Mechanics and regulation of cytokinesis in budding yeast

Abstract

Cytokinesis is essential for the survival of all organisms. It requires concerted functions of cell signaling, force production, exocytosis, and extracellular matrix remodeling. Due to the conservation in core components and mechanisms between fungal and animal cells, the budding yeast Saccharomyces cerevisiae has served as an attractive model for studying this fundamental process. In this review, we discuss the mechanics and regulation of distinct events of cytokinesis in budding yeast, including the assembly, constriction, and disassembly of the actomyosin ring, septum formation, abscission, and their spatiotemporal coordination. We also highlight the key concepts and questions that are common to animal and fungal cytokinesis.

1. Introduction to cytokinesis

Cytokinesis is the pivotal process of partitioning cellular constituents from one mother cell into two daughter cells following DNA replication and chromosome segregation. Successful completion of cytokinesis ensures faithful inheritance of genetic codes and is thus essential for maintaining species identity during its propagation [1-3].

Besides its apparent role in increasing cell number via symmetric division, cytokinesis also drives cell diversity via asymmetric division, i.e. a polarized cell dividing to generate two daughters of distinct fate and size. An extreme example is the polar body extrusion during oogenesis in which asymmetric division results in the formation of a large oocyte and a small polar body [4]. Asymmetric division is also exploited by stem cells for self-renewal and differentiation, and loss of the division asymmetry may cause tumorigenesis [5-7]. For example, in Drosophila, neuroblasts undergo asymmetric division to generate each a neuroblast and a smaller daughter cell – the ganglion mother cell (GMC) [5, 8, 9]. The GMC further divides to form two post-mitotic neurons. In the neuroblast, the atypical protein kinase C (aPKC) at the apical cortex promotes self-renewal, while Brain tumor (Brat) and Prospero at the basal cortex inhibit self-renewal and promote differentiation [10-12]. In mutants such as large giant larvae (lgl), the asymmetric localization of these fate determinants is lost, resulting in a tumor-like phenotype - hyperproliferation of neuroblasts at the expense of neurons [10-12].

Cytokinesis is also regulated to control the ploidy of the cells in different tissues and organs. For example, cardiomyocytes undergo nuclear division without cytokinesis, leading to the formation of binucleated cells. This increase in cell size or hypertrophy is believed to drive heart growth after birth [13, 14]. The function or presumed benefit of increasing ploidy is reflected, at least in part, in the observation that mononucleated and binucleated cardiomyocytes display differential electrical activity and calcium dynamics [15]. Hepatocytes also become binucleated postnatally as well as during liver regeneration [16, 17]. The significance of this polyploidization remains unknown, although it has been suggested that its function is to generate genetic diversity and enable adaptation to xenobiotic stress or other environmental insults [16]. Polyploidization due to failed cytokinesis could also have detrimental effects. For example, tetraploid, p53-deficient cells, generated by transient blockage of cytokinesis, caused tumor formation in nude mice, whereas diploid cells did not [18]. The resulting tumors displayed a wide range of genetic abnormalities including aneuploidy, chromosome rearrangements, and gene amplification [18]. This genomic instability is believed to drive tumorigenesis.

Because of its essential role in development and survival of an organism, cytokinesis has been studied in multiple systems for decades. These studies have revealed that cytokinesis requires complex, spatiotemporal coordination of multiple activities including force production, targeted vesicle fusion, and extracellular matrix (ECM) remodeling [1, 3, 19, 20]. In addition, the basic machinery and operating principles of cytokinesis are conserved from yeast to humans. This high level of conservation, coupled with its accessibility to genetic manipulations and live-cell imaging, has made the budding yeast Saccharomyces cerevisiae an attractive model for studying the molecular mechanisms underlying this fundamental process [21].

2. Cytokinesis in S. cerevisiae

Cytokinesis in both animal and fungal cells involves the selection of a division site, cleavage furrow ingression powered by a contractile actomyosin ring (AMR), and localized membrane and ECM remodeling that ultimately leads to abscission. While the core division machinery is conserved through evolution, the mechanisms of division site selection are highly species specific. In animal cells and the fission yeast Schizosaccharomyces pombe, the division site is specified by the mitotic spindle and nuclear position in anaphase and G2, respectively [1]. In S. cerevisiae, the division site, which is the site of budding, is specified in late G1 by Cdc42-controlled cell polarization (Fig. 1) [22]. Cdc42 is also required for polarized organization of actin cables and patches as well as for nascent septin ring assembly at the presumptive bud site. The septin ring then recruits Myo1, the sole myosin-II in budding yeast, via the septin- and myosin-binding protein Bni5 [23-25]. Thus, in all organisms, the specified division site is marked by myosin-II irrespective of the timing of specification during the cell cycle.

Fig. 1.

Assembly and execution of key cytokinesis events during the cell cycle.

