Adaptive mechanisms regulate preferred utilization of ketones in the heart and brain of a hibernating mammal during arousal from torpor

Abstract

Hibernating mammals use reduced metabolism, hypothermia, and stored fat to survive up to 5 or 6 mo without feeding. We found serum levels of the fat-derived ketone, d-β-hydroxybutyrate (BHB), are highest during deep torpor and exist in a reciprocal relationship with glucose throughout the hibernation season in the thirteen-lined ground squirrel (Spermophilus tridecemlineatus). Ketone transporter monocarboxylic acid transporter 1 (MCT1) is upregulated at the blood-brain barrier, as animals enter hibernation. Uptake and metabolism of ¹³C-labeled BHB and glucose were measured by high-resolution NMR in both brain and heart at several different body temperatures ranging from 7 to 38°C. We show that BHB and glucose enter the heart and brain under conditions of depressed body temperature and heart rate but that their utilization as a fuel is highly selective. During arousal from torpor, glucose enters the brain over a wide range of body temperatures, but metabolism is minimal, as only low levels of labeled metabolites are detected. This is in contrast to BHB, which not only enters the brain but is also metabolized via the tricarboxylic acid (TCA) cycle. A similar situation is seen in the heart as both glucose and BHB are transported into the organ, but only ¹³C from BHB enters the TCA cycle. This finding suggests that fuel selection is controlled at the level of individual metabolic pathways and that seasonally induced adaptive mechanisms give rise to the strategic utilization of BHB during hibernation.

natural hibernators face unique challenges in surviving physiological extremes that would normally lead to death in most species of mammals. Profound reductions in heart rate and oxygen consumption, in conjunction with near-freezing body temperatures, require a multitude of cellular and molecular adaptations for a hibernator to avoid injury (reviewed in Ref. 1). One of the more striking adaptations is the ability to survive 5–6 mo without feeding by switching over to a lipid-based metabolism. In the absence of food, survival relies on the liberation and mobilization of fatty acids stored in the hibernator's white adipose tissue. However, some organs, most notably the brain, cannot use fat as its sole source of fuel. In this paper, we examine the hypothesis that fat-derived ketone bodies provide the critical fuel for the brain and heart during a hibernator's prolonged period of starvation.

Brain function in mammals requires a relatively high rate of energy metabolism, and under normal circumstances, the predominant fuel is d-glucose. However, there are several circumstances such as starvation, diabetes, suckling neonates, and high-fat diets when ketone bodies are elevated in plasma and substantially replace glucose as a brain energy substrate [reviewed in Ref. 20]. Common to all of these scenarios is an ample supply of lipids either in the diet (mother's milk, high-fat foods) or stored reserves (brown and white adipose tissue). Lipolysis of triacylglycerols generates free fatty acids that are converted in liver mitochondria to ketone bodies. Efflux from the liver causes plasma ketone levels to rise from ∼0.03–0.3 mM to as high as 2.5 mM or higher in humans (5). An additional physiological condition that joins the list of elevated ketone levels is hibernation (13, 25). Elevated tissue and blood concentrations of ketone bodies suggest that these compounds are an important energy source in hibernators. In vitro studies using brain slices indicate that the hibernating brain also metabolizes ketones (19).

We are interested in the mechanism by which a hibernating mammal sustains metabolic activity under hypothermic conditions without food for a span of several months. In this paper, we report on the transport and metabolism of glucose and the ketone d-β-hydroxybutyrate (BHB) over a wide range of body temperatures (T_b = 7–38°C) in the heart and brain of the thirteen-lined ground squirrel (Spermophilus tridecemlineatus). Our findings show that serum concentrations of glucose and BHB are inversely proportional to each other at various body temperatures and activity states throughout the hibernation season. Although both glucose and BHB efficiently enter the heart and brain at a variety of body temperatures, there is a preference for BHB utilization in both tissues during hibernation.

MATERIALS AND METHODS

Animals.

All animal procedures were performed according to the guidelines for the care and use of laboratory animals at the University of Minnesota and were approved by the Institutional Animal Care and Use Committee. Thirteen-lined ground squirrels (Spermophilus tridecemlineatus) were live-trapped in Minnesota and housed at the designated animal facility at the University of Minnesota School of Medicine Duluth. Squirrels were kept individually in plastic top-load rodent cages filled with pine shavings and observed daily to assess overall animal health, levels of food and water consumption, and the initiation of torpor bouts. Diet consisted of Purina PMI Lab Diet 5001 rodent chow supplemented with black oil sunflower seeds with water available ad libitum. Squirrels were maintained at an ambient temperature of 23°C from late March to August, 15–17°C in September, and 11°C in October at a 12:12-h light-dark cycle (7 A.M./7 P.M.). This gradual step-down in air temperature has been used previously (3, 29) to duplicate the seasonal temperature decline that these animals normally experience in the wild and took place in an Institutional Animal Care and Use Committee-approved environmental chamber. Temperature of pine shavings in the rodent cages is ± 0.5°C of the thermostatic setting of the environmental chamber. From November through mid-March, the ambient temperature was 5°C, food was absent, and the animals were housed in total darkness. Lengths of individual torpor bouts were confirmed by the sawdust technique (22). Briefly, this method involves sprinkling a small amount of sawdust on the back of a torpid squirrel followed by daily observations to determine whether the sawdust is still in place. Body temperatures (rectal) were measured at the time of death using a Physitemp (Clifton, NJ) BAT-12 microprobe thermometer with a RET-2 rectal probe inserted ∼2 cm into the animal. The accuracy of the thermometer is ± 0.1°C between 0 and 50°C. Animals were killed by decapitation, heart puncture, or funnel freezing, depending on the individual experiment (described below).

d-β-hydroxybutyrate and glucose serum assay.

