Keratins coordinate tissue spreading by balancing spreading forces with tissue material properties Authors: Suyash Naik, Yann-Edwin Keta, Kornelija Pranjic-Ferscha, Édouard Hannezo, Silke Henkes, Carl-Philipp Heisenberg Published: 2026 ABSTRACT For tissues to spread, they must deform while staying intact. How spreading tissues balance flexibility with integrity is not yet well understood. Here, we show that keratin intermediate filaments adapt tissue mechanical resilience to the stresses arising in epithelial tissues during spreading. By analyzing the expansion of the enveloping cell layer (EVL) over the yolk cell in zebrafish embryos in vivo, we find that keratin network maturation in EVL cells is promoted by stresses building up within the spreading tissue. Through genetic interference and tissue rheology experiments, complemented by a vertex model with mechanochemical feedback, we demonstrate that stress-induced keratin network maturation in the EVL increases tissue viscosity, to prevent tissue rupture. Further, keratins are required in the yolk cell for mechanosensitive actomyosin network contraction and flow, the forces pulling the EVL. These dual mechanosensitive functions of keratins enable a balance between pulling force production and EVL mechanical resilience, ensuring uniform and robust tissue spreading. Introduction Epithelial cell layer spreading is a core feature of multiple developmental and disease-related processes. In Drosophila development, for instance, the spreading of the epidermis leads to the closure of an opening at the dorsal side of the embryo1,2. Likewise, in wound healing, the epidermal cell layer spreads, and fusion closes the wound3,4. Various cellular processes have been proposed to contribute to epithelial cell layer spreading, including cell spreading, cell migration, oriented cell division, and cell intercalation5,6. These processes can generate the mechanical forces driving active tissue expansion and determine the aptitude of tissues to undergo spreading. Zebrafish embryo morphogenesis is initiated by the spreading of the blastoderm over the large yolk cell in a process named epiboly7. During epiboly, the EVL, a simple squamous epithelial cell layer, formed at the surface of the blastoderm, undergoes massive spreading to eventually engulf the entire yolk cell at the end of epiboly (Fig. 1A)8,9. EVL spreading has been shown to depend on the formation and contraction of a large actomyosin band positioned within a thin cytoplasmic layer at the surface of the yolk cell, the yolk syncytial layer (YSL)10. This actomyosin band within the YSL forms around the entire circumference of the yolk cell close to where the leading edge of the EVL contacts the YSL, and its contraction and flow are thought to generate the mechanical forces pulling the EVL over the yolk cell (Fig. 1A)10,11. Both active spreading of EVL cells and oriented EVL cell divisions have been implicated in facilitating EVL spreading by modulating EVL surface tension12,13. However, EVL morphogenesis not only relies on the generation and transmission of active forces within the EVL and YSL, but also on changes in the material properties of the tissue determining its deformation in response to such forces14. Yet, the molecular and cellular mechanisms that determine EVL material properties and how they are spatiotemporally coupled to changes in active force production remain unclear. Understanding these mechanisms requires a closer examination of the cytoskeletal components within epithelial cells, particularly those that have been implicated in modulating tissue material properties.Fig. 1Keratin network maturation in the EVL.A Left: schematic representation of a zebrafish embryo at 60% epiboly stage (6 hpf) showing the expression and localization of keratin within the EVL and YSL (together with actin). Arrows mark the direction of EVL epiboly movements. Right: cross-section of the region outlined in the left panel marking the expression and localization of keratin, actin (within the YSL) and junctional components linking the margin of the EVL to the YSL. B Plot of maximum intensity projection images of keratin (top and middle rows) and actin (Utrophin; bottom row) expression in Tg(krt18:Krt18-GFP) (keratin) and Tg(acbt2:Utrophin-mcherry) (actin) embryos showing the progression of keratin expression within the EVL and YSL during epiboly (4–9.5 hpf). Scale bar: 100 µm (top row) and 25 µm (middle and bottom row). C Plot of averaged keratin intensity within the EVL in Tg(krt18:Krt18-GFP) embryos as a function of time during epiboly (4–9.5 hpf). N = 3 experiments, n = 5 embryos. Data are presented as mean line ± standard error of the mean (SEM) ribbon. D Plot of averaged density of the keratin network in Tg(acbt2:Utrophin-mcherry, krt18:Krt18-GFP) embryos as a function of time during epiboly (4–9.5 hpf). N = 3, n = 6 embryos. Data are presented as mean line ± SEM ribbon. E Plot of averaged apical cell area of individual EVL cells in Tg(acbt2:Utrophin-mcherry, krt18:Krt18-GFP) embryos as a function of time during epiboly (4–8.5 hpf). N = 3, n = 4 embryos. Data are presented as mean line ± SEM ribbon. F 3-dimensional (3D) plot of keratin intensity, network density, and EVL cell area as a function of time during epiboly. N = 3, n = 4 embryos. Data are presented as a surface whose width represents data spread, indicating their variability (standard deviation (SD) of density, intensity and area). G Plot of junctional versus apical keratin intensity measured in individual EVL cells in Tg(acbt2:Utrophin-mcherry, krt18:Krt18-GFP) embryos at 60% epiboly stage (6 hpf) N = 3, n = 3 embryos. Data are presented as scatterplot with points corresponding to individual cells. Source data are provided as a Source Data file. Keratin network maturation in the EVL. A Left: schematic representation of a zebrafish embryo at 60% epiboly stage (6 hpf) showing the expression and localization of keratin within the EVL and YSL (together with actin). Arrows mark the direction of EVL epiboly movements. Right: cross-section of the region outlined in the left panel marking the expression and localization of keratin, actin (within the YSL) and junctional components linking the margin of the EVL to the YSL. B Plot of maximum intensity projection images of keratin (top and middle rows) and actin (Utrophin; bottom row) expression in Tg(krt18:Krt18-GFP) (keratin) and Tg(acbt2:Utrophin-mcherry) (actin) embryos showing the progression of keratin expression within the EVL and YSL during epiboly (4–9.5 hpf). Scale bar: 100 µm (top row) and 25 µm (middle and bottom row). C Plot of averaged keratin intensity within the EVL in Tg(krt18:Krt18-GFP) embryos as a function of time during epiboly (4–9.5 hpf). N = 3 experiments, n = 5 embryos. Data are presented as mean line ± standard error of the mean (SEM) ribbon. D Plot of averaged density of the keratin network in Tg(acbt2:Utrophin-mcherry, krt18:Krt18-GFP) embryos as a function of time during epiboly (4–9.5 hpf). N = 3, n = 6 embryos. Data are presented as mean line ± SEM ribbon. E Plot of averaged apical cell area of individual EVL cells in Tg(acbt2:Utrophin-mcherry, krt18:Krt18-GFP) embryos as a function of time during epiboly (4–8.5 hpf). N = 3, n = 4 embryos. Data are presented as mean line ± SEM ribbon. F 3-dimensional (3D) plot of keratin intensity, network density, and EVL cell area as a function of time during epiboly. N = 3, n = 4 embryos. Data are presented as a surface whose width represents data spread, indicating their variability (standard deviation (SD) of density, intensity and area). G Plot of junctional versus apical keratin intensity measured in individual EVL cells in Tg(acbt2:Utrophin-mcherry, krt18:Krt18-GFP) embryos at 60% epiboly stage (6 hpf) N = 3, n = 3 embryos. Data are presented as scatterplot with points corresponding to individual cells. Source data are provided as a Source Data file. Keratin intermediate filaments are the most abundant and diverse cytoskeletal components in epithelial cells15,16. They form bundled filaments from keratin type I and type II heterodimers laterally associating in an antiparallel fashion into apolar tetramers that again align and anneal longitudinally into unit-length filaments17,18. Keratin filaments are semi-flexible, stable and highly elastic, different from the rather stiff and rigid actin and microtubule filaments19,20. They can self-assemble into intricate subcellular networks, the precise organization of which depends on their functional adaptation in different cell types21. Generally, the intracellular keratin network can organize into a rim network, supporting plasma membrane integrity and connecting desmosomal contacts, and a spoke network, surrounding the nucleus and transferring information from the cell exterior to the nucleus22–24. In polarized simple epithelia, such as the EVL, keratin networks are typically positioned near the apex where they are thought to function in resisting mechanical and chemical stresses and maintaining epithelial apicobasal polarity25–27. Keratins have been shown to be sensitive to mechanical forces by reorganizing and changing their mechanical properties upon stress application19,28,29. Yet, how keratin network mechanosensitivity functions in epithelial tissue morphogenesis remains unsettled30,31. The role of epithelial keratins in development and disease is only beginning to be understood32,33. Keratin mutations can cause diseases that lower the resilience of epidermal tissues to mechanical stress34. Moreover, studies in mouse embryos have shown that keratins are required for trophectoderm specification and extra-embryonic tissue growth and expansion30,31. Here we show that keratin intermediate filaments are required for EVL spreading during zebrafish epiboly. They function in this process by balancing EVL tissue viscosity with the external pulling forces mediating EVL spreading. This balancing function of keratins enables the EVL to undergo uniform spreading without rupturing. Results Keratins are specifically