Tissue-Specific Differences in the Development of Insulin Resistance in a Mouse Model for Type 1 Diabetes

Insulin resistance not only is a feature of obesity and type 2 diabetes mellitus (T2DM) but also can be present in patients with type 1 diabetes mellitus (T1DM) (1,2), resulting in a condition termed “double diabetes.” T1DM is characterized by absolute insulin deficiency resulting from autoimmune-mediated β-cell destruction (3), leading to uncontrolled glucose homeostasis. Although insulin resistance in T1DM may occur independently, the accelerator hypothesis postulates that it can be triggered by autoimmune diabetes itself (4). The causes and mechanisms linking T1DM and insulin resistance remain unknown.

Prolonged hyperglycemia may induce insulin resistance by mechanisms summarized as glucotoxicity (5). On the other hand, insulin deficiency also decreases the suppression of lipolysis in adipose tissue and liver, thereby raising circulating triglyceride (TG) and fatty acid (FA) levels, which in turn impair insulin-stimulated glucose transport by various mechanisms summarized as lipotoxicity (6,7). Both the diacylglycerol (DAG)/protein kinase C (PKC) isoforms pathway (6) and toll-like receptor 4 (TLR4)/fetuin A-mediated ceramide synthesis (8) have been shown to inhibit insulin signaling.

Furthermore, glucotoxicity and lipotoxicity strongly associate with lower mitochondrial function. In muscle, lower mitochondrial activity is frequently associated with insulin resistance in humans with or at risk for T2DM (9,10). Insulin deprivation can also reduce muscle mitochondrial ATP production by altered expression of mitochondrial genes (11), and insulin-resistant patients with long-standing T1DM may have lower ATP synthase flux (1). Studies have shown that reduced hepatic energy metabolism also correlates with insulin resistance, hyperglycemia, and altered hepatic lipid content in T2DM (12,13). However, others questioned the role of mitochondrial function in insulin resistance (14). The contribution of mitochondrial function to insulin resistance has not yet been clarified in T1DM.

We hypothesized that the onset of T1DM is associated with impaired insulin sensitivity and lower mitochondrial function. Although glucolipotoxicity may cause insulin resistance in the liver, muscle insulin resistance may develop independent of ambient glycemia. To test this hypothesis, we examined diabetes-related effects in female nonobese diabetic (NOD) mice independently of other confounding factors, such as sex, obesity, aging, and inherited mitochondrial abnormalities.

NOD and wild-type (WT) C57BL/6 mice were maintained under specific pathogen-free conditions on a 12-h light-dark cycle and received a standard rodent diet (ssniff M-Z Extrudate, 4.5% fat; ssniff Spezialdiäten GmbH, Soest, Germany) and water ad libitum. We studied only female mice because 80% of female but only 30% of male NOD develop diabetes within 30 weeks of age (15). Experiments were performed 3 days (acute diabetic NOD [A-DM]) or 8 weeks after onset of diabetes (insulin-treated chronic diabetic NOD [C-DM]) in randomly assigned mice. Nondiabetic NOD (N-DM) mice were matched for age and body weight. Mice were studied in the fed state and after 6 h of fasting. Insulin-treated C-DM were examined only in the fed state to avoid hypoglycemia. After decapitation, trunk blood and tissues were collected, weighed, and used for high-resolution respirometry or snap frozen in liquid nitrogen. All experiments were performed according to the guidelines for the care and use of animals (GV-SOLAS [Society for Laboratory Animal Science]) and approved by the local council of animal care in line with the requirements of the German animal protection act.

Diabetes was detected by testing for glycosuria (Diabur-test 5000; Roche Diagnostics, Mannheim, Germany) and confirmed by tail blood glucose measurements >250 mg/dL on 2 consecutive days (Precision Xtra Plus; Abbott, Wiesbaden, Germany). Insulin was administered subcutaneously in LinBit pellets (LinShin Canada, Toronto, ON, Canada) for 8 weeks, aiming to maintain blood glucose between 250 and 500 mg/dL. At glucose levels >500 mg/dL for >1 week, an additional insulin pellet was administered.

Mice were individually housed and placed in an eight-chamber indirect calorimetry system (PhenoMaster; TSE Systems, Bad Homburg, Germany). After 24 h of acclimatization, energy expenditure, respiratory quotient (RQ), physical activity, and food and water intake were simultaneously analyzed for 48 h.

