Prevention and Treatment of Obesity, Insulin Resistance, and Diabetes by Bile Acid–Binding Resin

The NSY mouse, which develops type 2 diabetes in an age-dependent manner with both impaired insulin secretion and insulin resistance (18,19), was used as an animal model of type 2 diabetes. NSY mice were maintained in the animal facilities of Osaka University Medical School (19). The Osaka University Medical School Guidelines for the Care and Use of Laboratory Animals were followed. All mice were maintained in an air-conditioned room (22–25°C) with a 12-h light/dark cycle and were given free access to food and water. Mice without colestimide treatment (control group) were given a high-fat diet, a purified ingredient diet with 45 kcal/g fat primarily from lard (no. D12451; Research Diets, New Brunswick, NJ), containing 23.7% protein, 41.4% carbohydrate, 23.6% fat, and 5.8% fiber, between 4 and 21 weeks of age (n = 7). For prevention or intervention with colestimide, mice were fed a high-fat diet mixed with 1.5% colestimide (a gift from Mitsubishi Pharma, Tokyo, Japan). The caloric density of the high-fat diet was 4.73 kcal/g; that of the colestimide-containing high-fat diet was 4.66 kcal/g. In the prevention study, mice were given a colestimide-containing high-fat diet between 4 and 21 weeks of age (n = 6). In the intervention study, mice were given a high-fat diet between 4 and 14 weeks of age, and then the diet was switched to a colestimide-containing high-fat diet at 14 weeks of age and the mice were treated for 7 weeks until 21 weeks of age (n = 7).

Plasma lipoproteins were analyzed by an on-line dual enzymatic method for simultaneous quantification of cholesterol and triglycerides by high-performance liquid chromatography at Skylight Biotech (Akita, Japan), according to the procedure described by Usui et al. (20). Blood glucose level (in milligrams per deciliter) was measured directly by the glucose oxidase method using Glutest E (Kyoto Daiichi Kagaku, Kyoto, Japan). Glucose tolerance was studied in mice by an oral glucose tolerance test (2 g glucose/kg body wt) after overnight fasting at 12 and 18 weeks of age, and the blood glucose concentrations were measured at fasting (0 min) and 30, 60, 90, and 120 min. The area under the curve of blood glucose was calculated according to the trapezoid rule. To assess the insulin secretion in response to glucose, plasma insulin concentrations were measured at fasting, 15, and 30 min after oral glucose administration (2 g/kg body wt) at 20 weeks of age. Plasma insulin concentration was measured with an insulin enzyme-linked immunosorbent assay kit (Morinaga Seikagaku, Yokohama, Japan). An insulin tolerance test was performed by an intraperitoneal injection of human insulin (0.5 units/kg body wt; Novolin; Novo Nordisk) in overnight-fasted mice at 19 weeks of age. Blood samples were obtained from the tail vein, and blood glucose concentrations were measured at fasting (0 min) and 15, 30, 45, and 60 min after insulin injection. Plasma adiponectin concentrations were measured by enzyme-linked immunosorbent assay using a mouse adiponectin enzyme-linked immunosorbent assay kit (Otsuka Pharmaceutical, Tokyo, Japan) at 21 weeks of age. BMI was calculated as body weight (in grams) divided by the square of body length (in centimeters), and plasma insulin level was measured at 21 weeks of age. All mice were killed under anesthesia by intraperitoneal administration of pentobarbital at 21 weeks of age, and the liver was immediately removed and frozen or fixed with 10% buffered formalin. The pancreas was also removed and fixed in formalin.

Frozen sections of the liver stained with Sudan III were examined using a fluorescence microscope with picture analysis program software, IPLab Spectrum (Scanalytics), and the area stained with Sudan III was quantified in four fields from each sample. Paraffin sections of the pancreas were stained with hematoxylin-eosin by the standard method. Islet mass was quantified in eight islets from each sample by a BZ-Analyzer (Keyence, Osaka, Japan).

