Address correspondence and reprint requests to Cai Song, Department of Psychology, Life Science Centre, Dalhousie University, 1355 Oxford Street, Halifax, Canada B3H 4J1 and Chinese Academy Engineer Institute for the Development of Endangered Medicinal Resource in Southwest of China, Guangxi Botanic Garden of Medicinal Plants, 189 Changgang Road, 530023 Nanning, China. E-mail: firstname.lastname@example.org
Eicosapentaenoic acid (EPA), a neuroactive omega-3 fatty acid, has been demonstrated to exert neuroprotective effects in experimental models of Parkinson's disease (PD), but the cellular mechanisms of protection are unknown. Here, we studied the effects of EPA in fully differentiated human SH-SY5Y cells and primary mesencephalic neurons treated with MPP+. In both in-vitro models of PD, EPA attenuated an MPP+-induced reduction in cell viability. EPA also prevented the presence of electron-dense cytoplasmic inclusions in SH-SY5Y cells. Then, possible mechanisms of the neuroprotection were studied. In primary neurons, EPA attenuated an MPP+-induced increase in Tyrosine-related kinase B (TrkB) receptors. In SH-SY5Y cells, EPA down-regulated reactive oxygen species and nitric oxide. This antioxidant effect of EPA may have been mediated by its inhibition of neuronal NADPH oxidase and cyclo-oxygenase-2 (COX-2), as MPP+ increased the expression of these enzymes. Furthermore, EPA prevented an increase in cytosolic phospholipase A2 (cPLA2), an enzyme linked with COX-2 in the potentially pro-inflammatory arachidonic acid cascade. Lastly, EPA attenuated an increase in the bax:bcl-2 ratio, and cytochrome c release. However, EPA did not prevent mitochondrial enlargement or a decrease in mitochondrial membrane potential. This study demonstrated cellular mechanisms by which EPA provided neuroprotective effects in experimental PD.
Parkinson's disease (PD) is characterized by a progressive loss of nigrostriatal dopaminergic neurons. The cause of PD is unknown, but environmental contaminants may be a trigger for disease onset. Among these contaminants, neurotoxin MPTP, or its active metabolite MPP+, has been most extensively used to study PD in vivo or in vitro (Dauer and Przedborski 2003).
PD research with the MPTP model has led to a number of hypotheses regarding the pathogenesis of PD. Perhaps most investigated is the oxidative stress hypothesis, which postulates that reactive oxygen species (ROS) and reactive nitrogen species (RNS) cause oxidative damage to cellular macromolecules (Dauer and Przedborski 2003; Drechsel and Patel 2008). MPP+ inhibits complex I of the mitochondrial electron transport chain (ETC.), leading to leakage of electrons that reduce molecular oxygen to superoxide (Dauer and Przedborski 2003; Drechsel and Patel 2008; Schapira 2010). Furthermore, pro-oxidant enzymes NADPH oxidase and cyclo-oxygenase (COX)-2 play a role in glial and neuronal ROS production, whereas nitric oxide synthases (NOS) catalyze the production of nitric oxide (NO). The role of these enzymes in MPTP neurotoxicity has been demonstrated in mice lacking them that have increased resistance to the neurotoxin (Anantharam et al. 2007; Hoang et al. 2009; Wang et al. 2005; Teismann et al. 2003; Wu et al. 2003; Dehmer et al. 2000). A depletion of major antioxidant GSH in the substantia nigra pars compacta (SNpc) and altered function of antioxidant enzymes further aggravates the oxidative damage (Marttila et al. 1988; de la Torre et al. 1996; Dauer and Przedborski 2003; Zeevalk et al. 2007; Thomas et al. 2008). Other reported mechanisms of MPTP neurotoxicity include activation of the COX-2- cytosolic phospholipase A2 (cPLA2) pathway, leading to increased metabolism of omega-(n)-6 polyunsaturated fatty acid (PUFA), arachidonic acid (AA), to potentially pro-inflammatory eicosanoids (Teismann et al. 2003). Decreased neurotrophic support in dopaminergic neurons has been reported as well (Baydyuk et al. 2010; Ding et al. 2011). The primary mode of cell death in PD is apoptosis (Dauer and Przedborski 2003), possibly induced by oxidative stress and inflammation. Up-regulation of pro-apoptotic protein bax in the SNpc of PD patients has been reported (Dauer and Przedborski 2003) and bax ablation prevents MPTP neurotoxicity in mice (Vila et al. 2001; Dauer and Przedborski 2003). Bcl-2, on the other hand, is antiapoptotic by inhibiting bax, and over-expression of bcl-2 protein protects mice from MPTP or 6-hydroxydopamine neurotoxicity (Offen et al. 1998).
From above evidence, it is clear that a therapeutic compound for PD should have multifactorial protection. Unfortunately, the standard PD treatment, while effective at symptom reduction, does not prevent or retard neuropathology, but causes side effects (Rezak 2007). Recent evidence suggests that natural omega-(n)-3 PUFAs may have neuroprotective effects (Dyall and Michael-Titus 2008). N-3 PUFA serve a wide range of functions, including signal transduction, neurochemistry, and gene expression (Youdim et al. 2000; Kitajka et al. 2004; Dyall and Michael-Titus 2008; Kawashima et al. 2010). N-3 PUFA can modulate the transcription of genes involved in inflammation, oxidative stress, and neurotrophic support, including inducible NOS (iNOS), COX-2, NADPH oxidase, brain-derived neurotrophic factor (BDNF), and antioxidant enzymes (Kitajka et al. 2004; Bernardo and Minghetti 2006; Bordet et al. 2006; Rao et al. 2007a, b; Dyall and Michael-Titus 2008). However, the neuroprotective effects of n-3 PUFA in experimental PD have only begun to be investigated (Bousquet et al. 2008, 2009, 2011a, b; Delattre et al. 2010; Meng et al. 2010; Luchtman et al. 2012). Most studies have focused on docosahexaenoic acid (DHA, C22:6, n-3), a major n-3 PUFA in neuronal membranes. There are currently no clinical studies investigating the protective effects of n-3 PUFA in PD.