After bud emergence, the septin ring expands into an hourglass, which, together with the myosin ring, marks the bud neck – the future division site (Fig. 1). Meanwhile, two arrays of actin cables and actin patches are polarized towards the bud cortex and the bud neck to drive bud growth and morphogenesis during S/G2/M phases by mediating exocytosis and endocytosis, respectively (Fig. 1) [22, 26]. Upon the entry into anaphase, actin filaments and myosin-II interact to form an AMR (Fig. 1) [23, 24, 27]. Thus, the timing of AMR assembly in budding yeast and animal cells is strikingly similar, despite their difference in the timing of myosin localization at the division site.

Upon activation of the mitotic exit network (MEN), the septin hourglass is instructed to convert into a double ring, which sandwiches the AMR (Fig. 1) [28, 29]. The AMR begins to constrict centripetally [23, 24]. Meanwhile, actin cables are polarized towards the bud neck to mediate the delivery of post-Golgi vesicles, which fuse with the plasma membrane (PM) to increase surface area as well as release cargoes such as the chitin synthase-II (Chs2) to drive primary septum (PS) formation [30, 31]. AMR constriction is followed closely by PS formation. The AMR guides PS formation in addition to presumed force production [25, 32, 33]. Reciprocally, in the absence of PS formation, the AMR undergoes asymmetric constriction towards one side of the bud neck, which might reflect a partial detachment of the AMR from the PM [31, 34-36]. Thus, the PS is thought to stabilize the AMR during its constriction. Conceptually, PS formation could be considered as the functional counterpart of ECM remodeling during cytokinesis in animal cells [37-39]. Near the end of PS formation, secondary septa (SS), which are synthesized by glucan synthases, begin to form at both sides of the PS [21, 40]. The SS is thought to plug a “tiny hole” at the center of the PS, leading to “abscission” i.e. separation of cytoplasm between mother and daughter cells [41].

Following SS formation, a kinase cascade, named Regulation of Ace2 and Morphogenesis (RAM) network, is activated, leading to the exclusive localization of the transcription factor Ace2 to the daughter cell nucleus [21, 40]. Ace2 controls the expression of cell wall hydrolases that digest the PS and part of the SS, resulting in cell separation.

In this review, we discuss the mechanics and regulation of cytokinesis in budding yeast, highlighting all the key molecules (Fig. 2), concepts, and questions with respect to specific events involved in cytokinesis.

Fig. 2.

Localization and dynamics of core cytokinesis proteins during the cell cycle. Modified from Wloka et al. 2013, which reports protein dynamics revealed by FRAP analysis.

3. Actomyosin ring assembly

3.1. Septins

3.1.1. Architecture, dynamics, and remodeling of the septin cytoskeleton during the cell cycle

Septins were first identified in budding yeast for their essential role in cytokinesis [42]. Septins are GTP-binding, cytoskeletal proteins that are conserved from yeast to humans and play critical roles in diverse functions including cytokinesis, cell migration, and morphogenesis [43-49]. Defects in septin functions are associated with diverse diseases including neurodegenerative diseases, infertility, and cancer.

Septins form rod-shaped, non-polar heterooligomeric complexes that polymerize end-to-end into linear filaments, which can be further organized into higher-order structures such as rings and gauzes [50, 51]. These structures are thought to impact the above-mentioned processes by acting as a cellular scaffold and/or diffusion barrier. In budding yeast, five mitotic septins form at least two distinct palindromic octameric complexes (Cdc11-Cdc12-Cdc3-Cdc10-Cdc10-Cdc3-Cdc12-Cdc11 and Shs1-Cdc12-Cdc3-Cdc10-Cdc10-Cdc3-Cdc12-Shs1) (Fig. 2) [52, 53]. These octamers form distinct cellular structures during different phases of the cell cycle [45, 47]. At the beginning of the cell cycle (in late G1), septins are recruited to the presumptive bud site (PBS), where they are assembled into a dynamic cortical ring (Fig. 1). This process depends on Cdc42 [29, 54-56], a small essential GTPase that plays a central role in cell polarization from yeast to humans [57]. Upon bud emergence, the cortical ring expands into a stable hourglass structure at the mother-bud neck, as indicated by fluorescence-recovery-after-photobleaching (FRAP) analysis (Fig. 2) [55, 58]. At the onset of cytokinesis, the hourglass is remodeled into a double ring, which sandwiches the constricting AMR during cytokinesis (Fig. 1). After cell division, the mother and daughter each inherits half of the double ring, which disassembles at the beginning of the next cell cycle while a new ring is formed at the PBS (Fig. 1).

How are the septin filaments packed in different cellular structures, especially in the hourglass and double ring? This key question was addressed by several pioneering studies using different experimental approaches. The original thin-section electron microscopy (EM) and more recent cryo-EM tomography suggest that paired septin filaments are organized into circumferential rings around the bud neck at a spacing of ∼28 nm [59, 60]. In contrast, polarized fluorescence microscopy suggests that septin filaments are aligned along the mother-daughter axis in the hourglass, and circumferentially around the mother-bud neck in the double ring [61, 62]. This approach allows the inference of filament orientation at the population, but not individual-filament, level. Therefore, a detailed architecture of the septin hourglass or double ring remained to be elucidated.

