Conessine as a novel inhibitor of multidrug efflux pump systems in Pseudomonas aeruginosa

P. aeruginosa PAO1 strain K767 (wild-type), MexAB-OprM overexpressed strain K1455 (PAO1-nalB), and MexB deletion strain K1523 (PAO1-∆mexB) were generously provided by Professor Dr. R. Keith Poole, Queen’s University, Kingston, Ontario, Canada.

Mueller-Hinton broth (MHB) and Tryptic soy agar (TSA) were purchased from Becton Dickinson Microbiology Systems (Sparks, MD, USA).

Minimum inhibitory concentration was tested by broth microdilution assay in accordance with the Clinical and Laboratory Standards Institute (CLSI) recommendation [18]. Antibiotics used in this study were selected based on substrate specificity of Ade efflux pump in A. baumannii: cefotaxime for AdeDE pump, novobiocin for AdeIJK pump, rifampicin for AdeDE and AdeIJK pump, erythromycin, levofloxacin, and tetracycline for AdeABC, AdeDE, and AdeIJK pump. In addition, cefotaxime, levofloxacin, novobiocin, and tetracycline have been reported as substrates for MexAB-OprM in P. aeruginosa. Stock solution of novobiocin (50 mg/L), rifampicin (1 mg/L), levofloxacin (18 mg/L), erythromycin (2 mg/L), cefotaxime (10 mg/L), and PAβN (10 mg/L) were prepared in sterile deionized water. Tetracycline (4 mg/L) and conessine (1 mg/L) were dissolved in 95% ethanol and 100%DMSO, respectively. Serial dilutions of conessine, PAβN, and antibiotics were prepared in MHB. In order to investigate the effect of each agent, 100 μL bacterial culture (1 × 10⁶ cfu/mL) was mixed with 100 μL each conessine, PAβN, or antibiotics. Synergistic effects of conessine (20 mg/L) or PAβN (25 mg/L) and antibiotics were assessed using checkerboard assay by adding 100 μL culture into a well containing 50 μL conessine or PAβN and 50 μL antibiotics. DMSO at a final concentration of 1% used as a negative control and PAβN, an efflux pump inhibitor was used as a positive control. Plates were then read after 18 h of incubation at 37 °C. Each assay with three biological triplicates was repeated at least twice. A 4-fold or greater reduction in MIC values after addition of conessine or PAβN was considered significant. Fractional inhibitory concentration index (FICI) value was calculated for each combination according to the following formula [19]: FICI = (MIC of efflux pump inhibitors in combination/MIC of efflux pump inhibitors alone) + (MIC of antibiotics in combination/MIC of antibiotics alone). Synergy, additivity, and antagonism were defined as FICI <1, =1, and >1, respectively.

H33342 accumulation assay was performed to evaluate the effect of efflux pump inhibitors on the activity of MexAB-OprM efflux pump [20]. Briefly, overnight bacterial cultures were inoculated into MHB and rotated at 250 rpm at 37 °C for 4–5 h. Bacterial cells were harvested by centrifugation (3000 rpm for 15 min) and the cells were washed with phosphatebuffered saline containing 1 mM MgSO₄ and 20 mM glucose. After centrifugation, the pellets were resuspended in the same buffer and OD₆₀₀ of each suspension was adjusted to 0.4. An aliquot of 100 μL of the bacterial suspension was added into a well in black microtiter plate containing each of 50 μL conessine (20 mg/L) or an efflux pump inhibitor, PAβN (25 mg/L).

The final concentration of DMSO in all assays was ≤1%. Plates were incubated at 37 °C for 15 min and 50 μL H33342 (2.5 μM) was added to each assay well. Fluorescence (excitation 355 nm, emission 460 nm) was measured at 37 °C every 2.30 min for 1 h using a Varioskan Flash spectral scanning multimode reader. Each assay was repeated at least twice. Differences in accumulation in the presence of efflux pump inhibitors compared with the absence of efflux pump inhibitors were analysed for statistical significance using Student’s t-test. P value ≤0.05 was considered significant.

Ability of conessine to permeabilize P. aeruginosa outer membrane was assessed by NPN uptake assay [21]. NPN, an uncharged lipophilic molecule, fluoresces weakly in aqueous environments but becomes strongly fluorescent in nonpolar environments such as cell membranes. Briefly, overnight bacterial cultures were inoculated into MHB and rotated at 250 rpm at 37 °C for 4–5 h. Bacterial cells were harvested at 3000 rpm for 15 min, washed with 100 mM NaCl and 50 mM sodium phosphate buffer (pH 7.0), and resuspended in the same buffer at A 600 = 0.1 in the presence of 0.05% of glucose. An aliquot of 100 μL of the bacterial suspension was added into a well in black microtiter plate containing each of 50 μL conessine (20 mg/L) or EDTA (100 μM) as a permeabilizer followed by adding 50 μL of NPN (40 μM). The final concentration of DMSO in all assays was ≤1%. NPN fluorescence intensity (excitation 322 nm, emission 424 nm) was monitored at 37 °C after 2.30 min for 1 h using a Varioskan Flash spectral scanning multimode reader (Thermo Fisher Scientific, Finland). Each assay was repeated at least twice. Differences in accumulation in the presence of efflux pump inhibitors compared with the absence of efflux pump inhibitors were analysed for statistical significance using Student’s t-test. P value ≤0.05 was considered significant.

