TFIIF-associating carboxyl-terminal domain phosphatase dephosphorylates phosphoserines 2 and 5 of RNA polymerase II.

The carboxyl-terminal domain (CTD) of the largest RNA polymerase (RNAP) II subunit undergoes reversible phosphorylation throughout the transcription cycle. The unphosphorylated form of RNAP II is referred to as IIA, whereas the hyperphosphorylated form is known as IIO. Phosphorylation occurs predominantly at serine 2 and serine 5 within the CTD heptapeptide repeat and has functional implications for RNAP II with respect to initiation, elongation, and transcription-coupled RNA processing. In an effort to determine the role of the major CTD phosphatase (FCP1) in regulating events in transcription that appear to be influenced by serine 2 and serine 5 phosphorylation, the specificity of FCP1 was examined. FCP1 is capable of dephosphorylating heterogeneous RNAP IIO populations of HeLa nuclear extracts. The extent of dephosphorylation at specific positions was assessed by immunoreactivity with monoclonal antibodies specific for phosphoserine 2 or phosphoserine 5. As an alternative method to assess FCP1 specificity, RNAP IIO isozymes were prepared in vitro by the phosphorylation of purified calf thymus RNAP IIA with specific CTD kinases and used as substrates for FCP1. FCP1 dephosphorylates serine 2 and serine 5 with comparable efficiency. Accordingly, the specificity of FCP1 is sufficiently broad to dephosphorylate RNAP IIO at any point in the transcription cycle irrespective of the site of serine phosphorylation within the consensus repeat.

fore, protein kinases and phosphatases that alter the state of CTD phosphorylation can serve as transcriptional activators or repressors depending on the point in the transcription cycle at which they function.
CTD phosphorylation occurs predominantly at serines 2 and 5 within the heptapeptide repeat. Genetic evidence indicates that the roles of serines in positions 2 and 5 are different. First, the partial substitution of serines in either position 2 or 5 have different effects on viability (8). Second, SRB (suppressors of RNA Polymerase IIB) mutations suppress the lethal effect of position 2 substitutions but not position 5 substitutions (9). Biochemical evidence has also confirmed differences between the two predominant serine positions. Serine 5 but not serine 2 phosphorylation recruits and activates the 5Ј-capping machinery (10,11). Furthermore, nutritional stress and heat shock can independently alter the pattern of CTD phosphorylation, indicating that phosphoserine 2 and phosphoserine 5 are functionally different (12)(13)(14).
A recent study using chromatin immunoprecipitation in Saccharomyces cerevisiae demonstrates that the pattern of CTD phosphorylation changes as RNAP II transcribes a given gene (15). Serine 5 phosphorylation is detected at the promoter regions, whereas serine 2 phosphorylation is increased as RNAP II leaves the promoter and transcribes the body of the gene. This dynamic phosphorylation of the CTD has functional consequences for the synthesis and processing of the primary transcript. During initiation, TFIIH phosphorylates the CTD at serine 5 (16,17). Phosphorylation presumably disrupts various protein-protein interactions important in the formation of the preinitiation complex (18). Serine 5 phosphorylation specifically facilitates the recruitment and activation of the mammalian capping enzyme (10,19). Capping of the 5Ј end of the RNA occurs as soon as the nascent transcript becomes accessible, usually at an RNA length of 25-30 nucleotides (20). Similarly in yeast, the recruitment and activation of Ceg1 (guanylyltransferase) and Abd1 (methyltransferase) are dependent on serine 5 phosphorylation (11,19,21,22). The retention of these capping enzymes displays a 5Ј-3Ј polarity on the gene (22). Ceg1 is released early in elongation, whereas Abd1 is released toward the 3Ј end of the gene.
Although it is unclear what happens to phosphoserine 5, an increase in serine 2 phosphorylation is observed as RNAP II elongates past position 200 (15). P-TEFb is the most likely candidate for serine 2 phosphorylation. P-TEFb preferentially phosphorylates serine 2 in early elongation complexes (23) and promotes processive transcript elongation by alleviating the negative effects of DSIF and NELF (24).
