Productive cell proliferation involves efficient and accurate splitting of the dividing cell into two separate entities. This orderly process reflects coordination of diverse cytological events by regulatory systems that drive the cell from mitosis into G1. In the budding yeast Saccharomyces cerevisiae, separation of mother and daughter cells involves coordinated actomyosin ring contraction and septum synthesis, followed by septum destruction. These events occur in precise and rapid sequence once chromosomes are segregated and are linked with spindle organization and mitotic progress by intricate cell cycle control machinery. Additionally, critical parts of the mother/daughter separation process are asymmetric, reflecting a form of fate specification that occurs in every cell division. This chapter describes central events of budding yeast cell separation, as well as the control pathways that integrate them and link them with the cell cycle.
Proliferation of most eukaryotic cells requires productive partitioning of intracellular components and the physical separation of this material into two distinct and independent living entities. This involves coordination of diverse processes and mechanical systems in time and space. In some cases these oppose each other, as exemplified by the construction, constriction, and disassembly of the actomyosin ring. General understanding of the molecular machines and regulatory networks underlying eukaryotic cell division and separation has been significantly advanced by studies in S. cerevisiae, which uses mechanisms that are conserved in fundamental organization, and often their precise components, across vast evolutionary distances.
Like other hemiascomycetes, Saccharomyces cerevisiae is an asymmetric organism that proliferates by forming a daughter bud that grows rapidly while the mother cell remains essentially the same size. This asymmetry persists through the process of division and cell separation, producing two individual cells with different properties. In mitosis, the mitotic spindle is positioned to ensure proper segregation of chromosomes, and separation proceeds in temporally distinct stages. First, beginning in late G1, the cytokinesis site is organized at the narrow mother/bud neck. Second, the cytoplasm is roughly halved by cytokinesis, via two interrelated but partly independent mechanisms: actomyosin ring contraction and formation of a specialized wall between the two cells called the septum. Finally, the septum is degraded and the mother and daughter cells physically dissociate. The process of cytokinesis and cell separation is closely linked to the positioning and division of the nucleus, which occurs without breakdown of the nuclear envelope. Nuclear processes are organized by the mitotic spindle, which is disassembled at about the same time as cytokinesis.
In three related parts, this chapter describes physical events and regulatory pathways that underlie separation of mother and daughter budding yeast cells into two distinct entities as they pass from mitosis into G1. Part 1 provides an overview of critical mechanical events of cell separation. This is followed in part 2 by a summary of the regulatory systems that control mitotic exit and link the mechanical processes of division to the cell cycle. These include the FEAR (Cdc fourteen early anaphase release) and MEN (mitotic exit network) pathways, which drive passage from the metaphase/anaphase transition in late mitosis to early G1, as well as the RAM network (regulation of Ace2 and morphogenesis), which controls septum destruction and thus the final events of cell separation. Part 3 discusses the regulatory connections that link these control pathways to the processes that drive cell division.
This chapter emphasizes mechanisms that orchestrate the execution and timing of the late events of cell division, in particular processes that happen after the actomyosin ring contracts. I mention some critical subjects largely in overview that are described extensively elsewhere. These include the late mitotic spindle and contraction of the cytokinetic apparatus, which are covered in other YeastBook chapters by Bi and Park (2012) and Winey and Bloom (2012), respectively, and are also reviewed extensively elsewhere (Tolliday et al. 2001; Walther and Wendland 2003; Balasubramanian et al. 2004; Moseley and Goode 2006; Moseley and Nurse 2009; Roncero and Sanchez 2010).
Mechanics of Mother/Daughter Separation
Two major things happen when budding yeast cells divide: partitioning and separation of the cytoplasm and division of the nucleus. These processes are closely interlinked to ensure that genetic material is properly segregated to the mother and daughter cells. Division of the cytoplasm comprises (A) construction and function of a contractile actomyosin ring and (B) deposition of a multilayered septum, followed by (C) destruction of the septum to allow final separation of the divided cells. Coinciding with these processes, (D) division of the nucleus and disassembly of the spindle occurs concurrently with cytokinesis.
Assembly and contraction of the actomyosin ring
Assembly of the S. cerevisiae cytokinesis site begins at the earliest stages of budding, with the formation of a narrow bud neck as cells pass from G1 into S phase. During the first stages of bud formation the septin proteins, which form filaments without intrinsic polarity (Frazier et al. 1998), are recruited from a soluble cytoplasmic pool to a polymeric form at the border between the mother cell and emerging daughter bud (reviewed in Weirich et al. 2008; Caudron and Barral 2009; McMurray and Thorner 2009; Oh and Bi 2011). As diagrammed in Figure 1, this septin filament system eventually forms a focused band that extends around the bud neck, close to the plasma membrane (Longtine and Bi 2003; Kinoshita 2006; Oh and Bi 2011). The septin lattice is initially highly dynamic, but reorganizes into a more stable structure as bud growth proceeds (Dobbelaere et al. 2003; Dobbelaere and Barral 2004; Vrabioiu and Mitchison 2006; Demay et al. 2011). The septin ring at the bud neck functions as a barrier that prevents diffusion of membrane proteins and other cell cortex material (Barral et al. 2000; Dobbelaere and Barral 2004; Vrabioiu and Mitchison 2006; Caudron and Barral 2009).
Other proteins involved in bud neck morphogenesis and cytokinesis are recruited to the septin collar as bud emergence occurs. These include Myo1, a type II myosin directly involved in actomyosin ring contraction, which localizes to the bud neck almost as soon as it forms but has no apparent function there until cytokinesis at the M/G1 transition. The septin band extends continuously through the bud neck until cytokinesis, when it splits in two, leaving rings on both the mother and daughter sides of the bud neck. This has been proposed to create a restricted membrane domain at the bud neck between the two remaining septin rings, allowing concentration of phosphoinositides and membrane-associated proteins involved in actomyosin ring contraction (Dobbelaere and Barral 2004).
Actomyosin ring contraction is accompanied by synthesis of a specialized multilayered septum between mother and daughter cells (Figure 1), which is built in distinct but nearly coincident phases. First, a thin primary septum composed primarily of chitin, a polymer of N-acetylglucosamine, forms behind the ingressing cytokinetic furrow. Primary septum synthesis and actomyosin ring contraction are intimately linked, with the ingressing furrow guiding an active zone of chitin synthesis. Closely following primary septum formation, the structure is thickened on both mother and daughter sides by localized deposition of additional chitin and other wall polymers (Figure 2), forming the secondary septum. This occurs as the actomyosin ring disassembles and the chitin synthase II machinery (see below) is endocytosed. Once the primary and secondary septa are complete, the enzymes responsible for chitin export are internalized, and the synthesis of other septum components is also downregulated.
Septation is a form of cell wall synthesis, and so in addition to dedicated enzymes it uses many of the same synthetic machines that assemble the dynamic cell wall covering the rest of the cell. Indeed, defects in protein mannosylation affect septation; this process occurs as proteins that span the plasma membrane or are secreted move through the endomembrane system, and is broadly important for the function of cell wall components (Schmidt et al. 2005). A forthcoming YeastBook chapter by Orlean and Strahl will discuss cell wall organization mechanisms in detail; additionally, the process by which this septum is built is the subject of a number of excellent reviews (Cabib et al. 2001; Roh et al. 2002; Roncero and Sanchez 2010).
Synthesis of the primary septum:
The primary septum is made of chitin deposited as long fibers that are not extensively cross-linked to other cell wall components (Cabib and Duran 2005). Chitin in the primary septum is synthesized in situ by chitin synthase II, of which the protein Chs2 is the catalytic subunit. This ∼110-kDa enzyme spans the plasma membrane and likely forms a channel through which the chitin polymer is extruded (Sburlati and Cabib 1986). Chs2 is exclusively involved in septation (Shaw et al. 1991). Consistent with this role, it is largely produced in mitosis, but is retained in the endoplasmic reticulum (ER) until mitotic exit (Chuang and Schekman 1996; Zhang et al. 2006; Teh et al. 2009). As discussed in part 3, when Chs2 is dephosphorylated as cells progress from M to G1, it proceeds through the secretory pathway to the site of cytokinesis and concentrates at the actomyosin ring (Schmidt et al. 2002; VerPlank and Li 2005; Teh et al. 2009). Chitin synthase II function is intimately connected with actomyosin ring contraction and polarization of membrane trafficking to the cytokinesis site and is also directly controlled by a mitotic exit regulatory system (Bi 2001; Schmidt et al. 2002; Oh et al. 2012).
While the primary septum’s main structure is provided by chitin fibrils extruded by Chs2, additional proteins concentrate at the structure to promote its integration with the rest of the septum. A fraction of the cell wall reorganizing enzyme Gas1 localizes to the primary septum upon cleavage of a GPI-linked portion of the protein that anchors it to the plasma membrane (Rolli et al. 2009). Similarly, Cwp1 is also trafficked to the site depending on when it is produced (Smits et al. 2006). These proteins help cross-link the outer layers of the chitinous primary septum to the adjacent glucan-rich secondary septum, ensuring that the multilayered structure does not delaminate.
The role of Rho GTPases in primary septum formation:
Rho family GTPases are critical for eukaryotic cytokinesis, and concentration of active Rho at the cell division site with proper timing is important for normal progression (Nishimura et al. 1998; Bement et al. 2005; Piekny et al. 2005). In budding yeast the GTPase Rho1, which plays important roles in spatial organization of cell growth and membrane trafficking, localizes to the cytokinesis site and performs key functions in multiple steps of septum biogenesis (Kohno et al. 1996; Qadota et al. 1996). Assembly of the actomyosin ring, which guides deposition of the primary septum, requires Rho1. This in part reflects its role in cytokinetic activation and targeting of the formin Bni1, which nucleates assembly of actin cables (Tolliday et al. 2002). Upon phosphorylation by the polo-like kinase Cdc5 in late M phase, the Rho guanine nucleotide exchange factor (GEF) Tus1 accumulates at the cytokinesis site and recruits Rho1 (Yoshida et al. 2006) by an unclear mechanism. The GEF then generates active Rho1–GTP at that site, which triggers formin-mediated actin cable assembly (Yoshida et al. 2009). Rho1 may also directly regulate other aspects of actomyosin ring contraction and primary septum deposition.
Synthesis of the secondary septum and remedial septation:
The secondary septum is a region of new cell wall material deposited next to the ingressing primary septum. Synthesis of β1–3 glucan by Fks1/glucan synthase is critical for this process (Cabib et al. 2001; Lesage et al. 2004, 2005; Lesage and Bussey 2006). The chitin synthase Chs3 also localizes to the site of secondary septum formation, providing additional structural reinforcement with chitin (Ziman et al. 1998; Cabib et al. 2001; Schmidt et al. 2002; Cabib and Schmidt 2003; Ortiz and Novick 2006). Delivery of other wall components by secretion is probably important, as there is a strong concentration of exocytic machinery to the site (Dobbelaere and Barral 2004; Zhang et al. 2006). For example, the “exocyst” component Sec3 (reviewed in Heider and Munson 2012; Liu and Guo 2012) acts in septation parallel to actomyosin ring contraction, suggesting that delivery of material through the secretory system promotes secondary septum formation (Dobbelaere and Barral 2004).
Budding yeast septation is robust and can occur in the absence of either primary septum synthesis or actomyosin ring contraction. This occurs via synthesis of a “remedial septum,” which is essentially a secondary septum deposited at the bud neck in a way that is disorganized yet sufficient to separate mother and daughter cells (Cabib and Schmidt 2003; Tolliday et al. 2003). In fact, complete absence of chitin in both the primary and secondary septa can be compensated by deposition of other polymers in this remedial structure, albeit poorly (Schmidt 2004). Intriguingly, elimination of the primary septum synthesis machinery selects for whole chromosome aneuploidies that enhance remedial septum formation (Rancati et al. 2008).
Rho1’s role in secondary septum formation:
In addition to its GEF-mediated recruitment, Rho1 has a short plasma membrane localization domain at its C terminus, containing a CAAX prenylation motif and a polybasic sequence (PBS) that mediates phospholipid association (Heo et al. 2006). This likely helps concentrate Rho1 at membrane regions enriched in phosphatidylinositol 4,5-bisphosphate, (PIP2), a phospholipid generated by the phosphoinositide kinase Mss4 (Homma et al. 1998; Yoshida et al. 2009). Intriguingly, a Rho1 C-terminal fragment concentrates at the bud neck during late division, consistent with localized PIP2 generation in a restricted region of membrane at the bud neck bounded by septin cortical barriers (Dobbelaere and Barral 2004). This mechanism for Rho1 recruitment may be crucial for secondary septum formation: when primary septum deposition is abrogated, increasing either Rho1 or PIP2 levels helps alleviate the resulting division defect (Yoshida et al. 2006). Rho1 activates Fks1 glucan synthase and initiates recruitment of Chs3, likely promoting secondary septum construction by locally activating these polymer synthesis machines (Mazur and Baginsky 1996; Qadota et al. 1996; Inoue et al. 1999). Accordingly, loss of the Rho–GTPase activating protein (GAP) Lrg1 results in much thicker secondary septa (Svarovsky and Palecek 2005). Rho1 is also involved in activation and recruitment of Sec3 and the exocytic apparatus, which delivers chitin synthases for construction of both primary and secondary septa (Guo et al. 2001; Roumanie et al. 2005; Baek et al. 2010; Wu and Brennwald 2010; Yamashita et al. 2010). Overall, these findings suggest that Rho1 directly coordinates multiple distinct processes to ensure that cytokinesis and septation proceed rapidly and robustly (Figure 3).
Once the cytokinetic furrow resolves and the primary and secondary septa are completed, mother and daughter cells remain linked by the septum. This connection is then cut, allowing the two cells to dissociate (Figure 4). Destruction of the septum requires enzymatic digestion of chitin in the inner layer, and to a more subtle extent, the digestion and/or remodeling of the secondary septum (Cabib et al. 2001; Walther and Wendland 2003; Roncero and Sanchez 2010).
