Secretogranin III Selectively Promotes Vascular Leakage in the Deep Vascular Plexus of Diabetic Retinopathy

Abstract

Diabetic retinopathy (DR), a leading cause of vision loss in working-age adults, induces mosaic patterns of vasculopathy that may be associated with spatial heterogeneity of intraretinal endothelial cells. We recently reported that secretogranin III (Scg3), a neuron-derived angiogenic and vascular leakage factor, selectively binds retinal vessels of diabetic but not healthy mice. Here, we investigated endothelial heterogeneity of three retinal vascular plexuses in DR pathogenesis and the therapeutic implications. Our unique in vivo ligand binding assay detected a 22.7-fold increase in Scg3 binding to retinal vessels of diabetic mice relative to healthy mice. Functional immunohistochemistry revealed that Scg3 predominantly binds to the DR-stressed CD31^- deep retinal vascular plexus but not to the relatively healthy CD31⁺ superficial and intermediate plexuses within the same diabetic retina. In contrast, VEGF bound to healthy and diabetic retinal vessels indiscriminately with low activity. FITC-dextran assays indicated that selectively increased retinal vascular leakage coincides with Scg3 binding in diabetic mice that was independent of VEGF, whereas VEGF-induced leakage did not distinguish between diabetic and healthy mice. Dose-response curves showed that the anti-Scg3 humanized antibody (hAb) and anti-VEGF aflibercept alleviated DR leakage with equivalent efficacies, and that the combination acted synergistically. These findings suggest: (i) the deep plexus is highly sensitive to DR; (ii) Scg3 binding to the DR deep plexus coincides with the loss of CD31 and compromised endothelial junctions; (iii) anti-Scg3 hAb alleviates vascular leakage by selectively targeting the DR-stressed deep plexus within the same diabetic retina; (iv) combined anti-Scg3 and anti-VEGF treatments synergistically ameliorate DR through distinct mechanisms.

Keywords: Scg3; VEGF; diabetic retinopathy; disease-targeted anti-angiogenic therapy; endothelial heterogeneity; synergistic combination anti-angiogenic therapy.

Figure 1

Constitutive expression of Scg3 in diabetic and healthy retinas of mice and humans. (a) Diabetic and healthy mouse retinas. Immunohistochemistry to detect Scg3 expression in the retinas of healthy and diabetic mice. Scg3 signals (red) are reduced in nuclear layers. CD31 signals (green) is an endothelial marker. Blue signals are for nuclei. (b) Diabetic and healthy human retinas. Scg3 expression is not upregulated in the diabetic mice and human retinas. Scale = 200 µm. RGC, retinal ganglion cells; IPL, inner plexiform layer; INL, inner nuclear layer; OPL, outer plexiform layer; ONL, outer nuclear layer; RPE, retinal pigment epithelium. Yellow boxes are the areas for zoom-in images.

Figure 2

VEGF and Scg3 differentially stimulate retinal vascular leakage in healthy and diabetic mice. (a,b) Representative images of FITC-dextran (70 kDa, green) in vivo leakage in healthy (a) and diabetic (b) mice receiving intravitreal injection of hVEGF (100 ng/1 µL/eye), hScg3 (250 ng/1 µL/eye) or PBS. Arrows indicate leakage spots. (c) Quantification of leakage spots in (a,b). (d) Quantification of leakage intensity in (a,b). Sample sizes (number of viewing fields/group) is indicated in the graphs (5 mice/group). ±SEM, * p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001; one-way ANOVA test.

Figure 3

Increase in Scg3 binding to diabetic but not healthy retinal vessels. (a) Schematic of in vivo ligand binding assay in live mice to quantify ligand binding to healthy and diabetic organs with or without blockade by ligand-specific antagonist. (b,c) Scg3-Phage binding to the retina (b) or choroid (c) in diabetic or healthy mice in the presence or absence of anti-Scg3 hAb. (d,e) VEGF-Phage binding to the retina (d) or choroid (e) in diabetic or healthy mice with or without aflibercept blocking. Sample sizes (mice/group) are indicated in the graph in (b–e). ±SEM; * p < 0.05, ** p < 0.01; n/s, not significant; one-way ANOVA test.

