Summary
Cranial sensory neurons develop in close proximity to blood vessels, however neurovascular interactions in the peripheral nervous system and their underlying signals are yet unknown. Here, we uncover two separate, novel roles for cranial vasculature in cranial sensory neurogenesis in zebrafish. The first involves precise spatiotemporal endothelial-neuroblast filopodial contacts and Dll4-Notch signaling to restrain neuroblast proliferation. Secondly, we find a role for signaling triggered by blood flow in promoting sensory neuron differentiation. Thus, we demonstrate that the cranial vasculature constitutes a hitherto unrecognized signaling component of the sensory ganglia niche and controls the pace of their growth and differentiation dynamics.
Introduction
In the adult brain, neural stem cells (NSC) reside in specialized microenvironments, referred as “niches” 1,2. There, NSCs remain relatively quiescent, but physiological or pathological circumstances can activate neurogenesis3. A key component in adult brain neurogenic niches is vasculature, which regulates several aspects of NSC behaviour4–7. In particular, two types of neurovascular interactions take place, one mediated by direct cell-cell contacts through NSC endfeet enwrapping endothelial cells (EC)7 and the other mediated by secreted signals from EC. Cell-cell contact signaling by EphrinB2 and Jagged1 expressed in ECs and Eph and Notch receptors in NSCs regulates NSC quiescence8. On the other hand, endothelia secrete VEGFs, Angiopoietins, BDNF, NT-3 and PEDF that control NSC self-renewal, progenitor cell proliferation or differentiation9–14. During embryonic central nervous system (CNS) development, vascular and neuronal development are also tightly coupled. In the developing mouse hindbrain and telencephalon, vasculature is required for radial glia and neural/glial progenitor cell proliferation15,16, while vasculature promotes NSC differentiation in the mouse cerebral cortex17. However, despite knowledge of neurovascular interactions in the CNS, there is little evidence for neurovascular crosstalk in the peripheral nervous system (PNS) during embryogenesis or adulthood.
The cranial PNS consists of several ganglia with sensory bipolar neurons that transmit external sensory information to the brain. Sensory neurons derive from neuronal precursors that are specified by neurogenin1 (neurog1) in cranial sensory placodes18. These neuronal precursors (neuroblasts) subsequently express Neurod, delaminate and coalesce into cranial ganglia where they undergo rounds of transit-amplification and differentiate into Islet2+ post-mitotic sensory neurons19–21 (Fig. 1a). Some of the autocrine signals regulating the balance between sensory neuroblast proliferation and differentiation have been described22–27. Since cranial blood vessels lie in close proximity to forming cranial ganglia it seems likely that vascular signals also influence their development28. However, this question has yet not been addressed.
In this study, we identify two novel roles for vasculature in cranial sensory neurogenesis. In vivo imaging of the developing sensory ganglia reveals that endothelial cells and neuroblasts establish direct contacts through filopodia. Genetic and pharmacological perturbations show that early filopodial contacts and Dll4 signaling inhibit neuroblast cell cycle entry. Later, blood flow initiation triggers cranial ganglia neuronal differentiation. Our results uncover, for the first time, that the cranial vasculature is an essential component of the sensory ganglia niche that regulates neuroblast proliferation and differentiation.
Results
Vasculature regulates the number of SAG neuroblasts through negative control of neuroblast cell cycle entry
We focused on neurovascular interactions in the statoacoustic ganglion (SAG), which forms the sensory neurons that innervate mechanosensory hair cells of the inner ear and the brainstem. The developing zebrafish SAG lies between three main vessels, the Primordial Hindbrain Channel (PHBC) that runs between the hindbrain and the otic vesicle (OV), the Lateral Dorsal Aorta (LDA) beneath the OV and the Primary Head Sinus (PHS) surrounding laterally the OV28 (Fig. 1b).
