Repurposing drugs to fast-track therapeutic agents for the treatment of cryptococcosis

Strains

Fungal strains used in this study are listed in Tables 1 and 2. Strains were maintained as glycerol stocks stored at −80 °C. Strains were grown on sabouraud dextrose agar (SDA) for 48 h at 30 °C before use.

Enzo drug library and drug stocks

The Screen-Well FDA-approved drug library was purchased from Enzo Life Sciences (catalogue no. BML-2841; Farmingdale, NY, USA). The 640 compounds in this library were provided at 2 mg/mL in dimethyl sulfoxide (DMSO). Flubendazole (FLB), mebendazole (MEB), benomyl (BEN), nisoldipine (NIS), nifedipine (NIF), felodipine (FEL), niguldipine (NIG), amphotericin B (AMB), itraconazole (ITZ), voriconazole (VRZ) and 5-flucytosine (5FC) were obtained from Sigma-Aldrich (St. Louis, MO, USA). Fluconazole (FLC) was obtained from Cayman Chemical (Ann Arbor, MI, USA). Stock solutions of amphotericin B, itraconazole and voriconazole were prepared at 1.6 mg/mL in water; 5-flucytosine was made at 6.4 mg/mL in water; fluconazole was made at 12.8 mg/mL in water and all remaining drugs were dissolved in DMSO at 2 mg/mL. All compounds were stored at −80 °C.

Primary drug screening

We evaluated all 640 FDA-approved compounds in the Enzo library for efficacy against C. deuterogattii strain R265. In the primary screen, each compound was tested once at 10 μg/mL and once at 40 μg/mL using 96-well flat-bottom microtitre plates. Amphotericin B was included as a standard antifungal control. Untreated cells and broth-only controls were also included. Cell suspensions of C. deuterogattii strain R265 were prepared following the reference methods of the Clinical and Laboratory Standards Institute (CLSI) for broth dilution antifungal susceptibility testing of yeasts (Clinical and Laboratory Standards Institute (CLSI), 2008b). Briefly, an isolated colony of C. deuterogattii strain R265 from a 2-day old plate was inoculated into 1 mL of Yeast Nitrogen Broth (YNB, pH 7; Sigma-Aldrich, St. Louis, MO, USA). A working cell suspension was prepared at 1 × 10⁶ to 5 × 10⁶ cells/mL by cell counting using a haemocytometer, and was diluted 1:100. Final cell concentrations were 0.5 × 10³ to 2.5 × 10³ cells/mL with a final volume of 200 μL per well. Plates were incubated at 37 °C (humidified) without agitation for 72 h. Initial ‘hit’ candidates were defined as compounds that visibly inhibited the growth of C. deuterogattii.

To identify compounds with the greatest potential for repurposing as novel antifungal agents, compounds that complied with the following were excluded from further assessment: (i) compounds that were active only at concentrations greater than 10 μg/mL; (ii) compounds that were known antineoplastic agents or had otherwise unwanted adverse profiles; and (iii) currently established antifungal agents or non-antifungal agents with antifungal activity that had already been described and assessed at the time of testing.

Broth microdilution test

The in vitro antifungal activities of ‘hit’ compounds were re-evaluated using the stocks supplied in the library as well as from independently sourced stocks, and testing was extended to additional pathogenic fungal species (Tables 1 and 2). Drug compounds were serially diluted twofold in YNB medium (final concentrations: 0.078125–40 μg/mL). Amphotericin B was included as an antifungal control for reference. Niguldipine, a member of the same class of 1,4-dihydropyridines (calcium channel blocker (CCB)) that had been identified as a non-‘hit’ compound in the screen was included as a negative control for the CCB series. Fungal cell suspensions were prepared following the CLSI standard method for broth dilution antifungal susceptibility testing of yeasts (Clinical and Laboratory Standards Institute (CLSI), 2008b) or filamentous fungi (Clinical and Laboratory Standards Institute (CLSI), 2008a). Plates were incubated without agitation at 35 or 37 °C (humidified) for 48 h for non-Cryptococcus species or 72 h for Cryptococcus species. The minimum inhibitory concentration (MIC) was determined as the lowest drug concentration at which a compound visibly inhibited growth (100% for FLB, MEB, BEN, NIF, NISO, FELO, NIG, AMB and 5FC; 80% for FLC, ITZ and VRZ). To obtain the minimum fungicidal concentration (MFC), 30 μL from each dilution well at the MIC concentration and higher was spread onto SDA plates. The MFC was defined as the lowest drug concentration that yielded three or fewer colonies after 72 h incubation (Espinel-Ingroff et al., 2002). All microdilution assays were tested in technical duplicates with at least two biological replicates unless otherwise specified.

