Cell Size Modulates SBF and Whi5 Chromatin Binding to Regulate the Start of the Budding Yeast Cell Cycle

Abstract

Cell growth and division are tightly coordinated to cell size. In budding yeast, increasing cell size promotes the G1/S transition, called Start, by activating the transcription factor SBF, which drives a large fraction of cell-cycle–dependent gene expression. Part of this regulation arises because the concentration of the SBF inhibitor Whi5 decreases as cells grow. However, cells lacking Whi5 can still maintain a relatively accurate size when the SBF activator Cln3 is also removed, indicating that there are additional size control mechanisms. To understand how cell size is mechanistically translated into the activity of SBF-regulated promoters, we quantified the binding kinetics of Whi5 and SBF in live cells using single-molecule fluorescence microscopy. We found that increasing cell size is associated with both a decreased chromatin affinity of Whi5 and an increased chromatin affinity of SBF, accompanied by a higher SBF:Whi5 cell copy-number ratio. Chromatin-binding trends under basal and Whi5 overexpression conditions indicate that Whi5 restricts SBF association with chromatin. The transition point at which SBF binding overtakes Whi5 binding coincides with the onset of the expression of the G1 cyclins CLN1 and CLN2, two SBF targets that are important for committing cells to division. Reduced Whi5 binding reflects changes in its chromatin-association rate, as Whi5 and SBF dwell times on chromatin remain ~10 s and are largely independent of cell size. Together, these results show how changes in SBF and Whi5 abundance and chromatin association transmit cell size information to the genome to regulate the size-dependent Start transition in budding yeast.

INTRODUCTION

Cell size homeostasis is a universal feature of biology ^1,2. While cell size homeostasis is commonly observed, the underlying molecular mechanisms have been difficult to isolate and determine. The understanding of molecular mechanisms is most developed in the budding and fission yeast model organisms ^3–6. In the budding yeast, Saccharomyces cerevisiae, size control primarily takes place in the G1 phase of new-born daughter cells before they are committed to cell division ^7–11. Cells that are born smaller delay entry into S phase to have more time to grow larger and compensate for their initial size ¹². In contrast, larger mother cells have homogenously short lengths of G1 phase and show no evidence of cell size control and get larger with each division cycle ^7,8,13.

The molecular network controlling the budding yeast G1/S transition has been linked to cell size regulation. Mostly, this has been attributed to the regulation of activity of the central cell cycle transcription factor SBF, a heterodimer composed of Swi4 and Swi6 ^14,15. In G1, SBF is bound to and repressed by Whi5 and activated by the G1 cyclin Cln3 in complex with the cyclin-dependent kinase Cdk1 (Fig. 2A) ^16–21. In addition to Cln3 and Whi5, the gene BCK2 has also been identified as a driver of the G1/S transition although the specific mechanism through which it works is unknown ^22–24. Once it is sufficiently active, SBF triggers a sharp positive feedback loop that commits cells to division at a point known as the Start of the cell cycle ¹⁰. This feedback loop operates through the SBF-dependent transcriptional activation of the two downstream G1 cyclin genes CLN1 and CLN2, which encode the Cln1 and Cln2 cyclin proteins that also form a complex with Cdk1 ²⁵. Cln1,2-Cdk1 complexes then complete the inactivation of Whi5 to fully activate SBF and irreversibly commit the cell to division ^25,26.

Figure 2. Bound fraction of SBF and Whi5 changes with size in daughter cells.

