LECT2 Functions as a Hepatokine That Links Obesity to Skeletal Muscle Insulin Resistance

  1. Toshinari Takamura1
  1. 1Department of Disease Control and Homeostasis, Kanazawa University Graduate School of Medical Sciences, Kanazawa, Ishikawa, Japan
  2. 2Division of Natural System, Graduate School of Natural Science and Technology, Kanazawa University, Kanazawa, Ishikawa, Japan
  3. 3Public Central Hospital of Matto Ishikawa, Hakusan, Ishikawa, Japan
  4. 4Department of Hospital Pharmacy, Kanazawa University Graduate School of Medical Sciences, Kanazawa, Ishikawa, Japan
  5. 5Department of Medicinal Informatics, Kanazawa University Graduate School of Medical Sciences, Kanazawa, Ishikawa, Japan
  6. 6Merck & Co. Inc., Rahway, NJ
  7. 7Division of Sports Medicine, Graduate School of Comprehensive Human Sciences, University of Tsukuba, Ibaraki, Japan
  8. 8Division of Material Engineering, Graduate School of Natural Science and Technology, Kanazawa University, Kanazawa, Japan
  9. 9Institute of Science and Engineering, Faculty of Natural System, Kanazawa University, Kanazawa, Japan
  10. 10Center of Infectious Diseases, West China Hospital of Sichuan University, Chengdu, China
  11. 11Department of Medical Life Systems, Faculty of Medical and Life Sciences, Doshisha University, Kyotanabe, Kyoto, Japan
  12. 12Department of Bioactive Molecules, National Institute of Infectious Diseases, Shinjuku-ku, Tokyo, Japan
  1. Corresponding author: Toshinari Takamura, ttakamura{at}m-kanazawa.jp.
  1. F.L. and H.M. contributed equally to this study.

Abstract

Recent articles have reported an association between fatty liver disease and systemic insulin resistance in humans, but the causal relationship remains unclear. The liver may contribute to muscle insulin resistance by releasing secretory proteins called hepatokines. Here we demonstrate that leukocyte cell–derived chemotaxin 2 (LECT2), an energy-sensing hepatokine, is a link between obesity and skeletal muscle insulin resistance. Circulating LECT2 positively correlated with the severity of both obesity and insulin resistance in humans. LECT2 expression was negatively regulated by starvation-sensing kinase adenosine monophosphate-activated protein kinase in H4IIEC hepatocytes. Genetic deletion of LECT2 in mice increased insulin sensitivity in the skeletal muscle. Treatment with recombinant LECT2 protein impaired insulin signaling via phosphorylation of Jun NH2-terminal kinase in C2C12 myocytes. These results demonstrate the involvement of LECT2 in glucose metabolism and suggest that LECT2 may be a therapeutic target for obesity-associated insulin resistance.

Introduction

Insulin resistance is a characteristic feature of people with type 2 diabetes (1) and plays a major role in the development of various diseases such as cardiovascular diseases (2) and nonalcoholic steatohepatitis (3,4). In an insulin-resistant state, impaired insulin action promotes hepatic glucose production and reduces glucose uptake by peripheral tissues. Insulin resistance is commonly observed in obese and overweight people, suggesting a potential role of ectopic fat accumulation in insulin-target tissues in mediating insulin resistance (5). However, the molecular mechanisms underlying insulin resistance are now known to be influenced by the abnormal secretion of tissue-derived factors such as adipokines (69), myokines (10,11), and hepatokines (1214), which traditionally have been considered separate from the endocrine system.

Leukocyte cell–derived chemotaxin 2 (LECT2) is a secretory protein originally identified in the process of screening for a novel neutrophil chemotactic protein (15). LECT2 (encoded by the Lect2 gene in humans) is expressed preferentially by human adult and fetal liver cells and is secreted into the bloodstream (16). The early study using Lect2-deficient mice showed that LECT2 negatively regulates the homeostasis of natural killer T cells in the liver (17). Anson et al. (18) more recently reported that LECT2 exerts anti-inflammatory and tumor-suppressive actions in β-catenin–induced liver tumorigenesis. To date, however, the role of LECT2 in the development of obesity and insulin resistance induced by overnutrition has not yet been established.

We previously demonstrated that overproduction of the liver-derived secretory protein selenoprotein P (SeP) contributes to hyperglycemia in type 2 diabetes by inducing insulin resistance in the liver and skeletal muscle (12). SeP has emerged from comprehensive liver screenings for secretory proteins whose expression levels are correlated with the severity of insulin resistance in patients with type 2 diabetes (12,19,20). Based on these findings, we have proposed that, analogous to adipose tissue, the liver may participate in the pathology of type 2 diabetes and insulin resistance through the production of secretory proteins called hepatokines (12). Other liver-secreted proteins such as fetuin-A (21), angiopoietin-related protein 6 (22), fibroblast growth factor 21 (23), insulin-like growth factors (24), and sex hormone-binding globulin (25) have recently been reported as hepatokines that are involved in glucose metabolism and insulin sensitivity. However, the identification of hepatokines involved in fat accumulation was not adequate. In this study, we identified LECT2 as a hepatokine whose expression levels were positively correlated with the severity of obesity in humans. Levels of LECT2 in blood also were elevated in animal models with obesity. Lect2-deficient mice showed an increase of insulin signaling in skeletal muscle. Conversely, treatment with recombinant LECT2 protein impaired insulin signaling in C2C12 myotubes. Our data demonstrate that LECT2 functions as a hepatokine that links obesity to insulin resistance in skeletal muscle.

Research Design and Methods

Human Clinical Studies

Liver samples to be analyzed by serial analysis of gene expression were obtained from five patients with type 2 diabetes and five nondiabetic subjects who underwent surgical procedures for malignant tumors, including gastric cancer, gall bladder cancer, and colon cancer. Liver samples to be subjected to DNA chip analysis were obtained from 22 patients with type 2 diabetes and 11 subjects with normal glucose tolerance using ultrasonography-guided biopsy needles. Detailed clinical information about these subjects is presented elsewhere (12,19).

Serum samples were obtained from 200 participants who went to the Public Central Hospital of Matto, Ishikawa, Japan, for a complete physical examination. Following an overnight fast, venous blood samples were taken from each patient. Serum levels of LECT2 were measured by an Ab-Match ASSEMBLY Human LECT2 kit (MBL Co.) (26,27).

