Tracing the evolution of the luminal membrane using the polarity marker Podocalyxin points at the biological significance of apicobasal polarity establishment in EHT cell emergence complexity

A, cartoons depicting the early and late steps of EHT cells emerging from the aortic floor (steps 1 and 2, as previously described in the zebrafish embryo, see (Kissa and Herbomel 2010; Lancino et al. 2018)) and with hypothetical evolution of the luminal membrane (in green) before (3 and 4) and after the release (2’, the cell detaches from the endothelial layer via junction downregulation leading to exposure of the luminal membrane with le extracellular milieu; 3’, the luminal membrane is consumed via endocytic recycling (E) and/or lysosomal degradation (Lys) prior to detachment; 4’, the luminal membrane in 4 is released inside the cell (twisted arrow) before detachment. Grey area= nucleus. B, PodocalyxinL2 (Podxl2) construct designed to establish transgenic fish lines. Cartoons representing full length (top drawing) and deleted Podxl2 (amino-acid sequence 341-587) in which the mucin domain (serine/threonine-rich O-glycosylation domain) is replaced by either eGFP or mKate2. The tetracystein-containing globular domain (subjected to N-glycosylation) was kept as favoring apical membrane retention. TMD, transmembrane domain; DTHL, C-terminal peptidic motif involved in partnership with PDZ domain containing proteins. C-D, EHT performing cells visualized using Tg(Kdrl:Gal4;UAS:RFP;4xNR:eGFP-Podxl2) embryos and time-lapse sequences initiated at 55 hpf obtained with spinning disk confocal microscopy (imaging was performed at the boundary between the most downstream region of the AGM and the beginning of the caudal hematopoietic tissue). Top grey panels show the green, eGFP channels for eGFP-Podxl2 only. Bottom panels show the merge between green and red (soluble RFP, in magenta) channels. Bars: 8µm. C, single plane images of 2 EHT pol+ cells extracted from a time-lapse sequence at t = 0 and t = 45 min, with the right cell (eht cell 2) more advanced in the emergence process than the left one (eht cell1). Note the enrichment of eGFP-Podxl2 at the luminal membrane (surrounding the cavity labelled with an asterisk) in comparison to the basal membrane (white arrow). Note also the evolution of the luminal membranes with time, with the aortic and eht cell 1 lumens still connecting at t = 0 (green arrow), the apparent fragmentation of the cytosolic vacuole (2 asterisks for eht cell 1 at t = 45 min) and the compaction of Podxl2-containing membranes for eht cell 2 at t = 45 min. More details on the evolution of the connection between the aortic/eht cell lumens are shown in Figure 1 – figure supplement 1A. D, single plane images of 2 EHT pol-cells extracted from a time-lapse sequence at t = 30 min and t = 210 min (see Figure 1 – video 3 for the full-time lapse sequence), with the right cell (eht cell 2) slightly more advanced in the emergence than the left one (eht cell 1, with the latest attachment point between the emerging cell and the aortic floor (pink arrow)). Note, in comparison with the cells in panel C, the ovoid shapes of cells, the absence of enrichment of eGFP-Podxl2 at luminal membranes (green arrows) and the accumulation of eGFP-Podxl2 at basal membrane rounded protrusions (white arrows).

Immature HE is not polarized and controls membrane delivery of intra-cytosolic vesicular pools

A, Tg(Kdrl:Gal4;UAS:RFP;4xNR:eGFP-Podxl2) embryo imaged using spinning disk confocal microscopy. Black and white images show eGFP-Podxl2 only. Images were obtained from a time-lapse sequence (initiated at 35 hpf) lasting for 435 min (7.25 hrs), with intervals of 15 min between each z-stack. Example of an HE cell with equal partitioning of eGFP-podxl2 between luminal and abluminal membranes (at t = 0 min), with eGFP-podxl2 containing intra-cytosolic vesicles (one labelled with a green asterisk) and undergoing mitosis at t = 30 min (HE cell 1’ and HE cell 1’’ are daughter cells). Note the inheritance of the largest micropinocytic-like vacuole by HE cell 1’ and its maintenance through time until EHT emergence initiation at t = 180 min (green asterisk in 1.5x magnified areas at t = 60 and 90 min). At t = 360 min (green box) both fluorescence channels are shown; bottom panel: green (eGFP-Podxl2), magenta (soluble RFP). The magenta arrow points at the basal side of the EHT pol + cell that does not contain any detectable eGFP-Podxl2; on the contrary, eGFP-Podxl2 is enriched at the luminal/apical membrane (note that exocytosis of the large vacuolar structure may have contributed to increase the surface of the apical/luminal membrane (the green asterisk labels the lumen of the EHT pol + cell). The green arrow points at the abluminal membrane of the EHT cell derived from HE cell 1’’ and that contains eGFP-Podxl2 (with no evidence of a significant expansion of a luminal/apical membrane); this indicates that this cell is more likely to be an EHT pol-cell that does not sort the vesicular cargo to the luminal/apical membrane. Bar = 10µm. B, model summarizing the evolution of HE cells that involves the tuning of apicobasal polarity thus leading to cells competent for giving birth to either EHT pol+ or EHT pol-cells (including the oriented release of large vesicular macropinocytic-like vacuoles preferentially toward the luminal membrane of future EHT pol + cells). The polarity status of the HE is proposed to evolve throughout the entire time window of the EHT, leading to asynchrony in its ability to give birth to EHT cells (emergence of EHT pol+ and EHT pol-cells are both observed until 72 hpf, see main text).

