1640
Biochemistry 2001, 40, 1640-1650
Familial Mutations and Zinc Stoichiometry Determine the Rate-Limiting Step of Nitrocefin Hydrolysis by Metallo-β-lactamase from Bacteroides fragilis† Walter Fast, Zhigang Wang,‡ and Stephen J. Benkovic* Department of Chemistry, The PennsylVania State UniVersity, 414 Wartik Laboratory, UniVersity Park, PennsylVania 16802 ReceiVed August 8, 2000; ReVised Manuscript ReceiVed NoVember 21, 2000
ABSTRACT:
The diverse members of the metallo-β-lactamase family are a growing clinical threat evolving under considerable selective pressure. The enzyme from Bacillus cereus differs from the Bacteroides fragilis enzyme in sequence, zinc stoichiometry, and mechanism. To chart the evolution of the more reactive B. fragilis enzyme, we have made changes in an active site cysteine residue as well as in zinc content to mimic that which occurs in the B. cereus enzyme. Specifically, by introducing a C104R mutation into the B. fragilis enzyme, binding of two zinc ions is maintained, but the kcat value for nitrocefin hydrolysis is decreased from 226 to 14 s-1. Removal of 1 equiv of zinc from this mutant further decreases kcat to 4.4 s-1. In both cases, the observed kcat closely approximates that found in the di- and monozinc forms of the B. cereus enzyme (12 and 6 s-1, respectively). Pre-steady-state stopped-flow studies using nitrocefin as a substrate indicate that these enzyme forms share a similar mechanism featuring an anionic intermediate but that the rate-limiting step changes from protonation of that species to the C-N bond cleavage leading to the intermediate. Overall, features that contribute 3.7 kcal/mol toward the acceleration of the C-N bond cleavage step have been uncovered although some of the total acceleration is masked in the steadystate by a change in rate-limiting step. These experiments illustrate one step in the evolution of a catalytic mechanism and, in a larger perspective, one step in the evolution of antibiotic resistance mechanisms.
Widespread use of β-lactam antibiotics, originally by β-lactam producing organisms and more recently by humans in clinical settings and in animal feed, provides an incredible selective pressure that drives the evolution of antibiotic resistance mechanisms in bacteria (1-3). While numerous methods of resistance have emerged including mutation of penicillin binding proteins, alteration of membrane permeability, and even production of efflux pumps, one of the major clinical concerns has been the evolution of widespectrum β-lactamases which hydrolyze the antibiotic to an inactive form (4, 5). Although the TEM-derived serine-based β-lactamases are currently considered the most clinically important among the extended-spectrum β-lactamases, metallo-β-lactamases (Class B or Class 3 lactamases) are of growing concern because they threaten rapid plasmidencoded dissemination, they have an extremely broad substrate profile, and they are not inhibited by clinically useful inhibitors of serine β-lactamases such as clavulanic acid. Excepting the monobactams, metallo-β-lactamases can hydrolyze members of virtually every β-lactam class (6). Although the kcat values for many of these substrates remain low (6, 7), the selective pressures on catalytic optimization promise the emergence of higher activity metallo-β-lactamases, a process already in progress. The dinuclear metallo† This work was supported in part by a National Institutes of Health grant (GM 56879-01) to S.J.B. and National Institutes of Health postdoctoral fellowships AI 10369 (W.F.) and GM 18061 (Z.W.). * To whom correspondence should be addressed. Phone: (814) 8652882. Fax: (814) 865-2973. E-mail:
[email protected]. ‡ Present address: Protein Science, Pharmacia Corporation, 7240267-411, 301 Henrietta Street, Kalamazoo, MI 49007.
