Research ArticleMetabolism

Overproduction of inter-α-trypsin inhibitor heavy chain 1 after loss of Gα13 in liver exacerbates systemic insulin resistance in mice

See allHide authors and affiliations

Science Translational Medicine  09 Oct 2019:
Vol. 11, Issue 513, eaan4735
DOI: 10.1126/scitranslmed.aan4735

Targeting ITIH1 in metabolic disease

The liver releases secretory proteins in response to metabolic stress. Kim et al. report a decrease in Gα13 in the liver of mice and humans with diabetes. Secretome analysis enabled identification of a specific protein (ITIH1) highly secreted by liver in association with insulin resistance and consequent hyperglycemia. Glycosyl modification of ITIH1 facilitated its deposition on the hyaluronan surrounding mouse adipose tissue and skeletal muscle, making a physical barrier between insulin and its receptor. Neutralization of secreted ITIH1 prevented systemic insulin resistance and ameliorated glucose intolerance in mice. This finding may contribute to developing a new strategy to treat metabolic diseases.

Abstract

The impact of liver disease on whole-body glucose homeostasis is largely attributed to dysregulated release of secretory proteins in response to metabolic stress. The molecular cues linking liver to whole-body glucose metabolism remain elusive. We found that expression of G protein α-13 (Gα13) was decreased in the liver of mice and humans with diabetes. Liver-specific deletion of the Gna13 gene in mice resulted in systemic glucose intolerance. Comparative secretome analysis identified inter-α-trypsin inhibitor heavy chain 1 (ITIH1) as a protein secreted by liver that was responsible for systemic insulin resistance in Gna13-deficient mice. Liver expression of ITIH1 positively correlated with surrogate markers for diabetes in patients with impaired glucose tolerance or overt diabetes. Mechanistically, a decrease in hepatic Gα13 caused ITIH1 oversecretion by liver through induction of O-GlcNAc transferase expression, facilitating ITIH1 deposition on the hyaluronan surrounding mouse adipose tissue and skeletal muscle. Neutralization of secreted ITIH1 ameliorated glucose intolerance in obese mice. Our findings demonstrate systemic insulin resistance in mice resulting from liver-secreted ITIH1 downstream of Gα13 and its reversal by ITIH1 neutralization.

INTRODUCTION

The metabolic network among tissues is mediated by hormones, cytokines, or other secretory proteins that have paracrine or endocrine actions on other organs, maintaining systemic nutrient and energy homeostasis. Perturbations in this cross-talk can provoke disorders in glucose metabolism, often accompanied by insulin resistance and diabetes. Insulin resistance is characterized by impaired insulin signaling in multiple metabolic organs including liver, adipose tissue, and skeletal muscle. Most of the nutrients absorbed by the intestine first pass through the liver, where two-thirds of blood glucose is assimilated. Hence, the liver senses and responds quickly to nutritional changes, regulating systemic glucose metabolism. Therefore, metabolic disturbances in the liver (e.g., steatosis) commonly precede the development of obesity, as well as insulin resistance in other organs (1, 2).

Numerous genes encoding extracellular proteins are expressed in the liver. Recent advances in comprehensive gene expression analysis and proteomic technology have contributed to the consideration of liver as an endocrine and secretory organ (3, 4). Abnormalities in liver function may have a detrimental effect on glucose metabolism in other organs, as indicated by changes in the protein secretory profile of steatotic liver in association with glucose intolerance and insulin resistance (5). Moreover, several liver-derived secretory factors affect the metabolism of peripheral organs, supporting the hypothesis that proteins secreted by liver control whole-body energy metabolism (57). However, the effects of metabolic challenges on liver-secretory proteins and their modes of action remain elusive.

G protein α subunits share the common biological feature of being activated in response to environmental changes sensed by G protein–coupled receptors (GPCRs) (8). Given that myriad GPCRs directly bind to a relatively small number of Gα proteins for signal transmission, Gα proteins regulate the complexity of diverging and converging signal transducing systems. Thus, Gα protein expression may have a profound effect on modulating physiological and biochemical activities. Among the major Gα protein family members, Gα13 is more highly expressed in liver than in other insulin target tissues (9). However, the biological function of Gα13 in the liver and its effects on whole-body energy metabolism have not been explored.

This study investigated the underlying basis of aberrant expression of liver-secreted proteins in mice under metabolic stress. The goal was to identify a mediator affecting glucose metabolism in extrahepatic tissues upon the onset of hyperglycemia. To define the hepatocyte-specific role of Gα13 in systemic glucose metabolism, we generated liver-specific Gna13 knockout (G13 LKO) mice and attempted to identify a molecular mediator and understand its effects on peripheral insulin resistance using proteomic techniques. Here, we report that Gα13 expression was markedly lower in the livers of mice or patients with hyperglycemia or diabetes. We demonstrate that selective ablation of Gna13 in mouse hepatocytes caused glucose intolerance and insulin resistance in other organs through overproduction of inter-α-trypsin inhibitor heavy chain 1 (ITIH1), a liver-derived secretory protein. In subjects with impaired glucose tolerance or type 2 diabetes, ITIH1 production was enhanced.

RESULTS

13 is down-regulated in the liver of diabetic mice

First, we examined Gα13 expression in major metabolic organs. In mice rendered obese through a high-fat diet (HFD), Gα13 was decreased in the liver and hepatocytes, whereas no difference was found in Gα13 expression in adipose tissue or skeletal muscle (P < 0.01; Fig. 1, A to C). The phenotypic changes were confirmed in genetically obese mice (P < 0.01; Fig. 1D). A strong inverse correlation existed between hepatic Gα13 expression and fasting blood glucose concentrations in both animal models (P < 0.01; Fig. 1E). A similar correlation was confirmed in the overall analysis for the obese diabetic mice in both animal models (P < 0.05; Fig. 1E), suggesting that decreased hepatic Gα13 was associated with hyperglycemia. To validate this association, we monitored hepatic Gα13 in human subjects with non-alcoholic fatty liver disease (NAFLD), in accordance with the degree of glucose tolerance (cohort #1). When compared to human subjects with normal glucose tolerance (fasting blood glucose below 110 mg/dl), there was a decrease in Gα13 in the liver of subjects with impaired glucose tolerance (fasting blood glucose ranging between 110 and 126 mg/dl) (Fig. 1F and table S1). The decrease in Gα13 was even greater in patients with overt type 2 diabetes (fasting blood glucose over 126 mg/dl) (Fig. 1F and table S1). A similar outcome was obtained with liver biopsy specimens stained immunohistochemically for Gα13. Liver biopsy specimens from patients with impaired glucose tolerance or type 2 diabetes showed negative-to-faint Gα13 staining compared to weak-to-moderate Gα13 staining intensity for liver specimens from subjects with normal glucose tolerance (Fig. 1G). Consistently, hepatic Gα13 expression was inversely correlated with indices of diabetes such as the HOMA-IR (homeostatic model assessment of insulin resistance) test, insulin and glucose concentrations, HbA1c, adipose tissue insulin resistance, and C-peptide (P < 0.05; Fig. 1H). When the samples were divided into two subgroups by the median value of hepatic Gα13 expression, clinical indices of diabetes were elevated in subjects with lower hepatic Gα13 (P < 0.01; fig. S1A). In another cohort study, similar results were found in patients with chronic hepatitis C virus infection who had clinical characteristics of type 2 diabetes (cohort #2) (P < 0.05; fig. S1, B to D). No significant correlations existed between hepatic Gα13 expression and lipid profiles assessed in the subjects in cohort #1 or cohort #2 (fig. S2, A to D). These results suggested that a decrease in Gα13 in the liver may affect glucose homeostasis.

Fig. 1 Decrease in hepatic Gα13 in obese mice and patients with diabetes.

(A) Immunoblot analyses for Gα13 expression of protein lysates of liver, adipose tissue (epididymal fat), and skeletal muscle (gastrocnemius) collected from C57BL/6 mice fed a normal diet (ND) or a high-fat diet (HFD) for 9 weeks. The relative band intensities of the immunoblots were quantified by densitometry and expressed as relative intensity normalized to β-actin (n = 4 per group). (B) Representative images of liver sections from C57BL/6 mice fed a normal diet (ND) or a high-fat diet (HFD) for 9 weeks [as in (A)] immunostained for Gα13 (brown) using a polyclonal antibody against Gα13 (n = 3 to 4 per group). (C) Immunoblot analysis for Gα13 expression in protein lysates of primary hepatocytes isolated from C57BL/6 mice fed either an ND or HFD for 12 weeks. The relative band intensities of the immunoblots were quantified by densitometry and expressed as relative intensity normalized to β-actin (n = 4 per group). (D) Immunoblot analyses for Gα13 expression in protein lysates of liver collected from wild-type (WT), obese ob/ob (top), and diabetic db/db (bottom) mice fed normal chow. Quantification of densitometry analyses for immunoblots is shown below the representative gels (n = 5 per group). The relative band intensities of the immunoblots were quantified by densitometry and expressed as intensity relative to β-actin. (E) Correlation analyses between hepatic Gα13 expression and fasting blood glucose concentrations in C57BL/6 mice fed a normal diet (ND) or high-fat diet (HFD) for 9 weeks (left, n = 8 or 13 per group) and wild-type (WT), obese ob/ob, or diabetic db/db mice (middle and right, n = 5 per group) fed a normal diet. An identical correlation was reanalyzed (far right box) using combined data for the same obese ob/ob and diabetic db/db mice (n = 23). (F) Immunoblot analyses for Gα13 expression in protein lysates of liver specimens from individuals with normal glucose tolerance (NGT), impaired glucose tolerance (IGT), and overt diabetes (T2DM) (NAFLD cohort #1). Quantification of densitometry analyses for immunoblots relative to β-actin is shown below the representative gels (n = 8 to 10 per group). (G) Representative images of liver sections obtained from the same individuals described in (F) stained for Gα13 expression (brown) using a polyclonal antibody against Gα13 (left, n = 3 to 4 per group) or stained with hematoxylin and eosin (H&E) (right, n = 1 to 2 per group). Scale bars, 100 μm. (H) Correlation analyses between hepatic Gα13 expression and diabetic indices such as the homeostatic model assessment of insulin resistance (HOMA-IR) test, insulin, fasting glucose, HbA1c, adipose tissue insulin resistance index (adipose tissue IR), and C-peptide in subjects with non-alcoholic fatty liver disease (NAFLD) or normal controls (n = 27, total number of subjects including normal controls and NAFLD patients). Values are expressed as means ± SEM. (**P < 0.01 versus ND or WT). Data were analyzed by two-tailed Student’s t test (A and D) or Pearson correlations (E and H). N.S., not significant.

