The HIR1 gene product is required to repress transcription of three of the four histone gene loci in Saccharomyces cerevisiae, and like its counterpart, the HIR2 protein, it functions as a transcriptional corepressor. Although Hir1p and Hir2p are physically associated in yeast, Hir1p is able to function independently of Hir2p when it is artificially recruited to the histone HTA1 promoter. A deletion analysis of HIR1 has revealed two separate repression domains: one in its N terminus, where seven copies of the β-transducin or WD40 motif reside, and the second in the remaining C-terminal amino acids. Overexpression of the WD repeats in a hir1Δ strain complemented its Hir− phenotype, while overexpression of the C terminus in a wild-type strain caused both Hir− and Spt− phenotypes. The Hir1p C terminus physically interacted in vivo with Hir2p, and both Hir1p repression domains interacted with full-length Hir1p. It was additionally found that the Hir1p WD repeats functionally interacted with the SPT4, SPT5, and SPT6 gene products, suggesting that these repeats may direct Hir1p to different protein complexes.
TRANSCRIPTIONAL repression has emerged as an increasingly important mechanism to control gene expression in eukaryotes. Although many eukaryotic repressors are site-specific, DNA-binding proteins, a number of them have been classified as transcriptional corepressors because they function without directly contacting DNA. In the absence of DNA binding, transcriptional corepressors can be targeted to gene promoters in several ways. For example, the Saccharomyces cerevisiae Tup1p and Ssn6p corepressors (Keleheret al. 1992; Tzamarias and Struhl 1994) are recruited to a subset of yeast promoters by their association with site-specific, DNA-binding proteins such as α2 and Mig1p (Komachiet al. 1994; Treitel and Carlson 1995; Tzamarias and Struhl 1995). Other transcriptional corepressors, such as yeast Sin4p (Jiang and Stillman 1992), are recruited to promoters because they are components of the RNA polymerase II mediator/holoenzyme complex (Liet al. 1995). We have obtained evidence that Hir1p and Hir2p, two regulators of histone gene transcription (Osley and Lycan 1987; Sherwoodet al. 1993), are transcriptional corepressors of the class exemplified by the Tup1 and Ssn6 proteins. The two Hir proteins do not appear to bind to DNA, but can act as direct transcriptional repressors when artificially recruited to yeast promoters (Spectoret al. 1997). We have speculated that they are targeted to specific histone gene promoters, including the promoter of the divergently transcribed HTA1-HTB1 locus, by their interactions with as yet unidentified DNA-binding proteins that act at defined upstream negative regulatory sites (Osleyet al. 1986; Lycanet al. 1987; Osley 1991; Sherwood 1993; Freemanet al. 1992). At these promoters, Hir1p and Hir2p, which appear to form a complex in vivo, function to keep the histone HTA1, HTB1, HHT1, HHT2, HHF1, and HHF2 genes repressed for most of the cell cycle (Spectoret al. 1997).
Hir1p and Hir2p may not act alone at histone gene promoters. Among the other proteins with which they are likely to act are the products of the HIR3/HPC1, HPC2, HPC3, HPC4, and HPC5 genes, which were also identified in screens for histone gene regulators (Xuet al. 1992), and the products of the SPT4, SPT5, and SPT6 genes (Clark-Adams and Winston 1987; Neigebornet al. 1987; Clark-Adams et al. 1998; Swanson and Winston 1990, 1992; Swansonet al. 1991; Maloneet al. 1993). The two HIR and three SPT gene products are functionally related because mutations in both sets of genes suppress his4-912δ insertion mutations in the same way as mutations in the genes that encode histones H2A and H2B (Clark-Adamset al. 1988; Sherwood and Osley 1991). A further link between the two groups of proteins was also shown by the observation that mutations in the SPT4, SPT5, and SPT6 genes decrease the level of transcription from the histone HTA1-HTB1 locus, which is subject to Hir-mediated repression (Compagnone-Post and Osley 1996). One possible explanation for this effect is that the three SPT gene products affect HTA1 and HTB1 transcription through Hir1p or Hir2p.
