RIG-I inhibits pancreatic β cell proliferation through competitive binding of activated Src

Nutrition is a necessary condition for cell proliferation, including pancreatic β cells; however, over-nutrition, and the resulting obesity and glucolipotoxicity, is a risk factor for the development of Type 2 diabetes mellitus (DM), and causes inhibition of pancreatic β-cells proliferation and their loss of compensation for insulin resistance. Here, we showed that Retinoic acid (RA)-inducible gene I (RIG-I) responds to nutrient signals and induces loss of β cell mass through G1 cell cycle arrest. Risk factors for type 2 diabetes (e.g., glucolipotoxicity, TNF-α and LPS) activate Src in pancreatic β cells. Elevated RIG-I modulated the interaction of activated Src and STAT3 by competitive binding to STAT3. Elevated RIG-I downregulated the transcription of SKP2, and increased the stability and abundance of P27 protein in a STAT3-dependent manner, which was associated with inhibition of β cell growth elicited by Src. These results supported a role for RIG-I in β cell mass loss under conditions of metabolic surplus and suggested that RIG-I-induced blocking of Src/STAT3 signalling might be involved in G1 phase cycle arrest through the Skp2/P27 pathway in pancreatic β cells.

Scientific RepoRts | 6:28914 | DOI: 10.1038/srep28914 Cell proliferation assay by 5-ethynyl-2′-deoxyuridine (EdU) labelling. DNA synthesis was analysed using an EdU Labelling Kit. MIN6 cells were cultured in 6-well plates on coverslips. EdU was added to the culture medium (50 μ M) for 2 h after treatment and cell proliferation was determined according to the manufacturer's instructions.
Immunofluorescence assay (IFA). IFA was used to observe visually the changes in localization and levels of phospho-STAT3 or the level of Src. MIN6 cells or isolated mouse islets were subjected to IFA after transfection and pharmaceutical treatment. Fluorescein-labelled antibodies diluted in PBS-BSA [anti-insulin/Cy3 antibody (1:50) and anti-phospho-STAT3/TRITC antibody (1:50) or anti-insulin/Cy3 antibody (1:50) and anti-Src/TRITC antibody (1:50)] were added to MIN6 cells or mouse islets and incubated overnight at 4 °C. Isolated mouse islets in suspension were centrifuged at 2000 × g for 5 min at 4 °C at each step 23,26 . Flow cytometric analysis. A trypsin-EDTA solution was used to digest MIN6 cells. Trypsinised MIN6 cells were collected by centrifugation at 500 × g for 5 min. The MIN6 cell pellets were washed with PBS three times and fixed in cold 75% ethanol at 4 °C overnight. Flow cytometry, preceded by propidium iodide (PI) staining, was used to determine the percentages of cells in the G0/G1, S and G2/M phases.
Luciferase reporter assay. According to the manufacturer's instructions, the luciferase reporter construct pGMSTAT3-Lu was transfected transiently into MIN6 cells cultured in 24 well plates, using the Lipofectamine 2000 reagent. The gene encoding β galactosidase, expressed in a plasmid driven by the cytomegalovirus (CMV) promoter (Clontech Laboratories, Palo Alto, CA, USA), was transfected simultaneously as an internal control. Six hours after transfection, the medium was replaced. The cells were treated 24 h after transfection and harvested for luciferase reporter assays, as described previously 29 . Data analysis. All data were representative of at least three experiments. Results are expressed as the mean ± SEM. Comparisons were performed using Student's t-test for two groups or ANOVA for multiple groups. P values < 0.05 were considered statistically significant.

Src is activated in MIN6 cells.
To explore the association of Src with β -cells mass and T2DM, especially its risk factors such as Glu-palm, LPS and TNF-α , MIN6 cells were subjected to different stimuli, such as treatment with 0.4 mM palmitate plus 16.7 mM glucose for 24 h, or 80 nM TNF-α for 6 h, or 10 μ g/mL LPS for 24 h, according to our previous reports 28,30 . The protein levels of Src and p-Src were then assessed by western blotting (Fig. 1A, B,C). The results showed an increase of p-Src in MIN6 cells that were exposed to glucolipotoxicity, or treated with TNFα or LPS (P < 0.05), while the protein level of Src was stable (P > 0.05). Primary islets were isolated from male db/ db mice (originally bred from C57BL/6J mice) and normal male ICR mice (C57BL/6J). The protein level of p-Src was also increased in primary islets from db/db mice compared with the control mice ( Fig. 1D, P < 0.05), suggesting that Src is activated in vivo in response to glucolipotoxicity. Activated Src in primary islets was also examined using IFA with an anti-p-Src antibody (red fluorescence Fig. 1E). Activated Src was significantly elevated in islets isolated from male db/db mice compared with normal male ICR mice (C57BL/6J). An anti-insulin antibody was employed to distinguish islet β cells from non-β cells in this experiment. The data suggested that Src tyrosine kinases are activated in MIN6 cells in response to risk factors for the development of T2DM.
