The Wnt Inhibitor Apcdd1 Coordinates Vascular Remodeling and Barrier Maturation of Retinal Blood Vessels

Abstract

Coordinating angiogenesis with acquisition of tissue-specific properties in endothelial cells is essential for vascular function. In the retina, endothelial cells form a blood-retina barrier by virtue of tight junctions and low transcytosis. While the canonical Norrin/Fz4/Lrp5/6 pathway is essential for angiogenesis, vascular remodeling, and barrier maturation, how these diverse processes are coordinated remains poorly understood. Here we demonstrate that Apcdd1, a negative regulator of Wnt/β-catenin signaling, is expressed in retinal endothelial cells during angiogenesis and barrier formation. Apcdd1-deficient mice exhibit a transient increase in vessel density at ages P10-P12 due to delayed vessel pruning. Moreover, Apcdd1 mutant endothelial cells precociously form the paracellular component of the barrier. Conversely, mice that overexpress Apcdd1 in retina endothelial cells have reduced vessel density but increased paracellular barrier permeability. Apcdd1 thus serves to precisely modulate Wnt/Norrin signaling activity in the retinal endothelium and coordinate the timing of both vascular pruning and barrier maturation.

Keywords: Apcdd1; Caveolin-1; Claudin-5; Frizzled-4; Norrin; Occludin; Wnt/β-catenin signaling; angiogenesis; blood-retina barrier; tight junction; transcytosis; vascular pruning.

Figure 1. Apcdd1 Is Expressed in Retinal Blood Vessels during Angiogenesis and Blood-Retina Barrier Formation

(A–H) Representative images of RNA in situ hybridizations for Apcdd1 and Claudin-5 in sections of P6 (A and B), P8 (C and D), P12 (E and F), and P17 (G and H) wild-type mouse retinas (n = 3 mice per time point). Apcdd1 is expressed by blood vessels in superficial and deep plexi (white arrows), as well as by neural progenitors/neurons (yellow arrows) in the NBL and INL. (I–L) Fluorescent RNA in situ hybridization for Apcdd1 and immunohistochemistry for Caveolin-1, an EC marker, in P12 (I and J) and P17 (K and L) wild-type retinas (n = 3 mice/time point). Apcdd1 colocalizes with Caveolin-1, indicating expression in the endothelium (I–L; white arrows). (J and L) Higher magnification of boxed regions in (I) and (K), respectively. (M–P) Double fluorescent RNA in situ hybridization for Apcdd1 (M and O, green) and merged panels (N and P; Apcdd1 green and PDGRFβ red) in P12 (M and N) and P17 (O and P) wild-type retinas. Apcdd1 colocalizes with Caveolin-1 but very little with PDGFRβ, indicating expression in the endothelium (I–P; white arrows) but not pericytes (M–P; yellow arrows). Apcdd1 mRNA is also expressed by non-endothelial cells within the INL (I and K; yellow arrows). Scale bars represent 50 μm in (A)–(H), 25 μm in (I), (K), and (M)–(P), and 15 μm in (J) and (L). GCL, ganglion cell layer; IPL, inner plexiform layer; NBL, neuroblast layer; OPL, outer plexiform layer; ONL, outer nuclear layer.

Figure 2. Loss of Apcdd1 Function Transiently Increases Vessel Density

(A–D) Immunohistochemistry for BSL-FITC (vessel marker) in wild-type and Apcdd1^−/− P5 retinas. (A) Wild-type and (C) Apcdd1^−/− retinas show comparable sprouting patterns at the leading edge of the leaflet. (B and D) Higher magnification of boxed regions in (A) and (C), respectively, illustrating sprouting ECs (white arrows). (E–L) P10 and P12 whole-mount retinas from both genotypes stained for BSL-FITC. Apcdd1^−/− retinas (G) and (K) have an increased vessel density in perivenous capillaries compared to wild-type controls (E and I). (F, H, J, and L) Higher magnification of boxed regions in (E), (G), (I), and (K), respectively. (M) Quantification of sprouts at vascular fronts for wild-type and Apcdd1^−/− retinas at P5 (n = 4 animals/group). (N) Quantification of vessel densities in the superficial plexus of wild-type and Apcdd1^−/− mice from P5 to P14 (n = 3–10 animals/group). Apcdd1 mutants have a transient increase in vessel density at P10 and P12. (O and P) Quantification of artery (O) and vein (P) diameters for wild-type and Apcdd1^−/− mice aged P8 to P14 (n = 3–10 animals/group). Apcdd1 mutants show increased venous diameters at P10 (P). Scale bars represent 200 μm in (A), (E), and (I), 50 μm in (F) and (J), and 25 μm in (B). A, artery; V, vein. Data are presented as mean ± SEM, *p < 0.05 **p < 0.01, Student’s unpaired t test.

