Hospital-associated infections (HAIs) present an elevated healthcare burden in both developed and developing countries (Chng et al., 2020). Acinetobacter baumannii is implicated in a considerable fraction of difficult to treat HAIs (Ayobami et al., 2019). Antimicrobial resistance, biofilm formation, and resistance to desiccation are among the competencies contributing to the environmental persistence and the epidemic potential of this species (Antunes et al., 2014). In addition to its intrinsic resistance to multiple antimicrobial classes, effective therapeutic options are being gradually depleted by the extraordinary ability of A. baumannii to acquire and upregulate resistance genes (Di Nocera et al., 2011). The emergence of multidrug-resistant (MDR) and extensively drug-resistant (XDR) A. baumannii has been increasing worldwide as well as in Egypt (Tal-Jasper et al., 2016; Elsayed et al., 2020). This forced the WHO to declare carbapenem-resistant A. baumannii as a category 1 (critical) priority pathogen for which novel therapeutic antimicrobials are urgently required (Tacconelli et al., 2018).

The New Delhi Metallo-β-lactamase-1 (NDM-1) is a carbapenemase that has been frequently linked to the XDR phenotype owing to its association with mobile elements loaded with other resistance genes (Wailan and Paterson, 2014). A. baumannii has been long recognized as an intermediate reservoir for bla_NDM-1 genes in which the harboring transposon (Tn125) was built and subsequently transmitted to other Gram-negative species (Toleman et al., 2012; Bontron et al., 2016).

Genome studies contribute significantly to better comprehend the molecular basis and evolution dynamics of antimicrobial resistance in nosocomial infectious pathogens (Hendriksen et al., 2019). Despite the large number of studies from Egypt that have discussed the epidemiology of healthcare-associated A. baumannii (Al-Hassan et al., 2019; Benmahmod et al., 2019; Wasfi et al., 2021), few studies have explored the whole-genome sequence of those circulating in Egyptian hospitals (Fam et al., 2020).

The objective of the current study was to explore the genomic features of four XDR bla_NDM-positive A. baumannii clinical isolates recovered from hospitalized patients at a large tertiary hospital in Egypt by whole-genome sequencing (WGS).

During the study period, a total of 20 A. baumannii isolates were recovered from 20 different hospitalized patients with age ranging between newborn (5days) and 65years old. Of these, 12 (60%) were females and 8 (40%) were males. More than half of the patients were hospitalized in intensive care units. Specimens were collected from different clinical sites (Table 1). Results are shown for the four bla_NDM-positive A. baumannii isolates.

A threatening rise in the incidence of carbapenem-resistant A. baumannii has been increasingly reported worldwide (Levy-Blitchtein et al., 2018; Moghnieh et al., 2018; Alcantar-Curiel et al., 2019) and in Egypt as well (Al-Hassan et al., 2019; Benmahmod et al., 2019; Mabrouk et al., 2020), leaving behind a substantial number of difficult to treat infections. For a deeper insight into the molecular mechanisms underlying carbapenem resistance in this highly problematic pathogen, a collection of 20 A. baumannii clinical isolates was screened for carbapenemase-coding genes by PCR. Four NDM producers were identified in clinical specimens recovered from ICU patients with severe infections. XDR phenotype was identified in all bla_NDM-positive A. baumannii with few reserved therapeutic options. These included tigecycline, colistin (for M02, M11, and M20), and amikacin (for M19) frequently associated with unfavorable pharmacokinetics and/or adverse effects particularly in critically ill patients (Spapen et al., 2011).

