Characterization of an upstream activation sequence and two Rox1p-responsive sites controlling the induction of the yeast HEM13 gene by oxygen and heme deficiency.

The Saccharomyces cerevisiae HEM13 gene codes for coproporphyrinogen oxidase, an oxygen-requiring enzyme catalyzing the sixth step of heme biosynthesis. Its transcription has been shown to be induced 40-50-fold in response to oxygen or heme deficiency, in part through relief of repression exerted by Rox1p and in part by activation mediated by an upstream activation sequence (UAS). This report describes an analysis of HEM13 UAS and of the Rox1p-responsive sites by electrophoretic mobility shift assays, DNase I footprinting, and mutational mapping. HEM13 UAS is composed of two subelements: a 16-base pair sequence binding a constitutive factor acting as a transcriptional activator, and a 5′-flanking 20-base pair GC-rich region. Both subelements were required additively for transcription, but each element alone was sufficient for almost normal control by oxygen/heme deficiency. Mutations in both elements decreased the induction ratio 3-4-fold. HEM13 UAS conferred a 2-4-fold oxygen/heme control on a heterologous reporter gene. Two Rox1p-responsive sites, R1 and R3, were identified, which accounted for the 6-7-fold repression by Rox1p. A factor bound to a sequence close to site R3. This DNA-binding activity was only detected in protein extracts of aerobic heme-sufficient ROX1 TUP1 cells, suggesting a possible role in site R3 function.

Genetic adaptive responses to hypoxic stress occur in many biological systems. Several systems for controlling these responses have been described in prokaryotes and eukaryotes. They involve the sensing of oxygen tension and the signaltransduction pathway leading to an alteration of gene expression. In bacteria, the sensor-regulator systems FixL, Fnr, and ArcB/ArcA are converted into active forms by oxygen deprivation (1)(2)(3)(4)(5). The rhizobial oxygen sensor FixL is a hemoprotein kinase whose activity is blocked by oxygen binding to the heme (1). In mammalian cells, a DNA-binding factor HIF-1 induced by hypoxia plays a central role in hypoxic gene activation, and there is some evidence that oxygen tension is sensed by a hemoprotein (Ref. 6 and references therein).
In the yeast Saccharomyces cerevisiae, oxygen control of gene expression is mediated essentially by two factors, which are active or synthesized only under aerobic conditions, the activator Cyp1p (Hap1) and the repressor Rox1p (reviewed in Refs. 7,8). Heme is believed to serve as an effector molecule; its biosynthesis has an absolute requirement for oxygen (9), and heme-deprivation appears to mimic the effects of oxygen deficiency in most of the cases tested (7,8); but the mechanism by which heme is implicated is not yet clear. Rox1p is a protein with a high mobility group motif that bends DNA when binding to a specific hypoxic consensus sequence and requires the general repressor complex Ssn6p⅐Tup1p to repress transcription of the target genes (10,11). The expression of the ROX1 gene is oxygen/heme-dependent via a complex regulatory network that includes the transcriptional activator Cyp1p, whose activity is heme-dependent (12)(13)(14)(15). The Rox1p concentration is therefore greatly reduced in the absence of oxygen or heme, and hypoxic gene expression is derepressed. The regulation by oxygen of the hypoxic genes studied so far in some detail, such as ANB1 (TIF51B) (isoform of translation initiation factor eIF-5A) (16,17), COX5b (isoform of cytochrome oxidase subunit Vb) (18,19), AAC3 (isoform of ADP/ATP translocator) (20), and ERG11 (lanosterol 14␣-demethylase) (21), is entirely accounted for by the repression by Rox1p in aerobiosis.
