Pyruvate decarboxylase and thiamine biosynthetic genes are regulated differently by Pdc2 in S. cerevisiae and C. glabrata

Abstract

Understanding metabolism in the pathogen Candida glabrata is key to identifying new targets for antifungals. The thiamine biosynthetic (THI) pathway is partially defective in C. glabrata, but the transcription factor CgPdc2 upregulates some thiamine biosynthetic and transport genes. One of these genes encodes a recently evolved thiamine pyrophosphatase (CgPMU3) that is critical for accessing external thiamine. Here, we demonstrate that CgPdc2 primarily regulates THI genes. In Saccharomyces cerevisiae, Pdc2 regulates both THI and pyruvate decarboxylase (PDC) genes, with PDC proteins being a major thiamine sink. Deletion of PDC2 is lethal in S. cerevisiae in standard growth conditions, but not in C. glabrata. We uncover cryptic cis elements in C. glabrata PDC promoters that still allow for regulation by ScPdc2, even when that regulation is not apparent in C. glabrata. C. glabrata lacks Thi2, and it is likely that inclusion of Thi2 into transcriptional regulation in S. cerevisiae allows for a more complex regulation pattern and regulation of THI and PDC genes. We present evidence that Pdc2 functions independent of Thi2 and Thi3 in both species. The C-terminal activation domain of Pdc2 is intrinsically disordered and critical for species differences. Truncation of the disordered domains leads to a gradual loss of activity. Through a series of cross species complementation assays of transcription, we suggest that there are multiple Pdc2-containing complexes, and C. glabrata appears to have the simplest requirement set for THI genes, except for CgPMU3. CgPMU3 has different cis requirements, but still requires Pdc2 and Thi3 to be upregulated by thiamine starvation. We identify the minimal region sufficient for thiamine regulation in CgTHI20, CgPMU3, and ScPDC5 promoters. Defining the cis and trans requirements for THI promoters should lead to an understanding of how to interrupt their upregulation and provide targets in metabolism for antifungals.

Introduction

Yeast pathogens have become increasingly dangerous in recent decades due to the use of wide-range antibiotics, surgeries, and catheters, while treatment of fungal infections has led to the appearance of antifungal resistance in yeast. The yeast species Candida glabrata is the second most common source of candidiasis, and acquires antifungal resistance relatively easily [1]. C. glabrata is more closely related to the relatively nonpathogenic yeast Saccharomyces cerevisiae as opposed to other Candida pathogens [2, 3]. C. glabrata and Saccharomyces cerevisiae are both Saccharomycetaceae species that diverged from other yeast species after a whole genome duplication event in a common ancestor [4]. We have focused on thiamine starvation in these two species because they are appealing model systems to compare evolutionary conservation of function, and it is a case where there are significant differences between the two species [5–8]. Additionally, by understanding the thiamine starvation pathway (THI pathway), we may be able to identify druggable targets in C. glabrata.

Thiamine is required for all life forms as it is required for glucose catabolism, and thiamine pyrophosphate (TPP) is the essential, functional cofactor for amino acid and carbohydrate metabolism [9]. S. cerevisiae can synthesize thiamine de novo and utilize extracellular thiamine sources to synthesize TPP. In S. cerevisiae, two transcription factors, Thi2 and Pdc2, and one regulatory factor, Thi3, modulate the activation of THI genes in response to thiamine starvation. Thi3 is related to the major TPP binding proteins, pyruvate decarboxylases, and when TPP is abundant in cells, TPP occupies a pseudo active site in Thi3p, destabilizing the transcriptional complex of Thi3, Thi2, and Pdc2 [10]. Under thiamine starvation conditions, TPP is dissociated from Thi3, and Thi3 interacts with the C-terminal domain of Pdc2. Pdc2 can then recruit transcription factors efficiently, and the transcription factor complex drives the transcription of THI genes [11]. While this model has not been fully supported, the thought is that due to this dynamic interaction between TPP and Thi3, the Pdc2 interaction with THI promoters is sensitive to intracellular TPP concentration with a Thi2/Thi3/Pdc2 complex forming at THI promoters [12].

C. glabrata lacks the transcriptional coactivator Thi2 and cannot synthesize the pyrimidine precursor for thiamine, making C. glabrata auxotrophic for thiamine. However, C. glabrata upregulates a small set of genes to acquire thiamine and thiamine precursors during thiamine starvation [6]. Overexpression of CgTHI3 can partially compensate for the loss of THI2 in S. cerevisiae, suggesting that Pdc2 can be activated independent of the Thi2 transcription factor [6]. Despite the loss of some THI genes, such as THI2, C. glabrata uniquely acquired the gene CgPMU3, which is upregulated during thiamine starvation and not found in any other species [8]. CgPMU3 is a newly evolved gene that encodes a TPP phosphatase, which is needed to utilize external TPP, and serves as a functional analog to ScPho3. Consequently, deleting CgPMU3 leads to a growth defect in an environment with TPP as the primary thiamine source, such as human serum [8, 13]. Thus, there are significant differences between the two species in what is upregulated in response to thiamine starvation, and what factors are required for that upregulation.

