Reversal of metabolic deficits by lipoic acid in a triple transgenic mouse model of Alzheimer's disease: a ¹³C NMR study

Abstract

Alzheimer's disease is an age-related neurodegenerative disease characterized by deterioration of cognition and loss of memory. Several clinical studies have shown Alzheimer's disease to be associated with disturbances in glucose metabolism and the subsequent tricarboxylic acid (TCA) cycle-related metabolites like glutamate (Glu), glutamine (Gln), and N-acetylaspartate (NAA). These metabolites have been viewed as biomarkers by (a) assisting early diagnosis of Alzheimer's disease and (b) evaluating the efficacy of a treatment regimen. In this study, 13-month-old triple transgenic mice (a mouse model of Alzheimer's disease (3xTg-AD)) were given intravenous infusion of [1-¹³C]glucose followed by an ex vivo ¹³C NMR to determine the concentrations of ¹³C-labeled isotopomers of Glu, Gln, aspartate (Asp), GABA, myo-inositol, and NAA. Total (¹²C+¹³C) Glu, Gln, and Asp were quantified by high-performance liquid chromatography to calculate enrichment. Furthermore, we examined the effects of lipoic acid in modulating these metabolites, based on its previously established insulin mimetic effects. Total ¹³C labeling and percent enrichment decreased by ∼50% in the 3xTg-AD mice. This hypometabolism was partially or completely restored by lipoic acid feeding. The ability of lipoic acid to restore glucose metabolism and subsequent TCA cycle-related metabolites further substantiates its role in overcoming the hypometabolic state inherent in early stages of Alzheimer's disease.

Introduction

Human brain consumes ∼60% of body's resting-state glucose, and the energy generated from glucose metabolism is essential for maintaining synaptic transmission.¹ Hence, disturbance of brain glucose uptake is expected to create a hypometabolic state that impinges on synaptic plasticity, because glucose availability tightly regulates glutamate (Glu) neurotransmission within the tricarboxylic acid (TCA) cycle and glutamine (Gln)–Glu cycles of neurons and glia. Reduction of brain glucose metabolism in patients with Alzheimer's disease has been shown by several clinical studies.²

Definite diagnosis of Alzheimer's disease is conventionally only possible by detection of β-amyloid plaques and neurofibrillary tangles in postmortem tissues; a clinical diagnosis is possible by testing mental status output in terms of different scales like the Mini-Mental State Examination or the Alzheimer's Disease Assessment Scale-cognitive subscale.³ However, the sensitivity of these scales is rather low, especially, during the early stages of Alzheimer's disease.⁴ Currently, early diagnosis and sensitive treatment monitoring are the important challenges that hamper effective management of Alzheimer's disease. Thus, biomarkers that would assist in accomplishing those two goals would be invaluable toward effectively managing Alzheimer's disease. It is established that biochemical changes accompanying energy metabolism (associated with N-acetylaspartate (NAA), myo-inositol (MI), Glu, Gln, and aspartate (Asp) precede the structural abnormalities in Alzheimer's disease. These developments have renewed interest in the metabolic aspect of Alzheimer's disease. In that light, magnetic resonance spectroscopy (¹H and ¹³C) has diagnostic value and aids early diagnosis of Alzheimer's disease. Using these tools, Glu, Gln, Asp, GABA, MI, NAA, and other metabolites (or neurochemicals) can be detected. These metabolites can serve as biomarkers, because they are found to be dysregulated in Alzheimer's disease.⁵ Beside diagnosis, assessment of therapeutic agents for Alzheimer's disease remains to be a challenge. Thus, there is a need for biomarkers of Alzheimer's disease that respond to the biochemical effects of the drug being tested. In this respect, magnetic resonance spectroscopy studies measuring metabolites have been shown as viable measurements to dynamically monitor therapeutic treatment.⁶

Lipoic acid (1,2-dithiolane-3-pentanoic acid), a disulfide compound, has been shown to have an insulin mimetic effect that results in increased glucose uptake and activation of the PI3K/Akt pathway, thus stimulating mitochondrial bioenergetics.^{7, 8, 9} A small clinical study following patients over 4 years, looking at the efficacy of lipoic acid in Alzheimer's disease treatment found beneficial effects in terms of stabilization of cognitive functions.¹⁰ Moreover, feeding lipoic acid increased glucose uptake and improved synaptic plasticity in the triple transgenic mouse model of Alzheimer's disease (3xTg-AD) mouse model of Alzheimer's disease,¹¹ reduced hippocampal memory deficits in the Tg2576 model of Alzheimer's disease,¹² improved long-term memory of aged NMRI mice,¹³ improved cognition in aged SAMP8 mice,¹⁴ and improved memory in aged rats.¹⁵ These multifaceted effects of lipoic acid can be mechanistically ascribed to its participation in thiol/disulfide exchange reactions that modulate the redox and energy status of the cellular environment.¹⁶

This study examines the ability of lipoic acid to modulate brain glycolytic and mitochondrial metabolic pathways in a 3xTg-AD. This mouse model closely mimics the pathology type (β-amyloid plaques and hyperphosphorylated tau resulting in neurofibrillary tangles) and synaptic deficits in an age-dependent manner as seen in humans¹⁷ and represents an advanced preclinical tool to study Alzheimer's disease and assess therapeutic efficacy of a candidate drug. However, the characterization of the metabolic components of glucose metabolism and the downstream TCA cycle-related metabolites has not been explored in this model. Experiments in this study were performed on whole brain of 13-month-old nontransgenic (nonTg) and 3xTg-AD mice; the study was aimed at (a) evaluating glucose metabolism by quantification of the subsequent TCA cycle-related metabolites, such as Glu, Gln, NAA, GABA, and Asp, and (b) assessing the ability of lipoic acid to modulate the levels of these metabolites.

