Distortion in the Ground-State ES Complex in

Mark J. Hokenson, Gregory A. Cope, Evan R. Lewis, Keith A. Oberg, and Anthony L. Fink*. Department of ... Samuel H. Schneider and Steven G. Boxer...
0 downloads 0 Views 103KB Size
6538

Biochemistry 2000, 39, 6538-6545

Enzyme-Induced Strain/Distortion in the Ground-State ES Complex in β-Lactamase Catalysis Revealed by FTIR† Mark J. Hokenson, Gregory A. Cope, Evan R. Lewis, Keith A. Oberg, and Anthony L. Fink* Department of Chemistry and Biochemistry, The UniVersity of California, Santa Cruz, California 95064 ReceiVed December 7, 1999; ReVised Manuscript ReceiVed March 20, 2000

ABSTRACT: Class A β-lactamases hydrolyze penicillins and other β-lactams via an acyl-enzyme catalytic mechanism. Ser70 is the active site nucleophile. By constructing the S70A mutant, which is unable to form the acyl-enzyme intermediate, it was possible to make stable ES complexes with various substrates. The stability of such Michaelis complexes permitted acquisition of their infrared spectra. Comparison of the β-lactam carbonyl stretch frequency (νCO) in the free and enzyme-bound substrate revealed an average decrease of 13 cm-1, indicating substantial strain/distortion of the lactam carbonyl when bound in the ES complex. Interestingly, regardless of the frequency of the CdO stretch in the free substrate, when complexed to Bacillus licheniformis β-lactamase, the frequency was always 1755 ( 2 cm-1. This suggests the active site environment induces a similar conformation of the β-lactam in all substrates when bound to the enzyme. Using deuterium substitution, it was shown that the “oxyanion hole”, which involves hydrogen bonding to two backbone amides, is the major source of the enzyme-induced strain/distortion. The very weak catalytic activity of the S70A β-lactamase suggests enzyme-facilitated hydrolysis due to substrate distortion on binding to the enzyme. Thus the binding of the substrate in the active site induces substantial strain and distortion that contribute significantly to the overall rate enhancement in β-lactamase catalysis.

β-Lactamases are the major source of resistance to β-lactam antibiotics and have been the targets of considerable mechanistic scrutiny. Despite numerous investigations, many details of the catalytic mechanism are still uncertain. The residues believed to be most important in catalysis by the class A family of β-lactamases include the strictly conserved active site residues, Ser70 and Lys73, as well as Glu166, Asn170, Ser130, and Lys234 (1). With the exception of Ser70, which acts as a nucleophile to form a covalent acylenzyme intermediate (Scheme 1), the mechanistic role of the other residues remains controversial. Kinetic studies indicate that catalysis is dependent on at least two ionizable groups with pK’s in the vicinity of 5 and 8.5. Glu166 is a likely candidate for the observed ionizable group with a pK1 in the vicinity of 5. The pK and the heat of ionization of the group responsible for the alkaline limb of the pH-rate profile have been shown to be consistent with a lysine residue (2), and Lys73 is a logical choice for the group responsible for pK2. In fact, using β-lactamase mutants, in which Lys73 and Lys234 were mutated to alanine, we have recently shown that pK2 reflects the ionization of both Lys73 and Lys234 (Lietz E., et al., submitted for publication). Such an assignment would require the catalytically active form to be the protonated state. Substantial evidence supports an acylenzyme kinetic mechanism (Scheme 1) in which the acylenzyme bond is formed by nucleophilic attack of the side chain of Ser70 on the β-lactam carbonyl (1). †This research was supported by a grant from the National Science Foundation. * To whom correspondence should be addressed. Phone: 831-4592744. Fax: 831-459-2935. E-mail: [email protected].

