Redundant Roles of Tead1 and Tead2 in Notochord Development and the Regulation of Cell Proliferation and Survival^▿

Mouse strains.

Tead1 and Tead2 mutant mice (accession numbers CDB0410K and CDB0404K, respectively) were generated as follows. The C57BL/6 mouse bacterial artificial chromosome genomic DNA clones RPCI24-359I13 and RPCI24-372P9, which contain the translation initiation sites of the Tead1 and Tead2 genes, respectively, were obtained from BACPAC Resources, Children's Hospital Oakland Research Institute, and were used for the isolation of genomic DNA fragments. To construct the targeting vectors, we first generated MC1-DT-A-pA/pMW118 by cloning a cassette consisting of the MC1 promoter, diphtheria toxin A (DT-A) gene fragment, and the poly(A) signal of the Pgk gene (34) into the KpnI-HindIII sites of a low-copy-number plasmid, pMW118 (Nippon Gene, Japan). The Tead1 targeting vector was constructed by cloning a 15-kb Eco47III-SalI fragment containing the second exon of the Tead1 gene into MC1-DT-A-pA/pMW118, followed by the replacement of an internal NaeI-AgeI fragment containing the second exon with a Pgk gene promoter-neo-poly(A) signal cassette flanked by loxP sites and a splice acceptor (SA) signal of En2-IRES-β-geo-poly(A) (SA-IRES-β-geo-pA, where IRES is internal ribosome entry site) cassette. The Tead2 targeting vector was similarly constructed by replacing the second and third exons with SA-IRES-β-geo-pA between a 6.5-kb AflIII-NruI fragment containing the first exon and a 6-kb AflIII-AflIII fragment containing the fourth exon. TT-2 ES cells (68) were electroporated with linearized targeting vectors, followed by positive and negative selection with G418 and DT-A, respectively. ES clones were first screened for homologous recombination by long PCR using LA-Taq (TaKaRa, Japan), as described previously (43). For the screening of Tead1 and Tead2 loci, the primer pair Tead1 S2 (5′-CGATGCCTGCTTGCCGAATATCATG-3′) and Tead1 R4 (5′-GGCTCCACAAGTCATGTGAAGTC-3) and the pair Tead2 S5 (5′-CTTCCTCGTGCTTTACGGTATC-3′) and Tead2 R10 (5′-CAAGGTTTGCAATGGCTAAGCAG-3′), respectively, were used. The positions of these primers are indicated in Fig. 1A. Homologous recombination of the Tead loci was confirmed for clones that were positive by PCR and by Southern hybridization of the ES genome with 5′ and 3′ external genomic probes, and the absence of random targeting-vector insertion was also confirmed with an internal probe for the lacZ gene that detects β-geo. The resultant ES clones were injected into 8-cell-stage embryos to produce chimeric mice. Chimeric founders from two independent ES cell lines were crossed with C57BL/6 mice, and the resulting mutant mouse lines were maintained on either an inbred C57BL/6 or outbred CD1 (Charles River) background. The mutant phenotypes were not affected by ES cell line origin or genetic background.

For genetic interaction studies, Tead1^+/−; Tead2^+/− mice were crossed with Yap^+/tm1Smil (hereafter referred to as Yap^+/tm1) mice (38), which were kindly provided by E. Morin-Kensicki and S. L. Milgram. Mice were housed in environmentally controlled rooms of the Laboratory Animal Housing Facility of the RIKEN Center for Developmental Biology, under the institutional guidelines for animal and recombinant DNA experiments.

Genotype determination by PCR.

The genotypes of adults and embryos were identified by PCR of DNAs prepared from ear punches and yolk sacs, respectively. A mixture of the following three primers was used: for examination of the Tead1 locus: Tead1-F1 (5′-CAGCCATATCACATCTGTAGAGG-3′), Tead1-F2 (5′-CTTTCTGTGGCACTGAGCCTCTCAG-3′), and Tead1-R2 (5′-ACTCTCCTTCTGCCTTAGACTCCTG-3′); for the Tead2 locus, Tead2-S2 (5′-CGATGCCTGCTTGCCGATATCATG-3′), Tead2-F2 (5′-ACATAGCTAGAACACGGCTGAGCTG-3′), and Tead2-R2 (5′-GAGCACTTAGCACCAAGCCTAGTTC-3′). The locations of the primers are indicated in Fig. 1A. The PCR conditions were as follows: 94°C for 3 min and 30 cycles of 94°C for 30 s, 59°C for 30 s, and 72°C for 1 min 30 s, followed by 72°C for 5 min. Genotype determination of the Yap^tm1 allele was also performed by PCR under conditions described previously (38).

Reverse transcription-PCR (RT-PCR).

