Effect of pH on the in vitro susceptibility of planktonic and biofilm-grown Proteus mirabilis to the quinolone antimicrobials

Authors


Correspondence

Colin P. McCoy, School of Pharmacy, Queen's University Belfast, 97 Lisburn Road, Belfast BT9 7BL, UK. E-mail: c.mccoy@qub.ac.uk

Abstract

Aims

To examine the effect of elevated pH, as reported during urinary catheter infections, on quinolone activity against the urease-producing pathogen Proteus mirabilis.

Methods and Results

Susceptibility of Pr. mirabilis to nalidixic acid, norfloxacin and ciprofloxacin was examined in media of pH 5 to pH 10 by determination of MICs, MBCs, minimum biofilm eradication concentrations (MBECs) and time-kill assays. Elevation of media pH from 5 to 9 caused a 10-fold decrease in bacteriostatic activity of nalidixic acid and was also associated with loss of the characteristic ‘paradoxical’ bactericidal activity. Alkaline pH, however, increased both bacteriostatic and bactericidal activities of the two fluoroquinolones tested against both planktonic and biofilm-associated Pr. mirabilis; MBC and MBEC values for ciprofloxacin decreased approx. 6000-fold and 10-fold, respectively, between pH 5 and pH 9. Rates of kill of all three agents were most rapid at pH 7, the optimal pH for bacterial replication.

Conclusions

pH has a pronounced effect on quinolone-mediated killing, which may be attributed to the dependence of cellular uptake on quinolone ionization state.

Significance and Impact of the Study

These results provide rationale for the use of these agents for Pr. mirabilis eradication in alkaline environments, including urinary catheter infections: the incidence, recurrence and recalcitrance of which pose a significant burden to healthcare providers.

Introduction

Heterocyclic carbonic acid derivatives, for example nalidixic acid (Lesher et al. 1962), cinoxacin (Lumish 1975) and oxolinic acid (Kershaw 1975), together with the 6-fluorinated agents, including ciprofloxacin (Bauernfeind and Petermuller 1983) and norfloxacin (Ito 1980), have found extensive therapeutic application in the treatment of urinary tract infections on the basis of their wide and potent spectrum of activity against common Enterobacteriaceae, of which Proteus mirabilis, Escherichia coli, Klebsiella aerogenes, Shigella spp. and Salmonella spp. are all sensitive (Bauernfeind 1983; Wolfson 1985; Jacoby 2011).

Despite the discovery of the first quinolone, nalidixic acid, half a century ago, the exact mechanisms involved in their antibacterial activity have never been fully elucidated (Hooper 2002). Quinolone-mediated killing, which occurs rapidly at concentrations exceeding the minimum bactericidal concentration (MBC), has recently been described as a two-part ‘poison’ hypothesis, involving formation of an initial reversible quinolone–topoisomerase–DNA complex, thereafter followed by lethal irreversible bacterial cell damage (Drlica 2008). The primary intracellular target in Gram-negative bacteria is DNA gyrase, the enzyme responsible for introducing negative supercoils into DNA; however, the secondary inhibition of RNA and protein syntheses at levels above the optimum bactericidal concentration results in the widely reported ‘paradoxical’ survival of bacteria associated with high concentrations of nalidixic acid (Crumplin 1975; Stevens 1980). This biphasic bactericidal response, reported against many uropathogens including E. coli, K. aerogenes and Pr. mirabilis (Crumplin 1975; Stevens 1980), is not, however, characteristic of all quinolones. For example, ciprofloxacin, a 6-fluorinated derivative, demonstrates lethality in the presence of protein synthesis inhibitors due to its ability to kill bacteria via an additional bactericidal mechanism independent of competent protein or RNA synthesis (Lewin 1991).

