OUP user menu

Facilitated recruitment of Pdc2p, a yeast transcriptional activator, in response to thiamin starvation

Kazuto Nosaka, Hiroyoshi Esaki, Mari Onozuka, Hiroyuki Konno, Yasunao Hattori, Kenichi Akaji
DOI: http://dx.doi.org/10.1111/j.1574-6968.2012.02543.x 140-147 First published online: 1 May 2012


In Saccharomyces cerevisiae, genes involved in thiamin pyrophosphate (TPP) synthesis (THI genes) and the pyruvate decarboxylase structural gene PDC5 are transcriptionally induced in response to thiamin starvation. Three positive regulatory factors (Thi2p, Thi3p, and Pdc2p) are involved in the expression of THI genes, whereas only Pdc2p is required for the expression of PDC5. Thi2p and Pdc2p serve as transcriptional activators and each factor can interact with Thi3p. The target consensus DNA sequence of Thi2p has been deduced. When TPP is not bound to Thi3p, the interactions between the regulatory factors are increased and THI gene expression is upregulated. In this study, we demonstrated that Pdc2p interacts with the upstream region of THI genes and PDC5. The association of Pdc2p or Thi2p with THI gene promoters was enhanced by thiamin starvation, suggesting that Pdc2p and Thi2p assist each other in their recruitment to the THI promoters via interaction with Thi3p. It is highly likely that, under thiamin-deprived conditions, a ternary Thi2p/Thi3p/Pdc2p complex is formed and transactivates THI genes in yeast cells. On the other hand, the association of Pdc2p with PDC5 was unaffected by thiamin. We also identified a DNA element in the upstream region of PDC5, which can bind to Pdc2p and is required for the expression of PDC5.

  • Saccharomyces cerevisiae
  • thiamin pyrophosphate
  • transcriptional regulation
  • THI gene
  • PDC5


The yeast Saccharomyces cerevisiae is able to synthesize thiamin pyrophosphate (TPP) de novo. In addition, it can efficiently utilize thiamin from the extracellular environment to produce TPP. The expression of genes involved in the synthesis of TPP and in the utilization of extracellular thiamin (THI genes) is coordinated when the supply of thiamin is limited, a mechanism called the yeast THI regulatory system (Hohmann & Meacock, 1998; Nosaka, 2006; Kowalska & Kozik, 2007). This control occurs at the transcriptional level, and TPP serves as an intracellular negative signal. Conversely, three positive regulatory factors, Thi2p, Thi3p, and Pdc2p, have been identified. Thi2p has a Zn2-Cys6 DNA-binding motif of the N-terminus in common with several yeast transcriptional activators (Titz et al., 2006). The C-terminal part of Thi2p is rich in acidic amino acids. Harbison et al. (2004) identified the elements of S. cerevisiae bound by transcriptional regulators, including Thi2p, using genome-wide chromatin immunoprecipitation technology. Several DNA sequences immunoprecipitated with an antibody specific for Thi2p were found upstream of the putative TATA box of THI genes, and one of these elements in PHO3, a THI gene which encodes a periplasmic acid phosphatase with high affinity for thiamin phosphates, had been demonstrated to be required for the induction in response to thiamin starvation (Nosaka et al., 1992). Thi3p is a TPP-binding protein whose sequence is about 50% identical to that of yeast pyruvate decarboxylase isozymes (Pdc1p, Pdc5p, and Pdc6p). As THI genes are expressed even under thiamin-replete conditions when the TPP-binding site of Thi3p is disrupted, Thi3p seems to act as a TPP sensor to exert transcriptional control (Nosaka et al., 2005). Pdc2p possesses putative DNA-binding domains similar to centromere binding protein B (Tanaka et al., 2001) and DDE superfamily endonuclease (Venclovas & Siksnys, 1995) at the N-terminus. The PDC2 gene is necessary for the expression of not only THI genes but also pyruvate decarboxylase structural genes (Hohmann, 1993). Thus, Pdc2p participates in the transcriptional regulation of TPP-synthesizing enzymes and TPP-dependent enzymes. The expression of PDC5 is also induced in response to thiamin starvation, whereas PDC1 is expressed abundantly in a thiamin-independent fashion (Muller et al., 1999). It is intriguing that Thi3p is not involved in the regulation of PDC5 in spite of being related to the intracellular level of TPP (Nosaka et al., 2005).

