GFAP Cre specifically and efficiently ablates FN expression in ACs ahead of the sprouting vascular front. (A-C) Visualization of nuclear GFAP lacZ (blue, A,B; black, C) revealed GFAP promoter activity in P5 retina (A). Border between ACs and vasculature is indicated by white line (B), as illustrated by co-labelling with fibronectin (red, B). High magnification imaging of the sprouting vascular front revealed GFAP lacZ expression in ACs ahead of vessels (arrowheads, C) and in ACs surrounding the vascular plexus (arrows, C) visualized by FN staining (red). High magnification images are projections of confocal z-stacks encompassing the depth of the vasculature. (D-I) GFAP Cre-mediated deletion of AC-derived fibronectin (FN^ACko) leads to reduced FN expression in mutant retinas (E,F) compared with control retinas (ctrl, D), as shown by FN immunofluorescence staining (red, D-I). (J-L) Counterstaining of ACs with PDGFRα (blue) and FN (red) reveals normal network development but reduced FN deposition in FN^ACko (K,L) compared with control littermates (J). White arrows indicate residual FN expression in FN^ACko retinas in the astrocytic network ahead of the vasculature (red, L) and sprouting vascular front (red) resulting in mosaicism. (M-O) High-magnification images showing the absence of astrocytic FN ahead of the sprouting vascular front in FN^ACko. Arrowheads indicate regions with complete loss of FN (N,O). Deficiency of astrocytic FN matrix leads to aberrant filopodia extension (asterisk, N). Arrows in N and O indicate regional residues of astrocytic FN.

RGD integrin binding in FN is dispensable for retinal migration but promotes filopodial alignment. (A,B) Mutation of integrin-binding RGD to an inactive RGE motif in astrocytic FN (FN^ACko/RGE) does not affect radial expansion of vessels towards the periphery and overall vascular patterning. (C) Vessel migration in RGE/wild-type mice and mice inheriting inactive RGE motif in astrocytic FN: wild type and RGE/wild type (n>8); control and FN^ACko/RGE (n=5). ^*P<0.05. Values represent mean+s.e.m. (D,E) Overview images of the retinal vasculature of mice deficient in endothelial integrin α5 (Itga5^ECko, E) illustrates no significant defects in vessel morphology compared with control littermates (D). (F) Retinal vessel migration of control and Itga5^ECko mice at P5. Number of mice: control, n=9; Itga5^ECko, n=6. ^*P<0.05. Values represent mean+s.e.m. (G-I) High-magnification imaging of filopodia (red, G-I) at leading tip cells displayed mis-alignment of filopodia to the astrocytic FN matrix (green) in mice deficient for endothelial Itga5 (H) and mice lacking astrocytic FN RGD (I) when compared with control mice (G). Arrowheads indicate misaligned filopodia.

Inhibition of VEGF binding to FN leads to reduced vessel migration due to decreased VEGFR2 signalling and PI3K activity. (A) ¹²⁵I-VEGF₁₆₅ binding to fibronectin (FN) 9-10/12-14 in the presence of increasing concentrations (μM) of peptides, expressed as a percentage of no-peptide controls (%). The half-maximal inhibitory concentration (IC₅₀) is analyzed to determine the inhibitory function of VEGF binding to fibronectin. (B) Control peptides (IC₅₀>100 μM) and FNIII_13-14 (IC₅₀=0.09 μM) were applied in the following experimental settings. Western blot analyses reveal activation and phosphorylation of VEGFR2 (pVEGFR2) in the presence of VEGF (lane 2). Treatment with inhibitory peptide FNIII_13-14 abolishes VEGFR2 phosphorylation (lane 3) down to baseline levels obtained in unstimulated conditions (lane 1). VEGFR2 protein levels are comparable in all experimental settings. (C) Western blot analyses show that the addition of VEGF results in activation of downstream target pERK whereas ERK protein expression levels are unaffected in the presence of VEGF and varying concentration of FNIII_13-14 inhibitory peptide. Treatment with increasing concentration of inhibitory peptide FNIII_13-14 results in concurrent decrease of ERK phosphorylation. (D) VEGFA-induced endothelial cell migration was inhibited with increasing concentration of peptide FNIII_13-14. Control peptide had no effect. Data are mean+s.e.m. (E) Distances vessels have migrated towards retinal periphery at P5 when treated with inhibiting peptide FNIII_13-14 and control peptide for 24 hours compared with uninjected control eyes. By measuring the total length of radial expansion of vessels we determined a significant reduction upon injection of FNIII_13-14 whereas control peptides have no effects. (F) When calculating the distance vessels have migrated during 24 hours of peptide treatment (day P4-P5), we found an almost 40% decrease in vessel migration upon FNIII_13-14 injection (n>5 injected with FnIII_13-14; n>5 control peptide injected mice; ^**P<0.01). Values represent mean±s.e.m. (G) Phosphorylation of VEGFR2 and AKT were assessed by total protein extraction and western blotting 15 hours after treatment with inhibiting peptides. Densitometric analysis revealed a significant reduction of pVEGFR2 compared with total VEGFR2 expression levels in FnIII_13-14 injected retinas. (H) The percentage of phosphorylated AKT in relation to total amounts of AKT after injection of peptides inhibiting VEGF-FN binding. ^*P<0.05. Values represent mean+s.e.m. (n=7 mice injected with FNIII_13-14; n=5 control peptide injected mice).

