β-Catenin C-terminal signals suppress p53 and are essential for artery formation

Increased activity of the tumour suppressor p53 is incompatible with embryogenesis, but how p53 is controlled is not fully understood. Differential requirements for p53 inhibitors Mdm2 and Mdm4 during development suggest that these control mechanisms are context-dependent. Artery formation requires investment of nascent endothelial tubes by smooth muscle cells (SMCs). Here, we find that embryos lacking SMC β-catenin suffer impaired arterial maturation and die by E12.5, with increased vascular wall p53 activity. β-Catenin-deficient SMCs show no change in p53 levels, but greater p53 acetylation and activity, plus impaired growth and survival. In vivo, SMC p53 inactivation suppresses phenotypes caused by loss of β-catenin. Mechanistically, β-catenin C-terminal interactions inhibit Creb-binding protein-dependent p53 acetylation and p53 transcriptional activity, and are required for artery formation. Thus in SMCs, the β-catenin C-terminus indirectly represses p53, and this function is essential for embryogenesis. These findings have implications for angiogenesis, tissue engineering and vascular disease.

T he p53 tumour suppressor plays a fundamental role in cell biology by inducing cell cycle arrest and apoptosis. Increased p53 activity is incompatible with normal embryogenesis, a process characterized by active growth. Indeed, inactivation of the key inhibitors of p53, Mdm2 and Mdm4, causes embryonic lethality 1,2 , and increased p53 activity during development yields a phenotype resembling CHARGE syndrome 3 . Although Mdm2 and Mdm4 are the major p53 inhibitors 4 , the differential requirement for Mdm2 and Mdm4 in cardiac and central nervous system development 5 raises the possibility that cell type-specific mechanisms restrain p53 during embryogenesis. This is especially intriguing during artery formation: although smooth muscle cells (SMCs) constitute the main component of the arterial wall, requirements for Mdm2 or Mdm4 have not been evaluated in this cell type, and how SMCs repress p53 function to proliferate, survive and build the arterial wall is unknown.
Development of a competent vasculature reflects the integration of several processes, including vasculogenesis, angiogenesis, remodelling and vascular maturation. Despite recent advances in this area 6 , understanding of vascular maturation remains limited. Mural cells-SMCs in large vessels or pericytes in the microcirculation-surround nascent endothelial tubes in a process called investment 7 to provide quiescence and stability 8 , and proliferate and synthesize extracellular matrix components to build a multilayered wall 9 , but how these activities are regulated is only partly understood.
b-Catenin, the non-redundant component of the canonical Wnt signalling pathway and structural element of the cadherin-catenin cell-adhesion complex 10 , has critical functions in embryogenesis and adult homoeostasis 11 . In the vasculature, endothelial cell b-catenin is required for angiogenesis in the central nervous system and contributes to endothelial arterial specification 12,13 . b-Catenin is also necessary for smooth muscle development in the lung 14 and in the proepicardium for coronary artery formation 15 . While b-catenin promotes proliferation and survival of cultured SMCs, and its expression and activity are induced in adult arteries during vascular remodelling [16][17][18][19] , whether it is required for vascular maturation during development is unknown.
To address this question, we inactivated b-catenin selectively in SMCs and tested a series of b-catenin mutants in vivo, and found an absolute requirement for SMC b-catenin in artery formation and mouse embryogenesis. Here, we provide evidence that b-catenin signalling in SMCs is essential for artery formation. Loss of this activity, particularly that emanating from the b-catenin C-terminal interaction domain, results in an incompetent circulation that cannot support embryogenesis. Interestingly, b-catenin performs this function in part by inhibiting p53 activity.

Results
SMC b-catenin is essential for artery formation. We studied SMC-selective inactivation of Ctnnb1 (b-catenin) by crossing Tagln-Cre transgenic 20 and Ctnnb1 flox/flox mice 21 . We found no surviving Tagln-Cre;Ctnnb1 flox/flox (SMbCKO; see Table 1 for mouse line abbreviations) progeny at E13.5 or beyond ( Fig. 1a; Supplementary Tables 1 and 2; Supplementary Fig. 1a,b). At E12.5, all control embryos looked normal, while all SMbCKO embryos were compromised, ranging in appearance from early stages-small size, dilated blood vessels and compromised overall development, but retaining recognizable morphologic features (Fig. 1b)-to advanced stages of resorption-still smaller, generally round and pale in appearance, and lacking morphological features. Mutant embryos appeared grossly normal until E10.5, indicating that vasculogenesis and early angiogenesis were largely unaffected. Evaluation of SMbCKO embryos revealed reduced wall thickness of the paired and fused dorsal aortas (PDA and FDA, respectively) as early as E10.5 ( Fig. 1c-e), plus increased cross-sectional area of the FDA at E11.5 (Fig. 1c,f); immunohistochemistry confirmed decreased b-catenin expression in the vessel wall and heart ( Supplementary  Fig. 1c). Because Tagln is transiently expressed in the heart, we also looked for a potential cardiac phenotype, but did not find (1) signs of cardiac dysfunction, such as oedema or pericardial effusion ( Supplementary Fig. 1d), (2) differences in the future left or right ventricles, compact or trabeculated myocardium (Supplementary Fig. 1e-g), (3) change in outflow tract (OT) dimensions in SMbCKO embryos at E10.5 ( Supplementary  Fig. 1h) or (4) major structural abnormalities at E11.5 ( Supplementary Fig. 1i)-these observations differ from a report of b-catenin inactivation using a different Tagln-Cre transgenic line, which describes severe right ventricular hypoplasia at E9.5 and demise between E10.5 and E11.5 (ref. 22). In our system, possibly due to different transgene integration site or modest variation in level, timing or distribution of Tagln-Cre expression or strain background, we observe that b-catenin loss in SMCs impairs artery formation, as reflected by thinned and dilated major vessels observed before any evident cardiac abnormality.
