Parkinson-associated risk variant in distal enhancer of α-synuclein modulates target gene expression

Abstract

Genome-wide association studies (GWAS) have identified numerous genetic variants associated with complex diseases, but mechanistic insights are impeded by a lack of understanding of how specific risk variants functionally contribute to the underlying pathogenesis. It has been proposed that cis-acting effects of non-coding risk variants on gene expression are a major factor for phenotypic variation of complex traits and disease susceptibility. Recent genome-scale epigenetic studies have highlighted the enrichment of GWAS-identified variants in regulatory DNA elements of disease-relevant cell types. Furthermore, single nucleotide polymorphism (SNP)-specific changes in transcription factor binding are correlated with heritable alterations in chromatin state and considered a major mediator of sequence-dependent regulation of gene expression. Here we describe a novel strategy to functionally dissect the cis-acting effect of genetic risk variants in regulatory elements on gene expression by combining genome-wide epigenetic information with clustered regularly-interspaced short palindromic repeats (CRISPR)/Cas9 genome editing in human pluripotent stem cells. By generating a genetically precisely controlled experimental system, we identify a common Parkinson's disease associated risk variant in a non-coding distal enhancer element that regulates the expression of α-synuclein (SNCA), a key gene implicated in the pathogenesis of Parkinson's disease. Our data suggest that the transcriptional deregulation of SNCA is associated with sequence-dependent binding of the brain-specific transcription factors EMX2 and NKX6-1. This work establishes an experimental paradigm to functionally connect genetic variation with disease-relevant phenotypes.

Extended data Fig. 1. Analysis of allele-specific expression of SNCA using quantitative reverse transcription polymerase chain reaction (qRT-PCR)

(a) Schematic illustration of the quantitative allele-specific SNCA expression analysis using a common primer pair and allele-specific Taqman® probes conjugated with different fluorophores to specifically detect a reporter-SNP (rs356165) in the 3′UTR of SNCA in a multiplex reaction. As indicated, 6-carboxyfluorescein (FAM) and 4,7,2′-trichloro-7′-phenyl-6-carboxyfluorescein (VIC) were used to detect the A- and G-allele respectively. (b) Representative multiplex qRT-PCR reactions (in duplicates) measuring allele-specific SNCA expression of allele-biased samples described in Fig. 1c,d. Allele-biased samples were generated by mixing hIPSC-derived neurons homozygous for either the A-(IPS-A) or G-allele (IPS-G) at rs356165 at indicated ratios. Also included is a plot showing c-DNAs synthesized in the absence of SuperScript® reverse transcriptase (no-RT) to control for genomic DNA contaminations. Plots are displayed as reporter dye fluorescence signal (ΔRn) in log scale as a function of run cycle.

Extended data Fig. 2. Analysis of in vitro differentiated hESC-derived mixed neuronal cultures

(a–d) Immunostainings of in vitro differentiated mixed neuronal cultures (differentiation day 28) for expression of neuron- and astrocyte-specific markers. Shown are representative images for staining of (a) neuron-specific beta-III-tubulin (TUJ1) and astrocyte specific glial fibrillary acidic protein (GFAP), (b) neuron-specific microtubule associated protein 2 (MAP2) and glutamatergic neuron-specific glutamate vesicular transporter 1 (vGLUT1), (c) TUJ1 and dopaminergic neuron-specific tyrosine-hydroxylase (TH) and (d) the pan-neuronal marker NeuN, which was used for relative quantification. (e) Quantification of a representative in vitro differentiation experiment in hESC line WIBR3 and BGO1. The quantification of rare TH-positive cells was only estimated. Source data are provided as Source Data for Extended Data Figure 2.

Extended data Fig. 3. Identification of PD associated risk variants overlapping with distal enhancers in the SNCA locus

(a) Detailed H3K4me1 and H3K27ac ChIP-Seq and DHSs-enrichment tracks for indicated CNS regions in the SNCA locus. Shown are the locations of NACP-Rep1 and PD-associated SNPs overlapping with two proximal enhancer elements (3′UTR enhancer and intron-4 enhancer) highlighted by light gray boxes. On the right, enlarged view of 3′UTR enhancer and intron-4 enhancer relative to top ranked PD-associated SNPs. (b) Enlarged view of intron-4-enhancer region showing H3K4me1 and H3K27ac Chip-seq enrichment tracks for substantia nigra and DHSs enrichment tracks for fetal brain relative to location of PD-associated SNPs. Shown below is number of predicted TF binding sites for reference (in red) and alternative SNP (minor allele) sequence (in blue) at each genomic position. Gray box indicates location of deletion described in Fig 2b and Extended Data Fig. 5a. (c) Summery of all PD-associated SNPs in the SNCA locus ranked by cumulative overlap with H3K4me1, H3K27ac and DHSs enhancer marks. Table summarizes the top 7 ranked SNPs with PD-association p-values, odd ratios (OR), number of predicted differential TF binding sites (Diff TFB) and location within enhancer elements as marked in (a). (d) Gene tracks showing H3K4me1 and H3K27ac enrichment in the SNCA locus for in vitro hESC-derived neurons (differentiation day 31).

