Abstract
Mitochondrial dysfunction is associated with many human diseases, including cancer and neurodegeneration, that are often linked to proteins and pathways that are not well-characterized. To help define the functions of such poorly characterized proteins and determine how their expression is regulated, we are performing deep mass spectrometry profiling to map the proteomes, lipidomes, and metabolomes of yeast strains and human cell lines, each lacking a single gene related to mitochondrial biology. To explore the resulting large-scale data, we devised a multi-omic data analysis and visualization tool that we use to find covariance networks that can predict molecular functions, identify correlations between profiles of related gene deletions, spotlight gene-specific perturbations that reflect protein functions, and elucidate a global respiration deficiency response that results from mitochondrial dysfunction. We used this approach to elucidate uncharacterized features of mitochondrial coenzyme Q (CoQ) biosynthesis—an essential pathway disrupted in many human diseases. In particular, our analyses defined previously unrecognized roles for Hfd1p—and its human homolog ALDH3A1—in producing the CoQ precursor, 4-hydroxybenzoate. We further leveraged this strategy to reveal that the RNA-binding protein, Puf3p, regulates expression of a select set of proteins involved in mitochondrial protein import, translation, complex assembly, and coenzyme Q (CoQ) biosynthesis—pathways essential to prime the full mitochondrial biogenesis program. Collectively, our results provide molecular insight into various aspects of mitochondrial biology and, more broadly, establish a high-throughput, multi-omic approach for quantifying diagnostic phenotypes and defining protein functions.