Checkerboard microdilution assay

Checkerboard assays determine the interaction between two drugs based on the fractional inhibitory concentration index (FICI) value. Interactions can be synergistic (FICI ≤ 0.5), additive/indifferent (0.5 < FICI ≤ 4) or antagonistic (FICI > 4) (Lai et al., 2016). Briefly, drug A (flubendazole, mebendazole or benomyl) was serially diluted from 0.31 to 0.0024 μg/mL and aliquoted in a 96-well flat-bottom plate and combined with drug B that was serially diluted and aliquoted in the following concentration ranges: amphotericin B (2–0.016 μg/mL); fluconazole (32–0.25 μg/mL); itraconazole (2–0.016 μg/mL); voriconazole (2–0.016 μg/mL); 5-flucytosine (32–0.25 μg/mL).

FICI values were calculated as: $$\text{FICI\hspace{0.17em}}=\frac{\text{MIC drug A in combination}}{\text{MIC drug A alone}}+\frac{\text{MIC drug B in combination}}{\text{MIC drug B alone}}$$ All drug combinations were tested against C. deuterogattii strain R265 and C. neoformans strain H99 as per CLSI guidelines. All plates were incubated at 37 °C and inhibition was read visually at 72 h.

Time-kill assay

Cryptococcus neoformans strain H99 was grown overnight to exponential phase in YNB media, shaking (180 rpm) at 37 °C. Cells were adjusted to 1 × 10⁶ cells/mL using a haemocytometer and sub-cultured in fresh YNB media. Three hours post sub-culturing, two 50 mL aliquots were taken. One culture was not treated, while the other was treated with 0.06 μg/mL flubendazole (1.5 × MIC). Cultures were returned to the shaking incubator and monitored for cell growth at 2, 3, 4, 5, 6, 9 and 24 h post-treatment. At each time point, each culture was sampled in technical duplicates, serially diluted in milli-Q water and spread onto SDA plates. All plates were incubated at 37 °C and observed at 48 h and 72 h for viability counts.

Primary drug screening identifies non-antifungal compounds with anti-cryptococcal activity

The preliminary screening of the Enzo library revealed a diverse variety of compounds that inhibited the growth of C. deuterogattii strain R265 (Fig. 1; Fig. S1; Table S1). A significant proportion (15%) of ‘hits’ from this initial screen were antineoplastic drugs. This antifungal affect may be attributed to the cytotoxic nature of these compounds since cryptococcal and humans cells are eukaryotic and may share cellular structures and processes that result in non-selective toxicity. For this reason, this group of drugs were not pursued as potential antifungal candidates. Using the selection criteria described in the “Methods” section, 16 of the remaining compounds identified in the primary screen were short-listed for further evaluation by MIC assays against C. deuterogattii. Antifungal activity was confirmed for ten compounds using drug stocks sourced from the library. The most potent compounds (MICs < 5 μg/mL) were aripiprazole, nisoldipine, oxatomide, flubendazole, nifedipine and anethole-trithione. Lack of commercial availability restricted further evaluation to flubendazole, an anthelmintic drug (Fig. 2), and nisoldipine and nifedipine, which are antihypertensive CCBs (Fig. 3). Felodipine and niguldipine, which are CCBs structurally related to nifedipine and nisoldipine, were also included in the assay.