Extended data Fig. 4. CRISPR/Cas9-mediated genome editing strategy for targeted insertion of PD–associated intron-4 enhancer elements in hESCs

(a) Schematic illustration of the CRISPR/Cas9-mediated 2-step genome editing strategy to delete and subsequently insert indicated intro-4 enhancer sequences containing the PD-associated risk SNPs rs356168 and rs3756054. Shown are the genomic organization of the SNCA locus, an enlarged view of wild-type and deleted (ΔE4) alleles, grand targeting sequences (underlined, PAM sequence in red), restriction sites, Southern blot (SB) probes and design of targeting vectors (TV; risk SNPs are highlighted in red). (b) Representative Southern blot analysis of indicated targeted WIBR3 hESCs (ΔE4/TV-A-T) compared with wild-type cells or hESCs carrying homozygous deletions (ΔE4/ΔE4). (c) Table summarizing intron-4 enhancer deletions and insertions of indicated haplotypes in WIBR3 hESCs. Correct targeting was determined by Southern blot analysis and genomic sequencing. Cell lines with targeted enhancer elements confirmed to be in cis with A-(FAM)-reporter SNP (determined by genomic sequencing-based phase-reconstruction) were maintained for subsequent analysis (compare Fig. 2b).

Extended data Fig. 5. Effect of intron-4-enhancer modification on total and allele-specific expression of SNCA

(a, b) Analysis of total SNCA expression in in vitro derived (a) neural precursors (same samples analyzed in Fig. 2c) and (b) mixed neuronal cultures (same samples analyzed in Fig. 2d) from targeted cell lines with indicated SNP genotypes at rs356168 and rs3756054 (compare Fig. 2b) compared to hESCs harbouring homozygous deletions of the intron-4 enhancer (ΔE/ΔE). qRT-PCR data are normalized to GAPDH and presented relative to the expression of ΔE/ΔE cells. (c) Expression analysis for the neuron-specific marker microtubule associated protein 2 (MAP2) and astrocyte specific glial fibrillary acidic protein (GFAP) by qRT-PCR in in vitro differentiated neurons described in (b). Data are normalized to 60S acidic ribosomal protein P0 (RPLP0) and presented relative to the expression of ΔE/ΔE neurons. Data are shown as mean ± s.d. of 3 biological replicates (each representing 3 technical replicates) for independent targeted clones as described in Fig. 2c,d (n indicates number of independently targeted clones per genotype; ΔE/ΔE, n = 4; A–T/ΔE, n = 4; G–C/ΔE, n = 3; A–C/ΔE, n = 3; G–T/ΔE, n = 2/3). (d) Alternative presentation of data displayed in Fig. 2c as dot blot grouped according to SNPs rs356168 and rs3756054 (excluding data for cells carrying homozygous deletions (ΔE4/ΔE4) of the intron 4 enhancer). (e) Table summarizing the results of corresponding two-way ANOVA analysis for allele-specific SNCA expression in neural precursors (each dot represents mean of 3 technical replicates, black bars indicate mean for each genotype). (f) Alternative presentation of data displayed in Fig. 2d as dot blot grouped according to SNPs rs356168 and rs3756054 (excluding data for cells carrying homozygous deletions (ΔE4/ΔE4) of the intron 4 enhancer). (e) Table summarizing the results of corresponding two-way ANOVA analysis for allele-specific SNCA expression in mixed neuronal cultures (each dot represents mean of 3 technical replicates, black bars indicate mean for each genotype). Allele-specific expression for each clone was analyzed in 3 independent biological replicate experiments and combined according to genotypes, n indicates number of independently targeted clones per group, † indicates an additional sub-clone derived from one of the two targeted clones for this genotype. Two-way ANOVA analysis (alpha = 0.05) was calculated based on allele-specific expression of all biological replicates. *P < 0.0001. (h) Schematic illustration of the experimental strategy for CRISPR/Cas9-mediated deletion of the intron-4-enhancer element. Genomic sequencing-based phase-reconstruction, to analyze the phase of the heterozygous enhancer SNP rs356168 in wild-type WIBR3 cells indicates that the functional G-allele at rs356168 is in cis with the G-allele of the reporter SNP rs356165. (i) Relative allele-specific SNCA expression (Boxplots showing median, 25^th and 75^th percentiles with whiskers indicating min and max) in neural precursors comparing wild-type cells to ΔE4/ΔE4 cells harboring homozygous deletions (expression is calculated relative to ΔE4/ΔE4 neural precursors). Note that the expression of the A-(FAM)-allele (displayed on the left Y-axis) is reciprocal to the expression of the G-(VIC)-allele (displayed on the right Y-axis). Statistically significant differences between groups were calculated using unpaired two-tailed t-test. n indicates number of independent sub-clones for wild-type cells or independently deleted clones for ΔE4/ΔE4 cells; allele-specific expression for each clone was analyzed in 3 independent biological replicate experiments at passage 1 and 2 respectively, each measured as 3 technical replicates. **P<0.0001. (j) qRT-PCR expression analysis for total SNCA in same samples as described in (i). Data are normalized to GAPDH and displayed relative to the expression of ΔE4/ΔE4 neural precursors. Data are displayed as mean ± s.d.; Source data and detailed statistical analysis are provided as Source Data for Extended Data Figure 5.