expressed within the EVL and YSL during gastrulation To explore the function of keratins during early zebrafish embryogenesis, we analyzed the expression of different keratins in embryos from early blastula to late gastrula stages (4.5–8.5 hpf). Previous studies have shown that 13 type I keratins and 3 type II keratins are specifically expressed within the developing EVL during this period35,36. To identify the temporal keratin expression profiles within the EVL during epiboly, we used RT-qPCR to map the expression of three keratin type II (keratin 4, 5, 8) and one keratin type I (keratin 18), abundantly expressed within the EVL36,37. We found all of these keratins to be expressed already at early blastula stages, with their expression continuously increasing until the end of gastrulation (Supplementary Fig. 1A). The expression was restricted to EVL progenitor cells at the blastoderm surface as evidenced by fluorescence in situ hybridization for keratin 8 mRNA (Supplementary Fig. 1B). To determine the spatial distribution of keratins within the epibolizing embryo, we took advantage of Tgkrt18:krt18-GFP embryos expressing GFP-tagged keratin 18 under its endogenous promoter. Keratin 18 expression was first detected in EVL progenitor cells within the early gastrula (4.0 hpf) arranged in short, bundled and unconnected filaments located predominantly at the apical surface of these cells (Fig. 1B, Supplementary Video 1 and Supplementary Fig. 1C). Keratin 18 continued to be selectively expressed within EVL cells until mid-gastrulation (5.5 hpf), when some additional expression was also detected within the forming YSL directly adjacent to the place where the leading edge of the EVL contacts the YSL (Fig. 1B). During this period and continuing until the end of gastrulation, the apical network of keratin 18 filaments within EVL cells became increasingly dense and interconnected (Fig. 1B), as evidenced by a continuous increase in keratin 18-GFP intensity and network density between 4.5 and 8.5 hpf (Fig. 1C, D, and Supplementary Video 1). This increase in keratin network intensity and density was accompanied by EVL cells increasing the apical area during epiboly (Fig. 1E), suggesting a close temporal correlation between keratin network maturation and EVL cell spreading (Fig. 1F). Initially, keratin filaments were predominantly localized to the apical surface of the EVL cells (Fig. 1B and Supplementary Fig. 1C), but from 6 hpf onwards, additional keratin 18 expression was detected in a bundle-like configuration along the apical junction of EVL cells (Fig. 1B, G), consistent with previous observation of keratins showing both junctional (“rim”) and apical (“spoke”) localization22,28,30,31. Finally, to check whether keratins in EVL cells can organize into networks different from t he keratin 18 network, we ubiquitous expressed a mCherry-tagged form of keratin 4 by mRNA injection at the 1-cell-stage. Interestingly, keratin 4-mCherry, previously shown to form different keratin dimers than keratin 1821, was selectively expressed within EVL cells and its subcellular expression pattern entirely co-localized with the keratin 18 network (Supplementary Fig. 1D), suggesting that different keratin dimers are part of the same network within EVL cells. Of note, EVL cells typically showed different levels of keratin expression both on the mRNA and protein level, with cells initially expressing higher levels also showing an earlier network maturation (Fig. 1B, G and Supplementary Fig. 1B). This eventually resulted in the EVL at later stages of gastrulation being composed of cells displaying keratin networks at clearly different stages of maturation (Fig. 1B, G). Collectively, these findings suggest that keratins are predominantly expressed within the EVL and adjacent YSL, forming an increasingly dense apical and junctional network from early blastula to late gastrula stages accompanying EVL cell spreading. Keratin network maturation is mechanosensitive EVL epiboly movements are driven by a large actomyosin cable forming within the YSL and pulling the margin of the EVL over the yolk cell10. Given that keratin networks can reorganize under stress29,38, we speculated that the observed maturation of the keratin network within EVL cells might be facilitated by EVL tissue tension building up during the course of epiboly10,39. To test this possibility, we analyzed whether and how EVL network maturation is affected in embryos where EVL tension is either increased or decreased. To modulate EVL tension, we expressed a constitutive active form of RhoA (CARhoA) specifically within the YSL, promoting YSL actomyosin contractility and pulling, or a constitutive active form of Myosin phosphatase (CAMypt) leading to reduced YSL contractility and pulling10,11. In CARhoA-expressing embryos, keratin 18 expression intensity and network density in EVL cells prematurely increased during the course of epiboly, while in CAMypt-expressing embryos, keratin expression and network maturation were delayed (Fig. 2A–D, Supplementary Videos 2–5 and Supplementary Fig. 2B). This keratin mechanosensitivity was detectable both when network maturation was analyzed as a function of developmental time or degree of epiboly progression (Fig. 2A–D and Supplementary Fig. 2B, C), suggesting that the effect of EVL tension on keratin network maturation is not just a secondary consequence of changes in EVL epiboly movements (Supplementary Fig. 2D).Fig. 2Pulling forces promote keratin expression within the EVL.A Schematic showing actomyosin contraction and flows within the YSL providing the mechanical forces pulling the EVL over the yolk cell during epiboly (for details see ref. 10) in a zebrafish embryo at 4 hpf (left) and 6 hpf (right). Green arrows, EVL epiboly movements; black arrows, actomyosin contraction within the YSL. B Maximum intensity projection images of keratin expression in representative Tg(acbt2:Utrophin-mcherry, krt18:Krt18-GFP) control embryos (YSL injection of 0.2% phenol red, top row) and embryos injected with 100 pg CAMypt (middle row) and 50 pg CARhoA (bottom row) into either the YSL at 3.3 hpf (CAMypt) or marginal cells at 3.3 hpf (CARhoA) at sphere stage (4 hpf, left column), shield stage (6 hpf, right column), and 90% epiboly stage (9.5hpf, right column). Scale bar: 100 µm. C Plot of average keratin intensity as a function of time (hpf) in Tg(krt18:KrtGFP) control (green, N = 3, n = 5 embryos), CAMypt (blue, N = 3, n = 5 embryos) and CARhoA mRNA injected embryos (pink, N = 3, n = 4 embryos) as described in (B). Data are presented as mean line ± standard error of the mean (SEM) ribbon. D Plot of average density of keratin network as a function of time (hpf) in Tg(actb2:Utrophin-mcherry, krt18:Krt18-GFP) control (green, N = 3, n = 6 embryos), CAMypt (blue, N = 3, n = 4 embryos) and CARhoA mRNA injected embryos (pink, N = 3, n = 3 embryos) as described in (B). Data are presented as mean line ± SEM ribbon. E 3-dimensional (3D) plot of keratin intensity, network density and EVL cell area as a function of time (hpf) in Tg(actb2:Utrophin-mcherry, krt18:Krt18-GFP) control (viridis, N = 3, n = 3 embryos), CAMypt (orange, N = 3, n = 3 embryos) and CARhoA mRNA injected embryos (pink, N = 3, n = 3 embryos) as described in (B). Data are presented as a surface whose width represents data spread, indicating their variability (standard deviation (SD) of density, intensity and area). F Schematic showing pipette aspirations of the EVL of a 70% epiboly (7 hpf) stage embryo where the regions of interest within the pipette and outside of it are marked as yellow boxes. G Maximum intensity projection images of keratin localization and intensity within the EVL before (left) and after (right) aspiration with a pipette in a representative Tg(krt18:KrtGFP) embryo. White lines outline the boundary of the pipette. Scale bar: 10 µm. H Box plot overlaid with scatter of keratin intensity within the pipette before (green) and after (blue) EVL aspiration in control (N = 4, n = 33 embryos), and Cytochalasin D (CytoD; 25 nM)-treated (N = 4, n = 25 embryos) Tg(actb2:Utrophinmcherry, krt18:Krt18GFP) embryos at 5 and 7 hpf. Box plots show the distribution with median (center line), interquartile range (box), and whiskers extending to values within 1.5 × IQR. (p values: ***<0.001; ANNOVA C(control versus CytoD treatment) = 3.883936e-05, ANNOVA C(pre- versus post-aspiration) = 3.045997e-14, ANNOVA interaction of C(stage)xC(control versus CytoD treatment) = 3.045997e-14). Source data are provided as a Source Data file. Pulling forces promote keratin expression within the EVL. A Schematic showing actomyosin contraction and flows within the YSL providing the mechanical forces pulling the EVL over the yolk cell during epiboly (for details see ref. 10) in a zebrafish embryo at 4 hpf (left) and 6 hpf (right). Green arrows, EVL epiboly movements; black arrows, actomyosin contraction within the YSL. B Maximum intensity projection images of keratin expression in representative Tg(acbt2:Utrophin-mcherry, krt18:Krt18-GFP) control embryos (YSL injection of 0.2% phenol red, top row) and embryos injected with 100 pg CAMypt (middle row) and 50 pg CARhoA (bottom row) into either the YSL at 3.3 hpf (CAMypt) or marginal cells at 3.3 hpf (CARhoA) at sphere stage (4 hpf, left column), shield stage (6 hpf, right column), and 90% epiboly stage (9.5hpf, right column). Scale bar: 100 µm. C Plot of average keratin intensity as a function of time (hpf) in Tg(krt18:KrtGFP) control (green, N = 3, n = 5 embryos), CAMypt (blue, N = 3, n = 5 embryos) and CARhoA mRNA injected embryos (pink, N = 3, n = 4 embryos) as described in (B). Data are presented as mean line ± standard error of the mean (SEM) ribbon. D Plot of average density of keratin network as a function of time (hpf) in Tg(actb2:Utrophin-mcherry, krt18:Krt18-GFP) control (green, N = 3, n = 6 embryos), CAMypt (blue, N = 3, n = 4 embryos) and CARhoA mRNA injected embryos (pink, N = 3, n = 3 embryos) as described in (B). Data are presented as mean line ± SEM