The Scaffolding A/Tpd3 Subunit and High Phosphatase Activity Are Dispensable for Cdc55 Function in the Saccharomyces cerevisiae Spindle Checkpoint and in Cytokinesis*

Protein serine/threonine phosphatase 2A (PP2A) is a multifunctional enzyme whose trimeric form consists of a scaffolding A subunit, a catalytic C subunit, and one of several regulatory B subunits (B, B (cid:1) , and B (cid:2) ). The adenovirus E4orf4 protein associates with PP2A by directly binding the B or B (cid:1) subunits. An interaction with an active PP2A containing the B subunit, or its homologue in yeast, Cdc55, is required for E4orf4-induced apoptosis in mammalian cells and for induction of growth arrest in Saccharomyces cerevisiae . In this work, Cdc55 was randomly mutagenized by low-fidelity PCR amplification, and Cdc55 mutants that lost the ability to transduce the E4orf4 toxic signal in yeast were selected. The mutations obtained by this protocol inhibited the association of Cdc55 with E4orf4, or with the PP2A-AC subunits, or both. Functional analysis revealed that a mutant that does not bind Tpd3, the yeast A subunit, as well as wild type Cdc55 in a tpd3 (cid:1) background, can form a heterodimer with the catalytic subunit. This association requires C subunit carboxyl methylation. The residual phosphatase activity associated with Cdc55 in the absence of Tpd3 is sufficient to maintain a partially synthetic medium 2% raffinose were A 600 0.3 and to another 2 h before addition of galactose 2%. in 2% to mid-log phase, and 2% galactose. in this are isogenic strain W303, KY520

and BЉ/PR72/PR130/PR59/PR48 (3). The core PP2A enzyme has also been shown to bind other cellular proteins, including striatin and SG2NA (4). Viral proteins, such as the SV40 small t antigen and the polyomavirus small and middle T antigens, can replace the cellular B subunits (5). The various cellular PP2A B subunits target the PP2A holoenzyme to different substrates and dictate its subcellular localization (reviewed in Ref. 3).
In Saccharomyces cerevisiae, two closely related genes, PPH21 and PPH22, redundantly encode the major PP2A catalytic subunit (6). TPD3 encodes the only A subunit, and two distinct B subunits, encoded by CDC55 and RTS1, are highly homologous to mammalian B and BЈ, respectively (7)(8)(9). Mutations of CDC55 are viable, but yield defects in cytokinesis and in the spindle checkpoint and result in abnormal cell morphology. Methylation of the C-terminal leucine residue of the PP2A catalytic subunit by a specific methyltransferase increases the affinity of this subunit for the regulatory subunits B/Cdc55 and BЈ/Rts1 and for Tpd3 both in vitro and in vivo (10 -13). Yeast Cdc55 and mammalian PP2A-B55 subunits share an extensive homology (53% identity and 67% similarity) (7), and the middle third of both proteins shows 83% similarity. Members of the B subunit family have been suggested to contain four to seven WD repeats (14,15).
Adenovirus E4orf4 protein is a multifunctional viral regulator. It down-regulates expression of genes that have been activated by E1A and cAMP (16,17), induces hypophosphorylation of various viral and cellular proteins (16,18), regulates alternative splicing of adenovirus mRNAs (18), and induces p53independent apoptosis in transformed cells (19 -21). E4orf4induced apoptosis is specific to transformed cells (22), and thus E4orf4 may have a therapeutic potential for cancer.
We have previously shown that E4orf4 interacts with the PP2A holoenzyme through a direct association with the B␣/B55 subunit (17), and can also bind PP2A complexes containing BЈ/B56 (23). We, and others, have recently shown that the interaction with PP2A complexes that include the B␣/B55, but not the BЈ/B56 subunit, is required for induction of apoptosis by E4orf4 (22)(23)(24). Binding the B subunit alone is not sufficient for induction of apoptosis, because an E4orf4 mutant, A3 (S95P), could bind the B subunit, whereas the A and C subunits were excluded from the complex, causing A3 to act as a dominant negative mutant (23). The presence of an active PP2A also contributes to other functions performed by the E4orf4 protein, including down-regulation of transcription (17,25,26) and alternative splicing (18).
