The continual occurrence of mutations in non-susceptible pathogenic strains and the associated determinants of antimicrobial resistance within hospitals, countries, and across the world are currently the greatest threats to international Public Health (Shelenkov et al., 2021). The wide use of antibiotics in immunocompromised patients at intensive care units and the absence of antibiotic stewardship programs in hospital settings have led to the occurrence of pathogens that are multiple drug-resistant (MDR) and the rise of extensively drug-resistant (XDR) strains. Therefore, clinicians were forced to depend on colistin as the last treatment option to combat these infections (Seleim et al., 2022). A. baumannii, a particularly problematic pathogen poses a significant threat to public health, by causing severe and invasive infections that are associated with high mortality rates in immunocompromised individuals and patients receiving intensive care. MDR and/or XDR A. baumannii have been reported to cause a significant degree of infections in Egypt during recent years (Ghaith et al., 2017; Al-Hassan et al., 2019). However, the direness of the situation is often masked in Egypt and other developing countries because of the lack of surveillance systems that can be attributed to restricted financial resources. Epidemiological surveillance is one significant tactic to assess and combat the burden exerted by problematic MDR pathogens in hospital settings. Overcoming antibiotic resistance is a critical challenge in the treatment of Acinetobacter infections (Wong et al., 2017), especially in view of the rise in carbapenem resistance among this species. Carbapenem resistance is generally associated with MDR and XDR phenotypes, which have been frequently demonstrated in Acinetobacter (Infectious Diseases Society Of, A 2012). Due to this problematic carbapenem-resistant A. baumannii (CRAB) clinical isolates, numerous Egyptian hospitals have reintroduced polymyxin use for therapy. Polymyxins E or colistin, the last therapeutic option for MDR Gram-negative pathogen infections, is widely used for the treatment of CRAB infections (Chamoun et al., 2021). Mechanistically, the interaction between the cationic non-ribosomal lipopeptides of colistin, and the lipid A component of lipopolysaccharide (LPS) in the outer membrane of the cell envelope destabilizes the latter. This subsequently allows the uptake of polymyxins into the periplasm and increases permeability by disrupting both outer and inner membrane integrity (Moffatt et al., 2019). Acquired colistin resistance develops primarily via drug target alteration. Substitutions or mobile genetic element insertion or deletion in the genes involved in the biosynthesis of lipid A, may result in structural or functional modifications in the lipid A moiety (Chamoun et al., 2021). Two key mechanisms for chromosomally mediated colistin resistance have been identified. The first involves LPS lipid A phosphoester group modifications, subsequent to pmrCAB operon mutations, which alter the affinity of lipid A to polymyxins by reducing its net negative charge. The operon pmrCAB comprises the pmrC gene that encodes a phosphoEthanolamine (pEtN) transferase, in addition to pmrA and pmrB, which encode the two component system (TCS), PmrA/PmrB. Overexpression of pmrC is induced by mutations in TCS (mainly pmrA) leading to colistin resistance via pEtN-mediated modification of lipid A [8,9]. Discrete genetic events are required for an adequate colistin resistance level in A. baumannii, which are up-regulation of pmrAB, pmrB point mutations, which leads to pmrC overexpression, and further addition of pEtN to lipid A (Beceiro et al., 2011). The second mechanism essentially involves termination of lipid A production due to various nucleotide substitutions, deletions, and insertions in one of its biosynthesis genes (lpxA, lpxC, and lpxD) that leads to frameshift mutations or produce truncated proteins that damage lipid A biosynthesis (Olaitan et al., 2014; Moffatt et al., 2019). In colistin-resistant isolates, there are non-synonymous mutations in the lpxC (P30L or S, N287D) and lpxD (E117K) genes that were previously reported from different origins (Oikonomou et al., 2015). Both colistin-susceptible and colistin-resistant isolates were found to carry the amino acid substitutions N287D (lpxC) and E117K (lpxD). It was also reported the possibility of these alterations in amino acids together with a mutation in the pmrCAB operon could result in synergistic effect that causes colistin resistance (Nurtop et al., 2019; Jovcic et al., 2021; Usjak et al., 2022). Several studies have focused on investigating the determinants of antibiotic resistance in colistin-resistant A. baumannii isolates. In this study, we investigated pmrCAB and lpxACD mutations in A. baumannii strains with retained susceptibility to colistin. Additionally, we made a comparative analysis of mutation patterns in colistin-susceptible isolates from distinct sequence types (STs).

