Oncogene (2008) 27, 3176–3185 & 2008 Nature Publishing Group All rights reserved 0950-9232/08 $30.00 www.nature.com/onc ORIGINAL ARTICLE Skp2 regulates the antiproliferative function of the tumor suppressor RASSF1A via -mediated degradation at the G1–S transition

MS Song1,4, SJ Song1,4, SJ Kim1, K Nakayama2, KI Nakayama3 and D-S Lim1

1Department of Biological Sciences, Korea Advanced Institute of Science and Technology, Daejeon, Korea; 2Division of Developmental Genetics, Center for Translational and Advanced Animal Research on Human Diseases, Tohoku University Graduate School of Medicine, Sendai, Miyagi, Japan and 3Department of Molecular and Cellular Biology, Medical Institute of Bioregulation, Kyushu University, Fukuoka, Japan

The tumor suppressor RASSF1A is inactivated in many consists of the essential components Skp1, Cul1, Rbx1 human cancers and is implicated in regulation of and an F-box protein, the latter of which binds to Skp1 microtubule stability, progression and apoptosis. through its F-box motif and serves to recognize and However, the precise mechanisms of RASSF1A action recruit target proteins (Cardozo and Pagano, 2004). The and their regulation remain unclear. Here we show that F-box protein Skp2 has been implicated in the Skp2, an oncogenic subunit of the Skp1–Cul1–F–box ubiquitin-dependent degradation of p27, E2F-1, free complex, interacts with, ubiquitinates, and cyclin E and FOXO1 (Marti et al., 1999;Nakayama promotes the degradation of RASSF1A at the G1–S et al., 2000;Huang et al., 2005). The APC–Cdh1 transition of the cell cycle. This Skp2-dependent destruc- complex degrades Skp2 at the onset of G1 phase, after tion of RASSF1A requires phosphorylation of the latter which Skp2 begins to accumulate late in G1 and its on serine-203 by cyclin D–cyclin-dependent kinase 4. abundance peaks during S and G2 phases (Bashir et al., Interestingly, mutation of RASSF1A-phosphorylation site 2004;Wei et al., 2004). 203 Ser to alanine results in a delay in cell cycle progression Cell cycle progression from G1 to S phase is governed from G1 to S phase. Moreover, enforced expression of by cyclin-dependent kinase (Cdk) 4 as well as Cdk2. Skp2 abolishes the inhibitory effect of RASSF1A on cell Cdk4 is activated by D-type cyclins at early to mid-G1 proliferation. Finally, the delay in G1–S progression after phase, whereas Cdk2 is activated by E- and A-type Skp2 removal is normalized by depletion of RASSF1A. cyclins during late G1 and S phase, respectively. The These findings suggest that the Skp2-mediated degradation activities of Cdk4 and Cdk2 are constrained by the p16 of RASSF1A plays an important role in cell proliferation Ink4a family and the p21 Cip/Kip family inhibitors, and survival. respectively (Sherr and Roberts, 1999). Cyclin D–Cdk4 Oncogene (2008) 27, 3176–3185;doi:10.1038/sj.onc.1210971; phosphorylates the retinoblastoma protein pRb, pRb- published online 10 December 2007 related proteins p107 and p130, and Smad3 (Sherr and Roberts, 1999;Matsuura et al., 2004). In addition, Keywords: Skp2;RASSF1A;cyclin D-Cdk4;ubiquiti- cyclin D–Cdk4 titrates p27 and p21 to cyclin D–Cdk4, nation;G 1–S transition thereby triggering the activity of the cyclin E–Cdk2 holoenzyme (Sherr and Roberts, 1999). Consistent with their growth-promoting functions, the cyclin D1 and the Cdk4 were found to be amplified in many human cancers (Bartkova et al., 1995;An et al., 1999). Introduction RASSF1A is a tumor suppressor gene on chromo- some 3p21 that is inactivated in many types of Two major types of ubiquitin ligase contribute to carcinoma (Dammann et al., 2000). Moreover, knock- regulation of cell cycle progression: the Skp1–Cul1– out mice defective for RASSF1A are prone to tumor F–box protein (SCF) complex is implicated in regulation development (Tommasi et al., 2005;van der Weyden of the G1–S transition, and the anaphase-promoting et al., 2005), further confirming that RASSF1A func- complex (APC) is required both for separation of sister tions as a tumor suppressor. chromatids at anaphase and for exit from M phase into Apparently RASSF1A appears to function in diverse G1 (Zachariae and Nasmyth, 1999). The SCF complex series of critical cellular processes involved in tumor suppression. RASSF1A has been shown to regulate Correspondence: Professor D-S Lim, Department of Biological mitotic progression by inhibiting the activity of APC- Sciences, Korea Advanced Institute of Science and Technology, Cdc20 (Song et al., 2004) and affect microtubule 373-1 Guseong-D. Yuseong-Gu, Daejeon 305-701, Korea. dynamics (Vos et al., 2004). It binds the p120 E4F E-mail: [email protected] 4These authors contributed equally to this work. factor that forms a complex with both Received 11 July 2007;revised 15 October 2007;accepted 5 November the Rb and p53 tumor suppressors (Fenton et al., 2007;published online 10 December 2007 2004). Furthermore, RASSF1A induces a G1 arrest by Regulation of RASSF1A by Skp2 at G1–S transition MS Song et al 3177 engaging the Rb family G1–S checkpoint through by immunoblot analysis. The abundance of endogenous inhibition of cyclin D1 (Shivakumar et al., 2002). RASSF1A in U2OS cells decreased markedly during S RASSF1A is also involved in apoptosis by linking phase and increased again as the cells approached the death receptor signaling to the apoptotic machinery G2–M transition (Figure 1a). Analysis of G1–S cell cycle (Baksh et al., 2005;Oh et al., 2006;Matallanas et al., progression after serum deprivation and restimulation in 2007). However, the precise mechanism by which NIH3T3 cells also showed the change in the level of RASSF1A is regulated remains unclear. We now show RASSF1A protein at the G1–S transition (Supplemen- that Skp2 interacts with and promotes the ubiquitin- tary Figure 1). We next determined whether this mediated degradation of RASSF1A, thereby inhibiting downregulation of RASSF1A expression in S phase is tumor suppressor function of RASSF1A. due to protein destabilization by measuring the half-life of RASSF1A during mitosis and at the G1–S transition. Endogenous RASSF1A was more unstable at G1–S, with a half-life substantially less than 1 h, than during Results mitosis (Figure 1b). In addition, the proteasome inhibitor MG132 prevented the downregulation of The tumor suppressor RASSF1A is degraded at the G1–S RASSF1A expression at G1–S (Figure 1c). These transition observations thus indicated that RASSF1A is degraded Previous studies have suggested a growth-inhibitory by the proteasome beginning at the G1–S transition. function of RASSF1A at G1 (Shivakumar et al., 2002; Rong et al., 2004) and during mitosis (Song et al., 2004; Vos et al., 2004). Thus, we reexamined the ability of Skp2 interacts with and promotes ubiquitination and RASSF1A to affect the cell cycle progression in various degradation of RASSF1A cell lines (U2OS, SaoS2, H1299, A549, MCF7, HeLa The changes in RASSF1A abundance during the cell and 293T). Interestingly, we found that ectopic expres- cycle were similar to those of the Cdk inhibitor p27 and sion of RASSF1A induced a delay in G1–S cell cycle opposite to those of Skp2 (Figure 1a). We thus progression in U2OS, A549 and MCF7 cells, whereas investigated the possible role of Skp2 in the degradation enforced expression of RASSF1A in SaoS2, H1299, of RASSF1A. To address this issue, we first assessed HeLa and 293T cells increased the cell population with a whether Skp2 interacts with RASSF1A with the use of a G2–M (data not shown). Consistent with a previous co-immunoprecipitation assay. Skp2 was the only F-box report (Rong et al., 2004), RASSF1A likely modulates protein among those tested (Skp2, b-TrCP1, b-TrCP2, both the G1–S and mitotic cell cycle regulation by two Fbw7, Fbl5, Fbl12) that interacted with RASSF1A in independent mechanisms in a cell line context-dependent vivo and in vitro (Figure 2a;data not shown). Forced manner. To further examine the contribution of expression of Skp2 resulted in a marked decrease in the RASSF1A regulation at G1–S cell cycle progression, steady-state amount of RASSF1A in U2OS cells, we monitored RASSF1A expression during the cell cycle whereas overexpression of other F-box proteins (data

