Increased Mitochondrial Oxidative Phosphorylation in the Liver Is Associated With Obesity and Insulin Resistance

Authors

  • David A. Buchner,

    Corresponding author
    1. Department of Genetics, Case Western Reserve University School of Medicine, Cleveland, Ohio, USA
    Search for more papers by this author
  • Soha N. Yazbek,

    1. Department of Genetics, Case Western Reserve University School of Medicine, Cleveland, Ohio, USA
    Search for more papers by this author
  • Paola Solinas,

    1. Center for Mitochondrial Diseases, Departments of Pharmacology and Medicine, Case Western Reserve University School of Medicine, Cleveland, Ohio, USA
    Search for more papers by this author
  • Lindsay C. Burrage,

    1. Department of Genetics, Case Western Reserve University School of Medicine, Cleveland, Ohio, USA
    2. Present address: Department of Pediatrics, University Hospitals Case Medical Center, Cleveland, Ohio, USA
    Search for more papers by this author
  • Michael G. Morgan,

    1. Department of Genetics, Case Western Reserve University School of Medicine, Cleveland, Ohio, USA
    Search for more papers by this author
  • Charles L. Hoppel,

    1. Center for Mitochondrial Diseases, Departments of Pharmacology and Medicine, Case Western Reserve University School of Medicine, Cleveland, Ohio, USA
    Search for more papers by this author
  • Joseph H. Nadeau

    Corresponding author
    1. Department of Genetics, Case Western Reserve University School of Medicine, Cleveland, Ohio, USA
    Search for more papers by this author

  • The first two authors contributed equally to this work.

(david.buchner@case.edu)

(joseph.nadeau@case.edu)

Abstract

Obesity is the result of excess energy intake relative to expenditure, however little is known about why some individuals are more prone to weight gain than others. Inbred strains of mice also vary in their susceptibility to obesity and therefore represent a valuable model to study the genetics and physiology of weight gain and its co-morbidities such as type 2 diabetes. C57BL/6J mice are susceptible to obesity and insulin resistance when fed an obesogenic diet, whereas A/J mice are resistant despite increased caloric intake. Analysis of B6- and A/J-derived chromosome substitution strains and congenic strains revealed a complex genetic and physiological basis for this phenotype. To improve our understanding of the molecular mechanisms underlying susceptibility to metabolic disease we analyzed global gene expression patterns in 6C1 and 6C2 congenic strains. 6C1 is susceptible whereas 6C2 is resistant to diet-induced obesity. In addition, we demonstrate that 6C1 is glucose intolerant and insulin resistant relative to 6C2. Pathway analysis of global gene expression patterns in muscle, adipose, and liver identified expression level differences between 6C1 and 6C2 in pathways related to basal transcription factors, endocytosis, and mitochondrial oxidative phosphorylation (OxPhos). The OxPhos expression differences were subtle but evident in each complex of the electron transport chain and were associated with a marked increase in mitochondrial oxidative capacity in the livers of the obese strain 6C1 relative to the obesity-resistant strain 6C2. These data suggests the importance of hepatic mitochondrial function in the development of obesity and insulin resistance.

Introduction

As the prevalence of obesity increases worldwide, so does the prevalence of co-morbidities such as type 2 diabetes (T2D), cardiovascular disease, and cancer. More than 150 million people worldwide are affected with T2D, whose complications include blindness, renal failure, and coronary disease. T2D occurs when β-cells in the pancreas no longer secrete sufficient amounts of insulin to overcome peripheral tissue insulin resistance and maintain euglycemia. The resulting hyperglycemia is currently controlled by lifestyle modifications or medications such as sulfonylureas or metformin which dramatically reduce morbidity. However, these therapies frequently fail to achieve optimal glycemic levels, which is particularly important to minimize the risk of microvascular complications (1).

Insulin resistance is a key factor in the pathophysiological development of T2D and remains the best indicator of future development of T2D among individuals with a family history of the disease (2). Insulin resistance occurs when target tissues such as skeletal muscle and adipose fail to properly respond to normal insulin levels, thereby requiring increasing amounts of insulin to maintain normal glucose uptake. The factors that lead to insulin resistance remain controversial, although an overabundance of adipokines, inflammatory mediators, and lipid deposits in nonadipose tissues are likely important (3). Among the controversies surrounding the etiology of insulin resistance is the question of whether mitochondrial dysfunction has a causal or compensatory role in the disease (4). Although mitochondrial dysfunction in skeletal muscle has long been the primary focus of T2D-related mitochondrial research, recent studies have demonstrated the importance of hepatic mitochondrial function and even called into question whether insulin resistance is a result of decreased or increased mitochondrial activity (5,6). Answering these questions may have important therapeutic implications for the treatment of T2D, as several well-characterized compounds exist that modulate mitochondrial function (7).

