Targeting Acinetobacter baumannii resistance-nodulation-division efflux pump transcriptional regulators to combat antimicrobial resistance

Whole genome sequencing studies have improved species identification within Acinetobacter compared to traditional 16S rRNA amplicon-based methods revealing a greater diversity in their environmental niches and more antibiotic-resistance genes than previously thought in this genus¹⁶. Recent phylogenomic studies of Acinetobacter species reveal no distinct clustering between clinical and non-clinical isolates¹⁷. Furthermore, surveillance studies reporting the presence of Acinetobacter drug-resistance genes outside of clinical settings, including in environmental¹⁸ and animal isolates¹⁹, highlight that Acinetobacter is a One Health problem²⁰. This raises concerns about future threats from gene transfer to non-resistant species²¹. Indeed, Acinetobacter species are noted for the intrinsic resistance genes found in both their core and accessory genomes^22,23. This intrinsic resistance, combined with significant genomic plasticity, allows Acinetobacter species to readily acquire or lose genes and enhances their adaptability^24,25. Such versatility also enables tolerance of extreme conditions—desiccation²⁶, oxidative stress²⁷, and acidic pH²⁸—further contributing to their overall resilience.

Efflux pumps are membrane transporters capable of actively expelling substrates—including antimicrobial agents—into the periplasmic or extracellular environment thus preventing the substrate from reaching its intracellular target. As such, efflux pumps are often the primary defense mechanism against antimicrobial compounds, though other protective mechanisms can be engaged (i.e., enzymatic modification of the drug, target site mutation, reduced membrane permeability)^29,30. Increased expression of efflux pumps alone can effectively confer multidrug resistance as many of the pumps are capable of expelling a diverse range of structurally dissimilar drugs³¹.

Different efflux pump families have been identified in Acinetobacter species including multidrug and toxic compound extrusion (MATE), ATP-binding cassette (ABC), proteobacterial antimicrobial compound efflux (PACE), major facilitator superfamily (MFS), small multidrug resistance (SMR), and resistance-nodulation-division (RND) pumps³². The most clinically relevant efflux pump family with respect to AMR infections caused by Acinetobacter species is the RND pumps^33,34. Since the expression of these pumps is typically tightly regulated, strategies that target these controlling regulatory pathways could be key to preventing overexpression of efflux pump genes and consequently impairing removal of toxic compounds³⁵. This approach may pave the way for the development of new antimicrobial compounds in the form of drugs that limit efflux pump gene expression which in turn make the bacteria more susceptible to other assaults.

Spanning the inner and outer membranes, RND efflux systems are typically comprised of an inner membrane RND transporter, a periplasmic adapter protein (PAP), and an outer membrane factor (OMF). The PAP and the RND transporter genes are typically encoded as a single operon which may or may not include the OMF gene³⁶. They form a continuous channel across the cell wall of Gram-negative bacteria thereby facilitating the extrusion of its substrates directly to the extracellular environment³⁷. RND efflux pumps have a broad substrate specificity and so contribute significantly to the MDR phenotype^38,39.

In Acinetobacter species, a total of nine RND family transporters have been identified²² of which the most clinically relevant efflux pumps are AdeAB(C), AdeIJK, and AdeFGH (Table 1). (Note that not all Acinetobacter species have AdeC; AdeAB(C) indicates that AdeC may be present.) Of these three, overexpression of adeAB(C) is most commonly seen in MDR clinical isolates^33,36. Present in the core genome of all Acinetobacter species, AdeIJK is believed to be the ancestral efflux pump of this genus²². AdeFGH is the least studied efflux pump, with its full clinical relevance yet to be determined^40,41.

