Oncogene (2007) 26, 6954–6958 & 2007 Nature Publishing Group All rights reserved 0950-9232/07 $30.00 www.nature.com/onc SHORT COMMUNICATION p21Waf1/Cip1/Sdi1 mediates retinoblastoma protein degradation

EV Broude1, ME Swift2, C Vivo1, B-D Chang1, BM Davis1, S Kalurupalle1, MV Blagosklonny1 and IB Roninson1

1Cancer Center, Ordway Research Institute, Albany, NY, USA and 2Department of Molecular Genetics, University of Illinois at Chicago, Chicago, IL, USA

Damage-induced G1 checkpoint in mammalian cells the proteasome. Proteasome-mediated Rb degradation involves upregulation of p53, which activates transcription is promoted by Mdm2 (Sdek et al., 2005) and gankyrin of p21Waf1 (CDKN1A). Inhibition of cyclin-dependent (Higashitsuji et al., 2000), E7 of papilloma virus (Boyer kinase (CDK)2 and CDK4/6 by p21 leads to depho- et al., 1996) and Tax of HTLV1 (Kehn et al., 2005). sphorylation and activation of Rb. We now show that Oncoprotein-induced proteasomal degradation of Rb is ectopic p21 expression in human HT1080 fibrosarcoma one of the mechanisms for Rb inactivation in carcino- cells causes not only dephosphorylation but also depletion genesis (Ying and Xiao, 2006), but Rb degradation has of Rb; this effect was p53-independent and susceptible to a not been described in DNA damage response. proteasome inhibitor. CDK inhibitor p27 (CDKN1B) also Changes in Rb phosphorylation are most commonly caused Rb dephosphorylation and depletion, but another detected by immunoblotting through changes in the CDK inhibitor p16 (CDKN2A) induced only depho- protein’s electrophoretic mobility. Examination of sphorylation but not depletion of Rb. Rb depletion was numerous Rb immunoblots published by different observed in both HT1080 and HCT116 colon carcinoma groups showed that in many (but not all) cases Rb cells, where p21 was induced by DNA-damaging agents. dephosphorylation, which results from drug treatment, Rb depletion after DNA damage did not occur in the cell senescence or ectopic p21 expression, is associated absence of p21, and it was reduced when p21 induction was with a reduction in the Rb protein signal. In the present inhibited by p21-targeting short hairpin RNA or by a study, we have asked (i) whether a decrease in the Rb transdominant inhibitor of p53. These results indicate that signal in response to p21 reflects protein degradation or p21 both activates Rb through dephosphorylation and merely altered immunoreactivity of dephosphorylated inactivates it through degradation, suggesting negative Rb, (ii) if p53 plays a p21-independent role in the de- feedback regulation of damage-induced cell-cycle check- crease in Rb, (iii) whether such decrease can be induced point arrest. by other CDK inhibitors that induce Rb dephosphory- Oncogene (2007) 26, 6954–6958; doi:10.1038/sj.onc.1210516; lation and (iv) if the decrease in Rb in drug-damaged published online 7 May 2007 cells is dependent on p21 induction. We have described previously a subline of HT1080 Keywords: p21; Rb; p27; damage response cells, p21-9, which expresses p21 from a promoter inducible by b-galactoside isopropyl-b-thio-galactoside (IPTG) (Chang et al., 1999). p21 induction in HT1080 p21-9 leads to Rb dephosphorylation and depletion p53-inducible cyclin-dependent kinase (CDK) inhibitor without any changes in Rb mRNA levels (Chang et al., p21 (CDKN1A) is the key mediator of damage-induced 2000). As shown in Figure 1a (left lanes), Rb was cell-cycle arrest. p21 interacts with different cyclin/CDK dephosphorylated and its signal intensity decreased 24– complexes and other regulators of transcription and 48 h after IPTG addition. To determine if the reduced signal transduction, exerting broad effects on cell survival, Rb signal indicates a decrease in protein levels rather gene expression and morphology (Roninson, 2002). p21 than altered antibody reactivity, we have used both a effects are partially mediated by Rb, which is inactivated mouse monoclonal and a rabbit polyclonal antibody in proliferating cells through phosphorylation by CDK2 against Rb. Both antibodies produced the same result and CDK4/6, both of which are inhibited by p21. As a (Figure 1a, left lanes), indicating that the decrease in Rb result, p21 induction leads to Rb dephosphorylation and signal intensity was indeed due to a decrease in the activation, with ensuing G1 arrest. protein level. Whereas p21 activates Rb by dephosphorylation, The decrease in Rb protein was not accompanied by several oncoproteins inactivate Rb by degradation via the appearance of Rb fragments that arise after caspase cleavage of Rb (Tan et al., 1997). Since the best-known mechanism of Rb degradation is via the proteasome Correspondence: Dr IB Roninson, Cancer Center, Ordway