All animal experiments were performed under approved protocol by Animal Ethics Committee of the Centre for Cellular and Molecular Biology (CCMB), Hyderabad, and were conducted in accordance with the guidelines established by Committee for the Purpose of Control and Supervision on Experiments on Animals, Ministry of Environment and Forests, Government of India. Two-month-old C57BL/6J male mice were housed in the CCMB Animal House in a room maintained at ∼22-25°C with relative humidity of 55–65% and 12 h/12 h light/dark cycle, which switches ON and OFF at 6:00 am and 6:00 pm, respectively.

Chronic exposure of AlCl₃ in 2-month-old mice has been used to mimic the AD pathology in rodents (Xiao et al., 2011; Di Paolo et al., 2014). Mice were divided into two major sets of study. In the first set of study, animals were divided into two groups: Ia. Control (n = 12) and Ib. AlCl₃ treated mice (n = 12) to assess the effects of Al exposure on memory and brain energy metabolism (Figure 1). Mice in Group Ib received AlCl₃ (40 mg/kg) intraperitoneally once a day for 60 days, while Group Ia received normal saline (0.9% NaCl) for the same period (Figure 2A Group I). Aluminum chloride was purchased from Acros Organics. AlCl₃ dissolved in normal saline (4 mg/ml) was administered (250 μl) to a 25-g mouse (40 mg/kg) between 1:00 and 2:00 pm once a day for 60 days.

In the second set of experiment, mice were divided into four subgroups: IIa. Control + Carboxy methyl cellulose, CMC (n = 7); IIb. AlCl₃ + CMC (n = 6); IIc. Control + RS (n = 5); and IId. AlCl₃ + RS (n = 5) to evaluate the effects of RS on memory and brain energy metabolism in AlCl₃ exposed mice (Figure 2A Group II). Mice in Groups IIb and IId received AlCl₃ (40 mg/kg, intragastrically) for 60 days, while those in IIa and IIc received normal saline for the same period. Following the 60 days of AlCl₃ treatment, the Groups IIc and IId mice were administered RS in CMC (2 g/kg, intragastric) for 30 days, while Groups IIa and IIb received CMC. RS was procured from Arya Vaidya Sala, Kottakal, India. As per Ayurvedic Pharmacopeia of India, the shelf life of RS is infinite, since the HgS (Cinnabar) is known to be highly stable. For the present set of studies, we prepared a 20% suspension of RS in 1% CMC, and used within 2 days. All treatments of AlCl₃ and RS in mice were carried out between 1:00 and 2:00 pm for the entire period of study.

After 60 days of the AlCl₃ treatment, memory of the mice was assessed using Morris Water Maze (MWM) test (Vorhees and Williams, 2006). The MWM consists of a circular pool with diameter and height, 120 and 47 cm, respectively, which was virtually divided into four equal quadrants. Different clues were provided on the wall of pool for spatial map of the environment. The pool was filled with water to a depth of 20 cm, and the water temperature was maintained between 22 and 25°C. An escape platform with 10 cm in diameter was submerged 0.5 cm under water level in the fourth quadrant. Mice were given 4 days of training, with four trials per day to locate the platform. Each training session was initiated from different quadrants and lasted for maximum period of 90 s. The time taken by mice (latency) to find the hidden platform was monitored with the help of Ethovision 3.1 software. The memory of mice was assessed on 7^th day by monitoring the time required by animals to reach the platform. Retention of memory in mice was assessed on 8^th day by removing the platform, and monitoring the time spent by animals in the platform zone. The learning and memory were also assessed following 30 days of RS intervention.

