NH2-terminal Modification of the Phosphatase 2A Catalytic Subunit Allows Functional Expression in Mammalian Cells*

Functional expression of recombinant wild-type phosphatase 2A catalytic subunit has been unsuccessful in the past. A nine-amino-acid peptide sequence (YP-YDVPDYA) derived from the influenza hemagglutinin protein was used to modify the NH2 and/or COOH terminus of the phosphatase 2A catalytic subunit. Addition of the nine-amino-acid sequence at the NH2 terminus allowed recombinant phosphatase 2A expression as a predominantly cytosolic phosphatase 2A enzyme. The 12CA5 monoclonal antibody that recognizes the nine-amino-acid hemagglutinin peptide sequence was used to immunoprecipitate the epitope-tagged phosphatase 2A catalytic subunit. Assay of the immunoprecipitated epitope-tagged phosphatase 2A demonstrated an okadaic acid-sensitive dephosphorylation of [32P] histone H1 and [32P]myelin basic protein similar to that measured with the wild-type enzyme. Functional phosphatase activity could be demonstrated for the NH2-terminal modified phosphatase 2A catalytic subunit following transient expression in COS cells or stable expression in Rat1a cells. In contrast, the COOH-terminal-modified phosphatase 2A catalytic subunit was very poorly expressed. The NH2-, COOH-modified subunit, having the nine-amino-acid hemagglutinin peptide sequence encoded at both termini of the polypeptide, was also expressed as a functional phosphatase 2A enzyme. Thus, NH2-terminal modification of the phosphatase 2A catalytic subunit results in a functional plasmid-expressed enzyme. The unique nine-amino-acid epitope-tag sequence also provides a method to easily resolve the recombinant phosphatase 2A from the endogenous wild-type gene product and related phosphatases expressed in cells.

network of Ser/Thr protein kinases. In fact, the phosphorylation of Ser/Thr residues of specific proteins is by far the major response observed following hormone/growth factor receptor-stimulated tyrosine kinase activity (Hunter, 1987). The involvement of specific protein Ser/Thr phosphatases in reversal of phosphorylation events is poorly defined, but it is becoming clear that dephosphorylation is a regulated event in hormone/growth factor action (Lee et al., 1991;Boyle et al., 1991).
The involvement of different Ser/Thr phosphatase gene products and isoforms in the control of cell phenotype is poorly understood, but progress is being made in defining proteins associated with, and regulated by, specific phosphatases. An example of the apparent importance of protein phosphastases in growth control is indicated by the selective binding of the phosphatase 2A enzyme by DNA tumor virus T antigens Pallas et al., 1990;Scheidtmann et al., 1991;Virshup et al., 1989;Walter et al., 1990;Yang et al., 1991). The growth control machinery of the cell is basically overtaken by T antigens to allow uncontrolled cell growth and virus replication. Both SV40 and polyoma T antigens bind the phosphatase 2A enzyme . In fact, the only known function in terms of stable protein interactions for SV40 and polyoma small T antigen is the binding of the phosphatase 2A enzyme (Pallas et al., 1990;Walter et al., 1990). One functional role of the small T antigen shown in vitro is the inhibition of SV40 large T antigen dephosphorylation by phosphatase 2A (Scheidtmann et al., 1991). Additionally, small T antigen was also shown to inhibit the phosphatase 2A-catalyzed dephosphorylation of the p53 growth suppressor gene (Scheidtmann et al., 1991).
The study of Ser/Thr phosphatases in mammalian cells using genetic strategies for expression of recombinant proteins has been frustrating. This frustration has been due to problems inherent to expression of Ser/Thr phosphatases using standard gene transfer techniques. First, attempts to express functional phosphatase catalytic subunits in mammalian cells using gene transfer techniques have met with little success (Green et al., 1987). The expressed phosphatase polypeptides have been denatured and inactive. Second, the catalytic subunits of Ser/Thr phosphatases are highly conserved (Cohen, 1989(Cohen, , 1990Wadzinski et al., 1990), and antibodies specific for different phosphatases and their isoforms have been lacking.
