Stress-induced Ceramide-activated Protein Phosphatase Can Compensate for Loss of Amphiphysin-like Activity In Saccharomyces cerevisiae and Functions to Reinitiate Endocytosis*

Saccharomyces cerevisiae cells lacking the amphiphysin-like orthologs, Rvs161 or Rvs167, are unable to thrive under many stress conditions. Here we show cells lacking Rvs161 require Cdc55, the B subunit of the yeast ceramide-activated protein phosphatase, for viability under heat stress. By using specific rvs mutant alleles, we linked this lethal genetic interaction to loss of Rvs161 endocytic domain function. Recessive mutations in the sphingolipid pathway, such as deletion of the very long-chain fatty acid elongase, Sur4, suppress the osmotic growth defect of rvs161 cells. We demonstrate that Cdc55 is required for sur4-dependent suppressor activity and that protein phosphatase activation, through overexpression of CDC55 alone, can also remediate this defect. Loss of SUR4 in rvs161 cells reinitiates Ste3 a-factor receptor endocytosis and requires Cdc55 function to do so. Moreover, overexpression of CDC55 reinitiates Ste3 endocytic-dependent degradation and restores fluid phase endocytosis in rvs161 cells. In contrast, loss of SUR4 or CDC55 overexpression does not remediate the actin polarization defects of osmotic stressed rvs161 cells. Importantly, remediation of rvs161 defects by protein phosphatase activation requires the ceramide-activated protein phosphatase catalytic subunit, Sit4, and the protein phosphatase 2A catalytic subunits, Pph21/Pph22. Finally, genetic analyses reveal a synthetic lethal interaction between loss of CDC55 and gene deletions lethal with rvs161, all of which function in endocytosis.

Failure to form these tubular structures leads to defects in intracellular trafficking (11).
Almost all rvs161-related phenotypes are suppressed by mutations in the sphingolipid biosynthetic pathway (18). For instance, loss of the SUR4/ELO3/VBM1 gene, which encodes a very long-chain fatty acid elongase required for the production of the C26 fatty acids found only in yeast complex sphingolipids, alters sphingolipid biosynthesis (sur4 cells accumulate the long-chain base, phytosphingosine (19)) and suppresses rvs defects (18,20). Sphingolipids are essential lipids and are conserved structurally throughout evolution (21). In addition to roles as structural components of membranes (22), sphingolipids act as signaling molecules regulating biological processes, such as cell growth, endocytosis, differentiation, stress, and apoptosis; the sphingolipid, ceramide, and multiple long-chain sphingoid bases are key signaling intermediates regulating these events in mammalian cells and in S. cerevisiae (23)(24)(25)(26)(27)(28).
Yeast CAPP is thought to be comprised of the subunits, Tpd3 (A), Cdc55 (B), and Sit4 (C), and its activation by addition of exogenous short-chain ceramides results in G 1 arrest (50). Putative CAPP activation through the addition of C 2 -ceramide, SUR4/ELO3/VBM1 gene inactivation, or overexpression of SIT4 can rescue the endocytic defects of cells lacking the Tlg1 or Tlg2 t-SNAREs involved in endocytosis (51) and confers normal growth and secretion to cells lacking the exocytic v-SNAREs, Snc1 and Snc2 (52). On the other hand, deletion of the CAPP B subunit, CDC55, or the catalytic subunits of PP2A abolishes the sphingoid base requirement in endocytosis (53). So the exact role of CAPP in regulating endocytosis is unclear. PP2A seems to regulate endocytosis in mammalian cells, and activation of PP2A is necessary for dynamin 2 dephosphorylation, resulting in the clathrin-dependent endocytosis of Na ϩ -ATPase and K ϩ -ATPase molecules (54).
How defects in sphingolipid biosynthesis suppress rvs phenotypes is unclear and likely very complex. Deletion of sphin-golipid genes suppresses the actin depolarization and growth defects of osmotic stressed rvs161 cells (55). However, steadystate actin cytoskeletal defects persist in rvs161 sur4 cells under conditions of glucose starvation, yet these cells are viable (20). Thus, understanding the molecular basis for how this suppression mechanism works should give valuable insight into how sphingolipid metabolites function as signaling lipids regulating many diverse cell events (23)(24)(25)(26)(27)(28).