To functionally evaluate if cranial vasculature regulates the development of cranial sensory neurons in SAG, we analyzed SAG growth dynamics in cloche (clo) homozygous mutants in the double TgBAC(neurod:EGFP) nl1;(Kdrl:ras-mCherry) s896 background, in which eGFP marks sensory neurons and mCherry marks ECs. Cloche encodes npas4l, a PAS-domain-containing bHLH transcription factor required for the formation of endothelial and hematopoietic cells29,30 and clo mutants have extensively been used as an avascular model31,32. In clo mutant embryos, the SAG appeared larger compared to siblings (sib) at 36 and 72 hpf (Fig. 1c). Quantification of SAG volumes revealed a statistically significant increase in clo mutant embryos (Fig. 1d). Moreover, the SAG displayed an abnormal shape and a significant expansion along its mediolateral axis (Fig. 1c,e). The increased volume in clo mutant embryos could reflect a loss of confinement of the SAG cells by taking the space left by the lack of blood vessels or, alternatively, a higher number of Neurod+ cells. To distinguish between both possibilities, we quantified in serial transverse sections the nuclei within the GFP-labelled SAG cells between 30 to 72 hpf in the clo mutant and sibling embryos (Fig. 1f and Supplementary Fig. 1). In siblings, there is a period of steep SAG growth between 30 to 36 hpf, (sib 30 hpf: 89 ± 5; sib 36 hpf: 128 ± 5). From 36 to 72 hpf, the mean number of Neurod+ cells is 105 ± 8, suggesting that after 36 hpf this population does not grow. Remarkably, we found that the number of Neurod+ cells was significantly higher in clo mutant embryos compared to siblings at 36 hpf (sib: 128 ± 5 vs clo-/-: 177 ± 5) and this increase persisted at all the following timepoints (Fig. 1f). Several mechanisms could account for the increased number of Neurod+ cells in clo mutant embryos: i) increased specification of neurog1+ cells within the otic epithelium and thereby increased delamination of Neurod+ cells, ii) reduced differentiation of Neurod+ cells into Islet2+ neurons or iii) increased proliferation of the transit-amplifying Neurod+ population. We did not find a higher number of neurog1+-expressing cells in the otic epithelium in clo mutant embryos compared to siblings (Fig. 1g,g’) nor significant differences in the number of differentiated Islet2+ cells in the SAG (Fig. 1h, h ’). Instead, in clo mutant embryos we detected a significantly higher number of proliferating Neurod+ cells after a BrdU pulse from 24 to 30 hpf (expressed as BrdU+/Neurod+ ratio) (Fig. 1i,i’).
Thus, our results show that at early timepoints, cranial vasculature regulates the number of SAG neuroblasts by negatively controlling cell proliferation of Neurod+ cells and therefore constitutes an essential component of the neurosensory niche during development.
Cranial blood vessels and sensory neuroblasts establish direct contacts through filopodia
In both the embryonic and adult brain, NSCs establish contacts with ECs through endfeet projections7,8,15. Therefore, we wondered if neuroblasts from the SAG also establish direct contacts with the adjacent vessels. To investigate this in detail, we live-imaged sensory neuroblasts from the SAG and ECs from the PHBC in double TgBAC(neurod:EGFP) nl1;(Kdrl:ras-mCherry) s896 reporter lines, at high spatiotemporal resolution. At 30 hpf, we did not observe neuronal extensions enwrapping ECs of the PHBC, but instead observed dynamic, thin filopodia extending from ECs to Neurod+ cells, as well as similar filopodia from Neurod+ cells extending to ECs (Fig. 2a,b, Supplementary Fig. 2, Supplementary Fig. 3 and Supplementary Video 1-4). The dynamics of filopodia were varied; in some instances, we observed filopodia extending and retracting in seconds (Fig. 2a,b cyan and magenta arrowheads respectively), in others, filopodia remained for several minutes (Fig. 2a,b, yellow arrowheads). Interestingly, some PHBC filopodia did not remain at the surface of the SAG but contacted deeper neuroblast somas (Supplementary Video 4). Moreover, stable filopodia-filopodia contacts between neuroblasts and ECs were also seen in some instances (Fig. 2c, Supplementary Fig. 3). As observed in transverse optical confocal reconstructions of the vessels and the SAG, filopodia were highly directional: PHBC filopodia extended only ventrally in the direction of sensory Neurod+ cells (Fig. 2d,e). Interestingly, filopodia were present from 24 to 36 hpf but not later, suggesting an early role of filopodia in vessel-neuroblast communication (Fig. 2f). Altogether, highly directed and temporally regulated filopodia-mediated cell contacts are established between both EC and sensory neuroblasts.
