Dissection of the promoter of the HAP4 gene in S. cerevisiae unveils a complex regulatory framework of transcriptional regulation

Abstract

In S. cerevisiae, the heteromeric Hap2/3/4/5 complex is necessary for induced transcription of a large number of genes involved in oxidative metabolism on non-fermentable carbon sources. The Hap4p subunit is the activator subunit and at the same time also the regulatory part of the complex, since it is the only one whose level is regulated by carbon source itself. HAP4 promoter analysis shows a 265 bp activating region at position −1006/−741 bp upstream of the ATG start codon. Specific and differential protein-binding to a 30 nt CSRE-like sequence within this region was observed with extracts from repressing and inducing carbon sources. Carbon source-dependent activation mediated by the 265 bp fragment, as well as protein binding to the 30 nt CSRE-like region, is dependent on the presence of CAT8 function, unveiling a complex framework by which the expression of the HAP4 gene is coordinated. Copyright © 2002 John Wiley & Sons, Ltd.

Introduction

When growing on glucose, Saccharomyces cerevisiae undergoes glucose or catabolite repression, a state in which mitochondrial respiratory chain function is repressed and the steady-state mRNA levels of hundreds of genes are lowered. Previous analysis of a number of nuclear genes encoding mitochondrial proteins showed that carbon source-dependent regulation of respiration is achieved mainly at the transcriptional level (Grivell, 1995). In S. cerevisiae, this transcriptional control is mediated by the Hap2/3/4/5p complex, which also regulates a large number of nuclear genes whose products are involved in energy conservation and oxidative carbohydrate metabolism. Whereas the genes encoding the DNA binding subunits HAP2, HAP3 and HAP5 are constitutively transcribed, transcription of HAP4, coding for the activator subunit, is strongly regulated by carbon source (Forsburg and Guarente, 1989). It is repressed by glucose and induced during the diauxic shift (DeRisi et al., 1997). Transcriptional activation by the HAP complex during growth on non-fermentable carbon sources is hence dependent on the availability of Hap4p. A yeast strain overexpressing HAP4 showed partial alleviated repression of respiratory function and altered the fermentative metabolism on glucose towards a more oxidative metabolism (Blom et al., 2000). This confirms the view that during growth on glucose the level of Hap4p becomes limiting, leading to downregulation of the target genes, and shows the pivotal role Hap4p plays in the balance between oxidative and fermentative metabolism. For a full understanding of how this delicate balance between respiration and fermentation can be affected by Hap4p, it is important to know how transcription of HAP4 responds to changes in environmental conditions and, moreover, which signalling pathways are involved in regulating this response. What is known thus far is the presence of a Mig1p binding site located around 350 bp upstream of the translation initiation codon. In vitro studies have shown that Mig1p is able to bind this sequence (Lundin et al., 1994), but in vivo evidence for a functional Mig1p binding site is lacking. In a mig1 disruption strain, the expression of HAP4 is not derepressed on glucose, but this could be due to the absence or inactivity of specific activators (Klein et al., 1999) and does not rule out the possibility that glucose repression of the HAP4 gene is, at least partially, mediated by Mig1p. In order to obtain more insight in elements involved in the transcriptional regulation of HAP4, we carried out a detailed analysis of the HAP4 promoter. Parts of the promoter were fused to an Escherichia coli LacZ reporter gene, and assayed for the ability to induce the reporter gene on different carbon sources. The binding of S. cerevisiae proteins to regions identified to be significant in the promoter analysis was also studied. This led to the conclusion that the HAP4 gene possesses a complex promoter which may harbour binding sites for one or more synergistically acting transcription factors that link the expression of the gene into other regulatory circuits controlling C-source responses in yeast.

Materials and methods

Strains and media

The yeast strains used in this work are listed in Table 1. Rich medium was 1% (w/v) yeast extract and 2% bacto-peptone (YP), selective medium contained 0.67% yeast nitrogen base without amino acids (Difco) supplemented with the appropriate amino acids and bases. Carbon sources were added at 2% (w/v) glucose (D), 2% (v/v) ethanol plus 2% (v/v) glycerol (EG), 2% galactose or 2% lactate. Media were solidified with 2% (w/v) agar.

