Sir,
IMP enzymes are one of the class B-type metallo-β-lactamases found in Acinetobacter baumannii. So far, nine IMP enzymes have been reported in this species, many as isolated cases (GenBank accession numbers DQ845788 and AB184977).1 IMP β-lactamases are often present in class 1 integrons, which, in turn, are embedded in transposons, resulting in highly transmissible genetic units. The majority of epidemiological investigations of the antimicrobial resistance encoded by class 1 integrons have depended on PCR primers that anneal to the conserved 5′CS and 3′CS ends of the integron. Analysis of PCR amplicons yields detailed insight into the presence and composition of resistance-encoding gene cassettes. However, cassette-targeted PCR analyses do not reveal the potential for and history of horizontal transmission of integrons. Novel insight into the horizontal transfer potential and mobility of integron-encoded resistance genes in pathogenic bacteria can be obtained from determination of the genetic regions surrounding the integrons. We have analysed the flanking sequences of the blaIMP-5-containing integron In76 in the clinical A. baumannii strain 65FFC2 to determine its unit of transfer. We report on the identification of a new integron mobilizable unit in this clinical bacterium.
A. baumannii strain 65FFC is an imipenem-resistant clinical strain (MIC >32 mg/L) isolated from the urine of a man hospitalized in the neurotraumathology ward of the University Hospital of Coimbra, Portugal, in 1998. It produces the IMP-5 metallo-β-lactamase from the blaIMP-5 gene cassette present in a class 1 integron (GenBank accession number AF290912). The strain is resistant to all β-lactam antibiotics (except ampicillin/sulbactam). The flanking regions of In76 as well as the integron sequence (10 216 bp) were determined by direct genomic DNA sequencing. Up- and downstream primer walking was based on initial annealing of the sequencing primer in conserved parts of the integron (GenBank accession number M73819). Sequencing was performed using BigDye chemistry (Applied Biosystems) and the sequences were edited and aligned in Sequencher v.4.2.2 (GeneCodes, Ann Arbor, MI, USA), and identified using BLASTN (www.ncbi.nlm.nih.gov).
A total of 3258 bp of sequence was determined upstream of the 5′CS region and 3564 bp downstream of the 3′CS region. In76 was found embedded in a Tn402-like transposon. A miniature inverted-repeat transposable element (MITE)-like structure of 439 bp was identified immediately adjacent to the beginning of the 5′CS of the integron, preceded by an incomplete putative transposase. Downstream of the 3′CS, two genes (tniBΔ1 and tniA) were identified that belong to a common defective transposition module of integrons. However, the tniA gene was interrupted by a second MITE with an identical 439 bp sequence to the one at the 5′CS flanking region. An interrupted putative transposase followed the MITE structure further downstream (Figure 1a). The nucleotide sequence obtained in this study has been deposited in the GenBank database (accession number JF810083).
Schematic representation of the flanking regions of (a) the integron In76 of A. baumannii 65FFC (this study) and (b) the integron of A. johnsonii.3 The regions between the dashed lines are 100% identical (1720/1720 bp and 3728/3728 bp).
Schematic representation of the flanking regions of (a) the integron In76 of A. baumannii 65FFC (this study) and (b) the integron of A. johnsonii.3 The regions between the dashed lines are 100% identical (1720/1720 bp and 3728/3728 bp).
The presence of the complete In76 integron and the two immediately flanking MITEs, together with the identification of a 5 bp target site duplication adjacent to both the MITEs, strongly suggests that the entire structure had inserted into the transposase gene through transposition.
Bioinformatic analysis also revealed that another 439 bp MITE structure, 100% identical to our isolate from Portugal, was previously found flanking a class 1 integron of an Acinetobacter johnsonii isolated from the digestive tract of an ocean prawn in Australia (GenBank accession number FJ711439).3 Although the MITE-like structures are inserted in the same relative position within the integron, they flank two different class 1 integrons with unrelated gene cassettes, and the entire structure is inserted in a different genetic context (Figure 1a and b). The acquisition of the integron by A. johnsonii by a MITE-facilitated transposition-like mechanism was also proposed by Gillings et al.3 The 100% identical MITE-like structure present in two different Acinetobacter species from different continents suggests that MITEs can disseminate horizontally and act as mobilizable vectors for resistance dissemination.
MITEs are non-autonomous mobile elements consisting of small repeat sequences, which do not encode proteins and are found randomly inserted in the genome of diverse bacteria.4 The mobilization of MITE-like structures by transposition has been shown in Archaea, eukaryotic and bacterial genomes.4,5 Recently another MITE-like structure, called the integron mobilization unit (IMU), has been described flanking a defective class 1 integron harbouring a carbapenem resistance gene, blaGES-5, in a clinical Enterobacter cloacae strain. Although the IMU sequence is not related to the MITE reported in our study, IMU-based mobilization of the integron could be demonstrated using a transposase provided in trans.5 Similar mobilization has been described for MITEs.3,4
There are no data in the scientific literature on the prevalence of MITEs in clinical bacteria. This is the second report of this 439 bp MITE-like structure, but the first in a clinical strain of the emergent pathogenic species A. baumannii. Moreover, it is a unique structure surrounding an IMP-type carbapenemase, suggesting that this resistance-encoding integron can be mobilized by a transposition-like mechanism. Our study emphasizes the need to determine the genetic context of mobile elements to fully understand the history of and future potential for dissemination of resistance genes.
Funding
This work was supported financially by the Center for Pharmaceutical Studies, University of Coimbra and the University of Tromsø. S. D. was financially supported by grant SFRH/BD/49061/2008 from Fundaçãopara a Ciência e a Tecnologia, Lisboa, Portugal, and the Research Council of Norway (YGGDRASIL fellowship).
Transparency declarations
None to declare.
Acknowledgements
We thank Michael Gillings for fruitful discussions.
