Insulin Is a Stronger Inducer of Insulin Resistance than Hyperglycemia in Mice with Type 1 Diabetes Mellitus (T1DM)*

Subjects with type 1 diabetes mellitus (T1DM) eventually develop insulin resistance and other features of T2DM such as cardiovascular disorders. The exact mechanism has been not been completely understood. In this study, we tested the hypothesis that excessive or inappropriate exposure to insulin is a primary mediator of insulin resistance in T1DM. We found that continuous exposure of mice with non-obese diabetes to insulin detemir, which is similar to some current conventional treatment of human T1DM, induced severe insulin resistance, whereas untreated hyperglycemia for the same amount of time (2 weeks) did not cause obvious insulin resistance. Insulin resistance was accompanied by decreased mitochondrial production as evaluated by mitochondrial DNA and levels of transcripts and proteins of mitochondrion-associated genes, increased ectopic fat accumulation in liver and skeletal muscle (gastrocnemius) evaluated by measurements of triglyceride content, and elevated oxidative stress detected by the GSH/GSSG ratio. Prolonged exposure of cultured hepatocytes to insulin induced significant insulin resistance, whereas the same length of exposure to a high level of glucose (33 mm) did not cause obvious insulin resistance. Furthermore, our results showed that prolonged exposure to insulin caused oxidative stress, and blockade of mitochondrion-derived oxidative stress by overexpression of manganese-superoxide dismutase prevented insulin resistance induced by the prolonged exposure to insulin. Together, our results show that excessive exposure to insulin is a primary inducer of insulin resistance in T1DM in mice.

Since the beginning of the insulin era, most patients with T1DM 2 have been able to live almost normally. However, as early as the 1940s, long before the concept of insulin resistance came to light, it was noticed that the application of insulin in T1DM was associated with the development of cardiovascular disorders (1)(2)(3)(4)(5). Since the 1960s, it has been established that the application of insulin in T1DM always leads to insulin resistance (6 -10). Mechanisms associated with cardiovascular disorders and insulin resistance in patients with T1DM have been studied extensively but have not been completely understood.
It has been shown previously that a high level of glucose and its byproducts such as glucosamine might play a critical role in the development of insulin resistance (11)(12)(13)(14)(15)(16)(17)(18)(19)(20)(21). However, it is unclear whether the effects of glucose and glucosamine on the development of insulin resistance can occur in the absolute absence of insulin. It is known that insulin can desensitize insulin signaling through activation of ERK1/2 MAPKs and/or Akt/ S6K (9,22). It is unclear whether this scenario actually represents physiology and pathophysiology. However, it is clear that ectopic fat accumulation in liver and skeletal muscles, increased oxidative stress, and decreased mitochondrial capacity/biogenesis are clustered together with insulin resistance. It has been shown that no matter how severe the obesity is, insulin resistance usually does not develop in the absence of ectopic fat accumulation (23). In contrast, when ectopic fat accumulation occurs, insulin resistance may ensue even in slim animals (24). Thus, ectopic fat accumulation seems to be critical for the induction of insulin resistance. Ectopic fat accumulation has been shown to be correlated with insulin resistance in T1DM (25). The role of oxidative stress in the development of insulin resistance is also definitive. The development of insulin resistance can be prevented in both cultured cells and animals when mitochondrion-derived ROS production is prevented (26 -28). A decreased mitochondrial number is a cardinal feature of insulin resistance/hyperinsulinemia (29 -31). The ratio between the mitochondrion-rich (type I) muscle fibers and glycolytic (type II) muscle fibers is decreased in subjects with insulin resistance/ hyperinsulinemia (32,33). The mitochondrial DNA copy number is decreased in subjects with insulin resistance/hyperinsulinemia (34). Suppression of mitochondrial biogenesis by antiretroviral nucleoside analogues is associated with the development of insulin resistance/hyperinsulinemia in patients with AIDS (35). Importantly, ectopic fat accumulation, oxidative stress, and decreased mitochondrial capacity can not only coexist but also promote each other. Insulin can certainly elevate fat accumulation, including ectopic fat accumulation, because a basic function of insulin is to promote lipogenesis while inhibiting lipolysis and fat oxidation. Mitochondrial production/biogenesis is always increased when levels of plasma insulin and Akt-dependent insulin signaling is low (36 -40), implying that insulin inhibits mitochondrial production directly or indirectly. We have recently shown that prolonged exposure to insulin indeed suppresses mitochondrial production in isolated hepatocytes and mice (41,42). Furthermore, insulin and PI 3-kinase signaling have been shown to stimulate peroxisome-derived production of ROS (43)(44)(45)(46), and ROS itself can stimulate mitochondrion-derived ROS production (47). Therefore, in this study, we have tested the hypothesis that inappropriate exposure to insulin plays a primary role in the development of insulin resistance in T1DM.

