Identification of a new family of protein phosphatase 2A regulatory subunits.

Protein phosphatase 2A (PP2A) is a major intracellular protein phosphatase that regulates multiple aspects of cell growth and metabolism. The ability of this widely distributed heterotrimeric enzyme to act on a diverse array of substrates is largely controlled by the nature of its regulatory B subunit. Only two gene families encoding endogenous B subunits have been cloned to date, although the existence of several additional regulatory subunits is likely. We have identified by two-hybrid interaction a new human gene family encoding PP2A B subunits. This family, denoted B56, contains three distinct genes, one of which is differentially spliced. B56 polypeptides co-immunoprecipitate with PP2A A and C subunits and with an okadaic acid-inhibitable, heparin-stimulated phosphatase activity. The three B56 family members are 70% identical to each other but share no obvious homology with previously identified B subunits. These phosphatase regulators are differentially expressed, with B56α and B56 highly expressed in heart and skeletal muscle and B56β highly expressed in brain. The identification of this novel phosphatase regulator gene family will facilitate future studies on the control of protein dephosphorylation and the role of PP2A in cellular function.

Protein phosphatase 2A (PP2A) is a major intracellular protein phosphatase that regulates multiple aspects of cell growth and metabolism. The ability of this widely distributed heterotrimeric enzyme to act on a diverse array of substrates is largely controlled by the nature of its regulatory B subunit. Only two gene families encoding endogenous B subunits have been cloned to date, although the existence of several additional regulatory subunits is likely. We have identified by two-hybrid interaction a new human gene family encoding PP2A B subunits. This family, denoted B56, contains three distinct genes, one of which is differentially spliced. B56 polypeptides co-immunoprecipitate with PP2A A and C subunits and with an okadaic acid-inhibitable, heparinstimulated phosphatase activity. The three B56 family members are 70% identical to each other but share no obvious homology with previously identified B subunits. These phosphatase regulators are differentially expressed, with B56␣ and B56␥ highly expressed in heart and skeletal muscle and B56␤ highly expressed in brain. The identification of this novel phosphatase regulator gene family will facilitate future studies on the control of protein dephosphorylation and the role of PP2A in cellular function.
Protein phosphatase 2A (PP2A) 1 is a major intracellular phosphatase that regulates such diverse cellular processes as DNA replication, transcription, signal transduction, and intermediary metabolism (1)(2)(3). PP2A is a heterotrimer, containing A, B, and C subunits. The catalytic activity of PP2A resides in the C subunit, a 36-kDa protein encoded by two 97% identical genes. The C subunit binds stably to the carboxyl-terminal region of the A subunit, a 65-kDa rod-shaped polypeptide consisting of 15 imperfect repeats. The B subunits bind to the amino-terminal region of the A subunit (Fig. 1A) and determine the substrate specificity of the complex (4 -8). Three distinct B subunits have been biochemically isolated from a variety of mammalian tissues (9 -14), and several studies have suggested the existence of additional B subunits (15,16). Additionally, several DNA tumor viruses encode polypeptides that can function as PP2A B subunits (17)(18)(19). The B subunits purified to date migrate in SDS-PAGE with the apparent molecular masses of 54 kDa (B54), 55 kDa (B55), and 72 kDa (B72). Three cDNAs encoding 55-kDa B subunits have been identified (10,11,20); the B55 family members are 80 -90% identical, and their level of expression varies by tissue type. A cDNA encoding the 72-kDa B subunit has also been cloned, and a splice variant encoding a 130-kDa protein has been identified (12). The sequence of the 54-kDa B subunit cDNA has not yet been reported. Interestingly, the amino acid sequences of the B55 and B72 subunits and the viral PP2A binding proteins show little homology to each other; thus, no common motif mediating the interaction of the B subunit with the PP2A A and C subunits has been discovered.
Heterotrimeric PP2A enzymes with different B subunits have distinct substrate specificities (7,8,21), a mode of phosphatase regulation that has important functional effects. For example, PP2A can turn SV40 DNA replication on or off, depending on the type of B subunit in the holoenzyme (6,22). Viral replication in vitro is controlled by the activity of viral initiator phosphoprotein, SV40 large T antigen. The heterotrimeric form of PP2A containing B55 removes a phosphoryl group from threonine 124 (a cyclin-dependent kinase site) (23) and inactivates T antigen's ability to initiate SV40 DNA replication, while the PP2A heterotrimer containing B72 removes inhibitory phosphoryl groups from serines 120 and 123 (casein kinase I sites) (24,25) and activates SV40 DNA replication (6).
