Beta-lactamase analysis and antibiotic susceptibility testing.

Analytical isoelectric focusing of crude cell-free sonicates was performed by the method of Matthew and Harris (36) on polyacrylamide gels (Ampholine PAGplate [pH 3.5 to 9.5]; Amersham Biosciences) in a Multiphor II system (Amersham Biosciences). The beta-lactamase bands were visualized by staining them with nitrocefin (Oxoid). The MICs of ampicillin, cefuroxime, cefotaxime, ceftazidime, and aztreonam were determined by Etest (AB Biodisk, Stockholm, Sweden) according to the manufacturer's instructions. MICs were also determined in the presence of clavulanic acid (2 μg/ml; SmithKline Beecham Pharmaceuticals). Transfer of ampicillin resistance was tested by conjugation experiments as described previously (20).

Nucleotide sequence determination.

The sequence of a 789-bp portion of the 861-bp chromosomal beta-lactamase gene of K. pneumoniae was determined after PCR amplification performed using the primers SHVseq1 (5′-ATG CGT TAT ATT CGC CTG TG) (positions 1 to 20 of the SHV-1 coding sequence; GenBank M59181) and SHVseq3a (5′-TTA GCG TTG CCA GTG CTC) (positions 861 to 844; GenBank M59181). The primers used for sequence determination included the two PCR primers and the internal primers SHVseq2 (5′-CCG CAG GAT TGA CTG CC) (positions 422 to 438) and SHVseq4 (5′-AGG CGG GTG ACG TTG TC) (positions 473 to 457). The sequence of a 383-bp portion of gyrA, starting at position 163 (codon 55) of the gene, was determined as previously described (11). The sequence of a 342-bp portion (codons 31 to 144) of the 939-bp malate dehydrogenase (mdh) gene was determined using PCR and the sequencing primers Pmdh-4 (5′-CCT TTC AGT TCC GCA ACA AAG) and Pmdh-5 (5′-AAC TGC CTT CAG GTT CAG AAC). These primers corresponded to fully conserved regions on an alignment of mdh genes from E. coli (GenBank accession no. M24777) (53) and K. oxytoca strain SB77 (Brisse, unpublished). Sequence determination of PCR products was carried out with the Big Dye Terminator Cycle Sequencing Ready Reaction kit (Perkin-Elmer) on an ABI 3700 or 3100 automated capillary DNA sequencer.

Specific PCR amplification of beta-lactamase gene families.

PCR primer pairs were designed to specifically amplify beta-lactamase genes of the SHV and LEN families and of a new third family named OKP (see Results): SHVfor (5′-GAG CGA AAG ATC CAC TAT CG) (positions 262 to 281) and SHVrev (5′-GGT ATC CCG CAG ATA AAT CA) (positions 786 to 767), LEN.for (5′-AAC TAC TGA CCG CGC AGC AT) (positions 563 to 582) and LEN.rev (5′-ATT ACG CTC GGC CAT ACT CG) (positions 810 to 791), and OKP-1F (5′-GCG GGC CGS GTG GGC TAT GTT) (positions 109 to 129) and OKP-2R (5′-MGC GGG GCC GCC RAC GCT CTT) (positions 426 to 406). PCR mixture conditions were as described previously (11). PCR amplification was achieved by submitting samples to an initial denaturation step (1 min at 94°C), followed by 35 amplification cycles (94°C for 1 min; 30 s at the annealing temperature of 60°C for SHV primers, 65°C for OKP primers, or 52°C for LEN primers; 72°C for 90 s) and a final elongation step of 5 min at 72°C. The seven phylogenetic-group reference strains were used as controls.

Sequence data analysis.

To determine the model of nucleotide substitution that best fit the data for each gene, the program Modeltest version 3.06 (49) was used to compare the maximum-likelihood scores of 56 different models computed using PAUP* version 4.0b10 (51), as described in the instructions for the Modeltest program. Phylogenetic trees were obtained using PAUP* version 4.0b10 and the neighbor-joining method, with distance options corresponding to the best model found for each gene. These models were HKY+G (21) for the beta-lactamase gene, the Jukes-Cantor (28) model for gyrA, and the Kimura two-parameter K80 model (31) for the mdh gene. Bootstrap analysis was performed with 1,000 replicates. Trees generated by PAUP* were saved and drawn using TreeView version 1.6.6 (44). The McDonald-Kreitman test of neutral evolution (37) was performed using DNASP version 3.53 (49a). Levels of synonymous substitution per synonymous site (K_s) were estimated using DNASP version 3.53, which uses the Nei and Gojobori method (39) after Jukes-Cantor correction for multiple substitutions. Computation of the K_s value for the divergence between E. coli and Salmonella enterica was performed by comparing pairwise values for one representative sequence of each E. coli phylogenetic group and all subspecies of S. enterica (9).