The total RNAs were prepared from E8.5 wild-type and Tead1^−/−; Tead2^−/− mutant embryos using an RNeasy Mini kit (Qiagen) and were used in a reverse transcription reaction with Ready-to-Go You-Prime First-Strand Beads (Amersham Biosciences) according to the manufacturer's instructions. The resulting cDNA pools were used for PCR amplification of the Tead1, Tead2, and β-actin cDNA fragments. The primers for each gene were as follows: Tead1-1, 5′-TACTGCCATCCACAACAAGC-3′ and 5′-TGCTGCACAAAGGGCTTGAC-3′; Tead1-2, 5′-CTCCGCTTTCCTTGAACAGC-3′ and 5′-GTCCACAGATTCGAGCAACG-3′; Tead2-1, 5′-CCTGATGCAGAAGGTGTGTG-3′ and5′-ACCTGAATATGGCTGGAGAC-3′; Tead2-2, 5′-GAGAAATTCAGTCCAAGCTG-3′ and 5′-TCTGGAACATTCCATGGTGG-3′; Tead2-3, 5′-GCCACCATGGAATGTTCCAG-3′ and 5′-GAGAACTCTGTCAGCTGCAG-3′; Tead2-4, 5′-TGAGCAGCCAGTATGAGAGC-3′ and 5′-CAGCAGACGGTACACAAAGC-3′; and β-actin, 5′-TGTATGCCTCTGGTCGTACCACAG-3′ and 5′-GATGTCACGCACGATTTCCCTCTC-3′.

The primers for Tead1-2, Tead2-4, and β-actin were originally described in Nishioka et al. (44). The PCR conditions were as follows: 94°C for 3 min and 30 cycles of 94°C for 30 s, 57°C for 30 s, and 72°C for 1 min 30 s, followed by 72°C for 10 min.

In situ hybridization.

Mouse embryos were staged by morphology (14). In situ hybridization was performed according to standard procedures (47, 61) with the following modifications. The sodium dodecyl sulfate in the hybridization buffer was replaced by 3-[(3-cholamidopropyl)dimethylammonio]-1-propanesulfonate (CHAPS), and the antibody reaction was performed in maleic acid buffer (100 mM maleic acid, 150 mM NaCl [pH 7.5]) instead of phosphate-buffered saline. The following probes were used to detect specific genes: Sox2 (a gift from R. Lovell-Badge); Otx2 (40); Krox20 (51); Hoxb1 (a gift from H. Koseki); Hoxb9 (a gift from K.-I. Yamamura); Afp and Brachyury (gifts from A. Shimono); Foxa1, Foxa2, and Foxc1 (47); Sox17 (28); Meox1 (6); Noto (1); α-globin (7); and Yap (49).

Analysis of cell proliferation and apoptosis.

Mitotic cells were detected by a standard immunohistochemical procedure using an anti-phospho-histone H3 rabbit polyclonal antibody (1:100 dilution) (Upstate) as the primary antibody. Apoptotic cells were detected by terminal deoxynucleotidyltransferase-mediated dUTP-biotin nick end labeling (TUNEL) using an Apoptosis Detection Kit (Chemicon) according to the manufacturer's instructions. Nuclei were detected by light counterstaining with hematoxylin. The numbers of total and phospho-H3- or TUNEL-positive cells were counted for two or three sections, respectively, per embryo. Mitotic and apoptotic indices were analyzed with Prism5 statistical software (GraphPad) using an unpaired, one-tailed t test and applying Welch's correction when appropriate.

Generation of Tead1^−/− and Tead2^−/− mutant mice.

We previously identified the TEAD family transcription factors as key regulators of the Foxa2 enhancer in the node and notochord (49). However, the in vitro and transgenic approaches we used did not reveal the roles of individual Tead genes in Foxa2 expression or in embryonic development. Therefore, to clarify the roles of individual Tead genes, we generated Tead mutant mice. Because the expression level of Tead2 during embryogenesis is much higher than that of the other Tead genes (29, 49), we first focused on the Tead2 gene. We disrupted Tead2 by replacing the second and third exons, which contain the translation start site and half of the TEA DNA-binding domain, with a cassette consisting of a SA-IRES-β-geo-pA signal (Fig. 1A). In mice homozygous for this allele, we detected no Tead2 transcripts containing the deleted exons by RT-PCR with probe Tead2-1, which amplified exons 2 to 4 (Fig. 1B). Unexpectedly, however, we did detect Tead2 transcripts by RT-PCR using probes amplifying some downstream exons (Tead2-2, exons 4 to 6; Tead2-3, exons 6 to 8), but transcripts of the exons further downstream (Tead2-4, exons 10 to 11) were not detected (Fig. 1B). Therefore, any protein produced from the truncated Tead2 transcripts would lack part of the DNA-binding and coactivator-interaction domains and should be nonfunctional; therefore, the Tead2 mutant allele should be a null allele. The genotypes of the embryos were routinely identified by PCR of the yolk sac DNAs (Fig. 1C). Tead2^−/− mice exhibited no overt abnormalities and were fertile, indicating that Tead2 is dispensable for development and growth (data not shown).

