Modeling single-cell phenotypes links yeast stress acclimation to transcriptional repression and pre-stress cellular states

Abstract

Stress defense and cell growth are inversely related in bulk culture analyses; however, these studies miss substantial cell-to-cell heterogeneity, thus obscuring true phenotypic relationships. Here, we devised a microfluidics system to characterize multiple phenotypes in single yeast cells over time before, during, and after salt stress. The system measured cell and colony size, growth rate, and cell-cycle phase along with nuclear trans-localization of two transcription factors: stress-activated Msn2 that regulates defense genes and Dot6 that represses ribosome biogenesis genes during an active stress response. By tracking cells dynamically, we discovered unexpected discordance between Msn2 and Dot6 behavior that revealed subpopulations of cells with distinct growth properties. Surprisingly, post-stress growth recovery was positively corelated with activation of the Dot6 repressor. In contrast, cells lacking Dot6 displayed slower growth acclimation, even though they grow normally in the absence of stress. We show that wild-type cells with a larger Dot6 response display faster production of Msn2-regulated Ctt1 protein, separable from the contribution of Msn2. These results are consistent with the model that transcriptional repression during acute stress in yeast provides a protective response, likely by redirecting translational capacity to induced transcripts.

Keywords: S. cerevisiae; cell biology; cell signaling; fitness; heterogeneity; stress response; yeast.

Figure 1.. Experimental approach.

(A) Schematic of Msn2 and Dot6 localization in the absence (left) and presence (right) of stress. (B) Diagram of microfluidic device used for time-lapse microscopy. (C) Representative nuclear localization scores (see Methods) for pre-stress growth, the acute-stress response, and the acclimation phase. (D) Cell or two-cell colony size was estimated by the number of pixels within the mask for each colony, and growth rates were calculated based of regression of those points during the pre- or post-stress phases. Cell volume change was reflected in the difference in pixel number before and after stress.

Figure 1—figure supplement 1.. Cellular response to salt within microfluidics device.

Time lapse of a single budding cell expressing Dot6-GFP (center panel) and Msn2-mCherry (right panel) before and after NaCl stress. Halogen images (left panel) were analyzed to identify cells and track them throughout the time course. Images were taken every 6 min, with salt added after 72 min. Top left image is timepoint T1, and timepoints continue from left to right and top to bottom. Salt was added after timepoint T12. The halogen images were used to measure colony size, while the Dot6-GFP and Msn2-mCherry images were used to measure transcription factor nuclear localization. The cell shown is from strain AGY1328; however, it is representative to what was also observed with other strains. For visualization, the brightness of the mCherry channel was increased by 50%.

Figure 1—figure supplement 2.. Growth rate estimates are robust.

(A) Growth rate was estimated based from the linear fit of collapsed-image pixel area versus time, from timepoints T1-T12 before addition of salt. The change in pixel area was highly linear for most cells (median R²=0.92, grey box plot). To test the robustness to time points considered, we performed a sliding-window analysis in which growth rates were calculated from subsets of timepoints. The linear fit remained high, and estimated growth rates were well correlated with the growth rates calculated from all pre-stress timepoints. (B) The median of sliding-window growth rates plotted against rates estimated from all timepoints is shown in black, whereas comparisons to sliding-window measurements match the colors from (A). (C–D) Same plots except for post-stress growth rate. As reported in the main text, there was a wider range of growth rates and thus a wider range of linear fits of the data (median R²=0.72). Nonetheless, the measured growth rates were highly correlated with those calculated from sliding windows of fewer timepoints. (E) As might be expected, growth rate of cells that recovered growth after NaCl stress were well estimated by a linear change in colony area, whereas cells that did not recover pre-stress growth rates showed a lower linear fit that was more heavily influenced by noise (confirmed by visual inspection). (F) The reduction in growth rate seen after NaCl treatment (here as in Figure 2B, blue plot) were specific to stress treatment (median ln(growth rate change)=–0.85), since most cells exposed to a shift in media without NaCl (grey bars) showed subtle changes in growth rate (median ln(growth rate change)=–0.20). Together, these analyses show that our estimates of growth rate are robust to time points used and that growth-rate changes discussed in the text are specific to NaCl stress.