The application of mRNA profiling to understand the molecular basis of insulin resistance has proven informative (8). The global nature of these techniques provides an unbiased approach to identifying pathways that contribute to disease pathology. In this report, we utilized multi-tissue global gene expression patterns to compare two mouse strains with different susceptibilities to diet-induced obesity and insulin resistance. Our results suggest the importance of hepatic mitochondrial function in the development of insulin resistance.

Methods and Procedures

Mice

Mice were housed in ventilated racks, had access to food and water ad libitum, and were maintained at 21°C on a 12-h light/12-h dark cycle. Strain 6C1 is susceptible to diet-induced obesity and 6C2 is resistant to diet-induced obesity. These strains were generated as previously described and maintained as homozygotes by brother-sister matings (9). All mice for these studies were obtained from breeding colonies at Case Western Reserve University that were fed LabDiet 5010 chow (PMI Nutrition International, St Louis, MO). Five-week-old male mice were placed on a high-fat simple carbohydrate diet (HFSC; D12331, Research Diets, New Brunswick, NJ) as previously described (9). The HFSC diet derives 58% of its kilocalories from fat (soybean and coconut oil), 26% of its kilocalories from carbohydrate (sucrose and maltodextrin), and 16% of its kilocalories from protein (casein). The Institutional Animal Care and Use Committee approved all procedures.

RNA isolation

Liver, skeletal muscle, and gonadal fat pads were isolated following 28 days and 100 days on the HFSC diet. For each strain and time point, 4 or 5 mice from separate cages were randomly selected for analysis (excluding the heaviest and leanest mouse from each strain and time point). Prior to dissection, mice were fasted for 16 h overnight and euthanized with cervical dislocation. Liver and skeletal muscle were stored in RNA later (Ambion, Austin, TX). RNA isolations were performed using the RNeasy mini kit (Qiagen, Valencia, CA). Liver RNA was isolated using the standard protocol and skeletal muscle RNA was isolated using the fibrous tissues protocol. Gonadal fat pads were snap-frozen in liquid nitrogen and stored at −80°C. RNA isolations were performed using the RNeasy lipid tissue kit (Qiagen) following the standard protocol.

Microarray hybridization

GeneChip Mouse Genome 430 2.0 arrays (Affymetrix, Santa Clara, CA) containing 45,000 probe sets corresponding to over 39,000 transcripts and variants were used for microarray analysis. Four biological replicates were analyzed for each tissue and time point except 100-day muscle for which five samples were analyzed for each strain. Hybridization procedures were carried out at the Gene Expression and Genotyping Core of the Case Comprehensive Cancer Center in accordance with Affymetrix protocols for single round amplifications.

Microarray data analysis

The VAMPIRE web based microarray analysis suite was used under default conditions to analyze gene expression (10). The significance threshold was set at a Bonferroni corrected value of P < 0.05. Gene Set Enrichment Analysis was performed under default conditions using the curated gene sets (Version 2.5, April 2008 release) (11). Pathways were considered significantly different between the 6C1 and 6C2 congenic strains if the false discovery rate was <0.2 and the familywise-error rate was <0.2.

Quantitative PCR

RNA from muscle, liver, and white adipose tissue was reverse transcribed using Superscript II (Invitrogen, Carlsbad, CA). Quantitative polymerase chain reaction (qPCR) was performed in triplicate using the DyNAmo HS Sybr Green qPCR kit (New England Biolabs, Ipswich, MA). Genes were chosen for qPCR validation that were well annotated in the Ensembl genome browser and that were highly significantly differentially expressed. The expression level for each gene was calculated using the ΔCt method relative to β-actin (Actb). The sequences of all primers used are listed in Supplementary Table S1 online.