Unlike the transmembrane machinery that synthesizes chitin and glucan polymers, cell separation enzymes are secreted and hence are beyond the reach of intracellular regulatory mechanisms. Yet, because of the critical importance of wall integrity, septum degradation must proceed via carefully targeted dissolution and reorganization of cell wall materials, coordinated with wall reinforcement. Septum destruction occurs nearly immediately after cytokinesis, displaying a temporal coordination with the completion of membrane abscission and septum synthesis that remains poorly understood.
Remarkably, the septum is degraded exclusively from the daughter cell side. This explains a long-noted observation: mother cells are left with a chitin-rich remnant of the septum called the “bud scar” after separation, while daughter cells have a region of new cell wall material at the corresponding site referred to as the “birth scar” (Belin 1972; Powell et al. 2003). This asymmetry reflects exclusive expression of septum destruction proteins in the newly born daughter cell, via mechanisms discussed in part 3.
Enzymes of septum destruction: chitinase:
Degradation of chitin in the primary septum is the key event in the final separation of mother and daughter cells. This is carried out by the chitinase Cts1, an enzyme with an N-terminal glycoside hydrolase domain plus a C-terminal carbohydrate-binding module that binds specifically to chitin, which may foster localized activity by anchoring it to specific regions of the septum and surrounding wall (Kuranda and Robbins 1991). Deletion of CTS1 causes dramatic defects in cell separation. The crystal structure of Cts1’s enzymatic domain (Hurtado-Guerrero and van Aalten 2007) yields insight into its specificity for the chitin oligosaccharide. The protein localizes to the region of the septum of a cell undergoing separation, in a distribution biased toward the daughter cell side (Colman-Lerner et al. 2001), consistent with its daughter-specific expression pattern (see part 3). Cts1 is highly glycosylated, but the details of its secretory trafficking, and whether any steps are regulated, are unknown.
Enzymes of septum destruction: endoglucanases/glucanosyltransferases:
In addition to Cts1, efficient septum removal requires a suite of known and putative hydrolytic enzymes that target wall components other than chitin. Some of these probably remove cell wall polymers in the secondary septum, while others locally reorganize the wall by transferring glucan chains from one macromolecule to another. These enzymes may function primarily to open up the secondary septum to facilitate delivery of chitinase to the primary septum. The proteins Dse2, Dse4, Egt2, and Scw11 are known or likely hydrolytic enzymes that function exclusively in mother/daughter separation. Additionally, optimal cell separation requires glucan remodeling proteins that function more broadly in wall organization during bud growth. These include the glucanosyl transferase Gas1 and members of a closely related group of putative glucanases collectively referred to as SUN proteins (Sim1, Uth1, Nca3; also Sun4 and possibly Tos1). These proteins probably do not function in primary septum degradation, and loss of any one causes relatively subtle separation defects. In the few cases tested, combined deletion of multiple genes enhances the separation defect (Mouassite et al. 2000a,b). Thus, these proteins probably play partially overlapping or redundant roles in secondary septum destruction and remodeling, in addition to functioning in cell wall organization during bud growth.
The best understood cell separation endoglucanase is Dse4, which was extensively characterized in studies that refer to it as Eng1 (Baladron et al. 2002). It is highly glycosylated, with a C-terminal catalytic domain that endohydrolytically cleaves 1,3-β-glycosidic bonds in vitro. Like Cts1, Dse4 localizes specifically to the daughter cell side of the septum in large budded cells that have not yet completed separation. Intriguingly, Dse4 may function as a transglucosidase, rearranging the linkages in glucan chains at high substrate concentrations rather than digesting and eliminating the polymer. Dse4 may contain a GPI linkage that attaches it to the plasma membrane and/or wall matrix (De Groot et al. 2003).
Other hydrolases specific to septum destruction:
Egt2, Dse2, and Scw11 are also likely hydrolases that function primarily in septum degradation (Adams 2004). All contain glucanase domains, but their enzymatic activities have not been directly assessed. Dse2 is required for efficient cell separation (Doolin et al. 2001), and like Dse4 may be GPI linked (De Groot et al. 2003). Scw11 is closely related to two other wall glucanases, Scw4 and Scw10, that are soluble proteins released from the cell wall upon treatment with reducing agent (Cappellaro et al. 1998). Genetic analysis indicates that Scw11 is involved in septum destruction while Scw4 and Scw10 have different functions in wall morphogenesis. Egt2 is a GPI-anchored protein, but it is not clear if its function requires cleavage of this moiety (Hamada et al. 1999). While most glucanases solely involved in septum destruction are expressed only in daughter cells, the EGT2 gene is regulated by Swi5 and thus expressed in both mother and daughter cells (Kovacech et al. 1996). Thus, Egt2 may have a special role in remodeling the septum on the mother cell side.
The role of the SUN family glucanases in cell separation is subtle, and complicated by their additional function in cell wall organization during budding (Mouassite et al. 2000a,b). All contain related domains with strong similarity to known glycosidases. Sun4 was identified by release from cell walls, where it is abundant. The UTH1 gene was initially identified in screens for mutations that increased cell lifespan. Uth1 has been proposed to act directly in both wall organization and autophagic destruction of mitochondria (Camougrand et al. 2003, 2004; Kissova et al. 2004), but this is controversial (Kanki et al. 2009). Recent evidence strongly suggests that Uth1 is primarily a cell wall hydrolase and many phenotypes seen when it is deleted are the result of strengthened cell walls (Ritch et al. 2010).
Gas1 belongs to a group of closely related GPI-linked proteins that contain glucanosyl transferase domains (Gas1–5) (Popolo et al. 2001, 2008; Ragni et al. 2007a,b; Rolli et al. 2011). These proteins transfer glucan chains extruded from β1–3 glucan synthase to other cell wall components, a critical role in wall organization. There are two extracellular pools of Gas1. Prior to cleavage of a GPI linkage from its tail, the protein remains associated with the plasma membrane (PM) and is present broadly. This membrane-anchored fraction functions in wall organization, in concert with Fks1/2, to build a lattice of cross-linked glucan. A significant fraction of Gas1 associates with chitin in the bud neck and close to the primary septum, and this anchorage probably happens after directed cleavage of the GPI anchor from its tail (Rolli et al. 2011). This fraction is involved in both bud neck morphogenesis and cell separation: loss of septum-localized Gas1 causes cell separation defects, suggesting that transglucosidase prepares the structure for the action of exoglucanases and chitinase. Intriguingly, another fraction of Gas1 plays an intracellular role in chromatin silencing (Koch and Pillus 2009), apparently unlinked to its role in cell wall organization.
An integrated model for the process of septum destruction:
Figure 5 illustrates two plausible models for septum destruction. In one view (Figure 4, A–C), chitinase and other cell wall remodeling enzymes are secreted into the periplasmic space beneath the secondary septum in the daughter cell. Additionally (and not shown in this model), some glucan-remodeling enzymes are present at the cytokinesis site because they were deposited before septation, most notably when the mother/bud neck formed early in the budding process. In this model, the glucan network of the secondary septum is then opened and remodeled to allow rapid diffusion of Cts1 to the primary septum. Once this payload of chitinase is delivered, it both binds and digests the lattice of chitin filaments. Simultaneous with or just behind this wave of hydrolytic and glucan-remodeling enzymes, β1–3 glucan (from Rho1-activated Fks1) and a small amount of chitin (from Chs1) are locally synthesized and transferred into a cross-linked mesh. Thus, the extracellular lattice’s integrity would be maintained as the secondary septum is dynamically remodeled.
Resolving the nucleus: spindle breakdown and karyofission
In addition to partitioning of mother and daughter cell cytoplasms, productive cell separation requires proper resolution of the divided nucleus. In budding yeast, as in other ascomycetes, nuclear division occurs without breakdown of the nuclear envelope. Once anaphase chromosome movement and spindle elongation are complete, a thin thread of nucleoplasm and central spindle microtubules encased by nuclear membrane remains (Winey et al. 1995). Final resolution of this closed mitosis requires dissolution of the central mitotic spindle (Maddox et al. 2000). Spindle disassembly happens just prior to cytokinesis and is linked to mitotic exit by a recently defined regulatory network (Woodruff et al. 2010). Additionally, the strand of doubly membrane-encased nucleoplasm that stretches between the mother and daughter cells must be cut. This process is here termed “karyofission,” and it is not well understood. This homotypic membrane fusion event appears to be linked to cytokinesis, but it remains unclear if machinery specifically dedicated to karyofission exists (Latterich et al. 1995; Lippincott and Li 2000). Intriguingly, nuclear division in filamentous fungi takes place in a common cytoplasm without the aid of a cytokinetic contractile apparatus or septation, suggesting the existence of an independent karyofission apparatus.
Mitotic Exit: Regulatory Systems That Control Mother/Daughter Separation
The mechanical events of late cell division and mother/daughter separation happen in rapid succession and precise order, reflecting a remarkable coordination of diverse and sometimes directly opposing processes in time and space. This is achieved by a regulatory system that links the cell cycle’s core cyclin–CDK oscillator to the assembly and function of molecular machines that divide the cell in two. The part of cell division that begins with anaphase chromosome movement and ends after resolution of cell separation is referred to as “mitotic exit.” Studies in yeast are achieving an unprecedented understanding of regulatory systems that control this phase of the cell cycle and drive the ordered events of late cell division and separation.
Unsurprisingly, mitotic exit control pathways are integrated with numerous other aspects of the cell division cycle. For example, checkpoints monitoring spindle assembly, DNA damage, completion of replication, and even bud neck morphogenesis, impinge on the M/G1 transition. While this chapter necessarily focuses on pathways that underlie mitotic exit and cell separation, I urge the reader to avoid drawing artificially rigid boundaries between the major cell cycle regulation systems. Indeed, understanding how these pathways are dynamically integrated is an important frontier in contemporary molecular biology whose exploration will certainly be led by ongoing studies in yeast.
Overview of mitotic exit pathways: APC/C, FEAR, MEN, and RAM
As cells make the transition from M to G1 the effects of mitotic cyclin–CDK are reversed and the regulatory pathways that promote late cell cycle events are activated. A key early insight into the control of mitosis was the determination that progression from metaphase to anaphase involves ubiquitin-mediated proteolytic destruction of mitotic cyclins (reviewed in Morgan 2007; Sullivan and Morgan 2007; Enserink and Kolodner 2010). It is now well established that cyclin destruction in mitosis is a key regulated step in the cell cycle, and failure to degrade mitotic cyclin blocks division at metaphase arrest (reviewed in Pines 2011). Additionally, reversing the effects of mitotic cyclin–CDK requires mitotic phosphorylations to be removed. In budding yeast this is largely carried out by the proline-directed phosphatase Cdc14 (Stegmeier and Amon 2004; Amon 2008).
As diagrammed in Figure 6, mitotic exit control can be roughly broken into four interlaced regulatory systems that are sequentially activated as cells pass from metaphase into G1. The first to act is the anaphase promoting complex/cyclosome (APC/C) (Pines 2011; Song and Rape 2011) in complex with the protein Cdc20, which initiates degradation of the mitotic cyclins as well as the material that keeps replicated metaphase chromosomes from separating. Mitotic cyclin–CDK promotes assembly of the APC/C, which is held inactive until checkpoints that monitor chromosome attachment to the mitotic spindle are satisfied (Peters 2002; Elia et al. 2003; Silva et al. 2011). The second and third mechanisms to act are the Cdc fourteen early anaphase release (FEAR) pathway and the mitotic exit network (MEN) (Dumitrescu and Saunders 2002; de Bettignies and Johnston 2003; Stegmeier and Amon 2004; Sullivan and Morgan 2007; Queralt and Uhlmann 2008a). The MEN is essential for mitotic exit, while the FEAR pathway is not strictly required. These pathways control Cdc14, which dephosphorylates mitotic CDK substrates, and also directly regulate processes important for productive cell division. Activation of the FEAR and MEN systems helps convert the APC/C to a different form, a complex with the protein Cdh1. A fourth pathway acts later by driving localization and activation of the transcription factor Ace2, which turns on expression of separation genes, as well as other mechanisms that promote septum destruction. This system also functions in cell morphogenesis control and is thus referred to as the regulation of Ace2 and morphogenesis (RAM) network (Nelson et al. 2003; Maerz and Seiler 2010).
“Hippo” pathways: the MEN and the RAM networks have similar core organization:
Before discussing the MEN and the RAM networks in detail as separate systems, it is worth noting their similar components and functional organization. As diagrammed in Figure 7, both comprise core elements of ancient pathways that contribute to the control of cell growth, proliferation, and morphogenesis in diverse eukaryotes. These pathways have been labeled Mst/hippo or Ndr/LATS signaling systems, after mammalian and Drosophila kinases involved (Edgar 2006; Harvey and Tapon 2007; Hergovich 2011, 2012; Hergovich and Hemmings 2009; Varelas and Wrana 2012). These networks appear deep in the tree of eukaryotic life, with related pathways present from humans to Protista (Manning et al. 2011; Tavares et al. 2012).
In these pathways, GCK group “Mst/hippo” kinases directly control AGC group “Ndr/LATS” kinases, which form a crucial association with highly conserved co-activating “Mob” subunits. The Ndr/LATS kinases function as the downstream-most components that control cell proliferation, gene expression, and morphogenesis. Flow of information in these systems involves Mst/hippo kinase phosphorylation of the Ndr/LATS kinase at a hydrophobic motif C-terminal to the kinase catalytic core, referred to as the hydrophobic motif (HM) site. This regulatory phosphorylation is broadly conserved in AGC group kinases and is thought to play regulatory roles that include enhancement of kinase activity and recruitment to substrate proteins in vivo (Biondi et al. 2002; Yang et al. 2002a,b; Sarbassov et al. 2005). It is not clear if HM site modification of Ndr/LATS kinases works the same way. The Ndr/LATS kinases have a conserved activation loop regulatory phosphorylation site, which in Ndr kinases gets modified by an intramolecular reaction in vitro and in vivo (Bichsel et al. 2004; Jansen et al. 2006). Ndr/LATS family kinase activations loops also generally contain an insert of unknown functional significance, which is not as highly conserved as the surrounding kinase domain (Bichsel et al. 2004; Maerz and Seiler 2010).
