Rsp5 WW Domains Interact Directly with the Carboxyl-terminal Domain of RNA Polymerase II*

RSP5 is an essential gene inSaccharomyces cerevisiae and was recently shown to form a physical and functional complex with RNA polymerase II (RNA pol II). The amino-terminal half of Rsp5 consists of four domains: a C2 domain, which binds membrane phospholipids; and three WW domains, which are protein interaction modules that bind proline-rich ligands. The carboxyl-terminal half of Rsp5 contains a HECT (homologous to E6-AP carboxyl terminus) domain that catalytically ligates ubiquitin to proteins and functionally classifies Rsp5 as an E3 ubiquitin-protein ligase. The C2 and WW domains are presumed to act as membrane localization and substrate recognition modules, respectively. We report that the second (and possibly third) Rsp5 WW domain mediates binding to the carboxyl-terminal domain (CTD) of the RNA pol II large subunit. The CTD comprises a heptamer (YSPTSPS) repeated 26 times and a PXY core that is critical for interaction with a specific group of WW domains. An analysis of synthetic peptides revealed a minimal CTD sequence that is sufficient to bind to the second Rsp5 WW domain (Rsp5 WW2) in vitro and in yeast two-hybrid assays. Furthermore, we found that specific “imperfect” CTD repeats can form a complex with Rsp5 WW2. In addition, we have shown that phosphorylation of this minimal CTD sequence on serine, threonine and tyrosine residues acts as a negative regulator of the Rsp5 WW2-CTD interaction. In view of the recent data pertaining to phosphorylation-driven interactions between the RNA pol II CTD and the WW domain of Ess1/Pin1, we suggest that CTD dephosphorylation may be a prerequisite for targeted RNA pol II degradation.

RSP5 is an essential gene in Saccharomyces cerevisiae and was recently shown to form a physical and functional complex with RNA polymerase II (RNA pol II). The amino-terminal half of Rsp5 consists of four domains: a C2 domain, which binds membrane phospholipids; and three WW domains, which are protein interaction modules that bind proline-rich ligands. The carboxyl-terminal half of Rsp5 contains a HECT (homologous to E6-AP carboxyl terminus) domain that catalytically ligates ubiquitin to proteins and functionally classifies Rsp5 as an E3 ubiquitin-protein ligase. The C2 and WW domains are presumed to act as membrane localization and substrate recognition modules, respectively. We report that the second (and possibly third) Rsp5 WW domain mediates binding to the carboxyl-terminal domain (CTD) of the RNA pol II large subunit. The CTD comprises a heptamer (YSPTSPS) repeated 26 times and a PXY core that is critical for interaction with a specific group of WW domains. An analysis of synthetic peptides revealed a minimal CTD sequence that is sufficient to bind to the second Rsp5 WW domain (Rsp5 WW2) in vitro and in yeast two-hybrid assays. Furthermore, we found that specific "imperfect" CTD repeats can form a complex with Rsp5 WW2. In addition, we have shown that phosphorylation of this minimal CTD sequence on serine, threonine and tyrosine residues acts as a negative regulator of the Rsp5 WW2-CTD interaction. In view of the recent data pertaining to phosphorylation-driven interactions between the RNA pol II CTD and the WW domain of Ess1/Pin1, we suggest that CTD dephosphorylation may be a prerequisite for targeted RNA pol II degradation.
RSP5 (reversion of Spt phenotype) is an essential gene in the yeast Saccharomyces cerevisiae. The gene was originally isolated in a screen for suppressors of SPT3 mutations (1). SPT3 is part of the multicomponent protein named SAGA (Spt-Ada-Gcn5 acetyltransferase), which has been implicated in at least two facets of RNA pol II 1 activity. These include the function of the TATA box-binding protein Spt15 and the activity of Gcn5 histone acetyltransferase (1,2). Moreover, Rsp5 was recently shown to form a physical and functional complex with RNA pol II (3,4).
In addition to its interaction with RNA pol II, Rsp5 possesses multiple and diverse substrates, suggesting that Rsp5 affects a wide range of cellular processes. For example, RSP5 is known to down-regulate the activity of several plasma membraneassociated proteins, including Fur4 uracil permease, Gap1 amino acid permease, and the integral membrane protein proton ATPase (5)(6)(7)(8). Rsp5 was recently proposed to be a component of a complex regulating an early stage of endocytosis (9). Mutations in RSP5 have also been shown to regulate stability of a minichromosome (10) and the cytoplasmic/mitochondrial distribution of Mod5, an enzyme that modifies tRNA (11). In addition, Rsp5-mediated ubiquitination has been implicated in mitochondrial inheritance (12).
