In the last decades, Acinetobacter baumannii infections have moved to the forefront of challenges encountered by clinicians worldwide. It is mainly recognized for causing a wide range of difficult-to-treat hospital-acquired infections, particularly in critically ill patients (Morris et al., 2019). Working in concert, the remarkable capacity for upregulating intrinsic resistance mechanisms and acquisition of foreign resistance genes contributed to an ever-expanding spectrum of antimicrobial resistance in A. baumannii. Leaving behind limited or no antimicrobial treatment options, extensively drug-resistant (XDR) and pandrug-resistant strains have been increasingly reported from different parts of the world (Hsueh et al., 2002; Leite et al., 2016; Hamidian and Nigro, 2019). Most of them are members of the high-risk global clones (GCs) 1 and 2 (also known as international clones; ICs) (Karah et al., 2012). Genome sequencing of the earliest strains of the high-risk GCs uncovered a wide repertoire of resistance genes being associated with genomic resistance islands (RIs) (Hamidian and Hall, 2017b). These are genomic regions encompassing variable assortments of transposons and integrons loaded with specific resistance genes (Fournier et al., 2006). They are one of the hallmarks of the horizontal transfer of resistance genes (Carraro et al., 2017).

The first known genomic RI, designated AbaR1 (A. baumannii Resistance 1), was identified in A. baumannii strain AYE from France carrying antimicrobial and heavy metal resistance genes within transposon fragments (Fournier et al., 2006). With a wide variability in size, genetic structure, and insertion sites (Bi et al., 2020), at least seven families of genomic RIs are currently known. These include AbaR-type islands (Post et al., 2010), AbaR4 (Hamidian and Hall, 2011), and A. baumannii genomic resistance islands (AbGRIs) types 1 to 5 (Nigro and Hall, 2012b; Nigro et al., 2013; Wright et al., 2014; Blackwell et al., 2017; Chan et al., 2020; Hua et al., 2021). Any or more than one RIs may be carried by MDR A. baumannii strains (Chan et al., 2015; Hamidian and Hall, 2017b). In addition to the Tn6019 backbone, AbaR contains multiple antibiotic resistance regions (MARRs) enclosed by two copies of Tn6018. AbaR-type RIs are commonly inserted within the ATPase-coding gene comM (Hamidian and Hall, 2018) often in GC1 strains. In the same location, two other RIs were identified. These include AbaR4, in which Tn2006 is inserted in a Tn6022 backbone (Hamidian and Hall, 2011), and AbGRI1, identified in GC2 strains (Nigro and Hall, 2012a,b; Hamidian and Hall, 2017b). AbGRI1 consists of Tn6022 (or its deletion derivatives) and Tn6172 joined by a plasmid-derived linker (Hamidian and Hall, 2017b). AbGRI1 variants often carry the resistance genes sul2, tet(B), strA, strB, and sometimes bla_OXA−23 (Bi et al., 2020). The three RIs, AbaR, AbaR4, and AbGRI1, are complex class III transposons carrying the transposition genes tniCABDE that target the comM gene for insertion (Hamidian and Hall, 2017b). The other four types of AbGRIs are IS26-bound transposons harboring variable combinations of resistance genes that are characteristic for each type. They are commonly identified in the chromosomes of A. baumannii strains that belong to GC2. They include AbGRI2 (Nigro et al., 2013), AbGRI3 (Blackwell et al., 2017), AbGRI4 (Chan et al., 2020), and AbGRI5 (Hua et al., 2021). AbGRI2 characteristically carries all or some of the resistance genes bla_TEM, aphA1, catA1, and a class I integron carrying sul1, aacC1, and aadA1 (Nigro et al., 2013). AbGRI3 commonly inserts within a putative GNAT family N-acetyltransferase gene. In addition to armA conferring resistance to all clinically useful aminoglycosides, AbGRI3 also carries msrE and mphE, with or without class I integron carrying the resistance genes aacA4, catB8, aadA, and sul1. In some cases, IS26-bracketed aphA1 also integrates into AbGRI3 (Blackwell et al., 2017). Recently, AbGRI4 was identified in GC2 and non-GC2 strains carrying the resistance genes aadB, aadA2, and sul1 in a class I integron. AbGRI4 uniquely targets an α/β-hydrolase gene (Chan et al., 2020). AbGRI5 is the latest RI to be identified in A. baumanni that resembles AbGRI3 in harboring armA, msrE-mphE, sul1, and class I integron that carries a different array of resistance genes [bla_PER−1-bla_CARB−2-aadA2-cmlA1-aadA1] compared to AbGRI3. In addition, AbGRI5 distinctively carries the macrolide resistance gene ere(B) (Hua et al., 2021).

