GIT1 regulates angiogenic factor secretion in bone marrow mesenchymal stem cells via NF-κB/Notch signalling to promote angiogenesis

Abstract

Objectives: Osteogenesis is coupled with angiogenesis during bone remodelling. G-protein-coupled receptor (GPCR) kinase 2-interacting protein-1 (GIT1) is an important protein that participates in fracture healing by regulating angiogenesis. This study investigated whether GIT1 could affect bone mesenchymal stem cells (BMSCs) to secrete angiogenic factors to enhance fracture healing by promoting angiogenesis and its possible mechanism.

Materials and methods: The angiogenesis of mice post-fracture was detected by micro-CT and immunofluorescence. Subsequently, vascular endothelial growth factor (VEGF) level in mouse and human BMSCs (hBMSCs) under TNF-α stimulation was detected. The hBMSCs were transfected with GIT1 shRNAs to further explore the relationship between GIT1 and VEGF and angiogenesis in vitro. Furthermore, based on previous research on GIT1, possible signal pathways were investigated.

Results: GIT1 knockout mice exhibited impaired angiogenesis and delayed fracture healing. And GIT1 deficiency remarkably reduced the expression of VEGF mRNA in BMSCs, which affected the proliferation and migration of human umbilical vein endothelial cells. GIT1 knockdown inhibited the activation of Notch and NF-κB signals by decreasing nuclear transportation of NICD and P65/P50, respectively. Overexpression of the canonical NF-κB subunits P65 and P50 markedly increased NICD-dependent activation of recombination signal-binding protein-jκ reporter. Finally, GIT1 enhanced the affinity of NF-κB essential modulator (NEMO) for K63-linked ubiquitin chains via interaction with NEMO coiled-coil 2 domains.

Conclusion: These data revealed a positive role for GIT1 by modulating the Notch/NF-κB signals which promoting paracrine of BMSCs to enhance angiogenesis and fracture healing.

Keywords: GIT1; NF-κB; Notch; angiogenesis; fracture healing.

Figure 1

GIT1 deficiency inhibits angiogenesis in fracture callus tissue. A, Representative immunostaining images for CD31 (in green) and EMCN (in red) in the callus tissues of GIT1 WT and KO mice 14 and 21 days post‐fracture. Scale bar, 100 μm. B, Quantification of blood vessel number and lumen area in CD31⁺/ EMCN⁺ blood vessels (A) in the callus tissues. Data are represented as means ± SEM (n = 3 for both WT and KO mice). *P < .05. C, To visualize and quantify callus vascularity, WT and GIT1 KO mice were perfused with 4% PFA followed by MICROFIL® injection compounds. Representative images for vascular micro‐CT reconstructions of harvested femora 14 and 21 days post‐fracture. D, Quantification of callus vascular parameters, including vessel volume and vessel number. Data are represented as means ± SEM (n = 3 for both WT KO mice). *P < .05

Figure 2

GIT1 deficiency inhibits expression and secretion of VEGF in BMSCs. (A) VEGF mRNA detected by qPCR in mBMSCs by direct adherent 24 hours culture from bone marrow adjacent to the fracture site in GIT1 WT and KO mice between 0 and 7 days post‐surgery. Data are represented as means ± SEM (n = 3 for both WT and KO mice). *P < .05. (B) hBMSCs were transfected with sh‐GIT1 or sh‐Scr for 48 hours and then subjected to control sham or TNF‐α (10 ng/mL) for 24 and 48 hours. VEGF concentration was detected by ELISA in hBMSC‐CM (n = 6). *P < .05. (C) Proliferation of HUVECs cultured with hBMSCs‐CM examined by CCK8 12 hours, 1, and 2 d after TNF‐α stimulation. Data are represented as means ± SEM (n = 8). *P < .05. (D, F, H) Representative images of Matrigel tube formation, transwell and scratch wound assays with hBMSCs‐CM cultures after TNF‐α stimulation for 48 hours. Scale bar, 100 μm. (E, G, I) Quantitative analysis of tube length and branch points during tube formation (D), the number of migrated cell (F) and migration rate of HUVECs (I). Data are represented as means ± SEM (n = 6). *P < .05

