
Antibiotic resistance (AR) is a silent pandemic that kills millions worldwide. A comprehensive computational modeling based on globally collected data revealed that in 2019, almost 5 million deaths were associated with antibiotic resistance, including 1.27 million deaths directly caused by antibiotic resistant bacterial infections (1). Among these deaths, over 50% were attributable to drug-resistant Gram-negative bacteria, such as
Antibiotic resistance in Gram-negative bacteria is a combination of acquired resistance developed upon exposure to antibiotics, and intrinsic resistance resulting from the bacterial physiological characteristics, such as impermeable cell envelop, efficient drug efflux pumps, and the formation of biofilms that prevent antibiotics from reaching the bacteria. Acquired resistance, including the alteration of drug targets, inactivation, or degradation of antibiotics, often renders the bacteria with high resistance. Conversely, intrinsic resistance facilitates the initial development of resistance in Gram-negative bacteria via mainly maintaining drug concentrations at a low level, buying the time to acquire a high degree of resistance (3-5). Among the resistance mechanisms, efflux pumps play pivotal roles for both intrinsic and acquired resistance in Gram-negative bacteria (6, 7).
Efflux pumps are membrane-bound proteins that serve as a primary defense mechanism via actively exporting toxic substances, such as antibiotics, detergents, and heavy metals, whenever bacteria encounter them. Combining reduced drug penetration via the outer membrane barrier, efflux pumps make Gram-negative bacteria inherently resistant to many antibiotics. There are five main families of efflux pumps in Gram-negative bacteria: the major facilitator superfamily (MFS), the small multidrug resistance (SMR) family, the multidrug and toxin extrusion (MATE) family, the ATP-binding cassette (ABC) superfamily, the resistance-nodulation-division (RND) superfamily, and the proteobacterial antimicrobial compound efflux (PACE) family (8). The tripartite RND pumps are uniquely found in Gram negative bacteria. Although it has been reported that
The structure of the whole complex of AcrAB−TolC has been resolved using various techniques that include X-ray crystallography and cryo-electron microscopy, providing insights into its structure and operational mechanism (28-31). Like other members of the tripartite RND family, AcrAB−TolC is a tripartite complex consisting of two trimeric proteins, AcrB (IMP) and TolC (OMF), bridged by hexameric AcrA (MFP) (Fig. 1A).
AcrB is a transmembrane protein with the size of 113.6 kDa (1,049 aa). In the AcrAB−TolC complex, AcrB exists as a homotrimer, and the three protomers are arranged in a triangular shape, forming a funnel-like structure that can cope with a wide range of molecules for their translocation out of the bacterial cell. Each protomer is composed of three domains: the transmembrane domain (TMD), the porter domain (PD), and the funnel domain (FD) (Fig. 1). The TMD of AcrB protomer is comprised of 12 transmembrane helices (TM1−TM12), which contain critical residues for proton translocation to conduct the energy transduction through electrochemical proton gradient across the membrane (32). A total of 36 TMs of TMDs in trimeric AcrB form the central cavity in the inner membrane, and provide a path for drug entrance (33). It has been shown that the conformational changes of the TMDs in each AcrB protomer are interconnected, affecting the function of the whole complex (34). The PD in AcrB consists of two N-terminal and two C-terminal subdomains (PN1, PN2, PC1, and PC2), and it is the major domain of AcrB for substrate binding, which during the drug efflux process, undergoes large conformational changes. Each PD of the protomer contains two substrate-binding pockets, the proximal binding pocket (PBP) between PC1 and PC2, and the distal binding pocket (DBP) between PC1 and PN2 (PN1), which are separated by 11 amino acids termed “a switch loop” (35, 36). The PD is a key structural component of AcrB that is essential for the recognition, binding, and translocation of various molecules (37). The FD, the third domain of AcrB, is a region located at the tip of AcrB toward the periplasmic end of the protein. A long loop made with two N-terminal β-sheets (Nβ 8 and 9) in the FD of each protomer stretches and penetrates the FD of another protomer, interlocking with each other and maintaining the architecture of the AcrB trimer, while the TMDs and PDs of protomers undergo conformational changes (Fig. 1) (38). A sequential peptide affinity tag experiment revealed that AcrZ, a 49 amino acid long (5.3 kDa) small protein, interacts with AcrB in one to one ratio (39). As a single helix, AcrZ interacts with MTD of AcrB in the inner membrane (28, 31, 40). A recent cryo-EM study observed that bending of AcrZ around position 16 (P16) is important for the physical interaction of AcrZ with AcrB (41). The same study also revealed that AcrZ binding to AcrB modulates the structural conformation and activity of AcrB through allosteric changes that is enhanced in a lipid environment where cardiolipin and palmityloleoyl phosphatidylglycerol are present (41).
