Identification of Off-Patent Drugs That Show Synergism with Amphotericin B or That Present Antifungal Action against Cryptococcus neoformans and Candida spp.

Amphotericin B (AmB) is the antifungal with the strongest fungicidal activity, but its use has several limitations, mainly associated with its toxicity. Although some lipidic and liposomal formulations that present reduced toxicity are available, their price limits their application in developing countries. Flucytosine (5FC) has shown synergistic effect with AmB for treatment of some fungal infections, such as cryptococcosis, but again, its price is a limitation for its use in many regions. In the present work, we aimed to identify new drugs that have a minor effect on Cryptococcus neoformans, reducing its growth in the presence of subinhibitory concentrations of AmB.

ABSTRACT

Amphotericin B (AmB) is the antifungal with the strongest fungicidal activity, but its use has several limitations, mainly associated with its toxicity. Although some lipidic and liposomal formulations that present reduced toxicity are available, their price limits their application in developing countries. Flucytosine (5FC) has shown synergistic effect with AmB for treatment of some fungal infections, such as cryptococcosis, but again, its price is a limitation for its use in many regions. In the present work, we aimed to identify new drugs that have a minor effect on Cryptococcus neoformans, reducing its growth in the presence of subinhibitory concentrations of AmB. In the initial screening, we found fourteen drugs that had this pattern. Later, checkerboard assays of selected compounds, such as erythromycin, riluzole, nortriptyline, chenodiol, nisoldipine, promazine, chlorcyclizine, cloperastine, and glimepiride, were performed and all of them confirmed for their synergistic effect (fractional inhibitory concentration index [FICI] < 0.5). Additionally, toxicity of these drugs in combination with AmB was tested in mammalian cells and in zebrafish embryos. Harmless compounds, such as the antibiotic erythromycin, were found to have synergic activity with AmB, not only against C. neoformans but also against some Candida spp., in particular against Candida albicans. In parallel, we identified drugs that had antifungal activity against C. neoformans and found 43 drugs that completely inhibited the growth of this fungus, such as ciclopirox and auranofin. Our results expand our knowledge about antifungal compounds and open new perspectives in the treatment of invasive mycosis based on repurposing off-patent drugs.

INTRODUCTION

Opportunistic invasive fungal diseases (IFDs) pose a life-threatening problem for the increasing population of immunocompromised patients in our society (1, 2). Their incidence has risen in the last few years, and their management has high associated costs, estimated around to be $30,000 to $50,000 per patient (3). Despite the availability of antifungal families, treatment of IFDs presents several limitations, such as antifungal cost, off-target toxicity, and restricted spectrum of action.

The main antifungal families are azoles, echinocandins, 5-fluoro-cytosine (flucytosine), and polyenes (amphotericin B [AmB]). In the case of polyenes, and in particular, AmB, several action mechanisms have been described. Classically, it has been stated that this antifungal binds to ergosterol and forms pores at the cell membrane (4–6). But AmB has other effects on the cells, such as ergosterol sequestration (7, 8) and induction of oxidative damage (9–13), which also contributes to the fungicidal activity of the drug.

AmB is the antifungal that presents the strongest fungicidal activity together with a wider action spectrum (see review in reference 14), but its toxicity restricts its use. To overcome this problem, several lipidic and liposomal formulations have been developed. However, their price prevents their frequent use in developing countries. An example are diseases caused by Cryptococcus neoformans, which is a basidiomycete that causes disease mainly in HIV-positive individuals (15), and it has a high prevalence in developing countries. Treatment for cryptococcosis is based on initial administration of AmB in combination with flucytosine, followed by fluconazole (16). Flucytosine use is also restricted due to its price (17) and limited availability in some regions.

The development of new drugs is an expensive and time-consuming process. However, nowadays, there are other strategies to find new antimicrobial compounds, based on screening of large collections of drugs (18). This approach has been successfully used in the last few years to identify drugs that present antifungal activity. For example, screening of synthetic drug libraries revealed two hydrazides, N′-(3-bromo-6-hydroxybenzylidene)-2-methylbenzohydrazide (BHBM) and its derivative 3-bromo-N′-(3-bromo-4-hydroxybenzylidene benzohydrazide), that target the synthesis of fungal sphingolipids and that could be promising drugs for the treatment of cryptococcosis (19, 20). A variation of these screenings is to test the activity of off-patent compounds, which is known as drug repurposing. The repurposing strategy has also been applied to look for active compounds (AC) against fungi, such as Candida albicans (21–26), C. neoformans (26–28), the emerging pathogen Candida auris (26, 29, 30), and multiresistant molds such as Lomentospora prolificans (26).

Since AmB is the antifungal with the strongest activity, we aimed to identify new drugs or combinations that could potentially enhance its fungicidal effect. For this purpose, we screened the Prestwick Chemical Library for compounds that increased the activity of AmB. This library has been used in screenings to find drugs with antifungal activity against C. neoformans and C. albicans (25, 28). It has even been used to identify synergistic drugs with amphotericin B and caspofungin against C. albicans biofilms (31). Once the AC were identified, their synergic effect with AmB was tested through checkerboard assay. The synergic combinations (AC + AmB) were tested for toxicity through an in vitro test with mammalian cells and in an in vivo test with zebrafish embryos. In parallel, other drugs that have full activity against C. neoformans were presented, some of them with known antifungal activity but others still not characterized.

RESULTS

Identification of the optimal subinhibitory AmB concentration against C. neoformans.

