The Antifungal and Synergistic Effect of Bisphosphonates in Cryptococcus

New treatment strategies are required for cryptococcosis, a leading mycosis in HIV-AIDS patients. Following the identification of Cryptococcus proteins differentially expressed in response to fluconazole, we targeted farnesyl pryrophosphate synthetase (FPPS), an enzyme in the squalene biosynthesis pathway, using nitrogenous bisphosphonates. We hypothesized that these would disrupt squalene synthesis and thereby produce synergy with fluconazole, which acts on a downstream pathway that requires squalene.

ABSTRACT

New treatment strategies are required for cryptococcosis, a leading mycosis in HIV-AIDS patients. Following the identification of Cryptococcus proteins differentially expressed in response to fluconazole, we targeted farnesyl pryrophosphate synthetase (FPPS), an enzyme in the squalene biosynthesis pathway, using nitrogenous bisphosphonates. We hypothesized that these would disrupt squalene synthesis and thereby produce synergy with fluconazole, which acts on a downstream pathway that requires squalene. The susceptibilities of 39 clinical isolates from 6 different species of Cryptococcus were assessed for bisphosphonates and fluconazole, used both independently and in combination. Effective fluconazole-bisphosphonate combinations were then assessed for fungicidal activity, efficacy against biofilms, and ability to resolve cryptococcosis in an invertebrate model. The nitrogenous bisphosphonates risedronate, alendronate, and zoledronate were antifungal against all strains tested. Zoledronate was the most effective (geometric mean MIC = 113.03 mg/liter; risedronate = 378.49 mg/liter; alendronate = 158.4 mg/liter) and was broadly synergistic when combined with fluconazole, with a fractional inhibitory concentration index (FICI) of ≤0.5 in 92% of isolates. Fluconazole and zoledronate in combination were fungicidal in a time-kill assay, inhibited Cryptococcus biofilms, prevented the development of fluconazole resistance, and resolved infection in a nematode model. Supplementation with squalene eliminated bisphosphonate-mediated synergy, demonstrating that synergy was due to the inhibition of squalene biosynthesis. This study demonstrates the utility of targeting squalene synthesis for improving the efficacy of azole-based antifungal drugs and suggests bisphosphonates are promising lead compounds for further antifungal development.

INTRODUCTION

Cryptococcal meningitis is a devastating infection of the central nervous system and a fatal AIDS-related pathology, causing approximately 180,000 deaths every year (1).While a majority of infections are caused by C. neoformans, the C. gattii species complex is of increasing concern due to its ability to infect immunocompetent hosts and its propensity to cause outbreaks (2, 3).

The preferred first-line treatment of cryptococcal infections is induction therapy with amphotericin B and 5-flucytosine, followed by maintenance therapy with fluconazole (4). Amphotericin B and 5-flucytosine operate on unrelated pathways yet produce strong synergistic Cryptococcus inhibition, and have been used to treat cryptococcal meningitis for decades (5, 6). Amphotericin B, however, is highly hepatotoxic and treatment requires intensive monitoring, while 5-flucytosine can be expensive and is not licensed for distribution in areas with the greatest need (7, 8). Fluconazole is sometimes used as both an induction and a maintenance therapy in developing regions; however, it has limited ability to fully resolve cryptococcosis, and resistance is increasing (9–12). Enhancing fluconazole with synergists is considered a promising approach to improve its antifungal efficacy, while limiting the development of resistance (13).

In a prior study, our group examined the proteomic response of Cryptococcus to fluconazole and found farnesyl pyrophosphate synthase (FPPS) was significantly upregulated during treatment (14). Within the mevalonate pathway, FPPS catalyzes the condensation of dimethylallyl pyrophosphate to form farnesyl pyrophosphate, the direct precursor to squalene (15). Squalene in turn feeds into the ergosterol biosynthesis pathway, which is targeted by fluconazole (16).

FPPS can be inhibited by nitrogenous bisphosphonates, an FDA-approved class of drugs that is used for the treatment of low bone density disorders, including osteoporosis and Paget’s disease (17, 18). Bisphosphonates have low toxicity, a high clinically achievable dose, and protective immunomodulatory activity (19, 20). Nonnitrogenous bisphosphonates are prescribed for similar indications, but have an FPPS-independent mechanism of action (21). The antifungal activity of bisphosphonates has not yet been explored.