Figure 3

GIT1 KO inhibits activation of the Notch signal. A, Representative Western blots for Hey1, Hes1 and VEGF expression in TNF‐α‐treated (24 and 48 hours) hBMSCs in the presence or absence of Notch inhibitor DAPT. B, Quantification of Hey1, Hes1 and VEGF in hBMSCs based on Western blots described in (A). GAPDH was used as loading control. Data are represented as means ± SEM (n = 3). *P < .05. C, qPCR expression analysis for Hey1, Hes1, and VEGF mRNA in hBMSCs treated as described in (A). Data are represented as means ± SEM (n = 3). *P < .05. D, Representative Western blots for GIT1, Hey1, Hes1 and VEGF expression in hBMSCs transfected with sh‐GIT1 or sh‐Scr for 48 hours, subjected to control sham operation or TNF‐α stimulation for 24 and 48 hours. E, Quantification of GIT1, Hey1, Hes1, and VEGF in hBMSCs based on Western blots described in (D). GAPDH was used as loading control. Data are represented as means ± SEM (n = 3). *P < .05. F, qPCR expression analysis for Hey1, Hes1 and VEGF mRNA in hBMSCs treated as described in (D). Data are represented as means ± SEM (n = 3). *P < .05

Figure 4

GIT1 KO inhibits activation of canonical NF‐κB signal. (A) Representative Western blots for total protein (GIT1, IKKα, IKKβ and P65) and phosphorylated protein (IKKα, IKKβ and P65) levels in GIT1 knockdown and control hBMSCs, exposed to TNF‐α stimulation for the indicated times. (B) Quantitative comparison of total protein and signalling activation levels between GIT1 knockdown and control hBMSCs using density scanning of the blots described in (A). GAPDH was used as loading control. Data are expressed as means ± SEM (n = 3). *P < .05 vs. hBMSCs infected with sh‐Scr. (C) Representative Western blots for cytoplasmic and nuclear protein (NICD, RELB, P65, P52 and P50) levels in GIT1 knockdown and control hBMSCs exposed to TNF‐α stimulation for 2 hours. (D) Quantification of cytoplasmic and nuclear protein (NICD, RELB, P65, P52, and P50) levels in hBMSCs based on Western blots described in (C). GAPDH and H3 were used as loading controls for cytoplasmic and nuclear proteins, respectively. Data are represented as means ± SEM (n = 3). (E) Representative immunostaining images for hBMSCs infected with sh‐GIT1 or sh‐Scr, treated with control sham operation or TNF‐α for 2 hours and co‐labelled for NICD, RELB, P65 (in green) and DAPI for subcellular co‐localization examination. Scale bar, 100 μm. (F) Reporter activity in HEK 293T cells co‐transfected with RBP‐jκ‐Luc and/or NICD‐, RELB‐, P65‐, P52‐ and P50‐expressing vectors. After 48‐hours transfection, luciferase activity was measured and fold increase vs. empty vector was calculated. Data are represented as means ± SEM (n = 5). *P < .05