Sensory neurons are not required for cranial vascular patterning
Communication between neurons and vessels in the CNS can take place from vessels to neurons, vice-versa and bidirectionally33–35. Thus, we next investigated if sensory neuron development has a role in vascular patterning near sensory ganglia. In this case, we analyzed homozygous neurog1hi1059 mutant embryos – which fail to specify sensory neuronal precursors18-in the double transgenic TgBAC(neurod:EGFP) nl1;(Kdrl:ras-mCherry) s896 reporter line. Strikingly, despite the total absence of SAG and other cranial ganglia in neurog1 mutant embryos, development of the PHBC and LDA displayed no obvious patterning defects at 30 hpf (Fig. 3a). The central arteries (CtAs) ingrowth into the hindbrain also did not display any delay nor obvious alteration at 48 hpf (Fig. 3a). We quantified the width of PHBC and the number of ECs at 48 hpf, and again found no significant differences between neurog1 mutants and siblings (Fig. 3b,c). In conclusion, our data indicate that cranial vascular development does not require signals from sensory neuroblasts and suggest that primarily the signaling direction goes from vessels to neurons.
Since imaging revealed directed filopodial contacts from cranial vessels to sensory neurons and vice-versa (Fig. 2a), we wondered about possible growth interdependency. We imaged neuronal filopodia dynamics in the absence of vessels and, endothelial filopodia in the absence of neurons. In clo mutant embryos, the number of neuronal filopodia was drastically reduced (Fig. 3d,d’). In contrast, in neurog1 mutant embryos the number of EC filopodia did not change, though they were shorter (Fig. 3e-e’’). These data indicate that neuronal filopodia require signals from ECs.
While sensory neurons in SAG do not appear to be required for vascular development, our data indicate that vasculature is required for SAG neurons to proliferate during a temporal window that coincides with the period of filopodial contacts. Moreover, neuroblasts only extend filopodia when ECs are present. Altogether, the results support the hypothesis that endothelial-neuronal filopodia contacts are required for vessels to signal to neuroblasts in the SAG to restrict growth.
Cell signaling mediated by filopodia regulates the number of SAG neuroblasts
Cell signaling through filopodia has emerged as a novel mechanism of signaling between adjacent cells in different embryonic tissues36–38. In order to assess the possible role of filopodial signaling in SAG growth, filopodia formation was inhibited with LatrunculinB (LatB), a toxin that binds to actin and has been used to impair filopodia function in vivo39,40. Imaging and quantification of filopodia in embryos treated with LatB for 6 hours indicated that the treatment was effective, the number of filopodia in ECs and in neuroblasts was reduced (Fig. 4a,b). Interestingly, inhibition of filopodia caused an increased number of Neurod+ cells in LatB-treated embryos and this phenotype was also due to a higher number of Neurod+ cells incorporating BrdU (Fig. 4c-e). No additional effects were found in treated clo mutant embryos (Fig. 4c-e). Hence, filopodia formation inhibition at early stages recapitulated the SAG growth phenotype found previously in the avascular mutant. These data support the hypothesis that the endothelial-neuronal filopodial contacts exert a signaling role onto SAG neuroblast proliferation.
Dll4 inhibition increases the Neurod+ population
To identify signals that could mediate the neurovascular crosstalk we revisited the published vascular signals regulating CNS neurogenesis for evidence of ligand expression in cranial vasculature or receptors in cranial sensory neurons41. In the Ventricular-Subventricular Zone (V-SVZ) of the adult brain, endothelial Jagged1 ligand restrains NSC proliferation8. Cranial vessels do not express Jagged1 but do express Delta-like 4 (Dll4), another Notch ligand, which is essential for proper vascular remodeling and sprouting42–44. We therefore hypothesized that Dll4 could be signaling to SAG neurons, which express Notch1 at the same early timepoints 45,46. Moreover, it has been postulated that pigmentation patterning in zebrafish is regulated by DeltaC-Notch signaling through filopodia47.
To test this hypothesis, we inhibited dll4 through the well-studied dll4 morpholino (MO)43,44 and assessed its effects on SAG growth. Remarkably, dll4 morphant embryos presented an expansion of the Neurod+ population similar to the one observed previously in clo mutant embryos (Fig. 5a). This expansion was reflected both in SAG volume (Fig. 5b) as well as in Neurod+ cell number (Fig. 5c). As a further control, dll4 morpholino was also injected in clo mutant embryos and the number of Neurod+ cells did not change, suggesting that the results are due to loss of dll4 expressed by the vasculature (Fig. 5a-c). These findings reveal that Dll4 is required to regulate Neurod+ cell number.