β-lactamases are considered to be more highly evolved than their mononuclear counterparts and generally have higher kcat/KM values (7, 8). Understanding the differences between the diverse metallo-β-lactamase family members and how they are adapting can lead to more efficient pharmaceutical design and can shed light onto the larger issue of evolution of antibiotic resistance mechanisms. Serine-based β-lactamases are thought to be related to the bacterial transpeptidases (9). However, the origin of metalloβ-lactamase remains unclear, although an RββR structure suggests a gene duplication event (10), and structural homology to glyoxalase II has been noted, as well as sequence homology to several bacterial cyclase/dehydratases, aryl sulfatase, and PHNP (11-13). Refined crystal structures have been reported for four members of the metallo-βlactamase family: the enzymes from Bacteroides fragilis (14), Bacillus cereus (8, 15), Stenotrophomonas maltophilia (16), and Pseudomonas aeruginosa (17). These structures can be thought of as snapshots of a relatively young class of enzyme at different stages of maturation. In fact, allelic variants of several metallo-β-lactamases have been reported, further demonstrating the genetic diversity of this class (1821). In particular, Fabiane et al. have suggested that the metallo-β-lactamase from B. cereus, originally characterized as a monozinc enzyme, is an evolutionary intermediate between the monozinc lactamases and the more active dizinc metallo-β-lactamases such as that from B. fragilis (8). This proposal is consistent with the different selective pressures placed on these two species in a clinical setting. B. cereus infections are usually associated with food poisoning for which antibiotics are not recommended, but members of the
10.1021/bi001860v CCC: $20.00 © 2001 American Chemical Society Published on Web 01/17/2001
Kinetics of Mutant B. fragilis Metallo-β-lactamase Scheme 1
Biochemistry, Vol. 40, No. 6, 2001 1641 Table 1: Mutagenic Primers Used in This Study (5′ to 3′) primer
sequence
C104R For C104R Rev NcoI For BamHI Rev
TGGCACGGCGATCGTATTGGCGGA TCCGCCAATACGATCGCCGTGCCA CTGAGAGTGCACCCCATGGCACAGAAA GGAATACCGGGTAGGATCCTACAATTC
antibiotic resistance as conferred by the metallo-β-lactamases. EXPERIMENTAL PROCEDURES B. fragilis group can result in postsurgical infections and are avoided by prophylactic treatment with cephalosporins (22), which provides a driving force for optimizing enzyme activity. Considerable mechanistic work on the metallo-β-lactamase enzyme from B. cereus has provided a kinetic and chemical mechanism in which C-N bond cleavage of cephalosporins (23) and benzyl penicillin (24, 25) occurs at or after the ratelimiting step and requires protonation of the leaving lactam nitrogen (Scheme 1A). The more efficient dizinc lactamase from B. fragilis has overcome this kinetic barier: the C-N bond cleavage of nitrocefin (a chromogenic cephalosporin) occurs before the rate-limiting step and through a mechanism that, unlike the B. cereus mechanism, does not require protonation of the leaving lactam nitrogen before the bond is broken (Scheme 1B) (26). In fact, the C-N bond cleavage rate is about 375-fold higher than that of the B. cereus lactamase. These differences in rate and mechanism are somewhat surprising because the amino acid composition of the zinc ligands as well as other crucial active site residues remain identical, although the equivalents of bound zinc may vary depending on experimental conditions. Despite these similarities, clearly both structural and mechanistic differences exist between these two proteins. Determining which structural features are responsible for these improvements is of interest as these changes have resulted in a more efficient catalyst of antibiotic hydrolysis. Although it is tempting to assign mechanistic differences as arising from specific features observed in the known structures, this method can be misleading because there is only 34% sequence identity between the two enzymes and significant differences are present both in the active site and throughout the protein. A more rigorous analysis can be conducted by grafting structural features of the various family members onto a common scaffold and evaluating the specific mechanistic consequences of each introduction. The two metallo-β-lactamase family members compared in this work are from B. cereus and B. fragilis. The specific features compared are position 104,1 (Arg in B. cereus and Cys in B. fragilis) which lies beneath the dizinc center, and the presence (or absence) of a second equivalent of zinc bound at the active site. These two familial differences are explored within the framework of the B. fragilis enzyme to determine their effect on the kinetic sequence and aspects of the catalytic mechanism. This analysis bridges the structural and mechanistic data for these two lactamase enzymes and, in a broader sense, charts a small step in the evolution of multiple 1 The numbering used throughout is taken from ref 14. To avoid confusion, residues from the B. cereus enzyme are numbered according to the structurally corresponding B. fragilis residue.
