A novel Tn1696-like composite transposon (Tn6404) harboring bla IMP-4 in a Klebsiella pneumoniae isolate carrying a rare ESBL gene bla SFO-1.

Genetic determinants of a clinical Klebsiella pneumoniae isolate (KP1814) coproducing IMP-4 and a rare ESBL gene SFO-1 was investigated. KP1814 belongs to a novel sequence type (ST) assigned to ST2270. WGS identified four circular DNA sequences in KP1814, including two multidrug-resistance (MDR) plasmids, one virulence plasmid, and one circular form. The MDR plasmid pKP1814-1 (299.9 Kb) is untypeable, and carries two large mosaic multiresistance regions (MRRs). bla SFO-1 and bla IMP-4 co-exists on MRR1, and bla SFO-1 is associated with an IS/Tn-independent genetic context. bla IMP-4 is carried by a novel In804-like integron (intlI-bla IMP-4-Kl.pn.I3-qacG2-aacA4-catB3∆) associated with a novel Tn1696-like transposon (designed Tn6404) flanked by IS5075. The other MDR plasmid pKP1814-3 is a 95,701-bp IncFII plasmid, and is a hybrid of a Shigella flexneri plasmid pSF07201 and an E. coli plasmid pCA08. All resistance genes of pKP1814-3 were detected in a ~16-kb IS26-flanked composite transposon carried by a Tn5396 transposon. The circular form (18.3 Kb) was composed of two parts belonging to pKP1814-1 and pKP1814-3, respectively. The plasmid pKP1814-2, carrying multiple virulence factors, encodes IncFIBK and IncFIIK replicons with a size of 187,349 bp. The coexistence of MDR and virulence plasmids largely enhances the bacterial fitness in the host and environment.

Introduction

SFO-1 is a rare class A ESBL first identified on a self-transferable plasmid in a clinical Enterobacter cloacae isolate from Japan in 1999[1]. By contrast to the known plasmid-borne β-lactamases which are produced constitutively, SFO-1 is induced by imipenem, and can hydrolyze most β-lactams except cephamycins and carbapenems[2]. The bla SFO-1 gene was later detected in an outbreak clone of multidrug-resistance (MDR) E. cloacae in Spain[3], and in three Klebsiella pneumoniae and one Escherichia coli clinical isolates in China[4,5]. However, the complete structure of bla SFO-1-harboring plasmids is still unknown.

IMP-4 was first detected in Hong Kong in 2001[6], and later widely disseminated in Australia and the mainland of China[7]. The spread of IMP-4 is frequently associated with class 1 integrons carried by plasmids[8,9]. In China, plasmid-borne bla IMP-4 has been sporadically reported in different provinces/cities (Chongqing, Hubei, Fujian, Zhejiang, Guangdong, Shanghai and Tianjin[10-16]. However, few studies have reported the full structure of bla IMP-4-carrying plasmids detected in China[16-18], largely limits our understanding on the dissemination mechanism of bla IMP-4. In this study, we report a carbapenem-resistant K. pneumoniae isolate that simultaneously carries bla IMP-4 and bla SFO-1 genes. The plasmid content and genetic determinants of the isolate were comprehensively analyzed to gain a better understanding of the spread mechanism of bla IMP-4 and bla SFO-1 genes in China.

Materials and Methods

Strain collection

The community-associated K. pneumoniae strain KP1814 was collected in the frame of a national-wide survey for antibacterial resistance among outpatients with community-associated infections[19]. This strain was isolated from an outpatient (the type of clinical specimen was unknown) in a secondary hospital in mid-south of China (Hubei province) in 2011.

Antimicrobial Susceptibility Testing

Antimicrobial susceptibility was determined using Vitek-2 with card GN-13 (Biomerieux, Marcyl’Etoile, France) and was interpreted according to the clinical breakpoints defined by EUCAST-v7.0 (http://www.eucast.org/clinical_breakpoints/). ESBL phenotype was confirmed by the double-disc synergy test (DDST) using cefotaxime (30 μg) and ceftazidime (30 μg) with clavulanate (10 μg) discs.

Plasmid analysis

Azide-resistant E. coli J53 was used as the recipient for the conjugative transfer of plasmids. Transconjugants were selected on MH media with sodium azide (100 µg/µl) and gentamicin (10 µg/µl) or ceftazidime (8 µg/µl). The selected transconjugants were later confirmed by PCR and susceptibility testing. The plasmid size was estimated by S1 nuclease pulsed-field gel electrophoresis (PFGE).

