Purification and Identification of a Novel Subunit of Protein Serine/Threonine Phosphatase 4*

The catalytic subunit of protein serine/threonine phosphatase 4 (PP4C) has greater than 65% amino acid identity to the catalytic subunit of protein phosphatase 2A (PP2AC). Despite this high homology, PP4 does not appear to associate with known PP2A regulatory subunits. As a first step toward characterization of PP4 holoenzymes and identification of putative PP4 regulatory subunits, PP4 was purified from bovine testis soluble extracts. PP4 existed in two complexes of approximately 270–300 and 400–450 kDa as determined by gel filtration chromatography. The smaller PP4 complex was purified by sequential phenyl-Sepharose, Source 15Q, DEAE2, and Superdex 200 gel filtration chromatographies. The final product contained two major proteins: the PP4 catalytic subunit plus a protein that migrated as a doublet of 120–125 kDa on SDS-polyacrylamide gel electrophoresis. The associated protein, termed PP4R1, and PP4C also bound to microcystin-Sepharose. Mass spectrometry analysis of the purified complex revealed two major peaks, at 35 (PP4C) and 105 kDa (PP4R1). Amino acid sequence information of several peptides derived from the 105 kDa protein was utilized to isolate a human cDNA clone. Analysis of the predicted amino acid sequence revealed 13 nonidentical repeats similar to repeats found in the A subunit of PP2A (PP2AA). The PP4R1 cDNA clone engineered with an N-terminal Myc tag was expressed in COS M6 cells and PP4C co-immunoprecipitated with Myc-tagged PP4R1. These data indicate that one form of PP4 is similar to the core complex of PP2A in that it consists of a catalytic subunit and a “PP2AA-like” structural subunit.

PP2A is one of the best studied protein serine/threonine phosphatases. Several reports have demonstrated an important regulatory role for PP2A in a variety of cellular processes. Because PP2A is involved in many cellular events, its activity must be tightly controlled in vivo. Two major mechanisms of regulation of PP2A have been described in the literature. The first involves post-translational modifications of the catalytic subunit of PP2A (PP2A C ). Phosphorylation of a tyrosine residue near the C terminus (9) or phosphorylation of an unidentified threonine residue (10) decreases catalytic activity in vitro. Conversely, carboxymethylation of the C-terminal leucine causes an increase in phosphatase activity (11). The second major mechanism of regulation of PP2A C is through its association with other proteins. PP2A exists as a heterotrimeric complex consisting of a catalytic subunit (C) associated with a structural subunit (A) and a variable subunit (B). The A subunit is thought to act as a scaffold for the B and C subunits (5,12,13). Three families of B subunits have been identified and include B, BЈ, and BЉ (2)(3)(4)(5)(6)12). The B subunit can modulate substrate selectivity in vitro (14,15) and is hypothesized to target PP2A to discrete compartments within the cell (2)(3)(4)(5)(6)13). In addition to the regulation of PP2A C by associated subunits and post-translational modification, the PP2A holoenzyme forms complexes with other cellular proteins, such as Tau (16) and Ca ϩ2 /calmodulin-dependent protein kinase IV (17).
The catalytic subunit of protein phosphatase 4 (PP4 C ) is 65% identical to PP2A C at the amino acid level (18) and has been placed in the type 2A family of phosphatases. PP4 C has been highly conserved between species sharing 91% amino acid identity between human and Drosophila (19). PP4 C is predominantly localized in the nucleus in rat brain and liver but is most highly expressed in testis (20). Additionally, PP4 C was demonstrated to be an essential enzyme in the development of Drosophila embryos (21). The expression of PP4 C was reduced to 25% of the normal protein level in a mutant strain of Drosophila termed centrosomes minus microtubules (cmm) (21). An interesting characteristic of the cmm phenotype is the presence of regions of cells that are unable to complete mitosis because no microtubules exist to connect chromatin and centrosomes. This phenotype implicates PP4 C in the regulation of the nucleation and/or stabilization of microtubules. These data taken together indicate that PP4 has a crucial cellular function, al-though a physiological substrate for PP4 has not yet been identified.
Appropriate regulation of PP4 is likely critical to its normal physiological function. The close homology of PP2A C and PP4 C sequence suggests that they may share certain regulatory properties. Indeed, it was recently shown that PP4 C , like PP2A C , is carboxymethylated (20). Furthermore, by analogy with both PP1 and PP2A, it is unlikely that PP4 exists as free catalytic subunits in the cell (5). However, PP4 C does not associate with the A subunit of PP2A in vitro (18). 2 In order to better understand and characterize the physiologically relevant forms of PP4, we now report the purification of a heterodimeric form of PP4 and identification of the first PP4 Cassociated protein, termed PP4 R1 .