In this study, we hypothesized that by targeting squalene synthesis and therefore reducing the precursor for the synthesis of ergosterol, nitrogenous bisphosphonates would exhibit antifungal activity. Furthermore, we reasoned that by attacking two independent points on sequential biosynthetic pathways, the combined effect of fluconazole and nitrogenous bisphosphonates would be synergistic. The aims of this study were therefore to demonstrate that nitrogenous bisphosphonates are antifungal against pathogenic Cryptococcus species and act synergistically with fluconazole, thereby providing a promising strategy for novel anti-cryptococcal therapies.

RESULTS

Nitrogenous bisphosphonates inhibit Cryptococcus and can synergize with fluconazole.

Nitrogenous bisphosphonates were tested for their ability to inhibit a diverse range of Cryptococcus species and strains. The MICs for fluconazole (FLC) and the nitrogenous bisphosphonates risedronate, alendronate, and zoledronate (RIS, ALN, and ZOL, respectively) in the 39 Cryptococcus isolates are presented in Table 1. The MIC ranges and geometric means for FLC, ALN, RIS, and ZOL across all 39 Cryptococcus isolates were as follows: FLC, 0.25 to 64, 3.06 mg/liter, respectively; RIS, 128 to 512, 378.49 mg/liter; ALN, 128 to 256, 158.43 mg/liter; and ZOL, 64 to 256, 113.03 mg/liter. ALN and ZOL were more inhibitory than RIS (P < 0.0001), and ZOL was more effective than ALN (P < 0.001). Cryptococcus deuterogattii was the most bisphosphonate-susceptible species (P < 0.05), despite containing strains with the highest FLC MICs.

TABLE 1.

MIC, MIC_C, dose reduction, and FICI data for fluconazole and nitrogenous bisphosphonates against Cryptococcus species and strains^a

