Dividing the spoils of growth and the cell cycle: The fission yeast as a model for the study of cytokinesis^†

Cell-Cycle Control of Cytokinesis

S. pombe has a single, essential mitotic cyclin-dependent kinase (CDK), named cdc2p: its function is conserved in distantly related yeasts such as S. cerevisiae where it is named CDC28, and in human cells, where it is known as CDK1 [Beach et al.,1982; Lee and Nurse,1987]. Cdc2p is required for mitotic entry; its activity is regulated in part by phosphorylation at Tyr15. The level of phosphorylation at this site is governed by the relative activities of the protein kinase wee1p and the phosphoprotein phosphatase cdc25p. The activities of cdc25p and wee1p are regulated at multiple levels to ensure that mitotic commitment occurs at the appropriate time; recent studies have revealed mechanisms that appear to coordinate growth, nutrient status and cell division [Petersen and Nurse,2007; Martin and Berthelot-Grosjean,2009; Moseley et al.,2009].

Reorganization of the actin and tubulin cytoskeletons for mitosis and cytokinesis depends upon commitment mitosis and CDK activity [reviewed by Marks et al.,1986; Hagan,1998]. Once mitosis has been initiated, elevated cdc2p activity prevents cytokinesis early in mitosis [Yamano et al.,1996; Chang et al.,2001; Dischinger et al.,2008]. Following CDK inactivation, cytokinesis requires the activity of a group of protein kinases known as the SIN (see below).

Positioning of Division Plane

Unlike mammalian cells, fission yeast undergo a “closed” mitosis, in which the nuclear envelope does not break down, and the spindle forms inside the nucleus [Hagan and Hyams1988; reviewed by Hagan,1998]. A consequence of this is that the molecular mechanism controlling the positioning of the division plane differs from that in animal cells. Though microtubules are not required for cytokinesis in S. pombe [He et al.,1997,1998; Sparks et al.,1999], they do play an important role in the spatial control of the division plane. The division site corresponds to the location of the late interphase nucleus, which is kept in the middle of the cell by a microtubule-dependent mechanism [Tran et al.,2000]. Moreover, the localization of pom1p at the cell tips [Bähler and Pringle,1998] is assured by the cytoplasmic MTs in S. pombe [reviewed by Martin,2009].

The PH-domain protein mid1p, is a key factor of positioning CR at the medial cell cortex [Chang et al.,1996; Sohrmann et al.,1996; Daga and Chang,2005]. Mid1p shuttles between the nucleus and cell cortex [Paoletti and Chang,2000] and positions the division plane by acting as an “anchor” for CR proteins. At least three molecular systems control mid1p localization in a cell-cycle specific manner. The first is that cdr2p forms nodes that physically anchor a population of mid1p at cell cortex during interphase [Almonacid et al.,2009]. These interphase mid1p nodes are excluded from cell tips by pom1p and passively restricted to the medial cortex [Celton-Morizur et al.,2006; Padte et al.,2006]. This pom1p-mediated tip occlusion system is also important for controlling cell polarity [Tatebe et al.,2008]. The finding that a partial reduction of cdc15p activity is able to restore normal septum placement in cells lacking mid1 and pom1 [Huang et al.,2007] suggests that the tip occlusion system also inhibits cdc15p and other CR proteins to prevent septation at the cell ends. This is reminiscent of the situation in bacteria, where systems exist to prevent cytokinesis at the cell tips under normal conditions; reviewed by [Oliferenko et al.,2009].

The second is nuclear export of mid1p at the G2/M transition, which targets mid1p to the cortex proximal to the nucleus [Sohrmann et al.,1996; Paoletti and Chang,2000; Daga and Chang,2005], which is likely to be mediated by plo1p [Bähler et al.,1998]; Plo1p directly binds to mid1p and seems to play an important role in formation of the medial mid1p ring; the protein kinase pdk1p also contributes to this process [Bimbo et al.,2005]. Consistent with this, both kinases localize to SPB and the medial ring in mitosis [Bähler et al.,1998; Bimbo et al.,2005]. The third regulatory mechanism is thought to be the presence of domains formation of tightly packed ER in the medial region, which helps to retain mid1p at the CR [Zhang et al.,2010]. Thus, redundant mechanisms including a positive spatial cue from the centrally positioned nucleus and inhibitory signals at the cell tips contribute to medial positioning of the division plane to produce similar-sized daughter cells.

