Yes-Associated Protein 65 (YAP) Expands Neural Progenitors and Regulates Pax3 Expression in the Neural Plate Border Zone

Yes-associated protein 65 (YAP) contains multiple protein-protein interaction domains and functions as both a transcriptional co-activator and as a scaffolding protein. Mouse embryos lacking YAP did not survive past embryonic day 8.5 and showed signs of defective yolk sac vasculogenesis, chorioallantoic fusion, and anterior-posterior (A-P) axis elongation. Given that the YAP knockout mouse defects might be due in part to nutritional deficiencies, we sought to better characterize a role for YAP during early development using embryos that develop externally. YAP morpholino (MO)-mediated loss-of-function in both frog and fish resulted in incomplete epiboly at gastrulation and impaired axis formation, similar to the mouse phenotype. In frog, germ layer specific genes were expressed, but they were temporally delayed. YAP MO-mediated partial knockdown in frog allowed a shortened axis to form. YAP gain-of-function in Xenopus expanded the progenitor populations in the neural plate (sox2+) and neural plate border zone (pax3+), while inhibiting the expression of later markers of tissues derived from the neural plate border zone (neural crest, pre-placodal ectoderm, hatching gland), as well as epidermis and somitic muscle. YAP directly regulates pax3 expression via association with TEAD1 (N-TEF) at a highly conserved, previously undescribed, TEAD-binding site within the 5′ regulatory region of pax3. Structure/function analyses revealed that the PDZ-binding motif of YAP contributes to the inhibition of epidermal and somitic muscle differentiation, but a complete, intact YAP protein is required for expansion of the neural plate and neural plate border zone progenitor pools. These results provide a thorough analysis of YAP mediated gene expression changes in loss- and gain-of-function experiments. Furthermore, this is the first report to use YAP structure-function analyzes to determine which portion of YAP is involved in specific gene expression changes and the first to show direct in vivo evidence of YAP's role in regulating pax3 neural crest expression.


Introduction
Yes-associated protein 65 (YAP) contains multiple proteinprotein interaction domains and functions as both a transcriptional co-activator and as a scaffolding protein. YAP was first identified and named based on its association with the Src-family tyrosine kinase and proto-oncogene, c-Yes [1]. YAP is a founding member of the WW domain-containing protein family [2,3]. The WW domain allows the binding of proteins containing a PPxY motif [4]. Proteins shown to bind to YAP via its two WW domains include: p53 family members (p73a, p73b, p63 [5]; Smad7 [6]; Runx2 [7]; and ErbB4 [8,9]. In addition to the two WW domains, YAP also contains other protein-protein interaction domains ( Figure 1A). Proteins that interact at the N-terminus of YAP include hnRNP U, a nuclear ribonucleoprotein shown to be important for RNA polymerase II transcription [10,11], the TEA domain-containing transcription factor (TEAD/TEF) family [12], and the Large tumor suppressor (LATS). The phosphorylation event involving LATS via the Hippo signaling pathway allows for the binding of 14-3-3, which leads to the subsequent sequestration of YAP to the cytoplasm [13]. At its C-terminus, YAP contains a postsynaptic density 95, discs large, and zonula occludens-1 (PDZ)-binding motif that allows for binding to PDZ domain-containing proteins.
Our initial interest in YAP came from the finding that YAP bound to the second PDZ domain of Na(+)/H(+) exchanger regulator factor 1/ezrin/radixin/moesin (ERM)-binding phosphoprotein of 50 kDa (NHERF1/EBP50) and co-localized to the apical membrane of polarized airway epithelia along with CFTR and c-Yes [14]. To determine the in vivo importance of this scaffolding complex, we used homologous recombination to remove YAP from the mouse and found that few embryos survived past embryonic day 8.5, a much earlier time point than would be expected for an associated lung development phenotype the ascribed functional and protein-protein interaction domains, including the TEAD-binding site (purple), the LATS phosphorylation site (orange), the two WW domains (red) that allow for PPxY binding, the Src Homology 3 (SH3)-binding domain (green), the coiled-coil region (blue), the transactivation domain (underline), and the PDZ-binding motif (pink). hnRNP U (yellow) binding has only been experimentally tested with human YAP, but related sites are in the fish and frog proteins. This diagram also illustrates the relative location of Xenopus laevis (x) and Danio rerio (z) MO-binding sites. (B) Injection of an equimolar cocktail of all three xYAP MOs at two concentrations (40 ng and 80 ng) resulted in efficient knockdown of endogenous, zygotic xYAP protein in stage 15 embryos as measured by western blot analysis. EF-2 expression from the same blot served as the loading control. (C) Three different xYAP MOs (80 ng; see A for binding sites) resulted in failed closure of the blastopore (arrows). (D) Reducing the concentration of the xYAP MO cocktail (left side) allowed blastopore closure, but resulted in dose-dependent A-P axis shortening. (E) Time-lapse video microscopy showed that zYAP MO (16 ng) injected embryos also exhibit perturbation in the completion of gastrulation. Asterisks mark the tissue front of epiboly movements. In uninjected and cMO-injected embryos, this front completely envelops the yolk by 10 hours post-fertilization (hpf). These fronts are still in the equatorial region in the 7.7-10 hpf YAP MO-injected embryos. doi:10.1371/journal.pone.0020309.g001 [15]. Detailed analyses of these mice illustrated that they suffered from defects in yolk sac vasculogenesis, chorioallantoic fusion, and anterior-posterior (A-P) axis elongation.
Given that YAP knockout mice struggled to progress normally through early development, in part because of nutritional deficiencies, we sought to better characterize a role for YAP during this time period by using embryos that develop externally: Xenopus laevis and Danio rerio. YAP morpholino (MO)-mediated complete loss-of-function prevented the completion of epiboly, delayed mesoderm induction, and severely impaired A-P axis elongation, phenotypes that were similar to YAP 2/2 mice. YAP gain-offunction experiments in Xenopus laevis expanded the progenitors of the neural plate and neural plate border zone, while concomitantly inhibiting expression of later markers of tissues derived from the neural plate border zone (neural crest, pre-placodal ectoderm (PPE), hatching gland), as well as epidermis and somitic muscle. Through gain-and loss-of-function experiments and endogenous chromatin immunoprecipitations (ChIP) for YAP, we show that YAP directly regulates pax3 expression via association with TEAD1 (N-TEF) and ultimately localizes to a highly conserved, previously undescribed, TEAD-binding site within the 59 regulatory region of pax3. Finally, structure/function analyses revealed that the PDZ-binding motif of YAP contributes to the inhibition of epidermal and somitic muscle differentiation, but a complete, intact YAP protein is required for expansion of the neural plate and neural plate border zone progenitor pools.