To further understand how mutating sphingolipid biosynthesis suppresses rvs defects, we determined the roles of PP2A/ CAPP in this process. Here we show that PP2A/CAPP activation, through overexpression of CDC55 or loss of SUR4, is an essential prerequisite for suppressing rvs growth defects. Suppression requires the PP2A catalytic subunits, Pph21/Pph22, and/or the CAPP catalytic subunit, Sit4. Importantly, CAPP/ PP2A activation reinitiates endocytosis but does not remediate the actin polarization defects of rvs161 cells. We propose that sphingolipid-dependent PP2A/CAPP activation results in the reinitiation of endocytosis, and this is essential for cells to thrive under stress in the absence of amphiphysin function.
Media and Growth Conditions-Yeast strains were grown in YP media (1% yeast extract, 2% bactopeptone) supplemented with either 2% glucose (YEPD), 2% glycerol, 2% glucose plus 3.4% NaCl, or 2% glucose plus 6% NaCl as indicated. Yeast strains were also grown in synthetic minimal media containing 0.67% yeast nitrogen base supplemented with the appropriate amino acids and carbon source. To assay for the ability to grow under different stress conditions, yeast strains were grown to exponential phase in YEPD or selective minimal media plus 2% glucose. 2 ϫ 10 5 cells were spotted as 10-fold serial dilutions onto various media plates or streaked for single colonies and incubated at 30°C or 37°C as indicated. Cell growth was examined after 48 or 72 h as indicated.
Fluorescence Microscopy of Polymerized Actin and Lucifer Yellow Internalization-For all microscopic observations, yeast strains were grown to exponential phase in YEPD or ura Ϫ plus 2% glucose. For NaCl stress-induced experiments, cells were pelleted by centrifugation and resuspended in YEPD plus 3.4% NaCl, YEPD plus 6% NaCl, or ura Ϫ media containing 2% glucose plus 3.4% NaCl. Actin was stained with rhodamine phalloidin as described by Adams and Pringle (59) with some modifications.
Briefly, cells grown to exponential phase were fixed by addition of formaldehyde to a final concentration of 3.7%. After 30 min, cells were pelleted by centrifugation, washed three times in 1ϫ phosphate-buffered saline, pH 7.3, resuspended in 45 l of 1ϫ phosphate-buffered saline containing 3 units of Rhodamine-phalloidin (Molecular Probes), and incubated at room temperature for 2 h. Cells were then washed eight times with 1ϫ phosphate-buffered saline and resuspended in 20 l of 1ϫ phosphate-buffered saline. To visualize the internalization of lucifer yellow (LY), 10 7 cells grown to exponential phase were resuspended in 90 l of YEPD and stained with 10 l of 40 mg/ml lucifer yellow solution (Sigma). The stained cells were then incubated at 30°C for 1 h. Cells were then washed four times with ice-cold LY buffer (50 mM sodium succinate, pH 5), resuspended in 10 l of cold LY buffer, and stored on ice before visualization. The actin cytoskeleton and lucifer yellow internalization was visualized using a Leica DMRBE fluorescence microscope and rhodamine isothiocyanate optics and a PlanAPO 100ϫ objective. Data were collected using a Hamamatsu DIG-15 charge-coupled digital camera and Open Labs software (version 2.1). For lucifer yellow internalization studies, a total of 350 cells were counted during each experiment and the values represented are the average values of three independent experiments.
Classification of the Actin Polarization State of Cells-Only cells with small buds were counted. Cells with actin patches concentrated in the small bud, with five or fewer patches in the mother cell and visible actin cables, were classified as polarized cells. Cells with most actin patches in the mother cell rather than in the small bud were classified as depolarized cells. A total of 350 cells was counted during each experiment, and the values are the average values of six independent experiments.
Determination of Ste3 Stability-The yeast strains used carry a functional STE3::GAL1-STE3-HA::TRP1 chromosomal allele. Cultures grown to exponential phase in YEPD or ura Ϫ plus 2% glucose at 30°C were pelleted, washed, and shifted to YP or ura Ϫ media containing 2% raffinose for 1.5 h. To induce expression of Ste3-HA, cultures were washed and shifted to YP or ura Ϫ media containing 2% galactose for 3 h. Cultures were then washed and shifted to YEPD, ura Ϫ plus 2% glucose, YEPD plus 3.4% NaCl, or ura Ϫ plus 2% glucose plus 3.4% NaCl for 2 h to shut off the expression of Ste3-HA. Samples were taken before induction and at the indicated times post initiation of shut-off with glucose. Ste3-HA level was assayed by Western analysis and immunoblotting using 16B12 mouse anti-HA antibody (Covance).
