Disarming Fungal Pathogens: Bacillus safensis Inhibits Virulence Factor Production and Biofilm Formation by Cryptococcus neoformans and Candida albicans

ABSTRACT Bacteria interact with each other in nature and often compete for limited nutrient and space resources. However, it is largely unknown whether and how bacteria also interact with human fungal pathogens naturally found in the environment. Here, we identified a soil bacterium, Bacillus safensis, which potently blocked several key Cryptococcus neoformans virulence factors, including formation of the antioxidant pigment melanin and production of the antiphagocytic polysaccharide capsule. The bacterium also inhibited de novo cryptococcal biofilm formation but had only modest inhibitory effects on already formed biofilms or planktonic cell growth. The inhibition of fungal melanization was dependent on direct cell contact and live bacteria. B. safensis also had anti-virulence factor activity against another major human-associated fungal pathogen, Candida albicans. Specifically, dual-species interaction studies revealed that the bacterium strongly inhibited C. albicans filamentation and biofilm formation. In particular, B. safensis physically attached to and degraded candidal filaments. Through genetic and phenotypic analyses, we demonstrated that bacterial chitinase activity against fungal cell wall chitin is a factor contributing to the antipathogen effect of B. safensis.

. Due to their eukaryotic nature and relatively close physiological similarity to human cells, pathogenic fungi are notoriously difficult to target for clinical therapy. In recent years, a novel concept in antimicrobial development has emerged to develop therapies that exclusively target microbial virulence factor elaboration instead of eradication of the pathogen itself (6). This approach can be seen as disarming pathogenic microorganisms and hence rendering them harmless and/or more susceptible to containment by antibiotics or the human immune system. Some pathogenic fungi inhabit the environment as their natural niche, and they may infect humans (for example, by inhalation) to cause disease. Here, we hypothesized that other microorganisms sharing environmental reservoirs, especially bacteria, may have evolved antagonistic mechanisms directed against disease-causing fungi. Specifically, we were interested in identifying microbial antipathogen activities targeting fungal virulence factor production.
Two closely related fungal pathogens of humans, Cryptococcus neoformans and C. gattii, naturally occur in environmental niches, such as soil, trees, plants, and bird excreta (7). Infections occurring through inhalation can result in lung colonization and meningoencephalitis in immunocompromised patients and are often life-threatening. It was recently estimated that 223,100 individuals suffer from cryptococcal meningitis annually, with a mortality rate of over 80% (8). Regions in sub-Saharan Africa where the number of persons suffering from AIDS/HIV is extremely high have especially frequent incidences of cryptococcosis. Indeed, cryptococcal infections are responsible for 15% of all AIDS-related deaths worldwide (8).
The pathogenic Cryptococcus species form a dark-brown/black, cell wall-associated pigment termed melanin upon host infection (9). Melanin is a major virulence factor in these pathogens and has antioxidant properties (10). Mutants with defects in melanin production usually display reduced virulence in vivo (11)(12)(13). In this study, we isolated 40 microorganisms from Cryptococcus-inhabited niches and investigated their potential to block or dampen fungal melanin production. Dual-species interaction studies revealed that several bacterial species of the genus Bacillus had antimelanization activity. One bacterium in particular, Bacillus safensis, strongly inhibited cryptococcal melanization and melanin shedding in a contact-dependent manner. Importantly, B. safensis did not significantly impact overall fungal growth but did block expression of other key cryptococcal virulence factors, including capsule production and biofilm formation. Antivirulence activity was not limited to basidiomycetes, as B. safensis also potently inhibited virulence factor elaboration by the ascomycete Candida albicans, another major human fungal pathogen that is obligately associated with mammals (14). Through genetic and phenotypic analyses, we demonstrate that chitinase activity is a factor contributing to the destablization of the fungal cell wall by B. safensis.

RESULTS
A screen of environmental microbes identified bacteria that inhibit C. neoformans melanin production. Our first objective was to investigate whether some of the microbes found in environmental niches that are coinhabited by Cryptococcus species may have antifungal activities. C. neoformans and C. gattii are found in the environment in soil and in bird droppings and are associated with trees and plants (7). We therefore assembled soil samples from Vancouver Island (a kind gift of Karen Bartlett, University of British Columbia) and plant samples from Vancouver (see Fig. S1A and S1B and Table S1 in the supplemental material). The soil samples were previously shown to be positive for the presence of C. gattii, and we hypothesized that this would increase our chances of identifying microbes with anticryptococcal activity (15). We isolated microbes from 24 different soil samples, covering nine different locations on Vancouver Island, and from two different plant samples from Vancouver. Between 1 and~100 colonies grew per plate after 48 h at 30°C, and we chose 1 to 4 colonies per plate for further investigation. Selection criteria were based on an effort to include a variety of sampling locations and differences in colony appearance and color. In total, 40 isolates of diverse sizes and with predominantly rod-shaped morphologies were assembled (Fig. S1C).
It has recently been proposed that targeting virulence factor production, rather than pathogen growth itself, may be a promising approach to prevent and treat disease while simultaneously reducing the likelihood of eliciting pathogen resistance (6). We therefore set out to determine whether any of the isolates had activity against cryptococcal production of the antioxidant pigment melanin. Melanin formation is a key virulence factor in cryptococci, and melanin-deficient mutants are usually strongly attenuated in virulence (11)(12)(13). We incubated each of the 40 isolates either individually or mixed with C. neoformans wild-type (wt) strain H99 on L-3,4-dihydroxyphenylalanine (L-DOPA) agar, a medium that induces strong melanin production by cryptococcal cells. We then manually selected a rectangular image of a region of the colony center (according to the scheme shown in Fig. S2A) and assembled each image fragment into a "melano-map" (Fig. 1A). None of the isolated microbes appeared to produce melanin under these conditions. Incubation with C. neoformans alone, and most of the fungusmicrobe coincubations, resulted in black colonies, indicative of melanin production. Strikingly, some of the isolates appeared to inhibit fungal melanization, leading to production of colonies with a light brown to beige color. Isolate M2 in particular had a strong antagonistic effect on melanization, and the color of dual-species colonies was light beige (Fig. 1A and Fig. S2B). We next determined the identity of those isolates that demonstrated antimelanization activity by 16S rRNA sequencing. All of the isolates were identified as bacteria, and most of them belonged to the genus Bacillus (Fig. 1B). Indeed, all bacteria found to possess antimelanization activity were Bacillus spp., while five randomly selected isolates without activity included both Bacillus spp. and species from other genera (Fig. 1B). Isolate M2 was identified as Bacillus safensis, a Grampositive, rod-shaped, motile bacterium found ubiquitously in soil (Fig. 1C).
B. safensis blocks cryptococcal melanization and melanin shedding in a contact-dependent manner. Due to the strong effect of B. safensis on fungal melanization, we decided to focus our further investigations on this bacterium. We first confirmed that B. safensis not only affected melanin formation by C. neoformans but also strongly inhibited melanization of C. gattii (Fig. 1D). We next verified that the observed antimelanization effect was not due to bacterial consumption of L-DOPA (which would potentially make less L-DOPA available for C. neoformans to synthesize melanin). To do so, we incubated C. neoformans in B. safensis-spent, filter-sterilized, L-DOPA medium. Fungal cells were still able to melanize, as indicated by a transition of the appearance of the medium from clear to black (Fig. S2C). These results suggested that the bacterial antimelanization activity was likely not simply due to competition for L-DOPA as a substrate but rather pointed to a specific antifungal mechanism.
To begin to elucidate how B. safensis impacted fungal melanin formation, we next investigated whether direct contact and bacterial viability were required for the observed melanin-inhibiting effect. We found that the inhibitory effect was observed only when the fungal and bacterial cells were spotted onto L-DOPA agar such that the cells were initially touching each other (Fig. 1E). Furthermore, heat-killed B. safensis did not exert antimelanization activity (Fig. 1E). We also noted that when the two microorganisms were spotted in contact with each other, bacteria appeared to swarm around C. neoformans colonies (Fig. 1E). These results indicated that both direct contact and live bacterial cells are required for the observed impact on melanin formation. Cryptococcal cells not only incorporate melanin pigment in their cell wall but also shed melanin into the cell periphery. This process contributes immune-modulatory activity during human infection. We noticed that fungal colonies coincubated with B. safensis on L-DOPA medium not only did not turn black but also appeared to shed much less melanin into the surrounding agar (Fig. S2D). We therefore next quantified a potential impact of the bacteria on the shedding of fungal melanin in liquid L-DOPA medium. Cultures incubated with B. safensis and C. neoformans did not turn black compared to the fungus-only control (Fig. 1F), and indeed, melanin shedding was strongly reduced upon coculture (Fig. 1G). Interestingly, we also detected shedding of melanin (or a melanin-like molecule) by B. safensis monocultures (Fig. 1G). B. safensis is known to possess laccases (the enzymes required for biochemical conversion of L-DOPA into pigment) (16); however, melanin shedding was relatively modest (Fig. 1G), did not translate into a change of color of the medium (Fig. 1F), and did not interfere with C. neoformans melanization (Fig. S2C).
