BRG1 restricts endogenous p53 and senescence
To address whether tumor cell lines that express wild-type p53 require BRG1, we used RNA silencing to specifically deplete BRG1 in HeLa, G401, and RKO cell lines originating from cervical, rhabdoid, and colon cancers respectively. BRG1 depletion resulted in stabilization of p53 protein in each of these cell lines and was accompanied by induction of the p53 transcriptional target p21Cip1 (). In contrast, RNAi knockdown of the BRG1 paralogue Brm did not increase p53 protein levels (Fig. S1
). In HeLa cells, the complex of the human papillomavirus E6 protein and the ubiquitin ligase E6AP stimulates p53 protein degradation, while mdm2 mediated p53 ubiquitination was reported to be inactive (Hengstermann et al., 2001
). BRG1 depletion did not alter levels of E6 protein (Fig. S7
). We also observed that RNA silencing of BRG1 in HPV negative p53 wild-type cell lines G401 and RKO led to a moderate increase in p53 protein and robust induction of p21Cip1 (). BRG1 silencing also increased wild-type p53 protein levels in U2OS cells (Fig. S2
). To confirm the effects of BRG1 depletion on p53 dependent transcription, we performed quantitative real time PCR (qRT-PCR) of p21 transcripts. There was two-fold increase in p21 transcripts in BRG1 depleted cells compared to control cells (). A luciferase reporter under the control of p21 promoter was also used to assess the transcriptional activity of p53 in BRG1 depleted RKO cells. BRG1 knockdown led to a 2.7 fold increase in luciferase activity ().
BRG1 depletion activates p53 and senescence in tumor derived cell lines
Importantly, BRG1 depletion in HeLa cells led to a striking increase in senescence-associated-ß-galactosidase (SA-ß-Gal), an indicator of the senescent phenotype () and consistent with the induction of p21. While we predict p21 induction was due to activation of p53, it might be a consequence of reduced levels of this SWI/SNF subunit affecting another pathway. To address the p53 dependency of p21 induction in BRG1 depleted cells, we generated a HeLa cell line constitutively expressing dominant negative p53 mutant (dnp53). These cells were resistant to p21 induction and cell senescence following BRG1 depletion, confirming the p53 dependence of these effects (Fig. S3
). Taken together, these data suggest that tumor cells must retain BRG1 expression to restrict transcriptional activation and growth arrest by p53.
BRG1 partners with CBP in regulation of p53
Two lines of evidence hinted that CBP complexes with BRG1. First, CBP recruits BRG1 to the interferon ß promoter (Agalioti et al., 2000
). Second, BRG1 and CBP proteins were co-purified in a complex with an activated form of the TGF-ß pathway effector Smad2 (He et al., 2006
). These findings prompted us to test whether HATs cooperate with BRG1 to inactivate p53 in proliferating cells. G401 cells were transfected with shRNA targeting CBP or p300. Knockdown of CBP reduced levels of CBP protein and conversely increased p53 and p21Cip1 proteins (). Activation of p53 was also observed in U2OS and HeLa cells following depletion of CBP (data not shown). CBP knockdown did not alter the levels of BRG1 but there was a slight increase in p300 levels. In contrast, CBP and BRG1 levels remained unchanged and p53 was not activated following silencing of p300. To test p53 transcriptional activity under these conditions, we transfected G401 cells with a luciferase reporter under the control of the p21 promoter along with control, CBP or p300 siRNAs (). There was a four-fold induction of luciferase in CBP knockdown cells compared to control and p300 siRNA transfections. To test the biological impact on the p53 pathway, we performed colony formation assays in G401 cells after RNAi knockdown of CBP and p300. CBP depletion inhibited colony formation in accord with activation of p53, whereas p300 shRNA did not (). Furthermore knockdown of p53 in colony assays revealed that cell growth inhibition due to CBP knockdown was p53 dependent. These results imply that both BRG1 and CBP are required for proliferation of cells that express wild-type p53.
From RNA silencing experiments in cancer cell lines and inhibition of p53 activity by BRG1-CBP partnership, we hypothesized that BRG1 physically interacts with CBP and this complex attenuates p53. We therefore undertook a biochemical approach to test the physical interaction between BRG1 and CBP. Endogenous CBP and p300 immunoprecipitations from HeLa cell extracts revealed that each contained BRG1 (). Interestingly, the CBP immunoprecipitate did not include p300 and the p300 immunoprecipitate did not pull down CBP, implying BRG1-CBP and BRG1-p300 are distinct complexes. Consistent with these results, BRG1 co-immunoprecipitated both CBP and p300 (). To ascertain whether BRG1 binds to CBP in vitro
, fragments of CBP fused to glutathione S-transferase (GST) were purified from E. coli (Turnell et al., 2005
). Baculoviruses were constructed to express in insect cells full-length BRG1 and an in-frame deletion (called NBRG) of amino acids 70-340 in the N-terminal proline-rich region (PRR) as this domain is not present in Brm. Binding assays showed that full-length BRG1 protein bound to the C/H3 region of CBP spanning residues 1460-1891 () while NBRG did not bind to GST-CBP fragments. It can be argued that NBRG is misfolded and on this basis is CBP binding defective. Therefore we introduced NBRG into H1299 cells and analyzed the immunoprecipitate for the presence of SWI/SNF subunit BAF47/Ini1 (Fig. S4A
). NBRG immunoprecipitates contained BAF47 similar to BRG1. Moreover we introduced NBRG into SW13 cells and assayed for induction of the BRG1 transcriptional target CD44 and found that NBRG induced CD44 similar to BRG1 (Fig. S4B
). BRG1 has been reported to co-immunoprecipitate with p53 though direct binding has not been proven (Lee et al., 2002
; Napolitano et al., 2007
). To confirm the BRG1-p53 interaction, we transfected BRG1 and p53 null H1299 cells with Flag-BRG1 and p53. Reciprocal immunoprecipitations of BRG1 and p53 indicate these proteins interact in vivo
(). Additional in vitro
binding experiments with purified and benzonase treated proteins showed that p53 bound to BRG1 and weakly to NBRG (). These results imply that direct binding of BRG1 to CBP and strong binding of BRG1 with p53 requires the N-terminal PRR of BRG1.