Western Analysis-Total cell extracts were obtained using a modified procedure from Hsiung et al. (60). Exponentially growing yeast cells were pelleted and resuspended in yeast lysis buffer (50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 0.1% Nonidet P-40, 10% glycerol) containing 50 mM sodium fluoride, protease inhibitors (0.4 mM phenylmethylsulfonyl fluoride, 1 g/ml pepstatin, 1 g/ml leupeptin, and 1 g/ml aprotinin), and phosphatase inhibitors (0.1 mM sodium orthovanadate, 5 mM EDTA, 5 mM EGTA, and 10 mM sodium pyrophosphate). Cells were then lysed with glass beads by using seven cycles of vortexing for 1 min followed by 1-min incubation on ice. Extracts were collected after centrifugation for 5 min at 3,000 rpm. The protein concentrations of the cell lysates were determined by the BCA protein assay (Pierce). Proteins were resolved by 10% SDS-PAGE and subsequently transferred to a nitrocellulose membrane. Membranes were blocked overnight at 4°C with 5% nonfat dry milk in Buffer A (10 mM Tris-HCl, pH 7.4, 150 mM NaCl) plus 0.05% Tween 20. Incubations with primary and secondary antibodies were performed at room temperature for 1 h in buffer B (Buffer A containing 1% milk and 10% goat serum). Membranes were washed four times after antibody incubations with Buffer A containing 0.05% Tween 20. The primary antibody was monoclonal mouse 16B12 anti-HA (Covance, 1:500 dilution) and the secondary antibody was polyclonal goat antimouse horseradish peroxidase (Amersham Biosciences, 1:1000 dilution). Proteins were detected using ECL chemiluminescence (Amersham Biosciences). For a loading control, actin protein was detected using monoclonal rabbit anti-␤-actin primary antibody (Rockland, 1:1000 dilution) and polyclonal goat anti-rabbit horseradish peroxidase secondary antibody (Amersham Biosciences, 1:2000 dilution).
Protein Phosphatase Assay-Phosphatase activity was measured for 15 min at 30°C by following the dephosphorylation of phosphohistone in the absence and presence of C 2 -ceramide as previously described (50).
Synthetic Lethal Interaction Studies-Haploid matings between cdc55 cells and strains harboring null alleles of genes having a lethal genetic interaction with loss of RVS161 were carried out using standard procedures (61). Heterozygous diploids were sporulated, and double null haploid progeny were obtained by tetrad dissection. For each mating cross, the haploid segregants were allowed to grow on YEPD medium at 30°C, and at least 96 asci were tested.

RESULTS
The B Regulatory Subunit of Yeast CAPP/PP2A, Cdc55, Is Required for the Viability of rvs161 Cells under Stress-While we were exploring whether CAPP had a role in sur4-dependent suppression of rvs defects, we discovered rvs161 cdc55 cells were inviable at 37°C, while both rvs161 and cdc55 cells grew (Fig. 1A). In addition, we noted that loss of CDC55 did not remediate high salt rvs growth defects (6% NaCl).
To understand why Cdc55 was required for growth of rvs161 cells during stress, we examined how its loss affected the growth of cells harboring rvs161 mutant alleles isolated by Brizzio et al. (17). These alleles cause either endocytosis (End Ϫ Fus ϩ ) or cell fusion/mating (End ϩ Fus Ϫ ) defects, thus separating specific Rvs161 functions. R35C, R59K, R113K, P158S, and P166S alleles cause endocytosis defects, whereas A175P and P203Q are cell fusion/mating-defective, based on several criteria (17). CDC55 was deleted in the two cell fusion (A175P and P203Q) and three endocytosis mutants (R35C, R113K, and P158S), and we determined whether these strains could grow at 37°C or when under osmotic stress.
Suppression of rvs161 Defects by Loss of SUR4 Requires CDC55-Because rvs161 endo Ϫ cells required Cdc55 during heat stress, we tested whether this B regulatory subunit had a role in sur4-dependent suppression of rvs defects. We focused on the rvs osmotic growth defect as a representative mutant phenotype (18). As previously published, loss of SUR4 suppressed the high salt osmotic growth defect of rvs161 cells (18,20) (Fig. 2A). Importantly, Cdc55 function was absolutely essential for rvs161 sur4 cells to grow under this condition ( Fig.  2A, rvs161 sur4 versus rvs161 sur4 cdc55). Moreover, loss of Sur4 function remediated the high salt osmotic defect of rvs161 R35C, rvs161 R113K, and rvs161 P158S cells (Fig. 2B, lanes 1  versus 2), and in all cases Cdc55 was required for this suppression (Fig. 2B, lanes 2 versus 3).
