SVA insertion in X-linked Dystonia Parkinsonism alters histone H3 acetylation associated with TAF1 gene

X-linked Dystonia-Parkinsonism (XDP) is a neurodegenerative disease linked to an insertion of a SINE-VNTR-Alu (SVA)-type retrotransposon within an intron of TAF1. This SVA insertion induces aberrant TAF1 splicing and partial intron retention, thereby decreasing levels of the full-length transcript. Here we sought to determine if these altered transcriptional dynamics caused by the SVA are also accompanied by local changes in histone acetylation, given that these modifications influence gene expression. Because TAF1 protein may itself exhibit histone acetyltransferase activity, we also examined whether decreased TAF1 expression in XDP cell lines and post-mortem brain affects global levels of acetylated histone H3 (AcH3). The results demonstrate that total AcH3 are not altered in XDP post-mortem prefrontal cortex or cell lines. We also did not detect local differences in AcH3 associated with TAF1 exons or intronic sites flanking the SVA insertion. There was, however, a decrease in AcH3 association with the exon immediately proximal to the intronic SVA, and this decrease was normalized by CRISPR/Cas-excision of the SVA. Collectively, these data suggest that the SVA insertion alters histone status in this region, which may contribute to the dysregulation of TAF1 expression.


Introduction
X-Linked Dystonia-Parkinsonism (XDP) is a progressive neurodegenerative movement disorder endemic to the island of Panay in the Philippines [1][2][3] that is characterized by the loss of medium spiny neurons in the striatum [2,4]. Previous studies mapped the genetic XDP causal locus to TAF1 [5,6] which encodes TATA Binding Protein (TBP)-Associated Factor-1 a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 (TAF1), the largest component of the multi-subunit TFIID complex involved in RNA polymerase II-mediated transcription [7][8][9][10]. In these analyses, all XDP individuals appeared to share a common haplotype consisting of seven sequence variants clustered in and around TAF1, specifically a 48-base pair (48-bp) deletion, five disease-specific single nucleotide changes (designated in the literature as DSC-1, -2, -3, -10, and - 12), and an intronic SINE-VN-TR-Alu (SVA)-type retrotransposon insertion [5,6]. Recent studies have since expanded the shared haplotype to encompass thirteen XDP-specific variants, including the original known markers, that fall within a genomic segment exclusive to TAF1 [11]. Of these disease-related variants, only the SVA has thus far been linked directly to clinical disease manifestation. Its 5' end contains a repeated hexamer, (CCCTCT) n , the length of which is polymorphic among XDP patients and is inversely correlated with age at disease onset [12,13]. Nevertheless, the possibility that any of the other variants within the haplotype may also contribute to pathogenesis is not excluded.
Alterations in TAF1 transcription have been described in XDP fibroblasts, induced pluripotent stem cells (iPSCs) and their neural derivatives, post-mortem brain tissue, and blood [6,11,[14][15][16] with some evidence that lower TAF1 levels may be inversely correlated to the length of the hexameric repeat within the SVA [13]. Furthermore, our group recently described additional transcriptional abnormalities in XDP neural stem cells occurring around the large TAF1 intron in which the SVA is inserted, specifically: (1) multiple aberrant splicing events terminating within TAF1 intron 32 immediately proximal to the SVA insertion, the most abundant of which was annotated as transcript TAF1-32i, (2) increased partial intron retention (IR) incorporating a proximal segment of intron 32 sequence within mature TAF1 mRNA, and (3) decreased exon usage immediately 3' to intron 32 [11]. Importantly, the excision of the SVA by CRISPR/Cas9-mediated genome editing rescued all of these defects and normalized expression of the full-length TAF1 transcript [11]. Other studies have also reported that SVA excision in XDP-derived iPSCs increases TAF1 expression [16]. Together, these findings strongly suggest that the decreased TAF1 expression consistently detected in XDP patients may be caused by aberrant transcriptional dynamics within intron 32 and flanking exons as a result of the SVA insertion. However, the precise mechanisms involved are yet to be elucidated.
Here, we assessed alterations in histone H3 acetylation in human post-mortem prefrontal cortex (PFC), a region characterized by relatively healthy neurons that are part of the striatalcortical circuit, and fibroblasts derived from XDP patients and unaffected family members. Furthermore, we assessed acetylated histone association with sites within the TAF1 gene in both XDP-derived fibroblasts and neuronal stem cells (NSCs). Lastly, we determined whether removing the SVA could reverse alterations in histone acetylation at specific TAF1 gene loci in NSCs derived from XDP and controls.

