Abstract
Melatonin mediates neuroprotection in several experimental models of neurodegeneration. It is not yet known, however, whether melatonin provides neuroprotection in genetic models of Huntington's disease (HD). We report that melatonin delays disease onset and mortality in a transgenic mouse model of HD. Moreover, mutant huntingtin (htt)-mediated toxicity in cells, mice, and humans is associated with loss of the type 1 melatonin receptor (MT1). We observe high levels of MT1 receptor in mitochondria from the brains of wild-type mice but much less in brains from HD mice. Moreover, we demonstrate that melatonin inhibits mutant htt-induced caspase activation and preserves MT1 receptor expression. This observation is critical, because melatonin-mediated protection is dependent on the presence and activation of the MT1 receptor. In summary, we delineate a pathologic process whereby mutant htt-induced loss of the mitochondrial MT1 receptor enhances neuronal vulnerability and potentially accelerates the neurodegenerative process.
Introduction
Huntington's disease (HD) is an autosomal-dominant chronic neurodegenerative disease that is universally fatal and lacks effective treatment. HD is caused by expansion of the CAG repeat in exon 1 of the HD gene (The Huntington's Disease Collaborative Research Group, 1993). Mutant huntingtin (htt) protein first kills selectively vulnerable medium spiny neurons (MSNs) in the striatum and thereafter the cortical neurons. As reported previously, we screened 1040 Food and Drug Administration-approved drugs for their ability to inhibit mitochondrial cytochrome c release triggered by proapoptotic stimuli (Wang et al., 2008). Melatonin inhibited cytochrome c release, activation of the caspase cascade, and cell death. Melatonin was also neuroprotective in a cellular model of HD (Wang et al., 2008).
Melatonin is an endogenously produced hormone secreted by the pineal gland and the retina (Cogburn et al., 1987). Within the body, the highest concentration of melatonin is in the brain (Reiter and Tan, 2002); within the cell, the highest concentration is in the mitochondria (Martín et al., 2000). Melatonin can be administered at high doses (Weishaupt et al., 2006) and is used clinically for sleep disturbances, circadian rhythm adjustment in the blind, and cancer (Ravindra et al., 2006). Most importantly, the potential benefits of melatonin have been evaluated in neurological diseases.
In addition to being a broad-spectrum antioxidant (Tan et al., 2002), melatonin is a ligand of several G-protein-coupled receptors. There are two mammalian isoforms of the melatonin receptor, MT1 and MT2. MT1 and MT2 are found in brain and peripheral tissues (Drew et al., 2001) (Allen Brain Atlas available at http://mouse.brain-map.org). Altered MT1 and MT2 expression levels have been reported in models of Alzheimer's disease (Savaskan et al., 2002, 2007), Parkinson's disease (Das et al., 2008), and normal aging (Sánchez-Hidalgo et al., 2009), as well as in senescence-accelerated mice (Caballero et al., 2008). In addition, melatonin blood levels are progressively reduced in HD patients (Aziz et al., 2009). Thus, the melatonin receptor–ligand axis may play a pathogenic role in HD and neurodegeneration.
Melatonin was shown to be protective in two HD-associated neurotoxin models, i.e., in rat brain homogenate treated with quinolinic acid (Southgate and Daya, 1999) and in striatal and cortical synaptosomes, isolated from rats treated with 3-nitropropionic acid (Túnez et al., 2004). However, melatonin has not been evaluated previously in transgenic mouse models of HD. Our data demonstrate that melatonin delays disease onset and prolongs lifespan in the R6/2 transgenic mouse HD model. In addition, we report the following. (1) MT1 receptor levels decrease in cultured striatal cells, mouse brain, and human striatum associated with mutant htt-mediated toxicity, and receptor depletion becomes greater as the disease progresses. (2) MT1 receptor knockdown by RNAi makes cells more vulnerable to cell death. Conversely, MT1 receptor overexpression increases resistance to cell death. (3) Melatonin administration counters MT1 receptor depletion attributable to mutant htt in vitro and in vivo. Together, these findings suggest that functional MT1 receptor depletion contributes to cell death from mutant htt.
Materials and Methods
Drugs.
Melatonin and luzindole were purchased from Sigma. 2-Iodo-melatonin was purchased from Tocris Bioscience.
Cell lines and induction of cell death.
Mutant htt ST14A striatal cells (8 Plx line) and parental ST14A striatal cells were kindly provided by Dr. Elena Cattaneo (University of Milan, Milan, Italy). Cell death was induced by shifting cells from the permissive temperature of 33°C to the nonpermissive temperature of 37°C and transferring them to serum-deprived medium (SDM) (Wang et al., 2005). Alternatively, cells were challenged by treatment with 1 mm H2O2 or with the combination of 10 ng/ml tumor necrosis factor α (TNFα) and 10 μm cycloheximide (CHX). In a second system, primary cerebrocortical neurons (PCNs) and primary striatal neurons (PSNs) were harvested from E14–E16 mice. Cultures of these cells were subject to oxygen–glucose deprivation (OGD) as described previously (Zhang et al., 2003a; Wang et al., 2009). In brief, culture medium was replaced with glucose-free Earle's balanced salt solution, and cells were incubated for 2 h with a defined concentration of melatonin. The cells were then placed in an anaerobic chamber with a BBL GasPak Plus (BD Biosciences), lowering the oxygen concentration to <100 ppm within 90 min. After 3 h, OGD was terminated by a return to normal culture conditions. Control cells were incubated for the same length of time in Earle's balanced salt solution with glucose in a normoxic incubator. Alternatively, PCNs (or PSNs) were treated for 18 h with 1 mm H2O2. In all experiments, cells were preincubated for 2 h with the test drug. Cell death was measured by the MTS [3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium] or the LDH release assay (Roche Applied Science).
