Introduction

Alzheimer’s disease (AD) is a chronic neurodegenerative disease characterized by amyloid plaques and neurofibrillary tangles (Blennow et al. 2006; Hardy and Selkoe 2002). AD is also a leading cause of dementia with an estimated prevalence of 42.3 million people worldwide in 2020 (Ferri et al. 2005), which seriously impair quality of life in AD patients. Moreover, this number will be increased to 81.1 million by 2040 (Ferri et al. 2005). Even more unfortunate is the fact that currently there is no effective strategy to treat or delay the development of AD (Hardy and Selkoe 2002; Holtzman et al. 2011). Therefore, further exploring the pathogenesis of AD will advance the development of clinical treatments for AD.

Although many factors involved in AD pathogenesis, the major cause is amyloid β (Aβ) deposition which results in brain neuronal dysfunction and apoptosis (Ferri et al. 2005). In addition, AD has also been implicated in metabolic disorders including glucose homeostasis and energy metabolism (Suzanne 2014). A stable energy metabolism is essential for keeping normal brain function. The brain possesses only 2% of the total body mass, but about 25% of glucose in the brain is metabolized for energy production (Bélanger et al. 2011). Moreover, brain glucose metabolism produces a series of metabolic intermediates including neurotransmitters and thereby maintains normal brain metabolism (Hoyer et al. 1996). Therefore, brain glucose hypometabolism has been implicated in the progression of AD (Mosconi 2005). Results from positron emission tomography and magnetic resonance imaging studies have shown that the disturbance of brain glucose metabolism can be regarded as an indicator of neuronal dysfunction in AD patients (Saint-Aubert et al. 2016; Li et al. 2016; Ballarini et al. 2016; Winkler et al. 2015). However, the detailed characteristics on brain glucose metabolism in AD are still far from being fully understood.

13C nuclear magnetic resonance (NMR) technique is a promising tool to track metabolic fate related to a specific metabolite (Shulman et al. 1990; Sibson et al. 2001; Zwingmann et al. 2003). The change of 13C incorporation from 13C-labelled substrate into other metabolites can be monitored by using 13C-NMR spectroscopy. In our previous study, a 13C NMR metabolomic approach in combination with intravenous [1-13C]-glucose and [2-13C]-acetate infusions was used to explore the changes in brain energy metabolism during type 1 diabetes (T1D) development (Wang et al. 2015). We found that mitochondrial metabolism was enhanced in astrocytes but impaired in neurons (Wang et al. 2015). Furthermore, we used 13C NMR metabolomics with [2-13C]-acetate and [3-13C]-lactate infusions to explore neuronal and astrocytic metabolism in the brain of db/db mice (Zheng et al. 2017). Our results showed that an unbalanced metabolic cooperation between astrocyte and neuron as well as an enhanced gluconeogenesis in type 2 diabetic (T2D) mice with cognitive decline (Zheng et al. 2017). The aim of the present study was to explore glucose metabolism in the brain of APP/PS1 transgenic AD mice at 10 months of age using a 13C NMR metabolomic approach in combination with intravenous [1-13C]-glucose infusion. This study will provide valuable information to further investigate the potential brain metabolic mechanism of amyloid pathology.

Materials and methods

Animals

Four-week-old APP/PS1 double transgenic mice (male, body weight = 23.7 ± 1.6 g, n = 7) and wild-type mice (male, body weight = 25.7 ± 2.3 g, n = 7) were purchased from the Mode Animal Research Center of Nanjing University (Nanjing, China) and housed in the Experimental Animal Center of Wenzhou Medical University (Wenzhou, China). During the whole feeding process, all mice had free access to standard rat chow and tap water in a specific pathogen-free colony with regulated temperature and humidity under a 12:12 h light:dark cycle. All animal experiments were performed in accordance with the Guide for the Care and Use of Laboratory Animals and approved by the Institutional Animal Care and Use Committee of Wenzhou Medical University. Experiments were reported according to the ARRIVE guidelines.

Morris water maze (MWM) test

The MWM test was conducted to evaluate the performance of spatial learning and memory in mice, as described in our previous study (Zheng et al. 2017). Briefly, the MWM test was performed using a circular pool (diameter = 110 cm; height = 30 cm) filled with opaque water at 26 ± 1 °C. The escape platform with a diameter of 7 cm was submerged 1 cm below the surface of the water. During the training period, if mice cannot find the escape platform within 60 s, the experimenter guided them to reach the escape platform. The training period continued for 4 days and 4 trials per day. After this period, the escape platform was removed, and mice were subjected to a 60s probe test at the same start location. The behavior of mice was tracked and recorded using an overhead video camera. The escape latency, the number of crossings over the original platform location as well as the percentages of swimming length and time in the goal area were calculated by a computer system equipped with ‘Viewer 2’ software (Biobserve GmbH, Bonn, Germany).