AcrA is the 42.2 kDa (397 aa) periplasmic protein comprising MFP in the AcrAB−TolC complex, which bridges the inner membrane protein AcrB and the outer membrane protein TolC (Fig. 1). AcrA exists in two conformationally different types of protomers (protomer I and II) in a hexameric assembly of the AcrAB−TolC complex. Each AcrA protomer is subdivided into four subdomains: the membrane-proximal domain (MPD), β-barrel domain, lipoyl domain, and α-helical domain (13, 42). A recent cellular electron cryo-tomography study revealed that there are two distinct conformations of MPDs in hexameric AcrA, in which one interacts with both AcrB and the inner membrane (protomer I), while the other interacts with only AcrB (protomer II) (42). Along with the MPD, the β-barrel domains flexibly linked to the MPDs are also involved in interaction with AcrB at its FDs. The lipoyl domains and the α-helical hairpins are arranged toward TolC in a cylindrical shape, and play roles in the stabilization of the AcrA hexameric assembly, interacting with each other, and the intermeshing cogwheel-like interaction with TolC providing the TolC interface, respectively (40).
TolC is the outer membrane protein of the complex, and forms a channel across the outer membrane of the bacterial cell. TolC-like proteins are ubiquitous in Gram-negative bacteria, which are involved in various extrusion processes of diverse molecules that are either produced by bacteria (metabolites, virulence factors, and toxins), or imported into them (antibiotics, biocides, bile salts and organic solvents) (43, 44). Unlike other Gram-negative bacteria that have several TolC-like proteins, entrobacteriaceae, such as
During the efflux process through the AcrAB−TolC drug efflux pump, substrates move from the AcrB binding pocket through the TolC channel, and into the extracellular space. The movements of substrates through this complex have been extensively studied using a variety of techniques, which include X-ray crystallography, cryo-electron microscopy, and molecular dynamics simulations (28-31, 46). The efflux process is initiated when a substrate of AcrAB−TolC binds to AcrB, and triggers a series of conformational changes in the complex (11, 47, 48). During the drug translocation process, conformational changes of AcrB occur in a cyclic manner, in which each protomer goes through the rotation of three states termed access, binding, and extrusion, or alternatively, loose (L), tight (T), and open (O), which are adopted from F0F1−ATP synthase (Fig. 1B) (48-50). The L state, the apo form of AcrB, is presumably the initial state of the conformation cycle. In the L state of a protomer, the PBP between PC1 and PC2 subdomains is open, while the DBP between PC1 and PN2 subdomains is closed in the PD (31, 51, 52). Upon binding of substrates, the L state shifts to the T state where the PBP becomes smaller, while the DBP is open, allowing the substrates to be translocated toward the FD. At the T state stabilized by substrate binding in the DBP, the TMD undergoes the conformational change that allows protons to access the acidic residues (D407 and D408) in the TMD, which provide the proton translocation driven T to O transition of AcrB (53). The exit gate between PN1 and PN2 subdomains is open at the O state, while both the PBP and DBP are closed. As a result, substrates move out through the FD toward the funnel of AcrA−TolC complex (31, 53). Finally, the release of protons from the TMD in the O state to the cytoplasm brings the AcrB to the initial L state, closing the exit gate, and opening the PBP (54). Of note, a recent study proposed an alternative explanation for the cyclic conformational rotation of AcrB, with the hypothesis that the conformational changes in AcrB trimer are initially triggered by proton binding, instead of substrate binding. However, this hypothesis has not yet been verified by others (55).