Whole-genome sequencing (WGS) and data analysis

Genomic DNA was extracted using Gentra Puregene Yeast/Bacteria Kit (Qiagen, Hilden, Germany). The genome was sequenced by Illumina Hiseq2500 (Illumina, San Diego, CA, USA) using a 2 × 125-bp pair-end libraries, and was further scaffolded by Pacbio RS II platform (Pacific Biosciences, California, US) using a 10-Kb library. De novo assembly for reads yielded by Hiseq2500 was done by CLC Genomics Workbench v8.0 (QIAGEN, Hilden, Germany) after quality trimming (Qs ≥ 20), and the scaffolding was performed by SSPACE standard version 3.0 with default settings[20]. Further gaps within scaffolds were closed using GapFiller with default settings[21]. Annotation was performed using the RAST server (rast.nmpdr.org) followed by manual curation using BlastP and ISfinder (https://www-is.biotoul.fr); multilocus sequence typing (MLST), resistome analysis and plasmid typing was done by uploading sequences to the CGE server (https://cge.cbs.dtu.dk).

Nucleotide Sequence Genbank Accession Numbers

The three plasmids and the circular form have been deposited at DDBJ/EMBL/GenBank under the accession of KX839207- KX839210.

Results and Discussion

The characterization of KP1814

KP1814 was resistant to numerous drugs, including ertapenem, and remained susceptible to amikacin, ciprofloxacin, and levofloxacin (Table S1). KP1814 showed an ESBL-negative phenotype by double-disc synergy test. However, WGS detected a rare ESBL gene bla SFO-1 which was overlooked using routine PCRs[19]. A carbapenemase gene bla IMP-4 and 23 acquired drug-resistant genes were further detected, including aac(3′)-IId (2 copies), aacA4, strA, strB, aac(6′)-Ib-cr (2 copies), and aadA5 (2 copies) for aminoglycoside resistance; bla TEM-1b and bla OXA-1 for β-lactam resistance; mph(A) (3 copies) for macrolide; catB3 and catB3∆ for phenicol; arr-3 for rifampicin; sul1 (3 copies) for sulphonamide; dfrA17 (2 copies) for trimethoprim. This genotype can fully explain the resistance profile observed. KP1814 additionally carries bla SHV-11 and fosA on the chromosome. KP1814 was assigned to the novel sequence type (ST) 2270 (2-3-1-20-61-4-133). S1 PFGE showed three clear bands with sizes approximately 300 kb, 180 kb, and 90 kb (data not shown).

The genome of KP1814 was completed to dissect its plasmid content. This resulted in one complete chromosome, 3 complete circular plasmids (pKP1814-1, pKP1814-2, and pKP1814-3) and one circular form with size of 299.9Kb, 187.3Kb, 95.6Kb, and 18.2Kb, respectively. This is in concordance with the PFGE results except the circular form that was not detected on the gel.

The characterization of pKP1814-1

The bla IMP-4 and bla SFO-1 genes co-locate on pKP1814-1. To the best of our knowledge, this is the first report on the co-existence of bla SFO-1 and bla IMP-4. pKP1814-1 was able to be transferred to the azide-resistant E. coli J53 via conjugation, and the conjugation frequency was estimated of ~5 × 10−5 per donor cell.

pKP1814-1 is a 299,858-bp plasmid (Fig. 1), and is untypeable by using PlasmidFinder. This is different from previously reported bla IMP-4-encoding plasmids that are frequently associated with incompatibility groups IncL/M, IncA/C and IncHI2 identified in Australia[22], and IncN in China[16-18]. pKP1814-1 shares a high homology with an IncA/C-IncH plasmid (the replicon type is IncU reanalyzed by PlasmidFinder in this study) pKOX-R1 (CP003684; 77% coverage, >98% nucleotide identity)[23], and an IncH plasmid pRpNDM1-1 (JX515588; 66% coverage, 99% nucleotide identity). The shared backbone between pKP1814-1 and pKOX-R1 is highly conserved with 82-bp difference (>99.9% nucleotide identity), including the replicon gene repA, umuC/D (for DNA repair), higB (encoding toxin), parA/B (encoding partitioning proteins), relB/stbD (for plasmid stability), conjugative transfer gene clusters dispersed in different locations (traIDJLEKBVCWUNX, traCBP, and traFHG) and other genes encoding function/hypothetical proteins (Fig. 1). Notably, pKOX-R1 and pRpNDM1-1 were identified in Taiwan and mainland China, respectively, suggesting a circulation of such plasmids in that geographic region. However, various replicon types carried by these plasmids highly challenges the epidemiological investigations.

