INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Transcriptional repression in eukaryotic cells often involves the assemblage of large complexes that repress through active mechanisms such as direct interactions with the basal transcriptional machinery and organization of chromatin into repressive structures (16, 18, 34). The repression of the hypoxic genes in baker's yeast provides an example of such a complex involving the DNA-binding protein Rox1 and the general repression complex Tup1-Ssn6 (20, 46, 47). Our studies have focused on a number of aspects of hypoxic gene regulation, including how differential levels of repression of the hypoxic genes are achieved, how the repression complex forms on the DNA, and how the complex inhibits transcription.

The hypoxic genes encode oxygen-related functions in respiration, heme, and membrane biosynthesis that are required at higher levels when molecular oxygen is limiting (46, 47). The expression of these genes is repressed under aerobic conditions by Rox1 binding to their regulatory regions (2, 5, 7). To achieve this oxygen-dependent repression, the ROX1 gene is transcriptionally induced aerobically and repressed anaerobically (2, 6). The level of Rox1-dependent repression of different hypoxic genes is variable, and we have divided these genes into two classes in terms of the strength of repression. The first includes unique genes that encode functions required under aerobic conditions, such as HEM13, OLE1, ERG11, and the autorepressed ROX1 itself. Because they are required aerobically, these genes can only be partially repressed. The second includes genes that have an aerobic homologue, such as HMG1-2, COX5A-5B, AAC1-2-3, and TIF51A-ANB1 (where the first gene is the aerobic and the last is the hypoxic homologue). These genes can be completely repressed. Variations in the quality and number of the Rox1 binding sites in the regulatory regions of the hypoxic genes contribute to this differential repression, but this is not the complete explanation (7). Our extensive analysis of one strongly repressed hypoxic gene, ANB1, revealed that there are two operators upstream of this gene, each consisting of two Rox1 sites in close proximity (7, 24). All four sites bind Rox1 with similar affinities, but the upstream operator, OpA, represses almost 10 times more effectively than does OpB, which is closer to the TATA box. This difference is not due to the location of these sites, but is a function of some intrinsic property of their sequences. We report here that this difference is due to the presence of a binding site for the protein Mot3 in OpA. This site is present in some but not all Rox1-repressed genes. Furthermore, we provide evidence here that Mot3 functions by either aiding in the recruitment of a general repression complex to the ANB1 promoter or helping the general repression complex function.

Rox1-dependent repression also requires the general repression complex Tup1-Ssn6 (2, 45). This complex has no DNA-binding activity, but rather interacts with a variety of regulon-specific DNA-binding proteins to target specific genes for repression. These regulons include, in addition to the hypoxic genes, the a mating type and haploid-specific genes, the glucose-repressed genes, DNA damage-inducible genes, flocculence genes, and others (10, 12, 21, 26, 30, 32, 41, 42). Two alternate mechanisms for Tup1-Ssn6-dependent repression have been proposed. There is ample evidence for the ability of this complex to organize chromatin (4, 8, 9, 23, 27, 36, 37, 39). Nucleosomes are phased by Tup1-Ssn6 in some repressed genes. This phasing is probably achieved through the ability of Tup1 to interact with hypoacetylated histones H3 and H4. The importance of this phasing has been demonstrated by the observation that deletions of the N-terminal coding region of either of these two histones caused a partial derepression of some Tup1-Ssn6-repressed genes. Finally, the TATA-binding protein (TBP) is excluded from binding to the TATA box by the Tup1-Ssn6 complex, consistent with a model of a positioned nucleosome blocking TBP access. On the other hand, there is evidence that Tup1-Ssn6 interacts directly with the basal transcriptional machinery (22, 33, 35, 40, 44). Anchoring either Tup1 or Ssn6 to DNA can inhibit transcription of chromatin-free DNA in vitro. Mutations have been isolated in the RNA polymerase II mediator complex that cause derepression of some Tup1-Ssn6-repressed genes, indicating a genetic interaction. While it may be possible that these two alternate repression mechanisms have some common components, at this point the link is not obvious, and we assume that they represent alternate and, for some genes, redundant mechanisms. In this study, we provide evidence for this view for ANB1. Nucleosomes show Tup1-, Rox1-, Mot3-, and histone H4-dependent phasing on the ANB1 regulatory region, but while deletion of TUP1, ROX1, or MOT3 results in at least partial loss of repression, deletion of the N-terminal coding sequence of H4 does not.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Strains and growth conditions. The strain RZ53-6 (MAT trp1-289 leu2-3,112 ura3-52 ade1-100) and the RZ53-6rox1 and RZ53-6tup1 derivatives have been described (5, 45). RZ53-6mot3 and RZ53-6rlm3 were derived from the wild type and rox1 strains, respectively, by displacement of the MOT3 gene with the mot3::kanMX construct described below. The strain P1/I8 contained deletions of both the HHT1-HHF1 and HHT2-HHF2 loci, which encode the isoforms of histones H3 and H4 (31). The cells were maintained carrying plasmids encoding the wild-type HHT1-HHF1 genes or the N-terminal deletions of histone H3 (hht1-2-HHF1) or H4 (HHT1-hhf1-8). An srb10 allele was transformed into this strain using the psrb10::LYS2 plasmid described below.