The homeostasis model assessment of insulin resistance (HOMA-IR) was calculated using the following formula: HOMA-IR = [fasting insulin (μU/mL) × fasting plasma glucose (mmol/L)]/22.5 (28). All patients provided written informed consent for participation in this study. All experimental protocols were approved by the relevant ethics committees at our institution and Matto Ishikawa Central Hospital and were conducted in accordance with the Declaration of Helsinki.

Animals

Eight-week-old C57BL/6J mice were obtained from Sankyo Laboratory Service (Tokyo, Japan). All animals were housed in a 12-h light/12-h dark cycle and allowed free access to food and water. A 60% high-fat diet (HFD; D12492) was purchased from Research Diets (New Brunswick, NJ).

Purification of LECT2

Murine LECT2 was expressed and purified as previously described (29), with minor modifications. Briefly, LECT2 was stably expressed in CHO cells. The protein was purified from the cultured medium by ion exchange chromatography. The fractions containing LECT2 were subsequently applied to a mono S column (GE Healthcare) equilibrated with 50 mmol/L sodium phosphate buffer (pH 7.5) and eluted with a linear gradient of 150–350 mmol/L sodium chloride (NaCl).

Lect2 Knockout Mice

Lect2 knockout mice were produced by homologous recombination using genomic DNA cloned from an Sv-129 P1 library, as described previously (17). All experimental mice were generated from intercross between heterozygous mice, and littermates were divided into groups. Because female Lect2 knockout mice had inconsistent phenotypes, only male mice were used in all experiments except those of starvation.

Materials

H4IIEC and C2C12 cells were purchased from the American Type Culture Collection (Manassas, VA). Human recombinant insulin was purchased from Sigma Aldrich (St. Louis, MO). Rabbit antiphospho-Akt (Ser473) monoclonal antibody, rabbit anti–total Akt polyclonal antibody, rabbit antiphospho-AMP-activated protein kinase (AMPK) (Thr172) monoclonal antibody, rabbit anti-AMPKα antibody, rabbit antiphospho-Jun NH2-terminal kinase (JNK) (Thr183/Try185), rabbit anti-JNK, rabbit anti–binding immunoglobulin protein antibody, rabbit antiphospho-eIF2α (Ser51) antibody, rabbit anti–nuclear factor-κB p65 antibody, rabbit antiphospho-IκB kinase-αβ (Ser176/180) antibody, rabbit anti-IκB kinase-α antibody, and rabbit antiphospho-IκBα(Ser32) antibody were purchased from Cell Signaling Technology (Danvers, MA). Rabbit antileukocyte cell–derived chemotaxin 2 polyclonal antibody (sc-99036) and rabbit anti–glyceraldehyde-3-phosphate dehydrogenase polyclonal antibody were purchased from Santa Cruz Biotechnology (Santa Cruz, CA).

Transient Transfection Experiment

C2C12 myoblasts were grown in 12-well multiplates. When 30–50% confluence was reached, cells were transfected with the Fugene 6 transfection reagent (Roche) with 1 μg of control or with mouse Lect2 expression plasmid DNA per well. After 24 h of transfection, the medium was replaced with Dulbecco’s modified Eagle’s medium (DMEM) containing 10% FBS. When the cells reached to 100% confluence 24 h later, the cells were differentiated into myotubes with DMEM containing 2% horse serum for 24–48 h. Then the cells were stimulated with 100 ng/mL human recombinant insulin for 15 min.

Small Interfering RNA Transfection in C2C12 Myoblasts

C2C12 myoblasts were transiently transfected with a total of 15 nmol/L of small interfering RNA (siRNA) duplex oligonucleotides using Lipofectamine RNAiMAX (Invitrogen), using the reverse-transfection method according to the manufacturer’s instructions. A JNK1-specific siRNA with the following sequence was synthesized by Thermo Scientific: 5′-GGAAAGAACUGAUAUACAA-3′ (sense). A JNK2-specific siRNA with the following sequence was synthesized by Thermo Scientific: 5′-GGAAAGAGCUAAUUUACAA-3′ (sense). Negative control siRNA was purchased from Thermo Scientific. Two days after transfection, cells were pretreated with LECT2 protein then stimulated with 100 ng/mL of human recombinant insulin for 15 min.

RNA Isolation, cDNA Synthesis, and Real-Time PCR Analysis

Total RNA was isolated from cells using the GenElute Mammalian Total RNA Miniprep Kit (Sigma Aldrich). Total RNA was isolated from mouse skeletal muscle and heart using RNeasy Fibrous Tissue Mini Kit (Qiagen). Total RNA was isolated from white adipose tissue using the RNeasy Lipid Tissue Mini Kit (Qiagen). RNA concentrations were measured by a NanoDropR ND-1000 spectrophotometer (NanoDrop Technology). cDNA was synthesized from 100 ng of total RNA using a high-capacity cDNA archive kit (Applied Biosystems, Foster City, CA). Real-time PCR analysis was performed by using TaqMan gene expression assays (Applied Biosystems). Primer sets and TaqMan probes were proprietary to Applied Biosystems (Assays-on-Demand gene expression products). To control for variation in the amount of DNA available for PCR, target gene expression in each sample was normalized relative to the expression of an endogenous control (18S ribosomal RNA or glyceraldehyde-3-phosphate dehydrogenase) (TaqMan control reagent kit; Applied Biosystems).

Treatment With Recombinant LECT2 Protein in C2C12 Myotubes

C2C12 myoblasts were grown in 24-well multiplates; after 100% confluence was reached, cells were differentiated into myotubes by culturing in DMEM supplemented with 2% horse serum for 42 h. C2C12 myotubes were serum-starved and incubated in DMEM for 6 h and then treated with LECT2 recombinant protein for various durations in the absence of serum. Following treatment with LECT2 recombinant protein, cells were stimulated with 100 ng/mL human recombinant insulin for 15 min.