MATERIALS AND METHODS
Materials-Goat anti-rabbit IgG-alkaline phosphatase conjugate, Bio-Scale DEAE2 column, and prestained SDS-polyacrylamide gel electrophoresis (PAGE) molecular mass standards were obtained from Bio-Rad. Phenyl-Sepharose, Source 15Q, Superdex 200, and GammaBind Plus Sepharose resins and gel filtration standards were from Amersham Pharmacia Biotech. Microcystin-Sepharose and microcystin were obtained from Upstate Biotechnology, Inc. (Lake Placid, NY) and Alexis Corp. (San Diego, CA), respectively. Bovine testes were purchased from Pel-Freeze (Rogers, AR) and stored at Ϫ80°C. Centricon-30 concentrators were from Amicon (Beverly, MA). Aprotinin, phenylmethylsulfonyl fluoride, pepstatin A, leupeptin, and SDS-PAGE molecular mass standards were from Sigma. Nylon-reinforced nitrocellulose membrane and Centrex UF-0.5 concentrators were manufactured by Schleicher & Schuell. Polyvinylidene fluoride membrane was obtained from Millipore Corp. (Bedford, MA). The sequencing reagents BigDye and Half-TERM were purchased from Perkin-Elmer and GenPak (Stony Brook, NY), respectively. NdeI was from New England Biolabs (Beverly, MA); EcoRI and XhoI were from Promega (Madison, WI). Plasmid purification kits and pcDNA3.1 were obtained from Qiagen (Valencia, CA) and Invitrogen (Carlsbad, CA), respectively. SulfoLink Gel was purchased from Pierce. The monoclonal anti-Myc antibody was a generous gift from Dr. Lee Limbird (Vanderbilt University).
Chromatography and Purification of PP4 Holoenzymes-All protein chromatographic steps were performed on the FPLC System (Amersham Pharmacia Biotech). The resuspended 50% ammonium sulfate pellets (above) were brought to at least 0.3 M ammonium sulfate by adding the appropriate volume of 4 M ammonium sulfate. The mixture was incubated on ice for 30 min and subsequently centrifuged at 13,000 rpm (Sorvall SS34) for 30 min. One-half of the supernatant was applied to a phenyl-Sepharose column (34 ml) equilibrated in 300 mM ammonium sulfate buffer X, and the column was developed with a linear gradient (184 ml) of 300 -0 mM ammonium sulfate (buffer X without phenylmethylsulfonyl fluoride) at 4 ml/min collecting 8-ml fractions. PP4-containing fractions from two successive phenyl-Sepharose columns were identified by immunoblot analysis, pooled, and diluted 2-fold with buffer X. The sample was loaded at 2 ml/min onto a Source 15Q column (8 ml) equilibrated in buffer X containing 100 mM NaCl; proteins were eluted with a linear gradient of 100 -450 mM NaCl (320 ml) at 4 ml/min collecting 8-ml fractions. Two peaks of PP4 eluted from Source 15Q at 250 and 300 mM NaCl (Fig. 1A). Fractions 19 -21 (250 -300 mM NaCl) containing PP4 pool 1 (see "Results") were combined and further purified over a Bio-Scale DEAE2 column (2 ml, equilibrated in 350 mM NaCl buffer X) using a linear gradient of 350 -1000 mM NaCl (50 ml), collecting 1-ml fractions at 1 ml/min. Finally, the PP4 fractions obtained from DEAE chromatography (600 -700 mM NaCl) were concentrated to 2 ml with a Centricon-30 concentrator and purified to near homogeneity by Superdex 200 gel filtration chromatography. The gel filtration column (120 ml) was equilibrated in buffer X containing 150 mM NaCl and developed at 1 ml/min collecting 2-ml fractions. Protein standards utilized for calibration of the gel filtration column included thyroglobulin (669 kDa), ferritin (440 kDa), catalase (232 kDa), aldolase (158 kDa), bovine serum albumin (67 kDa), ovalbumin (43 kDa), chymotrypsinogen (25 kDa), and ribonuclease A (13.7 kDa). Molecular weights of PP4 complexes were estimated from a graph of the logarithm of molecular weight as a function of elution volume of the standards.
Gel Electrophoresis and Immunoblotting-Samples were boiled for 10 min in Laemmli sample buffer (22) and then subjected to SDS-PAGE (10% gel, 0.75 mm, unless otherwise indicated). After electrophoresis, proteins were either visualized by silver stain (23) or transferred to a nitrocellulose membrane in 10 mM CAPS buffer, pH 11.0, containing 10% methanol. The blots were incubated with the appropriate affinitypurified anti-peptide antibodies (20,24), followed by incubation with goat anti-rabbit IgG-alkaline phosphatase conjugate. Immunocomplexes were visualized with 5-bromo-4-chloro-3-indolyl phosphate and nitro blue tetrazolium as substrates for alkaline phosphatase.