There appear to be two distinct kinds of Mst/hippo signaling pathways that are conserved as separate systems in a wide range of eukaryotes. In one, the downstream-most component is a LATS kinase bound to a Mob1 co-activator; in another it is an Ndr kinase bound to a Mob2 co-activator. As diagrammed in Figure 7, the MEN is a “LATS type” pathway, and the RAM network is an “Ndr type” pathway. As in yeast, metazoan pathways have distinct functions: the “salvador/warts/hippo” system is a LATS-type pathway that restrains cell proliferation and promotes apoptosis (Harvey and Tapon 2007; Pan 2010; Sudol and Harvey 2010; Zhao et al. 2011), whereas the Ndr-type “furry/tricornered/hippo” pathways drive cell division and control cell polarization (Emoto et al. 2006; Seiler et al. 2006; Gao 2007; Cameron and Rao 2010; Cornils et al. 2011a,b; Staley and Irvine 2012). The mechanistic implications of the similarity between Ndr and LATS branches (and more specifically the MEN and the RAM network) are unclear.
The APC/C is an E3 ubiquitin ligase that recognizes specific mitotic proteins and targets them for destruction by the proteasome, precipitating anaphase onset. It is important for mitotic exit and mother/daughter separation, and thus is discussed here in overview; recent reviews provide a more in-depth treatment (Pines 2011; Song and Rape 2011). In metaphase cells, APC/C is present but cannot trigger degradation of target proteins because its function requires association with one of two sequentially acting co-activating subunits, Cdc20 and Cdh1 (Sullivan and Morgan 2007). Cdc20 acts first, associating with APC/C at the end of metaphase to form a complex denoted APCCdc20.
Activated APCCdc20 initiates ubiquitination and degradation of mitotic cyclins, and also completely destroys securin, an inhibitor of the protease known as separase (in yeast, Esp1); separase destroys proteins that link replicated sister chromosomes at metaphase and promotes release of the phosphatase Cdc14 from inhibitory sequestration (Yeong et al. 2000; Sullivan et al. 2001; Yeong et al. 2002; Buonomo et al. 2003; Sullivan et al. 2004b). As discussed further below, APCCdc20-mediated destruction of spindle-associated proteins is important for the mechanical progress of anaphase onset. Formation of APCCdc20 is blocked by a checkpoint signal generated by kinetochores that have not formed a bipolar attachment to the mitotic spindle, linking the initiation of anaphase to completion of spindle assembly (Nezi and Musacchio 2009).
While APCCdc20 begins the destruction of mitotic cyclins, their complete degradation and persistent instability in the subsequent G1 phase requires the APC/C to associate with the Cdh1 subunit. This complex, known at APCCdh1, completes the inactivation of mitotic cyclin–CDK and also promotes destruction of other key mitotic proteins that are not targeted by APCCdc20; in fact, Cdc20 itself is a target of APCCdh1 (Prinz et al. 1998). One mechanistic reason for the transition between APCCdc20 and APCCdh1 is that activity of APCCdc20 is optimally maintained when cyclin–CDK levels are high, while APCCdh1 function is actually antagonized by mitotic CDK phosphorylation (Zachariae et al. 1998; Yeong et al. 2000; Toth et al. 2007; Holt et al. 2008; Pines 2011). Once mitosis is complete, APCCdh1 helps define the G1 state: it turns off mitotic exit by targeting the polo-family protein kinase Cdc5 for destruction in late M or early G1 and destabilizes mitotic cyclins throughout G1 (Wasch and Cross 2002; Visintin et al. 2008).
Cdc14: core regulator of mitotic exit events
Two major molecular processes must occur for cells to pass from M phase into G1. First, the mitotic cyclin–CDK complex has to be inactivated. Second, phosphorylation of key CDK substrates needs to be reversed. APCCdc20 is critical for initiating cyclin destruction, chromosome separation, and anaphase spindle elongation, but it is not sufficient for total destruction of mitotic cyclin or for completion of the M-to-G1 transition (Jaspersen et al. 1998; Morgan 1999; Sullivan and Morgan 2007; Holt et al. 2008). In budding yeast, completion of these tasks requires the phosphatase Cdc14 (Visintin et al. 1998; Jaspersen et al. 1999; Shou et al. 1999; Jaspersen and Morgan 2000; D’Amours and Amon 2004; Stegmeier and Amon 2004; Amon 2008). This CDK-counteracting phosphatase is inactive for most of the cell cycle because it is trapped in the nucleolus (Stegmeier and Amon 2004; Amon 2008). As cells pass from metaphase to anaphase, this entrapment is weakened and Cdc14 eventually floods into the cytoplasm to reverse CDK phosphorylations.
Notably, Cdc14 orthologs are not essential for mitotic exit in many organisms (Mocciaro and Schiebel 2010), including other hemiascomycetes such as Candida albicans (Clemente-Blanco et al. 2006). This likely reflects functional redundancy, with other phosphatases taking on the task of reversing mitotic CDK phosphorylations (Uhlmann et al. 2011). Nevertheless, Cdc14 is highly conserved, and in many organisms it may control specific mitotic subprocesses rather than mitotic exit as a whole. In both C. albicans and the distantly related yeast Schizosaccharomyces pombe, for example, Cdc14 orthologs are important for postcytokinetic processes such as septum destruction (Oliferenko and Balasubramanian 2001; Papadopoulou et al. 2010).
Cdc14 acts in direct opposition to mitotic CDK and controls diverse processes:
Cdc14 is a “dual-specificity” phosphatase (DSP) that strongly prefers dephosphorylate phosphoserines or phosphothreonines that are immediately followed by proline, a motif that corresponds to a minimal CDK phosphorylation site. Thus, the enzyme can reverse the phosphorylation state of mitotic CDK substrates. The structure of a human Cdc14 catalytic domain has provided insight into the enzyme’s evolution and site preference (Gray et al. 2003; Wang et al. 2004). Cdc14’s enzyme core consists of tandem DSP domains, in which the C-terminal “B domain” provides catalytic activity while a pocket formed between the two domains enforces specific site recognition.
Cells lacking Cdc14 function arrest division in telophase, with chromosome masses segregated between mother and daughter cells by a fully elongated mitotic spindle. Such cells can be driven into G1 and made to initiate new budding cycles by overproduction of the CDK inhibitor Sic1, but this occurs without completion of cytokinesis (Luca et al. 2001). Thus, while Cdc14 is not strictly indispensible for cell cycle progress if mitotic CDK can be inactivated, it is essential for mother/daughter separation. Accordingly, Cdc14 localizes to the bud neck during cytokinesis (Bembenek et al. 2005), suggesting that it functions at this site during cell separation and that control of its cytoplasmic localization might contribute to its regulation.
How does Cdc14 control division? It has numerous roles in mitotic exit, as has been extensively reviewed (Stegmeier and Amon 2004; Amon 2008; Clifford et al. 2008; Queralt and Uhlmann 2008a; Mocciaro and Schiebel 2010; Meitinger et al. 2012). One of its important functions is to promote the transition from APCCdc20 to APCCdh1, by dephosphorylating CDK sites on Cdh1 that block its association with the APC/C (Visintin et al. 1997; Zachariae et al. 1998; Jaspersen et al. 1999). However, Cdc14 has many more functions, and its reversal of a large number of CDK phosphorylations clearly influences the progress of late mitotic events. For example, it controls organization of the rDNA repeats, which form a highly distinct chromatin structure whose condensation and resolution is dependent on Cdc14 (D’Amours et al. 2004; Sullivan et al. 2004a). It also suppresses microtubule dynamics at anaphase onset by dephosphorylating the kinetochore component Ask1, thereby stabilizing the mitotic spindle as it begins the mechanical process of chromosome segregation (Li and Elledge 2003; Higuchi and Uhlmann 2005). Additionally, Cdc14 stabilizes spindles by dephosphorylating the microtubule binding protein Fin1, and shuts off the ability of anaphase-separated chromosomes to trigger a checkpoint signal by dephosphorylating the kinetochore component Sli15. Some key Cdc14 targets whose dephosphorylation is important for mother/daughter separation are discussed in the next section, but the current list is certainly incomplete.
How does a single phosphatase drive events that happen at different times in the transition from M to G1? Cdc14 might interact with different substrate targeting or localization subunits that direct its specificity and help define the sequential order of substrate dephosphorylation. Alternatively, Cdc14 appears to dephosphorylate different target proteins with different efficiency, with good Cdc14 substrates dephosphorylated faster and more completely than poor substrates (Bouchoux and Uhlmann 2011). Hence, as CDK activity levels drop, good Cdc14 substrates would become dephosphorylated earlier than poor substrates (Bouchoux and Uhlmann 2011; Uhlmann et al. 2011).
Cdc14 is controlled by regulated sequestration to the nucleolus:
Ectopic activation of Cdc14 can drive mitotic exit, and so an intricate regulatory system ensures that its activity is tightly controlled. During mitotic division Cdc14 localizes to the nucleolus and is essentially inactive until the end of metaphase. This reversible sequestration involves binding to a protein complex present on rDNA repeats in the nucleolus (Petes 1979). Specifically, Cdc14 binds the nucleolar protein Net1, also known as Cfi1, which is part of a system that transcriptionally silences rDNA repeats (Shou et al. 1999; Straight et al. 1999; Visintin et al. 1999). As cells begin anaphase, Cdc14’s anchoring to the nucleolus is weakened due to increased phosphorylation of key components of the anchoring system. The release of Cdc14 proceeds in two distinct stages. First, under the control of the FEAR pathway, Cdc14 is released from the nucleolus into the nucleoplasm, but only a very small amount enters the cytoplasm. Then, coincident with cytokinesis and resolution of the mitotic spindle, sustained release of Cdc14 into the cytoplasm is triggered by the MEN.
The polo-like kinase Cdc5: a regulator of both FEAR and MEN
The highly conserved polo-like kinase Cdc5 is critical for mitotic exit, and thus for the initiation of mother/daughter separation. The amount of Cdc5 present in cells is strongly linked with mitotic progress: it accumulates in S phase, reaches maximum levels in late M phase, and is rapidly degraded in G1. Like other kinases of this family, Cdc5 has an essential C-terminal domain known as the polo box domain (PBD), which mediates association with S-(pS/pT)-(P/X) motifs in which the central residue is phosphorylated (Elia et al. 2003). Indeed, a crystal structure of the PBD from a human polo-like kinase reveals that phosphopeptide binding is likely a highly conserved function of this domain (Cheng et al. 2003), and mutation of PBD residues needed for phosphopeptide recognition inactivates yeast Cdc5 (Song et al. 2000). The PBD recognition motif can correspond to a minimal CDK site, and thus an early view of Cdc5′s intimate relationship with mitotic CDK has been that CDK phosphorylation “primes” proteins for Cdc5 regulation. This is clearly the case for some Cdc5 substrates (Yoshida et al. 2006). Intriguingly, however, recent analyses indicate that phosphopeptide binding by the Cdc5 PBD is not truly essential, and instead is required for only a subset of Cdc5-dependent processes (Chen and Weinreich 2010; Ratsima et al. 2011).
Cdc5 controls both the FEAR pathway and the MEN. As discussed in detail below, in essence Cdc5 turns on both pathways and thereby increases the activity of Cdc14 (Figure 8). This is probably triggered by changes in Cdc5 kinase activity rather than in PBD-mediated docking to substrates (Ratsima et al. 2011). Intriguingly, since Cdc5 is a robust target of APCCdh1 in G1 phase, the kinase effectively plants the seeds of its own destruction by upregulating Cdc14 (Visintin et al. 2008).
As cells transition from metaphase to anaphase, the release of Cdc14 from its inhibited localization in the nucleolus begins. The FEAR pathway (Figure 9) controls early stages of this release. While this pathway is not strictly essential, it helps coordinate the timing of Cdc14-driven anaphase events.
Nucleolar Cdc14 sequestration: NET1 and the RENT complex:
Nucleolar sequestration of Cdc14 requires association with Net1. An N-terminal region of Net1 binds Cdc14 and strongly inhibits its ability to dephosphorylate substrates, so that Net1 both anchors and inactivates the phosphatase (Traverso et al. 2001). In addition to Net1, Cdc14’s sequestration requires a suite of other proteins. Most notably, Fob1 and Spo12 concentrate in the nucleolus and control the Net1–Cdc14 association. Fob1 blocks the progress of replication forks at specific locations in the rDNA repeats that form the bulk of the nucleolar region (Kobayashi and Horiuchi 1996; Mohanty and Bastia 2004). Spo12 is a binding partner of Fob1, and together they help keep the Cdc14 sequestration machinery in the nucleolus (Toyn and Johnston 1993; Stegmeier et al. 2002, 2004; Buonomo et al. 2003; Tomson et al. 2009; Bairwa et al. 2010). Fob1 anchors a Cdc14–Net1 complex in association with Sir2, a silencing protein, to the rDNA repeats. This complex is called the regulator of nucleolar silencing and telophase (RENT), and it suppresses transcription and recombination in the rDNA repeats in addition to localizing Cdc14 (Shou et al. 1999; Straight et al. 1999; Visintin et al. 1999; Huang and Moazed 2003; Kobayashi et al. 2004; Huang et al. 2006).
Triggering the first pulse of Cdc14 release:
Release from Net1 is a key part of Cdc14 activation, and thus the M/G1 transition. How is this interaction controlled? The phosphorylation state of Net1 and other RENT complex components is critically important. As diagrammed in Figure 9, mitotic CDK can phosphorylate Net1 at six key sites. Cdc5 is also required for Cdc14’s early anaphase release, but this role remains mysterious (Sullivan et al. 2001; Sullivan and Uhlmann 2003; Visintin et al. 2003; Azzam et al. 2004; Rahal and Amon 2008; Manzoni et al. 2010). CDK phosphorylation of Spo12 helps potentiate disengagement of the Cdc14–Net1 interaction (Tomson et al. 2009).
