Three’s company: The fission yeast actin cytoskeleton

Abstract

How the actin cytoskeleton assembles into different structures to drive diverse cellular processes is a fundamental cell biological question. In addition to orchestrating the appropriate combination of regulators and actin-binding proteins, different actin-based structures must insulate themselves from one another to maintain specificity within a crowded cytoplasm. Actin specification is particularly vexing in complex eukaryotes where a multitude of protein isoforms and actin structures operate within the same cell. Fission yeast Schizosaccharomyces pombe possesses a single actin isoform that functions in three distinct structures throughout the cell cycle. In this review, we explore recent studies in fission yeast that help unravel how different actin structures operate in cells.

Introduction

Fission yeast provides an important model to study actin function. The fission yeast actin cytoskeleton is comparatively simple, relying on only three actin structures for growth: actin patches, actin cables, and the contractile ring (Figure 1) ¹. An additional actin structure is also used during mating (Box 1). Work from numerous labs has identified most of the key components required for the regulation, assembly, maintenance and disassembly of these structures (Figure 1c).

Figure 1. Overview of the three major actin structures in fission yeast.

(a) Fluorescent image of the actin cytoskeleton in a population of fission yeast cells expressing the general F-actin marker GFP-CHD (calponin homology domain of Rng2). (b) Cartoon summarizing the subcellular distribution of actin structures during the cell cycle (centering on mitosis). The table below highlights the basic features and roles of the three actin structures. (c) Venn diagram summarizing the localization of highly conserved actin-binding proteins across actin patches (blue), actin cables (green) and the contractile ring (red). Actin-binding proteins are listed under generic and fission yeast protein names in groups outside the diagram based on their cellular distribution. Within the diagram, proteins overlapping two or more structures (black text) are further categorized using colored + signs to emphasize relative protein levels in each actin structure. Arp2/3 complex: consists of seven different subunits. Capping protein: heterodimer of Acp1 and Acp2. Hip1R: Huntingtin interacting protein-related, talin-like. Myosins and IQGAP associate with light chains (Myo1, calmodulin and Cam2; Myo2 and Myp2, Cdc4 and Rlc1; Myo51, calmodulin and Cdc4; Myo52, calmodulin; Rng2, calmodulin and Cdc4). (d) Regulation of actin filament turnover and myosin motors by tropomyosin and fimbrin ²³, ³³. Actin patches: High concentrations of fimbrin Fim1 prevent tropomyosin Cdc8 from binding the Arp2/3 complex-nucleated branched filaments, which allows efficient cofilin Cof1-mediated actin filament turnover and recruitment of myosin-I Myo1. Actin cables: tropomyosin favors myosin-V Myo52-directed motility on formin For3-nucleated straight parallel filaments. Contractile rings: Lower concentrations of fimbrin allow limited cofilin severing by partially inhibiting tropomyosin. Tropomyosin also favors myosin-II Myo2-mediated compaction of the formin Cdc12-nucleated straight antiparallel filaments.

Box 1. Mating in fission yeast

In addition to interphase actin patches and cables, and the mitotic contractile ring, fission yeast also utilizes the actin cytoskeleton for mating. Upon nitrogen starvation, cells of opposite h⁺ and h⁻ mating types grow projection tips towards each other by polarizing their actin cytoskeleton in response to diffusible signaling pheromones (Figure I) ¹⁰¹. Upon contact of their projection tips, localized cell wall degradation allows mating cells to fuse. Dramatic F-actin rearrangements occur throughout the polarization and conjugation process ¹⁰¹. The actin nucleation and elongation factor formin Fus1 localizes to the projection tip after mating cells contact, and is required for actin filament accumulation/maintenance and for subsequent localized cell wall degradation and cell fusion ^{102, 103}. However, before cell-cell contact the initial accumulation of actin filaments at the tips of polarized mating cells does not require Fus1 ¹⁰².