Molecular variants and deletion mutants to investigate the control of apicobasal polarity during the 30-32 and 48-50 hpf time-windows

A, Expression of Pard3ab and bb isoforms in the zebrafish embryo at 35 hpf using whole mount in situ hybridization (WISH). Expression in the dorsal aorta, in the trunk region (delimited areas), is marked by arrowheads as well as in the veinous plexus of the caudal hematopoietic tissue, in the tail. ey = elongated yolk. B, Pard3ab and bb zebrafish isoforms and natural variants to investigate the regulation of apicobasal polarity during the 30-32 hpf and 48-50 hpf time windows. The cartoons show the 3 sequential PDZ domains (PDZ1-3), the CR1 and CR3 conserved regions involved in oligomerization and atypical protein kinase C (aPKC) binding, respectively. For each isoform, the variants that we have cloned either contain or exclude (+/-exon) the peptides encoded by specific exons (see Materials and Methods); of relevance for our purpose (investigating the expression of potential interfering isoforms), we focused on isoforms deleted from exons encoding for part of the PDZ domains (covering the amino-terminus of Pard3ab PDZ2 and the carboxy-terminus of Pard3bb PDZ3). Note that deletion of exon 10 of Pard3ab triggers a frame shift in the ORF leading to a complete deletion of PDZ2 and to a premature stop codon upstream of PDZ3. In this case, this should lead to the synthesis of a deleted proteins with major loss of function or, alternatively, the degradation of the mRNA via alternative splicing activated non-sense mediated decay (NMD, as exemplified in the EMT process, see (Pradella et al. 2017)). The position of the primers used to amplify cDNA fragments is depicted (fw (forward), rev (reverse), magenta: primers for WISH probes; see Materials and Methods). Note that, in this study and for the PDZ1 domain of Pard3ab, we did not address the expression levels of the isoforms containing or not the insert encoding for the YVFR peptide. C, left: cartoons representing the zebrafish full-length runx1a amino-acid sequence and the dt-runx1 mutant deleted from the trans-activation domain and of the C-terminus (note that the construct encodes for a C-terminal fusion with eGFP that is released upon expression via a cleavable T2A peptide (introduced between the 2xHA tag and the N-terminus of eGFP, to prevent from potential steric hindrance). nls, nuclear localization signal. Right: anti-HA tag immunofluorescence obtained after z-projection of the dorsal aorta of a 50 hpf Tg(dt-runx1) embryo. Note the localization of the 2xHA-tagged dt-runx1 protein in nuclei (some of them are pointed by red arrowheads) and of eGFP in nuclei and the cytosol of aortic cells. Bar: 25µm.

Pard3 isoforms and variants are differentially regulated during the 30-32 and 48-50 hpf time-windows

qRT-PCR analysis of gene products of interest in trunks isolated from 30-32 hpf and 48-50 hpf embryos. Trunks were collected from eGFP positive or eGFP negative incrosses of heterozygous Tg(dt-runx1) fishes (dt-runx1 incross and sibling controls, respectively) and from Tg(kdrl:Gal4;UAS:RFP) embryos injected or not at the single cell stage with the sih morpholino (morpholino sih). Graphs show the measured mean fold changes relative to the expression of ef1α and to the expression in control embryos. Statistical tests: two sided unpaired two samples Wilcoxon test, with p-values of significant differences. A, gene expression levels of (left) the 3 isoforms of Pard3 proteins (Pard3ab, bb, b and ba, see Materials and Methods for accession numbers and primers) and (right) the hematopoietic marker cmyb and the polarity protein Scrib (for the cellular basolateral domain establishment/maintenance). B, Pard3ab and Pard3bb gene expression levels for dt-runx1 mutants (red boxes) and for the sih morpholino (yellow boxes), all relative to control conditions. C, Pard3ab +/-PDZ2 (top 2 panels) and Pard3bb +/-PDZ3 (bottom 2 panels) gene splicing variants expression levels for dt-runx1 mutants (left panels, red boxes) or the sih morpholino (right panels, yellow boxes), all relative to control conditions.

eGFP-Jam3b localization is reinforced at antero-posterior sites of the endothelial/EHT interface and at tri-cellular junctions

