Alpha‐synuclein promotes PRMT5‐mediated H4R3me2s histone methylation by interacting with the BAF complex

α‐Synuclein (αS) is a key molecule in the pathomechanism of Parkinson's disease. Most studies on αS to date have focused on its function in the neuronal cytosol, but its action in the nucleus has also been postulated. Indeed, several lines of evidence indicate that overexpressed αS leads to epigenomic alterations. To clarify the functional role of αS in the nucleus and its pathological significance, HEK293 cells constitutively expressing αS were used to screen for nuclear proteins that interact with αS by nanoscale liquid chromatography/tandem mass spectrometry. Interactome analysis of the 229 identified nuclear proteins revealed that αS interacts with the BRG1‐associated factor (BAF) complex, a family of multi‐subunit chromatin remodelers important for neurodevelopment, and protein arginine methyltransferase 5 (PRMT5). Subsequent transcriptomic analysis also suggested a functional link between αS and the BAF complex. Based on these results, we analyzed the effect of αS overexpression on the BAF complex in neuronally differentiated SH‐SY5Y cells and found that induction of αS disturbed the BAF maturation process, leading to a global increase in symmetric demethylation of histone H4 on arginine 3 (H4R3me2s) via enhanced BAF–PRMT5 interaction. Chromatin immunoprecipitation sequencing confirmed accumulated H4R3me2s methylation near the transcription start site of the neuronal cell adhesion molecule (NRCAM) gene, which has roles during neuronal differentiation. Transcriptional analyses confirmed the negative regulation of NRCAM by αS and PRMT5, which was reconfirmed by multiple datasets in the Gene Expression Omnibus (GEO) database. Taken together, these findings suggest that the enhanced binding of αS to the BAF complex and PRMT5 may cooperatively affect the neuronal differentiation process.

a-Synuclein (aS) is a key molecule in the pathomechanism of Parkinson's disease.Most studies on aS to date have focused on its function in the neuronal cytosol, but its action in the nucleus has also been postulated.Indeed, several lines of evidence indicate that overexpressed aS leads to epigenomic alterations.To clarify the functional role of aS in the nucleus and its pathological significance, HEK293 cells constitutively expressing aS were used to screen for nuclear proteins that interact with aS by nanoscale liquid chromatography/tandem mass spectrometry.Interactome analysis of the 229 identified nuclear proteins revealed that aS interacts with the BRG1-associated factor (BAF) complex, a family of multi-subunit chromatin remodelers important for neurodevelopment, and protein arginine methyltransferase 5 (PRMT5).Subsequent transcriptomic analysis also suggested a functional link between aS and the BAF complex.Based on these results, we analyzed the effect of aS overexpression on the BAF complex in neuronally differentiated SH-SY5Y cells and found that induction of aS disturbed the BAF maturation process, leading to a global increase in symmetric demethylation of histone H4 on arginine 3 (H4R3me2s) via enhanced BAF-PRMT5 interaction.Chromatin immunoprecipitation sequencing confirmed accumulated H4R3me2s methylation near the transcription start site of the neuronal cell adhesion molecule (NRCAM) gene, which has roles during neuronal differentiation.Transcriptional analyses confirmed the negative regulation of NRCAM by aS and PRMT5, which was reconfirmed by multiple datasets in the Gene Expression Omnibus (GEO) database.Taken together, these findings suggest that the enhanced Abbreviations ACTL6A/B, actin-like 6A/B; BAF, BRG1-associated factor; BDNF, brain-derived neurotrophic factor; CE, cytosol extraction buffer; ChIP-seq, chromatin immunoprecipitation sequence; DMEM, Dulbecco's modified Eagle's medium; Dox, doxycycline; DPF1, double PHD finger 1; esBAF, ES cell BRG1-associated factor complex; FCS, fetal calf serum; FDR, false discovery rate; GEO, gene expression omnibus; GO, gene ontology; H3R8me1, histone H3R8 mono-methylation; H3R8me2a, histone H3R8 asymmetric di-methylation; H3R8me2s, histone H3R8 symmetric di-methylation; H4R3me1, histone H4R3 mono-methylation; H4R3me2a, histone H4R3 asymmetric di-methylation; H4R3me2s, histone H4R3 symmetric di-methylation; HA, hemagglutinin; HDAC, histone deacetylase; logFC, log fold change; MCL, Markov cluster algorithm; nanoLC-MS/MS, nano liquid chromatography-tandem mass spectrometry system; nBAF, neuronal BRG1-associated factor complex; NCBI, The National Center for Biotechnology Information; NE, nuclear extraction buffer; npBAF, neural progenitor BRG1-associated factor complex; NRCAM, neuronal cell adhesion molecule; PD, Parkinson's disease; PHF10, PHD finger protein 10; PRMT5, protein arginine methyltransferase 5; RA, retinoic acid; RT-qPCR, quantitative reverse transcript real-time PCR; SMARC, SWI/SNF-related matrix-associated actin-dependent regulator of chromatin subfamily; SpC, spectral counts; SS18L1, SS18 like protein 1; SWI/SNF, SWItch/sucrose nonfermentable; TPM, transcripts per million; TSS, transcription start site; aS, alpha-synuclein.