Ground squirrel serum levels of the d-stereoisomer of β-hydroxybutyrate (d-BHB) were measured using the Stanbio Liquicolor β-hydroxybutyrate (Procedure # 2440) testing kit. Serum was prepared from whole blood collected at the time of death by heart puncture or from the neck by decapitation. Six microliters of serum were placed into one well of a 96-well plate with each sample assayed at least twice. The plate was incubated at 37°C for 3 min and then read in an absorbance spectrophotometer (Multiskan MCC/340, Labsystems, Helsinki, Finland) at 492 nm. Serum glucose levels were measured using the Stanbio Liquicolor Glucose testing kit (Procedure #1070). Three microliters of serum were placed into one well of a 96-well plate, and each sample was assayed at least twice. The plate was incubated at 37°C for 5 min and then read in an absorbance spectrophotometer at 492 nm.

Immunohistochemistry to determine cellular location of MCT1 and GLUT1.

We examined the cellular location and relative quantity of monocarboxylic acid transporter 1 (MCT1) and glucose transporter 1 (GLUT1) in active and hibernating ground squirrel brains by immunohistochemistry. Light microscopy was performed according to Gerhart et al. (8) on brain samples from summer active animals in August, animals entering hibernation in October, torpid animals in December, and spring active animals in April. Chicken polyclonal antisera were raised against the carboxyl terminus of rat MCT1 and against the carboxyl terminus of rat GLUT1. Antibodies against MCT1 were used to measure changes in the relative amount of MCT1 protein in ground squirrel brains with respect to time of year and animal activity. All immunohistochemistry reactions were done using the same batches of reagents in parallel to facilitate quantifiable comparisons among groups.

Ground squirrels were anesthetized with 5% halothane prior to heart puncture, perfused briefly with PBS, followed by perfusion with formal-acetic fixative (4% formaldehyde, 2% acetic acid). The perfusion time was 12 min, and tissues were stored at 4°C in fixative overnight before processing. Paraffin-embedded tissue sections were fixed onto glass slides and blocked with PBS containing 0.1% BSA and 1.5% normal goat serum. The primary antibody was diluted 1:1,200 in 0.1% BSA and applied to the sections for 1 h at room temperature. Sections were then incubated 30 min with biotinylated goat anti-chicken IgG (5 μg/ml in blocking solution) and 30 min with avidin-biotin-peroxidase complex reagent (both reagents from Vector Laboratories, Burlingame, CA). Color development was from 1 to 6 min in 0.6 g/ml 3,3′-diaminobenzidine (Sigma, St. Louis, MO).

Five digital photomicrographs of each squirrel cerebellum were obtained at ×200 magnification and analyzed with Scion Image (Scion, Frederick, MD). Uncalibrated optical density measurements were made using the grayscale images. For neuropil, five areas were selected in each micrograph with the freehand boundary tool, and mean optical densities were calculated. For microvessels, each micrograph was thresholded to a constant value, and the wand tool was used to select and measure the optical densities of 18 microvessel longitudinal sections in each micrograph. Thus, a total of 25 neuropil and 90 microvessel measurements were obtained for each brain. All measurements were exported to an Excel file, and average density values for neuropil and microvessels were calculated for each brain.

¹³C injection of ground squirrels.

Glucose and d-BHB labeled with the stable isotope ¹³C were used to measure fuel utilization during hibernation. Torpid adult thirteen-lined ground squirrels were given an intraperitoneal injection of either 1 ml of 1 M [2,4-¹³C₂]-d-β-hydroxybutyrate (¹³C-BHB) or 1 M [1-¹³C]-glucose (¹³C-glucose) purchased from Cambridge Isotope Laboratories (Andover, MA). Torpid animals injected December through March were selected based on the length of previous torpor bouts, where typical bouts ranged in length from 7 to 12 days as determined by the sawdust method (22). Therefore injections were in the middle of torpor bouts, ∼3–6 days after the previous interbout arousal (IBA). After injection, torpid animals were quickly returned to their nesting boxes under hibernating conditions of 24 h dark, ambient temperature of 5°C, and no food. We found that IP injections initiate the arousal process as body temperatures slowly rise over a period of hours. Torpid IP-injected animals (T_b = 5–7°C) were labeled for a minimum of 2 h and a maximum of 3 h following injection so that uptake and metabolism of ¹³C-BHB or ¹³C-glucose could be observed in the heart or brain by ¹³C magnetic resonance spectroscopy (¹³C MRS) at a variety of body temperatures without exhausting the pool of labeled substrate in either brain or heart. Because the rate of rewarming varied from animal to animal, we were able to take advantage of these individual differences to measure the extent of metabolic labeling at body temperatures ranging from 7 to 38°C. Incorporation of label below 10°C in both organs was not quantifiable due to greatly reduced metabolic activity. After labeling, body temperature was measured rectally, animals were killed by funnel freezing, and the heart and brain were snap frozen by liquid N₂. Perchloric acid (PCA) tissue extracts were prepared and used to experimentally determine incorporation of labeled substrates in brain and heart by ¹³C MRS.

