Physical and functional interaction of the yeast corepressor Tup1 with mRNA 5'-triphosphatase.

The Tup1-Ssn6 complex is an important corepressor in Saccharomyces cerevisiae that inhibits transcription through interactions with the basal transcription machinery and by remodeling chromatin. In a two-hybrid screen for factors that interact with the Schizosaccharomyces pombe Tup1 ortholog, Tup11, we isolated the pct1+ cDNA. The pct1+ gene encodes an mRNA 5'-triphosphatase, which catalyzes the first step of mRNA capping reactions. Pct1 did not interact with the S. pombe Ssn6 ortholog. In vitro glutathione S-transferase pull-down experiments revealed that Pct1 binds to the WD repeat regions of Tup11 and the functionally redundant Tup12 protein. Similarly, the S. cerevisiae Tup1 protein associates with the mRNA 5'-triphosphatase encoded by the CET1 gene. The highly conserved C-terminal domain of Cet1 interacts with Tup1 in vitro, and Tup1-Ssn6 complexes co-purify with the Cet1 protein, indicating that in vivo interactions also occur between these proteins. Over-expression of CET1 compromised repression of an MFA2-lacZ reporter gene that is subject to Tup1-Ssn6 repression. These genetic and biochemical interactions between Tup1-Ssn6 and Cet1 indicate that the capping enzyme associated with RNA polymerase II is a target of the corepressor complex.

Tup1 also mediates repression through the organization of chromatin. Ssn6-Tup1 interacts physically with class I histone deacetylases (HDACs), 1 and simultaneous mutation of the three HDAC genes, RPD3, HOS1, and HOS2, abolishes repression by Ssn6-Tup1 (27). The Ssn6-Tup1 complex interacts with at least two of these HDACs, Rpd3 and Hos2 (27). Hda1 has also been reported to interact with Tup1 in vitro and is required for deacetylation of histones H3 and H2B at the stress-responsive ENA1 promoter, whereas Rpd3 deacetylates the coding region of this gene (28). Microarray analyses reveal that expression of many Tup1-regulated genes is increased somewhat in the absence of Hda1, again suggesting that this HDAC plays a role in Tup1 functions (29). Tup1 interacts preferentially with underacetylated isoforms of histones H3 and H4 in vitro, genes repressed by Tup1 associate with underacetylated histones in vivo, and mutations in H3 and H4 synergistically reduce repression (30 -33). Altogether these observations indicate that Ssn6-Tup1, when tethered to promoter regions, recruit HDACs to deacetylate adjacent nucleosomes, which in turn promotes Tup1-histone interactions and stable recruitment of the corepressor (32).
Recently, a new role for Ssn6-Tup1 as a coactivator has been described (34). The ATF/CREB (activating transcription factor/ cAMP-response element-binding protein) repressor Sko1 recruits Ssn6-Tup1 to target promoters and represses transcription of the GRE2, AHP1, HAL1 genes, etc. in the absence of osmotic stress. However, in the presence of osmotic stress, the Hog1 kinase phosphorylates Sko1 and converts the Sko1-Ssn6-Tup1 repressor into an activator that recruits the SAGA (Spt-Ada-Gcn5-acetyltransferase) histone acetylase and SWI/SNF nucleosome remodeling complexes to osmotic-inducible promoters. At the GAL1 promoter, the Cti6 protein links SAGA with Ssn6-Tup1 to overcome Tup1-mediated repression, even though the corepressor remains bound to the promoter in this case (35).
In vitro protein binding experiments and two-hybrid analyses have defined at least three protein-protein interaction domains in the 713-amino acid Tup1 protein. The N-terminal 72 amino acids of Tup1 are required for interaction with Ssn6 and for self-multimerization (36). The histone-binding domain comprises amino acids 73-385 (30). WD repeats (amino acids 333-706) in the C-terminal region of Tup1 likely form a sevenbladed ␤-propeller structure (37) that interacts with DNAbinding repressors including ␣2 (7). Tup1-like proteins are found in other yeasts and higher eukaryotes, e.g. Groucho in flies (38) and TLE (transducin-like enhancer of split) in humans (39). Interestingly, the fission yeast Schizosaccharomyces pombe has two functionally redundant TUP1 genes, tup11 ϩ and tup12 ϩ (40). These Tup1 orthologs have the ability to bind to HDACs and histones H3 and H4 (40 -42). The tetratricopeptide repeat protein Ssn6 is also highly conserved from yeasts to humans (43), indicating that Ssn6-Tup1 corepressor complexes are commonly used for transcriptional repression in eukaryotes.
