Trichomonas vaginalis causes an estimated 170 million human cases of trichomoniasis annually, making it the most prevalent, non-viral sexually transmitted infection (STI) worldwide (Harp and Chowdhury, 2011). The complex epidemiology of trichomoniasis is due to high rates of unaccounted asymptomatic cases and poor surveillance of symptomatic cases (Workowski et al., 2021). This is problematic as untreated trichomoniasis may lead to reproductive tract sequelae in women, including adverse pregnancy outcomes and increased acquisition/transmission of other reproductive tract infections including HIV (Cotch et al., 1997, Kissinger, 2015). Current chemotherapeutic interventions rely on the flagship of the 5-nitroimidazole (5NI) class of compounds, metronidazole, the standard option since the 1950s (Dunne et al., 2003, Kissinger, 2015). However, 5NI resistance can be expected in 5 – 10% of symptomatic cases, particularly in females from lower socioeconomic tiers (Cudmore et al., 2004, Bosserman et al., 2011). Similarly, 5NI resistance is recognised but likely underestimated in the related, neglected veterinary pathogen Tritrichomonas foetus (Gookin et al., 2017), which causes an analogous STI in cattle and a gastrointestinal disease in felines. Notably, when the standard metronidazole treatment for resistance-presumed trichomoniasis fails, the only alternatives are: i) receiving a higher, often toxic dose of metronidazole (Cudmore et al., 2004), or ii) switching to a related 5NI, tinidazole, which is not easily accessible to all patients due to costs and availability within countries (Johnson, 2009).

Currently, there are no standards for definitive identification of 5NI resistance in trichomonads, beyond its presumptive link with treatment failure. At the molecular level, 5NI resistance is linked to changes in parasite antioxidant and metabolic pathways (Kulda, 1999). Antioxidant and metabolic enzymes such as NADH oxidase and Pyruvate Ferredoxin Oxidoreductase (PFOR) reduce 5NIs to toxic metabolites, and changes in these enzymes are linked to distinct 5NI resistance phenotypes under oxygenated and anaerobic conditions (Kulda, 1999, Graves et al., 2020). Oxygen-dependent (“aerobic”) 5NI resistance arises from altered oxygen-scavenging pathways, where the decreased activity of key antioxidant enzymes leads to increased intracellular oxygen levels, oxidising the toxic 5NI metabolites back to the non-toxic prodrug (Graves et al., 2020). In contrast, oxygen-independent (“anaerobic”) 5NI resistance is characterised by the parasites’ inability to reduce 5NIs, through decreased or complete loss of activity of 5NI-activiating enzymes (Kulda et al., 1993). As such, anaerobically resistant trichomonads are expected to be resistant to 5NIs under both anaerobic and oxygenated conditions, whereas oxygen-dependent 5NI resistance is affected by oxygen levels.

Due to these two resistance phenotypes and their dependencies, identifying 5NI resistance in trichomonads has relied on both oxygenated and anaerobic evaluation. However, current drug susceptibility platforms for trichomonads are growth-based, anaerobic systems (Upcroft and Upcroft, 2001, Natto et al., 2012), and have not been adequately evaluated or optimised under other atmospheric conditions (i.e. oxygenated formats). Notably, excess oxygen in current “aerobic assays” hinders microaerophilic metabolism and compromises growth in vitro (Upcroft and Upcroft, 2001). Consequently, current forms of oxygenated drug susceptibility testing (performed with excess oxygen in the system) are not comparable to the growth-based, anaerobic formats with which they are often run in parallel (Upcroft and Upcroft, 2001). Further, aerobic suppression of trichomonad growth means current aerobic assays often fail to meet the minimum metabolic utilisation that anaerobic quantitative assays rely upon for outputs (e.g., change in absorbance). Hence, current aerobic assays report qualitative ‘by-eye’ estimates of minimum inhibitory concentrations (MICs). Consequently, these aerobic systems cannot be automated or robustly evaluated by quantitative approaches.

Identification of 5NI resistance and quantification of isolate resistance levels requires standardised and higher-throughput platforms for both oxygen-dependent and oxygen-independent 5NI resistance. Current aerobic assays used to interrogate oxygen-dependent 5NI resistance is low-throughput and detrimental to trichomonad growth in vitro. We developed a microaerophilic growth assay that is capable of interrogating oxygen-dependent 5NI resistance, which can be run in complement with our previous anaerobic growth assay (Lam et al., 2021). Together, both these growth-based assays appropriately reflect the physiological environment these microaerophilic parasites reside in (i.e. microaerophilic to anaerobic conditions within the gastrointestinal and urogenital tract) (Zheng et al., 2015, Ng et al., 2018, Alessandri et al., 2020), whilst eliminating the traditional drawbacks of aerobic assays (i.e. excess oxygen impeding growth). We propose running these growth-based, physiologically relevant assays in parallel to quantify both 5NI resistance phenotypes in these trichomonads. Subsequently, we conducted drug susceptibility screens against 5NI-susceptible and 5NI-resistant trichomonad isolates (associated with clinical treatment failure) from two species (T. vaginalis (Voolmann and Boreham, 1993) and T. foetus (Rush and Šlapeta, 2021)) (Supplementary Table S1).