The primary structure of Rsp5 reveals a multidomain topology (Fig. 1). The amino-terminal half of Rsp5 consists of four domains: a C2 domain, which may bind membrane phospholipids and inositol phosphates in a calcium-regulated manner; and three WW domains (WW1, WW2, and WW3), which are protein interaction modules that bind proline-rich ligands. The carboxyl-terminal half of Rsp5 contains a HECT (homologous to E6-AP carboxyl terminus) domain (13). The HECT domain encodes an E3 ubiquitin-protein ligase activity, which ligates ubiquitin in the process of targeted protein degradation (14). Presumably, the C2 and WW domains act as membrane localization and substrate recognition modules for Rsp5, respectively (3,15,16).
The WW domain is a small protein interaction module composed of 40 amino acids that fold into a three-stranded, antiparallel ␤-sheet, which constitutes a hydrophobic pocket for binding proline-rich ligands (17)(18)(19). The WW domain is named after two conserved tryptophan residues spaced 20 -22 amino acids apart (20). Based on their binding specificity, the WW domains can be divided into two major and three minor groups (21). Group I WW domains, exemplified by the YAP and Dystrophin WW domains, bind proteins with PPXY motifs (18,22). Although the first proline of the PPXY motif greatly enhances binding to Group I WW domains, our laboratory (24) and others (23,25) have shown that in a few cases, this position can be substituted by specific amino acids such as leucine and serine. The description for Group I WW domain binding motifs indifferently uses the phrases PXY core motifs and PPXY motifs. Group II WW domains, exemplified by WW domains of FE65-and Formin-binding proteins, bind to PPLP motifs (26,27). Among the remaining three groups of WW domains, Groups III and V bind proline-rich sequences (21,28,56). However, Group IV WW domains were shown to interact with short sequences containing phosphorylated serine or phosphorylated threonine followed by proline. Interestingly, the binding of Group IV WW domains to their ligands was shown to be phosphorylation-dependent (29,30).
The RNA pol II large subunit is the major component of the transcriptional machinery responsible for regulated expression of Ͼ90% of genes (31). The yeast homolog of RNA pol II contains, in its CTD, a heptamer sequence (YSPTSPS) repeated perfectly 22 times and imperfectly four times (Fig. 1). An examination of the CTD sequence revealed a PXY core motif within the tandemly repeated heptamer peptide YSPTSP-SYSPTSPS (32). The CTD has many potential sites for phosphorylation by CTD kinases, most notably serines 2 and 5, and dephosphorylation by CTD phosphatases (31). In addition to these interactions, a well known function of the CTD is the ability to recruit many transcriptional components (33).
Our research is focused on the structural and functional characterization of the WW domain as a protein interaction and signaling module (34). Following the complete sequencing of the S. cerevisiae genome, we hypothesized that we could predict protein ligands binding to yeast WW domains based on determinations of their consensus binding motifs. Finally, we decided to further investigate a potential interaction between Rsp5 WW domains and the RNA pol II CTD because the PXY motif was found to be a minimal basic core that could interact with Group I WW domains (23)(24)(25).
Our strategy for identifying cognate ligands was to utilize phage display combinatorial peptide libraries and to determine the optimal ligand preference for each yeast WW domain, including the three Rsp5 WW domains. After obtaining the consensus sequences from binding phages, the yeast data base was searched for proteins containing the motifs in an effort to identify potential ligands for subsequent biochemical and genetic validation of WW domain-ligand complexes. Data base searches and pull-downs with the Rsp5 WW domains revealed a potential interaction with RNA pol II, and this interaction was extensively characterized. We report that Rsp5 WW2 (and possibly WW3) mediates binding to the CTD of RNA pol II. We were able to uncover a minimal CTD peptide that is sufficient to bind to Rsp5 WW2 in vitro and in the yeast two-hybrid assay. In addition, we have shown that phosphorylation of the minimal CTD sequence on serine, threonine, and tyrosine residues may act as a negative regulator of the Rsp5 WW2-CTD interaction. In view of the recent data pertaining to phosphorylationdriven interactions between the RNA pol II CTD and the Ess1/ Pin1 rotamase (30), we suggest that CTD dephosphorylation may be a prerequisite for targeted RNA pol II degradation. , and HECT domain are indicated with amino acid numbers representing the N and C termini of the protein and respective domains. The RNA pol II large subunit is illustrated with amino acid numbers representing the N and C termini of the protein as well as the location of the CTD with 26 heptamer repeats. Please note that the proteins are not drawn to scale.