Plasmids. All plasmid constructions were carried out using standard techniques (1). Enzymes were purchased from New England Biolabs and used as recommended by the vendor. PCRs were carried out with Taq polymerase (Perkin-Elmer) as recommended by the vendor. Genomic DNA for PCRs was prepared from RZ53-6 as described (19). Genomic sequences were obtained from the Saccharomyces Genome Database maintained at Stanford University. The sequence for a given gene is numbered with the first A in the ATG initiation codon as +1; bases 5'-wards are numbered negatively, and those 3'-wards are numbered positively.

Enzyme and RNA assays. -Galactosidase assays were carried out as described (19). All assays were performed multiple times with multiple independent transformants for each plasmid in each strain. Invertase assays were carried out as described (3, 14). Cells were grown repressed in SC containing 4% glucose or derepressed in SC containing 2% raffinose.

Protein purification. The MBP-Rox1 fusion was expressed in BL21-Codon Plus (DE3)-RIL cells (Stratagene). One liter of cells was grown in L-broth plus kanamycin (34 µg/ml) and chloramphenicol (20 µg/ml). The fusion was induced with 0.2 mM IPTG (isopropyl--D-thiogalactopyranoside). Cells were harvested and broken in a French press. The fusion protein was purified using amylose beads as described (2).

Gel retardation assays. The gel retardation assays have been described (2). The synthetic DNAs used are indicated in the appropriate figures. The radioactivity in the gel bands was quantitated using a Storm 860 PhosphorImager (Molecular Dynamics).

Micrococcal nuclease sensitivity assays. Cells used for the sensitivity assays were transformed with the multicopy plasmid YEp(112)ANB1 for the RZ53-6 derivatives and YEp(195)ANB1 for the P1/I8 strains. They were grown to midexponential phase in 400 ml of SC with tryptophan or uracil at 30°C with vigorous aeration. Chromatin preparations were carried out as described (1) with modification. The tup1 cells were quite flocculent, and it was difficult to obtain efficient spheroplast formation with enzyme treatment alone. Therefore, a short, vigorous mixing with glass beads (0.5 mm) followed the zymolyase treatment for all strains to maintain uniformity.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

