Characterization of a NDM-1- Encoding Plasmid pHFK418-NDM From a Clinical Proteus mirabilis Isolate Harboring Two Novel Transposons, Tn6624 and Tn6625

Acquisition of the blaNDM–1 gene by Proteus mirabilis is a concern because it already has intrinsic resistance to polymyxin E and tigecycline antibiotics. Here, we describe a P. mirabilis isolate that carries a pPrY2001-like plasmid (pHFK418-NDM) containing a blaNDM–1 gene. The pPrY2001-like plasmid, pHFK418-NDM, was first reported in China. The pHFK418-NDM plasmid was sequenced using a hybrid approach based on Illumina and MinION platforms. The sequence of pHFK418-NDM was compared with those of the six other pPrY2001-like plasmids deposited in GenBank. We found that the multidrug-resistance encoding region of pHFK418-NDM contains ΔTn10 and a novel transposon Tn6625. Tn6625 consists of ΔTn1696, Tn6260, In251, ΔTn125 (carrying blaNDM–1), ΔTn2670, and a novel mph(E)-harboring transposon Tn6624. In251 was first identified in a clinical isolate, suggesting that it has been transferred efficiently from environmental organisms to clinical isolates. Genomic comparisons of all these pPrY2001-like plasmids showed that their relatively conserved backbones could integrate the numerous and various accessory modules carrying multifarious antibiotic resistance genes. Our results provide a greater depth of insight into the horizontal transfer of resistance genes and add interpretive value to the genomic diversity and evolution of pPrY2001-like plasmids.


INTRODUCTION
Urinary tract infections (UTIs) are the most common bacterial infections (Gastmeier et al., 1998). Cases of UTIs can be classified as uncomplicated or complicated (Beahm et al., 2017). Clinically, Proteus mirabilis is most frequently a pathogen of UTIs, particularly in patients suffering from complicated cUTIs (Schaffer and Pearson, 2015). Although Escherichia coli is the primary urinary tract pathogen, P. mirabilis ranks third as the cause of UTIs and accounts for 4.1% of urinary tract infection isolates in CANWARD surveillance study, 4.6% in southern China, respectively (Karlowsky et al., 2011;Li et al., 2017). Because this pathogen is intrinsically resistant to nitrofurantoin, polymyxin, and tigecycline antibiotics (Ramos et al., 2018), acquiring additional carbapenemase antibiotic resistance is worrisome (Reffert and Smith, 2014). Currently, fosfomycin, which is previously used mainly as oral treatment for UTIs, has gained clinicians' attention worldwide because of its activity against multidrug-resistant bacteria (Reffert and Smith, 2014;Giske, 2015). In additionally, fosfomycin resistance rates are generally low but substantially higher when carbapenemase producers are considered (Giske, 2015). One such resistance gene is bla NDM-1 (New Delhi metallo-β-lactamase), which was initially identified in a Klebsiella pneumoniae strain (Yong et al., 2009). Isolates of this species that harbor the bla NDM-1 gene can hydrolyze nearly all β-lactam antibiotics except aztreonam. Therefore, the acquisition of bla NDM-1 by P. mirabilis would be problematic, as it would greatly reduce the therapeutic options for treating infections caused by it.
Here, we studied the bla NDM-1 -harboring plasmid, pHFK418-NDM, a known pPrY2001-like plasmid according to its replicon, which was first isolated from a clinical P. mirabilis HFK418 strain in China. We elucidated the complete sequence of pHFK418-NDM (which carries two novel transposons, Tn6624 and Tn6625) and compared it with six other pPrY2001-like plasmids to obtain insight into the horizontal transfer of resistance genes and the diversity and evolution of pPrY2001-like plasmids.