The primers were designed with additional restriction enzyme sites, BamHI or SmaI and EcoRI, for ligation into the pGEX-2TK vector. Amplified sequences of the domains were ligated in frame to the pGEX-2TK vector. Since the WW domains were amplified by polymerase chain reaction, the clones were verified by DNA sequencing and expression of the GST fusion protein. To identify RNA pol II among the proteins isolated as described under "Pull-down Experiments with Yeast Lysates," monoclonal antibodies H5 and H14 were used (gifts of David Bregman) (36). Antibody H5 recognizes a CTD heptapeptide with phosphorylated serine 2, and antibody H14 recognizes a CTD heptapeptide with phosphorylated serine 5 (37).
We modified two plasmids to perform an in vivo test of protein interactions (see "Yeast Two-hybrid Assay"). Briefly, pGAD424 (CLON-TECH) contains the Gal4 activation domain and a site for generating fusion proteins consisting of the activation domain and a translated gene of interest. The plasmid pGBT9 (CLONTECH) contains the Gal4 DNA-binding domain and also contains sites suitable for cloning. We modified the multiple cloning sites of pGAD424 and pGBT9 so that we were able to subclone the DNA fragment containing the second Rsp5 WW domain into pGAD424 and the DNA containing the minimal CTD peptide into pGBT9. The resulting clones were named pGAD424-Rsp5-WW2 and pGBT-OneCTD. For pGBT9-OneCTD, we annealed two complementary oligonucleotides synthesized by Life Technologies, Inc. The DNA fragments contained two unique restriction enzymes at the ends of the complementary sequences to ensure proper insertion into pGBT9. After annealing and ligating the DNA, clones were selected, sequenced for proper insertion of sequences, and named pGBT9-OneCTD.
Preparation of GST Fusion Proteins-Expression of the GST fusion proteins was induced after adding 1 mM isopropyl-1-thio-␤-D-galactopyranoside to bacterial cells at logarithmic growth and growing the cells for an additional 2 h at 30°C. The culture was centrifuged, and the cell pellet was redissolved in PBS and 1% Triton X-100, sonicated, and centrifuged. The clarified lysate was aliquoted into 1-ml Eppendorf tubes and stored at Ϫ80°C prior to use.
Examination of the quality of the expressed fusion protein was performed using the following procedure. The lysates were incubated in the presence of glutathione beads for 30 min, washed with PBS and 1% Triton X-100, washed with PBS, and redissolved in loading buffer (125 mM Tris-HCl (pH 6.8), 2% SDS, 50% glycerol, 55 mg of ␤-mercaptoethanol, and 0.001% Coomassie Brilliant Blue). The beads were boiled for 2 min, and proteins were resolved on a 10.5% gel by SDS-polyacryl-

TABLE II
BLAST data base searches for yeast proteins containing the optimal peptide sequences All searches were performed at the NCBI web site using the BLAST programs. For the general data base searches, PXY was searched using BLAST searches with the pattern -SPSYS, word size set to 2, and expected number set to 10 million. The other general data base searches, PPPPP, PPPPY, PPPPXY, and PPXYE, were run using Advanced BLAST, with the same settings for word size and expected number. For each general data base search, the total number of matches including open reading frames, an interesting match, and the function of the interesting match are listed in the columns, respectively. For the specific data base searches, each phage display consensus sequence for Rsp5 WW domains was used to search the yeast data base. The resultant searches were analyzed, and the unknown proteins, those proteins from predicted open reading frames, were ignored. For each section (Part A, (D/P)PP(S/P)YE; Part B, PPPPYS; Part C, (A/P)PPPYE; and Part D, LPXY), we included six interesting matches, the actual sequence, and protein function. HLMQPPYSIL Amino acid metabolism amide gel electrophoresis. After running the gel for ϳ1 h at 100 V or until the loading buffer ran to the bottom of the gel, the gel was stained with Coomassie Brilliant Blue Dye for 30 min, and the unbound dye was removed.