b-catenin promotes arterial SMC proliferation and survival. The arterial wall has two main cellular components: endothelial cells and SMCs. While CD31 staining demonstrated an intact endothelial monolayer in both control and SMbCKO embryos ( Fig. 2a; Supplementary Fig. 2a), staining for SMC markers showed lack of SMC investment and absence of a multilayered wall by E12.5 in SMbCKO embryos (Fig. 2b). We examined SMC-specific proliferation in the PDA wall by co-staining two markers of proliferation, phospho-histone H3 (pHH3) or Ki67, with smooth muscle a-actin (a-SMA), and found reduced SMC proliferation in SMbCKOs at E9.5 and E10.5 (Fig. 2c,d; Supplementary Fig. 2b). We also observed decreased cell proliferation in the FDA wall from E9.5 to E11.5 ( Supplementary  Fig. 2c). In addition, by both TUNEL assay and co-staining of cleaved Caspase 3 and a-SMA, we observed increased SMC apoptosis in the SMbCKO arterial wall (Fig. 2e,f; Supplementary  Fig. 2d). These findings suggest that the necessity for SMC b-catenin in formation of a multilayered arterial wall stems from its positive effects on cell proliferation and survival.
b-catenin is required for SMC population growth. To evaluate how b-catenin affects basic cell functions, we isolated aortic SMCs from Ctnnb1 flox/flox mice and transduced them with Cre-or GFP-expressing adenovirus to obtain b-catenin-deficient and control SMCs, respectively (Fig. 3a). In culture, SMCs lacking b-catenin grew more slowly and plateaued at a lower density (Fig. 3b). Moreover, with serum starvation, the b-catenindeficient cell population decreased faster and to a greater extent (Fig. 3c). Cell cycle analyses found more b-catenin-deficient SMCs in the G 0 /G 1 phase and less in S or G 2 /M (Fig. 3d), indicating that b-catenin is required for cell cycle progression. Serum deprivation caused more death of b-catenin-deficient than control cells (Fig. 3e), showing that SMC b-catenin is a prosurvival factor. On the other hand, SMC marker gene expression and transwell cell migration were not affected by b-catenin loss, suggesting a lack of effect on maintenance of SMC differentiation and motility (Fig. 3f,g; Supplementary Fig. 3).
b-catenin opposes p53 acetylation and activity in SMCs. Cdkn1a (p21) repression and Ccnd1 (cyclin D1) induction have ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms12389 been associated with activation of b-catenin/TCF signalling and proliferation in cultured SMCs 17,18 , so we tested expression of these genes, as well as that of proapoptotic Bax and antiapoptotic Bcl2l1, which have been related to b-catenin 23,24 . b-Catenindeficient SMCs expressed higher mRNA levels of Cdkn1a, Ccnd1, Bax, and Bcl2l1, but lower levels of Axin2, a known target of b-catenin (Fig. 4a). Since increased Ccnd1 and Bcl2l1 levels could not explain the observed phenotype, we focused on Cdkn1a and Bax (direct targets of p53), and hypothesized that b-catenin might restrain p53 function to promote SMC survival and proliferation. Using a p53 reporter assay 25 , we found a substantially higher p53 transcriptional activity in b-catenindeficient SMCs ( Fig. 4b; Supplementary Fig. 4a)-importantly, WT SMCs treated with the same Ad-Cre protocol did not exhibit increased p53 activity (Supplementary Fig. 4b). Restoration of b-catenin function in b-catenin-deficient SMCs with b-catenin S33Y , a constitutively active form 26 , reduced p53 activity toward control levels (Fig. 4b); b-catenin S33Y also abrogated p53 activity induced by p53 overexpression (Fig. 4c). In addition, SMCs treated with the b-catenin inhibitor XAV939 (ref. 27), which reduced b-catenin/TCF transcriptional activity in SMCs ( Supplementary  Fig. 4c), showed an increase in p53 activity (Fig. 4d). These observations indicate that b-catenin inhibits p53 activity in vascular SMCs.