Although CDK phosphorylation of Net1 and other RENT components causes release of Cdc14 (Azzam et al. 2004), this is counteracted by the protein phosphatase PP2A. This phosphatase is a multiprotein complex with different regulatory subunits that mediate substrate association; the regulatory subunit for Cdc14 release is Cdc55 (Queralt et al. 2006). Thus, a dynamic balance of kinase and phosphatase activities controls Cdc14 release from the nucleolus. PP2ACdc55 dominates this balance prior to anaphase, keeping Net1 dephosphorylated (Wang and Burke 1997; Queralt et al. 2006; Wang and Ng 2006; Yellman and Burke 2006). Suppression of PP2ACdc55 activity is thus a key trigger of mitotic exit.
When the FEAR pathway is activated, PP2ACdc55 activity toward Net1 is inhibited, and the balance of activities on the RENT complex tips toward CDK-mediated phosphorylation (Queralt and Uhlmann 2008a; Rossio and Yoshida 2011). The budding yeast separase ortholog Esp1 plays a critical role in this trigger, distinct from its essential role as the protease that dissolves the physical connection between replicated metaphase chromosomes. After APCCdc20 degrades an Esp1 inhibitor, securin, Esp1 triggers release of Cdc14 by promoting CDK phosphorylation of Net1 (Sullivan et al. 2001; Sullivan and Uhlmann 2003; Azzam et al. 2004). Intriguingly, this function is independent of Esp1 protease activity (Sullivan and Uhlmann 2003). The Esp1-associated protein Slk19 is necessary for optimal Cdc14 release: while Slk19 is actually cleaved by the protease, this does not seem to be functionally important for the FEAR pathway (Stegmeier et al. 2002).
The proteins Zds1 and Zds2, which associate with and regulate PP2ACdc55 (Yasutis et al. 2010; Wicky et al. 2011) help tip the kinase–phosphatase balance in favor of Cdk1 (Queralt and Uhlmann 2008b). They probably accomplish this by binding and retaining PP2ACdc55 in the cytoplasm, thereby lowering its nuclear concentration (Rossio and Yoshida 2011), but the precise mechanism remains unclear. Analysis of their localization in isolated nucleoli suggests that Zds proteins may regulate a nucleolar pool of the phosphatase (Calabria et al. 2012).
FEAR pathway is an oscillator:
FEAR-mediated release of Cdc14 can cycle with a defined periodicity even when mitotic cyclin–CDK activity levels are locked at discrete levels using nondegradable cyclins (Lu and Cross 2010; Manzoni et al. 2010). These pulses of Cdc14 release are entrained by Cdk1 to produce a single early burst of Cdc14 release. Normally, when mitotic CDK is only partially inactivated the FEAR system quenches itself by reducing the amount of CDK-phosphorylated Net1 in the cell, leading to resequestration in the nucleolus. Setting CDK activity at different levels reveals distinct oscillatory mechanisms. When CDK activity is high, cells stay arrested in mitosis and Cdk1 and Cdc5 drive Cdc14 release. Consistent with the negative feedback loop shown in Figure 8, Cdc5 is then degraded allowing resequestration of Cdc14 (Visintin et al. 2008). Resynthesis and reactivation of Cdc5 then reset the cycle. At lower CDK levels, Cdc5 likely does not reset the cycle, and the MEN (described below) plays an essential role in driving repeated bursts of Cdc14 release.
Consistent with its nonessential nature, FEAR-driven Cdc14 release cannot drive full exit from mitosis: it is incomplete and results in very little accumulation of the phosphatase in the cytoplasm. Complete mitotic exit requires the MEN, which promotes robust relocalization of Cdc14, leading to full dephosphorylation of cytoplasmic CDK substrates and numerous downstream events (Charles et al. 1998; Bosl and Li 2005; Toth et al. 2007). The core MEN pathway consists of three functional modules: the G protein Tem1, the protein kinase Cdc15, and the protein kinases Dbf2 and Dbf20 in complex with their co-activating protein Mob1 (Figure 10) (Visintin et al. 1998; Shou et al. 1999; Tinker-Kulberg and Morgan 1999; Bardin et al. 2000; Jaspersen and Morgan 2000; Pereira and Schiebel 2001; Visintin and Amon 2001).
MEN components concentrate dramatically at the spindle pole body (SPB), the budding yeast centrosome equivalent, with a strong initial asymmetric bias toward the SPB that enters the daughter cell cytoplasm (Cenamor et al. 1999; Menssen et al. 2001; Visintin and Amon 2001; Yoshida et al. 2002). This localization of Tem1, Cdc15, and Mob1–Dbf2/20 to the SPB is critical for the system’s activation in response to proper spindle position (Figure 11) and is largely accomplished through their association with the centriolin-related SPB protein Nud1 (Luca et al. 2001; Yoshida et al. 2002; Rock and Amon 2011). As discussed further in part 3 of this chapter, the MEN proteins also concentrate at the cytokinesis site just prior to actomyosin ring contraction and septation and probably function directly in those processes.
Core MEN components: the Dbf2–Mob1 module:
The paralogous protein LATS-family kinases Dbf2 and Dbf20 are the MEN’s downstream-most “LATS kinase” components (Figure 7) (Toyn et al. 1991; Toyn and Johnston 1994; Jaspersen et al. 1998; Lee et al. 2001a; Luca et al. 2001). Most analysis has focused on Dbf2, which appears to be the more important paralog (Toyn et al. 1991), using temperature-sensitive dbf2 alleles in dbf20Δ cells. Only Dbf2 is discussed here, though Dbf20 is assumed to have largely similar properties. Dbf2, like all other characterized LATS kinases, binds to a Mob family co-activating subunit, in this case Mob1 (Figure 10) (Komarnitsky et al. 1998; Luca and Winey 1998; Luca et al. 2001). Mob1 is essential for mitotic exit. Three-dimensional structures have been obtained for Mob1 orthologs from budding yeast, Xenopus laevis, and humans (Stavridi et al. 2003; Ponchon et al. 2004; Mrkobrada et al. 2006). The Mob1 core structure includes an electronegative surface that may create a binding region for the conserved positively charged Mob-binding peptide found N-terminal to the kinase domain in Dbf2 (Stavridi et al. 2003; Ponchon et al. 2004; Mrkobrada et al. 2006). In vitro studies showed that Dbf2 is basophilic, preferring to phosphorylate serine or threonine residues that are three residues C-terminal to arginine (R-X-X-[S/T]) (Mah et al. 2005). This short consensus motif, which is shared by other basophilic kinases, has been borne out for several in vivo substrates (Mohl et al. 2009; Oh et al. 2012).
Core MEN components: the Cdc15–Tem1 module:
Cdc15, which functions upstream of Dbf2/Mob1 in the MEN (Figure 10), is a ∼117-kDa protein with an Mst/hippo-related kinase domain near its N terminus. While Cdc15’s enzymatic activity does not seem to fluctuate over the cell cycle, its function is blocked prior to the metaphase/anaphase transition by CDK phosphorylation (Jaspersen and Morgan 2000). Cdc15’s C-terminal domain is crucial for its regulation and contains an important short segment next to the kinase domain that binds the Ras-related GTPase protein Tem1 (Asakawa et al. 2001; Bardin et al. 2003). Like other small GTPases, Tem1 probably activates its effectors only when loaded with GTP, but this has not been directly established. As discussed further below, Tem1 provides a key regulatory linkage that connects Cdc15’s ability to activate Mob1–Dbf2 to the status of mitotic processes like spindle orientation (Shirayama et al. 1994; Jaspersen et al. 1998; Jaspersen and Morgan 2000; Lee et al. 2001a,b; Menssen et al. 2001; Rock and Amon 2011). While specific amino acids in Cdc15’s Tem1-binding region are important for the association, the biochemical basis of the interaction is incompletely understood. In addition to this short Tem1-interacting segment, a C-terminal portion of Cdc15 appears to function as an inhibitory region. Finally, Cdc15’s central region mediates self-association, but it is not clear if this is direct or requires higher-order complex formation (Bardin et al. 2003).
Core MEN components: Nud1, a platform for SPB localization:
The SPB protein Nud1 is an essential MEN component (Figure 10). Loss of its function causes arrest at the M/G1 transition, and it is required for cytokinesis even when this block to G1 entry is overridden by overexpression of a CDK inhibitor (Gruneberg et al. 2000; Luca et al. 2001; Yoshida et al. 2002). It associates with the cytoplasmic face of the SPB, which is embedded in the nuclear envelope (Winey and Bloom 2012). Nud1 recruits other components of the MEN to the SPB, and this localization is critical for MEN activation (Jaspersen et al. 1998; Cenamor et al. 1999; Shou et al. 1999; Visintin and Amon 2001; Molk et al. 2004; Rock and Amon 2011; Valerio-Santiago and Monje-Casas 2011). In fact, Cdc15 regulation of the Mob1–Dbf2 complex is probably mediated by co-association on Nud1 at the SPB. Nud1 contains an ∼150-amino-acid region related to the metazoan protein centriolin, a cytokinesis factor that associates with centrioles and the site of cytokinetic abscission (Gromley et al. 2003). This conserved domain probably mediates association of membrane trafficking machinery (Gromley et al. 2005), but the importance of this association in yeast is unknown. Budding yeast Nud1 connects the γ-tubulin binding protein Spc72 to the SPB outer plaque and is thus an important anchor for cytoplasmic microtubules (Pereira et al. 1999; Gruneberg et al. 2000). It is not known if Nud1’s role in microtubule organization is related to its mitotic exit function.
MEN’s functional organization:
Controlling MEN: the core pathway:
As diagrammed in Figure 10, Cdc15’s direct phosphorylation of Dbf2’s C-terminal HM site is a key part of MEN activation, similar to homologous metazoan pathways (Hergovich and Hemmings 2009; Emoto 2011). In vitro studies show that Cdc15 phosphorylation of this HM site dramatically increases Dbf2 enzymatic activity (Mah et al. 2001, 2005). Furthermore, Mob1 greatly increases the efficiency of Cdc15’s in vitro phosphorylation of Dbf2. Corroborating these in vitro studies, genetic analysis indicates that Cdc15’s phosphorylation of Dbf2 is critical for its in vivo function (Mah et al. 2001). Mutation of Cdc15 phosphorylation sites on Dbf2 to alanine significantly abrogates MEN function, while mutation to glutamic acid (to mimic phosphorylation) renders Cdc15 dispensable.
Controlling MEN: modulating Tem1’s nucleotide state and SPB localization:
A possible logic diagram for MEN organization is diagrammed in Figure 11, which summarizes models and results from numerous investigators. As noted above, the core pathway comprising the Tem1–Cdc15–Dbf2/Mob1 cascade is activated by recruitment to the SPB by Nud1. This recruitment probably promotes Cdc15’s phosphorylation of Dbf2’s HM site, although the exact mechanism for this is not known. Additionally, Cdc15 is required to bring Dbf2/Mob1 to the SPB, suggesting that the upstream kinase controls the ability of Nud1 to bind the downstream kinase complex. The likely negative regulation of Cdc15 and/or Dbf2/Mob1 by mitotic CDK (Jaspersen and Morgan 2000) may be relieved by dephosphorylation of these proteins by Cdc14, after its release by the FEAR pathway prior to MEN activation (Konig et al. 2010).
Tem1 appears to be the first MEN component loaded onto Nud1 at the SBP, and photobleaching analysis shows that it is initially highly dynamic in early anaphase (Cenamor et al. 1999; Bardin et al. 2000; Menssen et al. 2001; Visintin and Amon 2001; Molk et al. 2004; Rock and Amon 2011; Valerio-Santiago and Monje-Casas 2011). Tethering Tem1 at the SPB by fusing it to another outer plaque component increases recruitment of Cdc15 (Valerio-Santiago and Monje-Casas 2011). However, this does not cause a corresponding increase in localization of the Dbf2/Mob1 complex, further suggesting that Cdc15 must be activated and modify components of the local protein environment before Dbf2/Mob1 can stably associate.
Tem1’s localization and nucleotide state are important in MEN control. The proteins Bub2 and Bfa1 were proposed to form a bipartite GAP that converts Tem1 to an inactive GDP-bound form (Bardin et al. 2000; Krishnan et al. 2000; Pereira et al. 2000; Wang et al. 2000; Lee et al. 2001b). However, other studies suggest that Tem1 by itself efficiently hydrolyzes GTP and exchanges GDP, and Bfa1 may in fact act as an inhibitor of GDP exchange (Geymonat et al. 2002). Regardless, assembly and function of the Bub2–Bfa1 complex is clearly a key aspect of MEN regulation (Fraschini et al. 2006; Geymonat et al. 2009). Like other MEN components, Bub2–Bfa1 localizes to the SPB, with a notable bias to the pole that migrates into the daughter cell cytoplasm; in fact, the complex strongly stabilizes SBP localization of Tem1 (Bardin et al. 2000; Pereira et al. 2000; Monje-Casas and Amon 2009), but not its initial recruitment (Valerio-Santiago and Monje-Casas 2011).
At least two distinct models have been proposed for Bub2–Bfa1’s control of mitotic exit. In one view, the Bub2–Bfa1 complex diffuses from the SPB to inhibit cytoplasmic pools of Tem1–Cdc15, with SPBs located in the mother cell generating a diffusible MEN-suppressing signal by producing active Bub2–Bfa1 (Fraschini et al. 2006; Caydasi and Pereira 2009). In a different view, Tem1 must turn on the MEN at the SPB, and rapid flux of Bub2–Bfa1 quickly removes the GTPase from this location. The fact that tethering Tem1 to the SPB triggers mitotic exit supports this hypothesis, which essentially posits that rapid loss of the Bub2–Bfa1 complex from mother-localized SPBs sweeps away Tem1 (Valerio-Santiago and Monje-Casas 2011).
Cdc5 inactivates Bub2–Bfa1, probably by directly phosphorylating Bfa1, and thereby positively regulates the MEN (Figure 11) (Hu et al. 2001; Hu and Elledge 2002; Pereira et al. 2002; Geymonat et al. 2003; Park et al. 2003). Separately, Cdc5 may directly promote recruitment of the Cdc15–Tem1 module to Nud1, although by unclear means (Luca et al. 2001; Yoshida et al. 2002; Park et al. 2008). This forms a feed-forward loop (FFL) in which Cdc5 both turns off inhibition of the Cdc15–Tem1 module and separately promotes this module’s Nud1 association and activation of Dbf2.