Quantification of PP4 C -The amount of PP4 C in bovine testis extracts was estimated by preparing a standard curve of purified PP4 C (protocol to be published elsewhere) and comparing the immunoblot intensities with samples from bovine testis soluble extracts. To visualize and quantitate PP4 C immunoreactivity, an 125 I-labeled donkey antirabbit IgG (Amersham Pharmacia Biotech) was utilized as the secondary antibody. The proteins corresponding to PP4 C were excised and counted using a ␥-counter. The amount of PP4 C in the purified PP4 holoenzyme was measured by quantifying the intensity of silver-stained proteins using NIH Image. Purified PP2A C was used as the standard, and its concentration was determined by the Bio-Rad protein assay.
Mass Spectrometry Analysis-The purified PP4 holoenzyme was concentrated to approximately 1 pmol/l using a Centrex UF-0.5 concentrator. Glutaraldehyde was added to the sample to achieve a 0.5% (v/v) concentration, and aliquots were removed and analyzed over a time period of 240 min. The reaction was quenched by the addition of the matrix, as described below. The sample was mass-analyzed with a PerSeptive Voyager Elite matrix-assisted laser desorption/ionization mass spectrometer. A nitrogen laser (337 nm) was used to irradiate the sample with sinapinic acid as the matrix. Approximately 0.5 l of sample was mixed with 0.5 l of a saturated solution of sinapinic acid in a 1:1 acetonitrile/water solvent containing 0.5% trifluoracetic acid on a target sample plate. After drying in air for several minutes, the target plate was placed in the mass spectrometer and then analyzed. An accelerating voltage of 25 kV was used to accelerate the ions in the time-of-flight analysis.
Preparation of Phosphorylated Substrates and Phosphatase Assay-Phosphorylated substrates were prepared as described previously for casein (25) and phosphorylase a and histone H1 (26). Purified proteins (approximately 0.14 ng of PP2A C and approximately 11.3 ng of PP4 C contained in PP4) were assayed for phosphatase activity in a 100-l reaction volume containing 25 mM Tris-HCl, pH 7.4, 1 mM dithiothreitol, 1 mM EDTA, 0.5 mg/ml bovine serum albumin, and 32 P-labeled substrate. Following a 10-min incubation at 37°C, the reaction was terminated by the addition of trichloroacetic acid (20% final concentration), and proteins were pelleted by centrifugation at 13,000 ϫ g for 10 min. Supernatants were collected and quantitated for [ 32 P] i , released by scintillation counting. Less than 20% of the total phosphorylated substrate was dephosphorylated during the assays, thus assuring that substrate was not limiting.
Determination of Peptide Amino Acid Sequences-PP4 obtained from the final gel filtration column was concentrated using a Centricon-30 concentrator, separated on a 6% (for analysis of 120 -125-kDa doublet) or 10% (analysis of 29-kDa protein) polyacrylamide gel and transferred to polyvinylidene fluoride membrane. The membrane was briefly stained with Coomassie R-250. Individual proteins were excised and digested in situ with lysyl endopeptidase alone or with lysyl endopeptidase followed by endoproteinase Asp-N. The resulting peptides were separated by high performance liquid chromatography and subjected to automated Edman degradation on a Perkin Elmer/Applied Biosystems Procise 492 protein sequencer. Sequencing was performed by Vanderbilt University Peptide Sequencing and Amino Acid Analysis Shared Resource. 2 S. Kloeker and B. Wadzinski, unpublished observations.
Cloning and Sequencing of PP4 R1 -The cDNA for PP4 R1 was identified by using the peptide sequences (derived from the purified proteins) to search the expressed sequence tag data base with the BLAST protocol. 3 The cDNA clones were obtained from The Institute for Genomic Research/American Type Culture Collection Human cDNA Special Collection (Rockville, MD) and sequenced in their entirety by fluorescent dideoxy terminator-based DNA sequencing (performed on an ABI310 sequencer in the Center for Molecular Neuroscience DNA Sequencing Facility, Vanderbilt University).
PP4 R1 Antibody Generation and Immunoaffinity Purification-A peptide (CASTHPASTRISED) identical to an amino acid sequence found in one of the bovine PP4 R1 peptides was synthesized with an amino-terminal cysteine residue. The peptide was coupled to keyhole limpet hemocyanin using m-maleimidobenzoyl-N-hydroxysuccinimide ester as described previously (27). The peptide-keyhole limpet hemocyanin (250 g) conjugates were injected subcutaneously into rabbits, first in complete Freund's adjuvant followed by several boosts in incomplete adjuvant. The antiserum was characterized by immunoblotting bovine testes soluble extracts in the absence and presence of the peptide. PP4 R1 anti-peptide antibodies (designated ␣PP4 R1 -1) were purified from rabbit antiserum by affinity chromatography (27). Briefly, the synthetic peptide was attached covalently to SulfoLink Gel, and the slurry was transferred to a column. Rabbit antiserum diluted 1:2 with Tris buffer (10 mM Tris-HCl, pH 7.5) was passed over the peptide-Sepharose conjugate three times. The resin was washed with Tris buffer and then with Tris buffer containing 500 mM NaCl. Bound antibodies were eluted by sequential addition of 100 mM glycine, pH 2.5, followed by 100 mM triethylamine, pH 11.5. The eluates were pooled, and the antibodies were concentrated and dialyzed against Tris-buffered saline containing 0.02% NaN 3 using Centriprep-30 concentrators.