BRG1 interacts with CBP/p300 and p53
The N-terminal PRR of BRG1 is crucial for p53 destabilization and inactivation
Despite a high degree of sequence identity between BRG1 and Brm (), RNAi depletion of BRG1 but not Brm activated p53. BRG1 but not Brm or NBRG co-precipitated p53 (). We therefore inferred a role for the PRR unique to BRG1 in p53 destabilization. Moreover, as NBRG did not bind CBP, we questioned whether it would exert dominant-negative effects on steady state levels of endogenous p53 and its activation. Over-expression of NBRG or Brm, but not luciferase, increased levels of p53 protein in HeLa cells (). Cycloheximide chase experiments revealed that heterologous expression of luciferase did not change the half-life of p53 protein but NBRG and Brm increased the half-life of p53 protein from <30 minutes to >60 minutes ().
BRG1 and CBP promote destabilization of endogenous p53
N-terminal PRR of BRG1 is required for p53 destabilization
In another series of experiments, transfection of wild-type p53 inhibited colony formation of C33a cells that lack functional p53 and are deficient for BRG1. Growth suppression of C33a cells by p53 was neutralized by co-transfection with BRG1 but not with NBRG (Fig. S5
). Interestingly p53-mediated inhibition of colony formation was reversed by the BRG1-KR point mutant that is defective for chromatin remodeling (Khavari et al., 1993
). Since the PRR of BRG1 is required for binding to CBP and p53, these results imply that the PRR is crucial for p53 destabilization and inactivation.
BRG1 and CBP are required for poly-ubiquitination of p53 in vivo
While several factors are known to ubiquitinate p53, we queried whether BRG1 and CBP have critical roles in promoting p53 poly-ubiquitination. Endogenous p53 was immunoprecipitated from HeLa cell lysates and blotted with antibodies to ubiquitin. Heterogeneous forms of poly-ubiquitinated p53 were abundant in vector control transfected extracts (, lane 2). RNAi depletion of BRG1 (lane 3) or CBP (lane 4) led to a dramatic reduction of ubiquitinated p53 species and accordingly increased steady-state p53 protein levels. Although CBP knockdown was not as complete, we observed a near total abrogation of p53 poly-ubiquitination in HeLa cells. An explanation how this partial CBP knockdown led to abrogation of p53 ubiquitination might be that multiple p53 ubiquitination effectors converge on CBP. Similar results were noted following shRNA silencing of BRG1 in U2OS cells (Fig. S2
). Ubiquitination of an unrelated protein was slightly increased in shBRG1 treated cells, indicating that RNAi to BRG1 did not inactivate the ubiquitin pathway (data not shown). Cycloheximide chase experiments indicated prolongation of p53 protein half-life to > 2 hours following RNAi knockdown of BRG1 (). CBP knockdown also increased the half-life of p53 protein to 120 min. These results reinforce our model that BRG1 and CBP collaborate to destabilize endogenous p53 protein in proliferating cancer cells.
DNA damage effects on BRG1 and CBP complexes
Since RNA silencing of BRG1 and CBP activated p53 and BRG1 interacts with CBP, we hypothesized that BRG1 might dissociate from CBP to allow p53 accumulation. In response to DNA damage, p53 levels began to increase after 6 hours (). To test whether HAT proteins CBP and p300 dissociate from BRG1, we analyzed CBP or p300 immunoprecipitations for the presence of BRG1 seven hours post doxorubicin treatment. CBP and p300 clearly associated with BRG1 in unstressed cells ( lanes 3,4). Upon DNA damage, reduced amounts of BRG1 were in complex with CBP and p300 (lanes 7,8). These results support our hypothesis that association of BRG1 with CBP promotes degradation of p53, and conversely, dissociation of BRG1 from CBP following DNA damage allows p53 stabilization and activation.
DNA damage inhibits BRG1-CBP/p300 interaction
Since BRG1 and CBP modify chromatin fluidity, we reasoned that their inactivation of p53 might occur at a p53 transcriptional target promoter region. To confirm p21 transcript induction by activated p53, we treated U2OS cells with doxorubicin and assayed p21 transcripts by quantitative real time PCR. Compared to untreated cells, levels of p21 RNA increased beginning at four hours post-treatment (). Subsequently we performed quantitative chromatin immunoprecipitation (qChIP) to determine occupancy of the p21 promoter by p53, BRG1, CBP and p300 in doxorubicin treated versus untreated U2OS cells. Commonly, p53 ChIP assays are conducted using the p53 monoclonal antibody DO-1 that recognizes an epitope on the transactivation region of p53 where mdm2 and other transcription complexes bind. Coincidently this region is subjected to phosphorylation in response to DNA damage signal. Therefore we deliberately chose to use rabbit antibodies that were raised against the full-length p53 protein to capture p53 at the p21 promoter. The qChIP assay revealed that total p53 and p300 present at the p21 promoter were only increased slightly subsequent to DNA damage (). In contrast, there was reduced CBP and BRG1 after DNA damage compared to untreated cells. These results support our hypothesis that p53 is inactivated at the site of recruitment by a BRG1-CBP complex in proliferating cells. Conversely, p53 activation corresponded with reduced levels of BRG1-CBP at the p21 promoter.
DNA damage reduces the levels of BRG1 and CBP on p21 promoter