Saccharomyces cerevisiae homologs of mammalian B and B' subunits of protein phosphatase 2A direct the enzyme to distinct cellular functions.

Protein phosphatase 2A (PP2A) is a major cellular serine/threonine protein phosphatase, present in the cell in a variety of heterotrimeric forms that differ in their associated regulatory B-subunit. Cloning of the mammalian B′ subunit has allowed the identification of a highly homologous Saccharomyces cerevisiae gene, RTS1. Disruption of the gene results in a temperature-sensitive growth defect that can be suppressed by expression of rabbit B′α or B′γ isoforms. The B′α subunit is much more effective in restoring normal growth at 37 °C than B′γ. Immunoprecipitated Rts1p was found associated with type 2A-specific protein phosphatase activity that is sensitive to 2 nM okadaic acid, but not to 100 nM phosphatase inhibitor-2, and to be phosphorylated in vivo. However, overexpression of RTS1 was unable to suppress the cold sensitivity, defective cytokinesis, and abnormal cell morphology resulting from defects in the CDC55 gene, which encodes the yeast homolog of a different B subunit of another form of 2A phosphatase, PP2A1. These results indicate that Rts1p is a yeast homolog of the mammalian B′ subunit and that the various regulatory B-subunits of PP2A are not functionally redundant but direct the enzyme to distinct cellular functions.

Protein phosphatase 2A (PP2A) is a major cellular serine/threonine protein phosphatase, present in the cell in a variety of heterotrimeric forms that differ in their associated regulatory B-subunit. Cloning of the mammalian B subunit has allowed the identification of a highly homologous Saccharomyces cerevisiae gene, RTS1. Disruption of the gene results in a temperaturesensitive growth defect that can be suppressed by expression of rabbit B␣ or B␥ isoforms. The B␣ subunit is much more effective in restoring normal growth at 37°C than B␥. Immunoprecipitated Rts1p was found associated with type 2A-specific protein phosphatase activity that is sensitive to 2 nM okadaic acid, but not to 100 nM phosphatase inhibitor-2, and to be phosphorylated in vivo. However, overexpression of RTS1 was unable to suppress the cold sensitivity, defective cytokinesis, and abnormal cell morphology resulting from defects in the CDC55 gene, which encodes the yeast homolog of a different B subunit of another form of 2A phosphatase, PP2A 1 . These results indicate that Rts1p is a yeast homolog of the mammalian B subunit and that the various regulatory B-subunits of PP2A are not functionally redundant but direct the enzyme to distinct cellular functions.
The 2A protein phosphatases are heterotrimeric complexes consisting of a core dimer, composed of a catalytic (C2) and a regulatory (A) subunit, associated with one of the variable regulatory B-subunits (4 -6). Two mammalian C2 (7-9) subunit isoforms have been identified which are highly conserved in the living kingdom. Three genes, PPH21, PPH22, and PPH3, encoding homologs of mammalian C2, have also been found in the budding yeast, Saccharomyces cerevisiae (10 -12). Disruption of any one of these genes has no significant effect, whereas disruption of PPH21 and PPH22 causes a severe growth defect (12). However, the double mutants can survive, provided that the PPH3 gene is intact (10,11). The PP2A catalytic subunits have been shown to be essential for normal cell cycle progression in S. cerevisiae and for the organization of the actin cytoskeleton (13), as well as for the maintenance of normal cellular morphology (11). Other reports implicate the phosphatase in regulation of glycogen metabolism in yeast (14). The regulatory A subunit has been identified in cells of diverse origin. Two isoforms of the A subunits have been characterized in mammalian cells (15) and one in Drosophila (16). In S. cerevisiae, the TPD3 gene encodes a homolog of the mammalian subunit (17). Mutations in this gene greatly diminish cell growth at 14 and 37°C. At the low temperature, cytokinesis appears to be affected, whereas at 37°C, transcription by RNA polymerase III is impaired.