The urinary tract, a common site of action of the quinolones, maintains a normal physiological pH range between 4·8 and 8 (Tallgren 2009). Elevations in urine pH to levels up to 9·1 have, however, been reported under pathological conditions, for example during infections by the urease-producing pathogen Pr. mirabilis as a result of urease-catalysed hydrolysis of urea into ammonia. This elevation in urinary pH results in the precipitation of crystals of calcium and magnesium phosphate, causing encrustation of the urinary catheter through the formation of crystalline biofilms. The resultant blockage of the catheter lumen leads to either incontinence, as urine leaks around the outside of the catheter, or painful distension of the bladder (Stickler 2006). Indeed, Pr. mirabilis has been isolated from the urine of approx. 40% of patients undergoing long-term catheterization (Mobley 1996) and, furthermore, represents the organism most commonly recovered from the catheter biofilms of patients experiencing recurrent catheter encrustation and blockage (Stickler et al. 1998). A thorough understanding of the effect of pH on antibacterial efficacy of the quinolones is therefore essential for optimizing clinical use of these agents in the chemotherapy of urinary tract infections, which represent one of the most common nosocomial infections (Foxman 2010). Early studies have reported significant differences in quinolone activity according to environmental pH; for example, Bauernfeind (1983) demonstrated enhanced bacteriostatic activity of the fluoroquinolone agents, ciprofloxacin and norfloxacin, against Pr. mirabilis with elevated pH, in contrast to the reduced antibacterial efficacy of their predecessors, cinoxacin and nalidixic acid, previously reported against this pathogen in alkaline media (Giamarellou and Jackson 1975). The effect of pH on bactericidal activity of the quinolones, with specific reference to this ‘paradoxical’ behaviour of nalidixic acid and the quinolone-mediated eradication of Pr. mirabilis biofilms, has, however, never been fully investigated, despite the predominance of biofilms in urinary catheter infections.

The aim of this present study was therefore to investigate the effect of pH on growth and susceptibility of a representative urease-producing urinary pathogen, Pr. mirabilis, in both planktonic and biofilm mode of growth, to nalidixic acid and two fluoroquinolone agents, norfloxacin and ciprofloxacin, to determine their potential efficacy for treatment of Pr.  mirabilis infections of the urinary tract where pH elevations are commonly reported. Bacterial growth curves were plotted at pH 5, pH 7, pH 9 and pH 10, and susceptibility of Pr.  mirabilis to nalidixic acid, norfloxacin and ciprofloxacin was assessed at these pH values by the determination of MIC, MBC and MBEC values, with kill kinetics also evaluated by performance of time-kill assays.

Materials and methods

Bacterial strain, growth media and antimicrobials

All studies were performed using Pr.  mirabilis ATCC 35508 obtained from LGC Standards (Middlesex, UK). The quinolone antibiotics, nalidixic acid, norfloxacin and ciprofloxacin, were purchased from Sigma-Aldrich (Poole, Dorset, UK), and all formulated bacteriological media were obtained from Oxoid Ltd. (Hampshire, UK). Mueller–Hinton broth (MHB) was adjusted to specific pH values by the addition of hydrochloric acid or sodium hydroxide.

Determination of antimicrobial susceptibility

Minimum inhibitory concentration (MIC) values were determined according to the CLSI broth microdilution method (Wayne 2012), with the following modification: pH-adjusted MHB was employed as the media. Five replicates of all test concentrations and controls were conducted. Culture media (10 μl) from wells showing no visible signs of turbidity were transferred to Mueller–Hinton agar (MHA) plates, and the minimum bactericidal concentration (MBC) was determined after overnight incubation at 37°C as the lowest concentration required to kill 99·9% of the initial inoculum. Minimum biofilm eradication concentrations (MBECs) were determined using the Calgary biofilm device (Innovotech Inc., Edmonton, Canada) as described by Ceri et al. (1999). Briefly, biofilms were established on the peg lids of MBEC plates over 24 h, according to manufacturer's instructions. Peg lids were then rinsed in phosphate-buffered saline (PBS) before exposing to serial dilutions of the test antibiotic prepared in MHB of the required pH. After 24 h, the lid assembly was rinsed twice in PBS, transferred to MHB supplemented with universal neutralizers (0·125% L-histidine, 0·125% L-cysteine and 0·25% reduced glutathione), as previously described (Carson et al. 2009), and sonicated for 10 min to retrieve biofilm bacteria. The MBEC was recorded after overnight incubation at 37°C as the lowest antibiotic concentration to prevent visible bacterial growth in wells of the microtitre plate.

