Dynamics of Rga1, a Cdc42 GTPase-activating protein, at the old and current cell division sites in budding yeast

Establishment and maintenance of cell polarity is critical for essential processes including cell movement, cell growth and division. The Rho GTPase Cdc42 plays a central role in polarity development from yeast to humans. In budding yeast, the axis of polarization is determined by the selection of a growth site (i.e. bud site). Yeast cells never bud within old cell division sites and always choose a new bud site. How yeast cells choose a single unused site for polarization is not well understood. Previous studies have shown that Rga1, a GTPase‐activating protein (GAP) for Cdc42, is uniquely required for proper bud‐site selection among Cdc42 GAPs. Rga1 localizes to old and current cell division sites to inhibit Cdc42 repolarization. To better understand how Rga1 regulates Cdc42 polarization, we examined Rga1 dynamics at the current and old cell division site(s) using methods in quantitative microscopy. Here we show that Rga1 is dynamic at both current and old division site(s). Time‐lapse imaging reveals that Rga1 protein is recruited to old bud site(s) in each cell division cycle, rather than being inherited and stably anchored at old division site(s). We also found that localization of Rga1 to old bud sites is dependent on the interdependent transmembrane proteins, Rax1 and Rax2, and Nis1, another protein that localizes to the bud neck and old bud sites. Our close‐up analyses of Rga1 localization in the G1 phase indicate that Rga1 at the current division site delocalizes during the T2 phase (the second phase of G1 partitioned by the nuclear exit of the transcriptional repressor Whi5) in wild‐type daughter cells. In contrast, in rax1Δ daughter cells, Rga1 delocalized before the T1‐T2 transition (when a new bud site is established). This earlier delocalization of Rga1 in rax1Δ daughter cells is likely to cause Cdc42 repolarization within the division site in rax1Δ daughter cells, leading to a bud‐site selection defect. This conclusion is further supported by computational modeling.