High resolution cryo-EM structure of the Pseudomonas aeruginosa OprM multidrug efflux pump provides target for rational inhibitor design

Abstract

OprM is the outer membrane channel component of resistance-nodulation-division (RND) efflux pumps in Pseudomonas aeruginosa, a high-priority human pathogen. OprM-dependent pumps play diverse roles in bacterial physiology and mediate resistance to critical antibiotic classes, including fluoroquinolones, aminoglycosides, and broad-spectrum β-lactam/β-lactamase inhibitor combinations such as ceftazidime/avibactam. As the outer membrane component of the tripartite efflux system, OprM gates substrate export through an iris-like periplasmic pore that opens upon association with its inner membrane partners. Here we present a single-particle cryo-EM structure of OprM at 2.08 Å, the highest resolution OprM structure to date. The structure reveals a closed leucine-lined periplasmic pore, consistent with previous X-ray structures. A detailed structure-based comparison of the closed and open states reveals the conformational changes accompanying periplasmic pore opening, including iris-like twisting of the coiled-coil helices, displacement of the gating Leu429 residues, and a broader reorganization of salt bridges than previously described. Conservation of the MexA-binding interface across multiple Mex proteins suggests a shared mechanism of OprM engagement. Three lipopolysaccharide (LPS) molecules are resolved at the β-barrel interface, bridging adjacent OprM monomers through a combination of hydrophobic interactions with the fatty acyl chains and specific polar contacts with the KDO, heptose, and phosphate groups. This binding mode resembles that of other generic β-barrels and is expected to be preserved in vivo. Together, these findings provide the most detailed structural characterization of OprM to date, offering new insights into its gating mechanism and membrane interactions with implications for the development of efflux pump inhibitors.

Graphical abstract

Highlights

• •Highest resolution (2.08 Å) structure of OprM to date.
• •Structural analysis of periplasmic pore opening reveals previously undescribed interactions and salt bridge rearrangements.
• •Cryo-EM structure confirms previously reported OprM closed state interactions.
• •Three LPS molecules at the β-barrel interface reveal the structural basis for OprM anchoring in the outer membrane.
• •Conserved MexA-binding interface suggests a shared OprM interaction mechanism across multiple Mex efflux pump components.

1. Introduction

Antimicrobial resistance (AMR) represents a serious threat to global public health, and it has been estimated that the attributable mortality of antibiotic-resistant infections may reach 39 million by 2050 without substantial therapeutic innovation (Naghavi et al., 2024). Pseudomonas aeruginosa is a Gram-negative bacterium that has been characterized by the World Health Organization (WHO) as one of the top ten pathogens responsible for serious infection of hospitalized patients, with an estimated prevalence of greater than 7% among hospital acquired infections (Magill et al., 2014; Weiner et al., 2016; WHO Bacterial Priority Pathogens List, 2024). In the contexts of both nosocomial and community acquisition, multidrug resistant (MDR) P. aeruginosa is an important cause of infection with significant morbidity and mortality in individuals with cystic fibrosis, chronic obstructive pulmonary disease, and other immunocompromising conditions (Ledger et al., 2024; Feng et al., 2017; Qin et al., 2022; Rossi et al., 2021; Rodrigo-Troyano et al., 2018).

A number of intrinsic mechanisms contribute to antibiotic resistance in P. aeruginosa, including the low-permeability barrier of the outer membrane, the production of biofilms, and the expression of beta lactamases and efflux pumps (López et al., 2017; Liu et al., 2024; Hancock, 1998; Trias et al., 1989; Strateva and Yordanov, 2009). Efflux pumps are protein complexes that span the inner and/or outer membranes and export exogenous and endogenously produced compounds from the cytoplasm into the periplasm or extracellular space (Fig. 1) (Webber, 2003). The often broad chemical transport specificities of bacterial efflux pumps make them highly effective mediators of antibiotic resistance (Nikaido and Pagès, 2012). Efflux pumps in the resistance-nodulation-division (RND) superfamily are particularly important contributors to AMR in Gram-negative bacteria (Nishino et al., 2009; Puzari and Chetia, 2017; Dulanto Chiang and Dekker, 2024). These protein complexes comprise a tripartite protein structure assembled from three separate components which span the inner membrane, periplasm, and outer membrane, respectively. The inner membrane pump (IMP) component is a trimeric H+/substrate antiporter that couples the proton motive force (PMF) and substrate export cycle. Previous work on model IMPs, including MexB from P. aeruginosa, has demonstrated that the IMP forms an asymmetric trimer that mediates an alternating-access transport cycle through three different conformations (Nikaido and Takatsuka, 2009; Yamaguchi et al., 2015). Through these alternating conformations, the pump opens a periplasm- or cytoplasm-facing permeation pathway to permit diffusion of small molecules including antibiotics into the central cavity, closes this pathway, and then opens a pathway that runs through the OMP to permit diffusion into the extracellular medium. The IMP associates on the periplasmic side with the membrane fusion protein (MFP), which bridges the inner and outer membrane components. The MFP then binds to the outer membrane protein (OMP), forming the exit channel (Fig. 1) (Ohene-Agyei et al., 2012; Dinh et al., 1994; Paulsen et al., 2006).

