Schizosaccharomyces pombe carboxyl-terminal domain (CTD) phosphatase Fcp1: distributive mechanism, minimal CTD substrate, and active site mapping.

Schizosaccharomyces pombe Fcp1 is an essential protein serine phosphatase that preferentially dephosphorylates Ser(2) of the RNA polymerase II C-terminal domain (CTD) heptad repeat Y(1)S(2)P(3)T(4)S(5)P(6)S(7). Here we show that: (i) Fcp1 acts distributively during the hydrolysis of substrates containing tandem Ser(2)-PO(4) heptads; (ii) the minimal optimal CTD substrate for Fcp1 is a single heptad of phasing S(5)P(6)S(7)Y(1)S(2)P(3)T(4); and (iii) single alanine mutations of flanking residues Tyr(1) or Pro(3) result in 6-fold decrements in CTD phosphatase activity. Fcp1 belongs to the DXDX(T/V) family of phosphotransferases that act via an acyl-phosphoenzyme intermediate. An alanine scan of 11 conserved positions of S. pombe Fcp1 identifies Thr(174), Tyr(237), Thr(243), and Tyr(249) as important for phosphatase activity. Structure-activity relationships at these positions were determined by introducing conservative substitutions. Our results, together with previous mutational studies, highlight a constellation of 11 amino acids that are conserved in all Fcp1 orthologs and likely comprise the active site.

The C-terminal domain (CTD) 1 of the largest subunit of RNA polymerase II (pol II) is composed of a tandemly repeated heptapeptide of consensus sequence Y 1 S 2 P 3 T 4 S 5 P 6 S 7 . CTD positions Ser 5 and Ser 2 undergo waves of phosphorylation and dephosphorylation during the transcription cycle, the purpose of which may be to regulate the transition from initiation to elongation modes and to control the recruitment, activity, and egress of the various mRNA-processing machines that act on the nascent transcript (1)(2)(3). The dynamic phosphorylation state of the CTD reflects a kinetic balance between the multiple CTD kinase and CTD phosphatase activities found in eukaryotic cells (2, 4 -7).
The enzyme Fcp1 is the major protein serine phosphatase responsible for removing phosphates from the CTD (2, 8 -16). Fcp1 orthologs are present in all known eukaryal proteomes, and the enzyme is essential for cell viability in budding and fission yeast (10,15). A partial deficiency of human Fcp1 is associated with an autosomal recessive developmental disorder characterized by cataracts, facial dysmorphism, and peripheral neuropathy (17).
Fcp1 catalyzes the metal-dependent hydrolysis of phosphoserine from the CTD in the context of the pol II elongation complex, isolated pol II, and (in the case of Schizosaccharomyces pombe Fcp1) synthetic CTD phosphopeptide substrates. Fcp1 also hydrolyzes the nonspecific substrate p-nitrophenyl phosphate (13,16). Fcp1 orthologs are relatively large polypeptides (723 aa in S. pombe; 732 aa in Saccharomyces cerevisiae; 961 aa in humans) consisting of a conserved central catalytic domain flanked by variable N-and C-terminal segments that are dispensable for CTD phosphatase activity in vitro (16,18), but are collectively required for Fcp1 function in vivo (19). The minimal phosphatase domain of S. pombe Fcp1 spans amino acids 156 -580 (18). The catalytic domain is itself composed of two domain modules: an N-terminal FCPH (FCP1 homology) domain spanning S. pombe Fcp1 residues 140 -326 (see Fig. 1) and a C-terminal BRCT (BRCA1 C terminus) domain, both of which are essential for Fcp1 phosphatase activity in vivo and in vitro (14,16,19). Although the catalytic contributions made by the BRCT domain are not known, a recent report implicates the BRCT domain in the initial binding of Fcp1 to the phosphorylated CTD (20). A wealth of data implicate the FCPH domain as the seat of the phosphatase active site.
A short conserved peptide motif ( 170 DLDQT 174 in S. pombe Fcp1) located near the N-terminal margin of the FCPH domain corresponds to the signature sequence of the DXDX(T/V) family of metal-dependent phosphohydrolases and phosphotransferases (21)(22)(23) (Fig. 1). DXDX(T/V) enzymes act via an acylphosphoenzyme intermediate in which the phosphate is linked to the first aspartate in the DXDX(T/V) motif (21, 24 -26). The second aspartate in the DXDX(T/V) motif serves as a general acid catalyst that donates a proton to the leaving group during formation of the acyl-phosphate intermediate. Formation and hydrolysis of the acyl-phosphoenzyme proceed via an associative mechanism through a pentacoordinate phosphorane transition state (27). Insights to the identity of the catalytic functional groups on the enzyme have emerged from crystallographic snapshots of DXDX(T/V) acyl-phosphatases captured at defined stages of the reaction cycle (25,27,28) and from mutational analyses of exemplary DXDX(T/V) family members, including Fcp1.
We previously employed site-directed mutagenesis to locate candidate catalytic residues of S. pombe Fcp1 (16,18). We found that alanine substitutions for Asp 170 , Asp 172 , Arg 223 , Asp 258 , Lys 280 , Asp 297 , and Asp 298 abolished phosphatase activity with either p-nitrophenyl phosphate or CTD-PO 4 as substrates. Structure-activity relationships were subsequently de-termined by introducing conservative substitutions at each essential position. Because the seven essential amino acids identified in S. pombe Fcp1 are conserved in Fcp1 orthologs from other organisms, we predicted that they comprise the phosphatase active site.