Materials and Methods
Animal use and ethics statement

xYAP and morpholinos
A Xenopus laevis full-length cDNA clone (XL211h05) of yesassociated protein 65 (xyap) was obtained from the National Institute for Basic Biology (Japan) and sequenced in both directions (GenBank Accession #FJ979828). Three morpholinos, MO1 (GGA GGT GGG AGC TAG GAC AGC GG), MO2 (GGA GAG GAC GCG GTA GGA GAC TGT G), and MO3 (GGG CTC CAT GGC TGC GGG GAG GTG G), were designed to the 59UTR of xyap for translational blocking ( Fig. 1A; GeneTools). Two splice blocking and putative early translational truncation MOs, exon 1 (GTA GAG GAG CAT ATA CCT GCC GTG A) and exon 2 (CCT GCA AAG AAC AAG TGG GAC AAT A) (GeneTools) were designed across exon/intron boundaries. In vitro translation reactions were performed using the TnT Quick Coupled Transcription/Translation System (Promega), according to the manufacturer's protocol. Each MO (80 ng) was injected into in vitro fertilized 1-cell Xenopus laevis embryos according to established methods [16]. To observe phenotypes associated with lower MO concentrations (0, 1.25, 2.5, 5, 10, 20, and 40 ng total), a cocktail of all three (MO1, MO2, and MO3) translational blocking MOs was injected into in vitro-fertilized 1-cell sibling embryos.

Cloning methods and constructs
For use in all of our gain-of-function analyses, we initially cloned an HA tag (ATG TAC CCA TAC GAT GTT CCA GAT TAC GCT) into the XhoI and EcoRV sites of the pSP64TXB vector so that proper expression could be detected by western blot analysis. xyap and xtead1 were subcloned in frame with the HA tag into the EcoRV and NotI sites of the pSP64TXB-HA vector. A set of xYAP mutant constructs, which included a constitutively active form of xYAP (cActive xYAP) with a mutated LATS phosphorylation site (S98A), a deletion (aa 61-81) of the TEAD-binding site (xYA-PDTBS), a deletion (aa 78-161, aa 199-236) of the WW domains (xYAPDWW), a deletion (aa 1-90) of the entire N-terminus (xYAPDN), and a deletion (aa 455-459) of the PDZ-binding motif at the C-terminus (xYAPDC), were also subcloned into the pSP64TXB-HA vector at the EcoRV and NotI sites.