We have recently demonstrated that E4orf4 induces PP2Adependent growth arrest in S. cerevisiae, by targeting PP2A to the anaphase-promoting complex/cyclosome (APC/cyclosome) and inhibiting the activity of the APC/C in mitosis (27). Our results, together with previously reported work, further suggested that E4orf4 could be enhancing a physiological interaction of PP2A with the APC/C. Thus, the work in yeast indicated that E4orf4 can be used as a tool to further characterize PP2A function and regulation. We have also shown that the yeast system can be utilized to select for non-apoptotic mutants of E4orf4, which do not bind PP2A (28). These results indicate that a biological selection assay based on E4orf4 toxicity in yeast can identify mutations in the E4orf4-PP2A complex, which affect interactions within the complex.
In this work we designed a protocol to select for Cdc55 mutants that lost the ability to transduce the E4orf4 toxic signal. These mutants define residues in Cdc55 required for the interaction with the core enzyme and with E4orf4. Functional characterization of the mutants and the WT protein revealed that a low level of binding of the regulatory B/Cdc55 subunit to the catalytic C subunit is retained in the absence of the scaffolding A/Tpd3 subunit. The minimal PP2A activity associated with this complex is sufficient for maintaining a partially active spindle checkpoint and an apparently normal cytokinesis.

EXPERIMENTAL PROCEDURES
Yeast Strains and Media-Yeast cells were grown either in YPD (1% yeast extract, 2% Bactopeptone, 0.015% L-tryptophan, 2% glucose) or in synthetic (SD) medium (29). Rapamycin was added to solid medium to a final concentration of 100 nM from a stock of 1 mM in 100% ethanol. Benomyl was added to solid medium to a final concentration of 25 g/ml from a stock of 3 mg/ml in dimethyl sulfoxide (Me 2 SO). Nocodazole was added to liquid medium to a final concentration of 10 g/ml from a stock of 2 mg/ml in Me 2 SO. ␣-Factor was added to liquid medium to a final concentration of 8 g/ml from a stock of 2 mg/ml. For induction with galactose, cells were grown in synthetic medium with 2% raffinose overnight. They were then diluted to A 600 0.3 and allowed to grow another 2 h before addition of galactose to 2%. Alternatively, cells were grown in 2% glucose to mid-log phase, washed once, and resuspended in 2% galactose. The yeast strains used in this study are isogenic with strain W303, and include KY520 (cdc55⌬) (27), Y2745 (cdc55⌬,ppm1⌬) (12), and KY679 (tpd3⌬) (27).
Mutagenesis-The Cdc55 open reading frame has been mutagenized at random by PCR amplification of CDC55 in the presence of 0.03 mM MnCl 2 , as previously described (22). The forward primer was CTC-CAAGCTTCTGTCTACC, and the reverse primer was ACATTATTCT-GGAATCC. Plasmid YCp50-CDC55 has been digested with SacI and MluI. The fragment containing the vector and sequences lying 916 bp upstream of the CDC55 coding frame and 330 bp downstream of it (defined as the linear vector) was gel-purified, as were the PCR products. The PCR products share homologous sequences with the vector, which include 120 bp upstream of the CDC55 open reading frame (positions Ϫ1036-Ϫ916 relative to the first ATG) and 334 bp downstream of it (1910 -2244). The linear vector and PCR products were transformed into KY520 yeast cells containing p414GALL-E4orf4, where they underwent homologous recombination to produce a circular plasmid, as previously described (32). Selection for E4orf4-resistant clones was performed on galactose plates. Site-directed mutagenesis was performed using the QuikChange XL site-directed mutagenesis kit (Stratagene, La Jolla, CA), according to the manufacturer's instructions.
Isolation of Plasmid DNA from Yeast-The yeast cells containing p414GALL-E4orf4 and mutant YCp50-CDC55 were grown 1 day on medium containing tryptophan but lacking uracil, and then plated on plates lacking uracil and containing 2-amino-5-fluorobenzoic acid to select for clones that lack the plasmid encoding the TRP marker (33). Plasmid DNAs were extracted from the clones obtained on 2-amino-5fluorobenzoic acid as previously described (34).
Immunoprecipitations and Phosphatase Assays-For co-immunoprecipitations, yeast extracts were prepared by bead beating cells for 4 min at 4°C in lysis buffer (250 mM NaCl, 50 mM Tris-HCl, pH 7.4, 5 mM EDTA, 0.5% Nonidet P-40, 0.1% Triton X-100, 50 mM NaF, 100 M sodium vanadate, 1ϫ complete protease inhibitor mixture (Roche Applied Science)). The lysates were spun to separate beads and debris from the clear lysate. The beads were washed twice more in the lysis buffer and immunoprecipitations were carried out in the same lysis buffer. Lysates and antibodies were rotated for 2 h at 4°C. Protein A beads (Amersham Biosciences) were added and rotated for an additional hour. The beads were then washed three times in lysis buffer.
