Caloric Restriction Impedes Age-Related Decline of Mitochondrial Function and Neuronal Activity

Abstract

Caloric restriction (CR) prolongs lifespan and retards many detrimental effects of aging, but its effect on brain mitochondrial function and neuronal activity—especially in healthy aging—remains unexplored. Here we measured rates of neuronal glucose oxidation and glutamate–glutamine neurotransmitter cycling in young control, old control (i.e., healthy aging), and old CR rats using in vivo nuclear magnetic resonance spectroscopy. We found that, compared with the young control, neuronal energy production and neurotransmission rates were significantly reduced in healthy aging, but were preserved in old CR rats. The results suggest that CR mitigated the age-related deceleration of brain physiology.

INTRODUCTION

Mitochondrial oxidative phosphorylation of glucose is the predominant mode of energy generation (adenosine triphosphate (ATP) production) in mammalian species. The mammalian brain has the highest energy demands of any organ based on its size, ¹ and majority of this energy is used to support neuronal activity and functional processes. ² Mitochondrial function declines with age in the brain and has been proposed to be a major factor in the loss of brain function with aging. The metabolic decline is even more rapid and profound in neurodegenerative disorders, such as Alzheimer's disease. ³ As a result, preserving brain mitochondrial integrity and metabolism with age could be critical for maintaining healthy brain function, often referred to as healthspan, and for extending lifespan. ⁴

Caloric restriction (CR) without malnutrition is one of several other interventions that have been introduced to preserve metabolism in aging process, and moreover CR has been shown to increase the lifespan of a broad range of species.^5,6 Several studies suggest that CR-induced increase in lifespan arises because of increased capacity for oxidative phosphorylation from elevated mitochondrial respiration. ⁷ Although CR effects on isolated mitochondria have been extensively studied, we posited that the effect of CR on brain mitochondrial function could be because of altered neuroenergetics. In this study, we determined whether CR could mitigate the declines of these measures in aging brain. We measured fluxes of neuronal tricarboxylic acid (TCA) cycle (index of mitochondrial function) and glutamate–glutamine neurotransmitter cycling (index of neuronal activity) using nuclear magnetic resonance (NMR) spectroscopy in vivo.

MATERIALS AND METHODS

Animal

Experiments were conducted with male Fischer 344 Brown-Norway F1 (F344BNF1) rats, which have shown to extend longevity under CR. ⁸ Rats were obtained from a CR colony at National Institute on Aging. At the CR colony of National Institute on Aging, all rats were fed ad libitum (NIH-31 diet) until 14 weeks of age. Then a group of rats were separated with CR diet. The CR regimen (NIH-31 fortified, see more details in Table 1) was initiated at 14 weeks of age at 10% restriction of food intake, increased to 15% restriction at 15 weeks, and up to 40% restriction at 16 weeks from which point food intake was maintained throughout the life of the animal (http://www.nia.nih.gov/research/dab/aged-rodent-colonies-handbook/caloric-restricted-colony). We ordered the ad libitum control rats at 5 months of age (N = 5) and 24 months of age (N = 6), and CR rats at 24 months of age (N = 6), all from the CR colony of National Institute on Aging. After arriving at our institutional facilities, rats were housed individually for a week in a specific pathogen-free facility and were fed the same diet everyday 1 hour before the onset of the dark cycle. All experimental procedures were approved by Institutional Animal Care and Use Committee at Yale University according to NIH guidelines.

Table 1.

Age, gender, diet, body weight, blood glucose, and ATP production of the rats

Animals	Age/gender	Diet	Body weight (g)	Blood glucose (mg/dl)	Neuronal ATP production (mmol/g/minute)
YC	5 Months/male	NIH-31	335 ± 8	112 ± 6	18.3 ± 0.5
OC	24 Months/male	NIH-31	540 ± 12 **	137 ± 9 **	9.2 ± 0.9 **
OCR	24 Months/male	NIH-31/NIA fortified a	380 ± 9	118 ± 8	14.5 ± 0.7*
F	—	—	27.3 **	24.6 **	32.8 **
Post hoc	—	—	YC = OCR << OC	YC = OCR << OC	OC < OCR << YC

ATP, adenosine triphosphate; CR, caloric restriction; NIA, National Institute on Aging; OC, old control; OCR, old caloric restriction; YC, young control. Data are presented as mean ± s.e.m. F values were computed by one-way, repeated measures ANOVA,

**

P < 0.01. Post-hoc testing was performed by Newman–Keuls test, where << indicates P < 0.01, < indicates P < 0.05, and = indicates P > 0.5.

a

Diet changed from 16 weeks of age. Compared with NIH-31, NIH-31/NIA fortified diet has 40% less calorie, but with complementary increases of vitamins, including vitamins A, D3, E, and B12. All diet were purchased from NIA.

