Mutations in TAC1B: a Novel Genetic Determinant of Clinical Fluconazole Resistance in Candida auris.

Candida auris has emerged as a multidrug-resistant pathogen of great clinical concern. Approximately 90% of clinical C. auris isolates are resistant to fluconazole, the most commonly prescribed antifungal agent, and yet it remains unknown what mechanisms underpin this fluconazole resistance. To identify novel mechanisms contributing to fluconazole resistance in C. auris, fluconazole-susceptible C. auris clinical isolate AR0387 was passaged in media supplemented with fluconazole to generate derivative strains which had acquired increased fluconazole resistance in vitro Comparative analyses of comprehensive sterol profiles, [3H]fluconazole uptake, sequencing of C. auris genes homologous to genes known to contribute to fluconazole resistance in other species of Candida, and relative expression levels of C. auris ERG11, CDR1, and MDR1 were performed. All fluconazole-evolved derivative strains were found to have acquired mutations in the zinc-cluster transcription factor-encoding gene TAC1B and to show a corresponding increase in CDR1 expression relative to the parental clinical isolate, AR0387. Mutations in TAC1B were also identified in a set of 304 globally distributed C. auris clinical isolates representing each of the four major clades. Introduction of the most common mutation found among fluconazole-resistant clinical isolates of C. auris into fluconazole-susceptible isolate AR0387 was confirmed to increase fluconazole resistance by 8-fold, and the correction of the same mutation in a fluconazole-resistant isolate, AR0390, decreased fluconazole MIC by 16-fold. Taken together, these data demonstrate that C. auris can rapidly acquire resistance to fluconazole in vitro and that mutations in TAC1B significantly contribute to clinical fluconazole resistance.IMPORTANCE Candida auris is an emerging multidrug-resistant pathogen of global concern, known to be responsible for outbreaks on six continents and to be commonly resistant to antifungals. While the vast majority of clinical C. auris isolates are highly resistant to fluconazole, an essential part of the available antifungal arsenal, very little is known about the mechanisms contributing to resistance. In this work, we show that mutations in the transcription factor TAC1B significantly contribute to clinical fluconazole resistance. These studies demonstrated that mutations in TAC1B can arise rapidly in vitro upon exposure to fluconazole and that a multitude of resistance-associated TAC1B mutations are present among the majority of fluconazole-resistant C. auris isolates from a global collection and appear specific to a subset of lineages or clades. Thus, identification of this novel genetic determinant of resistance significantly adds to the understanding of clinical antifungal resistance in C. auris.

KEYWORDS Candida, triazole, resistance, efflux, CRISPR, WGS, antifungal resistance, drug efflux F irst identified in 2009, Candida auris has rapidly become a health care-associated and multidrug-resistant pathogen of global concern (1, 2). While originally found to be the causative pathogen of virtually simultaneous outbreaks of invasive candidiasis in Asia, South Africa, and South America, C. auris has now been identified in more than 30 countries across 6 continents, including more than 900 confirmed clinical cases of C. auris infections in the United States (3). Further contributing to the clinical significance of this organism are its proclivity to colonize both environmental surfaces and patients, challenges associated with reliable identification in the clinical microbiology laboratory, and the markedly decreased susceptibility to currently available antifungal agents found in a large proportion of C. auris clinical isolates (4,5). While epidemiological data and clinical experience pertaining to the treatment of infections caused by C. auris are currently inadequate to support the establishment of epidemiological cutoff values and true clinical breakpoints, the Centers for Disease Control and Prevention (CDC) has proposed tentative breakpoints to help guide clinicians on the basis of available susceptibility data for C. auris clinical isolates. Applying these tentative breakpoints, approximately 3% of C. auris clinical isolates are resistant to echinocandins, one-third are resistant to amphotericin B, and 90% are resistant to fluconazole (MIC, Ն32mg/liter; modal MIC, Ն256mg/liter) (6). Additionally, one-third of clinical isolates show multidrug resistance, with elevated MIC levels for agents from two or more different classes of antifungals, and clinical isolates resistant to all available agents have been repeatedly reported (7,8).
The extent of fluconazole resistance among C. auris isolates is particularly concerning as this agent remains the most commonly prescribed antifungal, and many of the outbreaks of C. auris have occurred in resource-limited settings (2,(8)(9)(10)(11). While the pervasiveness of fluconazole resistance among C. auris clinical isolates substantially limits therapeutic options of C. auris infections, relatively little is known about the molecular mechanisms underpinning this resistance. One mechanism of fluconazole resistance repeatedly identified in C. auris is mutation of the gene encoding the sterol-demethylase enzyme targeted by the triazoles, ERG11. Three such mutations, encoding the amino acid substitutions VF125AL (commonly referred to as F126L), Y132F, and K143R, are frequently reported among fluconazole-resistant clinical isolates, and associations between these mutations and specific genetic clades of C. auris have been observed (2). Additionally, the mutations encoding the Y132F and K143R substitutions correspond to mutations known to contribute to triazole resistance in other species of Candida such as Candida albicans (12). While the direct impact of these ERG11 mutations has not been delineated in C. auris, heterologous expression of C. auris ERG11 alleles carrying mutations encoding either the Y132F or K143R amino acid substitution on a low-copy-number episomal plasmid was observed to decrease fluconazole susceptibility in a haploid strain of Saccharomyces cerevisiae (13). However, clinical isolates harboring the same ERG11 mutations and exhibiting a fluconazole MIC as low as 1 mg/liter have been described previously, as have fluconazole-resistant isolates of C. auris with no mutation in ERG11, suggesting the presence of yet-to-beidentified mechanisms of fluconazole resistance (8,14).
