Iron Overload and Diabetes Risk: A Shift From Glucose to Fatty Acid Oxidation and Increased Hepatic Glucose Production in a Mouse Model of Hereditary Hemochromatosis

Mice with knockout of the Hfe gene (Hfe^−/−) (13) were bred onto the 129/SvEvTac and C57BL/6J genetic backgrounds for more than 10 generations. Normal chow (Harlan Teklad TD-8640) contained 4.5% of calories as fat and 0.33 g/kg of carbonyl iron. High-fat diet (Research Diets D12451) contained 45% calories from fat. Mice were fed for a period of 2 months, beginning at age 4–6 months. Procedures were approved by the Institutional Animal Care and Use Committee of the University of Utah.

Quantative RT-PCR was performed as described previously (16). Muscle RNA was extracted using the PureLink Micro-to-Midi Total RNA Purification System (Invitrogen, Carlsbad, CA) according to the manufacturer's protocol. First-strand cDNA synthesis was carried out using Superscript III reverse transcriptase (Invitrogen) according to the manufacturer's protocol. Real-time PCR was performed with a thermal cycler (LightCycler, Roche Diagnostics, Basel, Switzerland) as previously described (20). Primers were designed using Primer3 software (http://primer3.sourceforge.net). Quantification of cDNA products was accomplished by the LightCycler software. Messenger RNA levels of specific genes were normalized to the average of the cyclophilin and Rpl13a levels for the same sample.

Oxygen consumption in isolated muscle mitochondria was measured using a fiber-optic oxygen sensor (Ocean Optics, Orlando, FL) as previously described (21). Hindlimb soleus muscles were suspended in 2.0 ml of 120 mmol/l KCl, 3 mmol/l HEPES, 5 mmol/l KH2PO4, 1 mmol/l EGTA and 1 mg/ml fatty acid free BSA, pH 7.2. Respiration assay was performed in a thermostated (25°C), water-jacketed, sealed glass chamber with constant magnetic stirring. Oxygen consumption rates were measured for the following substrates: 5 mmol/l glutamate + 2 mmol/l malate in the basal state (state 2), after the addition of 1 mmol/l ADP (to maximally stimulate the respiration, state 3), and in the presence of 1 μg/ml oligomycin to block ATP synthesis (state 4). Respiratory control ratios were calculated as the ratio of respiration rate in state 3 to that in state 4.

Soleus muscles were weighed and put in 24-well plates containing 400-μl Krebs-Ringer bicarbonate buffer (2.8 mmol/l glucose, 2 mg/ml BSA, 10 mmol/l HEPES [pH 7.4], 5 mmol/l NaHCO₃, 2.2 mmol/l CaCl₂, 1.2 mmol/l MgSO₄-7H₂O, 1.2 mmol/l KH₂PO₄, 4.8 mmol/l KCl, 126.2 mmol/l NaCl), pregassed for 30 min with 95% O₂/5% CO₂. Next, 3 μCi of [U-¹⁴C] glucose were added to each well for 2 h at 37°C. Glucose oxidation was quantified by measuring ¹⁴CO₂ release, trapped by paper membranes soaked with hyamine hydroxide. Background release from a well containing no tissue was subtracted.

Hearts were perfused in the working mode and glucose oxidation determined as previously described (22).

Palmitate oxidation by soleus muscle was assayed as previously described (23,24) in 500-μl Krebs-Ringer bicarbonate buffer containing 1% fatty-acid free BSA, 1 mmol/l carnitine, and 5 μCi of [9,10-³H] palmitate. Background ³H₂O release measured in a well without tissue was subtracted from experimental values.

Mice age 6 to 8 months old were studied for 3 consecutive days in a four-chamber open-circuit Oxymas system (CLAMS; Columbus Instruments, Columbus, OH) to measure oxygen consumption and carbon dioxide production.

Studies were performed in weight-matched nonsedated mice as described (25). The jugular vein was catheterized under avertin anesthesia, and using Micro-renathane tubing (Braintree Scientific, MRE 025). The animals were allowed to recover for 48 h before undergoing the clamp procedure. Mice were fasted overnight and transferred to a standard mouse housing cage with a tether arm attached to the catheter. A dual infusion pump (pump 33, Harvard Apparatus) was used to infuse insulin at a constant flow rate with no priming dose. A solution of 50% dextrose was infused at a variable rate to maintain a target value of 100–150 mg/dl and held at that level for 60 min. We also infusted [³H]glucose at a constant rate (0.1μCi/min). Glucose was measured using 3 μl of tail vein blood at 10-min intervals using a tail clip and glucometer (Glucometer Elite, Bayer, Tarrytown, NY). Insulin levels did not differ between the groups, and we also determined that there was no change in insulin clearance in Hfe^−/− mice as determined from C-peptide/insulin ratios (not shown).