A goal of the present study was to more fully delineate the phosphatase active site and the catalytic mechanism of Fcp1. Therefore, we conducted additional mutational analysis of the FCPH domain, guided by sequence comparisons to diverse Fcp1 homologs (Fig. 1). We report the effects of 17 mutations at 11 conserved positions of S. pombe Fcp1, which identify four new residues (Thr 174 , Tyr 237 , Thr 243 , and Tyr 249 ) as important for activity with both specific and nonspecific substrates. Based on the crystal structures available for related DXDX(T/V) phosphatases, we discuss plausible catalytic roles for the essential side chains of Fcp1.
Although the mutational studies illuminate the chemical mechanism of Fcp1, they do not instruct us on the basis for specific recognition of the phosphorylated CTD. Previously, we showed that S. pombe Fcp1 displays an inherent preference for a particular CTD phosphorylation array. Using equivalent 28-aa CTD peptides of identical amino acid sequence and phosphoserine content, which differed only in the positions of phosphoserine within the tetraheptad sequence, we found that Fcp1 was 10-fold more active in dephosphorylating Ser 2 -PO 4 than Ser 5 -PO 4 (16). Here we systematically investigated the structural features of the CTD Ser 2 -PO 4 substrate that promote optimal phosphatase activity. Our results show that: (i) Fcp1 acts distributively during the hydrolysis of substrates containing tandem Ser 2 -PO 4 heptads; (ii) Fcp1 fails to hydrolyze the N-terminal Ser 2 -PO 4 -containing heptad when the substrate is phased as (Y 1 S 2 P 3 T 4 S 5 P 6 S 7 ) n ; (iii) the minimal optimal CTD substrate for Fcp1 is a single heptad of phasing S 5 P 6 S 7 Y 1 S 2 P 3 T 4 ; and (iv) single alanine mutations of flanking FIG. 1. Conservation of the primary structure of the FCPH domain. The amino acid sequence of S. pombe (Spo) Fcp1 from residues 140 -326 is aligned to the sequences of Fcp1 orthologs from S. cerevisiae (Sce), Leptosphaeria maculans (Lma), Neurospora crassa (Ncr), Aspergillus nidulans (Ani), Candida albicans (Cal), Xenopus laevis (Xla), Drosophila melanogaster (Dme), Anopheles gambiae (Aga), Homo sapiens (Hsa), Caenorhabditis elegans (Cel), Dictyostelium discoideum (Ddi), and Encephalitozoon cuniculi (Ecu). Gaps in the alignment are indicated by dashes. The 16 residues of S. pombe Fcp1 that were subjected previously to mutational analysis are indicated by dots (•) above the sequence alignment. The seven essential residues defined as essential for phosphatase activity are shaded. The 11 amino acids targeted for mutational analysis in the present study are indicated by the "|" symbol. residues Tyr 1 or Pro 3 result in 6-fold decrements in CTD phosphatase activity.

EXPERIMENTAL PROCEDURES
Fcp1 Mutants-Amino acid substitution mutations (and diagnostic restriction sites) were introduced into the fcp1 ϩ cDNA by the two-stage overlap extension method as described previously (16). pET-Fcp1 was used as the template for the first-stage amplification. The mutated full-length cDNAs generated in the second-stage amplification were digested with NcoI and BamHI and then inserted into pET16m. The inserts of the resulting pET-Fcp1* plasmids were sequenced completely to confirm the desired mutations and exclude the acquisition of unwanted changes during amplification or cloning. The pET-Fcp1* and pET-Fcp1 plasmids were introduced into Escherichia coli BL21(DE3)-RIL, and the mutant Fcp1 proteins were purified from soluble bacterial lysates by nickel-agarose affinity chromatography as described previously for wild-type Fcp1 (16). Protein concentrations were determined by using the Bio-Rad dye reagent with bovine serum albumin as the standard.
CTD Phosphopeptides-The following CTD Ser 2 -PO 4 peptides containing an unmodified N-terminal amine and a C-terminal acid were synthesized and purified by the Sloan-Kettering Microchemistry Core Laboratory as described in Ref. 29 (the Ser 2 -PO 4 residues are underlined): 28-mer YSPTSPSYSPTSPSYSPTSPSYSPTSPS; 14-mer YSPTSPSYSPTSPS; 14-mer YSPTSPSYSPTSPS; 14-mer YSPTSP-SYSPTSPS; 12-mer PTSPSYSPTSPS; 10-mer SPSYSPTSPS; 8-mer SYSPTSPS; 10-mer (Y1A), SPSASPTSPS; 10-mer (P3A), SPSY-SATSPS. Additional CTD Ser 2 -PO 4 peptides containing an N-terminal amine and a C-terminal amide were as follows: 9-mer SPSYSPTSP; 8-mer SPSYSPTS; 7-mer SPSYSPT. The peptides were dissolved in 10 mM Tris-HCl (pH 7.4), 1 mM EDTA and stored at 4°C. The molar concentrations of the phosphopeptides were initially estimated from the absorbance at 274 nM using an extinction coefficient of 1.4 ϫ 10 3 M Ϫ1 for tyrosine. The content of Ser 2 -PO 4 was then determined for each peptide by measuring the release of inorganic phosphate after digestion with calf intestinal phosphatase (CIP; purchased from Roche Applied Science) as follows: CIP reaction mixtures (25 l) containing ϳ2.5 nmol of CTD phosphopeptide and CIP (2 units) were incubated for 60 min at 37°C. The reactions were quenched by adding 0.5 ml of malachite green reagent (BIOMOL Research Laboratories, Plymouth Meeting, PA). Release of phosphate was determined by measuring A 620 and interpolating the value to a phosphate standard curve. The phosphopeptide concentrations measured by CIP were in good agreement (Ϯ10%) with the concentrations calculated from the UV absorbance. The amounts of input CTD Ser 2 -PO 4 substrate specified in the Fcp1 reactions are based on the concentrations determined by CIP digestion.
