Beneficial Chromosomal Integration of the Genes for CTX-M Extended-Spectrum β-Lactamase in Klebsiella pneumoniae for Stable Propagation

Dominant F-type plasmids harboring the gene have been pointed out to be responsible for the dissemination of the CTX-M extended-spectrum-β-lactamase (ESBL)-producing K. pneumoniae. Recently, the emergence of K. pneumoniae isolates with the blaCTX-M gene in their chromosomes has been reported occasionally worldwide. Such a chromosomal location of the resistance gene could be beneficial for stable propagation, as was the Acinetobacter baumannii ST191 harboring chromosomal blaOXA-23 that is endemic to South Korea. Through the present study, particular clones were identified as having built-in resistance genes in their chromosomes, and the chromosomal integration events were tracked by assessing their genomes. The cefotaxime-resistant K. pneumoniae clones of this study were particularized as results of the fastidiousness for plasmids to acquire the blaCTX-M gene for securing the diversity and of the chromosomal addiction of the blaCTX-M gene for ensuring propagation.

T he acquired CTX-M-type extended-spectrum ␤-lactamases (ESBLs) belonging to the class A ␤-lactamases are grouped into five groups, 1, 2, 8, 9, and 25, by amino acid sequence similarity. Members within the same group share Ͼ94% identity and the members belonging to distinct groups share Յ90% identity (1). The five groups had been shown to originate from an intrinsic ␤-lactamase gene of different species of Kluyvera, i.e., both groups 1 and 2 from Kluyvera ascorbata (2,3), groups 8 and 9 from Kluyvera georgiana (4,5), and group 25, which remains to be identified but probably is from another member of Kluyvera. The insertion sequences (ISs), mostly ISEcp1 and, less frequently, ISCR1, hijacked bla CTX-M from the chromosome of the progenitor Kluyvera spp. and recruited it into a plasmid, which is then transferred to clinically relevant enterobacterial isolates. The ISs are recognized upstream of the bla CTX-M gene and provide portable promoter sequences stronger than the natural promoter sequences (6).
After the emergence of CTX-M ESBL-producing Escherichia coli in clinical settings in 1989 (7), CTX-M ESBL-producing Enterobacterales spread rapidly in the world, and their high prevalence is a grave concern in clinical settings, as the treatment options for patients infected by these pathogens are limited (8,9). Dominance of the bla CTX-M- 14 and bla CTX-M-15 genes in Enterobacterales clinical strains is achieved by their association with the mobile transposition element ISEcp1 and its location in prevalent F-type plasmids (10,11). Such conjugative plasmids, often carrying antimicrobial resistance genes, maintain their occupancy in bacterial populations by using plasmid addiction systems, such as the toxin-antitoxin (TA) systems that kill the plasmid-free daughter cell through a stable toxin and an unstable antitoxin (12). In addition, clonal expansion of a disseminated bacterial clone supports the successful spread of CTX-M ESBL-producing Enterobacterales. In contrast to the obvious dominance of the E. coli clone sequence type 131 (ST131) among the CTX-M ESBL producers (8), the bla CTX-M ESBL gene-carrying Klebsiella pneumoniae is known to be devoid of clonality (13,14). The emergence of K. pneumoniae isolates carrying the bla CTX-M gene in their chromosomes, devoid of clonality, were reported rarely from the beginning of the 2010s (15)(16)(17)(18).
From national antimicrobial resistance surveillance, designed as a cohort study for entire episodes of K. pneumoniae bloodstream infections occurring in a year in six general hospitals in South Korea, we were able to collect a total of 572 K. pneumoniae blood isolates, including 164 cefotaxime-nonsusceptible isolates (19). Of the cefotaxime-nonsusceptible K. pneumoniae isolates, 81.7% (134/164) harbored the bla CTX-M ESBL gene belonging either to group 1 or to group 9 (20). As shown in Fig. 1, certain STs, i.e., ST307, ST789, ST11, ST48, ST15, ST392, and ST14, have absolute high rates of cefotaxime resistance of over 78%. Particular STs also favor harboring the bla CTX-M gene. For instance, the cefotaxime-nonsusceptible ST48, ST789, ST392, and ST147, together with ST307, with one exception, harbored the group 1 bla CTX-M gene, while ST17 carried only the group 9 bla CTX-M gene. In the case of ST11 isolates, half of the cefotaxime-nonsusceptible isolates harbored the group 1 bla CTX-M gene, and the other half had the group 9 bla CTX-M gene. The phenomenon could be derived from the preferential acquisition of specific plasmids carrying a particular bla CTX-M gene or the clonal dissemination together with the expansion of the clones possessing the residential bla CTX-M gene in the chromosome. To determine how the clonality of cefotaxime resistance was achieved, bla CTX-M ESBL gene-carrying K. pneumoniae blood isolates from the cohort study were entirely sequenced, and a comparative analysis was carried out.
