Zebrafish Fak1a is functionally conserved with mammalian FAKs. (a) Zebrafish Fak1a and Frnk1a were successfully expressed in 293T cells as characterized by immunoblotting (IB) using anti-HA (left) and anti-green fluorescent protein (GFP) antibodies (right). (b) Zebrafish Fak1a and Frnk1a with an HA tag were expressed in FAK^−/− MEF cells as revealed by HA immunostaining shown in green (left column). They were present in focal contacts (boxed), and an enlarged image is shown at the top right corner of the respective panel. Focal contacts and cell nuclei were visualized by Paxillin (second column) and DAPI staining (third column), respectively. Merged images for Fak, Paxillin and DAPI staining are presented in the right column with enlarged insets for the boxed regions to better show the co-localization of Fak and paxillin in focal contacts. Cells were transfected with pEGFP-C3 vector only, fak1a or frnk1a to examine cell proliferation (c) and cell migration (d) as determined by a BrdU incorporation assay and Bodyen chamber migration assay, respectively (n = 3; ***p < 0.001). Detailed protocols are described in ‘Material and methods'.

Zebrafish fak1a morphants reveal severe gastrulation defects. (a) Zebrafish embryos were treated or untreated with a fak1a translation blocking morpholino (fak1a tMO₁) and immunoblotted (IB) against Fak1a and β-actin (loading control). The intensities of Fak1a bands were normalized to that of β-actin. The relative knockdown levels of Fak1a in treated embryos are shown by comparing band intensity to that of untreated embryos (right panel, n = 3; **p < 0.01). (b) The epiboly progression of fak1a tMO₁-treated embryos was either delayed (ii) or arrested (iii) compared to the normal standard control MO-injected embryos (i) at 10 hpf as presented in bright-field images. Black and red arrows point to the running fronts of the enveloping layer and deep cell layer, respectively. All images have the animal pole (A) placed at the top and the dorsal to the right (D). (c) Embryos were treated as indicated and classified into different categories as indicated, and percentages of embryos in each category are shown (n = 3, **p < 0.01; ***p < 0.001). The total number of embryos used in each treatment is shown at the bottom of each bar. (d) Proteins of embryos injected with or without 2.5/5 ng fak1a tMO₁ were extracted and immunoblotted using indicated antibodies. β-Actin was used as a loading control.

Loss of Fak1a perturbs the synchronized migration of enveloping and deep cell layers and the F-actin network. (a) Embryos untreated or injected with 5 ng of a fak1a translation blocking morpholino (tMO₁) were fixed at 8 hpf and subjected to F-actin/DAPI staining and Fak1a immunohistochemistry. Embryos were examined and photographed for the whole embryo image (upper) or a region flanking the yolk syncytial at a higher magnification (lower) under confocal microscopy. Representative photographs of different channels and merged images are shown. White arrowheads point to actin rings. Yellow and green arrowheads point to the running fronts of the enveloping (EVL) and deep cell layers (DCL), respectively. (b) Graphic demonstration of the average gap between EVL and DCL in untreated and fak1a tMO1-treated embryos (n = 3; n = 30, **p < 0.01). (c) The EVL/DCL gap smaller than 100 µm were considered normal, and the percentages of normal embryos are presented (n = 3; n = 30, ***p < 0.001). (d) The actin bundles (marked by white asterisks) between the actin ring and vegetal actin cap were clearly reduced in fak1a morphants, and the numbers of actin bundles are quantified in the right panel (n = 3; n = 10, ***p < 0.001). Disorganized YSL nuclei (yellow asterisks) were also observed in fak1a morphants (see DAPI staining and merged images). As indicated by the arrows in the top right corner, all images have the animal pole (a) placed at the top and the dorsal to the right (d).

Loss of Fak1a causes abnormal convergence and extension movements. (a) Embryos injected with a designated amount of a fak1a translation blocking morpholino (tMO₁) were subjected to WISH against indicated genes at the bud stage. Representative WISH staining photographs against ctsl1b/dlx3 or ctsl1b/ntl are presented for untreated and fak1a tMO₁-treated embryos (fak1a tMO₁). ctsl1b, dix3 and ntl staining were used to label the prechordal plate (pcp), ectodermal borders (ecb) and notochord (nt)/tail bud (tb), respectively. For simplicity, tissues are labelled in untreated embryos only. To reveal dorsal convergence, embryos were probed with ctsl1b/dlx3 as shown in the anterior view (left). A round prechordal plate is in the middle with two ectodermal borders forming a V-shape. Lines were drawn along the ectodermal borders that matched the prechordal plates. The V-shape formed an angle as indicated by A° and A′° in untreated and fak1a tMO₁-treated embryos, respectively. To reveal the dorsal extension, embryos were probed with ctsl1b/ntl as shown in the lateral view with the dorsal to the right (right). Lines were drawn from the anterior front of the prechordal plate and the posterior end of the tail bud to the centre of the embryo forming an angle towards the dorsal as indicated by B° and B′° in untreated and fak1a tMO₁-treated embryos, respectively. (b) The relative convergence and extension defects were quantified by calculating the ratios of A′°/A° (left panel) and B′°/B° (middle panel), respectively. The normality of anterolateral migration of the prechordal plate was examined by signals of ctsl1b and dlx3 in embryos injected with a designated amount of the fak1a tMO₁ (right panel). n = 3, *p < 0.05, **p < 0.01, ***p < 0.001, n.s., not significant. (c) Embryos were injected with different amounts of the fak1a tMO₁, cultured to the 8-somite stage, photographed (live embryo) or fixed, and subjected to WISH against ctsl1b/myo D. Representative dorsal view photographs are shown in (c), and statistical comparisons of embryos with normal somites are presented in (d). As described in (c,d), the anterior and posterior extensions of the dorsal axes were examined. Representative lateral-view photographs with the anterior to the left are shown in (e), and statistical comparisons of embryos with normal dorsal axis extension are presented in (f).

