Skip to main content
NCBI home page
Protein Science : A Publication of the Protein Society logoLink to Protein Science : A Publication of the Protein Society
. 2024 Dec 14;34(1):e5252. doi: 10.1002/pro.5252

Insights into substrate recognition and export tunnel preferences in the efflux transporter AcrB

Farrukh Makhamadjamonov 1, Michal Emil Karolak 2, Lesley Smyth 3, Abdessamad Ababou 2,4,✉
PMCID: PMC11645668  PMID: 39673478

Abstract

In Escherichia coli AcrB is a major multidrug exporter, which confers the bacterium resistance to many antibiotics with diverse structural and chemical proprieties. Studies have identified three possible tunnels (or channels) within AcrB that different substrates use before reaching the distal pocket, from which they are subsequently extruded. Recently, we reported that mutations in the AcrB gate loop may affect the conformational change kinetics involved in substrate export rather than directly affecting molecular interactions with this loop, and we highlighted the distinct export tunnel preferences between erythromycin and doxorubicin. To further understand the gate loop's role in AcrB's export activity and the rationale behind substrate preferences among the three possible export tunnels, namely tunnel‐1, ‐2, and ‐3, we investigated the structural and functional effects of several single and multiple mutations in the gate loop of AcrB. Our findings indicate that all three tunnels are energetically favorable for the substrates studied, with the majority forming more hydrogen bonds in any tunnel compared to the distal pocket. Moreover, our experimental and computational data revealed that some substrates with high molecular similarity exhibited different export tunnel preferences, as strongly suggested by their MIC values. To explain this unexpected outcome, we propose a generalized explanation that the conformational change kinetics in AcrB is substrate‐dependent.

Keywords: AcrB–substrate interactions, bacterial efflux pumps, conformational change kinetics, substrate recognition, X‐ray crystallography

1. INTRODUCTION

AcrB is the primary efflux transporter in Escherichia coli. It is a homotrimeric integral membrane protein with an exceptionally broad substrate specificity, capable of exporting structurally diverse and unrelated compounds such as antibiotics, detergents, dyes, organic solvents, and biofuels (Foo and Leong 2013; Ma et al. 1996; Nikaido 2009; Nikaido and Takatsuka 2009; Tsukagoshi and Aono 2000; White et al. 1997; Yu et al. 2003). As a component of the tripartite efflux pump AcrAB‐TolC, along with its many homologues, it is constitutively expressed in many pathogenic Gram‐negative bacteria (Johnson et al. 2020; Li et al. 2020). Over the past two decades, our understanding of AcrB's molecular export mechanism has steadily advanced, since early crystal structures determination of its resting and active states (Murakami et al. 2006; Nikaido 2011; Seeger et al. 2006; Thanassi et al. 1997). The functional rotating mechanism was proposed to explain how AcrB exported it substrates, based on three distinct conformation of each of its protomers, namely access/loose, binding/tight, and extrusion/open states (Murakami et al. 2006; Seeger et al. 2006). The reporting of few crystal structures of AcrB with few substrates, such as Minocycline (MIY), Erythromycin (ERY), and Doxorubicin (DOX), revealed further details on the functional rotating mechanism, linking the two tunnels (or channels), namely tunnel‐1 (located next to the vestibule at each protomer) and tunnel‐2 (access or proximal pocket), to the deep binding pocket (distal pocket), along with highlighting the importance of the gate loop, in tunnel‐2, separating the proximal and distal pockets (Eicher et al. 2012; Murakami et al. 2006; Nakashima et al. 2011). This gate loop is highly conserved across many RND efflux transporters from various bacteria (Figure 1a). However, in some members, such as AcrD, AmrB, or SmeB, as well as others not shown, the lack of conservation at positions 615 and 617, and possibly 620, may suggest differences in the gate loop's role in their export activity (e.g., substrate specificity, export rate, etc.). Further details on this gate loop can be found in our previous study (Ababou 2018).

FIGURE 1.

FIGURE 1

Open in a new tab

Gate loop in AcrB and RND efflux transporters, and crystal structure and thermal stability of AcrB and gate loop mutants. (a) Sequence alignment of AcrB and other RND efflux transporters. Alignment was performed over the whole sequences, using ESPript 3.0 (Robert and Gouet 2014) (Figure S1, Supporting Information), but for clarity only the region covering the gate loop is shown. Sequences are from Escherichia coli (ecAcrB, ecAcrD, ecAcrF, ecMdtF), Salmonella typhimurium (stAcrB), Yersinia entrocolitica (yeAcrB), Klebsiella pneumoniae (kpAcrB), Vibrio parahaemolyticus (vpVmeB), Acinetobacter baumannii (abAdeJ, abAdeB), Pseudomonas aeruginosa (paMexB, paMexD), Agrobacterium tumefaciens (atAmeB), Burkholderia pseudomallei (bpAmrB), Neisseria gonorrhoeae (ngMtrD), Stenotrophomonas maltophilia (smSmeB). Strictly conserved residues are boxed in white on a red background and highly conserved residues are boxed in red on a white background. The green bar locates the 11 residues in the gate loop and AcrB's residues used for mutations or deletion, where the numbering is based on AcrB sequence. (b) Structural comparison between AcrB (green, pdb code 4ZIT) and its mutants, RA (blue, pdb code 7O3M), RK (magenta, pdb code 7O3N), and AA (yellow, pdb code 7O3L), in P21 space group (for clarity, only access monomer A is shown). (c) A close‐up view, from (b), at the export channel (proximal pocket, gate loop, and distal pocket), depicting the overall structural conservation. (d) A close‐up view, from (b), showing the superposition of the gate loops with residues at positions 615, 617, and 620. (e) Different view from (d) showing the differences between the gate loop in AcrB and the mutants. (f) 2Fo‐Fc electron density map of DDM molecule binding to monomer A (access monomer in trimer‐1) and contoured at 1.0σ. (g) Thermal denaturation plot of the apparent fraction of unfolded protein (Fapp) as a function of temperature.

Further AcrB's studies, inspired from previous reports (Husain et al. 2011; Husain and Nikaido 2010), have shown the existence of a third tunnel, tunnel‐3, which is open to the central cavity within the AcrB trimer, and connected directly to the distal pocket without passing by the gate loop (Nakashima et al. 2011; Zwama et al. 2018). These tunnels have been shown, to some extent, to have some specificity for AcrB substrates suggesting that substrates with similar molecular proprieties should be exported through the same export tunnels. Importantly, mutations that are far from the substrate's preferred tunnels are unlikely to impact the drug susceptibility assay for that substrate.

However, several substrates have been reported to have conflicting tunnels preferences for their export route, such as Benzalkonium (BZK) and Rhodamin 6G (R6G) that were reported to have preference to tunnal‐3 (Zwama et al. 2018), while other reports suggesting tunnel‐2 as their preferred tunnel (Cha et al. 2014; Muller et al. 2017). Structural and functional investigation of AcrB in the presence of Fusidic acid (FSA) resulted in the suggestion of a new tunnel located between the transmembrane helices TM1 and TM2, which may be putatively specific for FSA and carboxylated drugs in general (Oswald et al. 2016; Tam et al. 2020). Subsequently, it has been reported, by the same group, that FSA to be shown in the tunnel‐1 in AcrB‐FSA complex structure, suggesting its export route may also include preference to tunnel‐1 (Tam et al. 2021). Several reports have shown that ERY prefers tunnel‐2 as part of its export route (Ababou and Koronakis 2016; Bohnert et al. 2008; Eicher et al. 2012; Nakashima et al. 2011; Wehmeier et al. 2009). AcrB mutation E673G increased ERY resistance 2‐fold, while structurally no contact is detected between E673 and ERY (Kobayashi et al. 2014). While Q569R mutation has no effect on the resistance of ERY, as Q569 is located at the entrance to the proximal pocket (shortest contact ~10.0 Å), the double mutant Q569R‐E673G showed also no change in ERY resistance (Kobayashi et al. 2014). AcrB double mutant I626R‐E673G which decreased ERY resistance 8‐fold while I626R alone decreased ERY resistance 4‐fold (Kobayashi et al. 2014). AcrB quadruple mutant A33W‐T37W‐A100W‐N298W, part of tunnel‐3, resulted in 4‐fold increase susceptibility for ERY (Zwama et al. 2018). Recently, AcrB R717Q and R717L mutations reported to decrease macrolide and fluoroquinolone susceptibility but increased the susceptibility for Novobiocin and Cloxacillin (Zwama and Nishino 2022). These experimental observations clearly suggest that our understanding of substrate's preference export route in AcrB is still incomplete, and possibly still superficial. Beyond the obvious experimental difficulties and the complexity to investigate such large efflux transporter, the major contributors to such incomplete understanding of AcrB export activity at the molecular level are due to:

  1. Lack of solved AcrB‐substrate complex structures.

  2. Lack of discovery of intermediate structure forms of these complexes.

  3. Lack of experimental data on the molecular dynamics of these complexes.

  4. Lack of molecular kinetics aspects of these complexes (i.e., substrate binding, protomer conformational changes).

To better understand AcrB's substrate export tunnel preferences, we conducted structural and functional studies on various AcrB gate loop mutants, including single mutants (F615A, F617A, R620A, R620K, R620E), double mutants (F615A‐F617A, F615A‐R620A, F617A‐R620A), a triple mutant (F615A‐F617A‐R620A), and a deletion mutant (F615G‐Δ[G616‐R620]). Using several antibiotics, well‐documented in the literature as AcrB substrates, we demonstrate that (1) most AcrB substrates are primarily directed toward the export tunnels through hydrogen bond formation; (2) most of these substrates form significantly fewer hydrogen bonds within the distal pocket, facilitating their transport toward extrusion; and (3) substrates with high molecular similarity can exhibit substantial differences in drug susceptibility assays, ranging from a significant effect to no effect at all from the gate loop mutation. To explain these seemingly contradictory outcomes, we propose that substrate‐dependent conformational change kinetics, may generally explain how AcrB's export activity occurs.