In angiogenic sprouts, the number of filopodia increases after dll4 downregulation48. When we assessed the number of filopodia in the developed PHBC, we detected a small increase in filopodial number and no differences in their length (Fig. 5d-f). Thus, the increase of Neurod+ cells in dll4 morphants cannot be attributed to a lack of filopodial processes but to a specific lack of signal. Finally, the number of ECs did not change in dll4 morphant embryos compared to embryos injected with a random morpholino (Fig. 5g,h). Therefore, the increase in Neurod+ cell number after dll4 down-regulation is not due to vascular developmental changes but to Dll4-Notch1 signaling.
Together, these results suggest that Dll4 signaling constitutes one of the key molecular mechanisms by which the vasculature regulates cranial sensory neurogenesis.
Absence of vasculature impairs sensory neuronal differentiation
Since more neuroblasts form in the SAG in the absence of vasculature, we next assessed if these neuroblasts differentiate, resulting in a higher number of Islet2+ post-mitotic cells at later stages. Unexpectedly we found, after plotting the number of immunostained Islet2+ nuclei at different timepoints, that there was a significantly lower number of Islet2+ cells in clo mutants from 54 hpf onwards (Fig. 6a,b). The differentiation curves indicate that, while at early timepoints sensory differentiation occurs partially independently of vasculature, from 54 hpf onwards, cranial vasculature is required to promote differentiation of Neurod+ cells to Islet2+ cells. As expected from the immunostaining data, the fluorescence from sensory axons was also reduced in clo mutant embryos, labelled by the Tg(isl2:GFP) zc7 reporter line which marks in vivo the cytoplasm of differentiated sensory neurons49 (Fig. 6c).
To determine if cranial vasculature regulates neurogenesis in other sensory ganglia besides the SAG, we assessed for filopodia contacts and quantified the number of Neurod+ and Islet2+ cells in different ganglia. Here, we did not perform a temporal analysis, but we chose discrete timepoints. At 30 hpf, we could also detect filopodia from the PHBC to the trigeminal and anterior lateral line ganglia (Supplementary Fig. 4c). At 36 hpf, we found a significantly increased number of Neurod+ cells in the trigeminal ganglion of clo mutants (Supplementary Fig. 4a,b), whereas at 72 hpf there was a reduced number of Islet2+ cells in the trigeminal and vagal ganglia (Supplementary Fig. 3d-g). Therefore, the neurovascular interaction appears to extend to other sensory ganglia.
To complete the analysis, we assessed if vasculature could regulate the development of other cell types such as the hair cells of the inner ear. For this aim, we generated clo homozygous mutants in the Tg(brn3c:mGFP) s356t background that labels the membranes of hair cells and their kinocilia50. Serial confocal stacks taken at 72 hpf and quantification of hair cells did not reveal significant differences in hair cell number in cristae nor maculae of clo mutants compared to siblings (Fig. 6d-f). Therefore, vasculature is specifically affecting neuronal differentiation of the inner ear but not hair cell number.
Initiation of blood flow triggers sensory differentiation
It has been reported that in zebrafish blood flow begins due to the completion of anastomosis of cranial blood vessels around 48 to 60 hpf51. The coincidence of this timing with the one of sensory differentiation defects in clo mutant embryos prompted us to explore whether the inducing signal of sensory differentiation was linked to blood flow. To this aim, we incubated embryos with Nifedipine, a drug that inhibits heartbeat and blood flow52. As expected from blood flow inhibition, we observed a drastic vessel diameter reduction in Nifedipine-treated embryos (Fig. 7a,b). In embryos treated with Nifedipine from 54 to 72 hpf, the number of Islet2+ cells decreased, recapitulating the defects found in clo mutant embryos (Fig. 7c-e). Treatment in the avascular model did not further enhance the phenotype, suggesting that the differentiation arrest is mostly attributed to the absence of blood flow in clo mutant embryos and not associated to side effects of Nifepidine (Fig. 7e). The same results were obtained when treatment was done with BDM, another drug used to inhibit blood flow53 (Supplementary Fig. 5). To narrow down the time window of blood flow requirement, we incubated embryos with Nifedipine from 60 to 72 hpf and from 48 to 60 hpf. Incubation from 60 to 72 hpf did not inhibit differentiation (Fig. 7f,g), whereas incubation from 48 to 60 hpf did (Fig. 7h,i). These results indicate that blood flow disruption of sensory differentiation takes place in a narrow time window and once signaling is initiated, blood flow is no longer required. Altogether, the results define for the first time a prominent role of cranial vasculature in sensory neurogenesis that can be segregated into two sequential and independent mechanisms. A first one, in which Dll4-Notch1 filopodia signaling from ECs to neurons regulates neuroblast transit-amplification, and a second one in which initiation of blood flow in vessels induces sensory neuron differentiation (Fig. 8).
