Neuroprotective effects of chronic exposure of SH-SY5Y to low lithium concentration involve glycolysis stimulation, extracellular pyruvate accumulation and resistance to oxidative stress

Abstract

Recent studies suggest that lithium protects neurons from death induced by a wide array of neurotoxic insults, stimulates neurogenesis and could be used to prevent age-related neurodegenerative diseases. In this study, SH-SY5Y human neuronal cells were cultured in the absence (Con) or in the presence (Li⁺) of a low lithium concentration (0.5 mm Li₂CO₃, i.e. 1 mm lithium ion) for 25–50 wk. In the course of treatment, growth rate of Con and Li⁺ cells was regularly analysed using Alamar Blue dye. Resistance to oxidative stress was investigated by evaluating: (1) the adverse effects of high concentrations of lithium (4–8 mm) or glutamate (20–90 mm) on cell growth rate; (2) the levels of lipid peroxidation (TBARS) and total glutathione; (3) the expression levels of the anti-apoptotic Bcl-2 protein. In addition, glucose metabolism was investigated by analysing selected metabolites in culture media and cell extracts by ¹H-NMR spectroscopy. As compared to Con, Li⁺ cells multiplied faster and were more resistant to stress, as evidenced by a lower dose-dependent decrease of Alamar Blue reduction and dose-dependent increase of TBARS levels induced by toxic doses of lithium and glutamate. Total glutathione content and Bcl-2 level were increased in Li⁺ cells. Glucose consumption and glycolytic activity were enhanced in Li⁺ cells and an important release of pyruvate was observed. We conclude that chronic exposure to lithium induces adaptive changes in metabolism of SH-SY5Y cells involving a higher cell growth rate and a better resistance to oxidative stress.

Glutathione, glycolysis, lithium, pyruvate, SH-SY5Y

Introduction

Multiple functional, structural, cellular and molecular changes occur in the brain during ageing, including cognitive decline and reduced brain weight and volume, probably caused by the progressive degeneration and loss of neurons (Shankar, 2010; West et al.1994). These disorders involve an impaired glucose uptake/metabolism (Cunnane et al.2011), mitochondrial dysfunctions (Beal, 2005; Müller et al.2010), increased oxidative stress (Dröge & Schipper, 2007; Muller et al.2007) and accumulation of abnormal proteins (Calabrese et al.2006; Trojanowski & Mattson, 2003).

During the past century, life expectancy has significantly increased. Unfortunately, prevalence of neurodegenerative diseases (e.g. Alzheimer's, Parkinson's, and Huntington's diseases) has also progressively increased, with the need to find means to prevent or cure these age-related disorders (Mattson, 2003; Müller et al.2010). Recent in vitro and in vivo studies suggest that lithium, which has been used for >50 yr for treatment of bipolar disorders (Machado-Vieira et al.2009; Manji et al.1999), could be used to prevent or treat age-related neurodegenerative diseases (Wada et al.2005).

Indeed, therapeutic relevant concentrations of lithium (0.5–1 mm) were found to stimulate neurogenesis (Chen et al.2000; Chuang, 2004), to increase neuronal differentiation of progenitor cells (Kim et al.2004) , to exert neuroprotective effects by increasing resistance to oxidative damages (Cui et al.2007; Kim et al.2011; Shao et al.2008) and to glutamate-induced excitotoxicity (Chen et al.2003; Chuang et al.2002), to decrease the production of amyloid-(Aβ) peptides (Phiel et al.2003) and to prevent τ phosphorylation (Engel et al.2006). Moreover, lithium was shown to promote neuronal survival by increasing the action of anti-apoptotic proteins and inhibiting pro-apoptotic signal transduction (Bielecka & Obuchowicz, 2008; Machado-Vieira et al.2009; Manji et al.2000; Marmol, 2008).

SH-SY5Y cells, derived from a human neuroblastoma, are used in our laboratory to investigate neuroprotective effects of lithium (Allagui et al.2009). SH-SY5Y cells were chosen because they have been widely used as an in vitro model for the study of neurodegenerative disorders (Alzheimer's or Parkinson's diseases) and to evaluate the protective effects of anti-apoptotic substances (Pasquariello et al.2009; Peng et al.2008; Xie et al.2010). We previously showed that long-term lithium treatment at therapeutic concentration (0.5 mm Li₂CO₃ for 25 wk) stimulated SH-SY5Y proliferation, modified expression level of HSP27 stress protein and enhanced resistance to oxidative stress (Allagui et al.2009). It was suggested that chronic treatment (over several months) by lithium could favour neurogenesis, decrease the vulnerability of neuronal cells to oxidative stress and induce post-translational changes of molecular chaperones.

