INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Transcription initiation by RNA polymerase II (RNAPII) requires the assembly of a multiprotein complex at the promoter. This complex consists of RNAPII, general transcription factors, and the SRB (suppressor of RNA polymerase B) or mediator proteins which are involved in the positive and negative regulation of transcription. Assembly of this preinitiation complex can be made to occur in a stepwise fashion in vitro (11), but most transcriptional initiation events in Saccharomyces cerevisiae appear to use a preassembled RNAPII holoenzyme containing most of the essential factors (57).

TFIIH is a general transcription factor that has an associated helicase, as well as protein kinase activity (26, 27, 52). One of the major targets for phosphorylation by TFIIH in the transcription initiation complex is the unique carboxy-terminal domain (CTD) of the largest subunit of RNAPII. This CTD consists of tandem repeats of the consensus sequence Tyr-Ser-Pro-Thr-Ser-Pro-Ser, which is repeated 52 times in human RNAPII and 26 or 27 times in S. cerevisiae (1, 22). Roles for the CTD during transcription initiation, promoter clearance, chain elongation, and transcript processing have been suggested (9, 43). RNAPII molecules with a hypophosphorylated CTD are preferentially recruited to the initiation complex in vitro (18, 40), whereas the elongating RNAPII in vivo usually has a hyperphosphorylated CTD (12, 45). The purified RNAPII holoenzyme contains hypophosphorylated forms of the CTD (37), whereas a purified form of the elongating RNAPII complex has a CTD that is heavily phosphorylated (47). Therefore, it seems that the transcription cycle involves cyclical phosphorylation and dephosphorylation of the RNAPII CTD (24).

Gene-specific roles have also been suggested for the CTD. Phosphorylation of the CTD by the kinase activity of the RNAPII holoenzyme component Srb10p before RNAPII binds to the promoter prevents the formation of productive transcription initiation complexes at repressed promoters (29). In addition, although the CTD is essential for yeast cell growth, likely reflecting its requirement for mRNA synthesis, at least several genes in S. cerevisiae can be transcribed by RNAPII molecules lacking the CTD (38, 42).

Dephosphorylation of the CTD must occur in order to regenerate the nonphosphorylated form of the enzyme that appears to be recruited to promoters. A CTD phosphatase activity was originally purified from HeLa cells and subsequently from S. cerevisiae (13, 14). The activity of this HeLa cell CTD phosphatase is stimulated by the RAP74 subunit of the general transcription factor TFIIF, and this stimulation can be inhibited by TFIIB (15). Partial cDNAs encoding the human protein were originally identified in a screen for RAP74-interacting proteins (4). The C-terminal domain of human RAP74 that interacts with the human CTD phosphatase FCP1 is necessary and sufficient for RAP74-mediated stimulation of CTD phosphatase activity in vitro (4). The homologous FCP1 gene is essential in S. cerevisiae, and the yeast Fcp1 protein also interacts directly with the RAP74 subunit of yeast TFIIF (3). The phosphatase catalytic domain of Fcp1p resembles similar domains found in a number of other database proteins of unknown function and has been designated the FCP homology (FCPH) domain (3, 36). This domain contains the phosphatase motif DXDX(T/V) ( = hydrophobic residue) at its catalytic center (3, 19, 36). This motif is characteristic of a new family of small-molecule phosphotransferases and phosphohydrolases (21) and is essential for Fcp1p to function in S. cerevisiae (36). It is different from the phosphatase motifs of the three classified protein phosphatase families (8), and Fcp1p might therefore be the founding member of a new class of eukaryotic protein phosphatases. Recombinant human or yeast Fcp1p can dephosphorylate the CTD (19, 36, 41) and the artificial substrate p-nitrophenylphosphate in vitro (36). Genome-wide expression studies show that Fcp1p is generally required for transcription by RNAPII in S. cerevisiae (36). The human CTD phosphatase was shown to function in recycling RNAPII in vitro (19).

In this study, we further delineated regions of yeast Fcp1p that are functionally important in vivo with the goal of understanding what factors influence CTD phosphatase. The integrity of the BRCT domain in Fcp1p is essential for function. We demonstrated a direct interaction between Fcp1p and TFIIB and showed that the binding sites in Fcp1p for TFIIB and RAP74 are very similar. We also identified related amino acid sequence motifs in TFIIB and RAP74 that are involved in the binding of TFIIB and RAP74 to Fcp1p. Strains with mutations in RAP74 were used to show that RAP74 utilizes this motif to interact with Fcp1p in vivo. As well, we showed that Fcp1p is able to strongly activate transcription when tethered to a promoter by a heterologous DNA-binding domain, suggesting that Fcp1p interacts with the RNAPII holoenzyme and recruits it to the promoter. This transcriptional activation by Fcp1p is mediated in part by its interaction with RAP74, a known component of the yeast RNAPII holoenzyme (34, 37).

