Contractile-Ring Assembly in Fission Yeast Cytokinesis: Recent Advances and New Perspectives

Abstract

The fission yeast Schizosaccharomyces pombe is an excellent model organism to study cytokinesis. Here, we review recent advances on contractile-ring assembly in fission yeast. First, we summarize the assembly of cytokinesis nodes, the precursors of a normal contractile ring. IQGAP Rng2 and myosin essential light chain Cdc4 are recruited by the anillin-like protein Mid1, followed by the addition of other cytokinesis node proteins. Mid1 localization on the plasma membrane is stabilized by interphase node proteins. Second, we discuss proteins and processes that contribute to the search, capture, pull, and release mechanism of contractile-ring assembly. Actin filaments nucleated by formin Cdc12, the motor activity of myosin-II, the stiffness of the actin network, and severing of actin filaments by cofilin all play essential roles in contractile-ring assembly. Finally, we discuss the Mid1-independent pathway for ring assembly, and the possible mechanisms underlying the ring maturation and constriction. Collectively, we provide an overview of the current understanding of contractile-ring assembly and uncover future directions in studying cytokinesis in fission yeast.

Introduction

Cytokinesis partitions a mother cell into two daughter cells at the end of each cell cycle. Failure in cytokinesis results in tetraploidy and contributes to tumorigenesis [Fujiwara et al., 2005; Ganem et al., 2007; Li et al., 2010; Sagona and Stenmark, 2010]. The actomyosin contractile ring in cytokinesis is conserved from fungi to humans [Pollard and Wu, 2010]. Research using the fission yeast Schizosaccharomyces pombe as a model system has provided novel insights into contractile-ring assembly. Briefly, contractile-ring assembly in fission yeast begins when Mid1, the anillin-like protein, recruits other proteins to assemble cytokinesis nodes [Paoletti and Chang, 2000; Wu et al., 2003; Motegi et al., 2004; Laporte et al., 2011; Padmanabhan et al., 2011]. The nodes later condense into a compact actomyosin contractile ring through a process that has been described as a search, capture, pull, and release (SCPR) mechanism, which depends on transient interactions between the myosin-II motors and linear actin filaments [Vavylonis et al., 2008]. The contractile ring then matures by recruiting additional components before its constriction [Wu et al., 2003]. Although contractile-ring assembly can be described by a seemingly simple model, the mechanisms and regulation underlying each step of the assembly process are complex and involve many proteins. Therefore, contractile-ring assembly in fission yeast has been an active field of research. Recent studies in fission yeast cytokinesis not only shed light on the assembly and architecture of cytokinesis nodes, but also elucidate several proteins’ contributions to the SCPR mechanism. Here, we review our current understanding of contractile-ring assembly, and discuss future perspectives in investigating cytokinesis in fission yeast.

Cytokinesis Node Assembly

In fission yeast, the anillin-like protein Mid1 specifies the division site [Chang and Nurse, 1996; Sohrmann et al., 1996; Bähler et al., 1998; Paoletti and Chang, 2000; Celton-Morizur et al., 2004; Daga and Chang, 2005; Almonacid et al., 2009]. In interphase, Mid1 localizes to both the nucleus and the plasma membrane at the equator of the cell [Sohrmann et al., 1996; Bähler et al., 1998; Paoletti and Chang, 2000]. At the G2/M transition, Mid1 is mainly concentrated on the medial cortex and joined by several other proteins to form an equatorial band of ~65 cortical punctate structures named cytokinesis nodes that have a Gaussian distribution along the long axis of the cell [Paoletti and Chang, 2000; Wu et al., 2003, 2006; Vavylonis et al., 2008]. These cortical nodes later coalesce into the contractile ring [Bähler et al., 1998; Wu et al., 2003, 2006], suggesting that these macromolecular cortical complexes (on the cytoplasmic side of the plasma membrane) are precursors of the contractile ring in fission yeast cells (Fig. 1A). Proteins in cytokinesis nodes include the anillin-like protein Mid1 [Bähler et al., 1998; Paoletti and Chang, 2000], the IQGAP protein Rng2 [Eng et al., 1998], the myosin-II motor (heavy chain Myo2, essential light chain Cdc4, and regulatory light chain Rlc1) [McCollum et al., 1995; Kitayama et al., 1997; May et al., 1997; Naqvi et al., 1999, 2000; Bezanilla et al., 2000; Motegi et al., 2000, 2004], the F-BAR protein Cdc15 [Fankhauser et al., 1995; Carnahan and Gould, 2003], and the formin Cdc12 [Chang et al., 1997; Wu et al., 2006; Coffman et al., 2009]. Their localizations to the nodes are actin-independent [Motegi et al., 2000; Paoletti and Chang, 2000; Wu et al., 2003, 2006]. In mid1Δ cells, equatorial cytokinesis nodes do not form, and cells are severely defective in division-site selection and contractile-ring assembly [Wu et al., 2003, 2006]. Mid1 phosphorylation by Polo kinase Plo1 is crucial for the initiation of cytokinesis-node assembly [Bähler et al., 1998; Almonacid et al., 2011] (for a detailed review of the regulation of Mid1 and division-site selection, see the review by Rincon and Paoletti [2012] in this issue).

Fig. 1. The assembly of cytokinesis nodes and the contractile ring in fission yeast.

(A) Contractile-ring assembly and cytokinesis in fission yeast. Interphase nodes are orange. Cytokinesis nodes and the contractile ring are red. Nuclei are blue. Actin filaments are green. (B) The assembly hierarchy of cytokinesis nodes. Hypothetical protein shapes/structures used in (C) are shown close to each protein name. Both the F-BAR protein Cdc15 and the Rng2-Cdc4 module (gray box) could recruit the formin Cdc12. (C) The architecture of a cytokinesis node. Yellow filaments are F-actin nucleated by Cdc12. PM, plasma membrane. (B and C) Modified from Laporte et al. [2011].