Immunoprecipitation phosphatase assays have been previously described (22). Briefly, immune complexes were prepared as described above and were washed twice more in PP2A buffer (50 mM imidazole, pH 7.2, 0.2 mM EGTA, 0.02% ␤-mercaptoethanol, 0.1% bovine serum albumin). The immune complexes were incubated with 50 nmol of peptide substrate RRA(pT)VA (35) in PP2A buffer for 60 min at 30°C. The amount of free phosphate released was measured by a color reaction generated in the presence of molybdate and malachite green. Antibodies used in this work were: anti-E4orf4 (19), anti-Tpd3 (from J. R. Broach), anti-Cdc55 antibodies, raised against a glutathione Stransferase-Cdc55 fusion protein, and anti-HA (Covance, Berkeley, CA).

Selection for Cdc55 Mutants That Cannot Mediate E4orf4
Toxicity in Yeast-The protocol used to select for CDC55 mutants that have lost the ability to transduce the E4orf4 toxic effect was based on the findings that E4orf4 inhibits growth of WT yeast but not of cdc55⌬ cells, and that ectopic expression of a WT CDC55 gene in the cdc55⌬ background restores E4orf4 toxicity (27,36). It also relies on the finding that E4orf4induced cell death requires an interaction between E4orf4 and an active PP2A, containing both catalytic and regulatory subunits (22,23). The mutants selected by this protocol may be deficient in their ability to bind either E4orf4 or the PP2A-AC heterodimer. To obtain cdc55 mutants, the CDC55 gene was mutated by a low-fidelity PCR-based protocol, as described previously (22). The mutated PCR products and the linear vector, described under "Experimental Procedures," were transformed into cdc55⌬ yeast cells containing p414GALL-E4orf4. Homologous recombination in the transformed yeast cells produced circular plasmids containing the mutated PCR products (32) (Fig. 1). Transformants that grew on galactose (i.e. resistant to E4orf4) presumably contained a mutated Cdc55 protein that could not transduce the E4orf4 signal. Growth of the transformants on galactose was compared with that of cdc55⌬ cells containing an empty vector or the vector expressing WT Cdc55. Only clones that grew as well as the vector-containing yeast were further analyzed. In the next stage, clones that became resistant to E4orf4 toxicity because of decreased E4orf4 expression levels or lack of expression of Cdc55 were identified by immunoprecipitations and Western blots, and discarded. Plasmid DNAs were extracted from the remaining yeast clones, and used in a second cycle of transformation and plating on galactose, in the presence of E4orf4, to confirm that they conferred resistance to the viral protein.
In the initial screen, 71 E4orf4-resistant colonies were obtained from 2,306 transformed yeast cells. 40 colonies of the 71 expressed high levels of E4orf4, as detected by Western blots. Of these colonies, only 12 expressed Cdc55 at more than 50% WT levels (clones 15, 19, 28, 29, 37, 68, 1, 66, 73, 81, 110, and 126) and three more expressed lower levels of Cdc55 (38, 41, and 87). Plasmid DNAs were extracted from the 15 resistant colonies expressing E4orf4 and Cdc55, and were tested for their ability to confer resistance to E4orf4 in a second cycle of transformation. All of them were indeed unable to mediate E4orf4 toxicity.
Analysis of the Ability of Mutant Cdc55 Proteins to Bind E4orf4 and the PP2A Tpd3 Subunit-Yeast clones containing the plasmids encoding E4orf4 and Cdc55 were induced on galactose and Cdc55 was immunoprecipitated from the yeast extracts. The presence of Cdc55, Tpd3, and E4orf4 in the immune complexes was detected using Western blots. In some experiments E4orf4 was immunoprecipitated and the presence of E4orf4 and Cdc55 in the immune complexes was investigated. As seen in Fig. 2 and Table I, the mutant Cdc55 proteins appeared to fall into four categories: (a) Cdc55 mutants that bind Tpd3, but not E4orf4 (clones 1, 28, 68, and 81); (b) Cdc55 mutants that bind E4orf4, but not Tpd3 (clone 126); (c) Cdc55 mutants that bind neither Tpd3 nor E4orf4 (clones 15, 19, 29, 37, 66, 73, and 110 (S309P)); and (d) Cdc55 mutants that are expressed at low levels, but bind Tpd3 at relatively high levels (clones 38, 41, and 87). The binding experiments were repeated three times, and binding and expression levels were quantified by densitometry. The average binding levels obtained in these experiments are presented in Table I.