Overnight fasted mice were anesthetized with urethane (1.5 g/kg, intraperitoneal). The body temperature of the animals was maintained at ∼37°C. The respiration rate of the animals was monitored using BIOPAC device (Santa Barbara, CA, United States). The tail vein was cannulated for infusion of ¹³C labeled substrate. [1,6-¹³C₂]Glucose was purchased from ISOTEC, Miamiburg, United States, and dissolved (0.225 mol/L) in deionized water for infusion. [1,6-¹³C₂]Glucose was administered as a bolus (1013 μmol/kg in 15 s), followed by an exponentially decreasing infusion rate every 0.5 min for the next 8 min, whereupon the rate was maintained to 51 μmol/kg/min until in situ freezing of the brain (Fitzpatrick et al., 1990; Patel et al., 2005; Tiwari et al., 2013). Mice in Group I were administered [1,6-¹³C₂]glucose for 10, 30, 60, and 90 min (n = 3 at each time point) to derive the ¹³C labeling kinetics of amino acids for metabolic analysis. Mice (n = 23) in second set of study were administered [1,6-¹³C₂]glucose for 10 min (one time point) for estimation of metabolic flux using the initial rate approximation. Blood was collected from orbital sinus just before the end of the infusion, and the brain was frozen in situ in liq.N₂ and stored at -80°C till further processing.

Brains were chiseled out and dissected into different regions such as cerebral cortex, hippocampus, and striatum under frozen conditions in a cryostat maintained at -20°C. Metabolites were extracted from frozen brain tissues (Patel et al., 2001). In brief, brain tissues were powdered with 0.1 mol/L HCl in methanol (1:2 w/v) using glass homogenizer maintained in a dry-ice-ethanol bath. [2-¹³C]Glycine (0.2 μmol, 99 atom%, Cambridge Isotopes, Andover, MA, United States) was added as a concentration reference. The powdered tissues were homogenized with ice-cold ethanol and centrifuged at 12,000 g for 30 min at 4°C to remove cell debris. The dried powder was dissolved in phosphate buffered deuterium oxide containing sodium trimethylsilylpropionate (0.25 mmol/L) for NMR analysis.

Blood plasma (100 μl) obtained from orbital sinus was mixed with deuterium oxide (450 μl) containing sodium formate (1 mmol/L), and passed through a centrifugal filter (10-kDa cutoff, VWR, Radnor, PA, United States) to remove macromolecules (Tiwari et al., 2013). Isotopic ¹³C labeling of glucose was determined in the plasma using ¹H NMR spectroscopy at 600 MHz spectrometer. The concentration of glucose was determined from the area of glucose resonance (5.2 ppm) relative to formate. The percent ¹³C enrichment of glucose-C1 was calculated by dividing the area of the ¹³C-coupled satellites by the total ¹H area (¹²C+¹³C) measured at 5.2 ppm.

¹H-[¹³C]-NMR spectra of brain tissue extracts were recorded at 600 MHz NMR spectrometer (Bruker AVANCE II, Karlsruhe, Germany) using a triple resonance probe equipped with three axes gradients (de Graaf et al., 2003). NMR parameters used are as described in detail in Bagga et al. (2013). In brief, a typical ¹H-[¹³C]-NMR experiment involved collection of two spin-echo ¹H NMR spectra under the condition of an OFF/ON ¹³C inversion pulse. Free induction decays (FIDs) were zero filled, apodized to Lorentzian line broadening, Fourier transformed, and phase corrected. The C-13 edited NMR spectrum was obtained by subtracting the spectrum obtained with ¹³C inversion pulse from that acquired under without inversion pulse. The concentrations of metabolites were calculated from the peak area in the spectrum obtained without ¹³C inversion pulse (OFF) relative to [2-¹³C]glycine. The isotopic ¹³C enrichments of amino acids at different carbon positions were determined from the ratio of area in the difference spectrum and the non-edited spectrum.

The aluminum and mercury content in the cerebral cortex of the Group II mice was measured using inductively coupled plasma mass spectrometry (ICP-MS). In brief, the weighed amount of cerebral cortex was digested using plastic pestle in concentrated nitric acid (1 ml). The resulting homogenate was diluted to 10 ml, and analyzed by ICP-MS for aluminum and mercury content.

The Aβ plaques were visualized by immuno-staining in first set of experiment following 60 days of AlCl₃ treatment. Bain samples were washed with phosphate buffered saline (PBS) followed by permeabilization with 0.1% Tween 20. The axial slices (30 μm) of brain were made using cryo-microtome, and stored in PBS with 0.1% sodium azide. For immuno-staining, sections were washed with PBS, blocked with 5% serum in PBS with 0.1% Tween 20, followed by overnight incubation at 4°C with primary Aβ antibody against the first 14 amino acids of the Aβ peptide (Abcam ab2539). The sections were washed thrice with PBS and treated with appropriate secondary antibody at 4°C for 1 h. The slides were washed again, mounted using Vectashield, and examined Confocal microscope (Leica SP5 AOBS) with a 10x dry objective.