A silicon catheter (Silastic Laboratory Tubing, Dow Corning, Midland, MI) was placed into the right-side jugular vein under Isofluran (CP Pharma, Burgdorf, Germany) anesthesia. Mice were allowed to recover for 4–5 days and fasted for 6 h on the day of the experiment (3:00–9:00 a.m.). To assess basal whole-body glucose disposal, d-[6,6-²H₂]glucose (98% enriched) (Cambridge Isotope Laboratories, Andover, MA) was infused at a rate of 4 μmol/kg/min for 120 min. The hyperinsulinemic-euglycemic clamp was performed with a primed (40 mU/kg), continuous infusion (4 mU/kg/min) (Huminsulin; Lilly, Giessen, Germany) for 180 min. Euglycemia was maintained by periodically adjusting a variable 20% glucose infusion. d-[6,6-²H₂]glucose was coinfused together with insulin solution (0.4 μmol/kg/min) and variable glucose infusion to obtain stable tracer concentrations during varying glucose infusion rates. Blood samples were taken at 10-min intervals during the last 30 min of basal and hyperinsulinemic-euglycemic clamps. Additional clamps were performed under identical conditions except for measuring FA levels at time points −15, 0, 5, 15, 30, 60, 90, 120, and 150 min to assess FA suppression and injecting 10 μCi of 2-deoxy-d-[1-¹⁴C]glucose at the end of the clamp to calculate rates of insulin-stimulated glucose uptake by gastrocnemius and soleus muscles (16). After the clamps, mice were exsanguinated through cervical incision and killed by cervical dislocation, and serum and organs were collected for analyses.

Ex vivo mitochondrial function was measured in fresh liver and gastrocnemius muscle using the Oxygraph-2k (Oroboros Instruments, Austria) as described (17). Defined respiratory states were obtained by the following protocol: 2 mmol/L malate, 10 mmol/L pyruvate, 10 mmol/L glutamate and 2.5 mmol/L ADP (state 3, complex I [CI]), 10 mmol/L succinate (state 3, CI + complex II [CII]), 10 μmol/L cytochrome c (mitochondrial membrane integrity check), carbonyl cyanide-p-trifluoromethoxyphenylhydrazone (stepwise increments of 0.25 μmol/L up to the final concentration of maximum 1.25 μmol/L, state u), and 2.5 μmol/L antimycin A. Addition of cytochrome c did not increase oxygen consumption, indicating integrity of the outer mitochondrial membrane after saponin permeabilization.

Tissue DNA was extracted with a Qiagen Tissue and Blood Kit (Qiagen, Hilden, Germany), and concentration was measured spectrophotometrically (GeneQuant; GE Healthcare, Munich, Germany). Mitochondrial DNA (mtDNA) copy number was quantified with RT-PCR (ABI Prism 7000; Applied Biosystems, Darmstadt, Germany) by using specific primers and 5′ FAM + 3′ TAMRA–labeled probes (Eurogentec, Liège, Belgium) for NADH dehydrogenase subunit 1 (forward primer: 5′-CTA-CAA-CCA-TTT-GCA-GAC-GC-3′, reverse primer 5′-GGA-ACT-CAT-AGA-CTT-AAT-GC-3′, probe 5′-CCA-ATA-CGC-CCT-TTA-ACA-ACC-TC-3′) and for lipoprotein lipase (forward primer: 5′-GGT-TTG-GAT-CCA-GCT-GGG-CC-3′, reverse primer 5′-GAT-TCC-AAT-ACT-TCG-ACC-AGG-3′, probe 5′-CTT-TGA-GTA-TGC-AGA-AGC-CC-3′). NADH dehydrogenase subunit 1 and lipoprotein lipase DNA copy numbers were determined by comparison with log-linear standard curves. The ratio of mtDNA to nuclear DNA is a measure of tissue concentration of mtDNA per cell.

