Nitric oxide is the primary mediator of cytotoxicity induced by GSH depletion in neuronal cells

Glutathione (GSH) is the most abundant low molecular weight thiol in mammalian cells and acts as the major cellular non-enzymatic antioxidant. This tripeptide is synthesized from its amino acid precursors in two steps: the enzyme γ-glutamylcysteine synthetase (γ-GCS) catalyzes the synthesis of γ-glutamyl-cysteine from glutamate and cysteine; glutathione synthetase (GS) then catalyzes the formation of GSH from glycine and γ-glutamyl-cysteine. It is present as a reduced form and two oxidized species: glutathione disulfide and glutathione mixed disulfide with protein thiols. Beside its function as an intracellular redox buffer, GSH participates as a cofactor of glutathione peroxidase in the detoxification of lipid and organic peroxides (Liddell et al., 2006). It also possesses several other roles related to its ability to modulate the thiol status of cysteine on several proteins, including receptors, molecules involved in signaling and nuclear transcription factors (Filomeni et al., 2002).

In the past few years, many studies have shown that GSH content decreases with age in many tissues from different animal species, but the underlying mechanism is not still clear (Liu et al., 2004). It has been proposed that GSH decrease could be a consequence of the formation of protein mixed disulfides. This event can broadly interfere with the catalytic efficiency of enzymes, which might be reflected in their reduced ability to mount adaptive responses under stressful conditions (Cotgreave and Gerdes, 1998; Droge, 2002; Rebrin et al., 2003; Rebrin et al., 2005). Moreover, GSH reduction is reported in patients and animal models of various neurodegenerative disorders. In particular, GSH levels were found to be lower in the substantia nigra of Parkinson's disease patients (Sian et al., 1994; Solano et al., 2008) and pyramidal neurons of Alzheimer's disease patients (Ballatori et al., 2009; Gu et al., 1998). The most convincing theory is that, in neurodegenerative diseases, the activity of both γ-GCS and GS progressively decreases through downregulation of gene expression (Ballatori et al., 2009; Sastre et al., 2005; Sethna et al., 1982). Additional roles for the antioxidant function of GSH are in the context of nitric oxide (NO) physiology and pathology (Hogg, 2002; Zhang and Hogg, 2005). NO is a gaseous free radical endogenously produced from L-arginine by the catalytic activity of NO synthases (NOSs). It has essential roles in diverse biological processes because of its facile but selective chemical reactivity towards a number of cellular targets. NO also undergoes reactions with oxygen, superoxide anion (O₂•⁻) and reducing agents to give products that themselves show selective chemical reactivity towards a variety of cellular targets, sometimes with a manifestation of toxic effects, such as nitrosative stress (Hughes, 2008). Such products of NO reactions include nitroxyl (HNO), oxides (NO₂, N₂O₄n and N₂O₃), peroxynitrite (ONOO) and S-nitrosothiols (RSNO). Direct reaction between NO and biological thiols can occur in anaerobic conditions, but is a very slow oxidation reaction that yields thiol disulfide and nitroxylanion. To form S-nitrosoglutathione (GSNO) and protein RSNO, the presence of oxygen is mandatory. In fact, NO is oxidized to N₂O₃, which is a better nitrosating agent (Hogg, 2002; Zhang and Hogg, 2005). As GSH is present intracellularly at millimolar concentrations, the formation of GSNO is plausible. GSNO generally has a longer lifetime than NO and protein RSNO, and is considered a more persistent or buffered pool of NO that can be released when or where it is required for signaling (Stamler et al., 1992). Therefore, GSH might contribute to overwhelm nitrosative stress. In fact, we previously demonstrated that, upon endogenous NO overproduction, a prompt increase in intracellular GSH levels was induced and was pivotal to prevent cell death (Baldelli et al., 2008). Moreover, upon treatment with physiological amounts of NO donors, an increase in GSH levels has been also reported (Moellering et al., 1999; Moellering et al., 1998).

On the basis of this knowledge, we investigated the possible role of GSH in modulating the level and reactivity of endogenous NO, with the aim of identifying the molecular mechanisms underlying aging and neurodegenerative disorders. Here, we report that, upon GSH depletion, endogenous NO is the primary factor affecting cell proliferation and viability through the NO-cGMP pathway. We also found that, upon inhibition of GSH synthesis, NO-mediated DNA damage and protein oxidation by NO, in terms of tyrosine nitration and S-nitrosylation, are operative in both neuroblastoma cells and primary cortical neurons (PCNs). This suggests that GSH could be an essential buffer of NO under physiological conditions and that its imbalance could be detrimentally linked to harmful effects from NO.