In contrast to the relative paucity of the A and C2 subunit isoforms, the B-subunits 2 of PP2A show great diversity. Three families, B, BЈ, and BЉ, have been found associated with the PP2A 1 , PP2A 0 , and PCS M , respectively (18 -21). Three isoforms of B subunit have been identified in mammals (21)(22)(23), and cDNAs encoding some 13 isoforms of the BЈ subunit have been isolated recently (24 -26). Cloning of the cDNA encoding the BЉ polypeptide suggests the existence of two alternatively spliced variants (27). Even though all of the B-subunits associate with the same dimeric C2⅐A core, no obvious homology has been found among the B, BЈ, and BЉ forms. The abundance of the diverse B-subunit isoforms and their association with the relatively invariant set of C2 and A subunits has led to the suggestion that the B-subunits are responsible for targeting specific phosphatase 2A holoenzymes to distinct cellular locales and/or for conferring specificity toward appropriate substrates (4,5,28).
Recent cloning of the mammalian BЈ subunits (24) allowed identification in GenBank of a highly homologous, 53-56% identical, yeast gene, RTS1 (accession number U06630), whose function is not clearly understood. This gene is unrelated to CDC55, the yeast homolog of the B subunit, whose disruption results in a cold-sensitive phenotype and morphologically aberrant cells (22). RTS1 was isolated independently by two laboratories, using different screening approaches. Evangelista et al. (29) isolated RTS1 as a multicopy suppressor of a ROX3 gene mutation. The ROX3 gene encodes an essential nuclear protein that functions in the global stress response pathway, controlling the level of CYC7 transcription. Shu and Hallberg (30) have isolated the same gene as a high copy suppressor of hsp60-ts mutant alleles. Disruption of RTS1 results in a temperature-sensitive phenotype and reduction of the mRNA levels of the mitochondrial chaperones Hsp60p, Cpn10p, and Mge1p at the restrictive temperature (30), whereas CYC7 mRNA levels were increased (29).
In this paper, we present evidence that RTS1 encodes the S. cerevisiae homolog of the mammalian BЈ subunit of PP2A 0 . The mammalian BЈ␣ and to a lesser extent the ␥ isoform were able to suppress the temperature-sensitive phenotype associated with the RTS1 disruption. Moreover, Rts1p was found associated with type 2A-specific protein phosphatase activity and was phosphorylated in vivo. However, the RTS1 gene could not restore normal growth or morphology to a CDC55-disrupted strain. These observations indicate that S. cerevisiae, like mammalian cells, possesses multiple forms of B-subunit. Most importantly, our results demonstrate that the various B-subunits of PP2A are not functionally interchangeable but direct the enzyme to distinct cellular functions.
Plasmid Constructions-Oligonucleotide primers were synthesized by the Biochemistry Biotechnology Facility at Indiana University School of Medicine on an Applied Biosystems synthesizer model 394. Restriction endonucleases were obtained from New England BioLabs or from Life Technologies, Inc. All recombinant DNA manipulations followed standard procedures (32).
Plasmid pDB20, containing the URA3-selective marker, was obtained from Kelly Tatchell (Louisiana State University, Shreveport). It was derived from the Escherichia coli-yeast shuttle vector YEp352 (33) by insertion of a 2-kb alcohol dehydrogenase (ADH1) promoter/terminator fragment (34) into the multiple cloning site of the vector. To construct the pDB21 vector with the LEU2-selective marker, the ADH1 promoter/terminator region was excised from pDB20 with EcoRI and PstI and ligated into the Yeplac181 vector (35). In both pDB20 and pDB21, a HindIII site was eliminated from the multiple cloning region such that the HindIII site between the ADH1 promoter and terminator was unique.
For expression in yeast, the complete BЈ␣ subunit coding sequence, tagged at the NH 2 terminus with a hexahistidine sequence (His-tag) was excised with NcoI and BamHI as a 1.74-kb fragment from plasmid BЈ␣⅐pET-15b (24), blunted, and subcloned into the blunted unique Hin-dIII site of pDB21 to yield pDB21(His-BЈ␣). The complete BЈ␥ subunit coding sequence was assembled by ligating the 1,481-bp AflIII-EcoRI fragment from clone BR6-2 with the 533-bp EcoRI-PvuII fragment from clone BR6-1 (24) at the EcoRI site. The resulting 2,014-bp fragment was subcloned into the blunt ended BamHI site of pTZ18U (U. S. Biochemical Corp.) to yield pTZ18U(BЈ␥). The hemagglutinin epitope tag YPY-DVPDYA (HA-tag) was introduced between the first and second codon of the open reading frame by polymerase chain reaction amplification of a 300-bp NdeI-HindIII fragment. The resulting fragment was ligated to the HindIII-SmaI 1,570-bp fragment of BЈ␥ cDNA excised from pTZ18U(BЈ␥). The 1,870-bp tagged BЈ␥ DNA was blunted and inserted into the blunt ended HindIII site of pDB21. This procedure yielded the plasmid pDB21(HA-BЈ␥).