ribbon. E 3-dimensional (3D) plot of keratin intensity, network density and EVL cell area as a function of time (hpf) in Tg(actb2:Utrophin-mcherry, krt18:Krt18-GFP) control (viridis, N = 3, n = 3 embryos), CAMypt (orange, N = 3, n = 3 embryos) and CARhoA mRNA injected embryos (pink, N = 3, n = 3 embryos) as described in (B). Data are presented as a surface whose width represents data spread, indicating their variability (standard deviation (SD) of density, intensity and area). F Schematic showing pipette aspirations of the EVL of a 70% epiboly (7 hpf) stage embryo where the regions of interest within the pipette and outside of it are marked as yellow boxes. G Maximum intensity projection images of keratin localization and intensity within the EVL before (left) and after (right) aspiration with a pipette in a representative Tg(krt18:KrtGFP) embryo. White lines outline the boundary of the pipette. Scale bar: 10 µm. H Box plot overlaid with scatter of keratin intensity within the pipette before (green) and after (blue) EVL aspiration in control (N = 4, n = 33 embryos), and Cytochalasin D (CytoD; 25 nM)-treated (N = 4, n = 25 embryos) Tg(actb2:Utrophinmcherry, krt18:Krt18GFP) embryos at 5 and 7 hpf. Box plots show the distribution with median (center line), interquartile range (box), and whiskers extending to values within 1.5 × IQR. (p values: ***<0.001; ANNOVA C(control versus CytoD treatment) = 3.883936e-05, ANNOVA C(pre- versus post-aspiration) = 3.045997e-14, ANNOVA interaction of C(stage)xC(control versus CytoD treatment) = 3.045997e-14). Source data are provided as a Source Data file. To more directly assess the effect of EVL tension on keratin network maturation, we locally increased EVL tension in 5 and 7 hpf embryos by aspirating the EVL using a micropipette and analyzing resultant changes in EVL network maturation. We found that keratin filaments showed increased accumulation in EVL cells upon aspiration (Fig. 2F–H, Supplementary Video 8 and Supplementary Fig. 2F), further supporting the notion that keratin network maturation within the EVL cells is promoted by EVL tension building up during the course of epiboly. To investigate the mechanisms underlying keratin network mechanosensitivity, we examined whether keratin mRNA expression within the EVL is responsive to reduced pulling forces from the YSL. Using qPCR, we compared keratin 4, 5, 8, and 18 expression levels in control embryos and embryos expressing CAMypt within the YSL, which attenuates actomyosin contraction-mediated pulling on the EVL10. Notably, this comparison revealed no significant differences in keratin 4, 5, 8, and 18 expression (Supplementary Fig. 2I), arguing against a force-dependent transcriptional regulation of keratin within the EVL. Interestingly, in our micropipette aspiration assay, increased keratin expression in the EVL was spatiotemporally accompanied by a marked enhancement of actin network assembly (Supplementary Fig. 2E, F). Given the well-established mechanosensitivity of the actomyosin cytoskeleton and its known physical interactions with the keratin network11,21,40,41, this observation raises the possibility that keratin mechanosensitivity in the EVL may depend on actin network dynamics. To test this hypothesis, we treated embryos with low concentrations of Cytochalasin D (CytoD) to impair actin network assembly in EVL cells and assessed keratin upregulation upon micropipette aspiration. Strikingly, CytoD-treated embryos showed a markedly reduced keratin response to aspiration (Fig. 2H), suggesting that proper actin network assembly is required for keratin mechanosensitivity in the EVL. To more directly assess whether actomyosin contractility affects keratin network assembly, we injected CAMypt mRNA into a single blastomere at the 64-cell stage to generate EVL clones with reduced contractility. We found that downregulation of actomyosin contractility resulted in a corresponding decrease in keratin network assembly within these clones (Fig. 2A–D and Supplementary Fig. 2G, H), supporting a causal role of actomyosin tension in promoting keratin assembly. Together, these findings demonstrate that the keratin cytoskeleton in EVL cells is mechanosensitive and that this mechanosensitivity is dependent on the mechanosensitive assembly of the actin network. Keratins regulate EVL tissue properties Keratins have previously been implicated in resisting mechanical stresses in epithelial cells. Thus, the mechanosensitive coupling between EVL keratin network maturation and pulling force generation within the YSL might constitute a mechanism to protect the EVL tissue against excessive deformations by mechanically strengthening it proportionally to the force exerted on it. To test this possibility, we knocked down/out the expression of the two keratin type II genes (keratin 4 and 8) primarily expressed within EVL cells36, reasoning that in the absence of keratin type II expression, no keratin dimers can be formed within the EVL and thus keratin network formation should be defective. Ubiquitous knock-down/out of keratin 4 and 8 expression by using morpholinos and Crispr-Cas9 or interference with keratin network formation by overexpressing a dominant negative version of keratin 18 (DNkeratin18)41, led to strongly diminished keratin 18 expression and network formation within EVL cells (Fig. 3A–C, Supplementary Videos 6–7 and Supplementary Fig. 3A, B). Loss of the keratin network in all these perturbations led to delayed EVL epiboly movements (Fig. 3D, Supplementary Videos 6 and 7 and Supplementary Fig. 3C) and frequent rupturing of the EVL towards the end of gastrulation, ultimately causing embryo lethality (Supplementary Fig. 3E, F). This phenotype could be partially rescued by co-injecting the keratin 8 morpholino - eliciting an epiboly phenotype similar but slightly weaker than the phenotype observed when injecting both keratin 4 and 8 morpholinos - with keratin 8 mRNA not recognized by the morpholino (Supplementary Fig. 3F–H, Supplementary Video 8). Rupture of the EVL in keratin deficient embryos was typically preceded by a reduction in junctional localization of E-cadherin (adherens junctions), Occludin b (tight junctions), and junctional plakoglobin a (desmosomal junctions) (Fig. 3G, Supplementary Videos 12–14 and Supplementary Fig. 3D, E), suggesting that keratins are essential for maintaining EVL tissue integrity by enabling proper formation and stabilization of cell-cell junctions between EVL cells.Fig. 3Loss of keratin expression diminishes EVL epiboly movements.A Maximum intensity projection images of keratin expression in representative Tg(krt18: Krt18GFP) embryos at sphere stage (4 hpf, left column), shield stage (6phf, middle column) and bud stage (10 hpf, right column) - with insets on the right lower corner showing single cells with their boundary marked by a red line - injected at the one-cell stage with 2 ng control MO (top row), 1 ng keratin 4 plus 1 ng keratin 8 MO (middle row), or with TraCr keratin 4 and keratin 8 gRNA (keratin 4/8 crispant F0; bottom row). Scale bar: 100 µm (main panel) and 10 µm (inset). B Plot of averaged density of keratin network in individual EVL cells as a function of time (hpf) during epiboly in Tg(actb2:Utrophinmcherry, krt18:Krt18GFP) embryos injected at the one-cell stage with 2 ng control MO (top row; green, N = 4, n = 4 embryos), 1 ng keratin 4 plus 1 ng keratin 8 MO (middle row; blue, N = 4, n = 4 embryos), or with TraCr keratin 4 and keratin 8 gRNA (keratin 4/8 crispant F0; bottom row; pink, N = 2, n = 6 embryos). Data are presented as mean line ± standard error of the mean (SEM) ribbon. C 3-dimensional (3D) plot of keratin intensity, network density and area of EVL cells as a function of time (hpf) during epiboly in Tg(actb2:Utrophinmcherry,krt18:Krt18GFP) embryos injected at the one-cell stage with 2 ng control MO (top row; green, N = 4, n = 4 embryos), 1 ng keratin 4 plus 1 ng keratin 8 MO (middle row; orange, N = 4, n = 4 embryos), or with TraCr keratin 4 and keratin 8 gRNA (keratin 4/8 crispant F0; bottom row; pink, N = 2, n = 6 embryos). Data are presented as a surface whose width represents data spread, indicating their variability (standard deviation (SD) of density, intensity and area). D Plot of EVL epiboly movement speed as a function of time (hpf) during epiboly starting at sphere stage (4 hpf) until late epiboly stages (9 hpf) in Tg(actb2: Utrophin-mcherry, krt18:Krt18GFP) embryos injected at the one-cell stage with 2 ng control MO (top row; green, N = 3, n = 4 embryos), 1 ng keratin 4 plus 1 ng keratin 8 MO (middle row; blue, N = 3, n = 4 embryos), or with TraCr keratin 4 and keratin 8 gRNA (keratin 4/8 crispant F0; bottom row; yellow, N = 2, n = 4 embryos). Data are presented as mean line ± SEM ribbon. E Maximum intensity projection images of the EVL at 13.5 hpf in Tg(actb2: Utrophin-mcherry, krt18:Krt18GFP) embryos injected with 1 ng keratin 4 plus 1 ng keratin 8 MO at the one-cell stage before (left) and after rupture (right). Scale bar:100 µm. F Bar plot of death rate with overlaid scatter points representing individual measurements in Tg(actb2:Utrophinmcherry, krt18:Krt18GFP) embryos injected at the one-cell stage with 2 ng control MO (green, N = 4, n = 4 embryos), 1 ng keratin 4 plus 1 ng keratin 8 MO (blue, N = 4, n = 4 embryos), or with TraCr keratin 4 and keratin 8 gRNA (keratin 4/8 crispant F0; yellow, N = 2, n = 6 embryos). Data are represented as bar plots showing mean values, with error bars representing standard deviation (SD). G Plot of average E-cadherin intensity as function of time in control MO (green, N = 3, n = 6) and 1 ng keratin 4 plus 1 ng keratin 8 MO (blue, N = 3, n = 5 embryos) injected Tg(cdh1-mlanYFP)xt17 embryos. Data are presented as mean line ± SEM ribbon. Source data are provided as a Source Data file. Loss of keratin expression diminishes EVL epiboly movements. A Maximum intensity projection images of keratin expression in representative Tg(krt18: Krt18GFP) embryos at sphere stage (4 hpf, left column), shield stage (6phf, middle