Fig 2: Neuronal cell swelling in adult Angpt4-/- mice.(A–B) Light microscopy images of peripheral retinal sections from control and Angpt4-/- mice stained with (A) toluidine blue (n = 7 mice/group) or (B) Masson's trichrome (n = 9 control and 5 Angpt4-/- mice). Framed area of inner nuclear layer (INL) is magnified on the right. Two different fixation, embedding and staining protocols indicate increased thickness of the INL in (A) and (B), data analyzed at least from four sections per mouse. (C–D) TEM micrographs from peripheral INL. (C) n = 8 control and 6 Angpt4-/- P13 mice and in adult mice (D) n = 8 control and 8 adult Angpt4-/- mice. Quantification (on the right) was done from cell somas locating in the interface between INL and inner plexiform layer that is composed mainly of amacrine cells (Jeon et al., 1998). Asterisks indicate cell swelling in (D) and black arrows Müller cell processes (C, D) that are displaced around swollen cells. A, amacrine cell; M, Müller glia; B, bipolar cell. (A–D) Median (line), average (square), 75th quartile (box), 5th and 95th percentile (whiskers) in whisker blots. *p<0.05 and **p<0.01 WT vs. Angpt4-/- in t-test. 10.7554/eLife.37776.017Figure 6—source data 1.Cell soma areas of individual neuronal cells in the interphase between INL and IPL.

Fig 3: Ligand-specific and redundant functions of Angpt4 and ANGPT4.(A) Angiopoietin-induced TIE2 translocation and activation in cell–cell junctions. TIE2-WT HUVECs were left non-stimulated (control) or stimulated with recombinant angiopoietins as indicated for 1 hr, fixed, and stained for total TIE2 (green), phosphorylated TIE2 (pTIE2, red) and nuclei (DAPI, blue). White cropping indicates examples of ROIs (sites where cell contacts occurred based on microscopic examination) for TIE2 and pTIE2 intensity measurement from randomly selected cells. (B) Quantification of TIE2 activation (pTIE2/total TIE2) in cell–cell junctions (n = 2 to 7 stimulations, total 330 to 915 cell junctions/stimulation were measured). (C) Sparse TIE2-GFP HUVECs were stimulated with angiopoietins overnight, fixed and imaged. In ANGPT1-stimulated cells, majority of TIE2-GFP is ECM bound (arrows). ANGPT2 promoted TIE2 translocation to the retracting cell edges (arrows) and less in ECM (dotted line indicates leading edge of the cell). Angpt4 and ANGPT4 did not induce long-term effect on TIE2 clustering and translocation into ECM. (D) Angiopoietin binding to acellularized ECM fraction from cultured HUVECs (n = 4 run in triplicate). (E) TIE2-WT HUVECs spreading on fibronectin, ANGPT1, ANGPT2, Angpt4 or ANGPT4. Cells were let to spread on coated coverslips for 1 hr and stained for actin. Quantification of image data is shown in (F). Cell area was normalized to fibronectin (FN) (n = 3 to 9 experiments, 333 to 1778 cells per coating were analyzed). (G) Angiopoietin-TIE2 binding affinity measured by ELISA. ANGPT1 binding to TIE2 (250 ng/ml) was set to one at ANGPT1 concentration 500 ng/ml (n = 4 experiments). (H) Angiopoietin binding to TIE2 in surface plasmon resonance assay. Human purified TIE2-Fc was immobilized on CM5 chip, and angiopoietins were injected onto the TIE2-Fc surface at 100 nM concentration. Mean ±SD, ANOVA followed by the Bonferroni post hoc test. ***p<0.001, **p<0.01, *p<0.05 vs. control; †††p<0.001, ††p<0.01 vs. ANGPT1; ###p<0.001, ##p<0.01, #p<0.05 vs. ANGPT2.

Fig 4: Angpt4 deficiency results in defective SMC maturation in veins.TEM micrographs of peripheral annular vein in P8 (A) and adult (11 months old) mice (B). SMCs are marked by red line. In the developing veins, SMCs are covered by basement membranes (BM, arrowheads) and SMC cytoskeleton appears translucent with some myofilaments (arrow). Perivascular ECM contains no fibrillar collagen. In adult mouse, venous SMC cytoplasm is mostly filled with myofilaments and perivascular matrix contains fibrillar collagen (Col). TEM micrographs of peripheral vein in 2 (C, D) and 11 (E, F) months old WT and Angpt4-/- mice, respectively. In both genotypes, perivascular cells (red line) are closely aligned with ECs and surrounded by BM (arrowheads) that indicates SMC identity. In Angpt4-/- retina, SMC cytoplasm lacks myofibril-rich cytoskeleton and fibrillar collagen (Col) matrix. No ultrastructural abnormalities are observed in ECs. (G–J) Whole mount retinal staining of peripheral veins in 2 months old WT (G, I) and Angpt4-/- (H, J) mice. SM22 (red) (G–J) and IB4 (green) (G, H) staining. In Angpt4-/- retina, SM22 level is low around veins (arrows, V). (K) Tie2 expression in arterial (a) and venous (v) SMC cell lines. HUVECs and fibroblasts are positive and negative controls, respectively. (L) qPCR analysis of venous specification marker genes at P12 retinas (Angpt4-/-n = 5 and WT n = 6). (M–N) Quantification of SMC maturation markers from TEM analyzed retinas (n = 6 WT and n = 7 Angpt4-/- mice). (O) Peripheral annular vein diameter is reduced in Angpt4-/- retina (n = 7 mice/group, the shortest diameter of peripheral branch measured from TEM cross-section). Data are presented as mean ±SD, ***p<0.001 and **p<0.01 Angpt4-/- vs. WT in t-test.

Fig 5: mRNA expression of angiopoietins and Tie2 in retinal vein development.mRNA expression was detected in whole mount retinas from WT mice by in situ hybridization. (A) At P3, astrocytes expressing Angpt4 precede the front of the developing vasculature (red blood cells indicated by arrows). (B) At P12, Angpt4 expression is notable around the developing vein in the peripheral retina (V, asterisks) and colocalizes with the astrocyte marker GFAP (white overlay in insert). (C) Angpt1 expression is not detected at P12 by whole mount in situ hybridization in the peripheral retina. (D) Angpt2 is expressed in retinal neurons (arrowhead) in the intermediate retinal plexus. (E) Expression of Tie2 is detected in the endothelium of arteries (A), veins and capillaries at P12. OR, ora serrata.