The Putative CAPP Catalytic Subunit, Sit4, Is Not Required for Sur4-dependent Suppressor Activity-It was important to determine what molecular species of PP2A/CAPP was required for sur4-dependent suppression. To delineate molecular structure, we first examined what was the consequence of deleting the PP2A catalytic subunits, Pph21 and Pph22 (49), or the putative CAPP catalytic subunit, Sit4 (50, 62), on osmotic stressinduced growth of rvs161 sur4 cells. Deletion of both Pph21 and Pph22, and not Sit4, attenuated sur4-dependent suppressor activity (Fig. 3). Loss of function of the PP2A/CAPP A regulatory subunit, Tpd3, also resulted in loss of viability of rvs161 sur4 cells (not shown). Thus, a heterotrimeric PP2A/CAPP species comprised of Tpd3/Cdc55/Pph21-22 is required for sphingolipid-dependent suppression of rvs161 growth defects under osmotic stress.
Activation of PP2A/CAPP through Overexpression of CDC55 Suppresses Multiple Growth Defects of rvs161 Cells-Our hypothesis is that during stress PP2A/CAPP is activated in a sur4-dependent manner and is essential in the absence of Rvs161 endocytic function. If this is the case, activation of PP2A/CAPP alone, through overexpression of Cdc55 should suppress rvs defects in the absence of any sphingolipid defects. Thus, we asked if increased expression of CDC55 could com-  MAY 1, 2009 • VOLUME 284 • NUMBER 18 pensate for loss of Rvs161. The rvs161 mutant phenotypes examined were lack of growth under osmotic stress, under nutrient (glucose) starvation, and on a nonfermentable carbon source (not shown), as well as, the inability to mate (18).

PP2A-dependent Suppression of RVS Defects
Cdc55-dependent Suppressor Activity Does Not Function through Regulating Actin Polarization-rvs161 cells harbor defects in actin cytoskeletal structure (13), which contribute to the observed pleiotropic phenotypes. To ascertain how Cdc55 activation of PP2A/CAPP suppresses rvs161 defects, we first asked whether it does so through regulating actin dynamics during stress. Cells were grown to exponential phase and shifted to media containing 3.4% NaCl, and at various times the actin cytoskeleton was visualized using rhodamine phalloidin.
Interestingly, when we overexpressed CDC55 in wild-type cells, either less cells depolarized their actin, or a greater number of cells repolarized more quickly after 30 min in high salt media (50% polarized) (Fig. 5, A, WT versus B,  WTϩYep24-CDC55). Moreover, the percentage of polarized cells observed in high salt after 2 h was almost identical to that seen under normal vegetative growth conditions (92% polarized). CDC55 overexpression in rvs161 cells did remediate rvs-associated actin polarization defects (77% polarized), but only under normal vegetative growth conditions. Under high salt conditions, actin was completely depolarized, and it remained depolarized (Fig. 5, A, rvs161 versus B, rvs161ϩYep24-CDC55). Similar results were seen in 6% NaCl. Thus, any regulation of actin dynamics by Cdc55 during stress requires Rvs161 function.

Loss of SUR4 Does Not Remediate the Actin Polarization Defects of rvs161 Cells under Lethal Osmotic Stress Conditions-Loss of SUR4
suppresses the actin polarization defect of rvs161 cells under osmotic stress, and it was suggested this remediation plays a role in sur4-dependent suppressor activity (55). However, studies were performed using 3.4% NaCl, a concentration sublethal for rvs161 cells (Fig. 5D). If sur4 suppressor activity functions through regulating actin depolarization/repolarization, and this is essential for rvs161 sur4 cells to grow under stress conditions, it should regulate actin dynamics at a NaCl concentration lethal to rvs161 cells, but not rvs161 sur4 cells.