Scheme 1

Ground-state distortion toward the transition state is believed to be a significant component of an enzyme’s catalytic machinery; however, examples that directly demonstrate such distortion are few. In the present investigation, we have used a combination of site-directed mutagenesis and FTIR spectroscopy to probe the structure of the noncovalent Michaelis complex. This was achieved by making the S70A mutant of β-lactamase, in which the active site serine is replaced by alanine, preventing formation of the acylenzyme and thus stopping the reaction at the ES complex, and by examination of the CO stretch frequency of the β-lactam carbonyl. By using hydrated thin-film attenuated total reflectance (ATR)1 FTIR it was possible to use much lower enzyme concentrations in the reaction solution than those required in typical FTIR transmission mode measurements (e.g., e1 mg/mL as compared to >10 mg/mL). It has been previously shown that the hydrated thin-film ATR technique effectively does not affect the structure of native proteins (3-6). This is because it is only the first layer of protein in contact with the IRE that may be structurally perturbed, and this contributes a negligible amount to the total signal. Fortuitously, neither the enzyme nor the product penicilloic acids absorb in the 1700-1800 cm-1 region, whereas substrates have a prominent absorbance due to the β-lactam carbonyl in the vicinity of 1760-1780 cm-1. Thus it should be possible to trap noncovalent ES complexes using 1 Abbreviations: FAP, N-(2-furylacryloyl)penicillin; WT, wild-type; IRE, internal reflectance element; ATR, attenuated total reflectance.

10.1021/bi9928041 CCC: $19.00 © 2000 American Chemical Society Published on Web 05/05/2000

Strain/Distortion in β-Lactamase Catalysis

Biochemistry, Vol. 39, No. 21, 2000 6539

the S70A enzyme and, from analysis of the carbonyl β-lactam band position, determine whether the enzyme induces strain or distortion of the β-lactam carbonyl bond in the complex.

Scheme 2: Subtraction Protocol Used to Extract the ES Spectruma

MATERIALS AND METHODS β-Lactamase from Bacillus licheniformis was produced and purified by overexpression from Escherichia coli as described elsewhere (Lietz, E. J., and Fink, A. L., manuscript submitted). Enzyme preparations used were homogeneous by IEF-PAGE with Coomassie Blue staining. Benzylpenicillin, cefoxitin, cefotaxime, phenoxymethylpenicillin, lysozyme, bovine pancreatic γ-chymotrypsin, and IPTG were purchased from Sigma. N-(2-furylacryloyl)penicillin (FAP) was from Calbiochem. Enzyme Kinetics. β-Lactamase activity toward FAP was determined spectrophotometrically using a Hewlett-Packard 8452A spectrophotometer. Path lengths employed were 1.0 or 0.1 cm. The hydrolysis of the amide bond in the lactam ring was followed at 340 nm, where  ) 3460 M-1 cm-1 and ∆ ) 1905 M-1 cm-1. Data analyses were performed using SigmaPlot (SPSS, Chicago, IL). For determination of Ki and Km, initial velocities were used. The pH dependence of kcat/Km was determined under first-order conditions (where Km . [S]). All assays were run at 30 °C, and the S70A concentration was usually 15 µM. Reactions were buffered by sodium acetate (pH 4.0-5.5), sodium phosphate (pH 6.08.0), sodium pyrophosphate (pH 8.0-9.5), and CAPS (pH 10.0-10.5). Buffer concentrations for all kinetic assays were 125 mM, with 0.5 M KCl. The rate of spontaneous hydrolysis of the substrates was subtracted from the enzyme-catalyzed rates. Ionization constants were determined by fitting the (kcat, Km, and kcat/Km) pH profiles to the expression in eq 1.

kobs ) klim/(1 + [H+]/K1 + K2/[H+] + K2/K1)

(1)

Sample Preparation for FTIR. Freshly prepared solutions of β-lactam substrates (penicillins and cephalosporins) were made in either DDI H2O or D2O, depending on the experiment performed, containing 10 mM potassium phosphate buffer (pH 7.0). To prevent significant substrate hydrolysis, the S70A β-lactamase mutant was allowed to react with a substrate for only 2-3 min prior to analysis. Variable incubation times were used in turnover experiments with wild-type enzyme. Typically, the concentration for each constituent was as follows: 450 µM substrate and 45 µM S70A β-lactamase (or 45 µM chicken egg white lysozyme or bovine pancreatic γ-chymotrypsin; see below). In some cases, 1:1, 2:1, or 4:1 substrate/enzyme ratios were used, with 45 or 90 µM enzyme. The buffer was 10 mM potassium phosphate, pH 7.0. All reaction mixtures were incubated at 22 °C. In the ATR experiments, samples (50 µL) were placed on a clean internal reflectance element (IRE), and the solvent was evaporated under a gentle stream of nitrogen gas. The substrate solution was either mixed with enzyme stock solution in a small vial and aliquots were removed for ATRFTIR analysis at the desired time intervals, or more typically the solutions were mixed directly on the IRE. During the solvent evaporation, a pipet tip was used to spread the sample uniformly over the surface of the IRE. Since the mixing time of β-lactamase with substrate is crucial, steps were taken to minimize their time of exposure. These steps include the use

