Probing the dynamics of a mobile loop above the active site of L1, a metallo-beta-lactamase from Stenotrophomonas maltophilia, via site-directed mutagenesis and stopped-flow fluorescence spectroscopy.

A structural feature shared by the metallo-beta-lactamases is a flexible loop of amino acids that extends over their active sites and that has been proposed to move during the catalytic cycle of the enzymes, clamping down on substrate. To probe the movement of this loop (residues 152-164), a site-directed mutant of metallo-beta-lactamase L1 was engineered that contained a Trp residue on the loop to serve as a fluorescent probe. It was necessary first, however, to evaluate the contribution of each native Trp residue to the fluorescence changes observed during the catalytic cycle of wild-type L1. Five site-directed mutants of L1 (W39F, W53F, W204F, W206F, and W269F) were prepared and characterized using metal analyses, CD spectroscopy, steady-state kinetics, stopped-flow fluorescence, and fluorescence titrations. All mutants retained the wild-type tertiary structure and bound Zn(II) at levels comparable with wild type and exhibited only slight (<10-fold) decreases in k(cat) values as compared with wild-type L1 for all substrates tested. Fluorescence studies revealed a single mutant, W39F, to be void of the fluorescence changes observed with wild-type L1 during substrate binding and catalysis. Using W39F as a template, a Trp residue was added to the flexile loop over the active site of L1, to generate the double mutant, W39F/D160W. This double mutant retained all the structural and kinetic characteristics of wild-type L1. Stopped-flow fluorescence and rapid-scanning UV-visible studies revealed the motion of the loop (k(obs) = 27 +/- 2 s(-1)) to be similar to the formation rate of a reaction intermediate (k(obs) = 25 +/- 2 s(-1)).

The ability of bacteria to acquire resistance to antibiotics is a serious problem that continues to challenge modern society (1). Excessive and often misuse of antibiotics in the clinic and for agricultural purposes has resulted in tremendous selective pressure for antibiotic-resistant bacteria (2). These bacteria utilize a variety of methods to become resistant, including modification of cell wall components to prevent antibiotic binding, expression of efflux pumps that transport the antibiotic out of the cell, and the production of enzymes that hydrolyze and render antibiotics ineffective (1,2).
The most common, least expensive, and effective antibiotics currently used are the ␤-lactams, such as carbapenems, cephalosporins, and penicillins (3,4). These antibiotics are mechanism-based inhibitors of transpeptidase, a bacterial enzyme required for the production of a strong viable cell wall (5,6). In response to their widespread use, an increasing number of bacterial strains have acquired the ability to produce ␤-lactamases, enzymes that hydrolyze and render ␤-lactam antibiotics ineffective. There are more than 300 distinct ␤-lactamases known, and Bush has classified these into four distinct groups based on their molecular properties (5,6). One of the more troubling of these is group 3, the metallo-␤-lactamases, which are Zn(II)-dependent enzymes that hydrolyze nearly all known ␤-lactams and for which there are no clinically useful inhibitors (7)(8)(9)(10)(11)(12)(13)(14). To date, there are no reports of a metallo-␤-lactamase being isolated from a major pathogen (15,16); however, these enzymes are produced by a variety of minor clinical pathogens such as Bacteroides fragilis, Pseudomonas aeruginosa, and Stenotrophomonas maltophilia, and the continued extensive use of ␤-lactam containing antibiotics will inevitably result in the production of a metallo-␤-lactamase by a major pathogen (2).
There is significant diversity within the metallo-␤-lactamases, and Bush, based on their amino acid sequence identities and substrate affinities, has further divided them into three subgroups (17). A similar grouping scheme based on structural properties has also been offered (18). The diversity of these subgroups is best exemplified by their vastly differing efficacies toward non-clinical inhibitors; these differences lead to the prediction that finding a single inhibitor for all metallo-␤lactamases may not be possible (14, 19 -27). To address this problem, we are currently characterizing a representative enzyme from each of the metallo-␤-lactamase subgroups with the goal of identifying common structural and mechanistic similarities that can be targeted for the generation of clinically useful inhibitors. This work describes our efforts on characterizing metallo-␤-lactamase L1 1 from S. maltophilia.