To further understand Tup1 functions in S. pombe, we performed a two-hybrid screen to isolate interacting proteins. We report here that Pct1 specifically interacts with S. pombe Tup11. Pct1 is an mRNA 5Ј-triphosphatase, which catalyzes the first step of mRNA capping reactions. In vitro and in vivo interactions between Tup1 and the S. cerevisiae mRNA 5Јtriphosphatase, Cet1, were also observed. Over-expression of Cet1 compromises Ssn6-Tup1 repression. These biochemical and genetic interaction data suggest that the capping enzyme associated with RNA polymerase II is a target of the corepressor complex.
Two-hybrid Screening-The S. cerevisiae strain L40, harboring lex-Aop-lacZ and lexAop-HIS3 reporters, was first transformed with plas-mid pBTM-tup11 to express a chimera of Tup11 and the DNA-binding domain of the E. coli LexA protein. Expression of LexA-Tup11 fusion protein in this transformant was confirmed by immunoblot using anti-LexA antibody (Clontech). Strain L40 containing pBTM-tup11 was further transformed with a library of S. pombe cDNA fragments fused to the activation domain (GAD) of the S. cerevisiae Gal4 protein. From about 6 million independent transformants, we isolated 134 histidine prototrophic (His ϩ ) clones, grown on plates lacking leucine, tryptophan, and histidine, based on the activation of the HIS3 reporter gene. Among these His ϩ clones, 90 clones developed blue color on 5-bromo-4-chloro-3-indolyl-␤-D-galactopyranoside (X-gal) plates due to the expression of the lacZ reporter gene. Positive clones were grown on nonselective plates and then Leu ϩ Trp Ϫ clones, which contained the pGAD library plasmid but not the pBTM-tup11 plasmid, were isolated. Plasmid DNAs were recovered from white, Leu ϩ Trp Ϫ His Ϫ colonies and were transformed into E. coli MC1066. These pGAD library plasmids were used for retransformation of strain L40 with pBTM-tup11 to confirm the twohybrid interactions. The nucleotide sequences of the insert cDNAs were determined by the dye-terminator method with the ABI PRISM 310 sequencing system using a primer from the GAL4 region.
GST-Tup1 fusion proteins were expressed in E. coli strain DH5␣ and purified using glutathione-Sepharose 4B beads (Amersham Biosciences) following the manufacturer's protocol. In vitro labeled proteins (see above) or yeast extracts were incubated with comparable amounts (as determined by SDS-PAGE analysis) of the different GST fusion proteins bound to beads in GST binding buffer at room temperature for 1 h or at 4°C overnight. Beads then were collected by centrifugation in a microcentrifuge at 500 ϫ g for 2 min. The beads were washed five times with GST binding buffer. The washed beads were resuspended directly in 1ϫ SDS-PAGE sample buffer. All samples were separated by SDS-PAGE and autoradiographed for detection of the 35 S-labeled protein or blotted for Western analysis.
T7-Cet1-His 6 proteins were purified using two affinity columns, His⅐Bind and T7⅐Bind columns (Novagen). GST-Tup1C and GST alone were purified using glutathione-Sepharose column (Amersham Biosciences). T7-Cet1-His 6 proteins (10 g) were incubated with 10 g of GST-Tup1C or GST at 30°C for 1 h. Glutathione-Sepharose beads (20 l) were added and incubated at room temperature for 1 h. After the beads were washed, proteins bound to the beads were separated by SDS-PAGE and visualized by Coomassie Brilliant Blue stain. Immunoblots were performed using anti-T7 antibody to detect T7-Cet1.

