VE-Cadherin Phosphorylation Regulates Endothelial Fluid Shear Stress Responses through the Polarity Protein LGN

Abstract

Fluid shear stress due to blood flow on the vascular endothelium regulates blood vessel development, remodeling, physiology, and pathology [1, 2]. A complex consisting of PECAM-1, VE-cadherin, and vascular endothelial growth factor receptors (VEGFRs) that resides at endothelial cell-cell junctions transduces signals important for flow-dependent vasodilation, blood vessel remodeling, and atherosclerosis. PECAM-1 transduces forces to activate src family kinases (SFKs), which phosphorylate and transactivate VEGFRs [3-5]. By contrast, VE-cadherin functions as an adaptor that interacts with VEGFRs through their respective cytoplasmic domains and promotes VEGFR activation in flow [6]. Indeed, shear stress triggers rapid increases in force across PECAM-1 but decreases the force across VE-cadherin, in close association with downstream signaling [5]. Interestingly, VE-cadherin cytoplasmic tyrosine Y658 can be phosphorylated by SFKs [7], which is maximally induced by low shear stress in vitro and in vivo [8]. These considerations prompted us to address the involvement of VE-cadherin cytoplasmic tyrosines in flow sensing. We found that phosphorylation of a small pool of VE-cadherin on Y658 is essential for flow sensing through the junctional complex. Y658 phosphorylation induces dissociation of p120ctn, which allows binding of the polarity protein LGN. LGN is then required for multiple flow responses in vitro and in vivo, including activation of inflammatory signaling at regions of disturbed flow, and flow-dependent vascular remodeling. Thus, endothelial flow mechanotransduction through the junctional complex is mediated by a specific pool of VE-cadherin that is phosphorylated on Y658 and bound to LGN.

Keywords: VEGF receptor; flow signaling; mechanotransduction; p120 catenin; vascular biology.

Figure 1. VE-cadherin Tyr658 phosphorylation regulates EC shear responses

(A) VE-cadherin domains. VE-cadherin contains an extracellular domain (ECD), a transmembrane domain (TMD), and an intracellular domain (ICD) containing catenin binding sites and several tyrosines capable of phosphorylation. The FRET-based tension sensor module was inserted at the indicated sites in the active and control sensors. (B) BAECs expressing either VE-cadherin wild-type (WT) or Y658F tension sensors were exposed to 15dynes/cm2 shear stress for 2 min, at which time the change in FRET is maximal. Cells were fixed and junction FRET signals were measured as described in STAR Methods. Values are means ±SEM. n>20 junctions per condition. Similar results were obtained in an additional three experiments. * p ≤ 0.05. (C) VE-cadherin^-/- cells stably expressing either WT or Y658F VE-cadherin were exposed to laminar shear stress at 15 dynes/cm2 for 24h. Cells were fixed, F-actin stained, and imaged. Scale bar = 20μm. (D) Quantification of (C). Percent of cells orientated within 23° relative to the axis of shear was determined. Values are means ± SD (n=3 experiments). ** p < 0.02, *** p < 0.001, **** p< 0.0001. (E) VE-cadherin^-/- cells re-expressing WT or Y658F VE-cadherin plus a NF-κB response element-driven reporter were subjected to oscillatory shear stress (OSS). NF-κB activity was assayed by fluorimetry as described in STAR Methods. Three experiments gave similar results. See also Figure S1.

Figure 2. VE-cadherin Y658 phosphorylation functions in vivo

(A) Endothelial cells in vitro were exposed to no flow, laminar flow at 15 dynes/cm² (LSS) or oscillatory flow at 1±3 dynes/cm² (OSS) for 16h. VE-cadherin Y658 phosphorylation was then assayed by Western blotting as in STAR Methods. (B) quantification of (A). (C) Aortas from adult WT mice were sectioned longitudinally and stained with anti-VEcad^pY658 and fluorescent Isolectin B4 (IB4) to label the endothelium. Scale bar = 200μm. (D) Quantification of (C). The ratio of endothelial inner curvature/outer curvature signal for each image was quantified (n=5 aortas). * p ≤ 0.05. (E) Aortas from 6-8wk WT or VE-cadherin^Y658F knock in mice were sectioned and immunostained as indicated. (F) Fluorescent signals from Inner curvature regions were quantified in 3-4 different aortas for each antibody. Scale bar = 100μm. * p ≤ 0.05. (G) WT or Y658F mice were subject to femoral artery ligation and blood flow in the hindlimb assayed by Doppler imaging on subsequent days. (H) Quantification of laser Doppler imaging results. * p ≤ 0.05, ** p ≤ 0.01. (I) MicroCT of vasculature in treated leg of WT and Y658F mice. (J) Quantification of vascular microCT from 4-5 mice. * p ≤ 0.001, **** p ≤ 0.0001 by ANOVA.

