Introduction

The IMP metallo-beta-lactamase can confer bacterial non-susceptibility or resistance to all beta-lactam antimicrobials, including the carbapenems which are frequently used to treat invasive Gram-negative infections. IMP-producing bacteria are most frequently reported from the South Pacific and Asia [1] and are present but less common in other regions of the world [2]. The bla_IMP-1 allele was first reported in Japan in 1994 as a gene cassette in a class 1 integron [3]. Also found in Japan, the bla_IMP-6 allele has been linked to human infections and hospital outbreaks [4–6]. Elsewhere in the South Pacific, reports from Australia and the Philippines have noted bacterial strains carrying bla_IMP-4 and bla_IMP-26 respectively [1]. Phylogenetic analysis of bla_IMP alleles shows that bla_IMP-6-related clades share phenotypic characteristics due to a Gly262Ser substitution compared to bla_IMP-1 [7]. Although bla_IMP-6 is found in class 1 integrons, bla_IMP-27, bla_IMP-64, and bla_IMP-95 that we have recovered in North America are almost exclusively found in class 2 integrons. Currently, these alleles have not been linked to human outbreaks or endemic bacterial strains and are recovered mostly commonly from One Health sources [8,9].

Class 2 integrons are non-mobile, with a truncated integrase gene, intI2, but maintain their promotor for gene cassette transcription. Class 2 integrons are frequently mobilized in Tn7 transposons [10]. Transposons play an important role in the spread of antimicrobial resistance in that they mobilize genetic elements, including antibiotic resistance genes, to allow their exchange between chromosome and plasmid [11]. The “cut and paste” ability of the Tn7 transposons provides the mechanism for the translocation [12,13] of chromosomally encoded class 2 integrons harboring the bla_IMP gene cassette to plasmids where they can be transferred horizontally within and between diverse bacterial species. We have observed bla_IMP within a novel class 2 integron which also harbors sat2 encoding streptothricin resistance. We found this class 2 integron co-harboring bla_IMP and sat2 gene cassettes in multiple taxonomic families of bacteria, both chromosomally and on mobile plasmids, suggesting the temporality of horizontal gene transfer (HGT).

The objective of this study is to characterize the genetic context of bla_IMP-27, bla_IMP-64, bla_IMP-95 alleles in isolates from the Comamonadaceae, Enterobacteriaceae, Moraxellaceae, Morganellaceae, and Shewenallaceae families recovered from One Health sources. Understanding the genotypic presentation is important for understanding these genes’ dissemination in a One Health context. This information may aid in our understanding of the temporality of horizontal gene transfer of the class 2 integron that harbors specific bla_IMP-27, bla_IMP-64, and bla_IMP-95 gene cassette alleles.

Materials and methods

Bacterial isolates harboring the carbapenemase genes, bla_IMP-27, bla_IMP-64, or bla_IMP-95, were originally recovered in Ohio and Missouri between 2015 and 2021 from various One Health sources including wastewater treatment plant effluent, livestock markets, fecal or lagoon samples from dairy and swine, and swabs from animal housing environments. Sampling plans were utilized at each site to ensure that individual samples each represented unique locations within a site. Permits were not required for sampling at these sites, access was provided voluntarily by the facility manager.

For recovery of carbapenem resistant Enterobacterales from samples, we used selective media supplemented with 0.5 μg/mL meropenem and 70 μg/mL zinc sulfate to identify phenotypic carbapenem resistance as previously described [9,14–16]. Carbapenemase production was confirmed in resistant isolates using the CarbaNP test [17], Bacterial species were initially identified by MALDI-TOF (Bruker Scientific, LLC, Billerica, MA), and species not expected to have intrinsic carbapenemase production were tested for the presence of known carbapenemase genes including bla_IMP by PCR as previously described [9,14–16].

