A cell group in the axial mesoderm regulates notochord morphogenesis in zebrafish. a Characterization of the prechordal plate (ppl) by whole-mount double in situ hybridization (ISH) of wild-type (WT) zebrafish embryos with the indicated markers. Scale bar: 100 µm. b Mapping of pcdh18a mRNA expression (red) relative to Tg(gsc:GFP) expression labelled by anti-GFP antibody (green). Notably, gsc:GFP labels the prechordal plate (ppl) and the trailing notochord (NC). Scale bar: 100 µm. c Inhibition of nodal signalling by SB505124 treatment (30 µM) from 4 to 8.5 h post-fertilization (hpf). Scale bar: 100 µm. d Confocal image of live zebrafish embryo at 5 hpf showing subcellular localization of Pcdh18a-GFP. Glycosylphosphatidylinositol-anchored mCherry (memCherry) marks cell membranes. Arrows indicate punctae of Pcdh18a localization at the membrane (open arrows) and intracellularly (closed arrows). Scale bar: 10 µm. e WT embryos or Tg(gsc:GFP) embryos were injected with Pcdh18a MO (0.5 mM). Morpholino-based knockdown of Pcdh18a leads to a wider and shorter notochord marked by ntl expression at 9 hpf (arrows). See Supplementary Fig. S1h for control experiments. Analysis of the shape of the notochord in a cross section of gsc:GFP transgenic embryos that were injected with the indicated constructs at 10 hpf. At 11 hpf, the body length was significantly shorter in the Pcdh18a-deficient embryos, as shown in an ISH-based analysis of notochord hgg/ntl (n = 20/32, arrows). Confocal microscopy-based analysis of cell shapes in the notochord of embryos that were microinjected with the indicated constructs at 12 hpf (cells with exemplary morphology were surrounded with a yellow circle). Scale bar: 100 µm. fpcdh18a MO-injected embryos show a wider axial mesoderm compared to Ctrl MO injected embryos (four embryos each). g Circularity of NC cells was measured in total of 1000 cells in five different embryos each. A circle has a circularity of 1.0, while noncircular shapes have a lower value of circularity. The error bars represent the SEM and significance as indicated (***p value < 0.001; unpaired Student’s t test)

Analysis of the influence of the ppl on notochord morphogenesis. a–d 3D-optic flow analysis. a Spherical mapping of the optic flow of the fluorescent signal of a gsc:GFP/memCherry embryo onto a spherical coordinate system with θ as azimuth angle and ϕ as polar angle. b 2D Mercator projection of flow field of mesoderm (red arrows) compared with ectoderm above (black arrows). Both embryos analysed at 5 hpf. c The corresponding heat map kymograph shows the relative velocity of the mesoderm roughly every 5 min. d Thick solid (dashed) lines are smoothed average profiles of wild-type (WT; Pcdh18a deficient) embryos, with individual embryo profiles shown in lighter lines in ϕ (red/orange) and θ (light/dark blue) directions. epcdh18a MO donor shield was transplanted into WT hosts and vice versa. After 3 h, the embryos were fixed and subjected to ISH against ntl. Donor cells are marked in red. Arrows indicate the width of the notochord. f Quantifications display mean value, standard error of mean (SEM), and significance level of six independent embryos per experiment as indicated (*p value < 0.01; unpaired Student’s t test). g Ablation of cell rows in the ppl (5th GFP positive cell row) or at the ppl-notochord border (15th GFP positive cell row) in the Tg(gsc:GFP) fish line. Embryos were injected with a nuclear marker (Histone 2B-mCherry) and cell rows were ablated at 7 hpf using ultrashort laser pulses of a two-photon microscope. Embryos were raised to 10 hpf, fixed, and subjected to ISH against ntl. After ablation of a cell row in the ppl, embryos develop an elongated notochord (n = 11/11), whereas the notochord progenitor cells move slower and a gap appears towards the ppl in embryos with ablation of a cell row at the ppl–notochord border. Consequently, the trailing ntl expression domain remains shorter and broader (n = 6/10, white arrows). Yellow arrows mark the ablated cell rows. Scale bar: 100 µm

Pcdh18a regulates recycling of the E-cadherin. a Confocal image of zebrafish embryo at 5 hpf. Embryos were microinjected with 0.1 ng of mRNA for the indicated constructs and were imaged in vivo at 5 hpf. Pcdh18a is localized in the cell membrane and in endocytic vesicles, together with E-cadherin (E-cad). b Quantification of the E-cad levels in the Pcdh18a-transfected L cells. Equivalent amounts of lysates from murine L cells or stably E-cadherin-GFP-transfected L cells that had been transfected with Pcdh18a were Western blotted and probed with an anti-GFP antibody; the results showed a 26% increase in the E-cadherin-GFP protein levels after Pcdh18a transfection. The sample blot shows different parts of the same blot and PCNA was used as a loading control. The experiments were performed in independent triplicate (*p value < 0.05; unpaired Student’s t test). c Endocytic routing of E-cad at 50% epiboly. WT embryos and Tg(rab5-GFP), Tg(rab7-GFP), and Tg(rab11-GFP) stable transgenic embryos were microinjected with 0.1 ng of the mRNAs for the indicated constructs. Arrows indicate E-cad localization with Rab proteins and Lamp1-positive vesicles. d Pearson’s co-localization coefficient was calculated from 70 µm thick confocal stacks of five different embryos, each from c. The error bars represent the SEM and significance, as indicated (*p value < 0.05, **p value < 0.01; unpaired Student’s t test). Scale bar: 10 µm

