High Genetic Diversity of Carbapenem-Resistant Acinetobacter baumannii Isolates Recovered in Nigerian Hospitals in 2016 to 2020

ABSTRACT Acinetobacter baumannii causes difficult-to-treat infections mostly among immunocompromised patients. Clinically relevant A. baumannii lineages and their carbapenem resistance mechanisms are sparsely described in Nigeria. This study aimed to characterize the diversity and genetic mechanisms of carbapenem resistance among A. baumannii strains isolated from hospitals in southwestern Nigeria. We sequenced the genomes of all A. baumannii isolates submitted to Nigeria’s antimicrobial resistance surveillance reference laboratory between 2016 and 2020 on an Illumina platform and performed in silico genomic characterization. Selected strains were sequenced using the Oxford Nanopore technology to characterize the genetic context of carbapenem resistance genes. The 86 A. baumannii isolates were phylogenetically diverse and belonged to 35 distinct Oxford sequence types (oxfSTs), 16 of which were novel, and 28 Institut Pasteur STs (pasSTs). Thirty-eight (44.2%) isolates belonged to none of the known international clones (ICs). Over 50% of the isolates were phenotypically resistant to 10 of 12 tested antimicrobials. The majority (n = 54) of the isolates were carbapenem resistant, particularly the IC7 (pasST25; 100%) and IC9 (pasST85; >91.7%) strains. blaOXA-23 (34.9%) and blaNDM-1 (27.9%) were the most common carbapenem resistance genes detected. All blaOXA-23 genes were carried on Tn2006 or Tn2006-like transposons. Our findings suggest that a 10-kb Tn125 composite transposon is the primary means of blaNDM-1 dissemination. Our findings highlight an increase in blaNDM-1 prevalence and the widespread transposon-facilitated dissemination of carbapenemase genes in diverse A. baumannii lineages in southwestern Nigeria. We make the case for improving surveillance of these pathogens in Nigeria and other understudied settings. IMPORTANCE Acinetobacter baumannii bacteria are increasingly clinically relevant due to their propensity to harbor genes conferring resistance to multiple antimicrobials, as well as their ability to persist and disseminate in hospital environments and cause difficult-to-treat nosocomial infections. Little is known about the molecular epidemiology and antimicrobial resistance profiles of these organisms in Nigeria, largely due to limited capacity for their isolation, identification, and antimicrobial susceptibility testing. Our study characterized the diversity and antimicrobial resistance profiles of clinical A. baumannii in southwestern Nigeria using whole-genome sequencing. We also identified the key genetic elements facilitating the dissemination of carbapenem resistance genes within this species. This study provides key insights into the clinical burden and population dynamics of A. baumannii in hospitals in Nigeria and highlights the importance of routine whole-genome sequencing-based surveillance of this and other previously understudied pathogens in Nigeria and other similar settings.

Antimicrobial resistance rates. Rates of phenotypic resistance to all the tested antimicrobials were high among the A. baumannii isolates, with at least 50% resistance recorded for 10 of the 12 tested antimicrobials. Only minocycline (29.1%) and tigecycline (26.6%) had lower resistance rates. The resistance rates were, however, significantly higher among known international clone lineages than among the singletons/noninternational clones (adjusted P # 0.009; Table 1). The IC7 ( pas ST25) strains had the highest rates of resistance; all seven isolates (100%) were resistant to meropenem, imipenem, doripenem, and all the other tested antimicrobials except tigecycline (0%) and minocycline (0%). The IC9 ( pas ST85) strains had similarly high rates of resistance to the carbapenems and other antimicrobials, as well as low rates of resistance to the tetracyclines (tigecycline, 25%; minocycline, 16.7%). Interestingly, the globally disseminated IC1 ( pas ST1) and IC2 ( pas ST2) strains had the highest rates of resistance to tigecycline (22.2% and 66.7%) but showed relatively lower resistance to the carbapenems than IC7 ( pas ST25) and IC9 ( pas ST85). Phenotypic susceptibility to colistin was also determined using the Vitek 2 system, but the results are not presented here as this is not the recommended method for correct determination of colistin resistance in A. baumannii.