Animal Preparation

All rats were fasted for 12 hours before scanning to decrease the plasma glucose level (to ˜5 to 6 mmol/L) such that upon infusion of ¹³C-labeled glucose the plasma level was approximately doubled (to ˜10 to 12 mmol/L) and hence the ¹³C-enrichment was ˜50% with [1,6-¹³C]-D-glucose. The animals were initially anesthetized with isoflurane (1% to 2%), tracheotomized, and ventilated (30% oxygen; ˜68% nitrous oxide). After surgery anesthesia was maintained with α-chloralose (initial 80 mg/kg, plus 40 mg/kg/hour) and D-tubocurarine-Cl (0.3 mg/kg) was administered for immobilization. The left femoral artery and vein were catheterized for continuous monitoring of arterial blood pressure and blood sampling, and for infusion of [1,6-¹³C]-D-glucose, respectively (Cambridge Isotopes, Andover, MA, USA). Blood gases (pCO₂, pO₂, and pH) were measured periodically in arterial blood samples (ABL5 blood gas system, Radiometer America, Westlake, OH, USA). Animal core body temperature was maintained near ˜37°C using a heating pad connected to a temperature-regulated water bath.

in vivo NMR Spectroscopy

We used in vivo proton-observed carbon-edited (POCE; 1 H-[¹³C]) NMR, which enables dynamic detection of mitochondrial function and neurotransmission rate. ⁹ The advantages of POCE over conventional application of ¹³C NMR (i.e., direct ¹³C detection) are twofold: ¹⁰ (1) greater sensitivity because it measures the ¹³C side bands of protons directly bound to ¹³C nuclei in the ¹H spectrum; and (2) the ability to resolve resonance from protons bound to both ¹²C and ¹³C nuclei, the latter allowing direct measurement of ¹³C-enrichment. Furthermore, POCE has the advantage over more conventional measures of glucose uptake to allow cell-type-specific (neurons, glia) metabolic rates to be measured as well as glutamate–glutamine cycling. POCE (or ¹³C NMR) studies often use [1-¹³C]glucose or [1,6-¹³C]glucose infusion to observe ¹³C turnover into various metabolites (e.g., glutamate, glutamine, aspartate, etc) in real time, from which metabolic fluxes can be extracted with compartmental modeling of neuronal-glial trafficking of metabolites (see below). With [1,6-¹³C]glucose infusion, in the first pass of the TCA cycle glutamate is labeled first in the C4 position (mainly in neurons) followed by glutamine in the C4 position (mainly in glia), whereas in the second pass of the TCA cycle glutamate is labeled initially in the C3 position (mainly in neurons) followed by glutamine in the C3 position (mainly in glia). From the in vivo POCE data it is possible to separate the C4 resonances of glutamate and glutamine, whereas the C3 resonance of glutamate is distinguishable. However, the C3 resonance of glutamine is derived by LCModel of in vivo POCE data in conjunction with POCE data of brain extracts (see below).

in vivo POCE spectra were obtained on an 11.7 T Agilent (Santa Clara, CA, USA) horizontal-bore spectrometer using a 14-mm-diameter surface coil tuned to proton frequency (499.8 MHz) and positioned on top of the animal head. The radio frequency decoupling on ¹³C (125.7 MHz) was achieved with two orthogonal 21-mm-diameter surface coils driven in quadrature mode and positioned on both sides of the animal head at 45° relative to the ¹H coil. The POCE spectra were obtained from a localized volume (8 × 4 × 6 mm³; Figure 1A). A total number of 288 POCE spectra (144 with ¹³C inversion and 144 without ¹³C inversion, each with 8 averages, required for POCE) were acquired and stored using a 6-second repetition time (with 3 seconds alternate on/off ¹³C inversion). However, the signal to noise of each individual spectrum was too low to allow accurate quantification of NMR intensities. Therefore, to increase the signal to noise and obtain better quantification of NMR signals, we averaged every 16 spectra (8 with ¹³C inversion and 8 without ¹³C inversion), which resulted in data points every 6 minutes and 24 seconds (see Figure 1C).

