The three-dimensional organization of multicellular organisms requires morphogenetic tissue rearrangements, which are particularly evident during embryonic development or in pathological conditions, such as wound healing or cancer. These rearrangements are largely coordinated by the interplay among cell-cell interaction, mechanical forces, and signaling pathways, derived, in part, from the surrounding environment. The adaptive morphogenetic outcome critically depends on how the cells integrate and interpret this signal complexity and transmit the message to their nuclei, eliciting a consequent transcriptional response (Wozniak and Chen, 2009).

In several organisms, including the teleost zebrafish, epiboly is the first morphogenetic rearrangement of the gastrulating embryo (Warga and Kimmel, 1990). These cellular rearrangements are relatively simple and thus have been used to understand basic principles of morphogenesis (Kimmel et al., 1995; Bruce, 2016). The zebrafish blastoderm, positioned at the animal pole, is formed by a single layer of loosely packed cells, known as deep cell layer (DEL), which is covered by the so called epithelial enveloping layer (EVL). The blastoderm interfaces with the yolk syncytial layer (YSL), itself a derivative of the blastoderm marginal cells. During epiboly, the blastoderm thins and moves towards the vegetal pole together with the external YSL (E-YSL). These epithelial spreading and thinning are driven, in part, by a circumferential actomyosin network formed by the E-YSL at its border with the EVL. The contraction of this actomyosin ring pulls the EVL to surround the yolk, further dragging the DEL toward the vegetal pole. This process culminates at the end of gastrulation with “blastopore closure” when the blastoderm seals around the yolk (Kimmel et al., 1995; Bruce, 2016).

In zebrafish, gastrulation commences during maternal-to-zygotic transition, a pivotal process marked by the activation of the zygotic genome but with a substantial involvement of maternal transcripts/proteins that endure throughout gastrulation, actively participating in its morphogenesis (Solnica-Krezel, 2020). Several studies have shown that teleost epiboly involves the function of the transcriptional co-activator Yes1 Associated Transcriptional Regulator (Yap1) and of the highly related TAZ (Gee et al., 2011; Hu et al., 2013; Porazinski et al., 2015; Vázquez-Marín et al., 2019). Yap1 was first described as a downstream nuclear effector of the Hippo signaling pathway, which is involved in the control of cell proliferation (Zhao et al., 2007). However, there is evidence that Yap1 can operate, perhaps independently of Hippo signaling, as a mechano-transducer that links mechanical forces to cellular response (Dupont et al., 2011). Cells constantly probe the forces generated in their extracellular environment through plasma membrane adhesive proteins and internal tension adjustments executed by modifications of their actomyosin network. Upon increasing intracellular tension, Yap1, which is normally retained in the cytoplasm, translocates to the nucleus, where it binds to TEA domain (TEAD)-containing transcription factors (TEAD1–TEAD4). The derived complex activates the transcription of target genes, which include F-acting regulators and cell adhesion components, thereby promoting a positive feedback loop that enables cells to counterbalance the external mechanical forces to which they are exposed (Panciera et al., 2017). Consistent with such mechano-transducer activity, yap1 has been shown to regulate medaka fish epiboly progression by controlling the expression of arhgap18, a gene encoding a Rho GTPase activating protein involved in the contraction of the actomyosin ring at the E-YSL/EVL interface (Porazinski et al., 2015). How yap1 activity is controlled and whether this is exerted by maternally or zygotically derived factors remained unanswered.

Vestigial-like protein 4 (Vgll4) is a transcriptional cofactor that can bind directly to TEAD proteins, via its tandem Tondu domain (TDU), thereby competing for the Yap-TEAD interaction and thus mechanistically acting as an inhibitor of Yap activity (Koontz et al., 2013; Jiao et al., 2014; Zhang et al., 2014; Li et al., 2023). In Xenopus and zebrafish the transcripts of the different vgll4 paralogs show a dynamic developmental expression pattern in different tissues (Faucheux et al., 2010; Barrionuevo et al., 2014; Xue et al., 2018) and in zebrafish the mRNA of two (Xue et al., 2018) (vgll4a; vgll4b) of the three paralogs (vgll4a, vgll4b and vgll4l) have been detected at the 1 cell stage (Xue et al., 2018), suggesting a possible early function. Although there are additional and species-specific phenotypes that involve later developing organs (Fillatre et al., 2019; Xue et al., 2019; Wang et al., 2020), loss of function studies in mice and analysis of genetic variants in the teleost Scatophagus argus suggest that Vgll4 participates in the regulation of body size (Feng et al., 2019; Suo et al., 2020; Yang et al., 2020; Sheldon et al., 2022), a trait that is often altered as the consequence of an abnormal gastrulation.

