Haemophilus influenzae bla_ROB-1 Mutations in Hypermutagenic ΔampC Escherichia coli Conferring Resistance to Cefotaxime and β-Lactamase Inhibitors and Increased Susceptibility to Cefaclor

Abstract

The clinical use of cefaclor has been shown to enrich Haemophilus influenzae populations harboring cefaclor-hydrolyzing ROB-1 β-lactamase. Such a selective process may lead to the increased use of extended-spectrum cephalosporins or β-lactams plus β-lactamase inhibitors and, eventually, resistance to these agents, which has not previously been observed in H. influenzae. In order to establish which bla_ROB-1 mutations, if any, could confer resistance to extended-spectrum cephalosporins and/or to β-lactamase inhibitors, a plasmid harboring bla_ROB-1 was transformed into hypermutagenic strain Escherichia coli GB20 (ΔampC mutS::Tn10), and this construct was used in place of H. influenzae bla_ROB-1. Strain GB20 with the cloned gene was submitted to serial passages in tubes containing broth with increasing concentrations of selected β-lactams (cefotaxime or amoxicillin-clavulanate). Different mutations in the bla_ROB-1 gene were obtained during the passages in the presence of the different concentrations of the selective agents. Mutants resistant to extended-spectrum cephalosporins harbored either the Leu169→Ser169 or the Arg164→Trp164 substitution or the double amino acid change Arg164→Trp164 and Ala237→Thr237. ROB-1 mutants that were resistant to β-lactams plus β-lactamase inhibitors and that harbored the Arg244→Cys244 or the Ser130→Gly130 replacement were also obtained. The cefaclor-hydrolyzing efficiencies of the ROB-1 variants were strongly decreased in all mutants, suggesting that if bla_ROB-1 mutants were selected by cefaclor, this drug would prevent the further evolution of this β-lactamase toward molecular forms able to resist extended-spectrum cephalosporins or β-lactamase inhibitors.

The introduction of conjugate vaccines effective against Haemophilus influenzae type b has influenced both the prevalence of infections caused by this respiratory pathogen (50) and the rates of antibiotic resistance. Nevertheless, H. influenzae is still an important causative agent of acute otitis media, pneumonia, and chronic bronchitis. Different surveillance programs have pointed out the constant increase in the proportions of ampicillin-resistant H. influenzae strains due to β-lactamase production from the mid-1970s to the end of the 1990s (14, 19, 35, 42), when the frequency of β-lactamase-producing strains reached 30 to 35% in countries such as the United States and Spain (2, 13, 15, 43, 44). In the postvaccination era, ampicillin resistance has dropped to apparently stable levels of 20 to 25% in the United States and 10 to 15% in Spain (18). Only two β-lactamases have been identified in H. influenzae, TEM-1 and ROB-1. The simultaneous presence of both of these enzymes in the same strain is extremely unusual (they are both found in about 0.5% of β-lactamase-positive strains) (36). The global prevalence of H. influenzae harboring ROB-1 is 7 to 11% among β-lactamase-producing isolates (12, 21, 26, 36). This enzyme has a particularly high level of hydrolytic activity against cefaclor, and about 70% of cefaclor-resistant H. influenzae strains are ROB-1 producers (21). High levels of consumption of this drug in the United States and Mexico may explain the high local prevalence of ROB-1 among β-lactamase-producing strains in both countries (21 and 26%, respectively) (D. J. Farrel, I. Morrissey, S. Bakker, E. Winsor, and D. Felmingham, Abstr. 42nd Intersci. Conf. Antimicrob. Agents Chemother., abstr. C2-1888, 2002). ROB-1 also hydrolyzes cefprozil, but it does not affect the oxyimino-cephalosporins (ceftriaxone or cefotaxime) and is effectively inhibited by β-lactamase inhibitors, such as clavulanate (8). The aim of this study was to evaluate the possible risk of emergence of ROB-1 molecular variants able to hydrolyze oxyimino-cephalosporins such as ceftriaxone or cefotaxime or to develop resistance to β-lactamase inhibitors. For this purpose, we used a novel approach, cloning of the bla_ROB-1 gene from H. influenzae in an engineered Escherichia coli hypermutable strain (GB20) lacking the ampC gene encoding chromosomal β-lactamase. Hypermutation increases the rate of emergence of possible β-lactamase mutants, and the lack of the ampC gene ensures that these resistance mutations do not occur in the chromosomal β-lactamase gene but in bla_ROB-1.

