We included all nonduplicate IMI-producing and NmcA-producing isolates showing antimicrobial resistance received at the French National Reference Center for Antimicrobial resistance (F-NRC) during 2012–2022 (n = 112) (Appendix 1 Table 1). Mass spectrometry showed that all strains belonged to the ECC. Since 2014, each year, 3–20 IMI/NmcA producers were identified, representing 0.03%–0.91% of all carbapenemase-producing Enterobacterales analyzed at F-NRC. No IMI/NmcA producers were found before 2014. (Appendix 2 Figure 1).

Disc diffusion antimicrobial susceptibility testing revealed resistance to third-generation cephalosporins for 1 strain (257D9, overexpression of ampC confirmed with CLOXA agar) of the 112 tested. We determined MICs for last-resort antibiotics against highly resistant bacteria on 30 IMI/NmcA producers belonging to several sequence types (STs) (Appendix 1 Table 2). Relebactam restored imipenem activity for 67% of the strains and vaborbactam restored susceptibility to meropenem for all strains with lower MICs than imipenem/relebactam. Then, 37% of the tested strains were susceptible to colistin. All 30 IMI/NmcA producers remained susceptible to cefepime, cefiderocol, and ceftazidime/avibactam.

We performed WGS on the 112 IMI/NmcA producers and identified 74 IMI-1 producers (Appendix 2 Figures 1, 2). Of those, 44 IMI-1–producing ECC were involved in an outbreak in Mayotte and La Réunion islands.

We confirmed ECC species identification using average nucleotide identity (ANI) calculation (Appendix 1 Table 3; Appendix 3). E. cloacae subsp. cloacae was the most prevalent species (n = 56 [50.0%]) (Figure). Multilocus sequence typing (MLST) assigned 42 known unique STs for 105 strains. The 7 remaining isolates belonged to new or undetermined STs. Major STs (>4 isolates) were ST820 (n = 45), ST250 (n = 5), ST657 (n = 5), ST1516 (n = 4), and ST1517 (n = 4) (Figure). Of note, 44 of the ST820 strains corresponded to the strain isolated in the Mayotte/La Réunion outbreak; the last IMI-1 E. cloacae subsp. cloacae of ST820, 193I8, was isolated in Paris and was not clonally related to the outbreak strains. That strain exposed >1,200 single-nucleotide polymorphisms (SNPs) corresponding with the other IMI-1 ECC ST820 isolates from Mayotte or La Réunion.

Genes encoding NmcA, IMI-1, IMI-4, IMI-12, and IMI-13 were localized on the chromosome, whereas those coding for IMI-2, IMI-6, IMI-17, IMI-19, IMI-25, IMI-26 and IMI-27 were carried on plasmids. We characterized genetic environments of bla_IMI/NmcA genes using Illumina (https://illumina.com) and MinION long-read (Oxford Nanopore, https://nanoporetech.com) sequencing. All chromosome-encoded bla_IMI/NmcA genes were located into a EcloIMEX-type genetic element (Appendix 2 Figure 4, panel A), except bla_IMI-13, which possessed a distinct genetic environment (Appendix 2 Figure 4, panel B). We detected already-characterized EcloIMEX-type and 6 new variants, named EcloIMEX-11−16 (Appendix 2 Figure 4, panel A). Those EcloIMEX elements were ≈15–≈39.4-kb long, possessed a highly conserved 5′ region, and were inserted between setB and yieP genes. We observed a strong correlation between bla_NmcA and EcloIMEX-1. In contrast, we identified bla_IMI-1 on 9 different EcloIMEX elements. We saw no correlation between the Enterobacter species and the type of EcloIMEX. The bla_IMI-13 gene was inserted in the chromosome between genes encoding a hypothetical protein and an Inovirus-type Gp2 protein. We identified several complete or partially deleted insertion sequences (IS) close to bla_IMI-13 (Appendix 2 Figure 4, panel B); however, the mechanism of bla_IMI-13 acquisition is unclear.

All bla_IMI-6 genes were carried on a IncFII(Yb)-type plasmid (160–200 kb) (Appendix 1 Table 4). Similarly, bla_IMI-2 genes were carried on a IncFII(Yp)-type plasmid for 75% (8/12) of the IMI-2 producers. The plasmidic replicase was not identified in the 4 remaining IMI-2 producers. The long-read sequencing performed on strains producing new IMI variants enabled a more precise identification of plasmid type and size (Appendix 1 Table 3). The close genetic environments of the bla_IMI genes included several IS that differed according to the bla_IMI variants (Appendix 2 Figure 4). Conjugation experiments performed in E. coli J53 used as recipient strain confirmed those plasmids were conjugative except the 1 carrying bla_IMI-17.

We built an SNP matrix for the 44 IMI-1 E. cloacae subsp. cloacae ST820 isolates involved in the Mayotte/La Réunion outbreak to confirm their clonality. Those strains were closely related (1–62 SNPs between 2 isolates). We also performed a Bayesian analysis to estimate the date of the most recent ancestor and the evolutionary rate of that population. We estimated the evolutionary rate of the clone to 3.94 × 10⁻⁷ substitutions per site and per year (95% highest posterior density [HPD] 2.50–5.33 × 10⁻⁷), corresponding to 1.63 SNPs per genome per year (95% HPD 1.04–2.21 SNPs). The common ancestor of the 44 IMI-1–producing E. cloacae subsp. cloacae ST820 isolates has an estimated date of 1994.7 (95% HPD 1990.8–2000.2) (Appendix 2 Figure 5).

Consistent with previous findings (6,9), our collection of IMI producers included uncommon species of ECC, such as E. cloacae subsp. cloacae, a rarely described species; IMI-1, IMI-2 and IMI-6 were the most prevalent variants. We identified no isolates of E. hormaechei, the most prevalent carbapenemase-producing ECC species (11,12).

Genetic environments and plasmid types of IMI-2 producers identified in this study were similar to those previously described (2,3,13); IncFII(Yp)-type plasmids were most common. The close genetic environment of bla_IMI-2 observed in our isolates has been reported on a plasmid identified in E. coli (2). The genetic environment of bla_IMI-6 was previously reported in an E. cloacae isolate described by Boyd et al. (6). Regarding the chromosome-encoded IMI and NmcA variants (n = 85), we described a variety of EcloIMEX elements (n = 11) including 6 novel elements; that the same EcloIMEX could be identified in different ECC species suggests that XerC/D recombinases enable the mobility of these bla_IMI-/NmcA–carrying EcloIMEX structures specifically between ECC species. Finally, the evolution rate of the IMI-1–producing E. cloacae subsp. cloacae ST820 clone (1.63 SNPs/genome/year) is similar to the 0.5–3 SNPs/year for a genome reported for a population of multidrug-resistant ECC in the United Kingdom (14) and the 2.5–3 SNPs/year for a genome identified for ST171 and ST78 carbapenem-resistant ECC (15).

In conclusion, in IMI/NmcA producers in France, we observed a large diversity of ECC species, STs, genetic supports, and genetic environments. Future work should elucidate why E. cloacae subsp. cloacae is highly prevalent among IMI producers; why bla_IMI/NmcA-carrying plasmids were almost always found alone in IMI-producing isolates that always do not carry any other resistance genes; and whether EcloIMEX genetic elements are mobilizable. Clinicians should remain aware of potential antimicrobial resistance among ECC species.