Species Identification and Antimicrobial Susceptibility Testing
The study was approved by the Medical Ethics Committee at the Affiliated Hospital of Qingdao University, China, and written informed consent was received from the patient. The P. mirabilis HFK418 strain was isolated from the urine specimen of a patient with epidemic encephalitis at the Affiliated Hospital of Qingdao University, China, in 2017. Referring to the method described in Ranjan et al. (2016), this strain was multiple tested for purity by routine laboratory methods, then the pure strain was cryopreserved at −80 • C in 50% glycerol. The pure isolate was revived in Luria-Bertani (LB) broth (BD Biosciences, United States) with 4 µg/ml meropenem to experiments. The P. mirabilis HFK418 isolate was identified and subjected to antimicrobial susceptibility testing using the VITEK compact-2 automated system (bioMérieux, France). In addition, fosfomycin MICs were further determined by fosfomycin E-tests (bioMérieux). CLSI (Clinical and Laboratory Standards Institute) 2018 breakpoints were used (M100-S28) (CLSI, 2018).

Antimicrobial Resistance Gene Screening and Plasmid Conjugal Transfer
The major acquired extended-spectrum β-lactamase (Dallenne et al., 2010;Hussain et al., 2014;Ranjan et al., 2017), fosfomycin (Dantas Palmeira et al., 2018), chloramphenicol (White et al., 1999), lincosamide (Garcia-Martin et al., 2018), and carbapenemase genes (Chen et al., 2015;Ranjan et al., 2016Ranjan et al., , 2017 were detected by PCR, after which all the PCR amplicons were sequenced on the ABI 3730 platform (Applied Biosystems, United States). The sodium azide-resistant E. coli J53Azi R strain was used as the recipient and the P. mirabilis HFK418 isolate as the donor for the conjugative transfer of the plasmids. The conjugal transfer tests were performed as described previously (Srijan et al., 2018), and the conjugation frequency was calculated as transconjugants divided by number of donors.

Carbapenemase Activity Assay
To determine whether the bla NDM-1 gene was expressed in both P. mirabilis HFK418 and the E. coli J53Azi R transconjugant HFK418-NDM-J53 strain, we performed an imipenem-EDTA E-test (AB-BioMérieux, Sweden) to assess the class B carbapenemase activity.

Sequencing and Sequence Assembly
Bacterial genomic DNA was extracted using the Wizard Genomic DNA Purification Kit (Promega, United States), followed by the MiSeq (Illumina, United States) and MinION (Oxford Nanopore, United Kingdom) sequencing. The short Illumina reads were trimmed to remove the poor quality sequences, and the resultant contigs were assembled using Newbler3.0 (Nederbragt, 2014). The longest single read obtained by the MinION sequencer was 98 kb, thereby crossing the repetitive shufflon regions in the plasmid (Laver et al., 2015). The long reads from the MinION combined with the short Illumina reads were hybrid assembled using SPAdesv3.11.1 (Bankevich et al., 2012). Hybrid assembly produced several scaffolds and BLASTN analysis confirmed that the scaffold in our study has the highest similarity to the plasmid p16Pre36-NDM (Accession no. KX832927) with coverage of 69% and identity of 96%. As most of the published plasmids are in a circle form, further bioinformatics analysis confirmed that this scaffold can be successfully cyclized using our in-house script. The correctness was then proved by mapping the highthroughput sequencing reads to the cyclized scaffold using CLC Genomics Workbench 9.0, with a mean reads mapping coverage of 111x. The consensus sequence acquired from CLC Genomics Workbench 9.0 was finally treated as the complete sequence of our plasmid pHFK418-NDM.

Nucleotide Sequence Accession Number
The complete nucleotide sequence of plasmid pHFK418-NDM has been deposited in the National Center for Biotechnology Information nucleotide database 2 under accession number MH491967.

Antibiotics
MIC (mg/L)/antimicrobial susceptibility * HFK418 HFK418-NDM-J53 J53 The genomic structures of the pPrY2001-like plasmids comprised two major regions: the backbone and accessory module. The backbone could be further divided into three parts: the replication genes (repA and its iterons), the conjugal transfer genes (tiv, rlx, and cpl), and the plasmid maintenance genes (parFG, MazFE, stbB, ssb, and flhC). Each plasmid's backbone was able to integrate two or more accessory modules by transposition or recombination events. pHFK418-NDM contains two accessory modules, the Tn6901 related region and the multidrug-resistant (MDR) region, while the MDR region contains Tn6625 and Tn10 (Supplementary Figures S2, S3).

