Integration of TGF-ß and Ras/MAPK Signaling Through p53 Phosphorylation

See allHide authors and affiliations

Science  09 Feb 2007:
Vol. 315, Issue 5813, pp. 840-843
DOI: 10.1126/science.1135961


During development and tissue homeostasis, cells must integrate different signals. We investigated how cell behavior is controlled by the combined activity of transforming growth factor–β (TGF-β) and receptor tyrosine kinase (RTK) signaling, whose integration mechanism is unknown. We find that RTK/Ras/MAPK (mitogen-activated protein kinase) activity induces p53 N-terminal phosphorylation, enabling the interaction of p53 with the TGF-β–activated Smads. This mechanism confines mesoderm specification in Xenopus embryos and promotes TGF-β cytostasis in human cells. These data indicate a mechanism to allow extracellular cues to specify the TGF-β gene-expression program.

Cross-talk between signaling pathways is necessary to effect efficient and fine-tuned regulatory control over metazoan development and physiology. TGF-β and receptor tyrosine kinase (RTK) ligands are pleiotropic cytokines affecting several aspects of cell behavior, ranging from differentiation and proliferation to movement and survival (1, 2). Previous work has shown that these signaling pathways are integrated: The Ras/MAPK cascade, which is downstream of RTK signaling, affects TGF-β–induced mesoderm development in vertebrate embryos and growth arrest in mammalian adult cells (1, 36). However, the mechanisms underlying this partnership have not been elucidated.

Smad2/3 and the tumor suppressor protein p53 physically interact and jointly regulate the transcription of several TGF-β target genes (7). p53 is activated by multiple stimuli through posttranslational modifications (8). Hence, p53 activation might serve to convey cues from extracellular signals within the TGF-β gene-expression program.

To investigate whether p53 acts as an integration node between Ras/MAPK and TGF-β pathways, we carried out loss-of-function studies in Xenopus embryos, where pluripotent cells of the animal pole (animal cap) can differentiate into mesoderm by the combined action of these signals. Endogenous p53 was depleted by microinjecting p53 antisense morpholino oligonucleotides (p53-MO) (fig. S1A). Control or p53-depleted animal cap cells were treated with combinations of fibroblast growth factor (FGF) and Activin proteins. FGF enhanced Activin-mediated induction of mesoderm markers (Fig. 1A, lanes 3 and 4), but this cooperation was lost in p53-depleted cells (Fig. 1A, lanes 7 and 8). Moreover, both p53 and FGF are required for the induction of a panel of Activin target genes (fig. S1B). These results are consistent with p53 being required downstream of FGF to foster TGF-β gene responses.

Fig. 1.

FGF potentiates TGF-β gene responses in a p53-dependent manner. (A) Reverse transcription polymerase chain reaction (RT-PCR) analysis for mesodermal markers (VegT, Snail, Xbra, and Mix.2) of animal caps explanted from Xenopus embryos injected with control-MO or p53-MO (40 ng). Where indicated, explants were treated with FGF1 (25 ng/ml) and Activin (6 ng/ml) and cultured for 2 hours before harvesting. The samples injected with p53-MO (lanes 5 to 8) were subjected to two additional PCR cycles for all the markers analyzed, in order to visualize residual mesodermal gene activations triggered by Activin in the absence of p53. ODC (ornithine decarboxylase) serves as a loading control. (B) FGF and Activin cooperate for the formation of a Smad2/p53 protein complex. Western blot analyses for endogenous p53 and Smad2 of protein complexes purified by DNA affinity purification (DNAP) with a wild-type (anti-p53 DNAP, lanes 2 to 5), or mutant (control DNAP, lane 1) biotinylated p53-consensus probe (7). (C) RT-PCR analyses for mesodermal marker genes induced in animal cap explants by wild-type (wt) or N-mut Xp53 mRNAs. Xenopus embryos were injected with p53-MO (40 ng) and different doses (10 pg, 30 pg, and 90 pg) of mRNAs encoding for the Xp53 isoforms. (D) Interaction of Xp53 and Smad2 requires p53 N-terminal phosphorylation. Extracts from Xenopus embryos injected with combinations of mRNAs (100 pg each) for Flag-tagged Smad2, wild-type or N-mut Xenopus p53, and Activin were precipitated by anti-p53 DNAP. The panels show Western blots for Smad2 and Xp53.