The aim of the present study was to improve our knowledge on neuroprotective effects of lithium in SH-SY5Y cells cultured in the presence of 0.5 mm Li₂CO₃ (Li⁺; i.e. 1 mm lithium ion) for 25–50 wk. Alamar Blue (AB) assay was used to compare the growth rate of control (Con) and Li⁺ treated cells. The resistance of Con and Li⁺ cells to oxidative stress was evaluated by: (1) measuring the effect of increasing doses of lithium carbonate (4–8 mm) or glutamate (20–90 mm) on cell growth; (2) analysing the changes of lipid peroxidation level and total glutathione (tGSH) content; (3) examining the expression level of the anti-apoptotic Bcl-2 protein.

In light of some recent works, which showed that lithium activated cerebral glucose metabolism in healthy men (Fan et al.2010; Kohno et al.2007), we have studied lithium effects on glucose metabolism by measuring the pH of culture media and analysing some metabolites in culture media and cell extracts by ¹H-NMR spectroscopy.

Materials and method

Chemicals and reagents

Lithium carbonate (Li₂CO₃), pro analysis grade, was from Prolabo/Rhone-Poulenc (France). AB assay was from Biosource (USA). Rabbit monoclonal anti-human/mouse Bcl-2 antibody (clone: E17) against Bcl-2 from Epitomics (USA) was purchased from Cliniscience (France). Super-Signal^® West Pico chemiluminescent substrate was from Pierce biotechnology (USA). Foetal calf serum (FCS) and other cell culture reagents were from Institute Jacques Boy (France). Proteases inhibitors, i.e. phenylmethanesulfonyl fluoride (PMSF), N-ethylmaleimide (NEM) and aprotinin and anti-rabbit IgG peroxidase antibody produced in goat were purchased from Sigma-Aldrich (France).

Long-term treatment of SH-SY5Y cells by Li₂CO₃

SH-SY5Y human neuroblastoma cells were chosen because they have been widely used to investigate functions related to neurodegenerative and neuroadaptive processes, neurotoxicity and neuroprotection (Xie et al.2010). The SH-SY5Y cell line is a thrice-cloned sub-line of SK-N-SH cells (ECACC #86012802), which were originally established from a bone marrow biopsy. SH-SY5Y cells were derived from immature neoplastic neural crest cells. They exhibit properties of stem cells and possess the capability of proliferating in culture for long periods.

SH-SY5Y cells were cultured in Dulbecco's Modified Eagle's Medium (DMEM) supplemented with 10% (v/v) FCS and 20 µg/ml gentamycin at 37 °C in humidified 5% CO₂ atmosphere. In order to avoid the rapid medium exhaustion (due to the quick proliferation of cells) and to reduce the stress phases (due to the number of re-seeding occurring during extended lithium exposures), DMEM containing a high-glucose level (25 mm) was used. Under such culture conditions, cells could be left in the same medium for 1 wk.

The experiments of chronic treatment by lithium were initiated by seeding cells in T25 flasks containing 8 ml culture medium at a density 20 000/cm² (i.e. 500 000/flask) and allowed to attach overnight. Then, Li₂CO₃ (150 mm stock solution adjusted to pH 7.0) was added in half of the cultures in order to obtain a final concentration of 0.5 mm. For 1 yr, once per week, cells from Con and Li⁺ treated cultures were dissociated by trypsin-EDTA treatment, counted and then re-seeded at low density, as above (20 000/cm²), lithium being added 12 h later in Li⁺ cultures. Con and Li⁺ treated cells were periodically collected for analysis.

AB assay for cell quantification

AB is a non-toxic, water-soluble dye that was designed to measure quantitatively cell proliferation. When added to cell cultures, the oxidized form of AB is converted to the reduced form due to chemical reduction of medium resulting from metabolic activity of cells. The redox reaction is accompanied by a shift in colour that can be easily measured by spectrophotometry. Magnitude of dye reduction correlates linearly with the number of cells (Al-Nasiry et al.2007), but can be affected by glycolytic and oxidative metabolism of glucose (Abe et al.2002). Preliminary experiments were carried out to determine the optimal density of seeding.

In order to compare proliferation rate of Con and Li⁺ cells, they were seeded in 24-well culture plates at a concentration of 20 000 cells/well/ml. For 6 d, the medium of one culture plate was replaced daily by fresh media containing 10% v/v AB and cultures were put back into the CO₂ incubator for 6 h. The absorbance at 570 nm (oxidized form) and 600 nm (reduced form) was recorded with an automated spectrophotometric microplate reader and percentage of AB reduction was calculated as recommended by the manufacturer (Biosource).

To analyse the relative resistance of Con cells and Li⁺ cells to toxic doses of glutamate or lithium, these chemicals were added to the culture medium to obtain final concentrations ranging from 4 to 8 mm for Li₂CO₃ and from 20 to 90 mm for glutamate. AB assay was performed after 6 d of growth. Percentage of AB reduction in Li⁺ culture media was normalized to the percentage of AB reduction measured in Con.