The intrinsic bla SHV gene was extracted, and the translated sequences were used to subtype the allele. A total of 9 subtypes were identified, and one isolate lost the gene through the interruption of IS26. The subtypes of SHV ␤-lactamases 28 (n ϭ 36), 11 (n ϭ 34), and 1 (n ϭ 28) were prevalent, and SHV-28 ESBL was identified in particular clones of ST14, ST15, ST4877, and ST307 ( Fig. 2; also see Fig. S1 in the supplemental material). The alleles of intrinsic SHV had fewer than 5 mismatches among 286 amino acids, with an amino acid identity of Ͼ98.2%. The molecular phylogeny from the multiple alignments of SHV alleles was poor, and no clear correspondence to the phylogenetic tree of STs (Fig. S1) or to that of the core genome ( Fig. 2) was observed.
A total of 29 isolates harbored CTX-M-15 coding genes in their chromosome; seven of these isolates harbored an extra bla CTX-M-15 gene in a plasmid, and one ST48 isolate carried a plasmid harboring the bla CTX-M-14 gene (Fig. 2). Characteristically, the chromosomes of all 16 ST48 isolates and the ST392-like isolate had two copies of the gene. The 16 ST48 isolates with two copies of the bla CTX-M-15 gene were closely related to each other, but no single strain was supported by the core genome multilocus sequence typing (cgMLST) (Fig. S2). Two isolates belonging to ST147 and ST307 carried two plasmids harboring the bla CTX-M-15 gene, and one ST11 isolate possessed both the bla CTX-M-15 gene-carrying plasmid and the bla CTX-M-14 gene-carrying plasmid.
Plasmids carrying the bla CTX-M gene. (i) Incompatibility types of the bla CTX-M gene-carrying plasmids. The characteristics and gene contents of representative plasmids are schematically presented in Fig. 3 and Table 1. Among the plasmids carrying the group 1 bla CTX-M genes, the dominant FIB:FII-type plasmids were observed to be diverse in terms of size and genetic components. However, the mosaic plasmids of FIB and FII carried characteristic gene contents of those plasmids, including clustered antimicrobial genes bracketed by varied ISs and heavy-metal resistance gene clusters in the FIB plasmids and the tra-type conjugative elements in FII plasmids. The FIB:FII plasmids harboring the bla CTX-M-14 gene were observed to have more mosaicism than those carrying the bla CTX-M-15 gene, as the length was ca. 100 kb longer, and more genetic elements unrelated to FIB:FII were identified. The plasmids of infrequent incompatibility types shared only the bla CTX-M gene and the vicinity with other plasmids. ISs were gathered near the genes for antimicrobial resistance, allowing the feasible mobility of the gene. The FIB:FII plasmids carrying the bla CTX-M gene in ST307 and ST463 isolates possessed the GNAT-related TacTA, while those in other STs have both HigBA and STM4031 (Fig. 2).
(ii) Comparison between the bla CTX-M gene-carrying plasmids. Pairwise comparison of the plasmids for the percent coverage of the sequences having Ͼ99.5% nucleic acid identity presented high percent coverage association by incompatibility type and by plasmid MLST (pMLST) of the plasmid (Fig. 4). The relatively high percent coverage of intertypes of plasmids was observed through mosaicism. For instance, the FIA:FIB plasmid has higher coverage than the FIB and FIB:FII plasmids. Curiously high percent coverage grouping by the ST of the host bacteria was observed among the FIB and R plasmids, even in an incompatibility type and a pMLST. The prevalent bla CTX-M-14 gene-carrying FIB:FII plasmids were likely categorized into two groups: those harbored (iii) Conjugation efficiency differed by the plasmid and the bacterial host. The transfer efficiency of plasmids belonging to representative incompatibility types was determined by liquid mating using the ST307, ST375, and ST17 recipients. The FIA: R⍀bla CTX-M-15 plasmid transferred effectively to ST17 and one of the two ST307 isolates with transfer frequencies of 3.0 ϫ 10 Ϫ5 to 1.3 ϫ 10 Ϫ4 and 3.4 ϫ 10 Ϫ6 to 1.1 ϫ 10 Ϫ5 , respectively, while the FIB:FII⍀bla CTX-M-14 plasmid transferred to the other ST307 isolate with a frequency of 5.1 ϫ 10 Ϫ6 to 5.2 ϫ 10 Ϫ5 ( Table 2) (Table S1). Intriguingly, the chromosomal sequences were similar to the 193,678-bp FIB:FII plasmid in the ST392-like isolate.