We hypothesized that Tead2 and the other Tead genes are functionally redundant and, therefore, that other Tead genes compensated for the absence of Tead2. Thus, to reveal a potentially masked Tead2 function, we generated compound mutants of Tead1 and Tead2. We selected Tead1 because a homozygous gene trap mutation of Tead1 is embryonic lethal around E11.5 (9) and because Tead1 and Tead2 show widespread overlapping expression between E7.5 and E9.0, except in the heart, where only Tead1 is expressed (49). Therefore, if Tead1 and Tead2 had any redundant functions in early development, a further reduction of the Tead gene dosages would exacerbate the mutant phenotype. To generate the Tead1 mutant mouse, the Tead1 gene was disrupted by replacing the second exon, which contains the translation initiation codon, with the PGK-neo-pA cassette and the SA-IRES-β-geo cassette described above (Fig. 1A). The Tead1^−/− mutant mice died around E11.5. The major defect was a reduction of myocardium trabeculation (data not shown), as previously reported for the gene-trap mutant (9). Some of our mutants died even earlier, perhaps because of the difference in the genetic backgrounds of the mutant mice (data not shown). We confirmed the proper homologous recombination of the Tead loci and the absence of randomly inserted targeting vectors in the mutant ES cells by Southern hybridization (data not shown). We also confirmed the absence of Tead1 transcripts by RT-PCR analyses using primer sets that amplify two regions of the downstream exons (Tead1-1, exons 7 to 9; Tead1-2, exons 10 to 11) in embryos that were homozygous mutants for this allele (Fig. 1B) and genotyped the embryos by PCR using yolk sac DNA (Fig. 1C).

To generate the Tead1^−/−; Tead2^−/− double mutant mice, we first generated Tead1^+/−; Tead2^+/− mice by crossing Tead1^+/− animals and Tead2^−/− animals. Since both the Tead1 and Tead2 genes are on the same chromosome (Chr7), the resulting Tead1^+/−; Tead2^+/− mice should have the targeting alleles on separate chromosomes (trans-heterozygous). The Tead1^+/−; Tead2^+/− mice did not exhibit any obvious abnormalities. To obtain cis-heterozygous animals, which have both targeted alleles on the same chromosome following a meiotic crossover event, the trans-heterozygous mice were crossed to wild-type mice. Of the progeny obtained with this cross, 14.6% were Tead1^+/−; Tead2^+/− cis-heterozygotes. We then intercrossed the cis-heterozygous Tead1^+/−; Tead2^+/− animals to obtain Tead1^−/−; Tead2^−/− double homozygous mutant mice. As anticipated from the observed Tead1^−/− embryonic lethality, we failed to obtain Tead1^−/−; Tead2^+/+, Tead1^−/−; Tead2^+/−, or Tead1^−/−; Tead2^−/− mice at weaning. Notably, we also did not identify any Tead1^+/−; Tead2^−/− mice at the weaning age, suggesting that this genotype is either embryonic lethal or early postembryonic lethal. Approximately half of the Tead1^+/−; Tead2^−/− embryos showed an anterior neural tube closure defect or exencephaly between E9.5 and E15.5 (see Fig. 8C and D and data not shown). Morphologically abnormal Tead1^−/−; Tead2^−/− embryos were consistently found at E9.5 but were not recovered after E12.5 (data not shown). To elucidate the events leading to the severe morphological abnormalities of the Tead1^−/−; Tead2^−/− mutant embryos, we next examined the abnormalities of the double homozygous mutant embryos in detail.

Tead1^−/−; Tead2^−/− embryos showed severe morphological defects at E8.5.

At E6.5, the Tead1^−/−; Tead2^−/− embryos were morphologically indistinguishable from sibling control embryos (data not shown). At E7.5, the mutant embryos were slightly smaller than their siblings but appeared normal histologically (Fig. 2A and B and data not shown). At E8.5, however, the mutant embryos were small and showed dorso-ventral undulations (Fig. 2C and D). Histological examination of the mutants showed that the anterior region was dorsally folded and there were two heart tubes, indicating that the two heart primordia failed to fuse (Fig. 2F). In the central region, in contrast to the wild-type embryos, the mutant neural plate had not fused along the dorsal ridges to form a neural tube. Instead, the neural plate remained flat, and the boundary between the neuroectoderm and the underlying mesoderm was not clearly demarcated (Fig. 2G). In the posterior region of the embryos, the boundary between the neuroectoderm and presomitic mesoderm was clearer (Fig. 2H). The notochord and somites were morphologically indistinct throughout the length of the body. At E8.75, the mutants failed to undergo embryonic turning, their neural plate was kinked dorso-ventrally, and an abnormal protrusion to the ventral side developed at the posterior end of the mutant embryo (Fig. 2J and K). In this region, abnormal cell accumulation was observed (Fig. 2K). At E9.5, the mutant embryos were much smaller than the control embryos and were highly disorganized morphologically (see Fig. 8A, B, and H). They typically developed a characteristic posterior-ventral protrusion and a bulbous allantois that failed to fuse with the chorion (see Fig. 8H).

