Search tips
Search criteria 


Logo of nihpaAbout Author manuscriptsSubmit a manuscriptHHS Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
Mol Cell. Author manuscript; available in PMC 2012 August 19.
Published in final edited form as:
PMCID: PMC3160603

Distinct interactions select and maintain a specific cell fate


The ability to specify and maintain discrete cell fates is essential for development. However, the dynamics underlying selection and stability of distinct cell types remains poorly understood. Here, we provide a quantitative single-cell analysis of commitment dynamics during the mating-mitosis switch in budding yeast. Commitment to division corresponds precisely to activating the G1 cyclin positive feedback loop in competition with the cyclin inhibitor Far1. Cyclin-dependent phosphorylation and inhibition of the mating pathway scaffold Ste5 is required to ensure exclusive expression of the mitotic transcriptional program after cell cycle commitment. Failure to commit exclusively results in coexpression of both cell cycle and pheromone-induced genes, and a morphologically-mixed inviable cell fate. Thus, specification and maintenance of a cellular state are performed by distinct interactions, which is likely a consequence of disparate reaction rates and may be a general feature of the interlinked regulatory networks responsible for selecting cell fates.


The precise specification and maintenance of particular cell fates in response to internal and external signals is crucial for life. While unicellular organisms make vital decisions to enter different life cycle stages in response to environmental change (e.g., to sporulate in poor conditions), multi-cellular organisms accurately pattern diverse cell types during development. In mammals, incorrect cell fate selection can result in developmental abnormalities, while poor maintenance may play a role in oncogenesis.

Despite its importance, we lack a precise molecular understanding of cell fate selection in multi-cellular organisms due to the presence of multiple overlapping pathways and associated spatiotemporal complexity. In particular, commitment points are frequently invoked in the explanation of differentiation processes, yet they remain largely conceptual due the lack of direct quantitative live-cell measurements of key regulatory proteins. This motivates the study of differentiation dynamics in unicellular organisms, which can be grown in more controlled environments and monitored with a variety of quantitative live-cell imaging techniques(Colman-Lerner et al., 2005; Yu et al., 2008). Fine temporal control of the cellular environment can be used to exogenously control signaling pathways (Charvin et al., 2008; Lee et al., 2008; Muzzey et al., 2009; Taylor et al., 2009), while time-lapse fluorescence imaging allows continual monitoring of the concentrations of key regulatory proteins, which can then be correlated with cell fates to determine causal biochemical relationships.

Here, we examine cell cycle commitment in budding yeast vis-à-vis pheromone-induced mating arrest, which exhibits all the features of terminal differentiation, including changes in gene expression, arrest of the cell cycle, and persistent alterations in morphology. Since the purpose of mating is to fuse two haploid cells, it must be restricted to the G1 phase, prior to the initiation of DNA replication. The point where a cell loses mating competence and commits to the cell cycle is called Start (Hartwell et al., 1974). Hence, upon exposure to mating pheromone, pre-Start cells arrest directly while post-Start cells complete one more round of division before arresting (Figure 1A). This physiology is reflected at the molecular level by inhibitory interactions at the interface between the cell cycle and mating pathways (see schematic in Figure 1B). Mutual inhibition ensures that the mating pathway only arrests the cell cycle pre-Start, while the cell cycle pathway only restrains mating post-Start.

A. Start marks the point of commitment to the mitotic cell cycle, where cells lose mating competence. B. Schematic of G1 cell cycle and pheromone-induced MAPK pathway regulation (mutual inhibition indicated with red arrows). C. Whi5-GFP enters the nucleus ...

The mating pathway is a mitogen activated protein kinase (MAPK) cascade that arrests the cell cycle prior to DNA replication primarily by inhibiting G1 cyclins in complex with the cyclin dependent kinase (Chang and Herskowitz, 1990; Jeoung et al., 1998; Peter et al., 1993; Tyers and Futcher, 1993). In haploid cells, pheromone binds a G-protein coupled receptor (Ste2 for α-factor and Ste3 for a-factor) located at the plasma membrane, which activates a heterotrimeric G protein by dissociating Gα from the Gαβγ (Gpa1–Ste4–Ste18) heterotrimer. Once free, the Gβγ subunit promotes Cdc24 activation of Cdc42 (Wiget et al., 2004), which in turn activates Ste20 (Lamson et al., 2002). Then, Ste20 triggers the MAPK cascade by phosphorylating and activating the MAPKKK Ste11 (Drogen et al., 2000). The scaffold protein Ste5 which interacts physically with both the kinases (Ste11, Ste7 and Fus3) and with the Gβγ subunit, is necessary for mating signaling by coupling receptor stimulation to MAPK pathway activity(Garrenton et al., 2009; Hao et al., 2008; Strickfaden et al., 2007; Takahashi and Pryciak, 2008; Whiteway et al., 1995). The downstream MAPK Fus3 activates the transcription factor Ste12 to induce the associated transcriptional program, including the CDK inhibitor Far1 (Chang and Herskowitz, 1990; Errede and Ammerer, 1989). Importantly, Far1 is activated by Fus3 phosphorylation (Chang and Herskowitz, 1992; Elion et al., 1993) to physically interact with and inhibit the G1 cyclins (Gartner et al., 1998), suggesting a stoichiometric mechanism common to CDK inhibitors(Sherr and Roberts, 1999).

Conversely, the G1 cyclins inhibit the mating pathway by promoting the phosphorylation and degradation of both Far1 (Peter and Herskowitz, 1994; Tyers and Futcher, 1993), and the scaffold Ste5, which is also removed from the membrane to disrupt signaling (Garrenton et al., 2009; Strickfaden et al., 2007). G1 progression is initiated by the upstream G1 cyclin Cln3 which forms a complex with the cyclin-dependent kinase Cdc28 (CDK1). Cln3-Cdc28 phosphorylates and partially inactivates Whi5, the inhibitor of the heterodimeric transcription factor SBF (Swi4/Swi6)(Costanzo et al., 2004; de Bruin et al., 2004; Wijnen et al., 2002). Partially active SBF, and the related transcription factor MBF (Mbp1/Swi6), promote the transcription of two further G1 cyclins CLN1 and CLN2 (CLN1/2), which form a positive feedback loop by completing Whi5 inactivation and SBF activation(Cross et al., 1994; Ferrezuelo et al., 2010; Flick et al., 1998; Skotheim et al., 2008; Wijnen et al., 2002). In cell cycle synchronized cultures, an increase in cyclin expression coincides with the phosphorylation and degradation of Far1, suggesting the possibility that feedback-driven increasing G1 cyclin activity plays an important role in determining Start (McKinney et al., 1993).

