Address correspondence and reprint requests to Ralf Dringen, Centre for Biomolecular Interactions Bremen, University of Bremen, PO. Box 330440 D-28334 Bremen, Germany. E-mail: email@example.com
Formaldehyde is endogenously produced in the human body and brain levels of this compound are elevated in neurodegenerative conditions. Although the toxic potential of an excess of formaldehyde has been studied, little is known on the molecular mechanisms underlying its neurotoxicity as well as on the ability of neurons to metabolize formaldehyde. To address these topics, we have used cerebellar granule neuron cultures as model system. These cultures express mRNAs of various enzymes that are involved in formaldehyde metabolism and were remarkably resistant toward acute formaldehyde toxicity. Cerebellar granule neurons metabolized formaldehyde with a rate of around 200 nmol/(h × mg) which was accompanied by significant increases in the cellular and extracellular concentrations of formate. In addition, formaldehyde application significantly increased glucose consumption, almost doubled the rate of lactate release from viable neurons and strongly accelerated the export of the antioxidant glutathione. The latter process was completely prevented by inhibition of the known glutathione exporter multidrug resistance protein 1. These data indicate that cerebellar granule neurons are capable of metabolizing formaldehyde and that the neuronal glycolysis and glutathione export are severely affected by the presence of formaldehyde.
Formaldehyde is the simplest aldehyde that shows high reactivity toward cellular macromolecules like DNA and proteins (Lu et al. 2010; Ospina et al. 2011). Formaldehyde is not only ubiquitously present as an environment pollutant (Dhareshwar and Stella 2008; de Groot et al. 2009) but is also generated in substantial amounts in the human body by enzyme-catalyzed reactions (O'Sullivan et al. 2004; Cloos et al. 2008; Hou and Yu 2010). However, owing to formaldehyde detoxification by cellular enzymes (Friedenson 2011; MacAllister et al. 2011), a steady state balance between formaldehyde-generating and -disposing processes is established that leads to a normal blood formaldehyde concentration of around 0.1 mM (Heck and Casanova 2004).
Formaldehyde-generating processes include methanol oxidation, methylamine deamination, and histone demethylation (O'Sullivan et al. 2004; Cloos et al. 2008; Lee et al. 2008; Hou and Yu 2010), while formaldehyde disposal occurs either by reduction to methanol via the cytosolic alcohol dehydrogenase (ADH) 1 (Friedenson 2011; MacAllister et al. 2011) or by oxidation to formate (Fig. 1). Formaldehyde oxidation is catalyzed by cytosolic ADH3 or by mitochondrial aldehyde dehydrogenase (ALDH) 2 (Friedenson 2011; MacAllister et al. 2011). Although ALDH2 acts directly on formaldehyde, the substrate for ADH3 is a formaldehyde-glutathione (GSH) adduct that is formed in an enzyme-independent reaction (Staab et al. 2009; Thompson et al. 2010; MacAllister et al. 2011). The formaldehyde-derived formate can either be exported from the cells via gamma-amino butyric acid receptors (Mason et al. 1990) and the monocarboxylate transporter 1 (Moschen et al. 2012) or can undergo further oxidation to CO2 (Fig. 1). Formate oxidation is a tetrahydrofolate (THF)-dependent two-step process where formate first reacts with THF to form 10-formyl THF in an ATP-dependent enzymatic reaction catalyzed by the cytosolic methylene THF dehydrogenase (MTHFD) 1 or by the mitochondrial MTHFD1L (Krupenko 2009; Krupenko et al. 2010) (Fig. 1). Subsequently, the cytosolic ALDH1L1 or its mitochondrial isoform ALDH1L2 oxidize 10-formyl THF to generate carbon dioxide (Krupenko 2009; Krupenko et al. 2010) (Fig. 1). To a lesser extent, catalase can also contribute to formate oxidation in a THF-independent manner (Skrzydlewska 2003).
The concentration of formaldehyde in the healthy human brain has been reported to be around 0.2 mM and 0.4 mM in the hippocampus and cortex, respectively (Tong et al. 2012). However, formaldehyde levels in the brain have been shown to increase even further with age and in Alzheimer′s disease (AD) (Tong et al. 2011, 2012). The reason for these increases in formaldehyde concentrations could be the elevated expression of formaldehyde-generating enzymes, as reported for the methylamine deaminating semicarbazide-sensitive amine oxidase (SSAO) in AD (Ferrer et al. 2002; del Mar Hernandez et al. 2005; Unzeta et al. 2007). Moreover, during aging increased activity of carboxymethyl esterase, an enzyme generating methanol by demethylation of post-translationally modified proteins, has been observed in the brain (Sellinger et al. 1988) which may foster elevated formation of formaldehyde by providing the substrate for methanol oxidation.
