Lactacidosis modulates glutathione metabolism and oxidative glutamate toxicity

Authors


Address correspondence and reprint requests to Pamela Maher, Salk Institute for Biological Studies, 10010 North Torrey Pines Road, La Jolla, CA 92037, USA. E-mail: pmaher@salk.edu

Abstract

J. Neurochem. (2010) 113, 502–514.

Abstract

Lactate and acidosis increase infarct size in humans and in animal models of cerebral ischemia but the mechanisms by which they exert their neurotoxic effects are poorly understood. Oxidative glutamate toxicity is a form of nerve cell death, wherein glutamate inhibits cystine uptake via the cystine/glutamate antiporter system inline image leading to glutathione depletion, accumulation of reactive oxygen species and, ultimately, programmed cell death. Using the hippocampal cell line, HT22, we show that lactate and acidosis exacerbate oxidative glutamate toxicity and further decrease glutathione levels. Acidosis but not lactate inhibits system inline image, whereas both acidosis and lactate inhibit the enzymatic steps of glutathione synthesis downstream of cystine uptake. In contrast, when glutathione synthesis is completely inhibited by cystine-free medium, acidosis partially protects against glutathione depletion and cell death. Both effects of acidosis are also present in primary neuronal and astrocyte cultures. Furthermore, we show that some neuroprotective compounds are much less effective in the presence of lactacidosis. Our findings indicate that lactacidosis modulates glutathione metabolism and neuronal cell death. Furthermore, lactacidosis may interfere with the action of some neuroprotective drugs rendering these less likely to be therapeutically effective in cerebral ischemia.

Abbreviations used:
CI

confidence interval

DMEM

Dulbecco’s modified Eagle medium

DTNB

5,5-dithiobis(2-nitrobenzoic acid)

EDTA

ethylene diamine tetraacetic acid

FCS

fetal calf serum

GST

glutathione S-transferase

HBSS

Hank’s buffered salt solution

HCA

homocysteic acid

MTT

3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide

NAC

N-acetyl cysteine

OGD

oxygen–glucose deprivation

Quis

quisqualate

SSA

sulfasalicylic acid

γ-GC

γ-glutamylcysteine

GCL

glutamate cysteine ligase

γ-GC-EE

γ-GC ethyl ester

Ischemic stroke is one of the most common neurological diseases and has a major impact on mortality and morbidity in industrialized countries. Major advances have been made in understanding the pathophysiology of cerebral ischemia on the cellular and molecular level. One major finding was that excessive stimulation of ionotropic glutamate receptors, especially of the NMDA subtype, contributes to neuronal death, a mechanism referred to as excitotoxicity (Choi 1987). However, clinical trials failed to show a benefit of NMDA inhibitors (Ikonomidou and Turski 2002). Thus, novel strategies for identifying neuroprotective compounds for ischemic stroke are needed.

An important finding in ischemic stroke is that lactate, as a product of glycolysis of blood-derived glucose, accumulates whereas oxidative phosphorylation is inhibited by the lack of oxygen resulting in tissue acidosis which is detrimental for cell survival (Kraig et al. 1987; Katsura et al. 1994; Siesjo et al. 1996). Furthermore, the elevation in blood glucose levels increases both lactate and acidosis in ischemic brain areas and infarct size in animals (Li et al. 1995; Anderson et al. 1999) and humans (Parsons et al. 2002). In humans, hyperglycemia is associated with a poorer stroke outcome compared with normoglycemia (Capes et al. 2001; Williams et al. 2002). Whereas the physiological lactate concentration in brain tissue is 1–2 mmol/kg, during focal ischemia brain lactate levels increase to 13–14 mmol/kg in normoglycemic and up to 22 mmol/kg in hyperglycemic animals (Wagner et al. 1992; Folbergrova et al. 1995). This is associated with a drop of brain pH from pH 7.2–7.4 to pH 6.5–6.6 in normoglycemic and to pH 5.9–6.2 in hyperglycemic stroke (Nedergaard et al. 1991; Wagner et al. 1992; Widmer et al. 1992; Folbergrova et al. 1995). The precise mechanisms by which lactate and acidosis exert their neurotoxic effects are poorly understood and their understanding may open new avenues for neuroprotective strategies.

Paradoxically, in vitro cultured primary neurons and neuronally differentiated NT2a cells are protected by acidosis against both glutamate excitotoxicity and oxygen glucose deprivation (OGD), an in vitro model of ischemic stroke, by acidosis-mediated inhibition of NMDA receptors (Giffard et al. 1990; Tang et al. 1990; Almaas et al. 2003).

Taken together, it can be concluded that there is strong evidence that lactacidosis worsens stroke outcome, but this has not been replicated in vitro. The factors that are responsible for the inadequate modeling of the in vivo situation are not known but hamper the detection of pathways that are responsible for lactacidosis toxicity in vivo. Thus, the detailed analysis of pathways by which lactacidosis exacerbates cell death in vitro using related models is warranted. Furthermore, as no high through-put model of lactacidosis exacerbated neurotoxicity has been described, an evaluation of a high number of potentially neuroprotective drugs in a lactacidotic environment such as in the ischemic brain has not been feasible.

GSH is the major antioxidant in the brain (Maher 2005). Oxidative glutamate toxicity is a mechanism by which high levels of glutamate inhibit cystine uptake by the cystine/glutamate antiporter, system inline image, thereby leading to GSH depletion and cell death (Tan et al. 2001). Experimentally, this mechanism can be investigated in isolation from excitotoxicity in neuronal cell lines (Miyamoto et al. 1989; Davis and Maher 1994) or immature primary cultured neurons still devoid of NMDA receptors (Murphy et al. 1990). Using the murine hippocampal cell line HT22, it was shown that after GSH depletion a massive increase in the production of reactive oxygen species ensues (Tan et al. 1998a), followed by activation of a cGMP-gated calcium channel leading to calcium influx (Li et al. 1997a), which initiates a form of programmed cell death with features of both apoptosis and necrosis (Tan et al. 1998b). Several protective pathways have been analyzed in detail using this simple, very reproducible model of neuronal cell death (Davis and Maher 1994; Maher and Davis 1996; Ishige et al. 2001a,b; Maher 2001; Sagara et al. 2002; Lewerenz et al. 2003; Sahin et al. 2006). Furthermore, oxidative glutamate toxicity can be partly responsible for nerve cell death in excitotoxic paradigms (Schubert and Piasecki 2001).