Research pathway (Ying and Xiao, 2006), we have tested the Institute, 150 New Scotland Avenue, Albany, NY 12208, USA. E-mail: [email protected] effect of a proteasome inhibitor ALLN (N-acetyl- Received 21 February 2007; accepted 18 March 2007; published online 7 leucinyl-leucinyl-norleucinal) on p21-induced Rb deple- May 2007 tion. In the experiment shown in Figure 1b, cells were p21 regulates Rb degradation EV Broude et al 6955 treated with IPTG for 20 h with ALLN added in one likely reflects the effect of GSE56 on the p53 response arm of the experiment for the last 4 h (longer treatment element present in the cytomegalovirus (CMV)-derived with ALLN could not be used because of the toxicity promoter, which drives p21 expression in HT1080 p21-9 of this compound). The addition of ALLN did not cells (Chang et al., 1999). p21 induction in GSE56- interfere with Rb dephosphorylation, but reduced the expressing cells resulted in complete dephosphorylation depletion of the protein (Figure 1b), suggesting that Rb and substantial depletion of Rb (Figure 1a, right lanes). depletion was due to proteasome-mediated degradation. The extent of Rb depletion was somewhat lower in the To determine if p53 contributes to Rb depletion in presence than in the absence of GSE56 (Figure 1a); this response to p21, we have analysed the effects of p21 on modest effect can be accounted for by the lower p21 Rb in HT1080 p21-9 cells, where p53 activity was levels. Hence, p53 is not required for p21-induced Rb blocked by GSE56, a transdominant inhibitor of p53 depletion in HT1080 cells. function (Ossovskaya et al., 1996). p21 induction by To determine if Rb depletion is a general consequence IPTG was about twofold lower in the presence of of its dephosphorylation, we compared the effects of p21 GSE56 (Figure 1a, right lanes). This inhibition most and two other Rb-dephosphorylating CDK inhibitors, p16Ink4A (CDKN1A), which inhibits CDK4/6, and p27Kip1 (CDKN2B), which inhibits CDK2. p16 and p27 were expressed in HT1080 cells from an IPTG- a IPTG (hr) inducible retroviral vector LNXRO2 (Chang and 0244824480 Roninson, 1996). IPTG-inducible expression of either p21 or p27 led to Rb dephosphorylation and depletion p21 (Figure 1c). In contrast, p16 caused only dephosphory- lation but not depletion of Rb. Hence, Rb depletion is +P y not a general consequence of its dephosphorylation. Rb -P We have investigated whether Rb depletion in +P response to DNA damage is determined by p21. Rb Antibod Doxorubicin treatment of HT1080 p21-9 cells induced -P Mouse Rabbit endogenous p21 and led to Rb dephosphorylation and α-tubulin depletion (Figure 2a, left). We have also analysed this effect in HT1080 p21-9 cells where p53 activity was p21-9 p21-9 GSE56 inhibited by GSE56. As shown in Figure 2a (right), GSE56 blocked p21 induction by doxorubicin almost b completely, and under these conditions, Rb showed only minimal dephosphorylation and no depletion. To verify that this effect of GSE56 was directly attributable to control IPTG ALLN IPTG+ALLN p21, we developed a derivative of HT1080 cells with +P Rb -P Figure 1 (a) Immunoblotting of p21 and Rb in HT1080 p21-9 load cells with IPTG-inducible p21 (Chang et al., 1999) (left lanes) or HT1080 p21-9 transduced with p53 inhibiting peptide GSE56 (right lanes) after the addition of 50 mM IPTG (Sigma, St Louis, MO, c p21 p16 p27 USA). GSE56 (Ossovskaya et al., 1996) was introduced into HT1080 p21-9 cells via a retroviral vector LXSE (Kandel et al., 0 h 24 h 0 h 24 h 0 h 24 h 1997); the transduced cells were selected by flow sorting for green fluorescent protein fluorescence. Cells were grown in Dulbecco’s +P modified Eagle’s medium (DMEM) with 10% FC2 serum. Rb Immunoblotting was carried out by standard procedures, using -P p21 mouse monoclonal antibody (Oncogene Research, San Diego, α-tubulin CA, USA) (top row), Rb rabbit polyclonal antibody (Neomarkers, Flemont, CA, USA) (middle row) and Rb mouse monoclonal antibody (BD PharMingen, San Diego, CA, USA) (bottom row). 1.2 a-Tubulin detected with a monoclonal antibody from Sigma was used as a loading control. Immunoblots were developed with 1.0 horseradish peroxidase-conjugated secondary antibodies (Amer- control sham Biosciences, Piscataway, NJ, USA) and ECL chemilumines- cence agent (Pierce, Rockford, IL, USA). Phosphorylated ( þ P) 0.8 IPTG 24 h and dephosphorylated (ÀP) forms of Rb are marked with arrows. (b) Immunoblotting of Rb in HT1080 p21-9 cells treated for 20 h 0.6 with IPTG alone or with IPTG and 20 mM proteasome inhibitor ALLN (Calbiochem/EMD Biosciences, San Diego, CA, USA) 0.4 added for the last 4 h of the 20-h period is shown. Ponceau staining was used as a loading control. (c) Immunoblotting of Rb in 0.2

