The Metallo-β-lactamase GOB Is a Mono-Zn(II) Enzyme with a Novel Active Site*

Metallo-β-lactamases (MβLs) are zinc-dependent enzymes able to hydrolyze and inactivate most β-lactam antibiotics. The large diversity of active site structures and metal content among MβLs from different sources has limited the design of a pan-MβL inhibitor. Here we report the biochemical and biophysical characterization of a novel MβL, GOB-18, from a clinical isolate of a Gram-negative opportunistic pathogen, Elizabethkingia meningoseptica. Different spectroscopic techniques, three-dimensional modeling, and mutagenesis experiments, reveal that the Zn(II) ion is bound to Asp120, His121, His263, and a solvent molecule, i.e. in the canonical Zn2 site of dinuclear MβLs. Contrasting all other related MβLs, GOB-18 is fully active against a broad range of β-lactam substrates using a single Zn(II) ion in this site. These data further enlarge the structural diversity of MβLs.

The expression of ␤-lactam degrading enzymes (␤-lactamases) is the most common mechanism of antibiotic resistance among bacteria (1,2). These enzymes have been grouped into four classes (A-D) according to sequence homology (3,4). Class A, C, and D enzymes use an active site serine residue as a nucleophile, whereas class B lactamases (generically termed metallo-␤-lactamases, M␤Ls) 9 employ one or two Zn(II) ions to cleave the ␤-lactam ring.
Molecular structures of M␤Ls from the three subclasses have been solved by x-ray crystallography (12,14,15,25,31). Comparison of the tertiary structure of enzymes belonging to the different subclasses reveals a common ␣␤/␤␣ sandwich fold, in which different insertions and deletions have resulted in different loop topologies and, ultimately, in different zinc coordination environments and metal site occupancies among B1, B2, and B3 enzymes (Fig. 1). M␤Ls bind up to two metal ions in their active sites. In B1 and B3 enzymes, one Zn(II) ion (Zn1) is tetrahedrally coordinated to three histidine ligands (His 116 , His 118 , and His 196 in Fig. 1) and a water/OH Ϫ molecule (3-H site) (14,15,28). The coordination polyhedron of Zn2 in B1 enzymes is provided by Asp 120 , Cys 221 , His 263 , and one or two-water molecules (DCH site). Notably, this site represents the active species in mono-Zn(II) B2 enzymes (25). Instead, two mutations (Cys 221 3 Ser and Arg 121 3 His) affect the Zn2 coordination geometry in B3 M␤Ls, and the metal ion is bound to Asp 120 , His 121 , His 263 , and one-or two-water molecules, while Ser 221 is no longer a metal ligand (DHH site) (28). A remarkable exception is provided by the deepest branching member of the M␤L B3 subclass, GOB from E. meningoseptica (4). In all reported GOB sequences, His 116 and Ser 221 are substituted by Gln and Met, respectively, suggesting the presence of an unusually perturbed metal binding site.
Here we show that E. meningoseptica GOB-18 represents a novel type of broad spectrum M␤L unique in being maximally active in the mono-Zn(II) form and in which the metal ion occupies only the Zn2, DHH site. This contrasts with two generally accepted ideas: broad spectrum M␤Ls are maximally active in the dinuclear form (B1 and B3 enzymes), and mono-Zn(II) enzymes are carbapenemases (B2 enzymes). Finally, the findings presented here confirm that Zn2 is central for M␤Lmediated catalysis and that the attacking nucleophile could be provided either by a non-metal center or by the Zn2 site. This claims for a revisited strategy for the design of broad spectrum M␤L inhibitors.

EXPERIMENTAL PROCEDURES
Source of GOB-18 Coding Sequence-An E. meningoseptica clinical strain from the Hospital Clemente Alvarez (Rosario, Argentina), identified by API 20NE (bioMerieux, Marcy l'Etoile, France), was used as the source of the genomic DNA for cloning the GOB-18 coding gene.