The close proximity of septin filaments to the PM and the high-density ribosomes in the surrounding cytoplasm could complicate the interpretation of filament orientation in the EM studies of thin sections of yeast cells. To overcome this potential problem, rapid-freeze-and-deep-etching EM was used to examine septin filaments associated with cell wall-removed, unroofed yeast cell cortices. This study revealed the presence of rings and gauze-like structures of septin filaments [50]. However, the cells used in this study were unsynchronized. Thus, it is uncertain during which stage of the cell cycle these septin structures are produced. Using the unroofing protocol developed in this study, with some modifications, a platinum-replica EM (PREM), which has been extensively used to analyze the actin and microtubule cytoskeleton in mammalian cells [63, 64], was performed to examine septin structures from cells synchronized at different stages of the cell cycle [51]. It was found that in the early hourglass, paired septin filaments of 300-400nm in length are densely packed along the mother-daughter axis [51]. These filaments likely contain septin octamers with Cdc11 as the terminal subunits [52, 53, 65]. In the late hourglass (before the onset of cytokinesis), a set of single filaments are added perpendicularly to the paired filaments at a regular spacing of ∼29 nm, which is approximately a septin octamer length (Fig. 3A) [51]. It is hypothesized that these filaments contain septin octamers with Shs1 as the terminal subunits [51, 53]. In the double ring, there are both paired and single filaments that are arranged circumferentially (Fig. 3B) [51]. This study provided the first view of the septin hourglass and double ring at the filament level.

Fig. 3.

Remodeling of the septin cytoskeleton from an hourglass to a double ring at the onset of cytokinesis. (A and B) Electron micrographs showing orientation of septin filaments in the hourglass and double ring. The paired filaments (green) are perpendicular to the single filaments (magenta), and their orientation with respect to the putative myosin-II filaments (purple) is shown. (C) A model of septin hourglass-to-double ring transition. AMR, actomyosin ring. Modified from Ong et al. 2014.

How is the septin hourglass remodeled into a double ring during the cell cycle? FRAP, photo-activation, and photo-conversion analyses indicate that the septin hourglass-to-double ring transition is accompanied by filament disassembly in the hourglass coupled with a double ring formation at the ends of the hourglass (Fig. 3C) [51]. What triggers filament disassembly and what dictates double ring formation at the ends of the hourglass remains unknown. Bud4, the anillin-like protein in budding yeast, is clearly required for double ring formation [66-68]. It is worth noting that anillin and anillin-like proteins are also involved in septin organization at the division site in other organisms [69-71].

3.1.2. Roles of septins in cytokinesis

Septins play distinct roles in cytokinesis via distinct architectures. They are the first proteins to arrive at the bud neck to scaffold the sequential assembly of the division machinery during the cell cycle. This “scaffolding” role is carried out by the septin hourglass, which also acts as a platform for anchoring, concentrating, and perhaps activating signaling molecules at the bud neck to carry out various cellular functions [45, 72]. From bud emergence to the onset of telophase, the septin hourglass is required for the localization of Myo1 to the bud neck [23]. This targeting event is mediated by the yeast-specific, septin-binding protein Bni5 before anaphase, and by the collective action of Bni5 and the IQGAP Iqg1 during anaphase [25, 73-75]. Actin ring formation begins at the onset of anaphase, and depends on the presence of Iqg1 [24, 27, 76] and Myo1 [23] at the septin hourglass. Thus, a key role of the septins is to scaffold AMR assembly before the onset of cytokinesis via the hourglass architecture.

The septin hourglass is also involved in shaping the AMR. When the hourglass is conditionally disrupted in a septin mutant, Myo1 and its essential light chain (ELC) Mlc1 are still able to localize at the division site in an F-actin-dependent manner, but they no longer form a smooth ring structure [77]. Thus, the septin hourglass is required for the assembly and shaping of the AMR.

At the onset of cytokinesis, the septin hourglass is remodeled into a double ring that sandwiches the AMR. The double ring is thought to act as a diffusion barrier to restrict diffusible factors to the division site during cytokinesis [78]. This barrier role becomes essential only when the AMR is compromised [66].

