Hypoxia Inhibits Subretinal Inflammation Resolution Thrombospondin-1 Dependently

Abstract

Hypoxia is potentially one of the essential triggers in the pathogenesis of wet age-related macular degeneration (wetAMD), characterized by choroidal neovascularization (CNV) which is driven by the accumulation of subretinal mononuclear phagocytes (MP) that include monocyte-derived cells. Here we show that systemic hypoxia (10% O₂) increased subretinal MP infiltration and inhibited inflammation resolution after laser-induced subretinal injury in vivo. Accordingly, hypoxic (2% O₂) human monocytes (Mo) resisted elimination by RPE cells in co-culture. In Mos from hypoxic mice, Thrombospondin 1 mRNA (Thbs1) was most downregulated compared to normoxic animals and hypoxia repressed Thbs-1 expression in human monocytes in vitro. Hypoxic ambient air inhibited MP clearance during the resolution phase of laser-injury in wildtype animals, but had no effect on the exaggerated subretinal MP infiltration observed in normoxic Thbs1^-/--mice. Recombinant Thrombospondin 1 protein (TSP-1) completely reversed the pathogenic effect of hypoxia in Thbs1^-/--mice, and accelerated inflammation resolution and inhibited CNV in wildtype mice. Together, our results demonstrate that systemic hypoxia disturbs TSP-1-dependent subretinal immune suppression and promotes pathogenic subretinal inflammation and can be therapeutically countered by local recombinant TSP-1.

Keywords: age-related macular degeneration; choroidal neovascularization; hypoxia; macrophages; mononuclear phagocytes; thrombospondin 1.

Figure 1

Hypoxia increases Mo resistance to RPE-induced elimination. (A) schematic representation of the experiments: mice were laser injured and either exposed to 10% O₂ hypoxia from d0–d4 and evaluated for the presence of subretinal IBA1⁺MPs at d4 or exposed to 10% O₂ hypoxia from d4–d10 and evaluated for MPs and the extent of Collagen 4⁺CNV at d10. Mice raised in normal 20.9% ambient O₂ served as controls. (B) Representative images of Collagen 4 (Coll4; red) and IBA-1 (green) immuno- stained RPE/choroid flatmounts of 10 days post laser-injured normoxic and hypoxic (d4–d10) mice. (C,D) Quantification of the density of subretinal IBA-1+ MPs per impact: (B) directly within the RPE-denuded lesion and (C) counted at a distance of 0–500μm to Coll4⁺CNV on the apical side of the RPE surrounding the lesion at d4 and d10 of normoxic and hypoxia-exposed 2-month-old mice (room air white columns, 10% O₂ hypoxia d0-d4 orange columns; 10% O₂ hypoxia d4–d10 blue columns) (n = 9–10 eyes; ^$ p = 0.0503; * p < 0.0001 Mann–Whitney versus their normoxic controls). (E) Quantification of the Coll4⁺CNV surface at d10 of normoxia and hypoxia-exposed (d4–d10) laser-injured mice (n = 9–10 eyes; * p = 0.0060 Mann–Whitney versus normoxic control). (F) Representative pictures of PU-1 OTX-2 co-stained co-cultures of PU1⁺human monocytes (green) and OTX-2⁺RPE cells (red) under normoxic-(20% O₂; white columns) and hypoxic-(2% O₂; black columns) conditions and their automated quantifications after 24 h co-culture (n = 5 wells; * p = 0.0079 Mann–Whitney versus the normoxic condition), (G) quantifications of PU1⁺human monocytes after 24 h of hypoxic (black column) or normoxic (white column) pre-incubation of Mo followed by 24 h of Mo/RPE coculture (n = 5 wells; * p = 0.0079 Mann–Whitney versus the normoxic condition) (H) automated quantifications of PU1⁺human monocytes after 24 h of hypoxic (black column) or normoxic (white column) pre-incubation of RPE followed by 24 h of Mo/RPE coculture. The in vivo results presented in (B–E) summarize two independently carried out experiments, the in vitro experiments (F–H) were repeated a minimum of five times and gave similar results. All values are reported as mean  ±  SEM. IBA-1: ionized calcium adapter molecule 1; Coll4: Collagen 4; PU1: hematopoietic transcription factor; OTX-2: Orthodenticle Homeobox 2; scale bar A = 400 μm, E = 500 μm.

