INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

-Lactam hydrolysis by the ubiquitous bacterial -lactamases represents the most common biochemical mechanism of resistance engineered by bacteria to protect themselves against the bactericidal effect of -lactam antibiotics. Based upon amino acid sequence alignments, four distinct classes of -lactamases can be distinguished (2, 3, 5, 16). Classes A, C, and D are serine -lactamases, which undergo acylation by the substrate -lactam ring. Those in class B are zinc-catalyzed -lactamases.

TEM-1-derived extended-spectrum -lactamases are numerous and widespread (6) because of the intensive usage of expanded-spectrum cephalosporins and other compounds resistant to the usual -lactamases. The capacity of TEM-1 to modify its substrate specificity represents a major clinical problem, and work to understand the mechanisms modulating this specificity was done previously with this enzyme (4, 11, 18, 19, 21, 26-28). Amino acid substitutions and pentapeptide insertion in the  loop of TEM-1 (residues 162-179) were found to increase the level of activity toward ceftazidime and subsequently to increase the level of ceftazidime resistance (9, 18, 19, 21). The  loop is a secondary structural element having a small distance between segment termini and is known to form a part of the active site; it also contains the catalytic residue Glu166, which is implicated in the deacylation step.

Studies of the capacity of other related class A -lactamase mutants to appear with modifications in substrate specificity have been poorly characterized. Because resistance to expanded-spectrum cephalosporins is still developing with the identification of newer -lactamases, it is important to understand how substrate specificity can be modified in other class A -lactamases and to determine if the genetic mechanism modulating substrate specificity could be generalized to all other members of this class.

PSE-4 is a class A -lactamase classified in the group 2c of Bush et al. (3). The particularity of these enzymes is the ability to hydrolyze carbenicillin as efficiently or better than benzylpenicillin. To investigate the potential of PSE-4 to extend its substrate specificity toward expanded-spectrum cephalosporins and to identify potential determinants of substrate specificity for ampicillin and carbenicillin, we used random replacement mutagenesis to randomize all of the amino acids of the  loop. Screening of the libraries with penicillins and the expanded-spectrum cephalosporin ceftazidime permitted the identification of subtle differences between ampicillin and carbenicillin hydrolysis, the former activity being more affected by amino acid substitution. It was also found that the mechanisms modulating ceftazidime hydrolysis were different in PSE-4 from those in TEM-1.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Bacterial strains and plasmids. The cloning vector pBGS18+ (23), a construct carrying the aphaA gene encoding aminoside phosphotransferase, lacZ peptide, the pBR322 origin of replication, and the SspI fragment of bacteriophage f1, was used for cloning procedures. The pMON711 phagemid used for mutagenesis contains a 1.1-kb EcoRI-HindIII fragment carrying the bla gene expressing PSE-4, cloned from the original Dalgleish strain. Escherichia coli CJ236 [dut ung thi relA pCJ105 (Cm^r)] was used to prepare uracilated single-stranded DNA for mutagenesis. E. coli JM101 [supE thi (lac-proAB) F' (traD36 proAB+ lacI^q lacZM15)] was used to produce -lactamases in large quantities for purification procedures, and E. coli DH5 F' [supE44 lacU169 (80, lacZM15) hsdR17 recA1 endA1 gyrA96 thi-1 relA1) was used to prepare plasmid DNA for sequencing, antibiotic susceptibility testing, and -lactamase expression studies.

Antibiotics, restriction enzymes, and biochemicals. All restriction and DNA-modifying enzymes were purchased from New England Biolabs, Ltd., Mississauga, Ontario, Canada. -Lactams, including nitrocefin, were purchased from Sigma Diagnostics Canada, Mississauga, Ontario, Canada, and Oxoid, Unipath, Ltd., Nepean, Ontario, Canada, respectively. The mutagenesis procedure was performed with the Muta-Gene Phagemid in vitro mutagenesis kit, second version (Bio-Rad Laboratories, Ltd., Mississauga, Ontario, Canada). In vitro protein synthesis was done with the E. coli S-30 extract system for circular DNA (Promega Corp., Madison, Wis.) distributed by Fisher Canada, Nepean, Ontario, Canada). [³⁵S]methionine and a ¹⁴C-labeled protein ladder were purchased from ICN Biomedicals, St-Laurent, Québec, Canada, and GIBCO-BRL, Burlington, Ontario, Canada, respectively.

