VEGF is an autocrine/paracrine neuroprotective factor for injured retinal ganglion neurons

Abstract

Vascular endothelial growth factor-A (VEGF) is the angiogenic factor promoting the pathological neovascularization in age-related macular degeneration (AMD) or diabetic macular edema (DME). Evidences have suggested a neurotrophic and neuroprotective role of VEGF, albeit in retina, cellular mechanisms underlying the VEGF neuroprotection remain elusive. Using purified adult retinal ganglion cells (RGCs) in culture, we demonstrated here that VEGF is released by RGCs themselves to promote their own survival, while VEGF neutralization by specific antibodies or traps drastically reduced the RGC survival. These results indicate an autocrine VEGF neuroprotection on RGCs. In parallel, VEGF produced by mixed retinal cells or by mesenchymal stem cells exerted a paracrine neuroprotection on RGCs. Such neuroprotective effect was obtained using the recombinant VEGF-B, suggesting the involvement of VEGF-R1 pathway in VEGF-elicited RGC survival. Finally, glaucomatous patients injected with VEGF traps (ranibizumab or aflibercept) due to either AMD or DME comorbidity, showed a significant reduction of RGC axon fiber layer thickness, consistent with the plausible reduction of the VEGF autocrine stimulation of RGCs. Our results provide evidence of the autocrine neuroprotective function of VEGF on RGCs is crucially involved to preserve injured RGCs such as in glaucomatous patients.

Figure 1

VEGF-elicited survival of retinal ganglion cells (RGCs). (A–C) Calcein-positive RGCs cultured for 6 DIV in control medium (A), conditioned medium from mesenchymal stem cells (CM-MSC) (B), or CM-MSC containing a rabbit polyclonal anti-VEGF-A₁₆₄ antibody (C). (D) RGC survival in CM-MSC, with or without the rabbit polyclonal anti-VEGF-A₁₆₄ antibody. (E) Linear correlation between VEGF-A₁₆₄ concentration in CM-MSC and RGC survival. Each point corresponds to an independent experiment. (F) RGC survival in the control medium, in conditioned medium from mixed retinal cells (CM-M), in CM-M plus the anti-VEGF antibody, or in control medium plus the anti-VEGF antibody alone. (G) Survival of RGCs cultured in control medium, CM-M, CM-M plus an anti-NF200 antibody, or control medium plus the anti-NF200 antibody. Data (means ± SEM) are normalized to the control condition in independent cultures (n = 4 in D, E; n = 11 in F; n = 6 in G). RGCs were seeded at initial density of 8000 cells/well. ***p < 0.001, **p < 0.01, and *p < 0.05 as compared to the control group, and ^##p < 0.01 compared between indicated groups (Kruskal–Wallis ANOVA followed by a Dunns post-hoc test).

Figure 2

Autocrine VEGF release by retinal ganglion cells (RGCs). (A,B) Representative confocal images showing live calcein-positive RGCs (6 DIV) with an initial low seeding density (8000 cells/well; A) vs. high density (30,000 cells/well; B). Scale bar represents 50 µm. (C) VEGF-A₁₆₄ concentrations in supernatants from cultured RGCs after 6 DIV, depending on the seeding density. Data are expressed as pg/ml and the means ± SEM from independent cultures with n/N referring to the number of experiments with detectable VEGF A₁₆₄ (n) over the total number of experiments (N). *p < 0.05, **p < 0.01 and ***p < 0.001 as compared to the seeding density of 8 × 10³; ^#p < 0.05 and ^##p < 0.01 between indicated groups (Kruskal–Wallis ANOVA followed by the Dunn’s post-hoc test). (D) Correlation between the measured VEGF-A₁₆₄ concentration and the actual number of surviving RGCs. The initial seeding densities were 8000 cells/well (green, n = 22), 18,000 cells/well (blue, n = 15), 30,000 cells/well (red, n = 15) or 50,000 cells/well (purple, n = 5). The non-linear regression curve is fitted with an exponential growth equation (r² = 0.65) and a significant correlation was evidenced (p < 0.001, Spearman test). (E) Suppression of RGC neuroprotection by VEGF-A traps added into the control culture medium: non-selective anti VEGF-A (pan antibody for VEGF isoforms), selective anti-VEGF-A₁₆₄ antibody, and the anti-VEGF Fab fragment ranibizumab. N refers to the total number of experiments in each group. (F) VEGF-A₁₆₄ concentrations (in pg/ml) in supernatants from RGC cultures (6 DIV) in the presence of various concentrations of three VEGF traps: anti-pan VEGF-A, anti-VEGF-A₁₆₄, or the VEGF trap ranibizumab. n/N refers to the number of experiments with detectable VEGF A164 (n) over the total number of experiments (N). *p < 0.05 as compared to the control (C) group (Kruskal–Wallis ANOVA followed by the Dunn’s post-hoc test). RGCs were seeded at initial density of 30,000 cells/ well in E,F.

