Carbapenem Resistance Conferred by OXA-48 in K2-ST86 Hypervirulent Klebsiella pneumoniae, France

We recovered 2 carbapenem-resistant K2-ST86 hypermucoviscous Klebsiella pneumoniae isolates from patients in France. The isolates had genetic attributes of hypervirulent K. pneumoniae but differed in ability to cause mouse lethality. Convergence of hypervirulent K. pneumoniae toward resistance could cause a health crisis because such strains could be responsible for severe and untreatable infections.

We recovered 2 carbapenem-resistant K2-ST86 hypermucoviscous Klebsiella pneumoniae isolates from patients in France. The isolates had genetic attributes of hypervirulent K. pneumoniae but differed in ability to cause mouse lethality. Convergence of hypervirulent K. pneumoniae toward resistance could cause a health crisis because such strains could be responsible for severe and untreatable infections.
was probably responsible for carbapenem resistance in Kpn2166. In the absence of rpsJ variants previously associated with tigecycline resistance, Kpn2166 resistance to tigecycline and quinolones probably resulted from a frameshift mutation in ramR (C→T at position 364) (7).
We identified IncL replicon in Kpn154 and IncN replicon in Kpn2166. The IncL replicon of Kpn154 was a canonical pOXA-48-like plasmid encoding bla OXA-48 (GenBank accession no. JN626286). In the Kpn2166 isolate, bla CTX-M-15 was encoded by a new ST9-IncN plasmid, included in an IS26-based composite transposon and downstream a truncated IS-Ecp1 insertion sequence.
We compared the pVIR-Kpn154 and pVIR-Kpn2166 plasmids with 16 complete hypervirulent K. pneumoniae plasmid sequences (Appendix Table) from the PATRIC database (http://www.patricbrc. org). A single-nucleotide polymorphism-based phylogenic tree showed a link between the major tree branches and the sequence types of hypervirulent K. pneumoniae owners but not with their geographic origin ( Figure 1, panel A). This finding suggests emergence was more likely caused by spread over distant geographic areas than by local expansion and that limited horizontal transfers between hypervirulent K. pneumoniae isolates probably resulted from the absence of known genes involved in conjugation. Phylogenetic tree analysis also showed 5 groups based on virulence gene synteny, with the predominant group typified by pLVPK ( Figure 1, panel B). Because plasmids of ST23-like hypervirulent K. pneumoniae all share a similar synteny of virulence genes, ST86 hypervirulent K. pneumoniae contains a diversity of plasmid synteny groups ( Figure 1, panel A), suggesting that rearrangements of virulence genes occurred several times along plasmid evolution at rearrangement hotspots active in non-ST23 genetic background. For example, pVIR-Kpn2166 and pVIR-Kpn154 differed from reference plasmid pLVPK by the permutations in the ≈100-kb region flanked by IS5 mobile elements ( Figure 1, panel C). pVIR-Kpn154 contained an additional copy of IS5, which was associated with another ≈30-kb permutation + translation event, suggesting that IS5 is a key factor in the evolution and diversity of hypervirulent K. pneumoniae plasmids. The chromosome of Kpn154 and Kpn2166 exhibited similar organization but differed by 128 insertion/deletion mutations and 1,928 single-nucleotide variants (Appendix Figures 2, 3). The Kpn154 chromosome-mediated ybt virulence locus, which encodes the yersiniabactin, was located in the integrative conjugative element ICEKp3 and was typed ybST202-1LV (6). In Kpn2166, ybt was located in an original isoform of ICEKp12, presenting an ≈34-kb deletion compared to the canonical 97,771-bp ICEKp12, and was typed ybST13-1LV (6).
Although Kpn154 and Kpn2166 have the same genetic background and share the same virulence score (6), they also have allelic and synteny differences in virulence genes. We therefore compared the virulence of these isolates in a sepsis model based on outbred Emerging Infectious Diseases • www.cdc.gov/eid • Vol. 26, No. 7, July 2020

Figure 1. Comparison of pVIR-Kpn2166 and pVIR-Kpn154
Klebsiella pneumoniae isolates from 2 patients in France (bold) with 16 hypervirulent K. pneumoniae virulence plasmids recovered from the PATRIC database (http://www. patricbrc.org). A) Single-nucleotide polymorphism-based phylogenetic tree built by RaxML from an alignment generated by Burrows-Wheeler Aligner and filtered to remove recombination using Gubbins as previously described mice challenged intraperitoneally, as described (Figure 2) (10). Mice injected with 10 3 CFUs of Kpn154 or hypervirulent K. pneumoniae reference strain K. pneumoniae NTUH-2044 died in <72 h, in contrast to mice inoculated with Kpn2166 or American Type Culture Collection (ATCC) 13883, showing that only isolate Kpn154 is hypervirulent. Higher bacterial doses (10 6 and 10 8 CFUs) of ATCC13883 and Kpn2166 did not lead to mouse lethality, confirming that Kpn2166 is not hypervirulent in this model, despite harboring all genetic attributes of hypervirulent K. pneumoniae except a functional rmpA2 gene and allelic variants of siderophore-encoding genes.

Conclusions
Our results show that a hypermucoviscous K2-ST86 strain can be avirulent in a sepsis mouse model and that hypervirulence cannot be clearly explained by siderophore production alone. Gene rmpA2, not required for the hypermucoviscosity phenotype as previously observed (12), might be required for hypervirulent phenotype because it is a main, but not the only, difference we observed between the ST86-K2 strains. Finally, these results highlight the importance of in vivo virulence investigation to identify hypervirulent K. pneumoniae, especially in the absence of an appropriate clinical scenario.
The threat of hypervirulent K. pneumoniae acquiring carbapenem resistance is becoming a reality in Asia, especially in China, where hypervirulence prevalence among carbapenem-resistant K. pneumoniae is 7.4%-15% (5). Most resistant isolates are non-K1/K2-ST11 and produce carbapenemase KPC-2;. they result from the transfer of the pLVPK-like plasmid into ST11 classical multidrug-resistant K. pneumoniae isolates, as observed in the 2 cases reported outside China (5). Inversely, the carbapenemase-producing isolate Kpn154 results from the transformation of K2-ST86 hypervirulent K. pneumoniae by plasmid encoding carbapenemase OXA-48, the most prevalent carbapenemase in France. Similar events occurred with the KPC-2 K2-ST86 isolate recently reported in Canada (13) and a few KPC-2 and NDM K1-ST23 cases documented in China and recently in the United States and United Kingdom (5,14,15). The combination of multidrug resistance and enhanced virulence has the potential to trigger the next clinical crisis and cause severe and untreatable infections in previously healthy persons.