SCF/c-Kit-activated signaling and angiogenesis require Gαi1 and Gαi3

The stem cell factor (SCF) binds to c-Kit in endothelial cells, thus activating downstream signaling and angiogenesis. Herein, we examined the role of G protein subunit alpha inhibitory (Gαi) proteins in this process. In MEFs and HUVECs, Gαi1/3 was associated with SCF-activated c-Kit, promoting c-Kit endocytosis, and binding of key adaptor proteins, subsequently transducing downstream signaling. SCF-induced Akt-mTOR and Erk activation was robustly attenuated by Gαi1/3 silencing or knockout (KO), or due to dominant negative mutations but was strengthened substantially following ectopic overexpression of Gαi1/3. SCF-induced HUVEC proliferation, migration, and capillary tube formation were suppressed after Gαi1/3 silencing or KO, or due to dominant negative mutations. In vivo, endothelial knockdown of Gαi1/3 by intravitreous injection of endothelial-specific shRNA adeno-associated virus (AAV) potently reduced SCF-induced signaling and retinal angiogenesis in mice. Moreover, mRNA and protein expressions of SCF increased significantly in the retinal tissues of streptozotocin-induced diabetic retinopathy (DR) mice. SCF silencing, through intravitreous injection of SCF shRNA AAV, inhibited pathological retinal angiogenesis and degeneration of retinal ganglion cells in DR mice. Finally, the expression of SCF and c-Kit increased in proliferative retinal tissues of human patients with proliferative DR. Taken together, Gαi1/3 mediate SCF/c-Kit-activated signaling and angiogenesis.

SCF is important for angiogenesis. Matsui et al. reported that SCF treatment in endothelial cells could activate pro-angiogenic reactions and enhance mobility and the formation of capillary tubes in endothelial cells [18]. Fang et al. reported that c-Kit deficiency hindered vascular endothelial stem cell proliferation and blocked angiogenesis in vivo [19]. Wang et al. reported that the activation of c-Kit by SCF could promote survival and suppress apoptosis in vascular smooth muscle cells [20]. SCF activates c-Kit signaling and is important for the formation of hematopoietic stem cells [17]. Herein, we examined the role of G protein subunit alpha inhibitory (Gαi) proteins in SCF-activated signaling and angiogenesis.

SCF-induced membrane c-Kit internalization in MEFs requires Gαi1 and Gαi3
Following SCF stimulation, c-Kit associates with several key adaptor proteins, including Grb2, Gab2, and Shc, and results in subsequent phosphorylation of Gab2 and Shc to promote downstream signaling [11]. We discovered that Gαi1/3 could associate with ligand-activated receptors (IL-4Rα, TrkB, VEGFR2, etc.), thus promoting receptor internalization and endocytosis and transducing downstream signals [27,28,33,35]. Herein we showed that SCF-activated c-Kit underwent membrane internalization ( Figure S2A). Cell membrane-localized c-Kit protein levels decreased remarkably in WT MEFs after SCF treatment ( Figure S2A). The membrane c-Kit internalization was fast and started within 1 min of SCF treatment (Figure S2A), and by 5 min, the majority of membrane c-Kit protein was internalized ( Figure S2A). Gαi1 and Gαi3 were required for SCF-induced c-Kit internalization, as membrane c-Kit internalization was prevented by Gαi1/3 DKO in MEFs ( Figure S2B). Total c-Kit protein levels were unchanged following SCF treatment in MEFs ( Figure  S2B).

Gαi1 and Gαi3 silencing prevents SCF-induced signaling and pro-angiogenic activity in endothelial cells
The roles of Gαi1 and Gαi3 in SCF-induced signaling in endothelial cells were studied. Co-IP assays were conducted and results showed that SCF-activated c-Kit immunoprecipitated with Grb2, Gab2, and Shc, as well as Gαi1 and Gαi3 in HUVECs ( Figure 3A). Expressions of c-Kit, Grb2, Gab2, Shc, Gαi1 and Gαi3 remained unchanged following SCF treatment ( Figure 3A, "Input"). Moreover, both Gαi1 and Gαi3 proteins formed a complex with c-Kit, Grb2, Gab2, and Shc in SCF-treated HUVECs ( Figure 3B). In HUVEC, SCF-induced cell proliferation ( Figure S3A), migration ( Figure S3B), and tube formation ( Figure  S3C), as well as the mRNA expression of VEGF ( Figure S3D) and PDGF-BB ( Figure S3E) were inhibited by the Erk1/2 inhibitor PD98059 or the PI3K-Akt-mTOR inhibitor LY294002. Importantly, PD98059 plus LY294002 ("PD+LY") completely blocked SCF-induced pro-angiogenic actions along with the mRNA expression of VEGF and PDGF-BB in HUVECs (Figures S3A-E). Thus PI3K-Akt-mTOR and Erk are two essential cascades required for SCFinduced pro-angiogenic actions in HUVECs.