A, cartoons representing homodimers of full-length JAMs with the C-terminal cytosolic part interacting with Pard3 (JAMs interact with the first PDZ domain of Pard3) as well as the constructs generated in this study and containing eGFP inserted between the Immunoglobulin-like (Ig) domains and the trans-membrane region (TMD). The constructs were obtained for zebrafish JAM2a and JAM3b. B, C, 52 hpf Tg(kdrl:eGFP-Jam3b; kdrl:nls-mKate2) embryos were imaged in the trunk region (AGM) using spinning disc confocal microscopy. The panels are either maximum z-projections (top two) or single plane z-sections (bottom two, focusing on the aortic floor) of aortic segments, with either the merged nls-mkate2 and eGFP-Jam3b fluorescence signals (magenta and green) or the eGFP-Jam3b signal only (black and white images). Bottom of the Figure: 2D-cartographies obtained after deploying aortic cylinders and showing the eGFP-Jam3b signals only. B, example of an EHT pol+ cell (cell 1, white arrows point at reinforcement of signal at antero-posterior junctions). On the 2D cartography, cell 1 (red) is contacting endothelial cells 2 and 3; note the reinforcement of eGFP-Jam3b signals along antero-posterior membrane interfaces perpendicular to blood flow (red arrows) as well as at the two tri-cellular junctions visible between cells 1, 2 and 3 (black arrows). C, example of two EHT pol-cells (cells 1 and 2, white arrows point at reinforcement of signal at antero-posterior junctions). On the 2D cartography, cells 1 and 2 (red) are contacting endothelial cells 3, 4, 6 and 3, 6 respectively; note the reinforcement of eGFP-Jam3b signals along antero-posterior membrane interfaces perpendicular to blood flow (red arrows) and endothelial cell 6 that has intercalated between endothelial cell 7 and EHT pol-cell 2 (blue arrow). In right margins, magenta and green arrowheads designate the aortic floor and roof, respectively. Bars = 10µm.

Junctional recycling at tri-cellular contacts is differentially controlled between the two EHT types

48-55 hpf Tg(kdrl:eGFP-Jam3b; kdrl:nls-mKate2) embryos were imaged using spinning disc confocal microscopy and illuminated for Fluorescence Recovery After Photobleaching (FRAP) in the trunk region (AGM, Aorta Gonad Mesonephros). A, B, panels are either maximum z-projections (top left) or single plane z-sections (bottom left and top right, focusing on the aortic floor) of aortic segments, with either the merged nls-mkate2 and eGFP-Jam3b fluorescence signals (magenta and green) or the eGFP-Jam3b signal only (black and white images). White arrows point at reinforcement of signal at antero-posterior junctional pools of an EHT pol+ cell (A) or of an EHT pol-cell (B), both marked by asterisks. Bottom right: 2D-cartographies obtained after deploying aortic cylinders and showing the eGFP-Jam3b signals only. Black arrows point at antero-posterior junctional pools, in particular at tri-junctional regions that exhibit increase in signal density (well visible in A, black arrows). 2 and 3 endothelial cells are contacting the EHT pol+ cell (A) and the EHT pol-cell (B), respectively. In right margins, magenta and green arrowheads designate the aortic floor and roof, respectively. Bars = 20µm. C - G, FRAP analyses. EGFP-Jam3b junctional pools corresponding to the brightest spots inside junctional regions of interest were bleached for FRAP measurements (these high intensity pools were localized at the level of bi- and tri-junctions for endothelial cells (EC) and in tri-junctional regions for EHT pol+ and EHT pol-cells; all these junctional pools were systematically visualized by deploying each aortic segment before bleaching as shown in the 2D-cartographies in A and B as well as in Figure 6 – figure supplement 1, see also Materials and Methods). FRAP analysis concerned 3 types of junctional interfaces: between endothelial cells (EC – EC, black and grey), EHT pol- and endothelial cells (pol-– EC, brown), EHT pol+ and endothelial cells (pol+ – EC, blue). C, D, evolution of mean fluorescence intensity for each type of junctional interface over time (10 min), after photobleaching (t = 0s). E, median maximum amplitude of recovery of all determinations and for each type of junctional interface (maximum of simple exponential fitted curves). F, G, early fluorescence recovery. Early evolution (over the first 30s) of the mean fluorescence intensity for each type of junctional interface over time after photobleaching (t = 0s). F, the fitted lines correspond to linear regressions of mean fluorescence intensities. G, median values of fluorescence recovery slopes (linear regressions) of all determinations and for each type of junctional interface. Statistical tests: two sided unpaired two samples Wilcoxon test.

Interfering with ArhGEF11/PDZ-RhoGEF function leads to the accumulation of hemogenic cells and impairs EHT progression

A-C, numeration and morphometric analyses of aorta and cell types for Tg(kdrl:eGFP-Jam3b; kdrl:nls-mKate2) ArhGEF11 exon 38 splicing morpholino-injected and control embryos (A), or for (Kdrl:eGFP-Jam3b; kdrl:ArhGEF11CRISPR-Cterdel+/+) homozygous ArhGEF11 C-ter deletion mutants and control siblings (B). 48-55 hpf embryos were imaged using spinning disk confocal microscopy. Aa, Bb, 2D-cartographies obtained after deploying aortic cylinders and showing the eGFP-Jam3b signals only with cell contours delineated either in blue (endothelial cells), yellow (hemogenic cells, see Materials and Methods for their morphological definition), red (morphologically characterized EHT cells, for controls), and small cells delineated by cyan boxes (morphologically uncharacterized EHT cells and putative post-mitotic cells remaining as pairs, included in the numeration as hemogenic cells). Cellular contours have been semi-automatically segmented along the cellular interfaces labelled with eGFP-Jam3b (see Materials and Methods). Scale bars: 10µm. Aa’, Bb’, left: numeration of endothelial, hemogenic and EHT-undergoing cells according to the position of their geometrical center (either on the aortic floor, or on the roof, or on the lateral side), for each condition; right: number of endothelial, hemogenic and EHT-undergoing cells in each condition calculated from the segmentation of 3 x 2D-projections per embryo and covering the entire aortic regions in the trunk. Aa’’, Bb’’, left: length of hemogenic cells (in the longest axis) in function of their orientation (°) relative to the blood flow axis (0 – 180°); right: distribution of the orientation of hemogenic cells relative to the blood flow axis, displayed as a mean distribution of cells per embryo. Aa’’’, Bb’’’, hemogenic cell elongation factors in arbitrary Units (scale factor given by the ratio between the first- and the second-best fitting ellipse diameters, the minimum value being 1 for a non-elongated object) represented as boxplot distribution of all segmented cells (left) or as the distribution of cell elongation factor per embryo (right), for controls and for interfering conditions as indicated. Aa’’’’, Bb’’’’, hemogenic cell area represented as boxplot distribution of all segmented cells (left) or as the distribution of cell area per embryo (right), for controls and for interfering conditions as indicated. C, aortic perimeter (in µm) for controls and mutant conditions as indicated. Statistical tests: two sided unpaired two samples Wilcoxon test. For the ArhGEF11 exon 38 splicing morpholino condition, analysis was performed on 2 x control (non-injected embryos) and 3 x embryos injected at the one-cell stage; for the CRISPR mutant condition, analysis was performed on 2 x wild-type siblings for control and 2 x homozygous mutant embryos whose DNA mutation was confirmed by sequencing. 3 consecutive aortic segments per embryo were analyzed to cover the whole length of the dorsal aorta, in the trunk region (covering a distance of 990µm per embryo).