We generated high-quality complete assemblies for representatives (five isolates) of the carbapenem-resistant lineages (Table S3) and extrapolated the genomic context results/observations to other clonal isolates. All the bla OXA-23 genes in the different clones in this study had similar immediate genetic contexts; they were all carried on a Tn2006 transposon flanked by IS4 family ISAba1 inverted repeat elements, or a similar Tn2006-like transposon that was missing the truncated DEAD helicase gene (Fig. 4). The nine oxf ST1114/1841 (IC2) isolates carried two distant (;1.5 Mb apart) copies of the bla OXA-23 gene on the chromosome, each flanked by ISAba1. Interestingly, one of these copies was inserted just upstream of the intrinsic bla OXA-51 -like gene (bla OXA-66 ). Among the five oxf ST231 (IC1) isolates, the bla OXA-23 carbapenem resistance gene was also chromosomally located. Like the oxf ST1114/1841 isolates, these isolates also had two copies of the bla OXA-23 gene, both proximal, carried on a Tn2006 transposon, and inserted into the chromosome without being associated with any other mobile elements; one copy was inserted between the ycfP and menH genes. There was only a single copy of the bla OXA-23 gene in the seven oxf ST229 (IC7) isolates, which was also present on the chromosome and, interestingly, located within an AbaR4-type genomic island inserted within the yifB (comM subfamily) gene. No other resistance gene was found in this genomic island.
As with both copies of the bla OXA-23 genes carried by the five oxf ST231 (IC1) isolates, the bla NDM-1 gene was also chromosomally located and was carried on a 10-kb Tn125 composite transposon flanked by the ISAba125 element (Fig. 5). This composite transposon was inserted within a larger 25-kb transposon flanked by an IS6 family insertion sequence (IS), IS1006, which also carried the aph(6)-Id and aph(30)Ib genes. The 13 isolates belonging to pas ST85/IC9 ( oxf ST1089 and oxf ST1936) had a different bla NDM-1 context. In these isolates, the bla NDM-1 gene was also located on the chromosome but was associated with an upstream ISAba125 element and located within a 7.9-kb Tn7382 transposon that also carried aph(39)-VI and had two flanking IS3 family ISAba14 direct repeats. The remaining six bla NDM-1 -carrying isolates carried the gene on a Tn125 composite transposon like the oxf ST231 (IC1) isolates, but we could not determine if this transposon was on the chromosome or a plasmid as there was no representative complete assembly sequenced for this set.
Plasmids detected in A. baumannii isolates. Plasmids belonging to 22 distinct replication initiation (Rep) protein types were detected in the isolates (Table S4). Of these, the Rep_3-family plasmids were the most common, with 18 different Rep types belonging to this group detected in the isolates. The remaining four plasmid Rep types detected belonged to the Rep_1 (n = 3) and RepPriCT_1 (n = 1) groups. The R3-T3 Rep type was the most common among the isolates (n = 31 isolates), followed by R3-T1 (n = 18), R3-T6 (n = 10), and R3-T60 (n = 10). Of all Rep types that were present in at least five isolates, none was unique to a specific international clone or ST, and the distribution of these plasmids was not consistent with the phylogeny of the isolates (Fig. S2).
A large ;111-kb plasmid (pABTJ2__22; GenBank accession no. CP004359.1) belonging to Rep type T3 of the Rep_3 family was the most common plasmid detected in all the isolates (n = 21), including the nine oxf ST1114/1841 (IC2) isolates and 12 of the 13 pas ST85 (IC9) isolates. We confirmed the presence of this plasmid and extracted its sequences from the three IC2 and IC9 isolates with complete assemblies. Interestingly, in all the three isolates, none of the pABTJ2__22 plasmids (ranging in size from 112 kb to 116 kb) contained any antimicrobial or disinfectant resistance genes. In the oxf ST1114/ 1841 strain, this was the only plasmid present; thus, all resistance genes detected in this ST were chromosomally located. The only plasmid-carried resistance gene in the isolates with complete assemblies was the aminoglycoside resistance gene, ant(20)-Ia, which was carried on a small ;6-kb plasmid present in the pas ST85 and oxf ST231 isolates.