To understand how Hir1p represses the transcription of selected genes, and to learn how the Hir2, Spt4, Spt5, and Spt6 proteins function in this process, we identified the regions of Hir1p that are responsible for repression in vivo. Hir1p was found to have two separate repression domains. The first, which comprises seven copies of a canonical β-transducin (WD40) repeat (Neeret al. 1994), complemented the Hir− defect of a hir1Δ strain when it was overexpressed. The second, which includes the remaining Hir1p C-terminal amino acids, produced both Hir− and Spt− phenotypes when it was overexpressed in a wild-type strain. These two regions of Hir1p also showed differential in vivo interactions with Hir1p, Hir2p, and Spt4p/Spt5p/Spt6p: The C terminus of Hir1p could be coimmunoprecipitated from yeast cell extracts with Hir2p and Hir1p, while the WD40 repeats physically interacted with Hir1p and functionally interacted with the three Spt proteins. Together, the results suggest that the two repression domains promote the interaction of Hir1p with diverse proteins and thus may contribute to the activity of Hir1p at selected promoters.
MATERIALS AND METHODS
Strains, growth conditions, and RNA analysis: The yeast strains used in this study are listed in Table 1. The lithium acetate method was used for plasmid transformations (Itoet al. 1983), with selection for the prototrophic marker present on each plasmid. Cells were grown in SD medium supplemented with appropriate amino acids and bases or in YPD medium (Kaiseret al. 1994). The latter medium was used for studies in which DNA replication was blocked with 0.2 M hydroxyurea for 30 min to measure the Hir phenotype of strains (Osley and Lycan 1987). All strains were grown at 30°, except for FY546, which was grown at 24° and then shifted to 37° for 90 min to inactivate the SPT6 gene product. The Spt phenotype of HIR and hir1Δ strains was measured by spotting serial dilutions of cells on SD medium with and without histidine, followed by incubation at 24° for 3–5 days.
Total RNA was prepared from yeast strains and hybridized to end-labeled probes specific for lacZ and RP51A mRNA, as described previously (Osleyet al. 1986; Moranet al. 1990). After S1 nuclease digestion, the protected fragments were separated on 4% acrylamide–8-M urea gels, which were dried and exposed to film or quantitated by phosphoimager analysis on a phosphoimager (Fuji, Stamford, CT) using MACBAS software.
Plasmids: HTA1-lacZ reporter plasmids with and without lexA operator sites are derivatives of plasmid pCALA1 and have been described previously (Rechtet al. 1996; Spectoret al. 1997). Plasmid pCALA1 is a LEU2 CEN4 plasmid in which the lacZ gene is under control of the wild-type HTA1-HTB1 promoter (Osleyet al. 1986).
Plasmid pBTM116 contains a full-length lexA gene under control of the ADH1 promoter (S. Fields, personal communication). Plasmids pBTM-HIR1 and pBTM-HIR2, which contain fusions of the full-length HIR1 or HIR2 coding region to the C terminus of lexA, have been described previously (Spectoret al. 1997). Truncated HIR1 genes were fused to the C terminus of lexA on plasmid pBTM116 using oligonucleotide-directed mutagenesis by PCR (Ausubelet al. 1989). For each fusion gene, a SmaI restriction site was created in frame with an ATG initiation codon, and SmaI-SalI restriction fragments that contained different segments of HIR1 were ligated to SmaI-SalI–digested pBTM116. In plasmids pBTM-HIR17WD, pBTM-HIR13WD, and pBTM-HIR11WD, a SmaI site and ATG inition codon were placed at the beginning of the first (7WD), fifth (3WD), and seventh (1WD) WD repeat of HIR1 with oligonucleotides 5′-CCAAAGGTCTCTCCCGGGATGAAAGTG-3′ (7WD), 5′-TATCACAACCCGGGGGATATGTCTTT-3′(3WD), and 5′-ACAAGTAGCCCGGGGCCAATGTTA-3′(1WD), and a translation termination codon and SalI site were placed immediately after the seventh WD40 repeat of HIR1 with the oligonucleotide 5′-AATTGGCTTGTCGACTTCATTATTTTC-3′. To create plasmid pBTM-HIR1Cterm, a SmaI site and ATG codon were placed at position +529 in the HIR1 ORF, immediately after WD40 repeat 7, using the oligonucleotide 5′-AATAATGACCCGGGCATGCCAATT-3′. A SmaI-SalI fragment that contained the natural HIR1 translation termination codon was then inserted into plasmid pBTM116.