While required for upregulation of thiamine biosynthesis genes, Pdc2 is essential for expression of pyruvate decarboxylase genes (PDC genes), including ScPDC1 and ScPDC5 in S. cerevisiae [14]. Deletion of ScPDC2 is lethal in medium with high glucose concentrations, and it is thought this is because the PDC genes are not sufficiently expressed to metabolize glucose [15]. Overexpression of ScPDC1 suppresses the growth defect of the Scpdc2Δ strain [12]. Pdc2 also interacts with the promoter of PDC5 (PDC5pr), which encodes a pyruvate decarboxylase isoform synthesized under thiamine starvation conditions [12, 14]. PDC5pr is utilized for investigating Pdc2 binding because it is dependent on Pdc2, but independent of Thi2 and Thi3, making it a simpler promoter to distinguish the binding site for Pdc2. Nosaka et al. identified an upstream region of PDC5pr between -416 and -346 nucleotides that is necessary for ScPDC5 expression, and this region shares sequence similarity with a binding site in THI promoters that we identified in C. glabrata [7, 12]. Because deletion of ScPDC2 is lethal, few genome-wide studies have examined Pdc2 properties in S. cerevisiae. For this reason, and because ScPdc2 appears to regulate different kinds of genes (THI and PDC genes) and have different requirements for cofactors, we examined the properties of Pdc2 in both species in detail.

We undertook this work to understand the roles of Pdc2 in both species and to understand the different requirements of Pdc2-dependent promoters. We demonstrate: 1) the Pdc2 activation domain is critical for essentiality in S. cerevisiae. 2) The PDC and THI promoters have different requirements for transcriptional regulators with Pdc2. 3) S. cerevisiae and C. glabrata have different requirements at Pdc2 dependent promoters which correlates with the loss of Thi2. 4) The activation domain of Pdc2 exhibits characteristics consistent with intrinsically disordered regions (IDRs) [16], but loss of parts of these IDRs does not affect promoters differentially. 5) Finally, we identify the minimal cis regions for different classes of promoters that are sufficient to confer thiamine starvation upregulation to a basal promoter.

Methods

Strains

Strains used in this study are listed in S1 Table. Experiments were performed in S. cerevisiae wild-type and C. glabrata wild-type, as well as strains in which the thiamine pathway regulators were deleted: Scpdc2Δ, Scthi3Δ, Scthi2Δ, Cgpdc2Δ, and Cgthi3Δ [6, 7, 17–19].

To swap Pdc2 proteins between S. cerevisiae and C. glabrata, PDC2 was first deleted with URA3, which replaced the open reading frame (ORF) via homologous recombination [20]. Deletions were verified by gain of the URA3 marker as well as by PCR to confirm loss of the gene and positivity for flanking PCR regions. To construct a Scpdc2::URA3 strain capable of growth in glucose medium, ScPDC1 was overexpressed under the control of the ScADH1 promoter on a LEU2⁺ plasmid (pRS315) [7] in S. cerevisiae wild-type before deletion of PDC2. We chose to perform all assays with standard growth conditions because the Scpdc2 strain grows slowly in ethanol/glycerol containing medium. The open reading frame of PDC2 from both species was then amplified and transformed into the pdc2::URA3 strains to precisely replace the ORFs. Strains were confirmed by loss of the selectable URA3 marker as well as by PCR. (All primer sequences listed in S2 Table).

To measure expression of C. glabrata THI promoters when CgPdc2 was truncated, the promoters were fused to yellow fluorescent protein (YFP) and incorporated into the genome, rather than introduced on a plasmid, to minimize noise from a plasmid reporter. The promoter-YFP was amplified to precisely replace the CgURA3 ORF in C. glabrata wild-type using homologous recombination. Strains were verified by loss of URA3, by PCR, and by presence of fluorescence during thiamine starvation, indicating expression of the promoter. CgPDC2 was then deleted in these fluorescent strains using URA3 as a selectable marker. Strains were verified by gain of URA3, by confirmatory PCR, and by loss of fluorescence during thiamine starvation, indicating PDC2 was deleted. URA3 was then replaced by CgPDC2, either full-length or truncated in 40 amino acid increments from the C-terminus (amplified from the plasmids described below), and strains were again confirmed by loss of the URA3 marker and by PCR.

Plasmids

To assess complementation of the Scpdc2Δ and Cgpdc2Δ strains, ScPDC2, CgPDC2, or fusions of the DNA binding domain and activation domain of these two proteins, were cloned by homologous recombination into a HIS3⁺ plasmid (pRS313) containing the CgPDC2 promoter [20, 21]. Plasmids were verified by PCR and by complementation in the native species’ deletion strain.

To assay induction of pyruvate decarboxylase (PDC) and thiamine (THI) pathway genes, we constructed plasmids where the promoter region of each gene is fused to YFP, allowing expression to be measured via fluorescence. The promoters (1 Kbp or 2 Kbp upstream of the start codon) were amplified by PCR and cloned by homologous recombination into a HIS3⁺ plasmid (pRS313) containing

YFP in a wild-type strain [20–22]. Plasmids were verified by PCR and fluorescence in a wild-type strain in the appropriate growth conditions.