(A) Diagram representing the regulation of SBF-dependent transcription during G1. (B) Representative images of the categorization done for the selection of daughter cells. Examples of overlays containing brightfield and Whi5-mCitrine images, showing pre-Start daughter cells coupled to mother cells at different stages of G1 or S phases. (C) Results for the single cell analysis of chromatin binding for Whi5-mCitrine-HaloTag (top) and for the distributions of log transformed radius of gyration for tracks obtained from small (20–40 fL) or large (40–60 fL) daughter cells (bottom). The gray line in the single cell analysis represents the trend obtained for the Whi5-WIQ mutant. (D) Results of the single cell analysis of chromatin-binding for Swi4-Halo (top) and for the distribution of radius of gyration values in small and large cells (bottom). (E) Results of the single cell analysis of chromatin-binding for Swi6-Halo mbp1Δ (top) and for the distribution of radius of gyration values in small and large cells (bottom). (F-H) Results of the single cell analysis of chromatin binding for Whi5-mCitrine-Halo, Swi4-Halo or Swi6-Halo mbp1Δ, respectively.

While the positive feedback loop driving commitment to cell division is now well accepted, there are multiple models for how budding yeast trigger this transition. In different growth conditions, Whi5 and Cln3 concentrations likely reflect cell size and cell growth rate respectively. The concentration of Whi5 is nearly inversely proportional to cell size so that its ability to inhibit SBF decreases as cells grow larger ^18,27,28. While Cln3 concentration is nearly constant in G1 during a particular growth condition, its concentration is higher when cells consume glucose and grow quickly compared to when they consume ethanol and grow slowly ²⁹. While unraveling how the G1/S transition is regulated across growth conditions is an important problem, here we restrict ourselves to consider how cell size triggers an increase in the rate of G1/S progression in a given growth condition.

For the case of cell size control in a given growth condition, there remains substantial controversy in the field for how the G1/S transition is triggered in larger cells ^18,30–35. The models can divided into three groups: activator titration models ^30–33,36, inhibitor dilution models ^18,37–41, and combined models containing both mechanisms ^34,35. Activator models propose that the increase amount of activator, like Cln3 or SBF, titrate the fixed DNA amount in G1 phase. The genome is thus used as a ruler to measure the number of activator molecules, like Cln3, which would be in proportion to cell size. The inhibitor dilution model proposes that the dilution of Whi5 serves as a cell size sensor to promote the G1/S transition in larger cells ⁴².

While the size control models above are consistent with the G1 cellular concentration dynamics that have been observed in budding yeast, they make different predictions for how regulatory proteins bind to the genome to control the G1/S transition. In this study, we assess cell size control models by measuring chromatin-binding kinetics of regulatory proteins in larger and smaller cells using single-molecule microscopy. We show that in daughter cells, which experience the most cell size control ^7,8, the chromatin-bound fraction of the cell cycle inhibitor Whi5 decreases as cells grow larger. This is consistent with the inhibitor dilution class of models. In contrast to Whi5, we find that the chromatin-bound fraction of the cell cycle-activating transcription factor SBF increases with cell size as cells grow during G1. Measurement of the copy numbers of these proteins in the nucleus results in an estimated ~20 copies of SBF bound to chromatin in the smallest cells and a plateau of ~100 chromatin-bound copies in larger cells. The transition point, where Whi5 abundance on chromatin is overtaken by that of SBF, correlates with initiation of transcription of the G1 cyclins CLN1 and CLN2, as measured by single-molecule Fluorescence In Situ Hybridization (smFISH). This transcriptional activation of G1 cyclins marks the Start commitment point to the cell cycle ^10,25. Finally, we characterize protein dwell times on chromatin and show that these changes mainly result from differences in the association rate of Whi5 and SBF, as their average dwell times are ~10 seconds across a range of cell sizes. Taken together, our data suggest a model where SBF association to chromatin, and ultimate activation of transcription, is controlled the genome-titration of SBF and the dilution of Whi5 that accompanies cell growth in G1 phase. Thus, our work marks a significant step forward towards a more mechanistic understanding of how the G1/S transition is triggered.