Species and strain					FLC:RIS				MICC (mg/liter)
									FICI	FLC:ALN				FICI	FLC:ZOL				FICI
	MIC (mg/liter)				FLC		RIS			FLC		ALN			FLC		ZOL
	FLC	RIS	ALN	ZOL	MICC	Δ	MICC	Δ		MICC	Δ	MICC	Δ		MICC	Δ	MICC	Δ
C. neoformans
571 216	0.25	512	256	128	0.13	2	32	16	0.73	0.13	2	32	8	0.77	0.06	4	16	8	0.46
WM556	0.5	512	128	128	0.25	2	32	16	0.56	0.06	8	32	4	0.5	0.13	4	16	8	0.42
WM385	4	512	256	256	2	2	32	16	0.72	2	2	64	4	0.71	0.5	8	64	4	0.38
H99	4	512	128	128	2	2	32	16	0.5	1	4	32	4	0.67	1	4	16	8	0.42
571 198	4	512	256	128	2	2	32	16	0.58	2	2	64	4	0.63	1	4	32	4	0.54
C3-1	4	256	128	128	1	4	64	4	0.54	1	4	32	4	0.5	1	4	16	8	0.38
WM625	8	256	256	256	4	2	32	8	0.67	4	2	32	8	0.58	2	4	32	8	0.46
Geometric mean	2.21	420	190	156	1	–	35.3	–	0.61	0.74	–	39	–	0.62	0.5	–	23.8	–	0.43
C. deneoformans
571 257	0.25	512	128	128	0.13	2	64	8	0.63	0.13	2	32	4	0.92	0.06	4	32	4	0.42
WM629	0.25	512	256	128	0.13	2	64	8	0.67	0.06	4	64	4	0.46	0.03	8	32	4	0.38
JEC21	0.5	512	128	128	0.13	4	32	16	0.56	0.25	2	32	4	0.63	0.13	4	32	4	0.5
JEC20	1	512	256	128	0.5	2	64	8	0.6	0.25	4	64	4	0.5	0.25	4	32	4	0.54
Geometric mean	0.42	512	181	128	1.18	–	53.8	–	0.61	0.15	–	45.3	–	0.6	0.09	–	32	–	0.46
C. gattii
Q00	1	512	128	128	0.5	2	64	8	0.6	0.5	2	16	8	0.54	0.25	4	16	8	0.46
PNG27	2	512	128	128	0.5	4	64	8	0.42	0.5	4	32	4	0.58	0.25	8	32	4	0.38
R794	2	512	128	128	1	2	64	8	0.77	1	2	32	4	0.67	0.5	4	16	8	0.38
V7	2	512	128	128	1	2	128	4	0.88	0.5	4	16	8	0.42	0.5	4	16	8	0.42
PNG20	4	512	256	128	1	4	64	8	0.5	2	2	32	8	0.58	1	4	32	4	0.46
NT-9	8	256	128	64	2	4	128	2	0.67	2	4	64	2	0.63	1	8	16	4	0.33
V20	8	256	128	128	4	2	64	4	0.75	1	8	32	4	0.5	2	4	16	8	0.38
Geometric mean	2.97	420	141	116	1.1	–	78	–	0.64	0.91	–	29	–	0.55	0.61	–	19.5	–	0.4
C. deuterogattii
MK914	2	512	128	128	1	2	128	4	0.6	0.25	8	16	8	0.42	0.5	4	16	8	0.31
R265	4	512	256	128	2	2	64	8	0.52	1	4	64	4	0.46	0.5	8	32	4	0.33
V5	8	256	128	128	4	2	64	4	0.67	2	4	32	4	0.54	1	8	16	8	0.26
03-201073	16	128	128	64	4	4	64	2	0.67	8	2	32	4	0.63	4	4	16	4	0.67
14.1433	16	128	128	64	4	4	64	2	0.63	2	8	32	4	0.38	2	8	8	8	0.25
14.1431	32	256	128	64	4	8	64	4	0.46	4	8	64	2	0.54	4	8	4	16	0.25
97/170	64	128	128	64	4	16	64	2	0.47	4	16	32	4	0.38	4	16	8	8	0.23
Geometric mean	11.9	232	141	86.1	2.97	–	70.7	–	0.57	2	–	35.3	–	0.47	1.64	–	11.9	–	0.31
C. tetragattii
Bt201	2	512	128	128	0.5	4	128	4	0.54	0.5	4	32	4	0.46	1	2	8	16	0.44
B5742	2	512	128	128	1	2	64	8	0.6	0.5	4	32	4	0.5	0.5	4	16	8	0.33
MMRL2651	4	512	256	64	2	2	64	8	0.67	1	4	64	4	0.54	0.5	8	16	4	0.38
WM779	4	512	128	128	2	2	64	8	0.63	2	2	32	4	0.63	1	4	16	8	0.33
MMRL2650	8	256	128	64	2	4	64	4	0.58	2	4	64	2	0.75	1	8	16	4	0.31
M27056	8	256	128	128	2	4	64	4	0.67	4	2	16	8	0.6	2	4	32	4	0.5
V00869	8	256	128	64	4	2	32	8	0.58	4	2	8	16	0.56	1	8	16	4	0.31
Geometric mean	4.42	380	141	95.1	1.64	–	64	–	0.61	1.49	–	29	–	0.57	0.91	–	16	–	0.37
C. bacillisporus
NIH179	1	512	128	128	0.5	2	64	8	0.6	0.25	4	32	4	0.5	0.13	8	32	4	0.46
571 159	2	512	128	128	0.5	4	128	4	0.46	1	2	32	4	0.63	0.5	4	8	16	0.38
97/427	2	512	128	128	1	2	128	4	0.63	1	2	16	8	0.6	0.5	4	16	8	0.33
WM161	2	512	256	128	1	2	32	16	0.56	1	2	32	8	0.54	1	2	8	16	0.48
B13C	2	512	256	128	0.5	4	128	4	0.5	1	2	32	8	0.63	0.13	16	8	16	0.33
NIH184	4	256	256	128	1	4	128	2	0.67	2	2	32	8	0.54	1	4	16	8	0.38
PNG30	8	256	128	64	1	8	128	2	0.52	2	4	32	4	0.58	2	4	16	4	0.42
Geometric mean	2.44	420	172	116	0.74	–	95.1	–	0.56	1	–	29	–	0.57	0.5	–	13.1	–	0.39

^aMIC, minimum inhibitory concentration (mg/liter); MIC_C, MIC in combination (mg/liter); FICI, fractional inhibitory concentration index (synergistic combinations are italicized); FLC, fluconazole; RIS, risedronate; ALN, alendronate; ZOL, zoledronate; Δ, fold decrease from MIC of drug alone to MIC_C of drug in combination; –, not applicable.

