Abstract

TEXT

Carbapenem-hydrolyzing β-lactamases belonging to Ambler classes A, B, and D have been found worldwide among the Enterobacteriaceae (24). Although carbapenem-hydrolyzing class D β-lactamases (CHDLs) are identified mainly in Acinetobacter species (19, 23), OXA-48 has been identified only in the Enterobacteriaceae. The bla_OXA-48 gene was first identified in a Klebsiella pneumoniae isolate from Turkey (22). Since then, several other OXA-48-producing isolates of various enterobacterial species (including Citrobacter freundii and Escherichia coli) have been reported, first mainly in Turkey and then in Belgium, France, Lebanon (16), Egypt, Israel (12), Senegal (18), Morocco (25), and Tunisia (9). In addition, there have been recent reports of OXA-48-related enzymes, such as OXA-181, being identified in India (7) and in other countries but always with a link to India (26), as well as OXA-163 from Argentina (21). It is noteworthy that OXA-48 and OXA-181 hydrolyze penicillins and carbapenems but spare extended-spectrum cephalosporins (27, 28). Many OXA-48 producers coexpress an extended-spectrum β-lactamase (ESBL) (most often CTX-M-15 or SHV-12), making those isolates resistant to all β-lactams available.

Previous studies indicated that plasmids carrying the bla_OXA-48 gene from different enterobacterial isolates, different clones, and different countries may share very similar features (6, 10). They were self-conjugative, very similar in size (reported to be ca. 60 to 70 kb), did not encode additional resistance markers, and were not typeable by using PCR-based replicon typing (PBRT) (4). This suggested the possibility of a wide spread of a single plasmid at the origin of dissemination of this carbapenemase gene. This prompted us to determine the sequence of one of these plasmids, to compare the identified structure with those of other OXA-48-positive plasmids, and to determine if some particular features could explain its successful spread. It is noteworthy that a 7.6-kb plasmid carrying the bla_OXA-181 gene and recovered from a K. pneumoniae isolate from Oman was fully sequenced. This revealed that it belonged to the broad-host-range ColE family of plasmids and that the bla_OXA-181 gene had been acquired by an ISEcp1-related one-ended transposition mechanism (27), two features clearly differentiating it from the bla_OXA-48 plasmids.

Plasmid pOXA-48a was extracted from the E. coli transconjugant recovered from the K. pneumoniae 11978 clinical isolate (22) by using the Qiagen Maxi kit (Qiagen, Courtaboeuf, France). The complete sequencing work flow of the Illumina genome analyzer IIx system was performed by the DNA Vision company (Gosselles, Belgium).

Sequence analysis of pOXA-48a revealed that it was 61,881 bp in size, with an average G+C content of 51.1% (Fig. 1). Each predicted protein was compared against the GenBank protein database using BlastP (http://blast.ncbi.nlm.nih.gov/Blast.cgi) with a minimum cutoff of 30% identity over 80% length coverage. Gene sequences were further compared and aligned with GenBank data using BlastN and ClustalW software. The IncL/M plasmids pCTX-M3 (GenBank accession number AF550415), which harbors the bla_CTX-M-3 ESBL gene, and pHK-NDM (GenBank accession number HQ451074), which harbors the metallo-β-lactamase bla_NDM-1 gene, were used as references for annotating pOXA-48a.

Open in a separate window

Fig 1

Major structural features of plasmid pOXA-48a from K. pneumoniae 11978 (22) in comparison with the IncL/M plasmids pCTX-M-3 (11) and pNDM-HK (14). White boxes indicate plasmid scaffold regions that are in common among plasmids. The locus Tra within the box is indicated. Resistance genes are indicated by orange boxes, except the β-lactamase genes, which are indicated by blue boxes. Transposon-related genes (tnpA, tnpR, and tnpM) and insertion sequences are indicated by red boxes. Replicase genes are indicated by purple boxes.

In silico analysis showed that pOXA-48a possessed the highest degree of sequence identity and gene synteny within its entire scaffold with the reference plasmid pCTX-M3 (97% identity at the nucleotide level) (11), with the exception of two loci. One region that differs from pCTX-M3 corresponded to Tn1999, which consists of two copies of the same insertion sequence, IS1999, flanking a DNA fragment that includes the bla_OXA-48 gene (1, 21). As previously shown, transposon Tn1999 was inserted within a tir gene similar to that responsible for inhibition of transfer of plasmid R100 (17). A target site duplication of 9 bp (CGTTCAGCA) was identified on each extremity of Tn1999 (1) (Fig. 2). Downstream of bla_OXA-48, a lysR gene encoding a regulatory protein of 304 amino acids was identified (98% amino acid identity with that of Shewanella oneidensis MR1) (13), followed by a truncated fragment of a gene encoding an acetyl coenzyme A carboxylase sharing 100% amino acid identity with that of S. oneidensis MR1 (13). In pCTX-M3, the ISEcp1-bla_CTX-M-3 transposable locus was inserted close to this target site for Tn1999, suggesting that this locus could possibly be a hot spot for integrating foreign DNA within IncL/M-type plasmids. The other region differing from other IncL/M plasmids was located between the replication module and the trb transfer locus. In pCTX-M3, this 27-kb resistance island was composed of a copy of Tn2 in which transposon Tn1548 carried the armA methylase encoding gene (2). That region contained at least 19 open reading frames (ORFs), including in particular genes encoding resistance to aminoglycosides (aadA2 and armA), trimethoprim (dfrA12), sulfonamides (sul1), and macrolides (mphE and msrE) (Fig. 1). In pHK-NDM, the bla_NDM-1 gene locus was inserted within the transposon carrying the armA gene, replacing the class 1 integron structure identified in pCTX-M3. In pOXA-48a, no foreign DNA was inserted in the backbone at this locus.