Expression vectors and transfection procedures.
The human MT1 receptor (cDNA clone MGC:104051 IMAGE: 30915497) cloned in the pCR4–TOPO vector was purchased from Invitrogen. The MT1 sequence was cut from pCR4–TOPO, epitope tagged, and transferred to the expression vector pcDNA3.1–Flag–hMT1a–GFP. Proper construction was confirmed by DNA sequencing (Biomyx Technology). Transient transfection of pcDNA3.1–Flag–hMT1a–GFP into mutant htt ST14A cells was performed using LipofectAMINE (Invitrogen) for 3 d. Cells were then shifted to nonpermissive conditions for 18 h. Cells were cotransfected with a plasmids encoding GFP. Morphology of cells was observed. This quantitative observation of cell death was made by counting the cells with morphologic features indicative of apoptosis (i.e., chromatin condensation, nuclear fragmentation, and cell shrinkage).
Western blotting.
Cells were collected in ice-cold lysis buffer [20 mm Tris, pH 8.0, 137 mm NaCl, 10% glycerol, 1% Nonidet P-40, and 2 mm EDTA with 5 mm Na2VO4, protease inhibitor mixture (Roche Molecular Biochemicals) supplemented with 0.2 mm PMSF]. The lysate was cleared by centrifugation at 19,720 × g for 10 min at 4°C, and the supernatant was analyzed by Western blotting. Samples of mouse brain were lysed in ice-cold RIPA buffer with protease inhibitors (Zhu et al., 2002; Wang et al., 2003). Antibodies to caspase-3 and caspase-9 were purchased from Cell Signaling Technology, ones to MT1 and MT2 from Millipore Bioscience Research Reagents, and ones to Rip2 from Alexis Biochemicals or BD Biosciences. Antibodies to caspase-1 were from Santa Cruz Biotechnology or BioSource and those to β-actin were from Sigma. Secondary antibodies and ECL reagents were from GE Healthcare.
Detection of factors released from mitochondria.
Separation of cytoplasmic and mitochondrial fractions of cells and tissues was conducted as described previously (Zhu et al., 2002; Wang et al., 2003). Antibodies to cytochrome c were purchased from BD Biosciences Pharmingen, ones to Smac/Diablo from Novus Biologicals, and ones to AIF from Sigma.
Isolation of brain mitochondria.
Brain mitochondria were isolated as described previously with some modification (Zhu et al., 2002; Krasnikov et al., 2005; Singh et al., 2006). Briefly, forebrains were quickly removed and homogenized in ice-cold buffer 1 (in mm: 320 sucrose, 10 HEPES, pH 7.4, and 1 EGTA, pH adjusted to 7.4 with KOH) using a tissue homogenizer (Kontes Glass). The homogenate was centrifuged twice at 3000 rpm for 3 min in a JA-20 rotor (Beckman Instruments), and the supernatant was further centrifuged at 11,000 rpm for 8 min. The pellet, containing crude mitochondria, was resuspended in buffer 1 and loaded atop a discontinuous Ficoll gradient of 7.5 and 10% in 320 mm sucrose, 10 mm HEPES-KOH, pH 7.4, and 0.02 mm EGTA, pH adjusted to 7.4 with KOH. The gradient was spun in a Beckman Instruments centrifuge with an SW-28 high-speed rotor at 23,400 rpm for 12 min. Pelleted non-synaptosomal mitochondria were rinsed twice, first with buffer 1 and then with buffer 2. Mitochondria were lysed in RIPA buffer containing protease inhibitors and were analyzed by immunoblotting.
Caspase activity assay.
Test cell cultures contain 5 μm melatonin, whereas controls are devoid of melatonin. Extracts were prepared and enzyme assays conducted as described previously (Wang et al., 2003) (Clontech). The ApoAlert caspase fluorescent assay kit was from Clontech, caspase-1-like substrate Ac-YVAD-AFC and caspase-9-like substrate Ac-LEHD-AFC were from Calbiochem, and caspase-3-like substrate Ac-DEVD-AFC was from BD Biosciences Pharmingen. Enzyme activity was determined with a Bio-Rad Versa Fluorometer. The evaluation of the effect of melatonin on enzymatic activities were performed in parallel with previously reported experiments evaluating methazolamide, but some of the controls are the same as in the previous report (Wang et al., 2008).
Human brain specimens.
Frozen postmortem samples of human brain were obtained from the Harvard Brain Tissue Resource Center (McLean Hospital, Belmont, MA). They were taken from the caudate and putamen of seven grade IV HD patients, four grade II HD patients, and six control samples from non-neurologic patients. There is no systematic difference between the average postmortem interval or age at death for persons in the three different groups (postmortem intervals, 21.9, 17.3, and 19.4 h; age, 58.9, 65.3, and 49.8 years for control, grade II, and grade IV group, respectively).
HD mice and treatment regimen.