Animal studies have demonstrated that formaldehyde has toxic effects on the nervous system, causes oxidative damage in various brain areas including the cerebellum (Songur et al. 2008; Zararsiz et al. 2011) and compromises memory and learning (Songur et al. 2010; Tong et al. 2011, 2012). However, there is limited information on the formaldehyde metabolism in brain cells and on the molecular mechanisms underlying its neurotoxicity. Recently, we have shown that cultured astrocytes efficiently oxidize formaldehyde to formate and that formaldehyde exposure accelerates glycolytic flux and GSH export from glial cells (Tulpule and Dringen 2011, 2012; Tulpule et al. 2012). Here, we show that also primary cerebellar granule neurons have the potential to oxidize formaldehyde and that presence of formaldehyde accelerates lactate production and multidrug resistance protein (Mrp) 1-mediated GSH export from viable neurons.
Materials and methods
Fetal calf serum and penicillin/streptomycin solution were purchased from Biochrom (Berlin, Germany), minimal essential medium from Gibco Life Technologies (Darmstadt, Germany) and Earle′s balanced salt solution from Life Technologies-Invitrogen (Darmstadt, Germany). Cytosine β-d-arabinofuranoside, poly-d-lysine, hydrogen peroxide, N-acetyl cysteine, l-glycine, and soybean trypsin inhibitor were obtained from Sigma-Aldrich (Steinheim, Germany). Formaldehyde was purchased from Merck (Darmstadt, Germany) while l-glutamine was from Fluka (Buchs, Switzerland). Bovine serum albumin (BSA), NAD+, NADH, NADP+, and NADPH were from Applichem (Darmstadt, Germany). The enzymes glucose-6-phosphate dehydrogenase, glutamate pyruvate transaminase, glutathione reductase, hexokinase, and lactate dehydrogenase (LDH) were from Roche (Mannheim, Germany), while formaldehyde dehydrogenase and formate dehydrogenase were from Sigma-Aldrich. Primers for RT-PCRs were purchased from MWG Biotech (Ebersberg, Germany), RedSafe™ from HiSS Diagnostics (Freiburg, Germany), and the RNeasy® Mini Kit from Qiagen (Hilden, Germany). Deoxyribonucleoside triphosphate (dNTP) mix, DNA loading dye, GeneRuler™, 50 bp DNA ladder, RevertAid™H Minus First Strand cDNA Synthesis Kit, magnesium chloride, Taq buffer, and Taq polymerase were purchased from Fermentas (St. Leon-Rot, Germany). Other chemicals of the highest purity available were from Fluka or Merck. 96-well microtitre plates (transparent and black) and 24-well plates were from Nunc (Wiesbaden, Germany).
Cerebellar granule neuron-rich cultures were prepared from the brains of 7–8-day-old Wistar rats according to a method described by Anggono et al. (2008). Briefly, the cerebellae were dissected and a single cell suspension was produced by tryptic digestion with 2.5% (w/v) trypsin in phosphate buffered saline (PBS; 10 mM potassium phosphate buffer, pH 7.4, containing 150 mM NaCl) containing 14 mM glucose, 60 mM MgSO4 and 0.3% (w/v) BSA followed by manual trituration with glass pipettes. In wells of poly-d-lysine coated 24-well dishes, 0.75 million cells were seeded in 1 mL of culture medium (90% minimal essential medium, 10% heat-inactivated fetal calf serum, 30 mM d-glucose, 25 mM KCl, 2 mM l-glutamine, 100 U/mL penicillin G, and 100 μg/mL streptomycin sulfate). The cells were cultured at 37°C in a humidified atmosphere with 5% CO2 in a cell incubator (Sanyo, Osaka, Japan). After 1 day in culture, the medium was replaced by culture medium containing 10 μM cytosine β-d-arabinofuranoside. The cells were used for experiments at a culture age between 7 and 10 days.
Astrocyte-rich primary cultures were prepared from the brains of newborn Wistar rats (Hamprecht and Löffler 1985). Cells were seeded in culture medium in 24-well plates (0.3 million cells per 1 mL) and incubated in the humidified atmosphere of a Sanyo incubator with 10% CO2. The culture medium was renewed every seventh day and the cultures were used for experiments at an age between 15 and 21 days.