We have previously demonstrated that Thi3p associates with Pdc2p directly, and to a lesser extent with Thi2p, and that these interactions are partially disturbed by TPP (Nosaka et al., 2008). We also found that Pdc2p and Thi2p transactivate gene expression, which is enhanced by thiamin starvation. No such enhancement was observed in the thi3Δ strain. However, Pdc2p expressed striking transactivation activity in a Thi3p-independent fashion when the C-terminal region containing the Thi3p-interacting domain was shortened (Nosaka et al., 2008). Based on these observations, we proposed a mechanism for the transcriptional activation of THI genes mediated by Pdc2p in response to thiamin starvation as follows. When intracellular TPP is abundant and occupies the TPP-binding sites of Thi3p, the C-terminal domain of Pdc2p masks the internal domain responsible for the transactivation activity. Upon thiamin deprivation, the dissociation of TPP from Thi3p is followed by the interaction of Thi3p with the C-terminal domain of Pdc2p, which in turn causes a conformational change in Pdc2p. As a result, the C-terminal domain is removed from the transactivation domain; thus, Pdc2p can exert full transactivation activity by recruiting general transcription factors efficiently. It is likely that Pdc2p binds the upstream region of THI genes, and Mojzita & Hohmann (2006) noted that Pdc2p actually binds DNA, although the experimental data were not published.

In this paper, we demonstrated, using chromatin immunoprecipitation (ChIP) assays, that Pdc2p interacts with the upstream region of THI genes, the sequences of which are different from the target sequence of Thi2p. It was also found that Pdc2p interacts with PDC5. Interestingly, the association of Pdc2p or Thi2p with the target DNA sequences of THI genes was enhanced by thiamin starvation, whereas the association of Pdc2p with the PDC5 promoter was unaffected. Furthermore, we identified a DNA element in the upstream region of PDC5, which can bind to Pdc2p and is required for the expression of PDC5.

Materials and methods


The TA-cloning vector pGEM® T-Easy (Promega) was used to clone PDC2 gene and the PDC5 promoter isolated from yeast genomic DNA by PCR using Ex Taq™ DNA polymerase (Takara Bio, Otsu, Japan) with specific primers. The expression vectors are listed in Table 1. In general, the target sequence was PCR-amplified from the vector pGEM-PDC2 or pGEM-PDC5-promoter using specific primers into which restriction sites were designed, and the fragment obtained was digested with the restriction enzymes and subcloned into expression vectors. The PDC5 promoter-lacZ plasmids (B593ΔX series) carried an in-frame fusion between the inserted promoter-associated start codon and the lacZ coding sequence. All PCR primers are available on request.

View this table:
Table 1

Expression vectors used in this study

pYES3/CTGAL1 promoter, 2µ origin, TRP1Invitrogen
pYES3/CT-THI2THI2 with a V5 epitope at the C-terminus in pYES3/CTNosaka et al. (1998)
pYES3/CT-PDC2PDC2 with a V5 epitope at the C-terminus in pYES3/CTThis study
pYES3/CT-PDC2(1–406)Amino acids 1–406 of PDC2 with a V5 epitope at the C-terminus in pYES3/CTThis study
pYES3/CT-PDC2(407–925)Amino acids 407–925 of PDC2 with a V5 epitope at the C-terminus in pYES3/CTThis study
B593ΔXlacZ reporter, YIp type, URA3NBRP-Yeast
B593ΔX-PDC5p832 bp of PDC5 promoter-LacZ fusion in B593ΔXThis study
B593ΔX-PDC5p-truncatedTruncated promoters region of PDC5 in B593ΔX, a total of eight plasmids as in Fig. 2This study
pRSET BBacterial expression vector, T7 promoter, ampr, pUC originInvitrogen
pRSET-PDC2(1–581)Amino acids 1–581 of PDC2 with a six histidine-tag at the N-terminus in pRSET BThis study
  • NBRP-Yeast, National BioResource Project-Yeast in Japan.

