Phospho-Carboxyl-Terminal Domain Binding and the Role of a Prolyl Isomerase in Pre-mRNA 3′-End Formation*

A phospho-carboxyl-terminal domain (CTD) affinity column created with yeast CTD kinase I and the CTD of RNA polymerase II was used to identify Ess1/Pin1 as a phospho-CTD-binding protein. Ess1/Pin1 is a peptidyl prolyl isomerase involved in both mitotic regulation and pre-mRNA 3′-end formation. Like native Ess1, a GSTEss1 fusion protein associates specifically with the phosphorylated but not with the unphosphorylated CTD. Further, hyperphosphorylated RNA polymerase II appears to be the dominant Ess1 binding protein in total yeast extracts. We demonstrate that phospho-CTD binding is mediated by the small WW domain of Ess1 rather than the isomerase domain. These findings suggest a mechanism in which the WW domain binds the phosphorylated CTD of elongating RNA polymerase II and the isomerase domain reconfigures the CTD though isomerization of proline residues perhaps by a processive mechanism. This process may be linked to a variety of pre-mRNA maturation events that use the phosphorylated CTD, including the coupled processes of pre-mRNA 3′-end formation and transcription termination.

The carboxyl-terminal domain (CTD) 1 of the largest subunit of RNA polymerase II plays a central role in controlling mRNA production in eukaryotes. In most eukaryotes the CTD consists of between 26 and 52 repeats with the consensus sequence, YSPTSPS (1). This unusual domain is not found in eukaryotic RNA Pol I or Pol III or in prokaryotic RNA polymerase. By mechanisms that are still largely not understood, RNA Pol II behavior is controlled by CTD kinases and phosphatases that modulate the phosphorylation state of the CTD (2). A body of evidence has accumulated indicating that the unphosphorylated CTD is present in RNA Pol II holoenzyme and is required during the initial stages of transcription (3)(4)(5)(6) but that a hyperphosphorylated CTD is present in elongating polymerase (7)(8)(9)(10).
The hypothesis that the phospho-CTD functions in pre-mRNA processing (11) has received firm experimental backing with the recognition that several components of the 5Ј mRNA capping enzyme complex are phospho-CTD-binding proteins (12)(13)(14). In addition, indirect evidence suggests that the phospho-CTD also binds splicing components (15)(16)(17)(18). Further, both the phosphorylated and the unphosphorylated CTD can interact with pre-mRNA 3Ј-end processing proteins (19,20). The combined evidence suggests that pre-mRNA processing may in general be cotranscriptional, with the (phospho)-CTD serving as a master coordinator of the events involved.
Recently, mutations in the peptidyl prolyl isomerase, ESS1/ PTF1, were found to be responsible for the temperature-sensitive phenotypes of two yeast mutants that displayed a pre-mRNA 3Ј-end formation defect (21). Yeast ESS1 was originally identified as essential gene (22), and is in fact the only essential peptidyl prolyl isomerase (PPIase) in Saccharomyces cerevisiae (23). The human protein, Pin1, and the Drosophila protein, Dodo, are close homologues of Ess1 and both can functionally replace Ess1 in yeast (24,25). All three proteins have a small 31-residue WW domain amino-terminal to the isomerase domain. Studies of the peptidyl prolyl isomerase activity of recombinant Ess1 (21) and Pin1 (26) indicate these enzymes display more than a 1000-fold higher specificity toward substrates in which the other amino acid at the isomerizable bond is negatively charged (Glu, phospho-Ser, and phospho-Thr). In higher eukaryotes a variety of techniques have been used to show that Pin1 stably associates with phosphorylated but not unphosphorylated mitotic regulators that are targets of Cdc2/ cyclinB (26 -29). The present study demonstrates that the hyperphosphorylated CTD of RNA Pol II contains high affinity binding sites for the WW domain of Ess1.