Even though reports about the structure of RIs carried by strains of this extremely problematic pathogen were published from several parts of the world (Lee et al., 2016; Blackwell et al., 2017; Kim et al., 2017; Chan et al., 2020; Leal et al., 2020; Hua et al., 2021), little is known about those circulating in Egyptian hospitals. Here, we used whole-genome sequencing (WGS) to analyze the diversity and configuration of RIs carried by 18 strains of A. baumannii isolated from patients admitted to one of the largest tertiary university hospitals in Cairo, Egypt, in 2020.

Updates on the genetic background of MDR and XDR A. baumannii are continuously being published from different parts of the world (Hamidian and Nigro, 2019; Gheorghe et al., 2021; Wareth et al., 2021) and from Egypt as well (Hassan et al., 2021; Jalal et al., 2021; Wasfi et al., 2021). However, reports about the association of resistance genes with RIs are relatively scarce. This is in part due to the need for multistep PCR mapping or whole-genome sequencing. To the best of our knowledge, this is the first report about the diversity and the genetic configuration of genomic RIs carried by A. baumannii isolates from Egypt. For this purpose, draft genomes of 18 non-duplicate MDR and XDR isolates were generated. Mostly from ICUs, the isolates were recovered from patients with bloodstream, respiratory tract, and wound infections. Very few treatment options were available. Draft genomes were employed for MLST analysis using both Pasteur and Oxford schemes. Oxford scheme-based analysis revealed the co-existence of two different alleles of the gdhB locus in 7/18 isolates generating two STs per isolate, a previously reported drawback for the scheme (Tomaschek et al., 2016). Nevertheless, MLST profiles generated by the Oxford scheme showed superior discrimination and concordance with the SNP-based phylogeny results. This was in line with other reports as well (Tomaschek et al., 2016; Gaiarsa et al., 2019). As reported before (Karah et al., 2012), most of the MDR and XDR A. baumannii strains belong to the Pasteur CCs 1 and two widely known as GC1 and GC2. In agreement with other studies (Al-Hassan et al., 2019; Fam et al., 2020; Wasfi et al., 2021), MLST analysis revealed the predominance of GC2 isolates in our collection (38.8%). Nevertheless, GC1 strains outweighed those that belonged to GC2 studied by others in our region (Ghaith et al., 2017; Jalal et al., 2021). Less representation (16.6%) was noted for the recently described GC9 (Müller et al., 2019) known to be endemic in Middle East countries (Al-Hassan et al., 2013, 2021; Bonnin et al., 2013; Ghaith et al., 2017; Jaidane et al., 2018; Salloum et al., 2018). Notably, a low prevalence of high-risk GCs was reported by older studies in Egypt (Al-Hassan et al., 2013; El Bannah et al., 2018), reflecting the progressive expansion of high-risk global clones in Egyptian hospitals over years. The emergence of successful STs not assigned to any high-risk GCs was also evident in the current study. These included the novel clonal complex CC2329^Oxf (CC113^Pas) represented by M03 (ST113/Pas2246^Oxf) and M14 (ST113/Pas ST2329^Oxf). Isolates that belong to this CC were reported from Egypt (Jalal et al., 2021), Saudi Arabia (Lopes et al., 2015), and Brazil (Leal et al., 2020). Another ST identified here, ST164^Pas/1418,Oxf was reported in the isolates from Myanmar (Tada et al., 2020), Vietnam (Wareth et al., 2021), Brazil (Coelho-Souza et al., 2013), Kenya (Musila et al., 2021), and as a major clone in Thailand (Khuntayaporn et al., 2021). SNP-based phylogenetic analysis showed that our isolates were clustered with others from different parts of the world. Transmission of genetically related strains over continents might be facilitated by travel-associated fecal colonization (Ostholm-Balkhed et al., 2013) and medical tourism (Benenson et al., 2018). Together with international strains with similar STs, our isolates were distributed over seven clusters in a pattern that fully matched the Oxford scheme-based STs.

