Genetic characterization of three qnrS1-harbouring multidrug-resistance plasmids and qnrS1-containing transposons circulating in Ho Chi Minh City, Vietnam

Plasmid-mediated quinolone resistance (PMQR) refers to a family of closely related genes that confer decreased susceptibility to fluoroquinolones. PMQR genes are generally associated with integrons and/or plasmids that carry additional antimicrobial resistance genes active against a range of antimicrobials. In Ho Chi Minh City (HCMC), Vietnam, we have previously shown a high frequency of PMQR genes within commensal Enterobacteriaceae. However, there are limited available sequence data detailing the genetic context in which the PMQR genes reside, and a lack of understanding of how these genes spread across the Enterobacteriaceae. Here, we aimed to determine the genetic background facilitating the spread and maintenance of qnrS1, the dominant PMQR gene circulating in HCMC. We sequenced three qnrS1-carrying plasmids in their entirety to understand the genetic context of these qnrS1-embedded plasmids and also the association of qnrS1-mediated quinolone resistance with other antimicrobial resistance phenotypes. Annotation of the three qnrS1-containing plasmids revealed a qnrS1-containing transposon with a closely related structure. We screened 112 qnrS1-positive commensal Enterobacteriaceae isolated in the community and in a hospital in HCMC to detect the common transposon structure. We found the same transposon structure to be present in 71.4 % (45/63) of qnrS1-positive hospital isolates and in 36.7 % (18/49) of qnrS1-positive isolates from the community. The resulting sequence analysis of the qnrS1 environment suggested that qnrS1 genes are widely distributed and are mobilized on elements with a common genetic background. Our data add additional insight into mechanisms that facilitate resistance to multiple antimicrobials in Gram-negative bacteria in Vietnam.

Plasmid-mediated quinolone resistance (PMQR) refers to a family of closely related genes that confer decreased susceptibility to fluoroquinolones. PMQR genes are generally associated with integrons and/or plasmids that carry additional antimicrobial resistance genes active against a range of antimicrobials. In Ho Chi Minh City (HCMC), Vietnam, we have previously shown a high frequency of PMQR genes within commensal Enterobacteriaceae. However, there are limited available sequence data detailing the genetic context in which the PMQR genes reside, and a lack of understanding of how these genes spread across the Enterobacteriaceae. Here, we aimed to determine the genetic background facilitating the spread and maintenance of qnrS1, the dominant PMQR gene circulating in HCMC. We sequenced three qnrS1-carrying plasmids in their entirety to understand the genetic context of these qnrS1-embedded plasmids and also the association of qnrS1-mediated quinolone resistance with other antimicrobial resistance phenotypes. Annotation of the three qnrS1-containing plasmids revealed a qnrS1-containing transposon with a closely related structure. We screened 112 qnrS1-positive commensal Enterobacteriaceae isolated in the community and in a hospital in HCMC to detect the common transposon structure. We found the same transposon structure to be present in 71.4 % (45/63) of qnrS1-positive hospital isolates and in 36.7 % (18/49) of qnrS1-positive isolates from the community. The resulting sequence analysis of the qnrS1 environment suggested that qnrS1 genes are widely distributed and are mobilized on elements with a common genetic background. Our data add additional insight into mechanisms that facilitate resistance to multiple antimicrobials in Gram-negative bacteria in Vietnam.