A sequence in ANB1 OpA is involved in Rox1-mediated repression. The repression region of the ANB1 gene consists of four Rox1 sites that we have divided into two operators, A and B, as illustrated in Fig. 1. The level of repression effected by OpA is nearly 10 times greater than that by OpB despite equivalent levels of noncooperative Rox1 binding to the sites within these operators (7). The two Rox1 sites act synergistically in repression from OpA, while the two Rox1 sites in OpB act additively. This difference did not appear to be due to a difference in spacing of the Rox1 sites in the operators or to the position of the operator relative to the TATA box. Rather, the sequence between the Rox1 sites of OpA appeared to contain information that rendered it a better repression site. This prompted us to examine the operator sequences more carefully. Comparison with the regulatory regions of other Rox1-regulated genes revealed a conserved sequence closely associated with the 5' OpA Rox1 binding site, TCGTTGCCT. This sequence has been noted before (24, 29), and a point mutation in it that affected ANB1 repression has been described (29), but its importance was overshadowed by the extensive studies of the Rox1 binding sites. This sequence was also affected by a previously described 10-bp deletion in OpA which changed the last T residue of the sequence to a C residue and caused severe loss of repression (7).

View larger version (26K): [in this window] [in a new window]   FIG. 1.   Mutations in the ANB1 regulatory region. The top diagram represents the wild-type ANB1-lacZ fusion gene in the plasmid YCp(33)AZ. The diagram includes the UAS (open box on left), the two operators with the Rox1 binding sites (solid boxes), and the Mot3 binding site in OpA (grey box), the TATA box (the triangle), and the coding sequence (open box on right). The middle diagram presents the OpA mutations constructed in an OpB deletion plasmid, YCp(33)AZOpB. The sequence of OpA is shown with the Rox1 binding sites indicated in solid boxes. The lower diagram presents the OpB mutations constructed in an OpA deletion plasmid, YCp(33)AZOpA. The sequence of OpB is shown with the Rox1 binding sites indicated in solid boxes.

The Mot3 DNA-binding protein acts through this OpA sequence. At this point in our investigations, we were alerted to a factor possibly working through this sequence. Sertil et al. (38) had identified the DAN1 gene, which was induced during anaerobiosis and regulated by a Rox1-independent regulatory system. Subsequently, they isolated a mutation that derepressed DAN1 and, surprisingly, ANB1. This mutation was complemented by the MOT3 gene (Charles V. Lowry, personal communication). Mot3 is a DNA-binding protein that contains two zinc fingers and appears to be involved in the regulation of a variety of genes (15, 25). It binds to the consensus sequence T(G/A)CCT(A/T/G), which matches the sequence found in OpA.

View larger version (77K): [in this window] [in a new window]   FIG. 2.   ANB1 RNA levels are partially derepressed by mot3. RNA was prepared from cells grown derepressed (anaerobically, lanes 1 to 4) or repressed (aerobically, lanes 5 to 8) from RZ53-6 (wild type [WT], lanes 1 and 5), RZ53-6mot3 (lanes 2 and 6), RZ56-6rox1 (lanes 3 and 7), and RZ53-6tup1 (lanes 4 and 8). The blot was hybridized to 32P-labeled DNA probes prepared from the ACT1 and ANB1 genes. The ANB1 probe cross-hybridizes to the aerobically expressed TIF51A RNA. An overexposed segment of lanes 5 and 6 is presented below the main autoradiograph for better visualization of the ANB1 band in the mot3 strain.

Mot3 functions as part of the Rox1-dependent repression complex. Mot3 could act to augment repression through the Rox1-Tup1-Ssn6 repression complex, or it could repress ANB1 expression independently. If the former were the case, the combination of loss of Mot3 repression plus loss of Rox1 repression should give no further increase in ANB1 repression beyond that observed with the loss of Rox1 repression. On the other hand, if the latter were the case, we would expect that ANB1 expression would be higher when both repression mechanisms were lost. An inspection of Table 1 indicates that mutations in the Mot3 site that cause partial loss of repression in a wild-type strain, OpA 1(10), 4, 5, and 6, do not cause any additional loss of repression in a rox1 strain, comparing 97 Miller units of activity for the plasmid with the wild-type Mot3 site compared to 101 to 116 Miller units for the mutant plasmids.