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Plasmids. Plasmids for the in vivo analysis of FCP1 deletions (Table 1) were constructed with a two-step strategy using the pRS series of CEN/ARS plasmids (53). pFK1 (pRS314-FCP1 positions 1 to 732) was digested with PstI, which cuts at codon 134 of FCP1, and XhoI, which cuts in the pRS314 polylinker downstream of the FCP1 transcription termination sequence. PCR products obtained by using primers MK44 (5'-GGG CTG CAG ATG CCT TCG ATG GTG TAC C-3') and MK46 (5'-GGG CTC GAG CTA GAT TAA CGT GTA GGG TTT TTC ATC C-3') or primers MK44 and MK47 (5'-GGG CTC GAG CTA TTT ATA AGT GCT AGG GTT CTT GG-3') (stop codons are underlined) were cut with PstI and XhoI and cloned into the pRS314-FCP1 vector backbone to create plasmids containing versions of fcp1 coding for amino acids 1 to 594 and 1 to 557, respectively, but lacking the transcription termination sequence. The termination sequence was subsequently inserted into these constructs cut with XhoI and KpnI by using a PCR product obtained with primers MK48 (5'-GGG CTC GAG CAG CYC AGA TGC CGT ATC TTT CC-3') and MK49 (5'-GGG GGT ACC AGT ACT TGT TGA GTA TTT AGG GG-3') and digested with KpnI and XhoI to give pMK96 (pRS314-fcp1-5 positions 1 to 594) and pMK97 (pRS314-fcp1-6 positions 1 to 557). Plasmid pMK98 (pRS314-fcp1-7 positions 134 to 594) was constructed by excising the PstI/KpnI fragment from plasmid pMK96 and inserting it into pMK90 (pRS314-fcp1-3 positions 134 to 732) cut with PstI and KpnI.

Protein purification. GST fusion proteins were expressed in Escherichia coli DH5 cells. Cells were grown at 30°C to an optical density at 600 nm of 0.4 and induced with 1 mM isopropyl--D-thiogalactopyranoside (IPTG). After 3 h of induction, the cells were harvested and resuspended in 1 M buffer A (10 mM Tris HCl [pH 7.9], 1 mM EDTA, 1 mM dithiothreitol [DTT], 1 M NaCl). The cells were lysed by sonication, and the supernatant after centrifugation was mixed with glutathione-Sepharose 6B (Pharmacia). The beads were washed four times with 1 M buffer A and then two times with 0.1 M buffer A. The bound proteins were eluted with 0.1 M ACB (10 mM HEPES [pH 7.9], 1 mM EDTA, 1 mM DTT, 10% glycerol, 100 mM NaCl) containing 50 mM reduced glutathione and dialyzed into 0.1 M ACB. The purified proteins were analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and bound to fresh glutathione-Sepharose 6B at the indicated concentrations for affinity chromatography experiments.

Protein-protein interaction assays. GST and various GST-TFIIB and GST-RAP74 fusion proteins were coupled to glutathione-Sepharose 6B at the indicated concentrations. The columns (40 µl) were equilibrated with 200 µl of 0.1 M ACB containing 5 mg of bovine serum albumin (BSA)/ml and next with 400 µl of 0.1 M ACB containing 1 mg of BSA/ml. Columns were loaded with 20 µl of rabbit reticulocyte lysate from the TNT transcription-translation system (Promega) programmed with 0.4 µg of the various plasmid DNAs which had been diluted 10-fold in 0.1 M ACB containing 1 mg of BSA/ml. The columns were washed with 400 µl of 0.1 M ACB, and the bound proteins were eluted in 120 µl of 0.1 M ACB containing 50 mM glutathione. Thirty-microliter aliquots of the eluted proteins were analyzed by SDS-PAGE and autoradiography.

Western blot analysis. Protein extracts from yeast strains grown in synthetic complete medium lacking the appropriate amino acids were prepared by glass bead lysis in the presence of trichloroacetic acid as described previously (36). After SDS-PAGE and transfer to a nitrocellulose membrane, protein analysis was performed using either the monoclonal antibody G2 (a generous gift from V. Svetlov and R. Burgess) directed against a conserved domain within the largest subunit of yeast RNAPII, an affinity-purified polyclonal antibody against Fcp1p (36), or a monoclonal antibody against the vacuolar H⁺-ATPase VMA (Molecular Probes) using standard procedures.

NMR studies. All nuclear magnetic resonance (NMR) experiments were performed at 25°C on a Varian Inova 600-MHz spectrometer equipped with a pulsed-field gradient unit and a triple-resonance (¹H, ¹³C, and ¹⁵N) probe with an actively shielded z gradient. Sensitivity-enhanced gradient two-dimensional (2D) (¹H and ¹⁵N) heteronuclear single quantum correlation (HSQC) spectra were recorded with a ¹⁵N sweep width of 1,600 Hz centered at 118 ppm and 64 complex points t1. An 8,000-Hz 1H sweep width centered at 4.753 was recorded with 512 complex points t2.

Yeast strain construction. All of the strains used in this study (Table 2) are derivatives of W303-1A (ade2-1 can1-100 trp1-1 leu2-3,112 his3-11,15 ura3-1 ssd1-d2). Integration of the TFG1 wild-type and tfg1-2 mutant alleles was done by cutting pMK99 and pMK100, respectively, with EcoRI and transforming W303-1A diploid cells. Yeast cells were transformed with linearized plasmids using the lithium acetate procedure, and integrants were selected on synthetic complete medium (SC) plates without His. Integration was confirmed by PCR, and haploid spores were obtained using a hydrophobic sporulation protocol (5), creating strains YMK202 and YMK204. The mutation was confirmed by sequencing PCR-amplified genomic DNA.