The detailed profiling of the resistance determinants and correlation to STs and resistance phenotypes was presented in the current study with special emphasis on resistance to β-lactams and aminoglycosides. The epidemiological linkage between specific bla_{OXA−51−like} variants and certain GCs was reported before (Zander et al., 2012; Karah et al., 2016; Jaidane et al., 2018) and was also evident in our isolates. This was noted for bla_OXA−64,bla_OXA−66,and bla_OXA−94 that were linked to GC1, GC2, and GC9, respectively. In agreement with others (Jalal et al., 2021), bla_ADC−79 was linked to GC1. Except for those that belonged to ST1050/2058^Oxf which carried bla_ADC−30, GC2 isolates carried bla_ADC−73, a single-nucleotide variant of bla_ADC−30 (Karah et al., 2016). ISAba1-amplified ADC-type β-lactamases, previously coupled to high-level cephalosporin resistance (Corvec et al., 2003), were confined to GC1 and GC2 isolates. Carbapenem resistance in our collection was mediated by two carbapenem-hydrolyzing enzymes, namely, OXA-23 and NDM-1. OXA-23 was the most common of all acquired β-lactamases produced by our isolates. It is also the most frequently described carbapenemase globally (Hamidian and Nigro, 2019). The gene bla_OXA−23 existed in Tn2006 that, in turn, was sometimes embedded in RIs. Despite the variety of the genetic platforms known to harbor bla_OXA−23, Tn2006 is the dominant vehicle for the acquisition of the gene worldwide (Nigro and Hall, 2016; Hamidian and Nigro, 2019). Less frequently, carbapenem resistance was mediated by NDM-1 (class B β-lactamases). In addition to one GC2 isolate (M20) and the phylogenetically related isolates M10 and M19, bla_NDM−1 was carried by all GC9 (ST85^Pas) isolates. GC9 was claimed to act as a reservoir for bla_NDM−1 in our region (Bonnin et al., 2013; Jaidane et al., 2018; Salloum et al., 2018; Al-Hassan et al., 2021). As described in our previous study (Zafer et al., 2021), bla_NDM−1 laid within an ISAba14-bracketed transposon also carrying aphA6 in M02, M11, and M10. However, the transposon could not be identified in other NDM-producer strains in our collection. Large-scale screening of this novel transposon is, therefore, recommended. Although reported in a higher prevalence in Egypt (El-Sayed-Ahmed et al., 2015; Abouelfetouh et al., 2019; Wasfi et al., 2021), a combination of the carbapenemase-coding genes bla_OXA−23 and bla_NDM−1 was found in only one isolate (M20). Class A β-lactamases, including TEM-1, CARB-16, and PER-7, had a considerable share in our collection too. While the exact environment of bla_CARB−16 could not be identified, both bla_TEM−1 and bla_PER−7 were exclusively carried within RIs, including AbGRI2 and RI-PER-7, respectively. The latter was first described as Tn1548-like-2 transposon by Karah et al. (2016) and later designated RI-PER-7 by Adams et al. (2020). Both reported the island in isolates that had Pasteur ST25, a double-locus variant of ST113^Pas in which RI-PER-7 was identified here. Shorter versions of the island were also reported to be carried by members of the family Enterobacteriaceae (Adams et al., 2020). WGSs of four isolates of the ST ST2246^Oxf were generated in a recent study in Egypt (Jalal et al., 2021). While the authors did not report the RI-PER-7 in their isolates, the signature genes of the island [mphE, msrE, armA, sul1, bla_PER−7, cmlA5, and arr2] were identified in 3/4 (75%) isolates. This newly emerging ST might thus serve as a reservoir for RI-PER-7. While the RI-PER-7-positive contig of M03 showed the highest similarity to a plasmid sequence and was thus anticipated to be plasmid-mediated, the exact location of the island in ST2246^Oxf is yet to be confirmed.