Feces were collected for 48 h during the 3–5 days of high-fat or colestimide-containing diet feeding in NSY mice at 20 weeks of age (n = 6 mice per group). Feces were dried and powdered, and fecal bile acids were extracted from 100 mg powdered feces with 90% ethanol according to the method of Tokunaga et al. (21); the extracted solutions were used to enzymatically determine bile acid concentration using a total bile acids kit (Wako, Osaka, Japan). Total lipids were extracted from 500 mg powdered feces with chloroform/methanol (2:1 vol/vol) according to the method of Folch et al. (22). The extracted total fecal lipids were gravimetrically determined.

Total RNA was extracted from frozen liver by the acid guanidinium-phenol-chloroform method using Isogen (Nippon Gene, Tokyo, Japan). cDNA was synthesized using ReverTra Ace α (TOYOBO, Tokyo, Japan). In RT-PCR, preliminary experiments were carried out with various numbers of cycles to determine the nonsaturating conditions of PCR amplification for the following genes. RT-PCR was performed with the following forward and reverse primers: cholesterol 7α-hydroxylase (CYP7A1, Cyp7a1) forward 5′-cactctacaccttgaggatgg-3′, reverse 5′-gacatattgtagctcctgatcc-3′; Farnesoid X receptor (FXR, Nr1h4) forward 5′-cgatcgtcatcctctctcca-3′, reverse 5′-atcagcatctcagcgtggtg-3′; SHP (Nr0b2) forward 5′-gcacctgcatctcacagcca-3′, reverse 5′-agggttgtggccggtctgat-3′; sterol regulatory element–binding protein 1c (SREBP-1c, Srebf1) forward 5′-gctgttggcatcctgctatc-3′, reverse 5′-tagctggaagtgacggtggt-3′; and hypoxanthine guanine phosphoribosyl transferase 1 (HPRT, Hprt1) forward 5′-ctcgaagtgttggatacagg-3′, reverse 5′-tggcctataggctcatagtg-3′. The mRNA levels were measured by the methods of nonradioactive RT-PCR and charged-coupled device imaging. CYP7A1-, FXR-, SHP-, and SREBP-1c–measured PCR products were normalized to the amount of cDNA of HPRT in each sample. Fatty acid synthase (FAS, Fas), carnitine palmitoyltransferase 1 (CPT-1, Cpt1a), and PEPCK (Pck1) mRNA were quantified by real-time PCR using a TaqMan probe on an ABI9700 (Applied Biosystems, Tokyo, Japan) and normalized to 18S rRNA (endogenous control). The primers and probes for FAS, CPT-1, PEPCK, and 18S rRNA were purchased from ABI.

All results are expressed as means ± SE. Phenotypic data were statistically analyzed by either Student’s t test or one-way ANOVA with post hoc analysis using Scheffe’s F test. Differences with P < 0.05 were regarded as significant. All statistical analyses were performed using StatView version 5.0 software (SAS Institute, Cary, NC).

To examine whether or not a bile acid–binding resin can prevent the development of diet-induced obesity and diabetes, NSY mice, an animal model of type 2 diabetes, were given a high-fat diet with (prevention group) and without (control group) colestimide starting at 4 weeks of age. Longitudinal analysis of body weight in these mice showed significantly lower body weight in mice with colestimide than in those without at all time points during the treatment (Fig. 1A). BMI was significantly lower in the prevention group than in the control group (Table 1). Despite the reduction in body weight gain, however, food intake in mice with colestimide was not lower but significantly higher than in those without (0.547 ± 0.024 vs. 0.436 ± 0.015 kcal · g body wt⁻¹ · day⁻¹; P < 0.05), indicating that prevention of obesity by colestimide was not due to a decrease in food intake.

To evaluate the effect of colestimide on glucose metabolism, we measured blood glucose levels. A significant reduction in nonfasting blood glucose was observed in mice with colestimide compared with those without (Figs. 1B and Table 1). Fasting blood glucose was also significantly lower in the prevention group than in the control group after 8 weeks of colestimide treatment (Fig. 1B).

