Insulin-Mimicking Bioactivities of Acylated Inositol Glycans in Several Mouse Models of Diabetes with or without Obesity

Insulin-mimetic species of low molecular weight are speculated to mediate some intracellular insulin actions. These inositol glycans, which are generated upon insulin stimulation from glycosylphosphatidylinositols, might control the activity of a multitude of insulin effector enzymes. Acylated inositol glycans (AIGs) are generated by cleavage of protein-free GPI precursors through the action of GPI-specific phospholipase C (GPI-PLC) and D (GPI-PLD). We synthesized AIGs (IG-1, IG-2, IG-13, IG-14, and IG-15) and then evaluated their insulin-mimicking bioactivities. IG-1 significantly stimulated glycogen synthesis and lipogenesis in 3T3-L1 adipocytes and rat isolated adipocytes dose-dependently. IG-2 significantly stimulated lipogenesis in rat isolated adipocytes dose-dependently. IG-15 also enhanced glycogen synthesis and lipogenesis in 3T3-L1 adipocytes. The administration of IG-1 decreased plasma glucose, increased glycogen content in liver and skeletal muscles and improved glucose tolerance in C57B6N mice with normal diets. The administration of IG-1 decreased plasma glucose in STZ-diabetic C57B6N mice. The treatment of IG-1 decreased plasma glucose, increased glycogen content in liver and skeletal muscles and improved glucose tolerance in C57B6N mice with high fat-diets and db/db mice. The long-term treatment of IG-1 decreased plasma glucose and reduced food intake and body weight in C57B6N mice with high fat-diets and ob/ob mice. Thus, IG-1 has insulin-mimicking bioactivities and improves glucose tolerance in mice models of diabetes with or without obesity.


Introduction
The peptide hormone insulin regulates glucose homeostasis by modulating the action of a multitude of metabolic enzymes in the liver, adipose tissue, and skeletal muscle. Upon insulin receptor activation and phosphorylation of the insulin-receptor substrates (IRSs), insulin signaling branches into several pathways leading to glucose uptake, lipogenesis, glycogen synthesis, glycolysis, and other anabolic processes [1]. However, glucose metabolism by activation of glycogen synthase, mitochondrial pyruvate dehydrogenase, and other regulatory enzymes through protein dephosphorylation, remain incompletely explained by these models for insulin action.
The existence of an alternative pathway that involves an insulinmimetic species of low molecular weight was proposed more than 30 years ago [2,3]. Second messengers referred to as inositol phosphoglycans (IPGs) are generated upon insulin stimulation from glycosylphosphatidylinositol (GPI) and control the activity of a multitude of insulin effector enzymes [4]. The minimal structural requirements of IPGs for insulin-mimetic action have been debated. Two types of IPGs, the inositol phosphoglycans A (IPG-A) type and IPG-P type were suggested. The IPG-A type is generated by cleavage of GPI anchors through the action of GPIspecific phospholipase C (GPI-PLC) [5]. The insulin-mimetic effect of IPGs has been demonstrated in adipocytes [6], hepatocytes [7], and muscle cells [6]. In addition, IPGs are involved in mitogenic signaling via type I cytokine receptors [8,9], suggesting a broader role of IPGs as second messengers in the signaling network of hormones and cytokines. However, the mechanistic details of IPG action remain largely unknown.
GPI-anchors with an inositol residue acylated with a fatty acid have been found in animals. A 2-palmitoyl group is added early in GPI biosynthesis, but is usually removed prior to translocation to the cell membrane [10]. These inositol-acylated GPI anchors or protein-free GPIs might serve as precursors of acylated inositol glycans (AIGs) through GPI-PLC or GPI-PLD. GPI-PLD and GPI-PLC also generate alkylacylphosphatidic or alkylacylglycerols, respectively from the hydrolysis of a truncated GPI. It has been reported that insulin activates GPI-PLC and GPI-PLD in adipocytes and hepatocytes [4,6,7,11]. It was shown that GPI-PLD converts protein-free GPIs into AIG [11]. These AIG models other than lipidated GPI anchors, have not been described in the literature.