Serum insulin (Mouse Insulin Elisa Kit; Mercodia, Uppsala, Sweden) and fetuin A (Mouse Fetuin A/AHSG DuoSet; R&D Systems, Abingdon, U.K.) were measured by ELISA. d-[6,6-²H₂]glucose enrichment in deproteinized plasma was quantified with gas chromatography mass spectrometry (Agilent Technologies, Waldbronn, Germany) after derivatization of glucose to pentaacetate. Serum TG, cholesterol (Roche/Hitachi, Roche Diagnostics, Mannheim, Germany), FA, and β-hydroxybutyrate (Wako Chemicals GmbH, Neuss, Germany) levels were assessed photometrically.

Tissue concentrations of thiobarbituric acid reactive substances (TBARS) were measured fluorometrically (BioTek, Bad Friedrichshall, Germany) in 10 mg of frozen gastrocnemius muscle and liver (18). Protein content in homogenates was assessed with the bicinchoninic acid assay (BCA Protein Assay Kit; Thermo Fisher Scientific, Bonn, Germany).

Mice were injected intraperitoneally with insulin 1 unit/kg body weight or saline (control) and killed after 10 min, and liver and muscle tissues were snap frozen in liquid nitrogen.

Total RNA extraction, cDNA synthesis, and real-time quantitative RT-PCR expression analyses were performed as described (19). Data of gene-specific probes (Assays-on-Demand; Applied Biosystems, Darmstadt, Germany) were normalized to 18S rRNA content (19).

Proteins were extracted by homogenization and centrifugation (21,000g for 15 min at 4°C) and measured by the BCA Protein Assay Kit (Thermo Fisher Scientific). Membrane and cytosol fractions were prepared by using differential centrifugation (20).

Ten micrograms of protein were loaded onto SDS-PAGE gels and transferred to polyvinylidene difluoride membranes (Merck Millipore, Schwalbach, Germany). Membranes were blocked with 5% milk in PBS and incubated with antibodies recognizing total Akt isoforms 1 and 2, extracellular signal-regulated kinase phosphorylated at Thr202 and Tyr204 (pERK-Thr202/Tyr204), c-Jun N-terminal kinase phosphorylated at Thr183 and Tyr185 (pJNK-Thr183/Tyr185), nuclear factor κ-light-chain-enhancer of activated B cells phosphorylated at Ser536 (pNFκB-Ser536), p38 mitogen-activated protein kinase phosphorylated at Thr180 and Tyr182 (p-p38-Thr180/Tyr182), Ser1101 phosphorylation of insulin receptor substrate 1 (IRS1) (pIRS1-Ser1101) and Ser307 phosphorylation of IRS1 (pIRS1-Ser307) (Cell Signaling Technology, Danvers, MA), GLUT4 (Abcam, Cambridge, U.K.), PKCε and PKCθ (BD Biosciences, Franklin Lakes, NJ), and the insulin receptor β-subunit (Santa Cruz Biotechnology, Santa Cruz, CA). Horseradish peroxidase–conjugated secondary antibodies (Promega GmbH, Mannheim, Germany) and enhanced chemiluminescence (Merck Millipore) were used for detection. Immunoblots were quantified using VersaDoc 4000 MP (Bio-Rad) and Quantity One version 4.6.9 (Bio-Rad) software. Stripped membranes were reprobed with α-tubulin (Calbiochem, Darmstadt, Germany) and GAPDH (Abcam) antibodies to allow for loading corrections.

Oil red O staining was performed in fresh right-side liver lobes fixed in 4% paraformaldehyde (21). Stained sections were examined by light microscopy at ×40 magnification (Leica DMRB; Leica Microsystems, Wetzlar, Germany). Liver and muscle samples were solubilized in ethanolic KOH and glycerolipids measured as described (22) by using a TG standard (Lyonorm Calibrator; PLIVA-Lachema Diagnostika, Brno, Czech Republic).

Membrane and cytosol fractions were extracted by homogenization and centrifugation (20) and DAG and ceramide levels measured by liquid chromatography-mass spectrometry (23).

Pancreatic tissue was embedded in paraffin, stained by hematoxylin-eosin, and examined with light microsopy (24). For β-cell and α-cell detection, double immunofluorescence staining was performed by using anti-insulin and antiglucagon antibodies (Dako, Hamburg, Germany), respectively. Images were acquired using inverted microscope (Leica Microsystems) and digital camera (Olympus Europa, Hamburg, Germany).

Glucose disposal (Rd) was calculated by Steele single-pool non–steady-state equations (25). Endogenous glucose production (EGP) is given as the difference between Rd and glucose infusion rate.