A total of 18 cultures that grew MDR/XDR A. baumannii were included in the study. Most patients were female and varied between neonates and older females. All the isolates were collected from ICU and NICU patients, and most cultures were obtained from patients with the greatest proportion of samples recovered from blood and wounds. Amplification of the bla _OXA-51-like gene confirmed that the collected isolates were A. baumannii. Most of the isolates belonged to GCs as shown in Figure 1 .

Treatment failure in A. baumannii is very common due to lack of understating the exact mechanisms of resistance to antibiotics together with the absence of novel therapies. In recent years, the alarming increase in the rates of CRAB strains in Egyptian hospitals have forced the reuse of colistin for treating these pathogens. Colistin is the antibiotic of choice for treating CRAB infections in healthcare facilities, a significant challenge to infection control in nosocomial settings. Elucidating the exact mechanisms that mediate and control colistin resistance is thus crucial to preserve its efficacy. To explore the relationship between chromosomally-mediated mechanisms of colistin resistance and the colistin resistance phenotype, we investigated 18 clinical isolates of A. baumannii collected from ICU and NICU patients by WGS. All of the isolates were previously characterized for MLST according to Pasteur and Oxford scheme. Some isolates were assigned two Oxford STs due to carrying two copies of gdhB locus, as described before (Gaiarsa et al., 2019). Most of the isolates belonged to high-risk GCs including GC1, GC2, GC7, GC9, and the most recently described clone, GC11 (Hansen et al., 2023).

Given the importance of studying resistance mechanisms in distinct strains, genetic investigation was carried out in colistin-susceptible strains. Detecting the MIC of colistin using the broth microdilution method is the only reliable method and remains the gold standard for assessing colistin resistance in A. baumannii approved by both the CLSI and by the EUCAST (CLSI, 2020; EUCAST, 2021). Of the 18 isolates analyzed, only a single strain, M19, demonstrated colistin resistance with an MIC ≥128 µg/ml. This percent of resistance was much lower than that was recently reported in Egypt in a study conducted on 17 A. baumannii isolates in which nine of them were colistin-resistant (Fam et al., 2020). It is noteworthy to mention that high rates of resistance toward different antibiotics were detected among our studied collection, and this could be attributed to the fact that high resistance patterns are usually observed in critically ill hospitalized patients due to existence of comorbidities and the overuse of antibiotics.