Nocodazole release time (h)

123 4 58106 7 9 11 12 13 14 15 RASSF1A p27

Skp2

Cyclin B

β-Actin

MG1 SG2-M

Mitosis G1-S MG132

RASSF1A RASSF1A β -Actin β-Actin CHX (h): 01042 4 1 2

Figure 1 RASSF1A is degraded at the G1–S transition of the cell cycle. (a) U2OS cells were released from nocodazole block (400 ng mlÀ1, 18 h) at indicated time point and subjected to immunoblot analysis with antibodies to the indicated proteins. (b) U2OS cells synchronized in mitosis by nocodazole block or at G1–S by thymidine block were released for the indicated times in the presence of cycloheximide (CHX, 50 mgmlÀ1) and then subjected to immunoblot analysis with anti-RASSF1A and -b-actin. (c) U2OS cells synchronized at G1–S were released from nocodazole block for 9 h in the absence or presence of 1, 5 or 10 mM MG132 and then subjected to immunoblot analysis with anti-RASSF1A and -b-actin.

Oncogene Regulation of RASSF1A by Skp2 at G1–S transition MS Song et al 3178

Lysate IP: Skp2IP: IgG

∆F) RASSF1A Lysate IP:FlagIP: IgG

VectorMyc-Skp2Myc-Skp2( * Cul1(short exposure) Skp2 RASSF1A Cul1(long exposure)

* IVT-Skp2: - + - + + Myc-Skp2 Skp1 IVT-RASSF1A: - - + + + Myc-Skp2(∆F) * Skp2 Heavy chain RASSF1A β-Actin Flag-RASSF1A

Skp2 Asn G1/S

Flag-RASSF1A : - + - +

Input His-Ubiquitin : + + + + IP: IgG IP: Skp2

Cul1 : - + + Ub-RASSF1A Rbx1 : - + + conjugates Skp1 : - + + Skp2 : - - + *

Input Flag-RASSF1A

RASSF1A-Ub 2.0 conjugates 1.6 1.2 0.8 35 S-RASSF1A 0.4 RASSF1A(fold)

Relative Ubiquitinated 0

siRNA: GFP Skp2

Time (days): 1 1 2 3 Skp2+/+ Skp2-/- RASSF1A RASSF1A p27 β-Actin Skp2 CHX (h):0241 01 2 4

β-Actin

Figure 2 Skp2 interacts with and induces ubiquitination and degradation of RASSF1A. (a) Lysate from HeLa cells was subjected to immunoprecipitation (IP) with anti-Skp2 or control immunoglobulin G (IgG), and the resulting precipitates as well as cell lysate were subjected to immunoblot analysis with anti-RASSF1A and anti-Skp2 (top). Immunoprecipitates with anti-Skp2 from the mixture of in vitro translated (IVT) Skp2 and RASSF1A were subjected together with the input proteins to immunoblot analysis with antibodies to RASSF1A or to Skp2 (bottom). (b) U2OS cells were transfected with a vector for Myc-tagged wild-type Skp2 or for an Myc-tagged Skp2 mutant (DF) that lacks the F box. The cells were subsequently synchronized at G1–S and subjected to immunoblot analysis with antibodies to RASSF1A, Myc or b-actin. Asterisk is nonspecific band. (c) Lysates from 293T cells transfected with a vector for Flag-RASSF1A were immunoprecipitated with anti-Flag, and the resulting precipitates were subjected to immunoblotting with anti-Cul1, anti-Skp2, anti-Skp1 and anti-Flag. Asterisks are nonspecific bands. (d) Ubiquitination reactions were performed with in vitro translated Cul1, Rbx1, Skp1 and Skp2 together with ubiquitin (Ub) and E1 and E2 enzymes. The ubiquitination of 35S-labeled RASSF1A was detected by SDS–PAGE and autoradiography. (e) 293T cells transfected with a vector for His-tagged ubiquitin and Flag-RASSF1A were synchronized with or without thymidine (G1–S) and then incubated with MG132 (20 mM) for 6 h before harvest. Cell lysates were immunoprecipitated with anti-Flag. The ubiquitination of RASSF1A was detected by immunoblotting with anti-Ub. An asterisk represents heavy chain. (f) U2OS cells were transfected for the indicated times with a vector for Skp2 small interfering RNA (siRNA) or for a control green fluorescent protein (GFP) siRNA, after which the abundance of the indicated proteins was examined by immunoblot þ / þ –/– analysis. (g) Skp2 or Skp2 MEFs were synchronized at G1–S by thymidine block and released in the presence of cycloheximide (50 mgmlÀ1) for the indicated times, after which the abundance of RASSF1A was examined by immunoblot analysis.

