Abstract
Antimicrobial resistance genes (ARGs) are associated with mobile genetic elements (MGEs) that conscript useful genes into the human–microbe and microbe–microbe battlefields. Thus, under intense selective pressure, ARGs have been constantly adapting and evolving, spreading among microbes. tmexCD-toprJ gene clusters, which encode resistance–nodulation–cell division (RND)-type efflux pumps, confer multidrug-resistance to clinically important antimicrobials, including tigecycline. Noteworthily, these gene clusters have emerged in gram-negative bacteria in humans, animals, and the environment worldwide by MGE-mediated transfer. Here we show a hidden MGE, strand-biased circularizing integrative element (SE), that is recently recognized to mediate transpositions of ARGs, associated with the spread of tmexCD-toprJ gene clusters. We identified multidrug-resistant isolates of Aeromonas species in a water environment in Vietnam that harbored multiple copies of tmexCD-toprJ in their chromosomes that were associated with SEs. In particular, Aeromonas hydrophila NUITM-VA1 was found to harbor two copies of a novel variant of tmexC3.3D3.3-topJ1 within cognate SEs, whereas Aeromonas caviae NUITM-VA2 harbored four copies of a novel variant of tmexC2D2.3-topJ2 within cognate SEs. Based on the nature of SE to incorporate a neighboring sequence into the circular form and reinsert it into target sites during transposition, we identified the order of intragenomic movements of tmexCD-toprJ gene clusters. Altogether, our findings suggest that most known subgroups of tmexCD-toprJ and their subvariants underwent transpositions among bacterial chromosomes and plasmids via SEs. Hence, a tmexCD-toprJ gene cluster ancestor may have been initially mobilized via SE, subsequently spreading among bacteria and evolving in new hosts.
Introduction
Antimicrobial resistance genes (ARGs) are estimated to have originated in environmental bacteria and subsequently spread to pathogenic bacteria in humans as acquired ARGs (1). Mobile genetic elements (MGEs) are the major driving force of gene transfer and amplification underlying ARG evolution (2). Spontaneous mutations and antimicrobial selection have led to the emergence of clinically problematic ARGs, including carbapenemase genes (e.g. blaNDM, blaKPC, blaOXA-48-like, blaIMP, blaVIM, and blaGES-5-like) (3), colistin resistance genes (e.g. mcr and arnT) (4, 5), and tigecycline resistance genes [e.g. tet(X) and tmexCD-toprJ] (6).
Carbapenem, colistin, and tigecycline are considered last-resort antimicrobials for infections caused by multidrug-resistant (MDR) gram-negative bacteria, such as Enterobacterales, Pseudomonas aeruginosa, and Acinetobacter baumannii, which are serious global public health threats (3, 4, 6). tet(X4) and other tet(X) variants, which encode flavin-dependent monooxygenases that catalyze tigecycline degradation, have emerged mainly in Enterobacterales and Acinetobacter species (3). tmexCD1-toprJ1, tmexCD2-toprJ2, tmexCD3-toprJ1 (initially designated tmexCD3-toprJ3), tmexCD4-toprJ4, and other tmexCD-toprJ subvariants, which encode resistance–nodulation–cell division (RND)-type efflux pumps that excrete multiple antimicrobials, including tigecycline, have emerged worldwide mainly in Enterobacterales and Pseudomonas species (3, 7–16).
Recently, a novel MGE family, named strand-biased circularizing integrative element (SE), was identified in Vibrio alfacsensis (17) and later in more diverse taxa of Gammaproteobacteria (18). SEs move between genomic locations via a copy-out-like route using tyrosine recombinases: SEs incorporate 6 bases next to the 5′-end of one specific strand into their circular intermediate form, which are then reinserted into a new target location. Therefore, SE-mediated transposition results in the insertion of 6 bases at newly formed attR recombination sites (17), and this 6-bp footprints allow tracing the order of SE integration events that occurred in the past in a genome.
In this study, we investigated MDR bacteria in a water environment in Vietnam and found MDR isolates of Aeromonas species that harbored multiple genes conferring resistance to last-resort antimicrobials, including tmexCD-toprJ gene clusters. Close examination of MGEs associated with multiple copies of tmexCD-toprJ in the chromosomes indicated involvement of SEs in the intragenomic amplification of the gene clusters. Comparison of sequences around known variants of tmexCD-toprJ in publicly available genome data revealed that most of them were likely integrated into bacterial chromosomes and plasmids via SEs.
