Adenovirus Protein E4orf4 Induces Premature APC^Cdc20 Activation in Saccharomyces cerevisiae by a Protein Phosphatase 2A-Dependent Mechanism

Abstract

Protein phosphatase 2A (PP2A) has been implicated in cell cycle progression and mitosis; however, the complexity of PP2A regulation via multiple B subunits makes its functional characterization a significant challenge. The human adenovirus protein E4orf4 has been found to induce both high Cdk1 activity and the accumulation of cells in G₂/M in both mammalian and yeast cells, effects which are largely dependent on the B55/Cdc55 regulatory subunit of PP2A. Thus, E4orf4 represents a unique means by which the function of a specific form of PP2A can be delineated in vivo. In Saccharomyces cerevisiae, only two PP2A regulatory subunits exist, Cdc55 and Rts1. Here, we show that E4orf4-induced toxicity depends on a functional interaction with Cdc55. E4orf4 expression correlates with the inappropriate reduction of Pds1 and Scc1 in S-phase-arrested cells. The unscheduled loss of these proteins suggests the involvement of PP2A^Cdc55 in the regulation of the Cdc20 form of the anaphase-promoting complex (APC). Contrastingly, activity of the Hct1 form of the APC is not induced by E4orf4, as demonstrated by the observed stability of its substrates. We propose that E4orf4, being a Cdc55-specific inhibitor of PP2A, demonstrates the role of PP2A^Cdc55 in regulating APC^Cdc20 activity.

Protein phosphatase 2A (PP2A) represents a major class of serine/threonine phosphatases that is evolutionally conserved across eukaryotes and plays a regulatory role in numerous cellular processes, including signal transduction, cell morphology, and cell cycle control (12, 15, 26, 43). The diversity of PP2A functions is due primarily to the existence of several PP2A holoenzyme variants. PP2A generally exists as a heterotrimer, composed of a catalytic C subunit, a structural A subunit, and a crucial regulatory B subunit that not only confers substrate specificity to the enzyme but also directs its cellular localization (13). In mammalian cells, there exist at least 18 known B regulatory subunit isoforms categorized into three classes (B, B′, and B′′) and a related fourth class (sometimes referred to as B′′′). These classes share very little or no homology, despite their common abilities to bind overlapping sites within the A subunit (12). The situation is simpler in yeast, in which the catalytic C subunit is encoded redundantly by two duplicated genes, PPH21 and PPH22; a single A subunit is encoded by TPD3; and only two B-type subunits exist, encoded by CDC55 (corresponding to mammalian B/B55 subunits) and RTS1 (corresponding to the B′/B56 family) (51). Nevertheless, it is difficult to determine what particular form of PP2A actually functions in specific PP2A-regulated processes in all eukaryotic cells. PP2A has previously been implicated in the control of mitotic events in yeast and higher eukaryotes that are essential for cell survival (12, 15); however, the precise role of PP2A regarding mitotic progression and cell cycle regulation has yet to be fully defined, especially with regard to the specific form of PP2A that is involved.

The E4orf4 (early region 4 open reading frame 4) protein of human adenoviruses is a 114-residue polypeptide containing an arginine-rich nuclear/nucleolar targeting sequence (27) shown previously to induce the p53-independent death of human cancer cells when expressed alone in the absence of other viral proteins (1, 19-23, 36-38). More importantly, E4orf4 protein was shown to bind to the B55α regulatory subunit of PP2A (1, 22, 37, 38), and analysis of a series of E4orf4 mutants revealed that functional interaction between E4orf4 and B55α correlates with tumor cell killing (22, 37). We have shown recently that E4orf4 protein interacts only with members of the B/B55 class of regulatory subunits in mammalian cells (20), and it is widely accepted that this interaction is important in eliciting toxicity in mammalian cells. Our group and others have also shown that E4orf4 is lethal when expressed in the budding yeast Saccharomyces cerevisiae (1, 17, 32). In yeast, E4orf4 protein binds to Cdc55, and the interaction between E4orf4 and the A and C subunits of the PP2A holoenzyme is entirely dependent on its binding to Cdc55 (32). The majority of E4orf4-induced toxicity is relieved in cdc55Δ mutant strains, suggesting that the PP2A^Cdc55 complex is involved in E4orf4-induced cell death (17, 32). While the specific mechanism by which E4orf4 reduces viability in yeast is unclear, PP2A^Cdc55 is likely to play a significant role. PP2A is known to play a role in the regulation of mitotic events, and we have observed previously that E4orf4 expression triggers mitotic arrest accompanied by elevated levels of Cdk1/Clb2-Cdc28 in both mammalian cells (20, 21) and yeast (32). Thus, it is reasonable to speculate that E4orf4 toxicity is elicited through the perturbation of signaling pathways critical for orderly progression through mitosis.

Mitosis is a tightly regulated process, and the switch-like transition between metaphase and anaphase depends on multiple control inputs, one of which is the precise timing of the activation of the anaphase-promoting complex (APC) or cyclosome. The APC is a multisubunit E3 ubiquitin ligase which degrades a number of proteins, including the cyclins that drive the metaphase activity of Cdk1, and securin (Pds1), a protein which inhibits the protease separase/Esp1. The coordinated activation of two forms of APC is dependent on its interaction with cofactors Cdc20 and Hct1 (28, 29, 44). Activation of the APC by Cdc20 is achieved through many regulatory pathways, one of which is the rise in Clb2-Cdc28/Cdk1 activity as cells enter mitosis. Thus, one mechanism controlling the timing of mitotic events is the balance between the inhibitory phosphorylation of Cdc28 on residue Y19 by the Wee1 ortholog Swe1 and the removal of the Y19 phosphate group by the Mih1 phosphatase, which activates Cdk1 (24). When Cdk1 activity rises, APC core subunits themselves become phosphorylated (18, 33). The completion of sister chromatid biorientation promotes the binding of Cdc20 to the APC (49). Active APC^Cdc20, in turn, degrades securin/Pds1 (5) and allows separase/Esp1 to cleave the cohesin Scc1. These events are critical for sister chromatid separation and the onset of anaphase (4, 5, 41, 42). Premature activation of the APC would be deleterious, as this would compromise the accurate segregation of chromosomes.