Experimental incubation of cells
If not stated otherwise, cultured neurons were washed with 1 mL pre-warmed (37°C) incubation buffer (IB; 145 mM NaCl, 30.4 mM KCl, 1.8 mM CaCl2, 1 mM MgCl2, 0.8 mM Na2HPO4, 20 mM HEPES, 5 mM glucose, pH 7.4) and exposed to formaldehyde in 200 μL IB containing formaldehyde and/or other compounds at 37°C. For incubations at 4°C, cells were washed with ice-cold (4°C) IB and incubated with 200 μL formaldehyde-containing IB at 4°C. The experimental incubations were terminated by collecting the incubation media and by washing the cells with 1 mL of ice-cold PBS. For experimental set-ups involving pre- and main-incubation the pre-incubation medium was aspirated, cells were washed once with 1 mL pre-warmed (37°C) IB and incubated further with 200 μL IB containing the indicated compounds. In experiments investigating the requirement of formaldehyde for the stimulated GSH export, 10 μL media samples were harvested at 1 h intervals to determine total glutathione (GSx) contents.
Determination of cell viability and protein content
Release of the cytosolic enzyme LDH and the membrane permeability for the fluorescent dye propidium iodide (PI) were assessed to determine compromised cell viability. LDH activity in cell lysates and media was measured using a microtitre plate-based photometric assay (Dringen et al. 1998). As a modification, some media samples were treated with Triton X-100 (final concentration of 0.5% (v/v)) to lyse potentially detached cells with intact cell membranes. PI staining of the cells was performed as previously described (Scheiber et al. 2010). The protein content of the cultured cells was determined after lysing of the cells with 200 μL 0.5 M NaOH by the Lowry method (Lowry et al. 1951), using BSA as a standard protein. The neuron cultures contained an average protein content of 51 ± 10 μg per well.
Determination of metabolites and enzyme activities
The concentrations of formaldehyde, formate, glucose, and lactate were determined by enzymatic assay systems as recently described (Liddell et al. 2009; Schmidt and Dringen 2009; Tulpule and Dringen 2012). Total glutathione contents (GSx = amount of GSH plus twice the amount of glutathione disulfide (GSSG)) and GSSG contents in cells and media were determined by the colorimetric Tietze cycling assay in microtitre plates (Hirrlinger and Dringen 2005). The specific activities of glutathione peroxidase, glutathione reductase, and catalase were determined by photometric assays as previously described (Dringen and Hamprecht 1997; Gutterer et al. 1999).
Cells seeded in 24-well plates were used for extraction of total RNA. The RNeasy® Mini Kit and the RevertAid™ H Minus First Strand cDNA Synthesis Kit were used for extraction of RNA and its reverse transcription for synthesis of cDNA following the instructions provided by the supplier. PCRs were performed as previously described (Tulpule et al. 2010) using the primer sets and annealing temperatures listed in Table 1. Gel electrophoresis was performed in a 2.5% agarose gel containing 1x RedSafe™.
Table 1. Primer sets and annealing temperatures used for reverse transcription PCRs
Annealing temperature (°C)
Amplicon size (bp)
Presentation of data
Data are presented as means ± SD of values that were obtained in n experiments (given in the panels of the figures) performed on independently generated cultures. Pictures showing PI staining or RT-PCRs were taken from representative experiments that were reproduced at least twice with comparable outcomes. The analysis of significance between groups of data was performed by anova followed by the Bonferroni post hoc test with *p < 0.05, **p < 0.01 and ***p < 0.001. Significance of differences between two data sets was analyzed with the paired t-test and is indicated by #p < 0.05, ##p < 0.01 or ###p < 0.001. p > 0.05 was considered not significant.
Viability of cultured neurons after exposure to formaldehyde
To test for potential neurotoxic effects of formaldehyde, cultured cerebellar granule neurons were incubated with formaldehyde in concentrations of up to 1 mM for up to 6 h (Fig. 2). Exposure of the cells to formaldehyde for 2 h did not significantly elevate the activity of LDH in the medium compared to controls, irrespective of whether the medium was treated with Triton X-100 or not (Fig. 2a). Accordingly, only a low number of PI-positive cells was found after a 2 h treatment without or with formaldehyde in concentrations of up to 1 mM (Fig. 2i–l). In contrast, all neurons in cultures that had been treated with 100 μM silver nitrate for 2 h or 6 h (positive control for toxicity) were PI-positive (Fig. 2h, m, r and w). Also, an incubation of neurons for 6 h with 0.1 mM formaldehyde did not compromise cell viability (Fig. 2b, c and t), while exposure of the cells to 0.5 mM or 1 mM formaldehyde caused a small but significant increase in the LDH activity in Triton X-100-treated media (Fig. 2b) and a matching loss in cellular LDH activity (Fig. 2c). However, exposure of the cells to 0.5 mM or 1 mM formaldehyde did not severely increase the number of PI-positive neurons (Fig. 2u and v), suggesting that the membranes of the cells that remained attached to the well under those conditions were intact.
