Six Groups of the OXY β-Lactamase Evolved over Millions of Years in Klebsiella oxytoca

Bacterial strains.A total of 182 K. oxytoca clinical isolates from a previous study (3) were initially analyzed by PCR using OXY-1- and OXY-2-specific primers (13). We next selected 18 K. oxytoca strains for sequencing of several genes (Table 1). Four of these strains were representatives of the previously described KoI and KoII phylogenetic groups (4). Strains SG266, harboring bla_OXY-4, and SG271, harboring bla_OXY-3 (14), were kindly provided by M.-H. Nicolas-Chanoine. The remaining 12 isolates (Table 1) were selected based on OXY-1 and OXY-2 PCR assays (see Results). Identification was initially performed using a VITEK apparatus (bioMérieux) and the indole test and confirmed for the 18 selected strains based on their gyrA sequences (4).

DNA preparation.DNA templates were prepared by suspending a freshly grown colony (approximately 30 mg) in 200 μl of purified water, heating it at 94°C for 10 min, and submitting the extracts to minicentrifugation at 7,500× g for 5 min. Supernatants were stored at− 20°C until use.

PCR amplification.In order to amplify the bla_OXY gene, primers OXY-E and OXY-G (Table 2) were designed based on the alignment of previously published bla_OXY-1 to bla_OXY-4 gene sequences. The amplified portion that permitted us to determine the sequence of the entire β-lactamase coding region stretched from 189 bp upstream of the start codon to 37 bp downstream of the stop codon. The 50-μl PCR mixture contained 20 mM Tris-HCl (pH 8.4), 50 mM KCl, 1.5 mM MgCl₂, 0.2 μM each primer, and 100 μM each deoxynucleoside triphosphate. Samples were submitted to an initial denaturation step (30 s at 94°C), followed by 35 amplification cycles (94°C, 30 s; 50°C, 30 s; 72°C, 30 s) and a final elongation step of 5 min at 72°C. PCR products were visualized under UV light after agarose gel electrophoresis.

Specific bla_OXY-1 and bla_OXY-2 PCR assays were performed using previously defined (13) primers OXY1-A and OXY1-B and primers OXY2-A and OXY2-B, respectively (Table 2), at an annealing temperature of 50°C.

The nearly complete sequence of the 16S rRNA gene (rrs) was obtained using PCR primers Ad and rJ (Table 2). PCR amplification conditions were as described above. Cycling conditions were 4 min at 94°C, followed by 35 cycles (94°C, 1 min; 49°C, 1 min; 72°C, 1 min) and 7 min at 72°C.

Three housekeeping gene portions were amplified by PCR. The sequence of a 940-bp portion of the RNA polymerase beta subunit gene (rpoB) was determined using primers VIC2 and VIC3 (Table 2). The sequence of a 383-bp portion of the gyrase subunit A gene (gyrA) was determined as previously described (4). The sequence of a 573-bp portion of the glyceraldehyde 3-phosphate dehydrogenase gene (gapDH) was determined using primers gapF-173 and gapR-181 (Table 2). PCR amplification conditions were as described for the β-lactamase gene, except that the annealing temperature was 55°C for rpoB and 60°C for gapDH.

PCR primer pairs were designed to specifically amplify β-lactamase genes of the bla_OXY-5 and bla_OXY-6 groups: OXY5-B, OXY5-C, OXY6-B, and OXY6-C (Table 2). PCR conditions were as described above, using an annealing temperature of 55°C.

Sequence determination.PCR products and sequence reaction products were purified by the ultrafiltration method (Millipore). The Ready Reaction Big Dye Terminator Cycle Sequencing v3.1 kit (Perkin-Elmer) and an ABI 3700 automated capillary DNA sequencer (Perkin-Elmer) were used. Primers used for sequencing were the same as those used for amplification. For the 1,102-bp bla_OXY gene product amplified with primers OXY-E and OXY-G, the internal sequencing primer OXY-H was used in addition (Table 2). For 16S rRNA gene sequencing, internal primers D, E, and rE (Table 2) were used in addition to the PCR primers.

Phylogenetic analysis.Sequence alignments were obtained using the MegAlign program of the LaserGene package (DNA Star Inc., Madison, Wisconsin). Phylogenetic trees were obtained with PAUP* version 4.0b10 (30), using the neighbor-joining method based on Kimura's two parameter distance. Bootstrap analysis was performed with 1,000 replicates. Trees generated by PAUP* were saved and drawn using TreeEdit v1.0a10 (built by A. Rambaut and M. Charleston in 2001 [http://evolve.zoo.ox.ac.uk ]). Numbers of synonymous substitutions per synonymous site (K_s) and of nonsynonymous substitutions per nonsynonymous site (K_a) were estimated using DNASP version 3.53 (29). Sequences of rpoB and gyrA used for calibration of the substitution rate were derived from Salmonellaparatyphi A strain ATCC 9150 and Escherichiacoli strain K-12.