Phosphatase Assay-Reaction mixtures (100 l) containing 50 mM Tris acetate (pH 5.5), 10 mM MgCl 2 , 10 mM p-nitrophenyl phosphate (pNØP), and Fcp1 as specified were incubated for 30 min at 37°C. The reactions were quenched by adding 900 l of 1 M sodium carbonate. Release of p-nitrophenol (pNØ) was determined by measuring A 410 and interpolating the value to a pNØ standard curve.
CTD Phosphatase Assay-Reaction mixtures (25 l) containing 50 mM Tris acetate (pH 5.5), 10 mM MgCl 2 , CTD phosphopeptide, and Fcp1 were incubated for 60 min at 37°C. The reactions were quenched by adding 0.5 ml of malachite green reagent. Release of phosphate was determined by measuring A 620 and interpolating the value to a phosphate standard curve.
Product Analysis by MALDI-TOF Mass Spectrometry-Reaction mixtures (300 l) containing 50 mM Tris acetate (pH 5.5), 10 mM MgCl 2 , 25 M phosphopeptide (YSPTSPS) 4 (ϭ 100 M Ser 2 -PO 4 ) and 0.3 M Fcp1 were incubated at 37°C. Aliquots (25 l) were withdrawn at the times specified, and the reaction was quenched by adding 10 l of 5 M formic acid. Aliquots (1 l) of the mixture were diluted 100-fold with 0.1% formic acid and loaded onto a 2-l column of Poros 50 R2 reverse-phase chromatography beads (PerSeptive Biosystems, Framingham, MA) packed into an Eppendorf gel-loading tip. The adsorbed peptides were eluted with 8 l of 30% acetonitrile, 0.1% formic acid. An aliquot (0.5 l) of the eluted peptide pool was analyzed by matrix-assisted laser-desorption/ionization (MALDI) time-of-flight (TOF) mass spectrometry (MS), using a Bruker UltraFlex TOF/TOF instrument (Bruker Daltonics, Bremen, Germany), as described previously (40). The spectra were obtained in the negative ion mode, by averaging multiple acquisitions. The distribution of tetra-, tri-, di-, and monophosphorylated CTD peptides in each sample (Fig. 3B) was calculated as the ratio of the observed peak heights of the individual tetra-, tri-, di-, or monophosphorylated peptides to the sum of the heights of the tetra-, tri-, di-, and monophosphorylated peptide peaks.

S. pombe Fcp1 Preferentially Dephosphorylates Ser 2 at
Neutral pH-We reported previously that purified recombinant S. pombe Fcp1 dephosphorylates a 28-aa CTD phosphopeptide (YSPTSPS) 4 consisting of four tandem repeats of the CTD heptad sequence in which all Ser 2 positions are Ser-PO 4 (16). S. pombe Fcp1 was 10-fold more active with this Ser 2 -PO 4 CTD substrate than with a 28-aa CTD phosphopeptide (YSPTSPS) 4 consisting of four heptads in which all Ser 5 positions are Ser-PO 4 (16). These in vitro experiments, which provided the first biochemical evidence that any Fcp1 ortholog has intrinsic preference for a particular phosphorylation array, were performed at pH 5.5, which is the optimal pH for the catalytic activity of S. pombe Fcp1 with both nonspecific (pNØP) and specific (CTD-PO 4 ) substrates (16). Our biochemical data for Ser 2 -PO 4 preference by S. pombe Fcp1 agree with genetic experiments showing that transient inactivation of S. cerevisiae Fcp1 in vivo results in a selective increase in the steady-state level of Ser 2 -PO 4 in the large subunit of template-engaged RNA polymerase II (41). Despite the concordance of these in vitro and in vivo studies, our biochemical evidence for Ser 2 -PO 4 preference by S. pombe Fcp1 has been called into question in a recent review article because our experiments were performed at pH 5.5 (42). Although S. pombe maintains neutral intracellular pH (between pH 7.1 and 7.3 (43)), it is sensible that S. pombe Fcp1, as a DXDX(T/V) phosphatase, would display optimal activity at mildly acidic pH in vitro, because the phosphoryl transfer mechanism calls for an unprotonated aspartate nucleophile and a protonated aspartate general acid catalyst. To explore this issue further, we evaluated the substrate specificity of purified S. pombe Fcp1 at pH 7.0 (Fig. 2). We see that Fcp1 preferentially dephosphorylates the Ser 2 -PO 4 substrate at neutral pH. From the slopes of the titration curves, we calculated that the specific activity with the (YSPTSPS) 4 substrate was 6-fold greater than the activity with the (YSPTSPS) 4 substrate. These data vitiate the notion that pH 5.5 somehow provoked S. pombe Fcp1 to prefer Ser 2 -PO 4 .