Construction and screenings of random libraries. The positions 162 to 164 162-164, 165-167, 168-170, 171-173, 174-176, and 177-179 random libraries were constructed by random replacement mutagenesis as described in detail by Petrosino and Palzkill (21). Mutagenic and priming oligonucleotides were synthesized with a DNA/RNA synthesizer, model 394 (Perkin-Elmer, Applied Biosystems Division, Foster City, Calif.), and purified with the EasyPrep-Oligo Prep Kit (Pharmacia Biotech, Baie d'Urfé, Québec, Canada). The restriction sites used to inactivate the bla gene in the first step of mutagenesis were BamHI and SpeI for the 162-164, 165-167, 168-170, 171-173, 177-179, and 174-176 libraries, respectively. In order to determine substrate specificity, equal amounts of plasmid DNA (100 ng) of each random library were transformed into E. coli DH5 by electroporation and plated on tryptic soy agar containing either ampicillin, carbenicillin, ceftazidime at various concentrations, or kanamycin at 50 µg/ml to select all amino acid sequences present in each library.

DNA preparation and sequencing. Plasmid single-stranded DNA for mutagenesis and plasmid double stranded DNA for sequencing were prepared by standard procedures (22). Sequencing was done by the dye terminator cycle sequencing technique with AmpliTaq DNA polymerase (Perkin-Elmer, Applied Biosystems Division), and DNA fragments were separated with an ABI Stretch 373 system (Perkin-Elmer, Applied Biosystems Division). DNA sequence analyses were done on a Sun Sparcs 1000C workstation with Genetics Computer Group software (GCG, version 9.0) of the University of Wisconsin and on a Macintosh computer with ABI software (Factura, Sequence Navigator, and AutoAssembler).

Antibiotic susceptibility testing. MICs were determined by the broth microdilution method. An inoculum of 10⁵ CFU of E. coli DH5 cells expressing either the wild type or mutant PSE-4 was prepared by dilution and inoculated into microtiter wells containing 100 µl of Mueller-Hinton broth containing binary dilutions of an antibiotic. The ranges of concentrations tested varied from 10 to 20,000 µg/ml for carbenicillin and ampicillin and 0.03 to 30 µg/ml for ceftazidime. The plates were incubated for 24 h at 37°C. The lowest concentration which inhibited bacterial growth, as monitored by visual inspection, was recorded as the MIC. Quality control was done with the E. coli reference strain, ATCC 25922.

Mutant enzyme expression. Expression of mutant -lactamase in bacterial cells was detected by immunoblotting. Cultures of 30 ml of tryptic soy broth were inoculated with E. coli DH5 expressing mutant -lactamases and incubated overnight. Cells were centrifuged and resuspended with 5 ml of 50 mM sodium phosphate buffer (pH 7.0). The suspension was then subjected to a sonication treatment (30-s) burst at 30% of maximum power. Cell debris was removed by centrifugation, and the lysate was recovered and stored at 20°C. Equal amounts of protein (20 µg) were loaded onto a sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) gel prepared by standard procedures (22). Electrophoresis was run for 1.5 h at 17 V/cm (Mini-PROTEAN II; Bio-Rad). Proteins separated by SDS-PAGE were transferred to polyvinylidene difluoride membranes by electroblotting with a TRANS-BLOT cell (Bio-Rad). -Lactamases were detected with anti-PSE-4 polyclonal antibodies (1/20,000 dilution) prepared in New Zealand White Rabbits and identified with a mouse anti-rabbit immunoglobulin G (1/5,000 dilution) coupled to alkaline phosphatase (Protoblot System; Promega Corp. [distributed by Fisher Canada, Ottawa, Canada]).

In vitro protein synthesis. To determine if the mutant genes were being transcribed and translated properly into well-folded proteins, we used the E. coli S-30 extract system for circular DNA for in vitro protein synthesis. Each reaction mixture contained 4 µg of plasmid DNA combined with amino acid mixtures minus methionine (5 µl), S30 premix (5 µl), S30 extract (circular) (15 µl), and 1 µl of [³⁵S]methionine (1,200 Ci/mmol at 15 mCi/ml). Reaction mixtures were incubated for 1 to 2 h at 37°C. Controls included vectors pBGS19+ and pBESTluc DNA.

