|Home | About | Journals | Submit | Contact Us | Français|
YAP is a multifunctional adapter protein and transcriptional coactivator with several binding partners well described in vitro and in cell culture. To explore in vivo requirements for YAP, we generated mice carrying a targeted disruption of the Yap gene. Homozygosity for the Yaptm1Smil allele (Yap−/−) caused developmental arrest around E8.5. Phenotypic characterization revealed a requirement for YAP in yolk sac vasculogenesis. Yolk sac endothelial and erythrocyte precursors were specified as shown by histology, PECAM1 immunostaining, and alpha globin expression. Nonetheless, development of an organized yolk sac vascular plexus failed in Yap−/− embryos. In striking contrast, vasculogenesis proceeded in both the allantois and the embryo proper. Mutant embryos showed patterned gene expression domains along the anteroposterior neuraxis, midline, and streak/tailbud. Despite this evidence of proper patterning and tissue specification, Yap−/− embryos showed developmental perturbations that included a notably shortened body axis, convoluted anterior neuroepithelium, caudal dysgenesis, and failure of chorioallantoic fusion. These results reveal a vital requirement for YAP in the developmental processes of yolk sac vasculogenesis, chorioallantoic attachment, and embryonic axis elongation.
Yes-associated protein (YAP65; referred to as YAP throughout) is a modular adapter protein first identified as a binding partner for the product of the proto-oncogene c-Yes (57). YAP contains multiple protein interaction domains, including a proline-rich amino terminus, a 14-3-3 binding site (3), WW domains (32, 57), SH3 binding motifs (12, 57), a coiled-coil, and a consensus PDZ binding motif at the extreme COOH terminus (43). YAP mRNA is broadly distributed, and expressed sequence tags have been identified in cDNA libraries from many different species and from many tissues, cell lines, and tumor samples. Although the cellular and subcellular localization of the YAP protein has been less well characterized, it is expressed in multiple cell types and can be localized to both cytoplasmic and nuclear compartments (3, 26, 32, 45, 60, 64, 69). This broad distribution and the modular structure of YAP together suggest multiple cellular functions.
The identification of YAP-interacting proteins has provided insight into potential roles for YAP in cell signaling within both cytosol and nucleus. YAP cytosolic interactions may impact cell signaling pathways by several possible mechanisms. For example, YAP binds the SH3 domain of c-Yes (57) and can also associate with cytoplasmic PDZ proteins via its COOH terminus (43). Therefore, YAP may play a role in the anchoring and/or targeting of c-Yes to position the kinase to respond to specific extracellular cues or to phosphorylate specific cellular substrates. Cytosolic YAP also may modulate growth factor receptor signaling. For example, YAP associates with the inhibitory Smad7 to attenuate transforming growth factor β signaling (14) and may affect signaling via the ErbB-4 receptor (32, 45). In the nucleus, YAP may function as a coregulator, modulating the activity of several transcription factors. In this manner, YAP interacts with RUNX family members (64, 69), which impact hematopoiesis and osteogenesis, as well as TEAD family members (39, 60), which are implicated in muscle cell and neural crest cell differentiation. Furthermore, we recently found that the proline-rich amino terminus of YAP associates in the nucleus with heterogeneous nuclear ribonucleoprotein U (26), a protein involved in mRNA processing and the control of gene expression. Finally, several pieces of data argue that regulated localization of YAP may impact apoptosis and cell cycle progression. Indeed, in the nucleus the YAP WW domain associates directly with the p53 gene family members p73α, p73β, and p63α and enhances the transcription of proapoptotic Bax and Mdm2 reporter constructs and endogenous Bax (55). In addition, phosphorylation by Akt stimulates YAP interaction with cytosolic 14-3-3 and attenuates p73-mediated apoptosis (55). YAP may also associate with p53BP2 (12), a protein known to inhibit the activity of the p53 tumor suppressor. Thus, YAP likely exists in cell type- and compartment-specific protein complexes that define its function throughout development and in the adult organism, and yet specific in vivo requirements for YAP remain undefined.
YAP shows significant similarity to the product of the related gene, Taz (30). With amino acid identity approaching 50%, YAP and TAZ may share common protein partners, and yet distinctions have been described (8, 21, 25, 26, 30, 37, 47) The extent to which these proteins show unique or overlapping function in vivo remains unclear. Thus, the potential for redundancy between these proteins emphasizes the probable complexity of YAP function and highlights the need for understanding in vivo requirements for YAP.