To overcome the problem of expressing functionally active phosphatase catalytic subunits, we hypothesized that the catalytic subunit polypeptide primary sequence would have to be modified. For this purpose, we began to manipulate the NH,and COOH-terminal regions of the phosphatase 2A catalytic subunit using mutagenesis strategies of the cDNA encoding the human phosphatase 2A gene product (Arino et al., 1988).
In this report, we demonstrate that addition of a nine-aminoacid peptide sequence at the NH, terminus of the phosphatase 2A catalytic subunit is sufficient to stabilize the translation product in a functional form. An antibody recognizing the nine-amino-acid "epitope-tag" sequence can be used to resolve the plasmid-expressed gene product from endogenous phosphatases, including the wild-type phosphatase 2A enzyme. The epitope-tagged phosphatase 2A enzyme can be functionally expressed both transiently and stably in mammalian cells. Thus, for the first time, genetic strategies seem feasible to alter the expression and to study the functional properties of specific Ser/Thr phosphatases in mammalian cells.

EXPERIMENTAL PROCEDURES
Materials-Restriction enzymes were obtained from New England BioLabs or United States Biochemical Corp. Thermus aquaticus DNA polymerase (AmpliTaq) was obtained from Perkin-Elmer Cetus. 01igonucleotide primers were synthesized on an Applied Biosystems (Foster City, CA) DNA synthesizer. Okadaic acid was purchased from Moana BioProducts, Inc. (Honolulu, HA). The IgG fraction rabbit anti-mouse IgG (whole molecule) was obtained from Cappel (Malvern, PA) and the 12CA5 monoclonal antibody was obtained from Berkeley Antibody Co., Inc. (Richmond, CA). Protein A-Sepharose and protein G-Sepharose were obtained from Pharmacia (Uppsala, Sweden). ['TI Protein A was from Du Pont-New England Nuclear. Protein kinase C was from Calbiochem (La Jolla, CA). The epitope-tag peptide, YPYDVPDYA, was synthesized using an automated Applied Biosystems solid-phase peptide synthesizer and purified on a preparative C-18 reverse-phase high performance liquid chromatography column.
Construction of Epitope-tagged PP2A cDNA-Oligonucleotides encoding the epitope sequence YPYDVPDYA were inserted in frame with the human liver phosphatase 2A cDNA (Arino et al., 1988) using the polymerase chain reaction (PCR).' A cDNA fragment encoding the NH2-terminal-taggedphosphatase 2A was generated by successive PCR and restriction digestion. The first PCR product was made using the sense oligonucleotide 1,5'-ATG TAT CCA TAT GAT GTT CCA GAT TAT GCT GAC GAG AAG GTG TTC AC-3', the antisense oligonucleotide 2, 5'-ATG TAG ACA GAA GAT CTG C-3', and the phosphatase 2A cDNA as template. Successive PCR using sense oligonucleotide 3, 5'-TTG CGG CCG CTA AGC TTG GCA CGA TGT ATC CAT ATG AT-3', antisense oligonucleotide 2, and the first PCR product as template generated a PCR product which was digested with the restriction enzymes HindIII and BstXI. This resulting cDNA fragment and the cDNA fragment of PP2A from the BstXI site to the HindIII site (3' terminus of phosphatase 2A) were ligated into the HindIII cloning site of the mammalian expression vector pCW1-NE0 (Woon et al., 1988) to produce the NHz-terminal epitope-tagged PPZA (Nqi-PP2A) cDNA under the control of the SV40 early promoter. A similar PCR strategy was used to produce the COOH-terminal-tagged and double-tagged phosphatase 2A (PP2A-CP' and NBPi-PP2A-CP', respectively). The first PCR product for COOH-terminal tagging of phosphatase 2A was produced using a degenerate sense oligonucleotide, 5'-GGGAATTCGA(CT)(CAT)-

T(GC)(CT)T(AGC)TGGTC(ATG)G-3', the antisense oligonucleo-
tide 4, 5'-AGC ATA ATC TGG AAC ATC ATA TGG ATA CAG GAA GTA GTC TGG GGT-3', and phosphatase 2A as template. Second step PCR was done using the same degenerate oligonucleotide, the antisense oligonucleotide 5, 5'-GGC TTA AGC TTT AAA ATT TCA TTA AGC ATA ATC TGG AAC ATC ATA-3', and the first PCR product as template. The resulting PCR product was digested with XhoI and HindIII, and the purified fragment was then ligated with the HindIII-XhoI fragment of PP2A or WP'-PP2A into the HindIII cloning site of pCW1-NEO. Orientation and proper construction of all of the constructs was verified by restriction enzyme analysis and DNA sequencing.