Table 1. Yeast strains used in this study

Strain	Genotype	Source
MC999-1A	MATa,ura3-52, his3Δ1, leu2-3, 112trp1	(Proft et al., 1995)
WAY.5-4A	MATa, ura3-52, his3-Δ1	(Hedges et al., 1995)
DG2	MATa, ura3-52, cat8::HIS3	(Hedges et al., 1995)

Recombinant DNA procedures

All DNA manipulations (restriction enzyme digestions, ligations) and E. coli transformations were performed according to standard procedures and recommended by the manufacturer. Sequencing was performed using the dideoxy method (Sanger et al., 1977) with M13 reverse primer (Promega) and PC1 (5′-GTGTGTGTATTTGTG-3′) within the cyc1 minimal promoter in pCZ.

Plasmids and plasmid construction

For the cloning of subfragments of the HAP4 promoter in front of a minimal CYC1 promoter fused to LacZ, restriction endonucleases were employed as indicated in Figure 1. A SmaI–SmaI fragment from −1006 bp to −741 bp was isolated from the HAP4 promoter and cloned into the dephosphorylated SmaI site of pUC18. Digestion of the SspI–SspI HAP4 promoter fragment (−1224 to +79) by Sau3AI yielded many fragments, and, amongst them, one from −560 bp to −420 bp and one from −420 bp to −160 bp were cloned into the BamHI site of pUC18. After sequence verification, an EcoRI–SalI fragment from the pUC18/SmaI fragment was inserted into an EcoRI–XhoI-digested pCZ plasmid (kindly provided by E. Boy-Marcotte). The Sau3AI fragments in pUC18 were cloned as EcoRI–SphI fragments in pCZ digested with EcoRI–SphI. For cloning of the 30 nt region containing a possible UAS, two complementary oligonucleotides (MWG186 + MWG187, see Table 2) containing the sequence were hybridized by heating for 2 min and subsequently slowly cooled to room temperature. After digestion with EcoRI, the double-stranded oligonucleotide was cloned into the EcoRI site of pCZ and sequenced, yielding pCZ–UAS. The 200 bp subfragment of the SmaI–SmaI region and the 170 bp subfragment lacking the 30 nt region (Figure 1) were obtained as PCR products using different combinations of primers (Table 2) and, after digestion, inserted in the EcoRI–SphI site of pCZ/UAS. All constructs were verified by sequence analysis.

Figure 1. Schematic overview of part of the HAP4 promoter. The restriction endonucleases used for subcloning are indicated. Sm, SmaI, sa, Sau3AI, ss, SspI. Numbering is relative to the ATG

Table 2. Synthetic oligonucleotides used in this study

Primer	Sequence (5′–3′)
Lower case are nucleotides which were added for convenient subcloning. Underlined sequence corresponds to the CSRE-like region.
MWG191	ccgc gcatgc CCG AGC TAC AAA TC
MWG192	ccgc gcatgc TTG ATT CGG TCG TGC
MV5	ggaattcc GGG TTT CTC TGA AGC
MWG186	gaattc GGGAA CAACG GAGGA ATAGA GCGAC AAG gaattc
MWG187	gaattc CTT GTCGC TCTAT TCCTC CGTTG TTCCC gaattc

