|Home | About | Journals | Submit | Contact Us | Français|
The p53 gene is mutated in many human tumors. Cells of such tumors often contain abundant mutant p53 (mutp53) protein, which may contribute actively to tumor progression via a gain of function (GOF) mechanism. We applied ChIP-on-chip analysis and identified the VDR (vitamin D receptor) response element as over-represented in promoter sequences bound by mutp53. We report that mutp53 can interact functionally and physically with VDR. Mutp53 is recruited to VDR-regulated genes and modulates their expression, augmenting the transactivation of some genes and relieving the repression of others. Furthermore, mutp53 increases the nuclear accumulation of VDR. Importantly, mutp53 converts vitamin D into an antiapoptotic agent. Thus, p53 status can determine the biological impact of vitamin D on tumor cells.
The p53 tumor suppressor is a major barrier against cancer progression. The p53 pathway is impaired in almost all human cancers (Vogelstein et al., 2000). About 50% of human cancers carry p53 mutations (Soussi and Wiman, 2007), mostly missense mutations resulting in overproduction of mutant p53 (mutp53) protein (Weisz et al., 2007b). This might imply a strong selection for mutp53 expression in carcinogenesis. Indeed, p53 mutations result not only in loss of tumor suppressing activities by the mutant allele, but also in trans-dominant inactivation of the remaining wtp53 (Shaulian et al., 1992). Importantly, at least some cancer-associated mutp53 variants acquire oncogenic activities, defined as gain-of-function (GOF) (Weisz et al., 2007a). Specifically, mutp53 can enhance proliferation, survival and tumorigenicity in mice (Bossi et al., 2006; Weisz et al., 2004). Furthermore, at least for some types of cancer, patients harboring particular missense p53 mutations in their tumors tend to be less responsive to chemotherapy (Soussi and Beroud, 2001).
Mechanistically, mutp53 can exert a dominant negative effect over the p53 family members p63 and p73 and inhibit their biochemical and biological activities (Irwin et al., 2003; Lang et al., 2004). Moreover, mutp53 can regulate specific sets of target genes independently of p63 and p73 (Lin et al., 1995; Zalcenstein et al., 2003; Weisz et al., 2004; Scian et al., 2004). Accordingly, the transcriptional activation domain of p53 is necessary for gene regulation by mutp53 as well as for its interference with apoptosis. Most cancer-associated p53 mutations occur in the DNA binding domain and abolish the ability of the protein to bind to the specific DNA sequences recognized by wtp53. Hence, the ability of mutp53 to regulate gene expression may require interactions with other proteins that tether it to the DNA, as suggested for NF-Y (Di Agostino et al., 2006) and NF-kB (Weisz et al., 2007a).
In this study, we employed chromatin immunoprecipitation coupled with microarray analysis (ChIP-on-chip) to identify DNA regions selectively associated with mutp53.
To elucidate the molecular basis for the ability of mutp53 to modulate specific gene expression, chromatin immunoprecipitation (ChIP) coupled with promoter microarray hybridization (ChIP-on-chip; see Experimental Procedures) analysis was performed on SKBR3 breast cancer-derived cells, which harbor an endogenous mutant p53R175H. About 70 promoters were bound with a p-value of < 0.001. Table 1 lists 30 genes whose promoters scored highest.
A bioinformatics analysis was next performed on the ChIP-on-chip data in order to identify transcription factor binding motifs overrepresented in mutp53-bound promoters. Every gene was scanned for binding sites from 1000 bp upstream to 200 bp downstream from its transcription starting site (TSS), for over-representation of 414 different binding motifs relative to the genes across the whole genome (Tabach et al., 2007). A similar analysis was performed on the putative promoters of mutp53-regulated genes identified in an expression microarray experiment, performed with p53-null H1299 lung adenocarcinoma cells stably transfected with p53R175H (Weisz et al., 2004). Table 2 lists transcription factors exhibiting a statistically significant association with mutp53 in at least one of the two experiments. Remarkably, the vitamin D receptor/retinoid X receptor (VDR/RXR) response element (VDRE; consensus: AGGTCAnnnAGGTCA), which mediates the transcriptional effects of vitamin D, scored positive in both the ChIP-on-chip and the expression microarray analysis. When a similar bioinformatics analysis was applied to a ChIP-on-chip data obtained from wtp53-expressing U2OS cells using the same arrays, it identified p53RE as the most significant motif but did not score VDRE (Table S1), thus confirming the validity of the analysis. Two motifs, HEN-1 and MEF-2, were found to be over-represented in promoters bound by both wtp53 and mutp53.
