/sodium acetate buffers (0.2 and 0.5 M NaCl) adjusted to pH 4.6 -5.6. The concentration of Zn(NO 3 ) 2 was adjusted to 25 times the K d at each pH. Before measurement of the rate of hydrolysis of nitrocefin by the IMP-1 mutants, the enzyme solution was preincubated for 10 min at 30°C in the adjusted buffers. The enzymatic parameters, K m and k cat , were obtained by fitting to the Michaelis-Menten or the Lineweaver-Burk equations.
Pre-steady-state Kinetic Studies-Stopped-flow spectrophotometry was carried out with Tris-HCl buffer (50 mM, pH 7.4, 0.5 M NaCl and 3.23% Me 2 SO) on a Photal RA-401 UV-visible spectrophotometer  (28). The residues are labeled according to the BBL standard numbering (20). The Zn(II) atoms are shown in green, oxygen atoms in red, nitrogen atom in blue, and sulfur atoms in yellow.
Rapid-scan studies were carried out with Tris-HCl buffer (50 mM, pH 7.4, 0.5 M NaCl and 3.23% Me 2 SO) on Photal RA-415S rapid-scan spectrophotometer at 25°C. The spectra were collected between 325 and 525 nm and monitored at time intervals of 2 s (gate time 200 ms, and 16 repeats).
Spectroscopic Titration of the Nitrocefin Hydrolyzed Product by Hydrochloric Acid and D120(81)E Mutant-A solution of the nitrocefin hydrolysis product was prepared by mixing 100 l of a Me 2 SO solution of nitrocefin (1240 M) and 100 l of an IMP-1 solution (54.6 M) with 2.9 ml of Tris-HCl buffer (50 mM, pH 7.4, 0.5 M NaCl) followed by preincubation for 30 min at 30°C. Spectroscopic titration of the nitrocefin hydrolysis product (10 M) produced using WT was performed by the stepwise addition of hydrochloric acid (0.2 M) and the D120(81)E mutant (56 M) in Tris-HCl buffer (50 mM, pH 7.4, and 0.5 M NaCl) at room temperature.

Inhibition of D120(81)A Mutant by Anions and Anion
Dependence on the Kinetics of the Hydrolysis-The change in the catalytic activity of D120(81)A by the presence of phosphate anion was examined by monitoring the change in absorbance at 491 nm with nitrocefin as a substrate for 5 min at 30°C in Tris-HCl buffer (50 mM, pH 7.4, 0.5 M NaCl) at various concentrations of phosphate ion. Under the same conditions, the effects of NaClO 4 , Na 2 SO 4 , K 3 [Co(CN) 6 ], and K 4 [Fe(CN) 6 ] were also examined.
Crystallization-Crystals of the IMP-1 mutants for x-ray crystallography were prepared by the vapor diffusion (hanging drop method), analogous to the method reported by Concha et al. (14). The drops were prepared by mixing 5 l of protein solution (10 mg/ml in 20 mM Hepes-NaOH, pH 7.5) and 5 l of reservoir solution containing 0.2 M sodium acetate, 0.1 M sodium citrate buffer, pH 6.5, 30% PEG 4000 against 500 l of reservoir solution. The crystals grew as thin plates after ϳ3 weeks at 20°C.
X-ray Data Collection and Processing-All diffraction data were collected at 100 K without additional cryoprotectants. The data sets for D120(81)A and D120(81)E were collected at the SPring-8 on beamline 40B2 and 41XU (Harima, Japan) at 2.0 and 3.0 Å resolutions with a wavelength of ϭ 1.00 and 0.98 Å, respectively. The data for the both crystals were integrated, merged, and scaled using HKL2000 (36).