High levels of Clb2-Cdc28/Cdk1 inhibit the activation of the second form of the APC, APC^Hct1, through phosphorylation of the Hct1 cofactor itself (14). Thus, failure to degrade the B-type cyclins, as well as Cdc20, prevent anaphase exit (50). The dampening of Clb2-Cdc28/Cdk1 activity is achieved through the initial degradation of Clb2, which is mediated by APC^Cdc20. This allows partial APC^Hct1 formation, which then becomes fully activated as Cdc20 is progressively degraded (48). Formation of an active APC^Hct1 together with the actions of the Cdc-fourteen early release (FEAR) and mitotic exit network (MEN) pathways, allows for proper mitotic exit and subsequent cytokinesis (6). Interestingly, Cdc55 has been implicated in the regulation of APC^Cdc20 and has been reported to play a role in mitotic exit, acting at the level of both the FEAR and MEN pathways (30, 31, 45-47).

We have shown recently with mammalian cells that binding of E4orf4 reduces the activity of B55α-containing PP2A holoenzymes when measured in vitro using purified complexes and substrates and that, in vivo, certain putative targets of this form of PP2A, including 4EBP-1 and p70^S6K, are hyperphosphorylated following E4orf4 expression (20). In addition, the PP2A inhibitors okadaic acid and I-PP2A were found to enhance E4orf4 killing, which was also shown to be enhanced with increasing concentrations of E4orf4 protein. Thus, we have postulated that the toxic effects of E4orf4 could be elicited when a sufficient proportion of the B55/Cdc55-containing PP2A holoenzyme population is inhibited to prevent the dephosphorylation of key substrates (20).

We propose that E4orf4 may provide a unique tool for specific interrogation of the role that PP2A^Cdc55 plays in the regulation of mitotic events and for identification of PP2A^Cdc55 substrates. In the present study, we hypothesized that E4orf4, through its interaction with Cdc55, may uncouple PP2A^Cdc55 regulation of mitotic events. This uncoupling would result in the unscheduled activation of APC^Cdc20, leading to degradation of specific APC substrates and mitotic defects. To understand the mechanism by which E4orf4-expressing cells develop these defects, which when left uncorrected ultimately culminates in decreased viability of cells, we examined the activity of APC^Cdc20 and APC^Hct1 in vivo to determine precisely the transition point affected by E4orf4 expression. We show here that, in a PP2A^Cdc55-dependent manner, E4orf4 induces the premature activation of APC^Cdc20 but not APC^Hct1, suggesting that PP2A^Cdc55 plays an important regulatory role in anaphase exit.

MATERIALS AND METHODS

Strains, plasmids, and media.

The yeast strains used in this study (Table 1) are all derivatives of W303. Wild-type hemagglutinin-tagged E4orf4 (HA-E4orf4) and HA-E4orf4 point mutants were subcloned from mammalian expression vectors into either the p424GAL1 or p425GAL1 (ATCC) DNA vector under the control of the GAL1 promoter. The same was done for FLAG-tagged E4orf4. HA-E4orf4 was also subcloned into the pYES2 (Invitrogen) DNA vector by the same method. The Clb2-HA₃ plasmid (GAL1pr, CEN, and URA) was obtained from Brenda J. Andrews (University of Toronto). Yeast transformations were performed using the one-step method of Chen et al. (3). Transformants were selected on synthetic complete medium containing 2% glucose and the appropriate auxotrophic supplements.

TABLE 1.

Yeast strains

Strain	Genotype	Source/reference
DHY189	MATaade2-1oc/ade2-1oc can1-100 his3-11,15 leu2-3,112 ade3Δ100 URA3::TUB1-YFP	Yeast Resource Center
HFY3	MATα ade2-1 ura3-1 leu2-3,112 rts1::HIS3 cdc55::TRP1	40
MSG36	MATα ade2-1 ura3-1 his3-11 trp1-1 leu2-3,112 HA3-CDC55	9
MSG107	MATα ade2-1 ura3-1 his3-11 trp1-1 leu2-3,112 GFP-CDC55	9
MSG136	MATα ade2-1 ura3-1 his3-11 trp1-1 leu2-3,112 RTS1-GFP CFP-SPC42(TRP)	9
YS69	MATα ade2-1 ura3-1 his3-11 trp1-1 leu2-3,112 RTS1-HA3	39
YS95	MATα ade2-1 ura3-1 his3-11 leu2-3,112 cdc55::TRP1	40
YS96	MATα ade2-1 ura3-1 trp1-1 leu2-3,112 rts1::HIS3	40
6803	MATα ade2-1 can1-100 leu2-3,112 his3-11,15 tetR-GFP-LEU2 TetOs-URA3PDS1−Myc18::LEU2 [psi+]	4
6565	MATaade2-1 can1-100 leu2-3,112 his3-11,15 ura3 GAL [psi+] SCC1 Myc18::TRP1	25
6142	MATaade2-1 can1-100 leu2-3,112 his3-11,15 ura3 GAL CDC5-Myc15::URA3 [psi+]	35
425	MATα Myc18-CDC20-TRP CDC16-HA3URA3 ade2-1 trp1-1 can1-100 leu2-3,112 his3-11,15 ura3	50

Cell synchronization experiments and FACS analysis.

Yeast strains transformed with galactose-inducible control, FLAG-E4orf4, or HA-E4orf4 DNA plasmids were grown in glucose-containing synthetic complete medium overnight and then rediluted into fresh medium and allowed to incubate for 2 h. Cells were then transferred to medium containing 2% raffinose, 0.2% glucose, and 0.2 M hydroxyurea (HU) (Sigma) for 5 h to synchronize cells in S phase. After a 5-h synchronization period (T = 0 h), cells were resuspended in fresh medium containing 2% raffinose, 2% galactose, and 0.2 M HU to maintain the S-phase arrest during induction of E4orf4 expression. Cells were harvested after 3 or 4 h of galactose induction. Cell samples were processed for Western blotting (described below) as well as fluorescence-activated cell sorting (FACS) analysis following the method outlined by Dien et al. (7). Propidium iodide-stained cells were acquired on a FACScan instrument, using Cell Quest software. Cell cycle histograms were created using FCS Express V3.

