N-acetyl-cysteine and celecoxib lessen cadmium cytotoxicity which is associated with cyclooxygenase-2 up-regulation

In many neurodegenerative disorders, aggregates of ubiquitinated proteins are detected in neuronal inclusions, but their role in neurodegeneration remains to be defined. To identify intracellular mechanisms associated with the appearance of ubiquitin-protein aggregates, mouse neuronal HT4 cells were treated with cadmium. This heavy metal is a potent cell poison that mediates oxidative stress and disrupts the ubiquitin/proteasome pathway. In the current studies, the following intracellular events were found to be also induced by cadmium: (i) a specific rise in cyclooxygenase-2 (COX-2) gene expression but not COX-1; (ii) an increase in the extracellular levels of the proinflammatory prostaglandin E2, a product of COX-2; and (iii) production of 4-hydroxy-2-nonenal-protein adducts, which result from lipid peroxidation. In addition, cadmium treatment led to the accumulation of high molecular weight ubiquitin-COX-2 conjugates and perturbed COX-2 glycosylation. The thiol-reducing antioxidant N-acetylcysteine, and, to a lesser extent, the COX-2 inhibitor celecoxib, attenuated the loss of cell viability induced by cadmium demonstrating that oxidative stress and COX-2 activation contribute to cadmium cytotoxicity. These findings establish that disruption of the ubiquitin/proteasome pathway is not the only event triggered by cadmium. This oxidative stressor also activates COX-2 function. Both events could be triggered by formation of 4-hydroxy-2-nonenal as a result of cadmium-induced lipid peroxidation. Proinflammatory responses stimulated by oxidative stressors that mimic the cadmium effects may, therefore, be important initiators of the neurodegenerative process and exacerbate its progress.


INTRODUCTION
The ubiquitin/proteasome pathway plays a major role in the intracellular quality control process by degrading mutated or abnormally folded proteins to prevent their accumulation as aggregates [reviewed in (1)]. In many neurodegenerative disorders, however, aggregates of ubiquitinated proteins are detected in neuronal inclusions [reviewed in (2)]. A correlation between neuronal inclusions and cell death remains to be established (3;4). Nevertheless, recent findings demonstrated that protein aggregates directly impair the function of the ubiquitin/proteasome pathway, the latter known to be essential for cell survival (5).
It has become increasingly evident that functional changes in the ubiquitin/proteasome pathway are critical to the neurodegenerative process. For example, a mutant form of ubiquitin, known as Ub +1 , was detected only in brains of Alzheimer's disease (AD) patients and not in age matched controls (6). Ub +1 -capped polyubiquitin chains were shown to be refractory to disassembly by de-ubiquitinating enzymes and to potently inhibit proteasome degradation of a polyubiquitinated substrate (7). Moreover, particular areas of AD brains were found to exhibit compromised proteasome activities when compared to age matched controls (8). In addition, a decline in ubiquitination activity was shown to decrease cell viability in a cellular model of Huntington's disease (9). In this model, transfected rat striatal neurons co-expressing a nonfunctional ubiquitin-conjugating enzyme and a mutant huntingtin containing polyglutamine expansions showed a greater loss of cell viability than transfectants expressing the huntingtin mutant alone (9). Together, these results strongly support the view that disruption of the ubiquitin/proteasome pathway plays an important role in neurodegeneration.
Oxidative stress is another mechanism found to be involved in neurodegeneration.
Increasing evidence supports its role in neuronal death in disorders such as AD and Parkinson's by guest on March 24, 2020 http://www.jbc.org/ Downloaded from disease (PD). Studies with autopsied brains of AD patients show a co-localization of high levels of oxidative stress products with neurofibrillary tangles (NFT) and senile plaques (10).
Furthermore, signs of oxidative stress, such as lipid peroxidation and a decline in reduced glutathione (GSH), were detected in the substantia nigra in brains of PD patients (11). The production of free radicals by oxidative stress promotes partial unfolding of cellular proteins, resulting in exposure of previously buried hydrophobic domains to proteolytic enzymes (12)(13)(14) and to ubiquitin-conjugating enzymes (15). This sudden increase in protein substrates may compromise the capacity of the ubiquitin/proteasome pathway to clear the abnormal proteins and cause their aggregation and accumulation within the cell. However, the link between oxidative stress and the accumulation of ubiquitinated proteins in neurodegeneration remains to be established.