Although Rng2, Cdc4, Myo2, Rlc1, Cdc15, and Cdc12 only colocalize with Mid1 beginning at G2/M, several proteins colocalize with Mid1 on the cortex during most of interphase [Moseley et al., 2009]. The interphase cortical structures that contain Mid1 and these proteins are hence named “interphase nodes,” which determine cell size and mitotic entry together with Pom1 kinase [Morrell et al., 2004; Martin and Berthelot-Grosjean, 2009; Moseley et al., 2009; Hachet et al., 2011]. Note that in wild type cells, the different nomenclature simply reflects the difference of node components in regard to cell cycle stage. Here, we discuss the connections between interphase and cytokinesis nodes and review recent advances on the assembly of cytokinesis nodes.

Cytokinesis Nodes and Interphase Nodes

Proteins in interphase nodes include three kinases Cdr2 [Breeding et al., 1998; Kanoh and Russell, 1998; Morrell et al., 2004], Cdr1 [Young and Fantes, 1987; Feilotter et al., 1991], and Wee1 [Parker et al., 1991; Wu et al., 1996; Masuda et al., 2011], kinesin-like protein Klp8, the putative Rho guanine nucleotide exchange factor (GEF) Gef2, and novel protein Blt1 [Martin and Berthelot-Grosjean, 2009; Moseley et al., 2009]. In mitosis, Klp8, Gef2, and Blt1 persist in cytokinesis nodes, but are not essential for contractile-ring assembly because without interphase nodes, Mid1 can still localize to the medial cortex at the G2/M transition and assemble cytokinesis nodes independently [Almonacid et al., 2009]. Nevertheless, evidence suggests that interphase node proteins can contribute to the regulation of cytokinesis [Almonacid et al., 2009; Laporte et al., 2011; Ye et al., 2012].

Cdr2, the SAD/GIN4 kinase in fission yeast, is the organizer of interphase nodes and involved in regulating cell size by inhibiting Wee1 kinase [Martin and Berthelot-Grosjean, 2009; Moseley et al., 2009]. Cdr2 is detected in condensing cytokinesis nodes but disappears from the division site shortly after the assembly of a compact contractile ring [Martin and Berthelot-Grosjean, 2009; Moseley et al., 2009]. Interestingly, Mid1, IQGAP Rng2, and F-BAR protein Cdc15 nodes are more dynamic in the absence of Cdr2, indicating that interphase node proteins play a role in stabilizing cytokinesis nodes [Laporte et al., 2011].

Gef2, Blt1, and Klp8 stay in the contractile ring until the end of ring constriction [Moseley et al., 2009; Ye et al., 2012]. Although the function of Klp8 remains unknown, strong synthetic genetic interactions in gef2Δ plo1-ts18 cells and quantitative microscopy indicate that Gef2 and Polo kinase Plo1 are required together to regulate Mid1 cortical level in cytokinesis. blt1Δ also shows mild genetic interaction with plo1-ts18, and its involvement in cytokinesis is partially through Gef2 [Ye et al., 2012]. The functions of Gef2 and Blt1 in interphase are not clear yet, although gef2Δ and blt1Δ cells are slightly longer than wild type cells [Moseley et al., 2009; Ye et al., 2012].

It seems surprising that the cell-size sensing machinery and the precursors of the contractile ring are colocalized, but this arrangement could promote timely assembly of the cytokinesis apparatus once a cell is committed to undergo mitosis. Nodes are not the only cortical structure observed in S. pombe. Rga4, the Rho GTPase (guanosine triphosphatase)-activating protein that inhibits growth on cell sides, localizes to punctate-like structures on the cortex [Das et al., 2007; Tatebe et al., 2008] that are distinct from nodes (our unpublished data). Eisosomes in fission yeast assemble into filaments on the cortex [Kabeche et al., 2011]. Clustering of the plasma membrane components may be the origin of the distinction between these different structures [Wachtler et al., 2003; Morrell et al., 2004; Takeda and Chang, 2005]. Whether cross-talks exist between these structures requires further investigation.

The Assembly and Architecture of Cytokinesis Nodes

Spindle pole bodies (SPB) are the yeast counterparts of animal centrosomes. SPB separation marks the onset of mitosis in fission yeast. All the cytokinesis node proteins except Mid1 arrive at nodes shortly before or around SPB separation [Wu et al., 2003; Laporte et al., 2011]. Recently, several studies have determined the hierarchy of cytokinesis-node assembly using complementary methods [Almonacid et al., 2011; Laporte et al., 2011; Padmanabhan et al., 2011]. First, localization dependencies of cytokinesis proteins at nodes were determined by using temperature-sensitive mutants and germinated spores of null mutants (Fig. 1B). Second, the appearance of the proteins at nodes was imaged and quantified with high temporal resolution. Third, their dynamics on nodes were obtained by fluorescence recovery after photobleaching (FRAP). Fourth, Mid1 phosphorylation and the physical interactions between Mid1 and other cytokinesis proteins were determined. Last but not least, the architecture of nodes was revealed by a modified single-molecule high-resolution colocalization (SHREC) technique that breaks the diffraction limit and results in a resolution of tens of nanometers [Churchman et al., 2005; Joglekar et al., 2009]. The architecture of cytokinesis nodes obtained is summarized in Fig. 1C.

