NMDA receptors coordinate brain vascular development via neuron-to-endothelial tip cell crosstalk in zebrafish

Abstract

N-methyl-D-aspartate receptors (NMDARs) are a major class of glutamate receptors crucial for neural development and function. Here, we report that NMDARs expressed on neurons regulate brain vascular development via neurovascular communication. Dysfunction of neuronal NMDARs impairs the formation of the zebrafish brain vasculature and abrogates the neural activity-induced enhancement of the brain vascular development. These defects are attributed to the reduced growth of vascular endothelial tip cells (ETCs) situating at the leading edge of brain angiogenic sprouts. At the molecular level, NMDAR dysfunction down-regulates neuronal expression of vascular endothelial growth factor and subsequent global Ca²⁺ activities of ETCs, thereby impairing ETC growth. Thus, our study uncovers an important role of NMDARs in brain vascular development, expanding the functional repertoire of NMDARs and the mechanistic understanding of the interplay between nervous and vascular systems during development.

Neurovascular interaction is critical for normal brain vascular development. Here, the authors show that neuronal NMDARs regulate brain vascular development via VEGF-induced global Ca2+ activities of endothelial tip cells.

Introduction

Interactions between the nervous and vascular systems are fundamental for proper brain development and function^1–9. During brain development, neurons and vascular endothelial cells (ECs) engage in close interplays via sharing a common neurovascular niche and exhibit similar specialized growth cone structures responding to some common environmental cues^1,4–11. Axon guidance cues, including Netrins, Slits, Semaphorins, Ephrin, Wnt, and Nogos, also regulate the navigation of vascular endothelial tip cells (ETCs), which guide angiogenic sprouts to their targets^12–18. Conversely, angiogenic factors, such as vascular endothelial growth factor (VEGF) and fibroblast growth factor-2, also play roles in neural development^{1,6,7,19–23}. Therefore, a comprehensive understanding of the neurovascular interaction is crucial for advancing our knowledge of the intricate interplay between the nervous and vascular systems.

Neural activity, the primary carrier of neural information, has been found to affect brain vascular growth^24–26. Previous studies show that physiological neural activities and enriched experiences can increase vascular density and branching in related local cerebral cortices^27–32, while hyperactivation of neural circuits leads to reduced local cerebrovascular development³³. In the mouse retina, melanopsin-dependent light responses or neuronal type-specific activities modulate vascular development and/or blood retinal barrier formation^34–37. However, a comprehensive understanding of the role of neural activity in the brain vasculature development, as well as the underlying mechanisms, remains elusive^24–26. N-methyl-D-aspartate receptors (NMDARs), a major class of ionotropic glutamate receptors primarily expressed in the central nervous system, play a pivotal role in neural activity-dependent neural development as well as brain functions^38–41. It is quite intriguing whether NMDARs are required for neural activity-dependent regulation of brain vascular development.

In the present study, by using in vivo time-lapse imaging of the developing brain vasculature in intact larval zebrafish in combination with genetic and optogenetic manipulations, we first showed that disruption of neuronal NMDARs or their signaling impaired the normal formation of the brain vasculature. In particular, NMDAR dysfunction completely inhibited the neural activity-induced enhancement of the brain vasculature development. Mechanistically, neural activity increased brain ETC growth via NMDARs, a process which was mediated by VEGF expression in neurons and VEGF-induced subsequent Ca²⁺ activities in ETCs. Our results indicate that NMDARs expressed in neurons are necessary for the neural activity-induced enhancement of brain vasculature development via VEGF - Ca²⁺ signaling cascade-dependent promotion of ETC growth.

Results

Dysfunction of NMDARs impairs the development of the zebrafish brain vasculature

To evaluate brain vascular development, we performed in vivo confocal imaging of the brain vascular network in transgenic Tg(kdrl:eGFP) zebrafish larvae, in which vascular ECs are labeled with the enhanced green fluorescent protein (eGFP) controlled by the EC-specific kdrl promoter⁴². In larval zebrafish, the expansion of brain vasculature is primarily achieved by the sprouting of newborn vessels from the choroidal vascular plexus (CVP) and primordial hindbrain channel, which respectively locates at the peripheral region of the midbrain and hindbrain, into the brain^43,44. Using the self-designed software Angiotome⁴⁵, we reconstructed midbrain vessels, extracted their centerlines, and calculated four primary parameters: vascular density (“Vascular density”), total length of vessel segments (“Total length”), number of all vessel segments (“Segment number”), and number of vessel segments emanated from CVP (“Segment from CVP”) (Fig. 1a, and Supplementary Movie 1).

Fig. 1. Neuronal NMDARs are required for brain vascular development in larval zebrafish.

a Schematic workflow of the analysis of the zebrafish brain vasculature from confocal imaging to morphology characterization. CVP, choroidal vascular plexus. A, anterior; L, lateral. Scale bars, 50 μm. b Representative centerlines of the midbrain vasculature of the zebrafish treated NMDAR antagonists MK-801 (10 μM) or D-AP5 (300 μM), or agonist NMDA (10 μM). c–f Summary data of vascular density (c), total length of vessel segments (d), number of vessel segments (e), and number of vessel segments emanated from CVP (f) in groups of control, MK-801, D-AP5, NMDA. g Representative images (up) and centerlines (bottom) of the midbrain vasculature in WT sibling and grin1b^+/-. Scale bars, 50 μm. h–k Summary data of vascular density (h), total length of vessel segments (i), number of vessel segments (j), and number of vessel segments emanated from CVP (k) in WT sibling and grin1b^+/- larvae. l Schematic showing the working mechanism of grin1b NQ and grin1b CT (left) and relevant plasmids for overexpression in neurons or ECs driven by the pan-neuronal elavl3 promoter or EC-specific kdrl promoter (right). Grin2n indicates Grin2a/2b/2c/2 d subunits. m–p Summary data of vascular density (m), total length of vessel segments (n), number of vessel segments (o), and number of vessel segments emanated from CVP (p) in groups of WT, tdTomato (neuron), NQ (neuron), NQ (EC), CT (neuron), CT (EC). q Representative images (up) and centerlines (bottom) of the midbrain vasculature in WT sibling and neuronal grin1b CKO. Scale bars, 100 μm. r–u Summary data of vascular density (r), total length of vessel segments (s), number of vessel segments (t), and number of vessel segments emanated from CVP (u) in sibling and neuronal grin1b CKO larvae. The numbers in the brackets indicate the number of larvae examined. One-way ANOVA was used for c–f and m–p, unpaired two-tailed Student’s t-test was used for h–k and r–u. Data are represented as mean ± s.e.m.

To investigate whether NMDARs are involved in brain vascular development, we first pharmacologically manipulated NMDAR activation by bath application of selective NMDAR antagonists (MK-801 and D-2-Amino-5-phosphonovaleric acid (D-AP5)) or agonist (NMDA) from 1.5 to 7 days post-fertilization (dpf), and then examined the effects on the brain vasculature at 7 dpf. In comparison with the control group, blockade of NMDARs by MK-801 or D-AP5 significantly impaired the brain vascular development in terms of all the four parameters mentioned above, while activation of NMDARs significantly promoted the brain vascular development (Fig. 1b–f, and Supplementary Fig. 1a). However, NMDAR blockade did not affect the growth of trunk vessels (Supplementary Fig. 1b). In addition, NMDAR blockade did not impair brain vascular integrity of 7-dpf larvae (Supplementary Fig. 1c, d) and the expression of blood-brain barrier-related signals including VE-cadherin and Wnt/β-catenin (Supplementary Fig. 1e–h). These results indicate that NMDAR activation is required for normal brain vascular development. As there are usually spontaneous neural activities in the brain under control conditions, and NMDAR activation is known to be closely associated with neural activity, the results above also imply that spontaneous neural activities contribute to the normal development of the brain vasculature.

NMDARs are heterotetrameric complexes composed of two obligatory GRIN1 subunits and two regulatory GRIN2 and/or GRIN3 subunits^46,47. To genetically manipulate NMDARs, we mutated the zebrafish grin1b gene, homologous to mammal GRIN1, with the transcription activator-like effector nuclease (TALEN) strategy by targeting two sites with one in the 5^th intron and the other in the 6^th exon (Supplementary Fig. 1i–k). As grin1b^+/- larvae were lethal after 6 dpf, we examined the brain vasculature of grin1b^+/- larvae and wild-type (WT) siblings at 3 dpf, and found that the brain vasculature was markedly impaired in grin1b^+/- larvae (Supplementary Fig. 1l, and Fig. 1g–k; p < 0.01), similar to the effects observed with pharmacological blockade. Consistently, the trunk vessels were relatively normal in grin1b^+/- larvae (Supplementary Fig. 1m).