Molecular phylogeny of the chromosomal beta-lactamase gene.

The nucleotide sequence of a 789-bp portion of the chromosomal beta-lactamase gene was established in the 28 strains. The sequence corresponds to 92% of the gene length and starts at position 37 of bla_SHV-1 (codon 13) and ends 33 nucleotides before the stop codon. All sequences were aligned without insertion-deletion events, and there were no missing or ambiguous data. The sample contained 24 distinct bla alleles (Table ​(Table1;1; see GenBank accession numbers for nucleotide substitutions). Alignment of the 24 nucleotide sequences showed 156 polymorphic sites, of which 14 were present in only one sequence and 142 were present in at least two sequences and thus phylogenetically informative.

The phylogenetic analysis revealed three groups of nucleotide sequences (Fig. ​(Fig.1).1). The first group included the three reference strains of the phylogenetic group KpI (ATCC 13883^T, ATCC 13884^T, and ATCC 11296^T/NCTC 5050^T); the SHV-1 reference strain, 1976E; and all six fecal isolates expressing SHV-1-like beta-lactamases. The second group included the two reference strains of group KpIII, the LEN-1 reference strain, the two environmental isolates, and the nine fecal isolates expressing LEN-1-like beta-lactamases. The third group comprised the two KpII reference strains and strains SB30 and OR95:2. The nucleotide sequence divergence was much lower within all three Kp groups than between any pair of them (Table ​(Table2),2), as illustrated by the long branches leading to the ancestral node of each group compared to the short branches inside the groups (Fig. ​(Fig.1).1). Of the three groups of bla sequences, the third was the most heterogeneous and showed two distinct subgroups, one comprising the two KpII reference strains SB18 and SB59 and one comprising strains SB30 and OR95:2 (Fig. ​(Fig.11).

Open in a separate window

FIG. 1.

Phylogeny of the chromosomal beta-lactamase gene of K. pneumoniae. The tree was obtained by the neighbor-joining method using the HKY85+G model of nucleotide substitution (see Materials and Methods). The root was determined using the sequence of beta-lactamase TEM-1, the closest known relative to the chromosomal beta-lactamase of K. pneumoniae. Because of the long branch leading to the TEM-1 sequence, the root position is represented by a triangle in order to increase the scale of the figure so that the relationships among K. pneumoniae beta-lactamase sequences are clearly visible. The scale indicates the proportion of dissimilar nucleotides among sequences after correction for multiple substitutions. Three major groups of sequences are visible, corresponding to the SHV, LEN, and OKP families of beta-lactamase genes. The values at the nodes correspond to the percent confidence obtained after 1,000 bootstrap replicates. Bootstrap values inside the two groups corresponding to the SHV and LEN families were well below 50% for most nodes and are not given.

There was excellent correspondence between the sequences of the chromosomal beta-lactamase genes and the pIs of the encoded enzymes (Table ​(Table1).1). Indeed, the first group of sequences were all from strains producing pI 7.6 (SHV-1-like) beta-lactamases, and the second group included all sequences from pI 7.1 (LEN-1-like) strains. Also, the divergence seen within the third group of sequences corresponded to the pI values. The four strains in this group had beta-lactamases with different pIs (SB18, pI 7.8; SB59, pI 8.1; SB30, pI 7.0; and OR95:2, pI 6.5). None of the sequences in the third group could be closely matched to sequences in public databases. The closest matches corresponded to bla_SHV (88% similarity to bla_SHV-1) and bla_LEN (89% similarity to bla_LEN-1). Therefore, we chose to assign them to a new beta-lactamase gene family, which we designated bla_OKP (for other K. pneumoniae beta-lactamases).

Correspondence between the phylogenies of the chromosomal beta-lactamase gene and gyrA.

The sequence of a 383-bp portion of the gyrA gene was established for 21 strains under study and compared to the gyrA sequences obtained previously for the seven reference strains of the K. pneumoniae phylogenetic groups KpI, KpII, and KpIII (11). All sequences were aligned without insertion or deletion, and no site had ambiguous or missing data. Eight distinct alleles were found among the 28 sequences (Table ​(Table1;1; see GenBank accession numbers for nucleotide substitutions). Eleven sites were polymorphic, two of which were singletons and nine of which were phylogenetically informative. Ten changes occurred in the third codon position, and one occurred in the first position. No change was observed at the amino acid level. The biggest pairwise difference was seen between alleles 1 and 7, with 3% (10 of 383) nucleotide changes.

The phylogenetic relationships among the gyrA sequences, depicted in Fig. ​Fig.2,2, corresponded fully to the phylogeny obtained for the chromosomal beta-lactamase gene seen in Fig. ​Fig.1.1. Indeed, all strains with bla_SHV sequences fell into group KpI, all strains with bla_LEN sequences fell into KpIII, and strains SB30 and OR95:2 formed a third branch, together with the KpII reference strains SB18 and SB59. Similar to the result observed based on their chromosomal beta-lactamase gene, strains SB30 and OR95:2 clustered in a distinct subgroup, different from the subgroup formed by the KpII reference strains. The nucleotide sequence divergence of gyrA within and between the groups KpI, -II, and -III is given in Table ​Table22.