Various segments of the HIR1 coding region were fused to the native HIR1 promoter by their insertion into a promoter cassette in plasmid YEp352, a 2μ URA3 vector. To construct this cassette, a SmaI site was created immediately before the HIR1 ATG codon by PCR-directed mutagenesis, using the oligonucleotide 5′TACCACTTTCATCCCGGGAGAGACCTT-3′, and an EcoRI-SmaI fragment that contained the HIR1 promoter was subcloned into plasmid YEp352. The promoter cassette was digested with SmaI and SalI and was used as a recipient for SmaI-SalI fragments isolated from HIR1 inserts in plasmid pBTM116. Plasmid YEpHIR1 contains the entire HIR1 ORF, YEpHIR11WD+Cterm contains the WD repeat 7 and C terminus of HIR1, YEpHIR11WD contains WD repeat 7, YEpHIR13WD contains WD repeats 5-7, YEpHIR17WD contains WD repeats 1–7, and YEpHIR1Cterm contains amino acids 392–840) from the HIR1 C terminus. To construct YCpHIR17WD, an EcoRI-SalI fragment from YEpHIR17WD was subcloned into plasmid YCp50, a CEN-URA3 vector.
The 2μ URA3 plasmids that carry hemagglutinin (HA)-tagged HIR1 and HIR2 genes and untagged HIR1 and HIR2 genes have been described previously (Spectoret al. 1997). Plasmid pBTM-SPT4 (LexA-Spt4p) was constructed by oligonucleotide-directed mutagenesis with the oligonucleotides 5′-GGTACAGTCCCGGGGATGTCTAGTGAAAGA-3′ (SmaI at start codon) and 5′-GTGATATCAGAGTCGACGGTTTTACTCAAC-3′ (SalI at stop codon). Plasmid pBM65 (2μ URA3 HA-SPT4) was generously provided by Fred Winston.
All gene constructions were confirmed by DNA sequence analysis using double-stranded DNA templates with dideoxy chain terminators (Sangeret al. 1977).
Immunological analysis: Whole-cell extracts were prepared as described previously (Spectoret al. 1997) from BJ5465 cells (Table 1) that had been cotransformed with high copy number plasmids carrying various combinations of epitope-tagged or untagged HIR1, HIR2, or SPT4 genes. Between 150 and 600 μg of crude lysate were incubated with 2.5 μl of monoclonal antibody 12C5A directed against the HA epitope or with 0.5 μl of rabbit polyclonal antibody against LexA under conditions that have been described previously (Sherwoodet al. 1993; Spectoret al. 1997). After the last wash, the immunoprecipitates were resuspended in 20 μl of SDS sample buffer and heated at 100° for 5 min (Ausubelet al. 1989). The eluate (5 μl) was analyzed on 7.5 or 12.5% SDS polyacrylamide gels for the presence of the tagged protein recognized by the primary antibody, and 15 μl was analyzed by 7.5 or 12.5% SDS-PAGE for the presence of coimmunoprecipitated proteins. After electrophoresis, the proteins were transferred to Immobilon filters, and Western blot analysis was performed with either anti-HA antibody (1:1000 dilution) or anti-LexA antibody (1:2000 dilution) using enhanced chemiluminescence for detection according to manufacturer's directions (Dupont-NEN, Wilmington, DE).