Fig. 1.

Schematic of RND efflux pumps in Gram-negative bacteria. Foreign bodies (red), like antibiotics, enter Gram-negative bacteria across the outer membrane. Each RND efflux pump is comprised of an Inner Membrane Pump (IMP, shades of pink), Membrane Fusion Protein (MFP, shades of green), and Outer Membrane Protein (OMP, shades of blue). As a defense mechanism, an RND efflux pump expels the foreign body from the cell using the PMF across the inner membrane as a source of energy. Protons are blue circles. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

In total, twelve RND efflux pump systems have been identified in P. aeruginosa (Webber, 2003; Nikaido and Pagès, 2012; Poole, 2004; Poole, 2005; Poole, 2000; Piddock et al., 2010; Zechini and Versace, 2009; Blair et al., 2014; Blair et al., 2015; Sun et al., 2014; Chuanchuen et al., 2002; Ranjitkar et al., 2019). OprM forms the shared efflux conduit in the outer membrane for several of these systems, including MexAB, MexVW, MexXY, MexJK. Different pumps export different compounds with broad transport specificities, including antibiotics. In particular, MexAB-OprM can transport sulfonamides, fluoroquinolones, cephalosporins, and carbapenems; MexVW-OprM can transport cephalosporins; and MexXY-OprM can transport aminoglycosides, tetracyclines, macrolides, chloramphenicol, and quinolones (Aires et al., 1999; Masuda et al., 2000a; Masuda et al., 2000b; Dreier and Ruggerone, 2015; Dulanto Chiang et al., 2022; Li, 2003). RND efflux pumps can be overexpressed through a variety of mutations that occur in promoter regions and regulatory proteins (Blair et al., 2015; Li et al., 2015). Clinical strains of P. aeruginosa can become hypermutators through mutations in their mismatch repair systems, and these strains may rapidly generate mutations that result in overexpression of RND efflux pumps (Dulanto Chiang et al., 2022; Mehta et al., 2019; Eliopoulos and Blazquez, 2003; Schaaff et al., 2002). In particular, MexAB-OprM and MexVW-OprM can confer resistance to broad-spectrum beta lactam-beta lactamase inhibitor combination antibiotics such as ceftazidime/avibactam when overexpressed (Dulanto Chiang et al., 2022; Cabot et al., 2014; Lahiri et al., 2015; Berrazeg et al., 2015).

Previous studies have demonstrated that deletion of OprM results in sensitization to important antibiotic classes, confirming the essential role in mediating resistance (Wang et al., 2010; Cunrath et al., 2019; Li and Poole, 2001). Given its central role in both intrinsic and acquired resistance, OprM thus represents an accessible, high-priority target for the development of novel therapeutic inhibitors. Additionally, OprM has been identified as a receptor required for successful infection by certain bacteriophages, including bacteriophage OMKO1 (Chan et al., 2018; Chan et al., 2025; Chan et al., 2016). Phage-driven inactivation of OprM can compromise efflux function and sensitize to antibiotics. Previous work has demonstrated that this may create a genetic trade-off that can be exploited therapeutically, in which phage pressure can sensitize bacteria to antibiotics, supporting the rationale for phage-antibiotic combination strategies (Chan et al., 2018; Chan et al., 2016; Burmeister et al., 2020).

High-resolution structural information can provide essential insights into the mechanisms underlying substrate recognition and transport and may provide the basis for rational inhibitor design and effective phage and monoclonal antibody-based therapies. Structures of OprM have been determined by cryo-EM in complex with MexA and MexAB, and a few OprM structures alone (Tsutsumi et al., 2019; Glavier et al., 2020). Although several X-ray crystal structures of OprM without other protein components have previously been reported, high resolution details at the single particle level remain to be characterized (Phan et al., 2010; Monlezun et al., 2015; Akama et al., 2004). Here we present a single particle cryo-EM structure of OprM at 2.08 Å, the highest resolution structure to date, and perform a detailed analysis of the conformational changes involved in the periplasmic pore opening, by comparing the closed OprM structure to the 3.2 Å open OprM-MexA structure (PDB ID: 6TA5, (Tsutsumi et al., 2019)).

