PP2A^Cdc55 regulates G₁ cyclin stability

Abstract

Maintaining accurate progression through the cell cycle requires the proper temporal expression and regulation of cyclins. The mammalian D-type cyclins promote G₁-S transition. D1 cyclin protein stability is regulated through its ubiquitylation and resulting proteolysis catalyzed by the SCF E3 ubiquitin ligase complex containing the F-box protein, Fbx4. SCF E3-ligase-dependent ubiquitylation of D1 is trigged by an increase in the phosphorylation status of the cyclin. As inhibition of ubiquitin-dependent D1 degradation is seen in many human cancers, we set out to uncover how D-type cyclin phosphorylation is regulated. Here we show that in S. cerevisiae, a heterotrimeric protein phosphatase 2A (PP2A^Cdc55) containing the mammalian PPP2R2/PR55 B subunit ortholog Cdc55 regulates the stability of the G₁ cyclin Cln2 by directly regulating its phosphorylation state. Cells lacking Cdc55 contain drastically reduced Cln2 levels caused by degradation due to cdk-dependent hyperphosphorylation, as a Cln2 mutant unable to be phosphorylated by the yeast cdk Cdc28 is highly stable in cdc55-null cells. Moreover, cdc55-null cells become inviable when the SCF^Grr1 activity known to regulate Cln2 levels is eliminated or when Cln2 is overexpressed, indicating a critical relationship between SCF and PP2A functions in regulating cell cycle progression through modulation of G₁-S cyclin degradation/stability. In sum, our results indicate that PP2A is absolutely required to maintain G₁-S cyclin levels through modulating their phosphorylation status, an event necessary to properly transit through the cell cycle.

Introduction

Transit through the cell cycle requires the activities of several cyclin-dependent kinases (cdk)¹ and the temporal expression of cell cycle-specific cdk regulators, termed cyclins.^2,3 The mammalian D-type cyclins regulate G₁-S cell cycle transition.⁴ Of the three D-type cyclins (D1, D2, D3), cyclin D1 is the most frequently overexpressed in human cancers.⁵ While mechanisms increasing D1 expression, such as gene amplification and altered transcription, are known, it is the loss of cyclin degradation that is thought to be the leading cause of overexpression.⁵ Cyclin D1 ubiquitin-dependent degradation is triggered by residue-specific changes in phosphorylation, which leads to its ubiquitylation by the SCF^{Fbxw/αB-crystallin} E3 ligase complex.⁶ The degree of phosphorylation is thought to be critical for modulating D1 cyclin stability, as the levels of SCF^{Fbxw/αB-crystallin} do not change during the cell cycle. While glycogen kinase 3β regulates the degradation of D1 through phosphorylation of Thr286,⁷ the protein phosphatase(s) required for dephosphorylation of this and possibly other residues is not known. Maintaining proper dephosphorylation is just as critical of an event as maintaining proper phosphorylation, as the loss of dephosphorylation would lead changes in the level of cyclin D1.

In S. cerevisiae, a single Cdk, Cdc28, associates with nine cyclins, including three G₁ cyclins (Cln1-3) and six B-type cyclins (Clb1-6).^8,9 G₁ arrest occurs in the absence of CLN1-3,¹⁰ thus G₁ progression requires the activity of at least one G₁ cyclin.^9,11 All three yeast G₁ cyclins are unstable and rapidly degraded at the end of G₁.^12-14 Phosphorylated forms of Cln1 and Cln2 are ubiquitylated by the SCF^Grr1 ubiquitin ligase complex and degraded by the 26S proteasome.^15,16 The F-box protein Grr1 recognizes the Cdc28-dependent autophosphorylation of multiple Cln1/2 PEST domain sites, resulting in SCF^Grr1-phospho-Cln1/2 association.^12,17-19 This autocatalytic targeting keeps Cln1/2 unstable throughout the cell cycle, including in pre-Start G₁ cells.¹³ Interestingly, deletion of GRR1 causes the accumulation of Cln1 and Cln2 and an elongated phenotype.¹⁷

Protein phosphatase 2A (PP2A) is a heterotrimeric serine/threonine phosphatase comprised of two regulatory subunits (A and B) and a single catalytic subunit (C).^20-23 B subunits direct substrate specificity and subcellular location and are necessary for PP2A to interact with and dephosphorylate a variety of substrates.^22-24 Mammalian PP2A is known to regulate multiple stages of the cell cycle. Cell cycle substrates include the retinoblastoma family of pocket proteins,²⁵ multiple cell factors required for mitotic exit^26,27 and the shugosin protein involved in sister chromosome segregation.²⁸

In S. cerevisiae, PP2A is represented by a single A subunit, Tpd3, two B regulatory subunits, Cdc55 (PPP2R2/PR55) and Rts1 (PPP2R5/PR61), and two catalytic subunits, Pph21 and Pph22.^29-32 Yeast PP2A is a positive regulator of mitosis in budding yeast, as its loss arrests cells at G₂/M.³² PP2A^Cdc55 regulates both the mitotic and meiotic cell cycles,³² thus it is a critical factor regulating cell cycle progression and viability. In trying to understand the reason for a synthetic lethal phenotype associated with loss of both Cdc55 and SCF function, a role for PP2A^Cdc55 was discovered involving the regulation of Cln2 phosphorylation state and stability. Thus, a novel pathway regulating cell cycle progression has been uncovered that involves PP2A-dependent regulation of G₁ cyclin protein phosphorylation state and stability.