R6/2 mice (The Jackson Laboratory) were used to establish a breeding colony. Male R6/2 mice were bred with congenic (i.e., B6SLJ) wild-types females (The Jackson Laboratory). In this study, R6/2 mice used were known to have a CAG repeat size of 110–115 (Wang et al., 2008). Starting at 6 weeks of age, test R6/2 mice were given daily intraperitoneal injections of 30 mg/kg melatonin, whereas control animals received the saline vehicle. Each group contained an equal number of male and female mice. Mice were evaluated weekly by placing them on a rotarod turning at 5 or 15 rpm and measuring the time until they fell (up to 7 min). Mortality was defined as the age of death or the age when the mouse was unable to right itself within 30 s or the age of death. In this study, we used nine mice in the melatonin group and compared them to six littermate mates in the saline control group. For molecular analysis of the effect of melatonin, matched littermates were killed at 25 weeks of age, and the brain tissue was analyzed. Experiments were conducted in accordance with protocols approved by the Harvard Medical School Animal Care Committee.
Stereology.
Serial-cut coronal tissue sections from the rostral segment of the neostriatum to the level of the anterior commissure (interaural 5.34 mm, bregma 1.54 mm to interaural 3.7 mm, bregma 0.10 mm; Franklin and Paxinos, 2000) were used for volumetric measurement of ventricular size. Ventricular areas were measured using Neurolucida Stereo Investigator software (MicroBrightField), with researchers (R.J.F. and K.S.) blinded to the experimental condition. All specimens were stained at the same time, photographed under the same conditions, and are age matched.
Measurement of huntingtin aggregates.
The extent of huntingtin protein aggregates was determined in serially cut coronal tissue sections from the most rostral segment of the neostriatum to the level of the anterior commissure (interaural 5.34 mm, bregma 1.54 mm to interaural 3.7 mm, bregma −0.10 mm). Neurolucida Stereo Investigator software was used to conduct unbiased stereological counts of huntingtin-positive aggregates (≥1.0 μm) in neostriatum from melatonin- and saline-treated R6/2 mice (25 weeks of age). The respective areas of the rostral brain, ventricles, and neostriatum were measured by imaging serial sections. To obtain unbiased measurements of aggregate density, counting frames were randomly sampled within each preparation from neostriatum. The number of huntingtin-positive aggregates was determined by the optical-dissector method reference. The areas of striatal neurons were determined by microscopic video capture with a Windows-based image analysis system (Optimas; Bioscan). Although this software identifies and measures profiles automatically, its reliability was verified by visual inspection of tissue preparations.
Immunohistochemistry.
Brains were cryoprotected in 30% sucrose and frozen in cold isopentane. Frozen sagittal sections (10 μm thick) were washed with 0.05% Tween 20 in PBS and permeabilized with 70% methanol and with 0.05% Triton X-100. The sections were blocked and incubated with MT1 antibody. After a second wash with 0.05% Tween 20 in PBS, the preparations were stained with FITC-conjugated secondary antibodies (a control procedure that omits staining with the primary antibodies determined the extent of background staining by the secondary antibody). After additional washes with 0.05% Tween 20 in PBS and DAPI counterstaining, sections were viewed with a Nikon Eclipse TE 200 fluorescence microscope.
RNAi.
Four siRNAs targeting MT1 expression were purchased from Qiagen. They were synthesized as pairs of oligoribonucleotides as follows. MT1 siRNA 1 targets sequence AACGCAATCATATACGGACTA and consists of 5′-CGCAAUCAUAUACGGACUAtt-3′ and 3′-ttGCGUUAGUAUAUGCCUGAU-5′ as sense and antisense strands, respectively. MT1 siRNA 2 targets sequence CGGGATCGCTATGAACCGCTA and consists of GGAUCGCUAUGAACCGCUAtt-3′, and 3′-GCCCUAGCGAUACUUGGCGAU-5′ as sense and antisense strands, respectively. MT1 siRNA 3 targets sequence CTCAGGAACGCAGGGAATATA and consists of 5′-CAGGAACGCAGGGAAUAUAtt-3′ and 3′-GAGUCCUUGCGUCCCUUAUAU-5′. MT1 siRNA 4 targets sequence 5′-GGCGCTGACGTCTATACTTAA-3′ and consists of 5′-CGCUGACGUCUAUACUUAAtt-3′ and 3′-ccGCGACUGCAGAUAUGAAUU-5′ as sense and antisense strands, respectively. Cells were transiently transfected with these siRNAs using HiPerFect Transfection Reagent (Qiagen). Forty hours after transfection, cells were analyzed using quantitative RT (qRT)-PCR and Western blotting. Apoptosis of mutant htt ST14A cells was induced in three different ways: (1) shifting to the nonpermissive temperature of 37°C with concomitant transfer to SDM; (2) exposure to H2O2 for 18 h; and (3) treatment with TNFα/CHX for 18 h.
Quantitative real-time RT-PCR.
Total RNA was isolated from mutant htt ST14A cells, mouse brains, or human striatum. The following pairs of primers were used: mouse MT1, 5′-TGAGTGTCATCGGCTCGATAT-3′ and 5′-TAGTAACTAGCCACGAACAGC-3′ (Niles et al., 2004); mouse MT2, 5′-CTCACTCTGGTGGCTTTGGTG-3′ and 5′-CTGCGCAAATCACTCGGTCTC-3′ (Kobayashi et al., 2005); human MT1, 5′-TGCTACATCTGCCACAGTCTC-3′ and 5′-CAGTAGCCCGTATATAATGGC-3′ (Niles et al., 1999); human MT2, 5′-TCATCGGCTCTGTCTTCAATA-3′ and 5′-ACTGGGTGCTGGCGGTCTGGA-3′ (Brydon et al., 2001); GAPDH, 5′-ATGGTGAAGGTCGGTGTCAACGGA-3′ and 5′-TTACTCCTTGGAGGCCATGTAGGC-3′ (Clontech). Real-time qRT-PCR was performed as described previously (Wang et al., 2005). The RT-PCR began by holding the temperature at 50°C for 30 min and then at 95°C for 10 min. Samples were then subject to 45 cycles of 95°C for 30 s, 55°C for 1 min, and 72°C for 30 s. Relative quantities of mRNAs were calculated from their Ct values (Wang et al., 2005). GAPDH was not variable in these experiments and was thus used as the control gene.