The dynamic phosphorylation of the CTD and the preferential phosphorylation of serine 2 and serine 5 can be viewed as molecular switches that control the progression of RNAP II and the recruitment of factors involved in the synthesis and processing of the primary transcript. The extent and specificity of CTD phosphorylation is maintained by the opposing actions of CTD kinases and CTD phosphatase(s). For example, Ctk1 (a putative P-TEFb homolog in yeast) and FCP1 (TFIIF-associating CTD phosphatase) appear to modulate the level of serine 2 phosphorylation in the RNAP II elongation complex (25). In addition, the level of phosphorylation of nontranscribing RNAP IIO in Xenopus laevis early embryos is maintained by MAP kinase Xp42 and FCP1 (26).
Unlike many CTD kinases that have been discovered and characterized, a single CTD phosphatase has been reported to date (for a review, see Ref. 27). Genetic studies have demonstrated that FCP1 is required for transcription in vivo, and its inactivation leads to a global defect in mRNA synthesis (28). The dephosphorylation of RNAP II is dependent on the interaction of FCP1 with a site on RNAP II that is outside of the CTD (29,30). FCP1 activity is stimulated by RAP74, the larger of the two subunits of TFIIF (29). TFIIB abrogates the stimulatory activity of TFIIF but has no influence on FCP1 activity in the absence of TFIIF.
FCP1 dephosphorylates RNAP IIO generated by serine/threonine CTD kinases but is not sensitive to vanadate, a tyrosine phosphatase inhibitor (31). Furthermore, the sensitivity of RNAP IIO in an elongation complex to FCP1 is dependent on its position with respect to the transcriptional start site (32,33). Although it has been established that FCP1 dephosphorylates phosphoserine 2 at the 3Ј end of the gene (25) and can recycle RNAP IIO to RNAP IIA (34), it is unclear if FCP1 can dephosphorylate phosphoserine 5 during transcript elongation (15). To understand the involvement of FCP1 at discrete stages in the transcription cycle, it is necessary to determine its substrate specificity. To examine FCP1 specificity, two independent experimental approaches were used in this study. First, FCP1 activity was assayed toward endogenous RNAP IIO populations contained in HeLa nuclear extracts. Second, FCP1 activity was assessed using purified calf thymus RNAP IIO substrates prepared in vitro by the phosphorylation of RNAP IIA with different CTD kinases.
Monoclonal Antibodies-Monoclonal antibody POL3/3 recognizes a conserved epitope within the largest RNAP II subunit that is distinct from the CTD (37,38). H5 and H14 are IgMs directed against phosphoepitopes within the CTD (39 -41) and were obtained from Covance. CC3 is an IgG isolated in a screen for chicken proteins with developmentally regulated expression (42). B3 is an IgM directed against nuclear matrix components (43).
Preparation of HeLa Nuclear Extracts-HeLa nuclear extracts were prepared from control cells and cells treated with actinomycin D at 1 g/ml for 1 h or serum-deprived for 24 h and stimulated with 20% serum for 1 h. After their respective treatments, the cell monolayers grown in 150-cm 2 dishes were washed with cold phosphate-buffered saline and gently removed by scrapping in buffer B (10 mM Hepes, pH 7.9, 1.5 mM MgCl 2 , 10 mM KCl, 0.5 mM DTT). Cells were centrifuged at 1,000 ϫ g for 10 min, and the pellets were resuspended in buffer B on ice. The cells were then homogenized with a Dounce homogenizer (10 times), and the lysates were centrifuged at 10,000 ϫ g for 10 min and fractionated into nuclear pellets and cytosolic supernatants. The nuclear pellets were resuspended in buffer C (20 mM Hepes, pH 7.9, 1.5 mM MgCl 2 , 25% glycerol, 420 mM NaCl, 0.2 mM EDTA, 0.5 mM DTT) and centrifuged at 15,000 ϫ g for 20 min. Pellets were resuspended in buffer D (20 mM Hepes, pH 7.9, 1.5 mM MgCl 2 , 25% glycerol, 1 M NaCl, 0.2 mM EDTA, 0.5 mM DTT) through a syringe to fragment the DNA and centrifuged at 15,000 ϫ g for 20 min. The nuclear extracts were dialyzed against buffer E (50 mM Tris, pH 7.9, 20% glycerol, 120 mM KCl, 0.1 mM EDTA, 1 mM DTT, 1 mM phenylmethylsulfonyl fluoride).