Funnel freezing.

The method of death used in brain extract analysis is an in situ freezing technique commonly referred to as funnel freezing (4, 9, 23). This method of rapid freezing with liquid N₂ meets American Veterinary Medical Association 1993 humane methods when animals are deeply anesthetized prior to the procedure. The animal is deeply anesthetized using 3–5% isoflurane with compressed breathing grade air. All procedures are carried out after the animal is deeply anesthetized—confirmed by loss of purposeful limb movement and loss of corneal reflex. An incision in the skin running along the midline from the eye level to the occiput exposes the skull. A plastic funnel (bottom diameter of 15 mm) is fitted into the skin incision in which the posterior rim of the funnel lies on the lamboidal suture. The funnel is properly secured by pulling the skin around it with sutures to prevent liquid N₂ from leaking into the respiratory airways so that the animal continues to breathe without interruption until the brain is frozen solid in situ. To ensure complete freezing of the brain, liquid N₂ addition is continued for 3 min until the eyes become frozen white. After the brain is frozen, the funnel is removed, the animal is decapitated, and the heads are stored at −80°C until the brains are processed into PCA brain extracts for metabolite analysis. The hearts are rapidly dissected out, placed in cryovials, snap frozen in liquid N₂, and stored in −80°C until processed into PCA heart extracts for metabolite analysis.

Preparation of tissue extracts.

Frozen brains were chiseled out of the skull at −25°C and pulverized under liquid nitrogen using a mortar. Frozen hearts are made into a fine powder with liquid N₂ and mortar and pestle. Frozen brain and heart powders are placed in ice-cold 0.9 M PCA (3 ml for brain and 2 ml for heart) and centrifuged for 15 min at 5,000 g, 4°C. The supernatant is collected; the remaining pellet is resuspended in 1 ml PCA, and allowed to sit for 15 min. The resuspended pellet is centrifuged again for 15 min at 5,000 g, 4°C, and the supernatant is added to the supernatant from the first centrifugation. The combined supernatants are neutralized to pH 7 with 9 M KOH and centrifuged again for 15 min at 5,000 g, 4°C. The supernatants are collected and lyophilized in a freeze-dryer. The lyophilized sample is resuspended in 600 μl 10% D₂O/90% distilled H₂O. The pH of each resuspended solution is corrected so that it is in the range of 7.1–7.2.

¹³C NMR data collection and processing.

¹³C NMR data were recorded at 25°C on a 14.1-Tesla UNITY INOVA spectrometer (Varian, Palo Alto, CA) using a pulse-acquire sequence with Nuclear Overhauser Effect and ¹H decoupling (repetition time of 15 s, acquisition time of 1 s with 128 or 256 scans to achieve sufficient signal to noise, S/N). A Lorentzian line broadening of 0.1 Hz was applied before Fourier transformation. The areas of ¹³C peaks of interest in ¹³C spectra (e.g., glucose C1, BHB C2 and C4, glutamate C4, etc.) were obtained by fitting the lines with Lorentzian functions using Varian software (VNMR ver. 6.1c). The area of each peak is directly proportional to the ¹³C concentration of the corresponding ¹³C-labeled molecule.

To determine relative changes in the concentration of metabolites throughout the hibernation season, the peak areas for each ¹³C-labeled metabolite were normalized to natural abundance ¹³C-taurine concentration in the heart and natural abundance ¹³C-myo-inositol concentration in the brain for each individual sample. Myo-inositol is a stable, naturally abundant ¹³C-labeled metabolite in the brain that is not labeled by ¹³C-BHB or ¹³C-glucose injections (11). Taurine has been identified as a stable, naturally abundant ¹³C-labeled heart metabolite that is also not labeled by an injection of either ¹³C-BHB or ¹³C-glucose (17).

Statistical analysis.

Statistical analysis was performed using JMP software (SAS Institute, Cary, NC). Prism software (GraphPad Software, San Diego, CA) and Excel (Microsoft) were used to create the graphs shown in the figures. Seasonal serum levels of d-BHB, glucose, and MCT1 expression are expressed as means ± SE and were analyzed by one-way ANOVA with Tukey's HSD multiple-comparison test. Calculating the Pearson correlation coefficient tested the strength of the relationship between serum d-BHB and glucose levels. Levels of the tricarboxylic acid (TCA) cycle-derived metabolite [4-¹³C] glutamate (Glu C4) in brain and heart tissue following injection with ¹³C-BHB or ¹³C-glucose were analyzed by two-way ANOVA for each tissue with substrate and T_b as main effects. The relationship between Glu C4 and T_b level in brain and heart was explored using linear regression of Glu C4 level as a function of T_b for individual animals at the time of death.

RESULTS

Circulating levels of BHB and glucose during hibernation.