To address the relationship between oxidative stress and disruption of the ubiquitin/proteasome pathway in the neurodegenerative process, we chose to treat mouse neuronal HT4 cells with cadmium (Cd 2+ ). This heavy metal increases lipid peroxidation in organs such as the brain, which is particularly sensitive to Cd 2+ -toxicity (16). Although Cd 2+ is not a Fenton metal and thus, by itself, is unable to generate reactive oxygen species (ROS), free radical scavengers and anti-oxidants lessen Cd 2+ -toxicity, suggesting that the heavy metal elicits an increase in free radical production [reviewed in (17)]. Earlier studies (18) suggested a close link between Cd 2+ cytotoxicity and the ubiquitin/proteasome pathway as yeast mutants deficient in a specific ubiquitin-conjugating enzyme (UBC7) or a proteasome subunit (PRE1) were shown to be hypersensitive to the heavy metal. Moreover, our previous studies with neuronal cells (19;20) demonstrated that Cd 2+ disrupts intracellular sulfhydryl homeostasis, leads to an accumulation of ubiquitinated proteins and to a loss in cell viability. The ubiquitinated proteins 6 that accumulate in cells upon Cd 2+ -treatment and the mechanisms mediating its toxicity in neuronal cells remain poorly defined.
Herein we report that, in mouse neuronal HT4 cells, Cd 2+ initiated a pro-inflammatory response by inducing up-regulation of COX-2 at the mRNA and protein levels without affecting COX-1. As a result of this response, there was an increase in the extracellular concentrations of the pro-inflammatory prostaglandin PGE2, which is a COX-2 product. Moreover, treatments with the heavy metal resulted in the stabilization of glycosylated and unglycosylated forms of COX-2 as well as an accumulation of ubiquitin-COX-2 conjugates. Cd 2+ also induced the formation of HNE-protein adducts in the neuronal cells indicating that its cytotoxicity may be mediated by HNE, a highly cytotoxic aldehyde product of lipid peroxidation. Based on the observations that Cd 2+ -induced oxidative stress is closely associated with COX-2 up-regulation and neuronal cell death, we evaluated the protective effect of two anti-oxidants [N-acetylcysteine (NAC) and ascorbic acid] and a COX-2 specific inhibitor (Celecoxib) on Cd 2+cytotoxicity. Of the three drugs tested, NAC, a thiol reducing anti-oxidant, was the most effective in preventing the loss of cell viability caused by the heavy metal. Ascorbic acid, a free radical scavenger, failed to prevent and even potentiated Cd 2+ -cytotoxicity in some instances.
Celecoxib, the COX-2 specific inhibitor, attenuated the loss of cell viability caused by the heavy metal under certain conditions. These studies provide evidence that both the disruption of the ubiquitin/proteasome pathway and the induction of a pro-inflammatory response are induced by oxidative stressors such as cadmium. These two mechanisms may be associated in the neurodegenerative process. Cell Cultures-HT4 cells were derived from a mouse neuroblastoma cell line infected with a retrovirus encoding the temperature-sensitive mutant of SV40 large T antigen. When grown at 39 o C (non-permissive temperature), HT4 cells differentiate with neuronal morphology, express neuronal antigens, synthesize and secrete nerve growth factor, and express receptors for nerve growth factor (21). The cells were maintained at 33 o C in Dulbecco's modified Eagle's medium containing 5% fetal bovine serum, as previously described (19).
Cell Treatments-Cultures of HT4 cells were treated at 37 o C with aqueous solutions of cadmium sulfate added to serum-containing medium. When specified, anti-oxidants (1mM NAC or 1mM ascorbic acid) or the COX-2 inhibitor Celecoxib (20µM) or inhibitors of transcription (15µg/ml actinomycin D) or translation (25µg/ml puromycin) or an inhibitor of N-linked glycosylation (10µg/ml tunicamycin), were added to the culture media 1h prior to Cd 2+treatment. NAC, ascorbic acid and puromycin were dissolved in water and the remaining listed 8 drugs were dissolved in DMSO. At the end of the indicated incubation times, the cultures were washed twice with PBS and the cells were harvested as previously described (22). Cell washes removed unattached cells, hence, subsequent assays, were performed on adherent cells only.