Discussion
The SAG and other cranial sensory ganglia reside in a vascularized milieu in adults and during embryogenesis. A fundamental question in developmental neurobiology is whether or not vasculature, in addition to transport of nutrients and oxygen, provides instructive cues for sensory neurogenesis. This is relevant to understand hearing loss and other sensory neuropathies linked to vascular defects54–56. Our detailed temporal analysis reveals that cranial ganglia indeed require vascular signals for proper growth and differentiation. We identified two segregated mechanisms: an early filopodial-based, cell-cell contact dependent mechanism regulating neuroblast proliferation and a later blood flow-dependent mechanism controlling neuronal differentiation.
In the V-SVZ, quiescent NSCs physically contact blood vessels with specialized endfeet7. By taking advantage of the imaging and cell reporter tools available in the zebrafish model we resolve, for the first time, that cell-cell communication takes place through dynamic filopodial extensions that project from ECs to neuroblasts. It is still possible that in other contexts, dynamic contacts also exist, but imaging in fixed tissue might have hampered their identification, as suggested by Obernier and colleagues57. Filopodia in angiogenic tip cells and migrating ECs have been described40, 58–60. To our knowledge, filopodia from endothelial stalk cells contacting neuronal cells have not been reported before. One could envision cell-cell contact with only ECs extending filopodia to neuronal somas. Interestingly, neuroblasts also emit filopodia, probably to increase the chances of cell-cell communication. Moreover, the study highlights the requirement of EC signals for neuronal filopodia growth.
Signaling roles for filopodia in Drosophila and vertebrates have been demonstrated, extending the cell-cell signaling distance of influence36–38. This is relevant in the facial mesenchyme where vasculature does not grow inside the sensory ganglia but adjacent. By blocking filopodia projections, we demonstrate that they have a signaling function in the negative control of neuroblast cell proliferation. Most likely only a subset of neuroblasts are contacted by endothelial filopodia, allowing a scattered pattern of cell cycle-arrested neuroblasts between others with proliferative capacities. In contrast to other contexts in which neurons control vascular patterning61, our results suggest that in the context of sensory ganglia and surrounding vessels, an absence of cranial ganglia does not have a major effect on early vasculogenesis.
During adult neurogenesis, cell contact-dependent neurovascular signaling is mediated by ephrinB2 and Jagged1 present in ECs that suppress cell-cycle entry in NSCs8. We propose that in the PNS Dll4, but not Jagged1, plays a similar role. Maintenance of a quiescent and undifferentiated population of neural progenitors is key for life-long regeneration. Thus, the role we have uncovered for cranial vasculature and Dll4 in the cranial sensory niche might be relevant in regenerative studies upon neuronal damage.
Several studies have highlighted the role of blood flow initiation in remodeling the vasculature, VEGF expression, angiogenesis and stimulation of hematopoietic stem cell formation44,62,63. It has been suggested that blood flow can exert tangential and perpendicular forces on ECs leading to actomyosin rearrangements, YAP-TAZ nuclear translocation and the induction of the mechano-sensitive transcription factor Klf2a63–65. Ultimately, blood flow triggers gene transcriptional changes in ECs and vessel maturation66. It is tempting to speculate that in our system, blood flow also triggers mechanical changes in ECs leading to the release of neuronal differentiation signals. Although we believe less plausible, expansion of the vasculature could also directly induce sensory differentiation by compression of the neuronal environment. Further studies should explore more precisely the changes taking place in cranial vessels upon blood flow initiation. Our current work identifies novel roles for head vasculature and shall contribute to our basic understanding of the signalling mechanisms regulating sensory neurogenesis in which head vasculature enters in scene.
