Loss of Glycine N-Methyltransferase Associates with Angiopoietin-Like Protein 8 Expression in High Fat-Diet-Fed Mice

Imbalance of lipid metabolism is a main cause of metabolic syndrome leading to life-threatening metabolic diseases. Angiopoietin-like protein 8 (Angptl8) was recently identified as a liver and adipose tissue-released hormone that is one of the molecules involved in triglyceride metabolism. However, the regulatory mechanism of Angptl8 is largely unknown. A high fat diet (HFD)-fed mouse model, which showed high cholesterol, high triglyceride, and high insulin in the blood, revealed the upregulation of hepatic and plasma Angptl8 and the downregulation of hepatic glycine N-methyltransferase (GNMT). The inverse correlation of hepatic Angptl8 and GNMT expression in the livers of HFD-fed mice was also confirmed in a publicly available microarray dataset. The mechanistic study using primary hepatocytes showed that the Angptl8 expression could be induced by insulin treatment in a dose- and time-dependent manner. Inhibition of PI3K/Akt pathway by the specific inhibitors or the dominant-negative Akt blocked the insulin-induced Angptl8 expression. Moreover, knockout of GNMT promoted the Akt activation as well as the Angptl8 expression. These results suggested that GNMT might be involved in insulin-induced Angptl8 expression in HFD-mediated metabolic syndrome.


Introduction
Obesity and metabolic syndrome due to over-nutrition and sedentary lifestyle are the risk factors for nonalcoholic fatty liver disease, cardiovascular disease, and type 2 diabetes [1,2]. In recent years, more attention has been paid to the proteins released from muscle, adipose, and liver tissues that affect lipid and glucose metabolism systematically [3][4][5]. The metabolism-related organokines include

Increase of Plasma Insulin in High-Fat Diet Mice
To access the regulation of Angptl8 in metabolic syndrome, we first established a HFD-induced obese mouse model. The mice were fed with normal diet (ND) or HFD for 12 weeks. The food intake in a HFD group and a ND group was not different (Table 1). However, the body weight, body mass index (BMI), and food efficiency ratio (FER) were higher (p < 0.05) in a HFD group compared to these in a  Figure 1A,B). The liver weight and hepatosomatic index (HSI) of mice in the HFD group was significantly higher than that of mice in the ND group (Table 1). After a 12 week feeding of ND or HFD, the fasting blood triglyceride (TG) levels of the mice in the HFD group did not differ from the ND group ( Figure 1C). However, the fasting blood glucose and total cholesterol levels of mice in the HFD group were higher than those of mic in the ND group ( Figure 1D,E). The steady-state blood insulin was increased in the HFD group compared to that in the ND group ( Figure 1F). On the other hand, the hepatic Angptl8 protein and mRNA levels were elevated in the livers of HFD-fed mice compared to its levels in the livers of ND-fed mice ( Figure 1G,H). In parallel, the plasma Angptl8 levels were increased in the HFD-fed mice compared to the ND-fed mice ( Figure 1I). feeding of ND or HFD, the fasting blood triglyceride (TG) levels of the mice in the HFD group did not differ from the ND group ( Figure 1C). However, the fasting blood glucose and total cholesterol levels of mice in the HFD group were higher than those of mic in the ND group ( Figure 1D,E). The steady-state blood insulin was increased in the HFD group compared to that in the ND group ( Figure  1F). On the other hand, the hepatic Angptl8 protein and mRNA levels were elevated in the livers of HFD-fed mice compared to its levels in the livers of ND-fed mice ( Figure 1G,H). In parallel, the plasma Angptl8 levels were increased in the HFD-fed mice compared to the ND-fed mice ( Figure 1I).

Insulin Induces Akt Phosphorylation and Angptl8 Expression
Since plasma insulin was increased in the HFD-fed mice, we hypothesized that insulin-mediated signaling pathways could be the upstream regulators of Angptl8. Accordingly, we isolated primary hepatocytes from the livers of ND-fed C57BL/6J mice. The primary hepatocytes were treated with different concentrations of insulin for 1 h. The data showed that the insulin treatment induced Angptl8 expression and phosphorylation of Akt at Threonine 308 (T308) and Serine 473 (S473) residues and in a dose-dependent manner ( Figure 2A). However, the insulin treatment did not alter the phosphorylation of ERK1/2. On the other hand, the treatment of 1 M insulin time-dependently increased Angptl8 expression within 1 h and the levels of Angptl8 reached a plateau after 1 h of insulin treatment ( Figure 2B). The insulin-induced Akt phosphorylation at T308 and S473 reached a peak after 1 h of the treatment. Similarly, the phosphorylation of ERK1/2 was not changed during the periods of insulin treatment. The results from qPCR showed that the insulin treatment induced a dose-and time-dependent upregulation of Angptl8 mRNA expression ( Figure 2C,D).