Western blot analysis.

Whole-cell extracts (WCEs) were prepared by resuspending cells in yeast lysis buffer (25 mM Tris-Cl, pH 7.4, containing 125 mM NaCl, 2.5 mM EDTA, 1% Triton X-100, and protease inhibitors) and vortexing them with acid-washed glass beads (Sigma). Protein amounts were quantified by Bio-Rad protein assay reagent, and 20 to 50 μg of total protein per sample was resolved by SDS-polyacrylamide gel electrophoresis (SDS-PAGE). Separated proteins were transferred to polyvinylidene difluoride (PVDF) membranes (Millipore) and immunoblotted with the indicated antibodies. Anti-HA (HA.11; Covance), anti-(M2) FLAG (Sigma-Aldrich), and anti-Myc (9E10; Covance) antibodies were all used at 1:1,000 dilutions. Rabbit polyclonal anti-Cdc28 (a gift from Raymond J. Deshaies, California Institute of Technology) was used at a 1:2,000 dilution as a loading control. Membranes were incubated with secondary antibody linked to horseradish peroxidase (Jackson ImmunoResearch) at a 1:10,000 dilution, and enhanced chemiluminescence (ECL) detection (PerkinElmer) followed.

Coimmunoprecipitation assay.

Yeast strains expressing HA₃-Cdc55 or Rts1-HA₃ (9, 39) were transformed with either vector control plasmid DNA or plasmid DNA encoding FLAG-E4orf4. Cells were grown overnight in 2% glucose-containing synthetic complete medium, transferred to 2% raffinose medium for 2 h, and finally resuspended in 2% galactose-2% raffinose medium for 6 h to induce E4orf4 expression. Cells were lysed with yeast lysis buffer (as described above), and 20 μg of WCE was used to detect protein expression by Western blotting. For each sample, 1 mg of WCE was precleared with protein G agarose beads (50% slurry) (Upstate) for 1 h, and the precleared lysates were subjected to immunoprecipitation with anti-HA antibody for 2 h at 4°C. Immunoprecipitates were washed five times with lysis buffer, heated in sample buffer for 5 min, and subjected to SDS-PAGE, PVDF transfer, and Western blotting with appropriate antibodies.

Kinase immunoprecipitation assay.

Yeast strains transformed with control vector DNA or plasmid DNA expressing Clb2-HA₃ under the GAL1 promoter were grown in 2% glucose-containing medium overnight, transferred to 2% raffinose for 2 h, and finally resuspended in fresh medium containing 2% raffinose and 2% galactose for 4 h to induce Clb2 expression. Cells were collected and lysed as described above. A total of 20 μg of WCE was used to detect Clb2 expression by immunoblotting. A total of 400 μg of WCE was immunoprecipitated with 0.5 μg of rabbit polyclonal Clb2 antibody (Santa Cruz) and 12 μl of protein A agarose beads (50% slurry) (Upstate) after 1 h of preclearing with protein A agarose beads. Immunoprecipitates were washed four times in lysis buffer and two times in kinase assay buffer (50 mM HEPES, pH 7.5, 1 mM EGTA, 2 mM MgCl₂, 1 mM dithiothreitol [DTT]). Immunoprecipitates were then incubated for 30 min at 30°C in 20 μl of reaction mix containing kinase assay buffer, 10 μg histone H1 (Sigma), and 10 μCi [γ³²P]ATP (specific activity, 6,000 Ci/mmol). Kinase reactions were stopped by the addition of 10 μl of 4× sample buffer and the heating of samples at 100°C for 5 min. Proteins were separated by SDS-PAGE using 15% polyacrylamide gels and then transferred to PVDF membranes. The level of γ³²P-labeled H1 was determined by autoradiography.

Cell growth and colony survival assays.

Spotting cell growth assays were performed as previously described (32). Briefly, cells were grown overnight in 2% glucose-containing synthetic complete medium and then resuspended in 2% raffinose medium for 2 h. Approximately equal numbers of cells were serially diluted, and equal volumes spotted onto 2% glucose-containing control plates and 2% raffinose-2% galactose plates and incubated at 30°C for 2 days, after which pictures were taken. The colony survival assays performed were done in a similar fashion. Cells grown overnight in 2% glucose-containing synthetic complete medium were first resuspended in 2% raffinose medium for 2 h (T = 0 h) and then switched to 2% raffinose-2% galactose medium to induce E4orf4 expression for 3, 6, and 9 h (T = 3 h, 6 h, and 9 h). Immediately, approximately 300 cells were plated onto glucose-containing plates (3 replica plates) and incubated at 30°C for 2 days, after which colony numbers were counted.

Microscopy.

Yeast strains carrying green fluorescent protein (GFP)-tagged Cdc55 or Rts1 (9) were transformed with control vector DNA or pYES2HA-E4orf4 plasmid. Cells were grown in 2% glucose-containing synthetic complete medium overnight and transferred to 2% raffinose-containing medium, and then HA-E4orf4 expression was induced for 5 h in 2% galactose-plus-2% raffinose-containing medium. Cells were collected and fixed with 4% formaldehyde, and GFP-Cdc55 or Rts1-GFP was visualized by fluorescence microscopy. Image stacks (0.5-μm optical sections with 10 to 12 stacks per image) were acquired using Volocity software. About 150 cells per sample were scored for GFP-Cdc55 and Rts1-GFP localizations. Representative images shown are the extended focus of a z series of 0.5-μm optical sections, and imaging processing was performed with Adobe Photoshop CS. Similar conditions were utilized for the visualization of yellow fluorescent protein (YFP)-Tub1 localization, as depicted in Fig. S2 in the supplemental material, except for the growth conditions under which S-phase synchronization was used (described above).

RESULTS

E4orf4 interacts with Cdc55 but not Rts1.