To assess the ability of small regions from THI promoters to confer thiamine regulation to an unregulated promoter CgPMU1, the THI promoter regions and CgPMU1 promoter were amplified by PCR to have overlapping sequences and all three PCR products were cloned by homologous recombination into a HIS3⁺ plasmid (pRS313) containing YFP. THI promoter regions were inserted into the CgPMU1 promoter approximately the same distance upstream of the start codon as in the native THI promoter (see schematic in Fig 8A). Plasmids were confirmed by PCR and whole plasmid sequencing (Plasmidsaurus).

Fig 8. Short regions of C. glabrata THI promoters, or ScPDC5 promoter, can confer thiamine starvation regulation to an unregulated promoter.

Regions from ScPDC5, CgTHI20, and CgPMU3 promoters were incorporated into the CgPMU1 promoter, which is not regulated by thiamine, fused to YFP. (A) Schematic demonstrating construction of the CgPMU1 promoter with regions from the ScPDC5 and C. glabrata THI promoters incorporated. (B) Expression was measured in high thiamine and thiamine starvation (no thiamine). The data presented is the mean and standard deviation of three independently grown samples.

To assay induction of THI promoters when CgPDC2 is truncated, truncated regions of CgPDC2, with the CgPDC2 promoter and terminator, were amplified by PCR and cloned by homologous recombination into a HIS3⁺ plasmid (pRS313). Plasmids were confirmed by PCR. (All primers used are listed in S2 Table).

Flow cytometry

To assay induction of pyruvate decarboxylase (PDC) and THI pathway promoters, fluorescence of cells containing plasmids with promoters fused to YFP was quantified by flow cytometry. Cells were grown at 30°C overnight in thiamine replete SD medium lacking histidine (Sunrise Science, CA). Cells were harvested by centrifugation, washed three times with sterile water, inoculated into thiamine replete (0.4 mg/L) and starvation (no thiamine added) conditions in SD medium lacking histidine, and grown at 30°C overnight (~18 hours). Mean fluorescence (in arbitrary units, a.u.) of each strain was measured using a flow cytometer with a 533/30 filter set (Accuri C6 Plus, BD Biosciences). In almost all cases, background fluorescence was less than 10,000 a.u.; however, there is variability of fluorescence based on precise growth conditions and we included positive and negative controls in each experiment.

Results

Both ScPdc2 and CgPdc2 regulate thiamine starvation regulated genes, but ScPdc2 is critical for pyruvate decarboxylase gene expression, making its loss lethal in standard growth conditions

Deletion of ScPDC2 is lethal to cells grown in high glucose conditions, whereas deletion of CgPDC2 does not result in an obvious growth defect, suggesting that the transcription factor has different specificities in the two yeast species. To understand the role of Pdc2 on the growth properties of S. cerevisiae and C. glabrata, we deleted PDC2 in both species. As Pdc2 is not essential for growth in C. glabrata, we were able to delete PDC2 normally. In S. cerevisiae, Pdc2 is essential for growth in 2% glucose conditions, and so we covered the Scpdc2∆ with ScPDC2 on a URA3⁺ containing plasmid. This allowed us to counter select against the plasmid with 5-FOA and transform in plasmids containing ScPDC2 or CgPDC2 on a HIS3⁺ containing plasmid to determine which can complement the growth defect of the Scpdc2∆ strain. We confirmed previous work that demonstrated that ScPDC2 is essential in standard glucose rich medium. Additionally, we determined that CgPDC2 was unable to rescue the lethal phenotype (Fig 1A). Finally, we confirmed that overexpression of ScPDC1 (under the control of the ScADH1 promoter) was able to rescue the lethality of the Scpdc2Δ (Fig 1A), which had been previously reported [12].

Fig 1. Pdc2 is critical for pyruvate decarboxylase gene expression in S. cerevisiae, but regulates thiamine starvation regulated genes in both S. cerevisiae and C. glabrata.

The activation domains (AD) dictate specificity. (A-B) Growth of a Scpdc2∆ strain with ScPDC2, CgPDC2, ScADH1pr-ScPDC1, or fusions of the DNA binding domain (DBD) and activation domain (AD) of both species’ Pdc2 proteins introduced on a plasmid. (C) Expression of CgTHI4 promoter fused to YFP in a Cgpdc2∆ strain and (D) ScTHI4 promoter fused to YFP in a Scpdc2∆ strain with PDC2 introduced on a plasmid. The fluorescence data presented is the mean and standard deviation of three independently grown samples.

Given that the DNA binding domain (DBD) in the N-terminus (1–485 aa in C. glabrata and 1–494 aa in S. cerevisiae) shares 78% identity (87% similarity) to one another and the C-terminal activation domains (AD) have 22% identity (42% similarity), we hypothesized that the DBDs were likely interchangeable and the AD is likely the critical determinant for gene regulation specificity. By placing fusions of ScPdc2 DBD–CgPdc2 AD and CgPdc2 DBD–ScPdc2 AD into the Scpdc2Δ strain, we were able to determine that the ScPdc2 AD is important for viability by its rescue of lethality in standard medium (Fig 1B). Whereas the S. cerevisiae and C. glabrata Pdc2 proteins have different functions for viability, both regulate thiamine starvation regulated genes as demonstrated with promoter-YFP fusion expression (Fig 1C—Cgpdc2∆ and 1D—Scpdc2∆); swapping Pdc2 in each species allows for the transcriptional induction of THI4, at least partially.