Bacillus spp. are known to secrete over 160 small molecules (17). We therefore evaluated the possibility that B. safensis may secrete a small molecule (or a mixture of molecules) with antimelanization activity. To this end, we grew B. safensis in yeast extract-peptone-dextrose (YPD), potato dextrose broth (PDB; Difco), or yeast nitrogen base-low-iron medium (YNB-LIM) and used the concentrated supernatants to test for Colony color map ("melano-map") of 40 environmental microbes, isolated from either soil or plants, following incubation on L-DOPA agar with or without C. neoformans wild-type strain H99. Fungal cells alone melanize and produce a black colony color. Cocultivation with some microbes results in impaired fungal melanization. Note that microbe M2 has strong antimelanization activity. Cn, C. neoformans; Ctrl, control; M1-M40, environmental microbe M1-M40. Stars indicate microbes with antimelanization activity. See Fig. S2A for additional information on assembly of the "melano-map." The experiment was performed twice with similar results, and results from one experiment are shown. (B) Proportion of environmental microbes with activity against cryptococcal melanization. 16S rRNA sequencing revealed that all 10 microbes showing antimelanization activity belong to the Bacillus genus. For evaluation of microorganisms without antimelanization activity, five microbes were randomly selected and analyzed by 16S rRNA sequencing. B. aryabhattai, Bacillus aryabhattai; B. megaterium, Bacillus megaterium; B. safensis, Bacillus safensis; B. subtilis, Bacillus subtilis; B. mycoides, Bacillus mycoides; B. thuringiensis, Bacillus thuringiensis; P. theicola, Pantoea theicola; P. xylanexedens, Paenibacillus xylanexedens. (C) DIC microscopy image of the Grampositive, rod-shaped bacterium B. safensis. Scale bar, 5 m. (D) B. safensis (Bs) inhibits melanization of the C. gattii wild-type strain R265 on L-DOPA agar at 30°C. Scale bar, 5 mm. The experiment was performed twice, and representative images are shown. (E) Direct cell contact and live bacterial cells are required for the antimelanization activity. Strains were incubated on L-DOPA agar at 30°C for 48 h. HK, heat killed. Scale bar, 5 mm. The experiment was performed three times, and representative images are shown. (F) The antimelanization activity of B. safensis is recapitulated in liquid L-DOPA medium. Strains were incubated at 30°C and 180 rpm for 46 h. Ctrl, medium-only control. (G) Quantification of melanin shedding for cryptococcal cells grown in liquid L-DOPA medium alone or in the presence of B. safensis (see panel F). Quantification is based on OD 405 measurements, and results are presented relative to the wild-type strain, H99, whose results have been set to 100%. nd, not detected. Results are the means ϩ standard deviations (SD) of three independent experiments, each performed in triplicate. ***, P Ͻ 0.0001. antimelanization activity. None of the three culture supernatants repressed melanin production (Fig. S2E). These results are consistent with a requirement for direct bacterial-fungal contact for suppression of C. neoformans melanization.
Bacterial-fungal interactions impacted C. neoformans growth dynamics. Having established that B. safensis has strong potency in inhibiting the production of a key cryptococcal virulence factor, we next determined whether the bacterium also affected fungal growth. Single-species incubations indicated that, as expected, B. safensis grew significantly faster in YPD than C. neoformans (Fig. S3). However, the two organisms had reached comparable cell densities after 72 h. Addition of the antibiotic gentamicin to YPD completely blocked bacterial growth without affecting fungal growth (Fig. S3). B. safensis grew even better in LB medium than in YPD; however, the LB medium did not sustain fungal growth (Fig. S3). Therefore, most of the fungus-bacterium coincubation studies described below were performed using YPD alone or YPD supplemented with gentamicin to inhibit B. safensis growth when necessary. Coincubation of C. neoformans with B. safensis had no effect on fungal growth dynamics in liquid L-DOPA medium compared to fungal monocultures ( Fig. 2A). L-DOPA medium is nutrient limited, and we next assessed a possible impact of bacteria on fungal growth in complex YPD medium. Notably, the presence of B. safensis delayed fungal growth at early time points. At 24 h, fungal cell numbers in cocultures were~180-fold reduced compared to fungal cultures incubated without bacteria. However, after 48 h, C. neoformans cell numbers in cocultures were only~5-fold reduced compared to fungus-only control cultures ( Fig. 2A). Coincubation with Escherichia coli had no effect on cryptococcal growth, pointing to differences in the interactions of both bacteria with this pathogen ( Fig. 2A). To investigate the growth dynamics of C. neoformans upon interaction with B. safensis in YPD in more detail, we performed a similar bimicrobial incubation experiment and evaluated fungal growth using the enumeration of CFU. The results confirmed that fungal growth was significantly (~340-fold) reduced after 24 h of coincubation compared to fungus-only incubations. At 48 h, however, viable cell numbers had reached comparable values with a relatively modest,~3-fold difference (Fig. 2B). In summary, B. safensis delays fungal proliferation at early time points but does not have a major impact on overall fungal growth after longer incubation times.
B. safensis targets the C. neoformans cell surface to suppress melanin production. Having demonstrated the capacity of B. safensis to inhibit C. neoformans melanization, we next sought to identify a potential mechanism. We reasoned that screening the recently established library of C. neoformans transcription factor (TF) mutants during coculture with B. safensis would provide clues to general fungal pathways or proteins potentially targeted by the bacterium (18). We performed the screen in liquid L-DOPA medium by culturing each of the 308 TF mutants (2 individual mutants per TF) alone Bacterial Inhibition of Fungal Virulence Traits ® or together with B. safensis for 5 days. Our original aim was to identify potential target candidates on the basis of evaluation of melanin production. However, the experiments were performed in 96-well plates and we obtained inconsistent results for melanin formation under those conditions. As an alternative and more general strategy, we examined the interactions using optical density (OD) measurements of fungal growth, and this approach provided a more robust evaluation of the response of C. neoformans to B. safensis. Specifically, when we compared final OD values at 405 nm (OD 405 ) of dual-species cultures for each mutant with the OD 405 values for fungal strains alone, we detected five mutants, namely, strains nrg1Δ, cep3Δ, sre1Δ, ert1Δ, and ada2Δ, with increased OD values during coculture compared to fungal monoculture in at least one of the two experiments performed (Fig. 3A). Subsequent testing of the five mutants revealed that none showed melanin defects on their own but that, similarly to the wt strain, melanization of each was inhibited by B. safensis. For each TF, two independent mutants were incubated in liquid L-DOPA medium with or without bacteria (308 TF strains in total) at 30°C for 5 days. Growth was determined via OD 405 measurements. Results are plotted relative to the fungus-only control. Each circle represents the mean of the results for two independent mutants of the same TF. The black dotted lines and the gray dotted lines indicate the overall means of all values Ϯ SD, respectively. The experiment (exp.) was performed twice, and TF mutant outliers (in at least one experiment) are indicated and color-coded according to previously published roles of these TFs in cell membrane and cell wall integrity (see scheme on the right). Rel. growth, relative growth of fungal cells in coculture versus monoculture, expressed in percentages. (B) Sorbitol bypassed B. safensis-mediated suppression of fungal melanization, and this effect was not observed in a C. neoformans nrg1Δ mutant. C. neoformans wild-type strain H99 or the nrg1Δ mutant was incubated alone or mixed with bacteria on L-DOPA agar with or without sorbitol for 48 h. Ctrl, control; Sorb, sorbitol. Scale bar, 5 mm. The experiment was performed twice, and representative images are shown. (C) Coincubation of C. neoformans nrg1Δ with B. safensis in YPD for 24 h resulted in fungal cell aggregation and abnormal cell wall chitin staining. White squares in the panels of the middle row indicate regions that are shown in a magnified view in the bottom row. Arrows point to abnormal, Љscratch-likeЉ CFW staining. DIC, differential interference contrast; CFW, calcofluor white. Scale bars indicate 5 m (middle row) and 2 m (bottom row). The experiment was performed twice, and representative images are shown. (D) B. safensis and CFW synergistically inhibit cryptococcal growth. CFU-based analysis of C. neoformans growth in YPD in presence of CFW or bacteria or a combination of the two was performed. Results are the means Ϯ SD of two independent experiments, each performed in duplicate. (E) A chitinase inhibitor partially rescues B. safensis-mediated inhibition of fungal melanization on L-DOPA. Strains were incubated at 30°C for 48 h. Ctrl, control; BisC, bisdionine C. Scale bar, 5 mm. The experiment was performed three times, and representative images are shown.
