Overexpression of Smad1 in GBM is associated with tumorigenesis and predicts poor prognosis
An overview of TCGA RNA-seq data derived from the Human Protein Atlas revealed that gliomas had the highest SMAD1 transcriptional expression among cancers (Supplementary Fig. 1A). The significant alteration of SMAD1 was identified across all tumor samples and paired normal tissues (Supplementary Fig. 1B). Low-grade glioma (LGG) and GBM both had higher SMAD1 expression. Among cancers with significant SMAD1 alterations, GBMs have the highest fold change (Supplementary Fig. 1C). Subsequently, samples with significant SMAD1 transcriptional expression were further characterized into different glioma datasets. It was revealed that SMAD1 was highly expressed in gliomas, covering different histology, grades, and subtypes (Fig. 1A). Moreover, survival analysis indicated that higher SMAD1 predicted poor GBM patient survival outcomes (Fig. 1B). To further explore whether SMAD1 was a valuable predictor for GBM prognosis, Gene Set Enrichment Analysis (GSEA) was performed based on genes correlated with SMAD1 expression derived from the TCGA GBM dataset. According to the C2All gene sets collected, the genes positively correlated with SMAD1 were enriched in GBM-related phenotypes (Fig. 1C). From the perspective of biological processes (GoBP), the genes positively correlated with SMAD1 were associated with astrocyte and glial cell differentiation, gliogenesis and neuronal stem cell population maintenance (Fig. 1D). To determine the significance of SMAD1 in human GBM, we cultured primary cells from patient-derived GBM tissue resections using 3D-culture method (Fig. 1E). The second passage of 3D-cultures were used for xenograft models and analysis. Western blot analysis detected the expression of Smad1 in patient-derived 3D-cultures (Fig. 1F). Immunofluorescence (IF) further illustrated that these patient derived spheres were Nestin positive and observed variable Smad1 expression (Fig. 1G). The second passage of spheres were dissociated and cultured according to the 3D method, with continuous observation of sphere growth (Fig. 1H). The tumor sphere growth was expressed as an amplification index by calculating the area of sphere at 120 h relative to the area of sphere at 24 h. A correlation analysis indicated that Smad1 protein levels were positively correlated with amplification of the patient-derived tumor spheres. To examine the correlation of Smad1 and GBM tumorigenesis, the third passage of 3D-cultures were implanted subcutaneously in nude mice. The xenografts derived from cultures with higher Smad1 expression grew more quickly than those derived from cultures with lower Smad1. As shown in Fig. 1I, the expression levels of Smad1 in these cells were positively correlated with tumor volume derived from the 3D-cultures, indicating that higher Smad1 corresponds to stronger tumorigenicity. To examine the general alteration of Smad1 protein in GBM, immunohistochemistry (IHC) was performed using a GBM tissue microarray containing 60 tumor samples and 4 normal samples (Fig. 1J). As shown in the representative images, Smad1 was mostly expressed in the nuclei of normal brain and GBM tissues. It showed that Smad1 protein levels were significantly higher in GBM than it in normal brains (Fig. 1K). Survival analysis based on the cutoff at median level of Smad1 in GBM indicated that higher Smad1 was associated with poorer clinical prognosis (Fig. 1L, Log-rank χ2 = 5.019, p = 0.025) regardless of age and sex (Supplementary Data 1 and Supplementary Table 1). Collectively, these data suggest that Smad1 can be regarded as a valuable prognostic factor for GBM.