Gαi1 and Gαi3 overexpression strengthens SCF-induced signaling and pro-angiogenic activity in endothelial cells
Since Gαi1/3 silencing, KO, or mutation largely inhibited SCF-induced signaling and pro-angiogenic activity in HUVECs, we next hypothesized that overexpressing Gαi1 and Gαi3 could exert opposite functions and augment pro-angiogenic activity in endothelial cells. Thus, the lentiviral particles with the Gαi1 (human)-expressing vector together with the lentiviral particles with the Gαi3 (human)-expressing vector were co-transfected into HUVECs, and puromycin was added to select two stable cell colonies, namely "oeGαi1/3-Slc1" and "oeGαi1/ 3-Slc2". The expressions of Gαi1 and Gαi3 increased robustly in oeGαi1/3 HUVECs, while that of Gαi2 remained unchanged (Figures 5A and B) compared to HUVECs with vector control ("Vec"). SCF-induced phosphorylation of Akt, S6K, and Erk1/2 was significantly augmented in oeGαi1/3-Slc1/2 HUVECs (Figure 5C). Overexpressing Gαi1 and Gαi3 promoted HUVEC proliferation and increased EdU-positive nuclei ratio ( Figure 5D). Moreover, in vitro migration ( Figure 5E) and capillary tube formation ( Figure 5F) were strengthened in oeGαi1/3 HUVECs. Figure 3. Gαi1 and Gαi3 silencing prevents SCF-induced signaling and pro-angiogenic activity in endothelial cells. HUVECs were treated with SCF (50 ng/mL) for 5 min, and the association of c-Kit, Grb2, Gab2, Shc, Gαi1, and Gαi3 was examined by co-immunoprecipitation (Co-IP) assays (A and B). Their expressions are shown as "Input" (A and B).

SCF shRNA inhibits pathological retinal angiogenesis in mice with diabetic retinopathy (DR)
We checked for alteration in the expression of SCF in streptozotocin (STZ)-administrated DR mice's retinal tissues. After 90 days of the last STZ administration, the retinal tissues of both DR and "Mock" control (citrate buffer-administrated) mice were collected. The mRNA expression of SCF in the retinal tissues of DR mice was significantly elevated ( Figure 7A). Moreover, protein upregulation of SCF was observed in the retinal tissues of a set of four representative STZ-administrated DR mice ( Figure  7B). After combining all 10 sets of blotting data, we found that the protein levels of SCF were significantly elevated in the retinal tissues of DR mice (Figure 7C).
AAV5-SCF shRNA ("shSCF-AAV5") or AAV5scramble shRNA control ("shC-AAV5") were injected intravitreously into the retina of DR mice on day-30 after the last STZ administration to examine whether increased SCF expression played a role in pathological retinal angiogenesis in DR mice. After another 60 days, the fresh retinal tissues were collected and examined. As shown, shSCF-AAV5 downregulated mRNA and protein expressions of SCF in shC-AAV5 DR mice's retinal tissues ( Figures   7D and E). Akt-Erk1/2 phosphorylation increased in shC-AAV5 DR mice's retinal tissues ( Figure 7F). Remarkably, SCF silencing by shSCF-AAV5 reduced Akt and Erk activation in DR mice's retinal tissues ( Figure 7F).