We noticed that three of the five mutants (strains nrg1Δ, sre1Δ, and ert1Δ) had deletions in genes that have previously been shown to contribute to cell membrane and/or cell wall integrity (CWI) in C. neoformans (18,19). We therefore hypothesized that B. safensis may target the fungal cell surface. To test this idea, we used L-DOPA agar supplemented with the cell membrane-and cell wall-stabilizing agent sorbitol and performed fungus-bacterium coincubation experiments. Strikingly, sorbitol rescued B. safensis-mediated inhibition of C. neoformans melanization, and this effect appeared to be temperature dependent, with more-pronounced melanization at 37°C (Fig. 3B). We verified that the sorbitol concentration used in these experiments did not negatively impact bacterial growth (Fig. S4A). We then investigated whether sorbitol may also bypass inhibition of fungal melanization mediated by the other bacterial species identified in our initial screen (Fig. 1A). With the exception of isolates M34 and M38 (both identified as B. subtilis; see Table S1), sorbitol bypassed inhibition of fungal melanization of all remaining bacteria (Fig. S4B). Next, we hypothesized that if B. safensis targets the fungal cell surface, melanization of the identified TF mutants (with defects in cell membrane/cell wall regulation) might not be rescued by sorbitol supplementation upon interaction with bacteria due to an additive negative effect on the cell surface. Indeed, sorbitol did not rescue melanin formation of the nrg1Δ and sre1Δ mutants under these conditions ( Fig. 3B and Fig. S4C). The ert1Δ mutant, however, behaved like the wild type in this assay, and sorbitol did rescue melanin formation in the presence of B. safensis. These results indicated that the cell wall integrity (CWI) pathway may be involved in resistance to B. safensis-mediated fungal targeting. Consistently, sorbitol did not rescue melanization of specific CWI pathway mutants of C. neoformans, i.e., strains hog1Δ (20) and mpk1Δ (21), upon exposure to bacteria (Fig. S4C).
Due to the strong defect in melanization of the nrg1Δ mutant during interaction with B. safensis, we next investigated single-species and dual-species fungus-bacterium cultures of this particular mutant in more detail. Surprisingly, nrg1Δ cells exposed to bacteria for 24 h in YPD displayed aberrant cell morphologies, with cell surfaces that appeared to have been "scratched" (Fig. 3C). Staining with the chitin-binding dye calcofluor white (CFW) confirmed that nrg1Δ cells exposed to B. safensis had altered cell surface morphologies (Fig. 3C). The "scratch"-like phenotype is highly reminiscent of a phenotype previously observed for the C. neoformans chitin synthase mutants chs3Δ and csr2Δ (22,23). These results therefore suggested that B. safensis targets the fungal cell wall at the level of chitin. To test this possibility, we next coexposed C. neoformans to bacteria and CFW. We reasoned that fungal growth would be synergistically inhibited if B. safensis indeed targets fungal chitin. Consistent with that assumption, a time course analysis of growth revealed that coincubation of fungal cells with either CFW or B. safensis modestly inhibited C. neoformans growth compared to the fungus-only control. Coapplication of bacteria and CFW, however, led to a synergistic inhibitory effect on fungal proliferation (Fig. 3D). These findings were specific for chitin because an analogous experiment using Congo red (a dye that binds cell wall ␤-1,3-glucan) did not demonstrate an impact on fungal growth (Fig. S4D). We also noticed that some wild-type C. neoformans cells in dual-species cultures with B. safensis appeared to stain slightly more strongly for cell wall chitin using CWF, and we confirmed this observation quantitatively ( Fig. S4E and S4F). The strong phenotype of the nrg1Δ mutant upon coincubation with B. safensis prompted us to examine its growth in the presence of bacteria. Interestingly, an assay based on enumeration of CFU revealed that cells of the nrg1Δ mutant were significantly more susceptible to B. safensis after 48 h of coincubation than those of the wild-type strain (Fig. S4G). To test whether B. safensis may also impact the fungal cell membrane, as suggested by the identification of some of the TF mutants with defects in membrane function, we next used the membrane-staining dye FM4-64 to stain C. neoformans cells from single-species and dual-species cultures. While FM4-64 stained the vacuolar membrane of fungus-only cultures in a distinctive ringlike fashion, fungal cells from cocultures displayed a diffuse staining pattern across the entire cell (Fig. S4H). These results indicate that B. safensis may interfere with the architecture of both the cell wall and the cell membrane.
On the basis of the similar phenotypes of B. safensis-exposed nrg1Δ cells and chitin synthase mutants and of the synergistic effect of the chitin-binding compound CFW on the growth of C. neoformans in the presence of bacteria, we hypothesized that bacteria may target fungal cell wall chitin by chitinase activity. We therefore determined the effect of the chitinase inhibitor bisdionine C (BisC) on fungal melanization in dualspecies cultures. Consistent with our hypothesis, BisC rescued C. neoformans melanization at least partially (Fig. 3E). Interestingly, commercially acquired chitinase did not prevent melanization of C. neoformans in these experiments (Fig. S4I). This result may have been due to the need for B. safensis to contact the fungal surface to impact chitin and also to the fact that the bacterium may produce more than one factor to exert an influence on melanization. Nonetheless, our results obtained with CFW and BisC suggest that bacterial chitinase activity is a factor contributing to the observed inhibition of cryptococcal melanin production.
We next reasoned that chitinase activity of B. safensis may release increased levels of chito-oligomers from fungal cells. Wheat germ agglutinin (WGA) specifically binds chito-oligomers (24), and we used it to stain C. neoformans cells following incubation with or without B. safensis. As predicted, we detected a modest but significant increase in WGA-staining intensity for fungal cells exposed to bacteria compared to controls ( Fig. S5A and S5B). We also noted that the cells of C. neoformans showed a significant increase in size following dual-species interaction (Fig. S5C). Overall, these results contribute to the conclusion that B. safensis targets the fungal cell wall at least in part through chitinase activity, resulting in destabilization of cell wall architecture and changes in morphology and ultimately leading to improper virulence factor expression.
B. safensis blocks C. neoformans virulence factor production. Besides melanin synthesis, cryptococcal cells also produce other disease-relevant virulence factors. These factors include production of a protective and immune-modulatory polysaccharide capsule and biofilm formation (5). Given the strong bacterial impact on fungal melanization, we next investigated whether B. safensis may also affect expression of these other fungal virulence factors. Strikingly, coincubation of C. neoformans with bacteria in 10% fetal calf serum (FCS; a capsule-inducing medium) resulted in a complete block in fungal capsule formation (Fig. 4A). Though the fungal cells were modestly reduced in overall cell size, no capsule could be detected in coincubations compared to fungal monocultures (Fig. 4B).
Because some of the polysaccharides present in the capsule are also part of the cryptococcal biofilm matrix (25), we hypothesized that B. safensis may also negatively impact fungal biofilm formation. Indeed, when we exposed C. neoformans to B. safensis under biofilm-inducing conditions at a ratio of 1:1, we detected a significant reduction in the overall level of biofilm formation of coincubations compared to incubations performed with a single fungal species (Fig. 4C and D). This effect was temperature dependent, as overall biofilm formation was more strongly reduced at 37°C than at 30°C. It should be noted that in this experiment the actual level of fungal biofilm formation was probably even lower than indicated, as biofilm formation was measured for the mixture of fungal and bacterial cells. On the basis of our previous findings of a possible role of B. safensis chitinase activity during interaction with C. neoformans, we next investigated whether addition of exogenous chitinase alone might impact capsule formation by C. neoformans. Consistent with previously published findings, chitinase significantly inhibited capsule formation ( Fig. 4E and F) (24). Overall, these findings indicate that B. safensis has strong potency in preventing C. neoformans capsule formation and significantly reduces fungal biofilm production.