column) and bud stage (10 hpf, right column) - with insets on the right lower corner showing single cells with their boundary marked by a red line - injected at the one-cell stage with 2 ng control MO (top row), 1 ng keratin 4 plus 1 ng keratin 8 MO (middle row), or with TraCr keratin 4 and keratin 8 gRNA (keratin 4/8 crispant F0; bottom row). Scale bar: 100 µm (main panel) and 10 µm (inset). B Plot of averaged density of keratin network in individual EVL cells as a function of time (hpf) during epiboly in Tg(actb2:Utrophinmcherry, krt18:Krt18GFP) embryos injected at the one-cell stage with 2 ng control MO (top row; green, N = 4, n = 4 embryos), 1 ng keratin 4 plus 1 ng keratin 8 MO (middle row; blue, N = 4, n = 4 embryos), or with TraCr keratin 4 and keratin 8 gRNA (keratin 4/8 crispant F0; bottom row; pink, N = 2, n = 6 embryos). Data are presented as mean line ± standard error of the mean (SEM) ribbon. C 3-dimensional (3D) plot of keratin intensity, network density and area of EVL cells as a function of time (hpf) during epiboly in Tg(actb2:Utrophinmcherry,krt18:Krt18GFP) embryos injected at the one-cell stage with 2 ng control MO (top row; green, N = 4, n = 4 embryos), 1 ng keratin 4 plus 1 ng keratin 8 MO (middle row; orange, N = 4, n = 4 embryos), or with TraCr keratin 4 and keratin 8 gRNA (keratin 4/8 crispant F0; bottom row; pink, N = 2, n = 6 embryos). Data are presented as a surface whose width represents data spread, indicating their variability (standard deviation (SD) of density, intensity and area). D Plot of EVL epiboly movement speed as a function of time (hpf) during epiboly starting at sphere stage (4 hpf) until late epiboly stages (9 hpf) in Tg(actb2: Utrophin-mcherry, krt18:Krt18GFP) embryos injected at the one-cell stage with 2 ng control MO (top row; green, N = 3, n = 4 embryos), 1 ng keratin 4 plus 1 ng keratin 8 MO (middle row; blue, N = 3, n = 4 embryos), or with TraCr keratin 4 and keratin 8 gRNA (keratin 4/8 crispant F0; bottom row; yellow, N = 2, n = 4 embryos). Data are presented as mean line ± SEM ribbon. E Maximum intensity projection images of the EVL at 13.5 hpf in Tg(actb2: Utrophin-mcherry, krt18:Krt18GFP) embryos injected with 1 ng keratin 4 plus 1 ng keratin 8 MO at the one-cell stage before (left) and after rupture (right). Scale bar:100 µm. F Bar plot of death rate with overlaid scatter points representing individual measurements in Tg(actb2:Utrophinmcherry, krt18:Krt18GFP) embryos injected at the one-cell stage with 2 ng control MO (green, N = 4, n = 4 embryos), 1 ng keratin 4 plus 1 ng keratin 8 MO (blue, N = 4, n = 4 embryos), or with TraCr keratin 4 and keratin 8 gRNA (keratin 4/8 crispant F0; yellow, N = 2, n = 6 embryos). Data are represented as bar plots showing mean values, with error bars representing standard deviation (SD). G Plot of average E-cadherin intensity as function of time in control MO (green, N = 3, n = 6) and 1 ng keratin 4 plus 1 ng keratin 8 MO (blue, N = 3, n = 5 embryos) injected Tg(cdh1-mlanYFP)xt17 embryos. Data are presented as mean line ± SEM ribbon. Source data are provided as a Source Data file. To understand how keratins function in EVL epiboly movements, we analyzed changes in EVL cell shapes and rearrangement in response to the pulling forces from the YSL during epiboly. Comparing wild-type with keratin 4/8 morphant embryos revealed that EVL cells in wild-type embryos coordinately elongated along the animal-vegetal (AV) axis of the gastrula, the axis of EVL spreading, while no such coordinated cell elongation was detectable in morphant embryos (Fig. 4A, B). This points at the possibility that tissue material properties, determining force transduction and mechanical resilience, might be altered by keratin network formation.Fig. 4Mechanical force percolation within the EVL is dependent on keratin expression.A Plots of EVL cell orientations with ellipses representing shape descriptors (long and short axis) of individual EVL cells with the line in the middle marking the orientation of the long axis at consecutive stages during epiboly (4.5, 6, 9, 12 hpf) in curvature corrected Tg(actb2:Utrophin-mcherry, krt18:Krt18GFP) embryos injected at the one-cell stage with 2 ng control MO (top row) or 1 ng keratin 4 plus 1 ng keratin 8 MO (middle row). Each cell is color-coded according to the orientation of the axis as shown in the color bar (viridis) at the bottom (Green: AV axis orientation, blue: dorsoventral/DV orientation). Scale Bar: 100 µm. B Rose plot overlaid with scatter of bin means of EVL cell orientations in Tg(actb2:Utrophinmcherry,krt18:Krt18GFP) embryos at 6 hpf injected at the one-cell stage with 2 ng control MO (N = 4, n = 93 cells) or 1 ng keratin 4 plus 1 ng keratin 8 MO (N = 4, n = 84 cells). C Bar plots with overlaid scatter points representing individual embryo measurements of tissue viscosity measured at the EVL margin using micropipette aspiration at 4–5.5 hpf and 6.5–7.5 hpf in Tg(actb2: Utrophin-mcherry, krt18:Krt18GFP) embryos injected at the one-cell stage with 2 ng control MO (N = 5, n = 37 embryos) or 1 ng keratin 4 plus 1 ng keratin 8 MO (C; N = 5, n = 33 embryos). Data are represented as bar plots showing mean values, with error bars representing SD. (p value: **<0.01; t-test p(4-5.5.hpf) = 2.797231e-12 and p (6.5-7.5hpf) = 1.792669e-08) D Representative quiver plots of EVL tissue recoil flow velocities after cell ablation/wounding in Tg(actb2:Utrophinmcherry) embryos at 6 hpf injected with 2 ng control MO (top) or 1 ng keratin 4 plus 1 ng keratin 8 MO (bottom). The arrows show the local velocity colored according to the magnitude (viridis). Scale bar: 25 µm. E Plot of averaged EVL tissue recoil velocity directly following cell ablation plotted as a function of distance from the wound center (0 µm) in Tg(actb2:Utrophinmcherry) embryos injected with 2 ng control MO (green, N = 4, n = 23 embryos) or 1 ng keratin 4 plus 1 ng keratin 8 MO (blue, N = 4, n = 17 embryos) immediately after EVL cell ablation. Data are presented as mean line ± standard error of the mean (SEM) ribbon. E’ Plot of the difference of averaged EVL tissue recoil velocity directly following ablation plotted as a function of distance from the wound center (0 µm) in Tg(actb2:Utrophinmcherry) embryos injected with 2 ng control MO versus 1 ng keratin 4 plus 1 ng keratin 8 MO. Data are presented as mean line ± SEM ribbon. F Plot of averaged EVL tissue flow velocity during wound closure after EVL cell ablation plotted as a function of distance from the wound center (0 µm) in Tg(actb2:Utrophinmcherry) embryos injected with 2 ng control MO (green, N = 4, n = 23 embryos) or 1 ng keratin 4 plus 1 ng keratin 8 MO (blue, N = 4, n = 17 embryos) at successive timepoints after EVL cell ablation (20 s, solid; 60 s, dotted; 120 s, dotted line with points). Data are presented as mean line ± SEM ribbon. F’ Plot of the difference of averaged EVL tissue flow velocity during wound closure after EVL cell ablation plotted as a function of distance from the wound center (0 µm) in Tg(actb2:Utrophinmcherry) embryos injected with 2 ng control MO versus 1 ng keratin 4 plus 1 ng keratin 8 MO at successive timepoints after the ablation (20 s, solid; 60 s, dotted; 120 s, dotted line with points). Data are presented as mean line ± SEM ribbon. G Plot of wound area as a function of time after EVL cell ablation in Tg(actb2:Utrophinmcherry) embryos injected with 2 ng control MO (green, N = 4, n = 23 embryos) or 1 ng keratin 4 plus 1 ng keratin 8 MO (blue, N = 4, n = 17 embryos). Data are presented as mean line ± SEM ribbon. (G’) Plot of the change of wound area as a function of time after EVL cell ablation in Tg(actb2:Utrophinmcherry) embryos injected with 2 ng control MO (green, N = 4, n = 23 embryos) or 1 ng keratin 4 plus 1 ng keratin 8 MO (blue, N = 4, n = 17 embryos). Data are presented as mean line ± SEM ribbon. Source data are provided as a Source Data file. Mechanical force percolation within the EVL is dependent on keratin expression. A Plots of EVL cell orientations with ellipses representing shape descriptors (long and short axis) of individual EVL cells with the line in the middle marking the orientation of the long axis at consecutive stages during epiboly (4.5, 6, 9, 12 hpf) in curvature corrected Tg(actb2:Utrophin-mcherry, krt18:Krt18GFP) embryos injected at the one-cell stage with 2 ng control MO (top row) or 1 ng keratin 4 plus 1 ng keratin 8 MO (middle row). Each cell is color-coded according to the orientation of the axis as shown in the color bar (viridis) at the bottom (Green: AV axis orientation, blue: dorsoventral/DV orientation). Scale Bar: 100 µm. B Rose plot overlaid with scatter of bin means of EVL cell orientations in Tg(actb2:Utrophinmcherry,krt18:Krt18GFP) embryos at 6 hpf injected at the one-cell stage with 2 ng control MO (N = 4, n = 93 cells) or 1 ng keratin 4 plus 1 ng keratin 8 MO (N = 4, n = 84 cells). C Bar plots with overlaid scatter points representing individual embryo measurements of tissue viscosity measured at the EVL margin using micropipette aspiration at 4–5.5 hpf and 6.5–7.5 hpf in Tg(actb2: Utrophin-mcherry, krt18:Krt18GFP) embryos injected at the one-cell stage with 2 ng control MO (N = 5, n = 37 embryos) or 1 ng keratin 4 plus 1 ng keratin 8 MO (C; N = 5, n = 33 embryos). Data are represented as bar plots showing mean values, with error bars representing SD. (p value: **<0.01; t-test p(4-5.5.hpf) = 2.797231e-12 and p (6.5-7.5hpf) = 1.792669e-08) D Representative quiver plots of EVL tissue recoil flow velocities after cell ablation/wounding in Tg(actb2:Utrophinmcherry) embryos at 6 hpf injected with 2 ng control MO (top) or 1 ng keratin 4 plus 1 ng keratin 8 MO (bottom). The arrows show the local velocity colored according to the magnitude (viridis). Scale bar: 25 µm. E Plot of averaged EVL tissue recoil velocity directly following cell ablation plotted as a function of distance from the wound center (0 µm) in Tg(actb2:Utrophinmcherry) embryos injected with 2 ng control MO (green, N = 4, n = 23 embryos) or 1 ng keratin 4 plus 1 