In addition to mutations in ERG11, increased expression of efflux pump-encoding genes is a common contributor to clinical triazole resistance among multiple species of Candida (15). Most notable of these is C. glabrata, in which nearly all of the clinical triazole resistance is attributable to overexpression of the ATP-binding cassette (ABC)type efflux pump-encoding genes C. glabrata CDR1 (CgCDR1), CgPDH1, and CgSNQ2 (16). The C. auris genome has recently been revealed to include a substantial number of efflux pump-encoding genes of both the ABC and major facilitator superfamily (MFS) classes, and triazole-resistant isolates of C. auris have been observed to exhibit efflux pump activity greatly exceeding (up to 14-fold higher) that of C. glabrata (17)(18)(19).
Furthermore, the increased expression of the C. auris ABC-type efflux pump-encoding gene CDR1 has previously been shown to substantially contribute to clinical triazole resistance (20,21). At present, however, the genetic determinants underpinning the increased expression of efflux pump-encoding genes in C. auris remain unidentified.
In this work, we took an unbiased approach utilizing in vitro evolution to create a collection of isogenic C. auris strains with increased fluconazole resistance, exhibiting an 8-to-64-fold increase in fluconazole MIC. Characterization of these strains and analysis of whole-genome sequencing data for over 300 globally distributed C. auris isolates implicated TAC1B (B9J08_004820), a close homolog of the well-characterized C. albicans transcriptional regulator CaTAC1, as a novel genetic determinant of clinical fluconazole resistance. Having identified TAC1B mutations to be present among a large proportion of fluconazole-resistant clinical isolates, we utilized a Cas9-mediated transformation system both to introduce the most common TAC1B mutation identified among resistant clinical isolates (encoding A640V) into the fluconazole-susceptible AR0387 and to correct the A640V-encoding mutation in the previously characterized and highly fluconazole-resistant AR0390 clinical isolate to the wild-type (WT) sequence. In both cases, the presence of this prevalent TAC1B mutation was found to be associated with significant increase in fluconazole MIC, demonstrating that mutations in TAC1B represent prevalent and significant genetic determinants of fluconazole resistance among clinical C. auris isolates. medium, RPMI 1640, supplemented with the same concentration of fluconazole for an additional 48 h to identify individual colonies exhibiting increased fluconazole resistance. Two individual colonies were randomly selected for characterization from the plate supplemented with 8 mg/liter of fluconazole (yielding strains FLU-A and FLU-B), and a single colony was randomly selected from the plate supplemented with 32 mg/ liter of fluconazole (yielding strain FLU-C). Two strains (including one strain each from the initial passages), FLU-A and FLU-C, were subsequently subjected to a second passage in 64 and 256 mg/liter of fluconazole supplemented media, respectively, yielding strains FLU-A2 and FLU-C2.
Fluconazole MICs were then determined for the parental AR0387 and each of the five fluconazole-evolved strains by broth microdilution in accordance with Clinical and Laboratory Standards Institute methodology with minor modifications as previously described (20). AR0387 exhibited a fluconazole MIC of 1 mg/liter, while the five fluconazole-evolved strains were found to have MICs ranging from 8 to 64 mg/liter (Fig. 2). Each of the second-generation evolved strains, FLU-A2 and FLU-C2, exhibited a further 2-to-4-fold increase in fluconazole MIC relative to the corresponding firstgeneration strains. Fluconazole MICs for all fluconazole-evolved strains were observed to be stable following storage at -80°C and multiple passages on fluconazole-free media.
Fluconazole-evolved strains exhibit alterations in membrane sterols without accompanying mutations in ERG11 or ERG3. As fluconazole-resistant C. auris clinical isolates are very often found to possess mutations in ERG11, sequencing of the ERG11 allele for each of the fluconazole-evolved strains was performed. Surprisingly, all evolved strains were found to have wild-type ERG11 sequences matching that of the parental AR0387. To assay for other changes to the ergosterol biosynthesis pathway which may have been contributing to fluconazole resistance, each of the fluconazoleevolved strains and the parental AR0387 were subsequently subjected to comprehensive sterol profiling. Briefly, each strain was grown to the exponential-growth phase in RPMI liquid media with or without 16 mg/liter of fluconazole (a concentration approximating the average serum concentration achieved in patients being treated for candidemia) (22,23).
Following growth in RPMI media without fluconazole, all of the fluconazole-evolved strains and the parental AR0387 were observed to have largely similar sterol profiles (Fig. 3). In all samples, ergosterol comprised more than 75% of total cellular sterols, with ergosta-5,7,22,24(28)-tetraenol and zymosterol observed to be the next most abundant sterols. AR0387 and four of the fluconazole-evolved strains (FLU-A, FLU-B, FLU-C, and FLU-A2) were also observed to have a small amount (2% to 4%) of lanosterol present, whereas this sterol was absent in the FLU-C2 strain.