Methods and Materials
Glycogen synthesis and lipogenesis in 3T3-L1 adipocytes 3T3-L1 fibroblasts were maintained in Dulbecco's modified Eagle's medium (DMEM) with 25 mM glucose plus 10% fetal bovine serum (FBS). Following 2 days at confluence, differentiation was initiated by the addition of DMEM containing 10% FBS, 167 nM insulin, 0.25 mmol/liter dexamethasone, and 0.5 mM isobutylmethylxanthine. Three days later, the medium was replaced with DMEM plus 10% FBS and 167 nM insulin. After 2 more days, the medium was switched to DMEM plus 10% fetal bovine serum. Adipocytes were used 6-14 days after completion of the differentiation protocol, when .90% of the cells expressed the adipocyte phenotype. Prior to experiments, cells were washed two times with DMEM medium containing 5 mM glucose, 0.2% bovine serum albumin, 25 mM Hepes (pH 7.4), 100 units/ml penicillin, 100 units/ml streptomycin and 0.29 mg/ml glutamine, and were incubated in the same medium for 2 h. Glycogen and lipid synthesis were measured in 24-well dishes. Following pretreatment with various concentrations of AIGs in DMSO for 30 min, cells were stimulated for 30 min in the absence and presence of 1-100 nM insulin. Then 1 mCi of [ 14 C]-glucose (approximately 220 cpm/nmol) was added to all wells. After a 60min incubation at 37uC, cells were washed three times on ice with PBS, and adipocytes were collected in 1 ml of distilled water. 400 ml of the cell suspension was added to 400 ml of PBS. The lipids were then extracted overnight with 5 ml of Betafluor (National Diagnostics), and glucose incorporation into lipid was measured by scintillation counting. 400 ml of the cell suspension was added to 600 ml of 50% KOH, and glycogen was precipitated and radioactivity was counted by a liquid scintillation counter. DNA content in 100 ml of the cell suspension was measured using the QIAGEN DNeasy kit, and counts per minutes were converted to picomoles glucose and normalized to the number of cell per dishes.

Lipogenesis in native rat adipocytes
Isolated native rat adipocytes were prepared and lipogenesis was measured as previously [12]. Rat adipocytes were incubated with 6-[ 3 H]-glucose (0.55 mM) and various concentrations of IG-1 and IG-2 for 1 hour. Incorporation of tritium into lipids was measured and is expressed as a percent of the maximal insulin response (%MIR).

Animal studies
All experiments were approved by the Center for Laboratory Animal Research, Tohoku University (Permit Number: 21-Idou-149), and all experiments were conducted according to the Guidelines of the National Institutes of Health and the Tohoku University Guidelines for Animal Care and Use. Male C57BL/6 mice were housed individually and given free access to a standard diet (65% carbohydrate, 11% fat, and 24% protein) or a high-fat diet (Quick Fat; 60.2% carbohydrate, 15.3% fat, and 24.5% protein; Clea Japan, Tokyo, Japan), starting at 5 weeks of age, for 5 weeks. db/db and ob/ob mice also were housed individually and given free access to a standard diet. C57BL/6 mice at 5 weeks of age received a single dose of 75 mg/g body weight streptozotoxin (STZ). 7 days after injections, the mice, whose fasting plasma glucose concentrations were over 400 mg/dl, were used as STZdiabetic mice.
Glycogen contents of the liver and ground hind limb muscles Glycogen was isolated from 30-50 mg of frozen liver and ground hind limb muscles by dissolving the tissue in 30% potassium hydroxide saturated with Na 2 SO 4 for 30 min at 100uC, followed by ethanol precipitation. Glycogen content was determined by the phenol-sulfuric acid spectrophotometric method at 490 nm [13] and expressed as micrograms of glycogen per milligram of liver.

Blood analysis
Blood glucose was assayed with Glucose Ace (Sanwa Kagaku Kenkyusho, Nagoya, Japan). Serum insulin and leptin were determined with ELISA kits (Morinaga Institute of Biological Science, Yokohama, Japan). Serum total adiponectin and tumor necrosis factor-a (TNF-a) concentrations were measured with an ELISA kit (Ohtsuka Pharmaceutical, Tokyo, Japan) and a TNF-a assay kit (Amersham Biosciences, Uppsala, Sweden), respectively. Serum total cholesterol, triglyceride, and FFA concentrations were determined with a Cholescolor liquid, Lipidos liquid (TOYOBO, Osaka, Japan), and NEFA C (Wako Pure Chemical, Osaka, Japan) kits, respectively.