Oncogene Regulation of RASSF1A by Skp2 at G1–S transition MS Song et al 3179 not shown) or expression of the mutant Skp2(DF), suggested that Skp2 targets RASSF1A for ubiquitin- which lacks the F-box domain, had no such effect dependent proteolysis at the G1–S transition. (Figure 2b). A co-immunoprecipitation assay revealed that some portions of endogenous SCF complex Cyclin D–cdk4 phosphorylates RASSF1A including Cul1 and Skp1, as well as endogenous Skp2 Given that the SCF complex recognizes most substrates were co-precipitated with RASSF1A (Figure 2c). These in a phosphorylation-dependent manner (Harper, 2002), results thus supported the notion that RASS1A we screened several Cdks for the ability to phosphor- degradation depends on a functional F box of Skp2. ylate RASSF1A. Among the various Cdks examined, We next determined whether the SCF–Skp2 complex only the cyclin D–Cdk4 phosphorylated RASSF1A directly ubiquitinates RASSF1A. Addition of in vitro (Figure 3a). We then determined the region of RASS- translated Skp2 to an in vitro reconstituted Skp1–Cul1– F1A that is phosphorylated by cyclin D–Cdk4 in vitro Rbx1 complex stimulated ubiquitination of 35S-labeled with the use of purified RASSF1A fragments fused to RASSF1A (Figure 2d). Expectedly, an in vivo ubiquiti- glutathione S-transferase (GST). Only the RASSF1A nation assay revealed that the ubiquitination of RASS- fragment comprising residues 180–239 was phosphory- F1A was evident during G1–S transition (Figure 2e). To lated by cyclin D–Cdk4 (Supplementary Figure 2a). further confirm that Skp2 regulates RASSF1A stability, Replacement of each of six serine, threonine or tyrosine we depleted U2OS cells of endogenous Skp2 by RNA residues in this fragment of RASSF1A with a nonpho- interference (RNAi) with a small interfering RNA sphorylatable alanine residue revealed that Ser203 is the (siRNA) specific for Skp2 mRNA. Skp2-depleted cells phosphorylation site for cyclin D–Cdk4, because phos- exhibited an increased level of endogenous RASSF1A phorylation of the fragment with the S203A mutation and p27 (Figure 2f). Consistently, the stability of was undetectable, whereas all other mutant fragments RASSF1A was greater in Skp2–/– MEFs than in wild- were still phosphorylated (Supplementary Figure type cells (Figure 2g). Taken together, these results thus 2b). Next, we prepared antibodies specific for the

+ - : GST - + : GST-RASSF1A + + : Cyclin D-Cdk4 IP : Cyclin CyclinD Cyclin E CyclinA B WT WT WT SA : GST-RASSF1A 32P-RASSF1A 32P-GST-RASSF1A - + + + : Cyclin D-Cdk4 32P-Cyclin D-Cdk4 32P-RASSF1A GST-RASSF1A p-RASSF1A p-RASSF1A Cyclin D p-RASSF1A Cyclin E (+ p-peptide) RASSF1A

Cyclin A RASSF1A

Cyclin B Cyclin D-Cdk4

G0-G1(%): 90 86 77 58 38 43 S (%): 6 10 19 37 45 27

G2-M(%): 4 4 4 5 17 30 p-RASSF1A

RASSF1A

Rb

β-Actin Serum stimulation (h): 81012141618 Figure 3 Cyclin D–Cdk4 phosphorylates RASSF1A. (a) Immunoprecipitates prepared from HeLa cells with antibodies to the indicted cyclins were incubated with GST-RASSF1A and [g-32P]ATP, after which the reaction mixtures were subjected to SDS–PAGE and autoradiography for the detection of 32P-labeled GST-RASSF1A as well as to immunoblot analysis with antibodies to GST or to the corresponding cyclins. (b) A purified GST-cyclin D–Cdk4 complex was incubated with GST or GST-RASSF1A (left) or with GST fusion proteins of wild-type (WT) or S203A mutant forms of RASSF1A (right) in the presence of [g-32P]ATP. The reaction mixtures were then subjected to SDS–PAGE and autoradiography for the detection of 32P-labeled GST-RASSF1A as well as to immunoblot analysis with antibodies specific for GST or RASSF1A or those specific for Ser203-phosphorylated RASSF1A (p-RASSF1A) that had been preincubated or not with the corresponding phosphopeptide antigen (p-peptide). (c) Human foreskin fibroblasts were synchronized in G0–G1 by serum deprivation, then restimulated with serum for the indicated times and subjected to immunoblot analysis with antibodies to the indicated proteins.