Pcdh18a domains and their importance with regard to endocytosis. a After photobleaching of a 3 µm spot at the cell membrane of E-cadherin-expressing embryos, new E-cadherin-GFP molecules moved into the bleached area from adjacent membrane regions, resulting in a return of 90% of fluorescence within 4 min 20 s (blue). Co-expression of Pcdh18a increased the speed of recovery and a 90% recovery was reached after 2 min 20 s (orange). Co-expression of Pcdh18a-ECD diminished FRAP of E-cadherin-GFP (green). Moving-average trendline was calculated with period 3. b Confocal images of zebrafish embryos at 5 hpf (50% epiboly). Embryos were injected with 0.1 ng mRNA of indicated constructs. In the deletion construct Pcdh18a-ECD, the intracellular domain was replaced by a mCherry domain. Pcdh18a-mCherry was localized to vesicles, whereas Pcdh18a-ECD-mCherrry was strongly localized to the cell membranes. Pcdh18a-GFP/Pcdh18a-mCherry and Pcdh18a-GFP/Pcdh18a-ECD-mCherry showed co-localization at the membrane and in vesicles suggesting homophilic interaction. Pcdh18a-mCherry/E-cadherin-GFP suggest heterophilic interaction. Pcdh18a-ECD-mCherry and E-cadherin-GFP were observed mainly at the membrane and did not co-localize suggesting that the intracellular domain of Pcdh18a is required for interaction and co-internalization with E-cad. Scale bar: 10 µm. cTg(gsc:GFP) embryos were injected with 0.1 ng of the e-cadherin-mCherry mRNA or co-injected with the pcdh18a MO (0.5 mM) and subjected to confocal microscopy analysis at 8 hpf. A cross section at the level of the ppl reveals enhanced E-cad localization at the plasma membrane and in endocytic vesicles, as shown by a projection of five fluorescence intensity histograms of five different embryos. Scale bar: 100 µm and 50 µm, respectively. d Bean plots shows the distribution, means, and standard deviations of the sizes of e-cadherin-GFP clusters in the lateral and axial mesodermal plate measured in 20 WT and pcdh18a morphant embryos

Pcdh18a affects E-cadherin dependent cell migration. a Wound-healing assay in HeLa cells. Cells were transfected with the indicated constructs and their migratory behaviour was monitored for 10 h after removing the insert. The dotted line shows limits of the confluent cell layer. b Quantification of the migration speed of HeLa cells. c Wound-healing assay in L cells. Cells were transfected with indicated constructs. After 24 h, L cells were treated with DMSO (1%) or Dyngo4a endocytosis inhibitor (1 µM in DMSO 1%). The migratory behaviour of cells was monitored in a time lapse for 34 h. d Quantification of the migration speed of L cells after blocking endocytosis. All wound-healing assays were conducted in independent triplicate, distances of the gap were measured at ten fixed positions. Mean values, SEM and significance are indicated (*p value < 0.05, ***p value < 0.005; unpaired Student’s t test). Scale bar: 200 µm

Directed cohort migration results in the formation of the rod-shaped notochord. a Visual rendering of the results of the simulations. See Supplementary Fig. S6 for the description of the cellular Potts model. b Embryos were microinjected with mRNAs for the indicated constructs (pcdh18a: 0.3 ng, e-cad: 0.4 ng, dyn2K44A: 0.2 ng). At 5 hpf, approximately 50 cells were grafted into the lateral embryonic margin of uninjected host embryos (n = 5). At 8 hpf, the migration and the directionality of the cell clusters were analysed. Animal pole was set to 0°, vegetal pole was set to 180°. Blue line indicates mean value of clonal coverage measured in ten different embryos per experiment and white lines indicate SEM. c Embryos from b were fixed and subjected to ISH for the lpm marker myf5. Horizontal cross sections revealed the formation of an ectopic rod-shaped structure of the Pcdh18a-positive clones in the lpm (yellow arrow). d Schematic summary of the function of the ppl in notochord morphogenesis. Pcdh18a/E-cadherin adhesion complexes (orange dots) increase cell adhesion within the ppl, leading to the cluster formation (left). In parallel, Pcdh18a controls endocytosis of E-cadherin adhesion complexes to allow fast cohort migration of the ppl cluster (right) to orchestrate intercalation of notochord cells. Scale bar: 100 µm