Strains and media

Escherichia coli strains DH5α and BL21(DE3)pLysS were used to amplify plasmids and express the recombinant proteins, respectively. Saccharomyces cerevisiae strains YPH500 (MATα ura3-52 his3-Δ200 leu2-Δ1 trp1-Δ63 ade2-101 lys2-801), NKC18 (thi3::HIS3 in YPH500), and NKC19 (thi2::HIS3 in YPH500) (Nosaka et al., 2005) were used in this study. As the PDC2 gene is necessary for the expression of pyruvate decarboxylase genes (PDC1 and PDC5) and the pdc2Δ strain is almost inviable in glucose medium (Hohmann, 1993), a deletion of PDC2 was introduced into the yeast strain whose PDC1 gene expression was driven by the ADH1 promoter as follows. The plasmid pGAD-PDC1 was made by replacing the 0.85-kb HindIII fragment of pGAD GH (Clontech) with a PCR-amplified PDC1 open reading frame with HindIII linkers, and the 3.35-kb SphI fragment containing the ADH1 promoter-PDC1-ADH1 transcription termination sequence from pGAD-PDC1 was inserted into YIp5 at the unique SphI site. Then, the recombinant plasmid was linearized at the unique BglII site in PDC1 and transformed into YPH500. The PDC2 gene of the thus constructed strain NKC20 (LEU2::ADH1promoter-PDC1-ADH1termination in YPH500) was disrupted by a PCR-directed integration method (Baudin et al., 1993) using HIS3 as a selectable marker. The newly constructed strain NKC21 (pdc2::HIS3 in NKC20) was a thiamin auxotroph, but grew normally in glucose medium containing thiamin. The mRNA levels of PHO3, THI20, and PDC5 in NKC21 were confirmed to be entirely depressed even under thiamin-deprived conditions (data not shown). To analyze the promoter activity of PDC5, all B593ΔX-derived plasmids were linearized with StuI to target integration to the ura3-52 locus and transformed into YPH500. Single-copy integration was confirmed by restriction mapping of PCR-isolated fragments from the genomic DNA. Standard media and growth conditions for yeast cells were as described previously (Nosaka et al., 2005). Thiamin was added to the yeast minimal medium to a final concentration of 1 µM (high-thiamin medium) or 10 nM (low-thiamin medium). The concentration of the carbon source (glucose, raffinose, and galactose) was 2%.

Chromatin immunoprecipitation (ChIP) assay

Yeast cultures (50 mL) were gently shaken with 1.5 mL of 36% formaldehyde for 15 min at 30 °C, and the cross-linking reaction was stopped with 2.5 mL of 2.5 M glycine. After two washes with cold PBS, the cells were suspended in 0.6 mL of lysis buffer (50 mM HEPES pH 7.5, 140 mM NaCl, 1% Triton X-100, 1 mM EDTA, and 0.1% sodium deoxycholate) containing 1 mM phenylmethylsulfonyl fluoride and 10 µL mL−1 protease inhibitor cocktail for Fungal and Yeast cells (Sigma), and lysed with glass beads in a bead beater (Biospec Products) by beating for three 60-s pulses with 5-min intervals on ice. After the lysate was drawn off the beads, the beads were again suspended with 0.6 mL of lysis buffer to recover the extracts. Then, the combined lysate was sonicated five times in ice-cold water using a Biorupter (Cosmo Bio, Tokyo) at 200 W for 30 s each time at 120-s intervals. Sonicated extracts were subsequently clarified by centrifugation. The lysate was divided into three fractions: the first and second (500 µL each) were used for immunoprecipitation, and the third (25 µL) was used as an input control. The first lysate was incubated with 40 µL of Dynabeads® Protein A (Invitrogen) complexed with an anti-V5 antibody (Invitrogen), which was prepared according to the manufacturer's instructions, at 4 °C for 2 h on a 360° rotator. The second lysate was incubated with mock Dynabeads as a negative control. Then, the lysate was removed, and the beads were washed twice with lysis buffer, twice with 1 mL of wash buffer (100 mM Tris-HCl pH 8.0, 250 mM LiCl, 0.5% NP-40, 1 mM EDTA, and 0.5% sodium deoxycholate), and once with 1 mL of TE buffer (10 mM Tris-HCl and 1 mM EDTA, pH 8.0). The beads were incubated with 250 µL of TE plus 1% SDS for 10 min at 65 °C to elute DNA. The aspirated DNA solution was incubated at 65 °C overnight to reverse the cross-linking. The immunoprecipitated DNA and the input control DNA were treated with proteinase K, and precipitated with ethanol after proteins were removed with phenol–chloroform. Thus, purified DNA was dissolved in 30 µL of TE.