Results

GRR1 is required for the viability of cdc55 mutants

A colony-sectoring screen was used to identify mutations synthetic lethal with loss of CDC55.³³ 45 non-sectoring mutants were identified and fell into six complementation groups. The viability of one group was linked to the presence of wild-type CDC55. A synthetic lethal interaction between loss of GRR1 and loss of CDC55 has been observed.³⁴ To determine if there was genetic linkage to GRR1, an isolate from a single cdc55 complementation group was transformed with pRS-GRR1-TRP1 and plated on 5-FOA Trp⁻ media. Cells now sectored and no longer displayed a FOA^s phenotype. A wild-type GRR1 was integrated at its endogenous locus, cells were tested for FOA^s, and identical results were obtained. Thus, a single gene complementation group was isolated harboring a recessive mutation(s) in GRR1. Sequencing of multiple clones revealed a single nonsense mutation located at nucleotide 892 (codon 298), changing C to T, resulting in the generation of a TAG stop codon (grr1^C892T).

Loss of GRR1 is synthetic lethal with the loss of PP2A^Cdc55 and not with the loss of PP2A^Rts1

There are no phenotypes shared by cdc55 and rts1 cells, so most likely little cross-talk occurs between PP2A^Cdc55 and PP2A^{Rts1 29, 30}. However, it was still important to show that the synthetic lethal interaction was specific to loss of PP2A^Cdc55. Thus, it was asked if loss of GRR1 was synthetic lethal with loss of RTS1 or the PPH21 and PPH22 PP2A catalytic subunits. FOA^s was used to assay for synthetic lethality. grr1 rts1 cells were viable, while grr1 pph21 pph22 cells required pRS-URA3-GRR1 for viability (Fig. 1). Thus, grr1 cells survive in the absence of PP2A^Rts1 but require PP2A^Cdc55 activity for viability.

Figure 1. Loss of GRR1 is synthetic lethal with loss of PP2A^Cdc55 but not with loss of PP2A^Rts1. Various strains were streaked onto Ura⁻ or FOA containing synthetic plates and incubated at 30°C for 2 d. All strains carry pGRR1-URA3 to test for the FOA^s phenotype, indicative of synthetic lethality.

Cdc55 synthetic lethality is specific to the SCF^Grr1 ubiquitin ligase complex

As Grr1 is part of the SCF complex, we asked if the synthetic lethal interaction between loss of CDC55 and GRR1 was due to the loss of SCF^Grr1 ubiquitin ligase function, or if it was SCF-independent and caused by loss of GRR1 alone. Thus, it was determined if a synthetic sick phenotype existed between loss of CDC55 and SCF skp1-11, skp1-12 and cdc53-1 temperature-sensitive alleles. skp1-11, skp1-12 and cdc53-1 strains are temperature-sensitive at 34°C. We asked if the loss of CDC55 in strains carrying these alleles caused cells to die at a temperature lower than 34°C. cdc55, skp1-11, skp1-12 and cdc53-1 cells grew at or above 34°C, while cdc55 skp1-11, cdc55 skp1-12 and cdc55 cdc53-1 cells were all inviable at this temperature (Fig. 2A).

Figure 2. Loss of CDC55 in strains harboring recessive SCF alleles reduces the temperature required for cell inviability. (A) Various SCF strains were dropped onto YEPD plates as 10-fold dilutions and grown at the indicated temperatures for 2 d. (B) Various strains harboring recessive F-Box alleles were dropped onto YEPD plates as 10-fold dilutions and grown at the indicated temperatures for 2 d.

The S. cerevisiae genome contains 21 orfs encoding putative or established F-box proteins; all are categorized based on the presence of an F box motif.³⁵ The three most studied are Cdc4, Met30 and Grr1. GRR1 is not essential, while CDC4 and MET30 are required for viability.^36,37 To further define F-box specificity, a genetic sick phenotype was tested for between loss of CDC55 and recessive alleles of CDC4 or MET30. We asked if the loss of CDC55 reduced the temperature required for the cell death of cdc4-1^ts or met30-6^ts cells. cdc55 cdc4-1^ts and cdc55 met30-6^ts cells, like cdc4-1^ts and met30-6^ts cells, grew at 30°C (Fig. 2B). Thus, the synthetic lethality between loss of CDC55 and GRR1 is specific to loss of SCF^Grr1 ubiquitin ligase activity.