We first examined actin dynamics in cells grown under sublethal 3.4% NaCl conditions. As published, the actin polarization defect of rvs161 cells was suppressed (Fig. 5A, rvs161 versus  rvs161 sur4). However, Cdc55 function is not required for rvs161 sur4 cells to grow at this salt concentration, because rvs161 sur4 cdc55 cells are viable (not shown). Cdc55 is only required for sur4-dependent suppressor activity under lethal 6% NaCl conditions ( Fig. 2A). Thus, we determined if loss of SUR4 suppressed the actin polarization defect of rvs161 cells grown under this lethal growth condition. This was not the FIGURE 5. Cdc55-dependent sur4 suppressor activity in rvs161 cells does not function through regulating actin polarization. Strains were grown in YEPD or selective minimal media at 30°C to exponential phase, and at time zero, NaCl was added to a final concentration of 3.4% (A and B) or 6% (C). Actin structure was visualized at the indicated times using Rhodamine-phalloidin and fluorescence microscopy. Histogram showing the average percentage of polarized cells versus total number of cells was determined (n ϭ 350). Error bars represent the standard deviation. D, cells lacking Rvs161 are viable on 3.4% NaCl. 10-fold serial dilutions of cells were spotted on YEPD, 3.4% NaCl, and 6% NaCl solid media. Cells were grown for 3 days at 30°C. MAY 1, 2009 • VOLUME 284 • NUMBER 18 case, because rvs161 sur4 cells remained depolarized in 6% NaCl for 2 h (Fig. 5C, rvs161 versus rvs161 sur4) and up to 5 h. Moreover, we observed steady-state actin defects in rvs161 sur4 cells grown exponentially in media containing 6% NaCl (not shown). Based on these results, we conclude that sur4-dependent suppressor activity and subsequent PP2A/CAPP activation does not regulate actin dynamics to ensure rvs161 cells remain viable under stress.

PP2A-dependent Suppression of RVS Defects
Sur4-dependent Suppressor Activity in rvs161 Cells Reinitiates Endocytosis of the Pheromone Receptor, Ste3, and Requires Cdc55-rvs161 cells harbor defects in endocytosis (15). Neither the overexpression of CDC55 nor loss of SUR4 suppressed the actin polarization defects of rvs161 cells. Thus we asked if the endocytosis defect of rvs161 cells was affected. We did so by determining the stability of the membrane-associated a-factor pheromone receptor, Ste3. Ste3 is subject to two modes of endocytosis, constitutive (or ligand-independent) and ligand-stimulated (63)(64)(65)(66)(67), leading to vacuolar-dependent degradation. Ste3 stability assays are frequently used when studying mutants defective in endocytosis (15,68). We constructed strains carrying an endogenous chromosomal integrated galactose-inducible HA-tagged STE3 allele. Cells were grown to exponential phase, pulse/chase experiments were performed by addition and removal of galactose, and stability of Ste3 was examined by Western analysis.
CDC55 Overexpression Restores Ste3 Endocytosis in rvs161 Cells-We next determined whether CDC55 overexpression and subsequent PP2A/CAPP activation leads to reinitiation of endocytosis. This was the case, as CDC55 overexpression alone reduced Ste3 levels in rvs161 cells grown under normal vegetative conditions. In fact, rvs161 cells overexpressing CDC55 had dramatically less steady-state levels of Ste3 compared with wild-type cells (Fig. 6C). The deletion of PEP4, which encodes the vacuolar proteinase A required for endocytosis-dependent degradation of Ste3 (63), restored Ste3 accumulation (Fig. 6, C, rvs161 Yep24-CDC55 versus D, pep4 rvs161 Yep24-CDC55). Therefore, Ste3 is being expressed and produced to wild-type levels, and its disappearance is solely due to vacuolar-dependent proteolysis.
If CDC55 overexpression suppresses rvs161 defects through reinitiation of endocytosis, it should do so under stress conditions; thus we tested this hypothesis. Under osmotic stress, Ste3 levels still did not accumulate in rvs161 cells overexpressing CDC55 (Fig. 6E, rvs161 versus rvs161 Yep24-CDC55).
CDC55 Overexpression or Loss of SUR4 Reinitiates Fluidphase Endocytosis in rvs161 Cells-rvs161 cells harbor defects in receptor-mediated and fluid phase endocytosis (15). Thus, we asked whether CDC55 overexpression could restore fluidphase endocytosis to mutant cells. We did so by monitoring the time-dependent internalization and vacuolar localization of the fluid-phase endocytic maker, lucifer yellow. CDC55 was overexpressed as described under "Experimental Procedures" using YIp-GPD-CDC55-URA3. Cells were designated as having either rim staining (plasma membrane), or rim and vacuolar staining.
