Cancer progression is associated with epigenetic alterations, such as changes in DNA methylation, histone modifications or variants incorporation. The p400 ATPase, which can incorporate the H2A.Z variant, and the Tip60 histone acetyltransferase are interacting chromatin-modifying proteins crucial for the control of cell proliferation. We demonstrate here that Tip60 acts as a tumor suppressor in colon, since mice heterozygous for Tip60 are more susceptible to chemically induced preneoplastic lesions and adenomas. Strikingly, heterozygosity for p400 reverses the Tip60-dependent formation of preneoplastic lesions, uncovering for the first time pro-oncogenic functions for p400. By genome-wide analysis and using a specific inhibitor in vivo, we demonstrated that these effects are dependent on Wnt signaling which is antagonistically impacted by p400 and Tip60: p400 directly favors the expression of a subset of Wnt-target genes and regulators, whereas Tip60 prevents β-catenin acetylation and activation. Taken together, our data underline the physiopathological importance of interplays between chromatin-modifying enzymes in the control of cancer-related signaling pathways.
Cancer progression is accompanied by alterations that allow the activation of pro-oncogenic pathways and inactivate anti-cancer barriers. These alterations can be genetic, such as point mutations, allele amplification or loss, translocations, but also epigenetics. These epigenetic modifications include DNA methylation on specific promoters/enhancers (1), which is known for long to induce the transcriptional silencing of tumor suppressors encoding genes, and also modifications of chromatin structure such as histone posttranslational modifications or histone variants incorporation (for review, see (2–4)). Consequently, chromatin-modifying enzymes that set up these modifications are frequently mutated or aberrantly expressed in cancer (for review, see (4)).
The histone acetyltransferase Tip60 has been shown to act as a haplo-insufficient tumor suppressor, since heterozygous Tip60+/− mice are more sensitive to lymphoma genesis upon c-Myc activation (5). Tip60 is required for the efficient activation of the DNA damage response (DDR) pathways (5,6) and its roles in ATM (7,8) and p53 (9–11) acetylation have been described as crucial events for DDR signaling and anti-proliferative cell fate, respectively.
Tip60 belongs to a multimolecular complex containing several other chromatin modifying enzymes, such as the p400 ATPase, which mediates the incorporation of the histone variant H2A.Z (12). We and others (6,13,14) have shown that Tip60 and p400 can have antagonistic functions in some situations (14,15). This relationship has a critical impact on the proliferation of colon tumor cell lines in culture (6) and the p400/Tip60 ratio is systematically altered in human colorectal tumors, in favor of p400, independently of the grade and the stage of the tumor. This imbalance greatly alters the sensitivity of cancer cell lines to apoptosis in vitro.
Colorectal carcinogenesis is a complex multistep process that includes changes in histomorphological appearance of the colonic mucosa and deregulations at molecular level. Some critical mutations impacting key regulatory pathways have been identified in the tumorigenesis of both inherited and sporadic colorectal cancers (for review, see (16)). APC, KRAS and p53 genes are the major mutated oncogenes or tumor-suppressor genes, impairing the function of Wnt/β-catenin, tyrosine kinase receptors and DDR signaling pathways, respectively. In addition, several other mutations have also been uncovered that affect important mechanisms such as DNA repair (MLH1, MSH2, etc.), TGF-β response (DCC, SMADs, etc.) or apoptosis (PTEN, BAX, etc.) ((16) and references therein).
In regard to their histopathological features as well as their biological, histochemical, genetic and epigenetic deregulations, Aberrant Crypt Foci (ACF) are now considered as the earliest preneoplastic lesions that can be seen in the colonic mucosa (for review, see (17,18)). ACF were first described on examination of methylene-blue-stained colonic mucosa of azoxymethane (AOM)-treated mice under light microscopy. The carcinogen AOM has been extensively used to study colonic carcinogenesis and chemoprevention (19,20). The AOM-induced colon tumors recapitulate many features associated with the progression of human sporadic colorectal cancers (21,22). Indeed, AOM initiates mutations in crucial genes controlling cell proliferation, such as KRAS or CTNNB, thereby activating colon cancer-related pathways (23). Finally, long-term feeding of rats with various fat diets modulates the AOM-induced colon carcinogenesis through Wnt/β-catenin signaling. Thus, such in vivo model is of functional relevance to study early steps of colon tumorigenesis.