Pyruvate dehydrogenase (PDH) activity was examined as described before (26) using a radioactive-enzymatic method. Frozen skeletal muscle tissue was homogenized in a buffer containing 25 mmol/l HEPES (pH 7.4), 3 mmol/l EDTA, 5 mmol/l sodium dichloroacetate, 1 mmol/l N-{α}-tosyl-l-lysylchloromethane, 1 mmol/l ADP, 5 mmol/l dithiothreitol, 2 mmol/l CaCl₂, 5 mmol/l MgCl₂, and 25 mmol/l KF, supplemented with 1.5% Triton X-100 and 5 μmol/l leupeptin. Before the addition of assay buffer, aliquots of homogenate were incubated at 30°C for 30 min to allow dephosphorylation and activation of PDH. Active PDH were determined in a reaction mixture containing 50 mmol/l HEPES (pH 7.4), 0.4 mmol/l thiamine pyrophosphate, 0.4 mmol/l CoA, 2 mmol/l dithiothreitol, 5 mmol/l NAD+, and 5 mmol/l MgCl₂, supplemented with 0.17% Triton X-100, 5 units/ml dihydrolipoamide reductase (E3), 20 μl of homogenate, and 2 μCi [1-¹⁴C]-pyruvate. The PDH reaction yielding acetyl-CoA + CO₂ was performed at 30°C for 7 min before termination of the reaction by the addition of 2 mol/l acetic acid in 2% SDS. The ¹⁴CO₂ released from the PDH reaction was trapped by paper membranes soaked with hyamine hydroxide and quantified with a scintillation counter (Beckman Coulter). PDH activity was expressed as nmol · min⁻¹ · mg tissue⁻¹.

Age-matched (6–8 month) male mice were killed. Serum triglycerides and free fatty acids were measured using assay kits from Sigma (St. Louis, MO) and Roche (Penzburg, Germany), respectively. Tissue was collected and tissue homogenates were used to assess pyruvate and glycogen levels. Pyruvate and lactate were measured by assay kits (Biovision, Mountain View, CA). Liver glycogen was measured as previously described (23).

Blood glucose was measured with a glucometer in fasted mice before and after intraperitoneal injection of sodium pyruvate (2 mg/g body weight).

Hind-limb muscle was collected and tissue homogenates were prepared for Western blot analysis. Levels of total and/or phosphorylated proteins were detected by immunoblotting using the following antibodies: Acetyl-CoA carboxylase (Ser79) (Cell Signaling Technology, Danvers, MA), AMPK (Thr172) (Cell Signaling Technology), ferritin (Abcam, Cambridge, MA), Akt (Cell Signaling Technology), and IRS2 (Abcam). Signals quantified by densitometry were normalized to glyceraldehyde-3-phosphate dehydrogenase (GAPDH) levels (Santa Cruz Biotechnology, Santa Cruz, CA) or, in the case of phosphoproteins, to the total levels of the same protein.

Descriptive statistics are represented as average ± SEM. A two-tailed Student t test was used to compare the differences between groups.

As previously shown for other tissues (13,27), skeletal muscle in Hfe^−/− mice is iron overloaded, with a 2.3-fold increase in ferritin compared with muscle of wild-type mice of the same strain (129SvEvTac, 129) as detected by Western blotting (P < 0.01, Fig. 1 A). We have reported that nonheme iron is also elevated in Hfe^−/− muscle (18).

We previously demonstrated that Hfe^−/− mice exhibited increased muscle glucose uptake (18). To further study the metabolic fate of absorbed glucose, glucose oxidation was measured in skeletal and cardiac muscle. Isolated soleus muscle from 129 Hfe^−/− mice exhibited a 33% decrease in glucose oxidation compared with wild type (Fig. 1,B, P = 0.05). Similar results were obtained in isolated working hearts (Fig. 1 B, P < 0.005). The decreased glucose oxidation, despite elevated glucose uptake in muscle, are manifest by increased lactate levels. We previously reported that lactate fluxes and serum lactate were elevated in Hfe^−/− mice (18), and we have verified that tissue lactate levels are also increased by 46% in Hfe^−/− muscle (9.1 ± 0.9 μg lactate/mg protein in wild-type muscle compared with 13.3 ± 1.1 μg/mg in Hfe^−/− muscle, N = 7, P = 0.01, not shown).