Fig. 3. VE-cadherin-p120ctn binding suppresses shear responses

(A) BAECs expressing control or p120ctn shRNA and VE-cadherin tension sensors (WT, Y658F, or the ΔCT 0 tension/maximum FRET control) were subjected to 0 or 2min shear, and FRET measured. Values are means ±SEM. n>20 junctions per condition. Similar results were obtained in three independent experiments. * p ≤ 0.05. (B) VE-cadherin^-/- cells expressing either WT or Y658F VE-cadherin plus either control or p120ctn shRNA were subjected to 0 or 15 min of shear; integrin activation was then measured by binding GST-FN9-11 and assaying by Western blotting, as described in STAR Methods. (C) Quantification of results in (B). Values are means ± SEM from 4 independent experiments. (D) VE-cadherin^-/- cells expressing VE-cadherin^Y658F and control or p120ctn shRNA were exposed to 15dynes/cm2 of laminar shear stress for 24h, fixed and stained for F-actin. Scale bar = 20μm. (E) Images from (D) were quantified for cell alignment. Values are means ± SEM, n=4 For percent of cells orientated within 23° relative to the axis of shear. ** p ≤ 0.01, *** p ≤ 0.001. See also Figures S2.

Figure 4. LGN competes with p120ctn for VE-cadherin binding and is important for shear mechanotransduction

(A) HUVECs treated with either scrambled (scr), p120ctn, or LGN siRNAs were lysed and immunoprecipitated with either anti-LGN or control IgG. Immunoprecipitates were Western blotted for LGN and VE-cadherin. Three independent experiments gave similar results. (B) Cells were transfected with either scrambled control, anti-p120ctn, or anti-LGN siRNAs and additionally infected with either VE-cadherin WT or Y658F lentiviral constructs. LGN was immunoprecipitated and bound VE-cadherin was immunoblotted as in (A). Three independent experiments gave similar results. (C) Human VE-cadherin intracellular domain was fused to a C-terminal Flag, an N-terminal, coiled-coil dimerization domain and a His₆ tag. (D) Recombinant VE-ICD-Flag protein on anti-Flag beads was incubated with lysates from cells over-expressing Myc-tagged LGN in the presence of variable amounts of cell lysate from control or HA-tagged p120ctn-expressing cells. Control lysate (-); LGN:p120ctn 1:5 (+); LGN:p120ctn 1:1 (++). Protein bound to VE-ICD-Flag resin was immunoblotted for HA, Myc, and Flag. The graph shows quantified results from 4 independent experiments. (E) HUVECs were lysed and immunoprecipitated for VE-cadherin, p120ctn, or LGN, then immunoblotted for VE-cadherin and anti-VE-cad^pY658. Three independent experiments gave similar results. (F) HUVECs expressing the VE-cad^WT tension sensor were treated with CRISPR cas9/LGN sgRNA to mutate LGN and then transduced with control virus or sgRNA-resistant Flag-tagged LGN. Cells were treated with or without shear stress for 2 min, and FRET measured. Values are means ± SEM, n>20 per condition. Similar results were obtained in two additional experiments. (G) HUVECs expressing the NF-κB reporter were transfected with scrambled control or anti-LGN siRNAs. Cells were treated with or without oscillatory shear stress for 6h, and lysates immunoblotted for the GFP reporter. Values are means ± SEM, n=3 independent experiments. (H) HUVECs treated with siRNA as in (G) were subjected to 12 dynes/cm² laminar shear stress for 12-16h, then fixed and stained for F-actin. (I) Quantified cell alignment from 3 independent experiments. Values are means ± SEM, n=3 Scale bar = 20μm. * p ≤ 0.05. See also Figures S3.