The activation of Wnt pathway is not sufficient to induce tumorigenesis, if barriers against malignant transformation remain active (24). Indeed, upon repression of Casein kinase α, a critical partner of the β-catenin-destruction complex, a massive activation of Wnt signaling is observed without causing malignant transformation since the p53 pathway is able to counteract the pro-tumorigenic effects of Wnt activation by increasing DDR (24) and thus, promoting subsequent anti-proliferative effects (25). Considering the crucial role of Tip60 in the p53 pathway and in the DDR activation (7, 10), we intended here to characterize the impact of the Tip60/p400 imbalance on the Wnt-involving colon tumorigenesis. Thus, our study brings important new insights to the continuously growing field of epigenetics and cancer (2).
P400 and Tip60 antagonistically control colon cancer progression
We previously showed that the p400/Tip60 ratio is modified in colon cancer (6), antagonistically impacting the viability of colon cancer cells, at least in culture conditions. To test whether such deregulations could be causal for cancer progression, we intended to characterize the impact of the Tip60/p400 imbalance on early steps of colon tumorigenesis in vivo. To that goal, we used mutant mice models harboring Tip60 (5, 26) or p400 (27) heterozygosities (the homozygous genotype being lethal in both cases) and we crossed them to generate double heterozygous mice (Supplementary Material, Fig. S1). Importantly, we did not observe any Tip60−/− or p400Δ/Δ homozygotes indicating that, although Tip60 and p400 have antagonistic functions, the depletion of one of them does not rescue the lethality observed upon inactivation of the other one.
We first analyzed the susceptibility of heterozygous mice to the formation of ACF, which are well-known preneoplastic lesions (18); we tested the appearance of such lesions upon administration of AOM (28), a compound commonly used to evaluate the susceptibility to preneoplastic lesions formation and early colon tumorigenesis (20, 29). In the Tip60+/− and/or p400+/Δ mice strains, no spontaneous tumor-associated phenotype is observed, neither in the colon or elsewhere. Moreover, no ACF is found in young (6 weeks) or old (18 months) mice, indicating that heterozygosity for Tip60 and/or p400 does not induce spontaneous hyperplasia. However, after carcinogen injection, Tip60+/− mice harbor a significantly higher number of ACF than wild-type ones (Fig. 1A). This indicates that the reduction of the Tip60 expression induces an increase of susceptibility to AOM-induced ACF formation.
In order to characterize the impact of the Tip60 heterozygosity at later steps of tumorigenesis, we analyzed the formation of adenomas after AOM injection in the presence of a proinflammatory treatment (Fig. 1B). As for early preneoplastic lesions, we observed a significant increase in the number of adenoma in Tip60+/− mice, as compared with wild-type ones (Mann–Whitney P-value < 0.02). These results demonstrate that the increased ACF formation upon defective Tip60 function translates into increased adenoma occurrence, indicating that Tip60 is a tumor suppressor in colon, as already shown for MYC-induced lymphoma (5).
We next assayed the sensitivity of p400+/Δ and double heterozygous mice to AOM-induced ACF formation. We found that p400+/Δ mice exhibit a susceptibility to AOM comparable with control mice (Fig. 1A). However, the disruption of one p400 allele almost abolishes the increase in ACF number observed in Tip60+/− mice (compare Tip60+/− with Tip60+/− and p400+/Δ). This observation indicates that p400 and Tip60 exert antagonistic effects on early stages of colon tumorigenesis in vivo. Interestingly, this also shows that p400 favors hyperplasia, at least in a Tip60+/− background. To our knowledge, these data are the first demonstration of in vivo oncogenic activities of p400. We failed to test adenoma formation for p400+/Δ mice, because these mice are more prone to die following proinflammatory treatment, probably due to the previously published (25) higher sensitivity to stresses upon p400 reduction.
