Correspondence: Bernard Turcotte, Department of Medicine, Room H5.74, McGill University Health Centre, 687 Pine Ave. West, Montréal, QC, Canada H3A 1A1. Tel.: 514 934 1934 ext. 35842; fax: 514 843 2819; e-mail: firstname.lastname@example.org
Rapidly increasing bacterial resistance to existing therapies creates an urgent need for the development of new antibacterials. Tirapazamine (TPZ, 3-amino-1,2,4-benzotriazine 1,4 dioxide) is a prodrug undergoing clinical trials for various types of cancers. In this study, we showed that TPZ has antibacterial activity, particularly at low oxygen levels. With Escherichia coli, TPZ was bactericidal under both aerobic and anaerobic conditions. Escherichia coli mutants deficient in homologous recombination were hypersusceptible to TPZ, suggesting that drug toxicity may be due to DNA damage. Moreover, E. coli strains deleted for genes encoding putative reductases were resistant to TPZ, implying that these enzymes are responsible for conversion of the prodrug to a toxic compound. Fluoroquinolone-resistant E. coli strains were as susceptible to TPZ as a wild-type strain. Methicillin-resistant Staphylococcus aureus strains were also susceptible to TPZ (MIC = 0.5 μg mL−1), as were pathogenic strains of Clostridium difficile (MIC = 7.5 ng mL−1). TPZ may merit additional study as a broad-spectrum antibacterial, particularly for anaerobes.
An increasing prevalence of antimicrobial resistance (Bancroft, 2007; Boucher et al., 2009) and decreased production of new antibiotics (Payne et al., 2007; Cooper & Shlaes, 2011) is making treatment of many medical conditions more difficult. Antimicrobial resistance is observed in various bacterial species. For example, one of the important pathogens is Escherichia coli, which affects hundreds and millions of humans annually (Da Silva & Mendonca, 2012). Escherichia coli readily acquires resistance to the antibacterial fluoroquinolones, which increases problems treating virulent strains (Da Silva & Mendonca, 2012). Another important pathogen is methicillin-resistant Staphylococcus aureus (MRSA), which has been proposed to be considered as a national priority for disease control in the United States (Klein et al., 2007). Likewise, the severity of Clostridium difficile infections has increased markedly in the last decade, resulting in c. 15 000 to 20 000 annual deaths in the United States (Rupnik et al., 2009). These and other drug-resistant pathogens make finding new antimicrobials a priority. Anticancer agents constitute a potential source of antibacterials, as they have already undergone extensive testing for bioavailability and pharmacokinetics. One such agent is TPZ (tirapazamine or 3-amino-1,2,4-benzotriazine 1,4 dioxide or SR4233; see Fig. 1a for chemical structure). TPZ is an anticancer prodrug that is more active at low oxygen levels, a feature used to target cancer cells in the centre of a tumour where oxygen levels are generally low due to poor vascularization (Brown & Wilson, 2004). TPZ has been used in Phase II and III clinical trials in combination with radiotherapy or other anticancer drugs such as cisplatin (von Pawel et al., 2000; Marcu et al., 2006; Ghatage & Sabagh, 2012). Whether TPZ could be a broad-spectrum antibacterial is unknown.
The hypoxic toxicity of TPZ is due to the enzymatic addition of one electron to TPZ by reductases, a process yielding a radical species that causes single- and double-strand DNA breaks, chromosome damage and cell death (Patterson et al., 1998; Denny & Wilson, 2000). The radical species is unstable, and at normal oxygen levels, it reacts with oxygen to restore TPZ and a much less toxic radical species (Patterson et al., 1998; Denny & Wilson, 2000). However, the exact mechanism of action of TPZ is not known, and the identity of the molecule that causes the oxidative damage is controversial (Patterson et al., 1998; Denny & Wilson, 2000). We have previously shown that TPZ has antifungal activity, as tested with budding yeast (Hellauer et al., 2005). However, the effect of TPZ in prokaryotes has not been documented.
In the present study, we report that TPZ has antimicrobial activity. Experiments performed with E. coli show that TPZ is bactericidal and is more active under anaerobic conditions. Other data, based on genetic experiments, suggest that TPZ causes DNA damage and that reductases are responsible for the conversion of TPZ to a toxic compound. Importantly, we show that TPZ is highly active against MRSA strains as well as various C. difficile strains.