Materials and Methods

Materials

[1-¹³C]Glucose (99%) was purchased from Sigma-Aldrich (St Louis, MO, USA) and deuterium oxide (99.9%) from Cambridge Isotope Laboratories (Andover, MA, USA). All other chemicals were the purest grade available from Sigma-Aldrich.

Mice Colonies and Lipoic Acid Feeding

Colonies of 3xTg-AD and nonTg mouse strain (C57BL6/129S; gift from Dr Frank Laferla, University of California, Irvine) were bred and maintained at the University of Southern California (Los Angeles, CA) following National Institutes of Health guidelines on use of laboratory animals and an approved protocol by the University of Southern California Institutional Animal Care and Use Committee. The triple transgenic mouse model of Alzheimer's disease (3xTg-AD) was first developed by Oddo et al.¹⁹ Twelve-month-old 3xTg-AD show extracellular β-amyloid deposits in the cortex and human tau immunoreactivity.¹⁹ Mice were housed on 12-hour light/dark cycles and provided ad libitum access to food and water. Twelve-month-old male mice were used for experiments. triple transgenic mouse model of Alzheimer's disease and nonTg mice were either fed with water containing 0.23% R-sodium lipoic acid (gift from Geronova Research) or normal water for 4 weeks. Thus, at the time of glucose infusion, mice were ∼13 months old.

Intravenous Glucose Infusion

The mouse to be infused was first restrained using a rotating tail vein injection restrainer (Braintree Scientific, Braintree, MA, USA). No anesthesia was used during the entire procedure to assess metabolism in nonanesthetized and awake mice; thus avoiding anesthesia-related interferences on brain metabolism. After restraining the mice, we tested the basal blood glucose levels as described below. The puncture made for testing the basal blood glucose levels was also used for inserting a vein catheter in the mouse tail (Braintree Scientific). The catheter was inserted following the manufacturer's instructions and was ensured to be in the tail vein by pushing some saline through the catheter. Any bulge at the bottom of the tail, resistance, or back flow of saline was considered as an improper insertion of catheter and performed again at a more proximal point in the same vein or the next vein was used. The glucose infusion protocol,¹⁸ previously shown in rat to achieve steady-state blood glucose concentration and brain glucose ¹³C enrichment rapidly, was slightly modified by scaling down glucose concentrations according to the weight of the mouse. In short, infusion consisting of a bolus, followed by exponentially decreasing amount of glucose for 8 minutes, and finally, infusion at a constant rate was performed for different durations (0, 8, 30, 60, and 120 minutes) to determine the time point for steady-state-like concentration. Sixty minutes were determined to be the approximate time point to reach the steady state for most metabolites in the nonanesthetized mice. Thus, further constant infusions were carried out for 60 minutes. Mice were kept in a quiet and warm environment to avoid too much stress during the glucose infusion. The constant infusion was carried out using a pump from Bio-Rad Laboratories, Hercules, CA, USA. The labeling pattern for [1-¹³C]glucose has been well described earlier.¹⁹

Tissue Collection and Extraction

The infusion was stopped after 60 minutes and catheter removed, followed by testing for final blood glucose levels as described below. The mouse was then immediately taken to a cold room followed by decapitation and quick removal of whole brain and dropping it in liquid nitrogen. The total time taken from end of infusion to start of decapitation and the total time taken from decapitation to dropping the whole brain in liquid nitrogen was ensured to be less than 1 minute each for all mice (to minimize the postmortem metabolic changes). Ischemia is expected to rapidly change the lactate concentrations due to decapitation and thus, relative concentrations were used to make any inferences about lactate metabolism. After freezing of the brain, it was weighed and perchloric acid extraction was carried out as described previously.²⁰ Briefly, frozen brain was powdered using a mortar and pestle (while keeping the brain cold by constantly adding liquid nitrogen). Powdered brain was mixed with 600 μL perchloric acid and centrifuged at 22,000 g for 20 minutes using a microcentrifuge, followed by neutralization of the supernatant with potassium hydroxide. After centrifugation at 22,000 g for 20 minutes (to remove precipitates of potassium perchlorate), the final brain extract supernatant was stored in −80 °C freezer until used for nuclear magnetic resonance (NMR) or high-performance liquid chromatography analysis. Weighing was carried out at each step to calculate the neutralization factor for each mouse brain extraction.