Enzyme assays and crude protein preparation

For assay of LacZ activity, yeast cells transformed with pCZ-derived plasmids were pre-grown overnight at 30° C in 5 ml selective media containing 2% glucose. Overnight cultures were diluted into 20 ml fresh media, and grown overnight to an OD_600nm of 1.5. Cells were centrifuged for 5 min at 5000 × g at 20° C, and resuspended in 1 ml sterile water; 250 µl and 500 µl aliquots were transferred to 40 ml media containing 5% glucose and 2% EG, respectively. Cultures were grown for an additional 5 h, after which cells were harvested by centrifugation for 5 min at 5000 × g, 4° C. Cells were washed with ice-cold sterile water, resuspended in 500 µl extraction buffer (160 mM Tris, pH 8.0; 400 mM (NH₄)₂SO₄; 10 mM MgCl₂; 1 mM EDTA; 10% glycerol; 1 mM PMSF; 7 mM ß-mercaptoethanol) and stored at −70° C until protein lysate isolation. For the isolation of crude lysates, cells were thawed, pelleted and suspended in 160 µl ice-cold extraction buffer with 1 mM PMSF and 7 mM ß-mercaptoethanol. An equal volume of 0.5 mm diameter sterile glass beads was added and the cells were disrupted by vortexing at 4° C for 15–20 min. Samples were centrifuged at 4° C for 30 min, and the supernatant was used for enzyme assay. The protein concentration was determined according to the method of Lowry (1951). For measurement of LacZ activity, 10–20 µg of protein lysate was diluted in a final volume of 750 µl extraction buffer, 250 µl substrate (ortho-nitro phenyl ß-D-galactopyranoside, Sigma; 4 mg/ml in extraction buffer) was added and incubated at 37° C. Absorption at 415 nm was measured.

Gel mobility shift assay

Electrophoretic mobility shift assays were performed using 4–8% polyacrylamide (29 : 1) gels with 0.5 × TBE buffer. Binding reactions were set up with 10–30 µg crude extract in gel shift binding buffer (4 mM Tris–HCl, pH 8.0, 40 mM NaCl, 4 mM MgCl₂, 4% glycerol and 20 ng BFB) containing 500 ng poly(dI-dC). Reactions were incubated for 5 min at RT. Approximately 50 fmol (100 cps) of probe was added to the mixture and incubated in a total volume of 20 µl at room temperature for an additional 20 min. Samples were loaded on acrylamide gel and run at 4° C at 15 mA constant current. For competition studies, cold competitor DNA was added and incubated for 5 min prior to addition of labelled DNA. PCR products were digested with EcoRI and end-labelled using γ³²P-dATP and Klenow enzyme. The double-stranded oligonucleotides used as probe in the gel shift analysis were prepared by annealing the single-stranded complementary oligonucleotides MWG186 and MWG187 (Table 2), subsequent labelling with T4 polynucleotide kinase and γ³²P-dATP. Probes were isolated from 8% polyacrylamide gel.

Partial protein purification

Total cell lysate was prepared from 3 l strain MC999, exponentially growing in YP medium containing 2% ethanol/glycerol. Cells were washed once with 100 ml H₂O, resuspended in 2 ml/g wet weight DTT buffer (100 mM Tris–SO₄, pH 9.4; 10 mM DTT) and incubated at 30° C for 20 min. After washing in 1.2 M sorbitol, cells were resuspended in spheroplasting buffer (1.2 M sorbitol; 20 mM potassium phosphate, pH 7.0), 7 ml/g (wet weight) of cells. Zymolyase (2 mg/g wet weight) was added and the preparation was incubated at 30° C until spheroplasts formed (about 2 h). Spheroplasts were washed with 1.2 M sorbitol, resuspended in 7 ml/g wet weight breaking buffer (0.6 M sorbitol, 10 mM Tris, pH 7.4, 1 mM EDTA, 1 mM PMSF) and lysed with 10 strokes of a Dounce homogenizer. After centrifugation at 10 000 × g for 15 min, supernatant was collected and dialysed overnight at 4° C against buffer A (50 mM KCl, 20 mM Tris–HCl, pH 7.5, 0.1 mM EDTA, 10% glycerol; 1 mM PMSF, 1 mM ß-mercaptoethanol). The dialysed supernatant was applied to a DEAE sepharose CL-6B column (Pharmacia Biotech), which had been equilibrated with buffer A, and proteins were eluted using a step gradient of 100–200–300–400 and 500 mM KCl in buffer A. Fractions were dialysed overnight at 4° C against buffer A with 20 mM KCl, concentrated and stored at −70° C. Protein concentrations were determined by the method of Lowry (1951).

