The pioneering work of Paul Ehrlich and Alexander Flemming kick-started the antibiotic era. Ehrlich postulated in 1906 the concept of a magic bullet that selectively targets the pathogenic microbe without affecting the host, while Flemming discovered penicillin, a compound effective against bacteria, from a mold with antibacterial properties in 1928 (1). Nowadays, antibiotics are viewed as one of the most successful forms of chemotherapy that revolutionized modern medicine. However, the first cases of resistance to penicillin were reported shortly after its introduction into the clinic in 1947. The use of antibiotics therefore comes with a price, namely: increased resistance (2). Currently, the effectiveness of many antibiotics has been affected because of the emergence of antibiotic resistant superbugs. This represents a global public health threat as evidenced by the high mortality rates of drug resistant microbes. For example, each year about 25000 patients in the EU die from infections with multidrug resistant bacteria, while this is the case for 63000 patients in the United States (1). Consequently, extra costs are imposed on the healthcare system because of infections with drug-resistant microbes. These are estimated to be €1.5 billion each year in the EU. The origins of multidrug-resistance are diverse and complex, although about 80% of all severe bacterial infections observed in the hospital are caused by multidrug-resistant Gram-negative pathogens (3). Pseudomonas aeruginosa (figure 1), a common inhabitant of the human respiratory system, represents an important example of a such a multidrug-resistant superbug because it possess an exceptionally high resistance to a wide variety of antibiotics and is responsible for about 10% of opportunistic infections in immunocompromised patients with cancer as well as patients with severe burns and cystic fibrosis (4). This bacterium is surrounded by a protective cell envelope that prevents the entry of many toxic compounds, explaining why P aeruginosa is inherently insensitive to many antibiotics. However, high-level multidrug-resistance is achieved through additional strategies of which the ability to rapidly export drugs represents a particular powerful mechanism (5,6). This is typically achieved by proteinaceous devices that actively pump out toxic molecules from the cell and are known as multidrug efflux pumps. These are made up of one or more protein components that either span the inner membrane or traverse the complete cell envelope (5,6). Interestingly, the genome of P aeruginosa contains many potential drug efflux systems that together are probably responsible for conferring high-level multidrug-resistance. Of these, the MexAB-OprM pump represents the best characterized one and comprises three subunits. These are localized to different compartments of the cell envelope with MexB and OprM embedded into the inner membrane (7) and outer membrane (8.9), respectively, and MexA in the periplasm (10). MexAB-OprM are assembled into a membrane-to-membrane complex that is directly responsible for the removal of antibiotics from the cell (10). MexB selects antibiotics that are to be removed and these are subsequently transferred to MexA in a step that is driven by the proton motive force. MexA bridges MexB and OprM in the periplasm, while OprM represents a channel-forming subunit that accepts antibiotics from MexA and facilitates the final step of their efflux - removal across the outer membrane into extracellular milieu. Considering the role of MexAB-OprM in conferring drug resistance, a detailed structural understanding of this efflux pump is, for example, required to design drugs that interfere with its function. It is therefore important to note that recent structures of wild-type MexAB-OprM obtained in the absence or presence of antibiotics provide profound insight into its molecular mechanism (11). Here I will discuss these structures as well as its functional understanding.
Overview of bacterial multidrug efflux transporters
It has been established that bacteria can be naturally resistant to certain antibiotics, while they are able to acquire high-level multidrug resistance through additional mechanisms. These include: (i) degradation or modification by enzymes, (ii) alteration of the drug target and (iii) preventing access to the drug target. The latter can be accomplished by reducing the permeability of the Gram-negative outer membrane through, for example, decreasing the expression of porins or increasing the expression of more-selective channels. Additionally, active removal from the cell represents another powerful mechanism to the prevent antibiotics from reaching their target. In fact, the ability of bacteria to rapidly export drugs has emerged as clinically one of the most important mechanisms to acquire high-level resistance. This is typically accomplished by so-called efflux pumps – protein complexes made up of one or more subunits that are located in the cytoplasmic membrane or traverse the entire cell envelope and drive the export of toxic molecules from the cell (figure 2 adapted from 5). These efflux systems are chromosomally encoded, thereby contributing to inherent and acquired antimicrobial resistance.
