Endothelial β-arrestins regulate mechanotransduction by the type II bone morphogenetic protein receptor in primary cilia

Abstract

Modulation of endothelial cell behavior and phenotype by hemodynamic forces involves many signaling components, including cell surface receptors, intracellular signaling intermediaries, transcription factors, and epigenetic elements. Many of the signaling mechanisms that underlie mechanotransduction by endothelial cells are inadequately defined. Here we sought to better understand how β-arrestins, intracellular proteins that regulate agonist-mediated desensitization and integration of signaling by transmembrane receptors, may be involved in the endothelial cell response to shear stress. We performed both in vitro studies with primary endothelial cells subjected to β-arrestin knockdown, and in vivo studies using mice with endothelial specific deletion of β-arrestin 1 and β-arrestin 2. We found that β-arrestins are localized to primary cilia in endothelial cells, which are present in subpopulations of endothelial cells in relatively low shear states. Recruitment of β-arrestins to cilia involved its interaction with IFT81, a component of the flagellar transport protein complex in the cilia. β-arrestin knockdown led to marked reduction in shear stress response, including induction of NOS3 expression. Within the cilia, β-arrestins were found to associate with the type II bone morphogenetic protein receptor (BMPR-II), whose disruption similarly led to an impaired endothelial shear response. β-arrestins also regulated Smad transcription factor phosphorylation by BMPR-II. Mice with endothelial specific deletion of β-arrestin 1 and β-arrestin 2 were found to have impaired retinal angiogenesis. In conclusion, we have identified a novel role for endothelial β-arrestins as key transducers of ciliary mechanotransduction that play a central role in shear signaling by BMPR-II and contribute to vascular development.

Keywords: BMPR2; beta‐arrestin; endothelial cells; primary cilia; shear stress.

Figure 1

β‐arrestin1 interacts with ciliary proteins under shear stress and is required for the formation of primary cilia in HUVECs. (a) Immunofluorescence images showing β‐arrestin1 (green), primary ciliary marker (red; Ac‐⍺‐tubulin), and DAPI (blue) in HUVECs. (b, c) Immunofluorescence staining and quantification of primary cilia (red; Ac‐⍺‐tubulin) in β‐arrestin1/2 siRNA (β‐arrs1 + 2 si) or control siRNA (Con si) treated HUVECs. Yellow arrows indicate primary cilia, and the number of primary cilia was quantified as percentage based on the number of cells (blue; DAPI). N indicates number of cells counted (400) and ****p < 0.0001. (d) Immunofluorescence images showing β‐arrestin1 (green), IFT81 (cyan), primary ciliary marker (red; Ac‐⍺‐tubulin), and DAPI (blue) in HUVECs. (e) Immunoprecipitation of GFP‐conjugated β‐arrestin1 and Myc‐conjugated IFT81 in HEK293 cells. Proteins were pulled down with GFP antibody and detected by Myc and GFP antibodies. Control GFP construct (con‐GFP) was used as immunoprecipitation control and GAPDH antibody was used as a loading control. (f, g) Immunofluorescence staining and quantification of β‐arrestin1 (green) in IFT81 siRNA (IFT81 si) or control siRNA (Con si) treated HUVECs. The number of β‐arrestin1 localized primary cilia was quantified as percentage based on the number of primary cilia (red; Ac‐⍺‐tubulin). n indicates number of primary cilia counted and ***p < 0.001. HUVECs, human umbilical vein endothelial cells.

Figure 2

BMPR2 and β‐arrestin1 are required for the shear stress response. (a) Confocal microscopy demonstrating BMP‐2‐induced internalization of RFP‐labeled β‐arrestin2 before (pre) and 30 min after stimulation (post) with 50 ng/ml of BMP‐2 in HEK293 cells transiently transfected with (top panel) BMPR1A/BMPR2 and (bottom panel) BMPR1B/BMPR2. (b) Immunoblot detection of phosphorylated eNOS (p‐eNOS), total eNOS (t‐eNOS), β‐arrestin1/2, BMPR‐II and GAPDH levels extracted from BMPR2 siRNA (BMPR2 si), or control siRNA (Con si) treated HUVECs with or without shear stress. (c) Cell alignment of BMPR2 siRNA (BMPR2 si) and control siRNA (Con si) treated HUVECs with or without shear stress. (d, e) Immunofluorescence staining and quantification of BMPR‐II (green) in HUVECs with or without shear stress. The number of BMPR‐II localized primary cilia was quantified as percentage based on the number of primary cilia (red; Ac‐⍺‐tubulin). n indicates number of primary cilia counted and ****p < 0.0001. (f, g) Immunofluorescence staining and quantification of β‐arrestin1 (green) in BMPR2 siRNA (si BMPR2) and control siRNA (si Con) treated HUVECs. The number of β‐arrestin1 localized primary cilia was quantified as percentage based on the number of primary cilia (red; Ac‐⍺‐tubulin). n indicates number of primary cilia counted and ***p < 0.001. BMPR, bone morphogenetic protein receptor; HUVECs, human umbilical vein endothelial cells; REP, red fluorescent protein.