The organization of individual actin filaments at the mating projection tip and during cell fusion is not known. F-actin and actin-binding proteins localize to structures that resemble actin patches in the mating projection tip ¹⁰¹, suggesting that these structures are analogous to interphase actin patches composed primarily of branched actin filaments. Multiple actin regulators in interphase actin patches have also been implicated in mating by localization and/or genetic phenotypes, including fimbrin Fim1, actin capping protein Acp1/2, and Arp2/3 complex ^{31, 63}. However, in addition to the formin Fus1, other proteins that primarily localize to long-straight actin filaments in actin cables and contractile rings are also implicated in mating, including tropomyosin Cdc8 and the type V myosin Myo51 ^{104, 105}. Therefore, a reasonable speculation is that the mating projection tip and subsequent cell fusion require both patch-like and cable-like actin filament structures. Elucidating how actin is assembled during mating and its role in cell fusion will require improved visualization of the organization and dynamics of the associated actin filaments, as well as a complete list of the associated actin-binding proteins.

Identification of important players in fission yeast has benefited from genetic screens and a small genome with low redundancy. Modification of the genome using homologous recombination allows for easy genetic manipulation and precise localization of fluorescently tagged proteins expressed at endogenous levels (Box 2). Localization studies have been extremely powerful in recent years with the emergence of sophisticated imaging techniques permitting high spatial and temporal resolution tracking of protein dynamics (Box 2). This review summarizes insights gleaned from in vivo experiments, and sophisticated in vitro studies with purified components (Box 3), that are beginning to shape our understanding of how different actin structures function in cells.

Box 2. Key in vivo methods

Manipulating the genome by gene targeting

The efficiency of S. pombe homologous recombination allows gene deletions, truncations, point mutations, inducible promoters, or protein tags to be quickly engineered anywhere in the genome. Precise analysis of phenotypes or protein localization patterns and dynamics dictates the use of endogenous protein levels if one is to avoid artifacts associated with multi-copy plasmids or overexpression. However, tagging is not always ideal since bulky tags can sterically interfere with protein function. Genetic, physiological and morphological tests can be used to determine if the tagged protein is fully functional. A good example here is actin (Figure IIa), where N- or C-terminal GFP tags at the endogenous locus are lethal when tagged actin is expressed as the sole actin source. The reason for the lethality is that GFP-actin incorporates into actin patches assembled by Arp2/3 complex but is unable to incorporate into cables and rings assembled by formins. In this case, expression of the tagged protein at a low level from a construct integrated into a nutritional marker locus, in addition to endogenous unlabeled protein allows visualization of actin dynamics in patches. Visualizing actin organization and dynamics in cables and rings requires other approaches such as immunofluorescence, rhodamine-phalloidin labeling and a GFP-calponin homology domain fusion protein (GFP-CHD).

Quantitative cell biology

Actin-based processes in fission yeast rely on strictly controlled assembly of proteins with highly ordered, reproducible timing and dynamics. Careful quantitation of endogenous protein behavior allows one to compare dynamic processes such as the speed of contractile ring constriction in different backgrounds (Figure IIb), where co-localization with a spindle marker allows the timing of constriction to be assessed relative to mitotic progression. Quantitative approaches are also used to define the transient co-localization of proteins at endocytic actin patches (Figure IIc), or rapid movement of proteins along cables. Use of FRAP (fluorescence recovery after photo-bleaching) allows further classification based on a protein’s ability to exchange between an actin structure and the cytoplasm. In addition to defining a protein’s role, precise in vivo quantitation has been used to determine protein concentrations and formulate mathematical models ^{15, 26, 47, 87, 106}.

Box 3. Key in vitro methods

Combining in vitro approaches with a genetically tractable system can be particularly insightful. Significant progress has recently been made towards understanding mechanistic details of the fission yeast actin cytoskeleton by utilizing in vitro approaches such as (Figure IIIa) total internal reflection fluorescence (TIRF) microscopy of individual actin filaments ^{31, 33, 35, 36, 55, 107}, and (Figure IIIb) myosin motility assays ^{23, 56, 58, 61, 108}.