DNA Cloning and Construction of the Expression Vector for GOB-18-Genomic DNA from the E. meningoseptica clinical strain used here was isolated essentially as described in ref (35). The complete coding sequence of the gob-18 gene was amplified by employing primers 1 and 2 described in Ref. 32. The DNA fragment was sequenced and two new primers were designed (see supplemental material) to clone the mature GOB-18 coding sequence cloned into BamHI-Hin-dIII sites of pETGEXTerm vector (13) in frame to the 3Ј-end of Schistosoma japonicus glutathione S-transferase gene, for expression purposes. The resulting recombinant pET-GOB-18 plasmid produces GOB-18 as a C-terminal fusion to glutathione S-transferase. The nucleotide sequence reported in this paper was assigned accession number DQ004496 in the combined EMBL/GenBank TM /DDBJ sequence data base.
Site-directed Mutagenesis-Sitedirected mutagenesis was achieved using a rapid PCR-based method with modifications (36 and see supplemental material).
Biochemical Characterization of Wild-type GOB-18 and Mutants-Size exclusion chromatography was done on a Superdex 200 HR 10/30 column (Amersham Biosciences). The molecular mass of purified GOB-18 was measured by mass spectrometry/electrospray using an LCQ Duo Ion Trap mass spectrometer at the mass spectrometry facility LANAIS-PRO, University of Buenos Aires. Circular Dichroism spectra of protein samples in 10 mM Tris-HCl, pH 7.5, and 50 mM NaCl were measured at 25°C, using a Jasco J-715 spectropolarimeter flushed with N 2 .
The GOB-18 and Gln 116 3 His GOB-18 apoproteins were prepared by treating samples ϳ0.  efflux and addition of 1 mM sodium dithionite. The reduced state of the metal ion was verified colorimetrically using o-phenanthroline. The Fe(III)-GOB-18 derivative was prepared taking advantage of Zn(II) dissociation at acidic pH values retaining the iron content. Holo-GOB-18 was dialyzed against 100 volumes of polybuffer TAMS (50 mM Tris, 50 mM sodium acetate, 50 mM MES, and 500 mM NaCl) adjusted to pH 4.5 (twice). Then pH was raised by dialysis against 100 volumes of 10 mM Tris-HCl, pH 7, and 50 mM NaCl (twice).
Metal Content Determination-The metal content of the purified enzymes was measured by inductively coupled plasmaatomic emission spectrometer or by atomic absorption spectroscopy in a Metrolab 250 instrument operating in the flame mode.
Determination of IC 50 Value-The concentration required to effect 50% inhibition of enzyme activity (IC 50 ) was determined by preincubating GOB-18 (2 nM) in 15 mM Hepes, pH 7.5, 200 mM NaCl, 30°C, with increasing EDTA concentrations for 1 min, prior to the initiation of the assay by the addition of 1 mM cefotaxime. A plot of steady-state hydrolysis rate versus inhibitor concentration provided the basis for the assessment of the IC 50 value.
Determination of the Kinetic Parameters-The hydrolysis of the antibiotics was monitored by following the absorbance variation resulting from the hydrolysis of the ␤-lactam ring. The reaction medium employed was 15 mM Hepes, pH 7.5, 0.2 M NaCl. All measurements were performed at 30°C. The kinetic parameters K m and k cat were derived from initial rate measurements, recorded on a Jasco V-550 spectrophotometer, and were estimated by nonlinear data fitting to the integrated form of the Michaelis-Menten equation.
pH Dependence-The pH dependence of GOB-18-mediated cefotaxime hydrolysis was determined by performing measurements in the polybuffer TAMS adjusted from pH 4.0 to pH 8.0 with increments of 0.5 pH units, at 1 mM substrate concentration (see supplemental material).
X-ray Absorption Spectroscopy-Samples of GOB-18 (ϳ1 mM) were prepared with 20% (v/v) glycerol, and loaded in Lucite cuvettes with 6-m polypropylene windows, before rapid freezing in liquid nitrogen. X-ray absorption spectra were measured at the National Synchrotron Light Source (Brookhaven National Laboratory, Upton, NY), beamline X9B, with a Si(111) double crystal monochromator; harmonic rejection was accomplished using a nickel focusing mirror. Data collection and reduction were accomplished according to published procedures (40). XAS data were collected on two independently isolated samples. As each data set gave similar results, the spectra were averaged; the data in Fig. 3 represent the average of the two data sets (12 scans total).
