Furry is required for cell movements during gastrulation and functionally interacts with NDR1

Gastrulation is a key event in animal embryogenesis during which germ layer precursors are rearranged and the embryonic axes are established. Cell polarization is essential during gastrulation, driving asymmetric cell division, cell movements, and cell shape changes. The furry (fry) gene encodes an evolutionarily conserved protein with a wide variety of cellular functions, including cell polarization and morphogenesis in invertebrates. However, little is known about its function in vertebrate development. Here, we show that in Xenopus, Fry plays a role in morphogenetic processes during gastrulation, in addition to its previously described function in the regulation of dorsal mesoderm gene expression. Using morpholino knock-down, we demonstrate a distinct role for Fry in blastopore closure and dorsal axis elongation. Loss of Fry function drastically affects the movement and morphological polarization of cells during gastrulation and disrupts dorsal mesoderm convergent extension, responsible for head-to-tail elongation. Finally, we evaluate a functional interaction between Fry and NDR1 kinase, providing evidence of an evolutionarily conserved complex required for morphogenesis.

Gastrulation is a crucial time in animal development during which major cell and tissue movements shape the basic body plan 1,2 . The morphogenetic movements of gastrulation rearrange the three germ layers precursors, positioning mesodermal cells between outer ectodermal and inner endodermal cells to shape the head-to-tail body axis. In order to break the initial "egg shape" of the embryo, cells need to polarize in a precise and coordinated manner. Cell polarity controls orientated cell division, cell shape changes, as well as cell movement. Additionally, cell polarity regulates the mechanical behaviors of the tissue, e.g. assembly of the extracellular matrix (ECM) during gastrulation and numerous other morphogenetic events [2][3][4] . The embryo of the frog Xenopus laevis is widely used as a model of cell polarization, migration, and morphogenesis due to its unique experimental advantages. The large size of the embryo and its cells allows extensive manipulation and high resolution live microscopy of explant cultures 3,5 .
At the beginning of Xenopus gastrulation, the presumptive anterior mesoderm cells located at the dorsal marginal zone (DMZ) roll inward at the midline of the blastopore lip in a process called involution. Involution follows bottle cell contraction and spreads laterally and ventrally leading to the formation of the blastopore, a ring of involuting cells that encircles the yolky vegetal endoderm cells. As involution proceeds, the blastopore progressively decreases in diameter, defining the posterior of the embryo, and closes at the end of gastrulation 2 . Simultaneously, on the dorsal side of the embryo, axial and paraxial mesoderm tissues undergo convergent extension which elongates the anterior-posterior axis and aids blastopore closure. During convergent extension, mesodermal cells polarize and intercalate with each other along the mediolateral axis, narrowing and extending the dorsal midline 6,7 . Gastrulation movements are orchestrated by a small, heterogeneous group of cells with inductive and morphogenetic properties located in the dorsal lip of the blastopore (DBL) of the amphibian

Results
Dorsal fry depletion causes axis elongation defects. We and others have previously determined that fry is expressed in the involuting mesoderm of the early gastrula, becoming restricted to dorsal tissues and lateral plate mesoderm of neurula, and remaining in somites, notochord, heart, eye, brain and pronephric kidney through tailbud stages 27,28 . We investigated in more detail its expression pattern before and during gastrulation. We found that fry transcripts are present almost exclusively in the animal half of the blastula and its expression domain encompasses the marginal zone in early gastrula ( Supplementary Fig. S1a,b). By late gastrula, fry transcripts are present in the axial mesoderm and the deep layer of the ectoderm, becoming restricted to the notochord, paraxial and lateral mesoderm in neurula ( Supplementary Fig. S1c,d). Since we were not able to detect the endogenous protein with available antibodies, we investigated Fry cellular localization in DMZ explants from fry-GFP mRNA injected embryos at a dose that did not cause an axis phenotype. Fry-GFP fusion protein 27 was mainly present in the cytoplasm and at the plasma membrane of dorsal mesodermal cells during gastrulation ( Supplementary Fig. S1e). In a previous report, Fry-GFP was detected in nuclei and cytoplasm of isolated mesodermal Xenopus cells 27 . This apparent discrepancy with our results might be related to the cellular context and the identity of the imaged cells. Cytoplasm and cell membrane localization of endogenous Fry protein has been reported in other organisms 14,15,21,24 and nuclear localization was found in Drosophila salivary gland and fat body cells 14 . Together, the evidence argues in favor of a highly mobile protein with distinct spatiotemporal cellular localization dependent on cell type and context.