Glucose tolerance test and insulin secretion

Following 100 days on the HFSC diet, mice were fasted overnight for 16–18 h. Blood samples (200 µl) were collected via the retro-orbital sinus at baseline (time 0) and following an intraperitoneal injection of dextrose dissolved in water (2 g/kg body weight) after 15, 30, 60, and 120 min, respectively. Each time point represents a different cohort of 8–10 mice, except for the 0 and 120 min time points which were collected from the same mice. Glucose was measured from the retro-orbital sinus using a hand-held glucometer (OneTouch Ultra, Life Scan, Milpitas, CA). Whole blood was then centrifuged and plasma was stored at −80°C. Insulin levels were determined using an ultrasensitive mouse enzyme-linked immunosorbent assay kit (Mercodia, Uppsala, Sweden).

Insulin tolerance test

Mice were fasted 4–6 h and then given an intraperitoneal injection of recombinant human regular insulin (0.75 µ/kg body weight; Novolin R; NovoNordisk, Bagsvaerd, Denmark). Glucose was then measured using a hand-held glucometer (OneTouch Ultra, Life Scan) from a tail blood sample at time 0, 30, and 60 min postinjection.

Mitochondrial isolation and oxidative phosphorylation

Liver mitochondria were isolated and oxidative metabolism measured as described by Hoppel et al. (12). The studies were done on 10 6C1 and 10 6C2 mice.

Statistics

Measurements are presented as means ± s.e.m. Comparisons were done using an unpaired Student's t-test.

Results

Localization of Obrq2 to a 30 Mb interval on chromosome 6

The obesity-susceptible congenic strain 6C1 and the obesity-resistant congenic strain 6C2 define the adiposity QTL Obrq2 (9). The body weight and BMI of strain 6C2 are 11.4% (4.77 g) and 5.4% (0.02 g/cm2) lower than that of strain 6C1 following 100 days on the high-fat high-sucrose (HFHS) diet (9). Obrq2 was previously mapped to a 40.9 Mb interval on chromosome 6 between the microsatellite markers D6Mit138 and D6Mit223. Strains 6C1 and 6C2 have now been genotyped with additional polymorphic markers to further define the regions of B6- and A/J-derived sequence in these strains (Figure 1). Analysis of this genotype information narrowed the Obrq2 candidate interval to 30.3 Mb between the single-nucleotide polymorphism markers rs13478633 and rs30218447 and eliminated 88 genes from the candidate interval.

Figure 1.

Map of chromosome six congenic strains. Analysis of strains 6C1 and 6C2 define Obrq2, a 30.3 Mb diet-induced obesity-resistance QTL. IR, insulin resistant; IS, insulin sensitive.

6C1 mice are insulin resistant relative to 6C2

Strain 6C1 mice fed the HFSC diet for 100 days demonstrated a statistically significant ∼25% increase in fasting plasma glucose levels relative to 6C2 (9) (Figure 2a). Fasting hyperglycemia can result from reduced pancreatic insulin secretion, increased hepatic gluconeogenesis, or both. The mild but significant increase in fasting insulin levels in strain 6C1 relative to 6C2 (9) (Figure 2b) suggests that the hyperglycemia in 6C1 is due to a failure of the liver and peripheral tissues to respond to insulin and downregulate gluconeogenesis as well as increase glucose uptake, rather than a defect in insulin secretion. Furthermore, a glucose tolerance test was performed. Identical glucose loads given to 6C1 and 6C2 mice led to significantly increased levels of both insulin and glucose in strain 6C1 (Figure 2a,b), suggesting that a defect in insulin secretion does not likely account for the hyperglycemia of strain 6C1 and that 6C1 is glucose intolerant compared to 6C2. In addition, the ability to clear glucose in response to an intraperitoneal bolus injection of insulin (0.75 U/kg body weight) was also dramatically decreased in 6C1 compared to 6C2 (Figure 2c), confirming the decreased whole body sensitivity to insulin. Collectively, these results suggest that impaired peripheral tissue responsiveness to insulin and/or increased hepatic gluconeogenesis contribute to the glucose intolerance and insulin resistance exhibited by mice carrying the Obrq2B6 allele, although direct measures of insulin action are needed to confirm these hypotheses.

Figure 2.

