CpG Demethylation Enhances Alpha-Synuclein Expression and Affects the Pathogenesis of Parkinson's Disease

Background Alpha-synuclein (SNCA) gene expression is an important factor in the pathogenesis of Parkinson's disease (PD). Gene multiplication can cause inherited PD, and promoter polymorphisms that increase SNCA expression are associated with sporadic PD. CpG methylation in the promoter region may also influence SNCA expression. Methodology/Principal Findings By using cultured cells, we identified a region of the SNCA CpG island in which the methylation status altered along with increased SNCA expression. Postmortem brain analysis revealed regional non-specific methylation differences in this CpG region in the anterior cingulate and putamen among controls and PD; however, in the substantia nigra of PD, methylation was significantly decreased. Conclusions/Significance This CpG region may function as an intronic regulatory element for SNCA gene. Our findings suggest that a novel epigenetic regulatory mechanism controlling SNCA expression influences PD pathogenesis.


Introduction
Parkinson's disease (PD) (MIM168600) is a neurodegenerative disease characterized by akinesia, rigidity, tremor, and postural instability. The prevalence is more than 1% of the population over 65 years old [1]. Most of the PD cases are sporadic, although 10% is hereditary and caused by mutations in SNCA, PARK2, LRRK2, PINK1, UCHL1, or DJ1, in the majority of cases [2]. Mutations in these genes are occasionally found in sporadic PD; however, the majority of sporadic cases show no evidence of mutation in any of the known causative genes.
The pathological hallmark of PD is Lewy bodies (LB), which are intracellular inclusion bodies consisting of aggregated alphasynuclein. Abnormal accumulation of aggregated protein is closely associated with the pathogenesis of many neurodegenerative diseases. The precise mechanism of aggregation remains unknown, but increased expression of aggregation-prone proteins can lead to their aggregation. For example, in Down syndrome, duplication of the 21 st chromosome, which contains the amyloid beta precursor protein (APP) gene, leads to accumulation of amyloid beta and Alzheimer's disease pathology [3]. In rare cases of PD, duplication or triplication of SNCA gene leads to alphasynuclein accumulation, with triplication producing a more severe phenotype than duplication, suggesting that SNCA expression level determines the severity of the pathology [4,5,6]. Animal models of neurodegenerative disorders are generated by over-expression of causal genes, further supporting the conclusion that increased gene expression is related to pathogenesis [7]. Additional evidence indicates that SNCA promoter polymorphisms increases alphasynuclein expression and increases susceptibility to sporadic PD [8,9,10]. On the basis of these reports, we hypothesized that there may be another unknown mechanism for increased SNCA expression in PD.
In addition to promoter polymorphisms, epigenetic modification can alter downstream gene expression. Epigenetic regulation includes histone modification and DNA methylation, of which CpG island methylation can be gene-specific; in several different cancers, CpG methylation inhibits binding of the transcription machinery, causing silencing of a specific oncogene, which leads to carcinogenesis [11,12,13]. In central nervous system disorders, CpG methylation has been associated with psychiatric disorders, such as autism and schizophrenia [14].
We sought to identify CpG islands in the SNCA gene, wherein methylation status was associated with alpha-synuclein expression. For this purpose, we had to find cell lines which express endogenous alpha-synuclein at sufficient levels to support a comparative analysis. To our surprise, we confirmed that nonneuronal 293 cells express alpha-synuclein protein, and expressed dopamine receptor genes DRD1 and DRD2. SNCA gene expression is affected by cell stress, such as serum deprivation, Interleukin-1b [15], Lipopolysaccharide (LPS) [16], nerve growth factor (NGF) [17], basic fibroblast growth factor (bFGF) [18], 1methyl-4-phenylpyridine (MPP+) [19], and dopamine [20]. By using these potential alpha-synuclein up-regulators, we have identified an SNCA CpG island that controls in vitro alphasynuclein expression, and then analyzed postmortem brain tissues to find PD-specific SNCA promoter demethylation in the substantia nigra.
Bisulfite conversion was performed with the EpiTect Bisulfite kit following the manufacturer's ''low-concentration'' protocol (QIA-GEN). For each samples, 100ng of genomic DNA was used and incubated at 95uC for 5min, 60uC for 25min, 95uC for 5min, 60uC for 85min, 95uC for 5min and 60CuC for 175min. Converted product was purified following the instructions of the Epitect Bisulfite kit. One mL from the final 20 mL elute was PCR-amplified with LA Taq (Takara) following the manufacturer's mixture condition and 30 PCR cycles of 98uC 10sec, 55.5uC (for CpG-1) or 63uC (for CpG-2) 30sec, 68uC 60sec. The PCR product was cloned to pCR9-TOPO vector (Invitrogen). At least 20 clones from each experiment were sequenced by M13 reverse or M13 forward (220) primers. There are 21 CpH sequences in the region. We used sequence results up to 3 unconverted CpHs for further analysis. For each samples, patterns of methylation and unconverted CpHs were used to detect similar clones. Up to four overlapping clones were allowed for measuring CpG methylation. Data analysis and methylation patter figures were generated using QUMA software [21].
Methylated-CpG Island Recovery Assay. A MethylCollector kit (Active Motif, Carlsbad, CA, USA) was used to perform the methylated-CpG island recovery assay (MIRA). The experiment was performed as instructed by the manufacturer. The methylated DNA target fragment was detected with PCR primers 59-AGCTGCGTTTGGCAAATAAT-39 and 59-GCC-CAAGTGTGATGGTTCTAA-39.

