INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

In Saccharomyces cerevisiae, the TUP1 gene encodes a protein required for repression of genes regulated by cell type, glucose, oxygen, DNA damage, and other signals (26, 38). Tup1p forms a complex in vivo with Ssn6p (24, 34). This complex does not bind DNA directly but is recruited to target gene promoters through interaction with a variety of sequence-specific DNA-binding proteins (2p for mating-type control [16, 28], Mig1p and Nrg1p for glucose repression [23, 31], Rox1p for oxygen repression [1, 39], and Crt1p for DNA damage [12]). Ssn6p may serve as an adapter between Tup1p and these DNA-binding proteins (33). Interestingly, Tup1p-LexA fusion proteins directly mediate repression of appropriate reporter genes, independently of Ssn6p (32). However, Ssn6p-LexA fusions require Tup1p for repression (15). Tup1p, then, appears to directly mediate repression, while Ssn6p does not.

In vitro protein binding experiments and two-hybrid analyses have defined a number of domains in the 713-amino-acid Tup1p protein. The 72 N-terminal amino acids of Tup1p are required for interaction with Ssn6p and self-multimerization (33). The histone binding and repression domain comprises amino acids 73 to 385 (6, 32). WD repeats (amino acids 333 to 706) in the C-terminal region of Tup1p likely form a seven-bladed -propeller structure (18, 29) that interacts with 2p (16).

Two mechanisms of repression have been proposed for the Ssn6p-Tup1p complex (7, 38). A number of factors necessary for repression, including Sin4p (4, 13), Sin3p/Rpd1p (36), Rpd3p (35), Srb10p/Are1p/Ssn3p, Srb11p/Ssn8p, and Srb8 (3, 17, 37, 38), are associated with subcomplexes within the RNA polymerase II holoenzyme. These findings suggest that Ssn6p-Tup1p may inhibit transcription through interactions with the transcription machinery. In support of this model, a modest amount of repression (two- to fourfold) can be achieved in vitro, in the presence of just the basal transcription machinery (10, 24).

A second model proposes that Tup1p mediates repression through the organization of chromatin. Tup1p interacts directly with the amino-terminal tail domains of histones H3 and H4 in vitro (6), and mutations in these histone domains synergistically reduce repression of multiple classes of Tup1p-regulated genes in vivo (6, 11). Moreover, the H3-H4 binding domain in Tup1p coincides with the repression domain. Ssn6p-Tup1p interactions with components of chromatin may lead to decreased accessibility of promoter regions, thereby effecting repression.

The above-described models for Tup1p repression are not mutually exclusive. Complete repression by Ssn6p-Tup1p may involve interactions with both the basal transcription machinery and the histones. For example, Ssn6p-Tup1p complexes might first halt transcription through altering the activity of the basal apparatus and then maintain the repressed state through organization of chromatin.

To further understand the mechanism of Tup1p repression, we sought functional homologs in other, related and unrelated yeasts. Here, we report a structural and functional analysis of TUP1 homologs from Kluyveromyces lactis and Schizosaccharomyces pombe. Our findings suggest that histone binding is a conserved feature of Tup1p repressor functions.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Yeast strains. S. cerevisiae YMH427 (MAT tup1::HIS3 ura3-52 trp1 his3 pho3 pho5 leu2-3,112::[LEU2, STE6-PHO5]) was used for complementation tests of the tup1 disruption. YMH465 (MAT leu2-3,112 trp1 pho3 pho5 ura3-52::[URA3, lexA4-CYC1-lacZ]) was used for monitoring the ability of LexA-Tup1p fusions to repress the CYC1 reporter. TY3 (MAT ura3-52 leu2-1 his3-200 trp1-1) was used for preparation of histone proteins. K. lactis IFO1267 was used for preparation of genomic DNA. The wild-type strain S. pombe 972 (h) was used for preparation of the genomic DNA and total RNA to be used as the PCR templates, and JY741 (h ura4-D18 leu1-32 ade6-M216) was used for construction of the tup11 and tup12 gene disruptants.