We performed loss-of-function experiments by knocking down fry translation using a previously-validated antisense morpholino oligonucleotide (fry-MO) 27 . As reported, injection of fry-MO into both dorsal blastomeres of 4-cell embryos resulted in a shortened anterior-posterior axis and reduced anterior head structures at the tailbud stage ( Fig. 1a-d) 27 . This effect was dose dependent and we classified embryos as "Shortened axis" or "Shortened axis & Head-less" as a more severe phenotype when the cement gland and the optic and otic vesicles were absent (Fig. 1c,d,g). Goto et al., have shown that full-length fry injection rescues cement gland and axis elongation defects in fry morphant embryos 27 . Here, we performed rescue experiments with a short chimeric version of Fry, FD + LZ 27 , that only possesses the N-terminal FD and C-terminal LZ domains and lacks the morpholino target site. In vertebrates, FD and LZ are the only two recognizable functional domains identified in the Fry protein. Together, they have an effect on secondary axis induction very similar to that of full-length fry 27 . We sought to investigate if these domains were sufficient to rescue the axis elongation phenotype of fry morphants. Co-injection of FD + LZ mRNA led to a significant reduction in the frequency of affected embryos and the severity of the fry-depletion phenotype (Fig. 1e,g). These results indicate that the FD and LZ domains can partially compensate for Fry loss-of-function. Intriguingly, dorsal overexpression of fry has no effect on axis formation 27 30 and goosecoid (gsc) 31 genes in the presumptive prechordal mesoderm were reduced in fry-MO injected embryos ( Fig. 2a-b''). Similarly, in the presumptive axial mesoderm, the chordin (chrd) 32 expression domain was reduced in fry-depleted embryos ( Fig. 2d-d''), while expression of notochord homeobox (not) 33 in the early chordamesoderm appeared unaffected ( Fig. 2i-i''). These findings are in agreement with a previous publication 27 and suggest that Fry modulates the cell fate of specific populations of the dorsal organizer, particularly cells of the presumptive prechordal mesoderm.
In the absence of Fry, the pan-mesodermal marker brachyury (xbra) 34 was expressed at the gastrula stage ( Fig. 2f-f ''). Furthermore, at late tailbud stage, chrd in the notochord and myogenic differentiation 1 (myoD) 35 in the somites were both expressed in the absence of Fry ( Supplementary Fig. S2a-d). Despite the shortening of the anterior-posterior axis, differentiated notochord and somites (MZ15 and 12/101 antibodies, respectively) were present in both mild and more severe fry morphants phenotypes ( Supplementary Fig. S2e-h). While both structures appeared morphologically abnormal, differentiation was not severely impaired. Our data indicate that Fry is not involved in general mesoderm induction or expression of terminal differentiation markers of axial mesoderm tissues.
Next, we investigated the effect of Fry loss-of-function in the expression patterns of these genes during Xenopus gastrulation. Consistent with gastrulation defects, the blastopore remained opened beyond late gastrula stages (St. 13) in most fry-depleted embryos ( Fig. 2c'' ,e'' ,g'' ,j'' ,k''). In addition, the expression domains of chrd and gsc were not only reduced, but abnormally positioned in these embryos (Fig. 2c-c'' ,e-e''). Chrd-positive cells marking the axial mesoderm appeared to reflect defects in involution ( Fig. 2e''). Furthermore, the expression domain did not extend along the anterior-posterior axis nor converge mediolaterally to the midline (Fig. 2e-e''). In addition, movement of gsc expressing prechordal mesoderm cells toward the animal pole was impaired in fry morphants (Fig. 2c-c''). The expression patterns of xbra and not at late gastrula stage were also abnormally positioned www.nature.com/scientificreports/ reflecting defects in midline elongation in fry-depleted embryos ( Fig. 2g-g'' ,k-k''). The neuroectodermal not expression that initiates at St13.5 indicates that gene expression patterns are not altered in fry-MO injected embryos due to developmental delay, but rather reflect defects in gastrulation movements in these embryos. At neurula stage most embryos closed their blastopore, however, the notochord does not extend normally in fry depleted embryos ( Fig. 2h-h'' ,l-l''). The elongation of dorsal midline tissues was quantified as the length of not gene expression domain in early and late-gastrula embryos 36 . As expected, dorsal mesoderm elongation was reduced in the absence of Fry and partially rescued by FD + LZ mRNA injection ( Supplementary Fig. S3). In light of these results, we decided to further characterize the defective morphogenetic movements associated with the loss of Fry function.