Results and Discussion
Two carbapenem- and tigecycline-resistant isolates of Aeromonas species, namely NUITM-VA1 and NUITM-VA2, respectively, were obtained from a water environment in Vietnam in 2021. Whole-genome sequencing analysis revealed that NUITM-VA1 (accession no. AP025277) and NUITM-VA2 (accession no. AP025280) were 97.1% and 97.9% identical to Aeromonas hydrophila strain ATCC 7966T (accession no. CP000462) and Aeromonas caviae strain CECT 838T (accession no. JAGDEN000000000), respectively. Multilocus sequence typing analysis revealed that NUITM-VA1 and NUITM-VA2 belonged to novel sequence types (STs), ST1063 and ST1064, respectively, of Aeromonas species.
A. hydrophila NUITM-VA1 harbored multiple clinically important ARGs, such as intrinsic genes of cephalosporinase (blacepH) and oxacillinase (blaaphH), tigecycline resistance genes [tet(X4) and tmexCD3-toprJ1-like], phosphoethanolamine transferase gene conferring colistin resistance (mcr-3.9), and efflux pump gene conferring fluoroquinolone resistance (qnrVC4), whereas A. caviae NUITM-VA2 harbored carbapenemase genes (blaNDM-1, blaKPC-2, and blaVIM-4), tigecycline resistance genes (tmexCD2-toprJ2-like), and 16S ribosomal RNA methyltransferase gene conferring aminoglycoside resistance (rmtB) (Figs. 1A and B). Noteworthily, NUITM-VA1 and NUITM-VA2 harbored multiple copies of tigecycline resistance genes. Consistent with this observation, NUITM-VA1 and NUITM-VA2 showed low susceptibility to most antimicrobials tested, including carbapenems, cephalosporins, aminoglycosides, fluoroquinolone, and colistin with some exceptions (Table 1).
Indeed, two copies of tmexCD3-toprJ1-like gene clusters and four copies of tet(X4) (including the pseudogene) associated with the insertion sequence ISCR2 were identified in NUITM-VA1, whereas four copies of tmexCD2-toprJ2-like gene clusters were identified in NUITM-VA2 (Fig. 1A and B). The identity for tmexC3, tmexD3, and toprJ1 in NUITM-VA1 with the corresponding component genes of tmexCD3-toprJ1 (accession no. CP066833) (10) were 99.91% (with one amino acid substitution, T235A), 99.90% (with V56E, Q283H, and G591V substitutions), and 100%, respectively. The identity for tmexC2, tmexD2, and toprJ2 in NUITM-VA2 with the corresponding component genes of tmexCD2-toprJ2 (accession no. CP054471) (9) were 100%, 99.97% (with V56E substitution), and 100%, respectively.
Recently, several subvariants of genes comprised in the tmexCD-toprJ cluster were identified, including tmexD1.2 (tmexD1 variant with V64I substitution) in Klebsiella pneumoniae pC5921_mex (IncFIB/IncHI1B/IncU plasmid, accession no. MZ532979) and tmexD2.2 (tmexD2 variant with V56E and P382A substitutions) in Klebsiella oxytoca pC7532_mex (IncFII/IncU plasmid, accession no. MZ532981) (15). In addition, known gene variants of the representative tmexCD-toprJ cluster (Table 2), including tmexC3.2 (tmexC3 variant with Q187H, T256M, and A386T substitutions) and tmexD3.2 (tmexD3 variant with V610L and L611F substitutions) in Klebsiella aerogenes pNUITM-VK5_mdr (IncC/IncX3 plasmid, accession no. LC633285) identified in our previous study (11). Based on these naming-standards, we designated the novel variants of tmexCD-toprJ identified in NUITM-VA1 and NUITM-VA2 as tmexC3.3D3.3-toprJ1 and tmexC2D2.3-toprJ2, respectively.
To elucidate the molecular mechanism of intragenomic transposition of tmexCD-toprJ, comparative analysis among tmexCD-toprJ-containing genomic regions in NUITM-VA1 and NUITM-VA2 was performed. Overall, tmexCD-toprJ and the regulator gene tnfxB encoded upstream of the gene cluster were found to be associated with an atypical MGE family, SE, that consists of four conserved genes of integrases (intA and intB), tyrosine recombinase-fold protein (tfp), and SE-associated recombination auxiliary protein (srap) (17, 18) (Figs. 1C and D).