These studies highlight the importance of the IQGAP protein Rng2 and Cdc4, which are the earliest to appear at nodes (~12 min before SPB separation) after Mid1. Except Mid1, other cytokinesis node proteins are not required for Rng2 and Cdc4 to localize. However, without functional Rng2 and Cdc4, Myo2 and Rlc1 cannot localize to nodes. Rng2 interacts with Mid1 in co-immunoprecipitation (co-IP) [Laporte et al., 2011; Padmanabhan et al., 2011], and rng2-M1 (with mutations H1329L and K1366E in the region interacting with Mid1) phenocopies mid1Δ [Padmanabhan et al., 2011]. The C-terminus (aa 1306–1489) of Rng2 interacts with the N-terminus (first 100 aa) of Mid1 in an in vitro binding assay [Almonacid et al., 2011], suggesting that Rng2 is directly recruited by Mid1. The N-terminus (aa 1-100) of Mid1, when targeted to nodes, is sufficient to assemble cytokinesis nodes and the contractile ring [Lee and Wu, 2012]. Cdc4 forms a complex with Mid1 in co-IP, but it is unknown if the interaction is direct or not. Rng2 and Cdc4 interact with each other via the IQ motifs in Rng2 [D’Souza et al., 2001], and their localizations to cytokinesis nodes are interdependent [Laporte et al., 2011; Padmanabhan et al., 2011]. Taken together, Rng2 and Cdc4 are more upstream in the node-assembly pathway but whether they are recruited as a complex is unknown.

Cdc4 has two recovery rates in FRAP analysis. Although the slow recovery resembles that of Rng2, the fast recovery depends on its interaction with the IQ motif of Myo2 [Naqvi et al., 2000; D’Souza et al., 2001; Laporte et al., 2011]. Consistent with the timing of appearance (~10 min before SPB separation) and the localization dependency, the weak interaction between endogenous Myo2 and Mid1 revealed by co-IP depends on functional Rng2 [Laporte et al., 2011]. Results from SHREC also suggest that Myo2 is further away from the plasma membrane compared to Mid1 and Rng2, with its motor head pointing into the cytoplasm and C-terminal tail folded. Rlc1, the regulatory light chain of myosin-II, displays the same timing of appearance and dynamics as Myo2, and its localization completely depends on Myo2 [Laporte et al., 2011].

F-BAR protein Cdc15 arrives at nodes ~5 min before SPB separation [Laporte et al., 2011]. The dephosphorylation of Cdc15, partly regulated by the Cdc14-family phosphatase Clp1 [Trautmann et al., 2001; Wolfe and Gould, 2004; Wolfe et al., 2006], is crucial for its division-site localization, conformation, and interactions [Wachtler et al., 2006; Clifford et al., 2008; Roberts-Galbraith et al., 2010]. Prematurely dephosphorylated Cdc15 localizes to cortical nodes and is able to cause medial localization of its interacting partners in interphase [Roberts-Galbraith et al., 2010]. Some discrepancy exists regarding the relationship between Mid1, Rng2, and Cdc15 (Fig. 1B). Physical interactions are reported between Mid1-Cdc15 (yeast two hybrid and co-IP assays; [Laporte et al., 2011]), Mid1-Rng2 (co-IP and in vitro binding assay using purified fragments; [Almonacid et al., 2011; Laporte et al., 2011; Padmanabhan et al., 2011]), and Cdc15-Rng2 (tandem affinity purification [TAP]- and co-IP; [Roberts-Galbraith et al., 2010]). On the one hand, Laporte et al. [2011] proposed that Mid1 can recruit Cdc15 to cytokinesis nodes independent of the Rng2-Cdc4 module, because Cdc15 nodes are detected in rng2-D5, rng2-346, rng2Δ, cdc4-8, and cdc4Δ mutants. On the other hand, Padmanabhan et al. proposed that Cdc15 is downstream of the Rng2-Cdc4 module because no Cdc15 nodes are detected in the rng2-M1 mutant at the restrictive temperature. It is possible that part of the discrepancy comes from the difference between rng2 mutants [Padmanabhan et al., 2011]. Indeed, the total Mid1 protein level is lower in rng2-M1 but not in rng2-D5 cells [Laporte et al., 2011; Padmanabhan et al., 2011]. Another possibility is that the higher autofluorescence at the GFP (green fluorescent protein) channel compared to the YFP (yellow fluorescent protein) channel might obscure a weak cortical Cdc15 signal in rng2-M1 [Padmanabhan et al., 2011]. Cdc15 levels in the nodes have not been quantified in the cdc4 and rng2 mutants [Laporte et al., 2011], so the possibility remains that both Mid1 and Rng2 are involved in recruiting Cdc15. Further studies are needed to address these different possibilities.

The last known component to join cytokinesis nodes before their condensation is the formin Cdc12, whose interaction with Cdc15 is well studied [Carnahan and Gould, 2003; Roberts-Galbraith et al., 2010]. Surprisingly, it was recently found that in addition to Cdc15, the Rng2-Cdc4 module can also recruit Cdc12 to cytokinesis nodes [Laporte et al., 2011] (Fig. 1B). Cdc12 is dispensable for other cytokinesis node proteins to localize [Laporte et al., 2011], although its function to nucleate actin filaments is essential for contractile-ring assembly [Kovar et al., 2003]. Soon after Cdc12 appears at the division site, cytokinesis nodes start to condense and the contractile ring assembles in ~10 min in wild type cells at 25°C.

