Transcriptional Response of Candida auris to the Mrr1 Inducers Methylglyoxal and Benomyl

ABSTRACT Candida auris is an urgent threat to human health due to its rapid spread in health care settings and its repeated development of multidrug resistance. Diseases that increase risk for C. auris infection, such as diabetes, kidney failure, or immunocompromising conditions, are associated with elevated levels of methylglyoxal (MG), a reactive dicarbonyl compound derived from several metabolic processes. In other Candida species, expression of MG reductase enzymes that catabolize and detoxify MG are controlled by Mrr1, a multidrug resistance-associated transcription factor, and MG induces Mrr1 activity. Here, we used transcriptomics and genetic assays to determine that C. auris MRR1a contributes to MG resistance, and that the main Mrr1a targets are an MG reductase and MDR1, which encodes a drug efflux protein. The C. auris Mrr1a regulon is smaller than Mrr1 regulons described in other species. In addition to MG, benomyl (BEN), a known Mrr1 stimulus, induces C. auris Mrr1 activity, and characterization of the MRR1a-dependent and -independent transcriptional responses revealed substantial overlap in genes that were differentially expressed in response to each compound. Additionally, we found that an MRR1 allele specific to one C. auris phylogenetic clade, clade III, encodes a hyperactive Mrr1 variant, and this activity correlated with higher MG resistance. C. auris MRR1a alleles were functional in Candida lusitaniae and were inducible by BEN, but not by MG, suggesting that the two Mrr1 inducers act via different mechanisms. Together, the data presented in this work contribute to the understanding of Mrr1 activity and MG resistance in C. auris. IMPORTANCE Candida auris is a fungal pathogen that has spread since its identification in 2009 and is of concern due to its high incidence of resistance against multiple classes of antifungal drugs. In other Candida species, the transcription factor Mrr1 plays a major role in resistance against azole antifungals and other toxins. More recently, Mrr1 has been recognized to contribute to resistance to methylglyoxal (MG), a toxic metabolic product that is often elevated in different disease states. MG can activate Mrr1 and its induction of Mdr1 which can protect against diverse challenges. The significance of this work lies in showing that MG is also an inducer of Mrr1 in C. auris, and that one of the major pathogenic C. auris lineages has an activating Mrr1 mutation that confers protection against MG.

15 mM MG. At MG concentrations of 10 mM (Fig. 1A) and 15 mM (Fig. S1), the mrr1aD mutant displayed a substantial growth defect relative to the parental isolate B11221 (WT), while the mrr1bD and mrr1cD mutants exhibited growth comparable to WT. None of the mutants (mrr1aD, mrr1bD, or mrr1cD) differed from the parental isolate B11221 (WT) in YPD alone or in the presence of 5 mM MG (Fig. S1). Like C. lusitaniae, the C. auris genome encodes multiple putative MG reductases; the closest orthologs to MGD1 and MGD2 were CJI97_000658 and CJI97_004624, respectively, in the B11221 genome assembly (58) and we will henceforth refer to these genes as MGD1 and MGD2. For reference, MGD1 and MGD2 correspond to B9J08_000656 and B9J08_004828, respectively, in the genome assembly of the C. auris reference strain B8441. By quantitative real-time PCR (qRT-PCR), basal expression of MGD1 was significantly decreased 24-fold in the mrr1aD mutant relative to B11221 WT (Fig. 1B), and expression of MGD2 trended lower in the mrr1aD mutant (;1.2-fold) but this difference did not reach statistical significance (Fig. 1C). MGD1 was also more highly expressed than MGD2 in the WT B11221 as in C. lusitaniae (58). Consistent with the transcriptional patterns, C. auris Mgd1 shares slightly more identity with C. lusitaniae Mgd1 than does C. auris Mgd2 (63% identity versus 61% identity).
In the C. auris B11221 background, expression of MDR1, another target of Mrr1 in other species including C. lusitaniae, also depended on Mrr1a, as the mrr1aD mutant and MDR1 (C) expression in B11221 WT (blue) and mrr1aD (red) cultures grown to exponential phase in YPD at 37°C. Data shown represent the mean 6 SD for three independent experiments. Ratio paired t test was used for statistical evaluation; * P , 0.05. (D) Volcano plot of all quantified genes in B11221 WT versus mrr1aD in the control condition. Each point represents a single gene; blue points indicate genes significantly more highly expressed in WT; red points indicate genes significantly more highly expressed in mrr1aD. Numbers adjacent to each colored point indicate the log 2 FC in mrr1aD versus WT. exhibited a significant 21-fold decrease in MDR1 expression compared to the WT parent (Fig. 1D). These results indicate that in C. auris MDR1 and MGD1 are co-regulated, as has been reported in C. albicans (44,45,(94)(95)(96), C. parapsilosis (97), and C. lusitaniae (37,39,40,58,98), and that higher expression of MGD1 and/or MDR1 contributes to growth in high concentrations of MG (Fig. 1A).
In C. lusitaniae and other Candida species, Mrr1 regulates dozens of genes in addition to MDR1 and MGD1 (37,40). To further elucidate the Mrr1a regulon in C. auris isolate B11221, we performed an RNA-seq analysis of in B11221 WT and its mrr1aD derivative in cells from exponential phase cultures grown at 37°C in YPD. In the control condition (YPD 1 dH 2 O), only four genes, including MDR1 and MGD1, were differentially expressed between the two strains with the cutoff of a log 2 fold change (log 2 FC) $ 1.00 or # -1.00 and a P-value less than 0.05 ( Fig. 1E and Data Set S1 for all data). MGD1 and MDR1 showed a 22-and 24-fold decrease, respectively, in mrr1aD compared with WT, consistent with our qRT-PCR data. CJI97_005632, which was 2.25fold lower in mrr1aD, is orthologous to the C. albicans genes RIM11 and C2_04280W_A, both of which are predicted to encode proteins with serine/threonine kinase activity, though it is worth noting that levels of the transcript were much lower than levels of MDR1 and MGD1. CJI97_000852, which was 2.77-fold higher in mrr1aD than in WT, has 16 orthologs of diverse predicted or known functions in C. albicans, including USO5, USO6, and RBF1 ( Fig. 1E and Data Set S1). Notably, MGD2 was not differentially expressed between B11221 WT and the mrr1aD mutant in our RNA-seq data (Data Set S1), consistent with our qRT-PCR results described above.