SCF^Grr1 substrate binding is required for viability of cdc55 cells

Grr1-bound substrates are targeted for ubiquitylation and subsequent degradation by the 26S proteasome.^38,39 Grr1 substrate binding is essential for SCF^Grr1 function. Thus, it was determined if Grr1 substrate binding was required for cdc55 cell viability. cdc55 cells were used harboring the grr1 alleles isolated by Hsuing et al.,¹⁷ which are able or unable to bind substrate.

The grr1B4Q allele has four basic residues (K498, R550, R680 and R709) that are all mutated to glutamine (Q). The residues are in a pocket formed by the concave surface of the LRR (leucine rich region) of Grr1 and are required for phosphorylated substrate binding (Fig. 3A). The grrR485Q allele has a substituted glutamine (Q) at residue R485, which resides on the convex surface of the LRR; this residue is not required for phosphorylated substrate binding.¹⁷ The grr1ΔC allele has 234 C-terminal residues deleted and is defective in substrate binding. FOA^s was used to test if cells harboring these alleles required PP2A^Cdc55 for viability.

Figure 3. Loss of CDC55 is synthetic lethal in grr1 strains harboring substrate-binding deficient recessive alleles. (A) Schematic of Grr1 and the LRR domain depicting various recessive mutations and C-terminal truncation. (B) Various strains were streaked onto Ura⁻ or FOA containing synthetic plates and incubated at 30°C for 2 d. (C) Table summarizing results and indicating the binding efficiencies of each allele and the stability of Cln2 in strains harboring each allele.

cdc55 cells deleted for GRR1 or harboring the grr1^C868T allele were FOA^s; thus, they cannot segregate the pCDC55-URA3 plasmid, which causes toxicity on FOA plates (Fig. 3B). cdc55 cells carrying the substrate binding defective grr1B4Q or grr1ΔC allele were also inviable, while those carrying the binding proficient grrR485Q allele grew on 5-FOA media. Thus, substrate binding and ubiquitylation activities of SCF^Grr1 are required for cell viability in the absence of PP2A^Cdc55. The results are summarized in Figure 3C.

Cdc55 regulates Cln2 stability

grr1B4Q and grr1ΔC alleles have defects in binding to and degrading Cln2¹⁷ and are synthetic lethal with loss of CDC55 (Fig. 3). Based on these results, we hypothesized that grr1 cdc55 cells die because of aberrant accumulation of Cln2. To begin to test this hypothesis, cdc55 cells were assayed for defects in Cln2 levels. Pulse/chase experiments were performed using a galactose-inducible CLN2-HA allele, and Cln2 stability was determined by western analysis. Cln2 was unstable in wild-type cells, with a half-life of ~30 min, whereas the level of Cln2 in grr1 cells was stable for up to 120 min (Fig. 4A). These results are in good agreement with published work.¹⁷cdc55 cells had a drastically reduced level of Cln2 when compared with wild-type cells, even at early time points (Fig. 4A), while rts1 cells gave similar results as wild-type cells (not shown). The reduction in Cln2 was not due to defects in expression, as northern analyses showed similar expression levels in all three strains (data not shown).

Figure 4. Cln2 instability in cdc55 cells is dependent on phosphorylation state and SCF^Grr1 function. (A) Wild-type Cln2 or a phosphorylation mutant allele (Cln2^4T3S) was pulse/chase expressed in the indicated strains using galactose induction/glucose shutoff. Experiments were performed at 30°C. Cln2 levels were determined by western analysis using anti-HA monoclonal antibodies. (B) Wild-type Cln level was induced and shutoff at the indicated temperatures. Cln2 levels were determined using western analysis. (C) Wild-type Cln2 was induced and visualized as described at the indicated temperatures. The loading control was a stable cross-reactive protein (**). For inducing Cln2-HA expression, 2% galactose was added and cells were grown for 45 min, followed by addition of 2% glucose and growth for 2 h. The zero time point is when glucose was added in order to shutoff Cln2 expression.

Since Cln2 phosphorylation signals rapid degradation,¹² it was determined if Cln2 instability in cdc55 cells was linked to phosphorylation state. A galactose-inducible CLN2-4T3S-HA allele was used, which has all seven putative serine/threonine phosphoacceptor sites required for degradation mutated to alanine; Cln2-4T3S is stable and functional, and its expression in single copy complements the cell cycle defect of a cln1 cln2 cln3 strain.¹² Cln2-4T3S was stable in cdc55 cells, and the level observed was comparable to that seen in wild-type cells (Fig. 4A). Proteolysis of Cln2 is triggered by Cdc28-Cln2 phosphorylation.^19,40,41 Loss of Cdc28 activity should stabilize Cln2 in cdc55 cells. A cdc55 strain was constructed harboring the cdc28-4^ts ts allele, and Cln2 stability was determined.⁴² Cln2 was stabile in cdc55 cdc28-4^ts cells at the restrictive temperature (Fig. 4B).