To determine whether the activation domains in each species dictate the ability to induce THI genes, we examined the Pdc2 fusions from Fig 1B. In C. glabrata, the AD is the primary determinant to induce transcription, suggesting the AD interacts with general transcriptional machinery to drive the upregulation. In S. cerevisiae, the story is different. The AD of S. cerevisiae seems to be the important determinant (compare ScPdc2 to CgPdc2 DBD/ScPdc2 AD), but the opposite fusion seems to remove the ability of Pdc2 to function in S. cerevisiae (compare CgPdc2 to ScPdc2 DBD/CgPdc2 AD). Because this same plasmid functions in C. glabrata (Fig 1C), and the entire C. glabrata Pdc2 protein can complement the Scpdc2∆ (Fig 1D), this suggests that there is antagonism between the S. cerevisiae DBD and the C. glabrata AD, but only in S. cerevisiae. Given that S. cerevisiae also has Thi2 (and ScTHI4pr is dependent on Thi2), it suggests that this antagonism might disrupt the interactions in the Pdc2/Thi3/Thi2 complex. This is not surprising because the ScPdc2 AD region (407–925 aa) has been shown to interact with Thi3 in two hybrid assays [11]. It is worth noting that the ScPdc2 DBD/CgPdc2 AD fusion still functions in C. glabrata which suggests it can still interact with Thi3 at least in C. glabrata. In sum, we believe the presence/absence of Thi2 leads to this contradictory data.

PDC2 is more important for regulating pyruvate decarboxylase (PDC) genes in S. cerevisiae relative to C. glabrata

Given that the Scpdc2Δ strain is inviable in glucose containing medium, but the Cgpdc2Δ is viable, we explored the regulation of PDC genes in both species. S. cerevisiae contains three PDC genes. ScPDC1 is thought to be the major PDC isoform in standard medium, ScPDC5 is thought to be a specialized thiamine repressible PDC, and of note both open reading frames are very similar to one another (88% identity) [14, 23, 24]. ScPDC6 (84% identity with ScPDC1) is thought to be important for growth on non-fermentable carbons sources [25]. ScTHI3 is 52% identical to ScPDC1, but it is not catalytically active (it is a regulator of Pdc2). Thus, we choose to focus on the two important PDCs in standard medium (ScPDC1 and ScPDC5). C. glabrata has only two copies of PDC genes (CgPDC1—CAGL0M07920g and CgPDC5—CAGL0G02937g) and CgTHI3. There are other weakly related proteins that are likely carboxylases with substrate specificity different from pyruvate.

To understand the regulation of the 4 genes (ScPDC1, ScPDC5, CgPDC1, and CgPDC5), we cloned 1 kb of each promoter upstream of YFP in a plasmid and examined YFP expression in wild-type, pdc2Δ, thi3Δ, and Scthi2Δ (Fig 2). We examined expression in both high thiamine and thiamine starvation conditions (no thiamine), because it was known that ScPDC5 is regulated by thiamine starvation [26]. We observe in S. cerevisiae that ScPDC5pr is expressed at a low level in high thiamine conditions and is induced >50-fold during starvation. We note that both ScPDC1pr and ScPDC5pr are dependent on ScPDC2, explaining why loss of PDC2 is so detrimental to a cell in standard glucose conditions (Fig 2A). Interestingly, Hohmann’s group demonstrated that loss of ScPDC1 is not lethal, and leads to an upregulation of ScPDC5 even in high thiamine conditions [23, 26, 27]. We confirmed that in our assay, ScPDC5pr-YFP expression is increased in a Scpdc1Δ strain relative to wild-type even in high thiamine conditions (S1 Fig). This indicates that there are feedback mechanisms that impact PDC expression independent of thiamine status, and that those feedback mechanisms likely are a response to lack of pyruvate decarboxylase activity the cell. It is unclear whether PDC expression can be independent of ScPdc2.

Fig 2. Pdc2 is more important for regulation of pyruvate decarboxylase (PDC) genes in S. cerevisiae than C. glabrata.

Expression of PDC1 and PDC5 promoters fused to YFP in high thiamine and thiamine starvation (no thiamine) in (A) S. cerevisiae and (B) C. glabrata strains where the thiamine regulatory factors were deleted. The data presented is the mean and standard deviation of three independently grown samples.

Finally, we note a consistent elevation of expression of both S. cerevisiae genes in the absence of THI3 or THI2, indicating that expression does not require these factors. This suggests that ScPdc2 has roles in transcription independent of thiamine starvation. Given that data has indicated that ScPdc2 is in a complex with Thi3 and Thi2 during thiamine starvation (and even in high thiamine to a lesser extent) [11, 12], our data suggest that Thi3 and Thi2 inhibit Pdc2 at PDC promoters. A simple explanation would be that the Pdc2/Thi3/Thi2 complex is unable to bind PDC promoters, and Pdc2 alone, or with other factors, binds PDC promoters.