Synergy was tested by checkerboard for combinations of FLC:RIS, FLC:ALN, and FLC:ZOL, and the MICs-in-combination (MIC_Cs) and fractional inhibitory concentration indices (FICIs) are reported in Table 1. Across all isolates, combining FLC and a bisphosphonate reduced the inhibitory dosage of both drugs 2- to 16-fold. FLC:ZOL was significantly more synergistic than FLC:ALN or FLC:RIS (P < 0.001), with an FICI ≤ 0.5 observed in 92% of isolates. There was no significant difference in synergy between FLC:ALN and FLC:RIS.

Fluconazole and zoledronate are fungicidal in combination.

To determine whether combining FLC and ZOL potentiated fungicidal activity, time-kill assays were performed for FLC, ALN, ZOL, FLC:ALN, and FLC:ZOL in C. neoformans H99 and C. deuterogattii V5 (Fig. 1). While FLC, ALN, and ZOL all had fungistatic inhibition of the two strains, FLC:ALN and FLC:ZOL combinations were fungicidal in a concentration-dependent manner at and above the MIC_C (Table 1). FLC:ZOL was significantly more fungicidal than FLC:ALN at 24 h (P < 0.05). There was no significant difference in the fungicidal action of FLC:ZOL between C. neoformans H99 and C. deuterogattii V5.

FIG 1.

Combinations of fluconazole and nitrogenous bisphosphonates are fungicidal at or above their MIC_C in a time-kill assay. C. neoformans H99 and C. deuterogattii V5 cultures were incubated with combinations of fluconazole and the indicated bisphosphonate at 1× (green triangles), 2× (orange triangles), and 4× (light blue diamonds) MIC_C (Table 1) or with FLC (yellow circles) or bisphosphonate (blue square) at MIC. The red diamonds represent an untreated no-drug control. Viable cell counts were determined by back-plating after 0, 3, 6, 12, 24, 48, and 72 h of treatment. FLC, fluconazole; ALN, alendronate; ZOL, zoledronate.

Squalene supplementation prevents zoledronate-mediated inhibition.

To verify that bisphosphonate-mediated killing was due to the inhibition of squalene biosynthesis and not due to an off-target effect, Cryptococcus cultures treated with ZOL at the MIC and FLC:ZOL at the MIC_C were supplemented with squalene to determine if this could restore growth (Fig. 2). FLC at MIC_C was subinhibitory in C. neoformans strain H99, C. deuterogattii strain V5, and C. deuterogatii strain 97/170. Squalene prevented inhibition by ZOL in a dose-dependent manner, with a 50% effective concentration (EC₅₀) for C. neoformans H99, C. deuterogattii V5, and C. deuterogatii 97/170 of 31.64, 6.49, and 6.19 mg/liter, respectively. For FLC:ZOL inhibition, this was 10.56, 2.75, and 3.02 mg/liter, respectively. The role of FPPS inhibition in bisphosphonate-mediated killing was further validated by testing nonnitrogenous bisphosphonates etidronate and clodronate for anti-cryptococcal activity. Neither exhibited any antifungal activity, even at the highest concentration tested, 512 mg/liter (not shown).

FIG 2.

The addition of squalene rescues ZOL-mediated inhibition of Cryptococcus. Suspensions of C. neoformans H99, C. deuterogattii V5, and the FLC-resistant C. deuterogattii strain 97/170 were treated with FLC (yellow circles) and ZOL (blue squares) at the MIC_C of each strain (Table 1) both separately and in combination (green triangles). Red diamonds represent the no-drug control. Squalene was added to each treatment group at the indicated concentrations. Data are presented relative to a no-drug control without FLC, ZOL, or squalene and represent the means from three biological replicates ± standard error of the mean (SEM).