To examine whether a bile acid–binding resin can improve diet-induced obesity and diabetes even after the development of these conditions, colestimide treatment was started at 14 weeks of age in mice fed a high-fat diet from 4 weeks of age (intervention group). Baseline body weight in the intervention group before treatment at 14 weeks of age was similar to that in the control group, both of which were significantly greater than that in the prevention group (Fig. 1A). Longitudinal analysis of body weight showed a reduction in body weight gain in the intervention group after colestimide treatment, and the body weight in the intervention group became closer to that in the prevention group, with no significant difference between the two groups at all time points after 3 weeks of intervention (17–21 weeks of age) (Fig. 1A). BMI in the intervention group also showed an intermediate value between those in the control and prevention groups at 21 weeks of age (Table 1). Despite the reduction in body weight gain, food intake in the intervention group was not significantly different from that in the control group (0.323 ± 0.007 vs. 0.334 ± 0.007 kcal · g body wt⁻¹ · day⁻¹), suggesting that the reduction in body weight with colestimide treatment was not due to reduced food intake. The colestimide-containing diet was well tolerated, and no apparent adverse effect was observed. The only apparent phenotypic changes observed were prevention of obesity, increase in food intake, and increase in the amount of feces.

In the prevention study, serum concentrations of total, VLDL, and LDL cholesterol were significantly lower in mice with colestimide than in those without at 15 weeks of age (Fig. 2). In the intervention study, VLDL cholesterol showed a significant reduction, and total and LDL-cholesterol showed tendencies toward a reduction in mice with colestimide compared with those without after 5 weeks of treatment (Fig. 2A). HDL cholesterol was not different among the groups. Serum concentrations of triglyceride, VLDL triglyceride, and LDL triglyceride in the prevention group were significantly lower than those in the control group (Fig. 2B). The intervention group showed tendencies toward a reduction in total triglyceride and VLDL triglyceride compared with the control group (Fig. 2B).

Liver weight was significantly reduced by colestimide treatment in both the prevention and intervention groups (Table 1). Histological analysis of the liver showed a significant reduction in the Sudan III–stained area in the prevention (69% reduction) and intervention (41% reduction) groups compared with the control group, indicating marked improvement of fatty liver by colestimide treatment (Table 1 and online appendix Fig. 1 [available at http://diabetes.diabetesjournals.org]).

Subcutaneous fat weight was significantly reduced in the prevention group, and the intervention group showed a tendency toward a reduction. Visceral fat weight (retroperitoneal, mesenteric, and epididymal fat) and weight of BAT were not significantly different among the three groups (Table 1). On histological examination of adipocytes from visceral fat and BAT, cell size and accumulation of lipid droplets were not significantly different among the groups (online appendix Fig. 3).

At 20 weeks age, control mice showed marked hyperglycemia, but this was markedly and significantly attenuated by treatment with colestimide in both the prevention and intervention groups (Table 1). Fasting serum insulin level was markedly higher in the control group, suggesting insulin resistance in this group (Fig. 3A). Administration of colestimide significantly attenuated hyperinsulinemia in both the prevention and intervention groups (Fig. 3A). To further confirm the effect of colestimide on insulin resistance, an insulin tolerance test was performed. The glucose-lowering effect of insulin was markedly impaired in the control group, which confirmed marked insulin resistance in this group (Fig. 3B). The glucose-lowering effect of insulin was markedly ameliorated by colestimide treatment in both the prevention and intervention groups (Fig. 3B), indicating that insulin sensitivity was improved by colestimide. To investigate the possible contribution of adiponectin to the improvement in insulin sensitivity, plasma adiponectin concentration was measured. Plasma adiponectin concentration was not significantly different among the three groups (Table 1).