We recently demonstrated that eicosapentaenoic acid (EPA, C20:5, n-3) also has some protective effects in experimental PD (Meng et al. 2010; Luchtman et al. 2012), but by unknown cellular mechanisms. While the precursor to DHA, EPA is thought to be a neuroactive lipid (Peet et al. 2010) and its low abundance in the brain may be because of its extensive utilization for cellular functions, including mitochondrial oxidation (Chen et al. 2009), signal transduction (Chen et al. 2009; Kawashima et al. 2010), and anti-inflammatory eicosanoid production (Youdim et al. 2000; Dyall and Michael-Titus 2008; Kawashima et al. 2010). Neurotrophic action by EPA has been observed as well (Taepavarapruk and Song 2010; Song et al. 2009; Kou et al. 2008). Here, we induced experimental PD by MPP+ treatment of fully differentiated human SH-SY5Y cells and tested that EPA can prevent MPP+-induced neurodegeneration by preventing the basic MPP+-induced cellular pathology, including mitochondrial dysfunction, oxidative stress, inflammation, and apoptosis. We also used primary rat mesencephalic neurons to confirm that EPA can protect against MPP+ in cells more specifically relevant to PD, and whether in these cells, EPA can modulate expression of BDNF and its receptor TrkB.
SH-SY5Y culture and differentiation
The human neuroblastoma cell line SH-SY5Y was originally derived from child neuroblastoma and frequently used as an in vitro model of dopaminergic neurons, especially when differentiated (Presgraves et al. 2004; Luchtman and Song 2010; Xie et al. 2010). The cell line was obtained from frozen stocks at the National Institute for Nutrisciences and Health (NRC-INH). Cells were cultured at 37°C in vented 75-cm2 flasks containing DME/F12 culture medium (VWR International, Mississauga, Ontario, Canada) supplemented with the following ingredients (VWR): 10% fetal bovine serum (FBS), 1% penicillin–streptomycin, 1% MEM non-essential amino acid solution, 1 mM sodium pyruvate, and 1.5 g/L sodium bicarbonate. For all experimental procedures, the cells were seeded in 6- or 96-well plates at a concentration of 0.5 × 106 cells/mL. Differentiation procedures were followed as described (Presgraves et al. 2004). All-trans-retinoic acid (RA) (Sigma Aldrich, Oakville, Ontario, Canada) was added the day after plating at a final concentration of 10 μM in DME/F12 with 10% FBS. After 4 days in the presence of RA (media changed every 2 days), cells were washed one time with fresh DME/F12 and then exposed to DME/F12 containing 80 nM Phorbol 12-myristate 13-acetate (PMA/TPA) (Sigma Aldrich) for 3 days. Then, media was changed to FBS-enriched media without RA or TPA and experimentation with MPP+ and EPA initiated. The passage of cells was always kept between p20 and 30 for experimental treatment.
Mesencephalic tissue containing substantia nigra was microdissected from Wistar rat embryonic E19 brains and maintained on ice-cold Hanks' balanced salt solution (HBSS) (PH7.3) according to a method described by Fath et al. (2009). Primary neuron culture methods were adapted from Mao and Wang (2003). In brief, tissue was cut into small (1 mm square) pieces and then incubated in phosphate-buffered saline (PBS) containing 0.1% trypsin (Gibco, Life Technologies, Invitrogen, Canada) for 20 min at 37°C for digestion. After centrifugation for 2 min at 180 g, tissue was triturated through a pipette in PBS containing 1 mg/mL bovine serum albumin (Sigma-Aldrich), 10 μg/mL DNase I (Sigma-Aldrich), and 0.5 mg/mL soybean trypsin inhibitor (Sigma-Aldrich). Then, cells were resuspended in DMEM/F12 medium containing 10% fetal bovine serum (Gibco), 1× B27 (Gibco), 10 g/L glucose (Sigma), 10 mg/L gentamicin (Gibco), 100 units/mL penicillin (Gibco), and 100 μg/mL streptomycin (Gibco). After counting on a hemocytometer using Trypan Blue, cells were diluted to a final concentration of 3 × 105 cells/mL and seeded in 0.01% poly d-lysine-coated 24-well or 96-well plates. After 24 h culture at 37°C, media was changed to another media containing 70% DMEM/F12, 30% neurobasal medium (Gibco) containing 0.5 mM l-glutamic acid (Gibco) and 1× B27. On the 4th day, 5 μM cytosine arabinoside (Ara-c, Sigma-Aldrich) was added to the media to inhibit the growth of glial cells. Media was changed every 5–7 days and Ara-c was maintained at the same concentration. Approximately 17–18 days were needed to culture cells before use. During experimentation with MPP+ and EPA, the same media containing Ara-c was used throughout.