Preparation and Purification of 32 P-labeled RNAP IIO-Calf thymus RNAP IIA was purified by the method of Hodo and Blatti (44) with modifications as described by Kang and Dahmus (5). Specific isozymes of 32 P-labeled RNAP IIO were prepared by phosphorylation of purified RNAP IIA with recombinant casein kinase II and [␥-32 P]ATP followed by phosphorylation in the presence of excess unlabeled ATP (2 mM) with either purified CTDK1/CTDK2, TFIIH, P-TEFb, recombinant MAPK2/ ERK2, or Cdc2 kinase. Each RNAP IIO isozyme was purified by step elution from DE53 as previously described (4). Because only the most carboxyl-terminal serine (casein kinase II site) is labeled with 32 P and lies outside the consensus repeat, dephosphorylation by CTD phosphatase results in an electrophoretic mobility shift of subunit IIo to the position of subunit IIa without loss of label.
CTD Kinase Assays-CTD kinase assays were performed as described previously (35). Reactions were performed in 20 l of CTD kinase buffer (20 mM Hepes, pH 7.9, 8 mM MgCl 2 , 0.5% glycerol, 0.1% Triton X-100, 1 mM DTT). Each reaction contained 2.75 fmol of 32 Plabeled GST-CTDa and an equivalent molar amount of 32 P-labeled RNAP IIA. Reactions were initiated by the addition of either TFIIH, P-TEFb, or MAPK2/ERK2 and incubated at 30°C for 30 min. Assays were terminated by the addition of 5ϫ Laemmli buffer, and RNAP II subunits were resolved on a 5% SDS-PAGE gel. The gel image was scanned in a Molecular Dynamics Image Scanner Storm 860 in the phosphor screen mode and analyzed by ImageQuant software.
CTD Phosphatase Assays-Both purified and recombinant human FCP1 were used in these studies. Human FCP1 was purified from HeLa cells as described previously (45) and used in assays involving endogenous RNAP IIO substrates. Recombinant human FCP1 was expressed in Sf21 cells, purified to homogeneity, and used in assays involving RNAP IIO substrates prepared in vitro by CTD kinases. Recombinant FCP1 was purified by chromatography on Ni 2ϩ -nitrilotriacetic acidagarose (Qiagen) and HiTrap SP and Q in series (both from Amersham Biosciences) and HiTrap N-hydroxysuccinimide-activated column that had been coupled with recombinant RAP74 (Amersham Biosciences). Purified FCP1 had a specific activity of 40,000 units/mg, whereas recombinant FCP1 had a specific activity of 270,000 units/mg. One unit of CTD phosphatase corresponds to the activity required to convert 1 pmol of free RNAP IIO to RNAP IIA in 1 min in the presence of a saturating amount of RAP74.
CTD phosphatase assays were performed as described previously (31) with minor modifications. Reactions were performed in 20 l of CTD phosphatase buffer (50 mM Tris, pH 7.9, 10 mM MgCl 2 , 20% glycerol, 0.025% Tween 80, 0.1 mM EDTA, 5 mM DTT) in the presence of 20 mM KCl. Reactions involving endogenous RNAP II in HeLa nuclear extracts contained ϳ200 -250 fmol of RNAP IIO. Reactions involving purified RNAP II contained 18 fmol of RNAP IIO. Both reactions were carried out in the presence of 7 pmol of RAP74. Reactions were initiated by the addition of FCP1 and incubated at 30°C for 30 min. Assays were terminated by the addition of 5ϫ Laemmli buffer, and RNAP II subunits were resolved on a 5 or 6% SDS-PAGE gel. CTD phosphatase assays of endogenous RNAP IIO contained in nuclear extracts were analyzed by Western blots using dilutions of 1:1,000 POL3/3, 1:250 H5, 1:250 H14, 1:1,000 CC3, or 1:1,000 B3 followed by 1:10,000 anti-mouse IgG horseradish peroxidase-conjugated secondary antibody (Promega). The blots were visualized by ECL Plus detection system (Amersham Biosciences). CTD phosphatase assays of FCP1 on purified casein kinase II-labeled RNAP IIO were analyzed by autoradiography. The corresponding blot and gel images were scanned on a Molecular Dynamics Image Scanner Storm 860 in blue fluorescence mode and phosphor screen mode, respectively.