-Galactosidase assays. Liquid-culture assays to measure -galactosidase produced by yeast strains that were transformed with the LexA operator-lacZ fusion plasmid pSH18-34 and various LexA-Fcp1p derivatives were performed by using cells permeabilized with Sarkosyl essentially as previously described (35). Cells were grown to mid-logarithmic phase in SC lacking Ura and His. The number of cells used for the various LexA-yFcp1 fusion proteins was adjusted in order to obtain reliable readings of optical density at 420 nm. For each measurement, -galactosidase activity was determined from three independent cultures and average values are given.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Regions of Fcp1p required for function in vivo. We showed previously that the catalytic FCPH domain of Fcp1p is essential for its function in vivo. We also found that cells of a strain expressing a truncated Fcp1p that lacks carboxy-terminal amino acid residues 627 to 732 are viable and that the same truncated protein is capable of functioning as a CTD phosphatase in vitro (3). To determine how much of the carboxy terminus of Fcp1p was dispensable for cell viability, a series of Fcp1p proteins that contained carboxy-terminal deletions were constructed with the choice of truncation site based on consideration of the predicted boundaries of the BRCT domain (amino acids 499 to 593) (Fig. 1A). This domain is found in a number of proteins that are involved in cell cycle checkpoint control in response to DNA damage (10), but its importance for CTD phosphatase activity is not clear. Each truncated form of Fcp1p was then tested by using a plasmid-shuffling protocol to replace a wild-type copy of FCP1 with a mutated version of the gene (Fig. 1B) (M. S. Kobor and J. Greenblatt, unpublished data for two other mutants). A form of the FCP1 gene encoding a protein that ended just after the BRCT domain at amino acid 594 (allele fcp1-5) supported viability, whereas a gene encoding a protein that ended at amino acid 557, which is in the middle of the two predicted BRCT subdomains (10), did not (Fig. 1B). These results suggested that the integrity of the BRCT domain in Fcp1p is important for the protein to function in vivo.

View larger version (68K): [in this window] [in a new window]   FIG. 1.   Domains of Fcp1p required for function in vivo. (A) Diagram showing predicted domain boundaries of Fcp1p and FCP1 alleles used in this study. The ability of these constructs to support yeast cell growth during the plasmid shuffling experiments whose results are shown in panels B and C is summarized on the right. (B) Strains with a wild-type copy of FCP1 on the URA3 CEN/ARS plasmid pRS316 and a chromosomal FCP1 deletion were transformed with TRP1 CEN/ARS plasmids carrying the wild-type FCP1 gene, no FCP1 gene, or an fcp1 gene encoding Fcp1p with a deletion of amino acid residues 595 to 732 or 558 to 732. The ability of these mutant alleles to complement the FCP1 chromosomal deletion was tested by plasmid shuffling on SC plates lacking Leu and Trp and containing 5-fluoroorotic acid (5 FOA) to counterselect the URA3 marker at 22°C. (C) FCP1 alleles encoding proteins that have either an amino-terminal deletion, a carboxy-terminal deletion, or a simultaneous deletion of both termini were tested for the ability to complement a chromosomal deletion of FCP1 by a plasmid shuffling assay at 22°C.

The BRCT domain is essential for mRNA synthesis and CTD dephosphorylation. To investigate further the importance of the BRCT domain in Fcp1p, we used site-directed mutagenesis to change its most conserved residue, namely, the tryptophan at position 575, to alanine. This residue is within the second predicted conserved region of the BRCT domain (10). We originally selected this residue exclusively based on its conservation among different BRCT domains because the structural and functional knowledge of BRTC domains was very limited when we began this study. Yeast strains carrying this allele, which we named fcp1-4, were viable at 30°C but did not grow at 37°C (Fig. 2A). To determine whether the W575A mutation in strains with the fcp1-4 allele caused a defect in mRNA synthesis, as do mutations in the catalytic FCPH domain, we shifted logarithmically growing cells to the nonpermissive temperature and measured the levels of poly(A)⁺ mRNA. There was a major decrease in the amount of poly(A)⁺ mRNA as soon as 1 h after the temperature shift (Fig. 2B). Consistent with this, we also found that the W575A mutation caused a defect in the dephosphorylation of RNAPII in vivo (Fig. 2C). In this experiment, we analyzed the relative amounts of hyperphosphorylated (RNAPIIO) and hypophosphorylated (RNAPIIA) forms of the largest subunit of RNAPII, Rpb1p, by probing Western blots of yeast extracts prepared at various time points after the shift to 37°C with a monoclonal antibody directed against a region of Rpb1p outside of the CTD. We were able to clearly distinguish the two forms of Rpb1p in these strains. Consistent with previous results (48), we observed an increase in the amount of RNAPIIO and a decrease in the amount of RNAPIIA even in wild-type strains after a shift to 37°C (Fig. 2C). However, this did not have an effect on mRNA synthesis as judged by the amount of poly(A)⁺ mRNA in these strains at the permissive temperature (Fig. 2B). At the permissive temperature, there was a larger amount of Rpb1p in the fcp1-4 mutant strain than in the wild-type strain. Interestingly, a much larger proportion of the Rpb1p in the fcp1-4 strain was present in either the hyperphosphorylated form or a variety of intermediate phosphorylated forms even before transfer to the nonpermissive temperature; at that time, the amount of the mutant Fcp1p in the cell was normal (Fig. 2C). A detailed analysis of the abnormal CTD phosphorylation pattern in the mutant strain at the permissive temperature will be described elsewhere (Kobor and Greenblatt, unpublished data). These results strongly suggest that the BRCT domain is important for the CTD phosphatase activity of Fcp1p in vivo. Moreover, the hypophosphorylated RNAPIIA form completely disappeared in the fcp1-4 strain upon a shift to 37°C; at that temperature, the mutant Fcp1p was partly degraded. The amount of the vacuolar H⁺-ATPase VMA, which we used as a loading control, did not change significantly (Fig. 2C). There was a very good correlation between the decrease in poly(A)⁺ mRNA and the change in the phosphorylation pattern of Rpb1p, giving further support to our earlier finding that inactivation of CTD phosphatase leads to a rapid shutdown of mRNA synthesis by RNAPII.