We first selected the minimal concentration of AmB that caused a modest inhibitory effect on C. neoformans. As shown in Fig. 1, AmB at 0.03 μg/ml caused a 20 to 30% decrease of C. neoformans growth. Higher concentrations (0.06 and 0.12 μg/ml) caused a more variable effect, depending on the day and plates, ranging the growth inhibition between 20 and 80%. In all cases, almost full inhibition of growth was found at concentrations of ≥0.25 μg/ml. In consequence, we selected the concentration of 0.03 μg/ml as the minimum subinhibitory concentration of AmB to be used for our initial screening.

FIG 1.

Effect of different AmB concentrations on growth in C. neoformans, C. krusei, and C. parapsilosis.

Screening for AmB synergistic compounds.

We next performed the screening to identify compounds that had a moderate effect on growth in C. neoformans but that had an inhibitory effect when combined with AmB 0.03 μg/ml. For this purpose, we prepared two plates as described in Materials and Methods as follows: one with the compounds alone (100 μM) and another one in which they were mixed with AmB (0.03 μg/ml). In each plate, we included wells without any drug as growth controls. After inoculation of the plates with the C. neoformans H99 strain and incubation for 48 h, we identified 14 potential drugs that enhanced the activity of AmB (Table 1). These AC were atracurium besylate, erythromycin, riluzole hydrochloride, nortriptyline hydrochloride, chenodiol, thioguanosine, diflorasone diacetate, oxiconazole nitrate, glimepiride, nisoldipine, promazine hydrochloride, chlorcyclizine hydrochloride, demeclocycline hydrochloride, and cloperastine hydrochloride. The combination of these drugs with subinhibitory AmB concentrations

TABLE 1.

Synergic compounds at 0.1 mM with AmB 0.03 μg/ml against C. neoformans

Compound name	Amount (μg/ml)	Therapeutic class	Target/action mechanism	Compound inhibition (%)	Combination inhibition (%)
Atracurium besylate	124	Neuromuscular	Nondepolarizing neuromuscular blocking agent. Antagonizes the neurotransmitter action of acetylcholine	40	95
Erythromycin	73	Infectiology/metabolism	Inhibits bacterial protein synthesis by binding to bacterial 50S ribosomal subunits	15	90
Riluzole hydrochloride	27	Central nervous system	Inhibitory effect on glutamate release, inactivation of voltage-dependent sodium channels, and interference in intracellular events that follow transmitter binding at excitatory amino acid receptors	25	95
Nortriptyline hydrochloride	30	Central nervous system	Inhibits the reuptake of the neurotransmitter serotonin at the neuronal membrane or acts at beta-adrenergic receptors	30	90
Chenodiol	40	Gastroenterology	Suppresses hepatic synthesis of both cholesterol and cholic acid, gradually replacing the latter and its metabolite, deoxycholic acid, in an expanded bile acid pool	30	90
Thioguanosine	31	Metabolism/oncology	Antineoplastic action. Inhibits the synthesis of DNA and RNA of cells.	40	85
Diflorasone diacetate	49	Endocrinology	Synthetic glucocorticoid that binds to the glucocorticoid receptor (GR) in the cytoplasm.	30	90
Oxiconazole nitrate	49	Infectiology/metabolism	Inhibits ergosterol biosynthesis, inhibits DNA synthesis, and suppresses the intracellular concentrations of ATP	5	95
Nisoldipine	39	Cardiovascular	By deforming the channel, inhibiting ion-control gating mechanisms, and/or interfering with the release of calcium from the sarcoplasmic reticulum. Inhibits the influx of extracellular calcium across the myocardial and vascular smooth muscle cell membranes	25	95
Promazine hydrochloride	32	Central nervous system	Antagonism at dopamine and serotonin type 2 receptors, with greater activity at serotonin 5-HT2 receptors than at dopamine type-2 receptors	20	80
Chlorcyclizine hydrochloride	33	Allergology/central nervous system	Histamine H1 receptor	15	100
Demeclocycline hydrochloride	50	Metabolism	Inhibits the translation by binding to the 30S and 50S ribosomal subunit, impairing protein synthesis.	35	90
Cloperastine hydrochloride	36	Respiratory	Not identified	5	90
Glimepiride	49	Endocrinology	Binds to ATP-sensitive potassium channel receptors on the pancreatic cell surface. Reduces potassium conductance causing depolarization of the membrane and stimulates calcium ion influx through voltage-sensitive calcium channels. Induces the secretion of insulin	5	80

induced more than 80% of fungal growth inhibition, while the compounds alone did not have any significant effect on C. neoformans growth (Table 1).

Checkerboard assay.

To confirm the synergistic activity, we selected nine compounds to perform checkerboard assays, based on different criteria, such as their administration route and degree of synergistic effect. This assay allows for the calculation of the fractional inhibitory concentration index (FICI) (see Materials and Methods). The main results were obtained at concentrations between 0.1 and 0.025 mM. For each combination, we selected the FICI for both 50 and 75% C. neoformans growth inhibition. The calculated FICI in all combinations was ≤0.5, except for nortriptyline hydrochloride, which provided an FICI of >0.5 only when 75% growth inhibition was examined (Table 2). These results confirmed the synergistic activity of the selected AC + AmB combinations.

TABLE 2.