Specification and anteroposterior patterning of the neuroectoderm were relatively normal in Tead1^−/−; Tead2^−/− embryos.

To dissect the abnormalities observed in the Tead1^−/−; Tead2^−/− embryos, we examined the expression of various molecular markers at E8.5. In all these experiments, we examined the expression of each marker gene in more than three mutant embryos and found consistent results unless noted. We first examined development of the neuroectoderm. At this stage in control embryos, the entire neural plate/tube expressed Sox2. In the Tead1^−/−; Tead2^−/− embryos, Sox2 expression was observed along the entire body axis, indicating that specification of the neuroectoderm occurred normally (Fig. 3A and B). To examine the anteroposterior patterning of the neural plate, we next analyzed the expression of the following regionally restricted genes: Otx2 (forebrain and midbrain) (52), Krox20 (third rhombomere [r3] and r5) (51), Hoxb1 (r4 and posterior to the thoracic level) (62), and Hoxb9 (posterior to the thoracic level) (5). In mutant embryos, all of these markers were expressed in the proper regionally restricted manner in the same anteroposterior order as in the control embryos (Fig. 3C to J). These results suggested that, irrespective of the gross morphological abnormalities, specification of the neural plate and its patterning along the antero-posterior axis took place relatively normally in the Tead1^−/−; Tead2^−/− embryos.

Endoderm development in the Tead1^−/−; Tead2^−/− embryo was relatively normal except for the posterior definitive endoderm.

We next examined endoderm development in the Tead1^−/−; Tead2^−/− embryos. At this stage, the endoderm is divided into two types, the visceral endoderm and the definitive endoderm. In normal development, the visceral endoderm, which transiently covers the embryonic region of the embryo, is displaced by the definitive endoderm and forms the visceral yolk sac by associating with the extraembryonic mesoderm. Visceral endoderm development was characterized by examining the expression of Alpha-fetoprotein (Afp) at E8.5 (15) (Fig. 4A). In mutant embryos, Afp expression was clearly observed in the extraembryonic area, suggesting that the visceral endoderm was developing normally (Fig. 4B). As a marker of definitive endoderm development, we used Foxa1, which is normally expressed widely in the definitive endoderm except for the posterior region surrounding the node and primitive streak (Fig. 4C) (47). This expression pattern was maintained in the Tead1^−/−; Tead2^−/− embryos, although the boundary between the Foxa1-positive and -negative regions was more evident in the mutants (Fig. 4D). Supporting the idea of normal endoderm development in the mutant, Foxa2 protein expression was observed in all the endoderm cells of the mutant embryos, as in the controls (data not shown) (47). In contrast, the expression of Sox17 in the definitive endoderm, which was normally restricted to the posterior region, was clearly down-regulated in the mutants (n = 3/4) (Fig. 4E and F) (28). Interestingly, the expression of Sox17 was increased (n = 3/4) in the mutant mesoderm-derived allantois, which was expanded laterally (Fig. 4F). These results suggest that development of the visceral and definitive endoderm were relatively normal in the Tead1^−/−; Tead2^−/− embryos, although development of the posterior definitive endoderm was partially compromised.

Paraxial mesoderm and lateral plate mesoderm were displaced toward the lateral margins in Tead1^−/−; Tead2^−/− embryos.

Histologically, the development of the paraxial mesoderm appeared most severely affected in the Tead1^−/−; Tead2^−/− embryos. No morphologically identifiable somites were formed, and in the central portion of the body, no paraxial mesoderm could be recognized on the ventral side of the neural plate (Fig. 2F to H). Therefore, we first examined the development of the paraxial mesoderm, which is characterized by the expression of Foxc1 (formerly known as MF1) (47). Foxc1 is also expressed in the head mesenchyme. Foxc1 expression in the Tead1^−/−; Tead2^−/− embryos was almost normal at E7.5 (Fig. 5A to D), but Foxc1-positive cells were displaced laterally as development proceeded, and it became localized to the position corresponding to the lateral plate or the extraembryonic mesoderm of normal embryos after E8.0 (Fig. 5E to G and data not shown). Foxc1 expression at the posterior end of the embryos retained a more medial position up to E8.75 (Fig. 5F and J) Although the segmented somites were not morphologically identifiable, the expression of Meox1 (formerly known as Mox1), which is specific to the somites (6) (Fig. 5I), was also observed in laterally displaced domains (Fig. 5K and data not shown). Interestingly, despite the absence of morphologically segmented somites, the expression of Foxc1 and Meox1 in the posterior region of the mutant embryos sometimes formed weak stripes (Fig. 5F, G, and K). The lateral plate mesoderm revealed by Foxf1 expression (38) was also displaced to the lateral margins in the mutants (data not shown). Taken together, these results suggest that, in terms of cell differentiation, development of the paraxial mesoderm (including head mesenchyme and somites) and lateral plate mesoderm occurred normally in the mutants, but they were displaced toward the lateral margins.