The concentration of BHB and glucose in ground squirrel serum was measured at various time points and activity states from September to June (Fig. 1). Interestingly, we observed an inverse correlation in glucose and BHB concentrations during the hibernation season. The mean circulating level of BHB increased from 0.26 mM in fed September–October active animals to ∼2.3 mM during torpor (T_b = 5–7°C) in December–March. During the same time span the level of serum glucose dropped from ∼8.5 mM in fed September–October active animals to about 5 mM during torpor in December–February, and down to 3.3 mM during torpor in March. The levels of both BHB and glucose reverted back to their prehibernating profile of higher glucose and lower BHB in fed active animals in April–June. Intriguingly, despite the fact that food was removed from all animals beginning November 1, animals undergoing regular interbout arousals (IBAs) showed an active profile of higher glucose and depressed BHB during these brief (<24 h) normothermic arousals from torpor that extended into the month of March (Fig. 1).

Fig. 1.

Mean serum concentrations of d-β-hydroxybutyrate (d-BHB) and glucose throughout the hibernation season. Measurements of both d-BHB (A) and glucose (B) in serum prepared from the same animals at the month(s) of year and activity state shown under the x-axis. Number of animals (n) and body temperature (T_b) at time of blood collection include September–October active (T_b = 21–38°C, n = 21); September–October torpor (T_b = 12–15°C, n = 8); December through February torpor (T_b = 5–9°C, n = 20); March torpor (T_b = 4–7°C, n = 7); December through March interbout arousal (T_b = 22–38°C, n = 13); and April through June active (T_b = 33–38°C, n = 10). Bars indicate mean serum d-BHB and glucose concentration. One-way ANOVA shows that time of year/activity state influenced the serum concentration of both d-BHB (F_5,69 = 17.62, P < 0.001) and glucose (F_5,69 = 7.15, P < 0.001). Error bars represent the standard error of the mean. In both A and B, bars that are not connected by the same letter (a, b, c) are significantly different according to Tukey's HSD test with Q = 2.93 and P < 0.05.

Transporters of glucose and BHB.

At the blood-brain barrier the primary transporter of glucose is glucose transporter-1 (GLUT1) and the transporter of BHB is monocarboxylic acid transporter-1 (MCT1). In the hibernating thirteen-lined ground squirrel, endothelial cells in capillaries and larger vessels expressed high levels of MCT1 compared with rats, as shown by immunohistochemistry (Fig. 2). MCT1 labeling of neuropil was also greater in squirrel brain than in rat brain. In both species, neuronal cell bodies were devoid of MCT1 transporter. Therefore, the chief difference in transporter levels between the hibernating and nonhibernating species is the extraordinary amount of MCT1 in the cerebral endothelium of the hibernator. This strongly suggests a greater capacity for the transport of ketones in the ground squirrel vs. the rat. GLUT1 was found in the cerebral endothelium and neuropil of both ground squirrel and rat. However, unlike MCT1, the level of endothelial and neuropil GLUT1 does not differ greatly between the hibernating ground squirrel and the rat (Fig. 2).

Fig. 2.

Immunohistochemistry of monocarboxylic acid transporter 1 (MCT1) and glucose transporter 1 (GLUT1) in rat and thirteen-lined ground squirrel (GS) brains. Dark staining shows site of immunolocalization. Major difference in transporter levels between the hibernating and nonhibernating species is the extraordinary amount of MCT1 in the blood vessels of the hibernator. MCT1-GS, MCT1 in cerebral cortex of torpid ground squirrel. MCT1-Rat, MCT1 in rat cerebral cortex. GLUT1-GS, GLUT1 in cerebral cortex of torpid ground squirrel. GLUT1-Rat, GLUT1 in rat cerebral cortex.

Induction of MCT1 expression at the blood-brain barrier was investigated to determine whether the increased levels of circulating BHB (Fig. 1) coincided with an increase in the capacity for transport across the blood-brain barrier. MCT1 immunohistochemistry of brain tissue from active (August, October, April) and torpid (December) squirrels was quantified to determine whether significant differences existed in MCT1 expression in neuropil and vessels throughout the hibernation season. In the thirteen-lined ground squirrel, endothelial cells in capillaries and larger vessels showed differential expression of the MCT1 transporter (Fig. 3A). Relative amounts of MCT1 throughout the hibernation season are unchanged in the neuropil, but significantly higher levels are seen in vessels during torpor (Fig. 3B). Therefore, the increase in circulating BHB in torpid animals is accompanied by an increase in MCT1, with the highest measured MCT1 levels occurring in December torpid animals.

Fig. 3.

Seasonal and/or animal activity-dependent expression of MCT1 in the brain of thirteen-lined ground squirrels. A: immunohistochemistry of MCT1 in thirteen-lined ground squirrel (GS) brains during the hibernation season, where dark staining shows site of immunolocalization. The scale bar applies to all panels and equals 100 μm. AUG, active animal in August; OCT, active animal in October; DEC, torpid animal in December; APR, active animal in April; cb, cell bodies; np, neuropil; v, vessels. B: graph of MCT1 optical density at the same months and activity states shown in panel A. Bars indicate mean MCT1 levels (n = 4 for each month). Error bars represent the standard error of the mean. MCT1 level differs significantly in vessels across all four time points (ANOVA: F_3,12 = 49.54, P < 0.001). Bars not connected by the same letter (a or b) are significantly different according to Tukey's HSD test with Q = 2.71 and P < 0.05. Tukey's HSD post hoc analysis shows that the level of MCT1 was significantly higher in vessels of December-torpid (DEC-Torpid) animals than in vessels from the other three active groups.

