PgmNr P347: The cost of noise in biochemical reactions and the evolutionary limits of cellular robustness.

Authors:
J. Van Dyken


Institutes
University of Miami, Coral Gables, FL.


Abstract:

Finite cell size and the probabilistic nature of biochemical reactions combine to generate substantial cellular “noise”, i.e., random fluctuations in molecular abundance. An open question is whether noise affects cell function, and, if so, how and by how much? Here I demonstrate that noise causes a quantifiable loss in cell fitness by slowing the average rate of biosynthesis. The efficacy of selection to suppress noise, however, decays superlinearly with cell size, leading to a stark taxonomic divide between prokaryotes and eukaryotes in the principle design imperatives of genomic and cellular architecture. 

 Extending previous work, I present an analytical framework for solving the steady-state statistics of molecular species in arbitrarily connected networks of non-linear chemical reactions in mesoscopic volumes, which I validate with extensive stochastic simulations. In general, substrate noise slows the average rate of non-cooperative bimolecular (i.e., Michaelis-Menten or hyperbolic) reactions, including reactions intimately tied to organismal fitness: nutrient uptake and translation. I find that gene expression is unusually sensitive to the cost of noise because, unlike with metabolic reactions, noise-induced slowdown is not buffered by network-level feedbacks. In addition, I find that transcriptional noise directly reduces fitness by slowing the average translation rate. In general, the loss in fitness caused by noise scales inversely with cell size, while random genetic drift scales positively with cell size. Together, the ability of natural selection to suppress noise decays superlinearly with cell size, leading to a stark, cell-size-mediated taxonomic divide in selection pressures for gene regulatory architecture. Greatly weakened selection efficacy in large cells may have facilitated the evolution of transcriptional complexity in eukaryotes as they evolved free of the selective constraints of noise minimization experienced by most prokaryotes. Furthermore, hyperbolic translation kinetics substantially influence the noise statistics of gene expression, with numerous practical implications. In particular, with hyperbolic translation, increasing mRNA/ribosome binding affinity actually reduces protein noise, even as it increases translational bursting. These results thus illuminate the costs of cellular noise, the targets of noise-amelioration, and the evolutionary limits of robustness.