Smad1 promotes GBM tumorigenesis and chemoresistance in vitro and in vivo
To determine whether Smad1 functions on GBM phenotypes, GBM cell lines were selected to establish Smad1 overexpression and depletion models based on Smad1 protein level (Supplementary Fig. 2A). Smad1-targeting CRISPR/Cas9 and corresponding controls were used to knockout (KO) SMAD1 expression in U87 and U251 cells containing higher levels of endogenous Smad1 (Supplementary Fig. 2B), whereas Smad1 overexpressing cells were generated from U118 and A172 cell lines with lower (relative) levels of Smad1 expression (Supplementary Fig. 2C). Smad1 KO resulted in a significant inhibition of cell proliferation and DNA synthesis as evidenced by cell growth assay (Fig. 2A), EdU incorporation (Fig. 2B), and colony formation assays (Fig. 2C). Smad1 depletion similarly resulted in a reduced number of invasive cells as shown by invasion assay (Fig. 2D). Conversely, ectopic Smad1 expression promoted cell proliferation, DNA synthesis, colony formation, and invasion (Fig. 2A-D). To assess the effect of Smad1 on tumorigenesis in vivo, we performed intracranial xenograft assay in nude mice using U87 cells with Smad1 depletion or control vectors. The tumor volumes were monitored by MRI (Fig. 2E). Smad1 depletion resulted in a significant decrease of tumor growth (Fig. 2E) and tumor cell proliferation (Fig. 2F). We subsequently determined the effect of Smad1 on cell apoptosis using Annexin V/7-aminoactinomycin D (7-AAD) staining and flow cytometry analysis. GBM cells were treated with a low dose of Doxorubicin (Dox, 2 µM) for 24 hours and apoptotic cells were analyzed. As shown in Fig. 2G, Dox induced a slight apoptosis of U87 and U251 cells, which mainly occurred in the late stage of apoptosis. Smad1 repression promoted Dox-induced apoptosis in U87 and U251 cells. On the contrary, Smad1 overexpression decreased Dox-induced apoptosis in A172 and U118 cells, suggesting the protective effect of Smad1 on cellular apoptosis induced by chemotherapy. Next, xenograft assay in nude mice was used to determine whether Smad1 depletion promotes chemosensitivity of GBM in vivo. U87 cells with Smad1 depletion or control vectors were injected subcutaneously into nude mice. Figure 2H illustrates the schematic time line of the U87 flank models. Briefly, low dose TMZ (20 mg/kg) was injected 21 days post-injection for 5 days, once a day. The tumor size was measured every 3 days from the 21st day post-subcutaneous injection. The tumors were collected 21 days after cell injection (Fig. 2I) and on the 15th day after TMZ treatment (Fig. 2L) respectively. Consistent with the results of the intracranial xenograft, tumor weight analysis of derived tumors at the 21th day after implantation showed that Smad1 depletion significantly inhibited the growth of subcutaneous transplanted tumors (Fig. 2I). The tumor volume of the two groups gradually decreased with the treatment of TMZ, suggesting that even low dose of TMZ exhibits inhibitory effects on tumor growth (Fig. 2J). Comparing the tumor inhibitory rate by calculating the tumor volume during treatment to the tumor volume before treatment (Fig. 2K) and the ultimate tumor volume (Fig. 2L), Smad1 knockout resulted in a significant tumor regression. Collectively, these data provide evidence that Smad1 is an important oncoprotein regulating GBM tumorigenicity and chemosensitivity.
The Regulation Of Smad1 In Gbm Mechanically Targets P53
Next, whether Smad1 regulates GBM tumorigenicity through canonical TGF-β/BMP signaling was explored. We examined the effects of small molecule inhibitors SB431542 (SB) against TGF-β type I receptors or Dorsomorphin (Dor) against BMP type I receptors. It was observed that SB had no effects on the growth of GBM cells, whereas Dor demonstrated cell growth inhibition (Supplementary Fig. 3A, B). Subsequently, ectopic Smad1-expressing cells were used to examine the effects of these two inhibitors on cell growth. Unsurprisingly, SB failed to affect cell growth, but Dor exhibited inhibitory effects on cell growth of A172 and U118 cells expressing ectopic Smad1 (Supplementary Fig. 3C), confirming a general understanding