Figure 3. Loss of Apccd1 Function Reduces Vessel Pruning

(A–H) Immunofluorescence for BSL-FITC and EdU in whole-mount retinas from wild-type littermates and Apcdd1^−/− mice. Representative images for EdU⁺ ECs (C, D, G, and H; yellow arrows) in P10 wild-type (A–D) and Apcdd1^−/− (E–H) retinas. (C, D, G, and H) Higher magnification of boxed regions in (A), (B), (E), and (F), respectively. (I–P) Immunofluorescence for BSL-rhodamine and Collagen IV in whole-mount P10 retinas from wild-type (I–L) and Apcdd1^−/− (M–P) mice. Empty matrix sleeves (J–L and N–P; white arrows) are defined as Collagen IV⁺ (green) and BSL-rhodamine⁻ (red). (J, L, N, and P). Higher magnification of boxed regions in (I) and (M), respectively. (Q) Quantification of EdU⁺ ECs per vessel area (n = 3 animals/group). (R) Quantification of arterial and venous EdU⁺ ECs per vessel length (n = 3 animals/group). Apcdd1^−/− mice have normal EC proliferation. (S) Quantification of empty matrix sleeves per vessel area at P10 and P14 (n = 3–7 animals per group). Apcdd1^−/− retinas have fewer empty matrix sleeves as compared to wild-type. Scale bars represent 200 μm in (A) and (I), 50 μm in (D), and 25 μm in (J). Data are presented as mean ± SEM, **p < 0.01, Student’s unpaired t test.

Figure 4. Loss of Apcdd1 Function Decreases Paracellular BRB Permeability

(A) Diagram for measuring paracellular permeability via leakage of biocytin tetramethylrhod-amine (TMR). (B–G) Representative images of wild-type (B, D, and F) and Apcdd1^−/− (C, E, and G) retinas at P10, P14, and P18 following injection of biocytin-TMR. Tracer leakage appears as intense bright puncta in the retinal parenchyma (B–G; white arrows). Apcdd1^−/− retinas exhibit less tracer leakage at P10 and P14 compared to wild-type controls. (H) Quantitation of signal intensities for biocytin-TMR leakage outside the retinal vasculature, normalized to the liver (n = 3–5 animals/group). More tracer leakage is seen in wild-type compared to Apcdd1^−/− retinas at P10 and P14. (I–P) Immunofluorescence images for Claudin-5 (I and M; red), Claudin-5 (red) and BSL (green) (J and N), Occludin (K and O; red), and Occludin (red) and BSL (green) (L and P) in the superficial plexi of the wild-type and Apcdd1^−/− retinas at P14. Claudin-5 and Occludin are localized to cell junctions in both genotypes (I–P; white arrows). Note the absence of Claudin-5 in regressing vessels (I, J, M, and N; yellow arrows). (Q) Western blot analysis for TJ proteins ZO-1 and Occludin and adherens junction protein Cadherin-5 in P14 and P18 retinal lysates. (R) Quantification of ZO-1, Occludin, and Cadherin-5 protein levels using LICOR. At P14, Apcdd1^−/− retina lysates show a dramatic increase in Occludin (n = 3–4 replicates/group). Scale bar represents 50 μm in (B)–(G). Data in (H) are presented as mean ± SEM, whereas in (R) as mean ± SD; *p < 0.05. **p < 0.01, Student’s unpaired t test.

Figure 5. Transcellular BRB Permeability Is Unaffected in Apcdd1 Mutants

(A) Diagram outlining albumin endocytosis in retinal ECs. (B–I) Representative images of wild-type (B, C, F, and G) and Apcdd1^−/− (D, E, H, and I) retinas at P10 and P14 following injection of albumin-Alexa Fluor 488. Apcdd1^−/− retinas have normal albumin uptake compared to wild-type retinas at both stages (C, E, G, and I; white arrows). (J and K) Quantification of intra-endothelial (J) and parenchymal (K) albumin normalized to liver (n = 3–4 animals/group). (L and M) Western blot analysis (L) and quantification (M) for Caveolin-1 and PLVAP (n = 3–4 replicates/group). Apcdd1^−/− retinal lysates have no significant differences in Caveolin-1 and PLVAP levels. Scale bars represent 50 μm in (B) and 25 μm in (C). Data in (J) and (K) are presented as mean ± SEM, and in (M) as mean ± SD; Student’s unpaired t test.