By generating mutants for the three vgll4 paralogs in zebrafish, here we have explored if vgll4 could be involved in controlling yap1 activity during epiboly. We show that maternal but not zygotic vgll4a is required to sustain yap1 signaling and E-cadherin/β-catenin adhesion complex, thereby promoting cell cortex configuration and actomyosin contraction, which drive epiboly progression and, ultimately, an appropriate elongation of the embryonic antero-posterior axis.

The genome of a fertilized egg is transcriptionally inactive. Thus, maternal RNAs and proteins present in the eggs are responsible for coordinating the first morphogenetic events that takes place during gastrulation, initially single-handed and then in cooperation with the progressively available zygotic gene products (Solnica-Krezel, 2020). In the zebrafish egg, these maternally derived molecules represent a large proportion of all possible gene products (Harvey et al., 2013). However, how many of them are critical for epiboly initiation and progression is still poorly defined. Genome wide (Kane et al., 1996) and specific maternal-effect mutational screens in zebrafish have identified only a limited number of mutants with an epibolic phenotype (Dosch et al., 2004; Wagner et al., 2004) and their subsequent characterization supports an important contribution of adhesion molecules and cytoskeletal components (Bruce, 2016). Other studies have thereafter identified a few maternally inherited transcriptional regulators, including eomes/tbr2, foxh1, pou5fl/oct4, nanog and yap1 (Bruce et al., 2005; Reim and Brand, 2006; Pei et al., 2007; Lachnit et al., 2008; Song et al., 2013; Porazinski et al., 2015; Gagnon et al., 2018; Veil et al., 2018). In this still fragmented scenario, our study adds a new component to the genetic network coordinating zebrafish epiboly, showing an important role for maternally inherited vgll4a, and, to some extent, for its paralog vgll4b. Notably, it also shows that vgll4a is required to support yap1 signalling and the expression of E-cadherin/β-catenin adhesion complex (directly or indirectly), thereby coordinating blastomere adhesion/cohesion. This observation aligns with recent studies demonstrating the versatility of Vgll4 in various biological contexts, extending beyond its established role as a Yap1 inhibitor (Fillatre et al., 2019; Xue et al., 2019; Quan et al., 2023; Wang et al., 2023).

Zebrafish epiboly can be divided in two phases: initiation or doming and progression. Our study underscore vgll4a requirement at doming with morphological consequences that become visible as epiboly progresses and later as the embryonic A-P axis extends. Our findings are substantiated by the distribution of tbxta, a marker for margin cells, at 4.5 and 8hpf. Additionally, disrupting myosin activity precisely at the 4hpf stage replicates the MZvgll4a mutant phenotype, both during epiboly and in larval stages. This observation establishes a functional link between these two events. A recent study strongly supports this possibility demonstrating reduced Xenopus A-P axis extension when tissue scale force production is wakened during gastrulation (Huebner et al., 2022). Furthermore, a short axis phenotype was also observed after interference with myosin activity in MZalkbh4 and Malkbh4 mutants (Sun et al., 2017), in which defective convergent extension movements were also reported. Thus, it is plausible that similar gastrulation defects may contribute to the phenotype observed in MZvgll4a embryos, although we have not investigated this possibility.