Cell Transfections-The epitope-tagged phosphatase 2A cDNAs were transiently expressed in COS-1 cells using the DEAE-dextran transfection procedure (Ausubel et al., 1987;Osawa et al., 1990) and stably introduced in Ratla cells by electroporation using 10 pg of ' The abbreviations used are: PCR, polymerase chain reaction; FPLC, fast protein liquid chromatography; HEPES, 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid; TPA, 12-0-tetradecanoylphorbol-13-acetate. cDNA (Bio-Rad gene pulser, 25 microfarads, 0.5 kV, Heasley et al., 1991). Stable transfectants were selected for resistance to G418 and positive clones were identified by Western immunoblotting. Both cell lines were grown in Dulbecco's modified Eagle's medium supplemented with 5% bovine calf serum, 5% newborn calf serum, 100 units of penicillin/ml, and 100 pg of streptomycin/ml. Preparation of Cellular Extracts and Subcellular Fractionutwn-Clonal Ratla cells or COS-1 cells (68 h post-transfection) were dislodged from the dish using trypsin, pelleted by centrifugation, and washed with phosphate-buffered saline. A cytosolic extract was obtained by repetitively passing cells in a low salt buffer (50 mM Tris, pH 7.4, 2 mM MgC12, 1 mM EDTA, 0.02 units of aprotinin/ml, and 10 pg of leupeptin/ml) through a 25-gauge needle, followed by centrifugation for 15 min at 10,000 X g to remove nuclei and cell debris. This extract was used for immunoprecipitations and immunoblotting. For subcellular fractionation, the broken cells were centrifuged 5 min at 2,000 X g, and the crude nuclear pellet was washed once with the low salt buffer. The supernatants from the low speed centrifugation were collected and centrifuged at 100,000 X g for 30 min. The supernatants (cytosolic extracts, C) were precipitated with trichloroacetic acid, washed with acetone, dried, and resuspended in Laemmli sample buffer. The nuclear (N) and 100,000 X g pellets (P) were also solubilized in Laemmli sample buffer.

Mono Q FPLC Fractionation-COS-1 cells (68 h post-transfection)
were collected by scraping two 100-cm dishes in 0.4 ml of ice-cold buffer consisting of 25 mM Tris, pH 7.2, 1 mM EDTA, 1 mM dithiotheitol, 0.5% Triton X-100, 0.02 units of aprotinin/ml, and 10 pg of leupeptin/ml. The cells were vortexed briefly and centrifuged 10 min at 10,000 X g to remove nuclei and cell debris. The extracts (500 pl) were applied to a Mono Q FPLC column equilibrated in buffer A (25 mM Tris, 1 mM EDTA, 1 mM dithiothreitol, pH 7.2). The column was eluted over a 19-min time period at a flow rate of 1 ml/min with a linear gradient of 0-500 mM NaCl in buffer A. Fractions were trichloroacetic acid precipitated and assayed by immunoblotting.