Radioactive Labeling of GST Fusion Proteins-The vectors containing the Rsp5 WW domains (pGEX-2TK) have a thrombin-cutting site and protein kinase A phosphorylation site between GST and the WW domain. The fusion proteins were incubated with glutathione-linked agarose beads and washed with PBS and 1% Triton X-100, PBS, and heart muscle kinase buffer (20 mM Tris (pH 7.5), 100 mM NaCl, and 12 mM MgCl 2 ). The beads were incubated for 30 min with 60 l of heart muscle kinase buffer plus 1 mM dithiothreitol, 150 units of protein kinase A (Sigma), and 30 Ci of [␥-32 P]ATP (NEN Life Science Products). The beads were washed in a flow-through column with PBS and 1% Triton X-100 and with PBS before elution with 10 mM free glutathione in 50 mM Tris-HCl (pH 8). The radioactive proteins were stored on ice until use as described under "Far Western Blotting" or "Pulldown Experiments with Yeast Lysates." Phage Display Analysis-Using each WW domain of Rsp5, we searched for preferred binding motifs using various combinatorial peptide libraries. Details concerning the construction of the libraries were discussed previously (38). Briefly, we used two libraries; one library displays 14-mer X 6 PPX 6 peptides, and the other displays 16-mer X 6 PXXPX 6 peptides, where X denotes any amino acid residue and P represents the position of the fixed proline residue. Each library consists of 1.5 ϫ 10 9 unique clones. After three rounds of "panning" each Rsp5 WW domain immobilized in wells of microtiter plates, phages were selected for their ability to bind the WW domains and not the GST fusion protein itself. The inserts of selected phages were sequenced and translated for peptide sequences. The consensus sequences identified for each Rsp5 WW domain were used to search the yeast data base for proteins containing the binding motifs. The general and specific data base searches, listed in Table II, utilized the Pam30 advanced BLAST algorithm, filters off, expected number set to 10 million, and word size set to 2. The PXY search employed the BLAST program using the SPSYS pattern (39).
Using 800 g of prepared yeast extracts, beads containing Rsp5 WW1, WW2, or WW3 were incubated in binding buffer (50 mM Tris-HCl (pH 7.4), 100 mM NaCl, 1 mM EDTA, 0.1% Tween, and 1% bovine serum albumin) for a minimum of 12 h at 4°C. Each preparation was washed twice with binding buffer without bovine serum albumin, and the binding proteins were released in loading buffer containing ␤-mercaptoethanol and dithiothreitol.
Next, the samples were assayed for proteins binding to respective WW domains. The samples were run on 10.5% SDS-polyacrylamide gel and blotted onto nitrocellulose membranes. The membranes were blocked in Western Wash solution (10 mM Tris-HCl (pH 7.4), 0.9% NaCl, and 0.1% Triton X-100) containing 5% milk and probed for 6 h at 4°C with their respective radiolabeled WW domains used in the binding experiments. Afterward, the blots were washed at least three times with Western Wash solution containing 5% milk and exposed to film.
Far Western Blotting-The GST fusion proteins were resolved on 10.5% SDS-polyacrylamide gel. After transferring the gel to nitrocellulose membranes, the blot was blocked for at least 4 h at 4°C with Western Wash solution containing 5% milk. The blots were probed with radioactively labeled GST-Rsp5 WW domains or GST-yeast CTD and washed as described above.
SPOTs Membrane Analysis-Small peptides were synthesized on derivatized cellulose membranes (Genosys Biotechnologies, Inc.) as described previously (40,41). Each membrane was blocked, incubated with radiolabeled probe, and washed using 1ϫ blocking buffer (supplied by Genosys Biotechnologies, Inc.) in Western Wash solution. The membranes were blocked for 6 h at 4°C, probed for 12 h at 4°C, and washed three times for 15 min each at room temperature (24,38).
Yeast Two-hybrid Assay-The two yeast plasmids (pGAD424-Rsp5-WW2 and pGBT9-OneCTD) were transformed individually into HF7c cells (42). For pGAD424-Rsp5-WW2, clones were selected on agar plates minus leucine; and for pGBT9-OneCTD, clones were selected on agar plates minus tryptophan. Simultaneously, the clone containing pGBT9-OneCTD was grown and additionally transformed with pGAD424-Rsp5-WW2. Clones containing both pGAD424-Rsp5-WW2 and pGBT9-OneCTD plasmids were selected on agar plates minus leucine and tryptophan. A selected clone containing both plasmids was grown on plates lacking leucine, tryptophan, and histidine and observed for its ability to grow.

RESULTS
Binding Specificities of the Rsp5 WW Domains Classify Them as Group I WW Domains-We decided to identify potential ligands for each WW domain of Rsp5 and used the following approach. After screening phage-displayed combinatorial peptide libraries, we determined the optimal peptide sequence for ligands to each WW domain and identified potential interacting proteins containing these sequences in the yeast proteome.
The results of "Phage Display Analysis" for each Rsp5 WW domain are presented in Table I. The consensus sequence for each Rsp5 WW domain is as follows: WW1, (D/P)PP(S/P)YE; WW2, PPPPYS; and WW3, (A/P)PPPYE. The preferred binding motif for each WW domain is a PPXY-like motif. Therefore, the three Rsp5 WW domains belong to Group I. Interestingly, two clones, R1-PXXP and 62-PXXP, one from WW2 and one from WW3 respectively, did not conform to the consensus sequence obtained from the other clones. These two clones possess the LPXY sequence, drawing a comparison with the PXY-type motif suggested previously (23)(24)(25).