Figure 1

The genetic map of pKP1814-1. A circular representation of pKP1814-1 was generated by using DNA plotter. From the inside outward, the first, second, and third circle show GC skew, GC content, and the regions with over 50% nucleotide identity to pKOX-R1(CP003684), respectively. The fourth and fifth circle show the genetic features in counterclockwise and clockwise directions, respectively. Genes are classified by different colours as shown in the legend, and the gene names of plasmid backbone are shown. Two MRRs are marked by gray arches.

The multi-resistance regions (MRRs) of pKP1814-1

All drug-resistant genes of pKP1814-1 were detected on two MRRs (Fig. 1). MRR1 is bracketed by an interrupted IRtnp of Tn21 (22 bp) and an 81-bp IRstr, and is composed of three subregions (Fig. 2A). The first subregion is a novel Tn1696-like mercury and multidrug resistance transposon (designed Tn6404 by Tn Number Registry). The two IRs of Tn6404 (IRtnp and IRmer) are interrupted by the insertion of IS5075 into two pieces (16 bp + 22 bp), respectively. This is consistent with that IS5075 targets a specific position in the terminal repeats of Tn21 family[24]. A similar Tn1696-like transposon is found in an IMP-4-encoding IncA/C2 plasmid pIMP-PH114 (KF250428) detected from a K. pneumoniae strain that was recovered from a patient who was hospitalized in the Philippines[25] (Fig. 2B).

Figure 2

Features of the two MRRs of pKP1814-1. (A) The structure of MRR1. Three subregions of MRR1 are highlighted in different colours. Black tall bars represent the IRs of Tns. The 38-bp IR (IRtnp) of Tn1696 is interrupted into two parts by the insertion of IS5075. Shorter red bars represent the IRs of IS/In. The group II intron Kl.pn.I3 is shown as a black rectangle. The genes of mer operon are labeled with their letter, and the other genes are marked in black; (B) Comparison of Tn6404, Tn1696 and Tn1696-like. The Tn1696-like transposon is detected on pIMP-PH114 (96158..123870); (C) The structure of MRR2. Two subregions of MRR2 are highlighted in different colours. The IR name of the subregion 1 is derived from pEl1573 (JX101693).

Tn6404 encodes a bla IMP-4-harboring In809-like integron, replacing the In4 detected in Tn1696. Notably, In809 with four cassettes (bla IMP-4-qacG2-aacA4-catB3) is widely disseminated in Acinetobacter spp and Enterobacteriaceae isolates in Asia-pacific region, e.g. Hong Kong, Singapore, Australia, and Philippines[25]. Here, the In809-like integron carries a novel cassette array (bla IMP-4-Kl.pn.I3-qacG2-aacA4-catB3∆), in which a group II intron Kl.pn.I3 is inserted into the attC site of the bla IMP-4 cassette (Fig. 2A). To date, various genetic context of bla IMP-4 has been identified, e.g. bla IMP-4 -qacG-aacA4-aphA15, bla IMP-4-Kl.pn.I3-mobC and bla IMP-4-Kl.pn.I3-qacEΔ-sul1 [7-10,16-18,22,26]. Of note, the structure bla IMP-4-Kl.pn.I3 seems unique in isolates from China revealed by blasting in GenBank, thus could be as an epidemiological marker for bla IMP-4 detected in China.

The 5′-conserved segment (CS) of the In809-like integron is identical to that of In4 carried by Tn1696 [27], encoding a complete intI1 gene bounded by a 25-bp IRi. However, the reported bla IMP-4-Kl.pn.I3-harboring integrons carry a truncated intI1 gene, mostly due to IS26 insertions[18]. The 25-bp IRi of the In809-like integron is located between the resII and resIII sites of the tnpR-Tn1696, in the same configuration previously described in In416 (AJ704863) detected on a VIM-4- and CMY-4-encoding plasmid pCC416 in Samonella [28]. The In809-like integron lacks a typical sul1-associated 3′-CS. Instead, a Tn5563a-like transposon locates at the 3′ end of the In809-like integron, and the IRR is inserted into catB3 resulting in a truncated gene. IRL of the Tn5563a-like transposon is missing caused by that the tnpR of Tn5563a is replaced by a tniR probably via recombination.