Proliferation of pancreatic β cells is inhibited in rodent models of T2DM. We next investigated whether activated Src is involved in the proliferation of pancreatic β cells in response to the stimuli related to T2DM. After exposure of MIN6 cells to glucolipotoxicity (0.4 mM palmitate plus 16.7 mM glucose), TNF-α (80 nM) or LPS (10 μ g/mL) for 24 h 28,30 , the percentage of EdU-positive β cells decreased compared with the control ( Fig. 2A, P < 0.05). Meanwhile, flow cytometry after PI labelling showed that the percentage of MIN6 cells increased in the G1 phase ( Fig. 2B, P < 0.05), but decreased in the S phase ( Fig. 2C, P < 0.05) in response to glucolipotoxicity, TNF-α or LPS. These data indicated that proliferation did not increase after the cells were treated with factors related to T2DM. These factors led to cellular senescence via G1 cell cycle arrest in MIN6 cells. Considering the data in Fig. 1, we hypothesized that activated Src did not improve cell viability and proliferation in MIN6 cells in response to the risk factors of T2DM.

RIG-I is upregulated in rodent models of T2DM.
To explore the relationship between the PRR RIG-I and risk factors related to T2DM, cells were exposed to risk factors of T2DM. Additionally, retinoic acid (RA) is a specific agonist of RIG-I. In MIN6 cells exposed to glucolipotoxicity (0.4 mM palmitate plus 16.7 mM glucose), LPS (10 μ g/mL) or RA (20 μ M) for 24 h, or transfected with plasmid RIG-I, qRT-PCR analysis indicated that the mRNA level of RIG-I was obviously increased (Fig. 3A, * P < 0.05). Western blotting showed that glucolipotoxicity, LPS, RA or transfection with RIG-I remarkably increased the RIG-I protein levels ( Fig. 3B P < 0.05). The mRNA level and protein level of RIG-I were also increased in primary islets from db/db mice compared with the corresponding controls (Fig. 3C,D; P < 0.05), suggesting that RIG-I increased in response to in vivo glucolipotoxicity. Meanwhile, RIG-I was also examined by IFA using an anti-RIG-I antibody (red fluorescence, Fig. 3E). RIG-I was increased in cells treated with glucolipotoxicity, LPS, RA or RIG-I-transfection compared with the control group. These data showed that RIG-I was upregulated in rodent models of T2DM, accompanied by activated Src signals.
Elevated RIG-I inhibits proliferation in a pancreatic β cell line. Insulin-like growth factor I (IGF-1) was reported to induce cell proliferation, and is closely associated with Src activity 31 . To explore the role of RIG-I in Src signalling and cell proliferation in pancreatic β cells, we performed a series of experiments using IGF-1 treatment. RA, a RIG-I specific agonist, was used in the experiment at different concentrations (0, 1, 5, 10, 20, 50 and 100 μ M) for different times (6, 12 and 24 h) with IGF-1 (10 nM) 28 to investigate the role of RIG-I in the pancreatic β cell line. MTT assays indicated that RA at more than 5 μ M inhibited the viability of MIN6 cells in a dose-dependent manner (Fig. 4A). We then investigated whether RIG-I could inhibit the proliferation of pancreatic β cells. MIN6 cells were subjected to starvation for 24 h in medium depleted of amino acids and serum, followed by IGF-I treatment in serum-free medium 28 , and then treated with RA (10 μ M) or transfected with plasmid RIG-I. The percentage of EdU-positive β cells increased in the cells treated with IGF-1 only, while the percentage

Elevated RIG-I inhibits proliferation via stabilization of P27 in pancreatic β cell line. Elevated
RIG-I inhibits Src-induced proliferation and the phenomenon is associated with the cell cycle. Thus, cyclin D1, cyclin E, CDK2, P27Kip1 and p21 were analysed at the mRNA and protein levels. MIN6 cells were subjected to starvation for 24 h in medium depleted of amino acids and serum, followed by IGF-I treatment in serum-free medium. Cells were transfected with plasmid RIG-I or treated with RA. QRT-PCR analysis indicated that the cyclin D1 and cyclin E mRNAs decreased slightly and CDK2, p27 and p21 mRNA levels remained stable ( Fig. 5A, P > 0.05). Meanwhile, western blotting showed that RIG-I transfection induced a decrease in cyclin E protein levels and increased P27 levels ( Fig. 5B,C, P < 0.05), whereas other protein levels were unaffected (P > 0.05). To explore the possible involvement of the RIG-I pathway in β -cells upon IGF-induced downregulation of P27 protein levels, MG132, a specific inhibitor of the proteasome, was used a subsequent experiment. MG132 remarkably upregulated the protein levels of P27Kip1 in response to IGF stimuli ( Fig. 5D, P < 0.05), but did not affect the change of P27Kip1 when RIG-I was elevated ( Fig. 5D, P > 0.05). These data suggested that RIG-I-induced accumulation of P27Kip1 might be attributed to increased stability of P27 and impairment of ubiquitin-dependent proteasome degradation of this protein.