Increased expression of efflux pump genes is a key determinant in shaping the antimicrobial resistance phenotype⁴². Specific activators and repressors mainly control expression at the transcriptional level, but there is also evidence of some crosstalk between the regulators such that the expression of efflux pump operons may be influenced in multiple ways⁴³. Specifically, the response regulator AdeR of the AdeSR two-component system may potentially interact with the BaeS histidine kinase of the BaeSR two-component system to induce adeAB(C) expression^44,45 (both systems discussed in the section below) (Fig. 1). Notably, the deletion of one efflux pump often results in the upregulation of another, underscoring the need for a comprehensive understanding of their regulatory networks³⁹. The primary regulatory mechanisms of the three most clinically relevant RND pumps in Acinetobacter species are very distinct from each other: AdeAB(C) and AdeFGH have dedicated local regulators, whereas the regulator of AdeIJK is quite distant from the operon. Global regulators involved in altering the expression of each of the efflux pumps, which are equally as important as the specific regulators, remain poorly understood⁴³.

A adeIJK, B adeFGH, and C adeAB(C) pump encoding operons. AdeFGH efflux pumps in A. baumannii. Solid arrows represent activation of expression, while blunt-ended arrows indicate repression of transcription. Dashed lines denote mechanisms of activation or repression that are not yet fully understood. The effectors for upregulation and downregulation of efflux pump operons are listed in boxes, with green and red arrows indicating upregulation and downregulation, respectively. For example, polyunsaturated fatty acids (PUFA) and (p)ppGpp upregulate the expression of AdeIJK, whereas pleural fluid (PF) and human serum albumin (HSA) downregulate it. Additionally, the figure highlights five druggable targets (I, II, III, IV, and V) of two-component systems that could potentially modulate efflux gene expression. Created in BioRender. Kumar, A. (2024) https://BioRender.com/z94p008.

Full size image

AdeAB(C) is the often-overexpressed efflux pump operon in clinical isolates with the MDR phenotype^46,47,48. The dedicated regulator controlling the activation of adeAB(C) expression is AdeSR, a two-component system (TCS) regulator consisting of a membrane-spanning histidine kinase (AdeS) and a cytoplasmic response regulator (AdeR)⁴⁹ (Fig. 1). The histidine kinase element detects environmental signals, including antibiotics, and undergoes a conformational change to facilitate autophosphorylation of a cytoplasmic-facing histidine residue. This phosphate group is then transferred to a well-conserved aspartate residue on the response regulator which triggers its binding to DNA^50,51. AdeSR is the most studied RND efflux system regulator thus far as the structure of AdeR has been solved^52,53 and binding site of AdeR to DNA is known^53,54. AdeR is unique compared to other response regulators in that it binds within the intercistronic region between adeR and adeA instead of within the defined promoter region of adeAB(C)⁵⁴, possibly facilitating recruitment of the σ subunit of RNA polymerase and activating downstream gene expression⁴⁵. AdeR is also able to bind DNA without activation by phosphorylation allowing for the low-level, basal expression of AdeAB(C)^53,55, and may engage in self-regulation enabling a rapid respond to external stimuli, both crucial features for survival in the sudden presence of deleterious drugs.

Studying drug-resistant isolates, researchers discovered that increased activation of AdeAB(C) was mostly due to substitutions in either the autophosphorylation site of AdeS or the receiver domain of AdeR^{56,57,58,59,60}. These substitutions led to enhanced sensitivity of AdeS to environmental stimuli or an increased rate of AdeR phosphorylation, respectively^58,59,61. Conversely, strains overexpressing AdeAB(C) due to inactivation of AdeS^51,62,63 have also been reported. These opposing findings emphasize the complexity of this regulatory system. Another intriguing observation is that deletion of adeS in A. baumannii led to increased sensitivity to sulbactam, which is not a substrate of AdeAB(C) while conferring resistance to tigecycline, which is a substrate⁵⁶. This “seesaw effect” could be due to an underlying, broader regulon involving AdeR. This is not uncommon for TCSs; indeed, there are indications that AdeRS influences A. baumannii motility, biofilm formation, and virulence, indicating a wider regulome^49,64,65. Additionally, biophysical and structural data have revealed two amino acid mutations within the dimerization site of AdeR that abolishes dimer formation and subsequent DNA binding⁵³. Further molecular dynamic studies of AdeS and AdeR—and mutations of each—are needed to better understand its response to antimicrobial substrates and the consequent regulation of AdeAB(C).