Materials. Except where noted, all chemicals were obtained through Sigma and all DNA modifying enzymes were purchased from New England Biolabs. Zn(II) and Co(II) standards were diluted from atomic absorption standard solutions into buffered solutions as described. Metallo-βlactamase from B. fragilis was expressed and purified as described earlier (27). All buffers designated metal-free were prepared with milli Q water (Millipore) and treated with Chelex 100 (Bio-Rad Laboratories) according to the manufacturer’s suggestions. Metal-free dialysis tubing was prepared as described earlier (27). Atomic absorption for Zn(II) content analysis was completed by using a Perkin-Elmer 730 atomic absorption spectrophotometer in the flame mode. Protein concentrations were determined by A280 using ) 39 000 M-1 cm-1 (27). Routine UV-vis spectra and steadystate time courses were determined by using an OLIS/Cary 14 UV-vis spectrophotometer. Construction, Expression, and Purification of C104R Lactamase. Overlap extension PCR2 was used for construction of the C104R mutant (28). Briefly, the wild type ccrA gene on plasmid pMSZ01 (27) was used as a template for two separate PCR reactions, one using a NcoI For primer and a C104R Rev primer, and the other using a BamHI Rev primer and a C104R For primer. All of the primers used were synthesized using an EXPEDITE nucleic acid synthesizer (PerSpective Biosystems, Inc., Framingham, MA) and are listed in Table 1. The products of these reactions were each pooled, precipitated and gel purified using a Qiaex II gel extraction kit (Qiagen). The resulting fragments were then combined along with NcoI For and BamHI Rev primers for another round of PCR resulting in a full-length fragment (approximately 850 bp) containing the desired C104R mutation. This fragment and the expression vector pET-27b(+) (Novagen) were digested sequentially by NcoI and BamHI; the PCR fragment and the 5.4 kb pET fragment were gel purified and ligated together using T4 DNA ligase (Promega) to create the kanamycin-resistant expression plasmid pWF02. This reaction mixture was electrotransformed into Escherichia coli DH5R cells (Life Technologies Inc.) and transformants selected with kanamycin. The exclusive presence of the desired mutation was verified by sequencing the entire lactamase gene from plasmid purified using a Midi purification kit (Qiagen). The pWF02 plasmid 2 Abbreviations: , extinction coefficient; HEPES, N-[2-hydroxyethyl]piperazine-N′-2-ethanesulfonic acid; SDS-PAGE, sodium dodecyl sulfate-polyacrylamide gel electrophoresis; Tris, tris(hydroxymethyl)aminomethane; Mes, 2-(N-morpholine)ethanesulfonic acid; UV-vis, ultraviolet-visible; V, steady-state rate of product formation; PAR, 4-(2-pyridylazo)resorcinol, TCEP, tris(2-carboxyethyl)phosphine; LMCT, ligand to metal charge transfer; PCR, polymerase chain reaction; bp, base pair; EXAFS, extended X-ray absorption fine structure; WT, wildtype.