To confirm the interaction of mutp53 with VDR/RXR-containing promoters and to assess the impact of the active form of vitamin D3, 1α25(OH)2D3, on such interaction, a pool of SKBR3 cells stably transfected with control shRNA (si-control) and another pool transfected with p53 siRNA (Fig. S1A) were subjected to ChIP analysis. Cells were either grown in medium containing charcoal-stripped serum to deplete residual vitamin D3 or treated with vitamin D3. Immunoprecipitation was performed using antibodies specific for either p53 or VDR, or beads only as a non-specific control. The immunoprecipitated DNA was subjected to qPCR analysis with primers specific for the regions encompassing putative VDREs in the promoters of the HIRA and TGF-β1 genes, which scored positive for mutp53 binding in the ChIP-on-chip screen. Significant binding of mutp53 to the VDRE-containing regions of both promoters was clearly detectable (Fig. 1A), but was not seen with a ChIP-control genomic sequence (bottom panel). Vitamin D3 (D3) treatment markedly increased VDR binding. Remarkably, knockdown of endogenous mutp53 reduced the binding not only of mutp53 but also of VDR. Importantly, vitamin D3 elicited a further increase in mutp53 binding relative to non-treated cells. Thus, mutp53 indeed associates with VDRE-containing regions within specific genes, and this is further enhanced by vitamin D3.
To further validate the binding of mutp53 to VDRE we performed an in vitro “Southwestern” assay, in which nuclear extracts are incubated with membrane-immobilized synthetic oligonucleotides followed by Western blot analysis to detect proteins that bound to the oligonucleotides. As seen in Fig. S1B, vitamin D3 treatment augmented the binding of VDR to a wt but not mutant VDRE sequence, as expected (IB: VDR). Importantly, mutp53 displayed weak specific binding to the VDRE under basal conditions, which was significantly augmented upon vitamin D3 treatment (IB: p53). On note, mutp53 did not bind to a wtp53 binding sequence (wtp53RE)
To investigate the functional consequences of the recruitment of mutp53 to VDRE-containing promoters, the firefly luciferase gene was placed downstream to 3 tandem copies of a canonical VDRE or a mutant incapable of VDR/RXR binding (mVDRE), In p53-null H1299 cells, cotransfection of expression plasmids encoding either p53R175H or p53R273H led to a mild increase in the transcriptional activity of the wild type but not mutant VDRE promoter. Vitamin D3 increased transcription from the wild type but not mutant VDRE (Fig. 1B, upper panel). Both cancer-associated mutp53 isoforms further increased vitamin D3-induced transcription. Importantly, wild type p53 failed to augment the transcriptional activity of this promoter. Neither the p53R175H22,23 triple mutant, possessing a defective transactivation domain (TAD), nor p73DD, which has a dominant negative effect over both p63 and p73, had any effect on transcription (Fig. S1C; see also Fig. S4E). Hence, the effects of mutp53 in our experimental models are mostly independent of other p53 family members, but require a functional p53 TAD.
Next, the same reporter plasmids were transfected into SKBR3 cells together with siRNA oligonucleotides specific for p53 (p53i), or LacZ as a control (LacZi). As expected, vitamin D3 induced a robust increase in transcription from a promoter containing wt but not mutant VDRE (Fig. 1B, lower panel). Importantly, knockdown of endogenous mutp53 reduced both basal and vitamin D3-induced transcription. Similar results were obtained with SW480 colorectal cancer cells expressing H273R mutp53 (Fig. S1D). Hence, mutp53 can cooperate with vitamin D3 to maximize VDRE-dependent transcription, at least from some promoters.
The VDR gene is a transcriptional target of wt p53 (Maruyama et al., 2006a) as well as of p63 and p73 (Kommagani et al., 2006; Kommagani et al., 2007). p63/p73 activity can be quenched by mutp53; indeed, p53R248W can elicit a dominant negative effect over transfected p73 in the regulation of VDR gene transcription (Kommagani et al., 2006; Kommagani et al., 2007). Knockdown of mutp53 in either SKBR3 or MDA-MB-231 cells caused a 30% to 50% reduction in total VDR protein (Fig 2A upper panel) but had no significant impact on VDR mRNA levels (data not shown), which suggests that mutp53 may lead to a mild stabilization of VDR. In H1299R175H-i cells, induction of mutp53 expression by Zn++ led to only a slight (30%) increase in VDR mRNA (Fig. 2B) and a mild (1.4 fold) increase in total VDR protein (Fig. 2C, lanes 1,2).