Elevated glucose and insulin levels in strain 6C1 relative to strain 6C2. (a) Strain 6C1 had increased fasting glucose levels as well as higher glucose levels during a glucose tolerance test (GTT) relative to strain 6C2. (b) Insulin levels are increased in strain 6C1 relative to 6C2 following the injection of equivalent glucose loads during the GTT. (c) Compared to 6C2, 6C1 has a decreased ability to clear glucose from the blood stream in response to a bolus intraperitoneal injection of insulin at a rate of 0.75 U/kg of body weight. *P < 0.05, **P < 0.01, ***P < 0.0001.

Global gene expression analysis

To examine the molecular basis of obesity and insulin resistance associated with Obrq2, gene expression patterns were examined in liver, white adipose, and skeletal muscle of strain 6C1 and 6C2 mice that had been fed the HFSC diet. Gene expression was measured with microarray hybridization following both 28 and 100 days on the HFSC diet and analyzed using the VAMPIRE analysis package (10). Both the early and late gene expression time points analyzed follow phenotypic differences in body weight between strains 6C1 and 6C2. For example, even following just 28 days on the HFHS diet, strain 6C2 weighs 10.1% less than strain 6C1 (9). Therefore, changes in gene expression may be secondary to phenotypic changes. Nonetheless, a total of 568 probes corresponding to 459 unique genes were differentially expressed between the two strains (Supplementary Table S2 online). To test the accuracy of the microarray data, qPCR was performed on 10 of the most significantly differentially expressed genes from liver, muscle, and white adipose tissue following 100 days on the HFHS diet. Only one of the genes was not detected due to low levels of expression. There was a high degree of correlation (r = 0.65) between the fold change in gene expression levels for both the microarray and qPCR experiments. Additionally, 24 out of the 29 genes tested (83%) demonstrated at least a 1.5-fold change between 6C1 and 6C2 in the same direction as suggested by the microarray experiments, suggesting a high level of confidence in the microarray data (Supplementary Table S1 online).

Eight of the significantly differentially expressed genes are located within the Obrq2 interval and therefore represent potential cis-acting eQTLs and candidate genes: ankyrin repeat and suppressor of cytokine signaling box-containing 15 (Asb15), carboxypeptidase A2, pancreatic (Cpa2), RIKEN cDNA D830026I12 gene, homeodomain interacting protein kinase 2 (Hipk2), leiomodin 2 (Lmod2), RNA binding motif protein 28 (Rbm28), smoothened homolog (Smo), and tissue factor pathway inhibitor 2 (Tfpi2). These cis-eQTL genes were differentially expressed in only a single tissue at a single time point, except Asb15 which was differentially expressed in both 28 and 100-day skeletal muscle and D830026I12 which was differentially expressed in 28 and 100-day white adipose tissue. The 451 differentially expressed genes that lie outside of the Obrq2 interval represents trans-eQTLs downstream of the causal genetic variants underlying Obrq2. Several genes with known functions related to obesity and insulin resistance were identified among these trans-eQTLs including insulin-like growth factor binding protein 2 (Igfbp2), uncoupling protein 1 (Ucp1), and peroxisome proliferative activated receptor, γ, co-activator 1α (Ppargc1a or Pgc-1α). Igfbp2 was overexpressed in adipose of the obesity-resistant strain 6C2 following 100 days on the HFSC diet. Overexpression of Igfbp2 has been previously shown to protect against the development of obesity and insulin resistance by inhibiting adipogenesis (13,14). Expression of Ucp1 was increased 1,250% in muscle of the lean strain 6C2 following 100 days on the HFHS diet. Since Ucp1 is expressed in brown adipose tissue, the expression in muscle suggests the presence of more intermuscular brown fat cells in 6C2 mice. The additional brown adipose tissue would lead to a greater percentage of nutrients being used for heat generation rather than energy production as has been previously observed in obesity-resistant 129S6 mice (15). This increase in uncoupling between mitochondrial respiration and thermogenesis is consistent with the lean phenotype of 6C2 despite similar food intake levels (9). Pgc-1α is a key mediator of hepatic metabolism (16) and was overexpressed in the liver of the obese strain 6C1 following 100 days on the HFSC diet. The potential link between Pgc-1α and Obrq2 will be expanded upon in the discussion section. Expression level variation in each of these genes may contribute to the obesity and insulin resistance associated with Obrq2. However, given that 459 genes are differentially expressed between 6C1 and 6C2, we turned to pathway analysis for additional molecular insight.

