Abstract
Drug resistance is a critical challenge in treating infectious disease. For fungal infections, this issue is exacerbated by the limited number of available and effective antifungal agents. Patients infected with the fungal pathogen Cryptococcus are most effectively treated with a combination of amphotericin B and 5-fluorocytosine (5FC). Infections frequently develop resistance to 5FC although the mechanism of this resistance is poorly understood. Here we show that resistance is acquired more frequently in isolates with defects in DNA mismatch repair that confer an elevated mutation rate. Natural isolates of Cryptococcus with mismatch repair defects have recently been described and defective mismatch repair has been reported in other pathogenic fungi. In addition, whole genome sequencing was utilized to identify mutations associated with 5FC resistance in vitro. Using a combination of candidate-based Sanger and whole genome Illumina sequencing, the presumptive genetic basis of resistance in 10 independent isolates was identified, including mutations in the known resistance genes FUR1 and FCY2, as well as a novel gene, UXS1. Mutations in UXS1 lead to accumulation of a metabolic intermediate that appears to suppress toxicity of both 5FC and its toxic derivative 5FU. Interestingly, while a UXS1 ortholog has not been identified in other fungi like Saccharomyces cerevisiae, where the mechanisms underlying 5FC and 5FU resistance were elucidated, a UXS1 ortholog is found in humans, suggesting that mutations in UXS1 may also play a role in resistance to 5FU in its role as a human cancer chemotherapeutic.
Introduction
One of the key challenges of the 21st century is the emergence and reemergence of pathogens. Opportunistic fungal pathogens comprise an important component of this problem as they infect the rapidly expanding cohort of immunocompromised patients [1]. These pathogens are responsible for millions of infections annually, with substantial mortality. Among the most dangerous are Cryptococcus species that cause approximately 220,000 infections a year, with more than 181,000 attributable deaths [2]. Cryptococcosis is particularly prominent in Sub-Saharan Africa, where the HIV/AIDS epidemic has resulted in a large population of susceptible individuals. Cryptococcosis is treated most effectively using a combination of 5-fluorocytosine (5FC) and amphotericin B [3,4]. However, in the parts of Africa where patients are most commonly afflicted with cryptococcosis, the medical infrastructure is insufficient to allow treatment with the highly toxic amphotericin B component of this dual therapy. Instead patients are typically treated with fluconazole monotherapy, with limited success. Excitingly, recent studies have shown that 5FC can be effectively paired with fluconazole to replace amphotericin B for treatment of patients in Africa [5]. However, 5FC is not yet approved or available for treatment in any African countries.
5FC acts as a prodrug, which enters cells via the cytosine permease Fcy2. 5FC itself is not toxic, but upon uptake into fungal cells, it is converted into toxic 5-fluorouridine (5FU) by cytosine deaminase, an enzyme that is not present in human cells [6]. In Cryptococcus, and other fungi, cytosine deaminase is encoded by the FCY1 gene. 5FU is then further processed by the product of the FUR1 gene, a uracil phosphoribosyltransferase, and inhibits both DNA and protein synthesis. Resistance is well understood in other fungal pathogens, like Candida albicans, where loss of function mutations in FCY1, FCY2, and FUR1 can mediate resistance [7]. In Candida lusitaniae, mutations in FUR1 can be readily distinguished from mutations in FCY1 and FCY2 because only fur1 mutations result in cross-resistance to 5FU [8]. Likewise, in Candida dubliniensis, natural missense fur1 mutations affect both 5FC and 5FU resistance [9]. However, little work has been conducted on 5FC resistance directly in Cryptococcus. One of the few early studies suggested that reductions in FUR1 activity may be linked to resistance to 5FC based on a high frequency of cross-resistance to 5FU [10]. However, this study took place prior to the cloning or sequencing of the FUR1 gene in Cryptococcus and attribution of resistance to FUR1 was based only on cross-resistance to 5FU. More recent studies of 5FC resistant Cryptococcus bacillisporus isolates found no mutations in FCY1, FUR1, or any of three putative FCY2 paralogs that explained drug resistance [11].