Introduction
a-Synuclein (aS) is a small protein composed of 140 amino acids encoded by the SNCA gene on chromosome 4q22 [1].aS is abundantly expressed in neuronal cytosol, but it is also present in the plasma membrane, mitochondria, cytoplasmic vesicles, and nucleus [2].The physiological function of aS has received significant attention at presynaptic nerve terminals, where it is known to regulate synaptic vesicle cycling [3].With respect to disease, aS is recognized as a key molecule in the pathogenesis of Parkinson's disease (PD), because it causes rare familial forms of PD (i.e., PARK1 and PARK4) and is the main component of Lewy bodies, a pathological hallmark of sporadic PD [4].The role of aS multiplication in familial PD provides a rationale for overexpressing aS to recapitulate PD pathology [4].
In recent years, the study of nuclear aS function has received increased attention [5].Indeed, aS interacts with nucleic acids and nuclear proteins, thereby affecting epigenetic modifications [6].Under physiological condition, histone acetylation is tightly controlled by lysine acetyltransferases and histone deacetylase (HDAC); however, this regulation is disrupted in neurodegenerative diseases, such as PD [7].In support of this, global HDAC inhibitors counteract aS-mediated neurotoxicity in cell and Drosophila PD models [8,9].In yeast models, aS is associated with reduced histone H3K36 di-methylation, together with mild changes in other histone modifications [10].We previously demonstrated that overexpression of aS activates euchromatic histone lysine methyltransferase 2 in retinoic acid (RA)-treated SH-SY5Y cells, which induces histone H3K9 mono-and di-methylation [11].Although many studies, including ours, have proposed the epigenomic modification potential of aS, detailed molecular interactions between aS and nuclear proteins remain to be understood.
The SWItch/Sucrose Non-Fermentable (SWI/SNF) family represents a fundamental histone remodeling complex that is conserved in eukaryotes [12].The mammalian homolog of SWI/SNF is known as the BAF (BRG1-associated factors) complex, which consists of approximately 15 molecular subunits [13].The neural progenitor BAF complex (npBAF), composed of the PHD finger protein 10 (PHF10) and actin-like 6A (ACTL6A) subunits, is required for self-renewal and proliferation in these cells [14].In contrast, postmitotic neurons lack PHF10, ACTL6A, and SS18; however, they have double PHD finger 1 (DPF1), ACTL6B, and SS18 like protein 1 (SS18L1) [13,14].This subunit switching is necessary for the transition from proliferating neural stem/progenitor cells to post-mitotic neurons [14].This complex maintains cell-type specific gene expression through the ATPase activity of its subunit, SWI/SNF-related matrixassociated actin-dependent regulator of chromatin subfamily A member 2/4 (SMARCA2/4) [12].It determines the direction of differentiation into various tissues, including the nervous system [12].With respect to disease, several subunits of the BAF complex have been implicated in the pathogenesis of PD.For example, the in silico prediction of PD-associated genes followed by in vivo Drosophila genetic screening identified SMARCA4.Subsequent knockdown of the fly homolog of SMARCA4 in dopaminergic neurons extended the lifespan of LRRK2 or SNCA transgenic Drosophila [15].In addition, altered expression of SMARCE1, another subunit of the BAF complex, was observed in a toxin-induced PD model using 6hydroxydopamine [16].
To better understand the function of aS in the nucleus, we screened nuclear extracts prepared from HEK293 cells for proteins that interact with aS by coimmunoprecipitation and mass spectrometry and further analyzed candidate proteins by in silico analysis.The results identified BAF complex components among the groups of binding proteins.In subsequent experiments, we found that aS overexpression caused dysregulation of the BAF complex in the neuronal differentiation process.

Identification of nuclear proteins that interact with aS
As an initial step in exploring the epigenetic impact of aS, we screened for nuclear proteins that interact with aS using cultured cells (Fig. 1A).Specifically, HEK293 cells stably expressing hemagglutinin (HA)-tagged aS were lysed in detergent-free buffer supplemented with nucleoprotectants and crude nuclear lysates were recovered from the insoluble fraction.Intermediate nuclear lysates contain intact nucleic acids, allowing proteins to condense around DNA.The nucleic acids were then digested with a nuclease cocktail (Fig. 1B).aS-interacting proteins were immunoprecipitated using an HA antibody or control mouse IgG, followed by detection using SDS/PAGE and silver staining.A thick band of 15 kDa corresponding to HA-aS was identified in all samples following immunoprecipitation (Fig. 1C).Of the nuclear extracts prepared by two different methods, the nuclear fraction with a superior background clearance was selected for further analyses (Fig. 1C).After immunoprecipitation, proteins from mock or HA-aS-expressing cells were analyzed by a nano liquid chromatography-tandem mass spectrometry system (nano LC-MS/MS).The numbers of proteins with spectral counts (SpC) of 5 or greater included 808 in mock cells and 1467 in HA-aSexpressing cells.Of these, 366 proteins exhibited at least a 4-fold increase in SpC compared with mock cells (Fig. 1D).Based on gene ontology (GO) cellular component annotation (GO: 0005634 nucleus), 229 proteins were classified as common nuclear proteins (Fig. 1E, Tables S1 and S2).Of note, known aSinteracting proteins, such as DNA-methyltransferase 1, casein kinase 1 alpha 1, and high mobility group box 1, were also identified (Table S1) [17][18][19].