Insulin Induces Akt Phosphorylation and Angptl8 Expression
Since plasma insulin was increased in the HFD-fed mice, we hypothesized that insulin-mediated signaling pathways could be the upstream regulators of Angptl8. Accordingly, we isolated primary hepatocytes from the livers of ND-fed C57BL/6J mice. The primary hepatocytes were treated with different concentrations of insulin for 1 h. The data showed that the insulin treatment induced Angptl8 expression and phosphorylation of Akt at Threonine 308 (T308) and Serine 473 (S473) residues and in a dose-dependent manner ( Figure 2A). However, the insulin treatment did not alter the phosphorylation of ERK1/2. On the other hand, the treatment of 1 M insulin time-dependently increased Angptl8 expression within 1 h and the levels of Angptl8 reached a plateau after 1 h of insulin treatment ( Figure 2B). The insulin-induced Akt phosphorylation at T308 and S473 reached a peak after 1 h of the treatment. Similarly, the phosphorylation of ERK1/2 was not changed during the periods of insulin treatment. The results from qPCR showed that the insulin treatment induced a dose-and time-dependent upregulation of Angptl8 mRNA expression ( Figure 2C,D).

Insulin-Induced Angptl8 is Mediated by PI3K/Akt Signaling Pathway
To investigate whether Angptl8 expression can be regulated by insulin-induced Akt signaling pathway, the primary hepatocytes were pre-treated with a PI3K inhibitor, LY294002, for 20 min, followed by a treatment of 1 µM insulin for 1 h. The results showed that the pre-treatment of LY294002 inhibited insulin-induced Akt phosphorylation at T308, as well as insulin-induced Angptl8 protein and mRNA expression ( Figure 3A,B). In parallel, the insulin-induced Angptl8 protein and mRNA expression was inhibited by a selective Akt inhibitor, MK2206 ( Figure 3C,D). To further confirm the role of Akt signaling on Angptl8 expression, a dominate-negative Akt plasmid (dnAkt), a kinase-dead form of Akt, or an empty vector was transfected into the primary hepatocytes, followed by the treatments of insulin. The results showed that the cells expressing dnAkt significantly inhibited the protein and mRNA expression of Angptl8 ( Figure 3E,F). Furthermore, we next studied whether the activation of Akt is sufficient to elevate Angptl8 expression. Ectopic expression of a constitutively active form of Akt, myrAkt, significantly increased Angptl8 expression with or without insulin treatment. These results demonstrated that the insulin-induced PI3K-Akt signaling pathway is an upstream regulator of Angptl8. To investigate whether Angptl8 expression can be regulated by insulin-induced Akt signaling pathway, the primary hepatocytes were pre-treated with a PI3K inhibitor, LY294002, for 20 min, followed by a treatment of 1 μM insulin for 1 h. The results showed that the pre-treatment of LY294002 inhibited insulin-induced Akt phosphorylation at T308, as well as insulin-induced Angptl8 protein and mRNA expression ( Figure 3A,B). In parallel, the insulin-induced Angptl8 protein and mRNA expression was inhibited by a selective Akt inhibitor, MK2206 ( Figure 3C,D). To further confirm the role of Akt signaling on Angptl8 expression, a dominate-negative Akt plasmid (dnAkt), a kinase-dead form of Akt, or an empty vector was transfected into the primary hepatocytes, followed by the treatments of insulin. The results showed that the cells expressing dnAkt significantly inhibited the protein and mRNA expression of Angptl8 ( Figure 3E,F). Furthermore, we next studied whether the activation of Akt is sufficient to elevate Angptl8 expression. Ectopic expression of a constitutively active form of Akt, myrAkt, significantly increased Angptl8 expression with or without insulin treatment. These results demonstrated that the insulin-induced PI3K-Akt signaling pathway is an upstream regulator of Angptl8.