A second peptide (CASENIFNRQMVARS) corresponding to amino acid residues 39 -52 in the predicted human sequence was synthesized with an amino-terminal cysteine residue. This peptide was utilized to generate a second rabbit polyclonal antiserum (designated ␣PP4 R1 -2) using similar methodology as described above.
Microcystin-Sepharose Affinity Isolation-Bovine testis soluble extracts (5 mg) were incubated with 30 l of microcystin-Sepharose (50% slurry) while rotating overnight at 4°C. The beads were pelleted and washed six times with 1 ml of buffer X containing 150 mM NaCl, and bound proteins were eluted with Laemmli sample buffer. Applied, unbound, and bound samples were analyzed by immunoblot.
Construction of Myc-tagged PP4 R1 cDNA-A cDNA fragment encoding the NH 2 -terminal Myc-tagged (MEQKLISEEDL) PP4 R1 was produced by successive polymerase chain reactions (PCRs) and restriction enzyme digestion. The first PCR product was made using sense oligonucleotide 1 (5Ј-CTC ATC TCA GAA GAG GAT CTG GCG GAC CTC TCG CTG CT-3Ј), antisense oligonucleotide 2 (5Ј-AAA AAG CAT ATG GTA TTG-3Ј), and human PP4 R1 cDNA as template. Successive PCR using sense oligonucleotide 3 (5Ј-GGC CGA ATT CAG GAT GGA ACA AAA ACT CAT CTC AGA AGA GGA TCT GGC-3Ј), antisense oligonucleotide 2, and the first PCR product as template generated a PCR product that was digested with the restriction enzymes EcoRI and NdeI. The resulting cDNA fragment and the cDNA fragment of PP4 R1 from the NdeI site to the XhoI site (3Ј-end of PP4 R1 ) were ligated into the EcoRI/XhoI cloning site of the mammalian expression vector pcDNA3.1 to produce the NH 2 -terminal epitope-tagged PP4 R1 cDNA under the control of the cytomegalovirus promoter. The construct was verified by restriction enzyme analysis and DNA sequencing.
Cell Transfection-COS M6 cells maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum were transiently transfected (10-cm tissue culture dishes) using the DEAE-dextran procedure as described previously (28,29). Transfected COS cells (one 10-cm dish) were dislodged by scraping in 1 ml of lysis buffer containing 25 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1 mM EDTA, 1 mM EGTA, 1 mM dithiothreitol, 0.5% Triton X-100, 6.3 g/ml aprotinin, 4 g/ml leupeptin, 10 mM benzamidine, 4 g/ml pepstatin A, and 1 mM phenylmethylsulfonyl fluoride. Cells were sonicated with two 10-s bursts followed by centrifugation for 15 min. The protein concentration of clarified supernatants was estimated by the BioRad protein assay using bovine serum albumin as the standard.
Immunoprecipitation of Myc-PP4 R1 and PP4 C -Approximately 1 mg of protein from cell lysates (in 1 ml of lysis buffer) was incubated with 40 l of anti-Myc ascites for 6 h while rotating at 4°C. Immune complexes were isolated by adding 40 l of a 50% slurry of GammaBind Plus Sepharose. After washing the beads six times with lysis buffer, bound proteins were eluted with 20 l of Laemmli sample buffer. The proteins were analyzed by immunoblot analysis with affinity-purified antibodies directed against the C terminus of PP4 C .
FIG. 1. PP4 exists in two complexes in bovine testis soluble extracts. A, Source Q profile of PP4 holoenzymes. Bovine testis soluble extracts were fractionated by phenyl-Sepharose chromatography, followed by Source 15Q chromatography. Select fractions of the Source 15Q column were analyzed by immunoblot using a mixture of antibodies specific for PP4 C and PP2A C . B, elution of pool 1 and pool 2 on gel filtration. Fractions 18 -23 and fractions 24 -27 from Source 15Q were pooled and individually subjected to gel filtration chromatography followed by immunoblot analysis with PP2A C -and PP4 C -specific antibodies. This experiment is representative of at least three independent fractionations. The elution of gel filtration standards is indicated by the arrows.