Recent work has demonstrated one source of increased rates of resistance to antifungal drugs in Cryptococcus: defects in the DNA mismatch repair pathway [12,13]. Natural isolates with DNA mismatch repair defects have been identified in both an outbreak population of Cryptococcus deuterogattii [12,14] and in Cryptococcus neoformans [13,15]. Defects in mismatch repair are also common in other human fungal pathogens, including Candida glabrata [16]. Depending on the population studied, multidrug resistance is sometimes linked to the hypermutator state in C. glabrata [17,18]. Here we demonstrate that DNA mismatch repair defects also enable rapid resistance to 5FC in C. deuterogattii (previously known as C. gattii VGII [19–21]). We then utilize whole genome Illumina sequencing, in combination with candidate-based Sanger sequencing, to identify the genetic basis for drug resistance in 10 independent isolates. We attribute resistance to mutations in FUR1 and unexpectedly, we also identify a novel pathway of resistance to 5FC involving mutations in the pathway responsible for producing the capsule, a core component of Cryptococcal virulence.
Results
In a previous study, we demonstrated that mismatch repair mutations conferred increased rates of resistance to the antifungal drugs FK506 and rapamycin [12]. Because these hypermutator strains are found among both environmental and clinical isolates, here we tested if a hypermutator state could also confer resistance to one of the front-line drugs used to treat Cryptococcosis: 5-fluorocytosine (5FC). A semi-quantitative swabbing assay was first employed to demonstrate that deletions of the mismatch repair gene MSH2 in Cryptococcus deuterogattii confer an elevated rate of resistance to 5FC (Figure 1A). This result was confirmed using a quantitative fluctuation assay approach (Figure 1B). This assay revealed a greater than 15-fold increase in the generation of resistance to 5FC in msh2Δ mismatch repair defective mutants. Similarly, a simple spreading assay using VGIIa-like strains that had previously been found to harbor an msh2 nonsense allele [12] demonstrated a much higher rate of resistance to both 5FC and 5FU than in the VGIIa non-hypermutator strains (Supplemental Figure 1).
In previous studies, mutator alleles in C. deuterogattii were not found to be generally advantageous in rich media [12]. However, under stressful conditions, such as drug challenge with FK506 and rapamycin, mutator alleles were highly beneficial. A competitive growth experiment was utilized to test the same concept with 5FC. Mutator strains became resistant to 5FC at a higher rate and thus rapidly outcompeted wildtype strains (Figure 2). However, in the absence of added stress, the mutator alleles showed no such advantage. This result suggests that drug challenge during infection may select for strains with elevated mutation rates that are able to acquire drug resistance more rapidly.
In other fungi, resistance to 5FC is typically mediated by mutations in one of three genes: FCY1, FCY2, or FUR1 [7,8,10,22]. As described above, mutations in FCY1 and FCY2 are typically distinguishable from fur1 mutations because mutations in FUR1 confer resistance not only to 5FC but also to 5FU. In contrast, fcy1 and fcy2 mutations confer resistance to only 5FC. To define the mechanism underlying 5FC resistance in C. deuterogattii, 29 resistant colonies were isolated and tested, originating from the wildtype (R265, 9 colonies) and from two independent msh2Δ mutants derived in the R265 background (RBB17, 10 colonies and RBB18, 10 colonies). Cultures were started from independent colonies and a single resistant colony was selected from each culture, so that only one resistant isolate is derived from any original colony derived from the frozen stock. All of the 5FC resistant isolates (Table 1) acquired were cross-resistant to 5FU (29/29) (Figure 3A), suggesting that resistance to 5FC in Cryptococcus deuterogattii was most commonly mediated by mutations in FUR1.