DNA–protein crosslinking by UV light

DNA probe (50 fmol) was incubated with 5 µg of the 100 and 300 mM protein fractions eluted from DEAE-sepharose under the optimal conditions for binding as described for gel mobility shift experiments. After 25 min at room temperature, the reaction mixtures were irradiated 30 cm below a UV light (256 nm) (UV dose = 180 erg/mm²/s or 1.8 J) for 40 min at 4° C in open Petri dishes. The cross-linked protein–DNA complexes were analysed by electrophoresis on 6% SDS–polyacrylamide gels. The gels were stained, dried and exposed to a PhosphorImager screen.

Results

An activating region is located between −741 bp and −1006 bp

The HAP4 gene is preceded by an unusually long intergenic region of around 2400 bp, with three transcription start sites, two mapped around −280 bp and one at −330 bp (Forsburg and Guarente, 1989). In order to map regions involved in the carbon source-dependent transcriptional regulation, subfragments of the promoter (see Figure 1) were cloned in front of a minimal CYC1 promoter. This contained the necessary elements for basal transcription but lacked activating sequences and was fused to the LacZ reporter gene (Guarente and Ptashne, 1981). Subsequently, the activity of the reporter gene was monitored in ß-galactosidase assays. The level of TATA-mediated gene expression of the CYC1–LacZ reporter was not detectable on either glucose and lactate after insertion of two Sau3AI fragments ranging from −420 bp to −560 bp and from −160 bp to −420 bp, the latter region containing the Mig1p consensus site. In more sensitive measurements, in which LacZ reporter activity was quantitated by means of a bioluminescence assay using Galacton (Clontech) as a substrate, very low, although not significantly induced, levels of ß-galactosidase activity could be measured in the two constructs on both glucose and lactate (data not shown). In contrast, the SmaI fragment from −741 bp to −1006 bp showed a low level of reporter gene expression on glucose, and this expression was more than 10-fold induced on lactate (Figure 2). Interestingly, the activating capacity of the SmaI fragment was lost in extracts from a cat8 mutant, affected in the derepression of gluconeogenesis (Figure 2). The absence of reporter gene expression in the cat8 mutant grown on derepressing carbon sources cannot be ascribed to the poor growth of the cells under these conditions, since expression of the reporter gene driven by two control promoter genes (ADE1 and ASN2) is not affected (data not shown). The results suggest that Cat8p is involved in the signalling pathway that controls HAP4 transcriptional regulation and, furthermore, indicate that major activating capacity resides within the 265 bp fragment.

Figure 2. Cat8p-dependent activating capacity. ß-galactosidase activity in wild-type (WAY.5-4A) and isogenic cat8 (DG2) strains were transformed with pCZ containing the SmaI HAP4 promoter fragment (−745/−1006) (left panel) and the 30 nt UAS sequence (right panel). Data are obtained from at least two independent cat8 and wt transformants; SD < 15%. Dark bars, glucose-grown cells; light bars, lactate-grown cells

Carbon-dependent activation by a 30 nt CSRE-like region

Closer analysis of the 265 bp SmaI fragment revealed a 30 nt region containing the sequence CGGN₆AGA, closely resembling a consensus sequence for a carbon source-responsive element (CSRE; CGGN₆GGA/C; Bojunga and Entian, 1999), involved in derepression of gluconeogenic genes. Deletion of 65 bp upstream of this 30 nt region (200 bp fragment in Figure 1) did not result in a significant decrease of reporter activity compared to the complete 265 bp region (data not shown). In contrast, a substantial decrease in both transcript levels on glucose and ethanol/glycerol was observed upon deletion of the CSRE-like region (170 bp region in Figure 1) (Figure 3 A). In order to determine the functional relevance of the 30 nt element, an oligonucleotide corresponding to the 30 bp sequence was cloned in front of the CYC1–LacZ fusion and assayed for the ability to confer carbon source regulation. The difference between repressed levels on glucose, derepressed levels on galactose and induced levels on lactate is quite distinctive (Figure 3 B), implying that a specific activating element is located in this region. Our interest in this region increased further when activation by the 30 nt region was studied in a Δcat8 background (Figure 3 C). In the disruption mutant induction of the CYC1–LacZ fusion on lactate was drastically reduced, suggesting that Cat8p might exert a regulatory function via the 30 nt region.