Determination of bacterial growth and quinolone rates of kill

Bacterial growth rate was determined by the turbidimetric method: samples from bacterial cultures in MHB of the required pH were removed every 30 min, and optical density at 550 nm was measured with a WPA C07000 Colourwave Colorimeter (Biochrom Ltd., Cambridge, UK), as previously described (Senior 1983). Growth rates were determined on three successive occasions with representative curves from one study displayed in Fig. 1. Time-kill assays were performed following CLSI methods (Wayne 1999). Turbidity of cultures grown to logarithmic phase was adjusted to obtain an inoculum density of 5 × 105 CFU ml−1, before supplementing the media with nalidixic acid (25 mg l−1), norfloxacin (1·5 mg l−1) or ciprofloxacin (0·625 mg l−1); concentrations were selected, based on the previously determined MIC values, to represent the lowest quinolone concentrations to inhibit bacterial growth at each pH. Antibiotic-free growth controls at each pH were also included. Cultures were continuously shaken on an orbital incubator at 37°C and samples removed after 0, 0·5, 1, 2, 3, 4 and 24 h of exposure (Haas et al. 2011). Viable bacteria were enumerated by preparing 10-fold serial dilutions in quarter-strength Ringer's solution with the problem of drug carry-over addressed by dilution, as previously described (Pankuch et al. 1994; Choi et al. 2011), and plating onto LSW agar, prepared as described by Walker et al. (1999). Bactericidal activity was defined as a ≥ 3-log10 reduction in viable count after 24-h contact with the antibacterial agent, with reference to the initial inoculum, and conversely, a ≤ 3-log10 reduction in viable count after 24-h contact was defined as bacteriostatic (Wayne 1999). Time-kill assays were repeated in triplicate, and results from one representative study are displayed in Fig. 2.

Figure 1.

Representative growth curves of Proteus mirabilis in MHB ranging from pH 5 to pH 10. Circles, squares, diamonds and triangles represent pH 5, pH 7, pH 9 and pH 10, respectively. Similar growth curves were generated by all three replicates at each pH value tested.

Figure 2.

Influence of pH on the time-kill kinetics of nalidixic acid (25 mg l−1) (a), norfloxacin (1.5 mg l−1) (b) and ciprofloxacin (0.625 mg l−1) (c) against Proteus mirabilis. The dashed line represents the limit of quantification. Circles, squares and diamonds represent pH 5, pH 7 and pH 9, respectively. Open and closed symbols refer to antibiotic present (added at time 0 h) and controls (without antibiotic), respectively. Each time-kill assay was repeated in triplicate with results from a representative study displayed.

Statistical analysis

The effect of pH on antibacterial activity of the three quinolone agents, nalidixic acid, norfloxacin and ciprofloxacin, was statistically evaluated by a one-way analysis of variance followed by Tukey's post hoc test for multiple comparisons between means of individual groups. A P value of <0·05 was used to denote significance.

Results

Bacterial susceptibility to quinolones

The effect of pH on susceptibility of the uropathogen Pr. mirabilis to nalidixic acid, norfloxacin and ciprofloxacin was examined by determination of MIC, MBC and MBEC values in MHB of pH 5, pH 7, pH 9 and pH 10, and results are shown in Table 1. An almost linear increase in relative potency, represented by a 10-fold decrease in MIC, of nalidixic acid was demonstrated upon reduction in media pH from 10 to 5. In contrast, MIC values of the two fluoroquinolones, norfloxacin and ciprofloxacin, increased 15- and 17-fold, respectively, upon lowering pH over the same range; this effect was most significant in acid pH where MIC values for norfloxacin and ciprofloxacin increased from 0·22 to 1·8 mg l−1 and from 0·08 to 0·66 mg l−1, respectively, between pH 7 and pH 5.

Table 1. Influence of pH on quinolone antibacterial activity against planktonic and biofilm-associated Proteus mirabilis ATCC 35508
pHNalidixic acid (mg l−1)Norfloxacin (mg l−1)Ciprofloxacin (mg l−1)
MICMBCMBECMICMBCMBECMICMBCMBEC
  1. a

    Reported as a bactericidal range. Above the concentrations reported in this range at pH 5 and pH 7, bactericidal activity of nalidixic acid was lost.

  2. b

    Biofilm eradication not observed at concentrations exceeding 150 mg l−1 norfloxacin at pH 10.