Liver-specific ablation of Gα13 exacerbates diet-induced insulin resistance in mouse peripheral tissues

To identify the pathophysiological role of hepatic Gα13 in systemic glucose metabolism, we generated a mouse strain with liver-specific ablation of Gna13 (G13 LKO) by breeding Gna13flox/flox mice with albumin-Cre transgenic mice (fig. S3A). qRT-PCR and immunoblotting analyses of different organs and primary hepatocytes verified specific deletion of Gα13 in hepatocytes (fig. S3B). Thereafter, we subjected wild-type and G13 LKO mice to a high-fat diet for 9 weeks and monitored metabolic profiles. No significant difference in diet-induced obesity was observed between genotypes (fig. S3, C and D). Liver morphology and serum biomarkers for hepatocellular injury remained unchanged (fig. S3, E and F). With respect to lipid metabolism, triglyceride content in both liver and serum was unaffected, whereas serum total cholesterol and low-density lipoprotein were slightly decreased in G13 LKO mice compared to wild-type mice (fig. S3G). Our results indicated that hepatocyte-specific deletion of Gna13 did not exacerbate liver steatosis in mice with diet-induced obesity.

Given the inverse relationship between hepatic Gα13 and hyperglycemia, we next explored the effects of liver-specific ablation of Gα13 on blood glucose and insulin concentrations in mice. Fasting glucose was higher in G13 LKO mice fed a high-fat diet compared to wild-type mice fed a high-fat diet, but this effect was absent in animals fed a normal diet (Fig. 2A and fig. S4A). In G13 LKO mice fed either a high-fat diet or normal diet, however, both glucose and insulin tolerance were impaired (Fig. 2, B and C, and fig. S4, B and C), whereas serum insulin concentrations after glucose challenge were not significantly different between genotypes (Fig. 2D and fig. S4D). Similarly, neither insulin nor C-peptide showed significant decreases in G13 LKO mice under fasted or fed conditions compared to wild-type mice; this was consistent with comparable pancreatic islet sizes observed during histological examination (Fig. 2, E to G, and fig. S4, E to G). These results suggested that glucose intolerance manifested in G13 LKO mice might be due to impaired systemic insulin sensitivity rather than a defect in glucose-stimulated insulin secretion.

Fig. 2 Impaired glucose homeostasis and insulin resistance in G13 LKO mice fed a high-fat diet.

(A) Measurement of fasting blood glucose concentrations in wild-type (WT) mice or mice with liver-specific ablation of Gna13 (G13 LKO) fed a high-fat diet (HFD) for 9 weeks (n = 5 per group). (B) Results of the glucose tolerance test (glucose gavage; 2 g/kg body weight) in WT and G13 LKO mice fed an HFD for 10 weeks (n = 6 per group). (C) Results of the insulin tolerance test (insulin injection; 1.5 insulin units/kg body weight) in WT and G13 LKO mice fed an HFD for 13 weeks (n = 12 per group). (D) Measurements of serum insulin concentrations during the glucose tolerance test in WT and G13 LKO mice fed an HFD for 10 weeks (n = 6 per group). (E) Measurements of fasting blood glucose concentrations in WT or G13 LKO mice fed an HFD for 5 weeks (n = 4 or 5 per group). Mice were subjected to fasting and then refeeding (fasted for 16 hours and then refed for 4 hours). (F) Measurements of serum insulin and C-peptide concentrations in the same mice as in (E) (n = 6 per group). (G) Representative images of pancreas sections from WT and G13 LKO mice fed an HFD for 9 weeks and stained for hematoxylin and eosin (n = 3 to 4 per group). (H) Immunoblots for phosphorylated-Akt (p-Akt) and total Akt expression in protein lysates of the livers from WT or G13 LKO mice fed an HFD for 5 weeks and injected with a single dose of insulin (2 insulin units/kg body weight; i.p., 15 min) or from WT or G13 LKO mice subjected to fasting and then refeeding (fasted for 16 hours and then refed for 4 hours). (I) Immunoblots for phosphorylated-Akt (p-Akt) and total Akt expression in protein lysates of primary hepatocytes isolated from WT or G13 LKO mice fed an HFD for 5 weeks and treated with a single dose of insulin (100 nM, 15 min). (J) Measurement of glucose production rate in hepatocytes from WT or G13 LKO mice fed an HFD for 5 weeks (n = 3 per group, experiments performed in triplicate). (K) Immunoblots for phosphorylated-Akt (p-Akt) and total Akt expression in protein lysates of epididymal fat tissue or soleus muscle from the same mice as in (E). Quantification of densitometry analyses for immunoblots relative to β-actin is presented below each blot (n = 3 per group). (L) Measurement of 2-deoxyglucose uptake rate in epididymal fat tissue or soleus muscle from WT or G13 LKO mice fed an HFD for 5 weeks (n = 4 or 6 per group). (M and N) The effect of hepatic Gα13 overexpression on glucose metabolism was determined by analyzing glucose (M) or insulin tolerance (N) in C57BL/6 mice injected via the tail vein with control lentiviruses (Lv-Con) or lentiviruses expressing mouse Gα13 (Lv-Gα13) (1.1 × 107 transduction units). Injected mice were subsequently subjected to HFD feeding for 8 or 9 weeks (n = 6 or 7 per group). For (B) to (D) and (M) and (N), insets represent area under the curve (AUC). Values are expressed as means ± SEM. *P < 0.05, **P < 0.01 for G13 LKO versus WT (A to C and K) or Lv-Gα13 versus Lv-Con (M and N). Data were analyzed by two-tailed Student’s t test (A to D, K, M, and N) or one-way ANOVA followed by Bonferroni post hoc tests (E, F, J, and L). N.S., not significant.

To better understand the effects on insulin signaling of Gna13 deletion in mouse hepatocytes, we measured phosphorylated Akt (phospho-Akt) as an insulin sensitivity marker in major metabolic organs of mice subjected to insulin challenge or a fasting and refeeding regimen. Wild-type and G13 LKO mice were fed a high-fat diet for 5 weeks to evaluate the metabolic impact of hepatic Gna13 deficiency on systemic insulin sensitivity without the possible confounding effect of inflammatory mediators induced by a long-term high-fat diet (1, 5). Phospho-Akt expression was unchanged in the liver or in primary hepatocytes isolated from mice fed a high-fat diet or normal chow (Fig. 2, H and I, and fig. S4, H and I). Glucose production in mouse hepatocytes was also comparable (Fig. 2J), presumably due to similar expression of gluconeogenic genes (fig. S4J). However, phospho-Akt was notably diminished in adipose tissue and skeletal muscle of G13 LKO mice fed either a high-fat diet or normal diet (Fig. 2K and fig. S4K). Consistently, liver-specific ablation of Gα13 completely abolished insulin-stimulated 2-deoxyglucose uptake in epididymal fat pad and skeletal muscle in ex vivo cultures (Fig. 2L). Moreover, the outcome of glucose tolerance assays after lentiviral overexpression of Gα13 in the liver of wild-type mice further confirmed the hypothesis that loss of Gα13 in hepatocytes caused systemic glucose intolerance (Fig. 2, M and N), potentially due to impaired insulin sensitivity in extrahepatic tissues. During challenge with a high-fat diet, Gα13 overexpression in the liver affected none of the parameters measured including body weight gain, food consumption, and liver and epididymal fat weight (fig. S4, L to O). Gα13 expression did not change in adipose tissue and skeletal muscle (fig. S4P), highlighting the potential role of hepatic Gα13 in systemic glucose tolerance. These results prompted us to investigate the potential liver-secreted molecules causing systemic insulin resistance.