Transcriptional activation by vitamin D3 requires translocation of VDR into the nucleus. We therefore examined the effect of mutp53 on VDR localization 3 hours after addition of vitamin D3. Remarkably, mutp53 had a dramatic effect on the amount of VDR in the nuclear fraction (Fig. 2A, 2C). Immunostaining analysis revealed that while in untreated H1299R175H-i cells VDR is largely cytoplasmic, vitamin D3 caused a substantial nuclear translocation of VDR in some cells (Fig. 2D, top two rows). Interestingly, induction of mutp53 by Zn++ caused nuclear translocation of VDR in many cells without vitamin D3 (third row). Moreover, cells expressing high amounts of mutp53 exhibited more pronounced nuclear VDR staining. Furthermore, combining mutp53 induction with vitamin D3 rendered VDR staining more prominent and almost exclusively nuclear (bottom row). The ability of mutp53 to enhance the nuclear translocation and accumulation of VDR probably contributes to the augmented vitamin D3-induced transcription.
Mutp53 may be tethered to VDRE elements through a complex with VDR/RXR. To explore the existence of such a complex, co-immunoprecipitation analysis was performed on SKBR3 cells with or without treatment with 100nM vitamin D3. As a control for wtp53, a similar experiment was performed on MCF7 cells treated with Nutlin-3A to induce p53 accumulation. As seen in Fig. 3A upper panel, pull-down of VDR coprecipitated p53 from both SKBR3 and MCF7 cell extracts, indicative of a complex comprising both proteins. This interaction was significantly increased by treatment with vitamin D3. Reciprocal co-immunoprecipitation yielded a similar picture (Fig. 3A lower panel). Interestingly, the co-immunoprecipitation of mutp53 with VDR was further augmented when a wt, but not mutant, VDRE consensus oligonucleotide was added to the cell extract during incubation (Fig. S2A), indicating that the VDRE sequence can recruit both proteins simultaneously.
The p53 protein is composed of three major domains; an N-terminal TAD, a central DNA binding domain (DBD) and a C-terminal regulatory domain (CTD). VDR was overexpressed in H1299 cells together with either full length H175Rp53, ΔN-H175Rp53 (aa 97-393), ΔC-H175Rp53 (aa 1-292) or the DBD of H175Rp53 (aa 97-292). Pull-down of VDR followed by Western blot analysis for p53 revealed that full length H175Rp53 and ΔN-H175Rp53 bound VDR, whereas ΔC-H175Rp53 and DBD-H175Rp53 did not (Fig. 3B); hence, the interaction between p53 and VDR is mediated through the p53 CTD.
Next, we knocked down mutp53 expression in SKBR3 cells, in SW480 colorectal cancer cells that harbor a combination of the p53R273H and p53P309S mutations, and in MCF7 cells that express wtp53. RT-qPCR analysis of transcripts of VDR target genes revealed that vitamin D3 caused a dramatic increase in CYP24A1 mRNA and a threefold induction of IGFBP3 mRNA in both SW480 and SKBR3 cells (Fig. 3C), which was attenuated by knockdown of endogenous mutp53. Hence, mutp53 is required for an optimal transcriptional response of at least some genes to vitamin D3. In contrast, knockdown of wtp53 in MCF7 cells had no significant effect on either endogenous or vitamin D3 -induced mRNA levels of CYP24A1 and IGFBP3.
p53 can transactivate its target genes by recruiting transcriptional coactivators such as CBP/p300. Furthermore, both wt and mutp53 interact with CBP/p300 through their TADs (Avantaggiati et al., 1997; Gu et al., 1997). We therefore assessed the binding of mutp53, VDR and p300 to VDRE in the CYP24A promoter. As seen in Figure 3D, mutp53, VDR and p300 showed comparable binding (4 to 8 fold enrichment over control) under basal conditions. As expected, vitamin D3 strongly increased the binding of VDR and p300. Binding of p53R175H to the CYP24A promoter also increased significantly, to 20 fold over background. Importantly, mutp53 knockdown significantly reduced both basal and vitamin D3-induced binding of all three proteins. Hence, mutp53 augments the recruitment of both VDR and p300 to VDRE-containing chromatin. To confirm the interaction of VDR, mutp53 and p300 with the same DNA segment we performed a re-ChIP experiment. As seen in Fig. 3E, ChIP for p53 followed by re-ChIP for VDR or p300 enriched for the presence of the CYP24A VDRE, when treated with vitamin D3.
