Prediction of the Evolution of Ceftazidime Resistance in Extended-Spectrum β-Lactamase CTX-M-9

ABSTRACT A random mutagenesis technique was used to predict the evolutionary potential of β-lactamase CTX-M-9 toward the acquisition of improved catalytic activity against ceftazidime. Thirty CTX-M mutants were obtained during three rounds of mutagenesis. These mutants conferred 1- to 128-fold-higher MICs of ceftazidime than the parental enzyme CTX-M-9. The CTX-M mutants contained one to six amino acid substitutions. Mutants harbored the substitutions Asp240Gly and Pro167Ser, which were previously observed in clinical CTX-M enzymes. Additional substitutions, notably Arg164His, Asp179Gly, and Arg276Ser, were observed near the active site. The kinetic constants of the three most active mutants revealed two distinct ways of improving catalytic efficiency against ceftazidime. One enzyme had a 17-fold-higher kcat value than CTX-M-9 against ceftazidime. The other two had 75- to 300-fold-lower Km values than CTX-M-9 against ceftazidime. The current emergence of CTX-M β-lactamases with improved activity against ceftazidime may therefore be the beginning of an evolutionary process which might subsequently generate a great diversity of CTX-M-type ceftazidimases.

The Netherlands) as recommended by the manufacturer. Briefly, 10 ng of plasmid was amplified by PCR with 200 M each deoxynucleoside triphosphate, 125 ng primers pBK-CMV1Ј and pBK-CMV2Ј (Table 1), and 2.5 U Mutazyme DNA polymerase. The primers included restriction sites for enzymes KpnI and EcoRI. The mutagenized amplicon was cleaved with these endonucleases (Roche Applied Science, Meylan, France) and cloned into the corresponding restriction sites of plasmid pBK-CMV downstream of the lacZ promoter. The recombinant plasmids were transformed into E. coli ⌬ampC, which was then cultured in 1 ml LB broth at 37°C for 1 h.
One hundred microliters of transformed cells was spread on Mueller-Hinton plates supplemented with 32 g/ml amoxicillin and 20 g/ml kanamycin to evaluate the total number of CTX-M-producing strains which were screened during the mutagenesis experiment. In parallel, 900 l of recombinants was selected for growth on Mueller-Hinton agar with increased concentrations of ceftazidime.
Three rounds of mutagenesis were performed consecutively. About 20,000 ␤-lactamase-producing clones were screened by using 4 g/ml ceftazidime during the first round. The mutant candidates for the second and third rounds were selected on the basis of inhibition diameters of Ն21 mm for cefotaxime or Ն14 mm for ceftazidime and a number of amino acid substitutions of Յ3 in order to increase the diversity of the recovered substitutions. During the second round, about 10,000 ␤-lactamase-producing clones were screened from six candidate mutants on agar containing 8 g/ml of ceftazidime. During the third round, about 10,000 ␤-lactamase-producing clones were screened from four candidate mutants on agar containing 16 g/ml of ceftazidime.
Selection of mutants for sequencing. The resistant phenotype of CTX-M mutant-harboring clones was established for ␤-lactams and kanamycin by the disk diffusion method. These clones were classified into two resistance phenotypes on the basis of inhibition diameters for cefotaxime and ceftazidime. Phenotype 1 exhibited similar inhibition diameters for ceftazidime and cefotaxime, and phenotype 2 exhibited lower inhibition diameters for ceftazidime than for cefotaxime. In both these phenotypes, the clones can be subgrouped on the basis of the inhibition diameter values for ␤-lactams. bla CTX-M genes were sequenced from two mutants in each phenotype group.
The sequences were determined by direct sequencing of PCR products, using the dideoxy chain termination procedure with an Applied Biosystems sequencer (ABI 377) (34). The nucleotide and deduced protein sequences were analyzed using software available at the website of the National Center for Biotechnology Information (http://www.ncbi.nlm.nih.gov). The ClustalW program (http: //infobiogen.fr) was used for alignment of multiple protein sequences (39).
