Role of Chromosome- and/or Plasmid-Located blaNDM on the Carbapenem Resistance and the Gene Stability in Escherichia coli.

The spread of New Delhi metallo-β-lactamase (NDM)-producing Enterobacterales represents a public health risk. The horizontal transfer of plasmids encoding the NDM gene, blaNDM, usually mediates its spread to other bacteria within the family. In contrast, Enterobacterales with a chromosome-located blaNDM is rarely reported. The phenotypic differences between chromosome- and plasmid-located carbapenemase genes are poorly understood. To determine the significance in terms of the location of drug resistance genes, we examined carbapenemase activity and stability of chromosome- and plasmid-located blaNDM. Escherichia coli M719 possessing both chromosomes- and plasmid-located blaNDM genes was used as a wild-type strain (WT) for the construction of mutants, ΔpblaNDM and ΔcblaNDM, wherein chromosome- or plasmid-located blaNDM, was knocked out, respectively. The mutant ΔpblaNDM showed lower hydrolyzing activity against imipenem and gene expression than the WT or ΔcblaNDM mutant. The MICs of both mutant strains were still above the breakpoint of imipenem and meropenem. Moreover, the chromosome-located blaNDM gene was stable for at least 30 days in the absence of antimicrobial pressure, whereas the ΔcblaNDM mutant lost blaNDM to 87% at 30 days compared to that of the initial inoculum. Organisms harboring the plasmid-located carbapenemase genes were found to provide a higher level of carbapenem resistance than those with chromosome-located genes. However, the latter organisms with chromosomal carbapenemase genes exhibited more stable carbapenem resistance than did the former ones. In summary, chromosomally located carbapenemase genes require further monitoring and more attention should be paid to them. IMPORTANCE Carbapenem-resistant Enterobacterales (CRE) carrying blaNDM have spread worldwide since they were first reported in 2009. Many studies using whole-genome sequencing have identified the genetic structures, plasmid scaffolds of blaNDM, and mechanisms of spread via horizontal transfer. Chromosome-located blaNDM and integration mechanisms from plasmids have rarely been reported, and their significance is not fully understood. Here, we showed that the chromosome-located blaNDM was associated with lower levels of carbapenem resistance and carbapenemase activity than the plasmid-located blaNDM. However, it conferred carbapenem resistance above the breakpoints and the loss of chromosome-located blaNDM was not observed in the absence of antibiotic pressure. This study suggests that CRE strains carrying chromosome-located blaNDM may persist in clinical and environmental settings for a long period even without antibiotic pressure and need to be monitored along with plasmid-located blaNDM.

OBSERVATION

Carbapenem-resistant Enterobacterales (CRE) represents a major health concern in clinical settings. Various types of carbapenemases have been identified, including NDM, IMP, VIM, KPC, and OXA-48 type. NDM-type carbapenemase originating from the Indian subcontinent has rapidly spread to different parts of the world, including Southeast Asia (1–5). The NDM gene, blaNDM, is typically located in plasmids that facilitate the spread of carbapenem resistance via horizontal gene transfer. Plasmids act as scaffolds that assemble arrays of antibiotic resistance genes, thereby generating multiple-drug-resistant phenotypes in members of the order Enterobacterales. However, few CREs have been reported to carry the blaNDM on their chromosome (6–9). Although the genetic contexts and integration mechanisms of chromosome-located carbapenemase genes have been reported occasionally, the difference in phenotype between isolates with the chromosome- and plasmid-located genes has not been extensively studied. Mathers et al. (10) identified five K. pneumoniae clinical isolates showing chromosomal integration of blaKPC. They compared the MICs of meropenem and ertapenem between these isolates and others harboring plasmid with blaKPC. They found that chromosomal genes exhibited weaker carbapenemase activity and lower expression of blaKPC. However, the genetic background, including sequence types (STs) and porin gene status, is variable among individual isolates, and the prediction of an accurate correlation between carbapenem resistance and location of the carbapenemase gene is challenging. Therefore, phenotypic assays using the same genetic background are required to evaluate the true significance in terms of carbapenem resistance between chromosomal or plasmid-located carbapenemase genes. Here, we isolated an E. coli strain possessing blaNDM-5 on plasmid as well as chromosome and used it in further experiments to explore the comparative significance of chromosome- and plasmid-located blaNDM in CRE.