Figure 3

β‐arrestins interact with and regulate signaling by BMPR‐II. (a) Interaction of β‐arrestins with BMPR‐II as assessed by coimmunoprecipitation of overexpressed FLAG‐tagged β‐arrestin1/2 with myc‐tagged full length BMPR2 and its 532X truncation. (b, c) β‐arrestin1 significantly reduces BMP‐induced Smad1/5/8 phosphorylation by BMPR‐II after stimulation with BMP‐2 (50 ng/ml) with siRNA knockdown of β‐arrestin1 or β‐arrestin2. pSmad/tSmad levels were normalized to ctl siRNA signal at 30 min (*p < 0.05 compared to other siRNA treated samples for that time point by two‐way ANOVA with Tukey's multiple comparison test). (d) Both β‐arrestin1 and 2 knockdown significantly decreases activity of a BMP reporter (Bre‐luc) transfected in HEK293 cells stimulated with BMP‐2 (10 ng/ml) (*p < 0.05 all groups significantly different; **p < 0.05 β‐arrestin1‐ and ctl‐siRNA significantly different by two‐way ANOVA with Tukey's multiple comparison test). (e) No difference in time course of BMP‐2‐induced endocytosis between WT and β‐arrestin1/2 KO HEK293 cells. The endocytosed pool of BMPR‐II was labeled with biotin (see Section 2) before pulldown with avidin beads. (f) No effect of β‐arrestin1/2 overexpression on receptor internalization in β‐arrestin1/2 KO cells. Endocytosis at 30 min was significantly decreased by the clathrin inhibitor Pitstop2 but not by the dynamin inhibitor dynasore. All experiments performed at least three times; shown are mean ± SEM. Blots and images shown are representative from at least three independent experiments. BMPR, bone morphogenetic protein receptor.

Figure 4

β‐arrestin1 is required for the endothelial cell shear stress response and eNOS expression. (a) Cell alignment of β‐arrestin1/2 siRNA or control siRNA treated HUVECs with or without shear stress. (b) Immunoblot detection of phosphorylated eNOS (p‐eNOS), total eNOS (t‐eNOS), β‐arrestin1/2 and GAPDH levels extracted from β‐arrestin1/2 siRNA, or control siRNA treated HUVECs with or without shear stress. (c) Immunofluorescence images showing eNOS (green) and DAPI (blue) in β‐arrestin1/2 siRNA or control siRNA treated HUVECs with or without shear stress. (d) Immunoprecipitation of ubiquitin‐eNOS in β‐arrestin1/2 siRNA treated HUVECs. Proteins were pulled down with eNOS antibody and detected by ubiquitin‐eNOS and total eNOS antibodies. Input lysates were detected by β‐arrestin1 and GAPDH antibodies. MG132 is a proteosomal inhibitor. (e) Immunoblot detection of phosphorylated eNOS (p‐eNOS), total eNOS (t‐eNOS), IFT81, IFT74 and GAPDH levels extracted from IFT81 and/or IFT74 siRNA, or control siRNA treated HUVECs with or without shear stress. HUVECs, human umbilical vein endothelial cells.

Figure 5

Endothelial‐specific knockout of β‐arrestins results in a retinal vascular development defect. (a) IB4 stained p7 retinal flat‐mounts of Arrb1/Arrb2 ECKO and wild‐type (control) mice. (b) Quantification of vascular radial expansion in Arrb1/Arrb2 ECKO and wild‐type (Ctl) mice. N = 7 wild‐type and 4 Arrb1/2 ECKO. *p < 0.05. (c) The number of branchpoints in Arrb1/Arrb2 ECKO and wild‐type (Ctl) mice. N = 7 wild‐type and 3 Arrb1/2 ECKO. *p < 0.05. (d) eNOS (green) and IB4 (red) stained p7 retinal flat‐mounts of Arrb1/2 ECKO and wild‐type mice. A indicates arteries and V indicates veins.