2. Results and discussion

To obtain the cryo-EM structure of Pseudomonas aeruginosa OprM, the outer membrane protein was expressed in Escherichia coli (E. coli) outer membranes and purified. Briefly, OprM was extracted from the membrane with n-Dodecyl β-D-maltoside (DDM) and exchanged into Non-Ionic Amphipol (NAPol). Size-exclusion chromatography (SEC) was used as a final step to obtain pure protein for further characterization (Supplemental Fig. 1a,b). The correct formation and expected trimeric stoichiometry of the OprM complex were confirmed via analytical ultracentrifugation (AUC) and mass photometry (MP) (Supplemental Fig. 1c,d). Both analyses showed a combination of full trimeric complex (∼159 kDa) and monomer (∼53 kDa). The AUC absorbance profile shows species at 5.15 S and 10.31 S, accounting for 78% and 17% of the total signal, respectively (Supplemental Fig. 1c). The interference profile shows a species at 3.87 S, presumed to be a NAPol micelle. The species at 5.15 S has a protein mass contribution of 47 ± 9 kDa with ∼50 kDa of NAPol (complex mass is 98 ± 20 kDa), whereas the species at 10.31 S has a protein mass contribution of 178 ± 30 kDa with ∼90 kDa of NaPol (complex mass is 271 ± 53 kDa). Similarly, MP indicated two species at 97 ± 25 kDa and 204 ± 58 kDa (Supplemental Fig. 1d). However, MP data displays more trimeric OprM than monomeric. While OprM is not consistently observed in a fully assembled trimeric state, the timescales governing its assembly and disassembly remain unclear. Cryo-EM grids of OprM were frozen and data were collected on a 300 kV Titan Krios G3 cryo-TEM. The micrographs obtained confirmed intact homotrimeric complexes (Supplemental Fig. 2).

We next determined the three-dimensional structure of the homotrimer by single-particle cryo-EM analysis, and iterative refinement of the OprM structure at 2.08 Å (Methods, Fig. 2a,b,c, Supplemental Fig. 2). The resulting cylindrical structure has two distinct regions, consistent with published OprM structures: a right-twisted β-barrel followed by a periplasmic left-twisted α-helical core. Each monomer contributes four β-strands and two extracellular loops to the 12-stranded β-barrel domain that spans the outer membrane. These six extracellular loops represent the main surface-exposed components of the transporter. The 12 residue Loop 1 (Leu116-Ala127), extends into the central axis of the pore and possibly plays a role in restricting diffusional access to the large β-barrel cavity (Phan et al., 2010). Loop 2 is 13 amino acids in length (Ser327 – Gly339) and extends from the outer surface of the β-barrel. A ring of conserved phenylalanine residues lines the lipid-water interface between the outer membrane β-barrel and the periplasmic α-helical core. The β-barrel is connected to the core region of OprM, comprising eighteen α-helices that extend into the periplasm by ∼100 Å. This α-helical core can be further divided into an α-barrel, an equatorial domain and a coiled-coil domain (Phan et al., 2010). Each monomer contributes two long and four short α-helices. Each pair of short helices stack end-to-end to form pseudo-continuous helices that span the same distance as the long helices. This region forms the bulk of the channel through which transported molecules pass. At the N-terminus, OprM undergoes lipidation on a cysteine residue, although not visible in the structure presented in this work. The lipid moiety is believed to insert into the outer membrane, providing a stable membrane anchor. Removal of the lipid group does not appear to affect protein function (Nakajima et al., 2000). Both the β-barrel and periplasmic pore have a triangular profile when viewed on the axis of the channel. The coiled-coils between helices 3/4 and 7/8 on the periplasmic end of OprM can interact with the six α-helical hairpin turns (2 per monomer) of an MFP to dock OprM to the inner membrane transporter (Glavier et al., 2020). Together, these four regions of OprM contribute to the assembly of the continuous tripartite channel and efflux pump system.

Fig. 2.