(This work was presented in part at the 41st Interscience Conference on Antimicrobial Agents and Chemotherapy, Chicago, Ill., 2001 [J. C. Galán, M. R. Baquero, and F. Baquero, Abstr. 41st Intersci. Conf. Antimicrob. Agents Chemother., abstr. C1-1490, 2001].)

MATERIALS AND METHODS

Bacterial strains and GB20 construction.

E. coli MI1443 (11) was used as the recipient for plasmids and for the determination of antibiotic susceptibilities. Phage P1 was grown on hypermutable E. coli strain AB1157 mutS::Tn10 with a defective mutS gene (32). Recovered phage P1 was then used to transduce E. coli MI1443 with a defective ampC gene, giving rise to E. coli GB20. Detailed descriptions of the strains are provided in Table 1.

TABLE 1.

Bacterial strains and plasmids used in this study

Strain or plasmid	Relevant genotype or phenotype	Source or reference
E. coli strains
MI1443	ΔfrdABCD ΔampC recA Smr	11
AB1157 mutS::Tn10	ara14 argE3 galK2 his4 lacY1 leu6 mtl1 proA2 str31 supE44 thi1 thr1 tsx33 xy15 mutS215::Tn10	32
GB20	ΔfrdABCD ΔampC mutS215::Tn10 recA Smr	This study
Plasmids
pACYC184	Tetr Chlr	9
pMON401	blaROB-1 in the BamHI site of pACYC184 (Ampr Chlr)	20
pMON410	The same as pMON401 but harboring the mutation S130G in ROB-1	This study
pMON411	The same as pMON401 but harboring the mutation R244C in ROB-1	This study
pMON420	The same as pMON401 but harboring the mutations R164W and A237T in ROB-1	This study
pMON421	The same as pMON401 but harboring the mutation R164W in ROB-1	This study
pMON422	The same as pMON401 but harboring the mutation L169S in ROB-1	This study
pBGS18−	KanrlacPOZ′	39
pBB1	blaROB-1 in the polylinker site of pBGS18−	This study
pBB10	The same as pBB1 but harboring the mutation S130G in ROB-1	This study
pBB11	The same as pBB1 but harboring the mutation R244C in ROB-1	This study
pBB20	The same as pBB1 but harboring the mutations R164W and A237T in ROB-1	This study
pBB21	The same as pBB1 but harboring the mutation R164W in ROB-1	This study
pBB22	The same as pBB1 but harboring the mutation L169S in ROB-1	This study
pBB120	The same as pBB1 but harboring the mutations S130G, R164W, and A237T in ROB-1	This study

Plasmids.

Plasmid pMON401 (Amp^r Chl^r) harboring the bla_ROB-1 gene (20) is a pACYC184 (Tet^r Chl^r) derivative (9). Plasmids pMON410, pMON411, pMON420, pMON421, and pMON422 are pMON401 derivatives obtained in different serial passage experiments in this work. Plasmid pBGS18⁻ (Kan^r) (39) was used in subcloning experiments, giving rise to recombinant plasmids pBB1 (wild-type bla_ROB-1 gene); pBB10 and pBB11, which carry mutations in the bla_ROB-1 gene that confer resistance to β-lactam plus β-lactamase inhibitor combinations; as well as pBB20, pBB21, and pBB22, which have mutations that lead to increased cefotaxime-hydrolyzing activity. Plasmid pBB120 harbors the bla_ROB-1 gene with the changes that confer both clavulanate and cefotaxime resistance. Detailed descriptions of the plasmids are included in Table 1.

Determination of mutation frequencies.

To ascertain the hypermutable nature of the E. coli GB20 construction, 50 μl of a 10⁻⁴ dilution of overnight cultures of each of the strains MI1443, GB20 (with and without pMON401), and AB1157 mutS::Tn10 was inoculated into five independent flasks containing Luria-Bertani (LB) broth; and the flasks were incubated overnight at 37°C under high agitation. A total of 100 μl of each overnight culture was plated onto LB agar plates containing 100 μg of rifampin per ml or 40 μg of nalidixic acid per ml. Simultaneously, 100 μl of a 10⁻⁶ dilution of each of the same cultures was plated on LB agar plates without antibiotics. The number of rifampin-resistant or nalidixic acid-resistant mutants was counted at 48 h, and the number of viable cells was counted at 24 h. The mutation frequency was defined as the ratio of rifampin- or nalidixic acid-resistant colonies versus the number of viable cells in antibiotic-free medium.

