Tup1 Paralog CgTUP11 Is a Stronger Repressor of Transcription than CgTUP1 in Candida glabrata.

TUP1 is a well-characterized repressor of transcription in Saccharomyces cerevisiae and Candida albicans and is observed as a single-copy gene. We observe that most species that experienced a whole-genome duplication outside of the Saccharomyces genus have two copies of TUP1 in the Saccharomycotina yeast clade. We focused on Candida glabrata and demonstrated that the uncharacterized TUP1 homolog, C. glabrata TUP11 (CgTUP11), is most like the S. cerevisiae TUP1 (ScTUP1) gene through phenotypic assays and transcriptome sequencing (RNA-seq). Whereas CgTUP1 plays a role in gene repression, it is much less repressive in standard growth media. Through RNA-seq and reverse transcription-quantitative PCR (RT-qPCR), we observed that genes associated with pathogenicity (YPS2, YPS4, and HBN1) are upregulated upon deletion of either paralog, and loss of both paralogs is synergistic. Loss of the corepressor CgCYC8 mimics the loss of both paralogs, but not to the same extent as the Cgtup1Δ Cgtup11Δ mutant for these pathogenesis-related genes. In contrast, genes involved in energy metabolism (CgHXT2, CgADY2, and CgFBP1) exhibit similar behavior (dependence on both paralogs), but deletion of CgCYC8 is very similar to the Cgtup1Δ Cgtup11Δ mutant. Finally, some genes (CgMFG1 and CgRIE1) appear to only be dependent on CgTUP11 and CgCYC8 and not CgTUP1. These data indicate separable and overlapping roles for the two TUP1 paralogs and that other genes may function as the CgCyc8 corepressor. Through a comparison by RNA-seq of Sctup1Δ, it was found that TUP1 homologs regulate similar genes in the two species. This work highlights that studies focused only on Saccharomyces may miss important biological processes because of paralog loss after genome duplication. IMPORTANCE Due to a whole-genome duplication, many yeast species related to C. glabrata have two copies of the well-characterized TUP1 gene, unlike most Saccharomyces species. This work identifies roles for the paralogs in C. glabrata, highlights the importance of the uncharacterized paralog, called TUP11, and suggests that the two paralogs have both overlapping and unique functions. The TUP1 paralogs likely influence pathogenicity based on tup mutants upregulating genes that are associated with pathogenicity.

INTRODUCTION

Candida glabrata and Saccharomyces cerevisiae are closely related species in the Ascomycota phylum with few, but significant, differences in environment and metabolism (1). C. glabrata is an opportunistic pathogen, has differences in drug and stress resistance and adherence relative to S. cerevisiae, and it is the second leading cause of candidiasis in the United States (2–4). The common ancestor of C. glabrata and S. cerevisiae underwent a whole-genome duplication (WGD) event, with both species losing most of these paralogs (often called ohnologs) (5–7). S. cerevisiae appears to have lost many transcription factor duplicates in particular, leading to the hypothesis that S. cerevisiae might have a simpler transcriptional network relative to other related yeast species (8). Preservation of the two paralogs suggests function and raises the following questions: what duplicates are maintained, and do those duplicates have an impact on growth characteristics?

Tup1 has been well characterized as a global transcription repressor in S. cerevisiae. Mutants that were able to take up dTMP were designated tup, for thymidine uptake mutants (9). Additionally, tup1 mutants were identified while screening for genes that regulate the mating-type locus (10). Tup1 belongs to a family of WD repeat repressor proteins (11, 12). On the C-terminal end, there are seven repeats of 43 amino acids with highly conserved residues, which are believed to be essential to Tup1 function (13). In contrast, the N-terminal is not as critical for repression (14). In S. cerevisiae, one unit of Tup1 works with four units of Cyc8 (also known as Ssn6) in the Tup1-Cyc8 repressor system (15).

The mechanism of Tup1-Cyc8 repression in S. cerevisiae is well studied. The complex represses over 150 yeast genes and up to 3% of S. cerevisiae genes, including diverse genes related to glucose metabolism, oxygen availability, and DNA damage (16–18). Tup1-Cyc8 affects transcription broadly through several proposed mechanisms. It is thought that sequence-specific DNA binding proteins recruit the system to promoters, but Cyc8 and Tup1 have different roles in repression. Cyc8 generally interacts more directly with the binding proteins whereas Tup1 facilitates other protein interactions, leading to repression of transcription (16, 19). While evidence suggests that the complex inhibits RNA polymerase II function, it also prevents transcription through epigenetic mechanisms. The Tup1-Cyc8 repressor system interacts with multiple class I histone deacetylases, making DNA less accessible for transcription (20). More specifically, Tup1 interacts with histones H3 and H4 to remodel chromatin (21, 22). Thus, while there are likely multiple mechanisms by with the Tup1-Cyc8 complex represses transcription, the complex is important for repressing important genes to the cell.

When either TUP1 or CYC8 is deleted in S. cerevisiae, mutants exhibit phenotypes related to the inappropriate expression of genes (18). For example, S. cerevisiae cyc8 and tup1 mutants flocculate, are defective in sporulation (13), and exhibit temperature-dependent phenotypes (23). Phenotypes are also noted in the presence of different carbon sources. For example, Sctup1 mutants can efficiently assimilate sorbitol, unlike their wild-type counterparts (24), and Tup1 is implicated in maltose metabolism (25).