Single particle Cryo-EM structure of OprM in NAPol. a Cartoon view of OprM homotrimer with monomers shown in blue, green and yellow. The dimensions of the protein are indicated. The outer membrane is represented by a grey rectangle. b Cryo-EM map of OprM with monomers in different colors. The grey density at the top is LPS. c Top and bottom cartoon views of OprM with transparent molecular surface. In the bottom view the three Leu429 are shown. d Close-up view of the structure from the periplasm, looking toward the outer membrane. Leu429 side chains from the 3 monomers are located ∼3 Å from each other. Asn427 acts as an additional barrier to the cap created by the 3 leucines. e Cutaway view of molecular surface for closed and open states (PDB: 6TA5) of OprM. Residues involved in interaction with MexA are highlighted as in Fig. 3a. f Interactions near the periplasmic pore in the closed and open state of OprM. Dotted lines indicate hydrogen bonds (black) and salt bridges (red) that stabilize the closed conformation. Only interactions between alpha helices 4 and 8 from monomer A (blue) with helices 7 and 8 from monomer B (green) are shown, with residues labeled with their respective colour. Images were created in ChimeraX. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

When not in complex with an MFP and IMP, OprM has a closed pore that prevents the molecule from being a constitutively open channel. As seen in previous OprM closed state structures, three highly conserved Leu429 contributed by each monomer form a hydrophobic pore at the periplasmic end of the molecule (Fig. 2d) (López et al., 2017; Phan et al., 2010; Monlezun et al., 2015; Akama et al., 2004). This site is the main constriction along the central axis in the structure, followed by an additional barrier from the amides of the Asn427 carboxamide groups (Fig. 2d). The pore is surrounded by a network of hydrogen bonds between Arg436-Asp433 and Gln225-Tyr413 across two different monomers (Fig. 2f). In the closed state, the channel near the periplasm completely closes (∼3 Å distance between Leu429 sidechain atoms but the molecular surface is closed), while previously published structures in an open conformation are ∼25 Å in diameter (Fig. 2e, Supplemental Fig. 3b-c) (Tsutsumi et al., 2019; Glavier et al., 2020). Every OprM structure determined in the absence of an MFP has been found in the closed state, all of which were solved by X-ray crystallography. The closed-state conformation observed here is consistent with previous X-ray structures, with an r.m.s.d. of 0.29 Å compared with the OprM X-ray structure 3D5K. Cryo-EM captures proteins in a more native, solution-like state, free from the crystal packing constraints that can artificially order or disorder sidechains in X-ray structures (Supplemental Fig. 3a). The published X-ray structure contains several bound Na⁺ and one Cl⁻ ion. There is weak density visible at the corresponding sites in the map of the present structure, though ions were not modeled to avoid over-interpretation. This nonetheless suggests that at least some of these ion-binding sites are not crystallographic artifacts. The observation of a closed state by cryo-EM supports the validity of previous OprM closed state structures, suggesting they were neither artifactual nor influenced by crystal packing (Tsutsumi et al., 2019; Eyal et al., 2005). Overall, the hydrophobic interactions gating the periplasmic region of OprM are a key structural characteristic for antibiotic efflux.

The C-terminal coiled-coil domain helices form an iris-like gating mechanism on the periplasmic end of the molecule. From the closed ground state, the periplasmic pore can fully open by a synchronized twist of the coiled-coil helices, which can be visualized by the interpolation of a continuous transition trajectory connecting the closed OprM structure presented in this work and the open OprM-MexA structure (PDB ID: 6TA5, (Phan et al., 2010)) as end point structures (Fig. 2e, Supplemental movie). This twisting motion of the helices is localized mainly in the coiled-coil, up to the equatorial domains, while the rest of the molecule is relatively rigid (Supplemental Fig. 3d). In the open state, existing H-bond distances increase 0.1–0.2 Å between some of the β-barrel strands. The pseudo-continuous helices are kept in alignment by the equatorial domain and seem to act like springs. A relatively small movement of the helices at the level of the equatorial domain translates into their large displacement at the periplasmic end of the molecule. Despite this large movement some H-bonding interactions between coiled-coil helices from adjacent monomers are well-preserved in the transition from the open to the closed conformation: Gln225 with Tyr413, Arg242 with Thr395 and Gln399, Gln246 with Tyr412, and Asn249 with Asp388 (Fig. 2f, Supplemental Table 2). Arg436 forms four salt bridges with Asp433 in the closed state. While these salt bridges have been shown to dissociate upon transition to the open state (Tsutsumi et al. 2019), they are not the only salt bridges disrupted during this process. Our analysis identifies seven additional salt bridges that are disrupted, three that remain intact, and six that are newly formed (Supplemental Table 3).

When the pore fully opens, the three Leu429 residues completely swing out of the way, and the coiled-coil domain widens to the full 25 Å diameter of the α-barrel domain. This is broadly consistent with the mechanism proposed by Phan et al. (2010), though that study, based on normal mode analysis of the closed state X-ray structure, underestimated the extent of periplasmic pore opening and could not resolve the structural details of the open state (Phan et al., 2010). The detailed interaction analysis presented here is structure-based and reveals additional interactions not previously described. Analysis with the program HOLE further confirms that the channel is mostly rigid in its upper half, with flexibility confined from the equatorial domain to the periplasmic pore region.