Following growth in RPMI media supplemented with fluconazole, the sterol profiles of each of the fluconazole-evolved strains were dramatically different from that of h. Percent growth was determined relative to respective untreated controls as assessed by absorbance at OD 600 . Growth inhibition of 50% relative to untreated control is shown as a dotted horizontal line. The fluconazole MIC for each isolate or strain is shown at the right. Error bars for each data point represent the standard deviations of results from three independent measurements of technical replicates. AR0387 (Fig. 4). While ergosterol was still the predominant sterol among all five fluconazole-evolved strains, lanosterol (46.0% Ϯ 7.4%) and 14-methyl-fecosterol (21.3% Ϯ 4.8%) were observed to be the two most prevalent sterols in AR0387. In Candida albicans, 14-methyl-fecosterol is a known substrate of the sterol-desaturase enzyme encoded by CaERG3, which catalyzes the conversion of 14-methyl-fecosterol to the toxic sterol associated with the antifungal activity of the triazoles, 14-methyl-ergosta-8,24(28)-dienol-3,6-diol (Fig. 5). Additionally, 14-methyl-ergosta-8,24(28)-dienol-3,6-diol comprised 3.8% Ϯ 1.9% all sterols present in AR0387, while this sterol was absent in the sterol profiles of all fluconazole-evolved strains. As mutations in the sterol-desaturaseencoding gene ERG3 have been observed to contribute to fluconazole resistance in other species of Candida and notable differences in the amounts of cellular 14-methylfecosterol and 14-methyl-ergosta-8,24(28)-dienol-3,6-diol were observed between AR0387 and the fluconazole-evolved strains, sequencing of the C. auris gene (B9J08_003737) with the highest degree of homology to C. albicans CaERG3 was performed. However, no mutation in C. auris ERG3 was observed in any of the fluconazole-evolved strains.
Fluconazole-evolved strains exhibit significantly reduced fluconazole uptake. As triazoles (including fluconazole) have previously been shown to enter the cells of C. albicans via facilitated diffusion, deficient drug importation was next examined as a potential mechanism contributing to the increased fluconazole resistance among the fluconazole-evolved strains (24). As previously described, the accumulation of [ 3 H]fluconazole was assessed for AR0387 and each fluconazole-evolved strain, as well as for a previously characterized strain of AR0387 where the CDR1 gene had been deleted (AR0387_Δcdr1), following 2 h glucose starvation in YNB media without carbon source supplementation (20). [ 3 H]fluconazole accumulation was observed to be reduced by

FIG 4
Sterol profiles of C. auris fluconazole-evolved strains grown in RPMI media supplemented with fluconazole reveal a lack of 14-methyl-ergosta-8,24(28)-dienol-3,6-diol. The major constituent sterols for AR0387 and fluconazole-evolved strains at the exponential-growth phase in RPMI media are shown as a proportion of total cellular sterols. Error bars for each data point represent the standard deviations of results from three independent measurements of technical replicates.
Candida auris TAC1B Contributes to Triazole Resistance ® approximately 50% in four of the fluconazole-evolved strains (FLU-A, FLU-B, FLU-C, and FLU-A2) relative to than that observed in AR0387, while accumulation in FLU-C2 did not significantly differ from that in AR0387 (5,438 versus 6,560 cpm, respectively; P ϭ 0.2842) (Fig. 6a). Importantly, there was no difference in [ 3 H]fluconazole accumulation between AR0387 and AR0387_Δcdr1 (6,560 and 6,813 cpm, respectively; P ϭ 0.9976) (Fig. 6b), confirming that the conditions used in this study for glucose starvation were adequate to remove the activity of this known C. auris resistance effector.
Mutations in TAC1B are associated with significantly increased expression of CDR1. Gain-of-function (GOF) mutations in zinc-cluster transcription factor genes, such as C. albicans genes CaUPC2, CaMRR1, and CaTAC1, represent a well-characterized mechanism of fluconazole resistance among other species of Candida (15). To determine if similar mutations might have been contributing to the fluconazole resistance among the fluconazole-evolved strains in these studies, the C. auris genes with the highest degree of homology to C. albicans transcriptional regulatory genes CaUPC2, CaMRR1, and CaTAC1, here named UPC2 (B9J08_000270), MRR1 (B9J08_004061), TAC1A (B9J08_004819), and TAC1B (B9J08_004820), were identified by BLAST and gene orthology analysis and sequencing was performed. As two C. auris genes possessing very high degrees of homology with CaTAC1 were identified (TAC1A and TAC1B share 25.6% predicted peptide sequence identity with each other and share 30.6 and 26.9% identity with CaTAC1, respectively), both were included in this study. While no mutations were identified in TAC1A or MRR1, all five fluconazole-evolved strains were found to have mutations encoding amino acid substitutions in TAC1B (Table 1). Both the FLU-A strain and the corresponding second-generation derivative FLU-A2 were found to harbor a mutation encoding the amino acid substitution R495G, while the FLU-B, FLU-C, and FLU-C2 strains were found to possess a mutation encoding the amino acid substitution F214S. Neither of these mutations corresponds to previously characterized GOF muta-  tions in CaTAC1 or to orthologous genes from other species of Candida. However, these mutations are predicted to alter residues near or within the conserved fungal transcription factor middle homology region (MHR) of Tac1Bp, and multiple mutations encoding amino acid substitutions in the MHR of CaTAC1 have previously been reported to be associated with fluconazole resistance (25). Additionally, a sole mutation in UPC2 encoding the amino acid substitution M365I was identified in FLU-C2, and this mutation similarly alters a residue predicted to reside within the MHR of Upc2p.
