Co-occurrence of 3 different resistance plasmids in a multi-drug resistant Cronobacter sakazakii isolate causing neonatal infections

ABSTRACT Cronobacter sakazakii 505108 was isolated from a sputum specimen of a neonate with severe pneumonia. C. sakazakii 505108 co-harbors 3 resistance plasmids of the IncHI2, IncX3, and IncFIB incomparability groups, respectively. These 3 plasmids have acquired several accessory modules, which carry an extremely large number of resistance genes, especially including those involved in resistance to carbapenems, aminoglycoside, tetracyclines, and phenicols and sulphonamide/trimethoprim. These plasmid-borne antibiotic resistance genes were associated with insertion sequences, integrons, and transposons, indicating that the assembly and mobilization of the corresponding accessory modules with complex chimera structures are facilitated by transposition and/or homologous recombination. This is the first report of fully sequence plasmids in clinical Cronobacter, which provides a deeper insight into plasmid-mediated multi-drug resistance in Cronobacter from hospital settings.


Introduction
Enterobacter sakazakii was initially defined in 1980 and reclassified into a new genus Cronobacter in 2007, 1 currently composed of 7 species. 2 Cronobacter species are motile, non-sporeforming, peritrichous rods within the Enterobacteriaceae family and ubiquitously distributed in nature. Cronobacter can cause serious infections in neonates and infants, especially those premature or with low birth weight, [3][4][5] and infections in elderly and immunocompromised adults have also been reported. 6,7 C. sakazakii, C. malonaticus and C. turicensis are considered as opportunistic human pathogens and account for the majority of clinical isolates of Cronobacter. [3][4][5] Cronobacter-induced neonatal infections manifest as necrotizing enterocolitis, meningitis, septicaemia and severe pneumonia with mortality rates of 40-80%, and in most cases are epidemiologically associated with ingestion of contaminated powdered infant formula. [3][4][5] The number of Cronobacter infection cases is underestimated due to misidentification of Cronobacter as other species such as Enterobacter cloacae.
Cronobacter isolates are generally susceptible to the most commonly clinically used antimicrobial agents, but resistance to one or more old-generation antimicrobials such as cephalothin, streptomycin, gentamicin and tetracycline has developed in a few Cronobacter isolates. 8,9 The production of chromosomal AmpC b-lactamases, including CSA-1 and CSA-2 in C. sakazakii, and CMA-1 and CMA-2 in C. malonaticus, confers the resistance exclusively to the first generation cephalosporins (e.g. cephalothin). 10 The tetA(B) gene and additional unknown determinants for tetracycline resistance have been reported in an environmental tetracycline-resistant Cronobacter isolate. 11 A multi-drug resistant (MDR) C. sakazakii isolate of animal origin co-harbors an IncI2 plasmid pWF-5-19C_mcr-1 (accession number KX505142) carrying mcr-1 (colistin resistance) and an IncB/O plasmid pWF-5-19C_NDM (accession number KX505142) containing fosA3 (fosfomycin resistance) and bla NDM-9 (carbapenem resistance). 12 pWF-5-19C_NDM is partially sequenced, while pWF-5-19C_mcr-1 represents the single fully sequenced antibiotic resistance plasmid in Cronobacter. Co-existence of bla VEB-1 (extended-cephalosporin resistance), qnrA (quinolone resistance) and arr-2 (rifampin resistance) in a plasmid-borne class 1 integron has been identified in a nosocomial MDR Cronobacter isolate. 13,14 These examples represent the few reports of plasmid-mediated MDR in clinical Cronobacter isolates, but neither the integron nor the plasmid has been fully sequenced.

CONTACT
This study deals with detailed genetic characterization of 3 resistance plasmids co-existing in a MDR C. sakazakii isolate causing severe neonatal pneumonia. These 3 plasmids carry a total of 22 non-redundant genes or gene loci involved in resistance to antimicrobials and heavy metals. This is the first report of fully sequenced antibiotic resistance plasmids in Cronobacter of clinical origin.