Vitamin D can also repress many genes, including CYP27B (Kim et al., 2007; Murayama et al., 2004), by a mechanism involving a negative vitamin D response element (nVDRE). As seen in Figure 3F, vitamin D3 caused a significant repression of CYP27B only in mutp53-depleted (p53i+D3) but not mutp53-proficient (LacZi+D3) SKBR3 cells, arguing that mutp53 prevents VDR-mediated transrepression. Mutp53 was not associated with the region of the CYP27B gene harboring the nVDRE (Figure 3G). Remarkably, while in cells depleted of mutp53 vitamin D3 strongly stimulated the binding of VDR, presumably to the nVDRE, this was attenuated by mutp53. Furthermore, while vitamin D3 caused a dissociation of p300 from this region in the absence of mutp53, p300 remained associated when mutp53 was present. These results imply that mutp53 interferes with the binding of VDR to nVDRE, enabling the relevant target genes to remain transcriptionally active.
Vitamin D has been reported to promote tumor cell death (Deeb et al., 2007; Duque et al., 2004). Indeed, vitamin D3 slightly augmented the death of wtp53-positive MCF7 breast cancer cells (Fig. 4A). This effect increased modestly when vitamin D3 was combined with cisplatin treatment (Cis-DDP), and was somewhat attenuated by endogenous wtp53 knockdown (p53i).
Surprisingly, vitamin D3 significantly reduced cisplatin and etoposide-induced apoptosis of SKBR3 cells, as measured by caspase activity, (Fig. 4B, Fig. S3B) or Anexin-V or PI assays (Fig. S3C, S3F). This protective effect was strictly mutp53-dependent, since it was abolished upon knockdown of endogenous p53R175H. Thus, while vitamin D often promotes apoptosis in cells lacking mutp53, mutp53 can convert vitamin D3 into a cytoprotective agent. A similar pattern was seen in an SKBR3-derived cell line expressing inducible p53 shRNA (Figure. S3B).
Mutp53-dependent antiapoptotic activity of vitamin D3 was also observed in two other breast cancer cell lines, MDA-MB-231 that harbors endogenous p53R280K (Fig. 4D and Fig. S3D) and MDA-MB-468 that expresses endogenous p53R273H (Fig. S3E), and a colorectal cancer cell line, SW480 that expresses p53R273H and p53P309S (Fig. S3F). Furthermore, vitmain D3 led to a mild but consistent increase in colony formation of H1299 cells transfected with various mutp53 isoforms but not with GFP control plasmid (Fig. S3I). However, in the breast cancer cell line BT474, which expresses p53E285K, vitamin D3 protection was not significant (data not shown). Hence, even though the observed effects are common to many mutp53-expressing tumor cells, the scope of the effect is cell context-dependent.
To compare the impact of wtp53 and mutp53 on the biological response to vitamin D in a genetically defined experimental model, we employed primary prostate epithelial cells immortalized by the introduction of hTERT and then transfected with either an empty vector (EP-156-neo cells) or a mutp53 expression plasmid (EP-156-p53R175H cells) (Kogan et al., 2006). EP-156-neo cells express endogenous wtp53 whereas EP-156-p53R175H cells overexpress p53R175H on the background of wtp53. As seen in figure 4E, treatment of EP-156-neo cells with vitamin D3, either alone or in combination with genotoxic agents, had a mild proapoptotic effect. However, mutp53 altered the response dramatically: instead of enhancing apoptosis, vitamin D3 now conferred strong protection (Fig. 4D and 4E). Hence, p53 mutations convert vitamin D3 from death-promoting into protective.