a E is the β-lactamase, S is substrate, L is lysozyme (see text), (E + S + ES) is the reaction mixture of enzyme and substrate, and ES represents the substrate bound in a Michaelis complex with the β-lactamase. At each step in the process, the subtraction scaling factor (n) was iteratively optimized by SAFAIR using the criteria indicated to evaluate the subtraction result.

of 50 µL samples as well as mixing the β-lactamase enzyme with substrate on the crystal. Once the bulk solvent had been removed, the IRE was placed in the FTIR sample compartment, and a spectrum was collected. The nitrogen used to dry deuterated samples was bubbled through deuterium oxide to prevent 1H exposure to the sample while drying. Deuterated Sample Preparation. β-Lactamase (S70A) and lysozyme were dialyzed against DDI water and lyophilized and then were dissolved in an appropriate volume of D2O, centrifuged, and the supernatant was lyophilized. Lyophilization and dissolution in D2O were performed twice more to ensure complete solvent exchange. FTIR Spectroscopy. Spectra were obtained using a Nicolet 800SX FTIR spectrometer equipped with an MCT detector. A total of 1024 interferograms were collected at either 4- or 2-cm-1 resolution. In ATR analysis, samples were examined on a trapezoidal ZnSe IRE crystal (73 × 10 × 6 mm, 45°) placed in a horizontal out-of-compartment ATR apparatus (SPECAC) (3). For some experiments, a germanium IRE was used. FTIR Data Processing and Analysis. All the data were analyzed with GRAMS/32 from Galactic Industries. The interferograms were converted to spectra with the Mertz method using medium Norton-Beer apodization. Contributions from water vapor, buffer, and solvent water were removed by subtraction. Two to four independent spectra of each sample type were scanned and examined individually to ensure consistent results. Subtractions were used to identify the spectral changes induced by the enzyme on the β-lactam carbonyl. The overall subtraction protocol is outlined in Scheme 2. In essence, this describes the subtraction of both the enzyme and the unbound-substrate signals from the spectrum of the mixture. In practice, obtaining reproducible results from these subtractions was not straightforward. Although signals are strong in ATR-FTIR, their intensities vary from sample to sample due to slightly different distributions of a sample on the surface of the IRE. One of the spectra must always be rescaled to match the other. To overcome this variability and still obtain the objectivity of an automatic processing

6540 Biochemistry, Vol. 39, No. 21, 2000 algorithm, the SAFAIR subtraction software package was employed. SAFAIR iteratively varies the scaling factor for a subtraction such that a user-selected “goodness” function is optimized (minimized). SAFAIR is well-suited to the extraction of small signals from intense spectra because it optimizes the subtraction-scaling factor based on the resulting difference spectrum rather than on the original spectra. The RMS deviation of replicate subtractions for a given substrate was on the order of (2%. The SAFAIR criteria used for this analysis were “linear” and “match". The match function optimizes the subtraction so that the shape of the difference (result) spectrum is as close as possible to a reference spectrum. Here, the reference spectrum was that of the substrate alone. The subtractions therefore removed all protein signal and left only the substrate signal. Note that there was a small contribution from the ES complex in the reaction mixture; however, this band was no more than 5-10% of the intensity in a small window of the data region that was used for the optimization. This signal can therefore be considered negligible in the result spectrum (S + ES). The linear function is optimal when the subtraction result is a straight line. Linearizing the 1850-1780 cm-1 data region caused a scaling factor to be selected that brought the slope of a region where there is only substrate signal (1800-1780 cm-1) to the same value as a baseline region (1850-1800 cm-1). This effectively removed the substrate signal without oversubtracting. Because of the effects of the refractive index of the protein on the substrate spectrum, which causes small shifts (