Blood Glucose Levels

Briefly, mice were fasted overnight (10 to 12 hours); its tail was warmed slightly using a lamp or a heating pad. The tail vein was located and a small puncture was made with a 25 G needle. The drops of blood that oozed out were tested for the blood glucose levels using a glucose meter and strips (Abbott, Green Oaks, IL, USA) as per manufacturer's supplied instructions. Blood concentrations were considered basal if they were below 45 mg/dL after fasting overnight. After glucose infusion, the levels typically rose to ∼150 to 200 mg/dL indicating successful tail vein glucose infusion. Glucose standards were used regularly to ensure the accuracy of the glucose meter. The ¹³C enrichments of blood glucose were measured by quantifying [1-¹³C]glucose in perchloric acid extract of the blood by NMR. The ¹³C enrichments did not differ much between the different groups and were ∼65% in all groups.

Nuclear Magnetic Resonance

The stored brain extracts were thawed and 630 μL of extracts were mixed with 70 μL of D₂O, 1.5 μL 1,4-dioxane (chemical shift reference and internal standard) and 4 to 5 crystals of sodium azide (preservative). All samples were analyzed using Varian VNMRS 600 Mhz instrument at 150.86 MHz for ¹³C. ¹³C spectra were acquired with proton decoupling and Nuclear Overhauser Enhancement with the following parameters: pulse angle of 45°, acquisition time of 1 second, and a relaxation delay of 5 seconds, 251 p.p.m. spectral width with 32,768 spectral points. A total of 7,312 scans were acquired at 25°C. Peak identification was carried out using chemical shift values from previous literature^{21, 22} after adjusting the chemical shift reference peak of 1,4-dioxane to 67.4 p.p.m. Relevant peaks in the spectra were identified and integrated using MestReNova software (Mestrelab Research, Escondido, CA, USA). The peak area of the internal standard 1,4-dioxane was used to normalize peak areas. The quantification of each peak was carried out by acquiring natural-abundance ¹³C spectra of Glu, Gln, Asp, NAA, and GABA in a single solution at different concentrations and constructing a standard curve of peak area versus ¹³C concentration for each isotopomer of these metabolites.

High-Performance Liquid Chromatography

Total (¹²C+¹³C) Glu, Gln, and Asp concentrations in the brain extract were measured, after precolumn derivatization with o-phthaladehyde and 2-mercaptoethanol, and separation on a reverse-phase column, by fluorometric detection as described previously.²³ To achieve baseline separation of Asp, Glu, and Gln from adjacent peaks while minimizing the total elution time, the following chromatographic program was used: elution with 25% methanol and 75% aqueous sodium phosphate buffer (50 mmol/L, pH 5.29) for 10 minutes followed by increase in the percentage of methanol to 49% in 15 minutes and to 100% in 8 minutes. The metabolites were quantified by comparison of the peak areas with those of standards. The percentage ¹³C enrichment of Glu C4, for example, was calculated from the concentration of [4-¹³C]Glu (after correction for natural-abundance ¹³C) and the total Glu concentration in each mouse.

Data Analysis

Student's two-tailed t-test was used for statistical analysis of paired data. The level of statistical significance and the values of n are indicated in the respective figures. *P⩽0.05, **P⩽0.01.

Results

¹³C Labeling of Brain Metabolites and the Time Course

Figure 1 shows a representative ¹³C NMR spectrum of the nonTg brain extract after 1 hour of [1-¹³C]glucose infusion. Well-resolved peaks of ¹³C-labeled isotopomers of lactate, Glu, Gln, Asp, GABA, NAA, MI, and glucose (C1 α and β) were observed. To ensure near-steady state ¹³C enrichment of cerebral metabolite pools, the time course of ¹³C enrichment was examined: the concentrations of the major ¹³C-enriched brain metabolites were measured in animals killed at the endpoint brain after 8, 30, 60, and 120 minutes of [1-¹³C]glucose infusion. As shown in Figures 2A and B, the concentrations of [4-¹³C]Glu and [4-¹³C]Gln in awake nonTg mice reached ∼maximum values in 60 minutes and showed no further increase. [3-¹³C] and [2-¹³C]Glu and Gln increased more slowly; and, little change was observed after 60 minutes. Hence, 1-hour infusion was deemed to be appropriate and used in subsequent experiments to examine possible differences in the concentrations of the ¹³C-metabolites between nonTg and 3xTg-AD mice as well as the effect of lipoic acid treatment. Awake animals were used to avoid the effect of anesthesia on cerebral glucose utilization.²⁴ A similar trend of time-dependent labeling was observed for the 3xTg-AD mice (Figures 2C and D) but the levels were much lower as compared with those in the nonTg mice.

Figure 1.

A representative ¹³C nuclear magnetic resonance spectrum of brain extract. A proton-decoupled Nuclear Overhauser Effect-enhanced ¹³C spectrum (150.86 MHz) of typical perchloric acid brain extract after [1-¹³C]glucose infusion showing the different isotopomers of glutamate (Glu), glutamine (Gln), aspartate (Asp), gamma-aminobutyric acid (GABA), N-acetylaspartate (NAA), glucose, and myo-inositol (MI). The chemical shift and internal standard, 1,4-dioxane is at 67.4 p.p.m. and isopropyl alcohol with three equivalent methyl carbons (solvent in the pH indicator) at 24.6 p.p.m. For [4-¹³C]Glu and [3-¹³C]Glu, doublets derived from contiguously ¹³C-labeled isotopomers are observed in addition to the singlets.

