Fig. 1
Workflow of proteo-genomics of soluble TREM2 (sTREM2) in cerebrospinal fluid (CSF) and association results using 3,350 EUR samples. A Our study included a study design with two stages: the first stage includes a GWAS using 3350 European samples from eight cohorts and 12,621,222 autosomal genotypic variants, and second stagein where multi-ethnic fine mapping using 250 non-European (non-EUR) samples from eight cohorts and 8,909,120 autosomal genotypic variants was performed. CSF sTREM2 was measured by SomaScan or MSD. In the first-stage GWAS analyses, using an additive linear model adjusting for age at CSF draw, sex, genotype platform/cohorts, and 10 PCs, we identified 4 loci associated with CSF sTREM2 levels: chromosome 3 RBMS3-TGFBR2 (novel), chromosome 6 TREM2 (novel), chromosome 11 MS4A (known), and chromosome 19 APOE (novel) as shown in Manhattan plot and locus zoom plots. For these 4 loci, we then conducted post-GWAS analyses. First, we used multi-ethnic fine mapping to detect the true causal variants underlying each locus. For each of the four loci, we then performed stepwise conditional analyses to identify the independent genotypic variants. To identify the functional genes underlying three novel loci, we performed colocalization analyses of each locus with the AD GWAS, GTEx eQTL, and MetaBrain eQTL. The regulatory role of these loci were annotated with the brain cell type-specific enhancer-promoter interaction map. For chromosome 3 RBMS3-TGFBR2 locus, in vitro functional validation using overexpression of TGFBR2 and RBMS3 in human primary macrophages was conducted. The overall genetic architecture overlapped between AD PRS and CSF sTREM2 was estimated using multivariate linear regression. Finally to determine whether CSF sTREM2 is causal for AD, two-sample Mendelian randomization was analyzed using CSF sTREM2 GWAS as exposure and the latest AD GWAS as outcome. B Manhattan plots of GWAS for cerebrospinal fluid (CSF) soluble triggering receptor expressed on myeloid cells 2 (sTREM2) in European individuals (EURs). P values are two-sided raw P values estimated from a linear additive model. The blue solid horizontal line denotes the genome-wide significance level (P = 5 × 10-8), and the red solid horizontal line represents the suggestive significance level (P = 1 × 10-6). X-axis depicts genomic coordinates by chromosome number and y-axis denotes the negative log10-transformed P value for each genetic variant. C) LocusZoom plot of GWAS of CSF sTREM2 at chromosome 3, 6, 11, and 19. The X-axis depicts genomic coordinates and the y-axis denotes the negative log10-transformed P value for each genetic variant