Cln2 instability in cdc55 cells is due to SCF^Grr1 ubiquitin ligase activity

As there was a genetic interaction between loss of CDC55 and SCF recessive alleles, it was asked if Cln2 instability in cdc55 cells was linked to SCF^Grr1 function. A galactose-inducible CLN2-HA allele was used, and Cln2-HA levels were determined in cdc55 cdc53-1, cdc55 skp1-11 and cdc55 skp1-12 cells at the restrictive temperature. In all cases, Cln2 became stable at high temperature (Fig. 4C, cdc55 cdc53-1 shown). Thus, PP2A^Cdc55 directly or indirectly regulates Cln2 stability through regulating its phosphorylation state, which targets phospho-Cln2 for ubiquitylation by SCF^Grr1.

Suppressor mutations that remediate the morphology defects of grr1 or cdc55 cells cannot cross-suppress

cdc55 cells are elongated at low temperature.^30,43 This phenotype is suppressed by loss of SWE1, a Cdc28 inhibitory kinase, or by expressing the CDC28 allele, CDC28AF, which activates Cdc28 through the loss of the inhibitory phosphorylation sites, Thr18 and Tyr19.^44,45grr1 cells are also elongated, and this phenotype is remediated by deleting CLN1 and CLN2 together.¹⁷ As both cdc55 and grr1 cells exhibit an elongated phenotype, we asked if any crosstalk existed between grr1 and cdc55 suppressors as a means to better understand where in the cell cycle cdc55 grr1 cells may be blocked.

Various null strain combinations were generated, and cell morphology was examined by light microscopy; cdc55 cells of the W303 background have a mild elongated phenotype at 30°C, thus, this temperature was used for phenotypic analysis. cdc55 swe1 and cdc55 CDC28AF cells displayed normal morphologies as compared with cdc55 cells,⁴⁵ as did grr1 cln1 cln2 cells compared with grr1 cells¹⁷ (Fig. 5). Deleting CLN1 or CLN2 alone in cdc55 cells did not suppress the morphology defect. In fact, deleting CLN1 and CLN2 together resulted in inviability. Deleting SWE1 or expressing CDC28AF did not suppress the grr1 morphology phenotype. Thus, no cross-talk exists between grr1 and cdc55 suppressors. It is interesting that deleting CLN1 and CLN2 in cdc55 or grr1 cells has opposite effects on viability and/or morphologies. These results suggest that PP2A^Cdc55 and SCF^Grr1 act antagonistically to precisely regulate critical Cln-dependent cell cycle events.

Figure 5. There is no cross-talk between cdc55 and grr1 morphology suppressor alleles. Various strains were grown in YEPD liquid medium at 30°C to exponential phase. An aliquot of cells were obtained and cell morphology was visualized using Nomarski optic and a Leica DRME fluorescence microscope. Greater than 90% of cells harbored the morphologies shown.

grr1 cdc55-1 are mononuclear at low temperature

cdc55 grr1 cells are inviable, which precluded the use of a double mutant as a tool to determine the effects loss of CDC55 and GRR1 together had on cell cycle progression. To circumvent this limitation, a grr1 strain was constructed harboring the cdc55-1 allele. cdc55-1 cells are viable at 30°C, but inviable at low temperature.³⁰ To begin to determine if and/or where in the cell cycle cdc55-1 grr1 cells were blocked, cells were arrested by incubating them at 30°C in YEPD media containing 200 mM hydroxyurea for 1 h, and subsequently released into YEPD.^46,47 Cells were then grown at either 16° or 30°C; FACS, Nomarski microscopy and DAPI staining, were used to monitor cell cycle progression. Unfortunately, FACS analysis was unsuccessful in indicating if and where cells were arrested. This is most likely due to the aberrant cell morphologies of cdc55, grr1 and cdc55-1 grr1 cells. DAPI staining did indicate that cdc55-1 grr1 cells remained mononuclear at low temperature, as did cdc55-1, while wild-type cells continued through the cell cycle.³⁰ Based on the results, we cannot definitively say whether cdc55-1 grr1 cells are arrested, or if they are, where in the cell cycle arrest occurs.

Cln2 and PP2A^Cdc55 physically interact

If PP2A^Cdc55 directly regulates Cln2 phosphorylation state and stability, a physical interaction should be observed between the two proteins. Co-immunoprecipitation experiments were used to determine if PP2A^Cdc55 associated with Cln2. It was first asked if Cdc55-myc pulled down a heterotrimeric PP2A species. Both HA-Pph21 and Tpd3 associated with Cdc55-myc (Fig. 6A). Next, Cdc55-myc was tested for its ability to pull down HA-Cln2 and/or HA-Cln2^4T3S. Both forms of Cln2 co-immunoprecipitated with Cdc55-myc (Fig. 6B); the association with wild-type Cln2 was weak; this is most likely due to its decreased steady-state level (Fig. 6B, 5% input Cln2 vs. Cln2^4T3S). Cln2 binding to Rts1 was not seen (not shown). Pph21-myc and/or Tpd3 were next tested for the ability to pull down HA-Cln2 and/or HA-Cln2^4T3S (Fig. 6C and D). Both HA-Cln2 and HA-Cln2^4T3S were found to be associated with Pph21-myc (Fig. 6C) and Tpd3 (Fig. 6D). Thus, Cln2 associates with a heterotrimeric PP2A^Cdc55 species.