In vitro spectrum of antifungal activity of CCBs and flubendazole

To investigate the spectrum of antifungal activity, MICs were determined for nisoldipine, nifedipine, felodipine, niguldipine and flubendazole against nine additional strains from four fungal species (Table 1). MIC assays using independently sourced drug stocks confirmed the MICs established for C. deuterogattii strain R265 and demonstrated a diverse range of antifungal activities against other fungal strains and species (Table 1). Flubendazole was the most potent compound against Cryptococcus species with MICs as low as 0.078 μg/mL, but was ineffective even at the highest tested concentration (40 μg/mL) against other fungal species. The series of CCBs presented an interesting array of activities across the fungal species. Nifedipine demonstrated consistent antifungal activity against the Cryptococcus species (5–10 μg/mL) and inhibited Candida glabrata, an emerging fungal pathogen, with an MIC of 20 μg/mL. In contrast, nisoldipine had higher and more variable MICs against different Cryptococcus species (5–40 μg/mL) and was the only drug that had activity against all nine fungal species. Niguldipine was not effective at the highest tested concentration against the fungal strains in this assay, while felodipine demonstrated activity against the Cryptococcus species with MICs between those of nifedipine and nisoldipine, but was only active against Cryptococcus species.

Antifungal activities of flubendazole

To test whether flubendazole is broadly active across Cryptococcus species, we extended the study to encompass all members of the C. gattii species complex, as well as the major molecular genotypes of C. neoformans (Table 2; Table S2), including veterinary, clinical and environmental isolates from various geographic locations. Flubendazole inhibited all isolates of both Cryptococcus species at consistently low concentrations (MIC 0.039–0.156 μg/mL). Furthermore, for isolates where fluconazole MIC data were available, flubendazole was found to be equally effective regardless of fluconazole susceptibility (Table 2; Table S2).

To explore the inhibitory effect of flubendazole on Cryptococcus cells, cell viability was assessed using a time-kill assay and determining the MFC in a microwell format. Analysis by time-kill curve over 24 h demonstrated that Cryptococcus cells exposed to flubendazole had slowly declining growth for up to 9 h post-treatment prior to loss of viability (Fig. 4; Table S3). Microbroth dilution assays generated MFC values that were similar to the respective MIC values (MFC 0.039–0.156 μg/mL; Table 2).

To explore interactions with current, commonly used antifungals, flubendazole and the other benzimidazole agents mebendazole and benomyl was tested in combination with amphotericin B, fluconazole, itraconazole, voriconazole and 5-flucytosine against C. deuterogattii and C. neoformans in a checkerboard assay. All combinations produced additive/indifferent results and no synergistic or antagonistic interactions were seen (Table 3; Tables S4–S5).

Repurposing CCBs as antifungal agents

Calcium channel blockers, widely used in the treatment of hypertension, are broadly divided into dihydropyridines and non-dihydropyridines classes (Catterall, 2011). Our screen identified nifedipine, felodipine and nisoldipine (Table 1; Fig. 3) as having antifungal activity against Cryptococcus. These are all 1,4-dihydropyridines and therefore present as intriguing leads for a potential new class of antifungal drugs.

Members of the 1,4-dihydropyridine CCB group share a core phenyl group attached to a 1,4-dihydropyridine that has two ester side groups (Fig. 3). Variations in side groups may account for the diverse antifungal activities observed, as compounds with the bulkier side groups (lacidipine and niguldipine) were ineffective against C. deuterogattii. Nifedipine, felodipine, nisoldipine and niguldipine target L-type calcium channels—one of five types of voltage-gated calcium channels (VGCCs) designated L-, N-, P/Q-, R- and T-type (Catterall, 2011). In the model yeast Saccharomyces cerevisiae, Cch1 and Mid1 are homologs of the mammalian VGCCs, and L-type CCBs are known to inhibit fungal growth via a Cch1–Mid1 protein channel complex located in the plasma membrane (Prole & Taylor, 2012; Teng et al., 2008). Cch1 acts as the calcium channel and has homologs in almost all fungi, while Mid1 is a regulatory subunit that interacts with Cch1 to mediate calcium influx (Hong et al., 2013). Hence, it appears likely that the antifungal effects of CCBs involve the Cch1–Mid1 calcium channel.