Despite considerable study of both the cell cycle and MAPK-mating pathways, Start has remained an abstract concept without a precise biochemical definition. We show that cell cycle commitment corresponds to activating the G1 cyclin positive feedback loop, which occurs when approximately 50% of the transcriptional inhibitor Whi5 has been exported from the nucleus. Genetic analysis of Start reveals separate functions for the Far1 and Ste5 inhibitory interactions at the interface between the cell cycle and mating pathways. While mutual inhibition between the G1 cyclins Cln1/2 and the cyclin inhibitor Far1 sets the commitment point, cyclin-dependent inhibition of the mating pathway scaffold Ste5 is required post-Start to ensure the exclusive expression of the mitotic transcriptional program. An ordinary differential equation model and in vivo kinetic measurements suggest that the observed separation of function is a consequence of the separate time scales associated with Far1 and Ste5 inhibition. Thus, selection and maintenance of a specific cellular state are performed by distinct interactions.


Quantitative Start assay

To determine the molecular basis of cell cycle commitment, we developed a single-cell microfluidics assay. Cells were grown asynchronously before exposure to a step-increase in pheromone (α-factor). This allowed us to classify cells based on the original operational definition of Start as the point of commitment to the cell cycle(Hartwell et al., 1974): upon exposure to α-factor, pre-Start cells arrest directly, while post-Start cells complete an additional mitotic cell cycle.

As a molecular reporter for cell cycle progression, we chose to monitor the nuclear concentration of the transcriptional inhibitor Whi5 through a C-terminal fusion protein expressed from the endogenous locus (Figure 1C). Our C-terminal WHI5-GFP fusion protein is very likely to be fully functional and expressed at WT levels because it does not significantly extend G1 duration or modify cell size(Costanzo et al., 2004; de Bruin et al., 2004; Skotheim et al., 2008). Our choice of reporter was motivated by the fact that the activity of Whi5 is regulated by its subcellular localization. When Whi5 is phosphorylated and inactivated by G1 cyclin-CDK complexes, it is excluded from the nucleus(Costanzo et al., 2004; de Bruin et al., 2004; Kosugi et al., 2009; Taberner et al., 2009). The rapid nuclear entry of Whi5 at anaphase is due to dephosphorylation by Cdc14(Taberner et al., 2009).

Throughout our assay, we use our phase and fluorescence images to measure the nuclear concentration of Whi5 (see methods). In the illustrative experiment shown in Figure 1D–F, cells are first grown for 4 hours in media without pheromone. During this period, cells cycle continuously as seen by periodic Whi5-GFP nuclear entry and exit (the red and blue traces in Figure 1F correspond to the two segmented cells of the same color shown in Figure 1D). Next, mating pheromone is added, which eventually arrests all cells with a high concentration of nuclear Whi5-GFP. Both the red and blue cells are post-Start at the time of pheromone addition and complete an additional mitotic cycle before arresting. After two hours, the pheromone in the media is removed and cells rapidly exclude Whi5-GFP from the nucleus and re-enter the cell cycle. This experiment demonstrates our ability to rapidly add or remove pheromone from the extracellular environment.

Nuclear Whi5 predicts cell fate

To identify variables that were predictive of cell fate, we recorded the amount of nuclear Whi5-GFP, cell area, cell type, and time spent in G1 prior to pheromone addition. Whi5-GFP concentration was measured relative to the maximum value after nuclear entry at anaphase (Figure 2A,B). Since pheromone pathway kinetics (Yu et al., 2008) and media switching (<1min) are faster than Whi5 kinetics (~5 min) and G1 duration (~30 min), we treat pheromone pathway activity as a binary variable that instantaneously changes upon pheromone addition. After pheromone addition, we tracked cells for an additional 2 hours and recorded the subsequent cell fate (arrest directly or divide once more).

Nuclear Whi5-GFP accurately predicts cell fate. A.B. Time courses of nuclear Whi5-GFP for a pre-Start (A) and a post-Start (B) cell at the time of pheromone addition (dashed line). For each cell we record the fraction of Whi5-GFP exported from the nucleus ...

To estimate the probability of each cell fate as a function of the observed variables we used logistic regression (Hosmer and Lemeshow, 2000). We noted a very sharp correlation between nuclear Whi5-GFP and cell fate, which was weaker for the other parameters (Figure 2C–F). Whi5-GFP predicted the cell fates of 97±1% of G1 cells correctly, compared with 65±2%, 65±2% and 66±2% for cell type, cell size and time spent in G1 respectively (Figure 2G). Note that a random guess predicts 50% correctly and the associated errors are estimated using cross-validation.

To validate the accuracy of the cell fates predicted by the logistic regression and to determine whether any other parameter might improve the nuclear Whi5-GFP fit we used L1-regularized multivariate logistic regression (Park and Hastie, 2007). This technique selects the relevant parameters (e.g., nuclear Whi5-GFP, cell size, etc) by penalizing the L1-norm of their coefficients. Thus, a variable is not included in the model unless it is able to provide significant additional predictive value independent of the other variables. Nuclear Whi5-concentration accurately predicted cell fate. Indeed, no additional measured variable yielded statistically significant additional information (Figure 2G and Figure S1A–D; Table S1). This indicates that Start is a well-defined point that is accurately predicted by the nuclear Whi5 concentration independent of cell size, type, and G1 duration.