2. RESULTS

2.1. Structures of the asymmetric trimers of AcrB mutants

Although we have co‐crystallized all the gate loop mutants with several antibiotics, the X‐ray diffraction data resulted in electron density maps that either have weak or no electron density to be assigned for those antibiotics within their reported binding sites or elsewhere in the structure (Eicher et al. 2012; Nakashima et al. 2011). Furthermore, the high similarity of the refined structures of R620A (RA), R620K (RK), and F615A‐F617A (AA) mutants, in P21 space group, with the structure of AcrB (Ababou and Koronakis 2016), and the limited structural information of these structures, we did not pursue further the structures of the other mutants. Indeed, comparison of RA, RK, and AA mutants' structures with our previously reported structure of AcrB (Ababou and Koronakis 2016) shows that overall, these mutants structures have high structural similarities as reflected by their low RMSD's (between 0.60 and 1.42 Å) (Figure 1 and Table S2, Supporting Information). However, slight structural differences may occur around the mutation positions, as shown in Figure 1d,e, where the highest changes was 3.8 Å between Cα of F617 in AcrB and A617 in AA. Detailed data collection and structures statistics are summarized in Table S1. Note for the refined structures one detergent molecule (dodecyl‐β‐D‐maltoside, DDM) per monomer was found in the mutants' structures (Figure 1f), as previously reported (Ababou 2018). However, we omitted the metal Ni in our refinement, even though the electron densities are clearly showing the presence of such metal (Figure S1), as previously reported and discussed in detail (Ababou 2018).

2.2. Gate loop mutations effect on the solution fold and stability of AcrB

Prior to perform any functional assay, we have checked whether the mutations affect the fold and stability of our mutants in solution. Consequently, we have used CD spectroscopy to collect the far‐UV CD spectra and the thermal unfolding data of AcrB and the mutants. The far‐UV CD spectra of AcrB and mutants show no substantial differences (Figure S2), suggesting that the mutants kept conserved secondary structures content and most probably overall similar structures to the WT in solution, as reported by our crystal structures of some of these mutants (Figure 1). The thermal denaturation results of the gate loop mutants revealed small shift in their melting curves as compared to the WT (Figure 1g). A minor stability of the WT over the mutants is observed as shown by the T m difference of less than 3.5°C, except for the mutant AA the T m was 4.1°C (Table S3). Such relatively small difference in T m suggests an overall similar stability between the WT and the mutants, and in particular at physiological temperature (Figure 1g). Hence, these results demonstrate that AcrB's gate loop mutations do not affect the fold and the overall stability of AcrB in solution, as previously reported (Ababou 2018; Ababou and Koronakis 2016).

2.3. Antibiotics susceptibilities of AcrB and mutants

Previously, we have shown the involvement of the gate loop in two mutants, AAA and ΔLoop, and two antibiotics ERY and DOX. To further investigate the importance of gate loop in the export activity of AcrB, in this work we have performed drug susceptibility assay for nine new mutants of the gate loop, along with F615A‐F617A‐R620A (AAA) and F615G‐Δ(G616‐R620) (ΔLoop), using 20 antibiotics (Figure S3). In the absence of antibiotics, except for cells selection using kanamycin, the cell growth of AcrB and all mutants were similar (Figure S4), indicating the absence of any effect of these mutations on either the growth or survival of the cells. Table S4 shows the effect of the acrB deficient E. coli (ΔacrB) and AcrB's single or combined mutations on the minimum inhibitory concentration (MIC) values for the 20 antibiotics. The MIC values of 12 and 26 μg/mL for Rifampicin (RIF) and Carbonyl cyanide 3‐chlorophenylhydrazone (CCP), respectively, were constant for ΔacrB, AcrB, and mutants (Table S4). Clearly, our results show that these antibiotics are not substrates of AcrB, as reported for RIF (Ohene‐Agyei et al. 2014). In fact, similar MIC values of RIF ranging from 8 to 16 μg/mL were reported (Schuster et al. 2016; Zwama et al. 2018); however, whether RIF is a substrate of AcrB is still unclear, in particular when the structure of AcrB binding to RIF was reported (Nakashima et al. 2011). CCP is rather well known to block efflux pumps by collapsing the proton motive force that energizes them (Mallea et al. 1998; Nikaido and Pages 2012). Although Phosphomycin (PPM) is not known to be a substrate of AcrB, we found a small but reproducible MIC ratio of AcrB to ΔacrB of 2. Consequently, RIF and CCP will not be included in any further investigations in this work.

For better analysis of the mutations' effect on AcrB's MIC for the antibiotics, we have calculated the MIC ratio of AcrB to the mutant (R), as shown in Table 1. For six antibiotics, DOX, Clindamycin (CLD), Tetracycline (TRC), Acriflavine (ACR), Nalidixic acid (NLD), and PPM, all the mutations have no effect on the susceptibility to these antibiotics (R = 1), while for Spiramycin (SPR) and Ciprofloxacin (CIP) the R value was also 1, except for only two mutants was 2, R620E (RE) and ΔLoop, and R620D (RD) and RE, respectively. For seven antibiotics, MIY, R6G, Moxifloxacin (MXF), Ofloxacin (OFL), Lomefloxacin (LMF), BZK, and PPM, all the mutations have increased susceptibility to these antibiotics (2 < R < 16). The highest susceptibility increase was for MIY with R = 16 for ΔLoop and R = 8 for the rest of the mutants. The lowest susceptibility increase was for OFL at R = 2 for all the mutants. For ERY and Novobiocin (NOV), all the mutations have increased susceptibility to these antibiotics (2 < R < 8), except of RA and RK where no susceptibility change was occurred (R = 1). For FSA, the mutants AA, AAA, and ΔLoop have increased susceptibility (R = 8) and no change for the rest of the mutants (R = 1). Instead for Chloramphenicol (CLP), all R620 single mutants and ΔLoop have R = 2 or 4, and the rest of mutants have no change in their susceptibility (R = 1). These results show that AcrB's gate loop is directly involved in the export of nearly 8 antibiotics and potentially not at all for the other 10 antibiotics.

TABLE 1.

Antibiotics susceptibilities of AcrB and gate loop mutants (MIC ratio, R, of AcrB to ΔacrB or mutants).

Substrate Abbr. MW DacrB F615A F617A RA RK RD RE AA F615A‐RA F617A‐RA AAA ΔLoop
Spiramycin SPR 841.1 16.0 1.0 1.0 1.0 1.0 1.0 2.0 1.0 1.0 1.0 1.0 2.0
Rifampicin RIF 823.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0
Erythromycin ERY 733.9 32.0 4.0 2.0 1.0 1.0 2.0 2.0 4.0 4.0 2.0 4.0 8.0
Novobiocin NOV 612.6 16.0 2.0 2.0 1.0 1.0 2.0 2.0 4.0 2.0 2.0 4.0 8.0
Doxorubicin DOX 543.5 64.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0
Fusidic acid FSA 516.7 16.0 1.0 1.0 1.0 1.0 1.0 1.0 8.0 1.0 1.0 8.0 8.0
Minocycline MIY 457.5 32.0 8.0 8.0 8.0 8.0 8.0 8.0 8.0 8.0 8.0 8.0 16.0
Tetracycline TRC 444.4 4.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0
Rhodamin 6G R6G 443.6 128.0 2.0 2.0 2.0 2.0 2.0 2.0 2.0 4.0 4.0 4.0 16.0
Clindamycin CLD 425.0 32.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0
Moxifloxacin MXF 401.4 32.0 4.0 4.0 2.0 2.0 2.0 2.0 4.0 4.0 4.0 4.0 4.0
Ofloxacin OFL 361.4 8.0 2.0 2.0 2.0 2.0 2.0 2.0 2.0 2.0 2.0 2.0 2.0
Lomefloxacin LMF 351.4 16.0 2.0 2.0 4.0 4.0 4.0 4.0 2.0 4.0 4.0 4.0 4.0
Ciprofloxacin CIP 331.3 4.0 1.0 1.0 1.0 1.0 2.0 2.0 1.0 1.0 1.0 1.0 1.0
Chloramphenicol CLP 323.1 8.0 1.0 1.0 2.0 2.0 4.0 4.0 1.0 1.0 1.0 1.0 4.0
Benzalkonium BZK 283.9 8.0 2.0 2.0 2.0 2.0 2.0 2.0 4.0 4.0 4.0 4.0 8.0
Acriflavine ACR 259.7 4.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0
Nalidixic acid NLD 232.2 4.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0
Carbonyl cyanide 3‐chlorophenylhydrazone CCP 179.6 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0
Phosphomycin PPM 138.1 2.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0
R=
128 64 32 16 8 4 2 1
Open in a new tab