The plasmid-mediated colistin resistance determinants mcr genes were not detected in our studied isolates. This goes in line with most of studies conducted on colistin-resistant A. baumannii isolates. Moreover, the studies that reported the existence of mcr genes failed to sequence any mcr variants (Khoshnood et al., 2020; Snyman et al., 2020). Instead, pmrAB mutations appears to be the driving mechanism of colistin resistance in A. baumannii in most of published reports (Seleim et al., 2022). A total of 34 point mutations were identified in our strains ( Figure 2 ) within the pmrCAB and lpxACD operons. The association of colistin resistance in A. baumannii with mutations in the putative two-component regulatory system PmrAB was first reported in 2009 (Adams et al., 2009). Although previous studies have reported mutations in the response regulator-coding gene pmrA (Arroyo et al., 2011; Lesho et al., 2013), we observed none. Colistin resistance is primarily a consequence of mutations in the pmrCAB operon, especially in the pmrB region (Nurtop et al., 2019), and we observed the highest number of mutations in the pmrC region. Sequence analysis revealed eight amino acid substitutions in PmrB and 20 amino acid substitutions in PmrC, all of which were detected in colistin-susceptible isolates, except for the single colistin-resistant strain, M19. Thus, it may be inferred that not all amino acid substitutions result in elevated colistin MICs. Seleim et al. (2022) reported that the pmrC expression level could not differentiate between colistin-resistant and colistin-susceptible isolates, and they observed that the correlation between pmrC expression levels and colistin MICs was not significant (Seleim et al., 2022). Additionally, this was also reported by Gerson et al. (2020). Thus, it is therefore concluded that colistin resistance mechanisms in A. baumannii are much more complicated than believed. The resistant strain M19 harbored unique and distinct mutations in pmrC and pmrB, and none in pmrA. These included three substitutions in PmrC, including N300D, A370S, and V486I, and two substitutions, namely H89L and N440H in PmrB. Notably, V486I and H89L were only observed in this strain. The H89L substitution in PmrB (Nurtop et al., 2019) and the V486I substitution in PmrC (Srisakul et al., 2022) have been previously reported in colistin-resistant A. baumannii isolates. While found here in colistin-susceptible isolates that belonged to three STs, some pmrC mutations (I131V and H499R) and pmrB mutations (P360Q) were previously linked to resistance to colistin (Arroyo et al., 2011; Misic et al., 2018). In agreement with our findings, N440H and A444V mutations of pmrB were previously predicted not to affect colistin susceptibility (Thi Khanh Nhu et al., 2016). To completely understand lipid A modification and the resulting colistin-susceptible phenotype, the effects exerted by pmrCAB mutations on the expression of pmrC require further investigation.

A total of six diverse LpxCD amino acid substitutions in colistin-susceptible A. baumannii strains were evident. The colistin-resistant isolate M19 harbored the mutation N287D in lpxC that was also present in the colistin-susceptible isolates. A previous study reported that mutations in lpxD or pmrB alone may suffice to induce colistin resistance, thus suggesting synergism between the effects of mutations within these genes in promoting the same (Nurtop et al., 2019). The diverse range of amino acid substitutions reported in this study, some of which have been previously described, suggest that the exact colistin resistance mechanisms in A. baumannii requires extensive investigation. Interestingly, conserved mutation patterns were mostly found in A. baumannii strains that belonged to the same ST/GC, particularly those inferred by the Pasteur scheme, regardless of their susceptibility to colistin. This was further confirmed by an expanded analysis in which more strains with the same STs as those identified here were included in the mutation analysis. Hence, we here emphasize the importance of carefully analyzing colistin resistance-related mutations with reference to a colistin-susceptible strain that belongs to the same ST as the one under investigation. Otherwise, the mechanisms underlying colistin resistance may be incorrectly inferred and the real contributors to colistin resistance may be overlooked. We found many studies that investigated pmrCAB and lpxACD mutations with reference to A. baumannii ATCC 19606 and ATCC 17978 that may lead to overestimation of mutation-related resistance and overlooking the real mechanism of resistance (Mavroidi et al., 2015; Haeili et al., 2018; Nurtop et al., 2019; Jovcic et al., 2021; Eze et al., 2022; Usjak et al., 2022; Kabic et al., 2023).

As the susceptibility profiles of the global strains included in our study were revised, only two strains with confirmed colistin resistance were identified (MS14413 and Ab-NDM-1). Interestingly, the only unique mutations identified in the two strains were found in pmrB, confirming the crucial role of this gene in driving colistin resistance. Notably, other mutations identified by the authors as a contributor to colistin resistance in Ab-NDM-1 were the pmrC mutations I115V, N284D, and I326T (Fernandez-Cuenca et al., 2020). These correspond to the mutations I131V, N300D, and I342T (according to the numbering of A. baumannii ATC 19606, GenBank CP059040) identified here as polymorphism. This further emphasizes the importance of the data presented here for the correct interpretation of pmrCAB and lpxACD mutations.