Different types of bacterial efflux pumps have been described that belong to one of the five major superfamilies (figure 2), namely: ATP-binding cassette (ABC) superfamily, the major facilitator superfamily (MFS), the small multidrug resistance superfamily, the resistance-nodulation cell division (RND) superfamily and the multidrug and toxic compound extrusion (MATE) superfamily. ABC-type efflux pumps utilize the energy of ATP hydrolysis to power export, while members of the other superfamilies employ the energy of transmembrane ion gradients (e.g. PMF) for this purpose. Unlike members of the RND superfamily that form a membrane-to-membrane pump made up of three subunits, members of the other superfamilies typically function as single entities in the cytoplasmic membrane to transport toxic molecules across the cytoplasmic membrane. RND-type efflux pumps are organized into a complex that traverses the entire Gram-negative cell envelope and actively pumps out toxic molecules from the cell into the extracellular milieu. These efflux pumps are crucial for high-level multidrug resistance and are therefore found in many pathogens. For example, the genome of P aeruginosa contains numerous potential efflux pumps, including 12 members of the RND superfamily and twenty MFS-type pumps (12), thereby explaining its resistance to a variety of antibiotics. A fully assembled RND transporter consists of a substrate-binding core subunit (RND pump) that is embedded into the cytoplasmic membrane and utilizes the energy of the PMF to translocate the substrate across the cytoplasmic membrane, a channel-forming outer membrane factor (OMF) that traverses the outer membrane and is responsible for expelling toxic molecules from the cell and a periplasmic membrane fusion protein (MFP) that is anchored to the cytoplasmic membrane and bridges the RND pump and OMF. The P. aeruginosa MexAB-OprM system represents one of the best characterized RND-type pumps and will be discussed in more detail below.
Biochemical and structural features of the multidrug transporter MexAB-OprM
The MexAB-OprM multidrug transporter is a tripartite efflux pump belonging to the RND superfamily and plays a key role in both intrinsic and acquired resistance. In vivo analysis with P. aeruginosa conditional strains, defective in this and other efflux pumps, showed that MexAB-OprM is able to expel a wide variety of antimicrobial agents, including quinolones, macrolides, tetracyclines, chloramphenicol and nearly all β-lactams (13). Thus, P. aeruginosa is resistant to almost every class of antibiotics. The subunits of the MexAB-OprM system are localized to the different compartments of the P. aeruginosa cell envelope. MexB is a protein of about 112 kDa that is embedded into the cytoplasmic membrane and contains 12 transmembrane domains. As RND pump, MexB determines the drug specificity of the efflux pump by selecting antimicrobials that are to be expelled from the cell as revealed by swapping MexB with the RND pump of another efflux pump that yielded a similar substrate scope as MexAB-OprM (14). The broad drug specificity of the MexAB-OprM system implies, therefore, that MexB is able to accommodate a variety of structurally diverse compounds. Subsequently, the selected antimicrobials are translocated across the cytoplasmic membrane in a PMF-dependent fashion. OprM is the channel-forming OMF and has a molecular mass of about 53 kDa. Interestingly, OprM is embedded into the outer membrane through its β-barrel domain similar to established outer membrane proteins, while it is also tethered to the outer membrane via lipid modifications like lipoproteins. MexA is a protein of about 41 kDa that is localized to the periplasm, although it is anchored to the cytoplasmic membrane by virtue of its lipid-modified N-terminus. It has been established that MexA is essential for in vivo activity of the efflux pump, which is in line with its function of bridging MexB and OprM.
Structural features of MexB
Crystal structures of the individual components of the MexAB-OprM system have been solved and collectively these provide an in-depth functional view of this efflux pump. Figure 3 shows the structure of MexB that was solved at 3.0 Å (7), revealing that its overall structure is a jellyfish-shaped trimer and comprises three large domains, namely: transmembrane domains (TMDs), pore domains and OMF docking domains. MexB contains 12 α-helical TMDs that anchor it into the cytoplasmic membrane and serve as a conduit for proton movement, which is coupled to the export of antimicrobials. The TMDs are organized into a channel with a central opening (indicated with an asterisk in the right panel) that is probably used for the translocation of drugs. Moreover, these are connected by several large periplasmic loops that are folded into distinct subdomains that make up the pore and docking domains, respectively. The pore domain is the main entry gate of drugs and contains the drug-binding pocket. The docking domain comprises two periplasmic subdomains (DN and DC) and interacts with MexA, thereby bridging the cytoplasmic and outer membranes.