Glucose tolerance tests
Glucose tolerance was assessed with intraperitoneal glucose injection. Glucose tolerance tests were performed on fasted (16 h) mice. Mice were given intraperitoneal glucose (2 g/kg body wt), and blood glucose was assayed immediately before and at 15, 30, 60, and 120 min after administration. Area under the curve (AUC) was calculated by the trapezoid rule for the glucose tolerance curve using the XLSTAT add-in for Microsoft Excel software (Mindware Software).

Intravenous administration of IG-1
Intravenous administration of IG-1 was performed on mice fed ad libitum. Mice received an injection of IG-1 or human regular insulin (1 units/kg body wt for normal diets; 2 units/Kg for high fat diets, db/db, and STZ-diabetic C57B/6N mice; Eli Lilly, Kobe, Japan) from tail veins, and blood glucose was assayed immediately before and at 30, 60, 90, 120, 180, 240 and 300 min after injection.

Statistical Analysis
All data are expressed as mean6SEM. Normality was tested with the Kolmogorov-Smirnov test. When data were normally distributed, the statistical significance of differences was assessed by one-way or two-way ANOVA. Multiple experimental groups were compared by use of Tukey-Kramer post hoc test. In all analyses, values of P,0.05 were accepted as statistically significant.
One shoot venous administration of IG-1 reduced plasma glucose and improved glucose tolerance in C57B6N mice fed normal diets When 1 mg of IG-1 was administered intravenously to the C57B6N mice fed with normal diets, plasma glucose concentrations were significantly reduced from 180 min to 240 min, as compared with DMSO administration (Fig. 3A). On the next day after one-shoot intravenous injection, fasting plasma glucose levels showed no significant difference between the IG-1 and DMSO groups. However, glycogen content in mice liver and skeletal muscles was increased significantly (P,0.05) (Fig. 3B,C).
Daily administration of IG-1 for 7 days did not significantly affect food intake, body weights, fasting plasma glucose, insulin and triglyceride levels in the mice with normal diets (data not shown). However, glucose tolerance tests were performed on C57BL/6 mice fed with normal diets that had fasted overnight and had received peritoneal injection of 2 g glucose/kg body weights. Fasting plasma glucose levels showed no significant difference between the IG-1 and DMSO groups. However, after glucose load, the AUC levels were significantly lower at doses of 5.18 mg/Kg or higher in the mice injected with IG-1 (P,0.05) (Fig. 3D,E). Fasting insulin levels showed no significant difference between the IG-1 and DMSO groups, however, plasma insulin levels were significantly reduced in the mice injected with IG-1 after 30 minutes (data not shown) (P,0.05).
The treatment with IG-1 decreased plasma glucose, improved glucose tolerance, and reduced food intake and body weight in C57B6N mice fed with high fat diets and db/db mice, models of obese diabetes mellitus One shoot venous administration of IG-1 (1 mg) significantly decreased plasma glucose concentrations from 90 min to 300 min in the C57B6N mice fed with high fat diets (Fig. 4A), as compared with DMSO treatment. On the day after one-shoot intravenous injection, fasting plasma glucose levels showed no significant difference between the IG-1 and DMSO groups (data not shown). However, glycogen content in mice liver and skeletal muscles was increased significantly in IG-1 treated mice (P,0.05) (Fig. 4B,C).
When IG-1 (40 mg/g body weight) was administrated to the C57B6N mice fed high fat-diets for 7 days, food intake of the treated mice was significantly lower than those of the control mice (Table 1). In addition, body weights were also significantly decreased in the IG-1-treated mice as compared with the control mice. Fasting plasma glucose, insulin and triglyceride levels were markedly decreased in the IG-1 treated mice than the control mice at the 7 days after IG-1 injection.