E4orf4 has been shown to interact with the B55/Cdc55 regulatory subunits of PP2A in human cells (1, 20, 22, 37, 38) and yeast (1, 17, 32). Additionally, in yeast, E4orf4 does not associate with the A and C subunits in the absence of Cdc55, suggesting that the E4orf4-PP2A interaction is through Cdc55 alone (32). To define the specificity of the E4orf4-Cdc55 interaction further, we examined whether or not E4orf4 also interacts with Rts1 through coimmunoprecipitation experiments. Yeast cells expressing HA₃-Cdc55 or Rts1-HA₃ (9) were transformed with either vector control plasmid DNA alone or a plasmid DNA expressing an N-terminal FLAG-tagged E4orf4 protein under the control of the inducible GAL1 promoter. Protein extracts were prepared from cells induced with galactose for 6 h, and epitope-tagged proteins were immunoprecipitated using anti-HA or anti-FLAG antibodies. Copurifying proteins were detected by immunoblotting. Figure 1A shows that, as previously described (32), FLAG-E4orf4 and HA₃-Cdc55 were found to interact in reciprocal coimmunoprecipitation experiments. In contrast, Rts1-HA₃ and FLAG-E4orf4 did not interact, even though both could be detected independently in the control immunoprecipitations as well as WCEs.

FIG. 1.

Involvement of Cdc55 and Rts1 in E4orf4 binding and cell killing. (A) Binding to E4orf4. Yeast strains containing endogenously tagged HA₃-Cdc55 (MSG36) or Rts1-HA₃ (YS69) were transformed with plasmid DNAs encoding a vector control or FLAG-E4orf4. Cell lysates were immunoprecipitated (IP) with anti-HA or anti-FLAG antibodies and subjected to SDS-PAGE and Western blotting with the indicated antibodies. (B) Cell killing. Cultures of W303 wild-type and cdc55Δ, rts1Δ, and cdc55Δ rts1Δ mutant yeast strains containing either vector control DNA or HA-E4orf4 DNA were spotted on noninducing 2% glucose or inducing 2% raffinose-2% galactose agar medium for cell growth assays, as described in Materials and Methods. Photographs were taken after plates were incubated at 30°C for 2 days.

It is widely believed that the cell death function of E4orf4 is mediated primarily through an interaction with the B55/Cdc55 class subunits. To confirm this, cell growth assays were conducted in the presence or absence of E4orf4 using appropriate yeast strains (40). Figure 1B shows that wild-type (WT) cells, cells with deleted Rts1 (rts1Δ mutant), and cells with both Cdc55 and Rts1 deleted (cdc55Δ rts1Δ mutant) were all susceptible to E4orf4-induced toxicity, whereas cells deleted for Cdc55 (cdc55Δ mutant) were significantly more resistant to killing. These results confirm that E4orf4 toxicity, with respect to PP2A, is specific to the Cdc55 subunit. The toxicity in cdc55Δ rts1Δ mutant cells suggests that there is a PP2A-independent perturbation that is uncovered when both targeting subunits are absent. Fig. S1 in the supplemental material shows that cell inviability rather than growth arrest occurred in wild-type E4orf4-expressing cells, as the majority of cells were unable to resume growth at 4 h post-E4orf4 expression, as determined by colony survival assays. These data suggest that misregulation of the PP2A^Cdc55 pool of PP2A induces cell death and demonstrates that any E4orf4-induced changes in PP2A activity or substrate specificity are solely, or at least predominantly, through Cdc55.

PP2A^Cdc55 localization is affected by E4orf4.

Previous studies with mammalian PP2A complexes suggested that E4orf4 may alter PP2A^Cdc55 activity (20, 22, 37, 38). We felt that PP2A^Cdc55 interaction with E4orf4 may also change the localization of PP2A^Cdc55, thus potentially altering access to critical substrates. During S and M phase, PP2A^Rts1 localizes to kinetochores and, upon anaphase exit, relocalizes to the bud neck, where it plays a role in septin dynamics and cytokinesis (8, 9). In contrast, PP2A^Cdc55 trimers (and even Cdc55 monomers) localize via targeting sequences in Cdc55 to bud tips in small or large budded cells and to the bud neck in large budded cells (9). Cdc55 and Rts1 are also found in the cytoplasm and nucleus in all cell cycle phases (9). To determine if E4orf4 binding affects the localization of PP2A^Cdc55 complexes, control DNA or HA-E4orf4 was expressed in yeast strains expressing GFP-Cdc55 or Rts1-GFP and Spc42-CFP fusion proteins (9). Spc42 is a major component of the spindle pole body (SPB). The SPB duplicates at the G₁/S transition, and at that time, the SPBs are in close proximity to kinetochores, prior to SPB separation (10, 16). We previously demonstrated that Rts1-GFP colocalizes with Spc42 during that time (9). Cells were visualized by fluorescence microscopy to determine the localization of GFP-Cdc55 or Rts1-GFP after 6 h of E4orf4 expression (n = 150 cells). Figure 2A (top) shows that in cells containing the control DNA, Rts1-GFP colocalized with Spc42-CFP at the kinetochores, as has previously been reported (9), but we did not consistently observe Rts1-GFP at the bud neck, most probably due to the transient, dynamic nature of this localization. Figure 2A (bottom) shows that in E4orf4-expressing cells Rts1-GFP localized in a fashion similar to that of the kinetochore. Figure 2B shows that in vector control cells, GFP-Cdc55 localized to the bud tip in small (39%) and large (4%) budded cells, to the bud neck in large (59%) budded cells, and to the nucleus in all cell types, as observed previously (9). In contrast, Fig. 2B shows that in E4orf4-expressing cells, GFP-Cdc55 localization was greatly reduced at the bud tip (10% in small and 0% in large budded cells) and bud neck (5%), with clear accumulation in the nucleus in all cell types. The localization data of GFP-Cdc55 in the presence and absence of E4orf4 has been summarized at the bottom of Fig. 2B. Expression of E4orf4 did not disturb the localization of PP2A^Rts1 but dramatically altered the normal localization of PP2A^Cdc55 at the bud tip and bud neck.

FIG. 2.