However, when the FUR1 gene was sequenced in this set of 5FC/5FU resistant isolates, unexpectedly, only three out of 29 isolates (10.3%) were found to have sustained mutations in FUR1 (R265-3, R265-4, and R265-6) (Table 1). Because fur1 mutations were the only known cause of 5FC/5FU cross-resistance, we performed whole genome Illumina sequencing on a subset of the remaining isolates to identify unknown genes underlying resistance. We sequenced 3 additional R265 isolates, 8 additional RBB17 isolates, and 9 additional RBB18 isolates, for a total of 20 5FC and 5FU resistant isolates.
From the sequenced genomes, reads were aligned to the R265 reference genome and SNPs and indels were identified. This analysis revealed that some of the presumed independent isolates were in fact siblings. Four groups of siblings existed (RBB17-3 and RBB17-4; RBB17-5 and RBB17-8; RBB18-2, RBB18-4, and RBB18-5; RBB18-6 and RBB18-9), resulting in a total of 3 independent R265 genomes, 6 independent RBB17 genomes, and 6 independent RBB18 genomes.
Of these 15 independent genome sequences, two contained unambiguous mutations in FUR1. One strain (R265-2), for which PCR amplification of the FUR1 locus had failed, showed an approximately 20 kb deletion. One end of the deletion lies within FUR1, consistent with the failed PCR. The other end of the deletion fell within a sequencing gap of the annotated V2 R265 reference genome. To identify the precise location of this second breakpoint, reads from R265-2 were mapped to a recent Nanopore and Illumina hybrid assembly of the R265 strain [23]. Interestingly, the second breakpoint was found within a gene encoding a weak paralog of FUR1 (5 x 10−10 protein BLAST e-value). This paralog (CNBG_4055) is also present in C. neoformans (CNAG_2344), suggesting that if it arose via duplication, it was before the last common ancestor to both species. Given that deletion of FUR1 confers resistance to 5FC and 5FU, it is unlikely that this paralog performs the same function as Fur1 (Figure 3A). Despite the protein similarity, no obvious nucleotide homology was found that may have mediated this large deletion conferring 5FC resistance. In fact, the FUR1 paralog is inverted relative to FUR1, reducing the likelihood that remnant homology may have generated a region susceptible to frequent homology-mediated deletion of FUR1 that would yield the type of regional deletion observed here.
The second fur1 mutation discovered by whole genome sequencing was a single base deletion that introduced a frameshift (R265-1) that had not initially been detected via Sanger sequencing. A Gly190Asp fur1 missense mutation was also identified in the msh2 mutant background (RBB17-5 and RBB17-8 sibling pair) (Table 1). However, this mutation was present in the sequencing of each strain at approximately 50% frequency, which would typically suggest a heterozygous variant. Because the starting strains used were haploid, and there was no indication of local duplication or any other indication of heterozygous variants in the genomes of these strains, it seems unlikely that these data are indicative of a heterozygous mutation. One alternate explanation is that the strains sequenced were mixed cultures or that the fur1 mutation reverted during the expansion of the culture for whole genome sequencing, which was not performed under selection. A test of individual colonies from the frozen culture of both sibling strains showed that 10 out of 10 colonies from each strain demonstrated both 5FC and 5FU resistance, suggesting that these strains were either a mixed culture of two different mutations that both confer resistance to 5FU and 5FC, or that fur1 mutations were lost during outgrowth for sequencing (Supplemental Figure 2).
A Trp167STOP mutation in FCY2 (CNBG_3227) was also detected in the sequenced set (RBB18-2, RBB18-4, and RBB18-5 sibling strains). While one of these mutations was nearly unambiguous (84% alternate allele, RBB18-4), RBB18-2 and RBB18-5 exhibited more mixed sequence at this locus. When individual colonies were isolated and retested from RBB18-2 and RBB18-5, they all showed resistance to both 5FC and 5FU (Supplemental Figure 2). Mutations in FCY2 were particularly unexpected because in other fungi they do not confer resistance to 5FU and because there are 2 additional paralogs of FCY2 present in the Cryptococcus genome. We attempted to test the ortholog of FCY2 from Cryptococcus neoformans using a deletion collection strain but found that the mutant in the collection retained a functional FCY2 allele. It is possible that this mutation may be a false positive, especially because all three of these sibling strains contained a second mutation in a gene that also plays a role in 5FC and 5FU resistance (below).