Interactome analysis suggests a linkage between aS and the BAF complex
For the functional classification of aS-interacting proteins, physical interaction-based clustering of 229 nuclear proteins and 137 non-nuclear proteins was performed by Markov cluster algorithm (MCL) using the String protein-protein interaction database (Fig. 2A, Fig. S1A) [20].Eight clusters were identified and annotated by the GO cellular components database.Of these, Cluster III ("npBAF") showed the highest binding strength value."npBAF" is a variant of the BAF complex and its biological behavior depends on its components.The cluster III protein names were then further analyzed by GO enrichment analysis."Neuronal cell specific BAF complex (nBAF)," or "npBAF," which are associated with neuron or precursor cells, exhibited a lower false discovery rate (FDR), whereas immature "ES cell specific BAF complex (esBAF)" annotation did not show significance (Fig. 2B).Mitochondria-related GO annotations were listed primarily in non-nuclear proteins, suggesting that the non-nuclear fraction was primarily of mitochondrial origin (Fig. S1A-C).

Functional correlation between aS overexpression and BAF complex
Interactome analysis of aS-interacting nucleoproteins suggested an association between aS and the npBAF complex.The BAF complex plays a fundamental role in epigenomic regulation and dysregulated BAF subunits cause transcriptional alterations.Based on these findings, we evaluated the transcriptional response following aS overexpression and modulation of the BAF complex.The corresponding RNA-seq datasets were obtained from Gene Expression Omnibus (GEO) datasets (Fig. 3A).After calculating the log fold-change (logFC) for each gene as overexpressed versus control, or knockout versus control, the correlation coefficient between aS overexpression and altered BAF activity-induced logFC was calculated.The results indicated that aS-overexpressing cells were negatively correlated with SMARCC2-KO and a portion of SMARCA4-expressing cells (Fig. 3B).Whole expressing genes from representative pairs showing significant correlations are indicated on the scatter plots (Fig. 3C).A logFC heatmap was generated with whole expressing genes (Fig. 3E).The results suggest that some of the transcriptome changes associated with aS overexpression may involve epigenetic pathways resulting from dysregulation of the BAF complex.

aS enhances PRMT5-mediated histone arginine methylation by interacting with BAF complex
To further clarify the regulatory effect of aS on the BAF complex, we examined changes in the conformation of the BAF complex before and after neuronal differentiation in SH-SY5Y cells, in which aS expression was inducible.Structurally, SMARCC1 and SMARCC2 form the core module that is essential for all BAF complexes.These fundamental BAF components were detected in the peptide screen as nucleoproteins interacting with aS and their binding was confirmed in co-immunoprecipitation experiments (Fig. 4A).More importantly, we also found an interaction between aS and protein arginine methyltransferase 5 (PRMT5), a nuclear protein detected in our screen that is known to interact with the BAF complex (Figs 2A and 4A) [21].
The npBAF-nBAF transition, an important step in neuronal differentiation, is characterized by the switch of PHF10, ACTL6A, and SS18 to DPF1, ACTL6B, and SS18L1 [22].To determine the role of aS during the BAF complex maturation process, we used SH-SY5Y cells, in which aS expression was inducible after RA/brain-derived neurotrophic factor (BDNF)mediated neuronal differentiation (Fig. 4B).The SMARCC2 interacting protein was used as a readout for the transition of the BAF complex.Prior to differentiation, SMARCC2 interacted with other BAF components, including SMARCC1, SMARCA4, and PHF10 (Fig. 4C).After differentiation with BDNF, binding of SMARCC2 to PHF10 was attenuated, whereas binding to DPF1 occurred.Because the transition from PHF10 to DPF1 is associated with the npBAF-nBAF transition [14], this model appeared suitable for assessing the maturation process of the BAF complex.Interestingly, the induction of aS overexpression clearly suppressed the PHF10-DPF1 substitution during neuronal differentiation (Fig. 4C).
PRMT5, which we identified as an aS-interacting protein, is a methyltransferase that acts as a histone remodeling factor together with the BAF complex [21,22].To support this, immunoprecipitation using this cellular system confirmed a direct interaction between PRMT5 and the core components of the BAF complex (Fig. 4D).It should be emphasized that the BAF-PRMT5 interaction was enhanced following aSinduction (Fig. 4D).Because the impact of PRMT5 in epigenetic alterations is well-documented as symmetrical di-methylation at H3R8 (H3R8me2s) and H4R3 (H4R3me2s), we examined the methylation status of H3R8 and H4R3 in cells under the following conditions: na€ ıve, neuronal differentiation, and neuronal differentiation under aS expression.The results indicated that H3R8me2s expression was unchanged under all conditions.H4R3me2s expression was reduced after differentiation by BDNF; however, this change was counteracted in cells overexpressing aS (Fig. 4E).Reciprocal changes H4R3me2s levels against intermediate H4R3 methylation (H4R3me1) confirmed the reliability of the experiment (Fig. 4E).