Among CRE isolates previously obtained from CRE surveillance at a tertiary-care hospital in Yangon, Myanmar (3–5), E. coli isolate M719 carrying blaNDM on the plasmid and chromosome was found using S1 nuclease pulsed-field gel electrophoresis (S1-PFGE) combined with Southern hybridization targeting blaNDM. This isolate belonged to ST8453 which was originally assigned to a clinical isolate from Yangon (5). Whole-genome sequencing using the MinION system on E. coli isolate M719 identified multiple resistance genes on the chromosome and plasmid genomes (Table S1 in Supplemental File 1). The chromosome of M719 was found to include an 11.2 kbp region that contained blaNDM-5 and three other antimicrobial resistance genes against aminoglycosides, sulfonamide, and trimethoprim, which were bracketed by IS26. This region was integrated between a quinone oxidoreductase gene and the dnaB gene and shared > 99% sequence identity with that integrated into the IncFII blaNDM-5-encoding plasmid (94.6 kbp) of M719 (Fig. 1). We also found 8 bp direct repeats adjacent to the IS26 copies, indicating intermolecular transposition from pM71901 to the chromosome (Fig. 1). IS26 is commonly detected in plasmids in Enterobacterales and surrounding regions of the chromosome and blaNDM (7, 11). The chromosomal integration of drug-resistant genes and virulence genes via IS26 has frequently been reported recently (12, 13). Therefore, chromosomal integration of blaNDM via IS26-mediated transposition and/or homologous recombination from plasmids are likely to occur in various CRE organisms.

FIG 1

Comparison of the chromosomal region (11.2 kbp) containing blaNDM-5 in E. coli M719 and the blaNDM-5-located region in the pM71901 genome. The area shaded in dark gray indicates the region possessing the identical DNA sequence (with > 99% similarity) between the two genomes. The 11.2 kbp region containing blaNDM-5 is highlighted in a red box. Nucleotide letters below the sequences of the M719 chromosome represent target site duplications.

We used M719 (WT) harboring one copy of the blaNDM-5 gene in its chromosome and plasmid for the construction of mutants by deleting the plasmid- or chromosomal-located blaNDM-5 gene, yielding a plasmid-located blaNDM-5 knockout mutant (ΔpblaNDM-5) and chromosomal blaNDM-5 knockout mutant (ΔcblaNDM-5) (Fig. S1 in Supplemental File 1). The MICs of imipenem and meropenem against mutants ΔpblaNDM-5 and ΔcblaNDM-5 were lower than those against the WT strain (Fig. 2A). Imipenem showed a lower MIC value against mutant ΔpblaNDM-5 than against mutant ΔcblaNDM-5 (P < 0.05), while significant difference was not observed between meropenem MIC values against the two mutants. This observation could be related to the difference in permeability of these two beta-lactams (14), although the detailed mechanism is unclear. Consistent with the above result, the hydrolytic activity of imipenem by ΔpblaNDM-5 was significantly lower than that of the WT and ΔcblaNDM-5 mutant (P < 0.05) (Fig. 2B and C). The hydrolytic activity of ΔpblaNDM-5 mutant also exhibited a significant decrease, compared to that of the WT. The double-knockout mutant, ΔpcblaNDM-5 was also constructed, and it was found to be susceptible to both carbapenems and showed no hydrolytic activity against imipenem. A recent study demonstrated that the permeability of meropenem was substantially lower than imipenem in K. pneumoniae and E. cloacae (14). Therefore, meropenem penetrated periplasm could be hydrolyzed immediately by a certain amount of the NDM enzyme, resulting in a higher MIC value. As a result, our meropenem MIC assay may have not shown a significant difference between the ΔpblaNDM-5 and ΔcblaNDM-5 mutants. These results indicated that organisms carrying plasmid-located blaNDM exhibited higher levels of carbapenem resistance and enhanced carbapenemase activity than mutants carrying chromosomal blaNDM only. A similar phenomenon has been reported in clinical isolates carrying chromosomal blaKPC, which exhibited lower MIC and tested negative or weakly positive in phenotypic tests probably due to a single chromosomal copy number with a lower mRNA level (10). We then investigated the correlation between the copy number of the carbapenemase gene and the mRNA level of the carbapenemase gene and the MICs against carbapenems (15–17). The WT containing two copies of blaNDM-5 (chromosome- and plasmid-located) showed the highest level of carbapenem resistance and blaNDM-5 mRNA level. The transcription levels of blaNDM-5 in mutant ΔpblaNDM-5 and ΔcblaNDM-5 were 0.4-fold and 0.7-fold compared to those of the WT, respectively (Fig. 2D). These results were consistent with those of MICs of carbapenems and hydrolytic activity against imipenem (Fig. 2C), showing that the mRNA level of blaNDM-5 and carbapenemase activity are positively correlated. We also analyzed the sequences of the promoter region upstream of chromosomal and plasmid-mediated blaNDM-5. We found that excluding promoter sequences involved in the difference in transcription, the 51 bp adjacent sequence of blaNDM-5 was completely conserved and identical to the promoter region of blaNDM-1 as described previously (Fig. S2 in Supplemental File 1) (18). The blaNDM-5 mRNA levels in mutant ΔcblaNDM-5 were 1.73-fold higher than those of mutant ΔpblaNDM-5, suggesting that the former harboring plasmid with blaNDM-5 showed higher levels of blaNDM-5 transcription and carbapenem resistance than did the latter with chromosomal blaNDM-5. The increase in the mRNA level of ΔcblaNDM-5 likely reflects the plasmid copy number of pM71901, and we subsequently estimated the plasmid copy number of the blaNDM-5 gene of pM71901 in the WT strain. The plasmid copy number of pM71901 was 2.12, which is similar to the copy number of IncFII plasmids, which are recognized as low-copy-number plasmids.