P53 activation involves three steps, (1) stress-induced p53 accumulation by several mechanisms, many of which affect the ability of Mdm2 to ubiquitinate p53, (2) release of p53 from the Mdm2-and Mdm4-mediated repression, and (3) recruitment and interaction with numerous cofactors 4 . Surprisingly, b-catenin loss in SMCs neither increased p53 expression ( Fig. 4e; Supplementary Fig. 4d), nor decreased the levels of Mdm2 and Mdm4 (Fig. 4e). Moreover, RO-5963-an Mdm2 and Mdm4 inhibitor 28 , at a dose sufficient to abrogate exogenous Mdm2 function in SMCs ( Supplementary Fig. 4e)-did not increase p53 function in SMCs expressing physiologic levels of these ubiquitin ligases (Fig. 4f). Similarly, siRNA-mediated knockdown of Mdm2  In d-f, data represent the mean ± s.e.m., groups were compared by two-way ANOVA and Bonferroni's multiple comparisons test.
and Mdm4 in SMCs failed to induce p53 activity ( Supplementary  Fig. 4f,g), which suggests that cultured SMCs do not require Mdm2/Mdm4 to restrain p53 activity. These observations support the idea that b-catenin inhibits SMC p53 activity by a mechanism independent of effects on Mdm2 and Mdm4. Conceivably, this mechanism could limit p53 binding to DNA or decrease its recruitment of transcriptional coactivators. To separate these functions, we used a Gal4-binding domain-p53 fusion protein (Gal4BD-p53), and found that activity of this protein on a promoter driven via Gal4 response elements (pFR-Luc) was greater in b-catenin-deficient SMCs (Fig. 4g). Thus, b-catenin decreases p53 transcriptional activity through a post-translational mechanism that may affect recruitment of transcriptional coactivators. Acetylation of p53 is indispensable for its transcriptional activity, triggering recruitment of coactivators, among other functions [29][30][31] , so we assessed this modification in SMCs and found that b-catenin loss, but not Mdm2/Mdm4 inhibition, robustly increased p53 acetylation (Fig. 4h,i). Altogether, our findings support the idea that b-catenin indirectly inhibits p53 transcriptional activity in vascular SMCs not by limiting its abundance or potentiating Mdm2-/Mdm4mediated repression, but through preventing p53 acetylation.
b-catenin restrains p53 in SMCs during artery formation. To evaluate p53 activity in vivo, we tested SMC expression of a known p53 target gene, Bax, in E11.5 embryos, and found that SMbCKO embryos exhibited higher expression than controls (Fig. 5a), consistent with increased p53 activity on loss of b-catenin. Thus, we hypothesized that repression of p53 is an essential aspect of the requirement for b-catenin in SMCs during artery formation. To test this idea, we studied mice with SM-selective inactivation of Trp53 (SMp53KO mice), and with simultaneous inactivation of Trp53 and Ctnnb1 (SMp53KO; bCKO). SMp53KO mice developed normally (not shown), indicating that SMC p53 activity is not required for artery formation. To test if unrestrained p53 activity contributes to the phenotype seen with loss of SMC b-catenin, we compared SMbCKO and SMp53KO;bCKO embryos at E12.5, and noted differences in the degree of resorption. Among SMbCKO embryos, 21% were in an early stage of resorption, while 79% appeared advanced (Fig. 5b,c). Interestingly, development of SMp53KO;bCKO embryos was less affected: 23% looked normal, 36% were in early resorption and only 41% in advanced resorption-a clear difference from findings with SMbCKO embryos (Fig. 5b,c)  we observed aortas with multilayered walls, an observation not seen in any SMbCKO embryos at this stage (Fig. 5d), and with reduced Bax expression ( Supplementary Fig. 5a-c). Thus loss of SMC p53 itself has no repercussion on vascular development, but suppresses the vascular phenotype resulting from inactivation of SMC b-catenin, supporting the idea that b-catenin represses p53 activity in SMCs in vivo during artery formation.
In addition, SMp53 þ / À ;bCKO embryos phenocopied the SMbCKO embryos (Fig. 5c,e), indicating that a single Trp53 allele provides sufficient p53 activity to prevent artery formation on loss of SMC b-catenin, and suggesting that low levels of p53 are effective in SMCs. Although the improvement in artery formation and overall development was important, p53 loss did not fully overcome the phenotype of SMbCKO embryos-thus, the critical requirement for SMC b-catenin during development extends beyond its suppression of p53 activity.