Backbone Regions in the pPrY2001-Like Plasmids
Our pairwise comparison analysis of the pPrY2001-like plasmids backbones showed that they shared >96% nucleotide identity across >42%, indicating that their backbones were relatively conserved. However, there were three major differences among all their backbones. (I) the parC gene (centromere, binding sites for parG) did not exist in pPrY2001, and the copy numbers of the 8-bp tandem repeat (TGTGTata) within the parC gene varied among the other plasmids (4 for p06-1619-1, pC131, and pPm60; 5 for pPp47, pHFK418-NDM, and p16Pre36-NDM).
(II) Compared with the conjugal transfer region in the other plasmids, the rlx gene from pPrY2001 is disrupted into rlx-3 and rlx-5 by insertion of ISPrre5 (named in this study).
(III) The hybrid backbone of plasmid p16Pre36-NDM was acquired from a pPrY2001-like plasmid and the IncC2 plasmid (the orf1847 and rhs2 marked genes) (Supplementary Figure S3).
Tn6624, a novel IS26-based transposon unit, has been inserted into the pHFK418-NDM plasmid from P. mirabilis HFK418. Delimited by 8-bp DRs (CATCGGCG), it has the following mosaic structure: IS26, a novel IS66-family ISPrre3, mph(E) (macrolide resistance), msr(E) (macrolide efflux protein), IS26, a fragment with an unknown function, and IS26. The mph(E)-msr(E)-IS26 fragment originated from the IS26-mph(E)-msr(E)-IS26 transposon unit and was initially identified in the chromosomal integrative conjugative element from Pasteurella multocida (Michael et al., 2012). Three copies of IS26 are present in Tn6624, which promotes the formation and transposition of Tn6624. Another novel 48,068 bp multidrug resistance transposon, Tn6625, was found in the pHFK418-NDM plasmid from P. mirabilis HFK418. The Tn1696, Tn6260, In251, Tn125, Tn6624, and Tn2670 mobile elements have been described in detail above, and all of them are included in the large composite Tn6625 transposon. Tn6625 carries twelve resistance genes, bounded by 3-bp DRs (TTG). Tn6625 contains integron In25, which has so far only been found in wastewater-isolated Providencia VIGAT3 (Guo et al., 2011). Thus, In251 was first isolated from clinical P. mirabilis HFK418, suggesting that it has been efficiently transferred from environmental micro-organisms to clinical isolates ( Figure 1A).
There are other transposon units also ( Tn6346, the truncated GIsul2 region, and Tn1548) in the MDR region of p16Pre36-NDM and pPrY2001, except as described above.
We found that Tn4352 and the truncated aacC2-tmrB region are integrated between Tn7-3 and Tn7-5 in pPm60.
Flanked by 8-bp DRs at both ends, Tn4352 is an IS26-bounded structure (IS26-aphA1a-IS26), and the aphA1a resistance gene confers resistance to kanamycin and neomycin (Wrighton and Strike, 1987). Although Tn4352 is complete in the MDR region-2 from pPm60, it is truncated in the MDR region-2 from p16Pre36-NDM and pPp47. Furthermore, the structure of Tn4352 is IS26-aphA1a in p16Pre36-NDM and IS26-aphA1a in pPp47. The orientation of Tn4352 in p16Pre36-NDM is direct, but reversed in pPp47and pPm60. The aacC2-tmrB region is present in plasmids pCTX-M3 and pU302L, is derived from transposon Tn2 from the Tn3-family, and contains a IS26 mobile element at its right-hand end (Partridge, 2011). The aacC2 and tmrB genes account for aminoglycoside and tunicamycin resistance, respectively. The truncated aacC2-tmrB region in pPm60 is composed of an aacC2-tmrB-orf192-orf228-orf1182 segment. The direction of the truncated aacC2-tmrB region is direct in pPm60, but reversed in pPp47. Owing to the insertion of a 28,064 bp exogenous region (with an unknown function), the truncated aacC2-tmrB region from pPp47 is segmented into two parts: ISCfr1-3 exists in the MDR region-1, while a fragment from ISCfr1-5 to the aacC2 gene is embedded in the MDR region-2. Similarly, Tn5563 is also located in the two MDR regions of pPp47. Tn5563 was originally discovered in plasmid pRA2 from Pseudomonas aeruginosa (Yeo et al., 1998), and two segments of Tn5563 in pPp47 are arranged as follows: the reverse segment (the mer operon and IRR) is present in MDR region-1 and the direct fragment (IRL and tnpR) is present in MDR region-2 (Figures 1A, 2, 3).