S. maltophilia is an important pathogen in nosocomial infections of immunocompromised patients suffering from cancer, cystic fibrosis, drug addition, and AIDS, and in patients with organ transplants and on dialysis (28 -30). This organism is inherently resistant to most antibiotics due to its low outer membrane permeability (31) and to ␤-lactams, due to the production of a chromosomally expressed group 2e ␤-lactamase (L2) and a group 3c ␤-lactamase (L1) (32,33). L1 has been cloned, overexpressed, and partially characterized by kinetic and crystallographic studies (34,35). The enzyme exists as a homotetramer of ϳ118 kDa in solution and in the crystalline * This work was supported by the National Institutes of Health Grants AI40052 and GM40052. Funds to purchase the CD spectrapolarimeter (DBI-0070206) and the stopped-flow UV-visible fluorescence spectrophotometer were provided by the National Science Foundation Grant CHE-0076936. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
‡ To whom correspondence should be addressed: state, tightly binding two Zn(II) ions per subunit. The Zn 1 site has three histidine residues and one bridging hydroxide as ligands, and the Zn 2 site has two histidines, one aspartic acid, one terminally bound water, and the bridging hydroxide as ligands (Fig. 1).
Efforts to solve the crystal structure of one of the metallo-␤lactamases with a bound substrate molecule have failed, most likely due to the high activity of the enzymes, even in the crystalline state, toward all ␤-lactam-containing antibiotics (35,36). Therefore, computational studies have been used extensively to study substrate binding, the role of the Zn(II) ions in catalysis, the protonation state of the active site, protein dynamics, and inhibitor binding (35,(37)(38)(39)(40)(41)(42).
The crystal structure of L1 reveals a series of amino acids that form a flexible loop, a structural feature shared with other metallo-␤-lactamases, which extends over its active site (residues 152-164). It has been suggested that this loop plays an important role during the catalytic cycle of the enzyme, clamping down on substrate, perhaps inducing strain to assist in hydrolysis or helping to stabilize a reaction intermediate (35). To monitor the motion of this loop in L1, we proposed to introduce a Trp residue on the loop to act as a fluorescent probe. However, this simple approach was complicated by the presence of five Trp residues in wild-type L1: Trp-39, Trp-53, Trp-204, Trp-206, and Trp-269 at 6.2, 25.2, 18.8, 39.8, and 19.6 Å from the active site, respectively (Fig. 1). Spencer and coworkers (43) had demonstrated previously that the quenching of fluorescence and subsequent return to original values in wild-type L1 could be used to monitor substrate binding and catalysis (43). It was, therefore, necessary to determine whether these fluorescence changes could be pinpointed to a single Trp residue. To this end, five mutant enzymes were generated in which each Trp residue, in turn, was changed to a Phe residue, resulting in five mutant enzymes: W39F, W53F, W204F, W206F, and W269F. Because of its proximity to the active site and predicted edge/face interaction with His-263, a metal binding ligand (35), we hypothesized that if indeed the observed fluorescence changes were due to a single Trp residue, it was likely Trp-39. If this strategy was successful a second mutation could be introduced into W39F, replacing Asp-160, a residue located on the flexible loop, with a Trp to act as a fluorescent probe, resulting in the double mutant W39F/ D160W. This work describes our efforts to determine whether we could generate an enzyme free of fluorescence changes during substrate binding and catalysis and subsequently create a double mutant that could be used to monitor the dynamics of the flexible loop in L1.

