We included all KPC-Kp sent to France’s National Reference Center for Antimicrobial Resistance during January 1–December 31, 2018. As previously described, we used isolates that were recovered from clinical and screening specimens and sent on a voluntary basis by any type of laboratory related to any health facility, such as private and public hospitals, nursing homes, and community laboratories (3,4). These laboratories were located throughout France, including overseas territories. KPC-Kp recovered by the National Reference Center for Antimicrobial Resistance represent ≈85%–90% of the KPC-Kp infection cases reported to the French Public Health Agency (R.A. Bonnin, L. Dortet, unpub. data). The collection used for WGS analysis represents a total of 63 nonduplicate isolates recovered from rectal screening (n = 45), urine (n = 12), blood cultures (n = 1), wound infections (n = 2), and respiratory samples (n = 2) and 1 isolate for which no recovery site information was available. Because the aim of the study was to evaluate the genetic diversity of KPC producers, we discarded from further analysis any duplicate isolates or isolates recovered from the same patient.

We performed antimicrobial susceptibility testing by using the disc diffusion method on Mueller-Hinton agar (Bio-Rad, https://www.bio-rad.com) and interpreted results according to European Committee on Antimicrobial Susceptibility Testing guidelines as updated in 2018 (http://www.eucast.org). We determined MICs for colistin by using broth microdilution (Sensititer Thermofisher, https://www.thermofisher.com). We performed carbapenemase detection by using Rapidec Carba NP (bioMérieux, https://www.biomerieux.com), followed by immunochromatographic detection of the carbapenemase enzyme using NG-Carba5 test (NG Biotech, https://ngbiotech.com).

We sequenced all KPC-Kp isolates by using Illumina technology as previously described (17). We extracted total DNA from colonies by using the Ultraclean Microbial DNA Isolation Kit (MO BIO Laboratories, https://www.mobio.com) according to the manufacturer’s instructions. We prepared the DNA library as previously described (17) and performed de novo assembly and read mappings by using CLC Genomics Workbench 12.0 (QIAGEN, https://www.qiagen.com). We identified the acquired antimicrobial resistance genes by using Resfinder 3.1 (https://cge.cbs.dtu.dk/services/ResFinder/) and the CARD database (https://card.mcmaster.ca). We annotated the genomes by using RAST (18). We performed phylogenic analysis by using CSIphylogeny 1.4 (https://cge.cbs.dtu.dk/services/CSIPhylogeny) and visualized the genomes by using FigTree 1.4.3 (http://tree.bio.ed.ac.uk/software/figtree). We performed sequences alignments by using ClustalW (https://www.genome.jp/tools-bin/clustalw). We analyzed single-nucleotide polymorphisms (SNPs) on the whole genome by using CSIphylogeny V1.4 with parameters as follows: select minimum depth at SNP position at 10×, minimum distance between SNPs at 10 bp, and minimum SNP quality score of 30.

We constructed the genetic contexts by using de novo assembly or by mapping with reference genomes from GenBank and verified by in-house PCR as previously described (17). We analyzed plasmid contents of clinical isolates by using PlasmidFinder 2.1 to search for the replicase gene and by conducting manual searches for genes showing homology with the replicase gene.

In 2018, a total of 3,931 carbapenem-resistant Enterobacteriaceae were collected, including 1,259 carbapenem-resistant K. pneumoniae, among which 1,010 were carbapenemase producers. OXA-48-like enzymes were the most prevalent carbapenemases (69.4%), followed by NDM (17.1%); 37 isolates (3.7%) had a combination of both of these carbapenemases. KPC enzymes represent only 3.0% of all carbapenemases, corresponding to 6.8% (69 isolates, including 6 duplicated isolates) of all carbapenemase produced by K. pneumoniae. KPC also was produced by 13 non–K. pneunomiae, including 5 KPC-2 producers (2 Escherichia coli, 1 Klebsiella oxytoca, 1 Enterobacter cloacae, and 1 Citrobacter koseri) and 8 KPC-3 producers (5 E. coli, 1 C. freundii, 1 E. cloacae, and 1 K. aerogenes). Accordingly, K. pneumoniae is the most prevalent species (84.1%) among KPC producers.