BAF-PRMT5 complex formation under increased aS expression alters NRCAM expression
Because the global level of H4R3m2s was affected under aS overexpression, we examined the distribution of histone methylation and the target genes.To identify genes targeted by the BAF-PRMT5 complex via H4R3m2s, Chromatin immunoprecipitation sequence (ChIP-seq) was performed using an H4R3me2s antibody with the aforementioned cell groups.The average number of peaks identified was 8674 in the na€ ıve group, 5865 in the neuronal differentiation group, and Fig. 2. Interactome and enrichment analysis of nuclear proteins interacting with aS.In silico analysis identified a physical interaction between aS and the BAF complex.(A) STRING interactome map of 229 identified nuclear proteins.The annotations with the most significant values in the GO cellular components database were assigned to the clusters.A dotted line indicates the border of the clusters.(B) Seven proteins classified into cluster III underwent a GO enrichment analysis for cellular components."npBAF" annotation showed the lowest FDR followed by "SWI/SNF complex," and "nBAF."nBAF, neuronal BRG1-associated factor; npBAF, neural progenitor BRG1associiated factor; SWI/SNF, SWItch/sucrose non-fermentable.8045 in the aS-induction group (Fig. 5A).More peaks were identified in the na€ ıve and aS-induced groups compared with the neuronal differentiation group, which is consistent with the results of western blot analysis (Fig. 4E).The ratio of peak annotations exhibited a similar distribution, with intron and intergenic regions densely occupied and less so for promoter regions (Fig. 5B).At the transcription start site (TSS), all three groups exhibited a low density of H4R3me2s (Fig. 5C).
Indeed, some peaks were flanked by the TSS (Table 1, Fig. 5E).In support of the influence of these histone modifications on NRCAM gene, we confirmed the NRCAM expression in SH-SY5Y cells with PRMT5inhibition and -overexpression.As we expected, treatment with GSK591, a potent PRMT5 inhibitor, significantly increased the transcript level of NRCAM (Fig. 5F).Besides, the overexpression of a methyltransferase-dead mutant of PRMT5 showed higher transcript level of NRCAM than that of wildtype PRMT5 (Fig. 5G).

Regulation of NRCAM expression by aS and PRMT5 is confirmed by multiple RNA-seq datasets
Thus far, we have shown that NRCAM expression is most significantly affected by aS induction through BAF-PRMT5 complex formation.To determine whether transcriptomic changes in NRCAM are similarly observed in other experimental models, we searched the GEO database to obtain RNA-seq datasets for cells in which PRMT5 was knocked down or pharmacologically inhibited (Fig. 6A).Prior to the analysis, datasets with transcripts per million (TPM) of NRCAM lower than 1 were excluded.Unfortunately, we could not find any BAF component-related datasets with sufficient expression of NRCAM, but we did identify several aS and PRMT5-containing datasets for further examination.Consistent with our results, one of two cellular models overexpressing aS showed negatively regulated NRCAM expression (Fig. 6B, upper panel).Conversely, three of five PRMT5 knockdown and four of eight PRMT5inhibited cell models showed a significant increase in NRCAM expression (Fig. 6B).None of the data analyzed was found to be inconsistent with our results.