Figure 2.. Cell-to-cell heterogeneity in the NaCl stress response.

(A-C) Shown are the distributions of the natural log of (A) colony growth rates before stress, (B) the change in growth rate after NaCl stress compared to before stress, and (C) the maximum change in cell pixel size during the acute-stress response versus during the pre-stress phase.

Figure 3.. Nuclear translocation dynamics of Msn2 and Dot6 are more coordinated before stress.

(A) Representative traces of Msn2 and Dot6 in the same cell. (B) The average number of coordinated peaks for Msn2 and Dot6, i.e. peaks called within 6 min (1 timepoint) of each other. (C) The average number of nuclear localization peaks per cell for Msn2 (red) and Dot6 (blue) during pre-stress and acclimation phases. (D–E) The average (black line)+/- one standard deviation (colored spread) of Msn2 (D) and Dot6 (E) nuclear localization during the time course. (F) Trace of the standard deviation of nuclear localization over the time course for Msn2 (red) and Dot6 (blue).

Figure 4.. Subpopulations of cells show distinct Msn2 and Dot6 translocation dynamics.

221 cells passing quality control metrics were partitioned into sub clusters based on their population-centered nuclear translocation dynamics shown on the right. Each row represents a cell and each column in a block represents a single timepoint; time of NaCl addition is indicated with an arrow. Data on the left show the log₂ ratio of nuclear versus total Msn2 (left) or Dot6 (right) according to the orange-scale key, see Methods. Data on the right show the same data normalized to the population median at each timepoint: yellow values indicate higher-than-median nuclear localization levels and blue indicates lower-than-median nuclear localization. Cell clusters identified by the package mclust are labeled to the right.

Figure 4—figure supplement 1.. Cellular profiles are recapitulated.

Patterns seen in data from Figure 4 (strain AGY1328, left side) are identifiable in a separate series of triplicated experiments with a second strain (AGY1813: BY4741 Dot6-GFP, Msn-mCherry, Ctt1-iRFP, right side). Data from AGY1813 were clustered independently by mixed-model clustering as described for Figure 4. This analysis identified six clusters of more than three cells (several small clusters were omitted from the figure). Patterns seen in Figure 4 shown on the left were visually identified and aligned with clusters from the second experiment on the right.

Figure 5.. Cell subpopulations display different growth rates before and after stress.

(A) Correlation between the natural log of pre- and post-stress growth rates for each cell, colored according to its cell cluster in Figure 4. (B–C) Distribution of median-centered growth rates before (B) and after (C) NaCl addition, for cell clusters shown in Figure 4. Boxes are colored yellow or blue if the distribution was significantly higher or lower, respectfully, from all other cells in the analysis (Wilcoxon Rank Sum test, FDR < 0.022). Dashed line indicates the median of all cells analyzed.

Figure 5—figure supplement 1.. The strength of transcription factor nuclear localization is weakly related to cell cycle phase and budding.

Cells grow the fastest in G1 and M-phase (Goranov et al., 2009), thus cells were binned if they were in G1 or M phase or S/G2 phases at the time of NaCl exposure. (A) The peak height of nuclear localization for Msn2 and Dot6 during the acute-stress phase are plotted, with cells in S/G2 or M/G1 indicated in red or blue, respectively. (B-C) The distribution of acute-stress peak heights for Dot6 (B) or Msn2 (C) was plotted for cells in M/G1 or S/G2. Wilcoxon rank sum tests show that cells in S/G2 had slightly higher nuclear accumulation of both factors (p<0.001).

Figure 5—figure supplement 2.. Cell clusters show similar relationships with pre- and post-stress growth rates.

As shown in Figure 5 except for the independent triplicated experiments shown in Figure 4—figure supplement 1. Similar cell clusters show similar relationships in pre- and post-stress growth rates, further confirming that the trends are reproducible across biological replicates, strains, and microscopy conditions.