Preparation of Cell Extracts for Western Blotting-Cell extracts were prepared and subjected to SDS-PAGE as previously described (19). Identification of COX-1, COX-2, ubiquitinated proteins and HNE-protein adducts was by Western blotting on 8% polyacrylamide gels. For detection of HNE-protein adducts the samples were run under non-reducing conditions in the absence of β-mercaptoethanol. The antigens were visualized by a horseradish peroxidase method utilizing the Super Signal West Pico detection system. Quantitative analysis of the immunostaining was by image analysis with the ImagePC program from NIH as described previously (23).  Immunoprecipitation -Immunoprecipitation of COX-2 from total cell extracts, normalized to protein concentration, was performed as previously described (24). The immunocomplexes were resolved on 8% SDS gels, followed by Western blot analysis probed with the anti-COX-2 antibody, proceeded by stripping and reprobing with the anti-ubiquitin conjugates antibody.
Glutathione Assay -Total GSH was quantified as described previously (26)   corresponds to COX-2 that is N-glycosylated at three sites (27). In cells treated with higher Cd 2+ concentrations, such as 3, 15, 30 and, in some cases, 45µM, two additional bands were identified as corresponding to COX-2 forms that are N-glycosylated at four sites (74kDa) and nonglycosylated (65kDa), respectively (27). The three COX-2 bands, corresponding to 74, 72 as well as 65kDa, are clearly visible in Fig. 5 (below). COX-2 levels rose above control as early as 10h after treatment with 15µM CdSO 4 , which is the Cd 2+ concentration that most effectively increases COX-2 levels to a range between 3 and 38-fold above control (Fig. 1E). In addition, a form of COX-2 (Ub-COX-2) that migrated at a high molecular weight, was clearly identified at the top of blots corresponding to cells treated for 24 and 48h with 15µM Cd 2+ (Fig.1C and D). At higher Cd 2+ concentrations (45µM) the levels of all COX-2 forms decrease. This effect is most likely due to a dramatic loss in cell viability caused by high Cd 2+ concentrations (see below).  levels were elevated to a maximum of 90ng of PGE2/ml media, a 360-fold increase over basal levels. After 48h of treatment, PGE2 production in cells treated with 30 and 45µM Cd 2+ was at least 3-fold lower than in those treated with 15µM of the heavy metal. This decline in PGE2 production most likely reflects the poor cell survival caused by higher Cd 2+ concentrations.

Cadmium Induces Accumulation of High Molecular Mass Ubiquitin-COX-2 conjugates and
Perturbs COX-2 Glycosylation -The COX-2 immunoreactive high molecular mass forms detected by Western blot analysis of total cell lysates probed with anti-COX-2 were further characterized for the presence of ubiquitin. Total lysates prepared from control and Cd 2+ -treated HT4 cells were subjected to imunoprecipitation with the anti-COX-2 antibody. Western blots of the immunoprecipitated proteins probed with the anti-COX-2 antibody (Fig. 4B) revealed three COX-2 forms (74/72 kDa doublet and 65kDa) as well as high molecular mass COX-2 (Ub-COX-2), a pattern of reactivity similar to the one observed in Fig.1. Only the high molecular mass COX-2 forms were detected when these blots were stripped and reprobed with the antibody that recognizes ubiquitin-conjugates (Fig. 4A). These findings suggest that Cd 2+ promotes stabilization of COX-2 as high molecular mass ubiquitin conjugates.
Treatment with endo H reduced the apparent molecular mass of most of the 72/74 kDa doublet to the 65 kDa band (data not shown). The latter was also the major band identified in Cd 2+ -treated cells preincubated for one hour with 10µg/ml of tunicamycin, a glycosylation inhibitor (Fig. 5B). The 65 kDa band corresponds, therefore, to unglycosylated COX-2. These results indicate that the shift in COX-2 apparent molecular masses seen in Cd 2+ -treated cells is due to changes in the extent of its glycosylation and not to its proteolytic degradation.