Ablation of Gα13 promotes the secretion of ITIH1 from mouse liver

To identify the hepatocyte-derived factors induced by ablation of Gna13, we prepared conditioned media from cultured hepatocytes from either G13 LKO or wild-type mice fed a high-fat diet for 5 weeks. Conditioned media from G13 LKO hepatocytes inhibited insulin-dependent Akt phosphorylation in differentiated 3T3-L1 or C2C12 cells compared to conditioned media from wild-type mouse hepatocytes (Fig. 3A). To elucidate the hepatocyte-derived soluble factors in the conditioned media, we performed semi-quantitative secretome analysis using conditioned media depleted of abundant plasma proteins (albumin and immunoglobulin). Of a total of 530 proteins detected, 104 were designated “secreted,” whereas others were grouped as “non-secreted,” based on the possession of an N-terminal signal sequence and the criteria annotated by UniProt (Fig. 3B). Of the secreted protein candidates, 67 were annotated as “liver-enriched” by UniProt or the Human Protein Atlas. The secretion of 42 proteins was increased in response to liver-specific ablation of Gna13, whereas secretion of 25 proteins was decreased. Of the top differentially secreted proteins showing twofold or more increased or decreased abundance, we narrowed our focus to ITIH1 because of its marked increase and potential effect on diabetes (Fig. 3C). In mice fed a high-fat diet, liver-specific ablation of Gna13 augmented hepatic and serum ITIH1 expression (Fig. 3, D to F), confirming the results obtained from proteomic analyses. Plasma protease C1 inhibitor was not examined due to the lack of a direct role in the development of diabetes and its controversial expression pattern depending on diabetes type (10, 11).

Fig. 3 Identification of ITIH1 as a hepatocyte-secreted protein enhanced by loss of Gα13.

(A) Immunoblots for phosphorylated-Akt (p-Akt) and total Akt expression in protein lysates of 3T3-L1 and C2C12 cells incubated with conditioned media (CM) that were collected from primary hepatocytes isolated from WT or G13 LKO mice fed a high-fat diet (HFD) for 5 weeks. (B) The numbers of proteins in primary hepatocyte conditioned media from WT and G13 LKO mice in (A) that were differentially secreted (n = 3). (C) List of top differentially secreted proteins from (B) detected by semi-quantitative secretome analysis. (D) Immunoblots for ITIH1 expression in protein lysates from the livers of WT or G13 LKO mice fed an HFD for 9 weeks. Quantification of densitometry analysis for bands is shown next to the immunoblot (n = 7 per group). β-Actin was the loading control. (E) Measurements for serum ITIH1 concentrations for WT or G13 LKO mice fed an HFD for 12 weeks. Albumin was used as the loading control. (F) Quantitative analysis of ELISA for serum ITIH1 for the same mice as in (E) (n = 7 per group). Values are expressed as means ± SEM. **P < 0.01 for G13 LKO versus WT mice (D and F). Data were analyzed by two-tailed Student’s t test (D and F). N.S., not significant.

Hepatic and serum ITIH1 are increased in subjects with impaired glucose tolerance or diabetes

To validate possible correlations between Gα13 and ITIH1 in clinical specimens, we analyzed ITIH1 expression in specimens from human subjects with diabetes where Gα13 expression was markedly decreased. In the first cohort of patients (cohort #1), ITIH1 was elevated in liver and serum samples from subjects with impaired glucose tolerance and in patients with overt diabetes (Fig. 4, A to C). Serum ITIH1 concentrations exhibited a strong negative correlation with hepatic Gα13 (r = −0.776, P < 0.01; Fig. 4D), suggesting an inverse relationship between hepatic Gα13 and ITIH1 production. Moreover, serum ITIH1 concentrations were highly associated with various diabetic parameters irrespective of lipid profiles (Fig. 4, E and F, and fig. S5, A and B). Similar results were observed in patients with hepatitis C virus infection (cohort #2), who showed a stepwise increase in ITIH1 production in both liver and plasma samples that was associated with the HOMA-IR test, a method used to assess insulin resistance (fig. S6, A and B). In contrast, expression of ITIH2, another liver-enriched member of the inter-α-trypsin inhibitor (ITI) protein family, was not elevated in the liver of diabetic subjects or G13 LKO mice (Fig. 4A and fig. S6, A and C). In addition, Itih1 and Itih2 mRNA expression was not significantly different between genotypes (fig. S6D), suggesting possible posttranscriptional regulation of ITIH1 by Gα13.

Fig. 4 Increased hepatic and serum ITIH1 concentrations in NAFLD subjects with diabetes.

(A) Immunoblot analyses for ITIH1 and ITIH2 expression in protein lysates of liver specimens from subjects with normal glucose tolerance (NGT), impaired glucose tolerance (NGT), or type 2 diabetes (T2DM) (NAFLD cohort #1). (B) Representative immunofluorescence images of liver sections from NAFLD patients with diabetes stained for ITIH1 (red) with DAPI counterstain (blue) (n = 3 to 4 per group, 60× magnification). (C) Quantitative analysis of ELISA for serum ITIH1 from the same subjects as in (A) (n = 7 per group). (D and E) Correlation analyses between serum ITIH1 concentrations and hepatic Gα13 expression (D, n = 17) and serum ITIH1 concentrations and insulin resistance–related indices including HOMA-IR, insulin, glucose, HbA1c, adipose tissue IR, and C-peptide (E, n = 21) in subjects with NAFLD or normal controls. (F) Changes in diabetic indices in relation to serum ITIH1 concentrations for human subjects categorized into two subgroups by the median value of serum ITIH1 (n = 10 per group). Data are shown as box and whisker plots. Box, interquartile range (IQR); whiskers, 5 to 95 percentiles; horizontal line within box, median. Values are expressed as means ± SEM. Data were analyzed by one-way ANOVA followed by Bonferroni post hoc tests (C), Pearson correlations (D and E), or Mann-Whitney tests (F).

Liver-secreted ITIH1 binds to hyaluronan surrounding mouse adipose tissue and skeletal muscle

Circulating ITIH1 is synthesized exclusively in hepatocytes and secreted into the bloodstream, stabilizing the extracellular matrix by covalent binding to hyaluronan (1214). Considering recent studies demonstrating excessive hyaluronan accumulation in insulin-resistant tissues (15, 16), we examined whether ITIH1 oversecreted in response to hepatic Gα13 deficiency might bind to hyaluronan surrounding adipose tissue and skeletal muscle in mice with impaired insulin sensitivity. ITIH1 expression in adipose tissue and skeletal muscle was higher in G13 LKO mice compared to wild-type mice both fed a high-fat diet (Fig. 5A), with no change in hyaluronan (Fig. 5B). Immunohistochemical studies on adipose tissue and skeletal muscle from G13 LKO mice showed enhanced ITIH1 immunoreactivity (Fig. 5C), whereas ITIH1 staining was reduced in tissues from mice with hepatic Gα13 overexpression (Fig. 5D), verifying the hepatic origin of ITIH1 secretion. In addition, ITIH1 staining in the mouse tissues was abrogated by pretreatment with hyaluronidase (Fig. 5C), demonstrating that ITIH1 required hyaluronan for binding to the tissues. Moreover, we found that ITIH1 deposition was augmented by hepatic Gα13 deficiency in extracellular matrix–enriched fractions from adipose and skeletal muscle tissues (Fig. 5E). Detection of fibronectin, but not β-tubulin, indicated appropriate enrichment of extracellular matrix proteins (fig. S7). In immunofluorescence assays, we corroborated the enhanced interaction of ITIH1 with hyaluronan in response to hepatic Gα13 deficiency (Fig. 5F, left and middle). Similar results were obtained in 3T3-L1 and C2C12 cells incubated with conditioned media from cultured G13 LKO hepatocytes (fig. S8). In liver tissue, ITIH1-hyaluronan complexes were not readily detected in either group, although ITIH1 staining intensity was markedly enhanced by hepatic Gα13 deficiency, presumably due to the lack of hepatic deposition of hyaluronan (Fig. 5F, right). Together, these results indicate that overproduction of ITIH1 by Gα13-deficient liver may result in ITIH1 deposition in adipose tissue or skeletal muscle through its interaction with hyaluronan.

Fig. 5 Increased ITIH1 deposition onto hyaluronan surrounding peripheral tissues in G13 LKO mice.

(A) Quantitative analyses of ELISA for ITIH1 in homogenates of epididymal fat tissue (n = 8 or 10 per group) and gastrocnemius muscle (n = 12 or 13 per group) from wild-type (WT) or G13 LKO mice fed a high-fat diet (HFD) for 9 weeks. (B) Quantitative analyses of ELISA for hyaluronan (HA) in the homogenates of epididymal fat tissue and gastrocnemius muscle of the same animals as in (A) (n = 8 per group). (C) Representative images of epididymal fat tissue and gastrocnemius muscle sections from mice as described in (A) stained for ITIH1 (brown) (n = 4 per group). The slides were pretreated with hyaluronidase (20 U/ml) or vehicle for 2 hours at 37°C. Scale bars, 100 μm. (D) Representative images of epididymal fat tissue and gastrocnemius muscle sections from wild-type (WT) C57BL/6 mice injected via the tail vein with control lentiviruses (Lv-Con) or lentiviruses expressing mouse Gα13 (Lv-Gα13) (1.1 × 107 transduction units). The injected mice were subsequently fed a high-fat diet (HFD) for 11 weeks (n = 4 per group). Scale bars, 100 μm. (E) Immunoblot analyses for ITIH1 expression in fractions enriched for extracellular matrix (ECM) prepared from epididymal fat tissue or tibialis anterior muscle of the same mice as in (A). ECM proteins were normalized to tissue weight and were stained with Coomassie blue stain after SDS-PAGE separation. (F) Representative immunofluorescence images of epididymal fat tissue, gastrocnemius muscle, and liver sections from the same mice as in (C) stained for ITIH1 (red) and hyaluronan binding protein (HABP) (green) (n = 3 per group). For liver tissue, blue color in merged images represents DAPI staining for nuclei. Arrowheads indicate merged color (yellow/orange). Scale bars, 25 μm. For (A) and (B), values are expressed as means ± SEM. *P < 0.05 for G13 LKO versus WT mice (A). Data were analyzed by two-tailed Student’s t test. N.S., not significant.

