Mutations affecting cell size have a modest effect on the coefficient of variation of size within the population.

(A) Distributions of cell size after normalizing by the population mean cell size for 2 WT (black) and 29 mutant strains (colored) of the budding yeast S. cerevisiae. Data are from (Chen et al., 2020). (B) The coefficient of variation (CVMutant = standard deviation/mean) for each size mutant normalized by the CV of WT cells plotted against the average cell size for each mutant (µMutant) population normalized to the average size of WT cells. Red dashed curve in panel A and red square in panel B corresponds to the cdh1Δ mutant with the largest change in CV compared to WT.

A mathematical model of the budding yeast cell cycle indicates that a ∼10% change in CV in response to a cell size mutation is typical.

(A) Schematic diagram of the model constructed by (Chandler-Brown et al., 2017) to describe progression through the asymmetric budding yeast cell cycle. (B) WT version of Start size control where the stochastic rate of progression through the transition depends linearly on the current cell size above a threshold V0. The resulting steady-state cell size distribution with model parameters fit to WT data is shown below. For comparison with simulations below, we find CVWT = 0.41 (C) The Start transition is modified to be a deterministic sizer where cells instantaneously progress in the cycle once they reach a threshold size. The resulting steady-state size distribution is shown below with a CV = 0.92 CVWT. (D) The Start transition is modified to take place at a constant stochastic rate, resulting in an effective pre-Start timer with some noise due to stochasticity. The resulting steady-state size distribution is shown below with a CV = 1.25 CVWT. In (C)-(D) all other aspects of the cell cycle model are kept constant.

Logistic regression models predicting passage through Start for mutant cells lacking key G1 regulators.

(A) Cell-ACDC (Padovani et al., 2022) was used to segment, track and analyze the cell cycles of the live cell mutants shown in panels. Phase contrast is used to identify cell boundaries. The budneck is marked with the Myo1-mKate fluorescent protein to detect cell division. (B) Deviance measurements for logistic regressions of progression through Start based on the indicated predictors show that cell size is the most informative single predictor for this transition. Data from (Chandler-Brown et al., 2017). In (B),(D) and (F), Full indicates logistic regression using all 3 predictors. Null indicates the control model with no predictors. (C) Steady-state cell size distribution of cln3Δ daughter and mother cells at all phases of their cell cycle, n = 377, CV = 1.09 CVWT. (D) Deviance measurements for indicated logistic regressions predicting the passage through Start of cln3Δ mutant cells. (E) Steady-state size distribution of cln3Δwhi5Δ daughter and mother cells at all phases of their cell cycle, n = 504, CV = 1.27 CVWT. (F) Deviance measurements for indicated logistic regressions of predicting passage through Start of cln3Δwhi5Δ mutant cells.

Parameter sensitivity analysis of the model identifies two types of asymmetry as the main drivers of the CV of the steady-state cell size distribution and of the daughter cell size distribution at the Start transition.

(A) Vectors representing the numerical gradients with respect to the logarithm of the parameters of the model. The first column represents the gradient of the CV of the steady state size distribution, and the second column represents the gradient of the CV of the daughter size distribution at the Start transition. Rows indicate each parameter’s contribution to the gradients as indicated by their names on the left, with colors representing the magnitude of each parameter’s contribution to the gradient. Below, a schematic representation of the cell cycle model is shown where colored arrows indicate the stochastic rates of transition with colors matching the names of the parameters on the rows above. (B) Resulting steady-state size distribution of mother and daughter cells upon perturbing the model parameters along the direction of the gradient of the CV of the same distribution. ε = 0.5, CV = 1.42*CVWT. (C) CV of the steady-state size distribution relative to the WT value indicated by the first black dot. The following 4 green dots correspond to parameter perturbations of the model, first along the positive and negative directions of the gradient as is done in panel (B) with ε = ±0.5, then by changing the G1 size control mechanism to a pure sizer or timer as is done in Fig. 2C,D. The boxplot represents the quartile distribution of relative CV when randomly and uniformly perturbing parameters with the same magnitude as before with ε = 0.5. Stars indicate statistical significance p<0.01 after performing a paired sample t-test between the directed parameter perturbations and the random parameter perturbations. (D) Relative CV of the steady-state size distribution of cells plotted against the asymmetry of the mother and daughter sub-distributions as quantified using the Jensen-Shannon divergence. Data of previous panels of this figure along with the experimental data for the Δcln3 and Δcln3Δwhi5 mutant cells shown in Fig. 3C,E. (E) Resulting daughter size distribution at the Start transition upon perturbing the parameters along the direction of the gradient of the CV of the same distribution. ε = 0.5, CV = 1.67 CVWT. (F) CV of the daughter size distribution at the Start transition relative to the WT value, similar to panel (C). (G) Relative CV of the daughter size distribution at Start against the asymmetry of the daughters born from 1st generation cells and the daughters born from 2nd generation or later cells sub-distributions as indicated by the Jensen-Shannon divergence. Data of previous panels of this figure along with the experimental data for Δcln3 and Δcln3Δwhi5 mutant cells shown in Fig. 3C,E.

Mother-daughter asymmetry is highly correlated with population CV for S. cerevisae cells growing in different nutrient conditions.

(A) Steady-state cell size distributions for S. cerevisae cells growing in the indicated nutrient condition. (B) Composite phase and fluorescence images of cells dividing into smaller daughter and larger mother cells in the indicated nutrient conditions. (C) CV of the steady state cell size distribution plotted as a function of the exponential growth rate for cells growing in that condition, computed from Coulter Counter distributions. The growth rate was altered by growing cells in synthetic complete media with different carbon sources. Error bars indicate standard deviation of measurements for 2 or 3 replicates. (D) CV at steady state plotted against the median of the volume asymmetry between pairs of mother cells (VM) and daughter cells (VD) at birth, using microscopy data allowing to monitor cell pedigree (one experiment per condition). Horizontal error bars indicate the standard deviation of the asymmetry ratio at division for cells growing in a given nutrient condition of a given genotype.