AdeAB(C) may also be modulated by other regulatory systems (Fig. 1). As mentioned above, BaeSR is a TCS and global regulator that modulates the expression of AdeAB in the presence of tigecycline in A. baumannii^44,66; however, DNA binding studies showed that BaeR does not directly bind to the upstream region of adeAB(C), indicating possible regulatory crosstalk with AdeSR⁴⁴. Another potential master regulator that may repress adeAB(C) is the broad stress-responsive protein DksA. DksA binds to RNA polymerase and allosterically modulates its activity, thereby modulating transcription^67,68. Additionally, the histone-like nucleoid structuring protein (H-NS), a DNA-binding factor, may also act as a potential repressor of adeAB(C)⁶⁹.

Activity of the TCSs is not only modulated by antibiotics but also by diverse environmental stimuli such as pH, temperature, and osmolarity^50,70. We previously demonstrated the responsiveness of AdeSR to saline stress in A. baumannii⁶⁴ and many molecules, including human serum albumin⁷¹ and polyamines within blood serum, have been shown to increase AdeAB(C) expression^72,73. This highlights the responsiveness of the AdeSR system to host-conferring conditions encountered during infection independent of the stress created by the presence of antimicrobial drugs. Understanding this is vital as any factor that affects mutation within the regulatory components may lead to enhanced antibiotic resistance due to efflux pump dysregulation. Appreciating all the triggers that influence the expression of AdeAB(C) could aid in the development of effective efflux inhibitors and help control MDR A. baumannii.

AdeN is the known primary regulator of the AdeIJK efflux pump (Fig. 1). AdeN belongs to the TetR family of regulators which consist of a highly conserved helix-turn-helix motif in the N-terminus and a ligand-binding domain in the C-terminus⁷⁴. TetR regulators typically dimerize and bind to a palindromic sequence located upstream of a gene, activating or repressing its expression. AdeN operates as a repressor so that in the presence of an as-yet-unknown effector, it fails to bind its DNA target and gene transcription proceeds. In this way, AdeN finely tunes the constitutive expression of AdeIJK at a critical threshold⁷⁵. Constant, low-level expression of AdeIJK plays a pivotal role in the intrinsic resistance of A. baumannii to antibiotics^36,75. Indeed, AdeIJK has been shown to efflux clinically important antibiotics such as β-lactams, tetracyclines, and fluoroquinolones³².

AdeN and AdeIJK are notably conserved and consistently present in the core genome^17,22,36 suggesting that they are crucial elements for Acinetobacter survival. For example, recent studies indicate that AdeIJK can contribute to membrane lipid homeostasis to protect against host-generated polyunsaturated fatty acids^76,77. Additional insight into the physiological substrates of AdeIJK may help identify the natural effector ligands that promote efflux pump gene expression and better understand its function beyond antibiotic efflux. Knowing this may tell us how Acinetobacter species can bypass the need for adaptive evolution of adeIJK under antibiotic stress and exert intrinsic resistance to antibiotics.

The genomic arrangement of adeN is quite unique for a primary regulator of an RND efflux system in that it is encoded quite distant from the adeIJK operon, more typical of a global regulator and unlike the local regulators of the other RND efflux systems which are proximate to the pump-encoding genes^75,78,79. Interestingly, no local regulator has been identified for the adeIJK operon³⁶. The reason behind this distinctive arrangement remains an intriguing question. Similar to other TetR family regulators, it is possible that AdeN is a global regulator with a wide regulome^75,80,81.