RESULTS

Single-molecule analysis of chromatin binding kinetics

To test molecular models of the budding yeast G1/S transition, we sought to measure the binding of SBF and Whi5 to chromatin in single cells. This would allow us to determine how the amount of specific chromatin bound proteins changes with cell size as cells progress through G1 phase. To do this, we made a fusion of the gene of interest at the endogenous locus with the coding sequence for HaloTag. The resulting chimeric protein was then labelled in vivo with cell permeable JaneliaFluor dyes (JF552 or PA-JF549) ^43,44. We used Highly Inclined and Laminated Optical sheet (HILO) microscopy⁴⁵ to increase the signal to noise ratio in our imaging (Fig. 1A), camera integration of 10 ms to detect both diffusive and chromatin-bound molecules, and intervals of 50 ms to sample binding events lasting at least few hundreds of milliseconds. We chose these imaging parameters based on previous single-molecule imaging of yeast and mammalian transcription factors ^46–50 (Fig. 1B). Characterization of the intensity of spots and their diffusion coefficient corroborated that they are single fluorophores and that their movement is consistent with chromatin binding, resulting in a slowing down of diffusion (Fig. S1A-D). By recording the position of spots over time, we obtained maps of individual molecules (tracks), each associated with a value of their radius of gyration, as a measure of their displacement. The distribution of radius of gyrations in a population of tracks, obtained from multiple cells, were fitted to two peaks in a Gaussian Mixture Model that we interpreted as chromatin-bound and diffusive, respectively (Fig. 1C & S1). Using the information in the distribution in the population, we then calculated the fraction of molecules bound to chromatin in a single cell by categorizing tracks into either bound or unbound, and assigning them to the cells from where they originated. This allows us to correlate bound fraction with the size of the individual cell.

Figure 1. Single cell characterization of chromatin binding.

(A) Proteins of interest were endogenously tagged with HaloTag and labelled with a cell permeable fluorophore coupled with a HaloTag ligand. HILO microscopy was used to increase the signal-to-noise ratio needed for single-molecule tracking. (B) Representative images of single-molecule tracking experiments. Brightfield image (left) and Whi5-mCitrine (middle) image were used to identify cells in the pre-Start G1 and to segment nuclei. Multiple timepoints of a freely diffusive molecule (top) or a chromatin-bound molecule (bottom) are shown along with the resulting ‘track’ that joins all localizations (right). (C) Procedure to characterize single cells. Tracking of molecules is done in a population of cells, resulting in a distribution of dynamic behaviours characterized by their radius of gyration. Two behaviours, chromatin-bound and diffusive, are identified in this distribution, and individual tracks, belonging to specific cells, are assigned to either of these two groups. Consequently, a fraction of chromatin-bound molecules is assigned to each cell. (D) Characterization of diffusive control mCitrine-Halo-NLS. The plot in the left compares bound-fraction against cell size. Each point in the plot represents a single cell. The plot in the right shows the log transformed radius of gyration for the cell population. (E) Single cell analysis (left) and distribution of the cell population (right) for the chromatin-bound control, histone H2B-HaloTag.

As positive and negative controls for our method of estimating chromatin bound fractions, we used a strain expressing the nuclear localized fluorescent protein mCitrine-Halo-NLS and a histone H2B-Halo fusion protein. mCitrine-Halo-NLS and H2B-Halo are expected to be largely freely diffusing or chromatin bound, respectively. As expected, molecules of mCitrine-Halo-NLS were mostly categorized as diffusive, with an average of 7.1% ± 1.1% categorized as chromatin-bound (Fig. 1D). In contrast, molecules of H2B-Halo were mostly chromatin-bound, with an average of 71.3% ± 2.5% chromatin-bound molecules per cell (Fig. 1E), similar to previous estimates of the bound fraction of histones in yeast ⁵¹. As expected, neither of these two proteins showed any relationship between cell size and chromatin binding (mCitrine-Halo-NLS: p value = 0.984; H2B-Halo: p value = 0.689).