MATERIALS AND METHODS
Reagents-The triglyceride assay kit (TR0100) and antisera against ␤-actin were from Sigma. Antibodies against total and phospho-Akt at Ser 473 , IRS-1(pSer 636/639 ) and PI 3-kinases (p55pTyr 199 ) were from Cell Signaling Technology (Danvers, MA). Antibodies against TFAM were from Santa Cruz Biotechnology (catalog No. CS-23588). Antibodies against Mn-SOD and total/phospho-IRS-1 at Tyr 416 were from Abcam. Insulin detemir was from Novo Nordisk. GSH/GSSG-412 TM assay kit was from Bioxytech (Foster City, CA). Blood glucose concentrations were measured by using a Breese 2 glucose meter (Bayer HealthCare). The protein assay kit was from Bio-Rad. Other materials were all obtained commercially and were of analytical quality.
Animal Experiments-Animals were housed under the usual day (12 h daylight) and night (12 h darkness) circadian rhythm and fed ad libitum. NOD/ShiLtj mice were purchased from The Jackson Laboratory (Bar Harbor, ME). When fasting blood glucose level reached ϳ300 mg/dl, NOD mice were treated with either detemir or the vehicle solution (saline, 100 l) via subcutaneous injections once every 12 h for 2 weeks. Euglycemia was reached and maintained for at least 2 days in each animal. Detemir doses varied among animals to achieve euglycemia. Application of detemir was similar to the so called current conventional treatment of human T1DM (48). All animal studies were approved by the Institutional Animal Care and Use Committee of The Hamner Institutes for Health Sciences and fully complied with the guidelines from the National Institutes of Health.
Cells-Mouse primary hepatocytes were isolated from C57BL/6 mice that were fed with the regular chow diet and cultured in Williams' medium E supplemented with 10% fetal bovine serum as described previously (49 -55). Mouse Hepa1c1c7 hepatoma cells were maintained in Dulbecco's modified Eagle's medium (DMEM, Invitrogen) supplemented with 10% fetal bovine serum.
Measurement of Mitochondrial DNA-Total DNA was extracted from tissues or cultured cells using Qiagen DNA extraction kit. DNA concentrations were determined using a by NanoDrop TM 1000 spectrophotometer (Thermo Scientific, Wilmington, DE). One ng of total DNA was used to determine the ratio of mitochondrial cyclooxygenase (COXII) to the nuclear intron of ␤-globin by real-time PCR using oligos listed in supplemental Table 1 (56). Data were confirmed by measuring the ratio of cytochrome b to glucagon genomic DNA of the same sample.
Insulin Tolerance Test (ITT)-ITT was performed after an overnight fast or 14 h after the last dose of detemir. Initial blood glucose levels were determined followed by intraperitoneal injection of human insulin (catalog No. I9278, Sigma) (0.75 unit/kg). Blood glucose levels were measured via tail vein blood at 15,30,60,90, and 120 min after the injection.
Measurement of Glutathione (GSH)/Glutathione Disulfide (GSSG) Ratio-Levels of GSH and GSSG in cell or tissue lysates were determined with a kit from OXIS International, Inc. (Foster City, CA) and normalized to protein levels. The GSH/GSSG ratio was calculated according to the instructions provided by the manufacturer.
RNA Extraction and Real-time PCR-Total RNAs were extracted from cells or tissues with an RNeasy mini kit (Qiagen) and reverse-transcribed into cDNAs, which were quantified by TaqMan real-time PCR with specific probes and primers from Applied Biosciences and normalized to levels of ␤-actin.