Immunoblotting-Proteins (100-300 pg) were solubilized in Laemmli sample buffer, boiled 10 min, centrifuged briefly in a microcentrifuge, and analyzed by SDS-PAGE. Proteins were transferred to nitrocellulose membranes and immunoblotted with 12CA5 monoclonal antibody or PTC-1 monoclonal antibody (Mumby et al., 1985). The blots were then incubated with rabbit anti-mouse polyclonal antibody followed by ['251]protein A and autoradiography.
Immunoprecipitations-To 100-300 pl of cytosolic extract was added Nonidet P-40 lysis buffer (50 mM Tris, pH 7.4, 150 mM NaCl, 2 mM MgClZ, 1 mM EDTA, 0.5% Nonidet P-40, 0.02 units of aprotinin/ml, and 10 pg of leupeptin/ml, final concentration) and 1-2 pl of 12CA5 ascites (approximately 5 pg of monoclonal antibody). The mixture was rotated end-over-end overnight at 4 "C followed by the addition of protein G-Sepharose (30 pl) or protein A-Sepharose (30 pl) for 1-2 h at 4 "C. The beads were washed two times with Nonidet P-40 lysis buffer and three times with low salt buffer. The epitopetagged phosphatase 2A protein was eluted from the antibody-bead complex with 25 p~ epitope-tag peptide (YPYDVPDYA) in low salt buffer, centrifuged 1 min at 1,000 X g in a microcentrifuge to pellet the beads, and the supernatant assayed for activity.
Assay of Protein Ser/Thr Phosphatase Activity-For assay of immunoprecipitated phosphatase activity, the protein fractions eluted from the antibody complex using the YPYDVPDYA peptide were placed in low salt buffer (100-pl final volume) containing 10 pg of bovine serum albumin and [32P]histone H 1 (2.5 X 10' cpmlpg) or [32P]myelin basic protein (1.4 X lo' cpmlpg). Samples were incubated at 30 "C for varying times and then protein precipitated on ice with 20% trichloroacetic acid. For okadaic acid inhibition studies, 1 pl of 100 X okadaic acid concentrations in Me2S0 was added to the phosphatase reaction mixture. Reactions were incubated for 30 min a t 30 "C before precipitation with trichloroacetic acid. The precipitated protein was pelleted by centrifugation a t 10,000 X g for 15 min. The supernatants were collected and quantitated for ['*Pi] released by scintillation counting. Fractions from Mono Q FPLC were assayed in the same buffer conditions in the absence of the YPYDVPDYA peptide.

RESULTS AND DISCUSSION
The difficulty encountered in expressing cloned Ser/Thr phosphatase 2A and related phosphatases has limited progress in analyzing the cellular functions of this important family of regulatory proteins. In a past report, recombinant phosphatase 2A expressed in mammalian cells has resulted in a denatured, inactive form of the polypeptide (Green et al., 1987). This problem is likely inherent to the improper folding of the protein during translation.
To analyze the expression and biological function of cloned phosphatase 2A, we initiated experiments that had two goals. First, to modify the primary sequence of phosphatase 2A at both the NHz and COOH terminus to determine if this would allow functional expression of the translation product; and second, to encode a unique antigenic epitope to discriminate the plasmid expressed phosphatase 2A from the endogenous phosphatase 2A catalytic subunit. Fig. 1 shows the strategy that was successful in accomplishing both goals. The nineamino-acid epitope YDYDVPDYA of the influenza hemagglutinin protein (Wilson et al., 1984) was encoded a t either the NH2, COOH, or both the NH2 and COOH termini of phosphatase 2A. The chimeric cDNAs encoding the nine-aminoacid epitope and the full-length phosphatase 2A polypeptide were inserted into the pCW1-NE0 expression plasmid (Woon et al., 1988) and used for both transient and stable transfection analysis.