Interfering with the function of ArhGEF11/PDZ-RhoGEF suggests an activity at the interface between endothelial and hemogenic cells that relies on restricting the mobility of junctional pools at tri-cellular junctions, with a prominent role of the +exon 38 variant during endothelial cell intercalation

A, 2D-cartography of an aortic segment of a Tg(kdrl:eGFP-Jam3b; kdrl:nls-mkate2) embryo injected at the one cell stage with the ArhGEF11 exon 38 splicing morpholino. All the arrows pointing at reinforced junctional contacts between cells were bleached and imaged for FRAP analysis (the bleached areas correspond to the spots of highest intensities in the regions of interest, as visualized at the fluorescent confocal microscope, see Materials and Methods). Black and Magenta arrows point at he-he-ec and he-ec-ec tri-cellular junctions, respectively (he: hemogenic cell; ec: endothelial cell). Green arrows point at he-ec bi-cellular junctions. ec: endothelial cell, he: hemogenic cell. Bar = 20µm. B, C, FRAP analysis of bleached eGFP-Jam3b localized in regions of interest in controls (grey) and ArhGEF11 exon 38 splicing morpholino injected embryos (blue) or homozygous ArhGEF11CRISPR-Cterdel+/+ mutants (green). Statistical tests: two sided unpaired two samples Wilcoxon test. Bb, Cc, evolution, after photobleaching (at t=0s), of the mean fluorescence intensity per condition at HE-EC, HE-EC-EC and HE-HE-EC bi- and tri-junctions over time (10 min). Bb’, Cc’, median values for maximum amplitude of recovery (maximum of simple exponential fitted curves, see Materials and Methods). Bb’’, Cc’’, early evolution, after photobleaching (at t=0s), of the mean fluorescence intensity per condition over time (for the first 30 seconds). The fitted lines correspond to the linear regression of the mean fluorescence intensities. Bb’’’, Cc’’’, median fluorescence recovery slopes (linear regression). D, Model (2D deployment of the aortic wall) representing the endothelial/hemogenic dynamic interplay and the proposed function of ArhGEF11 and of its +exon 38 peptide encoding variant at junctional and membrane interfaces. This interplay involves 2 essential dynamic events requiring junctional remodeling: 1 (left cartoon, magenta arrows), the mobility of he-ec-ec tri-junctional contacts accompanying the movement of endothelial cells (ex: for ec1 and ec2) along lateral sides of hemogenic cells which is required to decrease the number of adjoining cells (see Lancino et al. 2018). This takes place contemporarily to - or in alternance with -, the contraction of HE and EHT cells as they are progressing throughout the emergence and reducing contacting membrane surfaces along the longitudinal axis (the reduction in contacting membrane surfaces also involves membrane retrieval, hypothetically via endocytosis). Our data obtained with the CRISPR deletion mutants (a slight tendency to increase, on average, the turnover and mobile pool of these he-ec-ec junctions) suggest that ArhGEF11, in the wild type condition, should be slowing down the recycling of junctional components at tri-junctions, which hypothetically should contribute to increase adhesion strength. This may be required also to stabilize the junction-cytoskeleton interface involved in controlling the contraction/shrinkage of HE cells along the longitudinal axis, a hypothesis that is compatible with the mutant phenotype observed in this study, i.e the increase in frequency of more elongated HE cells and the decrease in HE cell progression throughout EHT (see Figure 7); 2 (right cartoon, cyan arrow and bottom cartoons a-c), the intercalation of an endothelial cell to isolate 2 adjacent hemogenic cells or 2 daughter cells after mitosis (not depicted). This is mandatory for EHT progression and completion which requires adjoining endothelial cells to protrude membrane extensions that will anastomose to seal the aortic floor (see Lancino et al. 2018). The accumulation of adjoining cells of rather small length and apparently impaired in EHT progression that we describe in Figure 7 upon MO interference (that may also indicate impairment in abscission completion) suggests that the ArhGEF11 +exon 38 peptide encoding variant is more specifically involved in controlling the remodeling of the he/he interface that leads to endothelial cell intercalation (bottom cartoons, the remodeling of he-he-ec junctions is leading to he-ec-ec junctions and takes place between b and c). The increase in the recycling parameters that we measure in this interfering condition (mobile pool and early speed of recovery) indicates that the activity of ArhGEF11, and in particular of its +exon 38 peptide encoding variant, negatively controls junctional recycling. Hypothetically, the junctional adhesion strengthening triggered by reducing the recycling of junctional components at the he/he interface may be required during the early phase and progression of intercalation to support increase in membrane tension and environmental constraints (intercalation contributes to reducing the length/surface of the he/he membrane interface preceding junctional remodeling (cartoons b and c); this reducing membrane interface is in addition submitted to the shear stress imposed by blood flow).