The DNA sample (1 µL) was subsequently subjected to PCR in a total volume of 20 µL using gene-specific primers (Supporting Information, Table S1). Preliminary reactions were performed to determine the optimal conditions to assure the linear amplification of each gene. In general, PCR was carried out with 28 cycles of 94 °C for 15 s, 54 °C for 15 s, and 72 °C for 10 s with Ex Taq DNA polymerase (Takara Bio). PCR products (50∼60 bp) were electrophoresed on an 8% polyacrylamide gel, stained with ethidium bromide, and photographed. The intensities of the bands in digitized images were quantified using the image j (1.42q) program, and the amounts of immunoprecipitated DNA were determined relative to the input DNA. Individual ChIP assays were repeated at least twice to confirm the reproducibility of the PCR-based experiment.

Electrophoretic mobility shift assay (EMSA)

Histidine-tagged Pdc2p(1–581) was expressed in bacterial cells and purified as described previously (Nosaka et al., 2005). The digoxigenin (DIG) gel shift kit (second generation; Roche Applied Science) was used for protein-DNA-binding assays. The oligonucleotide sequences used in this study are listed in Table S2. The double-stranded DNA probe was prepared by heating at 95 °C for 5 min and subsequent slow cooling to 65 °C. Then, the annealed fragment was isolated from a 5% polyacrylamide gel, and labeled by terminal transferase with DIG-11-ddUTP. The labeled probe (32 fmol) was incubated for 30 min at 25 °C with the recombinant Pdc2p(1–581) (2.5 µg) in 20 µL of EMSA buffer (20 mM HEPES pH 7.6, 30 mM KCl, 10 mM (NH4)2SO4, 1 mM EDTA, 1mM dithiothreitol, and 1% Tween 20) containing 1 µg of double-stranded poly(dI-dC), 1 µg of poly l-lysine, and 20 µg of BSA. The mixture was separated on an 8% polyacrylamide gel in 0.25× TBE and transferred to a nylon membrane (Biodyne B/Plus; Pall Gelman Laboratory) in 0.5× TBE using an electro-blotting system (Trans-blot SD Cell; Bio-Rad). Chemiluminescence of DIG-labeled DNA-protein complexes with anti-DIG-AP and CSPD on the nylon membranes was detected by an image analyzer (ImageQuant LAS 4000mini; GE Healthcare).

Results and discussion

Recruitment of THI regulatory factors to the target promoters

We used the ChIP assay to test whether Pdc2p associates with any of the THI genes. PHO3, THI4, and THI20 were chosen as representatives because they have consensus Thi2p recognition sites in their upstream regions (Nosaka, 2006). We also tested the interaction with PDC5, the expression of which is dependent on Pdc2p but not Thi2p (Nosaka et al., 2005). The Pdc2p with a V5-tag at the C-terminus was expressed from the GAL1 promoter in yeast cells grown in minimal medium containing 10 nM (low) thiamin, and tested for any association with upstream regions. Two or three different primer sets were initially designed for each gene, and, in the case of THI genes, one of these regions overlapped with the Thi2p-recognition site (Fig. a). The CYC1 promoter served as a negative control, as it was found not to associate with Pdc2p and Thi2p in preliminary experiments. As shown in Fig. b, the V5-tagged Pdc2p associated with all the THI genes tested. Of note, the V5-tagged Pdc2p was concentrated in regions containing a Thi2p recognition site in PHO3 and TH20 whereas in THI4 it was concentrated a modest distance from the site. In addition, ChIP assays showed an association between Pdc2p and the PDC5 promoter with the strongest signal located about 400 bp upstream from the start codon. The primer sets that exerted the strongest signal for each gene were employed for further ChIP assays.