Following the final day of culture, culture media with reagents (MPP+ and/or EPA-sodium salt; Sigma Aldrich) were added to the cells. EPA in sodium-salt form facilitates dissolution in media and was also previously determined to be incorporated in SH-SY5Y cells and exert functional effects (Langelier et al. 2005). A series of concentration and incubation time–response analyses of MPP+ and EPA were conducted in both SH-SY5Y cells and primary cells and cell viability measured. Thus, we tested the effect of MPP+ 0.1, 1, 10, 100, and 1000 μM at 24 and 48 h on cell viability. The effect of EPA 0.1, 1, 10, and 50 μM on cell viability was also determined. Higher doses of EPA were found to be toxic to the cells and are not physiologically relevant (Langelier et al. 2005). The concentrations and incubation times of MPP+ were selected where the neurotoxin caused a significant decrease in cell viability and coaddition of EPA could attenuate this effect. Then, experimental treatment consisted of four groups: (i) control (culture media), (ii) EPA (culture media with EPA), (iii) MPP+ (culture media with MPP+), and (iv) EPA and MPP+ (a simultaneous addition of MPP+ and EPA to the culture media). Using these treatment groups, mitochondrial function, cellular ultrastructure, oxidative stress, inflammation, and apoptosis were studied in SH-SY5Y cells. In primary neurons, the expression of BDNF and TrkB receptors were studied. See Figure S1 for an overview of experimental design.
Measurement of cell proliferation
As SH-SY5Y cells continue to proliferate following cessation of cell differentiation, albeit at a slower rate, and many of our measurements were sensitive to cell numbers, it was essential to measure cell proliferation by cell counting and normalizing data to cell numbers. Cells were seeded in six-well plates and 2 mL of cell suspension added to each well. Following experimental treatment, cells were washed in warm PBS and trypsinized by adding an equal volume of 0.05% pre-warmed trypsin (Sigma Aldrich) and PBS and incubating for 5 min at 37°C. Cells were then collected and centrifuged at 200 g for 5 min. The cell pellet was resuspended in warm PBS immediately before sampling using the Backman-Coulter Vi-cell (Backman-Coulter, Mississauga, Ontario, Canada), an automated Trypan Blue assay that takes 50 consecutive cell counts with high accuracy (Szabo et al. 2004).
Measurement of cell viability
Cells were seeded in 96-well plates and 100 (primary cells)–200 (SH-SY5Y cells) μL of cell suspension added to each well. Following experimental treatment, cell viability was measured with Thiazolyl Blue Tetrazolium Bromide (MTT) (Sigma Aldrich) according to manufacturer's instructions.
Measurement of mitochondrial membrane potential (ΔΨm)
SH-SY5Y cells were seeded in six-well plates and 2 mL of cell suspension added to each well. Following experimental treatment, JC-1 staining solution (a kit from Cayman Chemical, Ann Arbor, MI, USA), was added to each well (100x dilution of original stock) and the assay conducted according to manufacturer's instructions. JC-1 is a lipophilic cationic fluorescent dye, which selectively enters mitochondria and reversibly shifts from emitting a green (527 nm) to red (590 nm) fluorescence as ΔΨm increases. The overall condition of the cells can be assessed by calculating the proportion of JC1 aggregates to monomers (red to green).
Measurement of cellular ultrastructure with electron microscopy (EM)
The shape of mitochondria and other structural indicators of SH-SY5Y cellular pathology could be visualized with electron microscopy (EM). Following experimental treatment, cells were rinsed twice with PBS and scraped off the bottom of the wells. After collection with a centrifuge, 5% agarose was added to solidify the cells. Cells in agarose gel were then fixed by incubating in 2% glutaraldehyde in 0.1 M phosphate buffer for 2 h at 20°C. The fixative was washed off twice with PBS and post-fixed with 1% OsO4 in PBS for 1 h in refrigerator. Samples were dehydrated through ethanol gradients and embedded in Epon. Samples were polymerized in a vacuum oven for 24 h at 65–70°C.
Measurement of oxidative stress
SH-SY5Y cells were seeded in 96-well plates and 200 μL of cell suspension added to each well. Following experimental treatment, the level of intracellular ROS was quantified by 2′,7′ dichlorofluorescein diacetate (DCF-DA; Sigma Aldrich) according to manufacturer's instructions. The intracellular level of NO was quantified by 4,5-diaminofluorescein diacetate (DAF-2DA; Cayman Chemical) according to manufacturer's instructions. Because NADPHoxidase and COX-2 are major enzymes involved in the production of ROS (Wu et al. 2003; Anantharam et al. 2007; Hoang et al. 2009), we measured their mRNA and protein expressions with real-time PCR and western blot (see below).
Measurement of the antioxidant response
SH-SY5Y cells were seeded in six-well plates and 2 mL of cell suspension added to each well. Following experimental treatment, GSH concentrations were measured with a commercial kit (Sigma Aldrich) according to manufacturer's instructions. Antioxidant enzymes were measured with real-time PCR (see below).