Figure 6. Cln2 and PP2A^Cdc55 can co-immunoprecipitate. (A) Cln2 was induced for 3 h in galactose containing media and expression was shutoff by the addition of glucose for 1 h. Cell extracts were isolated and immunoprecipitated using an anti-myc monoclonal antibody (9E10). Pph21 was visualized using an anti-HA monoclonal antibody. Tpd3 was visualized using anti-Tpd3 polyclonal antibodies. Cdc55 was visualized using an anti-myc monoclonal antibody (9E10). con, wild-type strain harboring p3HA-PPH21; cln2, the control strain harboring YCp-GAL1-CLN2-HA-URA3 and pRS406-ADH-CDC55-13MYC; Cln2^4T3S, the control strain harboring YCp-GAL1-CLN2-4T3S-HA-URA3 and pRS406-ADH-CDC55-13MYC. (B) Cell extracts were isolated and co-immunoprecipitated as described above. con, wild-type strain harboring YCp-GAL1-CLN2-HA-URA3; cln2, the control strain harboring pRS406-ADH-CDC55-13MYC; Cln2^4T3S, a strain harboring YCp-GAL1-CLN2-4T3S-HA-URA3 and pRS406-ADH-CDC55-13MYC. Cdc55 was visualized using an anti-myc monoclonal antibody (9E10). Cln2 was visualized using an anti-HA monoclonal antibody. 5% of the cell lysate input is shown. (C Cln2 was immunoprecipitated with anti-myc antibodies. (D) Cln2 was immunoprecipitated with anti-Tpd3 antibodies.

Cln2 overexpression in cdc55 cells is toxic

One explanation for the results obtained so far is that accumulation of Cln2 in cdc55 grr1 cells is responsible for or aids in cell death. To further this hypothesis, we tested if the hyper-accumulation of Cln2 in cdc55 cells resulted in inviability; we hypothesized that accumulation of Cln2 would overload SCF^Grr1, causing the accumulation of “free” Cln2 in cdc55 cells mimicking what might occur in cdc55 grr1 cells. To overexpress Cln2, the galactose-inducible CLN2-HA allele was once again used, and cell growth was determined on galactose medium after several days. Wild-type cells tolerated the overexpression of Cln2, whereas its overexpression in grr1 cells was lethal (Fig. 7A). These results are in good agreement with published work.¹² cdc55 cells exhibited a grr1-like sensitivity to Cln2 accumulation. They could not tolerate overexpression of Cln2.

Figure 7. The overexpression of wild-type Cln2 or Cln2^4T3S in cdc55 cells is lethal. Various strains were streaked onto glucose or galactose containing synthetic medium. Strains were grown at 30°C for 2 d. Strains harbor the indicated plasmids. Growth on galactose results in the constitutive overexpression of the specific CLN2 allele.

If overexpressing Cln2 in cdc55 cells approaches the level accumulated in cdc55 grr1 cells, and the inability to degrade Cln2 causes cell death, then overexpressing GRR1 in cdc55 cells overexpressing Cln2 should suppress lethality. GRR1 and CLN2-HA were co-overexpressed in cdc55 cells, and viability was determined on galactose medium. As predicted, the increased expression of GRR1 suppressed the Cln2-dependent cell death of cdc55 cells (Fig. 7B). Based on the results as a whole, we suggest that hyperphosphorylated Cln2 accumulates in grr1 cdc55 cells, and this contributes to or is responsible for cell death.

Discussion

Conclusions

We have uncovered a relationship between PP2A activity and SCF-dependent degradation that is critical for maintaining G₁ cyclin levels and G₁-S transition. PP2A^Cdc55 directly dephosphorylates the G₁ cyclin, Cln2, thus maintaining its stability, while SCF^Grr1 is responsible for initiating phosphorylation-dependent ubiquitylation and degradation, both of which are necessary for the inactivation of G₁ cyclin-cdk activity and subsequent activation of a cyclin-cdk activity required for G₂-M.⁹ As mammalian cyclin D1 is overexpressed in human cancers,^5,48,49 and its stability is associated with defects in SCF activity,⁵⁰ targeting PP2A through inhibiting its activity may represent a plausible therapeutic for cancer treatment.