Figure 3. Activating capacity of the 30 nt region. (A) Relative ß-galactosidase activity under control of the 200 bp HAP4 promoter fragment and effect of deletion of the 30 nt activating region. Data are obtained from two to four transformants of wt strain MC999. Activity is relative to the absolute LacZ activity on glucose (24.4 U/mg). (B) Effect of different carbon sources (glucose, galactose and lactate) on relative ß-galactosidase activity in three independent transformants of wt strain MC999 transformed with pCZ harbouring the 30 nt activating region of the HAP4 promoter in front of the CYC1–LacZ fusion. Activity is relative to the absolute LacZ activity on glucose (0.18 U/mg). Dark bars, glucose-grown cells; light bars, lactate-grown cells. SD < 10%

Carbon-specific protein factors interacting with the 30 nt activating region

To determine whether proteins bind to the 30 nt activating region, a synthetic oligonucleotide (MWG186/187) corresponding to the 30 nt sequence was used in gel mobility shift assays. With protein extracts prepared from lactate-grown wild-type cells, three specific protein–DNA complexes could be observed, two major complexes shown as bands IV and III and one low-abundant complex indicated as I (Figure 4, lane 3). Addition of more protein to the incubation mixture did not result in a different binding pattern (not shown), illustrating that binding conditions are saturating and that formation of the complexes observed is not mutually exclusive. Specificity of the complexes was demonstrated by competition with a 50 molar excess of unlabelled MWG186/187 (Figure 4, lane 4), whereas addition of a 50 molar excess of unrelated oligonucleotide could not compete for binding (Figure 4, lane 5). Interestingly, band III was less abundant when protein extracts from glucose-grown cells were used (Figure 4, lane 2), suggesting a specific function of the protein factor responsible for complex III formation in gene activation mediated by the 30 nt region under non-fermentative conditions. Since reporter gene induction by this region did not occur in the Δcat8 strain, protein extracts prepared from repressed as well as induced Δcat8 cells were used in gel shift mobility experiments. As shown in Figure 4, deletion of CAT8 did not affect protein binding in glucose-grown cells (Figure 4, lane 6). However, under inducing conditions, lysates from Δcat8 cells form another complex with intermediate mobility (II, Figure 4, lane 7). Binding to the 30 nt region could not be competed by excess of an oligonucleotide containing the CSRE of the ACS1 promoter (Kratzer and Schueller, 1997), implying that the protein binding to the 30 nt region is not the same as that binding to CSRE_(ACS1).

Figure 4. Gel mobility shift experiment using the 30 nt region. Crude lysates were prepared and binding conditions set up as described in the Materials and methods section. Lane 1, no protein extract added; lane 2, protein extract from glucose-grown wild-type cells (WAY5-4A); lanes 3–5, protein extract from EG-grown cells; lane 4, competition with 50-fold molar excess unlabelled 30 nt oligo; lane 5, competition with 50 molar excess unrelated oligo; lane 6, protein extract from Δcat8 cells grown on glucose; lane 7, EG-grown Δcat8 cells. Asterisk indicates non-specific band