Methods
All methods were carried out in accordance with the guidelines and regulations of Massachusetts General Hospital and approved by the Massachusetts General Hospital licensing committees.

Human tissue samples
All procedures related to the collection, processing, and the use of XDP patient post-mortem brain samples were approved by Institutional Review Boards (IRB) at Makati Medical Center (Makati City, Philippines; protocol MMCIRB 2017-134) and Massachusetts General Hospital (MGH; Boston, USA; protocol 2016p-000427). XDP patients were informed about the option of post-mortem brain donation by their movement disorder neurologist or by their genetic counselor, clinical care, and seminars offered by the Sunshine Care Foundation (Roxas City, Capiz Philippines). Individuals interested in participating in brain donations were afforded with additional information. Informed consent for post-mortem brain donation was acquired from all participants. Final consent was obtained from next of kin after family members or health care providers contacted the brain bank at the time of death. Brain bank staff immediately traveled to the local funeral home to perform an autopsy limited to brain. Brains were placed in an ice water bath and transported to the brain bank facilities for processing following removal. For each brain, one hemisphere was cut into 16 standard coronal sections and frozen between Teflon coated cold metal plates on dry ice and the opposite hemisphere was fixed in formalin. Here, fresh frozen PFC was dissected from the coronal sections of 11 XDP patient brains. Control tissue from 3 PFC was provided by the Massachusetts Alzheimer's Disease Research Center (ADRC) with approval from the Massachusetts General Hospital IRB (1999p009556). Clinical information on the tissues used in this study are reported in Table 1.

Human fibroblast cultures
Human fibroblasts were provided by the MGH Collaborative Center for X-linked Dystonia Parkinsonism. The clinical characteristics of donor subjects, confirmation of XDP genotype, and derivation of fibroblasts were previously described [11,15,53]. In total we assessed histone AcH3 levels in fibroblasts derived from 6 XDP patients and 10 unaffected family members as well as histone AcH3 association across TAF1 in fibroblasts derived from 7 XDP patients and 14 unaffected family members. All fibroblasts were cultured in Dulbecco's Modified Eagle's Medium (DMEM) supplemented with 20% fetal bovine serum (FBS) and 1X Penicillin/Streptomycin/L-glutamine and grown in an incubator at 37˚C, 5% CO 2 .

Neuronal stem cells (NSCs) cultures
NSCs were provided by the MGH Collaborative Center for X-linked Dystonia Parkinsonism. Induced pluripotent stem cells reprogramming, neuronal differentiation and CRISPR/Cas9 nuclease-mediated genome editing were previously described [11]. We used 4 control, XDP, and ΔSVA lines for the outlined experiments.

Histone extractions
Histone extracts were obtained from XDP and control post-mortem PFC as well as XDPand control-derived fibroblasts as previously reported [54,55]. Briefly, tissues and cells were homogenized in 5% Triton buffer on ice and centrifuged at 500 g for 8 min at 4˚C to extract nuclei. Then, nuclei were washed twice in 5% Triton buffer and histones were extracted in 0.2 M HCl by vigorous shaking at 4˚C for 3 h. After centrifugation, the supernatants containing histones were collected and neutralized with 1 M NaOH. Protein concentrations were determined by Bradford assay.

Chromatin immunoprecipitation (ChIP)
ChIP for AcH3 was performed as previously described [54,56] . Briefly, proteins were crosslinked to DNA using formaldehyde for 1 h at 37˚C and washed three times in a PBS solution containing protease inhibitors at 4˚C. Cell pellets were resuspended in 100 μl of 1% SDS lysis buffer supplemented with 10 mM EDTA, 50 mM Tris-HCl (pH 8.1) and protease inhibitors. Next, DNA was sheared to 200-1000 bp using a probe sonicator (Dismembrenator Model 500, Fisher Scientific. PA). Prior to ChIP, a sample of each homogenate representing 1% of the total volume was removed and set aside as input. Samples were diluted with ChIP dilution buffer (0.01% SDS,