Discussion
As the name implies, synuclein was found in the nucleus and the synaptic membrane when it was first identified in the electric organ of Torpedo californica [23].In the human brain, misfolded, phosphorylated aS was detected in the nucleus as well as the cytoplasm of neural and glial cells in synucleinopathies, such as PD, dementia with Lewy bodies, and multiple system atrophy [24,25].Although aS does not have a canonical nuclear localization signal, its small molecular size enables it to pass through nuclear membrane pores and does not require a specific transport carrier for nuclear localization [26].The physiological or pathological function of aS in the nucleus has been reported in several studies, but remains poorly understood [27].In contrast, the transcriptional profiles of induced pluripotent stem cell-derived dopaminergic cells are affected by SNCA triplication, and nigral cells from PD patients show distinct transcriptional products compared with healthy controls, suggesting that aS contribute to the modulation of epigenetic status [28,29].
In this study, we focused on how nuclear aS modulates transcriptional profiles as well as the final output of epigenetic alterations.The most important difference between the present study and that of previous methods to identify aS-interacting proteins is the use of nucleoproteins as the starting material [30][31][32].This unique approach resulted in the discovery of numerous novel aS-interacting proteins in the nucleus.Furthermore, we identified eight clusters after the calculation with MCL, and cluster III was the histone remodeling complex, designated npBAF.Moreover, using public databases, we found a strong correlation between transcriptional changes resulting from aS overexpression and dysregulation of the BAF complex.These findings prompted us to explore the epigenetic role of aS in a BAF complex-dependent manner.The subunit switching system of the BAF complex has been studied primarily in neural stem/progenitor cells; however, the same switching phenomenon was successfully reproduced in RA/BDNF-treated, differentiated SH-SY5Y cells.Importantly, we observed that stoichiometric substitution of the components of the BAF complex after neuronal differentiation is abrogated by elevated aS expression.In addition, PRMT5, which can work together with BAF complex [33], was identified as an aS-binding partner.The interaction between PRMT5 and the BAF complex was also recapitulated in our cell model, in which the interaction of PRMT5-SMARCC2, a subunit of the BAF complex, was strengthened by increased expression of aS (Fig. 7).PRMT5 is a type II methyltransferase that catalyzes the symmetrical di-methylation of arginine residues [34].Substrates of BAF complex-related PRMT5 are histone H3R8 and/or histone H4R3 [21,35,36].Substrate specificity may be cell-type-dependent and only H4R3 was directed in our cellular model.The differentiation protocol for SH-SY5Y cells used in this study resulted in a reduction of H4R3me2s.Similar H4R3me2s regulation was reported in rat dopaminergic PC12 cells, in which the cells was maintained by a "writer" PRMT5 and an "eraser," PR Domain Zinc finger protein 4. Knockdown of PRMT5 resulted in precocious neuronal differentiation [37].These results suggest that the higher the occupancy of H4R3me2s on certain genes, the more neuronal differentiation is suppressed.It should be emphasized that elevated aS observed in the present study could counteract the reduction of H4R3me2s during neuronal differentiation.The aligned reads obtained from ChIP-seq with H4R3me2s antibody displayed the dip.This distribution is consistent with that of a previous study showing that the density in the TSS was lower compared with neighboring regions [38].Because the density of H4R3me2s in the TSS of aS-induction group was higher than that of the na€ ıve and neuronal differentiation groups, there may be genes specifically affected by aS induction.ChIP-seq analyses identified numerous genes with significant peaks and candidates, but NRCAM showed the largest difference, with one of the peaks harbored near its TSS.Although H4R3me2s is not potent enough to exert repressive histone remodeling, the density of H4R3me2s in the TSS correlates with gene expression levels [38].Thus, a small, but significant reduction of NRCAM mRNA levels by aS induction is consistent with altered nucleosome structure due to occupation by H4R3me2s.
NRCAM is a cell adhesion molecule belonging to the immunoglobulin superfamily and is involved in axon growth and guidance during nervous system development [39].The importance of NRCAM in regulating neuronal development is further supported by the fact that NRCAM knockout mice show abnormal behaviors reminiscent of autism and impulse control disorder [40,41].Of note, with respect to PD pathogenesis, a whole blood RNA-seq study conducted in a Turkish PARK4 pedigree showed a marked decrease in NRCAM expression levels in the prodromal phase [42].Although this result was not mentioned in the main text of the paper, it indicates that a similar phenomenon to the negative regulation of NRCAM by induction of aS expression shown in our data may occur in human patients.
In summary, we uncovered a previously unknown functional interaction between aS and the BAF complex.Increased aS expression may interfere with the maturation process of the BAF complex into a neuronal form.In addition, reinforcement of the PRMT5-BAF complex interaction by an excess amount of aS may affect the reduction of NRCAM, which is important for the regulation of neuronal function through a PRMT5-mediated repressive mark of H4R3me2s.These novel findings will advance our understanding of the functional role of aS under physiological or pathological situations and provide a fresh insight for future neuroscience research, including studies of the pathophysiology of devastating neurodegenerative diseases.