3.2. Myosin-II

Just like all myosin-IIs in animal cells, the sole myosin-II in budding yeast is a hexamer consisting of a dimer of the myosin-II heavy chain, Myo1, each of which is associated with an ELC (Mlc1), and a regulatory light chain (RLC) (Mlc2) (Fig. 2) [79]. Myo1 forms a two-headed structure with a kink in its tail [25], typical of non-muscle or smooth-muscle myosin-IIs [80]. This kink corresponds to the region containing seven proline residues [25]. Mlc1 and Mlc2 bind to the IQ1 and IQ2 motifs located between the head and tail of Myo1, respectively [79]. Mlc1 is not only the ELC for Myo1, but also a light chain for myosin-Vs (Myo2 and Myo4 in budding yeast) [79, 81, 82] and Iqg1 [76, 83]. While deletion of MYO1 causes pronounced defects in cytokinesis and cell separation, null cells are viable. In contrast, MLC1 is essential for cell viability and cytokinesis [76, 81-83]. This presumably reflects a Myo1-independent role of Mlc1 in targeted membrane trafficking and septum formation during cytokinesis that is mediated by myosin-V and Iqg1, respectively. Mlc2 binds to Myo1, but not Myo2 or Iqg1 [79]. Thus, Mlc2 is a dedicated light chain for Myo1. Deletion of MLC2 causes only a mild defect in Myo1 disassembly during cytokinesis [79, 84]. The number of binding partners of Mlc1 and Mlc2 might explain why ELCs are more conserved than RLCs in sequence and function through evolution for all organisms examined so far [79]. Collectively, these observations indicate that different components of the myosin-II in budding yeast play distinct roles in cytokinesis.

Different components of the myosin-II complex also target to the division site via different mechanisms. Distinct regions of Myo1 tail mediate its localization to the bud neck during different phases of the cell cycle for distinct functions [25]. Myo1 is recruited to the nascent septin ring at the presumptive bud site, and co-localizes with the septin hourglass from bud emergence to the onset of cytokinesis [25, 33, 74]. This Myo1-septin association is mediated by Bni5, which interacts directly with both the septins and the tail of Myo1 [25, 73-75]. In contrast, the localization of Myo1 at the bud neck from the onset of anaphase to the end of cytokinesis is mediated by Iqg1 [25]. Thus, both the Bni5- and Iqg1-based mechanisms contribute to Myo1 localization during anaphase, with the former tapering off while the latter escalating [25]. The Bni5 mechanism may mediate the role of Myo1 in the bud-to-mother retrograde flow of actin cables before anaphase [25, 85] while the Iqg1 mechanism is essential for AMR assembly and function during anaphase and cytokinesis [25], as Iqg1 and Myo1 both are required for actin ring formation [23, 24, 27]. IQGAP is also involved in myosin-II localization at the division site in fission yeast and Dictyostelium [86-89].

The RLC Mlc2 displays an identical localization profile to Myo1, and it localizes to the division site exclusively by binding to the IQ2 motif of Myo1 throughout the cell cycle [79]. In contrast, the ELC Mlc1 begins to accumulate at the division site at the medium-budded stage (∼G2/M phase) and disappears from the bud neck after cytokinesis and cell separation [76, 77, 79, 82, 83]. The neck localization of Mlc1 depends on the septins and actin filaments before and during cytokinesis, which is mediated chiefly by Myo1 and the formin Bni1, respectively [77].

Myosin-II also displays cell cycle-triggered changes in dynamics (Fig. 2). Nearly all the Myo1 molecules are localized to the division site ∼30 min after bud emergence, which corresponds to the small-budded stage [33]. Strikingly, Myo1 is mobile at the division site before the onset of anaphase and is progressively immobilized from anaphase to the onset of telophase and remains immobile during cytokinesis (Fig. 2) [33, 78]. The immobility of Myo1 depends on the “putative assembly domain” at its C-terminus, but not the head domain, the light chain-binding sites, or the actin cytoskeleton [33]. This observation suggests that Myo1 might undergo cell cycle-controlled filament assembly. Not surprisingly, Mlc2 and Myo1 display an identical profile of dynamics during the cell cycle (Fig. 2) [33], as Mlc2 is carried to the division site exclusively by Myo1 [79]. On the other hand, Mlc1 displays an immobility resembling the latter half of Myo1 during the cell cycle (Fig. 2), and this immobility, but not its localization, depends on Myo1 [33]. This and other observations described later suggest that Myo1 might form filaments that scaffold the assembly of the division machinery [33].

3.3. Formins and the actin ring

There are three filamentous actin (F-actin) structures in budding yeast: patches, cables, and the ring [26]. The actin patches consist of branched filaments nucleated by the Arp2/3 complex and are required for endocytosis [26, 90]. The actin cables and ring consist of linear filaments nucleated by the pair of formins, Bni1 and Bnr1 (Fig. 2) [32, 91, 92]. Bni1 and Bnr1 localize to the bud cortex and bud neck before cytokinesis to nucleate the assembly of two polarized arrays of actin cables that mediate vesicle and mRNA transport by myosin-Vs (Myo2 and Myo4 in budding yeast) to drive bud growth [26, 93, 94]. The localization and/or activities of Bni1 and Bnr1 are regulated by the bud cortex-localized small GTPases (Cdc42 and Rho1) and bud neck-localized septins (Shs1) and septin-associated kinases (Gin4), respectively [95-98]. At the onset of cytokinesis, Bni1 changes its location from the bud cortex to the bud neck and remains there until the end of cytokinesis, while Bnr1 disappears from the neck [99, 100]. Cells can survive the loss of either BNI1 or BNR1, but not both [32]. While both formins are collectively required for actin cable and actin ring assembly, Bni1 plays a much more predominant role in this process [32, 91, 92, 101]. Rho1 plays an essential role in actin ring assembly, presumably, by regulating Bni1 activity [101]. The activation of Rho1 at the division site is, in turn, regulated by the Polo kinase Cdc5 via the guanine-nucleotide-exchange factors (GEFs) for Rho1 [102, 103].