A search of the yeast data base using the consensus sequences identified the potential ligands listed in Table II (Part  2). The proteins are involved in diverse functions such as transcription, ubiquitination, and amino acid metabolism. The searches also yielded numerous open reading frames. Since the open reading frames represent hypothetical proteins of unknown functions, they were not listed.
We compared the specific data base searches in Part 2 of Table II with a data base search of the yeast proteome using BLAST for PXY and advanced BLAST for PPPPP, PPPPY, and PPXYE, listed in Part 1. The potential ligands identified in Part 2 were recognized in data base searches listed in Part 1. The phage-displayed peptide library screening produced results that are more restrictive than random data base searches.
The Second Rsp5 WW Domain Can Pull-down RNA pol II from Yeast Extracts-To rapidly screen for potential ligand(s) from those listed in Table II (Part 2), we attempted to pulldown binding proteins from total yeast cell lysates. Using GST fusion proteins of Rsp5 WW domains, we identified a protein of ϳ200 kDa in pull-downs with RSP5 WW2 and WW3 (Fig. 2, C  and D). In addition, two faint protein bands were observed migrating around 60 kDa in pull-downs with Rsp5 WW2 (Fig.  2C). Based on the molecular mass of the pull-down proteins and previously published data describing a genetic interaction between Rsp5 and RNA pol II (4), we hypothesized that the 200-kDa protein band was RNA pol II. We stripped the blots and reprobed them with monoclonal antibodies specific for the large subunit of RNA pol II and concluded that the protein, which migrated at 200 kDa, was most likely RNA pol II (Fig. 3,  C and D).
All Three Rsp5 WW domains Bind to the CTD of RNA pol II-To investigate the binding specificity driving the protein interaction between WW domains of Rsp5 and the yeast RNA pol II CTD, we used Far Western blot analysis. Bacterial lysates containing GST fusion proteins of yeast RNA pol II CTD, PPXY motifs found in p53BP2 and WBP-1, and a PPLP motif found in Mena were purified, loaded onto SDS gels, transferred to nitrocellulose membranes, and probed with each radioactively labeled Rsp5 WW domain. p53BP2 and WBP-1 contain the PPXY motif that was verified to bind YAP WW domains (18,38). Mena contains a PPLP motif and was shown to bind to FE65 (26). The clone labeled CTD contains 26 heptamer repeats of the yeast CTD. We demonstrated that each Rsp5 WW domain bound to p53BP2 and WBP-1 PPXY motifs and did not bind to the Mena PPLP motif. However, each Rsp5 WW domain bound to the yeast CTD (Fig. 4, A-C). Conversely, when we probed a blot containing equal amounts of Rsp5 WW domains and human YAP with the labeled yeast CTD, we confirmed that each WW domain had the capacity to bind the CTD and found that Rsp5 WW2 bound to the yeast CTD most efficiently (Fig. 4E).
Determination of the Minimal CTD Peptide Required to Interact with the Second Rsp5 WW Domain-Since Rsp5 WW2 was able to pull-down RNA pol II from yeast cell lysates and to interact directly with the CTD, we focused on the identification of the specific binding sites in the CTD. The amino acid residues of the CTD are enumerated as follows: 1 YSPTSPS 7 . In an attempt to identify the minimal CTD-derived peptide that interacts with Rsp5 WW2, we synthesized 14-mer CTD peptides

FIG. 4. Far Western blotting of GST-Rsp5 WW domains reveals an interaction with the yeast RNA polymerase II CTD. A-C, Far
Western blots are shown. Lysates of GST alone and GST fusion proteins of the PXY motifs from yeast CTD, p53BP2, and WBP-1 and of the PPLP motif from Mena were prepared; resolved by SDS-polyacrylamide gel electrophoresis; and transferred to nitrocellulose membranes. The membranes were probed with radioactively labeled GST-Rsp5-WW1 (A), GST-Rsp5-WW2 (B), or GST-Rsp5-WW3 (C). D, the loading of lysates on gels was normalized by Coomassie Blue staining. E, lysates of each GST-Rsp5 WW domain and human YAP were prepared, resolved by SDS-polyacrylamide gel electrophoresis, and transferred to a nitrocellulose membrane. Yeast GST-CTD was radioactively labeled and used to probe the membrane. F, the loading of lysates used in E was normalized by Coomassie Blue staining.
on SPOTs membranes and reduced the length of a repeat from the amino and/or carboxyl terminus by sequentially replacing the flanking sequences with alanine residues.