Upon overexpression of p53 in Xenopus animal cap cells, p53 cooperates with endogenous Smads to induce mesoderm markers (7) (fig. S1C); this p53 activity is counteracted by treatment of animal caps with the FGF-receptor inhibitory compound SU5402 or expression of a dominant-negative version of Raf, a critical component of the Ras/MAPK pathway (fig. S1, D and E). This suggests that endogenous RTK signaling promotes the mesoderm-inducing activities of p53. Moreover, p53 and FGF cooperate in mesoderm induction, as assayed by the induction of ectopic tail structures in whole embryos (fig. S1, F to J). To address the biochemical basis of this link, we treated human HepG2 cells with combinations of FGF and Activin proteins and then purified p53 from corresponding nuclear lysates by DNA-affinity purification (DNAP). As shown in Fig. 1B, treatment with FGF efficiently promotes the association of p53 and TGF-β–activated Smad2 within the same complex.

To gain insight into this process, we defined the structural determinants of p53 that are relevant for Smad binding. p53 binds recombinant Smad3 through the p53 N-terminal domain (fig. S2A). This segment carries several Ser/Thr residues (fig. S2B), whose phosphorylation has been implicated in p53 activation upon DNA damage (8). We found that endogenous FGF signaling also promotes phosphorylation of these residues in Xenopus embryos (fig. S2C).

To address whether N-terminal phosphorylation plays a causative role in guiding p53 activity toward the TGF-β pathway, we compared wild-type and N-mut Xp53, in which the N-terminal Ser/Thr residues have been replaced by Ala, for their ability to rescue TGF-β gene responses in p53-depleted Xenopus embryos. Mutation of the N-terminal phosphorylation sites severely impairs p53 mesoderm-inducing ability (Fig. 1C). Similar results were obtained with wild-type and N-mut mammalian p53 (fig. S2D). At the biochemical level, only wild-type and not N-mut p53 can complex with Smad2 (Fig. 1D and fig. S2E), indicating that this interaction, rather than being constitutive, must be enabled by p53 N-terminal phosphorylation. This appears to be a peculiar requirement, because wild-type and N-mut p53 display similar stability, transactivation capacity, and apoptosis-inducing activity (fig. S2, F to H). These results indicate that phosphorylation of N-terminal Ser/Thr residues is relevant for coupling p53 activity to Smad responses.

To investigate the relevance of p53 N-terminal phosphorylation for the activation of the TGF-β cytostatic program in human cells, we established a p53-complementation assay using the p53-null H1299 human lung cancer cell line (9). These cells retain an intact TGF-β transduction cascade and yet are unable to activate the TGF-β cytostatic program (fig. S3). Robust TGF-β–mediated induction of the cyclin-dependent kinase (CDK) inhibitors p21Waf1 and p15ink4b is rescued by adding back wild-type p53, whereas expression of p53-N-mut fails to do so (fig. S3B).

To identify the residues that must be phosphorylated in vivo to enable p53/Smad cooperation, we refined our analysis by comparing wild-type p53 with p53 mutants carrying Ala substitutions in (i) Ser15, Thr18, and Ser20, (ii) Ser6 and Ser9, or (iii) individual residues. As shown in Fig. 2A, all of these p53 isoforms similarly rescued the expression of mdm2, a TGF-β–independent p53 target, as well as the basal levels of p21Waf1. Moreover, Ser15, Thr18 and Ser20 were not required for inducibility of p21Waf1 and p15ink4b by TGF-β signaling. In contrast, phosphorylation of Ser9 and Ser6 was relevant for Smad cooperation. In line with these results, Ser9 phosphorylation was required to fully empower the mesoderm-inducing properties of p53 in Xenopus embryos (fig. S4). To extend these observations to TGF-β–induced cytostasis, we measured incorporation of bromodeoxyuridine (BrdU) in parental (p53-null) and p53-reconstituted H1299 cells. As shown in Fig. 2B, only wild-type p53, but not p53S9A, could rescue TGF-β–dependent growth arrest. Mechanistically, this is due to an impaired ability of p53S9A to complex with Smad2 (Fig. 2C). We then investigated whether endogenous Ras/MAPK signaling is relevant for p53 phosphorylation in Ser6/Ser9 (P-Ser). H1299 cells carry an activating mutation in N-Ras, leading to constitutive MAPK signaling (9). We found that inhibition of MAPK kinase (MEK), an effector of Ras upstream of MAPK, causes specific inhibition of P-Ser6 and P-Ser9 levels, with concomitant loss of TGF-β–mediated p21Waf1 induction in p53-reconstituted H1299 cells (fig. S5A). Hence, p53 phosphorylation in Ser9 and Ser6 serves as integration node in the cross-talk between Ras/MAPK and TGF-β.