Surrogate indexes of insulin sensitivity and secretion were calculated from fasting blood glucose (G) and serum insulin (I) levels as follows: QUICKI = 1 / [log(I in μU/mL) + log(G in mg/dL)]) (26) and HOMA-B = [20 × (I in μU/mL)] / [(G in mmol/L) – 3.5] (27).

Data are presented as mean ± SD in the text and tables, and as mean ± SEM in figures. Groups were compared by Bonferroni test or nonparametric unpaired t test (Mann-Whitney) with Hochberg post hoc analysis. P < 0.05 is considered to indicate statistically significant differences.

Body weight was similar among A-DM, N-DM, and C-DM mice and slightly higher than that of WT mice (Table 1). In the fed state, blood glucose was fourfold higher in A-DM and C-DM versus N-DM and WT (Supplementary Table 1). Insulin levels were lower in A-DM than N-DM (Table 1). During fasting, circulating glucose, TG, and FA levels were increased in A-DM compared with N-DM and WT (Fig. 1A–C). QUICKI was 13% and 9% lower in A-DM and N-DM, respectively, than in WT, suggesting glucose-independent insulin resistance in the NOD model (Table 1). HOMA-B was 3.5- and 5.5-fold higher in N-DM than in WT and A-DM, respectively, fitting with the previous observation that prediabetic NOD mice have higher in vivo and ex vivo glucose-stimulated insulin secretion than C57BL/6 mice (28). A-DM had lower HOMA-B and insulin levels in the fed state than N-DM, illustrating the impaired β-cell function resulting from the severe insulitis (data not shown) typical for the NOD model (24).

A-DM had substantially (60–75%) lower visceral and subcutaneous fat masses (Supplementary Fig. 1A and B). Muscle TG level was unchanged except for a small increase in ad libitum–fed N-DM (Supplementary Fig. 1C). Liver weight was comparable between groups, whereas hepatic TG level was lower in fasted A-DM (Supplementary Figs. 1A–D and 2).

Only WT and N-DM displayed normal diurnal energy expenditure patterns (Fig. 2A and B). A-DM showed a shift in RQ to 0.81 with preferential lipid oxidation (Fig. 2C and D), which was supported by a trend toward higher serum β-hydroxybutyrate levels (0.62 ± 0.22 mmol/L) versus N-DM (0.25 ± 0.07 mmol/L; P = 0.083). Finally, A-DM exhibited lower physical activity during the dark cycle (Fig. 2E) and increased overall food and water intake (Fig. 2F and G). Energy expenditure was similar among groups (Fig. 2H).

During the hyperinsulinemic-euglycemic clamps, steady-state blood glucose (Fig. 3A) and serum insulin (not shown) levels were comparable. A-DM had higher fasting EGP versus WT and N-DM (Fig. 3B). In A-DM, EGP was positively related to fasting glycemia (r = 0.95, P < 0.05, n = 9). Both A-DM and N-DM displayed lower insulin-mediated suppression of EGP than did WT (Fig. 3C), indicating hepatic insulin resistance. A-DM showed impaired insulin-stimulated Rd, which was 62% and 66% reduced compared with WT and N-DM (Fig. 3D). Uptake of 2-deoxyglucose by gastrocnemius and soleus muscles was also lower in A-DM (Fig. 3E and F). In A-DM, plasma FA level decreased similarly to WT and N-DM during the first 15 min but remained higher from 30 min to the end of the clamp (Fig. 2G). Overall percent suppression of FA in A-DM was comparable to N-DM (Fig. 2H), suggesting intact insulin sensitivity of adipose tissue.

In the fed state, CI respiration decreased by 38% in C-DM (Fig. 4A). During fasting, CI + CII and maximal electron transport system (ETS) capacities decreased by ∼20% in A-DM and N-DM (Fig. 4B). Results were normalized to mtDNA copy number, which did not differ between groups (not shown). Differences in mitochondrial respiration did not associate with changes in gene expression of peroxisome proliferator–activated receptor γ coactivator 1-α (PGC-1α), mitochondrial transcription factor A (TFAM), or nuclear respiratory factor-1 (NRF-1) (Fig. 4C and D).