Prior to substrate binding, the AcrAB−TolC complex is presumably at the resting state, where the AcrB trimer exists as a symmetrical structure, with all three protomers in the L state. In the resting state pump, the symmetrical (LLL) AcrB trimer is tied to a loosely packed AcrA hexamer that binds to the closed form of TolC (38). Upon substrate binding, the cyclic conformational change of AcrB occurs, converting its structure from symmetrical to asymmetrical form. Various asymmetrical AcrB trimers comprised of combinations of the different states (L− T−O, T−T−L, L−T−L) have been found through X-ray and cryo-EM analyses (31, 49). Since full AcrAB−TolC complexes with asymmetric AcrB trimers are supposed to be in the active process for substrate extrusion, the conformations of AcrA and TolC are adopted to complete the process, providing an open path for substrates to go through, out of the outer membrane. At the active stage of an AcrA hexamer, the β-barrels and MP domains associated with the AcrB trimer are reoriented, and the α-helical hairpins are tightly packed, sealing the enclosed cavity that is connected to the TolC (30, 46). The conformational changes of AcrA lead to relocation of the helices at the lower end of the TolC α-barrel in an iris-like opening, eventually forming an entire open path from AcrB to TolC toward the outer membrane (31). Several reviews have provided detailed explanations of the elaborate conformational changes in AcrAB−TolC complex during the substrate extrusion (11, 13).
Four different substrate access channels (CH1−4) have been suggested based on computational structure analyses of AcrB, of which entry sites for substrates toward the PBP and the DBP in AcrB are found to be either from the periplasm, or the outer leaflet of the inner membrane. Channel 1 (CH1) is located in the interface between PC2/TMD (TM8) that allows substrates to enter AcrB through the outer leaflet of the inner membrane (56). It has been observed that CH1 is open in the L and T states, but closed in the O state by the conformational shift of TM8 and PC2 (33, 56). The periplasmic Channel 2 (CH2) was observed in the cleft between PC1 and PC2, which leads the substrate to the PBP of the L and T states (33, 48, 56). The presence of the third channel (CH3) was observed at the central cavity, a large cavity at the bottom of the periplasmic domain and at the top of the TMS domain, formed by the residues of all three AcrB protomers in the inner membrane (57-59). It has been suggested that the CH3 can guide substrates directly to the DBP by passing the PBP, as well as the switch loop (33). Recent molecular modeling studies, followed by functional and structural analyses, have suggested a fourth channel (CH4) that begins at the TM1/TM2 groove, and extends to an interface pathway between PN2 and PC1 toward the DBP (60, 61). These different substrate access channels in distinct locations at the AcrB trimer contribute to substrate poly-specificity of AcrB, providing various routes for the extrusion of structurally unrelated diverse molecules out of the bacteria.