Antimicrobial susceptibility phenotype of the isolates was determined using broth microdilution (Sensititre MIC panels CMV3AGNF or CMV4AGNF and GNX2F, Thermo Fisher Scientific, Oakwood Village, OH) following Clinical and Laboratory Standards Institute (CLSI) guidelines [18]. Whole genome sequencing of the isolates was performed using the Illumina MiSeq platform. Sequence data were quality-checked, adapters removed, then assembled for analysis using SPAdes [19].

Sequences were aligned to a complete Tn7 reference that begins and ends at terminal repeats (IR) of Tn7 (not shown, repeat regions of MN628641.1). Isolates that did not show any evidence of Tn7 in chromosomal contiguous sequences and were found on plasmids were then aligned with an IncQ1 plasmid and original reference KY126032, known to harbor the integron, In2–69. Sequence data were deposited in NCBI GenBank under the Bioproject number PRJNA1197433.

Initial antimicrobial gene identification was accomplished using the ARGAnnot, ResFinder, CARD ontology, and AMRfinder online databases [20–23]. IncQ1 plasmids were confirmed using PlasmidFinder [24]. The ISFinder database [25] was used to identify HGT motifs with initial identification of transposon, insertion sequence, and integrons. For exploration of other truncated and incomplete genes the conserved domain database was used for identifying regions for targeted BLAST searches. Reference genes were then identified using refSeq, available from NCBI GenBank [26]. As well, other annotation tools available in Proksee [27] were used for additional gene annotation [28–32].

Results

This study analyzed 32 bacterial isolates, including 12 Shewanella spp., 2 Providencia spp., 4 Acinetobacter spp., 3 Proteus spp., 7 Escherichia coli, 3 classified only as Enterobacteriaceae bacterium, and 1 Acidovorax sp. (Table 1) as identified by NCBI. Of these, 17 carried bla_IMP-27, 13 carried bla_IMP-64, and 2 carried bla_IMP-95 (Table 1). Most of the isolates were originally recovered from feces or housing environments of dairy cattle (n = 18) or swine (n = 11), but two were from livestock auction market environments and one from treated human wastewater effluent. In most of the isolates (n = 14), bla_IMP-27 was located within a complete Tn7 transposon (Figs 1, 2). While truncated Tn7 motifs contained bla_IMP-27 (n = 2) and bla_IMP-95 (n = 2), and another 13 isolates maintained bla_IMP-64 on an IncQ1 plasmid and in 1 isolate bla_IMP-27 was located on an RP4 plasmid (Table 1, Fig 2).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Taxonomic family, bla_IMP gene index, mechanism for horizontal gene transfer, isolate source, date and US state of collection, and NCBI Genbank SRA accession number for 32 bacterial isolates harboring bla_IMP obtained from One Health sources.

https://doi.org/10.1371/journal.pone.0327200.t001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Annotation and alignment of the Tn7 transposon from 12 Shewanella spp. isolates harboring bla_IMP-27 recovered from One Health sources to the Tn7 reference MT469878.1 In Fig 1 reference genes are represented by light blue arrows, and arrow direction indicates direction of transcription.

Triangles flanking the Tn7 motif represent the attachment site and inverted repeats of attTn7 (IR). Ovals represent gene cassette recombination sites, and purple shaded elements represent genes not found in the reference, but throughout our samples the carbapenemase producing, bla_IMP. Dashed lines surrounding elements indicate truncations. When not found within contiguous sequence, double lines represent this break at the end of contigs.

https://doi.org/10.1371/journal.pone.0327200.g001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Annotation and alignment of the Tn7 transposon from 2 Providencia spp. isolates harboring bla_IMP-27 recovered from One Health sources to the Tn7 reference MT469878.1, and 4 Acinetobacter spp. with gene map characterizations with class 2 integrons and alignment of IncQ1 plasmids to integron In2-69, and an RP4 plasmid map In Fig 2 reference genes are represented by light blue arrows, and arrow direction indicates direction of transcription.