Measurement of TBARS in the muscle revealed that WT mice, unlike N-DM or A-DM, have lower lipid peroxide levels in the fed than in the fasting state (P < 0.01) (Supplementary Fig. 1E and F). Moreover, fed N-DM, A-DM, and C-DM mice had higher TBARS than WT mice (Supplementary Fig. 1E).

A-DM displayed reduced basal levels of insulin receptor (IR) (−32%) and IRS1 (−30%) in muscle compared with WT (Fig. 5A and B). Serine pIRS1-Ser1101 was increased by 149% and pIRS1-Ser307 by 246% upon intraperitoneal insulin injection (Fig. 5C and D). A-DM showed a profoundly decreased insulin-stimulated membrane-to-cytosol ratio of Akt (−61%) (Fig. 5E) and GLUT4 (−76%) (Fig. 5F). N-DM had reduced IR and intermediate decreases in Akt and GLUT4 membrane-to-cytosol ratio (Fig. 5A, E, and F).

Levels of pERK-Thr202/Tyr204, p-p38-Thr180/Tyr182, phosphorylated inhibitor of κB (IκB) at Ser32 (pIκB-Ser32) (Supplementary Fig. 3), and pNFκB-Ser536 (Fig. 5I) were unchanged. pJNK-Thr183/Tyr186 was increased in A-DM and N-DM (Fig. 5H). Only the higher membrane-to-cytosol ratio of PKCθ was clearly associated with decreases in peripheral insulin sensitivity in A-DM (+148% vs. WT, +53% vs. N-DM) (Fig. 5G).

A-DM had higher membrane-to-cytosol DAGs (Fig. 6A) as a result of changes in several specific DAG species (Fig. 6B). Neither total nor individual species of ceramides were different between the groups (Fig. 6C and D).

In the fed state, hepatic O₂ flux rates through CI, CI + CII, and maximal ETS capacity were increased in A-DM compared with N-DM and WT (Fig. 4E). In C-DM, CI and CI + CII respiration was increased versus WT but lower versus A-DM (Fig. 4E). No differences were observed in the fasted state (Fig. 4F). Hepatic mtDNA copy number was similar among all groups (data not shown). In contrast to muscle, increased respiration in the liver of A-DM was accompanied by higher transcription of PGC-1α TFAM and NRF-1 than in N-DM and WT (Fig. 4G).

Hepatic TBARS were unchanged in the fasted state but were higher in fed N-DM and A-DM (Supplementary Fig. 1E and F) and even more pronounced in C-DM. Similar to muscle, hepatic TBARS decreased during feeding in WT only (P < 0.001).

Protein abundance of hepatic IR, IRS2 (Fig. 5J and K), and IRS1 (data not shown) were similar in all groups. However, A-DM had 125% and 74% higher levels of pIRS1-Ser1101 and pIRS1-Ser307, respectively, than WT at baseline (Fig. 5L and M). N-DM mice also showed 137% higher pIRS1-Ser1101 (Fig. 5L). Insulin-stimulated translocation of Akt to the membrane was suppressed by 76% in A-DM and to a lesser extent (by 40%) in N-DM (Fig. 5N).

Levels of pERK-Thr202/Tyr204, p-p38-Thr180/Tyr182, IκB, pIκB-Ser32 (Supplementary Fig. 4), and pNFκB-Ser536 (Fig. 5R) were unchanged. Of note, the membrane-to-cytosol ratio of PKCε was 36% and 27% lower in A-DM and N-DM, respectively (Fig. 5O). This was exclusively attributed to an increase in PKCε in the cytosolic fraction (Fig. 5P). Furthermore, pJNK-Thr183/Tyr186 was increased by 64% in A-DM and by 67% in N-DM (Fig. 5Q), rendering pJNK the only factor associated with changes in hepatic insulin sensitivity.

In A-DM and N-DM, fasting serum concentrations of fetuin A were 2.0- and 2.9-fold higher than in WT (Fig. 1D).