Statistical analysis.
Statistical significance was evaluated by the t test. Data on viability or the extent of cell death were compared by ANOVA or repeated-measures ANOVA and by nonpaired Student's t test. Statistical analyses were done using StatView 4.0 software (SAS Institute). Error bars represent SEM.
Results
Melatonin inhibits the mitochondrial and the Rip2/caspase-1 cell death pathways
Release of cytochrome c from mitochondria triggers a set of biochemical changes that lead to neuronal cell death. These molecular pathways are activated in a number of neurodegenerative disorders, including HD (Friedlander, 2003; Wang et al., 2003). We therefore tested whether melatonin inhibited the release of cytochrome c from the mitochondria of mutant htt ST14A cells. Mutant htt ST14A cells were constructed from the parental ST14A striatal cell line by transforming it with a temperature-sensitive large T antigen. Mutant htt ST14A cells undergo cell death during shift from 33°C to 37°C with concomitant transfer to SDM (i.e., change from permissive to nonpermissive conditions) (Rigamonti et al., 2001). Cytochrome c localization was determined by immunocytochemistry. Cells were stained with antibodies to cytochrome c and the mitochondrial marker MitoTracker (Fig. 1A). Merging the images revealed that the highest concentration of cytochrome c immunoreactivity colocalized with mitochondria. Temperature shift with transfer to SDM induced the release of cytochrome c from the mitochondria into the cytoplasm. As indicated by the arrows, we observed a more diffuse staining pattern in cells kept under nonpermissive versus permissive conditions. Indeed, melatonin inhibited cytochrome c release as indicated by the preservation of punctate staining after temperature shift. Cells treated with melatonin revealed a punctate pattern of staining even under nonpermissive conditions. Melatonin inhibits the release of cytochrome c from the mitochondria to the cytoplasm (Fig. 1A).
Smac and AIF are two other proteins whose release from mitochondria results in activation of cell death pathways. In a variety of neurodegenerative diseases, the appearance of these factors in the cytoplasm has been associated with mitochondrial dysfunction (Friedlander, 2003; Wang et al., 2003; Zhang et al., 2003a; Adamczyk et al., 2005). During disease progression in HD mice, release of cytochrome c/Smac/AIF triggers a set of events that activate mitochondrial cell-death pathways (Kiechle et al., 2002; Friedlander, 2003; Wang et al., 2003). Cytoplasmic cytochrome c, apaf-1, and pro-caspase-9 form a macromolecular assembly called the “apoptosome,” resulting in caspase-9 activation. In its turn, caspase-9 activates caspase-3, thereby triggering the terminal events in the cell-death pathway (Li et al., 1997). Melatonin partially inhibited mutant htt-induced release of mitochondrial apoptogenic factors and caspase activation (Fig. 1A--C). Caspase activation was evaluated by immunoblotting (Fig. 1B) and complementary fluorogenic activity assays (Fig. 1C).
In addition to its activation of caspase-9 and caspase-3 and cleavage at the caspase-6 site (Graham et al., 2006), mutant htt stimulates the Rip2/caspase-1 pathway in R6/2 mice and in mutant htt ST14A cultures (Ona et al., 1999; Chen et al., 2000; Zhang et al., 2003b; Wang et al., 2005). We demonstrated previously that increased abundance of Rip2 protein (a caspase-1 activator) (Thome et al., 1998) correlates with greater toxicity from mutant htt. We determined whether melatonin inhibits Rip2 accumulation and counters caspase-1 activation. Indeed, we observed that procaspase-1 is activated in mutant htt ST14A cells during transfer to nonpermissive conditions (Wang et al., 2003, 2005). As demonstrated by immunoblotting and complementary fluorogenic assay, melatonin inhibits mutant htt-induced caspase-1 activation (Fig. 1D).
Rip2 binds to caspase-1 via a caspase activation and recruitment domain (CARD)–CARD (for caspase recruitment domain domain) homotypic interaction resulting in caspase-1 activation. Rip2 protein levels increase in neurologic diseases, including HD (Wang et al., 2005), Alzheimer's disease (Engidawork et al., 2001), and stroke (Zhang et al., 2003a). Moreover, reducing the level of Rip2 by RNA interference inhibits mutant htt-induced cell death of ST14A cells (Wang et al., 2005). Given that melatonin inhibits caspase-1 activation, we determined whether this effect involves blocking Rip2 upregulation. Indeed, melatonin inhibited mutant htt-induced Rip2 upregulation in mutant htt ST14A cells (Fig. 1E).
Melatonin has been reported to modulate mitochondrial pathways through inhibition of the mitochondrial permeability transition (Andrabi et al., 2004). The collapse of the mitochondrial transmembrane potential (ΔΨm) is a central event in cell death (Petronilli et al., 1994; Friedlander, 2003; Stavrovskaya and Kristal, 2005). Consequently, preservation of ΔΨm by melatonin is a possible mechanism by which it inhibits cell death. We tested this hypothesis using rhodamine 123 (Rh 123), a fluorescent, hydrophobic, and positively charged molecule that accumulates in electrostatically charged mitochondria. We observed punctate staining by Rh 123, a pattern indicative of a negative mitochondrial membrane potential. Moreover, transferring cells to nonpermissive conditions resulted in dissipation of ΔΨm and ensuing cell death (Fig. 1F). Furthermore, melatonin preserved the punctate distribution of Rh 123 fluorescence within the cell. We conclude that the ability of melatonin to maintain ΔΨm likely contributes to the ability of the compound to rescue mutant htt ST14A from temperature shift-induced cell death.