Mrr1a regulates only MDR1 and MGD1 in response to MG and benomyl. We have previously shown in C. lusitaniae that MG induces expression of the Mrr1-regulated genes MGD1 and MGD2 in an Mrr1-dependent manner, and MDR1 in a partially Mrr1-dependent manner (58). To determine if MG would induce expression of MGD1, MGD2, and/ or MDR1 in C. auris, we purified RNA for qRT-PCR from exponential-phase cultures of B11221 WT and mrr1aD treated with 5 mM MG or an equal volume of dH 2 O for 15 min. We found that MG treatment significantly enhanced expression of MGD1 in WT by 2.4fold but not in mrr1aD ( Fig. 2A). MGD1 was also induced by a 30-min treatment with 25 mg/mL benomyl (BEN), a known inducer of Mrr1-regulated genes in other Candida species (37,41,43,95,(99)(100)(101)(102)(103)(104), by 7.5-fold in the WT ( Fig. 2A). The different treatment times for MG and BEN were used to be consistent with previous studies using either compound in the related species C. lusitaniae (37,58). Expression of MDR1 was also more highly induced by treatment with either MG or BEN in WT compared with the mrr1aD mutant by 6-and 14.5-fold, respectively (Fig. 2B). Although MDR1 expression was significantly induced by MG and BEN in the mrr1aD, transcript levels of MDR1 were approximately 20-fold higher in the WT than in the mrr1aD under these conditions (Fig. 2B), suggesting that Mrr1a is required for maximum expression of MDR1 in response to stimuli.
To describe the complete Mrr1-dependent MG-and BEN-response regulon under our test conditions in C. auris, we also performed RNA-seq on exponential-phase cultures of B11221 WT and mrr1aD treated with MG or BEN as described above. In B11221 WT, MG led to the upregulation of 319 genes and downregulation of 133 genes compared with the control condition ( Fig. 2C and Data Set S1). In the mrr1aD mutant, MG led to the upregulation of 349 genes and downregulation of 143 genes compared with the control condition ( Fig. S2A and Data Set S1). Consistent with our qRT-PCR data in Fig. 2A, MG induced expression of MGD1 in the WT but not in the mrr1aD mutant (Table S1 and Data Set S1). Although expression of MDR1 was significantly induced by MG in both the WT and the mrr1aD mutant (Table S1 and Data Set S1), levels of MDR1 were substantially lower in the mrr1aD mutant even in the presence of MG ( Fig. 2D and Data Set S1), also in agreement with our qRT-PCR data. MGD1 and MDR1 strongly stood out as the only two genes in the MG response that were strongly dependent on Mrr1a (Fig. 2D).
Treatment with BEN led to upregulation of 160 genes and downregulation of 163 genes in the WT (Fig. 2E and Data Set S1). In the mrr1aD mutant, 181 genes were upregulated, and 229 genes were downregulated in response to BEN ( Fig. S2B and Data Set S1). Like MG, induction of MGD1 by BEN was completely dependent on Mrr1a (Table S1 and Data Set S1) and MGD2 expression was not induced by BEN (Data Set S1). Expression of MDR1 was also induced by BEN in both the WT and the mrr1aD mutant, but as with MG, MDR1 levels in the mrr1aD mutant did not reach that of the WT even with BEN treatment (Fig. 2F and Data Set S1). Again, MGD1 and MDR1, appear to be the only genes in C. auris whose induction of expression by either MG or BEN is dependent on Mrr1a. The Mrr1a-independent responses to MG and BEN are discussed further below.
B11221 has higher basal expression of MDR1 and of putative MG reductase genes compared with the clade I isolate AR0390. Many clade III isolates, including B11221, contain an N647T single nucleotide polymorphism (SNP) in MRR1a (25,105). In Iyer et al., this SNP was proposed to be a gain-of-function mutation due to the resistance of clade III isolates against azoffluxin, a novel antifungal compound that inhibits expression and activity of C. auris efflux pumps (105). As a first step to determine whether there were differences in activity between the Mrr1a protein encoded by the N647T allele found in clade III and the variant encoded by the allele found in clades I, II, and IV, we Data shown represent the mean 6 SD for three independent experiments. Ratio paired t test was used for statistical evaluation; ns P . 0.05; * P , 0.05; ** P , 0.01. (C to D) Volcano plots of all quantified genes in B11221 WT treated with either MG (C) or BEN (D). Each point represents a single gene; magenta points indicate genes that were significantly upregulated compared with the control condition, teal points indicate genes that were significantly downregulated compared to the control condition. MDR1 and MGD1 are shown along with the two most up-and downregulated genes in each condition. (E to F) Scatterplots of the average CPMs of all quantified genes in mrr1aD versus B11221 WT treated with MG (E) or BEN (F). Each point represents a single gene. Points below the dotted line indicate genes that were more highly expressed in the WT, and points above the dotted line indicated genes that were more highly expressed in the mrr1aD mutant. MDR1 and MGD1 are shown with red dots for reference. compared MG sensitivity of B11221 to that of clade I isolate AR0390. Interestingly, AR0390 grew substantially better than B11221 in the YPD control but showed a greater reduction in growth in YPD with 5 mM MG than did B11221 (Fig. S3). At concentrations of 10 mM (Fig. 3A) and 15 mM MG (Fig. S3), AR0390 exhibited a profound growth defect compared with B11221. To determine if differences in MG sensitivity were due to differences in MGD1 expression, we measured basal expression of MGD1 and its co-regulated gene MDR1 in B11221 and AR0390 using qRT-PCR. Both genes were significantly more highly expressed in B11221 by 42-and 4.2-fold, respectively ( Fig. 3B and C).