The notochord was not maintained in Tead1^−/−; Tead2^−/− embryos.

We previously identified TEAD family proteins as key regulators of the Foxa2 enhancer, which drives the expression of Foxa2 in the node and the notochord (49). Between E6.5 and E8.25, Foxa2 is expressed in the axial tissues, which include the gastrula organizer, the node, the notochord, and the presumptive floor plate of the neural tube/plate (47); and we observed similar expression in the Tead1^−/−; Tead2^−/− embryos (Fig. 6A and B and data not shown). However, by E8.5, Foxa2 expression in the mutant became discontinuous (Fig. 6C), and it disappeared except at the anterior tip of the embryo by E8.75 (Fig. 6D). The anterior expression domain may correspond to the prechordal plate and/or the ventral midline of the anterior-most neural plate.

As anticipated by the essential role of Foxa2 in node and notochord development (4, 60), Brachyury expression in the node and notochord (23) of the mutant embryos was normal at E8.25 but became discontinuous by E8.5 (Fig. 6F and G). At E8.75, Brachyury expression in the notochord was lost or nearly absent (n = 9/9), as was its expression in the node/posterior end of the notochord (n = 6/9), although this expression in some embryos was not significantly affected (n = 3/9) (Fig. 6H and data not shown). Defects in the notochordal development were further evidenced by a discontinuous and reduced expression of Noto, which is normally restricted to the posterior notochord (1), at E8.5 (data not shown). Brachyury expression in the primitive streak was maintained, indicating that the development of specifically the axial mesoderm was disturbed (Fig. 6G and H). That the development of the primitive streak was normal was further supported by its relatively normal expression of Fgf8, Wnt3a, and Tbx6 (8, 56, 69) in the Tead1^−/−; Tead2^−/− embryos (data not shown). Taken together, although the Tead1^−/−; Tead2^−/− embryos were clearly morphologically abnormal, major cell types in these embryos formed and were patterned in a relatively normal manner. In contrast, notochord development was clearly defective.

Tead1^−/−; Tead2^−/− embryos showed extraembryonic defects similar to those of Yap mutant mice.

Previous in vitro studies identified the YAP protein as a coactivator of TEAD family transcriptional factors (58). A Yap^{tm1Smil/tm1 Smil} (hereafter referred to as Yap^tm/tm1) mutant mouse was recently reported by Morin-Kensicki et al., and the abnormalities observed in the Yap mutant embryos resemble those of the Tead1^−/−; Tead2^−/− mutants (42). Because Yap^tm1/tm1 mutant embryos have defects in yolk sac vasculogenesis, we next examined whether the Tead1^−/−; Tead2^−/− mutants exhibited similar extraembryonic defects. Staining with an antibody to platelet endothelial cell adhesion molecule 1 (PECAM-1), which marks endothelial cells, revealed that endothelial cells differentiated but failed to organize into vessels (Fig. 7A and B). In addition, expression of the α-globin gene indicated the existence of erythroblasts in the disorganized yolk sac of the Tead1^−/−; Tead2^−/− embryos (Fig. 7C and D). Therefore, in the mutant yolk sac, specification of both components of the early yolk sac vasculature (endothelia and erythroblasts) occurred, but their development into an organized primitive vascular plexus was compromised. These features of the Tead1^−/−; Tead2^−/− yolk sac resembled those of the Yap^tm1/tm1 embryo. However, these phenotypic similarities were not caused by the reduced expression of Yap in Tead1^−/−; Tead2^−/− mutants because the Yap expression level in the Tead1^−/−; Tead2^−/− embryos was comparable to that in control embryos, as revealed by whole-mount in situ hybridization (Fig. 7E and F). Taken together, the similarities in both the embryonic and yolk sac phenotypes of the Tead1^−/−; Tead2^−/− and Yap^tm1/tm1 mutant embryos support the hypothesis that the TEAD1 and TEAD2 proteins use YAP as a coactivator protein in exerting their transcriptional activator function during embryogenesis.

Tead1 and Tead2 genetically interacted with Yap.