Tip60 and p400 antagonistically control colon cancer initiation through the Wnt pathway
To gain insights into the mechanisms by which p400 and Tip60 antagonistically control colon cancer initiation in vivo, we analyzed their roles on critical pathways involved in cancer progression. We performed transcriptomic experiments in colon carcinoma HCT116 cells by transfecting cells in duplicate using two-independent control, Tip60 or p400 siRNAs (leading to four control, four p400-depleted and four Tip60-depleted samples). The efficiency of siRNA-mediated knockdown is shown in Supplementary Material, Figure S2A. Total RNAs were then prepared and hybridized to microarrays containing most human genes (Fig. 2A). We observed that the two p400 siRNA have very similar knockdown efficiencies (see Supplementary Material, Fig. S2A and B). We therefore considered a RefSeq as regulated by p400 when its expression is significantly deregulated in the four p400-depleted samples compared with the four control samples. Such an analysis leads to a list of 2718 p400-regulated RefSeq (Supplementary Material, Table S1). In contrast, the efficiencies of the two siRNAs against Tip60 are very different (see Supplementary Material, Fig. S2A and B), precluding such an analysis. For each RefSeq, we thus compared the values of the two Tip60 siRNAs samples relative to the two control samples, leading to 4-fold change values. We considered a RefSeq as deregulated upon Tip60 knockdown when three out of four of these values are >1.25 (or <0.8 for repressed genes). Such an analysis led to the identification of 2412 Tip60-regulated RefSeqs (Supplementary Material, Table S2). Importantly, microarrays results were validated on selected genes for both Tip60 and p400 (Supplementary Material, Fig. S2C). In addition, 731 RefSeq are regulated by both p400 and Tip60, which is a highly significant overlap (expected 146, χ2 independence test P-value < 10−100), in agreement to the fact that they are present within the same complex and that many genes are known to be regulated, either similarly or antagonistically, by both Tip60 and p400. Strikingly, a gene ontology analysis (Fig. 2A) indicated that genes linked to the Wnt pathway are significantly deregulated upon Tip60 (P-value = 0.0431686) and p400 depletion (P-value = 0.00078871).
Given the critical role of this pathway in colon cancer initiation and progression, we next tested whether it participates in the consequences of p400 and Tip60 depletion on colon cancer initiation. To that goal, we used a recently developed inhibitor of Wnt signaling, the C59 compound, which exhibit a good in vivo bioavailability and thus can be used in mice (30). This inhibitor targets the membrane-bound O-acyltransferase porcupine (PORCN), an important Wnt-positive regulator (31). We induced ACF formation using AOM in wild-type and Tip60−/+ mice, and treated them, or not, with C59 (Fig. 2B). This inhibitor does not significantly affect the number of ACF observed in wild-type mice, in agreement with the fact that AOM mainly targets other pathways, such as KRAS-dependent ones. However, treatment with this inhibitor completely blocks the increase in ACF formation observed in Tip60 heterozygous mice, indicating that Tip60 is a tumor suppressor in colon by repressing Wnt signaling. In addition, the inactivation of one p400 allele (see Fig. 1A) mimics the inhibition of the Wnt pathway, since it affects ACF formation only in the Tip60 heterozygous context, indicating that p400 participates in ACF formation by favoring Wnt signaling.
Tip60 and p400 control the proliferation of normal and colorectal cancer-derived cells through the Wnt pathway
In order to provide insights into the molecular mechanisms involved, we next investigated whether the effects of the Tip60/p400 balance on the Wnt signaling can be observed in cultured cells. We found that, in tumoral HCT116 cells, Tip60 down-regulation using a specific siRNA increased the transactivating abilities of the Wnt-dependent Tcf/Lef transcription factors (measured using a luciferase-based reporter vector), whereas p400 knockdown decreased it (Fig. 3A, see Supplementary Material, Fig. S3A for siRNA efficiencies). Similar results were obtained upon activation of the Wnt pathway by the Wnt3a ligand. Moreover, they were also confirmed using an independent siRNA, ruling out off-target effects (Supplementary Material, Fig. S3B and C). Thus, in colon cancer cells, Tip60 represses the Wnt pathway, whereas p400 expression favors it, even upon overactivation by exogenous ligand. Strikingly, close examination of our microarrays results confirmed this finding for endogenous genes: we observed that Wnt-target genes affected by Tip60 knockdown are globally activated (mean fold change = 1.17), whereas Wnt-target genes, whose expression changes upon p400 knockdown, are mainly repressed (mean fold change = 0.91).