That nuclear Whi5 accurately predicts cell fate is reflected in the probability of cell cycle arrest decreasing from 80% to 20% as the nuclear Whi5-fraction exported prior to pheromone addition increases from 0.47 to 0.57 (inset Figure 2C). The transition from uncommitted to committed as a function of nuclear Whi5-GFP is very sharp (Hill coefficient >14). This remarkably strong correlation between cell fate and Whi5 levels provides a rational quantitative definition of cell cycle commitment: Start is the point where half the WT cells arrest, which occurs when approximately half the Whi5-GFP (52±3%) has been exported.

Sic1 is a stoichiometric inhibitor of the B-type cyclins responsible for initiating replication (Schwob et al., 1994) and has been suggested to also play a role in cell cycle commitment (Pathak et al., 2007). To investigate the potential role of Sic1 at Start, we created a strain expressing Whi5-mCherry and Sic1-GFP from their endogenous loci. Sic1 degradation follows Whi5 nuclear export suggesting that commitment precedes Sic1 degradation(Costanzo et al., 2004) (Figure S1E,F). Thus, Sic1 degradation in WT cells should be viewed as a dependent on, rather than causal of, cell cycle commitment.

Whi5 export correlates with Cln1/2 feedback induction

The correlation between cell fates and nuclear level of Whi5 prompted us to investigate its functional role in cell cycle commitment. Whi5 is bound to SBF on the CLN2 promoter, which likely inhibits CLN2 expression in early G1(Cross et al., 1994; de Bruin et al., 2004). We examined GFP expression from a CLN2 promoter integrated into the genome of a cell already expressing WHI5-GFP and the nuclear marker HTB2-mCherry(Mateus and Avery, 2000). After a simple subtraction, the Whi5-GFP signal does not interfere with the CLN2pr-GFP signal, and vice versa (Skotheim et al., 2008). In WT cells, the CLN2pr is measured to be induced when 76±2% Whi5-GFP has been exported (Figure S2). The rapid feedback-driven Whi5-GFP nuclear exit kinetics likely results in an overestimation of the amount of Whi5 exported by the time of CLN2pr induction because our measurement of the transcription induction depends on the maturation of a fluorescent reporter. We therefore decided to examine the correlation in cln1Δ cln2Δ cells in which the Whi5-GFP kinetics is substantially slower (Figure 3A). We recorded the amount of Whi5 exported at the time of CLN2pr induction. On average, 49±1% (± s.e.m.) Whi5-GFP has been exported by the time of CLN2pr induction (Figure 3B), which is very similar to the commitment point that occurs at 52±3%. Taken together, our analysis confirms the link between the Whi5 nuclear concentration at Start and CLN2pr induction and strongly suggests that commitment to cell division, i.e., Start, closely follows activation of the Cln1/2 positive feedback loop.

Induction of the CLN2 promoter correlates with exporting half the Whi5-GFP from the nucleus. A. CLN2pr expression (top) and the nuclear Whi5-GFP concentration (bottom) were monitored in single cells. Each color corresponds to a specific cell. For each ...

Start is determined by Far1-Cln1/2 mutual inhibition

To determine the key molecular interactions underlying Start, we performed a genetic analysis based on our quantitative commitment point assay. Since nuclear Whi5-GFP concentration is sufficient to predict commitment in ~97% of WT G1 cells (Figure 2), we restrict our analysis to estimating the probability of commitment as a function of only nuclear Whi5-GFP.

The notion that Start corresponds to the activation of the Cln1/2 positive feedback loop prompted us to examine cln1Δcln2Δ cells, which slowly progress through G1 using CLN3 as their only G1 cyclin. cln1Δcln2Δ cells had to export more Whi5-GFP (63±2%) before reaching the commitment point (Figure 4A) consistent with previous results (Oehlen and Cross, 1994). However, commitment in cln3Δ cells was not significantly different than in WT cells (Figure 4B). This suggests that auto-activation of the positive feedback loop occurs at a similar nuclear Whi5 concentration indicating that upstream signals acting through Cln3 may affect the timing, but not the constitution, of Start.

Genetic analysis of Start reveals the primacy of the Cln1/2-Far1 interaction in determining commitment to cell division. A–F. For each genotype, we performed the Whi5-based quantitative Start assay shown in Figure 2C for WT cells. Cells were grown ...

We next turned our attention to the set of inhibitory interactions at the interface between pheromone and cell cycle signaling. To investigate the importance of G1 cyclin phosphorylation and inhibition of specific pheromone pathway components, we examined mutants containing FAR1 and STE5 alleles lacking CDK phosphorylation sites (FAR1-S87A, STE5-8A; Figure 4C,D) (Gartner et al., 1998; Strickfaden et al., 2007). Relative to WT, FAR1-S87A commitment required more Whi5 export indicating that Far1 phosphorylation and degradation contributes to Start. However, commitment was not altered in STE5-8A cells. Moreover, commitment in STE5-8A FAR1-S87A double mutants is indistinguishable from commitment in FAR1-S87A single mutants (Figure 4E). Increasing the amount of Far1 delayed commitment (Figure 4F). The results from our genetic analysis clearly indicate that commitment in WT cells is determined by the Cln1/2-Far1 interaction and not by the Cln1/2-Ste5 interaction (figure 4G). Taken together, our results suggest the primacy of Cln1/2-Far1 mutual inhibition in defining cell cycle commitment.

Ste5 phosphorylation ensures separate cell fates

That the STE5-8A mutation did not change the commitment point was puzzling because Ste5 contains eight CDK-phosphorylation sites that regulate pheromone signaling in a dosage dependent manner (Strickfaden et al., 2007). This suggested a role for Ste5 as a CDK-activity sensor since at increasing kinase levels one would expect a greater proportion of the Ste5 sites to be phosphorylated and a concomitant decrease in Ste5 activity. Although commitment, as measured by bud emergence, was identical to WT cells, STE5-8A cells had other defects. Time-lapse analysis showed that 14% of STE5-8A cells budded and then shmooed without first completing cytokinesis (Figure 5A,B). These results were consistent with previous observations that a similar fraction of asynchronous STE5-8A cells arrested with 2N DNA upon α-factor exposure(Strickfaden et al., 2007).

