TL;DR
This study leverages genome-wide (GWAS) and epigenome-wide (EWAS) association studies to identify genetic and epigenetic factors influencing plasma protein levels associated with Alzheimer's disease (AD). They found that variations in genes related to TBCA and TREM2 are linked to AD risk, offering new insights into the molecular mechanisms underlying the disease.
What Problem Does This Paper Address?
Alzheimer's disease (AD) is a major global health challenge, and identifying reliable biomarkers for early detection and understanding disease mechanisms is crucial. While many blood proteins have been linked to AD, the underlying molecular factors regulating their levels and their direct impact on disease risk remain unclear. Traditional methods like neuroimaging and cerebrospinal fluid (CSF) analysis are invasive and costly. This paper aims to bridge this gap by investigating how genetic and epigenetic variations influence the levels of AD-associated plasma proteins, thereby shedding light on potential disease risk pathways.
Methods
The researchers employed a multi-pronged approach:
- Structured Literature Review: They systematically reviewed studies identifying plasma proteins associated with AD or its related phenotypes, focusing on those measured using the SOMAscan platform. This identified 282 relevant proteins.
- Genome-Wide Association Studies (GWAS): They conducted GWAS on the plasma levels of these 282 proteins in the Generation Scotland cohort. Bayesian penalized regression (BayesR+) was used to estimate the proportion of inter-individual variability in protein levels explained by genetic variation.
- Epigenome-Wide Association Studies (EWAS): Similarly, they performed EWAS on DNA methylation (DNAm) data from the same cohort to identify CpG sites associated with plasma protein levels. BayesR+ was also used here.
- Integration of GWAS and EWAS: They combined genetic and DNAm data to assess their independent and combined contributions to protein level variation.
- Mendelian Randomization (MR): This technique was used to infer causal relationships between plasma protein levels and AD risk or related biomarkers, and also to explore the relationship between gene expression and protein levels.
- Co-localization Analyses: These analyses were performed to determine if shared genetic variants underlie associations between different traits, such as protein levels and gene expression, or protein levels and AD risk.
The use of Bayesian penalized regression (BayesR+) is a notable methodological choice, as it can handle complex genetic architectures and account for relatedness within cohorts. The integration of GWAS, EWAS, and MR provides a comprehensive approach to understanding the molecular underpinnings of protein levels.
Datasets
The primary dataset used in this study was from the Generation Scotland cohort, specifically the STratifying Resilience and Depression Longitudinally (STRADL) subset.
- Genotyping data: Available for 1064 individuals.
- Plasma proteomic data: Measured using the SOMAscan platform for 1065 individuals.
- DNA methylation data: Measured using the Illumina Human MethylationEPIC BeadChip Array for 778 individuals.
The study also utilized publicly available data from eQTLGen, phenoscanner, and GoDMC for co-localization analyses.
Key Insights and Ideas
- TBCA and TREM2 as Key Players: The study highlights the significant association of plasma levels of Tubulin-Specific Chaperone A (TBCA) and Triggering Receptor Expressed on Myeloid Cells 2 (TREM2) with AD risk. Higher levels of these proteins were linked to lower AD risk.
- Genetic and Epigenetic Regulation of Protein Levels: A substantial number of genetic loci (64) and epigenetic loci (26) were found to be associated with the levels of 45 different plasma proteins, underscoring the complex interplay of genetics and epigenetics in determining protein abundance.
- Novel Associations Identified: The research identified seven novel protein quantitative trait loci (pQTLs) and 19 novel CpG sites associated with AD-related proteins, expanding our understanding of the genetic and epigenetic landscape of these biomarkers.
- Interplay of Genetic and Epigenetic Factors: The study demonstrated that both genetic and DNA methylation factors contribute to inter-individual differences in plasma protein levels, with their combined effects often explaining a significant portion of the observed variation.
- Causal Inference for Disease Risk: Mendelian randomization provided evidence for a causal relationship between plasma protein levels of TBCA and TREM2 and AD risk, suggesting these proteins are not merely correlates but may play a role in disease pathogenesis.
Key Results
- 64 independent genetic loci and 26 epigenetic loci were associated with 45 plasma proteins.
- Seven novel pQTLs were identified, including one for TREM2.
- Nineteen novel CpG sites were associated with 14 unique proteins.
- Plasma levels of TREM2 and TBCA showed strong associations with AD risk, with higher levels correlating with lower risk.
- CSF3 also showed a unidirectional association with AD risk.
- Co-localization analyses suggested that a single variant was associated with TREM2 levels and AD risk, while two separate variants were associated with CSF3 levels and AD risk.
- The study estimated that genetic factors explained a median of 13.0% of the variability in plasma protein levels, while DNA methylation explained a median of 10.0%.
Limitations and Future Work
The authors acknowledge several limitations:
- Limited scope of AD-related traits: The literature review did not encompass all possible AD-related traits, and there was heterogeneity in diagnostic criteria and phenotype definitions across studies.
- Platform specificity: The focus on the SOMAscan platform means that other blood protein correlates of AD reported in the literature were not captured.
- MR limitations: An insufficient number of variants were available for robust testing of horizontal pleiotropy in MR analyses.
- Potential for technical artifacts: Variants might influence SOMAmer reactivity or reflect technical artifacts, rather than true biological effects.
- Sample generalizability: The study sample consisted of Scottish individuals with a relatively homogenous genetic background, which may limit the generalizability of the findings to other populations.
Future work should focus on:
- Triangulating evidence: Integrating findings with other experimental and epidemiological approaches to confirm if blood proteins influence AD risk.
- Investigating TBCA dysregulation: Further research is needed to determine if TBCA dysregulation is a cause or consequence of disease risk mechanisms, particularly in carriers of APOE ε4 polymorphisms.
- Exploring MAPKAPK5 and TMEM97: The study suggests MAPKAPK5 as a potential downstream effector or correlate of TMEM97 in Aβ clearance, warranting further investigation.
Summary updated