Pyridylpiperazine efflux pump inhibitor boosts in vivo antibiotic efficacy against K. pneumoniae

Abstract

Antimicrobial resistance is a global problem, rendering conventional treatments less effective and requiring innovative strategies to combat this growing threat. The tripartite AcrAB-TolC efflux pump is the dominant constitutive system by which Enterobacterales like Escherichia coli and Klebsiella pneumoniae extrude antibiotics. Here, we describe the medicinal chemistry development and drug-like properties of BDM91288, a pyridylpiperazine-based AcrB efflux pump inhibitor. In vitro evaluation of BDM91288 confirmed it to potentiate the activity of a panel of antibiotics against K. pneumoniae as well as revert clinically relevant antibiotic resistance mediated by acrAB-tolC overexpression. Using cryo-EM, BDM91288 binding to the transmembrane region of K. pneumoniae AcrB was confirmed, further validating the mechanism of action of this inhibitor. Finally, proof of concept studies demonstrated that oral administration of BDM91288 significantly potentiated the in vivo efficacy of levofloxacin treatment in a murine model of K. pneumoniae lung infection.

Keywords: AcrAB-TolC; Antibiotic Efflux Pump; Antimicrobial Resistance; Cryo-EM; Efflux Pump Inhibitor.

Figure 1. Mechanism of action for BDM91288 (5).

(A) Side view of the cryo-EM density map of K. pneumoniae AcrB in complex with BDM91288 (5) at an overall resolution of 2.97 Å. (B) Side view of the structure model of the AcrB trimer based on the electron densities in (A) in complex with BDM91288 (5). Densities and structures in A and B comprise the L (blue), T (yellow), and O (red) protomers and the BDM91288 (5) inhibitor binding to the TM domain of the L protomer is indicated in green colour. The L protomer in complex with the inhibitor adopts a transition state between the O state and L state. The boundaries of the inner membrane (IM) are indicated for the L protomer. (C) Enlarged view of BDM91288 (5) (green, stick representation, with green mesh for the observed density) binding site at the cytoplasmic side of the TM domain nested between transmembrane (TM) helices 4, 5, 10 of the AcrB L protomer (blue). (D) Superimposition of the binding position of BDM88855 (orange, stick representation) (Plé et al, 2022) and BDM91288 (5). The interacting residues (blue sticks, with oxygen atoms in red) forming salt bridges between the distal piperazine ring and D408, and the proximal piperazine ring with D950, are indicated by dashed lines and numbers represent the distance in Å. Further interactions include the halogen bond between the BDM91288 (5) chlorine and the main chain carbonyl oxygen of K939 (TM helix 10), as well as the hydrogen bond between the distal piperazine ring with the main chain carbonyl oxygen of L404 (TM helix 4) (Fig. EV2).

Figure 2. Checkerboard assay evaluating levofloxacin antibiotic activity in the presence of different concentrations of BDM91288 (5) and PAβN.

Susceptibility studies were performed on K. pneumoniae strain (WT), and isogenic mutants containing a point mutation in acrB (AcrB^V448E), ramR (RamR^mut = K124*stop) or both (AcrB^V448E RamR^mut). Values indicate the bacterial viability expressed as the median percentage of resazurine reduction (yellow cells >5%, blue cells <5%). Data represent the mean bacterial viability of at least three biological replicates. Source data are available online for this figure.

Figure 3. BDM91288 (5) boosts the bactericidal effect of levofloxacin in a murine model of K. pneumoniae lung infection.

C57BL/6JRj mice (n = 15–24) were infected intranasally with 1 × 10⁴ K. pneumoniae ATCC 43816, treated or not 20 h post-infection with EPI BDM91288 (5) (30 mg/kg, orally) and 90 min later treated or not with levofloxacin (10 mg/kg or 50 mg/kg, intraperitoneally). Lungs of each animal were sampled at 4 h or 24 h post levofloxacin treatment to assess bacterial colonization. Plots show the mean ± SEM and represent a pool of 2 (24 h time point) or 3 (4 h time point) independent experiments. Statistical significance compared to levofloxacin 10 mg/kg group was assessed by the Mann–Whitney test (**P = 0.0014, *P = 0.0333). CFU: colony forming units. Source data are available online for this figure.

Figure EV1. Rate of berberine uptake by K. pneumoniae WT and K. pneumoniae AcrB^V448E in dependence of increasing BDM91288 (5) concentrations.

Data are a mean and SD of four independent experiments. Source data are available online for this figure.

Figure EV2. Structure of KpAcrB in complex with BDM91288 (5).

Enlarged view of the inhibitor binding site showing interacting residues (blue sticks) in less or equal than 4 Å distance from BDM91288 (5). The salt bridge between the distal piperazine ring and D408 (TM helix 4), the halogen bond between the BDM91288 (5) chlorine and the K939 (TM helix 10) main chain carbonyl oxygen, the hydrogen bond between the L404 (TM helix 4) main chain carbonyl oxygen and the distal piperazine ring, and the salt bridge between D950 (TM helix 10) and the proximal piperazine ring are indicated by dashed lines and numbers represent the distance in Å.

Figure EV3. Pharmacokinetic profile of a single dose BDM91288 (5) (30 mg/kg, orally, formulated in 10% hydroxypropyl-β-cyclodextrin) in mice.

(A) Concentration of BDM91288 (5) versus time in whole lung and plasma. Plots show the mean ± SD (n = 3). (B) Concentration of BDM91288 (5) versus time in lung after bronchoalveolar lavage (BAL), epithelial lining fluid (ELF) and plasma. BDM91288 (5) was given in combination with a single dose levofloxacin (10 mg/kg in water, i.p.). Plots show the mean ± SD (n = 4). Source data are available online for this figure.