Plasmids pDB20(HA-RTS1) and pDB21(HA-RTS1) were constructed by inserting the HA-epitope tag between the first and second codon of the open reading frame of RTS1 using YEp195RTS1 (29) as template for polymerase chain reaction amplification of a 672-bp NdeI-SpeI fragment. This fragment was then ligated at the SpeI site of the remaining 2.07-kb SpeI-BamHI portion of RTS1 excised from YEp195RTS1. The resulting 2.75-kb fragment was subcloned into the NdeI-BamHI sites of the pET-15b vector (Novagen). The 2.45-kb NdeI-Bst BI fragment ob-tained by partial digestion of pET15-b⅐RTS1 was then blunt ended and subcloned into the blunted HindIII site of pDB20 or pDB21.
Plasmids pDB20(HA-CDC55) and pDB21(HA-CDC55) were constructed as follows. The plasmid pTSV31(CDC55) containing the CDC55 gene on a 6-kb SacI-HpaI fragment (22) was used as template for polymerase chain reaction to amplify a 849-bp fragment containing an HA-tag at the 5Ј end of the coding sequence, flanked by an NdeI site. The 849-bp NdeI-BamHI fragment was joined at the BamHI site of the remaining 2.28-kb portion of the CDC55 sequence in which an XhoI linker had been added at the 3Ј end. The resulting NdeI-XhoI fragment was then ligated into the corresponding sites of pET15-b vector. The 2.43-kb NdeI-Bpu1102 I fragment, containing the HA-tag followed by the complete coding sequence and ϳ800 bp of 3Ј-untranslated sequence, was blunt ended and ligated into the filled-in HindIII site of pDB20 or pDB21. All polymerase chain reaction products were verified by dideoxynucleotide sequencing (36).
Western Blot Analyses-Yeast cells, transformed by the procedure of Ito et al. (37) or of Eble (38), were grown with vigorous shaking at 30°C in liquid synthetic medium. The cultures were harvested at late logarithmic or early stationary phase, washed with water, and pelleted in tared microcentrifuge tubes. The weight of packed cells was determined. The cells were either used immediately or stored frozen at Ϫ80°C. All subsequent steps were carried out at 0°C.
The cells were resuspended in 3 volumes of homogenization buffer consisting of 50 mM Tris-HCl, pH 7.5, 1 mM EDTA, 1 mM dithiothreitol, and 0.1 M NaCl, and containing a mixture of protease inhibitors (1 mM phenylmethylsulfonyl fluoride; 1 mM N ␣ -p-tosyl-L-lysine chloromethyl ketone; 0.1 mM benzamidine; 10 g of leupeptin/ml; and aprotinin, antipain, and pepstatin A, at 5 g/ml each). Where indicated, instead of NaCl, the buffer contained 50 mM NaF and 10 mM potassium phosphate.
Chilled acid-washed glass beads (0.45-0.50-mm diameter) were added up to the meniscus level, and the suspension was vortexed at a maximum speed in twelve 30-s bursts. The crude extract was collected by centrifugation for 2 min at 1,500 ϫ g followed by centrifugation at 18,000 ϫ g for 10 min at 4°C. The supernatant (fraction S Ϫ T) was removed, and the pellet was extracted with the same buffer containing 1% (v/v) Triton X-100. The suspension was left on ice for 15 min and then centrifuged at 18,000 ϫ g for 10 min at 4°C. The supernatant was collected (fraction S ϩ T), and the pellet was suspended in homogenization buffer without Triton X-100 (fraction P). Aliquots of all fractions were mixed with 5 ϫ SDS sample buffer at a 4:1 ratio (1 ϫ sample buffer is 12.5 mM Tris-HCl, pH 6.8, 0.5% SDS, 1% ␤-mercaptoethanol, 0.002% bromphenol blue, 10% glycerol) and analyzed by SDS-PAGE.