In C. glabrata, we do not observe the same complexity (Fig 2B). Loss of CgPDC2 impacts expression of CgPDC5 but only reduces CgPDC1 expression by two-thirds, likely explaining why deletion of CgPDC2 is not lethal–there is still significant expression of CgPDC1. We believe that factors other than Pdc2 might regulate these promoters, and the role of Pdc2 is diminished. Loss of THI3 does alter expression of CgPDC1 and CgPDC5 expression during thiamine starvation like S. cerevisiae, suggesting that Pdc2 could have a role in regulating these promoters, just in a diminished capacity. From these data, it is clear PDC2 is not as important for expression of PDCs, and it would suggest that deletion of PDC2 is not lethal in C. glabrata because its deletion does not severely reduce PDC expression, but in S. cerevisiae Pdc2 is crucial for PDC regulation.

Regulation of THI genes in both species

Given the different requirements for transcription of PDC genes between the species, the question arose: what are the requirements of THI gene promoters in the two species? To examine this, we chose three types of THI genes: the major thiamine uptake transporter THI10, two thiamine biosynthetic genes THI4 and THI20, and the TPPase CgPMU3 [8]. We did not study ScPHO3 because deletion of ScPDC2 did not have a large impact on its regulation [8, 28]. The choice of these genes spans both biosynthetic genes and acquisition genes. We cloned 1 kb of these promoters fused to YFP into a plasmid and examined expression. C. glabrata once again is a simpler case. All classes of THI gene promoters are highly dependent on CgPDC2 and CgTHI3 (Fig 3A).

Fig 3. C. glabrata THI promoters.

Expression of thiamine regulated promoters fused to YFP in high thiamine and thiamine starvation (no thiamine) in strains where the thiamine regulatory factors were deleted. C. glabrata promoters in (A) C. glabrata strains and (B) S. cerevisiae strains. The data presented is the mean and standard deviation of three independently grown samples.

To determine whether these C. glabrata promoters functioned similarly in S. cerevisiae, we moved the promoters to S. cerevisiae strains and assayed expression (Fig 3B). Expression of CgTHI4 and CgPMU3 behave as if they are standard THI promoters in S. cerevisiae, albeit at a much lower level. Even though these promoters function without Thi2 in C. glabrata, they now are dependent on the Pdc2/Thi3/Thi2 complex present in S. cerevisiae. Expression of CgTHI10 and CgTHI20 is lost in S. cerevisiae, underlying that while all C. glabrata THI promoters appear simple in their requirements, there is more complexity at the promoter that we do not understand; the promoters behave identically in C. glabrata but not the same in S. cerevisiae. Interestingly, CgPMU3 has a different cis region from all of the other C. glabrata THI promoters, raising the question of how a promoter that does not have the same cis architecture as other THI promoters is still responsive to the same trans environment in S. cerevisiae [7]. CgPMU3 is a recently evolved phosphatase gene that cleaves thiamine pyrophosphate (TPP), is required for C. glabrata to survive when TPP is the only thiamine source, and is regulated by the same transcription factors as other THI genes. Because CgPMU3 promoter does not use a cis element that is common to other THI promoters, we believe these data can begin to untangle the cis architectures required for regulation by thiamine status. This is because CgPMU3 evolved recently, whereas all of the other THI genes have long common evolutionary histories and have likely experienced similar selection pressures over evolutionary time.

In S. cerevisiae, each class of promoters has slightly different requirements (Fig 4). ScTHI10, ScTHI4, and ScTHI20 are dependent on ScPDC2 and ScTHI3. ScTHI2 is required for expression of ScTHI20, but not the ScTHI10 transporter gene, and only partially for the biosynthetic gene ScTHI4. These data are consistent with previous work that indicates Thi2 is necessary for the expression of some, but not all THI genes in S. cerevisiae [10, 12, 14]. While qualitatively we observe that S. cerevisiae promoters often function in C. glabrata, this is true for ScTHI10 and ScTHI4, both of which do not absolutely require Thi2 –i.e. these promoters do not need Thi2 and function fine with Pdc2 and Thi3. Interestingly, these S. cerevisiae THI promoters that do not require Thi2 are expressed at approximately the same level in either species. A simple explanation for this could be that the simple CgPdc2/CgThi3 complex effectively binds cis elements from either species, but the lack of Thi2 in C. glabrata prevents expression of some THI genes (like ScTHI20), and likewise, the C. glabrata promoters are expressed much less efficiently because of Thi2 interference, and potentially, the sequestering of ScPdc2 binding to PDC promoters.

Fig 4. S. cerevisiae promoters.

Expression of thiamine regulated promoters fused to YFP in high thiamine and thiamine starvation (no thiamine) in strains where the thiamine regulatory factors were deleted. S. cerevisiae promoters in (A) S. cerevisiae strains and (B) C. glabrata strains. The data presented is the mean and standard deviation of three independently grown samples.