Some, but not all, Tup1-Ssn6-repressed genes are partially derepressed in the mot3 strain. There are putative Mot3 binding sites in close proximity to Rox1 binding sites in the regulatory regions of the hypoxic genes ACC3, COX5B, and HEM13 but not ROX1, which is autorepressed. To determine whether Mot3 plays a role in their repression, we examined the effect of the mot3 deletion on the expression of lacZ reporter genes, comparing the level of derepression to that observed in a rox1 deletion strain (Table 2). As expected, the AAC3 and HEM13 fusions were partially derepressed, while the ROX1 fusion was not. Surprisingly, the COX5B was unaffected by the mot3 deletion, but this gene was not regulated strongly under these conditions, perhaps minimizing any Mot3 effect. Overall, these results suggest that Mot3 plays a general role in enhancing Rox1-dependent repression of a number of, but not all, hypoxic genes.

Mot3 binds specifically to OpA in vitro. To demonstrate that the putative Mot3 site in the ANB1 OpA can bind the Mot3 protein, we performed in vitro binding studies using a six-histidine-tagged version of Mot3 expressed in and partially purified from E. coli cells (25). Gel retardation studies were performed using a radiolabeled synthetic DNA containing OpA, and as can be seen in Fig. 3, a slower migrating band was visible in the presence of Mot3 (lane 2 and 3). To demonstrate that this band represented specific binding to the Mot3 sequence, competitor DNA was added that contained either the Mot3 site (lanes 4, 5, 7, and 8) or a deletion of the Mot3 site (lanes 6 and 9). Only the DNA containing the Mot3 site competed effectively to reduce the levels of the retarded band. OpA contains two Rox1 binding sites, and the labeled DNA bound two molecules of Rox1 (lane 10). Both competitors contained the two Rox1 sites also, and both competed equally well to reduce Rox1 binding (lanes 11 and 12).

View larger version (47K): [in this window] [in a new window]   FIG. 3.   Mot3 binds to OpA. (A) Gel retardation was carried out with 32P-labeled synthetic double-stranded DNA containing the OpA sequences (5'-TTTTCGTTTTTCCATTGTTCGTTCGTTGCCTCCTATTGTTCTCGAGCCTAAAA). The Rox1 sites are underlined, and the Mot3 site is in boldface. The DNA was synthesized so that the annealed molecules had single-stranded 5' ends that could be filled in for labeling (1). DNA used for competition either contained (as above) or lacked (5'-TTTTCGTTTTTCCATTGTTCGTTTTTTTTGCCCTATTGTTCTCGAGCCTAAAA) the putative Mot3 binding site. Competitor was added at 5 or 10 times the concentration of labeled DNA. The His-tagged Mot3 protein was prepared as described, and either 1 or 5 µl was added per binding reaction where indicated. MBP-Rox1 was prepared as described, and 1 ng was added per binding reaction where indicated. (B) Gel retardation was carried out with 32P-labeled synthetic double-stranded DNA (labeled as above) containing the OpA sequence lacking the two Rox1 binding sites (Mot3-DNA), 5'-TTTTTCC------CGTTCGTTGCCTGTTTTTTTGCCCT------CTCAAAA. The sequences underlined represent the Rox1 binding sites, with the dashes indicating deleted bases. The sequence in boldface is the Mot3 binding site. GST-Mot3 fusion (10 ng) was added in lane 2, and 10 ng of the cleaved Mot3 (plus free GST) was added to lane 3. No protein was added to lane F (free DNA).