Oncogene Regulation of RASSF1A by Skp2 at G1–S transition MS Song et al 3180 Ser203-phosphorylated form of RASSF1A. These anti- Phosphorylation and degradation of RASSF1A are bodies reacted with a GST-RASSF1A fusion protein required for cell cycle progression only when it was phosphorylated by cyclin D–Cdk4, To examine the effect of RASSF1A regulation on cell the reactivity was blocked by prior incubation of the proliferation, we analysed the cell cycle progression in antibodies with the phosphopeptide antigen, and the RASSF1A þ / þ and RASSF1A–/– MEFs. The entry into S antibodies did not recognize the GST-RASSF1A phase after release from serum deprivation occurred (S203A) mutant (Figure 3b). To determine whether much earlier in RASSF1A–/– MEFs than in wild-type endogenous RASSF1A is phosphorylated in vivo,we cells (Figure 5). Of consistent with this, induction of –/– synchronized human foreskin fibroblasts at the G0–G1 cyclin A was seen earlier in RASSF1A MEFs than in phase by serum deprivation. Cells were then released RASSF1A þ / þ . We next reintroduced a vector for wild- from growth arrest by incubation with fresh medium type or S203A mutant forms of RASSF1A into containing complete serum and, at different time points, RASSF1A–/– MEFs. Expression of the RASSF1A(S203A) were subjected to immunoblot analysis with the mutant lead to a significant delay in G1–S progression, phospho-specific RASSF1A antibodies and flow cytometric whereas wild-type RASSF1A induced only a slight analysis. Consistently, the maximal phosphorylation of delay (Figure 5). These observations thus suggested RASSF1A occurred at the G1–S transition (Figure 3c). that phosphorylation and degradation of RASSF1A are Taken together, these data indicated that cyclin D–Cdk4 required for G1–S cycle progression after serum phosphorylates RASSF1A at the G1–S transition. stimulation.

Interaction and degradation of RASSF1A by Skp2 Skp2 regulates the antiproliferative and antisurvival 203 require phosphorylation of RASSF1A at Ser function of RASSF1A at the G1–S transition We then evaluated the possible relationship between Given that RASSF1A has a tumor suppressor function RASSF1A phosphorylation and its degradation. The by inhibiting cell proliferation and survival (Dammann reduction in Cdk4 activity induced by overexpression of et al., 2000;Shivakumar et al., 2002;Song et al., 2004; the Cdk4-specific inhibitor p16 Ink4a or expression of a Baksh et al., 2005), we thus examined whether Skp2 Cdk4 siRNA, but not a Cdk2 siRNA in U2OS cells, inhibits the tumor suppression function of RASSF1A. resulted in the decrease in both the phosphorylation and Overexpression of wild-type and S203A mutant forms of degradation of RASSF1A (Supplementary Figure 3 and RASSF1A resulted in a significant loss of viability in data not shown). Given that Cdk4-dependent phosphor- U2OS cells (Figures 6a and b). Importantly, the ylation plays an important role in the degradation of inhibitory effect of wild-type RASSF1A was signifi- RASSF1A and Skp2 preferentially recognize the phos- cantly abolished by expression of Skp2, whereas no phorylated substrates, we assessed whether mutation of effect on viability was observed when cells were co- RASSF1A-phosphorylation site Ser203 to alanine affects transfected with Skp2 and the RASSF1A(S203A) the interaction of RASSF1A with Skp2. Co-immuno- mutant (Figures 6a and b). Moreover, overexpression 203 precipitation analysis revealed that replacement of Ser of RASSF1A resulted in a G1 arrest and this effect was with alanine abolished the interaction of RASSF1A with also inhibited by the additional Skp2 expression in Skp2 (Figure 4a). Furthermore, an in vitro binding assay U2OS cells (data not shown). Together, these observa- revealed that an RASSF1A peptide (residues 197–209) tions thus suggested that Skp2 inhibits the tumor containing a phosphoserine at position 203 interacted suppressor activity of RASSF1A. with Skp2 to a much greater extent than did the Skp2 regulates entry into S phase by mediating corresponding nonphosphorylated peptide (Figure 4b). degradation of the Cdk inhibitor p27 (Reed, 2003). Pulse-chase analysis showed that the RASSF1A(S203A) However, the observation that the phenotype of mutant was more stable than was wild-type RASSF1A Skp2–/–p27–/– mice is similar but not identical to that of in 293T cells (Supplementary Figure 4). Consistent with p27–/– mice (Nakayama et al., 2004) suggests that, this observation, the abundance of ectopic wild-type although p27 might be the main target of Skp2, Skp2 RASSF1A was markedly decreased, whereas that of the likely also mediates the ubiquitination of other sub- RASSF1A(S203A) mutant remained stable in RASS- strates. We therefore investigated whether G1–S pro- F1A-deficient H1299 cells transfected with a vector for gression might be affected by Skp2-mediated Skp2 (Figure 4c). In addition, MG132 blocked the degradation of RASSF1A. Analysis of cell cycle degradation of wild-type RASSF1A in these cells but progression after serum deprivation and restimulation had little effect on the abundance of the S203A mutant revealed that both Skp2 þ / þ p27–/– MEFs transfected with (data not shown). Finally, we examined whether a vector for RASSF1A and Skp2–/–p27–/– MEFs entered phosphorylation of RASSF1A at Ser203 might be S phase B3 h later than did Skp2 þ / þ p27–/– MEFs required for Skp2-mediated ubiquitination. An in vitro (Supplementary Figure 5). We also examined the effects ubiquitination assay revealed that ubiquitination of of RNAi-mediated depletion of Skp2 on cell cycle RASSF1A was abolished by the S203A mutation progression in human cells. Similar to previous findings (Figure 4d). Taken together, these data indicated (Bashir et al., 2004;Wei et al., 2004), loss of Skp2 203 that phosphorylation of RASSF1A on Ser by cyclin resulted in a marked delay in G1–S progression in U2OS D–Cdk4 is required for interaction and degradation cells (Figure 6c). This delay was accompanied by by Skp2. accumulation of RASSF1A and p27 (Figure 6c). To

Oncogene Regulation of RASSF1A by Skp2 at G1–S transition MS Song et al 3181 Flag-Skp2 : + + + + + + + + HA-RASSF1A(WT) : + - - - + - - - HA-RASSF1A(1-287) : - + - - - + - - HA-RASSF1A(1-119) : - - + - - - + - HA-RASSF1A(S203A) : - - - + - - - + Beads: RASSF1A-203nRASSF1A-203p Heavy chain * HA-RASSF1A(S203A) Binding: HA-RASSF1A(WT) 35S-Skp2 HA-RASSF1A(1-287)

