A brain-specific angiogenic mechanism enabled by tip cell specialization

Zebrafish strains and husbandry

Zebrafish (Danio rerio) were maintained at 28 °C under a 14 h–10 h light–dark cycle and raised under standard conditions in a certified animal facility (LA1500474) in accordance with European and national ethical and animal welfare guidelines. All of the animal procedures were approved by the corresponding ethical committee (Commission d’Ethique et du Bien Être Animal (CEBEA), Université libre de Bruxelles, protocol approval numbers: CEBEA-IBMM-2016:65 and CEBEA-07 GOS IBMM). Zebrafish staging was performed as described previously⁵¹. The following published transgenic and mutant lines have been used in this study: Tg(kdrl:EGFP)^s843 (ref. ⁵²), Tg(kdrl:ras-mCherry)^s896 (ref. ⁵³), Tg(7xTCF-Xla.Siam:GFP)^ia4 (ref. ⁵⁴), Tg(fli1a:Gal4FF)^ubs3 (ref. ⁵⁵), Tg(UAS:Kaede)^rk8 (ref. ⁵⁶), Tg(UAS:GCaMP7a)^zf415 (ref. ⁵⁷), Tg(gata1:DsRed)^sd2 (ref. ⁵⁸), gpr124^s984 (ref. ¹¹), wnt7aa^ulb2 (ref. ¹⁷), reck^ulb3 (ref. ⁵⁹), kdrl^hu5088 (ref. ⁶⁰) and col4a5^s510 (ref. ⁴⁵). The mmp25a^ulb26 and mmp25b^ulb27 alleles were generated in this study using CRISPR–Cas9 mutagenesis. All of the zebrafish experiments were performed on embryos and larvae younger than 5 days post-fertilization, before they became capable of independent feeding.

Mice

Mice were housed at 20 °C under a 12 h–12 h light–dark cycle under standard conditions and were maintained in a certified animal facility (LA1500474) in accordance with European and national ethical and animal welfare guidelines. The relative ambient humidity level ranged from 45 to 65%. All animal procedures were approved by the corresponding ethical committee (Commission d’Ethique et du Bien Être Animal (CEBEA), Université libre de Bruxelles, protocol approval number: CEBEA-08 GOS IBMM). Mice were maintained on the C57BL/6J background and, for experiments, mice of both sexes were used. BAT-GAL reporter (B6.Cg-Tg(BAT-LacZ)3Picc/J) mice⁶¹ and Mmp25-knockout mice²³ were provided by S. Piccolo and C. López-Otín, respectively. Vascular networks were quantified as the number of CNS-invading sprouts in the E10.5 midbrain and forebrain in five consecutive 60 μm sections, and as the organ surface-normalized vascular density (length or surface, depending on the vascular morphologies) in 60 μm sections of E10.5 forelimbs and E12.5 intestine, stomach, liver and lung.

CRISPR–Cas9-mediated gene disruption in zebrafish

Germline zebrafish mmp25a^ulb26 and mmp25b^ulb27 alleles were generated using CRISPR–Cas9 as described previously⁶². Target sites were selected using CRISPOR (v.5.01)⁶³. The following primers were annealed and cloned into the pT7-gRNA vector (Addgene, 46759): 5′-TAGGGGCAATGCCCTGCGAGTG-3′ and 5′-AAACCACTCGCAGGGCATTGCC-3′ for mmp25a; 5′-TAGGGGACAGCTACAGAGCAAAGA-3′ and 5′-AAACTCTTTGCTCTGTAGCTGTCC-3′ for mmp25b. sgRNAs were synthesized by in vitro transcription (HiScribe T7 Quick High Yield RNA Synthesis Kit; New England Biolabs) from BamHI-linearized pT7-gRNA vectors. Mmp25a was targeted in exon 4 (catalytic domain) and mmp25b was targeted in exon 2 (pro-domain). Synthetic capped zCas9 mRNA was transcribed from the XbaI-linearized pT3TS-nls-zCas9-nls vector (Addgene, 46757) using the mMESSAGE mMACHINE T3 Kit (Ambion). Co-injection of the sgRNAs (30 pg each) and nls-zCas9-nls mRNA (150 pg) was performed at the one-cell stage.