Determination of MICs.Antimicrobial susceptibility was determined by the dilution method on Müller-Hinton agar according to the recommendations of the French Society for Microbiology (6). The following drugs were used: amoxicillin, ceftazidime, clavulanate, and ticarcillin from GlaxoSmithKline; piperacillin, cephalothin, and cefoxitin from Sigma-Aldrich; cefotaxime from PanPharma; cefepime from Bristol-Myers Squibb; imipenem from Merck Sharp & Dohme-Chibret; aztreonam from Sanofi-Synthelabo; and tazobactam from Wyeth.

Isoelectric focusing.Crude extracts ofβ -lactamases were obtained by sonication. Isoelectric focusing was performed using a PhastSystem apparatus with PhastGel IEF 3-9 or 5-8 gels (Amersham-Pharmacia Biotech, Freiburg, Germany) and following the manufacturer's recommendations. β-Lactamase activity was revealed by staining the gel with 0.5 mg/ml of the chromogenicβ -lactam nitrocefin (Oxoid, Basingstoke, England). Theβ -lactamases used as references were TEM-1 (pI 5.4), TEM-52 (pI 6), SHV-3 (pI 7.0), OXA-30 (pI 7.3), SHV-1 (pI 7.6), OXY-3 (pI 7.7), OXA-35 (pI 8.0), and SHV-5 (pI 8.2).

Nucleotide sequence accession numbers.Theβ -lactamase, 16S rRNA, rpoB, gyrA, and gapDH gene sequences determined in this study have been submitted to the EMBL and GenBank databases and assigned the accession numbers listed in Table 1.

Correspondence between phylogenetic groups KoI and KoII and β-lactamase genes bla_OXY-1 and bla_OXY-2.In order to establish a possible correspondence between classifications of K. oxytoca strains (4) and OXYβ -lactamases (12), 183 isolates (182 clinical isolates and 1 from a rice plant) were identified by characterization of the gyrA gene and analyzed by PCR using bla_OXY group-specific primers (13). Based on gyrA characterization, 80 isolates could be assigned to KoI, whereas 96 isolates were assigned to KoII and 7 isolates represented a new gyrA phylogenetic group (called group KoVI; see below). Of the 80 KoI isolates, 75 were bla_OXY-1 PCR positive and bla_OXY-2 PCR negative. The five remaining isolates were atypical, being negative for both PCR assays. The 96 KoII isolates were bla_OXY-1 PCR negative and bla_OXY-2 PCR positive. The seven KoVI isolates showed the opposite result, being bla_OXY-1 PCR positive and bla_OXY-2 PCR negative. Therefore, we concluded that the OXY-1 and OXY-2 β-lactamase groups corresponded to the KoI and KoII phylogenetic groups, respectively, with the exception of five atypical KoI isolates. In addition, a new gyrA lineage was found, which apparently had a β-lactamase gene corresponding to or related to bla_OXY-1.

β-Lactamase gene sequencing.The nucleotide sequence of the β-lactamase gene was determined (using PCR primers OXY-E and OXY-G) for the five KoI isolates that were negative for primer OXY1-A and OXY1-B and primer OXY2-A and OXY2-B PCR assays, for the seven KoVI isolates, and for two reference strains each of the KoI and KoII phylogenetic groups (Table 1). Twelve bla_OXY-1 and bla_OXY-2 sequences, as well as the bla_OXY-3 (strain SG271) and bla_OXY-4 (strain SG266) sequences, were retrieved from public databases (Table 1). Considering only the coding region, 23 distinct sequences were found. Besides deletions at codons 11, 12, 25, and 26, there were 207 polymorphic nucleotide sites, of which 66 were variable in only one sequence (singletons) and 141 were variable in at least two sequences (informative sites). The deduced amino acid sequences showed 51 variable amino acid positions. The nucleotide substitutions were approximately three times more frequently synonymous (n = 158) than nonsynonymous (n = 51).