To dissect the trans effects of Pdc2 more thoroughly, we precisely replaced the open reading frames of PDC2 in each species, allowing us to determine if having the appropriate species’ version of Pdc2 allows for a restoration of expression. Here, we find that C. glabrata THI promoter-YFP plasmids require CgPDC2 to fully induce expression, and swapping the species’ Pdc2 eliminates most expression (Fig 5A). This indicates that in C. glabrata the THI promoters require some function of CgPdc2 that ScPdc2 cannot provide. Additionally, it suggests that just having CgPdc2 in S. cerevisiae is not sufficient for expression of C. glabrata promoters.

Fig 5. C. glabrata PDC2 is necessary but not sufficient for the expression of C. glabrata THI promoters, while S. cerevisiae THI promoters can be regulated by PDC2 from either species.

Expression of (A) C. glabrata and (B) S. cerevisiae THI promoters fused to YFP in high thiamine and thiamine starvation (no thiamine) in strains where the PDC2 open reading frame was precisely replaced with PDC2 from the opposite species. The data presented is the mean and standard deviation of three independently grown samples.

Alternatively, S. cerevisiae promoters function in S. cerevisiae, and these promoters are only somewhat sensitive to which version of Pdc2 is present (Fig 5B). This supports the previous arguments that S. cerevisiae promoters are able to function in S. cerevisiae even when CgPdc2 is the transcription factor. However, when expressed in C. glabrata, the S. cerevisiae THI promoters function best with the trans C. glabrata factors, as moving ScPdc2 into C. glabrata attenuates expression. This, not surprisingly, suggests that each species has optimized how the protein-protein interactions occur, and that Pdc2 is core to those protein-protein interactions. However, we cannot formally exclude the possibility that precise replacement of the open reading frame does not alter other factors, such as the stability of each protein in the other species.

PDC1 and PDC5 promoters

The biggest surprise was with CgPDC1 and CgPDC5, which appeared only subtly regulated by Pdc2 and Thi3 in C. glabrata, but now are highly induced during thiamine starvation in S. cerevisiae when ScTHI2 and ScTHI3 are deleted (Fig 6A). This effect is prominent with the CgPDC1 promoter. This suggests that CgPDC1 might have the remnants of ScPDC1 regulation, but in C. glabrata other factors are acting on PDC promoters masking the role of CgPdc2. This is not surprising given it seems likely that the PDC promoters have complex regulation independent of thiamine as pyruvate decarboxylase activity is critical for glucose metabolism. Surprisingly, wild-type S. cerevisiae has low expression of CgPDC1 and CgPDC5 and no apparent increase in expression in response to thiamine starvation. It is worth noting that if such complexities are present, we are unable to disentangle them with these experiments. In S. cerevisiae wild-type, the native ScPDC1 is being expressed to a high-level, potentially suppressing expression of CgPDC1, and in the Scpdc2Δ strain, where ScPDC1 is expressed from an ectopic promoter (ScADH1pr), a similar situation is likely happening. In both cases, it seems likely that CgPDC1 is not expressed in S. cerevisiae because there are high levels of ScPdc1 in the cell. These results are actually not surprising given that Hohmann’s group has demonstrated feedback mechanisms on ScPDC5 in the absence of ScPDC1 [23, 27]. The key result here is that loss of Thi3/Thi2 appears to release the constraint on Pdc2, allowing for high expression of CgPDC1. This indicates that the CgPDC1 promoter likely binds ScPdc2 without Thi2/Thi3. A similar process appears to happen with CgPDC5, but to a lesser extent.

Fig 6. PDC promoters in both species.

Expression of PDC1 and PDC5 promoters fused to YFP in high thiamine and thiamine starvation (no thiamine) in strains where the thiamine regulatory factors were deleted. (A) C. glabrata PDC promoters in S. cerevisiae strains. (B) S. cerevisiae PDC promoters in C. glabrata strains. The data presented is the mean and standard deviation of three independently grown samples.

When the ScPDC1pr-YFP plasmid is introduced into C. glabrata, it behaves the same as CgPDC1 in C. glabrata (compare Figs 2B–6B). This may indicate that both promoters have the ability to bind Pdc2 through the same cis elements and the trans milieu of C. glabrata dictates its expression. The converse is less obvious, but likely impacted by pyruvate decarboxylase activity feedback and the potential for CgPDC1 to be regulated by additional factors as we see in C. glabrata (compare Figs 2A–6A). When we introduce ScPDC5pr-YFP into C. glabrata, we see it is completely dependent on CgPDC2 and only partially dependent on CgTHI3 (Fig 6B). CgPDC5pr-YFP is expressed at a low level in S. cerevisiae, but does have similarities to its expression in C. glabrata (Fig 6A). We believe these data suggest that the trans environment is more important for dictating expression of the PDC genes in both species, but the cis elements are still required. We believe that PDC gene regulation warrants future studies to tease apart the requirements further and are beyond the scope of this study.

