PerspectiveCell Biology

Cdc42 Oscillations in Yeasts

See allHide authors and affiliations

Science Signaling  04 Dec 2012:
Vol. 5, Issue 253, pp. pe53
DOI: 10.1126/scisignal.2003630


A fundamental problem in cell biology is how cells define one or several discrete sites of polarity. Through mechanisms involving positive and negative feedback, the small Rho-family guanosine triphosphatase Cdc42 breaks symmetry in round budding yeast cells to define a single site of polarized cell growth. However, it is not clear how cells can define multiple sites of polarization concurrently. We discuss a study in which rod-shaped fission yeast cells, which naturally polarize growth at their two cell ends, exhibited oscillations of Cdc42 activity between these sites. We compare these findings with similar oscillatory behavior of Cdc42 detected in budding yeast cells and discuss the possible mechanism and functional outputs of these oscillations.

How cells define discrete sites of polarity at defined locations is a fundamental cell biological problem. In the budding yeast Saccharomyces cerevisiae, the small Rho-family guanosine triphosphatase (GTPase) Cdc42 breaks symmetry to specify a single incipient bud site in round cells lacking polarity cues (Fig. 1A). Symmetry breaking relies on self-amplifying positive feedback mechanisms that convert an initial stochastic distribution of Cdc42 into a single localized cluster of Cdc42 molecules (1, 2). Negative feedback mechanisms, revealed by observing transient oscillations of Cdc42 clusters, also contribute by improving the robustness of symmetry breaking (3). In S. cerevisiae, these reactions ensure the choice of one, and only one, site of polarized cell growth.

Fig. 1

Schematic of Cdc42 oscillations in budding yeast S. cerevisiae and fission yeast S. pombe. (A) Cdc42 (orange) oscillations in budding yeast, followed by bud site selection and growth. (B) Cdc42 (orange) oscillations during monopolar and bipolar growth in fission yeast. Blue arrows indicate cell growth.


How can cells define and maintain two or more sites of polarity concurrently? Multiple coexisting sites of polarization are necessary for achieving the complex connectivity of neurons or branching growth observed in fungal hyphae. Das et al. provided insight into this question by describing previously unknown oscillatory dynamics of Cdc42 in the fission yeast Schizosaccharomyces pombe using imaging and mathematical modeling (4). Rod-shaped fission yeast cells naturally grow in a bipolar manner at their two cell poles, a growth pattern achieved through a nonessential growth transition (named NETO for new end take-off) from mono- to bipolarity in which the end of the cell formed by the previous division (new cell end) begins to grow in G2 phase of the cell cycle. How cells that are initially monopolar later initiate growth at the second cell pole remains largely mysterious.

Das et al. reported that the amount of active Cdc42, detected through localization of a reporter composed of a CRIB (Cdc42/Rac interactive binding) domain fused to green fluorescent protein (CRIB-GFP), oscillated between the two cell poles in an anticorrelated manner: When the amount of CRIB-GFP decreased at one end, it increased at the other (Fig. 1B). Oscillations in Cdc42 activity happened not only in bipolar cells but also in cells that had not undergone NETO and were still monopolar. However, in monopolar cells, the amplitudes of Cdc42 activity at each pole were asymmetric, with the new cell end exhibiting amounts far below those of the growing end. In 50% of cells, these oscillations had a regular periodicity of about 5 min. These observations of Cdc42 oscillations add to a growing list of oscillatory systems. In addition to ones that lack a spatial component, such as circadian clocks and metabolic oscillations, multiple oscillatory systems have a clear spatial dimension: for example, periodic morphogenesis in budding yeast (5), pollen tube growth (6), or the bacterial Min system (7).

Building on proposed general mechanisms of oscillation (8), the authors developed a mathematical model that reproduced the general characteristics of Cdc42 oscillations in fission yeast through three major assumptions: (i) positive feedback promoted the accumulation of Cdc42 at a defined location; (ii) the positive feedback combined with delayed negative feedback to promote oscillations; and (iii) limiting amounts of Cdc42 regulatory factors ensured that the cell poles would compete for these factors and exhibit anticorrelative behavior.