Evaluation of the Effect of Anti-Oxidants and a COX-2 Inhibitor on Cadmium Cytotoxicity -To
investigate which mechanisms mediate Cd 2+ -cytotoxicity we evaluated the effects of two antioxidants, NAC and ascorbic acid, and a COX-2 inhibitor, Celecoxib, on HT4 cell viability. The results obtained after treatments with 15, 30 and 45µM of the heavy metal for 7h or 24h are shown in Fig. 6A and 6B, respectively. The lowest survival rate was observed as an approximate 90% decrease in cell viability after a 24h treatment with 45µM Cd 2+ . A 7h treatment with the same Cd 2+ concentration caused a 45% reduction in cell viability.
The thiol reducing agent NAC significantly (p<0.05) attenuated the loss in cell viability caused by 7h [ Fig.6A(a)] or 24h [ Fig.6B(a)] treatments with the three Cd 2+ concentrations tested.
Moreover, Celecoxib, a COX-2 specific inhibitor, significantly (p<0.05) lessened the cytotoxic effect of 7h incubations with the same three Cd 2+ concentrations [ Fig.6A(b)]. The COX-2 inhibitor was less effective after 24h incubations with the heavy metal by only significantly (p<0.05) preventing the loss of viability of cells treated with 15µM Cd 2+ [ Fig.6B(b)]. These findings suggest that the induction of a pro-inflammatory response is one of the early mechanisms activated by Cd 2+ and plays an important role in its cytotoxicity. However, in longer exposures to the heavy metal, other cytotoxic mechanisms may be sufficiently activated to override the effect of the pro-inflammatory response alone. COX-2 activity may then contribute to the acceleration of the cytotoxic process.
Remarkably, ascorbic acid failed to prevent the loss of cell viability caused by Cd 2+treatment. In some instances, this free radical scavenger even intensified Cd 2+ -cytotoxicity [ Fig.   6B(c)]. As discussed below, this effect could be due to a synergism between the anti-oxidant and cadmium in causing lipid peroxidation (see Discussion).

Inhibitors of transcription (actinomycin D) or translation (puromycin) did not
significantly alter the loss of cell viability caused by 24h treatments with Cd 2+ [Fig. 6B(d) and (e), respectively]. These results suggest that, under these conditions, de novo protein synthesis of COX-2 is not the sole contributor to the cytotoxic process.
It should be noted that some of the drugs tested in these experiments, namely Celecoxib, actinomycin D and puromycin, were slightly cytotoxic. Their effect on Cd 2+ -mediated loss in cell viability, therefore, reflects a comparison to cells treated with the drugs alone.

Cadmium Induces the Production of HNE-protein Adducts in HT4 neuronal cells -One
of the effects of Cd 2+ -induced oxidative stress is lipid peroxidation. Since HNE is a product of lipid peroxidation, we investigated if the heavy metal causes the formation of this highly cytotoxic aldehyde. Fig. 7 shows that 15, 30 and 45µM Cd 2+ induce the formation of HNEprotein adducts. These protein conjugates, with molecular masses above 132 kDa, are indicated at the top of the Western blot shown in Fig.7. Since HNE is known to be an extremely reactive electrophile, it is possible that the cytotoxic effect of Cd 2+ is potentiated by this aldehyde.
The Cd 2+ -induced HNE-protein adducts were not detected in HT4 cells pre-treated with NAC ( Fig. 7, right lanes). The latter is a sulfur-containing anti-oxidant, which reacts directly with free radicals and may, therefore, prevent lipid peroxidation.

NAC, but not Ascorbic Acid, Prevents the Cadmium-induced Rise in COX-2 levels as well as the
Decline in GSH levels and Attenuates PGE2-production -To further test the hypothesis that oxidation of protein thiols and ROS production are events that mediate Cd 2+ toxicity we attempted to block some of the heavy metal effects with the thiol reducing agent NAC. As seen middle panel) and decline in glutathione (Fig. 8B), the latter measured after a 24h treatment with 45µM Cd 2+ . In addition, NAC attenuated the Cd 2+ -induced rise in PGE2 levels detected in HT4 cells incubated with the heavy metal for 7h (Fig. 8C). Ascorbic acid (1mM) did not block the Cd 2+ -induced rise in COX-2 levels (Fig. 8A, right panel). More importantly, the decrease in GSH and the increase in PGE2 levels caused by the heavy metal were potentiated by ascorbic acid by a factor of 5-fold and 10 to 38-fold, respectively ( Fig. 8B and D).

