3.1 Diabetic cardiomyopathy occurred in HFD fed db/db mice.
Given that fatty acid (FA) oxidation accounts for 60% to 90% of mitochondrial ATP generation under normal conditions and FA accumulation is characteristic of the diabetic heart, we established an obesity related diabetic mouse model to induce diabetic cardiomyopathy by feeding db/db mice with HFD for 3 months. Our data showed that compared with the db/+ mice, the BW and FBG of db/db mice fed with HFD were significantly higher (Fig.1A), as well as HbA1c and TG (Fig.1B). In addition, the detection of inflammatory markers showed that the concentrations of IL-1β and IL-18 in plasma of db/db mice increased (Fig.1B). H & E staining showed that compared with db/+ mice, db/db mice fed with HFD appeared obvious myocardial hypertrophy, narrowing of left ventricular cavity, myocardial fibrosis and even breakage (Fig.1C). Immunohistochemical staining showed that the CTGF and COL1A1 labeled fibers in the myocardial interstitium of db/db mice were significantly increased, suggesting that HFD induced myocardial fibrosis in db/db mice (Fig.1D). Electron microscope showed that the myocardial myofilament bundles of db/+ mice were arranged neatly, the Z-line and M-line were clearly visible. By contrast, the myocardium of db/db mice was disordered or even broken, the Z- and M-line were blurred (Fig.1E). In addition, more apoptotic cells was observed by TUNEL staining in the myocardial interstitium of db/db mice fed with HFD (Fig.1F). Together, our data indicate that diabetic cardiomyopathy occurred in HFD fed db/db mice.
3.2 Mitochondria were impaired with mtDNA release in cardiomyocytes from HFD fed db/db mice.
As the cytosolic mtDNA derived from damaged mitochondria is a potential inflammatory mediator, we sought to identify the mitochondrial morphology and mtDNA release in DCM. We firstly performed electron microscopic analysis. In db/+ mice, the structure of myocardial mitochondria was complete, the shape was round or oval, and the mitochondrial cristae were complete, rich, and arranged in parallel. Whereas in db/db mice, the arrangement of mitochondria was disordered, swollen, and irregular, and most of the cristae were broken, fused, exfoliated, or even myelinated, and some vacuoles could be seen (Fig.2A). These data confirmed that the mitochondria in cardiomyocytes from DCM were severely impaired. Subsequently, we performed co-immunostaining of Mitofilin, the inner membrane protein, and dsDNA to assess the mtDNA release. As expected, we found that compared with db/+ mice, the signals of Mitofilin in cardiomyocytes of db/db mice were significantly decreased. Interestingly, we observed a significant increase in the number of dsDNA in the cytoplasm of cardiomyocytes from db/db mice (Fig.2B). To quantify the mtDNA release amount, we separated mitochondria and cytosol from the whole-cell for the qRT-PCR experiment. The primer Tert and primer Loop 1-3 were used to detect nuclear DNA and mitochondrial DNA, respectively (Fig.2C). Our results showed that Tert was not detected in the isolated and purified myocardial cytoplasmic DNA (Fig.2D), suggesting that the free dsDNA in the cytoplasm was not the nuclear source, and the cytoplasmic DNA extracted in this study was of high purity, and no obvious nucleolysis occurred. After that, we used primers Loop 1-3 to detect mitochondrial DNA in the isolated and purified cytoplasmic DNA. Consistently, the levels of free Loop1, Loop2, and Loop3 in the cytoplasm of db/db mice were significantly higher than those of db/+ mice (Fig.2E), indicating that the free dsDNA in the cytoplasm was mainly derived from mitochondria. Taken together, Mitochondria were impaired with mtDNA release in cardiomyocytes from HFD fed db/db mice.
3.3 The cGAS-STING-IRF3/NF-κB pathway was activated in hearts of HFD-fed db/db mice.
Given that mitochondrial damage led to mtDNA release into cytoplasm and cGAS is considered to be a cytoplasmic DNA biosensor, we next tested whether the cGAS-STING pathway was activated in hearts from HFD-fed diabetic mice. Expectedly, we found that the expression of cGAS and STING increased significantly in the cardiomyocytes of HFD fed db/db mice (Fig.3A). We also found the cGAS and STING gathered around the nucleus by immunostaining assay (Fig.3B-3C). In addition to activation of the cGAS and STING, the downstream targets, NF- κB and IRF3, was also activated in increased phosphorylated form (Fig.3A, 3D &3E), as well as the expression of NF- κB/IRF3-regulated IL-1β in the cardiomyocytes of HFD-fed db/db mice (Fig.3A). Likewise, the increased mRNA levels of cGAS and STING in HFD-fed db/db mice were confirmed by RT-PCR (Fig.3F), as well as the IL-1β and IL-18 (Fig. 3G). Taken together, these results suggested that The cGAS-STING-IRF3/NF-κB pathway was activated in hearts of HFD fed db/db mice.
3.4 PA-induced mitochondrial ROS led to mitochondrial damage and mtDNA release in H9C2 cells.
To investigate whether the lipotoxicity mediates the activation of the cGAS-STING-IRF3/NF-κB pathway in hearts of HFD fed db/db mice, we next used the H9C2 cell line treated by palmitate acid as a high fat-induced lipotoxic cell model. As shown in Fig. 4A, PA treatment led to an increase of ROS level and mitochondrial damage, which were both reversed by NAC, an inhibitor of ROS, indicating that PA-induced ROS led to mitochondrial damage. To confirm that PA treatment leads to mtDNA release into the cytoplasm, we performed co-immunostaining of mitochondria and dsDNA. As shown in Fig.4B, PA induced increase of cytosolic dsDNA in a dose-dependent manner. Further study by qRT-PCR analysis revealed that the increased cytosolic dsDNA induced by PA was derived from mitochondria (Fig.4C). To investigate the source of ROS in the process of mitochondrial injury induced by PA, we pretreated H9C2 cells with mitochondrial specific ROS scavenger TEMPO. We found that TEMPO could significantly reduce PA-induced intracellular ROS activation and improve mitochondrial membrane potential (Fig.4D). In addition, we evaluated the leakage of mtDNA by fluorescence confocal analysis of dsDNA, mitochondria, and nucleus in PA-treated H9C2 cells. The results showed that mtDNA leakage in the cytoplasm of H9C2 cells treated with PA increased, while TEMPO treatment of H9C2 cells in advance could significantly reduce mtDNA leakage induced by PA (Fig.4E). In summary, these data showed that PA caused mitochondrial damage and mtDNA leakage mainly by activating mitochondrial ROS.