We first explored the impacts of atmospheric conditions on trichomonad growth in laboratory-standard 5NI-susceptible isolates G3 (T. vaginalis) and KV1 (T. foetus). All isolates were cultured as previously described (Lam et al., 2021). Trophozoite growth was evaluated in 6-well plates containing the pH indicator phenol red (Sigma Aldrich, Australia) in sealed bags (Supplementary Fig. S1) containing either anaerobic sachets (BD:260683 from Becton Dickinson, Australia, or Oxoid:AN0010W from Thermofisher, Australia), microaerophilic sachets (BD:260685 from Becton Dickinson, Australia, or Oxoid:CN0020C from Thermofisher, Australia) or no sachets (aerobic; pouches unsealed). Trophozoite growth was evaluated at different seeding densities (Supplementary Fig. S1, Supplementary Table S2), taking into consideration previous reports (Upcroft and Upcroft, 2001), which required seeding 105 cells/well in aerobic assays. Absorbance at 570 nm, cell number and pH were evaluated as previously described at 24, 48 and 72 h (Lam et al., 2021) (Supplementary Methods S1). The number of trophozoite generations was inferred from initial seeding densities using an exponential (base 2) function.

Overall, cell counts demonstrated that higher seeding densities under all atmospheric conditions support fewer total trichomonad generations (Fig. 1A). Tritrichomonas foetus growth was less impacted by oxygen, irrespective of initial seeding densities (as noted by a recovery of cell counts by 48 h), but T. vaginalis growth was considerably impacted in aerobic conditions (Fig. 1A). Microaerophilic growth was comparable to growth under anaerobic conditions for both trichomonads (Fig. 1A), demonstrating that oxygen levels in microaerophilic conditions were tolerable, but excess oxygen was detrimental to growth (particularly T. vaginalis). Further, we demonstrated that microaerophilic growth was quantifiable, with the pH decrease (as a proxy for growth) under both microaerophilic and anaerobic conditions quantifiable using phenol red (a pH indicator) and measuring absorbance at 570 nm (Fig. 1B, Supplementary Fig. S2) (Lam et al., 2021). In contrast, whilst aerobic growth with phenol red correlated with a similar but smaller change in absorbance (Fig. 1B), we showed that the initial growth rate is hindered, particularly in T. vaginalis, compared with microaerophilic or anaerobic conditions (Fig. 1A). We also observed impaired growth in T. vaginalis led to clumped and less motile trophozoites when grown aerobically (data not shown); when T. vaginalis was grown aerobically in 96-well plates, we observed no motile trophozoites after incubation and no colour change when grown with phenol red (data not shown). Consequently, we propose that pure aerobic conditions cannot support a level of trichomonad growth capable of producing a sufficient and reliable shift in a reporter (i.e., a large enough dynamic range using phenol red), whereas microaerophilic conditions had comparable levels of growth to anaerobic formats. We therefore reason microaerophilic conditions will provide sufficient oxygen in the system to allow detection of oxygen-dependent 5NI resistance whilst also allowing sufficient trichomonad growth for a quantifiable reporter readout.

We then tested whether microaerophilic oxygen levels were sufficient to demonstrate oxygen-dependent 5NI resistance by drug susceptibility testing. Trichomonads were grown using either the microaerophilic pouch systems (as above), limited to one 96-well plate per sachet, or grown in the CampyGen 2.5 L jar (Oxoid:CN0025A from Thermofisher, Australia), limited to three 96-well plates per sachet; previously standardised TriTOX anaerobic growth assays (Lam et al., 2021) were conducted in parallel to microaerophilic growth assays. In each well, 10,000 trichomonads were seeded in 50 µL volumes (200 trophozoites/µL), containing 45 µL of complete TYI-S-33 media, and 5 µL of phenol red stock (1% w/v solvated in type II water). Drug susceptibility testing was done in known clinically resistant and susceptible trichomonads (Supplementary Table S1), and one newly axenised T. vaginalis isolate with suspected metronidazole resistance, as it had been isolated from a patient with chronic trichomoniasis who had repeatedly failed metronidazole treatment (Supplementary Methods S1).