Figure 6.. A multi-factor model best explains variation in post-stress growth rate.

(A) A representation of the nuclear localization measurements used in the multi-factor linear regression model. (B) Factors considered in the multi-factor linear regression model; those with significant contributions are highlighted with ***. (C) The variance in ln(post-stress growth rate) explained by the multi-factor linear regression model. P-value and R² are shown at the top of the plot and cell subcluster is indicated according to the key, showing that no single cluster dominates the correlation. (D) Principal component (PC) regression of post-stress growth rate and deconvolution of contributing factors according to the key. Variance explained is listed at the top of each bar (where PC2 does not contribute to post-stress growth rate).

Figure 6—figure supplement 1.. Linear regression of individual parameters on post-stress growth rate.

(A–J) Linear regressions of individual parameters listed in Supplementary file 2. Parameters in which the false discovery rate was <0.025 (p<0.003) are bold whereas other plots are deemphasized. There is a significant fit between cell/colony size at the experiment start time and post-stress growth rate (I); however, this parameter was not significant beyond the multiple-test threshold in the multi-factor linear model (see Supplementary file 3), suggesting that much of cell-size contribution is correlated with and thus absorbed by other factors in the model. (K) Fit from a multiple linear model similar to that shown in Figure 6 except in which pre-stress growth rate was not included (coefficient set to 0).

Figure 6—figure supplement 2.. Dot6 acute-stress peak height correlates with post-stress growth rate even across cells with no difference in pre-stress growth.

As an independent approach to disentangle the contribution of Dot6 behavior and pre-stress growth rate, we analyzed only the subset cells that show no difference in pre-stress growth (between dashed lines in A). This subset of cells retains a correlation between Dot6 peak height and post-stress growth rate with nearly the same predictive power (R²=0.12, compare to Figure 6D). (A) Correlation between pre- and post-stress growth rate over all analyzed cells. (B) Same as A except for cells between the dashed lines of A. The figure shows that for this subset of cells, there is no longer a correlation between pre- and post-stress growth rate. (C) Correlation between Dot6 acute-stress peak height and post-stress growth rate for cells shown in B. Cell colors correspond to clusters from Figure 4.

Figure 7.. Dot6 activation correlates with faster Ctt1 production.

(A) The average and standard deviation (n=4) of growth rates of wild-type (black lines) and dot6∆tod6∆ cells (blue lines) in the absence (solid) and presence (dashed) of 0.7 M NaCl added at 75 min (arrow). (B) Representative traces of single-cell Ctt1 production for pairs of cells that reach similar levels of Ctt1. (C) Correlation of Ctt1 production timing (time to change 5%) versus acute-stress peak heights. (D) The two-factor model correlates with measured Ctt1 production time, with only marginal contribution of Msn2 peak height (p=0.053). Adjusted R² is shown in both figures.

Figure 7—figure supplement 1.. Dot6 acute-stress response is correlated with pre-stress transcription factor behaviors.

We noticed in Figure 4 that many subpopulations showed inverse trends in pre-stress Msn2 versus Dot6 activation. Several clusters that had higher nuclear levels of Dot6 before stress had lower levels of Msn2, and vice versa. We therefore wondered if the relative activation of Dot6 versus Msn2 before stress was any indication of different cellular states. Acute stress peak heights and pre-stress area under the curve (AUC) are as defined in Figure 6B, and cell points are colored according to their cell cluster from Figure 4 as shown in the key. (A–B) Acute-stress peak height plotted against pre-stress AUC for Msn2 (A) or Dot6 (B). (C–E) Acute-stress peak height for Msn2 (C), acute-stress peak height for Dot6 (D), and the difference in acute-stress peak height (E) plotted against the difference in pre-stress AUC. The results show that Dot6 peak height is best explained by the relative pre-stress activation of Msn2 versus Dot6. These differences are likely capturing distinctions about pre-stress cellular states (see Discussion). P-values and R² of the fit are shown on each plot. All p-values were significant at a Benjamini-Hochberg corrected false discovery rate of 0.05.