Fry function is necessary for normal blastopore closure. Blastopore closure is a major contributor to gastrulation in amphibians involving the coordination of multiple morphogenetic movements in the embryo 3,37 .
To assess blastopore closure progression from the initial step of blastopore formation at the early gastrula until its closure at the early neurula, we measured the blastopore area on fixed embryos at different stages ( Fig. 3a-h). While most uninjected embryos closed their blastopore by late gastrula stage, the blastopore remained open in most of fry-depleted embryos at this stage. Yet, by mid-neurulation, all embryos close their blastopore with the exclusion of some with extruding yolk plugs (25%), suggesting either a non-specific developmental delay or a specific disruption of dorsal morphogenesis (Fig. 3i). Confirming the rescue of the fry morphant phenotype observed at the tailbud stage by co-injection of FD + LZ mRNA (Fig. 1), the blastopore closure defect induced by fry knock-down was also partially rescued by FD + LZ (Fig. 3i). The rescue of the fry knock-down argued against a non-specific delay, so we sought to further characterize the dynamics of blastopore closure. We acquired time-lapse sequences of gastrulating embryos from the onset of gastrulation until blastopore closure (Supplementary Movie S1). In these time-lapses (n = 12) we observed that blastopore formation, which is initially restricted to a small region on the dorsal side of the uninjected embryo, was laterally expanded in most fry morphants (10/12 embryos) (yellow dotted arrows, Supplementary Fig. S4). We saw a complete yolk-encircling ring of bottle cells forming in both groups of embryos. Additionally, by late gastrula stages, most fry-depleted embryos closed their blastopore concentrically (9/12 embryos) as opposed to the characteristic dorsal-dominated eccentric closure of the blastopore toward the ventral side (see Supplementary Movie S1 and Supplementary Fig. S4). Hence, Fry depletion in dorsal tissues alters blastopore formation and the dynamics of blastopore closure, but bottle cell formation appears coincident with control embryos.

Fry depletion affects movement of superficial involuting marginal zone cells during gastrulation.
Since blastopore closure is driven by multiple processes including convergent thickening, convergent extension, and involution 2,37 , we first investigated the impact of Fry depletion on the motion of the superficial cells during involution 38 . To achieve this, we injected dorsal blastomeres with H2B-eGFP mRNA with or without fry-MO and mounted the embryos for light-sheet fluorescence microscopy. This technique allowed us to visualize and track dorsal cell nuclei over several hours with negligible levels of photobleaching in intact, live gastrulat-   (Fig. 4a). As a measure of motion directionality, we evaluated the persistence of each cell as it moved towards the DBL. Persistence was defined as the ratio between the linear distance traveled by a cell and the total length of its path: cells that move in a straight line will have persistence of 1, whereas cells that move in a more erratic trajectory will have lower persistence. We observed that Fry depletion significantly reduced directional persistence with a mean of 0.93 ± 0.06 for cells of control embryos versus a mean of 0.85 ± 0.05 for cells of fry-depleted embryos (Fig. 4b). For control embryos, the majority of the cells (> 50%) had persistence values between 1 and 0.95 while very few cells had values below 0.70. By contrast, for fry-depleted embryos, fewer than 20% of the cells fall within the 1-0.95 persistence interval while the remaining 80% had lower persistence values (Fig. 4c). These results indicate that cells trajectories towards the DBL are severely affected in the absence of Fry (Fig. 4c,d). Next, we defined instantaneous velocity as the spatial displacement over time for two consecutive frames. In consistency with the observed blastopore closure delay, Fry depletion significantly reduced the instantaneous velocity of individual cells as they move towards the DBL (Fig. 4e).
To investigate whether the spatial organization of cells on the superficial involuting marginal zone (IMZ) was affected in fry-depleted embryos as a result of the observed velocity and persistence changes, we estimated the distance between nearby cell nuclei as a function of the distance from the DBL (see Methods). While cells 75-95 μm from the DBL showed a similar distance to neighboring cells in control and fry-depleted embryos, cells closer to the DBL (up to 65 μm) presented a significant reduction of the distance to neighbors in fry morphants (Fig. 4f). This result reveals an alteration of the geometrical organization in the pre-involution zone (up to 30 μm from the DBL) of the superficial IMZ as a result of Fry depletion. The observed increase of cell density or "bunching" in the proximity of the involution site in fry-depleted embryos, could indicate that while superficial IMZ cells are experiencing convergence forces, they are not being pulled over the blastopore lip by forces generated by post-involution tissues. These results demonstrate a dorsal-requirement for Fry during involution movements and suggest that convergent thickening might be operating in fry morphants allowing blastopore closure, even under conditions where convergent extension is likely defective 2,37 .