Materials and methods
Strains and culture conditions
Escherichia coli strains used in this study are listed in Tables S1 and S4, while S. aureus strains are listed in Table S2. For C. difficile, clinical isolates were provided by the Montreal General Hospital (Montréal, Québec) and were originally obtained from the Center for Disease Control and Prevention (Atlanta, USA). Bacterial strains were cultured in LB liquid medium and on LB agar plates at 37 °C. For generating anaerobic conditions, the BBL GasPak™ 100 system (BD diagnostics) or the AnaeroGen™ system (Oxoid) were used. TPZ was obtained from Sanofi-Aventis, and norfloxacin was obtained from Sigma (St. Louis, MO).
MIC determination of gyrase mutants
Fluoroquinolone-resistant strains (German et al., 2008) were grown overnight in LB medium. The overnight culture was re-inoculated at OD600 0.002, and the culture was grown to log phase (OD600 of 0.1). The culture was then diluted to OD600 0.002 in 2 mL LB with appropriate concentration of TPZ or norfloxacin. Cultures were incubated for 16–24 h, and the results were obtained by visual inspection of growth. Strain ATCC 25922™ was used as an internal (quality) control.
Assay with C. difficile
Clinical and Laboratory Standards Institute guidelines were followed to determine susceptibility of C. difficile strains to TPZ (Institute, 2007). Reference agar dilution procedure (Wadsworth method) was used to test the susceptibility to TPZ of eight different C. difficile strains (CLSI, 2007). Twofold serial dilution series of 128 μg mL−1 to 0.0586 ng mL−1 were tested. A single colony of the strains was inoculated overnight in freshly made Schaedler anaerobe broth (Oxoid, pH 7.6 ± 0.2) from streaked plates (premade plates of Columbia blood agar, 5% sheep blood, Oxoid). Using Wilkins-Chalgren Anaerobe broth (Oxoid), 0.5 MacFarland (MF) turbidity standard suspension of overnight cultures was prepared. A Steer's replicator or multiple inoculator (Cathra Automated Inc., St. Paul, MN) was used to apply the culture to solid medium (Rousseau & Harbec, 1987). The eight C. difficile strains were spotted in replicates (total 16 spots) on the same plate to reduce plate to plate variability. 100 μL of 105 dilution of the 0.5 MF (McFarland standard) was plated on brucella blood agar plates for CFU counts. Appropriate dilutions for CFUs were made in Schaedler broth.
Wild-type E. coli BW25113 was grown to log phase and treated with 50 μM TPZ (8.9 μg mL−1) (4X MIC). Aliquots were taken at time 0, 2, 6, 24 h. The dilutions were made in 0.9% saline solution, and cells were plated on drug-free LB agar plates. Three different dilutions were plated for each time point, and the average number of CFUs was determined for both the TPZ-exposed strains as well as the control. The wild-type strain was similarly treated with of 10 μM (1.8 μg mL−1) TPZ under anaerobic conditions.
TPZ has antibacterial activity
To assess the antibacterial activity of TPZ, wild-type E. coli strain BW25113 was grown overnight, and cells were then serially diluted and spotted on plates containing various concentrations of TPZ. Colony formation was monitored after an overnight incubation at 37 °C (Fig. 1b). From a starting culture of 8 × 105 CFU mL−1, TPZ greatly reduced colony formation at 8 μM (1.4 μg mL−1), while no colony was seen with 10 μM TPZ (1.8 μg mL). Remarkably, under anaerobic conditions, the concentration of TPZ required to prevent colony formation was c. 25 times lower (0.375 μM TPZ, 67 ng mL−1) than under aerobic conditions.