We then measured fatty acid oxidation in isolated soleus muscle. Hfe^−/− mice on the 129 or the diabetes- and obesity-prone C57BL/6J (C57) backgrounds demonstrated similar level of palmitate oxidation as determined by measuring ³H₂O release from ³H palmitate (Fig. 1,C). However, in the setting of the decreased glucose oxidation, this represents a 47% increase in the ratio of fatty acid to glucose oxidation in the Hfe^−/− compared with wild-type mice (Fig. 1 D, P < 0.01). A relative increase in fatty acid oxidation is supported by decreased serum triglycerides in the Hfe^−/− mice (52.3 ± 5.3 mg/dl in wild-type compared with 31.5 ± 8.6 mg/dl in Hfe^−/− mice, N = 6/group, P < 0.05) and a trend toward decreased serum free fatty acids (0.061 ± 0.03 mmol/l in wild-type compared with 0.056 ± 0.025 mmol/l in Hfe^−/− mice, P = 0.18).

There was no evidence of a change in glycogenolysis insofar as muscle glycogen levels did not differ between wild-type and Hfe^−/− mice (10.6 ± 1.6 mg/g muscle in type compared with 8.3 ± 2.2 mg/g in Hfe^−/− mice, N = 7/group, P = 0.22).

To determine if decreased mitochondrial function contributed to the decreased glucose oxidation, mitochondrial function was analyzed by measuring oxygen consumption by permeabilized muscle fibers. Compared with wild-type mice, mitochondria from Hfe^−/− soleus muscle exhibited only modestly decreased mitochondrial respiration. Under glutamate/malate-supported respiration, states 2, 3, and 4 respiration were apparently reduced by 19% (P = 0.05), 8.5% (P = 0.22), and 12.4% (P = 0.13), respectively (Fig. 2). ATP production also trended lower in Hfe^−/− mice (16%), although not statistically significantly (not shown, P = 0.31). Mitochondrial density, as assessed by the ratio of mitochondrial to nuclear DNA, did not differ between wild-type and Hfe^−/− mice (0.36 ± 0.06 vs. 0.32 ± 0.06 arbitrary units, respectively, P = 0.68).

Hfe^−/− mice, candidate gene expression was assessed in muscle using quantitative RT-PCR (Fig. 3 A). Carnitine palmitoyl transferase b (Cpt1b) mRNA was increased by 74% and Pdk4 mRNA by 94% in Hfe^−/− mice compared with wild-type mice (P < 0.05). Fatty acid synthase mRNA was decreased by 41% in the Hfe^−/− mice, although this difference did not reach statistical significance (P = 0.08). Transcripts of several other genes involved in nutrient transport or metabolism (Acl, Cd36, Acox1, Mcad, Lcad, Pdk1, Ldh α, Glut4, and Hif1α) did not differ. Because Pdk4 and Cpt1b are PPARδ targets, we also measured PPARδ mRNA levels, but PPARδ did not differ between wild-type and Hfe^−/− mice (7% decrease in Hfe^−/− mice, P = 0.48, not shown).

The change in pdk4 mRNA possibly explains decreased glucose oxidation because PDH kinase 4 (PDK4) phosphorylates and inhibits PDH, inhibiting pyruvate entry into the trichloroacetic acid; tricarboxylic acid (TCA) cycle (28). We therefore assessed the active fraction of PDH in soleus muscle, which was decreased by 34% in Hfe^−/− compared with wild-type soleus muscle (Fig. 3 B, P < 0.001). Total PDH activity did not differ between the two strains (18% increase in Hfe^−/− mice, P = 0.45, not shown).

Fatty acids are relatively preferred as a fuel in muscle of Hfe^−/− mice on a normal (high carbohydrate) chow diet. We therefore examined fatty acid oxidation in muscle of Hfe^−/− mice on a high-fat diet, and in this case, isolated soleus muscle from Hfe^−/− mice demonstrated an absolute increase in the capacity for fatty acid oxidation compared with wild type (Fig. 4,A, P < 0.05). Consistent with their roles in regulating fatty acid oxidation, levels of pAMPK and pACC were increased in skeletal muscle of Hfe^−/− mice on high fat (Fig. 4 B and C).

The preference for fatty acid oxidation in Hfe^−/− mice on a high-fat diet was confirmed in studies of mice in metabolic chambers. Hfe^−/− mice on a high-fat diet exhibited significant increases in oxygen consumption (Fig. 4,D) and heat production (Fig. 4,E) compared with wild type. These were not specific to mouse strain, as the differences were significant for mice on either the 129 or C57Bl6 backgrounds. Hfe^−/− mice on the high-fat diet also had significantly lower respiratory exchange ratios (Fig. 4,F). Differences between Hfe^−/− and wild type in heat production and respiratory exchange ratios were not apparent in mice fed the predominantly carbohydrate normal chow (not shown). In fact, Hfe^−/− mice on both the 129 and C57 genetic backgrounds have lower oxygen consumption rates on high carbohydrate chow (25%, P < 0.05, not shown), consistent with their decreased glucose oxidation rates on the same diet (Fig. 1 B).