Fourier filtered EXAFS data (⌬k ϭ 1-13 Å Ϫ1 ; ⌬r ϭ 0.5-2.1 Å, first shell or 0.1-4.5 Å for multiple scattering fits) were fit utilizing theoretical amplitude and phase functions calculated with FEFF v.8.00 (41). The zinc-nitrogen scale factor and the threshold energy, ⌬E 0 , were calibrated to the experimental spectrum for tetrakis-1-methylimidazole zinc(II) perchlorate, Zn(MeIm) 4 (42), and held fixed (at 0.78 and Ϫ21 eV, respectively), as were similarly determined values for iron (0.78 and Ϫ21 eV), in all subsequent fits to the data for GOB-18. First shell fits were then obtained for all reasonable coordination numbers, including mixed nitrogen/oxygen ligation, while allowing the absorber-scatterer distance, R as , and the Debye-Waller factor, as 2 , to vary; the best fits are presented in the supporting information. Multiple scattering contributions from histidine ligands were fit according to published procedures. Metal-metal (zinc-iron and iron-zinc) scattering was modeled with reference to the experimental EXAFS of Zn 2 (salpn) 2 and Fe 2 (salpn) 2 .
NMR-NMR spectra were recorded on a Bruker Avance II 600 spectrometer operating at 600.13 MHz at different temperatures, as indicated. 1 H NMR spectra were recorded under conditions to optimize detection of the fast relaxing paramagnetic resonances, either using the superWEFT pulse sequence or water presaturation. Spectra were acquired over large spectral widths with acquisition times ranging from 16 to 80 ms and intermediate delays from 2 to 35 ms. One-dimensional experiments with solvent presaturation were used to record isotropically shifted signals closer to the diamagnetic envelope.
EPR-EPR was recorded at 9.63 GHz with 2-milliwatt microwave power and 5-G (0.5 mT) magnetic field modulation at 100 kHz, using a Bruker EleXsys E500 spectrometer equipped with an ER4116DM cavity operating in perpendicular mode and an Oxford Instruments ESR900 helium flow cryostat and ITC503 temperature controller.

RESULTS
Biochemical Characterization of GOB-18-The gene coding for a GOB-type M␤L was cloned from a carpabenem-resistant E. meningoseptica clinical strain. Sequence analysis indicated a predicted molecular mass of 31.4 kDa for the encoded protein.
At the time of sequencing there were 17 known variants of GOB enzymes. This protein differed from those reported previously and was named GOB-18. This enzyme differed from the firstly reported GOB enzyme, GOB-1 (32), by three conserved mutations: Leu 94 3 Phe, Ala 137 3 Val, and Asp 282 3 Asn (standard consensus numbering) (5).
Recombinant GOB-18 was overproduced as a fusion to GST in the cytoplasm of E. coli BL21(DE3) pLysS cells, cleaved, and purified to homogeneity. GOB-18 is a monomeric enzyme according to size exclusion chromatography. Mass spectrometry confirmed the molecular mass expected from the gene sequence. The ␤-lactamase activity (measured as cefotaxime hydrolysis) was inhibited by EDTA with an IC 50 of 2.3 Ϯ 0.3 mM, indicating that GOB-18 holds a tightly bound divalent cation essential for activity. Inductively coupled plasma and atomic absorption analyses showed that recombinant GOB-18 contained significant amounts of zinc and, notably, iron. Although the relative amounts of these two metals varied among different enzyme preparations (0.45-0.75 iron/GOB-18 and 0.01-0.20 zinc/ GOB-18), the total metal content never exceeded one metal ion per protein molecule. Addition of excess Zn(II) did not alter the CD spectrum of GOB-18 as isolated, neither in the near nor in the far UV (data not shown). Overall, these data suggest that GOB-18 is a mono-metallic enzyme, in sharp contrast to the other B3 M␤Ls.