Extended data Fig. 6. Conditional GWAS and eQTL analysis to assess the effect of PD risk SNP rs356168

(a) Table showing baseline and conditional GWAS analysis in 5 publicly available PD GWAS cohorts (totaling 6014 cases and 9119 controls) to assess the extent to which rs356168 explains the observed associations in the SNCA locus of the top reported GWAS SNP rs356182 and the independent 5′ region SNP rs7681154 (b) qRT-PCR analysis for SNCA expression in postmortem frontal cortex tissue obtained from PD patients and controls stratified by risk genotype at rs356168 (number of brain samples from PD patients and controls are indicated for each genotype). Expression models were analyzed including adjustment for disease status, sex, pH, age at death, as well as for the interaction between PMI and disease status. Significance was assessed using a one sided test based on the a priori hypothesis of an association between the G-allele at rs356168 and increased expression of SNCA. P-values comparing each genotype are displayed in the graph. Alternatively analysis using linear regression shows a significant increase of total SNCA levels in carriers of G-allele at rs356168 (p = 0.031).

Extended data Fig. 7. PD-associated risk variants have little effect on enhancer-specific chromatin modifications at intron-4-enhancer

(a) Overview of isogenic cell lines (derived from WIBR3 hESCs) carrying distinct genotypes of the intron-4-enhacer element used for chromatin-immunoprecipitation (ChIP-qRT-PCR and ChIP-seq). (b, c) ChIP-qRT-PCR analysis in neurons derived from isogenic cell lines with indicated genotypes for binding of the enhancer-specific chromatin marks H3K4me1 (b) and H3K27ac (c) at the intron-4 and 3′UTR enhancer sequences compared with indicated negative control regions (calculated as percent of input). (d) Gene tracks of ChIP-seq analysis for the active enhancer mark H3K27ac in in vitro differentiated neurons derived from isogenic cell lines with indicated genotypes. (e) Quantitative read density analysis of H3K27ac Chip-seq data (as shown in d) displaying relative read density of the intron-4 enhancer compared to the 3′UTR enhancer to control for variability between ChIP experiments. Source data are provided as Source Data for Extended Data Figure 7.

Extended data Fig. 8. Sequence-specific binding of CNS expressed TFs EMX2 and NKX6-1 at SNCA intron-4-enhancer

(a) ChiP-qRT-PCR for binding of indicated TFs at the intron-4 enhancer element compared with negative control region in the SNCA locus (calculated as fold enrichment compared with IgG Isotype control) in hESC line BGO1 and hIPSC line IPS-PDC (derived from fibroblast AG20446). (b, d) Western blot analysis in nuclear extracts (NEs) (used for experiments displayed in Fig. 3 and Extended Data Fig 8c,e) from HEK293 cells overexpressing (b) Myc-DDK-tagged EMX2 or (d) NKX6-1 using indicated antibodies against DDK-(FLAG)-tag, EMX2 and NKX6-1 respectively. Lamin A was used to control for equal loading in NEs (c, e) EMSA analysis to determine TF concentration-dependent sequence-specific binding of (c) EMX2 and (e) NKX6-1 to oligonucleotides harboring the indicated genotype at rs356168 (A/G-allele). NEs from HEK293 cells overexpressing Myc-DDK(Flag)-tagged EMX2 (c) or NKX6-1 (e) were diluted with NEs from wild-type cells at indicated fractions to generate a TF concentration gradient. Relative EMX2 and NKX6-1 protein concentration in mixed samples were determined by Western blot analysis using an antibody against the DDK(Flag)-tag (panel below respective EMSA). Red arrows point to oligonucleotide-specific binding of overexpressed TFs. Source data are provided as Source Data for Extended Data Figure 8.