Hypha formation of the human-associated pathogen C. albicans is inhibited during interaction with B. safensis. Having demonstrated that B. safensis potently inhibits the elaboration of cryptococcal virulence factors without having a substantial antagonistic impact on growth, we next determined whether the bacterium exhibited anti-virulence factor activity during interactions with other fungal pathogens. Specifically, we performed bimicrobial interaction studies with another major human fungal pathogen, C. albicans. In contrast to C. neoformans and C. gattii, C. albicans is not found in the environment but is obligately associated with warm-blooded animals such as humans (14). One of the main C. albicans virulence factors is the capacity to transition between a round yeast form and an elongated hyphal form (the yeast-to-hypha transition) (26). We therefore focused on a potential effect of B. safensis on C. albicans hypha formation. Coincubation of bacterial and fungal cells under conditions that induced yeast growth (YPD, 30°C) did not affect C. albicans cell morphology or proliferation compared to fungus-only cultures (Fig. S6A). However, under hypha- Bacterial Inhibition of Fungal Virulence Traits ® inducing conditions (10% FCS, 37°C), B. safensis strongly inhibited C. albicans hypha formation compared to control cultures with the fungus alone (Fig. S6A). We estimated the levels of yeast and hypha production under these conditions by determining the wet weights of cell pellets following overnight sedimentation at 4°C and centrifugation. The wet weights of pellets of yeast-phase cells from monospecies and dual-species cultures were not significantly different. However, significantly more hyphal mass was formed by C. albicans-only cultures than by dual-species cultures, thus confirming our microscopic examination (Fig. S6B). It should be noted that the evaluation of C. albicans morphologies was semiquantitative and may have been affected by carryover effects of the presence of bacterial cells following sedimentation and centrifugation of the dual-species culture. Due to the relatively small size of B. safensis cells compared to C. albicans yeast and hyphal cells, however, these effects should have been negligible. To investigate whether B. safensis also affected C. albicans hypha formation on solid media, we incubated fungal cells alone or mixed with B. safensis or E. coli on water agar supplemented with 10% FCS. The fungal cells incubated alone showed strong filamentation following a 9-day incubation time, and the presence of E. coli did not affect this process (Fig. S6C). However, consistent with our previous observations, B. safensis potently inhibited hypha formation under these conditions (Fig. S6C).
B. safensis attaches to C. albicans hyphae and suppresses fungal adhesion capacity and de novo biofilm formation. To gain further insight into the influence of B. safensis on C. albicans morphogenesis, we next analyzed fungus-bacterium interactions on the single-cell level. We discovered that bacterial cells appeared to attach to fungal hyphae as soon as 4 h following coinoculation into YPD medium (Fig. 5A). This effect was dependent on bacterial viability because heat-killed cells did not attach to fungal filaments (Fig. S7). After 24 h of coincubation, hyphal filaments appeared to be much thinner and partially disintegrated compared to fungus-only controls (Fig. 5A). We also performed analogous experiments using different ratios of fungal to bacterial cells and found that C. albicans hyphal length inversely correlated with the number of bacteria in the inoculum (Fig. S8). Furthermore, the higher the initial bacterial inoculum, the more bacteria were found to attach to C. albicans filaments (Fig. S8). Overall, these results confirmed that direct cell-cell contact appears to be a crucial aspect of the interaction between B. safensis and fungal pathogens and that the antifungus mechanism may be directed against the fungal cell surface.
The yeast-to-hypha transition is a key process for the formation of C. albicans biofilms. Prior to biofilm formation, however, fungal cells need to adhere to a surface. The processes of adhesion and biofilm formation both have important clinical implications, as C. albicans biofilms often form on medical devices such as catheters and are extremely recalcitrant to antifungal therapy. In many cases, the only option is to surgically remove the catheters (27). We therefore next investigated whether B. safensis interfered with the adherence capacity and biofilm formation of C. albicans. Strikingly, the presence of bacteria inhibited adhesion of fungal cells to polystyrene (Fig. 5B) and drastically blocked de novo biofilm formation in a dose-dependent manner (Fig. 5C). Compared to the dense appearance of fungus-only biofilms, fungal cells exposed to bacteria were not able to form robust biofilms (Fig. 5D). Again, individual hyphal length was dependent on the dose of the initial bacterial inoculum. To evaluate the effects of B. safensis on C. albicans biofilms that had already formed, fungal biofilms were allowed to develop for 24 h, bacteria were added, and the biofilms were incubated for another 24 h. In this experiment, B. safensis did not have a major impact on C. albicans biofilm formation. A statistically significant but modest reduction in biofilm formation was detected only at the highest bacterial inoculum (Fig. 5E).
Bacterial chitinase activity contributes to the interaction of B. safensis with C. albicans. On the basis of our finding that B. safensis chitinase activity is likely to contribute to inhibition of virulence factor production by C. neoformans, we next hypothesized that similar chitin-targeted enzymatic activity may be responsible for the suppression of C. albicans hypha formation by B. safensis. We therefore analyzed the effects of exogenous chitinase on C. albicans filamentation. Consistent with our hy-pothesis, chitinase significantly reduced hypha formation by C. albicans, though not as potently as B. safensis (Fig. 6). These results indicate that chitinase appears to contribute to the observed effects but that additional B. safensis factors are likely to be involved in this interspecies interaction.
To determine whether B. safensis affected general C. albicans growth, we performed additional bimicrobial interaction studies. C. albicans growth was reduced after 24 h of coincubation with the bacterium compared to the fungus-only control but reached similar viable cell numbers after 48 h (Fig. 7A). These results are similar to those obtained for C. neoformans (see Fig. 2B). Finally, we used the dye WGA to stain C. albicans cells following incubation with or without B. safensis. We detected significantly stronger WGA staining of fungal cells that had been exposed to bacteria than in the controls ( Fig. 7B and C). Furthermore, the C. albicans cells were slightly reduced in size following dual-species interaction (Fig. 7D). This result indicates differences between C. albicans and C. neoformans in the morphological response of yeast cells to B. safensis ( Fig. 7D and Fig. S5C). Overall, these results indicate that B. safensis likely targets the cell walls of C. neoformans and C. albicans through chitinase activity, resulting in destabi-

DISCUSSION
There are 10 7 to 10 10 prokaryotes per gram of soil in nature (28,29). These enormous numbers make it highly likely that soil microbes, including bacteria, fungi, and protozoa, interact with each other, either directly or indirectly. Such interactions can have huge ecological, economic, and medical implications. Polymicrobial interaction studies are currently gaining momentum due to the realization that interspecies interactions strongly impact human health and disease (30)(31)(32). However, few studies have investigated the interactions among fungal pathogens naturally found in the environment and other niche-specific microbes. Specifically, whether environmental microbes impact fungal virulence traits is largely unknown. Here, we hypothesized that certain microorganisms in soil and on plants antagonistically interact with C. neoformans and C. gattii, two major human fungal pathogens found in these habitats. By screening 40 environmental microbes, we identified 10 bacteria of the genus Bacillus with activity against cryptococcal melanization.
One bacterium, B. safensis, had a particularly strong capacity to block the melanization of C. neoformans and C. gattii. Anti-virulence factor activities against other fungal virulence traits, including capsule production and biofilm formation, were also identified. Furthermore, antifungal activity was not limited to basidiomycete fungi but was also detected against a major ascomycete fungal pathogen, C. albicans. B. safensis potently inhibited the morphological transition from yeast to hyphae and efficiently blocked de novo biofilm formation. Through genetic and phenotypic screens performed using a C. neoformans TF mutant library, we identified the fungal cell wall as a target of B. safensis. Furthermore, we provide evidence that B. safensis likely employs chitinases to destabilize the cell wall. These findings led us to propose a model (Fig. 8) in which B. safensis comes into close contact with fungal cells and secretes chitinases and other factors to degrade the chitin fibers that form the innermost layer of the cell wall (33). Chitin is a polymer of N-acetylglucosamine and is one of the most abundant biopolymers in nature. Because melanin is incorporated into the fungal cell wall via anchoring to the chitin layer in normal cells, chitinase-mediated disaggregation events may take away the foundation for proper melanin deposition (Fig. 8). As indicated, other factors may influence the cell wall as well as other potential targets in C. neoformans, including the plasma membrane.