53.1325–50a>20001.807.19>2000.66250250
79.3837.5–150a>20000.220.441500.080.31125
926.0441.68>20000.070.071500.040.0425
1026.0483.33>20000.120.2312.5–150b0.040.0825

Variation in the MBC values of the three quinolones with pH followed the trends observed with the MIC values; however, no lethal activity against Pr.  mirabilis was displayed by concentrations of nalidixic acid exceeding 50 and 150 mg l−1 in MHB of pH 5 and pH 7, respectively. This is typical of the ‘paradoxical’ activity of nalidixic acid, as discussed previously. Conversely, this paradoxical effect was lost at pH 9 and pH 10, as nalidixic acid in these alkaline conditions effectively killed Pr. mirabilis at all tested concentrations higher than at the MBC. Acid pH had a more pronounced effect on the bactericidal activity of the two fluoroquinolones than on their inhibitory activity; the MBC values of ciprofloxacin increased by more than 800- and 6000-fold in pH 5 MHB compared with MHB of pH 7 and pH 9, respectively. No loss of the bactericidal activity of norfloxacin or ciprofloxacin against planktonic cells was observed when concentrations exceeding the MBC were tested in MHB ranging from pH 5 to pH 10.

The recalcitrance of Pr. mirabilis in the biofilm mode of growth, in comparison with the relative susceptibility of its planktonic counterpart, was demonstrated by the significantly higher MBEC values for the two fluoroquinolones in comparison with the MBC values, with the exception of the comparative susceptibilities to ciprofloxacin at pH 5. For example, at pH 9, biofilm-associated bacteria were more than 2000-fold less susceptible to the bactericidal activity of norfloxacin than their planktonic counterparts. Indeed, nalidixic acid in concentrations up to 2000 mg l−1 displayed no bactericidal activity against biofilm-associated cells at any pH. Furthermore, at pH 5, norfloxacin was ineffective against Pr. mirabilis biofilms when tested in concentrations up to 200 mg l−1. Again, susceptibility to the fluoroquinolones was higher at elevated pH; however, media pH had a less pronounced effect on biofilm eradication than killing of their planktonic counterparts; for example, the MBEC for ciprofloxacin decreased approx. 10-fold upon elevation of pH from 5 to 9, in comparison with the 6000-fold decrease in MBC observed between these two pH values. Despite a 12-fold reduction in MBEC for norfloxacin demonstrated upon elevation of pH from 9 to 10, no lethal activity towards Pr. mirabilis in the biofilm mode of growth was observed by concentrations of norfloxacin exceeding 150 mg l−1 in pH 10 MHB.

Bacterial growth and quinolone rates of kill

The effect of pH on growth of Pr. mirabilis was evaluated by plotting bacterial growth curves at pH 5, pH 7, pH 9 and pH 10. All variations in media pH reduced the rate of Pr. mirabilis growth relative to the optimal approximate generation time of 34 min in pH 7 MHB (Fig. 1). In pH 9 MHB, this reduction in growth rate was minimal; however, increasing alkalinity from pH 9 to pH 10 effectively doubled both the lag period and the generation time, with the latter increasing from 41 min in pH 9 MHB to 82 min in pH 10 MHB. After an extended lag phase in pH 5 MHB, the generation time of bacterial cells was 55 min, almost 2-fold longer than in pH 7 media.

Time-kill assays were performed using quinolone concentrations approximating the highest MIC observed against Pr. mirabilis at all pH values tested; this facilitated further evaluation of the effect of pH on quinolone bactericidal activity. Nalidixic acid at a concentration of 25 mg l−1 was bactericidal against Pr. mirabilis in MHB of pH 5 and pH 7 after 4-h and 3-h contact, respectively, as shown in Fig. 2(a). As expected from the determined MBC values, the concentration of nalidixic acid tested (25 mg l−1) was not sufficient for a bactericidal effect in pH 9 MHB, and transient antibacterial activity against Pr. mirabilis was displayed, such that after an initial lag in growth following the addition of nalidixic acid, a 1·5-log10 reduction in viable count was observed by 4 h. After 24 h, however, bacterial regrowth occurred to levels exceeding those of the starting inoculum. Norfloxacin (1·5 mg l−1) and ciprofloxacin (0·625 mg l−1) reached a ≥ 3 log10 reduction in viable count of Pr. mirabilis within 30 min in pH 7 and pH 9 MHB, with kill kinetics of norfloxacin again most rapid at pH 7 (Fig. 2b,c). Conversely, in pH 5 MHB, the concentrations of fluoroquinolones tested were sufficient to solely inhibit growth, with no bactericidal activity displayed.