In contrast to the adeSR regulatory system, the mutational landscape of adeN remains less explored. Studies indicate that amino acid substitutions and deletions in AdeN are primarily distributed within the dimerization domain of the regulator and occasionally within its predicted DNA binding domain^17,58,82,83. These mutations lead to varying levels of adeIJK upregulation^17,58,82,83. Insertion sequence (IS) elements are also frequently detected in adeIJK overexpressing mutants. For example, resistance to tigecycline has been associated with IS elements ISAba1, ISAba27, and ISAba125; and ISAba11 has been linked with ciprofloxacin resistance^83,84. The usual consequence of these insertions is disruption of adeN or its target sites, leading to increased expression of adeIJK⁸⁵. Interestingly, downregulation of adeN is observed upon ISAba13 insertion within the adeN promoter in response to erythromycin and host-derived unsaturated fatty acids⁷⁷. Together, this suggests that various stressors influence the IS type and the genomic location of the insertions; how this happens has not yet been discovered.

Overexpression of AdeIJK has also been observed independent of mutations in AdeN^17,82 suggesting that the adeIJK operon may be regulated by other factors—one of which may be the histone-like nucleoid structuring (H-NS) protein^69,86. In A. baumannii, inactivation of H-NS led to significantly elevated expression of adeIJK (as well as adeAB(C))⁶⁹, and hns complementation restored efflux-associated resistance phenotypes. Other studies demonstrated that stress response regulator DksA⁶⁷, alterations in (p)ppGpp levels⁸⁷, and oxidative stress-mediated soxR expression significantly altered AdeIJK expression⁸⁸ hinting that AdeIJK may partake in various cellular stress responses and is conditionally tuned by the involvement of multiple regulators. Indeed, high-level overexpression of AdeIJK is not often seen in clinical isolates of A. baumannii^58,82 as exceeding normal levels of tripartite, membrane-spanning structures is energetically costly, may impact membrane fluidity, or may even force the cell to expel essential metabolites⁸⁹. Why this is different from AdeAB(C)-overexpressing MDR isolates and the reason for the heterogeneity in AdeIJK expression are areas that require further investigation. Moreover, studies on these alternative regulators may clarify the downregulatory mechanisms of AdeIJK.

A LysR-type regulator, AdeL, serves as the local transcriptional controller of AdeFGH³⁶ (Fig. 1). This family of regulators is the most abundant in bacteria and governs crucial pathways including amino acid biosynthesis, oxidative stress response, ion transport, and antibiotic resistance⁹⁰. The regulator—which can act as an activator or repressor in response to an effector—usually functions as a homotetramer, each subunit containing a well-conserved helix-turn-helix DNA-binding domain at the N-terminus connected to an effector-binding domain⁹¹.

AdeFGH is normally maintained at low levels in A. baumannii^39,92 and is seemingly a minor contributor to the inherent antibiotic resistance of this microbe³⁹. Even though AdeFGH and AdeL are not found ubiquitously in Acinetobacter species, most infection-causing strains express them^36,93,94. Indeed, clinical A. baumannii isolates resistant to tigecycline and fluoroquinolones have been shown to overexpress AdeFGH^34,82,83,95. Under inducing conditions in the laboratory, AdeFGH has been observed to efflux of a range of antimicrobial substrates such as chloramphenicol, clindamycin, trimethoprim, sulfonamides, and nalidixic acid^39,92.

AdeL is a repressor of adeFGH⁹². Studies have shown that mutations in the C-terminus of AdeL created by selective pressure in vitro increased adeFGH expression up to 600-fold^81,92. Similarly, other works have suggested that mutations in the predicted ligand and DNA binding domains may activate constitutive adeFGH expression^41,58. The exact functional mechanism of these mutations is unknown as studies have yet to map the binding site of AdeL or determine its putative effectors.