To gain a deeper understanding of the broader transcriptional differences between B11221 and AR0390, we compared the basal global gene expression in YPD of the two strains using RNA-seq. First, we matched the 5,227 syntenic orthologs between the genomes of B11221 and the clade I reference strain B8441 to compare expression of each gene under the control condition. Of these, 755 genes were differentially expressed FIG 3 MDR1 and MGD1 are among the genes significantly more highly expressed in isolate B11221 compared with isolate AR0390. (A) Growth curves of B11221 (blue) and AR#0390 (orange) in YPD 1 10 mM MG. Data shown represent the mean 6 SD for three independent experiments. (B to C) qRT-PCR assessment of MGD1 (B) and MDR1 (C) expression in B11221 (blue) and AR0390 (orange) grown to exponential phase in YPD at 37°C. Data shown represent the mean 6 SD for three independent experiments. Ratio paired t test was used for statistical evaluation; * P , 0.05; **** P , 0.0001. (D) Volcano plot of all quantified genes, matched by syntenic ortholog, in B11221 and AR0390 in the control condition (YPD). Each point represents a single gene; blue points indicate genes significantly more highly expressed in B11221; orange points indicate genes significantly more highly expressed in AR0390. (E to F) qRT-PCR expression analysis for MGD1 (E) and MDR1 (F) in C. lusitaniae U04 mrr1D (gray) and its derivatives expressing CauMRR1a N647T (dark blue) or CauMRR1a (brown). Data shown represent the mean 6 SD for three independent experiments. One-way ANOVA was used for statistical evaluation; * P , 0.05; ** P , 0.01; *** P , 0.001; **** P , 0.0001. (G) Growth curves of C. lusitaniae U04 mrr1D (gray) and its derivatives expressing CauMRR1a N647T (dark blue) or CauMRR1a (brown) in YPD 1 15 mM MG. One representative experiment of three independent experiments is shown; error bars represent the standard deviation of technical replicates within the experiment. between B11221 and AR0390 in the control condition (jlog 2 FCj $ 1.00, FDR-corrected P , 0.05) (Fig. 3D, Data Set S1). The top 20 differentially expressed genes whose orthologs have known or predicted functions in C. albicans are reported in Table S2. Strikingly, the two genes which exhibited the largest difference in expression between B11221 and AR0390 were MGD2 (log 2 FC = 11.29) and MGD1 (log 2 FC = 8.53) (Fig. 3D, Table S2, and Data Set S1). A third gene with homology to MG reductases, CJI97_001800/B9J08_002257, was also more highly expressed in B11221, although the log 2 FC in expression of this gene in B11221 versus AR0390 was only 1.41 (Data Set S1). Low expression of MGD1, MGD2, and/or B9J08_002257 may contribute to the severe growth defect of AR0390 in the presence of MG. Consistent with our qRT-PCR data, MDR1 was also significantly more highly expressed in B11221 relative to AR0390 (log 2 FC = 4.42) ( Fig. 3D and Table S2). Although MGD2 and B9J08_002257 do not appear to be regulated by Mrr1a in our studies, it is nonetheless interesting to note the elevated expression of three putative MG reductases in the MDR1-overexpressing C. auris isolate B11221, as the co-expression of MDR1 with at least one MG reductase has been reported in numerous studies in other Candida species (37,40,44,45,58,(94)(95)(96)(97).
Clade III Mrr1a N647T exhibits a gain-of-function phenotype compared with clade I Mrr1a when expressed in C. lusitaniae. To compare the activities of the proteins encoded by the MRR1a alleles of B11221 and AR0390 more directly, we heterologously expressed each allele, henceforth referred to as CauMRR1a N647T and CauMRR1a, respectively, independently in a C. lusitaniae mrr1D mutant previously generated and characterized by our lab (37,40,58). All three C. lusitaniae clones expressing CauMRR1a N647T which we tested exhibited a 4-fold increase in FLZ MIC relative to the U04 mrr1D parent (16 mg/mL versus 4 mg/mL), confirming that C. auris clade III MRR1a can complement MRR1-dependent FLZ resistance in C. lusitaniae and adding support to the hypothesis that the N647T substitution in clade III MRR1a confers increased activity. However, the FLZ MIC of the three tested C. lusitaniae clones expressing CauMRR1a did not differ from that of U04 mrr1D (4 mg/mL), so FLZ MIC alone could not indicate whether this allele is functional in C. lusitaniae. One clone expressing each C. auris MRR1a allele was chosen at random for the remaining experiments described in this paper: clone #1 for CauMRR1a N647T and clone #5 for CauMRR1a. Using qRT-PCR, we then examined basal expression levels of C. lusitaniae MGD1 (CLUG_01281) and MDR1 (CLUG_01938/CLUG_01939) in the heterologous complements and the U04 mrr1D parent. Complementation with CauMRR1a N647T conferred a significant increase in basal expression of both MGD1 (Fig. 3E) and MDR1 (Fig. 3F) compared with the mrr1D parent, while complementation with CauMRR1a led to a small, but significant, decrease in expression of both genes relative to mrr1D ( Fig. 3E and F). These results are consistent with our previous observations that C. lusitaniae strains expressing certain Mrr1 variants with low basal activity demonstrate lower expression of some Mrr1-regulated genes, including MDR1 and MGD1, compared with an isogenic mrr1D strain suggesting that Mrr1 has both repressing and activating roles (37,58). Finally, we assessed the relative MG resistance of the isogenic C. lusitaniae strains expressing CauMRR1a N647T or CauMRR1a and the U04 mrr1D parent. The CauMRR1a N647T complement grew markedly better in 15 mM MG compared with U04 mrr1D whereas the CauMRR1a complement grew substantially worse than U04 mrr1D (Fig. 3G), consistent with the pattern of MGD1 expression we observed in these strains via qRT-PCR. None of the C. lusitaniae strains demonstrated growth differences in the YPD control, or in the presence of MG at concentrations of 5 mM or 10 mM (Fig. S4).