2.4. Prediction and analysis of antibiotics binding to AcrB and mutants

We performed structure prediction calculations for the binding of the antibiotics to AcrB and mutants to the proximal (i.e., tunnel‐2) and distal pockets, and the tunnel‐1 and tunnel‐3 sites, using molecular docking with their access and binding protomers (Figure 2a). All the antibiotics were docked with free rotation at all their rotatable bonds for best estimation of their binding mode within each binding site (Figures 2b, 3a, S5, and S6). The estimated binding free energies for the antibiotics in either proximal pocket or distal pocket are reported in Tables S5 and S6, and their energies difference are shown in Table S7. Clearly, antibiotics binding to the distal pocket is favored than the proximal pocket, in almost all cases, with the exceptions of CLD and SPR with relatively small differences, for 4 and 1 mutants, respectively (Table S7). Similarly, the binding to the distal pocket in AcrB is favored than either tunnel‐1 or tunnel‐3 for all antibiotics (Table S8). However, while comparison between the proximal pocket and tunnel‐1 or ‐3 shows most of antibiotics slightly favored the proximal pocket than the tunnels, ACR and NLD favors the tunnels instead. Comparison between the tunnels clearly shows that most of the antibiotics slightly favored tunnel‐1 over tunnel‐3 (Table S8). Interestingly, while our docking for the distal pocket included all the area between the domains PC1 and PC2 (Figure 2a), all the antibiotics revealed an overall well localized binding site in this pocket as represented by a small antibiotic NLD (29 atoms), except for the smallest one PPM (15 atoms), as shown in Figure 3b (see also Figures S7 and S8). This exception maybe due to the high number of hydrogen bonds occurred for PPM binding site when compared to most of other antibiotics (Table S10). Importantly, the estimated binding free energy for all the antibiotics binding to proximal pocket, distal pocket, tunnel‐1, and tunnel‐3 are favorable (Tables S5, S6, and S8).

FIGURE 2.

FIGURE 2

Open in a new tab

AcrB in complex with antibiotics in tunnel‐2. (a) AcrB monomer depicting the localization and the pathway of each tunnel. Tunnel‐1 (green), tunnel‐2 (magenta), and tunnel‐3 (cyan). The gate loop (GL) is colored in magenta, as part of tunnel‐2. (b) Antibiotics binding to the proximal and distal pockets, in AcrB and its representative gate loop mutants (AAA, ΔLoop), as predicted by molecular docking. For better visual inspection, the antibiotics were clustered into four groups based on their molecular weight. G1 for SPR, ERY, NOV, DOX, and FSA (842–516 g/mol). G2 for MIY, TRC, R6G, CLD, and MXF (458–401 g/mol). G3 for OFL, LMF, CIP, and CLP (362–323 g/mol). G4 for BZK, ACR, NLD, and PPM (284–138 g/mol). Mainly all the antibiotics populate similar binding region in either binding pocket, despite their size difference or the deletion of the gate loop.

FIGURE 3.

FIGURE 3

Open in a new tab

AcrB in complex with antibiotics. (a) Antibiotics binding to the tunnel‐1 and ‐3 in AcrB, as predicted by molecular docking. For better visual inspection, the antibiotics were clustered into four groups based on their molecular weight. G1 for SPR, ERY, NOV, DOX, and FSA (842–516 g/mol). G2 for MIY, TRC, R6G, CLD, and MXF (458–401 g/mol). G3 for OFL, LMF, CIP, and CLP (362–323 g/mol). G4 for BZK, ACR, NLD, and PPM (284–138 g/mol). (b) All nine binding modes of the antibiotic NLD and PPM within the large binding site of the distal pocket, in AcrB and its representative gate loop mutants (AAA, ΔLoop), as predicted by molecular docking.

To assess further the difference in the binding of these antibiotics, we inspected the number of hydrogen bonds and atomic contacts between the antibiotics and AcrB and mutants (Tables 2, 3, 4 and S9–S13). Although the distal pocket was energetically favored than the proximal pocket, the number of antibiotics with higher number of hydrogen bonds in the proximal pocket was greater than those in the distal pocket in AcrB and mutants with a ratio between 2 and 5 (Table 2). The same applies to the tunnel‐1 and ‐3 in AcrB (Table 4(a)), while energetically the distal pocket is favored than either of the tunnels (Table S8). Note that for few antibiotics the number of hydrogen bonds was equal in the proximal and distal pockets. However, close inspection of the number of atomic contacts reveals, in AcrB and most of the mutants, higher number of antibiotics with higher number of atomic contacts in distal pocket than in the proximal pocket (Table 3). Similar trend is observed for tunnel‐1 and ‐3, as most of antibiotics have higher number of contacts in distal pocket (Table 4(b)).

TABLE 2.

Number of hydrogen bonds difference between the distal and proximal pockets.

Dist‐Prox AcrB F615A F617A RA RK RD RE AA F615A‐RA F617A‐RA AAA ΔLoop
ACR 0 1 0 0 0 0 0 0 1 0 2 −1
BZK 0 0 0 0 0 0 0 0 0 0 0 0
CIP −2 −2 −2 −3 −2 −1 −2 0 −3 −2 0 −3
CLD −2 −1 −3 4 −1 −4 −1 −1 0 −1 −2 −1
CLP −1 −2 2 −3 −2 −4 −2 1 −2 −4 −2 −1
DOX −2 3 −4 0 −3 −2 −2 1 −3 −2 −1 −1
ERY 3 −1 2 5 2 5 2 −1 0 2 −1 −2
FSA 1 0 1 1 1 0 −1 0 −1 0 −1 0
LMF 0 0 2 0 0 0 0 3 0 2 0 −1
MIY −3 3 −1 −4 −4 −3 −1 −2 2 −2 −2 2
MXF −2 −5 −2 −1 −3 −2 −2 −4 −2 −1 −3 −2
NLD −4 −3 −3 −4 −4 −4 −4 −2 −3 −4 −2 −2
NOV 1 1 3 −2 −3 0 2 3 −1 1 4 2
OFL −1 −1 −2 −2 −2 −1 −1 −3 −1 −2 −1 −2
PPM 1 1 0 1 0 0 3 −1 2 1 2 4
R6G −1 −1 −3 1 0 1 1 −1 −1 −1 −3 −1
SPR −3 −5 −3 0 0 −1 −3 −2 −4 0 −3 −2
TRC −1 0 0 0 0 −1 0 6 −1 0 6 −2
Distal # a 4 5 5 4 2 2 4 5 3 4 4 3
None # b 3 4 4 6 7 6 4 4 4 5 3 2
Proximal # c 11 9 9 8 9 10 10 9 11 9 11 13
Ratio d 2.8 1.8 1.8 2.0 4.5 5.0 2.5 1.8 3.7 2.3 2.8 4.3
Open in a new tab
a

Number of antibiotics with more hydrogen bonds in distal pocket.

b

Number of antibiotics with equal hydrogen bonds in distal and proximal pockets.

c

Number of antibiotics with more hydrogen bonds in proximal pocket.

d

Ratio of hydrogen bonds number in proximal to distal pockets.

TABLE 3.

Number of atomic contact difference between the distal and proximal pockets.

Dist‐Prox AcrB F615A F617A RA RK RD RE AA F615A‐RA F617A‐RA AAA ΔLoop
ACR −6 −11 11 2 6 2 8 0 −11 −5 −3 3
BZK 0 11 17 5 3 −5 5 −4 −9 14 −2 −1
CIP −5 −2 −8 −2 −4 −6 −5 −9 5 0 −7 −17
CLD 11 −12 0 14 −7 −3 −11 −12 −8 −4 −17 9
CLP 10 −6 −9 −1 −4 11 6 −4 −5 −2 −9 13
DOX 37 4 13 32 36 35 38 17 17 25 3 −22
ERY −1 −1 14 15 24 −5 −1 −10 −4 9 −5 −12
FSA 29 3 33 23 24 22 26 −3 8 31 −4 3
LMF 5 −8 10 1 3 5 4 16 −8 2 0 1
MIY 9 8 8 8 6 15 7 3 7 10 1 −2
MXF −2 −3 26 13 −3 −3 0 20 −5 22 21 14
NLD 11 −7 16 12 9 9 10 10 −6 15 −6 9
NOV −13 −16 10 −14 32 25 6 −6 5 −6 −1 −15
OFL 1 −8 1 3 3 3 1 −10 −9 1 −12 14
PPM 15 12 10 16 14 8 3 20 12 16 15 8
R6G 15 −2 5 9 4 9 12 −5 8 1 −14 −8
SPR −7 22 −2 −4 4 −18 −6 −16 −4 2 −9 1
TRC 33 14 16 26 30 −12 32 11 13 14 10 11
Distal # a 11 7 14 14 14 11 13 7 8 13 5 11
None # b 1 0 1 0 0 0 1 1 0 1 1 0
Proximal # c 6 11 3 4 4 7 4 10 10 4 12 7
Ratio d 0.5 1.6 0.2 0.3 0.3 0.6 0.3 1.4 1.3 0.3 2.4 0.6
Open in a new tab
a

Number of antibiotics with more atomic contacts in distal pocket.

b

Number of antibiotics with equal atomic contacts in distal and proximal pockets.

c

Number of antibiotics with more atomic contacts in proximal pocket.

d

Ratio of atomic contacts number in proximal to distal pockets.

TABLE 4.

(a) Number of hydrogen bonds found for the antibiotics within the tunnel‐1 (T1) and tunnel‐3 (T3) and their differences with those in distal pocket. (b) Number of atomic contact difference between the distal pocket and tunnel‐1 (T1) or tunnel‐3 (T3).