Figure 1.

(A) The imaging voxel (including primarily the cortex) for the POCE (proton-observed carbon-edited) experiment. (B) Time-resolved ¹H[¹³C] or POCE-spectra from rat brain in vivo after the onset of [1,6-¹³C]-glucose infusion, from 0 (beginning) to 120 minutes (end point). (C) Time courses of ¹³C enrichments of Glu-C4 and Gln-C4, and best fits of the metabolic model for individual rats in the young control, old control, and old caloric restriction (CR) groups during [1,6-¹³C]-D-glucose infusion. (D) and (E) Quantitative values of V_TCA,N and V_cycle, respectively of the three groups of rats: young control (5 months, N = 5, V_cycle/V_TCA,N = 0.23 ± 0.02/0.49 ± 0.04 ≈ 0.47), old control (24 months, N = 6, V_cycle/V_TCA,N = 0.13 ± 0.01/0.25 ± 0.04 ≈ 0.52), and old CR (24 months, N = 6, V_cycle/V_TCA,N = 0.21 ± 0.01/0.39 ± 0.06 ≈ 0.54) rats. Data are presented as mean ± s.e.m. *P < 0.05, **P < 0.01, and ***P < 0.001, ns: non-significant, P > 0.5. (F) V_TCA,N plotted in hexose units (CMR_glc(ox),N) versus V_cycle. CMR_glc(ox),N and V_cycle showed a linear correlation among the three groups (r = 0.88, P < 0.01): young control (5 months, N = 5), old control (24 months, N = 6), and old CR (24 months, N = 6). The slope between CMR_glc(ox),N and V_cycle was 1.01, which is similar to that reported in previous studies of rat cerebral cortex.^16,17

During the scan, each rat received 1.5 mL of 0.75 mol/L [1,6-¹³C]-D-glucose infusion. The determination of glutamate and glutamine turnover required rapid attainment of a high and steady fractional enrichment of [1,6-¹³C]glucose in the blood throughout the time period of the experiment. The infusion protocol was optimized by adapting the 2-deoxy-D-[¹⁴C]-glucose method described by Patlak and Pettigrew ¹¹ to accommodate the large amount of [1,6-¹³C]-glucose infused during the experiment. The infusion rate, which was modified manually every 30 seconds, followed a decreasing exponential function during the first 8 minutes and was constant for the remaining period. ¹⁰ Total blood glucose concentration was measured from arterial blood samples (50 to 100 µL) drawn every 10 to 15 minutes (six samples) during the infusion with an Analox GM9D glucometer (The Vale London, UK). At the end of each experiment, the head and brain of the anesthetized animal were frozen in situ with liquid N₂. Brain extracts were prepared to obtain the end point of ¹³C enrichments of metabolites (see below).

Preparation of Blood Plasma and Brain Extracts

Plasma samples were prepared for further POCE spectral analysis by adding 200 µL of 2.5 mmol/L formic acid and 0.25 mmol/L TSP (3-trimethylsilyl tetradeuterated sodium propionate) and 400 µL 100 mmol/L phosphate buffer (pH 7) to 50 µL blood plasma. The solution was passed through a microcentrifuge filter tube (10 kDa cutoff, Nanosep Centrifugal Devices, VWR, Batavia, IL, USA). Concentrations and ¹³C enrichments of glucose were determined using POCE spectroscopy with 20 seconds repetition time to achieve fully relaxed measurements of metabolite resonances. Both TSP and formic acid were used as chemical shift and concentration references, respectively.