MG induces expression of MGD1 and MDR1 in C. auris B11221 and AR0390, but not in C. lusitaniae strains expressing C. auris MRR1a alleles. Next, we compared induction of MGD1 and MDR1 by MG in the C. auris strains B11221 and AR0390 via qRT-PCR. MG significantly induced expression of MGD1 by 2.4-fold in C. auris strain B11221 and by 4.0-fold in C. auris strain AR0390 (Fig. 4A) and expression of MDR1 by 6.0-fold in B11221 and 9.3-fold in AR0390 (Fig. 4B). AR0390 displayed lower expression of both genes in MG, but a higher fold change compared to B11221, further supporting the hypothesis that the N647T allele is gain-of function.
Finally, we compared induction of MGD1 and MDR1 by MG in the isogenic C. lusitaniae strains expressing either CauMRR1a N647T or CauMRR1a and the mrr1D parent. Additionally, we tested induction by BEN in these strains as a control. While the mrr1D parent exhibited a significant 1.8-fold induction of MDR1, neither C. lusitaniae strain expressing a C. auris Mrr1a allele demonstrated a significant change in MGD1 or MDR1 expression in response to MG ( Fig. 4C and D), indicating that C. auris Mrr1a may repress MRR1-independent MG induction of MDR1 in C. lusitaniae and that induction of MGD1 by MG in C. lusitaniae requires a functional MRR1 allele from its own species. Treatment with BEN led to significant increase in expression of MGD1 (Fig. 4E) and MDR1 (Fig. 4F) in all three C. lusitaniae strains. In response to BEN, MGD1 was induced by 1.9-fold in mrr1D, 2.9-fold in the CauMRR1a N647T complement, and 6.1-fold in the CauMRR1a complement (Fig. 4E). Likewise, expression of MDR1 was induced by 2.3-fold in mrr1D, 3.5fold in the CauMRR1a N647T complement, and 5.0-fold in the CauMRR1a complement in response to BEN (Fig. 4F). The striking difference in the ability of the C. lusitaniae strains expressing C. auris MRR1a alleles to respond to BEN versus MG suggests that there are differences in the mechanisms by which BEN and MG induce Mrr1-dependent transcriptional activation and that MG induction of C. auris Mrr1a is not supported by C. and MDR1 (B) in exponential-phase cultures of B11221 (blue) or AR0390 (orange) treated with MG as indicated. Data shown represent the mean 6 SD for three independent experiments. Ratio paired t test was used for statistical evaluation; ns P . 0.05; * P , 0.05; ** P , 0.01. (C to F) qRT-PCR analysis for expression of MGD1 (C, E) and MDR1 (D, F) in exponential-phase cultures of C. lusitaniae U04 mrr1D (gray) and its derivatives expressing CauMRR1a N647T (dark blue) or CauMRR1a (brown) treated with 5 mM MG for 15 min (C, D) or 25 mg/mL BEN for 30 min (E, F). Data shown represent the mean 6 SD for three independent experiments. Ratio paired t test was used for statistical evaluation; ns P . 0.05; * P , 0.05; ** P , 0.01. lusitaniae factors. These potential differences are a topic of future study and may shed light on mechanisms of Mrr1 activation in Candida species.
MG and BEN induced Mrr1a-independent transcriptional responses in C. auris. We have previously observed heterogeneity in MG resistance as well as MG-induced FLZ resistance among several C. auris isolates from different clades (58), and thus we were interested in whether the overall transcriptional response to MG was more similar or different in B11221 and AR0390. AR0390 had greater number of genes differentially expressed by MG compared with B11221; 438 genes were significantly upregulated, and 242 genes were significantly downregulated by MG (see Fig. S5 for the volcano plot of all genes). More genes had a larger fold change in response to MG in AR0390 compared with B11221, including MGD1 and MDR1 (Fig. 5A), consistent with the qRT-PCR results in Fig. 4A and B. However, there was a large overlap of 254 genes which were induced by MG in both strains (Fig. 5B), suggesting a common response across these two genetically distinct clades. These commonly induced genes include many with putative roles in amino acid biosynthesis; transmembrane transport; or acquisition and usage of sulfur ( Fig. 5C and Table S3). The complete comparison is available in Data Set S1.
Only 68 genes with syntenic orthologs across both strains were commonly repressed by MG (Fig. 5B). These genes include some with putative roles in metal transport or carbohydrate uptake and metabolism ( Fig. 5C and Table S3). We did not observe obvious patterns in genes that were only induced or repressed in one strain, and some genes that are listed as only induced or repressed in one strain were close to the cutoff in the other strain.