Quantitative RT-PCR of GSHpx, SOD, catalase, p47phox, bax, bcl-2, caspase 3, cPLA2, COX-2, BDNF, TrkB, and beta actin
SH-SY5Y cells and primary mesencephalic neurons were seeded in six-well plates and 2 mL of cell suspension added to each well. Following experimental treatment, cells were lysed by adding TRI-reagent (Sigma Aldrich) to each well. The RNA extraction method was followed as recommended by the manufacturer. Complementary DNA (cDNA) was synthesized from RNA using the Quantiscript reverse transcription kit (Qiagen, Toronto, Ontario, Canada). Near 1 μg RNA was used for first-strand cDNA synthesis and any possible genomic DNA contamination in the RNA samples was removed by a DNA-removal step provided with the kit. Nucleotide sequences of the primers (Table 1) were obtained from NCBI's Nucleotide database and PRIMER3 of Biology Workbench 3.2. After specificity validation in NCBI-nucleotide-BLAST, synthesis of primers was performed by Sigma Aldrich. PCR reactions were prepared using Quantitect SYBR Green master mix (Qiagen) and carried out using a Corbett Life Science (Sydney, Australia) Rotor-Gene 6000 system. The PCR sequences consisted of an initial incubation for 15 min at 95°C to activate the HotStarTaq DNA polymerase, followed by 94°C for 15 s (denaturing), 59°C for 30 s (annealing), and 72°C for 30 s (extension). After 40 cycles, a melting curve was generated for determination of primer specificity and identity. Gene expression levels were normalized to the RNA expression of housekeeping gene beta actin (relative quantification) with the ΔΔCT correction.
Table 1. Primer names and sequences
hCOX 2 F
hCOX 2 R
hBCL 2 F
hBCL 2 R
hCaspase 3 F
hCaspase 3 R
Western blot of p47phox and cytochrome-c
SH-SY5Y cells were seeded in six-well plates and 2 mL of cell suspension added to each well. Following experimental treatment, cells were collected and spun down at 10.000 g for 10 min and lysed with a radioimmunoprecipitation assay (RIPA) buffer containing 20 mM Tris; 150 mM NaCl; 1% Nonidet P-40; 0.5% Sodium Deoxycholate; 1 mM EDTA; 0.1% sodium dodecyl sulfate (SDS), and a protease inhibitor cocktail (Sigma Aldrich) (1 mL of protease inhibitor cocktail per 20 g tissue). Lysis was aided by sonication (XL2000; Misonix Inc, Farmingdale, NY, USA) and lysates centrifuged at 10 000 g for 10 min at 4°C. Aliquots containing 20–40 μg of protein were loaded and separated on 10% polyacrylamide gels at 100 V for 70 min in electrophoresis buffer (50 mM Tris, 192 mM glycine, and 3 mM SDS). Semidry gel transfer on Polyvinylidene difluoride (PVDF) membranes (Millipore, Bellerica, MA, USA) was performed using the Trans-Blot SD Semi-Dry Electrophoretic Transfer Cell (Bio-Rad laboratories, Mississauga, Ontario, Canada) for 60 min at 15 V in transfer buffer (50 mM Tris, 192 mM glycine, 3.4 mM SDS, and 20% methanol, pH 8,2–8,4). Blots were then washed in Tris-buffered saline–Tween-20 (TBST) (50 mM Tris, 133 mM NaCl, 2,6 mM KCl, pH 7,4) for 5 min, followed by blocking (TBST and 5% non-fat milk power) for 1 h at 20°C. Primary antibodies (AB) (Santa Cruz, Santa Cruz, CA, USA) were actin rabbit polyclonal IgG 200 μg/mL, cytochrome c rabbit polyclonal IgG 200 μg/mL, and p47phox rabbit polyclonal IgG 200 μg/mL, and were diluted 1 : 200 in blocking buffer. Secondary antibody was horseradish peroxidase (HRP)-conjugated donkey anti-rabbit IgG 200 μg/0.5 mL and diluted 1 : 500–1 : 10 000 in blocking buffer. Following blocking, blots were washed with TBST for 5 min and incubated with primary AB for 2 h at 20°C or overnight at 4°C, followed by the secondary AB for 1 h at 20°C. The blots were washed in between in TBS (3 × 5 min). Immunoreactive bands were detected using Supersignal West-Pico Chemiluminescent Substracte (Pierce, Rockford, IL, USA) on a Bio-Rad molecular imager Gel Doc XR system (#170-8170; Bio-Rad laboratories). Protein expression of cytochrome c and p47phox were measured relative to beta actin protein expression using ImageJ 1.41.
Data are presented as mean ± SEM. Statistical analyses were performed using Graphpad Prism 4.03 (Graphpad Software, La Jolla, CA, USA). For the concentration–response analyses, EPA/MPP+ effects were determined with one-way anova, followed by Dunnett post hoc analyses in case of significant main effects (p < 0.05). In the experiments for testing the effects of EPA treatment against MPP+, the possible interaction between both treatment factors was measured by two-way anova with Bonferroni post hoc tests if main or interaction effects were significant (p < 0.05).
Proliferation and cell viability in SH-SY5Y cells and primary neurons
No effect of MPP+ on SH-SY5Y cell proliferation was found after 24 h of incubation, but after 48 h, MPP+ 100 μM (q = 3.826, p < 0.05, Dunnett test after one-way anova), and 1000 μM (q = 4.0406, p < 0.05 Dunnett test) significantly suppressed cell proliferation (data not shown). Similar to Langelier et al. (2005), EPA had no effect on cell proliferation, even at the highest concentration (50 μM), and even after 6 days of continuous treatment, and also did not counteract the effect of MPP+ on proliferation.