S. pombe Pct1
Interacts with Tup11-Using LexA-Tup11 as bait, we screened a library of S. pombe cDNAs fused to the activation domain of the S. cerevisiae Gal4 protein (see "Experimental Procedures"). Those colonies showing histidine prototrophy and ␤-galactosidase activity were recovered and sequenced. One of the positive clones had a complete cDNA region identical to SPAC644.04 on chromosome I of S. pombe. Recently, SPAC644.04 was shown to encode an mRNA 5Јtriphosphatase, which removes the 5Ј-terminal phosphate of mRNA in the first step of mRNA capping reactions and is referred to as pct1 ϩ (47). The nucleotide sequence of the genomic pct1 ϩ DNA predicts that this gene has one intron. The intron sequence of pct1 ϩ was removed in our pct1 ϩ cDNA sequence, as expected.
Tup1 contains a C-terminal WD repeat domain. These domains are found in a variety of proteins with differing functions and are thought to mediate protein-protein interactions. To exclude the possibility that Pct1 binds to WD repeat proteins in general, we examined the interactions of Pct1 with the WD repeat protein Ste4. In a two-hybrid assay, LexA-Ste4 did not interact with Gal4-Pct1 (data not shown), indicating that Pct1 does not generally bind all WD repeat proteins.

Tup1 Interaction with mRNA 5Ј-Triphosphatase
In S. cerevisiae, Tup1 forms a complex with the Ssn6 protein (4 -6). Some sequence-specific DNA-binding proteins associate with both Tup1 and Ssn6 (7,8,14). Therefore, we asked whether Pct1 also interacts with the S. pombe Ssn6 ortholog (SPBC23E6.09) predicted by the S. pombe sequencing project. Pct1 did not bind to the S. pombe Ssn6 ortholog in the twohybrid assay (data not shown). This S. pombe Ssn6 two-hybrid construct is functional and interacts with the LexA-Tup11 construct used in the above experiments. These results suggest that S. pombe Tup11, but not Ssn6, interacts with the mRNA 5Ј-triphosphatase Pct1.
S. pombe Tup11 and Tup12 Interact with Pct1 through Their WD Repeat Domains-The S. cerevisiae Tup1 and the S. pombe Tup11 proteins have at least three protein-binding domains, i.e. an N-terminal domain that interacts with Ssn6 (36), a middle domain that interacts with histones H3 and H4 (30), and the C-terminal WD repeat domain mentioned above (7). To define the domain of Tup11 that interacts specifically with Pct1, two Tup11 deletion derivatives were tested for in vitro interactions with Pct1 (Fig. 1). The N-terminal half containing the Ssn6-and histone-binding domains (amino acid positions 1-298) and the C-terminal half containing the WD repeats domain (amino acids 297-614) were fused to glutathione Stransferase (named GST-Tup11N and GST-Tup11C, respectively; Fig. 1A) and expressed in E. coli. The pct1 ϩ gene was transcribed and translated in vitro in the presence of [ 35 S]methionine. GST-Tup11N, GST-Tup11C, or GST alone was purified from bacterial extracts with glutathione-Sepharose beads, and then comparable amounts of the GST fusion proteins were incubated with in vitro labeled Pct1. Pct1 bound to GST-Tup11C (Fig. 1B, lane 3) but not to GST-Tup11N (lane 2) or GST alone (lane 6). This in vitro binding experiment indicates that the Pct1 protein associates with the Tup11 protein through its WD repeat domain.
Because S. pombe has two redundant TUP1 orthologs, tup11 ϩ and tup12 ϩ (40), we examined the ability of Tup12 to bind to Pct1 (Fig. 1). Tup12 was also separated into two regions, an N-terminal half (Tup12N, amino acid positions 67-271) and a C-terminal half containing the WD repeats (Tup12C, amino acids 272-586), that were expressed in E. coli as GST fusions (Fig. 1A). Consistent with our results above, GST pull-down experiments indicated that in vitro translated Pct1 bound to GST-Tup12C (Fig. 1B, lane 5) but not to GST-Tup12N (lane 4). We conclude that Pct1 associates with both Tup11 and Tup12 through their WD repeat regions.