FLC:ZOL inhibits Cryptococcus biofilms.

Organization of microbes into biofilms impairs their sensitivity to antimicrobial stressors (22). Figure 3 shows the dose-dependent inhibition of the metabolic activity of C. neoformans, C. deuterogattii, and mixed-species biofilms. FLC and ZOL alone were not able to fully inhibit the metabolic activity of established C. neoformans biofilms at the tested concentrations (64× MIC = 256 mg/liter FLC and 8,192 mg/liter ZOL), while the FLC:ZOL combination fully inhibited C. neoformans biofilm activity at 32 × MIC_C (= 32 mg/liter FLC and 512 mg/liter ZOL). C. deuterogattii biofilm activity was fully inhibited at 32× MIC FLC (256 mg/liter) and 64× MIC ZOL (8,192 mg/liter), while FLC:ZOL produced total inhibition at 16× MIC_C (=16 mg/liter FLC, 256 mg/liter ZOL). Mixed-species biofilms were substantially less susceptible to FLC, ZOL, and FLC:ZOL treatment than their single-species counterparts, and were not entirely inhibited by any drug combination. Importantly, however, while biofilm activity was enhanced by the lower concentrations of FLC and ZOL, this was not seen in the FLC:ZOL combination treatment.

FIG 3.

Fluconazole:zoledronate combinations inhibit biofilms formed by single and mixed Cryptococcus species. Biofilms formed by C. neoformans H99, C. deuterogattii V5, and both species mixed together were treated with 2-fold multiples of the MIC of FLC (yellow circles) or ZOL (blue squares), or the MIC_C of FLC:ZOL (green triangles). Biofilm metabolic activity was quantified by XTT reduction assay. Data are the means from 4 biological replicates ± standard deviation (SD).

Combining fluconazole and zoledronate suppresses the development of resistance.

Combining antimicrobials usually lowers the probability of resistance developing to either agent (23). When C. neoformans H99 and C. deuterogattii V5 strains were cultured in subinhibitory concentrations of FLC and ZOL and passaged into increasing drug solutions, cells adapted rapidly to tolerate extremely high drug concentrations (Fig. 4). When propagated in FLC:ZOL, resistance was totally prevented at 4× MIC_C for C. neoformans H99 and 5× MIC_C for C. deuterogattii V5. The differences between combined and single-drug treatments were significant for both strains (P < 0.0001).

FIG 4.

Combining fluconazole with zoledronate retards the development of resistance to both agents. Actively growing cultures of C. neoformans H99 and C. deuterogattii V5 were treated with subinhibitory concentrations of agents (0.25× MIC for FLC [yellow circle] or ZOL [blue square]; 0.25× MIC_C for FLC:ZOL [green triangle]; Table 1) and passaged through increasing concentrations until 8× MIC or MIC_C was reached. Viability was assessed after 48 h of each treatment. The red diamonds represents a drug-free control. Data are the means from three replicates ± SEM.

FLC:ZOL resolves cryptococcosis in an invertebrate model of infection.

Caenorhabditis elegans nematodes provide an established rudimentary model of infection that enables in vivo assessment of anti-cryptococcal compounds (24) (Fig. 5). Infected C. elegans populations were rescued by treatment with combined FLC:ZOL (4 mg/liter FLC, 32 mg/liter ZOL), which restored endpoint survival to a level that matched (for C. deuterogattii V5 and C. deuterogattii 97/170) or almost matched (for C. neoformans H99) that observed with the avirulent Escherichia coli control (P < 0.001, P < 0.0001, P < 0.01, respectively).

FIG 5.

FLC:ZOL resolves Cryptococcus infection in a nematode model. Freshly hatched AU37 Caenorhabditis elegans nematodes were fed on Cryptococcus or E. coli (for an uninfected control) for 18 h at 25°C. Infected nematodes were rinsed and transferred into BHI broth supplemented with no-drug vehicle (red) or with 16 μg/ml FLC (yellow), 128 μg/ml ZOL (blue), or 4 μg/ml fluconazole plus 32 μg/ml zoledronate (green). The combined treatment reduced mortality to match (C. deuterogattii strains) or nearly match (C. neoformans) the uninfected control. Kaplan-Meier survival plots are representative of two independent experiments.

