Emergence and Transmission of Noise Created at Cell Division

Abstract

Molecules in cells collide and react randomly, creating stochastic fluctuations in synthesis and degradation which creates heterogeneity among genetically identical cells in the identical environments. However, the random motion of cellular components also creates spatial heterogeneity which at cell division means that the two sister cells can receive different amounts of each component. Such partitioning errors could in principle explain much of the heterogeneity that now is attributed to stochastic gene expression, but has been largely unexplored. This thesis aims to study the contribution of partitioning noise at cell division on cellular variability. The effect of partitioning errors on the overall heterogeneity can also accumulate over multiple divisions, depending on how quickly deviations are corrected during the cell cycle. Thus, its effect can only be properly estimated using accurate time-lapse measurements over many single cells while keeping track of sibling cells. In the first part, we present a microfluidic device that enables high- throughput and accurate measurements under exceptionally homogeneous growth condition both in time and space. In the second part, using Schizosaccharomyces pombe as a model organism, we systemically measure partitioning noise for high-abundance proteins with various localizations and reveal that a significant amount of partitioning noise composes protein noise at birth, ranging from 33% of cytoplasmic proteins to 57% of vacuolar proteins. Next, by leveraging our microfluidic device, we directly measure how partitioning noise is corrected over cell cycle. Surprisingly, all of the measured strains displayed a simple exponential correction curve with a half-life of 1 generation, as expected when cells produce the same amount regardless of their starting value. That is, the correction curve is fully described by a passive control model in which fluctuations are corrected by regression to the mean without feedback. We further use this model to identify the total contribution to protein noise from partitioning errors, and thus also to identify the noise component that appears to come from stochastic gene expression. Overall, our work demonstrates the significance of random partitioning on protein noise.