EXPERIMENTAL PROCEDURES
Escherichia coli strains DH5␣ and BL21(DE3) were obtained from Invitrogen and Novagen, respectively. The plasmid pET26b was purchased from Novagen. Primers for sequencing and mutagenesis studies were purchased from Integrated DNA Technologies. QuikChange™ mutagensis kit was purchased from Stratagene. DNA was purified using the Qiagen QIAQuick gel extraction kit or Plasmid Purification kit with Qiagen-tip 100 (Midi) columns. Wizard Plus Minipreps were acquired from Promega. Luria-Bertani media in powder form was purchased from Invitrogen. Isopropyl-␤-thiogalactoside, Biotech grade, was procured from Anatrace. Protein solutions were concentrated with an Amicon ultrafiltration cell equipped with YM-10 DIAFLO membranes from Amicon, Inc. Dialysis tubing was prepared using Spectra/Por regenerated cellulose molecular porous membranes with a molecular weight cut-off of 6 -8000 g/mol (44). Q-Sepharose Fast Flow was purchased from Amersham Biosciences. Nitrocefin was purchased from BD Biosciences, and solutions of nitrocefin were filtered through a Fisherbrand 0.45-m syringe filter (34). Cephalothin and penicillin G were purchased from Sigma and Fisher, respectively. Meropenem was generously supplied by Zeneca Pharmaceuticals. All buffers and media were prepared using Barnstead NANOpure ultrapure water.
Extinction coefficients for the mutants were determined utilizing the BCA protein assay kit purchased from Fisher, with wild-type L1 used to create the calibration curve. The concentrations of L1 and the mutants were determined by measuring the protein absorbances at 280 nm and using the published extinction coefficient of ⑀ 280 nm ϭ 54,606 M Ϫ1 ⅐cm Ϫ1 for wild-type L1 (34) and the experimentally determined extinction coefficients for the mutants. Analysis of metal content was performed on enzyme samples that were incubated for 1 h on ice in 50 mM HEPES, pH 7.5, containing a final concentration of 100 M ZnCl 2 . Weakly bound metal was removed by dialysis versus 2 ϫ 1 liter of metal-free (chelexed) 50 mM HEPES, pH 7.5. A Varian Liberty 2 inductively coupled plasma spectrometer with atomic emission spectroscopy detection was used to determine the metal content of multiple preparations of wild-type L1 and L1 mutants. Calibration curves were based on four standards and had correlation coefficient limits of at least 0.9950. The final dialysis buffer was used as a blank. The emission line of 213.856 nm is the most intense for zinc and was used to determine the zinc content in the samples. The errors in metal content data reflect the S.D. ( nϪ1 ) of multiple enzyme preparations. Circular dichroism samples were prepared by dialyzing the purified enzyme samples versus 3ϫ 2 liters of 5 mM phosphate buffer, pH 7.0, over 6 h. The samples were diluted with final dialysis buffer to ϳ75 g/ml. A Jasco J-810 CD spectropolarimeter operating at 25°C was used to collect CD spectra (34,45).
Steady-state kinetic assays were conducted at 25°C in 50 mM cacodylate buffer, pH 7.0, containing 100 M ZnCl 2 on a HP 5480A diode array UV-visible spectrophotometer. The changes in molar absorptivities (⌬⑀) used to follow the reactions were (in M Ϫ1 cm Ϫ1 ): nitrocefin, ⌬⑀ 485 ϭ 17,420; cephalothin, ⌬⑀ 265 ϭ Ϫ8,790; meropenem, ⌬⑀ 293 ϭ Ϫ7,600; penicillin G, ⌬⑀ 235 ϭ Ϫ936. When possible, substrate concentrations were varied between 0.1 to 10 times the K m value, and changes in absorbance (⌬A) versus time data were measured for a period of 60 s for each substrate concentration. Steady-state kinetic constants, K m and k cat , were determined by fitting initial velocity versus substrate concentration data directly to the Michaelis equation using Igor Pro (34). The reported errors reflect fitting uncertainties. All steady-state kinetic studies were performed in triplicate with recombinant L1 and L1 mutants from at least three different enzyme preparations.