Proteosomal dysfunction has been reported in HD (Bennett et al., 2007). We determined whether in our cellular system mutant htt causes malfunction of the ubiquitin–proteosome system. We found that, under nonpermissive conditions, mutant htt ST14A-expressing cells demonstrate a significant decline in proteosomal activity. Moreover, melatonin partially inhibits such stress-induced proteosomal dysfunction (Fig. 1G).
Melatonin-mediated protection requires the MT1 receptor
Neuroprotection by melatonin may be mediated by its ability to scavenge free radicals, by a traditional ligand–receptor interaction, or by a combination of the two. We addressed this question using two molecular analogs of melatonin: the melatonin-receptor antagonist luzindole and the agonist 2-iodomelatonin (Fig. 2A). As shown in Figure 2B, melatonin inhibited cell death as a result of temperature shift. To directly evaluate whether the effects of melatonin require receptor binding, we determined whether the receptor antagonist luzindole counters melatonin-mediated protection. In this in vitro system, luzindole almost completely eliminated neuroprotection by melatonin, indicative of a receptor-mediated effect as opposed to a free radical-scavenger mechanism (Fig. 2B). We repeated this experiment in other cell-death models, e.g., mutant htt ST14A cells exposed to H2O2 or treated with TNFα plus CHX (Fig. 2C,D). In both cases, luzindole blocked melatonin-mediated protection, confirming that receptor binding is central to the mechanism of protection.
The most vulnerable cells in HD are the MSNs of the striatum followed by cortical neurons (Li, 1999). Melatonin prevents cell death induced by OGD (Cazevieille et al., 1997; Wang et al., 2009). We tested whether melatonin inhibits cell death of PSNs induced by H2O2. Indeed, treatment with melatonin provided dose-dependent protection of PSNs challenged by this oxidative stress (Fig. 2E). In addition, melatonin diminished the extent of NMDA-induced death of primary cerebrocortical neurons (PCNs) (Fig. 2F). Again, luzindole eliminated this neuroprotection, demonstrating that the action of melatonin is mediated by its G-protein-coupled receptor (Fig. 2E,F).
We next evaluated whether the MT1/MT2 agonist 2-iodomelatonin shares the protective properties of melatonin. As expected from their common affinity for the MT1 and MT2 receptors, 2-iodomelatonin inhibited cell death of mutant htt ST14A. Moreover, luzindole essentially eliminated neuroprotection by 2-iodomelatonin (Fig. 2G). As observed with melatonin, 2-iodomelatonin inhibited the increase in Rip2 expression in stressed mutant htt ST14A cells. This effect was observed both in cells challenged by temperature shift and in ones exposed to H2O2 (data not shown). A clue to the mechanism of these phenomena is the ability of melatonin to inhibit mutant htt-induced release of mitochondrial cytochrome c in ST14A cells. Moreover, luzindole eliminates this molecular activity of the MT1 receptor agonist. The correlation between inhibition of cytochrome c release and protection of neurons from cell death implicates the effect of melatonin on mitochondria in neuroprotection (Fig. 2H).
Mutant htt-mediated toxicity selectively causes MT1 receptor loss
We investigated whether the toxicity of mutant htt correlates with changes in the expression of melatonin receptor MT1, receptor MT2, or both. We found that levels of MT1 protein dropped significantly when mutant htt ST14A cells were shifted to a nonpermissive temperature. Decreased levels of MT1 receptor were also observed when these cells were exposed to H2O2 (Fig. 3A). In both conditions, statistical analysis demonstrate that melatonin treatment significantly ameliorated MT1 loss (Fig. 3A). In the same cells, however, neither stress altered the level of MT2 protein (Fig. 3A). These data, as well as the ability of luzindole to inhibit melatonin-mediated protection, points to the MT1 receptor as a key modulator of this effect.
Next we compared the effect of stress on MT1 expression in parental and mutant htt ST14A cells. We found that temperature shift with SDM [which also induces cell death in the parental cell line but to a lesser degree than in the mutant htt line (Wang et al., 2005)] induced a moderate decrease in expression of MT1 (Fig. 3B). However, the change in MT1 expression was less dramatic in parental ST14A cells (51%) compared with mutant htt ST14A striatal cell lines (82%). This suggests that loss of MT1 receptor is not a specific mutant htt-mediated phenomenon but rather a more fundamental response to cellular stress.
Knockdown of MT1 sensitizes neurons to cell death, whereas its overexpression is neuroprotective
To confirm the importance of the MT1 receptor to the action of melatonin, we used the technique of RNA interference. Four DNA fragments were synthesized, each encoding an siRNA targeting the MT1 message. These molecules were transfected into mutant htt ST14A striatal cells with ∼90% efficiency. Of the four oligonucleotides, MT1 siRNA1 and MT1 siRNA2 effectively diminished levels of the MT1 mRNA and the encoded protein (Fig. 3C,D).
We investigated the effects of RNAi-mediated knockdown of MT1 in mutant htt ST14A cells. Test cells expressing MT1 siRNA 1 or siRNA 2 and an appropriate control were challenged by shift to nonpermissive conditions, exposure to H2O2, or treatment with TNFα/CHX. In all three systems, MT1 knockdown completely eliminated melatonin-mediated protection (Fig. 3E).