Genes involved in oxidative phosphorylation are overexpressed in the liver of 6C1

In addition to examining the effect of Obrq2 on the expression levels of individual genes, the expression level of genes in pathways was also studied. Gene Set Enrichment Analysis is an algorithm that seeks to identify subtle coordinated gene expression differences that are consistently identified among a priori sets of genes that are grouped based on similar functions or expression patterns (11). Three pathways were identified as differentially expressed (false discovery rate <0.2 and P < 0.2) between 6C1 and 6C2 (Table 1). The basal transcription factors pathway was increased in 6C1 muscle following 100 days on the HFHS diet. This suggests that overall transcription levels were elevated is strain 6C1 which can occur during muscle regeneration (17). This pathway was derived from Kyoto Encyclopedia of Genes and Genomes (18) with 15 of the 29 genes overexpressed in strain 6C1. None of the genes in this pathway are within the Obrq2 interval and therefore do not genetically differ between 6C1 and 6C2. Thus, the altered expression of the basal transcription factors pathway is due to a trans-acting effect of Obrq2 or is secondary to the phenotypic differences between 6C1 and 6C2. The nucleoside diphosphate kinase (Ndk) dynamin pathway was increased in 6C2 liver following 28 days on the HFHS diet. This pathway includes genes involved in endocytosis. Interestingly, dynamin regulates endocytosis of the insulin receptor and therefore modulates insulin signaling (19). This pathway was contributed to Gene Set Enrichment Analysis by BioCarta with 9 of the 19 genes overexpressed in strain 6C2. As with the basal transcription factor pathway, none of the genes in the Ndk dynamin pathway are within the Obrq2 interval, again suggesting either a trans-acting effect of Obrq2 or a secondary phenotypic effect. The third pathway identified was the oxidative phosphorylation (OxPhos) pathway which was elevated in 6C1 liver following 100 days on the HFHS diet. The OxPhos pathway consists of 76 manually curated genes (20), of which 56 were overexpressed in 6C1 (Figure 3). The increase in gene expression levels was consistent in all five electron transport chain complexes (Figure 4). Two OxPhos genes, Ndufb2 and Ndufa5, are located on chromosome 6 within the Obrq2 interval and therefore may contain genetic differences between 6C1 and 6C2 that contribute to the Obrq2 phenotype. There are no protein coding differences in either of these genes between B6 and A/J (Ensembl release 58), although the expression of both genes are slightly increased in strain 6C1 relative to 6C2 (Ndufb2, 1.07-fold; Ndufa5, 1.14-fold) consistent with the expression changes observed in the other OxPhos genes in the pathway.

Table 1.  Pathways that are differentially expressed between 6C1 and 6C2
inline image
Figure 3.

Genes in the oxidative phosphorylation pathway are upregulated in 6C1 relative to 6C2. Each column represents one sample (chip) and each row represents one gene. The range of colors corresponds to the range of expression values (red = high, blue = low). Genes contributing to the core enrichment are shown in red.

Figure 4.

Gene expression is upregulated in the liver of strain 6C1 relative to 6C2 following 100 days on the HFSC diet in each electron transport chain complex. Genes were classified according to the Kyoto Encyclopedia of Genes and Genomes (18).

Increased oxidative capacity of liver mitochondria in strain 6C1 relative to 6C2

The increase in liver OxPhos gene expression in strain 6C1 relative to 6C2 suggested that oxidative metabolism would be increased as well. To test this, mitochondria were isolated from liver as previously described (12). The weight of the liver in the obese strain 6C1 (2.4 ± 0.2 g, n = 10) was greater per mouse than in the obesity-resistant strain 6C2 (1.4 ± 0.1 g, n = 10; P < 0.001). However, the amount of mitochondria isolated per gram of liver was less in strain 6C1 (15.1 ± 1.3 mg/g wet weight liver, n = 10) compared to that harvested from strain 6C2 (20.9 ± 1.7 mg/g wet weight liver, n = 10; P < 0.05). As a result, the amount of mitochondria isolated from the individual livers is comparable between the obese (35.4 ± 4.3 mg/liver) and the obesity-resistant strains (29.9 ± 3.6 mg/liver; P > 0.3).