View larger version (55K): [in this window] [in a new window]   FIG. 2.   The BRCT domain is essential for mRNA synthesis and CTD dephosphorylation. (A) A strain with a W575A point mutation in the most conserved residue of the BRCT domain of Fcp1p is viable but has a temperature-sensitive phenotype. Yeast strains with wild-type FCP1 or the fcp1-4 mutation on TRP1 CEN/ARS plasmids in an fcp1::LEU2 background were grown on SC plates lacking Leu and Trp at 30 and 37°C. (B) Strains carrying the fcp1-4 allele have a severe defect in the synthesis of poly(A)+ mRNA after a shift to the nonpermissive temperature. Total cellular RNA was prepared from FCP1 and fcp1-4 strains at various times after a shift to 37°C. A 32P-labeled oligo(dT) probe was hybridized to 1 µg of total slot-blotted RNA (36). (C) The W575A mutation interferes with dephosphorylation of RNAPII. Whole-cell extracts prepared from FCP1 and fcp1-4 mutant strains at various times after a shift from 30 to 37°C were Western blotted with monoclonal antibody G2 to estimate the relative amounts of hyperphosphorylated and hypophosphorylated forms of the Rpb1p subunit. The same blots were also probed with antibodies against Fcp1p and VMA.

Fcp1p interacts directly with the general transcription factor TFIIB. Previous studies of the human system have shown that TFIIB can regulate CTD phosphatase activity in vitro (15). Consistent with this, a two-hybrid screen using human Fcp1a (4) as bait led to the identification of human TFIIB as an interacting protein (J. Langlois and J.G., unpublished data). Therefore, we examined whether the corresponding yeast proteins are able to interact with each other in vitro. Fcp1p could bind to recombinant full-length yeast TFIIB in an affinity chromatography experiment (Fig. 3A), and Fcp1p also bound a portion of the TFIIB that is present in yeast whole-cell extract (Kobor and Greenblatt, unpublished data). TFIIB consists of an amino-terminal zinc ribbon domain (amino acids 1 to 100) and a core domain (amino acids 100 to 345) containing two cyclin-related repeats (31) (Fig. 3B). When a series of GST fusion proteins containing various domains of TFIIB were used as ligands in affinity chromatography experiments, only the fusion proteins containing either the complete core domain or the first cyclin-like repeat were able to bind Fcp1p (Fig. 3C). This suggested that Fcp1p interacts with the first cyclin-related repeat of TFIIB.

View larger version (32K): [in this window] [in a new window]   FIG. 3.   Fcp1p interacts with the first cyclin-like repeat in TFIIB. (A) Fcp1p binds to recombinant full-length TFIIB. Purified recombinant His6-tagged yeast TFIIB was bound to Ni-agarose (Qiagen) microcolumns at the indicated concentrations, and 35S-labeled Fcp1p made by in vitro transcription and translation was tested for binding. After washing, the bound protein was eluted with high-salt buffer containing 1 M NaCl and visualized by autoradiography after SDS-PAGE. (B) Diagram showing predicted domain boundaries of yeast TFIIB. (C) The first cyclin-like repeat of the TFIIB core domain mediates the binding of Fcp1p. The indicated GST-yTFIIB fusion proteins were purified from bacteria and coupled to glutathione-Sepharose columns at a concentration of 4 mg/ml, and then 35S-labeled Fcp1p was tested for binding. The eluted Fcp1p protein was analyzed by SDS-PAGE and visualized by autoradiography.