Figure 4

Functional immunohistochemistry (FIHC) of Scg3 binding only to the deep plexus of the retinal vasculature in diabetic but not healthy mice. (a) Binding of clonal Scg3- or VEGF-Phage to superficial, intermediate and deep plexuses of healthy and diabetic retinas in the presence or absence of anti-Scg3 hAb. (b) Zoom-in Scg3 binding signals and CD31+ endothelial cells. Arrows indicate CD31-negative microcapillaries with Scg3 binding signals. Column 1-5 in (b) are zoom-in images for Box 105 in (a), as indicated by cognate numbers. (c) CD31 is markedly down regulated in the diabetic deep plexus but minimally in other plexuses of healthy and diabetic retina. White Box i-iv are areas for zoom-in images in (c), as indicated by cognate numbers. Red signals indicate vessel-bound Scg3- or VEGF-Phage; green signals indicate CD31 on retinal endothelial cells.

Figure 5

Synergistic combination therapy. Anti-Scg3 hAb or aflibercept was ivt injected into one eye at indicated doses with PBS for the fellow eye. After 24 h, DR leakage was quantified using Evans blue assay and normalized to fellow PBS eyes. + SEM. blind-coded. n = 5 mice/group, blind-coded. All data are normalized and compared against PBS fellow eyes of the same mice by paired t-test, as indicated by the lines on the top of the figure. Other comparisons indicated by individual brackets are analyzed using one-way ANOVA tests.

Figure 6

Differential calcium influx induced by Scg3 vs. VEGF in HRMEVCs. (a) Representative image of cells with calcium probe Fluo-8 AM at different times after treatment with Scg3 (1 µg/mL), VEGF (100 ng/mL) or PBS. (b) Zoom-in images of individual cells. Arrowheads indicate individual cells with the calcium probe. (c–e) Recording of the fluorescent intensity of the calcium probe in individual cells treated with PBS (c), VEGF (d) and Scg3 (e). Each line represents a single cell. (f) Quantification of the peak fluorescence intensity for VEGF and Scg3. (g) Quantification of latency for VEGF and Scg3. Yellow boxes in (a) are areas for cognate zoom-in images in (b). Colored lines in (c–e) represent individual cells.

Figure 7

Severity of DR leakage and anti-Scg3 therapy in Scg3^−/− diabetic mice. (a) Comparison of the retinal vasculature in wild-type and Scg3^−/− mice. Flat-mounted retinas were stained with Alexa Fluor 488-isolectin B4 to visualize retinal vessels. (b) Comparison of the superficial, intermediate and deep retinal plexuses of wild-type and Scg3-deficient mice. Vessels of flat-mounted retinas were stained, as described in (a). (c) Comparison of retinal structures of wild-type and Scg3-knockout (KO) mice. (d) Representative images of retinal endothelial cells indicated by red arrows in the superficial, intermediate and deep retinal vascular plexuses of wild-type and Scg3-null mice by transmission electron microscopy. (e) Severity of DR leakage in wild-type and Scg3^−/− mice. Data were normalized to wild-type mice. n = 8 mice for wild-type and 6 mice for Scg3^−/− group. (f) Therapeutic efficacy of anti-Scg3 hAb and aflibercept (2 µg/1 µL/eye) in Scg3^−/− mice. Data were normalized to control hIgG (2 µg/1 µL/eye). n = 6 mice for control hIgG, n = 5 mice for anti-Scg3 hAb and n = 6 for aflibercept group. ± SEM. blind-coded. * p < 0.05, *** p < 0.001, n/s, not significant; one-way ANOVA test.