For somatic gene disruptions, two sgRNAs targeting the same exon were synthesized using the following primer pairs: mmp2 sgRNA1: 5′-TAGGGGGAACTTTATGATGGGTG-3′ and 5′-AAACCACCCATCATAAAGTTCCC-3′; mmp2 sgRNA2: 5′-TAGGGGAACTTTATGATGGGTGA-3′ and 5′-AAACTCACCCATCATAAAGTTCC-3′; mmp14b sgRNA1: 5′-TAGGCCAGTCCATTTGATGGAGA-3′ and 5′-AAACTCTCCATCAAATGGACTGG-3′; mmp14b sgRNA2: 5′-TAGGATTCCCTGGGAAGTAAGCAT-3′ and 5′-AAACATGCTTACTTCCCAGGGAAT-3′; mmp25a sgRNA1: 5′-TAGGGGCAATGCCCTGCGAGTG-3′ and 5′-AAACCACTCGCAGGGCATTGCC-3′; mmp25a sgRNA2: 5′-TAGGGTCTGGTGAGGCTTATTTT-3′ and 5′-AAACAAAATAAGCCTCACCAGAC-3′; mmp25b sgRNA1: 5′-TAGGTAGGACTGGTTGAGCCGGTA-3′ and 5′-AAACTACCGGCTCAACCAGTCCTA-3′; mmp25b sgRNA2: 5′-TAGGAGGAGGCAGATATCCATAC-3′ and 5′-AAACGTATGGATATCTGCCTCCT-3′; lama1 sgRNA1: 5’-TAGGGAACGGCCGTCAGTTCCACT-3′ and 5′-AAACAGTGGAACTGACGGCCGTTC-3′; lama1 sgRNA2: 5′-TAGGCGGACTCTGCCACCACAGGT-3′ and 5′-AAACACCTGTGGTGGCAGAGTCCG-3′; lama1 sgRNA1-scrambled: 5′-TAGGGAACGGCCGTCAGTTACCTC-3′ and 5′-AAACGAGGTAACTGACGGCCGTTC-3′; lama1 sgRNA2-scrambled: 5′-TAGGCGGACTCTGCCACCGATGAC-3′ and 5′-AAACGTCATCGGTGGCAGAGTCCG-3′; lama2 sgRNA1: 5′-TAGGCGCAGACAGGCTCCGGTCAG-3′ and 5′-AAACCTGACCGGAGCCTGTCTGCG-3′; lama2 sgRNA2: 5′-TAGGTCAGCGGGTCACAGCTCAG-3′ and 5′-AAACCTGAGCTGTGACCCGCTGA-3′. lama2 sgRNA1-scrambled: 5′-TAGGCGCAGACAGGCTCCACGGGT-3′ and 5′-AAACACCCGTGGAGCCTGTCTGCG-3′; lama2 sgRNA2-scrambled: 5′-TAGGTCAGCGGGTCACATGCAGC-3′ and 5′-AAACGCTGCATGTGACCCGCTGA-3′; col4a1 sgRNA1: 5′-TAGGATAGGTCCTGGCGGTCCGGG-3′ and 5′-AAACCCCGGACCGCCAGGACCTAT-3′; col4a1 sgRNA2: 5′-TAGGCAGGTCCCAAAGGAACTGAT-3′ and 5′-AAACATCAGTTCCTTTGGGACCTG-3′; col4a2 sgRNA1: 5′-TAGGTGGCAGTCCCGGATCTCCAG-3′ and 5′-AAACCTGGAGATCCGGGACTGCCA-3′; col4a2 sgRNA2: 5′-TAGGAGGTTTGGATGGAGCTTCAG-3′ and 5′-AAACCTGAAGCTCCATCCAAACCT-3′; col4a3 sgRNA1: 5′-TAGGAAGGTTGTGCTGGGGTTCA-3′ and 5′-AAACTGAACCCCAGCACAACCTT-3′; col4a3 sgRNA2: 5′-TAGGAAGGATTCCCAGGATTGTGT-3′ and 5′-AAACACACAATCCTGGGAATCCTT-3′; col4a4 sgRNA1: 5′-TAGGTGGGTCGACAGGGCCCCCAG-3′ and 5′-AAACCTGGGGGCCCTGTCGACCCA-3′; col4a4 sgRNA2: 5′-TAGGAGAACCTTGGGGCCCCTGG-3′ and 5′-AAACCCAGGGGCCCCAAGGTTCT-3′; col4a5 sgRNA1: 5′-TAGGCCTGGGAAACCTGGAACACC-3′ and 5′-AAACGGTGTTCCAGGTTTCCCAGG-3′; col4a5 sgRNA2: 5′-TAGGCCGGGTTTAAAGGGTCAGCC-3′ and 5′-AAACGGCTGACCCTTTAAACCCGG-3′; col4a6 sgRNA1: 5′-TAGGCTTGGACCAGTGGGCAGCGG-3′ and 5′-AAACCCGCTGCCCACTGGTCCAAG-3′; col4a6 sgRNA2: 5′-TAGGATGGGGGCCCGGGACCAGTT-3′ and 5′-AAACAACTGGTCCCGGGCCCCCAT-3′; serpina1 sgRNA1: 5′-TAGGTGCTGCCTTGCTGGTAGCAA-3′ and 5′-AAACTTGCTACCAGCAAGGCAGCA-3′; serpina1 sgRNA2: 5′-TAGGCTGGTAGCAACGGCCTGGG-3′ and 5′-AAACCCCAGGCCGTTGCTACCAG-3′.

The efficiency of somatic gene disruption was scored by high-resolution melt analysis (HRMA) using the Illumina Eco real-Time PCR system, and further characterized using Illumina amplicon deep sequencing (Azenta Life Sciences).