SCPR Model and Beyond

Since the proposal that the contractile ring in fission yeast cells is assembled from the condensation of a broad band of nodes [Bähler et al., 1998; Wu et al., 2006], numerical simulations and live-cell imaging of cytokinesis node proteins and actin filaments were integrated to describe the mechanism. Monte Carlo simulations, using parameters obtained in vivo, recapitulate the condensation of nodes via transient connections between actin filaments and neighboring nodes. Vavylonis et al. [2008] therefore proposed the SCPR mechanism of contractile-ring assembly: an actin filament nucleated by formin Cdc12 in one node searches the cortex and can be captured by the myosin-II at another node, and the force generated by the myosin-II motor walking on the actin filament pulls the nodes closer before the release of the interaction. The SCPR model differs from the previously proposed spot/leading cable model for contractile-ring assembly in the numbers of actin-nucleation sites, orientations of actin filaments, and the importance of myosin-II motor activity [Chang et al., 1997; Chang, 1999; Arai and Mabuchi, 2002; Carnahan and Gould, 2003; Kamasaki et al., 2007; Mishra and Oliferenko, 2008; Coffman et al., 2009]. Key assumptions and predictions in the SCPR model were tested subsequently in a number of studies (see below). Consistent with the model, many different perturbations of contractile-ring assembly result in discontinuous aggregates (clumps) on the cortex rather than a continuous ring. Meanwhile, in vivo observations have led to the refining of the model [Ojkic et al., 2011; Laporte et al., 2012]. Here, we review the process of SCPR, and discuss players that contribute to each step of contractile-ring assembly (Fig. 2).

Fig. 2. The SCPR mechanism of contractile-ring assembly.

(A–D) Two nodes are pulled closer during the search (A), capture (B), pull (C), and release (D) process. Shaded areas (light blue) indicate the following events: (1) formin Cdc12 nucleates and elongates an actin filament from profilin-bound monomers. (2) A myosin-II motor captures the actin filament. (3) Tropomyosin stabilizes the actin filament. (4) Myosin-II motor activity induced by the UCS protein Rng3 pulls two nodes closer. (5) The crosslinking by α-actinin resists the movement. (6) The actin filament dissociates from myosin-II. (7) The actin filament is severed by cofilin. (8) The actin filament is capped at its barbed end by capping protein. (9) Cdc12 dissociates from the rest of the node.

Search

After the nodes have matured by the recruitment of the formin Cdc12, actin polymerization is crucial for the “search” step. Cdc12 is the essential formin that nucleates actin filaments at the division site [Chang et al., 1997; Kovar et al., 2003; Coffman et al., 2009]. Therefore, its activity and localization are critical for contractile-ring assembly. As previous studies suggest that the majority of actin filaments for contractile-ring assembly are nucleated by Cdc12 at the division site [Pelham and Chang, 2002; Coffman et al., 2009], actin filaments nucleated elsewhere in cells are not included in the SCPR simulation. Nevertheless, current data do not exclude the possibility that these filaments could be incorporated into the contractile ring.

The number of actin nucleation sites in the broad band of nodes determines the success and efficiency of contractile-ring assembly. Assuming there are ~65 nodes in each cell, the SCPR model requires that at least 50% of them should contain formins and nucleate 1–4 filaments from each node in order for the nodes to condense into a ring in ~10 min. In agreement with the model, 2–4 dimers of Cdc12 localize to >50% of nodes right before the nodes start to condense [Coffman et al., 2009; Laporte et al., 2011]. The high nucleation efficiency (~1 filament out of 3 dimers) of purified Cdc12 FH1FH2 domain in vitro [Scott et al., 2011] supports both the SCPR model and the in vivo data [Coffman et al., 2009; Laporte et al., 2011].

The orientation and elongation rate of each actin filament determines its chance to encounter another node. On the cell cortex, actin filaments are randomly oriented at the beginning of node condensation [Coffman et al., 2009], as applied in the SCPR model. Further study showed that actin filaments at the division site exhibit an average angle of 8° to the plasma membrane, possibly due to the position and orientation of Cdc12 in nodes, affinity of actin filaments with the plasma membrane, or restriction by the endoplasmic reticulum [Zhang et al., 2010; Laporte et al., 2011]. This angle may ensure that the actin filaments can be readily captured (see below). The length of actin filaments is controlled by actin-binding proteins and the processivity of Cdc12. Profilin Cdc3 [Balasubramanian et al., 1994] is required for the rapid elongation of actin filaments by Cdc12 [Kovar et al., 2003] (Fig. 2A). In the presence of profilin, Cdc12 associates with elongating actin filaments processively [Kovar and Pollard, 2004] and competes with capping protein better than other formins [Neidt et al., 2008]. Tropomyosin Cdc8 [Balasubramanian et al., 1992] helps the elongation of actin filaments by inhibiting disassembly [Skau et al., 2009]. The binding of tropomyosin to actin and thus its localization is regulated by acetylation [Skoumpla et al., 2007; Coulton et al., 2010].

Of note, the regulation of Cdc12 activity remains largely unknown. No Rho GTPase has been identified to activate Cdc12, although many diaphanous-related formins are regulated in this way. Because cdc12 is an essential gene, domain analyses were performed in the presence of the endogenous protein, adding complexity to the interpretation of the results [Yonetani et al., 2008; Yonetani and Chang, 2010]. The formation of interphase rings when a C-terminal Cdc12 truncation is overexpressed in the presence of endogenous Cdc12 is suggestive of some form of inhibition that acts on the long C-terminal tail of Cdc12 [Yonetani and Chang, 2010]. A formin damper or inhibitor, such as Smy1 or Bud14 in S. cerevisiae [Chesarone et al., 2009; Chesarone-Cataldo et al., 2011], has not been found yet in S. pombe. The temperature-sensitive mutant cdc12-112 forms many small clumps over the equator, consistent with the results of SCPR when actin filaments are too short [Hachet and Simanis, 2008; Ojkic and Vavylonis, 2010]. Because deletion of capping protein rescues the cdc12-112 phenotype, it has been suggested that this formin mutant lacks processivity [Kovar et al., 2005], which would result in shorter actin filaments and make capture events less likely.