Fluorescence spectra of 2 M wild-type L1 and L1 mutant samples were obtained at 25°C using an excitation wavelength of 295 nm on a PerkinElmer Life Sciences LS 55 luminescence spectrometer. Apo (metal-free) enzymes for fluorescence titrations were prepared by dialysis of samples versus 5ϫ 1 liter of 50 mM HEPES, pH 7.0, containing 10 mM phenanthroline followed by dialysis versus 5ϫ 1 liters of metal-free (chelexed) 50 mM HEPES, pH 7.0, and metal analyses were used to verify that the enzymes were metal-free. Data from fluorescence titrations were fitted to Stopped-flow fluorescence studies of nitrocefin hydrolysis by wildtype L1 and L1 mutants were performed on an Applied Photophysics SX.18MV spectrophotometer using an excitation wavelength of 295 nm and a WG320-nm cut-off filter on the photomultiplier. These experiments were conducted at 2 and 25°C depending on the enzyme studied using the same buffer as in the rapid-scanning visible studies. Rapidscanning UV-visible studies were performed under identical conditions using the above-mentioned instrument and utilizing an Applied Photophysics diode array detector.

RESULTS
Wild-type L1 and the L1 mutants were overexpressed in E. coli and purified as previously described (34), with the changes for W53F and W206F as noted under "Experimental Procedures." This procedure produced an average of 50 -60 mg of Ͼ95% pure, active protein per 4 liters of growth culture. Circular dichroism spectra were collected on samples of wild-type L1 and each of the mutants to ensure the proteins produced using the pET26b overexpression system had the correct secondary structure. The CD spectra of wild-type L1 and the mutants were similar and showed an intense, broad feature at 190 nm and a smaller feature at 215 nm (data not shown). These features are consistent with a sample with significant ␣/␤ content (35). Analyses of the CD spectra were performed on-line using DICHROWEB utilizing an algorithm called CDSSTR (www.cryst.bbk.ac.uk/cdweb/html/home.html) (47,48). Results are shown in Table I. The CD spectra demonstrate no significant changes in the amount of unordered content and only minor differences in ␣/␤ content that can be attributed to fitting errors using the CDSSTR program. Metal analyses on multiple preparations of wild-type L1 and the mutants demonstrated that the wild-type L1 and W39F, W53F, W204F, W206F, W269F, and W39F/D160W mutants bind 1.9 Ϯ 0.2, 1.9 Ϯ 0.2, 1.6 Ϯ 0.3, 1.8 Ϯ 0.2, 1.5 Ϯ 0.3, 2.0 Ϯ 0.1, and 1.9 Ϯ 0.2 Zn(II) eq per monomer (Table I). Within the limits of error, the metal analyses demonstrate that the single point mutations did not result in significant changes in metal binding ability of the enzymes.
Steady-state kinetic constants, K m and k cat , were determined for wild-type L1 and each of the mutants with four substrates. These values are presented in Table II. Cephalothin and nitrocefin, meropenem, and penicillin G were utilized as representatives of the three major classes of ␤-lactam-containing antibiotics, cephalosporins, carbapenems, and penicillins, respectively. The L1 preference for penicillins and carbapenems over cephalosporins, as exemplified by the k cat values, is in agreement with previous studies (34) and supports the L1 placement in the ␤-lactamase Bc family (17). For all substrates tested, the mutant enzymes exhibited slightly reduced k cat values (Ͻ10-fold) as compared with wild-type L1. Because K m values are often used as a first approximation of substrate binding (49), the K m values exhibited by the mutants were compared with those of wild-type L1. All five mutants, with the exception of W39F and W39F/D160W, exhibited similar K m values to wild-type L1. K m values for W39F were typically 4-fold greater than those of wild-type L1, whereas those of the double mutant approached and in one case (cephalothin) exceeded 10-fold that of wild-type L1. Because changes of 10-fold or greater are the standard to indicate significant differences (50), steady-state kinetics indicate that only W39F and W39F/ D160W approach this standard. However, both enzymes retained enough activity to be suitable for the purpose of this study. Initial fluorescence scans of wild-type L1 and the L1 mutants ( Fig. 2) revealed that W39F and W206F exhibited decreased fluorescence emissions at 340 nm, whereas W53F and W269F exhibited increases in their fluorescence emissions at 340 nm as compared with wild-type L1. W204F exhibited a fluorescence spectrum almost identical to that of wild-type L1, as did the double mutant, W39F/D160W. These spectra indicate that Trp-204 is fairly inconsequential to the overall natural fluorescence of wild-type L1. The decrease in fluorescence at 340 nm observed with the removal of Trp-39 or Trp-206 indicate that these residues contribute significantly to the natural fluorescence of wild-type L1. It appears as though the addition of the Trp residue at position 160 compensates for the loss of Trp-39. The increased fluorescence at 340 nm observed upon the removal of Trp-53 or Trp-269 indicates a mutual fluorescence quenching between these residues.