ChIP analysis of Pdc2p and Thi2p's association with the PHO3, THI4, THI20, and PDC5 genes. Yeast strains YPH500 (wild-type), NKC19 (thi2Δ), NKC18 (thi3Δ), and NKC21 (pdc2Δ) harboring pYES3/CT derivatives were first grown in 10 mL of low-thiamin medium with raffinose to an OD600 nm of around 0.8, pelleted, and grown for 16 h in 50 mL of the same medium but with galactose instead of raffinose. The cell cultures were used for ChIP experiments as described in Materials and methods. (a) Schematic diagram indicating the positions of PCR primers used to detect immunoprecipitated DNA. Base pair positions relative to the translation initiation codon are provided. White arrows indicate the Thi2p-recognition site. (b) Association of Pdc2p with various upstream regions. Individual ChIP assays were repeated twice and the average ratio of signal intensity of IP to input is indicated as ‘Relative IP’, on the y-axis. The CYC1 promoter (position −207 to −147) is used as a negative control. (c) Effect of thiamin, thi2Δ, and thi3Δ on Pdc2p. (d) Effect of thiamin, pdc2Δ, and thi3Δ on Thi2p. In the experiments of (c) and (d), individual ChIP assays were repeated three times. Representative results of polyacrylamide electrophoresis and the average ratio of signal intensity of IP to input are shown. Bars indicate standard deviation. Statistically significant differences compared to the wild-type cells grown in low-thiamin medium are shown (*P < 0.01; **P < 0.05). The superscript ‘H’ denotes that the results were obtained from wild-type cells grown in high-thiamin medium. (e) Association of Pdc2p(1–406) and Pdc2p(407–925) with the upstream regions. Individual ChIP assays were repeated twice and representative results of polyacrylamide electrophoresis are shown with the ratio of signal intensity of IP to input. nd, not detected.

We next investigated whether the associations between Pdc2p and the promoters of THI genes and PDC5 were influenced by the thiamin concentration in the medium and the absence of Thi2p or Thi3p. We found by Western analysis using an anti-V5 antibody that Pdc2p was expressed to a similar degree under our experimental conditions (data not shown). As shown in Fig. c, when the yeast cells were grown in 1 µM (high) thiamin medium, the association with Pdc2p was decreased in PHO3 and THI20, and to a lesser extent in THI4. This result suggests that the interaction of Pdc2p with the THI gene promoters is sensitive to the intracellular TPP concentration. Furthermore, when the ChIP assay was carried out using thi2Δ and thi3Δ mutant strains grown in low-thiamin medium, the coimmunoprecipitation of Pdc2p with THI genes was markedly decreased. Given that the association of Pdc2p with THI genes was not enhanced in the absence of Thi2p, it is unlikely that Pdc2p competes with Thi2p to bind to the target DNA. Conversely, Pdc2p's association with the PDC5 promoter was unchanged by the thiamin concentration or absence of Thi2p and Thi3p. As Thi2p recognition sites exist in the promoters of THI genes and Thi3p interacts with both Pdc2p and Thi2p (Nosaka et al., 2005), it is probable that the recruitment of Pdc2p to these promoters is facilitated by Thi2p bound to its target DNA via interaction with Thi3p. Thus, not only the transactivation activity but also the recruitment of Pdc2p seems to be enhanced in response to thiamin starvation.