C. glabrata has maintained two copies of TUP1 (CAGL0C03608g and CAGL0E00561g), whereas S. cerevisiae has lost one. Studies with TUP1 in C. glabrata have only focused on CAGL0C03608g (26); however, an uncharacterized paralog exists (we name this paralog CAGL0E00561g [CgTUP11]), raising the question of why these two paralogs have been preserved over evolutionary time. A phenotype has not been associated with the CgTUP1 or the CgTUP11 gene in C. glabrata, and the aim of this project was to explore their functions by looking for a phenotype for C. glabrata TUP1 and TUP11 mutants and determining what genes are regulated by the paralogs using transcriptome sequencing (RNA-seq). Additionally, we aimed to uncover which paralog is most similar in function to ScTUP1 and how loss of TUP1 homologs in each species impacts gene expression.

RESULTS

Many yeast species that experienced a whole-genome duplication have retained two copies of TUP1.

Examination of the C. glabrata genome indicated two paralogs related to ScTUP1. The two paralogs are CAGL0C03608g (643 amino acids [aa]), also annotated as CgTUP1, which has 69% identity with ScTUP1 (713 aa), and CAGL0E00561g (836 aa), which has 67% identity with ScTUP1. Of note, both paralogs share significant similarity over regions corresponding to the N and C terminals. Whereas many proteins share some similarity with the WD domain repeats in the C terminus (14), clear Tup1 homologs exhibit at least 50% identity over at least 400 aa of alignment. To determine which yeast species have more than one copy of TUP1 in the genome, we used ScTUP1, CgTUP1, and CgTUP11 to BLASTp search various pre-WGD and post-WGD genomes in the Saccharomycotina clade (Fig. 1). In the pre-WGD species, no species appear to have more than one copy of a TUP1 homolog, and in the post-WGD species, there is a clear division. The Saccharomyces species have one copy, and the other post-WGD species have two copies. While only correlative, these results suggest that there is a selective advantage to having two copies of TUP1 in the post-WGD species.

FIG 1

Characterization of Tup1 homologs in the Saccharomycotina clade. Using a simplified phylogenetic tree (32, 33), where the red circle represents the whole-genome duplication, we performed BLASTp on the genomes of the 16 species identified by (34). With visual inspection of potential homologs, we determined that >50% identity over ∼400 aa was a suitable cutoff to identify TUP1 orthologs. The species names in blue indicate species where there are two copies of TUP1 in the genome. Beside the names of the species are the systematic names of the identified genes.

Deletion of the two C. glabrata TUP1 homologs results in few clear phenotypes, but TUP11 appears to functionally replace ScTUP1.

To begin to understand the advantage of having two copies of TUP1, we deleted both paralogs, singly and in combination, as well as CYC8 in C. glabrata. We have named CAGL0E00561g TUP11 to indicate its relation to CgTUP1 (CAGL0C03608g). Characterization of multiple deletion strains indicated that unlike for Sctup1Δ strains, which rapidly flocculate and precipitate to the bottom of a culture tube, there is no clear flocculation phenotype. However, the growth rate of the double deletion strain is lower than for either single mutant or the wild type (see Fig. S1A in the supplemental material). Thus, we began an extensive screen for potential phenotypes.

Growth of S. cerevisiae and C. glabrata strains in synthetic defined (SD) medium with 2% glucose (A), 0.2% glucose (B), 2% ethanol replacing glucose (C), or 1.5 M potassium chloride (KCl) (D) added to standard medium. Logarithmically growing cells in SD with 2% glucose were washed with water, inoculated under various growth conditions at an optical density at 600 nm (OD600) of 0.01, and grown at 30°C in 96-well plates. OD600 was measured after 24 h for 2% glucose and after 48 h for the remaining conditions. The data presented are the means and standard deviations of six biological replicates. P value was determined by a Student t test comparing each species’ mutants to the wild-type strain, with a single asterisk indicating a P value of <0.05 and a double asterisk indicating a P value of <0.01. Download FIG S1, EPS file, 1.7 MB.

Streaking strains on agar plates, we were unable to determine clear phenotypes for the Cgtup1Δ, Cgtup11Δ, Cgtup1Δ tup11Δ, and Cgcyc8Δ mutants under the following conditions: temperature sensitivity, growth in added salts such as CaCl2, FeCl3, or KCl, growth in altered pH (pH = 2 or pH = 7), and sensitivity to ketoconazole and 2-deoxyglucose. The mutants behaved like the wild type, with only subtle growth defects in some conditions. We then chose a few conditions where there might have been a subtle phenotype on plates, grew the strains in liquid medium, and quantified growth (Fig. S1). We present a few quantified examples to demonstrate the variability in growth assays under different conditions. Of note is the statistically significant difference between the wild type and the Cgtup1Δ tup11Δ mutant, which does have a phenotype similar to that of the Sctup1Δ strain under standard growth conditions (synthetic medium with 2% glucose) and in 0.2% glucose. Additionally, we observed a growth enhancement in the Cgtup1Δ and Cgcyc8Δ mutants relative to the wild type in 2% ethanol; however, all cells grow poorly under this growth condition. Through multiple biological replicates we observed a lot of variability, and so while there was statistical significance in some mutants in some conditions, we do not feel comfortable asserting that there is a strong phenotype. We conclude that there are differences between wild-type and C. glabrata mutants but few phenotypes are easily observable.