Cloning of K. lactis TUP1 and S. pombe tup11⁺ and tup12⁺ genes. The K. lactis TUP1 gene was identified by Southern blot hybridization (27) under conditions of low stringency with a PCR product containing the WD repeat region of the S. cerevisiae TUP1 gene (corresponding to bp +1066 to +1552, relative to ATG) as a probe. A 1.1-kbp EcoRI fragment that reproducibly cross-hybridized with the S. cerevisiae probe was isolated from the K. lactis genomic DNA in pBluescript II KS(+). One of the positive clones, pKL5-2, carried nucleotide sequences similar to those encoding the WD repeats of S. cerevisiae TUP1 but truncated regions homologous to the N-terminal region. Therefore, a 0.7-kbp EcoRI-BglII fragment from pKL5-2 was used as a probe to identify a 2.0-kbp HindIII-BglII fragment containing the N-terminal region of the K. lactis TUP1 gene. This fragment was cloned into the plasmid pKL4-3. The plasmid pKLTUP1, carrying the entire K. lactis TUP1 gene, was constructed by ligating the 1.3-kbp HindIII-SpeI fragment from pKL4-3 to the 1.1-kbp SpeI-EcoRI fragment from pKL5-2. The nucleotide sequence of the insert DNA of pKLTUP1 was determined. The plasmid pYMC105 was constructed by insertion of the 2.3-kbp BamHI-SalI fragment, containing the entire K. lactis TUP1 gene, into the same site of YCp50, and this single copy vector was used for complementation analysis.

Construction of tup11 and tup12 disruptants. The tup11::ura4⁺ disruptant, JY741-tup11U, was constructed by transformation of JY741 with the BamHI- and HindIII-digested plasmid which had the insertion of the ura4⁺ DNA fragment at the BglII site of pYMS266. The tup12::LEU2 disruptant, JY741-tup12L, and the tup11::ura4⁺ tup12::LEU2 double disruptant, JY741-tup11U, tup12L, were constructed by transformation of JY741 and JY741-tup11U, respectively, with the BamHI- and XhoI-digested plasmid which had the insertion of the S. cerevisiae LEU2 DNA at the BglII site of ptup12int.

Construction of the S. cerevisiae-S. pombe Tup1p hybrids. The YCp50-based vector pYMC111 carrying the S. cerevisiae TUP1 promoter was constructed by insertion of a PCR product amplified with the primers 5'-CTCAAGCTTATTTTGCGCACGTTGGATTG-3' (corresponding to positions 939 to 920 relative to ATG) and 5'-CTCGGATCCCCATATTGGTTTGGATGGAAA-3' (corresponding to positions +3 to 17) into YCp50 after digestion with EcoRI and HindIII. The S. cerevisiae TUP1 and the S. pombe tup11⁺ DNA fragments were synthesized by PCR and cloned into the HindIII-SalI site of pYMC111 for expression in S. cerevisiae. Each gene was divided into three regions, roughly corresponding to the amino-terminal Ssn6p-binding domain, the central repression domain, and the C-terminal WD repeats of the S. cerevisiae protein. The PPP construct contained only S. pombe sequences, and the CCC construct contained only S. cerevisiae sequences. PPP was constructed by cloning of the entire coding region (corresponding to amino acid positions 1 to 614) of the S. pombe tup11⁺ gene amplified by PCR into pYMC111. PPC was constructed by replacing the region from positions 298 to 614 of PPP with the region from positions 329 to 713 of S. cerevisiae TUP1 at the SphI site of the S. pombe tup11⁺ gene. Similarly, CCP was constructed by replacing the region from positions 1 to 297 of PPP with the region from positions 1 to 328 of S. cerevisiae TUP1 at the SphI site of the S. pombe tup11⁺ gene. CPP was constructed by replacing the region from positions 73 to 328 of CCP with the region from positions 71 to 297 of S. pombe tup11⁺ from the plasmid pBTM-tup11 by using the MluI and SphI sites. CCC was constructed by ligating the region from positions 1 to 328 of CCP and the region from positions 329 to 713 of PPC. PPC2 was constructed by ligating the region from positions 1 to 351 of S. pombe tup11⁺ and the region from positions 434 to 713 of S. cerevisiae TUP1 by using the PstI site. Similarly, CCP2 was constructed by ligating the region from positions 1 to 433 of S. cerevisiae TUP1 and the region from positions 352 to 614 of S. pombe tup11⁺ by using the PstI site.