Since b-catenin has two major functions in the cell-as a structural component in cadherin-mediated cell adhesion, and as a signalling factor acting as a transcriptional coactivator in the canonical Wnt pathway 10 -we wanted to know which function was required for p53 regulation. We recently developed and validated b-catenin mutants that partially or totally lack transcriptional activity, but have intact cell-adhesion function 32 : (1) Ctnnb1 D164A , which has a single amino acid change (D164A) in the first armadillo repeat that prevents binding to the N-terminal coactivator B-cell lymphoma 9/BCL9L, and exhibits decreased transcriptional activity; (2) Ctnnb1 DC , a C-terminal truncation that changes codon 673 to a premature stop signal, abrogating interaction with several C-terminal interacting coactivators, and shows decreased transcriptional activity; and (3) Ctnnb1 dm , a double-mutant with both the D164A mutation and the C-terminal truncation, which lacks any detectable transcription-inducing activity (Fig. 6a). As above, we studied these signalling-specific mutations in vascular SMCs in the context of constitutively active b-catenin S33Y . b-Catenin dm did not repress p53 transcriptional activity (Fig. 6b), suggesting that b-catenin signalling function is essential, while its cell-adhesion role is not sufficient to restrain p53 activity. Signalling through b-catenin depends on protein-protein interactions with coactivators recruited by its N-terminal and C-terminal domains 10,33 . We used b-catenin D164A and b-catenin DC to compare the importance of N-or C-terminal interactions for p53 regulation. Similar to b-catenin dm , b-catenin DC failed to reduce p53 activity, whereas b-catenin D164A abrogated it (Fig. 6b). Moreover, while Gal4BD-p53 transcriptional activity was not affected by b-catenin dm or b-catenin DC , a decrease was observed with b-catenin D164A (Fig. 6c). These observations suggest that b-catenin C-terminal interactions are required to repress p53 activity in vascular SMCs, whereas N-terminal interactions are dispensable. Moreover, the fact that b-catenin dm -which has an intact TCF-binding domain 10,32 -did not repress p53 activity also suggests that b-catenin/TCF interaction is not sufficient to fulfil this repressive function. To test this idea further, we used an inhibitor of the b-catenin/TCF complex, PKF118-310 (ref. 34). A dose of PKF118-310 that reduced b-catenin/TCF transcriptional activity in SMCs (Supplementary Fig. 6a) did not increase p53 activity in these cells expressing endogenous or induced levels of p53 ( Supplementary Fig. 6b). Similarly, PKF118-310 decreased, rather than increased, Gal4BD-p53 activity ( Supplementary  Fig. 6c), suggesting that b-catenin/TCF interaction and associated Wnt target genes are not acting to restrain p53 activity in vascular SMCs.
Altogether these observations support the idea that b-catenin C-terminal interactions are necessary to restrain p53 activity through a b-catenin/TCF-independent mechanism. Interestingly, one of the known C-terminal interactors is the histone acetyltransferase Creb-binding protein (CBP) 10,35 , which also acetylates p53 (refs 29,36). b-Catenin prevents p53 acetylation (Fig. 4h), so it is possible that b-catenin/CBP interaction limits CBP availability for interaction with and acetylation of p53 in the nucleus of vascular SMCs. Consistent with this idea, we found that both b-catenin and CBP localized to the nucleus of SMCs by immunofluorescence (Fig. 6d) and were present in the nuclear fraction of SMCs (Fig. 6e), and nuclear b-catenin expression was also observed in the wall of aortas of E12.5 control embryos ( Supplementary Fig. 6d-f). Notably, blocking b-catenin/CBP interaction with ICG001 (ref. 37), which inhibited b-catenin/TCF transcriptional activity in SMCs ( Supplementary Fig. 6g), increased p53 acetylation (Fig. 6f); moreover, knocking down CBP in b-catenin-deficient SMCs decreased acetylation of p53 (Fig. 6g,h)-while CBP knockdown in control SMCs (Ad-GFP) did not have any effect on their relatively low levels of acetylated p53 ( Supplementary Fig. 6h).

b-Catenin C-terminal interactions promote artery formation.
To test the physiologic relevance of the above findings, we used mice in which mutant Ctnnb1 alleles 32 , like those described above, provided the only source of b-catenin in SMCs. We crossed Ctnnb1 dm/flox and Tagln-Cre;Ctnnb1 flox/WT mice to produce SMbC dm (SMC b-catenin replaced with the signalling-defective mutant), SMbCKO (complete loss of SMC b-catenin) and Ctnnb1 flox/WT (wild-type b-catenin) littermates. Interestingly, we found no SMbC dm mice in postnatal screening (Supplementary Table 3, Supplementary Fig. 7a), indicating that these mutants were embryonic lethal.
Notably, all E12.5 SMbC dm embryos were undergoing resorption-with 9% early and 91% advanced in their state of resorption. This finding was like that seen with SMbCKO embryos (Fig. 7a,b), and indicates that b-catenin's cell-adhesion function in SMCs is insufficient, whereas its signalling function is essential, for artery formation and normal embryogenesis.
To evaluate the importance of N-and C-terminal SMC b-catenin interactions in vivo, we designed breeding strategies to produce SMbC D164A and SMbC DC mice. In postnatal screening, these strategies yielded no SMbC DC mice (Supplementary Table 4, Supplementary Fig. 7a), but normal appearing SMbC D164A mice (Supplementary Table 5, Supplementary Fig. 7a) that reached adulthood. At E12.5, 19% of SMbC DC embryos looked normal, 44% were in early resorption and 37% in advanced resorption-a profile somewhat milder than that of SMbCKO embryos (Fig. 7b). In contrast, all SMbC D164A embryos appeared normal (Fig. 7b).