Input: Light chain 35S-Skp2 HA-RASSF1A(1-119)

Heavy chain Flag-Skp2

Lysate IP: Flag

HA-RASSF1A(WT) IP : + - + - HA-RASSF1A(S203A) IP : - + - + Flag-Skp2 IP : - - + + HA-RASSF1A:WT S203A

Flag-Skp2: - + - +

HA-RASSF1A HA-RASSF1A-(Ub)n

Flag-Skp2

β-Actin HA-RASSF1A

Flag-Skp2

Figure 4 Interaction and ubiquitination of RASSF1A by Skp2 depend on phosphorylation of RASSF1A at Ser203.(a) 293 T cells transiently expressing the indicated proteins were subjected to immunoprecipitation (IP) with anti-Flag, and the resulting precipitates as well as cell lysates were subjected to immunoblot analysis with anti-HA and -Flag. The asterisk marks a nonspecific band. (b) In vitro translated 35S-Skp2 was incubated with beads conjugated with a RASSF1A peptide (residues 197–209) containing phosphorylated (203p) or nonphosphorylated (203n) Ser203. The beads were then washed, and bound 35S-Skp2 was detected by SDS–PAGE and autoradiography. (c) RASSF1A-deficient H1299 cells transfected with vectors for HA-tagged wild-type or S203A mutant forms of RASSF1A in the presence or absence of Flag-Skp2, as indicated, were subjected to immunoblot analysis with antibodies to the indicated proteins. (d) Bead-immobilized immunoprecipitates prepared with anti-HA from 293T cells expressing HA-tagged wild-type or S203A mutant forms of RASSF1A were incubated with Flag-Skp2 immunoprecipitates from transfected 293T cells in a reaction mixture containing 20 mM MG132 and rabbit reticulocyte lysate, which was depleted of Skp2 with anti-Skp2. The reaction mixture was subjected to immunoblot analysis with antibodies to HA or to Flag. examine whether the accumulation of p27 or that of RASSF1A has been implicated in various cellular RASSF1A is mainly responsible for the delay in processes involved in tumor suppression. RASSF1A progression from G1 to S phase, we depleted cells of regulates the mitotic progression, microtubule dynamics both Skp2 and RASSF1A. The delay in G1–S progres- (Song et al., 2004;Vos et al., 2004) and G1–S transition sion induced by Skp2 removal was normalized by (Fenton et al., 2004). The frequently observed down- depletion of RASSF1A even in the presence of a high regulation of RASSF1A expression in human tumors level of p27, supporting the notion that G1–S progression has in many cases been attributed to promoter depends, at least in part, on degradation of RASSF1A methylation (Agathanggelou et al., 2005), however, (Figure 6c). Collectively, these results demonstrated that other potential causes have not been revealed yet. There the Skp2-mediated degradation of RASSF1A plays an are many evidence, that aberrant protein degradation of important role in cell proliferation and survival. cell cycle regulators is an important factor contributing to tumorigenesis (Bashir and Pagano, 2003;Pagano and Benmaamar, 2003). Discussion Skp2 is the substrate-recognition component of the SCFSkp2 ubiquitin ligase complex and is a key regulator Although molecular mechanisms by which RASSF1A of S-phase entry (Nakayama et al., 2000;Harper, participates in tumorigenesis are not fully elucidated, 2002;Cardozo and Pagano, 2004). Overexpression or

Oncogene Regulation of RASSF1A by Skp2 at G1–S transition MS Song et al 3182 RASSF1A+/+ RASSF1A-/- RASSF1A-/- RASSF1A-/- + Vector + Vector + RASSF1A(WT) + RASSF1A(S203A)

16 16 16 16 2N 2N 2N 2N 13 13 13 13 4N 10 4N 10 4N 10 4N 10 8 8 8 8 Serum 6 Serum 6 Serum 6 Serum 6 stimulation (h): 4 stimulation (h): 4 stimulation (h): 4 stimulation (h): 4

RASSF1A+/+ RASSF1A-/- RASSF1A-/- RASSF1A-/- + Vector + Vector + RASSF1A(WT) + RASSF1A(S203A)

Cyclin A

RASSF1A

β-Actin

468101316 468101316 4 6 8 10 13 16 468101316 Serum stimulation (h) Figure 5 Phosphorylation and degradation of RASSF1A are required for cell cycle progression. RASSF1A–/– MEFs were infected with retrovirus vectors for empty, wild-type or S203A mutant forms of RASSF1A. Infected cells were rendered quiescent by serum deprivation and then restimulated with serum for the indicated times before analysis of cell-cycle progression by flow cytometry (a) and immunoblot analysis of cell lysates with antibodies to the indicated proteins (b).