Western Blot Studies in C2C12 Myotubes

After the inulin stimulation, the cells were washed in ice-cold PBS, frozen in liquid nitrogen, and lysed at 4°C in 1× RIPA lysis buffer (Upstat Biotechnology) containing a Complete Mini EDTA-free cocktail tablet (Roche Diagnostics) and PhosSTOP phosphatase inhibitor cocktail tablets (Roche Diagnostics). Lysates then were centrifuged to remove insoluble material. Samples were sonicated with a BIORUPTOR (Cosmo Bio, Tokyo, Japan). Whole-cell lysates were then separated by 5–20% SDS-PAGE gels and were transferred to polyvinylidene fluoride membranes, using an iBlot gel transfer system (Invitrogen). Membranes were blocked in a buffer containing 50 mmol/L Tris, 150 mmol/L NaCl, 0.1% Tween 20, and 5% nonfat milk (pH 7.5) or 5% PhosphoBLOCKER reagent (Cell Biolabs, Inc.) for 1 h at 24°C. They then were probed with antibodies for 16 h at 4°C. Afterward, membranes were washed in a buffer containing 50 mmol/L Tris, 150 mmol/L NaCl, and 0.1% Tween 20 (pH 7.5) and then incubated with anti-rabbit IgG horseradish peroxidase–linked antibody (Cell Signaling) for 1 h at 24°C. Protein signals were detected using ECL Prime Western blotting detection reagent (GE Healthcare UK Ltd.). Densitometric analysis of blotted membranes was performed using ImageJ software (National Institutes of Health; http://rsbweb.nih.gov/ij/).

Glucose or Insulin Tolerance Tests in Mice

In preparation for glucose tolerance testing, mice were fasted for 12 h. After fasting, glucose was administered intraperitoneally, and blood glucose levels were measured at 0, 30, 60, 90, and 120 min. For insulin tolerance testing, mice were fasted for 4 h. After fasting, insulin was administered intraperitoneally, and blood glucose levels were measured. Blood glucose levels were determined by the glucose-oxidase method using Glucocard (Aventis Pharma, Tokyo, Japan). The measurable levels of blood glucose by Glucocard range from 20 to 600 mg/dL. Because mice fed an HFD are much more resistant to insulin than mice fed a standard diet, lower doses of glucose were injected in mice fed an HFD during glucose tolerance testing, as indicated in the legends of Figs. 3 and 4, to avoid the elevation of blood glucose levels to >600 mg/dL. In addition, more doses of insulin were injected to mice fed an HFD during insulin tolerance testing, as indicated in the legends of Figs. 3 and 4, to sufficiently decrease blood glucose levels.

Western Blot Studies in Mice

After 12 h fasting, mice were anesthetized by intraperitoneal administration of sodium pentobarbital. Then insulin (1 units/kg body weight) or PBS (vehicle) was injected through the vena cava. After 10 min, hind-limb muscle tissue, liver tissue, and epididymal white adipose tissue were removed and immediately frozen in liquid nitrogen. Tissue samples were homogenized using a Polytron homogenizer running at half-maximal speed (15,000 rpm) for 1 min on ice in 1 mL of 1× radioimmunoprecipitation assay lysis buffer (Upstat Biotechnology) containing a Complete Mini EDTA-free cocktail tablet (Roche Diagnostics) and PhosSTOP phosphatase inhibitor cocktail tablets (Roche Diagnostics). Tissue lysates were solubilized by continuous stirring for 1 h at 4°C and centrifuged for 15 min at 14,000 rpm. Protein samples were separated by 5–20% SDS-PAGE gels and were transferred to polyvinylidene fluoride membranes. Serine and tyrosine phosphorylation of specific target proteins was analyzed by Western blotting.

Hyperinsulinemic-Euglycemic Clamp Studies in Mice

Clamp studies were performed as described previously (12,30), with slight modifications. Briefly, 2 days before the study, 13-week-old male C57BL/6J wild-type and Lect2-deficient mice were anesthetized using sodium pentobarbital, and an infusion catheter was inserted into the right jugular vein. Before insulin infusion, mice were fasted for 6 h. Clamp studies were performed on conscious and unrestrained animals. Insulin (Novolin R; Novo Nordisk, Denmark) was continuously infused at a rate of 5.0 mU/kg/min, and the blood glucose concentration (monitored every 5 min) was maintained at 100 mg/dL through the administration of glucose (50%, enriched to approximately 20% with [6,6-2H2]glucose; Sigma) for 120 min. Blood was sampled through tail-tip bleeds at 0, 90, 105, and 120 min for the purpose of determining the rate of glucose disappearance (Rd). Rd values were calculated according to non-steady-state equations, and endogenous glucose production was calculated as the difference between the Rd and the exogenous glucose infusion rates (30).

Exercise Tolerance Test in Mice

Ten-week-old male C57BL/6J wild-type and Lect2-deficient mice were set in a running machine. After 5 min of warming up by running and 5 min of rest, mice started running at 11.2 m/min on a 0% incline. Running speed was increased every 5 min until the mice reached exhaustion, defined as when the mouse stopped running for 10 s on the electric tubes.

Acute Exercise Experiment in Mice

Eight-week-old male C57BL/6J mice were randomly divided into two groups: an exercise group and a rest group. All the mice in each group were warmed up for 10 min at 12.6 m/min on a 5% incline. After fasting for 3 h, blood was sampled through tail-tip bleeds. Mice in the exercise group were set in a running machine and started running at 12.6 m/min on a 5% incline. Mice were allowed to have a 5-min rest for every 30 min running. Meanwhile, the mice in the rest group were continually fasted. After 3 h running or resting, blood was sampled again through tail-tip bleeds. Then the mice were anesthetized and killed to isolate the liver tissue.

Starvation Experiment in Mice

Twenty-week-old female C57BL/6J wild-type and Lect2-deficient mice were starved for a total of 60 h; water was supplied. Body weight was measured and blood was sampled through tail-tip bleeds 12, 24, and 36 h after starvation; 60 h later, mice were intraperitoneally injected with insulin at a concentration of 10 units/kg body weight. Fifteen minutes later, mice were anesthetized and killed to isolate femoral muscle.

Blood Samples Assays in Mice

Serum levels of Lect2 were measured using the Ab-Match ASSEMBLY Mouse LECT2 kit (MBL). Serum levels of insulin were determined using a mouse insulin ELISA kit (Morinaga Institute of Biological Science, Inc., Yokohama, Japan), according to the manufacturers’ instructions.

Adenovirus-Mediated Gene Transfer in H4IIEC Hepatocytes

H4IIEC hepatocytes were grown to 90% confluence in 24-well multiplates. Cells were infected with adenoviruses encoding dominant-negative (DN) α1 and α2 AMPK, constitutively active (CA) AMPK, or LacZ for 4 h (8.9 × 106 plaque-forming units/well) (31). We simultaneously expressed α1 and α2 DN AMPK to maximize the effect on AMPK activity. After removing the adenoviruses, the cells were incubated with DMEM for 24 h. Then RNA was isolated from the cells by using GenElute mammalian total RNA miniprep kit (Sigma Aldrich).