Cell size increases the fraction of chromatin-associated SBF and decreases the amount of chromatin-associated Whi5 in daughter cells

To test how cell size regulates the G1/S transition, we sought to measure how the chromatin association of the key regulators Whi5 and SBF changes with increasing cell size. This is important because the activation of the Cln1,2-dependent positive feedback loop is regulated at SBF-bound promoter regions. Therefore, determining the amount of Whi5 and SBF on chromatin should give insight into their amounts at the key CLN1 and CLN2 promoter regions. We are particularly interested in pre-Start cells and how cell size drives cells through the Start transition. The complete dissociation of Whi5 from chromatin occurs after Start, when one can see its complete exit from the nucleus and the complete transcriptional activation of CLN2 ^10,25,52.

To determine how Swi4 and Whi5 chromatin association changes with cell size, we used a strain containing either genomically integrated WHI5-mCitrine-Halo or WHI5-mCitrine and SWI4-Halo alleles. Both of these fusion proteins are functional since we do not observe any changes to the cell size distribution of the population, which would be expected if the fusions disrupt these proteins functions ^17,18,53,54 (Fig. S2A). As Swi6 also is also part of the transcription factor MBF ⁵⁵, we also generated a strain carrying WHI5-mCitrine and SWI6-Halo alleles in an mbp1Δ background. We examined this background because Mbp1 forms the MBF transcription factor together with Swi6 and we only wanted to examine SBF. As expected, this strain displayed a slight cell size increase consistent with previous results ⁵⁶(Fig. S2B). In brightfield images, we distinguish mother-daughter pairs based on morphology: the larger, more ovoid cell is classified as the mother, and the smaller, rounder cell is classified as the daughter. We identified pre-Start cells by the presence of nuclear-localized Whi5-mCitrine (Fig. 2B). Applying the approach described above, we quantified the proportion of chromatin-bound molecules for each cell and its size.

Our analysis revealed that the fraction of Whi5-mCitrine-Halo that is chromatin-bound in daughter cells was lower in larger cells (Fig. 2C; p value <0.033). For example, in cells less than 30fL in volume, 48.9% ± 9.7 % (standard error of the mean) Whi5-mCitrine-Halo is chromatin-bound, while this fraction decreased to 29.1% ± 4.4 % in cells larger than 40fL. We also examined the chromatin-bound fraction of Whi5-WIQ-mCitrine-Halo, which is defective in binding to SBF and therefore to SBF-regulated promoters ⁵⁷. The fraction of less mobile, likely chromatin-bound Whi5-WIQ is 17.4% ± 1.9 % in large and small (Fig. 2C & S3C). These results demonstrate that the fraction of Whi5 bound to chromatin decreases with cell size in daughter cells until reaching near baseline levels at ~40fL. This baseline level was determined by analysis of the SBF non-binding mutant Whi5-WIQ-Halo. It also shows that, once Whi5 unbinds from chromatin, it can remain in the nucleus for some period of time without being immediately exported since if nuclear export immediately followed unbinding, no drop in the bound fraction would be observed and highly mobile nuclear Whi5 would be rarer.

In contrast to Whi5, both subunits of SBF, Swi4-Halo and Swi6-Halo, showed an increase in their chromatin-bound fraction as cell size increases. Small cells less than 30 fL in volume showed a bound fraction of 30.5 % ± 3.0 % of Swi4-Halo, while cells greater than 45 fL in volume showed a chromatin-bound fraction of 57.7% ± 7.4% (Fig. 2D; p value < 0.0006). For Swi6-Halo, the bound fraction increased from 21.2 % ± 3.5 % in cells <35 fL to 33.2 % ± 4.1 % in cells >50 fL (Fig. 2E; p value <0.028). Similar conclusions can be reached by analysing data from a population of cells, instead of single cells, after categorizing them into two groups (small and big cells) based on their volume (Fig. 2C-E & S3A).