Observation of Multiple PP4 Complexes in Bovine Testis
Soluble Extracts-During the initial stages of purification of PP4 holoenzymes from various tissues, a broad elution profile of PP4 immunoreactivity was observed on several chromatography columns including anion exchange and gel filtration (data not shown). One possible explanation of this behavior is that PP4, like PP2A, exists in multiple oligomeric complexes. To address this possibility, the elution of distinct partially purified PP4 species from gel filtration chromatography were compared. Because PP4 C protein is expressed in greatest quantity in the testis (20), this tissue was utilized for subsequent biochemical manipulations and holoenzyme purification. Bovine testis soluble proteins recovered from 25-50% ammonium sulfate precipitation were separated by sequential hydrophobic interaction and anion exchange chromatography. Fractions containing PP4 were identified by immunoblot analysis using a PP4 C antibody previously described (20). PP4 eluted over several fractions with two noticeable peaks from Source 15Q (fractions 20 and 25; Fig. 1A). Pools of each PP4 peak were individually applied to a gel filtration column (Fig. 1B). Immunoblot analysis of select gel filtration fractions indicated that PP4 in pool 1 exhibited an apparent molecular mass of approximately 270 -300 kDa (fractions 31-34), whereas PP4 in pool 2 was slightly larger, approximately 400 -450 kDa (fractions 29 -32). These data suggest that PP4 C exists in at least two distinct complexes in bovine testis. Immunoblot analysis of the gel filtration fractions using affinity-purified antibodies to the A and B␣ subunits of PP2A revealed that the PP2A subunits did not cofractionate with PP4 C (data not shown).
Purification of One PP4 Holoenzyme-As a first step toward identifying PP4-associated subunits, the purification of a PP4 holoenzyme from bovine testis was undertaken. PP4 (270 -300 kDa) was purified by sequential chromatography on hydrophobic, strong anion exchange, weak anion exchange, and gel filtration columns. Fig. 2 shows fractions from the gel filtration column as analyzed by immunoblot and silver stain. Silver stain analysis revealed a protein doublet (120 -125 kDa) cofractionating with the catalytic subunit of PP4. The molecular mass of the co-purifying protein doublet was carefully determined by comparing its migration with the migration of SDS-PAGE standards (data not shown). Taking into account the differences in molecular masses of these two proteins (see mass spectrometry analysis below), one might predict a 3:1 protein staining ratio if the complex existed in a 1:1 molar stoichiometry. The protein silver staining ratio of the 120 -125-kDa co-purifying proteins to PP4 C ranged from 1.4 to 2.7 as quantitated by NIH Image, suggesting a near stoichiometric complex. Extracts from 200 g of bovine testis (1200 mg of total protein) contained approximately 400 g of PP4 C , as determined by quantitative immunoblot analysis (see "Materials and Methods"). The final PP4 holoenzyme contained 7 g of PP4 C (based on the intensity of the silver-stained protein; see "Materials and Methods"), indicating an overall yield of approximately 2%. PP2A C was not detected in the silver-stained gels FIG. 2. Immunoblot (top) and silver stain analysis (bottom) of the purified PP4 holoenzyme. PP4 was purified from bovine testis soluble extracts by sequential FPLC on phenyl-Sepharose, Source 15Q, and DEAE. Pooled PP4 fractions from DEAE Sephacel were concentrated to 2 ml and further purified by gel filtration chromatography. The gel filtration fractions were analyzed by immunoblot with both PP2A C and PP4 C antibodies. The data shown are representative of five separate purifications. The elution of gel filtration standards is indicated by the arrows.

FIG. 3. Mass spectrometry analysis of purified PP4 holoenzyme.
Purified PP4 holoenzyme was concentrated to approximately 1 pmol/l, cross-linked with glutaraldehyde for 1 min, and analyzed by mass spectrometry. The peaks at 35143.8, 104499, and 140692 presumably correspond to PP4 C , PP4 R1 , and the PP4 complex (i.e. PP4 R1 -PP4 C dimer), respectively. of the purified PP4 preparation, but very low levels of PP2A C immunoreactivity were observed. The purified PP4 holoenzyme was further characterized by FPLC on Mono Q and phenyl-Superose columns. Silver stain analysis of these column fractions confirmed that the 120 -125-kDa protein doublet also co-purified with PP4 C on these analytical resins (data not shown). Moreover, when partially purified PP4 holoenzyme was incubated with microcystin-Sepharose (an affinity resin frequently used to identify PP1-and PP2A-associated proteins Refs. 30 -32), both PP4 C and the 120 -125-kDa protein doublet bound to microcystin (data not shown; see below). Together, these data provide compelling evidence that a PP4 C -associated protein, termed PP4 R1 , had been isolated.