MTT assays showed that, after correction for SH-SY5Y cell numbers (proliferation), MPP+ administration for 48 h (not 24 h) significantly decreased cell viability at 100 (q = 4.5, p < 0.01) and 1000 μM (q = 5.1, p < 0.01, Dunnett test after one-way anova) when compared with untreated cells (data not shown). In contrast, EPA induced a significant dose-dependent increase in cell viability when incubated with the cells without MPP+ for 48 h and at 1 (q = 4.8, p < 0.01), 10 (q = 5.9, p < 0.001), and 50 μM (q = 10.1, p < 0.001, Dunnett test) (data not shown). Importantly, this increase in cell viability was not simply because of an increase in cell numbers, as EPA did not affect total cell numbers, as reported before at a similar concentration (Langelier et al. 2005). To test the hypothesis that EPA can counteract the MPP+-induced reduction in cell viability, both compounds were added to the culture media for 48 h. Two-way anova revealed a significant interaction between MPP+ and EPA treatments (MPP × EPA: F(df 1,20) = 4.5, p < 0.05). Post hoc Bonferroni analyses showed that EPA at 50 μM significantly reversed the effect of MPP+ up to 100 μM (p < 0.001) (Fig. 1a). Lower doses of EPA were not effective at counteracting the effects of MPP+. On the basis of these results, we selected EPA 50 μM (for 48 h) in subsequent assays with SH-SY5Y cells for mitochondrial membrane potential, oxidative stress, inflammation, and apoptosis. This concentration of EPA borders physiological values (Langelier et al. 2005).
In primary mesencephalic cells, dose responses analyses revealed that MPP+ at 10 μM for 48 h was sufficient to decrease cell viability (q = 3.9, p < 0.01, Dunnett test after one-way anova) (data not shown). Higher doses (up to 50 μM) only slightly further worsened cell viability. A dose–response analysis for EPA showed that the n-3 PUFA incubated for 48 h significantly increased cell viability at 10 μM (q = 4.6, p < 0.001) (data not shown). Doses between 10 and 50 μM were equally effective at increasing cell viability, whereas doses above 50 μM were found to be toxic to the cells. To test the hypothesis that EPA 10 μM can counteract the MPP+-induced reduction in cell viability, both compounds were added to the culture media for 48 h. Two-way anova analyses revealed a significant interaction between MPP+ and EPA treatments (MPP × EPA: F(df 2,42) = 6.22, p < 0.01). Post hoc Bonferroni analyses showed that EPA at 10 μM significantly reversed the effect of MPP+ up to 10 μM (p < 0.01) (Fig. 1b). EPA did not protect against higher doses of MPP+. Thus, in subsequent assays for BNDF and TrkB mRNA expression, we tested the effect of EPA 10 μM against MPP+ 10 μM, both at 48 h.
BDNF and TrkB expressions in primary mesencephalic neurons
The mRNA expression of BDNF and TrkB receptor was measured in EPA/MPP+-treated (48 h) primary mesencephalic neuron-enriched cell cultures. In cells treated with EPA only, BDNF mRNA was significantly increased (p < 0.001) to 10-folds of the control. In the cells treated with MPP+ alone, a less but significantly increased mRNA expression of BDNF was also found (p < 0.05, Bonferroni test after two-way anova). EPA treatment did not reverse MPP+-induced change in BDNF (Fig. 2a). TrkB receptor expression was significantly increased by MPP+ (p < 0.001), but not by EPA. However, there was a significant interaction between MPP+ and EPA [F(df 1,28) = 5.4, p < 0.05] and post hoc tests showed that EPA significantly attenuated the MPP+-induced increase in TrkB expression (p < 0.001) (Fig. 2b).
Mitochondrial membrane potential (ΔΨm) in SH-SY5Y cells
MPP+ significantly decreased the ratio of red-to-green fluorescence in the cells with (p < 0.001, Bonferroni test after two-way anova) and without EPA treatment (p < 0.001, Bonferroni test) (Fig. 3). Interestingly, EPA by itself (no MPP+) increased the ratio, indicating that it enhanced the mitochondrial membrane potential compared with control (p < 0.05, Bonferroni test) (Fig.3), but did not reverse the effects of MPP+.
Cellular ultrastructure in SH-SY5Y cells
The effect of experimental treatment on the ultrastructure of fully differentiated SH-SY5Y cells, analyzed with EM, is shown in Fig. 4. EPA treatment resulted in visibly more slender mitochondria (Fig. 4b) compared with untreated control cells (Fig. 4a). A dramatically different picture emerged when cells were treated with MPP+ (Fig. 4c). Mitochondria appeared strongly enlarged, and electron-dense cytoplasmic inclusions distributed throughout the cytoplasm in MPP+-treated cells. These inclusions were likely lipid droplets, as the latter have a round (80–90 nm in diameter) shape with constant electron density. Remarkably, these structures were not present when MPP+-treated cells were cotreated with EPA (Fig. 4d).
Oxidative stress in SH-SY5Y cells
MPP+ increased the H2O2 production from SH-SY5Y cells (p < 0.001, Bonferroni after two-way anova) (Fig. 5a). The combined treatment of MPP+ and EPA resulted in a significant interaction effect [F(1, 39) = 12.32, p < 0.01, two-way anova] and post hoc Bonferroni tests showed that EPA attenuated the effect of MPP+ (p < 0.05) (Fig 5a), whereas the n-3 fatty acid by itself did not affect ROS production. MPP+ also increased NO production (p < 0.001, Bonferroni test) (Fig. 5b). Despite the fact that NO production was also significantly increased by MPP+ in EPA-treated cells (p < 0.001, Bonferroni test), the EPA and MPP+ treatment factors interacted significantly [F(2, 52) = 5.325, p < 0.001] and post hoc test showed that EPA reduced the effect of MPP+ (P < 0.05) (Fig. 5b), while it had no effect by itself.
