Introduction
Parkinson’s disease (PD) is a chronic neurodegenerative disorder characterized by degeneration of dopaminergic neurons projecting from the substantia nigra pars compacta (SNc) to the striatum, and the presence of proteinaceous inclusions called Lewy bodies, which are composed predominantly of fibrillar α-synuclein and ubiquitinated proteins (Braak et al.
2004; Ehringer and Hornykiewicz
1960). As a consequence of the loss of striatal dopamine (DA), a progressive impairment of the control of movements occurs, inducing akinesia, rigidity, and resting tremor (Dauer and Przedborski
2003). Although the etiology of PD remains unknown, it is believed to involve numerous risk factors, both genetic and environmental.
The ubiquitin-proteasome system (UPS) is the principal mechanism responsible for the degradation of damaged and misfolded intracellular proteins, and its failure leads to protein accumulation and cell death (Ciechanover and Brundin
2003). A number of studies have suggested that a failure of the UPS may be an important pathogenic factor in PD. In fact, it has been found that mutations of the components related to the UPS, e.g., parkin and ubiquitin carboxy-terminal hydrolase L1 (UCH-L1), lead to degeneration of the nigrostriatal pathway in certain forms of familial PD (Kitada et al.
1998; Leroy et al.
1998). Moreover, an impaired proteasomal function has been described in the SN of idiopathic PD (McNaught and Jenner
2001; McNaught et al.
2003). Besides UPS impairment, cell death in PD has also been linked to neuroinflammatory processes and excessive oxidative stress (Jenner
2003; Qian et al.
2010). In dopaminergic neurons, the oxidative deamination of DA by monoamine oxidase (MAO) and the auto-oxidation of DA results in the production of hydrogen peroxide which in turn can be converted to hydroxyl radicals to react with and cause damage to cellular molecules (Hermida-Ameijeiras et al.
2004).
The involvement of oxidative stress in PD is supported by a postmortem PD brain analysis which involved the evaluation of several parameters, such as protein and DNA oxidation, lipid peroxidation, and decreases in reduced glutathione level (Jenner
1998). The existence of ongoing inflammatory processes that may contribute to the progression of PD is supported, e.g., by the presence of activated microglia, the accumulation of cytokines and nuclear factor kappa B (NF-κB) pathway activation in the cerebrospinal fluid and the brain of PD patients (Hirsch and Hunot
2009; McGeer et al.
1988).
It has been proven that many neuropathological and behavioral features of PD can be replicated in animal models of PD, evoked by the use of UPS inhibitors. For instance, systemic and intracerebral administration of several UPS inhibitors, including lactacystin, epoxomicin, and PSI, induces degeneration of dopaminergic neurons with intracellular inclusions in the SNc as well as behavioral abnormalities in rodents (Fornai et al.
2003; Lorenc-Koci et al.
2011; Mackey et al.
2013; McNaught et al.
2002a,
2004; Vernon et al.
2011; Xie et al.
2010b). The toxicity of UPS inhibitors has also been reported in various cell cultures in vitro (Jantas et al.
2013; McNaught et al.
2002b; Reaney et al.
2006; Rideout et al.
2001). The inhibition of the UPS has been linked with the occurrence of neuroinflammatory processes and oxidative stress. For instance, systemic and intranigral administration of different UPS inhibitors provoke microglial activation in the SN along with the death of dopaminergic neurons (Ahn and Jeon
2006). This is consistent with in vitro data showing microglial activation in cells treated with lactacystin (Kwon et al.
2008). Furthermore, various parameters of oxidative stress have been examined in cell lines treated with UPS inhibitors (Lee et al.
2001). Therefore, it seems that compounds showing antioxidant and anti-inflammatory activity may protect dopaminergic neurons from the UPS failure-induced degeneration.
Celastrol, also called tripterine (3-hydroxy-24-nor-2-oxo-1(10),3,5,7-friedelatetraen-29-oic acid), a quinone methide triterpene, is a pharmacologically active compound extracted from a Chinese herb Tripterygium Wilfordii Hook F (Zhou
1991). Celastrol has strong antioxidant and anti-inflammatory activity and has been found to be effective in a number of animal models of inflammatory (Kiaei et al.
2005; Kim et al.
2009; Li et al.
2008a) and neurodegenerative diseases, e.g., Alzheimer’s and Huntington’s (HD) diseases (Allison et al.
2001; Cleren et al.
2005). Studies have shown that celastrol suppresses microglial activation, pro-inflammatory cytokine production, inducible nitric oxide formation, and lipid peroxidation (Allison et al.
