TEXT
Colistin is increasingly relied upon to treat infections caused by multidrug-resistant, carbapenemase-producing
Enterobacteriaceae (
1). Consequently, there is great concern surrounding plasmid-borne colistin resistance mediated by
mcr-1 (
2).
mcr-1 has now been reported in
Enterobacteriaceae from five continents: Asia, Europe, Africa, South America, and North America (reviewed in reference
3). A recent study detected
mcr-1 from
Escherichia coli collected in the early 1980s, suggesting that the emergence of
mcr-1 occurred much earlier than previously thought (
4). Given the global dissemination of
mcr-1 and the potential clinical consequences, understanding the molecular mechanisms underlying its mobility is critical. Here we provide a detailed examination of the genetic context surrounding
mcr-1 and present evidence that
mcr-1 is primarily mobilized by an IS
Apl1 composite transposon.
There is increasing evidence that IS
Apl1 plays a pivotal role in the mobilization of
mcr-1 (
5,
6). IS
Apl1 was first detected in
Actinobacillus pleuropneumoniae, an etiological agent of disease in swine (
7). It belongs to the IS
30 family and possesses features and activities similar to those of IS
30 members. It is flanked by 27-bp inverted repeats (designated IRL and IRR, for inverted repeat left and inverted repeat right, respectively) and contains a 927-bp open reading frame (ORF) that encodes a DD(E/D) superfamily transposase protein. Like IS
30 and other family members, IS
Apl1 moves through a covalently closed double-stranded DNA (dsDNA) circular intermediate containing 2 bp of host flanking DNA between the abutted transposon ends and, upon integration, generates 2-bp target site duplications (TSDs) (
7–9). Such circular intermediates are characteristic of a large number of IS families (
10) and appear to be produced by a replicative mechanism known as copy out—paste in (
11). Furthermore, the ends of these IS elements are often preferred target sites for insertion by the same IS (
12).
We employed the Geneious R9.1 software suite (Geneious, Auckland, New Zealand) to analyze 77
mcr-1-containing sequences, 19 of which were too fragmented to be useful (see Table S1 in the supplemental material). This represents every unique sequence present in NCBI as of this writing (August 2016). A common feature in all sequences is an ∼2,607-bp DNA segment containing
mcr-1 and a putative 765-bp ORF encoding a protein similar to a PAP2 superfamily protein (
Fig. 1). In 7 sequences, this segment is flanked by two copies of IS
Apl1 in the same orientation, reminiscent of a composite transposon (
Fig. 2A). In a further two sequences (pSA-MCR-1 and p100R), the downstream IS
Apl1 is in the reverse orientation. No TSDs flank these IS elements, but an analysis indicates that this is due to the inversion of a large segment of DNA that encompasses the downstream IS (data not shown). In 14 sequences, a single copy of IS
Apl1 is located upstream of
mcr-1 (
Fig. 3), while in the remaining 35 no intact copies of IS
Apl1 are present (see Fig. S1 in the supplemental material). Notably, in many of the structures lacking one or both IS
Apl1 copies, deletions and mutations at the extremities bestow some minor variability on the length of the
mcr-1 DNA segment. It has been previously shown that IS
30 family members exhibit a specific target preference for an imperfect palindrome sequence that has a slight GC bias in the center and an AT-rich distal region (
8). An alignment of all sequences containing at least one copy of IS
Apl1 reveals a significant bias for insertion in just such a structure, with a preponderance of thymine residues and adenine residues located directly upstream and downstream of the insertion site, respectively (
Fig. 1).
Of the 7 complete composite transposon sequences analyzed (see Table S1 in the supplemental material), five are flanked by the characteristic 2-bp TSDs of IS
Apl1 transposition (
Fig. 2A). Fortunately, for every example of the composite transposon insertion, there are multiple sequences in NCBI of the corresponding region without the insertion (empty sites). In every example, this region contains the same dinucleotide that forms the TSD (
Fig. 2B), as would be expected if
mcr-1 were mobilized as a composite transposon. Furthermore, in the two examples where the TSDs are missing (pECJP-59-244 and pS38), a comparison of the sequence with a corresponding region lacking the transposon identified deletions that removed the downstream TSD (
Fig. 2C). Taken together, these data provide compelling evidence that the
mcr-1 segment is mobilized as part of an IS
Apl1 composite transposon.