This study describes the early metabolic events occurring before and at the onset of insulin-dependent diabetes in NOD mice. The NOD model spontaneously develops autoimmune insulitis with insulin deficiency, resembling human T1DM. Because chemical (29,30) or surgical (31) induction of insulin deficiency may per se induce insulin resistance, NOD mice currently represent the most suitable model for studying metabolic changes associated with autoimmune diabetes. The study shows that the initiation of diabetes is associated with tissue-specific differences in metabolic flexibility and insulin sensitivity. Normoglycemic, nondiabetic NOD mice already exhibit hepatic insulin resistance associated with increased JNK phosphorylation and lipid peroxidation as well as lower muscle glucose transport. Acutely diabetic NOD mice also display muscle insulin resistance associated with increased intramyocellular DAG and PKCθ activation. Moreover, the liver transiently enhances its mitochondrial oxidative capacity at diabetes onset, possibly as an adaptation to increased lipolysis, whereas muscle oxidative capacity declines during later stages of diabetes.

A-DM mice show a reduction of visceral and subcutaneous fat, which reflects excessive fasting lipolysis and/or impaired stimulation of lipid synthesis by insulin in adipocytes. Blunted inhibition of lipolysis has been linked to the development of peripheral insulin resistance in adolescents with poorly controlled T1DM (32). In the current study, the intact reduction of FA during hyperinsulinemic clamps in NOD mice, unlike in humans, indicates that the absence of insulin-dependent control of lipid metabolism results from reduced insulin secretion rather than from insulin resistance of adipose tissue. The resulting rise in circulating FA and TG levels would favor redistribution of lipids toward ectopic fat storage in liver or muscle, which is associated with insulin resistance in these tissues (7,13). Of note, A-DM mice had unchanged muscle and even reduced liver TG content, indicating permanently ongoing lipid oxidation in the fed state. This was supported by a constant reduction and lack of the diurnal oscillation of RQ, reflecting restriction to fat oxidation, impaired switching to glucose utilization, and a trend toward ketonemia within 3 days after onset of hyperglycemia in A-DM. Likewise, T1DM patients with poor metabolic control have lower intrahepatic fat content along with moderate peripheral insulin resistance and increased whole-body lipid oxidation (2). On the other hand, N-DM mice had increased muscle fat content, likely reflecting higher lipid storage leading to unchanged circulating FA and TG levels. Both A-DM and N-DM mice also accumulated lipid peroxidation products in muscle and liver, indicating greater reactive oxygen species (ROS) production, which can further aggravate insulin resistance (33). Taken together, NOD mice are characterized by hyperlipidemia and oxidative stress in muscle and liver.

A-DM mice display markedly reduced muscle insulin sensitivity as assessed both in vivo and ex vivo. Impaired insulin-mediated glucose disposal and muscle glucose uptake, along with decreased membrane translocation of GLUT4 and Akt, could result from lipid-mediated intracellular alterations. In human T1DM, muscle insulin resistance is accompanied by decreased glucose transport into myocytes (34,35) and insulin-stimulated upregulation of GLUT4 mRNA (36) but increased muscle lipid content (37). FA-induced inhibition of insulin-stimulated glucose transport/phosphorylation (38) has been linked to intracellular accumulation of DAG, PKCθ activation, and serine phosphorylation of IRS1 (6). Supporting this contention, A-DM mice showed increased membrane-to-cytosol ratio of DAG and PKCθ, and Ser1101/Ser307 phosphorylation of IRS1 in skeletal muscle (Fig. 7). Moreover, the hyperglycemia-induced basal glucose uptake in A-DM mice could also contribute to PKC isoforms activation (40). On the other hand, plasma levels of fetuin A, an FA-induced hepatokine, were increased in A-DM and N-DM mice. Fetuin A positively correlates with hyperglycemia and insulin resistance during hyperlipidemia (41) and promotes lipid-induced insulin resistance by mediating the binding of FA to TLR4 (39). We found no changes in signaling downstream of TLR4, such as NFκB or ceramides, questioning the contribution of this pathway to muscle insulin resistance. Recent data also indicate that both high-saturated and high-unsaturated fat diets induce insulin resistance independently of TLR4 signaling (42).

In vivo hepatic insulin sensitivity is impaired in both N-DM and A-DM mice, implying that hepatic insulin resistance can develop before the onset of hyperglycemia. In the absence of hyperglycemia and hyperlipidemia, the mechanism of hepatic insulin resistance in N-DM mice likely differ from that observed in muscle of A-DM. Indeed, inhibition of IRS1 in the liver of N-DM and A-DM was not associated with changes in PKCε but, rather, with an increase in pJNK. Although previous studies yielded conflicting data on the protective effects of liver-specific JNK deficiency against lipid-induced insulin resistance, it should be emphasized that the enhanced phosphorylation of JNK in the NOD model occurs independently of lipids. Alternatively, increased fetuin A (41) and hepatic oxidative stress (43), which have been linked to JNK activation, could account for increased pJNK, inhibition of IRS1, and hepatic insulin resistance.