Cryptococcal virulence factors such as laccases (enzymes mediating melanin bio- synthesis) and capsule polysaccharides are delivered to the outer side of the cell via crossing the cell membrane and cell wall in vesicles (so-called "virulence delivery bags") that transport virulence factor building blocks (5). Consistent with the idea of an impact of B. safensis on the cell membrane and fungal secretion dynamics, we observed aberrant uptake of the lipophilic dye FM4-64 into B. safensis-exposed C. neoformans cells compared to unexposed fungal cells. The model of cell wall destabilization via chitin degradation also provides a possible explanation for the observed reductions in C. neoformans capsule production during dual-species interaction (Fig. 8). Capsule polysaccharides are assembled onto the ␣-1,3-glucan layer of the cell wall, beneath which the ␤-1,3-glucan/␤-1,6-glucan layer is anchored to chitin (5,34). Mutants with impaired capsule formation are usually strongly reduced in virulence (12,35,36). Hence, our results reveal that B. safensis targets a fungal structure in a way that does not affect overall cell survival but is essential for proper virulence factor assembly. Few studies have analyzed C. neoformans-bacterium dual-species interactions, and, to the best of our knowledge, the specific inhibition of cryptococcal virulence factor production has not been documented before. A pioneering study by Bulmer and colleagues described anticryptococcal activity exerted by two bacterial species isolated from pigeon guano, Pseudomonas aeruginosa and Bacillus subtilis (37). Another landmark publication by Casadevall and colleagues established a potential role of soil amoeba in shaping the virulence repertoire of C. neoformans during dual-species interactions in the environment (38). Amoebae show remarkable similarities with macrophages, key cells of the human immune system, and it was found that fungal capsule and melanin production protected C. neoformans following ingestion by amoebae (38). More recently, the soil bacterium Acinetobacter baumannii was found to inhibit C. neoformans growth in a fungal serotype-dependent manner. In contrast to our findings obtained with B. safensis, however, A. baumannii induced C. neoformans capsule enlargement (39).
It is remarkable that our isolation procedure appears to have selected for bacteria of the genus Bacillus. We used YPD and LB media for isolation, and it is possible that these nutrient-rich, complex media specifically promote growth of this environmental bacterial genus. Indeed, several studies in which bacteria were isolated from environmental soil or plant samples identified Bacillus as the predominant bacterial genus (40,41). The antimelanization effect, however, was not general with respect to the Bacillus genus, because some bacteria belonging to this genus did not display antifungal activity.
B. safensis is a Gram-positive, motile, spore-forming, ubiquitous soil bacterium that was first isolated from the spacecraft assembly facility at NASA's Jet Propulsion Laboratory in Pasadena, CA, USA (42). B. safensis is a close relative of Bacillus pumilus, a soil bacterium that was found also to cause contamination of spacecraft equipment (43). Spores of B. pumilus are extremely resistant to oxidative and UV stress (44). Interestingly, several studies have demonstrated antifungal chitinase production by B. pumilus, and this bacterium is used as a commercial probiotic and as a biopesticide in agriculture (45)(46)(47)(48). Moreover, bacteria of the genus Bacillus in general appear to possess particularly potent antifungal activity. A mixture of five bacterial species, three of which were Bacillus cereus, Bacillus megaterium, and Bacillus mojavensis, for example, was found to protect tobacco (Nicotiana attenuata) from Fusarium and Alternaria infection (49).
We used C. gattii-positive soil samples from Vancouver Island to increase the likelihood of identifying microbes with specific activity against cryptococcal cells (15). Vancouver Island is a hot spot for C. gattii and has been the site of an outbreak of cryptococcosis in recent years (50,51). Importantly, B. safensis also displayed potent antimelanization activity against C. gattii. It remains to be determined whether soils from areas without C. gattii or C. neoformans harbor similar microbial repertoires. As a first analysis, we focused only on easily cultivable microbes. We are aware that one limitation of our study is the fact that we did not isolate "uncultivable" environmental microbes that may exert important antifungal activities. Future studies could employ strategies such as the use of an iChip to isolate and culture uncultivable microbes from soil (52).
We detected moderate melanin shedding activity for B. safensis monocultures in liquid L-DOPA, indicating that these bacteria may produce an enzyme capable of metabolizing L-DOPA. Indeed, B. safensis produces at least one laccase (16). However, we argue that this activity does not interfere with the effect on cryptococcal melanization because (i) B. safensis-spent medium provides enough L-DOPA for C. neoformans to melanize; (ii) the measured values of bacterial melanin shedding were comparably small; and (iii) the bacterial colonies or cell periphery did not turn black, which would have been indicative of laccase-mediated pigment production.
We determined that bacteria required close contact with fungal cells to exert their antifungal effects. This was supported by several findings. First, heat-killed B. safensis cells did not block cryptococcal melanization. Second, B. safensis attached directly to C. albicans filaments. Finally, bacterial cells were found to swarm around fungal colonies on solid agar plates. Swarming in Bacillus is mediated by flagella and secretion of surfactins, lipopeptides that reduce tension between the substrate and the bacterial cells to allow gliding over surfaces (53,54). What the exact fungal cues are that direct bacterial chemotaxis is unknown, but fungal polysaccharide shedding may contribute to this phenomenon. The requirement of direct cell-cell contact for bacterial antagonism has been described before for the interaction between C. neoformans and Staphylococcus aureus, an important bacterial pathogen of humans. S. aureus attached to and killed C. neoformans in a process that was dependent on presence of fungal capsule polysaccharide (55). Furthermore, contact-dependent growth inhibition events have been described for P. aeruginosa-C. neoformans dual-species interactions. Bacteria inhibited C. neoformans via production of the quorum-sensing molecule pyocyanin (56). These studies indicate that direct bacterium-fungus interactions may be common but that the individual mechanisms of interaction can differ significantly.
B. safensis blocked C. neoformans capsule formation, and this prompted us to investigate fungal biofilm formation since the capsule matrix surrounding fungal biofilms is composed of the same glucuronoxylomannan (GXM) molecules that are the main building blocks of the polysaccharide capsule (25). Consistent with our hypothesis, C. neoformans biofilm formation was significantly inhibited by B. safensis. The values for biofilms formed by the fungus-bacterium cocultures are likely to be overestimates since this analysis did not discriminate between the two organisms and B. safensis formed considerable biofilms on its own.
C. albicans is a major opportunistic human fungal pathogen and can cause infections in immunocompromised persons that range from superficial infections of the skin and mucosal surfaces to life-threatening systemic infections with high mortality rates (14). A key virulence factor in this fungus is the capacity to transition between the yeast and hyphal forms, both of which are required for infection (26). Yeast cells can adhere to biotic surfaces (e.g., oral epithelial cells) and abiotic surfaces (e.g., catheters and polystyrene) and are believed to represent the morphology that mediates dissemination through the bloodstream. The hyphal form has been shown to mediate tissue penetration, both through active fungus-driven mechanical forces and by induced endocytosis by host cells (57)(58)(59). Hence, both candidal cell morphologies play important roles during pathogenesis. B. safensis significantly reduced C. albicans adhesion to polystyrene and strongly inhibited hypha formation. Because adhesion and hypha formation are prerequisites for biofilm formation in this pathogen, we hypothesized that B. safensis may have antibiofilm activities. Indeed, B. safensis potently blocked de novo C. albicans biofilm formation, while preformed C. albicans biofilms were resistant to B. safensis treatment. This is likely due to the fact that fungal cell walls are dynamically modified during growth, division, and morphogenesis. Indeed, fungi encode chitinases that have important roles in cell wall plasticity (60). These fungal chitinase activities, however, must be tightly regulated to prevent autodirected cell wall damage. Preformed C. albicans biofilm cells are encased in an extracellular matrix that probably protects these cells from B. safensis chitinase activity.