MATERIALS AND METHODS
Identification of Ess1 as a Phospho-CTD-binding Protein-To make the phospho-CTD affinity column, CTD kinase I was purified using polyclonal antibodies against a peptide from CTK1 as described (30). Phosphorylation of a ␤-galactosidase/yeast CTD fusion protein, ␤Ga-lyCTD (FPY1 in Ref. 30), was done as described (31) with sufficient CTD kinase I to cause a complete shift of ␤GalyCTD to the hyperphosphorylated form as observed on a stained gel. A duplicate blank reaction was done without kinase, and both reactions were desalted into PBS using Quick Spin Protein Columns (Roche Molecular Biochemicals). The AminoLink Plus Immobilization kit (Pierce) was used to make 250 l each of phospho-CTD and CTD matrix (pH 10 coupling protocol). The yeast strain, DY6, (a leu2 trp1 ura3-52 prb1-1122 pep4 -3 prc1-407) was grown at 30°C in minimal medium to an A 600 of 1 to 2. Cells were collected and disrupted by glass beads using a protocol (39) modified for a mini bead beater and substituting a protease inhibitor mix from Roche Molecular Biochemicals (catalog 1206 8934). The only other modification was the direct recovery of the extract from the disruption tube by centrifugation (15,000 RCF for 1 min) followed by pipette removal of as much supernatant from the beads as was possible. Prior to affinity chromatography, extracts were thawed and then clarified by centrifugation at 12,000 RCF for 30 s. The supernatant was diluted with 2 volumes of buffer (30 mM Tris, pH 7.8, 3 mM PMSF) and reclarified by centrifugation. Equal amounts of on-put (2.5 ml) were loaded onto paired 250-l columns containing identical amounts of phosphorylated or unphosphorylated ␤GalyCTD. Each column was washed five times with 400-l column equilibration buffer (10 mM Tris, pH 7.8, 100 mM NaCl, and 1 mM PMSF) and eluted with sequential steps consisting of 1 column volume of 110, 125, 140, 160, 190, 220, 250, 300, 350, 425, 500, 600, 700, 850, 1000, 1000, and 1000 mM NaCl in 10 mM Tris, pH 7.8, with 1 mM PMSF. Initial gel analysis using silver staining revealed a 21-kDa protein eluting specifically from the phospho-CTD column at 1 * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
‡ To whom correspondence should be addressed. E-mail: arno@ biochem.duke.edu. 1 The abbreviations used are: CTD, carboxyl-terminal domain; Pol I, II, III, polymerase I, II, and III, respectively; PMSF, phenylmethylsulfonyl fluoride; MALDI-MS, matrix-assisted laser desorption/ionizationmass spectrometry; PBS, phosphate-buffered saline; DTT, dithiothreitol; GST, glutathione S-transferase; RCF, relative centrifugal force. M NaCl. The corresponding high salt fractions from both columns were brought to 0.1% SDS, concentrated, and desalted with Microcon 10 units (Amicon) and separated on a 4 -15% SDS precast gel (Bio-Rad). Following a protocol from HHMI Biopolymer/Keck Foundation Biotechnology Resource Laboratory at Yale University, the gel containing the 21-kDa band was briefly stained with Coomassie Blue, after which the 21-kDa band was excised and destained for 3 h in 10% acetic acid and 50% methanol. The band was frozen and sent to the biotechnology facility at Yale for in-gel trypsinization and MALDI-MS analysis.
Expression of GSTEss1 Fusion Proteins-Fusion proteins were expressed using plasmids created by ligating ESS1 polymerase chain reaction fragments into the EcoRI/BamHI sites of the polylinker of pGEX2TK (Amersham Pharmacia Biotech). The primers CGGGATCC-ATGATCATTGTGGTACTTATTCC and GGAATTCCTAACCTACCGC-TTGATCAC were used to amplify the 190-amino acid version of Ess1 (GenBank TM 1015652) from genomic yeast DNA. The construct sequence was verified by DNA sequencing. Polymerase chain reaction fragments for the fusion protein containing the WW domain (GST-Ess1WW) and isomerase domain (GSTEss1ISO) were amplified from pGEX2TK-GSTEss1 using the primers CGGGATCCATGATCATTGTG-GTACTTATTCC/GGAATTCCTAGTTGGTGCCCTCAGGCTC (WW) and CGGGATCCGACCATCCAGTGCGTGTAAGATGC/GGAATTCCT-AACCTACCGCTTGATCAC (ISO) inserted into the EcoRI/BamHI sites of pGEX2TK (Amersham Pharmacia Biotech) and sequenced. Protein expression and glutathione purification were according to the protocol of the manufacturer (Amersham Pharmacia Biotech, XY-058-00-01) with minor modification. BL21 cells containing the expression plasmids were grown in LB medium with 50 g/ml ampicillin, and the fusion proteins were purified by batch adsorption. Complete elution of GST-Ess1 or GSTEss1WW required an overnight incubation with modified elution buffer (50 mM Tris, 10 mM glutathione, and 100 mM NaCl brought to pH 8.5-9 with NaOH). The fusion proteins were desalted into PBS using Quick Spin Protein Columns (Roche Molecular Biochemicals) and quantitated using the Bio-Rad Protein Assay (Bio-Rad) with IgG as a standard.
