Concomitant carriage of KPC-producing and non-KPC-producing Klebsiella pneumoniae ST512 within a single patient

Background: KPC-producing Klebsiella pneumoniae of clonal group 258 are prominent in healthcare settings in many regions of the world. The bla KPC gene is mostly carried by a multireplicon IncFII k -IncFI plasmid suspected to be highly compatible and stable in this genetic background. Here, we analysed the genetic diversity of an ST512 K. pneumoniae population in a single patient. Methods: Twelve K. pneumoniae isolates ( n =5 from urine samples and n =7 from rectal swabs) were recovered from one patient over a 2month period. Antimicrobial susceptibility testing, plasmid extraction and WGS were performed on all isolates. The ﬁrst K. pneumoniae isolate, D1, was used as a reference for phylogenetic analysis. Results: Antimicrobial susceptibility testing, plasmid analysis and WGS revealed concomitant carriage of carbapenem-resistant and carbapenem-susceptible K. pneumoniae isolates of ST512, with the absence of the entire bla KPC -carrying plasmid in the susceptible population. Furthermore, 14 other genetic events occurred within the genome, including 3 chromosomal deletions (of 71kb, 33kb and 11bp), 2 different insertions of IS Kpn26 and 9 SNPs. Interestingly, most of the events occurred in the same chromosomal region that has been deleted independently several times, probably after homologous recombination involving 259bp repeated sequences. Conclusions: Our study revealed (to the best of our knowledge) the ﬁrst case of in vivo bla KPC -carrying plasmid curing and a wide within-patient genetic diversity of a single K. pneumoniae ST512 clone over a short period of carriage. This within-patient diversity must be taken into account when characterizing transmission chains using WGS during nosocomial outbreaks.


Introduction
Carbapenem-resistant Klebsiella pneumoniae has emerged as a formidable threat in healthcare facilities. 1 KPC belongs to Ambler class A and emerged globally in Enterobacterales in the early 2000s. 2 The worldwide spread of KPC is multifactorial and has been related to the diffusion of a particular clonal group (CG), i.e. CG258, defined by a few single-locus variants, with ST258, ST11 and ST512 being the most predominant. 3 Notably, the emergence of the bla KPC-3 gene in Italy has been related to the spread of ST512 isolates. 4,5 Other genetic features associated with their spread are related to the genetic structure, a class II transposon containing the bla KPC gene (named Tn4401-like) and an IncFII Ktype conjugative plasmid. 2,6,7 After the worldwide dissemination of these successful clones and CG since 2000, all reports of K. pneumoniae of CG258 refer to KPC-producing strains, with the exception of a study that retrospectively analysed carbapenem-susceptible K. pneumoniae collection strains isolated between 1999 and 2013 in New York City and some of these belonged to ST258. 8 WGS has emerged as a powerful tool to study bacterial evolution. Evolution of the genome of KPC-K. pneumoniae ST258 during long-term human colonization revealed complex plasmid rearrangements and genome plasticity. 9,10 However, little is known about the genetic diversity that resides in the gastrointestinal tract within a single bacterial population at a specific timepoint.
Here, we describe the concomitant carriage of bla KPC -positive carbapenem-resistant and bla KPC -negative carbapenem-susceptible V C The Author(s) 2020. Published by Oxford University Press on behalf of the British Society for Antimicrobial Chemotherapy. All rights reserved. For permissions, please email: journals.permissions@oup.com. K. pneumoniae ST512 clinical isolates in a single patient. Our objectives were to analyse the genetic diversity of KPC-K. pneumoniae isolates and to evaluate the cohabitation of several subpopulations of K. pneumoniae ST512.

Bacterial isolates, MICs and growth conditions
Twelve clinical isolates of K. pneumoniae were isolated from a single patient over a 2 month period. The isolates were obtained from rectal swabs (n = 7) and urine samples (n = 5) and were named after the sampling date, D1 being the first isolate and D54 being the one isolated 53 days later. From rectal swabs, K. pneumoniae isolates were obtained after growth on selective medium supplemented with carbapenems (ChromID V R CARBA SMART; bioMérieux, Marcy-l'Étoile, France), whereas urine samples were spread on non-selective chromogenic medium (UriSelect TM ; Bio-Rad, Marnes-La-Coquette, France).
Antimicrobial susceptibility testing was performed using the disc diffusion method on Mueller-Hinton agar (Bio-Rad) and results were interpreted according to EUCAST guidelines. 11 MICs of carbapenems, ceftazidime, tigecycline, ciprofloxacin and aminoglycosides were determined by Etest (bioMérieux) and MICs of colistin were determined by broth microdilution (Sensititre TM ; Thermo Fisher, France).