G₁ cyclins regulate cell morphology and migration

The overexpression of cyclin D1 seems to be necessary for increased cell migration and metastasis.⁴ D1 overexpression inhibits Rho-activated kinase II and thrombospondin 1, factors involved in maintaining proper cell motility and adhesion by regulation of the cytoskeleton.⁵¹ Increased D1 expression also causes enhanced cytokinesis of mammary epithelial cells.⁵¹ A D1-dependent decrease in abundance of the Skp2 component of the SCF complex may be one factor responsible for the increase in cytokinesis.⁵¹ In S. cerevisiae, loss of Cdc55 results in enhanced activation of the pseudohyphal invasive growth pathway.^52,53 Moreover, cdc55 cells have morphological and cytokinesis defects at low temperature, giving rise to elongated daughter cells, some of which lack nuclei.³⁰ Loss of the Bem2 GTPase-activating protein required for Rho1/2 GTPase activity and bud site selection suppresses the morphology defect of cdc55 cells.³⁴ Interestingly, SCF^Grr1 function is required for actomyosin contraction during cytokinesis,⁵⁴ and the elongation defect of grr1 mutants can be suppressed by deletion of Cln1/2,^15,46 while Grr1 deletion suppresses the loss of Bem2.³⁴ Together, these results indicate that the yeast SCF/PP2A pathway links the cell cycle with morphogenesis, cell polarity and cell cycle progression. Similar defects to those described above, including increased cell migration and metastasis, are seen in mammalian cells when D1 levels are perturbed. Thus, SCF^Grr1 and PP2A^Cdc55, like their mammalian counterparts, are at the center of an axis that links temporal changes in G₁ cyclin levels with multiple pathways regulating cell size, morphology and invasiveness.

Cyclin regulation by phosphorylation

The instability of Cln2 was linked to its phosphorylation state, as a phosphorylation-deficient Cln2 was highly stable in cdc55 cells. C-terminal phosphorylation/dephosphorylation regulates cellular localization.^55,56 Cln2 localizes primarily to the cytoplasm but when hypophosphorylated, accumulates in the nucleus. This suggests Cln2 has a period of nuclear localization early on in the cell cycle.⁵⁶ Interestingly, the Cln2^4T3S phosphorylation mutant accumulates in the nucleus; its accumulation is toxic to wild-type cells (P. McCourt and J.T. Nickels, unpublished data). As mentioned above, cyclin D1 is subject to phosphorylation by glycogen kinase 3β, which targets it for degradation by SCF^{Fbx4/αβ/crystallin 6, 7}. A D1 cyclin Thr286Ala mutant accumulates in the nucleus and causes enhanced anchorage-independent growth.⁷

cdc55 and grr1 cells have an elongated phenotype

The cs phenotype of cdc55 cells is the result of increased stability of Swe1, the S. cerevisiae ortholog of S. pombe wee1.⁴⁴ Altering the cell cycle by loss of SWE1 or expression of a CDC28 allele lacking inhibitory phosphorylation suppresses the cs phenotype. grr1 mutants are elongated due to the accumulation of Cln proteins, as loss of CLN1 and CLN2 together is sufficient to suppress the elongated morphology.¹⁷ Cln2 overexpression in wild-type cells is not toxic and does not cause any noticeable phenotype.¹⁷ So Cln2 accumulation may act synergistically with another cell event(s), made defective by loss of CDC55 and GRR1 together, where its loss alone does not kill grr1 cdc55 cells but sensitizes them to Cln2-dependent death.

Cyclin regulation in mammals

The stabilities of the mammalian G₁ cyclins, D1 and E, depend on the activities of specific SCF complexes.^57,58 Evidence points to SCF^Fbw8 and SCF^{Fbx4-α/B-crystallin} regulating cyclin D1 stability,^59,60 while cyclin E stability depends on SCF^Fbw7 and the BCR (BTB-cul3-Rbx1) ubiquitin ligase complexes.^58,61 Cyclins D1 and E are highly stable in a number of cancers, a phenotype that is thought to aid in cell cycle proliferation and manifestation of the cancer phenotype.¹ In some cases, increased stability is associated with F-box gene mutations and deregulation of ubiquitin ligase activity.⁶²

There are only a few reports looking at the roles of PP2A and PP2A-like phosphatases in regulating cyclin stability. One report demonstrated that the serine/threonine phosphatase PP6 regulates cyclin D1 protein, where its transient expression in prostate cancer PC-3 cells caused a G₁ cell cycle arrest,⁶³ whereas knockdown of the PPP2CA α catalytic subunit causes an increase in cyclin D1 protein expression.⁶⁴ We now show that yeast PP2A^Cdc55 regulates the stability of the G₁ cyclin, Cln2, through regulating its phosphorylation state, which targets it for degradation by SCF^Grr1. These results indicate that PP2A activity in yeast and mammals must be maintained in order to initiate proper temporal expression and concomitant cyclin protein accumulation and, ultimately degradation, ensuring proper progression through the cell cycle, proper regulation of cell migration and maintaining proper cell morphology.