Fcp1 Is a Distributive Enzyme-The kinetics of dephosphorylation of the (YSPTSPS) 4  strate excess are shown in Fig. 3A. Approximately 14% of the input phosphoserine residues were hydrolyzed after 2 min; from this initial rate we calculated a turnover number of 0.39 s Ϫ1 . Phosphate release increased with time and reached a plateau at 60 min, at which time about 70% of the input phosphoserine residues had been hydrolyzed. The end point value at 60 min was not increased when the concentration of Fcp1 was increased 3-fold (not shown), suggesting that either: (i) about one-fourth of the input substrate peptide was unreactive with Fcp1 or (ii) one of the Ser 2 -PO 4 sites within the tetraheptad peptide was protected from Fcp1.
To explore this issue, aliquots of the reaction mixture were withdrawn at serial times, quenched with acid, and analyzed by MALDI-TOF mass spectrometry, which fractionates the peptide pool based on molecular mass. This allowed us to gauge the distribution of products as the input tetraphosphorylated CTD substrate was converted to species of lower mass differing by multiples of 80-Da increments (signifying the loss of integral numbers of phosphate groups). The first instructive findings emerged from the product distribution at the 60-min reaction end point, showing that there was no residual tetraphosphorylated CTD substrate and that 87% of the product corresponded to singly phosphorylated CTD peptide (1ϫPO 4 ) and 13% corresponded to doubly phosphorylated CTD (2ϫPO 4 ) (Fig. 3B). These results confirmed that all of the input CTD substrate was reactive with Fcp1 and that the reaction ceased after three of the four phosphates were hydrolyzed. The structural basis for this restricted end point is explored in depth below.
The evolution of the product distribution during the 60-min reaction (Fig. 3B) allowed a clear discrimination between processive and distributive modes of action by S. pombe Fcp1 on a multiply phosphorylated CTD array. In a purely processive mechanism under conditions of substrate excess, Fcp1 would sequentially hydrolyze all available Ser 2 -PO 4 on the peptide to which it initially bound before dissociating to a new substrate peptide. In that event, we would expect to detect at all times a product distribution consisting of a mixture of the unreacted 4ϫPO 4 substrate and the 1ϫPO 4 CTD "end-product," with a steady increase in the abundance of end-product as a function of time and little accumulation of CTD peptides with intermediate phosphorylation states. In contrast, in a purely distributive mechanism, Fcp1 would dissociate from the CTD peptide after hydrolysis of only one Ser 2 -PO 4 position and then be available to rebind to other CTD peptides in the substrate pool. In a distributive mechanism, we would expect to see a serial flux of product through each of the dephosphorylated species, i.e. that the triphosphorylated CTD (3ϫPO 4 ) predominates at the earliest times before it decays to a diphosphorylated CTD (2ϫPO 4 ), which is converted in turn to the 1ϫPO 4 CTD endproduct. Indeed, the distributive pattern is exactly what we observed (Fig. 3B). At the earliest time analyzed (2 min), virtually all of the input substrate had been consumed and converted primarily to 3ϫPO 4 CTD (78%) and to a lesser extent to 2ϫPO 4 CTD (22%). The 3ϫPO 4 species declined steadily in abundance over 20 min. The 2ϫPO 4 CTD accumulated steadily between 2 and 10 min, plateaued at 10 -20 min (comprising 62% of the CTD pool) and declined thereafter, concomitant with the steady accumulation of the 1ϫPO 4 product. Note that no 1ϫPO 4 product was detected at the earliest times (Յ5 min) and that this species did not appear until such time as a significant fraction of the CTD had been converted to 2ϫPO 4 CTD (Fig.  3B). These results provide strong evidence for a distributive catalytic mechanism for hydrolysis of Ser 2 -PO 4 . The kinetic profile also hinted that removal of the first phosphate was faster than removal of the second phosphate, which was in turn faster than removal of the third phosphate.
Fcp1 Dephosphorylates a CTD Phosphopeptide Containing Two Heptad Repeats-The preceding experiment using a tetraheptad CTD Ser 2 -PO 4 substrate showed clearly that Fcp1 is capable of dephosphorylating three of the four heptads. We considered two possible explanations for this result: (i) Fcp1 requires two Ser 2 -PO 4 moieties to bind and hydrolyze the CTD or (ii) Fcp1 requires a minimal length CTD peptide flanking each Ser 2 -PO 4 and that either the N-terminal heptad or the C-terminal heptad of the 28-aa substrate lacks the required flanking residues.