ITIH1 secretion in response to Gna13 ablation is due to O-GlcNAC transferase induction

To elucidate the molecular basis underlying overproduction and secretion of ITIH1 under the conditions of Gα13 deficiency or diabetes, we examined the effect of hyperglycemia on hepatic Gα13 expression in mice after an oral glucose gavage. A single glucose gavage in mice notably diminished Gα13 in the liver in conjunction with an increase in circulating ITIH1 (fig. S9A). A similar result was observed in cultured mouse hepatocytes exposed to high glucose concentrations (fig. S9B), supporting the causative effect of hyperglycemia on Gα13 repression in hepatocytes. Considering that most secretory proteins are modified by glycosylation to ensure protein stabilization and solubility (17, 18), we focused on the possibility of glycosyl modification of ITIH1. In particular, O-GlcNAcylation is one of the major posttranslational modifications elicited by high glucose concentrations due to hyperglycemia or diabetes (19). The loss of Gα13 in liver hepatocytes intensified O-GlcNAcylation of proteins along with increased ITIH1 production (Fig. 6, A and B). O-GlcNAcylation was markedly enhanced in G13 LKO mouse hepatocytes compared to wild-type mouse hepatocytes under normal glucose concentrations (Fig. 6B and fig. S9B), raising the possibility that O-GlcNAcylation might be directly controlled by Gα13 signaling.

Fig. 6 Increase in O-GlcNAc transferase–mediated O-GlcNAcylation of proteins after loss of Gα13.

(A) Immunoblot analyses for O-GlcNAc proteins (CTD110.6 or RL2 clones) or ITIH1 expression in protein lysates of livers from WT or G13 LKO mice fed a high-fat diet (HFD) for 5 weeks. (B) Immunoblot analyses for O-GlcNAc proteins (CTD110.6 or RL2 clones) or ITIH1 expression in protein lysates of primary hepatocytes isolated from WT or G13 LKO mice fed an HFD for 5 weeks. Primary hepatocytes were incubated with high glucose (25 mM) for 24 hours. (C and D) Immunoblot analyses for ITIH1 or O-GlcNAc transferase (OGT) expression in protein lysates of the liver (C) or in sera (D) from WT or G13 LKO mice. Glucose (2 g/kg body weight) was orally administered to mice, and the liver tissues were harvested after 6 hours. The relative band intensities were quantified by densitometry analyses for the immunoblots (n = 3 per group). β-Actin and albumin were the loading controls for liver and serum samples, respectively. (E) Immunoblot analyses for OGT in the protein lysates of liver or hepatocytes isolated from WT or G13 LKO mice fed an HFD for 9 or 5 weeks, respectively. (F) Immunoblot analysis for OGT and ITIH1 in protein lysates of liver from C57BL/6 mice injected via the tail vein with control lentiviruses (Lv-Con) or lentiviruses expressing mouse Gα13 (Lv-Gα13) (1.1 × 107 transduction units). Injected mice were subsequently fed an HFD for 10 weeks. Immunoblots were quantified by densitometry (n = 3 per group). (G) Immunoblot analysis for OGT and ITIH1 expression in protein lysates of primary hepatocytes infected with adenoviruses carrying an active mutant of Gα13 (Ad-G13QL) or green fluorescent protein (Ad-GFP) as a control. (H) Immunoblot analysis for ITIH1 expression in conditioned media (CM) from the same primary hepatocytes as in (G). Albumin was the loading control for conditioned media samples. (I) Immunoblot analyses for O-GlcNAc proteins (CTD110.6 clone) or O-GlcNAc transferase (OGT) in ITIH1 immunoprecipitates from the liver homogenates or sera of mice fed a high-fat diet (HFD) for 9 weeks. Values are expressed as means ± SEM. *P < 0.05, **P < 0.01 for Lv-Gα13 versus Lv-Con (F). Data were analyzed by one-way ANOVA, followed by least significant difference (LSD) post hoc test (C) or two-tailed Student’s t test (F).

Given that O-GlcNAC transferase, an enzyme catalyzing protein O-GlcNAcylation, is up-regulated in the clinical settings of insulin resistance such as hyperglycemia and diabetes (20, 21), we examined whether Gα13-dependent modification of O-GlcNAc was mediated by O-GlcNAC transferase. After a glucose gavage, G13 LKO mice displayed augmented expression of O-GlcNAC transferase in the liver, which was accompanied by increased ITIH1 in liver and serum relative to wild-type control mice (Fig. 6, C and D). Similar results were found in liver or cultured hepatocytes of G13 LKO mice fed a high-fat diet (Fig. 6E). We corroborated increases in ITIH1 and O-GlcNAC transferase associated with Gα13 repression in mice subjected to streptozotocin-induced diabetes (fig. S9, C to E). O-GlcNAC transferase expression was suppressed by Rho signaling downstream of Gα13, as evidenced by the outcomes using cultured primary hepatocytes exposed to chemical inhibitors and AML12 cells transfected with a constitutively active Rho mutant protein (fig. S9, F and G). In addition, lentiviral vector–mediated overexpression of Gα13 inhibited O-GlcNAC transferase expression in mouse liver in tandem with diminished ITIH1 in liver and serum (Fig. 6F and fig. S9H). Infection of cultured mouse primary hepatocytes with adenovirus carrying a constitutively active mutant form of Gα13 (Ad-G13QL) had a similar effect (Fig. 6, G and H). Moreover, immunoblotting of ITIH1 immunoprecipitates from liver or serum samples for O-GlcNAcylation verified ITIH1 O-GlcNAcylation (Fig. 6I). We further examined ITIH1 O-GlcNAcylation by liquid chromatography–tandem mass spectrometry (LC-MS/MS) using cultured HEK293A cells overexpressing FLAG-tagged human ITIH1 (fig. S10, A and B). We identified several putative O-GlcNAcylation sites on conserved amino acid residues including Ser590, Ser608, Ser820, and Ser824 (fig. S10, C and D). Of these, Ser590 and Ser608 residues were located on the C-terminal domain of ITIH1, whereas the other residues were located on the propeptide domain where the corresponding region was cleaved in the Golgi apparatus during the process of maturation and secretion of ITIH1.

O-GlcNAcylation may control protein stability by affecting protein ubiquitination (22). O-GlcNAC transferase overexpression increased ITIH1 stability along with a decrease in ITIH1 ubiquitination (fig. S11, A and B). Likewise, O-GlcNAC transferase knockdown using an shRNA-expressing plasmid against O-GlcNAC transferase (shOGT) suppressed ITIH1 O-GlcNAcylation and stabilization in cultured AML12 cells exposed to high glucose concentrations or overexpressing ITIH1 (fig. S11, C and D). Treatment of the AML12 cells with ST045849, an inhibitor of O-GlcNAC transferase, had a similar effect (fig. S11E), suggesting that ITIH1 was stabilized by O-GlcNAcylation. Given the role of O-GlcNAC transferase–mediated protein O-GlcNAcylation in the etiology of insulin resistance and diabetes (19, 21), we further tested the effect of hepatic O-GlcNAC transferase on systemic glucose tolerance in vivo. Treatment of G13 LKO mice with ST045849 (20 mg/kg per day, for 4 consecutive days) not only prevented an ITIH1 increase and O-GlcNAcylation in the liver and serum (Fig. 7, A and B) but also improved glucose tolerance (Fig. 7C). Likewise, overexpression of O-GlcNAC transferase in hepatocytes in vivo using a hydrodynamic injection technique enhanced both hepatic ITIH1 and serum O-GlcNAcylated ITIH1, whereas knockdown of hepatic O-GlcNAC transferase resulted in the opposite effects (fig. S11F). O-GlcNAC transferase expression was not changed in the other tissues examined, indicative of efficient and selective O-GlcNAC transferase gene delivery into the mouse liver (fig. S11G). Consistently, O-GlcNAC transferase knockdown lowered ITIH1 in the liver and serum of injected mice and rescued the glucose-intolerant phenotype of G13 LKO mice (fig. S11, H to J). To strengthen our hypothesis, we used a lentiviral vector carrying the human OGT gene and an albumin promoter. Specific O-GlcNAC transferase overexpression in hepatocytes increased ITIH1 in liver and serum compared to control mice (Fig. 7D) and impaired glucose and insulin tolerance (Fig. 7, E and F). Enhanced O-GlcNAcylation and stabilization of ITIH1 were verified in cultured primary hepatocytes (Fig. 7G). O-GlcNAC transferase expression was comparable in adipose tissue and skeletal muscle (Fig. 7H).