Genomic analysis
Total DNA was extracted from colonies using the Ultraclean Microbial DNA Isolation Kit (MO BIO Laboratories, Ozyme, Saint-Quentin, France) following the manufacturer's instructions. The DNA library was prepared using the Nextera XT-v2 Kit (Illumina, Paris, France) and then run on a HiSeq Illumina system (2%150 bp, paired end). The full sequence of the K. pneumoniae D1 genome was obtained using PacBio and Illumina's sequencing technologies. Assembly was performed using the RS_HGAP_Assembly.3 protocol from SMRT analysis toolkit v2.3 and with Canu. 12 The consensus sequence was polished with Quiver and manually corrected by mapping Illumina reads using breseq. 12 The acquired antimicrobial resistance genes were identified using the ResFinder server v3.0 (https://cge.cbs.dtu.dk/services/ ResFinder/). 13 The genome was annotated using Prokka. 14 SNP analysis was performed by mapping the reads from each genome (D2 to D54) against the K. pneumoniae D1 genome as a reference. Variants, SNPs, insertions and deletions were detected using breseq 12 or the variant detection tool of CLC genomic workbench v12.0 (QIAGEN, Les Ulis, France).

Plasmid content analysis
Plasmids were extracted using Kieser's extraction method and subsequently analysed by electrophoresis on a 0.7% agarose gel. 15

Nucleotide sequence accession numbers
The whole-genome sequences generated in the study have been submitted to the GenBank nucleotide sequence database under the accession numbers detailed in Table 1.

Ethics
This study was conducted in accordance with the Declaration of Helsinki and national standards. Informed consent was obtained from the patient.

Case report
In 2015, a patient suffered from acute pancreatitis (Balthazar score E) due to gallstones during a stay in Italy. The patient was hospitalized in Italy for 10 days, during which no nutrition was given, to rest the pancreas and bowels. Food was then reintroduced through a nasogastric tube and the evolution was favourable with exclusive enteric nutrition. The patient was repatriated to France for further medical care. Upon admission, screening for intestinal carriage of carbapenemase-producing Enterobacterales allowed identification of the presence of a KPC-producing K. pneumoniae (isolate D1), likely acquired during hospitalization in Italy. A cholecystectomy was performed 10 weeks after the acute episode and no complications occurred. During the 2 months of follow-up in France, 12 K. pneumoniae isolates (8 KPC-positive isolates and 4 KPC-negative isolates) were recovered from rectal swabs or urine samples ( Figure 1a).

Susceptibility testing and resistome
Susceptibility testing of the 12 isolates recovered revealed two different phenotypes regarding b-lactams, but the same coresistances. WGS revealed in all genomes three genes encoding aminoglycoside-modifying enzymes [aph(3 0 )-Ia, aac(6 0 )-Ib and aadA2], the natural fosfomycin resistance gene (fosA-like) and catA1-like, dfrA12 and sul1 conferring resistance to chloramphenicol, trimethoprim and sulphonamides, respectively. A substitution in GyrA (S83I) was also identified, conferring resistance to fluoroquinolones. The OmpK35 porin was inactivated by an insertion at position 121 of the gene (!G), leading to an early stop codon in the protein. In OmpK36, two amino acids were inserted at positions 135 (!Asp) and 136 (!Gly) in comparison with the WT sequence (NC_016845.1). This OmpK36 variant is known to contribute to increased MICs of carbapenems for K. pneumoniae of CG258. 16 Eight isolates were resistant to carbapenems (imipenem MICs of 4-8 mg/L and meropenem and ertapenem MICs of >32 mg/L), whereas four isolates were susceptible to broad-spectrum cephalosporins and carbapenems (imipenem MICs of 0.125 mg/L). The content of b-lactamase-encoding genes differed between isolates. WGS revealed that all carbapenem-resistant K. pneumoniae isolates contained four b-lactamase genes: bla KPC-3 carried by transposon Tn4401a, bla TEM-1 , bla OXA-9 (not functional due to a premature stop codon) carried by a multireplicon IncFIB-IncFII k plasmid of 113 639 bp (pKpQIL-like) and the naturally chromosomeencoded bla SHV-11 gene. Carbapenem-susceptible isolates possessed only the bla SHV gene of the b-lactamase-encoding genes.
MLST analysis indicated that all K. pneumoniae (KPC positive and KPC negative) belonged to ST512, suggesting that the two populations of K. pneumoniae ST512 seem to differ only by the presence or absence of Tn4401a or of the whole bla KPC -carrying plasmid. In order to distinguish between these two hypotheses, plasmid extractions analysed by electrophoresis revealed that an 100 kb plasmid was missing in all KPC-negative isolates (Figure 1b).