Formaldehyde clearance by neurons
The observed resistance of neurons to formaldehyde toxicity could be associated with their capacity to metabolize the applied formaldehyde. RT-PCRs were performed to test for expression in cerebellar granule neuron cultures (as well as in astrocyte primary cultures as control cells) of the enzymes involved in formaldehyde metabolism (Fig. 1). Indeed, bands of the expected sizes (Table 1) were observed for mRNAs of formaldehyde-metabolizing enzymes for both cell culture types (Fig. 3), suggesting that neurons may indeed have the capacity to reduce formaldehyde to methanol and/or to oxidize formaldehyde via formate to CO2 (Fig. 1). Among the investigated enzymes, at best a very weak signal was observed for the mRNA of ALDH1L1 in cultured neurons, while this band was clearly detectable for astrocyte cultures (Fig. 3).
To test whether neurons are indeed able to oxidize formaldehyde, cerebellar granule neurons were exposed to 0.5 mM formaldehyde for up to 3 h and formaldehyde and formate concentrations were determined in the media (Fig. 4a, circles). A control incubation without cells but otherwise under identical conditions revealed that hardly any formaldehyde was lost from the medium by evaporation or cell-independent oxidation to formate (Fig. 4a, triangles). In contrast, a decline in the extracellular formaldehyde concentration and an increase in the concentration of formate in the medium were observed on incubation of neurons with formaldehyde (Fig. 4a). The linear changes in the concentrations of formaldehyde and formate in the medium were used to calculate formaldehyde clearance and formate release rates of 210 ± 24 nmol/(h × mg) and 51 ± 10 nmol/(h × mg), respectively (Fig. 4b). Treatment of neurons with 0.5 mM formaldehyde for 2 h also increased the cellular level of formate significantly from a basal level of 4 ± 3 nmol/mg to 17 ± 3 nmol/mg (Fig. 4c). Under these conditions, a reduction of formaldehyde is unlikely to contribute to the disappearance of formaldehyde, since no increase in the concentration of extracellular methanol was observed (data not shown).
Formaldehyde forms adduct with cellular macromolecules like proteins and DNA (Lu et al. 2010; Ospina et al. 2011). Such processes could contribute to the decline in detectable formaldehyde observed for cultured neurons. To differentiate between a chemical reaction of formaldehyde with macromolecules and metabolic formaldehyde oxidation, the cells were incubated with 0.5 mM formaldehyde for 2 h at 37°C or 4°C, as the low temperature is known to completely prevent cellular formaldehyde metabolism but not its chemical reactivity (Tulpule and Dringen 2012). Formate was not detectable in the medium of cells that had been exposed to formaldehyde at 4°C (Fig. 4d), while the disappearance of formaldehyde was lowered by around 50% compared to cells that had been exposed to the aldehyde at 37°C (Fig. 4d). The cell viability was not affected by incubation of cells with formaldehyde at these temperatures (data not shown).
Effect of formaldehyde exposure on lactate production and glucose consumption by neurons
Formate accumulating within cells has the potential to affect ATP production as formate inhibits cytochrome c oxidase (Nicholls 1975; Wallace et al. 1997) which in turn may stimulate glycolysis (Walz and Mukerji 1988). To test whether exposure of neurons to formaldehyde affects cellular glucose metabolism, cerebellar granule neurons were treated with formaldehyde in a concentration of 0.5 mM. Already after 15 min of incubation, a significantly elevated extracellular lactate concentration was determined for formaldehyde-treated cells compared with control neurons (Fig. 5a). Calculation of the lactate release rates (Fig. 5b) from the linear increase in extracellular lactate contents revealed that presence of formaldehyde almost doubled the lactate release rate compared to control cells from 0.89 ± 0.16 to 1.57 ± 0.27 μmol/(h × mg) (Fig. 5b). The doubling of the release of lactate from formaldehyde-treated neurons was accompanied by a significant 50% increase in glucose consumption compared to control cells (Table 2). Formaldehyde treatment also increased the ratio of lactate production to glucose consumption, although this alteration did not reach the level of significance (Table 2).