Determination of lipid peroxidation

Lipid peroxidation was estimated by measuring the thiobarbituric acid reactive substances (TBARS) amount after addition of butylated hydroxytoluene (BHT) as antioxidant and alkaline hydrolysis to release protein-bound malondialdehyde (Carbonneau et al.1991; Hong et al.2000). Briefly, the mixture including 500 µl cell extract, BHT (0.02% final concentration) and NaOH (0.5 n final concentration) was incubated at 60 °C for 30 min. Proteins were precipitated by adding 500 µl trichloroacetic acid (15%) and the mixture was placed in ice for 10 min and centrifuged (1000 g for 10 min). Then, 800 µl supernatant was mixed with 100 µl thiobarbituric acid and heated at 90 °C for 30 min. After cooling, absorbance of the mixture was read at 535 nm. TBARS values were calculated by using a molar absorption coefficient of 1.56 × 10⁵m⁻¹ cm⁻¹.

Colorimetric microplate assay for tGSH

The assay utilizes an enzymatic recycling method, based on the reaction of the thiol group of glutathione (GSH) with 5′-thio-2-nitrobenzoic acid after the action of GSH reductase in the presence of NADPH₂ (Coutelle et al.1992). As recommended by this author, proteins from samples were eliminated by precipitation with metaphosphoric acid 10% (v/v) at 4 °C for 10 min. After centrifugation (10 000 g for 10 min), supernatants (containing GSH) were diluted (1:5) in phosphate buffer saline (PBS; 100 mm) with EDTA (5 mm), pH 7.8.

The determination of tGSH in sample extracts was carried out in parallel with GSH standards (1, 2, 5, 10, 15, 20 mm) using 96 well plates. Absorbance was measured at 405 nm using a kinetic program, which could monitor the reaction at 1 min intervals for 4 min. Samples were analysed in triplicate.

Analysis of Bcl-2 protein expression on Western blot

Cell layers were rinsed with ice-cold PBS, collected into a lysis buffer (PBS containing 0.5% NP-40 and protease inhibitors: 10 mm EDTA, 2 mm PMSF, 5 mm NEM, 1 µg/ml aprotinin, pH 7.4) and stored at −20 °C until use. In total, 100 µl lysis buffer was used for about 10⁶ cells. Cell lysates were sonicated for 10 s and protein concentration was determined using the Lowry's method (Lowry et al.1951). Equal amounts of protein (20 µg) from each homogenate were separated by SDS–PAGE according to Laemmli's (1970) method and electroblotted (100 V, 75 min) on to 0.45 µm pore size nitrocellulose membranes.

In order to saturate unspecific sites, membranes were incubated for 1 h at 37 °C in a Tris-buffered saline (TBS: 10 mm Tris, 140 mm NaCl, pH 7.4) containing 0.1% Tween 20 (TBST). Membranes were then incubated for 1 h at room temperature under continuous shaking in the presence of Bcl-2 rabbit monoclonal antibody diluted (1:1000) in TBST and allowed to stand overnight at 4 °C. After three washing steps, (1) TBS for 5 min, (2) TBS containing 0.1% Nonidet-P40 for 15 min and (3) TBS for 5 min (twice), the blots were incubated at room temperature, in darkness, for 2 h under continuous shaking in the presence of a peroxidase-conjugated secondary antibody diluted (1:10 000) in TBS containing 3% dried skimmed milk. After final washing, membranes were incubated with the chemoluminescent substrate for 2 min at room temperature and then exposed to Bio Max light-1 film (Kodak, USA).

Protein levels (arbitrary units) were determined by computer-assisted densitometric analysis (Photo-Capt software; 123 Multimedia, France) of immunoreactive bands visible on exposed film. Values were expressed as percent of controls (normalized to 100).

Analysis of Con and Li⁺ cell metabolism

Cellular metabolism was evaluated by monitoring pH decrease and by metabolomic analysis (¹H-NMR spectroscopy).

Con and Li⁺ cells were seeded at high density: i.e. 200 000 cells/cm², in order to inhibit cell proliferation by contact inhibition and to limit cell density differences between Con and Li⁺ cultures (which could result from their growth rate differences). Forty-eight hours later, culture supernatants were rapidly collected for pH measurement and frozen at −80 °C. The cells were washed with ice-cold PBS and immediately frozen at −80 °C until extraction procedure. Cellular metabolites were extracted at 4 °C as previously described (Tyagi et al.1996). Briefly, cells were thawed in 1 ml methanol: water (2:1, v/v) mixture at 4 °C and disrupted with an ultraturax (30 s in ice). Then, 1 ml chloroform was added and the mixture was vortexed for 2 min. Finally, 1 ml methanol: water (2:1, v/v) and 1 ml chloroform were added and vortexed for 2 min. The mixture was kept on ice for 30 min, centrifuged (4000 g, 10 min, 4 °C) and the upper fraction (containing the hydrosoluble metabolites) was collected and lyophilized. Cells extracts were reconstituted in 650 µl D₂O with 10 µl 10 mm 3-(trimethylsilyl)-1-propanesulfonate sodium salt (TMPS) solution and 200 µl D₂O with 6 µl 60 mm TMPS solution was added to 400 µl culture medium prior to NMR analysis.