Quantitative real-time PCR
Genomic DNA pulled down by ChIP as well as in the input samples was analyzed using SYBR Green (Qiagen, Valencia, CA) on an iCycler (Bio-Rad, Hercules, CA), with the following PCR protocol: 50˚C for 2 min, 95˚C for 10 min, 95˚C for 15 sec, 60˚C for 1 min for a total of 40 cycles. Each sample was run in duplicates and the average of the duplicate crossing threshold (Ct) values was used in this formula 100 � 2^(Adjusted input-Ct (IP)) to assess differences in AcH3 association with TAF1 exons and introns in fibroblasts and NSCs derived from control and XDP patients. The sequences of the primers used for real-time PCR were previously described [6,15] and are reported in Table 2. Specifically, intron 32-1b, intron 32-2a, intron 32-3c, and intron 32-4c refer to the primers used to amplify the region in intron 32 between exon 32 and the SVA as depicted in

Statistics
Normal distribution of data was not assumed regardless of sample size or variance. For graphical representation display individual values with the central line representing the median, and the edges representing the interquartile range, respectively. Comparison for unrelated samples were performed using a non-parametric Mann Whitney U test and a one-way ANOVA followed by Tukey's test at a significance level (a) of 0.05. Exact P values are reported.

Histone H3 acetylation is not altered in human post-mortem PFC and fibroblasts derived from XDP patients
In order to verify whether there were alterations in histone acetylation in the central nervous system (CNS) of XDP patients, total histone H3 acetylation (H3K9K14ac 2 ; AcH3) levels were assessed in human post-mortem PFC from XDP (n = 11) and non-neurological controls (n = 3) by western blots (Fig 1A). Our results demonstrate that there were no significant differences in AcH3 levels in XDP PFC compared to controls (Mann-Whitney U test = 10, p = 0.1841) ( Fig 1B). Next, we assessed total AcH3 levels in fibroblasts derived from XDP patients (n = 6) and unaffected family members (n = 10) by western blots (Fig 2A). As observed in XDP PFC, there were no significant differences in AcH3 levels in XDP-derived fibroblasts compared to control fibroblasts (Mann-Whitney U test = 12, p = 0.1898) (Fig 2B).

Acetylated histone H3 association with disease-specific variants along TAF1 gene is not altered in XDP-derived fibroblasts
Given that global histone acetylation levels may not represent the status of AcH3 associated with specific gene loci, we sought to determine whether there were alterations in AcH3 association with TAF1. Therefore, we examined sites within exons, the regions of intron 32 surrounding the SVA site at which abnormal splicing/IR occurs, and two additional disease specific variants (DSC10 and DSC12) within the XDP haplotype that are the closest to the SVA insertion ( Fig 3A) [5,6,11,57]. First, AcH3 association with DSC10 and DSC12 was assessed using chromatin immunoprecipitation (ChIP) with an anti-AcH3 antibody followed by qPCR in fibroblasts derived from XDP and controls. The results demonstrate that there was no significant change in AcH3 association with DSC10 or DSC12 in XDP-derived fibroblasts (n = 7) compared to control cells (n = 14) (Mann Whitney U test = 60, p = 0.5749, and Mann Whitney U test = 24, p = 0.2032, respectively) (Fig 3B and 3C).

Acetylated histone H3 association along the TAF1 gene is altered in XDPderived fibroblasts
Given that alteration in the transcription of exons surrounding the SVA insertion has been described in XDP [6,11,15], we measured AcH3 association across the TAF1 gene (Fig 4A), interrogating both constitutive and alternative exons that we and others have previously annotated [6,11,57], by ChIP with an anti-AcH3 antibody followed by qPCR in fibroblasts derived from XDP patients (n = 7) and unaffected family members (n = 14). The analysis revealed that there were no differences in AcH3 association with exon 2, exon 3, exon 16, exon 19, and exon 27 in XDP-derived fibroblasts compared to controls (Mann-Whitney U = 19, p = 0.8357; Mann-Whitney U = 18.50, p = 0.1810; Mann-Whitney U = 25, p = 0.7782; Mann-Whitney U = 25.50, p = 0.5220; and Mann-Whitney U = 25, p = 0.2441, respectively). Furthermore, there was a trend towards a significant decrease in AcH3 association with exon 36 in XDP-derived fibroblasts compared to controls (Mann-Whitney U = 17, p = 0.0577). Importantly, the analysis revealed a significant increase in AcH3 association with exon 17 as well as a significant decrease at exon 32, and exon 38 in fibroblasts derived from XDP patients compared to fibroblasts derived from unaffected family members (Mann-Whitney U = 0.5000, p = 0.0012; Mann-Whitney U = 6, p = 0.0041; Mann-Whitney U = 10, p = 0.0087, respectively) (Fig 4B-4K).