The bacterial surface charge has frequently been described in terms of ZP, which is the potential at the shear plane of the electrical double layer surrounding a cell in solution (Soon et al., 2011). Our results showed that the colistin-resistant cells display less negative ZP than the colistin-susceptible cells. The less negative ZP exhibited by the colistin-resistant isolate M19 in comparison to that of the colistin-susceptible cells has been previously explained to be a consequence of alterations in the structure and composition of the outer membrane (Soon et al., 2011). Colistin-resistant cells were previously reported to have more propensity for clumping in small clusters compared to colistin-susceptible cells (Soon et al., 2009). This was justified by the higher colloid aggregate stability within the particle carrying a lower magnitude of charge due to reduced electrostatic repulsion (Klodzinska et al., 2010). Lipid A phosphates esterification with 2-aminoethanol or 4-amino-4-deoxy-L-arabinose resulting in charge shielding has been reported before in colistin-resistant E. coli (Gatzeva-Topalova et al., 2005) and Pseudomonas aeruginosa (Gooderham and Hancock, 2009). Unexpectedly, M10, a colistin-susceptible isolate, exhibited the lowest negative ZP among all the isolates. Notably, this isolate is phylogenetically related to M19 (the colistin-resistant isolate) as described previously (Hamed et al., 2022), and thus the aberrant zeta potential may either be a consequence of cell membrane alterations or a reflection of ST-related polymorphism. Statistical analysis of the ZP values of all isolates revealed a significant difference between the ZP values of the isolates that belonged to the same ST/GC. This contradicts our hypothesis that the isolates that belong to the same ST/GC may have similar ZP values. More studies are recommended to better understand the relationship between STs and the ZP of A. baumannii.

The existence of the alternative pEtN transferase named ethanolamine phosphotransferase A-1 (EptA) in A. baumannii was previously demonstrated by Lesho et al. (2013). The authors described two PmrC homologs designated EptA-1 and EptA-2 that were localized outside the pmrABC operon. The same study has also reported the overexpression of eptA-1 and eptA-2 in the colistin-resistant isolates. Another study conducted on a pair of MDR A. baumannii by Deveson Lucas et al. (2018) indicated that overexpression of the “orphan” eptA in a pre-colistin-treatment A. baumannii strain resulted in increased colistin resistance (Deveson Lucas et al., 2018). This finding provides evidence for eptA encoding a functional pEtN transferase that mediates colistin resistance (Deveson Lucas et al., 2018). In our collection, EptA-coding genes were harbored by the GC2 isolates (M01, M04, M05, M13, M16, M17, and M20), two GC9 isolates (M02, M11) and a single GC1 isolate, M12. None of the eptA genes identified here was preceded by ISAba1 insertion. This confirms the assumption made by Potron et al. (2019) that overexpression of eptA genes, and subsequently colistin resistance is a consequence of an upstream insertion of an ISAba1 element (Potron et al., 2019).

The insertion of ISAba125 within a gene encoding an H-NS family transcriptional regulator was previously linked to eptA overexpression and colistin resistance (Deveson Lucas et al., 2018). This was not found in any of our isolates.

Our study reinforces the need for extensive investigations for the elucidation of the exact mechanisms that contribute to colistin resistance in A. baumannii, which consequently influences possible therapeutic options. Colistin resistance mechanisms in A. baumannii are complex and not easily understood. Multiple mutations in pmrCAB and lpxACD are unlikely to result in increased colistin resistance. The evaluation of mutations with reference to colistin-susceptible isolates of the same ST/GC is essential to avoid misinterpretation of ST/GC-related polymorphisms. Evidence was provided by expanding our analysis to include A. baumannii strains with the same STs as our isolates retrieved from a public database. Further, large scale studies including more colistin-resistant and colistin-susceptible strains are recommended to better establish the relationship between the ZP and colistin resistance and to confirm, with statistical evidence, the ST-related polymorphism of pmrCAB and lpxACD operons that may exist in both colistin-susceptible and colistin-resistant strains.

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 .

Conceptualization, MZ, AH, MA, HR, and SH. Methodology, MZ, AH, and SH. Software, SH. Writing—original draft preparation, MZ. Writing—review and editing, MZ, AH, MA, HR, and SH. Visualization, SH. Funding acquisition, MA. All authors have read and agreed to the published version of the manuscript.