The HEM13 gene encodes the enzyme coproporphyrinogen oxidase that catalyzes the sixth step in the heme biosynthetic pathway. The enzyme uses only oxygen as an electron acceptor for the oxidative decarboxylation of coproporphyrinogen (9). HEM13 is the only gene in the pathway that is closely regulated (8,9); its transcription is induced 40 -50-fold under conditions of oxygen or heme deficiency (22,23). Previous work has shown that coproporphyrinogen oxidase activity becomes ratelimiting for heme production when its substrate oxygen is limiting; the cells respond to oxygen limitation by increasing the amount of the enzyme (23). Hence, coproporphyrinogen oxidase may play a crucial role in linking heme synthesis to the oxygen/heme-dependent control of gene expression, and it is important to understand how the HEM13 gene is regulated. Cyp1p has been found to interfere with this regulation (24), mainly by its role on the synthesis of Rox1p that represses HEM13 under aerobic heme-sufficient conditions (12,13). But the repression of HEM13 by Rox1p is only partial, and a preliminary deletion-mutant analysis of the HEM13 promoter has revealed the presence of a region acting as an upstream activation sequence (UAS) 1 that is required for full induction under oxygen/heme deprivation (15). Such a UAS that responds positively to the absence of oxygen/heme is a novel regulatory element for yeast genes regulated by oxygen/heme. The present study describes the cis-and trans-elements of this UAS by directed mutagenesis and electrophoretic mobility shift assays. We have also identified the Rox1p-responsive sites and an oxygen/heme-regulated protein factor binding close to one of them.
DNA Manipulations and Site-directed Mutagenesis-Enzymes were obtained from New England BioLabs, Life Technologies, Inc., Boehringer Mannheim, and Appligene. Routine DNA manipulations, including the polymerase chain reaction (PCR), followed standard procedures (29). The Escherichia coli strain DH5␣ was used for cloning and propagating plasmids. Mutagenesis was carried out using the Altered Sites in vitro Mutagenesis System (Promega). The single-stranded DNA template was prepared from phagemid pALTER-1 (Promega) containing the 1.18-kb EcoRI-BamHI HEM13 fragment from pH13Z. The oligonucleotides used for mutagenesis, PCR, and sequencing were synthesized on an Applied Biosystem DNA Synthesizer at the Service de Synthèse d'Oligonucleotides, Institut Jacques Monod, and purified by polyacrylamide gel electrophoresis. All mutations were verified by sequencing on double-stranded plasmids using the Sequenase kit (U. S. Biochemical Corp.), HEM13-specific oligonucleotide primers, and [ 35 S]dATP.
Plasmids and ␤-Galactosidase Assay-Plasmid pH13Z ( Fig. 1), containing the HEM13-lacZ fusion, was obtained by ligating 1076 nucleotides of the 5Ј-untranslated region and the first 34 codons of HEM13 to the lacZ gene of the episomal plasmid YEp357 (30). The mutations generated in the HEM13 promoter were introduced into plasmid pH13Z by exchanging the 1.18-kb EcoRI-BamHI fragment, to yield plasmids pH13Z/m8 to pH13Z/m21. Plasmid pH13Z/m8 -10 was obtained by ligating the 0.74-kb EcoRI-SacII fragment of pH13Z/m10 into pH13Z/m8 cut with EcoRI and SacII. Similarly, pH13Z/m11-18, pH13Z/m14 -18, pH13Z/m11-19, and pH13Z/m14 -19 were obtained by exchanging the 0.74-kb EcoRI-ApaI fragment from pH13Z/m11 and pH13Z/m14 into pH13Z/m18 and pH13Z/m19. Plasmids pSX-H13Z and pSX-H13Z/m14 -18 ( Fig. 5) were constructed as follows. A 75-bp DNA fragment (fragment 4 in Fig. 1) corresponding to the region Ϫ384 to Ϫ310 was PCR-amplified using the two oligonucleotides 5Ј-CGCGGATCCGAAAGAGTAAAAGAAATCTTT and 5Ј-CCGGAATTCGGCTTCTGTTCCTGCGA, which carried sequences for BamHI and EcoRI restriction sites (underlined), and pH13Z or pH13Z/m14 -18 as templates. The amplified fragments cut with BamHI and EcoRI were first cloned into pOV10 (31). They were recovered by cutting with BamHI (blunt-ended) and XhoI and then ligated into the left-most SmaI and right-most XhoI sites of the CYC1-lacZ fusion plasmid pLG669Z (32), in place of the CYC1 and ANB1 gene regulatory sequences. Plasmid pLG669Z, from which the 1.53-kb SmaI-XhoI fragment had been deleted, pSX, served as control. These plasmids were transformed into the different yeast strains, and ␤-galactosidase activity was measured on SDS/chloroform-permeabilized cells Probes for DNA Binding Assays and Footprinting-The 93-bp fragment SfcI (blunt-ended)-SacI of the HEM13 promoter ( Fig. 1) was inserted into plasmid pUC19 digested with SmaI and SacI. It was recovered by cutting with PstI and EcoRI and labeled at the EcoRI site with Klenow enzyme and [␣ 32 P]dATP by the standard method (29). Similarly, the 190-bp SacI-Eco47III probe was prepared by inserting the SacI-HindIII fragment into pUC19, retrieving it by cutting with EcoRI and Eco47III, and labeling at the EcoRI site with Klenow enzyme and [␣-32 P]dATP. The 78-bp probe 1 (Ϫ549 to Ϫ472), the 60-bp probe 2 (Ϫ500 to Ϫ441), and the 29-bp probe 3 (Ϫ500 to Ϫ472) ( Protein Extract and Electrophoretic Mobility Shift Assay-Protein extracts were prepared from yeast cells (33), and the final protein pellet was suspended in 20 mM Tris-HCl, pH 8, containing 0.1 mM EDTA, 1 mM dithiothreitol, 20% glycerol, and 1 mM phenylmethylsulfonyl fluoride. Protein concentrations were determined by the method of Lowry using bovine serum albumin as a standard.