Figure 2.

Time course of ¹³C enrichment in awake nontransgenic (nonTg) and triple transgenic mouse model of Alzheimer's disease (3xTg-AD) mouse brain. The concentration (mmol/L) of different isotopomers for (A) nonTg glutamate (B) nonTg glutamine (C) 3xTg-AD glutamate (D) 3xTg-AD glutamine after infusion for 8, 30, 60, and 120 minutes (, [4-¹³C]Glu/Gln (glutamate/glutamine); , [3-¹³C]Glu/Gln; , [2-¹³C]Glu/Gln; , [1-¹³C]Glu/Gln); for the nonTg mice, n⩾3 per time point, for the 3xTg-AD mice, n⩾2 per time point.

Comparison of ¹³C-Metabolite Concentrations Among Different Mouse Groups

The concentrations of the ¹³C-labeled metabolites after 1 hour of ¹³C glucose infusion are shown in Table 1, as the mean±s.e.m for each mouse group. Within each group, the concentration is highest for [4-¹³C]Glu followed by [3-¹³C]Glu and [2-¹³C]Glu. For Gln, the concentration is highest for [4-¹³C]Gln followed by [3-¹³C]Gln and [2-¹³C]Gln. For Asp, the highest concentration is observed for [3-¹³C]Asp followed by [2-¹³C]Asp. The results are in good agreement with the known ¹³C-labeling pattern of brain metabolites from [1-¹³C]glucose infusion in rat brain, indicating that these techniques are readily transferable to mice also.²⁵

Table 1. Concentrations of the different isotopomers of ¹³C Glu, Gln, Asp, NAA, GABA, and MI.

Metabolite	nonTg (A)	nonTg+LA (B)	3xTg-AD (C)	3xTg-AD+LA (D)	A versus B	A versus C	C versus D
					P value
[4-13C]Glu	1.59±0.15	1.66±0.19	0.83±0.08	1.47±0.12	0.776	0.001(**)	0.001(**)
[3-13C]Glu	1.09±0.17	1.12±0.11	0.56±0.05	0.95±0.10	0.897	0.007(**)	0.005(**)
[2-13C]Glu	0.98±0.14	0.92±0.08	0.44±0.05	0.81±0.10	0.72	0.002(**)	0.006(**)
[1-13C]Glu	0.33±0.04	0.34±0.04	0.14±0.04	0.25±0.03	0.8	0.004(**)	0.037(*)
[4-13C]Gln	0.53±0.10	0.60±0.11	0.27±0.04	0.51±0.10	0.662	0.022(*)	0.034(*)
[3-13C]Gln	0.48±0.08	0.42±0.06	0.18±0.03	0.43±0.08	0.583	0.003(**)	0.015(*)
[2-13C]Gln	0.38±0.07	0.40±0.06	0.19±0.02	0.37±0.08	0.778	0.018(*)	0.034(*)
[1-13C]Gln	0.14±0.02	0.11±0.02	0.04±0.03	0.14±0.04	0.302	0.008(**)	0.046(*)
[4-13C]Asp	0.12±0.02	0.19±0.07	0.07±0.02	0.08±0.02	0.369	0.076	0.722
[3-13C]Asp	0.32±0.01	0.31±0.01	0.37±0.06	0.32±0.01	0.616	0.445	0.443
[2-13C]Asp	0.25±0.05	0.28±0.06	0.13±0.01	0.21±0.02	0.748	0.019(*)	0.013(**)
[1-13C]Asp	0.13±0.02	0.28±0.11	0.05±0.01	0.12±0.02	0.198	0.007(**)	0.012(**)
[3-13C]NAA	0.12±0.02	0.24±0.10	0.07±0.01	0.11±0.01	0.296	0.012(**)	0.021(*)
[2-13C]NAA	0.08±0.01	0.16±0.08	0.05±0.01	0.07±0.01	0.344	0.047(*)	0.281
[4-13C]GABA	0.25±0.02	0.27±0.07	0.13±0.01	0.21±0.02	0.781	0.001(**)	0.010(**)
[3-13C]GABA	0.20±0.03	0.25±0.06	0.12±0.01	0.20±0.02	0.558	0.005(**)	0.003(**)
[2-13C]GABA	0.42±0.04	0.51±0.10	0.20±0.01	0.45±0.08	0.609	0.000(**)	0.008(**)
[1-13C]GABA	0.12±0.02	0.21±0.05	0.10±0.01	0.14±0.01	0.219	0.128	0.071
[4,6-13C]MI	0.03±0.00	0.04±0.00	0.03±0.01	0.04±0.00	0.056	0.034	0.162
[2-13C]MI	0.02±0.00	0.02±0.00	0.02±0.01	0.02±0.00	0.423	0.270	0.410
[1,3-13C]MI	0.04±0.00	0.05±0.00	0.04±0.00	0.05±0.01	0.022(*)	0.276	0.051
[5-13C]MI	0.02±0.00	0.03±0.00	0.02±0.00	0.02±0.00	0.051	0.315	0.237

ASP, aspartate; GABA, gamma-aminobutyric acid; Gln, glutamine; Glu, glutamate; MI, myo-inositol; NAA, N-acetylaspartate.

