Knock‐out of multidrug efflux pump MexXY‐OprM results in increased susceptibility to antimicrobial peptides in Pseudomonas aeruginosa

Multidrug efflux systems of the resistance‐nodulation‐cell division family play a crucial role in resistance of Pseudomonas aeruginosa to a large variety of antibiotics. Here, we investigated the role of clinically relevant efflux pumps MexAB−OprM, MexCD−OprJ, and MexXY−OprM in resistance against different cationic antimicrobial peptides (AMPs). Our results indicate that a knock‐out in efflux pump MexXY‐OprM increased susceptibility to some AMPs by two‐ to eightfold. Our data suggest a contribution of MexXY‐OprM in resistance to certain AMPs in P. aeruginosa, which should be considered in the future development of new and highly active antimicrobial peptides to fight multidrug resistant infections.

The Gram-negative bacterium Pseudomonas aeruginosa is one of the most important opportunistic human pathogens causing a wide range of severe chronic and nosocomial infections. 1 Due to a large arsenal of intrinsic resistance mechanisms, P. aeruginosa is inherently resistant to many commonly used antibiotics including aminoglycosides, fluoroquinolones, and ß-lactams. In addition to the very low permeability of its outer membrane and the expression of antibiotic cleaving enzymes, P. aeruginosa possesses several multidrug efflux systems of the RND family. 2 Among the various resistance-nodulation-cell division (RND) efflux pumps in P. aeruginosa are MexAB-OprM, MexCD-OprJ, and MexXY-OprM, which are all capable of exporting several different classes of antibiotics and are involved in multidrug resistance in laboratory strains and clinical isolates of P. aeruginosa. 3,4 Antimicrobial peptides (AMPs) are an abundant and diverse group of molecules that are produced by many organisms as part of their first line defense. They are typically relatively short consisting of 10-60 amino acids, are positively charged with a net charge of +2 to +10 and have an amphiphilic character. 5 Since some of these peptides exhibit strong antimicrobial activities and their mechanism of action has been shown to address multiple targets within the bacterial cell resulting in low resistance development, they are considered as a promising class of new antimicrobial agents, in particular, against multidrug resistant Gram-negative bacteria including P. aeruginosa. 5,6 Besides intracellular targets such as inhibition of protein synthesis or binding of nucleic acids, many antimicrobial peptides disrupt the bacterial cell membrane by transmembrane pore-formation which subsequently leads to cell death. 7 Although the most relevant resistance mechanisms against polymyxins and AMPs in Gram-negative bacteria are membrane modification and proteolytic degradation, previous studies have also shown the involvement of transporters and efflux pumps in resistance to peptide antibiotics. 8,9 For example, it has been shown that the ABC transporter SapABCDF is a main transporter of AMPs in Haemophilus influenza, Salmonella typhimurium, and Proteus mirabilis. 8 In addition, the RND efflux pump MtrCDE has been shown previously to contribute to AMP resistance in the Gram-negative bacteria Neisseria gonorrhoeae and Neisseria meningitides. 10,11 More recently it has also been shown that the efflux pump MexXY-OprM contributes to the tolerance of P. aeruginosa to the peptide antibiotic colistin. 12 In this study, we investigated the role of clinically important RND efflux pumps MexAB-OprM, MexCD-OprJ, and MexXY-OprM in antimicrobial peptide resistance in P. aeruginosa by analyzing a set of efflux pump deficient and overexpressing strains.