Genotyping

Zebrafish gpr124^s984, wnt7aa^ulb2, kdrl^hu5088, reck^ulb3 and col4a5^s510 and mouse Mmp25 alleles were genotyped as described previously^{11,17,23,45,59,60}. The mmp25a^ulb26 and mmp25b^ulb27 alleles were genotyped by high-resolution melt analysis (Eco Illumina real-time PCR system) using the following primers: 5′-TTTCCACCTCCCTCAGTGTC-3′ and 5′-GTGGAAACGCAGAGGTGTGT-3′ for mmp25a; 5′-CGCACAGGACAGCTACAGAG-3′ and 5′-CTGCATTTCTCTAATGGCTCTCTCG-3′ for mmp25b.

MO, RNA and DNA microinjection in zebrafish

MOs targeting gpr124 (4 ng; splice blocking; ACTGATATTGATTTAACTCACCACA)¹¹, reck (0.4 ng; splice blocking; CAGGTAGCAGCCGTCACTCACTCTC)⁶⁴, wnt7aa (4 ng; splice blocking; TTCCATTTGACCCTACTTACCCAAT)¹⁷, lama1 (0.5 ng; translation blocking; ATCTCCATCATCGCTCAAACTAAAG), lama2 (1 ng; translation blocking GCCACTAAACTCCGCGTGTCCATGT), lama4 (0.5 ng; translation blocking; GCCATGATTCCCCCTGCAACAACTT), lama5 (0.25 ng; translation blocking; CTCGTCCTGATGGTCCCCTCGCCAT)⁶⁵, lamb1a (0.125 ng; translation blocking; TATTTCCAGTTTCTTTCTTCAGCGG), lamc1 (0.125 ng; translation blocking; TGTGCCTTTTGCTATTGCGACCTC)⁶⁶, col4a1 (1 ng; translation blocking; ACACATGGAAGCCGCATCTTCACAC)⁶⁷, col4a2 (2 ng; translation blocking; TTCTCACCCTCCATGCGAGCCTAAA), col4a5 (2 ng; translation blocking; ATGTTCCTCTGTTAAGCTAACTGCA), col4a6 (2 ng; translation blocking; AGGTAAAGTAGGCTATCCTCCTCGT) were obtained from Gene Tools and were injected at the zygotic stage at the indicated doses. Injection of a standard control MO (CCTCTTACCTCAGTTACAATTTATA, up to 8 ng) did not affect the brain vasculature.

Transgenic mosaic endothelial overexpression was achieved by co-injecting at the one-cell stage 25 pg of Tol2 transposase mRNA and 25 pg of the pTol2-fli1a:kdrl-2A-nls-mtagBFP2, pTol2-fli1a:mmp25b-2A-tagRFP, pTol2-fli1a:mmp25b^ΔZn2+-BD-2A-tagRFP or pTol2-fli1a:mmp25b^{Zn2+-BDH237A,H241A,H247A}-2A-tagRFP constructs⁶⁸.

Capped mRNAs were transcribed in vitro from NotI-linearized pCS2⁺ constructs, using the mMessage mMachine SP6 Kit (Thermo Fisher Scientific) and injected at the one-cell stage at a dose of 200 pg. The fragment encoding the Zn²⁺-binding domain (Zn²⁺-BD; His237–His247) was deleted in the ΔZn²⁺-BD mmp25b variant. Three histidines, essential for Zn²⁺ chelation, were substituted by alanines in the Zn²⁺-BD^{H237A, H241A, H247A} variant, abbreviated as Zn²⁺-BD^H-A. In the Pro⁻ mmp2 mRNA variant, the sequences encoding the prodomain (Ala30–Val107) were deleted. The sequences corresponding to the GPI-anchoring site of Mmp25b (Ser658–Gln697) were fused 3′ to the mmp2 ORF in the GPI⁺ mmp2 variant.

Transplantations

Host Tg(kdrl:ras-mCherry)^s896 and donor Tg(kdrl:EGFP)^s843 embryos were dechorionated with pronase (Millipore, 53702; 1 mg ml⁻¹) during 5 min at 28 °C in 1/3 Ringer solution, supplemented with penicillin (50 U ml⁻¹) and streptomycin (50 µg ml⁻¹). The embryos were subsequently incubated on agarose-coated dishes in the same medium. At the mid-blastula stage, 20 to 50 donor cells were transplanted into the blastoderm margin of stage-matched host embryos. After transplantation, embryos were incubated until the indicated stages. After assessing the contribution of EGFP⁺ transplanted cells using the Leica M165 stereomicroscope, mosaic vessels were recorded using time-lapse confocal microscopy. The contribution of cells of a defined genotype to the TC position was calculated as the fraction of the total number of mosaic vessels (CtAs or ISVs). The contribution to TC position in intraneural secondary branches was scored as the fraction of the stalk cell genotype in the initial brain-invading CtA.