In an effort to ascertain if the identified mutations in TAC1B and UPC2 may be associated with altered expression of potential resistance effectors among the fluconazole-evolved strains, the relative expression levels of ERG11, CDR1, and MDR1 were evaluated by reverse transcription-quantitative PCR (RT-qPCR). To accomplish this, AR0387 and each of the fluconazole-evolved strains were grown to the exponentialgrowth phase in RPMI media, and RNA was extracted as previously described. The expression of each gene of interest relative to AR0387 was assessed using the ΔΔC T (threshold cycle) method and the C. auris ACT1 housekeeping gene (B9J08_000486) (20). The level of expression of CDR1 was found to be significantly (3-to-5-fold) higher in all five fluconazole-evolved strains than in AR0387 (P Ͻ 0.0001 for all evolved strains compared to AR0387) (Fig. 7). This level of CDR1 expression is similar to that previously described among extensively fluconazole-resistant C. auris clinical isolates (20). Additionally, subtle variations in the levels of expression of the ERG11 and MDR1 genes, not and FLU-C than in AR0387 (P ϭ 0.0021, P Ͻ 0.0001, P Ͻ 0.0001, and P Ͻ 0.0001, respectively), while accumulation in FLU-C2 did not significantly differ from that seen in AR0387 (P ϭ 0.2842) and no difference in accumulation between AR0387 and AR0387_Δcdr1 was seen (P ϭ 0.9976|).
exceeding 2.1-fold higher than the levels seen with AR0387, were also observed among individual fluconazole-evolved strains. As copy number variations (CNVs) among genes encoding fluconazole resistance effectors, such as ERG11, have previously been reported among clinical isolates and laboratory strains of C. auris, qPCR amplifications from genomic DNA were performed to assay for CNV among the effectors ERG11, CDR1, and MDR1, as well as TAC1B, for each of the fluconazole-evolved strains. (26,27). For each gene of interest, three primer sets spanning the open reading frame were utilized. While no alteration in the copy number of ERG11, CDR1, or MDR1 was observed, the second-generation fluconazole-evolved FLU-A2 strain was found to show a 2-fold increase in the copy number of TAC1B, which was not evident in other evolved strains (see Fig. S1 in the supplemental material).
TAC1B mutations identified during in vitro evolution studies are also present among fluconazole-resistant C. auris clinical isolates. Interrogation of a data set consisting of whole-genome sequencing data for 304 globally distributed C. auris isolates representing each of the four major clades revealed 14 nonsynonymous TAC1B mutations and one deletion, excluding sites which are fixed in all isolates within a clade and which are present in both sensitive and resistant isolates ( Fig. 8; see also Table S1 in the supplemental material) (27). In total, mutations in TAC1B were identified among 165 (54%) isolates. Additionally, 50% (148) of the isolates in this collection with available susceptibility data were found to be fluconazole resistant (MIC of Ն32mg/ liter) and to possess a mutation in TAC1B. Furthermore, the two TAC1B mutations that arose during in vitro drug selection were found to be present among fluconazoleresistant clinical C. auris isolates, suggesting a possible role in clinical fluconazole resistance. R495G was found in a single clade I isolate, and the F214S change was found FIG 7 Significantly increased relative expression of C. auris CDR1 among fluconazole-evolved strains grown to the exponential-growth phase in RPMI media. The levels of expression of C. auris ERG11, CDR1, and MDR1 in AR0387 and the fluconazole-evolved strains were determined following culturing to the exponential-growth phase at 30°C in RPMI media. The expression level for each sample is shown relative to that of the respective gene in AR0387. Arrows between graphs indicate the lineage of each fluconazole-evolved strain from the parental AR0387. Error bars for each data point represent standard errors of the means of results from three biological replicates each performed with three technical replicates. The level of expression of C. auris CDR1 among all fluconazole-evolved strains was observed to be significantly higher (3.3 to 5.4-fold) than that of AR0387 (P Ͻ 0.0001 for each individual comparison). Differences in the levels of expression of CDR1 between the sequential evolved strains were not found to be significant (FLU-A versus FLU-A2, P ϭ 0.7186; FLU-C versus FLU-C2, P ϭ 0.3855).

Rybak et al.
