A Stochastic Model of the Yeast Cell Cycle Reveals Roles for Feedback Regulation in Limiting Cellular Variability

Abstract

The cell division cycle of eukaryotes is governed by a complex network of cyclin-dependent protein kinases (CDKs) and auxiliary proteins that govern CDK activities. The control system must function reliably in the context of molecular noise that is inevitable in tiny yeast cells, because mistakes in sequencing cell cycle events are detrimental or fatal to the cell or its progeny. To assess the effects of noise on cell cycle progression requires not only extensive, quantitative, experimental measurements of cellular heterogeneity but also comprehensive, accurate, mathematical models of stochastic fluctuations in the CDK control system. In this paper we provide a stochastic model of the budding yeast cell cycle that accurately accounts for the variable phenotypes of wild-type cells and more than 20 mutant yeast strains simulated in different growth conditions. We specifically tested the role of feedback regulations mediated by G1- and SG2M-phase cyclins to minimize the noise in cell cycle progression. Details of the model are informed and tested by quantitative measurements (by fluorescence in situ hybridization) of the joint distributions of mRNA populations in yeast cells. We use the model to predict the phenotypes of ~30 mutant yeast strains that have not yet been characterized experimentally.

A Hybrid Stochastic Model of the Budding Yeast Cell Cycle

“…As expected, the unregulated transcripts follow Poisson distributions, which are consistent with experimental measurements. The cell-cycle regulated transcripts, which follow long-tailed, non-Poisson distributions, are well-fit by two-component Poisson distributions as reported by [26,28]. (We note that in our model mClb2 represents both mClb1 and mClb2, and mCln2 = mCln1 + mCln2, whereas in the experiment these cyclin mRNAs are tracked independently.…”

“…FORTRAN code takes about 15 min to simulate 10,000 cell cycle on an Intel i7-3770 processor with 16G memory running a Linux environment. A similar system using a fully stochastic model may take more than one day (the FORTRAN code by Barik et al [26] is run using the same work station; it takes more than 30 hours to generate similar computational population of yeast cells).…”

“…Extensive experimental studies have been conducted particularly on the budding yeast (Saccharomyces cerevisiae) to explore gene regulation and signaling pathways of relevance to cell growth and division [7][8][9]. Moreover, various modeling approaches, such as deterministic models [10][11][12], Boolean networks [13][14][15][16][17][18][19], and stochastic models [20][21][22][23][24][25][26][27], have been adopted to explore the roles of different gene and protein interactions in robust progression through the cell cycle.…”

Genetic interactions derived from high-throughput phenotyping of 7,350 yeast cell cycle mutants

“…Mass doubling times are longer on YPGalactose (~150 min) and even longer on YPRaffinose (~200 min) 47 . Slower growth rates can enable positive regulators to build up such that mutants which would normally grow very slowly due to the stochasticity of cell cycle transitions can exhibit some level of rescue on YPG or YPR 23, 48 .…”

Stochasticity in multi-phosphorylation and quasi steady state approximation in stochastic simulation

“…However cellular functions are regulated by chemical reaction networks involving many proteins and these networks often consists of small regulatory network motifs with distinct properties (16). Therefore further investigations were carried out to understand the effects of network topology such as signaling cascades, feedback loops and feedforward loops on the propagation of chemical noise (17)(18)(19)(20)(21)(22). These protein interaction networks often consist of multisite reversible covalent modifications of target proteins such as phosphorylation, acetylation, methylation, ubiquitylation.…”