Approximately 50 g of protein of the S Ϫ T fraction and the corresponding volumes of the S ϩ T and pellet fractions were resolved by SDS-PAGE and transferred to nitrocellulose as described previously (39). After blocking, the filters were incubated for 1-2 h at room temperature with the appropriate antibodies. The anti-HA epitope 12CA5 monoclonal antibody was from Babco Laboratories. Antibodies against recombinant BЈ␣ subunit were raised in rabbits and affinity purified. The nitrocellulose membranes were incubated with 125 I-protein A (ICN), and the filters were subjected to autoradiography using Kodak X-Omat film with two DuPont Quanta III screens at Ϫ80°C. Protein concentration was measured on the S Ϫ T supernatant fraction by the method of Bradford (40), using bovine serum albumin as a standard.
Immunoprecipitation of HA-Rts1p-Yeast cell extracts were prepared as above except that Triton X-100 was added (1% final concentration) after disruption of the cells with glass beads. The Triton-soluble fraction (200 l) was incubated with 2-3 l of 12CA5 antibody on ice for 60 min. The antigen-antibody complex was harvested by the addition of 60 -100 l of a 50% protein G-agarose slurry (Life Technologies, Inc.). After gentle shaking for 60 min at 4°C, the samples were centrifuged for 10 s at 9,000 ϫ g, the supernatant (unbound fraction) was removed, and the protein G-agarose was washed five times with 500 l of 50 mM Tris-HCl, pH 7.5, 0.1 mM EDTA, 200 mM NaCl, and protease inhibitors. The beads were resuspended in 40 l of 50 mM Tris-HCl, pH 7.5, 0.1 mM EDTA, 0.2% ␤-mercaptoethanol, and used for phosphatase assays and for Western analyses. 32 P Labeling of Yeast Cells-RZ53-6⌬rts1 cells transformed with the plasmid pDB21(HA-RTS1) were grown in liquid synthetic medium without leucine to a density of 7.5 ϫ 10 7 cells/ml. The cells were then collected by centrifugation and suspended to a density of 6 ϫ 10 7 cells/ml in synthetic low phosphate medium (41). Two 25-ml cultures were incubated with shaking for 3 h at 30°C to deplete intracellular phosphate pools, after which time one flask received 4 mCi of 32 P i (DuPont NEN), and the incubation continued for 2 h. The labeled and unlabeled cells were processed as above for preparation of cell extracts and im-munoprecipitation of Rts1p in the presence of 50 mM NaF and 10 mM potassium phosphate. The immunoprecipitated material was resolved by 6% SDS-PAGE and transferred to polyvinylidene difluoride membrane. The membrane was washed with phosphate-buffered saline and blocked for 1 h with Boehringer Mannheim blocking solution. After blocking, the membrane was air dried followed by incubation for 1 h with anti-[HA]-antibody-peroxidase conjugate (kindly provided by Sophie J. Boguslawski, Boehringer Mannheim) and detection of HAtagged Rts1p by the chromogenic substrate BM Teton (Boehringer Mannheim), according to the manufacturer's specifications.
Protein Phosphatase Assay-Phosphatase activity was measured using 32 P-labeled phosphorylase a (1 mg/ml; specific radioactivity 3,000 -4,000 cpm/pmol) as a substrate (42). The assays were carried out in a 50-l reaction containing 50 mM Tris-HCl, pH 7.5, 0.1 mM EDTA, 0.2% ␤-mercaptoethanol, and 5 mM caffeine. Ten l of immunoprecipitated sample or 10 l of a 100-fold diluted extract was preincubated for 5 min at 30°C with 100 nM inhibitor-2 or 2 nM okadaic acid, before the addition of the substrate. Control reactions contained the vehicle buffers. The reactions were carried out at 30°C for 10 min and stopped by the addition of trichloroacetic acid to a final concentration of 16%. The released 32 P i was quantitated as described previously (42). To ensure linearity of the reaction, phosphate release was limited to less than 20%. One unit of phosphatase activity is defined as the amount of enzyme that releases 1 nmol of phosphate/min from the phosphorylase a substrate.
Microscopy-Yeast cells were grown in selective media, and aliquots of stationary phase cultures were applied to glass slides. The cells were examined using Nomarski optics with 400-fold magnification on a Nikon Microphot-FXA microscope, and the images were captured using a video camera connected to an Apple Macintosh computer.