Other Accessory Modules Outside the MDR Region of pPrY2001-Like Plasmids
We found that Tn6901 has a complete structure in pHFK418-NDM, but it is interrupted by insertion of the virulence-related region to generate two segments, Tn6901-5 and Tn6901-3 . Tn6901 is made up of an IRL-tnpA-res-tnpR-frmB (Sformylglutathione hydrolase)-glo (glyoxalase resistance)-frmA (S-glutathione dehydrogenase)-frmR (negative transcriptional regulator)-IRR structure in plasmid Rts1 from Proteus vulgaris, flanked by 5-bp DRs (Murata et al., 2002). Tn6901 is inserted between orf1528 and orf942 in the backbone of pHFK418-NDM, bracketed by 5-bp DRs. pHFK418-NDM, a pPrY2001like plasmid, is the only virulence gene-carrying plasmid, indicating that this plasmid can not only carry a large number of drug resistance genes, but also integrate virulence genes within it (Figure 1B).
It is known that xerC and xerD genes are site-specific recombinases in the lambda integrase family, where it was found that xer-mediated recombination events resulted in the transmission of resistance gene between plasmids and chromosomal locations (Merino et al., 2010). The dfrA6-ereA region is located downstream of the conjugal transfer region in p16Pre36-NDM, and has undergone xer-mediated recombination. The dfrA6-ereA region consists of xerC, recD, xerD, dfrA6 (trimethoprim resistance), ereA (erythromycin resistance), and dinB (Supplementary Figure S2).

CONCLUSION
The bla NDM-1 -harboring pHFK418-NDM plasmid, a pPrY2001like plasmid group member, was first recovered from a clinical multidrug resistant P. mirabilis HFK418 isolate in China. Our data have revealed that the pHFK418-NDM plasmid contains two novel transpositions, Tn6624 and Tn6625. Tn6625, a large composite transposon, has integrated a variety of mobile elements, such as the bla NDM-1 -carrying Tn125, mph(E)harboring Tn6624, and In251. In251 was first identified from the above-mentioned clinical isolate, suggesting that it had been efficiently transferred from environmental organisms to clinical isolates. The pHFK418-NDM plasmid was found to have the ability for conjugal transfer, and to harbor a large numbers of resistance and virulence genes.
The pPrY2001-like plasmids described above harbor a wide variety of antimicrobial resistance genes, with the exception of p06-1619-1. Their relatively conserved backbones have integrated a great variety of accessory modules in the form of resistance genes, gene clusters, insertion sequences, transposons, and integrons, all of which enhance the diversification and evolution of the pPrY2001-like plasmids. Our findings augment our current understanding on the horizontal transfer of resistance genes and the genetic diversity and evolution of pPrY2001like plasmids.

AUTHOR CONTRIBUTIONS
YZ, YT, and XZ conceived the study and designed the experimental procedures. DD, ZL, JF, NJ, and HZ performed the experiments. DD and ML analyzed the data. YZ, YT, XZ, BZ, and TZ contributed to reagents and materials. YZ, YT, and DD wrote the manuscript.

ACKNOWLEDGMENTS
We gratefully acknowledge Prof. Dujun Li for his helpful discussion and continuous encouragement.