Figure 5.
Gαi1 and Gαi3 overexpression strengthens SCF-induced signaling and pro-angiogenic activity in endothelial cells. HUVECs were transduced with the lentiviral human Gαi1-expressing construct plus the lentiviral human Gαi3-expressing vector, and two stable colonies, "oeGαi1/3-Slc1" and "oeGαi1/3-Slc2", were obtained after selection. Control HUVECs were transduced with vector control ("Vec"). HUVECs were then treated with SCF (50 ng/mL) for 15 min and listed mRNA and protein levels were examined (A-C). HUVECs were further cultured, and cell proliferation (D), in vitro migration (E), and capillary tube formation (F) were tested. * P< 0.05 versus "Vec". "N. S." denotes P > 0.05. The retinal vascular leakage, tested by Evans blue (EB) quantification, increased significantly in shC-AAV5 DR mice compared to the mock control mice (Figure 7G). IB4 staining assay results revealed enhanced retinal vasculature complexity with increased vascular branches and branch points in the retina of shC-AAV DR mice, further supporting retinal pathological angiogenesis (Figure 7H). Retinal trypsin digestion assay showed an increase number of retinal acellular capillaries in shC-AAV DR mice ( Figure 7I).
Specifically, in the DR mice retinal vascular leakage (Figure 7G), pathological angiogenesis (Figure 7H), and acellular capillary formation ( Figure 7I) were largely suppressed by SCF silencing through shSCF-AAV5. Thus, SCF silencing ameliorated pathological retinal angiogenesis in DR mice. mRNA expressions of VEGF (Figure 7J), PDGF-BB ( Figure  7K), and Ang-1 ( Figure 7L) increased substantially in the retinal tissues of DR mice, which was suppressed by shSCF-AAV5 injection (Figures 7J-L). Figure 7. SCF shRNA inhibits pathological retinal angiogenesis in diabetic retinopathy (DR) mice. The retinal tissues of DR mice (90 days after the last STZ administration) and "mock" control mice (with citrate buffer administration) were separated, expressions of SCF mRNA and protein were tested, and the results were quantified (A-C). Day-30 after STZ administration, mice were injected intravitreously with AAV5-packed SCF shRNA ("shSCF-AAV5", at 0.1 μL) or AAV5-packed scramble shRNA control ("shC-AAV5", at 0.1 μL). After another 60 days, listed mRNAs and proteins in the retinal tissues were assessed (D-F, J-L). Alternatively, mice were infused with Evans blue (EB) for 2 h, and the percentage of EB leakage was quantified (G). IB4 staining was carried out to visualize the retinal vasculature (H, scale bar = 50 μm), and the average number of vascular branches were quantified (H). The retinal trypsin digestion assay was performed and the number of acellular capillaries per view were recorded (I). "Mock" refers to mice administered with citrate buffer. * P< 0.05. The listed human tissues were homogenized and mRNA and protein expressions of SCF and c-Kit were examined (C-E). "Mock" refers to mice administered with citrate buffer. "GCL" is the ganglion cell layer; "ONL" is the outer nuclear layer; "INL" is the inner nuclear layer. * P< 0.05 (A and B). * P< 0.05 vs. "Ctrl" tissues (C-E).

SCF shRNA ameliorates degeneration of retinal ganglion cells (RGCs) in DR mice
In the pathogenesis of DR, pathological angiogenesis, energy crisis, oxidative injury, and inflammatory reaction, all lead to the degeneration of RGCs and are important mechanisms causing blindness [36,37]. The number of NeuN-stained RGCs in GCL (ganglion cell layer) decreased substantially in the retina of shC-AAV DR mice compared to the mock control mice (Figures 8A and B). Importantly, SCF shRNA by intravitreous injection of AAV5-SCF shRNA largely inhibited RGC degeneration in DR mice (Figures 8A and B).

SCF and c-Kit expressions increase significantly in proliferative retinal tissues of human patients with proliferative diabetic retinopathy (PDR)
Lastly, the expressions of SCF and c-Kit in human patients' proliferative retinal tissues were tested. We evaluated the previously-described human tissue samples [23,27]. Retinal proliferative membrane tissues of six different human PDR patients along with the retinal tissues of three age-matched traumatic retinectomy patients were obtained [23,27]. The mRNA (Figures 8Cand D) and protein ( Figure  8E) expressions of SCF and c-Kit increased substantially in human PDR patients' proliferative retinal tissues.