Although an integrated view of YAP protein interactions and function is lacking, the data suggest that YAP is an adaptor protein that modulates multiple signal transduction pathways in many cell types. These pathways have been explored in biochemical assays and in cell culture model systems, but little is known regarding YAP function in the intact organism. To investigate in vivo requirements for YAP, we generated mice carrying a targeted disruption in the Yap gene. The embryonic lethal phenotype of homozygous mutant mice indicates that YAP is essential for embryogenesis and that, in fundamental developmental events, TAZ does not compensate for YAP. These results demonstrate for the first time a critical role for YAP in early embryonic development.
A 507-nucleotide (nt) probe generated by PCR amplification from a mouse lung cDNA library (primer pair 5′-AGTTTCTGTCTCAGTTGGGACG-3′ [nt 13 to 34 of accession X80508] and 5′-CATGCTGTGGAGTGAGAGGCTC-3′ [nt 520 to 500 of X80508]) was used to screen a 129SvEv mouse genomic library packaged in Lambda Gem-ll with 13- to 15-kb inserts. Two of nine plaques subcloned into pBluescript SK (Stratagene) and partially sequenced provided genomic DNA sequence for the Yap targeting vector construct. The targeting vector 5′ arm of homology was generated by using a phage HindIII to mouse HindIII 1.3-kb fragment as genomic sequence upstream of the Yap coding sequence. The targeting vector 3′ arm of homology was generated by using a Yap exon 1 internal XhoI to phage XhoI 7.0-kb fragment that encompassed Yap exon 2 and downstream sequence. Subcloning these into HindIII and XhoI cloning sites of the pOSdupdel6142-SA vector (kindly provided by the UNC Animal Models Core) generated the targeting vector pOSdupdel6142-Yap65. The targeting vector replaces about 1 kb of genomic sequence with a reverse orientation, floxed MC1-neo cassette, and appends a PGK-TK cassette for negative selection.
Vector electroporation into 129SvEv embryonic stem (ES) cells and selection with Geneticin and ganciclovir, followed by PCR screening and Southern blot confirmation identified six ES cell clones with the targeted recombination event. ES cells from two expanded clones were injected into C57BL/6J blastocysts to generate chimeric founders. Founder males from each ES cell line were crossed to C57BL/6J female mice to give Yaptm1Smil mice heterozygous for the targeted allele that then were crossed to initiate Yaptm1Smil lines. We used allele nomenclature according to the guidelines established by the International Committee on Standardized Genetic Nomenclature for Mice and as implemented by the Mouse Genomic Nomenclature Committee. All ES cell work and generation of chimeric animals was conducted in the UNC Animal Models Core Facility according to well-established and approved protocols.
NCBI Build 33.1 of the mouse genome, released on 3 September 2004 (http://www.ncbi.nlm.nih.gov/mapview/map_search.cgi?taxid = 10090), indicated a putative locus, loc434366, identified by the gene recognition program GNOMON and in opposing orientation to the Yap gene that also is perturbed in this allele. The possible existence of a transcribed gene at this locus is supported by in silico identification of a promoter signal on that strand using promoter recognition computer algorithms (09/01/2004, McPromoter, http://genes.mit.edu/McPromoter.html; 09/01/2004, www promoter scan, http://bimas.dcrt.nih.bov/cgi-bin/molbio/proscan; 09/01/2004, Promoter 2.0, http://www.cbs.dtu.dk/services/Promoter/output.php) and by a single expressed sequence tag (accession CB600667) from the kidney cDNA library 12886: NIH_MGC_176. If translated, this gene would produce a highly unusual 95-amino-acid protein consisting largely of glutamic acid and lysine repeats. Nonetheless, to assess the potential impact on our further studies of having also disrupted this locus, we generated PCR primers in predicted loc434366 exons 2 and 3 (sequence below) and examined possible expression by reverse transcription-PCR (RT-PCR). This primer pair generates a 433-nt amplicon from reverse-transcribed RNA sequence but covers 3.2 kb of genomic sequence. We examined expression of loc434366 in a mouse kidney cDNA library (Stratagene) and in wild-type embryos at specific stages ranging from blastocyst (embryonic day 3.5 [E3.5]) stage to late gestation (E18.5) and found expression in kidney but no detectable expression in embryos until after E7.5 (data not shown).