Fig. 7 Overproduction and O-GlcNAcylation of ITIH1 induced by O-GlcNAc transferase.

(A) Immunoblot analysis for O-GlcNAc proteins (CTD110.6) or ITIH1 expression in protein lysates of livers from WT or G13 LKO mice injected with ST045849, an inhibitor of O-GlcNAc transferase (OGT; 20 mg/kg body weight). Mice were subjected to vehicle (left, n = 2 per group) or glucose gavage (right, n = 3 to 4 per group). Glucose (2 g/kg body weight) was orally administered to mice, and the liver tissues were harvested after 6 hours. (B) Immunoblot analysis for ITIH1 in serum samples from the same mice as in (A). (C) Analysis of the glucose tolerance test (2 g/kg body weight) in WT and G13 LKO mice injected with ST045849 (OGT inhibitor, 20 mg/kg body weight) (n = 6 per group). Inset represents area under the curve (AUC) for the glucose tolerance test (GTT). (D) Immunoblot analysis for ITIH1 expression in protein lysates of liver or in sera from 8-week-old C57BL/6 mice injected with lentiviral vector carrying the human OGT gene with an albumin promoter (Lv-OGTalb) or empty control vector (Lv-Con) via the tail vein (2 × 107 transduction units each). Albumin was the loading control for serum samples. (E and F) The effect of hepatic Gα13 overexpression on glucose metabolism was determined using a glucose test (E, n = 10 per group) or insulin tolerance test (F, n = 9 to 10 per group) in the same mice as in (D). Insets show the area under the curve (AUC) for the glucose tolerance test (GTT) or insulin tolerance test (ITT). (G) Immunoblot analyses for ITIH1 and O-GlcNAc ITIH1 in primary hepatocytes isolated from 8-week-old C57BL/6 mice injected via tail vein with lentiviral vector carrying the human OGT gene with an albumin promoter (Lv-OGTalb) or empty control vector (Lv-Con) (2 × 107 transduction units). Immunoblotting for O-GlcNAc proteins (CTD110.6 clone) was performed on immunoprecipitates of ITIH1 in primary hepatocyte cell lysates. (H) Immunoblot analyses for OGT in protein lysates of liver, epididymal fat tissue, and gastrocnemius muscle of 8-week-old C57BL/6 mice injected via tail vein with lentiviral vector carrying the human OGT gene with an albumin promoter (Lv-OGTalb) or empty vector control (Lv-Con) (2 × 107 transduction units) (n = 3 per group). The relative band intensities were quantified by densitometry analyses for the immunoblots. Values were expressed as means ± SEM. *P < 0.05 for G13 LKO + vehicle versus WT + vehicle; ##P < 0.01 for G13 LKO + ST045849 versus G13 LKO + vehicle (C); and *P < 0.05, **P < 0.01 for Lv-OGTalb versus Lv-Con (E, F, and H). Data were analyzed by one-way ANOVA, followed by least significant difference (LSD) (A), Bonferroni post hoc tests (C), or two-tailed Student’s t test (E, F, and H). N.S., not significant.

Antibody neutralization of ITIH1 overcomes systemic glucose intolerance and insulin resistance in mice

Having identified ITIH1 overproduction in response to a decrease in Gα13 in mouse liver and the consequent induction of insulin resistance in different organs, we finally examined whether antibody neutralization of ITIH1 had a beneficial effect on systemic glucose homeostasis in mice. C57BL/6 mice were placed on a high-fat diet or normal diet for 8 to 9 weeks, followed by daily injections of a custom-synthesized rabbit anti-ITIH1–neutralizing antibody or preimmune IgG as a control for an additional 2 weeks, and the metabolic outcomes were monitored. In this experiment, ITIH1 neutralization did not alter body weight or liver and epididymal fat weight (Fig. 8, A to C). Antibody neutralization normalized glucose disposal and insulin sensitivity with a decrease in serum ITIH1 (Fig. 8, D to F). Phospho-Akt was altered in 3T3-L1 or C2C12 cells incubated with conditioned media from G13 LKO hepatocytes in the presence of anti-ITIH1–neutralizing antibody (fig. S12A). We further examined the role of ITIH1 in insulin responsiveness using conditioned media from G13 LKO cultured primary hepatocytes deficient in ITIH1. siRNA-mediated Itih1 gene silencing abrogated the inhibitory effect of liver-specific Gα13 ablation on phospho-Akt expression and 2-deoxyglucose uptake (fig. S12, B and C). Immunoblotting assays using peptide competition and siRNA targeting of endogenous ITIH1 confirmed specificity of our custom-made antibody (fig. S12, D and E). ITIH1 neutralization using the anti-ITIH1 antibody, confirmed by a decrease in circulating ITIH1, improved glucose and insulin tolerance in G13 LKO mice fed a high-fat diet (Fig. 8, G to I), with no changes in body weight or fasting blood glucose concentrations (Fig. 8, J and K). Insulin-stimulated glucose uptake was improved in adipose tissue or skeletal muscle of wild-type mice fed a high-fat diet (Fig. 8L).

Fig. 8 Recovery of impaired glucose tolerance in HFD-fed G13 LKO mice treated with anti-ITIH1 antibody.

(A to C) Measurements for body weight gain and epididymal fat weights for 6-week-old C57BL/6 mice subjected to a high-fat diet (HFD) or normal diet (ND) for 10 weeks with daily injections of rabbit polyclonal anti-ITIH1 antibody or control pre-IgG antibody for the last 2 weeks of the HFD or ND diet (n = 7 to 8 per group). Body weight gain (A), liver–to–body weight ratio (B), and epididymal fat weight (C). (D, E, G, and H) Effects of anti-ITIH1 antibody treatment on the results of the glucose tolerance test (1.5 g/kg body weight) or insulin tolerance test (0.75 insulin units/kg body weight) in C57BL/6 mice fed an HFD or ND (D and E) and WT or G13 LKO mice fed an HFD (G and H). Mice were injected daily with purified anti-ITIH1 antibody or pre-IgG (i.p., 250 μg each) for the last 2 weeks of HFD or ND feeding. (D) Analysis of the glucose tolerance test in C57BL/6 mice fed an ND or HFD for 10 weeks (ND, n = 7; HFD, n = 8; HFD–pre-IgG, n = 7; HFD–anti-ITIH1 antibody, n = 8). (E) Analysis of the insulin tolerance test in C57BL/6 mice fed an ND or HFD for 11 weeks (ND, n = 7; HFD, n = 8; HFD–pre-IgG, n = 6; HFD–anti-ITIH1 antibody, n = 6). (F) Quantitative analysis of ELISA for ITIH1 in serum (ND, n = 7; HFD, n = 8; HFD–pre-IgG, n = 7; HFD–anti-ITIH1 antibody, n = 8). (G) Analysis of the glucose tolerance test in WT and G13 LKO mice fed an HFD for 11 weeks (n = 7 per group). (H) Analysis of the insulin tolerance test in WT and G13 LKO mice fed an HFD for 12 weeks (n = 6 per group). (I) Quantitative analysis of ELISA for serum ITIH1 in WT and G13 LKO mice fed an HFD (WT–pre-IgG or G13 LKO–pre-IgG, n = 7; G13 LKO–anti-ITIH1 antibody, n = 6). (J and K) Body weight gain (J) or fasting blood glucose concentrations (K) in G13 LKO mice fed an HFD for 13 weeks and treated with anti-ITIH1 antibody or control IgG as in (A) (n = 5 per group). Days 0 and 14 represent before and after treatments, respectively, with anti-ITIH1 antibody or pre-IgG control antibody. (L) Measurements of 2-deoxyglucose uptake in epididymal fat tissue of WT mice fed an HFD for 16 weeks. (Left) Basal 2-deoxyglucose uptake (n = 3 mice) and uptake after glucose gavage (n = 3 mice) in adipose tissue from mice treated with pre-IgG antibody control or anti ITIH1 antibody (right). Basal 2-deoxyglucose uptake (n = 3 mice) and uptake after glucose gavage (n = 3 mice) in skeletal soleus muscle of mice treated with pre-IgG antibody control or anti-ITIH1 antibody. For (D), (E), (G), and (H), insets represent area under the curve (AUC). Values are expressed as means ± SEM. *P < 0.05, **P < 0.01 for HFD versus ND (A, D, and E) or G13 LKO mice treated with pre-IgG versus WT mice treated with pre-IgG (G to I); #P < 0.05, ##P < 0.01 for mice fed an HFD and treated with anti-ITIH1 antibody versus mice fed an HFD and treated with pre-IgG antibody (D and E) or G13 LKO mice treated with anti-ITIH1 antibody versus G13 LKO mice treated with pre-IgG antibody (G to I). Data were analyzed by one-way ANOVA followed by Bonferroni (A to K) or least significant difference (LSD) (L) post hoc tests. N.S., not significant.