To better understand the overall fuel economy in the Hfe^−/− mice, we next measured hepatic glucose output (HGO) by means of the hyperinsulinemic euglycemic clamp. HGO was suppressed at maximal 10 mU/kg/min insulin in both wild-type and Hfe^−/− mice (not shown), but at submaximal 5 mU/kg/min insulin, HGO was increased by 4.5-fold in Hfe^−/− mice on normal chow and 1.7-fold on high fat (Fig. 5,A, P < 0.001). The increased HGO did not appear to stem from an overall stimulation of the pathway for gluconeogenesis, as the mRNAs for phosphoenolpyruvate carboxykinase (PEPCK) and glucose-6-phosphatase (G6Pase) were not increased in livers of Hfe^−/− mice (Fig. 5,B). Consistent with this lack of upregulation of gluconeogenic enzymes, pyruvate tolerance testing revealed no difference in glucose production between wild-type and Hfe^−/− mice (Fig. 5,C). The increased HGO also did not stem from insulin resistance or insulin deficiency: We previously reported that glucose disposal determined by glucose clamp is elevated in Hfe^−/− mice on normal chow at submaximal but not maximal glucose (18), and glucose disposal also trended toward an increase in the Hfe^−/− mice on high fat compared with wild type (157 ± 36 vs. 95 ± 10 mg/kg/min, P = 0.13, not shown). Furthermore, hepatic insulin signaling was augmented in the Hfe^−/− mice on normal chow. Fasted mice exhibited a 1.6-fold increase in phosphorylated Akt (pAkt) and a 2.3-fold increase in tyrosine phosphorylation of insulin receptor substrate-2 (pIRS2) (Fig. 5,D and E, P < 0.05 for both). The source of the higher HGO also did not appear to be increased glycogenolysis, as hepatic glycogen levels were increased by 65.7% in Hfe^−/− mice compared with wild-type mice (Fig. 5,F, P < 0.05). Pyruvate levels were increased by 32% in livers of Hfe^−/− mice (Fig. 5,F, P < 0.01). Unlike the case in skeletal muscle, however, in liver there was no evidence for a decrease in pyruvate entry into mitochondrial oxidative pathways. Levels of Pdk1 and Pdk2 were in fact significantly lower in the Hfe^−/− liver, which would favor increased PDH activity and decreased pyruvate levels; Pdk's 3 and 4 were undetectable. The mRNA for pyruvate carboxylase (PCX) which catalyzes pyruvate conversion to oxaloacetate for either gluconeogenesis or anaplerotic entry of pyruvate into the TCA cycle was slightly increased (25%, P < 0.05) in Hfe^−/− mice compared with wild-type mice (Fig. 5 G, P < 0.05). Both of these changes in liver would actually favor increased mitochondrial utilization of pyruvate in Hfe^−/− liver. In sum, therefore, the data above suggest that the increased pyruvate and HGO largely result from increased peripheral cycling rather than decreased hepatic utilization.

Increased tissue iron levels are associated with diabetes, both in human hereditary hemochromatosis (HH) and in dietary iron overload (1,3,15,29,30). Although this is at least partially the result of decreased insulin secretion (15,17,27), tissue iron overload also results in significant changes in glucose metabolism in skeletal muscle (18). In that case, however, the effect of iron is to increase glucose uptake, a change that would be predicted to be protective of diabetes. To better understand the changes in overall fuel economy caused by iron that might contribute to diabetes, we have investigated muscle and hepatic carbohydrate and fatty acid metabolism in a mouse model with targeted deletion of the gene most commonly mutated in human HH, Hfe. We show that Hfe^−/− mice exhibit a shift in fuel preference from glucose to fatty acid oxidation in muscle. The decreased glucose oxidation in muscle is associated with increased pyruvate/lactate recycling to liver, as demonstrated in our previous metabolic flux studies (18), in Fig. 5, and by the increased hepatic glucose output. These combined effects may contribute to the increased prevalence of diabetes in individuals with HH.