All of the data, so far, present a surprisingly complex set of transcriptional regimes. In S. cerevisiae there are cases where it appears that Pdc2 may act independent of the Thi2 and Thi3 regulators, where Pdc2 may require Thi3 but not Thi2, and where all three proteins likely form a Pdc2/Thi3/Thi2 complex. In C. glabrata the diversity of different complexes appears reduced, with PDC genes possibly only being partially dependent on Pdc2 and THI genes absolutely requiring both Pdc2 and Thi3 in a complex.

Both the DBD and AD of Pdc2 is required for the expression of C. glabrata genes and sequential loss of the intrinsically disordered regions (IDRs) in the AD impacts promoters in the same way

CgPMU3 having a different critical cis element relative to other promoters suggests that the architecture of the CgPMU3 promoter is different from other THI promoters, even though CgPMU3 uses both Pdc2 and Thi3 to upregulate expression during thiamine starvation [7]. We hypothesized that CgPMU3 has a different sensitivity to Pdc2 relative to other THI promoters, either not requiring all of Pdc2, or being differentially sensitive to parts of Pdc2. To test this hypothesis, we asked if CgPMU3 was more or less sensitive to fragmentation of Pdc2 -i.e. Does CgPMU3 only need the activation domain of Pdc2 to express during thiamine starvation? We determined that both the DBD and the AD are required for expression of ancestral THI promoters and CgPMU3, and that there does not appear to be a difference between the promoters–all promoters require both the DBD and AD of Pdc2 (Fig 7A). We also observed that Pdc2 has a large, disordered region in the AD [29] (S2 Fig), and tested to see if CgPMU3 was differentially sensitive to truncation of IDRs. The data indicate that there is not a large difference between how the THI promoters or CgPMU3 respond to truncation of Pdc2 (Fig 7B). The data are consistent with the region having IDRs, as truncations lead to a gradual loss of transcriptional activity, similar to other studies [16], but further studies are warranted to determine if there really is a significant, differential requirement at CgPMU3 relative to other THI promoters.

Fig 7. Both the DNA binding domain (DBD) and activation domain (AD) of Pdc2 are required for the expression of all C. glabrata genes and sequential loss of the IDRs in the activation domain impacts C. glabrata promoters in the same way.

(A) Expression of C. glabrata promoters fused to YFP in a Cgpdc2∆ strain with PDC2, or just the DNA binding domain (DBD) or activation domain (AD) of PDC2, introduced on a plasmid. (B) Expression during thiamine starvation of C. glabrata promoters fused to YFP in a Cgpdc2∆ strain with PDC2, truncated from the C-terminal end of the activation domain, incorporated into the genome to minimize noise from plasmid expression. The data presented is the mean and standard deviation of three independently grown samples.

Small sections of C. glabrata promoters can confer thiamine starvation regulation to an unregulated promoter

Given the multiple behaviors and dependencies of thiamine regulated promoters, we wanted to identify the smallest regions required to confer thiamine starvation regulation in three promoters. We choose 1) the ScPDC5 promoter because it appears to only be PDC2 dependent (in both species), 2) the CgTHI10 promoter because it requires both THI3 and PDC2 and does not function in S. cerevisiae, and 3) the CgPMU3 promoter because it also requires THI3 and PDC2 but has different cis elements and does function in S. cerevisiae. To do this, we cloned 30 bp to 100 bp fragments around the cis elements [7] known to be important for regulation into the CgPMU1 promoter (Fig 8A). We identified regions of varying lengths that are sufficient for conferring thiamine regulation to the CgPMU1 promoter: 90 bp of the ScPDC5 promoter, 60 bp of the CgTHI20 promoter, and 100 bp of the CgPMU3 promoter (Fig 8B). All of the promoter regions require PDC2, as deletion abrogates the upregulation during thiamine starvation, but ScPDC5 continues to be THI3 independent (S3 Fig). In the ScPDC5 promoter, there are two 22 bp repeating elements that share similarity to the putative Pdc2 binding element identified previously [12], but CgPMU3 does not appear to have a clear homologous Pdc2 binding sequence. Future studies will examine these sequences in detail.

Discussion

Here, we observe that ScPdc2 and CgPdc2 have different roles in regulating genes. In S. cerevisiae, where fermentation is critical for much of its lifestyle, it is likely there is an advantage in having the same transcription factor be core to coupling thiamine biosynthesis and the major enzyme that uses thiamine, pyruvate decarboxylase. Addition of Thi2 likely allows for regulation of ScPdc2 to be further subdivided between THI and PDC genes. In S. cerevisiae, Pdc2 appears to work with Thi2 and Thi3 to regulate THI genes, and then separately, Pdc2 works alone (or in another complex) to drive transcription of the PDC genes.

In C. glabrata, Pdc2 appears to be core to just THI gene regulation. Thi3 and Pdc2 work together to drive transcription, and the C-terminal domain of Pdc2 is important for that complex to function. This is supported by C. glabrata THI genes being similarly sensitive to loss of both CgPDC2 and CgTHI3. However, the PDC genes appear to be regulated by transcription factors other than Pdc2, although the cross species complementation experiments seem to suggest that the cis elements required for Pdc2 regulation are still present. The presence of potential Pdc2 regulation suggests there is cryptic dual regulation of PDC and THI genes in C. glabrata that we are not able to observe in our experimental set up. It is possible that Pdc2 regulates PDC genes in C. glabrata but only in another growth condition that we did not examine in our studies.