Precipitation of the Phospho-CTD with GSTEss1 Beads-The Amin-oLink Plus Immobilization kit (Pierce) was used to make both the GSTEss1 (ϳ2 mg/ml of matrix) and control beads (neutral coupling conditions). For the binding assay, GSTyCTD was phosphorylated essentially as described (32) using 10 M GSTyCTD and approximately 150 nM CTD kinase I. The reaction was desalted into PBS using Quick Spin Protein Columns (Roche Molecular Biochemicals), and 80 l of this solution was added to 20 l of control beads or GSTEss1 beads in PBS. Mixes were incubated at room temperature with inversion until the times indicated when both samples were spun (500 RCF for 10 s) to pellet the beads and an aliquot of the supernatants removed to 2ϫ SDS-polyacrylamide gel electrophoresis sample buffer and heated. Beads were then resuspended and the incubation continued until the next time point. Samples (10 l) were run on a 4 -15% SDS precast gel (Bio-Rad).
Far-Western and Western Analysis of Yeast Extracts-An extract of frozen yeast representing 20 ml of culture (grown to A 600 ϭ 1.2) was prepared according to method 2 of Ref. 33 to minimize phosphatase and protease activity; final volume ϭ 400 l in SDS sample buffer. Aliquots of extract (1 and 5 l) were run with prestained standards in five repeated sample sets on a 4 -15% SDS precast gel (Bio-Rad) and electroblotted to nitrocellulose (NC); all standards were completely transferred out of gel. The NC was incubated in BRB overnight with rocking (this and subsequent steps at 4°C) and then cut into strips each containing one sample set. For the far-Western analysis, strips were incubated separately with 6 nM fusion protein probe in BRB without DTT for 1 h. After rinsing 5 ϫ 12 min with PBS-Tw, they were reacted together with affinity-purified rabbit anti-GST IgG. 2 After rinsing 5 ϫ 12 min with PBS-Tw, the strips were incubated with 2°antibody (affinity-purified, alkaline phosphatase-coupled donkey anti-rabbit IgG (Jackson ImmunoResearch)) in BRB without DTT for 1 h, then rinsed 5 ϫ 12 min with PBS-Tw and once with 50 mM Tris-HCl, pH 9.5. The strips were then incubated with alkaline phosphatase color reagent substrates (Bio-Rad, 170-6432) at room temperature. For the Western, one strip was reacted with affinity-purified rabbit anti-yeast phospho-CTD IgG (prepared essentially as in Ref. 33) for 1 h, then washed 5 ϫ 12 min with PBS-Tw prior to addition of the 2°antibody and treatment as for the far-Western analyses.

RESULTS
Yeast CTD kinase I can quantitatively convert fusion protein substrates containing the CTD to a hyperphosphorylated form, with the accompanying characteristic decrease in electrophoretic mobility (31). To identify proteins that bind specifically to the hyperphosphorylated CTD, ␤GalyCTD was hyperphosphorylated and cross-linked to agarose beads. Total yeast extracts were loaded in parallel onto this and an identical column prepared with unphosphorylated ␤GalyCTD and eluted using a step gradient of NaCl. While many proteins eluted from both columns, a protein of about 21 kDa appeared to elute only from the phospho-CTD column at 1 M NaCl (Fig. 1A). This protein band was excised and subjected to in-gel trypsinization followed by MALDI mass spectrometric analysis of the tryptic fragments (Fig. 1B). Data base searches using the masses of the fragments gave an unambiguous identification of the protein as Ess1.
To confirm the ability of Ess1 to bind the phosphorylated CTD, a GSTEss1 fusion protein was purified and linked to agarose beads. As shown in Fig. 2, GSTEss1 beads were able to specifically deplete hyperphosphorylated GST-yCTD fusion protein from a solution containing both phosphorylated and unphosphorylated GSTyCTD. In addition to the unphosphorylated fusion protein, some forms of phosphorylated GSTyCTD were not effectively adsorbed, including a partially shifted intermediate clearly visible in Fig. 2. Extending the incubation an additional 30 min did not result in significant depletion of the remaining unbound material (data not shown). This may suggest that not all phosphorylations create a recognition site for Ess1.