Fry function is necessary for formation of the cleft of Brachet and the associated fibronectin network. At the start of gastrulation, a deep cleft named the cleft of Brachet forms around the marginal zone, separating mesendoderm from prospective pre-involution mesoderm and non-involuting ectoderm. The cleft is thought to form through changes in cell adhesion and vegetal rotation 41,42 . Leading-edge mesendodermal cells on one side of the cleft contact and move along the ectoderm blastocoel roof (BCR) on the other side of the cleft. The interface becomes the site where an ECM rich in fibrillar fibronectin is assembled that is essential for mesendoderm migration 43,44 . As mesodermal cells (first prechordal mesoderm and later chordamesoderm) move over the base of the cleft, the so-called inner lip, post-involution cells acquire the capacity to remain separated from the pre-involution and non-involuting ectodermal cells.
To assess the role of Fry in the formation of the cleft of Brachet, we analyzed hemisected fixed fry morphant and control embryos. In fry-depleted embryos, we found that while the cleft of Brachet was detected anteriorly, the boundary between the involuted mesoderm and pre-involuting cells was not observed near the blastopore lip (Fig. 5a-c and insets a'-c'). Given that abnormalities in the formation of the cleft of Brachet are frequently associated with tissue separation defects, we evaluated this behavior using ectoderm/BCR assays 44,45 . Involuted mesoderm test aggregates of uninjected and fry-MO injected embryos remained on the explanted BCR surface, demonstrating that internalized mesodermal cells are capable of maintaining separation from the explanted BCR cells (Supplementary Fig. S5). As expected, aggregates of uninjected and fry-MO injected embryos isolated from the inner layer of the BCR merged into the explanted BCR surface, as those cells do not express separation behavior ( Supplementary Fig. S5). Finally, as the cleft of Brachet is later stabilized by the assembly of a fibronectin fibrillar matrix 44 , we investigated its formation in early gastrula fry-depleted embryos. We observed that the normally abundant fibronectin fibrils that form in the cleft were reduced (Fig. 5d-f and insets d' ,e'), while other populations of fibrils on the BCR were not affected (Fig. 5d,e black arrowheads).
Taken together, our findings indicate that while the formation of the posterior-most domain of the cleft appears affected by loss of Fry function, post-involution cells have the ability to remain separated from noninvoluted ectodermal cells measured by BCR assays. Further, we suspect that in fry morphants, the failure of fibronectin matrix assembly alters the stability of the cleft.
Fry is required for elongation and mediolateral orientation of dorsal mesoderm cells. Convergent extension within dorsal midline tissues extends the body axis and aids blastopore closure 5,37 . To drive convergent extension, mesoderm cells must undergo directed cell rearrangement through mediolateral cell intercalation, a cell behavior marked by mediolateral cell elongation and alignment, called mediolateral intercalation behavior (MIB) 46 . To evaluate whether Fry plays a role in regulating the MIB, we assessed cell shape and orientation in DMZ explants isolated from embryos dorsally injected with fry-MO and the mRNA of a membrane marker (mem-mScarlet) 47 . We quantified the orientation of dorsal mesoderm cells within the explants by plotting the cell's major angle with respect to the mediolateral axis on a rose diagram. Mesodermal cells within mem-mScarlet injected explants orient along the mediolateral axis on fibronectin-coated substrate (Fig. 6a,c). By contrast, mesodermal cells in explants injected with fry-MO under the same culture conditions fail to orient (Fig. 6b,d). The degree of cell shape polarization or "polarity index" of dorsal mesoderm cells was measured as the ratio between cell major and minor axes 48,49 . Cells lacking Fry exhibit a lower polarity index (1.64 ± 0.02) relative to cells from embryos injected with the mRNA alone (1.92 ± 0.03) (Fig. 6e). These experiments demonstrate that Fry function is required for the expression of MIB in the dorsal mesoderm. In the absence of Fry, dorsal mesoderm cells presented a smaller area than cells in the control group evidencing additional changes in cell shape (Fig. 6f).  www.nature.com/scientificreports/ Human NDR1 kinase partially rescues axis elongation of fry-depleted embryos. In mammalian cells and invertebrates, Fry orthologs genetically and physically interact with NDR kinases regulating their activity [14][15][16]18,21,24,26 . We decided to investigate whether NDR1 kinase plays a role downstream of Fry in anterior-posterior axis formation and the morphogenetic events of gastrulation. To this end, we overexpressed a wild-type version of human NDR1 (hNDR1-wt), a constitutively active form named hNDR1-PIF 50 , which mimics an active kinase, and a kinase-dead version named hNDR1-kd 51 . Dorsal injection of all hNDR1 forms caused reduction of head structures and mild axis elongation defects in ~ 50% of the embryos (Supplementary Fig. S7a-d). These results indicate that dorsal overexpression of all three human NDR1 functional variants affects anterior-posterior axis development in Xenopus. Additionally, we evaluated the ability of hNDR1 variants overexpressed in ventral blastomeres to induce secondary axes, as this capacity has been reported for fry ventral overexpression 27 . In agreement with a previous report by Goto et. al., 27 hNDR1-wt had no effect on axis or head formation (Supplementary Fig. S7e). Similarly, neither hNDR1-PIF nor hNDR1-kd ventral overexpression affected axis formation or generated an ectopic axis ( Supplementary Fig. S7f,g).