Ethanol extracts were prepared from the frozen frontoparietal cortex (100 to 150 mg wet weight) using the procedure described by Patel et al, ¹² where [2-¹³C]-glycine (50 µL, 5 mmol/L) was added as an internal concentration reference at the beginning of tissue extraction. After centrifugation and lyophilization of the supernatant, the extract powder was suspended in 600 µL of a buffer solution containing 50 mmol/L phosphate (pH 7), 0.8 mmol/L formic acid, and 0.08 mmol/L TSP in D₂O/H₂O (2:1). The concentrations and ¹³C enrichments of glucose and other substances (such as glutamate (Glu) and glutamine (Gln) C4 and C3 resonance: Glu-C4, Gln-C4, Glu-C3 and Gln-C3) in the extracted brain samples were measured with POCE under fully relaxed conditions (i.e., repetition time of 20 seconds). Both plasma and brain extract samples were measured using a 500-MHz NMR spectrometer (Bruker Biospin Corp., Billerica, MA, USA). Because Glu-C3 in the POCE spectrum was partially overlapping with Gln-C3, the LCModel approach was used to separate them.

Metabolic Modeling

The time courses of brain amino acid ¹³C enrichments of Glu-C4, Glu-C3, Gln-C4, and the steady-state labeling of Gln-C3 (from extracts) were fitted to a two-compartment (neuron–astrocyte) metabolic model ¹³ using the plasma time courses of ¹³C-enriched glucose as the input. The goal of the metabolic modeling was to derive the rates of the neuronal TCA cycle (V_TCA,N) and the glutamate–glutamine neurotransmitter cycling (V_cycle). The relationship between ¹³C enrichments of plasma glucose and brain metabolites was described by coupled differential equations with restrictions of mass and isotope balance, using CWave 3.0 software (GF Mason, New Haven, CT, USA). The differential equations were solved in MATLAB 7.1 (Natick, MA, USA) using a first-order Runge-Kutta algorithm and fitting optimization was achieved with the Levenberg–Marquardt algorithm. ¹³ In this model, glutamate was assigned 10% to glia and 90% to neurons, whereas glutamine was assigned 100% to glia, the astrocytic TCA cycle flux (V_TCA,A) was set to 15% of total TCA cycle flux, the anaplerotic flux through pyruvate carboxylase (V_PC) was set to 20% of the rate of glutamine synthesis (V_gln), ¹⁴ and rates of the neuronal TCA cycle (V_TCA,N) and glutamate–glutamine cycling (V_cycle) were iterated using CWave for optimal fits of the model to the data. ¹³ The ATP production rate in neurons (V_ATP,N) was calculated assuming the steady-state oxygen-to-glucose index of 5.5^15,16 and that 34 ATP is generated per mole of glucose oxidation, ²

${\mathrm{V}}_{\text{ATP},\mathrm{N}}={\text{CMR}}_{\text{glc}(\text{ox}),\mathrm{N}}\times \frac{5.5}{6}\times 34$

where CMR_glc(ox),N is the rate of neuronal glucose oxidation and is the hexose unit representation of the neuronal TCA cycle flux. Because one glucose molecule is converted to two pyruvate molecules to generate two acetyl-CoA moieties to enter TCA cycle, CMR_glc(ox),N is given by ½V_TCAN.

Statistics

Data are presented as mean ± s.e.m. One-way, repeated measures analysis of variance with post-hoc testing (Newman-Keuls test) were used to determine statistical significance (P ⩽ 0.05).

RESULTS

Figure 1A shows the cortical region from which we obtained the POCE measurements and Figure 1B shows an example from spectral time course of ¹³C labeling over the 2-hour infusion (only spectra from four time points are shown here for presentation). It is observed that ¹³C labeling of Glu-C4 appears faster than that of Gln-C4, indicating ¹³C-glucose was mainly oxidized in neuronal compartment, and Gln-C4 was labeled after Glu-C4 via the glutamate–glutamine neurotransmitter cycling. Figure 1C depicts time courses of cortical ¹³C enrichments of Glu-C4 and Gln-C4 for a typical young control, old control, and old CR rats during [1,6-¹³C]-glucose infusion. The solid lines reflect the best fits of the constrained two-compartment metabolic model to the in vivo POCE data, which yielded the absolute rate estimates of glucose oxidation in neurons and the neurotransmitter flux in agreement with prior reports. ¹⁷ Figure 1D shows bar graphs of the rates of the neuronal TCA cycle (V_TCA,N) and Figure 1E shows the glutamate–glutamine cycling (V_cycle) for each group. We found significant differences of V_TCA,N and V_cycle between young control rat versus old control rat as well as old control rat versus old CR rat. Notably, both V_TCA,N and V_cycle dramatically declined in normal aging rats. Compared with the young control rats, the old control rats had 51% (P < 0.001) lower V_TCA,N (Figure 1D) and 58% (P < 0.001) lower V_cycle (Figure 1E). A previous human study using ¹³C MRS showed similar reductions in brain metabolism and neuronal activity with healthy aging. ¹⁸ Interestingly, the old rats treated with CR restored V_TCA,N and V_cycle by 44% (P < 0.01; Figure 1D) and 52% (P < 0.001; Figure 1E), respectively. Old CR rats had comparable V_cycle relative to that of the young control rats, but had a somewhat lower V_TCA,N. Similarly, the old control rats had significantly lower V_ATP,N compared with the young control rat, but CR retarded the decline. Caloric restriction also impeded other aging phenotypes, including changes in body weight and blood glucose level (Table 1), which were consistent with previous findings in rhesus monkey. ⁶