Structure Determination, Model Building, and Refinement-The structure of IMP-1 complexed with the inhibitor from P. aeruginosa at 2.0 Å resolution (PDB code 1DD6) (14) with neither Zn(II) ions nor water molecules nor inhibitor was used as the model for molecular replacement using the AMoRe software program (37). The CNS software program was used to refine the structures by positional and B-factor refinement (38). The interactive graphics program O was used to build the structures of the IMP-1 mutants (39). Topology and parameter files of oxidized Cys were utilized in the HIC-UP (x-ray.bmc.uu.se/ hicup) (40).

Construction of Recombinant Plasmids and the Mutation of Asp-120(81)-Site-directed mutagenesis was carried out with
Pfu Turbo TM polymerase using pKF19k/IMP as a template. The CD spectra of the IMP-1 mutants showed the same pattern as that for WT between 250 and 200 nm, indicating that no major structural change occurred as the result of mutation (data not shown). The Zn(II) contents of the D120(81)A and D120(81)E mutants were 2.2 and 2.3 per enzyme molecule, respectively, as determined by atomic absorption spectroscopy. A smaller binding affinity had been expected, but the replacement of Asp-120(81) with Ala or Glu did not significantly change the affinity of the protein for Zn(II) ion.
Steady-state Kinetics of Substrate Hydrolysis-The K m and k cat values for the hydrolysis of four ␤-lactams, cephaloridine, imipenem, nitrocefin, and benzyl penicillin, which are catalyzed by WT and the IMP-1 mutants, are shown in Table I. The inclusion of 1 mM Zn(NO 3 ) 2 in the buffer resulted in a slight change in K m and k cat . For all substrates measured, the IMP-1 mutants exhibited a reduced k cat compared with WT. D120(81)A and D120(81)E exhibited the k cat values that were 25-41,000 times and 40 -1750 times lower than those of WT, respectively. The IMP-1 mutants exhibited larger K m values for almost all the substrates examined than those of WT, except for nitrocefin, which showed a smaller K m value for D120(81)E. The hydrolysis of benzyl penicillin with D120(81)A and D120(81)E and of imipenem with D120(81)A in a buffer containing 1 mM Zn(NO 3 ) 2 could not be measured because the Zn(NO 3 ) 2 solution hydrolyzed these substrates very rapidly in the absence of the IMP-1 mutants.
pH Dependence of Hydrolysis under Conditions of Excess Zn(II) Ion-We reported previously that the activity of IMP-1 was restored by the addition of Zn(II) ions in an acidic buffer (34). The hydrolysis was measured in acidic buffer using nitrocefin (100 M) as the substrate, and the Zn(II) dissociation constant K d , based on Equation 2 of the IMP-1 mutants was evaluated by varying the concentrations of Zn(NO 3 ) 2 in acidic The pH dependence of the steady-state kinetic constant, k cat , was determined for the IMP-1 mutants with nitrocefin as a substrate. Two buffers were used as follows: acetic acid/sodium acetate buffers (0.2 and 0.5 M NaCl) (below pH 5.6) under the concentrations of Zn(NO 3 ) 2 at 25 times the K d in the range between pH 4.6 and 5.6, and MTEN buffer (pH 5.6 -9.0) containing Zn(NO 3 ) 2 (40 M for D120(81)A, 20 M for D120(81)E) in the range between pH 6.0 and 9.0. Plots of the logarithm of k cat for WT and the IMP-1 mutants, D120(81)A and D120(81)E, against pH are shown in Fig. 2. When the pH values of the buffers were in the range 4.6 -5.0, k cat and K m values were obtained from Lineweaver-Burk plots because nitrocefin precipitated below pH 4.8. The plots for WT and D120(81)A showed no inflections over the pH range measured. However, the plots for D120(81)E were increased slightly in the pH range from 5.2 to 9.0. Moreover, the logarithm of k cat of D120(81)E in the pH range from 5.2 to 4.6 was largely decreased.