Immunofluorescence and in situ hybridization

Zebrafish and mouse embryos were fixed in 4% paraformaldehyde (PFA) in PBS. For sections, embryos were washed in PBS and equilibrated in 30% sucrose in PBS (w/v) overnight at 4 °C. The embryos were then mounted in 7.5% gelatin (w/v), 15% sucrose (w/v) in PBS and stored at −80 °C. Zebrafish and mouse embryos were cut into 20 and 60 µm frozen sections, respectively, using the Leica CM1850 Cryostat (Leica) at −30 °C.

For immunofluorescence staining, the sections were washed three times with PBS Triton X-100 (0.4%; PBST) for 5 min, blocked using blocking buffer (PBST, 5% goat serum) for 1 h and then incubated with primary antibodies in blocking buffer solution overnight at 4 °C. After three washing steps in PBST for 5 min, the sections were exposed to secondary antibodies diluted in blocking buffer containing 0.001% DAPI overnight at 4 °C. After three washing steps in PBST for 5 min, the sections were mounted in DAKO fluorescence mounting medium (Agilent, S3023). The following primary antibodies and lectin were used: rabbit anti-laminin-111 (Merck, L9393, 1:250, used for zebrafish immunostaining, polyclonal immunization with an Engelbreth–Holm–Swarm mouse sarcoma extract), rat anti-laminin-111 (R&D systems, MAB4656, 1:250, used for mouse immunostainings, monoclonal reactivity towards LAMA1/B1), chicken anti-GFP (Aves Labs, GFP-1020, 1:200), rabbit anti-collagen type IV (Sigma-Aldrich, AB756P, 1:300), chicken anti-β-galactosidase (Abcam, ab9361, 1:300), anti-Erg1-Alexa Fluor (AF) 647 conjugate (Abcam, ab196149, 1:250) and isolectin B4-AF594 conjugate (Thermo Fischer Scientific, I21413, 1:200). The following secondary antibodies were used: goat anti-chicken AF488 (Thermo Fischer Scientific, A11039, 1:500), goat anti-rabbit AF594 (Thermo Fischer Scientific, A11012, 1:500), and donkey anti-rat AF647 (Thermo Fischer Scientific, A48272, 1:500).

For in situ hybridization, digoxigenin (DIG)-labelled antisense riboprobes were produced by in vitro transcription using the DIG RNA labelling kit and SP6 RNA polymerase (Roche). The templates were amplified from 48 hpf WT embryo cDNA, and cloned into NcoI/SacI-digested pGEMT using the following primers: kdrl: 5′-GCATGCTCCCGGCCGCCATGGTGGCAGGATTCACTTTGAGTGG-3′ and 5′-CATCCAACGCGTTGGGAGCTCTAGTGTAGGGCTCAATCCGCAG-3′; mmp25b: 5′-ATGAGTTTCTCAGGATATCTTGGTCTGG-3′ and 5′-TTATTGCGAGTTGAAGCCAATATGAAGC-3′; mmp14b: 5′-GCATGCTCCCGGCCGCCATGGTGGATGCAGCTCTTCTCTACACG-3′ and 5′-CATCCAACGCGTTGGGAGCTCCATGAGGCTGCTGGAAATGTGC-3′; mmp2: 5′-GCATGCTCCCGGCCGCCATGGTGCTCACACAGACAAAGAAGTGG-3′ and 5′-CATCCAACGCGTTGGGAGCTCTTTCCTGACATCAGCCGTCC-3′; mmp9: 5′-GCATGCTCCCGGCCGCCATGGCAAATCTGTGTTCGTGACGTTTCC-3′ and 5′-CATCCAACGCGTTGGGAGCTCCTCCTTGATTTGGCAGGCATCG-3′; lama1: 5′-GCATGCTCCCGGCCGCCATGGGTCACAACAAAGCCGACGACTG-3′ and 5′-CATCCAACGCGTTGGGAGCTCTGAGCGTTCCCTCAGCGCTGT-3′; col4a1: 5′-GCATGCTCCCGGCCGCCATGGGGTTCTAAGGGTGAAGGAGGTG-3′ and 5′-CATCCAACGCGTTGGGAGCTCCCCTCTTCATGCACACTTGAC-3′; col4a2: 5′-GCATGCTCCCGGCCGCCATGGCCTAAAGGAGATACCGGACCC-3′ and 5′-CATCCAACGCGTTGGGAGCTCCTACAGGTTCTTCATGCACAC-3′; col4a3: 5′-GCATGCTCCCGGCCGCCATGGGGACAAAAAGGACAGTGTGGTC-3′ and 5′-CATCCAACGCGTTGGGAGCTCGCAAGGTCACCTTGAGGCTGTTG-3′, col4a4: 5′-GCATGCTCCCGGCCGCCATGGCTGGGTCCCAGTGGTGCAAAAG-3′ and 5′-CATCCAACGCGTTGGGAGCTCCATTGGTTGGGGTCATTCATC-3′; col4a5: 5′-GCATGCTCCCGGCCGCCATGGGGTTTTCCAGGATCTAAAGGAG-3′ and 5′-CATCCAACGCGTTGGGAGCTCCGTCCTCTTCATACACACCAC-3′; col4a6: 5′-GCATGCTCCCGGCCGCCATGGCGTCCAGGAATAATAGGACC-3′ and 5′-CATCCAACGCGTTGGGAGCTCCTACAAGATCTTCATGCAGAC-3′; slc2a1a: 5′-GCATGCTCCCGGCCGCCATGGCAACTTGGCATTGTCATTG-3′ and 5′-CATCCAACGCGTTGGGAGCTCGGCTGTGATCTCTTCAAACG-3′; slc16a1a: 5′-GCATGCTCCCGGCCGCCATGGATGCCTCCAGCAACAGGAGG-3′ and 5′-CATCCAACGCGTTGGGAGCTCCTATACGACTCCATCTGCCTCCTTTT-3′; fabp11a: 5′-GCATGCTCCCGGCCGCCATGGGATCAAATCTCAATTTACAGCTGTTG-3′ and 5′-CATCCAACGCGTTGGGAGCTCTTCAAAGCACCATAAAGACTGATAAT-3′. Whole-mount chromogenic in situ hybridizations were performed as previously described⁶⁹ using anti-DIG-AP antibodies (Merck, 11093274910, 1:10,000). Combined immunostainings and FISH were performed as previously described⁷⁰, using anti-DIG POD antibodies (Merck, 11207733910, 1:1,000) and the TSA Plus Cy3 detection kit (Akoya Biosciences, NEL744001KT).