A,B. Composite phase and fluorescence images of WT (A) and STE5-8A (B) cells exposed to α-factor in late G1. The wild-type cell buds (pink arrow) and shmoos normally (white arrow) whereas a fraction of the STE5-8A cells first bud (pink arrow) ...

We reasoned that the mixed cell fate might be due to post-Start STE5-8A cells being exposed to α-factor when the G1 cyclins are responsible for inhibiting the mating pathway(Strickfaden et al., 2007). For STE5-8A cells past the commitment point, G1 cyclins would be able to continuously inhibit Far1 and thereby maintain their own expression to ensure cell cycle progression. However, without Ste5 phosphorylation there would be no mechanism through which G1 cyclins could down-regulate STE12-dependent expression, which would result in coexpression of the cell cycle and mating pathway programs. To test this hypothesis, we measured the probability of an aberrant cell cycle as a function of the time between commitment and pheromone addition. Indeed, cells exposed to α-factor just past Start had a high likelihood (~75%) of a defective cell cycle (Figure 5C,D). Further beyond Start, the aberrant arrest fraction likely decreases due to the activation of the mitotic cyclins that inhibit mating via a different mechanism (Ydenberg and Rose, 2009).

To test our prediction that aberrant cell fate selection correlates with mixed gene expression, we constructed STE5-8A and wild-type strains containing mating pathway (FUS1pr-GFP) and cell cycle (CLN2pr-mCherry) gene expression reporters (Figure 5E,F). We found that the aberrant STE5-8A cells showed significant co-expression of mating and cell cycle genes relative to both normally arrested STE5-8A and wild-type cells (Figure 5G). Furthermore, STE5-8A cells exhibiting a significant mitotic delay (>2h) were found to have an intermediate level of co-expression. Thus, specific phosphorylation of Ste5 by G1 cyclins is required for cell fate exclusivity.

Our observations give some insight into previously observed aberrant cell cycle arrests. Since far1Δ STE5-8A cells lack both the ability to arrest the cell cycle and to stop mating pathway activity when exposed to α-factor, each cell is expected to suffer a mixed cell fate every cell cycle it remains in α-factor. Indeed, ~80% of all far1Δ STE5-8A cells arrest aberrantly when exposed to α-factor (Figure 5D). This is consistent with previous results showing far1Δ STE5-8A sensitivity to α-factor and increased 2C arrest (Strickfaden et al., 2007). Furthermore, constitutive expression of Clb5 in a cln1Δ cln2Δ cln3Δ cell exposed to mating pheromone results in a similar coexpression of mating and cell cycle genes and 2C arrest(Oehlen et al., 1998).

Differential rate-constants may explain separation of function at Start

Our observation that Cln1/2-CDK phosphorylation of Far1, but not Ste5 impacts Start is consistent with cyclins phosphorylating and inhibiting Far1 before Ste5. In this model, the Cln1/2-Far1 inhibition is rapidly resolved to either a high or low Far1 state corresponding to mating arrest or the mitotic cell cycle respectively. Thus, if the cell cycle is selected, Far1 is rapidly phosphorylated and degraded while Ste5 phosphorylation occurs more slowly.

To address whether differences in inhibition rate could explain the observed separation of function we constructed a mathematical model based on ordinary differential equations (Figure 6A; Figure S3A,B). The goal of our modeling effort is not to fit data, but rather to gain a qualitative understanding for how differential rate constants may lead to a separation of function at the interface of competing signaling pathways. In our simplified model, cell cycle pathway activity is represented by the Cln1/2 level, while mating pathway activity is represented by the level of Ste5. Far1 is activated by Ste5 and inhibits Cln1/2, whereas Cln1/2 inhibits both Ste5 and Far1. To model the dynamics, Cln1/2 synthesis (k7), representing Cln3 activity is initiated at time=0. The mating pathway is then activated at a rate (k8) at time τ. Intuitively, if τ is large enough, Cln1/2 accumulates sufficiently to activate positive feedback which commits the cell to division (High Cln1/2, low Ste5 and Far1; Figure S3C). However, if τ is small the mating pathway is activated before Cln1/2 feedback and then Far1 inhibits Cln1/2 and the cell arrests (low Cln1/2, high Ste5 and Far1; Figure S3D). For each set of parameters there is a critical time, τcrit, above which the cell is committed to the cell cycle and below which the mating pathway is engaged. τcrit corresponds to a critical amount of Cln1/2 which can be accumulated before commitment.

A separation of kinetic time scales may result in a separation of function. A. Schematic of the simplified model (see text and supplement for details). We analyze the effects of STE5-8A and FAR1-S87A mutations by setting k1 or k3 to zero respectively. ...

To examine the idea that differences in time scales can lead to separation of function in the network we decided to fix all the rates except the Cln1/2 inhibition of Ste5 (k1) and Far1 (k3) whose ratio we varied from 500:1 to 1:500. For each ratio we determined the critical level of Cln1/2 beyond which the cell is committed to division: Cln1/2critWT(k1,k3). Next we examined the effects of the STE5-8A and FAR1-S87A mutations in our model by setting k1 or k3 to zero respectively and then repeating our calculation. The critical Cln1/2 levels for the STE5-8A and FAR1-S87A mutants are: Cln1/2critSTE5-8A(0,k3)and Cln1/2critFAR1-S87A(k1,0) respectively. We calculated how commitment was delayed, as represented by a decreased ability to activate the Cln1/2 feedback in each mutant relative to WT. The relative increase in the critical Cln1/2-level was denoted as IFAR1-S87A and ISTE5-8A so that

IFAR1S87A=Cln1/2critFAR1-S87A(k1,0)Cln1/2critWT(k1,k3) and ISTE5-8A=Cln1/2critSTE5-8A(0,k3)Cln1/2critWT(k1,k3).