A Time Course of CR-Assembly

An elegant study from the Pollard lab [Wu et al.,2003] used fluorescent protein tagged ring components, which had been tested as far as possible for their biological function, to examine the kinetics and timing of CR assembly. Using SPB markers, they were able to correlate this with mitotic progression. Though the behavior of many individual CR proteins had been analyzed previously, this study examined many CR markers in parallel, under similar filming and growth conditions, enabling the relative timing of their appearance at the CR to be investigated. It therefore provides an important landmark in our understanding of how the CR assembles. The process of CR assembly and cytokinesis is shown in cartoon form in Fig. 1.

When cells enter metaphase, F-actin and myosin II (myo2p) independently appear on the medial cortex and then interact with each other to form the CR [Naqvi et al.,1999; Motegi et al.,2000]. Prior to F-actin localization, accumulation of myo2p as cortical dots is induced, probably through interaction with mid1p [Motegi et al.,2004]. Importantly, whereas full-length myo2p will only accumulate at the medial region after onset of metaphase, a C-terminal tail region of myo2p can associate with cortical mid1p nodes before the G2/M transition, suggesting that interphase-specific auto-inhibition of localization to nodes may be an immanent property of myo2p [Motegi et al.,2004]. It has also been suggested that mitosis-specific dephosphorylation of Ser1444 controls the timing of association of myo2p with cortical nodes [Motegi et al.,2004]. Though expression of the nonphosphorylatable mutant myo2S1444A induces CR-assembly in G2-arrested cells [Motegi et al.,2004], a recent study casts doubt on the importance of this phosphoregulation because the same mutation does not affect the time course of myo2p accumulation into CR, and the phosphomimetic myo2S1444D mutant, can substitute for the wild-type protein without significantly affecting cytokinesis [Sladewski et al.,2009]. At present, it remains unclear how the auto-inhibition of myo2p is relieved at G2/M transition.

In addition to myo2p, an IQGAP-like actin-crosslinking protein rng2p, the formin cdc12p, and the F-BAR protein cdc15p are also incorporated into cortical mid1p nodes in early metaphase, independently of F-actin [Wu et al.,2003]. Dynamic redistribution of actin from cell tips, where actin patches are located in interphase, to the medial region occurs during formation of CR F-actin, which depends on the actin-depolymerizing factor, adf1p [Nakano and Mabuchi,2006]. Longitudinal F-actin cables are often seen in the middle of early metaphase cells [Arai et al.,1998]. However, preexisting actin cables are not essential for CR assembly, because functional CR is formed in cells lacking the cables such as for3-null mutant [Nakano et al.,2002]. Therefore, de novo polymerization of F-actin is the dominant pathways for CR assembly in S. pombe. Actin subunits depolymerized from actin patches in cell tips are bound to profilin cdc3p [Balasubramanian et al.,1994] and are repolymerized in the medial region of the cell, probably by cdc12p [Chang et al.,1997]. The cdc12p-capped barbed end of F-actin favors the use of the cdc3p-actin complex for polymerization [Kovar et al.,2003]. Recent biochemical and genetic analyses suggest that rng2p may be involved in actin-polymerization redundantly with cdc12p [Takaine et al.,2009]. Subsequent to this, myo2p dots associate with each other through interacting with F-actin and finally the CR is formed before anaphase. Thus, the timing of CR formation in S. pombe differs from animal cells, where assembly of the CR is induced after inactivation of CDK1 and anaphase onset. Formation of the S. pombe CR early in mitosis is mediated by mid1p, which acts as an organizing scaffold for the actin polymerization machinery and myo2p to the future division site. Mid1p has a C-terminal PH domain with low similarity to the metazoan cytokinesis proteins of the anillin family, which directly bind to both F-actin and myosin II [reviewed by D'Avino,2009]. Though mid1p may share the functional similarity with anillin, it is unclear whether these proteins have evolved from a common ancestral gene.