Despite several high-resolution structures of MexB, no cocrystals with drugs or inhibitors have been obtained and it is therefore not clear how antimicrobials are bound and processed by MexB. However, structures of AcrB, an RND pump from the homologous E. coli AcrBA-TolC efflux transporter, with different drugs and inhibitors have been reported. For example, the upper left panel of figure 4 shows the structure of the AcrB trimer with rifampicin (in orange spheres) and minocycline (in red spheres) (15). Rifampicin is a high molecular weight antibiotic, while minocycline is a low molecular weight antibiotic. This structure therefore suggests that AcrB has two different drug binding sites, namely a binding cavity for rifampicin and other high molecular weight drugs (proximal binding pocket) and a binding cavity for minocycline and other low molecular weight antibiotics (distal binding pocket) that is composed of predominantly hydrophobic residues, including phenylalanine, isoleucine, valine and alanine. The hydrophobic character of the distal binding pocket probably enables it to accommodate a large variety of antimicrobials
Moreover, the different structures of AcrB indicate that drugs are taken up from the periplasm into its binding cavities and are subsequently transferred through its central channel to the top of the AcrB trimer into the TolC channel. However, other pathways have been proposed, employing channels at the membrane-periplasm interface. Conceivably, hydrophobic drugs that are able to insert spontaneously into the cytoplasmic membrane seem to be well accepted by the RND pump unlike drugs with a hydrophilic character. Hydrophobic drugs may therefore be located in the periplasmic leaflet of the cytoplasmic membrane and enter the interior of the pump through one of the proposed channels at the membrane-periplasm interface. Both binding pockets are made up of four subdomains (PN1, PN2, PC1 and PC2) and are lined primarily by aromatic residues that are thought to determine drug specificity. Thus, the presence of two multisite drug-binding pockets explains the remarkable drug specificity of the AcrBA-TolC system. The right panel of figure 4 shows a structural comparison of AcrB (in orange) bound to rifampicin (as blue spheres) and minocycline (as green spheres) and MexB (in grey), revealing a near-perfect of both structures with rifampicin and minocycline located near the same helices. It is therefore likely that MexB also possesses two multisite drug-binding pockets similar to AcrB. A close-up of the rifampicin-binding is provided in the bottom panel of figure 4 with rifampicin in blue ball and sticks, interacting residues of AcrB in orange and functionally similar residues of MexB in grey. Of these, Asn719 and Arg717 interact with rifampicin through hydrogen bonds, while Phe617 interacts hydrophobically with rifampicin. Importantly, the proximal and distal binding pockets are separated by a conserved cluster of aromatic residues, including Phe617 in AcrB or Phe615 in MexB.
Structural features of MexA
In a complete RND-type efflux pump, the RND transporter and OMF don’t directly interact but they are connected instead by a membrane fusion protein that bridges the periplasm. It has been established that MexA is anchored to the periplasmic face of the cytoplasmic membrane through its lipid modifications and connects MexB and OprM. The atomic structure of MexA was elucidated at 2.40Å (10) and this is shown in figure 5, revealing that the monomeric structure of MexA (upper left panel) is made up of three principal domains, namely: (i) the α-domain that is formed by two α-helices that are organized into a helical hairpin, (ii) the globular β-domain that is adjacent to the α-domain and is made up eight short β-sheets and (iii) another globular domain following the central β-domain that consists of seven β-strands and one short α-helix and is therefore known as the α/ β-domain. Additionally, MexA contains a short membrane proximal domain next to the α/β-domain that was not well resolved in the crystal structure. Overall, six MexA protomers are folded into a funnel-like hexamer that is shown in ribbon (upper right panel) and surface representation (bottom left panel), respectively. A top view of the hexamer is shown in the bottom left panel. The α-domain of the hexamer interacts with OprM, while the β and membrane proximal domains interact with MexB.