Several lines of evidence indicate that GSH can promote neuronal survival in ischemic stroke. First, increased levels of GSH disulfide (GSSG), the oxidized form of GSH, and oxidative stress are detected in ischemic brain tissue (Baek et al. 2000). Second, pharmacological stimulation of GSH synthesis by tert-butylhydroquinone (Shih et al. 2005) or transgenic over-expression of GSH peroxidase (Weisbrot-Lefkowitz et al. 1998) reduces infarct size in animal models of ischemic stroke. In this study, we show that lactacidosis modulates GSH metabolism and sensitivity against oxidative glutamate toxicity in vitro. Our findings might contribute to a further understanding of how lactacidosis affects neuronal survival in vivo.

Materials and methods

Materials

Tissue culture dishes were from NUNC (Rochester, NY, USA); fetal calf serum (FCS) was obtained from Hyclone (Logan, UT, USA). Other materials and chemicals are listed as Appendix S1.

Cell culture and viability assays

HT22 cells were grown and passaged as described (Davis and Maher 1994). To analyze the modulation of GSH metabolism and oxidative glutamate toxicity by pH and lactate, Dulbecco’s modified Eagle medium (DMEM) with the cystine concentration reduced to 50% (100 μM) or in a few experiments cystine-free DMEM buffered with 10 mM 2-(N-morpholino)ethanesulfonic acid was used. The medium was either supplemented with 20 mM lactic acid or not and the pH roughly adjusted to either 7.4 or 6.2 using HCl. After equilibration in 10% CO2 at 37°C, the pH was re-adjusted and the medium was sterile-filtered and supplemented with 10% FCS. 2.5 × 103 HT22 cells were seeded in regular growth medium into 96-well plates. After 24 h, the medium was exchanged with cystine-reduced experimental medium at pH 7.4 with or without lactate or at pH 6.2 with or without lactate with the indicated concentrations of glutamate and neuroprotective compounds. After 24 h, the medium was replaced by regular growth medium. In some experiments, hydrogen peroxide (H2O2) was used instead of glutamate. For pre-conditioning experiments, regular growth medium was exchanged for the four test media 6 h after plating and after 18 h, cells were switched to regular medium for oxidative glutamate toxicity. Cell survival was assessed by phase contrast microscopy and the amount of viable cells per well assayed by the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) method as described previously (Hansen et al. 1989; Lewerenz et al. 2003) in at least three independent experiments. For micrographs, the cells were seeded in 35 mm dishes and treated similarly. A light microscope (Leitz DM-IRB) equipped with a phase contrast condenser, 5× objective lens, and a digital camera (Hamamatsu) was used to capture the images using Openlab software.

Rat cortical neurons were prepared from E18 cortices as previously described (Li et al. 1997b). 1 × 105 freshly prepared cortical neurons per well were seeded onto poly-l-lysine-coated 96-well plates using regular DMEM supplemented with 10% FCS. After 24 h, the medium was replaced with the lactacidosis media as described above and glutamate was added. After 24 or 48 h, the medium was replaced by regular growth medium and the surviving neurons quantified by the MTT assay.

Rat cortical astrocytes were prepared from E18 cortices and cultivated as described previously (Lewerenz et al. 2009). Confluent cultures were exposed to test media for 8 or 24 h.

Quantification of total GSH

HT22 cells plated at 3 × 105 in 60 mm dishes and grown for 24 h or confluent astrocyte cultures in 35 mm-dishes were used. Cultures were exposed to lactacidosis media with or without glutamate or cystine for the indicated time periods. For GSH recovery studies, cells were exposed to 200 μM diethyl maleimide for 2 h. This treatment depletes GSH by ∼80% in HT22 cells (Sagara et al. 1998). GSH levels were allowed to recover in test media with or without 10 mM N-acetyl cysteine (NAC) or 1 mM γ-glutamylcysteine ethylester (γ-GC-EE) for 4 h. Preparation of cell extracts and enzymatic measurement of total GSH was performed as previously described and normalized to cellular protein measured by the bicinchoninic acid-based method (Maher and Hanneken 2005). Measurement of GSH release was perfomed as described previously (Lewerenz et al. 2009) with the exception that 60 mm dishes and 2 mL medium were used. For experiments looking at GSSG, both total GSH and GSSG levels were measured using a kit from Oxis Research (Foster City, CA, USA) following the manufacturer’s instructions.

Measurement of extra- and intracellular pH

HT22 cells grown for 24 h in black-walled 96-well plates (Corning) were treated with lactacidosis media with or without 0.5 mM glutamate for 6 h. Intracellular pH was determined using 2′,7′-bis-(2-carboxyethyl)-5-(and-6)-carboxyfluorescein acetoxymethyl ester as described (Almeida et al. 2004) except that the medium was not changed during the course of the analysis and the dual fluorescence was measured on a Gemini plate reader with double excitation at 440 and 490 nm and emission at 535 nm. Extracellular pH was determined using a pH meter using the same conditions and cells seeded in 60 mm plates.