To confirm that we were incubating cells under conditions that could uncover a phenotype, we focused on growth in added sorbitol (24). We grew mutants and the wild type under conditions where sorbitol replaced glucose and confirmed that the Sctup1Δ mutant was able to grow much better than the wild type (Fig. 2A). Phenotypes where the mutant grows better are more convincing because the mutants often are somewhat sicker than the wild type under standard growth conditions (Fig. S1A). We then compared the C. glabrata strains and noted that none were able to grow better than the wild type in the same medium; however, the C. glabrata double mutant and the Cgcyc8Δ mutant do have a statistically significant defect in growth relative to single mutants or the wild type. We conclude that deletion of the CgTUP1 paralogs or the CgCYC8 gene does not confer the ability to grow in sorbitol in C. glabrata, unlike in S. cerevisiae. During our screening for phenotypes, we did uncover a phenotype for C. glabrata mutants in the presence of nonfermentable carbon sources. In the presence of 2% ethanol, the Cgtup1Δ mutant and Cgcyc8Δ mutant grew better than the wild type or other mutants, but because growth was highly retarded, we do not feel comfortable concluding that it is a strong phenotype (Fig. S1C). However, when we grew C. glabrata in 1% glycerol/1% ethanol, we observed a robust phenotype when (i) either TUP paralog was deleted, (ii) both were deleted, or (iii) CgCYC8 was deleted (Fig. 2B). We conclude that mutants of both species are capable of having advantages in alternative carbon sources, just different ones.

FIG 2

Growth of S. cerevisiae and C. glabrata strains in medium where sorbitol replaces glucose (A) or 1% glycerol/1% ethanol replaces glucose (B). (A) Cells of each strain were pregrown overnight at 30°C in liquid YEP with 3% glycerol to logarithmic growth phase, then washed, and inoculated into synthetic defined medium (SD) with 2% sorbitol replacing glucose at an optical density at 600 nm (OD600) of 0.05. OD600 was measured after 48 h. We were unable to observe the same phenotypes in YEP plus 2% sorbitol, indicating that medium composition is crucial to the observed benefit of sorbitol to the Sctup1Δ mutant. (B) Logarithmically growing cells in SD with 2% glucose were washed and inoculated into SD with 1% glycerol/1% ethanol replacing glucose at an OD600 of 0.05 and grown. OD600 was measured after 24 h. For both panels A and B, the data presented are the means and standard deviations of three biological replicates. P value was determined by a Student t test comparing each species’ mutants to the wild-type strain, with a single asterisk indicating a P value of <0.05 and a double asterisk indicating a P value of <0.01. Standard growth of the strains was confirmed in SD medium with 2% glucose (data not shown).

We next assessed whether the C. glabrata genes were capable of complementing an Sctup1Δ strain. A clear phenotype observable with Sctup1Δ mutants is the rapid settling of a culture in liquid medium because of increased flocculation. To determine which C. glabrata paralog would complement the flocculation phenotype, we measured spectrophotometrically the loss of absorbance from a culture as the cells settled by measuring the optical density at 600 nm (OD600) every 15 s. We then derived slopes of loss of absorbance to quantify the flocculation phenotype (Fig. 3). We cloned ScTUP1, CgTUP1, and CgTUP11 into a plasmid and transformed each plasmid into an Sctup1Δ strain. We determined that the ScTUP1 plasmid suppressed the flocculation phenotype, as expected, and CgTUP11 largely suppressed the flocculation, while CgTUP1 did not. We confirmed that the cloned genes are functional by looking at complementation in C. glabrata tup mutant strains when grown in medium where 1% glycerol/1% ethanol replaced glucose (Fig. S2). As shown in Fig. 2B, the Cgtup1Δ and Cgtup11Δ strains grew better than the wild type in the presence of 1% glycerol/1% ethanol. Adding back CgTUP1 on a plasmid to a Cgtup1Δ mutant or CgTUP11 on a plasmid to a Cgtup11Δ mutant restored growth to wild-type levels.

FIG 3

CgTUP11 is capable of suppressing the flocculation phenotype of the Sctup1Δ strain. The change in OD600 over time was used to quantify flocculation rates. The Sctup1Δ strain was transformed with either plasmid alone (a URA3+ vector) or plasmids containing ScTUP1, CgTUP1, or CgTUP11 and grown in liquid synthetic defined (SD) growth medium without uracil for ∼20 h at 30°C. Each sample was vortexed for 30 s, then the OD600 was read every 15 s for 1 min, and the slope of the decline was measured. The means and standard deviations of six biological replicates are presented. P value was determined by a Student t test comparing each TUP-containing plasmid to the vector alone strain, with a single asterisk indicating a P value of <0.05 and a double asterisk indicating a P value of <0.01.

