KPC-2 allelic variants in Klebsiella pneumoniae isolates resistant to ceftazidime-avibactam from Argentina: blaKPC-80, blaKPC-81, blaKPC-96 and blaKPC-97

ABSTRACT Ceftazidime-avibactam (CZA) therapy has significantly improved survival rates for patients infected by carbapenem-resistant bacteria, including KPC producers. However, resistance to CZA is a growing concern, attributed to multiple mechanisms. In this study, we characterized four clinical CZA-resistant Klebsiella pneumoniae isolates obtained between July 2019 and December 2020. These isolates expressed novel allelic variants of blaKPC-2 resulting from changes in hotspots of the mature protein, particularly in loops surrounding the active site of KPC. Notably, KPC-80 had an K269_D270insPNK mutation near the Lys270-loop, KPC-81 had a del_I173 mutation within the Ω-loop, KPC-96 showed a Y241N substitution within the Val240-loop and KPC-97 had an V277_I278insNSEAV mutation within the Lys270-loop. Three of the four isolates exhibited low-level resistance to imipenem (4 µg/mL), while all remained susceptible to meropenem. Avibactam and relebactam effectively restored carbapenem susceptibility in resistant isolates. Cloning mutant blaKPC genes into pMBLe increased imipenem MICs in recipient Escherichia coli TOP10 for blaKPC-80, blaKPC-96, and blaKPC-97 by two dilutions; again, these MICs were restored by avibactam and relebactam. Frameshift mutations disrupted ompK35 in three isolates. Additional resistance genes, including blaTEM-1, blaOXA-18 and blaOXA-1, were also identified. Interestingly, three isolates belonged to clonal complex 11 (ST258 and ST11) and one to ST629. This study highlights the emergence of CZA resistance including unique allelic variants of blaKPC-2 and impermeability. Comprehensive epidemiological surveillance and in-depth molecular studies are imperative for understanding and monitoring these complex resistance mechanisms, crucial for effective antimicrobial treatment strategies. IMPORTANCE The emergence of ceftazidime-avibactam (CZA) resistance poses a significant threat to the efficacy of this life-saving therapy against carbapenem-resistant bacteria, particularly Klebsiella pneumoniae-producing KPC enzymes. This study investigates four clinical isolates exhibiting resistance to CZA, revealing novel allelic variants of the key resistance gene, blaKPC-2. The mutations identified in hotspots surrounding the active site of KPC, such as K269_D270insPNK, del_I173, Y241N and V277_I278insNSEAV, prove the adaptability of these pathogens. Intriguingly, low-level resistance to imipenem and disruptions in porin genes were observed, emphasizing the complexity of the resistance mechanisms. Interestingly, three of four isolates belonged to clonal complex 11. This research not only sheds light on the clinical significance of CZA resistance but also shows the urgency for comprehensive surveillance and molecular studies to inform effective antimicrobial treatment strategies in the face of evolving bacterial resistance.

To ensure the consistency and accuracy of susceptibility results, the guidelines established by the Clinical and Laboratory Standards Institute were followed (18).Exceptions were the breakpoints for colistin, fosfomycin, tigecycline and ceftazi dime-avibactam for which the European Committee on Antimicrobial Susceptibility Testing Breakpoint tables for interpretation of MICs and zone diameters (https:// www.eucast.org/clinical_breakpoints)were used.

Molecular biology analysis
The bla KPC genes were initially confirmed using a multiplex PCR set up at the NRRLAR to detect epidemiologically relevant carbapenemases.The multiplex was set up for five epidemiologically relevant carbapenemases: bla OXA-48-like , bla KPC , bla NDM , bla IMP , and bla VIM (20).PCR for bla KPC was performed to confirm the nucleotide sequence by Sanger technology ABI PRISM 3100 (Applied Biosystems).GeneXpert-Xpert Carba-R (XCR) (Cepheid) was additionally used to confirm bla KPC detection.The amino acid alignment and structural visualization were performed using ESPript 3.0 and RCSB PDB, respectively (21,22).The integrity of porin genes ompK35 and ompK36 was preliminarily detected by PCR and agarose gel electrophoresis using a K. pneumoniae ATCC 13883 harboring wild-type ompK35 and ompK36 as a control strain A (23).