Hir1p contains two repression domains: The HIR1 gene product belongs to a diverse and expanding group of proteins that contain multiple copies of the β-transducin or WD40 motif, which has been been postulated to mediate protein–protein interactions (Fonget al. 1986; Hartleyet al. 1988; Dalrympleet al. 1989; Whitewayet al. 1989; Komachiet al. 1994; Neeret al. 1994; Sathe and Harte 1995; Kaufmanet al. 1997). Seven copies of this motif are present at the N terminus of the Hir1 protein (Lamouret al. 1995). To investigate whether these repeats were required for Hir1p to function as a transcriptional repressor, we deleted the entire WD40 domain (amino acids 1–389) and fused the remaining C-terminal region (amino acids 392–840) to LexA (Brent and Ptashne 1984), and we then directed the LexA-Hir1(Cterm) fusion protein to lexA operator sites at the promoter of an HTA1-lacZ reporter gene (Figure 1). We showed previously that a full-length LexA-Hir1 fusion protein acted as a direct transcriptional repressor in this context, and that it responded appropriately to several cell-cycle regulatory signals (Spectoret al. 1997). Thus, the LexA tethering assay provided a convenient method by which to monitor the functional state of Hir1p. The results indicated that the seven WD repeats were dispensable for repression by tethered Hir1p; in their absence, the 448 C-terminal amino acids (hereafter referred to as the Hir1p C terminus) repressed HTA1-lacZ transcription almost as strongly as a full-length Hir1 fusion protein (Figure 1A, Table 2). We next asked whether the seven WD repeats on their own (amino acids 1–389) repressed transcription when they were brought to the HTA1 promoter (Figure 1B). Although repression was slightly weaker, it was still significant (Table 2), indicating that Hir1p contains two separate repression domains—one that includes the N-terminal seven WD repeats and another that encompasses the remaining amino acids of the C terminus.
To determine how many WD repeats were required for transcriptional repression in this assay, we constructed two additional lexA fusion genes that contained either a single copy (WD repeat 7, amino acids 448–389) or three copies (WD repeats 5–7, amino acids 213–389) of the Hir1p WD repeats, and we examined their effects on HTA1-lacZ transcription (Figure 1B). Repression was observed in both cases, with the tethered protein that contained three WD repeats conferring a slightly stronger effect (Table 2). Nonetheless, because a single WD repeat caused significant repression when it was brought to the HTA1 promoter, the data suggest that a small number of these repeats might contribute to the repression function of the native Hir1 protein.
The Hir1 protein physically interacts with itself and with the Hir2 protein in vivo (Spectoret al. 1997). Having identified two repression domains in Hir1p, we wished to determine whether these domains showed different physical interactions with Hir1p or Hir2p. lexA-HIR1 fusion genes that contained the complete HIR1 ORF, the seven HIR1 WD repeats (lexA-HIR17WD), or the HIR1 C terminus (lexA-HIR1Cterm) were cotransformed with a HA-HIR1 or HA-HIR2 gene into a protease-deficient strain to reduce nonspecific proteolysis during extract preparation. The HA-Hir1 or HA-Hir2 proteins were precipitated from whole-cell extracts with an antibody against the HA epitope, and Western blot analysis with anti-LexA antibodies was used to follow the presence of the two LexA-Hir proteins both in the immune precipitates (P) and in the supernatant fractions (S) that remained after precipitation (Figure 2). The results showed that both the Hir1p WD repeats and C terminus could be coprecipitated with full-length HA-Hir1p (Figure 2A, lanes 2 and 6), although only a fraction (~10%) of each LexA fusion protein was associated with HA-Hir1p. This association was specific, however, because it was not observed in strains in which an untagged HIR1 gene was substituted for the HA-HIR1 gene (Figure 2A, lanes 4 and 8). In contrast, only the LexA-Hir1(Cterm) fusion protein was coprecipitated with HA-Hir2p (Figure 2B, lane 6), an association that was specific to the presence of the HA epitope in Hir2p (Figure 2B, lane 8). It also appeared that a larger fraction of the LexA-Hir1(Cterm) protein was associated with HA-Hir2p (~20%) than with HA-Hir1p, perhaps reflecting stronger heterotypic interactions through this region of Hir1p. The small amount of LexA-Hir1(7WD) that coprecipitated with HA-Hir2p (Figure 2B, lane 2) appeared to represent a nonspecific interaction because an equivalent amount of the fusion protein was present in the immuneprecipitate from cells cotransformed with an untagged HIR2 gene and a lexA-HIR1(7WD) gene (Figure 2B, lane 4).
We addressed the functional significance of these physical interactions by determining whether the HIR1 gene product was required for the Hir1p WD repeats or C terminus to repress transcription when tethered at the HTA1 promoter (Table 2, Figure 3). As we observed previously (Spectoret al. 1997), the tethered, full-length Hir1 protein was not as strong a repressor when Hir1p was absent (Table 2). The tethered C terminus also showed a partial Hir1p dependence, while the tethered WD repeats (one, three, or seven copies) generally exhibited the strongest requirement for Hir1p (Table 2). These dependencies represented true functional dependencies because each LexA fusion protein was made at approximately equivalent levels in both HIR and hir1Δ backgrounds (H. DeSilva, unpublished data).