CgTUP1 and CgTUP11 complement the Cgtup1Δ and Cgtup11Δ strains, respectively, in medium where 1% glycerol/1% ethanol replaces glucose. C. glabrata tup mutants were transformed with either plasmid alone (a HIS3 vector) or a plasmid containing CgTUP1 (in Cgtup1Δ) or CgTUP11 (in Cgtup11Δ). Logarithmically growing cells in SD without histidine with 2% glucose were washed and inoculated into SD with 1% glycerol/1% ethanol replacing glucose at an OD600 of 0.05 and grown at 30°C. OD600 was measured after 24 h. The data presented are the means and standard deviations of three biological replicates. P value was determined by a Student t test comparing each deletion strain with plasmid to the wild-type strain, with a single asterisk indicating a P value of <0.05 and a double asterisk indicating a P value of <0.01. It should be noted that while the Cgtup1Δ strain with vector does not have an asterisk indicating significance, the P value for this strain was 0.059, which is just above the threshold of 0.05, and this deletion strain was highly significant in Fig. 2B in the main text. Standard growth of the strains was confirmed in SD medium with 2% glucose (data not shown). Download FIG S2, EPS file, 0.8 MB.

CgTUP11 represses more genes than CgTUP1, but the two appear to have an overlap of target genes.

To understand the role that CgTUP1 and CgTUP11 have in transcription, we performed a series of RNA-seq (transcriptome sequencing) experiments with each individual deletion strain and the double-deletion strain, comparing them to the wild type. RNA was isolated from actively dividing cells in standard yeast extract, peptone, and dextrose (YEPD) medium. Comparison of RNA expression of the two Cgtup1Δ and Cgtup11Δ deletion strains to the wild type indicated that CgTUP11 was a more active repressor of gene expression than CgTUP1 (Fig. 4A). For example, there are 19 genes that increase over 20-fold in response to deletion of CgTUP11, and there were no genes that changed that much in response to CgTUP1 deletion. Additionally, the deviation of the mutant expression from that of the wild type is much more apparent in the Cgtup11Δ/wild-type comparison relative to the Cgtup1Δ/wild-type comparison (Fig. 4A).

FIG 4

RNA-seq of Cgtup1Δ, Cgtup11Δ, and Cgtup1Δ Cgtup11Δ strains compared to the wild type. (A) Comparison of Cgtup1Δ RNA expression (RPKM) (unfilled circles) and Cgtup11Δ RNA expression (RPKM) (filled circles) relative to wild-type RPKM. Two biological replicates of the Cgtup1Δ and Cgtup11Δ strains are compared to three biological replicates of wild-type and the mean expression is indicated. Genes that were expressed at less than 2 RPKM were not graphed, as their expression level was considered too low to be accurate. (B) Comparison of Cgtup1Δ Cgtup11Δ RNA expression (RPKM) (three biological replicates) relative to the wild type (three biological replicates). (C) Volcano plot of data from panel B, where the x axis is the log2 change in expression (RPKM) of the Cgtup1Δ Cgtup11Δ strain versus the wild type, and the y axis is the –log10 of the P value as determined by a Student t test.

Examination of the genes that appear to be most upregulated in response to deletion indicated that there is some overlap between the targets of repression of the two paralogs. For example, CgFBP1 and CgHSP30 are both derepressed in each deletion (Table S1). Because the individual mutants were only analyzed in duplicate RNA-seq experiments, we also analyzed the double mutant (Cgtup1Δ Cgtup11Δ) relative to the wild type (both performed in biological triplicate), which allowed us to determine statistical significance (Fig. 4B). Measuring expression of the double mutant relative to that of the wild type demonstrated that the double mutant had a larger change in expression of more genes than each single mutant, and thus, we concluded that the double mutant has more derepression than each single mutant. Plotting the same data as a volcano plot (Fig. 4C) identified 471 genes that that are derepressed in the double mutant 2-fold with a P value of less than 0.001.

RNA-seq data. Download Table S1, XLSX file, 1.5 MB.

Using a more stringent value of 4-fold change and a P value of <0.001, there are 248 genes that were derepressed in the double mutant and 15 genes that were repressed. These data suggest that the TUP1 paralogs are much more important for repression than for gene activation. There is no clear grouping of the genes that are responsive to Tup1 activation, other than CgEPA15 being the most repressed in the double mutant, at 105-fold repression. A gene ontology (GO) analysis of the 248 derepressed genes indicated a slight enrichment for carbohydrate metabolic processes, glycogen metabolic processes, and polysaccharide metabolism, which are consistent with targets of ScTUP1 and may explain the growth advantage on nonfermentable carbon sources. These data in total suggest that CgTUP11 is the most similar in function to ScTUP1 but that CgTUP1 is capable of repressing some of the same genes as well as other genes to a lesser extent. One simple explanation for the phenotypes could be that CgTUP11 is expressed at a higher level. While we cannot eliminate the possibility that there are different amounts of the two proteins, the transcript abundance of both genes was observed at a statistically identical level in the wild-type RNA-seq data (21.6 reads per kilobase per million [RPKM] for CgTUP11 versus 20.7 RPKM for CgTUP1).

RT-qPCR validates RNA-seq and indicates that targets have variability in sensitivity to CgTUP1, CgTUP11, and CgCYC8.