GNMT/PI3K/Akt Signaling Cascade Regulates Angptl8 Expression
To investigate whether GNMT may be involved in Angptl8 regulation, we first analyzed the GNMT expression in HFD-fed mice. The hepatic GNMT protein levels of mice in a HFD group were downregulated compared to that of mice in a ND group ( Figure 4A). The GNMT mRNA levels were also significantly decreased in the livers of HFD-fed mice compared to the mice in a ND group ( Figure 4B). To further confirm the relationship between GNMT and Angptl8 expression, we analyzed the mRNA levels of GNMT and Angptl8 in a publicly available microarray dataset, GSE39549, see References [32,33]. In this microarray dataset, the C57BL/6J mice were fed with ND or HFD for multiple time periods and the adipose tissues of mice were subjected to microarray analysis. The Angptl8 mRNA expression in the adipose tissues of the HFD-fed mice was higher than that of the ND-fed mice (p < 0.05), while GNMT mRNA expression was downregulated in the adipose tissues of the HFD-fed mice (p < 0.05) ( Figure 4C). Pearson's correlation showed that the Angptl8 mRNA expression inversely correlated with the GNMT mRNA expression (p < 0.01) ( Figure 4D). In the livers of GNMT-knockout (GNMT-KO) mice, total Akt and phosphorylated Akt at T308 was increased and the hepatic Angptl8 protein levels were also upregulated compared to that in the livers of wild-type mice ( Figure 4E). These results indicated that GNMT might be an upstream regulator of Angptl8. In the primary hepatocytes isolated from GNMT-KO mice, LY294003 and MK2206 still inhibited that Angptl8 expression ( Figure 4F). These results suggested a GNMT/PI3K/Akt signaling cascade that can regulate Angptl8 expression.