If TEAD1/TEAD2 functions by forming a complex with YAP during development, the simultaneous reduction of TEAD1/TEAD2 and YAP should reduce the amount of this complex and cause similar developmental defects as observed with the Tead1 and Tead2 compound mutants. Therefore, we generated the compound mutants of Tead1; Tead2; Yap and compared their phenotypes with those of the Tead1; Tead2 mutants at E9.5. Because we obtained Tead1^+/−; Tead2^+/−; Yap^+/tm1 male mice that showed no obvious abnormalities, we crossed these mice to Tead1^+/−; Tead2^+/− females to obtain the compound mutant embryos. Approximately half of the Tead1^+/−; Tead2^−/− embryos (n = 3/6) showed anterior neuropore closure defects (Fig. 8A to D). The introduction of a heterozygous mutation of Yap in this genetic context (Tead1^+/−; Tead2^−/−; Yap^+/tm1) caused additional abnormalities including smaller posterior tissues and defective embryonic turning (n = 2/2) (Fig. 8E). The Tead1^−/−; Tead2^+/− embryos were clearly developmentally delayed but resembled normal E8.5 embryos (n = 8/8) (Fig. 8F). The introduction of a heterozygous mutation of Yap in this genetic context (Tead1^−/−; Tead2^+/−; Yap^+/tm1) resulted in severe morphological defects that were essentially the same as those of the Tead1^−/−; Tead2^−/− embryos (n = 3/4) (Fig. 8G). Embryos with both of these genotypes developed a posterior-ventral protrusion and a bulbous allantois (Fig. 8G and H). Introduction of the heterozygous Yap mutation to the Tead1^−/−; Tead2^−/− mutants (Tead1^−/−; Tead2^−/−; Yap^+/tm1) did not significantly alter the phenotype (n = 7/7) (Fig. 8I). These results clearly demonstrated a genetic interaction between Tead1 and Tead2 and Yap.

Foxa2 expression in the notochord is directly regulated by TEAD proteins (49), and its expression was down-regulated in the Tead1^−/−; Tead2^−/− embryos (Fig. 6). Therefore, if TEAD1/TEAD2 uses YAP as a major coactivator, the Yap^tm1/tm1 embryos should also show reduced Foxa2 expression. Indeed, as anticipated by the discontinuous Brachyury expression in the notochord of Yap^tm1/tm1 embryos at E9.5 (42), the Foxa2 expression in the Yap^tm1/tm1 embryos between E8.75 and E9.5 was either strongly down-regulated except for the anterior tip (n = 2/7) (Fig. 8J) or discontinuous and gradually down-regulated in the posterior region (n = 5/7) (Fig. 8K). Taken together, the genetic interactions as well as the similarities of the mutant phenotypes support the hypothesis that YAP is the major coactivator protein of TEAD1 and TEAD2 in mouse embryogenesis.

Cell proliferation decreased, and the incidence of apoptosis increased in Tead1^−/−; Tead2^−/− embryos.

Yap was recently shown to have oncogenic activities (45, 72). Moreover, consistent with a role in growth regulation, Yap^tm1/tm1 embryos are smaller than sibling embryos by E9.5 (42). Because of the phenotypic similarities of the Tead1^−/−; Tead2^−/− and Yap^tm1/tm1 embryos (42) and because of the genetic interaction between Tead1 and Tead2 and Yap, we speculated that Tead genes are also involved in growth regulation. Because the growth retardation of later-stage mutant embryos could be secondarily caused by insufficient circulation owing to the defects in the heart and yolk sac, we analyzed embryos at E7.75 (early to late head-fold stage) when embryonic growth does not yet depend on circulation and when the abnormalities of the mutant embryos are not yet morphologically evident.

To learn if Tead1 and Tead2 play a direct role in embryonic growth, we examined the frequency of both mitotic and apoptotic cells in the mutant embryos. We found that the frequency of mitotic cells detected by anti-phosphorylated histone H3 was significantly reduced in the Tead1^−/−; Tead2^−/− mutants (Fig. 9A, B, and E) (4.0% to 2.8%, n = 5) with a tendency to decrease in all the germ layers; the difference reached significance in the mesoderm. In addition, TUNEL labeling revealed an increased frequency of apoptotic cells in the mutants (Fig. 7C, D, and F) (0.66% to 1.7%, n = 5). The tendency toward increased apoptosis was observed in all the germ layers and reached significance in the ectoderm. These results suggest that Tead1 and Tead2 directly promote cell proliferation and survival.

Tead1 and Tead2 are functionally redundant.