RIG-I induced downregulation of Skp2, which is involved in the stability of P27 in pancreatic β cells.
The periodicity of P27 depends on ubiquitin degradation, and the specificity of ubiquitination is provided by the E3 ligases. Skp2 is an E3 ligase that targets P27. To confirm whether Skp2 is involved in this process, IGF-1, RA  and plasmid with RIG-I were used to treat MIN6 cells. QRT-PCR and western blotting analyses showed that the SKP2 mRNA and protein levels were reduced (Fig. 6A,B) after MIN6 cells were treated with RA or transfected with the RIG-I plasmid, which were significantly different to cells treated with IGF-1 only (P < 0.05). To explore the role of Skp2 in P27 protein stability, MIN6 cells were transfected with a SKP2 plasmid. SKP2 overexpression  reversed RA-induced upregulation of P27 protein levels (Fig. 6C, P < 0.05), suggesting that RIG-I regulates P27's function through Skp2-induced protein degradation. Glucolipotoxicity is involved in the increment of RIG-I protein levels and proliferation inhibition of pancreatic β -cells; therefore, to explore the molecular mechanisms underlying role of RIG-I in glucolipotoxicity-induced proliferation inhibition, MIN6 cells were treated using 0.4 mM palmitate plus 16.7 mM glucose. The Skp2 protein level decreased and P27 protein level increased in response to glucolipotoxicity. Skp2 overexpression could reverse the increment in the P27 protein level elicited by glucolipotoxicity (Fig. 6D, P < 0.05). These data indicated that RIG-I induced downregulation of the Skp2 protein level, resulting in increased stability of P27, which might be associated closely with glucolipotoxicity.  the mechanism underlying altered SKP2 transcription, we used plasmid STAT3 to transfect MIN6 cells. After transfection, RA or glucolipotoxicity were used to stimulate the cells. QRT-PCR and western blotting analyses showed that STAT3 overexpression could reverse the RA-induced downregulation of SKP2 mRNA and protein levels (Fig. 7A,B, P < 0.05), suggesting that STAT3 is involved in the regulation of SKP2 transcription. To explore the molecular mechanisms underlying the role of STAT3 in glucolipotoxicity-induced dysfunction of pancreatic β -cells, MIN6 cells were treated using 0.4 mM palmitate plus 16.7 mM glucose. STAT3 overexpression reversed the reduced mRNA and protein levels of Skp2 in response to glucolipotoxicity (Fig. 7A,C, P < 0.05). These data suggested that STAT3 could be an important regulator controlling SKP2 gene expression in pancreatic β -cells.
Transcriptional activity of STAT3 was suppressed by RIG-I accumulation. STAT3 contains an Src homology 2 (SH2) domain that is activated by tyrosine phosphorylation in response to a wide variety of cytokines and growth factors 33 . To explore the relationship between STAT3 and factors that influence Src signals, cells were subjected to different treatments. We used si-RIG-I to transfect MIN6 cells and treated the cells with RA. The protein level of p-STAT3 decreased after the cells were treated with RA only, and si-RIG-I plasmid transfection reversed the decrease (Fig. 8A, P < 0.05). We then used IGF-1 to treat MIN6 cells grown in serum-free DMEM.  Plasmid RIG-I and RA were used to treat the cells. The protein level of p-STAT3 was inhibited more in cells treated with IGF-1 and RA or transfected with plasmid RIG-I compared with IGF-1 treatment only (Fig. 8B, P < 0.05). Using IFA with an anti-p-STAT3 antibody (green fluorescence Fig. 8C), p-STAT3 was observed to be increased significantly in the nucleus when treated with IGF-1 compared with the mock group grown in serum-free DMEM; however, it was reduced in cell nucleus when treated with RA. To explore the STAT3 transcription activity, MIN6 cells were transfected with a STAT3-dependent reporter construct pGMSTAT3-Lu and then treated as in Fig. 8B. Compared with the mock group or IGF-1 treatment, the luciferase activity was reduced significantly in the group subjected to RA (Fig. 8D, P < 0.05). The protein level of p-STAT3 was also decreased in primary islets from db/db mice compared with the corresponding control (Fig. 8E, P < 0.05), suggesting that STAT3 is activated in response to in vivo glucolipotoxicity. These data indicated that RIG-I was involved in the transcriptional activity of STAT3 in a phosphorylation-dependent manner.