Susceptibility to ␤-lactams. Antibiotic-containing disks were used for antibiotic susceptibility testing by the disk diffusion assay (Sanofi-Diagnostics Pasteur, Marnes la Coquette, France). MICs were determined by a microdilution method on Mueller-Hinton agar (Sanofi Diagnostics Pasteur, Marnes la Coquette, France) with an inoculum of 10 4 CFU per spot. MICs of ␤-lactam antibiotics were determined alone and combined at a fixed concentration of clavulanic acid (2 g/ml) or tazobactam (4 g/ml). Antibiotics were provided as powders by Glaxo SmithKline (amoxicillin, ticarcillin, cefuroxime, ceftazidime, and clavulanic acid), Wyeth Laboratories (piperacillin and tazobactam), Eli Lilly (cephalothin), Roussel-Uclaf (cefotaxime and cefpirome), Bristol-Myers-Squibb (aztreonam and cefepime), and Merck Sharp and Dohme-Chibret (imipenem).  ␤-Lactamase preparation. The mutant genes were amplified by PCR from plasmid pBK-CMV with primers NdeORFM9-A and SfiORFM9-B, which included restriction sites for enzymes NdeI and SfiI ( Table 1). The PCR products were digested with these two enzymes and were ligated into the corresponding restriction sites of modified plasmid pET9a (24). The resulting plasmids were transformed into E. coli BL21(DE3). E. coli transformants were selected on Mueller-Hinton agar supplemented with 30 g kanamycin and 0.5 g ceftazidime.
CTX-M mutant-encoding genes were overexpressed in Escherichia coli BL21(DE3) from pET9 derivative plasmids in 2xYT broth (Qbiogene, Irvine, CA) supplemented with 0.1 mM isopropyl-␤-D-thiogalactopyranoside (Sigma Chemical Co., St Louis, Mo.), as previously reported (11). After extraction of the enzymes by sonication, the extract was clarified by centrifugation and treatment with DNase I (Roche Applied Science, Meylan, France). CTX-M purification was carried out as previously described (6) by ion-exchange chromatography with an SP Sepharose column (Amersham Pharmacia Biotech) and gel filtration chromatography with a Superose 12 column (Amersham Pharmacia Biotech), using a fast protein liquid chromatography system. The total protein concentration was estimated by the Bio-Rad protein assay (Bio-Rad, Richmond, Calif.), with bovine serum albumin (Sigma Chemical Co., St Louis, Mo.) used as a standard. The purities of enzyme extracts were estimated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (6).
Determination of ␤-lactamase kinetic constants. Steady-state kinetic parameters were determined for mutants exhibiting ceftazidime, cefotaxime, and amoxicillin MICs of Ն64, Ն16, and Ն2048 g/ml, respectively. The Michaelis constant (K m ) and catalytic activity (k cat ) were determined with purified extracts by using a computerized microacidimetric method (23). Bovine serum albumin and Triton X-100 (Sigma Chemical Co., St Louis, Mo.) were added at final concentrations of 50 g/ml and 0.01%, respectively, to prevent denaturation (20) and aggregation of enzyme M-A6B1C1.

RESULTS
Construction of E. coli ⌬ampC. An ampC gene isogenic mutant of E. coli DH5␣ was constructed to eliminate phenotypic variations which may result from AmpC production during in vitro evolution. The ampC gene was replaced by kanamycin resistance cassette in E. coli ⌬ampC::Kan r . The cassette was then eliminated by using the helper plasmid pCP20. Each step of the construction was checked by PCR (Fig. 1). The resulting strain was designated E. coli ⌬ampC.
Collection of mutants. After the first round of mutagenesis, 106 clones were able to grow in the presence of 4 g/ml ceftazidime. Phenotype 1, which exhibited similar inhibition diameters for ceftazidime and cefotaxime, possessed 81 clones. Phenotype 2, which exhibited a smaller inhibition diameter for ceftazidime than for cefotaxime, possessed 25 clones. The clones belonging to phenotype 2 presented a great diversity of inhibition diameters for the other ␤-lactams, unlike the clones belonging to phenotype 1, which formed a homogenous group. A second round of mutagenesis was performed from six candidate mutants (one from phenotype 1 and five from phenotype 2), and 123 clones were obtained on agar containing 8 g/ml ceftazidime; 22 clones belonged to phenotype 1, and 101 clones belonged to phenotype 2. The last round of mutagenesis  was performed from four candidate mutants (two from phenotype 1 and two from phenotype 2); 60 clones had resistance phenotype 1, and 39 others had resistance phenotype 2. DNA sequencing. Deduced amino acid sequences revealed 38 different substitutions and one to six substitutions per enzyme (Table 2). One to three substitutions per CTX-M mutant were acquired after each cycle of mutagenesis.