Characterization of the protein binding the 30 nt region

In order to characterize the protein(s) binding to the 30 nt region, crude protein lysate was fractionated using a DEAE-sepharose column (Figure 5 A). Binding activity was localized in a fraction eluted at 100 mM KCl (indicated by III in Figure 5 B, lane 3) and in a fraction eluted at 300 mM KCl (I, II and IV in Figure 5 B, lane 5). In comparison to the low abundance of bands I, II and III observed with the same amount (1 µg) of total yeast lysate (lane 2), considerable enrichment of the binding protein(s) has been obtained. No other fractions were able to bind the 30 nt region. The specificity of the complexes formed was demonstrated by competitive displacement with excess of unlabelled 30 nt oligonucleotide and unrelated oligonucleotide. As shown in Figure 5 C, the complex indicated with III was formed specifically in the 100 mM fraction. Likewise, the formation of complex I and II in the 300 mM fraction was specific, whereas complex IV turned out to be non-specific, as it was competitively displaced by both unlabelled 30 nt and unrelated oligonucleotides. To establish the identity of the factor(s) binding to the 30 nt region, a UV-crosslinking experiment was conducted. Parallel irradiated and non-irradiated samples were analysed by native gel electrophoresis and SDS–PAGE (Figure 6 A B). Under denaturing conditions, complexes that are not cross-linked dissociate. Use of the fraction eluted at 100 mM KCl from the DEAE-sepharose column did not result in the cross-linking of any detectable protein binding to the 30 nt region on the SDS–PAGE gel, while under native conditions binding was established (Figure 6 A, B, lanes 5 and 6). With the 300 mM fraction, no radioactive signal on SDS–PAGE was detected without irradiation (Figure 6 B, lane 2), while under native conditions the complexes I and II, as described above, were clearly formed, migrating as large molecular weight complexes above 250 kDa (Figure 6 A, lane 2). Incubation with 300 mM fraction followed by UV irradiation showed two major labelled complexes at around 190 and 230 kDa under denaturing conditions (Figure 6 B, lane 3), while irradiation did not affect the formation of the complexes under native conditions (Figure 6 A, lane 3). The two complexes detected were competitively displaced upon addition of 100 molar excess of unlabelled oligonucleotide followed by UV irradiation (Figure 6 A, B, lanes 4). Once more, this shows that proteins bind specifically to carbon source regulatory elements within the 30 nt region.

Figure 5. Purification of the 30 nt binding protein from crude yeast lysate from EG-grown wild-type cells MC999. (A) Silver staining of different fractions eluted from the DEAE column; tot, total lysate; M, rainbow molecular weight marker (kDa) (Amersham Biotech). (B) Gel mobility shift assay with no lysate (-), 1 µg of total yeast lysate (tot) and 1 µg of the different fractions eluted from the DEAE column: 100 mM, 200 mM, 300 mM, 400 mM and 500 mM KCl. (C) Gel mobility competition experiment with fractions of 100 mM (lanes 2–4) and 300 mM (lanes 5 and 6) with 100 molar excess of unlabelled oligonucleotide (lanes 3, 6) or unrelated oligonucleotide (lanes 4, 7)

Figure 6. DNA–protein cross-linking by UV light. Upon incubation with radiolabelled 30 nt oligonucleotide and 300 mM or 100 mM eluted fraction, samples were irradiated, and half of the mixture was loaded on a 6% native polyacrylamide gel (A) or on a 6% denaturing SDS polyacrylamide gel (B). Lanes 1, radiolabelled 30 nt without protein; 30 nt with 5 µg 300 mM fraction without (lanes 2) or with irradiation (lanes 3) and excess of unlabelled 30 nt oligonucleotide (lanes 4); lanes 5 and 6, 30 nt oligonucleotide with 5 µg 100 mM fraction without (lanes 5) and with irradiation (lanes 6). To estimate the sizes of the complexes, the rainbow molecular weight marker (kDa) (Amersham Biotech) was used