In total, out of 29 original 5FC resistant strains (Table 1), six independent fur1 mutations were identified using Sanger and Illumina sequencing. One independent fcy2 mutation was identified by Illumina sequencing. We did not identify any fcy1 mutations, although fcy1 mutations confer resistance to 5FC in Cryptococcus neoformans (Supplemental Figure 3). In total, 13 sequenced genomes representing 11 independent isolates remained with no mutations in any genes known to have a role in 5FC or 5FU resistance. These genomes were examined to identify novel candidate mutations. To distinguish causal variants from background mutations, candidate genes were required to be mutated in at least two different independent isolates. Variant impact was also scored using SNPeff [24] and mutations were not considered if predicted to have low impact (i.e., synonymous, intronic, or non-coding variants). Mutations of moderate or higher impact were identified at a total of 56 sites (Supplemental Table 3). To further prioritize, we specifically focused on mutations that were present in isolates from more than one of the parental backgrounds. We identified UXS1, which sustained four novel mutations in seven isolates from two parental backgrounds (Figure 3B).
UXS1 encodes the enzyme that converts UDP-glucuronic acid to UDP-xylose [25]. This pathway is critical for the formation of the capsule, a core virulence trait of Cryptococcus, and for synthesis of other glycoconjugates. There is no UXS1 ortholog in either Saccharomyces cerevisiae or Candida albicans, where many of the resistance mechanisms for 5FC were elucidated. The mutations in UXS1 included a single base deletion in a 3 T homopolymer (R265-5), a single base insertion in a 7 C homopolymer (RBB18-8), and a missense mutation (Tyr217Cys, RBB18-6 and RBB18-9 sibling pair) that, like some of the previously identified FUR1 mutations, displayed mixed sequences at the mutation site (Figure 3B, Table 1). Finally, a uxs1 mutation (Asp306Gly) was identified in the three sibling isolates that also had fcy2 mutations (RBB18-2, RBB18-4, and RBB18-5 siblings). Both the uxs1 and fcy2 mutations were not present in 100% of the reads. However, both mutant alleles had allele frequencies >50%, suggesting the genome sequence was not just a mix of a uxs1 mutant strain and an fcy2 mutant strain, but instead that both mutations were present in at least a portion of the cells in the culture. Among the sequenced isolates, mixed allele frequencies appeared only in the hypermutator strains, suggesting that the rapid rate of mutation in these isolates may have contributed to difficulties acquiring or maintaining a clonal population during the expansion of cultures used to prepare DNA for whole genome sequencing, although more hypermutator strains were sequenced than wildtype strains. In sum, 9 sequenced genomes representing 8 independent isolates remained for which we were unable to identify a mutation that conferred resistance to 5FC and 5FU, all derived from msh2 mutant isolates.
To confirm the role of uxs1 mutation in resistance to 5FC and 5FU, a uxs1 deletion available from a C. neoformans deletion collection was employed (Figure 4A). This uxs1Δ strain was completely resistant to both drugs, suggesting that all three alleles isolated were likely loss of function mutations because they shared a drug resistance phenotype with the null mutant. We next sought to genetically define the mechanism by which drug resistance may be mediated by loss of uxs1 function. Multiple models were considered to explain why 5FC/5FU toxicity would require Uxs1. The first was that Uxs1 directly converts 5FU into a toxic product. If so, Uxs1 and Fur1 would function in the same pathway, as either mutant independently confers drug resistance. This hypothesis was tested using an overexpression allele of UXS1 that is driven by the actin promoter [26]. If this hypothesis were correct, we would expect to observe additional sensitivity conferred by the overexpression allele compared to wildtype. By reducing the amount of 5FU used to only 1 μg/mL, wildtype strains were only partially inhibited. However, introduction of an overexpression allele of UXS1 did not increase sensitivity (Figure 4B). This suggests that Uxs1 does not act by converting 5FU or a 5FU derivative into a toxic product.