Figure 3

Recombinant VEGF-A₁₆₄ directly stimulates in vitro the RGC survival through VEGF-R1 activation. (A) Quantification, by automatic counting, of densities of calcein-positive RGCs cultured for 6 DIV either in the low-nutritive condition (Control; white bar), or with application of two concentration of VEGF-A (1 ng/ml; grey bar and 10 ng/ml, black bar), or with application of the B27 supplement (2%, hatched bar), taken as positive control. (B) Quantification of densities of calcein-positive RGCs cultured for 6 DIV either in the low-nutritive condition (Control; white bar), or with application of 10 ng/ml VEGF (black bar), or with 10 ng/ml VEGF plus 5 nM ZM 323,881, a selective VEGF-R2 antagonist (ZM, grey bar) or with 5 nM of ZM 323,881 alone (oblique hatched bar). (C) Graph representing the densities of alive calcein-positive RGC cultured for 6 DIV either in the low-nutritive control condition (Control; white bar), or with application of 10 ng/ml VEGF-B (black bar) or 0.5 ng/ml VEGF-D (grey bar). (D,E) Gene amplification (PCR) of VEGF-Receptor1 (VEGF-R1, C) and VEGF-Receptor2 (VEGF-R2; D) performed on total cDNAs, previously obtained through a reverse transcription (RT) of total RNA extracted from freshly purified rat retinal ganglion cells (RGC) and rat full retina (Ret), whereas RNA from rat brain tissue (Br), was taken as positive control. These data revealed a high expression of these two-receptor transcripts in enriched RGCs. For each experiment, RGCs were seeded at initial density of 8000 cells/well. The respective RGC densities at 6 DIV were expressed as a percentage of the control. Data are means ± SEM from independent cultures (n = 20 in A, n = 18 in B, n = 11 in C). ***p < 0.001, **p < 0.01 and *p < 0.05 as compared to control (Kruskal–Wallis ANOVA, followed by a Dunns post-hoc test).

Figure 4

Treatment with VEGF trap reduces RNFL thickness in glaucomatous patients. (A) Circular peripapillary OCT scan analysis: An abnormal left eye with its corresponding fundus image (upper left). Red lines in the OCT B-Scan (upper right) indicate the inner and outer borders of the RNFL found by the algorithm. RNFL thicknesses (lower right) plotted over the thickness values measured in healthy subjects of the same age. Mean thickness of the RNFL in the six sectors (lower left). This scan shows a significant decrease (red segment) in the inferior temporal quadrant. B-C: IOP levels in eyes injected with VEGF traps (blue curves, n = 10) or non-injected control eyes (red curves, n = 10) from glaucomatous (B) or non-glaucomatous patients (C). D-E: RNFL thickness of eyes injected with VEGF trap (blue curves, n = 10) or non-injected control eyes (red curves, n = 10) from glaucomatous (D) or non-glaucomatous patients (E). **p < 0.01 compared to the baseline value in the treated group (Friedman test followed by a Dunn’s post hoc test).