ES cells, mice, and tissue scrapings from paraffin sections were genotyped by either Southern blot or PCR-based methods. We generated a 542-nt Southern probe by PCR amplification from subcloned genomic sequences 5′ to the Yap targeting construct using the primer pair 5′-GCTGCCATTTCAACTTTCTAC-3′ and 5′-CAGCAGTCTATCGCTTTGTG-3′. This probe recognizes an approximately 6-kb fragment from genomic DNA and an approximately 5-kb fragment from the targeted allele after enzymatic digestion with BamHI. To genotype by PCR, we used a primer trio that included a shared 5′ primer from sequence upstream of the Yap gene (YF, 5′-GAAGCTGTGGCACAAAGA-3′), a 3′ primer in sequence deleted in the targeted allele (YR, 5′-ATGCAAAGGCCACACTGT-3′) and a 3′ primer internal to the neo insert (NR, 5′-CGACGTTAACGGTACCAA-3′).
The transmitted Yaptm1Smil allele was maintained on a mixed 129SvEv/C57BL/6J background. Crossing to an outbred (CD1; Charles River) background to obtain larger litters for embryo analyses did not perceptibly alter the phenotype of homozygous mutant embryos. Litters for phenotype characterization were generated by natural matings and staged by assigning the first half-day of embryonic development (E0.5) as noon the day of vaginal plug observation.
Gene expression was assessed in wild-type embryos from blastocyst stage (E3.5) to late prenatal stage (E18.5) by RT-PCR. Positive controls for these reactions were heart or kidney cDNA libraries (Stratagene). Whole-embryo total RNA was extracted by using the TRIzol reagent (Invitrogen) according to the manufacturer's protocol with modifications for small tissue volume in embryos at ≤E10.5 as follows. (i) Embryos were dissected and rinsed several times in ice-cold phosphate-buffered saline and then transferred with a minimal volume to 40 μl of TRIzol. (ii) Embryos were homogenized by pulling through a 22-gauge 1 needle, and then the TRIzol volume was brought to 100 μl. (iii) Finally, after phase separation with the addition of 60 μl of chloroform, RNA was precipitated from the aqueous phase with 50 μl of isopropanol. cDNAs were generated according to manufacturer's protocol with the reverse transcriptase SuperScript II (Invitrogen). To test for expression of specific transcripts, we used PCR primer pairs that spanned at least one intron as follows: Yap, 5′-TCTGCGCAGCCAGTTGCCTA-3′ and 5′-GCTCATGCTGAGGCCGCTGT-3′; Yap2L (32), 5′-CCCTGATGATGTACCACTGCC-3′ and 5′-CCACTGTTAAGAAAGGGATCGG-3′; Taz, 5′-TCCCCAACAACTCCAGAAGAC-3′ and 5′-CAAAGTCCCGAGGTCACCAT-3′; and loc434366, 5′-GAATTGGAGTACAACTGGTTGTGAACAG-3′ and 5′-ACCTAATTGTGGGTATCCTTTCATTTG-3′.
Whole-mount in situ hybridizations on dissected E6.5 to E9.5 embryos were performed essentially as described previously (61). Digoxigenin-labeled riboprobes were generated using the sequence for Fgf8 (kindly provided by Gail Martin), Brachyury (T,kindly provided by Bernhard Herrmann), and α-globin (accession no. BC037630, I.M.A.G.E. clone 497128; Open Biosystems) fragment for sense and antisense probes subcloned into pBluescript II KS (Stratagene) using the PCR primer pairs 5′-AAACCATGGTGCTCTCTGGGGAAGA-3′ and 5′-TCTTGTGTTTCTTCCTACTCAGGCTTTATTC-3′ and Yap (mouse cDNA fragment subcloned into pBluescript II SK) using PCR primer pairs 5′-ACAGCCAGTGGCGTTGTCTCTG-3′ and 5′-CGGAACTATTGGTTGTCATTGTTCTCAATTC-3′.
Whole-mount immunohistochemistry for the PECAM1 antigen followed standard methods. Primary antibody against PECAM1 (Pharmingen) was followed by horseradish peroxidase-conjugated goat anti-rat (mouse absorbed) secondary antibody (Kirkegaard and Perry Laboratories) and visualized by using a nickel-enhanced, peroxidase-based diaminobenzidine reaction.
Paraffin sections (8 to 12 μm) of embryos in utero were hematoxylin and eosin stained by using standard histological methods with fixation in 4% paraformaldehyde to allow subsequent genotyping by PCR from tissue scrapings of unstained sections of each embryo.