It is important to note that these clear trends in chromatin binding of Whi5 and SBF are unique to daughter cells. Analysis of the chromatin-bound fraction per cell size in mother cells show no correlation, and greater scatter across cells sizes (Fig. 2F-H). This is consistent with these cells being at different stages of SBF activation in G1 irrespective of their size (Fig. S4).

The number of chromatin-bound SBF molecules increases with cell size from tens to hundreds

To relate the chromatin-bound fraction changes we observed to absolute numbers of molecules, we measured the copy number of Whi5 and SBF molecules per nucleus in our conditions. To do this, we measured the total intensity of cells carrying mNeonGreen fusions by integrating the intensity obtained from a Z-stack obtained using a spinning disc microscope. We corrected for cell autofluorescence by collecting data with two emission filters (one for green and one for red light) but using a single excitation wavelength ⁵⁸. We then divided the integrated intensity in the cell (or the nucleus) by our estimate of the fluorescence emitted by a single-molecule. This estimate was obtained by measuring the fluorescence intensity emitted by a nuclear pore complex that has a known number of mNeonGreen labeled subunits ⁵⁹ (see methods; Fig. 3A). Our results show that Swi4 copy number scales proportionally with cell size, starting with an average of less than 50 molecules per cell in small <20fL cells and increasing to more than 125 molecules in larger >50fL cells (Fig. 3B). The abundance of Whi5 and Swi6 was higher (~300 copies per cell), but the number of these molecules per cell increased less with cell size (Fig. 3B, Table S1). These estimates are similar to previous copy number estimates ⁶⁰, and consistent with the size-dependent concentration measurements previously reported, namely, that Whi5 is diluted in larger cells ⁴².

Figure 3. Number of chromatin-bound SBF increases with size in daughter cells.

(A) Procedure used for the estimation of protein copy numbers per cell. Individual foci produced by the NPC subunit Nup59-mNeon-Green were used to characterize the intensity produced by a single mNeonGreen molecule This value was used to divide the total cell intensity of mNeonGreen fusions, after subtracting background autofluorescence, to obtain an estimate of the copy number. (B) Distribution of copy numbers for Swi4, Swi6 and Whi5 fusions to mNeonGreen. Each dot represent a single cell. Line represents the average copy number at four different cell sizes. (C) Single-cell analysis showing the number of copies of Swi4 or Whi5 bound to chromatin with respect to the cell size for pre-Start daughter cells. A distribution of cell sizes is shown at the bottom of the plot. (D) Single cell analysis of the number of Swi4-HaloTag bound to chromatin in post-Start daughter cells.

After measuring how protein abundance changes with cell size, we can now convert our chromatin-bound fractions to estimates of the total amount of chromatin-bound proteins. This shows that the number of chromatin-bound SBF (Swi4) molecules was 39.9 ± 3.9 per cell in small daughter cells (cell size < 30 fL), increasing to 99.8 ± 12.4 molecules per cell in large daughter cells (cell size > 45 fL). Similarly, Swi6 binding increases from 80.0 ± 13.3 molecules per cell (cell size < 35 fL) to 130.0 ± 15.9 molecules per cell (cell size > 50 fL). By subtracting the Whi5-WIQ chromatin-bound fraction, the trend is inverted for Whi5, where cells with a volume < 30 fl have 102.1 ± 32.8 chromatin-bound molecules, which drops to nearly zero in larger cells. (Fig. 3C). In comparison, post-Start cells have a flatter trend for chromatin-bound Swi4 starting at about the maximal occupancy observed for pre-Start daughter cells, suggesting a plateau in the binding at just over 100 molecules (Fig. 3D).