2001; Sassa et al.
1990). Furthermore, celastrol has been reported to possess a potent antitumor activity, both in vitro and in vivo, which is mediated by multiple mechanisms including inhibition of the UPS and induction of apoptosis (Kannaiyan et al.
2011; Yang et al.
2006,
2010).
Recently, it has also been demonstrated that celastrol is able to prevent degeneration of nigrostriatal neurons in the MPTP-induced neurotoxicity in mice and in a genetic
Drosophila DJ-1A model of PD (Cleren et al.
2005; Faust et al.
2009). In view of the potential antiparkinsonian-like effects of this compound, we decided to test its potency in another PD model, i.e., the lactacystin-induced inhibition of the UPS, which may operate through different pathogenic mechanisms from the above-mentioned models. Therefore, the aim of our study was to determine whether celastrol may exert a neuroprotective effect both in vitro, in the lactacystin-induced toxicity in mouse primary cortical neurons and human neuroblastoma SH-SY5Y cells, and in vivo, in the rat PD model of lactacystin-induced degeneration of nigrostriatal dopaminergic system. Human neuroblastoma SH-SY5Y cell line is widely used to study the mechanism of cell death in relation to PD because it possesses many characteristics of dopaminergic neurons (Påhlman et al.
1990; Xie et al.
2010a). On the other hand, mouse primary cortical neurons exhibit typical neuronal phenotype (Lesuisse and Martin
2002) and we used them to examine the effects of treatment on two types of cells with different features.
Discussion
The main finding of the present study is that celastrol does not exert a neuroprotective effect under conditions of UPS inhibition. Furthermore, at higher doses, this compound accelerates toxicity triggered by lactacystin and induces cell death, when given alone in both in vitro and in vivo studies.
Toxicity of lactacystin has been confirmed in many studies using in vitro models (Jantas et al.
2013; McNaught et al.
2002b; Reaney et al.
2006; Rideout et al.
2001). It is generally accepted that the UPS inhibition is the main mechanism responsible for that effect (Fenteany et al.
1995). It was proven that lactacystin at a concentration of 10 μM, hence lower than those used in our experiments, induced a pronounced inhibition (75–90 %) of the chymotrypsin, trypsin-like, and post-glutamyl peptidase activity in PC12 cells (Fornai et al.
2003). Previous studies revealed that suppression of the UPS function by lactacystin led to the accumulation and aggregation of misfolded proteins, and resulted finally in induction of neuronal apoptosis via the release of cytochrome c from mitochondria and the activation of caspase-3-like proteases (Li et al.
2008b; Qiu et al.
2000).
In the present study, we demonstrated, using two different cell cultures (mouse primary cortical neurons and human neuroblastoma RA-SH-SY5Y cells) and a wide range of celastrol concentrations (from 0.01 to 10 μM) that celastrol, given concomitantly with lactacystin, did not protect cells from its toxic effect. Furthermore, we did not find any beneficial effect of celastrol, when that compound was given at various time points (from 1 to 18 h) before lactacystin exposure. The possible explanation for the toxic effect of celastrol in the present study is that this compound, besides its anti-inflammatory and antioxidant properties, is also a potent proteasome inhibitor with preferential inhibition of proteasomal chymotrypsin-like activity (Yang et al.
2006). At a concentration of 2.5 μM, celastrol inhibited chymotrypsin-like activity by
ca. 40–60 % in purified rabbit 20S proteasome, human cultured prostate tumor cells and Xenopus laevis A6 kidney epithelial cells (Walcott and Heikkila
2010; Yang et al.
2006). At the same concentration, it induced accumulation of ubiquitinated proteins in cells and changes in cell phenotype, accompanied with cytoskeletal disorganization and apoptosis (Walcott and Heikkila
2010; Yang et al.
2006). Hence, it may be speculated that inhibition of the UPS by celastrol may be one of the mechanisms responsible for its toxic effects, especially under conditions of UPS impairment by lactacystin.
Although we found no neuroprotective potency of celastrol, neither given as pretreatment, nor administered jointly with lactacystin, it should be mentioned that effect of celastrol may be biphasic, being protective at low doses and toxic at higher ones. For instance, in PC12 cells, its best protective effect against polyglutamine toxicity or the tert-butyl hydroperoxide (t-BHP)-induced oxidative stress was obtained at 0.1 μM (Sun et al.
2010; Zhang and Sarge
2007), and a toxic effect at 1.6 μM (Sun et al.