To further clarify the effect of colestimide on glucose metabolism and insulin secretion, an oral glucose tolerance test was performed. Blood glucose level at 120 min after oral administration of glucose was significantly lower in the prevention group than in the control group (Fig. 3C); that in the intervention group showed a tendency toward a reduction and was intermediate between that in the prevention and control groups. The insulin response to glucose was completely abolished in the control group, indicating a markedly impaired insulin response to glucose in this group (Fig. 3D). The insulin response to glucose was improved by colestimide treatment in both the prevention and intervention groups (Fig. 3D). Taken together, these data indicate that both insulin resistance and impaired insulin secretion were improved by colestimide treatment (Table 1 and Fig. 3). Histological examination of pancreatic islets showed that the enlarged islet area in the control group was significantly reduced by colestimide treatment in the prevention group (39% reduction) (Table 1). Islet area in the intervention group was not significantly different from that in the control group.

The reduction in body weight gain without a reduction in food intake in mice treated with colestimide suggests the possibility that colestimide may have reduced energy absorption, in particular, absorption of lipid. To examine this possibility, we measured fecal excretion of bile acids and lipids. Fecal excretion of bile acids was increased ∼3.5-fold by colestimide treatment, indicating that colestimide was successfully ingested by the mice and exerted its biological effect as a bile acid–binding resin (Fig. 4A). Fecal excretion of lipids was increased about sevenfold in mice with colestimide (Fig. 4B), indicating that colestimide markedly decreased the absorption of lipids from the intestine.

To clarify the molecular mechanisms underlying the metabolic changes by colestimide treatment, the expression profiles of genes related to lipid and glucose metabolism in the liver were examined. As expected, the mRNA level of CYP7A1, the rate-limiting enzyme of the cholesterol catabolic pathway, was upregulated about fourfold in mice with colestimide compared with those without (Fig. 5). The mRNA level of SHP, a repressor for the transcription of CYP7A1, was significantly decreased in mice with colestimide. The mRNA level of FXR, a nuclear receptor for bile acids and a regulator of SHP, was not significantly different among the three groups. The mRNA level of SREBP-1c, a major regulator of lipogenic genes, was decreased in mice with colestimide compared with those without. The changes in SREBP-1c expression were parallel to those in SHP expression.

To clarify the mechanism of colestimide in ameliorating fatty liver and the serum lipid profile, the expression levels of genes involved in biosynthesis and β-oxidation of fatty acids were examined. The mRNA level of FAS, a key enzyme of fatty acid synthesis and also known as a target for SREBP-1c, was significantly decreased in the prevention group and tended to decrease in the intervention group (Fig. 5). The expression of CPT-1, a key enzyme of β-oxidation of fatty acids, was not significantly different among the three groups. The expression level of PEPCK, a key enzyme of gluconeogenesis, was significantly lower in the prevention group than in the control group (Fig. 5). The expression level of PEPCK in the intervention group showed a tendency toward a reduction and an intermediate value between that in the prevention and control groups.

In this study, we examined the effect of a bile acid–binding resin (colestimide, which is clinically used as a cholesterol-lowering drug) on obesity and diabetes in a mouse model of type 2 diabetes. As observed in humans, colestimide treatment markedly improved hyperglycemia and dyslipidemia in mice. Moreover, diet-induced obesity and fatty liver were markedly ameliorated by colestimide. To our surprise, however, these phenotypic changes were observed without a change (intervention group) or even with an increase (prevention group) in food intake, indicating that the prevention of obesity and fatty liver was not due to a decrease in food intake. To clarify the mechanism of prevention and intervention in obesity without a change in food intake, we measured fecal excretion of lipids and found a significant increase, indicating that a reduction in lipid absorption from the intestine by colestimide is one of the mechanisms for its improvement of obesity. The improvement in obesity and fatty liver was associated with improvement in insulin resistance and impaired insulin secretion, leading to prevention and intervention in hyperglycemia. The beneficial effect of colestimide on metabolic phenotypes was observed even when treatment was started after the development of obesity and diabetes, although the effect was less marked than in the prevention group. These data, together with the data in humans showing that bile acid–binding resins improved glycemic control in patients with type 2 diabetes and dyslipidemia (16,17), indicate a close link of bile acid metabolism with glucose and lipid metabolism and suggest a novel therapeutic approach for the treatment of obesity and diabetes.