RESULTS
Mammalian BЈ Subunits Suppress the Temperature Sensitivity of the rts1 Mutant-The mammalian BЈ and the yeast Rts1p proteins have ϳ55% identity and 66% similarity if conservative replacements are taken into account. Disruption of the RTS1 gene results in temperature sensitivity for growth above 37°C. The homology between the mammalian BЈ subunit and the yeast gene product prompted us to investigate whether the BЈ proteins could functionally substitute for Rts1p and rescue the temperature-sensitive growth defect of an rts1 deletion mutant. When rabbit BЈ␣ and BЈ␥ subunit cDNAs were introduced into yeast cells, the parent strain RZ53-6 grew well at 30°C, regardless of the isoform used (Fig. 1), whereas overexpression of RTS1 led to somewhat slower growth. The growth at 30°C of a mutant rts1 strain transformed with RTS1, BЈ␣, or BЈ␥ subunit DNA was similar to that of the parent RZ53-6 strain, but the mutant strain transformed with vector alone grew less well even at the permissive temperature. Approximate doubling times in liquid SC medium without leucine were 5.1 h for the mutant cells transformed with vector only and 3.4, 3.7, and 3.3 h for cells transformed with BЈ␣, BЈ␥, and RTS1, respectively. The generation time of an isogenic wild-type strain was 3.2 h.
The difference in the ability of the individual subunits to rescue the mutant strain became readily apparent when trans-formed cells were incubated at 37°C (Fig. 1). Under these conditions, pDB21(His-BЈ␣) and pDB21(HA-RTS1) restored normal growth to the mutant, whereas pDB21(HA-BЈ␥) was less effective. The approximate generation times at 37°C for the cells transformed with vector only, BЈ␣, BЈ␥, and RTS1 were 32, 9.0, 23, and 8.0 h, respectively, compared with 6.7 h for the parent strain. These results indicate that the cDNAs encoding the mammalian BЈ subunits can functionally replace the yeast RTS1 gene even though the ␣ and ␥ isoforms are not equally efficient. Analysis of cell morphology revealed that, unlike BЈ␣, BЈ␥ subunit expression resulted in abnormal multiple and elongated cells (Fig. 2), a phenotype somewhat reminiscent of that observed in cdc55 mutants (22).
The ability of the mammalian BЈ cDNAs to rescue the temperature sensitivity of the yeast mutant was paralleled by the level of expression of the proteins. As shown in Fig. 3, BЈ␣, BЈ␥, and Rts1p polypeptides could be detected by Western blotting of yeast extracts. Although the majority of each species was found in the insoluble fraction, a significant amount was present in the Triton X-100-soluble fraction. The apparent molecular weight of the proteins on SDS-PAGE, 55,000 for His-BЈ␣ and 90 -95,000 for HA-Rts1p, deviates from the predicted 60,000 and 86,000, respectively, suggestive either of posttranslational modifications or abnormal mobility caused by intrinsic properties of the polypeptides. The observation that BЈ␣ expressed in E. coli (24) migrates with an apparent molecular weight similar to that seen in yeast implies that the abnormal mobility of this isoform is probably not due to phosphorylation. Western blots of extracts prepared from rts1 mutants transformed with pDB21(HA-RTS1) revealed the presence of two or three discrete bands between 92 and 95 kDa ( Fig.  3C and Fig. 4). When cell extracts were prepared in the presence of the serine/threonine phosphatase inhibitors NaF and potassium phosphate, an upward shift in the mobility of the lower band of the Rts1p was observed (Fig. 4A). Consistent with this initial observation, 32 P labeling of yeast cells and immunoprecipitation of Rts1p showed that Rts1p is phosphorylated in vivo (Fig. 4C). Three bands were detected by immunoblotting, and all three were labeled with 32 P. The slowest migrating species contained the highest ratio of 32 P to protein, suggesting that it was the most heavily phosphorylated.