In NUITM-VA1, one copy of tmexC3.3D3.3-toprJ1-containing SE was flanked by 5′-CATCGA-3′ in attL and 5′-TATCGA-3′ in attR, whereas the other SE copy was flanked by 5′-TATCGA-3′ and 5′-CATCGA-3′ (Figs. 1A and C). Given the nature of the 6-bp footprint of SE transposition (17, 18), it is likely that one SE copy became the donor for the other SE copy in the second location in NUITM-VA1, but the donor location could not be defined. In NUITM-VA2, four copies of tmexC2D2.3-toprJ2-containing SEs were detected in the chromosome. The 6-bp fingerprint 5′-TATCGA-3′ was identified next to the 5′-end of the SE copy at the location #1 (4,079,313–4,062,461 nt), as well as next to the 3′-end of the SE copy at the location #2 (3,001,983–3,018,834 nt), and the other SE copy at the location #3 (2,130,980–2,114,127 nt) (Figs. 1B and D). Thus, the SE copy at the location #1 was likely the donor of the other SE copies at locations #2 and #3. Moreover, the 6-bp fingerprint 5′-TATCGA-3′ was identified next to the 5′-end of the SE copy at location #2, as well as next to the 3′-end of the SE copy at the location #4 (1,131,379–1,114,526 nt), suggesting that the SE copy at the location #2 was the donor of the SE copy at the location #4.
Importantly, identical structures as those of tmexC3.3D3.3-toprJ1-containing SE in NUITM-VA1 and tmexC2D2.3-toprJ2-containing SE in NUITM-VA2 were detected in chromosomes of A. caviae WCW1-2 (accession no. CP039832) (19) and A. caviae K333 (accession no. CP084031) (20), respectively. Moreover, the tet(X4)-containing region in NUITM-VA1 was also detected in the WCW1-2 chromosome, suggesting that these genetic structures around ARGs in Aeromonas species isolates herein described are not an unusual event.
The known tmexCD-toprJ gene cluster can be divided into four major subgroups, consisting of tmexCD1-toprJ1, tmexCD2-toprJ2, tmexCD3-toprJ1, and tmexCD4-toprJ4 (3, 7–16). Hence, the presence of SEs in these tmexCD-toprJ subgroups and their subvariants was examined next (Fig. 2). The intact sequences of tnfxB-tmexCD-toprJ-containing SEs were identified in all subgroups except tmexCD4-toprJ4 (Fig. 2A). Although some SEs, such as tmexCD2-toprJ2-containing SEs in Raoultella ornithinolytica NC189 (accession no. CP054471 and MN175502) (9), tmexC2D2.3-toprJ2-containing SE in A. caviae NUITM-VA2 in this study, and tmexC3D3-toprJ1-containing SE in Pseudomonas terrae subsp. cibarius SDQ8C180-2T (accession no. CP073356) (12), harbored insertions of other MGEs within their SEs, such as insertion sequences, the 5’- and 3’-end of the SEs were completely conserved, indicating that they can form functional mobility structures. Interestingly, IncP-2 megaplasmids, such as P. aeruginosa pBT2436 (accession no. CP039989) (14, 21), was previously suggested to frequently carry tmexCD-toprJ; thus, plasmids also seemed to be involved in such SE-mediated transpositions.
Broken structures close to tmexCD-toprJ-containing SEs that lacked the conserved genes of SE and tmexCD-toprJ-associated tnfxB were also detected (Fig. 2B). This type of broken structures may have concealed the herein described close relationship between tmexCD-toprJ and SEs in previous studies. For example, tmexCD1-toprJ1 in K. pneumoniae pMH15-269M_1 (IncFIB/HI1B plasmid, accession no. AP023338) (8) and Citrobacter portucalensis pHN21SC92-1 (IncC plasmid, accession no. CP089438) (16) was previously suspected of being mobilized by IS26, but intact 3’-end sequence of SE was present on pMH15-269M_1 but not on pHN21SC92-1, suggesting the involvement of other MGEs rather than SE is an event that occurred after initial SE-mediated transpositions. For tmexCD4-toprJ4 in Klebsiella quasipneumoniae pHNLW22-2 (untypeable plasmid, accession no. CP089443) (13), no evidence suggested SE involvement on either the 3’ or 5’ sides of the gene cluster; nevertheless, since this tmexCD-toprJ variant has only been reported in one case to date, a possible association with SE cannot be ruled out.