Pre-steady-state Kinetics for the Hydrolysis of Nitrocefin Using WT and the IMP-1 Mutants-The reactions of nitrocefin with WT or D120(81)A were monitored by stopped-flow spectrophotometry at 391, 491, and 650 nm; the decrease in absorption at 391 nm and the increase in absorption at 491 nm were almost synchronous. No significant transient increase in the absorbance around 650 nm was detected. The time scale for the reaction of D120(81)A was 1/1000 slower than for WT.
However, the experiment with D120(81)E showed a nonsynchronous decrease and an increase in the absorbances at 391 and 491 nm, respectively, as shown in Fig. 3a. The increase in product appeared to reach completion more rapidly than the decrease in reactant. To solve this puzzle, rapid-scan spectroscopy was employed. When nitrocefin (8 M) and D120(81)E (8 M) were mixed in Tris-HCl buffer (50 mM, pH 7.4, 0.5 M NaCl, and 3.23% Me 2 SO), the absorption at 460 nm increased instead to 491 nm for the normal hydrolyzed nitrocefin as shown in Fig.  3b. This absorption at 460 nm showed no change for at least 50 min after the completion of the reaction of equimolar D120(81)E and nitrocefin (data not shown). However, for the reaction of nitrocefin (40 M) with D120(81)E (4 M), an ordinary absorption increase at 491 nm was observed (data not shown). To investigate the nature of this species having an absorbance maxima at 460 nm, the nitrocefin hydrolysis product was spectrophotometrically titrated with 0.2 M hydrochloric acid. The addition of hydrochloric acid caused a blue shift of 31 nm, and the absorption maxima was shifted to 460 nm with an ⑀ of 18,000 M Ϫ1 cm Ϫ1 when the pH of the medium reached a value of 2.3 (data not shown). The nitrocefin hydrolysis product was also titrated with D120(81)E. By the successive addition of D120(81)E to the nitrocefin hydrolysis product, the absorption spectra of the product was altered to that having an absorbance maxima at 460 nm (data not shown). Thus, the 460 nm species appears to be a protonated form of the nitrocefin hydrolysis product bound to D120(81)E.
Three-dimensional Structures of D120(81)A and D120(81)E-The three-dimensional structures of the D120(81)A and D120(81)E mutants were determined by the molecular replacement method using the known structure of WT complexed with the inhibitor (PDB code 1DD6) (14). Both mutants crystallize with a space group of P1, and four independent proteins, termed A-D, were located in the crystal lattice. The data collection and refinement statistics are summarized in Table III

. pH dependence plots for the hydrolysis of nitrocefin at 30°C for WT and the IMP-1 mutants D120(81)A and D120(81)E.
Data points were obtained in MTEN buffer (pH 5.6 -9.0) and acetic acid/sodium acetate buffer (pH 4.6 -5.6) in the presence of an excess of Zn(II) ion. the four independent molecules in the asymmetric unit for D120(81)A and D120(81)E gave root mean square deviations (r.m.s.d) of 0.23 and 0.49 Å, respectively. The IMP-1 mutants (molecule A was selected for each mutant) and WT were also superimposed, except for residues 62(26)-65 (29), which are disordered in WT, D120(81)A, and D120(81)E. The r.m.s.d. for the C-␣ atoms between the IMP-1 mutants D120(81)A and D120(81)E and WT were 0.35 and 0.56 Å, respectively. In a comparison of the deviations of the C-␣ atom of each residue between the IMP-1 mutants and WT, the residues moved most (Ͼ1 Å) were Glu-50(14) (1. A comparison of Zn(II) ion(s) and amino acid residue(s) as ligand(s) in the active sites is shown in Fig. 4, and the distances of the Zn(II) ion(s), ligands of the IMP-1 mutants, along with those of WT are summarized in Table IV. The active site in D120(81)A contains only one Zn(II) ion that is coordinated to His-116(77), His-118(79), and His-196(139) and two oxygen atoms of an acetate ion, a bidentate external ligand (Fig. 4b).