Individual embryos obtained from intercross litters dissected at E8.5 and genotyped by PCR from a yolk sac tissue sample were lysed in 20 mM Tris-HCl (pH 7.4), 150 mM NaCl, 0.1% sodium dodecyl sulfate, 1% Triton X-100, 1% deoxycholic acid, and 5 mM EDTA with a cocktail of protease inhibitors. Lysates were cleared by centrifugation at 14,000 × g for 20 min at 4°C. Protein concentrations were determined by using the BCA assay kit (Pierce Chemical Co.), and 20 μg of embryo lysate was resolved by sodium dodecyl sulfate-polyacrylamide gel electrophoresis, transferred to polyvinylidene difluoride membrane, and analyzed by Western blotting with anti-YAP 252. This antibody was generated against the C-terminal portion of human YAP and affinity purified (26). β-Actin served as a loading control.
Although expression of both the Yap and the Taz genes has been characterized in several cell lines and in Northern blots for both human and mouse tissue (30, 32, 45, 60, 64, 69), characterization of transcript distribution in whole tissue (39) and throughout development is lacking. To begin to explore in vivo roles for YAP, we examined expression of these genes in mouse embryos.
We examined developmental expression of Yap and Taz by RT-PCR (Fig. (Fig.1)1) from blastocyst stage (E3.5) to perinatal stages (E18.5). Yap expression was detected at every stage examined. In addition, we examined the expression of a known mouse Yap splice variant in embryos from E7.5 to E18.5. Using the primer pair described previously (32), we demonstrated the expression of two Yap transcripts both with (Yap2L) and without the inclusion of the 48-nt (32) variant exon 6 at all developmental stages examined (data not shown). In contrast, we detected Taz expression at all later stages but not in blastocyst stage embryos (Fig. (Fig.1).1). Thus, Yap and Taz appear prevalent across mouse embryo development but show differential expression at the earliest stages.
We further explored Yap expression by whole mount in situ in wild-type mouse embryos from E6.5 to E8.5 (Fig. (Fig.2).2). Again, we found widespread Yap expression overall, with dynamic stage- and region-specific domains of higher relative expression. At E6.5 (Fig. 2A to D), we observed Yap expression throughout extraembryonic ectoderm, epiblast, and nascent mesoderm in early streak-stage embryos (Fig. 2A and B) but found at later stages (Fig. (Fig.2C)2C) more intense expression in and adjacent to proximal epiblast regions. At E7.5 (Fig. 2E to G) we found widespread Yap expression with region-specific intensity distinctions (Fig. 2E and F). We observed strongest Yap expression in extraembryonic mesoderm and also in a ring of mid-proximodistal extraembryonic ectoderm. The strong extraembryonic ectoderm expression domain was proximal to the region forming the chorion. At this stage, Yap expression appeared lowest in visceral and definitive endoderm. By E8.5 (Fig. 2H,I) Yap expression remained widespread, with a strong expression domain at the distal tip of the allantois (Fig. (Fig.2H).2H). Overall, we observed a nearly ubiquitous, but dynamic, Yap expression pattern that modulated within detectable levels across developmental stages and distinct cell types.
To define in vivo requirements for YAP, we developed a targeting construct designed to disrupt Yap transcription. In the targeted allele Yaptm1Smil we replaced genomic sequence, including the Yap gene transcriptional and translational start sites and all but 10 bp of exon 1 with a reverse orientation MC1-neo cassette (Fig. (Fig.3A).3A). Electroporated ES cells subjected to positive and negative selection were screened by PCR, and homologous recombination confirmed with a 5′ external probe by Southern transfer (Fig. (Fig.3B).3B). Mice carrying the Yaptm1Smil allele were established by blastocyst ES cell injection and germ line transmission from chimeric founders.
Progeny from Yaptm1Smil heterozygous intercrosses were genotyped with PCR primer sets that generated a 259-bp product from the wild-type Yap allele and a 195-bp product from the Yaptm1Smil allele. Genotype analysis of E8.5 embryos obtained from a Yaptm1Smil heterozygous intercross (Fig. (Fig.3C)3C) demonstrates the presence of all three expected genotype classes.
To assess the impact of the mutant allele on YAP protein levels, we used affinity-purified polyclonal antibody generated against carboxy-terminal human YAP (26) to probe for endogenous mouse YAP in whole-embryo lysates obtained at E8.5 from Yaptm1Smil intercrosses. We did not detect the 65-kDa YAP protein in homozygous mutant (Yap−/−) embryos by Western blot analysis (Fig. (Fig.3D3D).