Photoconversion and FACS isolation of zebrafish brain ECs

Photoconversion of Tg(fli1a:Gal4FF)^ubs3;(UAS:Kaede)^rk8 PHBC or CtA ECs was performed using the Zeiss LSM710 confocal microscope (Carl Zeiss, objective lenses: Plan-Apochromat ×20/0.8 M27), as described previously⁷¹. In brief, anaesthetized embryos were mounted laterally in 1% low-melting-point agarose and the fluorescent Kaede protein was photoswitched by scanning the selected region of interest (ROI) using a 405 nm laser (five iterations of 50 s). After isolation from the agarose, the embryos were washed in Ca²⁺/Mg²⁺-free Hank’s Balanced Salt Solution (HBSS, Gibco) and dissociated at 28.5 °C for 30 min in TrypLE select (Thermo Fischer Scientific, 12563011). Dissociation was stopped by the addition of FBS and centrifugation. The cell pellet was resuspended in HBSS containing Ca²⁺/Mg²⁺ and 5% FBS, filtered and submitted for FACS analysis (BD Biosciences FACSAria III).

For scRNA-seq analyses, single photoswitched (red fluorescent) WT ECs were distributed in individual wells of 384-well plates containing 2.3 µl of Smart-seq2 lysis buffer (0.2% Triton X-100, 2 U µl⁻¹ RNase inhibitor, 2 mM dNTP mix and 1 µM Smart-seq2 primer (5′-AAGCAGTGGTATCAACGCAGAGTACT30VN-3′). The plates were stored at −80 °C before mRNA-seq using the Smart-Seq2 protocol⁷² and analysis using the Seurat v4 toolkit in Rstudio (v.1.1.463)⁷³. In brief, single-cell fastq files were demultiplexed by applying standard parameters of the Illumina pipeline (bcl2fastq v.2.19.0.316) using Nextera XT index kit v2 adapters. Mapping was performed to the zebrafish reference genome build GRCz11, with TopHat v.2.1.1 and Bowtie1 or Bowtie2 option. Adapter sequences were removed using Trim Galore v.0.4.4 before read mapping and doublets were removed using Samtools v.1.16.1 software. The generated BAM files containing the alignment results were sorted according to the mapping position, and raw read counts for each gene were calculated using the FeatureCounts function from the Subread package v.1.4.6-p5. For technical control, 92 ERCC RNAs were included in the lysis buffer and in the mapping.

For bulk RNA-seq analyses, Tg(fli1a:Gal4FF)^ubs3;(UAS:Kaede)^rk8 embryos were injected, or not, at the one-cell stage with gpr124, reck or wnt7aa MOs and PHBC ECs were isolated at 30 hpf as described above. Alternatively, embryos were treated with IWR-1 from 26 hpf onwards and CtA ECs were photoconverted and sorted at 36 hpf, as described above. Photoconverted PHBC ECs of 80 embryos were pooled and submitted for RNA extraction and RNA-seq, as previously described⁷¹. Transcriptomes were analysed and compared using DESeq2 (v.1.12)⁷⁴.