How is Tem1 activated? The Lte1 protein turns on MEN signaling and resembles a GEF (Bardin et al. 2000; Pereira et al. 2000; Jensen et al. 2002). Lte1 localizes to the bud cortex (Figure 12), leading to early models in which SPB migration into the daughter cell directly promotes GTP loading on Tem1. Recent studies suggest that Lte1 does not serve this purpose; rather, it appears to block the function of the protein kinase Kin4, which as discussed further below is an inhibitor of the MEN (Geymonat et al. 2009; Chan and Amon 2010; Bertazzi et al. 2011). Overall, the mechanism by which Lte1 activates the MEN remains uncertain.
Controlling MEN: spindle orientation and the “zone model”:
The MEN is the target of a signaling mechanism that blocks M/G1 progress when mitotic spindle is not properly oriented to segregate chromosomes to mother and daughter cells (Figure 12). This system, termed the “spindle orientation checkpoint,” ensures that mitotic exit does not occur until one of the two SPBs enters the daughter cell cytoplasm (Adames and Cooper 2000; Bardin et al. 2000; Bloecher et al. 2000; Pereira et al. 2000 and reviewed in Lew and Burke 2003; Fraschini et al. 2008; Burke 2009). Asymmetric regulation of Bub2–Bfa1 by the MEN-inactivating kinase Kin4 is the basis of a zone model for this checkpoint (Figure 12). As diagrammed in Figure 11, Kin4 phosphorylates Bfa1, and this prevents Cdc5 from turning off Bub2–Bfa1 (Hu et al. 2001; Hu and Elledge 2002; Maekawa et al. 2007). Kin4 concentrates at the mother cell cortex, the bud neck, and SPBs located in the mother cell cytoplasm (D’Aquino et al. 2005; Pereira and Schiebel 2005). Thus, it maintains Bub2–Bfa1 in a highly active state at SPBs present in the mother cell cytoplasm and appears to increase the rate of Bub2–Bfa1 turnover there (Caydasi and Pereira 2009; Chan and Amon 2010). In contrast, Kin4 disappears from SPBs that have entered the daughter cell, presumably allowing activation of SPB-associated Tem1 and downstream MEN components. This normally occurs only when the spindle is correctly aligned, when one of the SPBs moves into the daughter cell cytoplasm and enters a region of low Kin4 concentration (Figure 12).
If this zone model is correct, how do relatively small yeast cells set up and maintain such sharply defined subcellular regulatory regions? Kin4’s localization and activity are clearly important, and the mechanism that spatially restricts its function remains a subject of active investigation. The protein phosphatase PP2A, in complex with its regulatory subunit Rts1, is required both for restriction of Kin4’s localization to the mother cell cortex and for its recruitment to SPBs (Chan and Amon 2010). Additionally, the bud neck localized protein kinase Elm1 activates Kin4 by directly phosphorylating a critical activation loop site. Lte1, which is localized in the daughter cell, blocks Kin4’s ability to phosphorylate Bfa1 (Bertazzi et al. 2011); this requires the p21-activated kinase Cla4 (Seshan et al. 2002; Chiroli et al. 2003). Intriguingly, cells lacking cytoplasmic SPB-anchored microtubules exhibit a partial checkpoint defect, and the interaction of these microtubules with the bud neck has been proposed to activate the spindle orientation checkpoint (Adames et al. 2001; Moore et al. 2009).
MEN may function in metaphase spindle positioning and mRNA stabilization:
The prevailing view of MEN function is that its activity is closely coupled with the final stages of mitosis, leading into the initiation and completion of cytokinesis. Intriguingly, however, Mob1–Dbf2 and the MEN may have important functions significantly earlier. These include a possible role in orientation of the mitotic spindle in metaphase (Hotz et al. 2012) and stabilization of the early mitotic transcripts CLB2 and SWI5 (Trcek et al. 2011). In the latter case, a fraction of Dbf2 and Dbf20 associate with the mRNAs and suppress their degradation. Surprisingly, this appears to be independent of their kinase activity. For Dbf2, its loading onto mRNAs depends entirely on the promoter, not mRNA sequence context. Anaphase MEN activation appears to antagonize Dbf2/20’s mRNA stabilization role, and this may help close the window of the cell cycle in which the SWI5 and CLB2 mRNAs can produce protein.
Mob1–Dbf2/Cdc14 release positive feedback loop:
In vitro and in vivo studies indicate that Mob1–Dbf2 directly drives cytoplasmic accumulation of Cdc14 by phosphorylating it at sites near its C-terminal nuclear localization sequence (NLS) (Mohl et al. 2009). This inhibits the NLS’s function, shifting Cdc14’s distribution from the nucleus to the cytoplasm and thus allowing it to remove CDK phosphorylations on cytoplasmic proteins. The Mob1–Dbf2 complex appears to enter the nucleus prior to final release of Cdc14 to the cytoplasm (Stoepel et al. 2005), suggesting that it regulates Cdc14 in the nucleus. Additionally, Mob1–Dbf2 must phosphorylate at least one other nuclear protein to promote dissociation of Net1 and Cdc14, and this may be Net1 itself (Mah et al. 2005; Mohl et al. 2009). Intriguingly, the Net1-related protein Tof2 also sequesters a fraction of Cdc14 to the nucleolus, but this is refractory to the FEAR and is only released upon full MEN activation (Waples et al. 2009).
As cells complete synthesis of the primary and secondary septa, the RAM network turns on processes needed for septum destruction. This pathway is regulated by mitotic exit but is distinct from FEAR and MEN (and less extensively studied). By activating the transcription factor Ace2, it drives a sharp increase in transcription of mother/daughter separation genes. In addition to a mother/daughter separation function during the M-to-G1 transition, the RAM network functions during other parts of the cell cycle to promote sustained polarized growth and localized cell wall expansion. The RAM name (regulation of Ace2 and morphogenesis) reflects these dual roles (Nelson et al. 2003).
The RAM network thus far comprises six proteins that function in at least two distinct modules (Figures 13 and 14) (Racki et al. 2000; Bidlingmaier et al. 2001; Colman-Lerner et al. 2001; Weiss et al. 2002; Nelson et al. 2003). Cells lacking any of these six components have indistinguishable cell separation defects, and loss of multiple components does not further exacerbate this phenotype (Bidlingmaier et al. 2001; Nelson et al. 2003). The downstream-most module of the system is the Ndr family protein kinase Cbk1, which functions in complex with its co-activator Mob2 (Weiss et al. 2002). Cbk1 is regulated by the Mst/hippo protein kinase Kic1, which associates with the activating subunit Hym1 (Nelson et al. 2003; Brace et al. 2011). The very large protein Tao3 is critical for this regulatory interaction, but its role is not well defined (Du and Novick 2002; Jorgensen et al. 2002; Nelson et al. 2003). Additionally, the leucine-rich protein Sog2 is important for RAM network function, and most likely directly interacts with the Kic1–Hym1 complex (Nelson et al. 2003). All components except Sog2 are conserved, with putative paralogs present from fungi to metazoans (reviewed in Hergovich et al. 2006; Gao 2007; Maerz and Seiler 2010; Emoto 2011; Hergovich 2011; and see also Emoto et al. 2004; Chiba et al. 2009; Fang and Adler 2010; Cornils et al. 2011a,b).
Current knowledge about the RAM network’s control of septum destruction is discussed in part 3, in particular its control of the transcription factor Ace2. However, for discussions immediately below it is important to note that Ace2 is a direct in vivo phosphorylation target of Cbk1 (Mazanka et al. 2008; Mazanka and Weiss 2010). The RAM network’s role in defining cell architecture is less well understood and is largely beyond the scope of this chapter.
RAM components: the Mob2–Cbk1 module:
Cbk1, which stably associates with the Mob2 protein, is the sole Ndr family kinase in budding yeast (Racki et al. 2000; Bidlingmaier et al. 2001; Colman-Lerner et al. 2001) (Figure 13). Both Cbk1 and Mob2 are present throughout the cell cycle and appear to bind each other constitutively (Weiss et al. 2002); they have been identified as a complex in large-scale protein interaction analyses (Reguly et al. 2006; Stark et al. 2006, 2011). Cbk1’s activity fluctuates modestly over the cell cycle, with peaks occurring during bud formation and cell separation (Weiss et al. 2002), but the importance of this change is not clear. Indeed, alleles that drastically compromise Cbk1’s kinase activity have modest effects on the in vivo function of the Mob2–Cbk1 complex, suggesting that control of its specific activity is a relatively unimportant regulatory input (Racki et al. 2000; Colman-Lerner et al. 2001; Weiss et al. 2002; Jansen et al. 2006; Bourens et al. 2009).
Cbk1 has a distinctive phosphorylation consensus site preference that has proven extremely helpful in identifying its in vivo phosphorylation targets. Unbiased in vitro analysis of its specificity showed that Cbk1 strongly favors phosphorylation of serine or threonine in the motif H-X-R-R-X-(S/T), where X represents any amino acid (Mazanka et al. 2008). This sequence closely matches sites phosphorylated by metazoan LATS/warts family protein kinases in the yorkie/YAP/TAZ transcriptional co-activators (Zhao et al. 2007; Zhang et al. 2008; Wang et al. 2009), indicating that this consensus motif preference is conserved in Ndr/LATS family kinases. Cbk1’s preference for histidine at position −5 is very strong in vivo and in vitro; lysines can be substituted for arginines at positions −2 or −3, but, as for Dbf2, they cannot be eliminated (Mah et al. 2005; Mazanka et al. 2008).
Mob2, like the MEN component Mob1, is a member of the Mob-family of proteins that appear to function in complexes with Ndr/LATS protein kinases in all eukaryotes (Hergovich 2011). As with Mob1 and Dbf2/20, Mob2 association with Cbk1 is critical for normal protein kinase function (Weiss et al. 2002), suggesting that it acts as an essential co-activator. Additionally, binding of Mob2 may stabilize Cbk1 in vivo (Weiss et al. 2002; Nelson et al. 2003). The three-dimensional structure of Mob2 has not been determined, but its similarity to Mob1 suggests that it adopts a broadly similar conformation in its core (Luca and Winey 1998; Stavridi et al. 2003; Mrkobrada et al. 2006). A notable difference is that Mob1 contains a motif that binds zinc, while Mob2 does not. Unlike Mob1, there is no indication that Mob2 dimerizes, but in vitro and in vivo analyses suggest that Mob1 forms heterodimers with Mob2, potentially linking the Mob1–Dbf2 module with Mob2–Cbk1 (Mrkobrada et al. 2006).
The localization of Cbk1 and Mob2 is consistent with their dual roles in control of cell morphogenesis and mother/daughter separation gene transcription. They concentrate prominently at the mother/bud neck of large budded cells in telophase, prior to actomyosin ring contraction, and localize somewhat more faintly to the cortex of the growing daughter cell and the tip of the mating projection formed by cells responding to mating pheromone (Racki et al. 2000; Bidlingmaier et al. 2001; Colman-Lerner et al. 2001; Weiss et al. 2002). The proteins depend on one another for robust recruitment to cortical sites. Intriguingly, their cortical localization does not require other RAM network components (Weiss et al. 2002; Nelson et al. 2003). The proteins also accumulate in the daughter cell nucleus as cells pass from M to G1 in an Ace2-dependent manner, suggesting that they enter the nucleus as a ternary complex with the transcription factor (Colman-Lerner et al. 2001; Weiss et al. 2002).
RAM components: the Kic1–Hym1 module:
As noted above, Kic1 is an Mst/hippo-related kinase with an N-terminal catalytic domain highly similar to Cdc15’s (Sullivan et al. 1998). One distinctive feature is a conserved three amino acid motif C-terminal to the Kic1 kinase domain termed “site D” (Zeqiraj et al. 2009a; Filippi et al. 2011); in Kic1 the sequence is WDF. Kic1 was initially identified as a two-hybrid interactor with the calmodulin-related protein Cdc31 (Sullivan et al. 1998), which is essential for SPB duplication and aspects of mRNA export from the nucleus (Baum et al. 1986; Jani et al. 2009). While this interaction may be functionally important (Sullivan et al. 1998; Ivanovska and Rose 2001), its in vivo significance remains unclear.
Kic1 associates with the helical repeat protein Hym1, which is closely related to metazoan MO25 proteins (Sullivan et al. 1998; Bidlingmaier et al. 2001; Jorgensen et al. 2002; Nelson et al. 2003; Reguly et al. 2006; Stark et al. 2006, 2011) (Figure 13). Like the Cbk1/Mob2 module, both Kic1 and Hym1 localize to sites of cell growth and wall remodeling, including the daughter bud cortex, mating projection tip, and bud neck, with Hym1 recruitment to these sites dependent on Kic1 (Bidlingmaier et al. 2001; Nelson et al. 2003). Neither protein has been detected in nuclei. Overexpression studies suggest that Hym1 might have a role outside of the RAM network in G1 progression, possibly indicating that the protein interacts with kinases other than Kic1 (Bogomolnaya et al. 2004, 2006).
Hym1 is highly conserved, and analysis of mammalian MO25 reveals biochemical mechanisms that are probably critical for RAM network function. Mammalian MO25 binds the inactive Kic1-related pseudokinase STRAD with high affinity and associates more weakly with at least five different catalytically active mammalian Mst/hippo-related kinases (Hawley et al. 2003; Boudeau et al. 2004; Milburn et al. 2004; Zeqiraj et al. 2009a; Zeqiraj et al. 2009b; Filippi et al. 2011). The three-dimensional structure of mammalian MO25 has been solved alone and in complex with STRAD (Milburn et al. 2004; Zeqiraj et al. 2009a). It is broadly similar to armadillo-repeat proteins, with three-helix elements stacked atop one another to form a slightly concave solenoid. The STRAD:MO25 complex provides a plausible model for MO25’s binding and activation of Mst/hippo kinases. One set of interactions between MO25’s concave surface and the kinase N-terminal lobe helps position the kinase activation loop; this resembles the mechanism of CDK activation by cyclin binding (Jeffrey et al. 1995; Zeqiraj et al. 2009a,b; Filippi et al. 2011). Additionally, the conserved site D motif on the Mst/hippo kinases associates with an N-terminal pocket on MO25 (Zeqiraj et al. 2009a). Overall, it is possible that Hym1–Kic1 interaction is regulated in vivo to control the RAM network (and, more broadly, other Mst/hippo pathways), but this has not yet been addressed.