S. cerevisiae Tup1 Interacts with Cet1-To determine whether the corepressor Tup1 interacts with mRNA 5Ј-triphosphatase encoded by the CET1 gene in the budding yeast S. cerevisiae, we tested for interaction between these two S. cerevisiae proteins. We were unable to detect any two-hybrid interactions between S. cerevisiae Tup1 and Cet1 when using either full-length or deletion constructs (data not shown). How  Tup1N (lanes 2, 6, 10, and 14), GST-Tup1C (lanes 3, 7, 11, and 15), or GST alone (lanes 4, 8, 12, and 16). After the beads were washed, proteins bound to the beads were analyzed by SDS-PAGE. Shown are autoradiograms detecting the labeled proteins. Input (In, lanes 1, 5, 9, and 13) represents 10% of the labeled Cet1 proteins used in the binding reaction. B, deletion derivatives of Cet1 used for in vitro interaction assays. Full-length and truncated Cet1 are indicated schematically along with their ability to interact with Tup1. The amino acid positions of the proteins are indicated. labeled Cet1 as described above (Fig. 2). GST-Tup1C interacted well with full-length Cet1 (Fig. 2A, lane 3), whereas GST-Tup1N bound only slightly to Cet1 (lane 2). No interaction of GST alone with Cet1 was observed ( Fig. 2A, lane 4). These results reveal that the WD repeat domain of S. cerevisiae Tup1 can also associate with mRNA 5Ј-triphosphatase as was observed for S. pombe Tup11 and Tup12.
To determine whether Tup1 and Cet1 interact directly, we purified recombinant forms of these proteins expressed in E. coli. The full-length Cet1 protein was tagged with T7 (11 amino acids) in the N terminus and six copies of His in the C terminus and was purified using two affinity chromatography columns (Fig. 3A, lane 1). GST-Tup1C and GST alone were purified by glutathione-Sepharose affinity chromatography (Fig. 3A, lanes  2 and 3). T7-Cet1-His 6 was mixed with GST-Tup1C or GST and pulled down with glutathione-Sepharose beads. Comparable amounts of GST-Tup1C and GST were detected by Coomassie Brilliant Blue stain (Fig. 3A, lanes 4 and 5). Immunoblotting of the precipitates using an anti-T7 antibody showed that T7-Cet1-His 6 bound to GST-Tup1C (Fig. 3B, lane 2) but not to GST alone (lane 3). These results clearly indicate that Tup1 and Cet1 can interact directly.
To determine whether Tup1 associates with Cet1 in vivo, we expressed a GST-Cet1 hybrid protein or GST alone in the S. cerevisiae wild type or tup1 mutant cells. Protein extracts were prepared from these cells, and we analyzed proteins that copurified with GST-Cet1 or with GST alone using glutathione-Sepharose beads. Immunoblotting of the purified proteins using a polyclonal anti-Tup1 antibody revealed that whereas GST alone did not pull down Tup1 in either strain (Fig. 4, lanes 6  and 8), GST-Cet1 co-purified with Tup1 in the wild type strain (lane 5) but not in the tup1 strain (lane 7).
Tup1 Interaction with mRNA 5Ј-Triphosphatase encoding the GST-Cet1 hybrid protein or GST alone. As described above, protein extracts were prepared from these cells, and proteins that co-purified with GST-Cet1 or with GST alone were analyzed by immunoblotting using a monoclonal anti-HA antibody (Fig. 5). Equal levels of the Tup1-HA protein were expressed in the cells containing GST-Cet1 or GST alone (Fig.   5, lanes 1 and 2). Tup1-HA co-purified with GST-Cet1 (Fig. 5

, lane 3) but not with GST alone (lane 4).
Because Tup1 is in complex with Ssn6 in vivo (5, 6), we predicted that the Ssn6 protein would also co-purify with GST-Cet1. To test this hypothesis, we examined proteins pulled down in the above experiments with a polyclonal antisera raised against Ssn6 (Fig. 6). As predicted, Ssn6 co-purified with GST-Cet1 but not with GST alone (lanes 3 and 4), consistent with the above observations. Therefore we conclude that Tup1-Ssn6 corepressor complex interacts with mRNA 5Ј-triphosphatase Cet1 in S. cerevisiae.
