In the post-antibiotic era, where the emergence of carbapenem-resistant Enterobacterales (CRE) has been reported worldwide, colistin is increasingly being prescribed as a last-resort antibiotic due limited treatment options [1]. Colistin (polymyxin E) exerts bactericidal activity against most of the Gram-negative pathogens via mechanisms involving the outer membrane (OM) disruption and the neutralisation of lipopolysaccharides. The OM of Gram-negative bacteria comprises an asymmetric and compositionally unique bilayer, with phospholipid being the inner leaflet and outer leaflet studded with lipopolysaccharides (LPS) or lipooligosaccharides (LOS) [2]. The OM plays a key role as a permeability barrier against various toxic compounds including antibiotics and detergents. Regulations of the OM barrier are triggered by environmental signals; they play a significant role in promoting antimicrobial resistance by shielding the bacteria from toxicity and selectively allowing for the uptake of the nutritional compounds required for the bacteria through porins [3]. Additionally, the bacterial efflux pump systems that traverse both the cytoplasmic and outer membranes are thought to be synergistically working with the OM in conferring antibiotic resistance [4].

While the majority of the antibiotics such as beta-lactams rely on the pore-forming porins to pass through the OM for intracellular processes, the low permeability of the bacterial OM has been identified as a challenging barrier for antibiotic sensitivity. In order to overcome these obstacles, there is a need for research to shift focus on antimicrobials with OM-targeting properties. Although the clinical use of OM-targeting colistin has been revisited, the unwelcome trends of colistin resistance among the Gram-negative bacteria followed soon after this, demanding global attention. Since the discovery of mobilised colistin resistance (mcr) genes, which cause lipid A modification [5], a wide variety of colistin resistance related to horizontal gene transfer has been described [6]. The effects of mcr expression on the global physical properties of bacterial membranes were previously discussed elsewhere [7]. Resistance to colistin can also arise through mutations in the chromosomal genes that alter the membrane properties. The modifications of the PmrA/PmrB and PhoP/PhoQ two-component systems and the inactivation of the mgrB gene (a regulator of the PhoP/PhoQ system) are known to be majorly involved in colistin resistance via LPS modification [8]. Among these, mgrB mutations seem to be common and increasingly reported in Klebsiella pneumoniae as compared to other Enterobacterales ^{[9][10][11][12]}[9,10,11,12]. In K. pneumoniae, mgrB mutations which involve substitution, disruption, or inactivation have been identified as playing a prominent role in mediating colistin resistance [8].

2. Action of Colistin on Bacterial OM

The antibacterial activity of colistin targets the OM; as such, it is of a narrow spectrum and is potent mostly against Gram-negative bacteria, but not Gram-positive bacteria. It is active against most of the Enterobacterales including Klebsiella spp., E. coli, Salmonella spp., Enterobacter spp., Citrobacter spp., and Shigella spp., but Proteus spp. and Serratia spp. are intrinsically resistant to colistin. Categorised as cationic antimicrobial peptides (CAMPs), the colistin compound is positively charged and thus binds to the negatively charged phosphate groups of lipid A on the LPS via electrostatic interaction. The lipid A is an important building block of LPS which plays a critical role in maintaining bacterial permeability. Colistin acts competitively by displacing the divalent cations of calcium (Ca²⁺) and magnesium (Mg²⁺)—ions that otherwise help in stabilising the LPS molecules [13]. This leads to a series of processes of the loss of membrane integrity, the loss of the osmotic integrity of the cell membrane, the leakage of cell contents, and, subsequently, cell death [14]. Although LPS has been identified as the initial target of colistin activity, the exact mode of action of colistin remains unclear. Furthermore, colistin also exerts anti-endotoxin activity of lipid A through binding and neutralising the LPS molecules [13]. Mularski et al. (2016) examined the effect of coslitin on the OM of wild-type K. pneumoniae using cryo-electron microscopy (cryo-EM). Cryo-EM presented the clear visibility of the cytoplasmic membrane, peptidoglycan, OM, and fimbriae of the bacterial cell. The wide-type cells showed significant membrane damages after 2 h of exposure to colistin, including OM rupture, peptidoglycan discontinuity, and membrane blebs [15]. It has been demonstrated that the membrane damage by polymyxin molecules occurs in a concentration-dependent manner. At high polymyxin concentrations, the antibiotic molecules will form aggregates at the bacterial surface, leading to large physical defects [7]. Due to the non-specific nature of polymyxin interactions with membranes, polymyxins are avidly bound to the brush-border membrane of the kidney cells, leading to its undesired nephrotoxicity. Thus, attempts have been made to add antioxidants to the peptide molecules with the aim of suppressing polymyxin-induced membrane damage [16].