Triangles flanking the Tn7 motif represent the attachment site and inverted repeats of attTn7 (IR). Ovals represent gene cassette recombination sites, and purple shaded elements represent genes not found in the reference, but throughout our samples the carbapenemase producing, bla_IMP. Dashed lines surrounding elements indicate truncations. When not found within contiguous sequence, double lines represent this break at the end of contigs.

https://doi.org/10.1371/journal.pone.0327200.g002

We observed taxon-specific differences in the genetic context of bla_IMP gene alleles. Shewanella spp. and Providencia spp. isolates only contained bla_IMP-27 within a complete Tn7. Among multiple Shewanella spp. and Providencia spp. in our study, the presence of the Tn7 transposon, with the class 2 integron containing the bla_IMP-27 gene cassette, was located adjacent to the glmS gene (glucosamine 6-phosphate synthase). Eleven of the Shewanella spp. isolates and both of the Providencia spp. contain Tn7 just upstream from glmS, a known Tn7 attachment site, attTn7 [33]. Only one Shewanella sp. isolate (MU_12B) did not show this attTn7 to be present near glmS (Figs S1–S5). All Shewanella spp. isolates, except MU_12B, had complete Tn7s as well as the specific attachment site, attTn7, downstream of glmS as shown in annotation of the contiguous sequence from Shewanella sp. isolate D_7_23B (Fig S3). This is the most common and known attachment site for Tn7 within bacteria [33,34]. This is also seen in a contiguous sequence of Providencia sp. isolate G_1, that shows the complete Tn7 adjacent to the known glmS attachment site. All but one Shewanella spp., isolate MU_12B, show a transposition event to the glmS attachment site. In chromosomal sequence, the Tn7 flanking inverted, direct repeats (IR) at attachment sites (Figs 1, S1–S3, S6) provide evidence that Shewanella spp. and Providencia spp. isolates acquired bla_IMP-27 from a transposition event.

In MU_12B, multiple upstream elements atypical for a Tn7 attachment site were observed (Fig S4). Where glmS is found in MU_12B there is no sign of a Tn7 element, and where the Tn7 is located, there are a number of other functional mechanisms (Figs S4, S5). MU_12B also shows the Tn7 right end 22 bp repeat region adjacent to the Tn7 motif and the Tn7 left end 22 bp repeat region adjacent to intI2 (Figs S1, S2, S4, S6). The three gene cassettes for the class 2 integron in MU_12B are found on two different contigs (S6, S7). In addition to bla_IMP-27, Shewanella spp. isolates also harbored an additional carbapenemase gene bla_OXA-48 or bla_OXA-48-like on distant contigs (data not shown).

In contrast, Acinetobacter spp. isolates show evidence of prior acquisition, by harboring bla_IMP-27 or bla_IMP-95 within truncated Tn7 motifs in chromosomally associated contiguous sequences, allowing continual vertical transmission, without a fully functional Tn7 transposition mechanism surrounding the class 2 integron (Fig 2). Without the genes required for a transposition event (tnsABCDE), there is no genomic evidence of a functional transposition element (Fig 2, isolates D_4_50A, D_5_7A, D_5_41A, and USDA_S33). In these isolates, the Tn7 gene array was not found beyond what is described in the gene maps (Fig 2).

Proteus spp., E. coli, and the Enterobacteriaceae bacterium isolates show that the IncQ1 plasmid retains the class 2 integron motif but does not have evidence of a Tn7 transposition event as no evidence of any Tn7 motifs were found within chromosomal or other plasmid contigs for these isolates (Fig 2).