Draft genomes of the bla_NDM-positive isolates were obtained by Illumina sequencing for subsequent MLST and resistome analysis. In silico MLST and goeBURST analysis revealed that three out of four NDM producer A. baumannii belonged to the high-risk international clones (ICs), known for outbreak potential, worldwide dissemination (Karah et al., 2012), and multidrug resistance (Diancourt et al., 2010). M02 and M11 were assigned ST85^Pas/1089^Oxf that belong to IC9, recently described by Müller et al. (2019). Abundance of studies reporting bla_NDM-1-positive A. baumannii of ST85^Pas from Middle East countries (Bonnin et al., 2013; Decousser et al., 2013; Rafei et al., 2014; Salloum et al., 2018) has drawn attention on its probable endemicity in this region. IC2 was represented only by M20 (ST570^Pas/1701^Oxf), whose genome was loaded by the highest share of resistance genes. The abundance of IC2 A. baumannii in Egypt was also reported by others (Al-Hassan et al., 2019; Wasfi et al., 2021). To the best of our knowledge, this is the first report of bla_NDM-positive A. baumannii strain that belongs to ST164^Pas/ST1418^Oxf. Although MDR-resistant A. baumannii isolates that belong to ST164^Pas have been increasingly reported from different parts of the world (Coelho-Souza et al., 2013; Loraine et al., 2020; Tada et al., 2020), none was reported to carry a bla_NDM gene.

The SNP-based phylogeny analysis (Figure 1) showed that the isolates M02 and M11 were genetically related to two bla_NDM-positive A. baumannii strains of the same sequence type (1089^Oxf/85^Pas). These included A. baumannii strain Cl300 isolated in Lebanon in 2015 and strain ACN21 isolated in India in 2018. M20 was found to be genetically related to two NDM producer A. baumannii strains isolated in United States in 2016 (TP2 and TP3). Both TP2 and TP3 had the Oxford ST1578 a double locus variant of ST1701 to which M20 belongs. On the other hand, M19 showed no genetic relatedness to any of the NDM producer A. baumannii strains for which complete genomes were available in the NCBI. Inclusion of four draft genomes that belong to ST164^Pas and ST1418^Oxf retrieved from PubMLST genome collection revealed that M19 was most genetically related to A. baumannii strain CPO20190206 isolated from Denmark (ST164^Pas) and A. baumannii strain JorAb40 isolated from Jordan (ST1418^Oxf). Both strains were isolated in 2019 and, interestingly, none was found to carry a bla_NDM gene.

Resistome analysis disclosed a wide arsenal of resistance genes presented in Table 2 and correlated with the susceptibility profiles in Figure 2. Both intrinsic and acquired resistance mechanisms contributed to β-lactams resistance. The carbapenem-hydrolyzing class D β-lactamases (oxacillinases) provide both intrinsic (bla_OXA51-like genes) and acquired (bla_{OXA-23, 40, 58, 143,235}-like genes) resistance to β-lactams including carabapenems (Poirel and Nordmann, 2006; Ghaith et al., 2017). Overexpression of OXA-type β-lactamases has been linked to an upstream IS element, most frequently ISAba1, through which an additional promotor is provided (Evans and Amyes, 2014). bla_OXA-94 preceded by ISAba1 element was identified in M02 and M11 while no IS elements could be identified upstream to bla_OXA-91 or bla_OXA-66 carried by M19 and M20, respectively.

In addition to intrinsic OXA-type β-lactamases, the IC2 isolate (M20) also carried bla_OXA-23 the most widely disseminated oxacillinase acquired by carbapenem-resistant A. baumannii (Mugnier et al., 2010). Association of bla_OXA-23 with IC2 A. baumannii has been reported worldwide (Hamidian and Nigro, 2019). As with other IC2 isolates, bla_OXA-23 carried by M20 was found to reside in Tn2006 in which the gene is bracketed by two inversely oriented ISAba1 elements. Tn2006 is the most common structure harboring bla_OXA-23 either alone or incorporated into AbGRIs (Hamidian and Nigro, 2019).

Although the association of bla_OXA-91 and bla_NDM-1 in A. baumannii was not previously described in Egypt, co-existence of bla_OXA-51-like, bla_OXA-23 and bla_NDM-1 was reported by Wasfi et al. (2021).

