Role of MexAB-OprM in intrinsic resistance of Pseudomonas aeruginosa to temocillin and impact on the susceptibility of strains isolated from patients suffering from cystic fibrosis

Sir, Temocillin (6-α-methoxy-ticarcillin) is resistant to most β-lactamases, including AmpC and extended-spectrum β-lactamases, and is therefore considered a useful alternative to carbapenems in infections caused by several resistant Gram-negative pathogens.1 Yet, temocillin is inactive against Pseudomonas aeruginosa, possibly because of poor permeation across the outer membrane barrier and/or reduced binding to penicillin-binding proteins.1 However, the role of multidrug efflux systems has not been examined so far. Three multidrug efflux systems have been reported to export β-lactams in P. aeruginosa, namely (from least to most effective) MexXY-OprM, MexCD-OprJ and MexAB-OprM.2 We wondered whether temocillin could be the substrate of one or several of these transporters. Temocillin (Eumedica, Brussels, Belgium) and ticarcillin (disodium salt; Sigma–Aldrich, St Louis, MO, USA) were tested against: (i) the wild-type reference strain PAO1; (ii) a panel of laboratory strains with specific disruption(s) of the gene(s) encoding the three transporters mentioned above and MexEF-OprN, another efflux pump accommodating fluoroquinolones, trimethoprim and chloramphenicol, but not β-lactams,2 and producing different levels of AmpC; (iii) clinical isolates from patients hospitalized in intensive care units (ICUs) with hospital-acquired pneumonia (HAP); and (iv) strains from cystic fibrosis patients that were found to be hypersusceptible to carbenicillin and ticarcillin (TicHS phenotype) due to mutations in mexA or mexB.3 MICs were determined by microdilution in Mueller–Hinton broth (pH 7.4, 24 h) without or with the broad-spectrum efflux inhibitor Phe-Arg-β-naphthylamide (PAβN; 50 mg/L; Sigma–Aldrich).4 The expression of mexA and mexX was measured by quantitative real-time PCR, and that of mexC and mexE was measured by semi-quantitative RT–PCR. mexA, mexB and oprM were sequenced in strains from cystic fibrosis patients.3 The MIC of temocillin for PAO1 was ≥256 mg/L, but fell to 64 mg/L when tested in the presence of PAβN, a broad-spectrum competitive inhibitor of efflux transporters (Table 1), suggesting a role of active efflux in the intrinsic high-level resistance of P. aeruginosa to temocillin. The magnitude of the inhibitory effect of PAβN, however, varies depending on the substrate.4 To better quantify the impact of efflux on temocillin MICs, and also to identify the transporter(s) responsible for its efflux, we used isogenic strains deficient in the main efflux systems. Disruption of MexCD-OprJ, MexEF-OprN or MexXY only slightly affected the temocillin MIC (2–3 log2 reduction), consistent with the strongly repressed expression of these three systems in wild-type strains. In contrast, disruption of mexB, mexAB, oprM or mexAB-oprM decreased MICs to values as low as 2–4 mg/L, with a minimal additional effect of PAβN. Conversely, overexpression of mexAB, but not of mexXY, further increased the temocillin MIC compared with PAO1. This clearly indicates that MexAB-OprM-driven efflux strongly contributes to the intrinsic resistance of P. aeruginosa to temocillin, while the other Mex systems only play a minor role. We also confirmed the stability of temocillin to AmpC. MICs of temocillin and ticarcillin for P. aeruginosa strains with known efflux characteristics, as measured in Mueller–Hinton broth (MHB) or in MHB supplemented with the broad-spectrum efflux inhibitor PAβN (50 mg/L) ND, not determined. aReal-time quantitative PCR [threshold ratio compared with PAO1; values shown in bold are considered as denoting highly significant overexpression (≥2 and ≥5 for mexA and mexX, respectively, based on the recommendations of the manufacturer of the kit used for their detection; no threshold value set for oprM); values interpreted as denoting an absence (or quasi-absence) of detection are shown in italics]. bRT–PCR [semi-quantitative detection (+/−)]. cPAβN (broad-spectrum efflux inhibitor) used at 50 mg/L. dVettoretti et al. Antimicrob Agents Chemother 2009; 53: 1987–97. eRobertson et al. J Bacteriol 2007; 189: 6870–81. fMima et al. J Bacteriol 2007; 189: 7600–9. gComplete absence of detection. hS. Guénard and P. Plésiat (unpublished results). iMuller et al. Antimicrob Agents Chemother 2011; 55: 1211–21. jLi et al. Antimicrob Agents Chemother 1994; 38: 1732–41. kHamzehpour et al. Antimicrob Agents Chemother 1995; 39: 2392–6. lNo growth in the presence of PAβN (PAβN MIC = 25 mg/L for this strain). mMoya et al. PLoS Pathog 2009; 5: e1000353. nIsolated from ICUs in Belgium. MICs