2010). In HeLa cells expressing a mutant polyglutamine protein, a significant decrease in cell death was achieved only at concentrations between 0.4 and 1.6 μM (Zhang and Sarge
2007). The neuroprotective effect of low concentrations of celastrol (0.001 and 0.01 μM) was also found in our recent study, but only in cells incubated for longer time in a low-serum medium (Jantas et al.
2013). Therefore, it seems that celastrol has a narrow therapeutic window, and its protective or cytotoxic effect not only depends mainly on its concentration, but also on other factors, e.g., the cell type and culture conditions. However, in the present study, we did not find any protective effects of celastrol against lactacystin-induced toxicity, even when we used the former compound at the low, non-toxic concentrations (0.01 and 0.1 μM). Thus, it appears that under conditions of the strong UPS inhibition by lactacystin, celastrol is not able to protect cells.
Similarly to the above-mentioned neuroprotective effects of celastrol, there are some data demonstrating neuroprotective effect of lactacystin against toxicity induced by low doses of 6-OHDA (Inden et al.
2005; Yamamoto et al.
2007) or by glutamate (Maher
2008; van Leyen et al.
2005). On the other hand, treatment with other UPS inhibitors enhanced 6-OHDA-induced toxicity in different cell cultures (Elkon et al.
2001; Höglinger et al.
2003). These contradictory results may be explained by the fact that 6-OHDA at low doses increases protein degradation and the UPS activity, presumably in response to oxidative stress (Elkon et al.
2001,
2004). In contrast, higher doses cause their marked decline (Elkon et al.
2004). Therefore, it may be speculated that neuroprotection induced by UPS inhibitors is facilitated in those models in which the primary detrimental factor used to destroy neurons is not directly related to UPS inhibition.
On the basis of the above-mentioned studies, it appears that the final effect of the UPS inhibition (cell survival or death) depends strongly on the concentration of the inhibitor used and the duration of its effects (and consequently on the level of UPS inhibition over time). This assumption is consistent with the results of a microarray study (Yew et al.
2005) performed on mouse primary cortical neurons treated with lactacystin that showed different effects at varying post-treatment time points: an up-regulation of genes involved in the neuroprotective response to UPS inhibition at an early time point (e.g., heat-shock protein (HSP) 70, HSP22, genes of the UPS, and cell cycle inhibitors), followed by a proapoptotic response (genes involved in apoptosis, oxidative stress, and inflammatory responses). It has been shown that HSP70 plays a protective role by, for instance, preventing protein misfolding and aggregation (Yenari
2002), and its up-regulation appears to be specific for the cell death mediated by UPS inhibitors (Yew et al.
2005). In fact, celastrol has been found to enhance various HSP levels in vitro; however, the concentrations that enhance HSPs are only slightly lower than the concentrations that are toxic for dopaminergic cells (Chow and Brown
2007). Moreover, in t-BHP-induced cytotoxicity celastrol increases HSP70 expression at concentrations higher than those that are protective in this model (Sun et al.
2010). Therefore, it may be suggested that the increase in HSP70 expression is not a truly neuroprotective property of celastrol, but is rather a result of the defensive reaction of cells against its toxic effect. The above assumption is supported by the fact that the increase in HSP70 content was also observed in surviving dopaminergic neurons of rats treated with a low dose of lactacystin (Pastukhov et al.
2013).
The question arises why primary cortical neurons were more vulnerable to lactacystin toxicity compared to RA-SH-SY5Y cells (2.5 vs. 10 μg/ml, respectively) in the present study. These results are not consistent with the previously described greater sensitivity of various cell lines of tumoral origin to the UPS inhibition compared to normal cells (Adams et al.
1999; Almond and Cohen
2002). These discrepancies may be the result of RA-induced differentiation of SH-SY5Y cells which makes them more resistant to lactacystin toxicity. In fact, a very recent study on SH-SY5Y cells showed a protective effect of RA against the toxic effect of another UPS inhibitor epoxomicin (Cheng et al.
2013). In line with this assumption, we recently found that lactacystin at a dose as low as 0.25 μg/ml decreased cell viability in undifferentiated, but not RA-differentiated, SH-SY5Y cells (Jantas et al.
2013). Likewise, the toxic concentration of celastrol in undifferentiated SH-SY5Y cells was ten times lower than in RA-SH-SY5Y ones (Jantas et al.