Fluorescence titrations of wild-type L1 (excitation wavelength 295) showed an increase in fluorescence emission at 340 nm with the addition of up to ϳ0.9 eq of Zn(II) with no further increased emission upon the addition of greater amounts Zn(II). The fluorescence titration data were fitted to a quadratic equation (46) (lines in Fig. 3), and the resulting K D values were 2-3 orders of magnitude lower than the concentration of apoenzyme. These results suggest that these plots are active site titrations and not plots to determine Zn(II) binding K D values. It would appear that L1 binds Zn(II) preferentially to one of its Zn(II) binding sites. Fluorescence titrations of the L1 mutants revealed similar trends to that of wild-type L1 with the exception of W39F. W39F yielded virtually no change in fluorescence emission at 340 nm regardless of the amount of Zn(II) added (Fig. 3). Observing a smaller than expected change in fluorescence at 340 nm of W269F, we noted that this enzyme precipitates with the first few additions of Zn(II). The addition of EDTA to all of the samples except W39F at the end of each titration resulted in a decrease in fluorescence to the initial value (data not shown), demonstrating reversible Zn(II) binding.
To observe fluorescence changes upon the binding of substrate to wild-type L1 and the mutants, stopped-flow fluorescence studies were conducted as previously described (43). The reaction of enzyme with nitrocefin under single-turnover conditions at 25°C resulted in a rapid decrease in fluorescence followed by a rate-limiting return of fluorescence for all of the single point mutants except W39F (Fig. 4). Stopped-flow fluorescence spectra of W39F/D160W with nitrocefin differed from that of wild-type L1 and the other mutants in that with the addition of substrate, there was a rapid quenching of fluorescence that did not return to the value of the resting enzyme (Fig. 5). We found this behavior of W39F/D160W to be sub-  strate-specific for nitrocefin. When the experiment was repeated with meropenem as the substrate, the same quenching was observed; however, the fluorescence returned to the resting-enzyme value (data not shown). This behavior was confirmed with fluorescence scans of W39F/D160W with 1:1 nitrocefin and meropenem. The scans revealed a decrease of emission at 340 nm for the nitrocefin sample, whereas the meropenem sample was identical to that of the enzyme without substrate (Fig. 6). Close inspection of the stopped-flow fluorescence spectra of W39F/D160W with nitrocefin ( Fig. 5) revealed that the quenching is biphasic in nature, with an initial substrate concentration-dependent rapid phase followed by a slower phase seemingly independent of substrate concentration.