A, 2 EHT pol+ cells visualized using a Tg(Kdrl:Gal4;UAS:RFP;4xNR:eGFP-Podxl2) embryo and time-lapse sequence (initiated at 55 hpf) obtained with spinning disk confocal microscopy. The images are extracted from the same time lapse sequence than the one used for Figure 1C. Green channel (eGFP-Podxl2) only is shown. Green arrows point at the evolution of the connection between the aortic/eht cell lumens at t = 0, 10, 30 min. Asterisks label the lumen delimited by the luminal/apical membranes. Bar = 8µm. B, Model summarizing and interpreting the temporal evolution of the luminal/apical membrane (in green, the asterisks mark the lumen of the apparent vacuole-like intracellular membrane structures) after the release of an EHT pol+ cell from the aortic floor. Step 1, the pseudo-vacuole filled with fluid and delimited by eGFP-Podxl2 as visualized at t = 65 min in A (and also at 45 min in Figure 1C, 2 asterisks) is consumed partly via budding (after sorting of eGFP-Podxl2); these budding profiles can be seen on the left image (pink arrowheads) corresponding, for eht cell 1, to the plane 16 of the time point t = 80 min of the time lapse sequence (see the Z-stack in Figure 1 – video 2). Step 2, after sorting and budding, the cell remains with pseudo-endocytic Podxl2 containing membranes and the remaining vacuolar structures filled up with fluid regress, putatively by chasing water as illustrated in step 3. Step 4, the eht cell remains with pseudo-endocytic Podxl2 containing membranes that label newly born precursors of HSPCs. Note that since the cell remains in contact with the aortic floor while the pseudo-vacuole is regressing, the vacuole-like intracellular membrane proximal to aortic cells may never undergo fission (see Figure 1A, steps 4-4’) but gets consumed via budding and flattening upon water chase (steps 1-3 that are similar to Figure 1A, steps 3-3’).

HE cells are not polarized at 30 hpf

Tg(Kdrl:Gal4;UAS:RFP;4xNR:eGFP-Podxl2) 30 hpf embryo imaged using spinning disk confocal microscopy. Top 2 panels: green (eGFP-Podxl2) and red (soluble RFP, in magenta) channels. White arrows point at 2 individual EHT cells. Note that HE cell 1 protrudes long filopodia, some of which inside the aortic lumen (live imaging shows that they are moving along the blood flow, data not shown). Bottom 2 panels: 1.625X magnification of EHT cell 1 and 2 in top panels. Green channel (eGFP-Podxl2) only. Green arrows point at very large intracellular vesicles, the largest reaching approximately 30µm. Asterisks mark the cytoplasm. Bar: 100µm.

Evolution of non-polarized HE cells throughout emergence

Tg(Kdrl:Gal4;UAS:RFP;4xNR:eGFP-Podxl2) embryo imaged using spinning disk confocal microscopy. Images were obtained from discontinuous time-lapse sequences covering a period of 13 hours (from 35 – to – 48 hpf). Successive phases of the evolution of HE cells are visible, from a non-polarized status (with the accumulation of Podxl2-containing intra-cytosolic vesicles as well as cell division) to post-emergence EHT cells remaining beneath the aortic floor. The top panel is a z-projection of 69 consecutive z-sections interspaced by 0.3µm; green (eGFP-Podxl2) and red (soluble RFP, in magenta) channels are shown; 4 individual HE cells (1 – 4, green arrows)) are marked, with cells 1 and 4 more advanced in the process of emergence. All the other panels are z-sections focused on the aortic floor, allowing visualizing the progression of the EHT, in particular at 35 hpf (t = 0), a timing point at which the separation between the luminal and basal membranes are equally labelled with eGFP-Podxl2, attesting for the absence of apicobasal polarity (see cells 1 and 4, see also cell 1 at t = 2.5 hours (hrs) and cells 3’ and 3’’ at t = 5 hrs). Note, at t = 2.5 and 5.0 hrs, the presence of intracytoplasmic eGFP-Podxl2 labelled vesicles inside HE cells, suggesting vesicular transport (white arrowheads, to be compared with images Figure 2 – figure supplement 1, at 30 hpf). At 48 hpf (bottom panel), HE cells have emerged and remain, for some of them, in close contact with the aortic floor (after having performed mitosis (notably for HE cell 2 (the division is visualized at 5hrs) and for HE cell 3, each one giving rise to cells 2’ - 2’’ and 3’ - 3’’ respectively, all containing residual eGFP-Podxl2 containing membranes as EHT signature). Bar = 80µm.