(a) HB T1 T3 Dist‐T1 Dist‐T3 T1‐T3 (b) Contacts Dist‐T1 Dist‐T3
ACR 0 1 0 −1 −1 ACR 0 −7
BZK 0 0 0 0 0 BZK 0 −5
CIP 1 2 −1 −2 −1 CIP −12 −11
CLD 4 2 −1 1 2 CLD 14 31
CLP 4 3 −3 −2 1 CLP 23 11
DOX 5 3 −3 −1 2 DOX 25 10
ERY 3 2 2 3 1 ERY 23 6
FSA 3 1 −1 1 2 FSA 30 35
LMF 3 0 −3 0 3 LMF 11 21
MIY 4 4 −2 −2 0 MIY 3 5
MXF 0 1 2 1 −1 MXF 28 8
NLD 0 2 0 −2 −2 NLD 11 7
NOV 3 4 −1 −2 −1 NOV 8 −15
OFL 0 2 0 −2 −2 OFL 21 5
PPM 3 3 1 1 0 PPM 19 17
R6G 2 1 −2 −1 1 R6G −6 10
SPR 2 0 0 2 2 SPR 13 18
TRC 6 2 −5 −1 4 TRC 23 15
Distal/T1 # a 3 6 9 Distal # d 14 13
None # b 5 2 3 None # e 2 0
T1 and T3/T3 # c 10 10 6 T1/T3 # f 2 4
Open in a new tab
a

Number of antibiotics with more hydrogen bonds in distal pocket or T1.

b

Number of antibiotics with equal hydrogen bonds in distal and proximal pockets or in T1 and T3.

c

Number of antibiotics with more hydrogen bonds in T1 and T3 or T3.

d

Number of antibiotics with more atomic contacts in distal pocket.

e

Number of antibiotics with equal atomic contacts in distal and T1 or T3.

f

Number of antibiotics with more atomic contacts in T1 or T3.

2.5. Antibiotics' molecular characteristics versus MIC ratio analysis

To assess the presence of any correlation between the antibiotics' MIC ratio (R) for the mutants with their chemical and structural proprieties and ultimately with the antibiotics export route preferences, namely tunnel‐1, ‐2, or ‐3, we have performed PCA analysis using the antibiotics MIC ratio, molecular mass, geometry and chemical descriptors, and functional groups. To be as exhaustive as possible we have used 33 parameters to describe each antibiotic, excluding MIC ratio (Tables S14–S16). PCA analysis shows some putative clustering of some antibiotics, such as CLD, MXF, OFL, LMF, CIP, NLD, and ACR; however, close inspection reveals that their MIC ratio are different (R of 1, 2, or 4; Figures 4 and S9a–c). Furthermore, some antibiotics with R = 1 are far away from this cluster (ex: SPR or DOX). This shows that the MIC ratio may not have any direct or simple correlation with the antibiotic molecular characteristics, as identified by these 33 parameters. Nevertheless, we have searched for the minimum parameters that could reproduce similar PCA analysis outcomes. We found that 5 parameters reproduce similar clustering and/or trend of the PCA results using all 33 parameters, for F617A, RA, RK, RD, and RE, and with an overall pseudo‐rotation of 180 degrees of the original plot for the other mutants. Those parameters were the topological polar surface area (TPSA), LogP, the number of rotatable bonds (Rotatables), the number of hydrogen bond acceptors (HB Acceptor), and the number of hydrogen bond donors (HB Donor) (Figure 4).

FIGURE 4.

FIGURE 4

Open in a new tab

Principal component analysis (PCA) of representative mutants (F615A, AAA, ΔLoop), using 33 and 5 parameters (see details in section 2).

Although nearly eight antibiotics have their R equal to 1, our expectation using PCA analysis was to identify the overall common molecular characteristics that may explain such R value. The R value of 1 means that the gate loop has no effect on the antibiotic export and possibly the antibiotic prefers tunnel‐1 and/or tunnel‐3 as its preferred export tunnel(s) instead of tunnel‐2, where the gate loop is located. Since PCA analysis failed in that context, we have performed molecular similarity calculations of these antibiotics, as summarized in Table 5. Surprisingly, TRC and MIY are highly similar, with a value of 0.86, and yet R values of these antibiotics are 1 and 8 (or 16), respectively. Similarly, CIP has an R = 1 (excluding mutants RD and RE) and has high molecular similarity to MXF, LMF, and OFL, values between 0.77 and 0.89, but their R values are between 2 and 4. Clearly, molecular similarity for these antibiotics cannot explain the substantial difference in their R values.

TABLE 5.

Antibiotics similarities values.

SPR ERY NOV DOX FSA MIY TRC R6G CLD MXF LMF OFL CIP CLP BZK ACR NLD PPM
SPR 1.000 0.209 0.132 0.140 0.185 0.136 0.138 0.095 0.194 0.128 0.091 0.118 0.092 0.082 0.149 0.013 0.070 0.098
ERY 0.209 1.000 0.131 0.200 0.205 0.135 0.137 0.120 0.200 0.127 0.101 0.132 0.119 0.092 0.169 0.015 0.079 0.113
NOV 0.132 0.131 1.000 0.203 0.141 0.167 0.169 0.167 0.145 0.237 0.255 0.250 0.259 0.185 0.216 0.173 0.220 0.130
DOX 0.140 0.200 0.203 1.000 0.188 0.241 0.224 0.286 0.179 0.259 0.231 0.250 0.260 0.255 0.239 0.273 0.217 0.119
FSA 0.185 0.205 0.141 0.188 1.000 0.207 0.190 0.129 0.208 0.179 0.170 0.145 0.151 0.118 0.222 0.019 0.125 0.125
MIY 0.136 0.135 0.167 0.241 0.207 1.000 0.857 0.179 0.154 0.216 0.208 0.229 0.239 0.205 0.275 0.190 0.190 0.139
TRC 0.138 0.137 0.169 0.224 0.190 0.857 1.000 0.182 0.180 0.220 0.213 0.208 0.191 0.238 0.316 0.167 0.195 0.143
R6G 0.095 0.120 0.167 0.286 0.129 0.179 0.182 1.000 0.111 0.292 0.261 0.255 0.267 0.178 0.186 0.389 0.220 0.079
CLD 0.194 0.200 0.145 0.179 0.208 0.154 0.180 0.111 1.000 0.191 0.156 0.178 0.186 0.175 0.250 0.023 0.128 0.207
MXF 0.128 0.127 0.237 0.259 0.179 0.216 0.220 0.292 0.191 1.000 0.688 0.774 0.767 0.256 0.306 0.353 0.484 0.121
LMF 0.091 0.101 0.255 0.231 0.170 0.208 0.213 0.261 0.156 0.688 1.000 0.821 0.885 0.250 0.265 0.400 0.556 0.100
OFL 0.118 0.132 0.250 0.250 0.145 0.229 0.208 0.255 0.178 0.774 0.821 1.000 0.852 0.278 0.294 0.387 0.536 0.133
CIP 0.092 0.119 0.259 0.260 0.151 0.239 0.191 0.267 0.186 0.767 0.885 0.852 1.000 0.257 0.273 0.414 0.577 0.103
CLP 0.082 0.092 0.185 0.255 0.118 0.205 0.238 0.178 0.175 0.256 0.250 0.278 0.257 1.000 0.357 0.233 0.233 0.120
BZK 0.149 0.169 0.216 0.239 0.222 0.275 0.316 0.186 0.250 0.306 0.265 0.294 0.273 0.357 1.000 0.250 0.250 0.130
ACR 0.013 0.015 0.173 0.273 0.019 0.190 0.167 0.389 0.023 0.353 0.400 0.387 0.414 0.233 0.250 1.000 0.417 0.042
NLD 0.070 0.079 0.220 0.217 0.125 0.190 0.195 0.220 0.128 0.484 0.556 0.536 0.577 0.233 0.250 0.417 1.000 0.087
PPM 0.098 0.113 0.130 0.119 0.125 0.139 0.143 0.079 0.207 0.121 0.100 0.133 0.103 0.120 0.130 0.042 0.087 1.000
Open in a new tab

Note: The colors reflect the similarity extent (dark red for the highest and dark green for the lowest).

3. DISCUSSION

Over the past two decades, a substantial amount of structural and functional information on AcrB export activity has been reported. This has led to widely accepted insights and hypotheses regarding the resting state and active state structures (Murakami et al. 2002; Murakami et al. 2006; Nikaido 2011; Seeger et al. 2006), the functional rotation mechanism of AcrB (Eicher et al. 2012; Murakami et al. 2006; Nakashima et al. 2011), and the three possible export tunnels for substrates, known as channel or tunnel‐1, ‐2, and ‐3 (Nakashima et al. 2011; Yamaguchi et al. 2015; Zwama et al. 2018). However, the driving forces behind its substrates' specificity and export tunnels preferences remain to be fully understood. Consequently, in this work, we used several gate‐loop mutants of AcrB to further investigate their effect on AcrB's structure and drug susceptibility, as well as to elucidate the export tunnel preferences of several antibiotics.

Our recent crystal structures of AcrB and mutants AAA and ΔLoop, and the present mutants RA, RK, and AA, clearly show that the overall structures of the gate‐loop mutants are largely similar to the wild type AcrB (Ababou 2018; Ababou and Koronakis 2016). We anticipate that this similarity extends to all other mutants, as they involve either single or double mutations and exhibit a comparable fold and stability, as confirmed by their CD spectra and thermal unfolding data (Figures S2 and 1f and Table S3). However, drug susceptibility assays for several antibiotics revealed varying effects between AcrB and these mutants, ranging from significant differences to no effect, as indicated by their MIC values (Tables 1 and S4).