amplification of Skp2 has been reported in a large of a high level of p27 (Figure 6c). Thus it is likely that Skp2 number of human cancers (Yokoi et al., 2002) and targets RASSF1A to allow entry of cells into S phase and transgenic expression of Skp2 in mice leads to tumor targets p27 to regulate progression into mitosis. formation, suggesting that Skp2 is oncogenic (Gstaiger The ubiquitination of RASSF1A is regulated by the et al., 2001;Latres et al., 2001). Skp2 targets p27, cyclin interaction of Skp2 with RASSF1A in a phosphoryla- E and FOXO1 for degradation in a phosphorylation- tion-dependent manner (Figures 4a and b). Phospho- dependent manner (Marti et al., 1999;Nakayama et al., rylation of RASSF1A on Ser203 by cyclin D–Cdk4 2000;Huang et al., 2005). Here we provide several lines appears to be the key step required for Skp2 binding. of evidence that RASSF1A is also targeted by Skp2 at Except for the retinoblastoma protein family, Smad3 the G1–S transition: (1) the level of RASSF1A during has been the only Cdk4 substrate demonstrated so far cell cycle progression is inversely related to that of Skp2; (Matsuura et al., 2004). Here we have shown that (2) Skp2 directly interacts with and ubiquitinates RASSF1A is also phosphorylated by cyclin D–Cdk4 RASSF1A in a manner dependent on phosphorylation (Figure 3 ;Supplementary Figure 3). Mutation of this 203 of RASSF1A on Ser in G1–S transition;(3) over- cyclin D–Cdk4 phosphorylation site in RASSF1A expression of Skp2 results in a decrease in the cellular increases its abundance and antiproliferative function abundance of RASSF1A, thereby elevating cell viability (Figures 4c, d and 5). Cyclin D–Cdk4 thus likely inhi- and proliferation;(4) deletion of Skp2 or depletion of bits the accumulation of RASSF1A, thus facilitating cell Skp2 increases both the level and half-life of RASSF1A cycle progression from G1 to S phase. as well as induces a delay in G1–S progression;and (5) Recently, it has been reported that Aurora A appears entry into S phase is accelerated in RASSF1A–/– MEFs. to phosphorylate RASSF1A at Thr-202 and/or Ser-203, The phenotype of Skp2–/– mice is rescued in large part although further studies should be required for deter- by knockout of p27. However, the phenotype of mination of which one of two is predominant phospho- Skp2–/–p27–/– mice is not identical to that of p27–/– mice. residue in RASSF1A (Rong, 2007). Indeed, we also Although accumulation of p27 mainly contributes to the found that Aurora kinases A and B primarily phos- proliferation defect of Skp2–/– MEFs, Skp2 appears to phorylate RASSF1A on Ser-203 for early and late prevent the accumulation of p27 during S–G2 rather mitosis, respectively (SJ Song and D-S Lim, unpublished than at the G1–S transition (Nakayama et al., 2004). We data). Thus, phosphorylation of RASSF1A Ser203 by have now shown that the delay in G1–S progression Aurora kinases and Cdk4 appears to have clearly induced by Skp2 depletion was normalized by the distinct roles in cell cycle control at different stages of additional depletion of RASSF1A even in the presence the cell cycle.

Oncogene Regulation of RASSF1A by Skp2 at G1–S transition MS Song et al 3183 120 Vector 100 Skp2

Vector Flag-Skp2 80

60

40 Cell Survival (%) Vector 20

0

Vector (WT) (WT) RASSF1A RASSF1A(S203A) HA-RASSF1A Flag-Skp2: - - - + + + HA-RASSF1A: - WT SA - WT SA HA-RASSF1A (S203A)

HA-RASSF1A Flag-Skp2

β-Actin

GFP siRNA Skp2 siRNA Skp2 siRNA + RASSF1A siRNA

14 14 14 2N 13 2N 13 2N 13 12 4N 12 4N 12 4N 11 11 11 Nocodazole 9 Nocodazole 9 Nocodazole 9 release (h): 7 release (h): 7 release (h): 7

GFP siRNA Skp2 siRNA Skp2 siRNA + RASSF1A siRNA

Cyclin A p27

Skp2

RASSF1A

β-Actin Nocodazole release (h) : 7 9 11 12 13 14 7 9 11 12 13 14 7 9 11 12 13 14 Figure 6 Skp2-mediated degradation of RASSF1A is required for cell survival and proliferation. (a and b) U2OS cells were co- transfected with pEGFP vector and vectors for HA-tagged wild-type or S203A mutant form of RASSF1A in the presence or absence of Flag-Skp2, as indicated. At 48 h after transfection, transfected viable cells were photographed under both UV and transmitted light (a) and quantified by using bromophenol blue (b, top) and subjected to immunoblot analysis with antibodies to the indicated proteins (b, bottom). (c) U2OS cells transfected with a vector for Skp2 small interfering RNA (siRNA) or a control green fluorescent protein (GFP) siRNA in the presence or absence of a vector for RASSF1A siRNA were released from nocodazole block for the indicated times before analysis of cell cycle progression by flow cytometry (top) or immunoblot analysis of cell lysates with antibodies to the indicated proteins (bottom).

Since human cancers often show overexpression of Materials and methods Aurora A, Skp2, cyclin D1 and high levels of Cdk4 activity, the diminishing RASSF1A activity by Aurora Cell culture A, Skp2 or cyclin D–Cdk4 may contribute to tumor- U2OS, HeLa, 293T, human foreskin fibroblasts, H1299, igenesis. RASSF1A–/– MEFs, Skp2–/– MEFs, p27–/– MEFs and Skp2–/–

Oncogene Regulation of RASSF1A by Skp2 at G1–S transition MS Song et al 3184 p27–/– MEFs were grown in Dulbecco’s modified Eagle’s distributions were determined by a FACSCalibur flow medium supplemented with 10% fetal bovine serum. Cyclo- cytometer (Becton Dickinson and Co., Franklin Lakes, NJ, heximide or MG132 (EMD Biosciences, San Diego, CA, USA) USA) and analysed with MODFIT II software (Verity was added at a final concentration of 50 mgmlÀ1 or 10 mM, Software House Inc., Topsham, ME, USA). Immunofluores- respectively. Cells were synchronized in G0–G1, mitosis or G1–S cence was also performed as described (Song et al., 2004). by incubation for 36 h with 0.1% serum-containing medium, 16 h with nocodazole (100 ng mlÀ1;Sigma-Aldrich, St Louis, MO, USA) or thymidine (2 mM), respectively. Transfections were done Immunoblot analysis, immunoprecipitation, in vitro binding with the Effectene transfection reagent (Qiagen, Valencia, CA, assays and kinase assays USA), and transfection efficiencies of 70–90% in cells were Immunoblot analysis and immunoprecipitation were per- achieved. For retrovirus production, retroviral RASSF1A formed as described (Song et al., 2004). For in vitro binding 35 plasmid was transfected into 293T packaging cells to produce assays, in vitro translated S-Skp2 was incubated with bead- retroviruses as described previously (Lim et al., 2000). RASS- immobilized RASSF1A peptides in binding buffer (50 mM Tris, F1A–/– MEFs were infected at greater than 90% efficiency. pH 7.4, 150 mM NaCl, 5 mM MgCl2,1mM EDTA, 0.1% Triton X-100, phosphatase/protease inhibitors). The beads were washed four times with binding buffer and bound 35S-Skp2 was detected Plasmids and protein generation by SDS–PAGE and autoradiography. In vitro kinase assays were The vectors for HA-RASSF1A, GST-RASSF1A and retro- performed as described (Oh et al., 2006). viral RASSF1A plasmid were previously described (Song et al., 2004). The plasmids pcDNA3-Flag or -Myc-Skp2 and –Flag or -Myc-Skp2(DF) were kindly provided by M Pagano, In vitro ubiquitination assays and plasmids encoding other F-box proteins were kindly To examine the ability of Skp2 to ubiquitinate RASSF1A provided by JW Harper. The plasmids pcDNA3-HA-Cul1, in vitro, SCFSkp2 immunopurified from a mixture of in vitro pcDNA3-Rbx1 and pcDNA3-Skp1 were kindly provided by translated Cul1, Rbx1, Skp1 and Skp2 was incubated for 1 h at JB Yoon, and pcDNA3-HA-p16 was kindly provided by HW 30 1C with ubiquitin (1.25 mg mlÀ1;Sigma), rabbit E1 (EMD Lee. The RASSF1A(S203A) mutant was generated with the Biosciences), Ubc5a (EMD Biosciences), an ATP-regenerating use of a site-directed mutagenesis kit (Stratagene, La Jolla, system, HeLa cell extract and in vitro translated 35S-RASSF1A. CA, USA). GST fusion proteins were purified by standard In vitro ubiquitination assays using rabbit reticulocyte lysate were methods. Cyclin D–Cdk4 kinase was from Cell Signaling performed as described (Carrano et al., 1999). Technology Inc. (Danvers, MA, USA).