RIG-I interrupts the binding between Src and STAT3.
STAT3 is the downstream substrate of Src and binds to its SH2 domain corresponding sites. Src induces STAT3 dimerization and entry into the nucleus, resulting in the activation of the transcription and translation of target genes. To explore role of RIG-I in its competitive binding with Src and inhibition of Src activity 34 , MIN6 cells were grown in serum-free DMEM and treated with IGF. Western blotting was performed to detect binding of RIG-I and STAT3 with activated Src after co-immunoprecipitation with specific p-Src antibodies. After RIG-I transfection or RA treatment, binding of active Src with RIG-I increased, while binding of active Src with STAT3 was attenuated (Fig. 9A, P < 0.05). To investigate whether RIG-I modulated the Src and STAT3 interaction, the si-RIG-I plasmid was used in addition to RA treatment. Binding of active Src with STAT3 was increased compared to RA use only (Fig. 9B, P < 0.05). Furthermore, the STAT3 plasmid was also used to confirm these observations. Binding of active Src with RIG-I was inhibited and binding of active Src with STAT3 was not decreased after transfection with the STAT3 plasmid and treatment with RA compared with treatment with RA only (Fig. 9C, P < 0.05). Interestingly, the total Src level was not altered in these three experiments. EdU experiments were also carried out. Transfection with plasmid STAT3 could reverse the decrease in the percentage of EdU-positive β cells induced by RA (Fig. 9D, P < 0.05). These results suggested that RIG-I inhibits the Src/STAT3 association via competitive binding with active Src, leading to inhibition of pancreatic β cells proliferation.

Discussion
As a metabolic disease, T2DM is characterized by insulin secreted from pancreatic β cells that cannot meet the demands of insulin sensitivity or insulin resistance, which might result from over-nutrition with chronic hyperglycaemia and hyperlipidaemia. In the state of excess nutrition, proliferative modulator Src is activated in diabetic islets 35,36 , which was confirmed by our data (Fig. 1). However, such a state did not promote β -cell proliferation, and did not increase the compensatory function (Fig. 2). Perhaps in the initial stage of over-nutrition, activated Src exerts a positive effect on β cell growth; however, with development of glucolipotoxicity and metaflammation, Src signalling is impaired by other negative modulators.
The two systems of metabolism and immunity are so integrated that more and more pattern recognition receptors are being identified to play important roles in metabolic disorders 37 . For example, activated PKR and elevated TLR3 upon glucolipotoxicity and pro-inflammatory cytokines stimuli both inhibit β cell proliferation to downregulate their functions, leading to failure of β cell compensation in the progression of type 2 diabetes 23,28,38 . RIG-I is another PRR thatthat is present in islet β cells. As shown in Fig. 3, RIG-I was significantly upregulated in the models of T2DM. Previous studies indicated that TNF-α , a cytokine that regulates innate immune responses, induced the expression of RIG-I in endothelial cells. The ability of TNF-α to upregulate RIG-I required protein synthesis, expression of functional type I IFNRs, and STAT1 signalling 39 . Hyperglycaemia causes oxidative stress in pancreatic beta cells of T2DM 40 . ROS might enhance RIG-I levels in pancreatic beta cells 41 . Thus, it necessary to determine whether Rig-I is a negative regulator involved in Src signals and β -cell proliferation.
Before exploring the potential association between RIG-I and Src-induced proliferation in pancreatic β cells, we obtained insights into the effect of RIG-I on β cell growth. Previous reports suggested that RIG-I mediated β cell lesions 42 ; we showed that elevated RIG-I decreased β cell viability and abrogated IGF-stimulated proliferation through cycle arrest at the G1 phase. By contrast, knockdown of RIG-I could alleviate RA-mediated proliferation inhibition (Fig. 4). The results revealed that RIG-I is involved in inhibition of β -cells proliferation elicited by IGF-Src signals.