The integration units associated with the bla CTX-M gene and the targeting locus. The integration was mediated either by ISEcp1 (n ϭ 42) or by IS26 (n ϭ 3). Two integration events per chromosome, both mediated by ISEcp1, were observed in all 16 ST48 isolates, and single integration by ISEcp1 was observed in seven of the eight ST307 isolates and one each of the ST101, ST14, and ST15 isolates. Chromosomal integration by IS26 was found in one of the eight ST307 isolates and one each of the ST392 and ST392-like isolates. All of the targeted sites of integration were in the so-called core genome of K. pneumoniae, including the 16S rRNA and the coding sequences of OmpK35; except for the case found in ST15, the ISEcp1-bla CTX-M-15 -orf477 unit transposon was integrated into a 42,544-bp prophage bracketed by a complete 64-bp attL-attR sequence. The sequential order and the integration of the prophage carrying the bla CTX-M-15 unit transposon or that of the bla CTX-M-15 unit transposon into the preintegrated prophage were debatable. The integrations by ISEcp1 generated 5-bp direct repeats and were 2,971 to 29,048 bp in length, and those by IS26 were bracketed by 8-bp direct repeats and were 16,570 to 196,572 bp in length (Table S1). Comparison of the chromosomal integration units with plasmids for percent coverage of sequences having a nucleic acid identity of Ͼ99.5%, except for the ISEcp1-bla CTX-M-15 -orf477 unit transposon, showed that the integration units were more covered by the FIB (F-:A-:B-) and FIB:FII (K7:A-:B-) plasmids hosted by ST307 (Fig. 5).
Referring to the reference genome of K. pneumoniae (GenBank accession number NC_016845.1), integration occurred all over the chromosome, except between ca. 2 Mb and 3 Mb from the dnaA gene at the chromosomal replication origin (Fig. 6A) unit transposon from the primary integration unit found in ST48. For the second group, the primary integration identically targeted KPHS_45800 of the reference genome, and the ensuing integration was directed to KPHS_51830 or to KPHS_41460. The primary integration units interrupting KPHS_45800 had diverse structures, indicating indepen-  dent events of each isolate. The second integration presumably had preferential sequences of the ISEcp1-bla CTX-M-15 -orf477 unit transposon, and the sequence logo from the upstream and downstream sequences of the ISEcp1-mediated integration sites presented a consensus of the 5-bp AT-rich sequences upstream from the direct repeats, while the sequences further upstream and downstream did not (Fig. 6B) (21). Promoter sequences of the bla CTX-M gene. Promoter sequences of the bla CTX-M gene were provided from the upstream ISs (Fig. S3). All group 1 bla CTX-M genes had an ISEcp1 copy 48 bp upstream from the gene except for the bla CTX-M-3 gene of the longest region, 124 bp. The 76-bp elongated region indicates that the bla CTX-M-3 genecapturing event was independent from those of the other group 1 bla CTX-M genes. In the case of the group 9 bla CTX-M gene, the ISEcp1 copy was located 45 bp upstream from the bla CTX-M-14 gene, and ISCR1 was found 115 bp upstream from the bla CTX-M-9 gene. The promoter sequences were identical if they were given by the upstream ISEcp1 copy.

DISCUSSION
Third-generation cephalosporins are widely used in clinical settings to treat patients with K. pneumoniae bloodstream infections, and the increasing rate of cephalosporin resistance leads to increased use of carbapenems, encouraging the emerging carba- penem-resistant Enterobacterales. Currently, resistance to third-generation cephalosporins is mainly due to the acquisition of ESBL genes, and the genes mainly belong to the bla CTX-M type. The dissemination of the gene has mostly been favored due to its location in incompatibility F-type plasmids, which are widely distributed in E. coli and K. pneumoniae (8) and are freely acquired and lost by the bacterial host as the antimicrobial environment changes. The chromosomal location of the resistance determinants has taken the situation to a new level of steady spread of resistance regardless of the habitat of the bacterial host, and particular attention needs to be paid to the widespread clinical isolates harboring the chromosomal bla CTX-M gene.