FCP1
Dephosphorylates Phosphoserine 2 and Phosphoserine 5 within the CTD of Endogenous RNAP IIO-FCP1 specificity was initially examined with in vivo populations of heterogeneous RNAP IIO contained in HeLa nuclear extracts. The reactivity of endogenous RNAP IIO with phosphoserine-specific monoclonal antibodies was assessed before and after dephosphorylation with FCP1. Based on reactivity with synthetic peptides, monoclonal antibodies H5 and CC3 recognize phosphoserine in position 2 within the heptapeptide repeat, whereas H14 and B3 recognize phosphoserine in position 5 ( Fig. 1A) (13). POL3/3, directed against an epitope in the largest RNAP II subunit that is outside of the CTD (37,38), permits the detection of the largest subunit irrespective of the state of CTD phosphorylation. The degree and the specificity by which FCP1 dephosphorylates endogenous RNAP IIO were measured by the disappearance of immunoreactivity. Because a variety of studies have shown that changes in growth conditions can give rise to changes in the level and pattern of CTD phosphorylation, HeLa nuclear extracts from differentially treated cells provide a source of heterogeneous RNAP IIO. The FCP1 sensitivity of RNAP IIO present in HeLa nuclear extracts of control cells, cells treated with actinomycin D, and cells stimulated with serum was examined. Treatment of cells with the transcription inhibitor actinomycin D or ␣-amanitin increases the ratio of RNAP IIO/IIA (46). Serum stimulation and heat shock increase the level of RNAP IIO via the cellular activation of MAPK2/ERK2 (47)(48)(49) and change the pattern of CTD phosphorylation (12)(13)(14). The increase in IIo signals, as indicated by increases in POL3/3 immunoreactivity relative to control extracts, confirms the above-mentioned studies that RNAP IIO levels are elevated in response to environmental stimuli ( Upon treatment of the three HeLa nuclear extracts with 100 milliunits of FCP1, the major fraction of RNAP IIO is converted to RNAP IIA (Fig. 1B, panel POL3/3, lanes 2, 4, and 6). FCP1 processively dephosphorylates endogenous RNAP IIO. As indicated by the disappearance of immunoreactivity of H5 and H14, FCP1 is capable of removing phosphates from serine 2 and serine 5 in control HeLa nuclear extract (Fig. 1B, panels  H5 and H14, lanes 1 and 2). Likewise, FCP1 shows similar specificity toward RNAP IIO in HeLa nuclear extract from cells treated with actinomycin D (Fig. 1B, panels H5 and H14, lanes  3 and 4). Interestingly, phosphoserine 2 is relatively resistant to FCP1 dephosphorylation in HeLa nuclear extract from cells stimulated with serum (Fig. 1B, panels H5 and H14, lanes 5 and 6). These observations are confirmed by the use of CC3 and B3 in parallel Western blots (Fig. 1B, panels CC3 and B3). In the presence of higher concentrations of FCP1, the complete dephosphorylation of phosphoserine 2 in serum-stimulated HeLa nuclear extract is observed (data not shown). These results indicate that FCP1 is capable of removing phosphates from serine positions 2 and 5.
To investigate whether the resistance of phosphoserine 2 to FCP1 dephosphorylation is conferred with increasing time of serum exposure, HeLa cells were serum-starved for 24 h, and nuclear extracts were prepared from cells treated with 20% serum for 0, 10, 30, and 60 min. The percentage of RNAP IIO that remained relatively resistant to dephosphorylation did not change as a function of time, suggesting that resistance of phosphoserine 2 to FCP1 dephosphorylation is conferred by serum starvation rather than serum stimulation (data not shown).