Mitochondrial function was examined in isolated liver mitochondria following depletion of endogenous substrates (Table 2). The rate of oxidation of glutamate by liver mitochondria from the obesity-resistant strain 6C2 was coupled to the addition of 80 nmoles of adenosine diphosphate (ADP) with a respiratory control ratio of 6.2 and the efficiency of OxPhos was normal with an ADP/O ratio of 3. The rate of oxidation of glutamate was not increased by a 12.5-fold increase in ADP concentration to 2 mmol/l. Moreover, the respiratory rate is not further increased with dinitrophenol indicating that when phosphorylation is uncoupled, the oxidation component of OxPhos controls the velocity of respiration. Interestingly, the state 3 oxidative rate was 1.8 times faster in the obese strain 6C1 relative to the obesity-resistant strain 6C2 (Table 2). State 4 respiration was similarly increased (2.0-fold) leading to comparable respiratory control ratio and ADP/O ratios between the two strains. Oxidative rates with hi ADP (2 mmol/l) and the uncoupler dinitrophenol are also increased in the liver mitochondria from strain 6C1 relative to 6C2.

Table 2.  Increased mitochondrial oxidative phosphorylation in strain 6C1 liver relative to strain 6C2
inline image

Discussion

We have refined the Obrq2 genetic interval to a 30 Mb region on mouse chromosome 6 and extended the phenotypic analysis of this strain to demonstrate that strain 6C1 is glucose intolerant, insulin resistant as well as obese relative to strain 6C2. To further understand the molecular basis of the obesity and insulin resistance, we examined the global gene expression patterns associated with Obrq2 using microarray hybridization.

Pathway analysis was used to study the gene expression patterns of mice fed the HFSC diet, which revealed that insulin resistance and increased plasma glucose levels were associated with a broad increase in the expression of genes involved in liver mitochondrial OxPhos. The upregulation of OxPhos gene expression in strain 6C1 relative to 6C2 was subtle but widespread, comprising 56 of 76 OxPhos genes that spanned each complex of the electron transport chain (Figures 3 and 4). This observation is further highlighted by the 1.8-fold increase in oxidative metabolism in the liver mitochondria of strain 6C1 relative to 6C2 (Table 2). Additionally, liver mitochondria from both strains are well coupled with normal ADP/O ratios indicating comparable efficiency. Therefore, although equivalent amounts of mitochondrial protein are present in the livers of strains 6C1 and 6C2, the mitochondria in the obese strain 6C1 have a markedly increased oxidative capacity.

The relationship between insulin resistance and OxPhos has been widely studied, although the focus has largely been on skeletal muscle, in part due to the ease of accessibility (21). In skeletal muscle, insulin resistance is associated with a decrease in OxPhos gene expression (20). However, whether impaired mitochondrial function is a cause or consequence of insulin resistance remains controversial. Mild tissue-specific defects in either liver or skeletal muscle OxPhos due to apoptosis inducing factor (Aif) deficiency protects mice from obesity and insulin resistance, suggesting that the decrease in OxPhos activity in skeletal muscle is a consequence rather than a cause of insulin resistance (6). Additional evidence that decreased liver OxPhos activity can prevent insulin resistance comes from studies of gene expression patterns in human livers between obese, normoglycemic individuals and obese, and insulin resistant individuals. These comparisons identified a decrease in OxPhos-related gene expression levels in the obese, normoglycemic individuals relative to their insulin resistant counterparts (22,23). Gene expression analysis of the strains defining Obrq2 identified a similar expression pattern, with decreased OxPhos-related gene expression levels in the liver associated with protection from obesity and insulin resistance. Metabolite measurements in mice with a mild OxPhos reduction suggest a shift toward anaerobic glucose metabolism, which is considerably less efficient than aerobic glucose metabolism and may contribute to the reduction in adiposity (6).

The OxPhos gene expression differences associated with Obrq2 were only observed in the liver, with no changes in the OxPhos pathway detected in adipose or skeletal muscle. Decreased OxPhos activity in muscle due to insulin resistance is detectable using pathway analysis of gene expression patterns (20). Therefore, the lack of change in OxPhos gene expression levels in muscle between 6C1 and 6C2 suggests that the increase in liver OxPhos expression precedes any decrease in skeletal muscle. This further supports the notion that skeletal muscle mitochondrial dysfunction is a consequence of insulin resistance. The implication of these findings shifts the paradigm for the prevention of T2D from therapeutics to increase mitochondrial OxPhos to those that can mildly reduce it. This may have immediate clinical relevance as a number of compounds exist that inhibit OxPhos (7).