Lucifer yellow staining in wild-type cells was almost completely localized to the rim and vacuole (ϳ99%), whereas most rvs161 cells showed only rim staining (ϳ85%) (Fig. 7, WT GPD versus rvs161 GPD). Overexpression of CDC55 in rvs161 cells resulted in a ϳ3.5-fold increase in the percentage of cells with rim and vacuolar staining (Fig. 7, rvs161 GPD versus rvs161  GPD-CDC55). We next asked what PP2A/CAPP catalytic subunits were required. The putative CAPP catalytic subunit, Sit4, and the PP2A catalytic subunits, Pph21/22, were required for the observed Cdc55-dependent increase, as lucifer yellow staining returned to the rim of CDC55-overexpressing rvs161 cells lacking these catalytic subunits (92-98%) (Fig. 7, rvs161 GPD-CDC55 versus rvs161 sit4 GPD-CDC55 and rvs161 pph21  pph22 GPD-CDC55). Finally, we found no difference in lucifer yellow internalization between wild-type and rvs161 sur4 cells (not shown). Based on these results, we conclude PP2A/CAPP activation, either by CDC55 overexpression or loss of SUR4, results in reinitiating endocytosis, which is essential for the viability of rvs161 cells under stress. All strains carry a STE3::GAL-STE3-HA::TRP1 allele. Strains were grown to exponential phase in YEPD or selective minimal media at 30°C, shifted to YP or selective minimal media containing 2% raffinose for 1.5 h, and then shifted to YP or selective minimal media containing 2% galactose for 3 h to induce expression of Ste3-HA. To shut off expression of Ste3-HA, yeast strains were shifted to YEPD or selective media containing 2% glucose with (E) or without 3.4% NaCl (A-D). Samples were taken at the indicated times following the initial shut-off by glucose. Ste3-HA levels were determined by Western analysis. Actin levels (lower panel) were used as a loading control.

CAPP Activity Is Elevated in Cells Overexpressing CDC55 or
Lacking SUR4-Our hypothesis is that PP2A/CAPP activation is a necessary event for bypassing the need for Rvs161 function during stress. Thus, we asked if loss of SUR4 or overexpression of CDC55 causes an increase in in vitro CAPP activity. We measured basal/ceramide-independent (Fig. 8, shaded bars) and ceramide-activated (Fig. 8, solid bars) protein phosphatase activities as described (50).
In all strains tested, the majority of the basal/ceramideindependent catalytic phosphatase activity required the presence of the PP2A subunits, Pph21/Pph22, and not the CAPP catalytic subunit, Sit4 (Fig. 8, shaded bars). CAPP activity was absolutely dependent on Cdc55 and Tpd3 functions (Fig. 8, solid bars, WT versus cdc55, tpd3). These results are in good agreement with published work (50). In contrast to what was previously published, our results indicate two CAPP species can be activated and are comprised of Tpd3/ Cdc55/Sit4 and Tpd3/Cdc55/Pph21-Pph22. Strains lacking Pph21 and Pph22 together, or Sit4 alone, had severely reduced CAPP activity (Fig. 8, WT versus sit4 or pph21  pph22). Thus Cdc55 can target either Sit4 or Pph21/Pph22 for ceramide activation.

Lethal Genetic Interactions Exist between Loss of CDC55 and Loss of Factors Regulating Endocytosis and/or Actin Cytoskeletal
Structure-Our hypothesis is PP2A/CAPP functions under stress to regulate endocytosis. To begin to uncover factors regulated by PP2A/CAPP when cells are stressed, we determined whether lethal genetic interactions existed between loss of CDC55 and loss of genes synthetic lethal with loss of RVS161 (20,69,70). These genes regulate endocytosis and/or actin cytoskeletal organization (69,(71)(72)(73)(74)75). Double null haploid cells were generated and tested as described under "Experimental Procedures." Loss of CDC55 was synthetic lethal with loss of SLA2 (Table  2), which encodes an adaptor protein involved in membrane cytoskeleton assembly (72), and loss of VPS20, which encodes for a subunit of the endosomal sorting complex, ESCRTIII, required for transport of transmembrane proteins from multivesicular bodies to the lysosomal/vacuolar lumen (73). Loss of FIGURE 7. CDC55 regulates the internalization of lucifer yellow. LY internalization was assayed in cells containing the GPD promoter alone or CDC55 under the regulation of the GPD promoter. Cells were grown to early logarithmic phase in rich medium at 30°C, and then allowed to internalize LY at 30°C for 60 min. The average percentage of cells that accumulated LY versus total number of cells was determined (n ϭ 350) and standard deviations are indicated.  MAY 1, 2009 • VOLUME 284 • NUMBER 18

JOURNAL OF BIOLOGICAL CHEMISTRY 11937
CDC55 was also synthetic lethal at 37°C with loss of ABP1. ABP1 encodes for an actin-binding protein, which regulates actin cytoskeleton through binding to the SH3 domain of Rvs167 (69,71).