Indirect Calorimetry

Mice were housed in standard metabolic cages for 24 h. We used an indirect calorimetry system (Oxymax Equal Flow System; Columbus Instruments, Columbus, OH), in conjunction with the computer-assisted data acquisition program Chart5.2 (AD Instruments, Sydney, Australia), to measure and record oxygen consumption and carbon dioxide production at 5-min intervals. Heat generation was calculated per weight (kilocalories per kilogram per hour).

Measurement of Hepatic Triglyceride Content in Mice

Frozen liver tissue was homogenized in 2 mL ice-cold isopropanol after weighing. After incubation for 10 min with shaking at room temperature, the samples were centrifuged at 3,000 rpm for 10 min, and 1 mL of supernatant was transferred. Triglyceride content in each sample was measured using the commercial Triglyceride E-test WAKO kit (Wako Pure Chemical Industries, Osaka, Japan). Results were normalized to the weight of each liver sample.

Statistical Analyses

All data were analyzed using the Japanese Windows Edition of SPSS version 21.0. Numeric values are reported as the mean ± SEM. Differences between the two groups were assessed using unpaired two-tailed t tests. Data involving more than two groups were assessed by ANOVA. Glucose and insulin tolerance tests were examined using repeated-measures ANOVA.

Results

Identification of a Hepatic Secretory Protein Involved in Obesity

To identify hepatokines involved in the pathophysiology of obesity, we performed liver biopsies in humans and conducted a comprehensive analysis of gene expression profiles, as we previously described (12,19,32,33). We obtained ultrasonography-guided percutaneous needle liver biopsies from 10 people with type 2 diabetes and 7 normal subjects. We subjected the biopsies to DNA chip analysis to identify genes whose hepatic expression was significantly correlated with BMI (Supplementary Table 1). As a result, we found a positive correlation between hepatic Lect2 mRNA levels and BMI, indicating that elevated hepatic Lect2 mRNA levels are associated with obesity.

Circulating LECT2 Levels Correlate With Adiposity and Insulin Resistance in Humans

To characterize the role of LECT2 in humans, using enzyme-linked immunosorbent assays we measured serum LECT2 levels in participants who visited the hospital for a complete physical examination (26,27) (Supplementary Table 2). We found a significant positive correlation between serum LECT2 levels and BMI and waist circumference (Fig. 1A and B). LECT2 levels also showed a significant positive correlation with the HOMA-IR and a negative correlation with insulin sensitivity indices (Matsuda index) (Fig. 1C and D). In addition, serum levels of LECT2 positively correlated with those of SeP, an hepatokine that has already been reported to induce insulin resistance (12) (Fig. 1E). Moreover, LECT2 showed a correlation with levels of both HbA1c and systolic blood pressure (Fig. 1F and G), both of which were reported to be associated with insulin resistance (34,35). These results indicate that serum levels of LECT2 are positively associated with both adiposity and the severity of insulin resistance in humans.

Figure 1

Circulating LECT2 correlates with adiposity and insulin resistance. Graphs show individual correlations between serum levels of LECT2 and BMI (A), waist circumference (B), HOMA-IR index (C), insulin sensitivity (Matsuda index) (D), selenoprotein P (SeP) (E), HbA1c (F), and systolic blood pressure (G) in humans (n = 200).

AMPK Negatively Regulates Lect2 Expression in Hepatocytes

To confirm the elevation of LECT2 in animal models with obesity, we fed C57BL/6J mice an HFD for 8 weeks (Fig. 2AF). An HFD increased body weight in a time-dependent manner (Fig. 2A) and tended to increase triglyceride content in the liver (Fig. 2B). Hematoxylin and eosin staining showed mild steatosis in the mice fed an HFD (Fig. 2C). The expression of Lect2 was elevated in the livers of the mice fed an HFD, in accordance with steatosis-associated genes such as Fasn and Srebp1c (Fig. 2D). Serum levels of LECT2 showed a sustained increase since a week after beginning the HFD (Fig. 2E). In addition, we confirmed that an HFD, even for only a week, resulted in an increase of serum levels of insulin and a decrease of insulin-stimulated Akt phosphorylation in the skeletal muscle of C57BL/6J mice (Supplementary Fig. 1). Importantly, the livers from mice fed an HFD for 8 weeks showed decreased phosphorylation of AMPK (Fig. 2F), the energy depletion–sensing kinase that phosphorylates a variety of energy-associated enzymes and functions as a metabolic regulator that promotes insulin sensitivity (36). Since eating an HFD for a short period increased LECT2 concentrations, we then examined the effects of feeding on blood LECT2 levels. LECT2 levels were elevated in blood obtained from fed C57BL/6J mice compared with samples from the fasting mice (Fig. 2G). Moreover, AMPK phosphorylation decreased in the livers of the mice that had been fed (Fig. 2H). Since Lect2 expression was inversely correlated with AMPK phosphorylation in the liver, we hypothesized that AMPK negatively regulates LECT2 production in hepatocytes. Exercise is reported to increase phosphorylation and activity of AMPK not only in the skeletal muscle but also in the liver in relation to the intrahepatic elevation of AMP levels (37,38). Thus we examined the actions of aerobic exercise on Lect2 expression in the liver. C57BL/6J mice were loaded onto a running treadmill for a total of 3 h. Exercise decreased levels of Lect2 expression and LECT2 protein in the liver (Fig. 2I and J). Aerobic exercise for 3 h, but not resting, significantly reduced serum levels of LECT2 (Fig. 2K). Percentage changes from baseline showed that the reduction of serum LECT2 in the exercise group was significantly larger than that in the rest group (Fig. 2K). In addition, aerobic exercise increased AMPK phosphorylation in the liver (Fig. 2L). To determine whether AMPK suppresses Lect2 expression, we transfected H4IIEC hepatocytes with adenoviruses encoding either CA or DN AMPK. First, we found that transfection with CA AMPK significantly decreased mRNA levels of Lect2 in H4IIEC hepatocytes, similar to those of G6Pc that encode the key gluconeogenic enzyme glucose-6 phosphatase that is already known to be suppressed by AMPK (39) (Fig. 2M). In contrast, transfection with DN AMPK increased Lect2 expression (Fig. 2N). These results indicate that AMPK negatively regulates LECT2 production in hepatocytes.