The residual repression seen with all of the LexA fusion proteins was unlikely to depend on the Hir2 protein because deletion of the HIR2 gene reduced repression by full-length Hir1p, the Hir1p C terminus, or the seven Hir1p WD repeats less than 1.5-fold (M. A. Osley, unpublished data). Thus, one interpretation of these combined studies is that the physical association of the Hir1p C terminus with Hir2p and the Hir1p WD–WD interactions represent different functional states of Hir1p.
Effects of spt4, spt5, and spt6 mutations on transcriptional repression by tethered Hir1p: One explanation for the residual repression conferred by tethered, full-length Hir1p in a hir1Δ or hir2Δ strain is that proteins other than Hir1p or Hir2p assist Hir1p in this function. Candidates for such proteins are the products of the SPT4, SPT5, and SPT6 genes, a group of functionally related transcriptional repressors that may also act without directly binding to DNA (Clark-Adamset al. 1988; Swanson and Winston 1992). We recently showed that mutations in these three SPT genes decreased transcription of the HTA1-HTB1 locus, a target of repression by Hir1p (Compagnone-Post and Osley 1996). To test the idea that the Spt proteins might alter HTA1-HTB1 transcription through Hir1p, we asked whether mutations in each of the three SPT genes affected transcriptional repression by tethered, full-length Hir1p (Figure 4, Table 3). Repression by LexA-Hir1p was found to be reduced in all three spt mutants. The loss of repression was modest in spt5 and spt6 backgrounds, and the largest decrease was seen in the spt4 mutant (Table 3). These effects were specific to Hir1p because repression by a tethered, full-length Hir2p protein was unaffected by mutations in either the SPT4, SPT5, or SPT6 genes (Figure 4).
We next asked whether the requirement of tethered Hir1p for SPT4 depended on the presence of the seven Hir1p WD repeats or the Hir1p C terminus. We found that an spt4 mutation significantly reduced repression by the tethered WD repeats but had little effect on repression by the tethered C terminus (Figure 5 and Table 3). In addition, we observed that repression by the tethered WD repeats also required Spt5p and Spt6p (Table 3). Because the LexA fusion proteins are stable (H. DeSilva, unpublished data), these results reflected true functional dependencies. Together, the results suggest that the products of all three SPT genes play a role in the function of Hir1p as a transcriptional repressor, and that this role depends on the presence of the seven WD repeats.
Physical association between Hir1p and Spt4p: Because of the partial functional dependence of tethered, full-length Hir1p on Spt4p, we asked whether Hir1p and Spt4p might be physically associated in yeast cells. High copy number plasmids carrying a full-length HIR1 gene fused to lexA and an HA-SPT4 gene were cotransformed into a protease-deficient strain, and HA-Spt4p was precipitated from cell extracts with anti-HA antibody. The presence of LexA-Hir1p in the immuneprecipitates was examined by Western blot analysis, using an antibody directed against LexA (Figure 6A). LexA-Hir1p was found in the HA-Spt4p immunoprecipitates only if the cell extract had first been incubated with anti-HA antibody (Figure 6A, lanes 3 and 5); incubation with a nonspecific antibody did not result in LexA-Hir1p precipitation (Figure 6A, lanes 2 and 4). Using a second pair of epitope tagged genes, HA-HIR1 and lexA-SPT4, we next precipitated LexA-Spt4p from cell extracts with an antibody against LexA, and we examined whether HA-Hir1p was present in the immune precipitates by Western blot analysis with an anti-HA antibody (Figure 6B). Again, Hir1p was found to be associated with Spt4p only when the extract was incubated with an antibody against LexA (Figure 6B, lane 3), not with a nonspecific antibody (Figure 6B, lane 2). We estimated that only a low percentage (1–2%) of Hir1p coprecipitated with Spt4p. Thus, Hir1p–Spt4p interactions may be significantly weaker than either Hir1p–Hir1p or Hir1p–Hir2p interactions, suggesting a dynamic or unstable association.