From over 900 heterozygous intercross progeny genotyped postnatally (Table (Table1)1) , homozygous Yaptm1Smil mice were never recovered. In contrast, heterozygous mice were viable, fertile, and exhibited no overt abnormalities. To delimit the stage of embryonic lethality associated with homozygosity for Yaptm1Smil and to characterize the impact of YAP absence on development, we obtained embryos at several stages across the developmental timeline. From E6.5 to E9.5, we found genotype ratios consistent with the expected Mendelian distribution (Table (Table1,1, 102 = 2.46, DF2). Although we genotyped two partially resorbed homozygous mutant embryos at E10.5, no Yap−/− embryos were recovered at later stages of development.
Although we found no distinguishing characteristics that correlated with genotype in embryos examined at E6.5 (n = 45 embryos; from seven litters), we did observe morphological perturbations in some mutant embryos by E7.5. By E8.5 and E9.5, however, we observed a characteristic Yap−/− embryo appearance (Fig. (Fig.4).4). Compared to wild-type siblings (Fig. (Fig.4A)4A) at E7.5, Yap−/− embryos showed morphological variability ranging from overtly normal (Fig. (Fig.4B)4B) to severely perturbed (Fig. (Fig.4C).4C). Perturbations observed at E7.5 included very small embryos, the presence of a marked constriction at the embryonic-extraembryonic boundary (see also Fig. Fig.5A)5A) or complete separation of the epiblast and the extraembryonic ectoderm. We observed some morphological abnormalities in about half of the embryos analyzed at this stage. Nonetheless, most E7.5 homozygous mutant embryos were gastrulating, and many had generated a proper amnion and chorion.
In embryos from Yaptm1Smil heterozygous intercrosses characterized at E8.5 (Fig. 4D to G), we found Yap−/− embryos displayed a consistent and stereotyped morphology typified by a strikingly short and wide body axis, a distinctive multiple folding of the anterior epithelium, caudal dysgenesis, and failure of chorioallantoic attachment. Despite chorion development and allantoic extension, Yap−/− embryos recovered at these and later stages showed no evidence for attachment of allantois to chorion.
Among embryos recovered a E9.5 (Fig. 4H and I), Yap−/− embryos showed failure of ventral closure and turning, a convoluted anterior neurepithelium, and a characteristic bulbous allantois. In addition, by this stage Yap−/− yolk sacs had a distinctive rippled appearance.
To provide a better understanding of developmental progression in Yap−/− embryos, litters from heterozygous Yaptm1Smil intercrosses were fixed in utero and processed for paraffin sectioning and standard hematoxylin and eosin staining (Fig. (Fig.5).5). Wild-type embryos sectioned at E7.5 had generated a chorion and amnion, embryonic and yolk sac mesoderm, and an allantoic bud (Fig. (Fig.5A).5A). In addition, wild-type yolk sac mesoderm adjacent to yolk sac visceral endoderm showed cell morphology and organization consistent with early blood island development (Fig. (Fig.5B).5B). Consistent with the morphology of dissected embryos, Yap−/− embryos sectioned at E7.5 (Fig. (Fig.5A′5A′ and B′) had generated mesoderm but this tissue layer appeared disorganized (Fig. (Fig.5B′)5B′) relative to that of control embryos. The Yap−/− embryo shown in Fig. Fig.5A′5A′ and B′ demonstrated the constriction at the embryonic-extraembryonic boundary (Fig. (Fig.5A5A′).
We also compared stained sections from wild-type (Fig. 5C to H) and Yap−/− embryos (Fig. (Fig.5C′5C′ to H′) at E8.5. Both wild-type and homozygous mutant embryos showed distinct chorion, amnion, and neuroectoderm development (Fig. 5C, C′, D, and D′). In wild-type embryos, the thickened epithelium of the streak region extended posterior to the node (Fig. (Fig.5C).5C). In contrast, in Yap−/− embryos, despite development of a morphologically recognizable node (not shown), posterior regions showed a thinner tissue layer overlying the mesoderm extending to the amnion-allantoic bud junction (Fig. (Fig.5C′).5C′). This failure to maintain a posterior epiblast-like epithelium in the streak region is consistent with the observed caudal dysgenesis. In addition, anterior neurectoderm in homozygous mutant embryos showed excessive folding, extending as a single tissue layer into the amniotic cavity (Fig. (Fig.5D′).5D′). Some Yap−/− embryos developed anterior somites (Fig. (Fig.5E′)5E′) that looked much like those of wild-type sibling embryos (Fig. (Fig.5E),5E), and some initiated heart development (Fig. (Fig.5F′)5F′) to stages similar to that of wild-type sibling embryos (Fig. (Fig.5F).5F). Development of the Yap−/− allantois (Fig. (Fig.5G′)5G′) was similar to that of control siblings (Fig. (Fig.5G),5G), including extension into the yolk sac cavity and differential cell morphology along the allantoic proximodistal axis. In contrast, we observed striking differences in the development of the yolk sac mesoderm. Wild-type embryos showed an easily distinguished ring of yolk sac cells organized into blood islands with the concomitant distinctions between endothelial and hematopoietic precursor cells (Fig. (Fig.5H).5H). In Yap−/− embryos, although yolk sac mesoderm cells were abundant, well-defined blood island-like structures were rare and showed less clear cell type distinctions (Fig. (Fig.5H5H′).
