Essential role for MFG-E8 as ligand for alphavbeta5 integrin in diurnal retinal phagocytosis

Abstract

The integrin receptor alphavbeta5 controls two independent forms of interactions of the retinal pigment epithelium (RPE) with adjacent photoreceptor outer segments that are essential for vision. Alphavbeta5 localizes specifically to apical microvilli of the RPE and contributes to retinal adhesion that maintains RPE contacts with intact outer segments at all times. Additionally, alphavbeta5 synchronizes diurnal bursts of RPE phagocytosis that clear photoreceptor outer segment fragments (POS) shed in a circadian rhythm. Dependence of retinal phagocytosis and adhesion on alphavbeta5 receptors suggests that the extracellular matrix ensheathing RPE microvilli contains ligands for this integrin. Here we studied mice lacking expression of functional MFG-E8 to test the contribution of this integrin ligand to alphavbeta5 functions in the retina. Lack of MFG-E8 only minimally reduced retinal adhesion. In contrast, lack of MFG-E8, like lack of alphavbeta5 receptor, eliminated alphavbeta5 downstream signaling involving the engulfment receptor MerTK and peak POS phagocytosis, both of which follow light onset in wild-type retina. MFG-E8-deficient RPE in primary culture retained normal epithelial morphology and levels of apical alphavbeta5 receptors, but showed impaired binding and engulfment of isolated POS. Soluble or POS-bound recombinant MFG-E8 was sufficient to fully restore phagocytosis by MFG-E8-deficient RPE. Furthermore, MFG-E8 supplementation strongly increased POS binding by wild-type and MerTK-deficient RPE, but did not affect POS binding by RPE lacking alphavbeta5. Thus, MFG-E8 stimulates rhythmic POS phagocytosis by ligating apical alphavbeta5 receptors of the RPE. These results identify MFG-E8 as the first extracellular ligand in the retina that is essential for diurnal POS phagocytosis.

Fig. 1.

Wild-type but not MFG-E8-mutant RPE cells express and secrete MFG-E8. Lysates prepared from whole eyecups (A), purified RPE (B), and CM collected from primary RPE cells 1 day after medium change (C) were immunoblotted for β-gal, MFG-E8, and the RPE cell marker RPE65 as indicated.

Fig. 2.

Retinal adhesion in MFG^−/− mice is decreased at the morning peak, but normal at other times. (A) Immunoblots of individual MFG^+/+ and MFG^−/− retina samples were probed for RPE proteins as indicated. Only RPE65 at 0930 was decreased by 50 ± 6% in isolated MFG^−/−, retina compared with MFG^+/+ retina (mean ± SD, n = 5; P < 0.001; all other proteins, changes P > 0.05). (B) Quantification of solubilized RPE pigment retrieved with neural retina demonstrated that 19 ± 1% less pigment fractionated with MFG^−/− retina (filled bars) at 0930 than with MFG^+/+ retina (gray bars) (mean ± SEM, n = 12; ∗, P < 0.05). Samples harvested at 0800 and 1700 contained equal amounts of RPE pigment regardless of MFG-E8 expression (mean ± SEM, n = 4 at 0800; n = 7 at 1700; P > 0.05). (C) Relative RPE pigment retrieval with β5^−/− (empty bars) and MFG^−/− retina (filled bars) was each compared with their respective control strains.

Fig. 3.