GSH decrease is considered to be among the causative factors of aging and neurodegenerative disorders. Therefore, we investigated the effects of GSH depletion by treatment with 1 mM BSO, a specific inhibitor of GSH synthesis. Intracellular GSH content was analyzed in human neuroblastoma SH-SY5Y cells by high-performance liquid chromatography (HPLC). Fig. 1A shows significant GSH depletion just 3 hours after BSO treatment and almost undetectable levels after 24 hours. The effect of BSO on cell growth and viability was analyzed by direct count by Trypan Blue staining. We found that, after 24 hours, BSO profoundly affected cell number (Fig. 1B) without any increase in dead cells (Fig. 1C), indicating that proliferation arrest was occurring. This was confirmed by anti-BrdU immunohistochemistry, which demonstrated a dramatic decrease in BrdU incorporation in BSO-treated cells after 24 hours, indicative of cell-cycle arrest in G1 phase (Fig. 1D).

In a previous work, we demonstrated that an increase in GSH level could be induced by neuronal NOS (nNOS) overexpression and that such an increase was efficient in buffering NO toxicity (Baldelli et al., 2008). We therefore attempted to evaluate whether NO imbalance can induce cell-cycle arrest. Indeed, we analyzed NOS activity by measuring nitrites and nitrates (NO_x) in the culture medium. Fig. 2A shows that BSO treatment leads to significant accumulation of NO_x after 24 hours without any significant increase in nNOS protein (Fig. 2B). To evaluate whether other NOS isoforms were responsible for the NO_x increase, we measured endothelial and inducible NOS by western blot, but were not able to find detectable levels of these enzymes according to SH-SY5Y neural origin (data not shown).

These data led us to speculate that NO_x accumulates following NO imbalance because of lack of its efficient removal by GSH and that an increase in NO could be responsible for cell proliferation arrest. We therefore treated the cells with BSO in the presence of the nNOS inhibitor L-NAME (100 μM). Fig. 2C,D shows that, after 24 hours, inhibition of nNOS activity was sufficient to counteract growth arrest elicited by GSH depletion, as assessed through cell count by Trypan Blue staining and cell proliferation analysis by BrdU incorporation. The same results were obtained using other inhibitors of the NO pathway, such as the NO scavenger carboxy-PTIO and the nNOS inhibitor 7-nitroindazole (7-Ni) (data not shown).

Several transcription factors and upstream signaling pathways regulating cell proliferation are directly modulated by NO and/or its downstream effector cGMP (Hofseth et al., 2003; Komatsu et al., 2009; Yun et al., 1999). Among these, we assayed for possible modulation of p53 using western blot analysis. We found that p53 accumulates from 3 hours after BSO administration to 9 hours after, finally returning to the basal level after 24 hours (Fig. 2E). Next, we evaluated mitogen-activated protein kinases (MAPKs); significant activation of ERK1/2 by phosphorylation was found, with a time course mirroring that of p53 (Fig. 2E). Phospho-active forms of p38 and JNK were not detectable in either control cells or BSO-treated cells up to 48 hours (data not shown). To verify whether ERK1/2 and p53 modulation depends on NO imbalance, we carried out BSO treatment in the presence of L-NAME. As reported in Fig. 2E, L-NAME efficiently impedes p53 and phosphorylated ERK1/2 (ERK1/2-P) accumulation, suggesting that NO is the primary mediator of such effects. cGMP is considered the main downstream effector of the signaling pathway governed by NO and has a role in control of the cell cycle and proliferation. This messenger might be involved in phosphorylation processes mediated by MAPKs, including ERK1/2 (Thomas et al., 2008). We therefore treated SH-SY5Y cells with BSO in the presence of LY83583 (LY), an inhibitor of guanylate cyclase (GC), the enzyme that catalyzes the formation of cGMP. Fig. 2F,G shows that LY significantly abrogates the effect of BSO on cell proliferation, as assessed by cell count and BrdU assay. LY was able to inhibit p53 and ERK1/2-P accumulation (Fig. 2H) similarly to L-NAME, indicating that the NO-cGMP-ERK1/2-p53 signaling axis is involved in BSO-mediated effects. Next, we asked whether cGMP-dependent protein kinase (PKG) contributes to cell proliferation arrest. We treated SH-SY5Y cells with BSO in the presence of PKG inhibitor (KT5823). KT5823 (KT) was not able to hamper cell proliferation arrest at 24 hours, as assessed by cell count (supplementary material Fig. S1A) and BrdU assay (data not shown). In line with this result, the accumulation of p53 and ERK1/2-P was still observed in the presence of KT (supplementary material Fig. S1B), demonstrating that NO and cGMP directly converge on the ERK1/2-p53 pathway.

To more thoroughly investigate the role played by ERK1/2 and p53 in proliferation arrest, we assessed the effect of U0126, a synthetic ERK1/2 inhibitor. U0126 (260 nM) per se caused a significant reduction in ERK1/2 phosphorylation, indicating that the inhibitor worked effectively (Fig. 3A). Suppression of ERK1/2 activation led to significant reduction of BSO-induced p53 accumulation, suggesting that ERK1/2 is an upstream modulator of p53. Cell count showed that U0126 also completely abolished the anti-proliferative effect mediated by BSO after 24 hours, suggesting a crucial requirement for the ERK1/2 pathway in promoting NO-cGMP-mediated growth arrest (Fig. 3B).

