Understanding the Aetiology of Metabolic Diseases and Related Phenotypes through Human Genetics

Abstract

A key objective in human genetics is to elucidate how inherited genetic variations contribute to the phenotypic variations and disease susceptibility in the population. Whole-exome sequencing (WES) is a pivotal approach for uncovering gene-disease associations by analysing rare variants that can lead to significant changes in protein function, thereby establishing clear links between protein functions and diseases. Another important method, Mendelian randomisation (MR), leverages genetic variants as natural experiments to investigate potential causal relationships between modifiable risk factors and diseases. This thesis used both approaches to explore the genetic basis of metabolic diseases. Using the extensive WES and phenotyping data from approximately 450,000 participants in the UK Biobank (UKBB), I performed several studies on rare variant burden tests. I first conducted a comprehensive computational experiment to evaluate various rare variant selection strategies for gene-based burden tests. The results informed the development of the analytical pipeline and the selection of computational tools, with the aim of maximising the discovery power of disease-associated genes. I then applied the gene-burden test to identify protein-coding genes associated with type 2 diabetes (T2D) that showed heterogeneous effects between biological sexes. Four T2D risk-associated genes were identified, three of which demonstrated evidence of sex-specific effects. Although these findings were not replicated in a smaller independent cohort, they provided novel hypotheses for further investigations. Studying rare variants offers valuable insights into the genetic architecture of common complex diseases, helping to bridge the gap between monogenic and polygenic disease models. I investigated the contribution of rare variants in obesity susceptibility, focusing on nine genes initially identified in patients with severe early-onset obesity. This work first showed that rare variants previously identified in patients with suspected monogenic obesity that had loss-of-function (LoF) effects in molecular assays, did not necessarily translate into pathogenicity (disease-causing). This was evident from their high allele frequency, low variant-level obesity penetrance, and null association with adult BMI (a continuous proxy trait for obesity) in the UKBB. At the gene-level, I confirmed that MC4R, POMC, and PCSK1 exhibit haploinsufficiency effects on adult BMI using gene burden tests of rare variants with predicted LoF effects. From a clinical genetics perspective, this study underscored the utility of biobank datasets in informing the variant classification process for patients with suspected genetic causes of obesity. Finally, using the largest dataset on circulating metabolites measured by nuclear magnetic resonance in the UKBB in ~250,000 participants, I applied the MR approach to examine whether the circulating level of branched-chain amino acids (BCAA) – that are essential nutrients in human health – could be potentially a modifiable target for improving metabolic health. Through a bidirectional MR analysis, I showed that genetically predicted higher level of circulating BCAA causally reflect poorer metabolic health. The findings also suggested a potential feed-forward loop between elevated BCAA and dyslipidaemia, which may contribute to disease pathogenesis. This presents a novel hypothesis for future mechanistic studies. In summary, this thesis used human genetics methods to elucidate the genetic factors contributing to metabolic diseases. Key findings include the identification of rare variant effects in protein-coding genes associated with the susceptibility to T2D and obesity, and novel evidence of a bidirectional causal relationship between BCAA and lipid metabolism.