Recent studies have established that RIG-I restrains myeloid progenitor proliferation through competitive binding with active Src 4 . The recognition between RIG-I and Src acts through a cooperative interaction involving RIG-I recruitment domains (CARDs) and the Src SH1 domain. In inactive Src, the SH1 domain faces inward and is inaccessible to outside factors; therefore, the RIG-I CARDs preferentially associates with activated Src with an exposed SH1 domain. C-terminal to the RIG-I CARDs there is a classic PxxP motif. The PxxP motif can combine with the Src SH3 domain to prevent activated Src from inducing proliferation. Consistent with this finding,our results suggested that RIG-I could bind active Src competitively to inhibit pancreatic β cells proliferation elicited by IGF-Src signals (Figs 4 and 8).
RIG-I levels seemed to correlate with augmented β cells at G1 phase and reduced levels at S phase in a pancreatic β -cell line (Fig. 2), which was associated with several cell cycle modulators, such as cyclin D, P21, P27 and P53. In our experiments, accumulated RIG-I increased the protein level of P27, but there was no change at the mRNA level, suggesting post-translational regulation. Additionally, we found that MG132 could reverse significantly the decline of P27 in β cells without RIG-I function (Fig. 5). This evidence indicated that RIG-I-induced accumulation of P27 could be attributed to impaired ubiquitin-dependent proteasome degradation of P27. A similar mechanism underlying the degradation of P27 has been reported previously. For example, Prasad et al. demonstrated cytoplasmic P27 accumulation after treatment with MG132 in primary cervical cancer samples and cervical cancer-derived cell lines 43 . Mounting evidence suggests that the S-phase kinase-associated protein 1 (SKP1)/CUL1/F-box protein Skp2 acts as an E3 ubiquitin ligase and is involved in the ubiquitination and proteasome-dependent degradation of P27 [44][45][46][47] . Our analyses suggested the participation of Skp2 in RIG-I-mediated P27 stability. Correspondingly, Skp2 was downregulated by elevated RIG-I and glucolipotoxicity-stimulated P27 stability was abrogated when Skp2 was overexpressed (Fig. 6). Skp2 has been suggested to play a critical and specific role in regulating the cellular abundance of P27, and acts as an essential determinant of β cell proliferation. For example, diminished β cell mass, hypoinsulinaemia and glucose intolerance occurred in Skp2(− /− ) mice 48 .
Meanwhile, several studies indicated that the transcript level of SKP2 depends on STAT3. STAT3 is the most well characterized signalling pathway for increasing the nuclear localization and transactivation of genes, and is thought to promote cell growth and survival through multiple mechanisms, including increased expression of oncogenes, such as c-myc, Skp2 and cyclin D1. STAT3 can regulate the transcription of Skp2 by recruiting p300 and then binding to the enhancer region of the SKP2 gene. Our studies indicated that STAT3 controlled SKP2 gene expression as an important regulator in pancreatic β -cells, in which STAT3 overexpression can reverse the decrease in the Skp2 protein levels elicited by RIG-I (Fig. 7).
STAT3 is also the downstream substrate of Src 49,50,18 . STAT3 can be activated by tyrosine phosphorylated Src and is then translocated to the nucleus. In the nucleus, STAT3 dimmer and tetramers bind to specific DNA response elements to induce or repress gene transcription. The transcriptional activity of STAT3 was suppressed in in vivo models of T2DM 51,52 . Therefore, we hypothesized that elevated RIG-I might interrupt the binding of activated Src and STAT3. The hypothesis was confirmed by the co-immunoprecipitation and immunofluorescence assays, which showed that the physical interaction between RIG-I and phospho-Src resulted in little or no binding between STAT3 and phospho-Src and that elevated RIG-I inhibited the phosphorylation of STAT3 and its translocation to the nucleus (Figs 8 and 9). Furthermore, overexpression of STAT3 could compete with RIG-I to bind with activated Src and reverse the glucolipotoxicity-induced decrease of Skp2, thereby increasing the level of P27 to abrogate the RIG-I-dependent antiproliferative effect (Figs 7 and 9). The data showed that RIG-I competitively binds active Src and then inhibits the interaction between Src and STAT3, abrogating their proliferation effect in models of T2DM.
In summary, RIG-I is significantly activated by glucolipotoxicity and metaflammation, and is implicated in β cell mass loss through proliferation inhibition, leading to body decompensation in the course of T2DM. Via its PxxP motif, elevated RIG-I competes with STAT3 for activated Src and then blocks the Src/STAT3/Skp2 pathway to accumulate P27 function in pancreatic β cells, leading to cell cycle arrest at the G1 phase. These data provide a new insight into the pathogenesis of T2DM and suggest RIG-I as a novel drug target.