Susceptibility of fluoroquinolone-resistant strains to TPZ
The antibacterial property of TPZ prompted us to investigate its potency with antibiotic-resistant strains. As studies in human cells and budding yeast suggest that TPZ targets topoisomerase II (Peters & Brown, 2002; Hellauer et al., 2005), we focused on E. coli strains that carry fluoroquinolone-resistance mutations in the quinolone resistance-determining region (QRDR) of the gyrA or gyrB genes that encode DNA gyrase. Results were represented as ratios of mutant MIC to wild-type MIC to compare the relative susceptibility of the strains to norfloxacin and TPZ (Fig. 2). As expected (German et al., 2008), the gyrase mutant strains gave high values of MIC ratios for the fluoroquinolone norfloxacin. For example, mutants S83L and S83W, exhibiting ratios around 10, showed low susceptibility to norfloxacin, while mutants Q106H and D426N showed moderate susceptibility to the antibiotic. In contrast, when the same set of strains was tested with TPZ, most MIC ratios were close to one, showing that the mutants are as susceptible as the wild-type strain. Moreover, the GyrA G81C strain showed greater susceptibility than wild-type cells to TPZ. In conclusion, gyrase mutant strains that are resistant to fluoroquinolone are susceptible to TPZ. These results suggest that TPZ does not inhibit gyrase activity or, alternatively, it may act via a mechanism distinct from that of norfloxacin (see 'Discussion').
Bactericidal activity of TPZ
A time-kill assay was performed under aerobic and anaerobic conditions to assess the bactericidal activity of TPZ. A sharp decrease in viable cells was observed when bacterial cells were exposed to TPZ (Fig. 3). Under aerobic conditions and with TPZ at a concentration of 50 μM (8.9 μg mL−1), a marked decrease in survival (> 8 logs) was observed after a 4-h incubation (Fig. 3a). Culture regrowth was observed after 8 h. Growing cells collected after 8 h of TPZ treatment were subjected to a second time-kill assay that showed no resistance to TPZ (data not shown). Under anaerobic conditions, the shape of the kill curve obtained with 10 μM TPZ (1.8 μg mL−1) was similar to that obtained aerobically with 50 μM TPZ (8.9 μg mL−1) (Fig. 3b). We observed a difference of 106 CFU after 24 h between control media and TPZ-containing media under aerobic and anaerobic conditions. In summary, TPZ is bactericidal under both aerobic and anaerobic conditions.
TPZ hypersusceptibility from deletion of genes involved in recombination repair
Recombination is one mechanism by which E. coli can repair DNA damage. To determine whether DNA damage could provide a basis for the bactericidal activity of TPZ, we tested nine nonessential E. coli genes involved in recombination repair for TPZ susceptibility (Table S3). Strikingly, ∆recA, ∆recB, ∆recC, ∆recF, ∆recO and ∆recR strains were hypersusceptible to 7 μM TPZ (1.2 μg mL−1) under aerobic conditions, while colony formation by the wild-type strain was similar in the presence or absence of TPZ (Fig. 4a). However, with a lower concentration of TPZ (5 μM), only mutants recA, recB and recC were susceptible to TPZ. Similar results were obtained under anaerobic conditions except that, as expected, susceptibility was observed at a lower concentration of TPZ (data not shown). Since the rec genes listed above are all involved in DNA repair, the results suggest that TPZ is most likely involved in damaging DNA, either directly or indirectly (see 'Discussion').
Bacterial reductase mutants are resistant to TPZ
In eukaryotes, TPZ is activated by cytochrome P450 reductases. To determine whether reductases perform a similar function in E. coli or not, 198 known or putative reductases were identified in the E. coli genome (Materials and Methods and Table S4). The potential role of these reductases in mediating TPZ toxicity was determined by testing 198 strains carrying deletions of genes encoding reductases against a wide range of TPZ concentrations under aerobic and anaerobic conditions (data not shown). We observed that multiple reductases are potentially involved in prodrug activation. ArgC is a critical reductase for TPZ activation under both aerobic and anaerobic conditions (Fig. 4b and Table S3). In addition, YdhV, RsxD and YeiA also play a role in enhancing TPZ toxicity under both aerobic and anaerobic conditions. Finally, other gene products, such as YdbK, ProA, DmsB, NarG and YjhC, are also involved in lowering TPZ susceptibility, but to a lesser extent than ArgC (Table S3).