We next tested whether 5FC resistance in uxs1 mutants may occur through an indirect effect of the role of Uxs1 in synthesis of UDP-xylose. UDP-xylose is the donor molecule for xylose addition to glycans, a process that primarily occurs in the secretory compartment. If xylosylation of an unknown glycoconjugate is required to mediate 5FC toxicity, mutation of UXS1 would indirectly confer drug resistance. To test this, deletion mutants lacking transporters that move UDP-xylose into the secretory compartment (uxt1, uxt2, and a uxt1 uxt2 double mutant [27]) or that lack Golgi xylosyl-transferases that act in protein, glycolipid, and polysaccharide synthesis (cxt1 [28], cxt2, and a cxt1 cxt2 double mutant) were analyzed. None of these mutants demonstrated any change in sensitivity to 5FC or 5FU (Figure 4C). However, these data did not rule out a requirement for a (previously undescribed) cytoplasmic xylosyl protein modification. To test this hypothesis, a mutant that cannot generate UDP-glucuronic acid, the immediate precursor for UDP-xylose synthesis was used. This mutant (ugd1) is somewhat growth impaired relative to wildtype and cannot grow on YNB media. However, it does grow, albeit poorly, on rich YPD media, where it clearly exhibited sensitivity to 5FC. This result demonstrated that xylose modification, in any cellular compartment, is not required for 5FC toxicity (Figure 4D).
The previous models ruled out the lack of UDP-xylose for synthetic processes as an explanation for 5FC resistance. Another result of the loss of UXS1 function is the accumulation of UDP-glucuronic acid, the immediate precursor in the production of UDP-xylose. Past studies have shown that UDP-glucuronic acid accumulates to extremely high levels in uxs1 mutant cells [29]. To test whether this mediates resistance, we generated a uxs1 ugd1 double mutant, which should produce neither compound [29]. While the uxs1 ugd1 mutant was growth impaired, like the ugd1 single mutant, it was clearly sensitive to 5FC (Figure 4D). That uxs1 mutants are 5FC resistant, whereas uxs1 ugd1 double mutants are restored to 5FC sensitivity suggests that accumulation of UDP-glucuronic acid in uxs1 mutants mediates resistance to 5FC and 5FU (Figure 5).
Discussion
Treating fungal diseases is complicated both by the limited number of drugs that effectively treat infection without harming the patient and by the rapid rate at which fungi develop resistance to the few drugs that are effective. 5FC is a particularly emblematic example of this issue, as it is highly efficacious with limited toxicity. Human cells lack the ability convert 5FC to 5FU and toxicity is conferred only by the conversion of 5FC to the chemotherapeutic 5FU by a patient’s microbiota [30]. However, 5FC is ineffective when used for solo treatment because fungal resistance rapidly emerges. Here, we demonstrate that DNA mismatch repair mutants exhibit accelerated acquisition of resistance to 5FC. Evolutionary theory predicts that hypermutators should be rare in eukaryotic microbes because sex unlinks mutator alleles from the mutations they generate, eliminating the advantage of an elevated mutation rate and leaving only the general decrease in fitness from introduced mutations [31]. This result lends further support to the recent appreciation that mismatch repair mutants may be common in pathogenic fungi in part because treatment with antifungal drugs increases selection for mutations that generate resistance [12,13,15,16]. The potential instability observed in several of these mutations in msh2 mutants may suggest the capacity to revert nonbeneficial mutations once drug treatment ends, particularly in the context of a pathogen like Cryptococcus that is primarily environmental. Previous work showed this type of direct reversion of an auxotrophic ade2 mutation [12].