TFIIH and P-TEFb Preferentially Phosphorylate the CTD of RNAP IIA Relative to GST-CTDa-An alternative approach to establishing the specificity of FCP1 is to examine the ability of FCP1 to dephosphorylate RNAP IIO prepared in vitro by the phosphorylation of RNAP IIA with distinct CTD kinases. The phosphorylation of purified calf thymus RNAP IIA and GST-CTDa in the presence of increasing amounts of TFIIH (lanes 1-5), P-TEFb (lanes 6 -10), and MAPK2/ERK2 (lanes 11-15) is shown in Fig. 2. Both RNAP IIA and GST-CTDa are labeled with 32 P at their terminal serine by phosphorylation with casein kinase II. Because RNAP IIA and GST-CTDa are present in equimolar amounts in the same reaction, the efficiency with which each is shifted to the phosphorylated form is a measure of the substrate specificity of the CTD kinase present. The relative difference in the intensity of radiolabeled GST-CTDa and radiolabeled subunit IIa is a consequence of the difference in the efficiency of 32 P incorporation by casein kinase II.
TFIIH and P-TEFb efficiently convert RNAP IIA to RNAP IIO in a processive manner but show no or marginal activity toward GST-CTDa, respectively (Fig. 2, lanes 1-5 and lanes  6 -10). However, MAPK2/ERK2 converts both RNAP IIA and GST-CTDa to their phosphorylated forms in a distributive manner and with comparable efficiency (Fig. 2, lanes 11-15). This result indicates that the activities of TFIIH and P-TEFb are strongly dependent on the context in which the CTD is presented. In contrast, the activity of MAPK2/ERK2 appears to be insensitive to context. These findings suggest that the activities of TFIIH and P-TEFb are dependent on factors that are extrinsic to the CTD. Accordingly, reaction parameters defined on the basis of synthetic peptides or non-native substrates may differ significantly from those determined utilizing native RNAP IIA as substrate.
FCP1 Dephosphorylates RNAP IIO Prepared by Different CTD Kinases-The ability of FCP1 to dephosphorylate isozymes of purified calf thymus RNAP IIO, prepared by the in vitro phosphorylation of RNAP IIA with CTDK1/CTDK2, TFIIH, P-TEFb, MAPK2/ERK2, and Cdc2 kinase, was examined (Fig. 3A). A quantitation of the dephosphorylation of RNAP IIO as a function of increasing amounts of FCP1 is shown in Fig. 3B. RNAP IIO prepared by each of the five CTD kinases is efficiently dephosphorylated by FCP1. Furthermore, a comparable amount of FCP1 is required to dephosphorylate RNAP IIO prepared by each of the CTD kinases examined. Finally, the absence of subunits with mobilities intermediate between that of subunits IIa and IIo indicates that FCP1 processively dephosphorylates each isozyme of RNAP IIO. Kinetic assay, carried out at a fixed FCP1 concentration (5 milliunits), confirmed that FCP1 dephosphorylates RNAP IIO prepared by different CTD kinases with comparable efficiency (data not shown).
FCP1 Dephosphorylates RNAP IIO Prepared by TFIIH or Cdc2 Kinase at pH 7.9 but Not at pH 5.5-Synthetic peptides have been used extensively to characterize the activity of CTD kinases and more recently the activity of yeast FCP1 (50). Because of the reported low pH optimum of 5.5 required by yeast FCP1 to dephosphorylate synthetic peptides, the ability of human FCP1 to dephosphorylate isozymes of purified calf thymus RNAP IIO at pH 7.9 and pH 5.5 was examined (Fig. 4). FCP1 (50 milliunits) successfully converts RNAP IIO, prepared by the phosphorylation of RNAP II with either TFIIH or Cdc2 kinase, to RNAP IIA at pH 7.9 (lanes 1-4), whereas an equivalent amount of FCP1 shows no activity at pH 5.5 (lanes 5-8). This observation indicates that FCP1 activity toward native RNAP IIO is optimal near neutral or physiological pH.