The groups of genes that were differentially expressed in response to MG in both B11221 and AR0390 were also evident in the response of B11221 to BEN as well as the response of the mrr1aD mutant in response to MG and BEN. In B11221, a total of 46 genes exhibited significant induction by both MG and BEN, including MGD1 and MDR1. Many of the 44 other genes have predicted roles in assimilation and biosynthesis of sulfur-containing compounds or xenobiotic transport ( Fig. 5C and Table S1). MG also induced expression of many genes with predicted roles in the biosynthesis of amino acids. The two genes most highly upregulated upon MG treatment, in terms of fold change, in this strain were orthologous to the arginine biosynthesis genes ARG3 (log 2 FC = 4.77) and ARG1 (log 2 FC = 4.72) (Fig. 2C and Table S1). Conversely, BEN had a limited effect on expression of amino acid biosynthesis genes (Table S1). There were also common themes among the genes that were significantly repressed by both MG and BEN in B11221. Genes that were repressed by both MG and BEN included four orthologs of the HGT glucose transporter family, five genes with a predicted role in uptake of iron and/or copper, and ERG6, which encodes an enzyme in the ergosterol biosynthesis pathway (Fig. 5C and Table S1). The genes that were repressed by only one stimulus, MG or BEN, also included those involved in ergosterol biosynthesis and the uptake of iron, copper, or glucose (Fig. 5C, Table S1). In general, the transcriptional response of the mrr1aD mutant to MG and BEN was similar to that of B11221 WT ( Fig. S2 and Table S1). The complete data sets for MG and BEN responses are available in Data Set S1.

DISCUSSION
In this work, we have demonstrated that in C. auris, the zinc-cluster transcription factor Mrr1a, which is orthologous to Mrr1 in other Candida species, strongly regulates expression of a putative MG reductase MGD1 in addition to MDR1, and that Mrr1a plays a role in MG resistance, highlighting a function of Mrr1 that is distinct from antifungal resistance. We also compared basal global gene expression in B11221 and AR0390 and found that MDR1, MGD1, and MGD2 were among the genes significantly more highly expressed in B11221, consistent with the higher MG resistance of this isolate relative to AR0390. These differences were explained by our finding that MRR1a from B11221 encoded a higher activity variant than that from AR0390 as evidenced by a higher FLZ MIC, higher expression of MDR1 and MGD1, and higher MG resistance in the strain expressing CauMRR1a N647T compared with the isogenic strain expressing CauMRR1a. The allele from B11221 contains an N647T amino acid substitution (25,105) which is in the central region of the regulator where other gain of function substitutions have been found. Both alleles result in induction of MDR1 and MGD1 in response to BEN but not to MG in C. lusitaniae, suggesting that these two compounds activate Mrr1-dependent transcription through different mechanisms.
Under the conditions tested, Mrr1a regulation in the C. auris B11221 background was mainly of MGD1 and MDR1. Homologs of MDR1 and at least one gene encoding a known or predicted MG reductase are co-regulated by Mrr1 in C. albicans (44,45,(94)(95)(96), C. parapsilosis (97), and C. lusitaniae (37,40,58), suggesting that the co-regulation of these two genes has been conserved throughout multiple Candida species. Gaining a deeper understanding of the evolutionary and biochemical relationship between methylglyoxal reductases and efflux pumps, particularly Mdr1, may shed light on how genes with syntenic orthologs between B11221 and AR0390 that were significantly induced (indicated by "up" arrows) or repressed (indicated by "down" arrows) by MG in either or both strains. (B) Scatterplot of the log 2 FC of genes significantly induced by MG in AR0390 versus the log 2 FC of genes induced by MG in B11221. Only genes with syntenic orthologs between the two strains are shown. Each point represents a single gene; points above the dotted line indicate genes which exhibited a greater Log 2 FC in AR0390, and points below the dotted line indicate genes which exhibited a greater log 2 FC in B11221. MGD1 and MDR1 are indicated with red dots for reference. (C) Graphic summary of major groups of genes that were significantly up-or downregulated in response to MG in both B11221 and AR0390. Genes in bold text were also up-or downregulated in response to BEN in B11221.
Candida species sense and respond to environmental or physiological stresses, evade host defense mechanisms, and develop antifungal resistance. In all other Candida species with published Mrr1 regulons, however, Mrr1 appears to regulate expression of many more genes than the four we have described here in the C. auris strain B11221 (37,40,44,45,97). The surprisingly small number of C. auris genes whose expression was significantly altered by genetic deletion of MRR1a may be due to possible redundancy between MRR1a and the other two MRR1 orthologs in C. auris, MRR1b, and MRR1c, although further studies would be necessary to test this hypothesis. It is striking, however, that MRR1a alone seems to be necessary for expression and induction of MGD1, which is further supported by our observation that only the mrr1aD mutant had a growth defect in MG compared with parental B11221 (Fig. 1A).
Our demonstration of increased basal activity of the CauMRR1a N647T allele compared with the allele from AR0390 supports the hypothesis put forth by Iyer et al. (105) that the N647T substitution found in many clade III isolates is a gain-of-function mutation. Furthermore, this may explain why deletion of MRR1a leads to a mild decrease in azole resistance in B11221, but not in the clade IV isolate B11243 (46). In C. albicans, knocking out gain-of-function MRR1 causes a significant decrease in FLZ resistance, but knocking out MRR1 with wild-type transcriptional activity does not alter FLZ resistance (41,44,45,106). Similarly, knocking out gain-of-function MRR1 in C. lusitaniae also decreases FLZ resistance, although knocking out MRR1 alleles that do not encode a constitutively active protein generally leads to increased FLZ resistance (37).