There is no spherical electron density in the space corresponding to a Zn(II) ion, but a tetragonal electron density corresponding to cysteine-s-dioxide (Csw221) was derived from Cys-221 (158). The active site of D120(81)E contains two Zn(II) ions, referred to as Zn1 and Zn2. Zn1 is coordinated by His-116(77), His-118(79), and His-196(139), and Zn2 is coordinated by Glu-120(81), Cys-221(158), His-263(197), and one oxygen atom of an acetate ion (Fig. 4c). The bridging OH 2 or OH Ϫ and a water molecule coordinating to Zn2 are not positioned in the active sites of both mutants, as was found for CcrA. In WT, Asp-120(81)OD1 is considered to bind via a hydrogen bond with OH 2 or OH Ϫ , although OH 2 or OH Ϫ was not observed in the three-dimensional structure of WT as a bridging ligand between Zn1 and Zn2 as CcrA (28). However, in D120(81)E, the distances from Glu-120(81)OE1 to Zn1 and Zn2 are 2.5 Ϯ 0.6 and 2.5 Ϯ 0.3 Å, respectively; therefore, it is presumed that this oxygen atom binds both Zn1 and Zn2 as a bridging ligand. The distance between Zn1 and Zn2 in D120(81)E is 3.6 Ϯ 0.1 Å, somewhat longer than that of WT (3.3 Å).

Asp-120 Mutants of Metallo-␤-lactamase IMP-1
Inhibition of D120(81)A Mutant by Anion and Anion Dependence on the Kinetics of the Hydrolysis-D120(81)A was inhibited by phosphate ion after its purification by column chroma-tography using phosphate buffer. The catalytic activities of WT and the IMP-1 mutants for nitrocefin hydrolysis were dependent on the concentration of phosphate ion as shown in Fig. 6. The activity of WT was decreased by 30% at a phosphate ion concentration of 50 mM, but D120(81)A lost 90% of its activity at a concentration of 5 mM. The activity of D120(81)E, on the other hand, was increased by 30% at a concentration of 50 mM. This inhibitory action of phosphate ion against D120(81)A was analyzed as a competitive binding of phosphate ion with the enzymes, where v o,init is the initial velocity in the absence of an inhibitor. The inhibition constants, K i , were estimated to be 144 and 0.32 mM for WT and D120 (81) D120(81)A (b) and D120(81)E (c). The calculated 2͉F o ͉ Ϫ ͉F c ͉ maps (magenta) are represented at 1.5 and 1.0 levels for b and c, respectively. Zn(II) ions and a water molecule are represented by green and red spheres, respectively. Amino acid residues are displayed as sticks (carbon, oxygen, nitrogen, and sulfur atoms colored in gray, red, blue, and yellow, respectively). In WT, a bridging OH 2 or OH Ϫ is not observed. M Zn(NO 3 ) 2 was added to the reaction buffer, the inhibition of D120(81)A by phosphate ion was suppressed (K i ϭ 1.8 mM; data not shown).

DISCUSSION
In this study, we performed a detailed analysis of the role of Asp-120, the conserved amino acid in the active site of all metallo-␤-lactamases and which is thought to play an important role in enzymatic activity.
However, the critical role of Asp-120 in the mechanism of hydrolysis of di-Zn(II) metallo-␤-lactamases is not clear to date. In this study, Asp-120(81) of a metallo-␤-lactamase, IMP-1, was replaced with Ala or Glu by site-directed mutagenesis, and the point mutated enzymes, D120(81)A and D120(81)E, were prepared and characterized by spectroscopic and kinetic studies and by x-ray crystallography. The introduction of a mutation at Asp-120(81) caused no modification in the overall structure, based on the CD spectra, compared with that of WT, and two Zn(II) ions were retained in the proteins, as evidenced by atomic absorption spectroscopy.