Light microscopy image acquisition and processing

All images were acquired using the Leica M165 stereomicroscope, the Zeiss LSM710 or the Zeiss LSM900 confocal microscope equipped with the Leica Application Suite (LAS) v.4.2 or ZEN Blue v.3.1 microscopy software. Image analysis was performed using ImageJ v.1.53c. Zebrafish embryos were imaged live or after fixation in 4% PFA in PBS overnight at 4 °C. Mouse embryos were fixed (4% PFA in PBS), and stained after sectioning. Live imaging of dechorionated zebrafish embryos was performed after embryo immobilization with a low dose of tricaine in low-melting-point agarose (1% in E3 zebrafish medium supplemented with N-phenylthiourea and tricaine) in a glass-bottom Petri dish (MatTek Corporation). Confocal time-lapse images were recorded at a stable temperature of 28.5 °C, using an incubation chamber. Ca²⁺-oscillations were recorded by time-lapse imaging of Tg(fli1a:Gal4FF);(UAS:GCaMP7a) embryos, taking a z stack every 5 s during the 30 min before CtA sprouting (31–31.5 hpf). Circular ROIs (<5 µm diameter) were centred on oscillating PHBC ECs. F/F₀ was calculated to quantify changes in fluorescence, where F₀ is the baseline fluorescence. Ca²⁺ spikes were identified as events of F/F₀ ≥ 1.5.

For angiography, imaging was performed 1 h after injection of 1 nl of tetramethylrhodamine dextran 2,000,000 Da molecular mass (Thermo Fisher Scientific, D7139, 25 μg µl⁻¹ in PBS) in the heart of 72 hpf larvae using a micromanipulator. Tracer leakage assays were performed by injecting 1 nl of 150,000 Da FITC-labelled dextran (FD150S, 25 μg µl⁻¹ in PBS) intracardially and imaging 1 h after injection. Three-dimensional reconstructions were performed using the Imaris Filament Tracer software (Bitplane) before manual false-colouring to highlight extra- and intracerebral vessels exhibiting or not BBB properties.

Transmission electron microscopy

WT zebrafish embryos (32 hpf) were fixed overnight in 2.5% glutaraldehyde (Electron Microscopy Sciences), 4% PFA at 4 °C and post-fixed with 1% osmium tetroxide (Electron Microscopy Sciences) and 1.5% ferrocyanide (Electron Microscopy Sciences) in 0.15 M cacodylate buffer. The embryos were further stained with 1% uranyl acetate (Electron Microscopy Sciences), serially dehydrated and embedded in epoxy resin (Agar 100 resin; Agar Scientific). Resin blocks containing the processed embryos were trimmed to reach the ROI, which was evaluated by toluidine staining of thin sections (15 μm). Ultrathin 70 nm sections were then produced with a Leica EM UC6 ultramicrotome and mounted onto copper-Formvar-carbon grids (Electron Microscopy Sciences). Observations were made using the Tecnai 10 transmission electron microscope (FEI), and images were captured with a Veleta camera and processed using SIS iTEM v.5.1 software (Olympus).

Western blot analysis

Samples were denatured in Bolt LDS sample buffer and reducing agent (Thermo Fischer Scientific, B0007 and B0009) at 70 °C for 10 min. Gel electrophoresis was performed using 4–15% Mini-PROTEAN TGX Precast Protein Gels (Bio-Rad, 4561085). Proteins were transferred to nitrocellulose membranes. After blocking in 5% milk in Tris-buffered saline (TBS), the membranes were incubated with primary antibodies (1% milk in 0.05% Tween-20 TBS, TBST) overnight at 4 °C. After washing in TBST, membranes were incubated with secondary antibodies in 1% BSA in TBST, for 1 h at room temperature. Blots were revealed using Western Lightning Plus ECL (PerkinElmer, NEL103001EA).

The following primary antibodies were used: rabbit anti-HA (Merck, H6908, 1:1,000), chicken anti-GFP (Aves Biolabs, GFP-1020, 1:10,000), rat anti-laminin-111 (R&D systems, MAB4656, 1:250, monoclonal reactivity towards LAMA1/B1). The following secondary antibodies were used: goat anti-rabbit IgG HRP conjugate (Promega, W401B, 1:5,000), goat anti-chicken IgY HRP conjugate (Thermo Fischer Scientific, A16054, 1:40,000) and rabbit anti-rat IgG HRP conjugate (Merck, A9542, 1:5,000). Uncropped blots are provided in Supplementary Fig. 1.