Increasing the abundance of MT1 is expected to make cells more resistant to cell death. To test this hypothesis, we constructed the plasmid pcDNA3.1–MT1–GFP, a DNA molecule that expresses the MT1 receptor fused to GFP. Mutant htt ST14A cells were transiently transfected with either this plasmid or a control GFP-expressing plasmid. When shifted to 37°C with SDM, control cells underwent extensive cell death, whereas ones expressing the MT1–GFP fusion protein were relatively resistant. This qualitative observation was made rigorous by counting the cells with morphologic features indicative of apoptosis (i.e., cell shrinkage and chromatin condensation). Mitochondrial function was determined by the MTS assay, again showing that pcDNA3.1–MT1–GFP rescues mutant htt ST14A from cell death caused by this stress (Fig. 3F). The overexpression of the MT1 receptor (in the absence of exogenous melatonin) is sufficient to significantly attenuate the temperature shift-mediated cell death demonstrates that MT1 receptor itself is neuroprotective. Increasing the abundance of MT1 by either the transient transfection of MT1 receptor or the stimulation of melatonin (agonist of MT1) make cells more resistant to cell death. The correlation of MT1 concentration and cell rescue supports our contention that neuroprotection by melatonin depends on its interaction with this receptor molecule.
Melatonin delays disease onset and mortality in R6/2 mice
Having established the ability of melatonin to inhibit cell death in vitro, we investigated its effects in a mouse model of HD. We used the R6/2 mouse, a transgenic animal expressing a human gene encoding the N-terminal portion of polyQ-expanded huntingtin (Mangiarini et al., 1996). Beginning at 6 weeks of age, R6/2 littermate mice received daily intraperitoneal melatonin injections of 30 mg/kg or saline vehicle (n = 9 and 6, respectively; Fig. 4A--D). Disease onset, defined as the age when mice can no longer remain for 7 min on a rotarod turning at 15 rpm (Fig. 4A), was delayed 19% by melatonin treatment. Declining performance on a rotarod turning at 5 rpm (a more sensitive indicator of symptomatic disease progression) was also slowed by melatonin (Fig. 4B). Finally, melatonin prolonged the lifespan of these HD mice by 18% (Fig. 4D). In addition, using a larger sample size that includes both littermate and nonlittermate R6/2 mice that received daily intraperitoneal melatonin injections of 30 mg/kg (n = 12) or saline vehicle (n = 24), we found that disease onset was delayed 23% by melatonin treatment, whereas melatonin prolonged the lifespan of these HD mice by 21% (data not shown). In the course of disease progression, R6/2 mice lose a significant amount of body weight. The time course of weight loss is not significantly altered by melatonin in littermate mice (Fig. 4C) or in the entire sample, including littermates and non-littermates (data not shown). This remains true regardless of whether male and female mice are analyzed together or separately. It is noteworthy that, in many other studies using animal models for neurologic disease, drug and gene therapies are found not to change the pattern of weight loss (Ona et al., 1999; Chen et al., 2000; Chopra et al., 2007; Masuda et al., 2008; Wang et al., 2008).
Progressive ventriculomegaly is observed in R6/2 mice (Ferrante et al., 2002). Moreover, neuroprotective drugs counter this pathology (Ferrante et al., 2002; Stack et al., 2006; Wang et al., 2008). Consequently, we monitored ventricular hypertrophy in melatonin- and saline-treated R6/2 mice. We found that melatonin treatment significantly reduced ventriculomegaly in this animal model for HD (saline, 4.07 ± 1.02 mm3; melatonin, 0.91 ± 0.39 mm3; F(2,18) = 15.12, *p < 0.01) (Fig. 4E).
Progressive accumulation of huntingtin-protein aggregates in neurons is a pathological hallmark in R6/2 mice (Turmaine et al., 2000). This process was modestly slowed by melatonin treatment, although our results were not statistically significant (saline, 5.27 × 106 ± 1.18; melatonin, 5.06 × 106 ± 1.04; F(2,12) = 21.61, p = 0.27) (Fig. 4F). The failure of melatonin to prevent formation of huntingtin aggregates resembles observations using a filter-retardation assay in vitro (Heiser et al., 2000). Furthermore, several neuroprotective in vivo regimens failed to stop such protein aggregation (Ona et al., 1999; Chen et al., 2000; Wang et al., 2008).
The most important finding discussed above is that melatonin inhibits mutant htt-induced activation of cell-death pathways in cultured cells (Fig. 1). Analogous experiments in R6/2 mice demonstrated the ability of melatonin to effect equivalent changes in vivo. In particular, mitochondrial release of cytochrome c, AIF, and Smac was significantly diminished in R6/2 mice by administration of melatonin (Fig. 5A). As reported previously, we found that brain tissue from R6/2 mice contains activated caspase-9 and caspase-3 (Chen et al., 2000; Kiechle et al., 2002; Wang et al., 2003). Moreover, administration of melatonin diminished these proapoptotic changes (Fig. 5A,B).
The Rip2/caspase-1-mediated cell-death pathway is another physiological process important to neuronal dysfunction in R6/2 mice (Wang et al., 2005). Consequently, we determined whether the benefit of melatonin to these mice is associated with inhibition of this pathway. As described previously (Chen et al., 2000; Wang et al., 2003, 2005; Zhang et al., 2003b), Western blots of brain tissue from R6/2 mice revealed upregulation of Rip2 expression and increased caspase-1 cleavage (Fig. 5A). Consistent with our observations in cellular models for HD (Fig. 1D,E), we found that melatonin inhibits Rip2 upregulation and caspase-1 activation in R6/2 mice (Fig. 5A).