Gain-of-function analyses
For initial gain-of-function analyses, the animal poles of 1-cell Xenopus laevis embryos were injected with 2 ng of in vitro-transcribed ha-xyap mRNA (mMessage Machine, Ambion). For additional gain-of-function analyses, 2-cell Xenopus laevis embryos were co-injected with 1 ng of in vitro-transcribed ha-xyap or ha-xyap mutant mRNAs and 100 pg of in vitro-transcribed nls-b-galactosidase mRNA into the lateral, animal pole of one of the two blastomeres. Similarly, 100 pg of the in vitro-transcribed ha-xtead1 (xn-tef1) [17] or 100 pg of ha-xyap mRNAs, were injected alone or in combination. Every mRNA injection was repeated 2-4 times per construct using different parental frogs to normalize against variation in genetic backgrounds and micropipette delivery. Results from these independent experiments were then pooled.

zYAP and embryo manipulations
A Danio rerio full-length cDNA IMAGE clone 7066008 of yesassociated protein 65 (zyap) (NM_001115121) was obtained from Open Biosystems. A zYAP MO (59 CTC TTC TTT CTA TCC AAC TGA AAC C 39) was designed to the 59 UTR of zyap (GeneTools). In vitro translation reactions were performed using the TnT Quick Coupled Transcription/Translation System (Promega), according to the manufacturer's protocol.
1-cell embryos were injected with 16 ng of the zYAP MO, a standard control MO (GeneTools), or 300 or 600 pg of in vitro-transcribed ha-zyap mRNA. Once embryos reached the 1000-2000-cell stage, their chorion membranes were removed and embryos were placed on a custom fitted imaging mold (kindly supplied by Dr. Sean Megason) [18] for time-lapse videography. Embryos were subsequently allowed to progress to the prim-11 stage and fixed.

Chromatin Immunoprecipitation (ChIP)
ChIP assays were performed with the ChIP-IT Express kit (Active Motif) with some modifications. Three hundred stage 14-16 Xenopus laevis embryos were incubated, with gentle rolling, in 10 ml of 1% formaldehyde/0.16 modified Barth's solution (MBS) for 30 min at room temperature to crosslink genomic DNA and protein complexes. Crosslinking was stopped by incubating the embryos in 125 mM glycine/0.16 MBS with gentle rolling. Following two washes in 0.16 MBS, the embryos were snapfrozen and stored at 280uC. Chromatin was sheared with a Misonix 3000 cup horn by repeating 6 cycles of 30 sec: 1 sec pulse, 0.5 sec off at a power of 5. Samples rested on ice for 1 minute between each cycle. Shearing efficiency was determined by resolving a reverse-crosslinked, precipitated sample of chromatin on a 1% agarose gel. This sample was quantified using a Nanodrop, and 12.5 or 25 mg of chromatin was subsequently immunoprecipitated for 4 hours with 2 mg of affinity-purified YAP antibody or rabbit IgG (Genscript) in the presence of 0.25 mg/ml BSA and 0.1 mg/ml herring sperm DNA. Beads were washed once with ChIP buffer 1 (Active Motif) and twice with ChIP buffer 2 (Active Motif) prior to elution and proteinase K treatment. Five percent of the eluate was used to amplify the pax3 promoter TEAD-binding site region using the following xpax3-specific primers: forward (GCC TGA CAA TGG CAC CTT AT) and reverse (AGG CGC ACT TGT GTG ATT C). For subcloning this region, a proofreading DNA polymerase (cloned Pfu DNA polymerase, Stratagene) was used to PCR amplify the product from the isolated YAP co-immunoprecipitated Xenopus laevis genomic DNA. This PCR product was then gelpurified from a 1% agarose gel. Alanines were then added back to the ends using a non-proofreading DNA polymerase (Jumpstart Taq polymerase, Sigma-Aldrich). The products were then ligated into the pCRII-TOPO vector (Invitrogen) and sequenced.

Results
We previously showed that YAP 2/2 mice were embryonic lethal and exhibited severe developmental abnormalities that included defects in yolk sac vasculogenesis, chorioallantoic fusion, and A-P axis elongation [15]. Given that these defects could be due to nutritional deficiencies, we sought to better characterize a role for YAP during early development by using Xenopus laevis and Danio rerio, animal models for which the nutritional needs of the embryos are self-contained. In addition, these embryos permit easy knockdown of targeted protein expression via injection of genespecific MOs and efficient gain-of-function assessment via mRNA injections.