ng keratin 8 MO (blue, N = 4, n = 17 embryos) immediately after EVL cell ablation. Data are presented as mean line ± standard error of the mean (SEM) ribbon. E’ Plot of the difference of averaged EVL tissue recoil velocity directly following ablation plotted as a function of distance from the wound center (0 µm) in Tg(actb2:Utrophinmcherry) embryos injected with 2 ng control MO versus 1 ng keratin 4 plus 1 ng keratin 8 MO. Data are presented as mean line ± SEM ribbon. F Plot of averaged EVL tissue flow velocity during wound closure after EVL cell ablation plotted as a function of distance from the wound center (0 µm) in Tg(actb2:Utrophinmcherry) embryos injected with 2 ng control MO (green, N = 4, n = 23 embryos) or 1 ng keratin 4 plus 1 ng keratin 8 MO (blue, N = 4, n = 17 embryos) at successive timepoints after EVL cell ablation (20 s, solid; 60 s, dotted; 120 s, dotted line with points). Data are presented as mean line ± SEM ribbon. F’ Plot of the difference of averaged EVL tissue flow velocity during wound closure after EVL cell ablation plotted as a function of distance from the wound center (0 µm) in Tg(actb2:Utrophinmcherry) embryos injected with 2 ng control MO versus 1 ng keratin 4 plus 1 ng keratin 8 MO at successive timepoints after the ablation (20 s, solid; 60 s, dotted; 120 s, dotted line with points). Data are presented as mean line ± SEM ribbon. G Plot of wound area as a function of time after EVL cell ablation in Tg(actb2:Utrophinmcherry) embryos injected with 2 ng control MO (green, N = 4, n = 23 embryos) or 1 ng keratin 4 plus 1 ng keratin 8 MO (blue, N = 4, n = 17 embryos). Data are presented as mean line ± SEM ribbon. (G’) Plot of the change of wound area as a function of time after EVL cell ablation in Tg(actb2:Utrophinmcherry) embryos injected with 2 ng control MO (green, N = 4, n = 23 embryos) or 1 ng keratin 4 plus 1 ng keratin 8 MO (blue, N = 4, n = 17 embryos). Data are presented as mean line ± SEM ribbon. Source data are provided as a Source Data file. To investigate this possibility, we measured the material properties of EVL tissue at early and mid-gastrulation (5 and 7 hpf) using micropipette aspiration (Supplementary Fig. 4E)42,43. Analysis of the flow profile of the EVL into the pipette revealed a linear response, consistent with viscous properties, which allowed us to assess the viscosity of EVL tissue. Comparing the viscosity of EVL tissue between early and mid-gastrulation stages showed a significant increase in viscosity at mid-gastrulation, coinciding with the maturation of the keratin network (Fig. 4C, Supplementary Video 9 and Supplementary Fig. 2F). This suggests that keratin network maturation during epiboly contributes to the increase in EVL tissue viscosity. To test this function of keratins more directly, we performed EVL micropipette aspiration experiments in wild-type and keratin 4/8 morphant embryos. At early gastrulation, when differences in keratin network formation between wild-type and morphant embryos were still relatively small (Fig. 1B), we observed a slight reduction in viscosity in keratin 4/8 morphant embryos compared to wild-type (Fig. 4C). In contrast, at mid-gastrulation (Fig. 4C), when the differences in keratin network organization between morphant and wild-type embryos became more pronounced (Fig. 1B), viscosity was significantly reduced in the morphant embryos. This supports the idea that keratin network maturation enhances EVL tissue viscosity. To test whether this effect is due to keratin expression specifically within EVL cells, we knocked down keratin 4/8 expression within the YSL by injection of keratin 4/8 morpholinos directly into the forming YSL at the high (3.5 hpf) stage. Interestingly, this did not cause any detectable changes in coordinated EVL cell elongation along the AV axis compared to wild-type embryos, suggesting that keratins control EVL cell shape changes in a tissue-autonomous manner (Supplementary Fig. 4A, B). To further challenge this conclusion, we developed an EVL wound healing assay, allowing us to monitor autonomous EVL spreading during wound closure44,45. For wounding the EVL, we ablated 3–5 neighboring cells at random positions within the EVL in mid-gastrulation stage embryos (7 hpf) and observed how the coordinated spreading and movements of the neighboring cells extruded and replaced the ablated cells. In control embryos upon ablation, a supracellular actin cable formed at the leading edge of EVL cells neighboring the ablated cells (Fig. 4D, Supplementary Videos 10, 11). This was accompanied by the highly coordinated movement and spreading of EVL cells towards the site of cell ablation, eventually leading to the extrusion of the ablated cells and closure of the wounding site (Fig. 4F-F’). During this process, not only did leading-edge EVL cells elongate and move toward the wounding site, but cells located further away from the ablation site also responded, resulting in highly coordinated long-range tissue movements detected by particle image velocimetry (PIV) flow analysis (Fig. 4D–G). In contrast, EVL tissue movements were largely restricted to EVL cells directly neighboring the wounding site in keratin 4/8 morphant and crispant embryos (Fig. 4D–G, Supplementary Video 11 and Supplementary Fig. 4C, D). Similar differences in the spatial extent of tissue flows between wild-type and morphant/crispant embryos were observed when analyzing tissue recoil immediately following cell ablation, prior to initiation of wound closure by formation of the supracellular actin cable at the EVL leading edge (Supplementary Fig. 4C, D). Collectively, these findings support the notion that keratin expression within the EVL increases tissue viscosity, thereby extending the hydrodynamic length over which the EVL tissue deforms in response to forces applied at its margin. Keratins are required for mechanosensitive actomyosin contraction within the YSL Given that keratin expression within the EVL is mechanosensitive, we hypothesized that this behavior might constitute a feedback mechanism balancing EVL tissue viscosity with the forces pulling on its margin, thereby setting the rate of EVL epiboly movements. To address this hypothesis, we developed a vertex model46,47 of the EVL, enabling us to relate EVL tissue spreading dynamics to mechanosensitive keratin expression. Mechanical heterogeneity in epiboly has already been established within the context of a vertex model48. To this end, we wrote the energy \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$U={\sum }_{i}\frac{{\Gamma }_{i}}{2{A}_{0}}{({A}_{i}-{A}_{i,0})}^{2}+\frac{{\varGamma }_{i}}{2}{({P}_{i}-{P}_{i,0})}^{2}$$\end{document}U=∑iΓi2A0(Ai−Ai,0)2+Γi2(Pi−Pi,0)2 which penalizes deviations of the areas \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${A}_{i}$$\end{document}Ai and perimeters \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${P}_{i}$$\end{document}Pi of the cells \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$i$$\end{document}i, with cell stiffness constants \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\varGamma }_{i}$$\end{document}Γi, from their target values \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${A}_{i,0}$$\end{document}Ai,0 and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${P}_{i,0}$$\end{document}Pi,0, respectively. In order to make the tissue viscoelastic, we relaxed the target area \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${A}_{i,0}$$\end{document}Ai,0 over a characteristic viscous relaxation timescale \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\tau }_{i}$$\end{document}τi following \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\tau }_{i}\frac{{{dA}}_{i,0}}{{dt}}=-({A}_{i,0}-{A}_{i})$$\end{document}τidAi,0dt=−(Ai,0−Ai), while enforcing a minimum area \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${A}_{0}$$\end{document}A0, and scaled \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${P}_{i,0}$$\end{document}Pi,0 to work at a constant shape parameter \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${s}_{0}=\frac{{P}_{i,0}}{\sqrt{{A}_{i,0}}}=3.72$$\end{document}s0=Pi,0Ai,0=3.72, a value where the tissue is predicted to be still rigid46. Our model incorporates tissue spreading in 3D implicitly through a conserved cell volume \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${V}_{0}={h}_{i}{A}_{i}$$\end{document}V0=hiAi, where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${h}_{i}$$\end{document}hi is the height of the cell (see SI, section 3 for details). We next incorporated keratin into our model (Fig. 5D). As illustrated in Fig. 1B, keratin forms an intracellular network that is initially floppy, but reaches a percolation threshold at approximately \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${K}_{{th}}=150$$\end{document}Kth=150. Capturing the full complexity of its behavior would require simulating the coupled dynamics of actin and keratin networks, including keratin polymerization and depolymerization as well as actomyosin contractility at the subcellular level. To simplify, we instead implemented a mesoscale approximation in which the keratin network is acting in parallel with other mechanical components of the cell. Above the percolation threshold, keratin therefore contributes proportionally to both the cell’s effective stiffness and relaxation time with \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\varGamma }_{i}=\varGamma (1+\beta \varDelta {K}_{i})$$\end{document}Γi=Γ(1+βΔKi) and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\tau }_{i}=\tau (1+\beta \varDelta {K}_{i})$$\end{document}τi=τ(1+βΔKi), respectively, where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\varDelta {K}_{i}=\max \left(0,{K}_{i}-{K}_{{th}}\right)$$\end{document}ΔKi=max0,Ki−Kth and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\beta$$\end{document}β is a mechanical feedback constant.Fig. 5Model of tissue with a mechanosensitive feedback loop for keratin.A Plot of extension length of the aspirated EVL tissue in