Discussion

The genetic context of bla_IMP varied by taxonomic family, reflecting different HGT mechanisms. Shewanella spp. and Providencia spp. isolates all maintained bla_IMP-27 within a complete chromosomal Tn7 transposon while Acinetobacter spp. maintained bla_IMP-27 and bla_IMP-95 within a truncated, incomplete chromosomal Tn7. Other Enterobacteriaceae including Proteus spp., E. coli, and the isolates classified by NCBI as Enterobacteriaceae bacterium harbored bla_IMP-64 on an IncQ1 plasmid, and a single Acidovorax sp. isolate harbored bla_IMP-27 and sat2 gene cassettes on a RP4 plasmid, absent association to a class 2 integron, and adjacent to an ISL3. This suggests that taxonomy may influence both the specific bla_IMP allele present and the predominant HGT mechanism, whether via transposition or plasmid exchange.

Shewanella spp. and Providencia spp. isolates are the only taxonomic families where the complete Tn7 was found and only harbor the bla_IMP-27 gene allele. Acinetobacter spp. isolates were found with incomplete Tn7 chromosomally, missing the functional transposition elements (tnsABCDE), but with evidence of a prior transposition event due to the IR of attTn7 found up stream of the class 2 integron, and with no other functional Tn7 elements found in other contiguous sequences in these isolates (Fig 2). Acinetobacter spp. isolates harbor both bla_IMP-27 and bla_IMP-95. Curiously, all bla_IMP-64 genes were only found in IncQ1 plasmids present in Proteus spp., E. coli, and the isolates identified by NCBI only as Enterobacteriaceae bacterium. A single RP4 plasmid within an Acidovorax sp. isolate harbors a bla_IMP-27 and sat2 but is absent of the class2 integron, with a ISL3 transposase in its place.

Tn7 is a transposon known for the ability to “cut and paste” into a specific chromosomal target site. TnsABC + TnsD within Tn7 control and direct transposition to the attTn7, directly downstream of glmS [33,34]. This site-specific transposition location allows incorporation into the host bacteria, without the inactivation of any of the host bacteria’s genetic mechanisms [34]. Tn7 also has target site immunity, not allowing multiple copies of Tn7 to exist within the same host [34]. TnsABC + TnsE is the mechanism by which the Tn7 is directed into conjugative plasmids [11]. A single mutation in tnsA can induce extensive DNA replication, converting Tn7 into a replicative transposon [35], thus potentially increasing gene cassette copy number.

Tn7 transposons have long been identified as carriers of class 2 integrons [36], so it is unsurprising to note the class 2 integron containing a bla_IMP gene cassette within a Tn7, as is the case with all Shewanella spp. and Providencia spp. isolates. The functionality of the Tn7 and the ability to cut and paste from chromosome to plasmid, and then plasmid to chromosome, gives the ability of resistance gene cassettes to be found across diverse bacterial species [12,13]. When the class 2 integron is found within the Tn7, any associated HGT is the result of a transposition event.

Complete Tn7s found in our isolates differ from Tn7 alignment references (MT469878.1 in Figs 1 and 2, and MN62864.1 in Figs S1–S3) in that the first gene cassette found in class 2 integrons, typically dfrA1 (trimethoprim resistance) has been replaced by a bla_IMP (carbapenem resistance) (Fig 1). In the typical 2^nd and 3^rd gene cassettes, found in most isolates are genes for sat2 and aadA1 (resistance to streptothricin and streptomycin/ spectinomycin, respectively). The exceptions for the 2^nd gene cassette being in D_7_23B and W_P_2A where the typical sat2 gene has been substituted by linG and lnuF respectively (Figs 1 and 2). In isolates where the bla_IMP and sat2 gene cassettes are found in plasmids, the typical aadA genes in the 3^rd cassette position are absent (Fig 2).