We subsequently investigated whether NO-cGMP-mediated growth arrest at 24 hours could finally result in cell death. We found that BSO induced a significant cytotoxic effect after 48 hours, as evident from the increase in the percentage of dead cells (Fig. 4A). Therefore, we investigated the possible occurrence of nitration stress upon BSO treatment by analyzing 3-nitrotyrosine (NO₂-Tyr), a marker of protein nitration stress, using immunofluorescence microscopy (Abello et al., 2009; Ferrer-Sueta and Radi, 2009; Goldstein and Merenyi, 2008). As shown in Fig. 4B, BSO induces a time-dependent increase in NO₂-Tyr after 9 hours, reaching the maximum level at 48 hours, that was efficiently counteracted by L-NAME. The same results were obtained by dot-blot analysis of NO₂-Tyr (Fig. 4C). In addition, we looked at the level of Ser139-phosphorylated histone H2A.X (H2A.X-P), a hallmark of stress-induced DNA double-strand breaks (d'Adda di Fagagna et al., 2003). This analysis revealed that nuclei of BSO-treated cells displayed a parallel increase in H2A.X-P fluorescence, which was efficiently reversed by L-NAME, indicating that DNA damage was operative and caused by NO imbalance (Fig. 4B).

Cytochrome c oxidase (CcOX) is considered a preferential target of NO because of the presence of heme groups for which NO has great binding affinity (Antunes and Cadenas, 2007). Therefore, we measured the activity of CcOX after BSO treatment and found that it was maximally reduced after 48 hours (−60%). Inhibition of NO synthesis by L-NAME completely reversed the impairment of CcOX activity (Fig. 4D). As shown in Fig. 4E, the protein level of CcOX was not altered by BSO, neither was that of other mitochondrial proteins such as cytochrome c and Hsp60, indicating that the decrease in its activity is a consequence of NO imbalance.

Other important targets of NO toxicity are microtubules, because tubulin is highly susceptible to nitration on tyrosine residues (Dremina et al., 2005; Eiserich et al., 1999; Fiore et al., 2006). We investigated by fluorescent microscopy the possible effect of BSO treatment on microtubule lattices. Fig. 4E shows that, 48 hours after BSO treatment, cells display marked changes in the morphology of the microtubule network, and acquire a rounded shape with a reduced increase in the number of neurites. Treatment with L-NAME completely prevented microtubule impairment, suggesting that this phenomenon is NO dependent.

We have demonstrated that ERK1/2 inhibition by U0126 or LY is able to prevent cell growth arrest after 24 hours. We then evaluated the effect of such inhibition on SH-SY5Y cells treated with BSO at later time points. This analysis revealed that U0126 was still able to counteract cell proliferation arrest and death after 48 hours (Fig. 5A); however, NO cytotoxicity was not abolished. In fact, protein nitration and DNA damage were even more pronounced, as assessed by the analysis of NO₂-Tyr and H2A.X-P levels (Fig. 5B). This condition gives rise to cell death later, at 72 hours, when the percentage of dead cells upon BSO and U0126 treatment was comparable to that observed at 48 hours upon treatment with BSO alone (Fig. 5A). The same results were obtained with LY (data not shown).

To determine the mechanisms by which SH-SY5Y cells die upon BSO treatment, we examined for apoptosis induction. We carried out cytofluorimetric analysis of cell nuclei stained with propidium iodide after 48 hours. We found that a comparable percentage of cells in sub-G1 phase (apoptotic cells) were observed in control and BSO-treated cells (Fig. 5C). Also, the level of the precursors pro-caspase-8 and -9 was not altered at either 24 or 48 hours after BSO treatment, thus excluding their activation (Fig. 5D). To further validate the absence of caspase-dependent apoptosis, we carried out western blot analysis of poly (ADP-ribose) polymerase (PARP), a well-accepted target of caspase-3, the executor caspase downstream of caspase-8 and -9. As shown in Fig. 5D, the PARP-cleaved band does not accumulate after treatment with BSO.

It has been demonstrated that NO can induce cell death by a caspase-independent pathway (Langer et al., 2008; Uchiyama et al., 2002). We examined whether BSO could activate such a pathway, focusing on the mitochondrial protein apoptosis-inducing factor (AIF). AIF is known to translocate into the nucleus, after an apoptotic stimulus, where it participates in considerable DNA cleavage (Cande et al., 2002; Lorenzo and Susin, 2007). We performed immunofluorescence analysis and, as expected, found that AIF localized in the mitochondria of control cells, as evident from cytochrome c counterstaining (Fig. 6A). By contrast, in BSO-treated cells, AIF moved into the nuclei, particularly at 48 hours, as demonstrated by the superimposition of the AIF signal with that of nuclei and the loss of AIF mitochondrial localization. Treatment with BSO in the presence of L-NAME completely prevented mitochondrial export of AIF towards the nucleus, strongly indicating that NO toxicity is the sole inducer of caspase-independent apoptosis. In fact, cell counting by Trypan Blue staining showed that L-NAME and 7-Ni significantly reduced the cytotoxic effect of BSO at 48 hours (Fig. 6B). The involvement of cGMP in BSO toxicity was investigated by analyzing the effect of treatment with BSO in combination with the cGMP inhibitor LY. As reported in Fig. 6B, LY was able to completely counteract BSO-induced cell death at 48 hours, further implicating cGMP and NO in BSO toxicity. By contrast, the PKG inhibitor KT was not able to prevent BSO-induced cell death (Fig. 6B).