Our findings support a growing body of evidence suggesting a crucial role of calcium regulation and calcium channels in determining the growth and survival of yeasts under stressful conditions (Hong et al., 2010; Liu et al., 2006; Peiter et al., 2005). In C. neoformans, Cch1 facilitates calcium entry with Mid1 when there is a depletion of intracellular calcium (Hong et al., 2010). In the presence of the calcium chelator BAPTA, the growth of Δcch1 mutants is markedly diminished at elevated temperature (25–38.5 °C), a key virulence factor that is critical for human infection (Liu et al., 2006). Cch1 and Mid1 also appear linked to the regulation of intracellular redox under cell stress, as S. cerevisiae Δcch1 and Δmid1 (single- and double-knockout) mutants are sensitive to cold stress (10 °C) and iron toxicity due to a compromised ability to cope with oxidative stress (Peiter et al., 2005). Cell growth can be re-established following supplementation with either 10 mM calcium or reduced glutathione (an antioxidant), indicating that a functional Cch1–Mid1 complex is essential for stress tolerance (Peiter et al., 2005).

As well as possessing antifungal properties, CCBs have also been shown to act synergistically with fluconazole against fluconazole-resistant and -susceptible strains of C. albicans (Liu et al., 2016). Interestingly, CCBs alone or in combination with fluconazole do not significantly alter the gene expression level of CCH1 and MID1 (Liu et al., 2016) suggesting that either the cell is not sensing a disruption to calcium homeostasis, or CCBs act on targets other than calcium channels. Recently, a screen of the Prestwick Chemical Library® of FDA-approved small molecules identified fendiline hydrochloride, a non-dihydropyridine CCB, as having unique activity to potentiate phagosomal killing of intracellular C. neoformans in host macrophages (Samantaray et al., 2016). Although it has a very different structure to the antifungal CCBs identified here, fendiline is an L-type CCB, suggesting this class of CCBs may be the most promising for antifungal drug development. Furthermore, the antifungal activity of L-type CCBs in combination with standard antifungals have been characterised in C. albicans and Aspergillus (Afeltra et al., 2004; Liu et al., 2016). As hypertensive agents, L-type CCBs are currently administered in long term use with relatively minimal side effects (Chrysant, 2003). Of clinical relevance, some CCBs including nifedipine and nisoldipine can traverse the blood–brain barrier and may therefore be able to treat cryptococcal meningitis (Gaggi, Cont & Gianni, 1993). We therefore suggest L-type CCBs are promising leads for repurposing and redevelopment as potential new antifungal agents.

Repurposing flubendazole as an anti-cryptococcal agent

Flubendazole is an anthelmintic drug belonging to a family of benzimidazole carbamates. First introduced in 1980, it is currently used to treat parasitic gastrointestinal worm infections in humans and animals (Fig. 2) (Spagnuolo et al., 2010). Flubendazole acts by binding to and inhibiting the polymerisation of tubulin, a cytoskeletal protein that polymerises to form microtubules (Chatterji et al., 2011; Cruz & Edlind, 1997; Spagnuolo et al., 2010). Microtubules play major roles in various cell processes including division, motility, adhesion and intracellular transport (Chatterji et al., 2011). Benzimidazoles have been found to act selectively against fungal tubulin and not mammalian tubulin, possibly due to differences in their β-tubulin sequence (Cruz & Edlind, 1997).