Morganellaceae harboring bla_IMP-27 within a Tn7 transposon were previously reported for 23 Canadian isolates identified from routine surveillance for carbapenemase production among Gram-negative bacteria [37]. Curiously, they did not note the second gene cassette for sat2 that we observed being present in the resistome, with the caveat that sat2 is not in the ResFinder database and so may not have been detected. We have found that ResFinder consistently fails to detect sat2, emphasizing the need to consult multiple resistant gene annotation tools. We have found that using multiple annotation tools including ARGAnnot and CARD provide further confirmation of the correct annotation for the sat2. In other GenBank references, we have found the bla_IMP-27-sat2-aadA1 gene cassette motif to be consistently present (as noted in Boyd, 2021 Fig 1) [37–39]. Less common Enterobacterales, such as Providencia spp. and Proteus sp., have been previously noted to harbor bla_IMP [40]. Class 1 and class 3 integrons are established carriers of major bla_IMP alleles [41]. Alleles of bla_IMP and their distribution in bacterial species have been described [42]. A study where eleven isolates from American crows were found to carry two similar Tn7 genetic clusters to one of those within our study (W_P_2A), with the same bla_IMP-27 motif in a Tn7, on chromosomes of Providencia rettgeri, with a similar substitution in the second gene cassette, of an lnu(f)-like rather than a sat2 gene [8]. The bla_IMP-27 has also been previously found on transferable plasmids in Providencia spp. and Proteus spp. but without the plasmid replicon type identified. [43].

However, when a truncated Tn7 contains a class 2 integron, we can only speculate as to how the vertical transmission of the resistance gene motif is maintained and when and whether further mutations allow this element to be further spread horizontally. It is perhaps possible that the integron was acquired through transformation in the Acinetobacter spp., as there are no plasmids, or complete Tn7 motifs present, and 3 of the 4 isolates are predictive for CRISPR-Cas system [44].We first identified the IncQ1 plasmid harboring a bla_IMP-64 gene cassette in a class 2 integron [16]. IncQ plasmids are small, non-conjugative, and self-replicating. Although non-conjugative, they are efficiently transferred by a mobilization process which is attributable to other conjugative, helper plasmids, or by parasitizing of type IV secretion systems [45]. The presence of complete class 2 integrons on IncQ1 plasmids, absent evidence of transposition, suggests prior acquisition via transposition or transformation in an ancestral bacterial lineage, followed by stable maintenance through mutation. The IncQ plasmid is of importance in a One Health context as it has a very broad host range, notably it has been identified in wastewater as an important reservoir for antimicrobial resistance genes [46].

It is unclear how a RP4 plasmid picked up the two resistant gene cassettes bla_IMP-27-sat2 that are typically associated with a class 2 integron. The functionality of the resistance genes is determined by the promoter region of the integrase, and in this case, would therefore maintain the carbapenemase functionality from a promoter region within the ISL3. As all isolates were identified as carbapenemase producers, and no other genomic analysis shows any other carbapenem resistance genes present within the Acidovorax sp. isolate, it is evident that the bla_IMP-27 is functional.

As has been noted in other studies, the prevalence of certain bla_IMP alleles may be underestimated and unexamined depending on what carbapenemase gene PCR primers are used, laboratory assays and carbapenemase production confirmation analysis [37–39]. Our results may be limited because of the use of short read sequencing, where genetic motifs are determined based on assemblies and alignment when Tn7, integrons, and resistance genes are found in different contigs. Had we been able to utilize long-read sequencing, gene location would have been more precise. It is also possible that some mobile elements were not recognized within certain bacterial species. PlasmidFinder did not identify the RP4 plasmid, although both SPAdes (with plasmid specification) [19] and Galaxy [47] gave circularized closed assemblies, and careful annotation identified the origin of transfer, and known RP4 traJ and traJ-II genes. It is possible that although PlasmidFinder did not identify any plasmids within the Shewanella spp. and Providencia spp., and there were no obvious plasmids on assembly, that mechanisms could exist for the Tn7 to become mobilized to a plasmid if we could better identify certain plasmid motifs. As bioinformatic tools improve, it is likely that isolates like these obtained from One Health sources may further help us understand the temporality and spread of the bla_IMP-27 bla_IMP-64 and bla_IMP-95 gene alleles.

Supporting information