Rts1p Is Associated with PP2A Activity-Since the mammalian BЈ subunit isoforms were able to rescue the rts1 phenotype, an important question was whether Rts1p was associated with PP2A activity. Measurements of phosphorylase phosphatase activity in immunoprecipitates from extracts of yeast expressing HA-Rts1p indicated the presence of phosphatase activity that was insensitive to 100 nM inhibitor-2, a protein phosphatase 1-specific inhibitor, but almost completely inhibited by Transformants were grown on SC-leucine plates at 30°C for 2 days, then replicated and grown for 2 additional days at 30°C or 3 days at 37°C.
2 nM okadaic acid (Table I), a concentration that preferentially inhibits PP2A. Analysis by immunoblotting indicated that the monoclonal antibody 12CA5 almost quantitatively immunoprecipitated the HA-Rts1p (Fig. 5). Determination of phosphatase activity in extracts from rts1 mutant cells indicated that ϳ30% of the activity could be inhibited by 2 nM okadaic acid (data not shown). However, we do not know what proportion of this activity can associate with the yeast BЈ subunit homolog.
Rts1p Expressed in cdc55 Strains Does Not Suppress the Mutant Phenotype-Cdc55p is highly homologous to the mammalian B subunit of PP2A 1 (22). Disruption of CDC55 resulted in a cold-sensitive growth phenotype and abnormal multiple elongated cells at the restrictive temperature. Since the various B-subunits of PP2A associate with the same C2⅐A core, we examined whether Cdc55p and Rts1p could play interchangeable roles.
Attempts to transform RZ53-6 and rts1 mutant strains with the pDB21(HA-CDC55) plasmid did not yield transformants. These strains would not tolerate the presence of a high copy CDC55 plasmid, suggesting that high level of expression of Cdc55p might cause lethality in these strains. However, we do not believe this to be the case. More likely, the inability to transform these strains with pDB21(HA-CDC55) is due to the presence of the ade1 allele. Other yeast strains, such as F808 and DBY745, which harbor the same ade1 allele, could not be transformed by the multicopy CDC55 plasmid. In contrast, two unrelated strains that did not carry the ade1 allele, YNN27 and JC482, generated large numbers of transformed colonies with pDB21(HA-CDC55). Thus, it appears as though the ade1 mutation and the high copy CDC55 plasmid are unable to coexist for reasons that at this time are not understood. Perhaps high levels of Cdc55p and accumulation of the adenine precursor, phosphoribosylaminoimidazole carboxylate, combine to have a toxic effect on the cell.
The expression of Cdc55p and Rts1p in the cdc55 disruption mutant transformed with pDB20(HA-CDC55) or pDB20(HA-RTS1) was then investigated. Western blot analysis revealed that both HA-Cdc55p and HA-Rts1p were expressed in comparable amounts (Fig. 6). Cells overexpressing Cdc55p or Rts1p were grown at the permissive 30°C temperature and then transferred to 14°C. After 3 days, the cells were examined microscopically. The results demonstrated that although pDB20(HA-CDC55) restored normal morphology to the mutant cells (Fig. 7), pDB20(HA-RTS1) had no effect on the cdc55 phenotype. Therefore, even though both Rts1p and Cdc55p associate with the same C2⅐A complex, they must be involved in , the corresponding volume of the Triton-soluble (S ϩ T) and the insoluble fraction (P) were separated on a 9% SDS-polyacrylamide gel (panels A and B) or 6% SDS-polyacrylamide gel (panel C) and subjected to Western blotting. In panel A the probe was affinity-purified anti-BЈ␣ antibody, and 100 ng of recombinant BЈ␣ protein was included as a positive control. In panels B and C the probe was the12CA5 monoclonal antibody. Note the presence of two Rts1p bands in panel C. Extracts from cells transformed with pDB21 vector served as negative controls for each subunit. A protein band of ϳ48 kDa was observed in all cell extracts with the 12CA5 antibody. The nature of this cross-reacting species is unknown. The migration of molecular mass markers, expressed in kDa, is indicated to the right of each panel. 32 P labeling. Panel A, extracts from an rts1 mutant transformed with pDB21 vector or pDB21(HA-RTS1) were prepared in the absence (Ϫ) or the presence (ϩ) of phosphatase inhibitors (50 mM NaF and 10 mM potassium phosphate) as described under "Experimental Procedures," resolved by 6% SDS-PAGE, and analyzed by Western blotting. The blots were probed with 12CA5 antibody followed by autoradiography for 5, 18, and 72 h. The migration of glycogen phosphorylase (94 kDa) is indicated on the right. Panels B and C, yeast cell were 32 P labeled, and Ha-tagged Rts1p was immunoprecipitated as described under "Experimental Procedures." The immunoprecipitates from unlabeled and labeled cells were subjected to Western analysis. The immune complexes were detected colorimetrically (panel B), and the polyvinylidene difluoride membrane was subjected to autoradiography (panel C). Lanes 1 and 2, 12 l of immunoprecipitate from 32 P-labeled and unlabeled samples, respectively. different functions in the cell and cannot substitute for one another.