Several studies have investigated the effects of bacteria on C. albicans hypha formation (61)(62)(63)(64). Perhaps the best-studied interactions are those of C. albicans and P. aeruginosa, an important bacterial pathogen of humans (62,(65)(66)(67). Similarly to our findings with B. safensis, P. aeruginosa was demonstrated to attach to fungal filaments, and bacteria specifically killed fungal hyphae via mechanisms dependent on pili, phospholipase C, and phenazine secretion (62). Furthermore, a recent study demonstrated a role for the Enterococcus faecalis toxin EntV in blocking C. albicans filamentation, biofilm production, and virulence in a murine model of oropharyngeal candidasis (68). These studies indicate that certain bacteria interact with C. albicans and appear to preferentially target the hyphal morphology.
Several lines of evidence pointed toward a role of chitinase activity in mediating anti-virulence factor activities during fungus-bacterium dual-species interaction. First, we identified several C. neoformans cell surface-defective TF mutants that displayed altered interaction upon exposure to B. safensis compared with the C. neoformans wild type. Specifically, the nrg1Δ, sre1Δ, and ert1Δ mutants all have cell membrane and/or cell wall defects (18,19,69). Second, the cell membrane-stabilizing agent sorbitol rescued B. safensis-mediated inhibition of melanization, and this effect was not observed with the nrg1Δ mutant. Third, previously published data revealed a striking dysregulation of the chitin-synthesis machinery in nrg1Δ cells (19), and exposure to B. safensis resulted in nrg1Δ cell morphologies reminiscent of those of chitin synthase mutants (22,23). Fourth, coincubation with B. safensis and the chitin-inhibiting agent calcofluor white resulted in synergistic inhibition of C. neoformans growth. Fifth, the chitinase inhibitor BisC partially rescued C. neoformans melanization during coculture with B. safensis. And sixth, WGA staining revealed significantly more chito-oligomers on the cell surfaces of C. neoformans and C. albicans cells following exposure to B. safensis. Conceivably, B. safensis may also produce glucanases in addition to chitnases; however, we did not detect synergism of inhibition of C. neoformans growth between B. safensis and Congo red, a dye that inhibits ␤-1,3-glucan. Consistent with our results on antifungal chitinase potency, a recent study from Yoo and Choi demonstrated anti-C. albicans and anti-Cryptococcus activity of a chitinase isolated from the mushroom Coprinellus congregatus (70).
Despite the similarities between the effects on C. neoformans and C. albicans, we also noted an interesting, medium-dependent difference during bimicrobial interactions with B. safensis. While C. albicans cells were moderately reduced in size following exposure to bacteria compared to fungal monocultures, C. neoformans cells nearly doubled in size. This increase in cell size was observed in nutrient-rich YPD medium ( Fig. S5C) but not in capsule-inducing medium (Fig. 4B) and may have been related to titan cell formation by C. neoformans, a process occurring during in vivo infection and associated with increased tolerance of attack from immune cells (71,72). Cell enlargement upon contact with B. safensis may therefore be a fungal response aimed at self-protection from bacterial attack under these culture conditions.
Due to the absence of cell walls in human cells, the fungal cell wall represents an excellent target for the development of antifungal drugs. Indeed, current antifungal development programs specifically aim to target this pathogen structure. Nikkomycin Z, for example, is a compound that targets fungal cell wall chitin synthesis and is currently in clinical trial for the treatment of coccidioidomycosis (73,74). Humans produce chitinases, and it is hypothesized that these enzymes have antimicrobial functions (75). Initial studies have shown that administration of chitotriosidase, the recombinant human chitinase, to mice reduced cryptococcal and candidal infection (76,77). Moreover, it was recently shown that chitotriosidase recognizes cryptococcal chitin during infection and promotes pathologic type-2 helper T cell responses (78).
In summary, we have identified an environmental soil bacterium that specifically blocks virulence factor elaboration by human-pathogenic fungi without affecting overall fungal growth. We identified bacterial chitinase activity as a major contributing factor that mediates these effects and propose that identifying additional microbes with pathogenicity-specific activities may represent a promising approach to identify novel, pathogen-specific drug targets.

MATERIALS AND METHODS
Strains and growth conditions. C. neoformans var. grubii strain H99 (serotype A), C. gattii strain R265 (molecular type VGIIa; serotype B), and C. albicans strain SC5314 were used as wild-type controls. Other strains used in this study are listed in Tables S1 and S2. Fungal strains were routinely maintained on YPD agar (1% yeast extract, 2% Bacto-peptone, 2% D-glucose, 2% agar). Overnight cultures were grown in liquid YPD medium in a shaking incubator at 30°C and 180 rpm. The environmental microbes (see Table S1) and E. coli strain DH5␣ were cultivated on LB agar (1% Bacto-tryptone, 0.5% yeast extract, 1% NaCl, 2% agar), and overnight cultures were grown in liquid LB in a shaking incubator at 30°C and 180 rpm.
Isolation of environmental microbes. Environmental microorganisms were isolated from C. gattiipositive soil samples (from Vancouver Island, BC, Canada) (15) and from plant leaves of Aucuba japonica and Urtica dioica (from the University of British Columbia [UBC] campus, BC, Canada). For each soil sample,~50 mg was suspended in sterile phosphate-buffered saline (PBS), vigorously shaken, and incubated statically at room temperature (RT) for 20 min to allow undissolved particles to sink to the bottom of the tube. Next, 100 l of supernatant was plated onto LB agar and incubated at 30°C for 24 to 48 h. Single colonies were then restreaked twice to ensure enrichment of single organisms. For isolation of microbes from plants, upper leaf surfaces were gently pressed onto YPD agar for 5 to 10 s, and plates were incubated at 30°C for 24 to 48 h. Colonies were restreaked three times onto YPD. Microorganisms were kept in 25% glycerol at Ϫ80°C for long-term storage.
Identification of microbes. Selected microbes were identified at the genus level by 16S rRNA sequencing according to previously published protocols (79), with minor modifications. Briefly, genomic DNA (gDNA) was isolated by bead beating and phenol-chloroform extraction according to standard protocols. Universal primers 8F (5=-AGAGTTTGATCCTGGCTCAG-3=) and 1492R (5=-GGTTACCTTGTTACGA CTT-3=) were then used to amplify the 16S rRNA gene sequence. PCR experiments were performed in a total volume of 50 l and contained 32.5 l water, 10 l buffer HF (5ϫ), 4 l deoxynucleoside triphosphate (dNTP) mix (with a 2.5 mM concentration of each dNTP), 1 l forward primer (8F), 1 l reverse primer (1492R), 1 l gDNA (30 ng l Ϫ1 ), and 0.5 l Phusion DNA polymerase (New England Biolabs). The following PCR cycle was used for amplification: initial activation at 98°C for 30 s; 32 cycles at 98°C for 10 s, 56°C for 30 s, and 72°C for 45 s; and a final extension at 72°C for 10 min. PCR products were purified using a PCR purification kit (Fermentas) and sequenced using primer 8F and/or primer 1492R (Genewiz). Nucleotide sequences were then assessed for base-caller errors, and~700-bp gene sequences were compared with reference 16S rRNA gene sequences by BLAST analysis at the National Center for Biotechnology Information (NCBI) website (https://blast.ncbi.nlm.nih.gov/Blast.cgi). The 16S rRNA gene sequences were deposited in GenBank (see below).
Growth curves and CFU assays. A potential impact of interspecies interactions on fungal growth was evaluated by cell number determinations performed using a hemocytometer, optical density measurements, or CFU analyses.
The analysis of cryptococcal growth dynamics based on cell number was performed according to previously published reports (80), with minor modifications. Briefly, fungal and bacterial overnight cultures were washed twice in PBS and resuspended in PBS. Fungal cells alone (2 ϫ 10 3 cells ml Ϫ1 ) or in a mix of fungal cells (2 ϫ 10 3 cells ml Ϫ1 ) and bacterial cells (2 ϫ 10 5 cells ml Ϫ1 ) were then incubated in 5 ml L-DOPA or YPD in the dark at 30°C and 180 rpm. At the indicated time points, culture aliquots were diluted as required and fungal cell numbers determined using a hemocytometer. Note that fungal cells were readily differentiable from bacterial cells in the dual-species culture due to their significantly greater size and distinct morphology.