Next, the ChIP assay was performed with the V5-tagged Thi2p to study the possible variation in the association between Thi2p and target DNA. Of note, the primer set employed for the assay with THI4 was position C (Fig. a) different from that (position B) for the assay for V5-tagged Pdc2p, and the pdc2Δ mutant was used instead of strain thi2Δ. We confirmed that the protein level of thi2p was also unchanged by the experimental conditions (data not shown). As expected, V5-tagged Thi2p coimmunoprecipitated with all the THI genes except PDC5 (Fig. d). The association with the target DNA was also decreased by thiamin in the medium and the absence of Thi3p or, in this case, Pdc2p. These findings strongly suggest that both Pdc2p and Thi2p alone can bind target DNA and assist each other in recruitment to the THI promoters via interaction with Thi3p.

DNA-binding domain of Pdc2p

We next intended to map the DNA-binding domain in Pdc2p. Pdc2p is 925 amino acids (aa) long with an internal (aa 407–581) transactivation domain and C-terminal (aa 668–889) Thi3p-interacting domain (Nosaka et al., 2008). In addition, Pdc2p possesses putative DNA-binding domains at the N-terminus. We constructed plasmids to produce a truncated Pdc2p with a V5-tag and used them in ChIP assays. As expected from the presumed sequence, when the N-terminal 406 aa were removed, no association with THI genes and PDC5 was detected (Fig. e). Conversely, V5-tagged Pdc2p(1–406) coimmunoprecipitated with all the genes tested, although very weakly as compared with intact Pdc2p (Fig. e). In particular, its association with THI genes was markedly reduced. The elimination of the first ten N- or C-terminal aa from Pdc2p(1–406) led to abolishment of the ChIP signal (data not shown). Thus, the 1–406 aa region is necessary for Pdc2p to bind with target DNA. However, this region alone may not be adequate to exert full binding activity. Alternatively, because of a lack of the Thi3p-interacting domain (Nosaka et al., 2008), it is assumed that the recruitment of Pdc2p(1–406) to the promoter regions of THI genes was decreased.

Promoter analysis of PDC5

Meanwhile, we attempted to locate the promoter region responsible for the expression of PDC5. Although two ethanol-repression sequence (ERA) sites are recognized in the upstream region of PDC5 (Liesen et al., 1996), it is unclear whether these cis-acting elements are involved in the induction of PDC5 gene expression in response to thiamin starvation. We constructed a series of plasmids containing terminal and internal deletions of the PDC5 promoter and used the β-galactosidase activity to monitor their promoter activities. The results are summarized in Fig. 2. The LacZ gene with PDC5's upstream region truncated at position −418 from the start codon conferred almost full promoter activity. However, the upstream region truncated at position −390 showed significantly less promoter activity, and that at position −345 barely showed any activity. Furthermore, the deletion of the upstream region from −397 to −346 almost completely abolished the activity. These findings indicate that a sequence critical for PDC5 expression in yeast cells is located between nucleotides −417 and −346, which does not contain the ERA sites.

Effect of deletions in the upstream region on PDC5 expression. After the yeast strain YPH500 with the integrated B593ΔX derivatives was cultured through one round of 102-fold dilution in high- or low-thiamin medium, the cells were grown in the same medium to an OD600 nm of around 0.8 and β-galactosidase activity was determined using permeabilized cells (Nosaka et al., 2005). β-galactosidase activity was calculated as A280/reaction time (min) per cell culture (OD600 nm). Each value is the mean from two experiments. Black arrows indicate the ERA sites.

