Reconciled Rat and Human Metabolic Networks for Comparative Toxicogenomics Analyses 1914 views

Rats serve an integral role in drug development and biomarker discovery, and understanding metabolic differences between rats and humans is critically important to minimize unexpected toxicities in clinical trials. Despite a high degree of physiologic and genomic similarities between rats and humans, several metabolic differences have been described that could affect whether a biomarker is elevated or whether a compound is toxic to the liver. A comprehensive knowledgebase of functional differences between rat and human metabolism would dramatically improve the translation of preclinical studies to human trials. A genome-scale network reconstruction of metabolism serves as a repository for all known biochemical and transport reactions for an organism. In this dissertation, I have built the first genome-scale reconstruction of Rattus norvegicus metabolism, iRno, and a significantly improved reconstruction of Homo sapiens metabolism, iHsa. Comparative analyses with these models captured functional features known to distinguish rats from humans within purine, glycan, ascorbate, and bile acid metabolic pathways. Using reconciled biomass formulations, iRno and iHsa recapitulated realistic cellular growth rates under physiological constraints. After extensive manual curation and network reconciliation, I demonstrated the use of iRno and iHsa in systems toxicology by generating biomarker predictions for rat and human hepatocytes treated with 76 pharmaceutical compounds and environmental toxicants from a comparative toxicogenomics database. I developed a novel gene expression integration algorithm to generate biomarker predictions that can be evaluated across metabolites, treatments, and organisms. Biomarker predictions were validated with literature-based evidence for antipyretic and antigout medicines. Comparative analyses provided mechanistic insights into the selection of metabolite biomarkers common to rats and humans. Using metabolomics and transcriptomics profiles from rat hepatocytes, I performed high-throughput validation of biomarker predictions. In the future, I anticipate that these models will serve as powerful computational platforms for contextualizing experimental data and making functional predictions consistent with rat and human biology for clinical and basic science applications.

PHD (Doctor of Philosophy)

computational systems biology; toxicology; preclinical drug development; bioinformatics; metabolic network modeling; comparative genomics; rats

English

All rights reserved (no additional license for public reuse)

2016-04-25