that Smad1 is at least partially regulated by BMP signaling [21]. By comparing the cell growth inhibition, Dor treatment resulted in a lower inhibitory ratio in Smad1 overexpressing cells (Supplementary Fig. 3D). Colony formation assays additionally showed that Dor repressed cell proliferation of normal cells or Smad1 overexpressing cells, though the inhibitory effect on Smad1-overexpressing cells was relatively weaker (Supplementary Fig. 3E). These results re-iterated that BMP signaling plays important roles in regulating GBM cell growth, while Smad1 overexpression partially overcomes Dor induced inhibition. Given that Smad1 is translocated into the nucleus upon activation, we determined whether TGF-β/BMP signaling inhibition altered the cellular distribution of Smad1. As shown in Supplementary Fig. 3F, Smad1 predominately accumulates in the nucleus of vehicle treated U87 and U251 cells, and Dor and SB failed to alter Smad1 nuclear localization. These results suggest that Smad1 functions in GBM in a TGF-β/BMP- independent manner. To explore alternative pathways and mechanisms regulated by Smad1, we performed cDNA microarray on U87 cells with SMAD1 depletion or negative control constructs. Applying a 1.5-fold change (p < 0.05) as a cutoff, Smad1 KO affected the expression of 1103 genes in U87 cells (604 down-regulated and 458 up-regulated; Supplementary Data 2). Assessment of Gene Ontology of differentially expressed genes revealed that Smad1 depletion altered several pathways. Genes down-regulated by Smad1 KO were associated with chromosome organization and cell cycle process, while genes up-regulated by Smad1 depletion were associated with protein binding and secretion (Fig. 3A). Smad1 regulated genes play roles in functional processes, including cell cycle, apoptosis, and migration (Fig. 3B; Supplementary Data 3). Further enrichment analysis (Fig. 3C) indicated that down-regulated genes induced by Smad1 depletion were negatively associated with apoptotic process and cell death, while upregulated genes induced by Smad1 depletion positively regulated apoptotic process and cell death. Given the predominance of the p53-mediated apoptosis pathway, we examined the possibility of Smad1 in regulating p53 activity. To this end, we analyzed the occupancy of p53 on the promoters of Smad1 depletion-induced genes using chromatin immunoprecipitation sequencing (ChIP-seq) data published previously (GSE46641) [22]. Findings showed that of the 2079 genes containing peaks of p53 binding, the distance between p53-binding region (p53BR) and gene transcription starting site (TSS) of 196 genes was less than 20 kb. Interestingly, 29% (32/110) of the Smad1 depletion-induced genes had p53 binding regions within 20 kb (Supplementary Data 4), a rate significantly higher than the average p53 occupancy in the genome of Smad1 depletion-reduced genes (p < 0.0001, χ2). We next examined the effects of Smad1 on p53 transcriptional activity using luciferase reporter and qRT-PCR assay. Using a luciferase reporter containing a p53-responsive promoter sequence, we observed that Smad1 overexpression or depletion significantly increased or decreased the transcriptional activity of p53 (Fig. 3E). To detect the expression of p53 transcriptional targets, we listed genes upregulated after Smad1 KO (p < 0.05), identifying those which overlapped as direct p53 target genes (Fig. 3F) [23]. Three known tumor suppressive genes (BAX, AIFM3, and ATF3) were selected to evaluate the regulation of Smad1 on p53 transcriptional activity. In the subsequent qRT-PCR assay, we observed that Smad1 depletion significantly increased mRNA expression of p53 target genes, while Smad1 overexpression suppressed the expression of these genes (Fig. 3G). Moreover, we analyzed whether Smad1 inhibited the binding of p53 to target promoters by quantitative ChIP (qChIP). As a result, Smad1 depletion significantly increased the binding of p53 to target promoters, and Smad1 overexpression resulted in a decrease of p53 binding to target promoters (Fig. 3H). Collectively, these findings suggest that Smad1 exerts onco-functions in GBM via negative regulation of p53 transcriptional activity.