We performed a similar analysis with HA-Spt6p (F. Winston, unpublished data) and LexA-Hir1p, but we were unable to detect reproducible coprecipitation of LexA-Hir1p with this Spt protein (K. L. Lee, unpublished data). Hir1p, then, may preferentially interact in vivo with Spt4p, which also interacts with Spt5p and Spt6p (F. Winston, personal communication). Because of the strong functional interactions between the seven Hir1p WD repeats and both Spt4p (Figure 5) and Spt6p (Table 3), we also asked whether LexA-Hir1p (7WD) could be coprecipitated with either HA-tagged Spt protein; again, we were unable to obtain reproducible coprecipitation (K. L. Lee, unpublished data). These interactions will be reinvestigated with untagged Hir1 proteins when antibodies against native Hir1p become available.
Effects of overexpression of the Hir1p WD repeats and C terminus: Having identified two repression domains in Hir1p by the LexA tethering assay, we asked whether these domains functioned independently in the context of the native Hir1 protein. The seven Hir1p WD repeats or the Hir1p C terminus were expressed under control of the HIR1 promoter on a high copy number plasmid in a hir1Δ strain. The overexpressed WD repeats fully complemented the Hir− phenotype of this mutant (Osley and Lycan 1987), restoring its ability to repress HTA1 transcription when DNA replication was inhibited (Figure 7A). The presence of the same repeats on a CEN plasmid failed to complement the Hir− phenotype of the hir1Δ strain (Figure 7B), indicating that complementation required the overexpression of the Hir1p N terminus. To determine how many WD repeats were required for complementation, we also examined the effects of overexpression of three WD repeats (repeats 5–7; Figure 7B) because these same three repeats were able to repress transcription when tethered at the HTA1 promoter (Figure 1B). This construct was unable to complement the Hir− phenotype of the hir1Δ mutant. These results suggest that more than three WD repeats are required to substitute for the absence of native Hir1p at the histone gene promoter. Alternatively, the presence of specific WD repeats might be required to confer this regulatory effect.
Unlike the Hir1p WD repeats, the overexpressed Hir1p C terminus was unable to complement the Hir− phenotype of the hir1Δ mutant (Figure 7A). Instead, when this region of Hir1p was overexpressed in a HIR strain, it caused a Hir− phenotype, an effect not observed when the Hir1p WD repeats or full-length Hir1p were overproduced in the same wild-type background (Figure 7C). A surprising result was observed when a single copy of the Hir1p WD40 motif (repeat 7) was added back to the C terminus. This construct now complemented the Hir− phenotype of the hir1Δ strain (Figure 7A), and it suppressed the Hir− phenotype caused by overexpression of the C terminus in a wild-type strain (H. DeSilva, unpublished data).
A second phenotype associated with the absence of HIR1 is the suppression of the his4-912δ allele, or an Spt− (His+) phenotype (Sherwood and Osley 1991; Sherwoodet al. 1993). We therefore asked whether overexpression of either the seven Hir1p WD repeats or Hir1p C terminus complemented the Spt− phenotype of a hir1Δ his4-912δ strain (Figure 8A). Neither construct was able to correct the Spt− defect of this strain (Figure 8A). Next, we asked whether overexpression of either Hir1p domain produced a Spt− phenotype in a HIR his4-912δ background (Figure 8B). A strong Spt− phenotype was found to be associated with overexpression of the Hir1p C terminus, but not with overexpression of full-length Hir1p or the seven Hir1p WD repeats. The addition of a single WD40 repeat (repeat 7) to the Hir1p C terminus reversed its ability to cause an Spt− phenotype in the wild-type strain (Figure 8B), but this same construct could not complement the Spt− phenotype of a hir1Δ mutant (Figure 8A).
These phenotypic studies support the notion that the WD repeats and C terminus make different contributions to the function of Hir1p as a transcriptional corepressor. The WD repeats appear to be responsible for the specificity of Hir1p interactions, while the C terminus may contribute to the stoichiometry of Hir1p associations. Moreover, because overexpression of the Hir1p WD repeats could correct the Hir− but not the Spt− phenotype of a hir1Δ mutant, these data provide the first evidence that different requirements must be met for Hir1p to function at the HTA1 and his4-912δ promoters.