YAP is required for progression through gastrulation
The full-length Xenopus laevis yap (xyap) EST encodes a protein that is 78% identical to mouse YAP and contains all the described protein-protein interaction domains, as well as the transcriptional activation domain ( Figure 1A). Isolation of Xenopus laevis genomic DNA and subsequent PCR validated that our RT-PCR primer design amplified a PCR product across exon-intron boundaries (data not shown). RT-PCR and western blot analyses revealed that xYAP mRNA and protein are maternally expressed in the unfertilized egg through blastula stages, and are abundantly expressed from the onset of zygotic transcription (at mid-blastula transition) through tadpole stages ( Figure S1). These results are consistent with reports of ubiquitous maternal mRNA expression in Xenopus tropicalis, zebrafish and mouse, and widespread zygotic expression in multiple neural, neural crest, and mesoderm derived tissues, but limited expression in endoderm derived tissues of the post-gastrulation embryo [15,34,35,36,37].
Three xYAP MOs were designed around the translational start site ( Figure 1A) and their efficacies were confirmed in vitro (data not shown). An antibody directed against the C-terminus (274-454) of human YAP (hYAP) detected a band at the appropriate size from cold in vitro-translated xYAP product and stage 15 whole embryo lysates (data not shown). We used this hYAP antibody to test the efficacy of our xYAP MOs in vivo. Lysates from stage 15 MOinjected embryos showed efficient knockdown of endogenous, zygotic xYAP protein expression to undetectable levels ( Figure 1B), thus effectively mimicking the genetic knockout achieved in mouse [15].
Reducing the concentration of the xYAP MO cocktail below 40 ng resulted in partial endogenous, zygotic YAP protein expression (data not shown). This concomitantly allowed blastopore closure, but resulted in dose dependent A-P axis elongation defects ( Figure 1D). Embryos injected with 1.25 ng (n = 113) or 2.5 ng (n = 142) of the xYAP translation blocking MO cocktail appeared unaffected, whereas embryos injected with 5-20 ng (5 ng, n = 152; 10 ng, n = 155; 20 ng, n = 163) of this cocktail did not progress through gastrulation as rapidly as their control siblings, and had progressively shortened body axes ( Figure 1D). A similar phenotype was reported for a 5-7.5 ng YAP MO dose in zebrafish embryos [36].
Although the defective blastopore closure phenotype was reproducible using three different translational blocking xYAP MOs individually or in combination as well as two different splice blocking xYAP MOs, the phenotype was not rescued by coinjecting 2 ng of frog (xyap), mouse (myap), or human (hyap) mRNAs, even though they all were properly translated in Xenopus laevis embryos ( Figure S3). Therefore, we tested whether knockdown of YAP in another animal model, using similar methods, would produce a similar phenotype. Jiang et al. previously showed that a low dose of YAP MO that reduced fluorescence of a YAP 59UTR-eGFP reporter resulted in shortened embryos with small heads; however, these embryos were only analyzed at late stages (30-50 hpf) and thereby did not address the gastrulation defects observed in frog YAP morphants [36]. Therefore, we injected 16 ng of a zYAP MO into fertilized one-cell zebrafish embryos, which completely eliminates expression of zYAP in an in vitro translation assay (data not shown). Time-lapse videography showed that in the absence of zYAP epiboly movements were arrested ( Figure 1E). While the tissue front of epiboly gradually closed around the yolk cell between 5.25-10 hpf in uninjected and control MO-injected embryos (n = 15 per each group), this closure was not achieved in YAP morphants (n = 15; 100%). Thus, in three different vertebrates, loss of early YAP function interferes with the developmental networks that regulate completion of gastrulation movements.
To determine whether the MO-mediated gastrulation defects correlated with an effect on genes required for germ-layer formation, we performed qPCR analyses on well-established markers of each germ layer. Control MO (80 ng) or the xYAP MO cocktail (80 ng) were injected into one-cell Xenopus laevis embryos, and the embryos were collected when uninjected siblings reached mid-gastrulation (stage 11), a stage when germ layer markers are abundantly expressed. Genes normally expressed in the organizer at the onset of gastrulation were either unaffected (siamois) or moderately increased (nodal-related 3). In contrast, the expression levels of endodermal (sox17, p,0.013), neural ectodermal (sox11, p,0.021), and three out of five mesodermal (brachyury, p,0.013; goosecoid, p,0.011; wnt8 p,0.018) genes were significantly reduced in YAP MO-injected embryos (Figure 2A). However, analyses of several mesodermal markers by in situ hybridization in Xenopus laevis showed that these quantitative changes resulted from delayed expression rather than loss of mesoderm induction. While brachyury, eomesodermin, and chordin expression was markedly reduced in xYAP MO-injected embryos compared to sibling stage 11 embryos ( Figure 2B), at sibling stage 13 these genes, as well as several others, were expressed in patterns similar to control stage 11 embryos ( Figure 2B, n = 14-25 per sample, 100% for all markers). Thus, eliminating endogenous, zygotic xYAP protein expression does not prevent mesodermal gene induction, but does delay the expression of a number of mesodermal genes. These results indicate that the lack of progression through gastrulation in YAP morphant embryos is not due to a failure in germ layer inductions.

YAP gain-of-function also causes axis elongation defects
From these results, we predicted that increasing YAP protein above endogenous levels may cause gastrulation to be completed more rapidly. However, time-lapse video recordings of gastrulation movements in zebrafish embryos injected with two different doses of zyap mRNA did not detect any differences in the amount of time required for epiboly movements to close around the yolk plug ( Figure 3A, n = 5 per group). When these zyap mRNA injected embryos developed to later stages, however, significant perturbations were observed ( Figure 3B, 300 pg, n = 111, 100%; 600 pg, n = 94; 100%), including shortened A-P body axis and perturbed somitic and head morphologies. Likewise, injection of xyap, myap, or hyap (2 ng) mRNAs all caused similar phenotypes in Xenopus laevis embryos ( Figure 3C; n = 155 (xyap), n = 102 (myap), n = 93 (hyap), 100% affected). While these gain-of-function phenotypes resemble the reduced YAP phenotypes reported in zebrafish [36] and frog ( Figure 1D), analysis of gene expression (next section) indicates that YAP gain-of-function and loss-offunction results in strikingly different effects. Since these late phenotypes likely result from early alterations in tissue patterning and/or A-P axis formation, and given that YAP is a well-described transcriptional co-activator, we predicted that altering YAP levels would have distinct effects on the expression of genes involved in these early developmental processes.