2013). Other evidence for the increased resistance of RA-SH-SY5Y cells comes from the present study, namely after treatment of both types of cells with lactacystin and celastrol the increase in cell death was obtained in primary cortical neurons after subtoxic concentration of celastrol, whereas in RA-SH-SY5Y cells a toxic concentration of this compound was required to evoked the same effect.
The lack of the protective effect of celastrol against the lactacystin-induced toxicity in vitro was supported in the present study by utilizing the rat PD model in which degeneration of dopaminergic neurons was caused by unilateral administration of lactacystin directly into the SNc (Lorenc-Koci et al.
2011; Niu et al.
2009; Mackey et al.
2013; McNaught et al.
2002a). The advantage of the lactacystin model over other conventional animal models of PD is related to the fact that it replicates cardinal pathological features of PD, i.e., the nigral degeneration and aberrant protein degradation. We showed that lactacystin (5 μg/2 μl) induced a strong decrease (83 %) in DA level in the lesioned striatum 1 week after surgery, compared to the intact side. Parallel to the change in DA content, a significant decline in the levels of the DA metabolites DOPAC (intraneuronal metabolite), 3-MT (extraneuronal metabolite), and HVA (total metabolite) in the lesioned striatum was observed. Moreover, lactacystin-evoked acceleration of MAO-dependent N-oxidation (DOPAC/DA), COMT-dependent
O-methylation of DA (3-MT/DA), and the total DA catabolism (HVA/DA) which means that both extraneuronal and intraneuronal DA catabolism were increased after lactacystin administration. The enhanced DA catabolism, especially oxidative catabolism catalyzed by MAO, may at least partly contribute to neuronal death by inducing oxidative stress (Jenner
2003). In fact, there is evidence showing a close mutual relationship between inhibition of the UPS function and oxidative stress (Davies
2001; Lee et al.
2001).
In contrast to the lactacystin group, animals treated with any of celastrol doses showed no decrease in the striatal DA level. There were, however, some differences in the levels of DA metabolites between groups treated with two lower (0.3 and 1 mg/kg) and the highest (3 mg/kg) dose of celastrol. In the former two groups, only a small but significant decrease in the level of DOPAC in the ipsilateral striatum was revealed, whereas in the latter group, DOPAC and HVA levels were elevated, but 3-MT level was decreased in both sides of the striatum. Thus, celastrol seems to influence in the opposite way the intraneuronal DA catabolism, diminishing or increasing it, depending on the dose. This assumption is consistent with the direction of changes in DA catabolic ratios of DOPAC/DA which were decreased after lower doses (the effect more clearly visible on the contralateral side), but tended to rise after the highest dose. Such an increase in the oxidative pathway of DA catabolism may sensitize dopaminergic neurons and make them more vulnerable to lactacystin toxicity. In fact, only the treatment with the highest dose of celastrol accelerated the lactacystin-induced decrease in the striatal level of DA (from 83 to 97 %) and its metabolites compared to the intact side of the striatum. Furthermore, the highest dose of celastrol enhanced the lactacystin-induced acceleration of the MAO-dependent oxidative catabolism and the total catabolism of DA, but it was not accompanied by a simultaneous increase of extraneuronal DA catabolism (3-MT/DA). This implies that exacerbation of DA decline may result from the shifting of DA metabolism from
O-methylation towards the N-oxidation pathway. On the other hand, in groups treated with lactacystin and the lower celastrol doses, 3-MT/DA ratio substantially increased on the ipsilateral side compared to the lactacystin group. Such enhanced catabolism of DA through COMT-dependent
O-methylation is supposed to constitute an antioxidant defense mechanism against oxidative stress (Miller et al.
1996). However, since this increase was not accompanied by a substantial increase in DA level, the effect of celastrol seems to be insufficient to overcome the toxicity of lactacystin.
Our biochemical findings were confirmed in the present study by the immunohistochemical data. We demonstrated that lactacystin induced a strong reduction in the number and the density of TH-ir neurons in the lesioned SN one week after surgery. None of the two celastrol doses (1 and 3 mg/kg) prevented the lactacystin-induced loss of nigral TH-ir neurons. Furthermore, the higher celastrol dose was toxic by itself, having reduced the number and density of cells by
ca. 30 % on both sides of the SN. It is worth noting that the decrease in the number/density of TH-ir neurons on the intact side of the SN in rats treated with a 3 mg/kg of celastrol was greater in the lactacystin group (i.e., the group treated with lactacystin on the opposite side) compared to the rats administered with celastrol alone. The above finding suggests that lactacystin may to some extent spread to the contralateral side of the SN. This assumption is consistent with a most recent study showing a bilateral loss of TH-ir neurons in the SNc and ventral tegmental area, especially after high doses (10 and 20 μg) of unilaterally administrated lactacystin (Mackey et al.