Recombinant protein expression and purification

The human MMP25 and MMP2 catalytic domains were amplified from HUVEC cDNA and the zebrafish Mmp25b catalytic domain was synthesized after codon optimization. The fragments were cloned into the NcoI and XhoI restriction sites of pET21d. The catalytic domains span residues Tyr113 to Gly284 of zebrafish Mmp25b (UniProtKB: E7F1N5), Tyr108 to Gly280 of human MMP25 (UniProtKB: Q9NPA2) and Tyr110 to Asp452 of human MMP2 (UniProtKB: P08253). BL21 (DE3) E. coli cells were transformed with pET21d-zMmp25b-6xhis, pET21d-hMMP25-6xhis or pET21d-hMMP2-6xhis and grown in 100–300 ml LB medium (supplemented with 100 µg ml⁻¹ ampicillin). Protein expression was induced with 1 mM isopropyl β-d-1-thiogalactopyranoside (IPTG) when the culture reached an optical density at 600 nm (OD₆₀₀) of 0.9. After overnight incubation at 37 °C under agitation, cells were collected by centrifugation (5,000g, 20 min, 4 °C) and frozen at −80 °C. After resuspension in 50 mM Tris (pH 8), cells were mechanically lysed on ice (Microfluidics, 110SCE, 3 cycles). Inclusion bodies were recovered from the lysate by centrifugation (16,000g, 20 min, 4 °C) and solubilized in 8 M urea, 50 mM Tris (pH 7.6), 150 mM NaCl, 5 mM CaCl₂ and 50 µm ZnCl₂. The insoluble fraction was removed by centrifugation (16,000g, 20 min, 4 °C) and the supernatant was incubated overnight with 100 µl of Ni⁺/nitrilotriacetic acid agarose beads (Qiagen) at 4 °C. The beads were washed with 20 mM imidazole in TBS 8 M urea and elution was performed with 500 mM imidazole in TBS 8 M urea. Recombinant protein purity was assessed by SDS–PAGE and Coomassie blue staining, and protein concentrations were measured by the BCA protein assay (Thermo Fischer Scientific, 23223). Catalytic domains were refolded by dilution (1/20, v/v) in 50 mM Tris, 150 mM NaCl, 5 mM CaCl₂, 50 µM ZnCl₂, 0.005% Brij-35 (Thermo Fischer Scientific, 20150) for 1 h at 12 °C. The insoluble fraction was removed by centrifugation (21,400g, 10 min, 4 °C). Uncropped gels are available in Supplementary Fig. 1.

Mmp25 cleavage assays

For α-1 antitrypsin, 2 µM of α-1 antitrypsin (Athens Research and Technology, 16-16-0011609) was incubated with 2 µM rzMmp25b or 75 nM rhMMP25 overnight at 28 °C and 37 °C, respectively, in 50 µl Mmp25 cleavage buffer (50 mM Tris (pH 7.6), 150 mM NaCl, 5 mM CaCl₂, 0.005% Brij-35 (Thermo Fischer Scientific, 20150)).

For laminin-111, 15 µg of Matrigel (Corning, 354230) was incubated overnight at 37 °C with 1 µM of rhMMP25 in 50 µl Mmp25 cleavage buffer. The samples were concentrated by acetone precipitation before SDS–PAGE and western blot analysis for LAMA/B1 (R&D systems, MAB4656).

For collagen IV, 20 µg of collagen IV purified from human placenta (Merck, C7521) was incubated with 1 µM rhMMP25 overnight at 37 °C in 50 µl Mmp25 cleavage buffer. The samples were concentrated by acetone precipitation before SDS–PAGE and Coomassie blue staining.

For recombinant HA-tagged Col4a5 expressed in HEK293T cells, Zebrafish col4a5 was amplified from 48 hpf zebrafish cDNA, cloned in fusion to a C-terminal HA tag into pCS2⁺ (digested with BamHI and XhoI) and transiently expressed using PEI (polyethylenimine) in HEK293T cells (ATCC CRL-3216, authenticated by ATCC STR profiling, tested negative for mycoplasma contamination). The empty pCS2⁺ was used as negative control. Then, 48 h after transduction, the cells were washed twice in PBS, before collection and cell disruption using a disposable grinding pestle in Mmp25 cleavage buffer. After centrifugation (21,400g, 10 min, 4 °C), 4 µg of the supernatant was incubated overnight with 2 µM of rzMmp25b at 28 °C or 75 nM of rhMMP25 at 37 °C in 50 µl Mmp25 cleavage buffer.

For human COL4A1–6 putative cleavage sites expressed as GST–GFP linkers in E.coli, DNA sequences encoding a N-terminal fusion between a 12 amino acid fragment centred on the putative cleavage site of MMP25 in COL4A1–6 and GFP were cloned into pGEX-6P-1 downstream of the GST and the recognition sequence for site-specific cleavage by the PreScission Protease-encoding sequences. BL21 (DE3) E. coli were transformed with these constructs and protein expression was induced with 1 mM IPTG when OD₆₀₀ reached 0.7. After overnight incubation at 30 °C under agitation, cells were collected by centrifugation (5,000g, 20 min at 4 °C) and lysed in 50 mM Tris (pH 8) using the FastPrep-24 cell disrupter and Lysing Matrix B Bulk (M.P. Biomedicals). After three cell disruption cycles of 20 s, the cell lysates were clarified by centrifugation (21,400g, 10 min, 4 °C). Protein concentration of the supernatant was determined using BCA (Thermo Fischer Scientific, 23223). A total of 500 ng of the soluble fraction was incubated overnight in Mmp25-cleavage buffer with 75 nM of rhMMP25 or rhMMP2 at 37 °C, or with 1 × 10⁻³ IU of the control PreScission Protease (GenScript, N02799-100) at 25 °C in 50 µl Mmp25 cleavage buffer. Uncropped gels and blots are available in Supplementary Fig. 1.