Extended data Fig. 9. Single molecular mRNA FISH analysis and immunostaining for TFs EMX2 and NKX6-1 in mixed neuronal cultures

(a) Single molecule mRNA FISH for EMX2-(Cy5) and NKX6-1-(Alexa594) (displayed in false color) labeling EMX2 and NKX6-1 mRNA transcripts in WIBR3-derived in vitro differentiated neurons (differentiation day 21). Cultured neurons were dissociated before hybridization and attached to a glass slide before imaging. Representative images show multiple cells, which are either single-positive for EMX2 (green arrowhead) or double-positive for EMX2 and NKX6-1 (white arrowhead). (b–d) Representative images showing individual cells which are either double positive for the expression of EMX2 and NKX6-1 (b) or single positive for either NKX6-1 (c) or EMX2 (d). (e) Quantification of 100 individual cells for the presence of EMX2 and NKX6-1 transcripts. (f–k) Co-immunostaining for EMX2, NKX6-1,neuron specific beta-III-tubulin (TUJ1) and astrocyte specific glial fibrillary acidic protein (GFAP) in mixed neuronal cultures. Source data are provided as Source Data for Extended Data Figure 9.

Extended data Figure 10. NACP-Rep1 repeat length has no cis-acting effect on SNCA expression in hESC derived neurons

(a) Schematic illustration of the CRISPR/Cas9-mediated genome editing strategy to generate heterozygous deletions (ΔRep1/wild-type) in WIBR3 and BGO1 hESCs and subsequently insert indicated NACP-Rep1 length variants. Displayed are the genomic organization of the SNCA locus, an enlarged view of the wild-type and NACP-Rep1 deleted (ΔRep1) allele, targeting-targeting sequences (underlined, PAM sequence in red), restriction sites and Southern blot (SB) probe. Shown below is targeting vector (TV) design to insert the respective NACP-Rep1 elements (Rep1-257, Rep1-259, Rep1-261 and Rep1-263) with indicated repeat sequences. Only clones with heterozygous deletion of the repeat element on the same chromosome were identified based on the genotype of two heterozygous SNPs (rs58864428 and rs10030935) upstream and downstream of NACP-Rep1 and selected for subsequent experiments. (b) Representative Southern blot analysis of wild type and targeted WIBR3 hESCs with indicated NACP-Rep1 genotypes (modified alleles highlighted in red; no targeted wild-type allele represents NACP-Rep1-261 element). (c) Table summarizing NACP-Rep1 deletions and insertions in WIBR3 and BGO1 hESCs. (d) Representative fragment length analysis confirms expected NACP-Rep-1 repeat length in targeted cell lines. Red line indicates NACP-Rep1-261 peak at 269 bp. (e, f) Analysis of relative allele-specific SNCA expression in neurons (differentiation day 25) derived from targeted cells lines with indicated NACP-Rep1 alleles compared with untargeted controls in WIBR3 (e) and BGO1 (f) hESCs (expression was normalized relative to wild-type cells). Shown are mean values ± s.d. of three independent biological replicate experiments for each individual clone of the indicated genotype. Differences between individual clones or combined clones by genotypes were not significant based on one-way ANOVA testing for multiple comparisons between groups (alpha = 0.05) and did show a repeat length-dependent linear trend analyzed by linear regression. Source data are provided as Source Data for Extended Data Figure 10.

Fig. 1. Strategy to analysis cis-regulatory effects of genetic variants on allele-specific expression of SNCA

(a) Schematic illustration of the experimental strategy to study the function of PD-associated regulatory elements in hPSC-derived neurons by CRISPR/Cas9-mediated heterozygous deletion (ΔE) or exchange/insertion (ExE) of risk-associated enhancer sequences. (b) Expected effect on relative allele-specific gene expression in hPSC-derived neurons resulting from enhancer modifications described in (a). (c) Allele-specific SNCA expression analysis of allele-biased samples with indicated allele ratios. Allele-biased samples were generated by mixing hIPSC-derived neurons homozygous for either the A-(IPS-A) or G-allele (IPS-G) at rs356165. The relative expression of each allele was normalized to the total expression of SNCA (black bars) in the respective sample. (d) qRT-PCR analysis for relative expression of the A-(FAM)-allele (red bars) and the G-(VIC)-allele (blue bars) in allele-biased samples described in (c, calculated as proportion) compared with expected relative proportions (light and dark grey bars). (e) Analysis of total SNCA expression at different time points during in vitro differentiation of hESCs-derived neurons (BGO1 and WIBR3). (f) Allele-specific SNCA expression of hESC-derived neurons described in (e) normalized to differentiation day 10 cultures. Shown are mean values ± s.d. (n = 3). nd, not detected. Source data are provided as Source Data for Extended Data Figure 1.