Mot3 and Rox1 do not bind cooperatively to DNA. We envision three models whereby Mot3 can enhance Rox1-dependent repression: (i) Mot3 could interact with Rox1 to enhance its binding to DNA; (ii) Mot3 could bind independently to DNA and help recruit the Tup1/Ssn6 repression complex; or (iii) Mot3 could help the repression complex function, for example, by interacting with nucleosomes or the basal transcriptional machinery. To distinguish between the first and latter two possibilities, we tested for cooperative interactions between Rox1 and Mot3 binding to DNA in vitro. It is difficult to assess cooperative interactions directly by gel retardation because the pattern of bands is complex. There are two Rox1 binding sites plus one Mot3 site in OpA, resulting in five different DNA complexes that can form. This banding pattern can be seen in Fig. 4, where two sets of titrations were carried out: one in which increasing Rox1 was added to a constant amount of Mot3 (lanes 2 to 5), and the second in which increasing amounts of Mot3 were added to a constant amount of Rox1 (lanes 6 to 9). We also added increasing amounts of Rox1 in the absence of Mot3 (lanes 10 to 12) to help in identification of the complexes containing one or two molecules of Rox1. For this experiment, the GST-Mot3 fusion was digested with thrombin to release the Mot3 protein, ensuring that the GST moiety did not interfere with a potential Mot3-Rox1 interaction. The Rox1 protein used in these experiments was fused to the MBP, but this fusion repressed ANB1 expression in yeast cells as well as the wild-type Rox1 (data not shown), and therefore, if Rox1 interacts with Mot3, the fusion would do so, too. All the expected bands were visualized except the fully loaded DNA, which would contain two Rox1 and one Mot3 molecule. If Rox1 and Mot3 interacted cooperatively, we would expect that the amount of complex containing Mot3 (Fig. 4) would increase with increasing Rox1. This was not the case; the fraction of DNA to which Mot3 bound remained constant at about 0.3 in lanes 2 to 5. Similarly, the fraction of DNA in Rox1-containing complexes increased only about twofold from lanes 6 to 9. These effects were reproducible, but the absolute numbers varied from experiment to experiment. The results indicate that Rox1 and Mot3 do not interact directly.

View larger version (97K): [in this window] [in a new window]   FIG. 4.   Mot3 and Rox1 bind to OpA independently. Gel retardation was carried out with a synthetic 32P-labeled DNA (labeled as in Fig. 3) containing the two Rox1 and one Mot3 sites. The sequence was 5'-TTTTTCCATTGTTCGTTCGTTGCCTGTTTTTTTGCCCTATTGTTCTCAAAA, where the underlined sequences are the Rox1 binding sites and the sequence in bold is the Mot3 binding site. Protein was added to the indicated lanes as follows: none to lane 1; 10 ng of Mot3 and 0, 5, 20, and 50 ng of MBP-Rox1 to lanes 2, 3, 4, and 5, respectively; 10 ng of MBP-Rox1 and 0, 2.5, 10, and 40 ng of Mot3 to lanes 6, 7, 8, and 9, respectively; and 25, 50, and 100 ng of MBP-Rox1 to lanes 10, 11, and 12, respectively. The Mot3 used in this experiment was the GST-Mot3 fusion cleaved with thrombin. The deduced complexes are indicated to the right, where M represents Mot3 and R represents MBP-Rox1.

View larger version (55K): [in this window] [in a new window]   FIG. 5.   Mot3 and Rox1 do not bind cooperatively to DNA. (A) Gel retardation was carried out with the 32P-labeled Mot3-DNA shown in Fig. 3B. The same DNA was used as an unlabeled competitor in 1, 2.5, and 5 times the labeled DNA in lanes 5, 6, and 7, respectively. A DNA fragment containing the Mot3 and Rox1 sites (shown in Fig. 4) was also used as unlabeled competitor at 1, 2.5, and 5 times the labeled DNA in lanes 8, 9, and 10, respectively. GST-Mot3 fusion (10 ng) was added in lane 2, and 10 ng of the cleaved Mot3 (plus free GST) was added to each of lanes 3 to 10; 20 ng of MBP-Rox1 was added to each of lanes 4 to 10. (B) Gel retardation was carried out with either the 32P-labeled Mot3-DNA (lanes 1 to 4) or 32P-labeled OpA-DNA (lanes 5 to 8). MBP-Rox1 was added at 5 ng (lanes 2 and 6), 20 ng (lanes 3 and 7), and 100 ng (lanes 4 and 8).