GST pull-down assays. S. cerevisiae Ssn6p and S. pombe Tup11p proteins were produced and labeled with ³⁵S-methionine by the TNT Quick Coupled Transcription/Translation system (Promega). Histone proteins were purified from S. cerevisiae TY3 as described previously (7, 19). GST-2p, GST-S. cerevisiae Tup1p and GST-S. pombe Tup11p fusion proteins were expressed in E. coli DH5 and purified with glutathione-Sepharose 4B beads (Amersham Pharmacia Biotech) by following the manufacturer's protocol. In vitro-labeled proteins or unlabeled histones were incubated with comparable amounts (as determined by sodium dodecyl sulfate-polyacrylamide gel electrophoresis [SDS-PAGE] analysis) of the different GST fusion proteins bound to beads in binding buffer (20 mM Tris-HCl [pH 8.0], 150 mM NaCl, 1 mM EDTA, 0.1% Triton X-100, 1 mM glutamate, 1 mM dithiothreitol) at room temperature for 1 h. Then, the beads were collected by centrifugation in a microcentrifuge at 500 × g for 2 min. The supernatant was saved, and the beads were washed five times with binding buffer. The washed beads were resuspended directly in 1× SDS-PAGE sample buffer. All samples were separated by SDS-PAGE and autoradiographed for detection of the ³⁵S-labeled proteins or visualized by staining with Coomassie brilliant blue R-250 for observation of histones (27).

Far-Western blot analysis. Separation of histone proteins by SDS-PAGE, electroblotting onto nylon membrane, and binding to ³⁵S-labeled probes were done as previously described (5, 6). Full-length S. cerevisiae Tup1p and S. pombe Tup11p were synthesized and labeled with ³⁵S-methionine with pCite/Tup1 (6) and pCITE-tup11, respectively, as the templates and purified with microcon 10 spin columns (Amicon) as described previously (6).

Northern blot analysis. Cultivation of S. cerevisiae strains was as described previously (25). S. pombe strains were grown in YEL with 3% glucose and 2% Casamino Acids to a concentration of 0.5 × 10⁷ to 0.8 × 10⁷ cells/ml. The cells were collected by centrifugation, washed with sterile water, and shaken for 3 h in YEL with 2% Casamino Acids and 8% glucose (repressing conditions) or 0.1% glucose and 3% glycerol (derepressing conditions). Total RNA was prepared and separated on a formaldehyde-agarose gel, transferred to nylon membranes, and hybridized as described previously (25, 27).

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Cloning of TUP1 homologs from K. lactis and S. pombe. Sequences homologous to TUP1 were identified in genomic DNA from K. lactis by low stringency Southern hybridization with sequences encoding the WD repeat region of S. cerevisiae TUP1, and these sequences were subsequently cloned (see Materials and Methods). The nucleotide sequence predicts that the K. lactis Tup1p protein consists of 682 amino acids and has 61% overall identity to S. cerevisiae Tup1p (75% identity in the WD repeat domain and 45% identity in other regions) (Fig. 1). The predicted Ssn6p-binding domain of K. lactis Tup1p is more highly conserved with the corresponding region of S. cerevisiae Tup1p (56 of 71 amino acids identical) than that of C. albicans Tup1p (39 out of 72 amino acids identical) (2).

View larger version (70K): [in this window] [in a new window]   FIG. 1.   Comparison of primary structure of Tup1p homologs in yeast. (A) Alignment of Tup1p homologs. Amino acids identical among three or four yeast species are in gray or black, respectively. The predicted Ssn6p-binding region and seven WD repeats are indicated. Triangles represent the positions where introns were inserted in S. pombe tup11+. ScTup1, S. cerevisiae Tup1p; KlTup1, K. lactis Tup1p; CaTup1, C. albicans Tup1p (2); SpTup11, S. pombe Tup11p. (B) Comparison of functional domains within Tup1p homologs. Values indicate the percent identity with S. cerevisiae Tup1p. Dotted and closed boxes represent the Ssn6p-binding and WD repeat domains, respectively. The bars indicate the functional domains of S. cerevisiae Tup1p.