Phylogenetic analysis.The neighbor-joining tree (Fig. 1) obtained for the 30 sequences revealed six branches. The first included the six strains previously described as harboring bla_OXY-1 and the two KoI reference strains SB9 and SB71. The second branch included the six strains previously described as harboring bla_OXY-2 and two KoII reference strains, SB136 (= ATCC 49131) and SB175 (= ATCC 13182^T). These results firmly demonstrate the correspondence of β-lactamase groups bla_OXY-1 and bla_OXY-2 with phylogenetic groups KoI and KoII, respectively. The third and fourth branches each comprised a single isolate, corresponding to bla_OXY-3 and bla_OXY-4. The fifth branch was formed by the five atypical KoI isolates, whereas the sixth corresponded to the seven KoVI isolates. We propose to refer to the β-lactamase genes of these two new groups as bla_OXY-5 and bla_OXY-6, respectively.

• Open in new tab
• Download powerpoint

FIG. 1.

Phylogeny of the K. oxytoca β-lactamase gene. The neighbor-joining tree was rooted using as an outgroup the gene sequence of the SFO-1 β-lactamase (GenBank accession no. AB003148 ), the closest relative of bla_OXY. Values at the nodes correspond to bootstrap values obtained after 1,000 replicates. Six major branches of K. oxytoca β-lactamase sequences are visible, corresponding to the bla_OXY-1 to bla_OXY-6β -lactamase gene groups.

The mean nucleotide sequence similarity was clearly higher within all sixβ -lactamase groups (at least 99.7%) than between any pair of them (see Table 4), as the smallest intergroup difference, observed between the bla_OXY-1 and bla_OXY-5 groups, was 2.5%.

Amino acid sequence diversity of the two new β-lactamase groups, OXY-5 and OXY-6.Among the five isolates harboring bla_OXY-5, two alleles were distinguished, corresponding to two distinct amino acid sequences, OXY-5-1 and OXY-5-2 (Table 3). These two variants differed by two amino acids, at Ambler positions 39 and 153. Both variants differed from the OXY-1 β-lactamase of strain SL781 by five amino acids, located at Ambler positions 12 (deleted in OXY-5 variants), 100, 138, 140, and 165. None of these changes was specific for OXY-5, as they were also observed in OXY-2, OXY-3, and/or OXY-4 (Table 3).

Among the seven isolates harboring bla_OXY-6, four alleles were found, corresponding to four distinct amino acid sequences, OXY-6-1 to OXY-6-4 (Table 3). These variant enzyme sequences differed by their combination of three amino acid differences at Ambler positions 12 (deleted in strain SB324), 35, and 52, and they presented three common differences compared to the SL781 OXY-1 sequence, at positions 5, 87, and 89. None of these changes was specific for the OXY-6 β-lactamase group, as they were also observed in OXY-3 and/or OXY-4 sequences (Table 3).

The amino acid sequence similarity levels confirmed the distinctness of the two newβ -lactamase groups. The highest amino acid sequence similarity of OXY-5 and OXY-6 sequences was observed with OXY-1 (97.5 and 98.3%, respectively) and with OXY-4 (97.2 and 97.7%, respectively), whereas the intragroup value was 99.7% both for OXY-5 and for OXY-6. The most distant amino acid sequences were OXY-2 and OXY-3 (Table 4).

Isoelectric focusing.Analytical isoelectric focusing revealed a single β-lactamase band in all strains. The two distinct OXY-5 amino acid variants differed by their pI values (7.2 for OXY-5-1 and 7.7 for OXY-5-2). In contrast, the four distinct OXY-6 variants exhibited only two distinct pI values, 7.75 (for OXY-6-1, OXY-6-2, and OXY-6-3) and 8.1 (for OXY-6-4). These pI differences are in full agreement with the expected effects of the deduced amino acid changes (Table 3).

Antibiotic susceptibility testing.The MICs of 10 β-lactams and threeβ -lactam-β-lactam inhibitor combinations were determined (Table 5). The susceptibility level observed for strains harboring bla_OXY-5 and bla_OXY-6 were similar to the values observed for the strains of the OXY-1 and OXY-2 β-lactamase groups. The MICs of amoxicillin, ticarcillin, and piperacillin were slightly lower for strains SG271 (OXY-3) and SG266 (OXY-4). All strains showed resistance against amoxicillin and ticarcillin, which was inhibited in the presence of clavulanate. Resistance to piperacillin was intermediate and fell in the presence of tazobactam. Susceptibility to all otherβ -lactams tested was observed. The sequence of the promoter region was established for the five bla_OXY-5 and seven bla_OXY-6 strains. In agreement with the observed MICs, there were no mutations previously reported to increase the transcription rate.

bla _OXY-5- and bla_OXY-6-specific PCR assays.The 18 isolates were subjected to amplification with two primer pairs, primers OXY5-B and OXY5-C and primers OXY6-B and OXY6-C. These pairs were designed on regions containing group-specific nucleotide substitutions. OXY5-B and OXY5-C PCR amplification of the expected 127-bp fragment was positive for all isolates harboring the bla_OXY-5 β-lactamase gene and negative for isolates of all other groups. Conversely, amplification of the 107-bp fragment expected with primers OXY6-B and OXY6-C was positive only with isolates harboring bla_OXY-6 (data not shown).