Susceptibility of MRSA strains to TPZ
Both methicillin-sensitive (MSSA) and methicillin-resistant clinical isolates of S. aureus were tested using a spotting assay at various TPZ concentrations under hypoxic conditions (Fig. 5). The various strains showed similar colony formation in the absence of TPZ (Fig. 5, left panels); however, the inhibition of colony formation was evident at a TPZ concentration as low as 2 μM (0.36 μg mL−1) (Fig. 5, middle panels). There was a difference in viability among strain isolates, although we do not have genotypic information on individual strains to analyse them further. Interestingly, a slight increase in concentration to 3 μM TPZ (0.53 μg mL−1) completely inhibited growth of both MSSA and MRSA strains (Fig. 5, right panels). Colony formation was also measured under aerobic conditions. Staphylococcus aureus growth was inhibited with 100 μM TPZ or 17.8 μg mL−1 (data not shown). In summary, MRSA strains are susceptible to TPZ under both aerobic and anaerobic conditions, but the concentration of TPZ required to prevent S. aureus colony formation was 50 times lower under anaerobic conditions.
Susceptibility of C. difficile to TPZ
Eight clinical isolates of C. difficile were tested for susceptibility to TPZ by a spotting method described in 'Materials and methods'. The experiment was performed in duplicate with the following control: anaerobic growth in the absence of drug, aerobic growth and anaerobic growth with 4–10 μg mL−1 ampicillin. Clostridium difficile strains were susceptible at concentrations of TPZ that were 1000 times lower than needed for ampicillin. The NAP2 strain was the most susceptible to TPZ at 3.75 ng mL−1 (21 nM), while strain NAP5 was the least susceptible to TPZ; it showed some growth at 15 ng mL−1. Interestingly, all strains were susceptible with 30 ng mL−1 (0.17 μM) TPZ.
TPZ is an anticancer drug that is currently in phase II/III clinical trials. In this study, we have shown that TPZ has antibacterial activity. For example, E. coli was susceptible to TPZ at 10 μM (1.8 μg mL−1) under aerobic conditions (Fig. 1b). Remarkably, the effect of TPZ was much more pronounced under anaerobic conditions (0.375 μM, 67 ng mL−1), in analogy to animal cells that show increased susceptibility to TPZ under hypoxic conditions. In animal cells, there is good evidence that one electron addition to TPZ by reductases yields a radical species that causes single- and double-strand DNA breaks. In animal cells, homologous recombination is the major pathway for repair of DNA damage caused by TPZ (Evans et al., 2008). We have previously performed a genome-wide analysis of genes that alter susceptibility to TPZ in the yeast Saccharomyces cerevisiae (Hellauer et al., 2005). Results show that many genes involved in homologous recombination, the major yeast pathway for repairing double-strand breaks, are involved in protecting cells against TPZ.
There are some similarities between TPZ and metronidazole, a drug currently used against anaerobic pathogens such as C. difficile (Reysset, 1996; Lofmark et al., 2010). Both drugs show preferential activity under hypoxia; they are activated by enzymatic reduction and target DNA. So, what is the mode of action of TPZ in E. coli? Strains carrying deletions of genes involved in DNA repair were hypersusceptible to TPZ. For example, deletion of recA, recB or recC almost completely abolished colony formation, as tested with 5 μM TPZ (0.89 μg mL−1) under aerobic conditions (Fig. 4a). RecA is an essential component of the machinery involved in DNA repair via homologous recombination. Similarly, recB and recC encode components of the RecBCD complex, which is also involved in DNA repair. In contrast to the genes listed above, deletion of recF, recO or recR did not result in susceptibility to 5 μM TPZ. This observation can be explained by the fact that RecF, RecO and RecR are accessory proteins involved in RecA DNA loading (Kowalczykowski, 2000). Deletion of recD, recN or recJ resulted in moderate susceptibility to TPZ (data not shown). This observation is most likely due to the fact that these genes have overlapping functions (Lloyd & Buckman, 1991). Indeed, susceptibility to UV light is moderate with double mutants, while a triple mutant (∆recD ∆recN ∆recJ) is hypersusceptible (Lloyd & Buckman, 1991). Thus, the extent of DNA repair deficiency correlates with the susceptibility of E. coli to TPZ.