K. lactis TUP1 complemented an S. cerevisiae tup1 mutation, but S. pombe tup11⁺ did not. To see whether the above-described K. lactis and S. pombe TUP1 homologs were able to functionally substitute for the S. cerevisiae TUP1 gene, we expressed each gene in S. cerevisiae YMH427 (MAT tup1::URA3 STE6-PHO5), which is null for TUP1. As shown below, the 226-bp promoter region of K. lactis TUP1 was sufficient to express this gene in S. cerevisiae. The S. pombe tup11⁺ cDNA was expressed under the control of the S. cerevisiae TUP1 promoter.

View larger version (77K): [in this window] [in a new window]   FIG. 2.   Regulation of Tup1p-repressed genes in S. cerevisiae tup1 mutants expressing Tup1p homologs. Total RNA samples were prepared from cells of tup1-disrupted strain YMH427 having the plasmid carrying S. cerevisiae TUP1 (lane 1), K. lactis TUP1 (lane 2), or S. pombe tup11+ (lane 3) or the vector plasmid YCp50 alone (lane 4). Each RNA sample (2 µg per lane) was separated on an agarose gel in the presence of formaldehyde, blotted onto a nylon membrane, and hybridized with 32P-labeled a-specific STE2, the glucose-repressed SUC2, or the oxygen-repressed ANB1 probe. The same membranes were rehybridized with 32P-labeled ACT1 as an internal marker. The ANB1 DNA fragments were also hybridized with the Tr transcript which is not regulated by TUP1.

S. pombe tup11⁺ encodes a transcriptional repressor. The above experiments raise the question of whether the S. pombe Tup11p protein serves as a transcriptional repressor in vivo. To examine this question, this protein was fused to the DNA-binding domain of the bacterial LexA protein and expressed in S. cerevisiae YMH465, which carries a CYC1-lacZ reporter gene containing four copies of the LexA DNA-binding site upstream of the CYC1 upstream activation sequence. -Galactosidase activities were determined, and the results (in Miller units) were as follows: for LexA-S. cerevisiae Tup1p, 7.3 ± 0.7; for LexA-S. pombe Tup11p, 7.5 ± 0.4; and for LexA, 25.7 ± 2.4 (values are the means of 11 independent measurements ± standard deviations). As expected, this reporter was repressed in cells bearing S. cerevisiae Tup1p fused to LexA relative to cells bearing LexA alone. Importantly, expression of the reporter in cells bearing the S. pombe Tup11p-LexA fusion was also repressed and this repression was equal to that seen in the cells bearing S. cerevisiae Tup1p-LexA. These data indicate that S. pombe Tup11p can function as a transcriptional repressor in S. cerevisiae.

Chimeric Tup11p proteins bearing the Ssn6p-binding region of S. cerevisiae Tup1p complement tup1 null phenotypes. As shown above, the S. pombe tup11⁺ gene was not able to complement the S. cerevisiae tup1 mutation in spite of its ability to repress transcription in S. cerevisiae. To localize the functional differences between the S. cerevisiae and S. pombe Tup1p proteins, we tested the ability of hybrid Tup1p proteins to functionally substitute for the S. cerevisiae Tup1p. Several S. cerevisiae-S. pombe chimeric Tup1p proteins were constructed and expressed under the control of the S. cerevisiae TUP1 promoter in the tup1 disruptant strain YMH427. For these experiments, each coding region was divided into three sections, roughly corresponding to the amino-terminal Ssn6p-binding domain, the central repression domain, and the C-terminal WD repeats of the S. cerevisiae protein. The PPP construct contained only S. pombe sequences, and the CCC construct contained only S. cerevisiae sequences. Hybrid constructs (PPC, PPC2, CCP, CCP2, and CPP) contained mixtures of S. pombe and S. cerevisiae sequences as described in Materials and Methods.