Discussion

The transcriptional activator complex Hap2/3/4/5 plays a key role in the upshift of genes involved in oxidative metabolism in S. cerevisiae. The gene encoding the Hap4p activator subunit is the only one of which the transcription is regulated in response to available carbon source, and this regulation is the main determinant for the activity of the complex. The present study describes the characterization of the unusually long promoter region of the HAP4 gene as a first step in the unravelling of the mechanisms of transcriptional regulation of HAP4 gene expression. Initially, we generated 5′-HAP4 promoter deletion fused to a minimal CYC1–LacZ reporter and integrated the constructs in the ura site of the genome. However, we were not able to detect effects of 5′-HAP4 promoter deletions in a chromosomal context as there was a problem of insufficient sensitivity. To compensate for this by virtue of increased expression levels, we therefore made use of subfragments of the HAP4 promoter cloned in front of the minimal CYC1–LacZ reporter on episomal vectors, although we are aware that this is unlikely to reflect the complete complexity of expression control in the chromosomal context. This revealed a carbon source-dependent activation by the 265 bp fragment positioned from −741 bp to −1006 bp. The apparently long distance is not unlikely, taking into account that the transcription start site is located more than 250 bp upstream of the ATG (Forsburg and Guarente, 1989). Two other fragments downstream of the 265 bp activating region did not lead to a carbon source-dependent activation of the reporter gene. Based on the presence of a sequence closely resembling a CSRE consensus, a 30 nt region within the 265 bp sequence was selected to test its activating capacity. Indeed, this 30 nt region was able to confer carbon source-dependent regulation in a Cat8p-dependent manner. Deletion of the 30 nt sequence caused a significant decrease in reporter gene activity on both glucose and ethanol/glycerol, although the induction ratio was not affected. This suggests that other sequences within the 265 bp fragment might contribute to carbon-dependent regulation as well. Preliminar results further exploring the 265 bp activating fragment imply that several subfragments of the 265 bp fragment are indeed able to induce transcription in a carbon-dependent manner. Identification of the proteins binding to the 30 nt region has become nearer at hand with the partial purification of protein complexes migrating as high molecular weight structures at 190 and 230 kDa when cross-linked with the 30 nt region. Unfortunately, further purification of the 300 mM KCl elution fraction and attempts to identify possible candidate proteins using mass spectrometry have been unsuccessful until now.

Intriguingly, the carbon source-dependent activation by both the 265 bp region and the 30 nt region within this region, was dependent on the presence of a functional Cat8p, as demonstrated by both a drastically reduced level of reporter gene induction and slightly altered DNA–protein binding in the absence of Cat8p. Cat8p has been reported not only to control expression of genes encoding enzymes of gluconeogenesis (Bojunga et al., 1998; Hedges et al., 1995; Kratzer and Schueller, 1997; Rahner et al., 1999; RandezGil et al., 1997), but also of IDP2, encoding NADP-dependent cytosolic isocitrate dehydrogenase (Bojunga and Entian, 1999) that has no direct role in gluconeogenesis. The role of Cat8p may be a more general one, involving activation of other genes that are strongly derepressed under non-fermentative growth conditions. In this light, it is interesting that HAP4 is also strongly induced when glucose becomes limiting during the diauxic shift (DeRisi et al., 1997), thereby inducing transcription of genes involved in respiration. Idp2p has been suggested to play a defence role in the increase in respiration and concomitant formation of reactive oxygen species (Bojunga and Entian, 1999), in addition to its role in providing reducing equivalents (Minard and McAlister-Henn, 1999). In this respect, transcriptional regulation by Cat8p may thus be an efficient way of coordinating the increase in glyoxylate and gluconeogenic gene transcription with increase in respiratory function and defence mechanisms against reactive oxygen species. Proft et al. (1995) suggest that Cat8p activation might globally affect respiratory metabolism based on the observation that some cat8 mutant alleles showed a loss of cytochrome c oxidase and O₂ uptake activity. This is consistent with our findings that activation of HAP4 expression is dependent on Cat8p and recent experiments showing that Hap4p affects mitochondrial biogenesis as a whole (Lascaris et al., 2002, submitted for publication). However, Cat8p is not the sole determinant of carbon source-dependent transcriptional induction of HAP4, as deletion of CAT8 has no detectable effect on the steady-state transcriptional levels of HAP4 under control of its full-length 5′ promoter region. This has been demonstrated by Northern analysis (data not shown), as well as in a genome-wide expression study for targets of Cat8p (Haurie et al., 2001). A possible explanation might be that the role of Cat8p in transcriptional activation of HAP4 could be transient, or redundant to other factors that regulate transcription of HAP4. The fact that protein binding to the 30 nt region could not be competed by the CSRE of ACS1, a target of Cat8p, is an additional indication that the influence of Cat8p is not direct, but mediated via other factors. Based on the results described in this study, it can be concluded that we have just started to lift a corner of the veil that covers a complex regulation of HAP4 gene expression in S. cerevisiae. It might well be possible that the HAP4 promoter harbours additional regulatory features that would fine-tune its expression to regulate the balance between fermentative and oxidative metabolism in yeast.