Next, we compared the impact of STE5-8A mutation on increasing the commitment threshold relative to the impact of the FAR1-S87A mutation:

  • Impact of STE5-8A relative to FAR1-S87A=ISTE5-8A1(ISTE5-8A1)+(IFAR1-S87A1),

which is plotted in Figure 6B. When comparing the relative impact of the STE5-8A and FAR1-S87A mutations while varying the inhibition rates we found that the relative impact from the mutations is determined by the ratio between the Cln1/2-Far1 and Cln1/2-Ste5 inhibition rates (figure 6B, table S3). In a counterfactual analysis, we find that Ste5 inhibition would determine the commitment point if it were significantly faster than Far1 inhibition (k1[dbl greater-than sign]k3). Thus, commitment is determined by the fastest rate suggesting a separation of time scales of Far1 relative to Ste5 inhibition may underlie the observed separation of function.

Exogenously controlled CLN2 expression inhibits Far1 before Ste5

Our model predicts that separation of function at the cell cycle-MAPK interface arises due to more rapid inhibition of Far1 than Ste5 by G1 cyclins. To test this hypothesis, we examined a yeast strain expressing a FAR1-Venus fusion from the endogenous locus in addition to an integrated STE5-YFP allele. This strain also contains an integrated CLN2 allele driven by the MET3 promoter, which expresses CLN2 at physiological levels in media lacking methionine (Charvin et al., 2008). Upon pheromone arrest, Far1-Venus accumulates primarily in the nucleus, while Ste5-YFP accumulates at the shmoo tip. Control experiments on two strains containing only one of the two yellow markers indicates negligible contributions of Far1 and Ste5 to the shmoo tip and nuclear signals respectively. Pheromone-arrested cells, showing significant nuclear and shmoo yellow fluorescence, were exposed to a new media containing pheromone, but lacking methionine. This induced the exogenously controlled expression of MET3pr-CLN2 which initiated the degradation of Far1 prior to the removal of Ste5 from the membrane (see methods; Figure 6C–D, S3E,F). This result supports our argument that separation of function at network interfaces may arise from a separation of time scales.

We note that the effect of FAR1 and STE5 mutations was not a priori obvious because the number of links in a pathway does not necessarily predict the speed of signal propagation. Kinetic rate-constants may significantly differ, perhaps due to spatial considerations, and should ideally be determined in vivo. Here, Far1 inactivation via Ste5 inactivation must come after we observe Ste5 dissociation from the membrane. Our observations and analysis suggest that Ste5-mediated Far1 inhibition is slower than direct inhibition of Far1 by G1 cyclins, which likely underlies the separation of function shown in Figures 4 and and5.5. Thus, the separation of function of inhibitory interactions may be a general feature emerging from the interfaces between competing signaling pathways. Our data are consistent with theoretical analysis suggesting that linking fast (Cln1/2-Far1) and slow (Cln1/2-Ste5) feedback loops can produce rapidly inducible yet noise resistant switches (Brandman et al., 2005).


Here, we showed how the interaction between the cell cycle and the mating pathway determines Start. The point of commitment to the mitotic cell cycle corresponds to activating the G1 cyclin positive feedback loop controlled by the transcription factors SBF and MBF, which have >200 additional targets. In a related paper appearing in this issue, we show that increased expression of the G1 cyclins CLN1 and CLN2 precedes activation of the bulk of the >200 coregulated genes implying that genomewide changes in transcription are downstream of cell cycle commitment(Eser et al., 2011).

Our work emphasizes the importance of the G1 cyclin dependent inhibition of the MAPK scaffold Ste5 (Strickfaden et al., 2007). Once Cln1/2 has inhibited Far1 to pass Start, there must be an additional negative regulation due to Far1’s terminal location in the pheromone pathway. Without the ability of Cln1/2 to inactivate Ste5, the post-Start cell cycle cannot short-circuit pheromone signaling and downstream gene expression with potentially fatal consequences.

We show that commitment at Start corresponds precisely to the induction of Cln1/2 feedback and is controlled by mutual inhibition of Cln1/2 and Far1, thereby providing a precise biochemical definition for Start (Figure 7). That the commitment point is determined by the relative strengths of a direct mutual inhibition suggests a single axis through which evolution can rapidly tune the fundamental mating-mitotic switch. The complete separation of function of CDK phosphorylation of Ste5 and Far1 allows tuning the commitment point without risking cell cycle and mating pathway coexpression and reduced fitness.

Cell cycle commitment at Start is determined by CLN1/2 positive feedback in competition with Far1. Ste5 inhibition is required to truncate the pheromone signal post-commitment to ensure the exclusive expression of the mitotic transcriptional program.

Cause and consequences of a predictive measurement

The accuracy with which we determine Start stands in sharp contrast to the molecular noise often observed in single-cells and associated with cellular decisions (Bal·zsi et al., 2011). Indeed, that we are able to predict the cell fate of 97% of G1 cells exposed to mating pheromone strongly suggests that nuclear Whi5-GFP is a direct measurement of a well-defined continuous cell cycle phase in G1. Had we chosen different reporters, that less directly measure the cell cycle phase, it is unlikely that we could achieve such accuracy and we may have concluded that the process is fundamentally ‘noisy’. Instead, we conclude that the precise nature of cell cycle commitment likely reflects its fundamental importance in the yeast life cycle.

That nuclear Whi5 corresponds to cell cycle phase arises from molecular considerations. Whi5 export during G1 depends on the karyopherin Msn5 and is due to G1 cyclin phosphorylation disrupting nuclear localization sequences, while Whi5 import is constitutive throughout the cell cycle and depends on the classical nuclear import pathway(Kosugi et al., 2009; Taberner et al., 2009). Since acute removal of CDK activity leads to substantial nuclear re-entry within 10 minutes (Charvin et al., 2010), the nuclear Whi5-GFP concentration likely reflects a balance of import and export rates that is shifted by increasing CDK activity through G1. Thus, the monotonically decreasing Whi5 nuclear concentration is likely to be a direct measurement of G1 cyclin-CDK activity, whose dynamic range is precisely tuned to monitor progression through Start.