¹³C MRS of brain extracts.

Fuel utilization in hibernating thirteen-lined ground squirrels was studied by intraperitoneal injections of 1 M [1-¹³C] glucose (¹³C-glucose) or 1 M [2,4-¹³C₂] d-β-hydroxybutyrate (¹³C-BHB) followed by high-resolution NMR analysis of labeled metabolites. Spectra of labeled metabolites derived from ¹³C-labeled glucose and BHB in the brain of hibernating animals were obtained (Figs. 4 and 5). Both glucose and BHB were transported into the brain at body temperatures ranging from 10 to 38°C. BHB was readily metabolized as shown by the incorporation of ¹³C label into several brain amino acid resonances connected to the TCA cycle (Fig. 4A) such as glutamate, C4, C3, C2; and glutamine, C4, C3, C2, The glutamate; C4 resonance was the most highly labeled metabolite, consistent with the fact that glutamate C4 is the first carbon that becomes labeled following injection of [2,4-¹³C₂] β-hydroxybutyrate. The incorporation of ¹³C label in both glutamate and glutamine shows that the brain does not only take up BHB but that it is also efficiently metabolized via the TCA cycle.

Fig. 4.

¹³C-labeling of brain metabolites following injection of torpid (T_b = 5–7°C) ground squirrels with ¹³C-BHB. A: spectra of ¹³C-labeled metabolites in the 20 to 60 ppm region assayed on a 14.1 Tesla UNITY INOVA spectrophotometer (Varian, Palo Alto, CA). Body temperature at the time of death is indicated next to the respective spectra. Abbreviations for metabolites and the specific carbon labeled: BHB C2, β-hydroxybutyrate C2; Glu C4, glutamate C4; BHB C4, β-hydroxybutyrate C4. B: graph showing the relative levels of labeled metabolites normalized to natural abundance ¹³C-myo-inositol which is not labeled by injection of ¹³C-BHB. Metabolites shown on the x-axis are listed left to right in the order they are found on spectra from highest to lowest parts per million, respectively. Labeled BHB C2 and C4 have been combined as a single BHB value. Body temperature (Tb) range of animals at the time of death and number of animals (n) are defined in the upper right corner of the graph. Error bars show standard error of the mean. Asp, aspartate; BHB, β-hydroxybutyrate; Cr, creatine; GABA, gamma-aminobutyric acid; Gln, glutamine; Glu, glutamate; Lac, lactic acid.

Fig. 5.

¹³C-labeling of brain metabolites following injection of torpid (T_b = 5–7°C) ground squirrels with ¹³C-glucose. A: Spectra of ¹³C-labeled metabolites assayed on a 14.1-Tesla UNITY INOVA spectrophotometer (Varian, Palo Alto, CA). Body temperature at the time of death is indicated next to the respective spectra. myoIns, myo-inositol; Lac C3, lactate C3; Glc C1α, glucose C1 alpha; Glc C1β, glucose C1 beta; Glu C4, glutamate C4. B: graph showing the relative levels of labeled metabolites normalized to natural abundance ¹³C-myo-inositol, which is not labeled by injection of ¹³C-glucose. Metabolites shown on the x-axis are listed left to right in the order they are found on spectra from highest to lowest parts per million, respectively. Labeled Glc C1α and C1β have been combined as a single Glc C1 value. Body temperature (Tb) range of animals at the time of death and number of animals (n) are defined in the upper right corner of the graph. Error bars show standard error of the mean. Ala, alanine; Glc, glucose; others are defined in Fig. 4B.

Labeled glucose entered the brain over a variety of body temperatures (Fig. 5A) but showed little metabolism since there is only minor labeling of metabolites. There was little or no formation of labeled lactate C3 (a glycolytic product), and metabolites derived from the TCA cycle, such as glutamate and glutamine, were barely detectable (Fig. 5B). This result indicates that glucose is transported into the brain but metabolism via glycolysis, and/or the TCA cycle, is greatly reduced compared with labeling in summer active ground squirrels during August. Incorporation of ¹³C-glucose in summer active animals (T_b = 34.7–38.5°C) shows high levels of glutamate C4, as well as elevated levels of glutamate C3 and C2 during a brief 20-min labeling period (data not shown). This pattern of ¹³C-glucose labeling in summer active ground squirrels not only differs from that in torpid winter animals, but also resembles labeling seen in the brains of normothermic rats (9, 11).

¹³C MRS of heart extracts.

Spectra were collected from tissue extracts of the heart of hibernating ground squirrels injected with either 1 ml of 1 M ¹³C-BHB or 1 ml of 1 M ¹³C-glucose. As in the brain, labeled ¹³C-BHB was transported into the heart effectively during the hibernation season (Fig. 6A). The heart appears well adapted for using ketones as shown by the accumulation of metabolites derived from the TCA cycle in BHB-injected animals. Glutamate C4 levels were high with the glutamate C4 resonance showing a distinct fine-structure multiplet, indicating a high isotopic enrichment of the neighboring glutamate C3 carbon as well.