Phenotypic analysis of dt-runx1 expressing mutants: expansion of the thymus

A, maximum z-projections of confocal spinning disk sections of the thymus region of 5 dpf Tg(kdrl:gal4;UAS:RFP) (left, control) and incrossed Tg(kdrl:Gal4;UAS:RFP;4xNR:dt-runx1-eGFP) (right, dt-runx1) larva. The area for each thymus is delimitated (white lines); note the systematic expansion of thymic cells on the right sides of each image for dt-runx1 mutants in comparison to controls. Bar: 20µm. B, Area of projected thymus for the 2 conditions illustrated in A. Statistical tests: two sided unpaired two samples Wilcoxon test, with p-values of significant differences only.

Phenotypic analysis of dt-runx1 expressing mutants: evidence for apicobasal polarity of hemogenic cells

30-32 hpf embryos obtained from outcrossed Tg(Kdrl:Gal4;UAS:RFP;4xNR:mKate2-Podxl2) X Tg(kdrl:Gal4;UAS:RFP;4xNR:dt-runx1-eGFP) fishes and imaged in the trunk (AGM) region using spinning disk confocal microscopy. Images are depicting typical HE cells (in boxes) from hemogenic regions, with apical and basal membranes clearly separated from each other owing to reduction of their surface area (in comparison to flat aortic cells) and elongation in the antero-posterior axis. Luminal membranes are enriched with (magenta arrows) or more or less devoid of (green arrows) mkate2-Podxl2. Green: cytosolic eGFP released from the cleavage of dt-runx1-eGFP. Note that because of mosaicism, he cell 2 does not express mkate2-Podxl2. Bars:100µm.

Phenotypic analysis of dt-runx1 expressing mutants: slowing down of the emergence process and accumulation of EHT cells

Tg(kdrl:Gal4;UAS:RFP;4xNR:dt-runx1-eGFP) embryos imaged using spinning disk confocal microscopy and analyzed in the AGM/trunk region. A, z-projections of the dorsal aorta obtained from 52-55 hpf embryos. Top panel: fluorescence from the red channel is shown for the Tg(Kdrl:Gal4;UAS:RFP) control. Bottom panels: fluorescence in the green channel only is shown for the mutants (eGFP, released form the dt-runx1-eGFP cleavage). The black asterisks point at emerged cells that are in close contact with the aortic floor. Green arrows: EHT pol+ cells on aortic floor; magenta arrows: EHT pol+ cells in the lateral aortic wall; white arrows: uncharacterized emerging cells; he: hemogenic cells. B, evolution of an EHT cell extracted from a 7 hrs time-lapse sequence (0 to 420 min) and showing significant changes in its morphology throughout emergence. Note the sub-luminal and cytosolic localization of pools of Podxl2 (notably at t = 30 min, magenta arrows) suggesting enhanced trafficking of the protein and relative instability of apical polarity features, consistently with apparent fluctuation of the luminal membrane surface contacting the aortic lumen (green arrows, in particular at t = 240 min). At t = 420 min, the cell has emerged and Podxl2 containing membranes remain in close contact with the membrane contacting the aortic floor. Bars:20µm.

Model depicting the evolution of junctional interfaces and of the differential mobility of antero-posterior junctional complexes for EHT pol+ and EHT pol-cell emergence types

Top panel: EHT pol+ cell whose emergence depends on the constriction of circumferential actomyosin (orange, see Lancino et al. 2018) and the reduction of the membrane interface contacting endothelial neighbors shrinking along the longitudinal axis (horizontal double arrows) and in the 2D plane (endothelial neighbors are not depicted and are embedded in the blue X-Y plane). Presumably, the reduction of membrane interfaces relies on consumption via endocytosis. Green ovals: junctional complexes reinforced at antero-posterior poles. Green line: apical/luminal membrane with typical inward bending. Bottom panel: EHT pol-cell whose emergence depends on the dynamics of adhesion pools that move synchronously in 3D (X, Y, Z), both at antero-posterior poles (blue ovals) and at lateral sides of the emerging cell contacting endothelial neighbors (adjoining endothelial cells are not depicted and are embedded in the blue X-Y plane). Presumably, this type of emergence in which endothelial cells crawl over the EHT cell (curved arrows) involves, for the latter, a partial retrograde endocytic recycling of junctional complexes (opposite to the direction of emergence). Note that nuclei are not drawn at scale.

JAM2a and JAM3b expression and localization in diverse embryonic tissues

Localization of transiently expressed eGFP-Jam2a (A-D) and eGFP-Jam3b (E-G) in 52 hpf embryos, using spinning disk confocal microscopy. Plasmid constructs were expressed under the control of the heat shock Hsp70 promoter. For both constructs, expression was induced approximately 6 hrs before imaging, by 1hour balneation in 39°C embryo medium. All images are maximum z-projections. A, E, ependymal cells. White and red arrows point at reinforcement of eGFP-Jam2a and eGFP-Jam3b at apical intercellular junctions and at basolateral membranes, respectively. B, pronephric tubule cells. White arrows point at the reinforcement of the eGFP-Jam2a signal at apical sides of membranes of polarized cells constituting pronephric tubules. The red arrow points at baso-lateral localization of eGFP-Jam2a. C, F, skin epithelial cells. White arrows point at the localization of eGFP-Jam2a and eGFP-Jam3b at lateral junctional interfaces between two neighboring cells. Red arrows point at tri-cellular junctions at which the density of eGFP-Jam2a and eGFP-Jam3b is reinforced. D, skin epithelial cells. White arrows point at membrane protrusions. G, Striated muscle cells. Red arrows point at T-tubules (invagination of the sarcolemmal membranes); white arrows point at the plasma membranes of myofibrils. Bars: 20µm.