In contrast, cdc55 sla1, cdc55 sac6, and cdc55 vps21 mutants were all viable. Sla1 interacts with proteins that regulate actin dynamics and/or are required for endocytosis, such as Rvs167 (76). Sac6, also known as fimbrin, is an actin-bundling protein important in the organization and maintenance of the actin cytoskeleton (74), whereas Vps21, a mammalian Rab5 homolog, is a GTPase required for transport during endocytosis and for correct sorting of vacuolar hydrolases (72,75). Thus Cdc55, and presumably PP2A/CAPP, genetically interacts with a specific subset of factors involved in regulating endocytosis and the actin cytoskeleton.
PP2A/CAPP Can Only Compensate for rvs161 Function under Stress-Our results show that PP2A/CAPP can replace Rvs161 function under stress (Fig. 4). Thus, we asked whether PP2A/CAPP activation through CDC55 overexpression could compensate for loss of SLA1, SLA2, SAC6, or VPS20 and suppress the ts stress phenotype of cells lacking these genes. This was not the case, because sla1, sla2, sac6, and vps20 cells overexpressing CDC55 remained incapable of growing at high temperature. Thus PP2A/CAPP can only compensate for Rvs161 function under stress.
Thus, a second important result from our study is that two CAPP species can be formed in yeast cells. This is borne out by our in vitro CAPP assays, where decreased CAPP activity was observed in pph21 pph22 and sit4 cells (Fig. 8). Heterotrimeric CAPP activities have been purified from glioma cells and rat brain (39,42). They are members of the PP2A family, and they contain catalytic subunits orthologous to Pph21/Pph22. Small interference RNA-mediated depletion of this catalytic subunit in rat insulin-secreting cells (INS 832/13) significantly reduces CAPP activity (77), as does combined loss of PPH21 and PPH22 in yeast cells (Fig. 8).
The human ortholog of the putative CAPP catalytic subunit, Sit4, is the phosphatase, Pp6 (78). Sit4 may be the target of viral-induced cell death of yeast, through its association with Cdc55. Yeast cells are killed by overexpression of the adenovirus early region 4 open reading frame 4 (E4orf4), and this E4orf4-induced toxicity can be suppressed by deleting CDC55 (79 -81). This suggests E4orf4 uses Cdc55 to target a phosphatase. The catalytic subunit may be Sit4, because sit4 cells are highly sensitive to E4orf4 killing, and this phenotype cannot be suppressed by deletion of CDC55 (79). Human cells lacking Pp6 are also highly sensitive to killing by E4orf4 expression (79). To date, there is no direct evidence that Cdc55 and Sit4, or their metazoan counterparts, directly interact. However, association may only occur under very specific circumstances, like the stress of a viral assault.
Heat stress increases the levels of several ceramide species in S. cerevisiae through de novo biosynthesis (29), whereas the largest heat-induced increase of sphingoid base includes phytosphingosine and dihydrosphingosine (32,34). sur4 mutants constitutively accumulate the long-chain sphingoid base, phytosphingosine, and they have reduced levels of ceramide and complex sphingolipids (19). Based on these results, phytosphingosine is the only common sphingolipid accumulating that could directly activate CAPP.