AdeFGH overexpression is rarely observed during in vitro studies using basal media, even under the pressure of known substrates, leading to the idea that physiological substrates may act as AdeL effectors. For example, subinhibitory concentrations of the known substrate tigecycline have been shown to downregulate AdeFGH⁹⁶ while pleural fluid and human serum albumin upregulated expression of the pump operon^97,98. But note that host physiological substrates could be involved in an as yet unknown regulatory mechanism of AdeFGH. Unlike the previously discussed RND efflux pumps, AdeFGH has been shown to be influenced by only one global regulator, SoxR, which downregulates adeFGH expression⁸⁸.

A substantial body of evidence suggests that mutations in regulatory genes leading to low-level upregulation of RND pumps may not necessarily correlate with clinically significant levels of resistance^41,99. Nevertheless, these mutant subpopulations can be persistent or tolerant to antibiotics, potentially progressing to clinically relevant resistance through as yet unknown resistance mechanisms. Using immunosuppressed mouse models, Huo et al.⁴¹ clearly showed how AdeL mutations within A. baumannii persisters progressively enhanced resistance, transitioning from non-clinical levels to high-level AMR while improving bacterial fitness⁴¹. A related study in Pseudomonas aeruginosa showed how mutations in MexZ (the local regulator of the MexXY-oprM RND efflux pump) resulted in low-level upregulation of the pump, yet facilitated bacterial hiding within lung epithelial cells to reduce antibiotic exposure thereby improving tolerance to antibiotics⁹⁹.

Regulator mutations affecting efflux gene expression and the subsequent influence on infection-facilitating phenotypes (e.g., biofilm formation) have also been observed in A. baumannii^81,89. In one study, bacterial variants with mutations in the DNA binding domain of AdeL were prevalent in biofilm-forming cells following ciprofloxacin exposure while similarly treated planktonic cells had mutations predominantly in AdeN⁸⁴. The “selection” of AdeL variants in biofilm-forming cells correlates well with evidence that overexpression of AdeFGH enhances the transport of acylated homoserine lactones, key cell-cell communication molecules that increase biofilm formation⁴⁰. Another study observed that a mutation in AdeR resulted in overexpression of adeAB(C) leading to successful infection in a mouse lung model⁸⁹. Further, insertional inactivation of AdeN caused overexpression of adeIJK and a concomitant increase of the lethal A. baumannii invasion of mammalian cells¹⁰⁰. The same study showed that an adeN knockout strain caused elevated mortality in the Galleria mellonella (waxworm) infection model¹⁰⁰. Together, these examples point to the realistic opportunity to use efflux pump regulators to control A. baumannii virulence.

It is essential to recognize that modulation of efflux pump expression alone may not fully account for additional phenotypes. Most transcription factors are rarely limited to a single target¹⁰¹. Thus, it is crucial to determine whether the observed effects are attributable to the RND pump itself or to additional targets of the regulator¹⁰¹. The increasing evidence of RND regulators shaping efflux pump expression heterogeneity and thereby aiding bacterial evasion of antimicrobial treatment in dynamic physiological environments highlights the need to re-evaluate their impact to manage AMR and infection progression more effectively.

A variety of tactics can be envisioned to manipulate the regulatory systems of RND efflux pumps. Disruption of regulator function through small molecules or oligonucleotides that inhibit protein-protein, protein-DNA, and protein-effector interactions could be viable options (reviewed elsewhere³⁵), each avenue ultimately resulting in the down- or up-regulation of an efflux pump. The best strategy to employ would be dependent on the targeted efflux pump: downregulation of the pump operon would be the most effective option for activator-controlled AdeAB(C), whereas upregulation may be the better option for repressor-controlled AdeIJK and AdeFGH.

Targeted downregulation of an efflux pump could be achieved using the effector itself (or potentially mimics thereof) to interfere with the normal function of a regulator¹⁰². The same approach can be imagined using the known repressor proteins¹⁰³. Alternatively, inhibitors against the regulatory elements (protein- or DNA-based) could be used to block efflux pump expression. A promising example is the identification of an inhibitor of MarA, an activator of the RND AcrAB-TolC system in Gram-negative Enterobacteriaceae¹⁰⁴.