Uptake of radiolabeled amino acids

Measurement of system inline image activity was performed as described previously using sodium-independent uptake of 25 μM l-[35S]-cystine sensitive for inhibition by 10 mM glutamate or sodium-independent uptake of 10 μM l-[3H]-glutamate sensitive for inhibition by homocysteic acid (HCA) (Lewerenz and Maher 2009) except that 1.65 × 104 HT22 cells grown for 24 h in 24-well plates were used, Hank’s buffered salt solution (HBSS) adjusted to pH 6.8 and 6.2 was buffered with 10 mM 2-(N-morpholino)ethanesulfonic acid instead of HEPES and 500 μM quisqualate (Quis) was used as an alternative inhibitor.

For kinetic analysis of glutamate-sensitive l-[35S]-cystine uptake, the cystine concentration was increased up to 300 μM, in part by adding cold l-cystine for concentrations > 50 μM, with uptake time reduced to 1 min to prevent saturation.

Results

Lactate and acidosis reduce GSH and exacerbate oxidative glutamate toxicity

Treatment of HT22 cells with 20 mM lactate at pH 7.4 or 6.2 in the absence or presence of lactate slightly diminished the number of viable cells in the absence of glutamate and greatly reduced viability in the presence of a slightly toxic concentration of glutamate. The toxicity of glutamate was especially increased by the combination of acidosis and lactate (Fig. 1a–c). As shown in Fig. 1b, the EC50 for glutamate toxicity of 0.85 [95% confidence interval (CI): 0.84–0.87] mM at pH 7.4 in the absence of lactate was reduced to 0.69 (95% CI: 0.68–0.70) mM by 20 mM lactate, to 0.41 (95% CI: 0.36–0.45) mM by pH 6.2 and to 0.32 (95% CI: 0.31–0.33) mM by 20 mM lactate at pH 6.2. As oxidative glutamate toxicity is initiated by the depletion of GSH (Tan et al. 2001), we analyzed how lactate and acidosis modulate GSH levels in HT22 cells. Lactate, acidosis and the combination of both all significantly decreased total GSH levels normalized to cellular protein in both the absence and presence of glutamate (Fig. 2a) with the relative effects consistent with their effects on viability. GSSG levels remained below detectable levels under all conditions (not shown).

Figure 1.

 Acidosis and lactate exacerbate oxidative glutamate toxicity in HT22 cells. (a) HT22 cells in 96-well plates were exposed to the experimental media adjusted to pH 6.2 or 7.4 as described in Materials and methods without (pH 7.4, 6.2) or with 20 mM lactate (pH 7.4 Lac, pH 6.2 Lac) and treated without or with 0.5 mM glutamate for 24 h. The relative amount of viable cells was measured by the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay and MTT reduction was normalized to cells at pH 7.4 without lactate and glutamate. The graph represents the mean ± SEM of six experiments. (b) HT22 cells in 96-well plates were treated similarly as in (a) but with different glutamate concentrations. The relative MTT reduction was normalized to cells exposed to the different media alone. The graph represents the mean ± SEM of six experiments. (c) Representative micrographs of HT22 cells exposed to the indicated medium conditions and 0.5 mM glutamate overnight. Statistical analysis was performed by two-way anova (a) or non-linear regression (b), *p < 0.05, **p < 0.01, ***p < 0.001.

Figure 2.

 Effects of extracellular acidosis and lactate on GSH, H2O2 toxicity and intracellular pH. (a) HT22 cells in 60 mm dishes were exposed to the lactacidosis media with or without 0.5 mM glutamate and total cellular GSH and cellular protein were measured after 6 h. Total GSH normalized to cellular protein of cells treated at pH 7.4 without glutamate was set to 100% (43.4 ± 4.0 nmol/mg protein). The graph represents the mean ± SEM of eight experiments. (b) HT22 cells in 96-well plates were switched to the different media adjusted to pH 6.2 or 7.4 without or with 20 mM lactate 6 h after plating and exposed to 2.5 mM glutamate in regular growth medium 18 h later. The amount of viable cells was measured after 24 h with the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay and MTT reduction in wells not treated with glutamate was normalized to 100% for each pre-treatment condition. The graph represents the mean ± SEM of three experiments. (c) Cells plated as in Fig. 1A were exposed to the indicated concentrations of H2O2 for 24 h. The graph represents the mean ± SEM of six experiments with the MTT reduction in cells exposed to the different media alone normalized to 100%. (d) Cells plated as in (b) were treated with the lactacidosis media with or without 0.5 mM glutamate. After 6 h, intracellular pH was determined using 2′,7′-bis-(2-carboxyethyl)-5-(and-6)-carboxyfluorescein acetoxymethyl ester. The graph represents the mean ± SEM of four experiments. White bars: pH 7.4, black bars: pH 6.2, striped: +20 mM lactate. Statistical analysis was performed by two-way anova (a/d) or one-way anova (b/c) with Bonferroni post-tests, *p < 0.05, **p < 0.01, ***p < 0.001.

To test whether lactate and acidosis act via a similar mechanism on HT22 cells as glutamate, we pre-treated HT22 cells with test media containing 20 mM lactate at pH 7.4 or 6.2 in the absence or presence of lactate for 18 h before inducing oxidative glutamate toxicity in regular medium at pH 7.4. As shown in Fig. 2b, lactacidosis pre-treatment pre-conditioned HT22 cells against glutamate toxicity. Moreover, acidosis, both alone and in combination with lactate, also potentiated toxicity induced by H2O2 (Fig. 2c).

Next we asked how extracellular lactate and acidosis in the absence or presence of glutamate influence intracellular pH values. HT22 cells were treated with the lactacidosis media and 0.5 mM glutamate for 6 h and intracellular pH was determined using 2′,7′-bis-(2-carboxyethyl)-5-(and-6)-carboxyfluorescein acetoxymethyl ester. Extracellular acidosis drastically reduced intracellular pH in the absence and presence of glutamate (Fig. 2d). Although lactate did not change the extracellular pH within 6 h of incubation (Fig. S1), extracellular lactate induced a small but significant decrease in intracellular pH in the presence and absence of acidosis and glutamate (Fig. 2d).