To identify a role of RND efflux pumps in resistance to antimicrobial peptides in P. aeruginosa, we selected a set of efflux pump knock-out mutants and overexpressing strains (Table 1) and measured the minimal growth inhibitory concentrations (MICs) against a variety of natural and synthetic antimicrobial peptides ( Table 2). The determination of MICs was performed as described previously, 18 the cells were incubated in Mueller Hinton (MH) broth at 37°C for 18 h. Among the synthetic peptides were IDR-1018, 19 1037, 20 HHC-10, 21 and HHC-36 21 which are loosely based on linear bovine bactenecin variant Bac2A 21 and which have been reported previously to show a broad spectrum antibacterial activity against different pathogens including P. aeruginosa. 19,21,22 The MIC tests revealed a significant two-to eightfold increase in susceptibility against synthetic peptides Bac2a, 1037, HHC-10, and HHC-36 for the MexXYdeficient P. aeruginosa mutant K1525 (ΔMexXY-OprM) as well as for the triple mutant K2896 lacking all three efflux pumps MexAB-OprM, MexCD-OprJ, and MexXY-OprM in comparison with wildtype K767 (PAO1, Table 3). In contrast, no differences in MICs were observed between the tested strains of P. aeruginosa for indolicidin 23 and IDR-1018 (Table 3). In addition, MICs for gramicidin A, polymyxin B, and colistin did not reveal an increase in susceptibility in the triple mutant K2896 in comparison with WT K767 (Table 3). Furthermore, no differences in MICs were observed for IDR-1018, 1037, HHC-10, HHC-36, Bac2a, and indolicidin in the single knock-out mutants K1523 (ΔMexAB-OprM) and K1521 (ΔMexCD-OprJ) as well as for the MexAB-OprM overexpressing strain K1455, the MexCD-OprJ overexpressing strain K1536, and the MexXY overexpressing strain K2415 (Table 3). Overall, these MIC data suggested an involvement of the efflux pump MexXY-OprM in resistance to some antimicrobial peptides.   To further examine this hypothesis, we monitored the time-dependent killing of P. aeruginosa WT, mutant K1525 (ΔmexXY), and mexXY-overexpressing strain K2415 in the presence of peptide HHC-36, which showed the strongest effect in our MIC analyses. The cell densities of bacterial cultures were adjusted to~10 7 CFU/mL and HHC-36 was added at twofold MIC concentrations against WT (32 µg/mL) prior to the determination of CFUs after 0, 15, 30, 45, 60, and 90 min of incubation. The CFU counts for WT and K2415 still revealed cell numbers of 5.6 × 10 5 CFU/mL and 1.1 × 10 6 CFU/mL, respectively, after 90min of incubation with HHC-36 ( Figure 1). In contrast, at the end point of the experiment after 90 min, only 1.2 × 10 3 CFU/mL were detected in samples of efflux mutant K1525. Statistical significance of differences between CFU counts of K1525 and K767 as well as of K1525 and K2415 after 90 min was confirmed by a two-sided t-test for independent samples, resulting in p values < 0.001 in both cases.
In order to further verify these results, we amplified mexXY via PCR (primer sequences mexXY_forward: GAA CGTCCTCACAAGGGAAA, mexXY_reverse: GTGAACT  26 in P. aeruginosa due to the high intrinsic antimicrobial resistance to these agents (Table 3), we performed additional susceptibility tests with these peptides by using the more sensitive recombinant E. coli strains. For peptide ApidaecinIb, the MIC was eightfold higher in DH5α-pJET1.2::mexXY compared with the vector control strain (Table 4). In the case of Magainin 2, equal MICs of 4 µg/mL were determined for both E. coli strains.
Since MexXY-OprM expression has been shown to be induced by several of its substrates including the ribosometargeting antibiotics gentamicin and erythromycin, 27 we analysed mexX gene expression in response to HHC-36 and IDR-1018 using qRT-PCR (primer sequences GAGTACAC CGAAGCGCAGAC and GGCTGGGAGAAGTTCACGTA) as described previously. 22 Cells were grown in MH broth to midlog phase followed by a 2 h incubation with 20 µg/mL HHC-36 and IDR-1018, respectively. The obtained c t -values were normalized to the expression of the housekeeping gene rpoD. Samples were assayed three times in duplicate (n ≥ 6). Untreated cultures served as controls. Quantification of relative gene expression levels by qRT-PCR revealed an upregulation of mexX gene expression by 3.9 ± 1.2 fold in the presence of HHC-36 in comparison with the untreated control. In contrast, mexX gene expression was not altered in response to IDR-1018 as indicated by the relative gene expression levels of 0.9 ± 0.2 compared with untreated cells.