Serial passage experiments.

Aliquots (5 μl) of an overnight culture of E. coli GB20(pMON401) were inoculated into five tubes containing 5 ml of LB broth with tetracycline (20 μg/ml) and chloramphenicol (30 μg/ml) to ensure the maintenance of Tn10 and pMON401, respectively. The cultures were grown at 37°C under normal agitation and were submitted to daily serial passages with increasing concentrations of cefotaxime alone or amoxicillin plus a fixed concentration of a β-lactamase inhibitor (2 μg of clavulanate per ml). The β-lactam concentrations ranged from 2 dilutions below to 6 dilutions above the MIC for GB20(pMON401). Plasmid DNA was extracted from each of the five bacterial populations that grew at 4 to 5 and 6 dilutions above the MIC of cefotaxime or 4 to 5 dilutions above the MIC of amoxicillin-clavulanate. DNA was introduced into MI1443 by transformation; transformants were selected on LB agar plates containing chloramphenicol (30 μg/ml) and cefotaxime or amoxicillin-clavulanate at 2, 3, or 4 dilutions above the MIC. Ten transformants from each population were chosen; and different phenotypes were identified by disk diffusion tests with cephalothin, cefazolin, cefuroxime, cefotaxime, ceftazidime, ampicillin, and amoxicillin-clavulanate. The plasmid DNA of transformants corresponding to each phenotype was reintroduced by transformation into MI1443, with only chloramphenicol used as a selector. The DNA of derivatives of strain MI1443(pMON401) with possible bla_ROB-1 gene variants was sequenced. Finally, each bla_ROB-1 gene variant was subcloned into plasmid pBGS18⁻ to ascertain the specificities of the phenotypic changes independently from the host plasmid.

Construction of plasmids pBB1, pBB10, pBB20, and pBB120.

To avoid the possible interference of putative mutations in the bacterial resistance phenotype that could arise in other sites of plasmid pMON401 during serial passage experiments, the bla_ROB-1 gene was amplified from plasmids pMON401, pMON410, pMON411, pMON420, pMON421, and pMON422 by PCR and subcloned into pBGS18⁻, giving rise to recombinant plasmids pBB1, pBB10, pBB11, pBB20, pBB21, and pBB22, respectively. Plasmid pBB120 was constructed as described below.

Susceptibility testing.

The MICs of amoxicillin, ampicillin, piperacillin, amoxicillin-clavulanate (2:1 ratio), ampicillin-sulbactam (2:1 ratio), piperacillin-tazobactam (fixed concentration of tazobactam, 4 μg/ml), cephalothin, cefuroxime, cefaclor, cefotaxime, ceftazidime, and imipenem were determined as recommended by the NCCLS (27).

Amplification and sequencing procedures.

The primers used to amplify the bla_ROB-1 gene were ROBFE (5′-GGGGAATTCGATCAGAGTAATAATTTCTGAT-3′), which recognizes a region immediately upstream of the −35 region of the putative promoter, and ROBRH (5′-CGGCAAGCTTGAAAGCAAGTTTCAACGGCTT-3′), which hybridizes to a region 42 bp downstream of the stop codon containing an EcoRI and a HindIII restriction site (the respective restriction sites are underlined). The internal primers used to sequence the gene in both directions were ROBd1 (5′-CGTTTATGTATGGGATACAGAA-3′), which hybridizes to a region 185 bp from the beginning of the gene, and ROBr (5′-GGGTTACGTTATCGCCTAATT-3′), which hybridizes to a region 390 bp from the end of the gene. Amplification was performed in a mixture containing (per 50 μl) 2 mM MgCl₂, 1× PCR buffer II (Applied Biosystems, Weiterstadt, Germany), 200 μM (each) deoxynucleoside triphosphates, 25 pmol of each primer, and 2.5 U of Taq-Gold DNA polymerase (Applied Biosystems). The following PCR program was used: 94°C for 12 min and 35 amplification cycles of denaturation at 94°C for 60 s, annealing at 56°C for 60 s, and elongation at 72°C for 60 s. The amplification product was analyzed by electrophoresis on a 0.8% agarose gel predyed with ethidium bromide. The primers were removed from the PCR product with the QIAquick PCR purification kit (Qiagen, GmbH, Hilden, Germany) before the product was sequenced or cloned into a new plasmid.