From a dynamical systems point of view, Start is a bistable switch (Charvin et al., 2010; Novak et al., 2007; Skotheim et al., 2008). At a critical level of CDK activity the CLN1/2 positive feedback loop is irreversibly engaged as the dynamical system passes through a saddle-node bifurcation. The result that we can accurately predict cell fate, i.e., the location of the bifurcation point, is remarkable considering the numerous factors (some essential) that have been identified through genetic analysis to play a role in G1 kinetics (Jorgensen et al., 2002). Our present results do not exclude a role at Start for changes in the concentration of these factors because they may exist at different levels in different environmental conditions leading to a quantitatively different, but qualitatively similar, Start. However, for a given growth condition, the yeast cell cycle appears robust to concentration variations of other factors and the relevant dynamics collapse to a single axis, CDK activity, with a well-defined threshold.

General aspects of commitment

The existence of an alternate cell fate is essential for the definition of a commitment point, which emerges naturally from the interactions at the interface between any two pathways inducing competing cell fates. The fact that multiple signaling pathways induced by stress (Barbet et al., 1996; Gray et al., 2004; Trotter et al., 2001) can arrest cells in a low-CDK activity G1 state suggests the potential for a multiplicity of G1 commitment points. In these other arrests, commitment may be determined through alternate interfaces of the core G1/S cell cycle regulatory network with stress activated kinase pathways. Thus, cell cycle commitment defined vis-à-vis stress or nutrient deprivation rather than mating arrest may well occur at a different CDK-activity level. However, if the commitment point is defined by G1 cyclin inhibition, it is likely to coincide with the rapid increase in G1 cyclin synthesis occurring when half of the nuclear Whi5 has been removed. Thus, the possibility remains of a universal point of commitment.

Here, we defined cell cycle commitment in budding yeast to reveal a surprising degree of modularity in a regulatory interface essential for all living cells. Indeed, cell fate selection, as a result of interactions at the interface between the core cell cycle and extrinsic signaling pathways, is fundamental for successful proliferation and development. The loss of restriction point control regulating cell division in mammals is associated with cancer and developmental defects(Chen et al., 2009; Zetterberg et al., 1995). We expect our novel methods and conceptual framework to be extended to the study of human cells, thereby giving insight into both development and disease.

Experimental Procedures

Microscopy, flowcell and image analysis

Experiments were performed with a Cellasic microfluidic device ( using Y16 and Y4 plates with a flow rate of 5psi, which exchanged the media in the chamber in about 60s. Images were taken for up to 16 positions every three minutes with a Zeiss Observer Z1 microscope with an automated stage using a plan-apo 63X/1.4NA oil immersion objective. Automatic focusing was performed using Definite Focus hardware. WHI5-GFP and FUS1pr-GFP strains were exposed for 100ms and 25ms using the Colibri LED 470 module at 25% power. HTB2-mCherry and CLN2pr-mCherry strains were exposed for 10ms and 300ms using the Colibri 540-80 LED module at 25% power. All images were subsequently analyzed using custom MATLAB software that segments, tracks and calculates the mean nuclear fluorescence relative to the cytoplasm(Skotheim et al., 2008). See supplemental information in (Skotheim et al., 2008) for a detailed justification of the use of WHI5-GFP and CLN2pr-GFP dual labeled cells.

Yeast strains and growth conditions

Strains are congenic with W303 (see TablesS1–2). Prior to an experiment, cells were grown in log-phase (OD ~0.1) in synthetic complete medium with 2% glucose (SCD) and then sonicated (~5s 3W). All media used were mixed with 20µg/ml casein (Sigma) to inhibit α-factor surface adhesion (Yu et al., 2008). cln1Δ cln2Δ were treated slightly differently and were grown in log-phase in SCD lacking methionine (SCD - met) to express Cln2 from an integrated MET3 promoter. cln1Δ cln2Δ cells were switched to SCD after one hour in the flowcell and then treated as the other strains.

Measurement of Start

Cells were grown in the flowcell for at least 1.5 hours before being exposure to 240nM α-factor and tracked for at least 2 additional hours. For each G1 cell, we compare the amount of Whi5-GFP exported to the largest value within the last 30min (δ/γ). The Whi5-GFP baseline is calculated automatically for each cell. Cells with segmentation errors or nuclei outside of the plane of focus were discarded.

Mixed cell fate experiment

Cells were grown in the flow cell for 2 hours before being exposed to 240nM α-factor for 270min. Cells that first bud then shmoo without an intervening cytokinesis were scored as aberrant. Cells exhibiting prolonged mitosis until the movie limit, likely reflecting a bimodal population of cells either ultimately managing to go through cytokinesis or failing were scored separately as ‘mitotic delay’. FUS1-CLN2 coexpression was measured using the product of the baseline subtracted and peak normalized CLN2pr-mCherry and FUS1pr-GFP signals.

Far1 and Ste5 inhibition kinetics experiment

Cells were initially grown in SCD for 90min and then exposed to α-factor for 120min. Next we induced exogenous CLN2 by switching to media containing pheromone but lacking methionine. Nuclear Far1-Venus was quantified using the same algorithm as Whi5-GFP whereas we used custom matlab software to detect the brightest spot on the shmoo (manually selected) for the Ste5-YFP signal (Figure S3E,F).

Supplementary Material



In particular, we thank Fred Cross and Peter Pryciak for extensive discussions, insightful suggestions, and generous sharing of reagents. We also thank Richard Yu and Gustavo Pesce for generous sharing of reagents. We thank Chris Aakre, Amanda Amodeo, Nick Buchler, Stefano DiTalia, Carlos Gomez Uribe, Danny Lew, and Eric Siggia for comments on the manuscript. We thank Gilles Charvin for advice on microfluidics. Research was funded by the Burroughs Wellcome Fund, the NIH (GM092925), the NSF (CAREER award #1054025) and the Hellman Foundation.


Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.