Statistical Analysis-Data are presented as mean Ϯ S.E. Data were compared by Student's t test using GraphPad Prism version 5.0 for Windows (San Diego, CA). Differences at values of p Ͻ 0.05 were considered significant.

Treatment of NOD T1DM Mice with the Insulin Reagent Detemir Induces Insulin Resistance, whereas Hyperglycemia Does
Not Obviously Affect Insulin Sensitivity-Both hyperglycemia and treatment with insulin have been considered as inducers of insulin resistance in subjects with T1DM (9,57,58). To determine which of these two inducers is a stronger inducer of insulin resistance, a group of NOD diabetic mice with hyperglycemia (over 300 mg/dl) was treated with saline for 2 weeks, and another group of diabetic NOD mice was treated with two doses of the long acting insulin reagent detemir, with one dose in the early morning and another in the evening. Given that detemir is long acting, insulin was basically present in the blood at all times. Doses of detemir were gradually increased to achieve euglycemia. Detemir treatment started when fasting blood glucose levels reached ϳ300 mg/dl. The application of detemir in this study was similar to the current conventional treatment of human T1DM (48). NOD sibling mice of the same ages, with no diabetes, were used as negative controls. To achieve euglycemia, the dose of detemir was gradually increased to as high as 20 units/kg of body weight (Fig. 1A). As shown in Fig. 1B, blood glucose levels increased gradually and reached as high as 600 mg/dl at the end of the 2-week period in diabetic NOD mice treated with saline. Blood glucose levels in diabetic NOD mice treated with detemir decreased gradually and reached euglycemia during the last 2 days of the 2-week treatment when the dosage of detemir reached ϳ20 units/kg of body weight (Fig. 1B). To determine the levels of insulin sensitivity in different groups of mice, ITT was performed. As shown in Fig. 1, C and D, the NOD diabetic mice treated with detemir for 2 weeks showed no response to acute challenge with the regular fast acting insulin, whereas the NOD diabetic mice with hyperglycemia for 2 weeks without detemir treatment were equally as responsive to insulin challenge as the NOD mice without diabetes. Body weights for all groups were not significantly different (Fig. 1E). Intake levels of food and water in diabetic NOD mice tended to increase in comparison with the NOD mice with no diabetes (p ϭ 0.06 and 0.08, respectively) but were similar between the NOD diabetic mice treated with either saline or detemir (Fig. 1, F and G).
The changes in intake of food and water and body weight were not as dramatic as usually thought. There are several probable explanations. First, glucose reabsorption via the Na(ϩ)glucose transporter-2 (SGLT-2) in the kidney is increased when insulin is deficient (59), leading to milder loss of glucose and water as predicted. Second, the NOD mice used in this study were based on A/J mice, which are resistant to body weight change even under the high fat diet (60). Third, treatment with insulin was not sufficient until the last 2 days (Fig. 1, A and B). Thus, changes in the average body weight were not expected to be very significant. Taken together, these results show that: (a) continuous exposure to insulin for 2 weeks is a strong inducer of insulin resistance, whereas hyperglycemia for 2 weeks does not cause obvious insulin resistance in mice; and (b) no matter how strong the insulin resistance, blood glucose level can be brought to a normal level as long as sufficient amount of insulin is provided.
To determine levels of basal insulin signaling in different groups of animals, phosphorylation levels of p55 subunit of PI 3-kinase and Akt were measured in liver and gastrocnemius. Phosphorylation of both the p55 PI 3-kinase subunit and Akt was elevated by treatment with detemir in liver (Fig. 2, A and B).
PI 3-kinase activity was also increased by detemir (Fig. 2C). Similarly, phosphorylation of both the p55 PI 3-kinase subunit and Akt and PI 3-kinase activity were stimulated by detemir in gastrocnemius. (Fig. 2, D and E). These results show that the basal insulin signaling is increased in diabetic mice treated with detemir (insulin).