Another similarity between animal cells and E. coli is the requirement of reductases for TPZ toxicity. A number of mammalian reductases (e.g. NADPH:cytochrome P450 reductase, cytochrome P450, DT-diaphorase) have the capacity to metabolize TPZ in vitro (Patterson et al., 1998). Moreover, there is also evidence that nuclear reductases could also use TPZ as a substrate (Delahoussaye et al., 2001). Thus, multiple reductases can metabolize TPZ in animal cells. Our genetic analysis performed with E. coli suggests that this is also true for this bacterial species (Fig. 4b and Table S3). For example, deletion of argC greatly increases resistance to TPZ under aerobic and anaerobic conditions. ArgC catalyses the third step in the conversion of glutamate to arginine, specifically the NADPH-dependent reduction of N-acetylglutamyl phosphate. Thus, even though we do not have biochemical evidence for enzymatic reduction of TPZ by ArgC, it is likely that this enzyme performs such a function given its known reductase activity. A similar mechanism may also apply to other reductases that were identified in our genetic analysis (Table S3).
There is evidence that TPZ is an inhibitor of topoisomerase II in animal cells, while in budding yeast, overexpression of TOP2 (encoding topoisomerase II) results in hypersusceptibility to TPZ. Thus, TPZ could target DNA gyrase in E. coli. However, we were unable to demonstrate any clear inhibitory effect of TPZ on topoisomerase activity, as assayed in vivo and in vitro (Z. Shah, unpublished data). The observation that strains carrying mutations in the gyrA or the gyrB genes are resistant to fluoroquinolones, but not TPZ, correlates with the lack of evidence for TPZ acting as a gyrase inhibitor (Fig. 2). Thus, the mode of action of TPZ in bacteria may involve direct DNA damage.
Our study also showed that TPZ is active in S. aureus with both MSSA and MRSA strains. Resistance in MRSA strains is due to the mecA gene, which encodes PBP2A, a protein that has drastically reduced affinity to penicillin. Thus, it is not surprising that MRSA strains show susceptibility to TPZ at levels comparable to MSSA strains, as the mecA gene is not related to DNA damage. Importantly, our results also show that TPZ is highly active against various C. difficile strains (Fig. 6). For example, no colony formation was observed at 7.5 ng mL−1 for some strains where all strains tested were susceptible to 30 ng mL−1 TPZ.
What is the potential of TPZ as an antibiotic? As TPZ has been tested for cancer treatment in phase II/III clinical trials, its side effects and toxicity in humans are well documented (Graham et al., 1997; Senan et al., 1997; Denny & Wilson, 2000). The maximum tolerated dose of TPZ is 390 mg m−2 (Graham et al., 1997), while the dose used for one clinical trial was 220–300 mg m−2 (Le et al., 2006). We estimate that a dose of 300 mg m−2 corresponds to 65 μM TPZ (11.6 μg mL−1), a concentration well above (3 logs) the one required to prevent C. difficile colony formation in vitro (Fig. 6). It should be noted that experiments performed with E. coli under aerobic conditions (e.g. Fig. 1b) most probably underestimate TPZ toxicity as oxygen concentration in human tissues is approximately three times lower than that in the atmosphere (Brown & Wilson, 2004). In addition, because of its relatively simple chemical structure (Fig. 1a), TPZ can be synthesized at low cost. Moreover, various TPZ derivatives have been generated and could be easily tested to establish structure–activity relationships (Hay et al., 2007, 2008; Xia et al., 2011).
Antimicrobials for general use require a different safety profile than anticancer drugs and, as a result, it remains to be proven that TPZ would be a safe antimicrobial agent. However, there is a pressing need for new molecules to be used as a ‘last-resort’ to treat lethal multidrug resistant bacterial infections. In this respect, TPZ could be used in terminal cases where its beneficial effect would clearly outweigh any potential side effect. Thus, TPZ may represent an alternative to existing antibiotics, especially for treatment of anaerobic bacterial pathogens.
We are extremely grateful to Dr. Karl Drlica for critical review of the manuscript and for providing E. coli strains. We also thank Drs. Louis-Charles Fortier and Martine Raymond for review of the manuscript. We thank Sarnofi-Aventis for providing TPZ and the National BioResource Project (E. coli; NIG, Japan) for providing the Keio collection. Staphylococcus aureus strains were provided by Dr. Brigitte Lefebvre. We thank Dr. Vivian Loo and Susan Fenn for providing their expertise for experiments involving C. difficile and the personnel of the Laboratory of Clinical Microbiology (Royal Victoria Hospital) for advice. We also thank Dr. Marc Drolet and Valentine Usongo for help in performing gyrase assays. This work was supported by grants from CIHR (Grant # MOP-84231) and NSERC (Grant #184053-04).