An additional role of B subunits may be to act as targeting subunits that direct PP2A to specific subcellular locations. This method of regulation has been most clearly demonstrated for protein phosphatase 1, where the catalytic subunit is localized to its substrates, e.g. phosphorylase, phosphorylase kinase, and glycogen synthase, by association with a specific glycogen-binding subunit (2). Similarly, PP2A-B55␣ has recently been shown to associate with microtubules (26). While PP2A activity has also been found in membrane and nuclear fractions, the B subunits of PP2A that direct the heterotrimer to these sites have not yet been identified.
This current study was designed to identify novel PP2A B subunits. Using the yeast two-hybrid method (27) with the PP2A A subunit as bait, we identified a novel gene family encoding three polypeptides with a predicted size of approximately 56 kDa that were 70% identical to each other but with no significant similarity to either B55 or B72. Full-length polypeptides expressed in 293 cells bound to PP2A A and C subunits and co-immunoprecipitated with a heparin-stimulated, okadaic acid-inhibited phosphorylase phosphatase activity. Northern blots of human tissue showed that these genes have tissue-specific expression patterns, with two isoforms highly expressed in heart and skeletal muscle and one highly expressed in the brain. These results are further evidence that protein serine/threonine phosphatase diversity is generated in large part by association of a common catalytic subunit with an increasing array of regulatory or targeting subunits.

EXPERIMENTAL PROCEDURES
Yeast Two-hybrid Screen Strains and Constructs-A two-hybrid screen based on the method of Fields and Song (28) (Fig. 1) was performed using the Saccharomyces cerevisiae strain L40(MATa, his3⌬200, trp1-901, leu2-3, 112, ade2, LYS::(lexAop) 4 -HIS, URA3::(lexAop) 8 -lacZ (constructed by S. Hollenberg). AMR70 (MAT␣, his3, lys2, trp1, leu2, URA::(lexAop) 8 -lacZ) (constructed by R. Sternglanz, the gift of S. Hollenberg) was used in mating assays to test for specific interactions. Untransformed yeast were grown in YPD, and transformed yeast were grown in minimal media with necessary supplements (29). The HeLa cDNA library in pGAD-GH (with the selectable marker LEU2) contains inserts averaging 1500 bp in length and was the gift of Greg Hannon. pGAD-GH expresses inserts as fusion proteins with the transcriptional activation domain of GAL4 ( Fig. 1A) (30). All the baits and positive controls used in the two-hybrid method were cloned by use of polymerase chain reaction using Taq polymerase. High levels of substrate (ϳ100 ng) and low numbers of cycles (10 -15) were used to minimize inadvertent mutagenesis. Polymerase chain reaction primers (Table I) added appropriate restriction sites for inframe insertion into the polycloning sites of pBTM116 or pGAD-GH.
The bait plasmid was based on pBTM116 (originally constructed by P. Bartel and S. Fields; with the selectable marker TRP1 and modified by the insertion of the ADE2 gene (31) in the PvuII site) and expresses LexA fused to the full-length human PP2A A␣ subunit (5,18). Expression of this construct, LexA-65A, was confirmed by immunoprecipitation of the fusion protein from [ 35 S]methionine-labeled yeast extracts with an antibody to the PP2A A subunit. Constructs encoding mutant A subunits were also prepared to facilitate classification of isolated clones by the mating assay. These included COOH-terminal truncations A397 (LexA-A397) and A315 (LexA-A315) and the 12/5 loop substitution mutant (LexA-SUB), all described by Ruediger et al. (4,5) (Fig. 1A). The 12/5 (LexA-SUB) loop mutant was generated by site-directed mutagenesis; this mutation has been previously shown to eliminate B55 and SV40 small t antigen binding to the A subunit. Human B72 and C subunit genes were cloned into pGAD-GH in frame with the activation domain of GAL4 for use in control experiments (GAL4-B72, GAL4-C). Rat B55␣ cDNA (20) was cloned into the two-hybrid expression vector pVP16 (32) for use in control experiments (pVP16-B55).