Although mutp53 exerts a general stimulatory effect on VDR transcriptional activity, this by itself cannot explain how mutp53 reverses the impact of vitamin D on apoptosis. Rather, the answer might lie in differential effects on the expression of specific genes. Therefore, expression microarray analysis was performed on SKBR3 cells without or with endogenous mutp53 knockdown, with or without vitamin D3 treatment. The expression level of each individual gene in cells transfected with p53 siRNA but not treated with 1α25(OH)2D3 (= no mutp53, no vitamin D), averaged from two biological repeats, was taken as baseline. Relative expression levels of each gene in the other biological samples were calculated as fold change over baseline. Genes were then sorted according to their fold induction by vitamin D3, mutp53, or the combination of both. When we analyzed the genes most highly induced by mutp53, (Table S3), vitamin D3 association was strongly significant (31 out of 38 genes, excluding the two p53 probes; p=4×10-18, Student's t-test). Hence, in SKBR3 cells, most mutp53-upregulated genes are also VDR targets, implicating the interaction with VDR as a major regulatory mechanism by mutp53 in these cells. Figure 5A illustrates the effects of mutp53 and vitamin D on gene expression. The X axis shows the effect of vitamin D3 alone on each gene, calculated from the ratio between expression levels with and without vitamin D3 (50h) in mutp53 knocked-down cells. The Y axis shows the combined effect of vitamin D3 and mutp53, calculated from the ratio between expression levels with vitamin D3 in control cells and without vitamin D3 in mutp53 knocked-down cells. Most genes are along the diagonal, near the center of the plot, indicating lack of regulation by either vitamin D3 or mutp53. A few genes lie further up or down on the diagonal: these are regulated by vitamin D3, but the extent of regulation is not affected by mutp53. On that background, several gene clusters exhibit distinct response patterns that diverge from the behavior of the bulk transcriptome.
Cluster 1 (blue) contains genes that are highly induced by the combination of mutp53 and vitamin D3, a behavior consistent with the luciferase experiments. Many classical VDR target genes populate this cluster. Of note, the majority of those genes are also dependent on mutp53 for optimal basal level expression, as revealed by comparison of non-treated samples without (Fig. 5B, lanes 1,2) and with (lanes 6,7) p53 siRNA. This suggests that a functional interaction between VDR and mutp53 may exist even without vitamin D3 or at low residual levels of vitamin D3 produced by SKBR3 cells.
Cluster 2 (red) comprises genes repressed 3 fold or more by long (50 hours) treatment with vitamin D3 when mutp53 was absent, but this effect was strongly attenuated in the presence of mutp53 (Fig. 5C, compare lanes 5 and 10), These results are consistent with the data obtained for the CYP27B gene (Fig. 3F,G). Hence, in addition to augmenting the transactivation of many vitamin D3-inducible genes, mutp53 also relieves vitamin D3-mediated repression of a relatively large subset of genes.
Cluster 3 (green) contains genes whose expression was downregulated more than four fold by the combination of mutp53 and vitamin D3. The basal expression level of most genes in the cluster was repressed by mutp53 even without vitamin D. Vitamin D3, however, caused a further repression in cells expressing mutp53 but not in mutp53 knocked-down cells. Hence, these genes are repressed by vitamin D3 in a mutp53-dependent manner.
The microarray analysis in Fig. 5 demonstrated that mutp53 can modulate the activity of VDR in a gene-specific manner. The impact of mutp53 on the biological response to vitamin D suggested that mutp53 might modify the transcriptional program of VDR in a manner conducive to increased resistance to apoptosis. qRT-PCR analysis of individual transcripts further supports this conjecture (Fig. S4A). Thus, mutp53 augmented the ability of vitamin D to upregulate genes reported to promote survival and neoplastic transformation (e.g. SEMA3C (Moreno-Flores et al., 2003), Wnt5A (Ripka et al., 2007) and CSF3R) or to be overexpressed in cancer (e.g. Klk6 (Klucky et al., 2007)). Remarkably, mutp53 prevented the vitamin D-dependent repression of MAP2K5 and FGFR2, survival-promoting genes overexpressed in some cancers (Acevedo et al., 2007; Winter et al., 2007). Conversely, the combination of mutp53 and vitamin D repressed the proapoptotic XAF1 (Lee et al., 2006), CYFIP2 (Jackson et al., 2007), DAPK1 (Raval et al., 2007) and TXNIP (Billiet et al., 2008) genes. Interestingly, the first three are transcriptional targets of wtp53 (Jackson et al., 2007; Lee et al., 2006; Martoriati et al., 2005). The strong repression of those genes may underpin in part the cytoprotective program installed by vitamin D in mutp53-harboring cells.
