In this study, by combining live-cell imaging and mechanistic whole-cell modeling, we propose a mechanism for Cdc42-controlled septin ring assembly in budding yeast. Unexpectedly, we find that septins provide negative feedback by inhibiting Cdc42-GTP via the recruitment of Cdc42 GAPs and possibly by affecting spatial distribution of other regulatory molecules. By decreasing exocytic activity, we reveal that Cdc42-mediated recruitment of septins results in the formation of a septin cap rather than a ring. We further demonstrate that the hollowing of the septin cap into a ring is generated by polarized exocytosis that is also controlled by Cdc42 activity. Furthermore, we show that this transition results in relocation of active Cdc42 and all of its downstream effectors into the septin ring enclosure. Thus, septin ring formation defines the emergence of distinct daughter cell identity and also prevents it from intermixing with that of mother.
Feedback Loops between Cdc42 Activity, Septins, and Exocytosis Control Septin Ring Biogenesis
We found that several interlinked feedback loops play important roles in septin ring assembly (A). Unexpectedly, we found that septins, in a negative feedback loop, suppress the activity of the underlying Cdc42 cluster. This feedback mechanism completely depends on septins (A and 3B) and, at least partially, on Cdc42 GAPs (C and 3D) that associate with septins at the PBS (E).
Complex Interplay between Cdc42-GTP, Septins, and Exocytosis Shape the Septin Ring and Yeast Bud
Cdc42-GTP controls polarized exocytosis indirectly through formin-generated actin cables (Chen et al., 2012
) and directly through interactions with the exocyst (He and Guo, 2009
). At the same time, exocytosis boosts Cdc42 activity by delivering Cdc42 to the plasma membrane. Our study revealed a mutually antagonistic interaction between septin accumulation and exocytosis. Following the hypothesis that insertion of the exocytic membrane may effectively dilute septin accumulation, we directly demonstrated a negative spatial correlation between the local septin density and the intensity of exocytosis (C). Reciprocal inhibition of exocytosis by the accumulation of septins, as suggested earlier in mammalian cells, is also supported by our imaging results. Furthermore, our model simulations show that this negative interaction is, in fact, essential for the formation of a well-defined septin ring. Abrogation of this antagonistic interaction in the model invariably resulted in progressive fragmentation of the nascent ring due to random exocytic events that occur within the ring.
Negative feedback plays a key role in the regulatory networks of biological systems and has been implicated in generating rapid in-place oscillations of the Cdc42 cluster (Howell et al., 2012
) and traveling waves of the Cdc42 activity (Ozbudak et al., 2005
), both observed in a yeast mutant (rsr1
Δ) lacking spatial landmark-directed polarization. Unlike the latter phenomenon that was significantly reduced by actin depolymerization, translocation of the Cdc42 cluster in our study was observed only when the function of the exocytic machinery was compromised.
Highly Focused Exocytosis Is Crucial for Septin Ring Formation
Our study also uncovered an unappreciated role of polarized exocytosis in the spatial control of septin-mediated inhibition of Cdc42 activity at the PBS. In WT cells, frequent and spatially focused exocytosis could continuously displace the accumulating septins centrifugally, thus making the ring the only well-resolved septin structure observed at the PBS (Chen et al., 2011
). In contrast, reduced frequency of targeted exocytosis in latA-treated cells allows septins to accumulate at the PBS in the shape of a cap and inhibit Cdc42 activity within it, thus causing dissipation of the underlying Cdc42 cluster. Based on the rate of growth of total cell surface in our experiments, latA reduces the spatially-averaged rate of exocytosis by at least 5-fold. This is likely a gross underestimation because, in cells untreated with latA, cell surface growth is opposed by endocytosis whose rate could not be determined and thus accounted for in our simple estimate.
The disassembly of the cluster at the original PBS and its reestablishment at a new location followed by septin relocation to this site constitute the chasing phenomenon as seen in some latA-treated cells. Observation of chasing in the sec4-8
mutant at the nonpermissive temperature (Figures S6
C and S6D) further suggests that impaired exocytosis, not other effects of latA, is the cause of chasing. In the model, chasing is observed not only in the total absence of exocytosis but also with a weak exocytosis. The increase in the frequency of exocytosis first slows and then completely stops chasing indicating that, for the stable formation of a septin ring, the intensity of exocytosis must exceed certain threshold.
Does chasing have any adaptive function? We speculate that chasing may improve survival rate of cells with decreased exocytic activity. Reduced rate of exocytosis could potentially cause bud malformation resulting in failed cytokinesis and cell death. Septin-mediated negative feedback ensures that cells with subthreshold exocytic activity, rather than allowed to proceed into later developmental stages with a potentially defective bud, are redirected into the abortive pathway leading to the complete disassembly of the unsuccessful PBS. The following reestablishment of the PBS at a new location offers the cell a new chance to form a bud. Indeed, with the overall reduction in the average rate, exocytosis becomes more stochastic and thus uneven in time (C). Therefore, it is plausible that concomitant with a subsequent attempt at the PBS establishment, a sudden burst of exocytosis may provide just enough membrane to generate a stable ring opening. In support of this hypothesis, in both latA-treated WT and sec4-8 mutant cells, we observed septin ring formation that followed after a variable period of chasing. Together, our experimental data and modeling suggest that highly focused exocytosis is critically important for septin ring formation.