Pgc-1α is a key transcriptional regulator of both mitochondrial oxidative metabolism and hepatic gluconeogenesis (24,25). Expression levels of Pgc-1α were 2.1-fold higher in strain 6C1 relative to strain 6C2 (Supplementary Table S2 online, P < 10−8) and may therefore provide a link between the increase in both OxPhos gene expression levels and hepatic gluconeogenesis in strain 6C1. Pgc-1α is a transcription co-activator that controls these metabolic processes by upregulating expression of a series of genes including Pepck, G6pc, Hnf4a, Foxo1, NRF-1, NRF-2, among others (24). However, none of these genes are significantly upregulated in strain 6C1 relative to 6C2. In fact, Hnf4a is downregulated in strain 6C1, suggesting that Pgc-1α may be acting through a different pathway. Humans with T2D and fasting hyperglycemia also lack overexpression of hepatic Pepck and G6pc, which indicates that this alternate pathway may have significant clinical implications (26). One promising candidate in the Pgc-1α pathway is Lipin1, which is overexpressed in the liver of strain 6C1 relative to 6C2 following 28 days on the HFHS diet. Although this precedes the hepatic upregulation of Pgc-1α gene expression, Lipin1 has been shown to regulate OxPhos gene expression in a Pgc-1α dependent manner (27). Further studies will be required to determine the key downstream effectors of Pgc-1α signaling or demonstrate that the link between OxPhos gene expression and hepatic gluconeogenesis in Obrq2 is Pgc-1α-independent.

The differential expression of liver OxPhos genes occurred following 100 days on the HFHS diet. This change is preceded by expression differences in the liver between 6C1 and 6C2 in genes in the Ndk dynamin pathway. The dynamin complex is involved in both clathrin-dependent and clathrin-independent endocytosis (28). dynamin is a guanosine triphosphatase that directly interacts with Ndk, which regulates dynamin function by serving as a guanine nucleotide exchange factor (29). dynamin mediates the endocytosis of several key mediators of adiposity and insulin signaling including glucose transporter-4 in adipose and muscle (30,31) and MC4R in hypothalamus (32). Given that expression levels in the Ndk dynamin pathway differed between 6C1 and 6C2 in the liver, of particular interest is the role of dynamin in regulating endocytosis of the insulin receptor. Overexpression of a dominant negative allele of dynamin in a hepatoma cell line disrupted endocytosis of the insulin receptor (19). This disrupted the phosphorylation and activation of extracellular signal-regulated kinase-1 (ERK1) and ERK2 although phosphorylation of Akt remained unaffected. ERK1/2 are localized in the mitochondria of many different cell types where they interact with voltage-dependent anion channel 1 (Vdac) and histones H2A and H4, and causes a subtle but widespread increase in mitochondrial gene expression levels (33). It is therefore possible that alterations in the ERK1/ERK2 signaling pathway provide a direct link between the variation in the Ndk dynamin pathway detected in liver following 28 days on the HFHS diet and the increased mitochondrial OxPhos gene expression levels seen in 6C1 liver following 100 days on the HFHS diet.

In addition to understanding the molecular physiology of Obrq2, identifying its genetic basis also promises to shed light on the etiology of obesity and insulin resistance. Among the 8 cis-eQTLs discovered by gene expression profiling, Asb15, Smo, and D830026I12 expression differences were identified following just 4 weeks on the HFHS diet and therefore represent more likely variants causing the Obrq2 phenotype, rather than secondary to the development of obesity and insulin resistance. Further studies of these genes as well as the relationship between obesity, insulin resistance, and hepatic OxPhos in other dietary and genetic contexts will continue to provide new insights into the molecular and genetic basis of obesity and diabetes.

SUPPLEMENTARY MATERIAL

Supplementary material is linked to the online version of the paper at http:www.nature.comoby

ACKNOWLEDGEMENT

This research was supported by the Gene Expression and Genotyping Facility of the Case Comprehensive Cancer Center (P30 CA43703), the NIH NCRR grant RR12305, the NIH NCI Transdisciplinary Research on Energetics and Cancer (TREC) grant U54 CA116867, and by a fellowship from the Regione Autonoma Della Sardegna (RAS). We wish to thank Dr. Colleen Croniger for helpful discussions.

DISCLOSURE

The authors declared no conflict of interest.

Ancillary