® in 2 isolates from clade II and in 1 isolate from clade IV ( Fig. 8; see also Table S1). Notably, a mutation encoding the A640V amino acid substitution was found to be the most common among clinical isolates, found in 57 clade I isolates from 7 countries and always present with the ERG11 mutation encoding the K143R amino acid substitution. Nearly all (98.2%) of the isolates with A640V and K143R mutations displayed high-level fluconazole resistance (Ͼ64 mg/liter). Other common TAC1B mutations found included A657V in 15 clade I isolates and the frameshift mutation F862_N866del in 46 clade IV isolates. These mutations appeared in isolates with the ERG11 Y132F variant, and these isolates were found to have markedly high MIC values (Fig. 8), suggesting that these mutations may provide additive fluconazole resistance effects. Comparison of Tac1B protein sequences indicated that C. auris A657V corresponds to the CaTac1 GOF mutation A736V associated with increased triazole resistance in C. albicans. Additionally, we observed three novel TAC1B mutations in clade IV isolates lacking resistanceassociated mutations in ERG11, including K247E (n ϭ 5), M653V (n ϭ 7), and A651T (n ϭ 16), six resistant isolates from clade I which harbored two TAC1B mutations (A15T and S195C), and two different mutations affecting the P595 site (P595L in clade I and P595H in clade IV). Green circles indicate isolates harboring non-clade-specific nonsynonymous mutations or gain-offunction mutations in TAC1B (B9J08_004820), with filled circles corresponding to percent alternative allele of Ն0.8, while open green circles correspond to percent alternative allele of 0.67 to 0.79. The specific mutation is indicated for each isolate(s). Mutations in bold/dark blue arose in in vitro evolution experiments or were functionally tested in this study and found to be associated with increased resistance to fluconazole in C. auris. (b) Mutations and locations in TAC1B protein sequence associated with azole resistance are indicated using triangles. Mutations indicated with bold/dark blue (red triangles) arose in in vitro evolution experiments or were functionally tested in this study and associated with increased resistance to fluconazole in C. auris. The size of the triangle indicates the number of isolates from this study harboring the mutation (range, 1 to 57 isolates).
Mutations in TAC1B contribute to fluconazole resistance. As mutations in TAC1B were identified in a large proportion of fluconazole-resistant C. auris clinical isolates and as the mutation encoding the amino acid substitution A640V was found to be the most prevalent among the members of this large collection of clinical isolates, the direct impact of this mutation on fluconazole susceptibility was next determined using a Cas9-mediated transformation system. To accomplish this, the TAC1B allele from the previously characterized fluconazole-resistant C. auris AR0390 clinical isolate (also known as B11205, an isolate from clade I), which contains the mutation encoding the amino acid substitution A640V, was introduced into the fluconazole-susceptible AR0387 clinical isolate by the use of Cas9 ribonucleoproteins (Cas9-RNP) and the SAT-FLP system as previously described (20). Two independent positive-testing transformants were obtained, and the fluconazole MICs were determined by broth microdilution. Introduction of the TAC1B A640V allele into the native TAC1B locus was observed to increase the fluconazole MIC 8-fold relative to the parental AR0387 (Fig. 9). Conversely, when the same methods were used to introduce the wild-type TAC1B allele to isolate AR0390 (which harbors the TAC1B mutation encoding A640V), a 16-fold decrease in fluconazole MIC was observed (Fig. 9). The fluconazole MICs did not differ between independent transformants.

DISCUSSION
C. auris has rapidly become a fungal pathogen of global concern. Among the characteristics most notably distinguishing this organism from other species of Candida, the prevalence of fluconazole resistance is of clear clinical concern as fluconazole remains the most commonly prescribed antifungal worldwide. While mutations in the ERG11 gene are strongly associated with clinical fluconazole resistance in C. auris, the presence of this mechanism alone poorly explains the entirety of resistance observed clinically, and the roles of other genetic and molecular mechanisms contributing to fluconazole resistance in this organism remain largely unknown.
To date, the predominance of knowledge of the molecular mechanisms of fluconazole resistance among species of Candida comes from experience studying C. albicans. In this organism, the most commonly reported mechanisms of fluconazole resistance include mutations in the gene encoding the target of the triazole antifungals, CaERG11, and overexpression of either CaERG11 or genes encoding multidrug-efflux pumps such as CaCDR1, CaCDR2, and CaMDR1 (12,15,28). In fact, among a collection of 63 unrelated fluconazole-resistant C. albicans clinical isolates, all were found to exhibit one of these resistance mechanisms and the vast majority possessed a combination of mechanisms. Specifically, 87% were found to have missense mutations in CaERG11, 75% exhibited elevated expression of CaERG11, 77% exhibited elevated expression of CaCDR1 and CaCDR2, and 21% exhibited elevated expression of CaMDR1 to levels known to contribute to fluconazole-resistance (12, 28). The increased expression of C.  albicans fluconazole resistance effectors has been extensively studied and in the majority of isolates is directly attributable to GOF mutations in zinc-cluster transcription factor genes such as CaUPC2, CaTAC1, and CaMRR1 (contributing to the increased expression of CaERG11, both CaCDR1 and CaCDR2, and CaMDR1, respectively) (15,(28)(29)(30). Other C. albicans mechanisms of fluconazole resistance, such as loss-of-function (LOF) mutations in CaERG3, have also been reported but are much less commonly identified among clinical isolates (31).