Acidosis but not lactate inhibits cystine uptake by system inline image

In oxidative glutamate toxicity, impairment of cystine uptake leads to GSH depletion. Thus, we looked at whether acidosis and/or lactate inhibit cystine uptake. Glutamate-sensitive cystine uptake was reduced by ∼70% at pH 6.2 compared with pH 7.4, whereas the presence of 20 mM lactate did not influence system inline image activity (Fig. 3a). Cystine is protonated at pH 6.2 and it was suggested that the protonated form of cystine is not transported by system inline image (Bannai and Kitamura 1981). To test this hypothesis, we used radiolabeled glutamate in sodium-free conditions with HCA and Quis as specific inhibitors of system inline image as an alternative approach to measure system inline image activity without the use of cystine. Although system inline image activity was significantly and similarly reduced by acidosis when measured as HCA- and Quis-sensitive glutamate uptake, the ∼17% reduction in activity (Fig. 3b) was much smaller than the reduction in glutamate-sensitive cystine uptake (Fig. 3a). If the protonation of cystine is responsible for the reduction in cystine uptake, then the reduction of the transportable fraction of cystine should decrease the apparent affinity of system inline image for cystine. Although at pH of 6.2 the affinity became too low to reach saturation of uptake at technically feasible concentrations, the Km of system inline image was shifted from ∼26 (95% CI: 24–28) μM to ∼88 (95% CI: 65–120) μM when the pH was lowered to 6.8 whereas the Vmax was not significantly changed (pH 7.4: 1.34, 95% CI 1.31–1.38; pH 6.8: 1.46, 95% CI 1.21–1.72) (Fig. 3c). Taken together, our data are compatible with the hypothesis that cystine transport is impaired via two different mechanisms by acidosis. First, there is a minor inhibition of system inline image as indicated by reduced glutamate transport, and second and quantitatively more relevant, there is inhibition by the protonation of cystine which subsequently reduces the amount of transportable cystine.

Figure 3.

 Acidosis but not lactate inhibits system inline image function. (a) HT22 cells were grown in 24-well plates and uptake of 25 μM 35S-cystine (CSSC) was measured in HBSS either at pH 7.4, 6.2 or 7.4 in the presence of 20 mM lactate (pH 7.4 Lac) for 20 min in duplicate. The graph shows glutamate-sensitive cystine uptake normalized to protein after uptake in the presence of 10 mM glutamate in parallel wells was subtracted. Uptake at pH 7.4 for each experiment was normalized to 100%. The mean specific uptake (mean ± SEM) at pH 7.4 was 1.27 ± 0.14 nmol/mg. The graph represents the mean ± SEM of three experiments. (b) HT22 cells were grown as in (a). Uptake of 10 μM 3H-glutamate was measured in sodium-free HBSS in the absence or presence of 1 mM HCA or 500 μM Quis at pH 7.4 or 6.2. The graph shows HCA- or Quis-sensitive glutamate uptake normalized to protein. Uptake at pH 7.4 was normalized to 100%. The graph represents the mean ± SEM of three experiments. The mean specific uptake (mean ± SEM) at pH 7.4 was 7.74 ± 1.00 and 7.82 ± 0.99 nmol/mg for HCA- and Quis-sensitive glutamate uptake, respectively. (a) HT22 cells were grown in 24-well plates and 35S-cystine uptake at the indicated concentrations was measured in the absence or presence of 10 mM glutamate for 1 min at the indicated pH. The graph shows glutamate-sensitive cystine uptake after subtraction of uptake in parallel wells with glutamate and is the mean ± SEM of three independent experiments. Statistical analysis was performed by one-way (a) or two-way (b) anova with Bonferroni post-tests or non-linear regression (c), **p < 0.01, ***p < 0.001.

How lactate and acidosis modulate GSH synthesis downstream of system inline image

Intracellularly, cystine imported by system inline image is reduced to cysteine, which together with glutamate are the substrates for the first step of GSH synthesis, the production of γ-GC by glutamate cysteine ligase (GCL). Lactate reduced GSH levels in HT22 cells (Fig. 2a) but did not affect system inline image activity (Fig. 3a). Thus, we asked whether lactate and lactate plus acidosis differentially affect GSH metabolism downstream of cystine import. To test this, we depleted HT22 cells of GSH by treating with 200 μM diethyl maleimide for 2 h. The cells were allowed to recover GSH in lactacidosis test media for 4 h either in the absence or presence of 10 mM NAC or 1 mM γ-GC-EE. NAC both converts extracellular cystine to cysteine and is taken up directly into cells releasing cysteine after deacetylation (Raftos et al. 2007). γ-GC-EE is intracellularly converted to γ-GC, the substrate of GSH synthetase (GS), for GSH synthesis (Drake et al. 2002). When GSH was allowed to recover in the presence of NAC or γ-GC-EE, GSH recovery was significantly increased ∼2.1- and 2.4-fold, respectively, at pH 7.4, indicating that cystine uptake via system inline image but not GSH synthesis is rate-limiting in the absence but not in the presence of NAC or γ-GC-EE (Fig. 4a). Under all three conditions, lactate alone did not influence relative GSH recovery, while GSH recovery was significantly decreased by acidosis with a further decrease when lactate and acidosis were combined (Fig. 4b). The effect of acidosis was less in the presence of γ-GC-EE as compared with control and NAC (77 ± 8% vs. 56 ± 9% and 52 ± 10%, mean ± SD, two-way anova, Bonferroni post-test, p < 0.001). When lactate and acidosis were combined, the relative GSH recovery was significantly decreased when NAC was present compared with control (30 ± 7% vs. 47 ± 9%, mean ± SD, two-way anova, Bonferroni post-test, p < 0.001) but was increased in the presence of γ-GC-EE as compared with control (67 ± 7% vs. 47 ± 9%, mean ± SD, two-way anova, Bonferroni post-test, p < 0.001). Together, these data suggest that acidosis and lactate impair GSH synthetic pathways downstream of cystine import, more at the level of GCL than GS, and that lactate in the presence of acidosis particularly affects GCL activity.