Structural features of OprM
As OMF, OprM plays a crucial role in the final stage of drug efflux from the cell – translocation of antimicrobials across the outer membrane into the extracellular environment. Its high-resolution structure was solved at 2.4Å (9). This is shown in figure 6 and reveals that OprM is a channel-shaped homotrimer comprising two domains (upper left panel), namely: the β-barrel domain and the α-barrel domain. The latter can be subdivided into the equatorial and coiled-coil domain. The main part of the channel is made up of the large helices from the α-barrel, equatorial and coiled-coil domains. The structure also establishes that OprM is anchored to the outer membrane via two different mechanisms. Firstly, its β-barrel domain is embedded into the outer membrane, and secondly, OprM is also tethered to it by virtue of its lipid modified N-terminus. TolC is the channel-forming subunit of the E. coli AcrBA-TolC efflux pump and is closely related to OprM.
Analysis of the regions of TolC important for efflux activity indicated that the equatorial domain is crucial for this probably owing to its association with AcrA (16). This therefore implies that the same domain of OprM interacts with MexA. Moreover, the bottom panel shows that the OprM channel is closed at the top (extracellular surface) as well as at the bottom (periplasmic entrance). Closure of the exit at the top is caused by short loops of the β-barrel domain that protrude into the central pore, while the twisting of the periplasmic helices cause closing at this end of the channel pore. An additional hydrophobic constriction is present at this side of the channel that is formed by a triplet of leucine residues and is probably sufficient to block the passage of small molecules, including water. The finding that OprM is closed at both ends implies that the channel must open during drug efflux. However, the precise mechanism is not clear, although an iris-like movement has been proposed.
Structural features of the fully assembled MexAB-OprM efflux pump
Crystal structures of all subunits of the MexAB-OprM system have been determined individually, although no structural information of the fully assembled complex is available. Recently, however, the structure of the complete wild-type MexAB-OprM efflux pump was assessed by cryo-EM (11) at 3.6Å. This is presented in figure 7 in ribbon and surface representation, showing that the fully assembled complex adopted an overall structure resembling a vertically elongated rod with a 1:2:1 stoichiometry of OprM, MexA and MexB. The structure confirms the previously determined X-ray structures of the individual subunits and reveals that, as expected, OprM does not directly interact with MexB. Instead, OprM is connected to MexB through MexA, which forms a hexameric ring and associates with OprM via its membrane proximal domain. OprM is folded into a trimeric cylindrical channel that is associated with MexA only via its α-hairpin domain in a tip to tip interaction. The structure, moreover, shows that each protomer of MexB is in a different conformational state representing the access, binding and extrusion state, respectively, as has previously been reported for AcrB (ref). In the absence of drugs, the MexB protomers are closed towards the outside and these are therefore unable to bind drugs. Hence, this state probably resembles the resting state of the complex. In the presence of drugs, however, some of the MexB protomers are opened towards the outside, thereby enabling the binding of drugs in the distal binding pocket. MexB and MexA display two sites of interaction, namely: the β-barrel domain of MexA with the docking domain of MexB and the membrane proximal domain of MexA with the pore domain of MexB. In the crystal structure of OprM the periplasmic gate is closed (figure 6), while in the cryo-EM structure it is opened (figure 7 bottom right panel indicated by an asterisks). In both structures the top of the OprM channel at the cell surface is closed. Thus, assembly of the entire complex induces opening of the periplasmic gate. More specifically, binding of MexA is probably sufficient to open the gate because this probably results in reorientation of the OprM helices that block the periplasmic gate.