In mid1 mutant cells, although CR formation is not initiated at metaphase, myo2p associates with a cable-like F-actin structure elongating from the cell cortex during anaphase, resulting in formation of an abnormally shaped CR, which is frequently displaced from the cell middle [Huang et al.,2007]. Therefore, mid1p-dependent events assure that S. pombe division is symmetrical while another system, probably SIN-mediated, supports CR-assembly. Interestingly, at least two other species of the Schizosaccharomyces genus, S. japonicus and S. octosporus do not have a clear ortholog of mid1p and in these organisms CR assembly is initiated during anaphase [Alfa and Hyams1990] (KN, unpublished data). All three species (KN, unpublished data) have a conserved anillin-related protein, named mid2p in S. pombe, which is required for formation of the septin ring and cell separation in S. pombe [Berlin et al.,2003; Tasto et al.,2003].

Flp1p/clp1p is the S. pombe ortholog of the conserved CDC14-family of phosphoprotein phosphatases; it is implicated in many cellular processes, and is regulated by phosphorylation and localization [Cueille et al.,2001; Trautmann et al.,2001; Wolfe et al.,2006]. During mitosis, it localizes to the CR by associating with mid1p and controls the dynamics of CR components such as cdc15p and myo2p [Clifford et al.,2008]. Flp1p/clp1p also contributes to the dephosphorylation of cdc15p, probably in cooperation with other phosphatases [Fankhauser et al.,1995; Wachtler et al.,2006; Clifford et al.,2008]. Flp1p/clp1p is likely to have many targets, and since it is not an essential gene, other phosphoprotein phosphatases will also doubtless regulate aspects of cytokinesis in S. pombe. The CR is assembled stepwise during mitosis [Wu et al.,2003]; when it is fully compacted at the onset of anaphase B, completion of cytokinesis requires the SIN. The following sections will describe the SIN, what it does and how it is regulated.

The Septation Initiation Network

What Is the SIN?

The SIN is a group of protein kinases which are essential for cytokinesis. Signaling requires the activity of three protein kinases, each of which has a regulatory subunit (kinase-regulator); cdc7p-spg1p [Fankhauser and Simanis,1994; Schmidt et al.,1997; Mehta and Gould,2006], sid1p-cdc14p [Fankhauser and Simanis,1993; Guertin et al.,2000; Guertin and McCollum,2001] and sid2p-mob1p [Sparks et al.,1999; Hou et al.,2000; Salimova et al.,2000]. SIN signaling is modulated by the nucleotide status of the GTPase spg1p [Schmidt et al.,1997; Sohrmann et al.,1998] (Fig. 2). This is determined by the balance of spontaneous nucleotide exchange, a putative GEF, etd1p [Daga et al.,2005; Garcia-Cortes and McCollum,2009] and a GAP, cdc16p [Minet et al.,1979; Fankhauser et al.,1993], with which spg1p interacts through a scaffold, byr4p [Song et al.,1996; Furge et al.,1998; Furge et al.,1999]. The SIN is also activated by the mitotic regulator plo1p [Tanaka et al.,2001], which is the fission yeast orthologue of Drosophila POLO [Ohkura et al.,1995]. Loss of SIN signaling produces multinucleate cells, while constitutive activation of the SIN results in multiseptated cells [Minet et al.,1979; Fankhauser et al.,1993; Song et al.,1996]. Ectopic activation of the SIN by either overexpression of plo1, cdc7, or spg1 promotes CR and septum formation from any stage of the cell cycle, uncoupling the usual dependency of cytokinesis upon entry into mitosis [Fankhauser and Simanis,1994; Ohkura et al.,1995; Schmidt et al.,1997; Guertin et al.,2002].

SIN Signaling Requires Association of SIN Proteins with the SPB

Laser ablation of SPBs in fission yeast during mitosis suggests that at least one SPB must be intact during anaphase B for cytokinesis to occur [Magidson et al.,2006]. The S. pombe SPB is composed of cytoplasmic and nuclear components which are separated by the nuclear envelope and connected by fine striations [Ding et al.,1997]. The duplication of the SPB appears to be conservative [Ding et al.,1997; Grallert et al.,2004], generating “old” and “new” SPBs that can be distinguished using slow-folding RFP-tagged proteins [Grallert et al.,2004]. Studies of the localization of SIN proteins indicates that association with the SPB at various points of the cell cycle plays an important part in regulating the SIN, and therefore in the coordination of mitosis and cytokinesis [reviewed by Simanis,2003; Wolfe and Gould,2005; Krapp and Simanis,2008; Lattmann et al.,2009]. The distribution of the SIN proteins during mitosis is shown in Fig. 3.