A three-compartment metabolic model (Patel et al., 2005; Tiwari et al., 2013) was fitted to the time courses of ¹³C labeling of amino acids from [1,6-¹³C₂]glucose in the first set of study using a CWave software package (Mason et al., 2003) to determine the metabolic fluxes (Figure 1). The metabolic model consists of a series of differential equations describing mass balance and ¹³C isotope flowing from [1,6-¹³C₂]glucose (Supplementary Table 1). The ¹³C turnover curves of [3-¹³C]glutamate, [4-¹³C]glutamate, [2-¹³C]GABA, [3-¹³C]GABA, [3-¹³C]aspartate and [4-¹³C]glutamine were used for the analysis. The averaged concentrations of glutamate, glutamine, GABA and aspartate measured (Supplementary Table 2) in tissue extracts were used for metabolic modeling. Glutamine and GABA were assumed to be localized in astroglia and GABAergic neurons, respectively, whereas glutamate pool was distributed differently in brain regions as determined previously (Supplementary Table 1). The ratio V_cyc(Glu-Gln)/V_tca(Glu) and V_{cyc(GABA-Gln)}/V_tca(GABA) were constrained to values for the cerebral cortex, hippocampus and striatum as presented in Supplementary Table 1 (Tiwari et al., 2013). The values of V_dil(Gln), V_pc, V_tca(ANet), V_dil(A), V_tca(GABANet), V_shunt, V_dil(GABA), V_tca(Glu), V_dil(Glu) and V_xGlu(Glu-KG) were varied during modeling. The uncertainty in estimation of these parameters was determined using a Monte Carlo analysis as described previously by Mason et al. (1992). Briefly, 500 noisy data sets were generated by adding the random Gaussian noise to the noiseless set of data. The model was fitted to the data to generate a list of 500 values for each of the rates. These values were used to calculate average and standard deviation of each fitted parameters.

The rates of neuronal glucose oxidation in the second set of study were calculated based on the labeling of amino acids from 10-min infusion of [1,6-¹³C₂]glucose as described earlier (Patel et al., 2005; Bagga et al., 2013). The apparent cerebral metabolic rate of glucose oxidation (CMR_Glc(ox)) by glutamatergic neurons was obtained as follows:

where Glu_i and Asp_i are percentage ¹³C labeling of glutamate and aspartate, respectively, at carbon ‘i’ from [1,6-¹³C₂]glucose during 10-min infusion, Glc₁ is the labeling of [1-¹³C]glucose in blood plasma, and [Glu] and [Asp] are the concentrations of glutamate and aspartate, respectively.

The cerebral metabolic rate of glucose oxidation by GABAergic neurons was estimated as follows:

GABA_i is the percentage labeling of GABA at carbon ‘i’ from [1,6-¹³C₂]glucose, and [GABA] is the concentration of GABA in the cerebral cortex.

The total glucose oxidation was estimated as:

The Student’s t-test was applied to determine the significance of difference in the memory, concentration and metabolic rates in AlCl₃ treated mice when compared with controls in the first set of study. One-way ANOVA was carried to determine the statistical significance of difference in the concentration, ¹³C labeling, and cerebral metabolic rate of glucose oxidation with RS intervention in the second set of study. The post hoc Tukey’s honest test was carried out to further identify the statistical significance of difference among groups. All results are reported as mean ± standard error of the mean (SEM).

The learning and memory of AlCl₃ treated mice were assessed through the MWM test after 60 days of treatment and compared with parallel saline treated control mice. The escape latency, the time to locate the platform, is significantly longer (p = 0.031) on the 2^nd day of training in AlCl₃ treated mice when compared with controls (Figure 2B). Additionally, AlCl₃ treated mice (75.4 ± 6.1 s, n = 12) took significantly (p < 0.002) longer time during memory test to reach the hidden platform when compared with normal saline treated controls (43.9 ± 7.8 s, n = 12) (Figure 2C). Moreover, these mice spent (22.8 ± 1.6 s, n = 12) significantly (p < 0.001) less time in the platform zone than controls (34.4 ± 1.6 s, n = 12). These data suggest that AlCl₃ exposure leads to impairment in the memory in mice.