Biparental conjugation
Biparental conjugation assay of ceftazidime-avibactam-resistant isolates was performed on solid medium using Escherichia coli J53 (azide resistant) as acceptor strain.Transcon jugant strains were identified by conventional biochemical methods and selected on Tryptic Soy Agar plates containing 200 µg/mL of azide and 50 µg/mL of ampicillin or 10 µg/mL of ceftazidime.The horizontal transfer of the putative ceftazidime-avibactamresistant determinant was evaluated by disc diffusion and PCR.

Cloning of bla KPC allelic variants and susceptibility testing of recombinant clones
The complete bla KPC genes were amplified from whole DNA of the corresponding clinical strain by PCR using primers designed to introduce the NdeI and EcoRI restriction sites: KPC-F-NdeI (5′CATATGTCACTGTATCGCC3′) and KPC-R-EcoRI (5′GAATTCTTACTGCCCGTT3′).A proof-reading Pfu polymerase (Thermo Scientific, USA) was used in PCR reactions to avoid errors in the bla gene amplification.Amplified and purified amplicons were cloned into a pGEM-T Easy Vector (Promega, USA), and the resulting constructions were transformed into chemically competent E. coli TOP10F′ cells.The presence of the inserts and restriction sites were verified by DNA sequencing (Macrogen, South Korea).For subsequent cloning, the amplicons were digested from the original construction, and the released fragments were purified and then ligated in the NdeI and EcoRI sites of a pMBLe vector.Ligation mixtures were transformed in chemically competent E. coli TOP10F′ cells, and recombinant clones were selected in lysogeny broth (LB) agar supplemented with 20 µg/mL gentamicin.Recombinant plasmids of the selected clones were extracted and sequenced to verify the identity of bla genes and their proper insertion.
The MICs were determined for the E. coli TOP10 clones carrying pMBLe with the variants bla KPC-80 , bla KPC-81 , bla KPC-96 and bla KPC-97 along with the E. coli TOP10 wild type, E. coli TOP10 transformed with closed pMBLe, and E. coli TOP10 transformed with pMBLe with bla KPC-2 .

Whole genome sequencing and bioinformatic analysis
DNA was extracted using the QIAamp DNA Mini kit (Qiagen) following the manufactur er's instructions.The DNA concentration was determined using the Qubit 2.0 fluorometer (Thermo Fisher Scientific).The library was generated using the Nextera XT DNA Library Preparation Kit following the manufacturer's instructions.Whole genome sequencing was performed using the Illumina MiSeq platform to generate paired-end reads of 250 bp at the National Center of Genomics and Bioinformatics, ANLIS "Dr.Carlos G. Malbrán." Quality assessment of the reads was performed using FASTQC (V.0.11.5) (24), and Kraken2 (V.2.0.7-beta) was used to confirm the species (25).The reads were de novo assembled using SPAdes (3.13.0), and the assembly quality was evaluated using QUAST (V.5.0.2) (26).Genome annotation was performed using Prokka (V1.14.6) (27).The sequence type (ST) for each genome was also determined using ARIBA MLSTdb (V.2.14.6) (28).The resistance genes were determined using ARIBA Resfinder (V.2.14.6) (29), and the corresponding gene alleles were confirmed through in silico assemblies and AMRFinder Plus (V.3.8.4) (30).The analysis of the genetic environment of bla KPC was performed using TetTyper (31) and the capsular type with Kaptive 2.0 (32).The sequencing quality was optimal, and the identification of the corresponding bacterial species was consistent with MaldiTOF results (Table S1).Likewise, the assembly quality provided genome sizes and %GC content appropriate for the species under study (Table S2).