YAP expands neural progenitors and inhibits neural differentiation
Vertebrate A-P axis elongation is accomplished in part by elongation of the neural plate [38,39], and in Xenopus and zebrafish neurulae, yap mRNA is robustly expressed in the floor plate and lateral borders of the neural plate [36,40]. Therefore, we investigated whether neural progenitor fields were altered in YAP gain-of-function Xenopus laevis embryos. Given that our previous YAP gain-of-function experiments were injected into 1-cell Xenopus laevis embryos, we chose to more closely monitor where our injected mRNA ended up in the embryo by co-injecting xyap and b-galatosidase (b-gal; as a lineage tracer) mRNAs into one blastomere of the 2-cell embryo. The neural progenitor field, indicated by sox2 expression, was expanded as evidenced by a darker, longer, and/or wider expression domain compared to the uninjected side of the same embryo ( Figure 4A). Injection of the YAP MO cocktail (40 ng) caused a loss of sox2 expression on the injected side ( Figure 4A), indicating that the YAP gain-of-function and loss-of-function phenotypes are manifested early in body axis formation by different mechanisms. Expansion and perdurance of sox2 expressed resulted in concomitant repression of neural differentiation markers ( Figure 4B). neuroD, a bHLH neural differentiation transcription factor, p27 Xic1 , a cdk inhibitor shown to be important for cell cycle exit and subsequent neural differentiation [28], and n-tubulin, a post-mitotic neuron-specific tubulin, each were strongly repressed ( Figure 4B). These results are consistent with a study in the chick neural tube demonstrating that phosphorylated YAP is expressed in the sox2 + ventricular zone, and that electroporation of YAP increased the sox2 + domain, decreased the number of neuron-specific tubulin positive cells and reduced cell cycle exit [41]. Interestingly, we also observed that the expression of a muscle-specific bHLH differentiation marker, myoD, was strongly repressed (n = 52, 68%) ( Figure 4D) indicating that the effects of YAP were not confined to the neural ectoderm.
It is well documented that increased Notch expression and/or signaling correlates with increased numbers of neural progenitors, decreased differentiated neurons [31,42,43], and increased YAP expression [44]. Therefore, we were surprised to observe that xYAP gain-of-function reduced the mRNA levels of notch and hes1, a direct Notch signaling target gene [45,46,47] (Figure 4C). These results indicate that YAP's ability to repress neural differentiation is likely independent of Notch signaling.
The expansion of neural progenitors by increased YAP levels also reduced the expression domain of the differentiated epidermis, as marked by an epidermal-specific cyto-keratin [48] ( Figure 5B). Because interactions between the neural plate and the epidermis lead to the formation of a neural plate border zone that gives rise to the precursors of the peripheral nervous system, the neural crest and the pre-placodal ectoderm (PPE) [49], we analyzed whether these tissues were properly induced at neural plate stages. While the pax3 + domain surrounding the neural plate, which is required to specify neural crest [50,51]), was expanded ( Figures 5D, 6A), two genes expressed by premigratory neural crest (foxD3; zic1; [33,51]) and two PPE genes (six1, sox11; [52]) were dramatically reduced ( Figure 5A, B). In addition, the pax3 + precursors of the hatching gland [51] were virtually eliminated ( Figure 5D). Thus, in xyap-mRNA injected embryos the three different precursor populations that contribute to the formation of peripheral cranial structures did not develop at neural plate stages. To determine whether the placode and neural crest populations eventually develop, we targeted xyap-mRNA injections to the neural plate border zone blastomere precursors (blastomeres D12 and V12; [53]). At tail bud stages: 1) placode expression of six1 was significantly reduced in the otocyst ( Figure 5E) and the olfactory placode (data not shown); 2) foxD3-expressing neural crest cells were more dorsally located than in controls, indicating delayed migration into the periphery ( Figure 5F); and 3) trigeminal placode expression of neuroD was reduced ( Figure 5G). The perturbations in the differentiation of these precursors of peripheral cranial structures likely account for the cranial defects seen in the YAP gain-of-function late stage frog and fish embryos ( Figure 3B, C).