DISCUSSION
Several lines of evidence identify the RTS1 gene product as the yeast homolog of the BЈ subunit of PP2A 0 . First, the mammalian BЈ subunit proteins and Rts1p show a high level of amino acid homology (24). Second, two mammalian BЈ subunit isoforms are able to rescue the temperature-sensitive growth defect of the rts1 deletion strain (Fig. 1). Third, Rts1p coimmunoprecipitates with protein phosphatase activity that is insensitive to the PP1-specific inhibitor I-2 but is almost completely inhibited by okadaic acid (Fig. 5 and Table I) at a concentration (2 nM) ineffective against PP1.
The changes in the proportion of the slower mobility form of Rts1p, observed when cell extracts are prepared in the presence of protein phosphatase inhibitors, and the metabolic 32 P labeling (Fig. 4) clearly indicate that Rts1p is phosphorylated in vivo. The presence of at least three phosphorylated species of different electrophoretic mobility suggests that the protein is multiply phosphorylated at a minimum of three sites. Whereas the catalytic subunit of PP2A is known to be modified directly by phosphorylation (43) and by carboxyl methylation (44,45), the present paper provides evidence that the yeast BЈ subunit undergoes post-translational modification. Similar conclusions were recently reached for the mammalian isoforms (46). We do not know at this time the functional significance of the phosphorylation. However, control of activity by phosphorylation of regulatory subunits is well established for PP1 (4,47,48). Possibly a similar mechanism operates to control PP2A 0 activity and/or subcellular localization.
Expression of mammalian BЈ␣ and BЈ␥ subunits rescues the growth defect of the rts1 mutant at both 30 and 37°C. BЈ␣ restores growth completely, up to the wild-type level, whereas BЈ␥ only provides partial complementation. It is unlikely that this is the result of a different degree of expression of BЈ␥ protein, since Western analysis indicated that the polypeptide is present at a level comparable to that of Rts1p (Fig. 3). These results suggest that the two mammalian BЈ subunit isoforms may perform somewhat different cellular roles. The two isoforms are ϳ70% homologous; and interestingly the BЈ␥, but not the BЈ␣, polypeptide contains in its COOH-terminal region a putative bipartite nuclear localization signal that could direct the protein to the nucleus (49,50). Indeed, recently it has been reported (46) that the mammalian BЈ␥ but not ␣ can localize to the nucleus. Rts1p has been shown to reside in the cytoplasm (30). The potential nuclear localization of the BЈ␥, which would increase PP2A activity in the nucleus, may also be responsible for the abnormal morphology of the cells overexpressing the protein (Fig. 2), either by affecting nuclear functions or by depleting other pools of the enzyme.
