Interplay between CCN1 and Wnt5a in endothelial cells and pericytes determines the angiogenic outcome in a model of ischemic retinopathy

Abstract

CYR61-CTGF-NOV (CCN)1 is a dynamically expressed extracellular matrix (ECM) protein with critical functions in cardiovascular development and tissue repair. Angiogenic endothelial cells (ECs) are a major cellular source of CCN1 which, once secreted, associates with the ECM and the cell surface and tightly controls the bidirectional flow of information between cells and the surrounding matrix. Endothelium-specific CCN1 deletion in mice using a cre/lox strategy induces EC hyperplasia and causes blood vessels to coalesce into large flat hyperplastic sinuses with no distinctive hierarchical organization. This is consistent with the role of CCN1 as a negative feedback regulator of vascular endothelial growth factor (VEGF) receptor activation. In the mouse model of oxygen-induced retinopathy (OIR), pericytes become the predominant CCN1 producing cells. Pericyte-specific deletion of CCN1 significantly decreases pathological retinal neovascularization following OIR. CCN1 induces the expression of the non-canonical Wnt5a in pericyte but not in EC cultures. In turn, exogenous Wnt5a inhibits CCN1 gene expression, induces EC proliferation and increases hypersprouting. Concordantly, treatment of mice with TNP470, a non-canonical Wnt5a inhibitor, reestablishes endothelial expression of CCN1 and significantly decreases pathological neovascular growth in OIR. Our data highlight the significance of CCN1-EC and CCN1-pericyte communication signals in driving physiological and pathological angiogenesis.

Figure 1

Expression of CCN1 during postnatal vascular development in the retina. (A–D) Flat-mounted preparations of retinas from transgenic mice expressing the GFP reporter gene under the control of the CCN1 promoter were analyzed for cellular localization of the CCN1:GFP signal in the growing vasculature. Images shown depict colocalization in P4 retinas of the CCN1:GFP signal with the endothelium-specific marker IB4. (E–P) The CCN1:GFP signal does not colocalize with mural cell markers including NG2, desmin, myosin heavy chain (mhc) and smooth muscle α-actin (SMA). Retinas with completely formed vasculature at P16 expressed residual amount of the CCN1:GFP in the vascular wall (ON: optic nerve). (Q) The transcript levels of the endogenous CCN1 gene during postnatal development of the retinal vasculature was determined by qPCR and normalized to 18S ribosomal RNA. (n = 5).

Figure 2

OIR-induced pre-retinal neovasculatization is associated with the expression of CCN1 in pericytes. (A,B) Representative flat mount preparations of IB4-stained retinas from control and OIR mice. Mouse pups were placed at P7 under hyperoxia (75% oxygen) or ambient air (normoxia) for 5 days. Mice were returned to ambient air until P17. Areas of vaso-obliteration and preretinal neovascular tufts as determined by computer-assisted image analyses are shown with dotted yellow lines and blue outline respectively. (C,D) Steady state mRNA levels of CCN1 following the hyperoxic (P7 to P12) and ischemic (P12 to P17) phases of OIR as determined by qPCR. CCN1 mRNA levels were normalized to those of 18S rRNA. (n = 4). *p < 0.05 versus P7 (normoxia); **p < 0.001 versus P12 (normoxia); ***p < 0.05 versus P12 (normoxia); ****p < 0.05 versus P17 (normoxia). (E,F) Flat mounted preparations of retinas of the CCN1:GFP reporter mice subjected to OIR. Retinas were collected during the posthyperoxic phase at either P14 or P17 and stained with the endothelial marker endomucin (Edmucin) or the pericyte marker NG2. Note that CCN1:GFP-positive cells are endomucin-negative and that the CCN1:GFP signal localized predominantly in NG2-positive cells of neovascular tufts.

Figure 3

Effects of cell type-specific CCN1-deficiency on retinal vessel growth and development. (A) In the targeted CCN1 locus, exons 1 and 2 were flanked by loxP sites. Endothelial- or pericyte-specific promoter-mediated tamoxifen-induced cre excision removed the floxed fragment inactivating the CCN1 gene in the targeted cells. (B) Representative immunofluorescence images of IB4-stained whole retinal mounts of wild-type CCN1^+/+ mice and their littermate with either endothelial- or pericyte-specific CCN1 deletion. Vascular parameters were analyzed using the AngioTool Software (b for a, d for c, f for e). The outline of the vasculature is shown in white. The vasculature skeleton representation is shown in red and branching points in green. (C,D) Quantitative analysis of vascular parameters of representative retinas of CCN1^+/+ and CCN1 mutant mice. Graphical representations of the changes in vascular surface and lacunarity (i.e., the degree of “gappiness”) are shown. *p < 0.001 and **p < 0.05 (n = 5). (E) Representative immunofluorescence images of dual IB4 (red) and NG2 (green) staining of whole mount retinas of wild-type CCN1^+/+ mice and their littermate with either endothelial- or pericyte-specific CCN1 deletion.