¹H-NMR spectra were obtained at 300 K on a Bruker Avance DRX-600 spectrometer (Bruker, France) operating at 600.13 MHz and equipped with a 5 mm TXI cryoprobe. Fully relaxed ¹H spectra resulting from 60° excitation pulses and 5.5 s repetition time (SW of 9.4 kHz, 64 K data points) were obtained with pre-saturation pulse sequence to suppress residual intensity of ¹H water resonance peak by accumulating 256 free induction decays and 32 for cell extract and culture medium, respectively. Line broadening of 1 Hz was applied prior to Fourier transform. Characteristic metabolites (Desmoulin et al.1990; Govindaraju et al.2000) have been assigned by reference to literature data and on the basis of 2D homonuclear correlation spectroscopy experiments performed on the extracts (data not shown). The signal of TMPS (δ = 0 ppm) served as references for chemical shift and concentration; the number of protons giving rise to a signal was considered in the calculations for absolute concentrations of metabolites as previously described (Roncalli et al.2007).

Statistical analysis

Data were expressed as mean±s.d. Statistical significance was established by one-way or two-way analysis of variance (ANOVA) followed by Dunett's multiple comparison to control group (untreated) and Bonferroni's post-test for all pair-wise comparisons, respectively. Unpaired Student's t test was used for statistical comparison of two means and p < 0.05 was considered statistically significant.

Results

Effects of chronic exposure to lithium on SH-SY5Y cell proliferation rate

Figure 1 shows that the percentage of AB reduction in sub-cultures of SH-SY5Y, previously cultured in the absence or presence of 0.5 mm Li₂CO₃ for 36 or 47 wk, increased over culture time and was significantly higher in Li⁺ than in Con, suggesting that Li⁺ cells multiplied faster than Con cells. In agreement with this assertion, microscopic observations (Fig. 2) showed that the number of cells at mitosis and the cell density were higher in Li⁺ cultures than in Con cultures.

Fig. 1

Time-related changes in % Alamar Blue (AB) reduction in sub-cultures of SH-SY5Y previously cultured in the absence (Con) or presence of 0.5 mm Li₂CO₃ (Li⁺) for 36 (a) or 47 wk (b). Data are presented as mean±s.d. (n = 4). (a) Percentages of AB reduction (i.e. growth rate) were significantly higher in Li⁺vs. Con cells [two-way analysis of variance (ANOVA); F_6,28 = 967.0, p < 0.0001 for culture time effect; F_1,28 = 554.2, p < 0.0001 for chronic lithium treatment, Bonferroni's post-test for pair-wise comparison of Con vs. Li⁺, * p < 0.001]. (b) Percentages of AB reduction were significantly higher in Li⁺vs. Con cells (two-way ANOVA; F_6,28 = 767.0, p < 0.0001 for culture time effect; F_1,28 = 886.6, p < 0.0001 for chronic lithium treatment, Bonferroni's post-test for pair-wise comparison of Con vs. Li⁺, * p<0.001).

Fig. 2

Photographs of cultures were taken 2 d after the seeding of control SH-SY5Y cells (Con) or previously cultured for 47 wk in presence of 0.5 mm lithium carbonate (Li⁺). Arrows point to cells at mitosis. Haemalun-eosin staining; magnification: ×200.

In addition, cell counts carried out weekly, when stock cultures were sub-cultured, showed that the number of cells in Li⁺ cultures was always higher than in Con cultures (data not shown). This corroborates the results reported by Allagui et al. (2009) .