Purification of -lactamases. -Lactamases were prepared from 6 liters of Terrific Broth (Difco Laboratories, Detroit, Mich.) cultures of E. coli JM101 expressing wild-type or selected mutant enzymes supplemented with 50 µg of kanamycin per ml and 50 µg of ampicillin per ml. After 3 h of incubation at 37°C and shaking (200 rpm), IPTG (isopropyl--D-thiogalactopyranoside) was added to 1 mM final concentration, and incubation was done overnight under the same conditions. The -lactamases were isolated by an osmotic shock procedure (17). Cells and cellular debris were removed by centrifugation (8 min, 8,000 × g). The supernatant containing 20 mg of -lactamase was applied to PD-10 columns (Pharmacia Biotech) preequilibrated with 20 mM Tris-Cl (pH 8.0) buffer and then was eluted with 3.5 ml of the same buffer. Pooled -lactamases were applied to an Econo-Pac Q anionic-exchange column (Bio-Rad) and were separated with a linear salt gradient (0 to 70 mM NaCl) for 64 min at a flow rate of 2 ml/min (ConSep LC100; Millipore, Ltd., Mississauga, Ontario, Canada). -Lactamase activity in 1.5-ml fractions was assayed with 10 µl of nitrocefin (969 µM), and proteins were visualized by SDS-PAGE analysis. Fractions having detectable activity were pooled and concentrated by ultrafiltration with Centriprep 10 and Centricon 10 filters (Amicon Canada, Ltd., Oakville, Ontario, Canada). Concentrated -lactamases were further purified by gel filtration chromatography with a Hiprep Sephacryl S-100 column (Pharmacia Biotech). -Lactamases were eluted with 50 mM sodium phosphate buffer at pH 7.0 at a flow rate of 1 ml/min. Fractions having -lactamase activity were pooled and concentrated as described above, and purity was estimated by SDS-PAGE and NIH image software analysis (version 1.60). Calculation of the mean densities of protein bands indicated greater than 80% purity.

Nitrocefin assay and enzyme kinetics. Protein concentrations were determined with an adapted Bradford assay with a Bio-Rad concentrated reagent for protein assays and the macrodilution procedure (Bio-Rad, Mississauga, Ontario, Canada) and determined by linear regression analysis (Excel; Microsoft Canada, Inc., Mississauga, Ontario, Canada). Three independent measurements of concentrations were determined from the mean of three standard plots. To measure the level of -lactamase activity of the mutant enzymes expressed in E. coli DH5, 20 µl of a nitrocefin solution was added to 20 µl of cell lysates. The time required to change the color from yellow to red was recorded. Hydrolysis of -lactams was measured for purified wild-type and mutant -lactamases by spectrophotometry (Caryl spectrophotometer; Varian Australia, Pty., Ltd., Australia). All assays were done at 30°C in 50 mM sodium phosphate at pH 7.0 with 1.0-cm quartz cuvettes in a final volume of 1 ml. Hydrolyses for ampicillin ( = 912 M¹ cm¹), carbenicillin ( = 1,190 M¹ cm¹), and cefotaxime ( = 3,749 M¹ cm¹) were monitored at 232, 232, and 280 nm, respectively. Initial velocities were determined with substrate concentrations ranging from 5 to 1,000 µM. Each of the measurements was done three times. Kinetic parameters V_max and K_m were determined by a least-squares method combined with a dynamic weighting system (LEONORA, Analysis of Enzyme Kinetic Data software; Oxford University Press, 1995). Turnover number, k_cat was calculated by dividing V_max by the concentration of protein. Errors on measurements were calculated by algebraic and propagation rules. Since the purity of the mutant proteins was 80%, the numbers for k_cat are approximations.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Construction of random libraries. Randomization of the entire  loop was accomplished by randomizing three codons at a time for six separate regions of the loop, yielding six randomized libraries (162-164, 165-167, 168-170, 171-173, 174-176, and 177-179). Considering the total number of transformants obtained per the libraries, we used the Poisson distribution and determined that the least common amino acid sequence (W-W-W) had a probability of 99.9% of being present at least once in each library.