As the lethality of grin1b^+/- mutants precluded examination at later development stages, we then used a morpholino oligo (MO)-based approach to knockdown grin1b. Two MOs (MO1 and MO2) targeting the start codon region of grin1b were designed, and their effectiveness was verified by reduced NMDA-induced currents recorded from brain neurons in MO-injected morphants (Supplementary Fig. 2a–e). The brain vasculature in the morphants was significantly impaired (Supplementary Fig. 2f–j; p < 0.001), while trunk vessels remained intact (Supplementary Fig. 2k). Collectively, these results indicate that NMDARs are required for the normal development of the brain vasculature.

NMDARs expressed on neurons are essential for brain vascular development

Besides its dominant expression in neural cells, NMDARs are also expressed in ECs within the brain^48,49. To determine whether NMDARs expressed in neurons or ECs mediate the effects on brain vascular development, we specifically interrupted NMDAR signaling in neurons or ECs via overexpressing grin1b point mutations with the pan-neuronal elavl3 promoter or EC-specific kdrl promoter, respectively. The asparagine 598 (N598) of GRIN1 is a key amino acid residue in the pore of NMDAR channels, and mutations substituting N598 with glutamine (Q) display defective NMDAR channel functions and perinatally lethality in mice⁵⁰. We overexpressed the zebrafish grin1b point mutant N598Q (“grin1b NQ”) with non-fused red fluorescent protein tdTomato in neurons or ECs, and found that neuron-specific overexpression of grin1b NQ mutation significantly reduced NMDAR-mediated currents (“grin1b NQ (neuron)”) (Fig. 1l, and Supplementary Fig. 3a–f). In addition, the calmodulin-binding site of GRIN1, locating in its C terminal domain, is required for the transduction of NMDAR downstream signals, including CaMKIIα and MEK-ERK⁵¹. To inhibit NMDAR-activated downstream signals, we overexpressed the zebrafish grin1b C terminal (grin1b CT) to competitively bind with endogenous calmodulin⁵¹. As expectedly, NMDAR-mediated currents were not significantly affected in grin1b CT-overexpressing neurons (“grin1b CT (neuron)”) (Fig. 1l, and Supplementary Fig. 3a–f), but the phosphorylation levels of ERK and CaMKIIα (i.e. pERK and pCaMKIIα) were markedly reduced (Supplementary Fig. 3g, h; p < 0.05), confirming that overexpression of grin1b CT can effectively inhibit NMDAR downstream signals. In vivo imaging at 7 dpf showed that neuron-specific overexpression of grin1b NQ or grin1b CT significantly impaired the brain vasculature (Supplementary Fig. 3i, and Fig. 1m–p; p < 0.001), whereas EC-specific overexpression did not show significant effects (Fig. 1m-p). Consistently, neuron-specific overexpression of grin1b in grin1b morphants largely rescued the brain vascular defects (Supplementary Fig. 3j–n; p < 0.01). These results indicate that neuronal NMDARs are crucial for brain vascular development.

Further, we generated a zebrafish line with grin1b conditional knockout in neurons (“neuronal grin1b CKO”) using Cre-loxP system-mediated switch of gene-trap cassettes with fluorescent reporter expression in manipulated cells⁵². With this strategy, we first generated a grin1b knock-in zebrafish line (grin1b^zCKOIS) by inserting loxP-flanked inverted splice acceptor site-TagBFP-PolyA (BGHpA) cassette into the 5^th intron of the endogenous grin1b gene via CRISPR/Cas9 (Supplementary Fig. 4a, b)⁵². Through outcrossing with Tg(elavl3:Cre) zebrafish, the inverted cassette between the loxP sites was reversed, resulting in grin1b deficiency specific in neurons due to polyA sequence-mediated transcriptional stop before the exon 6, along with TagBFP expression (Supplementary Fig. 4b, c). By monitoring the brain vascular network in 7-dpf larvae obtained from parental grin1b^zCKOIS;Tg(elavl3:Cre) and Tg(kdrl:eGFP) lines, we observed brain vascular defects in heterozygous neuronal grin1b CKO larvae compared with WT siblings (Fig. 1q–u). The observed vascular defects were underestimated because of the incomplete expression and recombination of Cre in zebrafish neurons (Supplementary Fig. 4c). In addition, we did not obtain homozygous neuronal grin1b CKO larvae, likely due to early developmental lethality. Taken together, these results demonstrate that neuronal NMDARs are crucial for the normal development of the brain vasculature.

NMDARs are required for neural activity-induced enhancement of brain vascular development

Considering that the brain usually undergoes high levels of spontaneous neural activity under normal conditions, in particular during development, our results above inspired us to examine whether NMDARs contribute to the regulation of brain vascular development by neural activity. We enhanced neural activity via optogenetic stimulation (“Opto. Stim.”) with the photoactivable ChrimsonR in transgenic Tg(elavl3:ChrimsonR-tdTomato) zebrafish larvae, in which neurons express ChrimsonR fused with tdTomato (Fig. 2a). The efficacy of ChrimsonR activation was verified by the increase of brain-wide Ca²⁺ activities following red light stimulations in double transgenic Tg(elavl3:ChrimsonR-tdTomato);Tg(evlal3:GCaMP6s) larvae, in which neurons express both ChrimsonR and the Ca²⁺ indicator GCaMP6s (Fig. 2b–d; see Methods). Optogenetic stimulation was administered for 8 − 10 h per day during 3 − 6 dpf on dark-reared Tg(elavl3:ChrimsonR-tdTomato);Tg(kdrl:eGFP) larvae (see Methods). We found that optogenetic stimulation-induced neural activity significantly promoted brain vascular formation examined at 7 dpf (Fig. 2e–i; p < 0.01). This enhancement of brain vascular development was abolished by bath application of MK-801, which reduced the vascular development to a comparable level in larvae subjected to MK-801 alone (Fig. 2e–i, and Supplementary Fig. 5a).

Fig. 2. NMDARs are required for neural activity-induced enhancement of brain vascular development.

a Schematic showing the working mechanism of optogenetic activation via red light flash with a wavelength of 580 − 620 nm. b Pseudo-color images showing optogenetic stimulation-induced neural Ca²⁺ activities (∆F/F₀) in the midbrain of a 5-dpf zebrafish larva. Scale bar, 100 μm. c Heatmap showing neural Ca²⁺ activities of the midbrain evoked by consecutive five times of optogenetic stimulations (blue arrowheads) in six larvae. d Averaged trace of Ca²⁺ activities shown in (c). The trace in the insect shows the averaged Ca²⁺ activities from total five optogenetic stimulations in the left. The solid line and shaded band represent the averaged signal and its 95% confidence interval. Dash lines indicate the stimulations. e Representative centerlines of the midbrain vasculature of the four groups. f–i Summary data of vascular density (f), total length of vessel segments (g), number of vessel segments (h), and number of vessel segments emanated from CVP (i) in groups of control, Opto. stim., Opto. stim. + MK-801, MK-801. j–m Summary data of vascular density (j), total length of vessel segments (k), number of vessel segments (l), and number of vessel segments emanated from CVP (m) in groups of control, Elect. stim., Elect. stim. + MK-801, MK-801. n Representative centerlines of the midbrain vasculature of the four groups. Dark green centerlines indicate vessel segments in the optic tectum, and light green centerlines indicate vessel segments in other regions of the midbrain. o–r Summary data of vascular density (o), total length of vessel segments (p), number of vessel segments (q), and number of vessel segments emanated from CVP (r) in groups of control, Moving bar, Moving bar + MK-801, MK-801. The numbers in the brackets indicate the number of larvae examined. Unpaired two-tailed Student’s t-test was used for f–m and o–r. Data are represented as mean ± s.e.m.