Growth curve analyses were performed for C. neoformans and B. safensis monocultures in sterile, 96-well microtiter plates. Overnight fungal or bacterial cultures were washed twice in PBS and adjusted to an OD 600 of 0.1 in YPD or in YPD plus 50 g ml Ϫ1 gentamicin. The total medium volume per well was 200 l. The plates were sealed with a sterile adhesive foil and incubated in a microplate reader (Infinite M200; Tecan) at 30°C. Growth of the strains was then recorded by measuring the OD 600 at 30-min intervals for up to 72 h.
For CFU analyses, bacterial and fungal overnight cultures were washed twice in PBS and resuspended in PBS. Fungal cells alone (10 3 cells ml Ϫ1 ) or in a mix of fungal cells (10 3 cells ml Ϫ1 ) and bacterial cells (10 6 cells ml Ϫ1 ) were incubated in 1 ml YPD, YPD plus 200 g ml Ϫ1 Congo red (CR), or YPD plus 400 g ml Ϫ1 CFW at 30°C and 180 rpm. At the indicated time points, 50 l and 100 l of the appropriate dilutions were plated onto YPD plus 30 g ml Ϫ1 gentamicin. Plates were incubated for 2 to 3 days at 30°C and fungal colonies counted.
Melanin production. Melanin production was examined on solid or in liquid L-3,4-dihydroxyphenylalanine (L-DOPA; Sigma) medium containing 0.1% glucose. To screen for microbes with antimelanization activity, overnight bacterial and cryptococcal cultures were mixed at a 1:1 ratio and 7 l was dropped onto L-DOPA agar. Bacterium-and fungus-only controls were included. Plates were incubated in the dark at 30°C or 37°C for 48 h before being photographed. For some experiments, bacterial or fungal overnight cultures were washed twice in PBS and adjusted to an OD 600 of 20 or 4, respectively, before being mixed and dropped onto L-DOPA agar. Washing with PBS did not affect the bacterial antimelanization activity. To analyze the effects of chitinase on formation of melanin, fungal cells were resuspended in 1 mg ml Ϫ1 chitinase (from Streptomyces griseus; Sigma) solution before being spotted onto L-DOPA agar. To investigate if B. safensis viability is required for the antimelanization activity, bacterial cells were grown overnight, heat killed (15 min at 95°C), and exposed to C. neoformans on L-DOPA agar at 30°C for 48 h.
The effects of membrane stabilization or chitinase inhibition on B. safensis-mediated inhibition of C. neoformans melanization were tested by incubating monospecies or dual-species cultures on L-DOPA agar containing 1.5 M sorbitol or 2.5 mM bisdionine C (BisC), respectively.
To analyze the impact of B. safensis supernatant on melanin production by C. neoformans, bacteria were grown for 48 h at 30°C and 180 rpm in 50 ml YPD or potato dextrose broth (PDB; Difco) or yeast nitrogen base-low-iron medium (YNB-LIM [81]). Cultures were then centrifuged for 30 min at 4,200 rpm, and~10 ml of supernatant was subjected to filter sterilization (0.2 m pore size) and concentrated using 10 kDa-molecular-weight-cutoff tubes (Amicon Ultra) following the manufacturer's instructions. Concentrated supernatant from each bacterial preculture condition was then mixed with an overnight culture of C. neoformans at a ratio of 1:1 (vol/vol) and incubated on L-DOPA agar at 30°C for 48 h.
To analyze if B. safensis-spent L-DOPA medium provides enough substrate for C. neoformans to still melanize, bacterial cells were grown in L-DOPA at 30°C for 48 h and the supernatant was subjected to filter sterilization (0.2 m pore size) and reinoculated with C. neoformans. Uninoculated spent medium was included as a control. Samples were incubated at 30°C and 180 rpm for 48 h before being photographed.
Melanin shedding was analyzed according to previously published protocols (82), with minor modifications. Briefly, overnight fungal or bacterial cultures were washed twice in PBS and adjusted to 10 6 cells ml Ϫ1 or 10 9 cells ml Ϫ1 in PBS, respectively. The same volumes of fungal and bacterial cells were mixed (final ratio, 1:1,000), and 100 l was added to 5 ml L-DOPA. Fungus-only and medium-only controls were included, and samples were incubated in the dark at 30°C and 180 rpm for 46 h. Next, 1 ml culture was centrifuged at 1,000 ϫ g for 10 min, 100 l of supernatant was transferred to a 96-well plate, and the OD 405 was measured in a microplate reader (Infinite M200; Tecan).
Capsule formation. To investigate C. neoformans capsule formation in the presence or absence of B. safensis, stationary-phase fungal or bacterial cultures were washed once in PBS and resuspended in PBS. For capsule induction, 70 l of fungal cells alone or bacterial cells alone or of a mixture of fungal cells and bacterial cells was added to 2 ml 10% fetal calf serum (FCS; Gibco) or to 2 ml low-iron capsule induction medium (CIM; 5 g liter Ϫ1 glucose, 5 g liter Ϫ1 L-asparagine, 0.4 g liter Ϫ1 K 2 HPO 4 , 0.25 g liter Ϫ1 CaCl 2 · 2H 2 O, 0.08 g liter Ϫ1 MgSO 4 · 7H 2 O, 4.78 g liter Ϫ1 HEPES, 1.85 g liter Ϫ1 NaHCO 3 [dissolved in Chelex 100 resin-treated water], pH 7.4) supplemented with 100 g ml Ϫ1 chitinase or left unsupplemented and was incubated at 30°C and 180 rpm for 48 h. Samples were then stained with India ink to enable visualization of the polysaccharide capsule by differential interference contrast (DIC) microscopy.
TF mutant library screen. To identify a potential mechanism for the observed suppression of cryptococcal melanization by B. safensis, we performed a genomic screen using a C. neoformans transcription factor (TF) mutant library (18). The TF library encompasses 322 mutants with deletions in 155 different TFs (with at least two independent deletion strains per TF). In this study, we screened two independent mutants each for 154 of these TFs (a total of 308 mutants). TF mutants were grown overnight in 96-well plates in YPD at 30°C. TF mutant cultures were then diluted 1:40 in liquid L-DOPA and incubated either alone or with bacterial cells (final OD 600 ϭ 0.0025) in a final volume of 200 l L-DOPA. Plates were incubated in the dark at 30°C for 5 days. To assess overall growth, the OD 405 was then measured for each plate using a microplate reader (Infinite M200; Tecan).
Hypha formation. C. albicans hypha formation in the presence or absence of bacteria was investigated both on solid and in liquid media. For solid plate assays, overnight C. albicans, B. safensis, and E. coli cultures were washed twice in PBS and adjusted to 2 ϫ 10 6 cells ml Ϫ1 each in PBS. Fungal and bacterial cells were mixed in a 1:1 ratio, and 7 l was spotted onto solid water agar supplemented with 10% FCS. Fungus-only cultures were included as a positive control. Plates were incubated at 30°C for 8 to 9 days before being photographed.
For analysis of filament formation in liquid medium, overnight fungal and bacterial cultures were washed twice in PBS and cell numbers were adjusted to 10 7 cells ml Ϫ1 and 10 8 cells ml Ϫ1 in PBS, respectively. A 10-l volume of fungal cells Ϯ 10 l of bacterial cells was then added to 5 ml YPD or to 2 to 5 ml 10% FCS supplemented with or without 100 g ml Ϫ1 chitinase. YPD cultures were incubated at 30°C and 180 rpm for 24 h to induce yeast-phase growth, and cultures prepared in 10% FCS were incubated at 37°C and 180 rpm for 24 h to induce hypha formation. Samples were then analyzed by DIC microscopy. To estimate C. albicans yeast and hyphal mass production, samples were kept at 4°C overnight to allow the heavier fungal cells to sink to the bottom of the glass tube. Next, the supernatant, containing mostly bacterial cells, was carefully removed and the remaining cells were resuspended in PBS and transferred to preweighed microcentrifuge tubes. The supernatant was carefully removed in two centrifugation steps (3 min at 15,000 rpm), and the microcentrifuge tubes containing the pellets were weighed. Note that this procedure provides only an estimate of fungal yeast and hypha production, as some bacterial cells are likely to be carried over following the overnight static incubation step. Due to the small size of bacterial cells relative to fungal cells, however, these values should be negligible.