The SIN proteins associate with the SPB via a tripartite scaffold comprised of ppc89p, sid4p and cdc11p [Chang and Gould,2000; Krapp et al.,2001; Tomlin et al.,2002; Morrell et al.,2004; Rosenberg et al.,2006]. Mutants that compromise either sid4 or cdc11 block association of SIN proteins with the SPB, and prevent SIN signaling. Ectopic activation of the SIN in a sid4 mutant fails to promote septum formation, indicating that SPB association of SIN proteins is important for signaling [Balasubramanian et al.,1998]. The scaffold proteins are seen at the SPB throughout the mitotic cycle. The requirement for the SIN in CR assembly also suggests that the SIN has a cytoplasmic role, but it is noteworthy that all of its functions seem to require the SPB-associated scaffold molecules sid4p and cdc11p [Hachet and Simanis,2008].

During interphase, spg1p, byr4p, and cdc16p are all observed at the SPB; cdc16p and byr4p are interdependent for localization [Sohrmann et al.,1998; Cerutti and Simanis,1999; Li et al.,2000]. This tripartite complex [Furge et al.,1998] is presumed to be inactive for signaling the initiation of septation, but whether it has any other role at the SPB in interphase is unclear. The steady state level of byr4p is influenced by its ability to bind spg1p [Krapp et al.,2008], which may regulate the amount of GAP present in the cell. Following CDK activation and entry into mitosis, cdc16p is removed from the SPB. In the absence of cdc16p, byr4p prevents release and hydrolysis of GTP by spg1p in vitro [Furge et al.,1998], and may thus contribute to stabilizing active spg1p in early mitosis.

Spg1p in its GTP-bound form interacts with cdc7p during mitosis [Sohrmann et al.,1998; Mehta and Gould,2006], and is observed initially on both SPBs, where low levels of byr4p are also present [Cerutti and Simanis,1999; Li et al.,2000]. Cdc7p requires spg1p to localize to the SPB [Sohrmann et al.,1998], but not vice-versa, provided byr4p is present [Krapp et al.,2008]. After the onset of anaphase B, the constellation of SIN proteins seen at the two SPBs is different. As the poles separate, the cdc7p signal at the old SPB grows fainter while the new SPB becomes brighter, reaching a maximum as the SPBs approach the cell tips [Sohrmann et al.,1998; Cerutti and Simanis,1999; Grallert et al.,2004; Garcia-Cortes and McCollum,2009]. The GAP (byr4p-cdc16p) is reconstituted first at the old SPB, and the intensity of the signals at the old SPB increase throughout anaphase [Cerutti and Simanis,1999; Li et al.,2000], mirroring the behavior of cdc7p at the new SPB. Indirect immunofluorescence has shown that byr4p becomes asymmetric before cdc7p [Cerutti and Simanis,1999]. There are ∼ 400 molecules of cdc7p at the SPB in late mitosis [Wu and Pollard,2005] and time-lapse analysis of cdc7p has shown that its removal from the new SPB correlates with closure of the CR [Garcia-Cortes and McCollum,2009].

The protein kinase sid1p and its regulatory subunit cdc14p appear at the new SPB only after the inactivation of CDK and onset of anaphase B [Guertin et al.,2000; Dischinger et al.,2008]. Like cdc7p, the signal increases in intensity during anaphase B spindle elongation, peaking as the SPBs approach maximal separation. Sid1p and cdc14p are interdependent for localization to the SPB, which also requires functional spg1 and cdc7 [Guertin et al.,2000]. Sid2p and mob1p are also interdependent for localization and associate with both SPBs throughout mitosis, and also with the medial region at the time of septum formation. The sid2p-mob1p complex requires functional SIN and CR components to localize to the medial region [Sparks et al.,1999; Hou et al.,2000; Salimova et al.,2000]. Full activation of sid2p also requires functional cdc7p [Sparks et al.,1999]. Sid2p and mob1p localization to the SPB is reduced in cdc7 mutants and some alleles of spg1 [Sparks et al.,1999; Hou et al.,2000; Salimova et al.,2000].