Dual-Luciferase assay
CpG-2 was amplified by primers 59-TGTGGAGAAGCA-GAGGGACT -39, 59-ACAGGTTGATGGTGGAAAG -39 and cloned into pMOD shLuc (InvivoGen, San Diego, CA, USA) which encode firefly luciferase and the entire plasmid was demethylated in JM110 (Stratagene, La Jolla, CA, USA). In vitro CpG methylation was performed by incubation with M.SssI (New England Biotechnology, Cambridge, MA, USA) for 1 h at 37uC. Methylated and non-methylated plasmids were transfected into 293 cells with pGL4.74 [hRluc/TK] (Promega) to use sea pansy luciferase as an internal control calibrator. Dual-luciferase activity was measured using the Pick&gene luciferase detection kit (Toyo B-Net co, Tokyo, Japan) on a plate reader Phelios AB-2350 (ATTO, Tokyo Japan). Firefly activity was normalized by sea pansy activity to calibrate background and transfection efficiency.

Postmortem brain
Postmortem brains were obtained at University of Tokyo Hospital, University of Tsukuba Hospital, and National Hospital Organization Tokyo Hospital, with written consent from patients' families and approval of the ethical committees of each facility. Specimens were maintained at 280uC until analysis. Pathological diagnosis was made by neuropathologists using hematoxylin-eosin, Nissl, silver staining together with immunostaining by phosphorylated tau (AT8, Innogenetics) and phosphorylated alphasynuclein antibodies (psyn#64, Wako, Japan) following the established criterias [22,23].

Statistic analysis
Statistical analysis was performed with the aid of Instat3 software (Graphpad Software, Inc, LaJolla, CA). Significance was tested with ANOVA and Student's t-test.