INTRODUCTION
Fluoroquinolones are among the current first line of drugs in Vietnam for treating infections ranging from pneumonia to diarrhoea to bacteraemia. Since the mid-1990s, resistance to quinolones has been increasing steadily in multiple organisms causing respiratory tract infections, diarrhoea and bacteraemia (Holt et al., 2013;Nga et al., 2012;Nhu et al., 2014). The emergence of fluoroquinolone resistance in these pathogens and other bacteria represents a clear threat to the effective treatment of common bacterial infections. Resistance to quinolones occurs commonly via mutations in the gene of the target enzyme, DNA gyrase, but can also be plasmid mediated. Plasmid-mediated quinolone resistance (PMQR), in the form of a qnr gene, was first described in 1998 in a Klebsiella pneumoniae isolate from a patient with urinary tract infection in North America (Martínez-Martínez et al., 1998). Since this first report, multiple studies have described a range of PMQR determinants found within the Enterobacteriaceae and other bacterial families (Strahilevitz et al., 2009). To date, five qnr genes have been described and are distinguished on the basis of their sequence homology: qnrA, qnrB and qnrS1, and more recently qnrC and qnrD (Cavaco et al., 2009;Hata et al., 2005;Jacoby et al., 2006;Martínez-Martínez et al., 1998;Wang et al., 2009). The qnr genes appear to be highly promiscuous, having the capacity to become rapidly disseminated among related and unrelated hosts. The transmissibility of the qnr genes makes the genomic mechanisms facilitating their movement of considerable interest and of relevance in the community and in healthcare settings. Of the five qnr genes identified, only the genetic contexts of qnrA and qnrB have been comprehensively described. These genes are commonly located within complex sul1-type class 1 integrons (Garnier et al., 2006;Poirel et al., 2006). The context of qnrS is less well described, but there are reports of the gene being located within a gene cluster flanked by IS26 transposases (Chen et al., 2006;Hu et al., 2008). However, it is currently unknown if the qnrS gene is always associated with IS26 or if qnrS can be transferred and/or maintained by other, unrelated, mobile elements.
Several qnrS1-carrying plasmids have been described in the literature and have publicly available nucleotide sequences. These plasmids range in size and belong to various incompatibility groups including IncN (Dobiasova et al., 2013;Literak et al., 2010), IncI1 (Dobiasova et al., 2013), IncX1 (Dobiasova et al., 2013;Literak et al., 2010) and IncX2 (Literak et al., 2010;Sumrall et al., 2014). These qnrS1encoding plasmids have been identified in Asia and Europe and in a range of Gram-negative bacteria including K. pneumoniae, Escherichia coli and Enterobacter aerogenes (Dobiasova et al., 2013;Hu et al., 2008;Park et al., 2009;Sumrall et al., 2014). Two of the earliest and best-described qnrS1-encoding plasmids are pTPqnrS-1a and pK245. Plasmid pTPqnrS-1a is a 10 kb replicon, and was isolated from a multidrug-resistant (MDR) Salmonella Typhimurium DT193 in the UK (Kehrenberg et al., 2007). The second, pK245, was characterized in a clinical isolate of K. pneumoniae originating in Taiwan (Chen et al., 2006). In contrast to pTPqnrS-1a, pK245 is a large MDR plasmid of approximately 100 kb. The MDR phenotype of pK245 was demonstrated by transferring this plasmid into an antimicrobial-susceptible Escherichia coli strain by electrotransformation (Chen et al., 2006). The pK245-positive transformant showed an increase in MICs to multiple classes of antimicrobials, including aminoglycosides, b-lactams and (fluoro)quinolones and had an extendedspectrum b-lactamase (ESBL) phenotype (Chen et al., 2006). Comparative sequence analysis of available qnrS1 plasmids revealed that the genetic architecture surrounding the qnrS1 gene is identical between pTPqnrS-1a and pK245, and they additionally sharing a high sequence identity with the qnrS1 genetic region in other partial plasmid sequences, including pAH0376 from Shigella flexneri (Hata et al., 2005) and pINF5 from Salmonella Infantis (Kehrenberg et al., 2006).
In a study investigating the distribution of PMQR determinants in Enterobacteriaceae isolated from hospitalized patients and healthy volunteers from Ho Chi Minh City (HCMC), Vietnam, we found an exceptionally high prevalence of the qnrS1 genes in both hospital (63/139, 45 %) and community (49/413, 12 %) bacterial isolates (Vien et al., 2009). We therefore hypothesized that qnrS1 was embedded on a highly mobile and conserved genetic element, which was contributing to the spread and the apparent success of qnrS1 across the Enterobacteriaceae in this setting. In this current study, we aimed to characterize the dominant qnrS1-containing elements circulating in Enterobacteriaceae isolated from hospital patients and community volunteers in HCMC to understand if qnrS1 is being disseminated on one or more elements by defining their genetic context. To achieve this, we selected three qnrS1-containing plasmids, broadly representative of those found to be circulating in the hospital and community environments (Vien et al., 2012), for DNA sequencing and analysis in their entirety. The resulting sequence analysis of the qnrS1-containing mobile elements has broadened our knowledge of the genetic architecture surrounding the qnrS1 gene and added insight into MDR mechanisms that are circulating within these differing bacterial environments in Vietnam.