Fractional inhibitory concentration index for the selected compounds

Compound	FICI
	50%	75%
Erythromycin	0.35	0.43
Riluzole hydrochloride	0.39	0.37
Nortriptyline hydrochloride	0.40	0.62
Chenodiol	0.31	0.37
Nisoldipine	0.40	0.38
Promazine hydrochloride	0.34	0.43
Chlorcyclizine hydrochloride	0.16	0.37
Cloperastine hydrochloride	0.25	0.34
Glimepiride	0.15	0.28

Erythromycin showed a synergistic effect with AmB at concentrations between 0.1 mM (73 μg/ml) and 0.025 mM (18.25 μg/ml). Furthermore, it partially enhanced the antifungal activity even at low AmB concentrations of 0.008 μg/ml (Fig. 2A). Riluzole hydrochloride was also synergic at 0.1 mM (27 μg/ml) and 0.05 mM (13.5 μg/ml) in combination with 0.03 to 0.06 μg/ml AmB (Fig. 2B). Nortriptyline hydrochloride at 0.1 mM (30 μg/ml), in a minor scale, also potentiated AmB activity at 0.03 μg/ml and 0.015 μg/ml, inhibiting around 50% and 30% of the C. neoformans growth, respectively (Fig. 2C). For chenodiol, the better synergistic activity was observed using 0.1 mM (40 μg/ml), 0.05 mM (20 μg/ml), and 0.025 mM (10 μg/ml) combined with 0.015 to 0.06 μg/ml of AmB (Fig. 2D). Nisoldipine induced a drastic decreased in fungal growth even at 0.025 mM (9.75 μg/ml) and concentrations as low as 0.008 μg/ml of AmB (Fig. 2E). For promazine hydrochloride, chlorcyclizine hydrochloride, and cloperastine, where 0.1 mM corresponded to 32 μg/ml, 33 μg/ml, and 36 μg/ml, respectively, the synergistic effect was found with concentrations around 0.1 to 0.025 mM, which enhanced the antifungal activity of AmB at concentrations even as low as 0.008 μg/ml (Fig. 2F to H). Finally, in the presence of 0.03 μg/ml AmB, glimepiride reduced fungal growth 60% to 70% at concentrations of 0.1 (49 μg/ml), 0.05 (24.5 μg/ml), and 0.025 mM (12.25 μg/ml) (Fig. 2I).

FIG 2.

Effect of combination of erythromycin (A), riluzole (B), nortriptyline (C), chenodiol (D), nisoldipine (E), promazine (F), chlorcyclizine (G), cloperastine (H), and glimepiride (I) with AmB (0.008 to 0.5 μg/ml) on C. neoformans growth (H99 strain). In each case, the graph shows the effect of 0.1 (gray straight line, squares), 0.05 (gray straight line, triangles), and 0.025 mM (dotted gray line, inverted triangle) of each compound. The black straight line shows the growth of C. neoformans in the presence of AmB without any AC. The results were obtained from the checkerboard assays (see Materials and Methods for details). The experiment was performed in triplicates, and the average and standard deviation for each point are plotted.

Cytotoxicity assay in mammalian cells.

We next investigated whether the combinations of the identified AC with AmB had deleterious effects on mammalian cells. As shown in Fig. 3, erythromycin (Fig. 3A), chenodiol (Fig. 3B), and glimepiride (Fig. 3C), alone at 0.1 mM or in combination with AmB at 0.03 μg/ml, presented low cytotoxicity, with ∼80% cell viability after 24 h. Nisoldipine (Fig. 3D) presented about 80% cell viability when used alone but reduced to 70% cell viability in combination with AmB. Chlorcyclizine hydrochloride, riluzole hydrochloride, and nortriptyline hydrochloride induced a higher toxicity at 0.1 mM (around 50% death), although addition of AmB did not increase the inherent toxicity with the AC alone (Fig. 3E, F, and G). Promazine hydrochloride alone showed cell death around 50% and increased to 60% in combination with AmB (Fig. 3H). Finally, cloperastine (Fig. 3I) was the compound that showed the highest toxicity at the concentration tested (around 70% of mortality) (Fig. 3I).

FIG 3.

Toxicity of drug combination in the RAW264.7 cell line. Toxicity assays in macrophage-like RAW264.7 were performed as described in Materials and Methods. For each combination, the graphs show the death percentage of control cells (live control), lysis control, AmB (0.03 μg/ml), the corresponding AC (0.1 mM), and combination of both. The experiment was performed in triplicates on different days, and the bars show the average and standard deviation. Asterisks denote statistical difference between the samples and the live control cells. ERY, erythromycin; CHE, chenodiol; GLI, glimipiride; NISO, nisoldipine; CHL, chlorcyclizine hydrochloride; RLZ, riluzole; NOR, nortriptyline; PRO, promazine hydrochloride; CLO, cloperastine.

Toxicity to zebrafish embryos.

Physical-chemical characteristics of the test solution fell into the range of the valid criteria according to OECD 236 (32).

Table 3 shows the lethality data of the drug solutions (single or in combinations with AmB) for the zebrafish embryos (n = 16/group). Single compounds such as AmB, erythromycin, and glimepiride did not produce mortality for the time of exposure (0 to 96 h post fertilization [hpf]). Other compounds, such as nisoldipine, chlorcyclizine, cloperastine, promazine, or riluzole, exhibited relatively rapid zebrafish embryo mortality (during the first 24 to 48 hpf), while nortriptyline induced mortality at a later developmental stage (between 48 and 96 hpf). When the mixtures of AC with AmB were tested, we found that the combinations of AmB and erythromycin as well as AmB and glimepiride produced no mortality. The addition of AmB to chlorcyclizine, cloperastine, and promazine boosted early mortality (24 hpf) in comparison with the single compound data, indicating increased toxicity of the mixture. Negative control group (dilution water) and negative solvent group (0.4% dimethyl sulfoxide [DMSO] in egg water) did not exhibit mortality, while the positive control group showed mortality higher than 30%, in agreement with the OECD 236 guideline (32).