To confirm the targets identified in the RNA-seq data set, we harvested RNA from various strains grown in triplicate and performed reverse transcription-quantitative PCR (RT-qPCR) on candidate genes, which were some of the most derepressed genes in the double mutant in the RNA-seq data set (Fig. 5). We used the gene CgMIC10 for normalization, as it did not change expression in response to loss of TUP1 paralogs and was highly expressed (based on RPKM, it was in the top 5% of expressed genes). Whereas all of these genes change expression in a statistically significant manner in at least three of the mutant strains, some of the genes appear to be primarily regulated by CgTUP11 (such as CgYPS2, CgYPS4, CgMFG1, CgRIE1, and CgHBN1), others are primarily regulated by CgTUP1 (CgHXT2 and CgSOK2), and others appear to be regulated by both (CgFBP1 and CgHSP30). With many genes, there is a significant effect when both TUP homologs are deleted: i.e., the double-deletion strain is even more derepressed than individual mutants. Interestingly, all of these target genes appear to require CgCYC8 for repression, but the strongly CgTUP11-dependent genes (CgYPS2, CgYPS4, CgMFG1, CgRIE1, and CgHBN1) appear to not mirror the Cgtup11Δ Cgtup1Δ double mutant, suggesting that maybe another corepressor is important for full repression. There are two proteins annotated as being related to ScCYC8 in the C. glabrata genome; we deleted CAGL0D01364g, which is 84% identical to ScCYC8 over the region corresponding to ∼400 aa (bit score, 812), but CAGL0M01914g has 21% identity over the same region (bit score, 41). It is possible other corepressors are more important for CgTup11 interactions. Importantly, the RT-qPCR analysis of genes on independently grown cultures from the RNA-seq data set validates that the genes we have identified as targets in RNA-seq experiments are likely genuine. Further mechanistic analysis of how these two Tup1 proteins interact with Cyc8, and potentially other proteins, is needed to dissect the different classes of genes regulated by the TUP1 paralogs.

FIG 5

RT-qPCR of candidate C. glabrata genes in tupΔ and cyc8Δ strains. C. glabrata wild-type and mutant strains were grown in standard YEPD medium to logarithmic growth phase. RNA was harvested and reverse transcribed to cDNA, and quantitative PCR was performed using primers for candidate genes as determined by the RNA-seq data in Fig. 4. The amount of transcript was normalized to CgMIC10, which did not change expression with the loss of TUP genes. The data presented are the means and standard deviations of three biological replicates. P value was determined by a Student t test comparing each deletion strain to the wild type, with a single asterisk indicating a P value of <0.05 and a double asterisk indicating a P value of <0.01.

CgTUP11 complements the Sctup1Δ strain based on RNA-seq and CgTUP1 regulates only half of the same genes as ScTUP1.

To compare the regulation of C. glabrata genes with that of S. cerevisiae genes in response to deletion of TUP1 homologs, we performed RNA-seq with the Sctup1Δ strain, where we added back different versions of TUP1 in a plasmid. First, we examined the Sctup1Δ strain with an empty vector relative to a plasmid containing the wild-type version of ScTUP1 (Fig. 6A), allowing us to identify the targets for repression by ScTup1. To confirm that our Sctup1Δ strain behaved like in previous studies, we compared known ScTup1 targets in the Saccharomyces Genome Database (SGD) with our targets and found that >60% of known targets in SGD also changed expression in a statistically significant manner in our RNA-seq data set. We identified 71 genes that were derepressed 4-fold (P < 0.01) in response to loss of ScTUP1, and the GO annotation indicated that these genes are enriched for fungal cell wall and external encapsulating structure organization and sucrose metabolic processes, which is not surprising given the known phenotypes of the Sctup1Δ strain. We also identified 18 genes that increase expression 4-fold (P < 0.01) in the Sctup1Δ strain, and these genes are weakly enriched for mitochondrial electron transport. We then compared the Sctup1Δ strain with ScTUP1 on a plasmid with the strain containing CgTUP1 on a plasmid (Fig. 6B). In this case, there were still 52 genes derepressed 4-fold (P < 0.01), and 63% of those derepressed genes were represented in the 71 genes identified in the Sctup1Δ alone. These data suggest that CgTUP1 is capable of complementing only some of the defects of the Sctup1Δ strain. However, when we added back CgTUP11 to the Sctup1Δ strain (Fig. 6C), the strain behaved very similarly to the wild type (Sctup1Δ plus ScTUP1), with only 2 of the 71 ScTUP1-dependent genes still being upregulated 4-fold. Specifically examining the genes that have three-letter names associated with them (68 of the 71 genes) that are upregulated in the Sctup1Δ strain, all but ScGAT4 are repressed by the addition of CgTUP11 (Fig. 7). Addition of CgTUP1, however, represses only 53% of the ScTUP1 targets. Overall, the data are consistent with CgTUP11 complementation of the flocculation phenotype in the S. cerevisiae tup1 mutant (Fig. 3), indicating that CgTUP11 can functionally replace ScTUP1, and the RNA-seq data indicate that CgTUP1 can partially substitute for ScTUP1. Interestingly, there are new targets identified in the Sctup1Δ plus CgTUP1 strain that are not repressed, suggesting that CgTUP1 has a different specificity from ScTUP1 in S. cerevisiae (Fig. S3).