Similarly, whole-field electrical stimulation (“Elect. stim.”), which could induce brain-wide neuronal activities in larvae (Supplementary Fig. 5b–e; see Methods), also significantly promoted brain vascular development (Supplementary Fig. 5f, and Fig. 2j–m; p < 0.05), and these effects were completely suppressed by bath application of MK-801 (Fig. 2j–m; p < 0.001).

To further elucidate the role of NMDARs in brain vascular development under physiological conditions, we employed light moving bar stimulation, which could robustly evoke neuronal activities in the optic tectum (OT), a major region of the midbrain (Supplementary Fig. 5g–j; see Methods), and examined its effects on the growth of OT vessels in dark-reared larvae. Moving bar stimulation applied during 4.5 − 6 dpf significantly increased OT vascular formation examined at 6 dpf (Fig. 2n–r, and Supplementary Fig. 5k; p < 0.01), and these enhanced effects were also inhibited by bath application of MK-801 (Fig. 2n–r; p < 0.001).

Taken together, these results demonstrate that NMDARs are required for neural activity-induced enhancement of brain vascular development.

NMDARs mediate neural activity-induced promotion of ETC growth via increasing ETC Ca²⁺ activity

At the forefront of growing vessels, ETCs extend a few motile primary branches and multiple dynamical fine filopodia^53–56, and steer angiogenic sprouts to appropriate targets^56–58. To examine whether NMDARs enhance brain vascular development through regulating ETC behaviors, we performed in vivo imaging of brain ETCs in larvae at 4 dpf and characterized ETC morphological properties, including the average number of ETCs in the midbrain per larva (“ETC number”), number of filopodia per ETC (“Filopodium number”), and length of ETC filopodia (“Filopodium length”), and calculated the growth rate of ETCs during the stimulation process (“Growth rate”) (Fig. 3a; see Methods). We found that brain ETCs in neuronal grin1b CKO larvae exhibited impaired growth dynamics in comparison with WT siblings reared under normal condition (Fig. 3b–e). To test the effect of neural activity on the behavior of ETCs, we enhanced neural activity via optogenetic stimulation in dark-reared Tg(elavl3:ChrimsonR-tdTomato);Tg(kdrl:lyn-sfGFP) larvae, and found the growth of ETCs was significantly promoted (Supplementary Fig. 5l, and Fig. 3f–i; p < 0.05), and these effects were abolished by bath application of MK-801 (Fig. 3f–i; p < 0.05). Similar results were also observed when moving bar stimulation was used to increase neural activities in dark-reared larvae (Supplementary Fig. 5m, and Fig. 3j–m; p < 0.01). These results suggest that NMDARs mediate the neural activity-induced promotion of brain ETC growth.

Fig. 3. NMDARs mediate neural activity-promoted growth of ETCs.

a Left, morphology of a typical endothelial tip cell (ETC) in the midbrain vasculature of a 4-dpf Tg(kdrl:lyn-sfGFP) larva. Right, schematic showing the filopodia, branch and soma of the ETC. Scale bar, 10 μm. b–e Summary data of ETCs’ number (b), growth rate (c), filopodium number (d), and filopodium length (e) in sibling and neuronal grin1b CKO larvae. f–i Summary data of ETCs’ number (f), growth rate (g), filopodium number (h), and filopodium length (i) in groups of control, Opto. stim., Opto. stim. + MK-801, MK-801. j–m Summary data of ETCs’ number (j), growth rate (k), filopodium number (l), and filopodium length (m) in groups of control, Moving bar, Moving bar + MK-801, MK-801. The number in the brackets indicates the number of larvae (b, f, and j), the number of ETCs (c, d, g, h, k, and l), or the number of filopodia (e, i, and m). Unpaired two-tailed Student’s t-test was used for b and c, f and g, and j and k; two-sided Mann-Whitney test was used for d and e, h and i, and l and m. Data are represented as mean ± s.e.m.

We also compared the growth rate of ETCs of normally reared sibling larvae (in Fig. 3c) with these of darkness-reared control larvae (in Fig. 3g, k). Dark rearing caused a significant reduction of ETC growth rate (Sibling in Fig. 3c: 4.78 ± 0.79 µm/h, Control in Fig. 3g: 1.59 ± 0.36 µm/h, Control in Fig. 3k: 1.17 ± 0.43 µm/h; p < 0.001), suggesting a role of rearing illumination-associated neural activities in ETC growth.

How does neural activity regulate brain ETC growth via NMDARs? Previous studies have indicated that global Ca²⁺ activities across the soma and processes of ECs can increase EC migration and angiogenic capacity in cultured systems as well as zebrafish trunk vessels^59–61, while local Ca²⁺ activities occurred in ETC branches regulate branch extension or retraction in the zebrafish brain vasculature⁵⁶. To examine the role of Ca²⁺ activities in ETC growth, we tried to induce global Ca²⁺ activities in ETCs through 405-nm laser pulse-evoked Ca²⁺ release from NP-EGTA/Ca²⁺ complex^56,62, which was locally applied to the area of the ETC examined prior to imaging on transgenic Tg(kdrl:GCaMP6s);Tg(kdrl:HRAS-mCherry) larvae at 3 − 4 dpf (see Methods). We then monitored the Ca²⁺ activity and the growth of ETCs. Uncaging NP-EGTA/Ca²⁺ complex could induce global Ca²⁺ transients, which swept across ETC soma and branch, and significantly increased ETC growth (Fig. 4a–c, and Supplementary Movie 2; p < 0.05). Furthermore, we monitored the growth of ETCs displaying various levels of spontaneous global Ca²⁺ activities, and found that the global Ca²⁺ transient frequency positively correlated with the growth rate of ETCs (Fig. 4d; p < 0.001). We noticed that the frequencies of global Ca²⁺ transients in both DMSO and NP-EGTA sham groups were lower than that in the spontaneous group, possibly due to experimental manipulations including skin incision, micropipette penetration as well as chelation of intracellular Ca²⁺ by NP-EGTA. These results imply that global Ca²⁺ activities can promote ETC growth.

Fig. 4. NMDARs promote ETC growth via increasing ETC Ca²⁺ activities.

a Left, representative simultaneous time-lapse images showing evoked Ca²⁺ activity (top) and morphology (bottom) of an ETC. Right, traces of Ca²⁺ activities in the branch (top) and soma (bottom) of the ETC in responding to 405-nm laser-induced uncaging (dashed lines) of NP-EGTA/Ca²⁺ complex (dashed lines). T1 and T2 indicate the time points at which the images in (left) were captured. Scale bar, 10 μm. b Traces of Ca²⁺ activities in the branch (top) and soma (bottom) of an ETC in responding to 405-nm laser application (dashed lines) under DMSO condition as a control. c Summary data showing the comparison of evoked global Ca²⁺ transient frequency (x-axis) and growth rate (y-axis) of ETCs between NP-EGTA uncaging group and two control groups including DMSO and NP-EGTA sham uncaging (NP-EGTA sham). Each data point was obtained from individual larvae. d Positive correlation between the frequency of global Ca²⁺ transients and growth rate of ETCs. Black open dots indicate the data obtained from ETCs with various levels of spontaneous global Ca²⁺ activities (“Spont.”), and the pink, gray, and orange open dots represent the data shown in (c). The black line represents linear regression for all the data points. e Typical case from an ETC, showing Ca²⁺ activities under the three conditions. With MK-801 application, local Ca²⁺ transients in ETC soma or branches were still observed. Filled arrowheads, global Ca²⁺ transients; open arrowheads, local Ca²⁺ transients. f Summary data from 8 ETCs, showing the number of global Ca²⁺ transients per h in groups of Spont., Opto. stim., Opto. stim. + MK-801. Each data point was obtained from individual larvae. The number in the brackets indicates the number of ETCs. One-way ANOVA was used for c; two-sided Wilcoxon matched-pairs signed rank test was used for f; Simple Linear Regression analysis for was used for d. Data are represented as mean ± s.e.m.

To further examine whether neural activity affects global Ca²⁺ activities of brain ETCs via NMDARs, we induced neural activities at the neuronal soma region of the posterior OT in Tg(elavl3:ChrimsonR-tdTomato);Tg(kdrl:lyn-sfGFP) larvae via laser-based local optogenetic stimulations (Supplementary Fig. 6a, b), which could also increase the growth capability of nearby ETCs (Supplementary Fig. 6c–e). Then, we examined and found that local optogenetic stimulations markedly increased the number of global Ca²⁺ transients in nearby ETCs (Fig. 4e, f; p < 0.01). Furthermore, most of the global ETC Ca²⁺ transients were suppressed by bath application of MK-801 (Fig. 4e, f; p < 0.01). These results collectively demonstrate that NMDARs are required for neural activity-induced enhancement of ETC growth via regulating ETC global Ca²⁺ activity.