Effect of Mot3 on chromatin arrangement at ANB1. To determine how Mot3 contributes to the assembly of the repression complex on the ANB1 regulatory region, we used the previously established ability of this complex to alter chromatin structure in other regulons as a marker for its presence at the ANB1 locus in vivo. Micrococcal nuclease sensitivity assays were used to probe chromatin structure, and as shown in Fig. 6, we observed three reproducible differences among the patterns obtained with chromatin-bound DNA from wild-type and mutant cells or deproteinated (naked) DNA. Two sensitive sites (solid arrows) were observed in deproteinated DNA that were protected in repressed wild-type cells but not in derepressed tup1 or rox1 cells. These two sites map around the TATA box (Fig. 6), suggesting that the repression complex blocks access to the TBP. The protected region in wild-type cells is approximately 170 bp, in the size range expected for a nucleosome, as drawn in the diagram. This observation agrees with the finding that TBP binding to the ANB1 TATA box was significantly greater in tup1 than in wild-type cells (23). Interestingly, in the partially derepressed mot3 cells, these bands appeared less intense, suggesting that the partial depression results from either the incomplete assembly or partial function of the repression complex.

View larger version (62K): [in this window] [in a new window]   FIG. 6.   Repression complex alters chromatin structure at ANB1. (A) The autoradiograph represents a Southern blot of micrococcal nuclease-digested chromatin carried out as described in Materials and Methods. Chromatin was prepared from RZ53-6 (wild type) cells and its derivatives transformed with YEp(112)ANB1, as indicated. Micrococcal nuclease was added to final concentrations of 0 (), 1, 4, or 12 U/0.4 ml, and digestion was carried out for 10 min at 37°C. The samples in the lanes marked naked were prepared from RZ53-6 and deproteinated before the addition of 0, 1, 3, or 9 U of micrococcal nuclease per 0.4 ml for 10 min at 37°C. All samples were digested with BglII plus EcoRI after deproteination. This digestion generated a 1.4-kb fragment in the absence of nuclease. A 1-kb ladder size standard (New England Biolabs) was loaded in lane M; due to its high concentration, this DNA cross-hybridized to the probe, and the sizes (in kilobases) are indicated to the right. The open arrow represents the Rox1-dependent sensitive site, and the solid arrows represent the repression complex-dependent resistant sites. (B) Diagram of the ANB1 regulatory region. The Rox1 binding sites are represented as black boxes, the Mot3 site as a grey box, the TATA box as a triangle, and the coding sequence as an open box. The numbers designate base pairs starting from the BglII site. The BglII-SalI hybridization probe is indicated. The open and solid arrows are described above, and the ellipse represents the proposed positioned nucleosome.

Nuclease-protected sites result from an interaction between the repression complex and nucleosomes. Tup1 interacts with histones H3 and H4 (8), and although deletions in the N-terminal regions of either of these two proteins causes at least partial depression of some Tup1-Ssn6-repressed regulons, ANB1 repression is not affected (7). Nonetheless, given the differences in the nuclease sensitivity between wild-type cells and derepressed mutants, we investigated the nuclease sensitivity of ANB1 in these histone deletions. As shown in Fig. 7, in cells carrying the H4 N-terminal deletion, the two nuclease-sensitive sites protected in the wild-type cells (arrows) were not protected, giving the same pattern as that for rox1 and tup1 cells and deproteinated DNA. In cells carrying the H3 N-terminal deletion, these sites were only slightly sensitive, indicating a less dramatic effect of this allele.