Discussion
Elevation of Angptl8 levels in circulation has been recently considered as a potential biomarker for metabolic diseases [34]. This phenomenon could be observed in multiple animal models, such as non-alcoholic fatty liver disease and non-alcoholic fatty liver disease of HFD or methionine-choline deficient (MCD) diet-fed mice [35], as well as obesity-related syndrome of ob/ob or db/db mice [16]. In a physiological condition, e.g., in normal diet-fed mice, hepatic Angptl8 levels are rhythmically expressed under the regulation of glucocorticoid signaling and the LXRα pathway [25]. A recent study also showed that the Angptl8 mediated food-driven resetting of hepatic circadian clock [36]. In our study, the increase of total cholesterol, insulin, and glucose in the blood of HFD-fed mice indicated a metabolic syndrome in the mice. In agreement with the findings from others, the hepatic and plasma Angptl8 was increased in the HFD-fed mice. We therefore hypothesized that plasma insulin might be one of the factors that promotes Angptl8 production. Indeed, our in vitro study using primary hepatocytes demonstrated that the insulin-activated PI3K/Akt signaling pathway significantly induced Angptl8 expression in a dose and time-dependent manner. Blockage of the PI3K/Akt signaling cascade using the PI3K and Akt inhibitors, as well as the dominant-negative Akt, inhibited Angptl8 expression. For an unknown reason, we found the treatment of insulin resulted in the phosphorylation of Akt, but did not alter the phosphorylation levels of ERK1/2 in the primary hepatocytes. Further experimentation is required to study the protein switches that may control the differential activation of PI3K/Akt and ERK1/2 signaling upon insulin stimulation in primary hepatocytes. Our data indicated that elevation of plasma insulin is an important factor that induces the upregulation of hepatic Angptl8 expression. Consistent with this result, a previous study showed that the expression of Angptl8 was only upregulated in the liver, white adipose tissue and brown adipose tissue of obesity-induced hyperinsulinemic type 2 diabetes mice, but it was decreased in white adipose tissue of streptozotocin-induced hypoinsulinemic type 1 diabetes mice [37]. Together, our data and the results from other studies show that insulin-mediated PI3K/Akt signaling pathway is a major upstream regulator for HFD-induced Angptl8 expression, but not blood glucose. However, adipose tissue and muscle also act as endocrine organs that release various cytokines, such as leptin, adiponectin, tumor necrosis factor-alpha, and interleukin-6, and alter insulin sensitivity in obesity and metabolic syndrome [38]. A recent study showed that the Angptl8 expression was associated with hyperactive catabolism of the extracellular matrix (ECM) and inflammation [39]. Accordingly, we cannot exclude other factors, such as proinflammatory cytokines, may also be involved in HFD-induced Angptl8 regulation. Further experiments are required to investigate the effects of liver-, adipose tissue-, and musclederived cytokines in Angptl8 expression.   All angiopoietin-like proteins (Angptl1-8) are secretory glycoproteins. Among the family members, Angptl3, Angptl4, and Angptl8 are involved in lipoprotein metabolism by governing lipoprotein lipase activity [40]. Structurally, they share an N-terminal coiled-coil domain, which is required for the inhibition of lipoprotein lipase activity. Unlike Angptl3 and Angptl4, Angptl8 lacks a C-terminal fibrinogen-like domain, which has antiangiogenic functions [11]. Functionally, overexpression of hepatic Angptl8 elevated plasma triglyceride levels in wild type mice, but not in Angptl3-KO mice [12,14], suggesting that Angptl3 and Angptl8 work together to regulate triglyceride metabolism. However, Angptl4 plays a distinct function rather than Angptl3 and Angptl8. Angptl3 and Angptl8 promotes the post-prandial flux of triglyceride into adipose tissue in fed state, while Angptl4 prevents the uptake of circulating triglyceride into adipose tissue in fasted state [41,42]. Accordingly, elevation of Angptl8 may not be the only cause that increases plasma triglyceride. More studies are required to elucidate the regulatory network of angiopoietin-like proteins in pathological conditions.
In liver, GNMT is an abundant enzyme that contributes to the conversion of SAM, which is the methyl donor required for the methylation of phosphatidylethanolamine (PE). PE can be converted into phosphatidylcholine (PC) through the phosphatidylethanolamine N-methyltransferase (PEMT) pathway. Accumulation of hepatic SAM and PC is found in the livers of GNMT-KO mice [43]. The excess of SAM and PC contribute to the triglyceride synthesis and release. Indeed, the increase of hepatic SAM and the decrease of hepatic GNMT were linked to non-alcoholic fatty liver disease in a HFD-fed rodent model [44]. The down-regulation of GNMT can be through the DNA methylation of its promoter region and microRNA-mediated post-transcriptional regulation in disease models and clinical studies, liver fibrosis and cirrhosis, and HCC [45,46]. In our study, we showed that downregulation of GNMT might contribute to the elevation of circulating triglyceride levels through the upregulation of Angptl8. In our data, the loss of GNMT resulted in the increase of Angptl8 expression and Akt phosphorylation in primary hepatocytes isolated from GNMT-KO mice. The blockage of Akt phosphorylation by the PI3K inhibitor or the Akt inhibitor inhibited Angptl8 expression in the primary GNMT-KO hepatocytes. Since Angptl8 is a lipid metabolism-associated hormone that is released mainly from liver and adipose tissue, our study provides the possibility that HFD-induced downregulation of GNMT may increase the production and release of Angptl8, leading to the increase of plasma triglyceride levels in metabolic syndrome.
For its enzymatic function, GNMT is involved in transcriptional regulation of genes by inhibiting the conversion of SAM, leading to the alterations of genomic DNA methylation pattern. In addition, GNMT has been linked to detoxification and anti-oxidation pathways in HCC [47]. Although the molecular mechanism is not clear, phosphorylation of GNMT at Serine 9 residue leads to its nuclear translocation, which promotes the expression of genes with a detoxification function [48]. In the central nervous system, systematic knockout of GNMT resulted in the impairment of learning and memory by diminishing neurogenesis through the excess SAM-induced alteration of the bFGF-stimulated MAP kinase signaling cascade [49]. Recently, GNMT has been considered as an adaptor protein that binds to signaling molecules independent of its enzyme activity [30]. GNMT interacts with PREX2, a novel PTEN inhibitor [31], leading to the degradation of PREX2 through an E3 ligase HectH9-mediated proteasomal ubiquitination pathway. The GNMT-PREX2 interaction causes the increase of PTEN activity and the decrease of Akt phosphorylation as a consequence. In our study, we found that the loss of GNMT resulted in the increase of Akt phosphorylation in primary GNMT-KO hepatocytes and an elevation of Angptl8 expression. Our data indicated that the GNMT expression was associated with Angptl8 expression. Indeed, in our HFD-fed mice and from a microarray dataset analysis, GNMT protein and mRNA levels were inversely correlated with Angptl8 mRNA and protein expression. This association of GNMT and Angptl8 may be due to the inhibitory function of GNMT on the insulin-induced Akt signaling pathway. However, more experiments are required to explore the detail mechanism of GNMT-regulated Angptl8 expression.
In conclusion, our study suggests that downregulation of hepatic GNMT and upregulation of hepatic and circulating Angptl8 may be a potential biomarker for HFD-induced metabolic syndrome.
The inverse correlation of GNMT and Angptl8 expression may be through the inhibitory function of GNMT on insulin-induced Akt phosphorylation. Since GNMT and Angptl8 play important role in the homeostasis of lipid metabolism, targeting GNMT and/or Angptl8 may be a potential therapeutic strategy to overcome diet-induced metabolism syndrome.