Analysis of the genetic environment of bla_NDM-1 in the sequenced isolates showed different environments in different sequence types. ISAba14 element was inserted upstream to the cutA gene in M02, M11, and M20. This was previously documented by Bonnin et al. (2013) who failed to identify a downstream second copy of ISAba125 by PCR and suggested loss of functionality of this truncated transposon (ΔTn125). WGS of bla_NDM-positive isolates by a later study (Vijayakumar et al., 2020) uncovered the existence of a second copy of ISAba125 downstream to the ISAba14-interrupted transposon. Interestingly, analysis of the upstream region to the truncated transposon revealed the existence of the amikacin resistance gene AphA6 preceded by another copy of ISAba14 in ST85^Pas isolates (Figure 4). The two ISAba14 elements were thus thought to form an alternative composite transposon in which two resistance genes were enclosed for transposition (bla_NDM-1 and AphA6) rather than the widely known Tn125 in which bla_NDM-1 was hosted as the sole antimicrobial resistance gene. Transposition of this composite transposon might, therefore, favor the co-transfer of resistance to two of the last-line antimicrobial treatment options for MDR and XDR A. baumannii. Nevertheless, experimental analysis is required to examine the transposition potential of this transposon. In M19, IS1 family transposase was identified immediately upstream to ISAba125 that precedes the bla_NDM-1 gene. Insertion of IS1 element in this location was not identified in the nucleotide database of the NCBI.

Intrinsic to all A. baumannii, cephalosporin resistance is mediated by ADC (formerly known as bla_AmpC). In addition to the incomplete bla_ADC-73 carried by M20, a novel bla_ADC allele (bla_ADC-257) with an upstream ISAba1 element was identified in M19 recovered from a blood culture of a female patient admitted to the ICU with fever of unknown origin. With no similarity in the NCBI nucleotide database, bla_ADC genes carried by M02 and M11 were interrupted by an IS1008 element (Figure 5). No alternative intact copies of bla_ADC were identified in M02, M11, or M20. Other β-lactamases identified here included class A β-lactamases, more efficiently capable of hydrolyzing penicillins and cephalosporins than carbapenems (Jeon et al., 2015). These were coded by bla_Tem-1 carried by M20 and bla_CARB-16 in M19. However, their association with mobile elements could not be clearly determined.

In addition to the intrinsic aminoglycoside resistance gene ant(3``)-IIc (Zhang et al., 2017), the amikacin-modifying enzyme-coding gene aph(3`)-VIa (aphA6) was found in all isolates. The predominance of aph(3`)-VIa among the aminoglycoside modifying enzymes-coding genes was also reported by others (Aghazadeh et al., 2013; Sheikhalizadeh et al., 2017). Notably, the gene was also identified in the amikacin-sensitive isolate M19. Identification of aph(3`)-VIa in amikacin-susceptible isolates was also reported before (Aghazadeh et al., 2013; Sheikhalizadeh et al., 2017). In ant(2``)-Ia-positive isolates, the gene was found in pRAY plasmid variants. pRAY is a 6 Kb plasmid widely distributed in Acinetobacter species comprising the most common resistance mechanism to gentamicin and tobramycin (Hamidian et al., 2012).

Acquired 16S rRNA methyltransferases constitute the most important aminoglycoside resistance mechanism conferring resistance to most of the clinically important aminoglycosides (Galimand et al., 2003). Of them, armA has been widely reported from A. baumannii particularly those of the IC2 (Blackwell et al., 2017). Within a 20,844bp genomic resistance island closely similar to A. baumannii genomic resistance island 3 (AbGRI3; Blackwell et al., 2017), armA gene was identified in M20 (IC2). Other resistance genes hosted by the genomic island include aph(3`)-Ia, aac(6`)-Ib`, catB8, ant(3``)-Ia, sul1, msr(E), and mph(E). Another unique genetic structure in which genes coding resistance to two different antimicrobial classes was identified in M02 and M11 (Figure 6). This included the chloramphenicol efflux pump (FloR)-coding gene and sul2, conferring resistance to sulfamethoxazole/trimethoprim, enclosed by insertion elements. The closest match to this region was shown by Acinetobacter indicus chromosome (GenBank accession: CP071319.1) from which it may have been acquired with some genetic rearrangement.