Positive feedbacks are well understood in the budding yeast system and may apply to the fission yeast case. In S. cerevisiae, the scaffold protein Bem1 coordinates the location of Cdc42, Cdc24 [the guanine nucleotide exchange factor (GEF) for Cdc42], and any one of three PAK kinases (Cla4, Ste20, and Skm1) to positively reinforce zones of active Cdc42. Active Cdc42 binds the PAK kinase, which through Bem1 binding recruits Cdc24, thus promoting further local activation of Cdc42 (1). In fission yeast, mechanisms of positive feedback may be similar and could involve the scaffold Scd2, the Cdc42 GEF Scd1, and the PAK kinase Pak1 (also known as Orb2 or Shk1), because these three proteins form a complex with Cdc42 (911).

Negative feedback mechanisms in Cdc42 oscillations are less well characterized. Das et al. suggested that delayed negative feedback may be driven by the activity of the same PAK kinase, Pak1, likely involved in positive feedback. A mutation in Pak1 that decreases kinase activity led to asymmetric oscillations and increased amounts of active Cdc42 at the growing cell pole. Pak1 can phosphorylate both itself and the scaffold Scd2 (9), either of which could potentially lead to a decrease in active Cdc42. Similar inhibition of an activator was also proposed as a possible mechanism for driving negative feedback mechanisms in budding yeast (3): The PAK kinase Cla4 may phosphorylate Cdc24 (12). PAK kinase–mediated phosphorylation of Cdc24 was suggested to form part of a negative feedback loop important for ending the polar growth phase, although mutation of up to 35 phosphorylation sites in Cdc24 does not prolong this phase or produce any obvious polarization defect (13). However, phosphorylation by Cla4, or by another PAK kinase, could be involved in the negative feedback that enhances the robustness of symmetry breaking.

Alternatively, negative feedback could result from the activation of an inhibitor (3). The inhibitor could be a GTPase-activating protein (GAP) for Cdc42. Das et al. explored this possibility by deleting the gene encoding Rga4, the only known GAP for Cdc42 in S. pombe. However, in rga4∆ cells, Cdc42 oscillations were similar to those observed in wild-type cells. Perhaps another, uncharacterized GAP may be involved, possibly redundantly with Rga4. For the budding yeast system, GAP-mediated inhibition of Cdc42 remains a possibility as well.

Whether the negative feedback relies on inhibition of an activator or activation of an inhibitor, for oscillations to occur, a mechanism must be in place to dampen it. If the negative feedback relies on PAK kinase activity, this predicts the existence of a phosphatase for removal of the inhibitory phosphate group. An example of oscillations mediated by a phosphorylation cycle is the oscillations in phosphorylation and dephosphorylation of KaiC, which drive circadian rhythms of the cyanobacterium Synechococcus elongatus and has been reconstituted in vitro with a minimum set of factors (14). It is thus attractive to consider that Cdc42 oscillations could be controlled by similar cycles of phosphorylation and dephosphorylation.

Cdc42 oscillations in budding and fission yeast display many similarities. In addition to employing similar feedback mechanisms, Cdc42 oscillations in budding and fission yeast both occur with 5-min periods, although it remains unknown whether or how this parameter influences growth or polarity establishment. Both organisms may also have the capacity for both monopolarity and bipolarity. Indeed, regions in the parameter space of the Cdc42 oscillation model in budding yeast allowed for cell growth with two stable sites of polarization, suggesting that the shift from mono- to bipolarity can be achieved by quantitative changes of the same polarity circuit, rather than complete rewiring (3).

However, the two systems also display fundamental differences in cell geometry and in the physiological role of oscillations. The S. cerevisiae studies describe events of symmetry breaking in mutant backgrounds lacking protein landmarks to define the incipient bud site. In this case, the oscillations are transient and occur in an almost spherical mother cell, which does not grow considerably. By contrast, the oscillations in S. pombe occur throughout interphase in growing wild-type cells, which do not break symmetry per se, because the growth sites are already defined by the localization of protein landmarks and the cells are naturally rod-shaped. Perhaps due to these system differences, fluctuations in abundance of the proteins involved have distinct effects in the two organisms. In budding yeast, negative feedback serves to buffer against such fluctuations (3). By contrast, mathematical modeling of fission yeast growth predicted that the concentration of the regulatory factors, including GEFs, would be limiting for the establishment of symmetric oscillations in small monopolar cells, because the growing pole acts as a sink depleting the cytoplasmic pool of proteins involved (4). Cell growth would result in increased abundance of GEFs, thus promoting symmetric oscillations. Experimentally increasing or decreasing the concentration of the GEF Gef1 or increasing cell length fit the model predictions.