NMDARs mediate neural activity-induced ETC Ca²⁺ activities via VEGF signaling

VEGF is a crucial angiogenetic factor for vascular development and also a neurotrophic factor for neurogenesis and axon pathfinding^21,63. During brain vascular development, VEGF is primarily secreted by neural progenitors and mature neurons^6,64, and indispensable for blood vessel invasion, directional growth of capillaries, and vascular patterning^65–69. Therefore, we examined whether VEGF signaling acts downstream of neuronal NMDARs. First, we found that the mRNA levels of zebrafish vegfaa and vegfab in neurons, which were sorted from 3-dpf grin1b morphants (Supplementary Fig. 6f; see Methods), were markedly lower than those of control MO-injected larvae (Fig. 5a; p < 0.01). Second, we examined the protein level of VEGF-A in 5-dpf grin1b morphants and larvae with neuron-specific overexpression of grin1b CT, and found that VEGF-A was also reduced compared with that in control groups (Fig. 5b, c; p < 0.001). These results indicate that neuronal VEGF is downstream of NMDARs.

Fig. 5. Neuronal NMDARs regulate VEGF expression via ERK/CaMKIIα pathways.

a Vegfa (both vegfaa and vegfab isoforms) mRNA expression in sorted neurons from 3-dpf larvae with injection of control MO or grin1b MO1. b Representative western blotting showing decreased VEGF-A protein expression in 5-dpf larvae with injection of grin1b MO1 (top) or elavl3-driven neuronal overexpression of grin1b CT (bottom), in comparison with control groups. c Summary of VEGF-A protein expression in western blotting results represented in (b). d Representative western blotting showing decreased VEGF-A protein expression in larvae treated with PD98059 to block ERK activity or KN93 to block CaMKIIα activity. e Summary of VEGF-A protein expression in western blotting results represented in (d). A pool of 20 − 30 larvae were used in each group per independent experimental repeat. The number in the brackets indicates the number of independent experimental repeats (a, c, and e). Unpaired two-tailed Student’s t-test was used for a, c, and e. Data are represented as mean ± s.e.m.

Furthermore, inhibition of ERK activity with PD98059⁷⁰ or CaMKIIα activity with KN93⁷¹ significantly reduced VEGF-A expression in larvae (Fig. 5d, e; p < 0.001). Together with data showing the impaired phosphorylation level of ERK and CaMKIIα in larvae with neuron-specific dysfunction of NMDARs (see Supplementary Fig. 3f, g), these results indicate that neuronal NMDAR can regulate VEGF expression via ERK/CaMKIIα pathways.

Previous studies showed that VEGF-stimulated angiogenic responses are associated with elevated Ca²⁺ activities in ECs^59–61. To further test whether VEGF can induce Ca²⁺ activities in ETCs, we locally ejected exogenous VEGF protein via micropipettes to brain ETCs with relatively low-frequency spontaneous Ca²⁺ activities in 4-dpf Tg(kdrl:GCaMP6s) larvae. We found that VEGF application evoked global Ca²⁺ activities in ETCs (Fig. 6a, b, and Supplementary Movie 3). Moreover, both optogenetic stimulation-induced and spontaneous global Ca²⁺ transients of ETCs were suppressed by bath application of Ki8751 (Fig. 6c, d, and Supplementary Movie 4; p < 0.05), a selective inhibitor of the VEGF receptor 2, which mediates the major effect of VEGF on vascular development^72,73.

Fig. 6. NMDARs mediate neural activity-induced ETC Ca²⁺ activities via VEGF signaling.

a Typical traces of Ca²⁺ activities at the branch and soma of ETCs evoked by local puff application of 0.1% bovine serum albumin (BSA) or VEGF. The breaks in the right traces indicate the time used for unclogging the micropipette for puffing to the targeted ETC. Gray dashed lines indicate local puff pulses. b Summary of data showing global Ca²⁺ activities of ETCs induced by repetitive local puff application of 0.1% BSA or VEGF. Each pink square indicates a successful induction of global Ca²⁺ transients, and each gray square indicates a failed induction. The traces shown in (a) for 0.1% BSA and VEGF groups correspond to ETC 4 and ETC 5 in the relevant groups, respectively. c, d Suppression of optogenetic stimulation-evoked global Ca²⁺ activities of ETCs by application of the VEGFR2 inhibitor Ki8751. Tg(elavl3:ChrimsonR-tdTomato);Tg(kdrl:GCaMP6s) larvae at 4 dpf were used. Data were obtained under three consecutive conditions: spontaneous (“Spont.”), optogenetic stimulation (“Opto. stim.”), and optogenetic stimulation in combination with Ki8751 treatment (“Opto. stim. + Ki8751”). c Typical case from an ETC, showing Ca²⁺ activities under the three conditions. With Ki8751 application, local Ca²⁺ transients in ETC soma or branches were still observed. Gray dashed lines, global Ca²⁺ transients; open arrowheads, local Ca²⁺ transients. d Summary data from 7 ETCs, showing the number of global Ca²⁺ transients per h in groups of Spont., Opto. stim., Opto. Stim. + Ki8751. Each data point was obtained from individual larvae. e Working model. NMDARs activated by neural activity enhance neuronal VEGF expression through downstream ERK/CAMKIIα pathway and released VEGF further stimulates global Ca²⁺ activities and associated growth capability of ETCs, leading to enhanced vascular development in the brain. The number in the brackets indicates the number of larvae examined. Paired two-tailed Student’s t-test was used for d. Data are represented as mean ± s.e.m.

Taken together, these results indicate that neuronal VEGF works as an intermediator between neuronal NMDARs and the Ca²⁺ activity and growth of ETCs during brain vascular development.

Discussion

Our study uncovers the function of NMDARs in regulating brain vascular development. Specifically, NMDARs activated by neural activity promote neuronal VEGF expression through ERK/CaMKIIα pathways, and released VEGF subsequently increases the global Ca²⁺ activity and associated growth capability of ETCs, leading to enhanced vascular development in the brain (Fig. 6e).

Neural activity can shape neural development, ranging from neurogenesis, axon pathfinding, and synaptogenesis to circuit formation, and neuronal NMDARs are involved in these processes^{38–40,74–76}. Besides, neural activity can also regulate vascular development in the brain^27–33 and retina^34–37,77. In the mouse cerebral cortex, sensory-related neural activities or enriched environment rearing promote vascular development, and neurotransmitter release seems to be required in this process²⁷. In the mouse retina, melanopsin-dependent light responses regulate the patterning of blood vessels in the eye³⁴. In addition, dopamine derived from retinal ganglion cells inhibits early postnatal vessel growth³⁶, while cholinergic neuronal activities of starburst amacrine cells and glutamatergic neuronal activities of photoreceptors and bipolar cells regulate angiogenesis of the deep plexus and the blood retinal barrier formation^35,37. These studies focusing on specific regions of the central nervous system highlight crucial roles of neuronal type-specific activities in regulating vascular development and barriergenesis. Here, our study focused on the whole-brain vasculature development in larval zebrafish, and revealed that brain-wide neural activity-induced NMDAR activation promotes vascular development through VEGF-dependent enhancement of ETC growth. The study uncovered an important molecular and cellular mechanism underlying the regulation of brain vascular development by neural activity.

During development, the brain exhibits high levels of spontaneous neural activities that activate neuronal NMDARs³⁹. In our study, we found that pharmacological blockade or genetic perturbation of NMDARs caused impairment of brain vascular development under normal condition (see Fig. 1). In the nervous systems, ECs also express NMDARs, which can regulate hemodynamic responses^49,78, blood-brain barrier permeability⁷⁹, and vascular remodeling under pathological conditions⁸⁰. Collectively, it is plausible that NMDARs play an important function in interplays between vascular and neural systems in development, physiology as well as pathology.