Intravenous [1-13C]-glucose infusion

APP/PS1 and WT mice at 10 months of age were anesthetized with chloral hydrate after overnight fasting prior to operation. The glow lamp was used to maintain the rectal temperature of mice at 36.8–37.5 °C. [1-13C]-glucose (Sigma-Aldrich, > 99% pure) was prepared in normal saline at a concentration of 0.5 mol/l and then continuously injected through left jugular veins with a micro syringe pump at the rate of 0.1 ml/kg/min for 30 min. For a 13C NMR analysis, a substantial amount of samples was required to acquire sufficient sensitive signals due to the low abundance of 13C. After 15 min of infusion, therefore, the whole brain was taken out immediately and frozen in liquid nitrogen. The level of blood glucose in mice before and after injection [1-13C]-glucose was measured from a tail nick by a handheld glucometer (B/BRAUN omnitest plus).

13C nuclear magnetic resonance analysis

The brain tissue was extracted using the methanol–chloroform method as described previously (Wang et al. 2015). In brief, the brain tissue was weighed into a centrifuge tube and added with ice-cold methanol (4.0 ml/g) as well as distilled water (2.0 ml/g). The mixture was homogenized and mixed by vortex at 4 °C. Then, ice-cold chloroform (2.0 ml/g) and distilled water (2.0 ml/g) were added into the mixture, mixed again and allowed to stand for 15 min on ice. After centrifugation at 1000 g for 15 min, the supernatant was transferred into a new centrifuge tube, lyophilized for about 24 h and stored at -80 °C until use. The lyophilized sample was reconstituted with 0.6 ml of D2O (99.5%) containing 0.05% of sodium trimethlysilyl propionate-d4 (TSP) and transferred to a 5 mm NMR tube for NMR analysis.

13C NMR spectra were recorded at 298 K on a Bruker AVANCE III 600 MHz NMR instrument at 150.92 MHz, using an inverse-gated decoupling sequence to avoid the nuclear overhauser effect. The main parameters were set as follows: flip angle = 30°; relaxation delay = 2 s; spectral width = 33,333 Hz; scan number = 16,384; acquisition time = 1 s; data points = 64 K. NMR metabolic signals were assigned in accordance with our published data (Wang et al. 2015). The specific 13C-enrichment of metabolites after [1-13C]-glucose infusion was calculated assuming that the natural abundance of 13C is 1.1%.

Data analysis

In this study, all mice were randomly assigned to the experimental procedures, such as housing and feeding, the MWM test, animal operation as well as sample collection. NMR metabolomic analysis was conducted by masking animal group labels. The difference in metabolic and behavior data between APP/PS1 and WT mice was analyzed by Student’s t test in SPSS software (version 13.0, SPSS Inc., Chicago, IL, USA). The difference with a P value <0.05 was considered to be statistically significant.

Results

Learning and memory performance in APP/PS1 mice

Results from the MWM test of APP/PS1 and WT mice at 10 months of age are shown in Fig. 1. We found that the escape latency was remarkably longer in APP/PS1 mice at 10 months of age than age-matched WT mice on day 3 and day 4 as shown in Fig. 1a. The swimming path of WT and APP/PS1 mice at 10 months of age are illustrated in Fig. 1b. The percentage of total swimming length and time in the target quadrant as well as the number of crossings over the original platform location during the probe trial were shown in Fig. 1c–e, respectively. We can see that these behavioral parameters were significantly lower in APP/PS1 mice than in WT mice. Taken together, the MWM test demonstrates that learning and memory ability was impaired in APP/PS1 mice at 10 months of age.