relative protein amount HT1080 cells with IPTG-inducible expression of p21, p16 or p27, before (0 h) or after 24-h IPTG treatment (Chang and Roninson, 0.0 1996) is shown. The ratio of Rb to a-tubulin signals, as measured p21 p16 p27 by densitometry using Bio-Rad Versa Doc, is shown at the bottom.

Oncogene p21 regulates Rb degradation EV Broude et al 6956 a p21-9 p21-9 GSE56 a HCT116 HCTp21-/- HCTp53-/-

+P Rb p53 -P

p21 p21

α-tubulin +P Rb --++Doxorubicin -P

b p21 shRNA β-actin off on

+P 0312 0312 0312 Rb Doxorubicin (days) -P b Doxorubicin p21 control 25 50 200 Rad Aph α-tubulin +P Rb --++Doxorubicin -P Figure 2 (a) Immunoblotting of Rb and p21 in HT1080 p21-9 cells without (left lanes) or with GSE56 (right lanes), untreated or p53 treated for 16 h with 250 nM doxorubicin (Sigma). a-Tubulin was used as a loading control. (b) Immunoblotting of Rb and p21 in HT1080-p21shRNA cells with doxycycline-inducible p21-targeting shRNA is shown. Cells grown in the absence (left lanes) or in the p21 presence of 100 ng/ml doxycyclin (right lanes) were treated with doxorubicin and analysed as in (a). To derive HT1080-p21shRNA, HT1080 (from American type culture collection (ATCC) Manas- sas, VA, USA) was transduced with lentiviral vector pLV-tTR- β-actin KRAB-red (Wiznerowicz and Trono, 2003) (a gift from Dr D Trono; University of Geneva) expressing tTR-KRAB repressor; Figure 3 (a) Immunoblotting of p53, p21 and Rb in wild-type infectants were selected for dsRed fluorescence. Cells were then HCT116 cells (p53 parental) (left lanes) and HCT116 derivatives transduced with a lentivirus expressing shRNA targeting the with homozygous knockout of p21 (clone 80S4) (Waldman et al., GACCTGTCACTGTCTTGTA sequence of p21 and cloned into 1996) (middle lanes) or p53 (clone 379.2) (Bunz et al., 1998) (right tTR-KRAB-regulated vector LLCEP TU6X (Maliyekkel et al., lanes), on the indicated days of treatment with 200 nM doxorubicin. 2006), followed by puromycin selection. b-Actin (monoclonal antibody, Sigma) was used as a control. (b) Immunoblotting of Rb, p53 and p21 in HCT116 cells treated for 2 tetracycline/doxycycline-regulated expression of short days with the indicated concentrations of doxorubicin or 2.2 nM hairpin RNA (shRNA) against p21. shRNA induction aphidicolin, or 2 days after exposure to ionizing radiation (6 Gy). by doxycycline in this cell line (designated HT1080- p21shRNA) partially decreased p21 induction by 2 days of drug treatment, with only a partial decrease doxorubicin (Figure 2b). The remaining levels of p21 on day 3 (Figure 3a). The p53À/À cell line showed no were still sufficient to produce Rb dephosphorylation, detectable p21 without drug treatment and only weak but the cellular level of dephosphorylated Rb was much p21 induction by doxorubicin (Figure 3a). In p53À/À higher than that without p21 knockdown (Figure 2b). cells, doxorubicin caused substantial dephosphorylation These observations indicate that p21 mediates both of Rb, but Rb levels were essentially unchanged dephosphorylation and depletion of Rb in response to (Figure 3a). This suggests that low levels of p21 induced DNA damage. by the drug in p53À/À cells were sufficient to produce To determine whether p21-mediated Rb depletion partial dephosphorylation of Rb but not Rb depletion. occurs in a different cell type, we investigated the effects We next compared doxorubicin with other DNA- of doxorubicin on Rb in HCT116 colon carcinoma cells interacting agents that differ in their ability to induce and in HCT116 derivatives with a homozygous knock- p21. Doxorubicin (50 and 200 nM) and ionizing radia- out of either p21 or p53 (Waldman et al., 1996; Bunz tion (6 Gy) induced both p21 expression and Rb et al., 1998). Doxorubicin treatment of parental depletion, whereas 25 nM doxorubicin and the DNA HCT116 cells led to both dephosphorylation and replication inhibitor aphidicolin (2.2 nM) produced little depletion of Rb (Figure 3a). The p21À/À line, which or no p21 induction and did not deplete Rb (Figure 3b). expresses p53 but not p21, expressed much higher basal These results confirm the role of p21 in inducing Rb levels of Rb than the wild-type cells, and it showed dephosphorylation and degradation in response to no Rb dephosphorylation and no decrease in Rb after DNA damage.