The large majority of factors implicated in doming are maternally expressed, as both vgll4a and vgll4b are (Xue et al., 2018). The time-lag between the postulated molecular activity of maternal vgll4a at doming and the phenotypic consequences about 4 hours later is also not surprising, as a similar delay has been reported for other factors contributing to epiboly initiation (Lachnit et al., 2008; Sun et al., 2017). A recent study has shown that at sphere stage the central blastula becomes “fluid” as a consequence of loss of cell-cell adhesion and increased cell division (Petridou et al., 2019). The blastoderm margins instead maintain cell-cell adhesion and thus tension thanks to the activity of wnt11 non-canonical signalling (Petridou et al., 2019), known to control E-cadherin availability at the plasma membrane (Ulrich et al., 2005). If this differential fluid vs. tense state is perturbed, the ability of the blastula to react to mechanical forces is altered and epiboly becomes defective (Petridou et al., 2019). In our experiments we have not detected changes in wnt signalling using the global expression of three different downstream targets. Furthermore, vgll4a role in epiboly seems to be strictly maternal (Solnica-Krezel, 2020) and thus could act upstream of wnt11 signalling. Vgll4a-mediated control of wnt signalling will impinge on E-cadherin/β-catenin availability at the plasma membrane and thereby on differential blastoderm viscosity that favours epiboly progression. In support of this possibility, vgll4 paralogs appear to control the expression of non-canonical wnt signaling components, including wnt11f2, fzd8a and fzd10 (Fillatre et al., 2019). Nonetheless, we cannot exclude that very local changes in wnt pathway activation may occur, thereby explaining the MZvgll4a mutants phenotype as proposed by Petridou et al. (Petridou et al., 2019). Alternatively, vgll4a could interfere with non-canonical wnt signaling indirectly through the regulation of wnt/β-catenin canonical pathway, as the two branches of the pathway can have antagonistic effects in morphogenesis (Cavodeassi et al., 2005). Indeed, VGLL4 seems to interfere with the formation of a TEAD4-TCF4 complex, which promotes cell proliferation in colorectal cancer (Jiao et al., 2017). Furthermore, VGLL4 overexpression suppresses nuclear β-catenin levels and inhibits migration and invasion of gastric cancer cells, while its inactivation has opposite effects (Li et al., 2015). As an additional possibility, vgll4a could directly regulate E-cadherin expression as observed in both gastric and breast cancers (Li et al., 2015; Song et al., 2019), thereby influencing the membrane levels of E-cadherin/β-catenin complexes. According to our data the latter possibilities are however less likely. We observed a small, although significant, decrease of E-cadherin membrane localization only in EVL cells of MZvgll4a embryos at sphere stage, and no significant changes in the mRNA levels of cdh1 or in those of three read-out of wnt/β-catenin signalling, but our analysis, performed with entire embryos, precludes the identification of changes present only in a small cell subset. Furthermore, we cannot rule out that the MZvgll4a phenotype might be associated with abnormal E-cadherin stability or trafficking, as reported for the zebrafish wnt11f2 and oct4/MZspg mutants (Song et al., 2013; Petridou et al., 2019). This destabilization, in turn, could impinge on the actin cortex, which seems incomplete in most EVL cells (Figure 6), and thereby on the integrity of tight junctions. Although we have not addressed this point, the morphology of the EVL layer in MZvgll4a mutants resembles that of embryos treated with the dynamin inhibitor Dynasore (Lepage et al., 2014) or that reported for E-Cadherin mutants (Lachnit et al., 2008).

We focused on vgll4a and report that the related vgll4l has no role in epiboly, consistent with its lack of early expression (Xue et al., 2018) and its suggested functional diversification (Fillatre et al., 2019). On the contrary, we show both vgll4a and vgll4b seem to be important for oocyte fecundity and that MZvgll4b and MZvgll4a mutants present a very comparable A-P axis phenotype. However, in contrast to what observed for vgll4a, maternal vgll4b activity can be compensated by its zygotic counterpart. Furthermore, MZvgll4a;MZvgll4b double mutants are phenotypically very comparable to the two single mutants, and the additional inactivation MZvgll4l does not significantly worsen the short axis phenotype, compatible with the absence of maternal vgll4l expression. This suggests that the different paralogs do not compensate each other function and that vgll4a and vgll4b perhaps act in a different spatio-temporal window that converges in the same shorter axis phenotype. At later stages of development, vgll4b has been implicated in erythropoiesis, terminal differentiation (Wang et al., 2020), heart valvulogenesis (Xue et al., 2019) and establishment of the left-right asymmetry in coordination with vgll4l (Fillatre et al., 2019). Our vgll4a mutants do not show similar traits, at least in a gross morphological analysis, further supporting a possible diversification of the paralogs’ activity.