MICs of conessine and PAβN for P. aeruginosa wild-type strain K767 (PAO1), MexAB-OprM overexpressed strain K1455 (PAO1-nalB), and MexB deletion strain K1523 (PAO1-∆mexB) were determined. Intrinsic MICs of conessine in all strains were 40 mg/L while MICs of PAβN were between 512 and 1024 mg/L. At a concentration required to reduce MICs of antibiotics by at least 4-fold in P. aeruginosa strains, conessine (20 mg/L) and PAβN (25 mg/L) alone did not affect growth rate in all strains (data not present).

Susceptibility of P. aeruginosa strains to a range of antibiotics is shown in Table 1. Overexpression of MexAB-OprM conferred resistance to cefotaxime, erythromycin, levofloxacin, novobiocin, and tetracycline, except rifampicin. The experiments supported earlier report that resistance to rifampicin may involve mutations in rpoB gene rather than as a simple function of expression of efflux systems [22]. A mutant strain with MexB deletion displayed susceptibility to cefotaxime, levofloxacin, and novobiocin. The findings indicate that extrusion of cefotaxime, levofloxacin, and novobiocin was mainly specific to MexAB-OprM efflux system. On the other hand, resistance to erythromycin, rifampicin, and tetracycline may involve other efflux pump systems. Together, the results are consistent with previous studies on substrate specificity of Mex efflux systems in P. aeruginosa [23, 24].

To verify that a mechanism by which conessine potentiate antibacterial activity was through inhibition of MexAB-OprM efflux, this study determined whether conessine enhanced antibacterial activity against wild-type, MexAB-OprM overexpressed, and MexB deletion strain. As shown in Table 1, antibiotic spectrum was affected by the addition of conessine and PAβN. Conessine significantly reduced MICs of all antibiotics by at least 8-fold in wild-type and MexAB-OprM overexpressed strain. In the overexpressed strain, the levels were comparable to those obtained in wild-type strain for cefotaxime, levofloxacin, and tetracycline. Interestingly, with erythromycin, novobiocin, and rifampicin, MICs were 4- to 8-fold less than those in the wild-type strain. With mexB deletion strain, conessine increased its susceptibility to almost all antibiotics, except novobiocin. Previous studies documented that other efflux pumps such as MexAB-OprM and MexCD-OprJ could export novobiocin [23, 24]. However, no synergistic effect was observed in MexB deletion strain, indicating that conessine may inhibit only MexAB-OprM pump to restore novobiocin activity. The results clearly demonstrated that conessine could be an inhibitor of MexAB-OprM efflux pump. Synergistic activity between other antibiotics (except novobiocin) and conessine observed in MexB deletion strain suggested that conessine might inhibit other efflux systems present in P. aeruginosa. Conessine increased susceptibility to rifampicin in all strains when this agent was not exported by MexAB-OprM efflux pump, indicating that conessine enhanced antibiotic susceptibility in MexAB-OprM overexpressed strain independent of an impact on MexAB-OprM. Restoration of rifampicin efficacy by conessine might result in increase in antibiotic susceptibility. In addition, conessine has the same 8-fold reduction impact on antibiotic resistance in the wild-type, mexB deletion, and MexAB-OprM overexpressed strain, suggesting that conessine may act on other intrinsic resistance determinants.

A well-known inhibitor of RND efflux systems, PAβN [5], increased susceptibility to all antibiotics in the wild-type strain whereas in MexAB-OprM overexpressed strain, the inhibitor could not reduce MICs of tetracycline and cefotaxime. In MexB deletion strain, MICs of all antibiotics except novobiocin were affected by PAβN. The results are consistent with previous reports indicating that PAβN inhibited MexAB-OprM pump as well as other Mex efflux systems in P. aeruginosa such as MexCD-OprJ and MexEF-OprN [5, 25]. Remarkably, conessine could decrease MICs of levofloxacin, novobiocin, and rifampicin in all strains to the levels comparable to those when combined with PAβN. With cefotaxime, erythromycin, and tetracycline, conessine could lower the MICs in all the strains better than PAβN. The findings clearly demonstrated that conessine could reduce MIC against P. aeruginosa 2- to 8-fold lower magnitude in the tested strains, compared with PAβN.

H33342 accumulation assay was used to confirm that conessine directly inhibited efflux pump systems in P. aeruginosa [20]. Conessine at 20 mg/L demonstrated an increase in H33342 accumulation in a wild-type strain (Fig. 1a). However, the effect of conessine was less than PAβN. A similar pattern was observed from an overexpressed strain (Fig. 1b). In a pump-deficient strain (Fig. 1c), conessine could elevate the level of H33342 accumulation to the level comparable to those when combined with PAβN. The findings suggested that conessine inhibited MexAB-OprM efflux pump and other efflux pump systems in P. aeruginosa.

Ability of conessine to permeabilize outer membrane of P. aeruginosa was determined with intact cells by NPN uptake assay [21]. As shown in Fig. 2, conessine (20 mg/L) did not weaken outer membrane of all P. aeruginosa strains as evidenced by no increase in NPN uptake. Conversely, a positive control EDTA (100 μM) promoted NPN uptake across outer membrane of all tested strains as indicated by a significant increase in NPN uptake. The findings suggested that conessine has no effect on outer membrane permeability. Different P. aeruginosa genotypes exhibited different level of fluorescence especially in MexB deletion strain which demonstrated the highest level of fluorescence. Regarding MexAB-OprM overexpressed strain, the level of fluorescence was lower than in wild-type due to the efflux pump activity of MexAB-OprM. Conessine did not increase NPN uptake in all the tested strains, thus fluorescence level after the bacterial cells treated with conessine were the same. The level of fluorescence in the presence of conessine did not lower than the control, in particular for a pump-deficient strain (Fig. 2c) but because of the increase in fluorescence in the deletion strain.