A Ceftazidime-Avibactam-Resistant and Carbapenem-Susceptible Klebsiella pneumoniae Strain Harboring blaKPC-14 Isolated in New York City

KPC is currently the most common carbapenemase identified in the United States. More than 40 KPC variants have been described, of which KPC-2 and KPC-3 are the most frequent clinical variants. However, our understanding of the genetic structures and β-lactam resistance profiles of other novel KPC variants remains incomplete. Here, we report a novel blaKPC variant (blaKPC-14) and the complete genome sequence of blaKPC-14-harboring K. pneumoniae strain BK13048, which is susceptible to carbapenems but resistant to ceftazidime-avibactam. To the best of our knowledge, this is one of the earliest KPC-producing K. pneumoniae strains exhibiting resistance to ceftazidime-avibactam.

resistant Gram negative bacteria. Among the carbapenemases, Klebsiella pneumoniae carbapenemase (KPC), an Ambler class A serine ␤-lactamase, is particularly problematic, with major outbreaks in the northeastern United States, followed by its spread throughout the United States and worldwide (1). As a novel combination of a ␤-lactam and a ␤-lactamase inhibitor, ceftazidime-avibactam was highly active against KPC-producing bacteria. However, resistance to ceftazidime-avibactam has also been reported in patients who were treated with this combination, primarily due to amino acid substitutions in the KPC ␤-lactamase (2)(3)(4). In this study, we describe a novel KPC variant, KPC-14, isolated from K. pneumoniae strain BK13048, collected in 2003 from a New York City (NYC) patient. Surprisingly, this strain was both susceptible to carbapenems and resistant to ceftazidime-avibactam, a result indicating that this resistant KPC variant existed prior to 2015, when ceftazidime-avibactam was introduced.
Strain BK13048 was identified as a part of retrospective study screening of extended-spectrum cephalosporins and carbapenem-resistant K. pneumoniae from our archived strain collection. A molecular-beacon-based allelic discrimination real-time PCR assay (5) showed that strain BK13048 harbored a bla KPC-6-like variant. PCR and Sanger sequencing of the full-length bla KPC gene revealed a novel bla KPC variant, bla  . Nucleotide alignment of different bla KPC variants showed that bla KPC-14 differs from bla KPC-2 by a 6-bp deletion (nucleotide positions 721 to 726), resulting in a 2-amino-acid deletion at Ambler positions 242Gly and 243Thr. KPC-28 has the same 242Gly and 243Thr deletion, but an additional His274Tyr substitution distinguishes this variant from KPC-14 (6).
The MIC results from BK13048 showed an unusual profile: susceptibility to carbapenems but resistance to ceftazidime-avibactam. To investigate this finding, the fulllength bla KPC-2 , bla KPC-3 , and bla KPC-14 genes and the same promoter sequences were cloned into pET28a vectors, followed by electroporation into Escherichia coli DH10B cells (Invitrogen). Susceptibility testing of the bla KPC-14 , bla KPC-2 , and bla KPC-3 E. coli DH10B constructs showed that the bla KPC-14 construct demonstrated a ceftazidimeavibactam MIC of Ͼ16 g/ml, which is at least 64-fold higher than that of the bla KPC-2 or bla KPC-3 construct (MICs, Յ0.25 g/ml) (Table 1). Similarly, the MIC of ceftazidime was much higher for the bla KPC-14 plasmid construct (256 g/ml) than for the cloned bla KPC-2 (4 g/ml) or bla KPC-3 (16 g/ml) gene.
In contrast, the MICs of the different carbapenems (ertapenem, meropenem, and imipenem) for the bla KPC-14 construct were 8-to 32-fold lower than the MICs for the bla KPC-2 or bla KPC-3 construct ( Table 1). The susceptibility testing results presented above were consistent with a previous study by Oueslati et al. testing KPC-14 and KPC-28 using a different plasmid vector (pTOPO) (6). Those results demonstrated that the 242Gly and 243Thr amino acid deletions in KPC-14 decreased carbapenem activity but increased potency against ceftazidime and ceftazidime-avibactam (Table 1), and ceftazidime-avibactam resistance is likely due to increased activity against ceftazidime rather than reduced inhibition against avibactam. We further characterized and compared the kinetic parameters of KPC-14 and KPC-2. In brief, the sequences without the signal peptide (from bla KPC-14 and bla KPC-2 ) were obtained by PCR amplification using primers NdeI-KPC-2-F(30 -293) (5=-ACGCATATGG CGGAACCATTCGCTAAAC-3=) and Xhol-KPC-2-R-STOPdel (5=-TAACTCGAGCTGCCCGTTG ACGCCCAAT-3=), followed by insertion into plasmid pET28a in E. coli DH10B (Invitrogen). The KPC enzymes were then purified, and the steady-state kinetic parameters were determined as described previously (6, 7). The results showed that KPC-14 has a higher catalytic efficiency of ceftazidime and cefepime, but a lower hydrolysis activity of imipenem, than KPC-2 ( Table 2). In addition, no meropenem hydrolysis could be detected with purified KPC-14 under current conditions (measurement made over 5 min). The hydrolytic profile of KPC-14 was similar to that in the previous report from Oueslati et al. (6) and was consistent with the MIC observations presented above. Moreover, a previous experiment determining the 50% inhibitory concentrations (IC 50 ) of ␤-lactamase inhibitors also suggested that the 2-amino-acid 242Gly and 243Thr deletion had no impact on the inhibition properties of avibactam (6).