Of all the aminoglycoside resistance genes identified here, those conferring resistance to the clinically relevant aminoglycosides, amikacin, gentamicin, and tobramycin, were of much concern. Most importantly, the broad-spectrum armA gene conferring high-level resistance to all aminoglycosides (Galimand et al., 2003) was carried by all GC2 isolates. A high prevalence of armA has been previously reported in Egypt (El-Sayed-Ahmed et al., 2015). In addition, co-existing bla_OXA−23, bla_NDM−1, and armA were found in one GC2 isolate (M20). This combination was identified on plasmids of 8/25 (32%) isolates in an older study in Egypt (El-Sayed-Ahmed et al., 2015). Amikacin resistance in our isolates was also coded by aphA6 in 38.8% of the isolates. The gene is commonly identified in A. baumannii within the composite transposon TnaphA6 in which it is bracketed by two directly oriented ISAba125 elements (Nigro et al., 2011). Nevertheless, here it was most commonly identified within a bracket of an upstream ISAba14 element and a downstream ISAba125 element that was sometimes part of a composite transposon also carrying bla_NDM−1, as described above. Given that the ISAba125 elements are thought to drive the overexpression of aphA6 imparting high-level amikacin resistance, the effect of the upstream ISAba14 insertion is questionable. Notably, this unusual environment was associated with unrelated STs. While frequently found on plasmids (Nigro and Hall, 2012b; Hamidian and Hall, 2014; Hamidian et al., 2014), plasmid-associated aphA6 was found in one isolate (M05) in our collection. Encoding resistance to both gentamicin and tobramycin, aadB was spotted in 7/18 isolates on pRAY plasmid derivatives. In A. baumannii, aadB was identified within a class 1 integron or more commonly on the globally distributed pRAY plasmid derivatives (Hamidian et al., 2012). While the GC1 isolates M12 and M15 did not carry any pRAY plasmids or aadB, such plasmids were carried by their phylogenetically related isolates M06 and M09. Other pRAY-positive STs included ST85^Pas and ST164^Pas. In addition to armA, the long version of AbGRI3-1 RIs also carried aac4 that encodes the aminoglycoside modifying enzyme AAC(6')-Ib', conferring resistance to gentamicin rather than amikacin modified by AAC(6')-Ib (Ramirez and Tolmasky, 2010). In agreement with a recent study from Egypt, aac4 was merely carried by a subset of GC2 strains, perhaps carrying the same version of AbGRI3 (Jalal et al., 2021). In addition, a wide array of resistance genes to other antimicrobial classes including multidrug efflux pumps was abundantly identified in our collection.

The advances in the next-generation sequencing technology made possible the genome-wide resistome analysis of different bacterial species and the investigation of the association of resistance genes with mobile genetic elements. Despite the difficulties imparted by the fragmented nature of contigs assembled from the short-read sequencer output, we used a combined approach of de novo assembly and reference mapping to uncover the configurations of RIs carried by our isolates. Examining the integrity of the comM gene, the hotspot for insertion of AbaR, AbaR4, and AbGRI1-type islands was the first step in RI analysis. The gene was found to be interrupted in the majority of the isolates (16/18) equally by the resistance genes-free transposon Tn6022 and either AbaR4 or AbGRI1-like RIs (44.4%). A similar prevalence (46.7%) was reported for Tn6022 in A. baumannii isolates from Korea (Kim et al., 2017). Tn6022, the backbone of AbaR4, comprises five transposition genes (tniCABDE), a universal stress protein-encoding gene (uspA), and a sulfate permease-coding gene (sup) (Hamidian and Hall, 2011). It has been widely identified in lineage 2 GC1 isolates (Hamidian and Hall, 2011). Here, the transposon was carried by GC1 isolates and their phylogenetically related ones (M06 and M09). In addition, the transposon interrupted the comM gene in all GC9 isolates and only one GC2 isolate (M20). In Tn6022, the gene sup is a hotspot for insertion of Tn2006 with an embedded bla_OXA−23 generating AbaR4. While widely described in GC2 isolates (Kim et al., 2012, 2013), AbaR4 was identified here within the chromosome of M03 only. Interestingly, co-existing chromosomal and plasmid copies of AbaR4 were recently reported in Proteus mirabilis (Octavia et al., 2020). While not self-transmissible (Bi et al., 2019), AbaR4 mobility to the chromosome of P. mirabilis was proposed to be mediated by plasmids (Octavia et al., 2020). This demonstrates the impending threat of interspecies plasmid-mediated transfer of AbaR4.

Twenty-two AbGRI1 configurations with different backbones were characterized by Bi et al. (2020). Here, we identified three variant structures of AbGRI1. Of them, two variants had the typical AbGRI1 backbone [Tn6022 (or the deletion derivative, Tn6022Δ)-linker-Tn6172]. Despite their unique configurations, they were named here as AbGRI1-like-1 and AbGRI1-like-2 due to the lack of a universal nomenclature system for A. baumannii genomic RIs (Hamidian and Hall, 2018). AbGRI1-like-1 was identified in ST1050/2058^Oxf GC2 isolates. It was made up of Tn6022Δ, a plasmid-derived linker with int gene uniquely interrupted by an ISAba125 element and Tn6172. The ISAba125 element interrupting the int gene might act as a hotspot for the insertion of resistance elements flanked by two direct copies of ISAba125 (such as TnaphA6) by homologous recombination. The second variant AbGRI1-like-2 was formed of AbaR4, plasmid-derived linker, and Tn6172 to which a ΔISCR2-ΔTn10 fragment carrying tet(B) gene was inserted (Hamidian and Hall, 2017b). This configuration resembled that of an unnamed RI identified by Bi et al. (2020) that also carried a second copy of AbaR4 inserted within tet(B) gene. AbGRI1-like-2 was solely carried by M14. The third variant of AbGRI1 was a Tn2006-interrupted AbGRI1-1. It had an atypical AbGRI1 structure lacking the plasmid-derived linker. Even though AbaR4 was reported not to have any variants (Hamidian and Hall, 2018), the Tn2006-interrupted AbGRI1-1 was named by Seputiene et al. (2012) as AbaR4b. For convenience, this name was used in the context of this study. In addition to conferring resistance to carbapenems, AbaR4b confers resistance to tetracycline and streptomycin/spectinomycin and was first identified in A. baumannii strains from Lithuania (Seputiene et al., 2012). In Korea, AbaR4b was found in 22.4% of the tested isolates, and all belonged to GC2 (Kim et al., 2017).