To examine the mechanism by which treatment with detemir caused insulin resistance, we examined levels of IRS-1 phosphorylation at serines 636 and 639. As shown in Fig. 3, IRS-1 phosphorylation at serines 636 and 639 was increased in both liver and gastrocnemius of detemir-treated mice. These results imply that prolonged exposure to insulin induces insulin resistance, likely through IRS-1 serine phosphorylation. Note that these results do not suggest that appropriate application of the right type of insulin reagents for the appropriate amount of time via a physiological route cause insulin resistance.
Treatment of NOD mice with T1DM by Detemir Decreases Mitochondrial Production-To further examine the mechanism by which insulin induces insulin resistance, mtDNA and transcripts of some mitochondrion-related genes were quantified. As shown in Fig. 4A, levels of mtDNA tended to increase in both liver and gastrocnemius of NOD diabetic mice without detemir treatment (p Ͼ 0.05) but were decreased in liver (p Ͻ 0.05) and tended to decrease in gastrocnemius (p Ͼ 0.05) of NOD diabetic mice that were treated with detemir. Transcript levels of several mitochondrion-related genes including ATP synthase, estrogen receptor-related receptor ␣ (ERR␣), NADH dehydrogenase (Ndufv1), and mitochondrial transcription factor A (Tfam) were significantly decreased by continuous exposure to detemir in comparison with the NOD diabetic mice without insulin treatment in liver but not in gastrocnemius ( Table 1). The levels of many other mitochondrion-related gene transcripts were also decreased by insulin treatment without reaching statistical significance in liver. Protein levels of the key mitochondrial transcription factor TFAM was increased in the liver of diabetic NOD mice, but the increase was reversed by treatment with detemir (Fig. 4B). TFAM protein levels were also increased in the gastrocnemius of diabetic NOD mice, and this increase tended to be reversed by treatment with detemir although without reaching statistical significance (Fig. 4C). Together, these results suggest that mitochondrial production is increased in the absence of insulin in NOD diabetic mice but is decreased by treatment with continuous exposure to insulin (detemir).
Treatment of NOD Mice with T1DM by Detemir Increases Ectopic Fat Accumulation-As ectopic fat accumulation is a necessary component of insulin resistance (23,24), we evaluated the effect of detemir treatment on the levels of ectopic fat accumulation. As shown in Fig. 4D, in comparison with NOD mice without diabetes, the triglyceride content tended to decrease in both the liver and gastrocnemius of NOD diabetic mice with no detemir treatment (p Ͼ 0.05) but was increased significantly by treatment with detemir. Transcript levels of key fat oxidation genes including carnitine palmitoyltransferase 1␣ (CPT-1␣), medium chain acyl-CoA dehydrogenase (Acadm), long chain acyl-CoA dehydrogenase (Acadl), and very long chain acyl-CoA dehydrogenase (Acadvl) tended to decrease (p Ͼ 0.05), whereas those of key lipogenic genes such as sterol regulatory element-binding protein-1 (SREBP-1), SREBP-2, fatty acid synthase (FAS), and HMG-CoA reductase (Hmgcr) all tended to increase in the presence of continuous exposure to insulin in both liver and gastrocnemius although without reaching statistical significance (Table 2). These results indicate that the fat oxidation program is inhibited, whereas the lipogenic program is increased by continuous exposure to insulin (detemir), resulting in elevated ectopic fat accumulation.   (26 -28). Thus, we examined the effect of insulin on levels of oxidative stress by measuring the ratio between reduced GSH and oxidized GSSG. As shown in Fig. 4E, the GSH/GSSG ratio was not altered by hyperglycemia for 2 weeks, yet it was significantly decreased by continuous exposure to insulin (detemir) in liver. The GSH/GSSG ration was not altered by either hyperglycemia or detemir treatment for 2 weeks in gastrocnemius. These results suggest that oxidative stress was increased in liver but not in skeletal FIGURE 2. Continuous exposure to insulin increases the basal level of insulin signaling. Protein levels of insulin signaling components from mice described in Fig. 1 were evaluated by immunoblotting. A, PI 3-kinase ((PI3K) p55 subunit) level in liver. B, Akt level in liver. C, PI 3-kinase-␣ purified from liver lysates was used to convert PI(4,5)P 2 into PI(3,4,5)P 3 . The level of PI 3-kinase activity needed to produce PI(3,4,5)P 3 was measured and presented as the mean Ϯ S.E. D, levels of PI 3-kinase (p55 subunit) and Akt in gastrocnemius. E, PI 3-kinase-␣ purified from gastrocnemius lysates was used to convert PI(4,5)P 2 into PI(3,4,5)P 3 . The level of PI 3-kinase activity needed to produce PI(3,4,5)P 3 was measured and presented as the mean Ϯ S.E.  Fig. 1 were evaluated by immunoblotting with specific antibodies. muscles even after continuous exposure to insulin (detemir) for 2 weeks.