The Specificity of TFIIH, P-TEFb, MAPK2/ERK2, and Cdc2 Kinase as Determined by Reactivity with Phosphoserine 2-and Phosphoserine 5-specific Monoclonal Antibodies-The observation that CTD kinases can differ in their relative activity with RNAP IIA and GST-CTDa suggests that their specificity may be directly influenced by determinants in RNAP II that lie outside the CTD. Accordingly, it is important to know if the preferential phosphorylation of either serine 2 or 5 observed in synthetic peptides by distinct CTD kinases is also true when RNAP IIA serves as substrate. The ability of phosphoserine 2or phosphoserine 5-specific monoclonal antibodies to react with RNAP IIO prepared by the phosphorylation of 32 P-labeled RNAP IIA with TFIIH, P-TEFb, MAPK2/ERK2, Cdc2 kinase, and CTDK1/CTDK2 was examined. The upper panel of Fig. 5  (Autoradiogram, lanes 2-6; see also Fig. 3A, lane 1) confirms that each CTD kinase efficiently converts RNAP IIA to IIO. There is a subtle variation in the mobility of the largest subunits among the five RNAP IIO isozymes prepared by the different CTD kinases. This variation suggests that the RNAP IIO isozymes may differ in phosphate stoichiometry, pattern of CTD phosphorylation, and/or structural conformations that can give rise to changes in gel mobility. The Western transfer of these same samples and reaction with POL3/3 shows a broader distribution of the largest subunit with mobilities intermediate between that of subunits IIo and IIa (Fig. 5, panel POL3/3). However, results presented in the upper panel of Fig. 5 show that casein kinase II-labeled RNAP IIA is quantitatively converted to fully phosphorylated RNAP IIO. This observation suggests that the fraction of RNAP IIA phosphorylated at the casein kinase II site is preferentially phosphorylated by each CTD kinase. Accordingly, the casein kinase II site may play a role in the regulation of RNAP II phosphorylation.
Immunoblots of the various RNAP IIO isozymes were probed with phosphoserine 2-specific monoclonal antibodies H5 and CC3 and phosphoserine 5-specific monoclonal antibodies H14 and B3. As expected, none of these monoclonal antibodies reacted with RNAP IIA (Fig. 5, lane 1). Interestingly, RNAP IIO prepared by phosphorylation with CTDK1/CTDK2, TFIIH, P-TEFb, and MAPK2/ERK2 (lanes 2-5) reacted more strongly with H5, H14, and B3 than with CC3. RNAP IIO prepared by Cdc2 kinase (lane 6) reacts most strongly to H5, moderately strong with CC3 and B3, and very weakly with H14. The intensity of each band was quantified and is presented above each lane as phosphor-stimulated luminescence per unit area. Although a direct comparison of reactivity is difficult given that the signal intensity is the consequence of multiple factors, it is clear that the specificity of Cdc2 kinase differs markedly from that of CTDK1/CTDK2, TFIIH, P-TEFb, and MAPK2/ERK2 (compare lane 6 with lanes 2-5 in each panel). The relative reactivity of Cdc2 kinase-phosphorylated RNAP IIO with phosphoserine 2-specific monoclonals (H5 and CC3) is 2-4 times higher than that of the other RNAP IIO isozymes. Conversely, Cdc2 kinase-phosphorylated RNAP IIO has a markedly reduced reactivity with phosphoserine 5 monoclonals (H14 and B3) relative to the other RNAP IIO isozymes. The results indicate that Cdc2 kinase has a higher propensity to phosphorylate serine 2 and a lower propensity to phosphorylate serine 5 than do the other CTD kinases tested. DISCUSSION It has become increasingly clear that reversible phosphorylation of serine 2 and serine 5 in the CTD consensus repeat plays a critical role in the progression of RNAP II through the transcription cycle and in coupling RNA processing with transcript elongation. However, a precise definition of the changes in phosphorylation pattern that occur has been difficult due to the repetitive nature of the CTD. Synthetic peptides have played a key role in establishing the specificity of distinct CTD kinases as well as the specificity of monoclonal antibodies that have been used as structural probes. It is now clear that the preference of CTD kinases for serine phosphorylation at positions 2 or 5 can be influenced by factors extrinsic to the kinase. For instance, Ramanathan and co-workers (51) report that Cdk9 (enzymatic subunit of P-TEFb) preferentially phosphorylates serine 5 of a synthetic CTD peptide that is three repeats long, whereas Zhou et al. (23) report that P-TEFb phosphorylates serine 2 of RNAP II assembled in preinitiation complexes. Remarkably, the human immunodeficiency virus type 1 transactivator protein Tat can modify P-TEFb specificity such that the kinase phosphorylates both serine 2 and serine 5. Furthermore, the specificity of Cdk7/cyclin H is altered not only by MAT1 (to form the Cdk-activating kinase (CAK) complex), but the specificity of the CAK complex itself is altered by its association with the other subunits in the native holo-TFIIH complex (52,53).