Substrate specificity and tolerance of mutations. The libraries constructed were screened with different -lactams at various concentrations. We reasoned that this would allow us to determine the amino acid sequence requirements for the maintenance of the enzyme's function with respect to the selection agents. More precisely, we used carbenicillin and ampicillin for screening as a first step of the experiments and to examine the possibility that some amino acids in the  loop could provide specificity toward carbenicillin or ampicillin hydrolysis. Overall, there were no dramatic differences in the total number of functional amino acid sequences observed in mutants when ampicillin or carbenicillin was used at low concentrations (10 µg/ml other) than in the 177-179 region (53 versus 1%, respectively) (Table 1). At a higher antibiotic concentration, differences in the number of bacterial colonies between the 162-164 and 165-167 libraries of 10-fold appeared. In addition, the selective pressure exerted by ampicillin at a concentration of 10 µg/ml yielded similar percentages of mutants (having functional amino acid sequences) compared to carbenicillin with a concentration 100-fold higher (1 mg/ml). The screening results clearly demonstrated that the amino acid sequence requirement for wild-type PSE-4 levels of activity with ampicillin or with carbenicillin was more stringent for the positions 162 to 170 and 174 to 179 and suggested that amino acids in these regions are intolerant of substitutions. In contrast, the amino acid acids at positions 171 to 173 were more tolerant of mutations (e.g., there was a high percentage of functional amino acid sequence replacement). In a second round of screening, we determined if PSE-4 could extend substrate specificity to expanded-spectrum cephalosporins. The six  loop libraries were screened to find functional proteins by using ceftazidime as a selection agent with concentrations ranging from 0.5 to 1 µg/ml. At these concentrations, an E. coli strain expressing a wild-type PSE-4 was not able to grow, while mutants that were isolated were inferred to have extended substrate specificity. No mutants were obtained with 1 µg of ceftazidime per ml. By using 0.5 µg/ml, we selected mutants from the 165-167, 168-170, 171-173, and 177-179 libraries. Surprisingly, no PSE-4 mutants were obtained from the 162-164 and 174-176 libraries. The highest percentage (19%) of functional proteins selected with ceftazidime were obtained with the 177-179 library, while the smallest value (0.02%) was obtained with the 171-173 library, a difference of 1,000-fold. Furthermore, the percentage of functional amino acid sequences selected with ceftazidime in the 165-167, 168-170, and 177-179 libraries was 10- to 30-fold higher than the percentage of functional amino acid sequences obtained with high concentrations of carbenicillin or ampicillin.

View this table: [in this window] [in a new window]   TABLE 1.   Percentages of functional amino acid sequences for six randomized libraries of the  loop of PSE-4 selected with different -lactams at various concentrations

Sequence analysis of functional mutants. A total of 46 PSE-4 functional mutants selected with ampicillin (1,500 µg/ml) or carbenicillin (5,000 µg/ml) were completely sequenced from both DNA strands for the bla_PSE-4 genes. In general, the amino acid sequence requirements for a wild-type level of activity toward both penicillins were found to accommodate large variations among amino acid sequence (e.g., 14 amino acids out of a total of 18 were tolerant of substitutions) (Fig. 1). The four invariant residues were R164, E166, R178, and D179. D176 was rarely substituted; only 1 mutant out of 13 had an amino acid substitution of E for D, as seen in Table 3. The conservative nature of this unique substitution supports the importance of having an acidic amino acid at this position to yield a functional enzyme. Amino acid sequence requirements for a higher than wild-type level of activity toward ceftazidime were not stringent, because double and triple amino acid substitutions were obtained; we noted that G residues were frequently found in mutants resistant to ceftazidime.

View larger version (9K): [in this window] [in a new window]   FIG. 1.   Amino acid sequence variability of  loop mutants selected with ampicillin (1.5 mg/ml) or carbenicillin (5 mg/ml) (top) and ceftazidime 0.5 (µg/ml) (bottom). The wild-type PSE-4 -lactamase sequence is shown in boldface. Asterisks indicate no functional amino acid sequences found with ceftazidime as a selection agent.

Antibiotic susceptibility. To measure the level of susceptibility toward ampicillin, carbenicillin, and ceftazidime conferred by -lactamase mutants, we determined the MICs for E. coli expressing these enzymes. All of the mutant enzymes selected with carbenicillin at 5,000 µg/ml or ampicillin at 1,500 µg/ml still provided a high level of resistance toward both penicillins for E. coli cells (Tables 2 and 3). Furthermore, it seemed a general rule that these mutations in the  loop had a greater effect on ampicillin resistance (2- to 12-fold decrease of MICs) than on carbenicillin resistance (2- to 4-fold decrease). High-level resistance toward ceftazidime was observed with mutants selected with this antibiotic at a concentration of 0.5 µg/ml. MICs were 2- to 16-fold higher than those for the wild-type PSE-4 (Tables 2 and 3). In most cases, an increase in ceftazidime resistance was followed by a drastic increase in ampicillin and carbenicillin susceptibility. The 171DRK173-ceftazidime mutant was the unique exception in that group, showing similar levels of susceptibility to both penicillins compared to the wild type (Table 3). This observation is consistent with the high percentage of functional sequences obtained when the 171-173 library was screened with penicillins (Table 1).