Previously, it has been found that, when V_TCA,N is plotted in hexose units (i.e., CMR_glc(ox),N) against V_cycle, there is a close to 1:1 relationship between increments in glucose oxidation over an isoelectric baseline state and increments in glutamate–glutamine cycling. In the metabolic model, CMR_glc(ox),N was expressed as ½V_TCA,N because one glucose molecule makes to two pyruvate molecules and which in turn generates two acetyl-CoA molecules, etc. This result has been interpreted as indicating a high energetic cost for neuronal function that is approximately constant throughout the normal activity range, especially in the awake state.^16,17 Figure 1F shows a plot of CMR_glc(ox),N versus V_cycle for each of the rats in all three groups, where the slope of the line is 1.01, showing that the best fit is highly consistent with the previous findings.^16,17

DISCUSSION

We showed that CR impeded the age-related decline of brain mitochondrial respiratory function—old CR rats had comparable V_TCA,N, CMR_glc(ox),N, and ATP production rate relative to the young control rats. Although we were not able to determine whether this is because of mitochondrial biogenesis using POCE techniques in the study, our finding is consistent with the literature that long-term CR preserves mitochondrial function and induces bioenergetic efficiency (in isolated mitochondria), possibly because of decreased oxidant emission, increased antioxidant scavenging, and/or minimized oxidative damage to DNA and protein.^7,19

We also showed that CR impeded the age-related decline of neuronal activity—old CR rats had similar V_cycle compared with the young control rats. This is consistent with prior observations that metabolic and neuronal functions were highly associated;^3,4 that metabolic reserve was a determinant of cognitive aging. ²⁰ Metabolic reserve has been proposed as the ability of neuronal circuits to respond adaptively to perturbations in energy metabolism because of aging and disease processes, thereby maintaining their ability to support neuronal circuits and preventing declines in cognition. In line with this, old rats treated with CR (same strain used in the current study) have been previously shown to have enhanced memory compared with the age-matched controls. ²¹ Moreover, rodents with CR had lower incidence of Alzheimer's disease phenotypes. ²² Our findings suggest that CR is beneficial to aging brain by preserving metabolic and thus neuronal functions.

In addition to brain function, we showed that CR preserved other physiologic functions, including body weight and blood glucose level. The dramatic changes of these measures in normal aging rats may imply a developing pre-diabetic symptom. Collectively, these observations may provide a rationale for CR-induced extended healthspan and lifespan.

It has to be pointed out that CR-induced extended longevity may not be universal. Recent studies have showed that the lifespan response to a single level of CR (e.g., 40% CR) exhibits wide variation in mice with different genetic backgrounds. ²³ They also showed that there are cases where CR can shorten lifespan in inbred mice. However, brain metabolism and function was not examined in these studies. It will be important for future studies to identify whether CR has adverse effects on brain metabolism and functions during aging in rodent strains where deleterious effects on lifespan are observed.

In summary, we used NMR to show that during aging CR preserves mitochondrial energy production, energy demand, and neuronal activity with a long-lived rodent model. These results provide a rationale for CR-induced sustenance of brain health with extended lifespan. Understanding of nutritional effects on brain function may have profound implications in human aging and other age-related neurodegenerative disorders.