Animal Study
Twenty age-matched C57BL/6J mice (6 weeks old; male; Charles River. Technology, BioLASCO Taiwan Co, Ltd., Taipei, Taiwan) were randomly divided into ND-fed and HFD-fed groups with 10 mice each. There was no significant difference in initial body weight among the two groups (mean body weight of ND group = 18.36 ± 0.41 g, mean body weight of HFD group = 18.70 ± 0.40 g, p = 0.61). The mice were fed with a control diet (catalog number: D12450H, 10 kcal% fat, Research Diets, New Brunswick, NJ, USA) or a high-fat diet (catalog number: D12451, 45 kcal% fat, Research Diets) for 12 weeks. Body weight gain and feed consumption were monitored weekly to determine the feed efficiency ratio (FER). Body mass index (BMI) was calculated according to the relationship between body weight and naso-anal length squared. After the end point of experiment, the mice received a 12 h fasting and were anesthetized. The blood samples were collected by cardiac puncture, centrifuged at 800 g for 10 min at 4 • C, and the serum was stored at −80 • C. The livers were collected, weighted, and stored at −80 • C. The GNMT-KO mice in C57BL/6 background [50] and their littermate wild-type control were generated and maintained in specific pathogen-free conditions in accordance with the regulations at the Animal Center, Kaohsiung Medical University. All the animal experiments were performed in accordance with a protocol approved by the Institutional Animal Care and Use Committee of Kaohsiung Medical University (IACUC code: 102186; 25 Dec 2013).

Biochemical Analysis and ELISA
The serum glucose concentrations and the levels of total cholesterol and triglycerides were measured using the Fuji DRI-CHEM NX500V system (Fuji Film, Tokyo, Japan). The plasma insulin levels were determined using a RayBio mouse insulin ELISA kit (catalog number: ELM-Insulin, RayBiotech, Norcross, GA, USA). The plasma Angptl8 levels were measured using a mouse Betatrophin ELISA kit (catalog number: E11633m, Wuhan EIAab Science Co, Ltd., Wuhan, China). The absorbance measurement was performed using a Microplate Spectrophotometer (Epoch, BioTek, Winooski, VT, USA) at 450 nm wavelength.

Quantitative Real-Time PCR
Total RNA was extracted from liver tissues and primary hepatocytes using TRIzol reagent (Invitrogen, Carlsbad, CA, USA). cDNA was synthesized using an MMLV Kit (Invitrogen) according to the manufacturer's instructions. Quantitative real-time PCR (qPCR) was performed using the Fast SYBR Green Master Mix and run on a StepOnePlus real-time PCR system (Invitrogen) with a reaction program as follows: A total of 20 s at 95 • C, followed by 40 cycles at 95 • C for 3 s and annealing at 60 • C for 30 s. The mRNA expression of the indicated genes was determined using the 2 −∆∆Ct method and the data were normalized to those of the housekeeping gene glyceraldehyde-3-phosphate dehydrogenase (GAPDH). The specific primer sequences of the tested genes are listed as follows: mAngptl8, forward: CCAGTTGTGCTGCAAGGAAC, reverse: TTGCTTCTGTCTCCGCTCTG; mGNMT, forward: GTTGACGCTGGACAAAGA, reverse: AGCCTGTGCTGAGGATA; mGAPDH, forward: GGCAAATTCAACGGCACA, reverse: GTTAGTGGGGTCTCGCTCTG.

Statistical Analysis
All the observations were confirmed by at least three independent experiments. The results are presented as the mean ± standard deviation. We used two-tailed paired Student's t-tests for all the pair-wise comparisons. Comparisons between multiple groups were undertaken using one-way ANOVA followed by Dunnett's test. In all the comparisons and differences were considered statistically significant at p < 0.05.