As the enzyme NADPH oxidase is a source of ROS within cells, we tested the mRNA and protein expression of p47phox, a major subunit of this enzyme. As gene expression can be affected as early as 2–4 h following MPP+ treatment (Brill and Bennett 2003), we measured the effects of MPP+ at various incubation times (4, 12, 24, and 48 h). However, only the 48-h treatment with MPP+ significantly increased the p47phox mRNA (p < 0.001, Bonferroni test) (Fig. 5c) and protein expressions (p < 0.01, Bonferroni test) (Fig. 5d). When both MPP+ and EPA were added to the culture media, there was an interaction between both factors on both mRNA [F(2, 29) = 4.234, p < 0.05] and protein [F(1,15) = 9 .871, p < 0.01] expressions. Similar to ROS production, EPA attenuated the MPP+-induced increase in p47phox gene (p < 0.01, Bonferroni test) (Fig. 5c) and protein expression (p < 0.001, Bonferroni test) (Fig. 5d), whereas by itself, the n-3 fatty acid did not affect the enzyme at both mRNA and protein levels (Fig. 5c and d).
Antioxidant response in SH-SY5Y cells
MPP+ significantly increased GSH levels in SH-SY5Y cells with and without EPA treatment (p < 0.001, Bonferroni test after two-way anova) (Fig. 6a). EPA treatment by itself had no effect on GSH concentration, and also did not attenuate the effect of MPP+. Similar to p47phox, mRNA expression analyses of the effects of MPP+ and EPA on antioxidant enzymes were performed at various incubation times (4, 12, 24, and 48 h). Again, significant changes in superoxide dismutase (SOD), catalase, and GSHpx were only found after 48-h incubation. MPP+ significantly up-regulated the mRNA expression of GSHpx (p < 0.001, Bonferroni test) (Fig. 6b). While MPP+ also increased GSHpx mRNA expression in EPA-treated cells (p < 0.01, Bonferroni test), the n-3 fatty acid reduced the effect of MPP+ (p < 0.05) (Fig. 6b). This interaction was not statistically significant with a two-way anova test. Both SOD and catalase mRNA expression were increased by MPP+ to more than twice the control level (p < 0.001 and p < 0.05, respectively, Bonferroni test) (Fig. 6c and d), however, EPA did not attenuate these effects and also did not have an effect by itself on these genes.
Expression of cPLA2 and COX-2 mRNA in SH-SY5Y cells
Similar to previous genes, cPLA2 was only affected at 48 h of MPP+ treatment (the shorter incubation times tested negative). MPP+ significantly increased cPLA2 mRNA expression (p < 0.05) (Bonferroni test after two-way anova) (Fig. 7a). EPA treatment by itself had no effect on cPLA2 mRNA expression, but it interacted significantly with MPP+ treatment [F(2,29) = 4.83, p < 0.021] and post hoc analyses indicated that EPA significantly reduced the effect of MPP+ (p < 0.05, Bonferroni test) (Fig. 7a). Unlike cPLA2, COX-2 mRNA expression was decreased by EPA treatment as early as 4 h of incubation, whereas MPP+ did not have an effect at this incubation time (data not shown). This effect of EPA was observed in control (p < 0.05, Bonferroni test) as well as MPP+ (p < 0.05, Bonferroni test)-treated cells (data not shown). There were no effects of either EPA or MPP+ at 12- and 24-h incubation, but at 48 h of incubation, MPP+ significantly increased COX-2 mRNA expression (p < 0.01, Bonferroni test) (Fig. 7b). EPA by itself did not significantly affect COX-2 mRNA expression at 48 h (Fig. 7b), but it interacted significantly with MPP+ treatment [F(2,29) = 10.01, p < 0.01] and reversed the effect of MPP+ (p < 0.05, Bonferroni test) (Fig. 7b).
Expression of apoptotic markers bax, bcl-2, and caspase 3mRNA, and cytochrome c release in SH-SY5Y cells
The genes bax and bcl-2 were tested at different incubation times of MPP+ and EPA (4, 12, 24, and 48 h), but we did not find an effect of treatment until 48 h of incubation with either MPP+ or EPA. As bax was increased by MPP+ (p < 0.05, Bonferroni test after two-way anova) whereas bcl-2 was not, the bax:bcl-2 ratio was significantly increased by MPP+ in cells not treated with EPA (p < 0.05, Bonferroni test) (Fig. 8a). However, EPA treatment of the cells attenuated this effect and reduced the bax:bcl-2 ratio to control levels (p < 0.05) (Fig. 8a). MPP+ significantly increased cytochrome c release (p < 0.001, Bonferroni test) (Fig. 8b). EPA by itself also increased cytochrome c release, albeit less strongly (p < 0.05, Bonferroni test), but strongly interacted with MPP+ treatment [F(2,40) = 15.65, p < 0.0001]. EPA partially attenuated the effect of MPP+ (p < 0.01 Bonferroni test) (Fig. 8c). MPP+ significantly increased caspase 3 mRNA expression (p < 0.05) (Fig. 8c), whereas EPA treatment by itself did not affect caspase 3 and did not significantly interact with MPP+ (Fig. 8c).