DISCUSSION

Given that insulin resistance precedes the development of diabetes and etiologically determines metabolic syndrome in multiple organs, therapeutic approaches targeting single organs or a subset of intracellular signaling pathways have shown limited success or considerable side effects (23). The liver regulates overall glucose metabolism in response to changes in extracellular nutritional availability. Hepatic steatosis, which frequently accompanies insulin resistance, usually occurs before metabolic dysfunction in other organs (1, 2), suggesting the causal effect of liver pathophysiology on overall metabolic disturbances. Nevertheless, the notion of the liver as the origin and driver of global metabolic defects in the setting of hyperglycemia has drawn little attention. Our current findings show that hyperglycemia decreases Gα13 in hepatocytes, which exacerbates glucose intolerance and insulin resistance in extrahepatic organs through an O-GlcNAC transferase–dependent increase in circulating ITIH1. Our results suggest that there may be an increase in ITIH1 upon hyperglycemic challenge.

Studies investigating Gα13 have been limited as mice lacking Gα13 show a defect in angiogenesis during embryonic development resulting in lethality (24). We generated hepatocyte-specific Gna13 knockout mice using the Cre-loxP system. Upon feeding these G13 LKO mice a high-fat diet, the mice exhibited impaired glucose tolerance and marked insulin resistance in the absence of changes in obesity and lipid profiles, suggesting that metabolic abnormalities manifested by G13 LKO mice could be attributed to a defect in glucose disposal. Our data showing that the metabolic profiles of liver were unaffected by hepatic ablation of Gα13 support the notion that the Gα13 decrease in hepatocytes due to hyperglycemia was directly linked to enhanced secretion of liver-derived soluble factors but not to the initiation of hepatocyte malfunction or injury.

Insulin resistance, observed in subjects with glucose intolerance, entails a compensatory hyperinsulinemia in the diet-induced obesity mouse model that enables pancreatic β cells to overcome a decrease in insulin sensitivity in the peripheral tissues. Although it is well established that feeding mice a long-term high-fat diet induces inflammation and impairs pancreatic beta cell function, we assumed that the failure of beta cells might not have occurred in our model given the marked increase in insulin upon glucose challenge in mice fed a high-fat diet. In our study, wild-type and G13 LKO mice fed either a high-fat diet or normal diet showed comparable serum insulin concentrations when subjected to glucose gavage, which might reflect a mild effect of high-fat diet challenge on pancreatic beta cells in our mouse model. However, ITIH1 may also contribute to pancreatic beta cell function. A recent study showed higher accumulation of IαI (inter-α-trypsin inhibitor that comprises ITIH1, ITIH2, and bikunin) and hyaluronan in the pancreatic islets of patients with type 1 diabetes (25). Thus, we propose that increased ITIH1 may affect beta cell function under more severe diabetic conditions (e.g., pancreatitis and type 1 diabetes). Moreover, we cannot exclude the potential involvement of other factors affecting plasma insulin concentrations (e.g., insulin clearance by either liver or kidney).

Given that most GPCRs exist in various oligomeric complexes (26), it is improbable that a single GPCR or ligand might account for a variety of metabolic consequences. Moreover, some GPCRs can interact with more than one Gα protein, indicating possible engagement of several Gα proteins upon ligand binding. Considering the complex nature of GPCR-Gα protein coupling, certain ligands may activate several types of GPCRs/Gα proteins, transducing mixed signals to downstream effectors. In the current study, G13 LKO mice revealed a cell type–specific role for Gα13 and excluded interference from other G proteins. Moreover, Gα13 in liver was decreased under hyperglycemic conditions regardless of ligand activation of corresponding GPCRs. Therefore, it is possible that a high glucose concentration functions as a nutritional environment or putative ligand for glucose-sensing orphan GPCRs potentially affecting intracellular signals modulating Gα13, but this remains to be established in future experiments. Overall, the results from the Gα13 knockout mice need to be interpreted separately from the outcomes using common Gα13-activating ligands. Other approaches such as Gα13-selective designer GPCRs are needed to corroborate our proposed mechanism.

To obtain mechanistic insights into how high glucose concentrations may down-regulate Gα13, we examined whether O-GlcNAcylation affects Gα13 stability. A switch from high glucose (25 mM, 24 hours) to low glucose (5 mM, 12 hours) caused the recovery of Gα13 toward that of control in AML12 cells or primary hepatocytes (fig. S13, A and B). However, O-GlcNAcylation staining remained higher than control in both AML12 and primary hepatocytes. Nonetheless, the effect of high glucose on Gα13 O-GlcNAcylation was not clear in immunoprecipitation assays (fig. S13A). In addition, shOGT transfection did not prevent high glucose concentrations from repressing Gα13 (fig. S11C). Thus, Gα13 stability seems to be affected by glucose concentrations in a dynamic fashion and may not solely depend on O-GlcNAcylation. Further studies are necessary to understand the possible effects of O-GlcNAcylation on Gα13 signaling.

In the present study, proteomics-based approaches using semi-quantitative, label-free LC-MS/MS identified differential abundance of secreted proteins in their nascent form regulated by hepatic Gα13. Labeling of peptides and proteins through either isobaric tagging (e.g., iTRAQ) or metabolic labeling (e.g., SILAC) allows more sensitive detection of differential abundance than the present method. However, the label-free method enables detection of proteins in their original form for MS-based abundance measurements and, thus, is suitable for biomarker discovery and validation in terms of better quantification capability and reproducibility (27, 28). Of the top-ranked liver-enriched secretory proteins affected by a decrease in Gα13, ITIH1 was identified as a key molecule accounting for metabolic dysfunction. ITIH1 is predominantly synthesized and secreted from hepatocytes with varying expression under pathological conditions (e.g., ITIH1 was lower in patients with liver fibrosis or hepatocellular carcinoma) (29, 30). Although we do not necessarily exclude local ITIH1 expression in adipose tissue and skeletal muscle, our results indicate that a larger amount of liver-derived ITIH1 is incorporated into the extracellular matrix of adipose tissue and skeletal muscle in G13 LKO mice compared to wild-type mice, as indicated by the outcomes of experiments using extracellular matrix–enriched fractions and immunohistochemistry.

Extracellular matrix is composed of myriad extracellular proteins secreted by various cell types and modulates not only biological processes but also intercellular communication. Growing evidence suggests that extracellular matrix remodeling in peripheral tissues is of potential importance for glucose metabolism and insulin signaling under diabetic conditions (15, 16). Hyaluronan, one of the major components of extracellular matrix, is an anionic nonsulfated glycosaminoglycan, whose content and density are increased in insulin-resistant tissues (15, 16). The stiffness or rigidity of hyaluronan, which is affected by various interacting proteins and proteoglycans, determines cellular function (3133). Here, we postulated that binding of liver-derived ITIH1 to hyaluronan results in extracellular matrix remodeling in the pathogenesis of insulin resistance and diabetes. In our study, hepatic Gα13 ablation had no effect on hyaluronan content in adipose tissue and skeletal muscle in mice. Instead, our findings indicated enhanced colocalization and interaction of ITIH1 and hyaluronan in the tissues of G13 LKO mice, highlighting the role of ITIH1 as a hyaluronan-binding protein that contributes to peripheral insulin resistance. In the liver, both glucose metabolism and insulin sensitivity were not affected by Gα13 deficiency, implying the marginal effect of ITIH1 on hyaluronan-mediated insulin resistance in the liver, presumably because hyaluronan is metabolized and degraded specifically by liver sinusoidal endothelial cells via endocytosis (34). In line with this, it is reported that hepatic stellate cells are the unique cell type responsible for hyaluronan synthesis in the liver, supporting the contention that excessive accumulation of hyaluronan is observed mostly in fibrotic/cirrhotic liver (35). Consistently, ITIH1-hyaluronan complexes were not readily detected in the liver of G13 LKO mice despite the marked staining intensity for ITIH1 in liver. Our results demonstrate that the liver contributes to the maintenance of systemic glucose homeostasis by regulating insulin sensitivity in other metabolic organs through ITIH1 secretion.

O-GlcNAcylation of proteins as a dynamic posttranslational modification may enable cells to sense and respond quickly to glucose availability (1921). O-GlcNAcylation of protein extracellular domains has been found in a variety of organisms (3638). Similarly, secreted macrophage migration inhibitory factor is O-GlcNacylated by O-GlcNAC transferase, which affects its ability to regulate tumorigenesis (39), consistent with the observation that O-GlcNAcylation was aberrantly increased in several secretory proteins lacking an epidermal growth factor (EGF)–like domain (40). Our data showed that ITIH1 oversecretion due to Gα13 deficiency or hyperglycemia was accompanied by O-GlcNAC transferase expression induction in hepatocytes. We also discovered the ability of O-GlcNAC transferase to increase ITIH1 stability with diminished ubiquitination of ITIH1 by either immunoprecipitation or immunoblotting assay, which is in line with the report that O-GlcNAcylation controls protein stability through protein ubiquitination (22). The outcomes from LC-MS/MS analysis provided further evidence for O-GlcNAc modification of ITIH1. We addressed the engagement of Rho in our proposed signaling pathway, although the detailed mechanistic insight by which loss of Gα13 up-regulates O-GlcNAC transferase remains to be investigated. We demonstrated the inhibitory effect of hyperglycemia on Gα13 in mouse hepatocytes and the resultant overproduction of ITIH1 mediated by O-GlcNAC transferase–catalyzed O-GlcNAcylation. In our study, we cannot rule out the possibility that Gα13 signaling modulates O-GlcNAcase or its activity, which opposes O-GlcNAC transferase function. In addition, our results do not necessarily exclude the potential involvement of other liver-derived secretory proteins in the development of systemic insulin resistance as O-GlcNAC transferase overexpression might result in a generalized increase in protein O-GlcNAcylation that is not limited to ITIH1.