YAP cooperates with TEAD to expand pax3 + neural crest progenitors
While the pax3 + hatching gland cells were virtually eliminated, pax3 expression in the neural plate border zone that is required for neural crest specification [50] was extended, broadened, and/or stronger compared to the uninjected, control side of the embryo (Figures 5D, 6A). This expansion of the pax3 + neural crest progenitor field was concomitant with a decrease in genes expressed by specified, premigratory neural crest (zic1, foxD3; Figure 5B), suggesting that increased YAP holds these cells in a progenitor, undifferentiated state longer than in an unmanipulated embryo. Consistent with these gain-of-function results, embryos that were injected with the xYAP MO cocktail (40 ng) exhibited a complete loss of pax3 expression ( Figure 6B). The loss of pax3 in the neural crest progenitors, but not in the hatching gland precursors, could be rescued with xyap mRNA ( Figure 6B).
In both frog and chick, the underlying mesoderm plays an important role in specifying the tissues derived from the neural plate border zone (reviewed in [54]. Therefore, it is possible that the expansion of the pax3 + neural crest is secondary to YAP effects in the mesoderm. A recent study showed that it is the intermediate mesoderm, which lies directly underneath the neural plate border zone during neurulation, that is required for induction and maintenance of neural crest fate [55]. Therefore, we tested whether YAP activity in the intermediate mesoderm causes expansion of pax3 neural crest expression by targeting xyap-mRNA injections to the blastomere lineage that gives rise to the intermediate mesoderm but does not contribute to the neural crest (blastomere V21, [53]). In these embryos, pax3 + neural crest expression was expanded in only 1/25 embryos (Figure 6C), supporting the suggestion that YAP directly affects pax3 expression.
Much of the in vivo transcriptional co-activator activity of YAP results from interactions with members of the TEAD transcription factor family [56]. Recently, Naye et al. characterized two Xenopus TEADs, xtead1 (xn-tef) and xtead3 (xd-tef) [17]. Injection of xtead1 (100 pg) mRNA alone expanded the pax3 + neural crest progenitors, while a low dose of xyap (100 pg) mRNA alone had little effect ( Figure 6D). However, upon co-injection of equal amounts of xtead1 (100 pg) and xyap (100 pg) mRNAs, the percentage of embryos with an expansion of pax3 + neural crest progenitors was greatly increased (Figure 6D), indicating cooperativity between these proteins. Although TEAD gain-of-function alone expanded the pax3 + neural crest progenitors, the above experiments show that YAP enhances this effect, and the YAP MO experiments indicate that YAP is required for this effect. Therefore, we predicted that YAP acts as a transcriptional co-factor with xTEAD1 in regulating pax3 expression.
Results from a series of pax3 promoter transgenic deletions led Milewski et al. to suggest that a TEAD-binding site within a neural crest enhancer region was responsible for neural crest expression of pax3 [57]. However, we failed to find conservation of this previously described TEAD-binding site in the Xenopus tropicalis genome ( Figure 7A). Using the genomic alignment and conserved transcription factor binding site prediction program, ConTra [58], a predicted TEAD-binding site that was highly conserved in 15 different vertebrates was identified 58 base pairs upstream of the previously described mouse neural crest enhancer TEAD2binding site ( Figure 7A). To demonstrate direct involvement of xYAP in the control of pax3 transcription, we performed a ChIP analysis of the xpax3 promoter from wild-type stage 14-16 Xenopus laevis embryo DNA that was sheared to an appropriate size ( Figure 7B). Using primers made specifically to amplify the genomic region containing the conserved TEAD-binding site (yellow box in Figure 7A), endogenous xYAP co-immunoprecipitated with this region, illustrating the direct involvement of xYAP in regulating xpax3 transcription ( Figure 7C). This TEAD-binding site was specific since primers to another portion of the pax3 promoter were not pulled down with the YAP antibody ( Figure  S4A). Likewise, a region of the sox2 promoter, which possesses a putative TEAD-binding site, also failed to be pulled down with the YAP antibody ( Figure S4B). To confirm the presence of the TEAD-binding site within the YAP chromatin-immunoprecipitated piece of Xenopus laevis genomic DNA, a proofreading Taq polymerase was used to amplify and subclone the product ( Figure 7D). Interestingly, the conserved TEAD-binding site, but not the proposed mouse TEAD2-binding site, was located in this amplified fragment ( Figure 7D).

PDZ-binding motif of xYAP plays a role in epidermal and muscle differentiation
To better define which protein-protein interaction domain of xYAP is responsible for the expansion of the neural plate and neural crest progenitors as well as the correlative inhibition of neural, hatching gland, PPE, epidermal, and somitic muscle differentiation, we performed a series of structure-function analyses whereby mutant forms of xYAP ( Figure 8A) were expressed on one side of the embryo ( Table 1). The constitutively active form of xYAP (cActive xYAP), in which the LATS phosphorylation site is mutated rendering the protein unable to exit the nucleus, caused expansion of neural plate (sox2) and neural crest (pax3) progenitors and reduction of pax3 + hatching gland precursors at frequencies comparable to wild type xYAP ( Figure 8B, C). These results confirm that the nuclear activity of YAP is required for these phenotypes. In contrast, deletion of the TEAD-binding site (TBS), WW domains, N-terminus, or C-terminus each resulted in a dramatic (sox2) to moderate (pax3-NCP, pax3-HG) reduction in the frequency of the respective phenotypes, indicating that an intact protein is required. These results implicate the importance of multiple binding partners, not solely the interaction with TEAD. In contrast, loss of neural plate differentiation (p27 xic1 ) and formation of the PPE (sox11) were maintained at high frequencies with each xYAP mutant, indicating that interactions at one or more of the remaining domains are sufficient to downregulate these genes. Interestingly xYAP-mediated loss of epidermal (cyto-keratin) and somitic muscle (myoD) differentiation were specifically reduced by deletion of its PDZ-binding motif. These results implicate the involvement of a PDZ-containing interacting protein in the effects on these two tissues. The requirements for different YAP domains for the effects on these diverse embryonic tissues indicate that different binding partners are likely to mediate them.