To demonstrate that the transcriptional effects of mutp53 siRNA were not due to off-target effects, we performed a rescue experiment using an siRNA-resistant p53R175H expression plasmid. As shown in Fig. S2B,S2C, restoration of mutp53 expression reversed the effect of mutp53 knockdown on specific gene expression. Furthermore, it also reversed the proapoptotic effects of the knockdown (Fig. S2D). The expression pattern of these genes in additional experimental systems showed a similar trend to that of SKBR3 (Fig. S4B, S4C and S4D), attesting to the generality of the transcriptional cross-talk between mutp53 and VDR.
The differential behaviors of the different clusters and the identities of some of the genes offer an attractive explanation for the mutp53-dependent antiapoptotic effects of vitamin D3.
The present study reveals a functional and physical interaction between mutp53 and the vitamin D transcriptional regulatory pathway. Specifically, mutp53 is tethered to chromosomal regions containing VDRE elements, presumably through association with VDR, and to augment transcription from promoters containing such elements. Moreover, in cells harboring p53 mutations, mutp53 cooperates with vitamin D3 to elicit an antiapoptotic state. This surprising effect of mutp53 is likely due to its ability to modulate, qualitatively and quantitatively, the transcriptional program orchestrated by VDR, and appears to involve augmented expression of survival genes along with reduced expression of proapoptotic genes.
Vitamin D participates diverse biological processes such as calcium homeostasis, cell proliferation, cell differentiation (see (Deeb et al., 2007) for review). VDR is believed to be constantly transported into the nucleus; however after activation by vitamin D3, this transport is greatly enhanced (Yasmin et al., 2005). Ligand binding induces conformational changes in VDR, exposing surfaces for transcriptional coactivator binding and dimerization. The dimerization partner is usually retinoid X receptor (RXR), which is required for full transactivation by VDR. Dimerization enables high affinity interaction with the VDRE. Transcriptional coactivators, physically recruited by VDR, then initiate transcription. We show that VDR and mutp53 (and also wt p53) engage in a physical interaction, which is significantly enhanced by vitamin D3 treatment. The high endogenous levels of mutp53 in tumor cells probably enable this interaction to exert significant biological effects. Moreover, mutp53 increases the nuclear accumulation of VDR. Notably, in lung tumors, increased nuclear localization of VDR correlates with higher histological grade (Menezes et al., 2008). Remarkably, some enhancement of VDR nuclear accumulation by mutp53 can be seen even in the absence of added vitamin D. Thus, the binding of mutp53 to VDR might alter the latter's conformation in a way that mimics the effect of vitamin D3.
Increased nuclear localization of VDR is probably not the sole explanation for the effect of mutp53. Based on our findings, we propose the following additional mechanisms (Fig. 6). In the case of transactivation, VDR recruits mutp53 to VDREs in target genes, and mutp53 increases VDR-dependent transcription by augmenting the recruitment of additional transcriptional coactivators, such as p300 (Fig. 6 top). Indeed, p300 binds to the p53 TAD, which is intact in all cancer-associated hotspot p53 mutants.
The mechanism behind Vitamin D3-mediated transrepression is less well characterized. A distinct negative vitamin D response element (nVDRE) was identified in the promoters of several vitamin D3-repressed genes. nVDRE binds the transcriptional activator VDIR, which in turn recruits p300 and activates transcription. VDR, activated by vitamin D3, does not directly bind the nVDRE, but instead associates with VDIR. This leads to dissociation of p300 and recruitment of HDACs to repress transcription of the target gene. In contrast to the positive effect of mutp53 on VDR-mediated transactivation, mutp53 actually reverses VDR-mediated transrepression. In fact, the majority of genes repressed by vitamin D3 in SKBR3-p53i cells were derepressed in cells expressing endogenous p53R175H (Figure 5 red cluster). As an example, VDR binds the nVDRE-containing region of the CYP27B promoter only in the absence of mutp53 and presence of vitamin D3, and this leads to dissociation of p300 (Fig. 3G) and transcriptional repression (Fig. 3F). We therefore propose that while VDR can repress transcription by interfering with the binding and activity of positive transcription factors (TF-X, Fig. 6 bottom), such as VDIR, engagement of VDR by mutp53 relieves this interference, allows recruitment of transcriptional coactivators, and restores transcription.