Materials and Methods

Strains and plasmids

Most of the S. cerevisiae strains used in this study were derived from W303 (MATa ura3-1 leu2-3,112 his3-11,15 trp1-1 ade2-1 can1-100). For colony sectoring we used YJN2022 (MATa ade2 ade3 leu2 ura3 lys2 can1). For the colony-sectoring assay, YJN2022 was transformed with YCp-CDC55-ADE3-URA3, YCp-CDC55-LEU2 or YCp-GRR1-TRP1. The plasmids YCp-GAL1-CLN2-HA-URA3 or YCp-GAL1-CLN2-4T3S-HA-URA3 were used to examine Cln2 stability and the overexpression of Cln2 (gifts from Dr Stefan Lanker, Department of Medical and Molecular Genetics, Oregon Health and Science University). The vectors YCp-GAL1-CLN2-HA-TRP1, YCp-GAL1-CLN2-4T3S-HA-TRP1, YCp-GAL1-GRR1-MYC-URA3 and YCp-TRP1, were constructed during this study and used to examine growth defects caused by the overexpression of Cln2. Yeast transformations were performed using the procedure described by Ito et al.⁶⁵ E. coli XL1-Blue cells were used for routine plasmid propagation and grown in LB medium supplemented with ampicillin (200 µg/ml).

W303 strains harboring the grr1 mutant alleles, grr1B4Q, grr1R485Q, grr1ΔC and grr1::LEU2, were gifts from Dr Curt Wittenberg (Department of Molecular Biology, The Scripps Research Institute).¹⁷ Deletion strains were generated by the one-step disruption method of Rothstein.⁶⁶ CDC28AF strains were generated using pRS405-CDC28AF as previously described.⁶⁷ The cdc53-1 strain was a gift from Dr Katrina Cooper (Department of Cell and Molecular Biology, UMDNJ-SOM); the cdc28-4 strain was a gift from Dr Mark Rose (Department of Molecular Biology, Princeton University); and the skp1-11, skp1-12, met30-6 and cdc4-1 strains were gifts from Dr Steve Elledge (Department of Genetics, Harvard Medical School). Congenic W303 strains harboring these alleles were generated by several backcrosses and selection based on temperature sensitivity.

Media and growth conditions

Yeast strains were grown in YP media (1% yeast extract, 2% bactopeptone) supplemented with 2% glucose (YEPD), 2% raffinose or 2% galactose, or in synthetic minimal media containing 0.67% yeast nitrogen base supplemented with the appropriate amino acids and carbon source. For analyzing the genetic interaction between CDC55 and SCF, cells were grown to exponential phase in YEPD, and 2 × 10⁶ cells were dropped as 10-fold serial dilutions onto various media plates and incubated at 30, 34 and 37°C. Cell growth was examined after 72 h.

For analyzing the genetic interaction between grr1 alleles and CDC55, strains were transformed with YCp-GRR1-URA3, grown exponentially on Ura⁻ containing 2% glucose solid medium and then transferred onto solid media containing 1 mg/ml 5-fluoroorotic acid (5-FOA) and plasmid-dependent growth was determined as described.⁶⁸

To assay for the viability of strains overexpressing CLN2, cells were transformed with YCp-URA3, YCp-GAL1-CLN2-HA-URA3 or YCp-GAL1-CLN2-4T3S-HA-URA3, and grown on Ura⁻ containing glucose or Ura⁻ containing 2% galactose for 72 h. To test whether overexpression of GRR1 suppresses the lethality caused by CLN2 overexpression, strains were transformed with YCp-TRP1 or YCp-GAL1-GRR1-MYC-TRP1. It was examined if GRR1 overexpression could suppress the lethality caused overexpression of CLN2 by growth on Ura⁻ Trp^- containing 2% glucose or Ura⁻ Trp⁻ containing 2% galactose for 72 h.

Colony sectoring screen for mutations synthetic lethal with loss of CDC55

The colony-sectoring screen uses colony color selection based on adenine auxotrophy.^69,70 MATa ade2 ade3 leu2 ura3 lys2 can1 cdc55::KanMX2 (YJN2023) and MATα ade2 ade3 leu2 ura3 his3 can1 cdc55::KanMX2 (YJN2024) were transformed with plasmid, YCp50-CDC55-ADE3-URA3 [red colonies (cells harboring the plasmid)] are formed that contain white sectors {[cells that have lost (sectored) the plasmid]; as CDC55 is not an essential gene}. Mutations synthetic lethal with loss of CDC55 were generated using 2% ethane methyl sulfonate (EMS) mutagenesis (90% killing efficiency). Surviving cells were grown on YEPD for 5 d at 30°C in order to visualize the pure red colony phenotype. Approximately 145,000 colonies were screened. Positive clones were selected and tested for red phenotype stability on YEPD. Potential clones were then re-tested for CDC55 dependence by growth on 5-FOA plates. To determine dominance vs. recessive mutations, positive clones were mated to YJN2023 or YJN2024, and resulting diploids were analyzed for red color phenotype on YEPD. Complementation groups were determined by analyzing the red/white colony-sectoring trait of diploids produced by cross-mating positive haploid clones of the opposite mating type. As a final selection for positive clones, strains inviable on 5-FOA were transformed with YCp-CDC55-LEU2, grown exponentially in YEPD and plated onto Leu⁻ 5-FOA plates. Viable haploids were selected and considered to be synthetic lethal with CDC55.