Similar regions of Fcp1p bind TFIIB and RAP74. We used the TFIIB core domain as an affinity chromatography ligand to test which regions of yeast Fcp1p are involved in the Fcp1p-TFIIB interaction. Versions of Fcp1p with carboxy-terminal and amino-terminal deletions were tested for the ability to bind immobilized GST-TFIIB (amino acids 100 to 345). We had previously shown that two adjacent regions of yeast Fcp1p containing amino acid residues 457 to 666 and 667 to 732 can each bind yeast RAP74 independently (3) (see also Fig. 4B). These two regions were also able to bind to GST-TFIIB (amino acids 100 to 345) (Fig. 4A). In this case, a thioredoxin (TRX) fusion protein containing amino acid residues 457 to 666 of Fcp1p bound GST-TFIIB (amino acids 100 to 345) more strongly than a TRX-Fcp1 (amino acids 667 to 732) fusion protein. We were unable to detect binding when we tried to further subdivide these regions of Fcp1p, suggesting that more extensive deletions influence the proper folding of these regions (Kobor and Greenblatt, unpublished data). Of the proteins with carboxy-terminal deletions that we tested, the protein lacking amino acid residues 667 to 732 of Fcp1p was only slightly compromised in TFIIB binding whereas removal of amino acids 627 to 732 completely abolished the interaction (Fig. 4A). We next used these same constructs to perform a more detailed mapping of the binding regions for RAP74 in Fcp1p in order to compare them to the binding regions for TFIIB and also to the regions dispensable for viability. Similar to what was observed with TFIIB, Fcp1p (amino acids 1 to 626) had no detectable binding to RAP74 (Fig. 4B). The TRX-Fcp1 (amino acids 457 to 666) fusion protein bound somewhat less strongly to RAP74 than did the TRX-Fcp1 (amino acids 667 to 732) fusion protein or the full-length protein. It appeared that Fcp1p (amino acids 1 to 666) bound much less strongly to RAP74 than to TFIIB compared to the full-length protein. Therefore, we concluded that two adjacent regions in Fcp1p can interact with both TFIIB and RAP74 and that the interaction with both factors is abolished when amino acid residues distal to amino acid 627 of Fcp1p are deleted. One of the binding sites in Fcp1p, namely, amino acids 457 to 666, binds TFIIB about 4 times as strongly as it binds RAP74, whereas the other binding site in Fcp1p, namely, amino acids 667 to 732, binds RAP74 at least 10 times as strongly as it binds TFIIB.

View larger version (62K): [in this window] [in a new window]   FIG. 4.   TFIIB and RAP74 bind in similar ways to Fcp1p. Various 35S-labeled portions of Fcp1p or thioredoxin (TRX)-Fcp1 fusion proteins made by transcription and translation in vitro were chromatographed over affinity columns containing the indicated concentrations of either GST-yTFIIB (amino acids 100 to 345) (A) or GST-yRAP74 (amino acids 649 to 735) (B). GST was used as a control in all binding experiments. The columns were washed, and bound proteins were eluted with reduced glutathione, analyzed by SDS-PAGE, and visualized by autoradiography. Panel C summarizes the relative strengths with which various portions of Fcp1p bind to GST-yTFIIB (amino acids 100 to 345) and GST-yRAP74 (amino acids 649 to 735) presented in panels A and B.

An amino acid sequence motif common to TFIIB and RAP74 mediates their binding to Fcp1p. The similar patterns of yeast TFIIB and RAP74 binding to Fcp1p prompted us to search for similar amino acid sequences in these two general transcription factors. The sequence KEFGK is present in both yeast proteins TFIIB and RAP74 (Fig. 5A). The amino acid sequences in these regions are also similar in the TFIIB and RAP74 proteins found in humans, Xenopus laevis, Drosophila melanogaster, and Caenorhabditis elegans.

View larger version (36K): [in this window] [in a new window]   FIG. 5.   A conserved amino acid sequence motif is involved in the binding of TFIIB and RAP74 to Fcp1p. (A) Comparison of sequences from the first cyclin-related repeat of yeast TFIIB and the CTD of yeast RAP74. The alignment can be extended to TFIIB and RAP74 from other species (not shown). Mutations (mut) that were created are indicated below (RAP74) or above (TFIIB) the alignment. s.c., S. cerevisiae. (B) A K201E amino acid change in TFIIB reduces binding to Fcp1p. Wild-type GST-TFIIB and a K201E derivative (amino acids 100 to 345) were bound to glutathione-Sepharose at the indicated concentrations, and affinity chromatography with 35S-labeled Fcp1p was performed as described in the legend to Fig. 3. (C) Amino acid residues within the KEFGK motif of RAP74 are important for binding to Fcp1p. A series of amino acid changes were introduced into GST-RAP74 (amino acids 649 to 735) as indicated in panel A, and the resulting proteins were tested for binding to 35S-labeled Fcp1p in affinity chromatography experiments. All of the mutations that were tested reduced the binding of Fcp1p to GST-RAP74 (amino acids 649 to 735).

View larger version (51K): [in this window] [in a new window]   FIG. 6.   RAP74 (amino acids 649 to 735) and RAP74 (amino acids 649 to 735) K695E-K699E are properly folded protein domains. The double mutation of K695E and K699E does not affect the overall folding of the RAP74 carboxy-terminal region, as shown by 2D 1H-15N HSQC NMR spectra of the wild-type (WT) (A) and K695E-K699E mutant (MT) (B) proteins.

Effect of mutations in the Fcp1p-binding motif of RAP74 in vivo. The gene encoding the yeast homologue of the RAP74 subunit of human TFIIF is called TFG1 (30). To test the importance for yeast cell growth of the Fcp1p-binding motif of RAP74, a tfg1-2/TFG1 diploid yeast strain was constructed. The tfg1-2 allele encodes RAP74 (K695E, K699E), which fails to bind Fcp1p in vitro (Fig. 5C). After sporulation, we obtained a viable haploid tfg1-2 strain. Growth of this strain was normal under all of the conditions that we tested, with the exception that growth was somewhat impaired in the presence of the DNA-damaging agent methanosulfonic methyl ester (MMS) (Fig. 7A). This phenotype was also observed in fcp1 mutant strains (Kobor and Greenblatt, unpublished data), therefore providing a phenotypic link between Fcp1p and RAP74.