Sequence Analysis of the Mutants-The Cdc55 mutants were subjected to sequencing, and the results are shown in Table I. Because several of the mutants contained more than one mu-tation, a few substitutions were introduced as single mutations into the Cdc55 protein by site-directed mutagenesis. The additional mutants obtained by this procedure were tested for their ability to bind E4orf4 and Tpd3, and the results are summarized in Fig. 2C and Table I.
Binding of the Mutants to an Active PP2A-To determine whether the mutant Cdc55 proteins associated with an active PP2A-AC heterodimer, phosphatase assays were carried out, using the Cdc55 immune complexes and a phosphorylated PP2A peptide substrate, as previously described (22,28). The results shown in Fig. 3 demonstrate that most Cdc55 mutants that bound intermediate to high levels of Tpd3 (greater than 33% of WT levels), bound intermediate to WT levels of PP2A activity (more than 30% of WT levels). One clone associated consistently with 2-fold higher phosphatase activity compared with the WT Cdc55 protein (E290A). Clones that were severely deficient in Tpd3 binding (bound less than 15% WT levels) were also associated with dramatically reduced levels of PP2A activity (less than 20% of WT levels, clones 73, 66, 15, 37, 126, D262G, S368P, K118E, L102P, and S309P). However, two clones that bound Tpd3 well (clones 41 and 68 bound Tpd3 at more than 50% WT levels), appeared to bind low levels of an active PP2A core enzyme (around 15% WT levels).
Functional Analysis of Single Substitution Mutants-To test whether the various mutations affected Cdc55 functions, we examined their influence on two of the known phenotypes associated with CDC55. cdc55⌬ mutants demonstrate defective spindle checkpoint activity (37,38), which could result from a diminished ability to promote dephosphorylation of Cdc28 (37), and from enhanced derepression of the APC/C activity in the mutant cells, leading to a less effective inactivation of the APC/C by the spindle checkpoint pathway (27). In addition, Cdc55 antagonizes the Tor signaling pathway, and cdc55⌬ mutants exhibit rapamycin resistance as well as an enhanced association between the Tap42 and the C subunit of PP2A (39).
To test whether the various mutants were associated with a checkpoint defect caused by loss of CDC55, we examined their sensitivity to benomyl. Strains defective in the spindle checkpoint pathway are more sensitive to low levels of benomyl, a microtubule-depolymerizing agent, probably because of an increased proportion of cells that go through mitosis in the absence of an intact spindle (40,41). To simplify analysis, each class of cdc55 mutations to be examined was represented only by mutants with single amino acid substitutions. We expected that mutants, which were dramatically reduced in their ability to bind an active PP2A, would lose their resistance to sublethal concentrations of benomyl, whereas mutants associating with WT PP2A levels would retain benomyl resistance. Surprisingly, three mutants that associated with 15% or less of WT PP2A activity, L102P, K118E, and to a lesser extent S309P, were almost as resistant to benomyl as WT Cdc55. One mutant, D262N, which was associated with WT levels of PP2A activity was not as resistant to benomyl as the WT yeast (Fig. 4).
The macrolide drug rapamycin inhibits yeast growth by inhibiting essential activities of Tor kinase that are mediated at least partially through Tap42. cdc55⌬ strains are resistant to rapamycin, because Cdc55 is required to reverse Tor-mediated phosphorylation of Tap42, and in its absence even low levels of Tor kinase activity are sufficient for cell growth (39). The various single substitution mutants were grown in the presence of 100 nM rapamycin, a concentration to which the WT strain is sensitive, but on which the cdc55⌬ strain can grow. As shown in Fig. 4 mutants Y344A and Y344C that bind WT levels of phosphatase activity are as sensitive to rapamycin as the WT strain. Mutant D262N is rapamycin-resistant despite its ability to associate with WT levels of PP2A activity. Mutants that were associated with low levels of PP2A activity were more resistant to rapamycin than WT yeast. However, K118E was less resistant than other mutants retaining low level association with the active phosphatase, such as L102P and S309P.