FIG 2

Comparative features of E. coli M719 and its blaNDM mutants in terms of MIC to meropenem/imipenem, inactivation of imipenem, and blaNDM mRNA levels in their cultures. MICs of (A) meropenem and (B) imipenem stratified based on the location of blaNDM in Escherichia coli M719. Solid lines represent median MICs. The significance of P values was determined using the Steel-Dwass test (*, P < 0.05). (C) Hydrolysis of imipenem by the WT and knockout mutants (ΔpblaNDM and ΔcblaNDM) after 30 min incubation. Statistical comparisons were performed using the Steel-Dwass test for multiple comparisons. *, P < 0.05; ns, not significant. Error bars represent standard deviation. (D) Relative mRNA levels of blaNDM-5 as determined with qPCR. Values were normalized to the mRNA level in the WT. Statistical comparisons were performed using the Steel-Dwass test for multiple comparisons; *, P < 0.05. Error bars represent standard deviations.

In vitro and in silico experiments have demonstrated that plasmid-located drug resistance genes are frequently lost from plasmids under antibiotic-free conditions, i.e., deprived selection pressures (19–21). Therefore, we tested the stability of blaNDM in the absence of carbapenems. During the 30 days of daily passaging in the absence of meropenem, mutant ΔpblaNDM-5 retained chromosomal blaNDM-5, while mutant ΔcblaNDM-5 showed a gradual loss of the plasmid-located blaNDM-5 with a 13% reduction in the population (Table S2 in Supplemental File 1). In contrast, the loss of blaNDM-5 was not observed in the serially passaged culture in the presence of meropenem. These results suggested that the chromosome-located blaNDM-5 was stable in the absence of selective pressure of antibiotics.

In conclusion, we demonstrated unique features of plasmid/chromosome-located blaNDM-5. The plasmid-carrying blaNDM-5 exhibited higher levels of hydrolyzing activity and resistance against imipenem than did the chromosome-located gene because of its higher copy number, and frequent transmission to other bacteria via conjugation. Our results could apply to other antimicrobial resistant determinants that can be found on chromosomes and plasmids. However, it might not apply even to the same gene because the copy number depends on plasmid types and transcriptional activity should be affected by other factors than plasmid copy number. It should not also apply to the case in which a single copy of an antimicrobial-resistant determinant is sufficient to confer resistance on its bacterial host. Meanwhile, it is interesting to explore an upper threshold of β-lactamase genes’ copy number at which there is not an increase in the relative degradation of substrates of their encoding enzymes. This could be plausible because they are usually located in the periplasmic space or, for NDM-type β-lactamases, anchored to the outer membrane (22), which could potentially reach a saturation.

Although chromosomal blaNDM-5 showed essentially no conjugation ability and a relatively lower level of carbapenem resistance, it was stably carried in CRE even in the absence of antimicrobial selective pressure. The stability of blaNDM via chromosomal integration in outbreak clones or environmental isolates may pose a serious public health concern. Further studies are required to understand the biological behavior of CREs with chromosomal carbapenemase genes, which may be of epidemiological significance.