We explored the underlying genetic and genomic basis of 5FC resistance. The resistant mutants in C. deuterogattii selected here were cross-resistant to 5FU. Sanger and whole genome Illumina sequencing identified a presumptive genetic basis for drug resistance in 10 independent isolates. Analysis of resistance loci was relatively facile in wildtype strains, where an average of 1.66 coding mutations were identified by whole genome sequencing, including the putative resistance mutation, relative to the reference. However, this analysis was substantially more difficult in mutator strains where an average of 7.9 coding mutations were found per strain, with numerous additional noncoding or synonymous mutations. In addition, the phenomenon of mixed allele ratios in sequencing data was only observed in hypermutator strains. Likewise, sibling strains emerged from the selection, despite use of standard genetic best practices for isolating independent resistant mutants. This suggests that the initial freezer stock from each hypermutator strain had substantial existing mutations and population structure, which is not typically an issue for frozen Cryptococcus cultures. For the purposes of identifying the genetic basis of a trait that occurs at a high rate in wildtype, future studies would be advised to avoid mutations that increase mutation rate, as they contribute to background noise.
Mutations in UXS1 are particularly interesting as a mechanism of resistance in Cryptococcus because Uxs1 catalyzes the production of UDP-xylose, the donor molecule for essential components of Cryptococcal capsule polysaccharides. Strains lacking UXS1 are hypocapsular with altered capsule structure [29]. In addition, uxs1 mutants are avirulent in a murine tail-vein injection disseminated infection model [32]. This suggests that uxs1 mutants might be unlikely to emerge during exposure to 5FC in vivo, even though they represent a substantial proportion of the resistant isolates observed in this study. Future studies examining the mechanisms of resistance during treatment with 5FC in vivo will provide further insights into the possible contribution of uxs1 mutations to resistance in patients.
This study also illustrates the importance of examining drug resistance in the context of the pathogen being treated. Previous work in C. albicans and S. cerevisiae suggested that resistance would occur through mutations in FUR1, but both species are evolutionarily distant from Cryptococcus and lack a UXS1 ortholog. While these previous studies provided substantial insight into 5FC toxicity, studies in the pathogen of interest are essential. Surprisingly, one set of sibling strains (RBB18-2, RBB18-4, RBB18-5) that were cross resistant to 5FU had mutations in the FCY2 gene (CNBG_3227), which in other species confers resistance to 5FC but not 5FU. Unexpected cross-resistance between 5FC and fluconazole has been previously observed in fcy2 mutants of Candida lusitaniae but is proposed to occur through competitive inhibition of fluconazole uptake by 5FC that can no longer enter through Fcy2-mediated transport [8,33,34]. C. lusitaniae fcy2 mutants are not resistant to fluconazole without the addition of 5FC. In addition, multiple resistant strains were not assigned a presumptive causative mutation here and lacked mutations in any genes known to cause 5FC resistance from this or previous work (FUR1, FCY1, FCY2, and UXS1). Presumably unknown mechanisms are responsible for resistance to 5FC and 5FU in these strains as well, either in pathways unique to Cryptococcus or potentially more broadly conserved.
In addition, UXS1 mutations provide unexpected insight into interaction between nucleotide synthesis and generation of precursors for xylosylation. Surprisingly, accumulation of UDP-glucuronic acid appears to either inhibit the pyrimidine salvage pathway or activate thymidylate synthase (Figure 5). This suggests that UDP-glucuronic acid may have a role as a source of UDP for the cell, while UDP-xylose does not. While UXS1 orthologs are not found in C. albicans or S. cerevisiae, which lack xylose modifications, there is a UXS1 ortholog in humans. 5FU is commonly used as a chemotherapeutic drug [35], and resistance to 5FU is frequently associated with mutations in thymidylate synthase [36]. Data here suggest that uxs1 mutations may be acting in a similar fashion to either de-repress thymidylate synthase or inhibit Fur1 (Figure 5). Further exploration of the role of Uxs1 orthologs in humans during 5FU chemotherapy may be of interest.
Material and methods
Strains and media
The strains and plasmids used in this study are listed in Table S1. The strains were maintained in glycerol stocks at −80°C and grown on rich YPD media at 30°C (Yeast extract Peptone Dextrose). Strains with selectable markers were grown on YPD containing 100 μg/mL nourseothricin (NAT) and/or 200 μg/mL G418 (NEO).