Capture

When an actin filament nucleated from a node encounters another node (in the SCPR model, when the filament comes within the capture radius 100 nm from the centroid of the node), it might be captured by myosin-II (Fig. 2B). Tropomyosin Cdc8 stabilizes actin filaments (Fig. 2B) and increases the affinity of myosin-II for actin filaments [Stark et al., 2010]. It has been suggested that with the motor head of myosin-II being oriented away from the cortex in stationary nodes, it is more likely to catch the slightly inward-pointing actin filaments [Laporte et al., 2011]. Increasing myosin-II concentration speeds up contractile-ring assembly maybe partly by increasing capture events, or by producing more force and thus pulling the nodes together more quickly [Stark et al., 2010].

Other actin-binding proteins in nodes, such as Rng2 [Eng et al., 1998], could also capture actin filaments [Takaine et al., 2009], but the interaction would not result in shortening of the distance between nodes without the myosin-II motor activity. The interaction between Rng2 and actin filaments may generate tension and pull them closer to the myosin motor heads and thus increase the chance of capture by myosin-II. On the other hand, Rng2-actin interaction might interfere with the SCPR mechanism if actin filaments are stabilized but not directed to myosin-II. A Rng2 truncation lacking the actin-binding Calponin Homology Domain would be helpful for further analysis of its function and the importance of its actin-binding activity. Because Rng2 is essential, such a truncation may not be viable, although it would not be expected to affect the node assembly pathway [Laporte et al., 2011].

In the original SCPR model [Vavylonis et al., 2008], actin filaments stop growing once they are captured by neighboring nodes, and the tension-induced switch-off is important for the mechanism. Whether actin filaments stop growing after being captured in vivo remains untested.

Pull

A key assumption in the SCPR model is that the force generated by the myosin-II motor on the captured actin filament pulls the two nodes closer to each other. It predicts that mutants with defective myosin-II motor activity cannot condense the nodes properly. Indeed, Coffman et al. [2009] showed that when myo2-E1, a temperature-sensitive mutant with weakened myosin-II motor activity, is grown at the restrictive temperature, actin filaments are nucleated at the division site and associated with cytokinesis nodes, but node condensation is severely affected. Phosphorylation of the regulatory light chain was suggested to regulate myosin-II motor activity [Chew et al., 1998; Sanders et al., 1999; Loo and Balasubramanian, 2008]. Indeed, mutating phosphorylation sites on Rlc1 to alanine results in lower Myo2 motility in vitro and a delay in contractile-ring assembly at higher temperatures in vivo [Sladewski et al., 2009] (see below for discussion). The fission yeast UCS protein Rng3 [Wong et al., 2000] activates the motor activity of myosin-II (Fig. 2C) [Lord and Pollard, 2004; Lord et al., 2008]. When the temperature-sensitive rng3-65 cells are grown at the restrictive temperature, the movement of nodes is minimal; when the cells are shifted back to the permissive temperature, node condensation is recovered [Coffman et al., 2009]. Thus, myosin-II motor activity is essential for node condensation into the contractile ring.

If actin filaments are crosslinked into a network (Fig. 2C), then the pulling forces of myosin motors will be distributed to all the other nodes that are connected together through actin filaments. Although this could help coordinate node condensation, crosslinking may also bundle actin filaments along unproductive directions. Recent studies highlight the importance of actin-crosslinking proteins in cytokinesis. When semiflexible actin filaments are crosslinked, the stiffness of the network increases, thus making node condensation more difficult. α-Actinin forms a homodimer via its spectrin repeats and bundles actin filaments [Xu et al., 1998; Djinovic-Carugo et al., 1999]. In mammalian cells, the crosslinking by α-Actinin is required for structural support of the actin network during cytokinesis, because the depletion of α-Actinin results in a sudden collapse of the equatorial cortical network [Mukhina et al., 2007].

Ain1, the α-Actinin in S. pombe, also localizes to the contractile ring [Nakano et al., 2001; Wu et al., 2001]. Although both deleting and overexpressing ain1 cause delays in contractile-ring assembly, live-cell imaging shows completely opposite behavior of actin filaments and cytokinesis node proteins in these strains [Laporte et al., 2012]. In ain1Δ, the actin network becomes more dynamic and cytokinesis nodes usually first condense into a clump before a contractile ring is eventually formed. This is likely due to excess net pulling force because of less resistance via crosslinking. When excess Ain1 is present in the cell, the movement of nodes is attenuated and actin filaments form stable linear structures that may or may not assemble into a contractile ring. A balance between myosin pulling force and the damping effect of crosslinkers is reestablished when Myo2 is slightly overexpressed in these cells, which leads to successful contractile-ring assembly [Laporte et al., 2012]. These results suggest that the extent of crosslinking is critical for proper contractile-ring assembly. In addition to Ain1, Rng2 has also been shown to bundle actin filaments [Takaine et al., 2009]. In contrast, Cdc12 has no bundling activity [Scott et al., 2011]. Recent evidence suggests that actin crosslinker fimbrin Fim1 functions as a subsidiary to Ain1 during contractile-ring assembly, although Ain1 has a more prominent role during node condensation, probably due to the difference in the geometry/distance of the crosslinked actin filaments [Laporte et al., 2012].