(Figure IIIa) TIRF microscopy allows direct observation of individual filaments assembling from a pool of fluorescently labeled actin monomers near the surface of a slide or coverslip. Whereas traditional ‘bulk’ actin biochemistry techniques like the pyrene actin assembly assay only detect the concentration of polymer, TIRF microscopy provides the filament number, elongation rate at both ends, and filament architecture (bundles, branches, severing, etc.). For example, the fission yeast cytokinesis formin Cdc12 shown in (a) nucleates the assembly of straight actin filaments, and then remains processively associated with the elongating barbed end while dramatically slowing the elongation rate compared to a neighboring control filament without formin ¹⁰⁷. Moreover, multi-color TIRF microscopy allows simultaneous visualization of more than one component in complex reactions containing several factors that more closely recapitulate in vivo processes.

(Figure IIIb) In addition to employing bulk actin-activated ATPase assays to measure actin binding and motor activity, utilization of in vitro motility assays is also helpful. Motility assays require relatively small amounts of myosin and allow measurement of actin-binding and myosin-driven actin filament gliding rates utilizing epi-fluoresecence microscopy. The approach can be adapted to estimate motor duty ratios and processivity, and also provides a convenient means to test important parameters including ATP concentration and ionic strength.

Actin Patches – Endocytosis

Actin patches assemble at sites of endocytosis, accompanying polarized cell growth at the tips of interphase cells and middle of dividing cells ^{2, 3, 4}, which suggests that endocytic actin patch function accompanies cell wall synthesis and re-modeling. However, the identity of cargo endocytosed by actin patches in fission yeast is not known. In budding yeast, endocytic cargo includes membrane receptors, transporters, permeases, and SNAREs ^{5, 6}. In contrast to animal cells, endocytic internalization in yeast strictly depends on Arp2/3 complex-mediated actin assembly and does not rely on dynamin activity ^{5, 7, 8}, although the dynamin-like Vps1 plays a role in budding yeast endocytosis ⁹. As in animal cells, actin assembly cooperates with BAR (Bin-Amphiphysin-Rvs) and F-BAR proteins to promote membrane invagination and scission ^{5, 7, 10}. Strict dependence of yeast endocytosis on actin assembly is proposed to reflect the need to overcome turgor pressure in yeast cells ¹¹.

Initiation of actin patch assembly

Elegant work in budding yeast revealed that actin patch components assemble in a strictly ordered fashion ^{7, 12, 13}. Fission yeast express homologs of virtually all proteins present in budding yeast actin patches, and those that have been investigated assemble into patches with similar timing as in budding yeast ^{14, 15}. Comparison of endocytic pathways between the two yeasts and animal cells is important for identifying general principles ^{5, 7}.

Patch assembly starts with recruitment of early patch components that are thought to promote initial membrane invagination, including clathrin, Ede1, Yap1801, the epsin Ent1 ¹⁶, and the F-BAR protein Syp1 (Figure 2). Thirty to forty clathrin molecules are recruited 1 to 2 minutes before internalization and may form a cap at the tip of endocytic invagination ¹⁵. Approximately one minute after clathrin appearance, patches begin to accumulate endocytic adaptor proteins Sla1, End3, End4/Sla2, and Pan1 that link early endocytic components to actin assembly. Since End4 and Pan1 are the only endocytic adaptors examined in fission yeast ^{15, 17, 18}, dissecting the other components presents a challenge for future research. End4 is essential for endocytic internalization and polarized growth, while Pan1 is essential for viability. End4 localizes to patches independently of actin and may use its C-terminal talin homology domain to recruit the actin assembly machinery ¹⁷. Pan1 contains a WH2 domain expected to bind actin and a CA (central-acidic) region expected to bind Arp2/3 complex.

Figure 2. Endocytic actin patches.