During brain vascular development, VEGF is primarily secreted by neural progenitors and mature neurons^6,64. We found that NMDAR blockade by MK-801 slightly decreased the proliferation of neural precursors (Supplementary Fig. 7a–d), and the number of different types of neurons including total neurons, excitatory neuron, and inhibitory neurons (Supplementary Fig. 7e–j). Thus, in addition to the essential role of NMDARs in regulating VEGF production in individual neurons, the effects of NMDARs on neurogenesis may also contribute to the overall neuronal VEGF production and brain vasculature development.

Compared with siblings, 3-dpf grin1b^+/− larvae showed a significant reduction in the radius of midbrain vessels (Supplementary Fig. 8a, b). In contrast, no significant difference in vessel radius was observed in 7-dpf neuronal grin1b CKO or larvae treated with MK-801, D-AP5, or NMDA compared to controls (Supplementary Fig. 8c–i). This discrepancy may reflect developmental stage-dependent effects of NMDARs in angiogenesis. At early developmental stage, brain angiogenesis is vigorous, and disruption of grin1b leads to impaired VEGF production and ETC growth, and delayed vessel maturation as reflected by narrower vessels. As development progresses, vessel radius may be normalized, while the early growth deficits lead to reduced vascular branching and network length.

Situating at the leading edge of growing vessels, ETCs navigate through tissue microenvironments and guide angiogenic sprouts to their appropriate targets^56–58. Our study reveals that the brain ETC can sense neural activity-induced and NMDAR-dependent VEGF signaling to promote its growth. Besides VEGF, ETCs can respond to other cues released by nearby cells⁵³. For example, neural cell-derived Wnt and Norrin promote the formation and sprouting of ETCs via activating endothelial β-catenin^16,81,82, and Netrin1a evokes retraction of ETC filopodia via UNC5B receptors to inhibit sprouting angiogenesis¹². As the ETC can detect, integrate, and respond to multiple-modal signals originating from surrounding cells or tissues, it may act as a cellular substrate for the interaction between the vascular and nervous systems during development.

Methods

Zebrafish husbandry

Adult zebrafish (Danio rerio) were housed in an automatic fish housing system (ESEN, Beijing, China) and maintained at 28 °C. Unless otherwise specified, embryos and larvae were raised under a 14h-10h light-dark cycle in 10% Hank’s solution, which contains (in mM): 140 NaCl, 5.4 KCl, 0.25 Na₂HPO₄, 0.44 KH₂PO₄, 1.3 CaCl₂, 1.0 MgSO₄, and 4.0 NaHCO₃ (pH 7.2). To inhibit pigment formation, embryos were treated with 0.003% 1-phenyl-2-thiourea (PTU, P7629, Merck). All animal protocols involving zebrafish were reviewed and approved by the Animal Care and Use Committee of the Center for Excellence in Brain Science and Intelligence Technology, Chinese Academy of Sciences (Approval No. NA-046–2023).

Zebrafish lines

Zebrafish larvae at 1.5 − 7 dpf were used in this study. Tg(kdrl:eGFP)⁴² was used for brain vascular imaging; grin1b^zCKIOS;Tg(elavl3:Cre) was used for conditional knockout of grin1b in neurons; Tg(kdrl:lyn-sfGFP) was used for brain ETC imaging; Tg(elavl3:GCaMP6s) and Tg(kdrl:GCaMP6s) were used for calcium imaging in neurons and ETCs, respectively; Tg(elavl3:ChrimsonR-tdTomato) was used for the optogenetic activation of neurons; Tg(elavl3:YC2.1)⁸³ was used for cell sorting experiments; Tg(her4.1:eGFP), Tg(elavl3:H2B-GCaMP6s)⁸⁴, Tg(vglut2a:GFF);Tg(5XUAS:GFP)⁸⁵, and Tg (gad67:eGFP)⁸⁶ were used for examine the numbers of neural precursor, total neurons, excitatory neurons, and inhibitory neurons, respectively; Tg(7XTCF-Xla.Siam:GFP;fli1:DsRed)⁸⁷ was used to examine Wnt/β-catenin signaling. AB/WT zebrafish was used for western blot. Tg(actb2:loxP-eGFP-polyA-loxp-tdTomato was used to visualize Cre expression pattern (Hunter Biotechnology company, Nanjing, China).

Generation of grin1b^+/- line using TALEN strategy

Two TALEN expression vectors were constructed via unit assembly⁸⁸. To prepare capped mRNA, the vectors were linearized by NotI and in vitro transcribed using SP6 mMESSAGE mMACHINE Kit (AM1340, Ambion). Mixed mRNAs were injected into 1-cell stage Tg(kdrl:eGFP) embryos with 100 pg per embryo. To detect the mutations, the genomic DNA was extracted using ten of F0 embryos for PCR with primers (P1: 5′ -AAATCGGAGGCATCCCGCGT-3′; P2: 5′-ACACTGATTCGTCGGTGGGGA-3′). A 493-bp PCR product (300 ng) containing the TALEN targeting sites was digested with SacII at 37 °C for 4 h. The uncleaved band was recovered and cloned for sequencing analysis. The adult F0 founders carrying the mutation were then crossed with Tg(kdrl:eGFP) line to obtain F1 larvae. After confocal imaging of these F1 larvae, genotyping identification was performed to discriminate grin1b^+/- from WT sibling.

Generation of grin1b^zCKOIS zebrafish and conditional knockout of grin1b in neurons

The grin1b^zCKOIS line was generated with CRISPR/Cas9-mediated method as previously described⁵². Based on the CRISPR/Cas9-mediated strategy, we selected the sgRNA sequence of grin1b using the CRISPR/Cas9 design tool (http://crispor.tefor.net/crispor.py). The sequence of grin1b sgRNA is TTAGCGATTAGCAGTCTGGT (Reverse strand). The left arm was amplified from WT zebrafish genomic DNA with the following primers: Forward, 5′-GAGTTACGCAGATTATCTCAGAACG-3′; Reverse, 5′-TGAAAGCGAACCACACCAGG-3′. The right arm consists of the rest sequence of the intron 5, the CDS sequence from exon 6 to exon 21, and 3’ UTR of zebrafish grin1b. The rest sequence of the intron 5 was amplified from WT zebrafish genomic DNA with the following primers: Forward, 5′-CACGGCAAAGTTTCTAAATG-3′; Reverse, 5′-CTGTGATCGGATAAAGATTG-3′. CDS from the exon 6 to exon 21 was amplified from cDNA which was reversely transcribed from WT zebrafish mRNA with RevertAid First Strand cDNA Synthesis Kit (K1622, Thermo) with the following primers: Forward, 5′-GAAGACGAAGCCGCGGCCAT-3′; Reverse, 5′-TCACACCACTGTGCTCACAG-3′. 3’UTR was amplified from WT zebrafish genomic DNA with the following primers: Forward, 5′-AAGTGAAGAGAGAAAGAGAA-3′; Reverse, 5′-CAATCACAAATAGCAATCTG-3′.

The inverted cassette with loxP sites, splice acceptor, and BGHpA were amplified from the hey2^CKOIS plasmid⁴⁸. TagBFP was amplified from the donor plasmid KI(gfap-TagBFP). To establish the fish line, each embryo was injected with 1 nL solution containing 600 ng/µl zCas9 mRNA, 200 ng/µl sgRNA, and 20 ng/µl donor plasmid at 1-cell stage. To identify the grin1b^zCKIOS founders, adult fish were crossed with AB/WT zebrafish. Genomic DNA extracted from 3 dpf larvae was used for screening by PCR using the following primers: F1, 5′-CCCTTGAGTTTGCATGTGAAATGC-3′; R1, 5′-GCAATTCGGGTCGATGGTGAA-3′. To detect the inversion reversed cassette in Tg(elavl3:Cre); grin1b^zCKIOS, following primers were used for verification: F1, 5′-CCCTTGAGTTTGCATGTGAAATGC-3′; F2, 5′-TGCTAGGGAGGTCGCAGTAT-3′.

Generation of Tg(kdrl:lyn-sfGFP), Tg(kdrl:GCaMP6s) and Tg(her4.1:eGFP) lines

To generate the Tg(kdrl:lyn-sfGFP) line, the 741-bp lyn-sfGFP was amplified and inserted into the yTol2-kdrl backbone followed by sequencing verification. Then, 1 nL solution with an equal dosage of 25 ng/µl yTol2-kdrl-lyn-sfGFP and Tol2 transposase mRNA were co-injected into AB/WT embryo at the 1-cell stage. The injected embryos were raised to adult stages for screening the transgenic founders.