Purposeful upregulation of an RND efflux pump may seem counterintuitive to combat AMR, but the burden of overexpression on cellular fitness has been shown to impair bacterial growth⁸⁹. The physiological stress forces the cells to upregulate compensatory pathways involved in energy production, maintenance of cytoplasmic pH, and metabolite uptake in an aim to maintain homeostasis^105,106,107. Therefore, two-pronged AMR-fighting options may be available. A combined approach of efflux pump overexpression via regulator manipulation and prevention of a stress-induced, “life-saving” metabolic pathway shift is one hopeful strategy^105,106,107. Additionally, overexpression of efflux pumps can lead to “collateral sensitivity”, where improved resistance to one antibiotic increases the susceptibility of the bacteria to another, making a multidrug approach possible⁵⁶.

In addition to directly controlling efflux pump expression via their regulators, specifically manipulating the function of the regulators themselves is another plausible approach to control MDR A. baumannii. Efflux regulatory elements contain well-defined regions that can be strategically exploited through design of small molecule ligands¹⁰⁸. For example, AdeR has a conserved phosphorylation site and a magnesium binding motif that could be harnessed^53,109. Similarly, LysR and TetR family regulators have potential ligand-binding pockets⁵³. The unique, fairly conserved primary regulatory elements for each RND efflux pump in A. baumannii³⁶ could enable the development of targeted therapies that minimize collateral damage to beneficial microbiota and reduce the risk of broad-spectrum resistance. Synergistic treatments are also possible, combining inhibitors of regulatory proteins with existing antibiotics to reduce the required antibiotic dose and extend the longevity of currently used antimicrobial agents. Such targeting of transcriptional control mechanisms may open a new avenue in antimicrobial therapy, offering an alternative to the traditional approach of directly targeting the bacterial cell wall or replication machinery that are historically prone to evolve resistance¹¹⁰.

The most significant challenge in targeting regulatory elements of efflux pumps as potential therapeutic interventions is our limited understanding of these intricate regulatory mechanisms. Achieving optimal activation or inhibition of each regulator depends on knowing their binding affinities to their natural substrates. This, though, may be dependent on their confirmation which may change under different physiological conditions in vivo during an infection. Employing a machine-learning modeling approach may elucidate the regulome and physiological dynamics of the regulators and expedite therapeutic advances^108,111,112. Another challenge we face is that regulatory networks controlling RND efflux pumps can be redundant or complex⁴³. Interfering with one regulatory pathway may not sufficiently impair the bacteria as alternative pathways may be evoked. Manipulating transcriptional regulators may also lead to the inadvertent dysregulation of other essential bacterial functions and could lead to unintended toxicity to the patient. Understanding the full cellular role of each regulator is critical. A third challenge arises when considering combination therapies or exploiting the phenomenon of collateral sensitivity⁵². Ensuring that two molecules—a regulatory effector and an antibiotic, or two antibiotics, as examples—reach the infection site in the correct proportions is crucial¹¹³. This approach may also be highly dependent on the specific bacterial strain causing the infection and so would need to be specifically tailored. Finally, we also need to be aware that bacteria may develop resistance to inhibitors of regulatory proteins, just as they do to antibiotics, via mutations in the regulatory proteins themselves or through compensatory mechanisms.

Identifying highly conserved RND efflux pump transcriptional regulators and elucidating in detail their dynamics, structures, and effectors is crucial for developing broad-spectrum inhibitors against them. Exploration of the roles of these regulators independent of efflux pump expression is significantly lacking. And, as pathogens encounter many stressors during infection (oxidative, nitrosative, osmotic, nutritional, etc.) and elicit many adaptive metabolic strategies, computational and in vivo models are vital to understand the comprehensive roles played by these transcriptional regulators. There is much to learn, but also much to gain.