(a) Time course of endocytic actin patch assembly. Clathrin arrives at patches 100 seconds before internalization and endocytic vesicle scission at time zero, and is accompanied by early endocytic proteins(Early). Endocytic adaptor proteins (Adaptors) are recruited into patches 30–40 s before internalization. The Arp2/3 complex activators, Wsp1-Vrp1 complex and myosin-I Myo1, appear in patches 10 seconds before internalization and are accompanied by proteins thought to regulate their activity (Regulators). Arp2/3 complex is recruited a few seconds after Wsp1 and stimulates a burst of actin assembly and recruitment of actin-binding proteins. Coronin Crn1 arrives with a 5 second delay. F-BAR proteins (BAR) may cooperate with actin in promoting membrane invagination. Upon patch internalization, Myo1 stays behind, clathrin and Wsp1-Vrp1 rapidly dissipate, and some patches associate with actin cables and undergo retrograde flow. (b) Dendritic nucleation model of actin patch assembly and disassembly. Branched filament arrangement is inferred from the in vitro properties of the Arp2/3 complex, and the ultrastructure of budding yeast patches ¹⁴, ¹⁰⁸, ¹⁰⁹. (1) Inactive Wsp1 is recruited to the patch and activated; (2) Wsp1 binds actin monomer; (3) Wsp1-actin binds Arp2/3 complex; (4) ternary complex of Wsp1-actin-Arp2/3 complex binds to the side of actin filament; (5) Arp2/3 complex is activated and nucleates actin filament branch; (6) branch elongates; (7) capping protein Acp1/2 caps filaments; (8) fimbrin Fim1 cross-links filaments; (9) coronin Crn1 binds to filaments; (10) cofilin Cof1 severs filaments; (11) filament fragments diffuse away; (12) Cof1, Aip1, Srv2/CAP and profilin Cdc3 cooperate to disassemble filaments and recycle actin monomers. (c) Patches and contractile actin rings are distinct structures. The image is a 3D reconstruction of Z-series of spinning disk confocal images of red actin patch marker Cam2-mCherry (Myo1 light chain) and green actin ring marker Rlc1-mGFP (Myo2 light chain) in, from top to bottom, a cell with a broad band of pre-ring nodes, a cell with an unconstricted ring, and a cell with a constricting ring. Scale bar, 1 µm.

Arp2/3 complex activation

The ability of Arp2/3 complex to drive actin assembly is stimulated by one of the three nucleation promoting factors, Eps15-like Pan1, WASp Wsp1 and myosin-I Myo1, which are thought to have distinct but overlapping functions ^{15, 19, 20}. Consistent with its role as an adaptor, Pan1 arrives more than 20 seconds before Arp2/3 complex and is not sufficient to drive endocytosis in the absence of Myo1 and Wsp1 ^{15, 19}.

As in budding yeast, compared to Wsp1, Myo1 is a weaker atypical Arp2/3 complex activator that lacks the WH2 domain needed to fuel actin nucleation ^{14, 21}. Myo1 activation of Arp2/3 complex is stimulated by WIP/verprolin Vrp1, which contains a WH2 domain. However, Vrp1 fails to localize to patches in the absence of Wsp1 and likely functions in patch assembly in a complex with Wsp1 ¹⁴. Myo1 localizes at the base of the invaginating plasma membrane where it remains upon patch internalization ¹⁴. Thus, Myo1 may activate Arp2/3 complex in a distinct region around endocytic invagination or play other roles in endocytosis. Patch localization and function of Myo1 depend on F-actin and ATP-dependent motor activity, which is regulated by phosphorylation of the Myo1 motor domain ^{19, 22, 23}. Patch internalization requires myosin-I motor activity in both budding and fission yeast ^{21, 22}. Myo1 also plays a role in organization of sterol-rich membrane domains, but how this relates to Myo1’s role in endocytosis is unclear ²⁴.

Deletion of domains in Pan1, myosin-I and WASP responsible for binding Arp2/3 complex suggest that WASP is the primary Arp2/3 complex activator driving actin assembly for patch internalization in budding and fission yeast ^{21, 22, 25} (V. Sirotkin, unpublished). Mathematical modeling indicates that branched actin assembly driven by a wave of active Wsp1 can account for the time course of actin assembly in cells ²⁶, but requires that the binding of the ternary complex of Wsp1-Arp2/3 complex-actin to actin filaments is 400 times faster in cells than measured in vitro²⁷.

One potential link between membrane deformation and Arp2/3 complex activation is provided by BAR and F-BAR proteins which may cooperate with actin to promote membrane invagination and scission via their ability to stabilize or promote membrane tubulation ²⁸. There are four BAR and seven F-BAR proteins annotated in the fission yeast genome, but not all are involved in endocytosis. The F-BAR protein Cdc15 directly interacts with Myo1 ²⁹. The BAR protein amphiphysin Hob1 inhibits Wsp1 and depends on Wsp1 for localization to patches ³⁰. These interactions may direct Wsp1 and Myo1 activity to specific locations around the endocytic site, providing a potential feedback mechanism between membrane deformation and actin assembly. Other proteins, including Lsb1, Lsb4, Ldb17 and Bbc1, which interact with WASP or myosin-I in budding yeast may help regulate Myo1 and Wsp1 in fission yeast.