The mer operon (merRTPFADE) of Tn6404 is highly homologous with that of Tn5053 (L40585) (>99.9% nucleotide identity with 4-nuclotide difference), but is different with that of Tn1696 (~84–87% nucleotide identity; a replacement of merC by merF) (Fig. 2B), suggesting that the mer operons of Tn6404 and Tn1696 have independent origin. Notably, the Tn5563a-like transposon together with the mer operon consists of a 9.4-kb module, which is almost identical to that found on pJR-1 (CP005961) from P. mandelii (2-nucleotide difference) and on pUM505 (HM560971) from P. aeruginosa (3-nucleotide difference). This indicates that the 9.4-kb module may originate from Pseudomonas spp. Taken together, our findings support that the ubiquity and variety of elements in the Tn6404 is the result of frequent recombination processes.

The second subregion of MMR1 carries a module encoding a rare ESBL gene bla SFO-1, which locates adjacent to its regulator ampR as previously reported (Fig. 2A). The dissemination of ampR-bla SFO-1 is previously suggested to be mediated via IS26-composite transposon in K. pneumoniae and E. cloacae [3,4]. However, IS26 and/or other IS/Tn sequences were not detected here, and the genetic context of ampR-bla SFO-1 is identical to the available sequenced part (9568 bp) of an IncA/C plasmid pHD0149-2 (KJ721805) detected from an E. coli strain in China, suggesting a second dissemination mechanism harnessed by bla SFO-1.

The third subregion is a Tn5393c encoding strA-strB bracketed by two 81-bp IRs, and locates at downstream of the bla SFO-1-encoding module (Fig. 2A). The Tn5393c might originate from Aeromonas salmonicida, since it shared a high similarity with that found in pRAS2 (CP005961) from A. salmonicida (8-nucleotide difference).

MRR2 is flanked by a disrupted IRchrA (22 bp) and a 19-bp IR of ISCR1 (Fig. 2C), and is consisted of two subregions. The first subregion (IS5075-chrA-IS6100-mphR(A)-mrx-mph(A)-IS26-aac(3′)-IId-ISCfr1-bla TEM-1b) is bracketed by a 12-bp IRR of IS5075 and a 38-bp IRTEM, almost identical (3-nucleotide difference) to that detected in a Tn2-derived MRR of a bla IMP-4-harboring IncL/M plasmid pEl1573 (JX101693) from Australia[29]. It is suggested that such module can disseminate among different Inc groups via homologous recombination[25]. The second subregion of MRR2 carries an incomplete Tn1722 transposon only encoding a truncated tnpA and a tnpR. An ISCR1-containing class 1 integron In37::ISCR1 (intI1-aac(6′)Ib-cr-bla OXA-1-catB3-arr-3-qacE∆1-sul1-ISCR1) is inserted into the res site of Tn1722. Such insertion event has also been identified in an IncR plasmid pKPS30 (KF793937) and a bla VIM-13-encoding transposon (GQ396666), suggesting acquisitions of class 1 integron at a hotspot located within the res site of Tn1722 [30]. The Tn1722-derived MRR subregion is supposed to be obtained from pKPS30 via multiple recombination events, since they are only differed by 1 SNP and a tnpM deletion (Fig. 2C).

The IS26-flanked composite transposon of pKP1814-3

The other MDR plasmid pKP1814-3 is a 95,701-bp IncFII plasmid, and was able to be co-transferred with pKP1814-1 to the azide-resistant E. coli J53. This plasmid is a hybrid of a Shigella flexneri plasmid pSF07201 (KJ201887) and an E. coli plasmid pCA08 (CP009233) (Fig. S1). All resistance genes of pKP1814-3 were detected in a 16-kb IS26-flanked composite transposon (IS26-∆intI1-dfrA17-aadA5-qacE∆1-sul1-chrA-padR-IS6100-mphR(A)-mrx-mph(A)-IS26) carrying a class 1 integron In54, of which the 5′-CS was disrupted by the insertion of IS26 resulting in a truncated intI1 (Fig. 3). Such IS26-flanked transposons have been widely disseminated in plasmids, e.g. pEC958 (HG941719), pCA08 (CP009233), and pKF3-140 (FJ876827). They are in different combinations and arrangement highlighting the role of this module in the mosaic characteristics of MRRs. It is known that IS26-flanked transposon could switch to be a circular form as a translocatable unit via a replicative mechanism of IS26, and widely disseminate by incorporating at a new location either via replicative transposition, homologous recombination or a Tnp26-catalyzed conservative reaction[31,32]. The IS26-flanked composite transposon was further inserted into a Tn5396 transposon, thus is flanked upstream by a truncated tnpA-Tn5396 and downstream by a truncated tnpR-Tn5396, which was nearly identical to that found in pECO-824 (CP009860) with only 1-nucleotide difference. The Tn5396 composite transposon was further inserted into an IS1-like resulting in a truncated insB (Fig. 3), indicating multiple events involved in the acquisition of these IS/Tns here.