A wide range of substrates that include antibiotics, detergents, dyes, and other toxic compounds can be recognized and transported through AcrAB−TolC (Supplementary Table 1). The poly-specificity of AcrAB−TolC allows structurally diverse molecules to be transported by the pump, and numerous functional, structural, and mutagenesis studies have shown that the substrate specificity is initially determined by the access channels at the PD, which have certain substrate preferences (33, 49, 57, 62). The CH1 in the TMD/PC2 interface at the TM 7/8/9 groove near the vestibule guides substrates from the outer leaflet of the inner membrane to the PBP and DBP transition point (57). Phenicols, linezolid, and fusidic acid, as well as dodecyl β-D-maltoside (DDM) that has been co-crystalized with AcrB, are found to be preferred substrates for CH1 (36, 56, 60, 62). A mutagenesis study revealed that CH1 can be also a transport route for amphiphilic β-lactams with larger partition coefficient, such as nafcillin, cloxacillin, and benzyl-penicillin, of which the hydrophobic side chains are more likely to immerse in the outer leaflet of the inner membrane, while the β-lactam ring would still remain in the periplasm (63, 64). In general, it has been proposed that compounds with low molecular weight and low polar surface area are likely to enter the AcrB through CH1 (62). The CH2 pathway starting at the lateral cleft between the PC1 and PC2 interface is located about 15 Å above the putative membrane plane (36, 48, 49, 57). The location of CH2 seems to be suitable for capturing substrates in the periplasm. Although the substrate specificity of CH2 is still elusive, studies show that CH2 prefers high molecular weight drugs, such as erythromycin, rifampicin, minocycline, and doxorubin-dimer, in the classes of microlides, ansamycins, tetracyclines, and anthracyclines (36, 49, 57, 65). A site-directed mutagenesis revealed that planar aromatic cations with low molecular weight, such as ethidium bromide (EtBr), benzalkonium chloride (BZK), berberine (BER), and rhodamine 6G (R6G), are the preferred substrates for CH3 in the central cavity of TMD in the inner membrane (33). The fourth channel (CH4) comprised of the TM1/TM2 groove has been proposed based on several molecular modeling studies, as well as functional and structural analyses, and it has been shown that carboxylated molecules, such as fusidic acid, oxacillin, cloxacillin, dicloxacillin, and piperacillin, can be transported through CH4 (60, 62).
The substrate preference or specificity of each access channel appears to go toward certain physicochemical and structural properties of the substrates. For example, when some of the known substrates for AcrAB−TolC are projected in a principal component analysis (PCA) plot with respect to their molecular weight, hydrophobicity, and charge distribution, as well as their size and shape indicated by minimal and maximal projection areas, the drugs preferred by each channel are roughly clustered together, implying that the physicochemical and structural properties of the substrates play an important role in determining whether they can be recognized and transported through a certain channel in AcrB (Fig. 2) (11, 13). Hydrophilic aminoglycosides known to be unfavorable substrates for AcrAB−TolC are clustered away from the well-known AcrAB−TolC substrates in the plot (66). The PCA plot is made with 6 chemical features (molecular weight, hydrophilic-lipophilic Balance, logP, topological polar surface area, and minimum projection area and maximum projection area obtained from the UCSF ChimeraX); however, it is apparent that these 6 features are not sufficient to perfectly rationalize the substrate preference of each channel. For example, tetracycline, minocycline, and doxycycline, the drugs in the tetracycline class, have been reported as the substrates for CH2 (36). However, they are instead positioned close to the substrates that more likely use CH1 (Fig. 2). Piperacillin is also found to be apart from the other carboxylated β-lactams (oxacillin, cloxacillin, and dicloxacillin), although it has been shown that these carboxylated β-lactams are translocated through CH4 (60). Of note, piperacillin seems to indeed be somewhat different than other carboxylated β-lactams, behaving differently in the site-directed mutant (L300A) that is supposed to have altered CH4 function (60). To extend our comprehension of each channel’s substrate specificity, it would be necessary to integrate more chemical features of the substrates, including the presence of certain functional groups, such as carboxyl, amino, and hydroxyl groups that likely affect the substrate preferences between the access channels (67).
Once the substrates enter the AcrAB−TolC complex through different access channels, they either move through PBP to DBP (CH1 and CH2), or directly end up at DBP (CH3 and CH4) (62). The hydrophobic environment in DBP, comprising mainly phenylalanine, but also isoleucine, valine, alanine, and asparagine side chains, seems to be opted for taking in all the known AcrAB−TolC substrates as an anteroom to enter the exit funnel connected to AcrA−TolC (11, 68). Overall, the substrate poly-specificity of AcrAB−TolC is the outcome of a complex interplay between the physicochemical and structural properties of the substrate molecules, and the molecular features of the AcrAB−TolC complex.