View larger version (91K): [in this window] [in a new window]   FIG. 7.   Mutations in the N-terminal domain of histone H4 alters chromatin structure at ANB1. The autoradiograph represents a Southern blot of micrococcal nuclease-digested chromatin carried out as described in Materials and Methods and the legend to Fig. 4. Chromatin was prepared from P1/I8 cells carrying YEp(195)ANB1 and plasmids with the HHT1-HHF1 alleles (wild type), the HHT1-hhf1-8 alleles (H4N), or the hht1-2-HHF1 alleles. Micrococcal nuclease was added to final concentrations of 0 (), 2, 8, or 24 U/0.4 ml for the wild-type and H4N samples; 0, 0.25, 1, or 8 U/0.4 ml for the H3N samples; and 0, 1, 3, or 9 U for the naked DNA samples (prepared from the wild-type cells). Lane M, size markers (in kilobases). The arrows indicate bands representing the sites protected in the wild-type samples.

Srb10 does not play a role in ANB1 repression. Mutations in the SRB10 gene were isolated in screens for derepression of both the glucose-repressed genes and a mating type genes, regulons which are repressed by the Tup1-Ssn6 complex (22, 40, 44). Srb10 is a protein kinase member of the mediator complex, a component of the RNA polymerase II holoenzyme (11). Although it is unclear what role Srb10 plays in Tup1-Ssn6 repression, it seemed to be a good candidate to test for a role in the nucleosome-independent pathway. We constructed an srb10 deletion and srb10 hhf1-8 and srb10 hht1-2 double mutants. Neither the single deletion nor the double deletions affected repression significantly (Table 4). While there was a less than twofold increase in the srb10 deletion strain, there was no further increase in combination with the histone mutations. Fully derepressed expression was well above 150 Miller units in all cases. Clearly Srb10 does not play a major role in the nucleosome-independent pathway for repression of ANB1. This finding raises the intriguing possibility that the general repression complex interacts with the basal transcriptional machinery in different ways when anchored to different genes.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Role of Mot3 in hypoxic gene repression. We report here that the transcription factor Mot3 enhances repression by the Rox1-Tup1-Ssn6 complex. Mot3 was originally identified in two separate genetic screens, one for suppressors of adaptation to pheromone and the other for high-copy suppressors of the synthetic lethality of the mot1-24 spt3 double mutation (encoding general transcription factors) (15, 25). Using MOT3 overexpression and a mot3 deletion, Grishin et al. (15) showed that Mot3 negatively regulated (either directly or indirectly) a set of pheromone-inducible genes and positively regulated an eclectic set of other genes. They also demonstrated that a LexA-Mot3 fusion could act as an activator in an otherwise upstream activation sequence (UAS)-less reporter gene, but did not act as a repressor when a UAS was present. A mot3 deletion conveys no dramatic phenotype under a variety of growth and stress conditions except for a mild increase in UV sensitivity. Both studies suggested that Mot3 is a global transcription factor; it affects the expression of a variety of genes but may not be essential for the expression of any given gene. No insights were gained from these previous studies as to how it might function in the repression of hypoxic genes.

Role of chromatin in repression of ANB1. There is ample evidence that Tup1-Ssn6 can organize chromatin, and our findings suggesting that the complex positions a nucleosome over the TATA box are not surprising. It agrees with the report of Kuras and Struhl (23) that TBP cannot bind to the ANB1 regulatory region under repressed conditions. What is surprising is that the loss of this positioned nucleosome has no effect on repression. We found that in wild-type cells, the region around the TATA box was protected from micrococcal nuclease digestion and this protection was lost in cells lacking Rox1, Tup1, or the N-terminal region of histone H4. However, we previously reported (7) and confirmed here (Table 4) that this histone mutation does not cause derepression. Either nucleosome phasing plays no role in ANB1 repression, or there are redundant mechanisms, one nucleosome dependent and the other nucleosome independent. Clearly the evidence favors the second possibility. Mutations in histone H3 or H4 cause derepression in other systems, and we believe that our proposed positioned nucleosome is responsible for the inability of TBP to bind to the ANB1 TATA box under repressed conditions. The elimination of the positioned nucleosome without loss of repression in the H4 mutant provides us with the opportunity to genetically dissect the nucleosome-independent pathway, which we hope will ultimately the provide tools to study how both pathways operate in ANB1 repression.