Figure 2

AMPK negatively regulates Lect2 expression in the liver. A: Body weight of C57BL/6J mice fed an HFD (n = 15) or regular diet (RD; n = 16). Five-week-old male mice were fed an HFD for 8 weeks. B: Triglyceride contents in the livers of C57BL/6J mice fed an HFD or an RD for 8 weeks (n = 7 or 8). C: Hematoxylin and eosin staining of livers from C57BL/6J mice fed an HFD or an RD for 8 weeks. D: mRNA levels for Lect2, Fasn, and Srebp1c in the livers of C57BL/6J mice fed an HFD or an RD for 8 weeks (n = 7 or 8). E: Changes of blood levels of LECT2 in C57BL/6J mice fed an HFD (n = 8) or an RD (n = 8). Blood samples were obtained after fasting for 12 h. F: Phosphorylation of AMPK in the livers of C57BL/6J mice fed an HFD or an RD after a 12-h fast (n = 4). G: Blood levels of LECT2 from C57BL/6J mice following fasting for 12 h and subsequent feeding for 12 h (n = 7 or 8). H: Phosphorylation of AMPK in the livers of C57BL/6J mice following fasting for 12 h and subsequent feeding for 12 h (n = 3). I: Levels of Lect2 mRNA in the livers of C57BL/6J mice following running exercise for 3 h (n = 7–8). J: Protein levels of LECT2 in the livers of C57BL/6J mice following running exercise for 3 h (n = 7–8). K, left: Serum levels of LECT2 in C57BL/6J mice before and after running exercise for 3 h (n = 8; paired t test). K, right: Percentage changes of serum LECT2 after running exercise for 3 h (unpaired t test). L: Phosphorylation of AMPK in the livers of C57BL/6J mice following running exercise for 3 h (n = 3 or 4). M: Effects of CA AMPK on mRNA levels of Lect2 and G6pc in H4IIEC hepatocytes (n = 4). N: Effects of dominant-negative (DN) AMPK on mRNA levels of Lect2 and G6pc in H4IIEC hepatocytes (n = 4). Data in A and B and D–N represent the means ± SEM. *P < 0.05, **P < 0.01, ***P < 0.001. Fasn, fatty acid synthase; G6pc, glucose-6 phosphatase; Srebp1c, sterol regulatory-element binding protein-1c.

Lect2 Deletion Increases Muscle Insulin Sensitivity in Mice

Next we examined the role of LECT2 in the development of insulin resistance in mice. We found that expression of Lect2 in the liver was overwhelmingly dominant compared with that in other tissues in mice (Fig. 3A). This result suggests that the contribution of the other tissues (except the liver) on the circulating levels of LECT2 is very small or negligible in mice. Hence we used systemic knockout mice of LECT2 in the following experiments, although the animal models of liver-specific downregulation for Lect2 might be more suitable. We confirmed that serum LECT2 was undetectable in Lect2-deficient mice by using ELISA (Fig. 3B). Body weight, food intake, and heat production at rest were unaffected by Lect2 knockout (Fig. 3CE). However, the treadmill running challenge revealed that muscle endurance, as assessed by physical exercise, was significantly higher in Lect2−/− mice (Fig. 3F and G). A glucose or insulin loading test revealed that Lect2−/− mice showed lower blood glucose levels after glucose or insulin injection (Fig. 3H and I). Lect2−/− mice exhibited an increase in insulin-stimulated Akt phosphorylation in skeletal muscle (Fig. 3J and K) but not in the liver or adipose tissue (Supplementary Fig. 2A and B). Furthermore, JNK phosphorylation was unchanged in the liver and adipose tissue of these knockout mice (Supplementary Fig. 2C and D). Consistent with the results of insulin signaling, hyperinsulinemic-euglycemic clamp studies showed that the glucose infusion rate and peripheral glucose disposal were increased, whereas endogenous glucose production was unaffected by Lect2 deletion (Fig. 3L and Supplementary Fig. 3). In addition, expression of the genes involved in mitochondria and myogenesis, such as UCP3, Myh1, Myh2, and Ppard, were upregulated in the muscle of Lect2−/− mice (Fig. 3M). These results indicate that deletion of Lect2 increases insulin sensitivity in skeletal muscle in mice.

Figure 3

Lect2 deletion increases muscle insulin sensitivity in mice. A: Lect2 mRNA levels in various tissues of C57BL/6J mice (n = 4–8). B: Serum levels of LECT2 in Lect2-deficient and wild-type (WT) mice fed an HFD for 10 weeks (n = 9–13). Blood samples were obtained during the fed condition. C: Body weight of Lect2-deficient and wild-type mice fed a regular diet (n = 6–8). D: Food intake of Lect2-deficient and wild-type mice (n = 6–8). E: Heat production of Lect2-deficient and wild-type mice (n = 6–8). F, G: Running endurance was tested in Lect2-deficient and wild-type mice (n = 7 or 8). Running endurance is depicted as distance (F) and time (G). Intraperitoneal glucose (H) and insulin (I) tolerance tests in Lect2-deficient and wild-type mice (n = 7 or 8). Glucose and insulin were administered at doses of 2.0 g/kg body weight and 1.0 units/kg body weight, respectively. J, K: Western blot analysis and quantification of phosphorylated Akt in skeletal muscle of Lect2-deficient and wild-type mice (n = 5). Nineteen-week-old female mice were stimulated with insulin (administered through the vena cava) at doses of 1 unit/kg body weight; 10 min after insulin administration, the hind-limb muscles were removed. L: Glucose infusion rate (GIR), endogenous glucose production (EGP), and rate of glucose disposal (Rd) during hyperinsulinemic-euglycemic clamp in Lect2-deficient and wild-type mice (n = 6 or 7). M: mRNA levels of genes involved in myogenesis and mitochondria in skeletal muscle of Lect2-deficient and wild-type mice (n = 4 or 5). Data in A–I and K–M represent the means ± SEM. *P < 0.05, **P < 0.01, ***P < 0.001 (Lect2-deficient mice vs. wild-type mice).