Concentrations of the different isotopomers of ¹³C Glu, Gln, Asp, NAA, GABA, and MI in 13-month-old nonTg and 3xTg-AD mice±lipoic acid, after 1 h of [1-¹³C]glucose infusion are shown in the Table 1. Results in the column 2 to 5 are presented as average mmol/L±s.e.m; results in the columns 6 to 8 are the P values obtained from a two-tailed student t-test after comparing between the groups as indicated. *P<0.05, **P<0.01 (indicated in parenthesis); total n=28 and n⩾6 per group. The results for Glu, Gln, and Asp are corrected for natural-abundance (the results for NAA, GABA, and MI are not corrected for natural-abundance).

As shown in Table 1, ¹³C isotopomers of Glu showed a marked decrease in the 3xTg-AD mice compared with nonTg mice (values in parenthesis represent the % decrease); [4-¹³C]Glu (∼50%), [3-¹³C]Glu (∼49%), [2-¹³C]Glu (∼55%), [1-¹³C]Glu (∼58%). ¹³C isotopomers of Gln also showed a trend of prominent decrease in the 3xTg-AD mice compared with nonTg mice: [4-¹³C]Gln (∼49%), [3-¹³C]Gln (∼63%), [2-¹³C]Gln (∼50%), [1-¹³C]Gln (∼71%). Overall, more than 50% decrease of ¹³C Glu and Gln labeling was observed in the 3xTg-AD mice compared with the nonTg mice. Similar trends were also observed in the labeling for the different isotopomers of Asp (except [3-¹³C]Asp), NAA and GABA (except [1-¹³C]GABA) (Table 1).

Importantly, a pronounced recovery of the cerebral metabolites enriched from 1-¹³C glucose was observed in the brain extract of 3xTg-AD mice fed with lipoic acid; all Glu isotopomer levels were increased by lipoic acid feeding (values in parenthesis represent the % increase of 3xTg-AD mice fed lipoic acid in comparison with the 3xTg-AD mice not fed lipoic acid): [4-¹³C]Glu (∼77%), [3-¹³C]Glu (∼70%), [2-¹³C]Glu (∼84%), [1-¹³C]Glu (∼79%). ¹³C labeling of Gln isotopomers in the 3xTg-AD mice was also stimulated by lipoic acid feeding: [4-¹³C]Gln (∼89%), [3-¹³C]Gln (∼137%), [2-¹³C]Gln (∼94%), [1-¹³C]Gln (∼250%). Moreover, it brought levels of the different Gln isotopomers close to the levels of nonTg mice.

Mice fed lipoic acid showed similar trends in regards to [2-¹³C]Asp, [1-¹³C]Asp, [3-¹³C]NAA and all isotopomers of GABA. [3-¹³C]Lactate levels also decreased by ∼50% in the 3xTg-AD mice and was almost completely restored in the 3xTg-AD mice fed lipoic acid as shown in Supplementary Figure 1S. At variance with the 3xTg-AD mice, lipoic acid administration to the nonTg mice showed a minimal and statistically no significant effect (Table 1). Moreover, the levels of different MI isotopomers (natural-abundance ¹³C) did not show significant differences among the different groups. It is difficult to make conclusive remarks about myo-inositol levels by mainly looking at the natural enrichment levels owing to very small concentrations.

Fractional ¹³C (%) Enrichments of Glutamate, Glutamine, and Aspartate: comparison Among Different Mouse Groups

The total [¹²C+¹³C] concentrations of Glu, Gln, and Asp in the endpoint brain, measured by high-performance liquid chromatography, are shown in Figure 3 as the mean±s.e.m for each group. The concentrations are in good agreement with published values for rodent brain. There was no statistically significant difference in total [¹²C+¹³C] Glu, Gln, and Asp concentrations among different groups of mice (except the effect of lipoic acid in increasing Asp of the 3xTg-AD mice).

Figure 3.

Total concentrations (¹²C +¹³C) in 13-month-old nontransgenic (nonTg) and triple transgenic mouse model of Alzheimer's disease (3xTg-AD) mice±lipoic acid feeding. Total concentrations for (A) glutamate (Glu), (B) glutamine (Gln), and (C) aspartate (Asp) are presented as average mmol/L±s.e.m; P values obtained from a two-tailed student t-test comparing the specific groups are indicated under those groups. *P<0.05, **P<0.01; total n=28 and n⩾6 per group.

From the concentration of each ¹³C-metabolite (Table 1) and the total metabolite concentration (Figure 3), the % ¹³C enrichments at each carbon of Glu, Gln, and Asp were calculated for each mouse group, then the mean±s.e.m taken for each group. The results are shown in Figures 4, 5, 6. As shown in Figure 4, overall, % ¹³C Glu enrichments of the different isotopomers were significantly decreased in the 3xTg-AD mice as compared with the nonTg (values in parenthesis represent % decrease): [4-¹³C]Glu enrichment (∼48%) (Figure 4A), [3-¹³C]Glu enrichment (∼51%) (Figure 4B), [2-¹³C]Glu enrichment (∼56%) (Figure 4C), and [1-¹³C]Glu enrichment (∼56%) (Figure 4D). In comparison, the 3xTg-AD mice fed lipoic acid showed an increase of ∼50% enrichment for the four detectable Glu isotopomers over the untreated 3xTg-AD mice (Figure 4).