Two-hybrid Screen Methods-500 g of HeLa cDNA plasmid was used for the large scale transformation of the L40 strain carrying the LexA-65A bait plasmid (33), and interacting cDNA clones were selected by rapid growth on His Ϫ plates. Positive colonies were further screened for lacZ expression using the colony lift method (29). lacZ ϩ yeast were then allowed to lose the bait plasmid by removal of selection (Trp ϩ Leu Ϫ media) and were identified by accumulation of a red metabolite due to loss of the ADE2 gene also carried on the bait plasmid. Yeast containing only the library plasmids encoding putative interacting proteins were then tested for specific interaction by mating to AMR70 yeast-carrying plasmids encoding LexA-65A, LexA-SUB, LexA-A315, LexA-397, or LexA-lamin ( Fig. 1) (32,34).
Full-length clones were isolated from a human fetal brain cDNA library in bacteriophage lambda by standard methods (35) and sequenced on both strands using Sequenase 2.0 according to the manufacturer's instructions.
Expression of Hemagglutinin (HA)-tagged B Subunits-The protein coding sequences of B55␣ (20), B56␣, and B56␤ were cloned using the indicated polymerase chain reaction primers into pCEP-4/Lerner, a CMV promoter-driven expression vector based on pCEP4 (Invitrogen) that encodes a hemagglutinin epitope tag (36) at the amino terminus of the expressed polypeptide. 293 cells (human embryonic kidney cells transformed with adenovirus) grown in Dulbecco's modified Eagle's medium (Life Technologies, Inc.) plus 10% supplemented calf serum were transfected using lipofectamine (Life Sciences) following the manufacturer's instructions. Using a plasmid that expresses lacZ from the CMV promoter, the transfection efficiency was determined to be about 40%. Cells were lysed 24 -36 h after transfection on ice in 20 mM Tris-HCl, pH 7.5, 0.2% Nonidet P-40, 10% glycerol, 200 mM NaCl, 1 mM EDTA, 1 mM EGTA, 1 mM phenylmethylsulfonyl fluoride, 2 g/ml leupeptin, 1 mM benzamidine, and 2 g/ml pepstatin. Lysates were then centrifuged at 14,000 ϫ g for 3 min, and complexes were immunoprecipitated from the supernatant by the addition of 12CA5 mAb (a mouse monoclonal antibody that recognizes the hemagglutinin epitope) (36) and protein A-agarose. For immunoblotting, washed immunoprecipitates were solubilized in 1 ϫ SDS-PAGE loading buffer, separated by SDS-PAGE on 10% gels, transferred to nitrocellulose, then probed with rabbit anti-PP2A-A and PP2A-C antibodies and visualized by enhanced chemiluminescence (Amersham). Phosphatase activity in the immunoprecipitate was determined by quantitating the release of trichloroacetic acid-soluble 32 P from [ 32 P]phosphorylase (37). All phosphatase assays were performed in duplicate, and the amount of PP2A activity was calculated by subtracting radioactivity released from immunoprecipitates of untransfected cells or cells transfected with a HA-ERK1 construct.
Sequence analysis was performed with the Wisconsin package (38) and GenBank searches using the BLAST algorithm (39).

Identification of HeLa Cell cDNA Library Clones That Interact with the PP2A A Subunit in the Yeast Two-hybrid Screen-
The A subunit of PP2A was expressed in S. cerevisiae strain L40 as a fusion protein with the DNA binding protein LexA (LexA-65A) and used as the bait for a two-hybrid screen (Fig.  1). Two-hybrid interactions in this strain activate transcription of the HIS3 and lacZ genes. In pilot experiments, plasmids expressing the fusion proteins VP16-B55, GAL4-B72, and GAL4-C subunit were constructed and tested for interaction with LexA-65A. The GAL4-B72 fusion protein interacted strongly and specifically with LexA-65A (Fig. 1B), demonstrating the feasibility of identifying PP2A B subunits with the two-hybrid method.