Insulin Induces Insulin Resistance, but High Glucose Level Fails to Do So in Cultured Hepatocytes-To recapitulate the in vivo results described above, Hepa1c1c7 hepatoma cells were chronically exposed to insulin (10 nM) for 72 h in the presence of either a normal or a high level of glucose. During this period of incubation, culture media were changed twice a day with fresh media supplemented with insulin. Cells were then thoroughly washed with warm phosphate-buffered saline to remove any residual insulin before they were acutely treated with fresh media with insulin (1 or 10 nM) for 5 min as noted. Levels of insulin signaling were evaluated by immunoblotting. As shown in Fig. 5, phosphorylation levels of IRS-1 at Tyr 416 , the p55 subunit of PI 3-kinase, and Akt at both Ser 308 and Ser 473 were stimulated by acute treatment with insulin in the presence of both normal and high levels of glucose. It is noteworthy that incubation of cells with a high level of glucose did not alter the capability of cells in response to acute treatment with insulin, suggesting that exposure to a high level of glucose itself is not an inducer of insulin resistance. In contrast, chronic exposure of hepatocytes to insulin significantly blunted their response to acute insulin treatment in terms of phosphorylation levels of IRS-1 at Tyr 416 , the p55 subunit of PI 3-kinase, and Akt at both Ser 308 and Ser 473 . Together these results demonstrate that chronic exposure of cultured hepatocytes to insulin induces insulin resistance, whereas exposure to a high glucose level for the same length of time does not cause obvious insulin resistance.
Prolonged Exposure of Primary Hepatocytes Induces Oxidative Stress-It is well established that oxidative stress plays a critical/necessary role in the development of insulin resistance in both cultured cells and animals. For example, the development of insulin resistance induced by either TNF-␣ or dexamethasome is prevented by scavenging the mitochondrion-derived ROS in cultured cells (26). High fat-dietinduced insulin resistance is prevented when HFD-induced oxidative stress is blocked by deletion of the apoptosis induction factor (27). Insulin resistance in ob/ob mice is reversed by scavenging mitochondrion-derived ROS (28). To determine the role of oxidative stress in insulin-induced insulin resistance, the effect of insulin on oxidative stress in isolated hepatocytes was examined. As shown in Fig. 6A, the GSH/ GSSG ratio was decreased in isolated hepatocytes by incubation with insulin for 24 h. Similarly, the GSH/GSSG ratio was decreased by H 2 O 2 (positive control). Transcript levels of several oxidation-responsive genes including superoxide dismutase 1 (SOD1), Hmox1, and glutamine-cysteine ligase (Gclc) were also increased by treatment with insulin (Fig.  6B). These results suggest that prolonged exposure to insulin causes oxidative stress in hepatocytes.

JOURNAL OF BIOLOGICAL CHEMISTRY 27095
Oxidative Stress Is Required for Insulin Induction of Insulin Resistance in Isolated Hepatocytes-To determine the role of oxidative stress in insulin-induced insulin resistance, mitochondrion-derived ROS was scavenged by overexpression of Mn-SOD prior to prolonged exposure to insulin (36 h) (Fig.  7A). After a prolonged exposure to insulin (24 h), Akt phosphorylation stimulated by acute insulin challenge was significant in the presence of Mn-SOD overexpression (Fig. 7A, lane 4) equal to that in the positive control (lane 5) (Fig. 7A). Conversely, Akt phosphorylation in response to acute insulin stimulation was blunted in the absence of Mn-SOD after a prolonged exposure to insulin (Fig. 7A, lane 9). Next, we measured the effect of Mn-SOD on the expression of key gluconeogenic genes in the presence of prolonged exposure to insulin. In the absence of Mn-SOD overexpression, insulin suppression of key gluconeogenic genes including glucose-6-phosphatase (G6Pase) and phosphoenoylpyruvate carboxykinase (PEPCK) was significantly blunted by prolonged exposure to insulin (Fig. 7B). However, in the presence of Mn-SOD overexpression, insulin suppression of glucose-6-phosphatase and phosphoenoylpyruvate carboxykinase was not affected by prolonged incubation with insulin. These results suggest that mitochondrion-derived ROS plays a critical role in the development of insulin resistance caused by prolonged exposure to insulin.