The level of aluminum in the cerebral cortex of control mice was found to be 3.4 ± 0.7 μg per gram of brain tissue (n = 5). The treatment of AlCl₃ in mice for 2 months led to significant (p = 0.002) increase in aluminum content in AlCl₃ treated animals (431.4 ± 129.3 μg per g of brain tissue, n = 5) (Figure 3). There was no significant (p = 0.90) difference for the aluminum content between AlCl₃+RS and AlCl₃+CMC group.

The levels of neurometabolites were measured in non-edited ¹H-[¹³C]-NMR spectrum of brain tissue extracts relative to [2-¹³C]glycine (Figure 4A). The levels of cortical neurometabolites remain unchanged (p > 0.2) following 60 days exposure of AlCl₃ (Supplementary Table 2). The levels of glutamate and N-acetyl aspartate (NAA) were significantly (p < 0.01) lower in the striatum in AlCl₃ treated mice (n = 12) than in normal saline treated controls (n = 12). Moreover, a significant (p < 0.001) increase in myo-inositol level was observed in hippocampus of the AlCl₃ treated mice when compared with NS treated controls.

The cerebral metabolism was followed by ¹H-[¹³C]-NMR spectroscopy in brain tissue extracts in conjunction with timed infusion of [1,6-¹³C₂]glucose. The percent ¹³C labeling of plasma [1-¹³C]glucose averaged during entire period of infusion was found to be similar in AlCl₃ treated mice (44.9 ± 5.7%, n = 12) and normal saline treated controls (40.8 ± 4.3%, n = 12), suggesting that AlCl₃ did not perturb homeostasis of plasma glucose.

The ¹³C labeling of cortical amino acids was measured ex vivo in brain tissue extracts using ¹H-[¹³C]-NMR spectroscopy. The ¹³C labeling of [4-¹³C]glutamate, [3-¹³C]alanine, and [3-¹³C]lactate was visible at 10-min infusion time in AlCl₃ treated mice (Figure 4B). The labeling of different carbons of GABA, glutamine, and aspartate, which were very small at 10 min, could be seen from 30 min onward, and reached to steady state level by 60 min. A similar labeling pattern was observed in control mice. The percent ¹³C labeling of cortical metabolites at different carbon positions was calculated, and plotted with time to construct the ¹³C turnover curve of different amino acids from [1,6-¹³C₂]glucose. A similar analysis was performed to generate the ¹³C turnover curves of amino acids in striatum and hippocampus.

The metabolic rates associated with glutamatergic and GABAergic pathways were obtained by modeling of ¹³C labeling of amino acids from glucose. The best fit of the metabolic model (Supplementary Table 1) to the ¹³C turnover of amino acids in the cerebral cortex is presented in Figure 5. Each symbol represents ¹³C labeling measured from individual animals (n = 12), while the line depicts the prediction of labeling based on metabolic model and optimum values of different iterated parameters. The quality of fit of the model to the ¹³C turnover of amino acids in the hippocampus and striatum was similar to that in the cerebral cortex.

The metabolic rate of glucose oxidation associated with different cell types in the cerebral cortex was reduced following AlCl₃ exposure in Group II mice (Figures 8B,C). It is worth noting that the magnitude of reduction in CMR_Glc(ox) was higher (>32%) in the hippocampal and striatal regions following a waiting of one month as compared with that measured immediately in first set of study. The impact of one month waiting period was relatively more on the metabolic activity of GABAergic neurons in the striatum, which reduced by 28% as compared with only 8% reduction immediately after AlCl₃ treatment. The glutamatergic rate that was decreased by 19% immediately after AlCl₃ exposure was found to be reduced by 33% after one month of waiting period. The most dramatic effects of one month waiting period was observed in hippocampal region, wherein metabolic deficit was found in the range of 34–49% (Group II mice), while it was only 14–20% in the first set of experiment measured immediately after AlCl₃ exposure.