A HeLa cell cDNA library in pGAD-GH was then screened for proteins that interact with the LexA-65A bait (Fig. 1). 17 ϫ 10 6 transformants were plated on His Ϫ plates, and rapidly growing colonies were further screened by blue/white assay for lacZ expression and for specific interactions in a mating assay. 212 clones were isolated that interacted strongly and specifically with the A subunit. These 212 clones were separated into 16 groups based on dot blot hybridization and restriction digests of polymerase chain reaction-amplified inserts. The 5Јand 3Ј-ends of representative clones from each of the 16 groups were then sequenced. The three groups that are the subject of this report encoded polypeptides that are 70% identical to each other and are referred to as B56␣, -␤, and -␥1. Five clones encoded B56␣, two clones encoded B56␤, and one clone encoded B56␥1.
To identify regions of the PP2A A subunit bait required for interaction with these putative B subunits, yeast strain AMR70 carrying mutant bait constructs (as described under "Experimental Procedures") were mated to L40 yeast carrying only the prey plasmid; interaction was assessed by growth on His Ϫ media (Fig. 1). The B56 family of B subunits isolated in this screen did not interact with either the carboxyl-terminal truncated A subunits or the 12/5 loop mutant. This is similar to the result reported when B55 binding to PP2A AC complex was assessed in an in vitro assay (4,5). In contrast, GAL4-B72 interacted with a carboxyl-truncated A subunit, suggesting a different mode of interaction with the A subunit. None of the clones isolated in this two-hybrid screen encoded the PP2A C subunit, B55, or B72. This is consistent with the inability of the GAL4-C and VP16-B55 fusions to interact strongly with LexA-65A in control experiments and the reported absence of B72 mRNA in HeLa cells (12). Ruediger and co-workers (4,5) have demonstrated in in vitro studies that B55 binding to PP2A A subunit requires a C subunit binding site, suggesting that B55⅐C contacts are essential. Thus, the absence of human C subunit may have prevented formation of a stable two-hybrid complex between B55 and LexA-65A.
Full-length cDNA Clones Were Identified for the Partial Clones Found by the Yeast Two-hybrid Screen-Full-length cDNA clones were obtained for B56␣ and B56␤ by screening a human fetal brain cDNA library with 32 P-labeled probes from the 5Ј-end of clones from the HeLa two-hybrid library. Two independent phage clones containing similar amounts of additional 5Ј-sequence were obtained for B56␣, and 4 phage clones with additional 5Ј-sequence were obtained for B56␤. Since the reading frame of each isolate was known from the two-hybrid screen, in-frame stop and start sites could be used to deduce putative protein coding regions. The sequence of B56␣ was determined using the phage clones for bases 1-456 and the HeLa two-hybrid library clone for bases 456-3120. B56␤ sequence was entirely determined from the phage cDNA clone. For B56␥, only the 3Ј-and 5Ј-ends (350 bases) of the HeLa library clone were sequenced, and no attempt was made to find full-length clones. The full-length B56␣ cDNA is 3120 bp, with an open reading frame encoding a polypeptide of 56.1 kDa. The translational start was determined to be at nucleotide 572, since this was the first in-frame ATG and gives a start site offset only four amino acids from that of B56␤. Additional ATG codons in the B56␣ sequence were either far downstream (252 bp) or out of frame. The B56␤ cDNA is 2450 bp, and encodes a polypeptide of 57.3 kDa. The putative initiator methionine for B56␤ at nucleotide 326 is the first in-frame ATG upstream of the fusion site seen in HeLa prey library clones and has an optimum context for translational initiation (40). The only other in-frame ATG is at bp 230 but is followed by an in-frame stop codon at bp 242. We have not isolated a full-length cDNA encoding B56␥; the 1447-bp clone obtained in the two-hybrid screen codes for a predicted protein of 51.5 kDa. The first 1298 bp of B56␥1 are identical to a 3700-bp expressed sequence, HumORFY (GenBank accession number D26445) (41). However, HumORFY, encoding a 56.5-kDa polypeptide, appears to be an alternatively spliced message, as it completely diverges from B56␥1 7 codons upstream of the ␥1 stop site and encodes an additional 43 amino acids. This would predict that there are at least two splice variant forms of B56␥, a conclusion also supported by Northern blot analysis (see Fig. 5 and below). The three B56 cDNAs encode polypeptides with 70% identity to each other, with divergence limited to the extreme amino-and carboxyl-terminal regions. The B56 genes are also homologous to the S. cerevisiae gene RTS1/SCS1 (see "Discussion"). However, the predicted B56 amino acid sequences show no obvious similarity to previously cloned B subunits (Fig. 2).