Figure 4

Effects of pericyte-derived CCN1 on pre-retinal neovascularization following OIR. (A,B) Representative flat-mount preparations of IB4-stained retinas from CCN1^+/+, Cdh5-CreER^T2 CCN1^Δ/Δ and Cspg4-CreER^TM CCN1^Δ/Δ mice following OIR. (C,D) Areas of pre-retinal neovascular tufts and vaso-obliteration were determined by computer-assisted image analyses (C for A and D for B). (E,F) Compiled data showing percentage of neovascular tuft and vascularized areas in Cdh5-CreER^T2 CCN1^Δ/Δ and Cspg4-CreER^TM CCN1^Δ/Δ mouse retinas. Levels of neovascular and vascularized areas were set to 100% in wild-type OIR mice to facilitate comparisons among animals and groups. *p < 0.001 versus CCN1^+/+ and Cdh5-CreER^T2 CCN1^Δ/Δ (n = 4). (G–I) Double IB4 and NG2 staining of neovascular tufts in retinas from CCN1^+/+, Cdh5-CreER^T2 CCN1^Δ/Δ and Cspg4-CreER^TM CCN1^Δ/Δ mice depicting neovascular tufts in central retinal areas following OIR. Note the reduced size of tufts (i.e., “microtufts”) in Cspg4- CreER^TM CCN1^Δ/Δ mice.

Figure 5

Adenovirus-mediated expression of CCN1 induced Wnt5a expression and activity in cultured pericytes. (A) Cultured ECs and pericytes were transduced with an adenoviral vector expressing either luciferase (Ad-luc) or CCN1 (Ad-CCN1). After incubation in serum-free medium for 16 h, cell lysates were analyzed by western immunoblotting for CCN1 protein expression. (B,C) Expression Wnt ligands was analyzed by qPCR in Ad-luc- and Ad-CCN1-infected cells. *p < 0.001 versus Ad-luc. (D,E) The relative abundance of Wnt5a transcripts and Wnt3a and Wnt7a was calculated by comparing their ΔCTs in pericytes versus ECs. (F,G) Effects of CCN1 on canonical and non-canonical Wnt signaling activity. Following adenoviral-mediated gene transfer with either Ad-CCN1 or Ad-luc, cultured pericytes were transiently transfected with either pGL3-NFAT luciferase reporter or LEF/TCF-M50 Super 8× TOP Flash plasmid. Luciferase activity was determined in cell lysates and media. Values shown are from a representative experiment performed in triplicate. **p < 0.01 versus Ad-luc. Experiments were repeated three times using different cell preparations with similar results.

Figure 6

Functional and regulatory relationship between CCN1 and Wnt5a in ECs. (A,B) ECs were transduced with Ad-CCN1 or Ad-luc adenoviral vector, allowed to adhere to cytodex beads and then embedded in fibrin gels in the absence and presence of Wnt5a. Sprout formation was monitored by microscopy after 3 days. Quantitative analyses of spheroid sprouting was determined and expressed in arbitrary units. (C,D) ECs were pretreated with 5-FU to inhibit cell proliferation and the effect on cell migration was determined with cell scratch test. A representative experiment of the percentage closure from cell front after 8 h is shown.

Figure 7

Inhibition of non-canonical Wnt5a reduced pathological neovascularization following OIR. (A–D) Expression of Wnt5a, Wnt7a, Wnt7b and Wnt3a in mice with either endothelial- or pericyte-specific CCN1 deletion and their wild-type littermates following OIR. Wnt ligand gene expression was analyzed by qPCR in retinal lysates harvested at P17. **p < 0.01 versus CCN1^+/+. *p < 0.01 versus Cdh5-CreER^T2 CCN1^Δ/Δ. (E–G) Flat-mount images of IB4-stained retinas from OIR mice at P17 following treatment with either TNP470 or vehicle alone (Veh). Areas of pre-retinal neovascular tufts at P17 were delineated by a computer-assisted image software and highlighted in blue (c for a and d for b). The compiled data showing the percentage of neovascular and avascular areas are represented in F and G. **p < 0.05 versus +Veh. (H) Transcript levels of CCN1, YAP, SNAIL and G3BP2 as determined by qPCR in retinas from OIR mice injected with either control vehicle or TNP470. RNA levels at P17 (+Veh) were set to 100% to facilitate comparison among groups. *p < 0.01 versus P17 (+Veh). (I) Retinal flat-mounts from OIR CCN1:GFP reporter mice treated with vehicle or TNP470. Note that TNP470 treatment resulted in reduced neovascular tuft formation and reexpression of the CCN1:GFP reporter in ECs instead of perivascular pericytes in remnant tufts (arrows).

Figure 8

Regulation of EC proliferation and CCN1 gene expression by Wnt5a. (A,B) Monocultures and cocultures of ECs and pericytes (P) were transduced with either Ad-CCN1 or Ad-luc and further assessed for Wnt5a gene expression and BrdU incorporation in ECs. Cocultures of Ad-luc-infected ECs and Ad-CCN1-infected pericytes were incubated in the presence and absence of TNP470 (50 μM). *p < 0.01 versus Ad-luc. **p < 0.05 versus EC-Ad-CCN1 P-Ad-luc. ***p < 0.05 versus EC-Ad-luc P-Ad-CCN1. (C) Effects of Wnt5a on CCN1 gene expression on monolayer cell cultures of ECs and pericytes. Cultured ECs and pericytes were left untreated or treated with jasplakinolide for 2 h in the presence and absence of Wnt5a. CCN1 protein expression was determined by western immunoblotting. (D) Working model of the interplay between CCN1 and Wnt5a and their role in EC/pericyte crosstalk. Under ischemic conditions, EC/pericyte cooperation is altered by the shutdown of CCN1 in ECs and its expression by pericytes. Inasmuch as secreted CCN1 remains pericellular and acts largely on the cells that produce it, pericyte-derived CCN1 induces the expression of Wnt5a which acts in a paracrine manner to dampen CCN1 expression in ECs, stimulate their proliferation and promote retinal neovascularization. Thus, pericyte-derived CCN1 increases the sensitivity of ECs to ischemia.