Dopamine alters SNCA CpG island methylation and enhances alpha-synuclein expression
Our first goal was to identify the region of the SNCA CpG island responsible for transcription regulation. We tested various human cell lines for endogenous alpha-synuclein expression. Immunoblotting revealed that 293 cells, although not neuronal, expressed a significant amount of alpha-synuclein. In fact, the expression in 293 cells was considerably greater than in SH-SY5Y cells, which are often considered to be dopaminergic; thus, we decided to use 293 cells for further analysis.
Using 293 cells, we tested various drugs that have been reported to alter alpha-synuclein expression. Of 6 reagents tested, 1 lipopolysaccharide and dopamine caused a significant increase in the production of alpha-synuclein ( Figure 1B). Because dopamine seemed to enhance alpha-synuclein expression more than the lipopolysaccharide, we chose to analyze its effect on SNCA CpG island methylation.
The SNCA gene has two CpG islands, predicted by CpG detection software ( Figure 1C). One of the CpG islands (CpG-1) is located at the first exon, which is not a coding region for alphasynuclein. The second CpG island (CpG-2) is located in the first intron. We analyzed the methylation status of CpG-1 and -2 by MIRA to test the effect of dopamine. CpG-1 methylation remained unchanged whereas CpG-2 methylation was decreased by the addition of 100 mM dopamine ( Figure 1D). Since MIRA is a robust but not a sensitive method for measuring methylation, we next evaluated methylation with the more sensitive bisulfite sequencing method.
We next tested the correlation between dopamine induction of alpha-synuclein expression and CpG-2 demethylation. Addition of 100 mM dopamine further increased the expression of alphasynuclein mRNA ( Figure 3A). Since alpha-synuclein expression is affected by the cellular stress response [24], we tested whether dopamine caused cellular stress at the experimental concentration, and found that 100 mM dopamine did not alter cellular viability, as measured by the MTS method ( Figure 3B). We conclude, therefore, that demethylation was not caused non-specifically by dopamine toxicity, but is the result of a more direct mechanism. Finally, we tested whether CpG-2 methylation was essential for downstream gene expression. In vitro methylated CpG-2 had a significant negative effect on down-stream gene expression, as measured by luciferase assays ( Figure 3C). Thus, we concluded that CpG-2 methylation was essential for alpha-synuclein expression.

Methylation status of CpG-2 controls alpha-synuclein expression through dopamine receptor signaling
We next tried to understand the underlying mechanism of CpG-2 demethylation. Since 293 cells are not neuronal and there is little knowledge regarding their response to dopamine, we tested 293 cells for expression of dopamine receptors. Quantitative PCR clearly showed that 293 cells expressed DRD1 (D1 receptor) and DRD2 (D2 receptor), at levels comparable to those of dopaminergic SH-SY5Y cells ( Figure 3E). To determine whether demethylation of SNCA CpG-2 was receptor mediated, we added haloperidol, a D2 receptor antagonist, along with dopamine, and found that haloperidol clearly inhibited the effect of dopamine ( Figure 3F). These results strongly suggested that the effect of dopamine on CpG-2 demethylation was a D2 receptor-mediated phenomenon, rather than a direct effect of dopamine on the genome.

Significant demethylation of CpG-2 in the substantia nigra of PD patients
Since the CpG-2 region was associated with increased expression of alpha-synuclein in vitro, we next analyzed the methylation level of this region in the postmortem brain. Twelve brains from PD, Dementia with Lewy Bodies (DLB), and 7 from non-neurodegenerative patients were analyzed. The summary of clinical information and the availability of the brain regions from each case are shown in Tables 1 and 2.
Bisulfite sequence analysis was performed for all available brain regions and the % methylation was plotted against various clinical parameters ( Figures S1, S2, S3). When we compared methylation status of the anterior cingulated cortex and the putamen in controls and PD/DLB patients, there were no significant correlations ( Figure 4A, B, D, E). However, methylation status of the substantia nigra was significantly lower in PD patients than in the controls ( Figure 4F) and was unaffected by age ( Figure 4C). Finally, we plotted % methylation against disease duration and found that in ''normally'' methylated DLB case, the disease duration was shorter than the PD cases ( Figure 5A-C) suggesting extension of Lewy body pathology correlates with CpG-2 hypomethylation.