Smad1 Binds And Impairs P53 Acetylation Via Mh1 And Mh2, Respectively
GSEA results of GO BP (Fig. 1D) and GO MF (Supplementary Data 5) evaluation based on TCGA-GBM datasets indicated that genes correlated with Smad1 expression are related to histone deacetylation processes, p53-binding, and histone acetyltransferase binding. Given that acetylation is critically required for the activation of p53 and subsequent biological process [9, 11], the findings suggested that Smad1 may exert an inhibitory effect on p53 acetylation. To this end, we examined the alteration of p53 acetylation in GBM cells with Smad1 ectopic expression or depletion. We found that depletion of Smad1 enhanced (Fig. 4A), but Smad1 overexpression (Fig. 4B) decreased the acetylation of p53 at Lys382 (Ac382) and Lys373 (Ac373) in the indicated GBM cells without affecting overall p53 protein level. Moreover, IP analysis using acetylated-lysine antibodies revealed that Smad1 overexpression inhibited overall p53 acetylation in U118 and A172 cells (Fig. 4C). In 293T cells, Smad1 overexpression inhibited acetylation of ectopic p53 (Fig. 4D), indicating that Smad1 impairs p53 acetylation across a wide range of lysine sites. To confirm whether Smad1 inhibits new-acetylation of p53, Dox was used as an acetylation inducer [24, 25]. As shown in Fig. 4E, Dox chase-induced p53 acetylation in U118 and A172 cells was inhibited by Smad1 overexpression. Next, IP assay determined the endogenous interaction between p53 and Smad1 (Fig. 4F), revealing extensive co-localization of p53 and Smad1 in the nucleus of U87 and U251 cells (Fig. 4G). To identify the interaction region of Smad1 with p53, truncated mutants were established based on Smad1 protein domain structure (Fig. 4H). Smad1 without the MH2 domain (dMH2, depleting residues 182–496) exhibited comparable p53-binding ability as full-length Smad1, whereas MH1 deletion mutants (dMH1, depleting residues 1–331) completely abolished the p53 binding ability (Fig. 4I), indicating that the MH1 domain is essential for Smad1 binding of p53. Considering that the L3 Loop is an important structure mediating the interaction between SMADs and other proteins [26–28], the role of the L3 Loop in the interaction between Smad1 and p53 was explored. Although located in the MH2 domain, L3 Loop depletion (dLoop) markedly impaired the p53-binding ability of Smad1, indicating that the L3 Loop is another important domain affecting Smad1 binding of p53. An additional observation was that the lysine mutant of the L3 Loop (K418R) did not affect the binding of Smad1 to p53. An unexpected observation was that the loss of MH1 or the L3 Loop failed to rescue the inhibition of Smad1 on p53 acetylation, suggesting the inhibition of Smad1 on p53 acetylation is not determined by its p53 binding capacity. Alternatively, MH2 deletion led to the loss of the inhibitory effect of Smad1 on p53 acetylation. Moreover, the effect of MH1 and MH2 in Smad1-mediated endogenous p53 acetylation inhibition was confirmed in U87 and U251 cells (Fig. 4J). Collectively, these findings suggest that Smad1 binds p53 through the MH1 domain, but inhibits p53 acetylation via the MH2 domain.
Smad1 binds p300 via the MH2 domain and interferes with the interaction of p53 and p300
Due to the physical interaction between the acetyltransferase p300 and Smad1 [29–31], we examined whether Smad1 inhibits p53 acetylation by impairing p300-mediated p53 acetylation. p300 knock-down constructs mediated by siRNAs decreased the acetylation of p53 in U87 and U251 cells (Fig. 5A), indicating the involvement of p300 in p53 acetylation in GBM cells. p300 knock-down further abolished the increase of p53 acetylation induced by Smad1 depletion (Fig. 5B). Next, we probed the interaction between Smad1 and p300 (Fig. 5C), observing that this interaction was localized in the nucleus of GBM cells (Fig. 5D), as confirmed by the complex formation of Smad1, p53 and p300. A subsequent IP assay in 293T cells with ectopic p53 and p300 showed that Smad1 introduction inhibited the interaction between p53 and p300, and subsequent p53 acetylation (Fig. 5E, F). Using a GFP plasmid as an unrelated protein control, Smad1 depletion resulted in an increase of p53 and p300 interaction, while ectopic Smad1 weakened the binding of endogenous p53 and p300 in GBM cells (Fig. 5G). Next, the binding region of Smad1 to p300 was identified based on the truncated mutants of Smad1 (Fig. 5H). These showed that MH2 depleted Smad1 (dMH2) failed to bind p300, though other domain deletions or point mutations had no observable effects on Smad1 and p300 interaction. These findings indicated that MH2 is essential for Smad1 binding of p300. Subsequent IP assay in 293T cells co-expressing ectopic p53 and Smad1 deletion mutants indicated that dMH2 failed to inhibit the interaction between p53 and p300, while dMH1 possessed the capacity to inhibit the interaction of p53 and p300 (Fig. 5I). These observations addressed an underlying mechanism of p53 and Smad1 competitive binding on the same p300 domains. Given the evidence that the C-terminal domain is critical for p53-binding of p300 [32, 33], a C-terminal deletion mutant of p300 (del 1644-2414aa) was established. The IP results showed that both p53 and Smad1 lost the ability to bind p300 (Myc-p300MT; Fig. 5J), confirming the assumption that p53 and Smad1 bind p300 at the same domain. As a result, Myc-p300MT lost the regulatory control of p53 acetylation. Another interesting finding was that Smad1 itself was acetylated in p300WT overexpressed cells, but not in p300MT expressing cells (Fig. 5K). Considering that competitive inhibition is normally a reciprocal process, we further explored whether p53 inhibits Smad1 binding of p300 in U87 with depleted p53 or in p53 re-expressed cells (Fig. 7A). The result indicated that p53 depletion did not induce an increase of Smad1 and p300 interaction as expected. p53 re-expression also failed to affect the interaction of Smad1 and p300 (Fig. 5L). In other words, p300 bound Smad1 is not be replaced by p53 in GBM cells. Combined with the observations in Fig. 4H&I, these results document that the inhibitory effect of Smad1 on p53 acetylation is dependent on MH2-mediated p300 recruitment, thus preventing p300 binding and p53 acetylation.