Tg(krt18:Krt18GFP) embryos as a function of time upon micropipette aspiration. Colormap represents developmental time starting at the beginning of epiboly (4 hpf, blue) to later stages (8 hpf, orange). B Plot of keratin intensity measured inside the pipette in the aspirated EVL tissue in Tg(krt18:Krt18GFP) embryos as a function of time upon micropipette aspiration. Colormap represents developmental time starting at the beginning of epiboly (4 hpf, blue) to later stages (8 hpf, orange). C Representative plot of extension length as a function of time fit to the modified viscoelastic Maxwell model (inset, schematic representing the model) used to measure parameters for the vertex model. D Schematic representing a singular model EVL cell showing keratin filaments (green) at two distinct states of the network activity, dependent on the keratin \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${K}_{i}$$\end{document}Ki formation: unpercolated (left, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${K}_{i} < {K}_{{th}}$$\end{document}Ki {K}_{{th}}$$\end{document}Ki>Kth, mechanically active) with \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${K}_{{th}}$$\end{document}Kth being the threshold value of K above which the keratin mechanochemical feedback loop becomes activated. E Schematic diagram of the mechanochemical feedback loop built into the model representing the feedback from pulling forces from the YSL (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${F}_{{YSL}}$$\end{document}FYSL) stretching the tissue leading to increased pressure \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$p$$\end{document}p in the tissue that enhances keratin \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$K$$\end{document}K formation. This increase in keratin \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$K$$\end{document}K feedback into the area \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$A$$\end{document}A of the cells and increases tissue stiffness \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\varGamma$$\end{document}Γ and relaxation time \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\tau$$\end{document}τ. This increase in tissue stiffness \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\varGamma$$\end{document}Γ and relaxation time \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\tau$$\end{document}τ, in turn, feeds back on the pressure \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$p$$\end{document}p in the tissue. F Plot of model aspirated tissue heights as a function of time measured in the mean field model. Mean tissue height (dark red line) is represented as a solid line over curves representing estimated parameter variability shown as lines with higher transparency (light red lines). G Plot of model keratin intensities in aspirated tissues as a function of time in the mean field model. Mean keratin intensity (dark green line) is represented as a solid line over curves representing estimated parameter variability shown as lines with higher transparency (light green lines). Source data are provided as a Source Data file and in the Data Archive. Model of tissue with a mechanosensitive feedback loop for keratin. A Plot of extension length of the aspirated EVL tissue in Tg(krt18:Krt18GFP) embryos as a function of time upon micropipette aspiration. Colormap represents developmental time starting at the beginning of epiboly (4 hpf, blue) to later stages (8 hpf, orange). B Plot of keratin intensity measured inside the pipette in the aspirated EVL tissue in Tg(krt18:Krt18GFP) embryos as a function of time upon micropipette aspiration. Colormap represents developmental time starting at the beginning of epiboly (4 hpf, blue) to later stages (8 hpf, orange). C Representative plot of extension length as a function of time fit to the modified viscoelastic Maxwell model (inset, schematic representing the model) used to measure parameters for the vertex model. D Schematic representing a singular model EVL cell showing keratin filaments (green) at two distinct states of the network activity, dependent on the keratin \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${K}_{i}$$\end{document}Ki formation: unpercolated (left, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${K}_{i} < {K}_{{th}}$$\end{document}Ki {K}_{{th}}$$\end{document}Ki>Kth, mechanically active) with \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${K}_{{th}}$$\end{document}Kth being the threshold value of K above which the keratin mechanochemical feedback loop becomes activated. E Schematic diagram of the mechanochemical feedback loop built into the model representing the feedback from pulling forces from the YSL (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${F}_{{YSL}}$$\end{document}FYSL) stretching the tissue leading to increased pressure \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$p$$\end{document}p in the tissue that enhances keratin \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$K$$\end{document}K formation. This increase in keratin \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$K$$\end{document}K feedback into the area \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$A$$\end{document}A of the cells and increases tissue stiffness \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\varGamma$$\end{document}Γ and relaxation time \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\tau$$\end{document}τ. This increase in tissue stiffness \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\varGamma$$\end{document}Γ and relaxation time \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\tau$$\end{document}τ, in turn, feeds back on the pressure \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$p$$\end{document}p in the tissue. F Plot of model aspirated tissue heights as a function of time measured in the mean field model. Mean tissue height (dark red line) is represented as a solid line over curves representing estimated parameter variability shown as lines with higher transparency (light red lines). G Plot of model keratin intensities in aspirated tissues as a function of time in the mean field model. Mean keratin intensity (dark green line) is represented as a solid line over curves representing estimated parameter variability shown as lines with higher transparency (light green lines). Source data are provided as a Source Data file and in the Data Archive. To include keratin mechanosensitivity, we further assumed that its assembly from the cytosol increases with the stress applied on the tissue. To account for this, we used a simple, linear model of the evolution of the keratin concentration \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\tau }_{K}\frac{{{dK}}_{i}}{{dt}}=\alpha \,\max (0,{p}_{i})-{K}_{i}$$\end{document}τKdKidt=αmax(0,pi)−Ki, where keratin dissociates with time scale \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\tau }_{K}$$\end{document}τK and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\alpha$$\end{document}α is a coupling constant (note that other nonlinear rheology models have also been used to model the effect of keratin29). Mechanosensitivity in our model arises through biaxial stress, or equivalently, the in-plane pressure \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${p}_{i}$$\end{document}pi defined as \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${p}_{i}=\frac{1}{{V}_{0}}(\frac{{\varGamma }_{i}}{{A}_{0}}({A}_{i}-{A}_{i,0}){A}_{i}+{\varGamma }_{i}({P}_{i}-{s}_{0}\sqrt{{A}_{i,0}}){P}_{i})$$\end{document}pi=1V0(ΓiA0(Ai−Ai,0)Ai+Γi(Pi−s0Ai,0)Pi) (see SI, section 3). This formulation captures the pressure response arising from deviations in both cell area and perimeter relative to their preferred values. In this framework, EVL spreading corresponds to a substantial increase in cell area \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${A}_{i}$$\end{document}Ai, accompanied by dramatic thinning of individual cells. At steady state, this mechanosensitive feedback yields a linear relationship between keratin concentration and pressure, expressed as \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${K}_{i}=\alpha {P}_{i}$$\end{document}Ki=αPi under the assumption of positive pressure. A summary of this mechanochemical feedback loop is presented in Fig. 5E. To estimate the parameters for simulating EVL wound closure and epiboly, we used the EVL aspiration experiments described above (Fig. 5A, B, Figs. 2F–H and 4C), where we took the initial and pulled states as a proxy for the tissue at low and high keratin expression, respectively. In a first step, we fitted the low-keratin initial response and the high-keratin release curves separately for tissue height and keratin expression using a modified viscoelastic Maxwell model, allowing us to obtain estimates of the dissociation timescale \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\tau }_{K}$$\end{document}τK, pressure coupling \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\alpha$$\end{document}α, cell stiffness constant \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\varGamma$$\end{document}Γ, and relaxation timescale \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\tau$$\end{document}τ (Fig. 5C, Supplementary Fig. 7A–D and SI section 1). This model seeks to explicitly understand the local rheology at the length scale of a couple of cells and the time scale of minutes and is mathematically equivalent to the standard aspiration model of tissue spheroids42,48(see SI section 2 for a detailed derivation). In a second step, we developed a mean-field version of the full model that includes feedback (see SI section 5). We used this framework to compute tissue height and keratin expression in aspirated wild-type EVL tissue (Fig. 5F, G and Supplementary Fig. 9A–D), and to estimate the remaining coupling constant, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\beta$$\end{document}β, which represents the mechanical contribution of keratin. We obtained \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\beta=0.005$$\end{document}β=0.005, such that increasing keratin levels from \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${K}_{i}=0$$\end{document}Ki=0 to \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${K}_{i}=550$$\end{document}Ki=550, with a percolation threshold \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${K}_{{th}}=150,$$\end{document}Kth=150, results in a 2-fold increase of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\varGamma$$\end{document}Γ and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\tau$$\end{document}τ (Supplementary Fig. 9G,H). This parameter choice yields consistency with both the pipette aspiration measurements and the experimentally observed viscosity ratio between keratin-deficient and control embryos (Fig. 4C). The full set of fitted parameters is provided in Table S1. For keratin deficient model tissues, we instead set \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\beta=0$$\end{document}β=0, thereby eliminating any mechanical contribution from keratin. This choice, rather than the experimentally more realistic scenario \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\alpha=0$$\end{document}α=0 (corresponding to the absence of keratin production), was motivated by the ability to monitor the role of feedback arising from keratin expression while keeping the feedback loop formally open. Aside from this distinction, the two parameter choices (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\beta=0$$\end{document}β=0 and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\alpha=0$$\end{document}α=0) are mathematically identical. Using these parameters in our extended vertex model (see simulation library)49, we first simulated EVL wound closure behavior in wild-type and keratin-deficient embryos (Fig. 6A–F). To this end, we initialized a disordered tissue patch under tension and generated a model wound by deleting several cells and adding a contractile cable along the wound perimeter (Fig. 6A–D, Supplementary Video 15; see SI section 4 for details). Wound closure in the presence of a cable with tension \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\gamma=2$$\end{document}γ=2 was similar in both wild-type (β = 0.005) and keratin-deficient (β = 0) tissues, matching our experimental observations (Fig. 4G-G’).Fig. 6Vertex model of the EVL for tissue ablation and epiboly in the presence and absence of keratin mechanosensitivity.A Representative model tissue ablation showing keratin expression at \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$t=70s$$\end{document}t=70s in an EVL model tissue with periodic boundary conditions and mechanochemical feedback \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$(\beta=0.005)$$\end{document}(β=0.005). At \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$t=0s$$\end{document}t=0s 6 cells are removed to create a hole at the center of the tissue. A tension of 2 µN is applied on the edges surrounding the removal site to simulate a wound healing process. Colormap on the side represents the keratin intensities of the cells as a function of keratin \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$K$$\end{document}K with percolated keratin network (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${K}_{i} > {K}_{{th}}$$\end{document}Ki>Kth) appearing in green and unpercolated (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${K}_{i} < {K}_{{th}}$$\end{document}Ki {K}_{{th}}$$\end{document}Ki>Kth) appearing in green and unpercolated (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${K}_{i} < {K}_{{th}}$$\end{document}Ki100 pg) led to aberrant keratin network formation. DNkrt18 injections at 150 pg lead to a marked effect on keratin network organization at early stages of keratin network maturation. No toxicity was seen at these concentrations and embryo survival was indistinguisable from the phenol red injected embryos. For lowering actomyosin contraction within the YSL, 100 pg CAMypt mRNA together with 0.2% phenol red was injected into the YSL. As controls, embryos were injected with 0.2% phenol red alone into the YSL10,11. For increasing actomyosin contraction within the YSL, 0.5–1 pg CARhoA together with 2 pg H2A-mCherry mRNA were injected into marginal blastomeres at the 128-cell stage as previously described11. Keratin intensity and density analyses were restricted to the regions within the YSL where mCherry labeled nuclei were clearly detectable. For generating clones of EVL cells with reduced contractility, 100 pg CAMypt mRNA together with 0.2% phenol red were injected into a single blastomere of 64–128-cell staged embryos. The clones were identified by the localization of phenol red. Keratin intensity in the clone was compared to a similar sized region of interest (ROI) within the EVL at least 2 cells away from the boundary of the clone. For morpholino (MO) injections, keratin (krt)4, 8, and 18 MOs were designed as previously described37 krt4 MO: AGACCTGGTTGACATGATGCCTGTG krt8 MO: GGTTTTCTTGCTGTAGGTGGACATC krt18 MO: TGTAGCTTCTTCTCAGACTCATGGT 1 ng of each MO together with 0.2% phenol red was injected either at the one-cell stage (for uniform knock-down) or at the high-stage (3.3 hpf) (for YSL-specific knock-down). As control, 2 ng of a human beta globin MO (5′ - ‘CCTCTTACCTCAGTTACAATTTATA’ − 3′, Gene Tools) was injected. Morpholino rescue To rescue the keratin8 morphant phenotype, we co-injected a modified keratin8 mRNA construct along with the morpholino (MO). To reduce binding of the MO to the ATG sequence of the mRNA, the sequence of mRNA was modified such that the amino acids at position two to four (Ser, Thr, Tyr and Ser) of the coding sequence with alternate codons that code for the same amino acids with different sequences: forward MO insensitive krt8: gcatggacgagctgtacaagATGAGTACGTATAGTAAGAAAACC reverse MO insensitive krt8: TTCTTACTATACGTACTCATcttgtacagctcgtccatgcc The modified mRNA was cloned into a pCS2 plasmid and verified by sequencing. Embryos were injected at one-cell stage with 1 ng of keratin8 MO or control MO together with 0.2% phenol red and 100 pg of keratin8 mRNA for the rescue. Embryo mounting and imaging For inverted embryo imaging, embryos were dechorionated and mounted in 0.3%–0.5% low melting point (LMP) agarose in E3 (Invitrogen) on glass bottom dishes (MatTek) and then imaged on a Nikon CSU W1 with a CFI Plan Apo VC 60x WI/NA 1.2/WD 0.28-0.31 mm/Water, CFI Plan Apo λ 40x air/NA 0.95/0.17–0.25 mm or Leica Stellaris 8 (pipette aspiration experiments). For upright imaging, molds were made using 3% agarose with wells in which the embryos were mounted in a lateral position and covered with 0.5%-0.6% low-melting point agarose. Samples were then imaged on a Leica SP5 or Leica SP8 microscope with a HC FUOTAR L 25x/0.95 W (# 15506374), WD = 2.5 mm, wide angle (41°) objective. Fixed samples were mounted in 0.5–1% LMP agarose and put into prepared agarose moulds (2%) for upright imaging. Live embryos were imaged at 28.5 °C ± 1 °C. Analysis of EVL spreading To determine EVL progression throughout development, the margin of the EVL was tracked along the orthogonal section of laterally imaged embryos, allowing to accurately track EVL movement over the yolk surface. The interval time was set to be 15 min for the acquisitions. In case of shorter acquisition time intervals, the EVL progression was interpolated to gain tracking intervals of 15 min. The speed measured by this method was identical with the speed measured by Particle image velocimetry (PIV) flows described below (analysis of actomyosin flows)71. UV cell ablation and cortical actin laser cutting A Nikon CSU W1 SoRa+NIR spinning disk microscope with a home-built UV laser ablation system was used for ablating cells and monitoring the wound healing response. Ablations were performed by a single UV-laser ablation line over at least 4–5 cells with the intensity of the UV laser set at a level causing spontaneous cell delaminations. Z-stacks of 5–10 μm were acquired at 30 s intervals. EVL cells were segmented with Cellpose68,69 as described below (segmentation and tracking of EVL movements). Tissue flow velocity was measured using PIV with PIVlab71 and the cell segmentations were used to determine the orientation of the flows. Recoil flow analysis of the wound edge directly following the ablation was completed before actin accumulated on the wound edge. Wound closure flows were measured when actin had accumulated on the wound edge and recoil movements had ceased. For the actin network hydrodynamic length measurements within the YSL, the actin network at the apical surface of the YSL was cut using the same UV-laser setup as described above. Cuts were performed at an intensity such that the cortex was ablated without eliciting a wound healing response, as evidenced by the formation of an actomyosin cable closing the wound site. ROI spanning 40 µm on each side of the cut were used to calculate the flow fields using PIV tacking. The analysis of actin recovery in the cut was performed using a manual line ROI over the cut in Fiji and plotted in Python using matplotlib and pandas. The hydrodynamic length of these flows was determined by averaging the flow fields on each side of the cut and fitting the decay of the averaged flows to an exponential decay using scipy. Pipette aspirations Pipette aspirations were performed to measure the viscosity of the EVL as previously described. In short, pipette aspirations were carried out on embryos mounted in 3% methylcellulose in E3 on an inverted Leica SP5 or a Leica Stellaris 5 confocal microscope equipped with the micropipette aspiration system. For creep recovery experiments to measure viscosity, fire-polished and with heat-inactivated FBS passivated micropipettes (Biomedical Instruments) with an inner diameter of 60 μm, 30° bent positioned by micromanipulators (TransferMan Nk2, Eppendorf), and with a blunt end were placed on the EVL around 3–4 cells away from the EVL margin. A negative pressure was applied with