Methods
Zebrafish Strains and Genotyping
Zebrafish (Danio rerio) were maintained as previously described67 at the aquatic facility of the Parc de Recerca Biomèdica de Barcelona (PRBB). The following transgenic lines in different combinations were used: Tg(neurog1) hi1059 68 mutant to inhibit neurogenesis and Tg(cloche) m378 69 as an avascular model. Five different reporter lines were used: Tg(Kdrl:ras-mCherry) s896 70 and Tg(flk1:EGFP) s843 71 for vessels, TgBAC(neurod:EGFP) nl1 72 for neuroblasts, Tg(isl2:gfp) zc7 (also called Isl3)49 for sensory differentiated neurons and Tg(brn3c:mGFP) s356t 50 for hair cells.
Embryos were staged as previously described 73. Adult heterozygous for neurog1 hi1059 allele were identified by PCR genotyping by fin-clip genomic DNA68. Homozygous neurog1 mutant embryos were recognized by the significant reduction of GFP signal in Tg(neurog1) hi1059, BAC(neurod:EGFP) nl1, (Kdrl:ras-mCherry) s896 background. Avascular mutants were identified by the lack of vasculature in the reporter line for endothelial cells Tg(Kdrl:ras-mCherry) s896 or Tg(flk-1:EGFP) s843, lack of blood cells and circulation (visible under scope from 30-36hpf) and atrium enlargement, as described by Stainer et al. 29.
In both cases, wild-type and heterozygous-deficient siblings were undistinguishable and both have been used and referred as siblings (sib). Finally, since clo mutant embryos die around 7dpf 69, we studied embryos until 4dpf to exclude vessel directed effects from embryonic morbidity.
Live imaging, confocal microscopy and quantifications
Images of cryosections were acquired with a Leica DFC 7000 T microscope, and 40x oil immersion (0.75 NA).
Whole-mount embryos imaging was performed on a Leica SP8 confocal microscope, with 20x glycerol immersion lens (0.7 NA). Embryos were mounted in 1% low-melting agarose (Ecogene) in PBS or Danieau’s solution, for fixed and life embryos respectively, and placed onto a glass-bottom dish. For life embryos, 0.1% Tricaine (Sigma) was also added. The 488- and 561-nm laser lines were employed at a scan speed of 600 Hz.
For time-lapse imaging of filopodial, the resonant mode (8000 Hz) was used to achieve a high spatiotemporal resolution. Frames were captured simultaneously with 488 nm and 561 nm lasers at a time interval of 15 seconds. Z-size of 2 μm was used for videos and 0.5 μm for static stacks and transverse optical reconstructions. Confocal stacks and movies were flattened by maximum projections using FIJI (Image J).
Measurements were done manually with FIJI (Image J) for SAG and PHBC filopodia number, length and orientation using one single static image, from a defined region of interest (ROI). PHBC width was measured from maximum projection confocal images always at the same point, between CtA 2 and 3. PHBC EC number was calculated by perinuclear green cytosolic signalling from Tg(flk-1:EGFP) s843 and EC contours from Tg(Kdrl:ras-mCherry) s896 in a defined ROI. When ROIs were used, the same area was used to compare samples. SAG anteroposterior (AP) and dorsoventral (DV) axes length were measured from maximum projection confocal lateral images. The mediolateral (ML) SAG axis was calculated from 5 serial cryosections, in which the biggest measurement was used.
The number of Neurod+ nuclei was counted manually using DAPI staining in the TgBAC(neurod:egfp) nl1 labelled cells. neurog1+ cells were also counted manually after an ISH and DAPI staining. The number of HC was calculated using the Tg(brn3c:mGFP) s356t and counting kinocilia number.
The volume of the SAG was measured by calculating the area the 5 serial cryosections that include all SAG and multiplying them with each section thickness (20 μm).