Figure 2

Hypoxia decreases Thbs1- expression in Mos. (A) Scatter dot blot of the protein coding mRNAs of FACS sorted CD45⁺CD11B⁺Ly6G^negLy6C^high Mos from bone marrows of five 3-month-old mice that were raised in room-air or five age-matched mice that had been exposed to 10% ambient O₂ for 40 h. Only transcripts with a TPM greater than 100 in the normoxia group and a false discovery rate (FDR) smaller than 0.05 are depicted. The transcripts are plotted according to their expression levels (y-axis) and the log2-fold induction by hypoxia (x axis). The identified 18 transcripts that were more than two-fold overexpressed and the 19 transcripts that were more than two-fold under-expressed in Mos from hypoxic mice compared to Mos from room air raised mice are indicated. (B) Quantitative RT-PCR of Retina/RPE/Choroid, magnet sorted retinal microglial cells (MC), and magnet sorted bone marrow monocytes (BM-Mo) from room-air-(white columns) and 40 h hypoxia- (10% O₂; black columns) exposed mice. (n = 5 mice/group * p = 0.0032; Mann–Whitney versus the normoxic control). (C) Quantitative RT-PCR of human Mo exposed to normoxic culture conditions (white columns) or the indicated hours of 2% O₂ (black columns) (n = 5 wells/group, representative of two independent experiments, * p = 0.0079 (3 h) and 0.0159 (12 h); Mann–Whitney versus the normoxic control). All values are reported as mean  ±  SEM. Thbs1: Thrombospondin 1.

Figure 3

TSP-1 is necessary for hypoxia-induced inhibition of inflammation resolution after laser-injury. (A) Representative images of Collagen 4 (Coll4; red) and IBA-1 (green) immuno-stained RPE/choroid flatmounts of 10 day post laser-injured 2-month-old mice of the indicated strains, raised in 20.9% ambient O₂ (normoxia, white columns) or exposed to 10% O₂ hypoxia (d4–d10, blue columns) and quantification of subretinal IBA-1⁺MPs per impact on the RPE counted at a distance of 0–500μm to Coll4⁺CNV of the indicated strains (n = 9–10 eyes; * p = 0.0002; ^† p = 0.0114; Mann–Whitney versus the normoxic C57BL6/J mice, hypoxic Thbs1^−/− mice were not significantly different from hypoxic C57BL6/J mice). (B) Quantification of the Coll4⁺CNV surface at d10 of normoxic and hypoxia-exposed (d4–d10) laser-injured mice of the indicated strains (n = 9–10 eyes; * p < 0.0001; ^† p = 0.0114; Mann–Whitney versus the normoxic C57BL6/J mice). (C) Representative images of Coll4 (red) and IBA-1 (green) immuno- stained RPE/choroid flatmounts of 10 day post laser-injured, 2 month-old, wildtype mice exposed to 10% O₂ hypoxia (d4–d10) that were intravitreally injected at d4 and d7 with 2µL of PBS, or 2µL of PBS containing recombinant Thrombospondin 1 protein (TSP-1, 10µg/mL) and the quantification of subretinal IBA-1⁺MPs per impact on the RPE counted at a distance of 0–500μm to Coll4⁺CNV at d10 (n = 9–10 eyes; * p = 0.0001; versus the hypoxic PBS-treated mice). (D) CNV surface at d10 of the treated, hypoxia-exposed (d4–d10), laser-injured mice (n = 9–10 eyes; * p = 0.00143; Mann–Whitney versus the hypoxic PBS-treated mice). All values are reported as mean  ±  SEM. PBS: Phosphate buffered saline; rTSP-1: Recombinant thrombospondin 1; scale bar = 400 μm.