B56 Associates with the A and C Subunits of PP2A in Human Cells-By definition, PP2A B subunits form stable heterotrimers with the A and C subunits of PP2A. To evaluate the ability of these putative B subunits to form stable complexes with PP2A A and C subunits in vivo, B56␣ and B56␤ polypeptide coding sequences were cloned with a 5Ј-extension encoding FIG. 1. Application of the two-hybrid method to identify PP2A B subunits. A, subunit interactions in the PP2A heterotrimer and a schematic of the fusion proteins used. The PP2A B and C subunits are thought to bind to the amino and carboxyl regions of the A subunit as shown (4). The 65-kDa A subunit was expressed as a fusion with the LexA protein (LexA-65A). It was used to screen a HeLa cell cDNA library in pGAD GH (see "Experimental Procedures") for interacting proteins. Additional LexA constructs (LexA-315, LexA-397, LexA-SUB, and LexA-lamin) were used to characterize the interaction between the PP2A A subunit and putative B subunits. Interaction between the LexA constructs and the GAL4 fusion proteins activate transcription of HIS3 and lacZ genes in S. cerevisiae L40 and AMR70 cells. B, B56 clones interact specifically and distinctively with the PP2A A subunit. L40 cells expressing the indicated B subunit fusion proteins were mated to AMR70 cells expressing LexA-65A, LexA-modified A subunits, or the nonspecific bait LexA-lamin. The resulting diploids were tested for ability to grow on His Ϫ plates. ϩ signifies growth was similar to growth on His ϩ plates, Ϯ indicates growth was less than half the rate of growth on His ϩ plates, and Ϫ indicates no growth. SV40 small t antigen (t ag) was not tested in the mating assay; the data indicated here in italics are predicted from the results of Ruediger et al. (4,5). the hemagglutinin epitope into a CMV expression vector. To verify that these constructs indeed encoded proteins of the predicted molecular weight, cytosolic extracts from transfected cells were subject to SDS-PAGE and immunoblotting with the 12CA5 mAb (Fig. 3A). HA-B56␣ migrates slightly faster than HA-B55␣, and HA-B56␤ migrates at about the same speed as HA-B55␣.
Having established that the constructs indeed produced soluble proteins of the expected size, extracts from transfected cells were subjected to immunoprecipitation with 12CA5 mAb. Immunoprecipitated proteins were separated by SDS-PAGE and immunoblotted with antibodies that recognize PP2A A and C subunits. As Fig. 3, B and C, demonstrates, HA-tagged B56␣ and B56␤ polypeptides co-immunoprecipitate with PP2A A and C subunits in transfected human cells, indicating that B56 forms a stable complex with PP2A in human cells.
PP2A Phosphatase Activity Co-immunoprecipitates with HAtagged B56␣ and B56␤-PP2A heterotrimers contain a catalytic (C) subunit whose phosphatase activity is (i) inhibited by nanomolar concentrations of okadaic acid and (ii) stimulated toward nonspecific substrates such as phosphorylase a by heparin (presumably by facilitating dissociation of the catalytic subunit from the A and B subunits (42)). To determine whether the complex containing HA-B56 indeed contained active PP2Alike phosphatase, 12CA5 immunoprecipitates were incubated with 32 P-labeled phosphorylase a (Fig. 4). HA-B56␣ and HA-B56␤ both co-immunoprecipitate with a phosphorylase phosphatase activity that is 50 -80% inhibited by the inclusion of 1 nM okadaic acid (Fig. 4A). While both PP2A and PP1 can be inhibited by okadaic acid, PP1 is 10 -100-fold less sensitive than PP2A c , with an IC50 of 10 -20 nM. Thus, inhibition by 1 nM okadaic acid strongly suggests that the phosphatase activity present in these assays is PP2A. Phosphorylase phosphatase activity was also increased 2-4-fold in the presence of 15 g/ml heparin (Fig. 4B), a result expected for PP2A complexes. Thus, the data strongly suggest that the B56 polypeptides are part of a stable and active PP2A complex.