Release

Permanent interactions between nodes result in a series of clumps instead of a ring at the equator in the simulations of the SCPR model. Therefore, the release of interactions is critical for contractile-ring assembly. In the SCPR model, the release between two nodes happens with a constant rate [Vavylonis et al., 2008]. In vivo, several factors could contribute to the release of the interaction between two nodes (Fig. 2D). First, Myosin-II could dissociate from the captured actin filament. Myosin-II is a motor with low duty ratio and <15% of Myo2 is in the strong actin-bound state in vitro [Stark et al., 2010]. Second, actin filaments could be severed by cofilin. Several cofilin mutants that are defective in severing result in clump formation [Nakano and Mabuchi, 2006; Chen and Pollard, 2011]. Third, the formin Cdc12 could be displaced from the barbed ends of actin filaments. Although Cdc12 tightly binds barbed ends and competes with capping protein in vitro [Kovar et al., 2005], their relationship in vivo is less clear. In S. cerevisiae, Bud14 can displace formin Bnr1 from barbed ends that are immediately capped [Chesarone et al., 2009], but an S. pombe homolog to Bud14 is unknown. Last, part of the node could dissociate because of its intrinsic dynamics. Several cytokinesis proteins are dynamic in the nodes and recover quickly in FRAP analysis [Coffman et al., 2009; Laporte et al., 2011], and when Cdc12 or Myo2 in particular dissociates from the node (half times ~30 s), the tension between two nodes may be released. All of these factors could contribute to release during contractile-ring assembly, but their relative contributions are unknown.

The frequency of the release of the interaction appears to be tightly controlled in vivo. One example of the regulation of severing was discovered by comparing the level of fimbrin Fim1 at different locations. Fim1 localizes to both actin patches and the contractile-ring, with a higher concentration in patches. A recent study showed that in fission yeast cells, Fim1 and tropomyosin Cdc8 have antagonistic roles in cytokinesis, and the fimbrin-bound actin filaments are more susceptible to severing by cofilin [Skau and Kovar, 2010] due to the loss of protection by tropomyosin. This result explains why overexpression of Fim1 abolishes contractile-ring assembly [Wu et al., 2001]. The binding of Rng2 to the actin filament also protects it from being severed by cofilin in vitro [Takaine et al., 2009].

The unconnected filaments also undergo breakage [Vavylonis et al., 2008]. Recently, it was reported that tension created by the optical tweezer prevents actin filaments from being severed by cofilin [Hayakawa et al., 2011]. In the reconstituted system, the lifetime of actin filaments under tension are about twice as long as the relaxed ones [Hayakawa et al., 2011]. It is likely that the unconnected filaments in fission yeast cells are also more prone to be severed compared to the connected ones.

In the SCPR model, once an actin filament breaks, another one immediately grows out from the node to start the search again. Although this assumption has not been directly tested yet, the high nucleation efficiency of Cdc12 [Scott et al., 2011] is indicative of a sufficiently short interval between each SCPR cycle.

Modification of the Model

Although the original minimal SCPR model successfully recapitulates the critical elements of contractile-ring assembly in silico, two modified models were proposed recently to further address additional aspects of the assembly mechanism that we and others have observed in fission yeast cells. First, polarity was applied to the nodes in a modified model that introduces local alignment, a mechanism that allows the nodes to rotate and move around each other when they are closer than 0.4 lm [Ojkic et al., 2011]. It has been shown that Mid1 forms oligomers [Celton-Morizur et al., 2004], and the F-BAR protein Cdc15 forms filament-like structures when dephosphorylated [Roberts-Galbraith et al., 2010]. Therefore, the inclusion of local alignment in the model addresses protein–protein interactions between different nodes when they are very close to each other. In this modified model, a more homogenous and continuous distribution of nodes is achieved after condensation. Second, in light of the importance of crosslinking proteins and the stiffness of the actin network, a crosslinking parameter was introduced into a modified SCPR model [Laporte et al., 2012]. This modification can recapitulate the formation of the clump in ain1Δ cells and the stable meshworks in ain1 overexpressing cells (see the Pull section for discussion). In addition, the refined model supports a mechanism of cooperation between myosin-II and actin crosslinkers for successful node condensation.

The SCPR model focuses on early stages of contractile-ring assembly, and the later stages of node condensation are far more difficult to resolve than the initial stage. For example, although the actin filaments grow at random directions at early stages of contractile-ring assembly [Coffman et al., 2009], it is difficult to test whether this continues at later stages because the actin filaments are too dense at the contractile ring. Whether the increasing stiffness of the actin network restrains the orientation of nodes and the direction of actin filament elongation at later stages is unknown. The arrangement of F-actin in the contractile ring of wild type cells revealed by myosin S1 decoration and electron microscopy indicates that in a full-sized contractile ring, two semicircular populations of parallel filaments with opposite orientations exist during early anaphase B; in a constricting ring, filaments with opposite orientations are mixed homogenously throughout the ring [Kamasaki et al., 2007]. Investigating the orientations of F-actin in cells during late stages of node condensation, just before the formation of a compact ring, will be helpful to compare actin directionality at early versus late stages of node condensation. Although the architecture of the node before condensation indicates that the myosin-II motor head points toward the cytoplasm and does not support the formation of antiparallel myosin minifilaments, whether a change of node architecture may allow the minifilaments to form at later stages of node condensation remains unclear.