Fig. 6.

¹³C-labeling of heart metabolites following injection of torpid (T_b = 5–7°C) ground squirrels with ¹³C-BHB. A: spectra of ¹³C-labeled metabolites in the 20 to 60 ppm region assayed on a 14.1 Tesla UNITY INOVA spectrophotometer. Body temperature at the time of death is indicated next to the respective spectra. Abbreviations for metabolites and the specific carbon labeled: Tau C1, taurine C1; BHB C2, β-hydroxybutyrate C2; Tau C2, taurine C2; Glu C4, glutamate C4; BHB C4, β-hydroxybutyrate C4. B: graph showing the relative levels of labeled metabolites normalized to natural abundance ¹³C-taurine, which is not labeled by injection of ¹³C-BHB. Metabolites shown on the x-axis are listed left to right in the order they are found on spectra from highest to lowest parts per million, respectively. Labeled BHB C2 and C4 have been combined as a single BHB value. Body temperature (Tb) range of animals at the time of death and number of animals (n) are defined in the upper right corner of the graph. Error bars show standard error of the mean. Metabolite abbreviations are defined in the legend of Fig. 4B.

In contrast, the amounts of TCA cycle-derived metabolites generated by ¹³C-glucose are very small with no multiplets. This low incorporation of ¹³C label into amino acids via the TCA cycle with ¹³C-glucose was seen even in a ground squirrel that had a normothermic T_b of 35.3°C at time of death (Fig. 7A). Clearly, high levels of glucose enter the heart at a wide range of temperatures during transition from torpor to euthermia, but glycolytic products do not enter the TCA cycle. ¹³C-label is found in lactate C3 (Fig. 7B), but not lactate C2, as expected with glycolytic breakdown of [1-¹³C]-glucose. Accumulation of lactate is predicted by the upregulation of pyruvate dehydrogenase kinase isoenzyme 4 (PDK4) mRNA (2) and PDK4 protein (3) in the heart during hibernation. PDK4 phosphorylation of pyruvate dehydrogenase blocks the conversion of pyruvate to acetyl-CoA and thus results in an accumulation of lactate, rather than glycolytic products advancing through the TCA cycle.

Fig. 7.

¹³C-labeling of ground squirrel heart metabolites following injection of torpid (T_b = 5–7°C) ground squirrels with ¹³C-glucose. A: spectra of ¹³C-labeled metabolites assayed on a 14.1-Tesla UNITY INOVA spectrophotometer. Body temperature at the time of death is indicated next to the respective spectra. Abbreviations for metabolites and the specific carbon labeled Tau C1, taurine C1; Tau C2, taurine C2; Lac C3, lactate C3; Glc C1α, glucose C1 alpha; Glc C1β, glucose C1 beta; Glu C4, glutamate C4. B: graph showing the relative levels of labeled metabolites normalized to natural abundance ¹³C-taurine, which is not labeled by injection of ¹³C-glucose. Metabolites shown on the x-axis are listed left to right in the order they are found on spectra from highest to lowest parts per million, respectively. Labeled Glc C1α and C1β have been combined as a single Glc C1 value. The T_b range of animals at the time of death and the number of animals (n) are defined in the upper right corner of the graph. Error bars show standard error of the mean. Ala, alanine; Glc, glucose; others are defined in Fig. 4B.

Fuel preference and role of temperature.

It is well established that labeled carbons from both [1-¹³C] glucose and [2,4-¹³C₂] d-β-hydroxybutyrate enter the TCA cycle as [2-¹³C] acetyl CoA. The singly labeled acetyl group combines with oxaloacetate to form citrate labeled at the C4 position, which is subsequently converted to [4-¹³C] α-ketoglutarate. Labeled α-ketoglutarate can exit the TCA cycle via transamination to form Glu C4, which becomes the first TCA cycle-derived metabolite detected by ¹³C-NMR in the rodent heart (35) and brain (11). Glu C4 levels obtained in this study (Figs. 4–7) were used to quantify substrate preference and the influence of body temperature in heart and brain of thirteen-lined ground squirrels during arousal from torpor (Fig. 8). Levels of Glu C4 were consistently higher in ¹³C-BHB injected individuals than in ¹³C-glucose injected individuals in both brain (Fig. 8A; ANOVA: F_1,33 = 35.81, P < 0.001) and heart (Fig. 8B; ANOVA: F_1,30 = 33.60, P < 0.001). Interestingly, in the brain, an increase in body temperature results in higher levels of Glu C4 generated from both substrates (Fig. 8A); however, body temperature did not significantly influence the level of Glu C4 in the heart (Fig. 8B) when either ¹³C-BHB (F_1,15 = 3.3, P = 0.09) or ¹³C-glucose (F_1,14 = 3.5, P = 0.08) were injected. We conclude that body temperature plays a significant role in substrate utilization in the brain, but not the heart, and that BHB is the preferred fuel over glucose in both organs during arousal from torpor.

Fig. 8.