Actin patch assembly and disassembly

In fission yeast, Wsp1 and Myo1 arrive to patches 10 seconds before internalization and within seconds recruit Arp2/3 complex, which stimulates a burst of branched actin filament assembly culminating in patch internalization ¹⁴. Mathematical models indicate that patches contain a peak network of ~150 branched filaments that are 100–200 nm long ^{15, 26}. Filaments are short due to rapid capping by capping protein ^{31, 32}. How the actin network is arranged around the endocytic site to promote endocytosis is not clear.

Except for coronin Crn1, all actin-binding proteins accumulate in patches with the same timing as Arp2/3 complex. At peak, each patch contains 900 fimbrin Fim1 molecules that crosslink or bundle filaments composed of ~7000 actin subunits ¹⁵. Fimbrin may strengthen the actin network around the endocytic invagination, explaining why patch internalization fails in fim1Δ cells ³³. Other actin-binding proteins, including transgelin Stg1 and drebrin App1, may help to stabilize the actin network ³⁴. The abundance of fimbrin in patches is striking and plays an important role by preventing patch filaments from binding tropomyosin Cdc8, which otherwise would interfere with patch internalization and turnover by inhibiting myosin-I and cofilin (Figure 1d) ^{23, 33}.

Upon patch internalization, the actin network disassembles in 10 s to allow vesicle fusion with endosomes and to recycle actin. Mathematical modeling suggests that rapid disassembly requires severing of filaments into short fragments that diffuse away ²⁶. Cofilin severs filaments and dissociates Arp2/3-anchored actin branches ^{35, 36}. The yeast cofilin homolog Gmf1 also promotes debranching ^{37, 38}. Coronin arrives 5 s later than actin ¹⁵, and as in budding yeast may co-operate with cofilin and Aip1 in patch disassembly ^39–41. After internalization, patches undergo directed movement that depends on their stochastic association with actin cables and seemingly undirected movement independent of cables ⁴².

Contractile Ring – Cytokinesis

Although it is well established that actomyosin rings drive cytokinesis ^43–45, we are only now beginning to understand how these complex structures work. The late 1990s saw a revolution in our understanding of cytokinesis with the culmination of a number of genetic screens in budding and fission yeast revealing the fundamental players ⁴⁶. These proteins include actin, myosin-II, and a host of other cytoskeletal and signaling proteins conserved from yeast to man. With key components in hand, the next challenge lies in understanding their function, regulation, and how they interface with one another and the cell cortex to control ring assembly and constriction.

Mechanisms of ring assembly

Quantitative analysis of assembling rings and genetic studies have led to the proposal of two independent, yet complementary mechanisms of ring assembly (Figure 3). The first - ‘search, capture, pull, and release‘ (SCPR) - is based on the arrival and dynamics of different components at assembling rings (Figure 3b) ^{47, 48}. The anillin-like Mid1 (Dmf1) provides the spatial cue for ring placement in this model ^{49, 50}, emerging approximately 1 hour before cells enter mitosis at ~65 cortical nodes surrounding the nucleus at mid-cell ^{47, 51, 52}. Just prior to the onset of mitosis these nodes recruit additional Mid1 from the nucleus ^{50, 53}, followed by IQGAP Rng2, myosin-II Myo2, F-BAR protein Cdc15, formin Cdc12, tropomyosin Cdc8, and α-actinin Ain1 ^{52, 54}. Upon entering mitosis, formin Cdc12 nucleates the growth of actin filaments that associate with tropomyosin Cdc8 and undergo rapid elongation ⁵⁵. These filaments contact other nodes and their Myo2 motors. Ring compaction is driven by repeated cycles of Myo2-mediated tugging and release by cofilin Cof1-mediated filament severing^{33, 47, 54, 56, 57}. Multiple layers of regulation appear to converge on Myo2. The conserved UCS (Unc45-/Cro1p-/She4p-related) protein Rng3 works with Hsp90 to activate Myo2 upon its recruitment to the division site ^58–60, while regulatory light chain phosphorylation and tropomyosin Cdc8 enhance Myo2 motility favoring ring compaction ^{56, 61}. α-actinin Ain1 is presumably responsible for the crosslinking of actin filaments into short anti-parallel bundles ^{62, 63}.