Mid1-Independent Contractile-Ring Assembly

In mid1Δ cells, there are no equatorial cytokinesis nodes, and contractile-ring assembly is severely impaired [Wu et al., 2003]. However, most mid1Δ cells are viable, indicating the existence of other mechanisms to assemble the essential contractile ring. Some mid1Δ cells can slowly make a randomly positioned and randomly oriented ring from linear structures consisting of Rng2, Myo2, Rlc1, Cdc12, actin, and other cytokinesis proteins [Wu et al., 2003; Hachet and Simanis, 2008; Huang et al., 2008; Coffman et al., 2009]. Recent studies suggest that the SIN pathway is involved in contractile-ring assembly in cells lacking functional Mid1 [Hachet and Simanis, 2008; Huang et al., 2008; Roberts-Galbraith and Gould, 2008; Bathe and Chang, 2010] (For a detailed review of the SIN pathway, please see Johnson et al. [2012] in this issue). Hyperactivation of the SIN pathway induces contractile-ring assembly in interphase [Hachet and Simanis, 2008; Huang et al., 2008]. SIN-induced rings assemble from filamentous or linear structures that arise from random positions, resembling rings assembled in mid1Δ cells. When the SCPR mechanism is not compromised, SIN mainly functions in later stages of cytokinesis (see “Ring maturation and constriction” section). In SIN mutants, a compact ring can form but cannot mature and constrict before its collapse [Balasubramanian et al., 1998; Hou et al., 2000; Krapp et al., 2001; Hachet and Simanis, 2008]. Although contractile rings can form in mid1-6 and sid2-250 (SIN kinase) single mutants, it cannot form in the double mutants [Hachet and Simanis, 2008], suggesting the existence of parallel and/or sequential pathways for ring assembly in fission yeast. The Polo kinase Plo1 regulates both pathways and both are important for the formation of a functional contractile ring at the correct division site [Bähler et al., 1998; Tanaka et al., 2001; Almonacid et al., 2011].

The SIN-induced, Mid1-independent rings are able to constrict more slowly than normal rings [Hachet and Simanis, 2008], allowing completion of cytokinesis in some cells. It may be that without equatorial cytokinesis nodes in mid1Δ cells, the filamentous or linear structures could incorporate and spread the essential ring components and thus make a functional contractile ring [Roberts-Galbraith and Gould, 2008; Bathe and Chang, 2010]. However, the molecular mechanism of Mid1-independent ring assembly remains obscure. In addition, it is unknown why SIN-induced rings are less homogenous and constrict slower [Hachet and Simanis, 2008].

Taken together, contractile-ring assembly in wild type fission yeast can be successfully modeled using the SCPR mechanism, and the Mid1-independent ring assembly depends on the SIN pathway. In the SCPR model, each parameter consists of numerous events in the cells and involves many proteins. Contractile-ring assembly requires multiple rounds of SCPR. Therefore, defects in one step might lead to severe consequences. However, many proteins are involved in more than one step in the process. Thus, the in vivo results obtained are not always easy to interpret. In vitro assays have provided many insights, but in vivo analyses with high spatial and temporal resolution are required to distinguish different possibilities. It will be of interest to further test the assumptions and predictions of the SCPR model, and characterize proteins and processes that contribute to each step. In addition, integrating results from both node-dependent and Mid1-independent contractile-ring assembly will be a challenge in the future.

Ring Maturation and Constriction

In wild type cells, the condensation of nodes results in a compact ring without lagging nodes at ~10 min after SPB separation. The diameter of the ring stays constant for ~25 min before constriction begins [Wu et al., 2003]. During this stage, the contractile ring matures by concentrating many additional proteins to the ring and/or to the division site adjacent to the ring, including additional F-BAR protein Cdc15, capping protein, the unconventional myosin-II heavy chain Myp2 [Bezanilla et al., 1997, 2000; Wu et al., 2003; Kovar et al., 2005], Rho GTPases and their regulators [Mutoh et al., 2005; Nakano et al., 2005; Rincon et al., 2007; Wu et al., 2010; Arasada and Pollard, 2011], Arp2/3 complex and its activators [Carnahan and Gould, 2003; Takeda and Chang, 2005; Wu et al., 2006], septins [Wu et al., 2003; An et al., 2004], and many other proteins [Pollard and Wu, 2010].

Compared to cytokinesis-node and contractile-ring assembly, much less is known about ring maturation and constriction. Nevertheless, many studies indicate that the F-BAR protein Cdc15 and the SIN pathway play important roles during this stage. The number of Cdc15 molecules in the contractile ring increases 10-fold during ring maturation [Wu and Pollard, 2005]. Without functional Cdc15, a compact contractile ring can assemble but falls apart [Fankhauser et al., 1995; Balasubramanian et al., 1998; Wachtler et al., 2006; Hachet and Simanis, 2008]. The regulation of ring integrity is mediated at least partly through the SH3 domain of Cdc15. It interacts with the C2 domain protein Fic1 and the paxillin Pxl1, two of the proteins that appear at the division site during ring maturation and are involved in maintaining ring integrity [Ge and Balasubramanian, 2008; Pinar et al., 2008; Roberts-Galbraith et al., 2009]. Imp2, another F-BAR protein in fission yeast, cooperates with Cdc15 during this process [Roberts-Galbraith et al., 2009]. Failure to maintain ring integrity is also observed in SIN mutants with reduced Cdc15 recruitment to the division site [Hachet and Simanis, 2008].

In most mutants that exhibit a delay in contractile-ring assembly, the initiation of ring constriction is not delayed. For example in myo2-E1 cells, it takes much longer to assemble the contractile ring even at permissive temperature, but once the contractile ring is formed, it only undergoes a very short “dwell time” before constriction begins [Coffman et al., 2009; Stark et al., 2010]. In contrast, in cells with two copies of myo2, the contractile ring assembles prematurely, leaving a prolonged dwell time before ring constriction [Stark et al., 2010]. In mid1Δ and mid1 mutants, once the SIN-dependent ring is assembled after a delay, its diameter also starts to decrease immediately [Celton-Morizur et al., 2004; Hachet and Simanis, 2008; Huang et al., 2008]. Together, these results suggest that ring maturation and constriction are tightly controlled and probably regulated through a cell-cycle-dependent signal/mechanism. When a delay in ring assembly occurs, the ring may mature while being assembled. The recruitment of additional ring components during ring maturation could assist the defective assembly process in some mutants. For example, the SIN-dependent pathway could recruit additional Cdc15 to complement a defective SCPR mechanism in several mutants [Hachet and Simanis, 2008], and the arrival of Myp2 may support the eventual ring assembly observed in myo2-E1 mutants. Given that many proteins recruited in ring maturation are necessary for ring constriction, recruitment of these proteins during delayed assembly could allow the contractile ring to constrict at the normal time. Future studies should focus on characterizing the effect of proteins recruited during ring maturation on mutant ring assembly processes.

