4. DISCUSSION
In this study, we reported that the DPP-4 inhibitor linagliptin ameliorated hepatic steatosis elicited by the dual IR/IGF1R inhibitor OSI-906. We previously reported that a DPP-4 inhibitor, des-fluoro-sitagliptin, prevented diet-induced adipose tissue inflammation and hepatic steatosis (Shirakawa et al., 2011b). In another study, linagliptin reportedly improved insulin sensitivity and hepatic steatosis in diet-induced obese (DIO) mice (Kern et al., 2012). DPP-4 inhibition with linagliptin has been shown to decrease the expressions of SREBP1c, SCD-1 and FAS, all of which are known to be related tode novo lipogenesis, and increased the expression of PPARα in the livers of DIO mice (Kern et al., 2012). On the other hand, the expressions of genes involved in hepatic fatty acid oxidation were not affected by treatment with linagliptin in DIO or NASH mouse models (Kern et al., 2012; Klein et al., 2014). However, the mechanism by which DPP-4 inhibitors improve fatty liver in the presence of diet-induced obesity or diabetes remains unclear.
IR and IGF-1R-mediated signaling play a crucial role in hepatic insulin action, which regulates glucose and fatty acid metabolism. OSI-906 administration induces the acute inhibition of IR/IGF1R signaling and evokes hepatic steatosis (Tajima et al., 2017). In our mouse model, the dual inhibition of IR and IGF-1R with OSI-906 resulted in impaired insulin signaling in the liver. The liver weight and liver TG contents were significantly increased by the administration of OSI-906 for 7 days. Linagliptin significantly reversed the OSI-906-induced increase in liver weight and TG content. The OSI-906 group exhibited increased gluconeogenic gene expression, compared with the vehicle group, and treatment with linagliptin did not affect the expressions of these genes, consistent with an impairment in hepatic insulin action. The expressions of Gck , Srebp1c , Fas , and PPARα , which were also regulated by insulin signaling, showed no significant alteration by treatment with linagliptin in OSI-906 treated mice. Hence, the effect of linagliptin on hepatic steatosis seemed to be independent of the IR/IGF1R signaling pathway. Our data indicated that although insulin signaling is involved in the development of hepatic steatosis, an alternative pathway that improves liver steatosis without altering insulin signals likely exists in the OSI-906-treated model. The OSI-906 + Lina group had lower glucose levels at 4 hours after OSI-906 administration on days 1 and 2, compared with the levels in the OSI-906 group, but these effects were abolished after day 3. Hyperinsulinemia evoked by OSI-906 was also observed in the OSI-906 + Lina group. Therefore, the glucose-lowering effects of linagliptin did not contribute to the improvement in OSI-906-induced hepatic steatosis. In patients with metabolic syndrome, fatty liver occurs because of an increased uptake of fatty acids accompanied by an impairment in insulin action in adipose tissue. In this study, linagliptin did not reduce the plasma free fatty acid levels and hepatic expression of Cd36 , which is a free fatty acid transporter, in OSI-906-treated mice. Therefore, linagliptin seemed to ameliorate OSI-906-induced hepatic steatosis independently of the fatty acid flux to the liver, consistent with the regulation of fatty acid metabolism via insulin. Thus, linagliptin improved OSI-906-induced hepatic steatosis via a pathway that was independent of insulin signaling, glucose levels, or free fatty acid metabolism.
Liver-specific IR knockout (LIRKO) mice showed severe insulin resistance and glucose intolerance with increased hepatic glucose production, impaired glucose utilization in the liver, and abnormal mitochondrial function because of the failure of insulin activity (Cheng et al., 2009; Michael et al., 2000; Titchenell, Chu, Monks & Birnbaum, 2015). The loss of FoxO1 increases lipogenesis and decreases fatty acid oxidation in hepatocytes. Thus, liver FoxO1-null mice developed hepatic steatosis, accompanied by the upregulation of lipogenic genes. This regulation of lipogenesis by FoxO1 requires intact hepatic insulin signaling. The systemic inhibition of IR with S961 blunted the effects of insulin on hepatic glucose production in liver-specific IR/FoxO1 double knockout LIRFKO mice (I et al., 2015). Thus, systemic insulin signaling might be required to regulate hepatic glucose production and de novolipogenesis. On the other hand, lipodystrophy in adipose tissue-specific IR knockout (F-IRKO) mice causes progressive NAFLD (Softic et al., 2016). Treatment with OSI-906 also induced lipodystrophy and hepatic steatosis, similar to the F-IRKO mouse model. These findings indicate that not only hepatic insulin signaling, but also insulin signaling in adipose tissue plays a crucial role in the development in hepatic steatosis. However, in this study, linagliptin did not reverse the lipodystrophy induced by OSI-906 administration, implying that linagliptin exerts its effect on OSI-906-induced hepatic steatosis without altering either hepatic insulin signaling or insulin signaling in adipose tissue.