Lect2 Deletion Attenuates Muscle Insulin Resistance in Dietary Obese Mice

To elucidate further the role of LECT2 in the development of obesity-associated insulin resistance, we fed Lect2-deficient mice an HFD. HFD-induced body weight gain was smaller in Lect2-deficient mice compared with wild-type animals (Fig. 4A). To examine the mechanism by which Lect2-deficient mice were less obese after eating an HFD, we measured food intake and heat production in Lect2-deficient mice fed an HFD for only a week, when the body weight was comparable between the wild-type and knockout mice (Supplementary Fig. 4A). Food intake was unaffected (Supplementary Fig. 4B), but heat production as measured by oxygen consumption was significantly increased in Lect2-deficient mice fed an HFD (Supplementary Fig. 4C) in both light and dark phases (Supplementary Fig. 4D and E). Eleven weeks later after eating an HDF, serum levels of insulin and blood levels of glucose decreased in these knockout mice (Fig. 4B and C). A glucose or insulin loading test revealed that Lect2 knockout mice showed lower blood glucose levels after glucose or insulin injection (Fig. 4D and E). Consistent with the result of the insulin loading test, Western blotting revealed that insulin-stimulated Akt phosphorylation increased in the skeletal muscle of these knockout animals (Fig. 4F and G). In contrast, JNK phosphorylation significantly decreased in the skeletal muscle of Lect2-deficient mice (Fig. 4H and I). Furthermore, we examined muscle insulin signaling in Lect2-deficient mice fed an HFD for only 2 weeks, when the body weight was comparable between the wild-type and knockout mice (Supplementary Fig. 5AC). Insulin-stimulated Akt phosphorylation was significantly increased in the muscle of Lect2-deficient mice under conditions of an HFD for 2 weeks (Supplementary Fig. 5D and E). These results indicate that Lect2 deletion reduces muscle insulin resistance in dietary obese mice.

Figure 4

Lect2 deletion attenuates muscle insulin resistance in diet-induced obesity in mice. A: Changes of body weight of Lect2-deficient and wild-type (WT) mice fed an HFD (n = 9–13). B: Serum insulin levels of Lect2-deficient and wild-type mice fed an HFD for 11 weeks in a fed condition and after 12 h of fasting. (n = 9–13). C: Blood glucose levels of Lect2-deficient and wild-type mice fed an HFD for 11 weeks in a fed condition and after 12 h of fasting (n = 9–13). D: Intraperitoneal glucose tolerance tests in Lect2-deficient and wild-type mice fed an HFD for 9 weeks (n = 9–13). Glucose was administered at doses of 0.5 g/kg body weight. E: Intraperitoneal insulin tolerance tests in Lect2-deficient and wild-type mice fed an HFD for 10 weeks (n = 9–13). Insulin was administered at doses of 1.2 units/kg body weight. F and G: Western blot analysis and quantification of phosphorylated Akt in skeletal muscle of Lect2-deficient and wild-type mice (n = 3 or 4), respectively. H and I: Western blot analysis and quantification of phosphorylated JNK in skeletal muscle of Lect2-deficient and wild-type mice (n = 3 or 4), respectively. Mice were stimulated with insulin (administered through the vena cava) at doses of 1 unit/kg body weight; 2 min after insulin administration, the hind-limb muscles were removed. Data in A–E and G represent the means ± SEM. *P < 0.05, **P < 0.01, ***P < 0.001 (Lect2-deficient mice vs. wild-type mice).

Starvation Abolishes the Insulin-Sensitive Phenotype in Lect2-Deficient Mice

Next, to elucidate the role of LECT2 in a condition of severe undernutrition, we starved Lect2-deficient mice for 60 h. Starvation decreased body weight and blood glucose levels in a time-dependent manner, whereas there was no significance between wild-type and Lect2-deficient mice (Fig. 5A and B). Consistent with the changes of body weight, serum levels of LECT2 in wild-type animals significantly decreased during the period of starvation (Fig. 5C). Before starvation, serum levels of insulin in Lect2 knockout mice were lower compared with wild-type mice (Fig. 5D). However, the starvation reduced insulin levels to the extent to which the difference abolished between the two groups (Fig. 5D). Insulin-stimulated Akt phosphorylation in skeletal muscle also showed no difference between the two groups after 60 h of starvation (Fig. 5E). These results indicate that starvation abolishes the insulin-sensitive phenotype in Lect2-deficient mice.

Figure 5

Starvation abolishes the insulin-sensitive phenotype in Lect2-deficient mice. A: Body weight of 20-week-old female Lect2-deficient and wild-type (WT) mice (n = 5) during starvation. B: Blood glucose levels of Lect2-deficient and wild-type mice (n = 5) during starvation. C: Serum levels of insulin of Lect2-deficient and wild-type mice (n = 4 or 5) in a fed condition or after starvation for 36 h (paired t test). D: Serum levels of LECT2 of wild-type mice (n = 4 or 5) in a fed condition or after starvation for 36 h. E: Western blot analysis of phosphorylated Akt in skeletal muscle of Lect2-deficient and wild-type mice after 60 h of starvation. Mice were stimulated with insulin (administered intraperitoneally); 15 min after insulin administration, mice were anesthetized and hind-limb muscle samples were removed for analysis. Data in A–D represent the means ± SEM. Data in D were assessed by paired t tests. *P < 0.05.

LECT2 Impairs Insulin Signaling by Activating JNK in C2C12 Myotubes

First, to examine the effect of LECT2 on insulin signaling in vitro, we transfected C2C12 myocytes with a plasmid expression vector encoding mouse LECT2. Expression of endogenous Lect2 was negligible in C2C12 myocytes transfected with a negative control vector (Fig. 6A and B). We confirmed that C2C12 myotubes transfected with the Lect2 expression vector expressed Lect2 mRNA and released LECT2 protein into the culture medium (Fig. 6A and B). LECT2 transfection suppressed myotube differentiation in C2C12 cells (Fig. 6C). The cells transfected with the Lect2 vector showed a decrease in insulin-stimulated Akt phosphorylation (Fig. 6D) and an increase in basal JNK phosphorylation (Fig. 6E).

Figure 6

LECT2 impairs insulin signaling by activating JNK in C2C12 myotubes. C2C12 myoblasts in 30–50% confluence were transfected with negative control or mouse (m) Lect2 expression plasmid in AE. When the cells reached 100% confluence, they were differentiated into myotubes with DMEM containing 2% horse serum for 24–48 h. A: Lect2 mRNA levels in C2C12 myotubes transfected with control or Lect2 expression vector (n = 6). mRNA was obtained from the cells differentiated into myotubes for 24 h. B: LECT2 protein levels in culture medium of C2C12 myotubes transfected with control or Lect2 expression vector for 24 or 72 h (n = 3). LECT2 production was measured by ELISA. C: Representative images of C2C12 myotubes transfected with control or Lect2 expression vector. The cells were differentiated into myotubes for 48 h. D: Western blot analysis of phosphorylated Akt in C2C12 myotubes transfected with control or Lect2 expression vector (n = 4). The cells were stimulated by 100 ng/mL of insulin for 15 min. E: Western blot analysis of phosphorylated JNK in C2C12 myotubes transfected with control or Lect2 expression vector (n = 3). F: Western blot analysis of phosphorylated Akt in C2C12 myotubes pretreated with recombinant LECT2 protein (n = 4). The cells were pretreated with LECT2 protein. Three hours later, the cells were stimulated with insulin. G: Effects of recombinant LECT2 protein on JNK phosphorylation in C2C12 myotubes (n = 3). The cells were treated with 400 ng/mL of recombinant LECT2 protein for the indicated times. H: Concentration-dependent effects of recombinant LECT2 protein on JNK phosphorylation in C2C12 myotubes (n = 3). The cells were treated with LECT2 protein for 1 h. I: Effects of JNK-knockdown on LECT2 protein–induced insulin resistance in C2C12 myoblasts (n = 4). Data in A, B, and D–I represent the means ± SEM. *P < 0.05, **P < 0.01, ***P < 0.001 versus cells transfected with control vector or cells treated with vehicle.