Finally, to evaluate whether the B56 polypeptides bound to PP2A heterotrimers (forming heterotetramers) or formed novel heterotrimers, soluble extracts from transfected cells were analyzed by glycerol gradient velocity sedimentation followed by immunoblot analysis with 12CA5 mAb. The majority (Ͼ75%) of HA-B56␣ sedimented at the same velocity as HA-B55␣ and as aldolase, a 158-kDa protein (data not shown). These results are most consistent with the HA-B56 polypeptides forming a heterotrimeric complex with the PP2A A and C subunits.
Tissue-specific Expression of B56 Isoforms-The previously cloned PP2A regulatory subunits have shown tissue-specific and developmentally regulated patterns of gene expression, with specific isoforms preferentially expressed in muscle and brain (10,13). To determine the expression pattern of the B56 genes, a multiple human tissue Northern blot was probed with sequences from the three B56 genes (Fig. 5). B56␣ mRNA is approximately 3500 nucleotides, with expression in all tissues examined but with the highest expression in heart and skeletal muscle. B56␤ mRNA is 3000 nucleotides with highest expression in brain. The apparent sizes of B56␣ and B56␤ mRNAs (3.5 and 3.0 kb) are 380 -550 nucleotides longer than the cDNA clones we have obtained. B56␥ probe hybridizes with three mRNA species of 1.8, 2.1, and 4.4 kb. The 4.4-kb transcript is most highly expressed in heart and skeletal muscle, while the shorter transcripts are present largely in heart alone. The 4.4-kb species most likely corresponds to the 3700-bp HumORFY sequence, while the smaller more cardiac-specific mRNAs most likely correspond to the B56␥1 clone isolated in this study.
FIG. 2. The B56 family contains closely related proteins. B56 is encoded by one yeast and three human cDNAs with 70% amino acid identity. Predicted amino acid sequences were aligned using the Genetics Computer Group PILEUP program (38) and displayed using SeqVu (45). Only residues 219 -758 of RTS1 are shown. Amino acid identities are shaded, and similarities are boxed; gaps introduced to optimize the alignments are indicated by a dash .   FIG. 3. The A and C subunits of PP2A co-immunoprecipitate with HA-tagged B56. A, 293 cells were transiently transfected with a CMV expression plasmid encoding HA-tagged B56, B55, or ERK1. Soluble extracts from transfected cells were separated by SDS-PAGE on a 10% gel and then transferred to a nitrocellulose membrane. HA-tagged proteins were detected by immunoblotting with the 12CA5 mAb. Asterisk indicates a 12CA5 mAb cross-reacting band present in untransfected cells. B and C, soluble extracts from transfected cells were subjected to immunoprecipitation with 12CA5 mAb and protein A-agarose. Immunoprecipitates were eluted in 1 ϫ SDS-PAGE loading buffer, separated on a 10% polyacrylamide gel, and transferred to a nitrocellulose membrane, where they were probed with affinity-purified rabbit antibodies that recognize the PP2A A and C subunits. As a positive control, whole lysate was run in lane one (EXTRACT). Asterisk indicates IgG heavy chain present in the immunoprecipitates.

DISCUSSION
Reversible protein phosphorylation is one of the most widely utilized mechanisms to regulate cellular processes. The activity of many cellular enzymes is determined by the net protein phosphorylation level, which is determined by the balance of specific protein kinase and protein phosphatase activities. The substrate specificity of the regulatory protein kinases is determined largely by the diversity of their catalytic subunits, as demonstrated by the estimate that over 2000 distinct protein kinase catalytic subunits are encoded by the human genome (43). However, there appears to be a more limited number of protein serine/threonine phosphatase catalytic subunits. Their substrate specificity is determined instead by association with a variety of regulatory and targeting subunits (2).
We now report the identification, using the two-hybrid method, of a new gene family encoding PP2A regulatory subunits. These genes encode authentic PP2A B subunits based on the following evidence. First, all members of the gene family B56 (-␣, -␤, -␥) interacted with a LexA-A subunit bait in the two-hybrid assay and failed to interact with irrelevant baits such as LexA-lamin. Second, PP2A A and C subunits were shown to associate with epitope-tagged B56␣ and ␤ polypeptides in human cells in co-immunoprecipitation assays. Third, HA-B56␣ and -␤ co-immunoprecipitated with a phosphatase activity that was inhibited by 1 nmol of okadaic acid and enhanced by heparin, the results expected for a component of a PP2A complex.