Level of [4-¹³C] glutamate (Glu C4) after injection with ¹³C-BHB or ¹³C-glucose. A: brain level of ¹³C-Glu C4 relative to natural abundance ¹³C-myo-inositol plotted as a function of T_b at time of death (BHB, n = 18; glucose, n = 17). Relative level of Glu C4 increased significantly with T_b following injection of ¹³C-BHB (according to the equation 0.11 + 0.24 × T_b; R² = 0.30; F_1,16 = 6.86, P < 0.05) and ¹³C-glucose (according to the equation 0.21 + 0.04 × T_b; R² = 0.26; F_1,15 = 5.25, P < 0.05). In the brain, the slope of the relationship between Glu C4 level and T_b was significantly greater for ¹³C-BHB-injected individuals than for those injected with ¹³C-glucose (test for equality of slope: F_1,31 = 4.61, P < 0.05). B: heart level of ¹³C-Glu C4 relative to natural abundance ¹³C-taurine is plotted as a function of T_b at the time of death (BHB, n = 17; glucose, n = 16). Body temperature did not significantly influence the level of Glu C4 in the heart of arousing animals when either ¹³C-BHB or ¹³C-glucose was injected. Slopes of the best-fit regression lines for Glu C4 level and T_b are not significantly different from a slope of zero for ¹³C-BHB (F_1,15 = 3.3, P = 0.09) and ¹³C-glucose (F_1,14 = 3.5, P = 0.08). A and B: open circles represent Glu C4 levels of individual animals injected with ¹³C-BHB, and the solid line represents the best-fit linear regression through these points. Solid squares represent Glu C4 levels of individual animals injected with ¹³C-glucose, and the dotted line represents the best-fit linear regression through these points. In both brain and heart, the mean level of Glu C4 resulting from ¹³C-BHB injection was significantly higher than that from injection with ¹³C-glucose at all body temperatures (ANOVA: F_1,33 = 35.81, P < 0.001 and F_1,31 = 29.62, P < 0.001, brain and heart, respectively).

DISCUSSION

In this paper we report on the transport and utilization of two different metabolic substrates during the physiological extremes of mammalian hibernation; one fat-derived, the other a carbohydrate. We showed that metabolic substrates BHB and glucose can enter the heart and brain under conditions of depressed body temperature and reduced blood flow, but their utilization is highly selective. This finding suggests that fuel selection is controlled at the level of individual metabolic pathways during hibernation. Hibernating thirteen-lined ground squirrels use lipids stored in their white adipose tissue as the primary source of fuel. Metabolism of fatty acids in the liver produces BHB—an amazingly versatile ketone body that is small, highly soluble, and used as an energy substrate in a variety of tissues. Figure 1 shows that torpid squirrels experience profound elevation of BHB in their blood. Ketosis in this circumstance is believed to be a normal physiological condition. No evidence of metabolic acidosis is observed, most likely because insulin and glucagon levels are balanced and regulated by the pancreas and because compensatory mechanisms of the respiratory and renal organs maintain bicarbonate buffering.

Serum glucose concentrations are depressed during torpor, but increase to higher levels during activity, including IBAs (Fig. 1). In vivo high-field proton NMR (9.4 Tesla) of active and hibernating ground squirrel brains also showed an increase in brain glucose concentrations during IBAs with no increase in lactate (10). The increase in serum glucose levels during IBAs is intriguing because our experimental animals did not consume food for periods ranging from 1.5 to 4.5 mo. An explanation for these brief spikes in glucose is possibly due to gluconeogenesis, which increases during the warmer body temperatures of an IBA (7). Substrates for gluconeogenesis during hibernation include glycerol liberated via lipolysis of triacylglycerols and lactate (30). Another means of maintaining glucose homeostasis during hibernation is up-regulation of pyruvate dehydrogenase kinase 4 (2, 3), which blocks glycolytic flow into the TCA cycle by inhibiting the conversion of pyruvate to acetyl-CoA. The result of this inhibition is the generation of the gluconeogenic fuel lactate, as shown in the heart of animals injected with ¹³C-labeled glucose (Fig. 7).

Our results show fuel utilization at various stages of arousal from the torpid state because we quantified transport and metabolism at body temperatures from 10 to 38°C. During hibernation, the primary difference in utilization of BHB vs. glucose is that there was little or no labeling of metabolites derived from the TCA cycle in ¹³C-glucose-infused animals. Despite the fact that [2,4-¹³C₂] β-hydroxybutyrate has two labeled carbons, and [1-¹³C] glucose has one, the difference in levels of labeled TCA cycle derivatives glutamate and glutamine exceeded a twofold difference with ¹³C-BHB vs. ¹³C-glucose as shown for Glu C4 in Fig. 8. As the animal arouses, BHB is efficiently utilized as ¹³C-labeled BHB was consumed and corresponding levels of ¹³C-labeled metabolites increased (Figs. 4 and 6). This is in stark contrast to glucose, which entered the brain (Fig. 5) and heart (Fig. 7) but remained intact at high levels with very little labeling of metabolites derived from the TCA cycle, even at warm body temperatures (Fig. 8). This differential fuel utilization may serve as a means to conserve glucose due to the long-term fasting that accompanies hibernation.