Biochemical analysis has shown that cdc16p, spg1p, mob1p, and cdc13p (the mitotic cyclin for cdc2p) all bind to the N-terminal domain of cdc11p [Morrell et al.,2004]. Maintaining the association of cdc7p and sid2p with the SPB in anaphase B requires etd1p function [Daga et al.,2005]. Studies of the localization and activity of sid2p have given rise to a model placing it downstream of cdc7p and sid1p [Sparks et al.,1999; Guertin et al.,2000]. However, given its symmetrical distribution during anaphase, it cannot be excluded that it plays different roles at the new and old SPBs. Immunoelectron microscopy has shown that in interphase, sid2p is associated with the cytoplasmic face of the SPB [Sparks et al.,1999]. It is not known whether the other SIN proteins also localize to the outer face of the SPB during mitosis. Taken together, these data have led to the idea that the order of action of the SIN proteins is cdc7p-spg1p, then sid1p-cdc14p, and finally sid2p-mob1p [Guertin et al.,2000].

Autonomous Assembly of the Cytokinetic Machinery

Overexpression of C-terminal truncated cdc12p, probably corresponding to a dominant active form lacking the auto-inhibitory domain, induces cdc15p- and CR-dependent cytokinesis even in interphase cells [Yonetani and Chang,2010]. Overexpression of cdc15p is able to induce actin rearrangements to the medial region in G2-arrested cells. Cdc15p may be negatively regulated by phosphorylation in interphase [Fankhauser et al.,1995]. Cdc15p can self-assemble and may interact directly with the plasma membrane through its F-BAR domain [Roberts-Galbraith et al.,2010]. Dephosphorylation of cdc15p promotes its oligomerization and increases its affinity for one its binding partners, cdc12p [Roberts-Galbraith et al.,2010]. In view of the fact that cdc12p physically associates with cdc15p [Carnahan and Gould,2003], high doses of these proteins may be possible to induce actin-polymerization beneath cell cortex and ectopic CR assembly bypassing mitosis-specific regulation including phosphorylation. Ectopic activation of the SIN in interphase will also promote CR assembly and septum formation [Schmidt et al.,1997], bypassing the mid1p spatial cue [Hachet and Simanis,2008].

Though whole genome-based screening shows that more than 200 proteins are localized to the division site in S. pombe, only a small subset of these can trigger reorganization of F-actin, CR formation and septation when their activity is altered [Matsuyama et al.,2006; reviewed by Bathe and Chang,2010]. These may define rate-limiting steps or control points in CR assembly analogous to the pivotal role played by the rhoA GTPase in animal-cell cytokinesis (see below).

General Overview of Cytokinesis in Eukaryotic Cells

In animal cells, cytokinesis is brought about by membrane ingression, which has been named the cleavage furrow (CF). The CF is thought to be induced by geometrical asymmetry of the cortical tension in the cell. Several mechanisms for CF formation have been proposed [reviewed by Wang,2005]; currently, the most favored one is the purse-string model [reviewed by Mabuchi and Itoh,1992; Noguchi et al.,2001; Pollard,2010]. The equatorial CR is composed of F-actin filaments and myosin II, and their interaction produces the force required for membrane ingression at the site of division. The concentration of F-actin at the CR seems to be particularly evident in nonmotile cells such as early embryos or yeasts. Though the F-actin at the CR is often parallel to the equator of the cell, there are also examples where cortical F-actin or myosin II are observed perpendicular to the division plane in some dividing cells [Fishkind and Wang,1993; DeBiasio et al.,1996; Oegema et al.,2000; Murthy and Wadsworth,2005; Chen et al.,2008b]. According to the purse-string model, these cortical actomyosin filaments should not contribute to constriction of the ring. Moreover, there are some cell types, especially those which are adherent to a substrate, which are able to divide without using a CR [Neujahr et al.,1997; O'Connell et al.,2001; Kanada et al.,2005]; in these cases, it seems that coordination of the compression of the equatorial cortex and cortical expansion at the cell poles is sufficient to induce a CF. Globally, it is likely that these mechanisms are not mutually exclusive but interdependent, and the extent to which each of them contributes to CF ingression differs according to the cell type. Thus, though cytokinesis is a fundamental cellular event, the mechanism that brings it about in any given cell-type will be a mixture of conserved and cell-type-specific features.