METHODS
Bacterial strains. A total of 115 qnrS1-positive Enterobacteriaceae strains (38 Escherichia coli, 69 K. pneumoniae and eight from other Enterobacteriaceae species) were selected for analysis in this study. All of the strains have been described previously and were isolated from patients admitted to the tetanus ward of the Hospital for Tropical Diseases (HTD) in HCMC, Vietnam, between May and October 2004 and between June and November 2005 or from healthy volunteers participating in a typhoid vaccine study in 2005(Tran et al., 2010. The presence of the qnrS1 gene in these strains has been previously confirmed and described (Vien et al., 2009).
Three qnrS1-containing plasmids identified previously (Vien et al., 2012) were selected to be broadly representative of qnrS1-containing plasmids circulating in our setting (i.e. harbouring strain, hospital/ community infections and size) and were sequenced in their entirety, assembled and annotated gene by gene in comparison to sequences available in public databases. These plasmids (selected out of the 115 qnrS1-positive Enterobacteriaceae strains) were pE66An (Escherichia coli host), pK18An (K. pneumoniae host) and pK1HV (K. pneumoniae host). A summary of the basic features of these plasmids is given in Table 1.

V. Le and others 870
Journal of Medical Microbiology 64 Plasmid sequencing and annotation. The three qnrS1-containing plasmids were sequenced at the Wellcome Trust Sanger Institute in the UK using conventional Sanger sequencing methods as described previously (Parkhill et al., 2001). The plasmid sequences were annotated and analysed using Artemis (Rutherford et al., 2000) and aligned and compared using the Artemis Comparison Tool (Carver et al., 2008). Plasmid circularization and graphical representations were performed using DNA Plotter software (Carver et al., 2009).
PCR amplification for the qnrS1-containing transposon and RFLP typing. Total genomic DNA from 112 qnrS1 PCR amplificationpositive Enterobacteriaceae isolates was extracted using a Wizard Genomic DNA purification kit (Promega), according to the manufacturer's specifications. PCR amplification for the qnrS1-containing transposon from the extracted genomic DNA was performed under the following condition: 94 uC for 10 s, 55 uC for 30 s and 68 uC for 6 min for 35 cycles. Amplification was performed using the Expand Long Template PCR System (Roche) using the primers Trans-qnrS-F (59-CAGGAAGAGGCATTGTCAAAGG-39) and Trans-qnrS-R (59-GGTGCTTGTCAGCGTAAA-39). These primers were designed using Primer Express 5 software (Applied Biosystems, Life Technologies) and their specificity was assessed in silico using BLASTN (http://blast.ncbi. nlm.nih.gov/Blast.cgi). The resulting PCR amplicons were examined by electrophoresis and UV visualization on 2 % agarose gels containing 2 % ethidium bromide. The PCR amplicons containing the qnrS1 gene were typed using RFLP with three different enzymes: Eco RV, HindIII and PvuII (New England Biolabs). The restriction-digested PCR amplicons were analysed by gel electrophoresis for 2 h on a 1 % agarose gel, stained with 2 % ethidium bromide and examined under UV light. The restriction fragments were sized and compared for group typing using Bionumerics software (Applied Maths).
Primer-walking sequencing. The qnrS1-containing transposon from the K34N strain was sequenced using primer walking. The sequencing reaction was performed in a 20 ml reaction containing 4 ml Big Dye Terminator, 2 ml buffer, 20 ng genomic DNA and distilled water up to 20 ml. Each fragment was repeated four times using an ABI 3130XL machine (Applied Biosystems, Life Technologies). All sequences were assembled using Vector NTI software (Life Technologies).