YAP is well conserved
Through evolution, proteins within the WW domain-containing family have functionally diversified. Although no YAP homologue exists in yeast, its closest YAP relative, Rsp5, is a WW-containing protein exhibiting ubiquitin ligase activity. The Drosophila YAP homologue, Yorkie, exhibits little sequence conservation when aligned with its vertebrate YAP counterparts, especially at its Cterminal end where this protein lacks the conserved vertebrate transcriptional activation domain and the SH3-and PDZ-binding motifs. However, other invertebrates, such as the acorn worm, honeybee, wasp, sea anemone, sea urchin, and sea squirt, which also exhibit low vertebrate YAP identity (,40%), do possess the PDZ-binding motif. In order to utilize frog and fish to elucidate a common functional role in vertebrate development, it is important to establish that the YAP proteins in these animals contain similar functional domains. Indeed, xYAP and zYAP are 78% identical to the mouse homologue and contain all of the functional domains described in mammals (see also [36], [40]). Interestingly, the prolinerich region present at the N-terminus of the human homologue, which allows for hnRNP U binding, contains fewer prolines in nonmammals (human, 18; mouse, 15; frog, 6; zebrafish, 3).
The functional diversity of YAP in vivo, however, is just now beginning to be unraveled. In particular, there is a paucity of information regarding its function in early vertebrate developmental processes. Previously, we reported that mice lacking YAP exhibit severe developmental phenotypes that result in early lethality [15]. Given that the A-P axis defects may result from the extra-embryonic tissue defects, we exploited two more amenable models, Xenopus laevis and Danio rerio, to investigate the function of YAP during early development. Herein, we provide the first description of the mechanism by which YAP regulates the completion of gastrulation and the elongation of the A-P body axis. This protein is required for the proper timing of expression of early mesodermal genes, and for the expansion of the sox2 + neural plate and pax3 + neural crest progenitors at the neural plate border. We demonstrate that the effects of YAP, a transcriptional coactivator, on pax3 + neural crest progenitors are accomplished, at least in part, by co-regulation of the pax3 gene via interaction with the transcription factor, TEAD1.

YAP is required for progression through gastrulation
We show that MO-mediated elimination of YAP in vivo resulted in a failure of frog and fish embryos to complete the epiboly movements of gastrulation, and a lower MO dose knockdown reduced the elongation of the A-P body axis (this study, and [36]). While germ layer inductions occurred in the absence of endogenous, zygotic xYAP protein expression, the onset of mesodermal gene expression was perturbed, indicating that YAP is required during the early steps of body axis formation. While the mechanism is not known, these results demonstrate that the similar defects described in mice lacking YAP are not simply due to nutritional deficiencies.

Increasing YAP expands progenitors and inhibits their differentiation
When wild type xyap, myap, or hyap RNAs were injected into Xenopus laevis embryos, major morphological defects became apparent at tail bud stages. Because the tissue perturbations were widespread, we predicted that gene expression changes occurred during earlier patterning events. In fact, we observed that at neural plate stages two progenitor populations were expanded (sox2 + neural plate; pax3 + neural crest), whereas differentiation markers of these tissues as well as of somitic muscle and epidermis were repressed. These results are consistent with reports that overexpression of YAP expands mouse small intestinal progenitors [44] and chick neural tube sox2 + neural progenitors [41]. The mechanism by which the expansion of neural progenitors in frog embryos is accomplished is not yet known. The expansion of mouse intestinal progenitors is mediated by activation of the Notch pathway by YAP [44]; however, frog embryos injected with xyap mRNA showed reduced notch and hes1 RNA expression. Interestingly, xYAP did not expand all progenitors or all pax3expressing cells. YAP gain-of-function inhibited pax3 expression in hatching gland precursors, and reduced the expression of six1, a transcription factor that maintains the PPE in a progenitor state [52,59]. These results demonstrate that YAP-mediated expansion of progenitor populations has tissue specificity, even within the embryonic ectoderm. The mechanisms by which YAP gain-offunction and loss-of-function both result in a shortened axis are not known. One possibility is that YAP loss-of-function leads to a loss of the progenitor pool, as we show for sox2 and pax3, whereas YAP gain-of-function extends the time cells spend in a progenitor, undifferentiated state. In both cases, there would be a reduction in differentiated cells.