Fig. 2.

Requirement of p53 Ser9 phosphorylation for activation of the TGF-β cytostatic program. (A) Phosphomutant-p53 isoforms were tested for the ability to rescue TGF-β responsiveness in H1299 cells in comparison with wild-type mouse p53. p53 STSA carries Ala substitutions in Ser15, Thr18 and Ser20; p53S6, 9A, S6A, or S9A carry Ala substitutions in Ser6 and/or Ser9. Transfection and analysis of H1299 were as in fig. S3B. Fold inductions are the ratio of p21Waf1 or p15ink4b expression in the presence or absence of TGF-β stimulation, normalized on p53 levels. (B) Wild-type, but not p53S9A, rescues TGF-β–induced growth arrest in H1299 cells. Cells were transfected with the indicated p53 expression constructs as in (A) and assayed for BrdU incorporation. Columns represent the number of BrdU positive cells in the absence (cyan) or presence (red) of TGF-β stimulation, relative to the number of proliferating cells in the unstimulated control. (C) The interaction of p53 and Smad2 requires Ser9 phosphorylation. Nuclear extracts from H1299 cells transfected either with wild-type mouse p53 or p53S9A were precipitated by anti-p53 DNAP. The panels show Western blots for Smad2 and p53.

This prompted us to consider the possibility that, although p53 is a ubiquitous protein, FGF might spatially pattern p53's activity. In Xenopus, expression of different FGFs (eFGF, FGF3, and FGF8) is enriched in the marginal zone of the embryo, from which the mesoderm emerges, whereas lower FGF activity is present in the animal pole (10) (Fig. 3A). Using phosphospecific antibodies, we found that kinase activities targeting Ser9 and Ser6 are localized in the marginal zone; in contrast, phosphorylation in other residues appears constitutive (Fig. 3B). To determine whether endogenous FGF signaling is responsible for this graded p53 phosphorylation along the animal-vegetal axis, embryos were treated with the FGF-receptor inhibitor SU5402 or injected with DN-Raf mRNA. Blockade of FGF signaling causes specific down-regulation of P-Ser9 and P-Ser6 (Fig. 3C). Conversely, ectopic FGF expression in animal cap cells specifically raises P-Ser6 and P-Ser9 levels (Fig. 3D). Similarly, at the biochemical level, FGF is required for p53/Smad2 interaction because the formation of this complex is inhibited by SU5402 (fig. S6). However, introduction of Ser to Glu phosphomimicking substitutions in Ser6 and Ser9 (p53S6, 9E), renders p53 able to complex with Smad2 in an FGF-independent manner (fig. S6). Together, the results indicate that FGF patterns the phosphorylation status of p53 in the embryo, restricting its cooperation with TGF-β to the prospective mesoderm.

Fig. 3.