Overview of Colistin Resistance in K. pneumoniae

Before we divinge into the profound mechanism of  mgrB gene mutations leading to colistin resistance in K. pneumoniae, an overview of other molecular resistance mechanisms identified will be discussed here and summarised in Table 1. Kim et al. (2014) reported that the PmrA/PmrB and PhoP/PhoQ two-component systems are upregulated in K. pneumoniae upon exposure to colistin. The upregulation of pmrA/pmrB can be caused by mutations in pmrA or pmrB, followed by the upregulation of pmrC and the arnBCADTEF operon, which subsequently results in the synthesis of L-Ara4N (4-amino-4-deoxy-_L-arabinose) and PEtn (phosphoethanolamine) to lipid A [17]. The crosstalk between another two-component system, QseB/QseC, with PmrA/PmrB in Escherichia coli has been elucidated; however, the role of QseB/QseC in conferring colistin resistance in K. pneumoniae is not well characterised [18]. Nevertheless, recent studies showed that alterations in QseC contributed to polymyxins resistance ^[19][20][19,20]. Gene transformation experiments also proved that mutations in yciM and lpxM produced cells with decreased colistin susceptibility ^[19][21][19,21]. The role of yciM and lpxM in contributing to colistin resistance still needs to be mechanistically elucidated. It has been suggested that both mutations are associated with LPS production and lipid A modification because yciM is responsible for LPS biosynthesis regulation in E. coli [22], while lpxM encodes for lipid A acyltransferase [21]. Additionally, the CrrA/CrrB two-component system’s crosstalk with the PmrA/PmrB regulatory pathway has also been proposed to be mediated by the connector protein CrrC [23]. Mutations in crrB leading to CrrC expression have been found to result in reduced colistin susceptibility ^[19][23][19,23]. Apart from LPS and lipid A-associated resistance mechanisms, efflux pumps in K. pneumoniae, including AcrAB and KpnEF, can also be responsible for reducing colistin sensitivity. The acrB knockouts and kpnEF mutants have been demonstrated to display restored sensitivity towards a wide range of antibiotics, including polymyxin ^[24][25][24,25]. Before the discovery of the plasmid-mediated colistin resistance gene, mcr-1, which was reported in 2015 [5], chromosomal mutations were the only known mechanisms for acquired colistin resistance. The mcr-1 gene encodes for the PEtn transferase, which adds the PEtn residue to the lipid A moiety. The gene was initially discovered from the isolates from food animal farms, and it has since been identified to be circulating among the Enterobacterales. K. pneumoniae was one of the initial species that carried mcr-1 [5]. Since then, nine other mcr genes (mcr-2 to mcr-10) have been discovered and disseminated worldwide [26]. So far, mcr-1, mcr-3, mcr-7, mcr8, and mcr-10 have been reported in K. pneumoniae ^[27][28][29][27,28,29]. A cell morphology study showed that cells overexpressing mcr-1 presented varying marked increases in the thickness and density of the cell envelopes as compared with cells carrying empty plasmid ^[30][31][30,31]. The increased expression of mcr-1 was also concomitantly observed with a reduced cell growth rate and viability. This observation has been suggested to be related to the cell fitness and survival [31]. In an induction experiment using recombinant E. coli, it was observed that higher expression levels of mcr-9 genes were induced by sub-inhibitory concentrations of colistin, and this inducible expression was shown to be related to the QseC-QseB two-component system [20]. While both of the mgrB and mcr-1 genes confer colistin resistance through LPS modifications, Zhu et al. (2021) demonstrated that the coexistence of the mcr-1 gene and chromosomal mutations posed a fitness cost of the K. pneumoniae mutant. The study reported that mcr-1-negative strains harboured mutations in phoQ and mgrB more frequently, while crrA and pmrB mutations occurred more frequently in the mcr-1-positive strains [32]. A cloning study demonstrated that the mcr-1 gene has no impact on colistin resistance when it coexists with the inactivated mgrB gene in clinical K. pneumoniae [33]. The finding of this experiment suggests no synergistic effect of both genes, except for the leading role played by mgrB inactivation in conferring colistin resistance. The biological cost of colistin resistance is associated with bacterial growth rates, virulence, and transmissibility, which consequentially influence the evolution of antibiotic resistance. The insertional inactivated mgrB mutants were found to be stable even without the presence of antibiotic selection, and did not cause any significant fitness cost to the bacterial host experimentally [9]. A further epidemiological observation using the murine gut colonisation model corroborated that the reduced biological cost of mgrB mutants is an important mechanism for enhanced survival outside the host and host-to-host transmission in K. pneumoniae [34]. The consequences of mgrB mutation were further underscored by an experiment using a waxworm infection model to demonstrate that inactivated MgrB contributed to the heightened hypervirulence of K. pneumoniae [35].