RAM components: Tao3 and Sog2:
Tao3 (also known as Pag1) is a large (∼350 kDa) protein essential for RAM network function (Du and Novick 2002; Jorgensen et al. 2002; Nelson et al. 2003). Tao3 physically interacts with both Kic1 and Cbk1, and thus may facilitate communication between these two modules (Figure 13) (Nelson et al. 2003; Reguly et al. 2006; Stark et al. 2006, 2011). Like other RAM network components, Tao3 localizes to the bud cortex as well as the mother/bud neck in large budded cells, but it has not been detected in nuclei. Its localization does not require other RAM network components, and vice versa (Nelson et al. 2003). Thus, there are likely multiple independent mechanisms for recruitment of RAM network proteins to the cell cortex.
Tao3 is conserved from yeast to metazoans, and its likely orthologs function along with Ndr family kinases in control of cell proliferation and morphogenesis. These are broadly defined as the “furry” family of proteins, after a Drosophila melanogaster ortholog involved in the organization of cellular extensions such as bristles, wing hairs, and aristal laterals (Cong et al. 2001; He et al. 2005; Fang et al. 2010). Like the Ndr kinases, furry-related proteins are also involved in neuron morphogenesis (Zallen et al. 1999; Emoto et al. 2004; Gallegos and Bargmann 2004; Jia and Emmons 2006; Gao 2007). Phylogenetic analysis of these putative Tao3 orthologs indicates that the broad family has five sequence blocks that exhibit significant conservation from metazoans to yeast, and that a significant portion of the protein has HEAT repeats predicted to adopt an extended β-catenin–like structure (Gallegos and Bargmann 2004; Chiba et al. 2009; Goto et al. 2010)
Sog2 is a leucine-rich protein that interacts with Kic1 and most likely with Tao3 and localizes to the cell cortex (Nelson et al. 2003). The protein is otherwise not well understood, and is not as well conserved as other components of the RAM network.
Functional organization of the RAM network:
How does information flow within the RAM network to allow its control of cell separation and morphogenesis? While some important aspects of this system are known (Figure 13), far less is understood about it than either the MEN or the FEAR pathway. Since loss of any of the system’s components produces indistinguishable nonadditive defects in cell separation and morphogenesis, it is not possible to order the pathway through epistasis analysis of known phenotypes. In vitro protein kinase assays using immunoprecipitated Cbk1 suggest that it is the most downstream component: all other RAM network proteins of the system are essential for full in vitro enzymatic activity of the Cbk1 (Weiss et al. 2002; Nelson et al. 2003). Most effort toward understanding information flow within the RAM network and its coordination with mitotic exit has focused on regulation of the Mob2–Cbk1 module.
Regulatory phosphorylation of Cbk1’s hydrophobic motif site and activation loop:
While the protein kinase activity of immunoprecipitated Mob2–Cbk1 may be a useful proxy for RAM network function, there are several caveats to this analysis. Until recently, most such kinase assays used histone H1 as the in vitro test substrate, which is probably not ideal (Jansen et al. 2006; Mazanka et al. 2008). More direct assessment of Cbk1’s HM site and activation loop phosphorylation using phosphospecific antibodies has provided additional insights (Jansen et al. 2006; Brace et al. 2011). Both sites are phosphorylated in vivo (Jansen et al. 2006), an observation corroborated by large-scale mass spectrometry (Bodenmiller and Aebersold 2011). Phosphorylation of the HM site is dynamic over the cell cycle, peaking during cell separation and polarized growth of the daughter bud (Jansen et al. 2009). At peak times, only a small fraction of the kinase, between 3 and 5%, carries this phosphorylation (Brace et al. 2011). Phosphorylation of the activation loop site occurs through an intramolecular autophosphorylation reaction (Jansen et al. 2006), as with mammalian Ndr kinases (Tamaskovic et al. 2003), and remains relatively constant throughout the cell cycle.
Genetic analysis indicates that phosphorylation of both the HM site and activation loop site is important for Cbk1 function, although to different degrees (Jansen et al. 2006; Bourens et al. 2009; Panozzo et al. 2010). Neither modification is required for recruitment of Cbk1 to the daughter cell cortex and bud neck. Substitution of alanine at the activation loop phosphoacceptor site severely compromises Cbk1’s catalytic activity, making it nearly undetectable; yet this allele has a notably mild phenotype, exhibiting moderate defects in polarized growth, mating, and cell separation (Jansen et al. 2006; Bourens et al. 2009). In contrast, replacing the phosphoacceptor threonine at Cbk1’s HM site position completely abolishes its in vivo function, but not the in vitro kinase activity of Mob2–Cbk1 immunoprecipitated from asynchronous cells. It is pertinent here that HM site phosphorylation does increase catalytic activity of related kinases (Millward et al. 1998; Stegert et al. 2005), and substitution with glutamic acid at this site significantly increases Cbk1 kinase activity (Brace et al. 2011). Thus, since only a small fraction of Cbk1 is phosphorylated at the HM site, it is likely that analysis of immunoprecipitated Mob2–Cbk1 cannot accurately measure the degree of enzyme activation by this modification. Overall, the precise function of this modification remains unknown.
The RAM network and Ace2 in Cbk1 phosphoregulation:
Cbk1’s HM site phosphorylation requires all other RAM network components, again consistent with the Mob2–Cbk1 complex acting as the downstream-most part of the system (Jansen et al. 2006) (Figure 14). However, phosphorylation of Cbk1’s HM site is also greatly reduced in cells that lack Ace2, which as discussed in part 3 is one of the kinase’s in vivo substrates. This suggests a possible feedback loop in which the kinase’s substrate synergistically enhances the activation of its upstream regulator. The mechanism by which Ace2 enhances Cbk1 HM site phosphorylation is unknown; it might promote binding of Mob2–Cbk1 to Kic1–Hym1 or it might shield a fraction of Mob2–Cbk1 from a phosphatase that normally reverses HM site phosphorylation, perhaps by partitioning Mob2–Cbk1 to the nucleus. Finally, it is possible that one of Ace2’s transcriptional targets, rather than Ace2 itself, is responsible for modulating phosphorylation of the HM site. Regardless, it is unlikely that Cbk1 HM site phosphorylation fully depends on Ace2, as the phosphorylation site is essential for Cbk1’s role in polarized growth, whereas Ace2 is not (Weiss et al. 2002; Voth et al. 2005; Jansen et al. 2006).
Autophosphorylation of Cbk1’s activation loop site exhibits a more nuanced set of genetic dependencies (Figure 14) (Jansen et al. 2006). As expected, Mob2 is required. Tao3 is also required, suggesting that it either helps Cbk1 adopt a conformation suitable for autophosphorylation or shields it from a phosphatase activity. Intriguingly, this phosphorylation is reduced (though not absent) in cells lacking Hym1, but is unaffected in cells lacking Kic1 and Sog2.
It is likely that the Kic1–Hym1 module functions upstream of the Mob2–Cbk1 module by directly phosphorylating Cbk1’s HM site. Bacterially expressed Kic1 phosphorylates this site in vitro (Brace et al. 2011). Kic1 and Mob2–Cbk1 are reminiscent of Cdc15 and Mob1–Dbf2, where this regulatory interaction is well established (Mah et al. 2001; Visintin and Amon 2001). Similarly, in metazoans Mst/hippo family kinases like Kic1 phosphorylate the HM site of Ndr/LATS family kinases closely related to Cbk1 (Stegert et al. 2005; Hergovich and Hemmings 2009). Also, substitution with the phosphomimetic amino acid glutamic acid at the phosphoacceptor residue in Cbk1’s HM site renders upstream RAM components dispensable, at least for some functions, arguing that control of this regulatory site on Cbk1 is a major focus of the system’s upstream components (Panozzo et al. 2010).
Regulation of the RAM network by FEAR and MEN:
Consistent with mitotic exit control, Cbk1’s HM site is briefly phosphorylated during cytokinesis and cell separation, just prior to Ace2’s relocalization from the cytoplasm to the daughter cell nucleus (Jansen et al. 2006; Brace et al. 2011). As diagrammed in Figure 15, both the FEAR pathway and the MEN appear to coordinate this regulatory event with mitotic progress. Phosphorylation of the Cbk1 HM site is undetectable in both cdc5-1 and cdc14-2 cells arrested at restrictive temperature, consistent with FEAR regulation of the process or synergistic control by both FEAR and Cdc5. In contrast, phosphorylation of Cbk1’s HM site occurs in cdc15-2 cells arrested at restrictive temperature. These findings suggest that the FEAR pathway promotes the activation of upstream components of the RAM network, possibly by permitting interaction of Kic1–Hym1 with Mob2–Cbk1. Since cells lacking FEAR activity do not have dramatic problems with cell separation, activation of the RAM network is likely a role played by Cdc14, and any mechanism that releases it from the nucleolus (but not necessarily from the nucleus) is sufficient.
While necessary, phosphorylation of Cbk1’s HM site is not sufficient for the kinase to phosphorylate Ace2 and drive it into the daughter nucleus (Brace et al. 2011). Cbk1 (and possibly other components of the RAM network) appear to be held inactive by CDK phosphorylation that blocks Cbk1’s ability to interact with Ace2 until MEN activation. Upon sustained release of Cdc14, these modifications are reversed, likely permitting Cbk1 regulation of Ace2. Overall, the apparent dual control of Cbk1/RAM function at mitotic exit (Figure 15) emphasizes the importance of Cdc14 as a coordinator of late cell cycle events.
Control of Cell Separation Processes
The transition from M to G1 is marked by massive reorganization of numerous cellular components, coordinated with the mechanical action of the mitotic spindle and cytokinetic apparatus. Mitotic control systems therefore clearly drive a wide variety of processes, and these regulatory linkages remain incompletely understood. A few direct targets of mitotic exit pathways have been identified that are relevant to the separation of mother and daughter cells, and a few common regulatory themes emerge. Some target processes are controlled by APC-mediated destruction of key protein components. Other important proteins are held inactive in metaphase by mitotic CDK and activated upon dephosphorylation by Cdc14. Finally, in some cases the FEAR pathway, the MEN, or the RAM network directly regulate downstream events.
This section focuses mainly on control of septum destruction. The paralogous transcription factors Swi5 and Ace2 and their control are at the heart of this event and are important links between the regulatory systems that drive mitotic exit and the processes of cell separation and G1 onset. Thus, these important transcription factors are discussed in some depth.
Mitotic spindle stability
The mitotic spindle goes through major structural transitions as cells pass from M phase to early G1. It is highly stable only during early mitosis, and is destabilized upon mitotic exit. The protein Fin1 stabilizes the mitotic spindle, and the precise window in which it can act is under CDK control, so that it does not associate with the spindle when phosphorylated in S phase by the Clb5–CDK1 complex (Woodbury and Morgan 2007a). Upon dephosphorylation by Cdc14, Fin1 is released from inhibition and can strengthen the bipolar spindle (van Hemert et al. 2003; Woodbury and Morgan 2007b). As cells exit mitosis, coincident with spindle disassembly, APCCdh1 targets Fin1 for degradation.
Other spindle-stabilizing systems are controlled by mitotic exit. For example, CDK phosphorylation blocks spindle association of the midzone protein Ase1, and dephosphorylation by Cdc14 reverses this (Schuyler et al. 2003; Crasta et al. 2006; Khmelinskii et al. 2007; Khmelinskii and Schiebel 2008). Additionally, the Sli15–Ipl1 complex localizes from kinetochores (where it regulates their spindle attachment) to the spindle itself when Cdc14 reverses CDK phosphorylation of Sli15 (Pereira and Schiebel 2003). When localized to the central spindle, the Sli15–Ipl1 complex appears to help stabilize the structure.
Regulation of septation machinery and the cytokinetic apparatus
In addition to its role in spindle function, Cdc14 directly promotes cytokinetic events, including the formation of the primary septum. Like components of the MEN and RAM network, Cdc14 localizes to the site of cell division at the mother/bud neck, but it remains unclear if it functions there. In addition to Cdc14, the MEN plays important roles in separation and cytokinesis. Aspects of the cell cycle regulation of this part of mother/daughter separation are reviewed in other YeastBook chapters (Bi and Park 2012; Howell and Lew 2012) and are discussed only briefly here.
Mitotic control of chitin synthase trafficking:
The chitin synthase II catalytic subunit Chs2 is subject to multiple levels of cell cycle control that likely ensure that it is only available to form the primary septum. The CHS2 gene is subject to cell cycle regulation, and its expression peaks in early M phase (Pammer et al. 1992; Spellman et al. 1998). The Chs2 protein is then retained in the endoplasmic reticulum until cytokinesis, whereupon it moves through the secretory system to the bud neck, coincident with recruitment of exocytosis and endocytosis machinery to the site of cell division (Chuang and Schekman 1996; VerPlank and Li 2005; Zhang et al. 2006; McMurray et al. 2011). Chs2’s entry into the secretory system is antagonized by phosphorylation of CDK sites in its N terminus (Teh et al. 2009), and Cdc14 reverses this modification to allow trafficking of Chs2, the site of primary septum formation (Chin et al. 2012). Intriguingly, while CDK phosphorylation likely blocks Chs2’s secretion, it also greatly stabilizes the protein. This might reflect inhibition of proteolytic processing of a Chs2 zymogen. Moreover, the instability of dephosphorylated Chs2 might protect cells from inappropriate activation of this specialized chitin synthase (Martinez-Rucobo et al. 2009). The MEN has recently been demonstrated to directly regulate Chs2, through Dbf2 phosphorylation of at least one critical site (Oh et al. 2012). Dbf2 localizes to the division site as primary septum synthesis proceeds and appears to enhance Chs2’s synthesis of chitin. Intriguingly, this modification is also important for efficient dissociation of Chs2 from the division site, suggesting that Dbf2’s enhancement of primary septum synthesis ultimately helps turn the process off.