Loss of Fak1a perturbs hypoblast cell migration. Embryos were injected with 5 ng of a random control morpholino (MO) N-25 (control) or fak1a translation blocking MO (tMO₁), immobilized and monitored under differential interference microscopy. Time-lapse movies were taken for 10 min during the 75–90% epiboly stage to reveal the involuting cell migration of the anterior prechordal plate or convergent movement of lateral cells (see electronic supplementary material, movies S1–S4). (a) Representative snapshots of the prechordal plate or lateral cells at the end of representative recordings are shown. More than six cells were selected from an embryo to be traced in each movie, and their migrating routes are depicted by a rainbow line representing the recording time at 0–10 min. a′, anterior; a, animal pole. These experiments were repeated at least three times. (b) The moving direction (arrow direction) and migration distance (arrow length) of each traced cell are represented by an arrow. The origin (o′) of the coordinate plane stands for the starting point of each cell. a′, anterior; a, animal pole; d, dorsal side. (c) The polarity of each cell was measured as described in Results, and the analysis showed a significant loss in the polarity of the prechordal and lateral cell migration in fak1a tMO₁-treated embryos (n ≥ 3; n ≥ 36). (d) The migration velocity, migration rate, tortuosity (route/distance) and protrusion persistence of each recording were analysed with Simple PCI software, and comparisons between groups are shown. n.s., not significant, *p < 0.05; **p < 0.01, ***p < 0.001.

Fak1a functions non-cell-autonomously to regulate cell migration during gastrulation. (a–d) Rhodamine-labelled blastomeres were transplanted from embryos injected with 5 ng of StdMO (Std) or tMO₁ (MO) with rhodamine dextran to untreated hosts (UT) or tMO₁ morphant hosts (MO). Host embryos were then imaged (animal pole on the top and vegetal pole at the bottom) under epifluorescence microscopy, recorded and representative snapshots are shown in (a) STD > UT: StdMO-treated cells in an untreated host. (b) STD > MO: StdMO-treated cells in a tMO₁ morphant. (c) MO > UT: tMO₁-treated cells in an untreated host. (d) MO > MO: tMO₁-treated cells in a tMO₁ morphant. Arrows indicate the representative cellular protrusions in each embryo. (e) The average protrusion numbers per embryo were counted from each recording and shown. The total number of embryos used for each treatment is shown on the bottom of each bar. Values between groups with a significant difference (p < 0.05) are denoted by different letters. (f) The transplanted cells were traced and their curvillinear velocity (Vcl) and strait line velocity (Vcl) were calculated and shown. Values between groups with a significant difference (p < 0.05) are denoted by different letters. (g) The distribution of protrusions formed from donor cells transplanted to host is shown. The centre of a cell is regarded as the centroid of the rose diagram. A rose diagram is divided into eight equally parts with designated angles. The 0°, 90° and 180° points to the animal pore, dorsal side and vegetal pore, respectively. The percentages of protrusion per cell were calculated and plotted on rose diagrams. The effective protrusions were marked in red and the ineffective ones are marked in dark. Y-axis for the rose diagrams represents the percentage of protrusions in each direction bin. (h) The percentage of effective protrusions per cell in each group were shown (n = 3, *p < 0.05).

Overexpression of fak1a rescues gastrulation defects in wnt5b-deficient embryos. (a) Wild-type or wnt5b/ppt mutant embryos were untreated or injected with fak1a mRNA, cultured until 48 h post-fertilization (hpf), and photographed. The percentages of embryos with dead, pipetail or normal phenotypes were calculated and shown (right). The total number of embryos used for each treatment is shown at the bottom of each bar. (b) The average trunk lengths among the three groups were compared (n = 3, **p < 0.01, ***p < 0.001). (c) Embryos were injected with 3.75 ng wnt5b morpholino (MO) without (control) or with 200 pg fak1a mRNA and photographed at 10 hpf. wnt5b MO-injected embryos showed different degrees of epiboly defects, which was rescued by a fak1a mRNA co-injection. Representative photographs are shown at the top and quantitative analysis shown at the bottom. (d) Embryos were injected with designated MOs, cultured to the bud stage, lysed and subjected to a qPCR to determine fak1a or wnt5b expression. (e) Embryos were injected with a designated amount of the wnt5b MO, collected at the bud stage and subjected to immunoblotting against indicated antibodies specific to focal adhesion kinase (FA) (C-20) and different FAK phosphorylation sites. β-actin served as an internal control.