of temocillin and ticarcillin for P. aeruginosa strains with known efflux characteristics, as measured in Mueller–Hinton broth (MHB) or in MHB supplemented with the broad-spectrum efflux inhibitor PAβN (50 mg/L) ND, not determined. aReal-time quantitative PCR [threshold ratio compared with PAO1; values shown in bold are considered as denoting highly significant overexpression (≥2 and ≥5 for mexA and mexX, respectively, based on the recommendations of the manufacturer of the kit used for their detection; no threshold value set for oprM); values interpreted as denoting an absence (or quasi-absence) of detection are shown in italics]. bRT–PCR [semi-quantitative detection (+/−)]. cPAβN (broad-spectrum efflux inhibitor) used at 50 mg/L. dVettoretti et al. Antimicrob Agents Chemother 2009; 53: 1987–97. eRobertson et al. J Bacteriol 2007; 189: 6870–81. fMima et al. J Bacteriol 2007; 189: 7600–9. gComplete absence of detection. hS. Guénard and P. Plésiat (unpublished results). iMuller et al. Antimicrob Agents Chemother 2011; 55: 1211–21. jLi et al. Antimicrob Agents Chemother 1994; 38: 1732–41. kHamzehpour et al. Antimicrob Agents Chemother 1995; 39: 2392–6. lNo growth in the presence of PAβN (PAβN MIC = 25 mg/L for this strain). mMoya et al. PLoS Pathog 2009; 5: e1000353. nIsolated from ICUs in Belgium. To examine the clinical relevance of our observations, we measured temocillin MICs for isolates collected from ICU patients with HAP. All values were high, but those from isolates overexpressing mexA were higher than those for PAO1, corroborating the importance of this efflux system in temocillin resistance. In parallel, we found that isolates obtained from cystic fibrosis patients and showing hypersusceptibility to ticarcillin were also hypersusceptible to temocillin, with MICs ranging between 1 and 4 mg/L in most cases. Interestingly enough, however, the MICs for some isolates with single nucleotide mutations in MexA (G72S and Y197C) remained moderately elevated (32 mg/L), suggesting that these mutated proteins remained partly functional. Noteworthy, when considering all isolates examined here, differences between temocillin and ticarcillin MICs were much greater in isolates producing a functional or partially functional MexAB-OprM pump than in deficient strains (with temocillin MICs being 3–5 log2 dilutions higher than those of ticarcillin). This suggests that temocillin is a preferential substrate for the MexAB-OprM transporter compared with ticarcillin, pointing to a potential role of the 6-α-methoxy substituent in its recognition and efflux. While intrinsic resistance of P. aeruginosa to temocillin makes this antibiotic unusable in most conventional clinical set-ups, we see here that impairment of efflux lowers the MICs to values below the current clinical susceptibility breakpoint for Enterobacteriaceae (16 mg/L; UK and Belgium) or even the pharmacokinetic/pharmacodynamic breakpoint proposed for a 4 g daily dose (8 mg/L).5 This may further trigger current efforts in designing clinically useful inhibitors of the MexAB-OprM transporter, since such combined therapy could provide the clinician with a useful alternative to current antipseudomonal β-lactams, especially if considering temocillin's remarkable β-lactamase stability. The present data may also have potential immediate application for cystic fibrosis patients. These patients can be infected by Burkholderia cepacia, against which temocillin is active and, therefore, commonly used.1 Because of the large prevalence of P. aeruginosa isolates with the hypersusceptible TicHS phenotype in this patient population,3 temocillin could contribute to their eradication as well. Testing for temocillin susceptibility of P. aeruginosa isolated from cystic fibrosis patients appears, therefore, potentially useful. This work was supported by the Belgian Fonds pour la Recherche Scientifique Médicale (FRSM; grants 3.4597.06 and 3.4583.08). F. V. B. is Maître de Recherches of the Belgian Fonds de la Recherche Scientifique (F.R.S.-FNRS). J. M. B. is supported by the BioWin programme of the Belgian Région Wallonne (DG06). P. M. T. is an unpaid adviser of Eumedica (manufacturer of temocillin); he does not have any financial interests in this company. J. M. B., S. G., P. P. and F. V. B. have no conflicts of interest to declare. We thank the clinical microbiology laboratories from our collaborating hospitals for collecting clinical isolates from patients suffering from HAP (see list in Riou et al. Int J Antimicrob Agents 2010; 36: 513–22) and Eumedica, Brussels, Belgium, for providing us with microbiological standards of temocillin. We are also grateful to Virginie Mohymont for excellent technical assistance.