In previous studies rapid scans of wild-type L1 with nitrocefin revealed absorbance changes at three distinct wavelengths. These are a decay in the absorbance at 390 nm, an increase at 485 nm, and a rapid increase and decay at 665 nm, which can be attributed to substrate decay, product formation, and the appearance and decay of an intermediate, respectively. To date nitrocefin is the only substrate with which an intermediate can be spectroscopically observed. To investigate if the fluorescence quenching observed with W39F/D160W could be related to events during catalysis of nitrocefin, stopped-flow rapid-scanning UV-visible and fluorescence spectra of WT L1 and W39F/ D160W with nitrocefin at single turnover conditions were obtained under identical conditions (Figs. 5 and 7). The temperature was reduced to 2°C to slow the reaction and better our chances to obtain data suitable for kinetic fitting. Fitting the wild-type L1 data from the fluorescence experiments revealed quenching and recovery rates of 106 Ϯ 2 s Ϫ1 and 6.5 Ϯ 0.1 s Ϫ1 , respectively. The rate of quenching with W39F/D160W revealed two phases, 282 Ϯ 2 and 27 Ϯ 2 s Ϫ1 .
Fitting of the rapid-scanning data for wild-type L1, W39F, and W39F/D160W yielded rates of 69 Ϯ 2 and 7.1 Ϯ 0.1 s Ϫ1 , 18 Ϯ 3 and 6.8 Ϯ 0.8 s Ϫ1 , and 25 Ϯ 2 and 4.8 Ϯ 0.8 s Ϫ1 for intermediate formation and decay, respectively. These rate data are summarized in Table III. DISCUSSION In proteins fluorescence properties are typically associated with Trp residues and changes in the local environment of those residues during catalysis. Previous work by Spencer et al. (43) demonstrated that the binding of substrate to L1 results in a rapid quenching of fluorescence followed by slower return to resting values. The initial quenching of fluorescence was correlated to substrate binding and utilized to determine K s values for L1 with multiple substrates. The rate of return to the fluorescence of the resting enzyme was also correlated to the k cat value, as determined via steady-state kinetics, of the enzyme with that particular substrate. To determine whether this fluorescence could be attributed to a specific Trp residue, five mutant enzymes were generated in which each Trp residue was changed to a Phe.
Mutations were introduced into L1 after the Stratagene  QuikChange™ mutagenesis protocol, and the resulting enzymes were characterized to probe whether the single point mutations resulted in large structural changes in these mutants. Overexpression levels, total amounts of isolatable enzyme after protein purification, levels of bound Zn(II), and CD spectra of the mutants were compared with those of wild-type L1. All mutants exhibited overexpression and isolatable enzyme levels nearly identical to those of wild-type L1 (data not shown). Metal analyses of wild-type L1 and the L1 mutants showed that all of the enzymes within error bound nearly 2 mol of Zn(II) per monomer (Table I), indicating that the structures of the enzyme metal-binding sites were unaltered. Although slight variations in the CD spectra of the mutants, as compared with wild-type L1, were observed, fitting of the spectra utilizing the CDSSTR algorithm on the DICHROWEB website revealed virtually no change in the amount of unordered structure for all the mutant enzymes (Table I). These lines of evidence lead to the conclusion that none of the point mutations resulted in large structural changes in L1, particularly at the active site, and that any kinetic differences and changes in fluorescence properties could be attributed to the changed amino acid.
To compare the activity of the mutants with wild-type L1, we examined the steady-state kinetics of nitrocefin, cephalothin, meropenem, and penicillin G with each enzyme (Table II). The substrates tested were chosen to cover the three major categories of ␤-lactam antibiotics and because they exhibited low K m values in previous kinetic studies (34). We, therefore, believed that we could saturate the enzymes with substrate even if there was a large change in binding with the point mutations. It is not surprising that the smallest k cat values were obtained with cephalosporins, affirming the L1 classification as a group Bc metallo-␤-lactamase (preference toward penicillins and carbapenems) (17). For all substrates the mutants showed altered activities; however, decreases in k cat values were typically on the order of 2-3-fold, with W39F showing the greatest disparity at 3-6-fold reduced activity. In all cases the decreases were far less than the standard 1 order of magnitude change needed to indicate a significant difference. Therefore, we conclude that none of the point mutations significantly altered the activity of L1.