To study whether such changes in Wnt pathway activity can translate into changes in cell proliferation, we transfected HCT116 cells with siRNAs and assayed their proliferation abilities (Fig. 3B). No effect of Tip60 knockdown could be observed (left graph), as expected since these cells already have a largely defective Tip60 function, due to the change in the Tip60/p400 ratio (6). However, p400 knockdown significantly slows the proliferation rate. This is due to changes in the Tip60/p400 ratio, since it is reversed by the concomitant knockdown of Tip60. P400 knockdown is known in other cell types to lead to a cell proliferation arrest by inducing the expression of the cell cycle inhibitor p21 (13,14). However, in HCT116 cells, we do not observe any change in p21 mRNA expression (data not shown), consistent with our previously published data in HCT116 (6) showing that effects of Tip60/p400 ratio on cell growth are mediated through p53-independent pathway. Rather, we observed that these effects are mediated through the Wnt pathway, since the anti-proliferative effects of p400 knockdown were no longer observed in the presence of the Wnt3a ligand (Fig. 3B, right), which restores normal Wnt pathway activity in p400-depleted cells (see Fig. 3A). Thus, p400 promotes colon cancer cells proliferation by favoring the activity of the Wnt pathway.
Since the Wnt pathway and cell proliferation are strongly affected by cell transformation, and given that the consequences of Tip60 or p400 depletion in vivo are observed at early stages of colon cancer progression, we also analyzed the effects of Tip60 and p400 on the Wnt pathway activity and proliferation in non-transformed cells. Using a siRNA-based strategy, we found that Tip60 and p400 antagonistically control Tcf/Lef transcriptional activity in MEFs (Fig. 4A, left; see Supplementary Material, Fig. S4A for siRNA efficiencies), both in presence or in absence of the Wnt3a ligand, as already observed in HCT116 human cancer cells (see Fig. 3A).
To investigate the impact of such Wnt signaling regulation on the proliferation of normal cells, we generated MEFs from mice heterozygous for Tip60 and/or p400. We found, as expected, a 2-fold reduction in Tip60 and wild-type p400 (Supplementary Material, Fig. S1B and C). Importantly, p400 heterozygosity has no impact on Tip60 expression, and reciprocally. We observed that Tip60+/− MEFs grow faster than the wild-type ones (Fig. 4B), providing evidence that a two-fold decrease in Tip60 expression is sufficient to increase the growth rate of these normal cells. Strikingly, p400 heterozygosity completely reverses the increase of proliferation observed in Tip60+/− MEFs (compare Tip60+/− with Tip60+/− and p400+/Δ), underlining the importance of a correct p400/Tip60 ratio for the appropriate control of normal cell proliferation. To test whether these changes in normal cell proliferation are dependent on the Wnt pathway, we analyzed the impact of the Wnt3a ligand and the FH535 agent, an inhibitor blocking the β-catenin/Tcf transactivation (32,33), on the proliferation of wild-type or Tip60+/− MEFs (Fig. 4C). We found that Wnt3a stimulates more efficiently the proliferation of normal cells in a Tip60-reduced context. Importantly, this facilitating effect of the Tip60 heterozygosity is almost totally abolished by inhibiting the β-catenin/Tcf activity using FH535. Taken together, Figures 3 and 4 results indicate that Tip60 and p400 antagonistically controls the proliferation of normal and colon cancer-derived cells through the Wnt pathway, in perfect agreement with their Wnt-dependent role in colon cancer initiation in vivo.