Based on our analysis, we have chosen -YSPTSPSYSPT (Fig.  5, spot 16) as the "minimal" CTD peptide that binds Rsp5 WW2. Interestingly, a shorter CTD peptide, AAAAAAAPSYSPTSP (Fig. 5, spot 8), was also able to interact with Rsp5 WW2. Parallel SPOTs membranes were prepared with the same repertoires of peptides and probed with radioactive GST-Rsp5-WW1 and GST-Rsp5-WW3 fusion proteins, but no appreciable binding was observed.
Alanine Scanning of the 12-mer CTD Peptide-To identify the amino acid residues involved in the interaction between the minimal CTD peptide and Rsp5 WW2, we scanned the minimal CTD binding motif with alanine residues. Since the peptides are synthesized from the carboxyl terminus, one additional serine residue was added to the 11-mer peptide to provide an anchor and linker for each peptide to the membrane. Replacing tyrosine 1, proline 6, tyrosine 8, and serine 9 with alanine greatly diminished or abolished binding (Fig. 6, spots 2, 7, 9, and 10). Substitutions that had little effect were the following residues: serine 2, proline 3, proline 10, and threonine 11 (Fig.  6, spots 3, 4, 11, and 12). A significant increase in binding was observed when threonine 4 or serine 5 was replaced with alanine (Fig. 6, spots 5 and 6). The strongest positive effect on binding was observed when serine 7 was replaced with alanine ( Fig. 6, spot 8). The data suggested that the crucial amino acid residues required for an interaction between the minimal CTD binding motif and Rsp5 WW2 are 1 YXXXXPXYS 9 XX. These data confirmed the minimal PXY core shown in phage display screens and revealed the importance of flanking amino acid residues.
Effect of CTD Peptide Phosphorylation on Binding of the Second Rsp5 WW Domain-Next, we investigated whether CTD modification by phosphorylation alters binding to Rsp5 WW2. We synthesized 14-mer peptides of the CTD repeat on SPOTs membranes using phosphorylated serine, threonine, or tyrosine. The Rsp5 WW domains and GST alone were radiolabeled for probing the peptides. In this experiment, we confirmed that only Rsp5 WW2 bound the minimal CTD repeat.
We repeated the experiments with CTD peptides phospho- FIG. 5. Determination of the minimal CTD peptide required for binding to Rsp5 WW2. A, a SPOTs membrane was probed with radioactively labeled GST-Rsp5-WW2. B and C, peptides spanning 15 amino acid residues were synthesized on SPOTs membranes as illustrated. Alanine residues were used to substitute for each CTD amino acid residue to identify the shortest CTD peptide that binds Rsp5 WW2. The substitutions advanced from the amino terminus (spots 2-14), the carboxyl terminus (spots [21][22][23][24][25][26][27][28][29][30][31][32], and gradually from the amino and carboxyl termini (spots [15][16][17][18][19][20]. Spots 33-39 are imperfect CTD repeats. D, the probed SPOTs membrane was analyzed using a PhosphorImager to compare the relative intensity of signals. rylated at multiple sites including one phosphorylation in residues flanking the repeat and the other phosphorylation site in the complete repeat. We also varied patterns of multiple CTD phosphorylation. Although not every permutation of the phosphorylated CTD was examined, these experiments have shown that phosphorylation can negatively regulate binding between the CTD and Rsp5 WW2 (Fig. 7, columns II and III). However, although phosphorylation abolishes binding overall, Rsp5 WW2 may have a minor capacity for binding the CTD phosphorylated on serine 7 (Fig. 7, row Rsp5 WW2, compare column I, spots 2 and 6; column II, spot 2; and column III, spots 6, 7, and 9).
Influence of Phosphorylated "Imperfect" CTD Repeats on Binding of Rsp5 WW2-Since the CTD contains imperfect repeats as well, we decided to phosphorylate serine, threonine, and tyrosine residues on imperfect repeats to test the ability of Rsp5 WW2 to bind these peptides. As described under "Alanine Scanning of the 12-mer CTD Peptide," we observed a much greater binding intensity, using Rsp5 WW2, for a CTD peptide with an alanine substitution at serine 7 than for a minimal CTD peptide (Fig. 6, spot 8). We hypothesize that some imperfect repeats may have greater potential for binding Rsp5 WW2 (see "Discussion").
We synthesized naturally occurring imperfect CTD peptides, in unphosphorylated and phosphorylated forms, on SPOTs membranes and probed them with Rsp5 WW2. We observed that phosphorylation of imperfect repeats had a negative effect on binding to Rsp5 WW2 compared with the unphosphorylated imperfect repeats (Fig. 8).