3.5 PA-induced activation of the cGAS-STING pathway in H9C2 cells.
To elucidate the effect of PA-induced mtDNA release, we next evaluated the activation of the cGAS-STING pathway in PA-treated H9C2 cells. As shown in Figure 5A, PA treatment led to an elavated protein level of cGAS and STING in a dose-dependent manner in H9C2 cells. In addition, the downstream targets, phosphorylated IRF3 and NF- κB, were also activated by PA treatment in a dose-dependent manner, as well as IL-1β, which was regulated by IRF3 and NF-κB (Fig. 5A). Given that the function of STING is not only determined by its content but also by its location. We next performed co-immunostaining of STING and Golgi matrix protein 130 (GM130), a Golgi marker. In H9C2 cells without PA treatment, STING was weakly co-located with GM130, while PA treatment induced strong co-localization, which directly indicated the functional activation of STING (Fig.5B). Consistently, the concentration of IL-1β and IL-18 in the supernatant of H9C2 cells after PA treatment were also increased in a dose-dependent manner (Fig.5C), as well as the mRNA levels of cGAS, STING, IL-1β, and IL-18 (Fig.5D). Taken together, these results indicated that activation of the cGAS-STING pathway is involved in PA-induced myocardial inflammation.
3.6 Extracted mtDNA is sufficient to activate cGAS-STING signaling in H9C2 cells.
As the cGAS is not a mtDNA specific DNA sensor, other sorts of DNA can also activate it. To confirm that mitochondria-derived mtDNA is able to activate the cGAS-STING pathway, we isolated and purified mtDNA to transfect into H9C2 cells. Then the activation of the cGAS-STING pathway and downstream inflammatory activation level were detected by Western blot and qRT-PCR. As shown in Fig 6A&6C, cGAS and STING expression was activated after mtDNA transfection, accompanying the increased expression of IL-1β and IL-18. In addition, we performed co-immunostaining of STING and Golgi in PA-treated H9C2 cells to evaluate the activation of STING. The results indicated that STING aggregation to Golgi was significantly increased in mtDNA-transfected H9C2 cells (Fig 6B), suggesting that STING was functionally activated by mtDNA treatment. In summary, these data showed that in PA-induced myocardial inflammation, the released cytoplasmic mtDNA acted as the ligand of the cGAS-STING system.
3.7 Knockdown of STING blocked the PA-induced inflammation and apoptosis in H9C2 cells.
Given that STING functions as an effector in the cGAS-STING system, we sought to identify whether the inhibition of STING can reverse the effect of PA treatment in H9C2 cells. We employed siRNA to knockdown the STING mRNA. In H9C2 cells transfected with STING siRNA, the expression of STING protein was significantly decreased (Fig.7A), and the localization of STING in Golgi was significantly decreased (Fig.7B), indicating that the transfection of STING siRNA was effective. As expected, STING knockdown could significantly inhibit the activation of NF- κB and the increase of IL-1β in H9C2 cells treated by PA for 24 hours (Fig.7A,7C). In addition, STING knockdown could also significantly blocked the elevated secretion of IL-1β and IL-18 induced by PA treatment in the supernatant of H9C2 cells (Fig.7D). Moreover, we also observed a significant anti-apoptotic effect of STING knockdown on PA-treated H9C2 cells (Fig.7E). Taken together, these data directly indicated that knockdown of STING blocked the PA-induced inflammation and apoptosis in H9C2 cells.
3.8 Inhibition of STING ameliorated diabetic cardiomyopathy in HFD fed db/db mice.
Since that knockdown of STING blocked the PA-induced inflammation and apoptosis in H9C2 cells, we supposed STING as a potential therapeutic target of DCM. To this end, we used a specific inhibitor of STING, C176, to intraperitoneally inject into HFD fed db/db mice (Fig.8A). As shown in Fig.8B, inhibition of STING can reverse the cardiac dysfunction in db/db mice fed with HFD, showing an increase in E/A ratio and a shortening of isovolumic relaxation time (IVRT), suggesting an improvement in diastolic cardiac function. In addition, inhibition of STING could partially improve myocardial hypertrophy induced by HFD, but had no significant effect on myocardial contractile function (Fig.8B). To further study the pathological changes, we performed HE staining to observe the cardiac hypertrophy and immunohistochemistry to observe the myocardial fibrosis. The results showed that HFD feeding induced ventricular hypertrophy and myocardial fibrosis in db/db mice, which could be partially reversed by C176 treatment (Fig.8C). Also, HFD feeding induced the increase of inflammatory cytokine IL-1β in db/db mice, while C176 treatment reduced the production of IL-1β (Fig.8C), which was also confirmed by Western blot (Fig.8D). Besides, C176 treatment also blocked the HFD feeding induced activation of NF-κB, the downstream target of STING, in db/db mice by inhibition of phosphorylated P65 (Fig.8E). In a word, these results suggest that STING functions as a potential therapeutic target for diabetic cardiomyopathy.