TABLE 3.

Mortality in zebrafish embryos exposed to single drugs at 100 μM or their combination with AmB at 0.03 μg/ml

Compound or mixturea	% Mortality
	24 h	48 h	72 h	96 h	Total
Single AC
AmB	0	0	0	0	0
Chenodiol	0	100	100	100	100
Nisoldipine	100	100	100	100	100
Chlorcyclizine	0	100	100	100	100
Cloperastine	0	100	100	100	100
Erythromycin	0	0	0	0	0
Glimepiride	0	0	0	0	0
Promazine	0	100	100	100	100
Riluzole	100	100	100	100	100
Nortriptyline	0	0	0	100	100
DMSO 0.4%	0	0	0	0	0
Cont –	0	0	0	0	0
Cont +	18.75	43.75	100	100	100
Mixture
Chenodiol + AmB	0	100	100	100	100
Nisoldipine + AmB	100	100	100	100	100
Chlorcyclizine + AmB	100	100	100	100	100
Cloperastine + AmB	100	100	100	100	100
Erythromycin + AmB	0	0	0	0	0
Glimepiride + AmB	0	0	0	0	0
Promazine + AmB	56.25	81.25	100	100	100
Riluzole + AmB	100	100	100	100	100
Nortriptyline + AmB	0	0	0	100	100
DMSO 0.4%	0	0	0	0	0
Cont –	0	0	0	0	0
Cont +	6.25	50	100	100	100

^aCont –, negative control for death (nontreated embryos); Cont +, positive control for death (embryos treated with 3,4-dichloroaniline; see Material and Methods).

No obvious malformations were noticed in the zebrafish embryos exposed to either erythromycin, glimepiride, or nortriptyline alone or in combination with AmB when they were observed through the stereomicroscope. Similarly, these three AC or their combination with AmB did not induce significant changes on embryo hatching (see Table S1 in the supplemental material), as they hatched at a similar rate to the control group and within the physiological hatching period (usually between 48 and 60 hpf).

Erythromycin enhances the activity of AmB against Candida spp.

Among all of the compounds, we decided to focus on erythromycin because it did not increase the toxicity in cell lines nor zebrafish models, and it is a well-known macrolide antibiotic used to fight bacterial infections. We tested the effect against C. albicans, Candida glabrata, C. auris, Candida parapsilosis, Candida tropicalis, and Candida krusei.

The calculated FICI for all species, except for C. krusei, was <0.5 (Table 4), confirming that erythromycin also enhanced the AmB activity against most Candida species. For C. albicans, the synergistic effect was stronger with erythromycin concentrations ranging from 0.1 to 0.025 mM, which enhanced the activity of AmB at concentrations ranging from 0.03 to 0.12 μg/ml (Fig. 4A). When AmB (0.06 μg/ml) was combined with erythromycin (0.1 or 0.05 mM), there was a growth inhibition above 90% for C. albicans. Similar results were found for C. glabrata. For this species, erythromycin concentrations ranging from 0.1 to 0.025 mM combined with AmB (0.03 to 0.12 μg/ml) induced a statistically significant decrease in fungal growth (Fig. 4B). While AmB alone at concentrations of 0.03 and 0.06 μg/ml inhibited only around 7% and 35%, respectively, and addition of erythromycin (0.1, 0.05, and 0.025 mM) increased the inhibition to around 95%.

TABLE 4.

Fractional inhibitory concentration index for the combination of erythromycin with AmB against Candida spp.

Candida spp.	FICI of erythromycin with AmB
	50%	75%
C. albicans	0.33	0.33
C. glabrata	0.46	0.46
C. auris	0.43	0.37
C. parapsilosis	0.27	0.36
C. tropicalis	0.30	0.38
C. krusei	0.46	0.56

FIG 4.

Effect of the combination of erythromycin and AmB on C. albicans (A), C. glabrata (B), C. auris (C), C. parapsilosis (D), C. tropicalis (E), and C. krusei (F). In each case, the graph shows the effect of 0.1 (gray straight line, squares), 0.05 (gray straight line, triangles), and 0.025 mM (dotted gray line, inverted triangle) of erythromycin. Data shown in the graphs were obtained from the checkerboard assays (see Materials and Methods).

The synergic effect against C. auris was also tested, as this yeast is an important emergent pathogen that is resistant to different antifungals (33). As demonstrated for other species, the combination of AmB in concentrations ranging from 0.06 to 0.25 μg/ml with erythromycin ranging from 0.1 to 0.025 mM showed a synergistic effect against C. auris (Fig. 4C). AmB at 0.12 μg/ml only inhibited around 30% of the growth of C. auris; however, in combination with 0.1, 0.05, and 0.025 mM erythromycin, the inhibition increased to around 90%.

For C. parapsilosis and C. tropicalis, the synergistic effect was moderate (Fig. 4D and E). Finally, for C. krusei, the combination between both drugs did not show any synergistic effect (Fig. 4F).

Screening of compounds with antifungal activity against C. neoformans.