In vitro CAPP assays have extensively examined the lipid requirements for CAPP activation (38,41,42,82,83). CAPP can be activated by C 2 -, C 6 -, and C 18 -ceramides, as well as natural ceramides derived from bovine brain. Sphingosine, sphingolmyelin, and diacylglycerol have also been tested and do not activate CAPP. Structural requirements for CAPP activation by ceramide include the presence of the 4,5-trans double bond in the sphingoid base, and the primary and secondary hydroxyl groups of the sphingoid backbone. Although ceramide has long been recognized as the putative lipid activator of CAPP, there are reports that palmitate (84,85), linoleic acid (86), and cholesterol (87) can also stimulate activity. Thus, lipids other than ceramide can be activators of CAPP.
lcb1-100 ts cells are defective in endocytosis at high temperature; lcb1-100 ts cells harbor a weakened serine pamitoyltransferase activity required for the first step in sphingolipid biosynthesis (88). The addition of exogenous phytosphingosine to these cells restores endocytosis (89). Riezman and colleagues (53) showed loss of Cdc55, or overexpression of the yeast casein kinase 2, Yck2, or protein kinase C, Pkc1, restores endocytosis to lcb1-100 ts cells. Based on these results, they suggest Cdc55 functions to negatively regulate the activities of Yck2 and/or

synthetic lethal genetic interactions with genes regulating endocytosis
Haploid progeny from crosses were grown on rich media for 72 h at 30°C or 37°C.

PP2A-dependent Suppression of RVS Defects
Pkc1 in response to a sphingolipid generated signal. Our results show Cdc55-dependent PP2A/CAPP acts as a sphingolipid-dependent positive regulator of endocytosis, because its loss causes the reappearance of endocytosis defects. The fact that PP2A/CAPP activity can be differentially modulated by various lipid signals adds an additional tier of sophistication for regulating endocytosis. The overexpression of SIT4 or loss of SUR4, remediates the endocytic/exocytic defects of cells lacking specific v-or t-SNARES (51,52). Defects are suppressed through Sit4-dependent regulation of the phosphorylation states of the plasma membrane t-SNARES, Sso1/Sso2, or v-SNARES Tlg1/Tlg2 (51,52). The roles of Tpd3, Cdc55, Pph21, and Pph22 were not explored as they were in the present study, nor was it tested whether SIT4 overexpression or loss of SUR4 suppress the endocytosis defects of these mutants when they are under osmotic stress. So the exact molecular species of PP2A/CAPP were not resolved.
Our synthetic lethal studies indicate stress-induced PP2A/ CAPP regulates some Rvs161-dependent function ( Table 2). Rvs161 functions to "sense" and induce membrane curvature through its N-terminal BAR domain (1). rvs161 mutants accumulate vesicles at the plasma membrane (16,90), and synthetic lethal screening has establishing a putative role for Rvs161 in vesicle trafficking (70). Two-hybrid screens using Rvs161 as bait (20,91) have not uncovered putative interactions with proteins regulating vesicle secretion. Rvs167, which also contains a BAR domain and binds Rvs161, interacts by two-hybrid with Sec8, Sec20, and Exo70, which are involved in vesicle transport ( Table 2 (91). Based on our results, we have put forth a model to explain how multiple PP2A/CAPP species can be targeted and activated (Fig. 9). rvs161 cells harbor endocytosis defects under stress conditions, which causes cell death (Fig. 9A). During some types of stress, Cdc55 levels increase, modeled by our overexpression studies (Fig. 9B). The increase in Cdc55 levels shifts the cellular balance of PP2A from predominately a single Tpd3/Rts1/Pph21-22 heterotrimer to at least two species comprised of Tpd3/Cdc55/Pph21-22 and Tpd3/ Cdc55/Sit4; Rts1 is the other PP2A B regulatory subunit in yeast (Table 1), and it is in 10-fold excess compared with Cdc55 (92). The reshuffling of PP2A activities results in targeting cell factors regulating endocytosis, restoring viability to rvs161 cells.
The PP2A catalytic subunits, Pph21 and Pph22, and the CAPP catalytic subunit, Sit4, are required. Here, PP2A/CAPP activation occurs in a sphingolipid-independent manner. Microarray analyses have shown CDC55 expression increases in response to heat shock, nitrogen depletion, and during stationary phase (93). rvs61 cells are inviable when starved for nitrogen, which results in their premature entry into stationary phase (12).
When a specific stress generates a PP2A/CAPP-responsive sphingolipid signal, activation occurs and functions through the PP2A catalytic subunits, Pph21 and Pph22 (Fig. 9C). Cdc55 levels remain constant, and its affinity for Pph21 and Pph22, and not Sit4, is increased due to post-translational modifications and/or changes in Cdc55-protein interactions. Thus, the Cdc55 activity state is regulated by various stresses to ultimately activate and regulate the substrate specificities of multiple PP2A/CAPP species.