Tip60 counteracts β-catenin acetylation and accumulation
We next intended to characterize the mechanism by which p400 and Tip60 participate in Wnt signaling. First, we analyzed the expression of key players of this pathway. We found that depletion of Tip60, either in siRNA-treated or heterozygous MEFs, consistently increases β-catenin protein expression (Fig. 5A and B and Supplementary Material, Fig. S4B). Interestingly, β-catenin mRNA levels are only weakly, if anything, affected by Tip60 depletion, indicating that Tip60 mainly acts at a posttranscriptional level. Similar increase of β-catenin amount is also observed in human colon carcinoma HCT116 cells (Fig. 5D, Inputs panels) or in the normal Human Intestinal Crypt Cells HIEC cells (34) (Fig. 5E, inputs panels; Supplementary Material, Fig. S4C for silencing efficiency). Thus, Tip60 negatively controls β-catenin accumulation.
Since Tip60 is an acetyltransferase, and given that acetylation of a protein can modulate its stability, we tested whether β-catenin can be a substrate of Tip60. Surprisingly, Tip60 depletion increases β-catenin acetylation in MEFs (using two-independent Tip60 siRNAs, Fig. 5C), but also in human tumoral HCT116 (Fig. 5D), or normal HIEC (Fig. 5E) cells, indicating that Tip60 negatively regulates β-catenin acetylation. Acetylation of β-catenin by CBP/p300 or PCAF is known to lead to its stabilization and to the stimulation of its transactivating abilities (35–37). These findings led us to test whether Tip60 could prevent Wnt signaling by inhibiting β-catenin acetylation. To that goal, we made use of unacetylable K345A/R mutants of β-catenin (35). Both mutants activate the Tcf/Lef reporter promoter as efficiently as the wild-type β-catenin (Fig. 5F). However, they are not able to sustain the extra-activation observed upon Tip60 knockdown, in a striking contrast to the wild type. Thus, we conclude from Figure 5 experiments that the major mechanism by which Tip60 negatively regulates the Wnt pathway is by counteracting β-catenin acetylation and consequently its accumulation and transactivating abilities. Since the mutated site is a known target for CBP/p300, this suggests that Tip60 regulates Wnt activity by antagonizing CBP/p300 activating role on β-catenin.
P400 favors the expression of a subset of Wnt-target genes
In Figure 5 experiments, depleting p400 has no significant effect on β-catenin accumulation or acetylation, indicating that p400 modulates Wnt pathway activity independently of controlling β-catenin acetylation. Interestingly, p400 depletion in MEFs affects the expression of some of the Wnt/β-catenin-target genes, such as Itf2 (Supplementary Material, Fig. S5A) that encodes a transcription factor promoting the neoplastic transformation (38) and colon cancers (39). We also observed, by ChIP analysis, that the Itf2 promoter is enriched in p400 immunoprecipitates (Supplementary Material, Fig. S5B). This enrichment is specific, since no signal can be detected in the no antibody control and since the Itf2 promoter is more present in the p400 ChIP than the control RPLP0 gene. Moreover, the binding of p400 to the Itf2 promoter is similar to that observed on the CDKN1Ap21 promoter, to which p400 binding is well established (13). Thus, the Wnt-target Itf2 gene is directly regulated by p400. Interestingly, treating MEF cells with Wnt3a (which activates Itf2 transcription, Supplementary Material, Fig. S5C) leads to a drastic increase in the p400 ChIP efficiency of the Itf2 promoter (Supplementary Material, Fig. S5B), indicating that the binding of p400 to the Itf2 promoter increases upon the activation of the Wnt pathway. These data thus suggest that p400 participates in the Wnt-dependent transcriptional activation, at least on some Wnt-target genes.
To test whether such a finding can be generalized, we intended to globally characterize genes bound by p400 in colon-derived cells (Fig. 6A). Two-independent p400 ChIP from HCT116 cells were hybridized to Human ChIP-chip Promoter Arrays. Data analysis lead to the identification of 8809 RefSeqs (Supplementary Material, Table S3), corresponding to 3010 p400-bound genes. As expected, the CDKN1Ap21 promoter is identified in this analysis, and we validated other p400 target genes by conventional ChIP qPCR (Supplementary Material, Fig. S5D).