Identification of CpG-2 as SNCA regulator
We analyzed the effects of several reagents on alpha-synuclein expression in 293 cells and found that dopamine had the most pronounced effect without affecting cell viability. We also tried to screen reagents that altered CpG methylation along with increased alpha-synuclein expression level in SH-SY5Y cells; however, reagents we used did not increase its expression level without affecting cell viability significantly. To avoid the possibility of CpG demethylation by cell death, we decided to use 293 cells instead of SH-SY5Y cells.
Using dopamine as a modulator of SNCA expression, we searched for a CpG island which exhibited altered methylation which correlated with changes in gene expression. Of 2 CpG islands identified in the SNCA gene, we located the second CpG island at the first intronic region, and found that it showed a significant decrease in methylation, associated with induction of alpha-synuclein expression. This was first identified by MIRA method [25] and was confirmed by bisulfite sequencing [26], which yielded comprehensive results.  Table 1. Summary of the clinical information from the postmortem brains. CpG-1 lies from the 59 end of SNCA exon1 throughout the exon, and CpG-2 is located at intron1. Methylated CpGs in promoter regions regulate the expression of downstream genes via inhibition of transcription factor binding. Although the precise regulatory mechanism for SNCA transcription is unknown, transcription factors GATA at CpG-2 ( Figure 2A) [27] and ZSCAN21 [28] at upstream of CpG-2 have been reported to bind at intron1. In addition to this, gamma-synuclein, a close homolog of alpha-synuclein, is reported to have a similar regulatory region [29]; we thus concluded that CpG-2 in intron1 was a methylation dependent regulatory element of SNCA. We found that 293 cells expressed D1 and D2 receptors, which supported the rationale for our investigation of dopaminemediated CpG demethylation in this cell line although they are not neuronal cells. Additional experiments showed that haloperidol, the D2 receptor antagonist, clearly inhibited demethylation by dopamine; thus we concluded that this demethylation was mediated by dopamine receptor D2. D2 receptor activation is coupled with G protein G ai , which increases phosphodiesterase activity. CaMK and the protein kinase C pathway are activated by degradation of cAMP by phosphodiesterase [30]. This signaling pathway may be responsible for CpG-2 demethylation.

CpG-2 is significantly hypomethylated only in the substantia nigra of PD patients
Previous studies have compared alpha-synuclein expression in postmortem brain or peripheral blood from PD patients, with inconsistent results, partly because of artifacts of sample quality and degradation of RNA extracts from postmortem brain tissue [31,32,33]. In contrast to these, CpG methylation is reported to be stable up to 48 hours post mortem [34] and we also confirmed that in mice brain, methylation level of SNCA CpG were stable up to 24 hours ( Figure S5). These reports support the relevance of investigating SNCA CpG methylation in postmortem brains to derive  conclusions regarding gene expression. SNCA CpG-2 methylation levels differed among subjects and regions tested (Figure 4). This result was not surprising, since a report regarding analysis of CpG methylation in various CpGs suggested their diversity by age, sex, and race [35]. However, the result from substantia nigra was quite striking; the methylation level was almost zero in the PD group, whereas in the control group it was almost 100%, suggesting that methylation in the substantia nigra is an important component of PD pathogenesis. For concerns about DNA quality from substantia, nigra, we tested in several ways to assure that those were comparable between controls and PDs ( Figure S4). The demethylatyion of CpG-2 could be occurring in both neurons and glial cells especially because in the late disease course, massive neuronal loss can occur. However, this does not indicate that our findings is unrelated to the direct pathogenesis, since it has widely recognized that alpha-synuclein aggregation are observed also in the astrocytes and oligodendrocytes of PD brains [36,37,38,39]. In neurodegenerative diseases, it is still a mystery that neurons in similar functions start to degenerate simultaneously regardless of their spatial location in the brain. We can speculate that in the early developmental stage, even before the neuronal and the glial cells are differentiated, epigenetical abnormality can occur. After long incubational time, slightly over-expressed alphasynuclein can accumulate enough in the neurons to cause PD, while in the glial cells, efficient protein degradation system or the cell division might help them to prevent the accumulation.
Global CpG methylation decreases with age, and specific genes, such as cancer suppressors, are down-regulated by decreased promoter region CpG island methylation [40]. Our findings suggest that in PD patients, normally occurring SNCA CpG-2 hypermethylation does not occur, causing over-expression of alpha-synuclein and leading to its accumulation which, in turn, causes PD.