Case report
On April 28 2016, a female neonate with hyperpyrexia, bradypsychia, hyperspasmia, refusal to feed, recurrent apnea and severe skin jaundice was hospitalized in a public children's hospital in Nanjing City, China, and diagnosed to have bilirubin encephalopathy accompanied with severe pneumonia. Once hospitalized, the patient received a series of symptomatic treatments, especially including nutrition support therapy, exchange transfusion, neonatal phototherapy, mechanical ventilation; in addition, empirical intravenous antimicrobial treatment with latamoxef. Bacterial isolates were repeatedly recovered from the sputum specimens during routine sampling and cultivation from April 30th to May 5th, and one of these isolates was designated 505108. Based on the antimicrobial susceptibility test results, the antibiotic therapy was switched to intravenous administration with erythromycin since May 1st. Her symptoms associated with bilirubin encephalopathy and pneumonia gradually improved.
Each plasmid was composed of the backbone regions, together with the accessory modules that were recognized as acquired DNA regions associated with and bordered by mobile elements and inserted at different sites of the backbone ( Figure S1). A total of 22 non-redundant genes or gene loci, which were involved in resistance to antimicrobials (b-lactams including carbapenems, quinolons, aminoglycosides, tetracyclines, phenicols, sulphonamides, trimethoprims, rifampicins, bleomycin and acriflavin) and heavy metals (arsenic, copper, mercury, nickel/cobalt and tellurium), were found not only in the accessory modules but also in the backbones of these 3 plasmids (Table 1 and 2). p505108-NDM could be transferred into E. coli through conjugation, generating the transconjugant 505108-NDM-EC600 (Table 3). Repeated attempts failed to transfer p505108-MDR or p505108-T6SS into E. coli through conjugation and electroporation. Class B carbapenemase activity was observed in both The MDR-1 region # , the MDR-2 region # , Tn6362 # , the aphA1a region # , Tn2 # , the ISCfr9-ISCfr15 region, and 2 separate copies of DIS903D The Tn1696-Tn6322 region # , The sil-cop region # , Tn10 # , IS186B, and IS150 The bla NDM-1 region # , and ISKox3 The aphA1a region # None Note. @ reference plasmids included in genomic comparison; # accessory modules containing resistance genes as listed in Table 2.

General features of p505108-MDR
The p505108-MDR backbone had 95% BLAST query coverage and 99% nucleotide identity to the reference IncHI2 plasmid R478, 15 and these 2 plasmids shared the core IncHI2 backbone markers including repHI2A and repHI2B for replication initiation, parAB and parMR for partition, and the tra1 and tra2 regions for conjugal transfer ( Figure S1 and S2).
Whole genome comparison of p505108-MDR and R478 disclosed 10 different regions (DFRs), designated DFR-1 to DFR-10 ( Figure S2). A DIS903D element  (DFR-1) was inserted between parR and htdA within the tra2 region of p505108-MDR, probably making p505108-MDR non-conjugative. DFR-2 was located between orf564 and orf312 and organized as the hipB to orf411 backbone region, Tn6362, the orf189 to orf258 backbone region in p505108-MDR, but manifested as the Tn1696-Tn6322 region in R478; the acquisition of Tn1696-Tn6322 resulted in the loss of the above 2 small backbone regions from R478. Tn2 (DRF-3) was inserted into orf159 (splitting it into 2 separate parts) in p505108-MDR, which left 5-bp direct repeats (DRs; target site duplication signals of transposition) at both ends of Tn2. DFR-4 existed as the 11.4-kb sil-cop region (conferring resistance to silver and copper) that was inserted between the 2 backbone genes orf159 and orf819 in R478, but as the 25-kb MDR-1 region in p505108-MDR. The 6.9-kb aphA1a region (DRF-5) was observed between int and mucAB in p505108-MDR, and its acquisition led to truncation of mucA and loss of the orf318 to retA region. In R478, the class C tetracycline resistance transposon Tn10 (DRF-6) was inserted into orf300, while the IS186B element (DRF-7) existed between orf321 and ldrB. DRF-8 was composed of the relE to orf612 region and the ISCfr9-ISCfr15 region (both of which lacked resistance genes) in p505108-MDR, while it existed as an IS150 element flanked by 5-bp DRs in R478. The acquisition of ISCfr9-ISCfr15 by p505108-MDR and that of IS150 by R478 led to truncation of downstream orf606 and loss of the upstream relE to orf612 region, respectively. The 52.4 kb MDR-2 region (DFR-9) was inserted into klaB in p505108-MDR, leading to truncation of klaB as well as deletion of the downstream klaA-orf609 region. A second copy of IS903D (DFR-10) was inserted between orf2385 and orf450, leaving both of them truncated.
DFR-1, DFR-3 to DFR-7 and DFR-10 were entirely composed of accessory modules, while the other DFRs consisted of not only accessory modules but backbone regions; the acquisition of accessory modules induced deletion and/ or truncation of surrounding backbone regions ( Figure S2). Although p505108-MDR and R478 shared the overwhelming majority of their backbones, these 2 plasmids carried different profiles of accessory modules, most of which were inserted at different sites across the plasmid backbones.