As previously mentioned, in order for the tripartite RND efflux pump to be active, the periplasmic pore region is expected to open upon association with the inner membrane complex consisting of the IMP and MFP components (Tsutsumi et al., 2019; Glavier et al., 2020; Touzé et al., 2004; Wang et al., 2017). Specifically, cryo-EM structures have shown that the periplasmic pore of OprM is in an open state when bound to MexA (Tsutsumi et al., 2019; Glavier et al., 2020). The structures of OprM with MexA reveal the important residues involved in this protein-protein interaction, which were mapped to the structure presented in this work (Fig. 3a). Several hydrogen bonds are present in the open conformation: Gln104 in MexA binds to Ala220 or Tyr428 in OprM, and Lys108 in MexA binds to Gly216 or Gly424 in OprM (Supplemental Fig. 4b-c). In addition, the main chain of Ala105 in MexA forms a hydrogen bond with Arg420 in OprM, whereas Leu100 in MexA forms hydrophobic interactions with Val215 and Val217 of OprM. In the structure presented in this work, key residues such as Asp433 and Arg435 are inaccessible to a periplasmic MFP, as they are buried within the closed conformation. Tyr413 forms a hydrogen bond with Gln225 in the closed conformation (Fig. 2f), which is disrupted upon transition of OprM to the open conformation. Disruption of these key interactions might require physical interaction and energy from its IMP and MFP partners. The residues involved in these interactions are highly conserved among other homologs (OprN and OprJ) and must be essential for complex formation (Supplemental Fig. 5).

Fig. 3.

The OprM β-barrel in the outer membrane. a Residues interacting with MexA mapped on the OprM pore molecular surface, viewed from the periplasm. Comparison of the closed and open (PDB: 6TA5) states. Colors show the same residues in different monomers. b Top (from the extracellular space) view of the bound lipopolysaccharide (LPS, orchid) surrounding the β-barrel. Each LPS interacts with two OprM monomers. c Interactions between the LPS and the β-barrel. The glucosamine (GlcN) and Heptose (GMH) glycan rings and the phosphate group interact with charged atoms on the barrel. The fatty acyl chains from the Lipid A moiety form hydrophobic interactions with hydrophobic sidechains on the outside of the barrel. d Molecular surface of OprM with electrostatic charge. Hydrophobic areas (white) are proximal to the lipid groups in each LPS. The phosphate groups (orange) on the disaccharide backbone of LPS provide a negative charge that interacts with the positively charged amino acids of OprM (blue). Electrostatic surface properties were analyzed and visualized with ChimeraX. e β-barrel superposition of OprM (blue) with an OprM crystal structure (reddish, PDB: 3D5K) and an OprM cryo-EM structure (green, PDB: 6TA5). The view of the loops is from the extracellular space looking toward the membrane and periplasmic space. Extracellular loops 1 block the top of the β-barrel lumen. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

To date there are only structures of OprM interacting with MexA available (Tsutsumi et al., 2019; Glavier et al., 2020). However, at least four additional proteins, MexV, MexX, MexJ, and MexM bind OprM. These Mex proteins share identical amino acids corresponding to Leu100 and Ser107 of MexA and similarity among residues 96–108. The alignment of predicted AlphaFold2 structures shows high structural homology at the OprM binding interface (Supplemental Fig. 4a). The high degree of conservation in this region suggests that other Mex proteins may bind to OprM at the same site as MexA (Supplemental Fig. 4d).

In addition, the structure presented in this work contains density for three E. coli lipopolysaccharide (LPS) molecules; the most commonly found lipid in the outer membrane of P. aeruginosa (Lyon et al., 2022). Although P. aeruginosa and E. coli LPS differ substantially in their outer core glycans and O-antigen, the region visible in the OprM density shares the same overall topology between the two species, and common structural features mediate binding. Each LPS molecule bridges two OprM monomers at the interface of their contacting β-strands in the β-barrel (Fig. 3b,c). The fatty acid chains of Lipid A interact with the hydrophobic exterior of the OprM β-barrel, mimicking the outer membrane environment (Fig. 3c), while its negatively charged phosphate groups interact with a cluster of positively charged residues (Arg113, Arg115, Gln329) on the adjacent OprM monomer (Fig. 3d). E. coli fatty acyl chains are longer than those of P. aeruginosa; however, their hydrophobic, non-specific interactions with the β-barrel are unlikely to differ functionally. The observed LPS comprises a Lipid A moiety with an inner core of two keto-deoxyoctulosonate (KDO) and three heptose (GMH) residues. One Lipid A glucosamine (GlcN) residue interacts with Arg113 from β-strand 1, while the Lipid A phosphate group and the terminal heptose interact with Arg115. The two arginine residues are not strictly conserved, but the positive charge is conserved for the corresponding residue in OprJ and OprN (Supplemental Fig. 5). Lipid A is well-resolved in the density due to its direct protein contacts, whereas the non-interacting inner core glycans are more flexible and only visible at lower contour levels. Since OprM engages the conserved core components of LPS, the same binding mode can be expected in vivo, with the additional outer core glycans and O-antigen projecting into the extracellular space away from the barrel.