Fak1a and Wnt5b reciprocally rescue convergence defects in wnt5b and fak1a MO-injected embryos, respectively. (a,b) Embryos were untreated, injected with 7.5 ng of the wnt5b MO only or co-injected with both 7.5 ng of the wnt5b MO and 200 pg of fak1a mRNA. (c,d) Embryos were untreated, injected with 5 ng of the fak1a tMO₁ only, or co-injected with both 5 ng of the fak1a tMO₁ and 10 pg of wnt5b mRNA. Embryos were then cultured to the bud stage and subjected to WISH against ctsl1b, dlx3 and ntl to analyse convergence and extension as described in figure 4. Representative photographs are shown in (a) and (c), and quantitative analyses are shown in (b) and (d) (n = 3, **p < 0.01, ***p < 0.001).

Wnt5b and Fak1a modulate Rac1 and Cdc42 to control cell migration during gastrulation. (a,b) Embryos were untreated or injected with 7.5 ng of the wnt5b MO, 5 ng of fak1a or 1.25 pg of rac1 and cdc42 mRNA. Embryos were cultured to the bud stage and classified into normal, mild/severe convergent extension (mild/severe C/E) defects, and epiboly arrest categories. The percentages of embryos in different categories are shown. Numbers of embryos observed are given at the bottom of each bar. n = 3. Values between groups with a significant difference (p < 0.05) are denoted by different letters. (c) Embryos injected with the wnt5b or fak1a MO were collected at the bud stage to measure the expression of cdc42 or rac1 by a qPCR, respectively. Ef1α served as an internal control. (d) Embryos were treated as in (c) and lysed to measure the activities of Cdc42 and Rac1 by an ELISA activity assay (n = 3, *p < 0.05). (e) Embryos were treated with fak1a MO with or without rac1 and cdc42 mRNA, stained and examined as described in figure 3d. The actin bundles between the actin ring and cap are indicated by asterisks. The numbers of actin bundles in each treatment were presented in the right bar graph (n = 3, *p < 0.05, ***p < 0.001).

CRISPR/Cas9-mediated deletion of Fak1a results in mild gastrulation defects due to compensatory wnt5b expression. (a) The fak1a gene structure is shown with indicated exons in blue boxes. The target sites of a primer pair (F, forward primer; R, reverse primer) for amplifying a mutation detection 513-bp PCR fragment are indicated by arrows. The sense strand of wild-type fak1a is shown (upper strand). Exon 10 contains a PAM site shown in red. A gRNA target site of the exon 10 PAM site is labelled in blue. The Hae III site is shown in orange. The resulting sense strand of the fak1a Δ5 allele is shown (lower strand) with the translated amino acid in grey. (b) Partial chromatograms are shown for the wild-type and fak1a Δ5 alleles. Deleted nucleotides are indicated by asterisks. (c) Illustration of wild-type FAK1a and Fak1a Δ5 mutant proteins. The gRNA target site is indicated by an arrow. The mutant protein contains F1, F2 and a partial F3 domain that encodes 274 amino acids. (d) Hae III restriction digestion analysis. Genomic DNAs from tail fins of wild-type (+/+), heterozygous (+/−) or homozygous (−/−) fak1a were isolated and amplified by PCR using the fak1a forward and reverse primers indicated in A. The amplicons were digested, run in agarose gels and stained. A representative gel image is shown indicating selective molecular weight markers and sources of genomic DNAs. (e) Three different batches of wild-type and fak1a Δ5 mutant embryos were lysed and subjected to immunoblotting against a Fak c-terminal antibody. 293T cell lysate was used as a positive control, and zebrafish Rpl7a was an internal control. (f) Embryos were subjected to WISH against ctsl1b/dlx3 and ctsl1b/ntl to analyse convergence and extension as described in figure 4. In the scatterplot, each dot represents relative convergence, and the extension defect of each embryo underwent different treatments (n = 4). (g) Wild-type embryos and fak1a Δ5 MZ mutant embryos were injected with or without a designated subthreshold amount of the wnt5b MO, cultured to the bud stage, and the resultant gastrulation defects were classified into normal, mild convergence and extension (C/E) defects, severe C/E defects and epiboly arrest as shown in representative photographs (side view, dorsal up, anterior to the left). The calculated percentages of embryos are shown for each class. The numbers of embryos observed are given at the bottom of each bar. n = 3. Values between groups with a significant difference (p < 0.05) are denoted by different letters.

Wnt5b integrates with Fak1a to mediate cell movements during gastrulation. Gastrulation cell migration is controlled by the Wnt5b pathway to activate calcium and Dsh/Daam1. Herein, we identified that Wnt5b can also integrate with FAK signalling, and both Wnt5b and Fak1a then activate small GTPase Rac1 and Cdc42 to mediate actin dynamics during gastrulation. Convergent is mainly mediated by Fak1a-mediated cell migration and extension is exerted by cell intercalation, which requires both Fak1a and Wnt5b.