Figure 4.

 Acidosis impairs GSH synthesis. HT22 cells in 60 mm dishes were exposed to 200 μM diethyl maleimide (DEM) for 2 h which reduced GSH levels to 20 ± 5% of control values. The medium was then replaced with the experimental media at pH 7.4 or 6.2 with or without 20 mM lactate (Lac) and the cells were allowed to recover for 4 h. In some cases, 10 mM N-acetyl cysteine (NAC) or 1 mM γ-GC-EE were included in the medium during recovery. Intracellular total GSH was measured enzymatically and normalized to cellular protein. The graphs represent the mean ± SEM of 7–16 independent experiments. (a) Total GSH per mg cellular protein was normalized to cells at pH 7.4 without lactate not treated with DEM, NAC or γ-GC-EE (61.1 ± 3.3 nmol/mg protein). (b) Total GSH after recovery in the absence (Ctrl) or presence of NAC or γ-GC-EE. Total GSH recovery per mg cellular protein was normalized within each experimental group to that in cells at pH 7.4 without lactate (Ctrl: 50.2 ± 4.5 nmol GSH/mg protein; NAC: 105.3 ± 7.1 nmol/mg protein; γ-GC-EE: 137.9 ± 7.5 nmol/mg protein). Statistical analysis was performed by one-way (a) or two-way (b) anova with Bonferroni post-tests,*p < 0.05, ***p < 0.001.

How acidosis and lactate modulate GSH consumption

Net loss of intracellular GSH occurs when GSH is converted to GSSG by glutathione peroxidase and then exported (Dringen and Hirrlinger 2003), through the action of GSH S-transferases (GSTs) (Rinaldi et al. 2002) that couple GSH to various organic substrates thereby leading to their export (Lo and Ali-Osman 2007), and via the non-enzymatic coupling of GSH to organic molecules (Bates et al. 2009).

To determine how acidosis and lactate modulate GSH consumption, we exposed HT22 cells to cystine starvation using cystine-free DMEM to completely shut down GSH synthesis. Surprisingly, acidosis protected HT22 cells against cystine depletion (Fig. 5a) while the addition of lactate had no detectable toxic effect after 16 h and only a minor toxic effect after 24 h. Survival after 16 h of cystine starvation was not reduced by the presence of 1 mM glutamate indicating negligible amounts of cystine derived from serum (Fig. S2). Given these results, we looked at GSH levels in HT22 cells exposed to cystine starvation with or without acidosis and/or lactate. Acidosis more than doubled GSH levels after 4 h in cystine-free medium, whereas lactate had no effect on GSH levels (Fig. 5b).

Figure 5.

 Cystine starvation reveals that acidosis partially preserves intracellular total GSH and protects against cell death. (a) HT22 cells in 96-well plates were exposed to cystine (CSSC)-free Dulbecco’s modified Eagle medium adjusted to pH 7.4 and 6.2 without lactate (pH 7.4, 6.2) or with 20 mM lactate (pH 7.4 Lac, pH 6.2 Lac). After 16 h (left panel) or 24 h (right panel) the amount of viable cells was measured by the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay and MTT reduction normalized to cells treated with the same pH and lactate concentration in cystine-containing medium in parallel. The graphs represent the mean ± SEM of four (16 h) and six (24 h) independent experiments. (b) HT22 in 60 mm dishes and exposed to cystine starvation at pH 7.4 and 6.2 with or without 20 mM lactate (Lac) for 4 h. Intracellular total GSH was measured enzymatically and normalized to cellular protein and then normalized to the level in cultures treated with the same pH and lactate concentration in cystine-containing medium in parallel (53.9 ± 6.5 nmol/mg protein). The graph represents the mean ± SEM of five independent experiments. (c) HT22 cells grown as in (b) were exposed to cystine starvation at pH 7.4 and 6.2 for 4 h in the presence 200 μM acivicin to prevent degradation of GSH upon release. Released total GSH was normalized to cellular protein and total GSH release at pH 7.4 was normalized 100%. The graph represents the mean ± SEM of five independent experiments. At pH 7.4, GSH release (mean ± SEM) was 21.7 ± 2.3 nmol/mg protein, which represented 43.5 ± 3.9% of initial cellular GSH. Statistical analysis was performed by one-way anova with Bonferroni post-tests (a–c) or one sample t-test (d), *p < 0.05, **p < 0.01.

The increase in GSH could be explained either by decreased consumption or decreased export of GSH. Thus, we measured GSH export after 4 h of cystine starvation in the presence or absence of acidosis. Surprisingly, acidosis increased GSH export by ∼80% compared with pH 7.4 (Fig. 5c). In summary, acidosis but not lactate decreases GSH consumption and the preserved GSH pool protects HT22 cells against death.

How lactate and acidosis modulate GSH and oxidative glutamate toxicity in primary astrocytes and cortical neurons

In HT22 cells, acidosis and lactate reduce GSH synthesis thereby exacerbating oxidative glutamate toxicity. On the other hand, existing GSH pools are partially preserved by acidosis leading to protection against cell death induced by cystine starvation. Together, these results suggest that depending on the relative rates of GSH consumption and synthesis, acidosis can increase or decrease GSH levels in different cell types.