Changes in chromatin binding of Whi5 and Swi4 correlate with SBF-dependent transcription

The results above show changes in the chromatin-bound number of SBF and Whi5 occurring at sizes similar to those previously associated with CLN2 transcription and the Start transition ^25,61. To confirm that this is indeed the case here, we sought to measure the expression of the G1 cyclins CLN1 and CLN2 in single cells using single-molecule Fluorescence In-Situ Hybridization (smFISH) to detect mRNA transcripts of these genes in individual cells. To do this, we first generated cells carrying an MS2 24x array inserted at the 3’ end of CLN1 and CLN2, fixed and permeabilize the cells, and used fluorescent probe carrying Quasar 570 fluorophore targeting the MS2 array on mRNA molecules. We then took images in the green and red fluorescent channels to detect Whi5-mNeonGreen and the MS2 array, respectively, and in brightfield to measure the cell size. To test for cells actively expressing CLN1 and CLN2, we counted the number of cells that contained clear spots in the nucleus. Spots outside of the nucleus likely represent earlier transcription events (Fig. 4A). Our results showed few events in small cells, only 30.89% of 30fL cells having nuclear spots for CLN1 (N = 123, SE = 4.2%), and an increase in the frequency of transcription events in cells larger than 35fL, with 78.54% of 40fL cells having nuclear spots for CLN1 (N = 247, SE = 2.6%) (Fig. 4B & S5). These results corroborate our assumptions that the changes in chromatin binding correlate with changes in the expression of SBF-regulated genes in the conditions studied.

Figure 4. SBF-dependent gene activation matches with changes in its chromatin binding.

(A) Examples of smFISH for the SBF-dependent gene CLN1. Cell (yellow) and nucleus outlines (white dashed) are shown. Arrow shows an example of recently transcribed mRNA molecule used for the analysis. (B) Frequency distribution of cells with (pink) or without (blue) nuclear foci representing recently transcribed CLN1 events contrasted against cell size. Inset shows the same data but as fraction of cells with nuclear CLN1 foci in relation with cell size. (C) Single cell analysis showing the fraction of chromatin-bound Swi4-HaloTag against cell size for wild type (orange; data identical to Fig. 3C) and for a strain overexpressing Whi5 (red). (D) Frequency distribution of cells with (pink) or without (blue) nuclear foci of CLN1 against cell size in cells overexpressing Whi5. Inset shows the same data plotted as frequency of cells with nuclear CLN1 foci against cell size.

SBF binding to chromatin is inhibited by Whi5

The observation that SBF association with chromatin increased with cell size posed a conundrum because the concentration of both its subunits, Swi4 and Swi6, did not increase with cell size (Fig. S6B). One possible explanation for the increased association of SBF with chromatin in larger cells is that Whi5 prevents SBF from associating with chromatin. Consequently, increasing chromatin-binding of SBF could reflect the lower concentration of Whi5 in larger cells. To test this hypothesis, we overexpressed Whi5 by adding a second copy of the WHI5 gene under the control of the strong ACT1 promoter ³⁸. In these cells, the bound fraction of SBF (median = 0.19) is significantly reduced compared to wild type cells (median = 0.36, p value= 0.0006). The size-dependent increase is also slightly flattened compared to the wild type cells (p value= 0.09) (Fig. 4C). smFISH data confirmed that in cells overexpressing Whi5, the expression of CLN1 generally occurs at cell sizes larger than 40 fL (compared to 30 fL in WT), consistent with the changes in binding of SBF we observe (Fig. 4D). Taken together, these data indicate that SBF binding is regulated by Whi5.