For the unphosphorylated imperfect repeats, we observed that not every amino acid can increase binding intensity at position 7 of the minimal CTD peptide, which also corresponds to position X of the PXY core binding motif. For example, in two naturally occurring imperfect repeats in which position 7 harbors an asparagine or glycine residue, there was no observable binding of Rsp5 WW2 for the phosphorylated or unphosphorylated repeat (Fig. 8, spots 26 -34 and 41-48). Interestingly, a substitution of serine 7 with alanine increased binding 2-3-fold compared with the perfect repeat (compare spots 1 with spots 8 and 17 in Figs. 6 and 8, respectively). Another 2-fold increase was observed when threonines 4 and 11 were replaced with glycine and lysine, respectively (Fig. 8, spot 35). Finally, the interaction of Rsp5 WW2 with a phosphorylated imperfect CTD repeat at serine 2 or threonine 4 was just as good or greater compared with the binding of Rsp5 WW2 to an unphosphorylated perfect repeat (Fig. 8, compare spot 1 with spots 19 , 20, and 37). In summary, selected unphosphorylated imperfect CTD repeats show significantly stronger interactions with Rsp5 WW2 compared with the unphosphorylated perfect repeats.
The Minimal RNA pol II CTD Peptide Interacts with Rsp5 WW2 in a Yeast Two-hybrid Analysis-Rsp5 WW2 and the minimal RNA pol II CTD peptide were cloned into plasmids as fusions with the Gal4 activation domain (AD) and the DNAbinding domain (BD), respectively, to be assayed for protein interaction in vivo using the yeast two-hybrid assay. Clones containing Gal4-AD-Rsp5-WW2 (pGAD-Rsp5-WW2) or Gal4-BD-RNA pol II (pGBT-OneCTD) were plated on agar plates prepared without leucine, tryptophan, and histidine and were unable to grow (Fig. 9). However, a clone containing both plasmids was able to grow (Fig. 9). Thus, we showed that the in FIG. 6. Alanine scanning of a 12-mer minimal CTD peptide. A, radioactive GST-Rsp5-WW2 was used to probe the SPOTs membrane. B and C, peptides 12 amino acids long of the CTD repeat were synthesized on SPOTs membranes as illustrated. An alanine residue replaced each position of a perfect repeat. D, the probed SPOTs membrane was analyzed using a PhosphorImager to compare the relative intensity of signals.

FIG. 7. Analysis of phosphorylated CTD peptides on SPOTs membranes.
As listed to the right, GST alone and each GST-Rsp5 WW domain were radioactively labeled and incubated with three strips of spots (columns I-III) containing unphosphorylated and phosphorylated CTD peptides. Grid indicates the order of synthesized peptides, followed by the peptide sequence of each spot. J, phosphoserine; O, phosphothreonine; and Z, phosphotyrosine (in this case). vitro binding we observed for Rsp5 WW2 and the minimal CTD peptide can also occur in yeast cells. DISCUSSION We have characterized the binding between Rsp5 WW domains and the CTD of RNA pol II. The second WW domain of Rsp5 was able to pull-down the RNA pol II from yeast cell lysates and bind to CTD-derived peptides. Furthermore, we have determined the minimal length of the CTD peptides that can mediate the Rsp5 WW2-CTD interaction in vitro and in yeast cells. Through the course of our examination of the minimal CTD peptide, the selected imperfect CTD repeats were uncovered as potentially preferential sites for binding Rsp5 WW domains. In addition, we showed that the phosphorylation of the CTD residues has an overall negative effect on binding to Rsp5 WW2.
RNA pol II and Bul1 proteins have been shown to interact with Rsp5 functionally, biochemically, and genetically (3,4,32,43,44). Using the SPOTs membrane assay, we showed that peptides containing the PPXY motif, derived from Bul1, interacted 30 -40 times better than similar length CTD peptides containing a PXY motif. However, when Rsp5 WW2 was used in pull-down assays from yeast cell lysates, RNA pol II was the major protein band that was visualized. We did not observe a FIG. 9. Yeast two-hybrid analysis using the RNA polymerase II minimal CTD peptide and the second WW domain of Rsp5. One minimal CTD repeat motif was cloned into a plasmid containing the Gal4 DNA-binding domain (pGBT-OneCTD). The second Rsp5 WW domain was cloned into the plasmid containing the Gal4 activation domain (pGAD-Rsp5-WW2); for more details, see "Experimental Procedures". In the three panels, yeast previously transformed with the indicated plasmids was spread onto agar plates lacking leucine, tryptophan, and histidine.