The linear actin filaments nucleated by the formins require stabilization by tropomyosins (the major isoform Tpm1 and the minor isoform Tpm2) (Fig. 2) to form actin cables and the actin ring [101, 104, 105]. It remains unknown how the same kind of filaments are shaped into two distinct cellular structures to perform distinct functions.

3.4. IQGAP

IQGAPs are a family of scaffold proteins involved in diverse functions by binding to distinct partners via distinct domains [106, 107]. Mammalian cells have three isoforms, IQGAP1-3, that are expressed either ubiquitously (IQGAP1) or in a tissue-restricted manner (IQGAP2 and 3). While their roles, especially those of IQGAP1, in cell signaling, cytoskeletal organization, and cell-cell adhesion have been analyzed extensively, it remains unclear whether and how they are involved in mammalian cytokinesis. In contrast, the only or major known function of IQGAPs in budding yeast [24, 27], fission yeast [108], Dictyostelium [109], sea urchin [110], and C. elegans [111] is in the process of cytokinesis.

The sole IQGAP, Iqg1, in budding yeast plays at least two distinct roles in AMR assembly. It promotes actin ring formation by binding to actin filaments via its CHD (Calponin Homology Domain) [24, 25, 27, 76, 112], and it is required for the localization or maintenance of Myo1 at the division site during cytokinesis [25]. Both roles appear conserved in fission yeast [86, 87, 89, 113] and Dictyostelium [88, 109, 114].

3.5. A molecular model for AMR assembly

IQGAP (Iqg1) scaffolds AMR assembly by binding to formin (Bni1 and Bnr1)-nucleated actin filaments via its CHD and to myosin-II (Myo1, Mlc1, and Mlc2) in a manner that remains to be elucidated. These molecular complexes and/or assemblies are anchored to and shaped by the septin hourglass into a ring structure.

4. Actomyosin ring constriction

AMR constriction drives ingression of the cleavage furrow and guides vesicle fusion and septum formation. The constriction is powered by motor-dependent and –independent forces. The head domain of Myo1, which includes the motor domain and light chain-binding sites, is responsible for ∼20-30% of the constriction rate [25, 115]. Cells expressing only the tail of Myo1 form a “headless” AMR that can largely direct PS formation [25]. The nature of the “motor-independent” forces remains unclear, but it likely involves PS formation, which presumably drives PM ingression at the division site in the absence of the motor domain [25, 115].

Actin filaments play an essential role in AMR constriction, as their disruption by latrunculin A (LatA) does not block Myo1 localization to the bud neck but completely abolish its constriction during cytokinesis [23, 84]. Actin filaments could directly mediate Myo1 constriction as a part of the AMR, as indicated by the observation that Myo1 displays partial and abnormal constriction in an iqg1 mutant specifically defective in actin ring assembly [25, 112]. The architecture of the AMR remains unknown in any system. Thus, it is unclear how actin filaments and Myo1 interact with each other to produce a contractile force. Actin filaments could also regulate AMR constriction as components of the actin cables, via PS formation, by controlling myosin-V (Myo2)-dependent delivery of Chs2 to the bud neck during cytokinesis. Thus, actin filaments could regulate AMR constriction via two distinct actomyosin systems.

5. Actomyosin ring disassembly

In contrast to muscle contraction where the number of “contractile units” remains the same [116, 117], AMR constriction must be coupled with disassembly [118, 119]. As expected, Myo1 is progressively lost during AMR constriction [66, 84]. Three distinct mechanisms appear to control Myo1 disassembly, which involves its motor domain, its RLC Mlc2, and the IQGAP Iqg1, respectively.

While a “headless AMR” can largely drive cytokinesis in budding yeast, the constriction rate is decreased 20-30% [25, 115]. In addition, Myo1-tail displays a clear disassembly defect as well as a defect in the symmetry of ring constriction during cytokinesis [25]. As Myo1 lacking the light-chain binding sites does not show any similar phenotype [79], these data suggest that the Myo1 motor domain is involved in the control of AMR disassembly. Consistently, inhibition of motor activity of myosin-II in mammalian cells also causes defects in AMR disassembly [120]. However, it remains unknown how the motor domain regulates myosin disassembly during cytokinesis.