Presently, we demonstrate that n-3 PUFA EPA attenuated an MPP+-induced decrease in cell viability in fully differentiated SH-SY5Y cells, by antioxidant, anti-inflammatory, and antiapoptotic mechanisms. The effect was confirmed in primary mesencephalic neuron-enriched cultures containing SNpc; cells more specifically relevant to PD. These findings are consistent with our recent observations that an EPA diet in mice has some protective action against MPTP or MPP+ (Meng et al. 2010; Luchtman et al. 2012). Other studies have also shown that compounds primarily rich in DHA (Bousquet et al. 2008, 2009; Delattre et al. 2010) have neuroprotective effects in experimental PD.
Mitochondrial dysfunction and lipid metabolism
The inhibition of complex I of the mitochondrial ETC. is one of the key MPP+-mediated cellular neurotoxic processes described by others (Dauer and Przedborski 2003; Drechsel and Patel 2008; Schapira 2010). Neurotoxicity because of complex I inhibition results from the consequent production of ROS, mitochondrial depolarization, and opening of the mitochondrial permeability transition pore, leading to cytochrome c release (Dauer and Przedborski 2003; Drechsel and Patel 2008; Schapira 2010). MPP+ presently decreased mitochondrial ΔΨm, caused mitochondrial enlargement, and increased cytochrome c release, all indicating that the neurotoxin impaired mitochondrial health in SH-SY5Y cells. Despite the fact that EPA attenuated an MPP+-induced reduction in cell viability, the n-3 PUFA did not counteract a decrease in ΔΨm and mitochondrial enlargement. Remarkably, however, under control conditions (no MPP+ treatment), EPA increased ΔΨm, caused more slender mitochondria and increased cell viability. Possible ways by which EPA could affect mitochondrial function include the promotion of beta-oxidation, or the up-regulation of genes that regulate mitochondrial function (Clarke 2000; Kitajka et al. 2004). While it is remarkable that EPA attenuated an MPP+-induced reduction in cell viability without improving ΔΨm, the relationship between neuronal viability and mitochondrial function may be complex (Nakamura et al. 2000; Dennis and Bennett 2003). For instance, rotenone, a specific complex I inhibitor was reported to be less toxic than MPP+ to dopaminergic neurons, but depolarized ΔΨm to a greater extent. As opening of the mitochondrial transition pore and release of cytochrome c is highly redox sensitive, it is conceivable that EPA's antioxidant effect, possibly through NADPH oxidase inhibition, may have inhibited apoptosis rather than a direct effect on mitochondrial function.
Aside enlarged mitochondria, cytoplasmic inclusions, consistent with lipid droplets, were observed in MPP+-treated cells. Fatty acid accumulation in lipid droplets may be a defense mechanism against ROS-induced lipid peroxidation. Furthermore, accumulation of α-synuclein, a protein that is pathologically aggregated in PD (Sere et al. 2010) and expressed in SH-SY5Y cells, leads to activation of fatty acid synthase (FA) and increased production of fatty acids. MPP+ can up-regulate α-synuclein and promote its aggregation in SH-SY5Y cells (Kalivendi et al. 2004). While the mechanisms of lipid droplet formation by MPP+ in SH-SY5Y cells need to be further investigated, it was a remarkable observation that EPA fully prevented their presence.
Oxidative stress is a major pathway of MPP+ cellular toxicity. A high level of oxidative damage has been observed in the SNpc of PD patients. The SNpc is particular sensitive to oxidative stress, because of the large presence of iron and microglia in this area (Lawson et al. 1990; Drechsel and Patel 2008). Because our cell cultures were devoid of glia, it is likely that the MPP+ increased ROS production through mitochondrial dysfunction and activity of pro-oxidant enzymes NADPH oxidase and COX-2. Interestingly, these enzymes, including NOS, have been largely associated with activation of microglia, but recent studies suggest that activation of their neuronal forms appears deleterious as well and may precede the onset of glial activation (Anantharam et al. 2007; Hoang et al. 2009; Zawada et al. 2011; Cristóvão et al. 2012). The discovery of other members of the NOX family of NADPH oxidases showed that enzymes with ROS generation as primary function are not limited to phagocytes (Bedard and Krause 2007). Yet direct in vivo evidence for NADPH oxidase complex assembly within neurons and a direct role in ROS-mediated neurotoxicity is currently limited. Furthermore, our present evidence for NADPHoxidase activation by MPP+ in SH-SY5Y cells is limited by the fact that we only measured the mRNA and protein expression of the P47phox, a major subunit of the NADPH oxidase complex.
EPA presently reduced the MPP+-induced increase in ROS and NO production. One antioxidant mechanism of EPA may have been the suppression of pro-oxidant enzymes, as EPA reduced P47phox mRNA and protein, and COX-2 mRNA expression. MPP+ increased antioxidant GSH and mRNA expression of antioxidant enzymes GSHpx, SOD, and catalase. This engagement rather than suppression of the antioxidant response by MPP+ has been reported before (Thomas et al. 2008). Similar to the presently observed increases in TrkB, BDNF, and bcl-2, this response may be part of the array of pro- and antisurvival pathways engaged by MPP+ (Brill and Bennett 2003).