Dynamic association of SBF to chromatin is hindered by its interaction with Whi5

Having found that SBF is less chromatin-associated in Whi5 over-expressing cells, we next sought to determine why this is. In principle, it could be from reducing the rate free SBF binds chromatin or the time it stays bound there. We therefore sought to characterize the dwell time of SBF and Whi5 using single-molecule microscopy. Previous work using chromatin immunoprecipitation has demonstrated binding of SBF and Whi5 at promoters during G1 ^16,17, but the dwell time associated with this chromatin-binding has not been determined. To measure dwell times on chromatin, we used an approach similar to that we previously applied to characterize the subunit dynamics at the replisome ⁶². Briefly, as in the bound-fraction experiments described above, we used HaloTag derivatives labelled with PA-JF459 dye and HILO microscopy. The change to the protocol is that we now used camera integration times of 500 ms rather than 10 ms. This longer acquisition time resulted in motion blurring of fast-moving molecules but not chromatin bound molecules (Fig. 5A; see methods). We identified chromatin-bound molecules using a machine-learning approach ⁶³, and inferred the time a molecule was associated with the chromatin from the lifetime of the spots representing bound molecules after correcting for photobleaching (Fig. 5B & S7). Histone H2B-Halo was used as our bleaching control, as we expect that its focus disappearance largely represents bleaching rather than unbinding from the chromatin.

Figure 5. SBF binding to chromatin is transient and is inhibited by Whi5.

(A) Representative images of the imaging done to estimate dwell times. A brightfield images (left) is used to determine the cell outline. Whi5-mCitrine (middle) marks pre-Start cells and is also used to segment nuclei. Images of a chromatin-bound Swi4-HaloTag coupled to PA-JF549 are shown at different time points. (B) Example of a distribution of lifetimes for 105 Swi4-HaloTag foci (blue bars). The distribution is fit to a single exponential decay (purple), which after correcting it by the bleaching control (yellow), results in a corrected distribution of bound times (orange). (C) Estimated dwell time values for SBF and Whi5 in daughter cells categorized as small or large. Error bars represent standard error of the mean. (D) Estimated association rate values for SBF and Whi5 in daughter cells categorized as small or large. Error bars represent standard error of the mean.

Results from daughter cells show that Swi4 has an average dwell time of 8.1 seconds (C.I. 5.9–10.6 s), Swi6 has a dwell time of 12.2 seconds (C.I. 9.4–15.5 s), and Whi5 has a dwell time of 9.6 seconds (C.I. 6.9–12.7 s) (Fig. S7C). It is interesting that dwell time estimates are similar for SBF and Whi5 because this suggests most chromatin binding and unbinding of Whi5 occurs when it is associated with SBF. The duration of chromatin association of SBF does not differ significantly between daughters and mother cells (Fig. S6C) or between small and large daughter cells (Fig. 5C). Moreover, the average dwell time of SBF is similar in Whi5 overexpressing cells (Fig. 5C & S7), suggesting that Whi5 does not affect the dissociation of SBF from its binding sites on the genome. In contrast, estimates of the association rates, calculated using the measured bound fraction and dwell times, show a significant difference in small and large cells (Fig. 5D). These results show that regulation of transcription of SBF occurs by transient binding to promoters. Furthermore, since Whi5 affects the average occupancy of SBF on the genome, this result implies that forming a complex with Whi5 reduces the rate SBF binds its target sites on the promoter, and not the time it spends on the promoter after binding.

DISCUSSION

Our live cell study of SBF and Whi5 chromatin association gives mechanistic insight into how cell size triggers progression through the Start transition in budding yeast. As daughter cells grow larger, less Whi5 is associated with the chromatin. This is consistent with the inhibitor dilution model where the decreasing concentration of Whi5 in larger cells promotes cell cycle progression42. In addition to linking Whi5 dilution to changes in chromatin occupancy, we observe that SBF binding to chromatin increases with cell size, despite its approximately constant cellular concentration. This is consistent with SBF titration contributing to the size-dependence on Start. Moreover, it is likely these two mechanisms are intertwined as Whi5 inhibits SBF association with the chromatin (Fig. 6).

Figure 6. Model for SBF regulation in daughter cells.

Small newly born daughter cells have a few-fold excess of Whi5 relative to SBF (left). Whi5 limits SBF binding to chromatin and represses SBF-dependent transcription. As cells grow, Whi5 is diluted and the number of SBF copies increases (right). This shift in relative abundance increases the fraction of SBF not bound to Whi5, promoting SBF binding to target promoters and ultimately activating SBF-dependent transcription.