All CTX-M-9 mutants associated with resistance phenotype 1 harbored the Asp240Gly substitution, which is located in the ␤3 strand (Fig. 2). The CTX-M mutants harbored the Asp240Gly substitution alone (M-A1) or associated with substitution Val29Ala (M-A2) or Ile173Thr (M-A3) after the first round of mutagenesis. After the second round, the Asp240Gly substitution was associated with one (Ala219Asp in M-A1B2) or two (Gln87Leu and Arg276His in M-A1B1) additional substitutions. After the last round, three other substitutions appeared in mutants M-A1B2C1 (Gly289Trp, His197Arg, and Leu169Met) and M-A1B1C1 (His112Tyr, Thr230Ile, and Ala231Val).
The mutants involved in resistance phenotype 2 harbored a larger diversity of substitutions than those implicated in resistance phenotype 1. Twenty-four additional substitutions were observed ( Table 2). Substitutions at positions 167, 169, 179, and/or 164, which are located in the omega loop, appeared from the first round of mutagenesis (Fig. 2).
The Leu169Gln substitution was observed after the first round of mutagenesis alone in mutant M-A4. After the second round of mutagenesis, this substitution was associated with substitution Asp240Gly in its derivative M-A4B1. During the third round of mutagenesis, a second type of substitution, Leu169Met, was observed in mutant M-A1B2C1. This enzyme, which derived from mutant M-A1B2, additionally harbored substitutions Asp240Gly, His197Arg, Ala219Asp, and Gly289Trp. MICs of ␤-lactams. MICs of ␤-lactams were determined for the 30 clones collected for bla CTX-M sequencing (Table 3). Overall, ceftazidime MICs increased from 1 g/ml for CTX-M-9-producing E. coli ⌬ampC to 128 g/ml for the mutantproducing clones (A6B1C1). MICs of ceftazidime ranged from 8 to 16, 8 to 64, and 16 to 128 g/ml after the first, second, and third rounds of mutagenesis, respectively. Among the 30 CTX-M-producing clones, 16 had a high level of resistance to ceftazidime (MIC of Ն32 g/ml). Of these, eight were obtained after the second round of mutagenesis, and the other eight were obtained after the third round. Clone A6B1C1, which had the highest ceftazidime MIC (128 g/ml), was obtained after three cycles of mutagenesis.
For the 30 clones, the MICs of inhibitor-penicillin combinations, aztreonam, cefepime, cefpirome, and imipenem did not increase more than fourfold in comparison with those for CTX-M-9-producing E. coli ⌬ampC.
Kinetic constants. The mutants designated M-A1B1C1, M-A6B1C1, and M-A8B3, were overexpressed in E. coli BL21(DE3) from pET9a-derived plasmids and were purified by liquid chromatography. One to 3 milligrams of ␤-lactamase per liter of culture medium was obtained, and the purity was estimated to be Ͼ98%.
The kinetic parameters for these strains are shown in Table 4. M-A1B1C1 exhibited typical enzymatic features of CTX-M mutants. Lower K m values were obtained for penicillins (K m , 14 to 35 M) than for cephalosporins (140 to 450 M). Cephalothin was the best substrate (k cat , 3,800 s Ϫ1 ), and a higher k cat was observed for cefotaxime (550 s Ϫ1 ) than for ceftazidime (35 s Ϫ1 ). However, the k cat value against ceftazidime was 17-fold higher for M-A1B1C1 than for CTX-M-9. In addition, the ceftazidime K m value of M-A1B1C1 was lower than that of CTX-M-9 (450 M versus 600 M). Conversely, the kinetic constants of M-A6B1C1 and M-A8B3 were different from those of typical CTX-M enzymes, such as CTX-M-9. k cat values against penicillins and cephalothin were 7-to 25-fold lower for the two mutants than for CTX-M-9.  (17,20,26,38). All these CTX-M mutants conferred a higher level of resistance to ceftazidime than to cefotaxime. However, this substitution alone was responsible for a dramatic decrease in MICs of penicillins and other cephalosporins. Comparisons of MICs for clones A6, A6B1, and A6B1C4 suggest that substitutions Ala109Thr and Ala231Val may partially compensate for this pernicious action. The effect of substitution Pro167Ser may be due to a decrease in stability (20). Chen et al. showed that substitution Ala231Val caused a 2°C increase in the protein melting temperature of CTX-M-9 and a 1.1-kcal/mol increase in stability (11). This effect may explain the increase in MICs induced by the addition of substitution Ala231Val to substitution Pro167Ser.