Moreover, we found correlation between whole embryo phenotype and the structure of the FDA at E12.5: control embryos had a clear multilayered vessel wall, whereas SMbCKO aortas lacked SMC investment, as described before. In some SMbC DC embryos, we observed an important, although incomplete, recovery of the vessel wall, while in the SMbC D164A embryos, the aortic wall was fully restored (Fig. 7c). Since SMbC DC embryos were embryonic lethal, but showed an intermediate phenotype at E12.5, we tested the idea that SMbC DC could delay demise. At E15.5, a stage at which SMbCKO embryos were never observed, we found some SMbC DC embryos, although all of them were undergoing resorption (Fig. 7d,e). These findings indicate that expression of the Ctnnb1 DC allele as the only source of b-catenin in SMCs ameliorates defective artery formation and delays embryonic death. On the other hand, Ctnnb1 D164A was sufficient for artery formation that supported embryonic and postnatal development in a proportion of mice; however, that proportion of SMbC D164A mice observed in postnatal screening was lower than expected ( Supplementary Fig. 7b). At E15.5, stage at which all SMbC DC embryos were abnormal, all SMbC D164A embryos looked normal ( Supplementary Fig. 7c), suggesting that demise of some SMbC D164A mice may occur perinatally or postnatally. Altogether, these in vivo studies support the idea that b-catenin signalling function in SMCs is essential for artery formation. Moreover, b-catenin C-terminal interactions are required to fulfil this signalling role; by comparison, b-catenin N-terminal interactions are not essential for early events of vascular maturation, but may contribute to survival at later stages of development.

Discussion
This study demonstrates for the first time that SMC b-catenin is essential for developmental vascular maturation. SMC b-catenin deletion impairs artery formation with critical effects on SMC proliferation and survival. While other signalling pathways, including Notch 38,39 , sphingosine-1-phosphate 40,41 , PDGFB/PDGFRb 42-44 and TGF-b 8,45-47 participate in vascular maturation, the phenotype due to SMC b-catenin inactivation is distinct in terms of timing, vessels affected and lack of effect on cellular differentiation.
Mechanistically, we found that b-catenin inhibits p53 activity in SMCs, and loss of this function impaired artery formation. Interestingly, postnatal increase in p53 activity inhibits angiogenesis 48 . Thus, repression of p53 activity appears necessary for developmental vascular maturation and for postnatal angiogenesis; our findings point to this as a key function of SMC b-catenin during development.
This functional connection between b-catenin and p53 is surprising. To our knowledge, such an interaction has not been demonstrated in vascular biology; moreover, it contrasts with interactions described in cancer, wherein b-catenin induces p53 stability and activity through a tumour-suppressor ARF-dependent effect 49 , or p53 downregulates b-catenin by promoting its degradation 50 . While b-catenin-mediated repression of p53 may be specific to SMCs, relevance of this mechanism to other biological contexts has yet to be tested, and could be of particular interest for stem cell biology 51 .
How b-catenin suppresses p53 function in SMCs is not fully elucidated, but our findings offer important clues. With b-catenin deficiency, p53 activity was higher, but its expression was unchanged. Since p53 abundance is tightly regulated by several mechanisms, many of which affect the ability of Mdm2 to ubiquitinate p53 (ref. 4), this lack of change in p53 levels suggests that b-catenin does not impinge on this complex system. In addition, b-catenin loss did not decrease expression of Mdm2 or Mmd4; and pharmacologic and genetic inhibition of Mdm2/Mdm4 did not increase p53 activity. Moreover, while b-catenin loss increased p53 acetylation, inhibition of Mdm2/Mdm4 did not. Thus, the mechanism leading to increased p53 activity on b-catenin deletion in vascular SMCs differs from that characterized by p53 accumulation and inhibition of Mdm2/Mdm4 function 4 . Our observations suggest that vascular SMCs in culture do not require Mdm2/Mdm4 function to repress p53 activity, but they do require b-catenin. Consistent with this idea, adult SMC-selective inactivation of Mdm4 has no repercussions, whereas Mdm2 SMC-selective inactivation induces death of intestinal SMCs without evidence of a vascular phenotype 52 ; however, the low level of inactivation in this study limits conclusions about the role of Mdm2 in SMC biology in the adult vasculature. Whether Mdm2 and Mdm4 are required in vascular SMCs in vivo in particular during development has yet to be fully tested.