Discussion

Infection by urease-producing Gram-negative pathogens subsequently raises urine pH (Stickler 2006), thus making this study of quinolone activity against one such representative uropathogen, Pr. mirabilis, in various pH environments of major therapeutic relevance (Teoh Chan 1985). This study presents the first report of the pH dependence of the ‘paradoxical’ activity characteristic of nalidixic acid. This biphasic response is widely reported during routine assessment of antibacterial activity in media of neutral pH (Crumplin 1975) and was observed herein in pH 5 and pH 7 MHB. It has previously been reported that a mutation in an unknown locus conferring a nalidixic acid-resistant RNA polymerase would be necessary to develop a strain susceptible to all concentrations of nalidixic acid exceeding the MBC (Crumplin and Smith 1976); however, this study demonstrates, for the first time, that this can instead be achieved at elevated pH. Importantly, bactericidal activity against planktonic Pr. mirabilis was retained by nalidixic acid concentrations up to 50-fold higher than the MIC when tested in media of alkaline pH, more closely replicating the in vivo environment during infection with this urease-producing pathogen.

The observed differences in antibacterial activity according to media pH may be related to the effect of pH on the ionization state of the quinolones, previously proposed to modulate quinolone accessibility to their intracellular target (Lemaire et al. 2011). Quinolones display concentration-dependent killing; however, these agents must first permeate the phospholipid bilayer membrane. Two alternative pathways exist; the relative contribution of each to the overall drug uptake process reported to correlate with quinolone hydrophobicity. Uncharged species in their neutral or zwitterionic state readily penetrate across this membrane through a lipid-mediated pathway, in contrast to moieties with a net charge, which instead can only cross the outer membrane through pore-forming proteins, preferentially via the OmpF porin in cells of Enterobacteriaceae (Chevalier et al. 2000). For example, reduced accumulation of the hydrophilic fluoroquinolone, norfloxacin, has previously been demonstrated in E. coli cells lacking OmpF porins, and conversely, greater efficacy of its hydrophobic predecessor nalidixic acid was reported in mutant cells with a disrupted outer lipopolysaccharide membrane (Hirai et al. 1986; Delcour 2009). Moreover, the pH-dependent ionic character of the quinolones results in significant variations in water wettability of these agents according to environmental pH (Babic 2007). Nalidixic acid possesses a single pKa value of 6·0 ± 0·5 (Babic 2007); therefore, intracellular accumulation is favoured in acidic media where this agent exists almost completely as its unionized moiety with a high hydrophobicity index (Chevalier et al. 2000), in contrast to its almost complete ionization, and consequent impaired transport, in alkaline media, thus explaining the 10-fold decrease in MIC observed between pH 10 and pH 5. Recently, the 15-fold increase in relative potency of delafloxacin, a weak acid investigational fluoroquinolone with similar acido-basic characteristics to nalidixic acid in terms of possession of a single pKa value of 5·4 and absence of an protonatable substituent, against Staphylococcus aureus upon reduction in pH from 7·4 to 5·5 was explained by a 10-fold increase in cellular accumulation observed between pH 7 to pH 6. In contrast, the 4-fold decrease in potency of one zwitterionic fluoroquinolone agent, moxifloxacin, upon the same reduction in pH was associated with a 2-fold decrease in accumulation (Lemaire et al. 2011). Similarly, significantly lower MIC and MBC values of the fluoroquinolones, norfloxacin and ciprofloxacin (pKa1, pKa2: 6·22, 8·38 and 6·42, 8·29 for both agents, respectively) were demonstrated against Pr. mirabilis in pH 7 MHB in comparison with pH 5 MHB, in accordance with previous reports (Bauernfeind 1983), due to their existence as zwitterions at neutral pH and their consequent efficient permeation of the lipid bilayer, in comparison with their subsequent inability to traverse the phospholipid membrane at pH 5, due to the net positive charge carried by approx. 90% of these agents at this acidic pH. This regulated entry of ionized species into the bacterial cell through porins, preventing quinolone accumulation in levels that have a secondary inhibitory effect on the two macromolecular biosynthetic processes, RNA and protein syntheses, may therefore explain the retention of bactericidal activity against planktonic Pr. mirabilis by all concentrations of nalidixic acid exceeding the MBC, selectively in alkaline conditions.