To address the lack of complex structures of AcrB with its substrates, we used molecular modeling to investigate the molecular interactions between AcrB, and its mutants, and its substrates used in this work. Binding of all antibiotics to proximal pocket, distal pocket, tunnel‐1, and tunnel‐3 were energetically favorable (Tables S5–S8). Almost all the substrates were energetically favored in the distal pocket compared to the proximal and the other tunnels (Tables S7 and S8). This outcome is expected, as substrates must pass through the distal pocket to be extruded, despite differences in their molecular and chemical properties. Notably, while this trend correlates with the higher number of contacts in the distal pocket compared to the proximal pocket and other tunnels (Tables 3, 4, and S11–S13), the opposite was observed when comparing the number of hydrogen bonds (Tables 2, 4, S9, and S10). This suggests that hydrogen bond formation is a key driving force in directing the substrates toward specific tunnels; however, for the substrates like BZK, which lack the ability to form hydrogen bonds, other driving forces are likely at play.

Assuming the MIC ratio, R, of 1 means that the substrate export does not involve tunnel‐2, where the gate loop is localized, then the substrate must have preference to tunnel‐1 or ‐3. Interestingly, based solely on the number of hydrogen bonds in tunnel‐1 and ‐3 compared to proximal pocket, a clear preference trend emerges for such substrates. For example, DOX and TRC exhibit a higher hydrogen bonds number in tunnel‐1 and ‐3 compared to the proximal pocket (Tables 4 and S9), with higher number in tunnel‐1 compared to tunnel‐3. This suggests that these substrates may favor tunnel‐1 over tunnel‐3, as previously proposed for DOX (Ababou 2018). However, this trend does not hold for CLD, ACR, NLD, PPM, SPR (excluding R = 2 with RE and ΔLoop), and CIP (excluding R = 2 with RD and RE), as these antibiotics have fewer or similar number of hydrogen bonds in tunnel‐1 and/or tunnel‐3 compared to the proximal pocket. One possible explanation for this could be related to their size or molecular weight, as previously suggested in studies on the involvement of the gate loop in AcrB (Cha et al. 2014; Nakashima et al. 2011). Specifically, ACR (32 atoms), NLD (29 atoms), CLD (60 atoms), PPM (15 atoms), and CIP (42 atoms) might still prefer tunnel‐2 without the involvement of the gate loop, or they may have no particular preference among the tunnels. However, this explanation does not apply to SPR (135 atoms), the largest antibiotic used in this study. Therefore, both explanations have limitations.

To better analyze the relationship between R value and these substrates' molecular and chemical proprieties, along with their functional groups, we conducted a PCA analysis (Figure 4). Surprisingly, no clear clustering of substrates with R value of 1 was observed. Interestingly, the PCA analysis using five parameters (TPSA, LogP, Rotatables, HB Acceptor, and HB Donnor), revealed that hydrophobicity, polar surface, flexibility, and hydrogen bonds formation are the most significant molecular and chemical proprieties for these substrates (Figures 4 and S9a–c and Tables S14 and S15). To further investigate and identify any common molecular or chemical properties among these substrates, we performed molecular similarity calculations (Table 5). Using a similarity cutoff of 0.4 and above, we found ACR, NLD, and CIP have molecular similarities values between 0.41 and 0.58. This may provide a promising explanation for why these substrates share an R of 1. However, DOX, CLD, TRC, and SPR have low molecular similarities values (0.14–0.19), indicating that substrates with very low molecular similarity may still have similar export tunnel preferences. On the other hand, NLD shows substantial molecular similarity with MXF, LMF, and OFL (~0.5–0.6), yet the latter have R ≥ 2, indicating a preference for tunnel‐2. Interestingly, CIP has R = 1 (ignoring R of 2 with RD and RE) with high molecular similarity to MXF, LMF, OFL (~0.8–0.9), yet the latter have R ≥ 2. Strikingly, MIC values of MIY and TRC for AcrB were 3.7 and 1.8 μg/mL, respectively (Table S4)—a small difference—but despite their high molecular similarity (~0.86), MIY has R ≥ 8, indicating that MIY favors tunnel‐2.

It is important to note that, in principle, substrates with high molecular similarities are likely to have similar preferences for their export tunnels, as reported by others (Eicher et al. 2012; Husain et al. 2011; Husain and Nikaido 2010; Murakami et al. 2006; Nakashima et al. 2011; Oswald et al. 2016; Tam et al. 2020; Zwama et al. 2018). However, our data contradict, somehow, this hypothesis, as observed for TRC and MIY, for example, when their only molecular difference is the exchange of a H atom, ‐OH group, and ‐CH3 group in TRC by ‐N(CH3)2 group, H atom, and H atom in MIY, respectively (Figure S10). This limited molecular change did not affect dramatically either the molecular or the chemical proprieties of these two substrates (Tables S14–S16). Therefore, while a change in tunnel preference for TRC is plausible—given the higher number of hydrogen bonds in tunnel‐1 and ‐3 compared to the proximal pocket—it is difficult to dismiss the possibility that the MIC values may not accurately reflect TRC's true export tunnel preference. For clarity, we cannot eliminate the possibility of TRC to be exported through tunnel‐2 like MIY, passing by the gate loop, and yet no detectable effect of the gate loop mutations on the MIC values of TRC.

Recently, it was reported that BZK, R6G, and ACR have preference to tunnel‐3. The rational for this was the planar aromatic cation nature of these substrates (Zwama et al. 2018). While this may be in line with our inferences about ACR, our results for BZK and R6G contradict that, since these substrates have R values of 2–8 and 2–16, respectively (Table 2). Note that for R6G other reports agree with our results (Cha et al. 2014; Muller et al. 2017). In addition, FSA has R values of 1 for all mutants except AA, AAA, and ΔLoop with values of 8 (Table 1). Clearly, this suggests that FSA export tunnel is not unique but rather can be either tunnel‐1, based on number of hydrogen bonds compared to proximal pocket (Tables 4 and S9), or tunnel‐2. Recently, it was reported that FSA prefers an export tunnel at the groove between transmembrane helices TM1 and TM2 (Tam et al. 2020). However, recent experimental data, from the same group, clearly suggest that FSA can also use tunnel‐1 (Tam et al. 2021), which is in line with our present results.

Mutations in AcrB located in different tunnels or putative grooves can indeed help highlight the preferred export tunnels for various substrates, but this should not be considered a rule for determining their preferences. In fact, it is challenging to reconcile conflicting data that show substrates entering different tunnels, especially when they are the same (e.g., FSA, DOX) (Ababou 2018; Eicher et al. 2012; Tam et al. 2020; Tam et al. 2021) or highly similar to each other, as in our case of TRC and MIY. There are two main plausible explanations for this discrepancy.

The first explanation suggests that AcrB substrates may be promiscuous in their choice of export tunnel, based on accessibility and/or opportunity, and possibly in concentration dependent manner. Indeed, several studies have shown that substrates like BZK, ERY, DOX, R6G, FSA, and NOV can enter different tunnels (Eicher et al. 2012; Nakashima et al. 2011; Tam et al. 2021; Zwama et al. 2018). Furthermore, our computational analysis of AcrB–substrate interactions revealed that all these substrates were energetically favorable in all three tunnels and were more energetically favored in the distal pocket as compared to all other tunnels entrances (Tables S5–S8). Except for BZK, the antibiotics generally formed hydrogen bonds in all tunnels, suggesting that these substrates can be directed to any of AcrB's tunnels and become transiently stabilized prior to further movement toward the distal pocket, where the number of hydrogen bonds decreases (Tables 4, S9, and S10).

The second explanation considers the implication of the molecular dynamics AcrB, because for any substrate export, there are large conformational changes between the AcrB's protomers (Access/Loose (L), Tight/Binding (T), and Extrusion/Open (O)). Strong evidence for this will be the number of mutations reported to affect substantially the MIC values of several substrates, when structurally the mutations and the substrates have no contact. We reported the X‐ray structures of the resting and active states of AAA and ΔLoop mutants to be identical to AcrB, but their MIC values were largely different for ERY. We demonstrated that the main difference between these mutants and AcrB, leading to different export activity of ERY, was in their molecular dynamics behavior, as shown by our detailed MD simulation investigation (Ababou 2018; Ababou and Koronakis 2016). Mutation E673G was reported to increase ERY resistance 2‐fold, while structurally no contact is detected between E673 and ERY. While Q569R mutation has no effect on the resistance of ERY, as Q569 is located at the entrance to the proximal pocket (shortest contact is ~10.0 Å), the double mutant Q569R‐E673G also showed no change in ERY resistance, which canceled the effect of E673G (Kobayashi et al. 2014). Mutations V127A and Y49S increased ERY susceptibility by 2–4‐fold, when the shortest distance between V127 or Y49 and ERY is ~16.3 Å for both residues, and also D153E, which is located outside the export tunnel‐2 (shortest contact is ~24.0 Å), increased ERY susceptibility by 2–4‐fold (Soparkar et al. 2015). AcrB quadruple mutant A33W‐T37W‐A100W‐N298W, part of tunnel‐3, resulted in 4‐fold increase susceptibility for ERY, which prefers most exclusively tunnel‐2 (Zwama et al. 2018). In addition, recently R717Q and R717L mutations reported to decrease macrolide, and fluoroquinolone, susceptibility but increased the susceptibility for Novobiocin and Cloxacillin (Zwama and Nishino 2022). Several mutations were reported to improve AcrB efflux rate of biofuels (Foo and Leong 2013). Mutation Q737L significantly increased the export rate of α‐pinene by 400% and for n‐octane by 51%, while Q737 is in the docking domain at ~47 Å from F617 in the gate loop (Cβ–Cβ distance). Furthermore, mutation M987T deceased the export rate of α‐pinene and n‐octane by ~25%, while M987 is in the membrane domain at ~41 Å from F617 (Foo and Leong 2013). If the mutations described above, without direct contact with the substrate, can lead to an increase or decrease in drug susceptibility and/or efflux rate, then the most plausible explanation is a change in the molecular dynamics of AcrB. Thus, these mutations likely affect the conformational change kinetics of AcrB's protomers (slow, intermediate, or fast), which in some cases can be reflected in MIC values, as we have previously reported (Ababou 2018). Indeed, we have suggested that AcrB mutations, AAA and ΔLoop, may have affected the conformational change kinetics to explain the discrepancy between our structural and functional data, as inferred from the difference in molecular dynamics between AcrB and its mutants (Ababou 2018). This means, the interpretation of MIC values can be correlated to the time required to reach and populate each conformation adopted by the protomers (i.e., AcrB/fast, AAA/medium, ΔLoop/slow). For clarity, this is not about export rate/velocity but the rate of conformational change between L, T, and O conformations; however, the overall speed at which AcrB exports its substrates can be in part proportional to its conformational change kinetics, but undoubtedly other molecular and environmental factors can affect substantially the export velocity beyond AcrB's conformational change kinetics.