Effect of E4orf4 on the localization of Rts1 and Cdc55. (A) Rts1. Cells expressing Rts1-GFP (shown in green) and Spc42-CFP (shown in red) fusion proteins (MSG136) were transformed with control plasmid DNA or DNA expressing HA-E4orf4. Cells were prepared and visualized by fluorescence microscopy as detailed in Materials and Methods. The images shown are representative of Rts1-GFP kinetochore and Spc42-CFP spindle pole body localizations. The merged image shows that Rts1-GFP colocalizes near Spc42-CFP. (B) Cdc55. Control vector DNA or DNA expressing HA-E4orf4 was expressed in a strain that expresses GFP-CDC55 (MSG107). Shown are representative GFP-Cdc55 localizations in the nucleus and bud tip (left) and at the bud neck (middle). Shown on the right is the most common pattern exhibited when E4orf4 is expressed, GFP-Cdc55 being concentrated in the nucleus in the great majority of cells. The table at the bottom shows the percentage of unbudded, small budded, and large budded cells with GFP-Cdc55 at the indicated locations in the absence or presence of E4orf4. N/A, not applicable.

E4orf4 expression induces accumulation of G₂/M-phase cells and elevated Clb2-Cdc28/Cdk1 activity.

Previous studies suggested that E4orf4 induces mitotic arrest in both yeast (1, 17, 32) and mammalian (20, 21) cells. To examine this effect more thoroughly, vector control yeast cells or those expressing E4orf4 were grown asynchronously in noninducing glucose medium and then in raffinose medium, before being transferred to inducing galactose-containing medium for 6 h and then analyzed by FACS. Figure 3A shows that, at both 3 and 6 h post-galactose induction, wild-type cells expressing E4orf4 contained a greater population of 2N cells than controls, suggesting that E4orf4 induces an accumulation of G₂/M cells. In the case of cdc55Δ mutant cells, expression of E4orf4 yielded cell cycle profiles similar to those of the vector control, suggesting that the observed effects on the cell cycle by E4orf4 were PP2A^Cdc55 dependent. Figure 3B shows the results of a study using the same cultures described in the legend for Fig. 3A, from which immunoprecipitates prepared using antibodies against Clb2 were tested in vitro for Cdk1 activity using [γ³²P]ATP and histone H1 as the substrate, as described in Materials and Methods. Whereas wild-type cells expressing E4orf4 exhibited levels of Cdk1 activity significantly higher than those of the vector control, the activity in cdc55Δ mutant cells was similar to that of the control, regardless of the presence or absence of E4orf4. Thus, the accumulation of G₂/M cells, in parallel with high mitotic kinase activity, suggests that E4orf4 expression perturbs cell cycle timing and regulation of Clb2-Cdc28/Cdk1 activity by a mechanism that is dependent on PP2A^Cdc55.

FIG. 3.

Effect of E4orf4 on the cell cycle and Clb2-Cdc28/Cdk1 activity. (A) Analysis of cell cycle. Wild-type or cdc55Δ mutant cells transformed with either control plasmid DNA or that expressing E4orf4 were grown asynchronously, and cells harvested at 3 and 6 h postinduction in preparation for FACS analysis, as indicated and described in Materials and Methods. (B) Measurement of Cdk1 activity. WCEs from asynchronous wild-type or cdc55Δ mutant cells transformed with either control plasmid DNA or that expressing E4orf4 were grown for 6 h, and then extracts were immunoprecipitated using antibodies against Clb2. In vitro kinase assays were then performed using [γ³²P]ATP and histone H1 as substrate, as described in Materials and Methods. The figure shows an autoradiograph of analysis by SDS-PAGE of equal aliquots of the kinase assays.

Clb2 overexpression induces high Cdk1 kinase activity and cell death.

The lethality associated with E4orf4 expression in yeast cells could arise from high levels of Clb2-associated Cdk1 activity that we observed with E4orf4-expressing cells (32). To test the hypothesis that uncontrolled activation of Clb2-Cdc28/Cdk1 kinase is toxic, we assayed the viability of cells overproducing Clb2. Cells were transformed with DNA expressing galactose-inducible HA-tagged Clb2 or control vector DNA. After galactose induction, cells were collected, extracts were immunoprecipitated using anti-Clb2 antibodies, and the precipitates were tested for Clb2-Cdc28/Cdk1 kinase activity, as shown in Fig. 3B. Figure 4A (top) indicates the levels of Clb2 detected by immunoblotting using anti-Clb2 antibodies, and Fig. 4A (bottom) shows that Cdk1 kinase activity was greatly increased in cells that overexpressed Clb2 compared to the vector control. To determine the effect of increased Clb2 expression and Cdk1 activity on cell viability, spotting assays were also performed, as shown in Fig. 1B. Figure 4B shows that overexpression of Clb2 severely reduced cell viability. These results demonstrate that the unregulated activity of Clb2-Cdc28/Cdk1 is indeed toxic to cells. This opens the possibility that the Cdc55-dependent toxicity of E4orf4 at least in part arises from high levels of Clb2-Cdk1 activity.

FIG. 4.

Effect of overexpression of Clb2. Cells were transformed with control plasmid DNA or that expressing Clb2-HA₃ under the GAL1 promoter and then cultured in galactose-containing medium to induce Clb2 expression. (A) Clb2 expression and Cdk1 activity. Top, Western blotting using anti-Clb2 antibodies. Bottom, WCEs were subjected to immunoprecipitation with anti-Clb2 antibodies followed by Cdk1 kinase assays, as described in the legend for Fig. 3B. (B) Cell viability. Cells were grown in liquid medium and then spotted on 2% glucose-containing agar plates (noninducing) or 2% galactose-2% raffinose plates (inducing). Photographs were taken after plates were incubated at 30°C for 2 days.

E4orf4 expression correlates with reduced levels of substrates downstream of APC^Cdc20 in S-phase cells.

The observed increase in Clb2-Cdc28/Cdk1 activity due to E4orf4 expression led us to investigate whether inappropriate Cdk1 activation could be decoupling the timing of mitotic progression. This seemed plausible, as phosphorylation of the core APC subunits by Clb2-Cdc28/Cdk1 is known to be required for association of the APC with its cofactor Cdc20, leading to activation of APC^Cdc20 (18, 33). Another possibility could be that E4orf4 expression led to a mitotic arrest, which is consistent with the observation of an accumulation of cells in G₂/M that could have resulted from the high Clb2-Cdc28/Cdk1 activity. To determine whether the former was indeed the case, we tested the capability of E4orf4 to induce APC^Cdc20 activation in cells arrested in S phase by HU. We accomplished this by examining the levels of substrates downstream of APC^Cdc20 and DNA content, with respect to E4orf4 expression.