The finding that mutant D262N, which binds WT PP2A activity levels, behaves like a cdc55⌬ mutant may be explained by suggesting that this mutant cannot maintain an active conformation of the phosphatase toward substrates that are larger than the peptide used in the dephosphorylation assays. Alternatively, it may be unable to interact with physiological substrates of the enzyme. However, the finding that mutants K118E and S309P, which do not bind Tpd3 (Fig. 2) and are associated with a low phosphatase activity (15% of WT levels, Fig. 3), behave like WT Cdc55 when grown on benomyl, is much harder to explain. These results suggest that the scaffolding A/Tpd3 subunit and a high phosphatase activity may be dispensable for the activity of Cdc55 in the spindle checkpoint.
Cdc55 Maintains a Partially Active Spindle Checkpoint in the Absence of Tpd3-To test whether WT Cdc55 can maintain an active spindle checkpoint in the absence of Tpd3, we examined the viability of a tpd3⌬ strain in the presence of nocodazole, another microtubule-depolymerizing agent, and compared it to the viability of WT and cdc55⌬ yeast cells. Because the various strains differ greatly in their growth properties, we could not directly compare their growth on benomyl. Cells were arrested at G 1 by treatment with ␣-factor and then released from the G 1 block in the presence of nocodazole. Samples were removed at various times after release and analyzed for viability by a colony formation assay. Cells with an intact spindle checkpoint arrest in mitosis and retain viability, whereas cells defective in the spindle checkpoint proceed through mitosis in the absence of a spindle and lose viability (12). As shown in   3. Association of the Cdc55 mutants with an active PP2A. A, binding levels of Tpd3 by the various Cdc55 proteins were quantified by densitometry of gels represented in Fig. 2. Tpd3 binding by the WT protein was defined as 100% in each blot, and relative binding of Tpd3 by the mutants was calculated as percent of WT. Each bar represents the average of 2-4 blots. Error bars represent the S.D. B, Cdc55 proteins were immunoprecipitated with specific antibodies and phosphatase activity associated with the immune complexes was measured by an enzymatic reaction, using a phosphorylated peptide as a substrate. The level of phosphate release was measured by a colorimetric reaction, and relative phosphatase activity was determined as a percentage of WT Cdc55-associated activity. These experiments were repeated three times and error bars denote the S.D.
FIG. 4. Cdc55 mutants associated with low PP2A activity are resistant to benomyl. cdc55⌬ yeast cells transformed with the indicated Cdc55-expressing plasmids were grown overnight in the appropriate liquid medium. They were then serially diluted 5-fold and 5 l of cells in suspension were spotted onto glucose plates containing: no drug (control), 25 g/ml benomyl, or 100 nM rapamycin. The plates were incubated 2 days (no drug) or 3 days (with benomyl or rapamycin) at 30°C.

FIG. 5. A partially functional spindle checkpoint is maintained in the absence of Tpd3.
Exponentially growing cultures of W303-1A (wild type), tpd3⌬, and cdc55⌬ cells were incubated in the presence of ␣-factor for 3 h at 30°C and then transferred into fresh YPD medium containing 10 g/ml nocodazole. At different times after release from cell cycle arrest, cells were removed, sonicated, counted, and plated on YPD plates. Viable colonies were counted after 2 days. Triangles, wild type; diamonds, tpd3⌬; squares, cdc55⌬. absence of Tpd3 at higher than background levels, although binding is greatly diminished compared with Cdc55-HA-Pph22 association in the presence of Tpd3.
Ppm1 Is Required for the Spindle Checkpoint Function of Cdc55 and for the Tpd3-independent Interaction of Cdc55 with the PP2A Catalytic Subunit-It has been well documented that carboxyl methylation of the PP2A catalytic subunits, including yeast Pph21 and Pph22, is required for their efficient interaction with the A/Tpd3 and B/Cdc55 subunits. Deletion of the S. cerevisiae PPM1 gene, encoding a methyltransferase that modifies the PP2A catalytic subunits, leads to reduced binding of Cdc55 and Tpd3 to PP2A-C and increases benomyl sensitivity of yeast cells (12,13). To determine whether Ppm1 is required for the Tpd3-independent checkpoint activity of Cdc55, we tested the benomyl sensitivity of two Cdc55 mutants that cannot bind Tpd3 (126 and K118E) in the ppm1⌬ genetic background. As demonstrated in Fig. 7A, cdc55⌬ppm1⌬ yeast cells expressing Ppm1 and the K118E and 126 mutations were as resistant to 25 g/ml benomyl as were cells expressing Ppm1 with WT Cdc55. However, resistance was completely lost when the mutants were expressed in the absence of Ppm1. Both mutants behaved like a cdc55⌬ mutant on rapamycin. These results suggest that Ppm1 is required for the Tpd3-independent checkpoint activity of Cdc55.