Das et al. showed that NETO correlated with an increase in the amount of CRIB-GFP at the new cell end, supporting the idea that local concentrations and distributions of active Cdc42 control polarized growth (15). Cdc42 oscillations thus suggest a previously unknown way that cells could regulate the transition from monopolar to bipolar growth. Instead of a classical switch model, in which a signal activates growth at the new cell end at a defined time, the existence of oscillations suggests that the transition may occur stochastically between two existing potential states, likely helped by intrinsic noise.

The model may explain why the transition to bipolarity requires a minimal cell length (16). Previous studies had suggested that an existing growth site exerts long-range inhibition, preventing the formation of a new growth site nearby (17, 18). The results of Das et al. agreed with this idea, with one zone of active Cdc42 forming a local sink for activators. By increasing the total pool of Cdc42 GEFs and other factors while maintaining a constant cell tip size, cell growth may simply increase the availability of GEFs for both poles, which would trigger symmetric oscillations of activated Cdc42 at both cell tips, allowing bipolar growth. However, this hypothesis remains to be tested, because the authors did not show whether increasing GEF abundance, which triggers symmetric Cdc42 accumulation, was sufficient to promote bipolar growth in shorter cells.

Equal amounts of active Cdc42 at both cell poles may in fact not be sufficient to trigger bipolar growth. Bipolar growth depends not only on a minimal cell length but also on passage through S-phase and localization of the molecular landmarks Tea1, Tea4, and Pom1 (19). For example, G1-arrested cells remain monopolar, despite substantial growth in length, which, according to the model, should result in sufficient GEF abundance for symmetric accumulation of active Cdc42 and bipolarity. Specific mutant combinations have also been described that have symmetric Cdc42 localization, yet maintain monopolar growth (20). Although they did not investigate cell cycle requirements, Das et al. showed that Cdc42 oscillations still occurred in cells lacking the landmark Tea1, yet the amplitudes of Cdc42 activity at each pole were very asymmetric. This pattern resembles that shown for the pak1 mutant, which suggests that Tea1 could be involved in the proposed negative feedback loop. Intriguingly, Tea1 is a likely substrate of Pak1, and genetic evidence suggests a functional link between the two proteins (21).

Important questions remain about the relationship between Cdc42 oscillations and bipolar growth in S. pombe. Are effectors of Cdc42 also oscillating, or are oscillations in Cdc42 activity buffered by the abundance or residence time of the effectors? What specific oscillation parameters—amplitude, period, or anticorrelative nature—may correlate with, or even promote, bipolar growth? This is not immediately apparent. For example, although all cells appear to show anticorrelated changes in activity of Cdc42 at their two poles, only about 50% display a regular 5-min interval. Although future work may reveal such connections, another possibility is that the oscillations do not promote bipolarity per se but are a by-product of the negative feedback requirement for establishing a bipolar system. An interesting parallel can be made here with the bacterial Min system, which positions the site of division at the cell middle by inhibiting division at the cell poles. The Min system is oscillatory in Escherichia coli but not in Bacillus subtilis (7), suggesting that the oscillations may not be important for function. Reproduction of the E. coli oscillation system in B. subtilis opens the door to such question (22).

In closing, the discovery of Cdc42 oscillations by Das et al. provides a new system to study the mechanisms of biological oscillations and opens a new path for thinking about how S. pombe cells regulate the transition to bipolarity. Although many questions remain open, the discovery of Cdc42 oscillations in both budding and fission yeasts will most certainly lead to fundamental advances in our understanding of both the mechanisms driving oscillations and their functional output. Careful and detailed analysis of protein localization over time made these discoveries possible, demonstrating the power of live-cell imaging in unearthing fundamental cell behaviors.

References and Notes

Acknowledgments: We thank D. Lew for critical reading of the manuscript. Funding: Research in S.G.M.’s laboratory is supported by a Swiss National Science Foundation Research grant (31003A_138177) and a European Research Council Starting Grant (260493).
View Abstract

Navigate This Article