pubmed.ncbi.nlm.nih.gov

Mapping cis-regulatory chromatin contacts in neural cells links neuropsychiatric disorder risk variants to target genes - PubMed

Mapping cis-regulatory chromatin contacts in neural cells links neuropsychiatric disorder risk variants to target genes

Michael Song et al. Nat Genet. 2019 Aug.

Abstract

Mutations in gene regulatory elements have been associated with a wide range of complex neuropsychiatric disorders. However, due to their cell-type specificity and difficulties in characterizing their regulatory targets, the ability to identify causal genetic variants has remained limited. To address these constraints, we perform an integrative analysis of chromatin interactions, open chromatin regions and transcriptomes using promoter capture Hi-C, assay for transposase-accessible chromatin with high-throughput sequencing (ATAC-seq) and RNA sequencing, respectively, in four functionally distinct neural cell types: induced pluripotent stem cell (iPSC)-induced excitatory neurons and lower motor neurons, iPSC-derived hippocampal dentate gyrus-like neurons and primary astrocytes. We identify hundreds of thousands of long-range cis-interactions between promoters and distal promoter-interacting regions, enabling us to link regulatory elements to their target genes and reveal putative processes that are dysregulated in disease. Finally, we validate several promoter-interacting regions by using clustered regularly interspaced short palindromic repeats (CRISPR) techniques in human excitatory neurons, demonstrating that CDK5RAP3, STRAP and DRD2 are transcriptionally regulated by physically linked enhancers.

PubMed Disclaimer

Conflict of interest statement

Competing interests

The authors declare no competing financial interests.

Figures

Figure 1.
Figure 1.. Genome-wide mapping of physical chromatin interactions in functionally distinct neural cell types.

(a) Schematic of the study design for generating four functionally distinct cell types in the CNS and performing integrative analysis of chromatin interactions using pcHi-C, open chromatin regions using ATAC-seq, and transcriptomes using RNA-seq. The number of biological replicates based on independent experiments for each cell type is shown for each assay. (b) Proportions of interactions occurring within TADs for each cell type. (c) Histogram and empirical CDF plots of interaction distances for each cell type. (d) Proportions of interactions between promoter-containing bins (blue) and between promoter- and non-promoter-containing bins (purple) for each cell type. (f) Proportions of cell type-specific (blue) and shared (grey) distal open chromatin peaks at PIRs for each cell type.

Figure 2.
Figure 2.. Integrative analysis of chromatin interactions, epigenomic features, and gene expression.

(a) Histograms of the number of PIRs interacting with each promoter for each cell type. Means are indicated. Only protein coding and noncoding RNA promoters interacting with at least one PIR are included (15,316 promoters in excitatory neurons, 19,546 promoters in hippocampal DG-like neurons, 14,990 promoters in lower motor neurons, and 15,397 promoters in astrocytes). (b) Bar plots showing counts of epigenomic chromatin states inferred using ChromHMM in matched tissues overlapping significant (solid bars) versus randomly shuffled (striped bars) PIRs for each cell type. Means and the SEM for the number of overlaps across n = 100 sets of randomly shuffled PIRs are shown. (c) Comparative gene expression analysis across all cell types for expressed genes (normalized RPKM > 0.5) whose promoters interact exclusively with either enhancer-PIRs (n = 6,836 genes) or repressive-PIRs (n = 2,612 genes) (P = 9.4 × 10−63, t = 16.9, df = 6854.6, two-tailed two-sample t test). Violin plots show the distributions of gene expression values within each group, and boxplots indicate the median, IQR, Q1 – 1.5 × IQR, and Q3 +1.5 × IQR. Means are indicated with dotted horizontal lines. (d) Distributions of gene expression values across all cell types for expressed genes (normalized RPKM > 0.5) grouped according to the numbers of interactions their promoters form with enhancer-PIRs. Boxplots indicate the median, IQR, Q1 – 1.5 × IQR, and Q3 + 1.5 × IQR. Linear regression was performed on the mean gene expression values for n = 9 bins containing at least 10 genes (P = 2.1 × 10−3, F1,7 = 22.7, F-test for linear regression).