The experimental approach described above allowed the identification of drugs that inhibited the growth of C. neoformans in the control plates that did not contained subinhibitory concentrations of AmB. Although previous reports have already identified compounds with anticryptococcal activity using the same library, we decided to analyze our results to check if we could describe new compounds. Results from the new screening showed 43 AC that had inhibitory activity against C. neoformans (see Table S2 in the supplemental material). The majority of these compounds act in the central nervous system, followed by drugs belonging to cardiovascular therapeutic class and infectiology. To confirm the results of this screening, five AC were selected to be tested by sensitivity assays as described by EUCAST-AFST E.DEF 7.3 protocol using a range of concentrations from 0.4 to 0.0003 mM using four C. neoformans and four Cryptococcus gattii strains. Some of the compounds showed low MICs, such as ciclopirox ethanolamine (around 0.5 μg/ml), while others, such as triflupromazine, had higher MIC values (64 μg/ml). Hycanthone had very low activity against C. neoformans, with MIC values for both species above 64 μg/ml (Table 5).

TABLE 5.

MICs for the active compounds against C. neoformans and C. gattii isolates

Active compound	MIC (μg/ml)
	C. neoformans strain				C. gattii strain
	H99	KN99	CL0741	24067	NIH34	CBS10514	CBS10865	CL4999
Auranofin	4	2	4	2	2	2	2	2
Ciclopirox ethanolamine	0.5	0.5	0.5	0.5	0.5	0.5	0.5	0.5
Clofilium tosylate	32	32	32	32	16	16	16	16
Hycanthone	>64	>64	>64	>64	>64	>64	>64	>64
Perphenazine	16	32	32	32	32	16	32	16
Suloctidil	4	4	4	4	4	4	4	4
Tamoxifen	8	8	8	8	8	8	8	8
Thiethylperazine dimalate	32	32	32	32	32	32	32	32
Triclabendazole	8	8	8	8	8	8	8	8
Trifluoperazine dihydrochloride	16	16	16	16	16	16	16	16
Triflupromazine	64	64	32	32	64	32	64	32

DISCUSSION

The design of new antifungal treatments is a challenge in the field of medical mycology. In the last 10 years, there has been only one new antifungal licensed (isavuconazole). The changing epidemiology of fungal infections and their high incidence requires the identification and development of new drugs with antifungal activity. Since the commercialization of new molecules is a time-consuming process and very expensive, repurposing of off-patent drugs offers a feasible alternative to design new therapeutic approaches.

AmB shows the strongest antifungal activity among antifungals, but its administration is associated with toxicity. For this reason, in the present work, we attempted to identify new drugs that could enhance the activity of AmB at lower therapeutic doses that could reduce the toxicity at the same time. In a preliminary screening, we identified several compounds that could produce the desired effect. Furthermore, the synergism was confirmed by checkerboard assays. We also investigated the toxicity of the compounds in two different models, a mammalian cell line and the zebrafish embryo model, as a promising tool in toxicology. We showed that some of the compounds identified in this study as synergistic with AmB presented toxicity in these models at the concentrations used (such as chenodiol, chlorcyclizine, promazine, and riluzole). However, this did not limit their possible interest as synergistic compounds with AmB, since these compounds are already used in clinic for different purposes, and we only discarded compounds that showed increased toxicity when combined with AmB. This was the case for chlorcyclizine and cloperastine, which produced a slight increase in toxicity in the zebrafish model.

Glimepiride showed synergism with AmB and presented low toxicity in the in vitro and in vivo systems. This compound is used to treat diabetes mellitus type 2 by stimulating the release of insulin. However, it can cause side effects, (headache, nausea, and dizziness) due to endocrine imbalance. In our toxicity assays, we did not find any lethal toxicity in the combination of the drugs, but the fact that both AmB and glimepiride have side effects in humans indicate that this might not be the most suitable therapeutic option to treat cryptococcosis.

In the present study, one of the most promising compounds showing synergic activity with AmB was erythromycin. Either alone or in combination with AmB, erythromycin did not exert lethal toxicity in the in vitro or in vivo test models. In addition, no adverse effects on embryo development or hatching were evident after exposure to these drugs. Erythromycin is a macrolide antibiotic that inhibits protein synthesis in bacteria through binding to the ribosomal complex (34, 35), which could be a related mechanism for fungi. The mechanism by which this antibiotic enhanced the activity of AmB has not been studied yet. Nevertheless, it is known that AmB creates pores at the fungal membrane that increase permeability. In this way, AmB might promote influx of erythromycin into fungal cells, which could present an inhibitory effect for the fungal ribosomal complexes. In Saccharomyces cerevisiae and Schizosaccharomyces pombe, erythromycin inhibits mitochondrial protein synthesis by causing the separation of the tRNA from the mitochondrial ribosomes (36–38). Interestingly, AmB induces oxidative damage through a mechanism that involves the mitochondria (10–13), and inhibition of complex I conferred resistance to this antifungal (9). In consequence, it is reasonable to suggest that both drugs could synergistically increase mitochondrial damage.

Fungal infections often affect patients who are multitreated, and it is common that they receive antibiotics for a long period. For this reason, the fact that erythromycin enhances the antifungal effect of AmB might offer a feasible combination in clinic that eliminates fungal burden. Conversely, antibiotics can also decrease the bacterial population and in turn favor fungal replication; therefore, future studies using in vivo models are warranted. Another issue to consider is that due to their lipophilic properties and relatively high molecular weight, macrolides do not efficiently cross the blood-brain barrier (BBB) (39, 40), so their use during cryptococcal meningitis might be limited. However, in conditions of meningeal inflammation and disruption of the integrity of the BBB, their penetration into the central nervous system (CNS) might increase, so further investigations will be required to fully evaluate the role of erythromycin and AmB combination during cryptococcal meningitis. Conversely, erythromycin shows a good tissue distribution, so its use during candidemia and candidiasis might be a feasible strategy to improve the effectivity of AmB.