The metabolic switch resulting in a preference of BHB over glucose during arousal from torpor is due to several factors. We have shown that serum levels of BHB are elevated during torpor and that glucose levels are depressed (Fig. 1). Higher levels of BHB in ground squirrel liver (28), as well as an increase in the level of the ketone producing enzyme hydroxy methyl glutaryl Coenzyme A synthase 2 in liver (6) and intestine (18), indicates heightened ketone production during hibernation. Adaptive mechanisms for BHB utilization during hibernation include induction of the BHB transporter MCT1 in the brain (Fig. 3) and a six-fold increase in the level of the rate-limiting enzyme for ketolysis in the heart, succinyl CoA transferase (27). Warmer body temperatures during arousal accelerate BHB uptake from the blood and may explain the reduction in serum BHB levels during IBAs (Fig. 1A). Figure 9 is a model showing BHB utilization and glucose conservation in the heart and brain during hibernation based on our current and previous results. Ketone bodies such as BHB are released from the liver into the blood to fuel various organs, including heart and brain. Our data show that both BHB and glucose enter the heart and brain during torpor, but unlike glucose, BHB is metabolized via the TCA cycle, does not generate lactic acid, and is produced by catabolism of stored fat. Seasonal and/or torpor-induced adaptive mechanisms, such as MCT1 in brain and succinyl CoA transferase in heart, facilitate strategic use of BHB as a preferred fuel during hibernation.

Fig. 9.

Model showing the mechanism of β-hydroxybutyrate utilization and glucose conservation in the heart and brain during hibernation. Long solid lines with arrowheads indicate active metabolic pathways, dashed lines with arrowheads indicate pathways with reduced activity, and a solid block across a line indicates pathway stoppage. Short vertical arrows pointing up indicate an increase in concentration or activity. Short vertical arrows pointing down indicate a decrease in concentration or activity. BHB, beta-hydroxybutyrate; MCT1, monocarboxylic acid transporter 1; PDK4, pyruvate dehydrogenase kinase 4; SCOT, succinyl CoA transferase; TCA cycle, tricarboxylic acid cycle.

Elevation of ketone levels is also seen in starved hypothermic mice (12), and pathways of BHB utilization in the brain of rodents and humans appear to be substantially similar. A minimum of four proteins is responsible for the metabolism of ketone bodies, including MCT1 for transfer across cell membranes, and three enzymes linking the breakdown of BHB to acetyl-CoA prior to entry into the TCA cycle (20, 26, 31, 33). Expression of MCT1 protein in rat brain endothelial cells is regulated dynamically and is 25 times more abundant in suckling infants than adults. In addition, adult rats fed a ketogenic diet show an eight-fold induction of MCT1 after 4 wk (15, 16). Induction of MCT1 by high-fat diets resembles natural hibernation in which lipid is the major source of fuel (Fig. 3). In contrast to MCT1, the three enzymes for BHB metabolism appear to be constitutively expressed at a relatively high levels in rats, even when ketones are minimal (14, 21, 24, 34). Thus, brain energy metabolism may be fueled by ketones whenever substrates are sufficiently available.

As with glucose, catabolism of a single BHB molecule yields two molecules of acetyl-CoA that enter directly into the TCA cycle. However, BHB metabolism does not generate lactic acid and thus avoids the lactic acidosis that can occur with glycolytic breakdown of glucose. Furthermore, in contrast to pyruvate generated by glycolysis, metabolism of BHB has ∼31% greater energy yield as a result of the redox state of the mitochondrial respiratory chain complexes (32). The enhanced proton gradient across the mitochondrial membrane resulting from BHB metabolism allows for more energy to be available for ATP synthesis. The theoretical and empirical basis for this phenomenon was reported previously (33). The net result is that ATP levels may be maintained with less oxygen consumption, a distinct advantage for hibernation or conditions of metabolic stasis. The benefit of applying these principles to sustaining cell viability in humans during episodes of hypoxia and ischemia is underutilized and should be explored in future studies.

Perspectives and Significance

The biochemistry of hibernating mammals is remarkably similar to that of nonhibernators with the exception of regulatory events, allowing adaptation to a variety of environmental and physiological extremes. Identifying biochemical pathways that differ between the hibernating and active state within the same species has been used as a means to determine the mechanism of this adaptation. By studying the differential expression of genes and proteins, as well as changes in the levels of small molecules, a larger picture of how an animal induces and maintains the hibernating state has begun to emerge. The study described in this paper shows that ketones are the preferred fuel source when an animal arouses from torpor. During arousal, there is huge potential for reperfusion injury as the heart rate explodes from single digits to over 300 beats/min, but no injury occurs. The arousing hibernator also bridges the gap from hypothermia to normothermia, from minimal O₂ consumption to maximal O₂ consumption, yet the animal balances these extremes without physiological upset. The foundation of this balancing act is the molecular biology of adaptation played out in various organs throughout the body. Applying the lessons of hibernation molecular biology to the field of human medicine offers the advantage of millions of years of mammalian evolution and has the potential to identify novel therapies for pathologies such as stroke, atrophies from muscle and bone disuse, and hemorrhagic shock.

GRANTS

This work was funded by Defense Advanced Research Projects Agency (DARPA) contract W911NF-06-0088 to M. T. Andrews and National Institutes of Health grants R15HL08110 to M. T. Andrews and R01NS38672 to P. G. Henry.