Construction of β-lactamases with triple mutations.

The enzyme with the triple mutation Ser130Gly, Arg164Trp, and Ala237Thr was constructed by using the new primers ROBArg164Trp (5′-CTAGCCA^∗ATTGGTATGGGTT-3′), which harbors the mutation G⁵³²→A⁵³² (asterisk) and which results in the Arg164Trp substitution, and ROBd2 (5′-TTGGCAAAGGCATGACGATTG-3′), which hybridizes to nucleotide positions 380 to 400 in the bla_ROB-1 gene, together with primers ROBFE and ROBRH described above. A first PCR was performed with primers ROBFE and ROBArg164Trp by using DNA from the amoxicillin-clavulanate-resistant mutant as the template. A second PCR was performed with primers ROBd2 and ROBRH by using DNA from the cefotaxime-resistant mutant. A third PCR was carried out with primers ROBFE and ROBRH by using the PCR products obtained in the previous amplifications as the templates. The final amplification product was digested with EcoRI and HindIII, ligated to pBGS18−, digested with the same restriction enzymes, and transformed into MI1443. Transformants were selected on LB agar plates containing kanamycin (50 μg/ml), and the bla_ROB gene of recombinant plasmid pBB120 was sequenced to confirm the presence of the three desired changes.

Enzyme kinetics.

The β-lactamases extracted from E. coli MI1443(pBB1), MI1443(pBB10), MI1443(pBB11), MI1443(pBB20), MI1443(pBB21), and MI1443(pBB22) were prepared from 1 liter of an overnight culture of LB broth with kanamycin (50 μg/ml) at 37°C. The cells were harvested by centrifugation at 4,500 × g for 20 min, washed twice with 5 ml of 0.05 M phosphate buffer (pH 7.4), and resuspended in 2 ml of the same buffer. The suspension was sonicated for 10 min with 2-s pulses (Sonicator Heat system; Ultrasonics Inc.). Debris was removed by centrifugation (17,500 × g for 30 min at 4°C), and the β-lactamase activity of the supernatant was assayed by nitrocefin hydrolysis.

The β-lactamase activities of the extracts against different β-lactam antibiotics were measured spectrophotometrically (Uvikon-940 spectrophotometer). K_m and V_max values were obtained by double-reciprocal (Lineweaver-Burk) plots of the initial steady-state rates at different substrate concentrations. The inhibitor concentrations required to inhibit 50% of the β-lactamase activity (IC₅₀s) were determined by incubating the β-lactamase extracts with different concentrations of the inhibitors for 10 min at room temperature, and then 100 μM nitrocefin was added as the substrate (5).

RESULTS

Mutation frequency of constructed strain E. coli GB20.

The mutation frequency of the constructed variant, E. coli GB20 (ΔampC, mutS defective), was compared with those of its isogenic strain, E. coli MI1443 (ΔampC), and strain E. coli AB1157 mutS::Tn10. The frequencies of mutation to rifampin (100 μg/ml) and nalidixic acid (40 μg/ml) resistance for E. coli GB20 were 2.7 ± 0.9 × 10⁻⁷ and 5.5 ± 3.2 × 10⁻⁷, respectively. Identical values were also obtained when the strain harbored plasmid pMON401 (data not shown). These mutation frequency values are similar to those for E. coli AB1157 mutS::Tn10 strain from which the mutS allele was derived, which were 4.5 ± 1.4 × 10⁻⁷ and 2.6 ± 0.5 × 10⁻⁷, respectively. For wild-type strain E. coli MI1443, the mutation frequencies were 2.6 ± 3.3 × 10⁻⁹ and 5.7 ± 2.9 × 10⁻⁹, respectively. These results showed that the mutation frequency for E. coli GB20 was 100-fold higher than that for its isogenic strain, E. coli MI1443. These results, which reflect the means of five replicate experiments in each case, coincided with those expected for hypermutable strains with a defective mutS gene (25).

Serial passage experiments.

The original cefotaxime MIC for E. coli GB20(pMON401) was 0.06 μg/ml. After a week of daily passages with increasing cefotaxime concentrations, the strain was able to grow in the presence of cefotaxime at a concentration of up to 4 μg/ml. Transformants obtained from populations growing in five different tubes containing 2 or 4 μg of cefotaxime per ml exhibited a single phenotypic pattern of antibiotic resistance. Conversely, transformants from populations recovered from tubes containing only 1 μg of cefotaxime per ml yielded two different phenotypic patterns. This observation suggests that different cefotaxime-resistant mutational variants were selected in the presence of low antibiotic concentrations but were replaced by a more effective mutant in the presence of higher concentrations.