Vgll4 has been initially described as co-transcriptional repressor that interacts with Tead transcription factors, thereby acting as an as antagonist of Yap signalling (Guo et al., 2013; Zhang et al., 2014). This function has been thereafter validated in multiple contexts especially in tumour development (Yamaguchi, 2020). Our study not only shows that maternal vgll4a is required for epiboly but also that does so supporting rather than antagonising Yap1 signalling. The generation of a transgenic Yap/Taz-Tead reporter zebrafish line faithfully highlights pathway activation from 22 hpf (Miesfeld and Link, 2014). Using this same line we have been unable to detect reporter expression at epiboly stages. However, we show that overexpression of YAP is sufficient to rescue the phenotype of MZvgll4a mutants, suggesting that, likely, maternal vgll4a acts upstream of Yap1, given that ccn1 expression is decreased in MZvgll4a mutants. Furthermore, addition of the Yap inhibitor verteporfin to wt embryos replicates the MZvgll4a phenotype but has no effect on the MZvgll4a mutants themselves.

Our analysis has been performed at the whole embryo scale, taking epiboly progression, actomyosin organization and cell cohesion as read-out of vgll4a function in morphogenetic rearrangement. These rearrangements highly depend on mechanical and biochemical events at the cellular and subcellular scale (Petridou et al., 2017; Stooke-Vaughan and Campàs, 2018). Therefore, in this scenario, we cannot exclude that the delay in epiboly progression observed in MZvgll4a embryos represents the macroscopic result of a series of mechano-chemical feedback loops at the cellular scale (Hannezo and Heisenberg, 2019), involving both vgll4a and yap1 in the same or even different cell populations.

Nevertheless, and independently of the relative position of vgll4a and yap1 in epiboly regulation, our data suggests that vgll4a may be required for a proper balance between tissue tension/cohesion and contractility, two factors that contribute to mechanical stress (Wozniak and Chen, 2009). Indeed, decreasing myosin contractility at doming mimics the absence of maternal vgll4a function. We thus hypothesis that in the absence of vgll4a, cellular cohesion mediated by adhesion complexes is poor, impairing embryonic mechano-sensation (Figure 7). This impairment would decrease yap-dependent mechano-transduction, which, in turn, could affect the contraction of the actomyosin network, delaying epiboly progression (Figure 7). In support of this model, the yap mechano-regulatory program seems to be essential for sustaining intracellular tension during gastrulation, that, in turn, controls the assembly of the embryonic A-P axis (Sousa-Ortega et al., 2023). Our simplified model does not take into account other and yet undetermined factors that contribute to this process. For example, we were unable to rescue the speed of epibolic movements in MZvgll4a embryos with calyculin, a drug that increases myosin 2 contractility, either because there are other compensatory mechanisms or because we have not chosen the appropriate time window.

Independently from these considerations and in a broader context, our data suggest that upregulation of vgll4 expression may serve to enhance the mechano-sensing properties of some tissues, perhaps restoring an unbalanced back-and-forth dialogue between biochemical and mechanical cues, which has been described in pathological conditions, including different type of primary and metastatic cancers (Deng and Fang, 2018).

Maintenance of fish lines. Adult AB/TUE wild type, mutant (vgll4a, vgll4b, vgll4l) and 4xGTIIC:eGFP reporter line (Miesfeld and Link, 2014) zebrafish were maintained at 28°C on 14/10 h light/dark cycle. Embryos were raised at 28°C and staged according to the hours post fertilization (hpf) and their morphology. Embryos were growth in E3 medium (NaCl, 5 mM; KCl, 0.17 mM; CaCl₂, 0.33 mM; MgSO₄, 0.33 mM; 5.10% Methylene Blue). The ethical committee for Animal Experimentation of the Consejo Superior de Investigaciones Científicas (CSIC) and of the Comunidad Autónoma de Madrid approved the procedures used in the study.

Zebrafish mutants generation. Zebrafish mutant lines were generated using CRISPR/Cas9 technology. The gRNA were designed using the CHOP-CHOP tool (Labun et al., 2019) looking for potential disruption of enzyme restriction sites. gRNAs were synthesized using PCR templates as described in (Varshney et al., 2016). Cas9 protein (300 ng/μL; EnGen^® Spy Cas9 NLS, New England Biolabs) and gRNAs (100 ng/μL; Table S1) were co-injected into one-cell stage zebrafish embryos. F0 embryos were raised and outcrossed with AB/TUE wt. PCR amplification on genomic DNA isolated from tail clips of F1 zebrafish embryos was performed to identify disruption of specific restriction sites (Table S2). The DNA of potential mutants was thereafter sequenced and the selected embryos were raised to adulthood. We selected mutants in which the reading frame was disrupted and truncated as follow: vgll4a-S85Mfs15, vgll4b-P156Rfs5 and vgll4l P111Hfs61 (see Supplementary Figure S1 for more details).