The CgPMU3 promoter is recently evolved, does not have clear cis elements in common with other THI promoters in either species, and is Pdc2- and Thi3-dependent, raising the question of how this promoter functions. Because promoters between the two species are different enough from one another that they cannot be aligned, it is difficult to identify individual precise cis elements. However, having promoters from both species and a novel, newly evolved promoter narrowed down to ~100 bp now provides a lot of raw material for the eventual identification of sequences that are maintained for appropriate regulation. Future studies should be able to identify precise Pdc2 binding sites and those should correlate to conserved regions of each promoter. It will be interesting to determine whether the CgPMU3 promoter has acquired a cryptic Pdc2 binding site, or if a different cis element has recruited Pdc2 in a different fashion as other Pdc2 dependent promoters. One possibility is that we have not examined large enough promoter sequences, but typically promoters in these yeast species are small and we have demonstrated for some THI and PDC genes that examining 1 or 2 kb of the promoter recapitulates the same function (S4 Fig).

Interestingly, ScThi3 has been posited to be the thiamine sensor (or more precisely the TPP sensor), but in both species ScPDC5 is capable of being highly induced during thiamine starvation independent of Thi3. This suggests that Pdc2 or another unknown factor is capable of sensing thiamine concentrations. However, an alternative hypothesis is possible. Upregulation of ScPDC5 could be an indirect consequence of thiamine starvation. ScPdc1 protein could have a lower affinity for TPP than ScPdc5, and there could be feedback mechanisms leading to the increase in ScPDC5 expression during thiamine starvation because of pyruvate (or another metabolite) accumulating and activating ScPdc2 binding at the ScPDC5 promoter. Further studies are required to tease apart the regulation of PDC genes and answer the questions of when Pdc2 is binding and what complexes are binding at PDC promoters.

This work supports the argument that Pdc2 participates in multiple complexes to drive transcription of different classes of genes. The canonical complex has been Thi2/Thi3/Pdc2; however, we present evidence that loss of Thi3 actually enhances the expression of most PDC genes, indicating that the PDC genes bind Pdc2 as a different complex. Additionally, the ScPDC5 promoter suggests that cells sense thiamine concentration independent of Thi3. Our cloning of small cis regions sufficient to confer regulation should help us to understand what trans elements bind where. These cis elements require a complicated trans environment that we still do not understand. Future studies will identify the critical elements in these small regions and determine where transcription factors bind using chromatin immunoprecipitation.

Supporting information

S1 Table. Strains used in this paper.

(XLSX)

S2 Table. Primers used in this paper.

(XLSX)

S1 Fig. Loss of PDC1 leads to an upregulation of PDC5 in S. cerevisiae in both high thiamine and thiamine starvation conditions.

Expression of ScPDC5 promoter fused to YFP in high thiamine and thiamine starvation (no thiamine) in strains where the thiamine regulatory factors and PDC1 were deleted in S. cerevisiae. The data presented is the mean and standard deviation of three independently grown samples.

(EPS)

S2 Fig. The activation domains of C. glabrata and S. cerevisiae Pdc2 proteins are made up of intrinsically disordered regions.

Graph showing the predicted intrinsically disordered regions of the C. glabrata and S. cerevisiae Pdc2 proteins, where higher values correspond to a higher probability of disorder. Disorder tendency was calculated using the online tool IUPred2.

(EPS)

S3 Fig. The regions from PDC5 and THI promoters that confer regulation by thiamine to the CgPMU1 promoter are dependent on Pdc2.

Regions from ScPDC5, CgTHI20, and CgPMU3 promoters were incorporated into the CgPMU1 promoter fused to YFP, allowing this promoter to be upregulated during thiamine starvation. These promoter regions require Pdc2 and Thi3, as deletion abrogates expression, with the exception of ScPDC5, which remains Thi3 independent. The data presented is the mean and standard deviation of three independently grown samples.

(EPS)

S4 Fig. THI and PDC promoters in S. cerevisiae and C. glabrata are appropriately regulated by thiamine starvation regardless of the length of the promoter.

To determine if varying lengths of promoter sequence change the expression of THI genes, 2 Kb or 1 Kb upstream of the start codon was fused to YFP in a wild-type strain and expression of these promoters was measured in thiamine starvation. Expression of these promoters was fully repressed during high thiamine conditions, with the exception of the CgPDC1 promoter, which is not regulated by thiamine starvation. The data presented is the mean and standard deviation of three independently grown samples. (A) S. cerevisiae promoters in S. cerevisiae wild-type. (B) C. glabrata promoters in C. glabrata wild-type.

(EPS)

Data Availability

All relevant data are within the paper and its Supporting Information files.

Funding Statement

This work was funded by National Science Foundation grant (MCB 1921632 to D.D.W.), the Dennis M. Cook Endowed Gregor Mendel Chair in Genetics Endowment, the Villanova College of Liberal Arts and Sciences, and the Villanova Department of Biology. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.