The effect of the substitution of Asp-120(81) with Ala or Glu on steady-state kinetics was examined using four ␤-lactams as substrates. In Table I, D120(81)A and D120(81)E showed a remarkable reduction in activity, i.e. a decrease in k cat by 10 2 -10 4 -fold and an increase in K m by severalfold to 10 2 -fold, although the effect was dependent on the ␤-lactam used. Haruta et al. (41) reported that the substitutions of His-116(77), His-118(79), or His-196(139), which act as Zn(II) ligands, with Ala caused a decrease in IMP-1 activity compared with that of WT, due to the loss of Zn(II) ion. However, the activity of the IMP-1 mutants, D120(81)A and D120(81)E, was not recovered, even in the presence of an excess of Zn(II) ion. These results clearly indicate that Asp-120 in the MBLs is important for the hydrolysis of ␤-lactams, including those that are in widespread clinical use. The increased values of K m of the IMP-1 mutants also indicate that Asp-120(81) plays an important role in substrate binding. The strong binding of nitrocefin by WT, D120A(81), and D120E(81), as indicated by the small K m values, is probably due to the dinitrophenyl substituent of nitrocefin, which is oriented in a local hydrophobic environment of IMP-1.
Wang et al. (29,30,47) have shown that an intermediate is produced during the course of the hydrolysis of nitrocefin by the metallo-␤-lactamase from B. fragilis (CcrA) based on the observation of the transient appearance of an absorption band around 665 nm. Essentially the same change was observed for a metallo-␤-lactamase from S. maltophilia (L1) (35). Moali et al. (44) reported that this intermediate does not accumulate in the case of IMP-1. We carried out a pre-steady-state kinetic study using nitrocefin as a substrate, to determine whether an intermediate could be detected under equimolar conditions for D120(81)A and D120(81)E. However, the transient increase in the absorption at 650 nm was not observed for WT or the IMP-1 mutants. Kaminskaia et al. (42) reported that a di-Zn(II) complex, which was a model of metallo-␤-lactamase, showed an "anionic intermediate" of nitrocefin at 640 nm in 80% Me 2 SO. The pre-steady-state experiments for WT and the IMP-1 mutants were again repeated with reaction medium containing up to 80% glycerin, but no indication of a transient absorption at 650 nm was found. From these results, the rate-determining step for the IMP-catalyzed reaction of nitrocefin appears to proceed without the significant accumulation of an intermediate having an absorption maximum around 650 nm. This suggests that Asp-120(81) is not the proton donor to the anionic intermediate, and no accumulation of intermediate occurs because of the ease of proton transfer to the presumed intermediate or the operation of other mechanisms for IMP-1.
Pre-steady-state experiments of nitrocefin with D120(81)E gave strange results. The two curves (391 and 491 nm, Fig. 3a) were not synchronized, and a new band having a max at 460 nm appeared. This band was rather stable in an experiment using equimolar conditions (Fig. 3b). Bicknell et al. (45) reported that an intermediate having an absorption maximum at 450 nm could be detected under the conditions of temperatures lower than Ϫ58°C with a BcII enzyme. The species having an absorption maxima at 460 nm found in our study is probably identical to those reported by Bicknell et al. (45). The visible absorption of the nitrocefin hydrolysis product originates from the conjugated system that includes a dinitrophenyl substituent. A shorter wavelength than 491 nm would arise from a reduction in conjugation, which is supported by the spectroscopic titration of the nitrocefin hydrolysis product with hydrochloric acid. The species with an absorption maximum at 460 nm is probably formed by the binding of the nitrocefin hydrolysis product to D120(81)E and is not an intermediate in the reaction. This behavior was observed only for D120(81)E. One of the reasons for this is thought to be the shift in the location and orientation of the carboxylate side chain of Glu-120(81) from that of the Asp-120(81).