Embryo injections. Zebrafish embryos were injected at the 1 cell stage with 1 nL of the selected solutions using a IM-300/Narishige microinjector. Recombinant humanVGLL4 protein (0.19 ug/μL; Abnova, Ref: H00009686-P01) and vgll4a-HA mRNA (100 ng/μL) were used in rescue experiments. Phosphate buffer saline (PBS) containing 0.1% BSA or mutated vgll4a-HA mRNA (cloned from the vgll4a-S85Mfs15 line) were injected as controls. Myc-YAP, Myc-YAP-5SA and eGFP mRNAs were injected at 25 ng/ul.

Tissue processing and immunohistochemistry. Embryos were fixed by immersion in 4% paraformaldehyde in 0.1 M phosphate buffer pH 7.2 (wt/vol) overnight at 4 °C. Embryos were then washed in PBS with 0.5% Triton-X-100, incubated in a 15% sucrose-PBS solution (wt/vol), embedded and frozen in a 7.5% gelatin in 15% sucrose solution (wt/vol). Cryostat sections or whole embryos were stained using standard protocols and antibodies against the following antigens: HA (1:250, Sigma, H-6908), β-catenin (1:300, BD Bioscience, 610,154), E-cadherin (1:300, BD Bioscience, 610,181) and Yap (1:200, Cell Signaling Technology, #4912). Incubation with Phalloidin-TRITC (1:200, Sigma, P1951) was used to detect actin. Incubation with appropriate secondary antibodies was performed following standard procedures.

In situ hybridization (ISH). A specific tbtxa probe was generated by RT-PCR from cDNA of 6hpf embryos using the following primers (5′-GAT​CGG​AAA​TAT​GTC​TGC-3′ and 5′-GTTGTCAGTGCT GTGGTC-3′) with the Expand™ High Fidelity PCR System (Roche) and cloned with the StrataClone PCR Cloning Kit (Agilent). myod1 RNA probe was synthesized from cDNA encoding myod1 (Weinberg et al., 1996). Digoxigenin-UTP-labelled antisense riboprobes were synthesized in vitro using DIG RNA labelling Mix kit (Roche). Zebrafish embryos at different developmental stages were hybridized by standard procedures and visualized with NBT/BCIP (dark blue).

Construct generations. vgll4a was amplified from cDNA of wt or MZvgll4a (as control) embryos using the following primers (Fw: 5′-GGA​ATC​AAC​AGT​TAG​CGT​GCT-3’; Rv: 5′-aaC​TCG​AGT​CAA​GCG​TAA​TCT​GGA​ACA​TCG​TAT​GGG​TAA​GAC​TGA​CCA​ACA​TGA​TTG-3′). The amplicon, includes the hemagglutinin (HA) epitope included in the reverse prime, was cloned by TA-cloning in the pSC-A vector using the StrataClone PCR Cloning Kit (Agilent). The vgll4a-HA fragment was thereafter liberated from the pCS-A construct using EcoRI/XhoI digestion and cloned in the pCS2 vector. The pCS2 construct, linearized with NotI, was used to synthesize mRNA from the wt or mutated vgll4a version, tagged with HA, using the mMESSAGE mMACHINE SP6 Transcription kit (Invitrogene) following manufacturer’s instructions. After transcription, mRNAs were purified using the NucleoSpin^® RNA Clean-up kit (Machery Nagel). pQCXIH-Myc-YAP (Addgene plasmid # 33091) and pQCXIH-Myc-YAP-5SA (Addgene plasmid # 33093) (Zhao et al., 2007) were used to clone the Myc-tagged forms of human YAP in the pCS2 vector, after NotI (Takara) and EcoRI restriction digestion. The open pCS2 vector and the fragments were ligated using T4 DNA Ligase (NEB) and the resulting constructs used to synthesize the corresponding mRNA.