To obtain a more accurate determination of the size of the individual components of the PP4 complex, the sample was analyzed by mass spectrometry. This technique has recently been introduced as an excellent method to dissect the size of multiprotein complexes (33). The mass spectrum of crosslinked purified PP4 holoenzyme revealed two major peaks and several minor peaks (Fig. 3). The two major peaks are proteins of 35.1 and 104.5 kDa, presumably corresponding to PP4 C and PP4 R1 . The molecular mass of PP4 R1 measured by mass spectrometry (104,499 Da) was somewhat different from that observed by SDS-PAGE (120 -125 kDa). In contrast to the sharp PP4 C peak, the peak corresponding to PP4 R1 is quite broad. This broad PP4 R1 peak may be due to the presence of multiple species of similar molecular weight as detected by SDS-PAGE. Since the size of the 140-kDa peak is virtually identical to the sum of the molecular weights of PP4 C and PP4 R1 , it probably represents a heterodimeric complex containing stoichiometric amounts of the two proteins cross-linked with glutaraldehyde.
Phosphatase Activity of PP4 -Purified PP4 was assayed for phosphatase activity using casein, protein kinase A-phosphorylated histone H1, protein kinase C-phosphorylated histone H1, and phosphorylase a as substrates. The specific activity of purified PP4 for these phosphoproteins ranged from 0.624 to 13.1 pmol/min/g of PP4 C contained in the purified PP4 holoenzyme, whereas the specific activity of purified PP2A C toward identical substrates ranged from 67 to 1197 pmol/min/g. An immobilized phosphatase assay similar to the one described by Ding et al. (34) also was developed to test PP4 phosphatase activity toward proteins phosphorylated in vitro (bovine testis soluble extracts) or in vivo (HEK cells). Okadaic acid-sensitive protein dephosphorylation could be detected in this assay with purified PP2A C but not with purified PP4 (data not shown).
Identification of the PP4 R1 cDNA-Multiple peptides from both proteins of the putative bovine PP4 R1 were subjected to amino acid sequence analysis (Fig. 4). A total of nine peptides were sequenced; peptides 1 and 4, 3 and 9, and 5 and 6 had nearly identical sequences. The peptide sequences were used to search the expressed sequence tag data base to identify potential cDNA clones. Several human cDNA clones with translated sequences highly homologous to the sequenced peptides (Fig. 4) were identified and obtained from The Institute for Genomic Research/American Type Culture Collection Human cDNA Special Collection. One clone was of particular interest because it contained a Kozak consensus start sequence (35), an open reading frame of 933 amino acids with a predicted molecular weight of 105,189, and a poly(A) tail. Both the cDNA and predicted amino acid sequences of this clone are shown in Fig.  5A. The predicted molecular weight of the encoded protein is similar to the molecular weight determined by mass spectrometry of the purified PP4 R1 protein (104,500). Thus, these data indicate that the isolated cDNA clone contains most, and perhaps all, of the PP4 R1 coding sequence. Furthermore, all of the bovine peptides could be identified in one human cDNA sequence, and two peptides were shared by both proteins of the doublet, suggesting that the proteins of the doublet are very similar. Interestingly, PP4 R1 has 13 repeats similar to the nonidentical or "heat" repeats found in the A subunit of PP2A (PP2A A ) (36,37). The repeats identified in PP4 R1 are indicated by white type on black in Fig. 5A and are also listed in Fig. 5B. Amino acid residues in boldface type in Fig. 5B are residues identified in the consensus repeat sequence in PP2A A (38). Seven of the PP4 R1 repeats (indicated by an asterisk) were matched to the PP2 A repeats by employing the ProfileScan server. 4 The remaining repeats were found by manual examination of the amino acid sequence. PP4 R1 is distinct from PP2A A in that these motifs are not contiguous; instead, the repeats are separated by a novel region of approximately 332 amino acids between the sixth and seventh repeats. To date, this divergent region does not have homology to any known protein.
Characterization of an Antibody Generated toward PP4 R1 (␣PP4 R1 -1)-Polyclonal anti-peptide antibodies (designated ␣PP4 R1 -1) directed against a bovine sequence homologous to residues 912-924 of the expected human protein were produced FIG. 4. Amino acid sequence of peptides isolated from the purified PP4associated protein. The first peptide of each pair was obtained from the purified bovine protein, and the second peptide corresponds to the homologous region identified in the human sequence. The last column indicates whether the peptide was derived from the upper or lower protein of the PP4 C -associated doublet.
in rabbits. The antiserum recognized a doublet of identical molecular weight in both bovine testis soluble extracts and in the purified PP4 holoenzyme (Fig. 6A). Antibody binding could be blocked in the presence of excess peptide. The anti-peptide antibodies were affinity-purified from the rabbit antiserum and used to monitor the purification of the PP4 holoenzyme. Both PP4 R1 and PP4 C were significantly enriched throughout the purification protocol as shown in Fig. 6B.