Figure 3. Mechanisms of contractile ring assembly.

(a) Schematic of a fission yeast cell illustrating the timing of contractile ring (red) assembly, dwell/maturation, and constriction relative to spindle elongation (black) and SIN activity during mitosis. (b) An adaptation of the ‘search, capture, pull, and release’ mechanism of ring assembly ⁴⁷. Mid1 providesthe spatial cue for Rng2 and Myo2. Rng3 ensures that Myo2 motors are active. Cdc15 and Cdc12 arrive and stimulate actin assembly. Cdc8 binds filaments and promotes Cdc12-mediated actin elongation, limits cofilin Cof1-mediated filament severing, and specifies Myo2 motor activity leading to actomyosin ring compaction. (c) Schematic summarizing ring assembly from actomyosin cables (red) by the SIN-dependent ‘leading cable’ mechanism. The contribution of this mechanism to ring assembly becomes most apparent in a mid1Δ cps1-191 double mutant when the spatial organization of Mid1-dependent nodes is removed and septum formation is delayed by the cps1 mutant ⁶⁵. This delay ensures that ring assembly finishes in time to provide a tight spatial landmark for deposition of the septum. A mechanism by which the SIN communicates with the ring is outlined (inset): SIN promotes Clp1 phosphatase activity in the cytoplasm which (along with other unidentified phosphatases) dephosphorylates Cdc15 ⁷⁰–⁷³. Dephosphorylation promotes self-assembly, membrane deformation, recruitment of formin Cdc12 ²⁹, and associations with paxillin Pxl1 and C2 domain protein Fic1 to promote ring stability ⁷⁸.

The second mechanism of ring assembly (the ‘leading cable’ model) functions in conjunction with the SCPR model. Primary evidence for a second mode of ring assembly is that cytokinesis and growth are supported in mid1Δ cells lacking pre-ring nodes ^{49, 64, 65}. Ring assembly by the leading cable model relies on the septation initiation network (SIN) ^{64, 65}, which explains why rings can be tricked into assembling in interphase by ectopic activation of the SIN ^64–66. This model stems from the observation that under certain conditions actomyosin cables coalesce into rings following growth from a single spot on the cortex (Figure 3c) ^{62, 67}. The SIN is a conserved signal transduction pathway that triggers ring constriction and septum formation at the end of anaphase (Figure 3a) ⁶⁸. While SIN activity peaks at the time of ring constriction, significant activity is still evident earlier in mitosis and ensures that forming rings are compact and homogeneous ^{64, 69}. In the absence of pre-ring nodes (mid1Δ cells) rings are misplaced and disorganized, which often leads to cytokinesis failure ^{49, 50}. However, defects in organization (but not placement) can be suppressed by delaying septum formation (Figure 3). This finding indicates that assembly defects in the absence of Mid1/nodes are due to initiation of septation before an organized actomyosin ring has had a chance to form, as opposed to an inherent inability to support ring assembly. Rings fail to assemble in the absence of both Mid1 and SIN function ⁶⁴, highlighting the over-lapping roles of the two mechanisms.

Communication between the cell cycle and the ring

Two key ring components (Cdc15 and Myo2) are known substrates of the SIN-dependent Cdc14 family phosphatase (Clp1/Flp1) ^{70, 71}. Cdc14 phosphatases move from the nucleolus to the cytoplasm where they dephosphorylate cyclin-dependent kinase substrates, which favors cytokinesis and mitotic exit ⁶⁸. Phosphorylation of Clp1 by the terminal SIN kinase Sid2 promotes its retention in the cytoplasm via an interaction with the 14-3-3 protein Rad24 ^{72, 73}. When ring integrity is compromised, Clp1 activates a cytokinesis checkpoint, during which Clp1 activity is also utilized to stabilize ring structures ⁷⁴.