Examples of junctional contacts targeted by FRAP in the aortic landscape

After performing z-stack acquisitions in trunk regions followed by 2D-deployment of aortic segments (A-C), 4 different Tg(kdrl:eGFP-Jam3b; kdrl:nls-mKate2) 48-55 hpf embryos were illuminated for FRAP. Panels (A) and (B) are the entire segments from which cropped images of EHT pol+ and EHT pol-cells (black asterisks) and surroundings were extracted to build the panels A and B of Figure 6. Panels (C) and (D) are from 2 other embryos. The 2D-cartographies allow visualizing the junctions that were selected for FRAP, with the black arrows pointing at junctional interfaces between EHT and endothelial cells (all are in the area of tri-junctions) and the green arrows pointing at bi-junctions (Bj) or tri-junctions between endothelial cells (green asterisks). Bar = 20µm.

Searching for PDZ-domain containing RhoGEFs potentially involved in the EHT

A, cartoons representing the domains composing the 9 PDZ domain-containing RhoGEFs that were investigated in this study (all these RhoGEFs are encoded by different genes in the zebrafish; protein respective length is not drawn at scale). B, results obtained from 3 independent qRT-PCR experiments and performed on material extracted from trunk regions of 35 and 48 hpf embryos obtained from incrossed Tg(kdrl:Gal4;UAS:RFP;4xNR:dt-runx1-eGFP) adult fishes. Statistical tests: two sided unpaired two samples Wilcoxon test, with p-values of significant differences. GEF: guanine nucleotide-exchange factor; PDZ: postsynaptic density protein of 95 kDa, Discs large and Zona occludens-1; PH: Plekstrin homology; RGS: regulator of G-protein signaling; DH: Dbl (diffuse B-cell lymphoma) homology; RBD: Ras-binding domain; DEP: Dishevelled Egl-10 and Plekstrin; InsPx4-PTPase: PtdIns(1,3,4,5)P4 phosphatidylinositol phosphatase; TIAM: T-cell-lymphoma invasion and metastasis; LARG: leukemia-associated Rho guanine-nucleotide exchange factor; PRex: PtdIns(3,4,5)P3-dependant Rac exchanger-1 and 2.

Whole mount in situ hybridizations (WISH) performed on 30-32 hpf and 48-50 hpf embryos, with probes specific for all 9 PDZ-domain containing RhoGEFs that were investigated in this study. Note that all RhoGEFs are detected in the dorsal aorta of the trunk region (white arrowheads).

A N-terminal fragment of ArhGEF11/PDZ-RhoGEF localizes at junctional membranes with enrichment at antero-posterior sites of EHT cells

A, domains of ArhGEF11, including its actin binding site (L/IIxxFE) and the position of exon 38 encoded peptide. The bottom cartoon represents a protein deleted from its DH-PH domains and C-terminus that are replaced by eGFP (PDZ-PRD-RGS-eGFP) to follow the localization of the truncated fusion protein in expressing cells. B, localization of the ArhGEF11 PDZ-PRD-RGS-eGFP fusion protein after transient expression in the aorta (expression was obtained after injection of the plasmid, at the one cell stage, in the Tg(Kdrl:Gal4;UAS:RFP) fish line). Confocal images extracted from a time-lapse sequence (timing points t = 0 and t = 77 min) and showing either the eGFP channel only (left), or the merge between the green and the red channels (RFP labels the cytoplasm). Note the localization of PDZ-PRD-RGS-eGFP at membrane interfaces at early time point of the emergence and at the rim of an emerging cell (white arrowheads, second cell from the right) and its enrichment at antero-posterior sites of an EHT pol+ cell proceeding throughout emergence (right cell, at t = 0 and t = 77 min and bottom cartoons, green arrows point at localization/concentration of PDZ-PRD-RGS-eGFP).

MO and CRISPR approaches to investigate the function of ArhGEF11/PDZ-RhoGEF in the EHT

A, splicing MO interfering with integration of exon 38. Top panel: cartoon representing exons/introns (not drawn at scale) composing the 3-prime region of the gene encoding for ArhGEF11, with the position of the splicing MO. Middle panel: agarose gel showing the 2 alternative mRNAs encoding for ArhGEF11 in control animals (left track, control) and after injection of the MO at the one cell stage (right track, +MO at 2 and 5ng). Bottom panel: ArhGEF11 DNA sequences obtained after RT-PCR, cloning and sequencing for 1 control and 2 +MO clones (+MO1, +MO3). B, morphologies of control and +MO injected 48 hpf embryos. Note the absence of malformations. C, top panel: ArhGEF11 wild type and CRISPR-mediated 7bp deletion in the 3-prime region of exon 38 leading to a frame-shift in the ORF and a downstream premature stop codon; middle panel: CRISPR del/C-ter nucleotide/aa sequence and wild type nucleotide/aa sequence covering the extreme C-ter of the full-length protein; bottom panel: CLUSTAL 2.1 multiple sequence alignment of mouse and zebrafish C-termini highlighting the sequences of spliced variants of potentially similar activity in the regulation of ArhGEF11 activity on RhoA. Accession numbers: NP_001003912.1 and NP_001027010.1 for the mouse and Danio rerio sequence, respectively. D, morphologies of control and homozygous ArhGEF11CRISPR-Cterdel+/+ 48 hpf embryos ((Kdrl:eGFP-Jam3b; kdrl:ArhGEF11CRISPR-Cterdel+/+) embryos). Note the edema in the cardiac region (35 hpf embryos also exhibited retardation in blood circulation).

