Brain pericyte number is consistently lower and does not recover in foxf2a mutant larvae.

(A) Zebrafish brains were imaged using endothelial (red; Tg(kdrl:mCherry)) and pericyte (light blue; Tg(pdgfrβ:Gal4, UAS:GFP)) transgenic lines (arrows: brain pericytes). (A’-A’’) Brain pericyte soma (white arrows) and processes (yellow arrows) are closely associated with the endothelium. (B) Serially imaged wildtype and foxf2a mutant brains at 3, 5, 7 and 10 dpf. (C) Total brain pericyte numbers at 3, 5, 7 and 10 dpf. (D) Individual brain pericyte trajectories of serially imaged embryos over the same period. (E) Dorsal images of embryos for the indicated genotypes from a foxf2a heterozygous incross at 75 hpf. (F) Total brain pericytes at 75 hpf. Statistical analysis was conducted using multiple Mann-Whitney tests (C) and one-way ANOVA with Tukey’s test (F). Scale bars, 50 µm (A-B, E).

Loss of foxf2 affects embryonic pericyte numbers, but not endothelial cell pattern.

(A) foxf2a expression in wildtype brains at 72 hpf using Hybridization Chain Reaction (HCR) shows co- expression with pericyte marker ndufa4l2a. foxf2a is also lowly co-expressed in the endothelium with kdrl. Arrows show overlapping expression. (B) foxf2b is co-expressed with pericyte marker pdgfrβ also lowly expressed in the endothelium (kdrl). (C) foxf2b and pdgfrβ are expressed in a similar, overlapping pattern in pericytes of wildtype and foxf2a mutants. (D) Pericyte marker nduf4al2a and pdgfrβ are expressed in a similar, overlapping pattern in pericytes in wildtype and foxf2a mutants. (E) Image of endothelium used to generate the total blood vessel network length. (F) Total vessel network length from Vessel Metrics software. (G) Scatter plot of hindbrain CtA diameters. (H) Scatter plot of pericyte density and pericyte coverage (I). Statistical analysis was conducted using one-way ANOVA with Tukey’s test. Scale bars, 10 µm (A-D), 50 µm (F).

foxf2a mutants show strong brain vascular defects as adults.

(A-B) 3D projections of iDISCO-cleared immunostained whole wildtype and foxf2a-/- brains at 3 mpf, viewed ventrally. (C-D) Wildtype and foxf2a mutant 2 brain regions, viewed dorsally (arrows = defects in coverage). (E) Number of brain pericytes in 3 individual wildtype and mutant brains at 3 mpf detected using Imaris’ spot tool and machine learning. (F) Total vessel network length in 3 individual wildtype and mutant brains at 3 mpf using Ilastik and Imaris’ filament tool and machine learning. (G) Brain pericyte density calculated using number of brain pericytes per meter of vessel length. (H) Vessel diameter in 3 individual wildtype and mutant brains at 3 mpf using Imaris’ filament tool and machine learning. (I) Percentage of vessel segments in wildtype and mutant brains at 3 mpf segregated by vessel diameter (in 5 μm bins). (J) CUBIC-cleared wildtype and foxf2a mutant midbrain at 11 mpf (arrows: individual pericyte soma). C = caudal, D = dorsal, R = rostral, V = ventral. Statistical analysis was conducted using unpaired t-tests (E-H) and Anova with Dunnett’s post hoc test (I). Scale bars, 500 μm (A-B), 200 μm (C-D), 50 μm (J).

Loss of foxf2a has no impact on acta2-expressing brain vascular smooth muscle cells.