Statistical Analysis
All the data in the work were first tested for normal distribution using the Kolmogorov-Smirnov test and the Levene’s test for homogeneity of variances. For two-group comparison two-tailed Student’s t-test was used or Mann-Whitney U test for non-parametric tests. When more than two samples were being compared, one-way ANOVA followed by post-hoc test was performed. Values are expressed as median ± SEM. Graphs were performed with PRISM 7 (Graph-Pad) software. Error bars represent SEM, but for Fig. 2d and Fig. 5a, were shaded error bars represent SD. ns= non-significant, * p ≤ 0.05, * * p ≤ 0.01, * * * p ≤0.001.
Immunostaining and in situ Hybridization
Embryos at the desired stage were fixed with 4% paraformaldehyde (PFA) in PBS 0,1% Tween-20 (PBT) for 3 hours at RT or overnight at 4 °C. Fixed embryos were dehydrated, kept in 100% MeOH -20 °C 1h or overnight (O/N), and rehydrated. Embryos were permeabilized with Proteinase K 1:1000 in PBT, or cold Acetone for mouse anti-Islet1/2, and incubated with blocking solution (BS) for 2h at room temperature (RT) and with primary antibody O/N at 4°C. Primary antibody was washed 5 times 10min, and secondary antibody was applied for 2h RT or O/N 4 ° C.
Embryos were cryoprotected in 15% sucrose and embedded in 7.5% gelatine/15% sucrose. Blocks were frozen in 2-methylbutane (Sigma) to improve tissue preservation. Sagittal and transversal sections were cut at 20 μm thickness in a Leica CM 1950 cryostat. For BrdU incorporation experiments, dechorionated live embryos were incubated with 10mM BrdU, 10% DMSO in Danieau’s solution for 6 hours. After this they were washed twice in clean Danieau’s solution for 5 min prior 4% PFA fixation for 3 hours at RT. BrdU staining was performed in cryosections after 2N HCl DNA denaturation 1h RT. Anti-BrdU (bdbiosciences; mouse Cat: 555627, 1:100). Anti-GFP was used to enhance GFP signal from transgenic lines (Torrey Pines BioLabs Inc; rabbit Cat: TP401, 1:400). Anti-Islet1/2 was used to detect differentiated neurons (Cat: TP401; mouse Cat: 39.4D5, 1:200). Alexa-fluor 488 and 594 anti-rabbit and anti-mouse (Invitrogen, 1:400) were used as secondary antibodies. DAPI staining for cryosections was done at 1:10000 PBS for 5min, and 2×5min wash in PBS followed by mowiol mounting.
Whole-mount in situ hybridization was performed as previously described74 using neurog1 digoxigenin-labelled probe.
Inhibitor treatments
To block actin polymerization in filopodia, dechorionated live embryos were incubated in 0,1 μg/mL Latrunculin B (LatB) (Merk, Millipore), 0,4% DMSO in Danieau’s solution for 6 hours40; while control embryos were raised at 0,4% DMSO only. After treatment, embryos were washed twice in clean Danieau’s solution for 5min followed by immediate confocal imaging for filopodia visualization or 4% PFA fixation. In other cases, embryos were washed in Danieau’s solution until 36 hpf. It has been previously demonstrated that the concentration used of LatB specifically inhibits filopodial formation while it does not affect other cellular behaviours for up to 17 hours of treatment40.
Blood flow was blocked through the use of two different drugs. Dechorionated live embryos were incubated for 6, 12 and 24 hours in 40μM Nifedipine or 20mM BDM (2, 3-butanedione-2-monoxime) (Sigma-Aldrich) in Danieau’s solution, washed twice 5min and fixed in 4% PFA at the desired stage52,53.
Morpholino knockdown experiments
To inhibit protein translation of Dll4, 10ng of random and dll4 splicing morpholino antisense oligonucleotides (Gene Tools, Philomath, OR) were injected in the yolk of 1-2 cell stage embryos as previously described43. Efficiency of the knockdown was assessed by RT-PCR (data not shown).
Acknowledgments
We thank D Stainer for the Tg(cloche) m378 and Tg(kdrl:ras-mCherry) s896 lines, the PRBB Imaging Facility for technical support, T Schilling, A Bigas and I Fariñas for comments on the manuscript and lab members for discussions. The work was supported by BFU2014-53203 and BFU2017-82723P from MCINN to BA and the Unidad de Excelencia Maria de Maeztu (MDM-2014-0370).