The overlapping between p400-bound and p400-regulated genes is not very good (231 out of 2150 (1/9) p400-regulated genes are bound by p400), nor with Tip60-regulated genes (128 out of 2002 (1/15) Tip60-regulated genes), underlining the importance of indirect effects in genes regulation by p400 and Tip60. Strikingly, however, we found, by gene ontology analysis (Fig. 6A), that genes from the Wnt pathway are highly significantly enriched among p400-bound promoters (P-value = 2.88903e−06). In addition, when we compared with β-catenin-target genes, previously identified by ChIP-Seq experiments (40), we found that 158 out of 988 (1/6) β-catenin-target genes are also bound by p400, indicating that p400 binds to a subset of Wnt-target genes. Importantly, these 158 genes are mostly repressed upon p400 knockdown (mean of the log 2 (fold change) = −0.06, P-value = 0.001316), indicating that, on these genes, p400 has mainly an activating role. Thus, this genome-wide experiment confirmed that p400 binds to and participate in the activation of a subset of Wnt-target genes.
In addition, some genes encoding key Wnt pathway modulators were apparently bound by p400 (see Supplementary Material, Table S3), such as the Wnt-positive regulator PORCN, which is the target of the inhibitor we used in vivo. Careful analysis, by conventional ChIP followed by qPCR, of genes encoding several important Wnt modulators (Fig. 6B) indeed confirmed that genes encoding PORCN and another positive regulator, FZD2, as well as the gene encoding CAMKIID, an indirect repressor of the canonical Wnt pathway (41), are bound by p400. The binding of p400 on these genes is similar to that observed on SART3 and CDKN1Ap21 (positive genes in ChIP-chips experiments) and significantly higher than on negative genes RPLP0 and PTGER3. Interestingly, the knockdown of p400 using specific siRNAs decreased the expression of FZD2 and PORCN mRNA, and increased the expression of CAMKIID mRNA (Fig. 6C). All these effects result in a decreased activation of the Wnt pathway, probably explaining the effect of p400 depletion on Tcf/Lef luciferase reporter gene (Fig. 3A). Thus, taken together, these experiments indicate that p400 directly participates in the activation of a subset of β-catenin-target genes, and favors Wnt signaling by controlling the expression of key modulators of this pathway.
Here, we found that, by modifying the expression of Tip60 and p400, one can modulate the sensitivity to colon cancer initiation and progression, and that this function is linked to the antagonistic regulation of Wnt pathway activity by the two enzymes (Fig. 7). The role of Tip60 as a negative regulator of Wnt signaling, that we show here, was suspected (42), but never demonstrated. We show here that it is largely, if not only, dependent on its ability to counteract β-catenin acetylation and accumulation. This is an important and surprising finding, since it shows that a HAT (Tip60) can act antagonistically to other HATs (CBP/p300, PCAF, etc.), which acetylate β-catenin, leading to its stabilization and activation (35–37). Such an interference mechanism between HATs had not previously been shown, to our knowledge. Whether it is restricted to the Wnt pathway or can be generalized is an issue deserving further consideration.
In contrast, p400 acts as a positive regulator of the Wnt pathway. This is likely to be mediated by a different mechanism, since no effect of p400 on β-catenin accumulation could be found. Our ChIP-chips data allow us to propose that p400 binds to and favors the expression of a subset of Wnt-target genes in human cells (such as Itf2 in MEFs), as it has been shown for Tip49a (43). Although some genes bound by p400 are not (or only weakly) affected by p400 depletion in HCT116 cells (such as MYC for example), these genes may be very important in other cellular contexts. For example, CDKN1Ap21 expression is not affected by p400 depletion in HCT116 cells (microarrays results in Supplementary Material, Table S1), although it is very important to mediate the cell cycle arrest and senescence induction upon p400 depletion in other cell types (14,44). The direct control of Wnt-target genes is probably not the only mechanism by which p400 participates in the Wnt pathway (Fig. 7): indeed, we also observed that p400 negatively controls the expression of genes involved in the repression of the canonical Wnt/β-catenin pathway (such as Wif1 and sFrp1 in MEFs, or DKK1 and CAMKIID in human cells, Supplementary Material, Fig. S6 and Fig. 6C). Moreover, p400 is also responsible for direct-positive regulations of important Wnt pathway components such as FZD2 and PORCN (Fig. 6B and C). All these data indicate that the transcriptional modulations depending on p400 affect different levels of the Wnt pathway and participate to its activation (Fig. 7). The importance of such direct regulations is indicated by our findings that inhibiting p400 mimics the effects of targetting PORCN for ACF formation in vivo. Our data thus uncover p400 as an important new player in the Wnt signaling.