To address this issue, we synthesized a 14-aa diheptad CTD peptide YSPTSPSYSPTSPS phosphorylated at Ser 2 of each heptad. Reaction of Fcp1 with 100 M of the 2ϫPO 4 CTD peptide resulted in P i release proportional to the amount of input enzyme in the range of 0.16 -1.25 g of Fcp1 (Fig. 4A, titration curve (q)); 48% of the input phosphoserine was hydrolyzed at saturating enzyme levels (2.5-5 g of Fcp1). Apparently, only one of the two available Ser 2 -PO 4 heptads served as a substrate for Fcp1. To determine which phosphoserine was dephosphorylated and which was resistant, we synthesized two 14-mer monophosphorylated CTD peptides, which contained Ser 2 -PO 4 in either the N-terminal heptad (YSPTSPSYSPTSPS) or the C-terminal heptad (YSPTSPSYSPTSPS). Reaction of Fcp1 with 100 M of the 1ϫPO 4 C-terminal peptide resulted in quantitative P i release that was proportional to input Fcp1 (Fig. 4A, titration curve (f)). In contrast, Fcp1 was unreactive with 100 M of the 1ϫPO 4 N-terminal peptide at up to 5 g of input enzyme (Fig. 4A, titration curve (Ⅺ)). From the slopes of the titration curves, we calculated that Fcp1 specific activity with the 1ϫPO 4 N-terminal peptide was 0.5% of the specific activity with the 1ϫPO 4 C-terminal peptide substrate. A kinetic analysis under conditions of substrate excess showed that the rate and extent of P i release were similar for the 14-mer 2ϫPO 4 substrate and the 1ϫPO 4 C-terminal phosphoheptad substrate (Fig. 4B). From the slopes of the curves, we calculated turnover numbers of ϳ0.09 s Ϫ1 for the two diheptad CTD substrates.
These experiments demonstrate that: (i) a diheptad CTD peptide suffices for efficient hydrolysis of Ser 2 -PO 4 by Fcp1; (ii) a single Ser 2 -PO 4 residue suffices for efficient Fcp1 activity, provided that it is phased in the appropriate context with as few as five residues on the C-terminal side of the Ser 2 -PO 4 ; and (iii) Ser 2 -PO 4 cannot be hydrolyzed by Fcp1 when only a single Tyr residue is present on its N-terminal flank.

Defining the Minimal CTD Substrate for S. pombe Fcp1-
The requirements for CTD primary structure on the N-terminal side of the Ser 2 -PO 4 residue were probed by comparing Fcp1 activity with a series of N-terminal deleted, singly phosphorylated 12-mer (PTSPSYSPTSPS), 10-mer SPSYSPTSPS, and 8-mer (SYSPTSPS) peptides, which contained five amino acids C-terminal to Ser 2 -PO 4 and either 6, 4, or 2 amino acids on the N-terminal side, respectively (Fig. 5A). Fcp1 catalyzed quantitative release of P i from the 12-mer PTSPSYSPTSPS and 10-mer SPSYSPTSPS peptides with similar specific activities (these being 80 and 94%, respectively, of the specific activity with the 14-mer diheptad 1ϫPO 4 CTD substrate used in Fig. 4). In contrast, Fcp1 was much less active in dephosphorylating the 8-mer peptide SYSPTSPS (specific activity being 5% of that observed with the diheptad 1ϫPO 4 CTD substrate). We conclude that 4 amino acids N-terminal to the Ser 2 -PO 4 sufficed for optimal Fcp1 activity.
The requirements on the C-terminal side of the Ser 2 -PO 4 residue were probed with incrementally truncated 9-mer (SP-SYSPTSP), 8-mer (SPSYSPTS), and 7-mer (SPSYSPT) pep- tides, which contained the minimal 4 amino acids N-terminal to Ser 2 -PO 4 and either 4, 3, or 2 amino acids on the C-terminal side, respectively (Fig. 5B). All three of the C-terminal truncated phosphopeptides were dephosphorylated quantitatively by Fcp1 (Fig. 5B), with specific activities that were virtually identical to that observed with the diheptad 1ϫPO 4 substrate. We conclude that as few as two amino acids on the C-terminal side of Ser 2 -PO 4 suffice for optimal Fcp1 phosphatase activity and that a single heptad of phasing S 5 P 6 S 7 Y 1 S 2 P 3 T 4 comprises a minimal effective CTD substrate.
Role of CTD Residues Tyr 1 and Pro 3 in Fcp1-catalyzed Dephosphorylation of Ser 2 -To investigate the role of the amino acid side chains flanking phosphoserine as determinants of Fcp1 activity, we tested mutated versions of the 10-mer CTD phosphopeptide SPSYSPTSPS, wherein Tyr 1 or Pro 3 was replaced individually by alanine (Fig. 6). The titration profiles of the reaction of Fcp1 with 100 M of the Y1A (SPSASPTSPS) or P3A (SPSYSATSPS) substrates displayed a clear shift to the right compared with the wild-type CTD peptide (Fig. 6). The specific activities of Fcp1 with the Y1A and P3A 10-mer substrates were 16 and 17%, respectively, of the wild-type 10-mer CTD substrate. Thus, Tyr 1 and Pro 3 are both important for optimal CTD phosphatase activity of Fcp1. As discussed below, these results provide insight to the basis for preferential dephosphorylation of CTD Ser 2 -PO 4 by S. pombe Fcp1.
Alanine-scanning Mutagenesis of Fcp1-Previously, we initiated efforts to map the active site of Fcp1 by alanine scanning of the conserved residues of the FCPH domain; of the 16 positions already mutated (indicated by • in Fig. 1), seven were found to be essential for phosphatase activity (shaded boxes in Fig. 1) (16, 18). Here we extended the analysis by introducing single alanine changes at 11 new amino acids of the FCPH domain that are conserved in other Fcp1 orthologs. The mutated positions are indicated by | in Fig. 1. We focused in particular on basic (Lys 163 , Arg 164 , Lys 221 , and Lys 247 ), acidic (Asp 273 ), and polar (Gln 173 , Thr 174 , Tyr 237 , Tyr 249 , Thr 243 , and Ser 270 ) side chains. Wild-type Fcp1 and the 11 Fcp1-Ala mu-tants were produced in bacteria as His 10 -tagged fusions and purified from soluble bacterial extracts by nickel-agarose chromatography. SDS-PAGE analysis of the imidazole eluate fractions showed that the preparations were highly enriched with respect to the His-Fcp1 polypeptide (Fig. 7).