Figure 4.

¹³C enrichment percentage of the different isotopomers of glutamate in 13-month-old nontransgenic (nonTg) and triple transgenic mouse model of Alzheimer's disease (3xTg-AD) mice±lipoic acid feeding. Enrichment percentage for (A) [4-¹³C]Glu (glutamate), (B) [3-¹³C]Glu, (C) [2-¹³C]Glu, and (D) 1-¹³C]Glu are shown as average percentage±s.e.m; P values obtained from a two-tailed student t-test comparing the specific groups are indicated under those groups. *P<0.05, **P<0.01; total n=28 and n⩾6 per group.

Figure 5.

¹³C enrichment percentage of the different isotopomers of glutamine in 13-month-old nontransgenic (nonTg) and triple transgenic mouse model of Alzheimer's disease (3xTg-AD) mice±lipoic acid feeding. Enrichment percentage for (A) [4-¹³C]Gln (glutamine), (B) [3-¹³C]Gln, (C) [2-¹³C]Gln, (D) [1-¹³C]Gln are shown as average percentage±s.e.m; P values obtained from a two-tailed student t-test comparing the specific groups are indicated under those groups. *P<0.05, **P<0.01; total n=28 and n⩾6 per group.

Figure 6.

¹³C enrichment percentage of the different isotopomers of aspartate in 13-month-old nontransgenic (nonTg) and triple transgenic mouse model of Alzheimer's disease (3xTg-AD) mice±lipoic acid feeding. Enrichment percentage for (A) [4-¹³C]Asp (aspartate), (B) [3-¹³C]Asp, (C) [2-¹³C]Asp, (D) [1-¹³C]Asp are shown as average percentage±s.e.m; P values obtained from a two-tailed student t-test comparing the specific groups are indicated under those groups. *P<0.05, **P<0.01; total n=28 and n⩾6 per group.

An even more pronounced decrease in the 3xTg-AD mice compared with the nonTg mice was found when assessing % ¹³C Gln enrichments of the different isotopomers (values in parenthesis represent the % decrease); [4-¹³C]Gln enrichment (∼61%) (Figure 5A), [3-¹³C]Gln enrichment (∼65%) (Figure 5B), [2-¹³C]Gln enrichment (∼56%) (Figure 5C), and [1-¹³C]Gln enrichment (∼74%) (Figure 5D). Lipoic acid feeding increased the enrichment by ∼90% for the different Gln isotopomers of 3xTg-AD mice as shown in Figure 5 (with the corresponding P value).

In terms of % ¹³C Asp enrichments, a similar decrease in the 3xTg-AD compared with the nonTg mice was found (values in parenthesis represent the % decrease; except [3-¹³C]Asp wherein it represents an increase); [4-¹³C]Asp enrichment (∼33%) (Figure 6A), [3-¹³C]Asp enrichment (∼34%) (Figure 6B), [2-¹³C]Asp enrichment (∼50%) (Figure 6C), and [1-¹³C]Asp enrichment (∼56%) (Figure 6D). Lipoic acid feeding did not lead to an overall increase of the different Asp isotopomers as shown in Figure 6 (with the corresponding P value). However, it leads to an increase of % [2-¹³C]Asp by ∼40% (P=0.16) and % [1-¹³C]Asp by ∼100% (P⩽0.05).

Lipoic acid feeding to the nonTg mice did not elicit a statistically significant effect in either isotopomers of Glu, Gln, or Asp (except % [1-¹³C]Asp enrichment) (6.4% in nonTg versus 13.2% in nonTg+lipoic acid mouse group, P⩽0.05) (Figure 6D).

Discussion

Nuclear magnetic resonance allows for quantification of the different glucose metabolites and in essence allows monitoring of glucose metabolism and the downstream TCA cycle-related metabolites. The current study using an ex vivo NMR approach has an advantage over the in vivo approach in terms of facilitating the determination of metabolite levels in nonanesthetized mice and avoids interferences due to anesthesia. Moreover, the ex vivo approach permits quantification of the well-resolved ¹³C metabolite peaks by high-resolution NMR. However, the drawbacks of using our current ex vivo approach are possible postmortem changes in metabolite levels and inability to dynamically monitor the metabolite levels to calculate the metabolic rates as described earlier.²⁶

In this study, the metabolic status after [1-¹³C]glucose infusion was assessed in nonTg and 3xTg-AD mice with and without lipoic acid feeding. Of special interest were Glu and NAA, for Glu is the major excitatory neurotransmitter, whereas NAA is the most abundant amino molecule in the brain. Glu flux is believed to represent up to 80% of glucose metabolism, and to directly define Glu neurotransmitter rate in the intact brain.²⁷ The role of Glu in memory and cognition has been well documented earlier²⁸ and NAA, while unlikely to be a neurotransmitter per se, has been shown to be a neuronal and axonal marker.²⁹

¹³C Enrichment in Awake versus Anesthetized Rodent Brain

The % ¹³C enrichments of Glu, Gln, and Asp isotopomers after 1 hour of [1-¹³C]glucose infusion indicate the fraction of ¹³C labeling from precursor ¹³C glucose to the named ¹³C-isotopomer. It indicates the fraction of ¹³C labeling with respect to the total metabolite levels and would be helpful in identifying the metabolic differences among 3xTg-AD, nonTg mice, and the effect of lipoic acid feeding.

