The Metabolic Role of MAP3K15: Genetic and Phenotypic Insights from the 23andMe Research Database and Genetics-Driven Recruitment

Jennifer J. Brady, Kira Kalkus, Dominique T. Nguyen, Jingran Wen, Zachary L. Fuller, Yue Qi, Shaeloren S. Deering, Leah Selcer, Suyash Shringarpure, Michael V. Holmes, Xin WangmedRxiv 2024 · DOI

This paper, published by researchers at 23andMe, explores the metabolic impact of the gene MAP3K15. While previous studies linked this gene to Type 2 Diabetes (T2D) protection, this research uses the massive 23andMe database to show that loss-of-function (LoF) variants in MAP3K15 don't just lower risk—they significantly delay disease onset and protect against high cholesterol. Most importantly, the authors demonstrate a "Genetics-Driven Recruitment" (GDR) framework, successfully recalling participants based on rare genotypes to validate these findings with fresh clinical lab data.

TL;DR

Researchers leveraged 23andMe’s database of millions to show that MAP3K15 "natural knockouts" experience a 4.5-year delay in Type 2 Diabetes onset and have significantly lower cholesterol levels. The study highlights a powerful new paradigm for drug discovery: identifying individuals with rare protective mutations and recruiting them for clinical follow-up to validate therapeutic targets.

What Problem Does This Paper Address?

Drug discovery is notoriously difficult and expensive. One way to "de-risk" a potential drug target is to find "human knockouts"—people who naturally lack a functional version of a gene without ill effects. If these individuals are healthier (e.g., protected from diabetes), then a drug that inhibits that same gene is likely to be both safe and effective.

However, finding these individuals is a "needle in a haystack" problem because these loss-of-function (LoF) variants are extremely rare. Furthermore, even when identified in large biobanks, researchers often lack the ability to bring those specific individuals into a clinic for fresh blood work or deep phenotyping. This paper addresses both the biological role of MAP3K15 and the logistical challenge of genotype-based participant recall.

Methods

The researchers employed a multi-stage approach:

  1. Burden Testing: Instead of looking at single mutations, they "collapsed" various rare LoF variants in MAP3K15 into a single score to increase statistical power, testing for associations with 1,149 phenotypes.
  2. Polygenic Risk Interaction: They used Polygenic Risk Scores (PRS) to see how MAP3K15 LoF variants interact with a person's underlying genetic "background" risk for T2D.
  3. Genetics-Driven Recruitment (GDR): This is the paper's most novel methodological contribution. They identified 23andMe customers who carried a specific rare LoF variant (rs148312150) and invited them—along with age- and sex-matched controls—to participate in a new study involving mobile phlebotomy (blood draws) and clinical lab testing.
  4. Age-Stratified Analysis: They compared their results across different age cohorts to understand why previous studies (like those using the UK Biobank) might have missed certain lipid associations.

Datasets

  • 23andMe Research Cohort: A database of over 11 million genotyped individuals, with 5.2 million consented for recontact.
  • UK Biobank & FinnGen: Used as comparison datasets to replicate T2D findings and investigate discrepancies in cholesterol data.
  • PGS Catalog: Weights for the T2D Polygenic Risk Score were pulled from accession ID PGS000713.

Key Insights and Ideas

  • Buying Time: The most striking finding is that MAP3K15 LoF carriers at high polygenic risk for T2D saw a median delay of 4.5 years in their diagnosis compared to non-carriers. This suggests that inhibiting MAP3K15 could be a potent preventative strategy, even for those genetically predisposed to the disease.
  • The "Aha!" on Cholesterol: Previous studies in the UK Biobank failed to find a link between MAP3K15 and cholesterol. This paper suggests that's because the UK Biobank cohort is older (ages 40-69). In older populations, statin use and long-term environmental factors "mask" the genetic signal. By looking at 23andMe’s younger cohort (including those under 40), the protective effect on lipids became clear.
  • X-Chromosome Advantage: MAP3K15 is located on the X chromosome. This makes it a unique "natural experiment" because males are hemizygous; a single LoF variant results in a complete "knockout" of the gene, making the biological effects easier to observe than in autosomal genes.

Key Results

  • T2D Protection: Confirmed a strong protective association with T2D (OR = 0.836, p = 3.6e-63).
  • Novel Lipid Link: Discovered a novel association with protection against high cholesterol (OR = 0.961, p = 2.2e-8).
  • Clinical Validation: In the GDR pilot (31 cases vs. 31 controls), LoF carriers showed a modest reduction in median cholesterol and a significantly lower LDL/HDL ratio (p = 0.0444).
  • Blood Pressure Trend: Consistent with mouse models, there was a slight trend toward higher systolic blood pressure in LoF carriers, though it was not statistically significant in this small clinical sample.

Limitations and Future Work

  • Recruitment Constraints: The clinical pilot was limited to participants within a 50-mile radius of mobile phlebotomy services, leading to high attrition during the recruitment phase.
  • Sample Size: While the 23andMe database is huge, the number of people who actually completed the clinical blood draw was small (N=71), limiting the power to detect subtle changes in biomarkers like HbA1c.
  • Ethical Considerations: The authors highlight the "right not to know." Participants were recruited based on genetics unknown to them, and the researchers chose not to disclose their "case" or "control" status to avoid potential psychological harm from unvalidated genetic info.
  • Future Directions: The authors suggest that future GWAS should more routinely use age-stratification to uncover genetic effects that are obscured by medication or aging in older cohorts.

Summary updated

Abstract

Abstract MAP3K15 has been previously associated with protection from type 2 diabetes (T2D), prompting interest in the development of MAP3K15 inhibitors as a potential therapeutic option for diabetes. The trans-ancestry genome-wide association study (GWAS) meta-analysis and loss-of-function (LoF) burden testing methods that implicate association with T2D greatly benefit from large sample size. The direct-to-consumer genetic testing company, 23andMe, Inc., is the world’s largest research consented genetic database. We leveraged the 23andMe database to further inform the metabolic role of MAP3K15, using a variety of genetic analysis methods. We find that MAP3K15 LoF carriers show a significant delay of 4.5 years in the median age of T2D diagnosis among individuals at high polygenic risk and uncover a novel burden association of MAP3K15 LoF with protection against high cholesterol. We expanded these findings by establishing a capability to recruit consented participants on the basis of genetics unknown to them (specifically, a single LoF variant in MAP3K15, rs148312150), and obtained clinical laboratory evidence of a modest reduction in median cholesterol and LDL/HDL ratio in MAP3K15 LoF carriers. Our findings demonstrate the discovery power of the 23andMe database, including the feasibility of consented participant recruitment to inform therapeutic discovery and development.
BibTeX
@misc{brady2024metabolic,
  author = {Jennifer J. Brady and Kira Kalkus and Dominique T. Nguyen and Jingran Wen and Zachary L. Fuller and Yue Qi and Shaeloren S. Deering and Leah Selcer and Suyash Shringarpure and Michael V. Holmes and Xin Wang},
  title = {{The Metabolic Role of MAP3K15: Genetic and Phenotypic Insights from the 23andMe Research Database and Genetics-Driven Recruitment}},
  year = {2024},
  doi = {10.1101/2024.01.09.24301012},
  howpublished = {medRxiv},
}
Figure