To further confirm the acute action of LECT2 on insulin signaling, we treated C2C12 myotubes with recombinant LECT2 protein at nearly physiological concentrations. Treatment with 400 ng/mL of LECT2 protein for 3 h decreased insulin-stimulated Akt phosphorylation (Fig. 6F). In addition, treatment with LECT2 protein for 30–60 min transiently increased JNK phosphorylation in C2C12 myotubes (Fig. 6G). LECT2-induced JNK phosphorylation occurred in a concentration-dependent manner (Fig. 6H). To determine whether the JNK pathway mediates LECT2-induced insulin resistance, we transfected C2C12 myoblasts with siRNAs specific for JNK1 and JNK2. Because knockdown of JNK is known to alter the myotube differentiation in C2C12 myotubes (40), we used undifferentiated C2C12 myoblasts to purely assess the action of LECT2 on insulin signal transduction in the following experiments. Double knockdown of JNK1 and JNK2 rescued the cells from the inhibitory effects of LECT2 on insulin signaling (Fig. 6I). Inflammatory signals and endoplasmic reticulum stress are known to be powerful inducers of JNK (41). However, the markers of neither inflammation nor endoplasmic reticulum stress were changed in C2C12 myotubes overexpressed with Lect2 and in the skeletal muscle of Lect2 knockout mice (Supplementary Fig. 4). These in vitro experiments indicate that, at nearly physiological concentrations, LECT2 impairs insulin signal transduction by activating JNK in C2C12 myocytes.

Discussion

Our research reveals that the overproduction of the hepatokine LECT2 contributes to the development of muscle insulin resistance in obesity (Fig. 7). Recent growing evidence suggests a central role for fatty liver disease in the development of insulin resistance in obesity (4,42). Kotronen et al. (43) have reported that intrahepatocellular rather than intramyocellular fat is associated with hyperinsulinemia independent of obesity in nondiabetic men. Fabbrini et al. (44) have revealed that intrahepatic triglyceride, but not visceral adipose tissue, is a better marker of multiorgan insulin resistance associated with obesity. D’Adamo et al. (45) have shown that obese adolescents with high hepatic fat content show lower whole-body insulin sensitivity independent of visceral fat and intramyocellular lipid content. These articles indicate a strong correlation between fatty liver and muscle insulin resistance in humans, but it was still unknown whether fatty liver disease directly causes muscle insulin resistance in obesity. The liver is a major site for the production of bioactive secretory proteins called hepatokines (12,19). Many lines of evidence have reported that the dysregulation of the production of hepatokines such as SeP or fetuin A is involved in the development of systemic insulin resistance (12,13,46,47). This study demonstrates a previously unrecognized role for LECT2 in glucose metabolism and suggests that LECT2 is a strong candidate to explain a mechanism by which the fatty liver leads to whole-body insulin resistance in obesity.

Figure 7

The hepatokine LECT2 links obesity to insulin resistance in skeletal muscle.

The energy depletion–sensing kinase AMPK functions as a metabolic sensor that promotes insulin sensitivity (36). Exercise is known to increase the phosphorylation and activity of AMPK in skeletal muscle. Early reports have shown that exercise-induced AMPK phosphorylation also is observed in the liver (37,38). On the other hand, an HFD is reported to decrease AMPK phosphorylation in the liver, probably because of excessive accumulation of energy (48,49). Negative regulation of LECT2 by the energy depletion–sensing kinase AMPK supports the concept that LECT2 functions as a hepatokine that senses overnutrition. One limitation of this study is that the molecular mechanism by which AMPK reduces Lect2 expression is still unknown. Additional studies are needed to determine the transcriptional factors that negatively regulate Lect2 expression downstream of AMPK pathway.

JNK is a mitogen-activated protein kinase that is activated by various stimuli, including cytokines, reactive oxygen species, endoplasmic reticulum stress, and metabolic changes (41). JNK plays a major role in the development of insulin resistance induced by an HFD through phosphorylating insulin receptor substrates at specific serine and threonine residues (50,51). Several more recent studies suggest a role for JNK in the development of insulin resistance in skeletal muscle, as well as in the liver or adipose tissue. Ferreira et al. (52) reported an increase of JNK phosphorylation and a decrease of insulin-stimulated Akt phosphorylation in the skeletal muscle of patients with nonalcoholic steatohepatitis. Henstridge et al. (53) showed that muscle-specific overproduction of CA JNK induces muscle insulin resistance in mice. Conversely, Sabio et al. (54) revealed that muscle-specific JNK knockout mice exhibit improved insulin sensitivity in skeletal muscle. Hence the overproduction of LECT2 in the liver may contribute, at least in part, to JNK phosphorylation and the subsequent insulin resistance observed in the skeletal muscle of obese patients. However, the mechanism by which LECT2 increases JNK phosphorylation remains unresolved. Our results suggest that LECT2-induced JNK activation in cultured myocytes is independent of inflammation or endoplasmic reticulum stress (Supplementary Fig. 4). Identification of the LECT2 receptor and characterization of its downstream signaling will provide insights into the involvement of LECT2 in JNK phosphorylation.