The lag between enrichment of [4-¹³C]Glu and [4-¹³C]Gln follows from the predominance of Gln synthesis in glia and confirms earlier studies. The delays in enrichment of [3-¹³C]Glu, [3-¹³C]Gln, [2-¹³C]Glu, and [2-¹³C]Gln compared with [4-¹³C]Glu and [4-¹³C]Gln reflects the enrichment of the isotopomers in the first and subsequent ‘turns' of the TCA cycle. Accordingly, results match the expected isotopomer, cell location, and ‘turn' through TCA cycles in the rodent brain, previously described for rats and humans.

The ¹³C enrichments of 20.7% at [4-¹³C]Glu and 16.3% at [4-¹³C]Gln observed in our awake nonTg mice after 1 hour of [1-¹³C]glucose infusion (Figures 4, 5) are in reasonable agreement with the ¹³C enrichments of ∼19% ([4-¹³C]Glu) and ∼11% ([4-¹³C]Gln) reported in anesthetized rat brain in vivo after 1 hour of [1-¹³C]glucose infusion using identical infusion protocol.²⁵ In their study, the maximum ¹³C enrichments attained after 3 hours of infusion were 20% to 24% for [4-¹³C]Glu and 16% to 20% for [4-¹³C]Gln. It is possible that near maximum ¹³C enrichments were attained in our control mice after 1 hour (Figure 2) because glucose uptake and metabolism through the TCA cycle are faster in awake, than in anesthetized rodent brain. These results strongly suggest that 1-hour infusion is a reasonable time point to compare the percentage ¹³C enrichments of [4-¹³C]Glu and [4-¹³C]Gln among different groups of awake mice with and without lipoic acid treatment.

[1-¹³C]Glucose Metabolism

The concentrations of the ¹³C Glu, Gln, Asp, and NAA isotopomers in Table 1 show the extent of the ¹³C label transferred from the intravenously infused [1-¹³C]glucose to the downstream TCA cycle-related metabolites (Glu, Gln, Asp, and NAA). The prominent decrease of ¹³C Glu, Gln, Asp, NAA, and GABA isotopomers levels in the 3xTg-AD mice compared with the nonTg mice shows that significantly less ¹³C glucose label was transferred to the Glu, Gln, Asp, NAA, and GABA isotopomers. Interestingly, there was ∼50% decrease of [3-¹³C]lactate in the 3xTg-AD compared with nonTg mice; this could be explained by reduced brain glucose uptake in the 3xTg-AD mice, as shown previously.¹¹ Importantly, lipoic acid treatment restored the concentrations of these ¹³C-metabolites to levels close to those in the nonTg mice. These results demonstrate a substantial effect of lipoic acid toward increasing metabolite labeling in 3xTg-AD mice. Conversely, lipoic acid treatment on nonTg mice did not result in statistically significant differences in the Glu, Gln, Asp, GABA, and NAA ¹³C isotopomers. A possible reason is that glucose uptake and TCA flux are already maximal in nonTg mice and, thus, lipoic acid has no effect. Also, there was no difference in the levels of MI isotopomers among the different groups. It must be noted that 1 hour of [1-¹³C]glucose infusion would not necessarily result in any ¹³C enrichment of MI; hence its concentrations listed in Table 1 may mainly reflect natural-abundance ¹³C.

¹³C Enrichment of Glu, Gln and Asp

The total concentrations (¹²C+¹³C) of Glu, Gln, and Asp in various mice groups (Figure 3) showed only minor differences that did not reach statistical difference (except for increase of total Asp in the 3xTg-AD mice fed lipoic acid compared with the 3xTg-AD mice not fed lipoic acid). However, the % enrichments of Glu, Gln, and Asp isotopomers, calculated from the total concentrations and the concentrations of ¹³C-metabolites, were significantly decreased by ∼50% in the 3xTg-AD mice compared with the nonTg mice (Figures 4, 5, 6). Because this was accompanied with relatively minor differences in the total levels (¹²C+¹³C) of Glu, Gln, and Asp, it may be speculated that the glucose→TCA cycle-related metabolites flux is impaired in the 3xTg-AD mice. Thus, an alternate metabolic source in the AD brain may be supplying acetyl-CoA for TCA cycle-related metabolites like Glu, Gln, and Asp, thus ensuring their homeostatic levels.