To better understand the genetic structure associated with the bla KPC-14 gene, comprehensive whole-genome sequencing was performed using a combination of the Oxford Nanopore MinION and Illumina HiSeq platforms, followed by hybrid assembly using Unicycler (8). The complete sequencing of BK13048 showed that it contains a 5,213,293-bp chromosome with an average GϩC content of 57.6% and harbors 5,311 predicted open reading frames. In addition, it contains seven plasmids ranging from 5 kbp to 82 kbp. In silico multilocus sequencing typing (MLST) revealed that BK13048 belongs to sequence type (ST) 16 (9), which has been reported to cause nosocomial infections worldwide and is associated with bla NDM-1 -encoded carbapenemase and the presence of the extended-spectrum beta-lactamase (ESBL) gene bla CTX-M-15 (10). Analysis of acquired antimicrobial resistance (11) identified 14 antimicrobial resistance genes encoding resistance to ␤-lactams, aminoglycosides, fluoroquinolones, fosfomycin, sulfonamide, and trimethoprim (Table 3). In addition, in silico plasmid replicon typing (12) indicated that the seven plasmids belong to IncA/C, F, M, N, R, ColRNAI, and a novel incompatibility group ( Table 3).
The bla KPC-14 gene is located on an IncN plasmid, pBK13048_KPC14 (Table 1). pBK13048_KPC14 is 50,635 bp long with an average GϩC content of 53% and carries bla KPC-14 on the Tn4401b transposon (Fig. 1). Full plasmid sequence BLAST against NCBI GenBank (http://blast.ncbi.nlm.nih.gov/Blast.cgi) showed that pBK13048_KPC14 is highly similar to plasmid pKm38_N from Klebsiella oxytoca, which was isolated in 1997  (13), with 100% query coverage and overall 99.98% sequence identity (Fig. 1). In addition, pBK13048_KPC14 showed 94% query coverage and overall 99.97% sequence identity to one of the first sequenced bla KPC -harboring IncN plasmids, plasmid 12, isolated from NYC in 2005 ( Fig. 1) (14). In agreement with the structure of other IncN plasmids (13,15), pBK13048_KPC14 contains a 2-kb acquired region integrated downstream of uvp1 and harbors dfrA14, encoding trimethoprim resistance. In addition, pBK13048_KPC14 contains a second acquired region downstream of the nuc gene and carries bla KPC-14 (Fig. 1). This highlights the important role played by IncN plasmids in the spread of bla KPC during the early years of the carbapenem resistance epidemic. In silico IncN plasmid MLST showed that pKm38_N (isolated in 1997), pBK13048_KPC14 (2003), and p12 (2005) all belong to ST6 (repN-traJ-korA, allele profile 2-4-2), which is different from the sequence type harboring the bla KPC-28 -containing IncN plasmid pWI2-KPC28 (ST15, allele profile 7-6-3) from E. coli. Even though both KPC-14 and KPC-28 have the same 242Gly and 243Thr amino acid deletions, their genomic history suggests that pBK13048_KPC14 and pWI2-KPC28, as well as the bla KPC-14 and bla KPC-28 genes, likely evolved independently on different IncN plasmid backgrounds. A recent study from Italy described the emergence of two ceftazidime-avibactamresistant subpopulations of K. pneumoniae ST1685 (unrelated to the ST16 of BK14038), carrying KPC-14 and KPC-31 (Asp179Tyr substitution within the KPC ⍀-loop), in a patient following prolonged ceftazidime-avibactam treatment (16). Our study also suggested that the ceftazidime-avibactam-resistant KPC variants, e.g., KPC-14, could exist even without ceftazidime-avibactam exposure. These KPC variants, with reduced carbapenem hydrolytic capacities, raise a challenge for phenotypic and genotypic carbapenemase detection tests, since some of these assays may classify KPC-14 strains as carbapenemase producers (6,16). Consequently, molecular testing followed by a phenotypic carbapenemase activity assay has been proposed to detect and differ- entiate KPC variants associated with carbapenem susceptibility and ceftazidimeavibactam resistance (16). Taken together, we identified, completely sequenced, and characterized a novel bla KPC variant from K. pneumoniae BK13048, designated bla  , that revealed an unexpected resistance to ceftazidime-avibactam. Comprehensive sequence analysis and assembly using both the Illumina and Oxford Nanopore platforms revealed the genetic changes in bla KPC-14 and its plasmid structure. In contrast to other ceftazidimeavibactam-resistant bla KPC variants, the bla KPC-14 gene was not under ceftazidimeavibactam selection pressure, as evidenced by the fact that the isolation of BK13048 predated the U.S. introduction of this novel ␤-lactam and ␤-lactamase inhibitor combination in 2015.
Accession number(s). The complete nucleotide sequence of strain BK13048 has been deposited in GenBank as accession no. CP045015 to CP045022.

ACKNOWLEDGMENTS
This study was supported by a grant from the National Institutes of Health (grant R01AI090155 to B.N.K.). This work was also supported by grants R01AI100560, R01AI063517, and R01AI072219 (to R.A.B.) from the National Institutes of Health and by funds and/or facilities provided by the Cleveland Department of Veterans Affairs, the Veterans Affairs Merit Review Program, and the Geriatric Research Education and Clinical Center VISN 10 (to R.A.B.).