Conditional β-cat GOF or LOF in ECs in vivo modulates Cldn3 and Plvap expression and permeability in brain capillaries. P14 brain sections of 4OHT-treated LOF mice stained for Cldn3 or Plvap together with IB4. (A, a and g) Cre⁻ control mice. Arrowheads point to Cldn3-positive and Plvap-negative vessels, respectively. (d and j) Cre⁺ LOF resulted in Cldn3 down- and Plvap up-regulation in ∼50% of brain capillaries. Arrowheads point to Cldn3-negative and Plvap-positive vessels, respectively. β-Cat GOF in ECs in vivo by 4OHT treatment. E18.5 and P4 brain sections analyzed for Cldn3 or Plvap (green) together with IB4 (red). (b, c, h, and i) Cre⁻ control mice. Arrowheads point to weak Cldn3-positive vessels at E18.5 and a slightly higher level at P4 or strong Plvap at E18.5 and a slightly weaker level at P4. (e, f, k, and l) Cldn3 increase in Cre⁺ mice at E18.5 and P4. Plvap was down-regulated. Left panels show merge of IB4 and Cldn3 or Plvap; right panels show Cldn3 or Plvap alone in black and white. Insets show higher magnifications. (B) Evans blue injection i.p. 18 h before tissue harvesting in LOF mice. Arrowheads point to extravasated Evans blue in Cre⁺ mice at P14. (C) Quantitative RT-PCR for Plvap at P7 in LOF (n = 3; **, P = 0.0001) and GOF (n = 3; *, P = 0.0382). Cldn3 was statistically NS. Control is set to 100% (dashed line). Error bars represent SEM.

β-Cat LOF and GOF in MBEs in vitro. β-Cat^lox/lox and β-cat^{lox(ex3)/lox(ex3)} MBEs were treated with TAT-Cre in vitro to achieve β-cat LOF and GOF, respectively. (A, a–c and f–h) β-Cat^lox/lox with or without TAT-Cre, stimulated with controlCM (CCM) or Wnt3aCM. IF staining reveals increased junctional Cldn3 in Cre⁻/Wnt3aCM and a complete absence of β-cat in Cre⁺/Wnt3aCM. (d, e, i, and j) β-Cat^{lox(ex3)/lox(ex3)} MBEs with or without TAT-Cre, and IF stained for Cldn3 and β-cat. Arrowheads point to nuclear β-cat. (B) Quantitative RT-PCR for Cldn3 in Cre⁻/Wnt3aCM-treated cells from β-cat^lox/lox (n = 3; **, P = 0.0048) and in Cre⁺/controlCM cells from β-cat^{lox(ex3)/lox(ex3)} (n = 3; *, P = 0.0341) mice. Other tested genes were not significantly altered. Axin2 induction confirmed canonical Wnt signaling in Cre⁻/Wnt3aCM-treated cells from β-cat^lox/lox (n = 3; **, P = 0.0006) and in Cre⁺/controlCM cells from β-cat^{lox(ex3)/lox(ex3)} (n = 3; **, P = 0.0029) mice. Controls are set as 100% (dashed lines). Error bars represent SEM. (C) Western blot analysis for Cldn3 in β-cat^lox/lox and β-cat^{lox(ex3)/lox(ex3)} cells. Transcription and translation of an N-terminal truncated form of β-cat are detected by the anti–β-cat antibody and represented by the bottom band in β-cat^{lox(ex3)/lox(ex3)}/Cre⁺ cells. See graphs for densiomeric quantification of Western blot bands. Data are representative of at least three independent experiments.

Canonical Wnt signaling is active in ECs during brain angiogenesis and becomes progressively down-regulated during vessel maturation. (A) LacZ whole-mount stained (blue) E9.5 BAT-Gal embryos sectioned and counterstained for isolectin B4 (IB4; red). Panel b is a higher magnification of the boxed area in panel a. Arrows point to nuclear LacZ reporter gene staining. (B, a–d) Whole-mount hindbrain staining for LacZ (reflection; red) and IB4 (green) of BAT-Gal embryos (E13.5 and 17.5) analyzed by confocal microscopy. Arrowheads indicate LacZ-positive nuclei. (B, e–h) Staining of brain cryosections from postnatal BAT-Gal pups (P1 and P5) for LacZ (immunofluorescence [IF], red) and IB4 (green). Arrowheads indicate LacZ-positive nuclei. Positive nuclei outside the vascular system indicate active Wnt signaling in the brain parenchyma. (C) Quantification of LacZ-positive nuclei per 100-μm vessel length shows a significant decrease from E15.5 to 17.5 (five fields per hindbrain; three brains; **, P = 0.0003). Error bars represent SEM.

Wnt3aCM but not Wnt5aCM induced TJs and barrier properties in MBEs. Wild-type MBEs treated with controlCM, Wnt3aCM, or Wnt5aCM in vitro. (A, a, e, and i) IHC staining for Cldn3. (b, f, and j) IHC staining for Cldn5. Panels are the same magnification as indicated in panel k. (c, g, and k) IHC staining for active β-cat (clone 8E7). Arrowheads point to positive nuclei. (d, h, and l) Freeze-fracture EM. Arrowheads point to P-face–associated TJ ridges. (B) mRNA levels for Cldn1, Cldn3, Cldn5, Cldn12, Ocln, Plvap, and Axin2 measured by quantitative RT-PCR. Cldn3 (n = 3; *, P = 0.0319), Plvap (n = 3; *, P = 0.0098), and Axin2 (n = 3; **, P = 0.005); other tested genes were not significantly altered. Control is set to 100% (dashed line). (C) Western blotting for Cldn3 and Cldn5. α-Tubulin/tubulin α1A served as a loading control. Western blotting for 8E7 and total β-cat. No band could be detected in Wnt5aCM-treated cells. See graphs for densiomeric quantification of bands. Error bars represent SEM. Data are representative of at least three independent experiments.

Cldn3 up-regulation requires transcriptionally active rather than junctional β-cat in MBEs in vitro. (A, a–c; and B, a) MBEs were either treated with Wnt3aCM or control CM and infected with dnTCF4 adenovirus or a control. (A, a–c) IF stainings for Cldn3. dnTCF4 abrogated the effect of Wnt3aCM on junctional Cldn3. (B, a) Quantitative RT-PCR for Cldn3 (n = 3; **, P = 0.0095), Plvap (n = 3; *, P = 0.035), and Axin2 (n = 3; **, P = 0.004) with Wnt3aCM and for Plvap (n = 3; **, P = 0.002) with Wnt3aCM and dnTCF4. (A, d; and B, b) MBEs infected with LefΔN-βCTA lentivirus or a control. Junctional localization of Cldn3 increased in IF stainings. Quantitative RT-PCR for Cldn3 (n = 3; **, P = 0.002), Plvap (n = 3; **, P = 0.008), and Axin2 (n = 3; **, P < 0.0001). Controls are set as 100% (dashed lines). Green nuclei show an anti–V5-TAG (c) and an anti–HA-TAG (d) staining of dnTCF4 and LefΔN-βCTA, respectively, confirming infection of the target cells. Error bars represent SEM.