Electrotransformation. PCR-negative isolates for the qnrS1containing transposon were analysed for the presence of subfamilies of the known transposon. Plasmid DNA from these negative isolates was extracted using a Qiagen Midi Prep Plasmid DNA Extraction kit, as per the manufacturer's recommendations. Escherichia coli TOP10 cells (Invitrogen, Life Technologies) were transformed with isolated plasmid DNA by a Bio-Rad gene pulser, using conditions recommended by the manufacturer (Invitrogen, Life Technologies).
Transformants were selected on Luria-Bertani medium supplemented with 0.03 mg ciprofloxacin l 21 . Plasmid DNA from these transformants was extracted by the method of Kado & Liu (1981), examined on an agarose gel for the presence of only one plasmid and then subjected to PCR amplification for the qnrS1 gene to ensure transformation of the appropriate plasmid.
Southern blot analysis. Isolates containing a subfamily of known qnrS1-containing transposons were detected by Southern blotting with two different probes: qnrS1 and bla LAP-2 , using the primers described for amplification of the qnrS1 region. Plasmid DNA from the transformants was extracted using a Qiagen Midi Prep Plasmid DNA Extraction kit, as per manufacturer's recommendations. These plasmids were then digested with EcoRI and duplicates were run on a gel. The gel was subsequently transferred to a membrane. The membrane was cut into two pieces and each was hybridized with one of the probes, qnrS1 or bla LAP-2 . If an isolate had signal with both probes binding to the same fragment, it was assigned as carrying a subfamily of a qnrS1-containing transposons.

RESULTS AND DISCUSSION
Global comparison of the three qnrS1-containing plasmids Figure 1 shows a global DNA alignment of the three sequenced plasmids and a previously sequenced qnrS1containing plasmid (pK245) identified in a K. pneumoniae isolate from Taiwan as a comparator (Chen et al., 2006). These alignments showed that the two plasmids isolated independently from different bacterial genera within the hospital environment (pE66An and pK18An) exhibited substantial gene synteny with each other, but generally shared a lower degree of DNA homology with the plasmid identified in a community isolate (pK1HV).

Plasmid pE66An
The qnrS1-encoding plasmid pE66An was extracted from an Escherichia coli strain isolated from a rectal swab taken from a patient admitted to the tetanus ward of the HTD in HCMC. Plasmid pE66An is 80 105 bp with an approximately neutral G+C content of 51.52 mol% (Fig. 2). After complete annotation of the plasmid sequence, 109 predicted coding sequences (CDSs) were identified; the protein products of seven of these CDSs were predicted to be associated with resistance to a variety of antimicrobial classes. These antimicrobial resistance genes were: aacC3 (gentamicin), sulII (sulfonamides), tetR and tetA (tetracyclines), qnrS1 (quinolones), bla LAP-2 (b-lactams) and bla CTX-M-14 (third-generation cephalosporins). An association between the qnrA gene and ESBL-encoding genes has been reported previously (Castanheira et al., 2007;Hamouda et al., 2008;Lavigne et al., 2006). The bla CTX-M-14 gene within pE66An was adjacent to the element ISEcp1. The? insertion element ISEcp1 has been shown previously to mediate the transfer of bla CTX-M-14 (Bou et al., 2002). As the bla CTX-M-14 gene is in association with this ISEcp1 insertion element, the potential for dissemination of this gene to other plasmids or transposable elements is likely to be enhanced.
Comparative analysis showed that two regions within plasmid pE66An, a 16.7 and a 10.2 kb region, exhibited significant DNA sequence similarity to regions within the previously described Klebsiella oxytoca plasmid pKOX105 (Carattoli et al., 2010) and three smaller regions (5.9, 5.2 and 3.5 kb) within plasmids pIP843 (Cao et al., 2002), pRAx (Fricke et al., 2009) and pKF3-94 (Zhao et al., 2010) from HCMC, Madagascar and China, respectively (Fig. 2a). Plasmid pKOX105 was isolated from a K. oxytoca isolate present in the intestinal microbiota of an individual in a long-term care facility in Bolzano, Italy, in 2005. Both pE66An and pKOX105 are IncN plasmids, with each containing the highly conserved IncN plasmid backbone (Carattoli et al., 2010). Plasmid pE66An contains two regions of 10 280 and 5183 bp that encode sequences predicted to be responsible for the conjugal transfer. Indeed, our previous work has shown that pE66An has the capacity to be efficiently conjugated at high frequency into a suitable recipient strain. We therefore concluded that this conjugation system is functionally active (Vien et al., 2009).