Initial characterization of the generic phosphatase activity of the Fcp1-Ala proteins was performed using 10 mM pNØP as a substrate (Fig. 8A). The extent of conversion of pNØP to pNØ was directly proportional to the concentration of the recombinant wild-type protein (Fig. 8A, titration curve (q)), which released 17 nmol of pNØ per microgram of protein in 30 min. The specific activity of the T243A protein was Ͻ0.5% of the activity of wild-type Fcp1, whereas Y249A, Y237A, and T174A were 2%, 5, and 9% as active as wild-type Fcp1, respectively (Fig. 8A). Our criterion for a significant mutational effect is one that elicits at least a 10-fold decrement in specific activity compared with wild-type Fcp1. Thus, we conclude that the Thr 174 , Tyr 237 , Tyr 249 , and Thr 243 side chains are important for Fcp1 activity. The specific activities of the other Fcp1-Ala proteins, expressed as the percent of wild-type, were as follows: K163A (85%) R164A (34%), Q173A (48%), K221A (67%), K247A (140%), S270A (160%), and D273A (90%). We surmise that the Lys 163 , Arg 164 , Gln 173 , Lys 221 , Lys 247 , Ser 270 , and Asp 273 side chains do not contribute significantly to catalysis of the generic phosphatase reaction.
The CTD phosphatase activity of the Fcp1-Ala mutants was measured by the release of P i from the 10-aa CTD phosphopeptide SPSYSPTSPS (Fig. 8B). The enzyme preparations were assayed in parallel; the reaction mixtures contained 100 M CTD peptide and 2 g of input Fcp1, an amount sufficient for saturating levels of P i release by wild-type Fcp1 (see Fig. 4A). Conducting the screening assays in this fashion highlighted the most severe mutational effects on CTD phosphatase activity (Fig. 8B). The T174A, Y237A, Y249A, and T243A proteins that were severely defective in CTD dephosphorylation (with extents of P i release in the range of 2-4% of wild-type Fcp1) are the same four mutants that displayed Ͻ10% of wild-type activity in hydrolyzing pNØP. The extents of P i release from the CTD phosphopeptide by the other Fcp1-Ala proteins, expressed as the percent of wild-type value, were as follows: K163A (86%), R164A (42%), Q173A (76%), K221A (90%), K247A (110%), S270A (110%), and D273A (96%). The seven Fcp1-Ala proteins that displayed near wild-type CTD phosphatase activity in the single-point screening assay were the same as those that retained activity in hydrolyzing pNØP.
Structure-function Relationships at Essential Residues of S. pombe Fcp1-To further evaluate the contributions of Thr 174 , Tyr 237 , Thr 243 , and Tyr 249 to the phosphatase reaction, we tested the effects of conservative substitutions. Threonine was replaced by serine and valine; tyrosine was changed to phenylalanine. The recombinant T174S, T174V, Y237F, T243S, T243V, and Y249F proteins were purified from soluble bacterial extracts by nickel-agarose chromatography (Fig. 7B). Hydrolysis of pNØP was measured as a function of enzyme concentration for wild-type Fcp1 and the six conservative mutants (Fig. 9A). Introduction of a serine in lieu of Thr 243 restored activity to 48% of the wild-type level, compared with Ͻ1% activity for T243A. The valine substitution had no salutary effect (2% of wild-type activity), implying that the hydroxyl group at position 243 is essential for catalysis. Replacement of Thr 174 with serine revived activity to 40% of the wild-type level, compared with 9% activity for the T174A mutant, whereas the valine mutant (with 2% of wild-type activity) was even more defective than T174A. Thus, the hydroxyl group of Thr 174 is the functionally relevant substituent. Conservative replacement of Tyr 237 with Phe resulted in virtually complete gain of function, to 95% of wild-type activity, compared with the Y237A mutant, which was 2% as active as wild-type Fcp1. The hydroxyl group of the aromatic residue at position 237 is apparently not important for Fcp1 function. This is consistent with the fact that a Phe side chain is present at the equivalent position of S. cerevisiae Fcp1 (Fig. 1). Conservative replacement of Tyr 249 with Phe elicited a partial gain of function (to 19% of wild-type activity, versus 5% for Y249A), indicating that the hydroxyl moiety of Tyr 249 , although not strictly essential, enhances Fcp1 phospha-tase activity by a factor of 5. The Tyr 249 side chain is strictly conserved in all of the Fcp1 orthologs aligned in Fig. 1.
The hierarchy of conservative mutational effects on CTD phosphatase activity (Fig. 9B) was generally concordant with the effects on generic phosphatase activity. T174S and T243S were much more active than T174V and T243V. The Y237F mutant displayed seemingly full CTD phosphatase activity, whereas Y249F was one-sixth as active as wild-type Fcp1.