To thoroughly dissect the molecular mechanisms involved in BSO toxicity, we carried out experiments with SH-SY5Y cells transfected with a vector containing a cDNA encoding the human antioxidant enzyme copper-zinc superoxide dismutase (hSOD1 cells). These cells are resistant to NO-induced apoptosis (Ciriolo et al., 2000). Cells transfected with the empty vector (mock cells) were used as control. SOD1 can intercept O₂•⁻ and inhibit its reaction with NO to form ONOO⁻, thus limiting cellular stress. In hSOD1 and mock cells, GSH decrease upon treatment with 1 mM BSO followed the same trend as that observed for untransfected SH-SY5Y cells (data not shown). We then measured NO₂-Tyr levels in the presence of BSO and/or L-NAME by immunofluorescence analysis. As shown in Fig. 7A, mock cells treated with BSO display an increase in NO₂-Tyr with respect to untreated cells that is efficiently counteracted by treatment with L-NAME. By contrast, hSOD1 cells, which have increased levels of SOD1 (Fig. 7A), were not affected by BSO treatment; indeed, no formation of NO₂-Tyr adducts was evident (Fig. 7A). Furthermore, we analyzed cell proliferation and death of mock and hSOD1 cells after BSO addition. Fig. 7B,C shows that, in mock cells, BSO induces significant growth arrest at 24 hours and death at 48 hours; both effects are counteracted by L-NAME. On the other hand, BSO did not induce any detrimental effects on cell proliferation and viability of hSOD1 cells (Fig. 7B,C), implying a functional role for ONOO⁻ in the induction of both cell growth arrest and death.

To further evaluate whether GSH depletion is detrimental only in the presence of NO, we tested other two cell lines: mouse neuronal NSC34 cells expressing high levels of nNOS (Baldelli et al., 2008) and human cervical cancer HeLa cells, which do not express any NOS isoforms (Bulotta et al., 2001; Nisoli et al., 2003). These cells were treated with BSO in the presence or absence of L-NAME, and viable and dead cells were counted after 48 hours. Fig. 8A shows that NSC34 cells are susceptible to BSO as well as SH-SY5Y, and that L-NAME is able to inhibit both cell death and NO₂-Tyr accumulation (Fig. 8B). On the contrary, no toxicity was evident in HeLa up to 48 hours (Fig. 8C), time when significant inhibition of GSH synthesis (−80%) was achieved (data not shown), indicating that GSH depletion alone is not sufficient to cause cytotoxicity.

To assess the validity of our results, we examined the effects of GSH depletion on an ex vivo experimental model represented by mouse PCNs. In these cells, a different concentration of BSO was used (25 μM), which has previously been reported to be efficient in decreasing GSH levels in mouse PCNs (White and Cappai, 2003). Fig. 9A shows that, under our experimental conditions, BSO induces a significant decrease in GSH, starting after 24 hours.

To examine the effect of GSH depletion on viability, PCNs were treated with BSO in the presence or absence of 100 μM L-NAME, and morphological changes were evaluated by optical microscopy. The analysis of PCNs exposed to BSO for 48 hours identified widespread death, and neuritic membrane beading and blebbing consistent with apoptosis (Fig. 9B). Pre-treatment with L-NAME prevents these changes, suggesting that NO-mediated toxicity is also found in PCNs. Moreover, staining with Hoechst 33342 (Hoechst) was used to visualize the morphological changes of nuclei and to estimate cell viability and/or apoptosis commitment. The quantitative data obtained from nuclei counts are reported in Fig. 9C. In particular, whereas control cells exhibit uniformly dispersed chromatin and intact nuclei, BSO-treated cells display a significant increase in nuclei with condensed chromatin and the appearance of apoptotic bodies (70%). The presence of L-NAME markedly decreased the number of PCNs with nuclear morphological changes, as the percentage of condensed-fragmented nuclei returned to control values. The same protective effects were obtained by treating PCN with BSO in combination with LY (Fig. 9C). On the contrary, the use of KT did not prevent death, demonstrating that, in PCNs, BSO toxicity is mediated by the NO and cGMP pathway without the involvement of PKG. Fig. 10A shows representative images of nuclei stained with Hoechst after 48 hours of BSO treatment. Data reported in supplementary material Fig. S2 show that BSO is able to significantly induce PCN death after 24 hours, which was efficiently prevented only by L-NAME and LY co-treatment.

Finally, we measured the possible occurrence of BSO-mediated nitration stress by examining the NO₂-Tyr levels through immunofluorescence analysis. As shown in Fig. 10A,B, BSO induces accumulation of NO₂-Tyr levels at 48 hours that is efficiently counteracted by L-NAME. The same results were obtained by dot-blot analysis of NO₂-Tyr levels (Fig. 10B).