Epidemiological and phenotypic analyses
The epidemiological data and phenotypic results are detailed in Table 1.Briefly, the isolates were received at the NRRLAR between November 2019 and December 2020 from four institutions.Among these, one isolate was recovered from a patient in Neuquén Province Hospital, while the remaining three were from different hospitals in Buenos Aires City (CABA) with no apparent epidemiological connection.The average age of the patients was 58.5 years, two females and two males.Three isolates were obtained from infection sites and one from a screening procedure.Of note, the patients infected or colonized with K. pneumoniae M25752 and K. pneumoniae M25923 had previously undergone treatment with ceftazidime-avibactam (Table 1).
The presence of a carbapenemase activity was confirmed by Triton-Hodge microbio logical assay and GeneXpert-Xpert Carba-R, with all four isolates testing positive.None of the variants was detected by the colorimetric tests Blue Carba Test or CarbaNP.Addi tional tests, such as double disk synergy, mCIM, and immunochromatography, yielded variable results for the different isolates (Table 1).
All isolates demonstrated resistance to ceftazidime, ceftazidime-avibactam, cefotaxime, and aztreonam.All isolates had low-level resistance to imipenem, except for KPN-bla KPC-81 , which showed susceptibility.Regarding meropenem, all isolates were susceptible, except for KPN-bla KPC-97 with an intermediate MIC result.Avibactam or relebactam restored carbapenem and aztreonam susceptibility, as shown in Table 2. Avibactam failed to restore ceftazidime susceptibility, whereas on other cephalospor ins tested, the inhibitory effects of avibactam varied.Only fosfomycin was uniformly susceptible on all the isolates.Only KPN-bla KPC-80 was susceptible to colistin, while all strains were resistant to tigecycline, ciprofloxacin and trimethoprim-sulfamethoxazole (Table 2).

Molecular analysis of K. pneumoniae isolates harboring bla KPC variants
The nucleotide sequences of bla KPC revealed that the four isolates harbored novel allelic variants derived from bla KPC-2 , designated as bla KPC-80 , bla KPC-81 , bla KPC-96 and bla KPC-97 .The nucleotide alignment of the different variants showed that bla KPC-80 differs from bla KPC-2 by a 9-bp insertion/duplication (CCTAACAAG, nucleotide position 796 to 804), resulting in the insertion of three amino acids (ProAsnLys) between Lys269 and Asp270 (K269_D270insPNK) near the Lys270-loop.bla KPC-81 differs from bla KPC-2 by a 3-bp deletion (ATC, nucleotide position 514 to 516), leading to the deletion of one amino acid at position Ile173 (del_I173) within the Ω-loop.bla KPC-96 differs from bla KPC-2 by a single-nucleotide point mutation (TxA, nucleotide position 718), resulting in the substitution Tyr241Asn (Y241N) within the Val240-loop.bla KPC-97 differs from bla KPC-2 by a 15-bp insertion (AACAGCGAGGCCGTC, nucleotide position 829 to 844), leading to the insertion of five amino acids between Val277 and Ile278 from the KPC-2 numbering scheme (V277_I278insNSEAV), within the Lys270-loop (Fig. 1).The amino acid alignment showing the mutated region of the new allelic variants with respect to bla KPC-2 is shown in Fig. 1.
With the aim to determine whether the bla KPC allelic variants were located on mobile or conjugative plasmids, we performed biparental conjugation assays.As a result, a transconjugant was obtained only for M25399-KPN-bla KPC-80 when selecting with azide and ceftazidime.In addition, the transconjugant strain was resistant to ceftazidime-avi bactam with a MIC value of 8 dilutions higher than that of the wild-type E. coli J53 (32 µg/mL vs. 0.125 µg/mL, respectively).