Our results are consistent with the close coupling of metabolism to iron availability such as has been demonstrated in lower eukaryotes (31,–33). The decreased glucose oxidation in skeletal muscle is due, at least in part, to decreased PDH activity. The mechanism is likely the observed increase in Pdk4 mRNA, PDK4 activity being largely regulated at the transcriptional level (28). A candidate mediator for the regulation of Pdk4 may be increased AMPK activity (34) that we previously reported in Hfe^−/− mouse muscle (18). Other factors contributing to the decreased glucose oxidation include a modest degree of mitochondrial dysfunction and increased fatty acid oxidation (Randle effect) (35). The mitochondrial dysfunction in Hfe^−/− muscle is likely, however, to be a relatively minor contributor given the increased capacity for fatty acid oxidation apparent in mice on high-fat diets.

The mechanism for the increased fatty acid oxidation is also likely multifactorial. We previously reported higher serum adiponectin levels in Hfe^−/− mice (18) that would contribute to higher rates of fatty acid oxidation through increased AMPK activation and decreased ACC activity (36,37). Increased Cpt1b expression may also play a role, as might changes in malonyl-CoA, which were not assessed. Perhaps related to the current findings, mice overexpressing erythropoietin in muscle exhibit a similar phenotype to iron-overloaded mice, including increased fat oxidation, decreased glucose oxidation, and protection from diet-induced obesity, although tissue iron levels were not measured in that model (38). Insofar as iron sufficiency should facilitate unimpaired erythropoiesis, it would seem advantageous for an iron-sufficient mouse to shift to the more energy-efficient but oxygen-inefficient fuel source of fatty acids to make use of that full capacity for oxygen transport. Indeed, under the opposite condition of hypoxia, fatty acid oxidation decreases, enzymes for fat metabolism downregulate, and glucose oxidation increases (39,40). Consistent with this proposed adaptive relationship between iron and fuel choice, Hfe^−/− mice consume more oxygen and produce more heat when fat is available in the diet. A decreased ability to transition between utilization of carbohydrate and lipid fuel sources, so-called “metabolic inflexibility,” is a characteristic of the metabolic syndrome and type 2 diabetes (19). Chronically increased fat oxidation, especially in the setting of decreased mitochondrial function, might also contribute to the accumulation of lipid products that have been implicated in the pathogenesis of diabetes (41). Whether these pathways contribute to increased diabetes risk in human hemochromatosis, however, is unknown.

The data are most consistent with the source of the increased hepatic glucose production being increased recycling of lactate and pyruvate, as was also previously documented by isotopomer analysis of the fate of 1,2[¹³C]glucose (18). It appears that this increase in hepatic glucose production is mainly substrate driven as the levels of mRNA for gluconeogenic enzymes were not increased (Fig. 5,B) and glucose production from exogenous pyruvate was not enhanced (Fig. 5,C). The data also do not support changes in insulin signaling as the underlying cause of the phenotype. We previously demonstrated no change in basal or insulin-stimulated (in vivo) pAkt in skeletal muscle (18), and herein we shown that the increased hepatic glucose production occurs in the face of paradoxically increased insulin signaling in liver (Fig. 5 D and E). Why insulin signaling is upregulated is not known. Levels of the insulin-sensitizing adipokine have been shown to be elevated in Hfe^−/− mice (18), but other unknown factors may also contribute, including compensation for decreased insulin levels (17) or decreased inflammatory signaling in macrophages based on their lower iron content (42). Why this increased insulin signaling does not translate to decreased levels of gluconeogenic enzymes is unclear and is under investigation.

In sum, we have shown here and in previous work that the diabetes risk engendered by iron operates not only through its toxic effects on β cells mediated by increased oxidative stress and mitochondrial dysfunction. Rather, iron also exerts regulatory effects on metabolism and fuel choice. Iron overloaded skeletal muscle demonstrates decreased glucose oxidation and increased fatty acid oxidation. Liver of Hfe^−/− mice exhibits increased glucose output, with the patterns of hepatic gene regulation and metabolite levels suggesting that the gluconeogenesis is largely substrate-driven and results from the altered fuel choice in muscle. These findings should help elucidate the association between iron and diabetes, not only in hemochromatosis, but also in nonhemochromatotic individuals with the metabolic syndrome or type 2 diabetes associated with dietary iron excess.

This work was supported by the National Institutes of Health (DK-81842 [D.A.M.] and HL73167 [E.D.A.]), the Research Service of the Veterans Administration, and the University of Utah Center for Clinical and Translational Research (UL1-RR025764).

No potential conflicts of interest relevant to this article were reported.

J.H. wrote the manuscript and researched data. D.J., B.L., M.S., and J.S. researched data. E.D.A. researched data and contributed to discussion. D.A.M. wrote the manuscript and reviewed/edited the manuscript.

Parts of this study were presented in abstract form at the 70th Scientific Sessions of the American Diabetes Association, Orlando, Florida, 25–29 June 2010.