Correspondence betweenβ -lactamase and housekeeping gene diversity.In order to compare the phylogeny based on the β-lactamase gene with the evolutionary history of other regions of the K. oxytoca chromosome, the nucleotide sequences of portions of the three housekeeping genes rpoB (940 nucleotides [nt] sequenced), gyrA (383 nt), and gapDH (573 nt) were determined in the 18 study isolates. For the three protein coding genes, the alignment revealed no insertion or deletion event, and there was no ambiguous data. The proportion of polymorphic sites was 8.72% (82/940 nt) for rpoB, 8.35% (32/383 nt) for gyrA, and 6.8% (39/573 nt) for gapDH, whereas it was 23% (203/876 nt) for bla_OXY (Table 6). The smaller amount of variable sites in the housekeeping genes was mainly due to the rarity of nonsynonymous substitutions (Table 6), indicative of stronger selective pressure against amino acid changes in the three housekeeping genes, relative to the bla_OXY gene.

The neighbor-joining trees obtained based on rpoB, gyrA, and gapDH were very similar. Therefore, we concatenated the three genes to estimate more robustly the phylogeny of the strains (Fig. 2). A very close agreement with the phylogeny previously obtained on the basis of the β-lactamase gene (Fig. 1) was observed. Strains with bla_OXY-2, bla_OXY-3, bla_OXY-4, and bla_OXY-6 genes clustered into four separate branches, each corresponding to theirβ -lactamase group. These four branches clearly represent distinct phylogenetic groups. For each gene, the nucleotide divergence within groups was much lower than that observed among groups (Table 7). The isolates harboring bla_OXY-1 and bla_OXY-5 β-lactamase genes were very close (Fig. 2 and Table 7). Therefore, we consider bla_OXY-5-harboring strains as belonging to phylogenetic group KoI (see Discussion). Groups KoII and KoIII appeared to be the most external lineages, as observed with the bla_OXY gene.

• Open in new tab
• Download powerpoint

FIG. 2.

Phylogeny obtained based on 1,896 nt positions from the three housekeeping genes rpoB, gyrA, and gapDH. The neighbor-joining tree was rooted using K. pneumoniae type strain SB132 (= ATCC 13883^T). Values at the nodes correspond to bootstrap values obtained after 1,000 replicates. Five branches of K. oxytoca are distinguished, corresponding to phylogenetic groups KoI to KoVI. bla_OXY-5-harboring strains fall into the KoI branch.

The nearly complete (1,454-bp) 16S rRNA gene sequence was determined for the two KoI and the two KoII reference strains and for two bla_OXY-5-harboring and three bla_OXY-6-harboring strains (Table 1). Excluding 12 positions showing within-strain heterogeneity among gene copies (5), the 16S rRNA sequences of bla_OXY-5- and bla_OXY-6-harboring strains were completely identical to that of KoI, with the exception of two substitutions for one bla_OXY-6-harboring strain (SB3051; data not shown).

Ancient presence of the chromosomal β-lactamase gene in K. oxytoca.In the absence of bacterial fossils, the molecular clock hypothesis, according to which the rate of synonymous substitution at synonymous sites (K_s) along evolutionary lineages is roughly constant over time, is a way to estimate the timing of evolutionary events. In order to estimate the time since divergence of the K. oxytoca phylogenetic groups, we used genes rpoB and gyrA, both of which showed clock-like behavior. To calibrate the K_s values of rpoB and gyrA, we used two extreme estimations for the time of divergence between E. coli and Salmonella enterica, 30 and 140 million years (23, 25, 27). The K_s of rpoB between E. coli and S. enterica is 0.223, whereas the K_s for gyrA is 0.367. Table 8 gives the K_s values observed among K. oxytoca groups. Using the 30-million-year calibration, rpoB and gyrA K_s values indicate a split between KoI and KoII 15 and 11 million years ago, respectively, whereas the two groups with the greatest difference, KoIII and KoIV, would have separated 23 or 18 million years ago, respectively. The divergence of the bla_OXY-5 strains from KoI would be as recent as 3 to 2 million years. Using the 140-million-year calibration, the separation of the KoIII and KoIV groups would be estimated to be as old as 109 or 84 million years, based on rpoB and gyrA.