View larger version (84K): [in this window] [in a new window]   FIG. 7.   Effects in vivo of mutations in the Fcp1p-binding domain of RAP74. (A) The K695E-K699E mutation in the CTD of RAP74 causes sensitivity to the DNA-damaging drug MMS. TFG1 wild-type and tfg1-2 mutant strains were plated at 2,000 cells per plate on SC plates lacking His and containing 0.01% MMS and grown at 30°C for 3 days. Control plates did not contain any MMS. (B) Synthetic phenotypes of fcp1 tfg1-2 double-mutant yeast strains. Tenfold serial dilutions of TFG1 and tfg1-2 mutant strains that also had a chromosomal deletion of FCP1 and carried the indicated FCP1 alleles on TRP1 CEN/ARS plasmids were grown for 3 days at various temperatures on SC plates lacking His, Trp, and Leu. All of the strains with mutant fcp1 alleles are more temperature sensitive in the tfg1-2 background than in the TFG1 background, except for the strain carrying the fcp1-5 allele, which encodes a protein that lacks the RAP74-binding site. The diagram shows the positions of the mutations in Fcp1p. (C) Western blot analysis of fcp1 tfg1 double-mutant strains. Whole-cell extracts prepared from strains carrying the FCP1, fcp1-1, and fcp1-4 alleles in the TFG1 and tfg1-2 background were grown at 22°C, subjected to SDS-PAGE, and Western blotted with monoclonal antibody G2 to estimate the relative amounts of the hyperphosphorylated and hypophosphorylated forms of the Rpb1p subunit. A similar experiment was also performed with strains carrying the FCP1, fcp1-3, and fcp1-5 alleles that were grown at 30°C. The blots were underexposed to visualize the difference in phosphorylation status of the CTD of Rpb1p in the double-mutant strains. The same blots were also probed with antibodies against VMA. wt, wild type.

fcp1 tfg1-2

RAP74 can be a target for transcriptional activation in vivo. A number of components of the RNAPII holoenzyme complex have been shown to be able to activate transcription when artificially tethered to a promoter via a fusion to a heterologous DNA-binding domain (17, 25). This activation is thought to be due to enhanced recruitment of the RNAPII holoenzyme (50). Since human Fcp1p is a component of a human RNAPII holoenzyme complex and since Fcp1p interacts with TFIIB and RAP74, which are components of the yeast RNAPII holoenzyme (37), we tested whether yeast Fcp1p, when fused to the DNA-binding domain of LexA, could activate the transcription of a lacZ reporter gene containing eight upstream LexA-binding sites in vivo. As shown in Fig. 8A, a LexA-Fcp1p fusion strongly activated transcription. Removing an increasing number of amino acid residues from the carboxy terminus of the LexA-Fcp1 fusion protein resulted in a sharp drop in transcriptional activation. LexA-yFcp1 (amino acids 1 to 666) had only about 5% of the activation potential of the full-length fusion protein, and the shorter fusion protein LexA-yFcp1 (amino acids 1 to 626), which does not bind RAP74 or TFIIB (Fig. 4), failed to activate transcription above the levels of the LexA DNA-binding domain alone. All of these fusion proteins were expressed to comparable levels in the yeast cells (M. S. Kobor, L. D. Simon, and J. Greenblatt, unpublished data). Therefore, these results revealed a good correlation between the ability of Fcp1p to activate transcription in vivo when brought into the vicinity of a promoter and its ability to bind to RAP74 and TFIIB in vitro.