FIG 6

RNA-seq of Sctup1Δ with complementing plasmids. (A) Comparison of Sctup1Δ plus vector RNA expression (RPKM) and Sctup1Δ with ScTUP1 RNA expression (RPKM). Three biological replicates of each strain are compared and the mean expression is plotted. Genes that were expressed at less than 2 RPKM in both strains were not graphed. (B) Comparison of Sctup1Δ plus CgTUP1 RNA expression (RPKM) and Sctup1Δ with ScTUP1 RNA expression (RPKM). (C) Comparison of Sctup1Δ plus CgTUP11 RNA expression (RPKM) and Sctup1Δ with ScTUP1 RNA expression (RPKM).

FIG 7

Diagram of genes identified in RNA-seq of Sctup1Δ strain with complementing plasmids. There are 68 three-letter-named genes in S. cerevisiae that are upregulated more than 4-fold (P < 0.01) in the Sctup1Δ strain, indicated by the yellow circle. Thirty-six of those genes are repressed by addition of either CgTUP1 or CgTUP11 (red), and the other 32 are only repressible by CgTUP11 (pink).

Diagram of genes identified in RNA-seq of the Sctup1Δ strain with complementing plasmids. There are 68 three-letter-named genes in S. cerevisiae that are upregulated more than 4-fold (P < 0.01) in the Sctup1Δ strain, indicated by the circle on the left (yellow and orange). Thirty-six of those genes are repressed by addition of either CgTUP1 or CgTUP11 (yellow), and the other 32 are only repressible by CgTUP11 (orange). Genes in the red circle are new targets identified in the Sctup1Δ plus CgTUP1 strain that are not repressed, suggesting that CgTUP1 has a different specificity from ScTUP1 in S. cerevisiae. Finally, only two genes are derepressed in the Sctup1Δ plus CgTUP11 strain, indicating that CgTUP11 functionally replaces ScTUP1; GAT4 is not a target of either CgTUP11 or CgTUP1, and DIT1 is a new target of repression by both CgTUP11 and CgTUP1. Download FIG S3, EPS file, 1.9 MB.

To determine whether in vivo the TUP1 homologs regulate similar genes, we examined the top 100 genes from our RNA-seq data sets that were regulated by ScTUP1 and by the two C. glabrata homologs (Fig. S4). We then assessed whether there was a clear homolog in the other species and determined whether that homolog was also upregulated in a significant manner (P value < 0.05 in a Student t test). It is worth mentioning that for many gene families, including those encoding the yapsins (YPS), adhesins (EPA), and hexose transporters (HXT), it is difficult to discriminate true orthologs, and so we collapsed gene families into one representative and assessed whether they were similarly regulated. We determined that 37% of the S. cerevisiae genes that are upregulated (with a P value <0.05) in the Sctup1Δ strain are also upregulated in the Cgtup1Δ tup11Δ strain. Conversely, 43% of the upregulated Cgtup1Δ tup11Δ genes were also upregulated in the Sctup1Δ strain. We conclude from these analyses that ScTUP1, CgTUP11, and CgTUP1 do target some of the same genes, but likely because of speciation, and divergence of the two paralogs in C. glabrata, there are differences in the specific targets in each species.

While ScTUP1, CgTUP1, and CgTUP11 do target some of the same genes, there are differences in the specific targets. (A) The top 100 genes regulated by ScTUP1, as determined by RNA-seq, were examined in the C. glabrata genome to determine if there was an ortholog and if that ortholog was regulated by CgTUP1 and CgTUP11 in a significant manner (P value < 0.05 in a Student t test). (B) Conversely, the top 100 genes regulated by CgTUP1 and CgTUP11, as determined by RNA-seq of the Cgtup1Δ tup11Δ strain, were examined in the S. cerevisiae genome to determine if there was an ortholog and if that ortholog was regulated by ScTUP1. Download FIG S4, EPS file, 2.0 MB.

DISCUSSION

We have identified a TUP1 paralog (CgTUP11, or CAGL0E00561g) in C. glabrata that is equally as important as, if not more important than, the known CgTUP1 (CAGL0C03608g) gene in standard growth medium. After the WGD and dramatic loss of most paralogs, the retention of two TUP1 paralogs in post-WGD species is likely important. It is possible that this allows for more specialization or exploitation of different niches, but it is noteworthy that the Saccharomyces species complex appears to have only one homolog. It is possible that S. cerevisiae has simplified its ability to repress genes: i.e., there can be an ON/OFF switch for stress or no stress. Conversely, C. glabrata and related species may use the two paralogs to tailor multiple repressive regimes for varied stress conditions. Although we do not know the independent and overlapping roles of the duplicate TUP1 genes in C. glabrata, lessons from Schizosaccharomyces pombe, in which there are two homologs, may be informative. In S. pombe, Tup11 and Tup12 have both functionally redundant and distinct functions, but Tup12 appears to have more specific repression activity than Tup11 (27).