In this work, we utilized in vitro evolution to create a collection of isogenic C. auris strains with elevated fluconazole MICs in an effort to identify novel genetic determinants of fluconazole resistance. Strains with significant (8-fold to 32-fold) increases in fluconazole MICs were obtained after a total of only 96 h of growth in fluconazolesupplemented media, and this resistance was observed to be stable after subsequent culturing on fluconazole-free media. This rapid emergence of increased fluconazole resistance is particularly concerning, considering that patients being treated for infections caused by C. auris have acquired antifungal-resistant infections while receiving therapy and that the fluconazole-resistant strains in these experiments were obtained after only a single passage in media supplemented with fluconazole at clinically relevant concentrations (8 to 32 mg/liter) (23). The rapidity of this emergence of fluconazole resistance may be influenced by the haploid nature of the C. auris genome, where acquisition of even a recessive mutation may make a larger immediate contribution to resistance than has been found to occur with other diploid species of Candida such as C. albicans, where a loss-of-heterozygosity event may be required to observe the maximal impact of a single resistance mutation (32). Furthermore, a notable proportion of C. auris clinical isolates have been observed to exhibit increased copy numbers of resistance-associated genes, as was identified in this study with the increased TAC1B copy numbers seen in the second-generation fluconazole-evolved FLU-A2 strain (which concomitantly harbored a R495G-encoding mutation in TAC1B) (27).
Surprisingly, no mutations in C. auris ERG11 were identified among the fluconazoleevolved strains created in this study, and all five strains were found to harbor one of two mutations in TAC1B (encoding F214S and R495G), a gene with a high degree of homology to the well-characterized C. albicans transcriptional regulator gene CaTAC1. Intriguingly, two of the three first-generation evolved strains, FLU-B and FLU-C, were found to have acquired the same TAC1B mutation (F214S) even after selection following exposure to different concentrations of fluconazole (8 and 32 mg/liter, respectively). Furthermore, RT-qPCR revealed that each of the five TAC1B mutant strains exhibited elevated expression of the ABC-type efflux pump-encoding gene CDR1, as has been observed with clinical C. albicans isolates possessing a GOF mutation in CaTAC1 (29). Importantly, the degree of CDR1 overexpression observed among fluconazole-evolved strains was similar to that previously reported among fluconazole-resistant clinical C. auris isolates (20).
The clinical relevance of mutations in C. auris TAC1B was further corroborated upon large-scale analysis of whole-genome sequencing data for over 300 C. auris clinical isolates. Among the members of this global collection, the majority (54%) of the isolates were found to have mutations in TAC1B, and 90% of all isolates with TAC1B mutations were resistant to fluconazole. A total of 14 nonsynonymous TAC1B mutations and one deletion were identified, and this included both of the mutations identified in fluconazole-evolved strains, as well as a single mutation (encoding A640V) which was found among 57 clinical isolates of C. auris from clade I. Subsequently, Cas9-RNPmediated genetic manipulations demonstrated that the mutation in TAC1B encoding the A640V amino acid substitution, the most common mutation found among fluconazole-resistant clinical isolates of C. auris, was sufficient alone to elevate fluconazole resistance by 8-fold. Thus, these data definitively demonstrate that mutations in C. auris TAC1B represent a novel genetic determinant of clinical fluconazole resistance.
Additional characterization of the fluconazole-evolved C. auris strains created in this study was also performed to identify potential changes in cellular sterol composition and fluconazole uptake which might also be associated with increased fluconazole resistance in C. auris. Previously, analysis of sterol profiles had revealed significant changes in cellular sterol composition among fluconazole-resistant clinical isolates of Candida possessing mutations in ergosterol biosynthesis genes. In one example, a clinical C. albicans isolate with pan-triazole resistance and harboring mutations in both CaERG11 and CaERG5 was found to produce no detectable ergosterol while instead accumulating a large amount of ergosta-5,7-dienol (82% of total sterols) (33). Moresubtle sterol profile changes were also reported in a multidrug-resistant clinical isolate of C. parapsilosis possessing a loss-of-function mutation in CpERG3 (encoding G111R). This isolate was found to accumulate large amounts of ergosta-7,22-dienol (72% of total sterols), a substrate of CpERG3 not typically present in such large proportions (22).
Thus, to assay for changes in ergosterol biosynthesis which might be associated with the fluconazole resistance phenotype, comprehensive sterol profiling of the fluconazole-evolved C. auris strains was undertaken. Following growth in RPMI media without fluconazole supplementation, the only notable difference in levels of sterols among the fluconazole-evolved strains and AR0387 was a lack of detectable lanosterol, the substrate of ERG11, in the second-generation evolved FLU-C2 strain. It is tempting to speculate that this lack of lanosterol may be associated with the mutation in UPC2 which is unique to this fluconazole-evolved strain, as GOF mutations in CaUPC2 have been shown to increase expression of ergosterol biosynthesis genes, including CaERG11 (28). However, further interrogation of the C. auris UPC2 regulon and this potential resistance-associated mutation is clearly needed to definitively associate this phenotype with the observed mutation in C. auris UPC2.