Thus, we extended our studies to primary cortical neurons and astrocytes and asked how acidosis and lactate modulate GSH levels and oxidative glutamate toxicity in these cell types. In astrocytes, after 8 h of exposure to acidosis, we observed a trend to lower GSH levels (−19%, 95% CI: −46 to +9) which became significant (−27%, 95% CI: −56 to −1) when acidosis and lactate were combined. Interestingly, none of the different media had an effect on GSH levels when GSH synthesis was impaired by the presence of 2.5 mM glutamate (Fig. 6a, left panel). After 24 h, both acidosis and lactate prominently and significantly decreased GSH and lactate exacerbated acidosis-induced GSH depletion. Furthermore, at this time point, the combination of acidosis and lactate decreased GSH in the presence of 2.5 mM glutamate (Fig. 6a, right panel). Thus, we asked whether the delayed effect of lactacidosis on GSH levels when system inline image activity was compromised might be explained by a combination of the opposing effects of lactacidosis on GSH synthesis and GSH preservation as observed in HT22 cells. To test this idea, astrocytes were exposed to cystine-free medium for 8 h. As predicted, when GSH synthesis was completely inhibited by the absence of cystine, acidosis led to an increase in GSH levels (Fig. 6b). None of these treatments resulted in a significant loss of viability in the astrocytes (data not shown).

Figure 6.

 Acidosis and lactate modulate total GSH in primary cortical astrocytes and sensitivity to oxidative glutamate toxicity in immature primary cortical neurons. (a, b) Cortical astrocytes in 35 mm dishes were exposed to medium at pH 7.4 or 6.2 with or without 20 mM lactate (Lac) for 8 h (a, left panel) or 24 h (a, right panel) with or without 2.5 mM glutamate or cystine (CSSC) starvation for 8 h (b). Intracellular total GSH was measured enzymatically and normalized to protein. Graphs represent mean ± SEM of five (8 h) and four experiments (24 h). Mean ± SEM total GSH at pH 7.4 without glutamate and with cystine was 82.9 ± 12.3 nmol/mg protein after 8 h and 85.0 ± 12.3 nmol/mg protein after 24 h. For better comparison, total GSH per mg protein at pH 7.4 without glutamate was normalized to 100% and to the mean relative GSH at pH 7.4 in the presence of 2.5 mM glutamate (8 h: 59.2%, 24 h: 26.3%) or in medium containing cystine-free Dulbecco’s modified Eagle medium (17.7%). (c) Freshly prepared primary cortical neurons (E18) in 96-well plates were exposed to lactacidosis media at pH 7.4 or 6.2 with or without 20 mM lactate (Lac). The amount of viable cells was measured by the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay after 24 h (left panel) or 48 h (right panel). The graphs represent the mean ± SEM of three (24 h) and two (48 h) independent experiments each done in quadruplicate. Statistical analysis was performed by one-way (a) or two-way (b) anova with Bonferroni post-tests or non-linear regression with one exponential decay (c), *p < 0.05, **p < 0.01, ***p < 0.001.

Similar to HT22 cells, glutamate treatment of immature primary neurons leads to GSH depletion and cell death (Murphy et al. 1990; Ratan et al. 1994; Li et al. 1997b). However, oxidative glutamate toxicity at 24 h follows a different pattern with increasing concentrations of glutamate as compared with HT22 cells. Whereas in HT22 cells a sigmoidal dose–response curve is observed with ∼100% toxicity at higher concentrations, in primary neurons, the dose-dependence of glutamate-induced cell death matches a one-phase exponential decay. Furthermore, while acidosis and lactate induce a similar trend in the survival of the primary cortical neurons as compared with the HT22 cells with a decrease in the EC50 of glutamate-induced cell death by acidosis and lactate (pH 7.4: 0.60 mM; pH 7.4L: 0.54 mM; pH 6.2: 0.43 mM; pH 6.2L: 0.37 mM; no significant differences), the plateau of survival at high glutamate concentrations is increased by acidosis and by lactate plus acidosis (pH 7.4: 21.4%, 95% CI 10.9–31.8; pH 7.4L: 35.8%, 95% CI 20.8–50.8; pH 6.2: 45.2%, 95% CI 38.6–51.8; pH 6.2L: 67.5%, 95% CI 59.5–75.6) (Fig. 6c, left panel). We hypothesized that this pattern reflects the two effects of acidosis on GSH levels characterized in the HT22 cells; the inhibition of GSH synthesis and the preservation of GSH pools, with the difference being that the GSH pool is more stable in the primary cortical neurons than in the HT22 cells. Thus, the stabilization of the GSH pool by acidosis overrides the impaired synthesis, leading to an overall protective effect. To test this hypothesis, we extended the incubation time with glutamate to 48 h. If the preservation of the GSH pool by acidosis explains the acidosis-induced protection, then this effect should decay with time when synthesis is inhibited. Indeed, after prolonged exposure to glutamate, the beneficial effects of acidosis and acidosis plus lactate were abolished (pH 7.4: 12.5%, 95% CI 4.9–19.6; pH 7.4L: 15.3%, 95% CI 1.5–29.0, pH 6.2: 25.5%, 95% CI 15.0–36.1; pH 6.2L: 23.1%, 95% CI 14.9–31.1) (Fig. 6c, right panel).

The neuroprotective activity of compounds that protect against oxidative glutamate toxicity is differentially affected by lactacidosis