Mechanism of drug efflux
Insight into the substrate processing by RND pumps was obtained by solving the crystal structure of AcrB with and without antimicrobials (17). This showed that each of the AcrB protomers is in a different conformational state that correspond to one of the functional states of the drug export cycle. Based on this it was proposed that antimicrobials are exported by a three-step rotating mechanism. Specifically, drugs are bound to only one protomer, which is referred to as the binding protomer. While another protomer, the extrusion protomer, is open towards the top (funnel) of AcrB. The docking domain contains a large opening that connects to TolC, the channel-forming subunit. The last protomer, the access protomer, is characterized by a vacant drug binding pocket but is not open towards the funnel. The binding site in the binding and extrusion protomers are conformationally different. In the binding protomer, it is expanded, whereas its volume in the extrusion protomer is decreased in such a way that antimicrobials can’t bind. The binding pocket of the access protomer resembles that of the extrusion protomer. The exit from the binding site towards the top or funnel (docking domain) of AcrB is controlled by different central helices. The pore domain contains various channels that extend from the interior of AcrB to the periplasm and these are probably used by drugs to enter the binding pocket. In the binding protomer, drugs reach the binding pocket through such an uptake channel. However, the exit pathway towards the funnel is blocked by one of the central helices. This uptake channel is not observed in the extrusion protomer and the exit pathway towards the funnel is open. The access protomer contains the pathway from the periplasm is open but the binding cavity is small. Conceivably, drugs are pushed into the top of the funnel by a decrease of the binding pocket as the protomers cycle through the access, binding and extrusion state, thereby simultaneously opening the exit from the pocket towards the funnel in a PMF-dependent fashion. The protomers of MexB were also observed in three different conformational states suggesting that MexB processes substrates by a similar mechanism as AcrB as outlined in figure 8 (adapted from 11). In the first step, the gate loop of the pore domain that blocks the access towards the drug binding pocket is pushed downward as the drug concentration within the cell increases, thereby opening the drug binding pocket. Subsequently, MexB ejects drugs into the MexA-OprM channel by a functional rotation mechanism as described above. As the drug concentration decreases, the entry towards the drug binding pocket of the access protomer is closed and the complex returns from the binding state to the resting state. The fully closed binding pocket in the resting state prevents backflow of drugs.
Gram-negative bacteria employ different strategies to achieve high-level multidrug resistance with the ability to rapidly export drugs from the cell as a particular powerful mechanism for this purpose. P aeruginosa represents an important example of a multidrug-resistant superbug owing to its exceptionally high resistance to a wide variety of antibiotics. This is to large extend caused by the presence of numerous potential efflux pumps in the proteome of P aeruginosa, including 12 members of the RND superfamily and twenty MFS-type pumps. Of these, the MexAB-OprM system represents the best characterized multidrug efflux pump, which is made up of three subunits. These are localized to different compartments of the cell envelope with MexB and OprM embedded into the inner membrane and outer membrane, respectively, and MexA in the periplasm. The complete MexAB-OprM pump forms a membrane-to-membrane complex with MexB responsible for selecting antimicrobials that are to be removed and these are subsequently transferred to MexA in a step that is driven by the proton motive force. MexA bridges MexB and OprM in the periplasm, while OprM represents a channel-forming subunit that accepts drugs from MexA and facilitates their removal out of the cell into the extracellular environment. Importantly, a detailed structural understanding of this efflux pump is required to fully exploit it as novel drug target. Although crystal structures of each individual subunit have been solved, no structural information of the entire complex is available. Recently, however, the structure of the fully assembled wild-type MexAB-OprM efflux pump was established by cryo-EM (figure 7), revealing an overall structure that resembled a vertically elongated rod. Based on this structure, a mechanism of drug efflux was proposed (figure 8) in which MexB exports antimicrobials by a three-step rotating mechanism towards the MexA-OprM channel. It is expected that this recent structure and increased functional understanding of the MexAB-OprM pump will open up new avenues in the design of novel drugs that that interfere with its function. For example, several promising inhibitors of bacterial multidrug efflux pumps have been designed to block the mechanism of drug efflux (18).
1. Aminov RI. 2010. A brief history of the antibiotic era: lessons learned and challenges for the future. Front Microbiol. doi: 10.3389/fmicb.2010.00134.
2. Munita JM, Arias CA. 2016. Mechanisms of Antibiotic Resistance. Microbiol Spectr. doi: 10.1128/microbiolspec.VMBF-0016-2015.