Phenotypic characterization of cloned KPC allelic variants
In order to study the phenotypic effect of the KPC enzymes without the genetic background of the clinical strains, bla KPC genes were cloned and transformed in fully susceptible competent cells (E. coli TOP10) (Table 3).The presence of pMBLe carrying bla KPC-80 (K269_D270insPNK), bla KPC-96 (Y241N) and bla KPC-97 (V277_I278insNSEAV) resulted in an increase in the MICs of imipenem for the recipient E. coli TOP10, raising it by two dilutions and placing it in the intermediate category (Table 3).Conversely, there were no observed changes in the MIC when pMBLe carried bla KPC-81 (del_I173).The influence on meropenem MICs was minimal for all four constructions.As a control, pMBLe carrying bla KPC-2 was used, leading to an increase in MICs for both carbapenems.When imipenem was evaluated in the presence of both avibactam and relebactam, a significant two-dilution reduction in imipenem MIC was observed for the pMBLe constructs with bla KPC-80 , bla KPC-96 and bla KPC-97 , lowering the values back to the TOP10 baseline.
In the case of cephalosporins and monobactams, the MICs for cefotaxime, ceftazi dime, ceftaroline and aztreonam increased by a factor of 4 to 9 when compared with E. coli TOP10 transformed with the closed pMBLe.Interestingly, the MICs for these new variants in the case of ceftazidime were significantly higher than those observed in the bla KPC-2 transformant.Avibactam had the capacity to restore the MICs of all cephalospor ins and monobactams to the baseline TOP10 values, except for ceftazidime, where the MIC reduction was more modest, ranging between one and two dilutions.The weakest inhibitory effect of avibactam on ceftazidime MIC was observed for bla KPC-96 .

Whole genome analysis of the clinical isolates harboring bla KPC allelic variants
Upon further analysis of these results, bioinformatic tools such as Resfinder-ARIBA identified KPC variants by reporting nucleotide identity percentages lower than 100% compared with the sequences in the database (Table 4).The genetic environment of bla KPC-80 (K269_D270insPNK), bla KPC-81 (del_I173) and bla KPC-97 (V277_I278insNSEAV) was identified as Tn4401a, while bla KPC-96 (Y241N) was found within a Tn4401b-1 element.
In order to determine additional factors that could contribute with ceftazidime-avi bactam resistance phenotype, the integrity of the porin genes ompK35 and ompK36 was analyzed (Table 4).The strains carried mutations caused by insertions that ren dered reading frame shifts and, as a consequence, unfunctional OmpK35 proteins in all genomes, except in M25399 (bla KPC-80 ) (Table 4).ompK36, ompK37 and ompS maintained the wild-type sequence in all genomes.Additionally, the four strains harbored the wild-type PBP3 encoding fts1 gene.Furthermore, none of the isolates carried bla PER or AmpC genes (Table 4), also known to contribute with ceftazidime-avibactam resistance (6).The detailed accompanying resistome can be found in table S3.Briefly, all strains harbored bla SHV-11 and fosA, which is consistent with the K. pneumoniae species.They also carried oqxA and oqxB, sul1, and D87N mutation in DNA-gyrase.Additionally, the qacEdelta1 gene responsible for resistance to quaternary ammonium biocides was present in all four isolates.