We used marker analysis to characterize the yolk sac vasculature defect revealed by our histological analysis. The PECAM1 antigen is a marker for endothelial precursor cells in yolk sac and mouse embryo by E8.5 (1). By E8.5, wild-type and heterozygous embryos stained with an antibody against PECAM1 (Fig. 6A to C) show a well-formed yolk sac vascular plexus, as well as allantoic and embryonic blood vessels. In sharp contrast, although Yap−/− embryos showed cells immunopositive for PECAM1, these cells entirely lacked organization into vessels in the yolk sac (Fig. (Fig.6D).6D). Instead, PECAM1-labeled cells appeared to be profusely distributed across most of the yolk sac in homozygous mutant embryos. Conversely, the developing vasculature in the Yap−/− allantois appeared fairly normal (Fig. (Fig.6E).6E). Although abnormally positioned, blood vessels also were evident (Fig. (Fig.6F)6F) in the Yap−/− embryo proper. Thus, although the endothelial cell component of the early vasculature appears to be specified during Yap−/− embryogenesis, in the yolk sac endothelial cells fail to organize into vessels.
The yolk sac vasculature develops with the concomitant specification and organization of both endothelial and hematopoietic precursor cells. Early development of distinct cell types occurs in a ring of yolk sac tissue as organized structures referred to as blood islands which later anastomose to generate the yolk sac vascular plexus. Because our histological analysis of Yap−/− embryos showed a large number of disorganized yolk sac mesoderm cells, we explored development specifically of the hematopoietic component of the yolk sac vasculature by using a riboprobe to detect alpha globin expression. Alpha globin is expressed in erythrocyte precursors present in the murine yolk sac by E7.5 (35) and is present in these erythroblasts of the yolk sac primitive plexus at E8.5 (Fig. (Fig.6H).6H). Whole-mount in situ analysis for alpha globin showed expression of this erythroblast marker in cells of the Yap−/− yolk sac (Fig. 6I to K) but in a domain lacking the distinctive organization normally observed. Thus, in Yap−/− embryos, we observe specification of both components of the early yolk sac vasculature—endothelia and erythroblasts—but see failure of these cells to develop with the organization of a proper primitive vascular plexus.
We used whole-mount in situ analyses of expression domains for several marker genes to help define the extent of patterning achieved in the Yap−/− embryo proper (Fig. (Fig.7).7). We used two markers, fibroblast growth factor 8 (Fgf8) and homeobox gene expressed in ES cells (Hesx1) (22), to reveal the extent of anteroposterior pattern present in the convoluted neurectoderm of Yap−/− embryos by E9.5. By organogenesis stages, Fgf8 normally is expressed in neurectoderm in both an anterior domain and at the midbrain-hindbrain boundary, in addition to branchial arch, limb bud and tail bud domains (7). Consistent with the restricted bands of expression observed in wild-type embryos (Fig. (Fig.7A),7A), we found that Fgf8 was present in Yap−/− embryos (Fig. 7B and C) in one or two narrow stripes across the folded neuroepithelium, indicating elements of proper AP patterning. Similarly, Hesx1 marked the anterior-most neurepithelium (22) in both wild-type and mutant embryos (data not shown). In addition, the posterior tailbud domain showed proper Fgf8 expression in Yap−/− embryos (Fig. 7B and C).