Lactacidosis both increases ischemic brain damage (Kraig et al. 1987; Katsura et al. 1994; Siesjo et al. 1996) and exacerbates oxidative glutamate toxicity in HT22 cells (Fig. 1). Thus, the latter paradigm might be suitable to screen drugs for their neuroprotective activity in the presence of lactacidosis. We have identified numerous compounds that protect against oxidative glutamate toxicity via different mechanisms and might provide useful starting points for therapeutically relevant drugs (Ishige et al. 2001b; Sagara et al. 2002; Maher 2006; Liu et al. 2008). Thus, we tested three compounds, the flavonoids fisetin and quercetin, both known to reduce ischemic brain damage (Rivera et al. 2004; Maher et al. 2007) and the tyrphostin AG494 (Sagara et al. 2002). Interestingly, the EC50s of the compounds for protection were differentially affected by lactacidosis (Fig. 7a). Whereas the EC50 (pH 7.4: 3.66 μM, 95% CI 3.61–3.71; pH 6.2L: 3.69 μM, 95% CI 3.64–3.74) of fisetin was not affected by lactacidosis, lactacidosis increased the EC50 of quercetin more than twofold (pH 7.4: 2.04 μM, 95% CI 2.14–3.7; pH 6.2L: 4.21 μM, 95% CI 4.13–4.29) and the EC50 of AG 494 ∼ninefold (pH 7.4: 0.29 μM, 95% CI 0.20–0.40; pH 6.2L: 2.28 μM, 95% CI 2.03–3.06). The maximal protection by the three compounds was also differentially modulated by lactacidosis. Whereas the maximal protection by fisetin and quercetin in the presence of lactacidosis was decreased to 82% (95% CI 81–83) and 83% (95% CI: 82–84), respectively, compared with pH 7.4, the maximal protection by AG494 in the presence of lactacidosis showed a tendency to a more prominent decrease (68.7%, 95% CI 53.7–83.7%). Next, we tested whether the differential ability to protect against oxidative glutamate toxicity relates to the ability of the compounds to preserve GSH in the presence of lactacidosis. Whereas at pH 7.4 10 μM fisetin and 10 μM quercetin significantly increased GSH levels in the presence of 2.5 mM glutamate for 8 h from ∼60% to ∼94% and ∼98%, respectively, of control levels, only fisetin showed a significant effect on GSH levels in the presence of lactacidosis (Fig. 7b).

Figure 7.

 The activity of neuroprotective compounds is differentially affected by lactacidosis. HT22 cells in 96-well plates were exposed to 2.5 mM glutamate at pH 7.4 or 6.2 together with 20 mM lactate in the presence of the indicated concentrations of the compounds. The amount of viable cells was measured by the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay. To calculate protection, MTT reduction with 2.5 mM glutamate alone was normalized to 0% for each medium. MTT reduction without glutamate or compound at the respective pH and lactate concentrations was normalized to 100%. The graphs show the mean ± SEM of three independent experiments each done in triplicate. (b) HT22 cells in 60 mm plates were exposed to medium at pH 7.4 or 6.2 with 20 mM lactate in the presence or absence of 2.5 mM glutamate and fisetin (Fis), quercetin (Quer) or AG 494 for 8 h. Intracellular total GSH was measured enzymatically and normalized to protein. Total GSH per mg protein was normalized to control cells not treated with glutamate or compound at the same pH and lactate concentration (74.7 ± 4.2 nmol/mg protein at pH 7.4 and 55.3 ± 5.1 nmol/mg protein at pH 6.2 with 20 mM lactate). Graphs represent mean ± SEM of four experiments. Statistical analysis was performed by non-linear regression (a) or two-way anova with Bonferroni post-tests (b), **p < 0.01, ***p < 0.001.

Discussion

In experimental stroke, lactate accumulation from the metabolism of blood-derived glucose in the presence of hypoxia along with the accompanying tissue acidosis is detrimental for tissue survival (Siesjo et al. 1996). In stroke patients, a clear relationship between hyperglycemia, local lactate production in the ischemic brain tissue and the conversion of the penumbra to infarction has been demonstrated (Parsons et al. 2002). The clinical application of this finding is rigorous blood glucose control during the acute phase of stroke management on Stroke Units.

However, the mechanisms underlying lactacidosis-induced neurotoxicity in ischemia are still poorly understood. This is for the most part because of the paradox that although excitotoxicity has been identified as one of the most important mechanisms in ischemic brain damage in vivo as well as in the corresponding in vitro paradigm, OGD (Kochhar et al. 1988; Park et al. 1988; Perez-Pinzon et al. 1995), acidosis protects primary neuronal cultures against excitotoxicity and OGD in vitro (Giffard et al. 1990; Tombaugh and Sapolsky 1990; Rytter et al. 2003).

In contrast, in oxidative glutamate toxicity using the hippocampal cell line HT22, neuronal cell death is not propagated via activation of ionotropic glutamate receptors (Maher and Davis 1996) but instead involves GSH loss and oxidative damage (Tan et al. 2001). Oxidative damage is also found in ischemic stroke and cerebral ischemia leads to prominent GSH oxidation (Shivakumar et al. 1995). Thus, the paradigm of oxidative glutamate toxicity is ideally suited to investigate how neuronal GSH metabolism is modulated by lactate and acidosis.

We show that both lactate and acidosis corresponding to conditions found in hyperglycemic stroke (Nedergaard et al. 1991; Wagner et al. 1992; Widmer et al. 1992) alone as well as in combination exacerbate oxidative glutamate toxicity in HT22 cells. We used the MTT assay as a surrogate marker for cell number (Hansen et al. 1989) which accurately reflects survival in oxidative glutamate toxicity in HT22 cells (Maher and Davis 1996). However, part of the decreased MTT reduction in the presence of lactacidosis might be due to reduced proliferation not death, although both are markers of cytotoxicity. Furthermore, lactacidosis pre-treatment induced a pre-conditioning effect in HT22 cells against oxidative glutamate toxicity at regular pH, indicating that both insults act via similar mechanisms. Considering that GSH depletion is the crucial determinant for cell death in oxidative glutamate toxicity and that both lactate and acidosis decrease GSH levels in a manner similar to their effects on cell survival, our results indicate that both lactate and acidosis exacerbate cell death by decreasing GSH. Furthermore, as protons are components of the GSH/GSSG redox pair, the reductive power of GSH itself is pH sensitive and decreased pH impairs GSH action, even at the same concentration (Schafer and Buettner 2001). The deleterious effect of lactacidosis is not restricted to oxidative glutamate toxicity as a similar pattern of increased cell death by oxidative stress was observed when HT22 cells were exposed to H2O2.