To generate the Tg(kdrl:GCaMP6s) line, the 1406-bp GCaMP6s-CAAX was amplified and inserted into the yTol2-kdrl backbone followed by sequencing verification. Then, 1 nL solution with an equal dosage of 25 ng/µl yTol2-kdrl-GCaMP6s-CAAX and Tol2 transposase mRNA were co-injected into AB/WT embryo at the 1-cell stage. The injected embryos were raised to adult stage for screening the transgenic founders.

To generate the Tg(her4.1:eGFP) line, the eGFP fragment was amplified and subcloned into the plasmid yTol2:her4.1:mCherryT2ACreERT2⁸⁹ to replace the mCherryT2ACreERT2 cassette. This original plasmid was a kind gift from Dr. Michael Brand (Technische Universität Dresden, Dresden, Germany). Then, 1 nL solution with an equal dosage of 25 ng/µl yTol2:her4.1:eGFP and Tol2 transposase mRNA were co-injected into AB/WT embryo at the 1-cell stage. The injected embryos were raised to adult stage for screening the transgenic founders.

Morpholino oligo-based knockdown and microinjection

Morpholino oligos (MOs) were purchased from Gene Tools (Philomath, Oregon) and were dissolved in nuclease-free water. Each MO was injected into 1-cell stage Tg(kdrl:eGFP) embryos with 0.75 − 1 ng. The targeting sequences of MOs are as follows:

grin1b MO1: 5’-TTTGCCCTGGATGGCAGAGAAGTCG-3’;

grin1b MO2: 5’-AAAGGCGAACAGAACCAGGCGCATT-3’;

control MO: 5’-CCTCTTACCTCAGTTACAATTTATA-3’.

Construction and microinjection of overexpression plasmids

Plasmids including elavl3-tdTomato, elavl3-tdTomato-2A-grin1b, elavl3-tdTomato-2A-grin1b NQ, elavl3-tdTomato-2A-grin1b CT, kdrl-tdTomato-2A-grin1b NQ and kdrl-tdTomato-2A-grin1b CT were cloned in the tol2mini-MCS vector⁹⁰. P2A sequence encoding the self-cleavage 2A peptide was used for unfused co-expression of grin1b CT, or grin1b NQ with tdTomato⁹¹. grin1b coding sequence was amplified by RT-PCR. grin1b NQ sequence was obtained using Fast Mutagenesis system Kit (Transgene, China). The sequences of the elavl3 and kdrl promoters were amplified from elavl3-YC2.1 and kdrl-MCS plasmids, respectively^83,92. Tol2 transposase mRNA was transcribed using SP6 mMESSAGE mMACHINE Kit (Ambion) from NotI linearized pCS-TP⁹³. 1 nL solution with an equal dosage of 25 ng/µl overexpression plasmids and Tol2 transposase mRNA was co-injected into Tg(kdrl:eGFP) embryos at 1-cell stage for further experiments.

Neuron sorting and qPCR

Tg(elavl3:YC2.1) zebrafish larvae at 3 dpf, injected with grin1b MO1 or control MO, were digested by trypsin⁹⁴. The dissociated cells were sorted using a 488 nm channel, with the total number of sorted cells exceeding 10⁵ for subsequent experiments. Total RNA was extracted using TRIzol reagent according to the manufacturer’s instructions for real-time PCR.

qPCR primers for zebrafish genes were used as follows:

vegfaa-F: 5’- ACGAAACGTCACTATGGAGGTGCT-3’;

vegfaa-R: 5’-CTTTGGCCTGCATTCACACTTGGT-3’;

vegfab-F: 5’-GCTGTTCGCGTGCTCCAGTTATTT-3’;

vegfab-R: 5’-ACCTCCATGAATGGCACTACGTCA-3’;

gapdh-F: 5’-ATACACGGAGCACCAGGTTG-3’;

gapdh-R: 5’TCGTTGAGAGCAATACCAGCA-3’.

Western blotting

For each group, a pool of 20–30 zebrafish larvae at 5 dpf was lysed in RIPA buffer (P0013D, Beyotime) supplemented with a protease inhibitor cocktail (050M4063, Merck). During lysis, the tissue was thoroughly ground on ice, followed by lysis on ice for 30 min. The lysate was then centrifuged at 13000 x g for 10 min at 4 °C. The supernatant was collected and mixed with loading buffer containing SDS, Tris-HCl, β-mercaptoethanol, glycerol, and bromophenol blue, followed by boiling at 100 °C for 10 min. Equal amounts of protein samples were loaded onto the 12% SDS-PAGE gels for electrophoresis and subsequently transferred onto PVDF membranes. Then, the membranes were incubated in a blocking solution (5% BSA, 0.1% Tween-20 in TBS) for 2 h at room temperature. Following blocking, the membranes were incubated with primary antibodies for overnight at 4 °C. On the second day, the membranes were washed three times for 15 min each with TBST buffer (0.1% Tween-20 in TBS) and incubated with HRP-conjugated donkey anti-mouse or anti-rabbit secondary antibodies for 2 h at room temperature. After additional washing, the membrane was treated with ECL Western Blotting Substrate (32209, Pierce) for signal detection. The primary antibodies used in the current study include: anti-VEGF (1:1000; sc-152, Santa Cruze), mouse anti-beta Actin (1:10000; M200010, Abmart), rabbit anti-beta Actin (1:20000; B9901, Immunoway), anti-pCAMKIIα/β (1:1000; 06-881, Millipore), anti-CAMKIIα (1:1000; 05-532, Millipore), anti-pERK (1:1000; 4370 s, Cell Signaling), and anti-ERK (1:1000; 9107, Cell Signaling), anti-VE-cadherin (1:1000; LS-C210049, LSBio), anti-GAPDH (1:5000;104945-1-AP, Proteintech). For the quantification analysis of western blotting results, expression of interested protein was firstly normalized to the reference protein within each group. Subsequently, values in the experimental group were normalized to those in the corresponding control group. For grin1b CT-related western blotting experiments, the injected embryos were screened at 3 dpf under a mercury-arc fluorescence microscope, and only those exhibiting high tdTomato expression in the brain were pooled and used for the subsequent protein lysis.

In vivo confocal imaging

In vivo imaging was performed on larvae at 3–7 dpf at room temperature. The larvae were immobilized in 1% low melting-point agarose (2070-100GCN, Merck) for imaging. Three-dimensional confocal imaging was conducted using an Olympus Fluoview 1000 (Olympus) or Nikon FN1 confocal microscope (Nikon), respectively equipped with a 20 × (NA 0.95) or 25 × (NA 1.1) water immersion objective, with a Z-axis resolution of 3 or 4 μm. Laser beams of 405 nm, 488 nm, or 561 nm were used to excite BFP (for photolysis of NP-EGTA/Ca²⁺ complex), eGFP/sfGFP/GCaMP6s (for morphology or Ca²⁺ activity), or tdTomato/mCherry/ChrimsonR (for morphology or photoactivation), respectively. For Ca²⁺ imaging on ETCs or neurons, imaging intervals of 6 − 10 seconds and 1 second were used, respectively. Z-axis projected images of ETCs were used for analysis.

Quantitative analysis of the midbrain vasculature

To analyze the properties of the brain vasculature, raw images were processed by the Angiotome 1.0⁴¹. Three main sequential steps were performed: vessel signal segmentation, vasculature extraction, and skeletonization. With computer-assisted algorithms, the volume of neural tissue, the volume of vasculature, and the centerlines of the zebrafish midbrain vasculature were obtained. Vascular density (volume of vasculature/volume of neural tissue), total length of vessel segments, number of all vessel segments, and number of vessel segments emanated from CVP were used to quantify the brain vascular growth. For analysis of vessels in the optic tectum in the moving bar experiments, the 3-D template of the optic tectum was prepared. Using the imaging data of 6-dpf Tg(elavl3:YC2.1) larvae, the optic tectum tissue containing the fiber area and cell body area was selected by polygons slices by slices. Subsequently, the properties of vasculature in the optic tectum were analyzed by Angiotome 1.0.