Genome sequencing
DNA was isolated for sequencing by expanding individual colonies to 50 mL liquid cultures in YPD at 30°C. Cultures were then frozen and lyophilized until dry. DNA was extracted using a standard CTAB extraction protocol as previously described [37]. Illumina paired-end libraries were prepared and sequenced by the University of North Carolina Next Generation Sequencing Facility. Raw reads are available through the Sequence Read Archive under project accession number PRJNA525019.
Genome assembly and variant calling
Reads were aligned to the V2 R265 reference genome [38] using BWA-MEM [39]. Alignments were further processed with SAMtools [40], the Genome Analysis Toolkit (GATK) [41], and Picard. SNP and indel calling was performed using the Unified Genotyper Component of the GATK with the haploid setting. VCFtools [42] was utilized for processing of the resulting calls and variants were annotated using SnpEff [24]. Variant calls were visually examined using the Integrated Genome Viewer (IGV) [43]. FungiDB was also used to determine putative function and orthology of genes containing called variants in the dataset [44].
Strain construction
A ugd1Δ mutant was constructed in the KN99a background as follows. Primers pairs JOHE45233/JOHE45085, JOHE45086/JOHE45087, and JOHE45088/JOHE45234 were used to amplify 1 kb upstream of UGD1, the neomycin resistant marker, and 1 kb downstream of the UGD1 gene, respectively. To generate the deletion allele for C. neoformans transformation, all three fragments were cloned into plasmid pRS426 by transforming S. cerevisiae strain FY834 as previously described [45]. Recombinant S. cerevisiae transformants were selected on SD-uracil media and verified by spanning PCR with primer pair JOHE45233/JOHE45234. The resulting PCR product was introduced into C. neoformans laboratory strain KN99a by biolistic transformation and transformants were selected on YPD containing neomycin. Putative ugd1Δ deletion mutants were confirmed by PCR.
uxs1Δ single mutants and ugd1Δ uxs1Δ double mutants were generated via a genetic cross. First, the KN99α uxs1Δ mutant from the Hiten Madhani deletion collection was mated with the wild-type KN99a laboratory strain. Through microdissection, spores were isolated, germinated, and genotyped via PCR for the gene deletion and the mating type locus to isolate a MATa uxs1Δ mutant in the KN99 background. Second, the KN99a uxs1Δ mutant was mated with wild-type H99. Spores were dissected and genotyped via PCR for the gene deletion and the mating type locus to isolate H99 uxs1Δ single mutants. Finally, the H99 uxs1Δ single mutant was crossed with KN99a ugd1Δ to generate ugd1Δ uxs1Δ double mutants, and the H99 ugd1Δ single mutant.
Spot dilution assays
Single colonies were inoculated into 5 mL of liquid YPD and grown overnight at 30°C. Cell density was determined using a hemocytometer and the cultures were diluted accordingly such that 100,000 cells were aliquoted on to the most concentrated spot and subsequent spots consisted of 10-fold dilutions per spot. Each strain was spotted onto YPD or YNB alone and onto media also containing 5FC or 5FU at the indicated concentration. Plates were incubated at 30°C until photographed.
Swab assays
Swab assays were conducted as previously described [12]. Briefly, independent colonies were inoculated in liquid YPD media and cultured with shaking until saturation. Sterile cotton swabs were then used to spread culture to a plate containing drug in order to select for resistant colonies. This assay is only semi-quantitative, as the inoculum is not strictly controlled between independent cultures when swabbing.
Acknowledgements
This study was supported by NIH/NIAID R37 MERIT award AI39115-21 and NIH/NIAID R01 AI50113-15 to J.H.; NIH/NIAID R21 AI109623 to T.L.D; and NIH/NIAID F30 AI120339 to L.X.L. This study utilized a Cryptococcus gene deletion collection deposited at the Fungal Genetics Stock Center and made freely available ahead of publication by the Madhani laboratory and funded by NIH R01 AI100272.