1642 Biochemistry, Vol. 40, No. 6, 2001 was transformed into BL21(DE3) E. coli (Novagen) when used for expression and subsequent purification of the mutant protein via the same procedures as used for WT lactamase. Extended Dialysis of WT Metallo-β-lactamase. In a sealed container, under a blanket of argon, WT lactamase (2 mL of 236 µM) was dialyzed against metal-free, argon-purged, pH 6.0 cacodylic acid (25 mM, 500 mL). Spectra/Por dialysis tubes (MWCO 10K) fitted with a screw cap (Spectrum) were used to facilitate removal of aliquots which were then diluted in dialysis buffer to 4-22 µM and stored at -70 °C before zinc quantification by atomic absorption. After each aliquot was taken, the dialysis buffer was changed and the argon blanket replaced. Samples were taken after 2, 4, 6, 8, 10, 12, 14, and 15 days of dialysis at 4 °C. Zinc content is described as a ratio of [zinc]/[protein]. Determining Zinc Content of Proteins Using PAR under Denaturing Conditions. Protein samples were diluted to a final concentration of 2-4 µM into freshly prepared metalfree 50 mM Hepes buffer, 4.52 M guanidine hydrochloride, 97 µM 4-(2-pyridylazo)resorcinol (PAR), pH 7.6. Zinc standards were prepared in concentrations ranging from 0 to 10 µM in the same buffer as described above. All samples were allowed to incubate at room-temperature overnight before A500 was measured. The zinc standards were fit to a linear plot and used to calculate the zinc concentrations of the protein samples which were expressed as [zinc]/[protein]. Addition of EDTA to the zinc standards resulted in an A500 close to the value predicted at 0 µM zinc, indicating that there is no metal contamination introduced by the PAR chelator. RemoVing Zinc from WT or C104R under NatiVe Conditions. In a 300 µL cuvette, 1 mM TCEP (Molecular Probes) and various concentrations of PAR (30, 150, 450, 750, 1050, and 1350 µM) in 50 mM metal-free Hepes, pH 7.6, were mixed with 3 µM of protein sample. The increase of A500 due to PAR2Zn formation was observed for about 15 min until the reading stabilized. This value was compared with a zinc standard plot determined separately at each PAR concentration to determine the apparent zinc concentration of the protein sample under native conditions. This concentration was divided by the final protein concentration and subtracted from the total equivalent zinc bound value for each protein as determined by the denaturing experiments described above. The resulting values of equivalent zinc removed from the protein were plotted versus the [PAR]/ [protein] ratios. Batchwise Preparation of Zinc-Depleted Proteins. To prepare a larger quantity of zinc-depleted protein than for the spectrophotometric samples above, a batchwise process was used. Because TCEP complicates the kinetic measurements of WT lactamase (data not shown), all buffers were purged with argon immediately before use, and TCEP was omitted. For preparation of the WT zinc-depleted sample, protein (36.3 µM) was incubated with a 60-fold excess of PAR in MTEN buffer (50 mM Mes, 25 mM Tris, 25 mM ethanolamine, 100 mM NaCl) at pH 7.0 for 20 min at room temperature and subsequently spun through a 1.5 mL Sephadex G-25-150 gel filtration column preequilibrated with argon-purged, metal-free, MTEN buffer at pH 7.0. Protein concentration of the resulting solution was determined by A280 as before and the resulting preparation diluted as needed for other experiments. The monozinc C104R mutant was
Fast et al. prepared in a similar manner by incubating protein samples (28 µM) for 20 min at room temperature with a 75-fold excess of PAR in argon-purged metal-free 50 mM Hepes, pH 7.6, followed by a spin column preequilibrated in the same buffer. If a different buffer was required for kinetic assays, the first spin column was followed by a second preequilibrated in the appropriate buffer. Denaturing PAR assays for zinc concentration showed no detrimental effect from adding a second spin column and verified the final [zinc]/[protein] value. Cobalt Reconstitution of Zinc-Depleted C104R. Monozinc C104R mutant was prepared in metal-free 50 mM Hepes, pH 7.6 buffer as described above. The sample was divided into two aliquots (53.7 µM). To the first aliquot, 1.5 equiv of Zn(II) was added. To the second aliquot, 1.5 equiv of Co(II) was added. Both samples were incubated on ice for 0.5 h and the spectrum of each recorded. Because a slight amount of PAR (
