Additional file 2 of Distinct sex-specific DNA methylation differences in Alzheimer’s disease

Additional file 2: Supplementary Table 1. Quality control (QC) information on DNA methylation samples and probes for each dataset contributing to the sex-specific meta-analyses. Under Probes QC, shown are the number of probes remaining after each QC procedure. Under Samples QC, shown are the number of samples remaining after each QC procedure. Supplementary Table 2. At P < 10-5, sex-specific meta-analyses identified a total of 23 CpGs and 4 CpGs signicantly associated with AD diagnosis in female samples and male samples, respectively . For each CpG, annotations include the location of the CpG based on hg19/GRCh37 genomic annotation (chr, position), nearby genes based on GREAT (GREAT_annotation), the type of associated genomic feature (RefGene_Group), Illumina gene annotations, location with respect to CpG islands (Relation_to_Island), and overlap with enhancers identified in Nasser et al. [53] study (PMID: 33828297). Inverse-variance weighted fixed-effects meta-analysis models were used to combine cohort-specific results from logistic regression models that included covariate variables age, batch, and immune cell-type proportions. A total of 9 CpGs had the same direction of change in males and females (highlighted in gray). Odds ratios (OR) describe changes in odds of AD (on the multiplicative scale) associated with a one percent increase in methylation beta values (i.e., increase in methylation beta values by 0.01) after adjusting for covariate variables. Highlighted in red are CpGs that mapped to promoter regions. 95% CI = 95% confidence interval for odds ratio. Supplementary Table 3. In female samples, a total of 41 DMRs were significantly associated with AD diagnosis at 5% Sidak corrected P-value. Among them, 6 DMRs overlapped with enhancer regions from Nasser et al. study (Nature 2021; PMID: 33828297) (Enhancer = TRUE). Highlighted in red are DMRs that mapped to promoter regions. Direction indicates hypermethylation (+) or hypomethylation (-) in AD subjects, which was determined based on hyper- or hypo- methylation of the majority of the CpGs (located within the DMR) in meta-analysis. Supplementary Table 4. CpGs within top 10 most significant DMRs in females. Direction indicates hypermethylation (+) or hypomethylation (-) in AD samples in the ADNI and AIBL datasets. Supplementary Table 5. In male samples, a total of 24 DMRs were significantly associated with AD diagnosis at 5% Sidak corrected P-value. Among them, 7 DMRs overlapped with enhancer regions from Nasser et al. study (Nature 2021; PMID: 33828297) (Enhancer = TRUE). Highlighted in red are DMRs that mapped to promoter regions. Direction indicates hypermethylation (+) or hypomethylation (-) in AD subjects, which was determined based on hyper- or hypo- methylation of the majority of the CpGs (located within the DMR) in meta-analysis. Supplementary Table 6. CpGs within the top 10 most significant DMRs in males. Direction indicates hypermethylation (+) or hypomethylation (-) in AD samples in the ADNI and AIBL datasets. Supplementary Table 7. Information on brain samples used in cross-tissue meta-analysis. Supplementary Table 8. Results of analysis of female samples. In (a) and (b), we analyzed matched DNAm-RNA from the ADNI dataset (adni.loni.usc.edu), and tested association of DNA methylation at significant CpGs with expression levels of genes located nearby. At 5% FDR, for CpGs in the promoter regions (i.e., within +/- 2k bp from TSS), DNAm at 23 CpGs (mapped to 5 DMRs) were significantly associated with expressions of their target genes. For CpGs in distal regions (>2k bp from TSS), we tested association between the CpGs with 10 genes upstream and 10 genes downstream from the CpG location. Only 1 CpG was significantly associated with expression of its target gene at 5% FDR. In (c), we performed a meta-analysis for gene expressions of the target genes using two prefrontal cortex brain samples datasets in AD (GEO accessions: GSE33000, GSE44772), to test association between gene expression and AD, adjusting for age, sex and surrogate variables for cell types. Supplementary Table 9. Results of analysis of male samples with matched DNAm-RNA data in the ADNI dataset. In (a) and (b), we tested association of DNA methylation at significant CpGs with expression levels of genes located nearby. At 5% FDR, for CpGs in the promoter regions (i.e., within +/- 2k bp from TSS), DNAm at 12 CpGs (mapped to 2 DMRs) were significantly associated with expressions of their target genes. For CpGs in distal regions (>2k bp from TSS), we tested association between the CpGs with 10 genes upstream and 10 genes downstream from the CpG location. A total of 13 distal CpGs (mapped to 5 DMRs) were significantly associated with expressions of their target genes at 5% FDR. In (c), we performed a meta-analysis for gene expressions of the target genes using two prefrontal cortex brain samples datasets in AD (GEO accessions: GSE33000, GSE44772), to test association between gene expression and AD, adjusting for age, sex and surrogate variables for cell types. Supplementary 10. In femlaes, a total of 64 CpG - mQTL pairs were significant in both brain and blood samples analyses. The blood mQTLs and brain mQTLs were obtained from the GoDMC database and xQTL server, respectively. Supplementary 11. In males, a total of 19 CpG - mQTL pairs were significant in both brain and blood samples analyses. The blood mQTLs and brain mQTLs were obtained from the GoDMC database and xQTL server, respectively. Supplementary Table 12. In females, a total of 155 mQTLs in the blood overlapped with the 24 GWAS nominated LD blocks in Kunkle et al. [60] (PMID: 30820047). The mQTLs in blood were obtained from the GoDMC database. Annotations for CpGs include location of the CpG based on hg19/GRCh37 genomic annotation (Chr, Position), Illumina gene annotation (UCSC_RefGene_Name), the type of associated genomic feature (UCSC_RefGene_Group), and location with respect to CpG islands (Relation_to_Island). Supplementary Table 13. In males, a total of 864 mQTLs in the blood overlapped with the 24 GWAS nominated LD blocks in Kunkle et al. [60] (PMID: 30820047). The mQTLs in blood were obtained from the GoDMC database. Annotations for CpGs include location of the CpG based on hg19/GRCh37 genomic annotation (Chr, Position), Illumina gene annotation (UCSC_RefGene_Name), the type of associated genomic feature (UCSC_RefGene_Group), and location with respect to CpG islands (Relation_to_Island). Supplementary Table 14. Overlap of AD-associated DMRs with AD GWAS loci reported in Kunkle et al. [60]. Supplementary Table 15. Sensitivity analysis for model that additionally adjust for smoking scores, which was computed using the SSc method as implemented in R package EpiSmokEr (PMID: 31466478). All 27 sex-specific CpGs from Supplementary Table 2 remained highly significant, with meta-analysis P-values ranging from 5.83 x 10-8 to 2.59 x 10-5. Supplementary Table 16. Sensitivity analysis comparing logistic regression model that additionally adjusts years of education vs. model not adjust education in the analysis of ADNI dataset. Supplementary Table 17. Results of internal validation that compared logsitic regression models with or without education effect. A 10-fold cross-validation using the ADNI dataset showed the estimated average AUCs for the best performing logistic regression models with and without education were 0.707 and 0.710 in females, and 0.650 and 0.604 in males. The MRS was computed as the sum of methylation beta values for significant CpGs weighted by their estimated effect sizes obtained in the meta-analysis. In males, significant CpGs used for the MRS included 2 out of the 5 significant CpGs in the meta-analysis of methylation-by-sex interaction effect which were also available in AddNeuroMed dataset. In females, significant CpGs used for MRS included 9 out of 23 CpGs in meta-analysis that compared AD vs. CN samples which were also available in AddNeuroMed dataset.