Determination of Cln2 stability

Cultures were grown to exponential phase in Ura⁻ containing 2% raffinose at 30°C. For inducing Cln2-HA expression, 2% galactose was added, and cells were grown for 45 min, followed by addition of 2% glucose and growth for 2 h. Strains carrying the temperature-sensitive alleles, cdc28-4 or cdc53-1, were shifted to 34°C or 37°C, respectively, for 30 min before expression of Cln2-HA was induced. Cln2-HA level was determined by western analysis and immunoblotting using 12CA5 mouse monoclonal anti-HA antibody.

Western analysis

For Cln2 westerns, total cell extracts were isolated using a modified procedure from Hsiung et al.¹⁷ Exponential grown cells were pelleted and resuspended in yeast lysis buffer (50 mM TRIS-HCl, pH 7.5, 150 mM NaCl, 0.1% NP-40, 10% glycerol) containing 50 mM sodium fluoride, protease inhibitors (0.4 mM phenylmethylsulfonyl fluoride, 1 µg/ml pepstatin, 1 µg/ml leupeptin and 1 µg/ml aprotinin) and phosphatase inhibitors (0.1 mM sodium orthovanadate, 5 mM EDTA, 5 mM EGTA and 10 mM sodium pyrophosphate). Cells were lysed with glass beads using seven cycles of vortexing for 1 min followed by 1 min incubation on ice. Total cell extracts were obtained by centrifugation for 5 min at 3,000 rpm. 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 (10mM TRIS-HCl pH 7.4, 150mM NaCl) + 0.05% Tween20. 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 4× after antibody incubations with Buffer A containing 0.05% Tween-20. The primary antibody was 12CA5 monoclonal mouse anti-HA (1:1,000 dilution), and the secondary antibody was a polyclonal goat anti-mouse HRP (Amersham; 1:2,000 dilution). Proteins were detected using ECL (Amersham). A stable non-specific cross-reactive band was used as a loading control.¹⁷

Co-IP method and western analysis

Yeast cells were grown overnight in medium containing 2% raffinose at 30°C to 0.3 OD600, followed by induction with 2% galactose for 3 h. Cells (6 × 10⁸) were collected by centrifugation at 2,000 rpm for 5 min. For co-immunoprecipitation of Cdc55-Myc13 with Cln2, TPD3 and Pph21, cells were lysed with glass beads in lysis buffer (50 mM of Tris-Cl, pH 7.4, 50 mM of NaCl, 2 mM of EDTA, 1 mM of DTT, 0.5 mM of PMSF and protease inhibitor cocktails). Cell debris and unbroken cells were removed by centrifugation at 3,000 rpm for 5 min at 4°C. The supernatant was incubated with 0.5% of Triton X-100 on ice for 30 min. Three μl of 9E10 antibody was added and incubated on ice for 3 h. Twenty μl of protein A beads (blocked with 1% BSA) were then added, followed by rotation at 4°C for 90 min. After washing 3× with lysis, buffer plus 1% Triton X-100 and once with 20 mM Tris-Cl, pH 7.4, the beads were resuspended in 40 μl of 2X SDS loading buffer, incubated at 95°C for 5 min and resolved by SDS-PAGE for western blotting. The parental yeast strain used was Y162 (MATα, pph21::HA3-PPH21 ura3 leu2 trp1 his3 ade2 can1). The Cdc55-13myc plasmid used for endogenous integration was JB1314 (Cdc55-13myc-PRS406) and was linearized using NheI. Pph21 was visualized using 12CA5 monoclonal mouse anti-HA (1:1,000 dilution). Tpd3 was visualized using anti-Tpd3 polyclonal antibodies (1:500 dilution). Cdc55-13myc was visualized using 9E10 monoclonal mouse anti-MYC (1:1,000 dilution).

Cell cycle analysis

Strains were grown to exponential phase in YEPD at 30°C. Cells were then transferred to YEP media and incubated for 12 h at 30°C. Cells were arrested with 200 mM hydroxyurea. They were shifted to 16 and 30°C for 1 h prior to cell cycle release. Cells were then incubated at 16 and 30°C; FACS and light microscopy and DAPI staining were used to monitor cell cycle progression. Viability was determined at all time points using plate viability assays.

Disclosure of Potential Conflicts of Interest

No potential conflicts of interest were disclosed.