Principles of bursty mRNA expression and irreversibility in single cells and extrinsically varying populations

Abstract

The canonical model of mRNA expression is the telegraph model, describing a gene that switches on and off, subject to transcription and decay. It describes steady-state mRNA distributions that subscribe to transcription in bursts with first-order decay, referred to as super-Poissonian expression. Using a telegraph-like model, I propose an answer to the question of why gene expression is bursty in the first place, and what benefits it confers. Using analytics for the entropy production rate, I find that entropy production is maximal when the on and off switching rates between the gene states are approximately equal. This is related to a lower bound on the free energy necessary to keep the system out of equilibrium, meaning that bursty gene expression may have evolved in part due to free energy efficiency. It is shown that there are trade-offs between having slow nuclear export, which can reduce cytoplasmic mRNA noise, and the energy required to keep the system out of equilibrium -- nuclear compartmentalization comes with an associated free energy cost. At the population level, I find that extrinsic variation, manifested in cell-to-cell differences in kinetic parameters, can make the system more or less reversible -- and potentially energy efficient -- depending on where the noise is located. This highlights that there evolutionary constraints on the suppression of extrinsic noise, whose origin is in cellular heterogeneity, in addition to intrinsic randomness arising from molecular collisions. Finally, I investigate the partially observed nature of most mRNA expression data which seems to obey detailed balance, yet remains unavoidably out-of-equilibrium.