View this table: [in this window] [in a new window]   TABLE 2.   Amino acid sequences, MICs, and -lactamase activity of PSE-4 mutants selected from the 162-164, 165-167, and 168-170 libraries with different -lactams

View this table: [in this window] [in a new window]   TABLE 3.   Amino acid sequences, MICs, and -lactamase activity of PSE-4 mutants selected from 171-173, 174-176, and 177-179 libraries with different -lactams

Mutant -lactamase expression. The total protein content of E. coli expressing mutated -lactamases was used for immunoblotting analysis to verify the effect of the mutation on the level of expression for these enzymes. Different levels of expression were observed from mutants selected in each randomized library screened with penicillins (Fig. 2) and varied from low to wild-type levels of expression. The majority of mutant -lactamases selected from the 162-164, 165-167, 168-170, 171-173, and 174-176 libraries were generally expressed at low levels compared to the wild type (Fig. 2). Amino acids in the  loop that were more affected by substitutions (e.g., yielding poorly expressed enzymes) were localized in the 162-164 and 174-176 libraries. In the former, the majority of the mutant enzymes were poorly expressed, and some were barely detectable (Fig. 2A, lanes 2, 4, and 5). The substitution of L162 and D163 with other amino acids was detrimental to enzyme expression. The hydrophobic character at position 162 seemed to be essential for steady-state expression of the enzyme. Moreover, loss of the acidic residue D163 was also detrimental to enzyme expression. The presence of its carboxylate side chain seemed to be important for the structural integrity of PSE-4. In the 174-176 library, some of the enzymes were undetectable. The only expressed enzymes were those with changes VGD, HED, and DLD (Fig. 2E, lanes 1, 2, and 5). The presence of a G at position 175 appeared to be a prerequisite for a well-expressed enzyme. Wild-type levels of expression for the enzymes having HED and DLD changes were more difficult to explain. Comparisons between the DLD mutant and the unstable VLD mutant indicated that the substitution of L by D presumably stabilized the loss of G175.

View larger version (40K): [in this window] [in a new window]   FIG. 2.   Expression levels of wild-type PSE-4 and mutant -lactamases selected with ampicillin (1.5 mg/ml) or carbenicillin (5 mg/ml) examined by immunoblotting. The -lactamases were detected with anti-PSE-4 polyclonal antibodies. The numbers followed by asterisks indicate mutants selected with ceftazidime (0.5 µg/ml). (A) Mutants from the 162-164 library. Lanes: +, wild-type PSE-4 -lactamase; 1, MDR; 2, LSR; 3, IDR; 4, LNR; 5, TDR. (B) Mutants from the 165-167 library. Lanes: 1, WVG; 2, LGG; 3, LYP; 4, WEP; 5, LET; 6, WEA; 7, GEP. (C) Mutants from the 168-170 library. Lanes: 1, GQG; 2, GAG; 3, EMG; 4, ELG; 5, VVN; 6, VMN. (D) Mutants from the 171-173 library. Lanes: 1, GGA; 2, GGG; 3, KGG; 4, AGV; 5, GWG; 6, VVR; 7, EVW; 8, DGA. (E) Mutants from the 174-176 library. Lanes: 1, VGD; 2, HED; 3, RNE; 4, LGD; 5, DLD; 6, LSD; 7, LVD. (F) Mutants from the 177-179 library. Lanes: 1, RRD; 2, ERD; 3, VRD; 4, GRD; 5, ARD; 6, GGG; +, wild type.

View larger version (41K): [in this window] [in a new window]   FIG. 3.   Transcription and translation efficiency of the mutated bla genes from mutants of the 162-164 library. The proteins were synthesized in vitro with the E. coli extract S-30 system and analyzed by autoradiogram (A) and immunoblotting (B). The 30-kDa protein synthesized with pBGS19+ plasmid DNA corresponded to the product of the aphA gene of the vector. Lanes: , negative control (no DNA); 1, pBGS19+ (vector); 2, pMON70016 (TDR); 3, pMON70017 (LNR); 4, pMON70018 (LSR); 5, pMON70020 (LSR); +, pMON711 (wild type). The two LSR mutants have different serine codons, pMON70018 (AGC) and pMON70020 (TCC).