In the absence of plasmid-mediated quinolones resistance genes, nonsusceptibility to levofloxacin in all NDM producers investigated here was mediated by target site mutations. These affected the QRDRs within GyrA (S83L) and ParC (S80L) enzymes. The mutation pattern identified in our isolates was commonly reported as the predominant mechanism responsible for fluoroquinolones resistance in A. baumannii (Hamed et al., 2018; Nodari et al., 2020; Roy et al., 2021).

Resistance to colistin, the last line of defense against XDR A. baumannii, was evident in only one isolate (M19) that, fortunately, retained susceptibility to amikacin and tigecycline. Colistin resistance in M19 was accompanied by multiple nonsynonymous mutations affecting pmrABC and IpxACD genes. Missense mutations identified in pmrB (H89L) and pmrC (I42V) genes carried by M19 were also reported in colistin-resistant A. baumannii studied by Nurtop et al. (2019) in Turkey. It is worth mentioning that the amino acid affected by pmrB mutation identified here is located outside the histidine kinase domain, the main determinant of colistin resistance in A. baumannii (Arroyo et al., 2011; Beceiro et al., 2011; Lesho et al., 2013). Moreover, all lpxACD mutations identified here were previously reported in both colistin-susceptible and colistin-resistant isolates (Oikonomou et al., 2015; Haeili et al., 2018; Nurtop et al., 2019). Accordingly, novel unidentified resistance mechanisms might stand behind the high-level resistance (MIC≥128μg/ml) of M19 to colistin. Further investigations including gene expression analysis are therefore required to confirm or role out the impact of such mutations on colistin susceptibility.

Diverse efflux pumps, whose overexpression has been linked to multidrug resistance, were identified in the sequenced isolates. RND efflux pumps known by their broad substrate profiles (Coyne et al., 2011), including AdeABC, AdeIJK, and AdeFGH. were identified in all isolates. RND efflux pumps contribute to intrinsic resistance of A. baumannii to several classes of antimicrobials. Other multidrug efflux pumps carried by all isolates included AbeM, a member of the multidrug and toxic compound extrusion family efflux pumps and the small multidrug resistance efflux pump AbeS (Coyne et al., 2011). Except for FloR conferring resistance to phenicols in M02 and M11 only, efflux pumps of the MFS were disseminated in all sequenced genomes. With narrow substrate profiles, AmvA, AbaF, and AbaQ are known to extrude erythromycin, fosfomycin, and quinolones, respectively (Coyne et al., 2011; Perez-Varela et al., 2018). The macrolide-specific ABC pump MacAB was also found in all isolates.

It is worth mentioning that the current study suffers from some limitations, most importantly is using short-read sequencing technology instead of a hybrid long- and short-read sequencing approach known to produce more accurate genome organization. Consistent with other studies (Leal et al., 2020), resistance to some antimicrobials could not be correlated to known resistance genes highlighting the need for further investigations including gene expression analysis or identification of novel resistance determinants. Finally, only four genomes were sequenced here thus correlating resistance genes with particular STs could not be fully achieved.

The current study is one of the few studies reporting WGS of A. baumannii clinical isolates from Egypt. The isolates showed XDR phenotype and were recovered from ICU patients. High-risk international clones were identified, predominantly IC9 (ST85^Pas) widely reported from Middle East countries. Diverse mobile elements were associated with resistance genes with novel insertion sites and genetic organizations. Co-existence of amikacin and carbapenem resistance genes on an ISAba14-bracketed transposon was uniquely identified in ST85^Pas/ST1089^Oxf. bla_NDM-1 gene was identified, for the first time, on the chromosome of an A. baumannii strain that belongs to sequence type ST164^Pas/S1418^Oxf. WGS of the highly problematic MDR and XDR pathogens may aid in the identification of emerging resistance genes and their dissemination dynamics. Co-existence of resistance genes within mobile genetic elements could also be identified. This may aid in optimizing treatment guidelines to avoid selection of resistance to last-line antimicrobials. WGS also permits monitoring the emergence of novel global MDR clones and facilitates comparative genomic analysis and developing cheaper molecular techniques for routine screening.

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 clinical and chemical pathology department, Kasr Al-Aini Hospital, Cairo university. Only bacterial isolates were collected for the routine laboratory work to ensure patient care and informed consents were not required.

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.

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