Analysis of ETC morphological properties

The morphology of brain ETCs was measured by ImageJ software (NIH). To measure the length of ETCs, the root position (start), where the ETC buds from the parental vessel, and the tip position (end), where the free terminal of the longest branch is located, was defined. The path between the start and end position was then traced, and the length was calculated using the Simple Neurite Tracer plugin. The growth rate of ETCs was calculated by dividing the difference of length of the ETC at two time points by the time interval between the two time points. The length of filopodia was measured with the Simple Neurite Tracer plugin, and the number of filopodia was counted with the Cell Counter plugin. Data of ETC growth rate were obtained from zebrafish larvae at 4 − 4.5 dpf, with two imaging time points, namely early (4 dpf) and late (depending on the time length of stimulation in related experiment) time points, respectively. Data of the ETC number, filopodium number, and filopodium length were obtained from the early imaging time point at 4 dpf. In general, ETCs were captured and characterized in four conditions, including (1) ETCs appeared at both time points; (2) newborn ETCs only displayed at late time point but not early time point; (3) retracted ETC displayed only at early but not late time point; and (4) ETCs forming contact with targeting vessels at late time point. For analyzing ETC number, filopodium number, and filopodium length, ETCs under conditions (1), (3) and (4) were included. For analyzing the growth rate of ETCs, ETCs only under the condition (1) were included.

Detection of the brain vascular integrity

10 nl dextran (D22912, Invitrogen) or mixture of dextran (1 mg/ml) and DAPI (1 mg/ml) were injected into the circulation system via common cardinal vein in 7-dpf larvae. For dextran leakage detection, the images were taken immediately after injection. For dextran and DAPI leakage detection, the images were taken at 30 min after injection.

Immunofluorescence

The treated 7-dpf Tg(her4.1:eGFP) zebrafish larvae were fixed with 4% PFA at 4°C overnight, followed by dehydrating in 30% sucrose. Then, the larvae were embedded with OCT and cyrosected with the thickness of 50 μm. The brain slices were selected and washed with 0.5% PBST. After blocking with 5% goat serum in 0.5% PBST for 1 h at room temperature, the slices were incubated with PH3 primary antibody (1:100, 07–424, upstate) at 4°C overnight. On the next day, the brain slices were washed with 0.5% PBST for 3 times, and incubated with Alex-546 goat-anti-rabbit secondary antibody (1:500, A-11035, Invitrogen) for 2 h at room temperature. After washing, the slices were images for quantification.

Quantification of cell numbers in the larval brain

Zebrafish larvae treated with MK-801 or NMDA were imaged at 7 dpf. Tg(her4.1:eGFP) was used for counting the neural precursor number. The neural precursors were primarily distributed in the ventricle zone of the brain, with each cell spanning across about 9 μm in Z-axis. Considering the small population of neural precursors in the brain, and to avoid possible errors generated by missed cells during counting, we manually counted her4.1+ cell number of the midbrain in all image frames with Image J. Tg(elavl3:H2B-GCaMP6s), Tg(vglut2a:GFF);Tg(5×UAS:GFP) and Tg(gad67:eGFP) lines were used for counting the cell numbers of total neurons, excitatory neurons, and inhibitory neurons, respectively. We manually counted fluorescence-labeled neuron numbers of the midbrain in image frames every 9 μm in Z-axis with Image J.

Quantification of the neuronal labeling efficiency

The neuronal labeling efficiency of transiently overexpressed elavl3:tdTomato-P2A-grin1b CT was quantified using ImageJ. For each zebrafish larva, three optical sections of the optic tectum, spanning from the dorsal to ventral brain, were selected for analysis. Images were first threshold-adjusted, after which the soma regions were delineated using the selection tool. The percentage of labeled neurons within the soma region was then quantified using the “Area Fraction” function in the Measure function of ImageJ.

Quantification of the Wnt/β-catenin signals

The brain vascular Wnt/β-catenin signals were measured using Wnt/β-catenin reporter Tg(7XTCF-Xla.Siam:GFP; fli1:DsRed) zebrafish line. The midbrain vasculature was traced using the Simple Neurite Tracer plugin in ImageJ to measure the total length of vasculature and the TCF-positive vessels. To assess the relative extent of brain vascular TCF signaling, we calculated the ratio of TCF-positive vessel length to the total vascular length.

Analysis of Ca²⁺ imaging data

Ca²⁺ imaging data were analyzed with ImageJ software (NIH)⁵⁶. The mean greyscale value of each region of interest (ROI) in each frame was defined as F. The average value of the first 20 frames of ascending ordered F was defined as F₀. The calcium activity in neurons and ETCs was then calculated as ΔF/F0, where ΔF = F − F₀. A value exceeding two times the standard deviation was identified as a Ca²⁺ transient. The minimum duration of Ca²⁺ transients was 10 seconds. A Ca²⁺ transient occurring in the soma and all branches of an ETC was defined as a global Ca²⁺ transient, meanwhile a Ca²⁺ transient occurring only in the soma or in the branch of an ETC was defined as a local Ca²⁺ transient.

Bath and local puff application of drugs

To manipulate NMDAR channel functions, the antagonist MK-801 (10 µM, Merck) and D-AP5 (300 µM, Merck), or the agonist NMDA (10 µM, Merck) were dissolved in 10% Hank’s solution and bath-applied to larvae from 1.5 to 7 dpf. The medium was refreshed daily. To inhibit ERK and CAMKII activities, the inhibitors PD98059 (40 µM, IMA1002, Gene Operation) and KN93 (10 µM, 422711, Calbiochem) were bath-applied to 4-dpf larvae for 10–12 h, respectively. To activate VEGF signaling in ETCs, local puffing of VEGF (20 μg/mL, R&D Systems) was performed by using a micropipette (tip diameter, 1 μm) positioned near the target ETC, introduced from the contralateral optic tectum. The VEGF solution in the pipette was ejected using gas pressure pulses (3 psi, 50-ms duration, 1-second intervals, 10 pluses) delivered by a Picospritze III (Parker Picospritzer). To inhibit VEGF signaling during ChrimsonR optogenetic stimulation, Ki8751 (1 μM, 2542, Tocris) was bath-applied to zebrafish larvae for 30 minutes before imaging.

Application of optogenetic activation, moving bar, and electrical stimulations

For experiments involving ChrimsonR-based optogenetic activation, larvae were maintained under darkness. Optogenetic activation was implemented through LED at 580 − 620 nm or laser beams at 561 nm. For LED stimulation, a 10-w LED was controlled by the Master 8 (A.M.P.I.), and positioned 7 cm away from the larva in the “Opto. stim.” group and 5 cm away in the “Opto. stim. + MK-801” group to evoke comparable levels of Ca²⁺ activities in the optic tectum across groups. Each LED stimulation pulse lasted for 2 seconds, with a 30-second interval between pulses. To monitor brain vasculature development, LED stimulation lasted for 8 − 10 h per day during the daytime on Tg(elavl3:ChrimsonR-tdTomato);Tg(kdrl:eGFP) larvae from 3 to 6 dpf. Zebrafish larvae were maintained until 7 dpf for brain vasculature imaging. To monitor the growth capability of ETCs, LED stimulation lasted for 8 − 10 h, imaging was taken before and after the stimulation on Tg(elavl3:ChrimsonR-tdTomato);Tg(kdrl:lyn-sfGFP) larvae at 4–4.5 dpf. For 561-nm laser stimulation, photoactivation was performed focally at a single pixel point in the posterior area of the optic tectum (region for soma of neurons) with a laser power of 0.6 mW. Each 561-nm laser stimulation lasted 1 h with a 30-second interval between cycles, and each cycle consisted of 3 pulses with a duration of 400 ms and an interval of 10 ms. Tg(elavl3:ChrimsonR-tdTomato);Tg(kdrl:GCaMP6s) and Tg(elavl3:ChrimsonR-tdTomato);Tg(kdrl:lyn-sfGFP) larvae at 4 dpf were used to examine Ca²⁺ activities and growth capability of ETCs, respectively. Tg(elavl3:ChrimsonR-tdTomato);Tg(elavl3:GCaMP6s) larvae at 5 dpf were used to examine neural activities.