Vitamin D3 can induce apoptosis either alone or in conjunction with other drugs (Colston et al., 1992; Nakagawa et al., 2005). Moreover, vitamin D3 has been reported to possess anti-tumor activities (Kerner et al., 1989). VDR knock-out mice exposed to the chemical carcinogen DMBA develop more skin tumors and lymphomas than wild type mice, and nearly all VDR knock-out mice display in-situ hyperplasia of the mammary gland (Zinser et al., 2002a; Zinser et al., 2002b). The notion that vitamin D3 is capable of exerting anti-cancer effects has spurred attempts to develop vitamin D3 analogues as cancer chemotherapy agents.
However, in addition to the well documented proapoptotic effects, there are circumstances where vitamin D3 exerts antiapoptotic effects, including increased cell survival following UV trauma and protection of some cancer cell lines from killing by cytotoxic drugs. An inhibitory effect of vitamin D3 on TNFα-induced apoptosis and on TRAIL and Fas ligand-induced apoptosis, accompanied by a decrease in Bax and upregulation of Bcl2 and p21, has also been described (Duque et al., 2004). However, in SKBR3 and MDA-MB-231 cells p21 and Bcl2 protein levels did not change significantly following either mutp53 knockdown or vitamin D3 treatment (Fig. S3J). Interestingly, VDR is upregulated in several types of cancer, including breast and ovarian carcinomas (Friedrich et al., 1998; Friedrich et al., 2002). Moreover, elevated VDR was reported to correlate with tumor stage (Menezes et al., 2008; Sahin et al., 2005). Thus, the VDR pathway can lead to either death or survival, depending on the cellular context.
To date, no satisfactory mechanism was proposed to account for the seemingly conflicting effects of VDR/vitamin D3. Our findings now provide a possible explanation. Perhaps the most significant finding is that mutp53 can convert vitamin D3 from a proapoptotic agent into an antiapoptotic one. In support of this notion, vitamin D3 suppresses death receptor-mediated apoptosis in OVCAR3 ovarian carcinoma cells (Zhang et al., 2005b), which harbor endogenous mutp53. Conceivably, some cancers might have evolved a mechanism that allows them to capitalize more efficiently on the survival route of VDR while evading its proapoptotic effects. Our data suggests that acquisition of gain-of-function p53 mutations may constitute one such mechanism. Obviously, p53 mutations alone are not sufficient to drive the conversion of the VDR pathway into an antiapoptotic one, as not all mutp53-expressing tumor-derived cell lines are equally protected by vitamin D3. Thus, additional alterations most likely cooperate with mutp53 to orchestrate the antiapoptotic response to vitamin D3. Identification of such cooperating factors remains an important challenge.
Vitamin D and its derivatives are being extensively explored as cancer-preventive and even cancer-therapeutic agents. Our findings call for extra caution in exploring this approach. Conversely, individualized cancer therapy might be implemented also in the context of vitamin D3 treatment, where the p53 mutation status of the tumor may serve as an important aid for outcome prediction.
H1299 human non-small cell lung cancer cells and SKBR3 breast cancer cells were maintained in RPMI-1640 medium supplemented with 10% fetal calf serum (FCS) and antibiotics. Where indicated, charcoal stripped serum was included instead of regular serum. MCF7 human breast cancer cells stably expressing short hairpin RNA (shRNA) targeting p53, as well as vector control cells (gift of R. Agami, Netherlands Cancer Institute) were grown in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% FCS, 2 mM L-glutamine, and antibiotics. The MDA-MB-231 breast cancer cell line and the SW480 colorectal cancer cell line were maintained in DMEM supplemented with 10% FCS and antibiotics. EP156 epithelial cells were grown as described (Kogan et al., 2006). SKBR3-D8 cells were grown under the same conditions as parental SKBR3, plus puromycin (1μg/ml) and blasticidin (10μg/ml). All cell lines were grown at 37°C in a humidified atmosphere of 5% CO2 in air.
Expression plasmids for wild-type human p53 and mutant p53R175H were gifts from C.C. Harris (National Cancer Institute, Bethesda, MD). Reporter plasmids were constructed by inserting a 64bp DNA fragment containing either 3 tandem VDRE sequences (AGGTCAnnnAGGTCA) separated by 5 bp spacers or a derivative that contained two point mutations in each half VDRE site (mVDRE: ATATCAnnnATATCA) upstream of the luciferase reporter gene within the PGL3-promoter plasmid (Promega).