Indirect DNase I Footprinting Assays-Large-scale EMSA reactions were performed as above. A total of eight identical assays were prepared using, for each, 4 ng (about 120,000 cpm) of DNA probe, 60 g of protein, and 3 g of poly(dI-dC) in a final volume of 40 l. After 15 min incubation, 1 l of 200 mM MgCl 2 , 100 mM CaCl 2 plus 1 l of DNase I (0.2 g) were added to each sample. Incubation was continued for a further 2 min at 22°C and stopped by placing on ice and adding 0.1 volume of 0.5 M EDTA. The DNA-protein complexes were resolved by electrophoresis as described above. The positions of free and proteinbound DNA were determined by autoradiography; the corresponding bands were excised, and DNA was isolated by electroelution. DNA was then precipitated with ethanol, heated at 90°C for 3 min in denaturing buffer, chilled on ice, and loaded onto a 6% polyacrylamide sequencing gel along with the products of DNA sequencing reactions of the probe performed by the method of Maxam and Gilbert (as described in Ref. 29).

RESULTS
Binding of a Factor to the HEM13 UAS Element-Our deletion mutant analysis of the HEM13 promoter showed that the 190-bp SacI-Eco47III DNA fragment contains a UAS that is required for basal level expression of the HEM13 gene and its full induction by oxygen and heme deficiency (15). To locate the cis-acting element(s) within this DNA fragment and identify any putative factor(s) that may bind to it, we performed electrophoretic mobility shift assays (EMSA). The 190-bp SacI-Eco47III DNA fragment yielded a single retarded band, complex A, with protein extracts prepared from wild-type cells grown under aerobic, anaerobic, and heme-deficient conditions. This band was also seen with protein extracts prepared from FIG. 2. Protein binding to the SacI-Eco47III fragment of the HEM13 promoter. A, the labeled 190-bp SacI-Eco47III fragment was incubated for 15 min at 4°C with 1.0 g of poly(dI-dC) plus crude protein extracts (30 g) prepared from wild-type strain S150-2B (WT) and from its congenic cyp1⌬ and rox1⌬ mutant derivatives. O 2 , the cells were grown aerobically; N 2 , the cells were grown anaerobically; H Ϫ , the congenic heme-deficient hem1⌬ mutant strains were used. Electrophoresis was carried out at 4°C. First lane, probe incubated without protein extract. B, EMSA competition experiments with protein extracts from aerobic wild-type cells were carried out as above but with a molar excess (50-to 150-fold) of unlabeled PCR-amplified 75-bp fragment 4 (see Fig. 1), wild-type (WT), or carrying mutation m18 or m19. The sample in the first lane was incubated without added protein. cyp1⌬, cyp1⌬ hem1⌬, and rox1⌬ mutant cells (Fig. 2A). The formation of complex A was completely competed out by a 100-fold excess of the unlabeled DNA fragment, and preliminary experiments with overlapping segments of the SacI-Eco47III fragment used as competitors suggested that the protein-binding DNA sequence was located around position Ϫ333 (DraI restriction site) (results not shown). This was confirmed by determining the protected DNA region in complex A by indirect DNase I footprinting experiments (Fig. 3). No extended footprint was detected, but highly reproducible single base protections were observed at positions Ϫ328 and Ϫ342, together with sites more sensitive to cleavage at positions Ϫ340, Ϫ337, and Ϫ333. Site-directed mutations were made which changed 5-6 nucleotides around each of the protected sites; mutation m19 changed the sequence Ϫ342 TCTACTϪ337 to GAGATC, and mutation m18 changed the sequence Ϫ331 AC-GAG Ϫ327 to GATCT (see Fig. 4). The ability of the factor(s) to bind to the resulting altered sites was assessed by competition experiments. EMSAs were carried out with the labeled SacI-Eco47III fragment in the presence of the 75-bp PCR-amplified DNA fragments (fragment 4 in Fig. 1) carrying mutation m18 or m19 as competitors. A 150-fold molar excess of the mutated DNA fragment m18 did not prevent the formation of complex A, whereas the mutated DNA fragment m19 significantly impaired its formation, although not as completely as did the nonmutated wild-type fragment (Fig. 2B). These results indicate that the two sites contribute to complex formation but to different