DISCUSSION
Acinetobacter baumannii infections remain a global public health problem, and much remains to be understood about the population structure of this species, especially in understudied settings. In this study, we discovered a large diversity of A. baumannii in the clinical setting of southwestern Nigeria hospitals, with 35 different sequence types, 16 of which were novel. Studies conducted in previously uncharacterized settings have reported similar findings (7,19). A 2018 retrospective study in Colombia found seven novel STs out of the 11 detected that had been circulating for over 8 years (16). Given the relatively small number of samples in our study, the observed high diversity and lack of phylogeographic clustering may also be indicative of widespread dissemination and underreporting of A. baumannii infections in Nigeria. Clinical microbiology diagnostics remains a challenge in Nigeria and most developing countries. Very few patients are cultured at all, and A. baumannii isolates are particularly difficult to identify using conventional diagnostics (7,20,21). The institution of genomic surveillance for AMR in Nigeria has been coupled with efforts to improve basic microbiology capacity, and it is expected that these measures will help plug the existing gaps in the diagnostics and surveillance of these pathogens of major public health importance (22).
This study adds to our understanding that certain clones (ICs 1 to 9) predominate globally and account for a large proportion of the antimicrobial resistance problem in A. baumannii. Nonetheless, we furthermore show that novel and emerging clones are also important in settings of endemicity, as evidenced by the relatively high resistance rates ($50% resistance to seven of the 12 antimicrobials tested) of these novel STs/ noninternational clones, as well as their possession of carbapenem resistance determinants with genetic contexts identical to those present in the international clones. Corresponding Institut Pasteur STs are as follows: oxf ST1089, pas ST85; oxf ST1936, pas ST85; oxf ST231, pas ST1; oxf ST930, pas ST32; oxf ST862, pas ST149; oxf ST2450, pas ST1093; oxf ST2456, pas ST2. *, split across two contigs.
Given that carbapenem resistance contributes to clonal expansion and successful dissemination among A. baumannii strains, the local emergence of a wide variety of carbapenem-resistant variants is a noteworthy trend (15,23). Moreover, within the known international clones, IC1 ( pas ST1) and IC2 ( pas ST2) are still regarded as the most important A. baumannii lineages causing infections and hospital outbreaks globally (3,(24)(25)(26). Our study, however, shows an increasing local prevalence of isolates belonging to the recently classified IC9 ( pas ST85) lineage relative to the globally prevalent IC2 lineage. Even more notable is the associated increased prevalence of the previously rare bla NDM-1 carbapenem resistance gene. STs belonging to IC9, mostly oxf ST1089, and most of which also carry the bla NDM-1 gene or other variants such as bla NDM-6 , have been reported more frequently predominantly in Africa and the Middle East (27)(28)(29)(30)(31)(32)(33), but also in many other parts of the world (17,(34)(35)(36).