The hyperphosphorylated CTD produced by CTD kinase I can also be used as an effective radiolabeled probe to identify phospho-CTD-binding proteins in blots of cellular extracts. Preliminary far-Western analysis of proteolyzed Ess1 suggested the WW domain could bind the phospho-CTD. 3 To test this hypothesis we created expression constructs for GST fusion proteins containing only the WW (GSTEss1WW) or isomerase (GSTEss1ISO) domains. Far-Western analysis of the fusion proteins (Fig. 3) shows similar association of the hyperphosphorylated CTD probe with GSTEss1 (lanes 1-3) and GSTEss1WW (lanes 4 -6) but no association with GSTEss1ISO (lanes 7-9). Clearly, under these conditions the phosphorylated CTD appears to associate only with fusion proteins including the WW domain. As a control to demonstrate specificity, total Escherichia coli proteins were run in lane 10 of Fig. 3. There should be no natural phospho-CTD-binding proteins in E. coli, and indeed far-Western analysis reveals no phospho-CTDbinding proteins.
To demonstrate that in vivo hyperphosphorylated RNA polymerase II from yeast can associate with Ess1 and to determine whether the WW domain could mediate this interaction, we analyzed total cellular extracts of asynchronously grown yeast by far-Western using the fusion proteins as probes (Fig.  4). In these experiments a protein of Ͼ200 kDa was the dominant interacting species, bound equivalently by GSTEss1 (lanes 5 and 6) and GSTEss1WW (lanes 8 and 9). This protein is almost certainly the hyperphosphorylated form of the largest subunit of RNA Pol II (subunit IIo), as it has an identical electrophoretic mobility to IIo detected using affinity-purified anti-phospho-CTD polyclonal antibodies (lanes 11 and 12). Note that neither the control lanes probed with GST (lanes 2 and 3) nor the lanes probed with GSTEss1ISO (lanes 14 and 15) gave any detectable signal.

DISCUSSION
Our results effectively demonstrate that the phosphorylated CTD binds directly to Ess1 and that it is the WW domain that mediates this binding. Although a direct association between the phospho-CTD and the human Pin1 has not been demonstrated, Albert et al., have recently shown that beads contain-ing Pin1 can precipitate a fraction of the hyperphosphorylated RNA Pol II from HeLa cell extracts (34). Given our results with Ess1, this finding strongly suggests that Pin1 also interacts directly with the phosphorylated CTD.
Also recently, it has been shown that the WW and not the isomerase domain of Pin1 is responsible for Pin1 binding to proteins containing mitosis-specific phospho-epitopes (29). Further, the WW domain of Ess1 was also able to bind most of these mammalian proteins. At present, the connection between Ess1/Pin1 association with the phospho-CTD and with mitotic regulatory proteins remains to be determined.
It has been demonstrated that isomerase activity of both Pin1 (27) and Ess1 (21) has a strong preference for proline FIG. 2. GSTEss1-containing beads deplete the phosphorylated but not the unphosphorylated CTD from solution. GSTyCTD and phosphorylated GSTyCTD (Pn GSTyCTD) were added to 0.2 vol of control beads or GSTEss1 beads and incubated at 25°C. At the times indicated the beads were pelleted and supernatants sampled. The initial solution (10 l) and time point supernatants (10 l) were run on a 4 -15% SDS gel which was stained with Coomassie Blue and analyzed with a PhosphorImager. Note that the phosphorylated intermediate clearly visible in the stained gel between the GSTyCTD and Pn GSTyCTD does not adsorb to the GSTEss1 beads. Removal of the hyperphosphorylated species, Pn GSTyCTD, reveals the radioactivity associated with this minor band.

FIG. 1. Identification of Ess1 as a phospho-CTD-binding protein.
Total yeast extracts were loaded onto paired columns containing phosphorylated and unphosphorylated CTD. A, gel analysis of the elution of these columns with an identical step gradient of increasing NaCl revealed a protein of 21 kDa eluting at 1 M NaCl from the phosphorylated CTD column (labeled P CTD) but not the unphosphorylated CTD column (labeled CTD). This protein is indicated by the arrow to the right of the Coomassie Blue-stained 4 -15% SDS gel. The 21-kDa band in this gel was excised and subjected to in-gel trypsinization and subsequent MALDI mass spectrometry analysis. B, the portion of the MALDI-MS spectrum containing the major products is shown. The experimentally determined masses (MH ϩ ) for these fragments are indicated. When the masses of all fragments produced were matched against possible fragments using ProFound (OWL data base), the program identified Ess1. Peptides covering 57% of Ess1 were identified, including most of the major fragments. Indicated in the parentheses are the tryptic fragments of the 170-amino acid version of Ess1 (GenBank TM accession number 758286) that correspond to the masses observed. Bradykinin and an ACTH fragment (18 -39), with protonated monoisotopic masses of 1060.57 and 2465.2, respectively, were included as internal calibrants. bonds in which the proline is preceded by an acidic residue (Glu, phospho-Ser, and phospho-Thr). These functional similarities, the conservation of amino acid sequence and domain organization, and the ability of Pin1 to substitute for Ess1 in yeast all indicate a very high level of functional conservation between these two proteins.