Deletion of MLC2 causes a delay in Myo1 disassembly [79]. Similarly, myosin-II purified from RLC-null cells of Dictyostelium discoideum displays a defect in filament disassembly [121]. Strikingly, the cytokinesis defects of RLC-null cells in budding yeast [79], fission yeast [122], and Dictyostelium [123] are suppressed by deletion of the RLC-binding site on the respective myosin heavy chain. These observations suggest that the RLC in these organisms acts to relieve an auto-inhibitory effect of the RLC-binding site on myosin disassembly during cytokinesis.

Iqg1 undergoes anaphase promoting complex/cyclosome (APC/C)-mediated degradation during AMR constriction [84, 124]. It appears to intimately associate with Myo1, Mlc1, and Mlc2 at the bud neck in wild-type cells as well as at the ectopic sites in APC/C mutants where Iqg1 becomes stabilized [84, 124]. Thus, Iqg1 likely controls AMR disassembly via its presumed interaction with Myo1 [25] as well as its demonstrated interaction with Mlc1 [76, 83]. It remains to be seen whether a similar mechanism operates in other systems.

6. Membrane insertion at the division site

Cytokinesis in animal, fungal, and plant cells requires targeted vesicle fusion at the division site [19, 20, 125]. This serves to supply membranes to cover the new cell ends created during cell division and deliver enzymatic cargoes to remodel the ECM at the division site [21]. In budding yeast, the entire growth machinery is switched from the bud cortex to the bud neck during mitotic exit to promote cytokinesis (Fig. 1). This includes polarity regulators, actin cables, myosin-V, and secretory components such as the exocyst, a conserved vesicle-tethering complex [22, 26, 93, 126]. The formin Bni1 plays a dual role in cytokinesis, promoting vesicle transport and AMR constriction by nucleating actin cable and actin ring assembly, respectively [32, 91, 92, 101, 127, 128]. It remains unknown how the growth machinery is targeted to the division site and how vesicle fusion is spatiotemporally coupled to the AMR during cytokinesis.

7. Primary septum formation

Increasing evidence suggests that ECM remodeling is a common feature of fungal and animal cytokinesis. In C. elegans and mice, chondroitin synthesis is essential for embryonic cytokinesis [37, 38]. In sea urchin eggs [129] and human cells [130-132], hyaluronan, a prevalent ECM component, is enriched at the division site. In budding yeast, AMR constriction is followed closely by the centripetal growth of a primary septum, an electron-lucent, chitinous structure that formed only during cytokinesis (Fig. 1). The PS is synthesized by the chitin synthase II (Chs2) [133, 134]. Chitin (polymers of N-acetyl-glucosamine; also exists in Drosophila and zebrafish) is structurally similar to hyaluronan (polymers of glucuronic acid and N-acetyl-glucosamine disaccharide). In addition, both polymers are synthesized by members of the glycosyltransferase-2 family. Thus, the PS can be considered as a yeast counterpart of an animal ECM. In budding yeast, deletion of CHS2 causes rapid and asymmetric AMR constriction, suggesting that the PS might stabilize the AMR during its constriction [31, 34, 35]. This PS-AMR interplay also applies to the fission yeast S. pombe [135]. Similarly, hemicentin, a secreted ECM protein, is required for preventing retraction of the nascent cleavage furrow during cytokinesis [39]. Thus, a common role of ECM remodeling at the division site appears to stabilize the AMR during furrow ingression. Chs2 is delivered to the bud neck by exocytosis at the onset of cytokinesis [30]. The AMR and the septin double ring play distinct and collectively essential roles in restricting Chs2 to the division site [66, 78]. A protein complex consisting of Inn1, Hof1, and Cyk3 is involved in Chs2-mediated PS formation (Fig. 2) [36, 136-139]. Inn1 contains a C2 domain at its N-terminus that is essential for PS formation, and several PXXP motifs at its C-terminus that bind to the SH3 domains of Hof1 and Cyk3 [36]. Hof1 is a F-BAR protein [32, 140-142] and Cyk3 contains a transglutaminase domain [143, 144]. These proteins interact with each other and share an essential role in AMR-independent cytokinesis [41, 137, 138, 143]. Deletion of INN1 completely blocks PS formation but not Chs2 localization, suggesting that Inn1 is involved in Chs2 activation [36]. Overexpression of Cyk3 stimulates PS formation in wild-type cells [145] and suppresses the growth and septal defects of innΔ cells, suggesting that Cyk3 is also involved in Chs2 activation [36]. Thus, a critical question is how Inn1 and Cyk3 promote PS formation by activating Chs2.