3. Eichenberger EM, Thaden JT. 2019. Epidemiology and Mechanisms of Resistance of Extensively Drug Resistant Gram-Negative Bacteria. Antibiotics. doi: 10.3390/antibiotics8020037.
4. Bassetti M, Vena A, Croxatto A, Righi E, Guery B. 2018. How to manage Pseudomonas aeruginosa infections. Drugs Context. doi: 10.7573/dic.212527.
5. Delmar JA, Su CC, Yu EW. 2014. Bacterial multidrug efflux transporters. Annu Rev Biophys. 43: 93-117.
6. Du D, Wang-Kan X, Neuberger A, van Veen HW, Pos KM, Piddock LJV, Luisi BF. 2018. Multidrug efflux pumps: structure, function and regulation. Nat Rev Microbiol. 16: 523-539.
7. Sennhauser G, Bukowska MA, Briand C, Grütter MG. 2009. Crystal structure of the multidrug exporter MexB from Pseudomonas aeruginosa. J Mol Biol. 389: 134-145.
8. Akama H, Kanemaki M, Yoshimura M, Tsukihara T, Kashiwagi T, Yoneyama H, Narita S, Nakagawa A, Nakae. 2004. Crystal structure of the drug discharge outer membrane protein, OprM, of Pseudomonas aeruginosa: dual modes of membrane anchoring and occluded cavity end. J Biol Chem. 279: 52816-52819.
9. Phan G, Benabdelhak H, Lascombe MB, Benas P, Rety S, Picard M, Ducruix A, Etchebest C, Broutin I. 2010. Structural and dynamical insights into the opening mechanism of P. aeruginosa OprM channel. Structure. 18: 507-517.
10. Akama H, Matsuura T, Kashiwagi S, Yoneyama H, Narita S, Tsukihara T, Nakagawa A, Nakae T. 2004. Crystal structure of the membrane fusion protein, MexA, of the multidrug transporter in Pseudomonas aeruginosa. J Biol Chem. 279: 25939-25942.
11. Tsutsumi K, Yonehara R, Ishizaka-Ikeda E, Miyazaki N, Maeda S, Iwasaki K, Nakagawa A, Yamashita E. 2019. Structures of the wild-type MexAB-OprM tripartite pump reveal its complex formation and drug efflux mechanism. Nat Commun. 10(1):1520. doi: 10.1038/s41467-019-09463-9.
12. Stover CK, Pham XQ, Erwin AL, Mizoguchi SD, Warrener P, Hickey MJ, Brinkman FS, Hufnagle WO, et al. 2000. Complete genome sequence of Pseudomonas aeruginosa PAO1, an opportunistic pathogen. Nature. 406: 959-964.
13. Masuda N, Sakagawa E, Ohya S, Gotoh N, Tsujimoto H, Nishino T. 2000. Substrate specificities of MexAB-OprM, MexCD-OprJ, and MexXY-oprM efflux pumps in Pseudomonas aeruginosa. Antimicrob Agents Chemother. 44: 3322-3327.
14. Murata T, Kuwagaki M, Shin T, Gotoh N, Nishino T. 2002. The substrate specificity of tripartite efflux systems of Pseudomonas aeruginosa is determined by the RND component. Biochem Biophys Res Commun. 299: 247-251.
15. Nakashima R, Sakurai K, Yamasaki S, Nishino K, Yamaguchi A. 2011. Structures of the multidrug exporter AcrB reveal a proximal multisite drug-binding pocket. Nature. 480: 565-569.
16. Yamanaka H, Tadokoro S, Miyano M, Takahashi E, Kobayashi H, Okamoto K. 2007. Studies on the region involved in the transport activity of Escherichia coli TolC by chimeric protein analysis. Microb Pathog. 42: 184-192.
17. Murakami S, Nakashima R, Yamashita E, Matsumoto T, Yamaguchi A. 2006. Crystal structures of a multidrug transporter reveal a functionally rotating mechanism. Nature. 443: 173-179.
18. Mahmood HY, Jamshidi S, Sutton JM, Rahman KM. 2016. Current Advances in Developing Inhibitors of Bacterial Multidrug Efflux Pumps. 2016. Curr Med Chem.23: 1062-1081.