Table 2. Lactate release and glucose consumption by cultured neurons
Cultured neurons (n = 3) were incubated without or with 0.5 mM formaldehyde for 2 h and the concentration of lactate release into the medium and the concentration of glucose consumed from the medium were determined. The significance of differences (paired t-test) between the data obtained for cells incubated in the absence and the presence of formaldehyde is indicated as ##p < 0.01.
Formaldehyde treatment has been reported to stimulate GSH export from glial cells (Tulpule and Dringen 2011; Tulpule et al. 2012). Similarly, the application of formaldehyde to neurons caused a time- and concentration-dependent increase in extracellular GSx content (Fig. 6a and d) that was accompanied by a matching decline in cellular GSx contents (Fig. 6b and e), while the sum of cellular plus extracellular GSx remained almost constant for all the conditions used (Fig. 6c and f). In the absence of formaldehyde, cultured neurons exported, at best, low amounts of GSH while upon application of formaldehyde a rapid extracellular GSx accumulation was observed for the first 2 h of incubation which slowed down thereafter (Fig. 6a). After 6 h of incubation with 0.5 mM formaldehyde, the specific cellular and extracellular GSx contents were 1.3 ± 0.8 nmol/mg and 11.2 ± 1.6 nmol/mg, while control cells that had been incubated without formaldehyde contained 14.9 ± 0.8 nmol GSx/mg and had released 4.0 ± 0.4 nmol GSx/mg into the medium. Half-maximal effects on the cellular and extracellular GSx contents of neurons after incubation for 2 h were found for around 0.3 mM formaldehyde, while presence of 1 mM formaldehyde induced maximal effects (Fig. 6d and e).
Formaldehyde-accelerated GSH export from glial cells is mediated by Mrp1 (Tulpule and Dringen 2011; Tulpule et al. 2012). Mrp 1 transports anionic compounds such as GSH as well as its conjugates in an ATP-dependent manner (Keppler 2011). Since cultured neurons contain Mrp1 mRNA (data not shown), Mrp1 was considered a potential transporter to mediate formaldehyde-induced GSH export from neurons. Indeed, presence of the Mrp1 inhibitor MK571 (Hirrlinger et al. 2002; Minich et al. 2006) almost completely prevented the formaldehyde-induced extracellular accumulation of GSx (Fig. 7a) and the corresponding cellular decline of GSx (Fig. 7b), while the sum of cellular plus extracellular GSx was not affected by the absence or presence of formaldehyde and/or MK571 (Fig. 7c). Quantification of the GSSG content which contributes to the determined GSx amounts revealed that the very low GSSG levels in cerebellar granule neurons or their media were not increased after exposure of the cells to formaldehyde and/or MK571 (Fig. 7). None of these conditions compromised the cell viability as indicated by the absence of any significant increase in extracellular LDH activity (data not shown).
To investigate whether the formaldehyde-induced GSH export depends on the presence of formaldehyde, neurons were pre-incubated without or with 0.5 mM formaldehyde for 1 h after which the cells were exposed to fresh medium containing no or 0.5 mM formaldehyde (Fig. 8). The accelerated GSH export was only maintained by cells that were exposed to formaldehyde during the main incubation (Fig. 8a, filled triangles), while removal of formaldehyde (Fig. 8a, open triangles) slowed the further increase of extracellular GSx contents to values that were comparable to those of control cells that had no contact to formaldehyde (Fig. 8, circles). This is also reflected by the export rates calculated for the main incubation (Fig. 8b). The GSH export of cells that had been exposed to formaldehyde only during the pre-incubation did not significantly differ from the values determined for cells that had no contact with formaldehyde [about 1.5 nmol/(h × mg)], while the presence of 0.5 mM formaldehyde during the main incubation increased the GSH efflux rate by three-fold (Fig. 8b).
To test whether a formaldehyde treatment inactivates the enzymes involved in GSH synthesis, cerebellar granule neurons were pre-incubated without or with 0.5 mM formaldehyde for 3 h and subsequently incubated for 3 h in formaldehyde-free medium with the extracellular GSH precursors glutamine (1 mM), N-acetyl cysteine (1 mM), and glycine (1 mM). Within this main incubation, the GSx content of formaldehyde-treated cells was restored to 17.2 ± 1.3 nmol/mg and was almost identical (p > 0.05, n = 3) to the values obtained for control cells that had been pre-incubated in the absence of formaldehyde (16.8 ± 2.3 nmol/mg).