  • Bal·zsi G, van Oudenaarden A, Collins JJ. Cellular Decision Making and Biological Noise: From Microbes to Mammals. Cell. 2011;144:910–925. [PMC free article] [PubMed]
  • Barbet NC, Schneider U, Helliwell SB, Stansfield I, Tuite MF, Hall MN. TOR controls translation initiation and early G1 progression in yeast. Mol Biol Cell. 1996;7:25–42. [PMC free article] [PubMed]
  • Brandman O, Ferrell JE, Li R, Meyer T. Interlinked fast and slow positive feedback loops drive reliable cell decisions. Science. 2005;310:496. [PMC free article] [PubMed]
  • Chang F, Herskowitz I. Identification of a gene necessary for cell cycle arrest by a negative growth factor of yeast: FAR1 is an inhibitor of a G1 cyclin, CLN2. Cell. 1990;63:999–1011. [PubMed]
  • Chang F, Herskowitz I. Phosphorylation of FAR1 in response to alpha-factor: a possible requirement for cell-cycle arrest. Mol Biol Cell. 1992;3:445–450. [PMC free article] [PubMed]
  • Charvin G, Cross FR, Siggia ED. A microfluidic device for temporally controlled gene expression and long-term fluorescent imaging in unperturbed dividing yeast cells. PLoS ONE. 2008;3:e1468. [PMC free article] [PubMed]
  • Charvin G, Oikonomou C, Siggia ED, Cross FR. Origin of irreversibility of cell cycle start in budding yeast. PLoS Biol. 2010;8:e1000284. [PMC free article] [PubMed]
  • Chen HZ, Tsai SY, Leone G. Emerging roles of E2Fs in cancer: an exit from cell cycle control. Nat Rev Cancer. 2009;9:785–797. [PMC free article] [PubMed]
  • Colman-Lerner A, Gordon A, Serra E, Chin T, Resnekov O, Endy D, Pesce CG, Brent R. Regulated cell-to-cell variation in a cell-fate decision system. Nature. 2005;437:699–706. [PubMed]
  • Costanzo M, Nishikawa JL, Tang X, Millman JS, Schub O, Breitkreuz K, Dewar D, Rupes I, Andrews B, Tyers M. CDK activity antagonizes Whi5, an inhibitor of G1/S transcription in yeast. Cell. 2004;117:899–913. [PubMed]
  • Cross FR, Hoek M, McKinney JD, Tinkelenberg AH. Role of Swi4 in cell cycle regulation of CLN2 expression. Mol Cell Biol. 1994;14:4779–4787. [PMC free article] [PubMed]
  • de Bruin RA, McDonald WH, Kalashnikova TI, Yates J, 3rd, Wittenberg C. Cln3 activates G1-specific transcription via phosphorylation of the SBF bound repressor Whi5. Cell. 2004;117:887–898. [PubMed]
  • Drogen F, O'Rourke SM, Stucke VM, Jaquenoud M, Neiman AM, Peter M. Phosphorylation of the MEKK Ste11p by the PAK-like kinase Ste20p is required for MAP kinase signaling in vivo. Curr Biol. 2000;10:630–639. [PubMed]
  • Elion EA, Satterberg B, Kranz JE. FUS3 phosphorylates multiple components of the mating signal transduction cascade: evidence for STE12 and FAR1. Mol Biol Cell. 1993;4:495–510. [PMC free article] [PubMed]
  • Errede B, Ammerer G. STE12, a protein involved in cell-type-specific transcription and signal transduction in yeast, is part of protein-DNA complexes. Genes Dev. 1989;3:1349–1361. [PubMed]
  • Eser U, Fettig M, Johnson A, Skotheim JM. Commitment to a cellular transition precedes genome-wide transcriptional change. to appear. Molecular Cell. 2011 [PMC free article] [PubMed]
  • Ferrezuelo F, Colomina N, Futcher B, Aldea M. The transcriptional network activated by Cln3 cyclin at the G1-to-S transition of the yeast cell cycle. Genome Biol. 2010;11:R67. [PMC free article] [PubMed]
  • Flick K, Chapman-Shimshoni D, Stuart D, Guaderrama M, Wittenberg C. Regulation of cell size by glucose is exerted via repression of the CLN1 promoter. Mol Cell Biol. 1998;18:2492–2501. [PMC free article] [PubMed]
  • Garrenton LS, Braunwarth A, Irniger S, Hurt E, Kunzler M, Thorner J. Nucleus-specific and cell cycle-regulated degradation of mitogen-activated protein kinase scaffold protein Ste5 contributes to the control of signaling competence. Mol Cell Biol. 2009;29:582–601. [PMC free article] [PubMed]
  • Gartner A, Jovanovic A, Jeoung DI, Bourlat S, Cross FR, Ammerer G. Pheromone-dependent G1 cell cycle arrest requires Far1 phosphorylation, but may not involve inhibition of Cdc28-Cln2 kinase, in vivo. Mol Cell Biol. 1998;18:3681–3691. [PMC free article] [PubMed]
  • Gray JV, Petsko GA, Johnston GC, Ringe D, Singer RA, Werner-Washburne M. "Sleeping beauty": quiescence in Saccharomyces cerevisiae. Microbiol Mol Biol Rev. 2004;68:187–206. [PMC free article] [PubMed]
  • Hao N, Nayak S, Behar M, Shanks RH, Nagiec MJ, Errede B, Hasty J, Elston TC, Dohlman HG. Regulation of cell signaling dynamics by the protein kinase-scaffold Ste5. Mol Cell. 2008;30:649–656. [PMC free article] [PubMed]
  • Hartwell LH, Culotti J, Pringle JR, Reid BJ. Genetic control of the cell division cycle in yeast. Science. 1974;183:46–51. [PubMed]
  • Hosmer DW, Lemeshow S. Applied Logistic Regression. Second Edition. John Wiley & Sons, Inc.; 2000.
  • Jeoung DI, Oehlen LJ, Cross FR. Cln3-associated kinase activity in Saccharomyces cerevisiae is regulated by the mating factor pathway. Mol Cell Biol. 1998;18:433–441. [PMC free article] [PubMed]
  • Jorgensen P, Nishikawa JL, Breitkreutz BJ, Tyers M. Systematic identification of pathways that couple cell growth and division in yeast. Science. 2002;297:395–400. [PubMed]