There are 14 examples, involving 4 distinct insertion sites, in NCBI where
mcr-1 is delimited by IS
Apl1 on the 5′ end only (see Table S1 in the supplemental material). However, 2 sequences (PEC2_1-4 and pKP81-BE) were not analyzed further due to a large deletion that removed most of the 3′ end, including the putative ORF. In the remaining 12, we identified 3 variants of the one-ended examples based on differences at the 3′ end. In the first example (
Fig. 3A), present in 9 of the 12 one-ended variants, the
mcr-1 segment ends where the putative ORF downstream of
mcr-1 terminates. There are significant variations in the last 6 bp of this segment, where mutations and deletions have resulted in distinct sequences (
Fig. 3A). In the second and third examples, an additional 33 bp and 90 bp, corresponding to the last 33 and 90 bp of IS
Apl1, respectively, are present immediately downstream of the ORF (
Fig. 3B). There are intact 27-bp IRRs of IS
Apl1 included in these regions. In the latter two structures, 2-bp TSDs flank this second IRR and the IRL of the upstream IS
Apl1 copy, with the exception of pVT553, where a subsequent deletion removed the downstream TSD (
Fig. 3B). Similarly to that observed in the analysis of the composite transposon structures, as shown by comparisons to corresponding empty sites in NCBI, the same dinucleotides as those in the TSD are present. In contrast, only pHNSHP45 has a putative TSD (GA) in the absence of this second IRR. However, an alignment of this region with multiple other homologous sequences deposited in GenBank reveals that this putative TSD is present in some but not others due to a poly(A) tract at this location (data not shown). Hence, it is unclear whether this is a true TSD or a sequencing artifact caused by the poly(A) sequence. Overall, these data indicate that IS
Apl1-mediated transposition of the
mcr-1 segment can also occur when an additional IS
Apl1 IRR is present downstream of the region without an assembled composite transposon. In contrast, there is scant evidence for the mobilization of
mcr-1 in the absence of this additional IRR.
There are 35 mcr-1 segments in NCBI that have no flanking ISApl1, 12 of which represent unique sequences (see Table S1 in the supplemental material). In every unit, the segment is delineated at the 3′ end by the putative ORF, though the same mutations and deletions that were present at the 3′ end in the one-ended structures are also present (see Fig. S1 in the supplemental material). Similarly, in 8 of the 12 sequences, the element is delineated on the 5′ end by a conserved trinucleotide (5′-ATA-3′), a sequence found immediately downstream of the IRR of ISApl1 (in both the composite transposon and single-copy ISApl1 variant cases) (see Fig. S1). In the remaining 4, deletions and mutations in this area have resulted in changes at this position (see Fig. S1). Notably, in one sequence (LVOP01000041), a partial copy of an upstream ISApl1, including the IRR, is still present (see Fig. S1).
In addition to providing a working hypothesis on the mobilization of mcr-1 by ISApl1, our analysis also allowed us to construct a plausible hypothesis on how this structure emerged.
An alignment of all composite transposon sequences revealed a conserved dinucleotide between the 3′ end of the IRR of the upstream IS
Apl1 and the beginning of the
mcr-1 segment (AT) and between the end of the
mcr-1 segment and the 5′ end of the IRL of the downstream IS
Apl1 (CG) (
Fig. 1). Furthermore, in every structure with just an upstream copy of IS
Apl1, the same conserved dinucleotide (AT) is also present between the IRR of the upstream IS and the beginning of the
mcr-1 segment (
Fig. 1). This suggests that the original composite transposon was formed by two separate insertions of IS
Apl1 flanking the
mcr-1 segment, generating characteristic 2-bp TSDs during the process. Subsequently, as this composite transposon moved, it generated new TSDs at each location, thus masking the original flanking TSDs.
Much of the uncertainty surrounding the movement of
mcr-1 has centered on the observation that many sequences lack one or both copies of IS
Apl1 (
13). However, Szabó et al. have shown with IS
30 that the loss of one or both IS elements in the composite transposon Tn
2706 can contribute to replicon stabilization through both transposition and illegitimate recombination (
14). In 37% to 57% of the plasmids analyzed by Szabó and colleagues, IS
30 was partially or completely removed by a process that generated mismatches and deletions in the area of excision (
14). An analysis of
mcr-1 segments having one or no copies of IS
Apl1 also shows a preponderance of deletions and mutations surrounding the area where IS
Apl1 would have been located (see
Fig. 3). For example, an alignment of 5 plasmids with high homology to pHNSHP45 (GenBank accession number KP347127) reveals that, while the plasmids are almost identical, there is a high concentration of mutations and deletions at the extremity of the
mcr-1 segment, precisely at the location where the downstream copy of IS
Apl1 is located in the composite transposon.
Overall, these data provide compelling evidence that mcr-1 is mobilized primarily as a composite transposon composed of directly orientated copies of ISApl1. In view of the established IS30 family transposition mechanism, it seems probable that the entire composite transposon transposes via a circular intermediate in which the left and right ISApl1 copies abut. Over the course of evolution, this composite transposon has lost one or both copies of ISApl1, most likely through a process of illegitimate recombination. ISApl1 loss may have served to increase the stability of mcr-1 in plasmid vectors, facilitating widespread dissemination of this gene. Functional experiments should further clarify the nature of how mcr-1 transposes and its origins, with significant implications for understanding dissemination of this important resistance gene.