View larger version (36K): [in this window] [in a new window]   FIG. 3.   Identification of functional domains in S. pombe Tup11p by creation of chimeric proteins. The tup1 strain YMH427 was transformed with plasmids harboring the indicated S. cerevisiae-S. pombe Tup1p hybrids. The resulting transformants were assayed for mating ability, flocculation, and STE6-PHO5 expression (APase activity [in milliunits], the average of three measurements with a margin of error of <20%). The open boxes indicate regions derived from S. cerevisiae Tup1p, and the closed boxes indicate regions derived from S. pombe Tup11p. The amino acid positions of the junctions are indicated. Flo, flocculation; , nonflocculent; +, flocculent; Non, nonmating ability.

S. pombe Tup11p binds to 2p but not to Ssn6p of S. cerevisiae in vitro. Since the WD repeats of S. cerevisiae Tup1p were exchangeable with those of S. pombe Tup11p, we predicted that the S. pombe repeats would interact with DNA-binding proteins that recruit Tup1p to target promoters in S. cerevisiae. Therefore, we tested the ability of S. pombe Tup11p to bind to 2p in vitro. The 2p protein was fused to GST and expressed in E. coli. S. pombe Tup11p was transcribed and translated in vitro in the presence of ³⁵S-methionine. GST-2p or GST alone was purified from bacterial extracts with glutathione-Sepharose beads, and then equal amounts of the GST fusion proteins were incubated with in vitro-labeled S. pombe Tup11p. Bead-bound fractions were analyzed by SDS-PAGE (Fig. 4A). S. pombe Tup11p bound to GST-2p (Fig. 4A, lane 1) but not to GST alone (Fig. 4A, lane 2). These data are consistent with our findings that the chimeric proteins CCP and CPP, whose WD repeats are derived from S. pombe Tup11p, function in S. cerevisiae.

View larger version (36K): [in this window] [in a new window]   FIG. 4.   Interaction of S. pombe Tup11p with S. cerevisiae 2p and Ssn6p in vitro. (A) Binding of S. pombe Tup11p to 2p. In vitro 35S-labeled S. pombe Tup11p was incubated with beads bound to GST-2p (lane 1) or to GST alone (lane 2). After the beads were washed, proteins bound to the beads were analyzed by SDS-PAGE. Shown are autoradiograms detecting the labeled proteins. Input (lane 3) represents 10% of the labeled Tup11p used in the binding reaction. (B) Binding of S. pombe Tup11p with S. cerevisiae Ssn6p. In vitro 35S-labeled S. cerevisiae Ssn6p was incubated with beads bound to GST-S. cerevisiae Tup1p (7-253) (lane 1), GST-S. pombe Tup11p (1-298) (lane 2), or GST alone (lane 3). Input (lane 4) represents 10% of the labeled Ssn6p used in the binding reaction.

S. pombe Tup11p binds specifically to histones H3 and H4. S. cerevisiae Tup1p binds to histones H3 and H4 directly (6). Interestingly, the histone binding domain of S. cerevisiae Tup1p was replaced with an analogous region of S. pombe Tup11p in the chimeric protein CPP, which effectively substituted for S. cerevisiae Tup1p in vivo. This observation raises the possibility that S. pombe Tup11p also binds to histones. To test this idea, we isolated histones from S. cerevisiae and incubated them with the GST-Tup1p fusion proteins described above. As expected, GST-S. cerevisiae Tup1p bound to histones H3 and H4 but not to H2A and H2B (Fig. 5A, lanes 2 and 3). Strikingly, the GST-S. pombe Tup11p fusion protein also bound specifically to histones H3 and H4 (Fig. 5A, lanes 4 and 5). In contrast, GST alone did not bind effectively to any of the histones (Fig. 5A, lanes 6 and 7).