Figure 6. Apcdd1^–/– Brain ECs Display Decreased Paracellular but Increased Transcellular Permeability In Vitro

(A and B) Immunofluorescence for Claudin-5 and ZO-1 in wild-type (A) and Apcdd1^−/− (B) BECs. There is no detectable difference in junctional localization for either protein in wild-type or Apcdd1^−/− BECs. (A and B; white arrows). (C) Graph of trans-endothelial electrical resistance (TEER) in monolayers of wild-type or Apcdd1^−/− BECs. Green area indicates the period of TEER measurement after incubation in low serum (1%) media (red arrow indicates start of TEER measurement). (D) Quantitation of the areas under the curve for the green region depicted in (C) (n = 10 independent experiments). Apcdd1^−/− BECs show increased TEER compared to wild-type BECs. (E and F) Western blot and quantification for Claudin-5, ZO-1, and Occludin in wild-type and Apcdd1^−/− BECs. Apcdd1^−/− BECs have increased Occludin expression (n = 5 independent experiments). (G and H) Distribution of albumin-Alexa Fluor 594 puncta in wild-type (G; white arrows) and Apcdd1^−/− (H; white arrows) BECs following incubation to measure transcellular permeability. (I) Quantification of EC-associated albumin in wild-type and Apcdd1^−/− BECs. Apcdd1^−/− BECs have a significant increase in albumin uptake (n = 5 independent experiments). (J and K) Western blot (J) and quantitation (K) for Caveolin-1, Cavin-2, and PLVAP (n = 3–5 independent experiments). Apcdd1^−/− BECs show increased PLVAP expression. Scale bars represent 15 μm. Data in (D) and (I) are presented as mean ± SEM and in (F) and (K) as mean ± SD; *p < 0.05 **p < 0.01, Student’s unpaired t test.

Figure 7. Endothelial Apcdd1 Overexpression Results in Decreased Vessel Density but Delayed Paracellular Barrier Formation in the Retina

(A) Schematic diagram for overexpression of Apcdd1 in ECs. (B–E) Representative images of BSL-FITC and mCherry expression in TRE3-Apcdd1-IRES-mCherry⁺ (sTg) (B and D) and Cadherin5-tTA⁺; TRE3-Apcdd1-IRES-mCherry⁺ (dTg) (C and E) P10 retinas. mCherry is expressed in the vasculature of dTg (C and E) but not sTg (B and D) animals. (D and E) Higher magnifications of boxed regions in (B) and (C), respectively. (F and G) Representative images of sTg (F) and dTg (G) retinas at P14 following injection of biocytin-TMR. Tracer leakage appears as intense bright puncta in the retinal parenchyma (F and G; white arrows). (H) Quantification of vessel densities in the superficial plexus of sTg and dTg P10 retinas (n = 5–8 animals/group). dTg mice have decreased vessel density at P10. (I) Quantitation of signal intensities for biocytin-TMR leakage outside the retinal vasculature, normalized to the liver (n = 4–6 animals/group). Doubly transgenic retinas show increased tracer leakage compared to singly transgenic retinas at P14. (J) Diagrams of canonical Wnt signaling activity and tight junction structure in non-regressing sealed vessels and regressing leaky vessels for wild-type and Apcdd1^−/− retinas. In wild-type mice, Apcdd1 reduces Wnt levels from high (red) to intermediate (orange) in a subset of vessels that will undergo regression. Furthermore, Apcdd1 prevents formation of mature tight junctions (three lines) in vessels destined for regression. In Apcdd1^−/− retinas, more vessels have high levels of Wnt signaling and thus form a mature barrier but do not undergo regression. Therefore, Apcdd1 precisely sculpts the levels of Wnt activity in retina ECs to coordinate vessel pruning and barrier maturation. Scale bars represent 200 μm in (B) and (C), 50 μm in (D) and (E), and 25 μm in (F) and (G). sTg, single transgenic; dTg, double transgenic. Data are represented as mean ± SEM; *p < 0.05, Student’s unpaired t test.