In addition to antimicrobial resistance and conjugal transfer functions, plasmid pE66An also encodes genes that suggest that it may be able to adapt to a variety of hosts and environments. For example, we identified a CDS with 99.5 % identity to frmA, a class III alcohol dehydrogenase identified in a Pasteurella piscicida isolate (Kim & Aoki, 1994), and 98.1 % identity to a class III alcohol dehydrogenase identified in an Escherichia coli isolate (Hochhut et al., 2006). The function of the protein product encoded by frmA is involved in resistance to formaldehyde and other aldehydes. We predict that frmA provides a selective advantage for bacterial hosts in hospital environments where disinfectants containing aldehydes are used (Chen et al., 2006).

Plasmid pK18An
Plasmid pK18An was carried by a K. pneumoniae isolate taken from a rectal swab of a tetanus patient in the HTD Fragments with substantial DNA homology to other sequenced plasmids and the qnrS-encoding region are highlighted. qnrS1 plasmids in Enterobacteriaceae in HCMC. The sequence of plasmid pK18An showed it was a circular replicon of 51 160 bp, with a G+C content of 51.32 mol% (Fig. 2). The annotation of pK18An identified 72 CDSs, the functions of six of which were predicted to be associated with resistance to antimicrobials, including aacC3 (gentamicin), sulII (sulfonamides), strA and strB (streptomycins), dhfrV (trimethoprim) and qnrS1 (quinolones). Plasmid pK18An was also found to harbour a bla LAP-2 gene in close proximity to qnrS1, but that carried an IS5 element insertion and so was likely to have been inactivated. The sulII-strA-strB genes were located in close proximity to each other and adjacent to an IS26 element. The strA-strB genes are often linked with the sulII sulfonamide-resistance gene, commonly encoded on broad-host-range non-conjugative plasmids in a range of Gram-negative bacteria found in humans and animals. The usage of streptomycin in clinical and animal medicine has diminished dramatically over the last 10-20 years, yet the persistence of sulII-strA-strB implies that factors other than a direct selection pressure from the antimicrobial are important for the maintenance of these genes (Sundin & Bender, 1996).
Like pE66An, plasmid pK18An shared two large regions, of 10.2 and 9.4 kb, with extensive DNA homology to plasmid pKOX105 (Carattoli et al., 2010) (Fig. 2b). Similarly, plasmid pK18An also contained two regions containing CDSs that are predicted to be responsible for the conjugal transfer, but these operons were disrupted by numerous IS26 elements. These sequence data probably explain why it was not possible to conjugate pK18An into a recipient Escherichia coli under laboratory conditions (Vien et al., 2009). Whilst pK18An was not conjugative, the plasmid sequence was littered with multiple IS elements, particularly surrounding antimicrobial resistance genes, suggesting that such elements may facilitate the independent transfer of these genes to other plasmids.
Plasmids pK18An and pE66An both contained restriction modification systems; pK18An contained an ecoRIIM gene and pE66An contained the ecoRIIR and ecoRIIM genes. The ecoRIIR and ecoRIIM genes share 98.8 % nucleotide identity with the ecoRII endonuclease gene (Bhagwat et al., 1990), and have 100 % nucleotide identity with the Escherichia coli modification methylase gene, ecoRII (Som et al., 1987). In addition to assisting with defence against bacteriophage infection, this restriction modification system has also been reported to contribute to the spread and maintenance of plasmids encoding these systems (Kobayashi, 2001).

