BackgroundIt is demonstrated that the soluble TREM2 in the cerebrospinal fluid (CSF) is associated with Alzheimer disease (AD). Recent studies by our group found that it is the MS4A4A and/or MS4A6A locus that plays a role in modulating sTREM2 and is also associated with AD risk. The goal of this study is to validate and extend findings from previous studies using a larger cohort, identify functional genes, and determine the overlap of the genetic architecture of CSF TREM2 levels with other traits.MethodWe used 3,033 individuals from Knight ADRC, ADNI, DIAN, and PPMI, and 8,499,039 variants (MAF >0.01) to identify protein quantitative trait loci (pQTLs) that modify TREM2 levels via a linear regression including age, sex, first 10 genetic principal components and array/cohort as covariates. TREM2 protein measurements were obtained from SomaScan 7k platform. Colocalization analyses were conducted for AD risk as well as functional genes from eQTLGen. LDSC and GNOVA and Mendelian randomization (MR) is being used to determine the overlap of the genetic architecture of sTREM2 levels with those for AD risk, onset and progression, as well as cardiovascular and lipid traits.ResultThe pQTL analysis identified the previously discovered MS4A gene region on chr11 (rs72918674, P = 7.009e‐58) as well as a peak on chr3 (rs73823326, P = 2.227e‐09). The lead variant resides in the intron of MS4A6A gene. Conditional analysis on this variant revealed rs10897026 as an additional independent signal located in the intron of MS4A4A. Chr11 peak colocalized with AD risk (PP‐H4 = 0.97) and MS4AE and MS4A6A expression QTL in blood (PP‐H4 = 0.97 and 0.96, respectively). Four regulatory region variants are in high linkage disequilibrium (LD) with chr3 index variant. RBMS3 and TGFBR2 are two genes flanking the chr3 peak, both expressed in microglia.ConclusionThis is the largest study to date aiming at identifying genetic modifiers of CSF sTREM2. The findings validated discoveries from previous studies and identified a novel signal on chr3 at genome‐wide significance. We propose two new genes, RBMS3 and TGFBR2 on chr3, could be involved in TREM2 biology. Our findings provide new insight to unravel sTREM2 modulators and their role in AD.

Background: Parkinson's disease (PD) is a movement disorder characterized by dopaminergic neurodegeneration in the substantia nigra of the brain. Another neurodegenerative disorder, Alzheimer's disease (AD), is a progressive memory loss disorder with pathological changes mainly at hippocampus and entorhinal cortex. As dementia is also present in late stage of PD, there might be common molecular mechanism between the pathogenesis of AD and PD. Evidence has nominated Leucine Rich Repeat Kinase 2 (LRRK2), Progranulin (GRN), Glycoprotein nonmetastatic melanoma protein B () since these proteins were linked to AD and PD. Identifying genetic variants of 5 proteins will be critical to shed lights on common disrupted pathways shared by AD and PD. Method:To identify genetic variants associated with these 5 proteins, we performed protein quantitative trait loci (pQTL) analyses of log10 transformed CSF GRN, GPNMB, CTSB, ENTPD1 and LRRK2 assayed in SomaScan platform. We considered ∼11 million sequenced variants from 787 individuals recruited at PPMI in the discovery and ∼4.3 million imputed variants from 799 Knight ADRC samples in replication. Age, sex and 10 principal components were included as covariates. Genome-wide threshold (5.0×10 −8 ) was used in the meta-analyses of discovery and replication for significance.Result: On chromosome 12, exonic missense LRRK2 variant rs33939927 (MAF = 1.4%) was identified as shared pQTL for GRN (P = 7.87×10 −10 ), CTSB (P = 8.99×10 −9 ), GPNMB (P = 3.59×10 −9 ), and ENTPD1 (P = 1.24×10 −13 ). The second LRRK2 exonic missense variant rs34637584 (MAF = 11%) was pQTL consistent for 3 proteins (GRN: P = 2.34×10 −8 , GPNMB: P = 8.55×10 −13 , and ENTPD1: P = 1.75×10 −13 ). The third LRRK2 exonic missense variant rs35303786 (MAF = 3.4%) was identified for GPNMB (P = 1.53×10 −8 ) and ENTPD1 (P = 2.31×10 −8 ). The risk allele of 3 SNPs were associated with higher amount of corresponding CSF proteins. No significant LRRK2 pQTL was identified. Additional analyses including colocalisation with AD and PD GWAS results are ongoing.

Background: Brain metabolism perturbation in dementia is not well understood.Metabolite levels are quantitative traits that have been linked to genetic loci. Existing metabolite quantitative trait loci (MQTL) were found mostly using tissues other than cerebrospinal fluid (CSF) and brain, which being less than ideal for studying neurodegenerative diseases. The first CSF MQTL study was published last year, but with limited