Although Mrr1a does not appear to play a major role in C. auris azole resistance (46), our findings suggest that it contributes to resistance against MG, which may be encountered in the host environment. We have previously shown that Mrr1 also contributes to MG resistance in C. lusitaniae in a manner that is partially dependent on MGD1 and MGD2 (58). Indeed, gain-of-function mutations in MRR1 may arise in various Candida species due to selective pressures other than azoles. In C. lusitaniae, we have reported the emergence of gain-of-function mutations in MRR1 among isolates from a patient with no history of clinical antifungal use (40). In C. auris, most sequenced clade III isolates exhibit both the MRR1a N647T allele and the ERG11 F126L allele (25), the latter of which has been shown to be a major contributor to azole resistance (31). Although it is not known whether the MRR1a or ERG11 mutation occurred first in the clade III lineage, it seems plausible that if the ERG11 mutation did occur first, evolution of the MRR1a N647T allele in C. auris is likely to be the result of selection for MGD1 expression and/or an unknown role for Mdr1 that is unrelated to azole resistance. Therefore, we hypothesize that Mrr1 may act, either directly or indirectly, as a response regulator for carbonyl stress in Candida species, and future studies will investigate a possible role for Mrr1 in resistance against other physiologically relevant reactive carbonyl compounds.
Curiously, although both variants of C. auris Mrr1a were inducible by BEN when expressed in C. lusitaniae, they were not inducible by MG under the conditions tested ( Fig. 4E and F). One possible hypothesis for this observation is that Mrr1 must interact with at least one particular binding partner to induce transcription in response to MG, and that C. auris Mrr1a does not bind efficiently to this C. lusitaniae Mrr1-binding protein or complex. Differential requirements for Mrr1-dependent transcriptional activation by chemical stressors have reported in C. albicans. For example, the transcription factor Mcm1 is required for Mrr1-dependent induction of MDR1 in response to BEN but not to H 2 O 2 (101), and the redox-sensing transcription factor Cap1 is required for MDR1 induction by H 2 O 2 and may play a role in MDR1 induction by BEN (44). Furthermore, gain-of-function Mrr1 in C. albicans requires the Swi/Snf chromatin remodeling complex to maintain promoter occupancy, and the kinase Ssn3, which is a subunit of the Mediator complex, may act in opposition to Mrr1 or its coactivators (38). Thus, although C. auris Mrr1a can complement Mrr1-dependent basal and BEN-induced expression of MDR1 and MGD1 in C. lusitaniae, it may be incompatible with certain elements of the C. lusitaniae MG-responsive transcriptional machinery. Further studies on the differences between C. lusitaniae and C. auris Mrr1, particularly in the presence of MG, may elucidate more detailed mechanisms of Mrr1 activation.
In general, we observed substantial upregulation of genes with predicted roles in transmembrane transport, sulfur metabolism, and amino acid biosynthesis in response to MG in all three strains tested. Many genes downregulated in response to MG in all three strains have predicted roles in metal acquisition, particularly iron, and carbohydrate metabolism. In both B11221 WT and mrr1aD, BEN treatment led to differential expression of similar groups of genes as MG in addition to induction of genes with predicted roles in oxidative stress response. Our studies of the transcriptional response of C. auris to MG and BEN contribute to the understanding of how Candida species may adapt to oxidative and/ or carbonyl stress, two types of stress that a pathogen is likely to encounter in the host environment. In humans, elevated serum MG has been reported in diabetes as well as in renal failure, which are both risk factors for Candida infection (107,108). There is also evidence that neutrophils (109) and macrophages (110,111) generate MG during the inflammatory response, consistent with elevated levels of MG in sepsis patients (77). In our transcriptomics analysis of three C. auris strains exposed to 5 mM MG for 15 min, upregulation of numerous genes involved in amino acid uptake, metabolism, and biosynthesis was one of the most striking responses to MG (Table S1 for comparison of MG and BEN in B11221 WT and mrr1aD and Table S2 for the comparison of genes induced by MG in B11221 and/ or AR0390). In particular, induction of ARG genes is interesting considering the report that C. albicans upregulates expression of arginine biosynthesis genes when phagocytosed by macrophages or in response to sublethal concentrations of hydrogen peroxide, tert-butyl hydroperoxide, or menadione in vitro (112). This induction of ARG genes in C. albicans by macrophages is dependent on the gp91 phox subunit of the macrophage oxidase, and thus is likely a direct response to oxidative stress rather than arginine depletion (112). In our data set, ARG3 and ARG1 exhibited the highest log 2 FC in response to MG in the B11221 background, independently of MRR1a (Table S1). We also observed, in all three C. auris strains, induction of several MET genes, which are involved in methionine synthesis and are an important branch of sulfur assimilation in yeast. Other genes involved in sulfur acquisition and assimilation that were induced by MG include the sulfate importer SUL2, a gene orthologous to both CYS3 and STR3 of S. cerevisiae, and numerous genes associated with iron-sulfur cluster formation (Table S1). A gene orthologous to MUP1 of S. cerevisiae and C. albicans was induced by MG in B11221 WT and AR0390 but fell short of the log 2 FC $ 1.00 cutoff in mrr1aD (Table S1 and Data Set S1). Induction of genes involved in sulfur metabolism, including the MET pathway, SUL2, CYS3, STR3, and MUP1, has previously been observed in Saccharomyces cerevisiae exposed to 1g/L acetaldehyde (113), another reactive aldehyde metabolite that is structurally similar to MG. Thus, sulfur acquisition and metabolism may be an important part of the carbonyl stress response in yeast.