Discussion
Akt-mTOR and Erk cascade activation are vital for SCF/c-Kit-induced HUVEC survival, migration, and capillary tube formation in vitro and angiogenesis in vivo [18,38]. Herein, we discovered that Gαi1/3 are essential proteins mediating SCF-activated signaling and angiogenesis. In MEFs and HUVECs, SCFinduced Akt-mTOR and Erk activation was prevented by Gαi1/3 silencing, KO, or DN mutations but was strengthened following ectopic overexpression of Gαi1/3. SCF-stimulated HUVEC proliferation, migration, and capillary tube formation were substantially suppressed after Gαi1/3 shRNA, KO, or DN mutations but were greatly enhanced following Gαi1/3 overexpression. In vivo, Gαi1 and Gαi3 endothelial knockdown potently reduced SCFinduced Akt-mTOR and Erk activation in retinal tissues and retinal angiogenesis in mice.
Following SCF stimulation, the Grb2-Sos complex recruitment to c-Kit was through the association with tyrosine-phosphorylated Shc, which mediated Src family kinase (SFK) phosphorylation and activated Erk-MAPK signaling downstream [39]. PI3K could be activated by SCF by binding to Gab2 [40,41]. Herein, Gαi1/3 was associated with SCF-activated c-Kit in MEFs and HUVECs, which was essential for binding and activation of key adaptor proteins (Grb2, Gab2, and Shc) and transducing signals downstream. Gαi1/3 DN mutation disrupted SCF-induced binding of adaptor proteins to c-Kit and prevented Akt-mTOR and Erk activation downstream.
Nishida et al., reported that Gab2 was tyrosine phosphorylated in response to SCF stimulation [42]. SCF-activated Akt and MAPK activation was largely impaired in bone marrow-derived mast cells with Gab2 KO [42]. Sun et al., supported a role of Gab2 in mediating PI3K activation by SCF-activated c-Kit [43]. Following SCF stimulation phosphorylated Gab2 associated with c-Kit and Shp-2, required for downstream signaling transduction [43,44]. Here we found that SCF-induced c-Kit-Gab2 association and Gab2 phosphorylation were largely inhibited by Gαi1/3 depletion or DN mutations. Thus Gαi1/3 shall act as upstream proteins mediating SCF-induced Gab2 activation.
With SCF binding, the receptor c-Kit clusters as dimers and internalizes by endocytosis possibly in clathrin-coated pits [45,46]. Like other RTKs, SCF-activated c-Kit internalization is a controlled process assembled by the endocytic machinery including clathrin chains, adaptor proteins, dynamin, and other cytosolic factors [45,46]. This process is important for binding to adaptor proteins, downstream signaling activation, and receptor recycling [45][46][47][48].
Hypoxia upregulates c-Kit in endothelial cells, leading to remarkably enhanced angiogenic responses to SCF [49]. In mouse ocular neovascularization models, expressions of c-Kit and SCF are markedly enhanced in ocular tissues [49]. Conversely, blockade of the SCF/c-Kit cascade in c-Kit mutant mice or using anti-SCF antibody remarkably ameliorates pathological ocular neovascularization [49]. Our results demonstrated that expressions of SCF and c-Kit increased significantly in human PDR patients' proliferative retinal tissues. mRNA and protein expressions of SCF increased dramatically in STZ DR mice retinal tissues. Importantly, SCF silencing, through intravitreous injection of SCF shRNA AAV, inhibited pathological retinal angiogenesis and RGC degeneration in DR mice. In addition to our previous findings showing Gαi1/3 upregulation in PDR patients' proliferative retinal tissues [27], we propose that augmented SCF-c-Kit-Gαi1/3 cascade is vital for the pathological angiogenesis in PDR, representing a promising therapeutic target against PDR and other ocular neovascularization diseases.
In DR, pathological angiogenesis in retinas will lead to severe ischemia and hypoxia environment, causing glutamate toxicity, oxidative injury and inflammation [36,37,50]. These events will eventually lead to degeneration of RGCs and vision loss [36,37,50]. Here SCF mRNA and protein expression was robustly increased in STZ DR mice retinal tissues. SCF silencing, through intravitreous injection of SCF shRNA AAV, potently suppressed pathological retinal angiogenesis and restored RGCs. Thus, in DR retinas, SCF silencing-induced amelioration of pathological angiogenesis should be far more powerful in restoring RGCs than possible decreased SCF-mediated direct neuroprotection [51].

Other assays
Cellular functional assays, including the EdU test for cell proliferation, cell migration, and in vitro capillary tube formation, have been described previously [23,27,52]. Protocols for western blotting, qRT-PCR, and Co-IP assays were described in the previous studies [31,53,54]. We followed a previously described protocol for the isolation of cell plasma membrane [55] with minor modifications [28]. All the primers and viral constructs were synthesized by Genechem (Shanghai, China).

Human tissues
The human tissues used herein have been described previously [23,27] The protocols were approved by the Ethics Committee of Soochow University (#BR-2019-012).

Statistical analysis
All data are presented as mean ± standard deviation (SD). All in vitro cell experiments and in vivo animal experiments were repeated at least five times. The blot data or qRT-PCR data quantifications were based on five replicate experiments unless otherwise stated. Statistical differences were calculated by Student's t-test (comparing two groups) or by one-way analysis of variance (ANOVA) followed by Dunnett's post hoc test. P< 0.05 was considered statistically significant.