  • Kosugi S, Hasebe M, Tomita M, Yanagawa H. Systematic identification of cell cycle-dependent yeast nucleocytoplasmic shuttling proteins by prediction of composite motifs. Proc Natl Acad Sci U S A. 2009;106:10171–10176. [PubMed]
  • Lamson RE, Winters MJ, Pryciak PM. Cdc42 regulation of kinase activity and signaling by the yeast p21-activated kinase Ste20. Mol Cell Biol. 2002;22:2939–2951. [PMC free article] [PubMed]
  • Lee PJ, Helman NC, Lim WA, Hung PJ. A microfluidic system for dynamic yeast cell imaging. Biotechniques. 2008;44:91–95. [PubMed]
  • Mateus C, Avery SV. Destabilized green fluorescent protein for monitoring dynamic changes in yeast gene expression with flow cytometry. Yeast. 2000;16:1313–1323. [PubMed]
  • McKinney JD, Chang F, Heintz N, Cross FR. Negative regulation of FAR1 at the Start of the yeast cell cycle. Genes Dev. 1993;7:833–843. [PubMed]
  • Muzzey D, Gomez-Uribe CA, Mettetal JT, van Oudenaarden A. A systems-level analysis of perfect adaptation in yeast osmoregulation. Cell. 2009;138:160–171. [PMC free article] [PubMed]
  • Novak B, Tyson JJ, Gyorffy B, Csikasz-Nagy A. Irreversible cell-cycle transitions are due to systems-level feedback. Nat Cell Biol. 2007;9:724–728. [PubMed]
  • Oehlen LJ, Cross FR. G1 cyclins CLN1 and CLN2 repress the mating factor response pathway at Start in the yeast cell cycle. Genes Dev. 1994;8:1058–1070. [PubMed]
  • Oehlen LJ, Jeoung DI, Cross FR. Cyclin-specific START events and the G1-phase specificity of arrest by mating factor in budding yeast. Mol Gen Genet. 1998;258:183–198. [PubMed]
  • Park MY, Hastie T. L-1-regularization path algorithm for generalized linear models. J Roy Stat Soc B. 2007;69:659–677.
  • Pathak R, Blank HM, Guo J, Ellis S, Polymenis M. The Dcr2p phosphatase destabilizes Sic1p in Saccharomyces cerevisiae. Biochem Biophys Res Commun. 2007;361:700–704. [PMC free article] [PubMed]
  • Peter M, Gartner A, Horecka J, Ammerer G, Herskowitz I. FAR1 links the signal transduction pathway to the cell cycle machinery in yeast. Cell. 1993;73:747–760. [PubMed]
  • Peter M, Herskowitz I. Direct inhibition of the yeast cyclin-dependent kinase Cdc28-Cln by Far1. Science. 1994;265:1228–1231. [PubMed]
  • Schwob E, Bohm T, Mendenhall MD, Nasmyth K. The B-type cyclin kinase inhibitor p40SIC1 controls the G1 to S transition in S. cerevisiae. Cell. 1994;79:233–244. [PubMed]
  • Sherr CJ, Roberts JM. CDK inhibitors: positive and negative regulators of G1-phase progression. Genes Dev. 1999;13:1501–1512. [PubMed]
  • Skotheim JM, Di Talia S, Siggia ED, Cross FR. Positive feedback of G1 cyclins ensures coherent cell cycle entry. Nature. 2008;454:291–296. [PMC free article] [PubMed]
  • Strickfaden SC, Winters MJ, Ben-Ari G, Lamson RE, Tyers M, Pryciak PM. A mechanism for cell-cycle regulation of MAP kinase signaling in a yeast differentiation pathway. Cell. 2007;128:519–531. [PMC free article] [PubMed]
  • Taberner FJ, Quilis I, Igual JC. Spatial regulation of the start repressor Whi5. Cell Cycle. 2009;8:3010–3018. [PubMed]
  • Takahashi S, Pryciak PM. Membrane localization of scaffold proteins promotes graded signaling in the yeast MAP kinase cascade. Curr Biol. 2008;18:1184–1191. [PMC free article] [PubMed]
  • Taylor RJ, Falconnet D, Niemisto A, Ramsey SA, Prinz S, Shmulevich I, Galitski T, Hansen CL. Dynamic analysis of MAPK signaling using a high-throughput microfluidic single-cell imaging platform. Proc Natl Acad Sci U S A. 2009;106:3758–3763. [PubMed]
  • Trotter EW, Berenfeld L, Krause SA, Petsko GA, Gray JV. Protein misfolding and temperature up-shift cause G1 arrest via a common mechanism dependent on heat shock factor in Saccharomycescerevisiae. Proc Natl Acad Sci U S A. 2001;98:7313–7318. [PubMed]
  • Tyers M, Futcher B. Far1 and Fus3 link the mating pheromone signal transduction pathway to three G1-phase Cdc28 kinase complexes. Mol Cell Biol. 1993;13:5659–5669. [PMC free article] [PubMed]
  • Whiteway MS, Wu C, Leeuw T, Clark K, Fourest-Lieuvin A, Thomas DY, Leberer E. Association of the yeast pheromone response G protein beta gamma subunits with the MAP kinase scaffold Ste5p. Science. 1995;269:1572–1575. [PubMed]
  • Wiget P, Shimada Y, Butty AC, Bi E, Peter M. Site-specific regulation of the GEF Cdc24p by the scaffold protein Far1p during yeast mating. Embo J. 2004;23:1063–1074. [PubMed]
  • Wijnen H, Landman A, Futcher B. The G(1) cyclin Cln3 promotes cell cycle entry via the transcription factor Swi6. Mol Cell Biol. 2002;22:4402–4418. [PMC free article] [PubMed]
  • Ydenberg CA, Rose MD. Antagonistic regulation of Fus2p nuclear localization by pheromone signaling and the cell cycle. J Cell Biol. 2009;184:409–422. [PMC free article] [PubMed]
  • Yu RC, Pesce CG, Colman-Lerner A, Lok L, Pincus D, Serra E, Holl M, Benjamin K, Gordon A, Brent R. Negative feedback that improves information transmission in yeast signalling. Nature. 2008;456:755–761. [PMC free article] [PubMed]
  • Zetterberg A, Larsson O, Wiman KG. What is the restriction point? Curr Opin Cell Biol. 1995;7:835–842. [PubMed]