Mice and humans have four Tead genes, the encoded proteins of which are highly homologous, especially in their TEA domain (DNA-binding domain) and C-terminal coactivator-binding domain (26, 30). Because they have similar activities in vitro and their expressions widely overlap during embryogenesis (26, 49, 58), it is likely that the Tead genes play redundant roles in embryogenesis. However, this possibility has not been investigated. By generating mouse mutants for the Tead1 and Tead2 genes, we have demonstrated here that Tead2^−/− mutant mice are apparently normal while Tead1^−/−; Tead2^−/− double mutants show severe and widespread defects. This result clearly indicates that the roles of Tead1 and Tead2 in mouse embryogenesis largely overlap. Although we failed to detect clear abnormalities in the Tead2^−/− embryos between E9.5 to 15.5 (data not shown), Kaneko et al. observed an exencephaly phenotype in a minor population of their Tead2^−/− embryos (31). Because both Kaneko's group and our group deleted exons 2 and 3 to disrupt the Tead2 locus, the reasons for the phenotypic differences are currently unknown. However, it is interesting that approximately half of the Tead1^+/−; Tead2^−/− embryos also showed exencephaly. This similarity in mutant phenotypes between Tead2^−/− and Tead1^+/−; Tead2^−/− embryos also supports the notion that Tead1 and Tead2 play overlapping roles. Because Tead2 is not expressed in the embryonic heart (26, 49), the heart-specific defects of the Tead1^−/− mutants are likely to be attributable to the absence of a functionally redundant Tead2 gene in the heart, rather than to a distinct function of Tead1 in the heart. In support of this hypothesis, the heart defect observed in the Tead1^−/− mutant (11) was a defect in the proliferation of myocardium (H. Sato and H. Sasaki, unpublished data), and defective cell proliferation was a characteristic feature of the Tead1^−/−; Tead2^−/− double mutants. Therefore, Tead1 and Tead2 appear to play essentially the same roles in embryogenesis. In further support of the functional redundancy among Tead genes, Tead4 is dispensable for ES cell differentiation and postimplantation development (44, 67), and Tead3^−/− mice are apparently normal and fertile (A. S. and H. Sasaki, unpublished data). From these observations, we propose that the mammalian TEAD proteins have essentially the same activities in postimplantation development, with their functional differences being attributable to differences in their expression domains.

Tead1 and Tead2 are required for maintenance of notochord development.

In spite of the gross morphological abnormalities of Tead1^−/−; Tead2^−/− mutant embryos at E8.5, the differentiation and patterning of the neuroectoderm and endoderm were relatively normal, while stark abnormalities were observed in the mesoderm. Although cell differentiation occurred, the relative positions of the paraxial mesoderm and lateral plate mesoderm were displaced toward the lateral margins. The underlying mechanism of this phenotype is not known, but a disturbance of the growth coordination between adjacent germ layers may explain such morphological defects. The Tead1^−/−; Tead2^−/− embryos showed a reduction in cell proliferation that was more evident in the mesoderm than in the overlying ectoderm. Because mouse embryos develop in an inside-out manner continuous with the balloon-like yolk sac up to E8.5, it is likely that the rapidly expanding ectodermal sheet in the mutant could not be kept inside the more slowly growing sheets of paraxial and lateral plate mesoderm. As a consequence, the ectodermal sheet may have protruded by penetrating through the gap between the bilaterally located mesodermal sheets, physically displacing them to more lateral positions.

Only the notochord showed a clear developmental defect in the Tead1^−/−; Tead2^−/− embryos. This result is consistent with our previous finding that TEAD proteins are key transcriptional regulators of the Foxa2 enhancer (49). Because Foxa2 is a transcription factor essential for the development of the notochord and the node (4, 60), Tead genes should be required for node/notochord development. In Tead1^−/−; Tead2^−/− embryos, the node and notochord formed initially, but they were not maintained after E8.5. Therefore, of the four TEAD proteins, TEAD1 and TEAD2 appear to be the major regulators of the Foxa2 enhancer after E8.5. The activation of the Foxa2 enhancer at earlier stages might be controlled by other TEAD proteins either exclusively or redundantly with TEAD1 and TEAD2. Alternatively, the earlier expression of Foxa2 could be redundantly regulated by an unidentified enhancer that is independent of Tead. Supporting the latter possibility, only some of the enhancers have been identified for Foxa2, and the enhancer controlling its earlier expression in the gastrula organizer remains to be identified (48). Nevertheless, the abnormal development of the notochord clearly indicates the involvement of Tead1 and Tead2 in this process, and this function of the TEAD proteins is supported by both in vitro and in vivo genetic evidence (49).

YAP is a major coactivator protein of TEAD1 and TEAD2.