Following growth in RPMI media supplemented with 16 mg/liter (a concentration representative of the plasma concentration of fluconazole in patients being treated for candidemia), a stark difference in the sterol profiles of AR0387 and all fluconazoleevolved strains was observed. Whereas ergosterol remained the principal sterol present in all fluconazole-evolved strains (comprising more than 50% of sterols), lanosterol predominated the total cellular sterols in the parental AR0387 isolate (46%). Additionally, 14-methyl-ergosta-8,24(28)-diene-3,6-diol, a sterol proposed to be toxic and important for the antifungal activity of the triazoles, was uniquely found in AR0387 (4% of total sterols) in conjunction with a large amount of 14-methyl-fecosterol (21%), the direct precursor of this speculated toxic sterol (34). However, this change in the sterol profile of AR0387 under conditions of fluconazole treatment is consistent with previously reported changes in fungal sterols with suprainhibitory concentrations of fluconazole (33,35). Therefore, these changes in the sterol profile of AR0387 may be related to the lower fluconazole MIC for this clinical isolate (1 mg/liter) whereas the maintained ergosterol content observed in fluconazole-evolved strains may be a direct consequence of the identified mutations in C. auris TAC1B or a result of the decreased fluconazole sensitivity of these fluconazole-evolved strains. Further investigation of the TAC1B regulon and the impact of TAC1B mutations on ergosterol biosynthesis and sterol profiles is required.
While the mechanism of triazole uptake in fungi has remained unknown to date and decreased drug uptake is not a mechanism which has been demonstrated to contribute to fluconazole resistance among clinical isolates of Candida, it is has been suggested that clinical isolates of Candida with reduced fluconazole sensitivity may exhibit altered fluconazole uptake (24). To interrogate the potential role of altered fluconazole uptake in the resistance observed among the fluconazole-evolved strains created in this study, analysis of [ 3 H]fluconazole uptake was performed. Among the fluconazole-evolved strains, this analysis revealed an intriguing decrease in [ 3 H]fluconazole uptake in 4 of the 5 strains compared to AR0387. All strains except for FLU-C2 were found to accumulate approximately 50% less [ 3 H]fluconazole under energy-depleted conditions. While the reason for the higher degree of [ 3 H]fluconazole uptake in the FLU-C2 strain remains unknown, again it is tempting to speculate whether the mutation in UPC2 which is unique to this strain may be involved, particularly as it contrasts with the diminished uptake in the related first-generation evolved FLU-C strain. Further inves-tigation of the role of altered fluconazole uptake in resistance among clinical isolates of C. auris and the impact of mutations in either TAC1B or UPC2 on fluconazole uptake is clearly merited.
While the findings from the experiments described here are intriguing and of clear clinical significance, this study also had limitations. The in vitro evolution experiments described were performed in a single genetic background and using only fluconazole. Thus, it cannot be determined if different mechanisms of triazole resistance would be observed if similar studies were conducted with different clinical isolates of C. auris or different agents of the triazole class of antifungals. Additionally, while similarities between the fluconazole-evolved strains in this work and clinical isolates of Candida possessing known GOF mutations in close homologs of TAC1B are apparent, such as the increased expression of CDR1, further characterization of the C. auris TAC1B regulon is needed to fully understand the potential similarities and differences between the mutations in TAC1B and known mechanisms of fluconazole resistance in other species of Candida. That withstanding, taken together, the findings of these studies serve to demonstrate that mutations in TAC1B both represent a potent genetic determinant contributing to clinical fluconazole resistance in C. auris and are prevalent among the members of a large global collection of fluconazole-resistant clinical isolates. Further studies characterizing the interplay between mutations in ERG11 and TAC1B and the delineation of the TAC1B regulon in C. auris are needed.

MATERIALS AND METHODS
Isolate, strains, and growth media used in this study. Clinical isolates AR0387 and AR0390 were made available by the CDC and FDA AR Isolate Bank as part of the C. auris collection of isolates. All constructed strains and clinical isolates were grown in YPD liquid media (1% yeast extract, 2% peptone, and 2% dextrose) at 30°C in a shaking incubator unless otherwise indicated. Frozen stocks of all strains and clinical isolates were prepared with 50% sterile glycerol and were maintained at -80°C.
MIC determination. Fluconazole (Sigma) was prepared in dimethyl sulfoxide (DMSO). As previously described, a modified version of the Clinical and Laboratory Standards Institute document M27 methodology utilizing broth microdilution and RPMI liquid media and reading absorbance at 600 nm on a BioTek Synergy 2 microplate reader (BioTek, Winooski, VT) was used to determine the fluconazole MIC as the lowest concentration at which 50% inhibition of growth was obtained (36). All susceptibility testing was performed in technical triplicate and biological duplicate.
Comprehensive sterol profiling. Fluconazole-evolved strains and the parental clinical isolate were grown to the exponential-growth phase at 30°C in RPMI liquid media with or without fluconazole supplemented at 16 mg/liter. Alcoholic KOH was used to extract nonsaponifiable lipids. A vacuum centrifuge (Heto) was used to dry samples, which were then derivatized by adding 100 l 90% N,O-bis(trimethylsilyl)-trifluoroacetamide-10% tetramethylsilane (TMS) (Sigma) and 200 l anhydrous pyridine (Sigma) while heating at 80°C for 2 h as previously described (22,34). gas chromatography-mass spectroscopy (GC-MS) (with a Thermo 1300 gas chromatography system coupled to a Thermo ISQ mass spectrometer; Thermo Scientific) was used to analyze and identify TMS-derivatized sterols through comparison of the retention times and fragmentation spectra for known standards. Sterol profiles for each sample were determined by analyzing the integrated peak areas from GC-MS data files using Xcalibur software (Thermo Scientific). All sterol analysis was performed in biological triplicate. Error bars for each data point represent the standard deviations of results from three independent measurements of technical replicates. Samples were then filtered and dried on glass fiber filters and then washed with another 5 ml of stop solution, and the filters and cells were then transferred to a 5-ml scintillation vial. A Beckman Coulter scintillation analyzer was then used to quantify the radioactivity of each filter following the addition of 3 ml of scintillation cocktail (Ecoscint XR, National Diagnostics). Experiments were performed with six biological replicates, and all results were normalized to cpm per 1 ϫ 10 8 cells. Statistical comparisons were made using a one-way analysis of variance (ANOVA) followed by a Tukey test, and the P values presented represent adjusted values.