Chenodiol, which is used to dissolve gallstones (41–43), also showed synergy with AmB. This compound is a bile acid that acts as a detergent and dissolves lipid particles. Besides, it reduces cholesterol biosynthesis (44). Although fungi do not contain cholesterol in their membranes, it is possible that chenodiol reduces the main sterol of fungal membranes, ergosterol, either by direct binding and sequestration or by inhibition of its synthesis, which would increase the susceptibility to AmB.

The following three compounds showed 3 to 4 times more antifungal activity when combined with AmB: riluzole hydrochloride, nortriptyline hydrochloride, and promazine hydrochloride. However, they showed some toxicity in vitro and in vivo, either alone or in combination with AmB. These medicines act on the central nervous system (CNS). Interestingly, other antipsychotic drugs, such as quetiapine, and olanzapine and the antidepressant sertraline, have already demonstrated activity against C. neoformans with MIC values of 0.5 mg/ml for quetiapine and 0.25 mg/ml for olanzapine. Although we have not studied the mechanisms of action of the described compounds, there are already reports that antipsychotic drugs could act on the cellular membrane of eukaryotes increasing their permeability (45–47). In the case of sertraline, it shows potent in vitro anticryptococcal activity by blocking protein synthesis (48, 49). For other purposes, riluzole could inhibit primary cancer cell proliferation, probably by inducing the arrest of cells in the G₂/M phase (50). In addition, these authors described that riluzole hydrochloride also induced the production of reactive oxygen species (ROS) and apoptosis in liver cancer cells by inhibiting glutamate release.

Nisoldipine inhibited around 25% of the growth of C. neoformans, and when combined with AmB, the inhibitory effect increased to around 95%. However, there was a statistically significant increase in the cell toxicity when combined with AmB. Nisoldipine acts as a calcium channel blocker, and it has been previously reported to inhibit growth of some Cryptococcus spp., Candida spp., S. cerevisiae, and Aspergillus fumigatus (51). Another study demonstrated that another calcium channel blocker, nifedipine, showed a synergistic effect with itraconazole and was also active against itraconazole-resistant strains of A. fumigatus (52). Liu et al. combined four calcium channel blockers, amlodipine, nifedipine, benidipine, and flunarizine, with fluconazole and found synergism even against C. albicans fluconazole-resistant strains (53). These results suggest that inhibiting calcium channels might be another strategy to augment the efficacy of AmB.

In general, we found a good agreement between the toxicity data found in cell lines and in the zebrafish embryo model, although some differences were also detected, with some combinations and compounds more toxic in the zebrafish embryo model. These differences may be primarily due to the complexity of the zebrafish embryo system in which drugs usually show different toxicokinetics and undergo metabolism to some extent. In addition, zebrafish embryos have developed some of the main organs, such as heart, liver, and nervous system, by 72 hpf, which can be potential off-targets for the drugs. Therefore, the zebrafish model is a complementary model to the cell system for screening toxicity of drugs.

In our experimental approach, we always carried out a control plate with the compounds of the Prestwick Library without AmB, so it was unavoidable to identify compounds that inhibited the growth of C. neoformans. We found 43 drugs that inhibited the growth of C. neoformans. From these compounds, we selected 11 to confirm the antifungal activity against different C. neoformans and C. gattii strains. Anticryptococcal activity was confirmed for 10 compounds (Table 5; see also Table S2 in the supplemental material). Previous studies have also identified compounds among the Prestwick Chemical Library with fungicidal effect against C. neoformans. Butts et al., using the Prestwick library, described 31 compounds active against C. neoformans through the adenylate kinase release approach (28). When compared with those results, we found 12 compounds in common, thioridazine, perhexiline maleate, chlorprothixene, fluspirilene, trifluoperazine, methiothepin maleate, prochlorperazine dimaleate, suloctidil, thonzonium bromide, clomiphene, tamoxifen, and thiethylperazine dimalate. Suloctidil was one of the compounds with the lowest MIC value in both studies. This drug was also active against C. albicans in planktonic and biofilm cells in vitro and was also able to reduce fungal burden in vivo (28, 54). Recently, Truong et al. (51) used the repurposing strategy to identify AC against Cryptococcus deuterogattii. The authors identified 54 active drugs using the Enzo drug library that contains 640 off-patent drugs. Among them, they suggested the use of flubendazole, an anthelmintic drug, as a new anticryptococcal drug due to its activity against all pathogenic Cryptococcus spp., including isolates resistant to fluconazole. They also identified that the calcium channel blockers nifedipine, felodipine, and nisoldipine are active against all pathogenic Cryptococcus spp. (51).

In the present work, we have also identified other compounds that were not previously reported by other authors using the same chemical library. There are several explanations for this discrepancy, but differences in the experimental conditions, such as the concentration used, or the detection method are key points. Another major difference is the screening strategy, as we were interested in compounds with effects on fungal growth (fungicides and/or fungistatic). Nevertheless, previous reports identified drugs with fungicidal activity but using an assay based on adenylate kinase release from the cells. Expanding the screening for compounds with both fungicidal and fungistatic activity might have highlighted more compounds, as some of the antifungals used in clinic are fungistatic or even can behave as fungistatic or fungicidal depending on the species.

The compound that showed lowest MICs against C. neoformans and C. gattii was ciclopirox ethanolamine (0.5 μg/ml). Ciclopirox ethanolamine is an antifungal agent used topically to treat a variety of fungal infections, mainly dermatophytosis, seborrheic dermatitis, and cutaneous candidiasis (55). It acts through the chelation of polyvalent metal cations, such as Fe³⁺ and Al³⁺. Its activity was tested in vitro against distinct fungal species (56, 57), but its use for the treatment of systemic mycosis has not yet been reported.