Target DNA sequence of Pdc2p

As no transcriptional regulator other than Pdc2p involved in the expression of PDC5 has been identified to date, it is possible that the sequence recognized by Pdc2p is located in the region between −418 and −346. We, therefore, carried out an EMSA to determine whether Pdc2p can bind to this restricted region. Several double-stranded oligonucleotides 30–40 bp in length were designed from the above region and used as the DNA probes (Table S2). The recombinant Pdc2p(1–581) purified as a histidine-tagged protein from E. coli cells (Fig. S1) was used in this experiment, as full-length Pdc2p and Pdc2p(1–406) were not expressed in E. coli cells. As a result, when the probe #5–1 corresponding to the region from −410 to −379 was mixed with Pdc2p(1–581), a band migrating more slowly than the free probe was detected (Fig. b). In addition, the DNA probe in which one nucleotide was deleted from the 3′- or 5′- side of #5–1 did not confer retardation (data not shown). This shifted band was depleted by competition with a 125-fold molar excess of unlabeled #5–1, suggesting that Pdc2p can specifically bind to this sequence. It is likely that this sequence acts as a cis-acting element indispensable for PDC5 expression. Furthermore, we noticed that a DNA sequence partially homologous to #5–1 was located immediately upstream of the Thi2p-recognition site of PHO3 (Fig. a). Then, to determine whether Pdc2p also recognizes this homologous sequence, several oligonucleotides were prepared for an EMSA. As shown in Fig. 3, unlabeled #3–2 (corresponding to the region from −273 to −234), and to a lesser extent #3–1 (−256 to −227), were found to partly compete with #5–1 for binding to Pdc2p(1–581). Nevertheless, no shifted bands appeared when #3–1, #3–2, and their elongated oligonucleotides were used as labeled probes (data not shown), suggesting that the interaction between Pdc2p and the PHO3 upstream region is not strong enough to be detected under our in vitro assay conditions. We have previously demonstrated that, in addition to the region from −234 to −215 containing the Thi2p-recognition site, the deletion of −273 to −245 in the PHO3 promoter causes the decrease in expression (Nosaka et al., 1992). Pdc2p can probably bind to the region from −273 to −234 and transactivate the PHO3 gene together with Thi2p, which binds the closely spaced site. Until now, we could not identify the Pdc2p-recognition site in the THI4 and THI20 promoters by EMSA using oligonucleotides with a sequence similar to #5–1 or #3–2. From these findings, we assume that Pdc2p binds the recognition sites of THI genes with low affinity, and therefore, the presence of Thi2p with Thi3p is required for the satisfactory recruitment of Pdc2p to THI promoters. On the other hand, the interaction between Pdc2p and the PDC5 promoter region is so strong as not to need another regulatory protein for the recruitment to the PDC5 promoter.

Direct association of Pdc2p with the PDC5 promoter region. (a) Schematic diagram of the PHO3 and PDC5 promoters with the sequences of oligonucleotides used in EMSA. Base pair positions relative to the translation initiation codon are provided. (b) EMSA. Protein-binding reactions to the labeled oligonucleotide #5–1 were performed using the purified Pdc2p(1–581) as described in Materials and methods. In the competition experiment, unlabeled nucleotides were added to the reaction mixture at a 125-fold more excess.

In conclusion, our study showed that the N-terminal domain of Pdc2p interacts with the upstream region of THI genes and PDC5. In the mechanism for THI gene expression mediated by Pdc2p in response to thiamin starvation, not only the transactivation activity but also the recruitment to THI promoters seems to be enhanced via interaction with Thi3p (Fig. 4). It is highly likely that under thiamin-deprived conditions, the ternary Thi2p/Thi3p/Pdc2p complex is formed and transactivates THI genes in yeast cells. Conversely, the association of Pdc2p with PDC5 was unaffected by thiamin concentration in the medium. To date, a mechanism underlying the regulation of PDC5 expression by TPP remains uncertain. As Pdc1p, a major isoform of yeast pyruvate decarboxylase, functions as a negative regulator for expression of PDC5 (Eberhardt et al., 1999), it will be interesting to investigate the relation between the TPP-binding of Pdc1p and the transcriptional control of PDC5.

Schematic representation of transcriptional regulation of THI genes and PDC5 in response to thiamin starvation. Pdc2p, Thi2p, and Thi3p are nuclear proteins, and the intracellular TPP concentration has no influence on the nuclear localization of Thi3p (Nosaka et al., 2005). Both Pdc1p and Pdc5p are also enriched in the nucleus (Mojzita & Hohmann, 2006), although the glucose fermentation occurs in the cytosol.

Supporting Information

Additional Supporting Information may be found in the online version of this article:

Fig. S1. SDS–PAGE analysis of the purified protein.

Table S1. Primers used in ChIP assay.

Table S2. Oligonucleotides used in EMSA.


This work was supported in part by a research grant from the Vitamin B Research Committee of Japan.


View Abstract