Assessment of the relative resistance of Con and Li⁺ cells toward toxic doses of lithium or glutamate

Cells maintained for 36 or 47 wk in the absence or presence of 0.5 mm Li₂CO₃ were then exposed to lithium carbonate (4–8 mm) or glutamate (20–90 mm) for 4 or 6 d (Fig. 3). High lithium concentrations (6–8 mm) induced a dose-dependent decrease of the % AB reduction in both Con and Li⁺ cultures that was higher after 6 d (Fig. 3a, ^#p < 0.001) than after 4 d (Fig. 3c), highlighting a dose- and time-dependent decrease of cell number by culture. Interestingly, the dose-dependent decrease of % AB reduction (growth inhibition) was lower in Li⁺ than in Con cultures, after the fourth day (***p < 0.001) and the sixth day of incubation (* p < 0.05, ** p < 0.01 and *** p < 0.001 for 6, 7 and 8 mm lithium, respectively), which indicates that Li⁺ cells had developed some resistance against adverse effects of high lithium concentrations. A dose-dependent decrease of % AB reduction was also measured when cells were maintained with high glutamate concentrations (Fig. 3c, d) and the decrease was more pronounced after 6 d (Fig. 3c, # p < 0.001) than after 4 d (Fig. 3d). A relative resistance of Li⁺ cells toward glutamate was observed after the 6-d incubation period (Fig. 3c). However, the decrease of % AB reduction (growth inhibition) was found to be significantly lowered only in Li⁺ cultures exposed to 20 or 40 mm glutamate, *** p < 0.001 and ** p < 0.01, respectively. No significant difference of the magnitude of % AB reduction was measured between Li⁺ and Con cultures for experiments with 60 and 90 mm glutamate. Thus, no protective effect was observed for glutamate concentrations, which led to a growth inhibition higher than 70%.

Fig. 3

Dose effects of lithium carbonate (4–8 mm) or glutamate (20–90 mm) on growth rate of cells previously maintained for 36 or 47 wk in the absence (Con) or presence of 0.5 mm Li₂CO₃ (Li⁺). Percentage of Alamar Blue (AB) reduction was respectively performed in Con and Li⁺ sub-cultures after 6 or 4 d of growth in the presence of high concentrations of lithium (a, b) or glutamate (c, d). Data are presented as mean±s.d. (n = 3). Effects of toxic concentrations of lithium and glutamate are attenuated in Li⁺ pre-treated cells [* p < 0.05, ** p < 0.01, *** p < 0.001 by two way analysis of variance (ANOVA)/Bonferroni's post-test for pair-wise comparison of Con vs. Li⁺]. The decrease of the % AB reduction was significantly higher after 6 d of growth in the presence of high lithium (a) or glutamate (c) concentrations (two-way ANOVA/Bonferroni's post-test for pair-wise comparison of 4 d vs. 6 d, ^#p < 0.001) that than after 4 d (b, d). UT, untreated cells.

Assessment of the relative resistance capacities of Con and Li⁺ cells against oxidative stress

As shown in Fig. 4a, GSH content was significantly increased (20–24%, * p < 0.05) in Li⁺ cells cultured for 35 or 48 wk in the presence of 0.5 mm lithium carbonate. In parallel, lipid peroxydation level (TBARS) was lowered in Li⁺ cells (* p < 0.05), as compared to Con cells (Fig. 4b).

Fig. 4

Total glutathione (tGSH) and thiobarbituric acid reactive substances (TBARS) levels in control (C) or 0.5 mm lithium carbonate (Li⁺) cells (35 and 48 wk) after 6 d of growth. Data are presented as mean±s.d. (n = 4). (a) tGSH levels are significantly higher in Li⁺ cells than in Con cells [two-way analysis of variance (ANOVA), Bonferroni's post-test for pair-wise comparison of Con vs. Li⁺, * p < 0.05]. (b) TBARS levels are significantly lower in Li⁺ cells than in Con cells (two-way ANOVA, Bonferroni's post-test for pair-wise comparison of Con vs. Li⁺, * p < 0.05).

In order to check whether high GSH levels could limit the oxidative stress induced by high doses of lithium or glutamate, and explain the higher resistance of Li⁺ cells (Fig. 3), TBARS levels were measured in Con and 50 wk Li⁺ pre-treated cells then exposed for 6 d to toxic doses of lithium carbonate (5–7 mm) or glutamate (10–60 mm).

As shown in Fig. 5, both lithium and glutamate induced a dose-related increase of TBARS in Con cells (^#p < 0.05). A significant increase of TBARS level occurred for 6 mm Li₂CO₃ and 20 mm glutamate, concentrations at which the inhibition of growth was initiated (Fig. 3a, c). In Li⁺ cells, no dose-dependent increase of lipid peroxidation level was measured (Fig. 5a). As a consequence, TBARS levels were significantly lower in Li⁺ cells (*** p < 0.001) than in Con cells for experiments with 6 and 7 mm Li₂CO₃. Similarly, up to the concentration of 20 mm glutamate, the TBARS were significantly lower in Li⁺ cells (* p < 0.05). Conversely, no significant reduction of TBARS level was found in Li⁺ cells for 40 and 60 mm glutamate, as compared to Con cells.