-Lactamase activity and enzyme kinetics. From high levels of -lactamase hydrolytic activity to very low levels were detected, depending on the region of the  loop in which mutations occurred, the type of amino acid substitutions done, and the selection agents used (e.g., penicillin versus cephalosporin) (Tables 2 and 3). The results indicated that no direct correlations existed between ampicillin and carbenicillin MICs and -lactamase activity measured in vitro. Mutants with mutations in the 168 to 170 (VVN and VMN) and 171 to 173 (GGA, GGG, KGG, GWG, EVW, DGA, and ASV) regions showed relatively unchanged MICs, but the mutants showed a decrease in -lactamase activity compared to that of the wild type. Thus, wild-type levels of -lactamase activity were not required to give high levels of resistance to both ampicillin and carbenicillin. Ceftazidime mutants showed no detectable -lactamase activity, and this phenotype correlated with a drastic increase in ampicillin and carbenicillin susceptibility.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

Random replacement mutagenesis was used to identify amino acids in the  loop that could modulate substrate specificity and to examine the effect of amino acid substitutions on the level of PSE-4 expression. Screening of the six  loop libraries with high concentrations (1 mg/ml) of ampicillin or carbenicillin yielded differences in the percentage of functional -lactamases. The 162-164 and 165-167 libraries gave a 10-fold-higher percentage of functional amino acid sequences with the carbenicillin selection compared with the ampicillin selection. This finding indicated that the sequence requirements for a wild-type level of resistance toward both penicillins were more stringent for ampicillin than for carbenicillin at these positions. This particularity seemed to apply to specific regions of the  loop, especially the 162 to 164 and 165 to 167 regions. Furthermore, the selective pressure exerted by 10 µg of ampicillin per ml was equivalent to a 100-fold-higher concentration of carbenicillin. With these results, we hypothesized that ampicillin hydrolysis was affected more by substitutions and that specific interactions must be present to maintain efficient hydrolysis. In contrast, carbenicillin hydrolysis required fewer specific interactions with amino acid residues in the  loop and possibly other residues near or in the active site. Sequence analysis of functional mutants selected with ampicillin or with carbenicillin identified four invariant residues, R164, E166, R178, and D179, which are critical for ampicillin or carbenicillin hydrolysis. N170 and D176 also appeared to be important for a wild-type level of function. These residues were previously identified from high-resolution X-ray structures (10, 12, 13, 24, 29) and site-directed mutagenesis studies (1, 7, 8, 14, 15, 26, 29) to be implicated in the catalytic activity and active-site conformation of -lactamases.

Contrasting of the sequence requirements (e.g., amino acid sequences) for function with high concentrations of ampicillin or carbenicillin did not permit the identification of a single amino acid in the  loop that could interact specifically with either one of the penicillins studied. The small number of mutants sequenced for each selection could explain the lack of correlation between the percentage of functional amino acid sequences and the sequence requirements for function. Furthermore, the MICs of ampicillin and carbenicillin for the mutants isolated with high concentrations of either penicillin were affected in a similar manner by the substitution; this was observed for the majority of mutants. There appears to be no major determinant of specificity in the  loop of PSE-4 that could discriminate for the preferential hydrolysis of ampicillin over carbenicillin.

To investigate the potential of PSE-4 to extend its substrate specificity toward expanded-spectrum cephalosporins, we screened the six  loop libraries with ceftazidime at a concentration of 0.5 µg/ml. This concentration restrained the growth of E. coli cells expressing wild-type PSE-4. Thus, potential screens were considered to have an extended-spectrum profile and were assumed to be able to hydrolyze ceftazidime more efficiently. Functional amino acid sequences consistent with a higher-than-wild-type level of resistance toward ceftazidime were identified in a precise region of the  loop. In contrast to TEM-1 -lactamase, the structural modification in the 162-164 and 174-176 regions of PSE-4 by amino acid substitutions was not sufficient to increase the level of ceftazidime hydrolysis.

