Targeting cell-type-specific, choroid-peripheral immune signaling to treat age-related macular degeneration

Abstract

Age-related macular degeneration (AMD) is a leading cause of blindness featuring pathogenic neovascularization of the choroidal vasculature (CNV). Although systemic immunity plays a role in AMD, the ocular signals that recruit and activate immune cells remain poorly defined. Using single-cell RNA sequencing, we prospectively profile peripheral blood mononuclear cells from 65 individuals including AMD and controls, which we integrate with existing choroid data. We generate a network of choroid-peripheral immune interactions dysregulated in AMD, including known AMD-relevant gene vascular endothelial growth factor (VEGF) receptor 2. Additionally, we find CYR61 is upregulated in choroidal veins and may signal to circulating monocytes. In mice, we validate that CYR61 is abundant in endothelial cells within CNV lesions neighboring monocyte-derived macrophages. Mechanistically, CYR61 activates macrophage anti-angiogenic gene expression, and ocular Cyr61 knockdown increases murine CNV size, indicating CYR61 inhibits CNV. This study highlights the potential of multi-tissue human datasets to identify disease-relevant and potentially therapeutically modifiable targets.

Keywords: CYR61; angiogenesis; choroid; macrophage; neovascularization.

Graphical abstract

Figure 1

Schematic of experimental workflow Created in BioRender.

Figure 2

Cell heterogeneity of circulating immune cells and choroid in AMD (A) Uniform manifold approximation and projection (UMAP) plot showing cell type heterogeneity of PBMCs (from n = 65 patients). (B) Dot plot showing marker gene expression for cell types in PBMCs. (C) UMAP plot showing cell types that are shared and unique between PBMCs and choroid. (D) UMAP plot showing cell type heterogeneity of choroid and PBMCs. See also Figures S1, S2, and Table S1.

Figure 3

Choroid-peripheral immune cell signaling dysregulated in AMD (A and B) Stacked bar plot showing the number of up- and downregulated genes for each cell type in PBMCs and choroid. DEGs are those with pseudobulked |log₂FC| > 0.5 and p value < 0.05 (Wilcoxon rank-sum test). (C) Full intercellular network of ligand-receptor interactions that are dysregulated in AMD. The lines connecting icons are shaded by the LR_score calculated by SingleCellSignalR. Genes expressed by PBMC cell types are shown in color according to the legend, whereas those expressed choroidal cell types are colored in gray and labeled in parentheses. V-shaped icons indicate receptor genes, whereas diamond icons indicate ligand genes. See also Figure S3.

Figure 4

Choroidal ligands or receptors dysregulated in AMD that can signal with circulating immune cells (A) Violin plots showing nine ligands/receptors that are dysregulated in choroidal cell types (pseudobulked |log₂FC| > 0.5 and p values < 0.05, Wilcoxon rank-sum test). The specific cell types in which the genes are dysregulated are labeled in parentheses. Each point represents the average expression for one patient (n = 10–11 patients/group). (B) Representative image of CYR61 localization in whole-mounted mouse CNV lesions in the laser-injury-induced mouse model of wet AMD. Each image is an en face maximal projection of a confocal z stack that spans the entire neovascular lesion across 19 slices. CD31 and CCR2 are markers of endothelial cells and monocyte-derived macrophages, respectively. See also Figures S4–S6.

Figure 5

Effect of CYR61 on endothelial cells and monocytes/macrophages (A) Heatmap showing HUVEC genes whose expression is dysregulated after 24-h incubation with CYR61 (adjusted p value < 0.05; n = 5 wells/group). (B) Heatmap showing THP1 gene expression changes after treatment with CYR61 (24 h; n = 5 wells/group). Genes shown are those with adjusted p value < 0.05 and |log₂FC| > 1.5 in THP1 cells treated with 5 μg/mL. (C and D) Pathway analysis of the 1,110 genes with adjusted p value < 0.05 upregulated in BMDMs following CYR61 treatment (24 h, 0.5 μg/mL). The top 10 pathways, ranked by p value, are shown. (E) Heatmap showing genes dysregulated in BMDMs by CYR61 treatment (24 h, 0.5 μg/mL) with adjusted p value < 0.05 and |log₂FC| > 1.5 (n = 4–7 wells/group). (F) Bar graphs showing changes in the expression of three pro-inflammatory and three pro-angiogenic genes in BMDMs following 24-h treatment with CYR61 (0.5 μg/mL). Statistical significance was assessed using Mann-Whitney test. Bars show mean ± SEM for n = 4 wells/group (technical replicates). (G) Venn diagram showing the number of ligand-receptor interactions with a ligand dysregulated when BMDMs are treated with CYR61 and/or a receptor that is dysregulated in HUVECs when co-incubated with CYR61-activated BMDMs. These ligand-receptor interactions were from the LRdb database downloaded from https://github.com/SCA-IRCM/LRdb. (H) Fifty-eight macrophage-to-endothelial cell ligand-receptor interactions that are dysregulated when macrophages are activated by CYR61. (I) Macrophage ligands dysregulated by CYR61, which can signal to endothelial cells (n = 4–7 wells/group). ∗p < 0.05. See also Figure S7.

Figure 6

CYR61 inhibits murine CNV (A) Immunostaining for CYR61 and CD31 in mouse CNV lesions in eyes injected intravitreally with negative control or anti-Cyr61 locked nucleic acid (2 μL/eye, 150 μM). Each image is an en face maximal projection of a confocal z stack that spans the entire neovascular lesion across 12 or 13 slices. (B) Bar graph showing quantification of mean CYR61 fluorescence intensity. Bars for each group indicate mean ± SEM for images of n = 7–8 CNV lesions from five or six eyes from n = 3 mice. Statistical significance was assessed using a Mann-Whitney test. (C) Laser-induced CNV lesions in mice administered negative control or anti-Cyr61 locked nucleic acid via intravitreal injection (2 μL/eye, 150 μM). Images are en face views of the lesion on a flat-mounted choroid. Mice were injected with FITC-dextran to visualize neovascular lesions. The white dashed line circumscribes the CNV lesion. (D) Scatter dot plot showing relative CNV sizes in mice receiving intravitreal injection of negative control or anti-Cyr61 locked nucleic acid. Sizes of CNV lesions were averaged by eye and then normalized to the average negative control CNV size in each experiment. Line indicates mean ± SEM for 31 eyes from n ≥ 15 mice across three independent experiments. Statistical significance was assessed using a Mann-Whitney test. ∗p < 0.05. See also Figure S7.