Nitration of proteins is considered a less reversible process with respect to S-nitrosylation (Abello et al., 2009). The former is usually associated with cell death, the latter with signal transduction pathways leading to either cell death or survival (Brune, 2003). Therefore, we investigated the possible occurrence of S-nitrosylation upon BSO treatment by fluorescence microscopy using a specific S-nitrosocysteine (S-NO) antibody. Fig. 10A shows that BSO significantly upregulates the levels of S-NO proteins and that such a phenomenon was abolished by L-NAME. To verify whether the S-nitrosylation process was selective to some proteins or nonspecific, we carried out the biotin-switch method on PCNs treated with BSO. Then, we identified S-NO proteins by western blot analysis of the biotin adducts by incubating cell membrane with HRP-conjugated streptavidin. Fig. 10C shows that a wide array of proteins are S-nitrosylated, indicating that S-nitrosylation is not a triggering event for BSO toxicity.

Growth factors and stimulation of N-methyl-D-aspartic acid receptor (NMDA-R) are known to activate the synthesis of NO by nNOS in neuronal cells (Clementi et al., 1995). Whereas growth factors are protective, the excessive stimulation of NMDA-R results in cytotoxicity (Gu et al., 2010). We therefore investigated the effect of NO imbalance mediated by GSH depletion in the presence of such nNOS activators. In particular, PCNs were treated for 24 hours with BSO in the presence of 20 ng/ml epidermal growth factor (EGF) or 20 μM NMDA. The protective effect of EGF and the harmful effect of NMDA in BSO-untreated cells were confirmed. In fact, PCNs treated with EGF or NMDA displayed lower or higher levels of condensed-fragmented nuclei than controls, respectively (Fig. 11A). The co-treatment EGF-BSO significantly increased cell death with respect to BSO alone (50% versus 30%) (Fig. 11A), indicating that NO increase mediated by EGF (supplementary material Fig. S3A) acts in synergy with that deriving from BSO. This phenomenon was more evident upon NMDA-BSO co-treatment (BSO 30%; NMDA-BSO 60%), according to the higher rate of NO production upon NMDA-R stimulation. The same enhanced effect was obtained upon EGF-NMDA co-treatment (NMDA 33%; EGF-NMDA 60%) (Fig. 11A).

The buffering role of GSH against NO was confirmed in PCNs by analyzing the intracellular content of GSH after NMDA treatment. Fig. 11B shows that GSH is significantly decreased as soon as 15 minutes after NMDA addition. This time corresponds to the maximum increase in NO production, as measured by the vital NO fluorescent probe DAF-2DA (Fig. 11C).

Finally, we determined the levels of NO₂-Tyr and S-NO proteins upon EGF or NMDA treatment. Fig. 11D shows that EGF alone does not result in any nitration or nitrosative stress, according to its protective role. However, it exerted an enhancing effect on eliciting NO-mediated protein damage upon co-treatment with either BSO or NMDA (Fig. 11D). In agreement with its cytotoxic action, NMDA was able to mediate the accumulation of both S-NO and NO₂-Tyr proteins, which was further enhanced by co-treatment with BSO or EGF (Fig. 11D). Addition of L-NAME completely abolished the nitration and S-nitrosylation events in all the conditions tested (data not shown).

Many neurodegenerative diseases and the physiological aging process share oxidative and nitrosative stress as common terminal processes. In association with these phenomena, progressive alteration of the GSH pool is also observed (Ballatori et al., 2009), but to date the synergistic effect of such events in favoring aging and its related disorders has not been clarified. Previously, we demonstrated that chemical inhibition of GSH synthesis results in an increase in free radicals and that either SOD1 or nNOS modulation are able to affect GSH homeostasis. We suggested that possible cross-talk between the two enzymes and GSH was operative, especially in assuring neural cell integrity (Aquilano et al., 2006; Baldelli et al., 2008). The data here obtained add another piece to the puzzle of the role of GSH as a direct antioxidant. In particular, it has emerged that GSH, besides providing protection from O₂•⁻, counteracts the potential toxicity of physiological levels of NO. The importance of GSH in counteracting NO-mediated cytotoxicity has been largely proved in a variety of cell types. In particular, upon treatment with NO donors, an increase in GSH and augmented cellular susceptibility to NO challenge, after chemical depletion of GSH, have been reported (Moellering et al., 1999; Moellering et al., 1998). In a previous work, we found that GSH was fundamental to preventing the harmful effects of an increase in NO production after nNOS overexpression (Baldelli et al., 2008). The most intriguing finding of the present work is that, when the GSH pool is altered, NO derived from basal nNOS activity can be equally detrimental both in neuroblastoma cells and in PCNs. Interestingly, in PCNs, we observed that NO produced downstream of NMDA-R stimulation promptly causes a decrease in the GSH pool, thus reinforcing the idea of a buffering role for GSH against NO. In fact, NO₂-Tyr accumulation is significantly induced in neuronal cells and prevented by using an NO scavenger or inhibiting NO synthesis. Also, the accumulation of H2A.X-P, which reflects the occurrence of DNA strand breaks, is totally prevented upon NO inhibition. The effectiveness of SOD1 overexpression in avoiding BSO-mediated nitration damage to proteins finally highlights the synergistic activity of O₂•⁻ and NO in triggering oxidative stress. This suggests that ONOO⁻ could be the primary cause of cell damage upon GSH depletion, although other pathways of protein nitration could be operative, such as those governed by its derivatives or by NO₂• produced by various heme-peroxidases or by the interaction of NO with tyrosyl radicals (Abello et al., 2009).