Collectively, our findings revealed that a decrease in hepatic Gα13 promoted overproduction of O-GlcNAcylated ITIH1 in hepatocytes and its secretion into the systemic circulation. The hypothesis that ITIH1 binds directly to hyaluronan on the surface of adipose tissue and skeletal muscle resulting in extracellular matrix stabilization and consequent insulin resistance provides a new conceptual framework by which the liver may regulate systemic glucose homeostasis in response to varying glucose concentrations.

There are some limitations to our study. First, we used several in vivo gene transfer techniques in G13 LKO mice to modulate Gα13/O-GlcNAC transferase expression. A genetic loss-of-function approach targeting ITIH1 in the liver is needed to provide support for our proposed mechanism. Second, despite our attempt to examine the possible role of O-GlcNAcylation on Gα13 function, the basis underlying how the hyperglycemic condition represses Gα13 needs to be investigated further. Third, the candidate GPCR/corresponding ligands responsible for the observed signaling and functional pathways were not identified here. Another approach is needed to identify the candidate GPCR ligands and study them in depth (e.g., high-throughput ligand screening). Fourth, a detailed experimental approach is necessary to understand the biology of ITIH1 and hyaluronan in insulin desensitization. Last, ITIH1 concentrations in patients with impaired glucose tolerance or type 2 diabetes were quite high (23 to 166 μg/ml), which would make it difficult to use a neutralizing antibody to reduce circulating ITIH1 in clinical situations due to the high antibody dosage that would be needed. Thus, other therapeutic options would be needed to efficiently accomplish the tight management of blood ITIH1 concentrations, such as siRNA/shRNA-based gene therapy. Nevertheless, our findings suggest that ITIH1 should be investigated further as a target for developing new treatments for diabetes.

MATERIALS AND METHODS

Study design

The objective of the study was to determine the effects of hepatic Gα13 on whole-body glucose metabolism in mice. We aimed to elucidate the underlying mechanisms of Gα13’s potential effects using primary hepatocytes and G13 LKO mice. We used lentiviral or hydrodynamic injection–mediated in vivo delivery of genes encoding Gα13 or O-GlcNAc transferase to mice with diabetes induced by feeding the mice a high-fat diet, gavaging them with glucose, or treating them with streptozotocin. We validated the clinical relevance of our preclinical findings in two independent cohorts of human subjects, who had impaired glucose tolerance or overt diabetes induced by NAFLD (cohort #1) or chronic hepatitis C virus infection (cohort #2). Secretome analysis using LC-MS/MS technology was conducted to identify putative targets responsible for the development of systemic insulin resistance in G13 LKO mice. Overall, glucose tolerance in the mice was analyzed by performing a glucose tolerance test, an insulin tolerance test, and fasting blood glucose/insulin measurements. The study was extended to examine antibody neutralization of ITIH1 in mice to assess the potential of ITIH1 as a therapeutic target. For in vivo experiments, age-matched mice were randomly allocated to different groups, but the experimenters were not blinded. Only mice in poor health or with an insufficient metabolic response to diabetic insults were excluded from data analysis (one or two mice per group for Figs. 7F and 8E).

Animals

Animal experiments were conducted under the guidelines of the Institutional Animal Use and Care Committee at Seoul National University. All animals were maintained in a 12-hour light/dark cycle and fed ad libitum. All mice used were male and had C57BL/6 background. C57BL/6, ob/ob, or db/db mice were purchased from the Jackson Laboratory (Bar Harbor, ME). Gna13flox/flox mice (a gift from S. Offermanns, Max Planck Institute, Germany) (41) were crossed with albumin-Cre transgenic mice (the Jackson Laboratory) to generate G13 LKO mice. Gna13flox/flox mice without detectable Cre gene were used as wild-type (WT) littermates. For diet-induced obesity experiments, mice at the age of 8 to 12 weeks were fed either a high-fat diet (HFD; 60% kcal fat; Research Diets D12492, New Brunswick, NJ) or a normal chow diet (ND) for 5 to 16 weeks. For an acute insulin injection experiment, mice were fasted overnight and intraperitoneally injected with insulin (2 insulin units/kg body weight; Humalog, Lilly, Indianapolis, IN) and subsequently euthanized 15 min afterward. For a fasting/refeeding transition model, mice were fasted for 24 hours, followed by refeeding for 4 hours. For a streptozotocin-induced diabetes model, 10-week-old C57BL/6 mice were intraperitoneally injected with 200 μl of either citrate buffer (pH 4.5) (Sigma-Aldrich, St. Louis, MO) or streptozotocin [50 mg/kg in citrate buffer (pH 4.5)] (Sigma-Aldrich) once a day for 5 consecutive days and were euthanized 4 weeks afterward. For an oral glucose gavage experiment, 10-week-old C57BL/6 mice were fasted overnight, were given a single bolus of glucose (2 g/kg body weight) by gavage, and euthanized at indicated times thereafter. In another set of experiments, WT or G13 LKO mice at 12 weeks of age were injected with ST045849 (TimTec, Newark, DE) at the dose of 20 mg/kg body weight, as resolved in a 2.5% DMSO/2.5% Tween 20/95% PBS solution, once a day for 4 consecutive days and euthanized 6 hours after the last treatment. Glucose tolerance test or glucose gavage experiments were performed in an identical manner, where the assays were initiated 6 hours after the last injection.

Human samples

Written informed consent was obtained from all subjects before participating in research studies. Studies using human samples were reviewed and approved by the independent Institutional Review Board of the Seoul Metropolitan Government Seoul National University Boramae Medical Center. The subject cohort included patients from two clinical studies (cohorts #1 and #2). Liver biopsies and serum/plasma samples were obtained from diabetic patients with NAFLD (cohort #1) or chronic hepatitis C virus infection (cohort #2). For subjects with biopsy-proven NAFLD (cohort #1), measurements of serum biochemical parameters, type 2 diabetes status, and histological assessment were carried out for 31 human subjects. These subjects included those with normal glucose tolerance (fasting blood glucose concentrations were lower than 110 mg/dl, n = 11), impaired glucose tolerance (fasting blood glucose concentrations ranged between 110 and 126 mg/dl, n = 10), or overt type 2 diabetes (fasting blood glucose concentrations were more than 126 mg/dl, n = 11). Together, these subjects had a broad range of values on the HOMA-IR test (ranging from 0.49 to 22.01). None of the human subjects were taking antidiabetic medications. Three or five subjects in each group (n = 3 for normal glucose tolerance and impaired glucose tolerance; n = 5 for type 2 diabetes) were taking antihyperlipidemic medications (statins) at the time the measurements were taken (see table S1 for the characteristics of these individuals). For subjects with chronic hepatitis C virus infection (cohort #2), serum biochemical parameters and glucose tolerance status were assessed in 25 subjects (13 males, 12 females) with a broad range of HOMA-IR values (ranging from 1.3 to 21.0). The patients were subdivided and analyzed according to the HOMA-IR index. Subjects in cohort #2 had an average age of 57.0 ± 2.0 years, with a body mass index (BMI) of 24.8 ± 0.8 kg/m2 and with a fasting blood glucose concentration of 119.4 ± 8.7 mg/dl.

Glucose tolerance test and insulin tolerance test

For glucose tolerance tests (GTT), mice were orally administered with glucose (1.5 to 2 g/kg body weight) after overnight fasting, and blood was drawn to measure glucose concentrations at indicated times thereafter. For insulin tolerance tests (ITT), insulin (0.75 to 1.5 insulin units/kg body weight; Humalog, Lilly) was intraperitoneally injected into mice after fasting for 6 hours, and blood was drawn to measure blood glucose at indicated times. Insulin concentrations were measured at each indicated time in serum collected from the mice subjected to GTT.

Glucose production assays in primary hepatocytes

The glucose concentration was determined using the Amplex Red Glucose/Glucose Oxidase Assay Kit (Invitrogen) according to the manufacturer’s instructions. The cells were lysed, and the protein concentration was determined to normalize the values for glucose production.

Conditioned media preparation from primary hepatocytes

Cultured primary hepatocytes from mice fed an HFD for 5 weeks were washed with PBS and incubated in serum-free Opti-MEM medium (Gibco BRL). Conditioned media (CM) were collected after 24 hours and centrifuged at 3000g for 5 min, and the supernatant excluding debris was stored at −80°C until use. For secretome analysis, CM was depleted of abundant serum proteins (e.g., albumin and immunoglobulin) using commercially available immunodepletion resin (MIDR002-020, R&D Systems), followed by concentration at 4800g centrifugation for 90 min at 4°C with Amicon Ultra centrifugal filter units (10 kDa MWCO; Millipore, Burlington, MA).