Antibodies In vivo ubiquitination assays Guinea pig polyclonal antibodies to RASSF1A were prepared Cells transfected with a plasmid encoding His-tagged human with a GST fusion protein containing an N-terminal fragment ubiquitin and Flag-RASSF1A were subjected to thymidine of human RASSF1A (residues 1–119) as the antigen. Rabbit block (2 mM) for 26 h or asynchronized. Cells were incubated polyclonal antibodies specific for Ser203-phosphorylated RASS- with MG132 (20 mM) for 6 h before harvest and then analysed F1A were generated with a synthetic peptide containing for in vivo ubiquitination. Cells were lysed by incubation for 1 phospho-Ser203 (CSVRRRTpSFYLPKD). Specific antibodies 10 min at 95 C with two volumes of tris-buffered saline TBS were affinity purified with the appropriate antigen. Other (10 mM Tris–HCl (pH 7.5);150 m M NaCl) containing 2% antibodies included HA (Roche Applied Science, Indianapolis, SDS. After the addition of eight volumes of 1% Triton X-100 IN, USA);Flag or b-actin (Sigma);p27 (BD Biosciences, San in TBS, the lysates were sonicated for 2 min and then Jose, CA, USA);p21, cyclin A, cyclin B, cyclin D, cyclin E, incubated with protein G agarose. The beads were removed Skp2, Myc, GST or Cdk4 (Santa Cruz Biotechnology, Santa by centrifugation, and the lysates were then subjected to Cruz, CA, USA);Skp2 (Zymed, Carlsbad, CA, USA) and immunoprecipitation with anti-Flag coupled to protein G p-pRb (S780, Cell Signaling Technology Inc.). agarose. The beads were washed first with 0.5 M LiCl in TBS and then twice with TBS, boiled and subjected to immunoblot analysis with anti-ubiquitin. Cell lysates were subjected to RNA interference immunoblotting with anti-Flag. The sequences of siRNAs for Skp2 and RASSF1A were previously described (Bashir et al., 2004;Song et al., 2004). The DNA sequences for these siRNAs were inserted into the plasmid Acknowledgements pSUPER (Oligoengine, Seattle, WA, USA) and introduced into cells by transfection with the use of the Effectene transfection We are grateful to M Pagano, JW Harper, JB Yoon, HW Lee, reagent (Qiagen). The plasmid pSHAG1–Cdk4 or –Cdk2 RNAi F Liu for DNAs and to G Pfeifer for RASSF1A wild-type and for a Cdk4 or Cdk2 siRNA was kindly provided by F Liu. null MEFs. This study was supported by a grant from the Korea Research Foundation and the 21st Century Frontier Cell cycle analysis Functional Project of KISTEP (Ministry of Analysis of DNA content during cell cycle progression Science and Technology of Korea), and the National Research was performed as described (Oh et al., 2006). Cell cycle Laboratory Program of Korea.

References

Agathanggelou A, Cooper WN, Latif F. (2005). Role of the Ras- with high tumor cell proliferation. Am J Pathol 154: association domain family 1 tumor suppressor gene in human 113–118. cancers. Cancer Res 65: 3497–3508. Baksh S, Tommasi S, Fenton S, Yu VC, Martins LM, Pfeifer GP et al. An HX, Beckmann MW, Reifenberger G, Bender HG, (2005). The tumor suppressor RASSF1A and MAP-1 link death Niederacher D. (1999). Gene amplification and overexpres- receptor signaling to Bax conformational change and cell death. Mol sion of CDK4 in sporadic breast carcinomas is associated Cell 18: 637–650.