Excision of the SVA normalizes deficits in acetylated histone H3 association with TAF1 exon 32 in XDP-derived NSCs
Given that AcH3 association with exon 32, the region adjacent to the SVA-type insertion in the TAF1 gene, is decreased in XDP-derived fibroblasts, we used NSCs derived from control, XDP and XDP ΔSVA to determine if the presence of the SVA could cause alterations in AcH3 at exon 32 using ChIP followed by qPCR. Our results revealed a significant effect on AcH3 association with exon 32 between XDP-derived NSCs and controls (one-way ANOVA [F(2, 9) = 6.867, p = 0.0155). Importantly, while there were no significant differences in AcH3 association with exon 32 between XDP-and control-derived NSCs, there was a significant increase in AcH3 association with exon 32 in XDP ΔSVA cells compared to XDP-derived NSCs (Tukey's test, p = 002178, and p = 0.0125, respectively) ( Fig 5). Next, we assessed whether excision of the SVA would alter histone H3 acetylation associated with TAF1 intron 32 using the same primers as depicted in Fig 3. Our results demonstrated no significant effect on AcH3 association with intron 32-1b as well as no significant difference in AcH3 association with intron 32-1b between XDP-and control-derived NSCs and XDP ΔSVA NSCs (one-way ANOVA [F(2, 9) = 0.2936, p = 0.7525; Tukey's test, p = 0.9572 and p = 0.7366, respectively) (S1A Fig). Similarly, there was no significant effect on AcH3 association with intron 32-2a, and there was no significant difference in AcH3 association with intron 32-2a in XDP-derived NSCs compared to both control-derived NSCs and XDP ΔSVA NSCs (one-way