Initially we utilized the method of Edelhoch (51) to adjust the extinction coefficient of wild-type L1 to reflect the loss of a Trp residue in each of the L1 mutants, resulting in a value of 48,916 M Ϫ1 ⅐cm Ϫ1 for each. However, fluorescence scans of the mutant enzymes indicated that a change from Trp to Phe did not alter the natural fluorescence of L1 in a consistent manner (Fig. 2), indicating that the environment of the Trp residue is critical in determining its contribution to the natural fluorescence of the enzyme. It is logical then to assume that the local environment of a Trp residue is also important to its ability to absorb radiation at 280 nm, and therefore, simply assigning the same extinction coefficient to all of the L1 mutants was not valid. To account for the environmental differences, we utilized a BCA protein assay in combination with absorbance at 280-nm data to individually determine extinction coefficients for each L1 mutant (Table I).
Two fluorescence studies were conducted with each L1 mutant and compared with wild-type L1 to determine whether previously observed properties could be attributed to a single Trp residue. First, each enzyme was analyzed with stoppedflow fluorescence using nitrocefin as the substrate (43,52). W53F, W204F, W206F, and W269F all exhibited fluorescence spectra similar in shape to that of wild-type L1 (Fig. 4), showing the quenching of fluorescence upon the binding of substrate to the enzyme and return of fluorescence during substrate turnover. W39F exhibited no change in fluorescence upon the binding and turnover of substrate. There are two possible explanations for this observation. First, as previously observed with other mutants, substrate does not readily bind to W39F, and therefore, no change in fluorescence can be observed (52). Second, the Trp residue at position 39 is responsible for the observed fluorescence changes in wild-type L1, and the absence of this Trp residue in W39F is the reason no changes in fluorescence can be observed. In those previous studies, the mutant that did not yield fluorescence changes was also observed to have a nearly 1000-fold decrease in its k cat value and a 25-fold increase in its K m value with nitrocefin as compared with wild-type L1. Steady-state kinetic data on W39F, however, shows no significant change in the k cat or K m values of W39F as compared with wild-type L1 (Table II). It is reasonable then to conclude that the absence of fluorescence changes with W39F is not due to an inability of the enzyme to bind and turnover substrate and can be attributed to removal of the Trp residue at this position. A second experiment, fluorescence titrations of apo-wild-type L1 and each of the L1 mutants with Zn(II), confirms this conclusion. With wild-type L1, a steady increase in fluorescence emission at 340 nm is observed upon the addition of Zn(II). At ϳ0.9 eq of Zn(II), the fluorescence emission reaches a maximum, and there is no further increase (Fig. 3). The addition of EDTA to the sample at the end of the titration resulted in a decrease in fluorescence to the initial value (data not shown). These results demonstrate reversible Zn(II) binding and indicate that binding to one site is preferred, and only the binding of Zn(II) in this site can be monitored with fluorescence. Each of the L1 mutants, with the exception of W39F, yielded spectra with features similar to that of wild-type L1. The small change noted in the fluorescence at 340 nm of W269F and failure to reach levels observed in the fluorescence scans are attributed to precipitation of this enzyme upon the addition of Zn(II). No change in fluorescence was observed during the titration of apoW39F with Zn(II). Assuming that the added Zn(II) is populating the Zn(II) binding sites in W39F, which from metal analysis data is logical, the absence of fluorescence changes can be attributed to the missing Trp residue at position 39. Furthermore, due to its proximity to the Zn 2 binding site and proposed face/edge interaction with His-263, it is reasonable to conclude that Zn(II) preferentially binds to the Zn 2 binding site in wild-type L1. This result is in contrast to previous studies on ␤-lactamase II where the Zn 1 site has been shown to be tighter binding (53).