Treatment with colestimide reduced lipid accumulation in the liver and subcutaneous fat but not in visceral fat. A high-fat diet significantly increased subcutaneous fat (2.52 ± 0.24 vs. 4.24 ± 0.23 g for standard diet vs. high-fat diet, respectively) but not visceral fat (3.06 ± 0.19 vs. 2.82 ± 0.07 g, respectively), indicating that colestimide administration reduced the fat accumulation in the organs affected by a high-fat diet, i.e., in the liver and subcutaneous fat. These data, together with no significant changes in plasma adiponectin level by the treatment, suggest that factors other than visceral fat mass played a role in the improvement of obesity and insulin resistance by colestimide, at least in the mouse model used in the present study. The marked improvement of fatty liver observed in this study may be responsible for this. Accumulating lines of evidence in both animal models (23,24) and humans (25,26) suggest the close association of fatty liver with insulin resistance and diabetes. The reduction of lipid accumulation in the liver was reported to reverse hepatic insulin resistance and normalize gluconeogenesis in type 2 diabetic patients without changing peripheral glucose metabolism (25).

Gene expression analysis of the liver indicated reduced expression of genes for both fatty acid synthesis and gluconeogenesis by treatment with colestimide, suggesting that this bile acid–binding resin affected not only cholesterol metabolism, but also glucose and fatty acid metabolism in the liver. Treatment with colestimide decreased the expression of SREBP-1c, a major regulator of lipogenic genes, and its downstream target FAS, a key enzyme of fatty acid synthesis, but did not affect CPT-1, a key enzyme in β-oxidation of fatty acids. The expression level of PEPCK, a key enzyme of gluconeogenesis, was also suppressed by colestimide, which was associated with the amelioration of hyperglycemia. The expression of SHP, an upstream regulator of SREBP-1c, PEPCK, and CYP7A1, was decreased, suggesting that changes in lipid and glucose metabolism (in addition to cholesterol metabolism) by the bile acid–binding resin were mediated by decreased expression of SHP. This is further supported by the fact that constitutive expression of SHP in the liver was reported to change the expression of the genes described above in the opposite direction to that observed in the present study, resulting in fatty liver (27). Very recently, targeted disruption of SHP was reported to decrease lipid contents in the liver and increase insulin sensitivity in the liver, which was associated with decreased expression of PEPCK (28). The data in the present study, together with the data described above, provide a molecular basis for a link between bile acid and glucose metabolism and suggest the bile acid metabolism pathway as a novel therapeutic target for the treatment of obesity, insulin resistance, and type 2 diabetes.

The changes in phenotypes and gene expression profiles by treatment with a bile acid–binding resin in the present study are apparently different from those expected from the data reported for mice treated with a bile acid, cholic acid. Feeding mice with cholic acid was reported to inhibit high-fat diet–induced obesity, hyperglycemia, and hypertriglyceridemia (29,30). Treatment with bile acid–binding resins, which decrease the reabsorption of bile acids, was therefore expected to show phenotypic changes opposite to those reported for cholic acid feeding. The phenotypic changes by colestimide administration in the present study, however, were similar to those reported for mice treated with cholic acid, in that obesity, fatty liver, hyperglycemia, and hypertriglyceridemia were ameliorated rather than worsened (Fig. 2B). This discrepancy may be explained by the net effect in the expression levels and activity of FXR, SHP, and liver X receptor (LXR), which resulted from the bile acid–binding property and cholesterol-lowering effect of colestimide. Colestimide inhibits the absorption of most bile acids, such as chenodeoxycholic acid and lithocholic acid, but not cholic acid (14). Administration of bile acid–binding resins was reported to increase, but not decrease, the relative proportion and absolute value of cholic acid without changing the total bile acid pool size (31). Administration of cholic acid, in contrast, was reported to increase not only cholic acid, but also other components of bile acids (30). Thus, bile acid–binding resins and cholic acid are similar in increasing absolute value of cholic acid but different in their effect on other bile acids. FXR is reported to be activated by taurocholic acid, deoxycholic acid, and taurodeoxy cholic acid but not by cholic acid (32). These differences may explain the reason why the changes in the expression levels and activity of FXR and SHP are different between the treatment with colestimide and cholic acid. The decrease in the expression of SREBP-1c and FAS was observed both in the present study and in a study with cholic acid administration (33), despite the difference in the expression levels of the upstream regulator, SHP. Colestimide treatment decreased the expression of SHP (Fig. 5), while cholic acid administration increased it (33). This may be explained by the net effect of SHP and LXR because the expression of SREBP-1c is critically regulated by SHP (negative regulator) and LXR (positive regulator), as is shown by a study with LXR knockout mice (33) and the promoter assay of the SREBP-1c gene (34–36).