After 7 days of IG-1 treatment, glucose tolerance tests were performed on C57BL/6 mice with high fat diets. Fasting plasma glucose levels showed no significant difference between the IG-1 and DMSO groups in mice with high fat-diets. However, after glucose load, the AUC levels were significantly lower at doses of 12.3 mg/Kg or higher in the IG-1-treated mice with high fat diets (Fig. 4D,E) (P,0.05). Fasting insulin levels showed no significant difference between the IG-1 and DMSO groups, however, plasma insulin levels were significantly reduced in the mice injected with IG-1 (41.263.5 mg/Kg) after 30 minutes in those two mice (data not shown) (P,0.05). These findings indicate that IG-1 exerted therapeutic effects on diabetic animal models of diet-induced obesity.
Intravenous administration of IG-1 (1 mg) significantly reduced plasma glucose concentrations in streptozotocin (STZ)-diabetic mice in 180-300 min after administration (Fig. 5A). Intravenous administration of IG-1 significantly increased glycogen contents in mice liver and skeletal muscles of the STZ-diabetic mice (P,0.05) (Fig. 5B,C). When IG-1 was administrated to the STZ-diabetic mice for 7 days, fasting plasma glucose was significantly lower at a dose of 51.563.7 mg/Kg in the IG-1-treated STZ-diabetic mice than in the untreated STZ-diabetic mice (P,0.05) (Fig. 5D). These findings indicate that IG-1 exerted therapeutic effects in an animal model of insulin-deficient type 1 diabetes.
One shoot venous administration of IG-1 (1 mg) significantly decreased plasma glucose concentrations from 90 min to 240 min in the db/db mice, obese diabetic animal models (Fig. 6A), as compared with DMSO treatment. On the day after one-shoot intravenous injection, fasting plasma glucose levels showed no significant difference between the IG-1 and DMSO groups. However, glycogen contents in skeletal muscles were increased significantly (P,0.05) (Fig. 6C).
When IG-1 was administrated to the db/db mice for 7 days, food intake of the treated mice was significantly lower than those of the control mice (data not shown) (P,0.05). In addition, body weights were also significantly decreased in the IG-1-treated mice as compared with the control mice (data not shown) (P,0.05). Fasting plasma glucose, insulin and triglyceride levels were markedly lower in the IG-1-treated mice than in the control mice at the 7 days after IG-1 injection (data not shown) (P,0.05).
After 7 days of IG-1 treatment, glucose tolerance tests were performed on db/db mice. Fasting plasma glucose levels showed no significant difference between the IG-1 and DMSO groups in db/db mice. However, after glucose load, the AUC levels were significantly lower at doses of 10.0 mg/Kg or higher in IG-1treated db/db mice (P,0.05) (Fig. 6D,E). Fasting insulin levels showed no significant difference between the IG-1 and DMSO groups, however, plasma insulin levels at 30 minutes after glucose load were significantly lower in the mice injected with IG-1 at doses of 33.264.4 mg/Kg, (data not shown) (P,0.05).
IG-1 treatment for 28 days reduced food intake and body weight in the ob/ob mice, a genetic obese animal model (Table 2). IG-1 treatment also decreased fasting plasma glucose, insulin, triglyceride and free fatty acids levels in the ob/ob mice (Table 2). In addition IG-1 treatment increased serum adiponectin and decreased serum TNF-a in the ob/ob mice ( Table 2).
Administration of IG-1 for 7 days significantly improved glucose tolerance dose-dependently at doses of in 5.18 mg/Kg or higher in the C57B/6N mice on normal diets (Fig. 3E), 12.3 mg/Kg or higher in the C57B/6N mice on high fat diets (Fig. 4E) and 10.0 mg/Kg or higher in db/db mice (Fig. 6E). Thus, long-term treatment with IG-1 improves glucose tolerance in normal mice and in mice models of type 2 diabetes, obesity and metabolic syndrome. The treatment with IG-1 for 7 or 28 days reduces daily food intakes, body weights, and fasting plasma concentrations of glucose, insulin, triglyceride and free fatty acids, and hepatic triglyceride contents in mice models of type 2 diabetes, obesity and metabolic syndrome (table 1 and 2).