MS analysis

For protein digestion, bands of interest were excised from SDS–PAGE gels, washed twice with distilled water and shrunk in 100% acetonitrile. In-gel proteolytic digestion was performed by the addition of 4 µl of trypsin (Promega; in 50 mM NH₄HCO₃) and overnight incubation at 37 °C.

For MS, protein digests (supernatants) were analysed using nano-liquid chromatography–electrospray ionization–MS/MS on the timsTOF Pro (Bruker v.5.3) system. Peptides were separated by nanoUHPLC (nanoElute, Bruker) on a 75 μm inner diameter, 25 cm C18 column with integrated CaptiveSpray insert (Aurora, IonOpticks) at a flow rate of 200 nl min⁻¹, at 50 °C. LC mobile phase A was 0.1% formic acid (v/v) in H₂O, and mobile phase B was 0.1% formic acid (v/v) in acetonitrile. Digests (1 µl) were loaded at a constant pressure of 600 bar, directly on the column. After injection of the digest (1 µl), the mobile phases were linearly increased from 2% B to 13% over 18 min, from 13% B to 19% over 7 min, from 19% B to 22% over 4 min, and from 22% B to 85% in 3 min.

Data acquisition on the timsTOF Pro was performed using Hystar v.5.1 and timsControl v.2.0. The TIMS accumulation time was 100 ms and mobility (1/K₀) ranged from 0.6 to 1.6 V s cm⁻². Analyses were performed using parallel accumulation serial fragmentation (PASEF) acquisition method⁷⁵. Per total cycle of 1.1 s, one MS spectrum was followed by ten PASEF MS/MS spectra.

For data processing, tandem mass spectra were extracted, charge-state deconvoluted and deisotoped by Data analysis (Bruker) v.5.3. All MS/MS samples were analysed using Mascot (Matrix Science; v.2.8.1), searching the Human Proteome database (https://www.uniprot.org/uniprotkb?query=(proteome:UP000005640), 101,673 entries) assuming semi-specific trypsin digestion. Three missed cleavages were tolerated. Mascot was searched with a fragment ion mass tolerance of 0.050 Da and a parent ion tolerance of 15 ppm. Carbamidomethyl of cysteine was specified as a fixed modification in Mascot. Oxidation of methionine, hydroxylation of lysine and proline, deamination of asparagine and glutamine, and acetylation of the N-terminus were specified in Mascot as variable modifications.

Peptide and protein identifications were performed using Scaffold (v.Scaffold_5.10.0, Proteome Software). Peptide identifications were accepted by the Scaffold Local FDR algorithm if establishing a probability higher than 96.0% to achieve an FDR lower than 1.0%. Protein identifications were accepted if the probability was higher than 5.0% to achieve an FDR lower than 1.0% and containing at least two identified peptides. Protein probabilities were assigned by the Protein Prophet algorithm⁷⁶. Proteins that contained similar peptides and could not be differentiated based on MS/MS analysis alone were grouped to satisfy the principles of parsimony. Proteins sharing significant peptide evidence were grouped into clusters.

Statistics and reproducibility

Seurat v.4 was used to analyse the scRNA-seq datasets. Bulk RNA-seq data were analysed using DESeq2 v.1.12. Statistical analyses were performed using RStudio v.1.1.463 and GraphPad Prism v.9. Pearson correlation analyses and visualizations were performed using ggcorrplot v.0.1.3. Normally distributed data are represented as mean ± s.d. and were analysed using one-tailed one-way ANOVA (with post hoc Dunnett’s test) and two-tailed Student’s t-tests for multiple and single comparisons, respectively. Non-normally distributed data are represented as median ± interquartile range and were analysed using one-tailed Kruskal–Wallis tests (with post hoc Dunn’s test) for multiple comparisons and two-tailed Mann–Whitney U-tests for single comparisons. No statistical methods were used to determine the sample size. The sample size was determined by the technical constraints of the experiments, as well as our and other’s previous work on zebrafish neurovascular development^{11,12,13,14,17,71}. One-cell stage embryos are undistinguishable irrespective of their genotype, and were therefore randomized during injections. The allocation of organisms into experimental groups was randomized. Experimental groups of an experiment were always raised in parallel, under identical conditions. For zebrafish and mouse Mendelian genetics experiments, genotyping was always performed after phenotypic assessment. The researcher is therefore inherently blinded to the experimental conditions. In MO and somatic gene disruption experiments, investigators were not blinded. The sex of animals was not determined (embryonic or larval zebrafish) or was not analysed (embryonic mice) at the developmental stage of interest. The number and nature of observations (n), mean or median, type of error bar and statistical tests used for analysis are indicated in the figure legends. Images of immunofluorescence, in situ hybridization, transmission electron microscopy, and protein gels or blots are representative of experiments that were repeated independently at least three times. All attempts at replication were successful.

Reporting summary

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