Discussion
In this study, we demonstrated for the first time that total histone H3 acetylation is not altered in human post-mortem PFC or fibroblasts derived from XDP patients. Although previous studies have shown evidence that TAF1 protein may exhibit HAT activity, our results indicate that global AcH3 levels are not affected in XDP brain and cell lines in which TAF1 expression is reportedly reduced. We did, however, detect local differences in AcH3 association with sites within the TAF1 gene, albeit only within coding regions. AcH3 association was not altered in XDP fibroblasts at sites within intron 32 flanking the SVA, nor at two disease-specific single nucleotide changes, DSC10 and DSC12, within introns 32 and 18, respectively. Within coding regions, in contrast, there was a significant increase in AcH3 association with exon 17 and a significant decrease in exons 32 and 38 in XDP-derived fibroblasts compared to controlderived cells. Lastly, the excision of the SVA by CRISPR/Cas9-mediated gene editing induced an increase in AcH3 association with TAF1 exon 32 in XDP ΔSVA NSCs. Together, our findings suggest that the SVA-type insertion alters AcH3 association with specific regions of the TAF1 gene in XDP.
While decreases in TAF1 expression are well documented in XDP [6,11,12,[14][15][16], the molecular mechanisms involved in this process remain unknown. Alterations in the epigenome may provide a mechanism where alterations in the chromatin landscape may decrease TAF1 transcription in XDP. Interestingly, it has been suggested that retroelements, including SVAs, may regulate gene expression [44][45][46][47][48][49][50][51]. Similarly, histone modifications, including methylation, can regulate SVA-type retrotransposon insertions to prevent further retrotransposition or the expression of toxic RNA [26][27][28]. Furthermore, previous studies have demonstrated that retrotransposons may indirectly alter permissive histone marks such as acetylation by promoting repressive marks such as methylation [29]. Therefore, the expression of retrotransposons, including SVAs, is finely regulated in the genome, a process that is dysregulated during ageing and age-related diseases [58,59]. However, the exact involvement of histone acetylation in these regulatory mechanisms remain to be clarified. Our findings here demonstrating a decrease in AcH3 association with TAF1 exon 32 in both fibroblasts and NSCs derived from XDP patients suggest that alterations in histone acetylation in the region flanking the SVA site may be involved in regulating TAF1 expression in XDP. However, further investigation is required to clarify the role of histone acetylation in regulating SVAs across the genome. TAF1 activity has also been linked to histone acetylation, specifically, the C-terminal domain was shown to possess HAT activity [30][31][32][33][34][35], suggesting that a loss of TAF1 activity could lead to a decrease in histone acetylation. Furthermore, the two C-terminal bromodomains on TAF1 have been shown to selectively bind to and propagate acetylation of histones [30]. Here, we did not detect a significant change in total acetylation of histone H3 in postmortem PFC or fibroblasts derived from XDP patients. However, we did not assess alterations in histone acetylation in the striatum, the brain region most affected in XDP and characterized by the loss of medium spiny neurons and a decrease in TAF1 expression [4]. Future studies will assess whether significant alterations in histone acetylation occur in this brain region.
Alterations in global histone acetylation are not always representative of altered acetylation at individual gene promoters. For instance, we previously reported that, while global histone acetylation is not altered in both animal and cellular models of Huntington's disease (HD), the decrease in AcH3 levels occurs at specific genomic loci and there is a decrease in AcH3 association with specific genes downregulated in HD [54]. Similarly, we previously reported that TAF1 transcription of exons decreased distal to intron 32 as measured by RNA CApSeq in both fibroblasts and NSCs derived from XDP patients [11]. Specifically, we previously reported a decrease in the expression of TAF1 transcripts containing exons 32-36 and 38 [6,15]. In this study, we report that AcH3 association with the TAF1 gene is altered across this locus in XDPderived fibroblasts. Specifically, while there was a significant increase in AcH3 association with exon 17, there was a significant decrease in AcH3 association with exon 32, the region flanking the SVA, as well as with exon 38, and a trend towards a significant decrease with exon 36. Collectively, our findings suggest that decreases in AcH3 association with exons 32, 36, and 38 may provide a mechanism whereby these specific TAF1 transcripts are altered and may contribute to the overall decrease in TAF1 expression in XDP.
Recently, it has been demonstrated that the excision of the SVA normalized TAF1 expression in both XDP-derived iPSCs and NSCs [11,16], and this excision also restored normal splicing in XDP-derived NSCs [11]. Although the exact molecular mechanisms involved in this process are yet unclear, our findings suggest that alterations in histone acetylation may play a role in this process, whereby the presence of the SVA may alter AcH3 levels at this locus in XDP. In addition, although AcH3 association is not altered in XDP-derived NSCs compared to control cells, the excision of the SVA in XDP-derived NSCs was able to increase AcH3 association with exon 32 compared to unedited cells, further suggesting that the SVA is directly linked to altered histone H3 acetylation in XDP. However, genome wide studies, assessing AcH3 association across TAF1 as well as the entire genome are required to better characterize and determine the role of histone acetylation in XDP pathogenesis.
Lastly, alterations in HAT and HDAC levels and activity have been described in several neurodegenerative diseases, such as HD [43,60,61], Alzheimer's disease (AD) [62,63], Parkinson's disease (PD) [64,65], amyotrophic lateral sclerosis (ALS) [66,67], and spinal muscle atrophy (SMA) [68]. Therefore, targeting HDACs and HATs has been suggested as potential therapeutic approach for the treatment of neurodegenerative diseases [69][70][71]. For instance, targeting HDAC2 with the specific inhibitor mithramycin improved neuronal plasticity in cellular model of AD [63], while HDAC6 inhibition improved cognitive decline associated with HD [72], AD [73], tauopathy [74], and Charcot-Marie-Tooth disease [75]. Furthermore, three different HDAC inhibitor clinical trials were approved by the Food and Drug Administration (FDA) for treating neurodegenerative disorders, including sodium phenylbutyrate for the treatment of HD (Clinical Trial NCT00212316) [76] and ALS (Clinical Trial NCT00107770) [66], and valproic acid for the treatment of SMA (Clinical Trial NCT00227266) [68]. Our data suggests that altering AcH3 levels may regulate TAF1 transcription at least at specific loci, therefore, future studies will assess not only the role of specific HATs and HDACs in order to provide greater mechanistic insight into the pathogenesis of XDP but will also assess the potential therapeutic efficacy of epigenetic inhibitors in XDP-derived cell lines.