To probe the role of Asp-120(81) in the mechanism of hydrolysis of IMP-1, the pH dependence of hydrolysis was examined using nitrocefin as a substrate. We reported previously that IMP-1 is inactivated in acidic medium as the result of the dissociation of Zn(II) from the holoenzyme (34). Therefore, prior to the examination of pH dependence of the IMP-1 mutants, the K d was obtained as reported previously, and these values are similar to those reported for WT (Table II) (34). The pH dependence of the hydrolysis for the IMP-1 mutants was then measured under conditions of excess Zn(II) by the method reported previously (34). The K m and k cat values for WT and D120(81)A were almost constant between pH 4.6 and 9.0, but the k cat for D120(81)E increased by 10 2 -fold when the pH was increased from 5.2 to 9.0 (Fig. 2). These results suggest that Asp-120(81) is not a factor in decreasing the pK a for the water bridging two Zn(II) ions, because the substitution of Asp-120(81) with Ala do not show any pH dependence between pH 4.6 and 9.0. It is likely that the positive charges of the two Zn(II) ions are the main cause for lowering the pK a for the water bridging two Zn(II) ions. These results are supported by pH dependence studies of the Asp-mutants, D120(103)C, of CcrA from B. fragilis reported by Yanchak et al. (31) and D120(88)C and D120(88)S of L1 from S. maltophilia reported by Garrity et al. (43); a water bridging two Zn(II) ions has a pK a Ͻ 5.0. The logarithm of k cat of D120(81)E decreased with decreasing pH from 5.2 to 4.6. This indicates the possibility of the participation of an acid-base group with pK a values between pH 4.6 and 5.2, but we were unable to identify which residue is responsible for this pK a value at present.
To investigate the reasons for why the kinetics studies of the IMP-1 mutants showed strange results, crystals of the IMP-1 mutants were prepared, and x-ray crystallography was carried out. The overall structure was conserved despite the mutations of Asp-120(81), but the local structure at the active center was changed (Fig. 4). The flaps, residues 62(26)-65(29) of the IMP-1 mutants, were also disordered in the crystal structures, similar to WT. This is probably due to the high flexibility of the flaps. C-␣ atoms of the IMP-1 mutants were superimposed on that of WT, and each r.m.s.d. for C-␣ atoms was investigated. The residues moved more than 1 Å in r.m.s.d., Asp-49(13), Glu-50 (14), Val-61(25), Pro-68(32), Val-76(40), Asp-77(41), Lys-186(129), Ile-187(130), and Glu-210(150) positioned on hairpin loops exposed to solvent and residues, and Glu-296(219) and Ser-297(220) positioned at the C terminus. Most interestingly, the C-␣ atoms of  in the active sites of both structures of D120(81)A and D120(81)E were moved 1.2 and 1.3 Å away from the zinc-binding site compared with those in WT, respectively (Fig. 5). Moreover the Zn-Zn distance of D120(81)E is also longer by 0.3 Å than that of WT (Table IV).  is conserved in the active sites of all known MBLs and is the residue that coordinates to Zn2. Suá rez et al. (33) and Oelschlaeger et al. (46) reported that the Zn-Zn distance could possibly increase to 4.5-5.0 and 4.3-5.2 Å by using CcrA and IMP-1 in the substrate complex as a model in molecular dynamics simulations, respectively. This suggests that His-263(197) is flexible, to some extent, and this flexibility may cause an increase in the Zn-Zn distance.