shows the expression and fractionation of the epitope-tagged phosphatase 2A. In panel A the transient expression in COS cells and stable expression in Ratla fibroblasts is shown. Transient expression in COS cells revealed that the NH2-terminal (WP'-PP2A)
and NH2-and COOH-terminal (Nep'-PP2A-Cep') fusion proteins were expressed with an M, of 38,000 and 40,000, respectively. Interestingly, the COOH-terminal epitope-tagged phosphatase 2A fusion protein (PP2A-CP') was not detectable. In stable Ratla transfectants, only the NHz-terminal epitope-tagged phosphatase 2A fusion protein was detected. Multiple independent Ratla clones demonstrated varying levels of NeP"PP2A expression. Assay of [32P]histone H1 phosphatase activity in crude cell lysates of Nep'-PP2A-expressing cells did not show a dramatic change in total histone dephosphorylation activity relative to wild-type cells (not shown). This is not unexpected given the levels of expression and the presence of high histone phosphatase activity found in the crude lysates. Many independent G418-resistant Ratla clones were screened for expression of the PP2A-CePi and NeP'-PP2Ac ' p ' fusion proteins (not shown). No clones were isolated expressing the PP2A-CP' fusion and very low level expression of the Nepi-PP2A-CeP' fusion protein was observed with Ratla transfectants. These findings indicated that modification of the phosphatase 2A COOH terminus appeared to diminish expression of the polypeptide, particularly in stable Ratla cell transfectants. A similar lack of expression of the COOHterminal epitope-tagged Phosphatase 2A fusion protein was observed with stable transfectants of Chinese hamster ovary and NIH3T3 cells (not shown). In transient assays, the NeP'-PP2A-CePi fusion protein was expressed reproducibly at   Fig. 2 shows the relative subcellular distribution of the endogenous wild-type phosphatase 2A catalytic subunit in nuclear and high speed pellet and supernatants of lysed COS and Ratla cells. When cells are lysed in the absence of detergents by passage through a 26-gauge needle the majority of the phosphatase 2A catalytic subunit is found in the high speed supernatant. Comparatively low levels are found in the nuclear and high speed pellet fractions. Thus, the phosphatase 2A enzyme is primarily a soluble protein.
The plasmid expressed NHz-terminal and NHz-and COOH-terminal epitope-tagged phosphatase 2A fusion proteins have similar fractionation profiles when transiently expressed in COS cells (Fig. 2, panel C). Similarly, the NHzterminal epitope-tagged phosphatase 2A fusion behaved primarily as a soluble cytoplasmic protein when stably expressed in Ratla cells (panel C). These findings demonstrate that the recombinant NHz-terminal epitope-tagged phosphatase 2A fusion proteins were soluble when expressed in mammalian cells and fractionated using differential centrifugation essentially identical to the endogenous wild-type enzyme. Fig. 3 shows the fractionation by Mono Q FPLC of the wild-type endogenous phosphatase 2A and NHz-terminal epitope-tagged phosphatase 2A fusion protein. Using protein kinase C-phosphorylated [32P]histone H1 as a phosphatase 2A substrate (Jakes and Schlender, 1988), it is readily apparent that the elution profile of the NHz-terminal epitopetagged phosphatase 2A fusion protein is virtually identical to the endogenous wild-type phosphatase 2A enzyme. The two overlapping peaks of phosphatase activity observed in the control and transfected cells is related to the partial resolution of the heterotrimeric phosphatase 2A complex (first eluting peak, fractions 11-12) uersus the heterodimeric phosphatase 2A enzyme (second eluting peak, fractions 12-15). This finding is based on the Mono Q elution profiles for purified heterodimeric (AC) complex and the heterotrimeric phosphatase 2A enzyme, which has the C subunit associated with the A and B regulatory subunits (Kamibayashi et al., 1991).