View larger version (57K): [in this window] [in a new window]   FIG. 5.   Interaction of S. pombe Tup11p with histones. (A) GST pull-down analysis. Semipurified histones were incubated with beads bound to GST-S. cerevisiae Tup1p (amino acids 7 to 253) (lanes 2 and 3), GST-S. pombe Tup11p (amino acids 1 to 298) (lanes 4 and 5), or GST alone (lanes 6 and 7). After washing, bound (lanes 2, 4, and 6) and unbound (lanes 3, 5, and 7) fractions were analyzed by SDS-PAGE and visualized by staining with Coomassie brilliant blue R-250. Input (lane 1) represents the total amount of histones used in each binding reaction. (B) Far-Western blot analysis. Samples of yeast histone proteins were separated by SDS-PAGE, electroblotted onto a nylon membrane, and probed with an 35S-labeled S. cerevisiae Tup1p (ScTup1) or S. pombe Tup11p (SpTup11) probe. A parallel lane was stained with Coomassie brilliant blue R-250. (Coomassie).

S. pombe Tup1p homologs function as repressors in S. pombe. During the course of this study, another sequence similar to Tup1p was added to the S. pombe protein database (accession no. 2555018), which we propose to call tup12⁺. Since these two TUP1 homologs might provide redundant functions, we created mutants with double disruptions in tup11⁺ and tup12⁺. We examined the expression of a glucose-repressible gene, fbp1⁺, in these strains, since in S. cerevisiae TUP1 regulates some glucose-repressible functions. Total RNA was prepared from isogenic wild-type (JY741), tup11 (JY741-tup11U), tup12 (JY741-tup12L), or tup11 tup12 double mutant (JY741-tup11U, tup12L) cells cultivated in either high (8%) or low (0.1% with 3% glycerol) glucose medium. Northern blot analysis showed that fbp1⁺ was repressed in wild-type cells and the tup11 and tup12 single disruptants in the presence of 8% glucose (Fig. 6, lanes 1, 3, and 5). However, fbp1⁺ expression was derepressed in the tup11 tup12 double disruptants (Fig. 6, lane 7), reaching levels of up to 50% of those observed in the wild-type strain under derepressing conditions (Fig. 6, lane 2). Interestingly, under derepressing conditions the transcription levels of fbp1⁺ also increased two- and threefold in the tup12 and tup11 tup12 double disruptants, respectively (Fig. 6, lanes 6 and 8), but not in the tup11 disruptant (Fig. 6, lane 4). Thus, tup12⁺ may limit fbp1⁺ expression even under derepressing conditions. These data indicate that Tup11p and Tup12p are required for full repression of fbp1⁺ and that these proteins provide redundant functions in S. pombe.

View larger version (37K): [in this window] [in a new window]   FIG. 6.   Transcription of the glucose-repressed fbp1+ gene in tup11 and tup12 mutants. Total RNA samples were prepared from cells of JY741 (tup11+ tup12+) (lanes 1 and 2), JY741-tup11U (tup11::ura4+) (lanes 3 and 4), JY741-tup12L (tup12::LEU2) (lanes 5 and 6), JY741-tup11U, or tup12L (tup11::ura4+ tup12::LEU2) (lanes 7 and 8) grown in repressing (R; 8% glucose; lanes 1, 3, 5, and 7) or derepressing (D; 0.1% glucose and 3% glycerol; lanes 2, 4, 6, and 8) conditions. Each RNA sample (10 µg per lane) was separated on an agarose gel in the presence of formaldehyde, blotted onto a nylon membrane, and hybridized with a 32P-labeled PCR product harboring the coding region of fbp1+. The same membrane was rehybridized with a 32P-labeled leu1+ PCR product as an internal control. Normalized levels of fbp1+ RNA relative to leu1+ (values averaged from three independent experiments) are shown below the lane numbers. The level of fbp1+ in the wild-type strain (WT) under derepressing conditions was set to 1.0 for comparison purposes. Standard deviations were within 10% for all samples except lanes 6 and 8, for which standard deviations were 21 and 27%, respectively.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Tup1p is well characterized as a corepressor in S. cerevisiae (7, 38). Here, we report the cloning and characterization of TUP1 homologs from K. lactis and S. pombe. We found that the TUP1 gene from the budding yeast K. lactis functionally substituted for S. cerevisiae TUP1. Interestingly, the TUP1 gene from another budding yeast, C. albicans, also complemented an S. cerevisiae tup1 mutation (2), indicating that the mechanism of transcriptional repression by Tup1p is conserved in at least three separate budding yeasts.