Discussion
Some intracellular insulin actions might be mediated by insulinmimetic species of low molecular weight, such as IPGs, generated from the hydrolysis of GPI. Analysis of isolated insulin-mimetic IPGs from insulin sensitive mammalian tissue or lower eukaryotes revealed two structurally and functionally distinctive classes of IPGs [14]. Type-A IPGs consist of myo-inositol and D-glucosamine, whereas type-P IPGs contains 3-O-methyl-D-chiro-inositol and D-galactosamine [15][16][17]. IPG-As mimic the lipogenic activity of insulin in adipose tissue and inhibit cAMP-dependent protein kinase [14]. IPG-Ps mimic the glycogenic effect of insulin on muscle and liver and stimulate pyruvate dehydrogenase phosphatase [14][15][16]. While the composition of type-A IPGs implies a structural relationship to the GPI glycolipids, in contrast, the origin of IPG-P messengers is unclear. The similarity between IPG-A and GPI glycans is supported by the fact that antibodies raised against GPI-anchors cross-react with IPG preparations [18]. Moreover, insulin-stimulated glycogen synthesis in K562 cells requires the biosynthesis of GPI anchors [19]. GPI-anchored proteins are found in lipid rafts on virtually all eukaryotic cell surfaces [20]. They are particularly abundant on the surface of protozoan parasites. GPIs anchor proteins to the outer leaflet of the plasma membrane or occur in free from. The GPI core structure consists of EtN-P-6-Man-a1,2-Man-a1,6-Man-a1,4-GlcN-a1,6-myo-D-inositol-1-P-lipid that can be decorated by many different species-and tissue-specific modifications [21]. Lipolytically and proteolytically degraded GPI-anchored proteins from parasites [22] and yeast [23] have been shown to regulate glucose homeostasis in the rodent animal model.
A number of active IPG-As carry a 1,2-cyclic phosphate at myoinositol [24][25][26][27][28][29]. It has been suggested that IPG-As are generated by cleavage of GPI anchors through the action of GPI-PLC [30,31]. However, the gene for mammalian GPI-PLC has never been identified. The details of both generation of IPGs and IPG downstream signaling remain unclear. Despite intensive efforts using synthesized IPGs, there is no agreement on the chemical composition of an active IPG-A molecule. The GPI core glycan pseudopentasaccharide without any modifications was shown not to have biological activity [32]. It was recently reported that some synthetic IPGs lack insulin-mimetic activity [33].   So, we speculated that another insulin-mimicking variant of IPGs might be the AIGs that are generated from protein-free GPI anchors through the action of a GPI-PLD and GPI-PLC. AIG have ester-linked fatty acid at the C2 hydroxyl of inositol. Inositol acylation occurs during GPI biosynthesis. The presence of this lipid modification makes the GPI anchor inherently resistant to the action of bacterial PI-PLC [34].
The gene encoding GPI-PLC, which cleaves IPG or AIG from GPI, has still not been identified in mammalian and human genomic surveys. In contrast, accumulating evidence suggests that GPI-PLD (GPLD1) has an important role in insulin signaling and pathogenesis of insulin resistance and diabetes. Gene structure and tissue expression of GPLD1 has been well elucidated [35]. GPI-PLD is abundant in mammalian serum [36]. The source of the circulating enzyme is liver and pancreatic b-cells [37][38][39]. GPI-PLD expressed in pancreatic islet is regulated by glucose and insulin [37]. Glucose stimulates both GPI-PLD and insulin secretion from beta TC3 cells [38]. However, GPI-PLD secretion   Insulin-Mimicking Bioactivities of Acylated Inositol Glycans PLOS ONE | www.plosone.org differs from insulin secretion by a higher rate of basal release and a lower magnitude of response to secretagogues [39]. Serum level of GPI-PLD was increased in human and rodent type 1 diabetes, but it was reduced by treatment with insulin [40,41]. It was reported that insulin induced GPI hydrolysis by GPI-PLD [42]. Deeg et al reported that insulin resistance is associated with increased serum levels of GPI-PLD [43]. GPLD1 was reported to contribute to low fat diet-induced improvement of insulin sensitivity, but not low-carbohydrate diet-induced insulin sensitivity [44]. GPI-PLD is reported to contribute the pathogenesis of insulin-resistant pre-eclamsia through the association with IPG [45]. Grass pea (Lathyrus sativus) seeds have been used in some traditional medicines to ameliorate diabetic symptoms. It was reported that IPG generated from Lathyrus sativus seeds by hydrolysis with GPI-PLD, possesses insulin-mimetic activities [46]. GPI-PLD associated with HDL through the interaction with apoA-1, and apoA-IV [47,48]. GPI-PLD was reported to contribute to triglyceride metabolism [49]. Statin regulated serum level of GPI-PLD [50]. GPI-PLD was reported to contribute to lipid metabolism in liver and pathogenesis of nonalcoholic fatty liver disease [51].