Our previous studies led us to propose that the Hir1 corepressor is targeted to histone gene promoters by its association with a factor that binds at promoter-specific negative regulatory elements or with another protein that directly contacts this factor, and that once at a promoter, it contacts a downstream target to bring about repression (Sherwood 1993; Spectoret al. 1997). Hir1p is associated in vivo with a second corepressor, Hir2p, but when Hir1p is artificially recruited to the histone HTA1 promoter, it can repress transcription in the absence of Hir2p (Spectoret al. 1997). This result helped support the notion that the role of Hir2p is to bring Hir1p to relevant promoters, where it then acts as a direct transcriptional repressor. Hir1p also acts directly or indirectly at other genes besides those encoding core histones, notably alleles of HIS4 and LYS2 that contain δ insertion mutations, although it is not known how it is targeted to these loci (Sherwood and Osley 1991). In this study, we performed a functional analysis of Hir1p to understand its role in the transcription of the genes it controls. The results showed that the Hir1 protein contains two separate repression domains. The first domain encompasses seven copies of the β-transducin or WD40 repeat at the Hir1p N terminus (amino acids 1–390), and the second includes the remaining 448 C-terminal amino acids. In native Hir1p, these domains most likely mediate the interactions of Hir1p with different proteins to bring about its transcriptional effects.
The N-terminal domain is both necessary and sufficient for the function of Hir1p at histone gene promoters because overexpression of the seven WD repeats allowed repression of the histone HTA1 gene in a strain lacking Hir1p (Figure 7). Thus, the notion that the primary function of Hir2p is to recruit Hir1p to these promoters is probably not the case because it was the Hir1p C terminus, not the WD repeats, that physically interacted in vivo with Hir2p (Figure 2). The WD repeats, therefore, might interact directly with a DNA-binding factor at the HTA1 promoter, bringing Hir1p to this promoter without the mediation of Hir2p. It is unlikely that these repeats directly contact downstream targets of repression as well, because when they were tethered at the HTA1 promoter, either Hir1p (Figure 3) or Spt4p/Spt5p/Spt6p (Figure 5, Table 3) had to be present for repression to occur.
The Hir1p C terminus, while unable to correct the Hir− phenotype of a hir1Δ mutant, was identified as a separate repression domain because it repressed HTA1 transcription when tethered at the promoter (Figure 1). Unlike the WD repeats, however, the tethered C terminus repressed transcription independently of both Hir1p (Figure 1) and the three Spt proteins (Table 3). This suggests that the Hir1p C terminus might contact downstream repression targets directly without the intervention of other proteins.
Based on the assumption that the two repression domains have specialized roles, we propose the following model for how native Hir1p inhibits HTA1 transcription: the WD domain is postulated to contact the factor that binds at the HTA1-negative site, bringing Hir1p to the promoter, while the C-terminal domain is presumed to have direct contact with a downstream target, causing repression. We also suggest that when the WD repeats are overexpressed in a hir1Δ mutant, truncated Hir1p is still brought to the HTA1 promoter, but the interaction of the WD repeats with Spt4p/Spt5p/Spt6p now allows Hir1p to contact this downstream target. Thus, the Hir1p C terminus and Spt4p/Spt5p/Spt6p might be functionally equivalent. Although this predicts that overexpression of the WD repeats would be unable to repress transcription in a strain that is deleted for both HIR1 and SPT4, this cannot be tested directly because a hir1Δspt4Δ double-mutant is inviable (F. Winston, personal communication).
What, then, is the role of Hir2p in the function of Hir1p as a repressor of HTA1 transcription? Based on the results of this and a previous study (Spectoret al. 1997), we postulate that while both Hir1p and Hir2p are required for repression at the HTA1 promoter (Sherwoodet al. 1993), and while a fraction of Hir1p and Hir2p are physically associated in vivo (Figure 2), Hir1p and Hir2p may in fact function independently. This conclusion is based on the observation that Hir2p is dispensable for repression by tethered full-length Hir1p (Spectoret al. 1997), the tethered Hir1p WD repeats, and the tethered Hir1p C terminus (M. A. Osley, unpublished data). Hir1p and the Spt4/Spt5/Spt6 proteins, on the other hand, may function together at the HTA1 promoter because all three Spt proteins are required for the tethered Hir1p WD repeats to repress transcription. We have observed previously, however, that in spt4, spt5, and spt6 mutants, the HTA1 promoter is partially repressed rather than derepressed, as the tethering studies would have predicted (Compagnone-Post and Osley 1996). These conflicting data can be reconciled if the absence of the Spt proteins frees Hir1p to enter into alternative protein interactions to repress HTA1 transcription.