To determine whether C subunit methylation was required for the Tpd3-independent association of Cdc55 with Pph22, WT Cdc55 and mutant 126 were introduced into a cdc55⌬ppm1⌬ yeast strain expressing HA-Pph22 together with an empty vector or the vector expressing Ppm1. Fig. 7B demonstrates that binding of both WT and mutant Cdc55 proteins to HA-Pph22 was reduced in the ppm1⌬ genetic background, as compared with their binding in yeast cells expressing Ppm1.
These results are consistent with the conclusion that mutant 126 retains resistance to sublethal benomyl levels by binding low levels of the catalytic subunit in the absence of Tpd3. In contrast with their activity in the spindle checkpoint, the low phosphatase levels associated with the mutants appear not to be sufficient for the role of Cdc55 in inhibition of TOR-mediated signaling.
Defects in Tpd3 Binding Are Not Associated with Morphological Defects of cdc55 Mutations-cdc55⌬ cells show defects in cytokinesis, often yielding multinucleate elongated cells that possess pronounced constrictions along their lengths (7). To investigate whether Cdc55 mutants that are unable to bind Tpd3 show this morphological defect, exponentially growing cultures of cdc55⌬ cells expressing WT Cdc55, the K118E and 126 mutants, or no Cdc55 proteins were harvested and cells were visualized by light microscopy. As seen in Fig. 8, the mutant cells are not associated with the cdc55 defect, and have FIG. 6. Cdc55 binds the PP2A catalytic subunit in the absence of Tpd3. Extracts from wild type, cdc55⌬, and tpd3⌬ cells transformed with HA-PPH22 or untransformed wild type cells were immunoprecipitated with anti-Cdc55 antibodies. Immunoprecipitates (IP) or extracts before precipitation (lysate) were separated by SDS-PAGE and then transferred to polyvinylidene difluoride membranes. The membranes were probed sequentially with anti-HA, anti-Cdc55, and anti-Tpd3 antibodies.
FIG. 7. Ppm1 is required for the Tpd3-independent spindle checkpoint function and for the interaction of Cdc55 with the PP2A catalytic subunit. A, cdc55⌬ppm1⌬ yeast cells transformed with PPM1 or an empty vector together with wild type CDC55, an empty vector, or mutants K118E and 126 were grown overnight in the appropriate liquid medium. Cultures were then serially diluted 5-fold and 5 l of cells in suspension were spotted onto glucose plates containing: no drug (control), 25 g/ml benomyl, or 100 nM rapamycin. The plates were incubated 2 days (no drug) or 3 days (with benomyl or rapamycin) at 30°C. B, cdc55⌬ppm1⌬ yeast cells were transformed with PPM1 or an empty vector, with a plasmid expressing HA-Pph22 or an empty vector, and with wild type CDC55, an empty vector, or mutant 126. Protein extracts were immunoprecipitated (IP) with anti-Cdc55 antibodies. Immunoprecipitates or extracts before precipitation (lysate) were separated by SDS-PAGE and then transferred to polyvinylidene difluoride membranes. The membranes were probed sequentially with anti-HA, anti-Cdc55, and anti-Tpd3 antibodies. a WT-like appearance. Thus, regulation of cytokinesis can be maintained by Cdc55 proteins that are unable to bind Tpd3.
Based on our results we conclude that Cdc55 can interact with the PP2A catalytic subunit in the absence of Tpd3, albeit at low levels. The residual phosphatase activity associated with Cdc55 under these conditions is sufficient to maintain some of the Cdc55 functions.

DISCUSSION
In this work we used a selection protocol based on PP2A-dependent E4orf4-induced growth arrest in S. cerevisiae to find biologically interesting Cdc55 mutants. The characterization of these mutants revealed the surprising and novel finding that Cdc55 can associate with the catalytic PP2A C subunit in the absence of the scaffolding A/Tpd3 subunit. Furthermore, the low levels of phosphatase activity associated with these complexes are sufficient to maintain a partially active spindle checkpoint and an apparently normal cytokinesis.
Types of Mutants Identified by the Selection Protocol-E4orf4 induces PP2A-B/Cdc55 subunit-dependent growth arrest and cell death in both yeast and mammalian cells (22-24, 27, 36). Furthermore, E4orf4 must interact with an active PP2A enzyme to induce cell death (22,23). Thus selection of Cdc55 mutants that are unable to transduce the E4orf4-induced Cdc55dependent growth arrest in yeast was expected to identify residues in the yeast Cdc55 subunit involved in its interaction with the PP2A core heterodimer and with E4orf4, as well as mutations that inhibit expression or destabilize the Cdc55 protein. The second type of mutations has not been further studied in this work.