Fig. 5

Dose-related effects of lithium carbonate (a) or glutamate (b) on lipid peroxidation levels, assessed by thiobarbituric acid reactive substances (TBARS) (nmol/mg protein), in control (C) or 0.5 mm lithium carbonate (Li⁺) cells (50 wk). Values are presented as mean±s.d. (n = 4). (a) Treatment with high lithium concentration, lipid peroxidation levels are significantly lower in Li⁺ than in Con cells [two-way analysis of variance (ANOVA): F_1,15 = 95.93, p < 0.0001 for effect of chronic treatment with 0.5 mm Li₂O₃ for 50 wk; F_3,15 = 6.338, p=0.0055 for effect of acute high Li₂O₃ concentration, Bonferroni's post-test for pairwise comparison of Con vs. Li⁺,* p <0.05, *** p <0.001]. (b) Treatment with high concentration of glutamate (two-way ANOVA; F_1,20 = 23.33, p=0.0001 for effect of chronic treatment with 0.5 mm lithium for 50 wk; F_4,20 = 4.19, p=0.0126 for effect of acute glutamate concentration, Bonferroni's post-test for pairwise comparison of Con vs. Li⁺, * p <0.05). Student's t test for pairwise comparison between treated cells and untreated cells (UT), ^#p < 0.05.

Effect of lithium chronic treatment on Bcl-2 expression level

As shown in Fig. 6a, Bcl-2 expression was strongly increased in Li⁺ cells as compared to Con cells. Densitometric analysis of immunoreactive bands (Fig. 6b) demonstrated that this increase was by 171% (p < 0.01). Repeated analyses in the course of chronic Li⁺ treatment gave similar results.

Fig. 6

Bcl-2 protein expression changes. (a) Immunoblot showing the expression of Bcl-2 protein in SH-SY5Y controls cells (C) or (Li⁺) cells cultured in the presence of 0.5 mm lithium for 28 (28w) or 53 (53w) wk. Equal amounts of proteins (20 µg) were loaded by lane. (b) Relative expression changes of Bcl-2 protein in SH-SY5Y Con cells or Li⁺ cells cultured in presence of 0.5 mm lithium for 53 wk. Results are expressed as percent±s.e.m. of Con, which are arbitrarily given the value 100 and correspond to the mean of four analyses. Statistical significance was assessed by Student's t test. p ⩽ 0.05 was considered statistically significant.

Analysis of long-term effects of lithium on cellular metabolism

Evolution of pH in culture medium

During chronic exposure to lithium (>25 wk), culture media of Li⁺ cells were found to acidify more quickly than those of Con cells. No similar observation was done for shorter lithium exposure period (<4 wk; data not shown). Previous works carried out in our laboratory (data given by Allagui) with LiCl gave similar results (data not shown). In order to check whether pH changes could be due to differences in growth rates (as seen in Fig. 1), Con and Li⁺ cells were seeded at high density in order to obtain immediate confluence and contact inhibition. Addition of 0.5 mm lithium carbonate did not result in pH medium change (pH 7.2 ± 0.1). But 48 h later, pH was found significantly much more acidic in Li⁺ culture (*** p < 0.001) than in Con (Fig. 7), suggesting a higher metabolic activity of lithium-adapted cells.

Fig. 7

pH values in culture supernatants. Data are presented as mean±s.d. (n = 3). After 2 d of growth, pH values were significantly lower in medium from 0.5 mm lithium (Li⁺) cells [two-way analysis of variance (ANOVA); F_1,12 = 204.5, p < 0.0001 for effect of the chronic lithium treatment, Bonferroni's post-test for pair-wise comparison of Con vs. Li⁺, *** p < 0.001].

Metabolomic analyses

Concentrations of glucose and various metabolites were measured in fresh medium and in the culture supernatants of Con and Li⁺ cells, 48 h after seeding (Table 1). Lactate was the most abundant metabolite in Con supernatants. Surprisingly, pyruvate concentration was very high in Li⁺ supernatants. Based on metabolite concentrations measured in culture media (fresh medium, Con or Li⁺ supernatants, Table 1), the concentration decrease that corresponds to a consumption and the concentration increase that corresponds to a release were calculated (Fig. 8). Thus, over 2 d, Con and Li⁺ cells consumed respectively 42 and 72% of glucose from the fresh culture medium. By contrast, glutamine and valine (an essential amino acid) uptake was very low and similar in both Con and Li⁺ cells.

Fig. 8

Analysis of glucose, glutamine and valine consumptions and of metabolites release in control (C) and 0.5 mm lithium (Li⁺) culture supernatants. Data are presented as mean±s.d. Statistically significant difference of the means by Student's t test, * p < 0.05, *** p < 0.001.