degrees. The fact that no other retarded band appeared in these competition experiments suggests, but does not prove, that a single factor binds to this approximately 17-bp region (Ϫ342 to Ϫ326). The nucleotide sequence of the 3Ј-end of this region (CGACTC on the noncoding strand), whose mutation m18 abolished complex formation, showed some similarity to the recognition sequences of the Gcn4p, Bas1p, Yap1p yeast transcriptional activators (TGACTC, see Ref. 34 for references). But none of these factors is implicated in complex A formation, since a normal complex was formed in EMSA using protein extracts prepared from gcn4⌬ bas1⌬ and yap1⌬ mutant cells (data not shown).
Mutational Analysis of the HEM13 UAS-The effects of the two mutations m18 and m19 on HEM13 gene expression were then assessed in vivo. The functional significance of a 5Ј-flanking GC-rich sequence (Ϫ363 to Ϫ344) was also explored by generating extensive mutations, m11 to m15, in this sequence, FIG. 4. Mutational analysis of the HEM13 UAS. The sequence of the HEM13 UAS is shown. The nucleotide substitutions in mutants m11 to m19 are indicated. Plasmids derived from pH13Z carrying a HEM13-lacZ fusion with the various mutations were introduced into the wild-type strain S150-2B and its hem1⌬ derivative. Transformants of S150-2B were grown in aerobiosis (O 2 ) and in anaerobiosis (N 2 ). Transformants of hem1⌬ were grown in aerobiosis (H Ϫ ). ␤-Galactosidase activity (A 420 nm ϫ 1000/min/A 600 nm of cells) was assayed in triplicate on at least two different cultures (A 600 nm ϭ 0.5-0.7), each inoculated with 10 -12 independent transformants. Values reported are the means with standard errors in parentheses. and analyzing their effects, individually or in combination with mutations m18 and m19, on HEM13 expression and regulation. The mutations were introduced into plasmid pH13Z (Fig. 1) carrying a gene fusion between HEM13 and E. coli lacZ, and the resulting plasmids were transformed into the wild-type strain S150-2B and its hem1⌬ derivative deficient in heme synthesis. HEM13 expression was analyzed by measuring ␤-galactosidase activity in the S150-2B transformants grown under aerobic (O 2 ) and anaerobic (N 2 ) conditions to determine the effects of oxygen deprivation and in the hem1⌬ transformants grown aerobically to determine the effects of heme deficiency (H Ϫ ) (Fig. 4). Mutations m11, m12, and m15 had no effect on basal aerobic HEM13 expression but slightly altered regulation by heme and oxygen. Mutation m14 caused a 2-fold decrease in HEM13 expression under all conditions. Mutations m18 and m19 had similar effects, but to different extents, consistent with their different effects on the formation of the DNA-protein complex A. Mutation m18 decreased aerobic expression 5.5-fold, the expression induced by oxygen deficiency 2-fold, and heme deficiency 4-fold. The effects of m19 were about one-half of those of m18. But the effects of mutations m18 and m19 were more dramatic when combined with mutations m11 or m14, although this synergistic effect did not occur under all physiological conditions. Mutation m11, which had no effect alone, caused a further 2-fold decrease in aerobic activity of the mutant promoters m18 and m19 (pH13Z/m11-19 and pH13Z/m11-18). The activity of the double mutant promoter m14-18 (pH13Z/m14-18) was decreased, under oxygen and heme deficiency, six and four times more than might be expected from a simple additive effect. The net effect of the double mutation m14 -18 was a 3-4-fold reduction in the extent of anaerobic and heme Ϫ induction as compared with the wild-type promoter, although neither of the two mutations alone was critical for induction. The double mutation m14 -19 had smaller effects than mutation m14 -18, especially in anaerobiosis where control was almost normal. This suggests that the region encompassing mutations m19 and m18 discriminated between the effects of oxygen and heme deficiency, as has been proposed (15). The fact that deleting the 190-bp SacI-Eco47III fragment causes a 4-fold decrease in the induction ratio (15) indicates that the 40-bp DNA segment studied here is probably the only regulatory element present in the SacI-Eco47III fragment.