Supplementary data on the ArhGEF11/PDZ-RhoGEF exon 38 splicing morpholino phenotype

Tg(kdrl:eGFP-Jam2a; kdrl:nls-mKate2) control (A, C, E) or MO-injected (B, D, F) embryos were imaged using spinning disk confocal microscopy (48-55 hpf time-window). Aorta segments (330 µm each) were imaged in the trunk region (AGM). The 2D-cartographies with delimited cellular contours in panels A and B are presented Figure 7A with the corresponding un-modified 2D-cartographies shown on top for unmasking of contours. Panels A and C illustrate 2 different segments from the same embryo. Panels B, D, F illustrate segments from 3 different embryos. (a-f) Maximum z-projections of merged nls-mkate2 and eGFP-Jam2a signals for control (a, c, e) and for ArhGEF11 exon 38 splicing morpholino (b, d, f) conditions. For panels (a) and (b), maximum z-projections of the eGFP-Jam2a signal only are also shown. (a’-f’) Single z-plane images of merged nls-mkate2 and eGFP-Jam2a signals for control (a’, c’, e’) and for ArhGEF11 exon 38 splicing morpholino (b’, d’, f’) conditions. For e’, the image is a composition of 2 different z-planes from the same field (the boundaries are marked with white ticks). In right margins, magenta and green arrowheads designate the aortic floor and roof, respectively. (a’’-f’’) 2D-cartographies (bottom, with delineated cellular contours) obtained from eGFP-Jam2a signals for control (a’’, c’’, e’’) and for the ArhGEF11 exon 38 splicing morpholino (b’’, d’’, f’’) conditions, respectively. Cell contours are delineated either in blue (endothelial cells), yellow (hemogenic cells), red (morphologically characterized EHT cells, red arrows), and small cells delineated by cyan boxes (morphologically uncharacterized EHT cells and putative post-mitotic cells remaining as pairs). Cellular contours have been semi-automatically segmented along the cellular interfaces labelled with eGFP-Jam3b (see Materials and Methods). White and black arrows designate hemogenic cells with their nucleus visible on the z-section. Analyses were performed on 2 x control non-injected embryos and 3 x embryos injected at the one-cell stage with the ArhGEF11 exon 38 splicing MO. Bars = 20µm.

Supplementary data on the ArhGEF11/PDZ-RhoGEF CRISPR C-ter deletion phenotype

(Kdrl:eGFP-Jam3b; kdrl:ArhGEF11CRISPR-Cterdel+/+) homozygous ArhGEF11 C-ter deletion mutants (CRISPR ArhGEF11: panels B, D, F) and control siblings (Control: panels A, C, E) were imaged using spinning disk confocal microscopy (48-55hpf time-window). Note that the genetic background of the CRISPR fish line is (Kdrl:GAL4; UAS:RFP) thus allowing the red cytosolic staining of aortic and hemogenic cells. Aorta segments (330 µm each) were imaged in the trunk region. The 2D-cartographies with delimited cellular contours in panels A and B are presented Figure 7B with the corresponding un-modified 2D-cartographies shown on top for unmasking of contours. Panels A, C and D, F illustrate 2 different segments from the same embryo. (a-f) Maximum z-projections of merged nls-mkate2 and eGFP-Jam3b signals for control (a, c, e) and for the CRISPR mutant (b, d, f) conditions. For panels (a) and (b), maximum z-projections of the eGFP-Jam2a signal only are also shown. (a’-f’) Single z-plane images of merged nls-mkate2 and eGFP-Jam3b signals for control (a’, c’, e’) and for ArhGEF11 exon 38 splicing morpholino (b’, d’, f’) conditions. For b’ and d’, the image is a composition of 2 different z-planes from the same field (the boundaries are marked with white ticks). In right margins, magenta and green arrowheads designate the aortic floor and roof, respectively. (a’’-f’’) 2D-cartographies (bottom, with delineated cellular contours) obtained from eGFP-Jam3b signals for control (a’’, c’’, e’’) and for the CRIPR mutant (b’’, d’’, f’’) conditions, respectively.

Cell contours are delineated either in blue (endothelial cells), yellow (hemogenic cells), red (morphologically characterized EHT cells, red arrows), and small cells delineated by cyan boxes (morphologically uncharacterized EHT cells). Cellular contours have been semi-automatically segmented along the cellular interfaces labelled with eGFP-Jam3b (see Materials and Methods). White and black arrows designate hemogenic cells with their nucleus visible on the z-section. Analyses were performed on 2 x wild-type siblings for controls and 2 x mutant embryos whose mutation was confirmed by DNA sequencing. Bars = 20µm.