Localization of extracellular matrix component FN and collagen IV in the retinal vasculature. (A-D) Immunofluorescence staining for fibronectin (FN, red A-C) revealed strong deposition ahead of the growing vasculature (white arrow, A), whereas astrocytic FN expression is reduced in the mature vascular plexus. Boxed areas in A are shown at higher magnification, demonstrating the lack of astrocytic FN deposition (red) in the mature retinal vascular plexus. Blood vessels were visualized with isolectin-B4 (green, A-C) and co-labelling with PDGFRα (blue, C,D) confirmed astrocytic deposition of FN. Note that the apparent vascular staining with PDGFRα reflects astrocytic endfeet in close association with the blood vessels. High magnification imaging showed alignment of filopodia to astrocytic matrix (arrows, B). (E,F) In contrast to FN, collagen IV (red) is distributed in the abluminal BM of retinal vessels (green, E) and absent from endothelial tip cells and filopodia (F). High magnification images are projections of confocal z-stacks encompassing the depth of the vasculature.

Loss of astrocytic FN results in reduced migration of vessels. (A,B) Visualization of the retinal vasculature (P5) by isolectin-B4 immunofluorescence staining revealed reduced migration towards the retinal periphery in mice deficient of astrocytic FN. Vessel migration was quantified by measuring total length as illustrated by the white arrow (A,B). (C) Radial expansion of superficial vessels at P5 in control (n>5), heterozygous (FN^ACko/+, n>3) and FN^ACko mice (n>5). ^**P<0.01 one-way ANOVA. Values represent mean+s.e.m. (D) The number of filopodial protrusions per vessel length at the migrating vascular front of P5 mice (n≥20). ^***P<0.0001. Values represent mean+s.e.m. (E) The number of branch points per image field of P5 FN^ACko and control mice, n≥10. ^***P<0.0001. Values represent mean+s.e.m.

Loss of astrocytic FN impairs VEGFR2-mediated stimulation of migration via the PI3K pathway. (A) Western blot analysis showed reduced phosphorylation of VEGFR2 but not of total VEGFR2 protein levels in retinas of P5 FN^ACko compared with control mice. (B) PI3K activity determined by pull-down assays with pY-peptides was markedly reduced in FN^ACko retinas. Control and FN^ACko retinas for p-peptide, n=7. ^*P<0.05. Values represent mean+s.e.m. (C) Phosphorylation of serine residue 473 (pSer) of AKT relative to total levels of AKT was significantly reduced in FN^ACko retinas (C) but not in Itga5^ECko retinas (C). (D) Densitometric analyses of P5 control (n=4) and Itga5 retinas (n=5). Values represent mean+s.e.m.

Compound deletion of astrocytic FN and heparan sulfate (HS) results in an enhanced migratory defect. (A-C) Immunofluorescence staining with a phage display antibody (HS4C3V) recognizing sulphated HS chain (green, A-C) revealed a similar expression pattern to that observed for FN (Fig. 1A-C). HS is localized along the astrocytic network ahead of the vasculature (B,C) labelled by the AC marker PDGFRα (blue, A-C). Furthermore, HS is present in the BM surrounding the blood vessels (A,C) visualized by isloectin-B4 (red, A-C). (D,E,G,H) Overview images of the retinal vasculature of control mice (D,G), mice deficient in astrocytic HS (Ext1 ACko; E) and mice deficient in both HS and FN (FN/Ext1 ACko; H). (F) Percentage of retinal vessel migration at P5 (n>10 control mice; n=6 Ext1 ACko; n=3 FN/Ext1 ACko; ^***P<0.0001). Values represent mean+s.e.m. (I) Proposed model of astrocytic VEGF-FN interaction regulating vessel migration. Right to left: FN is expressed by ACs ahead of the growing vasculature and assembled in a fibrillar network providing instructive cues for vessel migration. ACs produce and secrete VEGFA in the extracellular milieu and this VEGFA is retained by the FN matrix. Endothelial tip-cell filopodia express integrin receptors that guide filopodial alignment along the ACs and mediate adhesion to the FN matrix. VEGFA binds to VEGFR2 expressed on leading EC, thereby triggering VEGF receptor signalling in tip cells and downstream effectors PI3K. Loss of astrocytic FN results in a diffused VEGFA gradient, reduced VEGF/VEGFR2 signalling and decreased PI3K activity in endothelial tip cells. As a consequence, migration of the vascular plexus is delayed. Selective replacement of integrin RGD-binding motif by an inactive RGE motif in ACs or endothelial-specific deletion of integrin α5 lead to abundant filopodial misalignment and probably to defective tip-cell adhesion. However, in both cases radial vessel migration is not or is only marginal affected. Combined deletion of both matrix components FN and HS from ACs results in greater reduction of radial vessel expansion compared with loss of FN or HS, respectively, indicating the importance of VEGF retention to ACs in vessel migration.