The N-terminal F-BAR domain of Cdc15 recruits formin Cdc12 during ring assembly and self assembles into filaments that are thought to deform membranes ^{29, 75}. The C-terminal SH3 domain of Cdc15 recruits additional proteins that stabilize the ring ^76–78. Consistent with being a downstream target, Cdc15 is dephosphorylated upon SIN activation and relies on the SIN for localization at the ring ⁶⁴. cdc15 mutants also phenocopy SIN mutants in failing to form homogenous rings ⁶⁴. Amazingly, more than 33 different phosphorylation sites have been identified in the central region of Cdc15, eleven of which are thought to be Clp1 targets ⁷⁵. Dephosphorylation of this region (by Clp1 and unidentified phosphatases) promotes an open conformation facilitating self-assembly, and association with binding partners (Figure 3c) ⁷⁵. Mutagenesis of phosphorylation sites showed that premature dephosphorylation favors interphase assembly of Cdc15 and other key components (e.g., Rng2 and Myo2) at mid-cell. Thus, cell cycle-specific dephosphorylation of Cdc15 during mitosis appears to drive ring assembly in the leading cable model. Although the Myo2 tail is phosphorylated at S1444 ^{61, 79}, mutagenesis of this site causes only minor defects ⁶¹.

Summary

Contractile ring assembly relies on two over-lapping pathways: one built on the spatial organization of ring components provided by the nodes, the other on signaling from the SIN which promotes the function of Cdc15. The fact that Mid1 anchors Clp1 at assembled rings suggests direct cooperation between the two pathways during ring assembly ⁷⁰. Yet many key questions remain. How does Mid1 recruit Myo2 and Rng2 during ring assembly? How do Cdc15 filaments impact the cortex? How is the formin Cdc12 regulated? The IQGAP Rng2 is a multi-domain actin-binding protein essential for cytokinesis ^{80, 81}, but how does it work? What are other key targets of the SIN pathway? While we have learnt a tremendous amount about fission yeast cytokinesis in recent years, there are many more questions than answers.

Actin Cables – Polarity

Actin cables are the third major actin filament based structure, which provide polarized tracks for type V myosin-directed delivery of vesicles and organelles to the expanding cell tips (Figure 4). Actin cables are composed of bundles of short parallel actin filaments, assembled by the actin nucleation and elongation factor formin For3, whose barbed ends are oriented towards the cell cortex (Figure 4) ^82–84. Less is known about the assembly, maintenance and disassembly of actin cables compared with actin patches and contractile rings. However, genetic, cellular and mathematical modeling studies have provided a working model for For3-mediated actin cable assembly (Figure 4) ^85–87.

Figure 4. Actin cable assembly in fission yeast.

(a) Fluorescent image of fission yeast expressing the general F-actin marker GFP-CHD, and a corresponding cartoon diagram of microtubules and actin cables in a single cell. Interphase cells contain ~four dynamic microtubules whose plus ends periodically interact with the cell tip, and deliver polarity factors that direct actin cable assembly ⁹². Actin cables are polarized tracks utilized by myosin-V motors to deliver materials to cell tips ⁹⁹, ¹⁰⁰. Actin cables are bundles of short parallel actin filaments assembled by the formin For3 ⁸²–⁸⁴. (b) Model for actin cable assembly. A-Inactive For3 diffuses to the cell tip (For3: 1 and 2). B-Microtubules deposit the +TIP Tea1/Tea4 complex at cell ends where it recruits the polarisome complex, which includes For3 and its activators Rho-GTPase Cdc42, Bud6 and Pob1 ⁸⁶, ⁸⁹, ⁹³. C-Activated For3 (1) transiently mediates processive actin filament assembly. D-Seconds later For3 (1) is partially inactivated, remains filament-bound but does not facilitate further elongation, and releases from the cortex along with its associated short filament. D and E-Activation and transient processive actin filament assembly by a neighboring For3 (2) pushes partially active For3 molecules and associated filaments inward by retrograde flow as the cable grows ⁸⁵. F-Actin cables are disassembled distal to the cell tip by an unknown mechanism, and inactive For3 and actin monomers are subsequently recycled back to the tip.