Our studies with mutant Ctnnb1 alleles and inhibitors in vascular SMCs suggest that b-catenin N-terminal interactions and b-catenin/TCF interaction are dispensable to repress p53 activity, while C-terminal interactions are required. In particular, b-catenin forms that can interact with CBP also prevent p53 acetylation, a modification required for p53 activation 30 . In addition to CBP, other proteins interact with the b-catenin C-terminus 10 ; while our findings with the ICG001 inhibitor are consistent with disruption of a direct b-catenin-CBP interaction, it is possible that other C-terminal interactors are involved and that the linkage between b-catenin and CBP is indirect. These additional factors participate in chromatin remodelling or connect b-catenin with the transcription initiation or elongation machinery; they are not b-catenin specific, and interact with other coactivators and transcription factors including p53 (refs 53,54). Our studies with the Gal4BD-p53 chimeric protein reveal that b-catenin C-terminal interactions are required to repress p53 activity, independent of possible effects on DNA binding. In this context, it may be that p53 activity depends on recruitment of acetyltransferases and other coactivators necessary for transcription, and that the presence of the b-catenin C-terminus affects availability of acetyltransferases/coactivators in the SMC nucleus; b-catenin loss may increase this pool of acetyltransferases/coactivators and thereby potentiate p53 acetylation and activity. Alternatively, mechanisms involving altered expression of b-catenin-CBP-dependent, but TCF-independent, target genes whose products impinge on p53 acetylation and activity are also possible.
Notably, we found that SMC b-catenin C-terminal interactions are physiologically relevant for artery formation. Indeed, C-terminal interactions are required, while N-terminal interactions dispensable for arterial assembly; this is consistent with our cell culture studies and with the idea that mammals have a less strict requirement for b-catenin N-terminal signalling compared with Drosophila 55 . Whereas Ctnnb1 D164A as the only source of b-catenin protein in SMCs fully restored artery formation, p53 loss in SMbCKO embryos resulted in an important but partial recovery, suggesting that b-catenin C-terminal interactions go beyond modulation of p53 activity. Such p53-independent mechanisms might involve b-catenin/TCF interaction and activation of b-catenin target genes. Although cyclin D1 has been characterized as a direct transcriptional target of b-catenin in studies using promoterreporter constructs 56 , the unexpected increase in cyclin D1 mRNA observed in b-catenin-deficient SMCs is consistent with an indirect linkage between b-catenin and cyclin D1 expression 57 , and may be due in part to the higher fraction of b-catenindeficient cells in the G0/G1 phase of the cell cycle, during which SMC cyclin D1 levels are high 58 .
Our studies indicate the significance of SMC b-catenin signalling function in vivo during arterial wall formation; however, relevant upstream regulators of this pathway remain unknown. Wnt ligands, released in an autocrine or paracrine (from endothelial cells or other sources) manner, might signal to SMCs for b-catenin stabilization and nuclear translocation during vascular maturation. This hypothesis remains to be tested.
SMC growth plays a pathogenic role in vascular disease 59 ; our studies suggest that inhibition of b-catenin C-terminal interactions might serve as a therapeutic strategy. In addition, our findings show that SMC b-catenin is essential for the formation of multilayered arterial walls, which may be relevant for therapeutic angiogenesis and tissue engineering 60,61 . Moreover, tumour vasculature forms by angiogenesis and arterio-venogenesis 62 ; our findings raise the possibility that SMC b-catenin contributes to formation and maturation of tumour vasculature by promoting SMC investment.
In conclusion, we have demonstrated that SMC b-catenin is essential for artery formation, in part by acting as a repressor of p53 and thereby promoting cell proliferation and survival. This novel role of b-catenin requires its signalling function, in particular, its C-terminal interactions. In addition to elucidating a critical function for b-catenin in development, these findings have implications for therapeutic angiogenesis, tissue engineering, tumour vascularization and vascular disease.

Methods
Mice. Tagln-Cre 20 , Ctnnb1 flox/flox21 and Trp53 flox/flox mice 63 as well as Ctnnb1 D164A/WT , Ctnnb1 DC/WT and Ctnnb1 dm/WT mice 32 have been described. Tagln-Cre mice (STOCK Tg(Tagln-cre)1Her/J, (C57BL/6 Â SJL)F2) and Ctnnb1 flox/flox (B6.129-Ctnnb1 tm2Kem /KnwJ) mice were obtained from the Jackson Laboratory. Trp53 flox/flox mice in a mixed C57BL/6 Â 129 Â FVB background (provided by Dr Liang Zhu, Albert Einstein College of Medicine) were crossed with Ctnnb1 flox/flox mice for four generations before use in experimental breeding, and the Ctnnb1 D164A/WT , Ctnnb1 DC/WT and Ctnnb1 dm/WT were kept on a C57BL/6 background. In general, test genotypes were compared with littermate controls. Male and female 8-16-week-old mice were used as breeders for timed mating. Embryos from E9.5 to E19.5 and 21-day-old pups, male and female, were studied. Noontime of the day of detecting vaginal plugs was designated as E0.5. Mice and embryos were genotyped using tail or yolk sac DNA, respectively, plus validated PCR protocols (see references for each mouse line). All animals were housed in pathogen-free conditions, the procedures followed the rules and regulations of the AAALAC and were approved by the Institutional Animal Care and Use Committee (IACUC) of Albert Einstein College of Medicine.