In animal cells, the central spindle and overlap region of astral microtubules (MTs), which extend from opposite spindle poles, promote CR assembly at telophase. Centralspindlin, a protein complex consisting of MKLP1, a kinesin-6 dimer, and Rho-family GTPase-activating protein (RhoGAP) subunits, plays the central role in this system [Mishima et al.,2002]. Cell cycle-dependent kinase 1 (CDK1), which promotes mitotic events such as centriole duplication, chromosome condensation, and spindle formation, phosphorylates MKLP1 during metaphase, which prevents it from interacting with MTs [Mishima et al.,2004]. After CDK1 is inactivated at the onset of anaphase, the protein-phosphatase CDC14 dephosphorylates MKLP1, and Centralspindlin moves on MTs and accumulates at their equatorial plus-ends. The POLO-like kinase, which regulates multiple mitotic events in addition to cytokinesis, induces association of ECT2, a guanine nucleotide-exchange factor for Rho GTPase, with Centralspindlin and localizes this protein complex to the equatorial region of the anaphase cell [reviewed by Petronczki et al.,2008; Wolfe et al.,2009]. As a result, Rho is specifically activated there and induces CR formation by promoting actin-polymerization and activation of myosin II via effectors such as formin and Rho-kinase. Furthermore, in the equatorial region of Caenorhabditis elegans oocyte Centralspindlin also reduces Rac GTPase-activity, which inhibits the formation of an Arp2/3-complex-dependent branched actin network and thereby ensures formation of the formin-dependent CR, which is composed of straight F-actin [Canman et al.,2008]. In addition to this, the tips of astral MTs which contact the cell poles relax cortical tension at the polar regions, which leads to a relative increase in equatorial contractility in C. elegans oocyte [reviewed by Werner and Glotzer,2008].

Cytokinesis is accomplished by the scission of a “bridge” connecting the daughter cells. Dynamic reorganization of membranes including scrambling of outer and inner sides of plasma membrane, SNARE–mediated vesicle fusions, and membrane scission by ESCRT complex is required for this step [reviewed by Prekeris and Gould,2008]. Vesicle fusion is particularly important for cytokinesis in higher plant cells where formation of the cell plate progresses centrifugally [reviewed by Jurgens,2005; Otegui et al.,2005; Dhonukshe et al.,2007]. Interestingly, an archaeal ortholog of Vps4, a protein stimulating ESCRT-mediated membrane scission, is required for cell division in the crenarchaeon Sulfolobus acidocaldarius [Lindas et al.,2008]. Therefore, it is possible that some molecular systems in cytokinesis may be conserved across evolutionary kingdoms; alternatively, their mechanistic similarity may reflect convergent evolution. Future study of the molecular basis for cytokinesis in various cells and organisms will doubtless help to answer this question.

Assembly and Arrangement of CR F-Actin in Fission Yeast

In this section, we will address two questions: first, what is the origin of CR F-actin? Second, is the CR formed by a spot/leading cable mechanism, or from a band of nodes? Considerable effort has been devoted to analysis of CR assembly [Vavylonis et al.,2008; Yonetani et al.,2008; Coffman et al.,2009]; reviewed by [Mishra and Oliferenko,2008; Roberts-Galbraith and Gould,2008; Bathe and Chang,2010; Pollard and Wu,2010]. These studies have given rise to two models for the initial assembly of CR F-actin in fission yeast. One is the node model, which assumes that short F-actin filaments are nucleated in random directions from about 65 nodes, each of which contains two cdc12p dimers [Wu and Pollard,2005], and then condensed into a continuous ring by a search, capture, pull and release (SCPR) mechanism that is driven by myosin II motor activity [Wu et al.,2006; Pollard,2008; Vavylonis et al.,2008; Pollard and Wu,2010]. The other is the spot model, which proposes that F-actin cables emanating from a single or several spot(s) of clustered cdc12p encircle the cell equator to form a ring [Chang et al.,1997; Chang,1999; Arai and Mabuchi,2002; Wong et al.,2002; Carnahan and Gould,2003; Motegi et al.,2004; Kamasaki et al.,2007]. These two models are consistent in that they agree that CR F-actin is polymerized de novo at mitosis (and does not arise from conversion of interphase actin cables or patches), but they are differ significantly with regard to how CR F-actin is formed: from dozens of cdc12p nodes or from a small number of cdc12p spots. Recent studies have shown that the cdc12p spot formation depends on its FH3 domain and the actin-bundling protein ain1p, and is dispensable for CR assembly [Yonetani et al.,2008; Coffman et al.,2009]. Thus, it is likely that recruitment of cdc12p to the medial region and concomitant initial CR F-actin assembly may be mediated through multiple distinct mechanisms operating individually or in parallel, reminiscent of the situation in mammalian cells, where multiple redundant mechanisms also operate (see above). Coffman et al. found that cdc12p appears in, at least, four distinct forms: speckle (one cdc12p dimer), node (two cdc12p dimers), spot (a probable cluster of cdc12p dimers) and ring [Coffman et al.,2009]. It is possible that the abundance of each form is dictated by the extent of aggregation of cdc12p. Analysis of how oligomerization of cdc12p affects its actin polymerization activity may help to elucidate the way in which multiple modes of initial CR F-actin assembly vary.