(A) foxf2a+/+ and foxf2a-/- larvae at 5 dpf showing vSMC coverage in the brain using endothelial (red; Tg(kdrl:mCherry)) and vSMC (light blue; Tg(acta2:GFP)) transgenic lines. (B) Scatter plot of total vSMCs at 5 dpf. (C) foxf2a+/+ and foxf2a-/- larvae at 10 dpf showing vSMC coverage in the brain. (D) Scatter plot of total vSMCs at 10 dpf. (E) 3D projections of iDISCO-cleared immunostained whole wildtype and foxf2a-/- brains at 6 mpf, viewed ventrally. (F) Total vSMC length at 6 mpf using Imaris’ filament tool and machine learning. (G) Total vessel network length at 6 mpf Ilastic and Imaris’ filament tool and machine learning. (H) vSMC coverage per total blood vessel network length at 6 mpf. C = caudal, D = dorsal, R = rostral, V = ventral. Statistical analysis was conducted using unpaired t-tests. Scale bars, 20 µm (A, C), 200 μm (E).

foxf2a mutant brain pericytes show increased soma size and process length.

(A) Wildtype and foxf2a-/- mutant brain pericytes at 3 and 10 dpf with tracings of individual pericytes (indicated by arrows). (B) Brain pericyte soma area at 3 and 10 dpf. (C) Multispectral Zebrabow labelling reveals pericyte-process interactions in the larval brain. (arrows: pericyte interaction points). (D) Total process length per pericyte at 3 and 10 dpf. (E) Varying pericyte-pericyte interactions at 10 dpf (arrows: interaction points). (F) Number of each type of interaction at 10 dpf. (G) Length of overlap when process interaction occurs. Statistical analysis was conducted using multiple Mann-Whitney tests in B, a one-way ANOVA with Tukey’s test at 3 dpf and a Kruskal-Wallis test with Dunn’s multiple comparisons test at 10 dpf in D. Scale bars, 25 µm (A), 20 µm (C), 5 µm (E).

foxf2a mutant pericytes degenerate.

(A) foxf2a-/- mutant pericyte at 10 and 13 dpf with the degenerating process and cell body with a wildtype control from the same brain region (arrows: individual pericyte). (B) Bar graph with process blebbing phenotype penetrance in wildtype and mutant brains (n = total samples examined). (C) Time-lapse of a foxf2a-/- mutant midbrain from 4-5 dpf (arrows: individual pericyte). (D) Inset of mutant pericyte undergoing degeneration (arrows: blebbing). Scale bars, 20 µm (A, C).

foxf2a mutants regenerate brain pericytes normally after genetic ablation.

Zebrafish brains were imaged using endothelial (red; Tg(kdrl:GFP)) and pericyte (light blue; Tg(pdgfrβ:Gal4, UAS:NTR-mCherry)) transgenic lines. (A) Wildtype and mutant brains at 3 dpf in control (DMSO) and treated (MTZ) groups. (B) Total brain pericytes at 3 dpf. (C) Wildtype and mutant brains at 10 dpf. (D) Total brain pericytes at 10 dpf. Statistical analysis was conducted using a one-way ANOVA (B) or Kruskal-Wallis test with Dunn’s multiple comparisons test (D). Scale bars, 50 μm (A, C).

Model of foxf2a mutant brain pericyte defects over the lifespan.

Wildtype pericytes develop normally in the embryo and establish extensive, continuous coverage over vessels by adulthood. foxf2a mutant pericytes exhibit abnormal morphology during development that worsens over the lifespan, with mutant vessels developing atypical morphology and discontinuous coverage.

Key Resources Table

Experimental N’s and details used for figures.

foxf2a mutants exhibit regional loss.

Serial imaging of a foxf2a-/- mutant brain at 3, 5, 7 and 10 dpf (arrows: brain pericytes; boxes: regional loss). Scale bars, 50 µm.

foxf2 knockouts have severe pericyte deficiency during development.