DISCUSSION
In this study, we described the emergence of four ceftazidime-avibactam-resistant K. pneumoniae isolates carrying novel KPC variants derived from KPC-2.The clinical isolates exhibited uniform resistance to extended spectrum cephalosporins and monobactams.Among the isolates tested, only one exhibited the typical phenotype of complete susceptibility to carbapenems (M25752-bla KPC-81 ), as frequently seen in KPC variants (34).In contrast, the remaining three isolates displayed a mild level of carbapenem resistance.This resistance may be attributed to the intrinsic hydrolytic activity of the newly identified variants, as indicated by the results obtained when these variants were cloned into a susceptible strain which caused a significant elevation of imipenem MICs to intermediate values.The exception was bla KPC-81 , where the basal TOP10 imipenem MIC values did not change after cloning.Furthermore, the confirmation of the per se putative hydrolytic activity of bla KPC-80 , bla KPC-96 and bla KPC-97 came through the restoration of full susceptibility to carbapenems upon the addition of avibactam or relebactam.
The susceptibility to carbapenems, specifically meropenem and imipenem, varied among the isolates: except for KPN-bla KPC-97 , which showed intermediate susceptibil ity (MIC: 2 µg/mL) to meropenem, the other three clinical strains were susceptible to  meropenem, while only the KPN-bla KPC-81 showed susceptibility to imipenem.When we evaluated the MICs of E. coli clones harboring the bla KPC allelic variants, we observed that the clones remained susceptible to carbapenems but behaved as resistant to cephalosporins and aztreonam.Furthermore, when we evaluated the same antibiotics in the presence of avibactam, we observed a decrease in the MICs in all cases except for meropenem, whose MICs remained low even in the absence of DBO inhibitors.The same behavior was observed in E. coli expressing bla KPC-96 with ceftazidime and ceftazidime-avibactam, where the MICs differed only by one dilution.Three hotspot mutations associated with ceftazidime-avibactam resistance have been described: the omega-loop (Arg164 to Asp179), the Val240-loop (Thr237 to Thr243), and the Lys270-loop (Ala266-Glu275) (7,11,35).Each of these sites appears to exhibit a different tolerance to different types of mutations.The omega-loop exhibits flexibility in accommodating both insertions and deletions, while the Val240 loop primarily tolerates deletions and point mutations.In contrast, the Lys270-loop is primarily amenable to insertions, with a majority of these insertions corresponding to duplicated sequences (7,11,35).All the variants described in this work fall within this framework of mutations.
When we closely examined each of the new KPC variants, we found that KPC-80 and KPC-97 exhibit insertions/duplications in the Lys270 loop.Interestingly, this loop, while being the second most mutated one linked to ceftazidime-avibactam resistance among KPC variants in the literature, contains fewer mutations compared with other Class A β-lactamases (7).Notably, the same mutations found in KPC-80, derived from KPC-2 in the form of PNK duplication, were also identified in the KPC-41 mutant, originating from KPC-3.In this latter pair, it is observed that KPC-41 has a higher affinity for ceftazidime and lower inhibition by avibactam compared with KPC-3 (9).The PKN duplication has been linked to an enhanced ability to hydrolyze ceftazidime and reduced inhibition by avibactam when compared with KPC-3.In our investigations involving the bla KPC-80 clone, we also observed a parallel increase in ceftazidime MICs (Tables 2 and 3).However, it's noteworthy that in this case, the inhibitory effect of avibactam remained unchanged and effective.To substantiate this phenotypic observation, further in vitro evaluation of the kinetic parameters of the KPC-80 enzyme is required.
KPC-96 carries a single-point mutation in the Val240-loop, a crucial loop that contains residues involved in the active site and interacts with residues from the omega-loop, such as Val240, Tyr241 and Thr243 (11).Notably, it was this particular variant, when expressed in E. coli TOP10 and subjected to avibactam inhibition that exhibited the weakest response on ceftazidime (Table 3).This is the first report of a clinical isolate harboring the Y241N mutation, which until now has only been obtained in vitro in laboratory settings following exposure to ceftazidime-avibactam (36).
Finally, KPC-81 has a deletion in the omega-loop.Mutants with deletions in this site are defined as "specialized in the hydrolysis of ceftazidime" as they have been reported to be susceptible to all beta-lactams except for ceftazidime and ceftazidime-avibactam.While the precise mechanism remains not fully understood, it is believed to involve increased binding of ceftazidime (11).The bla KPC-81 isolate described here aligns with these previous findings, exhibiting resistance to cephalosporins and monobactams while maintaining susceptibility to carbapenems upon cloning into E. coli TOP10 (Table 3).Whenever there was an increase in MICs, the introduction of avibactam successfully restored them to their original baseline values (Table 3).Similar KPC allelic variants have been reported worldwide derived from KPC-3 or KPC-2.For instance, in France, KPC-28 (deletion Δ242-GT-243 derived from KPC-3, with H273Y) was reported (10), and in New York, KPC-14 (deletion Δ242-GT-243 derived from KPC-2) was reported from a clinical isolate in 2003, prior to the introduction of avibactam to clinical treatment (37).In 2020, KPC-14 and KPC-33 (both featuring D179Y mutation) were detected in a patient previously treated with ceftazidime-avibactam in Italy (38).Unfortunately, the emergence of ceftazidime-avibactam-resistant KPC mutants was accelerated with the introduction of this drug in the clinical practice (39).Still, the KPC enzyme mutation alone does not entirely explain the ceftazidime-avibactam resistance phenotype.It is known that deficiency in the OmpK35 porin also contributes to an increase in this MICs (40).In this work, we found frame shift mutations in the gene encoding for OmpK35 porin in the clinical isolates carrying KPC-81, KPC-96 and KPC-97, although the overall contribution of impermeability to the phenotype seems negligible here.
Of the four K. pneumoniae isolates harboring KPC allelic variants described here, three belonged to CC11, M25752 and M25197 to ST258, and M25923 to ST11 (Table 4).In the ST258 isolates, bla KPC was detected within the Tn4401a or b-1 genetic element, while in ST11-M25923, it was associated with Tn4401a.Historically in Argentina, the ST258 has been associated to the successful dissemination of bla KPC-2 primarily in conjunction with Tn4401a and not b1 (12) while ST11 has disseminated bla KPC in the Tn3-derived genetic elements bearing non-Tn4401 structures (41).Finally, it is well stablished that K. pneumoniae is an opportunistic microorganism known to causes infection through several virulence factors including surface antigens, fimbriae, capsule, outer membrane proteins and siderophores (42).Two of the isolates, M25399 and M25923, were found to harbor genes that encode for K10 and K39 capsular types, respectively.These capsular types have been very frequently described and well characterized as they are known to confer resistance against the bactericidal activity of antimicrobial peptides, comple ment, phagocytosis and opsonization (43,44).The other two isolates, M25752 and M25197, harbored capsular loci that has been identified and labeled as KL106, but the corresponding serotypes resulted as non-typeable because they remain uncharacterized (32).