To evaluate Fry and NDR1 functional interaction in axis elongation and development of anterior structures, we performed injections of the different hNDR1 mRNAs individually or with fry-MO. Dorsal co-injection of hNDR1-wt or hNDR1-kd failed to rescue Fry depletion (Fig. 7b,c,e,f), while co-injection of hNDR1-PIF mRNA resulted in a significant suppression of the tailbud fry morphant phenotype (Fig. 7d,f). Although most embryos co-injected with hNDR1-PIF mRNA and fry-MO presented axis elongation defects and therefore scored as "shortened axis", < 80% body length relative to the average length of control embryos, the truncation of the axis was clearly less severe in these embryos relative to fry morphants (Fig. 7b,d and see Supplementary Table S1).

hNDR1-PIF partially rescues impaired convergent extension in fry-depleted embryos.
To address whether the defective MIB exhibited by fry-depleted cells results in impaired convergent extension, we analyzed the elongation of DMZ explants 52 . Unlike control explants that elongate as a result of convergent extension (Fig. 8a,b), elongation of DMZ explants from fry-depleted embryos was strongly inhibited (Fig. 8c,h). Additionally, coinjection of FD + LZ mRNA along with fry-MO significantly restored explant elongation (Fig. 8d,h), consistent with the axis elongation rescue at tailbud stage (Fig. 1). These results indicate that defective convergent extension is, at least in part, responsible for the shortened axis phenotype.
Next, we evaluated whether the different functional variants of hNDR1 were able to rescue convergent extension in fry-depleted embryos. The analysis of DMZ explants elongation showed that hNDR1-PIF partially compensates for Fry loss-of-function (Fig. 8f,h). However, when hNDR1-wt or hNDR1-kd were co-injected with www.nature.com/scientificreports/ fry-MO, explants elongated to a small degree, but the differences with explants from embryos injected only with fry-MO were not significant (Fig. 8e,g,h). Complementary to these results, we quantified dorsal mesoderm elongation in intact neurula-stage embryos by measuring the length of the not expression domain. We observed a modest but significant rescue when co-injecting hNDR1-PIF along with fry-MO ( Supplementary Fig. S6). Together, these results show that in the absence of Fry function, only constitutively active hNDR1 is able to rescue convergent extension, suggesting that kinase activity is required. The lack of significant rescue by hNDR1-wt further suggests that activation of this kinase still requires Fry function. Our data are in agreement with previous findings showing that Fry functionally interacts with NDR1 kinase, and support our hypothesis that Fry function in dorsal axis elongation and convergent extension is mediated, at least in part, by the activation of NDR1 kinase.

Discussion
Despite the fact that Fry is an evolutionarily conserved protein from yeast to humans, very little is known about its function. At the beginning of gastrulation, fry transcripts are present in the marginal zones supporting a role in the initial gastrulation movements. We found that Fry loss-of-function consistently affects the movements and spatial configuration of the superficial and deep IMZ cells. As gastrulation proceeds, fry mRNA is enriched in the axial and paraxial mesoderm and the deep ectodermal layer, all tissues that undergo convergent extension. In this regard, our studies demonstrate a requirement of Fry in regulating cell behaviors that underlie morphogenetic movements of midline tissues.