DISCUSSION
T1DM, caused by a deficiency in insulin production, was incurable before the insulin era. The discovery and application of insulin have made it possible to control hyperglycemia in T1DM and have vastly prolonged the life span and improved the quality of life in patients with T1DM. Nevertheless, the application of insulin in T1DM has inadvertently led to cardiovascular disorders and insulin resistance (1)(2)(3)(4)(5)9). The exact   mechanism by which insulin causes cardiovascular disorders and insulin resistance in T1DM has not been completely understood. Our results from this study indicate that excessive or inappropriate exposure to insulin is a primary inducer of insulin resistance in T1DM. Plasma levels of insulin normally fluctuate with several factors. First, insulin secretion is stimulated by nutrients and thus is increased after each food intake but returns to a low basal level when all nutrients leave the blood. Second, there is a circadian rhythm in insulin secretion from pancreatic ␤-cells with the lowest level of insulin occurring between midnight and 6:00 a.m. and reaching a peak between noon and 6:00 p.m. in a given day. Third, insulin secretion oscillates within a range of seconds to periods of 9 -14 min under both in vivo and in vitro conditions (61,62). Evolutionarily and physiologically, these fluctuations and oscillations of insulin secretion may play a critical role in storing energy when food is abundant and releasing stored energy to be consumed when food is not available. In other words, fluctuations and oscillations of the plasma insulin level are necessary for both efficient storage and consumption of energy. Disruption of these physiological variations of insulin secretion may promote energy storage while inhibiting efficient energy consumption. Indeed, it has been noticed that fluctuations, circadian rhythm, and oscillations of insulin secretion are lost in subjects with chronic hyperinsulinemia, a loss that is closely associated with accelerated fat accumulation (obesity) (63). In addition, physiological fluctuations, circadian rhythm, and oscillations of plasma insulin level may be necessary for maintaining appropriate cellular growth. It is very reasonable that insulin promotes cellular growth when nutrients are available such as after food intake, and cellular growth should be stopped when nutrients are scarce such as during fasting. Actually, cellular molecules and organelles are even consumed for an alternative fuel source during starvation through autophagy, which is not only important in providing an alternative fuel source but also necessary for maintaining cellular health such as prevention of cancers (64,65). In other words, loss of physiological fluctuations and oscillations of insulin secretion may lead to undesired cell growth such as cancers. Indeed, it has been established that the prevalence of cancers is increased in subjects with overweight/obesity and hyperinsulinemia (66).