Comparing the Mono Q elution profiles of purified AC and ABC forms of phosphatase 2A with the elution profile of NHpterminal epitope-tagged phosphatase 2A, indicated that the NHz-terminal epitope-tagged phosphatase 2A eluted in both peaks corresponding to the two enzyme forms (not shown). This finding indicates that the NHp-terminal epitope-tagged fusion protein associates with the phosphatase 2A regulatory subunits.
By using the 12CA5 antibody, it was also possible to immunoprecipitate the epitope-tagged phosphatase 2A fusion proteins (Fig. 4). The NH2-and double NHz-and COOHterminal phosphatase 2A fusion proteins were shown to be active in immunoprecipitates from lysates of the appropriate COS or Ratla cell transfectants. Immunoprecipitation of functional epitope-tagged phosphatase 2A could be accomplished from both crude lysates and following fractionation on Mono Q columns. It was found that the phosphatase 2A activity was significantly increased when the epitope-tagged phosphatase 2A fusion protein was released from the antibody complex using the YPYDVPDYA antigenic peptide, indicating the free enzyme was more active than when complexed to the antibody. The basis for the increased activity is unclear but may be related simply to substrate accessibility.
The NHz-terminal phosphatase 2A fusion protein released from the immune complex with epitope peptide was subsequently resolved by SDS-PAGE in parallel to serial dilutions of purified phosphatase 2A catalytic subunit (not shown). The proteins were then immunoblotted with a phosphatase 2A antibody raised against the COOH-terminal 14 amino acids of phosphatase 2A, which will equally recognize the wild-type and NHz-terminal epitope-tagged proteins. Using this method to determine the relative amount of NHp-terminal phospha-

FIG. 3. Mono Q-FPLC analysis of transiently expressed
NeD'-PP2A. COS-1 cells transfected with pCW1-NE0 (control) or pCW 1-NE0 encoding the NePi-PP2A cDNA were solubilized and centrifuged to remove nuclei and cell debris. Extracts (500 pl) were applied to a Mono Q FPLC column and eluted with a linear 0-500 mM NaCl gradient. Panel A, fractions were collected (1 ml) and the proteins precipitated with trichloroacetic acid (10%). Proteins were resolved by SDS-PAGE, the proteins transferred to nitrocellulose, probed with 12CA5 monoclonal antibody, then rabbit anti-mouse, and [12sI]protein A as described under "Experimental Procedures." Following autoradiography, the blots were reprobed with anti-phosphatase 2A monoclonal antibody, rabbit anti-mouse, and   4B). Thus, the attachment of the nine-amino-acid sequence to the phosphatase 2A polypeptide had little effect on its enzymatic activity. The immunoprecipitated NH2-terminal phosphatase 2A fusion protein is inhibited by okadaic acid with an ICbo of 0.3-0.7 nM (Fig. 5). This is similar to what is observed with the purified phosphatase 2A catalytic subunit. Cummulatively, these findings indicate that the soluble character, enzymatic activity, chromatographic behavior, and okadaic sensitivity characteristics of the NH,-terminal epitope-tagged phosphatase 2A are virtually indistinguishable from the endogenous wild-type catalytic phosphatase 2A subunit.
The ability of the YPYDVPDYA peptide sequence when fused to the NH2 terminus to allow functional expression of cloned phosphatase 2A catalytic subunit may involve changes in protein folding during translation. Why NH2-terminal modification of the phosphatase 2A catalytic subunit permits functional expression relative to COOH-terminal modification is obviously unknown, but its utility is apparent. The epitope-tagged phosphatase 2A, for the first time, provides functional expression and detection of recombinant serine/ threonine phosphatases. This has not been accomplished in mammalian cells with any other procedure for this set of gene products. The expressed fusion protein is easily resolved from endogenous cellular phosphatases by immunoprecipitation, and the antigenic epitope peptide can be used to dissociate the enzyme from the antibody complex. This strategy may have general utility for the study of other phosphatases as well. In fact, any phosphatase for which a cDNA has been isolated can be epitope-tagged and tested for expression in mammalian cells.