Cells were seeded in 24-well culture dishes. Each well was transfected with VDRE-luciferase or mVDRE-luciferase, together with increasing amounts of various p53 expression plasmids and ß-galactosidase (ß-gal) plasmid. Luciferase activity was assayed 48 hours post transfection. Each plasmid combination was transfected into three identical wells. Luciferase assays were performed using (D)-luciferin (Roche). Luminescence was determined with the aid of a Rosys-Anthos Lucy 3 luminometer. Luciferase values were normalized to ß-gal activity.
Immunofluorescence microscopy was performed as described (Kalo et al., 2007) Incubations with antibodies for VDR and/or p53 were carried out for 45 min at 22°C, with three extensive washes after each step. Cells were mounted with Prolong Antifade (Invitrogen-Molecular Probes). Fluorescence digital images were recorded on a Zeiss Axio Imager microscope (Carl Zeiss, Goettingen, Germany).
Co-immunoprecipitation was performed as described (Kalo et al., 2007). Cells were treated with 100nM 1α25(OH)2D3 for 3h. Cells were then scraped into ice-cold PBS and lysed with lysis buffer. Monoclonal anti-p53 antibody PAb240 or control anti-SV40 Large T antigen Pab419 antibody were incubated with 30μl protein A for 1h at room temperature and added to the lysate. Immune complexes were precipitated overnight at 4°C. The immunoprecipitated material was washed, pellets were resuspended in SDS sample buffer and subjected to Western blot analysis.
Chromatin immunoprecipitation was performed as described (Stambolsky et al., 2006) employing protein A beads cross-linked by DMP to anti p53 polyclonal antibodies, anti-HA antibodies or anti VDR antibodies (C20 Santa Cruz Biotechnology).
DNA samples were extracted using PCR clean-up mini-columns (Promega). Real-time PCR was performed using Sybr Green as described above.
Samples of immunoprecipitated DNA and 0.02% of the input DNA (calibrated to be the equivalent of background binding) were amplified by linker-mediated-PCR and subjected to ChIP-on-chip analysis as described (Odom et al., 2004), using 1μg of polyclonal anti p53 antibody (H47, homemade) and the Hu13K human promoter array. A whole-chip error model (Simon et al., 2001) was used to calculate confidence values for each spot.
Search for enriched transcription factor binding motifs in microarray data employed the MathInspector library of 414 PSSMs maintained by Genomatix (Release 4.1) (Quandt et al., 1995) and a customary promoter to PSSM assignment score (Elkon et al., 2003). A threshold on this score was defined, above which a PSSM was considered assigned to a promoter. For this purpose, the promoters of co-regulated genes were used for a range of potential values of the calculated threshold score, employing the hypergeometric statistics, the group's specificity score (Hughes et al., 2000) of the motif relative to the genes in the cluster. A threshold score that minimizes the hypergeometric probability function was then identified and adopted.
To monitor apoptosis, cultures were subjected to a caspase activity assay (Promega). To that end, cells were seeded in 96 well plates. The next day, either 1α25(OH)2D3 or DMSO was added. After an additional 24h cells were subjected to different treatments (e.g. anti-cancer drugs) for various times and harvested by adding lysis buffer to the cells, and fluorescence was monitored several hours later. Values were normalized either for cell number or for readings of a WST1 kit assay (Roche), as indicated in the corresponding figure legends. For cell viability assessments, the trypan blue dye exclusion method was used, counting a minimum of 150 cells/assay and expressing data as percentage of dead (dye including) cells.
We wish to thank N. Goldfinger for assistance, and R. Agami and C.C. Harris for plasmids. Supported by EC FP6 funding (Contract 502983), Center of Excellence grant from the Flight Attendant Medical Research Institute, grant R37 CA40099 from the National Cancer Institute, the Dr. Miriam and Sheldon Adelson Medical Research Foundation, and the Yad Abraham Center for Cancer Diagnosis and Therapy. This publication reflects the authors’ views and not necessarily those of the European Community.
Mutation of the p53 tumor suppressor gene is the most common genetic alteration in human cancer. Once mutated, p53 not only loses its tumor suppressor activity but gains oncogenic functions. Indeed p53 is being examined as an important target for cancer therapy. Likewise, Vitamin D and its analogs are being evaluated as potential anti-cancer agents. Our findings provide a mechanism for mutp53 GOF, based on the interaction between p53 and VDR. The results we obtained may have clinical implications and suggest that p53 status should be considered when contemplating vitamin D analogs for cancer therapy.