During amphibians gastrulation, two intrinsic convergence behaviors of the IMZ contribute to blastopore closure: convergent thickening and convergent extension. While convergent extension is exclusively conducted by the presumptive dorsal tissues of the mid-gastrula, convergent thickening operates in all pre-involuting circumblastoporal tissues in a symmetric fashion. Moderate dorsal patterning defects have little impact on convergent extension as embryos lacking a notochord can form tadpoles that are indistinguishable from wild-type 53,54 . As it was demonstrated in severely UV ventralized embryos, convergent thickening can drive blastopore closure on its own in the absence of convergent extension 37,55,56 . We observed that, similar to severely ventralized embryos, dorsal Fry depletion alters the initial blastopore formation and blastopore closure dynamics, suggesting that www.nature.com/scientificreports/ convergent thickening still operates in the absence of Fry. Additionally, our cell tracking recordings of frydepleted embryos reveal alterations in the movement of superficial IMZ cells and the spatial organization of the cells near the blastopore lip, revealing a lack of coordination between the forces that operate in the system. The fibronectin component of the ECM plays a critical role in the regulation and maintenance of gastrulation movements and blastopore closure 48 . Fibronectin in the cleft of Brachet provides physical support for mesendoderm cell migration 43,57,58 , as well as a source of signaling 59,60 . Loss of fibronectin can lead to alterations in the migratory kinetics of leading-edge mesodermal cells 48 and therefore might also contribute to the abnormal development of anterior structures in fry-depleted embryos 29,61 . As in fibronectin knock-down embryos, convergent extension is disrupted in fry-depleted embryos. However, DMZ cells lacking Fry fail to elongate and orient when provided an exogenously fibronectin substrate, suggesting the existence of an additional mechanism affecting the morphogenetic behavior of these cells. Based on these migratory defects and loss of cell polarization and orientation, it is possible that Fry regulates cytoskeleton dynamics. In Drosophila, Fry has been found to alter actin organization during egg chamber elongation and hair wing morphogenesis, as well as to regulate microtubule sliding in neurons 13,15,62 . Given that Fry binds to microtubules and regulates their dynamics in Drosophila and mammalian cells 21-23 , controlling chromosome alignment, mitotic spindle orientation and morphogenesis, www.nature.com/scientificreports/ it is also possible that defects in cell division could contribute to the phenotype of fry morphants. Whether Fry regulates cytoskeletal components involved in morphogenesis and cell division remains to be determined. Fry also promotes dendrite attachment to the ECM in Drosophila dendritic arborization neurons, but the mechanism remains elusive 63 . Whether Fry is required for intercellular adhesion or cell-ECM adhesion via integrins in our system, will require further investigation. Convergent extension is under the control of multiple molecular pathways 64 , among them, importantly, the non-canonical Wnt/PCP signaling pathway 4,[65][66][67][68][69] . In addition to regulating cell polarity and the coordination of morphogenetic behaviors, the PCP pathway regulates polarized ECM deposition required for convergent extension 66,70 . Many characteristics of the Fry loss-of-function phenotype, such as blastopore closure defects, convergent extension impairment, dorsal mesoderm MIB disruption, and the reduction of the fibronectin fibrillar matrix assembly along the mesoderm surface, resemble PCP pathway perturbations 66 . While evidence of genetic interactions between PCP components and the Fry-Tricornered pathway (Tricornered, Drosophila ortholog of NDR1) have been described for dendritic self-avoidance 71 , further studies will be required to determine whether there is physical and/or functional interaction between Fry and the PCP signaling pathway in Xenopus.
Previous studies, and our findings suggest the involvement of Fry function in early cell fate decisions in Xenopus regulating anterior specification. The development of head abnormalities in the fry morphants could likely be attributed to early prechordal mesoderm genes misexpression. However, as it was shown in prior studies of PCP mutants, defects in anterior patterning can arise when patterns of cell motility are disrupted 72 . In this sense, it will have to be determined whether the morphogenetic defects associated with Fry loss-of-function have an impact on early patterning.