Effects of formaldehyde on antioxidative enzymes and peroxide resistance
To investigate a potential effect of formaldehyde on the activity of antioxidative enzymes in neurons, the cells were incubated for 2 h without or with 0.5 mM formaldehyde and the activity of antioxidative enzymes were determined. The specific activities of glutathione peroxidase, glutathione reductase, and catalase of control cells that had been incubated without formaldehyde were 10.3 ± 6.4 nmol/(min × mg), 30.1 ± 3.9 nmol/(min × mg), and 26.7 ± 4.1 μmol/(min × mg), respectively. These values were not significantly altered (p > 0.05, n = 3) on incubation of the cells with formaldehyde.
To test whether the formaldehyde-induced loss of cellular GSH (Fig. 6) may increase the vulnerability of the cells toward oxidative stress, cerebellar granule neurons were pre-incubated with 0.5 mM formaldehyde for 3 h and subsequently exposed for 4 h to hydrogen peroxide in concentrations of up to 300 μM (Fig. 9). Both the pre-incubation with formaldehyde and the main incubation with peroxide did not severely compromise cell viability, as the extracellular LDH activity increased at best by up to 15% compared to control cells (absence of formaldehyde during pre-incubation and absence of peroxide during main incubation).
Cultured cerebellar granule neurons were used as a model system to investigate the consequences of a treatment of neurons with formaldehyde. Exposure of these cells to formaldehyde in concentrations of up to 1 mM was not acutely toxic. This resistance of cultured cerebellar granule neurons against toxicity of low concentrations of formaldehyde for short time-frames is consistent with literature data for cortical neurons (Song et al. 2010), PC12 cells (Lee et al. 2008), astrocytes (Tulpule and Dringen 2011, 2012), and oligodendroglial OLN-93 cells (Tulpule et al. 2012). Cell viability of cerebellar granule neurons was only compromised during prolonged exposure of the cells to formaldehyde which may be caused by the observed alterations in cellular metabolism of formaldehyde-treated neurons and/or by the impairment of cellular functions because of its reactivity with cellular macromolecules (Lu et al. 2010; Ospina et al. 2011). Animal experiments have revealed that formaldehyde causes neuronal damage by oxidative stress (Gurel et al. 2005; Zararsiz et al. 2006, 2007, 2011). However, at least for viable neurons formaldehyde exposure did not increase the cellular or extracellular levels of GSSG, suggesting that cultured cerebellar granule neurons did not suffer from severe oxidative damage under the conditions used here. Furthermore, under the conditions used an acute formaldehyde exposure did not inactivate GSH synthesis, did not significantly lower the specific activities of enzymes involved in peroxide detoxification and did not increase the vulnerability of the cells against hydrogen peroxide. Thus, cultured cerebellar granule cells appear to have sufficient antioxidative potential to prevent oxidative stress during an acute exposure to formaldehyde.
Rapid metabolism of formaldehyde can be excluded as potential reason for the relative resistance of cerebellar granule neurons against formaldehyde toxicity, as even after 3 h exposure still around 70% of the applied formaldehyde was detectable in the incubation medium. Nevertheless, cultured cerebellar granule neurons have the potential to metabolize formaldehyde and also consistently express the mRNAs for enzymes known to contribute to formaldehyde reduction to methanol (ADH1) and for formaldehyde oxidation to formate (ADH3 and ALDH2), confirming literature data on the expression of these enzymes in neurons in brain sections (Martinez et al. 2001; Galter et al. 2003; Su et al. 2011). Also, the mRNAs of the 10-formyl THF synthesizing enzymes, MTHFD1 and MTHFD1L, were found in cultures of cerebellar granule neurons. While MTHFD1 has been previously shown to be expressed in neurons (Anthony and Heintz 2007), the presence of MTHFD1L has been demonstrated at least for adult brain tissue (Prasannan et al. 2003). mRNA of ALDH1L2 was detected for cultured cerebellar granule neuron and astrocytes, which is consistent with the presence of this enzyme in the brain (Krupenko et al. 2010). However, only a marginal expression of ALDH1L1 mRNA was found for the neuron cultures as expected from the reported highly specific expression of ALDH1L1 in astrocytes (Cahoy et al. 2008). The very weak signal for ALDH1L1 mRNA observed for cerebellar granule neuron cultures may result from a few contaminating astrocytes in these cultures.
