VEGF but not PlGF disturbs the barrier of retinal endothelial cells

•VEGF165 disturbs barrier of retinal endothelial cells above a threshold concentration.•Loss of claudin-1 is associated with long-term treatment of VEGF165.•PlGF does not disturb barrier of retinal endothelial cells, effect of VEGF121 is weak.•VEGF-A inhibition is sufficient to normalize barrier function.•Anti-VEGF drugs ranibizumab and aflibercept are equally efficient.

Elevated permeability of retinal endothelial cells (REC), as observed in diabetic retinopathy (DR), is induced by extended exposure to ≥25 ng/ml vascular endothelial growth factor A165 (VEGF165) for up to 3 d and this effect is more pronounced when equimolar amounts of basic fibroblast growth factor (bFGF) and insulin-like growth factor (IGF-1) are present. Down-regulation of the tight-junction protein claudin-1 and its loss from the plasma membrane is associated with induced higher permeability, whereas other tight-junction proteins (e.g. claudin-3, claudin-5, ZO-1) show only subtle changes in our experimental setting. Using immortalized bovine REC (iBREC) as a well-established model, we investigated effects of other members of the VEGF family, i.e. VEGF121, placental growth factor (PlGF-1 and PlGF-2) and viral VEGF-E which activate different sets of VEGF receptors, on barrier function after extended treatment: iBREC were incubated with 1-100 ng/ml of the growth factors for up to 2 days before barrier function was assessed by measuring transendothelial resistance (TER). Presence of TJ-proteins was determined by western blot analyses and immunofluorescence staining. Similar experiments were performed to evaluate whether the primary actions of PlGF-1, PlGF-2 or VEGF121 are modulated by bFGF or IGF-1 when all growth factors (each at 25 ng/ml, but 10 ng/ml IGF-1) act simultaneously at equimolar concentrations. We also studied the potential normalization of the barrier disturbed with combinations of growth factors by addition of the VEGF-specific Fab fragment ranibizumab or the recombinant protein aflibercept which binds VEGF and PlGF. Whereas 1 ng/ml VEGF-E were sufficient to impair the iBREC barrier, a higher concentration of 100 ng/ml VEGF121 was needed to reduce TER and expression of claudin-1 over 2 days. By PlGF-1 or PlGF-2, the barrier was not affected even at the highest concentration tested (100 ng/ml) and these factors also did not modulate the effect of VEGF165. The weak barrier derangement caused by VEGF121 was slightly enhanced by bFGF and IGF-1. After induction of the barrier breakdown with various combinations of all growth factors included in the study, normal TER and claudin-1 expression was re-established by ranibizumab. Both VEGF inhibitors ranibizumab and aflibercept similarly reinstated lost claudin-1, even when applied at a small fraction of the clinically relevant concentrations. These results show that VEGF-A, but not PlGF impairs the barrier function of iBREC and that the longer isoform VEGF165 is more potent than VEGF121. To induce barrier dysfunction in iBREC, activation of VEGF receptor 2 - probably in concert with neuropilin-1 - seems to be sufficient because VEGF-E and VEGF165, but not PlGF-1/-2 reduced TER or claudin-1 expression.