Inflammation may also play a major role in the pathogenesis of PD (McGeer and McGeer 2008). Activated microglia are the main source of pro-inflammatory mediators, including cytokines and ROS (McGeer and McGeer 2008). Previously, we found that EPA treatment of mice fully attenuated MPTP-induced pro-inflammatory cytokine production (Luchtman et al. 2012). In our SH-SY5Y cultures, however, a source of inflammation was likely the cPLA2–COX-2-mediated AA cascade (Klivenyi et al. 1998), presently suggested by the increase in cPLA2 and COX-2 mRNA expression, although glia–neuron interactions may be necessary for this pathway to be neurotoxic (Wang et al. 2005). This is consistent with our finding that MPP+ in brain slices increases AA content and cPLA2 mRNA expression (Meng et al. 2010). Activation of the arachidonic acid cascade may lead to the production of potentially pro-inflammatory eicosanoids, such as PGE2. The fact that EPA down-regulated cPLA2 and COX-2 is consistent with its potent anti-inflammatory effects.
Evidence suggests that neurotrophic support is compromised in PD (Mogi et al. 1999) and experimental models of PD (Baydyuk et al. 2010; Ding et al. 2011), and partial deletion of the TrkB receptor leads to a reduced number of dopaminergic cells in the SN and the formation of α-synuclein aggregates in older mice (Von Bohlen und Halbach et al. 2005). We presently found that both BDNF and its cognate receptor were up-regulated by MPP+, which was mostly likely again an engagement of the cellular survival pathway in response to relatively acute exposure to stress and injury. Engagement of BDNF action after acute brain/cellular injury has been described extensively (Ebadi et al. 1997). Luellen et al. (2006) also reported an increase in Parkinsonian neurotoxin-induced BDNF protein expression. EPA by itself markedly increased BDNF, but when EPA and MPP+ were cocultured together, the expression of BDNF was not normalized. Bousquet et al. (2009) also reported an n-3 PUFA diet-induced increase in striatal BDNF mRNA expression, but MPTP treatment did not result in a decrease of either BDNF or TrkB mRNA expression in that study. Why EPA presently did not attenuate the MPP+-induced increase in BDNF, but did attenuate the increase in TrkB is unknown.
Oxidative stress and inflammation can induce apoptosis (Dauer and Przedborski 2003; Drechsel and Patel 2008). MPP+-induced ROS formation can stimulate opening of the mitochondrial transition pore, leading to cytochrome c release. Furthermore, bax promotes aberrant transition pore opening and evidence of increased bax-mediated apoptotic signaling in PD and the MPTP model has been demonstrated (Vila et al. 2001). Thus, an increased ratio of bax to antiapoptotic bcl-2 expression may result in transition pore opening, cytochrome c release, and activation of caspase 3. Presently, the bax:bcl-2 ratio was increased, as well as caspase 3 mRNA expression and cytochrome c release, all indicating that MPP+ promoted apoptosis in SH-SY5Y cells. More importantly, while EPA did not prevent increased caspase 3 mRNA expression, it significantly attenuated the bax:bcl-2 ratio and reduced the release of cytochrome-c, thereby reducing the potential formation of an apoptosome. These results are consistent with the antiapoptotic effects of EPA observed before in vivo (Lonergan et al. 2002, 2004).
The fully differentiated SH-SY5Y cells are one of the most commonly used cell lines to study PD (Presgraves et al. 2004; Luchtman and Song 2010; Xie et al. 2010). Nonetheless, to further validate the SH-SY5Y model and to increase the specificity of our findings to PD, we also used primary mesencephalic neuron-enriched cell cultures containing SNpc. The fact that MPP+ decreased cell viability and that EPA attenuated this effect in both cell types strengthens the validity of this finding altogether. One finding that stood out in the primary cells was that lower doses of MPP+ were required to achieve neurotoxicity in primary neurons. One possible mechanism for this result is that RA activated neuronal survival pathways in SH-SY5Y cells (Cheung et al. 2009), although the RA/TPA protocol is known to increase the sensitivity of SH-SY5Y cells to MPP+ compared with undifferentiated cells (Presgraves et al. 2004). A second side note is that we presently did not determine whether it is EPA or a secondary EPA metabolite, like docosapentaenoic acid (DPA, C22:5, n-3) or even DHA, that was responsible for the effects observed. It was unlikely to be DHA, as incorporation of newly formed DHA into phospholipids from its precursor EPA is strongly inhibited in SH-SY5Y cells (Langelier et al. 2005).
In sum, we have shown that EPA can significantly reduce MPP+-induced cellular toxicity, by antioxidant, anti-inflammatory, neurotrophic, and antiapoptotic mechanisms. Because of the importance of oxidative stress in the MPTP models, more research needs to be conducted to approach critical questions such as to what extent EPA (or another n-3 PUFA for that matter) affects mitochondrial ROS production or ROS production from major oxidant enzymes such as NADPH oxidase. Overall, our results are consistent with accumulating reports that n-3 PUFA may have beneficial effects in PD, and warrant further investigation. For instance, it would also be of interest to experimentally compare EPA, DHA, and formulas containing EPA and DHA in different proportions and determine what product has the most potent neuroprotective outcome.
This study was supported by grants to Cai Song from Amarin Neuroscience Ltd and Atlantic Innovation Fund, Canada. We greatly appreciate the gift of SH-SY5Y cells from Dr. Michael Maine, at the National Research Institute for Nutrisciences and Heath (INH, NRC). Author QingJia Meng is considered co-first author. There were no conflicts of interest.