In preprints: expanded insight into epithelial spreading during zebrafish epiboly Free

The focus in tissue mechanics has largely been on forces generated by the actin cytoskeleton (Charras and Yap, 2018; Rauzi, 2020). Naik and colleagues expand on this by showing a role for keratin intermediate filaments in balancing the structural integrity of the EVL with the need to be deformable (Naik et al., 2025). Keratin intermediate filaments can confer mechanical resilience to epithelial cells and play roles in epithelial differentiation, collective cell migration, and response to tensile forces (Dmello et al., 2019; Latorre et al., 2018; Long et al., 2006; Nahaboo et al., 2022; Ramms et al., 2013). Naik and colleagues showed that several keratin members are primarily expressed in the EVL and YSL. The density of the EVL keratin networks increased concomitantly with EVL spreading, which they reasoned was related to a build-up of tension in the EVL. Consistent with this, manipulating the pulling activity of the YSL affected keratin dynamics in the EVL: increasing YSL myosin activity (and thus EVL tension) accelerated keratin network maturation, while decreasing myosin activity (and EVL tension) slowed it. The findings were corroborated by locally increasing EVL tension using micropipette aspirations, which resulted in keratin accumulation in aspirated EVL cells. Overall, these results indicated that keratin network maturation in the EVL is mechanosensitive.

The EVL is a network of interconnected cells, so a challenge is ensuring that cells deform collectively to enable coordinated sheet spreading without tearing. The authors proposed that the mechanosensitive accumulation of keratin networks at sites of high tension may enable the EVL to rapidly enhance its mechanical resilience in response to stronger deforming forces. In agreement with this, Naik and colleagues showed that knocking down several keratin members resulted in reduced viscosity of the EVL and increased EVL rupture towards late epiboly. They further explored how keratin influences the coordination of cell movements using a wound-healing assay. When clusters of EVL cells were laser ablated in control embryos, it triggered elongation of distant cells towards the wound, highlighting the robust ability of the EVL to preserve its barrier integrity through long-range, coordinated cell movements. In contrast, when keratin was knocked down, only cells closely neighbouring the wound site elongated in response, implying a loss of long-range coordination. Since keratins enhance EVL viscosity, they proposed that this allows mechanical forces to be propagated through the tissue, so individual EVL cells deform together. Since EVL cells closest to the YSL should be subject to a stronger pulling force, this mechanical coupling could ensure that EVL cells farther from the leading edge also deform accordingly.

Although their main focus was on the role of keratin in the EVL, an interesting finding by Naik and colleagues was that keratin filaments played a role in the YSL. Their in silico model predicted that reduced keratin, and thus reduced resistance to deformation, would result in more rapid EVL advancement, but their experimental results showed that epiboly was instead delayed. Reconciling these findings, they found that keratin was also required in the yolk for aligned actomyosin flows, which are essential for generating the pulling force on the EVL (Behrndt et al., 2012). There seems to be an interplay between keratin and actin in the yolk, as micropipette aspiration revealed that keratin is required for actin to accumulate at sites of local tension in the yolk. This finding is noteworthy as it highlights the involvement of keratin in both the EVL and the yolk, suggesting a broader coordination, not just between cells within a tissue, but also across different tissues.