It is of note that the B56 gene family has no obvious similarity to previously identified gene families encoding polypeptides that bind to the amino-terminal end of the PP2A A subunit (B72/130, B55, B56, and polyoma and SV40 small t antigens). Thus, it has not been possible to define an interaction domain that is required for binding to the A subunit of PP2A. The B56 family has a very highly conserved (80% identical) central region, while both the carboxyl terminus and the amino terminus are significantly more divergent. This suggests that the conserved region is required for interaction with the A and possibly the C subunit, whereas the ends may perform different functions such as regulation of substrate specificity or intracellular location of PP2A.
The lack of sequence similarity between the different B subunit families suggests that they each bind differently to the PP2A core A⅐C complex and exert their effects on substrate specificity in this manner. In support of this hypothesis, we note that test preys encoding B55, B72, and B56 fusion proteins each interacted differently in the two-hybrid screen when tested for association with the mutant and wild type A subunits (Fig. 1B). Thus, B72, unlike B55 and B56, interacted with carboxyl-truncated A subunit A397, and the B56 preys interacted with full-length A subunit significantly more strongly than did the B55 preys. B55 interaction with the A subunit is reportedly stabilized by the presence of the C subunit (4); this is supported by our finding that the B55 prey construct interacted only weakly with the A bait in yeast, where no mammalian C subunit exists. B72 and B56, like SV40 small t antigen, may bind more tightly to the A subunit and thus have less of a requirement for a B-C interaction. Additionally, within each B subunit family there are non-conserved sequences that may contribute to unique interactions with the PP2A A and C subunits or contribute to interactions with additional cellular proteins. For example, PP2A heterotrimers containing either B55␣ or B55␤ differ in their response to effector molecules such as protamine and heparin (44). Differential expression in diverse tissues of B56 and B55 family members also implies that each isoform performs a specific function.
A yeast gene 68% similar to the B56 gene family has been independently identified by two groups (Fig. 2) (24,25). SCS1 (suppressor of chaperonin sixty-1), was isolated as a high copy suppressor of several temperature-sensitive alleles of hsp60 (a mitochondrial chaperonin) (24). SCS1 in budding yeast is a cytosolic protein that when overexpressed appears to positively regulate transcription of additional chaperonin genes. The identical gene, termed RTS1, was cloned as a multicopy suppressor of ROX3, which encodes a transcriptional regulator involved in the response to anaerobic conditions. Thus, one function of the B56 homologues in yeast appears to be as regulators of the transcriptional response to environmental stress.
The PP2A regulatory subunits we have identified are similar in size and tissue distribution to a previously purified regulator FIG. 4. PP2A activity binds to HA-tagged B56. 293 cells were transiently transfected with the CMV expression plasmids. The cytosolic fractions were incubated with 12CA5 mAb and protein A-agarose. Phosphatase activity in immunoprecipitates was assayed using [32 P]phosphorylase a as a substrate. The phosphatase activity was also assayed in the presence of 1 nmol of okadaic acid (O.A.) (upper) or 15 g/ml heparin (lower). Duplicate assays varied by less than 10%, and all reactions converted less than 25% of phosphorylase a to phosphorylase b. PP2A activity was calculated by subtracting the background 12CA5precipitatable activity present in non-transfected cells. The specific activity was 4 -10-fold higher than background. of PP2A with an M r of 54,000 as assessed by SDS-PAGE (13). If the B56 genes indeed encodes the 54-kDa protein, we would predict based on Northern blot analysis that the cardiac B subunit, also known as BЈ, is encoded by B56␣ or B56␥.
PP2A has been implicated in the control of the cell cycle and the initiation of DNA replication. Since the PP2A A and C subunits are distributed in multiple cellular compartments, one function of a B subunit may be to target the heterotrimer to the nucleus, where it can participate in the regulation of these processes. The expression of the B55, B56, and B72 genes largely in terminally differentiated tissues such as heart and brain suggests that the PP2A B subunit that has direct involvement in cell cycle regulation or DNA replication remains to be identified.