E4orf4 expression reduces levels of Pds1/securin in S-phase cells.

Pds1/securin is an APC^Cdc20 substrate that is rapidly degraded when APC^Cdc20 is activated. It is well established that degradation of ubiquitinated Pds1 by APC^Cdc20 is required for dissolution of the cohesin complex and sister chromatid segregation at anaphase (4, 5, 41, 42). During S phase, Pds1 is normally stable and blocks the activity of Esp1, which cleaves Scc1. We reasoned that if E4orf4 expression results in unregulated Cdc28-Clb2/Cdk1 activity, premature APC^Cdc20 activation might occur. We tested this using a strain which contained endogenously expressed functional Myc-tagged Pds1 fusion protein (Pds1-Myc₁₈) (4). Cells transformed with either vector control plasmid DNA or DNA expressing FLAG-E4orf4 were blocked in S phase with 0.2 M HU under noninducing conditions (T = 0 h). After half of the culture was collected for analysis, cells were grown under inducing conditions in 0.2 M HU for 3 h (T = 3 h) to induce E4orf4 expression while maintaining the S-phase arrest. Pds1-Myc₁₈ levels in WCEs were determined by Western blotting using anti-Myc antibody and those of E4orf4 using anti-FLAG antibody. FLAG-E4orf4 protein could be detected after 2 h of induction and was maximal by 3 h (data not shown). Figure 5A shows that after 3 h of induction to allow for E4orf4 expression, Pds1-Myc₁₈ in WCEs was reduced compared to that in vector control WCEs, with equal amounts of protein loaded, as demonstrated by Cdc28 levels using anti-Cdc28 antibodies. Figure 5B shows the results of FACS analysis of these cultures, indicating that the majority of cells incubated in the presence of HU were present in a single major fraction typical of S-phase cells, whereas untreated cultures exhibited a biphasic pattern characteristic of asynchronously growing cells. Thus, these results suggest that E4orf4 expression reduces Pds1 levels in S-phase-arrested cells.

FIG. 5.

E4orf4 promotes premature proteolysis of Pds1 (securin) in cells arrested in S phase by HU. Yeast cells containing endogenous Myc-tagged PDS1 (strain 6803) were transformed with either control plasmid DNA or that expressing E4orf4. Cells were synchronized in S phase with 0.2 M HU for 5 h (T = 0 h), after which E4orf4 expression was induced in the continued presence of HU for another 3 h (T = 3 h), as described in Materials and Methods. (A) Analysis by Western blotting. Equal aliquots of cell extracts were subjected to Western blotting using antibodies against Myc (top), FLAG (middle), or Cdc28 (bottom). (B) FACS analysis. Cells from the same samples described in the legend for panel A were analyzed by FACS, as described in Materials and Methods.

E4orf4 expression reduces levels of Scc1/cohesin in S-phase cells.

Normally, Pds1 binds to Esp1 and inhibits its protease activity against the cohesin subunit Scc1 (4, 5). After Pds1 is ubiquitinated by APC^Cdc20 and degraded by the proteasome, Esp1 becomes free to cleave Scc1, thus promoting the dissolution of the cohesin complex and sister chromatid separation (42). To study the effect of E4orf4 on the stability of Scc1-Myc₁₈ (25), vector control and E4orf4-expressing cells were again examined according to the same protocol of HU treatment as that described in the legend for Fig. 5. Figure 6A shows that Scc1-Myc₁₈ levels significantly decreased by 4 h of E4orf4 expression, whereas they remained high in control cells. Equal amounts of protein, as demonstrated by Cdc28 levels, were loaded. Figure 6B shows that by FACS, HU retained the majority of cells in S phase, and Fig. 6C indicates that E4orf4 expression resulted in elevated Clb2-Cdc28/Cdk1 activity. It should be noted that upon repeated analyses, the observed increase in Cdk1 activity was often more modest than that of the experiment shown; however, this difference could be due to the presence of considerably lower levels of Clb2 during S phase. Nonetheless, these results suggested that the observed premature decrease in Pds1 levels in E4orf4-expressing cells appeared to promote the premature release of Esp1, thus resulting in the inappropriate loss of Scc1.

FIG. 6.

E4orf4 promotes premature proteolysis of Scc1 (cohesin) in cells arrested in S phase by HU. An experiment similar to that described in the legend for Fig. 5 was conducted, except the fate of Scc1 (cohesin) was examined with yeast containing endogenous Myc-tagged SCC1 (strain 6565). (A) Analysis by Western blotting. (B) FACS. (C) Cdk1 assays. Whole-cell extracts were immunoprecipitated with anti-Clb2 antibody, and equal aliquots were assayed for Cdk1 kinase activity, as described in the legends for Fig. 3B and 4A.

E4orf4 expression does not cause bypass of S-phase checkpoints.

We observed that expression of E4orf4 in S-phase-arrested cells appears to reduce levels of Pds1 and Scc1, both of which are degraded as a consequence of the activation of APC^Cdc20. This is consistent with E4orf4 expression inducing premature activation of APC^Cdc20. However, if E4orf4 expression causes HU-treated cells to bypass S-phase (DNA damage/replication) checkpoints, Pds1 levels would decrease as a consequence of entry into anaphase. To determine if this was the case, the morphology of spindles labeled with YFP-tubulin was examined in HU-arrested cells after 4 h of E4orf4 induction (see Fig. S2 in the supplemental material). Indeed, HU-arrested cells expressing E4orf4 contained either monopolar or short bipolar spindles located in the mother cell, consistent with an S-phase arrest. This analysis verified that E4orf4 and concomitant uncoupling of Clb2-Cdc28/Cdk1 activity did not allow cells to enter anaphase. Importantly, these results suggest that the E4orf4-dependent reduction of Pds1 and Scc1 in HU-treated cells is due to premature activation of APC^Cdc20.

E4orf4 expression does not reduce Cdc5 and Cdc20 levels in S-phase cells.