Open in a separate window

FIG. 2.

Phylogeny of the gyrA gene of K. pneumoniae. The tree was obtained by the neighbor-joining method using the Jukes-Cantor model of nucleotide substitution (see Materials and Methods). The tree was rooted at the midpoint of the branch leading from group KpII to the other two groups. The scale indicates the proportion of dissimilar nucleotides among sequences after correction for multiple substitutions. Three major groups of sequences are visible, corresponding to the K. pneumoniae phylogenetic groups KpI, KpII, and KpIII defined previously (11).

The correspondence between the chromosomal beta-lactamase gene sequences and the phylogenetic groups based on gyrA sequences was verified when 34 additional clinical strains, previously assigned to different phylogenetic groups, were analyzed by PCR (data not shown). The bla_SHV gene was detected only in the 21 KpI strains tested, and the bla_LEN gene was detected only in the seven KpIII strains tested. The PCR assays performed with primers OKP-1F and OKP-2R were positive only in the six KpII strains tested. The gyrA sequences of these strains showed that they were more closely related to KpII reference strains than they were to strains SB30 and OR95:2.

Diversity of chromosomally encoded beta-lactamase.

Among the 24 distinct nucleotide sequences obtained from the 28 strains, there were 54 codon differences that implied amino acid changes. In total, 17 distinct amino acid sequences were found (Tables ​(Tables11 and ​and3).3). Among the 10 KpI strains (all producing SHV beta-lactamases), one of the amino acid changes observed occurred at a site (residue 252) where variation had not been described previously (http://www.lahey.org/studies/webt.htm). The strain carrying this bla_SHV allele, K. pneumoniae subsp. rhinoscleromatis ATCC13884^T, thus produces an SHV variant with a leucine at position 252 instead of the proline found in previously described SHV enzymes. The other KpI strains showed deduced amino acid sequences identical to that of SHV-1, SHV-11 (the Leu35Gln variant, which was described as non-ESBL) (41), or SHV-33 (the Pro226Ser variant, which does not confer resistance to broad-spectrum cephalosporins) (15).

Among the 14 KpIII strains (all producing LEN beta-lactamases), one plant isolate and one fecal isolate had amino acid sequences identical to that of the LEN-1 enzyme (3). Two other fecal isolates had beta-lactamase amino acid sequences that were indistinguishable from that of the LEN-2 enzyme (15). The remaining nine KpIII strains showed seven distinct amino acid sequences differing from LEN-1 by two to four amino acids (Table ​(Table3).3). These new variants are also different from LEN-2 and a recently described variant LEN enzyme (GenBank AF452105). They thus represent seven new LEN variants.

The deduced amino acid sequences of the OKP beta-lactamases of the KpII reference strains SB18 and SB59 differed by six residues (Table ​(Table3).3). The other two OKP beta lactamases, found in strains SB30 and OR95:2, differed by 3 amino acids, and they differed from the beta-lactamase sequence of strain SB18 by 18 and 19 amino acids, respectively (Table ​(Table3).3). Along the 263-codon portion that was sequenced, 28 to 32 amino acid differences (87.8 to 89% similarity) were observed among the four OKP beta-lactamases and SHV-1, and 27 to 33 amino acid differences were observed between the OKP beta-lactamases and LEN-1 (87.5 to 89.7% similarity).

The deduced amino acid sequences of all beta-lactamase variants identified in this study were conserved at all positions considered to be important for enzymatic activity (25, 32). The Asp-104 residue, suspected to be responsible for the very slight differences in the substrate profiles of SHV-1 and TEM-1 (32), was conserved in all enzyme variants. Changes associated with clinically relevant phenotypic resistance, in particular the change Gly238Ser associated with the majority of ESBL SHV enzymes, were not observed.

To test whether the evolution of the chromosomal beta-lactamase gene has been subjected to natural selection, we performed the McDonald-Kreitman test of neutral evolution (37). This test compares the synonymous and nonsynonymous variation between and within groups. Under neutrality, the ratio of synonymous to nonsynonymous substitutions that are fixed between groups should be the same as the ratio of synonymous to nonsynonymous polymorphisms within groups. The test, performed for the pairwise comparison of the three K. pneumoniae groups KpI, -II, and -III, did not indicate any evidence of selection on the beta-lactamase gene (Table ​(Table44).

We thank L. G. Burman for contributions to early steps of this project and L. Gezelius for technical assistance with isoelectric focusing and MIC determinations. We are grateful to P. A. D. Grimont for providing the two environmental strains.