DISCUSSION

This study has demonstrated that zoledronate is consistently and potently synergistic with fluconazole across a comprehensive collection of strains and species of Cryptococcus, and that this synergy is due to the combined inhibition of squalene and ergosterol biosynthesis. Targeting sequential biosynthetic pathways is known to often produce drug synergy, and a number of recent studies have examined the synergistic effect of combining azole antifungals with inhibitors of enzymes in the mevalonate or ergosterol pathways. Terbinafine inhibits squalene epoxidase, and synergy with fluconazole has been reported in a wide array of pathogens in both in vitro and pharmacokinetic studies (25–27). However, synergy was often relatively weak, with an FICI close to 0.5, and was not consistently seen for all strains of the tested species. Statins, which are commonly prescribed to control human cholesterol levels, inhibit HMG-CoA reductase in the mevalonate pathway and synergy with azoles has been observed for a variety of fungal pathogens, including some isolates of Cryptococcus (28). Unfortunately, their usefulness is limited by a narrower spectrum of antifungal activity and adverse off-target effects that are toxic to already-weakened mycosis patients (29–31). Finally, morpholine antifungals like amorolfine inhibit Δ¹⁴-sterol reductase in the ergosterol biosynthesis pathway (32), but while amorolfine has been successfully applied with azoles as a topical treatment for onychomycosis, systemic use for more aggressive pathogens like Cryptococcus is unlikely (33–35). Together these studies demonstrate the value of targeting sequential pathways, but also show that different targets and inhibitors produce substantial variations in synergistic outcome.

Although all of the nitrogenous bisphosphonates tested target FPPS, zoledronate was consistently more active as an antifungal and a synergist than alendronate or risedronate. This superior activity parallels the relative potency demonstrated by zoledronate at inhibiting osteoclasts, where it is at least five times more effective than other bisphosphonates (21). The presence of an imidazoline ring that is distal to the bisphosphonate head in zoledronate interacts with basic amino acids in human FPPS, enabling tighter binding than the less polar amine and pyridine rings present in alendronate and risedronate, respectively (36, 37). FPPS in both C. neoformans and C. deuterogattii has cognate basic amino acids at the same location, and improved ligand binding may be the cause of the greater antifungal potency of zoledronate. Small, polar inhibitors are also better at passively diffusing across the cell membrane, so the more polar zoledronate may also be more able to penetrate the cell and access its target enzyme than the other agents (38).

The fluconazole-zoledronate combination produced significant synergy in over 92% of the Cryptococcus strains tested and this was independent of fluconazole susceptibility; indeed the highly fluconazole-resistant isolates of C. deuterogattii had among the most significant FICI values and their MIC_C for fluconazole was reduced to clinically achievable levels. Species in the C. gattii complex largely infect immunocompetent hosts and it is unlikely that the resistant C. deuterogattii isolates had been exposed to prophylactic fluconazole (39, 40), hence their reduced fluconazole susceptibility is likely to be intrinsic to their physiology and not acquired (41). As fluconazole-zoledronate synergy appears to work regardless of fluconazole susceptibility, this suggests that intrinsic resistance is not a generalized response to chemical stressors and that it can be reversed by depriving the cell of squalene. Furthermore, the presence of zoledronate strongly inhibited the capacity to acquire fluconazole resistance under conditions where resistance was rapidly induced by fluconazole monotherapy. These data show that zoledronate-mediated synergy can be a tool for both preventing fluconazole resistance from developing and for increasing susceptibility, even in highly resistant strains.

In this study, we have shown that synergistic azole-bisphosphonate pairs could have potential in antifungal therapy. The fluconazole-bisphosphonate combination is fungicidal for planktonic Cryptococcus cells and inhibits fluconazole-resistant Cryptococcus biofilms. Demonstration of efficacy in a nematode infection model provides preliminary in vivo data that the combination can work in a rudimentary animal system. The bisphosphonates currently on the market have shortcomings as systemic antifungals, however, as they have poor neural pharmacokinetics and their propensity for binding to bone mineral restricts bioavailability in tissues where the fungal burden is highest (42, 43). The results of this study should therefore serve as proof-of-concept that FPPS is a valid target for antifungal and combination therapy, with zoledronate a promising lead compound for further development.