DISCUSSION
The aim of this study was to identify mechanisms activated by disruption of the intracellular oxidation-reduction homeostasis and accumulation of ubiquitinated proteins, which could contribute to neuronal cell death. For this purpose, we treated mouse HT4 neuronal cells with cadmium, since the heavy metal is an oxidative stressor that depletes GSH, increases the levels of protein-mixed disulfides and leads to the accumulation of ubiquitinated proteins and neuronal cell death (19).
The present investigations reveal that Cd 2+ specifically enhances COX-2 gene expression, at the mRNA and protein levels, without affecting COX-1 expression in the HT4 neuronal cells.
In addition, the heavy metal induces a rise in the production of PGE2, a pro-inflammatory prostaglandin that is a product of COX-2. This result demonstrates that the de novo synthesized COX-2, resulting from Cd 2+ induction, is enzymatically active. Moreover, the increases in COX-2 expression were detected as early as a 10h treatment with 15µM Cd 2+ suggesting that the induction of a pro-inflammatory response is an early event in the cellular reaction to the heavy metal.
Whether the accumulation of ubiquitinated proteins and/or COX-2 up-regulation are a cause or a consequence of Cd 2+ cytotoxicity in our system remains unclear. Nevertheless, the cellular events brought about by the heavy metal in HT4 neuronal cells (19 and herein) mimic those we reported for proteasome inhibitors (22). Like oxidative stress induced by cadmium, proteasome inhibition leads to the accumulation of ubiquitinated proteins, to the induction of a pro-inflammatory response and to a loss in cell viability. Therefore, the data presented herein together with our previous findings from studies with cadmium or proteasome inhibitors (19;22) suggest that disruption of the ubiquitin/proteasome pathway cooperates with COX-2 function in a neurodegenerative pathway. The fact that COX-2 plays a role in neurodegeneration is supported by several studies. For example, up-regulation of COX-2 precedes the appearance of NFT-containing neurons and neurodegeneration in patients with Fukuyama-type congenital muscular dystrophy, a neurodegenerative disorder transmitted through autosomal recessive inheritance (28). Similarly, both NFT-containing and damaged neurons in Down's syndrome and AD were found to exhibit high expression of COX-2 (29). ROS produced by COX-2 could be responsible for its role in the neurodegenerative process. It is well established that ROS are byproducts of prostaglandin biosynthesis by cyclooxygenases and these ROS are regarded as important contributors to tissue damage resulting from, for example, brain injury and ischemia (28;30).
Interestingly, ROS are currently being considered to act not only as toxic metabolites but also as signaling molecules that may regulate cellular processes such as apoptosis (31;32). ROS derived from Cd 2+ treatment may cause lipid peroxidation (33). One of the products of lipid peroxidation, HNE, is a highly reactive aldehyde recognized as one of the most important mediators of the cytopathological effects of oxidative stress (34). Our observation that Cd 2+ caused formation of HNE-protein adducts, indicates that HNE may, in fact, be one of the molecules triggering some of the intracellular responses to Cd 2+ . In support of this hypothesis, a study involving a screen of 15 different oxidized fatty acid metabolite products of lipid peroxidation, revealed that HNE was the only specific inducer of COX-2 in rat liver epithelial RL34 cells (35). Moreover, HNE-modified proteins were shown to become ubiquitinated (36) and act as non-competitive inhibitors of the proteasome (37). Other studies demonstrated that HNE can directly modify the proteasome and inhibit its trypsin-like and caspase-like activities (36). Consequently, under the conditions tested in our studies, HNE may be one of the molecules contributing to COX-2 up-regulation as well as to the accumulation of ubiquitinated proteins.
To confirm that ROS production mediates some of the Cd 2+ effects we treated HT4 cells with two anti-oxidants, namely NAC and ascorbic acid. NAC is a thiol-reducing agent known to increase GSH pools and to react with ROS in cells (38;39). The present studies demonstrate that NAC attenuates the Cd 2+ effects on cytotoxicity, formation of HNE-protein adducts, COX-2 induction and PGE2 production as well as the decline in GSH, confirming that oxidative stress is an important mechanism mediating cadmium toxicity. The demonstration that NAC decreases the HT4 neuronal cell loss in viability induced by 45µM Cd 2+ underscores the capability of the thiol reducing agent to protect cells from oxidative stress. Due to their anti-oxidant properties, thiol reducing agents such as NAC are emerging as potential therapeutic drugs to manage neurodegenerative disorders such as PD (38;39).