Cell culture and differentiation
Human neuroblastoma cell lines SH-SY5Y (ATCC CRL-2266; RRID: CVCL_0019) were obtained from American Type Culture Collection (ATCC, Manassas, VA, USA).Human Embryonic Kidney cell lines HEK293 (RCB1637; RRID: CVCL_0045) were obtained from RIKEN BRC Cell BANK (Ibaraki, Japan).HEK293 cells stably expressing HA-tagged human wild-type aS [43] and tetracyclineregulated inducible human wild-type aS expressing SH-SY5Y cells [44] were cultured in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal calf serum (FCS) (HyClone Laboratories, Logan, UT, USA) at 37 °C in 5% CO 2 in humidified air.Cell lines were authenticated during the course of this study by morphological analysis using optical microscopy and growth curve determination, as suggested by ATCC.Cell lines were tested to confirm that they were mycoplasma-free.G418 (InvivoGen, San Diego, CA, USA) was added to the medium to maintain aS-expressing HEK293 cells.To differentiate SH-SY5Y cells, the cells were incubated in 3% FCS medium supplemented with 10 lM RA (Sigma, Fig. 7. Graphical summary of the effect of aS expression on BAF complex, PRMT5, and its downstream targets.During neuronal differentiation, the BAF complex forms a neuron-specific assembly and regulates neuronal gene expression.In the presence of aS, the neural progenitor BAF (npBAF) cannot mature into neuronal BAF (nBAF).In this context, the potentiated BAF complex-PRMT5 interaction enhances H4R3me2s repressive histone marks, resulting in the negative regulation of downstream neuronal genes.Abbreviations used are: DPF1, Double PHD fingers 1; nBAF, neuronal BAF complex; npBAF, neural progenitor BAF complex; PHF10, PHD finger protein 10; PRMT5, protein arginine methyltransferase 5; SMARCA4, SWI/SNF related, matrix associated, Actin-dependent regulator of chromatin, subfamily A, member 4; SMARCC1, SWI/SNF related, matrix associated, Actin-dependent regulator of chromatin, subfamily C, member 1; SMARCC2, SWI/SNF related, matrix associated, Actin-dependent regulator of chromatin, subfamily C, member 2; TSS, transcription start site.

Silver stain and mass spectrometry
Nuclear lysates (2000 lg) were incubated with anti-HA antibody-conjugated agarose beads (Ezview TM ; Sigma) on a carousel at 4 °C overnight.The captured proteins were eluted by adding 50 lg of HA peptide (MBL, Tokyo, Japan).The quality of the samples was evaluated by SDS/PAGE followed by silver staining (FUJIFILM Wako Pure Chemical Co., Osaka, Japan).For mass-spectrometry, samples purified by SDS/PAGE were in-gel digested with trypsin and the resulting peptides were analyzed by mass spectrometry using nanoLC-MS/MS, which was performed by Filgen, Inc (Aichi, Japan).The resulting proteins were classified as "Nuclear protein" and "Non-Nuclear protein" based on GO annotation "GO:0005634 nucleus" using BIO- MART 2.52.0 in R. Interactome and functional enrichment analyses were performed for the categorized proteins using STRING 11.5 with the basic settings: Network type: "physical subnetwork type," meaning of network edges: "confidence," active interaction sources: "Textmining," "Experiments," "Detabase," minimum required interaction score: "highest required confidence (0.900)," max number of interactors to show: "none," "none" [45].Cluster analysis was performed using the MCL with a 3.0 inflation parameter.The binding strength value for each cluster was calculated using the log 10 (observed/expected) formula in STRING 11.5.

Immunoprecipitation
Nuclear fraction lysates containing 750 lg of protein were incubated overnight on a carousel at 4 °C with 2 lg of antibody, followed by incubation with 40 lL of Pierce protein G magnetic beads (Thermo Fisher Scientific, Waltham, MA, USA) for 1 h.After washing four times with NEN buffer, the protein complexes were eluted in 29 Laemmli buffer and subjected to western blot analysis.

Western blot analysis
Histone-enriched lysates were prepared as previously described [11].Immunoprecipitated samples or histone enriched lysates were used for western blot analysis.After lysate preparation, protein concentrations were determined using the BCA protein assay kit (BioRad, Hercules, CA, USA).Lysates were dissolved in Laemmli buffer, electrophoresed on polyacrylamide gels, and transferred to polyvinylidene fluoride membranes (BioRad).The membranes were blocked in 5% skim-milk containing TBS-Tween 20 and incubated with primary antibody in Western blocking reagent (Roche Applied Science, Penzberg, Germany) at 4 °C overnight.After incubation with HRP-conjugated secondary antibody for 1 h at room temperature, the proteins were detected using the Immobilon Western chemiluminescent HRP substrate (Millipore) using ChemiDoc TM MP imaging system (BioRad).

Chromatin immunoprecipitation sequencing
ChIP-seq was performed according to a previously described method with slight modification [11].After lysing the cells in nuclear lysis buffer, samples were sonicated with a protocol adjusted to obtain DNA fragments of 150-200 bp length.Immunoprecipitation was done using 40 lL protein A magnetic beads (Thermo Fisher, Waltham, MA, USA) with 4 lg H4R3me2s antibody (Epigentek) or control rabbit IgG.The obtained DNA fragments were comprehensively sequenced using Illumina HiSeq2000 (outsourced to Hokkaido System Science, Sapporo, Japan).The reads were aligned to a human reference genome (CRCh38) using bowtie1.H4R3me2s peaks for each sample were identified by MACS2 using a P value cutoff of 0.01, after removing the blacklist and Grey-ListChIP annotation regions [46].Group comparisons were performed using data merged with SAMTOOLS.The peak distributions were analyzed with Chipseeker.Read density of the TSS was analyzed using DEEPTOOLS.