Being a heme-containing enzyme, CcOX is an important sensor of NO, strongly inhibiting its activity. In 1994, four independent laboratories reported that NO, either exogenously added to several preparations containing mitochondria or endogenously produced in intact cells, inhibited CcOX. However, in the three laboratories that used exogenous NO, the effect on CcOX was found to be reversible (Bolanos et al., 1994; Brown and Cooper, 1994; Cleeter et al., 1994; Schweizer and Richter, 1994) and in competition with O₂, whereas in the other study, which relied upon endogenous or sustained NO formation, inhibition of CcOX activity was found to be progressive and persistent. The molecular mechanism responsible for the irreversible inhibition of CcOX by ONOO⁻ was further investigated and revealed by Sharpe and Cooper, and Pearce et al. (Pearce et al., 1999; Sharpe and Cooper, 1998). Regardless of the specific mechanism leading to irreversible damage of CcOX by ONOO⁻, it is now clear that this interaction might have as yet unconsidered patho-physiological implications (Cooper et al., 2003).

This work also revealed that NO imbalance and the consequent activation of the NO and cGMP pathway, rather than alteration of the GSH redox system, are the main cause of cellular homeostasis impairment. The involvement of NO is also supported by the fact that other neural cell lines expressing nNOS, such as NSC34 (Aquilano et al., 2007), but not NOSs-lacking cells, such as HeLa (Bulotta et al., 2001; Nisoli et al., 2003), are susceptible to BSO toxicity. Surprisingly, PKG, which is known to be the main protein activated by NO and cGMP, does not seem to be involved in our experimental systems, as the use of a PKG inhibitor does not change the cytotoxic effect of BSO. In the absence of GSH, prevention of proliferation arrest and neuronal death is completely achieved by inhibiting NOS and GC, implying that alternative pathway(s) can be activated. In this context, we found that impairment of cell proliferation is due to the NO- and cGMP-mediated activation of the ERK1/2-p53 pathway. Actually, the use of the MEK1/2 inhibitor U0126 and the GC inhibitor LY clearly demonstrated that the transient activation of ERK1/2 is fundamental to inducing p53 and arresting cell proliferation. In fact, by impeding ERK1/2 phospho-activation, both p53 accumulation and growth arrest are inhibited. However, both ERK1/2 and p53 rapidly return to basal levels, excluding their contribution to the induction of BSO-mediated apoptosis. This assumption is also confirmed by the fact that cell death, even if at later time points, is equally obtained upon BSO treatment in the presence of ERK1/2 or GC inhibitors, probably due to the irreversible and additional accumulation of nitration damage.

According to previous studies, NO is able to inhibit at nanomolar concentration the activity of cysteine protease caspases through S-nitrosylation (Kim et al., 1997). However, apoptosis can proceed in the absence of caspase activation by a caspase-independent pathway (Tait and Green, 2008). A master regulator of such a pathway is AIF, which, after alteration of mitochondrial homeostasis, migrates into the nucleus, where it participates with endonuclease G in nuclear fragmentation (Lorenzo and Susin, 2007). It has been established that NO is able to trigger mitochondrial alteration and promote AIF release from mitochondria (Vieira and Kroemer, 2003). Our data are consistent with NO-dependent induction of AIF nuclear translocation, as inhibition of NO production upon BSO treatment is able to completely prevent this phenomenon. Inhibitor of GC and L-NAME are effective in inhibiting cell death; therefore, it is likely that cGMP plays a crucial function in transducing the signal to the apoptotic machinery. To the best of our knowledge, cross-talk between cGMP signaling and AIF has scarcely been investigated and deserves further research. The strength of this present study is that PCNs show the same detrimental effects upon GSH depletion. Indeed, the increase in NO-mediated damage and cell death was efficiently prevented by inhibiting NOS. In particular, both nitration and nitrosylation stress on proteins is strongly induced after GSH depletion, with S-nitrosylation probably representing a side-effect of NO imbalance and not being directly involved in signal transduction leading to cell death. This hypothesis is supported by the widespread increase in S-nitrosylated proteins upon BSO challenge. From the results obtained, it can be assumed that GSH deficiency could have a comparable detrimental effect with respect to the excitotoxic activity of NMDA-R stimulation. Likewise, although EGF is implicated in cell protection, it can enhance the toxic activity of NO upon BSO and NMDA treatment.