The IS26-bracketed RIs AbGRI2 and AbGRI3 co-existed in all GC2 isolates. While AbGRI2 occurred in only one configuration (AbGRI2-15), two variants of AbGRI3 with different resistance spectra were found. AbGRI2 has been reported most commonly in GC2 isolates carrying all or some of the resistance genes bla_TEM, aphA1, catA1, and [sul1, aadA1, and aacC1] within a class I integron (Nigro et al., 2013). AbGRI2-15 was first identified by Liepa et al. (2021) in A. baumannii strain collected from a patient in Lebanon. The island is a short version of AbGRI2 in which only aphA1 and bla_TEM−1 remained after IS26-mediated deletion and recombination events, as reported by the author. Each of the GC2 isolates carried either of the two variants AbGRI3-1 or AbGRI3-4. AbGRI3-1, also known as Tn6180, was first identified in A. baumannii strains MDR-TJ (GenBank accession: CP003500.1) and TYTH-1 (GenBank accession: CP003856.1) conferring resistance to four classes of antimicrobial agents (Blackwell et al., 2017). It was the least frequently identified among AbGRI3-type islands identified in a collection of A. baumannii isolates from Singapore in a study by Blackwell et al. (2017). In the same study, AbGRI3-4 was the dominant one identified in 46.6% of isolates. AbGRI3-4 is a short version of AbGRI3, most importantly, retaining armA. Blackwell et al. (2017) also highlighted the successful dispersion of AbGRI3-4 in isolates from India and Sweden.

Genomic island sul2 (GIsul2) is an integrating element first described by Nigro and Hall (2011) in the chromosome of A. baumannii strain ATCC 17978 isolated in France in 1951. Within GIsul2 is located a dihydropteroate synthase type 2-coding gene (sul2) preceded by an ISAba1. While thought to provide a promotor for sul2 (Nigro and Hall, 2011), high-level sulfonamide resistance in ATCC 19606 conferred by sul2 was not accompanied by an upstream ISAba1 (Hamidian and Hall, 2017a). GIsul2 was reported as the main vehicle for the mobilization of the sul2 gene (Hamidian and Hall, 2017a). It is also the donor of [CR2-sul2] elements in AbGRI1-type RIs integrated through homologous recombination. Here, GIsul2 was solely identified in GC1 isolates and those found to be phylogenetically related.

Beyond merely defining the resistance genes standing behind MDR and XDR in A. baumannii, WGS was used in the current study for defining the genomic RIs carried by clinical isolates of A. baumannii from Egypt. Epidemiology analysis of the isolates showed a significant representation of the high-risk GCs, particularly GC2. Novel ST (ST2329^Oxf), resistance gene (bla_ADC−258), and IS (ISAba61) were identified. RIs were carried by the majority of the isolates most frequently co-existing in GC2. They were loaded with several genes conferring resistance to various antimicrobial classes and hotspots for the acquisition of more resistance genes. AbGRI1-type RIs showed up with the widest diversity, including two novel configurations.

The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found in the article/Supplementary Material.

The study was performed in accordance with relevant guidelines and regulations, and no experiments were performed on humans and/or human tissue samples. The study was approved by the local Ethical Committee of the Clinical and Chemical Pathology Department, Kasr Al-Ainy Hospital, Cairo University. Bacterial isolates were only collected as part of the routine patient care, and informed consent was not required.

MZ, AH, MA-A, HR, and SH contributed to the study design, performance of experiments, and data analysis. SH performed the genome assembly, bioinformatics analysis, and wrote the first draft of the manuscript. All authors read and approved the final version of the manuscript.

The authors thank the Deanship of Scientific Research at King Saud University for funding this work through Project No. RGP-038.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.