Genomic analysis and phylogeny
To establish whether a gain or a loss of the bla KPC -carrying plasmid occurred in the K. pneumoniae ST512 population, SNPs and genetic events underlying in vivo evolution analyses were explored using isolate D1 as a reference (Figure 1c). Even though the 12 isolates were highly related, a total of 15 different genetic events were identified: the loss of the whole bla KPC -carrying plasmid (in 4 isolates), 3 different deletions of chromosomal regions (71 959 bp in 5 isolates, 33 752 bp in 2 isolates and 11 bp in 2 isolates), 2 different insertions of a copy of ISKpn26 (at position 2309021 in 5 isolates and at position 4117119 in 1 isolate) and 9 different SNPs (involving 7 isolates). Interestingly, the 33 kb deleted region (4106137-4139889) was part of the larger 71 kb deleted region (4068121-4139091). No mobile element could be found surrounding these two deleted regions, but the presence of three copies of repeated sequences (RSs) of 259 bp were present (Figure 1d). These regions share over 96% nucleotide identity ( Figure S1, available as Supplementary data at JAC Online). Recombination events involving RS1/RS3 and RS2/RS3 are likely responsible for the deletions of the 71 and 33 kb fragments, respectively.
Comparison with deposited genomes in public databases indicates that isolate D33-3 was the closest to other K. pneumoniae ST512 genomes. In the chromosome of D1 (KPC positive), a deletion of 11 bp occurred and this deletion was also present in D33-2 (KPC negative), in addition to the insertion of a copy of ISKpn26. Furthermore, the loss of pKPC seems to have occurred in another branch, between D33-3 (KPC positive) and D19-2 (KPC negative) (Figure 1c). These observations make it highly likely that there was loss of the bla KPC -carrying plasmid by the KPC-positive population rather than its acquisition by the KPC-negative population.

Discussion
An unexpected genetic and phenotypic variability of K. pneumoniae ST512 in a single patient was observed as a result of several unrelated genetic events. Concomitant isolation of carbapenem-resistant and -susceptible isolates recovered over a short period of time (2 months) was due to the presence or not of the bla KPC-3 -carrying plasmid (pKPC). In our study, this diversity was clearly underestimated by the use of a selective medium for rectal samples that only allowed the growth of carbapenemresistant bacteria. Hence the carbapenem-susceptible population could only be identified from urine samples. Despite this major bias, three isolates from different branches of the distribution were recovered on the same day (Day 33) as proof that this genetic diversity was present in the patient's microbiota.
Interestingly, over the 15 genetic events described, 5 involved the same chromosomal region (from 4068121-4139091) in 10 isolates. This region included over 60 coding sequences and, among them, the ramA gene (accession number KC843634) was entirely deleted (in 71 and 33 kb deleted genomes), inactivated by a nonsense mutation or putatively transcriptionally affected by the insertion of a copy of ISKpn26 in the intergenic region or by an 11 bp deletion in the ramR regulator (Figure 1d). ramA, an araCfamily transcriptional regulator, is part of the ramR-romA-ramA operon and is involved in the expression of the AcrAB efflux pump, leading to increased MICs of tigecycline and fluoroquinolones. We could not observe any correlation between MICs of these antibiotics and any of the genetic events. However, since this region was inactivated several times by independent mechanisms, it is tempting to speculate that its inactivation confers to this K. pneumoniae ST512 a competitive advantage in that clinical context.
The genetic analysis indicated that a loss of the bla KPC-3 -carrying plasmid likely occurred in the K. pneumoniae ST512 population. pKPC is very closely related to the successful pKpQIL plasmid (99.98% nucleotide identity and 100% query coverage), the first KPC-producing plasmid that was sequenced in 2006 and known to have spread through all of Europe by its tight association with CG258. 17,18 A plasmid stability assay did not find an increased capacity of K. pneumoniae D1 to lose the pKPC plasmid in comparison with other K. pneumoniae isolates carrying pKpQIL-like plasmids (data not shown). Previous genomic analyses of KPC-K. pneumoniae during long-term carriage have reported large plasmid rearrangements, deletions of the entire Tn4401 or plasmid transfer between Enterobacterales, but loss of the entire bla KPCcarrying plasmid by a member of K. pneumoniae CG258 has never been reported to date. 9,19 The pKpQIL-like plasmids are thought to be highly compatible with the K. pneumoniae CG258 genetic background and to have contributed to the worldwide dissemination of KPC carbapenemase. 2,20 Information regarding the antibiotic selection pressure that occurred during the patient's hospitalization was available (Figure 1a). Given the genetic support of resistance genes linked to these antimicrobial agents, cefixime seems to be the sole molecule capable of maintaining selective pressure to prevent the loss of the KPC plasmid; however, it was prescribed on Day 44, far after the isolation of the first non-KPC producer on Day 19. Thus, most of the patient's bacterial follow-up was done when no antimicrobial selective pressure occurred and we witnessed the natural history of a colonizing K. pneumoniae ST512 isolate.
Overall, we report here a wide genetic diversity of K. pneumoniae ST512 in a single patient who underwent low antimicrobial selective pressure. This diversity must be taken in account when trying to infer transmission routes using WGS during nosocomial outbreaks.