Our results also indicate that, although Tip60 and p400 physically interact, they probably affect Wnt signaling by independent mechanisms. This is reminiscent to what has been described on the CDKN1Ap21 promoter that p400 represses by incorporating H2A.Z and that Tip60 activates, once the p400-induced repression has been relieved, through histone acetylation (13). It is tempting to speculate that p400 and Tip60 have acquired, during evolution, antagonistic functions on many different processes, some of which being mediated by their physical interaction and the direct repressive effect of p400 on Tip60 histone acetyltransferase activity (15), others through independent mechanisms. Strikingly, in most cases (regulation of p21 expression (14), senescence induction (44, 45), DDR activation (6) and Wnt signaling (this study)), p400 has a proproliferation role, whereas Tip60 has an anti-proliferative role.
Importantly, our data indicate that the status of Tip60 and of p400 are crucial in vivo elements controlling the sensitivity to preneoplastic lesions in a Wnt-dependent manner. Our results allow us to propose that the Tip60/p400 ratio exerts a fine-tuning control on the Wnt pathway activity (Fig. 7), impacting early steps of colon tumorigenesis and cancer cells proliferation. We observed changes in colon cancer initiation through the modulation of the Wnt pathway in heterozygous animals, i.e. when Tip60 or p400 expression levels are modified by 2-fold. Strikingly, we previously uncovered an approximate 2-fold difference in the Tip60/p400 ratio in a collection of colon cancers as compared with normal adjacent tissues (6). Our results thus show that the imbalance found in human cancers is probably sufficient to affect Wnt signaling and, as a consequence, could participate in colon cancer progression in humans.
Tip60 has already been shown to be a tumor suppressor for myc-induced lymphoma. In this case, depleting Tip60 was shown to prevent the activation of the oncogene-induced DNA damage response (OIDDR) pathway (5). Our data thus demonstrate that Tip60 can also prevent cancer progression by an alternate mechanism, i.e. by repressing Wnt signaling. Interestingly, Tip60 is involved in many other anti-proliferative processes, such as p53 activation (46,47), apoptosis (12,14) or senescence induction (44). Since Tip60 is known to be underexpressed in a variety of human cancers (5,48), the consequence of this underexpression is probably both an activation of the procancer Wnt pathway and an inhibition of various anti-cancer barriers, including the p53 pathway, the OIDDR and senescence. Our data thus underline the importance of the Tip60 function and of the Tip60/p400 imbalance for cancer initiation and progression.
Strikingly, we show for the first time that p400 possesses some oncogenic properties, at least in a Tip60-reduced context, i.e. when cancer progression is dependent on the Wnt pathway. Given the importance of the Wnt pathway in many different cancers, p400 expression and activity is probably important for the progression not only of colon cancer, but also of other cancers. Importantly, since p400 directly regulates a subset of Wnt-target genes and some genes encoding key Wnt modulators, it probably participates in setting up epigenetic modifications, such as the presence of H2AZ, on these promoters. In this respect, our data highlight the importance of the epigenetic control of Wnt signaling. Given that epigenetic modifications are reversible, one can imagine that epigenetic drugs, in particular drugs that target p400, may reverse these epigenetic modifications and decrease the activation of Wnt signaling. Our results thus underline the importance of developing and testing epigenetic drugs targeting this new Wnt pathway modulator.
In addition, our findings showing that Tip60/p400 balance affects the proliferation of normal cells in a Wnt pathway-dependent manner provide evidence that the Tip60/p400 status can also influence the cell physiology prior to tumorigenesis. Thus, any alteration of the ratio could impact critical physiological roles of the Wnt/β-catenin pathway, such as the transcriptional control of pluripotency genes, the self-renewal of stem cells and many differentiation processes (49). It could also be involved in other important pathologies involving the Wnt signaling (diabetes, osteoporosis, obesity, coronary diseases, etc., for review, see (50)). The contribution of Tip60 and p400 in normal development or pathologies, through Wnt signaling, clearly deserves further investigation.