Multiple studies in invertebrates and ours here in Xenopus point to a role of Fry protein in cellular mechanisms driving morphogenesis. Fry was originally identified in Drosophila melanogaster where its mutation causes disorganized epidermal cell morphology 13 . A very similar phenotype found in tricornered gene mutants suggested that these proteins physically interact and function in a common pathway 13,14,18 . Our rescue experiments show that human constitutively active NDR1, in contrast to the wild-type and kinase-dead variants, can partially compensate for the loss of Fry function, suggesting that Fry is necessary for NDR activation during axis development in Xenopus. Our results argue in favor of a conserved functional interaction between Fry and NDR kinases in animal development. The partial rescue of Fry-depletion phenotypes by NDR1 suggests that this kinase might not be the only protein mediating Fry function in these processes. In fact, Fry appears to have NDR-independent functions in mammalian cells [22][23][24] and in nematodes 16 . Moreover, in Xenopus embryos, ventral overexpression of Fry induces a secondary axis, while it has been shown here and in a previous work 27 that NDR1 does not have this capacity. The specific function of Xenopus Stk38 (stk38; orthologue of human NDR1) in axis elongation and convergent extension movements will require further investigation. Further, the precise mechanism by which overexpression of hNDR1 functional variants lead to an abnormal axis development will have to be determined. Similarities between the phenotypes associated with loss-of-function and gain-of-function are not uncommon to the events that regulate gastrulation movements 11,68,73 . We can speculate that hNDR1-kd is acting in a dominantnegative fashion by sequestering necessary interacting proteins. Alternatively, the observed phenotype could be explained by a kinase-independent mechanism resulting in a gain-of-function. Discrimination between these possibilities will require further analysis and exceeds the scope of this work.
In the present work, we show for the first time that Fry is a regulator of cell movements and morphogenesis during gastrulation. Genetic studies in yeast, nematode and fruit fly have revealed a critical role of Fry in morphogenesis and cell polarity associated with NDR activation; however, the molecular and cellular mechanisms are not well understood. Future research on the identification of additional upstream and downstream factors will provide important insight into the mechanism of action of Fry.

Materials and methods
Ethics statement. This study was carried out in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the NIH and also the ARRIVE guidelines. The animal care protocol was approved by the Comisión Institucional para el Cuidado y Uso de Animales de Laboratorio (CICUAL) of the School of Applied and Natural Sciences, University of Buenos Aires, Argentina (Protocol #64).

Xenopus embryo preparation.
Xenopus laevis embryos were obtained by natural mating. Adult frogs reproductive behavior was induced by injection of human chorionic gonadotropin hormone. Eggs were collected, de-jellied in 3% cysteine (pH 8.0), maintained in 0.1 X Marc's Modified Ringer's (MMR) solution and staged according to Nieuwkoop and Faber 74 . The embryos were placed in 3% ficoll prepared in 1 X MMR for microinjection.
Constructs for mRNA synthesis. Human NDR1 constructs were generously provided by Alexander Hergovich. The hNDR1-PIF (constitutively active), hNDR1-wt (wild-type) and hNDR1-kd (kinase-dead) 50,75 cDNAs were excised from pcDNA3.HA.hNDR1-PIF, DNA3.HA.hNDR1-wt and pcDNA3.HA.hNDR1-kd (K118A) by BamHI and XhoI digestion and cloned into pCS2 + . The Fry-GFP construct was generously provided by Toshiyasu Goto 27 . Constructs for pCS2 + .H2B-eGFP 76 and pCS2 + .HA.FD + LZ 28 have been described previously. pCS2 + .mem-mScarlet was generated using mScarlet 77  In situ hybridization and immunostaining. Whole-mount in situ hybridization was carried out as previously described 78 . Fry (Dharmacon), chordin (gift from Edward De Robertis) and the myoD constructs (gift from Oliver Wessely) were linearized as previously described 28,32,79 . The Brachyury construct (gift from Neil Hukriede) was linearized with BglII, the not construct (gift from David Kimelman) was linearized with HindIII, the gsc construct (gift from Neil Hukriede) was linearized with EcoRI, and the otx2 construct (gift from Ira Blitz) was linearized with EcoRI. All linearized constructs were transcribed with T7 for antisense probe synthesis with the exception of gsc which was transcribed with SP6. For whole-mount immunostaining with MZ15 (DSHB Cat# MZ15, RRID:AB_760352) and 12/101 (DSHB Cat# 12/101, RRID:AB_531892), we followed the protocol previously described 79 . Fibronectin immunostaining with 4H2 (DSHB Cat# 4H2, RRID:AB_2721949) and subsequent clearing and mounting for confocal microscopy was performed as previously described 43  Image analysis. Images of fixed whole embryos were collected with a Leica DFC420 camera attached to a Leica L2 stereoscope. Histological slides were imaged using a digital camera (Infinity 1; Lumera Corporation) attached to a light-field microscope (CX31: Olympus). For gastrulation time-lapse sequences, late blastula (St. 9) uninjected and fry-MO injected embryos were selected and transferred to custom acrylic chamber with 0.1 X MMR. Gastrulation was recorded for approximately 13 h at 18 °C. Images were taken every 3 min using an Imaging Source SMK 23G445 camera attached to a Zeiss Axiovert S 100 scope. Morphology of gastrulating embryos was quantified from fixed embryos using ImageJ software (https:// fiji. sc/). Blastopore closure was calculated as the ratio of the blastopore area over the area of the vegetal hemisphere of the embryo. Elongation of dorsal midline tissues evaluated as the length of the not expression domain divided by the length of the whole-embryo, as previously described 36 .