Regulation of the Hof1–Cyk3–Inn1 complex:
As discussed more fully by Bi and Park (2012) the assembly of a trimeric complex composed of the SH3-domain containing proteins Cyk3 and Hof1 and the proline-rich protein Inn1 is critical for the initiation of a properly placed primary septum (Jendretzki et al. 2009; Nishihama et al. 2009; Meitinger et al. 2010, 2011). This complex localizes to the actomyosin ring before it contracts, a process that involves coordinated phosphorylation of Hof1 by Mob1–Dbf2, Cdc5, and mitotic CDK. The overall effect of this regulation is to shift Hof1 from a septin-bound pool to an actomyosin association (Korinek et al. 2000; Vallen et al. 2000; Sanchez-Diaz et al. 2008; Meitinger et al. 2010, 2011). As a result, Inn1 and Cyk3 promote the localized activity of Chs2 and thus growth of the primary septum in concert with actomyosin ring contraction. This interaction requires Cdc14, which removes mitotic CDK phosphorylations on Inn1 that block its interaction with Cyk3′s SH3 domain (Palani et al. 2012).
Disassembly of the contractile apparatus:
In cytokinesis, the final separation of cytoplasms involves removal of the actomyosin ring components and resolution of the membrane junction that the contractile apparatus has brought together. In metazoans, this process is known as “abscission” (Barr and Gruneberg 2007); it is not clear if budding yeast undergo an identical event, although some similar processes are clearly needed (Dobbelaere and Barral 2004; Yoshida et al. 2009). One key event in yeast is the disassembly of the actomyosin ring. APCCdh1 plays an important role in this process, both during and after ring contraction (Tully et al. 2009). Cells lacking APCCdh1 show delayed disappearance of the contractile ring myosin Myo1, the IQGAP protein Iqg1, and the myosin light chains Mlc1 and Mlc2. Aberrant actomyosin ring disassembly leads to pronounced defects in the completion of the septum, with many cells failing to complete closure of a small region in the septum’s center. Part of this APCCdh1 function involves destruction of Iqg1, and Mlc1 likely plays a second reinforcing role (Ko et al. 2007; Tully et al. 2009).
Control of septum destruction and G1 entry: Ace2 and Swi5
The septum is rapidly destroyed minutes after it is built, starkly exemplifying precise coordination of opposing processes by the cell cycle regulatory system. Obviously, premature action of enzymes that efficiently degrade chitin and destroy or remodel other septum components would greatly complicate the structure’s construction during contraction of the cytokinetic furrow. Furthermore, septum destruction is spatially regulated, proceeding from the daughter cell side.
The control of this process in time and space is largely accomplished by RAM network regulation of both transcription and translation, making septum destruction an event occurring only once in a cell’s life, when it is a newly born daughter. This section describes current understanding of the control mechanisms underlying Ace2-driven transcription of genes involved in mother/daughter separation and daughter cell identity and then discusses recent insights into the RAM-mediated translational regulation of some of these genes by the mRNA binding protein Ssd1.
Ace2 works at the M/G1 transition along with the paralogous but functionally distinct transcription factor Swi5, which controls the expression of a distinct set of genes that close mitotic exit and function in G1 processes. Figure 16 shows the proteins’ cycle of nuclear localization, plus regulatory events important for their control. A discussion of their similarities and differences helps describe mechanisms of the mother/daughter separation process. However, since Swi5 is not critical for mother/daughter separation, I discuss Ace2 more extensively.
Functional organization of Ace2 and Swi5:
As diagrammed in Figure 17, Ace2 has C2H2 binuclear zinc finger motifs proximal to its C terminus, which likely recognizes the DNA motif GCTGG[G/T/C] (Dohrmann et al. 1992, 1996; Harbison et al. 2004; Badis et al. 2008). Separate motifs mediate nuclear import and export: a NLS and a nuclear export sequence (NES). Both have been experimentally verified, and the NES binds exportin Crm1/Xpo1 in vitro and in vivo (Jensen et al. 2000; Mazanka et al. 2008). A region in the middle of Ace2 mediates association with Cbk1 in two-hybrid studies (Racki et al. 2000) and appears to be functionally important in domain swap experiments with Swi5 (McBride et al. 1999; Sbia et al. 2008).
Ace2 and Swi5 are closely related paralogs, likely resulting from an ancient genome duplication event (Byrne and Wolfe 2005, 2006). While their N termini have diverged, their Zn-finger domains are highly similar, and they can bind the same DNA sequence motifs (Dohrmann et al. 1992, 1996). Although Swi5 appears to lack an NES, it has a related C-terminal NLS (Butler and Thiele 1991; Moll et al. 1991; Dohrmann et al. 1992). Despite their similarities, Ace2 and Swi5 behave differently in vivo and with a few exceptions do not regulate the same genes (Doolin et al. 2001; Voth et al. 2005). Paradoxically, Swi5 associates in vivo with binding sites at the promoters of genes that are exclusively Ace2 driven, but does not turn on their transcription. This failure of Swi5 to activate promoters to which it binds is due to local “antiactivation” mediated by the Rpd3 acetyltransferase complex recruited by the Forkhead-related DNA binding proteins Fkh1 and Fkh2 (Voth et al. 2007).
Ace2 and Swi5 at the dawn of G1: Ace2 controls cell separation genes; Swi5 closes mitotic exit:
Ace2 regulates genes encoding the most important septum destruction proteins, in a coherent pulse of transcription that coincides precisely with mother/daughter separation and is among the sharpest in the yeast cell cycle (Spellman et al. 1998; Pramila et al. 2002; Lu et al. 2007). These genes include CTS1, SCW11, DSE2, DSE3, and DSE4, which as discussed above encode known or putative septum hydrolases (Dohrmann et al. 1992; O’Conallain et al. 1998, 1999; McBride et al. 1999; Bidlingmaier et al. 2001; Colman-Lerner et al. 2001; Doolin et al. 2001; Di Talia et al. 2009). Ace2 also clearly drives expression of the bud site selection gene BUD9 as well as DSE1, which may be involved in bud scar organization and bud site selection (Bidlingmaier et al. 2001; Voth et al. 2005; Draper et al. 2009; Frydlova et al. 2009). Additionally, Ace2 may promote transcription of the SUN4 glucanase (Bidlingmaier et al. 2001; Di Talia et al. 2009). For brevity, this group of Ace2-regulated genes are henceforward referred to as mother–daughter separation (MDS) genes. In a fascinating mechanism of cell cycle control that is beyond the scope of this chapter, Ace2 also lengthens G1 in daughter cells by inhibiting expression of the early G1 cyclin CLN3, helping them to attain a size suitable for budding (Laabs et al. 2003; Di Talia et al. 2009).
The targets of Swi5 notably include the protein Sic1 (Toyn et al. 1997), a potent CDK inhibitor. Its expression and activation during the M-to-G1 transition creates a negative feedback loop that makes mitotic exit an irreversible process (Luca et al. 2001; Lopez-Aviles et al. 2009; He et al. 2011). Swi5 also drives the expression of the cell separation gene EGT2 and the transcriptional repressor ASH1, an mRNA that is transported to the daughter cell and ultimately translated there to suppress mating type switching. Since the ASH1 mRNA must be transported to the daughter cell prior to cytokinesis, this further emphasizes the fact that Swi5′s regulation of transcription happens just prior to division.
Ace2 and Swi5 are produced in mitosis but sequestered in the cytoplasm:
Ace2 and Swi5 are initially regulated at the level of their own transcription. Both genes are activated in early M phase under the control of a transcription factor complex known as SFF, which includes the Forkhead-related transcription factor Fkh2, the general transcription factor Mcm1, and the co-activator Ndd1 (Hollenhorst et al. 2000, 2001; Koranda et al. 2000; Zhu et al. 2000). SFF drives coherent expression of the “CLB2 cluster” of genes, which in addition to ACE2 and SWI5 includes other genes involved in mitotic control and cytokinesis (Spellman et al. 1998).
Unsurprisingly given their shared origin, some of the same mechanisms control the nuclear accumulation of both Ace2 and Swi5. Nuclear entry of each protein is inhibited by mitotic CDK activity, with phosphorylation directly inactivating their NLS and hence trapping them in the cytoplasm until mitotic exit (Figure 16, see diagrams next to cell outlines). Appending Swi5′s NLS to a heterologous protein confers cell-cycle–dependent nuclear localization, involving CDK phosphorylation of sites near the NLS; this was one of the earliest identifications of a physiologically important Cdk1 substrate (Moll et al. 1991). Similarly, Ace2 is present in early mitotic cells but cannot enter the nucleus because its NLS is not functional until mitotic exit has commenced (Dohrmann et al. 1992; O’Conallain et al. 1999). As in Swi5, the Ace2 NLS is flanked by minimal CDK phosphorylation motifs (Figure 17); and alanine replacement at these sites allows nuclear entry of Ace2 (O’Conallain et al. 1999; Sbia et al. 2008; Mazanka and Weiss 2010; Brace et al. 2011).
In Ace2’s case, the Clb3:Cdk1 complex likely phosphorylates the key NLS-inactivating CDK sites, and large-scale mass spectrometry confirms that Ace2 is an in vivo substrate of mitotic Cdk1 (Archambault et al. 2004; Holt et al. 2009). In addition to sites adjacent to its NLS, Ace2 is phosphorylated on numerous other minimal CDK motifs, suggesting extensive regulation by cyclin–CDK (Archambault et al. 2004). These sites appear to participate in the Ace2’s inactivation in G1 (Mazanka and Weiss 2010), but their function in mitosis remains unclear.
Cdc14 relieves inhibition of Swi5 and Ace2 nuclear import:
Cdc14 efficiently removes CDK phosphorylations that inhibit Swi5′s nuclear import, making nuclear entry of Swi5 strongly dependent on mitotic exit (Visintin et al. 1998; Traverso et al. 2001; Stegmeier and Amon 2004). Similarly, Ace2 was found to bind Cdc14 in a mass spectrometric analysis (Breitkreutz et al. 2010). Furthermore, Ace2 is heavily phosphorylated in mitotically arrested cells, Cdc14 removes most of this phosphorylation in vitro, and dephosphorylation of Ace2’s CDK consensus motifs occurs abruptly upon release from metaphase arrest in a manner dependent on Cdc14 (Archambault et al. 2004; Sbia et al. 2008; Mazanka and Weiss 2010). As discussed further below, dephosphorylation by Cdc14 is both necessary and sufficient for Swi5 to induce transcription of its target genes, while for Ace2 this is necessary but not sufficient.
Ace2 localizes to the daughter cell nucleus just prior to cytokinesis:
The timing and daughter-specific expression of MDS genes arises from M/G1 activation and asymmetric partitioning of Ace2, which is diagrammed and compared with Swi5 in Figure 16. To summarize work of multiple groups, Ace2 is initially isotropically distributed in mitotic cells, but is excluded from nuclei until well into anaphase. The protein becomes weakly visible in nuclei of mother and daughter cells in early telophase. If nuclear export is blocked at this stage by either genetic or pharmacological inhibition of the exportin Crm1/Xpo1, Ace2 accumulates strongly in both nuclei. Thus, at this time of the cell cycle Ace2 undergoes active import and export from nuclei, and there is no evident asymmetry between mother and daughter in these rates. A few minutes prior to actomyosin ring contraction, Ace2 rapidly shifts to nearly exclusive accumulation in the daughter cell’s nucleus, where it remains until shortly after mother/daughter separation is complete. Ace2’s localization to the daughter cell nucleus is accompanied by a corresponding reduction in its cytoplasmic concentration in both mother and daughter cells and occurs independently of the protein’s ability to bind DNA (Dohrmann et al. 1992; O’Conallain et al. 1999; Racki et al. 2000; Bidlingmaier et al. 2001; Colman-Lerner et al. 2001; Doolin et al. 2001; Weiss et al. 2002; Nelson et al. 2003; Voth et al. 2005; Bourens et al. 2008; Mazanka et al. 2008; Sbia et al. 2008; Di Talia et al. 2009; Mazanka and Weiss 2010; Brace et al. 2011).
Soon after separation is completed in early G1, and coincident with disappearance of MDS transcripts, Ace2 disappears from daughter cell nuclei; this is not due to degradation but rather to its relocalization to the cytoplasm where it is retained by a complex cytoplasmic trapping mechanism (Sbia et al. 2008; Mazanka and Weiss 2010; Brace et al. 2011). The Ace2-driven asymmetry of MDS gene expression has been visualized as daughter-specific accumulation CTS1 mRNA and daughter cell fluorescence in strains carrying a GFP variant under the control of the DSE1 promoter (Colman-Lerner et al. 2001; Bourens et al. 2008). This partitioning is quite stringent, and the DSE1 promoter has been exploited to construct a genetically encoded system called the “mother enrichment program” that conditionally kills only daughter cells, to assist studies of aging and replicative lifespan (Lindstrom and Gottschling 2009; Lindstrom et al. 2011).
Ace2 partitioning to the daughter cell nucleus resembles segregation of Ash1, a transcriptional repressor that blocks mating type switching in daughter cells. However, these asymmetries are generated by different mechanisms. Ash1 accumulates mainly in the nuclei of daughter cells because its mRNA is first moved as a translationally inactive ribonucleoprotein particle (RNP) to the daughter cell by the type V myosin Myo4 traveling on cytoplasmic actin cables and then translated as the cells divide. (Darzacq et al. 2003; Paquin et al. 2007; Hasegawa et al. 2008; Paquin and Chartrand 2008). Ace2 localization does not require cytoskeletal components needed for Ash1 asymmetry (Weiss et al. 2002). Rather, the segregation of Ace2 is most likely diffusion mediated and involves selective inactivation of its nuclear export in daughter cells by the RAM network; by contrast, Ash1 segregation does not require RAM network function (Weiss et al. 2002). A significant fraction of Ash1 is present in the nuclei of mother cells (McBride et al. 1999), suggesting that ASH1 mRNA segregation is not as stringent a mechanism for protein localization as the partitioning of Ace2.