Activation of a Heterologous Gene by HEM13 UAS-The functionality of the HEM13 UAS in a heterologous expression system was tested by inserting the 75-bp PCR-amplified DNA fragment 4, containing HEM13 UAS wild type or carrying the double mutation m14 -18, into a CYC1-lacZ fusion lacking the regulatory elements of the CYC1 and ANB1 genes. HEM13 UAS activated the transcription of the fusion gene 20-fold in aerobiosis (Fig. 5). Oxygen deficiency caused a further 2-fold activation, and heme deficiency caused a 4-fold activation. The double mutation m14 -18 completely abolished the activity of HEM13 UAS, under both basal aerobic and inducing conditions. These results demonstrate that HEM13 UAS acts as a UAS element in a heterologous context and confers heme and oxygen control on the reporter gene to the same extent as it does in its native context. Identification of the Rox1p-responsive Sites-HEM13 expression is repressed by the repressor Rox1p under aerobic hemesufficient conditions (11)(12)(13)15). The HEM13 upstream sequence contains three sequences that are identical or similar to the hypoxic consensus sequence for Rox1p recognition, YYYAT-TGTTCTC (10,17): R1 (Ϫ185 CCCATTGTTCTC Ϫ174) centered about 20 nucleotides upstream from the TATA element, R2 (on the noncoding strand, Ϫ238 TGCTTTGTTCAA Ϫ249), and R3 (Ϫ475 TCAATTGTTTAG Ϫ464) (Fig. 1). They were destroyed individually and in combination by introducing multiple base pair substitutions within the consensus core sequence ATTGTT to explore the relative role of these presumptive Rox1p-mediated control elements (Fig. 6). Derivatives of plasmid pH13Z carrying these various mutations were transformed into the wild-type strain S150-2B and an isogenic rox1⌬ deletion mutant strain, and the transformants were assayed for ␤-galactosidase activity (Fig. 6). Mutating R2 did not affect the expression, whereas gene expression was increased 1.3-1.5-fold after mutating R1 and 3.6 -3.8-fold after destroying R3. Mutating both R1 and R3 had an additive effect and led to an expression activity similar to that measured with the wild-type plasmid in a rox1⌬ background. Since incapacitating these two cis elements R1 and R3 had the same effect as genetic elimination of the trans-acting factor, we conclude that they are the only functional Rox1p-responsive elements in the HEM13 promoter. The 6 -7-fold repression by Rox1p observed with these multicopy gene fusion constructs correlates well with the 5-6fold ROX1-dependent repression measured for the chromosomal HEM13 gene (15).