Cdc55 mutations obtained by the approach described here can be divided into 4 classes. Type 1 mutations reduce E4orf4 binding to Cdc55, but do not interfere dramatically with Tpd3 binding. Type 2 mutations affect Tpd3, but not E4orf4 binding. Type 3 mutations interfere with both E4orf4 and Tpd3 binding, and type 4 mutations bind enhanced levels of Tpd3 compared with their expression levels. Because many of the mutants obtained in the screen contained more than one amino acid substitution, single substitutions were introduced by site-di-rected mutagenesis to determine more precisely which residues were responsible for disrupting the Cdc55 interactions.
The selection protocol described here yielded only one type 2 mutation that interfered with Tpd3, but not with E4orf4 binding (clone 126). A single substitution mutant (K118E), which was derived from mutant 126 and shares similar properties with it, contains a charge-reversal mutation in the inter-repeat loop between WD2 and WD3. Similar charge reversal mutations in WD3, WD4, and the inter-repeat loop between them have been previously described to affect the interaction between B␥ and the AC subunits of mammalian PP2A (14).
Several mutations belong to class 3 and interfere with both E4orf4 and Tpd3 binding. Some of these mutants have two substitutions, both of which contribute to the altered interactions. For example, clone 37 contains two substitutions, F33S and V514D. F33S reduces mildly the interactions with Tpd3 and E4orf4. V514D reduces Tpd3 binding to intermediate levels, and reduces significantly E4orf4 binding. The combined effect of both substitutions leads to complete inhibition of binding to both Cdc55 binding partners. In addition to mutants containing substitutions that additively affect Cdc55 interactions, we also found single substitutions that affected binding to both proteins. These include L102P, D262G, S309P, and S368P. These substitutions introduce into the polypeptide amino acids, such as glycine and proline, that may contribute to more dramatic changes in local conformation, thus altering surfaces involved in binding both Tpd3 and E4orf4. Indeed, mutation D262N reduces mostly E4orf4 binding, and affects Tpd3 binding only mildly, whereas substitution D262G completely inhibits both interactions. These results suggest that D262 may be directly involved in E4orf4 binding, and that adjacent residues, affected only by a more drastic change in local conformation induced by the glycine substitution, could be required for the interaction with Tpd3.
The fourth class of mutants includes three mutants that are expressed weakly, but appear to bind relatively enhanced levels of Tpd3. These mutants contain substitutions in several residues that are not conserved between yeast and mammalian cells, and are located toward the C terminus of Cdc55. Thus some mutations in Cdc55 may lead to changes in the protein that allow a higher affinity of interaction with PP2A-AC. The fact that evolution did not favor such alterations suggests that some feature of PP2A control could involve quick association and disassociation of different regulatory subunits and the core enzyme. Indeed, it has been shown that methylation of the PP2A-C subunit, which is required for association of the B and BЈ subunits with the core enzyme (11)(12)(13)43), is altered during the cell cycle (44). These findings indicate that the B subunit content of the heterotrimeric enzyme may be subject to cell cycle control. In another example, the composition of PP2A heterotrimers was transiently altered during the initial stage of retinoic acid-induced granulocytic differentiation (45).
The physical interaction between Cdc55 and Tpd3, detected by Western blots, and the interaction between Cdc55 and an active PP2A-AC core enzyme, measured by immunoprecipitation phosphatase assays, are generally in agreement (Figs. 2  and 3). Mutations inhibiting the interaction with Tpd3, or reducing significantly Cdc55 expression led to significantly reduced levels of Cdc55 association with an active PP2A. Mutations that affect E4orf4 binding, but not Tpd3 binding, or those that reduce binding to Tpd3 only mildly, do not significantly affect the interaction with an active PP2A. However, two mutants, clones 41 and 68, which bind Tpd3 efficiently (more than 50% of WT levels), showed a more significant inhibition of association with an active PP2A (15% of WT levels). These mutants may have lost the ability to interact with the C subunit, and as a result may be unable to form a stable and functional heterotrimeric complex. In contrast, mutation E290A appears to be consistently associated with a 2-fold higher PP2A activity compared with WT Cdc55, although it binds normal levels of Tpd3. This mutation possibly enhances the interaction of Cdc55 with the C subunit.
