Multi-omics integration analysis identifies novel genes for alcoholism with potential overlap with neurodegenerative diseases.

Overview of the study.Series of analyses were undertaken to identify the candidate causal genes associated with risk of AUD and DPW. We used the stratified linkage disequilibrium score (LDSC) regression to test whether the heritability of AUD and DPW is enriched in regions surrounding genes with chromatin markers in a specific tissue. This analysis helped us to identify the large eQTL/mQTLs datasets in the relevant tissues to perform the multi-omic integration analysis using SMR. The candidate causal SNPs and genes prioritized using SMR were further filtered according to threshold of association in GWAS and linkage disequlibrium (Heidi P and COJO). The complex loci with multiple genes were further validated and prioritized by exploring differential gene expression data from brains of people with alcohol use disorder and controls. Integration of eQTL data from monocytes also helped to prioritize candidate genes specifically expressed in the myeloid cells. The cell type specific epigenetic data from the human brain was also used to identify the causal SNP/s associated with DPW and AUD.

LDSC analysis using tissue specific chromatin data.LDSC analysis showed significant enrichment of promoter-specific markers (H3K4me1/me3) in the fetal and adult brain for the SNPs identified in (A) DPW and (B) AUD GWAS meta-analysis. Y-axis represents the annotations and X-axis represents the −log 10 P value for enrichment calculated using partition heritability method as implemented in LDSC. The dotted red line represents the threshold of multiple test correction according to Bonferroni.

Results of SMR based integration analysis of DPW and AUD GWAS meta-analysis with eQTL/mQTL from fetal and adult brain.X-axis represents the chromosomes and Y-axis shows the standardized direction of effect (Z scores) of co-localized SNP on gene expression/ methylation and GWAS phenotype. Z scores were derived from the effect size (betas) and standard errors (SE) from the SMR analyses. Positive Z score shows that increase in mRNA expression or methylation is associated with excessive drinking or increased risk of AUD, while negative Z score depicts the vice-versa. Genes marked on the plots represent the genes that passed the strict threshold of co-localization (FDR SMR-P < 20%; SMRHeidi P > 0.05; GWAS P < 5 × 10−5; eQTL P < 5 × 10−8) and multiple levels of transcriptomic and/ or epigenetic evidence. SMR P-values for the co-localized SNPs are obtained using the Wald test.

MAPT was identified as a candidate gene associated with increased DPW.A Locus zoom plot showing DPW and eQTL (DLFPC) associations at 17q.21.3. X-axis represents the positions along chromosome 17 and the y-axis represents the P values of each SNP at this locus. P-values for the DPW GWAS are obtained from aggregated weighted Z statistics (Liu et al, 2019). P-values for DLFPC-eQTL meta-analysis were obtained using conventional inverse-variance-weighted meta-analysis as implemented in the SMR software package. Color of each dot presents the R2 for LD at the locus (Red = 0.8–1.0; Orange 0.6–0.79; Green 0.4–0.59; Blue 0.2–0.39 and dark blue <2.0). B The co-localized SNPs were found to be overlapping with the chromatin interaction region that loops back to the promoter of the MAPT gene. C In independent transcriptomic data from the human brain (N = 92), mRNA expression of MAPT was found to be associated with the alcohol consumption. The analysis was performed using DeSeq2 program and P-value for association resulted from the Wald test. The shaded area around trend-line depicts the 95% confidence level intervals plotted using “geom_smooth(method = “lm”)” from ggplot2.

SPI1 was nominated as candidate gene associated with increased DPW and AUD.A Locus zoom plot showing DPW, AUD, and mQTL (Fetal brain) associations at 11p.11.2. X-axis represents the positions along chromosome 11 and y-axis represents the −log10 (P) values of each SNP at this locus. P-values for the DPW GWAS (Liu et al, 2019) and AUD GWAS are obtained from aggregated weighted Z statistics. P-values for mQTL meta-analysis were obtained using conventional inverse-variance-weighted meta-analysis. Color of each dot presents the R2 for LD at the locus (Red = 0.8–1.0; Orange 0.6–0.79; Green 0.4–0.59; Blue 0.2–0.39 and dark blue <2.0). Yellow line represents the position of rs56030824 identified as a functional variant co-localized with AUD, DPW, and mQTLs in the fetal brain. The tracks show the peaks for promoter marks in 4 major cell types of the brain. Rs56030824 was found to overlap with promoter-specific marks (H3K4me3 and H3K27ac), specifically in microglia. B Effect sizes for DPW GWAS and SPI1 expression in CD14+ monocytes were found to be correlated i.e. decreased alcohol intake was associated with decreased SPI1 expression. Rs56030824 showed the strongest association with DPW and mQTL in the common variant category. The shaded area around trend-line depicts the 95% confidence level intervals plotted using “geom_smooth(method = “lm”)” from ggplot2. C rs56030824 is a strong eQTL and associated with SPI1 expression in CD14+ monocytes. The box in the Box Plot is extending from the 25th percentile to the 75th percentile, with median horizontal line within the box. The whiskers are extending to one and a half times the interquartile range.