Our data implicate C. glabrata TUP1 and TUP11 in pathogenicity. For example, previous work has suggested that CgTup1 is recruited by Yap7 (part of the Yeast AP1 family) to repress YHB1; YHB1 encodes a gene for flavohemoglobin, which detoxifies nitric oxide (26). We were unable to see a significant differential effect of nitric oxide stress on the Cgtup1Δ tup11Δ strain relative to the wild type (data not shown), but it raises the possibility that the C. glabrata TUP1 paralogs might be important for survival in mammalian cells. Additionally, yapsin (aspartyl proteases) genes are important for C. glabrata survival in macrophages and cell wall structure and thereby have direct involvement in the species’ pathogenicity (28). Examination of the YPS gene family in the RNA-seq data indicated that YPS2, YPS4, YPS6, YPS8, YPS9, and YPS10 have increased expression in the absence of both TUP1 homologs. Given that 6 of the 11 YPS genes have increased expression (and YPS4 has a 152-fold increase in expression in the double mutant), a better understanding of how the C. glabrata Tup1 homologs contribute to pathogenicity is needed.

Finally, it is surprising that we observed few clear, robust phenotypes in the Cgtup1Δ tup11Δ or the Cgcyc8Δ strain, especially in light of many genes being repressed by these proteins. The simplest explanation would be that the genes were not actually deleted, but the RNA-seq data are strongly indicative of the appropriate gene deletion. Sctup1Δ strains grow slower than the wild type and can access alternative carbon sources. It is easy to think of ScTUP1 as repressing many stress genes, and loss of ScTUP1 results in many stress genes being upregulated, causing slow growth. In C. glabrata, a survey of many stress conditions did not identify clear phenotypes. Additionally, the deletion of CgCYC8 suggested that there are genes that are strongly dependent on the TUP1 paralogs but not as dependent on CYC8, such as CgYPS4 or CgHBN1. This suggests that the well-defined complexes described for S. cerevisiae are not so canonical in other species, and possibly other proteins are required for repression in Tup1-containing complexes (17, 19). The altered lack of repression of genes by CgTUP1 in the Sctup1Δ strain (both partial repression of ScTUP1 targets and additional genes not thought to be regulated by ScTUP1) supports the notion that there are likely multiple complexes that target different genes. Additionally, supporting the argument of additional complexity in C. glabrata is the observation that very few genes are downregulated in response to deletion of the TUP1 homologs in C. glabrata. This is in contrast to the case with S. cerevisiae, for which there are demonstrated activation roles for the Tup1-Cyc8 complex (29, 30). Further dissection of the Tup1 and Tup11 complexes is warranted to understand the differential roles of these complexes in gene repression in post-WGD species.

MATERIALS AND METHODS

Strains and plasmids.

Strains used in this study are listed in Table S2. Genes were deleted in the C. glabrata wild-type strain using antibiotic resistance markers, KANMX6 and NATMX6, which replaced open reading frames via homologous recombination (primers listed in Table S3). Deletions were verified using gain of the selectable marker as well as PCR to confirm loss of the open reading frame and positivity for flanking PCR regions.

Strains used in this study. Download Table S2, PDF file, 0.5 MB.

Primers used in this study. Download Table S3, PDF file, 0.5 MB.

URA3+ plasmids (pRS316) containing TUP genes were used for cross complementation experiments with an Sctup1Δ strain. Empty pRS316 plasmid (vector) was used as a negative control, and pRS316 containing wild-type ScTUP1 was used as a positive control. Plasmids containing CgTUP1 and CgTUP11 were also transformed into the Sctup1Δ strain. The primers for construction of the plasmids by gap repair are listed in Table S3 (31).

HIS3+ plasmids (pRS313) containing TUP genes were used for cross complementation experiments with Cgtup1Δ and Cgtup11Δ strains. The primers for construction of the plasmids by gap repair are listed in Table S3.

Phenotypic assays.

To investigate visible phenotypic differences between deletion strains and wild-type, strains were grown on various plate and liquid medium conditions. The experimental plate conditions were 1 M potassium chloride, 0.25 M calcium chloride, 110 μM inositol, 0.5 μg/mL of ketoconazole, and high temperature (37°C) in yeast extract, peptone, dextrose (YEPD) standard medium. Each plate was divided into six subsections for S. cerevisiae wild-type and Sctup1Δ and C. glabrata wild-type, Cgtup1Δ, Cgtup11Δ, and Cgtup1Δ tup11Δ strains. From 5 mL of YEPD liquid cultures grown overnight, colonies were streaked for single colonies. Growth was assessed after 24 h at 30°C for all conditions except high temperature, which was grown at 37°C for 24 h.

To assess phenotypes in liquid media, the S. cerevisiae wild-type and Sctup1Δ strains and C. glabrata wild-type, Cgtup1Δ, Cgtup11Δ, Cgtup1Δ tup11Δ, and Cgcyc8Δ strains were grown under various conditions. In standard yeast extract and peptone (YEP) medium, different carbon sources replaced the 2% glucose: 1% ethanol and 1% glycerol together, 0.2% glucose, 2% ethanol, and 2% acetic acid. In standard yeast extract, peptone, and dextrose (YEPD) medium, various compounds were added: 9 mM 2-deoxyglucose, 1 mM FeCl3, 2.8 mM Congo red, 1 M potassium chloride (KCl), 10 mM Tris (pH 7.9), and 0.5 μg/mL of ketoconazole. Strains were grown in standard YEPD overnight, and then cells were harvested and washed three times with water. Cultures were inoculated at an optical density at 600 nm (OD600) of 0.05 in triplicate. After 24 to 48 h of growth at 30°C, the optical density of each culture was measured to determine differences in growth. A subset of these conditions was retested using synthetic defined (SD) medium instead of standard YEP (Fig. 2B and Fig. S1). For the data presented in Fig. S1, six biological replicates of each strain were inoculated at an OD600 of 0.01 in 96-well plates rather than in culture tubes.