Chronic low-grade inflammation is known to contribute to the development of steatosis (Tilg & Moschen, 2010). Interestingly, the inhibition of IR and IGF1R by OSI-906 showed a tendency to decrease inflammatory gene expressions. These results indicate that OSI-906-induced hepatic steatosis is not provoked by the inflammatory response, unlike high-fat diet-induced hepatic steatosis. We previously reported that DPP-4 inhibition prevented diet-induced hepatic steatosis and adipose tissue inflammation in a diabetic mouse model (Shirakawa et al., 2011a). We also showed that the PAI-1 expression levels in adipose tissue were elevated in diet-induced hepatic steatosis (Shirakawa et al., 2011a). In the present study, PAI-1 expression in the liver tended to decrease with linagliptin treatment in the absence or presence of OSI-906, though a significant difference was not observed. Linagliptin reportedly reduced advanced glycation end products (AGE)-related oxidative stress in the kidneys of a type 1 diabetes mouse model (Nakashima, Matsui, Takeuchi & Yamagishi, 2014), whereas no significant differences in the expressions of genes related to oxidative stress were observed in the current study. Thus, in this OSI-906-induced hepatic steatosis model, the identification of a new molecular mechanism for hepatic steatosis via insulin resistance, other than inflammation or oxidative stress, might be useful for understanding the development of hepatic steatosis.
Proteomic and phosphoproteomic analyses revealed some possible mechanisms underlying OSI-906-induced progression in hepatic steatosis and its amelioration by treatment with linagliptin. We focused on molecules such as PLIN2 and sirtuins, which were suggested by a proteomic analysis to be responsible for the effect of linagliptin on hepatic steatosis. PLIN2, a lipid droplet-coating protein, is related to lipid accumulation in the liver and promotes hepatic steatosis (Hall et al., 2010; Libby, Bales, Orlicky & McManaman, 2016; Najt et al., 2016). PPAR γ is also known to be required for the induction of PLIN2. In a quantitative proteomic analysis, OSI-906 administration significantly up-regulated PLIN2, whereas treatment with linagliptin almost canceled the OSI-906-induced elevation in PLIN2. These results imply that the regulation of PLIN2 in the liver is associated with the development of hepatic steatosis induced by OSI-906 and its recovery as a result of DPP-4 inhibition using linagliptin. Similarly, CYP2B10 was up-regulated in OSI-906-treated liver, and treatment with linagliptin reversed this induction. Regarding MUP20, its expression was reduced by OSI-906 and restored by linagliptin. The regulation of Cyp2b10 and Mup20 were consistent with the context reported for high-fat-diet-induced steatosis and its recovery by treatment with resveratrol (Baur et al., 2006). Cyp2b10 is related to lipid metabolism in the liver, and one of the regulators of Cyp2b10 is a constitutive androgen receptor (CAR) (Ohno, Kanayama, Moore, Ray & Negishi, 2014). Major urinary proteins (MUPs), which are secreted proteins, belong to the lipocalin family and are predominantly produced by the liver. A previous report showed that MUP1 increases energy expenditure and improves glucose tolerance (Zhou, Jiang & Rui, 2009). Since resveratrol acts as a sirtuin activator, these data suggest that treatment with linagliptin improves OSI-906-induced hepatic steatosis by regulating sirtuin activity and its upstream and downstream metabolic pathways.
Sirtuin, a nutrient sensor, activates AMPK and downstream fatty acid oxidation in the liver. In fact, NNMT was significantly up-regulated in OSI-906-treated liver in a quantitative proteomic analysis, supporting the impact of sirtuin-mediated metabolic regulation in these mice. Treatment of a human hepatocyte cell line with exendin-4 reportedly increased the transcript levels of sirtuin1 (SIRT1) (Lee et al., 2012). There are several signal transduction pathways that have been proposed to improve hepatic steatosis as a result of incretin-based therapy, such as cAMP-PKA signaling, PI3K-PDK1-Akt/PKB signaling, and AMPK signaling (Samson & Bajaj, 2013). Whether incretin exerts direct effects or facilitates indirect pathways affecting hepatocyte metabolism remains controversial. Moreover, a recent report showed the importance of hepatocyte-derived DPP-4 and obesity on adipose inflammation and insulin resistance (24).
To address the conflicts in the actions of DPP-4 inhibitors and the GLP-1 receptor agonist on the liver, we examined the effect of linagliptin and the GLP-1 analog liraglutide on hepatocyte AML-12 cells. Linagliptin did not inhibit gluconeogenic gene expressions in OSI-906-treated AML-12 cells, consistent with an insulin signaling-independent action of linagliptin on hepatic steatosis. On the other hand, liraglutide showed a tendency to reduce the expressions of gluconeogenic genes in OSI-906-treated cells. These results suggest that the DPP-4 inhibitor and the GLP-1 analog had different effects on insulin signaling in hepatocytes. In addition, the influence of linagliptin on Tnf expression differed between in vivo andin vitro studies, implying an indirect action of linagliptin on hepatocytes that does not involve the regulation of DPP-4 enzymatic activity.
Taken together, these results show an effect of DPP-4 inhibition on hepatic steatosis that is induced by the acute inhibition of IR/IGF1R signaling through an insulin signaling-independent pathway (Figure 8). Our findings support the non-canonical pleiotropic effects of DPP-4 inhibitors without a glucose-lowering effect and suggest the potential of DPP-4 inhibitors as a new treatment for NAFLD. Further investigation of the underlying mechanism is required.