In PCNs, as well as the condensation of nuclear chromatin, typical nuclear apoptotic bodies are formed; this deserves further study to better characterize the apoptotic process (caspase dependent or caspase independent). It is now attractive to verify whether the observed NO imbalance can also be effective in mice in whichBSO and L-NAME have been successfully used to inhibit GSH and NO synthesis in several tissues and organs (Cattan et al., 2008; Orucevic and Lala, 1996; Stevanovic et al., 2009; Tunctan et al., 2006; Watanabe et al., 2003). Work is in progress in our laboratory to dissect this issue.

It has been reported that the level of GSH in peripheral blood leukocytes is a useful biomarker to detect redox imbalance in inherited disorders affecting mitochondrial function (Atkuri et al., 2009). Therefore, we can speculate on the great importance of measuring GSH levels to monitor disease status and response to antioxidants therapies in order to avoid its decrease and the consequent nitrative or oxidative stress.

In conclusion, our work demonstrates that neuronal cells are highly vulnerable to physiological flux of NO in the absence of GSH. In particular, our study showed that the intracellular depletion of GSH is able to induce cellular stress in NO-producing cells through a NO-dependent mechanism, such as inhibition of CcOX activity, DNA damage, and S-NO and NO₂-Tyr protein accumulation. Moreover, NO seems to be the only mediator of cell proliferation arrest through the ERK1/2-p53 signaling pathway and apoptosis through AIF nuclear translocation. The results obtained give support to the idea that GSH represents an important buffer of NO toxicity, both under physiological conditions and upon excitotoxic NMDA-R stimulation. They also help to understand the molecular mechanisms underlying neuronal cell loss during aging and its related neurodegenerative disorders.

Protease inhibitor cocktail, rabbit polyclonal anti-S-NO, mouse monoclonal anti-catalase, anti-SOD1, anti-β-tubulin, anti-actin and anti-p53, Triton X-100, propidium iodide, poly-L-lysine, dimethylsulfoxide (DMSO), NMDA, EGF and paraformaldehyde were from Sigma. Rabbit polyclonal anti-iNOS, anti-eNOS, anti-Hsp60 and anti-PARP, mouse monoclonal anti-cytochrome c, anti-nNOS and anti-GAPDH were from Santa Cruz Biotechnology (Santa Cruz, CA). Nitrocellulose membrane and IgG (H + L) HRP-conjugated mouse and rabbit secondary antibodies were from Bio-Rad Laboratories (Hercules, CA). 7-Ni, L-NAME, carboxy-PTIO, MEK1/2 inhibitor U0126, GC inhibitor LY83583 and the PKG inhibitor KT5823 were from Merck-Chemicals (Darmstadt, Germany). Anti-CcOX subunit IV, Hoechst 33342, and AlexaFluor568- and AlexaFluor488-conjugated secondary antibodies were from Invitrogen (Carlsbad, CA). Polyclonal rabbit anti-ERK1/2-P and anti-NO₂-Tyr were from Cell Signaling Technology (Danvers, MA). Mouse monoclonal anti-H2A.X-P, anti-procaspase-8 and -9 were from Millipore (Billerica, MA). ChemiGlow chemiluminescence substrate was from Alpha Innotech Corporation (San Leandro, CA). BrdU Cell Proliferation assay kit was from GE Healthcare (Buckinghamshire, UK). Rabbit polyclonal anti-AIF was from Biomol Research Laboratories (Plymouth Meeting, USA). All other chemicals were obtained from Merck Ltd (Darmstadt, Germany).

SH-SY5Y neuroblastoma cells and HeLa cervix carcinoma cells were purchased from the European Collection of Cell Cultures (Salisbury, UK). The NSC34 cells were a kind gift of Neil Cashman (Montreal Neurological Institute, Montreal, Quebec, Canada). SH-SY5Y, HeLa and NSC34 cells were grown in DMEM-F12, RPMI and D-MEM, respectively, supplemented with 10% fetal bovine serum (FBS), 2 mM glutamine, 100 U/ml penicillin/streptomycin (Lonza Sales, Basel, Switzerland) and maintained at 37°C in an atmosphere of 5% CO₂ in air. SH-SY5Y cells overexpressing human SOD1 (named hSOD1) were obtained as previously described (Aquilano et al., 2003).

BSO, a highly selective and potent inhibitor of the enzyme γ-GCS, was added in culture medium at a concentration of 1 mM unless otherwise stated. The NOS inhibitors L-NAME and 7-Ni were added in culture medium at concentrations of 100 μM and 10 μM, respectively. The GC inhibitor LY and the PKG inhibitor KT were added at concentrations of 2 μM and 4 μM, respectively. The NO scavenger carboxy-PTIO was added at a concentration of 2 μM. MEK1/2 inhibitor U0126 was added at a concentration of 260 nM. All these chemicals were added 1 hour before BSO treatment and maintained throughout the experiment. EGF and NMDA were used at concentrations of 20 ng/ml and 20 μM, respectively, and added together with BSO. All the concentrations of the chemicals used and the duration of the experiment were selected after performing time-course and dose-response curves.