MATERIALS AND METHODS

Cryptococcus strains.

Seven isolates each of C. neoformans (formerly C. neoformans var. grubii, VNI and VNII), C. gattii (formerly C. gattii VGI), C. deuterogattii (formerly C. gattii VGII), C. tetragattii (formerly C. gattii VGIV), and C. bacillisporus (formerly C. gattii VGIII), and four isolates of C. deneoformans (formerly C. neoformans var. neoformans, VNIV) were used in this study (Table 1). Isolates were primarily sourced from Duke University (NC, USA) and Westmead Hospital (NSW, Australia), and had been identified to the species level using multilocus sequence typing (3, 44).

Antifungal agents.

Fluconazole (Sapphire Bioscience) was dissolved in dimethyl sulfoxide (DMSO) to produce a stock solution of 16 g/liter. Nonnitrogenous bisphophonate drugs etidronate (ETI) and clodronate (CLO), and nitrogenous biosphophonate drugs RIS, ALN, and ZOL (Sigma-Aldrich) were dissolved in 0.1 N NaOH to 51.2 g/liter. As DMSO can be antimicrobial, controls containing DMSO were used in the inhibition assays and both solvents were kept consistent across dilutions.

Susceptibility and synergy.

MICs of FLC, ETI, CLO, RIS, ALN, and ZOL were determined according to CLSI standards for broth microdilution, standards M27-A3 and M27-S3 (45), with the exception that bisphosphonate stocks were dissolved in 0.1 N NaOH due to limited solubility in pure DMSO. Test concentrations were 0.125 to 64 mg/liter for FLC and 1 to 512 mg/liter for bisphosphonates, prepared in yeast nitrogen base (YNB) (Sigma-Aldrich). The final inoculum was ∼2 × 10³ cells/ml, verified by back-plating. MICs were interpreted after 72 h of incubation at 35°C, reported at MIC₈₀ for FLC and MIC₁₀₀ for bisphosphonates.

MIC_Cs of combined FLC and bisphosphonates were assessed using a checkerboard dilution assay (46), with drug pairs diluted in perpendicular directions. MIC_Cs were recorded as the most synergistic dosage of both drugs in a checkerboard. Synergy between FLC and each bisphosphonate was evaluated using fractional inhibitory concentration indices (FICI), calculated as the sum of ratios between the combined MIC_C and the individual MIC of each drug in a given pair; i.e., FICI = MIC_A ÷ MIC_CA + MIC_B ÷ MIC_CB. MICs and MIC_Cs are reported as the mode and FICIs are the means of three replicates, with comparisons made using a nonparametric Steel’s test.

Time-kill assays.

Time-kill studies were performed as described previously (47). Cryptococcus neoformans H99 and C. deuterogattii V5 strains were taken from an overnight culture in YNB and diluted to 5 × 10⁵ cells/ml in 10 ml YNB containing FLC, ALN, or ZOL at the MIC, or combinations of FLC:ALN and FLC:ZOL at 1, 2, or 4× the MIC_C. Treated cells were incubated at 37°C with shaking at 210 rpm. Aliquots (30 μl) were taken at 0, 3, 6, 12, 24, 48, and 72 h postinoculation and back-plated on yeast peptone dextrose (YPD) agar. Colonies were enumerated after 48 h at 30°C to determine viable cell counts. Experiments were conducted in duplicate, alongside a no-drug control.

Squalene rescue.

Squalene supplementation was performed based on previously described methods (48). Actively growing C. neoformans H99, C. deuterogattii V5, and C. deuterogattii 97/170 were prepared to ∼2 × 10³ cells/ml in YNB with ZOL at MIC or FLC:ZOL combined at MIC_C in a 96-well plate (Corning). Treated cell suspensions were supplemented with exogenous squalene (Sigma-Aldrich, dissolved in acetone) at doubling dilutions from 2 to 1,024 mg/liter. The effect of squalene on cryptococcal growth was assessed spectrophotometrically at 550 nm in a Bio-tek 800TS microplate reader. Experiments were carried out alongside controls containing DMSO or FLC at the MIC_C. Data were summarized as mean optical densities ± standard error of the mean (SEM) across three replicates. Nonlinear regression analysis was performed in GraphPad Prism 7.