The structural basis for this observation is difficult to explain without crystallographic data for PSE-4. The amino acids implicated in salt bridge formation in the TEM-1  loop (K73: E166, R161: D163, R164: E171, R164: D179, and D176: R178) (12) are conserved in PSE-4 (Fig. 4). From this homology, it was predicted that similar interactions would be present in the PSE-4  loop; removal of these structural constraints would allow the entry of the bulky side chain of ceftazidime, as was postulated for TEM-1 ceftazidime mutants (19, 21, 25). Two tentative explanations could be envisaged: first, the conformation and the network of interactions between the amino acids of the  loop of PSE-4 are different from those of TEM-1. Second, the need for specific interactions with ceftazidime is required for efficient hydrolysis. An additional piece of data supporting the structural differences in the  loop between PSE-4 and TEM is the lower level of expression for PSE-4 mutants selected on ampicillin, while TEM-1 ampicillin mutants were found to be expressed at levels similar to that of the native enzyme (19, 21). Thus, it is possible that other types of amino acid interactions are as important as the putative salt bridge interaction in the stabilization of the PSE-4 structure. The sequence requirements for a stably expressed protein are more stringent for PSE-4 than for TEM-1. Low levels of protein expression resulting from a single or multiple mutations could be due to many factors, such as thermodynamic alterations, altered folding kinetics, or increased protease susceptibility resulting from amino acids exposed to the solvent. Since a correlation between enzyme susceptibility to proteases and thermal stability exists (20), it is convenient to think that such phenomenon explains the observations with the majority of the  loop PSE-4 mutations. Further experiments measuring protein stability will be essential to identify the exact causes of low levels of expression in these PSE-4 mutants.

View larger version (5K): [in this window] [in a new window]   FIG. 4.   Sequence alignment of amino acids from the PSE-4 and TEM-1  loops.

Sequence requirements for clinically relevant levels of ceftazidime resistance (16 to 32 µg/ml) were found to be stringent in some TEM-1 mutants. These functional sequences were only found in the 165 to 167 region. Determinants A, Y or F165, Y or H166, and G167 only have been described previously (19). Such specificity determinants were observed with ceftazidime mutants selected from the 165-167 library; a G167 was found in two mutants, while a Y was found at position 165. However, these new observations are quite unique and merit further investigations to determine if these changes have the same effect in other class A enzymes.

The catalytic activity of mutants selected on ampicillin and carbenicillin was affected more when changes were in specific regions of the  loop, especially the 162 to 164, 168 to 170, and 171 to 173 positions. We specify that 100% of wild-type -lactamase activity was not a prerequisite for high levels of ampicillin and carbenicillin resistance. Since a direct correlation exists between the amount of enzyme and the level of activity, the decreased in vitro activity seen for the majority of mutants was due to the low level of -lactamase expression. Kinetic analysis of the 162MDR164 mutant revealed that the substitution enhanced the apparent affinity and reduced the catalytic activity of the enzyme. The relief of critical structural interactions in the  loop could disturb the correct positioning of chemical groups involved in the catalysis of substrates. This hypothesis was introduced previously in the random replacement mutagenesis studies of TEM-1 derivatives that were able to hydrolyze ceftazidime and that showed no activity toward the preferred substrate (19, 21). Further evidence to support this hypothesis comes from structural data extracted from the crystal of the P54 -lactamase mutant of Staphylococcus aureus (10). It was found that the elimination of the salt bridge between R164 and D179 by substituting for the latter with an N substantially disordered the  loop and resulted in a drastic decrease in activity. It was also found that deacylation of the acyl-enzyme complex of penicillin G and P54 enzyme was the rate-limiting step. Moreover, kinetic analysis of cefepime hydrolysis by the D179G and R164N mutant variants of TEM-1 -lactamase showed lowered values for the dissociation constants K_s for both mutants (25). Circular dichroic analysis of the R164N enzyme indicated a decrease in helicity for this mutant compared to the native structure. The structural modifications of the active site were proposed to accommodate the kinetic and the structural data (25).

In conclusion, the data obtained from this study confirmed the important roles of R164, E166, N170, D176, R178, and D179 in the structure and function of PSE-4. Sequence requirements of the  loop consistent with a stably expressed protein were more stringent for PSE-4 than for TEM-1. The determinants of carbenicillin specificity were not found in the  loop region of PSE-4, and, finally, the mechanism responsible for substrate specificity toward expanded-spectrum cephalosporins of PSE-4 seemed to be quite different from that of TEM-1.