Very recently, Wang et al. (15) reported that mice with targeted disruption of SHP were resistant to diet-induced obesity and showed that SHP has a novel function as a negative regulator of energy production in BAT. In addition, Watanabe et al. (30) reported that administration of cholic acid to mice induced energy expenditure in BAT, preventing obesity and insulin resistance. Unlike SHP-null mice or cholic acid administration, no significant changes were observed in body temperature under ad libitum conditions (online appendix Fig. 2A) in uncoupling protein-1 expression in BAT (online appendix Fig. 2B) or in histological appearance of BAT (online appendix Fig. 3) by colestimide treatment. A significant increase, however, in the expression of peroxisome proliferator–activated γ coactivator-1α (online appendix Fig. 2C), a dominant regulator of energy metabolism, and a tendency toward an increase in the expression of type 2 iodothyronine deiodinase (DIO2) (online appendix Fig. 2C), which mediates local activation of thyroid hormone, were observed in BAT, as in the case of SHP-null mice (13). Significant change in body temperature in SHP-null mice was observed under fasting conditions, but not in ad libitum conditions (15), suggesting that changes in energy metabolism, which cannot be detected by changes in the body temperature measured under ad libitum conditions, may have also contributed to the prevention of diet-induced obesity in the present study. A reduction in lipid absorption from the intestine by colestimide, as well as the mechanism discussed above, may have contributed to the repression of body weight gain.

Despite the beneficial effect of bile acids (such as cholic acid and chenodeoxycholic acid) on glucose, lipid, and energy metabolism (29,30,33), the use of bile acids in humans is limited because of toxic side effects, especially in the liver, and an increase in LDL cholesterol by inhibiting LDL receptor activity (37). The beneficial effect of bile acid–binding resins, as observed in the present study, together with their well-established safety and efficacy in the treatment of hypercholesterolemia in humans, indicate the potential use of this kind of drug for the treatment of not only hypercholesterolemia, but also obesity, insulin resistance, and diabetes. Randomized controlled clinical trials are necessary to clarify the efficacy and safety of this kind of drug on glucose, lipid, and energy metabolism in humans.

In conclusion, the present study showed that colestimide, a bile acid–binding resin, improved not only dyslipidemia, but also obesity, insulin resistance, and type 2 diabetes by suppressing SHP, a pleiotropic regulator of metabolic pathways. These data suggest a close link between bile acid metabolism and glucose metabolism, possibly via SHP, and point to a novel therapeutic target for the treatment of obesity, insulin resistance, and type 2 diabetes. Further studies on insulin signaling and on loss or gain of function of SHP in insulin-sensitive tissues, as well as studies in other animal models, would provide additional insights into the mechanisms by which colestimide improves insulin resistance and diabetes. Given the well-established efficacy and safety of bile acid–binding resins in the treatment of hypercholesterolemia in humans, clinical studies on the efficacy and safety of this kind of drug in the treatment of obesity, insulin resistance, and diabetes are warranted.

This study was supported by a Grant-in-Aid for Scientific Research on Priority Areas, a Grant-in-Aid for Scientific Research, and a Grant-in-Aid for Exploratory Research from the Ministry of Education, Culture, Sports, Science and Technology, Japan.

M.K. is a Research Fellow of the Japan Society for the Promotion of Science.

We thank Y. Tsukamoto and M. Moritani for their skillful technical assistance.