3. MgrB as a PhoP/PhoQ Regulator

Olaitan et al. (2014) observed that mutations in mgrB contribute more in colistin resistance among the K. pneumoniae compared to the mutations involved in other two-component systems, such as pmrA/pmrB and phoP/phoQ [8]. In order to ascertain the importance of these two-component systems in mgrB-mediated lipid A modifications, the lipid A moieties synthesised by mgrB-phoQ and mgrB-phoQ-pmrAB mutants as well as the mutants complemented with phoPQ were compared. It was found that the previous mutants (mgrB-phoQ and mgrB-phoQ-pmrAB) resembled that of the wild-type, while the mutants complemented with phoPQ resembled the lipid A produced by the mgrB mutant. This further confirmed that the inactivation of the mgrB gene giving rise to colistin resistance can be PhoPQ- but not PmrAB-dependent [35]. MgrB is a small, 47-amino acid regulatory transmembrane protein which negatively regulates the PhoP/PhoQ system in Gram-negative bacteria [36]. A pioneering study by Poirel et al. (2015) demonstrated that a premature stop codon in the sequence of mgrB, leading to truncated MgrB with only 29 amino acids, is a key target for colistin resistance in K. pneumoniae [11]. However, it is understudied how such a small protein exerts a great impact on the PhoP/PhoQ pathway. Previous work has suggested that MgrB acts by directly inhibiting PhoQ histidine kinase activity, thereby modulating a negative feedback loop ^[36][37][36,37]. Studies have suggested that MgrB may interact with other proteins aside from PhoQ for stress responses. In acidic conditions, the inactivation of MgrB resulted in the increased accumulation of RpoS, the stress regulator, through the regulation of IraM expression ^[34][38][34,38]. In the absence of functional MgrB, PhoQ over-activation and the accumulation of RposS were found to be significant contributors during the environmental survival of K. pneumoniae [34]. In E. coli, a recent work suggested that the MgrB transmembrane and periplasmic regions are necessary membrane anchors establishing the physical interaction with PhoQ but are not sufficient to inhibit PhoQ. Further investigation also pinpointed a number of functionally important residues spread across the protein that are important for PhoQ’s inhibitory function. The study also tested the expression levels of MgrB orthologs from K. pneumoniae and other related enterobacterial species in E. coli. As expected, MgrB from K. pneumoniae, Serratia spp., and Salmonella Typhimurium showed comparable reporter activities to that of E. coli MgrB, indicating that the MgrB orthologs are generally highly conserved [39]. Interestingly, although MgrB orthologs are found several Enterobacterales, further studies are clearly needed to decipher how mutations in mgrB have emerged as a predominant mechanism for acquired colistin resistance, as compared to other species, especially E. coli.