Considering that E4orf4 expression appeared to reduce Pds1 and Scc1 levels, studies were conducted to determine if the substrates of the second form of APC, APC^Hct1, were also affected, by examining the levels of two of its substrates, Cdc5 and Cdc20 (2, 11, 28, 35). Yeast strains expressing Cdc5-Myc₁₅ or Cdc20-Myc₁₈ (35, 50) were transformed with either control vector DNA or DNA expressing HA-E4orf4, and experiments with HU-treated cells were conducted as described in the legends for Fig. 5 and 6. Figure 7A shows that after 4 h of E4orf4 expression in the presence of HU, Cdc5-Myc₁₅ levels were similar to those in control cells. Figure 7B shows that similar results were also observed with Cdc20-Myc₁₈, a component of the first APC^Cdc20 and a substrate of APC^Hct1. Again, equal amounts of protein, as demonstrated by Cdc28 levels, were loaded. These results suggested that unlike substrates downstream of APC^Cdc20, those of APC^Hct1, specifically Cdc5 and Cdc20, were not prematurely degraded when E4orf4 was expressed in HU-treated cells.

FIG. 7.

E4orf4 does not induce degradation of APC^Hct1 substrates Cdc5 and Cdc20 in cells arrested in S phase by HU. Experiments similar to those described in the legends for Fig. 5 and 6 were conducted, except the fates of Cdc5 and Cdc20 were examined with HU-treated cells. The levels of Cdc5 (strain 6142) and Cdc20 (strain 425) were determined by Western blotting using antibodies against Myc. (A) Myc-tagged Cdc5; (B) Myc-tagged Cdc20.

Reduction of Pds1 and Scc1 by E4orf4 in S-phase cells is PP2A^Cdc55 dependent.

We have shown previously that much of the toxicity of E4orf4 in yeast requires a functional interaction with the Cdc55 subunit of PP2A, as mutations in E4orf4 that block this interaction also reduce toxicity (32). To establish further the role of PP2A^Cdc55 in the E4orf4-induced reduction of Pds1 and Scc1, studies of HU similar to those described in the legend for Fig. 5 to 7 were performed using two types of E4orf4 mutants. The class I E4orf4 mutant R81AF84A fails to bind B55/Cdc55 and is not toxic in either yeast or mammalian cells (22, 32). The class II mutant K88A is capable of binding B55/Cdc55, but this interaction is believed to be nonfunctional, as it fails to induce significant toxicity. Figure 8A shows that Pds1-Myc₁₈ levels were again decreased following expression of wild-type E4orf4, while with both the R81AF84A and K88A mutants, Pds1-Myc₁₈ appeared to be present at levels similar to those in control cells. We performed similar experiments to probe the levels of Scc1 in the presence of mutant E4orf4. Figure 8C shows that the same outcome was evident with Scc1-Myc₁₈. Equal amounts of protein, as demonstrated by Cdc28 levels, were loaded. These results predict that mutation of E4orf4 should result in increased cell viability. Figure 8B and D show growth assays confirming that these mutants do in fact increase cell viability. These results suggest that the reduction of both Pds1 and Scc1 by E4orf4 is dependent on its functional interaction with Cdc55.

FIG. 8.

Degradation of Pds1 and Scc1 in cells arrested in S phase by HU is dependent on a functional interaction between E4orf4 and Cdc55. An experiment similar to those described in the legends for Fig. 5 and 6 were performed, but, in addition to cells transformed with control vector DNA or DNA expressing FLAG-tagged wild-type E4orf4 (Fig. 7A to D), mutant E4orf4 protein R81AF84A or K88A was also expressed in 6803 (Myc-tagged Pds1, Fig. 7A and B) or 6565 (Myc-tagged Scc1, Fig. 7C and D) cells. The stability of Pds1 and Scc1 was examined as described in the legends for Fig. 5 and 6, and the viability of these cells was determined as described in the legend for Fig. 4B. (A) Pds1 stability; (B) Pds1 cell viability; (C) Scc1 stability; (D) Scc1 cell viability.

DISCUSSION

We have demonstrated in this study that E4orf4 in yeast functions via the Cdc55-specific form of PP2A to uncouple the timing of Clb2-Cdc28/Cdk1 activity, resulting in the S-phase reduction of levels of Pds1, normally a substrate APC^Cdc20 in anaphase, and of Scc1, whose stability depends on the presence of Pds1. These results are summarized in Fig. 9. We believe that the inappropriate reduction of these critical mitotic proteins prior to anaphase by E4orf4 at least partially explains the mechanism by which E4orf4 causes growth defects, ultimately leading to cell death in yeast.

FIG. 9.

Model for activation of Clb2-Cdc28/Cdk1 and APC^Cdc20 by E4orf4. This figure illustrates the steps involved in APC activation, degradation of APC substrates, and the potential role of E4orf4 and PP2A^Cdc55 in this process. E4orf4 induces Cdk1 activity, and PP2A is believed to positively regulate the stability of Swe1 and may negatively regulate formation of APC^Cdc20 via dephosphorylation of core proteins. Cdk1 is known to promote the activation of APC^Cdc20 and to inhibit activation of APC^Hct1. Our results suggest that activation of Cdk1 by E4orf4 expression could occur, at least in part, from the inhibition of PP2A^Cdc55 that could affect Swe1 or other factors that regulate Cdk1 activity. Inhibition of PP2A^Cdc55 could also promote activation of APC^Cdc20. P, phosphorylated.

The effects of E4orf4 expression are exerted principally through the Cdc55 regulatory subunit of PP2A, as demonstrated by the specific functional binding between these two proteins and the distinct localization change of PP2A^Cdc55 in the presence of E4orf4. As Cdc55 controls the intracellular localization of PP2A^Cdc55 (9), the observed localization of PP2A^Cdc55 primarily to the nucleus raises the possibility that E4orf4 redirects PP2A activity toward inappropriate substrates and/or prevents normal PP2A activity against bona fide substrates. Our work with mammalian cells indicated that binding of E4orf4 to the B55/Bα subunit inhibits the formation of active Bα-containing PP2A holoenzymes (20). It is unknown whether or not the inhibited holoenzyme is inactive against all substrates or if novel phosphoproteins become inappropriate targets. Our hypothesis is that the binding of the E4orf4 protein to B55 subunits in mammalian cells blocks the access of at least some substrates to the substrate binding site shared by the PP2A B and C subunits and that this inhibition causes both mitotic arrest and cell killing by E4orf4 (20). Although we have not measured directly the effects of E4orf4 on yeast Cdc55-specific phosphatase activity, because of the high degree of similarity of structure and function with mammalian PP2A (12), we suspect that a similar inhibition may also occur with yeast Cdc55-containing PP2A holoenzymes.