Histopathological and morphometric analysis. Embryos from E9.5 to E12.5 were dissected under a Stemi 2000-C stereomicroscope (Zeiss), fixed overnight with 4% formaldehyde in phosphate buffer (Fisher Scientific, SF100-4), dehydrated through an ethanol gradient, treated with xylene and embedded longitudinally in paraffin. Cross-sections (5 mm) were stained with hematoxylin and eosin. Hematoxylin and eosin-stained cross-sections of hearts and arteries were photographed with a digital microscope (COOLSCOPE, Nikon). Morphometric analysis of PDAs and FDAs, heart and OT was performed using ImageJ software. Total cross-sectional and vessel lumen areas were measured, with the distance between the total area edge and lumen edge indicating wall thickness. The total area of future left and right ventricles were measured, as well as the area of compacted and trabeculated myocardium in each ventricle. The ratio of myocardium area to total area was also calculated. The length of OT and OT wall thickness were also measured, and the latter was expressed as a fraction of the whole OT cross-section.
Immunohistochemistry. Tissue sections were deparaffinized and rehydrated, endogenous peroxidase activity neutralized and antigens retrieved by boiling in sodium citrate solution (Vector Labs H-3300). Sections were blocked in 2% bovine serum albumin (BSA)/10% normal serum (same species of secondary antibody) and avidin-blocking solution (Vector) in phosphate-buffered saline (PBS), incubated with primary antibody in 2% BSA/10% normal serum and biotin-blocking solution (Vector) in PBS, incubated with biotinylated secondary antibody 1:500 (Vector), incubated with ABC (Vector PK-6100) or ABC-AP reagent (Vector AK-5000), incubated with DAB/substrate/chromogen system (Dako K3467; brown) or AP substrate kit I (Vector SK-5100 reagent, red) and counterstained with hematoxylin (Vector H-3404). Specificity of staining was confirmed by omission of the primary antibody. Images were obtained using a digital microscope ( Mouse aortic SMCs. Primary SMCs were obtained from 5 to 6-week-old Ctnnb1 flox/flox and Ctnnb1 WT/WT mouse aortas using collagenase-elastase digestion 64 as follows: aortas were collected from the aortic arch to the iliac bifurcation, rinsed with cold serum-free DMEM with antibiotics, moved to the cell culture hood, stripped of adventitia, minced with scissors and digested with Collagenase 0.5 mg ml À 1 (Sigma type I, C-0130) and Elastase 0.125 mg ml À 1 (Sigma type III, E-0127) in serum-free DMEM with antibiotics at 37°C until most cells were in suspension; then, enzymes were inactivated by adding DMEM with serum, and the cell suspension was recovered and centrifuged at 400g for 5 min, the SMC pellet was washed with complete medium, and SMCs plated in Fibronectin-coated dishes. Mouse aortic SMCs were maintained in DMEM (Invitrogen) containing 20% fetal bovine serum (FBS, HyClone), 100 U ml À 1 penicillin, 100 mg ml À 1 streptomycin and 1% L-glutamine, and subcultured weekly. Passages 2 or 3 were used for transduction with GFP-or Cre-expressing adenovirus (Ad5.CMV-GFP or Ad5.CMV-Cre) to obtain control or cells lacking b-catenin, respectively, and passages 4 to 8 were used for experiments. To evaluate cell population growth, 0.2 Â 10 5 cells of each group were plated in 35 mm dishes in triplicate for every time point, allowed to seed overnight, serum-starved for 48 h and then stimulated with and maintained in 20% FBS, and counted every 2 days using trypan blue exclusion and a hemocytometer. To evaluate cell population decline, 3 Â 10 5 cells of each group were plated in 35 mm dishes in triplicate for every time point, allowed to seed overnight, maintained under serum starvation and counted every day. polyacrylamide gel electrophoresis and blotted onto PVDF membranes (Immobilon-P, Millipore). After blocking in TBST (Tris pH 8.0, NaCl 150 mM, 0.1% Tween 20) plus 4% (w/v) non-fat milk or BSA, blots were incubated at 4°C overnight with primary antibodies. Signals were detected with horseradish peroxidase-conjugated secondary antibody and chemiluminescence (ECL, GE Healtcare). Equivalent protein loading was tested with anti-GAPDH (Santa Cruz, sc-25778, 1:5,000 dilution). Primary antibodies used: anti-b-catenin (Santa Cruz sc-7963, 1:500 dilution), anti-a-SMA (Sigma A2547, 1:5,000 dilution), anti-Sm22a (Abcam 14106, 1:10,000 dilution), anti-p53 (Santa Cruz sc-6243, 1:500 dilution), anti-MDM2 (Santa Cruz sc-965, 1:200 dilution), anti-MDMX (Santa Cruz sc-74468, 1:200 dilution), anti-Lamin A/C (Santa Cruz sc-6215, 1:200 dilution) and anti-CBP (Santa Cruz sc-369, 1:100 dilution). All uncropped blots can be found in Supplementary Fig. 8.