To assess the phenotype for growth in liquid medium containing sorbitol instead of glucose, S. cerevisiae wild-type and Sctup1Δ strains and C. glabrata wild-type, Cgtup1Δ, Cgtup11Δ, Cgtup1Δ tup11Δ, and Cgcyc8Δ strains were precultured in YEPD overnight and then in YEP with 3% glycerol overnight. The cells were washed three times with water and inoculated into SD medium with either 2% sorbitol or 2% glucose at an OD600 of 0.05 in triplicate. After 24 to 48 h of growth at 30°C, the optical density of each culture was measured. It is worth noting that we were unable to observe a difference in growth in YEP plus 2% sorbitol, whereas the effect was dramatic in SD, suggesting that components of the media can influence the observation of phenotypes.

To measure flocculation rates, the change in OD600 over time was measured after vigorous vortexing. The Sctup1Δ strain was transformed with URA3 plasmids containing ScTUP1, CgTUP1, CgTUP11, or no gene (vector). These strains were grown in SD medium without uracil for ∼20 h at 30°C. Each sample was vortexed for 30 s, and the OD600 was recorded every 15 s for 1 min.

RT-qPCR.

C. glabrata wild-type, Cgtup1Δ, Cgtup11Δ, Cgtup1Δ tup11Δ, and Cgcyc8Δ strains were grown in triplicate in YEPD for ∼20 h, then inoculated at a low density in fresh YEPD medium, and grown for 5 h at 30°C to logarithmic growth phase. RNA was extracted using the Zymo Research Corp. Direct-zol RNA MiniPrep Plus kit and reverse transcribed to cDNA using the Bio-Rad iScript cDNA synthesis kit. Quantitative PCR was performed with a CFX qPCR machine (Bio-Rad) using Bio-Rad SsoAdvanced Universal SYBR green Supermix in a 25-μL reaction mixture. The amount of transcript for each gene was normalized to CgMIC10, which has consistent expression across the various strains, unlike the more common normalization control CgACT1. Each gene was also amplified using 10-fold genomic DNA dilutions as an amplification control. Based on RNA-seq data, we targeted genes that exhibited elevated transcription levels in tup1 mutants. Genes and primer sequences can be found in Table S3.

RNA-seq.

RNA sequencing was performed on two sets of strains, one set to look at expression in C. glabrata and one set to look at expression in S. cerevisiae. C. glabrata wild-type, Cgtup1Δ, Cgtup11Δ, and Cgtup1Δ tup11Δ strains were grown in YEPD standard medium for 6 h at 30°C. Wild-type and Cgtup1Δ tup11Δ strains were grown in triplicate, but Cgtup1Δ and Cgtup11Δ strains were only grown in duplicate. The Sctup1Δ strain was transformed with URA3 plasmids containing ScTUP1, CgTUP1, CgTUP11, or no gene (vector). These strains were grown to logarithmic growth phase in SD medium without uracil for 6 h at 30°C in triplicate.

RNA was purified using the Zymo Research Corp. Direct-zol RNA MiniPrep Plus kit, and the concentration of RNA was determined using a Qubit 3.0 fluorometer and the Qubit RNA HS assay kit. The RNA library was prepared using the NEBNext Ultra II RNA library prep kit for Illumina protocol from New England BioLabs, Inc. Samples were diluted in 0.1× Tris-EDTA (TE) to 4 nM and the Illumina NextGen MiSeq sequencer was used to sequence the samples, generating FASTQ files for each sample that were >1 million reads. Single-end reads were paired, trimmed, and aligned to the reference C. glabrata or S. cerevisiae genomes using Geneious (using default settings), and RPKM for each gene was exported for analysis. FASTQ files are deposited in the NCBI SRA database (accession number PRJNA782995).

To analyze the RNA sequencing data, genes were sorted by ratio of expression between the mutant and wild-type strains. In the case of C. glabrata, the Cgtup1Δ tup11Δ strain was used as the mutant for obtaining ratios. For S. cerevisiae, Sctup1Δ with the empty vector was compared to Sctup1Δ with ScTUP1. A two-tailed Student t test was used to compare expression levels between mutant and wild type. The genes were then sorted by the P values acquired from the t test. For the top 100 genes that were highly expressed (P < 0.05), expression was compared to the expression of the corresponding homolog in the other species and it was noted whether the difference in expression between strains was significant.

Data availability.

RNA-seq data files are available as FASTQ files in the NCBI SRA database (accession no. PRJNA782995). Experiments with mutants and the wild type and their biological replicates are available as accession numbers SAMN23411010 through SAMN23411031. Analysis of FASTQ files is available in Table S1.