Beyond the preference of the isomerase domain for an acidic residue prior to proline, the specificity requirements for the domains of Ess1/Pin1 are not well understood. The initial report examining the specificity of Pin1 found that its ability to bind peptides could be used as a basis for designing efficient isomerase substrates (26). This seemed to suggest the peptide binding site and prolyl isomerase site were the same; however, our study and the study mentioned above (29) strongly suggest it was the specificity of the WW domain that was responsible for peptide binding. Relevant to our work, a sequence found to be a preferred binding site, YpSP, represents a sequence po-tentially present in the phosphorylated CTD repeats. Peptide substrates with this sequence were also the best isomerase substrates. While the binding sites of the WW and isomerase domains are clearly distinct, they may still have similar recognition determinants.
WW domain associations with Pol II are not limited to the phosphorylated CTD. It has been shown that several WW domains, including one found in the mouse ubiquitin ligase, NEDD4, one found in human YAP (35), and at least one found in yeast Rsp5 (36), can associate with the unphosphorylated CTD. In agreement with our results, it has also been shown that the WW domain of Ess1 fails to interact with the unphosphorylated CTD (35). Interestingly, two studies have looked at the sequence requirements for peptide binding to the WW domain of human YAP, and both identified a sequence present in the CTD, PxY, as the most important for the binding site motif (37,38). The earlier and more thorough study found the position prior to proline was best occupied by another proline but was tolerant of Leu, Ser, Val, or Ala (37). The sequence SPxY is present in the CTD sequence, probably explaining the ability of human YAP to bind the unphosphorylated CTD.
Previous work has revealed a direct physical connection between 3Ј-end processing and the CTD. Mammalian cleavage stimulation factor (CstF) can be eluted from either a phosphorylated or an unphosphorylated CTD column, and the CstF p50 and p77 components have been shown to bind directly to the unphosphorylated CTD (19). Recently, both the unphosphorylated and the phosphorylated CTD have been found to be potent stimulators of the 3Ј cleavage reaction (20). A functional link between Ess1 and 3Ј-end processing was established by the isolation of mutations causing a 3Ј-end formation defect which mapped to the isomerase domain of Ess1 (21). Our demonstration that Ess1 stably associates with the phosphorylated CTD immediately suggests a mechanistic explanation for these observations: phospho-CTD binding by the WW domain localizes Ess1 to elongating Pol II, where it is positioned to isomerize pSP bonds present in the phospho-CTD. In addition, the separation of the binding and isomerase activities of Ess1 into independent domains creates the potential for processive isomerase function. Changes in CTD structure resulting from the activity of Ess1 would be expected to affect 3Ј-end formation by altering interactions between the CTD and cleavage/ polyadenylation components. Involvement of CTD kinase I in this process is consistent with the finding that a suppressor of CTDK-I deficiency-caused cold sensitivity contains several regions homologous to components of the 3Ј-end processing machinery in yeast and mammals. 4 The possibility that Ess1 reconfigures the phosphorylated CTD though processive isomerization of its proline residues certainly adds an unexpected twist to the investigation of eukaryotic pre-mRNA transcription and 3Ј-end formation.  1-9). Note that all the proteins include proteolyzed species although the smaller species are abundant only for GSTEss1ISO. Lane 10 contains 5 l of E. coli cells (BL21) from a culture grown to saturation. Lane 11 contains prestained markers with the indicated masses.
FIG. 4. The predominant Ess1 binding protein is the hyperphosphorylated form of the largest subunit of RNA Pol II. Multiple identical sets of lanes containing 1 and 5 l of total yeast extract separated by prestained markers (lanes 1, 4, 7, 10, and 13) were run on a 4 -15% SDS gel and transferred to nitrocellulose. Strips were cut and probed with GST, GSTEss1, GSTEss1WW, affinity-purified rabbit antiyeast phospho-CTD IgG, or GSTEss1ISO as indicated. Far-Western strips were reacted with affinity-purified rabbit anti-GST IgG, and all strips were developed using alkaline phosphatase-coupled goat antirabbit IgG as a 2 o antibody. The strips with GST and with GSTEss1ISO were developed twice as long as the others.