Figure 5

SLD structure containing GIT1 CC2 domain plays a critical role in interaction with NEMO CC2 domain. (A) Association of endogenous GIT1 and NEMO in hBMSCs without treatment or after TNF‐α stimulation for 2 hours, was examined by IP with GIT1 antibody, immunoblotting for NEMO or IP with NEMO antibody, and immunoblotting for GIT1. (B) Fixed and permeabilized hBMSCs from (A) were first incubated with rabbit NEMO and mouse GIT1 antibodies. Subsequently, cells were incubated with anti‐mouse MiNUs PLA probes followed by ligation and amplification. Interacting proteins were visualized using red fluorophore‐labelled oligonucleotides. Fixed and permeabilized cells were incubated with mouse IgG and rabbit IgG for use as controls. Cells were then incubated with anti‐mouse MiNUs PLA probes and visualized using red fluorophore‐labelled oligonucleotides, as above. (C, D) Functional domains of GIT1 and NEMO. (E) HEK293T cells were co‐transfected with Flag‐GIT1‐WT or Flag‐GIT1 deletion mutants and HA‐NEMO for 48 hours and subsequently treated with TNF‐α for 15 minutes. IP was performed with Flag antibody and probed for HA to detect interaction of NEMO and GIT1 or GIT1 deletion mutants. Cell lysates were also examined directly by immunoblot analysis with HA or Flag antibodies. (F) HEK293T cells were co‐transfected with HA‐NEMO‐WT or HA‐NEMO (ΔCC2) and Flag‐GTI1‐WT. Interaction of GIT1 and NEMO or NEMO (ΔCC2) was examined by IP with HA antibody and immunoblotting for Flag antibody. Cell lysates were examined directly by immunoblot analysis with antibodies. ANK, ankyrin‐rich repeat domain; ARF‐GAP, amino‐terminal ADP‐ribosylation factor‐GTPase‐activating protein domain; LZ, leucine zipper domain; TBD, thioredoxin binding region; ZF, zinc finger

Figure 6

GIT1 enhances affinity of NEMO for K63‐linked ubiquitin chains via CC2 of GIT1. A, HEK293T cells were transfected with Fag‐GIT1‐WT, HA‐NEMO‐WT, Myc‐ubiquitin‐WT, and Myc‐K6‐, K11‐, K27‐, K29‐, K33‐, K48‐ or K63‐linked‐Ub for 48 hours and subsequently treated with TNF‐α for 2 hours following MG132 (10 μM) treatment for 1 hour. HA‐immunoprecipitation was performed and analysed using anti‐Myc antibody by immunoblot analysis. B, HEK293T cells were transfected with HA‐NEMO‐WT, Myc‐ and K63‐linked‐Ub, and Flag‐GIT1, or Flag‐GIT1(ΔCC2) for 48 hours. Harvested protein was treated with TNF‐α for 2 hours following MG132 (10 μM) treatment for 1 hour, immunoprecipitated with anti‐HA and immunoblotted with anti‐Myc

Figure 7

GIT1 does not affect K63‐linked RIP1 ubiquitination induced by TRAF2. A, Association of endogenous TRAF2 and NEMO in GIT1 knockdown and control hBMSCs without treatment or after TNF‐α stimulation for 2 hours was examined by IP with NEMO antibody, immunoblotting for TRAF2 and RIP1 or IP with TRAF2 antibody, and immunoblotting for NEMO and RIP1. B, C, Fixed and permeabilized hBMSCs treated as described in (A) were first incubated with rabbit NEMO antibody and mouse TRAF2 antibody (B) or mouse TRAF2 antibody and rabbit RIP1 antibody (C). Subsequently, cells were incubated with anti‐mouse MiNUs PLA probes followed by ligation and amplification. D, HEK293T cells were transfected with His‐RIP1‐WT, Myc‐ and K63‐linked‐Ub and sh‐GIT1, or sh‐Scr for 48 hours. Harvested protein was treated with TNF‐α for 2 hours following MG132 (10 μM) treatment for 1 hour, immunoprecipitated with anti‐His, and immunoblotted with anti‐Myc

Figure 8

Working model of positive NF‐κB/Notch signal regulation by GIT1. GIT1 interacts with NEMO via their mutual CC2 structures and enhances NEMO affinity for K63‐linked ubiquitination of RIP1 induced by TRAF2 (Figure 8). It leads to phosphorylation of the IKKα/β/NEMO complex and facilitates activation of the canonical NF‐κB signal, but not the non‐canonical signal. It regulates the Notch signalling by increasing translocation of NICD into the nucleus and subsequently leading to interaction of NICD and RBP‐jк, requiring cleavage from the Notch molecules to be induced by γ‐secretase and blocked by DAPT. Therefore, endogenous GIT1 enhances nuclear import of the canonical NF‐κB subunits P65 and P50 and NICD, leading expression of the downstream target genes of the NF‐κB/Notch signal, such as VEGF and other angiogenic factors in BMSCs, during the early stages of fracture healing