Fig. 2. Identification of a functional cis-acting PD-associated SNP in an intronic enhancer element of SNCA

(a) Heatmap of H3K4me1 and H3K27ac ChIP-Seq and DHSs-enrichment tracks for several CNS regions in the SNCA locus (for details see Extended Data Fig. 3a; Data provided by NIH Roadmap Epigenomics Consortium; http://www.roadmapepigenomics.org). Shown are the locations of NACP-Rep1 and PD-associated SNPs overlapping with two proximal enhancer elements (3′UTR-enhancer and intron-4-enhancer) highlighted by light gray boxes. (b) Schematic illustration of the CRISPR/Cas9-mediated strategy to delete and subsequently insert intron-4 enhancer elements with indicated PD-associated risk SNP genotype at rs356168 and rs3756054. Targeted clones carrying inserted risk alleles in cis with the A-(FAM)-reporter SNP (confirmed by genomic sequencing-based phase-reconstruction) were used for subsequent analysis described in (c,d). (c, d) Relative allele-specific SNCA expression in (c) neural precursors and (d) mixed neuronal cultures (differentiation day 25) derived from targeted cell lines with indicated intron-4 enhancer alleles compared to hESCs carrying homozygous enhancer deletions (ΔE4/ΔE4) (expression was normalized relative to ΔE/ΔE cell lines). Data are presented as dot plot; each dot represents mean of 3 technical replicates. Allele-specific expression for each clone was analyzed in 3 independent biological replicate experiments and combined according to genotypes. Black lines indicate mean expression for each genotype; n indicates number of independently targeted clones per genotype, † indicates an additional sub-clone derived from one of the two targeted clones for this genotype. Statistical differences between genotypes were calculated using one-way ANOVA (alpha = 0.05) followed by Tukey’s multiple comparison test based on allele-specific expression of all biological replicates. *P < 0.001; ** P<0.0001; source data and detailed statistical analysis are provided as Source Data for Extended Data Figure 2.

Fig. 3. Sequence-specific effect of PD-associated risk variants on binding of CNS expressed TFs EMX1 and NKX6-1 at SNCA intron-4 enhancer

(a) ChIP-qRT-PCR analysis for binding of indicated TFs at the intron-4-enhancer element compared with a negative control region in the SNCA locus (calculated as fold enrichment compared with IgG Isotype control). Statistical significance was determined using t-test followed by the Holm-Šídák method to correct for multiple comparisons (alpha = 0.05). *P < 0.0001. (b, c) EMSA analysis for SNP-genotype-specific binding of (b) EMX2 or (c) NKX6-1 to oligonucleotides (oligo) harboring the indicated genotype at rs356168 (A/G-allele). Binding was analyzed in nuclear extracts (NEs) from wild-type (293) or EMX1 (b, EMX) or NKX6-1 (c, NKX) overexpressing HEK293 cells. Red arrows point to oligonucleotide-specific binding which is lost in the presence of unlabeled competitor oligonucleotides (indicated genotype at rs356168; 200X). (d) SNCA expression analysis following doxycycline (DOX)-induced overexpression of EGFP, EMX2 or NKX6-1 for 3 days in terminally differentiated neurons (differentiation day 21). Shown are mean values ± s.d. (n = 10) of relative SNCA expression in DOX-induced cells compared with the corresponding untreated controls (NoDOX). Results are representative of two different experiments. Statistical significance was calculated using t-test followed by the Holm-Šídák method to correct for multiple comparisons (alpha = 0.05). *P < 0.0001. Source data are provided as Source Data for Extended Data Figure 3.

Fig. 4. Proposed model describing the correlation between SNP-dependent TF binding, SNCA expression and PD risk

Carriers of the A-allele at rs356168 (PD protective allele) show efficient binding of the brain-specific TFs EMX2 and NKX6-1 at the distal intron-4-enhancer, which results in a suppressed distal enhancer and consequently lower expression of SNCA associated with a reduced risk to develop PD. In contrast, carriers of the G-allele at rs356168 (PD-risk allele) show reduced TF binding, which results in an active distal enhancer leading to increased expression of SNCA and increased risk to develop PD.