For drug susceptibility testing under both microaerophilic and anaerobic conditions, we selected a panel of 5NIs and a nitrofuran including metronidazole (Mtz, Sigma Aldrich), tinidazole (Tnz), ronidazole (Rnz), ornidazole (Orn), secnidazole (Sec) and furazolidone (Furz, a nitrofuran), with two structurally unrelated compounds, the microbial natural product trichostatin A (TSA) and the synthetic benzimidazole mebendazole (Meb), which both have distinct, oxygen-independent mechanisms of actions as positive controls. Compounds were screened in a 10- to 12-point, two-fold dilution series in a minimum of technical and biological duplicates. Absorbances at 570 nm at 48 h were used to plot compound dose-response curves (Supplementary Fig. S3) to acquire 50% inhibitory concentrations (IC50s), and Z′ was calculated as a measure of assay performance (Zhang et al., 1999); assays were rerun if Z′ < 0.4. Resistance factors (RFs) for each clinical isolate was calculated by dividing the mean IC50 against the IC50 of a known 5NI-susceptible isolate as a proxy for drug resistance levels (Supplementary Methods S1), where this fold change represents the change in dosage required to kill 50% of the resistant trichomonad population. In this study, we use RF = 2 as a threshold for defining resistance in both anaerobic and microaerophilic conditions.

Our two structurally independent positive controls (TSA and Meb) had similar IC50s for all T. vaginalis isolates under all atmospheric conditions (Supplementary Tables S3 and S4). In T. foetus, increased IC50s were calculated for both compounds under microaerophilic conditions compared with anaerobic conditions. However, these were comparable between resistant and susceptible lines, with an RF ∼1 observed for TSA and Meb in both species under both atmospheric conditions (Fig. 2). Regardless, the RF for these clinical trichomonad isolates against TSA and Meb under both atmospheric conditions suggests the apparent resistance developed in these isolates are likely to be specifically against 5NIs compounds with an oxygen-dependent mode of action (MOA) (Kulda, 1999).

Both clinically obtained T. vaginalis isolates demonstrated Mtz resistance under anaerobic and microaerophilic conditions relative to the Mtz-susceptible G3 isolate (Fig. 2A). The RF for Mtz was 5.70 and 7.41 for C10 under microaerophilic and anaerobic conditions, respectively (Table 1, Fig. 2A); whereas the RF for Mtz was 10.6 and 11.8 for B7268 under microaerophilic and anaerobic conditions, respectively (Table 1, Fig. 2A). These in vitro results concurred with clinical observations, and their original isolation from patients who presented with chronic trichomoniasis despite Mtz treatment, eventually cured by high and extended doses of Tnz (Voolmann and Boreham, 1993). The B7268 isolate remained susceptible to Tnz under anaerobic conditions whilst the C10 isolate had an RF above our threshold for oxygen-independent Tnz resistance (Table 1, Fig. 2Aa), which is consistent with the resolution of trichomoniasis in both patients upon prescription of increased Tnz dosage (Voolmann and Boreham, 1993). Interestingly, B7268 remained resistant to Tnz under microaerophilic conditions with RF = 9.24, whereas C10 did not demonstrate oxygen-dependent Tnz resistance in our microaerophilic assay with RF = 1.76 (Table 1, Fig. 2Ab). Further testing of susceptibility to other 5NIs demonstrated that these two clinical isolates appeared to be cross-resistant, namely: Rnz, Orn and Sec under both atmospheric conditions, with particularly high RFs for B7268 against Sec (Fig. 2A). Overall, these results indicate that both isolates, broadly termed as Mtz-resistant, in fact had more complex and divergent 5NI resistance phenotypes which correlated with their clinical presentation and treatment successes/failures.

Whilst our clinical T. vaginalis isolates demonstrated broad 5NI resistance and both oxygen-dependent and oxygen-independent Mtz resistance (Fig. 2A), our clinical T. foetus isolate remained susceptible to most of our 5NIs tested under anaerobic conditions but resistant under microaerophilic conditions (Table 1, Fig. 2B). Notably, this T. foetus clinical isolate displayed Mtz and Sec resistance under anaerobic conditions, with RFs = 3.09 for Mtz and RF = 4.03 for Sec (Table 1, Fig. 2Ba). Under microaerophilic conditions, however, this clinical T. foetus isolate appeared resistant across the 5NI panel, with RFs consistently above 7 for all 5NIs tested (Table 1, Fig. 2Bb). Importantly, Mtz and Rnz had an ∼10-fold decrease in potency under microaerophilic conditions in this isolate (Fig. 2Bb), a quantifiable decrease in drug potency compared with previous minimum inhibitory/lethal concentration estimates (Rush and Šlapeta, 2021). Further, the oxygen-dependent resistance to Mtz and Rnz appears to correlate with 5NI cross-resistance in a microaerophilic environment, displaying similar RFs of ∼10 among other 5NIs: Tnz, Orn and Sec (Table 1, Fig. 2Bb).