The study was conceived through the observation of the resistance isolates belonging to some STs and the almost absolute prevalence of CTX-M-15 and CTX-M-14. Antimicrobial resistance in the dominant clone ST25 is indeed meager, and the ST307, ST789, ST11, and ST48 clones that make up no more than 4.5% of total K. pneumoniae blood isolates presented high rates of cefotaxime resistance between 78% and 100% (20). Even though an extra consideration for the presence of clonally related isolates is needed, the disproportional groups of the bla CTX-M gene carried by each ST isolate were caused by, at least partially, the preferred plasmid type carrying the gene. It was likely that the particular clone has a preferential plasmid type and vice versa.
The base pair distance between the right end of the inverted repeat for ISEcp1 and the start codon of the bla CTX-M gene could give a brief point of comparison for the international genetic contexts. The 48-bp distance for the group 1 bla CTX-M genes found in the present study was frequently identified in other parts of the world, i.e., France, India, and China (22)(23)(24). However, the 124-bp distance for the bla CTX-M-3 gene and the 45-bp distance for the bla CTX-M-14 gene seemed unique to this study, and no identical sequence was found even from the nucleotide collection of the GenBank database. The plasmids carrying the bla CTX-M gene mostly belonged to the incompatibility F-type as a mosaic FIB:FII. Among the FIB:FII plasmids, those in the isolates belonging to the most prevalent CTX-M ESBL-producing ST307 seemed discrete in terms of its TA systems. The GNAT-related toxin, first identified in Salmonella, is known to inhibit translation and arrest further growth of the bacterial host (25). The rare incompatibility types were R, which is also famous as an antimicrobial resistance-associated plasmid, I1, carrying the group 1 bla CTX-M type, and the N, A/C2, and B/O/K/Z types, harboring the group 9 bla CTX-M gene. Compared to the previous reports (10,13), the plasmid types became much more disproportionate. Among the 115 CTX-M ESBL-producing K. pneumoniae blood isolates in this study, more than a quarter of the isolates harbored the bla CTX-M gene in their chromosomes. All of the chromosomal bla CTX-M genes were subtype 15 and were harbored by K. pneumoniae hosts mostly belonging to ST48 and ST307, although the disproportionate distribution of bla CTX-M gene-harboring clones was considered. ST307 and, less dominantly, ST48 are globally notorious K. pneumoniae carbapenemase (KPC)-producing clones, and both have been reported as KPC producers in South Korea (26).
More than a quarter of the ST307 isolates harbored the chromosomal bla CTX-M-15 gene. The gene was included in varied integration units in terms of length, which differed by the 3= region of the unit transposon of ISEcp1-bla CTX-M-15 -orf477. The integration units found in ST307 targeted diverse loci in the chromosome, emphasizing the genome plasticity of the notorious clone. One of the isolates harbored the gene in an integration unit bracketed by a pair of IS26 elements of a direction of the KHPS_51550 K ϩ -transporting ATPase-coding sequence. Interestingly, the other seven ST307 isolates had one copy of IS26 interrupting KHPS_51550, disclosing the IS26anchored chromosomal integration of the bla CTX-M-15 gene-including segment in a plasmid. In this case, the ISEcp1 copy upstream from the bla CTX-M-15 gene was truncated, and the lost mobility was replaced with the IS26 copy. Two other IS26-mediated integration cases in ST392 and ST392-like clones also included the truncated ISEcp1 copy upstream from the bla CTX-M-15 gene. Integration of the entire FIB:FII plasmid in the ST392 chromosome and that of the duplicated composite transposon in the ST392-like chromosome was bracketed by a pair of IS26 elements in a particular direction. In the latter case, a supposed rolling-circle tandem amplification by IS26 resulted in the double copy of the bla CTX-M-15 gene in a chromosome.
More than half of the isolates harboring the residential bla CTX-M gene belonged to ST48, which had two copies of the gene at a distant locus in the chromosome. The ST48 isolates were clonally distinct, and the chromosomal integration events seemed to be independent. The integration was always mediated by ISEcp1, and the integration hot spots were the zinc transporter zupT gene and the genes encoding phosphoenolpyruvate-dependent phosphotransferase II and putative glycoside hydrolase. The zupT gene was presumed to be a primary integration site for an ISEcp1 unit transposon from an R plasmid, because the unit size is longer than 9 kb and the genetic contexts included plasmid-associated components. At the other hot spot, the identical 2,971-bp unit transposon of ISEcp1-bla CTX-M-15 -orf477 was identified as prevailing in the second integration, probably from the primary integration unit. In one exceptional ST48 case, secondary integration was observed at the other gene as a 3=-terminal truncated form of the unit transposon of ISEcp1-bla CTX-M-15 -orf477.