Figure 3

Features of the MRR of pKP1814-3. The components of the MRR are shown separately according to the obtaining order proposed. The IRR of IS1 is missing due to the insertion of Tn5396. The red bars represent the IRs of IS/Tn/In. The genes are in different colours according to the classification described in Fig. 1.

The circular form of KP1814

KP1814 harbored an 18,238-bp circular form. Of note, ~58.6% of the circular form was almost identical to a region (from chrA to orf39) of pKP1814-1, and ~79% was nearly identical to the IS26-flanked transposon of pKP1814-3 (Fig. S2). The circular form shares a 6.78-kb module (chrA-padR-IS6100-mphR(A)-mrx-mphA-IS26) with pKP1814-1 and pKP1814-3. Intriguingly, a recent study detected various Tn1548-like circular forms in Acinetobacter baumannii [33]. Taken together, this suggests an emerging role of circular forms in the dissemination of antibiotic resistance genes.

The characterization of the virulence plasmid pKP1814-2

pKP1814-2 is a 187,349-bp circular plasmid encoding IncFIBK and IncFIIK replicons (Fig. S3). The plasmid backbone of pKP1814-2 has a high synteny (43–46% coverage) to pKP1-19 (CP012884) and pKPN-d90 (CP015132), mainly comprising genes encoding replication/maintenance/modification (repA, parA/B, umuC/D, relB/E, samA/B, ssb, psiA/B, higA-like), F-like type IV conjugative transfer (traMJYALEKBVCFQHGTDIX) and other genes encoding function/hypothetical proteins (Fig. S3). The shared regions of the three plasmids were highly conserved with ~98.3% nucleotide identity (about 1500-bp difference). Numerous virulence-associated genes and one heavy-metal associated resistance gene were detected on pKP1814-2. This plasmid carried a gene encoding klebicin B, a bacteriocin involved in competitive exclusion of other bacteria to form nutritionally restricted niches[34]. Of note, an mrk gene cluster (mrkABCDF) encoding type 3 fimbria was detected. This cluster is commonly found on the chromosome of K. pneumoniae, and was occasionally identified on a few plasmids in Enterobacteriaceae. Such plasmid-borne mrk cluster is suggested to profoundly enhance the ability of biofilm formation and increase plasmid conjugation efficiency[35,36]. pKP1814-2 harbored a gene cluster glgB-glgC-glgA-glgP-pgm involved in the glycogen metabolic pathway, which has been linked to environmental survival, symbiotic performance and colonization, and virulence[37]. This cluster may originate from Klebsiella spp. as the two best matches with 100% coverage were found on pCAV1374-228 (CP011634) of a K. oxytoca strain, and on p_IncFIB_DHQP1002001 (CP016810) of a K. pneumoniae strain.

Additionally, a thermoresistance cluster, encoding heat-shock proteins ClpC, Hsp20, and ∆FtsH, was identified on pKP1814-2. This cluster could be exchanged across species as it was also detected on two E. cloacae plasmids: pENT-22e (CP009855) and pENT-4bd (CP008907). An intact ftsH gene was further found out of the cluster. These heat-shock proteins are known to provide protection against stressful conditions and to contribute to the bacterial pathogenesis[38-40]. The thermoresistance cluster additionally carried a kefB gene encoding a potassium/proton antiporter, which is known to play a role in protecting E. coli from electrophile toxicity[41]. It is assumed that those genes and gene clusters coding for carbon-source metabolism, stress response, fimbria clusters, efflux pump, and bacteriocin maybe involved in virulence properties of the host as well as in increased survival and fitness under the hospital environment, overall contributing to the successful dissemination and maintenance of these plasmids in hosts.

In conclusion, we report the complete structure of a novel plasmid carrying bla SFO-1 and bla IMP-4. Our study, along with others, raises the concern that co-existence of MDR and virulence plasmids enables to increase the fitness and viability of hosts largely challenging the clinical treatments and outcomes.

Supplementary information