Genetic Interaction of Corepressor Tup1 and mRNA 5Ј-Triphosphatase Cet1-To determine whether over-expression of Cet1 might affect Tup1-mediated repression, we over-expressed the GST-Cet1 fusion protein in wild type strains containing an MFA2-lacZ reporter gene (Table I). Transcription of the MFA2-lacZ gene is normally repressed in haploid ␣ cells through recruitment of Tup1-Ssn6 by the ␣2-Mcm1 repressor complex. Because ␣2 is not expressed in haploid a cells, the MFA2-lacZ gene is not repressed in these cells. We found that repression of the reporter gene was partially lost in ␣ cells upon over-expression of the GST-Cet1 protein (MFA2.1␣ with pCyGST-CET1U). Although the reporter gene was not fully derepressed compared with expression levels observed in a cells (MFA2.1a with pRD56), the loss of repression in ␣ cells was similar to that observed upon mutation of histones or mutation of RNA polymerase II mediator components (19,22,24,26,30). Cet1 over-expression did not affect the expression level of the corepressors Tup1 and Ssn6, as the amounts of Tup1 and Ssn6 are fairly equal in cells over-expressing GST-  lanes 1 and 3) or GST (G, lanes 2 and 4) and incubated with glutathione-Sepharose beads as described in the legend for Fig. 4. Immunoblots were performed using anti-Ssn6 antibody. Input (lanes 1 and 2) represents 10% of protein extracts used in the binding reactions.  5 and 6). Immunoblots were performed using anti-HA antibody (lanes 1-4). Input (lanes 1 and 2) represents 10% of the protein extracts used in the binding reactions. Tup1 Interaction with mRNA 5Ј-Triphosphatase Cet1 or GST alone (Figs. 4 -6, lanes 1 and 2). In contrast, over-expression of GST-Cet1 did not further increase expression of the MFA2-lacZ gene in a cells. These data indicate that Cet1 levels influence the function of the Tup1-Ssn6 corepressor. Derepression of the reporter gene in the Cet1-over-expressing strain might be because of sequestration of Tup1 from the repression complex. Alternatively, the increased amount of Cet1 might bypass normal sequestration of Cet1 by Tup1. In either case, these genetic data strongly suggest a functional, in vivo interaction between Tup1 and Cet1.
Hrs1/Pgd1 (hereafter referred to as Med3), a component of the RNA polymerase II transcriptional mediator complex (49,50), also interacts with Tup1 and Ssn6 (18). Med3 over-expression also derepressed transcription of the genes repressed by Tup1 and Ssn6 (18), indicating that the mediator complex is another target of the repressor complex. Because these two targets represent different steps in gene expression, it is possible that interference with both of these pathways simultaneously would lead to a greater effect on repression than does interference with either one alone. Therefore we over-expressed both Cet1 and Med3 in ␣ cells and measured MFA2-lacZ expression (Fig. 7). Simultaneous over-expression of GST-Cet1 and GST-Med3 resulted in an additive loss of repression. A 6-fold higher ␤-galactosidase activity was observed in the cells over-expressing both proteins compared with cells containing empty vectors, whereas over-expression of GST-Cet1 caused a 3.5-fold and over-expression of GST-Med3 a 2.5-fold loss of repression. This additive effect of Cet1 and Med3 overexpression suggests that Tup1-Ssn6 independently targets these proteins. These findings are also consistent with observations that Tup1 interacts with Cet1 and Med3 through separate domains (18).
We tried to determine whether cet1 mutant cells exhibited loss of Tup1-Ssn6-mediated repression. Because the CET1 gene is essential for cell growth (51), 2 we constructed a temperature-sensitive cet1 allele using the improved degron method (52). After shifting cet1-td cells containing the MFA2-lacZ reporter gene to the nonpermissive temperature, the HAtagged Cet1 protein was rapidly degraded within 30 min (data not shown). However, no derepression of the MFA2-lacZ gene was observed 30, 60, or 120 min after shifting to 37°C. Another group has reported that conditional mutations in Ceg1, a partner of Cet1 in RNA capping, lead to destabilization of mRNAs (53). Thus, it is possible that any increased transcription of lacZ that occurred in our experiments upon Cet1 loss would be mitigated by increased mRNA turnover, precluding our ability to analyze directly the effects of Cet1 loss on Tup1-mediated repression.