MFG^−/− mice lack the synchronized daily activation of MerTK and POS phagocytosis. (A and B) Comparative immunoblotting determined relative phosphorylation levels of MerTK in wild-type MFG^+/+, MFG^−/−, and MFG^+/− retina. Protein levels of MerTK, β5 integrin, αv integrin, or RPE65 did not vary among samples. Thirty minutes after light onset, MerTK tyrosine phosphorylation was increased by 3.3 ± 0.8-fold and 2.4 ± 0.6-fold in MFG^+/+ and MFG^+/− retina, respectively, compared with 1 h before light onset (mean ± SD, n = 3; ∗, P < 0.05). MerTK tyrosine phosphorylation did not significantly change in MFG^−/− retina during the same period (1.1 ± 0.8-fold change; n = 3; P > 0.1). (C and D) (Top) Opsin immunofluorescence (green) of paraffin sections of retinas showed that MFG^+/+ RPE (C) contained more phagosomes than MFG^−/− RPE (D) 30 min after light onset (red, retinal nuclei). Arrowheads show examples of phagosomes. (Middle) Areas marked by white rectangles are enlarged. (Bottom) Conversion of opsin labeling to black and white facilitated counting of opsin-positive phagosomes in the RPE layer. (Scale bars: Upper, 50 μm; Middle, 10 μm.) (E) Phagosome quantification at different times of day demonstrates peak number of phagosomes after light onset in MFG^+/+ RPE (gray bars) but not in MFG^−/− RPE (filled bars) or β5^−/− RPE (empty bars; shown previously by EM) (5).

Fig. 4.

MFG^−/− RPE cells in culture are phenotypically normal, but show impaired POS binding in the absence of exogenous factors. (A–D) Maximal projections of representative fields show similar appearance of double immunofluorescence labeling of adherens junction protein α-catenin (A and B) and β5 integrin (C and D) in primary, unpassaged RPE from MFG^+/− (A and C) and MFG^−/− (B and D) littermates. (Scale bars: 20 μm.) (E and F) Parallel samples and β5^−/− RPE were fed with isolated FITC-POS in DMEM with or without 4% FBS before fixation and quantification of bound (E) and internalized (F) POS. Omitting FBS reduced POS binding by MFG^−/− RPE to 47 ± 4% compared with MFG^+/− RPE cells (P < 0.01, Student's t test). β5^−/− POS binding was not significantly altered by FBS (P > 0.05). Bars represent mean numbers of bound and internal POS per MFG^+/− (gray bars), MFG^−/− (filled bars), and β5^−/− (light gray bars) RPE cell ± SD (n = 3).

Fig. 5.

Recombinant, soluble, or POS-bound MFG-E8 requires αvβ5 integrin to promote POS binding. (A and B) Addition of 0.25 μM MFG-E8 (B) increased total FITC-POS uptake (binding plus engulfment) by RPE-J cells, compared with no additive (A) as seen in maximal projections of x–y confocal stacks (Upper) and in single x–z confocal scans (Lower). Green, FITC-POS; red, RPE-J nuclei. (Scale bars: 10 μm.) (C–F) Fluorescence quantification showed that soluble MFG-E8 increased POS binding (C) and internalization (D) in a dose-dependent manner. High doses inhibited both phases of uptake. Addition of MFG-E8 increased total POS uptake of primary MFG^−/− RPE cells (F), compared with no additive (E). Green, FITC-POS; red, RPE tight junction protein ZO-1; blue, RPE nuclei. (Scale bars: 10 μm.) (G) Fluorescence scanning quantified FITC-POS uptake by RPE as indicated with (+) and without (−) 0.25 μM soluble MFG-E8. MFG-E8 significantly increased POS binding (compare filled top of + bars with dark gray top of − bars) by MFG^+/+ RPE (mean change 1.8-fold; P < 0.05), MFG^−/− RPE (3.1-fold; P < 0.05), and MerTK-deficient RCS RPE (3.0-fold; P < 0.05), but not by β5^−/− RPE (1.1-fold; P > 0.1). (H) POS-bound recombinant MFG-E8 increased POS binding by MFG^−/− RPE (2.7-fold; P < 0.05). Addition of CM retrieved from MFG-E8-coated POS after 1 h of mock incubation did not change MFG^−/− RPE activity [compare − and + bars (Left) with − and + bars (Right)]. All bars represent mean ± SD (n = 3). Significance was determined comparing values with and without MFG-E8 using Student's t test. POS internalization did not significantly change (P < 0.05 for all samples; light bottom parts of bars).