P300-mediated Acetylation Is Vital For Smad1 Oncofunctions
Considering the interaction between Smad1 and p300, we speculated that Smad1 is acetylated by p300 in GBM cells. To this end, whether Smad1 was acetylated in GBM cells was examined. As shown in Fig. 6A, there was a considerable acetylation of Smad1 in the indicated GBM cell lines, and p300 knock-down resulted in an obvious decrease of Smad1 acetylation in U87 and U251 cells (Fig. 6B). To explore the acetylation site of Smad1 regulated by p300, candidate sites were predicted using ASEB, a web service of the lysine-acetyl-transferase (KAT)-specific acetylation site prediction platform (http://cmbi.bjmu.edu.cn/huac) [34]. Findings suggested that p300 could be potentially acetylating Smad1 at 4 lysine sites. Based on these lysine sites, we constructed unit point and multi-site mutants of Smad1, converting the indicated lysine residues to Arginine (R) (Fig. 6C). These mutants were expressed in U87 cells and the acetylation of ectopic Smad1 was detected by IP assay using an acetylated-lysine antibody (Fig. 6D). Findings revealed that the acetylation of K81R, 2KR was comparable to that of wild-type Smad1, while the acetylation of mutants containing K373R (including K373R and 4KR) were repressed, suggesting K373 is an important site for determining Smad1 acetylation. Subsequently, the SBE reporter assay indicated that Smad1 overexpression resulted in a significant promotion of SMAD responsive transcriptional activity, while K373R expression lost this promotion (Fig. 6E). Accordingly, K373R failed to induce the expression of ID2, a typical SMAD target gene which was markedly induced by wild-type Smad1 overexpression (Fig. 6F). However, K373R retained acetylation inhibition on p53 as wild-type Smad1 does in GBM cells (Fig. 6G), indicating that acetylation status does not affect the inhibition of Smad1 on p53 acetylation. Subsequent functional study in K373R-expressing cells revealed the important role of acetylation in growth promotion of GBM cells (Fig. 6H, I). However, K373R contained the anti-apoptotic effects of wild-type Smad1 in GBM cells (Fig. 6J). Moreover, as shown in Fig. 5L, neither p53 depletion nor re-expression affected Smad1 acetylation in GBM cells, disclosing that SMAD1 has the priority over p53 in the acetylation mediated by p300. Therefore, these findings demonstrate a regulatory relationship among p53, Smad1 and p300 in GBM cells (Fig. 6K).