Oncogene p21 regulates Rb degradation EV Broude et al 6957 In summary, we have found that p21 (which inhibits The sequence of events, wherein p21 first activates Rb CDK2 and CDK4/6) and p27 (a selective CDK2 through dephosporylation and then deactivates it inhibitor) cause not only dephosphorylation but also through depletion, suggests negative feedback regula- degradation of Rb. The finding that p16 (a selective tion that limits the duration of Rb-mediated checkpoint CDK4/6 inhibitor) does not deplete Rb would at first arrest. It is noteworthy that senescent (permanently glance suggest that phosphorylation by CDK2, but not arrested) fibroblasts express primarily p16 rather than by CDK4/6, protects Rb from proteolytic degradation. p21, and p16 does not cause Rb degradation. We note, However, this interpretation seems unlikely, because however, that this pattern is not common to all cellular expression of p16 was shown to inhibit not only senescent fibroblast cultures, since some senescent CDK4/6 but also CDK2, by causing reassortment of fibroblasts show apparent Rb depletion by immuno- cyclin–CDK-inhibitor complexes (Jiang et al., 1998). On blotting (Stein et al., 1999). In contrast to normal the other hand, p21 and p27 are structurally much senescent fibroblasts, p21 expression is maintained over more similar to each other than to p16 outside their a long term in p16-deficient tumor cells rendered CDK-binding domains, and it is plausible that their senescent by DNA damage (Chang et al., 2002), and interactions with non-CDK proteins are responsible in such cases Rb appears dispensable for long-term for the degradation of dephosphorylated Rb. This maintenance of cell-cycle arrest. p21-mediated transcrip- interpretation is also consistent with the observed tional inhibition of multiple cell-cycle progression genes dephosphorylation but not degradation of Rb in cells (Chang et al., 2000) provides one of the mechanisms that expressing low levels of p21 (Figures 2a and 3a). The can sustain cell-cycle arrest in the absence of Rb, and requirement for additional events beyond dephosphory- additional mechanisms will undoubtedly be elucidated lation can explain why the depletion of dephosphorylated in the future. Rb is detectable in some, but not all, human cell lines p21 is almost never mutated in cancers, and its pattern exposed to DNA damage (see, for example, Figure 2 of expression in human tumors is consistent with a role in Smith et al., 1994). of both a tumor suppressor and an oncogene (Roninson, Rb degradation is likely to be mediated by the 2002). The finding that p21 both activates Rb by proteasome, since it is inhibited by ALLN that has dephosporylation and inactivates it by degradation proteasome inhibition as its primary effect (Drexler, provides another explanation for the dual tumor 1997). p21-induced Rb degradation is unlikely to be suppressor/oncogenic functions of p21. mediated by Mdm2, because it was not prevented by the inhibition of p53, a positive regulator of Mdm2. Acknowledgements Although there is no evidence for a direct binding between p21 and Rb, p21 was reported to be associated We thank Dr Bert Vogelstein for HCT116 derivatives, Dr with 20S proteasome that also degrades Rb (Sdek et al., Didier Trono for pLV-tTR-KRAB-red and Natalie Warholic 2005), and it is conceivable that p21 could drive some as and Jennifer Huntington for technical assistance. This work yet unidentified Rb-containing multimeric complexes to was supported by NIH grants R01 AG17921, R01 CA62099 the proteasome. and R01 CA95727 (IBR).