Binding of a Factor Upstream of the R3 Site-In addition to FIG. 5. Activation of transcription of a heterologous gene by HEM13 UAS. The 75-bp fragment 4 from Ϫ384 to Ϫ310 (UAS-H13), wild-type or carrying the double mutation m14 -18, was cloned into the left-most SmaI and the right-most XhoI sites of the CYC1-lacZ fusion plasmid pLG669Z in place of the CYC1 and ANB1 regulatory sequences. The resulting plasmids were then transformed into strain S150-2B and its hem1⌬ derivative. The transformants were grown, and ␤-galactosidase was assayed as described in the legend to Fig. 4. the Rox1p-responsive site R3, sequences related to the binding site for Cyp1p were found in the HEM13 promoter region upstream of the SacI restriction site (15). It was also reported that Cyp1p could act as a weak repressor (ϳ2-fold) of HEM13 in aerobiosis and a weak activator (ϳ2-fold) in the absence of heme (12,13,24), independently of its role on ROX1 expression (15). The possible binding of Cyp1p, and/or other potential trans-acting factor(s) to the region upstream of SacI, was evaluated using EMSAs with various labeled restriction fragments extending from the NruI to XmnI sites; shorter, overlapping, PCR-amplified DNA fragments were also used (probes 1-3 in Fig. 1). The 93-bp Sfc1-SacI DNA fragment (Fig. 7) and probes 1-3 (data not shown) yielded a DNA-protein complex B that was specific as judged by competition with a 50-fold molar excess of unlabeled DNA. Complex B was formed only with protein extracts prepared from aerobic cells; it was absent when proteins were prepared from anaerobic or heme-deficient cells or from rox1⌬ and tup1⌬ mutants. Competition experiments (see Fig. 7B, third lane for the 29-bp probe 3) localized the sequence involved in the formation of complex B to the DNA segment Ϫ500 to Ϫ472, immediately upstream of the R3 site. Further attempts to delineate the binding site within this 29-bp segment by footprint analysis and mutagenesis were unsuccessful. DISCUSSION We have defined the cis and trans components of the two major regulatory pathways that control the induction of HEM13 expression in response to oxygen/heme deficiency: a 6 -7-fold aerobic repression exerted by Rox1p and a 3-4-fold activation mediated by a novel UAS element. The two mechanisms provide a 20 -30-fold induction to HEM13, which is slightly less than that observed experimentally (40 -50-fold). The additional 2-fold activation to reach full induction level might be mediated by Cyp1p under heme deprivation and by unknown factor(s) under anaerobiosis (15). But Cyp1p probably acts indirectly since it apparently does not bind HEM13 promoter probes, even when using protein extracts enriched in Cyp1p and under conditions optimized for detecting DNA⅐Cyp1p complexes (33) (data not shown).
HEM13 UAS-The HEM13 UAS is about 40-bp long and is composed of two subelements, a 16 -17-bp sequence binding a protein factor A and a 5Ј-flanking 20-bp GC-rich region. Both subelements are required additively for basal level of transcription, but each element alone appears to be sufficient for almost normal control of transcription by oxygen/heme. The protein-DNA complex A likely represents functionally important interactions, since different mutations affect both in vitro binding and in vivo promoter activity in a similar manner; this implicates protein factor A as a transcriptional activator. The DNA sequence of the factor A-binding site contains two blocks of dyad symmetry, CGAGTCG and TC-N 11 (AT-rich)-GA, which could be recognition motifs for factor A (see bottom of Fig. 3). Our results tend to favor the second one, but more work is needed to determine this with certainty. The role of the GC-rich region is not clear. It could bind a factor that was not detected in vitro in our EMSA conditions (perhaps via the rotationally Wild-type strain S150-2B (ROX1) and its congenic derivative rox1⌬ were transformed with HEM13-lacZ fusion plasmids (pH13Z) carrying the mutations mR1, mR2, mR3, and mR1-R3, and the ␤-galactosidase activity was determined as described in the legend to Fig. 4. The mutations in each plasmid are indicated by an X in place of an R element. symmetric CGG triplets), and this factor might participate with factor A in the additive transcriptional effect observed. Or its intrinsic DNA structure could impose constraints on the neighboring factor A-binding site. Unusual structures have been described for d(CCGCGG) (35) and d(GGGCCC) (36) sequences, which are both present within the GC-rich region. Sequence comparisons revealed no obvious similarity of HEM13 UAS to the binding sites for known transcription factors (34) and to sequences in the upstream regions of other oxygen-regulated genes (7), including the hypoxic SRP1 and TIP1 genes whose regulation is independent of Rox1p (37).
The way in which HEM13 UAS receives the signal generated by oxygen/heme deficiency and increases transcriptional activity in response to it is unknown at present. It is unlikely that factor A is responsible for the oxygen/heme deficiency activation switch, since no apparent differences were observed in the pattern of complex A by EMSA analysis of induced and noninduced extracts, and mutations abrogating complex A formation did not greatly impair the oxygen/heme control. However, factor A might interact with a sensor-regulator factor, which could also contact the GC-rich region or a protein binding to it, thus explaining why each of the HEM13 UAS subelement is sufficient for control. Isolation and analysis of factor A should help clarify the problem of HEM13 UAS function.