Because the specificity of CTD kinases is context-dependent and can be influenced by associating factors or even the substrate itself, experiments were carried out to determine their relative activity with respect to the phosphorylation of RNAP IIA and GST-CTDa. In the presence of equimolar amounts of RNAP IIA and GST-CTDa, both TFIIH and P-TEFb preferentially convert RNAP IIA to RNAP IIO while showing no activity or marginal activity toward GST-CTDa, respectively. Con- versely, MAPK2/ERK2 phosphorylates RNAP IIA and GST-CTDa with nearly identical efficiency. Finally, although both TFIIH and P-TEFb convert RNAP IIA to RNAP IIO in a processive manner, MAPK2/ERK2 converts both RNAP IIA and GST-CTDa to their respective phosphorylated forms in a distributive manner.
To determine the substrate specificity of FCP1, two experimental approaches were used. The first approach takes advantage of the finding that changes in growth conditions result in changes in the pattern of CTD phosphorylation (12)(13)(14). This study demonstrates that FCP1 can dephosphorylate endogenous RNAP IIO populations from HeLa nuclear extracts pre-pared from differentially treated cells. As determined by diminished reactivity with phosphoserine-specific monoclonal antibodies, FCP1 catalyzes the removal of phosphates from both serine 2 and serine 5.
In the second approach, a panel of RNAP IIO isozymes were individually prepared in vitro by the phosphorylation of purified calf thymus RNAP IIA with TFIIH, P-TEFb, MAPK2/ ERK2, Cdc2 kinase, and CTDK1/CTDK2. The pattern of phosphorylation by each CTD kinase was examined using a panel of monoclonal antibodies specific for serine 2 or serine 5 phosphorylation. The results presented here suggest that TFIIH, P-TEFb, and MAPK2/ERK2 have a higher propensity to phosphorylate serine 5 than does Cdc2 kinase. Conversely, Cdc2 kinase has a higher propensity to phosphorylate serine 2 than do the other CTD kinases. This is supported by the finding that the relative phosphoserine 2/phosphoserine 5 reactivity is similar among the panel of RNAP IIO prepared by TFIIH, P-TEFb, and MAPK2/ERK2 but differs markedly from that of RNAP IIO generated by Cdc2 kinase. The unanticipated reactivity of H5 with RNAP IIO isozymes prepared with TFIIH, P-TEFb, and MAPK2/ERK2 may result from either partial phosphorylation at serine 2 or cross-reactivity of H5 with serine 5 in the context of native RNAP II.
Results presented here using free RNAP II as substrate are in general agreement with studies using synthetic CTD peptides (51,54,55). TFIIH, P-TEFb, and MAPK2/ERK2 all phosphorylate free RNAP II at serine 5, whereas Cdc2 kinase phosphorylates the two substrates at both serines 2 and 5. As noted above, Zhou et al. (23) report that P-TEFb phosphorylates serine 2 in RNAP II assembled in a preinitiation complex. Just as the ability of FCP1 to dephosphorylate RNAP II differs dramatically between free RNAP II and RNAP II in an elongation complex (32,33), the specificity of CTD kinases may differ between free RNAP II and RNAP II in specific protein-DNA complexes. Alternatively, RNAP II assembled in a preinitiation complex in the presence of HeLa nuclear extract may be associated with a factor(s) that alters the specificity of enzymes that modify the CTD.