Embryo microsurgery (DMZ explants). Following microinjection, early gastrula embryos (St. 10.5) were transferred to 1 X Modified Barth´s Saline (MBS) solution. Vitelline membranes were removed with forceps and DMZ explants were microsurgically isolated using hair tools 49,80 . To asses CE, the isolated explants were placed in plastic dishes, immobilized with a cover glass and cultured until co-cultured intact siblings reached late neurula stage (St. 19). The percentage of elongation was calculated as previously described 52,68 . For confocal microscopy, explants were prepared in Danilchik's for Amy (DFA) culture media and mounted in custom acrylic chambers on glass prepared in advance with 20 μg/ml fibronectin (Sigma) 81 . The explants were imaged with a confocal microscope (SP5, Leica Microsystems). For Fry-GFP visualization in mesodermal cells, DMZ explants were allowed to adhere and heal for 30-40 min prior to imaging. For cell membrane visualization with mem-mScarlet, explants were cultured until sibling embryos reached late gastrula stage (St. 13/14). Cell orientation, polarity index, and cell area were calculated as previously described 49,66 using image analysis software (ImageJ, https:// fiji. sc/).
Light-sheet fluorescence microscopy. Following microinjection, late blastula embryos (St. 9) were placed in pre-warmed (22 °C) 0.2% agarose (Biodynamics) prepared in 0.1 X MMR. Embryos were drawn into 4 cm long FEP tubes (Rotilabo, D-76185 inner diameter 1.58 mm, outer diameter 3.18 mm) with a needle and syringe 82 . The tube was plugged with solid 1% agarose and the samples were imaged using a custom-built lightsheet fluorescence microscope (LSFM) 39 . This microscope uses a cylindrical lens to generate a static Gaussian light-sheet that is focused into a sample using an illumination objective (Olympus UMPLFLN 10XW, 0.3 NA). The resulting light-sheet has a thickness of around 5 μm over a field of view of around 1 mm × 1 mm. The emitted fluorescence is collected using a detection objective (Olympus UMPLFLN 20XW, 0.5 NA) followed by a filter wheel, a tube lens and an sCMOS camera (Andor Zyla). Samples were scanned across the light-sheet using a stepper motor and the whole microscope was controlled using a custom-made graphical interface. 3D stacks of injected embryos were acquired upon visualization of the blastopore formation and every 3 min using an axial step of 5 μm over a distance of 400-600 μm. The resulting images were segmented and tracked using ilastik 40 . We analyzed the tracking results using custom-made Python scripts. We defined a cell's instantaneous velocity v as v = �r/�t , where r is the cell's spatial displacement over two consecutive frames and t the corresponding frame rate (3 min in all our experiments). We defined persistence of motion as the ratio between the linear distance traveled by a cell and the total length of its migration path 83 . In order to measure velocity and persistence of motion of dorsal superficial involuting marginal zone (IMZ) cells, we selected cells that: (i) were tracked for at least 45 min, (ii) moved a distance of at least 20 μm and (iii) were closer than 80 μm to the DBL at the last www.nature.com/scientificreports/ time point that they were tracked. The distance to the nearest neighbors was calculated as follows: for a single time-point (t = 150 min), the center of each nucleus as reported by ilastik was fed into the Delaunay algorithm (Python 3.6, SciPy 1.3) to obtain the list of nearest neighbors (NN). The Euclidean distance from each nucleus to their NN was calculated and then averaged. The distance of each cell from the blastopore lip was defined as the closest distance from the nucleus to a curve that was manually drawn over the lip for the frame of interest (region size = 30 μm; overlapping region size = 5 μm).
BCR assays. Following microinjection, early gastrula embryos (St. 10.5) were transferred to 1 X Modified Barth´s Saline (MBS) solution. BCR assays to assess tissue separation behavior were performed as previously described 45 . Briefly, prechordal or deep layer ectodermal cells derived from uninjected or fry-MO injected embryos were dissected into small cell aggregates (test aggregates) and placed upon an uninjected BCR. Explants were immobilized with a cover glass and separation behavior was scored after 45 min.