The apoprotein was devoid of lactamase activity that could be recovered by addition of Zn(II). Remarkably, apo-GOB-18 remetallated with Zn(II) bound only one equivalent of Zn(II) and its activity largely exceeded that of the enzyme as isolated (Fig. 2). On the other hand, Fe(II)-GOB-18 and Fe(III)-GOB-18 displayed negligible activities compared with that of Zn(II)-GOB-18. In fact, Fe(III)-GOB-18 was poorly active even upon addition of Zn(II), suggesting that both metal ions compete for the same binding site. The kinetic parameters of Zn(II)-GOB-18, reported in Table 1, show that mono-Zn(II)-GOB-18 presents a broad substrate spectrum for ␤-lactams, similarly to that shown by dinuclear B3 enzymes. Addition of 20 M Zn(II) to the reaction medium did not modify the enzyme kinetic parameters.
The pH dependence of GOB-18-mediated cefotaxime hydrolysis showed a plateau between pH values of 6 and 8, and an acidic limb, resulting in a complete inactivation at pH Յ 4.5. This loss of activity could be reverted by raising the pH to 6.5, indicating that the inactivation is reversible. Metal content analysis at different pH values showed that the decrease in activity at low pH paralleled the dissociation of Zn(II) from the enzyme, despite the fact that most of the Fe(III) remained bound up to pH 4 (see supplemental material).
Overall, these data show that Zn(II)-GOB-18 is the active ␤-lactamase species. GOB-18 preparations with larger zinc/ iron ratios were the most active ones, in line with this observation (see supplemental material). Expression of the complete GOB-18 gene (including a transit peptide sequence) in E. coli results in an enzyme secreted to the periplasm confering ␤-lac-tam resistance to the bacterial host. Isoelectrofocusing analysis of the periplasmic fraction after osmotic shock revealed that the enzyme is present exclusively in the Zn(II) form, suggesting that iron uptake could be due to an artifact of protein overexpression in E. coli cytoplasm.
X-ray Absorption Spectroscopy-EXAFS data recorded at the zinc and iron edges, k 3 ((k), and the corresponding Fourier transforms for GOB-18 are shown in Fig. 3. EXAFS curve-fitting results are presented as supporting information. Fits to Fourier-filtered first shell scattering (0.5-2.1 Å) in the case of Zn(II)-GOB-18 suggest coordination to 4 nitrogen/oxygen donors at a distance of 2.01 Å (Fit Zn-1). Fit residuals did not improve significantly upon inclusion of a sulfur, or a mixed shell of nitrogen/oxygen. Multiple scattering analysis indicates ligation to 2 histidine residues (Fit Zn-2). The low Debye-Waller factor for the second multiple scattering path (C 1 -N 1 ) most likely reflects some rotation of the imidazole plane about the axis normal to the zinc-nitrogen bond. Inclusion of a single zinc-carbon scattering interaction at a distance of 2.71 Å, representing carboxylate carbon scattering, modestly improved the fit (ϳ15%) (Fit Zn-3).
In the case of Fe(III)-GOB-18, single shell fits give an average of 5 nitrogen/oxygen scatterers at 2.11 Å (Fit Fe-1). Inclusion of a sulfur atom increased the fit residual. The larger metal-ligand distance is consistent with a higher coordination number for iron relative to zinc. Separate shells of 3 oxygen atoms at 2.04 Å and 2 nitrogen atoms at 2.20 Å could be resolved (Fit Fe-2), and multiple scattering fits indicate the presence of two His ligands (Fit Fe-3). Inclusion of an iron scatterer in fits for Zn(II)-GOB-18 (Fit Zn-4) and a zinc scatterer in fits for Fe(III)-GOB-18 (Fit Fe-3) slightly improved the fits but led to different metal-metal distances (3.61 and 3.55 Å, respectively). This allows us to discard the existence of a heterodimetallic site in GOB-18. Thus, EXAFS data, together with the biochemical studies, suggest that Zn(II) and Fe(III) compete for the only metal binding site in GOB-18. Based on this, we decided to exploit Fe(III) as a spectroscopic probe of the metal site.