FGF phosphorylates p53 on Ser6 and Ser9 through CK1ϵ/δ. (A) Schematic diagram showing the distribution of FGF/MAPK signaling in the Xenopus embryo at late blastula stage (10). (B to D) Analysis of the phosphorylation status of human p53 (100 pg) injected in Xenopus embryos. p53 was purified by immunoprecipitation, and phosphoresidues were detected by Western blot. (B) p53 mRNA was injected in the animal pole or in the marginal zone of Xenopus embryos. p53 phosphorylation on Ser6 and Ser9 is enriched in the marginal zone, where FGF signaling is stronger. (C) p53 mRNA was injected in the marginal zone region alone or in combination with DN-Raf mRNA (1 ng). When indicated, injected embryos were cultivated in the presence of the FGFR1 inhibitor SU5402 (60 μM). (D) FGF enhances p53 phosphorylation of Ser9 and, to a minor extent, of Ser6. p53 mRNA was injected in the animal pole region alone, or in combination with eFGF mRNA (0.8 pg). Consistent results were obtained on endogenous p53 in human cells (fig. S5B). (E) CK1ϵ induces expression of mesodermal genes in a p53-dependent manner. Xenopus CK1e mRNA was injected at four different doses (200 pg, 400 pg, 800 pg, and 1.6 ng) together with control-MO or p53-MO (40 ng). Animal cap explants were dissected at late blastula stage as in Fig. 1A. Additional markers are shown in fig. S8A. (F and G) Whole-mount in situ hybridizations for the pan-mesodermal marker Xbra in control-MO and CK1ϵ-MO–injected embryos. (H) CK1ϵ/δ are required for Ser6 and Ser9 phosphorylation in human cells. The panel shows Western blot analysis of p53 phosphorylation in H1299 cells. Wild-type mouse p53 was transfected in combination with control-siRNA (small interfering RNA) or anti-CK1ϵ/δ siRNA. CK1ϵ depletion was monitored by Western blot. (I) CK1ϵ/δ is required for the TGF-β cytostatic program through p53 Ser9 phosphorylation.

Next, we wished to gain insight into the kinase responsible for inducing p53 phosphorylation in response to FGF/Ras/MAPK signaling. Both Ser6 and Ser9 conform to a CK1 consensus: There are seven mammalian CK1 genes, but p53 has been shown to associate specifically with CK1ϵ and CK1δ (11). In Xenopus embryos, inhibition of these kinases with dominant-negative CK1ϵ (DN-CK1ϵ) (12, 13) antagonizes FGF-mediated Ser6 and Ser9 phosphorylation (fig. S7). Biologically, increasing levels of CK1ϵ promote mesoderm induction in a p53-dependent manner (Fig. 3E and fig. S8); conversely, loss-of-CK1ϵ by microinjection of DN-CK1ϵ or CK1ϵ morpholino inhibits endogenous and p53-mediated mesodermal gene expression (Fig. 3, F and G, and fig. S9). Thus, CK1ϵ lies downstream of FGF to promote p53 phosphorylation and Smad cooperation in Xenopus mesoderm development.

We next investigated the relevance of CK1ϵ/δ-mediated p53 phosphorylation on the activation of the TGF-β cytostatic program in human cells. To this end, p53-reconstituted H1299 cells were transfected with siRNAs to deplete endogenous CK1ϵ and CK1δ. CK1ϵ/δ knockdown leads to down-regulation of P-Ser6 and P-Ser9 levels (Fig. 3H) and to loss of TGF-β–mediated p21Waf1 induction (Fig. 3I, compare lanes 3 and 4 with lanes 7 and 8). By contrast, a phosphomimicking substitution of Ser9 with Glu (p53S9E) renders p53 able to sustain TGF-β–mediated p21Waf1 induction even in the absence of CK1ϵ/δ (Fig. 3I, compare lane 4 with lane 8 and lane 6 with lane 10). Hence, p53S9E acts epistatically to CK1ϵ/δ. This indicates the key role of p53 N-terminal phosphorylation as mediator of the positive effect of CK1ϵ/δ in supporting TGF-β cytostatic responses.

We have established a role for p53 as signaling integrator, outside of its widely investigated response to genotoxic stress (8). We provide evidence that p53 activity, rather than stability, can be qualitatively patterned by RTK/Ras-induced phosphorylation through CK1ϵ/δ. This phosphorylation step enables a robust biochemical interaction of p53 with TGF-β–activated Smads, leading to mesoderm induction in embryos and, in human cells, to the deployment of the TGF-β cytostatic program.

These data establish a mechanistic link between three key regulators of cell proliferation that are dysregulated in human cancers: Ras, p53, and TGF-β. This could provide an explanation for the p53-dependent tumor-suppressive function of Ras/MAPK reported in primary cells (14, 15). Activated Ras may well have general growth-promoting effects but, in the presence of wild-type p53, this would be balanced by the positive role played on p53/Smad cooperation that would sustain TGF-β growth control and thus limit neoplastic transformation.

Supporting Online Material

Materials and Methods

Figs. S1 to S9


References and Notes

View Abstract

Navigate This Article