Plasmid pK1HV
Plasmid pK1HV was isolated from a K. pneumoniae strain cultured from a healthy child, resident in HCMC. Plasmid pK1HV was the largest of the three sequenced plasmids at 133 191 bp with a G+C content of 52.5 mol% (Fig. 2). Plasmid pK1HV contained 167 predicted CDSs, of which the overwhelming majority were of unknown function. However, pK1HV was also found to carry 11 genes that are associated with resistance to various classes of antimicrobials, including aadA2 (streptomycin), ble (bleomycin), aphA1 (gentamicin), aac(39)-IV (gentamicin), hph (hygromycin), sulII (sulfonamides), forR (chloramphenicol), tetR and tetA (tetracyclines), bla LAP-2 (b-lactams) and qnrS1 (quinolones). Similar to pE66An, pK1HV also carried an IS26-tetR-tetA complex, which is a common mechanism facilitating the transfer of tetracycline resistance. Plasmid pK1HV also harboured a type 1 integron containing the dfrA12-orfF-aadA2 cassette, an antimicrobial resistance region that remains common in contemporarily isolated MDR Gram-negative organisms (Gestal et al., 2005). The presence this dfrA12-orf-aadA2-containing type 1 integron in pK1HV (isolated from the community) again poses questions regarding the ongoing selection of genes encoding resistance to streptomycin.
All of the predicted genes on this extended fragment of plasmid DNA were mostly conserved but functionally unknown, or were genes proposed to encode components required for conjugal transfer. The remaining regions of pK1HV carried the identified antimicrobial resistance genes and an array of IS elements (Fig. 2c). Whilst pK1HV did not contain the mucAB operon, like pE66An and pK18An, it did contain the imp operon, which had 85 % nucleotide identity to impA and impB on the IncI1 plasmid TP110 from a Salmonella Typhimurium strain isolated in the UK in 1968 (Lodwick et al., 1990).

Characterization of qnrS1-containing transposons
Annotation of the three sequenced qnrS1-containing transposons extracted from the plasmid sequences of pE66An, pK18An and pK1HV and their alignments with other described qnrS1-containing fragments are shown in Fig. 3. The alignment of qnrS1-containing fragments from pE66An, pK18An and pK1HV showed that they were identical, except for pK18An, which contained a 980 bp insertion. The qnrS1 gene in all of the three plasmids was located within a transposon structure composed of two identical IS26 elements at either terminal portion of the transposon. In addition to containing the qnrS1 gene, these transposons also carried the bla LAP-2 gene, which confers resistance to narrow-spectrum b-lactams. Additionally, this transposon shared a common backbone with other available qnrS sequences and contained an IS2 element, a putative resolvase (ydaA) predicted to belong to a family of stress proteins (Beliaev et al., 2002), and three other proteins of unknown function (Park et al., 2009;Wu et al., 2008) (Fig. 3). The G+C content of the qnrS-containing transposons was *50 mol%, which was slightly lower than the mean G+C content of the sequenced plasmids in their entirety, consistent with the notion that the qnrS1-encoding transposons have been inserted into these plasmids via horizontal gene transfer. Screening for qnrSI transposons in commensal Enterobacteriaceae isolated in HCMC Using the newly generated DNA sequences of the qnrS1containing transposons from the three sequenced plasmids (Fig. 3), we designed PCR primers to amplify and compare qnrS1-containing transposons from DNA extracted from 112 qnrS1-positive hospital-and community-acquired Enterobacteriaceae that have been described previously (Vien et al., 2011(Vien et al., , 2012see Methods). The locations of the primer-binding sites are shown in Fig. 3 and the predicted sizes of PCR amplicons were 6859 bp (pE66An qnrS1 transposon, type A) and 8059 bp (pK18An qnrS1 transposon, type B). Seventy-one of the 112 isolates (63.4 %) were PCR amplicon-positive for the qnrS1-containing transposon and were of sizes consistent with those described in pE66An and pK18An. The 71 PCR amplicons with known qnrS1-containing transposons were subjected to RFLP analysis with EcoRV, HindIII and PvuII.
In addition to the two described qnrS1-containing transposons that were identified in the three sequenced plasmids, the RFLP mapping patterns from these 71 amplicons also revealed a transposon with a third structure (K34N strain, type C). Using a primer-walking sequencing method, we found that the qnrS1-containing transposon in K34N strain was 6652 bp and identical to the qnrS1containing transposon from E66An strain, except for a 200 bp deletion in a gene of unknown function (Fig. 3). We were therefore able to distinguish three related yet distinct qnrS1-containing transposons of 6859, 8059 and 6652 bp, which we arbitrarily named types A, B and C, respectively.
