Thrombospondin 1 inhibits inflammatory lymphangiogenesis by CD36 ligation on monocytes

Abstract

Lymphangiogenesis plays an important role in tumor metastasis and transplant outcome. Here, we show that thrombospondin-1 (TSP-1), a multifunctional extracellular matrix protein and naturally occurring inhibitor of angiogenesis inhibits lymphangiogenesis in mice. Compared with wild-type mice, 6-mo-old TSP-1-deficient mice develop increased spontaneous corneal lymphangiogenesis. Similarly, in a model of inflammation-induced corneal neovascularization, young TSP-1-deficient mice develop exacerbated lymphangiogenesis, which can be reversed by topical application of recombinant human TSP-1. Such increased corneal lymphangiogenesis is also detected in mice lacking CD36, a receptor for TSP-1. In these mice, repopulation of corneal macrophages with predominantly WT mice via bone marrow reconstitution ameliorates their prolymphangiogenic phenotype. In vitro, exposure of WT macrophages to TSP-1 suppresses expression of lymphangiogenic factors vascular endothelial growth factor (VEGF)-C and VEGF-D, but not of a primarily hemangiogenic factor VEGF-A. Inhibition of VEGF-C is not detected in the absence or blockade of CD36. These findings suggest that TSP-1, by ligating CD36 on monocytic cells, acts as an endogenous inhibitor of lymphangiogenesis.

Figure 1.

TSP-1^−/− mice display mild isolated lymphangiogenesis. (A–D) Immunofluorescence analysis of LYVE-1⁺⁺⁺/CD31⁺ lymphatic vessels (L, arrows) emanating from the limbal arcade (Li) toward the center of the cornea in 6-mo-old TSP-1^−/− (A-C) or WT (control, D) mice. Blood vessels (B, arrowheads) in the limbal arcade stain intensely with anti-CD31, but not with anti–LYVE-1. (E) Electron micrograph of the corneal stromal lymphatic vessel in TSP-1^−/− mice showing erythrocyte-free lumen (Lu). Arrowheads denote the thin endothelial lining with absent basement membrane and absent pericyte coverage. Inset shows higher magnification view. (F and G) WT control and TSP-1^−/− complete corneal flat-mounts stained with LYVE-1 (Li, limbus, arrowhead; L, lymphatic vessel, arrow). Note isolated lymphatic vessel traversing the whole cornea in G (L, arrow). (H) Morphometric comparison demonstrates a significantly increased spontaneous corneal lymphangiogenesis in TSP-1^−/− mice compared with WT controls (*, P < 0.05; n = 10; all mice aged 6 mo). (I) Morphometric quantification of lymphatic vessels in the ear skin from young (p10) or adult (p30) WT and TSP-1^−/− mice (*, P < 0.05; n = 5 per group, compared with WT controls). Error bars indicate SD.

Figure 2.

TSP-1 regulates the lymphangiogenic response in a mouse model of inflammation-associated corneal neovascularization. Representative segments of corneal flat mounts from WT and TSP-1^−/− mice stained with CD31 are shown in green (blood vessels) and LYVE-1 is shown in red (lymphatic vessels; A; Li, Limbal vascular arcade: bottom, center of the cornea: top; B; L, lymph vessels). (C) Morphometric quantification of lymphvascularized area in WT and TSP-1^−/− mice (*, P < 0.05; n = 10 per group; age 8 wk). (D) Morphometric analysis of vascularized areas with lymphatic vessels in sutured corneas of WT mice treated with three subconjunctival injections and topical eye drops (x3 daily) of recombinant human (rh) TSP-1 in comparison to control (*, P < 0.05; n = 8 per group, age 8 wk). Error bars indicate SD.

Figure 3.

Increased transcription of the lymphangiogenic growth factor VEGF-C in TSP-1^−/− mice. VEGF-C mRNA levels in corneas of 8-wk-old (A) and 6-mo-old (B) TSP-1^−/− mice in comparison to WT controls. (C) Flow cytometric detection of CD11b⁺ macrophages in the corneas harvested from 6-mo-old TSP-1^−/− mice (24.6%) compared with WT controls (19.7%). (D) Expression of VEGF-C in thioglycollate-elicited macrophages from TSP-1^−/− mice (age 8 wk) compared with WT mice detected by real-time PCR. Corneas were pooled from five mice per group in each experiment. Expression of VEGF-C relative to the GAPDH in four PCR reactions per RNA sample is indicated. *, P < 0.05 compared with WT control. Error bars indicate SD.

Figure 4.

Increased inflammation-induced lymphangiogenesis in CD36^−/− mice. Representative segments of corneal flat mounts from WT (A) and a CD36^−/− (B) mice. (C) Morphometric comparison of lymphvascularized area between WT control and CD36^−/− mice (*, P < 0.05; n = 10 per group; age 8 wk). Li, limbal vascular arcade bottom; center of the cornea, top; flat mount stained with LYVE-1, red (lymphatic vessels); L, lymph vessels. Error bars indicate SD.

Figure 5.

CD36 on macrophages regulates the lymphangiogenic response in the cornea. (A) The macrophage marker CD11b (green) and the TSP-1 receptor CD36 (red) are colocalized (yellow) on monocytic cells in the normal corneal stroma. (B) The specific lymphatic marker LYVE-1 (red) is not colocalized with antibodies against CD36 (green), and is expressed by blood vessels detectable at the limbus. (C) Morphometric comparison of corneal lymphvascularized area between WT (n = 8) control and CD36^−/− (n = 9) mice (reconstituted with WT bone marrow) in a mouse model of inflammatory corneal neovascularization. Error bars indicate SD.

Figure 6.

TSP-1 down-regulates expression of VEGF-C in macrophages via CD36. Relative expression of VEGF-C (A, B, D, and E), VEGF-D (F), and VEGF-A (G) message and VEGF-C protein (C) in thioglycollate-elicited macrophages from WT or CD36^−/− mice: (A) freshly harvested, (B–G) cultured in the absence or presence of TGFβ2, TSP-1, or CD36 blocking and control peptides. Total RNA harvested from macrophages was subjected to real-time PCR analysis to detect expression of VEGF-C, VEGF-D, and VEGF-A mRNA and to detect VEGF-C protein culture supernatants were analyzed by ELISA. Although freshly harvested CD36^−/− macrophages express significantly lower levels of VEGF-C compared with WT controls (A) these macrophages retain their ability to respond to exogenously provided TGFβ2 in culture by increasing expression of VEGF-C (both message [B] and protein [C]) compared with WT controls. Inhibition of VEGF-C expression detectable in WT macrophages cultured with TSP-1 (5 µg/ml) is not detected in similarly treated CD36 ^−/− macrophages (D). In the presence of CD36-binding TSP-1–derived peptide capable of blocking CD36–TSP-1 interaction (CD36 blocking peptide), the inhibitory effect of TSP-1 on the VEGF-C expression in WT macrophages is significantly reversed (E). Treatment of WT macrophages to TSP-1 also inhibited the expression of the lymphangiogenesis factor VEGF-D (F), but not that of predominantly hemangiogenic factor VEGF-A (G). Expression of VEGF-C, VEGF-D, and VEGF-A relative to the GAPDH in four independent PCR reactions per RNA sample is indicated. *, P < 0.05 compared with WT, untreated or indicated control. Error bars indicate SD.