For moving bar stimulation, larvae were reared in darkness after fertilization. During stimulation, larvae were placed in a transparent 3-cm dish, which was then set inside a transparent 10-cm glass dish covered with aluminium foil on top. Then, the dish was positioned 10 cm above an LCD screen. A white moving bar was projected from the LCD screen, with dimensions of 20 cm in length and 0.5 cm in width, and a moving speed of 2 cm per second. The dish was centered within the white bar moving area. To assess the effectiveness of the moving bar stimulation, calcium imaging was performed on Tg(elavl3:GCaMP6s) larvae at 5 dpf. To monitor brain vascular development, Tg(kdrl:eGFP) larvae were exposed to moving bar stimulation from 4.5 to 6 dpf, while larvae in the control group were placed in the same area but shielded from the stimulation by aluminium foil covering both sides of the glass dish. After the moving bar stimulation, brain vascular imaging was conducted immediately. To monitor ETC development, Tg(kdrl:lyn-sfGFP) zebrafish larvae were exposed to the moving bar stimulation for 8 h per day during the daytime from 4 − 4.5 dpf. Brain ETCs of zebrafish larvae were imaged before and after the moving bar stimulation.

For electrical stimulation, larvae were placed in egg water (60 μg/ml ocean salt) within a rectangular container with 5 cm in length, 0.8 cm in width, and 0.5 cm in depth. Parallel aluminium foil sheets were attached to the two long sides of the container, with wires connected to each sheet. Electrical stimulation was provided using a YC-2 programmable electrical stimulator (Chengdu Instrument Factory, China). Each experiment consisted of 5 electrical stimuli, with a 5-minute interval between them. Each stimulus consisted of 100 pulses with a duration of 0.5 ms and an interval of 10 ms. Following stimulation, the absence of lethality and the presence of locomotor activities in larvae were used to assess the reliability of the electrical stimulation protocol. To examine the efficiency of the electrical stimulation, calcium imaging was performed on Tg(elavl3:GCaMP6s) larvae at 5 dpf. To monitor brain vascular development, electrical stimulations were administered to Tg(kdrl:eGFP) zebrafish larvae every 12 h from 2 to 7 dpf. Control larvae were placed in the container without receiving any electrical stimulation. After the electrical stimulation, brain vascular imaging was conducted at 7 dpf.

Photolysis of NP-EGTA/Ca²⁺ complex

NP-EGTA AM (6637, Setareh Biotech) was dissolved in DMSO to obtain 16 mM stock solution. The working solution containing 4 mM NP-EGTA and 0.05% Pluronic™ F-127 (P3000MP, Invitrogen) in ES was prepared before the experiment. To induce global Ca²⁺ activity in ETCs, the NP-EGTA working solution was locally puffed into the region neighboring the targeted ETC in Tg(kdrl:GCaMP6s) zebrafish at 3–4 dpf⁵². The puffing of the NP-EGTA working solution was repeated with an 8-min interval with gas pressure (3 psi, 50-ms duration, 1-second intervals, 4 pluses) throughout the experiment. Local photolysis of the NP-EGTA/Ca²⁺ complex was achieved by 405-nm laser scanning of around 4 × 4 μm regions of interest at both branch and soma sites at 0.3 − 0.5 mW. Each photolysis stimulation consisted of repeated sequential scanning of each square (10 μs per pixel) for a total duration of 1.5 − 2 s. The laser power and stimulation duration were titrated for each ETC based on its dorsoventral position and responsiveness to ensure consistent Ca²⁺ uncaging. The experimental workflow was as follows: (1) high-resolution morphological snapshot before uncaging; (2) time-lapse imaging for spontaneous Ca²⁺ activity for 20 min; (3) high-resolution morphological snapshot at the onset of uncaging; (4) time-lapse Ca²⁺ activity imaging and uncaging with two sessions lasting around 60 min in total; and (5) high-resolution morphological imaging after uncaging. During uncaging, a total of 10 times of 405-nm laser stimulation were applied to each ETC with an interval of 4 min. The cases with ≥ 2 Ca²⁺ transient responses evoked by NP-EGTA uncaging were used for further analysis. For analysis, the frequency of global Ca²⁺ transients were calculated as induced and spontaneous global Ca²⁺ transient number divided by the actual Ca²⁺ imaging duration (Global Ca²⁺ transient (/h)). Growth rate was determined as the change in ETC length divided by the exact time interval between high-resolution morphological snapshots, which included breaks during and after uncaging sessions (Growth rate (µm/h)). In control groups, larvae were treated with DMSO only or treated with NP-EGTA followed by mock 405 nm laser application.

In vivo whole-cell recording

Zebrafish larvae were first paralyzed with α-bungarotoxin (100 μg/ml, Merck) and then embedded dorsal side up in a glass-bottomed chamber by 1.2% low melting-point⁹⁵. A cut of skin above the ventricle of the optic tectum was made by a micropipette, creating a small hole for subsequent insertion of recording micropipettes. After the dissection, the larva was transferred to a recording setup and perfused with an extracellular solution containing (in mM): 134 NaCl, 2.9 KCl, 2 CaCl₂,2 MgCl₂,10 HEPES, and 10 glucose (290 mOsmol/l, pH 7.8). The micropipette was pulled to resistances of 20 − 30 MΩ by a Flaming/Brown P97 puller (Sutter Instruments) and filled with an internal solution containing (in mM): 100 K-gluconate, 10 KCl, 2 CaCl₂, 2 Mg-ATP, 2 Na-GFP, 10 HEPES, and 10 EGTA (280 mOsmol/l, pH 7.35). The equilibrium potential of Cl^- was about −60 mV according to the Nernst equation. Recordings were performed with an EPC-10 amplifier (Heka, Germany), with signals filtered at 2.9 kHz and sampled at 10 kHz. Data were discarded if the series resistance (20 − 100 MΩ) varied >20% or the membrane potential was above −40 mV during the recording. To record NMDA-induced currents, a brief puff of NMDA (100 mM, 100 ms) was focally applied near the soma of tectal neurons via a micropipette with 2-μm tip opening through an air pressure (10 psi), which was controlled by Picospritzer III (Parker Instrumentation). The peak amplitude of NMDA-induced currents was measured under different holding potentials ranging from −100 to +90 mV (with an increment of 10 mV), and then plotted against the holding potential to get the I-V curve.

Quantification and statistical analysis

All statistical analyses were performed using GraphPad Prim (V9.5). The normality of data was first examined with the Kolmogorov-Smirnov test or the Shapiro-Wilk test. For normally distributed data, the significance of difference between the data obtained from two groups, or from the same group before and after manipulations, was determined by using unpaired or paired two-tailed Student’s t-test, respectively. In multiple comparisons, One-way ANOVA test was used. For non-normally distributed data, two-tailed Mann-Whitney test or two-tailed Wilcoxon signed rank test was used for inter-group or intra-group comparison, respectively. Simple Linear Regression Analysis was utilized in this study to evaluate the linear relationship between the variables. The p value less than 0.05 was considered to be statistically significant. All results were represented as mean ± s.e.m.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Author contributions

J.L.D. supervised the project. J.L.D., T.T.L., S.Y.G., H.Y.L. and B.X. designed experiments. T.T.L. and J.L.D. wrote the manuscript with contributions from B.X., S.Y.G., H.Y.L. and J.A.Y. S.Y.G. performed most of the experiments except otherwise stated. H.Y.L. performed experiments of grin1b CKO, optogenetic stimulation and endothelial tip cells. L.S. performed photolysis of NP-EGTA/Ca²⁺ complex. T.T.L. performed experiments on grin1b NQ/CT and molecular pathway investigations. R.W.Z. performed electrical recordings. B.X. and Y.Z. performed blood-brain barrier-related experiments. L.S. and J.L.Z. performed neuronal activity analysis. X.D.Z., Q.C., S.G.Y. and J.H. constructed the transgenic zebrafish lines and provided essential reagents.

Peer review

Peer review information

Nature Communications thanks Dritan Agalliu, Phillip O’Herron and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Data availability

All data supporting the findings of this study, including raw images, are available either within the Article and its Supplementary Information files or from the corresponding authors upon request. Source data are provided with this paper. The zebrafish line generated in this study is preserved at the Center for Excellence in Brain Science and Intelligence Technolog (CEBSIT), and is available from the corresponding authors upon request and signature of a Material Transfer Agreement between the CEBSIT and the requestor. Due to file sizes, raw imaging data are also available on request, and requests will be processed and fulfilled within four weeks. Source data are provided with this paper.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-025-66543-9.