In agreement with our results, previous work has indicated that MexAB-OprM is not involved in resistance to peptide antibiotics including polymyxin B and the human defense peptides LL-37, HBD-2, and HBD-3. 22,28,29 although it was shown that MexAB is involved in the development of phenotypic tolerance to colistin in P. aeruginosa biofilms. 30 Furthermore, it has been demonstrated that susceptibility of P. aeruginosa to LL-37 was independent of efflux pumps MexCD-OprJ, MexEF-OprN, MexXY-OprM, and MexGHI-OpmD. 22 However, our MIC and kill-curve data suggest that MexXY-OprM contributes to resistance against several short AMPs in P. aeruginosa. This is in alignment with a recent study by Puja and colleagues who demonstrated that MexXY-OprM contributes to resistance against the peptide antibiotic colistin in P. aeruginosa. 12 Interestingly, while this study found identical MICs against colistin for wildtype PAO1 and the corresponding MexXY-deficient mutant, kill-curve experiments clearly showed a much stronger effect of colistin on cells of the mexXY mutant strain compared with wild-type. 12 The authors suggested multiple reasons for how MexXY-OprM might be involved in colistin resistance. Among these were a connection of mexXY with genes associated with regulation and modification of LPS, and the export of LPS components or transport of other molecules to the bacterial surface which counteract the electrostatic binding of colistin. A link between LPS and MexXY has also been suggested previously. 31 In addition, Puja et al. suggested the possibility that colistin could be a direct substrate for MexXY-OprM. 12 Initially a previous study excluded polymyxin antibiotics as substrates for this efflux pump, however, this hypothesis was based mainly on MIC data and Puja et al. provided a more detailed study. 29 To date, efflux-mediated resistance to antimicrobial peptides has been reported for several bacteria. In an early study, Shafer et al. 11 demonstrated that the Neisseria gonorrhoeae efflux system MtrCDE confers resistance to the cationic host defense peptides protegrin-1, LL-37, and tachyplesin-1, while Tzeng et al. 10 showed an involvement of MtrCDE in AMP resistance in Neisseria meningititis. In Yersinia enterocolitica the temperature-sensitive inducible efflux pump and potassium antiporter system RosAB has been proposed to mediate adaptive resistance to polymyxin B by an active peptide export and, in addition, by an acidification of the cytoplasm. 32 In Klebsiella pneumoniae an enhanced susceptibility toward fluoroquinolone, tetracycline, and aminoglycoside antibiotics as well as to human bronchoalveolar lavage fluid components HNP-1, HBD-1, and HBD-2 and to the bronchoalveolar lavage fluid itself was noticed upon knock-out of the inner membrane transporter protein AcrB of RND efflux system AcrAB-TolC. 33 In addition, this AcrAB-TolC knock-out was also correlated with a lowered resistance of Klebsiella pneumoniae against polymyxin B. 33 In conclusion, our data indicate that the P. aeruginosa efflux pump system MexXY-OprM modulates susceptibility against synthetic and natural antimicrobial peptides including 1037, HHC-36, HHC-10, Bac2a, and Apidaecin Ib. Cationic AMPs represent one promising class of antibiotics effective against both Gram-positive and Gram-negative pathogens. 5 However, since our data demonstrate that MexXY-OprM is implicated in AMP resistance in P. aeruginosa, this characteristic should be considered in future developments of new and highly active antimicrobial peptides. In addition, more studies are necessary to obtain a better understanding of the molecular mechanism involved in efflux-mediated peptide resistance in P. aeruginosa.

ACKNOWLEDGMENTS
The authors would like to thank Keith Poole for kindly providing P. aeruginosa strains and Robert Hancock for kindly providing synthetic AMPs used in this study. This work was supported by funds from the Karlsruhe Institute of Technology (KIT) and start-up funds from Carleton University.