Tn6060, a Transposon from a Genomic Island in a Pseudomonas aeruginosa Clinical Isolate That Includes Two Class 1 Integrons

Infections caused by multidrug-resistant gram-negative bacteria are a continuing and growing problem (3). A significant contributor within this cohort is Pseudomonas aeruginosa, a species that shows a great diversity of resistance profiles, with resistance to ß-lactam antibiotics being most notable (16). In recent years, a number of multidrug-resistant P. aeruginosa strains have been isolated that contain class 1 integrons with novel resistance cassettes, including several that contain determinants conferring resistance to β-lactams (5, 9, 16). Here we report the recovery of a VIM-1 metallo-β-lactamase-producing P. aeruginosa clinical isolate with a novel transposon that included two class 1 integrons.

Repeated urinary specimens from a patient in Sydney, Australia, with a long-standing, treated prosthetic knee infection yielded a highly multidrug-resistant P. aeruginosa strain. The isolate was carbapenem resistant and underwent further analysis for the presence of a metallo-β-lactamase. When noted to be VIM-1 positive, the isolate, recovered in 2008 (8) and here designated strain 37308, was referred and analyzed in detail. Initial screening by PCR with the class 1 integron-specific primers HS915 and HS916 (7) suggested the presence of such an element. To investigate the genetic context of this class 1 integron, a fosmid genomic library was constructed (1). intI1-positive clones were identified by PCR with the primers HS915 and HS916, and one such clone, F32, was completely sequenced. The cloned region was 43,062 bp in length. The strain was found to contain two closely linked class 1 integrons, both of which were contained within a transposon, here designated Tn6060.

Tn6060 was inserted within a previously identified genomic island, which, in the P. aeruginosa strain PACS171b (recovered from a cystic fibrosis patient [4]), is 36,840 bp long (accession number EU595750). The genomic island is not present in the P. aeruginosa strain PAO1. The equivalent island region in clone F32 is very similar (99% identity) to that of PACS171b over the length of available sequence (17,616 bp) in the former, with the notable exception of two deletions of DNA in the F32 clone of 113 bp and 2,859 bp in length. Insertion of Tn6060 generated 5-bp flanking direct repeats of the genomic island sequence, characteristic of transposition of Tn21-like transposons. The duplicated target comprised bases 9304 to 9308 in relation to the PACS171b genomic island sequence (Fig. 1A) .

• Open in new tab
• Download powerpoint

FIG. 1.

Tn6060 and its location in the Pseudomonas aeruginosa strain 171b genome fragment cloned as fosmid 32. (A) Sequence shown is the immediate region of the pathogenicity island as identified in PACS171b (accession number EU595750) that surrounds the transposon. Flanking numbers indicate the boundaries of the pathogenicity island referenced to the same database entry. Underlined sequence is the direct repeat generated by transposon insertion. (B) Detailed structure of transposon Tn6060. IRi and IRt represent the inverted repeats found at the integron end and transposition module end, respectively, of class 1 integrons/transposons. The hatched region in the middle represents a region of identity to pRMH760. Bent arrows identify PCR primers referred to in the text. Other key integron features are as defined previously (12, 15). Resistance cassettes are as defined previously (5, 9). (C) Diagrammatic representation of the two integrons, highlighting the relationship between the two integrons, sequence common to pRMH760, and direct transposon repeats. Hatched double lines indicate regions of identity to pRMH760 (accession number AY123253.3). The region labeled 3-4, when inverted, is sequence contiguous with region 1-2 in pRMH760. The pair of 5-base sequences indicated by arrows become direct repeats flanking IRi/IRt in the inverted sequence (see the text). RB, resolvase binding site; SA, small-array integron; LA, large-array integron; TM, Tn402-like transposition module.

The structure of Tn6060 is shown in Fig. 1B. The transposition module extending from the 38-bp inverted repeat to the point of insertion of the right-hand integron (Fig. 1B) displays only a 2-bp difference from sequence of Tn1403 (15). The point of insertion of the large-array integron (comprising the cassettes aadA1, aphA15, aacA4, and bla_VIM1) with reference to IRi of that integron is the same as that for In28 in Tn1403. However, the integron itself is very different from In28 and instead has a structure most closely related to that of In70.2 (14), with a copy of ISPa7 inserted in the 5′ conserved segment (5′-CS) at the same point where it appears in In70.2. In addition, this integron, while comprising a defective transposition module, lacks the insertion elements normally associated with the In4 family (including In28) or In5 family of class 1 integrons (10, 12, 15). Beyond the IRt inverted repeat is a 547-bp region identical to sequence within the urf2 gene, which is part of the Tn21-like mercury resistance transposons (11). The sequence abuts the IRi end of a second class 1 integron with a different, smaller cassette array, aacC3-cmlA5. This second integron is identical to the first with the exception of a different cassette array. Sequence outside the 5′-CS of this small-array integron (i.e., to the left of IRi in Fig. 1B) is identical to that adjacent to IRi of In70.2 in the sequence under GenBank accession no. AJ581664 up to the point where the latter sequence is available. The region between IRi of the small-array integron and the end of the Tn1403-like inverted repeat is 286 bases in length.