In Drosophila, the TEAD protein Sd uses Vg as a coactivator protein (22, 53). Mice and humans each have four Vg homologues (Vgll1-Vgll4), and their involvement in skeletal muscle development as TEAD coactivator proteins has been suggested (10, 37, 59). On the other hand, biochemical studies in mammals identified YAP and its related protein, Wwtr1, as TEAD coactivator proteins (39, 58). However, the major coactivator(s) of the TEAD proteins in vivo is not known. We show here that Tead1^−/−; Tead2^−/− embryos were small at E8.5 and, despite relatively normal antero-posterior patterning, developed a discontinuous notochord and defective yolk sac vasculogenesis. All of these features are in common with the phenotype observed in Yap^tm1/tm1 mutant embryos (42). Furthermore, we also showed that Tead1 and Tead2 genetically interacted with Yap during embryogenesis and that a TEAD target gene, Foxa2, was down-regulated in Yap^tm1/tm1 mutant embryos. Although the Yap-related gene Wwtr1 is also expressed at this stage (42), Wwtr1^−/− mutants develop to term and show glomerulocystic kidney disease postnatally (24). Furthermore, Vgll1^−/−, Vgll3^−/−, and Vgl1l^−/−; Vgll3^−/− mutant mice exhibited no overt abnormalities (H. Sato and H. Sasaki, unpublished data). The similarity of the mutant phenotypes and the genetic interactions strongly support the hypothesis that YAP is the major coactivator protein of TEAD1 and TEAD2 during early embryogenesis. Likewise, these results indicate that YAP functions primarily with TEAD proteins in early development. Although this does not rule out the possible involvement of Wwtr1 or Vgll proteins in the transcriptional activities of other TEAD proteins or of any TEAD proteins at different developmental stages, it is plausible that YAP is a major coactivator protein of all four TEAD proteins throughout development. This hypothesis can be genetically tested in the future by generating mice carrying mutations for the other Tead genes and by testing genetic interactions among the mutants of Tead and other potential coactivators.

Tead1 and Tead2 regulate cell proliferation and survival.

A striking feature of the Tead1^−/−; Tead2^−/− embryos is their small size. Although growth defects could be secondarily caused by insufficient nutrition from circulation defects, our analysis at E7.75, which precedes heart formation, revealed that Tead1 and Tead2 directly regulate cell proliferation and survival. Although the impact of the deletion of Tead1 and Tead2 on proliferation and apoptosis was most evident in the mesoderm and ectoderm, respectively, the reason for such variation among germ layers is currently unknown. In further support of the idea that cell proliferation is controlled by TEAD proteins, we also observed significantly reduced bromodeoxyuridine labeling specifically in the myocardium of E9.5 Tead1^−/− embryos (H. Sato and H. Sasaki, unpublished data). Recently, Yap was identified as an amplified gene in mammary tumors and liver cancers, and it was shown to have transforming properties, including the acceleration of proliferation and suppression of apoptosis (45, 72). Because our study strongly suggests that YAP is a major coactivator protein of TEAD1 and TEAD2, such growth-promoting activities of Yap may well be mediated by the transcriptional activities of TEAD1, TEAD2, and/or other TEAD proteins. Although Yap also has proapoptotic activity, this activity is mediated by p73 (55). Therefore, it is likely that the anti- and proapoptotic functions of Yap are mediated by distinct transcription factors, TEAD and p73, respectively.

The growth-regulatory activities of Yap appear to be evolutionarily conserved because overexpression of the Drosophila Yap homologue Yki also causes increased proliferation and reduced cell death (25). In Drosophila, Yki activity is negatively regulated by the Hippo (Hpo) signaling pathway, in which the Hpo-Sav complex activates the Wts kinase-Mats complex, which in turn inactivates Yki (reviewed in reference 46). The Hpo signaling pathway is likely to be conserved in mammals because (i) mammalian homologues exist for all the components; (ii) the human homologues of Hpo, Wts, and Mats can rescue their corresponding Drosophila mutants (35, 57, 64); and (iii) some of the components have tumor-suppressive activities in mammalian cells (36). In fact, a recent study demonstrated that Hpo-Yap signaling also regulates organ size in mouse (13). However, the transcription factor that acts downstream of Yap/Yki is currently unknown. Consistent with the essential role in cell survival played by Drosophila Sd (17), an intriguing hypothesis is that TEAD/Sd is one of the downstream effectors of the Hpo signaling pathway in regulating cell proliferation and survival.

Conclusion.

Our genetic analyses of the roles of Tead1 and Tead2 indicate that these two TEAD proteins have largely redundant functions in mouse development. The major roles of Tead1 and Tead2 appear to be the promotion of cell proliferation and suppression of cell death, but Tead1 and Tead2 also clearly regulate development of the notochord. The similar mutant phenotypes and evidence of genetic interactions suggest that TEAD1/TEAD2 uses YAP as a major coactivator protein. Whether these features can be generalized to all the TEAD proteins or are specific to TEAD1/TEAD2 awaits the results of ongoing genetic analyses of the requirements for the Yap and Tead genes.