A clearer functional distinction exists between Cdc55p, the B FIG. 5. Immunoprecipitation of Rts1p. HA-tagged Rts1p was immunoprecipitated from extracts of rts1 cells overexpressing the protein or transformed with the vector only. Two hundred l of Triton-solubilized extract was reacted with 3 l of 12CA5 antibody, and the complex was absorbed onto protein G-agarose as described under "Experimental Procedures." Aliquots of the extract (S) and the unbound (U) and the immunoprecipitated (IP) material were resolved by SDS-PAGE and subjected to Western blotting. Most of the HA-Rts1p was found in the immunoprecipitated fraction. The cross-reacting 48-kDa polypeptide described in Fig. 3 was present in the extract and unbound fractions, but not in the immunoprecipitated complex, indicating that the native form of this protein is not recognized by the 12CA5 antibody.  7. Effect of B and B subunit expression on cdc55 phenotype. The cdc55 mutant was transformed with pDB20 vector alone (panel A), pDB20(HA-RTS1) (panel B), and pDB20(HA-CDC55) (panel C). The transformants were patched on a SC-uracil plate and incubated at 14°C for 3 days. Cells were scraped from the plate, resuspended in water, and examined using Nomarski optics with 400-fold magnification. Images were captured as described under "Experimental Procedures." subunit of PP2A 1 , and Rts1p. Overexpression of Rts1p cannot suppress the cold-sensitive phenotype of a cdc55 mutant. The cells grown at 14°C remain defective in septation and cytokinesis (Fig. 7). Therefore, in keeping with the idea of spatial or functional targeting, the BЈ and B subunits of PP2A must channel the same phosphatase toward different cellular substrates or locales. Furthermore, the somewhat slower growth observed in wild-type RZ53-6 cells overexpressing RTS1 may be suggestive of enrichment of the PP2A holoenzyme containing Rts1p at the expense of other essential pools. Both the mammalian and yeast PP2A may be involved in cell cycle control. Carboxyl methylation of the catalytic subunit of PP2A has been reported to correlate with cell cycle progression (51), and the activity of a microtubule-associated PP2A fluctuates as the cells traverse the cell cycle (28). The B subunit of Drosophila PP2A was shown to be required for correct chromatid migration in the anaphase (52). Mutations in both the yeast catalytic subunit gene (13) and B subunit gene (22) result in cell cycle defects, although the two display dissimilar morphological aberrations: multiple elongated buds in the cdc55 mutant at the restrictive temperature and small, deformed buds in the pph21 mutant. The rts1 strain cells at either 30 or 37°C appear to be significantly enlarged, again indicating a distinct function of the BЈ subunit.
Mutations in the CDC55 and RTS1 genes result in phenotypes similar to those displayed by A subunit defects. Mutations in the yeast TPD3 gene, encoding the regulatory A subunit of PP2A, result in both cold-and temperature-sensitive phenotypes (17). At 13°C, the cells resemble cdc55 mutants in that they show elongated buds, consistent with defective cytokinesis. The same mutation does not allow growth at 37°C, a phenotype analogous to that of the rts1 mutant. The ability of one mutation to confer a dual phenotype is consistent with the regulatory A subunit being a component of different phosphatase holoenzymes containing either the Cdc55p or the Rts1p.
Several lines of evidence point toward an involvement of RTS1 in modulating cellular responses to stress conditions. First, overexpression of RTS1 suppresses a rox3 temperature mutation (29) and several hsp60-ts alleles (30), both genes known to respond to stress. Hsp60p resides in mitochondria, and Rox3p is localized in the nucleus but affects the level of expression of a mitochondrially located protein, iso-2-cytochrome c. The fact that RTS1 is required for expression of the mitochondrial chaperonins Cpn10p and Mge1p (30), which are defective in the hsp60-ts strains, suggests that this protein phosphatase subunit plays a role in regulating the expression of genes whose products are destined for mitochondria and/or involved in stress responses. Supporting this notion is the observation that, although RTS1 is not an essential gene, its disruption not only impairs growth at elevated temperature but also decreases utilization of nonfermentable carbon sources, a feature consistent with impaired mitochondrial function.
Second, disruption of the RTS1 gene itself confers temperature sensitivity and results in elevated expression of the CYC7 gene in response to osmotic stress and heat shock (29). These responses may occur through a common stress response element (53). The osmotic stress response is mediated, at least in part, by the activation of the Hog1p protein kinase pathway (54 -56). Thus, we infer that the protein kinases and PP2A 0 may share common substrates or that they may be targets for each other. Other protein phosphatases, PTP2 and PP2C, have also been implicated in the control of the osmosensing mitogenactivated protein kinase pathways in S. cerevisiae (56) and in Schizosaccharomyces pombe (57) In conclusion, the present study demonstrates that the RTS1 gene encodes the S. cerevisiae homolog of the mammalian BЈ subunits of PP2A 0 and that the Rts1p is phosphorylated in vivo. Mammalian BЈ␣ and, to a lesser extent BЈ␥, are able to complement the temperature-sensitive defect of rts1, but RTS1 cannot replace CDC55, the yeast homolog of the B subunit of PP2A 1 . Thus, the B-subunits are not redundant but confer functional specificity to the various forms of PP2A.