The MDR-1 region from p505108-MDR
The MDR-1 region (Fig. 1) was a derivative of, although dramatically genetically differed from, the sil-cop region of R478 because these 2 modules were located at the same site of the IncHI2 backbone and had the same terminal regions. The sil-cop region of R478 carried a Tn7-like core transposition module tnsABCD and the silver (sil) and copper (cop) resistance loci. Various derivatives of the silcop region were found in several other IncHI2 plasmids. 16 Being dramatically distinct from the sil-cop region of R478, the MDR-1 region lost the entire sil and the most parts of tnsABCD and cop, but instead acquired several intact or residue mobile elements associated with resistance genes, especially including the dfrA18 region and an unusual In0 with paired terminal inverted repeats (TIRs). The dfrA genes were often associated with ISCR1 17 as observed in the dfrA18 region of p505108-MDR, in which ISCR1 was truncated due to its connection with upstream IS26. The prototype In0 was an empty class 1 integron, but a 1.9-kb Tn5393 remnant carrying the streptomycin resistance module strAB was integrated at a site downstream of the PcW TGN-10 promoter of In0 in p505108-MDR, interrupting In0 into 2 separate parts.

The MDR-2 region from p505108-MDR
The MDR-2 region (Fig. 2) had a complex chimera structure, which carried 5 resistance-conferring substructures, namely the chloramphenicol resistance unit IS26-catA2-IS26, the tetracycline resistance unit IS26-tetR(D)-tetA (D)-IS26 [also designated TntetD 18 ], the extended-spectrum b-lactam resistance unit IS26-bla SHV-12 -IS26, In46 and In615. The 3 IS26-flanking units lacked short DR sequences at both ends and were identified as IS26-composite transposon-like mobile elements. [19][20][21] A total of 7 copies of IS26 were found in the MDR-2 region, and the common IS26 component acts as an adaptor to mediate massive recombination and transposition events and thereby plays a pivotal role in assembly of large MDR regions with complex mosaic structures. 22,23 In46 from the MDR-2 region contained the 5 0 -conserved segment (5 0 -CS) and the sole gene cassette aacA4cr, but lacked the inverted repeat at the integrase end (IRi) and the 3 0 -terminal region composed of 3 0 -CS and inverted repeat at the tni end (IRt), which was resulted from connection of In46 with upstream IS26 and downstream ISCfr8. In615 was a complex class 1 integron that contained 2 resistance gene-carrying variable regions (VRs: VR1 and VR2). ISCR1 mobilized the nearby VR1 together with 3 0 -CS2 from one integron to 3 0 -CS1 of another integron, facilitating the formation of complex class 1 integrons. 17 The primary structure of VR1 from In615 was the gene cassette array aacA27-ereA2, in which ereA2 was interrupted into 2 separate parts by insertion of the IS1247-aacC3-arr7 transposition unit with 4-bp DRs at both ends. 24 The qnrB4-bla DHA-1 region (VR2) connected with ISCR1 was found several complex class 1 integrons carried on plasmids, including pCFI-1 (accession number JN215523), pCFI-2 (accession number JN215524), pCFI-3 (accession number JQ356870), pNMDHA (accession number GU943791), pRBDHA (accession number AJ971343) and pMPDHA (accession number AJ971344). 25 Tn6362 from p505108-MDR Tn1696 (Fig. 3), a unit transposon of the Tn21 subgroup of Tn3 family, was assembled from insertion of class 1 integron In4 into the resolution (res) site of a backbone structure IRL (inverted repeat left)-tnpA (transposase)-tnpR (resolvase)-res-mer-IRR (inverted repeat right). 26 As  a derivative of Tn1696, Tn6362 (Fig. 3) retained the mer-IRR region but had 2 major modifications: i) IRL was interrupted by IS4321R (the IS1111 family members IS4321 and IS5075 target and are inserted at a specific position in the 38-bp TIRs of Tn21 subgroup transposons), 27 which was further interrupted by IS102; and ii) a pbrR-zntA-lspA region, probably involved in zinc uptake, was acquired instead of the tnpAR-res:In4 region. Tn6362 was bracketed by 5-bp DRs, indicating that mobilization of Tn6362 into p505108-MDR was a transposition process requiring the core transposition determinants (TIRs, tnpAR and res), and that the lesion or loss of these core determinants occurred post transposition.
The aphA1a regions from p505108-MDR and p505108-T6SS The presence of 2 highly similar aphA1a regions (Fig. 4) in the 2 co-existent plasmids p505108-MDR and p505108-T6SS were validated, although highly unusual, by a set of PCR amplifications that targeted several key jointing fragments of these 2 aphA1a regions and their surrounding backbone regions, using genomic DNA of the 505108 isolate as template.
The prototype aphA1a region (as observed in p505108-MDR) was likely connected with an IS1R element, which resulted from transposition or homologous recombination, generating the aphA1a region of p505108-T6SS with deletion of DISKpn21-5 0 relative to p505108-MDR. pESA3 and its close derivatives including  Figure S6 for the PCR results. p505108-T6SS have been widely identified as virulence plasmids in pathogenic C. sakazakii strains. 32 Notably, acquisition of the aphA1a region by p505108-T6SS made it to be a carrier of not only virulence determinants but also antibiotic resistance markers.