Our data on the absolute numbers of molecules bound to chromatin has additional implications for the regulation of cell cycle-dependent gene expression. We estimate that the number of chromatin-bound Swi4 molecules starts at about 40 in small G1 cells and rises to about 100 as cells arrive at Start. Given that there are approximately 200 G1/S transition-related promoters, and that some of these promoters contain multiple SCB sites ^64–66, not all SBF-regulated promoters can be bound to SBF even when they are all active. This conundrum is resolved once we consider the rate with which SBF interacts with chromatin. Our data show that both Whi5 and SBF have a similar dissociation rate on chromatin, with average dwell times of ~10 seconds. This is within an order of magnitude to the reported dwell times of other yeast transcription factors such as Ace1p, with a ~3 second average ⁶⁷, and Gal4, with a reported average of 25–75 seconds ⁶⁸. These estimates are also within the range of transcription factor dwell times reported for mammalian cells, which span from a few to hundreds of seconds ⁶⁹. The fast turnover of SBF on chromatin means that a smaller number of transcription factors can regulate a larger number of binding sites on the genome. This is because SBF will be bound to each of its sites a fraction of the time depending on its specific binding constants so that the same molecule can regulate several sites in any given minute.

Our data show that changes in SBF and Whi5 binding to chromatin are largely explained by changes in their association rates rather than their dissociation rates. This suggests that Whi5 primarily reduces the probability of SBF associating with chromatin, rather than actively promoting its dissociation. Together with the matching dwell times of SBF and Whi5, and prior evidence for their direct interaction in solution ^16,17, these results support a model in which SBF diffuses and binds to chromatin as a complex with Whi5. Reduced chromatin binding of SBF, but not its complete abrogation, suggests that either a minimal number of copies of SBF binding is required for promoter activation, or that the presence of Whi5 prevents promoter-activation. Changes in the binding of Whi5 may at least partly reflect changes in its phosphorylation state ⁷⁰, as phosphorylated Whi5 is unable to bind chromatin through SBF. Further work will be needed to determine whether Whi5 dissociation from chromatin precedes its phosphorylation or simply reports on it.

Our quantitative analysis has highlighted the need for additional biochemical studies of the Whi5-SBF interaction. We found that there is more Whi5 bound to chromatin than Swi4 in the early G1 phase. However, it was previously shown that Whi5 ChIP signal depended on its GTB region that likely interacted with Swi6’s C-terminal domain ⁵⁷. Consistent with these results, we also found that the WIQ WHI5 mutation disrupts its chromatin association. This raises the question that if Whi5-SBF were a simple protein-protein interaction, and Whi5’s chromatin association depended on SBF, then how could more Whi5 than SBF molecules be bound to chromatin at any time? One possibility is that SBF has multiple binding sites for Whi5. Another is that Whi5 may bind to another copy of Whi5 that is bound to SBF. Additional research is required to resolve this intriguing mechanism.

Taken together, our findings based on live cell single-molecule microscopy provide a mechanistic framework for understanding how cell size regulates cell cycle progression through the G1/S transition in budding yeast. We demonstrate that SBF association with chromatin increases with cell size, while Whi5 association decreases, illustrating the intertwined dual contributions of activator titration and inhibitor dilution to cell cycle regulation. These results bridge the gap between known concentration dynamics and the biochemistry taking place at the promoters driving commitment to the Start of the cell cycle. Moreover, our quantitative analysis of the dynamic interplay between Whi5 and SBF on chromatin identify new questions related to their biochemical mechanism. Beyond yeast, these insights provide a conceptual framework that may inform our understanding of size control and cell cycle regulation in multicellular organisms, including humans, where similar principles governing cellular growth and division have been shown to apply.

Supplementary Material

Supplementary Information

Document S1. Figures S1-8 and Tables S1-4