Auranofin also showed activity against C. neoformans with an MIC value of 4 μg/ml. This compound is used for treatment of rheumatoid arthritis. Our results are in agreement with previous findings that reported that this drug was active against some fungal species (23, 58). Although the inhibitory mechanism is not known, it has been suggested that auranofin exerts its action through the production of reactive-oxygen-mediated cell death.

Tamoxifen has also been described for its antifungal activity by different groups (59–63). It is an anticancer drug that acts as an antiestrogen in the mammary tissue. The mechanism of action still is not fully known, but it is known that this drug can bind to calmodulin (63). This protein is required for growth at high temperature in C. neoformans (64).

Triclabendazole is another compound also presenting activity against C. neoformans and C. gattii (65) with MICs around 4 to 8 μg/ml. It is an anthelminthic drug that binds to the β-tubulin molecule, causing an ultrastructural disruption by impairing the maintenance of the integrity of the surface membrane (66, 67). Other antifungals, such as griseofulvin, carbendazim, and thiabendazole, bind to tubulin and thus prevent microtubule polymerization. Also, there are a great number of patents of distinct molecules that have tubulin as an alternative target with antifungal activity against a great diversity of pathogenic fungi (68).

In conclusion, drug repurposing showed to be an efficient tool to find potential drugs that could enhance the AmB activity and helped to find new active drugs against several pathogenic yeasts. In this sense, the combination of erythromycin or glimepiride with AmB reduced the antifungal effective concentration while showing no adverse effects in the toxicity assays, which included an in vitro test with the mammalian cell system and an in vivo toxicity test with zebrafish embryos. The synergism between AmB and erythromycin could be a new promising drug combination to improve the current antifungal therapy against several invasive fungal diseases. Finally, we also identified drugs with anti-Cryptococcus activity, such as ciclopirox or auranofin, which should be further characterized to be considered as potential treatments for cryptococcal disease.

MATERIALS AND METHODS

Strains and growth conditions.

The Cryptococcus neoformans H99 strain (69) was used for the initial screening. Cryptococcus neoformans strains KN99, CL0741, and 24067 and C. gattii strains NIH34, CBS10514, CBS10865, and CL4999 were also used. The following Candida spp. strains were also used: C. albicans SC5314 (70), C. glabrata CL-9555, C. tropicalis CL-10621, C. krusei ATCC 6258, C. parapsilosis ATCC 22019, and C. auris CL-10013. The yeast strains were incubated in Sabouraud medium (liquid or solid containing 1.5% agar; Oxoid) at 30°C.

Antifungals and chemical library.

We used the Prestwick Chemical Library to identify drugs that potentially enhance the activity of AmB. This library contains 1,280 off-patent drugs approved by the Food and Drug Administration (FDA), European Medicines Agency (EMA), and other agencies. The library was provided in 96-well format, each containing 100 μl of each compound at a concentration of 10 mM in DMSO (dimethyl sulfoxide). In addition, we also used amphotericin B (Sigma-Aldrich).

Antifungal susceptibility testing.

The MIC of AmB to the C. neoformans H99 strain was tested according to antifungal susceptibility testing (AFST) following EUCAST-AFST E.DEF 7.3 protocol (71). Briefly, C. neoformans cells were inoculated in Sabouraud agar plates for 48 h, and a suspension of 1 × 10⁵ to 5 × 10⁵ cells/ml was prepared in distilled H₂O. Antifungal susceptibility plates were prepared with a range of AmB concentrations from 16 to 0.03 μg/ml (1/2 dilutions) in RPMI medium containing 2% glucose and buffered at pH 7.0 with 165 mM morpholinepropanesulfonic acid (MOPS). One hundred microliters of the C. neoformans suspensions were added to the antifungal susceptibility plates (prepared as 2× stocks) and incubated at 35°C without shaking for 48 h. After incubation, the optical density (OD) was measured at 530 nm, and the MIC was calculated. MIC was defined as the antifungal concentration that produced a 90% inhibition of growth compared to that of the control well without the antifungal. The strains Candida parapsilosis ATCC 22019 and C. krusei ATCC 6258 were used as quality control isolates in the AFST experiments. This experiment was performed three times (triplicate).

Susceptibility of fungal strains to selected compounds from the Prestwick Chemical Library was evaluated as described above (EUCAST-AFST E.DEF 7.3 protocol). The concentration range for all of the tested compounds was from 64 to 0.12 μg/ml.

Screening of drugs from the Prestwick Chemical Library which enhance the activity of amphotericin B on antifungal activity.

The use of the Prestwick Chemical Library in our conditions allowed the identification of compounds that showed synergism with amphotericin B but also of off-patent drugs that had antifungal activity against C. neoformans. The original stocks of the 1,280 compounds present in the Prestwick Chemical Library were diluted (1:10) to obtain a concentration of 1 mM in intermediate plates with 2.2× RPMI medium supplemented with glucose. From these intermediate plates, 20 μl of each well was again diluted (1:5) in new plates prepared with 80 μl of 0.06 μg/ml AmB in 2× RPMI medium supplemented with 4% glucose per well. Parallel plates without AmB were carried out as control to test the activity of the compounds alone. These plates were then inoculated with 100 μl of a C. neoformans suspension (1 × 10⁵ to 5 × 10⁵ cells/ml) to achieve a final concentration of 100 μM and 0.03 μg/ml of the AC and AmB (subinhibitory for C. neoformans), respectively. In the final plates, columns 1 and 12 did not contain any compound, so they were used to include the following controls: (i) sterility controls, (ii) AmB alone at subinhibitory (0.03 μg/ml) and at inhibitory (0.25 μg/ml) concentrations, and (3) a growth control with 2× RPMI medium supplemented with 2% glucose and 1% DMSO.