Concluding remarks
Cronobacter species have the ability to survive in powdered infant formula, and C. sakazakii, C. malonaticus and C. turicensis represent dangerous opportunistic pathogens of neonates. 33 Cronobacter species tend to be more sensitive to most antibiotics than other Enterobacteriaceae species. There are few reports describing the MDR in Cronobacter isolates of both environmental and clinical origins, and molecular mechanisms of antimicrobial resistance in Cronobacter are poorly understood. C. sakazakii 505108, causing severe neonatal pneumonia, co-harbors 3 resistance plasmids belonging to the IncHI2, IncX3 and IncFIB incomparability groups, respectively. These 3 plasmids carry an extremely large number of resistance genes, and most of these plasmidborne resistance genes were associated with insertion sequences, integrons and transposons, constituting various large accessory modules with chimera structures. Mobilization of these accessory resistance modules into plasmid backbones are promoted by transposition and homologous recombination. MDR in Cronobacter isolates leads to limited choice of antibiotics for treatment, resulting in a greater risk of death. Therefore, surveillance of plasmid-mediated MDR in clinical Cronobacter isolates is of paramount importance.

Bacterial strains and identification
Bacterial species identification was performed using 16S rRNA gene sequencing 34 and PCR-detection of a 492-bp cgcA sequence 35 and a 151-bp gyrB sequence 36 specific for C. sakazakii. The MLST scheme for C. sakazakii was derived from the PubMLST database (https://pubmlst. org/cronobacter/).The major plasmid-borne carbapenemase genes were screened for by PCR. 37 All the PCR amplicons were sequenced on ABI 3730 Sequencer (LifeTechnologies, CA, USA) with the same primers as used for PCR.

Sequencing and annotation
Genomic DNA was isolated from the 505108 isolate using a Qiagen large construct kit and sequenced from a mate-pair library with average insert size of 5,000 bp, using a MiSeq sequencer (Illumina, CA, USA). DNA contigs were assembled using Newbler 2.6. 38 Gaps between contigs were filled using a combination of PCR and Sanger sequencing using an ABI 3730 Sequencer. Open reading frames and pseudogenes were predicted using RAST 2.0 39 combined with BLASTP/BLASTN 40 searches against the UniProtKB/Swiss-Prot 41 and RefSeq 42 databases. Annotation of resistance genes, mobile elements and other features was performed using CARD, 43 ResFinder, 44 ISfinder 45 and INTEGRALL. 46 Multiple and pairwise sequence comparisons were performed using MUSCLE 3.8.31 47 and BLASTN, respectively. Gene organization diagrams were drawn in Inkscape 0.48.1.

Phenotypic assays
Activity of Ambler class A/B/D carbapenemases in bacterial cell extracts was determined by a modified CarbaNP test. 48 Bacterial antimicrobial susceptibility was tested by the broth dilution method and interpreted as per CLSI guidelines. 49

Disclosure of potential conflicts of interest
No potential conflicts of interest were disclosed.

Funding
This work was supported by the grants AWS15J006 and 2016YFC1202600 from the National Major Research & Development Program of China.