Acetylation Reverses The Gain Of Function Of Missense Mutant P53 And Enhances Chemosensitivity
The consensus is that TP53 mutations are associated with malignant phenotypes in cancers by gain-of-function (GOF) [35]. The finding that Smad1 inhibited p53 acetylation in U251 and U118 cells containing p53 missense mutations suggested that Smad1 negatively regulates the acetylation targeting either wild-type or mutant p53. To further confirm the finding that Smad1 impairs the acetylation of mutant p53, we established a p53 depleted U87 cell line mediated by CRISPR/Cas9 (Fig. 7A). Based on this p53-depleted cell line, p53 hotspot missense mutants, including R175H, R273C, and R248Q, were introduced with the aid of site mutations according to the sgRNA sequence (sgMut). As shown in Fig. 7B, Smad1 overexpression resulted in a decrease of acetylation of the indicated p53 mutants without affecting overall p53 expression. Given that promoting acetylation restores mutant p53 DNA binding and transcriptional activity and reverse its GOF in cancers [15, 16], our findings in p53 mutant-containing U251 and U118 cells suggest that the inhibitory effect of Smad1 on p53 acetylation is the basis for the oncogenic effects of mutant p53. Considering that the C-terminal domain (CTD) is a determinant of p53 DNA binding and tumor suppression [36], we explored whether acetylation can reverse the tumor promotion of mutant p53 in GBM. To this end, the CTD constitutive acetylation mimic (3KQ) was constructed based on p53 and R175H lentivirus with amino acid mutants to glutamine (Q) at K373, K381 and K382. In the following functional study of U87 stable expressing cells (Fig. 7C&D), we failed to observe the cell growth inhibition of wild-type (WT) p53 overexpression in 3D-spheroids assay (Fig. 7C; Supplementary Table 8a) and in colony formation assay (Fig. 7D; Supplementary Table 8b). On the other hand, the mimetic acetylation of wild-type p53 (WT3KQ) clearly showed a significant inhibition on tumor cell growth. Different to WT-p53, R175H introduction resulted in a significant promotion of tumor cell growth, suggesting the GOF of missense mutant p53. Interestingly, R175H acetylation simulator (R175H3KQ) significantly impaired the cell growth promotion of R175H, consistent with the view that acetylation of CTD can reverse the carcinogenic effect of mutant p53 [16]. With respect to cell apoptosis, wild-type p53 overexpression did not show obvious promotion in Dox-induced cellular apoptosis, while a significant increase of apoptosis was observed in WT3KQ expressing cells. Surprisingly, R175H introduction resulted in a moderate, but significant increase of cell apoptosis as compared to Mock or WT, and this promotion was further strengthened by its acetylated simulator (Fig. 7E; Supplementary Table 8c). DNA damage in these Dox treated cells was detected by Histone H2A.X Tyr142 phosphorylation (pH2A.X) assay (Fig. 7F). It showed that R175H expression increased DNA damage induced by Dox, and that acetylated simulator further enhanced this induction. These findings suggested that R175H and its acetylation leads to deeper DNA damage, enhancing the apoptosis induced by chemotherapy. To further confirm this in vitro finding, in vivo tumorigenesis assay was performed using luciferase labeled U87 cells containing these constructs. U87 cells with the indicated constructs were stereotactically implanted into the right hemisphere of mouse brain and received TMZ treatment once every 2 days until the 14th day post-implantation (total of 7 doses). The intracranial tumorigenesis was monitored through bioluminescence imaging 2 days after the last treatment (Fig. 7G). It showed that acetylation mimetic wild-type p53 enhanced tumor response to chemotherapy, substantiating that missense mutant tumors are endowed with higher chemosensitivity than wild-type tumors. Consequently, tumors with the R175H acetylation mimetic exhibited the highest chemosensitivity to TMZ (Fig. 7H). Collectively, these observations demonstrate that releasing p53 acetylation from the Smad1/p300 inhibition mechanism not only strengthens the function of wild-type p53, but also reinforces the chemosensitivity of GBM cells with missense mutant p53.
Smad1 inhibits p53 acetylation in vivo
To further test whether Smad1 plays a role in regulating p53 acetylation in vivo, we evaluated the correlation between Smad1 and p53 acetylation by IHC of GBM tissue array. First, we examined the level of p53 and p53 acetylation in GBM with patient’s survival information (Fig. 8A). We found that the level of acetylated p53 was positively correlated with p53 expression (Fig. 8B). There was no correlation between p53 protein levels and survival of GBM patients based on the cutoff at median expression (Fig. 8C), while a significant correlation was observed between p53 acetylation and survival outcomes (Fig. 8D). Considering the high mutation rate of TP53 in GBM [37], these findings suggest that acetylation is a benign indicator even in patients containing mutant p53. Finally, we analyzed the correlation between Smad1 and p53 in serial cross-sections of GBM tissue stained by Smad1, p53 and acetylated p53 respectively (Fig. 8E). No expression association was detected between p53 and Smad1 (Fig. 8F), though consistent with in vitro findings, a negative correlation was observed between Smad1 and p53 acetylation (Fig. 8G). To further confirm the findings, we evaluated the correlation between Smad1 and acetylated p53 (Acetylated Lys382) by multiplex immunohistochemistry in the same GBM tissue sections (Fig. 8H). The quantitative analysis confirmed the negative relationship between Smad1 and p53 acetylation (Fig. 8I).