Although RNAP IIO isozymes prepared by TFIIH and P-TEFb were slightly better substrates for FCP1 relative to RNAP IIO prepared by Cdc2 kinase, each RNAP IIO prepared in vitro can be efficiently dephosphorylated by FCP1 and successfully converted to RNAP IIA. This result indicates that FCP1 can remove phosphates from both serine positions 2 and 5. FCP1 specificity determined here using purified calf thymus RNAP IIO made by known CTD kinases corroborates the specificity established using endogenous RNAP IIO contained in HeLa nuclear extracts. These findings establish that FCP1 has a broad specificity and is capable of dephosphorylating different isozymes of RNAP IIO present at different stages in the transcription cycle. FIG. 4. FCP1 activity at pH 7.9 and pH 5.5. RNAP IIO, prepared by the phosphorylation of 32 P-labeled RNAP IIA with TFIIH or Cdc2 kinase, was incubated in the presence of increasing amounts of FCP1 at pH 7.9 or 5.5 and analyzed as described under "Experimental Procedures." Each reaction contained ϳ18 fmol of RNAP IIO, 7 pmol of RAP74, and a specified amount of FCP1 (increasing from lane 1 to 4 and from lane 5 to 8) at either pH 7.9 (lanes 1-4) or pH 5.5 ( lanes  5-8). mU, milliunits. A recent study on the specificity of Schizosaccharomyces pombe FCP1, in which phosphate release from synthetic CTD peptides was determined, report that FCP1 preferentially dephosphorylates phosphoserine 2 over phosphoserine 5 (50). The apparent discrepancy between these results and those presented here might be a consequence of differences in the source of FCP1, the substrate, and/or reaction conditions. In the study reported here, human FCP1 was assayed at pH 7.9 with native RNAP IIO substrates, whereas the study by Hausmann and Shuman (50) used fission yeast FCP1 at pH 5.5 with phospho-CTD peptides. The pH optimum for yeast FCP1 hydrolysis of p-nitrophenol phosphate and phospho-CTD peptides is 5.5, whereas the pH optimum for human FCP1 toward native RNAP IIO is about 7.9, with no apparent activity at pH 5.5. Furthermore, in the presence of equimolar amounts of RNAP IIO and GST-CTDo (both prepared by MAPK2/ERK2), FCP1 successfully dephosphorylates native polymerase but not GST-CTDo at pH 7.9 (data not shown). These results indicate that at or near neutral or physiological pH, free native RNAP II, but not the recombinant CTD, is the preferred substrate for FCP1. Last, the amount of FCP1 required to dephosphorylate RNAP IIO at pH 7.9 is substantially less than the amount used for the dephosphorylation of phospho-CTD peptides at pH 5.5.
The pattern of CTD phosphorylation can influence the recruitment of RNAP II to the promoter as well as coordinate transcript elongation and RNA processing. Using chromatin immunoprecipitation in yeast, Komarnitsky et al. (15) demonstrate that serine 5 phosphorylation is primarily detected at the promoter region, whereas serine 2 phosphorylation is observed in the coding region. Due to limitations on the use of monoclonal antibodies as structural probes, it is still unclear whether phosphates on serine 5 are removed before the addition of phosphates on serine 2. For example, it is possible that phosphorylation at position 2 alters the affinity of monoclonal antibodies directed against position 5 and vice versa.
The observation that serine 2 phosphorylation increases in promoter distal regions in fcp1 mutants indicates that FCP1 is responsible for the turnover of serine 2 phosphates during transcript elongation (25). The results here showing that FCP1 can remove phosphates from serine 2 are in agreement with these studies and are consistent with the idea that FCP1 can dephosphorylate RNAP IIO concomitant with or shortly after termination to replenish the pool of RNAP IIA (34). Conversely, the finding that serine 5 phosphorylation is not altered in fcp1 mutants suggests that FCP1 may not participate in the potential dephosphorylation of phosphoserine 5 during early elongation (25). Using similar assays with some modifications, Schroeder et al. (22) found that FCP1 can modulate phosphoserine 5 levels and direct the dissociation of capping enzymes. Differences in the temperature at which the fcp1 mutants were assayed may contribute to the discrepancy in these results.
It is important to definitively establish whether serine 5 phosphates turnover either partially or completely during early elongation. If serine 5 phosphates are indeed removed early during transcript elongation, the broad specificity of FCP1 makes it a likely candidate for acting at this stage of transcription. However, it is possible that another CTD phosphatase is responsible for the turnover of serine 5 phosphates and that FCP1 activity is down-regulated in early elongation. The broad specificity of FCP1 also makes it a likely candidate for 1) modulating RNAP II activity during different stages of transcription and 2) participating in the mobilization of RNAP IIO from storage sites. The key to understanding FCP1 involvement in the regulation of gene expression lies in determining the factors that influence its recruitment to different RNAP II-containing complexes.