MATERIALS AND METHODS
All microarrays data from this publication have been submitted to the GEO public functional genomics data repository (GSE45616). Statistical significance was calculated using Student's t-test, χ2 independence test or Mann–Whitney test. For protocol complements, see Supplementary Material, Experimental Procedures.
The experiments involved with animals were conducted accordingly to French governmental norms and the protocol was approved by the IPBS-CNRS Ethic Committee (Approval #20090326/62).
Mice heterozygous for Tip60 (26) were obtained from Dr B. Amati and back-crossed with C57Bl/6J females (Charles-River, L'Arbresle, France). Frozen embryos of the p400 strain (27) were purchased from RIKEN-BRC (Ibaraki, Japan) and reviviscence was done using C57Bl/6J mice.
Induction of preneoplastic lesions
Six- to 8-week-old animals received, twice at 1 week apart, an intraperitoneal injection of AOM (Sigma-Aldrich, St-Quentin-Fallavier, France) in saline solution at 15 mg/kg of body weight or vehicle alone. Mice were then treated or not using 5 mg/kg per day of the Wnt inhibitor C59 (BioVision Inc, Milpitas, CA, USA) in the drinking water immediately after the first AOM injection. Three weeks after the second injection, animals were sacrificed and colons were slit opened longitudinally and examined for ACF, after fixation in formalin and staining with 0.05% methylene-blue solution (Sigma-Aldrich).
To accelerate the appearance of advanced lesions, mice were also treated with 1% DSS (MP Biomedicals, Illkirch, France), a proinflammatory molecule, for 1 week in the drinking water immediately after AOM injection. Ten weeks later, colons were dissected, fixed in formalin and photographed.
Analysis of microarrays results
Analysis of large-scale gene expression in HCT116 cells previously transfected with siRNAs (see Supplementary Material, Fig. S2A for efficiencies and Experimental Procedures for details) were done using Nimblegen (Roche Nimblegen, Madison, WI, USA) DNA microarrays (Human Gene Expression 12x135K Arrays).
Chromatin immunoprecipitation and ChIP-chips experiments
ChIP experiments were performed as described (14) and analyzed by qPCR using specific primers.
ChIP-chips experiments were done using two-independent p400-immunoprecipitated chromatin samples from HCT116 cells. After the p400-ChIP, samples were amplified using the WGA kit (Sigma-Aldrich). DNA labeling and hybridization on Human ChIP-chip 385K RefSeq Promoter Arrays (Roche Nimblegen) were performed. Bioinformatic analysis allowing identification of peaks was done using the Nimblescan software. A promoter was recorded as bound by p400 if binding peaks (P < 0.01) were identified in the two experiments at <500 pb distance, which corresponds to the ChIP resolution. Clustering of bound genes was done using GeneCoDis3 online software (http://genecodis.cnb.csic.es) (51).
This work was supported by the Ligue Nationale Contre le Cancer as an ‘Équipe labellisée’ (Grant #VP7537) and from the Fondation of the Association pour la Recherche contre le Cancer. L.M. and F.E. were awarded by the Fondation de France for this work.
We thank Dr B. Amati for providing us Tip60+/− mice from Dr J. Lough's lab. We greatly thank Dr C. Neuveut for providing us all β-catenin expression vectors and Dr J.-F. Beaulieu for the gift of HIEC cells. We thank N. Naud for her help with ACF analysis. We used the INSA-GeT-Biopuces and the LBCMCP Toulouse Genotoul TRI cytometry plateforms as well as the luminometry facility of the LBME (UMR5099, Toulouse).
Conflict of Interest statement: None declared.
- signal transduction
- cell proliferation
- colorectal cancer
- adenosine triphosphatases
- dna methylation
- tumor suppressor genes
- colon cancer
- signal pathway
- signal transduction pathways
- tumor progression