In the B11221 background, we observed modest overlap in the genes and groups of genes that were up-or downregulated in response to either MG or BEN. MDR1 and MGD1 were among the genes induced by both compounds, and induction of MGD1 by either MG or BEN was completely dependent on MRR1a. Although BEN, which originated as an agricultural fungicide, is widely recognized as an inducer of expression of Mrr1-regulated genes in Candida species (37,41,43,95,(99)(100)(101)(102)(103)(104), the mechanism by which this induction occurs is not yet known. BEN is thought to cause oxidative stress in yeast (114,115), which is consistent with our observation of an upregulation of genes with a predicted role in oxidative stress response in BEN-treated C. auris cultures (Table S1). Additionally, in mammalian cells, BEN exposure has been shown to inhibit aldehyde dehydrogenase enzymes (116)(117)(118)(119), which may lead to an accumulation of reactive aldehydes, although this possible mechanism has not yet been investigated in fungi.
We also note similarities between the results of our study of MG-and BEN-treated C. auris and the recently published transcriptional analysis of the clade I C. auris strain NCPF 8973 exposed to 75 mM farnesol (120). In response to farnesol, the authors reported upregulation of many genes with predicted roles in transmembrane transport, such as MDR1 and CDR1, and downregulation of numerous genes predicted to be involved in metal acquisition and homeostasis, including multiple ferric reductases and iron permeases (120). As farnesol may cause oxidative stress in Candida species (120)(121)(122)(123) and in S. cerevisiae (124,125), the overlap in transcriptional changes in response to MG, BEN, and farnesol likely provides valuable insight into how C. auris and other Candida species sense and adapt to physiologically relevant stressors. In fact, MG itself may serve as a stress signal in various organisms. In plants, for example, intracellular MG increases in response to drought (126,127), salinity (126,(128)(129)(130)(131), cold stress (126), heavy metals (128), or phosphorous deficiency (131), and overexpression of certain genes involved in MG detoxification has been shown to enhance salt tolerance in tobacco (126) and in Brassica juncea (132). Investigating whether MG detoxification is linked to abiotic stressors such as salt, temperature, or desiccation in Candida species would be an interesting avenue of future research, particularly in C. auris due to its persistence on hospital surfaces and high salt tolerance.

MATERIALS AND METHODS
Strains, media, and growth conditions. The sources of all strains used in this study are listed in Table S4. All strains were stored long term in a final concentration of 25% glycerol at 280°C and freshly streaked onto YPD agar (10 g/L yeast extract, 20 g/L peptone, 2% glucose, 1.5% agar) once every 7 days and maintained at room temperature. Unless otherwise noted, all overnight cultures were grown in 5 mL YPD liquid medium (10 g/L yeast extract, 20 g/L peptone, 2% glucose) on a rotary wheel at 30°C. Media was supplemented with 25 mg/mL BEN (stock 10 mg/mL in DMSO) or 5 mM, 10 mM, or 15 mM MG (Sigma-Aldrich, 5.55 M) as noted. Escherichia coli strains were grown in LB with 15 mg/mL gentamicin (gent).
Plasmids for complementation of C. auris MRR1a. Plasmids for complementing C. auris MRR1a into C. lusitaniae were created as follows: the open reading frame of MRR1a was amplified from the genomic DNA of C. auris isolates B11221 (for CauMRR1a N647T ) and AR0390 (for CauMRR1a) using a forward primer with homology to the 59 flank of C. lusitaniae MRR1 and a reverse primer with homology to the 39 flank of C. lusitaniae MRR1 for recombination into the C. lusitaniae MRR1 complementation plasmid pMQ30 MRR1-L1191H1Q1197 * (58). Plasmid pMQ30 MRR1-L1191H1Q1197 * was digested with AscI (New England BioLabs) and AgeI-HF (New England BioLabs). The PCR products and digested plasmid were cleaned using the Zymo DNA Clean & Concentrator kit (Zymo Research) and assembled using the S. cerevisiae recombination technique described in (133). Recombined plasmids were isolated from S. cerevisiae using a yeast plasmid miniprep kit (Zymo Research) before transformation into NEB®5-alpha competent E. coli (New England BioLabs). E. coli containing pMQ30-derived plasmids were selected for on LB containing 15 mg/mL gentamicin. Plasmids from E. coli were isolated using a Zyppy Plasmid Miniprep kit (Zymo Research) and subsequently verified by Sanger sequencing with the Dartmouth College Genomics and Molecular Biology Shared Resources Core. MRR1a complementation plasmids containing the correct sequences were linearized with NotI-HF (New England BioLabs), cleaned up with the Zymo DNA Clean & Concentrator kit (Zymo Research) and eluted in molecular biology grade water (Corning) before transformation of 1.5 mg into C. lusitaniae strain U04 mrr1D as described below. All plasmids and primers used and created in this study are listed in Table S4.
Transformation of C. lusitaniae with C. auris MRR1a complementation constructs. Mutants in C. lusitaniae were generated using an expression-free CRISPR-Cas9 method as previously described (37,58,134). In brief, cells suspended in 1M sorbitol were electroporated immediately following the addition of 1.5 mg of C. auris MRR1a complementation plasmid that had been previously linearized with NotI-HF (New England BioLabs) and Cas9 ribonucleoprotein containing crRNA targeting the NAT1 gene. Transformants were selected on YPD agar containing 600 mg/mL hygromycin B (HygB). Successful transformants were identified via PCR of the C. lusitaniae MRR1 locus as previously described (37,58). CRISPR RNAs (crRNAs; IDT) and primers used to validate transformants are listed in Table S4.
MIC assay. MIC assays for FLZ were performed in RPMI 1640 medium (Sigma, containing L-glutamine, 165 mM MOPS, 2% glucose at pH 7) as described in Demers et al. (40) and Biermann et al. (58) using the broth microdilution method. The final concentration of FLZ in each well ranged from 64 mg/ mL to 0.125 mg/mL. Plates were incubated at 35°C and scored for growth at 24 h and 48 h; the results are reported in Table S4. The MIC was defined as the drug concentration that abolished visible growth compared with a drug-free control.