Despite the severely compromised bactericidal activity of all tested quinolones against Pr. mirabilis in the biofilm mode of growth, we give the first demonstration of enhanced fluoroquinolone-mediated biofilm eradication at elevated pH. The inefficacy of nalidixic acid for biofilm eradication, combined with the 100- to ≥2000-fold greater MBEC values than the corresponding planktonic MBC values of the fluoroquinolones, however, highlights the importance of preventing biofilm formation, for example with the use of drug-eluting medical devices. Numerous factors are involved in the recalcitrance of biofilm bacteria, including genotypic, metabolic and phenotypic changes, for example, the presence of persister bacterial cells (Lewis 2007), unique genetic adaptation (Adnan et al. 2010) and environmentally responsive growth rates (Bester et al. 2010). To target bacteria growing within biofilm communities, quinolones must first diffuse through the exopolysaccharide matrix at a rate dependent to some extent on ionic character (Stewart and Costerton 2001). One explanation for the enhanced biofilm eradication by the fluoroquinolones at elevated pH concerns the stronger interactions between these agents and the negatively charged matrix components, with consequential hindered biofilm penetration, when existing as their protonated moieties, at pH 5, than at alkaline pH when these agents carry a net negative charge. Furthermore, the lower reduction in MBEC values compared with the difference in planktonic MBC values observed upon increasing pH from 5 to 9 may, in part, be attributed to the chemical heterogeneity within biofilms and the generation of local microenvironments with significantly lower pH resulting from the accumulation of acidic waste products (Stewart and Franklin 2008), which thereby limits the influence of media pH on subsequent antibacterial potency.

Rapid bactericidal activity, characteristic of the quinolones (Wolfson 1985), was observed in this study, with the markedly improved potency and faster bactericidal activity of the fluoroquinolones, norfloxacin and ciprofloxacin, in comparison with the older agent, nalidixic acid, evident in Fig. 2. Furthermore, these time-kill assays demonstrate the additional importance of media pH on determining kinetics of bactericidal activity against planktonic organisms. Irrespective of the effect of pH on bacteriostatic and bactericidal activities of the three quinolones examined, kill kinetics for all three agents were most rapid at pH 7, the optimal pH for bacterial multiplication. The influence of bacterial growth rate on antimicrobial susceptibility is well established (McKenney and Allison 1997) and may in this case be explained by consideration of the mechanism of quinolone action: these agents inhibit DNA gyrase (Lewin 1991), the activity of which is proportional to metabolic activity, therefore maximizing effects of the quinolones when cell division was most rapid (McKenney and Allison 1997). For example, kill kinetics of nalidixic acid at a concentration of 25 mg l−1 against Pr. mirabilis were faster in pH 7 MHB, where activities of the biochemical targets were optimal, than in pH 5 MHB, despite the 3-fold higher MIC at pH 7.

This study demonstrates the critical importance of replicating environmental conditions during in vitro bacteriological testing to obtain a comprehensive understanding of therapeutic efficacy. Alkalinization of the urine by urease-producing bacteria may be regarded as a bacterial paradox, not only contributing to the pathogenicity of urinary infections due to crystalline biofilm formation, but also inherently increasing quinolone bactericidal activity, most likely due to changes in their degree of ionization and consequent cellular uptake, thus supporting their clinical use in the chemotherapy of urinary tract infections. For the first time, we report the exploitation of the Pr. mirabilis-induced pH changes for fluoroquinolone-mediated biofilm eradication, the predominant mode of bacterial growth in urinary catheter infections. Furthermore, we have highlighted the previously unreported bactericidal activity of nalidixic acid at high concentrations and an alkaline environment, with this elevated pH mirrored during in vivo infection with urease-producing uropathogens.

Acknowledgements

This work was funded by the Department for Employment and Learning, Northern Ireland. The authors declare that they have no conflict of interests in the research.

Ancillary