In conclusion, our present study provides a glimpse into Aeromonas species, one of the most common environmental bacteria that have been rapidly and silently becoming resistant to clinically important antimicrobials. Notably, our present study highlights for the first time the role of SE-mediated transpositions for the evolution of the MDR gene clusters, tmexCD-toprJ. Our previous study suggested that Aeromonas and Pseudomonas species in the natural environment are not only important reservoirs of ARGs, but are also carriers of evolutionary changes in ARGs, such as blaGES-5-like carbapenemase genes (22). The MGE-mediated spread of ARGs among bacteria and their epidemiology concerning some specific examples, such as mcr1 and ISApl1 (23), and blaNDM and Tn125 (24), have been previously analyzed in detail. The present study provides more direct epidemiological evidence of transpositions of tmexCD-toprJ mediated by SEs, which can be identified by the footprints of SEs. This finding is quite significant for investigating the global spreading of ARGs, including tmexCD-toprJ, and will pave the way for future genomic epidemiological investigations on antimicrobial-resistant bacteria.
Materials and methods
Bacterial isolation and antimicrobial susceptibility testing
Carbapenem- and tigecycline-resistant of environmental isolates of Aeromonas species, A. hydrophila NUITM-VA1 and A. caviae NUITM-VA2 were obtained from the Kim-Nguu River in Hanoi, Vietnam, in March 2021. Environmental water sample was collected and cultured using Luria-Bertani (LB) broth containing 4 mg/L of meropenem at 37°C overnight, and then further selected and isolated using CHROMagar COL-APSE (CHROMagar Microbiology) containing 4 mg/L of tigecycline. Bacterial species identification was performed using MALDI Biotyper (Bruker). Antimicrobial susceptibility testing using Escherichia coli ATCC 25922 as quality control was performed with agar dilution according to the CLSI 2020 guidelines. For tigecycline, AST was additionally performed in the presence or absence of 75 mg/L of the efflux pump inhibitor 1-(1-naphthylmethyl)-piperazine (NMP) as used in the previous studies (7, 11).
Whole-genome sequencing and subsequent bioinformatics analysis
Whole-genome sequencing of NUITM-VA1 and NUITM-VA2 was performed using MiSeq (Illumina) with MiSeq Reagent Kit v2 (300-cycle) and MinION (Oxford Nanopore Technologies; ONT) with the R9.4.1 flow cell. The library for Illumina sequencing (paired-end, insert size of 300–800 bp) was prepared using the Nextera XT DNA Library Prep Kit, and the library for MinION sequencing was prepared using the Rapid Barcoding Kit (SQK-RBK004). ONT reads were basecalled using Guppy v5.0.11 in the super-accuracy mode and then assembled de novo using Canu v2.1.1 (https://github.com/marbl/canu) with the default parameters. The overlap regions in the assembled contigs were detected using LAST (https://gitlab.com/mcfrith/last) and then trimmed manually. Sequencing errors were corrected by Racon v1.4.20 (https://github.com/isovic/racon) twice with the default parameters using ONT reads and then corrected by Pilon v1.20.1 (https://github.com/broadinstitute/pilon) twice with the default parameters using Illumina reads, resulting in their complete circular chromosomes.
Genome annotation and average nucleotide identity analyses were performed using the DFAST server (https://dfast.nig.ac.jp). ARGs were detected using Staramr v0.7.2 (https://github.com/phac-nml/staramr) with the custom ARGs database, including all known tigecycline resistance genes. The circular representation of bacterial chromosomes was visualized using the Proksee server (https://proksee.ca). Linear comparison of sequence alignments of genomic regions containing ARGs and MGEs was performed using BLAST and visualized by Easyfig v2.2.2 (http://mjsull.github.io/Easyfig/).
Nucleotide sequence accession nos
Complete chromosome sequences of A. hydrophila NUITM-VA1 and A. caviae NUITM-VA2 have been deposited at GenBank/EMBL/DDBJ under accession nos. AP025277 and AP025280, respectively.
Funding
This work was supported by grants (JP22gm1610003, JP22fk0108133, JP22fk0108139, JP22fk0108642, JP22wm0225004, JP22wm0225008, JP22wm0225022, JP22wm0325003, JP22wm0325022, and JP22wm0325037 to M. Suzuki; JP22fk0108132 and JP22wm0225008 to I. Kasuga; JP22wm0125006 and JP22wm0225008 to F. Hasebe; JP22fk0108604 and JP22gm1610003 to K. Shibayama) from the Japan Agency for Medical Research and Development (AMED), grants (20K07509 and 21K18742 to M. Suzuki; 19K21984 and 21K18742 to I. Kasuga; 21K18742 to T. Takemura; 21K15440 to A. Hirabayashi) from the Ministry of Education, Culture, Sports, Science and Technology (MEXT), Japan, a grant from Mishima Kaium Memorial Foundation to H. Yano, and a grant (MS.108.02-2017.320 to H. H. Tran) from the National Foundation for Science and Technology Development (NAFOSTED), Vietnam.
Competing interests
None declared.
Ethical approval
Not required.