Spectroscopic Characterization of Fe(III)-GOB-18-The UVvisible spectrum of Fe(III)-GOB-18 revealed an absorption band centered at 375 nm, which was not present in the apoenzyme nor in the Zn(II)-reconstituted form (Fig. 4A). This feature can be attributed to a typical His-Fe(III) charge transfer band, similar to those observed in lipoxygenase, iron superoxide dismutase, and other non-heme, non-Fe-S iron proteins (43). EPR spectra of GOB-18 recorded at 6 K and 25 K (Fig. 4B) indicated an isolated, "rhombic" (i.e. mean E/D ϳ 1 / 3 ) Fe(III)

TABLE 1 Kinetic parameters for the hydrolysis of different ␤-lactam substrates by fully loaded wild type Zn(II)-GOB-18 and Zn(II)-Gln 116 3 His GOB-18 mutant
Kinetic parameters derived from a nonlinear fit of Michaelis-Menten equation to initial rate measurements. Reported values are the average of at least three independent enzyme preparations. ND, not determined. ion. The almost complete loss of the M S ϭ ͉Ϯ1/2͘ (or M S ϭ ͉Ϯ5/2͘) signals at 25 K suggests a zero field splitting, ⌬, of less than about 30 cm Ϫ1 . The asymmetry of the lines at g eff ϳ 4.3 and at g eff ϳ 9 indicates significant strain in E/D, although the shoulder on the crossover at g eff ϳ 4.3 (Fig. 4B, inset) indicates a lower degree of strain than in the often-seen "adventitious iron" signal and provides strong evidence for the Fe(III) being tightly protein-bound. In contrast to the iron-loaded glyoxylase II (44), no signals attributable to an Fe(III)-Fe(II) coupled system were observed from GOB-18. 1 H NMR spectra recorded under conditions that allow the detection of paramagnetic signals revealed a set of isotropically shifted resonances spanning from 100 to 20 ppm (Fig. 4C): a broad feature (A), accounting for three resonances centered at 80 ppm, a signal of fractional intensity at 38 ppm (B), and a two-proton resonance at 30 ppm (C). When the spectrum was recorded in D 2 O, a loss of intensity corresponding to two protons was noticed in the broad envelope A, revealing the presence of two exchangeable resonances. The broad nature of these resonances is comparable with that reported for mononuclear Fe(III) enzymes such as protocatechuate dioxygenase (45), allowing us to discard the presence of a coupled, di-iron site. The two exchangeable resonances located at ϳ80 ppm can be attributed to two His ligands, in agreement with the analysis of the EXAFS data. The broad nature of these signals prevented us to perform one-dimensional or two-dimensional nuclear Overhauser effect spectra to assign them.

Substrate
Altogether, EPR, NMR, and EXAFS data allow us to rule out the formation of a dinuclear Fe(III) site in GOB-18, supporting the above assumption that Fe(III) and Zn(II) bind to the same site. EXAFS and NMR reveal the presence of two His ligands at the metal site, with five nitrogen/oxygen ligands for Fe(III). The 1 H NMR spectrum shows only 5-6 resonances corresponding to metal ligands, two of them corresponding to imidazolic His NHs, suggesting that one or two water molecules would be needed to achieve a penta-coordinated site.
This picture can be accounted for by two possibilities: (i) that the metal ion binds to a modified Site 1 in GOB-18, in which His116 has been replaced by a Gln residue, with His 118 , His 196 , and a H 2 O/OH Ϫ moiety completing the coordination sphere or (ii) by placing the metal ion into Site 2 (DHH ligand set), with Asp 120 , His 121 , His 263 , and a H 2 O/OH Ϫ as metal ligands.