Susceptibility testing revealed that all KPC-Kp were resistant to all broad-spectrum cephalosporins (ceftazidime, cefotaxime, cefepime) and monobactam (Figure 1). All KPC-Kp were resistant to ertapenem. According to European Committee on Antimicrobial Susceptibility Testing breakpoints, 62.1% (41/66) KPC-Kp isolates remained susceptible to imipenem and 30.3% (20/66) to meropenem (Figure 1). Ceftazidime/avibactam (98.5% susceptibility) and colistin (92.2%) remained the most potent agents (Figure 1, panels B and C). However, we identified 1 isolate resistant to ceftazidime/avibactam but susceptible to carbapenems (Figure 1). This isolate produces a new variant of KPC, named KPC-39, that has been reported to possess increased ceftazidime catalytic activity but also to have concomitantly lost its carbapenemase activity (19). Among other antimicrobial families, 84.9% KPC-Kp isolates were susceptible to tigecycline, 30.3% to sulfamethoxazole/trimethoprim, and 19.7% to ciprofloxacin (Figure 1, panel D). Resistance to aminoglycosides varied from 40.9% for amikacin to 75.8% for tobramycin.

We performed WGS on 63 nonduplicate KPC-Kp isolates and identified their resistomes by using Illumina technology. In this collection, 44 isolates possessed the bla_KPC-3 gene (69.8%), and 18 (28.5%) possessed the bla_KPC-2 gene. One isolate harbored a novel single-nucleotide variant of bla_KPC-3, bla_KPC-39 (Figure 1, panel B). Two isolates produced 2 carbapenemases, including 1 isolate coharboring bla_KPC-2 and bla_VIM-1 and another 1 coharboring bla_KPC-2 and bla_NDM-4 (Figure 2). We identified additional antimicrobial-resistance determinants in all isolates (Appendix 1). Approximately 46% of the KPC-Kp isolates carried an extended spectrum β-lactamase encoding gene, including 22 isolates harboring bla_CTX-M-15; 4 isolates coharboring bla_CTX-M-14 and bla_CTX-M-15; and 3 isolates with bla_CTX-M-14, bla_CTX-M-3, or bla_CTX-M-65. The other acquired β-lactam resistance determinants encoded for the narrow-spectrum β-lactamases OXA-9 and TEM-1. To decipher quinolone resistance, we analyzed the presence of plasmid-mediated resistance determinants and known mutations in gyrase and topoisomerases (20–22). Resistance to quinolones was mediated by either mutation in gyrase gyrA affecting the residues S83 (p.S83I, n = 42; p.S83F, n = 7; and p.S83Y, n = 2) or D87 (p.D87N, n = 7; p.D87G, n = 3; and p.D87A, n = 2) or parC affecting the residues p.S80 (p.S80I, n = 51) or the production of Qnr (Qnr66-like, QnrB6, or QnrS1). Aminoglycoside resistance was caused by the production of the 16S RNA methylase RmtB (n = 7) or an aminoglycoside-modifying enzyme (encoding by aac(3′)-IIa, aadA1, aadA2, aac(6′)-Ib-cr, or strA/strB). Colistin resistance (n = 5) resulted systematically in chromosome-encoded resistance with alteration of the mgrB gene. In these isolates, 2 possessed a nonsense substitution (p.Q30*), 1 missense involved in colistin resistance (p.C27W) (23), an ISKpn26-like inserted (with direct repeats [DRs] of 4 bp: TTAA), and a missense mutation leading to the disappearance of the start codon. No isolate had plasmid-encoded resistance mcr-1.