A test for the retention of memory without platform indicated that the time spent around the platform area is significantly (p = 0.001) less in AlCl₃ exposed mice. Intervention of RS to AlCl₃ treated mice for one month period was found to decrease the escape latency (28.1 ± 5.1 s, n = 7) significantly (p = 0.001) as compared with the vehicle (CMC) treated AlCl₃ mice (82.6 ± 7.4 s, n = 5). Furthermore, analysis of retention of memory indicated that time spent around the platform zone is also increased significantly (p = 0.04) following RS administration (AlCl₃+CMC 20.0 ± 1.9 s, n = 5; AlCl₃+RS 29.7 ± 1.4 s, n = 6, Figure 7B). It is noteworthy that RS intervention in control mice had no significant effect on the escape latency (p = 0.71) (Figure 7A) or on the time spent around the platform (p = 0.58). Although the average latency to reach the platform appears to be higher in Control+RS mice than AlCl₃+RS group, the difference is not significant (p = 0.069). Moreover, the latency of Control+RS mice (54.0 ± 8.6 s) was not significantly (p = 0.71) different from Control+CMC (42.7 ± 9.0 s).

There was no significant (p = 0.90) difference for the aluminum content between AlCl₃+RS and AlCl₃+CMC group (Figure 3). Moreover, the analysis of mercury indicated no significant increase in mercury level in RS treated mice (Supplementary Figure 1). These data suggest that RS intervention has improved memory in AlCl₃ exposed mice without any adverse effect on memory in controls.

As mentioned above, the ¹³C labeling of cortical [4-¹³C]glutamate (p = 0.049), [3-¹³C]glutamate (p = 0.001), [3-¹³C]aspartate (p = 0.001), and [4-¹³C]glutamine (p = 0.009) was found to be reduced significantly in AlCl₃ treated mice as compared with the normal saline treated controls (Table 1). The RS intervention in AlCl₃ exposed mice increased ¹³C labeling of cortical [4-¹³C]glutamate (AlCl₃+CMC 1.61 ± 0.05, n = 5; AlCl₃+RS 1.91 ± 0.05 μmol/g, n = 7, p = 0.057) and [3-¹³C]aspartate (AlCl₃+CMC 0.12 ± 0.01, n = 5; AlCl₃+RS 0.17 ± 0.01 μmol/g, n = 7, p = 0.034) when compared with CMC treated mice (Table 1 and Supplementary Figure 2). Though, the ¹³C labeling of [4-¹³C]glutamine was increased with RS supplementation in AlCl₃ treated mice (AlCl₃+CMC 0.19 ± 0.01, n = 5; AlCl₃+RS 0.25 ± 0.02, n = 7, μmol/g), it did not reach to statistical significance (p = 0.067). In addition to improvement in the ¹³C labeling of these amino acids, a robust increase in the labeling of [2-¹³C]GABA, [3-¹³C]aspartate, and [4-¹³C]glutamine in the subcortical regions, namely hippocampus and striatum, was observed (Table 1). It is worth noting that RS did not affect the ¹³C labeling of amino acids in control mice (p > 0.20).

The cerebral metabolic rates of glucose oxidation in the second set of study, calculated using Equations 1–3, are presented in Figure 8. The RS intervention improved the rate of glucose oxidation in the cerebral cortex (AlCl₃+CMC 0.270 ± 0.011, n = 5; AlCl₃+RS 0.341 ± 0.011, μmol/g/min, n = 7, p = 0.028) in AlCl₃ exposed mice. Analysis of energy metabolism at the cellular level revealed that recovery of CMR_Glc in glutamatergic neurons (AlCl₃+CMC 0.180 ± 0.006, n = 5; AlCl₃+RS 0.219 ± 0.007, μmol/g/min, n = 7, p = 0.044) was better than that in GABAergic neurons (p = 0.07). It is noteworthy that RS intervention did not affect (p > 0.3) the cortical glucose oxidation in control mice. The improvement in energy metabolism with RS intervention was relatively higher in the subcortical regions (∼40–70%; striatum and hippocampus) as compared with that in the cerebral cortex (∼20–30%) of AlCl₃ treated mice. These data indicate beneficial effect of RS on the energy metabolism of neurons in AlCl₃ treated mice.