Previous studies have highlighted bla OXA-23 as the predominant carbapenem resistance gene among A. baumannii strains and shown that bla NDM-1 is rare among A. baumannii strains (37). Our findings, however, show that although OXA-23 carbapenemases are the most common enzymes among A. baumannii in southwestern Nigeria, NDM-1 prevalence is also notably high in our setting and NDM-1 appears to be spreading between different lineages. The Tn125 composite transposon carrying the bla NDM-1 gene in the five oxf ST231 (IC1) isolates and six other isolates with distinct STs ( oxf ST862, oxf ST930, oxf ST2450, and oxf ST2456 [IC2]) is identical in structure and composition to that previously described in an A. baumannii strain isolated in Germany in 2007 and other subsequent studies (37,38). This bla NDM-1 -carrying Tn125 transposon, which is believed to have originated in A. baumannii, has also been demonstrated to be very frequently and efficiently mobilized, facilitating its dissemination within and between A. baumannii strains and other Enterobacterales (39). BLAST searches of this transposon against the National Center for Biotechnology Information's nonredundant nucleotide database revealed the presence of this transposon and its derivatives predominantly among A. baumannii and non-baumannii Acinetobacter species, but also in different plasmids and chromosomes of E. coli, K. pneumoniae, and other Enterobacterales. The bla NDM-1 among the pas ST85 (IC9) isolates, however, had a different context-a truncated Tn125 transposon captured within flanking ISAba14 direct repeats and recently named Tn7382 (40). This transposon is only sparsely described in literature but is predominantly found among A. baumannii strains belonging to IC9 (36, 40). Another bla NDM-1 -carrying and Tn125-derived transposon, designated Tn6924, was recently described in a pas ST85 isolate from Lebanon, indicating the epidemiological importance of this lineage in the increased prevalence and potential dissemination of bla NDM-1 among A. baumannii strains (41). The increased NDM-1 prevalence in our study and in A. baumannii in general, as facilitated by these highly mobilizable transposons, as well as the potential for intra-and interspecies spread, has notable implications as this enzyme is very potent and has a wider spectrum of hydrolysis for beta-lactams and carbapenems than OXA-23 and OXA-58, thus grossly limiting the already limited number of treatment options for A. baumannii infections (42)(43)(44)(45).
Although all the bla OXA-23 genes in our study had similar contexts, these Tn2006 and Tn2006-like transposon structures were found in diverse chromosomal backgrounds, thus adding to previous knowledge of their rapid and frequent genome mobility (6). One interesting observation was the presence of two copies of the bla OXA-23 gene in the oxf ST1114/1841 and oxf ST231 strains. A previous study by Zhang and colleagues reported the duplication of a plasmid-borne bla OXA-23 gene in an A. baumannii strain in the presence of subinhibitory concentrations of carbapenem (46). This duplication was reported to confer improved fitness in carbapenem-containing media but was also associated with a fitness cost in antibiotic-free media. Our findings, however, indicate the maintenance of multiple bla OXA-23 copies in the chromosomes of multiple strains in distinct lineages, despite the reported fitness costs and resulting instability. This is an important finding as it suggests a key evolutionary adaptation that could lead to the increased potency of OXA-23 carbapenemases and expansion of A. baumannii populations harboring bla OXA-23 , as well as an increased risk of mobilization and onward dissemination of the gene.
This study revealed a diverse array of plasmids, with 22 distinct replication initiation protein types detected in the bacterial isolates and Rep_3 being the most common. The distribution of these plasmids was not consistent with the phylogeny of the isolates, suggesting extensive horizontal exchange between different A. baumannii lineages. The most common plasmid identified, pABTJ2__22, did not contain any antimicrobial or disinfectant resistance genes, despite being present in many of the isolates. In fact, among all the plasmids confirmed in the clones with complete assemblies ( oxf ST231 [IC1], oxf ST1114/1841 [1C2], pas ST25 [IC7], and pas ST85 [IC9]), only one plasmid carried a resistance gene [ant(20)-Ia]. While this may be surprising given the important role that plasmids are known to play in the dissemination of resistance genes in A. baumannii (3,47), it is worth noting that the other isolates without complete assemblies could have possessed other resistance genes carried on plasmids that were not confirmed.
Our inability to generate complete assemblies for all the lineages also represents another limitation of this study. As the complete transposon structures were not assembled into single contigs in the strains without complete assemblies, it is not possible to be definitive about the transposon structures of these strains. Nevertheless, the only missing bits in the structures were repeat sequences, which are difficult to assemble with short-read data, thus providing strong evidence for the presented structures. Another limitation of this study was that we did not determine the phenotypic susceptibility of the carbapenem-resistant strains to colistin, which is one of the few remaining antimicrobials with activity against carbapenem-resistant A. baumannii. However, there were no colistin resistance determinants detected in the isolates, suggesting that they may be susceptible to colistin.