Three-dimensional Modeling-Three-dimensional models for Zn(II)-GOB-18 were built by homology modeling from the two available x-ray structures of B3 M␤Ls: FEZ-1 and L1 (38 and 21% sequence identity, respectively), including metal-ligand constraints based on spectroscopic data. We examined the possibility of binding a single Zn(II) ion to each of the two canonical metal binding sites. Best GOB-18 models for both mono-zinc active site conformations show low backbone root mean square deviation compared with FEZ-1 and L1 (0.6 and 1.5 Å, respectively), suggesting that the protein fold can accommodate a single Zn(II) ion in either binding site. Quantum chemical geometry optimizations based on density functional theory on unconstrained models of these metal sites with water or hydroxide as the fourth ligand suggested a regular tetrahedral geometry for a Zn(II) ion at a modified Site 1 and a distorted tetrahedron for the metal ion bound to Site 2. Predicted averaged zinc-nitrogen/oxygen distances are within Ϯ0.04 Å from the values retrieved from EXAFS data, preventing us from discerning between the two sites (see supplemental material for additional "Experimental Procedures").
Site-directed Mutagenesis of Putative Metal Ligands-To unequivocally identify the metal binding site in GOB-18, we constructed mutants of possible metal ligands. In one of them  Table S1. Asp 120 , which is conserved in all M␤Ls as a Zn2 ligand, was changed to Ser; and Gln 116 , distinctive of GOB enzymes, was replaced by a His. The Gln 116 3 His mutant, in particular, would be expected to provide a 3-H binding site similar to that found in all B1 and in most B3 M␤Ls (12,15,28,31).
Gln 116 3 His GOB-18 was purified as a mixture of zinc and iron forms, typically with 0.06 -0.70 zinc/protein and 0.20 -0.60 iron/protein. Remarkably, the total metal content of this mutant never exceeded one equivalent per protein, as determined for wild-type GOB-18. Apo-Gln 116 3 His GOB-18 remetallated with Zn(II) bound only one equivalent of Zn(II). The kinetic parameters of zinc-Gln 116 3 His GOB-18 showed a partially decreased activity when compared with that of wildtype GOB-18 ( Table 1). Addition of 20 M Zn(II) to the reaction medium gave rise to minor changes in the mutant enzyme activity (data not shown). UV-visible and EXAFS data at the iron edge of holo-Gln 116 3 His GOB-18 were identical to those recorded for the wild-type enzyme. Similarly, the paramagnetic 1 H NMR spectrum of this mutant showed minor perturbations in the Fe(III) ligand signals when compared with that of the wild-type enzyme (Fig. 4C). The fact that the metal site is largely unperturbed suggests that residue 116 is not a metal ligand neither in wild-type GOB-18 nor in Gln 116 3 His GOB-18.
Asp 120 3 Ser GOB-18 displayed a markedly decreased metal binding ability for both iron and zinc. The iron binding capacity of Asp 120 3 Ser GOB-18 was abolished, and the Zn(II) content reached at most 0.45 Zn(II)/protein (even after exhaustive dialysis against metal containing buffer). Asp 120 3 Ser GOB-18 was inactive toward different assayed antibiotics. These data reveal that Asp 120 is a metal ligand, being also essential for catalytic efficiency.
B1 and B2 M␤Ls have a Cys ligand in the Zn2 site (15,25). The present three-dimensional model suggests that the only Cys residue present in GOB-18 is far away from the active site. This Cys residue is only accessible to 5,5Ј-dithiobis-(2-nitrobenzoic acid) under denaturing conditions. To definitively discard the possibility that this Cys residue may serve as a metal ligand in GOB-18, we constructed a Cys 201 3 Ser mutant. The metal content, catalytic efficiency, and spectroscopic data of Cys 201 3 Ser GOB-18 were similar to those of the wild-type enzyme, indicating the absence of a Cys ligand in the metal binding site, in agreement with the present three-dimensional model and EXAFS data.