None of this sequence is related to that of Tn1403, including the Tn6060 inverted repeat at this end, suggesting that this region is derived from another, yet to be identified transposon.

There are precedents for replicons/transposons containing two integrons, and these probably arise by homologous recombination between conserved regions of integrons (2, 11) that then generates a tandem “head-to-tail” array in the product. In contrast, the two integrons here are arranged “tail to tail” (Fig. 1B) and are therefore present as large inverted repeats across the extensive sequences in common. In clone F32, this has resulted in an inversion between the two integrons at some point prior to sequencing. This is evident from the fact that the 5-bp sequence (CCGGG), immediately abutting IRi, of the small-array integron is the reverse complement of the sequence (CCCGG) located next to IRt of the large-array integron (Fig. 1C). This 5-bp duplication would have arisen from the transposition event that resulted in one of the class 1 integron/Tn402-like transposons in Tn6060 being inserted into the pRMH760-related sequence. As also shown in Fig. 1C, alignment of the indicated sequences with the pRMH760 (11) sequence also implies an inversion event.

To assess the orientation of the two integrons within the P. aeruginosa genome in strain 37308, we carried out PCR using four primers in pairwise reactions. One primer in each pair (HS983 [5′ TGGCAAGCATGTGTCAAAGAATGTG 3′] and HS984 [5′ CCGGCCTGACGCCTAGTTA 3′]) targets sequences beyond the two IRi sites, and the second in each pair targets sequences unique to the large-array (HS965 [5′ CTGTCGGATACTCACCACTC 3′]) and small-array (HS959 [5′ GGTAACTGTCGCTTGGTTGAG 3′]) integrons (Fig. 1B). In the orientation found in F32 (Fig. 1B), only two of the four PCRs will generate a product. These are HS983 with HS965 and HS984 with HS959. If the two integrons invert, the other two primer pair combinations will generate a product. Interestingly, all four PCRs carried out with DNA extracted from an overnight growth culture of strain 37308 generated a product, although the two pairs targeting the orientation opposite to that shown in Fig. 1 were significantly brighter (data not shown), implying that this second tail-to-tail orientation was the “parent” in this clone. In any event, it is clear that this region inverts at a high rate, particularly since this PCR method will detect only inversions involving DNA that is homologous between the respective primer pairs. This region between the primers constitutes only a small fraction of the DNA common to the two integrons.

We are unable to offer a precise genealogy for the creation of Tn6060. A number of events must have been involved, however, driven by both homologous recombination and site-specific resolution. One early event presumably involved homologous recombination between two class 1 integrons at different locations or between the 5′- and 3′-CS of two integrons, since the insertion points of the two integrons are different. Also, the right-hand integron (Fig. 1C) derived from the In28-like parent does not possess Tn1403-like sequence beyond IRt, even taking into account the inversion. Thus, the IRi end is located in the same position as In28 is in Tn1403, but the IRt end is not. Additionally, it is significant that both integrons have an identical class 1-specific 5′-CS with respect to the presence of ISPa7 and 3′-CS. Since it is unlikely that the integron that is analogous to In28 with respect to the insertion point in the transposition module had an In70.2-like backbone, we hypothesize that this integron's 5′-CS was derived from the second integron by a gene conversion-like event after the two came together.

One of the difficulties in describing likely ancestry is the fact that while both cassette arrays have been described previously, contextual information is limited in previously published reports. The large array, which includes VIM-1, is present in In70.2, and this has the same backbone over the available sequence (14). However, the published sequence extends only to the resI site just beyond IRi and stops before the end of the 3′-CS and the IRt boundary. Given that In70.2-like elements are common in P. aeruginosa isolates in Europe, it would be interesting to determine if this integron backbone is being disseminated by Tn6060. We found that Tn6060 is transposable to a conjugative plasmid (mean frequency, 9.4 × 10⁻³/donor over three independent assays), so this is possible. Despite the fact that the two arrays have been seen individually in P. aeruginosa clinical isolates (9, 14), they have not till now been reported together. However, this may be due partly to the experimental design of many studies, in which integron arrays are detected via primers that flank conserved regions of class 1 integrons (6). Where multiple arrays are present, biases in conventional PCR may preclude detectable amplification of larger arrays at the expense of smaller ones. Also, the point has been made previously (11, 15) that a variety of genetic elements and processes contribute to the spread of resistance and that detailed examination of the context in which such elements as integrons are located can be particularly insightful in comparing and understanding multidrug-resistant isolates (7, 13). More systematic and detailed sequence analyses may therefore shed light on the mechanisms by which integrons and their arrays move through pathogens.

• American Society for Microbiology