The substrate polyspecificity of AcrAB−TolC with different drug access channels would be a great benefit, granting them the capability to simultaneously cope with various structurally unrelated antibiotics in different classes. Of note, the strong cooperative efflux between different drugs, such as chloramphenicol and cefamandole, and the simultaneous extrusion of the CH3-preferred drugs and the CH1- or CH2-preferred drugs have been observed (62, 69). In the current clinical setting, combination therapies with more than one antibiotic have become a routine procedure, due to increased antibiotic resistance (70). In-depth understanding of the substrate polyspecificity mechanism for AcrAB−TolC, a major efflux pump in Gram-negative bacteria, might enable the development of a better strategy for a more effective combination therapy.
AcrAB−TolC-associated multidrug resistance has been a significant clinical implication for the treatment failure of antibiotics against Gram-negative bacterial infections (71-74). In addition, the involvement of AcrAB−TolC in other cellular functions, such as biofilm formation and pathogenicity, in Gram-negative bacteria makes the pump an attractive target for the development of new strategies to combat antibiotic resistant infections (21-25). As such, inhibition of AcrAB−TolC can be an effective way to enhance the effectiveness of antibiotics and reduce the occurrence of antibiotic resistance. Various AcrAB−TolC inhibitors, including small molecules and peptides, have been identified (13, 26, 75).
Phenylalanine-Arginine-β-Naphtylamide (PAβN) is a broad-spectrum efflux pump inhibitor that has the inhibitory activity for various efflux pumps in Gram-negative bacteria, including MexAB−OprM, MexCD−OprJ, MexEF−OprN, AdeABC, and AcrAB−TolC. PAβN was initially identified through a 200,000 small molecule combination screening with levofloxacin against MexAB−OprM, MexCD−OprJ, or MexEF−OprN overexpressing
There are also inhibitors that specifically target AcrAB−TolC: 1-(1-naphtylmethyl)-piperazine (NMP), MBX2319, WK2, and BDM88855. An arylpiperazine, NMP, was identified in a high throughput screening against an AcrAB overexpressing
More recently, new strategies have been explored to identify AcrAB−TolC inhibitors, such as interfering with the pump assembly. Recent studies demonstrated that designed peptides that can mimic TMs of AcrB interrupt AcrB trimerization, leading to the inhibition of AcrAB−TolC mediated drug efflux (82). In another way to disrupt the efflux pump assembly, AcrA binding molecules, such as NSC series compounds, have been developed, and have shown potentiation activity for certain antibiotics, including novobiocin and erythromycin (83, 84). Recent articles have comprehensively reviewed some of the AcrAB−TolC inhibitors, describing their structure-activity relationships, chemical synthesis, mode of action, and
The AcrAB−TolC efflux pump is one of the major efflux pumps found in Gram-negative bacteria, and most likely is the main efflux pump contributing to multidrug resistance, especially in Enterobacteriae (11, 13). The pump is capable of extruding a wide range of structurally diverse compounds from the bacterial cell, including many antibiotics, thus reducing their intracellular concentrations, and rendering them ineffective. Studies over several decades have greatly extended our understanding of AcrAB−TolC in terms of its structure, substrates, and operational mechanism. However, there are many more gaps to fill to fully understand the nature of its sophisticated operation mechanism, as we glimpsed in this review, such as its substrate recognition process by the different access channels. The extrusion of molecules through AcrAB−TolC seems to be elaborately controlled by certain rules orchestrating the molecular features of the AcrAB−TolC protein complex and the diverse physicochemical properties of the substrates. In-depth comprehension of the mechanisms of AcrAB−TolC efflux pump-mediated antibiotic resistance will lead to the development of new strategies to combat antibiotic-resistant infections, such as the development of inhibitors that can effectively block the function of AcrAB−TolC, as well as new antibiotics that are less susceptible to AcrAB−TolC-mediated resistance.
This work was supported by the National Research foundation of Korea (NRF) grant funded by the Korea government (MSIT) (NRF-2020K1A4A7A02095129/2023M3A9G6057281 to S.J.).
The author has no conflicting interests.
![]() |
![]() |