Conclusions
In this study, we report the emergence of resistance to ceftazidime-avibactam due to the emergence of allelic variants of bla KPC-2 .These bla KPC alleles harbored muta tions that were in hotspots of the mature protein.This underscores the importance of both epidemiological surveillance and molecular investigations in monitoring and comprehending the diverse mechanisms of resistance that pose challenges to effective antimicrobial treatments.Moreover, it is essential to exercise caution and judiciously use of ceftazidime-avibactam in the clinical practice in order to prevent the emergence of resistance.Preserving the effectiveness of this antibiotic is crucial to curb the spread of resistance.

FIG 1
FIG 1 Amino acid alignment of KPC-2 with the novel allelic variants.(A) The amino acid alignment was obtained with ESPript program.The image shows the α-helices and η-helices marked with loops; the β-strands are indicated with black arrows.The color-labeled lines indicate the position of the omega-loops (violet), Val240-loop (green), and Lys270-loop (yellow).(B) Fragment of a representative view of the KPC-2 fold (PDB 4ZBE, https://www.rcsb.org/3d-view/4ZBE)complexed with avibactam (bars) with colored and labeled loops Ω-loop, Val240, and Lys270 near the active site (33).

TABLE 2
Susceptibility testing of K. pneumoniae clinical isolates a

TABLE 3
Susceptibility testing of E. coli TOP10 cells transformed with mutant-KPC genes a a AVI, avibactam; REL, relebactam.

TABLE 4
Bioinformatics data analysis of K. pneumoniae isolates