Growth kinetics. Growth kinetic assays were performed as previously described in Biermann et al. (58). In brief, exponential-phase cultures of C. auris or C. lusitaniae were washed and diluted in dH 2 O to an OD 600 of 1; 60 mL of each diluted cell suspension was added to 5 mL fresh YPD. To each well of a clear 96-well flat-bottom plate (Falcon) was added 100 mL of YPD or YPD with MG at twice the desired final concentration and 100 mL of cell inoculum in YPD. Plates were arranged in technical triplicate for each strain and condition and incubated in a Synergy Neo2 Microplate Reader (BioTek, USA) according to the following protocol: heat to 37°C, start kinetic, read OD 600 every 60 min for 36 h, end kinetic. Results were calculated in Microsoft Excel and plotted in GraphPad Prism 9.0.0 (GraphPad Software).
Quantitative real-time PCR. Overnight cultures of C. auris or C. lusitaniae were diluted 1:50 into 5 mL fresh YPD, and grown to for 4 h at 37°C. To each culture was added MG to a final concentration of 5 mM (4.5 mL stock), BEN to a final concentration of 25 mg/mL (12.5 mL stock), or 4.5 mL molecular Candida auris Response to Methylglyoxal and Benomyl mSphere biology grade dH 2 O. Cultures were returned to the roller drum at 37°C for 15 min (MG or dH 2 O) or 30 min (BEN), then centrifuged at 5,000 rpm for 5 min. The differences in time of exposure in the experimental scheme was used to maintain consistency with published experiments in other species, and not because of known differences in kinetics of activity for the two inducers. RNA isolation, gDNA removal, cDNA synthesis, and quantitative real-time PCR were performed as previously described (40). Transcripts were normalized to C. auris or C. lusitaniae ACT1 expression as appropriate. Results were calculated in Microsoft Excel and plotted in GraphPad Prism 9.0.0 (GraphPad Software). Primers are listed in Table S4. RNA sequencing. Overnight cultures of C. auris were diluted to an OD 600 of 0.1 in 5 mL fresh, prewarmed YPD, and incubated on a roller drum at 37°C for five to six doublings (approximately 6 h). Cultures were diluted once more to an OD 600 of 1 in 5 mL fresh, pre-warmed YPD and returned to the roller drum at 37°C for another five to six doublings. To each culture was added MG to a final concentration of 5 mM (4.5 mL), BEN to a final concentration of 25 mg/mL (12.5 mL), or 4.5 mL molecular biology grade dH 2 O. Cultures were returned to the roller drum at 37°C for 15 min (MG or dH 2 O) or 30 min (BEN), then centrifuged at 5,000 rpm for 5 min. Supernatants were discarded and RNA isolation was performed on cell pellets as described above for qRT-PCR. gDNA was removed from RNA samples as described above. DNA-free RNA samples were sent to the Microbial Genome Sequencing Center (https://www .migscenter.com/) for RNA sequencing.
Analysis of RNA-seq. RNA-seq data were analyzed by the Microbial Genome Sequencing Center (https://www.migscenter.com/) as follows: Quality control and adapter trimming was performed with bcl2fastq (https://support.illumina.com/sequencing/sequencing_software/bcl2fastq-conversion-software.html). Read mapping was performed with HISAT2 (135). Read quantification was performed using Subread's featureCounts (136) functionality. Read counts were loaded into R (https://www.R-project.org/) and normalized using edgeR's (137) Trimmed Mean of M values (TMM) algorithm. Subsequent values were then converted to counts per million (cpm). Differential expression analysis was performed using edgeR's Quasi Linear F-Test. In the supplementary file, the sheet named "All Quantified Genes" contain the results of the exact test for all genes in addition to the normalized counts per million for all samples. Differentially expressed genes were determined using the cutoff of jlog 2 FCj . 1 and P , 0.05.
Generation of Venn diagrams. Venn diagrams of differentially expressed genes across different strains and conditions were computed using the Venn diagram tool from UGent Bioinformatics & Evolutionary Genomics, which is accessible at https://bioinformatics.psb.ugent.be/webtools/Venn/.
Statistical analysis and figure preparation. All graphs were prepared with GraphPad Prism 9.0.0 (GraphPad Software). Ratio paired t-tests and one-way ANOVA tests were performed in Prism; details on each test are described in the corresponding figure legends. All P-values were two-tailed and P , 0.05 were considered significant for all analyses performed and are indicated with asterisks in the text: * p ,0.05; ** p ,0.01; *** p ,0.001; **** p ,0.0001.
Data availability. The data supporting the findings in this study are available within the paper and its supplemental material and are also available from the corresponding author upon request. The raw sequence reads from the RNA-seq analysis have been deposited into NCBI sequence read archive under BioProject PRJNA801628 (https://www.ncbi.nlm.nih.gov/bioproject/PRJNA801628).

SUPPLEMENTAL MATERIAL
Supplemental material is available online only. DATA SET S1, XLSX file, 5.6 MB.

ACKNOWLEDGMENTS
We thank Joachim Morschhäuser and the FDA-CDC Antimicrobial Resistance Isolate Bank for providing strains. We thank Judith Berman for the pGEM-URA3 plasmid used for yeast cloning. We thank Elora Demers for primers.
A This study was supported by grants R01 5R01 AI127548 to D.A.H. Core services were provided by STANTO19R0 to CFF RDP, P30-DK117469 to DartCF, and P20-GM113132 to BioMT. Sequencing services and specialized equipment were provided by the Genomics and Molecular Biology Shared Resource Core at Dartmouth, NCI Cancer Center Support Grant 5P30 CA023108-41. The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH.
The authors have declared that no competing interests exist.