View larger version (35K): [in this window] [in a new window]   FIG. 8.   Transcriptional activation by LexA-Fcp1 fusion proteins. (A) A full-length LexA-Fcp1 fusion protein and a series of carboxy-terminal and amino-terminal deletion constructs were tested for the ability to activate the lacZ reporter construct pSH18-34, which has eight LexA-binding sites upstream of the promoter. The ability to activate transcription is correlated with ability to bind RAP74 and TFIIB. -Gal, -galactosidase; MU, Miller units. (B) The interaction between Fcp1p and RAP74 is important for transcriptional activation by LexA-Fcp1p. LexA or a LexA-Fcp1 fusion protein was expressed in a TFG1 or tfg1-2 strain, and transcriptional activation of reporter constructs having one, two, or eight LexA-binding sites was measured. Transcriptional activation by LexA-Fcp1p is synergistic and is dependent on the interaction with RAP74. This dependence is more pronounced at higher levels of activation. For simplicity, the values for LexAp alone were omitted but they were always below 2 Miller units. All measurements were done in triplicate, and the averages are given. Standard deviations were below 20% for all points. AD, activation domain.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Fcp1p is an RNAPII CTD phosphatase required for most or all mRNA synthesis in yeast. A characteristic feature of Fcp1p is the presence of a BRCT domain, which is commonly found in proteins that are involved in checkpoint control in response to DNA damage. We demonstrated that the integrity of this domain is essential for Fcp1p to function in S. cerevisiae. Even at the permissive temperature, a strain with a point mutation in the BRCT domain of Fcp1p accumulates an excessive proportion of hyperphosphorylated RNAPII. Upon a shift to the nonpermissive temperature, the hypophosphorylated form of RNAPII disappears and mRNA production is shut down. This further supports our earlier finding, obtained by using yeast strains carrying the fcp1-1 and fcp1-2 alleles, that Fcp1p is responsible for dephosphorylating the CTD in vivo and that failure to do so can result in a shutdown of RNAPII transcription (36). A recent X-ray structure analysis of the BRCT domain of the human DNA repair protein XRCC1 showed that its two predicted subdomains form one compact domain consisting of a four-stranded parallel beta sheet surrounded by three alpha helixes with extensive intramolecular contacts (58). Based on this structure, it is very likely that our partial deletion of the BRCT domain of Fcp1p, which was not able to complement a chromosomal FCP1 deletion, profoundly affected the overall fold and structure of the domain. The conserved tryptophan residue that we altered to create a temperature-sensitive mutant forms part of the highly conserved hydrophobic pocket and makes contacts with a number of other important residues. Our mutation might create a version of the BRCT domain that is partly unfolded and quickly unfolds upon a shift to the nonpermissive temperature, leading to degradation of the mutant protein. BRCT domains have been shown to be dimerization domains involved in the formation of either homodimers (55) or heterodimers (56) and can also interact with proteins like RNA helicase A that do not contain BRCT domains (2). Interestingly, mutating the conserved Trp residue in XRCC1 to Asp did not affect the interaction with DNA ligase III (56). It will be very important to identify the interacting protein partner(s) of this essential domain of Fcp1p. No other protein within the general transcription machinery is known to have a BRCT domain.

In addition to the known interaction between Fcp1p and RAP74 (3), our studies revealed a direct interaction with the general transcription factor TFIIB, which had been shown previously to inhibit RAP74-stimulated CTD phosphatase activity (15). We discovered that helix E1 within the first cyclin-like repeat in the core domain of yeast TFIIB mediates the interaction with Fcp1p. This region was implicated previously in the formation of the TFIIB-TBP-DNA complex and binding to the acidic activation domain of VP16. Mutations within this region cause a temperature-sensitive phenotype and have an effect on basal transcription in vitro (7, 20). However, these mutated TFIIB proteins retained the ability to respond to the acidic activator Gal4-VP16 in vitro and in vivo, suggesting that this region is important for basal but not activated transcription (7, 20). The additional interaction of this region with CTD phosphatase is intriguing, and more work is necessary to distinguish between effects caused by the inability to form the TFIIB-TBP-DNA complex and failure to interact with Fcp1p.

Although we did not further address the function of TFIIB in the regulation of CTD phosphatase activity, we showed that there are two independent binding sites for TFIIB within the carboxy-terminal region of Fcp1p. These binding sites are very similar to the binding sites for RAP74, although the stronger binding site for RAP74 lies within amino acids 667 to 732 whereas the stronger binding site for TFIIB lies within amino acids 457 to 666. A truncated version of Fcp1p that ends at amino acid 626 and lacks binding sites for TFIIB and RAP74 is still active in a CTD phosphatase assay in vitro (3).

Nevertheless, our in vivo deletion analysis demonstrated that a protein that fails to interact with both TFIIB and RAP74 is able to support growth. CTD phosphatase interacts directly with RNAPII (15; Kobor and Greenblatt, unpublished data), and there might be additional interactions with components of the RNAPII holoenzyme which might suffice in vivo to bring CTD phosphatase into the vicinity of its substrate and compensate for the lack of strong RAP74-Fcp1p and TFIIB-Fcp1p interactions under most circumstances. Another explanation for this finding is the possibility that an unidentified CTD phosphatase distinct from Fcp1p can compensate for the loss of Fcp1p interaction with TFIIB and RAP74. TFIIH, SRB10, and CTDK-1 are distinct CTD kinases which might phosphorylate the CTD on different residues during the transcription cycle. Therefore, the possible existence of distinct CTD phosphatases specific for different phosphorylated residues cannot be excluded.

We identified a previously unrecognized short amino acid sequence of high similarity between TFIIB and RAP74 that maps to a CTD of RAP74 and helix E1 in at the carboxy-terminal end of the first cyclin repeat of TFIIB (6, 44). The degree of amino acid conservation within this motif is particularly high among the TFIIB and RAP74 proteins of S. cerevisiae, humans, and X. laevis, although less so for D. melanogaster and C. elegans, suggesting that it has an important biological function. We showed that this motif mediates the interactions of TFIIB and RAP74 with Fcp1p. Our data suggest that the charge distribution of this region is a major component of the interactions and are consistent with earlier findings that RAP74 can stimulate CTD phosphatase activity and that this stimulation can be inhibited by TFIIB (15). Our data imply that TFIIB may compete with RAP74 for binding to Fcp1p. However, we have not been able so far to determine whether the interactions between RAP74 and TFIIB with the carboxy-terminal half of Fcp1p are mutually exclusive or whether the three proteins can form a ternary complex.