Figure 3.
Figure 3.. Cell-type-specific PIRs and TF motif enrichment analysis.

(a) Classification of significant promoter-PIR interactions with interaction score ≥ 5 in at least one cell type based on their overall cell type-specificities. Counts of interactions in each specificity category are summarized in Supplementary Figure 3a. Cell types are hierarchically clustered based on their interaction scores over all interacting loci. (b) Top enriched GO terms from the “GO Biological Process 2018” ontology in Enrichr for genes participating in cell type-specific (groups 1–4) versus shared (group 15) interactions with distal open chromatin peaks. 459, 837, 217, 307, and 1,925 genes were used as inputs for groups 1–4 and 15, respectively. Enriched GO terms are ranked by their combined scores (calculated by multiplying the log of the P value via Fischer’s exact test with the z-score of the deviation from the expected rank). An expanded list of enriched GO terms is available in Supplementary Table 3. (c) Enrichment of consensus TF motif sequences at open chromatin peaks in cell type-specific PIRs by motifs (rows) and cell types (columns). 1,145, 1,271, 843, and 2,566 peaks were used as input for the excitatory neurons, hippocampal DG-like neurons, lower motor neurons, and astrocytes, respectively. The color of each dot represents the degree of enrichment (calculated using the cumulative binomial distribution in HOMER) for each motif and cell type, and the size of each dot represents the gene expression of the corresponding TFs for each motif. Entries with similar or identical consensus TF motif sequences are grouped for brevity.

Figure 4.
Figure 4.. Validation of PIRs in human neural cells.

(a) In-vivo-validated enhancer elements with neural annotations overlap a higher proportion of open chromatin peaks in the neural cells (757 of 919 elements) compared to enhancer elements with non-neural annotations (415 of 649 elements) (P < 2.2 × 10−16, χ2 = 67.5, df = 1, Pearson’s chi-squared test with Yates’s correction). (b) Counts of enhancer elements participating in chromatin interactions (589 of 1,568 elements) with neural and non-neural annotations. (c) Counts of enhancer elements interacting exclusively with their nearest genes (blue), more distal genes (pink), or both (orange), and the number of target genes for each scenario (right). (d) Open chromatin peaks in cell-type-specific PIRs (regions 1, 2, and 3) interact with the CDK5RAP3 promoter. Both enhancer elements (pink) and CTCF binding sites in excitatory neurons (dark blue) are localized to all three regions, and all interactions occur within a TAD in the cortical plate (chr17:45,920,000–47,480,000). (e) LacZ staining in mouse embryos reveals tissue-specific patterns of enhancer activity. (f) CRISPRi silencing of region 1 results in significant downregulation of CDK5RAP3 expression in excitatory neurons (P = 9.1 × 10−4, t = 4.65, df = 10 two-tailed two-sample t test). The neighboring genes MRPL10, PNPO, and NFE2L1 were unaffected (P = 9.1 × 10−2, t = 1.87, df = 10, P = 4.1 × 10−1, t = 0.853, df = 10, and P = 8.0 × 10−1, t = 0.259, df = 10, respectively, two-tailed two-sample t test). Three independent replicates per condition and two sgRNAs per replicate were used for each experiment. Boxplots indicate the median, IQR, minimum, and maximum. (g) CRISPRi silencing of region 2, but not region 3, results in significant downregulation of CDK5RAP3 expression in excitatory neurons (P = 2.2 × 10−3, t = 5.11, df = 6 and P = 3.3 × 10−1, t = 1.05, df = 6, respectively, two-tailed two-sample t test). Two independent replicates per condition and two sgRNAs per replicate were used for each experiment. Boxplots indicate the median, IQR, minimum, and maximum.

Figure 5.
Figure 5.. Genetic analysis of chromatin interactions with complex neuropsychiatric disorder-associated variants.