The original amoxicillin-clavulanate MIC for E. coli GB20 was 4/2 μg/ml, and after serial passages, growth was obtained in the presence of 64/2 μg/ml. Transformants obtained from all tubes containing this highly resistant population had a single phenotypic pattern of resistance. As in the case of cefotaxime, when transformants were obtained from cultures growing in tubes with only 16/2 μg of amoxicillin-clavulanate per ml, three different phenotypic patterns of antibiotic resistance were found. Again, the interpretation is that mutational variants were selected at low amoxicillin-clavulanate concentrations; however, only the more effective mutant (that with the “winner” genotype) was able to survive in the presence of the higher concentration.

Genetic characterization of mutants.

Both strands of the bla_ROB-1 gene from E. coli strains that achieved the highest levels of amoxicillin-clavulanate or cefotaxime resistance after antibiotic challenge were fully sequenced. The mutations found are shown in Table 2.

TABLE 2.

Mutants of ROB-1 β-lactamase conferring resistance to cefotaxime and/or amoxicillin-clavulanate

Variant	Amino acid substitution at position:					Phenotype
	130	244	164	169	237
ROB-1						Wild type
Mutant-IR1	S130G					IR-ROB
Mutant-IR2		R244C				IR-ROB
Mutant-ES1			R164W		A237T	ES-ROB
Mutant-ES2			R164W			ES-ROB
Mutant-ES3				L169S		ES-ROB
Mutant-IR1-ES1	S130G		R164W		A237T	IR-ES-ROB

Three types of mutant variants were obtained in the cefotaxime challenge experiments. Mutant ES1 was the only transformant found in all five tubes with the highest cefotaxime concentration. This variant showed two nucleotide substitutions, C436→T436 and G656→A656, resulting in the amino acid substitutions Arg164Trp (near the Ω loop) and Ala237Thr (next to the KSG motif, according to the numbering scheme of Ambler et al. [1]). Other extended-spectrum ROB (ES-ROB) variants were found only in the presence of lower cefotaxime concentrations. One of the mutants had the single mutation C436→T436, which gave rise to the Arg164Trp substitution; and the other mutant had the single change T452→C452, which resulted in the amino acid substitution Leu169Ser.

Two types of resistant mutants were detected in the amoxicillin-clavulanate challenge experiments. In the case of transformants from five of five tubes growing in the presence of high amoxicillin-clavulanate concentrations, a single type of inhibitor-resistant ROB (IR-ROB) mutant was detected. The mutant had the single mutation A335→G335, which results in the amino acid change Ser130Gly, located in the SDN motif of the β-lactamase molecule. A second type of IR-ROB mutant was detectable only among transformants growing in some tubes with low amoxicillin-clavulanate concentrations. That mutant had a single mutation, C674→T674, that conferred the amino acid change Arg244Cys.

Antibiotic susceptibilities of E. coli MI1443 containing pBB1 pBB10, pBB11, pBB20, pBB21, pBB22, and pBB120 recombinant plasmids.

The MICs of the different β-lactam antibiotics for E. coli MI1443 harboring plasmid pBGS18⁻ and the recombinant plasmids (pBGS18⁻ derivatives with the bla_ROB variants) are shown in Table 3.

TABLE 3.

Susceptibilities to β-lactams of E. coli MI1443 and derivatives producing wild-type ROB-1 and mutant ROB-1 β-lactamases