In addition to the recruitment of more components, several lines of evidence suggest that the ring undergoes reorganization during maturation and constriction. First, the disappearance of Mid1 from the ring at the onset of ring constriction suggests that other proteins take over its role to anchor the contractile ring to the plasma membrane. The postanaphase array of microtubules and β-glucan synthase Bgs1/Cps1 contribute to the anchoring of contractile ring at the equator of the cell [Pardo and Nurse, 2003]. Second, different dynamics of cytokinesis proteins in FRAP assays may suggest that the organization of cytokinesis nodes, fully assembled rings, and constricting rings are different, although variance exists in how these experiments were performed [Clifford et al., 2008; Yonetani et al., 2008; Coffman et al., 2009; Laporte et al., 2011]. The reorganization likely prepares the ring for constriction.

Myosin-II motor activity is required for both the assembly and constriction of the contractile ring. Because the diameter of the contractile ring does not change while the ring matures, it will be interesting to investigate how the myosin-II motor activity is regulated at this time. It is possible that after node condensation into a compact ring, the motor activity is turned off during ring maturation, and activated again for constriction. It has been suggested that Rlc1, phosphorylated by Pak1, inhibits motor activity of Myo2 until ring constriction [Naqvi et al., 2000; Loo and Balasubramanian, 2008]. However, an in vitro assay suggested the opposite result [Sladewski et al., 2009]. A Pak1 mutant with impaired kinase function accelerates ring constriction when anaphase progression is slowed down in the ase1Δ mutant [Loo and Balasubramanian, 2008], but the same phenotype was not observed with nonphosphorylatable Rlc1 in ase1⁺ background [Sladewski et al., 2009]. Therefore, whether the myosin-II motor is switched off during ring maturation remains elusive.

Alternatively, the lateral redistribution of ring components along the arc length during ring maturation, as observed in 4D projections of GFP-Cdc15 and Rlc1-GFP [Wu et al., 2006; Hachet and Simanis, 2008], suggests that the myosin motor may remain active at this stage. If so, ring constriction must be prevented by other mechanisms so that the ring diameter remains constant during ring maturation. It has been shown that turgor pressure in fission yeast cells is very high [Minc et al., 2009]. It is possible that the force produced by active myosin-II walking on actin filaments is enough to slide actin filaments laterally along the plasma membrane during ring assembly, but not sufficient to overcome the turgor pressure to constrict the ring during ring maturation. It will be enlightening to investigate if septum formation and membrane insertion provide additional forces to overcome the high turgor pressure during ring constriction.

Ring constriction is regulated by the SIN pathway. SIN components localize to SPBs, but Sid2 kinase and its interacting partner Mob1 [Salimova et al., 2000] also localize to the contractile ring [Sparks et al., 1999; Hou et al., 2000]. Sid2 phosphorylation of Clp1 is required for the retention of Clp1 in the cytoplasm, and the mutation of Sid2 phosphorylation sites on Clp1 causes defects in cytokinesis [Chen et al., 2008]. Other Sid2 substrates present in the contractile ring remain to be identified.

Our understanding of ring maturation and constriction is far from complete. Investigating when and how different proteins are added to the contractile ring will shed light on the process of ring maturation. In addition, it is important to investigate what kind of reorganization takes place while the ring matures. Unlike in S. cerevisiae [Young et al., 2010], the cytokinesis apparatus in S. pombe has not been successfully isolated and purified yet, and its more dynamic nature and bigger diameter make it challenging. In contrast, quantitative live-cell imaging promises to be a powerful tool in determining the concentration and dynamics of ring components. Three-dimensional reconstitution or a system that allows the cells to be imaged vertically to overcome the poor z-resolution will be particularly helpful in studying the architecture of the ring and will potentially reveal the change of ring organization during ring constriction.

Conclusions and Perspectives

In this review, we have summarized recent advances on contractile-ring assembly in fission yeast. Cytokinesis nodes assemble in a hierarchical order and condense into a compact contractile ring through a process described as the SCPR mechanism. Each step in the mechanism is contributed by a subset of proteins that regulate myosin-II activity and actin dynamics. Although the mechanism is less clear, the compact ring matures by recruiting additional components and undergoes remodeling before and perhaps also during its constriction.

Gaining new information to refine the established model in fission yeast will benefit the field of cytokinesis significantly. Although thoroughly investigating the function and regulation of key players such as the anillin-like Mid1, IQGAP Rng2, myosin-II complex, the F-BAR protein Cdc15, and formin Cdc12 will allow us to further examine and polish the mechanism of contractile-ring assembly, we need to keep in mind that many other proteins are likely involved in this process. In fission yeast, ~250 different proteins localize to the division site [Matsuyama et al., 2006], at least ~130 proteins have been reported to be involved in cytokinesis, and many of them are conserved from yeasts to humans [Pollard and Wu, 2010]. It is necessary to analyze the genetic and physical interactions among these proteins systematically and elucidate whether they contribute to contractile-ring assembly.