an increment of 10 Pa using a Microfluidic Flow Control System Pump (Fluiwell, Fluigent) and Dikeria micromanipulation software70. Images were acquired every second. A negative pressure of 200 mbar was applied until the aspirated tissue flowed in the pipette with constant velocity, followed by pressure release. The length of the aspirated tissue was measured using a custom-made Fiji macro and plotted over time to calculate the aspiration and retraction speed. The viscosity was calculated as previously described42,43. Pipette aspirations were also used to study the mechanosensitive response of the keratin and actin cytoskeleton within the EVL (keratin and actin) and the YSL (actin). To this end, the EVL of Tg(acbt2:Utrophin-mcherry,krt18:Krt18GFP) or Tg(krt18:Krt18GFP) embryos was aspirated under constant pressure, and the mean intensities of keratin and/or actin at the midplane of the pipette and a region of 30 µm around the aspiration site were recorded. For determining keratin mechanosensitivity within the EVL when the actin cortex is disassembled, Tg(acbt2:Utrophin-mcherry,krt18:Krt18GFP) embryos were exposed to 25 nM Cytochalasin D 10 min before the start of aspiration. For the YSL aspirations, a z-stack of 50 µm was taken and the actin intensity in Tg(acbt2:Utrophin-mcherry,krt18:Krt18GFP) embryos was measured on the YSL surface in intervals of 30 s. Analysis of YSL actomyosin flows To measure actin flow velocities within the YSL, Tg(actb2:Utrophin-mcherry) or Tg(actb2: Lifeact-GFP) embryos were mounted on a Nikon CSU W1 spinning disc microscope in 0.5% LMP agarose and imaged every 30 s. A 500 × 500-pixel ROI within the YSL was selected with maximum intensity projections centered along the AV axis close to the margin of the EVL. The respective ROI was then used for PIV analysis using PIVlab71 and post-processed by a custom-made Python script. Flow velocities represented in the kymographs were averaged over three consecutive time points. Alignment of these flows was analyzed by measuring the angular mean of the flow vectors in the ROI as described above. Segmentation and tracking of EVL cells To visualize EVL cells, Tg(actb2:Utrophin-mcherry) or Tg(acbt2:Utrophin-mcherry, krt18:Krt18GFP) transgenic embryos were used to mark apical cell junctions. Maximum intensity projections of Z-stacks covering approximately 150 µm depth (for movies acquired with a SP8 confocal microscope) or 30 µm depth (for movies acquired on a Nikon spinning disc microscope) were generated for subsequent analysis. Segmentation of the EVL cells was performed using a “human in the loop” pipeline using the Cellpose segmentation software. Initial segmentations were generated using models trained on manually annotated datasets. These models were then employed to make further segmentations. Post-processing involved manually correcting any segmentation errors and using these corrections to further refine the segmentation models. Heightmaps were generated with the Local Z-projector plug-in in Fiji in order to correct for the curvature of the embryo. Deproj functions were used to integrate the segmentations and the heightmaps to measure EVL cell area and alignment. Data were processed and plotted using custom Python scripts. Keratin network segmentation For analyzing the keratin cytoskeleton in EVL cells, Tg(krt18:Krt18GFP) embryos were imaged at 63× magnification as described above (embryo mounting and imaging). As most of the keratin network within EVL cells formed a single sheet at the apical surface, Z-projections of the keratin network over the entire EVL cell were used. First, a Gaussian filter and denoising were applied to obtain the filamentous skeleton. Next, the images were thresholded using an adaptive median threshold and the area fraction covered by the network as an estimate of network density was determined using Fiji. For junctional keratin accumulation, the cell margins were outlined using EVL segmentations as described above (segmentation and tracking of EVL cells) with a thickness of about three pixels to obtain a skeleton junction image. The skeleton junction image was then used to segment the junctional pool of keratin in each cell and the intensity of the junctional pool was measured in the segmented image. The total intensity of keratin per cell was measured using the whole apical segmented area inside the junctional segmentations. Fluorescence in situ hybridization WT or Tg(krt18:Krt18GFP) embryos at 9 hpf were fixed in 2% paraformaldehyde at RT for 1h, followed by stepwise fixation with and subsequent storage in methanol at −20 °C. After overnight fixation in methanol the embryos were stepwise rehydrated to PBST (phosphate buffered saline, 0.1% Tween-20 pH 7.3) followed by incubation in hybridization buffer (10% Dextran sulfate, Formamide 10%, tRNA 1 mg/ml, SSC 2x, BSA 0.02%, Vanadyl-ribonucleoside complex (NEB S1402S) 2 mM) at 30 °C. Stellaris RNA FISH probes for keratin8 mRNA were designed from zebrafish mRNA sequences using LGC Biosearch Technologies’ Stellaris® RNA FISH Probe Designer and 30 probes per probe-set were ordered. The embryos were incubated in this probe-set overnight at 30 °C. The hybridized embryos were washed with wash-buffer 3 times before mounting in gold mounting media and imaged using an inverted Leica SP5 or a Leica Stellaris 5 confocal microscope as described above. Statistics and reproducibility The statistical analyses were performed using scipy stats package in Python. The number of transplants or embryos (n) and experimental replicates (N) analyzed are indicated in the figure legends. At least three independent experiments were performed in all cases whose data is provided in the Source Data file. Error bars are indicated as ribbons, whiskers or as applicable and described in each figure legend. Sample sizes were not determined using statistical tests. The statistical tests performed are collected in a Jupyter notebook available on github. Each legend describes the statistical tests performed along with the associated p values. To compare more than two groups, ANNOVA was performed. To increase statistical power, a correction for multiple comparisons was used. To compare two groups paried and unpaired t-test and Mann-Whitney test were used. Reporting summary Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article. Supplementary information Supplementary information Description of Additional Supplementary Files Supplementary Movie 1: Keratin expression within the EVL during epiboly Supplementary Movie 2: Keratin expression within the EVL in embryos with reduced YSL pulling force Supplementary Movie 3: Keratin network maturation in EVL cells of embryos with reduced YSL pulling force Supplementary Movie 4: Keratin expression within the EVL of embryos with enhanced YSL pulling force Supplementary Movie 5: Keratin network maturation in EVL cells of embryos with enhanced YSL pulling force Supplementary Movie 6: Keratin expression within the EVL of embryos with reduced keratin type II expression Supplementary Movie 7: Keratin network maturation in EVL cells of embryos with reduced keratin type II expression Supplementary Movie 8: Keratin expression within the EVL of embryo with reduced keratin8 expression and rescue with modified mRNA Supplementary Movie 9: Changes in keratin expression upon EVL aspiration Supplementary Movie 10: Changes in keratin expression during EVL wound closure after cell ablation Supplementary Movie 11: Wound closure in control and keratin-deficient embryos Supplementary Movie 12: E-cadherin expression in control and keratin-deficient embryos Supplementary Movie 13: Occludin-b expression in control and keratin-deficient embryos Supplementary Movie 14: Jup-a expression in control and keratin-deficient embryos Supplementary Movie 15: Simulated EVL wound closure in model with and without keratin feedback onto mechanics Supplementary Movie 16: Simulated EVL tissue undergoing epiboly Supplementary Movie 17: Simulated recoil of EVL tissue upon wound formation Supplementary Movie 18: Effect of normal and reduced YSL pulling force on control and keratin-deficient simulated EVL tissues Reporting Summary Transparent Peer Review file Supplementary information Description of Additional Supplementary Files Supplementary Movie 1: Keratin expression within the EVL during epiboly Supplementary Movie 2: Keratin expression within the EVL in embryos with reduced YSL pulling force Supplementary Movie 3: Keratin network maturation in EVL cells of embryos with reduced YSL pulling force Supplementary Movie 4: Keratin expression within the EVL of embryos with enhanced YSL pulling force Supplementary Movie 5: Keratin network maturation in EVL cells of embryos with enhanced YSL pulling force Supplementary Movie 6: Keratin expression within the EVL of embryos with reduced keratin type II expression Supplementary Movie 7: Keratin network maturation in EVL cells of embryos with reduced keratin type II expression Supplementary Movie 8: Keratin expression within the EVL of embryo with reduced keratin8 expression and rescue with modified mRNA Supplementary Movie 9: Changes in keratin expression upon EVL aspiration Supplementary Movie 10: Changes in keratin expression during EVL wound closure after cell ablation Supplementary Movie 11: Wound closure in control and keratin-deficient embryos Supplementary Movie 12: E-cadherin expression in control and keratin-deficient embryos Supplementary Movie 13: Occludin-b expression in control and keratin-deficient embryos Supplementary Movie 14: Jup-a expression in control and keratin-deficient embryos Supplementary Movie 15: Simulated EVL wound closure in model with and without keratin feedback onto mechanics Supplementary Movie 16: Simulated EVL tissue undergoing epiboly Supplementary Movie 17: Simulated recoil of EVL tissue upon wound formation Supplementary Movie 18: Effect of normal and reduced YSL pulling force on control and keratin-deficient simulated EVL tissues Reporting Summary Transparent Peer Review file Source data Source data Source data