In summary, the total levels of ¹³C label transferred from glucose are decreased in the 3xTg-AD mice and similarly, the % ¹³C enrichments are also decreased. These results strongly suggest that either (1) there was a decrease of glucose uptake by the brain or (2) the utilization of that glucose for conversion to TCA cycle-related metabolites was impaired or (3) there was a decrease in brain glucose uptake followed by impaired glycolytic metabolism. If only the levels of ¹³C labeling are considered, it is tempting to speculate that an impairment of glucose uptake might be the causal reason of decrease in labeling as shown by the ∼30% to 40% decrease in the total brain glucose uptake in the 13-month-old 3xTg-AD mice.¹¹ However, taking the % enrichment and the total (¹²C+¹³C) levels into consideration, it seems unlikely that only low glucose uptake could be contributing to the lower ¹³C labeling and lower % ¹³C enrichment. If only lower glucose uptake were contributing the decrease of ¹³C labeling and % ¹³C enrichments, then it should have also affected the total levels of metabolites (¹²C+¹³C) proportionately. However, data in this study show that the decrease in total metabolite levels is rather minimal in the 3xTg-AD mice (in contrast to the substantially decreased ¹³C labeling and % ¹³C enrichments). Thus, both substrate supply and metabolic rate alterations are likely to be present in the 3xTg-AD mice that lead to a lower amount of glucose being converted to TCA cycle-related metabolites. These results further substantiate the metabolic alterations present in the different Alzheimer's disease rodent models.^{30, 31, 32, 33} Magnetic resonance studies conducted in the APP-PS1 model of Alzheimer's disease show decreased NAA and Glu,³⁰ along with an increase of MI.³⁴ A systematic evaluation on the APP-PS1 model found the ratio of choline to creatine (in the cortical and subcortical areas) as a noninvasive biomarker in these mice but did not find the utility of Glu, Gln, and MI as biomarkers.³⁵ Longitudinal monitoring of the APP-PS1 mice showed lower Glu, NAA, and taurine with more apparent changes of NAA in the female mice.³⁶ Decrease of NAA, Glu, and glutathione was found in the APP_Tg2576 mice.³¹ Similarly, results of decrease in Glu, Gln, NAA, and GABA, along with metabolic perturbations in the other metabolites, was found by a metabolomic study of the CRND8 transgenic mouse model of Alzheimer's disease.³² Using ¹H-[¹³C]-NMR on the APP-PS1 mouse model of Alzheimer's disease, it was found that the levels of [4-¹³C]Glu/Gln and [2-¹³C]GABA were reduced after [1,6-¹³C₂]glucose infusion at an early time point but not at isotopic steady-state levels.³⁷ In summary, metabolic analysis of the different mouse models of Alzheimer's disease as referenced above (along with several not referenced here), have shown metabolic perturbations. However, selecting the most relevant Alzheimer's disease mouse model based on the hypothesis and the type of study should be an important consideration. Mouse models like the APP-PS1 are very aggressive in the display of Aβ pathology and would serve useful to monitor a potential therapeutic in reducing Aβ pathology. Additionally, metabolic alterations are well characterized in the APP-PS1 mouse model by several studies and have helped to further our understanding of this widely used mouse model of Alzheimer's disease. However, the lesser aggressive 3xTg-AD mouse model has several advantages over the APP-PS1 mouse model in terms of having age-dependent appearance of pathology and synaptic impairments in addition to the close mimicking of the pathology type (plaques and tangles). Moreover, a very important advantage of the 3xTg-AD mouse model is the development of one hallmark pathology (plaques) leading to the development of other signature lesion (tau pathology)—a feature missing in all the previous mouse models of Alzheimer's disease.¹⁷ Thus, understanding the metabolic differences in glucose utilization by the brain (in the 3xTg-AD mouse model) would further our understanding of this advanced preclinical mouse model that is regularly used to assess potential therapeutics for Alzheimer's disease. To our knowledge, this is the first study that documents the metabolic alterations in the 3xTg-AD mouse model of Alzheimer's disease and complements well with the earlier studies in other transgenic mouse models but shows a much steeper decline of brain glucose metabolism. Moreover, the ability of lipoic acid to restore glucose metabolism and the subsequent TCA-related metabolites specifically in the 13-month-old 3xTg-AD mice with little effect in the nonTg mice point to the specificity of the restorative effect.

It must be noted that exogenous lipoic acid equilibrates among different intracellular and extracellular compartments but cannot substitute for covalently bound lipoic acid (as the cofactor of mitochondrial complexes such as pyruvate dehydrogenase and α-ketoglutarate dehydrogenase). It is likely that thiol/disulfide exchange reactions facilitated by lipoic acid are involved in activation or stimulation of cysteine-rich member of insulin signaling, such as the insulin receptor itself and insulin receptor substrate.^{7, 38} This, in turn, leads to a positive feedback loop that stimulates greater uptake, utilization, and metabolism of glucose and the subsequent TCA cycle-related metabolites. These effects may be viewed as an insulin-like effect of lipoic acid, providing further support to the role of insulin resistance and Alzheimer's disease.^{39, 40}

Overall, the brain of 13-month-old 3xTg-AD mice show a hypometabolic state that was encompassed by prominently lower glucose→TCA cycle-related metabolites flux (Glu, Gln, Asp, and NAA and accompanying shift of substrate supply); administration of lipoic acid was successful in overcoming this hypometabolic state.

The authors declare no conflict of interest.