Despite significant strides in developing novel antimicrobials, the treatment of bacterial infections has become increasingly challenging due to the widespread emergence of multidrug resistance (MDR), particularly among Enterobacteriaceae. MDR pathogens represent a clear and present danger that affects every populated continent of the globe (Bassetti et al., 2015). For instance, MDR bacteria are estimated to contribute to 45% of deaths in Africa (WHO, 2014), and in 2019 alone, MDR Escherichia coli and Klebsiella pneumoniae were associated with over 100,000 deaths globally (Murray et al., 2022). In Enterobacteriaceae, MDR is often mediated by extended-spectrum β-lactamases (ESBLs) (Zhang et al., 2015; Beyene et al., 2024). While TEM and SHV variants were historically predominant, the CTX-M-type ESBLs, which hydrolyze third-generation cephalosporins like cefotaxime, have now become the most prevalent globally (Castanheira et al., 2021). CTX-M type ESBLs are classified into five groups, CTX-M-1, CTX-M-2, CTX-M-8, CTX-M-9, and CTX-M-25, with the CTX-M-1 group being the most predominant. Recent meta-analyses in sub-Saharan Africa confirm CTX-M-15 (CTX-M-1 group) as the most reported ESBL type across human and animal health sectors (Olaitan et al., 2025). The pandemic clone E. coli ST131 producing CTX-M-15 type ESBL remains a key driver in Africa, mirroring global trends (Nicolas-Chanoine et al., 2014). The bla_CTX-M gene, which encodes CTX-M types ESBLs, is typically located on mobile genetic vectors called plasmids that often harbor multiple antimicrobial resistance (AMR) genes (Mikhayel et al., 2024). This facilitates the rapid dissemination of MDR phenotypes.

Although plasmids contribute significantly to AMR gene spread, they may impose a fitness cost on the bacteria due to the energetic burden of their replication and maintenance, thus increasing the chances of instability or loss (Carroll and Wong, 2018). Additionally, plasmids are susceptible to loss during cell division through segregational instability. To overcome these limitations and ensure stable inheritance of critical traits, plasmid-encoded genes, including bla_CTX-M, may integrate into the bacterial chromosome, provided such insertions do not truncate essential genes or alter vital regulatory pathways (San Millan et al., 2015). Chromosomal integration of bla_CTX-M is frequently mediated by mobile elements such as ISEcp1 and IS26 (Gomi et al., 2022; Komori et al., 2024). Most studies have reported chromosomal bla_CTX-M gene insertions as small segments (<5,000 bp) containing few or no additional resistance genes (Mshana et al., 2015). However, emerging evidence suggests that larger chromosomal islands harboring bla_CTX-M alongside multiple AMR genes [e.g., aac(3)-IIa, qnrB1, aac(6′)-Ib-cr5, bla_OXA-1, dfrA14, catB3, tet(A)] also exist. These bla_CTX-M-carrying islands are usually mobilized by transposable elements, which facilitate their insertion at various chromosomal sites. Notably, Yoon et al. (2020) described bla_CTX-M-carrying chromosomal segments in K. pneumoniae, while Goswami et al. (2020) reported similar findings in E. coli.

Previously, our research group reported large MDR chromosomal islands carrying bla_CTX-M in E. coli and Enterobacter cloacae from Zambia (Shawa et al., 2021; Shawa et al., 2025). However, the overall prevalence and genetic landscape of chromosomal bla_CTX-M across the African continent remain poorly characterized. With the growing availability of whole genome sequencing (WGS) data from African countries submitted to the National Center for Biotechnology Information (NCBI), this study used in silico analysis to estimate the prevalence of chromosomally integrated bla_CTX-M among Enterobacteriaceae of African origin and characterize the genetic context of these chromosomal insertions.

In September 2024, whole genome sequences of Enterobacteriaceae chromosomes (E. coli, K. pneumoniae, Enterobacter spp., Salmonella spp., and Shigella spp.) from African clinical sources were downloaded from the NCBI nucleotide database. To this end, we explored the NCBI using search terms including the bacterial species, host species, and country (e.g., “Escherichia coli, Homo sapiens, Kenya”) and filtered the results using a customized sequence length range of 1,000,000 to 6,000,000 bp. Fasta files of the output sequences were collected as a single folder for each country through a bulk download. The number of unique sequences per dataset was determined using Seqkit (rmdup function) (Shen et al., 2016).

