Integration of genetics and proteomics reveals circulating TIMP2 role in cognitive performance

Murine studies have identified blood proteins influencing mice cognitive function, yet translating these findings to humans remains challenging. We report an innovative bioinformatic approach to investigate proteins' potential effects on the human brain. We generated genetic proxies to predict plasma protein levels through a clumping-thresholding approach, and validated their robustness in two independent cohorts. We then assessed their association with cognitive performance in humans.

First, we identified circulating proteins in mice affecting brain aging through a systematic review. We retrieved protein quantitative trait loci (pQTLs) summary statistics associated with these proteins in human plasma from the Fenland study (Pietzner et al., 2021). Clumping-thresholding polygenic scores (protPRSs) were computed as proxies for plasma protein levels in 1,380 cognitively unimpaired (CU) individuals from the Knight-ADRC cohort and 410 CU individuals from the ALFA+ study. We validated the predictive ability of the protPRSs by analyzing their association with plasma protein levels measured by Somalogic.

As a proof of concept, we assessed associations of genetically predicted protein levels with cognitive performance using linear models adjusted for age, sex, and education in ALFA+ study participants. Stratified models by sex, APOE-ε4 carriership, and Aβ status were also assessed. We annotated significant pQTLs for biological significance through enrichment analysis. Gene-level tissue specificity was evaluated by integrating differential gene expression data from GTEx v8.

We identified 12 circulating proteins affecting mice brain aging. Most computed protPRS significantly predicted plasma protein levels across the two independent cohorts: Knight-ADRC (10/13) and ALFA+ (7/13). In ALFA+, a significant association was found between TIMP2-protPRS and better cognitive performance, measured by Preclinical Alzheimer"s Cognitive Composite (PACC) and episodic memory composite (EM). Associations of TIMP2-protPRS with PACC remained significant in stratified models. The pQTLs incorporated into the TIMP2-protPRS were located on chromosome 3, potentially crucial for regulating TIMP2 protein levels. Notably, significant variants were annotated within the TLR9, SMIM4, STAB1, and RTF1 genes. Integrating genetic data with expression profiles across 54 tissues revealed tissue-dependent up-regulation of genes associated with TIMP2 pQTLs, particularly in brain regions.

This innovative approach integrating genetic and proteomics data, overcomes translational challenges from animal studies and provides a complementary framework for assessing the association of circulating proteins with brain performance in humans. We demonstrated that genetically predicted plasma TIMP2 levels, known for its rejuvenating effect on mice brains, are linked to better cognitive performance in humans, identifying TIMP2 as a potential therapeutic target for age-related brain diseases.