Previous studies had suggested that E4orf4 affects mitotic processes (17, 20, 21, 32), and indeed, these ideas are consistent with the results of the present study. E4orf4 was found to increase Clb2-Cdc28/Cdk1 activity in a manner dependent on its functional interaction with Cdc55. Our studies examining the relative abundance of APC substrates (Pds1, Cdc20, and Cdc5) following expression of wild-type or mutant forms of E4orf4 in HU-treated S-phase cells suggested that the toxicity of E4orf4 results from the premature activation of APC^Cdc20 and the failure to activate APC^Hct1. Comparable results have also been obtained when unsynchronized growing cells were examined (data not shown).

A previous study proposed that APC activity is reduced as a consequence of E4orf4 expression (17). Those authors suggested that the E4orf4-induced inhibition of APCs, in combination with high Clb2-Cdc28/Cdk1 activity, created conflicting signals in the cell that result in death. This hypothesis is in apparent conflict with our present results, which suggest the APC^Cdc20 is inappropriately active in S-phase E4orf4-expressing yeast cells, resulting in reduced levels of Pds1 as well as the Esp1 substrate Scc1. It is important to note that the previous study did not examine the impact of E4orf4 on S-phase levels of proteins whose stability is controlled by the APC. Thus, an explanation for the observed difference between our study and that of Kornitzer et al. (17) is that the impact of E4orf4 on PP2A targeting and activity during S phase may be mechanistically distinct from that in mitosis.

In this study, we show that E4orf4 expression results in an increase in Clb2-Cdc28/Cdk1 activity in S-phase cells. As Cdk1 is known to phosphorylate APC subunits and contribute to APC^Cdc20 activity, it seems reasonable to conclude that the premature activation of APC^Cdc20 seen with E4orf4-expressing cells resulted from the PP2A^Cdc55-dependent increase in Clb2-Cdc28/Cdk1 activity. Thus, one aspect of the toxicity of E4orf4 is that its activity against PP2A uncouples Cdc28 and APC activity from the cell cycle state. This idea is validated by the observation that overexpression of Clb2 was lethal. This proof-of-principle experiment supported the hypothesis that the increased Clb2-Cdc28/Cdk1 activity induced by E4orf4 expression may be a critical step in the mechanism leading to cell death. It is important to consider that our cell viability experiments are based on the analysis of cells that may be perturbed in both S phase and M phase, and the additive effect on these two cell cycle states leads to a cell cycle delay that ultimately fails, returning defective cells to the cell cycle and resulting in death.

The altered localization of PP2A^Cdc55 primarily to the nucleus in E4orf4-expressing cells leads to the possibility of enhanced PP2A^Cdc55 activity directed toward nuclear substrates and reduced activity toward cytoplasmic substrates. How this spatially restricted PP2A^Cdc55 activity affects the regulation of Cdc28, either directly or indirectly, in S-phase cells is still unknown. Our present study would favor a model in which PP2A^Cdc55 is a negative regulator of Clb2-Cdk1 in S phase, possibly invoking the activation of Swe1 and inactivation of Mih1 in the nucleus. Figure 9 summarizes our current model for the possible involvement of E4orf4 in mediating the activation of Clb2-Cdc28/Cdk1. Studies are under way to examine more thoroughly the effects of E4orf4 expression on specific factors that regulate Cdk1 activity.

While APC^Cdc20 appeared to be prematurely activated in the presence of E4orf4, APC^Hct1 was not, as suggested by the stability of its substrates Cdc20 and Cdc5. At present, we are still uncertain if expression of E4orf4, which results in elevated levels of Clb2-Cdc28/Cdk1, prevents the formation of APC^Hct1 through inhibitory phosphorylation or if the activity of this complex itself is inhibited. The simplest hypothesis would be that high levels of Cdk1 maintain Hct1 in a highly phosphorylated state, thus making it incapable of forming a complex with the APC core proteins. Studies are currently under way to examine this question.

E4orf4 induces cell death through PP2A^Cdc55, as cell viability is mostly rescued in cdc55Δ mutant strains (17, 32). This line of evidence is again confirmed in this study. However, with the observation that toxicity is never fully relieved in cdc55Δ mutant cells expressing E4orf4 (Fig. 1B), we recognize the potential existence of secondary E4orf4 targets that may decrease viability in a manner independent of the Cdc55 pathway. Such a possibility is evident in the decreased viability of cdc55Δ rts1Δ mutant cells expressing E4orf4 (Fig. 1B). Without the presence of regulatory subunits, E4orf4 likely can no longer transduce its effects through PP2A and may be doing so through other phosphatases of the same class. This seems plausible, as considerable functional redundancy exists between phosphatases in yeast (34). Thus, the E4orf4-induced toxicity observed with cdc55Δ rts1Δ mutant cells may result from off-target effects. Nevertheless, we believe that in cells expressing PP2A, E4orf4 mediates its toxic effects mostly through the pool of PP2A^Cdc55. This contention is demonstrated in Fig. 1B, in which E4orf4 is shown to be even more lethal in rts1Δ mutant cells. In such a situation, cells presumably contain only the PP2A^Cdc55 form, thus making Cdc55 the predominant pathway by which E4orf4 induces cell death.

In this present study, we have found that E4orf4 uncouples PP2A^Cdc55 regulation of mitotic events, leading to unscheduled degradation of specific APC substrates and a defect in mitotic exit. The role that PP2A plays within mitosis has been well studied over the past few years; however, much is still unknown. Equally unknown is the identity of PP2A substrates. As E4orf4 is believed to be a specific inhibitor of PP2A^Cdc55, it may serve as a unique and powerful tool for identifying the cellular processes and pathways regulated by PP2A^Cdc55.