2013). Although the dose of lactacystin we used (5 μg) was too low to damage neurons on the contralateral side of the SN, it could sensitize neurons to celastrol toxicity.
The other finding of our in vivo study was that celastrol, besides induction of degeneration of dopaminergic neurons, may exert systemic toxicity by producing weight loss and mortality in rats. These findings are consistent with studies of Zhu et al. (
1996) who revealed toxic effects of celastrol on digestive, urogenital and blood circulatory system in dogs. In our study, only the highest dose of 3 mg/kg of celastrol produced adverse effects and 27 % of rats from this group died before the end of the experiment. Interestingly, a similar dose, given once or twice daily, was found to be effective against MPTP- and 3-nitropropionic acid-induced toxicity in mice and rats, respectively, as well as against the tumor in mice, without inducing toxic side-effects (Cleren et al.
2005; Yang et al.
2006). Moreover, the i.p. LD50 of celastrol for Wistar rats was determined to be much higher (20.5 mg/kg) (Li et al.
2013). On the other hand, one study demonstrated a high mortality (40 %) in mice during celastrol (4 mg/kg) treatment (Raja et al.
2011). Therefore, further studies are required to determinate the safety and tolerability of that compound.
Although in the present study, no beneficial effects of celastrol against lactacystin toxicity were found, it should be noted that some researchers showed neuroprotective effects of that compound in other animal models of PD. For instance, Cleren et al. (
2005) demonstrated that celastrol significantly diminished the MPTP-induced loss of dopaminergic neurons in the mouse SN and attenuated striatal DA and DOPAC depletion, probably by attenuation of the MPTP-induced increases in TNF-α and NFκB immunoreactivity, or induction of HSP70. In another study, utilizing a genetic
Drosophila DJ-IA model, celastrol increased brain DA level and the number of TH-ir neurons in the dorsomedial cluster (Faust et al.
2009). The main reasons for such contrasting results obtained in the present and the above-mentioned studies are most probably the differences in the models of PD which reflect diverse aspects of the pathophysiology of PD and may involve different cellular mechanisms of neurodegeneration. In particular, lactacystin, which is an irreversible UPS inhibitor, induces long-term inhibition of the UPS, since
ca. 40–50 % decrease in chymotrypsin-like activity can be observed in mouse ventral midbrain (VM) 3–4 weeks after lactacystin administration (Li et al.
2010; Zhu et al.
2007). On the other hand, the
Drosophila DJ-IA model that refers to an early-onset form of familial PD does not seem to be directly related to UPS dysfunctions (Bonifati et al.
2003; Faust et al.
2009). It has been demonstrated that DJ-1 mutations result in mitochondrial defects and increased neuronal vulnerability to oxidative stress (Takahashi-Niki et al.
2004; Wang et al.
2012). However, DJ-1-deficient mice do not show any UPS dysfunctions, in either the striatum or VM (Yang et al.
2007). The question arises whether MPTP is able to decrease UPS activity in vivo. It has recently been found that only continuous administration of MPTP with an osmotic minipump produces a long-lasting (2 weeks) inhibition of the UPS and formation of inclusion bodies stained for ubiquitin and α-synuclein in the mouse SN. In contrast, conventional MPTP injection induces only a brief UPS inhibition (<24 h) and fails to produce protein aggregation (Fornai et al.
2005). Interestingly, under conditions of conventional MPTP administration, the reversible UPS inhibitor PSI either does not enhance neurodegeneration (Kadoguchi et al.
2008), or even protects dopaminergic neurons against the MPTP-induced toxicity in mice (Oshikawa et al.
2009), thus confirming the earlier observation from in vitro models that the severity and duration of the UPS inhibition may determinate the final, beneficial or adverse effect of the UPS inhibitors.
In conclusion, the presented in vitro and in vivo studies demonstrate, for the first time to our knowledge, that celastrol, a compound which has been shown to be neuroprotective in some models of PD, is no longer protective under conditions of UPS inhibition. Since UPS failure seems to contribute to the pathogenesis of PD, the compounds with a profile similar to celastrol, which, besides anti-inflammatory and antioxidant properties, are also potent proteasome inhibitors, may accelerate the progression of the disease. Therefore, any putative possible pathogenic factors, including those related to the UPS inhibition, must be taken into account in a search for potentially neuroprotective drugs useful in PD treatment strategies.