References

Boyer SN, Wazer DE, Band V. (1996). E7 protein of human human tumor cells by chemotherapeutic drugs. Oncogene papilloma virus-16 induces degradation of retinoblastoma 18: 4808–4818. protein through the ubiquitin–proteasome pathway. Cancer Drexler HC. (1997). Activation of the cell death program by Res 56: 4620–4624. inhibition of proteasome function. Proc Natl Acad Sci USA Bunz F, Dutriaux A, Lengauer C, Waldman T, Zhou S, Brown 94: 855–860. JP et al. (1998). Requirement for p53 and p21 to sustain G2 Higashitsuji H, Itoh K, Nagao T, Dawson S, Nonoguchi K, arrest after DNA damage. Science 282: 1497–1501. Kido T et al. (2000). Reduced stability of retinoblastoma Chang BD, Roninson IB. (1996). Inducible retroviral vectors protein by gankyrin, an oncogenic ankyrin-repeat protein regulated by lac repressor in mammalian cells. Gene 183: overexpressed in hepatomas. Nat Med 6: 96–99. 137–142. Jiang H, Chou HS, Zhu L. (1998). Requirement of cyclin Chang BD, Swift ME, Shen M, Fang J, Broude EV, Roninson E-Cdk2 inhibition in p16(INK4a)-mediated growth sup- IB. (2002). Molecular determinants of terminal growth pression. Mol Cell Biol 18: 5284–5290. arrest induced in tumor cells by a chemotherapeutic drug. Kandel ES, Chang BD, Schott B, Shtil AA, Gudkov AV, Proc Natl Acad Sci USA 99: 389–394. Roninson IB. (1997). Applications of green fluorescent Chang BD, Watanabe K, Broude EV, Fang J, Poole JC, protein as a marker of retroviral vectors. Somat Cell Mol Kalinichenko TV et al. (2000). Effects of p21Waf1/Cip1/Sdi1 on Genet 23: 325–340. cellular gene expression: implications for carcinogenesis, Kehn K, Fuente CL, Strouss K, Berro R, Jiang H, Brady J senescence, and age-related diseases. Proc Natl Acad Sci et al. (2005). The HTLV-ITax oncoprotein targets the USA 97: 4291–4296. retinoblastoma protein for proteasomal degradation. Onco- Chang BD, Xuan Y, Broude EV, Zhu H, Schott B, gene 24: 525–540. Fang J et al. (1999). Role of p53 and p21waf1/cip1 in Maliyekkel A, Davis BA, Roninson IB. (2006). Cell cycle senescence-like terminal proliferation arrest induced in arrest drastically extends the duration of gene silencing after

Oncogene p21 regulates Rb degradation EV Broude et al 6958 transient expression of short hairpin RNA. Cell Cycle 5: Stein GH, Drullinger LF, Soulard A, Dulic V. (1999). 2390–2395. Differential roles for cyclin-dependent kinase inhibitors Ossovskaya VS, Mazo IA, Chernov MV, Chernova OB, Strezoska p21 and p16 in the mechanisms of senescence and Z, Kondratov R et al. (1996). Use of genetic suppressor differentiation in human fibroblasts. Mol Cell Biol 19: elements to dissect distinct biological effects of separate p53 2109–2117. domains. Proc Natl Acad Sci USA 93: 10309–10314. Tan X, Martin SJ, Green DR, Wang JY. (1997). Degradation Roninson IB. (2002). Oncogenic functions of tumour suppres- of retinoblastoma protein in tumor necrosis factor- sor p21(Waf1/Cip1/Sdi1): association with cell senescence and CD95-induced cell death. J Biol Chem 272: and tumour-promoting activities of stromal fibroblasts. 9613–9616. Cancer Lett 179: 1–14. Waldman T, Lengauer C, Kinzler KW, Vogelstein B. (1996). Sdek P, Ying H, Chang DL, Qiu W, Zheng H, Touitou R et al. Uncoupling of S phase and mitosis induced by anticancer (2005). MDM2 promotes proteasome-dependent ubiquitin- agents in cells lacking p21. Nature 381: 713–716. independent degradation of retinoblastoma protein. Mol Wiznerowicz M, Trono D. (2003). Conditional suppression of Cell 20: 699–708. cellular genes: lentivirus vector-mediated drug-inducible Smith ML, Zhan Q, Bae I, Fornace Jr AJ. (1994). Role of RNA interference. J Virol 77: 8957–8961. retinoblastoma gene product in p53-mediated DNA damage Ying H, Xiao ZX. (2006). Targeting retinoblastoma protein response. Exp Cell Res 215: 386–389. for degradation by proteasomes. Cell Cycle 5: 506–508.

Oncogene