The work presented here aims at clarifying the active structure of GPI precursor-derived insulin-mimetic IPGs using synthetic AIGs. We investigated the insulin-mimetic potential of five synthetic AIGs in primary murine adipocytes, 3T3-L1 adipocytes, as well as type 1 and type 2 diabetic mice. IG-1 has insulinmimicking bioactivities in primary murine adipocytes, 3T3-L1 adipocytes, as well as type 1 and type 2 diabetic mice and genetic obese mice. In contrast, IG-2 only stimulated lipogenesis in primary murine adipocytes, but not in 3T3-L1 adipocytes. IG-1 and IG-2 are expected to be generated from the hydrolysis of a protein-free GPI precursor by GPI-PLD and GPI-PLC, respectively. In contrast, less bioactive IG-15 with a 1-palmitoyl-group and alpha 1,3 linkage might not to be generated from GPI precursor. It has been reported that a minor fatty acid constituent found at the 2-position of myo-inositol in mammalian GPI is myristate [52]. We synthesized IG-21 (D-glucosamine-(a1,6)-(2myristoyl)-D-myo-inositol) and other short-chain fatty acid (C2 to C12)-linked inositol glycans species and checked insulin-mimic bioactivities in 3T3-L1 adipocytes. However, we could not detect insulin-mimic bioactivities of those lipophilic inositol glycans (data not shown), suggesting that palmitoylation at the 2-position of myo-inositol might be critical for insulin-mimic bioactivities of lipophilic inositol glycans. The minimal active AIG structure remains elusive and new synthetic materials will be required for the study of insulin-mimetic AIGs.
Our results strongly support the notion that GPI derived AIGs have insulin-mimetic activity. These exciting results support the hypothesis that these truncated GPI products, IG-1 or IG-2 may be involved in insulin signaling in muscle, fat, and liver, as well as in body glucose and lipid homeostasis. This study confirmed that IG-1 has insulin-mimicking bioactivities and improves glucose tolerance in several mice models of diabetes with or without obesity, such as male C57BL/6 mice fed normal diets (normal model), mice fed high fat diets (model of diabetes with mild obesity), db/db mice (genetic model of diabetes with severe obesity), ob/ob mice (genetic model of severe obesity), STZdiabetic mice (model of lean insulin-deficient diabetes), respectively. Our preliminary study demonstrated that IG-1 inhibited glucose-induced insulin secretion in Min-6 insulinoma cells (data not shown). Inhibition of insulin secretion by IG-1 may be a mechanism for turning off insulin action. Therefore, AIGs would be expected as useful tools in the treatment of diabetes, obesity and metabolic syndrome.
It is very interesting that long-term treatment with IG-1 reduced food intake and body weight, and increased adiponectin concentrations in diabetic obese mice. Anabolic actions of IG-1 might be mediated through the neural network from liver and adipose tissues to brain, resulting in decreased appetite and body adiposity, as well as increased adiponectin concentrations [53]. Another possibility is that IG-1 might move into the brain and regulate appetite as an anabolic signaling factor. IG-1 might affect adipose tissues as an anabolic signaling factor and regulate adiponectin secretion. Further investigation is needed to elucidate a possible role of IG-1 on appetite regulation.
We need further investigation to identify the insulin-sensitive GPI precursor and elucidate the molecular mechanism of GPI-PLD activation by insulin to generate IG-1 and phosphatidic acids. Further investigation is required to identify the down-stream signaling molecules activated by IG-1. We will clarify the extent of interaction between IG-1 and PAs on metabolic signaling of insulin. A clinical study should be conducted to elucidate the role of IG-1 in insulin resistance in type 2 diabetes, obesity and metabolic syndrome.