Cdc42 and Septin Ring Define Distinct Cell Identity of the Nascent Bud
Using the interactions depicted in A, the exocytosis-dependent transition from a septin cap to a ring can be explained using the diagram in B. Insertion of new membrane at the peak of exocytosis probability function (blue) causes septin density to change from a unimodal profile to a profile with a depression in the middle. This change partially relieves the septin-mediated repression of Cdc42 activity in the center but increases it on the sides of the cluster causing the profile of the Cdc42 cluster to become even narrower. Delivery of Cdc42 with the vesicles can also contribute to the increase of Cdc42 activity within the nascent ring opening. Because the probability of exocytosis is positively regulated by Cdc42-GTP and, independently, negatively by septins, its spatial profile changes as shown in B. Thus, the probability of exocytosis in the center of the septin depression increases further whereas in the area occupied by the nascent septin ring it drops. Subsequent exocytic events further amplify Cdc42 activity inside the forming ring, leading to its dramatic rise that is robustly observed in our experiments (A, 1B, and A) and fully reproduced in our model behavior (C and 4D). This increase in the activity of Cdc42 on the membrane of the nascent tiny bud can be explained by the synergy of the direct positive feedback due to the exocytosis-dependent Cdc42 transport and the indirect exocytosis- and septin-dependent positive feedback loop characterized in this study (A, dashed line).
The exclusive localization of activated Cdc42 within the nascent septin ring results in inevitable relocation into the bud of all Cdc42 effectors and their interactors. Because the exocytic activity is also confined to the bud, the enclosure of the septin ring defines a membrane domain with distinct lipid and protein composition. This membrane domain is a first manifestation of the unique daughter identity that is prevented from remixing with that of mother by the septin diffusion barrier.
Septin Ring and Bud Neck Are Shaped by a Common Morphogenetic Process
The interplay between the morphology of the septin ring and the cell shape has long attracted attention in the field (Caviston et al., 2003; Gladfelter et al., 2005
). We observed that cells deleted for Cdc42 GAPs formed broad dome-shaped, shmoo-like protrusions in the presence of latA (F; Movie S2
). Septins were either distributed uniformly over the entire protrusion or concentrated into irregular fragments at the base of the protrusion. Such phenotypes can be explained using the relationships established in our study (A and 7B). The GAP mutant (bem2
Δ or rga1
Δ) cells formed large Cdc42 clusters with abnormally high total activity of the GTPase (). Consequently, septins recruited to these clusters formed broader and therefore less dense caps, providing weaker suppression of the Cdc42 activity than in WT cells. In the absence of actin cables that focus the spatial distribution of exocytosis, insertion of new membrane is directed largely by the broad spatial profile of Cdc42-GTP. Membrane vesicles randomly inserted within the Cdc42 cluster further widen the cluster and dilute the septin density. Moreover, the generation of two distinct zones, the septin-dense “ring” that excludes exocytosis and the exocytosis-generated “hole” that excludes septins, does not occur. Instead, the broad initial distribution of Cdc42-GTP and exocytosis generate a protrusion in which the Cdc42 activity, septin density and exocytosis remain spatially distributed as shown in the left column of B.
To test our hypothesis that the shape of the spatial profile of exocytosis is responsible for the morphological dichotomy between a “bud” that is separated from its mother by a slim hourglass-shaped neck and a tapered conical “shmoo,” we resorted to a computational model that permits evolution of the cell shape with time (see Supplemental Experimental Procedures
for details). In these simulations, starting with a spherical cell with a polar septin cap, the cell membrane was continuously expanded using two distinct spatial profiles shown in C. A characteristic “neck” with a diameter narrower than that of the expanding “bud” was formed only under the assumption of a ring-shaped zone where the probability of exocytosis was practically negligible. In these simulations, septins were segregated into this zone, forming a septin-dense ring whereas the membrane of the nascent bud was essentially septin-free. In contrast, the bell-shaped profile of membrane expansion led to the generation of a shmoo-like protrusion with continuously distributed septin density. Thus we conclude that the lack of a well-defined septin ring results in the loss of both the distinct membrane identity and the bud-like shape of a nascent protrusion.
In this study, we provide a coherent systemic model describing how an initially homogeneous cell can give rise to two distinct membrane identities that are then amplified and transformed into two different cellular fates. In its present form, the model captures only the most salient features of this process and, therefore, is inevitably incomplete. Many additional factors not considered here explicitly, such as membrane phospholipids, other small Rho GTPases, cell wall and factors that affect its remodeling, are likely to contribute to the emergence of bud identity and elucidation of their role is the subject of future work. The main driver of eukaryotic cell polarity, Cdc42, plays dual role in this process by both defining the identity of daughter membrane and creating the diffusion barrier in the form of a septin ring. Both functions are vitally important for the robust segregation of mother and daughter cellular identities. Future studies will likely reveal further contributions of polarity regulators and, in particular, Cdc42, in the differential distribution of cell-fate factors in other asymmetrically dividing cells.