Fig. 1
figure 1

Spatial learning and memory in wild-type (WT) and APP/PS1 mice assessed by the Morris water maze: a escape latency during a 4-day training period; b swimming path in the probe test; c percentage of swimming length in the target quadrant; d percentage of swimming time in the target quadrant; e the number of crossings over the original platform location in the probe test; f blood glucose level in WT (black bar) and APP/PS1 (red bar) mice before and after intravenous [1-13C]-glucose infusion. Significant level: *P < 0.05; **P < 0.01; ***P < 0.001

Full size image

Metabolic changes in the brain of APP/PS1 and WT mice after [1-13C]-glucose infusion

Figure 2a, b show typical 13C NMR spectra of brain tissue samples in APP/PS1 and WT mice at 10 months of age after [1-13C]-glucose infusion, respectively. It can be seen from 13C NMR spectra that 13C isotope of [1-13C]-glucose has been incorporated into the specific carbon sites of metabolites, including glutamate (Glu: C2, δ55.7; C3, δ28; C4, δ34.5), glutamine (Gln: C2, δ55; C3, δ27.1; C4, δ31.7), γ-aminobutyric acid (GABA: C2, δ35.3; C3, δ24.5; C4, δ40.2), aspartate (Asp: C2, δ53.2; C3, δ37.5), lactate (Lac: C2, δ69.4; C3, δ21), succinate (Suc: C2/C3, δ35), alanine (Ala: C3, δ17), N-acetyl aspartate (NAA: C2, δ54.1; C3, δ40.5; C6, δ22.8) and taurine (Tau: C1, δ48.3; C2, δ36.2). Furthermore, OPLS-DA based on 13C NMR data was used to discriminate metabolic patterns between APP/PS1 and WT mice at 10 months of age (R2Y = 0.755, Q2 = 0.475). Figure 2c, d illustrates the score and loading plots of OPLS-DA between APP/PS1 and WT mice at 10 months of age, respectively. At 10 months, a clear separation based on 13C NMR data was observed between APP/PS1 and WT mice (Fig. 2c). From its corresponding loading plot, we can see that most signals of 13C-labeled metabolites were higher in WT mice than APP/PS1 mice, as shown in Fig. 2d.

Fig. 2
figure 2

13C NMR metabolomic analysis: typical 13C NMR spectra of brain tissue extract in APP/PS1 (a) and wild-type (WT, b) mice after infused with [1-13C]-glucose; score (c) and loading (d) plots of OPLS-DA analysis based on 13C NMR data. Metabolite: Ala, alanine; Asp, aspartate; Glu, glutamate; Gln, glutamine; GABA, γ-aminobutyric acid; Lac, lactate; NAA, N-acetyl aspartate; Suc, succinate; Tau, taurine

Full size image

Metabolism of [1-13C]-glucose in the brain of APP/PS1 and WT mice

Metabolism of [1-13C]-glucose in the brain of mice is illustrated in Fig. 3. First of all, [1-13C]-glucose was metabolized to [3-13C]-pyruvate through glycolysis, and then transformed into [3-13C]-Lac by anaerobic glycolysis or [3-13C]-Ala via transamination. In addition, [3-13C]-pyruvate can be converted to [2-13C]-acetyl-CoA by pyruvate dehydrogenase (PDH) in both astrocytes and neurons or [3-13C]-oxaloacetate by pyruvate carboxylase (PC) in astrocytes and then enter the tricarboxylic acid (TCA) cycle. The PC and PDH pathways produce different labeling TCA-derived metabolites. For example, when [3-13C]-pyruvate enters the TCA cycle via the PC pathway, [3-13C]-oxaloacetate and [2-13C]-α-ketoglutarate (α-KG) were labeled, further forming [2-13C]-Glu as well as [2-13C]-Gln. However, from the PDH pathway, [2-13C]-acetyl-CoA and [4-13C]-α-KG were labeled. Then, [4-13C]-α-KG can be further transaminated to [4-13C]-Glu, which is subsequently decarboxylated to [2-13C]-GABA in GABAergic neurons or converted to [4-13C]-Gln in astrocytes. On the second turn of the TCA cycle, the C2 and C3 of Glu, the C2 and C3 of Gln as well as the C3 and C4 of GABA were labeled.