FIG. 8. Phosphorylation scanning of perfect and imperfect CTD repeats on SPOTs membranes. A, GST-Rsp5-WW2 was radioactively labeled and incubated with the SPOTs membrane. B and C, perfect and imperfect CTD repeats were synthesized on SPOTs membranes as illustrated. These imperfect repeat sequences are spots 11, 17, 26, 35, and 41. Each position containing serine, threonine, or tyrosine was modified with a phosphorylated amino acid residue: phosphoserine (J), phosphothreonine (O), or phosphotyrosine (Z). D, the probed SPOTs membrane was analyzed using a PhosphorImager to compare the relative intensity of signals.
protein migrating between 110 and 140 kDa, the expected molecular mass of Bul1. Assuming that levels of RNA pol II and Bul1 are similar in cells, there could be many explanations for these differences. One favorable explanation is that the numerous (total of 26) CTD PXY-containing repeats may create a scaffolding that assembles Rsp5 WW domains into a large and stable complex with the RNA pol II CTD. The relatively high "local concentration" of the PXY motif combined with the unique structure of the CTD tail could account for such strong complex and/or efficient pull-down results.
The deletion analysis of CTD using alanine substitutions resulted in an interesting observation. After identifying a minimal CTD peptide that binds Rsp5 WW2 (AYSPTSPSYSPTA-AA), the continued sequential substitution of the CTD with alanine revealed a single peptide (AAAAAAPSYSPTSPA) that bound to Rsp5 WW2 as well as the minimal CTD peptide (AYSPTSPSYSPTAAA). All the intermediate and subsequent peptides in the sequential substitution were negative for binding. This result is difficult to explain. One possibility is that the stretch of alanine residues preceding the PXY core in this peptide (AAAAAAPSYSPTSPA) folds into a conformation similar to the minimal CTD peptide and presents the PXY core for binding. Another possible explanation is that CTD repeats can achieve unique conformations only when a certain length of amino acids and specific amino acid residues of the repeat are present (45). More important, we have chosen an 11-mer CTD sequence as the minimal length peptide for binding Rsp5 WW2 in our study. By NMR analysis, we are investigating the structure of Rsp5 WW2 in complexes using the CTD peptides, as well as other PPXY-containing peptides, to understand these results at the molecular level.
Our complex of Rsp5 WW2 with the 11-mer CTD peptide was confirmed by two methods. The yeast two-hybrid assay confirmed the ability of Rsp5 WW2 to form a complex with the 11-mer peptide of CTD in cells. In addition, we generated an NMR structure of Rsp5 WW2 in complex with the unphosphorylated 11-mer CTD peptide. This structure recapitulates the details previously solved in the complex of the YAP WW domain with the WBP-1 target peptide containing PPXY (19). 2 A complicated picture has emerged where the status of RNA pol II CTD phosphorylation correlates with transcriptional activity (46). In general, either hypophosphorylated RNA pol II molecules are in the nuclear milieu awaiting targeting by association with general transcription factors, or RNA pol II rests in quiescence at promoter sites (47). The RNA pol II with a hyperphosphorylated CTD clears the promoter and allows transcription to proceed (31,48). CTD phosphorylation and dephosphorylation can dynamically change its conformation and drastically alter the proteins that interact with the CTD and therefore the RNA pol II holoenzyme (33). During the preinitiation stage of transcription, the dephosphorylated CTD interacts with proteins involved in transcriptional activation; upon CTD phosphorylation, the proteins that interact with the CTD are involved in co-transcriptional pre-mRNA processing (49,50).
One of the implications of our data is that the unphosphorylated or partially dephosphorylated CTD may signal an interaction with Rsp5 ubiquitin ligases to target RNA pol II for degradation in the proteasome. Based on our results, we consider two possibilities. CTD phosphatases may mediate a complete dephosphorylation of the CTD, allowing Rsp5 WW domains to form a complex with the CTD at many potential sites; or CTD phosphatases may dephosphorylate specific CTD repeats, possibly selected imperfect repeats, and restrict the abil-ity of Rsp5 WW domains to form a complex with the CTD. The latter possibility seems more likely and is supported by several reports showing that a form of RNA pol II with the phosphorylated CTD is preferentially targeted by ubiquitin ligases (51)(52)(53).
In contrast, several reports have shown that Ess1, which contains an isoprolyl isomerase and a Group IV WW domain, is able to interact with the phosphorylated RNA pol II CTD and to regulate activity by controlling several processes, which include the formation of the 3Ј-end mRNA and termination of RNA pol II-mediated transcription (30,54,55). 3,4 Since Ess1 contacts the phosphorylated CTD through its WW domain, an interesting regulatory scenario emerges. The CTD of RNA pol II has an ability to form complexes with two WW domaincontaining proteins, Rsp5 and Ess1, which harbor dissimilar enzymatic activities and uniquely affect RNA pol II in terms of transcriptional control mechanisms. A functional test of these regulatory complexes in yeast cells expressing Rsp5 and/or Ess1 with specific point mutations that affect their domain(s) could illuminate novel signaling steps controlling the stability of RNA pol II or maintenance of transcription.