In silico prediction of AMR genes was performed by ResFinder (Florensa et al., 2022), and bla_CTX-M-harboring genomes were annotated using dfast version 1.3.2 (Tanizawa et al., 2018). The chromosome sequences were also subjected to in silico multilocus sequence typing (MLST) using the mlst database,¹ which partly uses the PubMLST database² (Jolley and Maiden, 2010). Chromosomal insertions harboring the bla_CTX-M gene were detected by aligning the annotated files to appropriate reference genomes using Mauve (Darling et al., 2004), and the results were visualized in genoPlotR version 0.8.11 (Guy et al., 2010). To further characterize the insertions, nucleotide sequences were subjected to BLASTn against the NCBI database, and the resulting hit tables were filtered in R using the dplyr package version 1.1.4 (Wickham et al., 2023), with filtering criteria set to “% identity > 99.” Additionally, plasmid replicons were detected using the PlasmidFinder database (Carattoli et al., 2014), while genomic islands of horizontal origin were predicted using IslandViewer 4 (Bertelli et al., 2017), which includes IslandPath-DIMOB (Hsiao et al., 2003), SIGI-HMM (Waack et al., 2006), and IslandPick (Langille et al., 2008). Previously published Zambian sequences were excluded from this analysis, as they have been characterized in prior studies (Shawa et al., 2021).

To distinguish between E. coli and Shigella spp., one sequence was subjected to a local BLAST using lacY and ipaH genes. To this end, lacY (from E. coli str. K-12 substr. MG1655, Accession Number: NC_000913.3) and ipaH (from S. flexneri 2a str. 301, Accession Number: NC_004337.2) were downloaded from the NCBI and concatenated into one fasta file. A local database was then created from the merged fasta file using the “makeblastdb” command while specifying the “-dbtype nucl” and “-parse_seqids” arguments. Finally, the query sequence was subjected to a BLAST search against the local database using the “blastn” command. S. flexneri 2a str. 2,457 T (Accession Number: AE014073.1) was used as a positive control for ipaH.

Core-SNP-based phylogenetic trees for the bla_CTX-M-carrying genome sequences of each species were created using parsnp version 2.1.4 (Treangen et al., 2014) using E. fergusonii (Accession Number NZ_CP083638.1) and K. quasipneumoniae (Accession Number NZ_LR588411.1) as outgroups for E. coli and K. pneumoniae, respectively. Finally, the output parsnp tree files were visualized and edited in iTOL version 7 (Letunic and Bork, 2024).

In this study, we utilized publicly available WGS data to investigate the burden and genetic context of chromosomally integrated bla_CTX-M genes among Enterobacteriaceae species from Africa. Chromosomal bla_CTX-M was detected in ~16% of genomes analyzed, spanning nine out of 18 countries. The most frequently identified allele was bla_CTX-M-15, found in two-thirds of the sequences. The bla_CTX-M gene was located within the chromosomal insertions flanked by ISEcp1 or IS26 elements, highlighting the central role of these mobile elements in facilitating stable chromosomal integration of this clinically significant ESBL determinant. Notably, some of these insertions also harbored additional AMR genes, underscoring their potential to encode chromosomal MDR.

The high prevalence of chromosomal bla_CTX-M may reflect strong selection pressure arising from widespread antimicrobial use in clinical settings. Chromosomal integration offers a potential evolutionary advantage by reducing fitness costs and structural instability typically associated with plasmid carriage, as well as mitigating the risk of plasmid loss through segregational events during cell division (Gu et al., 2015). The strains examined in this study appear to have acquired secondary chromosomal integration of plasmid-derived bla_CTX-M via two main mechanisms. In the majority of sequences, the ISEcp1 gene was located upstream of bla_CTX-M, consistent with ISEcp1-mediated transposition (Poirel et al., 2005). The presence of MDR islands bounded by ISEcp1 further suggests that this element can transpose large DNA segments. These MDR islands often included additional IS elements, indicating that chromosomal integration may occur through a multistep process involving successive transposition events. Nevertheless, the identification of nearly identical insertions in both plasmid and chromosomal sequences supports the possibility of a single-step integration.

Notably, in strains lacking ISEcp1, insertions were flanked by IS26 elements. Previous studies have shown that IS26-mediated transposition can occur via a cointegration mechanism, leading to duplication of IS26 and the formation of 8 bp TSDs (Harmer et al., 2014). Alternatively, a single IS26 element with adjacent “passenger” genes can form a translocatable unit, which may insert next to another IS26 copy through conservative transposition that does not generate TSDs (Harmer et al., 2014). We identified 8 bp TSDs flanking two directly oriented IS26 elements on either end of a 75 kbp chromosomal insertion encoding MDR. This observation is consistent with IS26-mediated replicative transposition. However, comparison with matching GeneBank sequences revealed notable differences in the arrangement of genes within these insertions. Given the presence of multiple IS copies across the segment, it is likely that the genes were integrated through a multistep process involving several mobile gene elements. We speculate that the initial event may have been a copy-in transposition involving a single IS26 element and associated genes, leading to the formation of a second IS26 and TSDs. This could have been followed by a series of conservative (non-replicative) transposition events, whereby translocatable units targeted pre-existing IS26 sites within the growing composite island (Harmer and Hall, 2024).