RNA extraction and RT-qPCR
Total RNA was isolated from SH-SY5Y cells using the RNeasy Mini kit (Qiagen, Venlo, the Netherlands) and reverse-transcribed with oligo hexamer and random primers (1 : 1) using the Primescript TM II 1st strand cDNA synthesis kit (Takara).RT-qPCR was performed using Universal SYBR Green Supermix (BioRad) and the threshold cycles were calculated with a Step one plus real-time PCR system (Applied Biosystems, Waltham, MA, USA).Individual C t values of the target genes obtained from biological replicate samples under each condition were normalized to the C t values of the corresponding internal control.Fold-change was calculated by the DDC t method.Primer pairs are listed in Table S3.

Statistics
Functional enrichment analysis of aS-interacting proteins was conducted on STRING 11.5 and FDR was calculated with Benjamini-Hochberg procedure [20].GRAPHPAD PRISM software (Boston, MA, USA) was used for histone immunoblotting quantification and RT-qPCR, in which t-test or two-way ANOVA test with a Dunnett's multiple comparisons post-hoc test was applied.R software was used for statistic calculation of RNA-seq analysis, in which correlation coefficients and two-sided P-values were calculated using logFC values for each gene with Spearman method.H4R3me2s peaks for each sample were identified by MACS algorism using a P-value cutoff of 0.01.

Fig. 1 .
Fig. 1.Comprehensive screening of proteins interacting with aS in the nucleus.(A) Schematic representation of the procedure for nuclear fractionation, immunoprecipitation, and nanoLC-MS/MS analysis.(B) Nuclear and cytosolic fractionation by two-step centrifugation in applicable buffers.(C) Representative image of the silver-stained gel following HA-immunoprecipitation and SDS/PAGE of HA-aS stably expressing HEK293 cells and control.Similar images were obtained in triplicate.HA-aS was effectively immunoprecipitated in each fraction.Nuclear fraction lysed in NE2 buffer shows specific bands compared with the cytosolic fraction and untreated HEK293 cells.(D) aS-interacting proteins identified by two-step filtering: over 5 SpC in HA-aS samples and at least a 4-fold increase in SpC compared with untreated cells.(E) Classification of 366 aS-interacting proteins as "nuclear protein" and "non-nuclear proteins" according to GO annotation.CE, cytosolic extraction buffer; HA, hemagglutinin; NE, nuclear extraction buffer; SpC, spectral counts; aS, alpha-synuclein.

Fig. 3 .
Fig. 3. Systematic and comparative analyses of RNA-seq datasets of aS-overexpressing cells.Comparative analyses of RNA-seq datasets implied a functional interaction between aS and the BAF complex.(A) Flow diagram showing the method to obtain the relevant datasets from the NCBI/GEO database.RNA-seq datasets obtained from human-derived cultured cells related to aS-overexpression, SMARCC1knockdown, SMARCC2-knockdown, and SMARCA4 overexpression.(B) Correlation analysis between aS-overexpressed cells and BAF component-modulated cells.The numbers shown in the figure are Spearman's correlation coefficients of the logFC for all expressed genes.Several datasets showed a high correlation despite the cell-type difference.(C, D) LogFC scatter plots of expressed genes.Significant negative correlations (P < 0.0001) were calculated in the comparison of aS-overexpressing SH-SY5Y cells with SMARCC2-KO HAP1 cells (C) and ATPase-deficient-SMARCA4-expressing C33A cell (D).(E) LogFC heatmap of expressed genes.Non-hierarchal clustering procedure revealed an obvious reciprocal pattern in aS-overexpressing datasets and BAF conformational change datasets.logFC, log fold change; SMARCA4, SWI/SNF related, matrix associated, actin-dependent regulator of chromatin, subfamily A, member 4; SMARCC1, SWI/SNF related, matrix associated, actin-dependent regulator of chromatin, subfamily C, member 1; SMARCC2, SWI/SNF related, matrix associated, actindependent regulator of chromatin, subfamily C, member 2; WT, wild-type.