Cell cycle analysis. SMCs (1 Â 10 6 cells) were incubated with 20 mM BrdU for 1 h, collected and fixed with ice-cold 70% ethanol while vortexing, then washed and incubated with 2N HCl/0.5% Triton X-100, resuspended in 0.1 M sodium tetraborate and resuspended in 0.5% Tween 20/1% BSA/PBS. In all, 20 ml of Anti-BrdU-FITC (Cat. No. 347583) were added and cells incubated for 30 min at room temperature, then cells were washed and resuspended in 0.5 ml of PBS containing 10 mg ml À 1 RNase A and 10 mg ml À 1 of propidium iodide for fluorescence-activated cell-sorting analysis in the Flow Cytometry Core Facility at Albert Einstein College of Medicine. Data were analysed with FlowJo 8.7 (FlowJo, LLC).
Survival assay. SMCs (5 Â 10 4 ) were plated on a 22 Â 22 mm coverslip in triplicate for each condition, allowed to become a subconfluent monolayer and then treated with 20 or 0% FBS for 48 h. Dead and live cells were distinguished by adding 150 ml of combined LIVE/DEAD assay reagents (2 mM Calcein AM and 4 mM ethidium homodimer-1) (L-3224, Life Techologies) to each coverslip. The percentage of dead cells for each coverslip was determined as the mean of (100 Â dead/(live þ dead)) of 16 randomly selected epifluorescent photomicrographic fields (Axio Observer.Z1 microscope, Zeiss). ImageJ software was used to count cells.
Transwell migration assay. A 6.5 mm Transwell system with 8.0 mm pore polyester membrane insert (Corning 3464) was used. SMCs were serum-starved for 24 h, collected and plated at 5 Â 10 4 cells per insert (upper chamber) in DMEM, as described before, containing either 0.2% or 20% FBS. The inserts were place into wells (lower chamber) containing either 0.2% or 20% FBS. After 8 h, cells remaining on top of the membrane were removed and cells that reached the bottom fixed and stained with Hema3 (Stat Pack Cat #123-869). The number of cells per high-power field ( Â 200) was counted in 10 random fields per membrane. Triplicates for every genotype and condition were used every time.
Acetylation of p53. Evaluation of p53 acetylation in SMCs was done by a PathScan Acetylated p53 Sandwich ELISA kit (Cell Signaling 7236C), following manufacturer instructions. Briefly, total cell lysates were obtained, sonicated and protein concentration measured. Then, 100 ml of each sample containing equal amount of protein was added to a microwell coated with a p53 mouse monoclonal antibody and incubated at 4°C overnight. The microwells were extensively washed, and an acetylated-lysine rabbit monoclonal antibody added for 1 h at 37°C. After washing, an anti-rabbit horseradish peroxidase-linked antibody was added for 0.5 h at 37°C. After washing, 100 ml of TMB substrate was added for 10 min at 37°C. Then, 100 ml of STOP solution was added and absorbance at 450 nm read in a Synergy 2 Microplate Reader (BioTek). The magnitude of the absorbance is proportional to the quantity of acetylated p53. Triplicates for every genotype and condition were used every time. When indicated, cells were treated with 10 mM RO5963 (Calbiochem 44153), 50 mM of ICG001 (SelleckBio S2662) or vehicle (DMSO); or transfected with CBP siRNA (Santa Cruz sc-29243) or control siRNA (Santa Cruz sc-44231) using X-tremGENE siRNA transfection reagent (Roche 04476093001).
Isolation of nuclear fraction. Control mouse aortic SMCs were collected by trypsinization, washed with ice-cold PBS and collected by gentle centrifugation. Isolation of cellular fractions was done using the Qproteome nuclear protein kit (QIAGEN 37582), following manufacturer's instructions. Briefly, 5 Â 10 6 SMCs were resuspended in hypotonic Lysis buffer NL. Then detergent was added and cells were vortexed to rupture the plasma membrane. Centrifugation was used to separate the cytosolic fraction (supernatant) from nuclei (pellet). The nuclear pellet was washed and incubated in a high-salt buffer. Centrifugation was used to separate the nucleic-acid-binding proteins (supernatant) from the nuclear debris (including genomic DNA). Extraction of the 'insoluble' nuclear proteins was done by incubating nuclear debris with a buffer containing Benzonase, which digests DNA and releases proteins intimately associated with DNA.
Statistics. Comparisons between two groups were evaluated by two-tailed t-test, and between more than two groups by one-way or two-way ANOVA (analysis of variance) followed by post hoc test when appropriate. Analysis of phenotypic categories and comparison of observed versus expected genotype frequencies was done by w 2 -test or Fisher's exact test. Significance was accepted for values of Po0.05. All in vitro experiments were replicated at least three independent times. For developmental animal studies, no statistical method was used to predetermine sample size. In general, breeding strategies were designed to produce embryos with the genotypes of interest and controls within the same litter-thus littermates were compared-and no randomization was necessary to allocate animals in experimental groups. No embryos were excluded from the analysis, and investigators were not blinded.
Data availability. The authors declare that all data supporting the findings of this study are available within the article and its Supplementary Information files or from the corresponding author on reasonable request.