Table 1

Concentrations of selected metabolites in the culture media

Metabolites (mm)a	Fresh medium	Con supernatant	Li+ supernatant
Glucose	25a	14.38 ± 2.85**	6.90 ± 0.96** #
Glutamine	3.94 ± 0.27	3.04 ± 0.65	3.39 ± 0.37
Valine	0.91 ± 0.17	0.72 ± 0.06*	0.71 ± 0.04*
Lactate	2.21 ± 0.01	31.31 ± 2.75**	34.39 ± 1.47**
Pyruvate	0.68 ± 0.01	4.55 ± 0.30**	34.11 ± 0.92** #
Alanine	0.14 ± 0.02	2.06 ± 0.26**	3.15 ± 0.07** #

a

Glucose 25 mm and 3-(trimethylsilyl)-1-propanesulfonate sodium salt served as quantitative reference for fresh medium and supernatants, respectively.

Data are expressed as mean concentration±s.d. (n = 4) of metabolites from fresh medium and culture supernatant of control (Con) and 0.5 mm Li₂CO₃ (Li⁺) cells (one-way analysis of variance comparing metabolite concentrations for the three media, Dunnett's post test compared to fresh medium, * p<0.05, ** p < 0.01).

Unpaired Student's t test for comparison between metabolite concentrations in Con vs. Li⁺, ^#p < 0.01.

The glycolytic end-products as lactate, pyruvate and alanine were released by Con and Li⁺ cells (Fig. 8). Surprisingly, pyruvate release in Li⁺ culture was >7-fold higher (p < 0.001) than in Con, whereas lactate and alanine release was increased by 12% (p < 0.05) and 56% (p < 0.001) respectively, as compared to Con.

Analysis of intracellular metabolites showed a moderate increase of lactate (40%, p < 0.05) and alanine concentrations (20%, p < 0.05) in Li⁺ cells (Table 2), whereas there was no significant difference of pyruvate and valine concentrations in Li⁺vs. Con cells.

Table 2

Intracellular concentrations of selected metabolites

Metabolites (nmol/mg protein)a	Con extract	Li+ extract
Glucose	76 ± 34	43 ± 10*
Glutamine	186 ± 28	98 ± 3*
Glutamate	221 ± 24	145 ± 7*
Lactate	758 ± 59	1058 ± 59*
Pyruvate	40 ± 7	50 ± 7
Alanine	108 ± 3	130 ± 14*
Succinate	35 ± 4	28 ± 1*
Valine	0.015 ± 0.001	0.014 ± 0.002

a

3-(trimethylsilyl)-1-propanesulfonate sodium salt served as quantitative reference for ¹H-NMR analysis.

Unpaired Student's t test for comparison between metabolite concentrations in cellular extract of controls (Con) vs. 0.5 mm Li₂CO₃ (Li⁺), * p < 0.05 (n = 3).

The enhanced glucose removal from the media and the increased release of glycolytic end-products indicates an activation of the glycolytic pathway in Li⁺ cells. The fact that intracellular concentrations of succinate (−20%, p < 0.05) and glutamate (−35%, p < 0.05) are significantly lower in Li⁺ cells (Table 2) suggests that activity of the Krebs cycle was reduced.

Discussion

The aim of this work was to investigate whether chronic treatment by a low lithium carbonate concentration (0.5 mm) could be a means to prevent some of the age-related neurodegenerative disorders, such as neuronal loss, oxidative stress and/or impairment of glucose metabolism.

Many studies have shown that lithium treatment enhanced neurogenesis in vivo (Chen et al.2000; Kim et al.2004), stimulated proliferation of glial or neuronal cells (Levine et al.2002; Misiuta et al.2006) and delayed cell death induced by a variety of insults (Hashimoto et al.2002; Nonaka & Chuang, 1998). In our laboratory, we previously demonstrated that chronic exposure to low lithium concentrations enhanced SH-SY5Y neuronal cell growth rate (Allagui et al.2009). In the present study, using AB reduction to quantify cell number by culture, we confirmed that SH-SY5Y cells permanently cultured in presence of 0.5 mm carbonate lithium (Li⁺) proliferate faster than Con. As SH-SY5Y cells possess properties of stem cells (Xie et al.2010), we hypothesize that chronic exposure to low lithium concentration could favour the neurogenesis and brain regenerative capacity in regions where adult neurogenesis was reported, in particular the subgranular zone of the hippocampal dentate gyrus (Eriksson et al.1998; Ming & Song, 2011).

In addition, growth inhibition and lipid peroxidation level induced by high Li doses (>6 mm) were significantly lower in Li⁺ cultures than in Con, which suggest that some adaptative processes render Li⁺ cells less sensitive to high (toxic) lithium concentrations.

Concerning glutamate, its toxic effects were attenuated in Li⁺ culture for concentration <40 mm, suggesting that mechanisms underlying toxicity of the highest glutamate concentrations are not linked to oxidative stress.