Although the S. pombe tup11⁺ gene was not able to complement the S. cerevisiae tup1 mutation, our data strongly suggest that S. pombe tup11⁺ also encodes a corepressor protein. First, the amino acid sequence of S. pombe Tup11p is significantly homologous to Tup1p homologs identified in a number of species, including Neurospora crassa Rco-1 (E value of basic local alignment search technique for protein sequences [BLASTP], e¹²⁴), Dictyostelium discoideum Tup1 (E value of BLASTP, e¹¹⁵), and S. cerevisiae Tup1p (E value of BLASTP, e¹⁰⁰). Second, when S. pombe Tup11p was artificially recruited to a promoter region in S. cerevisiae via a LexA DNA-binding domain, it repressed transcription of the downstream gene. This same strategy was used by others to define S. cerevisiae Tup1p, Ssn6p, and Rpd3p as corepressors (14, 15, 32). Third, a chimeric protein (CPP), which contains the 72 N-terminal amino acids of S. cerevisiae Tup1p and the 543 C-terminal amino acids of S. pombe Tup11p functionally complemented the S. cerevisiae tup1 null allele. This result indicates that S. pombe Tup11p has corepressor functions but is defective in interacting with S. cerevisiae Ssn6p, as confirmed by our biochemical analysis. Fourth, S. pombe Tup11p binds to S. cerevisiae 2p, suggesting it can be recruited to target promoters by DNA-binding repressors. Fifth, S. pombe Tup11p specifically interacts with histones H3 and H4, as does S. cerevisiae Tup1p. Many studies indicate that interaction of Tup1p with histones likely plays an important role in the repression mechanism (6, 7). Finally, disruption of tup11⁺ in combination with disruption of tup12⁺, another TUP1 homolog gene in S. pombe, causes a defect in glucose repression of fbp1⁺. Thus, S. pombe Tup1p homologs likely function as transcriptional repressors, as does their S. cerevisiae counterpart.

Our finding that S. pombe Tup11p does not bind to S. cerevisiae Ssn6p is consistent with the low conservation of the amino-terminal regions of these proteins (Fig. 1). In contrast, the Ssn6p-binding region of S. cerevisiae Tup1p is highly conserved in Tup1ps of other the budding yeasts (K. lactis and C. albicans), which can complement the S. cerevisiae tup1 mutations. Interestingly, a database search has revealed a homolog of Ssn6p in S. pombe (accession no. 3116127, E value of BLASTP, e¹²⁹), and in separate studies, we have shown that S. pombe Tup11p interacts with S. pombe Ssn6p by two-hybrid and in vitro binding analyses (data to be presented elsewhere). These findings suggests that Tup11p also forms a corepressor complex with Ssn6p to repress transcription in S. pombe.

The primary structure of the histone binding region in S. cerevisiae Tup1p is not well conserved in either S. pombe Tup11p or K. lactis Tup1p (Fig. 1). This region is also significantly shorter in C. albicans Tup1p, as only 117 amino acids are found between the Ssn6p-binding domain (2) and WD repeats of this protein in contrast to the 260 amino acids in the analogous region of S. cerevisiae Tup1p. This diversity of sequence suggests that higher order folding of these regions may be important for forming a histone binding domain.

The mechanism of transcriptional repression by Tup1p-Ssn6p might also be conserved in higher eukaryotes. The Groucho protein in flies and the TLE1 protein in humans have WD repeats and bind to histone proteins (22). TLE1 also binds to a mammalian Ssn6p homolog (9). Groucho binds directly to Hairy-related or Runt domain DNA-binding proteins through the sequence WRPW or WRPY, respectively (8), indicating that it may be recruited to promoters in a manner similar to that of yeast Tup1p. However, the DNA-binding proteins known to interact with S. cerevisiae Tup1p do not have WRPW and WRPY motifs, so the details of recruitment are not conserved.