We found that the mechanisms by which acidosis and lactate impair GSH synthesis are multi-factorial. Cystine import via system inline image, the critical first step in GSH biosynthesis that provides the rate limiting amino acid, cysteine, is prominently inhibited by acidosis but not lactate. We show that the decrease in cystine uptake is caused by a decrease in the apparent affinity of system inline image for cystine at low pH. This finding is in line with the previously published hypothesis that system inline image only transports the anionic form of cystine (Bannai and Kitamura 1981). As the pK values of the two inline image groups of cystine are 7.48 and 9.02, respectively (Meister 1963), the amino group with the lower pK is increasingly protonated with decreasing pH and the concentration of transportable cystine is subsequently decreased, which is consistent with the reduction in the apparent affinity of system inline image for cystine at low pH observed by us. In addition, we show that independently of cystine protonation, system inline image is slightly inhibited by acidosis because system inline image-mediated glutamate uptake is also reduced by acidosis. Although previous reports suggested that system inline image in astrocytes is inhibited by lactate but not acidosis (Koyama et al. 2000), lactate did not modulate cystine uptake in the HT22 cells.

In addition, we found that GSH synthesis downstream of cystine import is also inhibited by both lactate and acidosis in cells. The inhibition of GSH synthesis by acidosis is consistent with the observation that GCL has a pH optimum of ∼8.3–8.8 with the activity completely inhibited at pH 6.0 (Mandeles and Block 1955). In addition, GS has a pH optimum of 8.2–8.6 with the activity reduced to ∼10% at pH 6.0 (Snoke et al. 1953). As extracellular acidosis prominently decreases intracellular pH in HT22 cells and this is enhanced by lactate, our results showing impairment of GSH recovery in cells exposed to lactate, acidosis or both are consistent with the reported properties of both enzymes. Nevertheless, as the effect of lactate on intracellular pH is less than the effect of lactate on oxidative glutamate toxicity, GSH levels and recovery, we cannot exclude that lactate also inhibits GSH synthesis independently of its effect on intracellular pH.

In contrast, acidosis increased GSH levels when GSH synthesis was inhibited by the use of cystine-free DMEM for the culture medium. Our data showing increased intracellular GSH as well as enhanced GSH release in response to acidosis suggest that GSH consumption is inhibited by low pH. Pathways that lead to net consumption of GSH include the reaction of GSH with multiple organic molecules present in cells. GSH reacts non-enzymatically with nitroalkene fatty acids (Bates et al. 2009) or nitric oxide (Chiueh and Rauhala 1999). In addition, GSH adducts can be derived from 4-hydroxy-2-nonenal eicosanoids, isoprostanes, estrogens and catecholamines by enzymatic catalysis through GSTs (Blair 2006). GST activity in neurons has been reported to be prominently inhibited by acidosis (Ying et al. 1999). Nevertheless, these findings have to be interpreted with caution because the pH dependence of apparent GST activity is dependent on the artificial substrate used for assaying activity (Pabst et al. 1974). High GST activity has been shown to induce GSH depletion (Rinaldi et al. 2002). A reduction of GSH adduct formation during acidosis might underlie the increase in GSH levels that we observed when GSH synthesis is reduced by cystine starvation.

Taken together, our results indicate that GSH synthesis is inhibited on multiple levels by lactate and acidosis. Most importantly, acidosis inhibits both GSH synthesis and consumption. Thus, acidosis can be predicted to exacerbate or ameliorate GSH depletion during cell stress depending on the relative rates of de novo-synthesis, recycling, export and consumption of GSH. Indeed, our data using immature primary neurons strongly support this hypothesis. In immature primary neurons, survival in the presence of increasing concentrations of glutamate follows a one-phase exponential decay in contrast to the sigmoidal dose response curve seen with HT22 cells. The plateau at high glutamate concentrations in immature primary neurons where system inline image can be assumed to be completely inhibited might indicate a relatively stable pool of GSH. Acidosis increases the plateau of survival in line with our finding that acidosis protected HT22 cells against cystine depletion by stabilizing the GSH pool. Nevertheless, we found that this effect wears off with time. Similar findings in astrocytes support the idea that both mechanisms are relevant for GSH metabolism in the brain during acidosis.

Finally, we show that the neuroprotective activity of some compounds against oxidative glutamate toxicity is differentially affected by lactacidosis. Of note, two compounds tested, fisetin and quercetin, increased GSH at pH 7.4 but quercetin failed to increase GSH in the presence of lactacidosis while the ability of fisetin to preserve GSH was much less impaired. Thus, the neuroprotective action of compounds that are still active in the presence of lactacidosis must, at least in part, rely on mechanisms unrelated to GSH. In contrast, compounds that fail to protect in the presence of lactacidosis might exclusively rely on boosting GSH for neuroprotection. Furthermore, other neuroprotective pathways might also be subject to lactacidosis-mediated inhibition as indicated by our results obtained with the tyrphostin AG494, whose protective effects were severely inhibited by lactacidosis but which did not modulate GSH levels in the presence of glutamate at pH 7.4 or in the presence of lactacidosis. Therefore, we propose that compounds whose neuroprotective activity against oxidative glutamate toxicity is not impacted by lactacidosis are promising candidates for further investigation for the treatment of cerebral ischemia. Thus, oxidative glutamate toxicity in the presence of lactacidosis could be a valuable tool for pre-screening compounds for therapeutic purposes in stroke.

In conclusion, we show that lactacidosis has opposing effects on GSH synthesis and GSH consumption. Thus, whether lactacidosis decreases or increases intracellular GSH levels depends on the amount of GSH synthesis, the cell type and the duration of the insult. Furthermore, lactacidosis differentially modulates the activity of neuroprotective compounds. Our findings might have relevance beyond stroke. Diseases that are accompanied by either generalized or localized oxidative stress and lactacidosis include critical illness (Crimi et al. 2006; Antonini et al. 2008), diabetes (Vantyghem et al. 2000) and multiple sclerosis (Friese et al. 2007; Gonsette 2008).

Acknowledgements

The study was supported by the American Heart Association grant 075514Y to Pamela Maher.

Ancillary