Co-isolation of PP4 R1 and PP4 C on Microcystin-Sepharose-The availability of PP4 R1 -specific antibodies (␣PP4 R1 -1) al- FIG. 5. Sequence of PP4 R1 (A) and identification of "heat" repeats (B). A, the peptides obtained from purified bovine PP4 R1 are underlined. Regions homologous to the repeats found in the A subunit of PP2A are printed in white type on black. B, amino acid sequence alignment of the PP4 R1 repeats. Amino acid residues in boldface type are loosely conserved amino acids previously defined in the repeats of PP2A A . The repeats with asterisks were identified by ProfileScan as having homology to the PP2A A "heat" repeats. lowed examination of PP4 binding to microcystin-Sepharose. This affinity resin previously has been utilized to isolate PP1 and PP2A complexes from crude tissue extracts (30 -32). Soluble proteins from bovine testis were incubated with the microcystin-Sepharose in the absence or presence of excess unconjugated microcystin, and bound proteins were eluted by Laemmli sample buffer. Immunoblot analysis of the applied, flowthrough, and bound samples revealed that PP4 C and PP4 R1 co-purified on microcystin-Sepharose (Fig. 7). The interaction of both proteins with the resin was blocked by preincubation of bovine testis soluble extracts with 5 M unconjugated microcystin.
Expression and Immunoprecipitation of PP4 R1 -To aid in the detection and isolation of the recombinant protein, the PP4 R1 cDNA was engineered with a Myc tag on the amino terminus. In Fig. 8A, cell lysates prepared from vector and Myc-PP4 R1transfected COS M6 cells were subjected to immunoblot analysis with the anti-Myc antibody. A protein migrating slightly faster than the purified bovine PP4 R1 was observed only in the lysates from cells transfected with Myc-PP4 R1 cDNA. To determine if recombinant Myc-PP4 R1 protein could associate with endogenous PP4 C , Myc-PP4 R1 was immunoprecipitated from cell lysates with the anti-Myc antibody, and the immune complexes were analyzed for PP4 C and PP4 R1 by immunoblotting. PP4 C co-immunoprecipitated with Myc-PP4 R1 in lysates prepared from Myc-PP4 R1 -transfected cells, but not in lysates from vector-transfected cells (Fig. 8B). These data confirm that the PP4 R1 cDNA clone encodes a protein that can associate with PP4 C . Moreover, PP4 R1 appears to be a specific PP4 C -associated protein because PP2A C was not detected in the Myc-PP4 C immune complex (data not shown).

DISCUSSION
The high homology of PP2A C and PP4 C led to the hypothesis that PP4, like PP2A, may associate with regulatory subunits to form multimeric complexes. To identify any proteins associated with the catalytic subunit of PP4, one oligomeric form of this phosphatase was purified. The purified complex consisted of PP4 C and a PP4 C -associated protein (PP4 R1 ). Sequence information obtained from PP4 R1 peptides was utilized to identify a cDNA clone. The amino acid sequence encoded by this cDNA has a predicted molecular weight of 105,189, which is very similar to the size of PP4 R1 measured by mass spectroscopy (104,500). Furthermore, the cDNA contains the Kozak consensus start sequence (35), indicating that the most 5Ј methionine is an ideal place for the initiation of translation. These data suggest that the cDNA clone may indeed represent a fulllength PP4 R1 cDNA. Although the Myc-tagged PP4 R1 migrated slightly faster on SDS-PAGE than the purified protein, we would like to point out that the expressed protein was of human origin, whereas the purified protein was obtained from bovine tissue. Thus, the differences in sizes could be species variations of PP4 R1 . Alternatively, the difference in migration on SDS-PAGE of the purified and expressed proteins could be attributed to proteolysis of the expressed PP4 R1 protein. The PP4 R1 cDNA does, however, encode a functional PP4 C binding protein, because when expressed in mammalian cells it can be immunoprecipitated with the catalytic subunit of PP4.
The purified PP4 preparation exhibited very little phosphatase activity toward substrates readily dephosphorylated by PP2A C . The activity that was measured could be accounted for with as little as 1% PP2A C contamination, and it is important to point out that a hint of PP2A C immunoreactivity was observed in the purified PP4 holoenzyme. Two reasonable explanations for the low apparent PP4 activity include inhibition of PP4 C catalysis by PP4 R1 or a narrow PP4 substrate profile (i.e. the ideal or physiological substrate has not yet been tested). Additionally, we cannot rule out the possibility that PP4 was inactivated during the purification procedure. However, this explanation seems unlikely, because PP2A catalytic activity was not inactivated following similar purification conditions (data not shown), and partially purified PP4 bound to microcystin-Sepharose, suggesting that the catalytic site is intact.