Quantitative RT-PCR analysis. Total RNA was isolated from wt and mutant embryos (n = 30, for each genotype) using TRIzol (Sigma) according to the manufacturer´s instruction. Each experiment was performed with biological triplicates. 5 μg of total RNA was used to synthesize the first-strand cDNA using the First-Strand cDNA synthesis kit (GE Healthcare) with a pd(N)₆ primer. Each quantitative RT-qPCR reaction was performed using the GoTaq qPCR Master Mix kit (Promega). For a 10 µL reaction, 4 µL of cDNA (2.5 ng/μL) was mixed with 1 µL of primers (2.5 µM; Table S3) and 4 µL master mix. Reaction was incubated at 95°C for 10 min, then at 95°C for 15 s and 40 cycles and at 60°C for 60 s. The levels of the eef1a1l1 mRNA were used as housekeeping reference (Xu et al., 2016).

Pharmacological treatment. Embryos were incubated for different time-windows at 28°C in E3 medium containing appropriate volumes of DMSO alone or DMSO containing either Blebbistatin (203,390, Calbiochem) Calyculin A (BML-EI192, Enzo). or Verteporfin (5,305, Tocris). Treatment with 50 µM Blebbistatin was carried out for 30 min using three different time-windows (3, 4 and 5 hpf). Verteporfin (5 µM) was added to embryos at the 1 cell stage up to 75% epiboly (8 hpf). Calyculin A was used at four different concentrations (0.2 µM, 0.5 µM, 0.7µM and 1 µM) and added at 4hpf for 30 min.

Imaging. Sections were analysed with DM or confocal microscope. Zebrafish embryos at different stages were observed and photographed using a stereomicroscope and DFC500, DFC350 FX cameras (Leica Microsystems). For sections or whole embryo staining, LSM710 confocal laser scanning microscope coupled to an AxioObserver inverted microscope (Zeiss) were used to obtain digital images, which were then processed and analysed with ImageJ (Fiji) software. Images shown in Figure 1, Figure 4, Figure 5, Fig. S1, S4 and S6 were assembled using the Photoshop CS5 software.

Quantification and statistical analysis. All quantifications were performed using the ImageJ (Fiji) software. 72hpf embryonic larvae were treated with tricaine and photographed in lateral views at ×13 magnification to determine their body length. For epiboly progression quantifications, embryos hybridized with a tbxta probe were photographed and the total length of the egg and the distance between the margin cells and the vegetal pole were measured. Blastopore closure quantification was performed as previously reported (Köppen et al., 2006). The area of individual and centrally positioned blastoderm cells from wt and mutant embryos was determined by tracing their perimeter on confocal images of whole embryos stained with β-catenin antibody. Only one image per embryo was used. The position and width of the actomyosin ring was determined on z-stack files obtained from the confocal images of whole embryos. The z-stack files were used to generate maximum intensity projections. The mid-region of the actomyosin ring and DEL/EVL margins were positioned in the centre of the image to avoid possible visual distortions due to the embryo curvature and a rectangular region of interest (ROI) was selected covering the central area of the embryo. Ten different measurements of the width of the actomyosin ring as well as ten different measurements of the distance between DEL-EVL were obtained along the ROI using the “straight” selection tool. The ten measurement per embryo were averaged to obtain a more accurate value of the ring width and also the DEL-EVL distance and their means were used in the comparative analysis. Immunofluorescence intensity quantifications were performed using mid-plane of z-stack files obtained from the confocal imaging of β-catenin, F-actin and E-cadherin staining on whole embryos (EVL cells) or cryostat sections of different embryos (DEL cells). A rectangular ROI was used to measure the mean grey value of the selected image as indicator of signal intensity. GraphPad Prism 7 statistic software was used to analyze the data using t-test for two groups with parametric distribution and with Mann-Whitney test for two groups with no parametric distribution. One-way ANOVA test was used for more than two groups with parametric distribution with Tukey’s multiple comparisons test to analyze differences between groups. Kruskal–Wallis test for no parametric distribution with Dunn’s multiple comparisons test to analyze differences between groups. Fisher exact test was used in the analysis of Figure 1G. Statistical difference between pools of treated and control embryos was determined with the Pearson´s X2 test. Error bars indicate s. e.m. in all graphs.