Benzimidazoles were originally developed as agricultural fungicides prior to their use as anthelmintic drugs (Horton, 2000). While the antifungal activities of benzimidazoles and their potential to be developed as anti-cryptococcal agents have been known for over two decades (Patent US5434163) (Cruz, Bartlett & Edlind, 1994; Del Poeta et al., 1998; Edlind & Cruz, 1995), no further development as antifungal agents has occurred since 1998 (Del Poeta et al., 1998). Our study indicated that flubendazole is fungicidal at extremely low concentrations against C. neoformans and C. deuterogattii (Table 2). Our findings show that flubendazole equally inhibits pathogenic and environmental isolates of Cryptococcus, as well as strains that are resistant to fluconazole. This indicates that flubendazole may be an effective treatment against genetically and physiologically diverse Cryptococcus species and strains. As resistance to fluconazole is frequently induced during prolonged treatment and has been associated with clinical failure in Cryptococcus-infected patients (Bicanic et al., 2006; Jarvis et al., 2014; Smith et al., 2015), flubendazole may have promise as a second-line therapy when fluconazole fails, particularly in resource-poor regions where there are often no suitable alternatives.

Flubendazole interferes with normal cell growth as early as 3 h post-treatment and continues to render treated Cryptococcus cells unviable (Fig. 4), and in a microwell format with 72 h incubation the MFC values were either the same or one increment higher than the respective MIC values (Table 2). This indicates that flubendazole is a fungicidal drug, and with the correct clinical administration, it could effectively clear fungal burden. As high fungal burden is a strong determinant for mortality and overall poor clinical outcomes, and currently amphotericin B is the only fungicidal agent that is effective against cryptococcosis, this reinforces the potential utility of flubendazole as an anti-cryptococcal agent (Jarvis et al., 2014). It is noted however, that unlike amphotericin B, flubendazole does not act synergistically with other standard antifungal agents, consistent with previously published data (Table 3) (Joffe et al., 2017).

The current findings support a recent screening of 727 compounds in the National Institute of Health Clinical Collection (Joffe et al., 2017). Flubendazole and three related anti-helminthic benzimidazoles (mebendazole, albendazole and triclabendazole; sharing the benzimidazole scaffold) were among 17 compounds active against C. neoformans (Joffe et al., 2017). However, that study focused on mebendazole on the basis that the drug is able to penetrate the blood–brain barrier. Low MIC and MFC values were reported, and efficacy against biofilms and against cryptococcal cells internalised in macrophages were shown, along with additive activity when combined with amphotericin B (Joffe et al., 2017). Flubendazole may mirror the antifungal capabilities observed for mebendazole as they share structural similarities, and there is evidence that like mebendazole, flubendazole can traverse the blood–brain barrier (Téllez-Girón et al., 1984) making these compounds favourable candidates for the treatment of cryptococcal meningitis. However, as flubendazole is formulated to treat gastrointestinal worms, it is not yet known whether it would be able to reach therapeutic concentrations in the brain required to elicit an antifungal effect.

The precise mechanism underlying the antifungal activity of flubendazole is unclear, however it is thought that it targets β-tubulin in C. neoformans (Cruz & Edlind, 1997). The kill-curve demonstrated initial growth after treatment followed by declining viability and death at 9 h, which is consistent with a mode of action that works by interfering with cell growth and division. In Echinococcus granulosus, a tapeworm parasite, flubendazole has been found to disrupt a wide spectrum of cellular homeostatic mechanisms, including energy metabolism and calcium homeostasis. In the latter it appears to increase cytosolic calcium from extracellular and intracellular stores (Cumino, Elissondo & Denegri, 2009), demonstrating an interesting overlap with the CCBs outlined above.

Overall, flubendazole (and more generally the benzimidazole scaffold) may be a promising starting point for the treatment of cryptococcal disease. This view is further encouraged by a lack of evidence of serious adverse side effects in patients undergoing anti-helminthic treatment and in mice experimentally treated with flubendazole (Ceballos et al., 2009; Feldmeier et al., 1982; Lassègue et al., 1984; Téllez-Girón et al., 1984). In preliminary cryptococcal meningitis studies, flubendazole demonstrated fungicidal activity in a hollow fibre infection model and efficacy in vivo, reducing fungal burden in both a rabbit and mice model (Nixon et al., 2018). However, whether flubendazole can accumulate in the brain to concentrations that are clinically achievable and able to elicit an antifungal effect, and whether this can be sustained in the absence of adverse reactions, remains unknown and requires more extensive in vivo testing.