Strain	Variant	Phenotype	MIC (μg/ml)a
			AMX	AMC	AMP	SAM	PIP	TZP	CEF	CEC	CXM	CTX	CAZ	IPM
MI1443			≤1	≤0.5	≤1	1	1	≤1	1	≤0.5	2	0.03	0.25	0.12
MI1443 (pBGS18−)			≤1	≤0.5	≤1	1	1	≤1	1	≤0.5	2	0.03	0.25	0.12
MI1443 (pBB1)	Wild type	ROB	≥2,048	4	≥2,048	8	1,024	2	4	512	16	0.06	8	0.25
MI1443 (pBB10)	Mutant IR1	IR-ROB	2,048	32	512	32	512	512	1	≤0.5	2	0.03	1	0.12
MI1443 (pBB11)	Mutant IR2	IR-ROB	2,048	16	512	16	512	4	1	≤0.5	2	0.03	2	0.12
MI1443 (pBB20)	Mutant ES1	ES-ROB	128	2	32	4	8	≤1	8	16	64	4	128	0.12
MI1443 (pBB21)	Mutant ES2	ES-ROB	512	2	256	2	128	≤1	4	2	32	1	128	0.25
MI1443 (pBB22)	Mutant ES3	ES-ROB	1,024	2	1,024	4	512	≤1	16	64	64	0.25	64	0.25
MI1443 (pBB120)	Mutant-IR1-ES1	IR-ES-ROB	16	8	8	2	16	4	1	4	4	0.03	32	0.25

^aAbbreviations: AMX, amoxicillin; AMC, amoxicillin-clavulanic acid (2:1); AMP, ampicillin; SAM, ampicillin-sulbactam; PIP, piperacillin; TZP, piperacillin-tazobactam; CEF, cephalothin; CEC, cefaclor; CXM, cefuroxime; CTX, cefotaxime; CAZ, ceftazidime; IPM, imipenem.

The E. coli strain with pBB20, which harbors the ES-ROB mutant enzyme with the substitutions Arg164Trp and Ala237Thr (the winner genotype in cefotaxime challenge experiments), had increased levels of resistance to extended-spectrum cephalosporins, ranging from 16-fold increased levels of resistance to ceftazidime to about 60-fold increased levels of resistance to cefotaxime, the selector antibiotic. However, this mutant was more susceptible to penicillins and, surprisingly, to cefaclor (16-fold) than the E. coli strain harboring the wild-type enzyme. The E. coli strain with pBB21, which carries Arg164Trp as the sole change, was 16 times more resistant to ceftazidime and cefotaxime than the E. coli strain harboring the wild-type enzyme; and the level of resistance of the E. coli strain with pBB21 to cefaclor was reduced 256-fold. For the E. coli strain with pBB22 (Leu169Ser), the ceftazidime MIC increased eightfold and the cefotaxime MIC increased fourfold, while the cefaclor MIC decreased eightfold.

The E. coli strain with pBB10, which harbors the IR-ROB enzyme with the Ser130Gly replacement (the winner genotype in amoxicillin-clavulanate challenge experiments), was eightfold more resistant to β-lactam-β-lactamase inhibitor combinations than the E. coli strain with pBB1 and the wild-type enzyme. Interestingly, the susceptibilities to all cephalosporins tested increased, particularly that to cefaclor (≥1,024-fold). In the case of the E. coli strain with pBB11, which harbors the second IR-ROB variant with the Arg244Cys amino acid change, the level of resistance to the inhibitors increased 4-fold, but again, the level of resistance to cefaclor decreased dramatically (≥1,024-fold).

The E. coli strain with the hybrid mutant enzyme (IR-ES-ROB) showed a phenotype of reduced levels of resistance to both amoxicillin-clavulanate (8 μg/ml, versus 4 μg/ml for the wild type) and ceftazidime (32 μg/ml, versus 8 μg/ml for the wild type). Other investigators have observed this trend with other β-lactamases (34, 40), once more revealing the difficulty in obtaining both resistance phenotypes in the same enzyme molecule.

Kinetic parameters.

The kinetics of substrate hydrolysis by the ROB-1 β-lactamase in comparison with those by the ROB-1 mutants with a single mutation (IR-ROB or ES-ROB) or double mutations (mutant ES1) are shown in Table 4. With respect to cefotaxime, the catalytic efficiencies (V_max/K_m values) of the β-lactamases of the ES-ROB mutants with the single Leu169Ser or Arg164Trp change or the double Ala237Thr and Arg164Trp changes were 3.5, 5.5, and 1,000 times higher, respectively, than that of the ROB-1 β-lactamase. On the contrary, the catalytic efficiency toward cefaclor in these mutants was decreased 3, 17, and 6 times, respectively. The catalytic efficiency of the β-lactamase of the more efficient mutant (with double Ala237Thr and Arg164Trp mutations) toward ampicillin was reduced 27.5-fold.

TABLE 4.