DISCUSSION
The potential complexity of the CTD serine phosphorylation array comprises 2 2n different structures, where n is the number of heptad repeats (n ϭ 52 in mammals and 29 in S. pombe). The effector functions of the CTD can be fine-tuned by remodeling the phosphorylation array, a process that entails tipping the balance between the activities of CTD kinases and CTD phosphatases at some or all of the serine phosphorylation sites. Eukaryotic cells are known to contain multiple CTD kinases (4) and recent evidence highlights the existence of multiple CTD phosphatases (5)(6)(7). Among the CTD phosphatases, Fcp1 has been studied most intensively. Early work indicated that the activity of S. cerevisiae and human Fcp1 in dephosphorylating pol II was stimulated by the general transcription factor TFIIF (9,30). Thus, most biochemical studies of S. cerevisiae and human Fcp1 have focused on the TFIIF-dependent phosphatase activity.
Our emphasis has been to understand the intrinsic catalytic properties and specificity of Fcp1, which is a necessary step if one is to ultimately evaluate how positive and negative regulatory factors, covalent modifications of Fcp1 (31,39) or cis-trans isomerization of the CTD prolines (46), impact on Fcp1 function. We have developed S. pombe Fcp1 as a model system in view of the relative ease of obtaining active recombinant protein and the fact that S. pombe Fcp1 displays a comparatively vigorous intrinsic phosphatase activity with either pNØP or synthetic CTD phosphopeptide substrates. We showed previously that recombinant S. pombe Fcp1 was 10-fold more active at its optimal pH of 5.5 in dephosphorylating a 28-mer CTD peptide containing Ser 2 -PO 4 than an otherwise identical 28-mer containing Ser 5 -PO 4 . Here we confirmed that S. pombe Fcp1 also prefers Ser 2 -PO 4 at pH 7.0. The present study substantially advances our knowledge of Fcp1 mechanism by: (i) demonstrating that S. pombe Fcp1 is intrinsically distributive with respect to the phospho-CTD substrate; (ii) defining the minimal CTD Ser 2 -PO 4 substrate; and (iii) underscoring the role of CTD primary structure as a determinant of Fcp1 activity.
The issue of whether Fcp1 is intrinsically processive or distributive is significant in light of a previous report that human Fcp1 appeared to act processively in dephosphorylating 32 Plabeled pol II in the presence of TFIIF (32). S. pombe Fcp1 is clearly distributive with respect to the CTD. Although we cannot rule out a model whereby the fission yeast and human enzyme have fundamentally different mechanisms, we speculate that the apparent processive action of the human enzyme on intact pol II does not arise because the enzyme remains bound to the CTD until the CTD is exhaustively dephosphorylated ("true" processivity), but rather that human Fcp1 is bound elsewhere to the body of pol II (15,30) and that this association allows the polyphosphorylated pol II CTD in cis to reiteratively bind, react, and dissociate at the Fcp1 active site ("apparent" processivity). Were Fcp1 truly processive with respect to the CTD, then one would have to posit the existence of mechanisms to protect the CTD from wholesale dephosphorylation after a single Fcp1 binding event (e.g. via sequestration of some heptads by virtue of binding to other proteins, or via factors that promote Fcp1 dissociation from the CTD).
Knowledge of the atomic structure of the CTD and how such parameters as CTD length, amino acid sequence, and phosphorylation arrays influence CTD-PO 4 effector functions has emerged from recent studies of the interaction of the mRNA capping apparatus with the phosphorylated CTD (29,33,34). A key inference from these studies is that the CTD-PO 4 is conformationally plastic and capable of adopting diverse structures that are templated by the proteins to which the CTD binds. Thus, it is incumbent to delineate for each CTD-inter-acting protein or enzyme the basis for CTD recognition.
The present analysis of the hydrolysis by S. pombe Fcp1 of synthetic CTD phosphopeptides illuminates the contributions of CTD length, heptad phasing, and individual conserved CTD side chains to CTD phosphatase activity. These insights could only have been obtained using defined substrates. The reports that Fcp1 enzymes from species other than S. pombe appear to catalyze the hydrolysis of phosphoserine from the CTD in the context of the pol II elongation complex or isolated pol II, but not in the context of a recombinant CTD fusion protein (12,30), have been cited as casting doubt on the value of results obtained for S. pombe Fcp1 with CTD phosphopeptide substrates (42). We would emphasize here that orthologous enzymes from different species cannot be presumed to have the same inherent biochemical properties, or to engage in the same repertoire of protein-protein interactions that impact on their enzymatic activity. Indeed, this point is underscored by the demonstration that the orthologous triphosphatase and guanylyltransferase components of the fungal mRNA capping apparatus interact differently with each other, and with the pol II CTD, in S. pombe versus S. cerevisiae (33,44,45). Rather than viewing the clearly demonstrated capacity of S. pombe Fcp1 to act on CTD phosphopeptides in vitro as an aberration, we see S. pombe Fcp1 as being liberated from the dependence on protein-protein contacts (e.g. with the "body" of pol II or with TFIIF) for CTD phosphatase activity seen with human Fcp1. Moreover, the ability to dephosphorylate CTD phosphopeptide substrates is not a property unique to S. pombe Fcp1 among the growing collection of protein phosphatases known or suspected to act on the pol II CTD in vivo. Yeo et al. (5) have identified several small CTD phosphatases (SCPs) in human cells, which belong to the DXDX(T/V) enzyme family that includes Fcp1. They showed that recombinant human SCP1 has optimal phosphatase activity at pH 5.0 (thereby resembling S. pombe Fcp1) and that human SCP1 readily dephosphorylates the same pair of synthetic 28-mer CTD Ser 2 -PO 4 and Ser 5 -PO 4 peptides we use here and in our prior studies of CTD-PO 4 effector functions (16,18,29,33). Human SCP1 displayed a modest (2-fold) preference for Ser 5 -PO 4 over Ser 2 -PO 4 (5). Thus, different DX-DX(T/V) phosphatase family members that act on the CTD clearly do have different inherent specificities.