After 48 h of incubation at 35°C, readings at 530 nm were performed. We scored as negative those drugs that did not have any effect on C. neoformans by themselves (<40% inhibition), and positive AC were those that in combination with AmB 0.03 μg/ml induced more than an 80% decrease of growth.

In the screening described above, one of the control plates contained only the compounds from the Prestwick Library. For this reason, it was unavoidable to also identify compounds that caused almost full inhibition of C. neoformans growth (more than 80% of growth inhibition). The inhibition of these compounds was further tested in microdilution plates using EUCAST protocol and a concentration range of 64 to 0.12 μg/ml in 96-wells plates, prepared as described above. The inoculum was prepared as described above, and the results were obtained after 48 h of incubation at 35°C through spectrophotometer readings at 530 nm.

Checkerboard assay.

To confirm the data obtained in the screening for drugs that enhance the activity of AmB, checkerboard assays were performed expanding the concentration range for the AC from 0.4 to 0.0003 mM and for AmB from 0.5 to 0.008 μg/ml.

For this purpose, each compound or AmB solutions where prepared at 50× in DMSO. So, the solutions were diluted in RPMI (1:50) plates to ensure that, in the assay, DMSO concentration did not exceeded 1%. Finally, 50 μl from each solution concentration was mixed in the final assay plate. The C. neoformans inoculum was prepared as described above in the antifungal susceptibility testing section following the EUCAST-AFST E.DEF 7.3 protocol. The OD at 530 nm was measured after 48 h of incubation at 35°C, and the fractional inhibitory concentration index (FICI) was calculated according to the following equation: ΣFIC = FIC (compounds) + FIC (AmB), where the FIC is the ratio of the MIC of the combination with the MIC alone.

The combination of AmB with the AC was considered synergic when the FICI is ≤0.5, indifferent when FICI is >0.5 and ≤4, and antagonist when FICI is >4 (72, 73).

Toxicity assays. (i) In vitro cytotoxicity assay.

In vitro cytotoxic activity was evaluated using the macrophage cell line RAW 264.7. Macrophages (10⁵ cells/well) were exposed to 0.1 mM of each compound alone or combined with 0.03 μg/ml AmB and incubated at 37°C and 5% CO₂ for 24 h. Cytotoxicity was obtained by measuring the release of lactate dehydrogenase (LDH) to the medium using the CytoTox 96 nonradioactive cytotoxicity assay kit (Promega) following the manufacturer’s recommendations. Untreated cells (viable cells), dead cells (with lysis solution), and wells with only medium were used to determine the background LDH activity and served as controls. Results were represented as percentage of mortality in comparison to the wells in which full death was induced with the lysis buffer provided in the kit.

(ii) In vivo toxicity assessment with zebrafish embryos.

Fish husbandry and embryo collection. Zebrafish (Danio rerio) progenitors were wild type, obtained from a local pet store and maintained at standard laboratory conditions of 26°C on a 14/10 h dark/light photoperiod in a recirculation system for at least 2 months before using them as progenitors. The water was prepared according to UNE-EN ISO 7346-3:1998. Fishes were fed twice a day with a commercial diet (Zeigler), and every 2 days the diet was complemented with living Daphnia magna.

One day before the experiment, parent animals were separated from the rest and caged in tanks (one female and two males) overnight. Spawning was induced when the light was turned on the following morning. Each spawning was examined for mortality, and when lower than 15%, the eggs were accepted for their use in the toxicity tests.

Waterborne exposure of zebrafish embryos. The zebrafish eggs from two spawning were washed with dilution water (UNE-EN ISO 7346-3:1998 or ISO 15088:2007). Only fertilized eggs, which staged at 2 to 4 cells were disposed into 96-well plates with 200 μl of the test solution (n = 16 embryos/test solution). Plates were covered and incubated in a calibrated climate chamber at 28.5 ± 0.1°C with a 14 h/10 h light/darkness cycle and for 96 h, with fresh medium renovation every 24 h.

Test solutions included the negative control with dilution water, the solvent control with 0.4% DMSO, the positive control with 3,4-dichloroaniline (4 μg/ml in dilution water), and the drug solutions. Chenodiol, nisoldipine, chlorcyclizine hydrochloride, cloperastine hydrochloride, erythromycin, glimepiride, promazine hydrochloride, riluzole hydrochloride, and nortriptyline hydrochloride were either tested alone at 0.1 mM or combined with 0.03 μg/ml AmB.

Dissolved oxygen, pH, and conductivity were assessed in freshly prepared test solutions to comply with test solution requirements according to OECD 236 recommendations (32). We included the following test solutions: negative control with dilution water, solvent control with 0.4% DMSO, and positive control with 3,4-dichloroaniline (4 μg/ml in dilution water).

Toxicological assessment. Zebrafish mortality and development was assessed at 0, 8, 24, 48, 72, and 96 h post fertilization (hpf) through stereomicroscope observation. Hatching was evaluated by visual inspection of all zebrafish embryos every 3 h from 48 hpf to 60 hpf.

Statistical analysis.

To evaluate the significance of the results of the cytotoxicity assay, we used analysis of variance (ANOVA) followed by Tukey test using the GraphPad Prism 6 (GraphPad Software Inc., San Diego, CA).