Oncogene Regulation of RASSF1A by Skp2 at G1–S transition MS Song et al 3185 Bartkova J, Lukas J, Strauss M, Bartek J. (1995). Cyclin D1 results in accumulation of cyclin E and p27(Kip1), polyploidy and oncoprotein aberrantly accumulates in malignancies of diverse centrosome overduplication. EMBO J 19: 2069–2081. histogenesis. Oncogene 10: 775–778. Nakayama K, Nagahama H, Minamishima YA, Miyake S, Ishida N, Bashir T, Pagano M. (2003). Aberrant ubiquitin-mediated proteolysis Hatakeyama S et al. (2004). Skp2-mediated degradation of p27 of cell cycle regulatory proteins and oncogenesis. Adv Cancer Res 88: regulates progression into mitosis. Dev Cell 6: 661–672. 101–144. Oh HJ, Lee KK, Song SJ, Jin MS, Song MS, Lee JH et al. (2006). Role Bashir T, Dorrello NV, Amador V, Guardavaccaro D, Pagano M. of the tumor suppressor RASSF1A in Mst1-mediated apoptosis. (2004). Control of the SCF(Skp2-Cks1) ubiquitin ligase by the APC/ Cancer Res 66: 2562–2569. C(Cdh1) ubiquitin ligase. Nature 428: 190–193. Pagano M, Benmaamar R. (2003). When protein destruction runs Cardozo T, Pagano M. (2004). The SCF ubiquitin ligase: insights into amok, malignancy is on the loose. Cancer Cell 4: 251–256. a molecular machine. Nat Rev Mol Cell Biol 5: 739–751. Reed SI. (2003). Ratchets and clocks: the cell cycle, ubiquitylation and Carrano AC, Eytan E, Hershko A, Pagano M. (1999). SKP2 is protein turnover. Nat Rev Mol Cell Biol 4: 855–864. required for ubiquitin-mediated degradation of the CDK inhibitor Rong R. (2007). Mitotic kinase Aurora-A phosphorylates RASSF1A p27. Nat Cell Biol 1: 193–199. and modulates RASSF1A-mediated microtubule interaction and Dammann R, Li C, Yoon JH, Chin PL, Bates S, Pfeifer GP. M-phase cell cycle regulation. Oncogene;e-pub ahead of print: 17 (2000). Epigenetic inactivation of a RAS association domain family June 2007. protein from the lung tumour suppressor locus 3p21.3. Nat Genet 25: Rong R, Jin W, Zhang J, Sheikh MS, Huang Y. (2004). Tumor 315–319. suppressor RASSF1A is a microtubule-binding protein that Fenton SL, Dallol A, Agathanggelou A, Hesson L, Ahmed- stabilizes microtubules and induces G2/M arrest. Oncogene 23: Choudhury J, Baksh S et al. (2004). Identification of the E1A- 8216–8230 . regulated transcription factor p120 E4F as an interacting partner of Sherr CJ, Roberts JM. (1999). CDK inhibitors: positive and negative the RASSF1A candidate tumor suppressor gene. Cancer Res 64: regulators of G1-phase progression. Dev 13: 1501–1512. 102–107. Shivakumar L, Minna J, Sakamaki T, Pestell R, White MA. (2002). Gstaiger M, Jordan R, Lim M, Catzavelos C, Mestan J, Slingerland J The RASSF1A tumor suppressor blocks cell cycle progression and et al. (2001). Skp2 is oncogenic and overexpressed in human cancers. inhibits cyclin D1 accumulation. Mol Cell Biol 22: 4309–4318. Proc Natl Acad Sci USA 98: 5043–5048. Song MS, Song SJ, Ayad NG, Chang JS, Lee JH, Hong HK et al. Harper JW. (2002). A phosphorylation-driven ubiquitination switch (2004). The tumour suppressor RASSF1A regulates mitosis by for cell-cycle control. Trends Cell Biol 12: 104–107. inhibiting the APC-Cdc20 complex. Nat Cell Biol 6: 129–137. Huang H, Regan KM, Wang F, Wang D, Smith DI, van Deursen JM Tommasi S, Dammann R, Zhang Z, Wang Y, Liu L, Tsark WM et al. et al. (2005). Skp2 inhibits FOXO1 in tumor suppression through (2005). Tumor susceptibility of Rassf1a knockout mice. Cancer Res ubiquitin-mediated degradation. Proc Natl Acad Sci USA 102: 65: 92–98. 1649–1654. van der Weyden L, Tachibana KK, Gonzalez MA, Adams DJ, Ng BL, Latres E, Chiarle R, Schulman BA, Pavletich NP, Pellicer A, Petty R et al. (2005). The RASSF1A isoform of RASSF1 promotes Inghirami G et al. (2001). Role of the F-box protein Skp2 in microtubule stability and suppresses tumorigenesis. Mol Cell Biol lymphomagenesis. Proc Natl Acad Sci USA 98: 2515–2520. 25: 8356–8367. Lim DS, Kim ST, Xu B, Maser RS, Lin J, Petrini JH et al. (2000). Vos MD, Martinez A, Elam C, Dallol A, Taylor BJ, Latif F et al. ATM phosphorylates p95/nbs1 in an S-phase checkpoint pathway. (2004). A role for the RASSF1A tumor suppressor in the regulation Nature 404: 613–617. of tubulin polymerization and genomic stability. Cancer Res 64: Marti A, Wirbelauer C, Scheffner M, Krek W. (1999). Interaction 4244–4250. between ubiquitin-protein ligase SCFSKP2 and E2F-1 underlies the Wei W, Ayad NG, Wan Y, Zhang GJ, Kirschner MW, Kaelin Jr WG. regulation of E2F-1 degradation. Nat Cell Biol 1: 14–19. (2004). Degradation of the SCF component Skp2 in cell-cycle Matallanas D, Romano D, Yee K, Meissl K, Kucerova L, Piazzolla D phase G1 by the anaphase-promoting complex. Nature 428: et al. (2007). RASSF1A elicits apoptosis through an MST2 pathway 194–918. directing proapoptotic transcription by the p73 tumor suppressor Yokoi S, Yasui K, Saito-Ohara F, Koshikawa K, Iizasa T, Fujisawa T protein. Mol Cell 27: 962–975. et al. (2002). A novel target gene, SKP2, within the 5p13 amplicon Matsuura I, Denissova NG, Wang G, He D, Long J, Liu F. (2004). that is frequently detected in small cell lung cancers. Am J Pathol Cyclin-dependent kinases regulate the antiproliferative function of 161: 207–216. Smads. Nature 430: 226–231. Zachariae W, Nasmyth K. (1999). Whose end is destruction: cell Nakayama K, Nagahama H, Minamishima YA, Matsumoto M, division and the anaphase-promoting complex. Genes Dev 13: Nakamichi I, Kitagawa K et al. (2000). Targeted disruption of Skp2 2039–2058.

Supplementary Information accompanies the paper on the Oncogene website (http://www.nature.com/onc).

Oncogene