Knowing now that a properly phased single heptad S 5 P 6 S 7 Y 1 S 2 P 3 T 4 suffices for efficient S. pombe Fcp1 phosphatase activity in vitro and that the side chains flanking the phosphoserine are important for Fcp1 activity, we are in a position explain why S. pombe Fcp1 preferentially dephosphorylates Ser 2 -PO 4 versus Ser 5 -PO 4 . Inspection of the analogously phased Ser 5 -PO 4 peptide Y 1 S 2 P 3 T 4 S 5 P 6 S 7 shows that the C-terminal flanking dipeptide is similar in both cases (either Pro-Thr or Pro-Ser), whereas the N-terminal flanking peptides are divergent (Ser-Pro-Ser-Tyr versus Tyr-Ser-Pro-Thr), particularly with respect to the presence of a tyrosine immediately upstream of phosphoserine in the preferred Ser 2 -PO 4 CTD. Loss of the Tyr 1 side chains reduced Fcp1 activity by a factor of 6, an effect that accounts for most of the disparity between the Ser 2 and Ser 5 phosphatase activities of S. pombe Fcp1. The similar reduction in activity elicited by the P3A mutation of the CTD underscores that Fcp1 is a proline-directed protein serine phosphatase.
The present mutational analysis of Fcp1 identified four new side chains important for both generic and CTD-specific phosphatase activities: Thr 174 , Tyr 237 , Thr 243 , and Tyr 249 . Together with the seven essential amino acids identified previously (Asp 170 , Asp 172 , Arg 223 , Asp 258 , Lys 280 , Asp 297 , and Asp 298 ), they likely comprise the phosphatase active site. It had been suggested on the basis of primary structure comparisons and comparative mutational analyses that Fcp1 belongs to a subgroup of DXDX(T/V) acyl-phosphatases that includes bacteriophage T4 polynucleotide 3Ј phosphatase (18,(35)(36)(37)(38). The signature of this subfamily is the essential 297 DD 298 dipeptide located at the distal end of the FCPH domain of Fcp1 (Fig. 1). The ten catalytically essential side chains of the T4 phosphatase include six aspartates, two arginines, one lysine, and one serine (35)(36)(37)(38). 2 The amino acid composition of the T4 phosphatase active site is generally similar to the constellation of essential residues of S. pombe Fcp1, which consists, to date, of five aspartates, one arginine, one lysine, two threonines, and two tyrosines.
Crystallographic snapshots have been obtained for Methanococcus jannaschii phosphoserine phosphatase (PSP), Lactococcus lactis phosphoglucomutase (PGM), and T4 polynucleotide 3Ј phosphatase at various stages of the catalytic cycle (25,27,28,38). These structures suggested plausible catalytic roles for five of the essential side chains of Fcp1 in either nucleophilic attack on Ser-PO 4 (Asp 170 ), general acid-base catalysis (Asp 172 ), transition-state stabilization (Lys 280 ), or metal coordination (Asp 297 and Asp 298 ). Reference to the available crystal structures of PSP and PGM suggests that the newly identified essential Fcp1 residue Thr 243 is the functional equivalent of a conserved serine side chain (Ser 99 in PSP and Ser 114 in PGM) that interacts via O␥ with one of the trigonal phosphate oxygens of the pentacovalent transition state. The equivalent Ser 211 residue in T4 phosphatase is essential for activity. 2 The catalytic serine is located 31 to 47 aa upstream of a conserved lysine that coordinates one of the other trigonal phosphate oxygens of the transition state. Note that Thr 243 of Fcp1 is situated 37 aa upstream of its corresponding essential lysine, Lys 280 . The structure-activity relationships at Fcp1 Thr 243 and T4 phosphatase Ser 211 are concordant (i.e. either threonine or serine at these positions supported the activity of both enzymes). Thus, the requirement for a hydroxyamino acid at Fcp1 residue 243 is consistent with the hydrogen bond to the phosphate seen in the PSP and PGM crystals.
Thr 174 is important for Fcp1 activity and is located within the signature DXDXT phosphatase motif; all Fcp1 orthologs contain threonine at this position (Fig. 1). The corresponding side chain in PSP (Thr 15 ) engages in a hydrogen bond via O␥ to the carboxylate O␦2 of the essential Asp 11 side chain. PSP Asp 11 O␦2 coordinates the essential divalent cation cofactor, whereas O␦1 forms the covalent acylphosphate linkage (25). We posit a similar interaction between Fcp1 Thr 174 and Asp 170 and suggest that this hydrogen bond helps orient the Asp nucleophile for metal coordination and attack on the phosphorylated substrate.
Definitive evaluation of the Fcp1 mechanism and the substrate specificity determinants suggested by our studies of S. pombe Fcp1 will ultimately hinge on crystallization of Fcp1 bound to a defined CTD phosphopeptide. The delineation of the phased heptamer S 5 P 6 S 7 Y 1 S 2 P 3 T 4 as the minimal effective CTD substrate should facilitate cocrystallization experiments.