Levels of MT1 receptor decline as HD progresses; melatonin halts its depletion
It is evident from Figure 3A that depletion of MT1 receptor is associated with cell death in mutant htt ST14A cells. Knockdown of MT1 expression eliminates melatonin-mediated neuroprotection (Fig. 3E). In addition, overexpression of MT1 results in increased resistance to cell death (Fig. 3F). To evaluate the potential role of the MT1 receptor in vivo, we determined the abundance of protein in the brains of R6/2 mice. We found the MT1 receptor to be less abundant in brain tissue from symptomatic R6/2 mice than in equivalent samples from age-matched wild-type littermates. In contrast, levels of MT2 receptors were similar in brains from R6/2 and control mice. As observed for the receptor protein, there was less MT1 mRNA in R6/2 than in wild-type brains (Fig. 6A). Remarkably, treating R6/2 mice with melatonin countered the loss of MT1 receptor mRNA (p < 0.001). This increase in MT1 mRNA was reflected in a nonstatistically significant increase on MT1 protein. Consistent with observations in cellular models for HD, levels of MT2 receptor protein and mRNA were similar in brains of wild-type, saline-injected R6/2, and melatonin-treated R6/2 mice (Fig. 6B). The most significant reductions in MT1 receptor expression occurred in the cortex (Fig. 6C), with more subtle effects in the striatum (data not shown).
The highest subcellular levels of melatonin are observed in mitochondria (Martín et al., 2000). Consequently, we measured levels of melatonin receptors in brain mitochondria from R6/2 and wild-type mice. We found that both MT1 and MT2 receptors are expressed in brain mitochondria (Fig. 6D), suggesting that the mitochondrial cell-death pathways may be mediated at least in part by the interactions of melatonin with its receptors, although it is possible that melatonin has additional, non-receptor-mediated targets. In whole-cell extract, the level of MT1 is lower than that of MT2 receptor (Fig. 6D, right). Consistent with its role in cell death pathways, MT1 receptor is enriched in the mitochondrial fraction (Fig. 6D, left). Furthermore, MT1 receptor is decreased fourfold in brain mitochondria from R6/2 mice compared with wild-type littermates (Fig. 6E).
Expression of MT1, MT2, and the encoding mRNAs in human striatum
To determine the relevance of our cellular and mouse models of HD to clinical cases, we investigated the expression of MT1 and MT2 in human striatum. Levels of MT1 and MT2 receptor proteins and their respective mRNAs were measured in postmortem samples of striatum from four persons who died at HD grade II and seven persons who survived until grade IV. Data from these samples were compared with that from controls, i.e., postmortem tissue from six persons deceased of non-neurologic causes. We observed progressive depletion of both the MT1 receptor protein and the encoding MT1 transcript as demonstrated by the visible MT1 lanes in the control samples, reduced intensity of MT1 bands in grade II samples, and the absence of MT1 bands in most grade IV samples (Fig. 7A). Positive bands were visible in all lanes using β-actin, demonstrating that sufficient protein was loaded in each lane. Minor differences in loading of protein were accounted for in the accompanying quantitative graph (Fig. 7A). In contrast, there was little correlation of the levels of MT2 mRNA or MT2 protein with the severity (or even presence) of HD (Fig. 7B). These analyses of melatonin receptors in human striatum confirm the relevance to clinical HD of our findings in cellular and mouse models.
Discussion
Melatonin interferes with the mitochondrial and the Rip2/caspase-1 cell-death pathways
We have identified aberrant molecular events that favor activation of cell-death pathways in cellular and mouse models of HD. In particular, depletion of the MT1 receptor within mitochondria reduces the protective properties of melatonin, therefore increasing cellular vulnerability to proapoptotic challenge. An important result came from investigations on MT1 expression in conditionally immortalized mutant htt ST14A cells. In this system, we observed depletion of MT1 mRNA and protein during induction of cell death. In the mouse model of HD, melatonin treatment increases RNA expression of the MT1 receptor. However, the data do not exclude the possibility that melatonin treatment could also directly alter the expression of the huntingtin gene, affecting disease onset, progression, and survival. It is noteworthy that similar changes in the expression of melatonin receptor occur in other neurodegenerative diseases and during aging (Savaskan et al., 2002; Caballero et al., 2008; Sánchez-Hidalgo et al., 2009). Moreover, we observed low levels of MT1 (both protein and mRNA) in the brains of R6/2 mice and in postmortem specimens of HD human striatum. RNAi-mediated knockdown of MT1 receptor reduced the ability of melatonin to inhibit cell death in vitro, increasing cellular vulnerability. Conversely, overexpression of MT1 increased resistance to cellular stress.
The benefit of melatonin to stressed cells has been attributed to its ability to scavenge free radicals (Lena and Subramanian, 2003). More likely than not, melatonin counters cell death through both a receptor-mediated mechanism, as described above, and an antioxidant mechanism because the administration of a neuroprotective dose of melatonin does not completely restore MT1 expression. These actions become progressively attenuated as melatonin levels decline during aging (Mocchegiani et al., 1994; Carranza-Lira and Garcia Lopez, 2000). Changes in expression of the melatonin receptor may also contribute to this downward trend. There is a large variability in melatonin productivity depending on mouse genetic background (Kasahara et al., 2010), and inbred strains of laboratory mice such as C57BL/6J were found not to have detectable melatonin in their pineal glands.