To study C. albicans hypha formation with or without B. safensis on the single-cell level, overnight fungal and bacterial cultures were washed once in PBS and resuspended to levels of 10 7 cells ml Ϫ1 and 10 9 cells ml Ϫ1 in PBS, respectively. For coculture, 10 l fungal cells and 10 l bacterial cells (ratio of 1:100) were added to 3 ml 10% FCS and incubated in a humidified incubator at 37°C with a 5% CO 2 atmosphere. Fungus-only cultures were used as controls. After 4 h and 24 h, aliquots were taken and analyzed by DIC microscopy.
To investigate the impact of various fungus-bacterium ratios on C. albicans filamentation, hypha induction assays were performed in sterile, polystyrene, flat-bottom, 24-well microtiter plates (Corning).
Overnight fungal or bacterial cultures were washed three times in PBS and cell numbers determined using a hemocytometer. C. albicans was adjusted to 2 ϫ 10 4 cells ml Ϫ1 in 10% FCS, and 500 l was added per well (C. albicans-only control). For fungus-bacterium cocultures, the same number of fungal cells (10 4 cells per well) was exposed to bacterial cells at ratios of 1:1, 1:10, 1:100, and 1:1,000 in a total volume of 500 l 10% FCS per well. Plates were then incubated for 4 h in a humidified incubator at 37°C with a 5% CO 2 atmosphere. Next, all wells were washed three times with PBS. The surface-adherent fungal cells with or without attached bacteria were fixed with 4% formaldehyde, and the plates were kept at 4°C for short-term storage until microscopic analysis.
To study the effect of B. safensis viability on the capacity to attach to C. albicans hyphae, fungal cells were exposed to live or heat-killed (15 min, 98°C) bacterial cells in 10% FCS at 37°C and 5% CO 2 for 24 h. Aliquots were then analyzed by fluorescence microscopy.
Adhesion assays. C. albicans adhesion assays with or without B. safensis were performed in sterile, polystyrene, flat-bottom, 24-well microtiter plates (Corning). Overnight fungal or bacterial cultures were washed three times in PBS and cell numbers determined. C. albicans was adjusted to 2 ϫ 10 4 cells ml Ϫ1 in 10% FCS, and 500 l was added per well (C. albicans-only control). For fungus-bacterium cocultures, the same number of fungal cells (10 4 cells per well) was exposed to bacterial cells at ratios of 1:1, 1:10, 1:100, and 1:1,000 in a total volume of 500 l 10% FCS per well. Plates were then incubated for 45 min in a humidified incubator at 37°C with 5% a CO 2 atmosphere. All wells were washed three times with PBS to remove bacteria and nonadherent C. albicans cells. The surface-adherent fungal cells were fixed with 4% formaldehyde, and plates were kept at 4°C for short-term storage until microscopic analysis. Quantification of C. albicans adherence was performed using an inverse microscope (Zeiss ID03). The number of adhered cells was determined by counting cells in 20 random high-power fields (HPFs) (size, 0.238 mm 2 per well). The experiment was performed twice in quadruplicate (160 HPFs in total examined per condition).
Biofilm assays. C. neoformans biofilm experiments were performed with or without B. safensis by crystal violet staining as previously described (83), with minor modifications. Briefly, overnight fungal and bacterial cultures were washed three times in PBS and resuspended in Dulbecco's modified Eagle media (DMEM; Gibco) to reach a final concentration of 10 7 cells ml Ϫ1 as individual bacterial or fungal cultures or as a mixture of the two organisms (1:1 ratio, with 10 7 cells ml Ϫ1 each). Next, 300 l of fungal cells, bacterial cells, or mixed fungal cells-bacterial cells were transferred into individual wells of sterile, polystyrene, flat-bottom, 24-well microtiter plates (Corning). Wells with DMEM only were included as controls. Plates were incubated at 30°C or at 37°C for 48 h in plastic bags (to avoid medium evaporation). Each well was then washed twice with sterile water and air-dried for 5 min at RT. For biofilm quantification, 100 l 0.3% crystal violet solution was added to each well (including the medium-only control wells) and plates were incubated at RT for 5 min. Next, each well was thoroughly washed twice with sterile water and biofilms were destained with 200 l 100% ethanol for 5 min at RT. Finally, 75 l of destaining solution was transferred to a new 96-well microtiter plate and the OD 550 measured in a microplate reader (Infinite M200; Tecan). Medium-only control values were subtracted from all measurements.
For analysis of de novo C. albicans biofilm formation with or without B. safensis, overnight fungal and bacterial cultures were washed three times in PBS and resuspended in 10% FCS to a final concentration of 2 ϫ 10 4 cells ml Ϫ1 . Next, 500 l of cells was transferred to wells of sterile, polystyrene, flat-bottom, 24-well plates (Corning). For fungus-bacterium cocultures, equivalent numbers of fungal cells (10 4 cells per well) were exposed to bacterial cells at ratios of 1:1, 1:10, 1:100, and 1:1,000 in a total volume of 500 l 10% FCS per well. Wells with 10% FCS medium only were included as controls. Plates were then incubated for 24 h in a humidified incubator at 37°C with a 5% CO 2 atmosphere. Next, wells were washed three times with PBS and air-dried for 5 min at RT. Biofilm quantification was then performed using crystal violet staining as described above for C. neoformans.
To investigate the effect of B. safensis on preformed C. albicans biofilms, fungal cells were allowed to form biofilms for 24 h according to the protocol described above. Supernatants were then removed, and 10 4 , 10 5 , 10 6 , or 10 7 bacterial cells in 10% FCS (corresponding to a fungus-bacterium ratio of 1:1, 1:10, 1:100, or 1:1,000 with respect to initial inocula) were added to fungal biofilm-containing wells. Plates were then returned to 37°C and 5% CO 2 and further incubated for 24 h. Next, each well was washed three times with PBS and biofilm production quantified via crystal violet staining as described above.
Cell membrane and cell wall staining. To assess the impact of B. safensis on the C. neoformans cell membrane, monospecies or dual-species cultures were grown overnight in YPD at 30°C and 180 rpm and cells were washed in PBS and stained with FM4-64 (Invitrogen) at a final concentration of 20 M in the dark at RT for 20 min (84). Cells were then washed twice in PBS and analyzed by fluorescence microscopy. For analysis of a potential impact of bacteria on fungal cell wall chitin, monospecies or dual-species cultures were grown overnight in YPD at 30°C and 180 rpm, washed three times in PBS, and fixed in 4% formaldehyde at 4°C for 30 min. Next, cells were stained with calcofluor white (CFW; Sigma) at a final concentration of 25 M in the dark at 37°C for 30 min. Cells were then washed three times in PBS and analyzed by fluorescence microscopy. For simultaneous detection of chito-oligomers, cultures were prepared and washed in PBS as described above and then stained with fluorescein-conjugated wheat germ agglutinin (WGA; Invitrogen) at a final concentration of 5 g ml Ϫ1 in the dark at 37°C for 30 min. Cells were then washed three times in PBS, stained with CFW as described above, and visualized by fluorescence microscopy using appropriate filter sets.

ACKNOWLEDGMENTS
We thank Joseph Heitman (Duke University) for C. neoformans H99, Karen Bartlett (University of British Columbia) for soil samples and C. gattii R265, Malcolm Whiteway (Concordia University) for C. albicans SC5314, J. Andrew Alspaugh (Duke University) for the nrg1Δ mutant, Yong-Sun Bahn (Yonsei University) for the C. neoformans transcription factor mutant library, Jennifer Lodge (Washington University) for the lac1Δ mutant, and Hiten Madhani (University of California, San Francisco) and the NIH (R01AI100272) for providing the hog1Δ and mpk1Δ mutants as part of the 2015 C. neoformans knockout collection (distributed via the Fungal Genetics Stock Center). Furthermore, we thank Daniel Mosquin and Douglas Justice (University of British Columbia) for help in plant identification and Audrey Tupin (University of British Columbia) for help with microscopy. We also thank Kaila Pianalto (Duke University) and all members of the Kronstad laboratory for helpful discussions.
F.L.M. is the grateful recipient of a research fellowship from the Deutsche Forschungsgemeinschaft (grant MA6248/1-1). J.W.K. is a Burroughs Wellcome Fund Scholar in Molecular Pathogenic Mycology. This work was also supported by a grant from the Canadian Institutes of Health Research (to J.W.K.). The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.