The mode of LPS binding to OprM resembles that observed for generic β-barrels such as FhuA (PDB: 2FCP, (Ferguson et al., 1998)), involving non-specific interactions, rather than the highly specific, metal cation-mediated binding seen in homotrimeric porins such as OmpE36 (PDB: 5FVN, (Arunmanee et al., 2016). No direct functional role for LPS in the OprM transport mechanism has been reported. To date, the only documented case of LPS serving a direct functional role in a Pseudomonas β-barrel is OprH, where LPS binding stabilizes the outer membrane and enhances antibiotic resistance by displacing divalent cations (Edrington et al., 2011).

The OprM structure reveals side chain conformations in its well-ordered extracellular exposed loops at high resolution. Among these, loops 1 and 2 shield the β-barrel from the extracellular space. These loops exhibit a compact orientation consistent with previously published structures of both the closed and open OprM states, and the cryo-EM structure presented in this work confirms that this positioning is not an artifact of crystal contacts (Fig. 3e). Sequence alignment of OprM, OprN, and OprJ shows that these extracellular loops differ between transporters (Supplemental Fig. 5), and this sequence divergence likely reflects evolutionary adaptations to the distinct substrate specificities of each RND efflux system. The OprM β-barrel narrows from 13 Å at its widest to 6 Å at the extracellular loops, imposing an additional structural barrier to substrate passage (Supplemental Fig. 3b–d).

3. Conclusion

We present the highest resolution structure of OprM to date, providing new insights into the gating mechanism, protein-protein interactions, and membrane environment of this clinically important efflux pump component. The closed-state conformation observed by cryo-EM is consistent with previous X-ray structures. The structure-based comparison of closed and open states reveals a broader reorganization of interactions during periplasmic pore opening than previously appreciated, including additional salt bridge rearrangements not described before. Conservation of the MexA-binding interface across multiple Mex proteins points to a shared mechanism of OprM engagement that may be exploited therapeutically. The resolution of bound LPS molecules at the β-barrel interface offers the first structural view of OprM in a membrane context. Together, these findings advance our understanding of RND efflux pump function in P. aeruginosa and provide a structural framework to guide the rational design of efflux pump inhibitors as a strategy to combat antimicrobial resistance.

4. Methods

4.1. Bacterial strain, plasmid design, and cloning

The full-length oprM gene (PA0427) from P. aeruginosa strain PAO1 (Genome assembly accession: GCF_000006765.1) was synthesized and cloned into the pET-21b(+) expression vector (Novagen), with C-terminal hexahistidine (6×-His) tag and a TEV protease cleavage site. The resulting construct, oprM_pET-21b(+), was subsequently transformed into E. coli C43 (DE3) (OverExpress C43(DE3), Sigma Aldrich) cells for protein expression.

4.2. Protein expression

Transformed E. coli C43 (DE3) were plated onto LB agar plates containing 50 μg/ml carbenicillin and grown overnight in a 37 °C incubator. Transformed colonies were added to 50 mL starter cultures of LB media supplemented with 30 μg/ml carbenicillin and grown at 37 °C shaking at 220 rpm. After 1 h of growth, 25 mL of starter cultures was used to inoculate a flask containing 1 L of 2xYT medium supplemented with 30 μg/ml carbenicillin. Cultures were grown at 37 °C at 220 rpm until OD₆₀₀ reached 0.6–0.8 and then induced with isopropyl β-D-1-thiogalactopyranoside (IPTG) at a final concentration of 0.2 mM. Cells continued to grow overnight at 22 °C and 220 rpm. After 16–18 h, cells were harvested at 5422 xg for 10 mins and pellets were stored at −80 °C for future use.