Conclusion. Acinetobacter baumannii strains in the hospital setting in southwestern Nigeria are highly phylogenetically diverse, are highly resistant to antimicrobials, and may be underreported, indicating the urgent need to improve diagnostic capacity for and surveillance of A. baumannii infections both in Nigeria and in other understudied settings. Our findings also suggest that there is frequent dissemination of carbapenem resistance genes between the different A. baumannii lineages, as well as integration and possible maintenance of these genes in the chromosomes. More local studies are needed to characterize the hospital burden of A. baumannii infection in Nigeria and identify contributors to environmental and clinical spread.

MATERIALS AND METHODS
Ethical considerations. This study was approved by the University of Ibadan/University College Hospital (UI/UCH) Ethics Committee (UI/EC/19/0632). Patients were not actively recruited for this study, and all associated patient data were anonymized before being retrieved for analysis.
Isolate collection. All Acinetobacter isolates included in this study were isolated between 2016 and 2020 and submitted to Nigeria's AMR surveillance reference laboratory at the University College Hospital, Ibadan, Nigeria. These isolates were collected as part of routine surveillance of WHO global priority pathogens in Nigeria and were isolated from blood, cerebrospinal fluid, and rectal swab samples. Where available, associated metadata on sample type, collection date, and patient hospitalization status were obtained from the reference laboratory metadata database. All cryopreserved presumptive Acinetobacter isolates were resuscitated on CHROMagar Acinetobacter medium (CHROMagar, Paris, France) and preliminarily identified using the Vitek 2 automated system (bioMérieux, Inc., Marcy-l' Etoile, France) with Gram-negative identification cards (reference number 21341) according to the manufacturer's instructions.
Antimicrobial susceptibility testing. We determined the phenotypic susceptibility of the isolates to clinically relevant antimicrobials using the Vitek AST N281 cards (reference number 414531) on the Vitek 2 automated system. The following antimicrobials were tested: cefepime, ceftazidime, ciprofloxacin, doripenem, gentamicin, imipenem, levofloxacin, meropenem, minocycline, piperacillin-tazobactam, ticarcillin-clavulanic acid, and tigecycline. The MIC values of all tested antimicrobials except tigecycline were interpreted according to the Clinical and Laboratory Standards Institute (CLSI) clinical breakpoints (48). Tigecycline MIC values .2mg/mL were interpreted as resistant as both the EUCAST guidelines (49) and the CLSI guidelines did not provide clinical cutoffs for tigecycline in A. baumannii. All interpretations were done using the AMR R package (version 1.8.1; https://msberends.github.io/AMR/).
Genomic DNA extraction and whole-genome sequencing. We extracted genomic DNA from all the presumptively identified Acinetobacter baumannii complex isolates, prepared double-stranded genomic DNA libraries, and sequenced the libraries on an Illumina platform as previously described (7). After preliminary analyses of the short-read whole-genome sequencing (WGS) data, we selected representatives of the different A. baumannii lineages identified in our data set and carried out long-read whole-genome sequencing of these isolates using the Oxford Nanopore technology to obtain completely assembled genomes for comprehensive analyses. Genomic DNA was reextracted from the selected isolates using the A&A Genomic Mini AX Bacteria1 kit (A&A Biotechnology, Gda nsk, Poland) to obtain less fragmented DNA. Long-read sequencing libraries were then generated using the Rapid Barcoding Sequencing kit (SQK-RBK004) and sequenced on a MinION Flow Cell (R9.4.1) with MinKNOW version 22.08.9 (Oxford Nanopore Technologies, Inc., Oxford, United Kingdom). We then carried out superaccuracy base calling and demultiplexing on the generated reads using Guppy version 6.3.8.
Whole-genome sequence analysis. We performed de novo genome assembly, species identification, and quality control of all short-read genomes using the Global Health Research Unit (GHRU) protocol (https://www.protocols.io/view/ghru-genomic-surveillance-of-antimicrobial-resista-bpn6mmhe). All assemblies with .300 contigs, genome sizes of ,3.3 Mb or .4.7 Mb, an N50 score of ,25,000, and containing .5% of contaminating single nucleotide variants of core genes were excluded from downstream analyses. Long-read sequences were assembled using the Trycycler pipeline (50), and the generated circularized assemblies were then polished using Medaka v1.7.2. To generate high-quality complete assemblies, the generated long-read assemblies were then polished with the short reads using Polypolish (51).