In this study we treated NOD mice with T1DM by continuous presence of insulin (detemir) in a way that resembles some current treatment of human T1DM (48). Obviously, this regimen takes away all physiological fluctuations, circadian rhythm, and oscillations of plasma insulin levels and is bound to have inadvertent effects. As expected, the basal Akt-dependent classical insulin signaling and ectopic fat accumulation are increased in liver and skeletal muscle. Ectopic fat accumulation plays a critical role in a high fat diet or obesity-induced insulin resistance (23,24). The ectopically accumulated fat and its intermediate catabolites such as FFA, diacylglycerol, and ceramide can induce insulin resistance through various signaling pathways including MAPK (c-Jun NH 2 -terminal kinase (JNK), p38, and ERK1/2), protein kinases C, IB kinases, S6 kinase (S6K), and endoplasmic reticulum stress (67)(68)(69)(70)(71). Additionally, FFA can induce insulin resistance via production of mitochondrion-derived ROS (mtROS), which is known to activate all the above described signaling pathways involved in insulin resistance (72). FFA can promote mtROS production in at least three possible ways. First, ectopically accumulated FFA may overload mitochondria with acetyl-CoA and NADH leading to increased proton gradient and mitochondrial membrane potential, which ultimately elevates mtROS production (73,74). Second, long chain acyl-CoA activated from FFA blocks the import of ADP (a necessary substrate for ATP synthesis) through the adenine nucleotide translocator (Ant) into the mitochondrial matrix and consequently increases mtROS production (75)(76)(77)(78)(79). Third, when fat accumulation reaches a certain level, the secretion of inflammatory cytokines such as TNF-␣ is increased in the blood. TNF-␣ is a known inhibitor of Ant (80). Thus, accumulated fat can promote mtROS production indirectly via TNF-␣. TNF-␣ happens to be a potent stimulator of inflammation, which is widely considered by some as a primary player of insulin resistance (81). In regard to TNF-␣, we believe that the elevated plasma level of TNF-␣ is initially a consequence of excessive fat accumulation. This statement is based on several known facts. First, the secretion of TNF-␣ and other cytokines from  OCTOBER 2, 2009 • VOLUME 284 • NUMBER 40 adipose tissue occurs usually after fat accumulation has reached a certain level. Second, TNF-␣ is lipolytic (82), i.e. the presence of TNF-␣ is more likely meant to melt away excessively accumulated fat. Third, TNF-␣ is known to be a strong stimulator of mitochondrial production and thermogenesis (83), indicating that the presence of TNF-␣ is originally meant to burn off the fat released from the adipose storage. Fourth, TNF-␣ is an established inducer of insulin resistance via oxidative stress as described above. It is reasonable to assume that the presence of TNF-␣ is meant to prevent further fat accumulation that is driven by insulin. In the continuous presence of excessive fat accumulation caused by inappropriate exposure to insulin or overeating and/or physical inactivity, TNF-␣ may cause excessive inflammation while working hard to rid the body of excessively accumulated fat and prevent further fat accumulation. Additionally, it should be emphasized that, mtROS recently has been shown to be critical/necessary for the development of insulin resistance in both cultured cells and animals (26 -28). Moreover, increased ectopic fat accumulation without oxidative stress does not lead to insulin resistance (84,85). In this study we have observed that inappropriate or prolonged exposure to insulin in both mice and cultured hepatocytes causes increased oxidative stress.

Insulin Induction of Insulin Resistance in T1DM
The primary source of ROS is mitochondria, and mitochondrial dysfunction has been linked to insulin resistance. We and others have shown that mitochondrial mass/production is decreased in the presence of insulin resistance and hyperinsulinemia (29 -35, 41) but is increased when levels of plasma insulin and Akt-dependent insulin signaling are low (36,37,39). It has also been shown that mitochondrial production is increased in subjects with frank diabetes, which have an absolute or relative insufficiency of insulin production (86). Mitochondrial production is increased in adipose tissue when the insulin receptor is fat-specifically knocked out (40). Furthermore, insulin has been shown to inhibit the function of PGC-1␣, a potent stimulator of mitochondrial production, through blockade of PGC-1␣ gene transcription or inactivation of the PGC-1␣ protein (87,88). These observations imply that insulin can inhibit mitochondrial production. In this study, we have found that continuous treatment of NOD diabetic mice decreases mtDNA and transcript levels of some mitochondrion-associated genes more obviously in the liver than in skeletal muscle gastrocnemius. Although the transcript levels of the primary transcription factor (nuclear respiratory factor 1 (NRF-1)) and co-activator (PGC-1␣) in the regulation of mitochondrial production are not decreased, those of Tfam are significantly reduced by the continuous exposure to insulin. TFAM protein level was also decreased by the continuous exposure to insulin. Tfam is downstream of NRF-1 and PGC-1␣ and directly controls the sole promoter of mitochondrial DNA (89). Also, Tfam is necessary for the replication of mtDNA (89). Therefore, it appears that insulin can inhibit mitochondrial production through suppression of Tfam expression probably independently of the expression levels of NRF-1 and PGC-1␣. In a recent publication and another manuscript, our results show that prolonged exposure of isolated hepatocytes to insulin suppresses mitochondrial production and function in an Akt-dependent manner and increased basal Akt-dependent insulin signaling in mice may be responsible for the insulin resistance induced by a high fat diet (41,42). The reduction of mitochondrial mass/ capacity by excessive exposure to insulin is obviously more vulnerable to be overloaded by nutrients and may lead to increased mtROS production. It should be noted that there is still some debate about the role of mitochondrial mass in the development of insulin resistance. For example, insulin resistance can occur in the presence of increased mitochondrial mass (90). Likewise, decreased mitochondrial mass does not necessarily lead to insulin resistance (91). It is the level of mitochondrionderived ROS production but not mitochondrial mass itself that appears to be the linchpin between mitochondria and insulin resistance (92,93).