Sample preparation for proteomics of hepatocyte CM samples

For liquid chromatography–tandem mass spectrometry (LC-MS/MS) analysis of CM samples from primary hepatocytes, protein concentration of CM was measured using Quick Start Bradford 1 × Dye Reagent (Bio-Rad Laboratories, Hercules, CA). Then, a fraction of proteins (100 μg) was prepared in 50 mM ammonium bicarbonate and was reduced and alkylated by treatment with dithiothreitol (Bio-Rad Laboratories) and iodoacetamide (Sigma-Aldrich, St. Louis, MO). Trypsin (Promega, Madison, WI) was added to digest samples at a protein-to-enzyme ratio of 50:1 (w/w), and the solution was incubated at 37°C for 16 hours. Digested samples were separated into 12 fractions using high pH on a C18 column as the first dimension.

ITIH1-enriched sample preparation for proteomics of O-GlcNAcylation

HEK293A cells overexpressed with human ITIH1 (Origene, Rockville, MD, USA) were subjected to SDS-PAGE. ITIH1 bands were cut from the SDS-PAGE gel for LC-MS/MS analysis of ITIH1 O-GlcNAcylation. Briefly, a fraction of proteins prepared in 50 mM ammonium bicarbonate was successively reduced and alkylated by treatment with dithiothreitol (Sigma-Aldrich) and iodoacetamide (Sigma-Aldrich). In-gel digestion was conducted by incubation with trypsin at 37°C overnight, and the sample was subsequently resolved by online reversed-phase chromatography using a C18 column as a cleanup procedure.

LC-MS/MS analysis for hepatocyte secretome

Spectra raw data were acquired on a linear trap quadrupole (LTQ)–Orbitrap (Thermo Fisher, San Jose, CA) with EASY-nLC II (Thermo Fisher Scientific). An autosampler was used to load 6-μl aliquots of the peptide solutions into an EASY-Column; C18 Trap column with an inner diameter of 100 μm, a length of 20 mm, and a particle size of 5 μm (Thermo Fisher Scientific). The peptides were desalted and concentrated on the trap column for 15 min at a flow rate of 2 μl/min. Then, the trapped peptides were separated on an EASY-Column; C18 analytic column with an inner diameter of 75 μm, a length of 100 mm, and a particle size of 3 μm (120 Å; from Thermo Fisher Scientific). The mobile phases were composed of 100% water (A) and 100% acetonitrile (ACN) (B), and each contained 0.1% formic acid. The voltage applied to produce the electrospray was 2.0 kV. During the chromatographic separation, the LTQ-Orbitrap was operated in a data-dependent acquisition mode. The MS data were acquired using the following parameters: full scans were acquired in Orbitrap at a resolution of 60,000 for each MS/MS measurement, and six data-dependent collision-induced dissociation (CID) MS/MS scans were acquired in LTQ with 10-ms activation time performed for each sample, 35% normalized collision energy (NCE) in CID, and ±1.5 Da isolation window. Previously fragmented ions were excluded for 180 s. Then, the datasets generated by LTQ-Orbitrap were analyzed using the Proteome Discoverer (version 1.3.0.339, Thermo Fisher Scientific) and Scaffold (version 4.4.1, Proteome Software Inc., Portland, OR) Platform and searched against the UniProt mouse protein database (release 2014_06) using SEQUEST and X!tandem. Peptide identifications were accepted if they could be established at greater than 90% probability by the Scaffold Local FDR algorithm. Protein identifications were accepted if they could be established at a greater than 95% probability. These were also accepted if they contained at least two identified unique peptides. Protein probabilities were assigned by the Protein Prophet algorithm.

Secreted proteins were defined as either the presence of N-terminal signal peptide or the annotation in UniProt by Gene Ontology cellular compartment (GO:0005615 or GO:0005576) or the keyword for cellular components (secreted). Liver-enriched proteins were considered as those annotated as “tissue-enriched in liver” or “liver-specific” in UniProt and/or Human Protein Atlas.

ITIH1-neutralizing antibody treatment

The rabbit polyclonal antibody against mouse ITIH1 was produced by immunization with a KLH-conjugated synthetic peptide predicted from mouse ITIH1 sequence (C-DKAREVAFDVE). The antibody was then purified from the serum of final bleed through IgG purification. Preimmune IgG (pre-IgG) was similarly purified and used as a control. Mice were daily injected with purified antibody or pre-IgG (diluted in sterile PBS, i.p., 250 μg each) for the last 2 weeks during the diet feeding.

Statistical analyses

Values are expressed as means ± SEM. Statistical significance was tested using a two-tailed Student’s t test or one-way ANOVA with Bonferroni or least significant difference (LSD) multiple comparisons where appropriate. Differences were considered significant at P < 0.05.

SUPPLEMENTARY MATERIALS

stm.sciencemag.org/cgi/content/full/11/513/eaan4735/DC1

Materials and Methods

Fig. S1. Decrease in hepatic Gα13 in patients with diabetes.

Fig. S2. Correlations between hepatic Gα13 and lipid profiles in patients with diabetes.

Fig. S3. A metabolic and liver function profile in G13 LKO mice fed a high-fat diet.

Fig. S4. Impaired glucose homeostasis and insulin resistance in G13 LKO mice fed a normal diet.

Fig. S5. Correlations between serum ITIH1 concentrations and lipid profiles in patients with NAFLD.

Fig. S6. ITIH1 and ITIH2 concentrations in the liver of patients with diabetes or G13 LKO mice fed a high-fat diet.

Fig. S7. Immunoblotting for extracellular matrix and cytoplasmic protein markers.

Fig. S8. Immunofluorescence staining for ITIH1-hyaluronan complexes in vitro.

Fig. S9. Increase in O-GlcNAC transferase–mediated O-GlcNAcylation and ITIH1 concentrations after loss of Gα13.

Fig. S10. LC-MS/MS analysis of O-GlcNAc modification of human ITIH1.

Fig. S11. O-GlcNAC transferase–mediated stabilization of ITIH1 through suppression of proteasomal degradation.

Fig. S12. The effect of ITIH1 silencing on insulin sensitivity in vitro and effects of anti-ITIH1 antibody treatment of mice in vivo.

Fig. S13. The effect of a high to low glucose transition on O-GlcNAcylation and Gα13 concentrations in mice in vivo.

Table S1. Characteristics of subjects with NAFLD in cohort #1.

Data file S1.

References (4251)

REFERENCES AND NOTES

Acknowledgments: We thank S. Offermanns for providing the Gna13flox/flox mice. Funding: S.G.K. was supported by the National Research Foundation (NRF) funded by the Korean government (Ministry of Science, ICT and Future Planning) (no. 2017K1A1A2004511). W.K. was supported by the Korea Health Technology R&D Project through the Korea Health Industry Development Institute funded by the Ministry of Health & Welfare, Republic of Korea (no. HI17C0912). C.S.C. was supported by a grant from the Bio and Medical Technology Development Program of the NRF, funded by the MSIP (no. 2014M3A9D5A01073886). T.H.K. was supported by the Basic Science Research Program of the Ministry of Education (NRF no. 2016R1A6A3A01011043 and NRF no. 2018R1A6A3A11048112). This study was supported, in part, by a fund from the Korea Institute of Oriental Medicine (no. K16820). Author contributions: T.H.K. and S.G.K. conceived the project, designed overall in vivo and in vitro experiments, and wrote the manuscript. T.H.K. performed most of the experiments and data analysis with contributions from J.H.K., M.J.H., and C.Y.H., who performed glucose and insulin tolerance tests and virus and hydrodynamic tail vein injection experiments. S.-Y.P. and C.S.C. performed exploratory experiments for metabolic phenotyping of animals and assisted in the analysis and interpretation of data. T.H.K. and M.J.H. performed primary hepatocyte isolation and preparation of conditioned media samples for LC-MS/MS analysis. Y.-I. K. and J.-Y. C. performed and analyzed LC-MS/MS for secretome studies. I.J.C. and C.H.L. generated adenoviral Gα13 active mutant. J.W.L. provided input on O-GlcNAc transferase experiments and constructs. W.K. collected, analyzed, and provided human samples. S.G.K. supervised the overall study and provided funding. All authors discussed the results and commented on the manuscript. Competing interests: S.G.K. is a consultant for Pharmaking Co. and Wooshin Labotachi Co. in Korea. W.K. consults and lectures for Gilead, Boehringer-Ingelheim, Eisai, Samil, Ildong, LG Chemistry, CJ Healthcare-Kolmar, GreenCross, BuKwang, and Standigm and receives research support from Pfizer and Roche. C.S.C. has stock options in MD Healthcare. The other authors declare that they have no competing interests. S.G.K., T.H.K., and J.-Y.C. are co-inventors on patent no. 10-2019-0034314 entitled “Method of screening nucleic acid-based material targeting ITIH1 for treating disease related to hyperglycemia” and patent no. 10-2019-0042917 entitled “Method of screening small molecular material targeting ITIH1 for treating disease related to hyperglycemia.” S.G.K., T.H.K., and W.K. are co-inventors on patent no. 10-2019-0109433 entitled “Use of ITIH1 for detecting insulin resistance in disease associated with glucose intolerance” and patent no. 10-2019-0109892 entitled entitled “Pharmaceutical composition comprising antibody specifically binding to ITIH1 for improving insulin resistance.” Data and materials availability: All data associated with this study are in the paper or the Supplementary Materials. G13 LKO mice with liver-specific ablation of Gna13 are available through an MTA from Sang Geon Kim.
View Abstract

Stay Connected to Science Translational Medicine

Navigate This Article