(A) Serial imaging of wildtype and foxf2a-/-;foxf2b-/- double mutant brains at 3, 5, 7 and 10 dpf (arrows: brain pericytes). (B) Scatter plot of total brain pericyte numbers at 3, 5, 7 and 10 dpf. (C) Individual mutant trajectories over the same period. Statistical analysis was conducted using multiple unpaired t-tests with Welch corrections. Scale bars, 50 µm (A).

pdgfrβ mRNA and transgene have similar expression in wildtype and foxf2a mutants.

foxf2b expression is not upregulated in mutants. A) Integrated intensity of pdgfrβ expression by HCR. B) Integrated density of confocal imaged pdgfrβ transgene expression in wildtype, foxf2a heterozygote and foxf2a homozygote mutants. C) Integrated density of foxf2b expression in foxf2a mutants by HCR. D) Relative expression by qPCR of foxf2a and foxf2b expression in foxf2a mutants. Statistics in A, C use the Student’s t-test while B and D use Anova with Bartlett’s test. None of the comparisons are significant (ns).

Expression of foxf2a and foxf2b in single cell sequencing data from Daniocell.

foxf2a is expressed strongly in mural cells, pericytes and smooth muscle (vascular and visceral), with lower expression of foxf2b in the same cell types. Available at: https://daniocell.nichd.nih.gov/.

foxf2a mutant adult brains have normal size as compared to wildtypes.

(A) 3-month-old and 11-month-old foxf2a-/- mutant and wildtype brains were dissected and imaged dorsally under Brightfield. (B) Standard length measured from snout to base of the tail. (C) Brain length was measured from the tip of the forebrain to the end of the cerebellum. (D) Widest portion of midbrain measured. (E) Ratio of brain length relative to standard length. Statistical analysis was conducted using two-way ANOVAs with Šídák’s multiple comparison test. Scale bars, 50 μm (A).

Workflow of computational analysis of vascular network in adult brains.

(A) Pericyte cell bodies annotated using the spot tool and machine learning in Imaris (green: cell bodies; purple: processes). A mask was created from the annotations, with each cell body annotated in yellow. The total number of cells was exported for statistical analysis. (B) Vessels were annotated in Ilastik to create a vessel probability map (yellow: blood vessels; blue: background). The map was imported as a channel into Imaris, where it would be further annotated with machine learning using the surface tool (green: blood vessels; blue: background). A mask was created from the surface, which was used to create a 3D network using the filament tool. Total network length and segment diameters were exported.

foxf2a mutants show strong brain vascular defects in adulthood.

(A-B) 3D projections of iDISCO-cleared immunostained whole wildtype and foxf2a-/- brains at 3 mpf with Tg(pdgfrβ:Gal4, UAS:GFP, kdrl:mCherry), viewed ventrally (arrows = defects in coverage). (C-D) 3D projections of iDISCO-cleared immunostained whole wildtype and foxf2a-/- brains at 6 mpf with Tg(acta2:GFP, kdrl:mCherry), viewed ventrally (E-F) 3D projections of CUBIC-cleared whole wildtype and foxf2a-/- brains at 11 mpf with Tg(pdgfrβ:Gal4, UAS:GFP, kdrl:mCherry) viewed ventrally. R = rostral, C = caudal, D = dorsal, V = ventral. Scale bars, 500 μm (A-D), 700 μm (E-F).

Pericyte heterogeneity in the adult zebrafish brain.

Immunolabeling for mural cell transgenes (kdrl:mCherry and pdgfrβ:Gal4, UAS:GFP) on zebrafish brain vibratome sections. Vascular smooth muscle cells (vSMC), and pericyte (ensheathing, mesh and thin strand) subtypes are present in the adult zebrafish brain. Scale bars, 5 µm.

foxf2a mutants show morphologically unusual pdgfrβ-expressing cells and blood vessels in the adult brain.

Immunolabeled sections in equivalent regions of wildtype and foxf2a-/- mutant brains at 11 mpf. (A) Region of the brain with an inset of pdgfrβ-expressing mural cells, likely vSMCs (arrows: large calibre vessel). (B) Region of the brain with an inset of pericytes (arrows: individual cell bodies. Scale bars, 50 µm (A-B).

Abnormal blood vessels become apparent in adult foxf2a mutant brains.

Immunolabeled sections in equivalent regions of wildtype and foxf2a-/- mutant brains at 11 mpf. Large aneurysm-like structure with downregulated kdrl compared to the matched wildtype region. Scale bars, 50 µm (A-B).