Fig. 3
figure 3

Brain glucose metabolism in wild-type (WT, black bar) and APP/PS1 (red bar) mice after [1-13C]-glucose infusion. Data are presented as relative enrichment level ± SE; N.D., no detection; Significant level: *P < 0.05; **P < 0.01

Full size image

It can be seen from Fig. 3 that APP/PS1 mice had significantly lower 13C labeling from [1-13C]-glucose in all carbon positions of Glu (C2, 4.18 ± 0.71 vs. 5.47 ± 0.63, P = 0.005; C3, 3.72 ± 0.53 vs. 5.06 ± 0.58, P = 0.001; C4, 7.53 ± 1.43 vs. 9.76 ± 1.02, P = 0.007) and GABA (C2, 5.08 ± 0.88 vs. 6.66 ± 0.87, P = 0.008; C3, 2.33 ± 0.44 vs. 3.00 ± 0.49, P = 0.025; C4, 2.46 ± 0.61 vs. 3.45 ± 0.52, P = 0.009) than WT mice. Moreover, relative to WT mice, reduced 13C fluxes into all carbon positions of Gln were also detected in APP/PS1 mice, but a significant difference between them was only found in the C3 position (1.87 ± 0.30 vs. 2.27 ± 0.33, P = 0.041). For APP/PS1 mice the 13C-enrichment of Asp was significantly decreased in the C2 (9.31 ± 2.10 vs. 11.82 ± 1.95, P = 0.047) and C3 (7.74 ± 1.36 vs. 10.05 ± 1.35, P = 0.011) positions compared with WT mice. From Fig. 3, we also found that the 13C flux from [1-13C]-glucose into Suc C2/C3 was significantly decreased in APP/PS1 mice relative to WT mice (3.31 ± 1.15 vs. 5.05 ± 0.88, P = 0.01). As compared with WT mice, APP/PS1 mice had a reduction of lactate-alanine shuttle as indicated by decreased 13C fluxes into Ala C3 (4.14 ± 1.22 vs. 6.32 ± 1.24, P = 0.009) as well as Lac C3 (7.73 ± 2.27 vs. 9.97 ± 1.44, P = 0.054). Additionally, it is worth noting that a high level of [1-13C]-glucose can still be observed in the brain of APP/PS1 mice (13.11 ± 8.87), but not detected in WT mice (Fig. 3). Figure 1f also shows that blood glucose level in APP/PS1 mice was significantly higher than that in WT mice after the same amount of [1-13C]-glucose infusion.

Table 1 shows change of metabolite ratio in the brain of WT and APP/PS1 mice after [1-13C]-glucose infusion. We found that the ratios of GluC4/GlnC4 (P = 0.06) and GluC2/GlnC2 (P = 0.01) were lower in the brain of APP/PS1 mice than WT mice. However, as compared with WT mice, GlnC4/GABAC2 (P = 0.02) and GlnC2/GABAC4 (P = 0.01) ratios were significantly increased in the brain of APP/PS1 mice (Table 1). There were no significant differences between WT and APP/PS1 mice in other metabolic ratios, as shown in Table 1.

Table 1 Change of metabolite ratio in the brain of wild-type (WT) and APP/PS1 mice
Full size table

Discussion

The main clinical symptoms of AD are spatial learning and memory impairment (Blennow et al. 2006). As expected, the water maze test showed that APP/PS1 mice had an obvious memory and cognitive impairment at 10 months of age in comparison with age-matched WT mice. Brain glucose metabolism has been closely linked to brain function (Mergenthaler et al. 2013; Lauretti et al. 2017; Videbech 2000; Baxter et al. 1989). In the present study, therefore, 13C NMR-based metabolomics in combination with an in vivo injection of [1-13C]-glucose was applied to elucidate glucose metabolism in the brain of APP/PS1 mice at 10 months of age.

Energy metabolism plays a critical role in maintaining normal brain function (Bélanger et al. 2011). It is well known that glucose is an important substance for supplying energy via tricarboxylic acid (TCA) cycle. In the present study, the 13C flux of [1-13C]-glucose into the TCA cycle was detected by 13C NMR spectroscopy as indicated by 13C-labeled Suc, one of key intermediates in the TCA cycle. We found that the 13C-enrichment of Suc C2/C3 was significantly decreased in the brain of APP/PS1 mice at 10 months of age after [1-13C]-glucose infusion relative to age-matched WT mice, indicating a reduction of energy metabolism. Interestingly, we also found that glucose was accumulated in the brain of APP/PS1 mice but not in WT mice. These findings indicate a decrease in brain glucose utilization as well as brain energy deficiency during the development of amyloid pathology. One possible explanation could be that the accumulation of amyloid plaques in APP/PS1 mice damaged the mitochondrial function (Xie et al. 2013; Cabezas-Opazo et al. 2015; Anandatheerthavarada et al. 2003). The reduction of brain glucose metabolism has also been reported in AD patient (Mosconi 2005). In our previous study, the 13C-enrichment of Suc C2/C3 was also decreased in the brain of T1D rats at 15 weeks of age after [1-13C]-glucose infusion, indicating a reduced brain energy metabolism (Wang et al. 2015). Furthermore, using 13C NMR technique with intravenous [2-13C]-acetate and [3-13C]-lactate infusions, we found that brain energy metabolism was decreased in astrocytes but not in neurons of T2D mice at 12 weeks of age (Zheng et al. 2017). Taken together, our results suggest that the reduction of brain energy metabolism may be a link between amyloid pathology and diabetic encephalopathy.