The targeted integration locus had a peculiar consensus AT-rich sequence 5 bp upstream from the direct repeats. This integration site was preferred not only by the major clones but also by the minor ST101, ST14, and ST15 clones. An ST15 isolate was infected by a prophage, and the chromosomal integration of the unit transposon ISEcp1-bla CTX-M-15 -orf477 targeted the AT-rich region within the prophage. Based on the rareness of integration mediated by IS26, complex contexts of the IS26-associated integration unit and identified truncated ISEcp1 copy made it possible to form a reasonable hypothesis: IS26 would be the second-best choice, following ISEcp1, for the chromosomal integration of the bla CTX-M gene.
The cefotaxime-resistant K. pneumoniae clones of this study were particularized because of the fastidiousness for plasmids to acquire the bla CTX-M gene and of the chromosomal accumulation of the bla CTX-M gene. The fit clones in clinical settings may have performed a consequent dissemination through acquired resistance, during which the present population of K. pneumoniae blood isolates was likely being made. The diverse genetic contexts bracketing the bla CTX-M-15 gene, chromosomal locus of integration, and the hospitals from which the isolates were recovered provided enough evidence to make an assumption. Thus far, to the best of our knowledge, nationwide dissemination of K. pneumoniae clones with the residential bla CTX-M gene has never been reported, and we consider it important to keep a close watch on its status.
This study showed an evolutionary path for antimicrobial resistance in clinical isolates to sustain their life while surrounded by an abundance of antimicrobials. The evolutionary strategy could be summed up in two parts, securing diversity and ensuring propagation. For diversification, different types of plasmids were equipped for enough trials of the bacterial host, and various accessible mobile genetic elements were used to acquire the gene. Such a plan is important for the bacterial host to avoid being at a standstill. To ensure stable propagation, the antimicrobial resistance determinant was appointed as a residential gene in the chromosome. The bacterial host then could better deal with encountering the life-threatening antimicrobials.

MATERIALS AND METHODS
Isolates used in the study. Among the 134 isolates collected from the cohort study, a total of 115 isolates harboring the bla CTX-M genes were recoverable in good shape, and those 115 isolates were used for the study.
Whole-genome sequencing. From the 115 K. pneumoniae isolates, genomic DNA was extracted with the GenElute bacterial genomic DNA kit (Sigma-Aldrich, St. Louis, MO). The entire genomes were sequenced using both Illumina and Nanopore technologies. Libraries were prepared for Illumina using both the Swift 2S Turbo DNA library kit (Swift Biosciences, Ann Arbor, MI) and Swift 2S Turbo combinatorial dual indexing primer kit (Swift Biosciences) and for Nanopore using the ligation sequencing kit (Oxford Nanopore, Oxford, UK). Reads were assembled using Spades (ver. 3.11.1) (27). Annotation of the complete sequences was carried out using prokka 1.13.7 (https://github.com/tseemann/prokka) (28).
Phylogenetic analysis. A total of 16 housekeeping proteins of the 115 K. pneumoniae core genomes were used to produce a multiple alignment with muscle v3.8 (29). The phylogeny was analyzed using PhyML v3.0 with the Whelan and Goldman matrix, and a gamma correction was made. To ensure the robustness of the topology, 100 bootstraps were calculated for the concatenated sequences. To root the phylogenetic tree, the genome of Klebsiella oxytoca CAV1374 (NZ_CP011636.1) was used.
Plasmid transfer by bacterial conjugation. For bacterial conjugation, spontaneous mutants resistant to both nalidixic acid and sodium azide were generated from drug-susceptible K. pneumoniae clinical isolates B16KP0003 of ST17, F16KP0005 of ST375, and E16KP0152 and C16KP0023 of ST307, which are devoid of any obvious plasmid by electrophoresis, for recipients. K. pneumoniae ST789 C17KP0019/ FIA:R⍀bla CTX-M-15 and ST432 E16KP0235/FIB:FII⍀bla CTX-M-14 were selected as donors. Equal amounts of exponential cultures of the donor and recipient isolates were mixed, incubated in Mueller-Hinton broth devoid of any drug for 12 h, and spread on brain heart infusion agar (Difco Laboratories) containing nalidixic acid (30 mg/liter), sodium azide (100 mg/liter), and cefotaxime (10 mg/liter). Each colony was confirmed by PCR, and the plasmid transfer frequency was calculated as the number of transconjugants per donor. The experiments were performed in duplicate and repeated at least three times. Data availability. The genomes of the 115 K. pneumoniae isolates were deposited in the GenBank nucleotide database under accession numbers CP052136-CP052744 (see Table S2 in the supplemental material) and under BioProject PRJNA625837.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only. FIG S1, TIF