DISCUSSION
S. cerevisiae Tup1 and its orthologs appear to function through conserved mechanisms of transcriptional repression. In this study, we have discovered a new aspect of these functions, mediated through interactions of the S. pombe Tup11 corepressor with Pct1, an mRNA 5Ј-triphosphatase that catalyzes the first step of mRNA capping reactions. Pct1 binds to Tup11 in vivo and in vitro. Moreover, Pct1 binds to the WD repeats of the Tup12 protein, which is redundant in function with Tup11. S. cerevisiae Tup1 also interacts with S. cerevisiae Cet1 in vitro and in vivo, indicating that interaction between Tup1 corepressor and mRNA 5Ј-triphosphatase is conserved in yeasts. Over-expression of Cet1 compromises Ssn6-Tup1 repression, indicating that the interactions between Tup1 and Cet1 impact Tup1 functions.
Both Tup1 and Ssn6 are associated with GST-Cet1 in vivo.
Our in vitro binding experiments indicate that Tup1 binds directly to Cet1 and Tup11 and Tup12 bind, probably directly, to Pct1. However, S. pombe Ssn6 did not interact with Pct1 in the two-hybrid assay, and the interaction between Ssn6 and Cet1 in vivo may not be direct. Interestingly, Tup1 binds to the histones H3 and H4, but Ssn6 does not (30). Together, these observations are consistent with the idea that Tup1 plays the main role in repression by the Ssn6-Tup1 complex (1). However, Med3 interacts with both Tup1 and Ssn6, supporting additional roles for Ssn6 in Ssn6-Tup1 corepressor functions (18). Genetic screens have identified a number of genes that are required for Tup1 repression, including multiple components of RNA polymerase II holoenzyme: Med3, Srb7, Srb8, Srb10, Srb11, Sin4, Rox3, and Rgr1 (18 -26). These findings strongly suggest that subunits of RNA polymerase II holoenzyme are targets of Tup1 repression. Med3 also binds the Tup1 protein (18), and when over-expressed, Med3 compromises Ssn6-Tup1 repression just as Cet1 does. Srb7 also associates with Tup1, and disruption of the Tup1-Srb7 interaction causes derepression of a target gene (19). The mRNA capping enzyme in S. cerevisiae is composed of Cet1 and Ceg1 (mRNA guanylyltransferase) and is recruited to the phosphorylated C-terminal domain of the largest subunit of RNA polymerase II (54,55). Therefore, at least three proteins associated with RNA polymerase II appear to be targeted by Tup1. Med3 and Cet1 seem to be independent targets of Tup1, because the effects of overexpression of Med3 and Cet1 are additive. The reason the corepressor targets multiple proteins is not known, nor is it clear whether Tup1 targets all of these polymerase II-associated proteins at every promoter or whether it interacts with different targets at different promoters. Different tetratricopeptide repeats in Ssn6 are required for repression of different genes, and differential changes in H3 and H4 acetylation associated with Tup1-regulated genes have been observed (27,33). Thus, different repressive architectures may exist at different genes regulated by this corepressor. mRNA capping is considered a co-transcriptional event rather than a post-transcriptional event because it occurs when the transcript is only 25-30 nucleotides long (56,57), coincident with the time of C-terminal domain phosphorylation and recruitment of the capping enzyme (54,55). Interestingly the capping enzymes are also bound to and up-regulated by the elongation factor Spt5 and HIV (human immunodeficiency virus) Tat protein (58,59). The association of Tup1 with the mRNA capping enzyme might inhibit capping functions. However, the purified recombinant version of the C-terminal WD repeat region of Tup1, which binds to Cet1, did not inhibit the RNA 5Ј-triphosphatase activity of Cet1. 2 Recently Cet1 was identified as a general repressor of RNA polymerase II transcription that functions at the reinitiation step of the transcription cycle (60). It is possible that Tup1 utilizes this repressor activity of Cet1 to down-regulate transcription of target genes.