Several studies have demonstrated that lithium produces a protective effect against oxidative stress. This effect was generally attributed to the increase of GSH levels and glutamate-cysteine ligase expression (rate-limiting enzyme for GSH synthesis) as well as to the increase of GSH S-transferase activity (Cui et al.2007; Shao et al.2008; Wang et al.2004). In vivo, an increase of GSH peroxidase activity was also observed in the hippocampus of rats treated for 40 d with lithium (Vasconcellos et al.2006).

Accordingly, a higher level of GSH was found in Li⁺ cells as compared to Con. In addition, lithium chronic treatment induced, under our experimental conditions, a clear overexpression of the major anti-apoptotic protein Bcl-2. Similarly, chronic administration of lithium in rodents induced an increase of Bc1-2 levels in several brain areas: frontal cortex; dentate gyrus; striatum (Manji et al.2000). It is widely accepted that Bcl-2 protects cells against oxidative stress. In cells overexpressing Bcl-2, an increased amount of GSH was observed, suggesting a role of Bcl-2 in controlling the cellular redox status (Ellerby et al.1996). More particularly, Bcl-2 was found (1) to stabilize mitochondrial function and suppress oxy-radical accumulation, (2) to reduce the formation of reactive oxygen species and (3) to inhibit the leakage of hydroperoxide from mitochondria (Lee et al.2001).

Glucose is the primary source of energy required for brain activity and an impairment of glucose uptake/metabolism is suspected to be implicated in the pathogenesis of a variety of neurodegenerative diseases (Cunnane et al.2011; Liu et al.2008). Therefore, glucose metabolism was investigated by measuring the pH increase in growth medium (an index of glycolytic intensity) and by metabolomic analysis (¹H-NMR spectroscopy) of the culture supernatants and cellular extracts. Our results indicate that glucose uptake and glycolytic rate were stimulated in Li⁺ treated cells, as evidenced by the rapid pH increase, the increase of glucose consumption, the enhanced production of pyruvate and, to a lesser extent, the enhanced production of lactate and alanine. The lithium-induced stimulation of glycolysis does not exclude a stimulation of the pentose-phosphate pathway, as observed by Fan et al. (2010) . Such a metabolic situation would be suitable for supplying ribose to actively growing Li⁺ cells (for nucleotides synthesis).

The most surprising effect of lithium treatment was the very strong increase of pyruvate release in the extracellular medium. No similar results were reported in the literature. Such an unusual metabolic pattern raises some serious thoughts about the underlying mechanism. Everything happens as if pyruvate could not enter the Krebs cycle to be converted into acetyl-CoA and would rather be excreted. Probably, the over-produced pyruvate is transported out of the cells by a proton-linked monocarboxylate transporter (Halestrap & Price, 1999), which could explain the rapid acidification of Li⁺ culture media.

Using [U-¹³C]-glucose in primary culture of cortical neurons, Fan et al. (2010) showed that lithium (1 mm for 3 d) enhanced glycolytic activity and part of the Krebs cycle activity, particularly the anaplerotic pyruvate carboxylation. Pyruvate oxidation by the tricarboxylic acid cycle and citrate synthase flux were shown to be significantly reduced in the presence of 1 mm Li⁺, suggesting a direct inhibitory effect of Li⁺ on the tricarboxylic acid cycle flux (Fonseca et al.2005). Similarly, Krebs cycle activity in Li⁺ cells seems to be reduced since a light decrease of intracellular succinate and glutamate levels was detected, as compared to Con. The lack of Krebs cycle stimulation can explain the release of the over-produced pyruvate in the extra cellular space of Li⁺ cells.

Interestingly, it was reported that extracellular pyruvate protects neurons against a variety of insults, including glutamate (Miao et al.2010) and H₂O₂ toxicity (Nakamichi et al.2005; Wang et al.2007). The pyruvate protective effect could result from its ability to undergo non-enzymatic decarboxylation in the presence of H₂O₂ (Desagher et al.1997). Pyruvate was also found to exert significant cytoprotection when it was administered after H₂O₂ insult and to suppress superoxide production by sub-mitochondrial particles, suggesting that mechanisms other than direct non-enzymatic reaction contributed to pyruvate's antioxidant effects (Nakamichi et al.2005; Wang et al.2007).

Overall, these works show that chronic exposure of neuronal cells to low lithium concentrations leads to adaptive changes in metabolism resulting in: (1) a stimulation of cell proliferation; (2) an increase of glucose uptake and glycolytic activity; (3) a rise of anti-apoptotic (Bcl-2) and anti-oxidant (GSH and above all pyruvate) cell capacities. These properties of lithium could be useful to prevent and/or counteract dysfunctions underlying brain ageing and neurodegenerative diseases, such as a defect in glucose transport and metabolism, excessive reactive oxygen species (ROS) production and neuronal loss.

Statement of Interest

None.