4.3. Protein purification

Cells were thawed and resuspended in 50 mM Tris-HCl pH 7.4, 200 mM NaCl for 15 mins enhanced with 200 μM 4-(2-aminoethyl)benzenesulfonyl fluoride (AEBSF, Goldbio), 200 μM DNase I (GoldBio), 50 μM lysozyme (GoldBio), and one cOmplete protease inhibitor cocktail tablet (Roche). Cells were lysed by two passages through an EmulsiFlex-C3 (Avestin) at ∼15,000 psi. Membranes were pelleted by ultracentrifugation in a Type 70 Ti Beckman rotor at 130,000 ×g for 1 h at 4 °C. The membrane pellets were then resuspended in 50 mM Tris-HCl pH 7.4, 200 mM NaCl, 40 mM imidazole, and solubilized with n-Dodecyl β-_D-maltoside (DDM, Anatrace) at a final concentration of 1%. Membranes were solubilized for 1 h at 4 °C. Insoluble material was pelleted by ultracentrifugation in a Type 70 Ti Beckman rotor at 130,000 ×g for 1 h at 4 °C. The solute was immediately loaded onto a HisTrap HP prepacked His tag protein purification column (Cytiva) on an AKTA Purifier (GE Healthcare). The column was equilibrated in buffer A: 50 mM Tris-HCl pH 7.4, 200 mM NaCl, 0.05% DDM, and 40 mM imidazole (GoldBio). Protein was eluted with 70% buffer A and 30% buffer B: 50 mM Tris-HCl pH 7.4, 200 mM NaCl, 0.05% DDM, and 1 M imidazole. Fractions were collected at 2 mL increments and run on an SDS-PAGE gel. The gels were analyzed via Coomassie Stain (Abcam) and Western blot incubated with an α-His-HRP antibody (Sigma-Aldrich) and imaged with SigmaFast 3,3’-Diamino-benzidine tablets (Sigma-Aldrich).

Desired fractions were pooled and concentrated using an Amicon Ultra-15 Centrifugal Filter Unit with a 50 kDa MW cut-off (Millipore). The protein was reconstituted into the amphipol, NAPol, at three times the concentration of the protein complex sample (Anatrace). After rocking for 1 h at 4 °C, 1 g of treated biobeads (Bio-Rad) were added to the mixture overnight. The next day, the biobeads were filtered from the protein. The sample was further purified by size-exclusion chromatography (SEC) using a Superose6 Increase 10/300 column (GE Healthcare) at a flow rate of 0.5 mL/min. The buffer consisted of filtered 50 mM Tris-HCl pH 7.4, 200 mM NaCl and fractions were collected at 0.5 mL increments. The eluted fractions were analyzed via Coomassie Stain and Western blot as described above. The fractions of interest were concentrated using an Amicon Ultra-0.5 mL Centrifugal Filter Unit with a 50 kDa MW cut-off (Millipore) and stored at 4 °C for further analysis.

4.4. Mass photometry

OprM was diluted to 20-25 nM in filtered 50 mM Tris-HCl pH 7.4, 200 mM NaCl. Mass photometry measurements were acquired on a Refeyn OneMP mass photometer with Acquire MP 2023 R1(Refeyn Ltd) (Young et al., 2018). All measurements were carried out in a silicone gasket (Grace Bio-Labs) on a glass microscopy slide. Protein samples (10 μL) were added to the gaskets containing 10 μL buffer and images were acquired for 60s. All data collections were analyzed using DiscoverMP version 2021 R1 (Refeyn Ltd) (Young et al., 2018). Masses were calculated from measurements on calibrated standards.

4.5. Sedimentation velocity analytical ultracentrifugation (SV-AUC)

Sedimentation velocity was carried out at 50,000 rpm (195,650 x g at 7.0 cm) and 20 °C on a Beckman Coulter ProteomeLab XL-I analytical ultracentrifuge and An50-Ti rotor. A sample of OprM with NAPol in SEC buffer (50 mM Tris-HCl pH 7.4, 200 mM NaCl) was studied in 12 mm two-channel centerpiece cells, and data were analyzed in SEDFIT in terms of a continuous c(s) distribution of sedimenting species (Schuck, 2000). The solution density, viscosity, protein extinction coefficient, and protein partial specific volume were calculated in SEDNTERP (Philo, 2023). The protein refractive index increment was calculated in SEDFIT (Schuck, 2000). The partial specific volume for NAPol was calculated based on its chemical composition following the method of Durchschlag and Zipper, and a refractive index increment of 0.14 cm³g⁻¹ was assumed (Durchschlag and Zipper, 1997). Absorbance and interference c(s) distributions were analyzed simultaneously using the fitted ƒ/ƒ_o membrane protein calculation module in GUSSI to obtain the protein and amphipol contributions to the sedimenting complex of interest (Brautigam, 2015).