Meanwhile, in one E. cloacae strain, the MDR insertion was flanked by int immediately upstream of ISEcp1, suggesting that chromosomal integration may have occurred via site-specific recombination. However, the involvement of ISEcp1 cannot be excluded, given its role in mobilizing resistance elements. Notably, this MDR segment was detected in numerous plasmid and chromosome sequences in the GeneBank, all of which lacked the flanking int gene but possessed ISEcp1 and IS26 on opposite ends. This pattern suggests that structural rearrangement, potentially mediated by ISEcp1 and/or IS26, may have occurred following initial integration.

The observation of similar genetic architectures shared by clonally unrelated strains highlights the frequency with which chromosomal integration occurs by horizontal gene transfer. For instance, the qnrS1 gene was found precisely 4,640 bp downstream of bla_CTX-M in two genetically distinct E. coli strains (ST38 and ST8130), as well as a strain previously identified as S. sonei but reassigned to E. coli ST484. This suggests that an identical DNA segment was horizontally acquired by diverse genetic backgrounds. In parallel, we also observed evidence of the spread of chromosomal bla_CTX-M by clonal expansion. Three of the four K. pneumoniae strains from South Africa belonged to the same ST (2497) and carried the same AMR profiles. These three strains were recovered from patients admitted to the same hospital in Cape Town (Marais et al., 2024), supporting a likely scenario of patient-to-patient transmission. Similarly, the two K. pneumoniae ST39 from Uganda and the two ST437 strains from Tanzania shared indistinguishable resistance gene arrangements, consistent with local clonal dissemination. In Sudan, while only two of the 17 K. pneumoniae strains were typed (both as ST11), the remaining 15 untyped strains also harbored the same set of AMR genes, suggesting either the expansion of a single dominant clone or recurrent horizontal acquisition of a common resistance island by unrelated strains. Generally, most strains from different countries did not share STs, and when the same ST was observed across borders, the associated AMR profiles differed. For example, K. pneumoniae ST11 from Ghana and Sudan had bla_CTX-M-15 and bla_CTX-M-14, respectively, and the two strains exhibited distinct resistance architectures, indicating independent acquisition events and no evidence of inter-country clonal spread. Phylogenetic analysis revealed regional clustering of E. coli and K. pneumoniae strains, but the presence of bla_CTX-M-harboring MDR insertions did not correlate with strain clonality, highlighting the role of horizontal gene transfer in MDR dissemination.

Interestingly, the MDR region identified in the Malawian strain CAC124 (ST5640) was detected in the chromosomes of over 30 E. coli ST131 strains (Supplementary Table S3). While CAC124 most likely acquired this segment through horizontal gene transfer, its consistent presence within the ST131 lineage suggests subsequent clonal dissemination. The broad geographic distribution of E. coli ST131 carrying a chromosomally integrated MDR-encoding genetic island could suggest an emerging clonal wave of extraintestinal pathogenic E. coli, particularly given that ST131 is already recognized as a pandemic clone.

The presence of multiple genomic islands in the analyzed genomes highlights the high frequency of horizontal gene transfer events and the potential role of factors beyond AMR genes for virulence and fitness enhancement. Furthermore, detecting plasmid replicons on chromosomes confirms the horizontal origins of these factors. While exceptionally large plasmids (>1.7 Mbp) have been reported in rare cases (Kothari et al., 2019), all the replicon-harboring contigs in this study were larger than 4.9 Mbp (Table 2), ruling out their possibility of being plasmids.

This study is not without limitations. First, some strains could not be assigned STs due to missing loci or nucleotide ambiguities within the MLST regions. Second, we assumed that all genomic sequences exceeding 1,000,000 bp represented chromosomal sequences, which may not always be true, as described above. Third, our approach may have excluded chromosomal segments <1,000,000 bp, as many genome assemblies are incomplete. Lastly, a strain previously reported as S. sonei by Microflex LT MALDI-TOF MS (Bruker Daltonik, GmbH, United Kingdom) (Portal et al., 2024) was reassigned to E. coli ST484 based on our WGS analysis. This discrepancy is not unexpected, as Shigella species are phylogenetically nested within E. coli, prompting ongoing discussions to reclassify Shigella as a sublineage of E. coli (Lan and Reeves, 2002).

We used publicly available WGS data to characterize the genetic landscape of chromosomally integrated bla_CTX-M among Enterobacteriaceae strains from Africa. Approximately 16% of the analyzed genomes carried chromosomally encoded bla_CTX-M, many of which could be linked to plasmid-derived origins. In several cases, the bla_CTX-M-containing chromosomal insertions also harbored additional AMR genes, resulting in genotypically MDR chromosomal segments. While widespread clonal dissemination was observed in the globally dominant E. coli ST131 lineage, clonal spread among African strains appeared localized, often limited to individual hospital settings. Our findings underscore an evolutionary strategy by which Enterobacteriaceae stabilize and preserve MDR traits through chromosomal integration, potentially ensuring long-term persistence even in the absence of plasmid-mediated transmission.