While this study advances our understanding of the role of keratin in epithelial morphogenesis, several questions remain. Although keratin filaments were mostly discussed collectively in this work, distinct keratin members might play specific roles in tissue morphogenesis. Individual keratin types may interact within the network and play non-redundant roles (Mazzalupo et al., 2003; Nanes et al., 2024), so whether different keratin isoforms contribute in a spatially and temporally distinct manner in the EVL would be interesting to address in the future. In addition, the precise molecular interactions between keratins and other cytoskeletal elements, such as actin filaments, need further exploration. There is a well-established functional cross-talk between keratin and actin filaments in epithelia, contributing to mechanical resilience and migration (Kölsch et al., 2009; Pora et al., 2020; Yoon and Leube, 2019). It was shown that keratin networks in the EVL accumulate under increased tension; however, how this accumulation of keratin relates to potential changes in actin within the EVL remains unclear. In terms of their activities, how actin-driven forces in the EVL might be integrated with the structural support role of keratin is an important yet missing piece of the puzzle. One possibility is that this interaction could take place at the level of cell–cell junctions (Outla et al., 2025; Prechova et al., 2022). Addressing these interactions could shed light on the general principles of cytoskeletal coordination in tissue-scale epithelial morphogenesis. Finally, the role of keratin in mechanical resilience was not discussed in the context of cellular rearrangements, which occur in the EVL during epiboly (Campinho et al., 2013; Keller and Trinkaus, 1987; Köppen et al., 2006). This raises the question of how the mechanical resilience provided by keratin filaments in EVL cells is balanced with the need for cells to occasionally break their adhesions to rearrange with neighbouring cells.

While Naik and colleagues focused on the role of keratin in reinforcing tissue integrity, the preprint by Minsuk and colleagues highlights the importance of active cellular rearrangements for EVL spreading (Minsuk et al., 2025). Using the simulation framework Tissue Forge (Sego et al., 2023), they constructed an agent-based model in which EVL cells are treated as ‘individual agents’, allowing the model to capture the cellular rearrangements of individual EVL cells in response to being stretched. In their simplified model, EVL cells are represented as centre-based particles with attractive forces (bonds, representing cell adhesion) and repulsive forces (preventing cell overlap) between them. Neighbour exchanges are simulated by the breaking of bonds and formation of new ones, with EVL boundaries not explicitly depicted. Asymmetries in morphogenesis from axial development are also not included; instead, a uniform pulling force from the yolk cell is modelled. Despite these simplifications, their model was able to capture the extensive cell rearrangements that occur in the EVL, with greater rearrangements near the leading edge, consistent with reports in live embryos (Keller and Trinkaus, 1987; Köppen et al., 2006). In addition, straightening of the leading edge emerged spontaneously and robustly in the model without being explicitly programmed. Edge straightening has been observed in live embryos but has not been deeply investigated before.

In their simulations, the authors found that imposing a constraint that promotes hexagonal cell packing in the EVL was necessary to prevent sheet tearing under tension, enabling yolk engulfment and closure. Relaxing this constraint increased cellular rearrangement rates at the margin and, interestingly, also increased the rate of EVL leading edge straightening. This relationship suggests that irregular cell packing and faster cell rearrangement near the leading edge produces a more fluid-like tissue, which facilitates straightening. The possibility that leading edge shape could serve as a readout of tissue fluidity could motivate future experiments to measure and manipulate this feature in vivo. Since increased cell rearrangement in embryos would require active remodelling of junctions, these results also raise questions about the cues that initiate this process near the leading edge and how junctional proteins and the cytoskeleton are regulated in response.

The authors were also interested in how synchrony of EVL cell spreading arises, another understudied behaviour seen in live embryos. They could enforce synchrony in the model by making the YSL pulling force proportional to the distance of each marginal EVL cell from the vegetal pole, so that lagging cells are pulled more strongly. A regulatory mechanism to synchronise the advancing EVL margin has not previously been postulated. Whether such a synchronisation mechanism exists in live embryos is an open question, and one can speculate about some sort of feedback between EVL cells and the YSL to regulate force distribution and maintain coordinated spreading.

The two recent preprints from Naik et al. (2025 preprint) and Minsuk et al. (2025 preprint) provide insights into how the EVL deforms viscoelastically in response to pulling from another tissue, and how the behaviours of individual EVL cells can be coordinated to prevent tearing of the sheet. How the EVL balances its requirements for mechanical resilience with dynamic rearrangement is an area worthy of future investigation and should consider the different tissues involved during epiboly, as well as the different cytoskeletal and junctional components at play.

We thank Tony Harris for helpful comments on the manuscript.