(ii) Residue 240. Aspartate 240 is located at the end of the ␤3 strand, at the entrance of the binding site. The residue at position 240 is not conserved among ␤-lactamases of class A. Acid residues (Glu or Asp) are observed in TEM-1 and SHV-1 penicillinases and CTX-Ms. Residue Gly240 is present in the ESBLs PER, VEB-1, and BES-1, which have hydrolytic activity against ceftazidime (6,32). The Asp240Gly substitution was obtained in eight CTX-M mutants, as in natural enzymes CTX-  Benzylpenicillin  295  25  115  14  12  12  15  4  Amoxicillin  90  20  67  35  5.5  12  7  12  Cephalothin  3,000  150  3,800  140  420  250  200  13  Cefotaxime  450  120  550  200  17  M-15, CTX-M-16, CTX-M-27, and CTX-M-32, which exhibited improved activity against ceftazidime (4,8,16,31). CTX-M mutant M-A1 has the same amino acid sequence as CTX-M-16. Its derivative M-A1B1C1, which harbored the additional substitutions Ala231Val, Gln87Leu, His112Tyr, Thr230Ile, and Arg276His, conferred high-level resistance against all ␤-lactams, which therefore is the best compromise in the resistance phenotype. This enzyme had greater hydrolytic activity against ceftazidime (k cat , 35 s Ϫ1 ) than did natural CTX-M enzymes, including those harboring substitutions Pro167Ser and Asp240Gly. These results suggest that the natural CTX-M enzymes harboring the substitution Asp240Gly are the most probable phylum for new mutants conferring the highest level of resistance to ␤-lactams.  (19,22,35). Substitutions Arg164Ser/His and Asp179Ala/Gly/ Asn/Glu are common among TEM/SHV-type ESBLs (15,21). These changes are involved in the extension of substrate specificity to extended-spectrum cephalosporins such as ceftazidime by increasing flexibility in the omega loop. However, these substitutions have never been observed in natural CTX-M enzymes. Substitutions at positions 164 and 179 were obtained after one round of mutagenesis. During the three round of mutagenesis, the substitutions were obtained in nine mutants and always in combination with additional substitutions. The substitutions at positions 231 and/or 276 were harbored in eight of these mutants, which conferred a higher level of resistance to ceftazidime than to cefotaxime. Surprisingly, substitutions Arg164His and Asp179Asn/Tyr were both obtained simultaneously in the same mutants. Mutant M-A8B3, which harbored substitutions Arg164His and Asp179Asn in combination with Gln188Arg, Ala231Val, Arg276Ser, and Gly289Glu, conferred 64-fold higher resistance to ceftazidime than the parental enzyme CTX-M-9. However, its hydrolytic activities against oxyimino cephalosporins, including ceftazidime, were greatly inferior to those of CTX-M-9. In fact, the increase in activity against ceftazidime for M-A8B3 resulted in an impressively low K m value (2 M, versus 600 M for CTX-M-9).
(ii) Residue 169. Residue 169 is highly conserved in ␤-lactamases. In vitro mutagenesis experiments with the TEM-1 enzyme and the recent characterization of the ESBL SHV-57 showed that substitutions Leu169Pro and Leu169Arg resulted in activity against ceftazidime with a concomitantly significant compromise in the resistance to penicillins (25,40). Two different substitutions were obtained at position 169 in our CTX-M mutants, Leu169Gln and Leu169Met, which seem to confer similar behavior against penicillins and ceftazidime. These substitutions induced an inversion of the resistance phenotype against ceftazidime and cefotaxime in comparison to the phenotype induced by the parental enzyme CTX-M-9, as observed with the other mutants harboring substitutions at positions 164, 167, and 179 located in the omega loop. The association of substitution Asp240Gly with Leu169Gln in mutant A4B1 raised the MICs of ceftazidime and cefotaxime in comparison with those for mutant A4 (ceftazidime MICs, 32 versus 8 g/ml; cefotaxime MICs, 8 versus 0.5 g/ml).