Because the body axis of Yap−/− embryos appears short and wide (Fig. 4F and G), we used a midline marker to assess mediolateral patterning (Fig. 7D to F). Brachyury (T) normally is expressed in the primitive streak, in the tail bud, and in the notochord precursor cells that are distributed along the midline of the embryo (62). In Yap−/− embryos (Fig. 7E and F) we found T expression in the streak and the tailbud. Although a Texpression domain also demarcated a midline in Yap−/− embryos, this domain appeared unusually wide and the T-expressing cells were distributed in a discontinuous manner. This pattern of T expression suggests that Yap−/− embryos initiate midline development, specifying notochord precursor cells, but the unusual mediolateral spread and discontinuity of the Tdomain suggests abnormalities in development along the mediolateral axis, which is consistent with the abnormally wide axis morphologically evident in mutant embryos (Fig. (Fig.4G4G).
Embryos lacking YAP arrest development around E8.5 and display defects in yolk sac vascular development (Fig. (Fig.5H′,5H′, 6D, and H), chorioallantoic fusion (Fig. (Fig.4I),4I), and embryonic axis elongation (Fig. 4F, G, and I). The complexity of the mutant embryo phenotype, with profound defects in both the embryo proper and the extraembryonic regions, and the observed widespread developmental Yap gene expression (Fig. (Fig.2)2) together support multiple roles for YAP in development. In general, the source of failed embryonic progression in Yap−/− embryos does not appear to result from primary problems with tissue specification since histological (Fig. (Fig.5),5), and marker analyses (Fig. (Fig.66 and and7)7) indicate relatively normal development of distinct cell types in both the yolk sac and the embryo proper. Instead, our results may reveal vital requirements for YAP in morphogenetic movements or maintenance of proper cell number during embryogenesis.
One striking outcome of analyses of mouse mutant phenotypes in recent decades has been the understanding that, although profound developmental disturbances in the embryo proper may be tolerated until late gestational stages or even to birth, disruption in the cardiovascular and placental systems likely results in death in utero (4,5, 50). Consistent with this observation, Yap−/− embryos, which die in the first half of gestation, show significant perturbation in early yolk sac vascular development, as well as in an early step in placental development. Current understanding of vascular development in the yolk sac suggests a multistep progression initiating with vasculogenesis (11). Although the molecular regulation of vasculogenesis is only beginning to be characterized (6, 11, 24), the Vegf signaling pathway is essential. Indeed, the vascular defect in Yap−/− yolk sac (Fig. (Fig.55 and and6)6) appears similar to the yolk sac phenotype of embryos that lack VEGFR1 (15) in which an overabundance of yolk sac endothelial cells underlies vasculogenesis failure (16). Although we have observed by RT-PCR retained expression of VEGFA, VEGFR1, and VEGFR2 in Yap−/− embryos (data not shown), further study may support a role for YAP along this pathway. Thus, we can now add YAP to the short list of genes known to be required for yolk sac vasculogenesis and can look for a potential role for YAP in the regulation of endothelial cell number. Strikingly, the Yap−/− embryo also provides a genetic dissection of the process of vasculogenesis by region: in yolk sac versus in both the embryo proper and the allantois. This regional distinction corresponds largely, albeit not strictly (11), with a cellular contingent difference. In yolk sac, blood vessels develop adjacent to a visceral endoderm cell layer and, in some regions, in association with blood cell precursors. In contrast, in both embryo (11) and allantois (10) vasculogenesis proceeds largely in the absence of these other cell types. Cell-cell interactions between yolk sac mesenchyme and visceral endoderm appear to be important for yolk sac vascular development (2, 13, 46). Therefore, it may be that the role of YAP lies in the cross talk between yolk sac mesenchyme and adjacent yolk sac visceral endoderm. Alternatively, YAP may mediate steps essential for vessel development specifically in the presence of hematopoietic precursors, a developmental process that remains largely uncharacterized.
YAP also appears fundamental to an early step in placental development: attachment of the allantois to the chorion. The bulbous appearance of the Yap−/− allantois by E9.5 (Fig. (Fig.4I)4I) is like that observed in other embryos in which the allantois forms and elongates but fails to fuse with the chorion (20, 28, 33, 34, 36, 42, 44, 48, 50, 52, 56, 59, 63, 67). We found that Yap is expressed in the cell components potentially involved in the process (9) including chorionic ectoderm, chorionic mesoderm, and allantois (Fig. 2E, F, and H). In addition, we observed proper initial development of these tissue types in mutant embryos, suggesting again that YAP functions not generally for cell type specification but principally in the cell-cell interaction events required for attachment. The apparent higher intensity Yap expression in distal-tip allantois (Fig. (Fig.2H)—the2H)—the very region required for initial attachment—motivates further study. Curiously, for many of the genes with identified roles in chorioallantoic attachment and fusion, mutation causes defects in this process in only a fraction of the homozygous mutant embryos. This observation supports the existence of multiple pathways to reach fusion, each of which may sometimes suffice alone. Remarkably, we found failed chorioallantoic attachment in 100% of Yap−/− embryos observed. Therefore, YAP appears fundamental to this critical step in the initiation of the placental circulation.