Cultured cerebellar granule neurons are able to metabolize formaldehyde and cleared exogenous formaldehyde in a time- and temperature-dependent manner with a rate of about 200 nmol/(h × mg), which is similar to values recently reported for cultured astrocytes (Tulpule and Dringen 2012). Extracellular and cellular formate levels of formaldehyde-treated neurons accounted to 25% and 10%, respectively, of the formaldehyde that had disappeared within 2 h from the medium, while no methanol generation from formaldehyde was observed (data not shown). The rather low rate of formate generation by cerebellar granule neurons contrasts the situation reported for astrocytes where formaldehyde is almost quantitatively oxidized to formate (Tulpule and Dringen 2012). Evaporation of formaldehyde can be ruled out as the potential reason for the observed disappearance of formaldehyde as in the absence of neurons the decline in medium concentration of formaldehyde was only marginal. However, some interaction of formaldehyde with cellular macromolecules (Lu et al. 2010; Ospina et al. 2011) is likely to contribute to the observed loss of formaldehyde, since incubation of cells at 4°C prevented only half of the disappearance of formaldehyde, but completely prevented its oxidation to formate. In addition, since cultured neurons contain the mRNA for the enzymes responsible for formate oxidation, the complete enzymatic oxidation of some formaldehyde to CO2 is likely to also contribute to the observed cell-dependent loss of formaldehyde.
Cultured cerebellar granule neurons produced lactate at a basal rate of around 0.8 μmol/(h × mg) which is similar to the glycolytic rates previous reported for cultured neurons (Itoh et al. 2003; Zwingmann and Leibfritz 2003), but lower than that reported for cultured astrocytes (about 1.3 μmol/(h × mg)) (Scheiber and Dringen 2011; Tulpule and Dringen 2012). This was expected since astrocytes are reported to be more glycolytic than neurons (Itoh et al. 2003; Zwingmann and Leibfritz 2003) possibly because of an inhibited pyruvate dehydrogenase complex (Halim et al. 2010) and a lower capacity for NADH shuttling into the mitochondria (Berkich et al. 2007; Neves et al. 2012). Exposure of neurons to formaldehyde rapidly increased lactate release, doubled their lactate release rate and increased glucose consumption by 50%. This accelerated lactate release is likely to be predominantly a consequence of the accumulation of formaldehyde-derived formate in the cells which is a known inhibitor of cytochrome c oxidase (Nicholls 1975; Wallace et al. 1997), as inhibition of mitochondrial respiration has been reported to accelerate glycolytic lactate production in cultured neurons (Walz and Mukerji 1988). However, as neurons metabolize formaldehyde with a rate of 0.2 μmol/(h × mg), a part of the NADH that is used for pyruvate reduction may have been derived from formaldehyde and/or formate oxidation.
A formaldehyde-induced stimulation of glycolytic flux has also been observed for cultured astrocytes (Tulpule and Dringen 2012). However, the onset of the increase in astrocytic lactate production was delayed by about one hour and the lactate release rate was only increased by up to 50% (Tulpule and Dringen 2012), while the extracellular lactate concentration in formaldehyde-treated neuron cultures was already significantly elevated after 15 min and the lactate release rate was doubled compared to control neurons. These differences between cultured astrocytes and neurons in the modulation of glycolytic flux after formaldehyde application may be a consequence of the more efficient formate export from astrocytes (Tulpule and Dringen 2012) and the higher basal glycolytic flux rate in astrocytes (Bolanos et al. 2010).
Cultured neurons release GSH with a low specific rate, confirming literature data (Hirrlinger et al. 2002). The basal GSH release rate determined for cultured neurons (1.1 ± 0.5 nmol/(h × mg)) is lower than values determined for cultured astrocytes (1.5 ± 0.3 nmol/(h × mg)) but higher than that of OLN-93 cells (0.4 ± 0.3 nmol/(h × mg)) (Tulpule and Dringen 2011; Tulpule et al. 2012). Application of formaldehyde to neurons strongly accelerated this basal GSH export in a concentration- and time-dependent manner in a process that was completely prevented by the Mrp1-inhibitor MK571, as recently also reported for cultured glial cells (Tulpule and Dringen 2011; Tulpule et al. 2012). Thus, formaldehyde-induced stimulation of Mrp1-mediated GSH export appears to be a common feature of neural cells, although the extent of stimulation differed between the brain cell types as the basal GSH release rate increased in presence of 0.5 mM formaldehyde by about four-fold, 10-fold, and three-fold in cultured neurons (present report), astrocytes, and OLN-93 cells (Tulpule and Dringen 2011; Tulpule et al. 2012), respectively. As formaldehyde application increases the Vmax-value of the GSH export from astrocytes 10-fold without changing the Km-value, a formaldehyde-induced activation of Mrp1 and/or recruitment of additional transporters to the plasma membrane have been suggested as reasons for the accelerated GSH export (Tulpule et al. 2012). Such processes may also be involved in the accelerated Mrp1-mediated GSH export from formaldehyde-treated neurons. Furthermore, the observation that the formaldehyde-induced effect on Mrp1-mediated GSH transport is terminated by removing formaldehyde suggests that a rapidly reversible process is involved in the stimulated GSH export from neurons that requires the acute presence of formaldehyde.