All reported AcrB structures determination, using X‐ray crystallography, present only two conformational states, namely the symmetric trimer, known as the resting state with all protomers adopting the same conformation (LLL) (Das et al. 2007; Murakami et al. 2002; Veesler et al. 2008), and the asymmetric trimer, known as the active state with each protomer in different conformation (LTO) (Murakami et al. 2006; Nakashima et al. 2011; Seeger et al. 2006). Also, it was not easy to observe experimentally any intermediate structural forms of any other RND efflux transporter beyond their LLL and LTO conformational states. However, recently several Cryo‐EM structures of RND efflux transporters reported intermediate conformational states of their trimer complex. CmeB from Campylobacter jejuni has been reported in two forms in which all the protomers adopt O conformation (OOO) and another form with one protomer adopted an L like conformation (L*), called resting conformation (L*OO) (Su et al. 2017). AdeB from Acinetobacter baumannii has been reported in three forms, in substrate free state, consisting of form I OOO with the protomers in close contact in inner membrane part, form II OOO with one protomer is moved away from the rest in the inner membrane part, and form III OOO with all protomers moved away from each other in the inner membrane part. In the presence of ethidium (Et), AdeB was reported to adopt three forms consisting of form Et‐I OOT. form Et‐II in L*TO, and form Et‐III LTO (Morgan et al. 2021). Recently, the structure of AcrD from E. coli, in the absence and presence of its substrate, gentamicin, revealed a trimer structure adopting the conformational state L*TO (Zhang et al. 2023). In addition, the heavy‐metal RND efflux transporter CusA was reported to adopt different conformational states of its trimer form, namely, OOT, OTT, OOO, and TTT, suggesting the protomers function independently (Moseng et al. 2021). These findings, by Cryo‐EM technique, clearly suggest the possibility for RND efflux transporters to adopt different intermediate conformational states, with the possibility that the protomers to function independently of each other, during their export activity. Importantly, these cryo‐EM structures revealed that conformational change kinetics is a crucial factor in these intermediate forms, and ultimately in the export activity of these RND efflux transporters. Furthermore, recently experimental evidence on the change in molecular dynamics of AcrB has been reported to modulate the binding and efflux of substates in the presence of an inhibitor, which by analogy can be considered as a mutation that increases the antibiotic susceptibility of AcrB (Reading et al. 2020). Consequently, our conformational change kinetics explanation for AcrB's gate loop mutations effect on its export activity, may be strengthened further because what we did predicted to be possible for AcrB, has been corroborated experimentally, and importantly it can be observed for other RND efflux transporters, as shown above.

As discussed earlier, two primary explanations are suggested: the accessibility and opportunity for the substrate to use any export tunnel, and the conformational change kinetics of AcrB's protomers. TRC and MIY have comparable MIC values for AcrB and have high molecular similarity, yet their R values differ significantly with 1 and 8 (or 16), respectively. If both substrates were expected to use the same tunnel‐2, this discrepancy in R values for TRC suggests that TRC might prefer tunnel‐1 and/or tunnel‐3. While conformational change kinetics is most favorable explanation for this striking R difference between MIY and TRC, based on recent evidence and inferences (Ababou 2018; Morgan et al. 2021; Moseng et al. 2021; Reading et al. 2020; Su et al. 2017), it is crucial to find an explanation that integrates both conformational change kinetics and substrate effects and form a general explanation instead. Thus, considering the case of MIY and TRC, as well as and the efflux rate changes for biofuels (Foo and Leong 2013), it is plausible that the conformational change kinetics may be substrate‐dependent. This suggests a more general explanation where the substrate, and possibly its chosen export tunnel, can influence the conformational change kinetics, as illustrated in Figure 5. This generalized explanation is further supported by our PCA analysis, based on those five minimum parameters of TPSA, LogP, Rotatables, HB Acceptor, and HB Donor (Figures 4 and S9a–c). The substrate's flexibility, hydrogen bond formation, and hydrophobicity will likely impact the molecular dynamics of AcrB's protomers, thereby affecting their conformational change kinetics.

FIGURE 5.

FIGURE 5

Open in a new tab

Depiction of AcrB mutations effect on its conformational change kinetics between resting state (Access/Loose (L)—LLL trimer) and active state (Access/Loose (L), Binding/Tight (T), Extrusion/Open (O)—LTO timer). (a) Substrate independent conformational change kinetics pathway, where the trimer's unknown conformational state is denoted as XYZ. (b) Substrate‐dependent conformational change kinetics pathways, where the trimer's unknown conformational state is as denoted as (XYZ)S, and S denotes the substrate.

4. CONCLUSION

Our work demonstrates that investigating AcrB's gate loop mutants has provided crucial insights into AcrB export activity at the molecular level. Molecular modeling and computational analysis of interactions between antibiotics and AcrB revealed that all substrates exhibit favorable binding free energies across all three tunnels. With few exceptions, hydrogen bonds appear to be the primary driving force directing substrates toward any of the three tunnels, supporting the idea of substrate promiscuity. Additionally, we found that, energetically, the distal pocket is more favorable than any of the tunnels, despite generally having fewer hydrogen bonds. This suggests a sequential mechanism for the export of most AcrB substrates: (1) the substrate is primarily captured and directed by hydrogen bond formation and is transiently stabilized; (2) as the protomer changes conformation from access/loose to binding/tight, the substrate is pushed through the tunnel into the distal pocket, where it forms more atomic contacts (highly favorable energetically) but fewer or no hydrogen bonds, allowing loose binding within this pocket to facilitate substrate movement toward extrusion; and (3) when the protomer changes conformation from binding/tight to extrusion/open, the substrate is easily pushed from the distal pocket into the extrusion channel.

Our MIC R values highlight the substrates' preferences for one of the three tunnels. Interestingly, while R values of 1 would typically indicate that an antibiotic may prefer tunnels other than tunnel‐2 (which contains the gate loop mutations), this simple explanation was challenged by MIY and TRC. These two substrates have high molecular similarity and comparable MIC values for AcrB. To explain these experimental observations, we propose two possible explanations: (1) substrate promiscuity in choosing their export tunnel, likely based on accessibility and/or opportunity, possibly in a concentration‐dependent manner; and (2) generalized conformational change kinetics, where the kinetics of protomer conformational changes are suggested to be substrate‐dependent. This explanation is supported by our MIY and TRC data, PCA analysis, and is corroborated to some extent by recent reports on the cryo‐EM structures of intermediates in a few RND efflux transporters, as well as experimental evidence on the impact of molecular dynamics on AcrB's export activity.

5. MATERIALS AND METHODS

5.1. Expression and purification of AcrB mutants

All AcrB's gate loop mutants were prepared, expressed, and purified as previously reported (Ababou 2018). In addition to our previous mutants, AAA, and ΔLoop, in this work we have prepared single and double mutation of the bulky residues in the gate loop, namely F615, F617, and R620. The mutants are F615A, F617A, RA, RK, RD, RE, AA, F615A‐R620A (F615A‐RA), and F617A‐R620A (F617A‐RA).

5.2. Structural stability of AcrB mutants

CD spectra and thermal unfolding data were collected as previously reported (Ababou 2018). The protein concentration was held constant at 5–10 μM. All CD measurements were corrected by subtracting the buffer spectra. The unfolding experiments were fitted as previously reported (Ababou 2018).

5.3. Crystallization, data collection, and structure refinement of AcrB mutants

All AcrB mutants were crystallized as previously reported (Ababou and Koronakis 2016). X‐ray diffraction data were collected at Diamond Light Source (Didcot, UK). X‐ray data sets were indexed and integrated using iMosflm (Battye et al. 2011) and scaled using Scala or Aimless in the CCP4 suite (Winn et al. 2011). The structures were solved by molecular replacement using Phaser (McCoy et al. 2007) or Molrep (Vagin and Teplyakov 2010). Structures were solved using PDB file 3D9B (Veesler et al. 2008) containing residues 1–1033. All Structures refinement were performed as previously reported (Ababou and Koronakis 2016). The structures were completed with iterative rounds of manual model‐building with Coot (Emsley et al. 2010) and refinement in Phenix (Adams et al. 2010). Figures were prepared using PyMol (www.pymol.org). The atomic coordinates and structure factors of AA, RA, and RK structures in space group P21 have been deposited to the Protein Data Bank, and released with the accession code 7O3L, 7O3M, and 7O3N, respectively.