(a) Enrichment analysis for eleven complex neuropsychiatric disorders or traits. The color and size of each dot represent the enrichment P value (two-tailed one sample z-test) and the raw fold enrichment (determined as the number of SNPs overlapping significant PIRs divided by the mean number of SNPs overlapping n = 100 sets of randomly shuffled PIRs), respectively. The total numbers of SNPs are available in Supplementary Table 6. (b) Proportions and counts of GWAS SNPs with at least one linked SNP participating in chromatin interactions. (c) Counts of GWAS SNPs across all diseases with at least one linked SNP interacting exclusively with their nearest genes (blue), more distal genes (pink), or both (orange), and the number of target genes for each scenario (right). (d) PIRs with MP SNPS in an intron for PTPRO interact with the STRAP promoter. All interactions occur within a TAD in the cortical plate (chr12:14,960,000–16,040,000). Biallelic deletion of this PIR in three independent clones results in significant downregulation of STRAP expression in excitatory neurons (P = 3.4 × 10−4, t = 18.5, df = 3, two-tailed two-sample t test). Error bars represent the SEM. (e) A PIR containing SCZ SNPs interacts with the DRD2 promoter. All interactions occur within a TAD in the cortical plate (chr11:113,200,000–114,160,000). Monoallelic deletion of this PIR in three independent clones results in significant downregulation of DRD2 expression in excitatory neurons (P = 6.2 × 10−3, t = 6.92, df = 3, two-tailed two-sample t test). Error bars represent the SEM.

Figure 6.
Figure 6.. Genetics variants contribute to chromatin interaction bias and alterations in gene expression.

(a) Quantile-quantile plots showing the proportions of interacting 10-kb bins exhibiting significant allelic bias at an FDR cutoff of 5% (two-tailed binomial test with BH correction) in excitatory neurons (n = 22,162 bins) and lower motor neurons (n = 21,479 bins). (b) A sample interaction with significant allelic bias in excitatory neurons (P = 5.4 × 10−4, two-tailed binomial test) and lower motor neurons (P = 4.2 × 10−7, two-tailed binomial test). The interaction occurs between the SYT17 promoter and a PIR with bipolar alcoholism SNPs at an open chromatin peak. Heterozygous phased WTC11 variants at the PIR as well as bar graphs of allele-specific read counts are shown. (c) Enrichment of significant eQTLs from GTEx V7 at significant versus randomly shuffled PIRs in matched tissue types for excitatory and hippocampal DG-like neurons (P < 2.2 × 10−16 for both cell types, two-tailed one sample z-test). Means and the SEM for the number of overlaps across n = 100 sets of randomly shuffled PIRs are shown. (d) Distributions of interaction scores for chromatin interactions overlapping significant versus randomly sampled nonsignificant eQTL-TSS pairs in excitatory and hippocampal DG-like neurons (P = 2.3 × 10−4 for excitatory neurons and P = 1.8 × 10−6 for hippocampal DG-like neurons, two-tailed two-sample Kolmogorov-Smirnov test). Additional details are available in the methods. Violin plots show the distributions of gene expression values within each group, and boxplots indicate the median, IQR, Q1 – 1.5 × IQR, and Q3 + 1.5 × IQR.

Similar articles

Cited by

References

    1. Mumbach MR et al. Enhancer connectome in primary human cells identifies target genes of disease-associated DNA elements. Nat Genet 49, 1602–1612, doi:10.1038/ng.3963 (2017). - DOI - PMC - PubMed
    1. Zhu Z et al. Integration of summary data from GWAS and eQTL studies predicts complex trait gene targets. Nat Genet 48, 481–487, doi:10.1038/ng.3538 (2016). - DOI - PubMed
    1. Claussnitzer M et al. FTO Obesity Variant Circuitry and Adipocyte Browning in Humans. N Engl J Med 373, 895–907, doi:10.1056/NEJMoa1502214 (2015). - DOI - PMC - PubMed
    1. Smemo S et al. Obesity-associated variants within FTO form long-range functional connections with IRX3. Nature 507, 371–375, doi:10.1038/nature13138 (2014). - DOI - PMC - PubMed
    1. de la Torre-Ubieta L et al. The Dynamic Landscape of Open Chromatin during Human Cortical Neurogenesis. Cell 172, 289–304 e218, doi:10.1016/j.cell.2017.12.014 (2018). - DOI - PMC - PubMed

Publication types

MeSH terms

Substances