Phylogenetic analysis revealed that the 63 KPC-Kp belonged to 15 different clones (STs) circulating in France (Figure 2; Appendix 1). Although many studies have asserted that CG258 is responsible for the spread of bla_KPC (6,8–11), in our collection, only 8 isolates (12.7%) belonged to CG258 (4 each for ST258 and ST512). Furthermore, epidemiologic investigations revealed no link between these isolates (Figure 2). Because KPC is not widely disseminated in France, we did not expect to observe such clonal diversity. Indeed, the epidemiology of KPC in France is not comparable to what was reported in nearby countries in Europe where KPC is endemic, such as Greece and Italy, and where the spread of bla_KPC-2/-3 is clearly linked to CG258 (24,25). Among the 8 isolates we identified that belonged to CG258, 3 were recovered from patients with travel history in Greece (isolate 175C3 and isolate 177H5) and Italy (isolate 160C2). In France, the 3 most prevalent clones are ST307 (with 15 isolates), ST147 (12 isolates), and ST13 (7 isolates). By using the 21 SNP cutoff value proposed by David et al. to identity a single hospital outbreak caused by a ST258 and ST512 cluster (8), we identified that the ST307 clone was overrepresented because of an outbreak that included 11 isolates (Figure 2; Appendix 2). However, this ST307 also included 4 isolates that were not related to this outbreak, such as the 195I4 strain, which was isolated from a patient who traveled in Crete (Greece) and possesses an additional carbapenemase-encoding gene (bla_NDM-4). The second most prevalent clone, ST147, seemed to have disseminated upon distinct events (Appendix 2). Most of the ST147 isolates have been recovered from different areas with no epidemiologic link between the patients (Figure 2). A link with Portugal has been identified for most (9/11) patients infected or colonized with a KPC-3–producing K. pneumoniae of ST147 (Figure 2). The same link with Portugal was observed for 4 patients infected or colonized with a KPC-3–producing K. pneumoniae of ST231. Strains from ST383 represented a small outbreak for which cross contamination was evidenced (<20 SNPs between 4 isolates [Appendix 2]). One isolate (171J7) was distantly related to other clones and corresponded to K. variicola. KPC-2-producing K. pneumoniae isolates of ST11 were predominantly linked to patients who had a history of travel in Asia (China and Vietnam), where this ST is known to be the main vector of bla_KPC dissemination.

Analysis of the close genetic context of bla_KPC highlighted diversity in the genetic structures at the origin of the acquisition of the carbapenemase-encoding gene. The well-known Tn4401a (in 29 isolates) and Tn4401d (in 26 isolates) were the most prevalent structures identified (Figure 2 and 3; Appendix 1). The KPC-Kp of the 2 main clones ST307 and ST147, bla_KPC, is carried on Tn4401a in ST307 and Tn4401d in ST147. Two unrelated isolates (ST11–167I9 and the K. variicola 171J7 isolate) harbored bla_KPC in the Tn4401b isoform. In the remaining 6 isolates (ST273–171J9, ST147–199D1, ST1788–189B3, ST11–171G8, ST11–190F6, and ST11–171J10), bla_KPC-2 is localized in an NTE element (Figures 2, 3). Although 3 isolates belonged to ST11, they displayed 200–800 SNPs of differences along their core genome, indicating that they were unrelated (Appendix 2). The links with 3 different countries (Portugal for ST11–171G8, China for ST11–190F6, and Vietnam for ST11–171J10) are consistent with this unrelatedness. Analysis of NTE elements revealed 4 different structures even if common features were observed (Figure 3). For instance, the presence of a fragment of ISKpn6 downstream of bla_KPC and a copy of ISKpn27 upstream were always present (Figure 3). DRs of TATAGG bracketing ISKpn27 indicated a transposition process that occurred inside the resolvase gene of Tn3. Immediately upstream of bla_KPC-2 (74 bp), the presence of the inverted repeat right of Tn3 is present in all NTE, indicating that all these structures were related. However, the NTE differed by the size of the deletions that are present between ISKpn27 and bla_KPC-2 (from 280 bp in NTE-190F6 to 940 bp in NTE-199D1). We could observe a remnant of bla_TEM-1 in longer structures, but it was not functional anymore. Analysis of the 4 NTE revealed that in NTE-199D1, several copies of IS26 bracketed the whole structure, indicating that this IS might be involved in its acquisition by transposition or a recombination event. IS26 has been recently demonstrated to be able to transpose and thus create a class I transposon by targeting another copy of IS26 (26). NTE-171J10 is inserted in the fip gene of IncN-type plasmids with the presence of DRs surrounding the NTE-171J0 (Figure 3). The fip gene has already been demonstrated to be an integration hot spot in IncN-type plasmids (27,28). DRs as well as putative inverted repeats of Tn3-family transposon are present at the integration site (Figure 3). Moreover, the presence of the complete Tn3 transposase gene indicated that NTE-171J10 might be functional. In NTE-189B3, a new class I transposon carrying a protein of unknown function has been identified. DRs bracketed ISApu1 and ISApu2, indicating a transposition process mediated by these close insertion sequences (Figure 3).