Additionally, we made another important observation, that hyperglycemia is not an inducer of insulin resistance under the conditions described in this study. This observation appears to contradict the popular perception that hyperglycemia plays a primary important role in the development of insulin resistance in T1DM. For example, it has previously been shown that the normalization of blood glucose in rats with T1DM by preventing glucose reapportion from kidneys with phlorizin reverses insulin resistance in adipocytes (20,21). However, the normalization of blood glucose level does not improve other insulinmediated functions such as glycogen synthesis in skeletal muscles (94). In cultured cells (adipocytes and skeletal muscle cells), a high level of glucose can only induce insulin resistance in the co-presence of insulin (15)(16)(17)(18)(19). This is consistent with our results that a high level of glucose alone does not cause insulin resistance in cultured hepatocytes. Thus, the partial recovery of insulin sensitivity by normalization of the blood glucose level using phlorizin in animals (described above) might not be caused directly by a reduction of blood glucose level. Furthermore, a high level of glucose may lead to increased production of glucosamine via the hexosamine pathway. Glucosamine infusion in rats can induce severe insulin resistance in skeletal muscles (14). Transgenic overexpression of the rate-limiting enzyme glutamine-fructose-6-phosphate aminotransferase of the hexosamine pathway in mice markedly reduces the wholebody glucose disposal (13). A similar effect of glucosamine has also been observed in cultured cells (myotubes and adipocytes) (11,12). These reports strongly imply that glucosamine plays an important role in the induction of insulin resistance. However, none of these studies has proved that glucosamine induces insulin resistance in the absolute absence of insulin, which is always present in the blood of animals (endogenous or exogenous from insulin infusion during insulin clamp studies) or in the sera of the culture media. Besides, our observation is consistent with other earlier studies showing that hyperglycemia and hyperlipidemia do not cause as much trouble as atherosclerosis, which usually follows insulin resistance/hyperinsulinemia, in the absence of insulin (1)(2)(3)(4)(5).
Why does hyperglycemia in the absence of insulin not cause insulin resistance? Glucose has to be activated by hexokinase/ glucokinase to be utilized via glycolysis, the pentose pathway, or mitochondrial oxidation or to be stored as glycogen. It has been established that expression of the hexokinase/glucokinase gene is insulin-dependent (95)(96)(97)(98). Thus, it is conceivable that glucose may not be activated due to lack of hexokinase/glucoki-nase in the absence of insulin. As a result, in the absence of insulin, glucose cannot cause either NADPH dehydrogenase (pentose pathway)-or mitochondrion-derived oxidative stress, which is critical/necessary for the development of insulin resistance as detailed above. Hyperglycemia, however, over a long period of time can certainly cause trouble via its physical (hyperosmolarity) or chemical features but probably not through insulin resistance.
In summary, we have observed that continuous exposure to insulin, which is similar to some current conventional treatment of human T1DM, leads to a cluster of insulin resistance, ectopic fat accumulation, and oxidative stress. Prolonged exposure of isolated hepatocytes to insulin causes oxidative stress and insulin resistance; blockade of mitochondrion-derived oxidative stress prevents development of the insulin resistance induced by excessive exposure to insulin. Together, our results suggest that the inappropriate application of insulin in T1DM is a primary inducer of insulin resistance in mice.