Just as deletion of the portion of Fcp1p that binds TFIIB and RAP74 in strains that carry the fcp1-5 allele has little effect on cell growth, so does the tfg1-2 mutation in RAP74 that prevents it from binding to Fcp1p. However, the sensitivity of the tfg1-2 mutant to the DNA-damaging agent MMS provides an indirect phenotypic link to FCP1. The failure of CTD phosphatase to interact with RAP74 may lead to loss of viability when there is severe DNA damage or other stresses which we have not yet discovered. Importantly, double-mutant analysis strongly supports a functional interaction between RAP74 and the carboxy-terminal region of Fcp1p in vivo. Strains carrying a mutant fcp1 allele are more temperature sensitive in the tfg1-2 background than in the TFG1 background, except in the case of fcp1-5, which lacks the RAP74-binding sites. These results are consistent with the hypothesis that stimulation of CTD phosphatase activity by RAP74 occurs in vivo and becomes more important when CTD phosphatase activity is weakened by mutations in Fcp1p outside its RAP74-binding region. These results also imply that the KEFGK motif in RAP74 mediates its interaction with Fcp1p in vivo. Consistent with this, we found a larger proportion of intermediately phosphorylated and hyperphosphorylated Rpb1p in double-mutant strains, except when the RAP74-binding carboxy-terminal region of Fcp1p was deleted.

Our studies also revealed that the amino-terminal region of Fcp1p has some function in vivo. Deletion of both the amino-terminal region and CTD of Fcp1p is lethal. Although we do not know whether the resulting protein is unstable or nonfunctional, an fcp1-3 tfg1-2 mutant strain carrying an amino-terminal deletion of Fcp1p is temperature sensitive. Taken together, these data suggest that the amino-terminal region of Fcp1p has an important function that is revealed only when Fcp1p cannot interact with RAP74. Strains that lack either the carboxy terminus or the amino terminus of Fcp1p have a larger proportion of intermediately phosphorylated Rpb1p.

Our observation that transcriptional activation by LexA-Fcp1p is strongly reduced in the tfg1-2 mutant strain provided further evidence that RAP74 interacts with Fcp1p in vivo. The artificial recruitment of the RNAPII holoenzyme by a LexA-Fcp1p fusion protein and subsequent activation of the reporter construct transcription may occur because Fcp1p can associate with RNAPII holoenzyme complexes (4). We have observed that a LexA-Fcp1p fusion is able to complement a chromosomal fcp1 deletion, suggesting that it can perform the normal cellular functions of Fcp1p (Kobor, Simon, and Greenblatt, unpublished data). The substantial effect of the tfg1-2 mutation on activation by LexA-Fcp1p indicated that this activation probably involves a direct interaction with RAP74. The fact that portions of Fcp1p that do not contain the phosphatase catalytic domain are able to activate transcription indicates that activation by LexA-Fcp1p is independent of CTD phosphatase activity and suggests that recruitment of the RNAPII holoenzyme is the most likely mechanism for this activation process. The acidic carboxy-terminal region of Fcp1p could also act as a bona fide transcriptional activator. Interestingly, previous studies have shown an important role for human RAP74 in transcriptional activation by serum response factor (33), and an interaction between the activation domain of the model activator Gal4-VP16 and RAP74 has been reported (60). Although the interaction sites for the serum response factor, VP16, and Fcp1p on RAP74 do not overlap, it is tempting to speculate that some endogenous yeast activators act at least partially through interaction with RAP74 in vivo. The fact that the effect of the tfg1-2 mutation on activation by LexA-Fcp1p decreases when fewer LexA-binding sites are present suggests that LexA-Fcp1p also interacts with some protein other than RAP74. This protein could be TFIIB or some other component of the RNAPII holoenzyme.

CTD phosphatase activity may be regulated partly at the level of transcriptional initiation, as is suggested by its physical and functional interactions with TFIIF and TFIIB. It has been reported that CTD phosphatase activity is necessary to recycle RNAPII after a round of transcription is completed in vitro (19), and it is possible that both TFIIF and TFIIB play a role in the temporal regulation of this process. Alternatively, it is also possible that CTD phosphatase is active at a different step during the transcription cycle. Although TFIIF binds directly to RNAPII and helps to assemble RNAPII into the preinitiation complex, it can also interact with elongating RNAPII and stimulate its rate of chain elongation (28, 32, 49). A role for TFIIB in steps after holoenzyme recruitment to the promoter has also been proposed (51). One or more factors in HeLa nuclear extract affect the ability of Fcp1p to dephosphorylate the CTD in transcription elongation complexes in vitro (19, 39), and the human immunodeficiency virus (HIV) Tat protein, which stimulates chain elongation by RNAPII, binds to human Fcp1p and inhibits its CTD phosphatase activity (41). In light of our studies reported here, it may be important that the interaction between the HIV Tat protein and human FCP1 involves the BRCT domain of human FCP1 (J. Archambault and J. Greenblatt, unpublished data). It is conceivable, therefore, that CTD phosphatase activity is differentially regulated at more than one stage of the transcription cycle. As well, the possibility cannot be excluded that Fcp1p has relevant substrates other than the CTD in vivo. In this regard, it is interesting that a number of the general transcription factors, including RAP74, have been reported to be phosphoproteins (46, 54), and phosphorylation of RAP74 has been suggested to be important for HIV Tat-mediated stimulation of transcriptional elongation (59). In addition, it is possible that Fcp1 has functions that are not related to its phosphatase activity, as suggested by the finding that human FCP1 can act as an elongation factor in vitro independently of its catalytic function (19).