Kinetics parameters for ROB-1 β-lactamase and mutant derivatives^a

Variant	Cefaclor			Ampicillin			Cefotaxime
	Vmax (S−1)	Km (μM)	Vmax/Km (M−1 S−1)	Vmax (S−1)	Km (μM)	Vmax/Km (M−1 S−1)	Vmax (S−1)	Km (μM)	Vmax/Km (M−1 S−1)
ROB-1	635	227	2.8 × 106	2,564	186	1.37 × 107	0.64	142	4.5 × 103
Mutant IR1	21.4	177	1.2 × 105	471	220	2.1 × 106	—b	—	—
Mutant IR2	17.8	578	3 × 104	214	102	2 × 106	—	—	—
Mutant ES3	15.5	16	9 × 105	—	—	—	2	120	1.6 × 104
Mutant ES2	18.8	120	1.6 × 105	—	—	—	4.2	167	2.5 × 104
Mutant ES1	182	373	4.8 × 105	64	127	5 × 105	728	88	8.2 × 106

^aAll values are the means of three independent experiments.

^b—, not determined.

The catalytic efficiencies (V_max/K_m values) of the IR-ROB β-lactamases with the Ser130Gly and the Arg244Cys changes toward cefaclor were reduced 23 and 95 times, respectively, in comparison with that of the wild-type enzyme. The catalytic efficiency toward ampicillin was also reduced 6.5-fold.

Activities of β-lactam inhibitors against ROB-1 β-lactamase variants.

The IC₅₀s of clavulanate, sulbactam, and tazobactam for the wild-type ROB-1 β-lactamase and the different ROB-1 variants are shown in Table 5. The IC₅₀s of tazobactam for the variants compared with that for the wild-type enzyme increased the least: 70-fold for the ROB-1 variant carrying the Ser130Gly change and only 1.5-fold for the mutant with the Arg244Cys substitution. The IC₅₀s of clavulanate, sulbactam, and tazobactam increased the least for the ES-ROB mutant with the double Arg164Trp and Ala237Thr mutations, with the IC₅₀s being 30, 8.5, and 60 times lower, respectively, than the IC₅₀ for ROB-1.

TABLE 5.

β-Lactamase inhibitor IC₅₀s for ROB-1 β-lactamase and ROB-1 mutant enzymes

Variant	IC50 (μM)
	Clavulanate	Sulbactam	Tazobactam
ROB 1a	0.009	0.25	0.09
Mutant IR1	1.5	80	6.9
Mutant IR2	0.2	3.6	0.15
Mutant ES1	0.0003	0.03	0.0015

^aThe reference values for ROB-1 β-lactamase are <0.01 and <1 μM for clavulanate and sulbactam, respectively (8).

DISCUSSION

Several methods for prediction of the evolution of an enzyme that increases the catalytic activity toward its substrate have been described (22, 41, 47, 49). It is expected that in vitro evolution models could yield the same mutants as in vivo processes of evolution (3, 17). In this sense, the use of hypermutable strains is a good means of prediction of natural evolution (30; A. J. O'Neill, I. Chopra, J. L. Martínez, and F. Baquero, Letter, Antimicrob. Agents Chemother. 45:1599-1600, 2001). Nevertheless, the use of E. coli mutant strains for prediction of the evolution of plasmid-mediated β-lactamases has been hindered by the frequent emergence of chromosomal mutants, particularly those involving the AmpC β-lactamase (22). Host strain E. coli GB20 (ΔampC) was used in this work to overcome this problem, and use of this strain can be recommended for prediction by simple serial challenge procedures of a number of mutations in β-lactamases leading to the extension of the hydrolytic spectrum, including β-lactams and β-lactamase inhibitors.

In this study, different resistance genotypes were obtained in the presence of low antibiotic concentrations, but a single genotype was recovered in each experiment in the presence of the highest antibiotic concentration. This result supports the hypothesis that the variability of resistant mutants may be higher in the presence of lower antibiotic concentrations (23). This, in fact, is what is suggested by experiments for the concentration-dependent selection of mutants (28). Beyond a critical concentration, the allele that conferred the greatest fitness advantage eliminated all others (clonal displacement) (3). Our results indicate the need for exploration for the presence of mutants in the presence of low antibiotic concentrations, as some of them will be impossible to detect in the presence of higher ones.