We found no association of mitochondrial respiration with insulin sensitivity, excluding mitochondrial function as a potential mediator of insulin resistance in NOD mice. In muscle, mitochondrial oxidative capacity and transcript levels of mitochondrial biogenesis–related genes were comparable between A-DM and N-DM. However, C-DM showed lower oxidative capacity, in line with the observation that muscle ATP synthesis is decreased in vivo in human T1DM (1). This secondary decrease of mitochondrial function could result from mitochondrial damage caused by lipid peroxide accumulation (44), inhibitory effects of prolonged insulin treatment on mitochondrial biogenesis and function (45), or direct effects of the chronic diabetic state (1). In contrast, hepatic mitochondrial oxidative capacity as well as biogenesis were clearly higher in A-DM mice. Increased oxidative capacity may represent an early adaptation to increased lipid and glucose flux after diabetes onset. Expression of genes related to oxidative phosphorylation is upregulated in the liver of NOD mice 2 weeks after diabetes onset (46) as well as in the liver of obese humans with T2DM (47). Decreased Akt could mediate the increase in hepatic PGC-1α, FA oxidation (48), and overall mitochondrial respiration (49). Of note, upregulated hepatic mitochondrial oxidative capacity seems to be transient because it declines in C-DM. Similar to muscle, this could be a consequence of mitochondrial damage due to lipid peroxide accumulation.

In conclusion, insulin resistance in a murine nonobese model of T1DM develops in the liver before the onset of diabetes and is associated with increased oxidative stress and pJNK as possible cellular mediators. Muscle insulin sensitivity is impaired already 3 days after the onset of diabetes and accompanied by increased levels of DAG, PKCθ, and serum fetuin A. Furthermore, mitochondrial oxidative capacity is transiently enhanced in the liver and declines in muscle with longer disease duration. Knowledge of these mechanisms could help to develop new strategies to prevent and treat insulin resistance and subsequent complications in patients with T1DM.

Acknowledgments. The authors thank Mario Kahn of the Department of Internal Medicine, Yale University School of Medicine; Kay Jeruschke, Daniella Herzfeld de Wiza, Heidi Müller, and Conny Köllmer of the Institute for Clinical Biochemistry and Pathobiochemistry, German Diabetes Center, Düsseldorf; and Volker Burkart, Alexander Strom, Fariba Zivehe, Ilka Rokitta, and Olesja Ritter of the Institute for Clinical Diabetology, German Diabetes Center, Düsseldorf, for their expertise.

Funding. This study was supported in part by Schmutzler Stiftung, Skröder Stiftung, the German Research Foundation (SFB575, project 13, to M.R.), the German Federal Ministry of Education and Research to the German Center for Diabetes Research, the Ministry of Science and Research of the State of North Rhine-Westphalia, the German Federal Ministry of Health, and grants from the U.S. Public Health Service (R01-DK-40936, P30-DK-45735).

Duality of Interest. This study was supported in part by grants from the European Foundation for the Study of Diabetes (Novo Nordisk and GlaxoSmithKline grants). No other potential conflicts of interest relevant to this article were reported.

Author Contributions. T.J. and G.S. conceived of the experiments, researched data, contributed to the discussion, and wrote the manuscript. K.K., D.M.O., E.P., J.K., and B.K. researched data and edited and reviewed the manuscript. J.W., A.L.R., L.J., P.N., H.-J.P., and D.Z. researched data. G.I.S. contributed to the discussion and edited and reviewed the manuscript. J.S. conceived of the experiments, contributed to the discussion, and edited and reviewed the manuscript. M.R. conceived of the experiments, contributed to the discussion, and wrote the manuscript. M.R. 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.

Prior Presentation. Parts of this study were presented in abstract form at the 72nd Scientific Sessions of the American Diabetes Association, Philadelphia, PA, 8–12 June, 2012; and at the 48th Annual Meeting of the European Association for the Study of Diabetes, Berlin, Germany, 1–5 October 2012.