Ascorbic acid is an anti-oxidant and free radical scavenger effective against peroxyl-and hydroxyl-radicals, superoxide, singlet oxygen and peroxynitrite [reviewed in (40)]. In our studies, however, ascorbic acid failed to prevent Cd 2+ cytotoxicity. It acted instead as a prooxidant in the presence of the heavy metal potentiating its decrease in GSH. This free radical scavenger also increased the levels of cytotoxicity and PGE2 production induced by Cd 2+ . A recent report demonstrated that ascorbic acid in the presence of transition metals stimulates the decomposition of products of lipid peroxidation to fatty acid metabolites such as HNE (41). Cd 2+ is a transition metal and a synergism between ascorbic acid and the heavy metal may, therefore, be responsible for enhancing its cytotoxicity.
We also investigated if COX-2 inhibition would prevent the loss of cell viability induced by Cd 2+ . Our studies revealed that Celecoxib, a COX-2 specific inhibitor, increased the survival by guest on March 24, 2020 http://www.jbc.org/ Downloaded from rate of HT4 cells treated with 15, 30 and 45 µM Cd 2+ for 7h and 15µM Cd 2+ for 24h. This effect may be due to Celecoxib blocking a cyclooxygenase-mediated toxic pathway that is independent of eicosanoid biosynthesis. In this pathway lipid hydroperoxides, which are the precursors of fatty acid metabolites such as HNE, are produced enzymatically by cyclooxygenases (41). COX-2 was found to be more efficient in producing this enzymatic reaction than COX-1 (41). Our studies demonstrated that Celecoxib, however, failed to significantly promote cell survival of HT4 cells treated for 24h with 30 and 45µM Cd 2+ . The Celecoxib effect on Cd 2+ cytotoxicity, therefore, becomes concentration-dependent after longer treatments with the heavy metal. These results suggest that COX-2 activation plays an important role in the early phase of the cytotoxic process initiated by Cd 2+ . COX-2 inhibition, therefore, could be an important target for therapeutic intervention in the initial stages of neurodegeneration associated with oxidative stress. Accordingly, other studies demonstrated that treatment of AD patients with non-steroidal anti-inflammatory drugs (NSAIDs), which inhibit cyclooxygenases, slows the progression of the disease proportionally to the duration of the treatment (42).
Our studies indicate that Cd 2+ may prevent COX-2 turnover, since treatment with the heavy metal promotes the accumulation of ubiquitin-COX-2 conjugates. These findings are consistent with our previous demonstration that ubiquitin-COX-2 conjugates appear in HT4 cells treated with proteasome inhibitors (22), suggesting that COX-2 may be turned over by the ubiquitin/proteasome pathway. In addition, Cd 2+ most likely perturbs COX-2 glycosylation.
Accordingly, an unglycosylated 65 kDa form of COX-2 was consistently detected in Cd 2+treated but not in control cells. The heavy metal also increases the protein levels of the 72 and 74 kDa COX-2 forms, which are N-glycosylated at three and four sites, respectively (27). N-Glycosylation of cyclooxygensases at three sites is required for them to achieve a native (27).
In summary, these studies demonstrate that disruption of the intracellular sulfhydryl homeostasis and formation of HNE-protein adducts mediate Cd 2+ cytotoxicity. The heavy metal induces the accumulation of ubiquitinated proteins, cell death (19) and a pro-inflammatory response manifested by up-regulation of COX-2 (shown in this report). It is possible that neurodegenerative factors, such as genetic make-up, age or environmental toxins that mimic the cadmium effects, contribute to the development of initial cellular lesions containing ubiquitinated proteins. These neuronal inclusions, which are hallmarks of neurodegeneration, may, themselves, lead to neuronal cell death. The neurodegenerative process, however, could be exacerbated by stimulation of a pro-inflammatory response that would contribute to a more rapid decline in neuronal survival. Inhibition of COX-2 activation in the initial phases of the neurodegenerative process may, therefore, be an important target for preventive therapy.