All of the single substituted amino acids that affect E4orf4 or Tpd3 binding are identical or similar to the corresponding amino acids in the mammalian B␣ subunit. Indeed, a representative mutation from each class of mutations was introduced into the mammalian B␣ subunit and was shown to confer binding properties similar to those demonstrated in yeast (results not shown).
It has been suggested that monomeric subunits of PP2A, found outside the heterotrimer, are unstable in both Drosophila (46,47) and mammalian cells (14). However, our results indicate that in yeast cells this is not always the case, because several Cdc55 mutants that were unable to integrate into the PP2A heterotrimer were expressed at WT levels (for example, clones 37, 66, 126, and K118E). These results are in agreement with previous reports showing that when individual subunits cannot participate in PP2A heterotrimer formation in S. cerevisiae their expression levels remain unaltered (12,13,48,49).
PP2A Can Function as a Complex Consisting of Cdc55 and the Catalytic Subunit-When the mutants obtained in the E4orf4-based selection process were tested to find if they retained Cdc55 functions, it appeared that mutants deficient for Tpd3 binding could still maintain an intact spindle checkpoint, as analyzed by their ability to grow on sublethal concentrations of benomyl (Figs. 4 and 7). These mutants could also undergo apparently normal cytokinesis (Fig. 8). Furthermore, WT Cdc55 also retained partial checkpoint activity in the absence of Tpd3 (Fig. 5 and Ref. 38). Two possibilities may theoretically be compatible with these findings. Cdc55 may bind the catalytic subunit in the absence of the A/Tpd3 subunit and direct the PP2A C-Cdc55 heterodimer to dephosphorylate spindle checkpoint and cytokinesis-related substrates. Alternatively, Cdc55 may have PP2A-independent activities. Our results support the suggestion that Cdc55 and the C subunit form a heterodimer. Both mutant 126 and WT Cdc55 could bind low levels of Pph22 without binding Tpd3. The association of mutant 126 with Pph22 was inhibited when the methyltransferase PPM1 gene was deleted (Fig. 7B). It has been previously reported that Ppm1 is responsible for PP2A-C methylation, which is required for association of Tpd3 and regulatory subunits B/Cdc55 and BЈ/Rts1 with the catalytic subunit (12,13). It now appears that PP2A-C subunit methylation also affects its binding to Cdc55 in the absence of Tpd3. Furthermore, loss of Cdc55-C dimer formation in the ppm1⌬ background was accompanied by loss of growth on benomyl. Although we cannot rule out the possibility that methylation is also involved in an as yet unidentified PP2A-independent function of Cdc55, our results are consistent with the simpler model suggesting that the spindle checkpoint is maintained by the Cdc55-C subunit dimer. Although it has been previously reported that B-C complexes could be found when purified PP2A complexes were subjected to cross-linking (42), to our knowledge this is the first time that an A-subunit-independent association between B and C subunits was demonstrated in vivo, and was assigned biological activities.
It has been previously reported that in S. cerevisiae, Tpd3 cellular concentration is limiting for both catalytic and regulatory Rts1 subunit binding (49). However, Cdc55 levels were shown to be much lower than levels of all other subunits. During normal mitosis or when the spindle is damaged by drugs, it may be essential that Cdc55 target PP2A activity to spindle checkpoint components fast, to prevent premature cell cycle progression. Binding to the non-limiting PP2A-C subunit directly may provide a more efficient solution to such emergencies than competition for PP2A AC heterodimers.
Several Cdc55 mutants, which associated with low levels of phosphatase activity (less than 15% of WT levels: K118E, 126, L102P, and S309P), retained WT Cdc55 activity in the spindle checkpoint but could not antagonize Tor signaling as well as WT Cdc55 (Figs. 4 and 7). Mutants K118E and 126 were also shown to support an apparently normal cytokinesis (Fig. 8).
These results may suggest that lower levels of Cdc55-associated phosphatase activity are required for the Cdc55 functions in the spindle checkpoint and in cytokinesis than are required to inhibit TOR signaling. Low levels of Cdc55-C heterodimer formation also appear not to be sufficient for E4orf4-mediated signaling, because mutant 126, which does not bind Tpd3, was obtained by virtue of conferring resistance to E4orf4, and a tpd3⌬ strain was reported to be resistant to E4orf4 (27).
In summary, study of the interaction between the viral regulator E4orf4 and PP2A, a key cellular protein, provided novel insights into the structure and function of PP2A. Furthermore, the set of cdc55 mutants obtained here will be useful in the future when structural data on Cdc55 becomes available, and can be correlated with these mutants.