Open in a separate window

Fig 2

Genetic environment and content of Tn1999. The different gene names are indicated. The sequence of the 9-bp target site duplication is underlined.

The RepA replication protein of pOXA-48a exhibited only two amino acid differences, out of 351 amino acids, from RepA of both plasmids pCTX-M3 and pHK-NDM. Only one replication module was identified in pOXA-48a, as for other IncL/M-type plasmids known to possess broad-host-range properties that have since been identified in the Enterobacteriaceae, Erwinia amylovora, Ralstonia eutropha, Pseudomonas aeruginosa, and Agrobacterium tumefaciens (11, 17). Altogether, these results indicate that plasmids pOXA-48a, pCTX-M3, and pHK-NDM may have a common origin in the environment, a hypothesis which is reinforced by the identification of bla_OXA-48-like genes in the waterborne environmental species S. oneidensis and Shewanella xiamenensis (23, 28).

In order to better address whether plasmid pOXA-48a may possess a broad host range of replication, electrotransformation assays were carried out using DNA of pOXA-48a and different recipient species, namely, E. coli JM109 (control), Acinetobacter baumannii CIP70.10, Pseudomonas aeruginosa PU21, and S. oneidensis MR-1, with selection based on ticarcillin (50 μg/ml) (13, 29). Different protocols were used depending on the species of the recipient strain (8, 30). Whereas no transformant was obtained with A. baumannii and P. aeruginosa, pOXA-48a replicated efficiently in E. coli and S. oneidensis. This indicates that pOXA-48a possesses a broad host range of replication and that such an IncL/M-type scaffold could have been at the origin of the capture of the bla_OXA-48 gene from its natural progenitor (Shewanella spp.) and its dissemination among the Enterobacteriaceae. Our results also suggest that dissemination of the bla_OXA-48 gene in A. baumannii or P. aeruginosa is unlikely to occur if it remains associated only with this type of plasmid.

The transfer operon identified in pOXA-48a was very similar to the transfer operon in pCTX-M-3, being formed by two distinct regions. This common feature suggests that the propensity for spreading of the bla_OXA-48 gene in the Enterobacteriaceae might be as high as that of the bla_CTX-M-3 gene. This transfer operon included the trb and tra operons, known to encode type IV pili, which are involved in plasmid conjugation. The tra operon was very similar to that of pCTX-M-3, except for three genes, namely, traX-traY-excA, encoding proteins more distantly related (54%, 79%, and 34% identities, respectively) to those encoded by pCTX-M-3. Due to the significant divergence of this gene array, a recombination event could explain such gene array replacement. The lack of identification of pOXA-48a as an IncL/M plasmid by the PBRT method can be explained by the location of one of the corresponding primers in the excA gene.

In order to evaluate whether the modifications observed in pOXA-48a could influence its conjugation rate compared to that of another IncL/M-type plasmid, conjugation frequencies were determined by using as a donor the same E. coli J53 background harboring either pOXA-48a or pNDM-HK, along with E. coli JM109 as the recipient, with a selection based on ticarcillin (50 μg/ml) and nalidixic acid (20 μg/ml). No difference was observed, conjugation rates being 3.3 × 10⁻⁵ and 2.5 × 10⁻⁵ for pOXA-48a and pNDM-HK, respectively).

Since the PBRT technique was inefficient for classifying pOXA-48a, three primer pairs were newly designed to amplify conserved regions of its backbone (Table 1). They were designed in the repA, traU, and parA genes, encoding proteins implicated in replication, transfer, and partitioning, respectively. The presence of the bla_OXA-48 gene was confirmed using previously designed primers (Table 1). An international collection of strains (n = 19) harboring the same bla_OXA-48 gene was used for typing bla_OXA-48-positive plasmids. Plasmid analysis performed by using the Kieser technique (15) confirmed that they were all ca. 62 kb in size. PCR experiments confirmed that this same plasmid backbone was identified in all the bla_OXA-48-positive isolates (Table 2). In addition, PCR performed as described previously (17) confirmed that transposon Tn1999 targeted the same location in all plasmids. Altogether, our results show that the current dissemination of OXA-48 producers, mostly in countries located near the Mediterranean Sea and in Western Europe, may be due largely to the spread of a single plasmid.

We report here the whole sequence of the major vehicle of the bla_OXA-48 carbapenemase gene. Although some other widespread carbapenemase determinants, such as the bla_KPC and bla_NDM genes, have been shown to disseminate through different plasmid scaffolds (3), we showed here that the current spread of the bla_OXA-48 gene is linked to the wide diffusion of an identical IncL/M plasmid scaffold.

ACKNOWLEDGMENT

Footnotes

REFERENCES