The active site in D120(81)A was compared with that of WT (Fig. 4b). The active site in WT contains two Zn(II) ions, but the active site in D120(81)A contains only one Zn(II) ion, despite the fact that the Zn(II) content, as determined by atomic absorption spectroscopy, was two per protein molecule. This is probably due to the removal of Zn2 by chelation by the excess acetate ion present, because of the length of time required for crystal growth and the oxidation of Cys-221(158) to SO 2 , resulting in a modified Cys-221(158) and Csw221(158), buried in the Zn2 coordination sphere. Zn1 was coordinated by an acetate ion acting as a bidentate ligand and three His(s) and was not coordinated by H 2 O. These results are likely due to the increased positive charge because of the substitution of Asp with Ala at the 120 position (81 residue). This increase in positive charge could be responsible for the inhibition of D120(81)A by phosphate ion. D120(81)E has two Zn(II) ions at the active site, as does WT, but its coordination environment is quite different from that of WT (Fig. 4c). It is assumed that an oxygen atom of Asp-120(81)OD1 of WT is not coordinated to Zn2 but binds to the bridging OH 2 or OH Ϫ by a hydrogen bond, similar to CcrA (Fig.  1) (14,28). In D120(81)E, one oxygen atom (Glu-120(81)OE1) would be able to bridge both Zn1 and Zn2 because the distances of Zn1-Glu-120(81)OE1 and Zn2-Glu-120(81)OE1 are 2.5 Ϯ 0.6 and 2.5 Ϯ 0.3 Å, respectively (Table IV). The Zn-Zn distance of D120(81)E (3.6 Ϯ 0.1 Å) is longer by 0.3 Å than that of WT (3.3 Å). This indicates that the dissociation of H ϩ from OH 2 is suppressed in D120(81)E because OH 2 would not bridge the two Zn(II) ions but was disordered in the structure, and the positive charges of the two Zn(II) ions would be dispersed among the two coordinated oxygen atoms of Glu. This type of coordination is in agreement with the pH dependence for hydrolysis for the reaction catalyzed by D120(81)E, i.e. the pK a of the Zn1-bound H 2 O is increased since it is not bridged. This decrease in positive charge in the active site of D120(81)E is also in agreement with the binding of the nitrocefin hydrolysis product to D120(81)E. D120(81)A was strongly inhibited by phosphate ion with nitrocefin as a reporter substrate, as shown in Fig. 6. This inhibition was weak for WT. The substitution of Asp with Ala changes the local charge in the active site, and a plausible cause for this is the increase in positive charge around the substrate-binding site and is in agreement with the x-ray structure of D120(81)A. The binding of phosphate ion to a Zn(II) ion or the catalytic active site would result in inhibition by lowering the extent of substrate binding. If this inhibition occurs through the binding of phosphate with the active site, the dissociation constants, assuming a competitive inhibition, are K i ϭ 144 and 0.32 mM for WT and D120(81)A, respectively. The effect of several anions on the inhibition of D120(81)A mutant was also evaluated. Whereas ClO 4 Ϫ and SO 4 2Ϫ had no effect, [Co(CN) 6 ] 3Ϫ and [Fe(CN) 6 ] 4Ϫ showed K i values of 1.1 and 4.0 mM, respectively. Moreover, when 100 M Zn(NO 3 ) 2 was added to the reaction buffer, the inhibition of D120(81)A by phosphate ion was suppressed. From these results, strong inhibition of D120(81)A by phosphate ion occurred after the removal of the Zn(II) ion from the active site by chelation. The anion would bind to the active site, since the local positive electric charge of the active site in D120(81)A is increased. CONCLUSION The role of Asp-120 in IMP-1, which is a conserved amino acid in the active site of all MBLs, was investigated by the preparation of some Asp-120-mutants, kinetic studies, and xray crystallographies. The following four points can be made. (i) Asp-120 in IMP-1 plays the important roles in catalysis and the substrate recognition, as evidenced by steady-state kinetics studies. (ii) Asp-120 in IMP-1 does not act as a proton donor to the anion intermediate of nitrocefin and a proton shuttle, as evidenced by pH dependence studies of the hydrolysis. (iii) The hydrolytic mechanism for IMP-1 is different from those for CcrA and L1, and one of main reasons is the ease of proton transfer because there is no accumulation of the anion intermediate of nitrocefin, as evidenced by pre-steady-state kinetics studies using IMP-1 (44) and the IMP-1 mutants. (iv) From the results of x-ray crystallography, the structure of D120(81)E (the oxygen atom of Glu-120(81)OE1 is presumably not coordinated to bridging H 2 O or OH Ϫ ), the oxygen atom of the carboxylate of Asp-120 is essential to coordinate the bridging H 2 O or Asp-120 Mutants of Metallo-␤-lactamase IMP-1