Mouse PNC cultures were obtained from cerebral cortices of E15 C57BL-6 mice embryos, as previously described (Filomeni et al., 2010). All the experiments were performed according to the Animal Research Guidelines of the European Communities Council Directive (86/609/EEC).

Cells were directly counted by optical microscopy on a hemocytometer after Trypan Blue staining. Alternatively, cells were stained with 50 μg/ml propidium iodide according to Nicoletti et al. (Nicoletti et al., 1991) and analyzed by a FACScalibur instrument (Beckton and Dickinson, San Josè, CA). The percentage of cells in each phase of the cell cycle was evaluated by WinMdi 2.9 software (Joseph Trotter, Scripps, San Diego, CA). Cell proliferation was assayed by a BrdU Cell Proliferation assay kit, as previously described (Baldelli et al., 2008).

Cell viability in BSO-treated primary mouse neurons cultures was monitored by phase-contrast microscopy or staining the cells with the nuclear dye Hoechst (Lieberthal et al., 1998). Cells were seeded on glass cover slips, fixed with 4% paraformaldehyde and incubated with Hoechst for 10 minutes at room temperature. Fluorescence images were acquired on Olympus IX 70 equipped with Nanomover® and softWoRx DeltaVision (Applied Precision, WA) with a U-PLAN-APO 60× objective. Nuclei of ten different images were counted and the results were reported as the percentage of condensed-fragmented nuclei with respect to total cell number.

Cell pellets were resuspended in lysis buffer containing 10 mmol/L Tris-HCl, pH 7.4, 5 mmol/L EDTA, 150 mmol/L NaCl, 0.5% IGEPAL CA-630 and protease inhibitor cocktail. 20 μg total protein was electrophoresed on 8.5% or 10% SDS-polyacrylamide gels and blotted onto nitrocellulose membrane. Membranes were stained with primary antibodies against NO₂-Tyr (1:200), H2A.X-P (1:1000), pro-caspase-8 (1:1000), pro-caspase-9 (1:1000), p53 (1:5000), ERK1/2-P (1:200), catalase (1:5000), GAPDH (1:5000), β-tubulin (1:5000), Hsp60 (1:5000), actin (1:5000), CcOX subunit IV (1:1000), cytochrome c (1:2000), nNOS (1:500), eNOS (1:200), iNOS (1:200) and PARP (1:500). For dot-blot analysis, 20 μg protein was vacuum transferred to nitrocellulose membrane using a Bio-Rad dot-blot apparatus (Bio-Rad Laboratories). Nitrocellulose membrane was stained with anti-NO₂-Tyr (1:200). After incubation with the appropriate HRP-conjugated secondary antibodies, proteins were detected using a Fluorchem Imaging system (Alpha Innotech-Corporation, St Leandro, CA) upon staining with ChemiGlow chemiluminescence substrate. Densitometric analyses of protein bands were performed by the Quantity one software (Bio-Rad Laboratories, Hercules, CA). Immunoblots reported in the figures are representative of at least four experiments that gave similar results. Catalase, β-tubulin or GAPDH were used as loading control.

Intracellular GSH was determined as previously described (Baldelli et al., 2008). Data are expressed as nanomoles of GSH per milligram of protein.

Accumulation of NO_x in culture medium was measured by the Griess reaction, as previously described (Aquilano et al., 2007).

Biotin switch assay was performed as previously described (Jaffrey and Snyder, 2001). After protein separation by non-reducing SDS-PAGE and western blot, biotinylated proteins were detected by incubation of nitrocellulose membrane with HRP-conjugated streptavidin.

Immunofluorescence analysis was performed as previously described (Aquilano et al., 2010). Cells were incubated with anti-NO₂-Tyr, anti-AIF, anti-β-tubulin, anti-cytochrome c, anti-pH2A.X or anti-SOD1 (1:50). The levels of S-NO proteins were detected according to Iwakiri et al. (Iwakiri et al., 2006). Fluorescence images were acquired on Olympus IX 70 equipped with Nanomover (60× objective) or with a Nikon Eclipse TE200 epifluorescence microscope with a 100× objective (Nikon, Firenze, Italy). All images were captured under constant exposure time, gain and offset. The images reported in the figures are representative of at least four experiments that gave similar results.

CcOX was assayed spectrophotometrically as previously described (Ciriolo et al., 2000). Activity was expressed as Units (micromol cytochrome c oxidized min⁻¹) per milligram protein.

Proteins were assayed by the method of Lowry et al. (Lowry et al., 1951).

The results are presented as means ± s.d. Statistical evaluation was conducted by ANOVA, followed by the post Student-Newman-Keuls analysis. Differences were considered to be significant at P<0.05.

We gratefully acknowledge Palma Mattioli (Department of Biology, University of Rome Tor Vergata, Italy) for her assistance in fluorescent microscopy and image analysis. We are also grateful to the animal house staff of the University of Rome Tor Vergata, Italy. This work was partially supported by grants from Ministero della Salute and MIUR.