DISCUSSION
GOB Enzymes Are Monometallic Lactamases-The biochemical, spectroscopic, and mutagenesis data herein shown allow us to propose that GOB-18 is a mono-metallic M␤L, with the catalytic Zn(II) ion bound to Asp 120 , His 121 , His 263 , i.e. the canonical Zn2 site (DHH ligand set in subclass B3) (Fig. 5). The coordination sphere would be completed by one or two solvent molecules in the Zn(II) and Fe(III) variants, respectively. The presence of a naturally occurring Gln residue rather than a His at position 116 in GOB-18 resembles the situation found for subclass-B2 M␤Ls, in which an Asn is found (25,27). In both cases, the absence of the otherwise conserved His 116 ligand would induce the loss of the Zn1 site. However, engineering a putative 3-H site in GOB-18 does not lead to a dimetallic enzyme, suggesting additional structural differences in the metal site of GOB-18.
GOB enzymes differ by a few point mutations, apparently far from the active site. This allows us to confidently extend our results to all GOB lactamases. GOB M␤Ls thus present features of B2 (being Zn2-only enzymes) and B3 (based on sequence homology) lactamases. However, two major differences of GOB M␤Ls with B2 enzymes should be noted: (i) GOB is active even in the presence of 0.1 mM Zn(II), while B2 enzymes are inhibited upon binding of a second Zn(II) equivalent and; (ii) B2 enzymes are selective carbapenemases, while GOB is a broad spectrum M␤L. Thus, GOB M␤Ls are unique in being the only broad spectrum mono-Zn(II) ␤-lactam hydrolases and in presenting active site mutations within one subclass.
Toward a General Catalytic Mechanism for Metallo-␤-Lactamases-The effect of the Zn(II) content on the enzymatic activity in M␤Ls is still a controversial issue. The Zn1 (3-H) site present in all B1 and in most B3 enzymes, GOB M␤Ls being the only exception, has been regarded as the catalytic site providing the attacking zinc-bound nucleophile. As inferred from a series of elegant spectroscopic studies, the Zn2 site (DCH ligand set in B1 enzymes and DHH ligands in B3 lactamases) contributes to ␤-lactam hydrolysis by binding to the bridging ␤-lactam nitrogen (17,18,46). The trapping of hydrolyzed moxalactam in the active site of S. maltophilia L1 recently confirmed that both Zn1 and Zn2 are implicated in substrate binding and catalysis in B3 M␤Ls (47).
The structure of both the free and product-complexed form of the B2 enzyme CphA revealed only one catalytic Zn(II) ion bound to the DCH site (25). This metal ion is also involved in stabilization of the negative charge on the bridging nitrogen atom of the ␤-lactam ring, as in B1 and B3 enzymes. Thus, the Zn2 site would play the same role in all M␤Ls, i.e. the stabilization of the negative charge developed at the ␤-lactam nitrogen to favor cleavage of the C-N bond, and the proton delivery through a water molecule bound to it. This is further stressed by the fact that in all studied cases, the rate-determining step is the C-N bond cleavage (that involves the Zn2 ion) (16,17,46,48). This suggests that a strategy aimed for the design of a general inhibitor could be pursued by targeting the Zn2 site.
The issue that remains unsolved is the identification of the nucleophile in Zn2-only enzymes. The absence of the Zn1 site in B2 lactamases has led to the proposal of a non-metal activated nucleophile, a hypothesis later supported by theoretical and enzymological studies (48,49). However, the possibility of a Zn2-bound nucleophile cannot been definitively ruled out. The results herein presented reveal that Zn2-only M␤Ls are not only restricted to B2 subclass and that they can display a broad substrate spectrum. Therefore, a general mechanism of Zn2-only enzymes needs to be definitively elucidated and compared with those of dinuclear and Zn1only enzymes.
Several reports have indicated that the mono-Zn(II) variants of B1 and B3 enzymes are active (50,51). Since the first solved M␤L structure revealed one metal ion bound to the 3-H site (12), this has been regarded as the catalytic center in monometallic variants. However, spectroscopic data have revealed that when BcII (a well studied B1 lactamases) binds one equivalent of metal ion, it is distributed between the canonical Zn1 and Zn2 sites (52,53). These data should be reanalyzed at the light of the present evidence, considering that active mono-Zn2 enzymes may also occur among B1 and B3 enzymes other than GOB.