The 41 isolates with an undetermined qnrS1-containing mobile element were further investigated for the presence of a subfamily of the known transposon. There have been multiple reports regarding the association between qnrS1 and bla LAP-2 genes (Cano et al., 2009;Dahmen et al., 2010;Huang et al., 2008;Park et al., 2009;Poirel et al., 2006Poirel et al., , 2007. Yet, due to a lack of a PCR amplicon, we hypothesized that some isolates contain both the qnrS1 and the bla LAP-2 sequence, but the adjacent regions demonstrate a different structure. A laboratory Escherichia coli strain was successfully transformed with a qnrS1-encoding plasmid extracted from 27 of the 41 isolates with an undefined qnrS1-containing transposon structure. Plasmid DNA from these 27 transformants was extracted and digested with EcoRI and individually probed by Southern blotting targeting the qnrS1 and bla LAP-2 genes. With the resulting plasmid hybridization of the plasmid DNA extracted from the 27 isolates, only two  3. A schematic representation of sequenced qnrS1-containing transposons. Graphical representation of the synteny between the qnrS-containing transposons between the three plasmids sequenced here (pK18An, pE66An and pK1HV) and other sequenced fragments containing the qnrS-encoding region. The plasmids and the host organism in which they were first identified are given. The region with the greatest DNA homology is identified and includes the highlighted ORFs for qnrS (red), a putative IS2 element (grey), a gene encoding a putative resolvase protein (ydaA) and three other ORFs encoding hypothetical proteins of unknown function. Additional genes are colour coded: blue, bla LAP-2 ; grey, IS elements; white, ORFs without a name encoding hypothetical uncharacterized proteins. The locations of the binding sites for PCR amplification of the transposon are highlighted.
produced a detectable signal with both qnrS1 and bla LAP-2 probes on the same digestion fragment, implying that these two plasmids (and their corresponding hosts) also carry a qnrS1-containing transposon of the same subfamily as described in the sequenced plasmids (Fig. 4).
We stratified the proportion of the three qnrS1-containing transposons across all the hospital-and communityacquired isolates that were compared (Table 2). Forty-five of the 63 hospital isolates (71.4 %) harboured a qnrS1-containing transposon that was identical to pE66An (type A), one isolate (1.6 %) carried a qnrS1-containing transposon identical to pK18An (type B) and three isolates (4.8 %) carried a qnrS1-containing transposon identical to K34N (type C). Similarly, in the 49 community isolates, there were 18 (36.7 %), one (2 %) and three (6.1 %) isolates carrying type A, B and C qnrS1-containing transposons, respectively. However, 14 isolates (22.2 %) from the hospital and 27 isolates (55.1 %) from the community qnrS1 gene-positive isolates were negative by PCR targeting the qnrS1-containing transposon, suggesting that the qnrS1 gene in these strains is embedded on a different genetic element, which was undeterminable by the described PCR amplification methods used here.

CONCLUSIONS
Hospital-acquired infections with antimicrobial-resistant organisms can be dangerous, and we have recently shown the problems that can be associated with highly virulent clones of K. pneumoniae (Chung The et al., 2015). Here, we have shown that there is a dominant qnrS1-containing transposon circulating in qnrS1-positive Enterobacteriaceae in HCMC, Vietnam, which we found to be present in 71.4 % of qnrS1-positive isolates from a hospital and in 36.7 % of qnrS1-positive isolates from the community. Moreover, we determined the complete nucleotide sequences of three qnrS1-encoding plasmids. These sequences permitted a description of the circulating genes contributing to an MDR phenotype in three bacterial isolates from a hospital and the community, and also provided insights into the means of adaptation of these plasmids within a variety of hosts and environments. Notably, the DNA sequences of the two plasmids isolated from two different bacterial genes in a hospital setting exhibited substantial homology, thus presenting evidence of genetic transfer among nosocomial commensal bacteria. Finally, the annotation of plasmid pE66An provides further evidence for the association of the PMQR gene qnrS and the ESBL gene bla CTX-M-14 .