Constriction and Disassembly of CR and the Functions of Myosin II in CR Constriction

In the purse-string hypothesis, myosin II motors are presumed to generate a force driving CR constriction via interaction with CR F-actin. Inhibition of the ATPase activity of myosin II using blebbistatin blocks cytokinesis in mammalian cells without affecting CR F-actin assembly [Straight et al.,2003]. Interestingly, studies using blebbistatin also revealed that myosin II is involved in turnover of CR F-actin [Guha et al.,2005; Murthy and Wadsworth,2005].

S. pombe has two myosin II isoforms, myo2p and myo3p (also called myp2p), both of which localize to the CR [Bezanilla and Pollard,2000]. Myo2p is essential for cytokinesis, while myo3p is required for growth under stress conditions. Localization of the myosins to cortical nodes and the CR has been analyzed extensively [Naqvi et al.,1999; Motegi et al.,2000; Mulvihill et al.,2001; Wu et al.,2003; Motegi et al.,2004; Martin-Garcia and Valdivieso,2006; Sladewski et al.,2009]. The precise function of myosin II in S. pombe cytokinesis remains unclear. F-actin accumulates at the division site in myo2-null cells but fails to form a ring [Kitayama et al.,1997]. Biochemical experiments using purified myo2p revealed that a G345R mutation (the temperature-sensitive myo2-E1 allele), significantly reduces the actin-binding and motor activity of myo2p [Lord and Pollard,2004]. In the myo2-E1 mutant cells, node condensation into a ring abolished at restrictive temperature and is delayed at the permissive temperature [Coffman et al.,2009]. Moreover, CRs in myo2-E1 cells constrict twofold slower than those in wild-type cells [Stark et al.,2010]. These observations are consistent with the SCPR model in that myosin II functions both in assembly of the F-actin ring and contraction of the CR. Myo3p does not appear as an equatorial band of nodes in early mitosis and joins the CR after myo2p has formed a ring [Wu et al.,2003]; this may indicate that myo3p plays an important role in CR constriction. Alternatively, since myo3p associates physically with chitin-synthase, it may contribute to CR closure via septum formation [Martin-Garcia and Valdivieso,2006]. Since a null-mutation of myo3 exacerbates the phenotype of myo2-E1 cells, the two myosin-IIs seem to share overlapping functions in cytokinesis [Motegi et al.,2000].

Folding of the myosin head domain requires a co-chaperone of the (UNC-45/CRO1/She4p) family of proteins. In C. elegans, UNC-45 colocalizes with myosin II during cytokinesis in the early embryos [Barral et al.,2002; Kachur et al.,2004] and is required for assembling thick filaments of myosin II in sarcomeres [reviewed by Kachur and Pilgrim,2008]. S. pombe rng3p belongs to the same UCS -family and associates with myo2 and other four myosin proteins cotranslationally [Amorim and Mata,2009]. Rng3p is involved in CR formation cooperatively with myo2p [Wong et al.,2000] and promotes motor activity of myo2p for F-actin [Lord and Pollard,2004; Lord et al.,2008]. Rng3p localizes to the CR just before CR constriction in wild-type cells [Lord and Pollard,2004], suggesting that rng3p may induce myo2p-dependent CR constriction. Interestingly, rng3p is concentrated in nodes at metaphase in myo2-E1 cells [Lord et al.,2008].