We show that overexpression of Lect2 does not alter the inflammatory response in cultured myotubes. Early reports suggest that Lect2 exerts different effects on inflammation depending on various pathological conditions. Inflammation observed in autoimmune disorders such as collagen antibody–induced arthritis or concanavalin A–induced hepatitis is reported to be suppressed by Lect2 (17,55). Lect2 also attenuates β-catenin–induced inflammation associated with hepatocellular carcinoma in mouse models (18). On the other hand, a more recent report showed that Lect2 activates lipopolysaccharide-stimulated macrophages via the CD209a receptor, resulting in an improvement in the prognosis for survival in mice with bacterial sepsis (56). Because we found no expression levels of CD209a in C2C12 myotubes in real-time PCR experiments (data not shown), Lect2-induced insulin resistance in cultured myocytes is likely to be independent of inflammatory response via the CD209a receptor. However, it is unknown whether Lect2 affects macrophages observed in the adipose tissue of obesity. The actions of Lect2 on low-grade inflammation seen in obesity are now under investigation.

Interestingly, although Lect2 knockout mice showed an increase of insulin signaling in the skeletal muscle when fed an HFD or regular chow, this increase was abolished after 60 h of starvation. Serum levels of LECT2 were increased by an HFD (Fig. 2E), whereas they were decreased by starvation in wild-type mice (Fig. 5C). Hence it seems most likely that the difference in serum LECT2 levels between wild-type and knockout mice was enhanced by an HFD, whereas it was reduced by starvation. The abolishment of the insulin-sensitive phenotypes in Lect2 knockout mice after starvation may be explained by reduction of the difference in serum LECT2 levels. These results suggest that Lect2 plays a major role in the regulation of insulin sensitivity in overnutritional conditions, but not in the undernutritional ones.

Our data reveal that a 60% HFD for 1 week resulted in a concurrent decrease of insulin signaling in skeletal muscle and an increase of circulating levels of LECT2 in C57BL/6J mice. A previous clinical report showed that overfeeding and inactivity for only 3 days impaired insulin sensitivity in healthy young men (57). Impairment of insulin sensitivity occurred before changes in body composition such as total fat mass and visceral fat area. However, additional clinical studies are required to determine whether eating an HFD for several days indeed induces simultaneous alterations of circulating LECT2 and muscle insulin sensitivity in humans.

C2C12 myocytes transfected with plasmid encoding LECT2 showed an impairment of myotube differentiation and insulin signal transduction (Fig. 6C and D). The presence of LECT2 protein in the culture medium (Fig. 6B) suggests that LECT2 derived from the pLect2 acted on the cells in an autocrine or paracrine manner. Because the half-life of LECT2 protein was predicted to be short because of the low molecular weight of LECT2 (16 kDa), we initially overexpressed Lect2 in the cultured myocytes to examine the chronic actions of LECT2. In the next experiments, we directly treated well-differentiated C2C12 myotubes with recombinant LECT2 protein for 3 h (Fig. 6F) to exclude the possibility that LECT2-induced suppression of myotube differentiation causes secondary insulin resistance in pLect2 experiments. The results obtained from the experiments using the recombinant LECT2 protein suggest that LECT2 directly induces insulin resistance in C2C12 cells independent of its action on myotube differentiation.

Okumura et al. (55) reported that treatment with LECT2 ameliorated collagen antibody–induced arthritis in mice. This report suggests that LECT2 suppresses the inflammatory response that progresses after autoantibodies develop. Several clinical studies showed that the onset of inflammatory polyarthritis, such as rheumatoid arthritis, is accelerated by obesity (58,59). Because our current data reveal a positive correlation between BMI and serum LECT2 levels in humans, it seems that overproduction of LECT2 fails to exert a sufficient suppressive action on inflammatory polyarthritis in people with obesity. However, it is still unknown whether LECT2 acts on the process of autoantibody production by B lymphocytes in the acquired immune system. Further basic and clinical studies are needed to investigate the relationship between LECT2 and obesity-associated arthritis.

Our current cross-sectional data show that serum levels of LECT2 positively correlate with the severity of insulin resistance in human subjects. However, many lines of evidence demonstrated that various circulating proteins whose expression is altered in obesity, such as adiponectin and resistin, participate in the development of insulin resistance (60). Hence our study does not necessarily place LECT2 as only a single causal factor of insulin resistance. In addition, further prospective studies are needed to confirm the causal relationship between LECT2 and insulin resistance in people with obesity.

In summary, our experiments identified LECT2 as an obesity-upregulated hepatokine that induces insulin resistance in skeletal muscle. Lect2 may be a potential target for the treatment of obesity-associated insulin resistance.

Article Information

Acknowledgments. The authors thank Drs. Kuniaki Arai and Tatsuya Yamashita, Kanazawa University, for liver biopsies, and Drs. Isao Usui and Hajime Ishihara and Prof. Toshiyasu Sasaoka, Toyama University, for supplying their technical expertise on Western blot analyses of phosphoproteins. The authors thank Maki Wakabayashi, Yuriko Furuta, and Yoko Hashimoto, Kanazawa University, for technical assistance. The authors also thank Mutsumi Tanaka of the Alfresa Pharma Corporation for measuring blood levels of LECT2 in humans. The authors thank Fabienne Foufelle of Université Pierre et Marie Curie for providing the adenovirus vector encoding for DN-AMPK, as well as In-kyu Lee of Kyungpook National University for providing the adenovirus vector encoding for CA-AMPK.

Funding. This work was supported by Grants-in-Aid for Scientific Research (C-23591301 to T.T. and C-25461334 to H.M.) from the Ministry of Education, Culture, Sports, Science, and Technology of Japan.

Duality of Interest. This work was supported in part by Takeda Science Foundation (to H.M.). No other potential conflicts of interest relevant to this article were reported.

Author Contributions. F.L. researched the data and wrote the manuscript. H.M. conceived and designed the experiments, researched the data, contributed to the discussion, wrote the manuscript, and reviewed and edited the manuscript. K.C., H.Tak., A.K., K.M., N.T., H.H., N.M.-N., Y.T., H.N., Y.M., T.N., M.N., K.Ta., T.S., K.I., K.To., and Y.S. researched the data. T.O., K.-i.M., S.M., H.Tan., and S.Y. contributed to the discussion. S.K. and T.T. contributed the discussion, wrote the manuscript, and reviewed and edited the manuscript. T.T. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

  • Received May 6, 2013.
  • Accepted January 14, 2014.

Readers may use this article as long as the work is properly cited, the use is educational and not for profit, and the work is not altered. See http://creativecommons.org/licenses/by-nc-nd/3.0/ for details.

References

| Table of Contents

This Article

  1. Diabetes vol. 63 no. 5 1649-1664
  1. Supplementary Data
  2. All Versions of this Article:
    1. db13-0728v1
    2. 63/5/1649 most recent

ORCID