We performed a single nucleotide polymorphism (SNP)-based phylogenetic reconstruction analysis to determine the phylogenetic relationships between the identified A. baumannii strains. The raw reads of all samples were mapped to a reference sequence (GenBank accession no. GCA_000830055.1) using the BWA-MEM algorithm with BWA v0.7.17 (52), and possible duplicates were marked and removed using Picard v2.21.6 (http://broadinstitute.github.io/picard). Variant sites were called based on the alignment to the reference sequence using BCFtools v1.9 (53), and low-quality variants were removed. Variant sites were extracted using SNP-sites v2.4.1 (54) and concatenated into pseudogenomes for each of the samples, as well as the reference sequence, after which all pseudogenomes were combined to form a pseudoalignment. We then used RAxML-NG v1.1.0 (55) to construct a maximum likelihood (ML) phylogeny with 50 starting trees and 1,000 bootstrap replicates using the general time reversible gamma (GTR1G) model with the Lewis method for ascertainment bias correction (56).
Multilocus sequence types (MLSTs) were determined from the assembled genomes using the R package MLSTar v0. 1.5 (57) with the Oxford and Institut Pasteur MLST schemes (58,59). The detected sequence types were assigned to IC groups if they had no more than a double locus variation from the Oxford STs in the nine defined ICs (60,61). Lipooligosaccharide outer core loci and capsular polysaccharide loci were identified using Kaptive v2.0.3 (62). Identified loci with at least a "good" confidence match were reported. All genomes and assembly fragments were annotated using Bakta v1.5.1 (63) with database version 4.0.0. Antimicrobial resistance genes carried by each isolate were identified using AMRFinderPlus v3.10.24 (64) with database version 2022-04-04.1. The intrinsic bla ADC -family and bla OXA-51 -like genes were excluded from the analysis. Plasmid replicons were identified in the short-read assemblies using the AcinetobacterPlasmidTyping database (65). Only the best-matching replicons (highest percent identity) for each unique contig were reported. Using only the complete assemblies, the genetic contexts of carbapenem resistance genes were observed in Artemis, and genomic resistance islands were identified using the IslandViewer 4 tool (66). Insertion sequence (IS) elements were identified using the BLAST tool on the ISfinder database (https://www-is.biotoul .fr/blast.php). Using GView Server (https://server.gview.ca/), we mapped the draft assemblies of the remaining isolates in each clone to the complete assembly of the long-read sequenced representative strain to determine representativeness. Genomic context results for the representative strain were extrapolated for isolates that had a perfect mapping (.95% coverage) to the representative assembly. For carbapenem-resistant strains without complete assemblies for representative sequences, we identified the contigs carrying the resistance genes and associated elements using BLAST (https://blast.ncbi.nlm.nih.gov/Blast.cgi). Genetic structure comparisons were carried out using Clinker (67).
Statistical analysis. All statistical analyses were carried out in R v4.2.1. Proportions were compared between groups using Fisher's exact test with false-discovery rate correction for multiple testing. Dunn's test with Bonferroni correction for multiple testing was used following a Kruskal-Wallis test to carry out pairwise comparisons of the numbers of resistance determinants between multiple groups. P values less than 0.05 were considered statistically significant.
Data availability. The raw reads of all 86 A. baumannii genomes have been deposited in the European Nucleotide Archive (https://www.ebi.ac.uk/ena) with study accession no. PRJEB29739. Accession numbers for each sample are listed in Table S1 in the supplemental material.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only.

ACKNOWLEDGMENTS
We thank Ayorinde Afolayan, Gabriel Temitope Sunmonu, and Gitte Petersen for technical assistance. We also acknowledge Nonyelum Osuagwu for coordinating the collection of isolates from facilities (Clina-Lancet Laboratories and EL-LAB Medical Diagnostics) that are not originally part of Nigeria's AMR surveillance system.