Functions of Actin-Depolymerizing Factor in Cytokinesis

FRAP analysis has shown that CR F-actin and cortical F-actin turn over rapidly both in fission yeast and mammalian cells [Pelham and Chang,2002; Guha et al.,2005; Murthy and Wadsworth,2005]. The actin-depolymerizing factor (ADF)/cofilin-family (hereafter refer to as ADF) proteins facilitate depolymerization and severing of actin filaments and hence play a key role in regulation of F-actin dynamics in various motile processes; reviewed by [Pantaloni et al.,2001]. In some animal cells, ADF proteins are required for completion of cytokinesis, but not for assembly of the CR [Gunsalus et al.,1995; Somma et al.,2002; Kaji et al.,2003; Ono et al.,2003]. It is likely that these ADFs contribute to turnover of CR F-actin or its active disassembly.

In S. pombe cells, the sole ADF, named as adf1p, localizes to the CR and is essential for accumulation of F-actin at the division site during cytokinesis [Nakano and Mabuchi,2006]. This is consistent with the observation that Xenopus ADF/cofilin, XAC, localizes to the leading edge of the CF in a fertilized egg and also participates in its formation [Abe et al.,1996]. Complementation-assays in fission yeast using mutant ADFs suggest that the actin-severing activity of adf1p may primarily contribute to its function in cytokinesis rather than its actin-depolymerizing activity. Actin-severing increases number of filament ends, which is required to induce dynamic turnover of actin subunits, since internal subunits are never exchanged in actin filaments. Andrianantoandro and Pollard, however, revealed that high concentration of adf1p nucleates actin polymerization whereas it effectively severs F-actin at lower concentrations [Andrianantoandro and Pollard,2006]. Thus, in fission yeast, adf1p would promote turnover of CR F-actin by a combination of its actin-severing and -nucleation activities. Another F-BAR protein, imp2p and a paxillin-related protein, pxl1p may also be implicated in CR disassembly or constriction [Demeter and Sazer,1998; Ge et al.,2005; Pinar et al.,2008], though their mechanism of action in this context remains unclear.

Possible Functions of CR and Mechanisms for Its Constriction

It is unequivocal that in dividing animal cells net inward forces are generated along the equator, whether by the CR or by other mechanisms, and they drive the constriction of the medial cortex [Rappaport,1967; Hiramoto,1975]. On the other hand, contribution of CR to the generation of contractile force is elusive in fission yeast as discussed above. In the budding yeast, S. cerevisiae, the CR is not essential for cytokinesis, and septation alone manages to carry out the closure of bud neck [Bi et al.,1998]. The SIN pathway is not only required for localization of β-1,3-glucan synthase, bgs1p, at the division plane for formation of the primary septum [Liu et al.,2002], but is also essential for maturation of CR [Hachet and Simanis,2008]. Hyperactivation of the SIN pathway induces ectopic F-actin ring formation [Schmidt et al.,1997]. Taken together, CR constriction and septation would be interdependent in both mechanical and signaling aspects, and it may be difficult to distinguish separately their functions in cytokinesis. We believe, however, that fission yeast CR shares enough functional molecular bases with those of animal cells to serve as a model system.

Whatever the actual function of CR constriction is, its mechanism(s) is still an open question. As discussed above, the purse-string model should be challenged by examining whether myo2p assembles into minifilaments in vitro. Fine and time-resolved EM observation of CR and the medial region of dividing fission yeast cells may also help to address this issue. An attractive model for generation of the contractile force has been proposed, which posits that polymerization and depolymerization of CR F-actin can produce mechanical force for the ring contraction in the presence of end-tracking crosslinkers even without myosin II motors [Zumdieck et al.,2007]. In this model, tracking of a depolymerizing filament end by the crosslinker, which also binds to the second filament, is harnessed to generate sliding force between the two filaments. At present, the model remains to be validated, but it clearly seems worth pursuing, given the general and critical importance of both ADFs and formin proteins in cytokinesis.