Together, our findings demonstrate the importance of a standardised approach for experiments which co-define oxygen-dependent and oxygen-independent 5NI resistance, and the importance of an assay platform that can provide quantitative and comparable results within and between atmospheric conditions. Notably, our microaerophilic format can be used to benchmark minimum standards of diagnosing oxygen-dependent trichomonad 5NI resistance. With no fixed phenotypic standards for 5NI resistance in trichomonads, studies continue to report a wide range of MIC estimates under aerobic and/or anaerobic conditions as indicative of oxygen-dependent and oxygen-independent 5NI resistance, respectively (Marques-Silva et al., 2021), some without adequate benchmarking to a comparable control. We propose that derived IC50s or MICs from clinical isolates under anaerobic or microaerophilic conditions must be compared with a relevant, 5NI-susceptible control, particularly as our work demonstrates that microaerophilic IC50s/MICs of 5NIs are higher than anaerobic IC50s/MICs by up to 20-fold (Supplementary Tables S3-S6). Thus, drug susceptibility testing of clinical trichomonad isolates under both microaerophilic and anaerobic conditions must be conducted in parallel, and with a 5NI-susceptible isolate to provide a ratio of IC50s (RFs compared with a known-susceptible isolate) for an accurate assessment of 5NI resistance. These quantifiable results can further inform 5NI resistance levels; in this work, we use RF = 2 as a threshold for resistance in either anaerobic or microaerophilic formats, a similar approach as described by Lossick et al. (1986). However, while still a valuable baseline, our RF = 2 cut-off should only be considered as a suggestion presently, where this baseline will be subject to change as more reliable, quantitative 5NI susceptibility testing data is accumulated within this field, henceforth providing more statistical power for further analyses to better define 5NI resistance in trichomonads. Moreover, these cell-based anaerobic/microaerophilic assays remain only the first step in characterising trichomonad 5NI resistance, with other molecular or genetic tools urgently required to validate such phenotypic assessments.

To date, only one diagnostic tool, a real-time PCR assay identifying mutations in a nitroreductase gene (ntr6), is clinically available for the molecular identification of Mtz resistance in T. vaginalis (Paulish-Miller et al., 2014, Marques-Silva et al., 2021). Mutations in nitroreductase genes may reduce or eliminate its enzymatic activity, limiting the formation of toxic 5NI radicals (Paulish-Miller et al., 2014), which, by current definition, is a basis for oxygen-independent 5NI resistance (Graves et al., 2020). However, as multiple key antioxidant and metabolic pathways are associated with overall 5NI resistance (Graves et al., 2020), there likely exist multiple genes that are associated with oxygen-dependent and/or oxygen-independent resistance, which are accepted to have different underlying mechanisms (Kulda, 1999). Whilst studies have identified a panel of single nucleotide polymorphisms that are associated with T. vaginalis Mtz resistance (Bradic et al., 2017), further experiments are needed to identify biomarkers for 5NI resistance and evaluate their suitability in oxygen-dependent and oxygen-independent resistance phenotypes in both trichomonad species.

Furthermore, new quantitative and automatable platforms for the interrogation of oxygen-dependent and -independent 5NI resistance will support studies to understand the prevalence, relationship, and clinical relevance of both 5NI resistance phenotypes. In particular, these assays can test the hypothesis that oxygen-dependent resistance precedes oxygen-independent resistance, and continued exposure to 5NI treatments may eventually lead to “complete” resistance due to a loss in drug activation pathways (Kulda et al., 1993, Kulda, 1999). If oxygen-dependent resistance precedes oxygen-independent resistance, then detection of the former may be used as a basis to alter the current clinical treatment regime, lowering the risk of developing further complex resistances. For example, by switching to Tnz for T. vaginalis infections sooner to limit selecting and developing anaerobically Mtz-resistant parasites that are difficult to clear through dose escalation (Voolmann and Boreham, 1993). Together with clinical observations, these growth assays may be used as a tool to phenotypically assess resistance status. Additionally, these may directly advise clinicians on the efficacy of the current treatment and inform dose/drug changes as required.

As treatment failure for trichomoniasis is increasing, there is a need for a standardised method of resistance surveillance. To our knowledge, this is first quantitative study interrogating both 5NI resistance phenotypes in trichomonads. We demonstrate our assays are robust and quantifiable to assess oxygen-dependent and oxygen-independent 5NI resistance in both T. vaginalis and T. foetus through drug susceptibility testing under microaerophilic and anaerobic conditions, respectively. Importantly, our assays have a quantitative ‘readout’ of growth, thereby eliminating qualitative ‘by-eye’ estimations of pharmacodynamic parameters such as maximum effect (Emax) and MICs. More importantly, we have outlined the need for caution in the levels of oxygen required for in vitro growth, and the implications for how 5NI resistance should be examined. We hope that these assays will provide the phenotypic platforms for standardising 5NI resistance as a basis for optimising genetic and molecular tools to complement the current resistance surveillance program in these important and neglected parasites.