8. Secondary septum formation

Secondary septa (SS) are electron-dense structures formed at both sides of the PS (Fig. 1). Their formation is triggered near the end of PS formation [41]. SS not only protect the newly generated cell ends but also “plug” the tiny hole at the center of the PS, leading to complete membrane closure or abscission between the daughter cells [41]. SS are structurally similar to lateral walls, which are composed of mannoproteins and 1,3-β-D-glucan. The glucan is synthesized by the glucan synthases, with Fks1 and Fks2 being their catalytic subunits [146-149]. While Fks1 functions mainly during budding and Fks2 mainly during nutritional starvation, mating, and sporulation, they are collectively essential for cell wall synthesis, SS formation, and cell survival during vegetative growth [147, 148]. The chitin synthase III also plays a role in SS formation, as cells lacking the catalytic subunit Chs3 form abnormal SS [134]. Consistently, Chs3 becomes essential for cytokinesis in the absence of Chs2 (lacking the PS) [134] or Myo1 (lacking the AMR) [41, 103].

SS formation requires Cdc42 inactivation and Rho1 activation during cytokinesis [41, 150, 151]. These GTPases may mutually antagonize each other in part via the paxillin homologue Pxl1 [41, 152, 153] and Gps1/Aim44 [151, 154]. Rho1 controls SS formation by acting as the regulatory subunit of the glucan synthases Fks1 and Fks2 [148, 155]. Thus, Rho1 plays at least two sequential and distinct roles in cytokinesis: promoting actin ring assembly via the formin Bni1 and stimulating SS formation via the glucan synthases. Cyk3 binds to Rho1 and inhibits its activation during PS formation [41]. This ensures that SS are formed only after the PS [41]. Cyk3 also stimulates PS formation [145]. Thus, Cyk3 defines a mechanism for spatiotemporal coordination of PS and SS formation during cytokinesis.

9. Cell separation

Cell separation refers to the physical cut between the mother and daughter cells. This is achieved through degradation of the PS and part of the SS by cell wall-digestive enzymes including the endochitinase Cts1 [156] and the glucanases Dse4/Eng1 and Egt2 [157, 158]. Expression of Cts1 and Eng1 is controlled by the transcriptional factor Ace2 [159-161], whereas the expression of Egt2 is controlled by both Ace2 and its paralog Swi5 [158]. The activation and localization of Ace2 is, in turn, controlled by the RAM pathway, which comprises of Cbk1 kinase, its activating subunit Mob2, and upstream regulatory components [161-163]. Upon activation, Ace2 enters the daughter nucleus to drive the expression of the hydrolytic enzymes. After cell separation, Ace2 exits the nucleus and is kept in the cytoplasm by cyclin-dependent kinases [164]. Cbk1 kinase is evolutionarily conserved [40]. However, it is unknown whether there is a post-abscission or -cytoplasm cleavage process involving ECM remodeling at the division site that is catalyzed by a Cbk1 homologue in mammalian cells.

10. Spatiotemporal coordination of cellular events in cytokinesis

As described above, cytokinesis involves multiple cellular events including AMR constriction, targeted membrane deposition, and septum formation. These events must coordinate in time and space to ensure the execution of cytokinesis with high efficiency and fidelity. Here, we discuss some potential coordination mechanisms.

10.1. Igniting the cytokinesis machine by MEN

The mitotic exit network (MEN) is a conserved GTPase-dependent signaling pathway that coordinates spindle position, mitotic exit, and cytokinesis at the end of the cell cycle (Fig. 4A) [1, 40, 165-167]. The small GTPase Tem1 and its bipartite GTPase-activating protein (GAP) Bub2-Bfa1 preferentially associate with the bud-bound spindle pole body (SPB; counterpart of centrosome in animal cells). Upon the entry of this SPB into the daughter compartment, the bud cortex-localized Lte1 activates Tem1. This is not done by acting as a GEF via its GEF-like domain but by inhibiting the association of the GAP-stimulating kinase Kin4 with the daughter SPB [167-169]. Activated Tem1 binds to its effector, a Hippo-like kinase Cdc15, which, in turn, activates the LATS-like kinase Dbf2. The activation of this kinase cascade ultimately results in the release of the phosphatase Cdc14 from the nucleolus into the nucleus and cytoplasm [170]. Cdc14 then removes the phosphorylation from the substrates of the cyclin-dependent kinase (CDK), culminating in mitotic exit and cytokinesis [1, 40, 167].

Fig. 4.

The mitotic exit network and its regulation of cytokinesis. (A) The mitotic exit network. SPB, spindle pole body. (B) Spatiotemporal control of AMR constriction and PS formation by MEN and IQGAP. AMR, actomyosin ring; PS, primary septum.