Once Trp-39 was established as the residue responsible for fluorescence changes observed during substrate binding and catalysis in wild-type L1, W39F was used as a template to create a double mutant, W39F/D160W, using the methods previously described. The replacement of Asp-160 with a Trp residue afforded the ability to directly monitor the dynamics of a flexible chain of amino acids that extends over the active site of L1 (54). This double mutant was characterized using the same methods as for the single point mutants of L1, and all lines of evidence lead to the conclusion that this second mutation did not result in significant structural changes for this enzyme. W39F/D160W overexpression, isolatable enzyme levels, and levels of bound Zn(II) ( Table I) were identical to those of wild-type L1. The CD spectra of this double mutant was virtually identical to that of wild-type L1, as revealed by the fitting of the spectra utilizing the CDSSTR algorithm on the DICHROWEB website (Table I). We were confident then that any kinetic differences and, more importantly, changes observed in fluorescence studies could be attributed to the Trp residue engineered into the flexible loop of the protein.
Steady-state kinetic constants, K m and k cat , were determined for W39F/D160W, with the same representative substrates used with the single point L1 mutants. For all substrates tested, the double mutant exhibited k cat values similar to W39F, with only slightly reduced values for cephalothin and meropenem and a value approximately twice as large as with penicillin G. Importantly, the k cat value with nitrocefin was virtually identical to that of W39F and wild-type L1. The only noticeable difference in K m values, as compared with W39F, was with penicillin G, where a 3-fold increase was observed. The K m value with nitrocefin was approximately a 6-fold increase over that of wild-type L1; it is, however, identical, within error to that of W39F. Therefore, it is reasonable to conclude that the increased K m as compared with wild-type L1, can be attributed to perturbations in the active site due to the change from Trp to Phe at position 39, since there appears not to be any further increase caused by the addition of the Trp residue on the flexible loop portion of the enzyme.
Stopped-flow fluorescence studies of W39F/D160W with nitrocefin showed a biphasic quenching of fluorescence that did not return to initial values (Fig. 5). The initial decline is rapid and dependent on nitrocefin concentration, which we interpret to be a binding type event where the loop rapidly closes on substrate as it docks in the active site. The second phase of the quenching is significantly slower and is not substrate concentration-dependent. The observed rate for this step (k obs ϭ 27 Ϯ 2 s Ϫ1 ) is identical within error to the rate of intermediate formation (k obs ϭ 25 Ϯ 2 s Ϫ1 ) observed in rapid-scanning experiments performed under identical conditions (Fig. 7). Previous work predicted the breakdown of this intermediate during a protonation event to be the rate-limiting step of the reaction (55). This conclusion is supported by the data in this work, as the rate of breakdown of the intermediate is noted to be the slowest step observed in these studies. We conclude, therefore, that once bound to the active site, the loop assists in destabilizing substrate to form the intermediate in a non-ratelimiting event.
Because the fluorescence never rebounds from the quenched state with W39F/D160W, we predict that the loop is somehow locked in the closed position, trapping an enzyme-bound product species. This appears to be specific to nitrocefin, however, since in the same stopped-flow fluorescence experiments using meropenem as the substrate, the observed quenching returns to initial values (data not shown). Fluorescence scans of free enzyme and 1:1 enzyme-substrate solutions using wild-type L1, W39F, and W39F/D160W with nitrocefin and meropenem as substrates (Fig. 6) confirm this conclusion. However, suppressed fluorescence observed with the 1:1 wild-type L1/nitrocefin solution (data not shown) indicates that this enzymebound product is not unique to the double mutant but is a phenomenon observed with L1 and cephalosporins.
In this work we have demonstrated that the fluorescence properties of L1 can be attributed to a single Trp residue, specifically Trp-39. We exploited this fact to create a mutant enzyme with a Trp label on the flexible loop portion of the enzyme, and stopped-flow fluorescence and rapid-scanning UVvisible were used to demonstrate the motion of this loop is kinetically linked to the non-rate-limiting formation of a reaction intermediate. Furthermore we have presented evidence for an enzyme-bound product when cephalosporins are used as substrates with L1. This knowledge forms the basis for the development of a clinically useful inhibitor of L1 and other metallo-␤-lactamases. It may be possible to engineer a substrate mimic containing substituents able to irreversibly bind to the loop, thereby locking the molecule in the active site, or block the loop and effectively inhibit the enzyme.