Neurotransmitter metabolism is of great importance for maintaining neurotransmitter homeostasis, particularly the Gln-Glu-GABA cycle (Deelchand et al. 2009; Haberg et al. 1998). In astrocytes, Glu is absorbed and synthesized to Gln by glutamine synthetase, which exists predominantly in astrocytes (Norenberg and Martinez-Hernandez 1979). Then, Gln can be transported into neurons and synthesized to Glu by phosphate-activated glutaminase. Moreover, Glu can also be decarboxylated to GABA by glutamic acid decarboxylase in GABAergic neurons. Glu and GABA are the major excitatory inhibitory neurotransmitters in central nervous system, respectively. Therefore, the neurotransmitter shuttling between astrocytes and neurons (so-called the Gln-Glu-GABA cycle) maintains brain neurotransmitter homeostasis. Our results show that the 13C-enrichments of the C2, C3, and C4 of Glu and GABA as well as the C3 of Gln were significantly reduced in the brain of APP/PS1 mice after [1-13C]-glucose infusion as compared with age-matched WT mice. These findings indicate that the Gln-Glu-GABA cycle was down-regulated in the brain of APP/PS1 mice. Moreover, relative to WT mice, the ratio of GluC2/GlnC2 was significantly decreased, while the ratios of GlnC4/GABAC2 and GlnC2/GABAC4 were significantly increased in the brain of APP/PS1 mice, indicating a decrease in Gln transport from astrocyte to glutamatergic and GABAergic neurons. Previously, we have reported that the conversion of Gln to Glu/GABA was also significantly decreased in 15-week-old T1D rats (Wang et al. 2015) as well as 12-week-old T2D mice (Zheng et al. 2017) by 13C NMR in combination with an in vivo injection of [2-13C]-acetate. Therefore, we speculate that the down-regulation of the Gln-Glu-GABA cycle and the reduction of Gln transport may be common metabolic characteristics between amyloid pathology and diabetic encephalopathy.

Asp, as another excitatory neurotransmitter, is derived mainly from oxaloacetate by transamination. In this study, the 13C-enrichments of the C2 and C3 of Asp were significantly lower in the brain of APP/PS1 mice than WT mice after [1-13C]-glucose infusion. This result also reveals a decreased neurotransmitter metabolism in the brain of APP/PS1 mice.

Lactate-alanine shuttle has been reported to regulate the nitrogen exchange between astrocytes and neurons in the brain (Zwingmann et al. 2000; Schousboe et al. 2003; Waagepetersen et al. 2000). However, our results show that the lactate-alanine shuttle was inhibited in the brain of APP/PS1 mice, as indicated by a decrease of 13C flux into Ala C3 and Lac C3 from [1-13C]-glucose. Previously, we also found an identical phenomenon in the brain of 15-week-old T1D rats (Wang et al. 2015). These results indicate that an inhibition of the lactate-alanine shuttle may occur in both amyloid pathology and diabetic encephalopathy.

Conclusions

13C NMR metabolomics combined with intravenous [1-13C]-glucose infusion was conducted to investigate brain glucose metabolism in APP/PS1 mice at 10 month of age. We found that brain glucose utilization was significantly lower in APP/PS1 mice than WT mice, thereby resulting in the reduction of energy and neurotransmitter metabolism. Moreover, we also found that relative to WT mice the lactate-alanine shuttle was inhibited in the brain of APP/PS1 mice. Compared with our previous studies, several common metabolic features may exist in both amyloid pathology and diabetic encephalopathy. In our future work, several suggestions should be considered: (1) brain metabolic rate of glucose can be measured by a multi-time-point 13C-glucose infusion; (2) it could be of great interest to explore glucose metabolic differences between different brain regions; (3) sex difference in the metabolic fate of glucose should be investigate in order to draw the general conclusion; (4) it is important to explore information on gene and protein levels to advance understanding of brain glucose metabolism of amyloid pathology.