RAM network activates Ace2 and traps it in the daughter cell nucleus:
Removal of CDK phosphorylation near Ace2’s NLS permits the protein to enter nuclei, and is necessary but not sufficient for its localization and transcriptional activation of MDS genes (O’Conallain et al. 1999; Mazanka et al. 2008; Sbia et al. 2008; Mazanka and Weiss 2010; Brace et al. 2011). This contrasts with Swi5: NLS unmasking during mitotic exit is both necessary and sufficient for Swi5 to accumulate to high levels in nuclei and activate transcription of its target genes (Moll et al. 1991; Jans et al. 1995; Doolin et al. 2001; Sbia et al. 2008). The reason for this difference is that Ace2 has a highly active NES, and its equilibrium concentration in nuclei is not high enough to drive expression of its target genes when both its import and export are unrestrained. In contrast, Swi5 lacks an NES and thus rapidly accumulates in nuclei when its NLS is uninhibited (Sbia et al. 2008).
Cbk1 directly regulates Ace2’s nucleocytoplasmic shuttling by phosphorylating and inactivating its NES. Two Cbk1 consensus motifs are present within Ace2’s NES (Figure 17). Cbk1 phosphorylates these in vitro, and mass spectrometry revealed in vivo phosphorylation of both residues (Mazanka et al. 2008; Gnad et al. 2009; Bodenmiller and Aebersold 2011). A phosphospecific antibody raised against one of these sites indicates that the modification occurs shortly after release from mitotic block, coincident with the transcription factor’s nuclear localization, and that it requires Cbk1 function (Mazanka and Weiss 2010; Brace et al. 2011). Mutations that inactivate Ace2’s NES, including phosphomimetic substitutions at either Cbk1 site, cause it to localize strongly to both mother and daughter nuclei and make MDS gene expression and septum destruction independent of the RAM network (Racki et al. 2000; Mazanka et al. 2008; Sbia et al. 2008). Binding of the exportin Crm1/Xpo1 to Ace2’s NES has been reconstituted in vitro, and phosphorylation of either Cbk1 site disrupts the association (Mazanka and Weiss 2010). Importantly, eliminating these Cbk1 sites prevents Ace2’s concentration in the daughter nucleus, resulting instead in a faint mother–daughter distribution like that seen just after mitotic exit.
Cbk1 also regulates Ace2 by phosphorylating a more C-terminal site (S436). This is not required for Ace2’s localization to the daughter nucleus, but rather enhances its ability to activate target genes (Mazanka et al. 2008). Thus, Cbk1 controls Ace2 in two synergistic ways. NES inactivation drives accumulation in the daughter cell nucleus, but is not fully necessary for induction of Ace2-responsive genes. Likewise, S436 phosphorylation is not absolutely necessary for Ace2 function when the NES phosphorylation sites are present.
Turning Ace2 and Swi5 off:
The transcription of Swi5- and Ace2-driven genes occurs in an exceptionally sharp burst in early G1 daughter cells. (Spellman et al. 1998; Pramila et al. 2006; Li et al. 2009). Such a pulse requires both an “on” and “off” switch. As shown in Figure 16, the bursts of Ace2 and Swi5 localization follow subtly different courses that reflect important differences in their regulation (Sbia et al. 2008; Mazanka and Weiss 2010). In many cases, inactivation phases of biochemical pulse-generating systems involve destruction of key components (e.g., Pomerening et al. 2003). In fact, Swi5′s inactivation follows this route: the protein is degraded in early G1, likely triggered by the G1 CDK Pho85 (Tebb et al. 1993; Measday et al. 2000; Sbia et al. 2008). In vivid contrast, Ace2 remains stable throughput the cell cycle and is turned off by a complex system that results in its nuclear export and stable cytoplasmic sequestration (Sbia et al. 2008; Mazanka and Weiss 2010). This includes periods in G1 in which Mob2–Cbk1 and the RAM network is active in the control of cell morphogenesis (Racki et al. 2000; Bidlingmaier et al. 2001; Colman-Lerner et al. 2001; Weiss et al. 2002; Nelson et al. 2003; Jansen et al. 2006, 2009).
The first phase of Ace2 inactivation involves dynamic nuclear retention that determines the length of the pulse of Ace2-driven transcription. In early G1, Cbk1 continuously rephosphorylates and thus inactivates Ace2’s NES in the nucleus (Mazanka and Weiss 2010). Inhibition of Cbk1 causes rapid loss of Ace2 NES phosphorylation and immediate export of the protein from the nucleus, indicating that Cbk1 acts in the presence of an unknown countervailing phosphatase and that a dynamic balance of competing kinase and phosphatase activities determines how long Ace2 stays in the nucleus. During normal G1 transit Cbk1 eventually loses this competition, and Ace2’s NES is dephosphorylated and thus activated (Mazanka et al. 2008; Sbia et al. 2008; Mazanka and Weiss 2010).
Ace2 cannot reenter the nucleus once it is exported in G1 (Bourens et al. 2008; Mazanka et al. 2008; Sbia et al. 2008; Mazanka and Weiss 2010), due to a multistage mechanism that traps it in the cytoplasm. This sequestration requires either of two G1 CDKs: Pho85 or Cdk1 (Mazanka and Weiss 2010). In principle, these G1 CDKs could trap Ace2 in the cytoplasm entirely by phosphorylating sites that inactivate Ace2’s NLS, recapitulating control by M phase Cdk1. However, Ace2 becomes extensively phosphorylated at CDK sites prior to its export from the nucleus, and a mutant form of Ace2 (ace2-AP) lacking all 21 minimal CDK sites is only partially defective in G1 cytoplasmic sequestration (Mazanka and Weiss 2010). Moreover, in cells arrested at G1/S by Cdk1 inhibition, both Ace2-AP and wild-type Ace2 proteins are identically blocked from entering the nucleus. How is Ace2 kept in the cytoplasm without direct phosphorylation of its CDK sites? Importantly, G1 cytoplasmic trapping of Ace2-AP is essentially absent in pho85Δ cells, but loss of Pho85 kinase activity does not cause this effect (Mazanka and Weiss 2010). Thus, Pho85’s role here is independent of its kinase function, and perhaps a Pho85–cyclin complex binds Ace2 in G1 and blocks its nuclear import, with CDK phosphorylation promoting either Pho85–cyclin association or its binding to Ace2.
Thus, Ace2 inactivation may occur in multiple stages, by contrast with the relatively simple degradation of Swi5 (Figure 18; also see Figure 16). Early G1 CDK phosphorylation of Ace2 (by Cdk1, Pho85, or a transcription-associated CDK) may inactivate its NLS while it is still in the nucleus, ensuring that unmasking of Ace2’s NES rapidly turns off expression of its target genes. Later in G1, Pho85 and/or Cdc28 activate Ace2 cytoplasmic trapping, which involves a kinase-independent function of Pho85. Such a multistage system might allow both a fast block to Ace2 nuclear reentry that sharpens the off phase of MDS gene expression, as well as a slower-maturing retention mechanism that is more stable and resistant to protein phosphatases. These models remain speculative, however. It is also not clear if there is any benefit to retaining stable Ace2 in the cytoplasm; conceivably, it could be rapidly released under certain growth conditions (Butler and Thiele 1991).
An unknown mechanism restricts the RAM network’s control of Ace2 to the daughter cytoplasm:
Ace2’s rapid concentration in the daughter cell nucleus prior to actomyosin ring contraction is a compelling example of the partitioning of a transcription factor that is initially distributed evenly in a contiguous cytoplasm. This process appears to be entirely mediated by diffusion, with the daughter nucleus acting as a trap for Ace2. Intriguingly, Ace2’s nuclear accumulation does not occur if nuclear division occurs entirely in the mother cell, suggesting that the daughter cell cytoplasm has a special ability to promote inactivation of the Ace2 NES (Mazanka et al. 2008). Since Mob2–Cbk1 (and other RAM network proteins) concentrate at the bud cortex, it is tempting to speculate that their localization is causal. However, while Ace2 physically interacts with the Mob2–Cbk1 module, it has never been seen to colocalize with them at the cell cortex, and it is not clear why Ace2 activated in the cytoplasm would not diffuse into the mother cell. Phosphatase activities localized to the mother or bud neck could maintain a steep gradient in Cbk1 activation. Indeed, a point mutation in Cbk1’s C-terminal tail that probably hyperactivates the kinase increases localization of Ace2 in the mother cell nucleus (Panozzo et al. 2010). However, Ace2 asymmetry cannot exclusively involve spatial control of Cbk1’s HM site, as acidic substitutions at Cbk1’s HM phosphorylation site create a gain-of-function allele but do not cause symmetrical Ace2 distribution (Panozzo et al. 2010; Brace et al. 2011). Therefore, the mechanism responsible for the diffusion-mediated partitioning of Ace2 remains obscure.
RAM network controls translation of cell separation proteins
Recent analysis has shown that Cbk1, and by extension the RAM network, controls the translation of specific proteins by directly regulating the mRNA binding protein Ssd1. In brief, Ssd1 is an RNAseII-related protein broadly conserved in fungi that appears to lack catalytic activity; it has been implicated in numerous processes. It binds specific mRNAs and suppresses their translation, and Cbk1 efficiently directly negatively regulates this Ssd1 function in vivo (Hogan et al. 2008; Jansen et al. 2009; Kurischko et al. 2011a). This is critical for cell wall remodeling during bud growth; however, the Ssd1 translational control system’s important role in cell morphogenesis outside of the process of mother/daughter separation is not discussed further in this chapter. As diagrammed in Figure 19, Cbk1 regulates Ssd1 through direct in vivo phosphorylation of sites in Ssd1’s N-terminal region (Jansen et al. 2009).
Ssd1 associates with mRNAs that encode some of the most important MDS proteins, such as Cts1, Dse2, and most of the SUN-family glucanases (Hogan et al. 2008; Jansen et al. 2009); the CTS1 and SUN4 mRNAs are particularly strongly enriched in Ssd1 pull-down experiments. Translation rate measurements show that Ssd1-associated messages are significantly suppressed when Cbk1 activity is acutely blocked (Jansen et al. 2009). Ssd1 can localize to cytoplasmic granules known as P-bodies, which are associated with translational suppression (Balagopal and Parker 2009; Jansen et al. 2009; Kurischko et al. 2011a), but Ssd1 can clearly carry out its translational repression function in cells lacking these structures. Ssd1 likely assembles a translationally silent mRNP in the nucleus: it associates with unspliced and centromeric mRNAs, interacts with the phosphorylated C-terminal tail of RNA polymerase II, and requires a sequence that can function as an NLS (Phatnani et al. 2004; Jansen et al. 2006; Hogan et al. 2008; Kurischko et al. 2011b). The mechanisms and precise function of Ssd1’s control of translation remain poorly understood, but since RAM network proteins concentrate at the bud neck during cell separation, the Ssd1 circuit could allow the cytokinesis site to exert post-transcriptional control over the expression of proteins needed at different stages of the process (Nelson et al. 2003; Jansen et al. 2006; Kurischko et al. 2011a).
Model for septum destruction control: cascading feed-forward loops
The RAM network’s control of cell separation is a remarkable multilayered system. By regulating both Ace2 and Ssd1, the pathway activates both the transcription and translation of a specific set of genes. As diagrammed in Figure 20, Mob2–Cbk1 phosphorylation of these effectors creates a FFL network motif. More specifically, this configuration forms a coherent feed-forward loop, in which the system’s two branches exert an effect of a similar sign on the downstream component or process (Mangan and Alon 2003). Such network motifs are significantly enriched in biology and can create systems that turn on slowly but off quickly (Mangan and Alon 2003; Dekel et al. 2005; Kalisky et al. 2007). Intriguingly, upstream control of Ace2 and Mob2–Cbk1 during the M/G1 transition may constitute a similar network motif, with Cdc14 activating both Mob2–Cbk1 and Ace2. This produces a pathway in which two such loops are linked in a cascading fashion, consistent with the construction of complex biological regulatory systems from recurring simple network motifs (Milo et al. 2002; Yeger-Lotem et al. 2004; Kashtan and Alon 2005; Alon 2007). While much remains to be learned about the septum destruction control system, it is possible that this network organization helps ensure that the process does not start before cytokinesis is complete.
Studies of the asymmetric division of budding yeast cells have greatly illuminated molecular machines and regulatory systems that allow the spectacular feat of cell division and has identified systems that are broadly important for eukaryotic cell proliferation. In particular it has become clear that mitotic exit control pathways coordinate multiple independent processes, which act in rapid sequence to split the dividing cell into two unequal progeny. However, while the molecular components involved in both the mechanics and control of this event are better known, much remains mysterious. It seems clear that we are only beginning to identify targets of the APC/C, Cdc14, the MEN kinases Cdc15 and Dbf2, and the RAM network that are relevant to the processes that productively cut the dividing cell in two.
It is still not clear how the relative timing of cytokinetic events is enforced: for example, why do cells not start destroying the septum before they have finished building it? Similarly, we do not fully understand how mitotic exit control pathways are integrated such that they act sequentially. The events leading to RAM network activation are not well understood, nor is it known how these proteins get to the cell cortex and bud neck or if this is even important for their function. On the cytological side, mechanisms for cutting the divided nucleus, forming the secondary septum, and halting septum synthesis remain unknown. While the activation and nuclear localization of Ace2 is well worked out, the mechanism responsible for its asymmetric partitioning is not. Finally, the RAM network’s role in cell morphogenesis and translational control is only dimly understood: it is not clear how Ssd1 suppresses translation of associated mRNAs and if it promotes their localized translation. Future research in these areas will define how diverse systems are integrated to achieve effective cell division.
I thank Angelika Amon, David Stillman, and members of my research group at Northwestern University for helpful comments on this work. My laboratory’s research is supported by National Institutes of Health grant GM084223 and by funding from the Chicago Biomedical Consortium with support from the Searle Funds at The Chicago Community Trust.
Communicating editor: J. Thorner
- Received April 6, 2012.
- Accepted August 30, 2012.
- Copyright © 2012 by the Genetics Society of America