Substitutions not observed among clinical CTX-M, TEM, and SHV enzymes. (i) Residue 276.
Residue 276 is in the terminal ␣-11 helix. CTX-M enzymes do not contain Arg244, which interacts with the carboxylate function of substrates and inhibitors in most ESBLs. This interaction is critical for the catalysis and the inhibition process of TEM-1 penicillinase (9). Residue Arg276 of CTX-M enzymes was predicted to be a substitute for Arg244. A substitution at position 276 was obtained during three distinct mutational events. However, the Arg276His substitution in association with substitutions Asp179Gly, Asn106Ser, and Thr86Ala caused only a modest increase in MICs against ceftazidime (8 g/ml conferred by mutants M-A7 versus 16 g/ml conferred by mutant M-A7B3). The MICs of inhibitor ␤-lactam combinations were not significantly modified by these substitutions, in contrast with the substitution at position 244 in TEM enzymes (9).
(ii) Residues 106 and 109. Residues 106 and 109 are located in the loop harboring residue 104, in the vicinity of residue 132, which are implicated in the binding of ␤-lactams in CTX-M catalytic cavity (35). Residues Asn106 and Ala109, observed in clinical CTX-M enzymes, were replaced in 15 CTX-M mutants by residues Ser106 and Thr109, which are observed in TEM and SHV enzymes (19,22,35). Residues 106 and 109 are implicated in the placing of side chains of residues 104 and 132, respectively (35). In association with the substitution Pro167Ser, Ala109Thr increased the ceftazidime MIC (32 g/ml conferred by mutant M-A6B1 versus 8 g/ml conferred by mutant M-A6). In addition, the Ala109Thr substitution seems to compensate for the pernicious effect of substitution Pro167Ser on activity against penicillins (32 to Ͼ2,048 g/ml versus 8 to 512 g/ml) and the other cephalosporins (0.5 to 128 g/ml versus 0.12 to 32 g/ml). Mutant M-A6B1C1, which harbored the association Pro167Ser and Ala109Thr in combination with substitutions Gly146Arg, Thr227Ala, and Gln254Pro, had the highest MIC of ceftazidime (128 g/ml). This was due to the very low K m (8 M) of the enzyme with regard to ceftazidime, as observed with mutant M-A8B3.
(iii) Residue 166. The role of Glu166 in both the acylation and deacylation steps of TEM ␤-lactamases has been well argued from kinetic and modeling studies (27,28). Recent crystallographic data indicate that the Lys73 side chain may replace Glu166 in the acylation step of CTX-M ␤-lactamases (11,12,18,37). In this case, a substitution at position 166 would inactivate the CTX-M enzyme. The Glu166Val-harboring mutant M-A6B1C2 conferred a dramatic decrease in the MICs of ␤-lactams, except for ceftazidime (32 g/ml), as observed with the Glu166Ala-harboring mutant of PER-1 (7). The unexpected activity of M-A6B1C2 against ceftazidime suggests the intervention of other residues in the process of deacylation of ceftazidime in this mutant, which harbored the additional substitutions Pro167Ser and Ala109Thr. Recent crystallographic data showed that residue Lys73, which is able to interact with residue Ser70 and the catalytic water molecule, was greatly mobile in the enzyme CTX-M-9 (11). Hydrolysis might be catalyzed in M-A6B1C2 by a symmetric process mediated by residue Lys73, which may activate residue Ser70 during the acylation step and the catalytic water molecule during the deacylation step. However, this process seems to be substrate dependent and efficient only against ceftazidime.
In conclusion, 30 clones with ceftazidime MICs (8 to 128 VOL. 50, 2006 IN VITRO EVOLUTION OF CTX-M-9 AGAINST CEFTAZIDIME g/ml) higher than those induced by the CTX-M-9 enzyme (1 g/ml) were obtained. The substitutions involved in this in vitro evolution include both those previously observed in natural CTX-M enzymes and new substitutions. This wide diversity emphasizes the evolutionary potential of CTX-M enzymes. The substitutions may emerge in the future under selection pressure driven by the therapeutic use of ceftazidime.