PP4 R1 migrated as a doublet of 120 -125 kDa on SDS-PAGE and exhibited a fairly broad peak (ϳ105 kDa) as determined by mass spectrometry. The apparent size discrepancy may be due to aberrant migration of PP4 R1 on SDS-PAGE. Several possibilities could account for the protein doublet. First, the smaller protein may represent a proteolytic fragment of the larger protein giving rise to the appearance of two proteins on SDS-PAGE and a fairly broad peak on mass spectrometry. We are FIG. 6. Characterization of PP4 R1 antibodies (␣PP4 R1 -1) (A) and assessment of PP4 purification (B). A, bovine testis soluble extracts and purified PP4 protein were analyzed by immunoblotting with a polyclonal antibody directed toward PP4 R1 (in the absence and presence of the synthetic peptide). B, pools of the peak PP4 fractions from each step of the purification were analysed by silver stain (right panel) or subjected to immunoblot analysis with specific PP4 C and PP4 R1 antibodies (left panel) (5 g of protein at each step, except 0.5 g in the final gel filtration step).
aware of some proteolysis, because another minor co-purifying protein (29 kDa) was identified as a fragment of PP4 C by amino acid sequence analysis (data not shown). A second explanation is that the two proteins are identical except that one is posttranscriptionally or post-translationally modified. The mass spectrometry data is consistent with this possibility; however, at present no direct evidence exists to support this interpretation. Finally, the two proteins may represent highly homologous isoforms of PP4 R1 that differ slightly in molecular weight.
PP2A A and PP4 R1 both contain nonidentical or "heat" repeats. Each of the 15 PP2A A repeats is predicted to consist of two tightly packed amphipathic helices (36). A linear organization of these repeats would form a rodlike protein with an axial ratio of 10.5:1 (39,40). PP2A C has been shown to bind to the C-terminal repeats of the A subunit, whereas the simian virus 40 and polyomavirus T antigens, and perhaps the B subunit of PP2A, bind to the N terminus region (36). By analogy to PP2A, it is quite likely that PP4 C binds to the C terminus of PP4 R1 , since this is where PP2A A and PP4 R1 share the greatest homology. PP4 R1 is distinct from PP2A A in that the nonidentical repeats are not contiguous in PP4 R1 . It is difficult to interpret the implications of this unique PP4 R1 feature both structurally and functionally. Since we did not identify any PP2A "B-like" subunits in our purified preparation, it is conceivable that the divergent region separating the repeats may function similarly to the B subunit in dictating substrate specificity and subcellular localization. However, we cannot rule out the possibility that a B-like subunit of PP4 was lost during the purification. Finally, the larger PP4 complex (440 kDa), which was not extensively characterized, may contain distinct PP4 subunits because the PP4 R1 to PP4 C ratio of immunoreactivity in this complex was substantially lower (data not shown) than the ratio in the purified 270 -300-kDa complex.
The data presented in this report suggest that the purified PP4 complex is composed of one PP4 catalytic subunit and one associated protein; however, it is possible that two dimers may combine to form a tetramer. A complex corresponding to the predicted size of a tetramer was not detected by mass spectrometry but would be consistent with the gel filtration data. Since the anomalous elution of PP2A A on gel filtration may be explained by the predicted rodlike structure (39,40), the size discrepancy of purified PP4 estimated by gel filtration and mass spectrometry could be reconciled by similar factors.
Several studies had indicated that PP4 existed in a high molecular weight complex with proteins distinct from the reg-ulatory subunits of PP2A (18). 2 The biochemical purification of one PP4 holoenzyme has provided direct evidence to support this hypothesis and has resulted in identification of the first PP4 C -associated protein (PP4 R1 ). Interestingly, the predicted amino acid sequence of PP4 R1 has homology to the A subunit of PP2A and suggests similar functions. The PP4 R1 cDNA clone and encoded protein have led to the development of polyclonal antibodies that can be employed to examine tissue distribution and subcellular localization of the endogenous protein. Additionally, the purification and characterization of other PP4 complexes may reveal more similarities with and/or differences from PP2A complexes. These studies will contribute to the overall understanding of the regulation of the type 2A-like phosphatase family. cDNA. Cell lysates were prepared and analyzed by immunoblot with an antibody directed toward the Myc epitope. The third lane contains purified PP4 holoenzyme; PP4 R1 in the purified preparation was visualized with affinity-purified PP4 R1 antibodies (␣PP4 R1 -1). B, ␣-Myc immunoprecipitation. Anti-Myc immune complexes from cellular lysates obtained in A were analyzed by immunoblot with PP4 R1 (␣PP4 R1 -2) antiserum and affinity-purified PP4 C antibodies. The top half of the immunoblot was probed with ␣PP4 R1 -2 antiserum because the affinity-purified ␣PP4 R1 -1 antibodies fail to recognize the human protein. The data are representative of three experiments.