Several MEN components, including the mitotic kinases Cdc15 and Dbf2 as well as the phosphatase Cdc14, localize to the division site to control cytokinesis [166, 171]. MEN regulates AMR constriction, as the ring can form but fails to constrict in MEN mutants [28, 32]. Several mechanisms could account for this observation. First, MEN controls the remodeling of the septin hourglass into a double ring, which sandwiches the AMR and renders its constriction “permissible”. The AMR is tethered to the “rigid” septin hourglass before the onset of cytokinesis, which makes constriction difficult, if not impossible (Fig. 3C). After the remodeling event, the AMR is tethered to the PM between the septin double ring (Fig. 3C), which permits, but is not necessarily sufficient for, constriction and furrow ingression. Second, MEN controls timely AMR assembly and constriction during the cell cycle (Fig. 4B). CDK phosphorylation of Iqg1 delays actin ring formation until anaphase [172] and Cdc14 dephosphorylation of Iqg1 promotes actin ring assembly from anaphase to, presumably, the end of cytokinesis [173]. As Iqg1 is also essential for Myo1 localization during cytokinesis [25], MEN is likely to control AMR assembly and constriction via Iqg1. MEN also controls AMR assembly via Cdc14 regulation of Bni1 localization during cytokinesis [174]. Currently, it is unclear whether MEN controls AMR constriction independently of its assembly.

MEN also controls PS formation, as the PS fails to form in the MEN mutants [138, 175]. Chs2 is synthesized and retained at the ER via CDK phosphorylation [176]. Cdc14 reverses this phosphorylation, which allows Chs2 to translocate from the ER to the bud neck to drive PS formation (Fig. 4B) [177, 178]. Dbf2 also phosphorylates Chs2 to control its association with the AMR during cytokinesis (Fig. 4B) [145]. Finally, MEN controls the localization of Inn1, Hof1, and Cyk3 to the bud neck during cytokinesis (Fig. 4B) [36, 137, 138]. These proteins are important for PS formation. Thus, MEN controls PS formation via multiple and distinct mechanisms.

10.2. Spatial coordination of actomyosin ring constriction and primary septum formation during cytokinesis

AMR constriction is always followed closely by the centripetal growth of the PS, suggesting that the two processes must be aligned with precision to achieve efficient cytokinesis. These processes are mutually dependent (Fig. 4B) [34, 35]. The AMR guides PS formation in addition to force production [25, 32, 33] whereas the PS stabilizes the AMR during its constriction [31, 36]. A similar scenario involving the interplay between the AMR and localized ECM remodeling at the division site might take place in animal cells [39].

How do the AMR constriction and PS formation spatially coordinate at the molecular level? Perhaps the best understood mechanism involves Iqg1 that links AMR to PS formation (Fig. 4B). Iqg1 is required for actin ring assembly [24, 27] and also for Myo1 localization during cytokinesis [25]. Thus, Iqg1 has a clearly defined role in AMR assembly. Iqg1 is essential for cell viability and cytokinesis whereas Myo1 is not, despite the fact that myoΔ cells display obvious defects in cytokinesis and cell separation [23, 24, 179, 180]. This observation suggests that Iqg1 must play additional role(s) in AMR-independent cytokinesis (Fig. 4B). Indeed, overexpression of Cyk3 rescued the growth and cytokinesis defects of iqgΔ cells without restoring the AMR, supporting a role of Iqg1 and Cyk3 in AMR-independent cytokinesis [143]. Subsequently, it was shown that Cyk3, Hof1, and Inn1 interact with each other to promote PS formation (Fig. 4B) [36, 137, 138]. Both Inn1 and Hof1 also interact with Iqg1 [136, 139, 172, 181]. Taken together, these observations suggest that Iqg1 plays a pivotal role in both AMR constriction and PS formation during cytokinesis.

11. Closing remarks

The collective power of diverse model systems, gene-editing tools, and imaging technologies has led to rapid progress in the field of cytokinesis over the last two decades. However, analysis has been largely focused, and rightfully so, on individual mutants, proteins, and protein complexes. The major challenge in the future is to determine how different proteins are wired together at a systems level to execute cytokinesis with robustness and high fidelity.

Comparative analysis of cytokinesis in different models using diverse experimental approaches has proven to be an effective strategy for illustrating the general principles of cytokinesis, as each model offers distinct experimental advantages and may provide insights from a unique angle. Such an analysis has led to the notion that cytokinesis in animal and fungal cells, including the budding yeast discussed here, entails the same processes (AMR constriction, targeted membrane trafficking, and ECM remodeling) and core components, despite inevitable differences in details. Such an approach should be continuously employed to tackle the fundamental questions regarding cytokinesis: what is the architecture of the AMR in different organisms? Is the core design of the AMR evolutionarily conserved or different despite the common components (actin and myosin filaments)? Is the AMR a pure force-generating machine in animal cells, as currently believed, or is it also involved in guiding membrane traffic and ECM remodeling at the division site, as suggested in budding yeast? How are exocytic vesicles targeted to and fused at the division site? Is abscission mechanistically similar in animal and fungal cells? These compelling questions not only spark the scientific curiosity but also press the need for developing more advanced tools to study the nature and coordination of micro systems pertaining to cytokinesis.