In addition to a direct Mrp1-mediated GSH export, a Mrp1-mediated export of the GSH-containing intermediates of the formaldehyde metabolism, S-hydroxymethyl-GSH and S-formyl GSH, could also contribute to the observed accelerated GSH export in formaldehyde-treated neurons, as GSH conjugates are known as good Mrp1 substrates (Cole and Deeley 2006; Waak and Dringen 2006). As a result of their known lability (Uotila 1981; Krieter et al. 1985) both S-hydroxymethyl-GSH and S-formyl-GSH are likely to liberate GSH rapidly after export from the cells. Considering a potential export of S-formyl-GSH from formaldehyde-treated neurons, it is tempting to speculate that this export may be responsible for the extracellular accumulation of both GSH and formate. However, as formate release from the cells (100 nmol/(2 h × mg)) is around 10 times faster than GSH release (10 nmol GSH/(2 h × mg)), a simultaneous release of formate and GSH in the form of S-formyl-GSH could only explain the observed accelerated GSH export but cannot be responsible for the majority of the observed formate export.
The concentrations of formaldehyde that were effective in stimulating glycolytic lactate production and GSH export and to generate formate in cultured neurons (present report) or cultured glial cells (Tulpule and Dringen 2011, 2012; Tulpule et al. 2012) are in the range of formaldehyde concentration that have been reported for blood (0.1 mM), brain cortex (0.4 mM), and hippocampus (0.2 mM) (Heck and Casanova 2004; Tong et al. 2011, 2012). Thus, the observed metabolic consequences of an exposure of brain cells to formaldehyde are likely to be relevant for the in vivo situation, especially under conditions that have been connected with elevated brain formaldehyde levels such as aging, AD, or multiple sclerosis (Tong et al. 2012, 2011; Khokhlov et al. 1989 cited in Miao and He 2012; Ferrer et al. 2002; del Mar Hernandez et al. 2005; Unzeta et al. 2007; Airas et al. 2006).
The observed formaldehyde-induced alterations in the metabolism of brain cells may severely compromise brain functions. Excess of formate generation has adverse effects on brain cells as extracellular formate is known to be neurotoxic (Kapur et al. 2007), and elevated cellular levels of formate will lower mitochondrial ATP production (Wallace et al. 1997). In addition, accelerated proton-coupled export of lactate and formate will result in metabolic acidosis which is known to compromise cell viability (Yao and Haddad 2004; Rose 2010). Furthermore, accumulation of extracellular GSH and depletion of neuronal GSH has been shown to cause excitotoxicity (Regan and Guo 1999b; Lee et al. 2010) and a combination of extracellular GSH accumulation with energy depletion further potentiates neuronal death (Regan and Guo 1999a). Finally, acidosis may even impair neuronal restoration of GSH as a lowered pH has been reported to inhibit GSH synthesis (Lewerenz et al. 2010). Thus, elevated formaldehyde generation, formaldehyde-derived formate generation as well as formaldehyde-induced alterations in glucose and GSH metabolism in brain are likely to trigger a number of events that will synergistically compromise viability and functions of brain cells. These processes may contribute to the reported formaldehyde-induced oxidative damage, neurotoxicity, and impaired cognitive potential (Songur et al. 2010; Tong et al. 2011, 2012).
In summary, cerebellar granule neurons are capable of oxidizing formaldehyde and severely increase their lactate production and GSH release during exposure to formaldehyde. Such processes may contribute to the known neurotoxic potential of formaldehyde that have been reported for various brain regions including the cerebellum (Songur et al. 2008; Zararsiz et al. 2011). Since aging and disorders such as AD and multiple sclerosis have been connected both with elevated brain formaldehyde levels (Tong et al. 2012, 2011; Khokhlov et al. 1989 cited in Miao and He 2012), and with oxidative damage and impaired energy metabolism (Haider et al. 2011; Correia et al. 2012; Leuner et al. 2012; Sohal and Orr 2012; Soler-Lopez et al. 2012), further studies are now required to investigate whether elevated brain formaldehyde levels as well as a potentially accelerated formate generation contribute to the disturbances observed in brain metabolism for these conditions.
M.C. Hohnholt thank Dr. Karen Smillie and Professor Michael M. Cousin (Edinburgh, Scotland) for training her in the preparation of cerebellar granule neuron cultures.
Conflict of interest
The authors have no conflict of interest to declare.