Biofilm inhibition.

Inhibition of established biofilms was assessed as detailed previously (22). Approximately 10⁶ planktonic cells/ml of C. neoformans H99 or C. deuterogattii V5, or ∼5 × 10⁵ cells/ml of each of the two strains combined, were suspended in Dulbecco’s modified Eagle medium (DMEM) (Sigma-Aldrich) in 96-well polystyrene plates (Corning) and incubated at 37°C for 72 h to form biofilms. Following incubation, the medium was removed and adherent biofilms were rinsed 3× with phosphate-buffered saline (PBS) to remove any nonadherent cells, and the biofilms remaining were examined microscopically. Single- and mixed-species biofilms were treated with FLC, ZOL, and FLC:ZOL from 1 to 64 × MIC or MIC_C in doubling increments and were incubated for 48 h at 37°C (49). Treated biofilms were then rinsed as above and their metabolic activity was quantified by XTT (2,3-bis-(2-methoxy-4-nitro-5-sulfophenyl)-2H-tetrazolium-5-carboxanilide salt) reduction assay. The activity of each biofilm was calculated relative to its untreated counterpart, and was summarized as the mean of four technical replicates obtained from each of three independent experiments.

Propagation of antifungal resistance.

To assess whether combined FLC:ZOL treatment limits the development of resistance to either agent, approximately 10³ actively growing C. neoformans H99 or C. deuterogattii V5 cells were first grown in YNB with subinhibitory concentrations of FLC, ZOL, and FLC:ZOL at 0.25× MIC or MIC_C. Treated suspensions were incubated at 37°C with shaking for 48 h and viability was determined by staining with 0.4% trypan blue (Thermo Fisher). An aliquot containing approximately 10³ cells was taken from each treatment and passaged into sequentially increasing drug concentrations, starting from 0.5 and finishing at 8× the MIC or MIC_C, with all cultures incubated with shaking at 37°C for 48 h between each subculture. Viability staining and enumeration were repeated for each treatment. Cultures were propagated alongside a no-drug control. Experiments were performed in triplicate and data were summarized as the mean ± SEM.

C. elegans infection model.

Nematode infection and treatment were performed with modifications to previous methods (24, 50). Fresh eggs of Caenorhabditis elegans AU37 (Δglp-4; Δsek-1) were rinsed in M9 buffer (3 g/liter KH₂PO₄, 6 g/liter Na₂PO₄, 0.5 g/liter NaCl, 1 g/liter NH₄Cl) and hatched on nematode growth medium (NGM) (3 g/liter NaCl, 2.5 g/liter peptone, 5 mg/liter cholesterol, 1 mM CaCl₂, 1 mM MgSO₄, 40 mM KPO₄ [pH 6.0], 20 g/liter agar) for 24 to 36 h. AU37 nematodes were used due to their ability to form stable populations at 25°C and their broader pathogen sensitivity (51). Synchronized nematodes were counted, and 200 were dispensed onto NGM agar plates inoculated with C. neoformans H99, C. deuterogattii V5, or C. deuterogattii 97/170, or else E. coli OP50 as a nonpathogenic control. Plates were incubated for 18 h at 25°C, after which 50 nematodes were isolated, rinsed, and transferred to each of four wells in a 6-well polystyrene microplate (Corning Life Sciences, Corning, NY, USA) containing 40% brain heart infusion (BHI) (Sigma-Aldrich, St. Louis, MO, USA) broth in M9 with 90 mg/liter kanamycin, 200 mg/liter streptomycin, and 200 mg/liter ampicillin. These were then treated with FLC (16 mg/liter), ZOL (128 mg/liter), or FLC:ZOL (8 mg/liter and 64 mg/liter, respectively), or drug-free medium. Nematodes were monitored by phase-contrast microscopy with an IS10000 inverted microscope (Luminoptic). Dead nematodes were removed and enumerated every 12 h for 10 days. Mantel-Cox log-rank comparison tests were performed in GraphPad Prism 7. Two independent replicate experiments were performed.