In addition to showing critical requirements for YAP in early steps of yolk sac vascular and placental development, Yap−/− embryos also reveal an essential role for YAP in shaping the main body axis of the embryo. Failure to elongate the body axis in Yap−/− embryos suggests a role for YAP in the regulation of cell number in the embryo proper and/or in the fundamental morphogenetic movements of gastrulation. We found that the Yap−/− embryo phenotype overall appeared remarkably similar to that shown by mouse embryos carrying mutations influencing fibronectin (FN)-α5β1 integrin interactions (17-19, 66, 68). Indeed, mouse embryos in which FN-α5β1 interactions have been perturbed show a short, wide body axis, and yolk sac vascular defects as do Yap mutant embryos. Moreover, the YAP binding partners of the Src kinase family have been shown to mediate integrin-stimulated cell migration (41) and mouse embryos triple mutant for three of these genes—Src, c-Yes, and Fyn—also show a shortened body axis similar to the FN-α5β1 mutant classes (31) and thus also to Yap mutant embryos. Furthermore, a possible role for YAP in the convergent extension cell movements critical for vertebrate body axis elongation is suggested by evidence of a role for the zebra fish Fyn and c-Yes homologs in this process (29). Thus, YAP might serve by several potential interactions to promote the morphogenetic movements critical for proper axis elongation.
In addition, YAP could impact axis elongation through the regulation of proliferation, apoptosis, or both processes. Indeed, YAP is known by biochemical and cell culture assays to influence apoptotic and cell cycle regulatory pathways (12, 55). Moreover, recent evidence from Drosophila supports a role for YAP in regulating these processes (27). Nonetheless, the available genetic evidence from mice indicates that these known interactions alone are insufficient to explain the phenotype since, for example, mouse embryos homozygous mutant for either p73 (65) or p63 (40) survive to birth. Although not characterized extensively in wild-type gastrula-stage mouse embryos, apoptosis and proliferation appear to be fairly evenly distributed with potentially higher rates of both processes associated with distal and anterior epiblast (23, 38, 49, 51, 53, 54, 58). Although we are in the process of rigorously examining both proliferation and apoptosis in the absence of YAP, our histological preparations of E7.5 Yap−/− embryos (Fig. (Fig.5A′5A′ and B′) do not suggest marked changes in the number of mitotic or dying cells. At later stages, with more perturbed mutant embryo development, it becomes difficult to distinguish primary from secondary impacts on cell survival and proliferation.
We observe expression of both Yap and Taz throughout most of mouse embryo development, with differential expression observed only at the blastocyst stage (Fig. (Fig.1).1). Despite this predominantly concomitant expression and our finding by RT-PCR analysis of retained Taz expression in mutant embryos (data not shown), Yap−/− embryos show severe developmental perturbation. This result argues for critical requirements for YAP that cannot be compensated for by potential redundant functions of the related protein TAZ. Possibly, a single time window for this vital YAP role occurs at peri-implantation stages, prior to TAZ expression onset. Nonetheless, it seems likely that YAP plays many roles throughout development some of which will occur during concomitant TAZ expression. Thus, the phenotype of the mutant embryo supports unique in vivo requirements for YAP. Teasing apart the mechanisms of YAP function in development may provide insight into the rules defining the independent roles of these related proteins.
The study of YAP interactions and function in vitro and in cell culture has identified several binding partners and potential signaling pathway interactions for this protein. We generated a targeted null allele to reveal YAP requirements at the level of the whole organism. We found that the Yap−/− phenotype overall is not easily accounted for by individual interactions with known binding partners or signaling pathways. This result may be explained to some extent by combinatorial effects, with multiple YAP interactions impacting embryogenesis. These observations also may suggest, however, that protein-protein interactions between YAP and heretofore unrecognized binding partners could play critical roles in the early stages of embryogenesis.
We thank Stephen Gee for critical reading of the manuscript, Sean Barron for help with mouse colony maintenance, the UNC Histopathology Facility for sections, and the UNC Animal Models Core for generating Yaptm1Smil chimeric founders.
This study was supported by National Institutes of Health grant HL63755 to S.L.M.