5.4. Antibiotic susceptibility assay

Functional data were performed using E. coli strain JW0451‐2 (ΔacrB) cells complemented with plasmid encoding the AcrB gate loop mutants as previously reported (Ababou 2018). Cells adjusted OD at 600 nm of ~0.05 were transferred to 100 μL 2TY medium containing carbenicillin, for the drug susceptibility test, using serial dilution in 96‐well plate for a starting concentration of SPR at 1686.1 μg/mL, RIF at 49.4 μg/mL, ERY at 1467.9 μg/mL, NOV at 1269.2 μg/mL, DOX at 870.0 μg/mL, FSA at 129.2 μg/mL, MIY at 29.6 μg/mL, TRC at 28.9 μg/mL, R6G at 958.0 μg/mL, CLD at 461.4 μg/mL, MXF at 26.3 μg/mL, LMF at 23.3 μg/mL, OFL at 10.8 μg/mL, CIP at 19.9 μg/mL, CLP at 19.4 μg/mL, BZK at 71.0 μg/mL, ACR at 541.9 μg/mL, NLD at 63.6 μg/mL, CCP at 409.2 μg/mL, PPM at 45.5 μg/mL. The bacterial growth was measured using 96‐well flat‐bottomed plates (Nunc, Thermo Scientific), and a Synergy HTX Multi‐Mode microplate reader (BioTek Instruments). The experiments were performed in triplicate, except for RIF in nine duplicates, and CCP and PPM in five duplicates.

5.5. Molecular docking and structural analysis

The unsolved AcrB mutant structures (F615A, F617A, RD, RE, F615A‐RA, F617A‐RA) were modeled by mutating the residue(s) of interest in each monomer of trimer‐1 of AcrB structure (Ababou 2018). The modeled structures and the crystal structures of AcrB and its mutants were energy minimized prior to the molecular docking with the antibiotics, using Amber14 package (Case et al. 2014). Antibiotic structures were optimized, and their CM2 partial charges were calculated using the semi‐empirical PM3 Hamiltonian as implemented in DivCon program (Dixon and Merz Jr. 1997; Gogonea and Merz Jr. 1999). The molecular docking calculations were performed using AutoDock Vina (Trott and Olson 2010). AcrB and its mutants and the antibiotics were prepared for docking using AutoDock Tools (Morris et al. 2009), where the calculated partial charges of the antibiotics were used instead of those assigned by AutoDock Tools. The molecular docking calculations were performed for the following docking sites: proximal pocket in access/loose monomer, distal pocket in the tight/binding monomer, binding pockets in tunnel‐1 and tunnel‐3 in access/loose monomer. To avoid any potential bias in our docking calculations, the 4docking search space was set to 22 Å × 22 Å × 22 Å which is larger than all the actual docking sites. For extensive sampling of docking positions, we set the exhaustiveness parameter to 80.

All calculated docking structures were analyzed for close contacts and hydrogen bonds using in‐house scripts and codes in Fortran 90. Only the best docking position for each antibiotic was considered based on its estimated binding energy. To consider the subtle difference in the structures and the molecular dynamics of the docking sites, which may lead to the presence or absence of a hydrogen bond we have changed slightly the cut‐off distance for hydrogen bonds from 2.40 to 2.55 Å. Also, the atomic contacts were determined using a distance cut‐off of 3.80 Å, excluding hydrogen atoms.

5.6. Antibiotics molecular, similarity, clustering, and PCA analysis

The chemical and geometrical descriptors for the antibiotics were produced using ChemAxon (www.chemaxon.com), and the cheminformatics toolkit ChemmineR package in R program (Cao et al. 2008). Antibiotics structural similarity search and clustering were performed using LiSiCA software (Lesnik et al. 2015). Principal component analysis (PCA) was performed using R program.

AUTHOR CONTRIBUTIONS

Farrukh Makhamadjamonov: Investigation; validation. Michal Emil Karolak: Investigation; validation. Lesley Smyth: Resources; writing – review and editing. Abdessamad Ababou: Conceptualization; investigation; methodology; validation; visualization; writing – original draft; writing – review and editing; software.

Supporting information

Data S1. Supporting Information.

PRO-34-e5252-s002.pdf (34.5KB, pdf)

Data S2. Supporting Information.

PRO-34-e5252-s001.pdf (2.6MB, pdf)

ACKNOWLEDGMENTS

We thank the beamline staff at the Diamond Light Source (Didcot, United Kingdom). We thank King's College London and University of East London for supporting this work.

Makhamadjamonov F, Karolak ME, Smyth L, Ababou A. Insights into substrate recognition and export tunnel preferences in the efflux transporter AcrB . Protein Science. 2025;34(1):e5252. 10.1002/pro.5252