Assessment of copy number variation by qPCR and relative gene expression by reverse transcription-quantitative PCR. For assessment of gene copy number variation, genomic DNA was isolated from each isolate or strain, and qPCR was performed by the use of three independent primer sets spanning the open reading frame of each gene of interest and the housekeeping gene ACT1, using SYBR green per the manufacturer's instructions and as previously described (26). For assessment of relative gene expression levels, C. auris isolates and strains were inoculated into 2 ml of RPMI broth buffered with morpholinepropanesulfonic acid (MOPS) to pH 7.0 and grown overnight at 30°C for initiation. Overnight cultures were then diluted to an optical density at 600 nm (OD 600 ) of 0.1 in 10 ml of RPMI media with or without 16 mg/ml of fluconazole and placed in a 50-ml conical tube. Cultures were then incubated for 10 h and then confirmed to be in the exponential-growth phase under these conditions, after which the cells were collected by centrifugation, with storage of the cell pellets at -80°C until isolation of RNA was performed. Synthesis of cDNA was performed using a RevertAid RT kit (Thermo Scientific) per the manufacturer's instructions. C. auris ACT1, ERG11, CDR1, and MDR1 were then amplified from cDNA using SYBR green, PCR master mix, and previously described parameters (20). All experiments were performed in biological and technical triplicate. The 2 ϪΔΔCT method was used to calculate the relative levels of expression of each gene of interest, and standard errors were determined using ΔC T values as previously described (37,38). Error bars for each data point represent standard errors of the means of results from three biological replicates performed with three technical replicates. Statistical comparisons were made using a one-way ANOVA followed by a Tukey test, and the P values presented represent adjusted values. Primers are listed in Table S2 in the supplemental material.
Antifungal susceptibility testing for global collection of isolates. Fluconazole susceptibility testing was included for 294 of the 304 isolates included in whole-genome analyses. A total of 270 isolates were tested at the CDC as outlined by Clinical and Laboratory Standards Institute guidelines. Briefly, custom prepared microdilution plates (Trek Diagnostics, Oakwood Village, OH, USA) were used for fluconazole. Resistance to fluconazole was set at Ն32 mg/liter. This interpretive breakpoint was defined based on a combination of these breakpoints with those established for other closely related Candida species, epidemiological cutoff values, and the biphasic distribution of MICs between the isolates with and without known mutations for antifungal resistance (https://www.cdc.gov/fungal/candida-auris/c -auris-antifungal.html).
Cas9-ribonucleoprotein-mediated transformations. C. auris Cas9 and electroporation-mediated transformations were performed as previously described (20) with minor modification. The C. auris TAC1B alleles from AR0387 (TAC1B WT ) and AR0390 (TAC1B A640V ) were amplified from genomic DNA and then cloned into plasmid pBSS2 using restriction enzymes SacII and EagI, yielding plasmids pBSS2-TAC1B WT and pBSS2-TAC1B A640V . Repair templates for each allele of interest were then amplified from each plasmid using primers that also introduced approximately 50 bases of homology targeting the TAC1B loci to the 3= end of the repair templates. Primers are listed in Table S2. Electrocompetent C. auris cells were prepared as previously described. Approximately 4 M concentrations of dual Cas9-RNP constructs targeting both the TAC1B allele and the sequence immediately downstream of the open reading frame and 1 g of repair template were mixed with cells prior to electroporation performed according to the C. albicans protocol on a GenePulsar Xcell (Bio-Rad) (24). Cells were then allowed to recover for 4 to 6 h in YPD with incubation in a shaking incubator at 30°C. Transformants were then selected by plating recovered cells on YPD plates supplemented with 400 g/ml of nourseothricin. Integration of the repair template at the targeted loci was then confirmed by PCR for all transformants. The FLP recombinase was then induced by growing positive-testing transformants in YPM (1% yeast extract, 2% peptone, and 2% maltose) to mediate loss of the SAT1-FLP cassette. All final strains that were identified as having lost the SAT1-FLP cassette by replica plating as previously described were then again confirmed by sequencing (18).
Data availability. All data from Illumina sequences analyzed in this project are available in the NCBI SRA under BioProject accession no. PRJNA328792, PRJNA470683, and PRJNA493622. A set of isolates are available from the CDC and FDA Antimicrobial Resistance (AR) Isolate Bank (https://www.cdc.gov/ drugresistance/resistance-bank/index.html).

SUPPLEMENTAL MATERIAL
Supplemental material is available online only. FIG S1, TIF file, 2.6 MB.