xYAP directly regulates pax3 transcription
The effects of altering YAP levels on pax3 expression in the neural crest progenitors suggested that YAP directly regulates pax3 transcription. Increasing evidence suggests that the interaction of YAP with the TEAD family of transcription factors is critically important for proper vertebrate development, [34,37,41,60], including neural progenitor expansion [41]. Therefore, we searched for highly conserved TEAD-binding sites in the 59 regulatory Figure 8. xYAP deletion mutants exhibit differential activities. (A) Cartoons of the xYAP mutants created to determine which protein-protein interaction domain(s) is important for the in vivo gain-of-function phenotypes described in Figures 4-6. Deletions or mutations are indicated by color loss: the TEAD-binding site (xYAPDTBS, purple), the LATS phosphorylation site (cActive xYAP, orange), the two WW domains (xYAPDWW, red), the N-terminus (xYAP, DN-term) containing both the hnRNP U and TEAD-binding sites, and the PDZ-binding motif (xYAPDC-term, fuchsia) at the C-terminus. (B) The percentage of embryos showing expansion of sox2-expressing neural plate cells or expansion of pax3-expressing neural crest progenitor (NCP) cells after injection of each of the mutant forms of xYAP. Note that cActive xYAP, which prevents YAP from leaving the nucleus, is as effective as wild type YAP. However, all other mutant forms reduce this phenotype. Sample sizes are presented in Table 1. (C) The percentage of embryos showing reduced gene expression after injection of each mutant form of xYAP. Deletion of the WW domains or of the PDZ-binding motif interfered the most with repression of pax3 + hatching gland (HG) progenitors. Loss of neural plate differentiation (p27 xic1 ) and a PPE marker (sox11) were maintained at high frequencies with each xYAP mutant, indicating that interactions at one or more of the remaining domains are sufficient to downregulate these genes. However, xYAPmediated loss of somitic muscle (myoD) and epidermal (cyto-keratin) differentiation was specifically reduced by deletion of its PDZ-binding motif. doi:10.1371/journal.pone.0020309.g008 Table 1. xYAP deletion mutants exhibit differential activities. region of pax3 and found a previously undescribed, TEAD-binding site within this region. Increased expression of TEAD1 phenocopied the xYAP-mediated expansion of pax3 in the neural crest progenitors and significantly enhanced this phenotype following coexpression of xtead1 with levels of xyap mRNA that were ineffective on their own. Importantly, we demonstrated the in vivo relevance of this predicted association by ChIP analysis. Endogenous YAP localized to this newly identified TEAD-binding site within the 59 regulatory region of pax3, but not to a region of the pax3 promoter lacking putative TEAD-binding sites ( Figure  S4A). In addition, a region of the sox2 promoter containing a putative TEAD-binding site also did not co-immunoprecipitate with YAP ( Figure S4B), even though in chick neural tube YAP and TEAD1 appear to co-regulate sox2 expression [41].
We have yet to confirm whether endogenous TEAD1 resides on this region or whether other TEADs are present. For example, there is evidence that TEAD1 and TEAD2 can functionally compensate for each other in early mouse development [61,62]. Nonetheless, these experiments demonstrate a new developmental role for both TEAD and YAP in cooperatively driving pax3 expression in neural crest progenitors. Our structure/function analyses, however, indicate that the expansion of the pax3 neural crest progenitors likely involves YAP binding to proteins in addition to TEADs, because deletion of other domains also reduced this effect. In fact, the different effects of YAP on different ectodermal and mesodermal genes appear to require different protein interaction domains, confirming that the ability of YAP to bind to multiple, diverse proteins endows this protein with multiple diverse functions. Here, we have illuminated a few key developmental roles for YAP, which appear to be consistent across three vertebrates. Moving forward, it will be interesting to see whether it is YAP's transcriptional activation abilities or its function as a scaffolding protein that is more important for each specific effect. An intriguing notion is that YAP may act as a critical scaffolding protein within the nucleus to assist in the regulation of transcription or regulate the state and/or remodeling of chromatin. Figure S1 mRNA and protein expression of xYAP during Xenopus laevis development. RT-PCR analyses showed that xyap RNA was maternally expressed in an unfertilized egg and early cleavage (stage 3), decreases slightly between late cleavage (stage 6) and the mid-blastula transition (stage 9), but was then expressed abundantly through subsequent stages of Xenopus laevis development through feeding tadpole (stage 40). The (+) indicates lanes that included reverse transcriptase in the RT-PCR reaction, while the (2) indicates lanes that lacked the reverse transcriptase in the RT-PCR reaction. Western blot analysis showed that xYAP protein was maternally present at cleavage stages (stages 2-7), was detectable at the onset of epiboly and gastrulation (stages 9-10), and increased dramatically from mid-gastrula (stage 11) onwards. The (+) represents the positive control lane, which contains a cold in vitro translated xYAP product.