The substitutions most frequently observed among TEM-derived extended-spectrum β-lactamases occur at position Arg164 and imply the presence of smaller amino acids in both TEM laboratory-derived mutants (Ser, Gly, Asn, Cys) and clinical isolates (Ser, Hys, Cys) (31, 48; http://www.lahey.org). In the case of the laboratory-derived ROB-1 variants obtained in this study, Arg164 was replaced by the larger molecule Trp. Vakulenko et al. (48) have described that the mutant enzyme with the Arg164Trp change is responsible for the same cefotaxime or cefepime MICs as those for the wild-type enzyme but that the change drastically increases the susceptibility to ampicillin. The same substitution in ROB-1 increased the MIC for extended-spectrum β-lactams when a single mutation or double mutations were present (16- and 60-fold, respectively). This fact suggests that both the TEM-1 and the ROB-1 enzymes have different structures around the active site, which may reflect the distant phylogenetic relationship between them (24).

The second substitution observed in the ES-ROB winner genotype, Ala237Thr, is frequently found in TEM variants able to hydrolyze cefotaxime or ceftazidime (TEM-5 or TEM-24). Nevertheless, this change does not increase the level of resistance to β-lactams in the absence of other mutations and has been considered a resistance modulator (6). A third ES-ROB mutant carried the single substitution Leu169Ser. Replacements at this position have been already studied in TEM-1 by random mutagenesis or the DNA shuffling technique, with increased activity against ceftazidime being detected (31, 46).

A single nucleotide change in the bla_ROB-1 gene, A335→G, which leads to the amino acid substitution Ser130Gly in the conservative domain SDN (7), confers a fourfold increase in the level of resistance to β-lactam-β-lactamase inhibitor combinations. This substitution has previously been described in several β-lactamases such as IRT-17 (4), SHV-10 (33), OXY-2 (37), and ACI-1 variant (J. C. Galán, M. R. Baquero, M. Reig, F. Baquero, and J. Blázquez, Abstr. 40th Intersci. Conf. Antimicrob. Agents Chemother., abstr. 1919, 2000). The Arg244Cys variant with the IR-ROB phenotype has been described in TEM derivatives such as IRT-1, IRT-16, and IRT-18 (http://www.lahey.org). These mutants, like the one obtained in this work, maintained their susceptibilities to tazobactam.

The hybrid IR-ES-ROB β-lactamase was constructed, and each mutation (Ser130Gly and Arg164Trp plus Ala237Thr) conferred the most efficient phenotype. This hybrid mutant showed weak activities against amoxicillin-clavulanate and ceftazidime and showed even weaker activities against ampicillin and cefotaxime. The possibility that this mutant could be selected in the clinical environment is low. It is known that, despite some exceptions (16, 29, 38), there is frequent incompatibility between the IR and ES phenotypes (34, 40).

Serial passage experiments with H. influenzae in the presence of amoxicillin-clavulanate apparently yielded a few variants for which the MIC was increased, but the increase was not more than fourfold and the variants were not analyzed further (10). It is noteworthy that inhibitor-resistant β-lactamases or extended-spectrum β-lactamases have never been reported in H. influenzae. It is also noteworthy that when extended-spectrum β-lactamases were introduced into H. influenzae, the strains remained susceptible, according to NCCLS criteria, while the MICs for the strains increased significantly (45).

The emergence in H. influenzae of resistant ES-ROB or IR-ROB enzymes could cause important clinical problems. When E. coli was used as the host for these enzymes, the cefotaxime and amoxicillin-clavulanate MICs increased 128 and 64 times, respectively. Considering that similar increases may occur in a typical H. influenzae strain for which basal cefotaxime and amoxicillin-clavulanate MICs are 0.03 and 0.5 μg/ml, respectively, the MIC of cefotaxime could reach 4 μg/ml and that of amoxicillin-clavulanate could reach 32 μg/ml. Interestingly, in the IR-ROB and ES-ROB mutants obtained in this work, the susceptibility to cefaclor was increased significantly (1,024 times for IR-ROB and from 4 to 256 times for ES-ROB) with respect to that for the E. coli strain harboring the wild-type ROB-1 enzyme. The cefaclor MIC for a typical H. influenzae isolate with ROB-1 is 32 μg/ml, but those for strains harboring the IR-ROB mutants are expected to be less than 0.125 μg/ml and those for ES-ROB mutants are expected to range from 0.125 to 8 μg/ml. Although treatment with cefaclor could select for H. influenzae strains harboring the ROB-1 β-lactamase (21), the same antibiotic could have prevented the evolution of this enzyme to IR-ROB or ES-ROB enzymatic variants.