Review Editor: Aitziber L. Cortajarena

REFERENCES

  1. Ababou A. New insights into the structural and functional involvement of the gate loop in AcrB export activity. Biochim Biophys Acta Proteins Proteomics. 2018;1866(2):242–253. [DOI] [PubMed] [Google Scholar]
  2. Ababou A, Koronakis V. Structures of gate loop variants of the AcrB drug efflux pump bound by erythromycin substrate. PLoS One. 2016;11(7):e0159154. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Adams PD, Afonine PV, Bunkoczi G, Chen VB, Davis IW, Echols N, et al. PHENIX: a comprehensive Python‐based system for macromolecular structure solution. Acta Crystallogr D Biol Crystallogr. 2010;66(Pt 2):213–221. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Battye TG, Kontogiannis L, Johnson O, Powell HR, Leslie AG. iMOSFLM: a new graphical interface for diffraction‐image processing with MOSFLM. Acta Crystallogr D Biol Crystallogr. 2011;67(Pt 4):271–281. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Bohnert JA, Schuster S, Seeger MA, Fahnrich E, Pos KM, Kern WV. Site‐directed mutagenesis reveals putative substrate binding residues in the Escherichia coli RND efflux pump AcrB. J Bacteriol. 2008;190(24):8225–8229. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Cao Y, Charisi A, Cheng LC, Jiang T, Girke T. ChemmineR: a compound mining framework for R. Bioinformatics. 2008;24(15):1733–1734. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Case DA, Babin V, Berryman JT, Betz RM, Cai Q, Cerutti DS, et al. AMBER 14. San Francisco, CA: University of California; 2014. [Google Scholar]
  8. Cha HJ, Muller RT, Pos KM. Switch‐loop flexibility affects transport of large drugs by the promiscuous AcrB multidrug efflux transporter. Antimicrob Agents Chemother. 2014;58(8):4767–4772. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Das D, Xu QS, Lee JY, Ankoudinova I, Huang C, Lou Y, et al. Crystal structure of the multidrug efflux transporter AcrB at 3.1A resolution reveals the N‐terminal region with conserved amino acids. J Struct Biol. 2007;158(3):494–502. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Dixon SL, Merz KM Jr. Fast, accurate semiempirical molecular orbital calculations for macromolecules. J Chem Phys. 1997;107:879–893. [Google Scholar]
  11. Eicher T, Cha HJ, Seeger MA, Brandstatter L, El‐Delik J, Bohnert JA, et al. Transport of drugs by the multidrug transporter AcrB involves an access and a deep binding pocket that are separated by a switch‐loop. Proc Natl Acad Sci U S A. 2012;109(15):5687–5692. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Emsley P, Lohkamp B, Scott WG, Cowtan K. Features and development of Coot. Acta Crystallogr D Biol Crystallogr. 2010;66(Pt 4):486–501. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Foo JL, Leong SS. Directed evolution of an E. coli inner membrane transporter for improved efflux of biofuel molecules. Biotechnol Biofuels. 2013;6(1):81. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Gogonea V, Merz KM Jr. Fully quantum mechanical description of proteins in solution. Combining linear scaling quantum mechanical methodologies with the Poisson–Boltzmann equation. J Phys Chem A. 1999;103(26):5171–5188. [Google Scholar]
  15. Husain F, Bikhchandani M, Nikaido H. Vestibules are part of the substrate path in the multidrug efflux transporter AcrB of Escherichia coli . J Bacteriol. 2011;193(20):5847–5849. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Husain F, Nikaido H. Substrate path in the AcrB multidrug efflux pump of Escherichia coli . Mol Microbiol. 2010;78(2):320–330. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Johnson RM, Fais C, Parmar M, Cheruvara H, Marshall RL, Hesketh SJ, et al. Cryo‐EM structure and molecular dynamics analysis of the fluoroquinolone resistant mutant of the AcrB transporter from salmonella. Microorganisms. 2020;8(6):943. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Kobayashi N, Tamura N, van Veen HW, Yamaguchi A, Murakami S. Beta‐Lactam selectivity of multidrug transporters AcrB and AcrD resides in the proximal binding pocket. J Biol Chem. 2014;289(15):10680–10690. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Lesnik S, Stular T, Brus B, Knez D, Gobec S, Janezic D, et al. LiSiCA: a software for ligand‐based virtual screening and its application for the discovery of Butyrylcholinesterase inhibitors. J Chem Inf Model. 2015;55(8):1521–1528. [DOI] [PubMed] [Google Scholar]
  20. Li J, Xu Q, Ogurek S, Li Z, Wang P, Xie Q, et al. Efflux pump AcrAB confers decreased susceptibility to Piperacillin‐Tazobactam and Ceftolozane‐Tazobactam in Tigecycline‐non‐susceptible Klebsiella pneumoniae . Infect Drug Resist. 2020;13:4309–4319. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Ma D, Alberti M, Lynch C, Nikaido H, Hearst JE. The local repressor AcrR plays a modulating role in the regulation of acrAB genes of Escherichia coli by global stress signals. Mol Microbiol. 1996;19(1):101–112. [DOI] [PubMed] [Google Scholar]
  22. Mallea M, Chevalier J, Bornet C, Eyraud A, Davin‐Regli A, Bollet C, et al. Porin alteration and active efflux: two in vivo drug resistance strategies used by Enterobacter aerogenes . Microbiology (Reading). 1998;144(Pt 11):3003–3009. [DOI] [PubMed] [Google Scholar]
  23. McCoy AJ, Grosse‐Kunstleve RW, Adams PD, Winn MD, Storoni LC, Read RJ. Phaser crystallographic software. J Appl Cryst. 2007;40(Pt 4):658–674. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Morgan CE, Glaza P, Leus IV, Trinh A, Su CC, Cui M, et al. Cryoelectron microscopy structures of AdeB illuminate mechanisms of simultaneous binding and exporting of substrates. mBio. 2021;12(1):e03690‐20. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Morris GM, Huey R, Lindstrom W, Sanner MF, Belew RK, Goodsell DS, et al. AutoDock4 and AutoDockTools4: automated docking with selective receptor flexibility. J Comput Chem. 2009;30(16):2785–2791. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Moseng MA, Lyu M, Pipatpolkai T, Glaza P, Emerson CC, Stewart PL, et al. Cryo‐EM structures of CusA reveal a mechanism of metal‐ion export. mBio. 2021;12(2):e00452‐21. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Muller RT, Travers T, Cha HJ, Phillips JL, Gnanakaran S, Pos KM. Switch loop flexibility affects substrate transport of the AcrB efflux pump. J Mol Biol. 2017;429(24):3863–3874. [DOI] [PubMed] [Google Scholar]
  28. Murakami S, Nakashima R, Yamashita E, Matsumoto T, Yamaguchi A. Crystal structures of a multidrug transporter reveal a functionally rotating mechanism. Nature. 2006;443(7108):173–179. [DOI] [PubMed] [Google Scholar]
  29. Murakami S, Nakashima R, Yamashita E, Yamaguchi A. Crystal structure of bacterial multidrug efflux transporter AcrB. Nature. 2002;419(6907):587–593. [DOI] [PubMed] [Google Scholar]
  30. Nakashima R, Sakurai K, Yamasaki S, Nishino K, Yamaguchi A. Structures of the multidrug exporter AcrB reveal a proximal multisite drug‐binding pocket. Nature. 2011;480(7378):565–569. [DOI] [PubMed] [Google Scholar]
  31. Nikaido H. Multidrug resistance in bacteria. Annu Rev Biochem. 2009;78:119–146. [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Nikaido H. Structure and mechanism of RND‐type multidrug efflux pumps. Adv Enzymol Relat Areas Mol Biol. 2011;77:1–60. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Nikaido H, Pages JM. Broad‐specificity efflux pumps and their role in multidrug resistance of Gram‐negative bacteria. FEMS Microbiol Rev. 2012;36(2):340–363. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Nikaido H, Takatsuka Y. Mechanisms of RND multidrug efflux pumps. Biochim Biophys Acta. 2009;1794(5):769–781. [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Ohene‐Agyei T, Mowla R, Rahman T, Venter H. Phytochemicals increase the antibacterial activity of antibiotics by acting on a drug efflux pump. Microbiol Open. 2014;3(6):885–896. [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Oswald C, Tam HK, Pos KM. Transport of lipophilic carboxylates is mediated by transmembrane helix 2 in multidrug transporter AcrB. Nat Commun. 2016;7:13819. [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Reading E, Ahdash Z, Fais C, Ricci V, Wang‐Kan X, Grimsey E, et al. Perturbed structural dynamics underlie inhibition and altered efflux of the multidrug resistance pump AcrB. Nat Commun. 2020;11(1):5565. [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Robert X, Gouet P. Deciphering key features in protein structures with the new ENDscript server. Nucleic Acids Res. 2014;42(W1):W320–W324. [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Schuster S, Vavra M, Kern WV. Evidence of a substrate‐discriminating entrance channel in the lower porter domain of the multidrug resistance efflux pump AcrB. Antimicrob Agents Chemother. 2016;60(7):4315–4323. [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Seeger MA, Schiefner A, Eicher T, Verrey F, Diederichs K, Pos KM. Structural asymmetry of AcrB trimer suggests a peristaltic pump mechanism. Science. 2006;313(5791):1295–1298. [DOI] [PubMed] [Google Scholar]
  41. Soparkar K, Kinana AD, Weeks JW, Morrison KD, Nikaido H, Misra R. Reversal of the drug binding pocket defects of the AcrB multidrug efflux pump protein of Escherichia coli . J Bacteriol. 2015;197(20):3255–3264. [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. Su CC, Yin L, Kumar N, Dai L, Radhakrishnan A, Bolla JR, et al. Structures and transport dynamics of a Campylobacter jejuni multidrug efflux pump. Nat Commun. 2017;8(1):171. [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Tam HK, Foong WE, Oswald C, Herrmann A, Zeng H, Pos KM. Allosteric drug transport mechanism of multidrug transporter AcrB. Nat Commun. 2021;12(1):3889. [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Tam HK, Malviya VN, Foong WE, Herrmann A, Malloci G, Ruggerone P, et al. Binding and transport of carboxylated drugs by the multidrug transporter AcrB. J Mol Biol. 2020;432(4):861–877. [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. Thanassi DG, Cheng LW, Nikaido H. Active efflux of bile salts by Escherichia coli . J Bacteriol. 1997;179(8):2512–2518. [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. Trott O, Olson AJ. AutoDock Vina: improving the speed and accuracy of docking with a new scoring function, efficient optimization, and multithreading. J Comput Chem. 2010;31(2):455–461. [DOI] [PMC free article] [PubMed] [Google Scholar]
  47. Tsukagoshi N, Aono R. Entry into and release of solvents by Escherichia coli in an organic‐aqueous two‐liquid‐phase system and substrate specificity of the AcrAB‐TolC solvent‐extruding pump. J Bacteriol. 2000;182(17):4803–4810. [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. Vagin A, Teplyakov A. Molecular replacement with MOLREP. Acta Crystallogr D Biol Crystallogr. 2010;66(Pt 1):22–25. [DOI] [PubMed] [Google Scholar]
  49. Veesler D, Blangy S, Cambillau C, Sciara G. There is a baby in the bath water: AcrB contamination is a major problem in membrane‐protein crystallization. Acta Crystallogr Sect F Struct Biol Cryst Commun. 2008;64(Pt 10):880–885. [DOI] [PMC free article] [PubMed] [Google Scholar]
  50. Wehmeier C, Schuster S, Fahnrich E, Kern WV, Bohnert JA. Site‐directed mutagenesis reveals amino acid residues in the Escherichia coli RND efflux pump AcrB that confer macrolide resistance. Antimicrob Agents Chemother. 2009;53(1):329–330. [DOI] [PMC free article] [PubMed] [Google Scholar]
  51. White DG, Goldman JD, Demple B, Levy SB. Role of the acrAB locus in organic solvent tolerance mediated by expression of marA, soxS, or robA in Escherichia coli . J Bacteriol. 1997;179(19):6122–6126. [DOI] [PMC free article] [PubMed] [Google Scholar]
  52. Winn MD, Ballard CC, Cowtan KD, Dodson EJ, Emsley P, Evans PR, et al. Overview of the CCP4 suite and current developments. Acta Crystallogr D Biol Crystallogr. 2011;67(Pt 4):235–242. [DOI] [PMC free article] [PubMed] [Google Scholar]
  53. Yamaguchi A, Nakashima R, Sakurai K. Structural basis of RND‐type multidrug exporters. Front Microbiol. 2015;6:327. [DOI] [PMC free article] [PubMed] [Google Scholar]
  54. Yu EW, McDermott G, Zgurskaya HI, Nikaido H, Koshland DE Jr. Structural basis of multiple drug‐binding capacity of the AcrB multidrug efflux pump. Science. 2003;300(5621):976–980. [DOI] [PubMed] [Google Scholar]
  55. Zhang Z, Morgan CE, Cui M, Yu EW. Cryo‐EM structures of AcrD illuminate a mechanism for capturing aminoglycosides from its central cavity. mBio. 2023;14(1):e0338322. [DOI] [PMC free article] [PubMed] [Google Scholar]
  56. Zwama M, Nishino K. Proximal binding pocket Arg717 substitutions in Escherichia coli AcrB cause clinically relevant divergencies in resistance profiles. Antimicrob Agents Chemother. 2022;66:e0239221. [DOI] [PMC free article] [PubMed] [Google Scholar]
  57. Zwama M, Yamasaki S, Nakashima R, Sakurai K, Nishino K, Yamaguchi A. Multiple entry pathways within the efflux transporter AcrB contribute to multidrug recognition. Nat Commun. 2018;9(1):124. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Data S1. Supporting Information.

PRO-34-e5252-s002.pdf (34.5KB, pdf)

Data S2. Supporting Information.

PRO-34-e5252-s001.pdf (2.6MB, pdf)

Articles from Protein Science : A Publication of the Protein Society are provided here courtesy of The Protein Society

RESOURCES