Different requirements must be met for Hir1p to function at the his4-912δ promoter. Overexpression of the Hir1p WD repeats was unable to correct the Spt− phenotype of a hir1Δ strain, indicating that the Hir1p C terminus has an essential function in this effect. We had previously suggested that the Spt− phenotype of hir mutants could be an indirect effect of the unbalanced production of histones in these strains, a physiological situation that results in suppression of the δ insertion (Clark-Adamset al. 1988; Sherwood and Osley 1991). The results of the present study, however, provide the first evidence that Hir1p acts directly at both the HTA1 and his4-912δ promoters because the restoration of histone stoichiometry by WD overexpression did not concomitantly produce an Spt+ phenotype. Despite this conclusion, we still do not know whether Hir1p acts independently or with the Spt4/Spt5/Spt6 proteins to affect his4-912δ transcription.
The effects of overexpression of the Hir1p C terminus provide additional support for the notion that this region of Hir1p forms different protein associations from those of the WD repeats. Hence, the dominant Hir−Spt− phenotype produced by the overexpressed C terminus could have resulted from the disruption of specific protein–protein interactions at both the HTA1 and his4-912δ promoters. For example, because the Hir1p C terminus could be coprecipitated with Hir2p in vivo (Figure 2), the overexpressed C terminus might bind to Hir2p and prevent it from interacting with other proteins required for HTA1 repression, leading to a Hir− phenotype. The Spt− phenotype could also have resulted from the disruption of protein complexes that act at the his4-912δ promoter. As discussed above, our data suggest that the Hir1p C terminus must be present for Hir1p to function at this promoter. Overexpression of the C terminus, which interacts in vivo with Hir1p and Hir2p (Figure 2), might favor Hir1p–Hir1p interactions and thus prevent the association of Hir1p with Spt4p or other proteins at the his4-912δ locus.
The observation that the mutant phenotypes produced by overexpression of the Hir1p C terminus in a wild-type strain were reversed by the addition of a single WD repeat suggests that the interactions of the WD repeats are dominant to those of the C terminus in native Hir1p. This supports the notion that WD-protein interactions target the C-terminal repression domain to appropriate locations. For example, the WD repeats could promote Hir1p interactions with a DNA-binding factor at the HTA1 promoter, thereby targeting the Hir1p C terminus to this promoter, where it could now engage in appropriate contacts with downstream targets to inhibit transcription.
The presence of WD repeats in a protein is postulated to provide multiple interfaces for interactions with other proteins, although the broad sprectrum of proteins that contain these repeats suggests that the range of targets is also broad (Neeret al. 1994). The Tup1p transcriptional corepressor contains WD repeats with some structural and functional similarity to those of Hir1p (Komachiet al. 1994; Tzamarias and Struhl 1995). A single Tup1p WD repeat contacts the α2 repressor at a cell-specific genes (Komachiet al. 1994), raising the issue of the specificity of WD associations: If the Hir1p WD repeats also contact a site-specific, DNA-binding protein, how do Hir1p and Tup1p recognize their correct DNA-binding targets? Perhaps the structure of the two WD repeats is sufficiently different, or association with other proteins, e.g., Ssn6p with Tup1p (Williamset al. 1991), confers an additional level of specificity.
We thank Dessislava Dimova and Judith Recht for their critical reading of the manuscript, Mark Treitel for construction of the lexA-SPT4 fusion gene, Roger Brent for the gift of anti-LexA antibodies, and Fred Winston and Stan Fields for strains or plasmids. This work was supported by National Institutes of Health grant GM40118 to M.A.O.
Communicating editor: A. P. Mitchell
- Received August 5, 1997.
- Accepted November 17, 1997.
- Copyright © 1998 by the Genetics Society of America