4.6. Cryo-EM sample preparation

3 μL of OprM (absorbance at 280 nm = 2.0) were applied to freshly glow-discharged Quantifoil R1.2/1.3, copper, 300 mesh grids (Electron Microscopy Sciences, Protochips, Inc.) for 45 s at 15 mA (Pelco easiGlow, Ted Pella, Inc.). The grids were blotted for 7 s (100% humidity, 22 °C, and 5 blot force) and were then plunge-frozen in liquid nitrogen-cooled liquid ethane using a FEI Vitrobot Mark IV system (Thermo Fisher Scientific) after 0 s wait time.

4.7. Cryo-EM data acquisition

The cryoEM dataset was collected on a Titan Krios G3 (Thermo Fisher Scientific) operating at 300 kV with an Imaging Filter Quantum LS and a K3 direct electron detector (Gatan). The pixel size was 0.415 Å/pixel in super resolution mode and micrographs were collected as dose-fractionated movies with SerialEM in counting mode, with 40 frames and a total dose of 50.70e/Ų (Mastronarde, 2005). Detailed parameters for cryoEM data collection are listed in Supplemental Table 1.

4.8. Cryo-EM image processing

The single particle data analysis was performed following the standard procedures in cryoSPARC v4.5.3 with few modifications as summarized in Supplemental Figs. 2 and Supplemental Table 1 (Punjani et al., 2017; Punjani et al., 2020). 6259 micrographs were imported to cryoSPARC v4.5.3 and movies were gain corrected, aligned, and binned by a factor of two. Micrographs were denoised using a grey scale normalization factor of 1.5. Particles were selected using Blob Picker and a total of 4,367,911 particles were extracted with a binning of 2 (box size of 128 × 128, 1.66 Å/px). After one round of 2D-classifications, 1,656,108 particles were used for ab initio modeling followed by heterogeneous refinement, both with four classes and C3 symmetry. One of the classes contained 818,360 particles and 3.40 Å resolution. These particles were re-extracted at a box size of 320 × 320. Then, the particles underwent three rounds of ab initio modeling followed by heterogeneous refinement, both with three classes and C1 symmetry. After the final round of heterogenous refinement, the hand was flipped for the best class, which contained 192,828 particles, and was subjected to non-uniform refinement. This yielded a 2.37 Å density map and underwent reference-based motion correction. These particles were used for another non-uniform refinement, which yielded a 2.18 Å density map. Symmetry expansion (C3) was performed, and the resulting particles were used for a final local refinement, which yielded our final density map at 2.08 Å.

4.9. Model building and refinement

The structure was built starting with the coordinates of the OprM crystal structure 3D5K by fitting into the maps with Coot (Phan et al., 2010; Emsley et al., 2010). The structure was iteratively rebuilt in Coot and real-space refined in Phenix v1.21 (Adams et al., 2010). Poorer density regions were rebuilt using DeepEMHancer maps and models generated with Model Angelo (Sanchez-Garcia et al., 2021; Jamali et al., 2024). The geometry of the final structure was further improved with ISOLDE (Croll, 2018). Electrostatic surface properties (calculated using the Linearized Poisson-Boltzmann Equation mode with a solvent radius of 1.4 Å) were analyzed and visualized using ChimeraX (Pettersen et al., 2021). Pore and channel analysis of OprM were calculated with HOLE and visualized using ChimeraX (Pettersen et al., 2021; Smart et al., 1996).

4.10. Sequence alignment

Sequences for OprM and homologous species were obtained from UniProtKB (UniProt, 2024; Madeira et al., 2024). Multiple sequence alignments were completed using Clustal Omega and edited with Jalview (Waterhouse et al., 2009). The final alignment with sequence similarities highlighted, was produced with ESPript 3.0, with sequence similarities depiction parameters set to %Equivalent, global score of 0.7, and flashy colour scheme output (Robert and Gouet, 2014).

CRediT authorship contribution statement

Anna C. Ratliff: Writing – review & editing, Writing – original draft, Investigation, Formal analysis. Istvan Botos: Writing – review & editing, Investigation, Formal analysis. Amanda M. Putti: Writing – review & editing, Writing – original draft, Investigation. Prashant P. Patil: Writing – review & editing, Writing – original draft, Investigation, Formal analysis. Arthi Ramkumar: Writing – review & editing, Investigation. Rodolfo Ghirlando: Writing – review & editing, Investigation, Formal analysis. John P. Dekker: Writing – review & editing, Supervision, Project administration, Funding acquisition, Conceptualization. Susan K. Buchanan: Writing – review & editing, Supervision, Project administration, Funding acquisition, Conceptualization.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

Supplementary Video 1

Morph movie between the closed and open states of OprM.

Data availability

Data will be made available on request.