Our investigations on melatonin were motivated by the ability of the compound to inhibit calcium-induced cytochrome c release from purified mitochondria (Wang et al., 2008). When administered to cells, melatonin blocked transfer to the cytoplasm of cytochrome c and blocked the release of mitochondrial factors Smac/Diablo and AIF. Because mitochondria are the organelles dedicated to generation of energy, they affect every aspect of cellular metabolism. Moreover, maintaining a rich store of ATP is particularly important at times of cellular stress, e.g., when cells are challenged by the expression of mutant htt. A molecular event characteristic of the cell-death process is dissipation of the mitochondrial membrane potential (ΔΨm). Not surprisingly, loss of ΔΨm is observed in models for many neurodegenerative disease, including ones for HD (Andrabi et al., 2004). We have now demonstrated that melatonin inhibits mutant htt-induced ΔΨm loss (Fig. 1F). This activity further inhibits the release of mitochondrial proapoptotic factors and makes melatonin all the more potent at countering cell death.
Rip2 dysregulation mediates aberrant caspase-1 activation in HD (Wang et al., 2003) and in cerebral ischemia (Zhang et al., 2003a). Other researchers have observed increased levels of Rip2 in Alzheimer's disease (Engidawork et al., 2001). We recently reported that melatonin inhibits caspase-1 activation in an experimental model of cerebral ischemia (Wang et al., 2009). We now show that melatonin counters the upregulation of Rip2 in both mutant htt ST14A cells transferred to nonpermissive conditions (Fig. 1) and R6/2 transgenic mice (Fig. 5). Moreover, Rip2 protein stimulates autocatalytic activation of procaspase-1 (Thome et al., 1998). The ability of melatonin to rescue stressed cells is apt to result, at least in part, from its targeting of this upstream event in the cell-death pathway. Alternatively, or in addition, the primary target of melatonin may be the mitochondria, because upstream modulation of cell death pathways may result from a feedback-loop response.
Melatonin and its analogs are candidate drugs for HD therapy
The potential benefits of melatonin have been evaluated in patients with and in animal models for Alzheimer's disease (Srinivasan et al., 2006), amyotrophic lateral sclerosis (Weishaupt et al., 2006), Parkinson's disease (Willis and Armstrong, 1999; Medeiros et al., 2007), stroke (Kilic et al., 2004; Dominguez-Rodriguez et al., 2007), epilepsy (Gupta et al., 2004; Savina et al., 2006), and traumatic brain injury (Ozdemir et al., 2005).
Melatonin is an endogenous hormone with great importance to health and disease. A gradual decrease in melatonin production has been observed in both a mouse model of aging (Mocchegiani et al., 1994) and investigations on human subjects (Carranza-Lira and Garcia Lopez, 2000). Recently, a significant delay was reported in melatonin peaks in humans with early-stage HD. Furthermore, there is a progressive reduction of melatonin blood levels associated with increasing disease severity (Aziz et al., 2009). Low levels of circulating melatonin is not specific to HD; it has also been observed in patients with Alzheimer's disease (Ozcankaya and Delibas, 2002) and Parkinson's disease (Blazejova et al., 2000).
2-Iodomelatonin, an agonist of MT1 and MT2, shares the neuroprotective properties of melatonin. Conversely, luzindole, an antagonist of these two receptors, eliminates neuroprotection by melatonin. The structures of these three molecules are shown in Figure 2, and their respective effects on cell death are noted. These observations are of practical importance: melatonin analogs and other agonists of the melatonin receptors are potential drugs for treating HD. In this manner, we hope to identify molecules that have a longer half-life than melatonin (20–50 min) (Waldhauser et al., 1984).
We and others have conducted drug trials in HD transgenic mice. Thirty-two potentially therapeutic agents are analyzed for their relative effectiveness in monotherapies (table not shown). Only five of these test compounds were observed to extend survival by >20%. Melatonin was one such drug, delaying mortality by 21%. The low toxicity of melatonin and its ability to cross the blood–brain barrier make it a potentially useful compound for the treatment of HD.
In conclusion, we have demonstrated that the interaction of melatonin with the mitochondrial MT1 receptor is critical for neuroprotection. Neither the localization of MT1 receptors to mitochondria nor their selective loss in HD has been reported previously. In HD, pathologically low levels of circulating melatonin and reduced expression of MT1 receptor increase the vulnerability of neurons to cell death. We have also demonstrated that therapies that restore high levels of melatonin–MT1 receptor complexes provide a potential avenue for neuroprotection in HD.
Footnotes
-
This work was supported by grants from the National Institutes of Health/National Institute of Neurological Disorders and Stroke (R.M.F., X.W., R.J.F.), the Muscular Dystrophy Association (X.W.), the Huntington's Disease Society of America (R.M.F.,) and the Veterans Administration (R.J.F.). We thank Dr. Ethan M. Shimony and Dr. Diane Carlisle for editorial assistance, Dr. Irina Stavrovskaya, Jennifer Moody, and Diane Sheldon for experimental assistance, and Dr. Ahmad Aziz for insightful discussion.
- Correspondence should be addressed to either of the following: Dr. Robert M. Friedlander, University of Pittsburgh School of Medicine, Department of Neurological Surgery, University of Pittsburgh Medical Center Presbyterian Hospital Suite 449, 200 Lothrop Street, Pittsburgh, PA 15213, friedlanderr{at}upmc.edu; or Dr. Xin Wang, Brigham and Women's Hospital, Harvard Medical School, Department of Neurosurgery, Boston, MA 02115, xwang{at}rics.bwh.harvard.edu