Rox1p-responsive Elements-Two sites, R1 and R3, mediate the repression by Rox1p under aerobic heme-sufficient conditions. A factor(s) B binds to the region immediately upstream of site R3, and its synthesis or DNA-binding activity is oxygen/ heme-dependent and requires Rox1p/Tup1p functions. This tight regulation of complex B formation, which parallels HEM13 regulation by Rox1p, suggests that these DNA-protein interactions are functionally significant. An oxygen-and hemedependent DNA-binding activity has recently been identified; it binds with low specificity to a region of the COX6 promoter previously shown to act as a repressor site in the absence of heme (38,39). Since this activity was only detected under high salt conditions (Ͼ125 mM NaCl, compared with 50 mM KCl for our EMSA conditions), it is unlikely to be factor B. Factor B is also unlikely to be the oxygen-and heme-dependent factor RC2 described earlier, which binds specifically to CYC1 UAS1 and competes with Cyp1p (40,41). Various levels of repression stringency have been reported for the Rox1p-responsive sites found in the ANB1, COX5b, and AAC3 genes (Fig. 8). It has been suggested that the poly(dT-dA) sequences present immediately upstream of the Rox1p-consensus elements are necessary for repressor function (17,20). These poly(dT-dA) tracts are very long in the ANB1 and AAC3 genes, shorter in COX5b, and very short and imperfect in HEM13 (Fig. 8). In the case of COX5b, a specific factor, Ixr1p(Ord1p), is also implicated in the aerobic repression (42). FIG. 8. Regulatory elements controlling the activation of well characterized hypoxic genes in response to oxygen deficiency. The scale at the top gives the distance in bp from the translation start codon (A of ATG is ϩ1). The thin arrows mark the transcription start sites and T indicates TATA elements. Thick arrows symbolize sequences similar to the hypoxic consensus sequence for Rox1p recognition, YYYATTGTTCTC: those marked with two asterisks play a major role in Rox1p-mediated repression; those marked with one asterisk play a minor role; those without asterisk have no effect on gene expression. Hatched boxes indicate poly(dT-dA) sequences; numbers under these boxes indicate matches with and lengths of the poly(Ts). The approximate position of DNA-protein complexes is shown beneath the line for each gene. This figure is based on data from the following sources: HEM13 (this work), ANB1 (16,17), COX5b (19,39), AAC3 (20). FIG. 7. Binding of a factor to a 29-bp sequence upstream of the R3 site. A, the labeled SfcI-SacI fragment was incubated for 15 min at 20°C with 0.75 g of sonicated salmon sperm DNA plus protein extracts (20 g) prepared from wildtype strain S150-2B (WT) and from its congenic cyp1⌬, rox1⌬, and tup1⌬ mutant derivatives. O 2 , the cells were grown aerobically; N 2 , the cells were grown anaerobically; H Ϫ , the congenic heme-deficient hem1⌬ mutant strains were used. Electrophoresis was carried out at 20°C. The sample in the first lane was incubated without proteins. B, EMSA competition experiments with protein extracts from aerobic wild-type cells were carried out as above but with a 50-fold excess of the unlabeled SfcI-SacI fragment (second lane) or probe 3 (third lane) (see Fig. 1 for the definition of the probes).
Ixr1p(Ord1p), a two high mobility group-box protein initially isolated as a protein that binds to and bends platinated DNA (43), recognizes a 44-bp COX5b DNA segment encompassing the Rox1p element, probably in a structure-specific rather than a sequence-specific manner (42). Poly(dT-dA) sequences, by virtue of their intrinsic structure, can alter chromatin structure so that transcription factors gain better access to their cognate binding sites located in the vicinity (44). In the absence of long poly(dT-dA) sequences, other factors could be required that provide similar functions. The function of factor B (and Ixr1p ?) may be to help Rox1p bind to its recognition sequence and anchor the general repressor complex Ssn6p⅐Tup1p with the proper protein-protein interactions to ensure its repressor activity.