A systematic analysis of gene-gene interaction in multiple sclerosis

Lotfi Slim, Clément Chatelain, Hélène de Foucauld, Chloé-Agathe AzencottbioRxiv 2020 · DOI

TL;DR

This paper introduces a novel gene-level epistasis analysis framework, epiGWAS, to uncover complex gene-gene interactions in Multiple Sclerosis (MS). By analyzing genome-wide association study (GWAS) data and integrating biological pathway information, they identified known and novel gene interactions, offering new insights into MS pathogenesis and potential therapeutic targets.

What Problem Does This Paper Address?

Multiple Sclerosis (MS) is a complex autoimmune disease with a significant genetic component. While Genome-Wide Association Studies (GWAS) have identified numerous genetic loci associated with MS, these studies primarily focus on individual gene effects (single-locus analysis). This approach often misses crucial "epistatic" interactions, where the combined effect of two or more genes is greater than the sum of their individual effects. Understanding these gene-gene interactions is vital for a comprehensive understanding of MS and for developing more effective treatments. The authors aim to address this gap by systematically investigating gene-level epistasis in MS.

Methods

The core of this work is the extension of the epiGWAS framework, originally designed for SNP-level epistasis detection, to a gene-level analysis. The authors employ two primary mapping strategies to link genetic variants to genes:

  1. Physical Mapping: This involves selecting SNPs located within the boundaries of a gene, focusing on variants that might directly affect protein structure and function.
  2. eQTL (expression Quantitative Trait Loci) Mapping: This approach uses SNPs that are significantly associated with the expression levels of nearby genes. This allows the study of epistasis through variations in gene expression.

For both mapping strategies, they systematically test all pairs of genes within specific MS-related biological pathways (obtained from MetaCore). The analysis involves calculating interaction scores and then ranking these interactions. To further refine the findings, they apply filtering criteria based on the potential functional impact of variants (for physical mapping) and the consistency of gene expression effects (for eQTL mapping).

Datasets

The study utilizes the Multiple Sclerosis (MS) dataset from the second release of the Wellcome Trust Case Control Consortium (WTCCC2). This dataset comprises genotypic data from MS cases and controls. For biological context and pathway information, they use MetaCore disease maps related to multiple sclerosis. Additionally, they leverage cis-eQTL data from the eQTLGen consortium for the eQTL mapping strategy.

Key Insights and Ideas

  • Gene-level epistasis is crucial for understanding complex diseases: The study highlights that focusing solely on individual genes misses critical interactions that contribute to disease risk.
  • Complementarity of mapping strategies: Both physical mapping (focusing on protein structure) and eQTL mapping (focusing on gene expression) successfully identified gene interactions, and their combination provided a more robust understanding.
  • Known interactions validated: The method successfully rediscovered known gene interactions relevant to MS, such as the direct binding between GLI-1 and SUFU, and the regulation of IP10 transcription by NF-κB. This validation lends confidence to the approach.
  • Potential for novel discoveries: The analysis identified numerous novel gene pairs with epistatic interactions, suggesting new biological pathways and mechanisms involved in MS.
  • Directionality of synergy: The study explored the directionality of gene-gene synergy, indicating whether the interaction increases or decreases disease risk, which is crucial for therapeutic development.

Key Results

The analysis yielded a significant number of gene-gene interactions:

  • Physical Mapping: Identified 4 gene pairs with epistasis involving missense variants. One notable pair was GLI-1 and SUFU, known for their direct binding and role in oligodendrocyte precursor cell differentiation.
  • eQTL Mapping: Identified 117 gene pairs with epistasis mediated by eQTLs. A key finding was the interaction between IP10 and NF-κB, both of which have known roles in MS.
  • Enrichment Analysis: The discovered subnetworks showed an enrichment of known therapeutic targets, suggesting their relevance for future drug development.
  • Directionality: For eQTL strategy, the interaction coefficient $\alpha_{12}$ was positive in 47% of cases, and the directionality of synergy was split equally between positive and negative.

Limitations and Future Work

  • Biological Validation: While the study provides statistical evidence and biological context, the authors acknowledge that more biological and experimental validation is needed to confirm the discovered interactions.
  • Computational Cost: Exhaustive analysis of all possible gene-gene interactions is computationally intensive, necessitating a posteriori filtering.
  • Interpretation of Statistical vs. Biological Epistasis: The authors note the potential mismatch between statistical epistasis (as detected by GWAS) and true biological epistasis, which can make direct biological interpretation challenging.
  • Future Work: The authors suggest that the discovered novel subnetworks can aid in understanding MS by unraveling new disease mechanisms. They also propose that the developed pipeline can be generalized to study gene-gene interactions in other diseases. The identification of synergistic effects is particularly relevant for developing combination therapies, aligning with new FDA guidelines.

Summary updated

Abstract

Abstract Multiple sclerosis is a complex autoimmune disease which genetic basis has been extensively investigated through genome wide association studies. So far, the conducted studies have detected a number of loci independently associated with the disease but few have investigated the interaction between distant loci, or epistasis. In the present work, we perform a gene level epistasis analysis of multiple sclerosis GWAS from the Wellcome Trust Case Control Consortium 2. We systematically study the epistatic interactions between all pairs of genes within 19 multiple sclerosis disease maps from the MetaCore pathway database. We report 4 gene pairs with epistasis involving missense variants, and 117 gene pairs with epistasis mediated by eQTLs. Our epistasis analysis is able to retrieve known interactions linked to multiple sclerosis: direct binding interaction between GLI-I and SUFU, involved in oligodendrocyte precursor cells differentiation, and regulation of IP10 transcription by NF- κ B, thus validating the potential of epistasis analysis to reveal biological interaction with relevance in a disease specific context.
BibTeX
@misc{slim2020systematic,
  author = {Lotfi Slim and Clément Chatelain and Hélène de Foucauld and Chloé-Agathe Azencott},
  title = {{A systematic analysis of gene-gene interaction in multiple sclerosis}},
  year = {2020},
  doi = {10.1101/2020.05.28.121939},
  howpublished = {bioRxiv},
}
Figure