Fig. 4 .
Fig. 4. Interaction between aS, BAF complex components, and PRMT5 demonstrated by co-immunoprecipitation. (A) Representative image of the immunoprecipitation assay in HEK293 cells stably expressing HA-conjugated aS showing the interaction between aS, SMARCC1, SMARCC2, and PRMT5.(B) SH-SY5Y cells under three experimental conditions were used in the experiment: na€ ıve, neuronal differentiation, and neuronal differentiation with aS-induction.Neuronal differentiation was induced by a combination of RA and BDNF.aS expression was induced by a tetracycline-regulated transcriptional activation system.(C) Representative image of the immunoprecipitation assay using SMARCC2 antibody in SH-SY5Y cells of the three groups.Neuronal differentiation treatment potentiates the interaction between SMARCC2 and DPF1, specific component of nBAF, besides the interaction with npBAF component PHF10 is reduced.In the presence of aS, PHF10-DPF1 transition is not observed.(D) Representative image of the immunoprecipitation using PRMT5 antibody showing the enhancement of the interaction between PRMT5 and the BAF complex by aS induction.(E) Global histone H3R8 and H4R3 methylation levels in the three groups.H4R3me2s reduction is specifically reduced in neuronally differentiated cells.Concurrently, H4R3me1, a precursor of H4R3me2, is increased.These changes are recovered by aS induction.(A, C, D) Similar images of these assays were obtained in triplicate.(E) For immunoblot quantification, a two-way ANOVA test with a Dunnett's multiple comparisons post-hoc test was applied.*P < 0.05 against "na€ ıve", n = 4.All graphs were presented as mean AE SEM.BDNF, brain-derived neurotrophic factor; Dox, doxycycline; DPF1, double PHD fingers 1;HA, hemagglutinin; H3R8me1, histone H3R8 mono-methylation; H3R8me2a, histone H3R8 asymmetric dimethylation; H3R8me2s, histone H3R8 symmetric di-methylation; H4R3me1, histone H4R3 mono-methylation; H4R3me2a, histone H4R3 asymmetric di-methylation; H4R3me2s, histone H4R3 symmetric di-methylation; IP, immunoprecipitation; MW, molecular weights; nBAF, neuronal BRG1-associated factor; npBAF, neural progenitor BRG1-associated factor; PHF10, PHD finger protein 10; PRMT5, protein arginine methyltransferase 5; RA, retinoic acids; SMARCC1, SWI/SNF related, matrix associated, actin-dependent regulator of chromatin, subfamily C, member 1; SMARCC2, SWI/SNF related, matrix associated, actin-dependent regulator of chromatin, subfamily C, member 2; SMARCA4, SWI/SNF related, matrix associated, actin-dependent regulator of chromatin, subfamily A, member 4.

Fig. 5 .
Fig. 5. Identification of target genes undergoing H4R3me2s modification by ChIP-seq analysis using aS-expressing SH-SY5Y cells.(A) Number of H4R3me2s peaks determined by MACS pipeline.A fewer number of peaks are annotated in the neuronal differentiation group compared with the na€ ıve and aS-induction groups, n = 3. (B) Peak distribution in the three groups.Most of the peaks are harbored in intron and intergenic lesions and a few peaks are localized around the transcription start site (TSS), n = 3. (C) Read density plot in the TSS AE 5.0 kb region.Low density H4R3me2s is shown in the TSS compared with the neighboring regions.aS-induction group have the highest methylation level in the TSS, n = 3. (D) RT-qPCR analysis of potential target genes.The mRNA levels of NRCAM, TMEM60, and THAP5 are significantly reduced by aS-induction.A two-way ANOVA test with a Dunnett's multiple comparisons post-hoc test was applied.*P < 0.05 against "Neuronal diff.",n = 4. (E) The Integrative Genome Viewer views of ChIP-seq showing H4R3me2s occupancy at the promoter region of NRCAM gene.(F, G) RT-qPCR analysis of the NRCAM genes in SH-SY5Y cells treated with 1 lgÁlL À1 GSK591 (F) and stable expression of wild-type or methyltransferase-dead PRMT5 (G), in which Welch's t-test was performed.*P < 0.05 against "aS induction" or "Wild-type", n = 3.All graphs were presented as mean AE SEM.PRMT5, protein arginine methyltransferase 5; RT-qPCR, reverse transcription quantitative real-time PCR.

Fig. 6 .
Fig. 6.Impact of aS and PRMT5 on NRCAM expression.Validation using publicly available RNA-seq datasets.(A) Flow diagram showing the method used to obtain the relevant datasets from the NCBI/GEO database.RNA-seq datasets were obtained from human-derived cultured cells related to the knockdown or inhibition of PRMT5.(B) Forest plots of log-fold changes (logFC) of NRCAM against controls are shown.Similar to our experimental results (Fig. 5), a part of the datasets exhibiting aS overexpression shows NRCAM reduction, and approximately half of the PRMT5 knockdown or inhibition datasets show up-regulated NRCAM.None of the data analyzed was found to be inconsistent with our results.PRMT5, protein arginine methyltransferase 5.

Table 1 .
Target genes affected by the BAF-PRMT5 complex with increased aS expression.Results of ChIP-seq analyses.Fifteen genes were peak-annotated with at least P < 0.00001 significance by comparing aS-overexpressed SH-SY5Y cells with normal differentiated or untreated cells.