Fission yeast actin cables are a wonderful example of cytoskeletal crosstalk, since their assembly site is initially established by microtubules ^88–92. The general scheme is that the polarity factor Tea1–Tea4 complex is transported to the cell tip by microtubule plus ends via a +TIP complex consisting of EB1 Mal3, CLIP-170 Tip1 and kinesin Tea2. Upon deposition at the cell tip through the Mod5 receptor, the Tea1–Tea4 complex recruits polarisome complex polarity factors including the formin For3 and the For3 activators Rho-GTPase Cdc42, Bud6 and Pob1 ^{86, 89, 93}. Establishment of the actin cable assembly site is further complicated by the existence of parallel/overlapping pathways with several positive feedback loops, including signaling through the small Rho-family GTPase Cdc42 ⁸⁸.

Upon activation, For3 is believed to initiate rapid processive actin filament assembly like other formins. Tagging with three tandem copies of GFP revealed that in cells For3-3xGFP has surprisingly transient dynamics ⁸⁵. For3 remains associated with the cell cortex for only a few seconds before moving inwards by retrograde flow with the elongating actin cable at the same rate as actin assembly ⁸⁵. Therefore For3 may facilitate actin filament assembly for only a few seconds before becoming partially inactivated so that it remains associated with the filament, but does not promote further elongation. Partially inactive For3 and its filament are subsequently carried into the cell interior by actin filament assembly mediated by other fully activated For3 molecules at the cell tip (Figure 4). The specific mechanisms by which For3 is activated, partially inactivated, and fully inactivated are not known.

Unlike actin patches and the contractile ring, the list of actin filament binding proteins that localize to actin cables is surprisingly small (Figure 1C). Bona fide actin cable components include the filament stabilizing protein tropomyosin Cdc8 and coronin Crn1. Although it is possible that other factors are absent, it is more likely that (1) they are present at low concentrations, and (2) the negative impact on cables upon their genetic loss/reduction has been difficult to interpret given the notorious difficulty of imaging actin cables in live cells. One obvious missing component is an actin filament crosslinker (Figure 1C and 4), which is likely to be fimbrin Fim1, α-actinin Ain1 or transgelin Stg1. Furthermore, nothing has been reported about the mechanisms by which actin cables are disassembled at the cell interior. Presumably the actin severing factor cofilin Cof1 plays a role, but this has not been established. Determining how Cof1 and other factors disassemble old actin filaments away from the cell tip, but not new actin filaments near the cell tip, is certain to be a fascinating story that may depend on both spatial and temporal differences of an extending actin cable.

Global regulation and self-organization of the fission yeast actin cytoskeleton

Endocytic actin patches, the contractile ring, and polarizing actin cables are each tailored for a particular cellular function in part by unique compositions of actin-binding proteins that determine the arrangement and dynamics of their actin filaments. Actin patches, whose lifetimes are less than a minute, are composed of short-branched filaments (Figure 2). Contractile rings and actin cables, composed of straight filaments bundled in antiparallel or parallel fashion, persist much longer than patches (Figures 3 and 4). How cells assemble and maintain these functionally diverse actin filament structures within the same cytoplasm is a largely unanswered question.

Actin filament organization is partly determined by the nucleation factor. For example, in patches the Arp2/3 complex produces branched filaments with barbed ends oriented towards the membrane where its activators are located (Figure 2). Similarly, the actin cable formin For3 produces parallel filaments because For3 is activated exclusively at the cell cortex (Figure 4). The contractile ring formin Cdc12 produces anti-parallel filaments in the ‘search, capture, pull and release’ model because it is activated from randomly arranged pre-ring nodes. However, differences in actin nucleation factors alone are not sufficient to produce functionally diverse actin filament structures. Filaments within patches, cables and rings are bound by partially overlapping sets of biochemically diverse actin-binding proteins that are integral to the establishment, maintenance, and disassembly of the actin filaments within these structures (Figure 1c). Therefore, a major challenge is to determine how specific subgroups of actin-binding proteins co-localize to a particular structure and collectively influence actin filament organization and dynamics. We hypothesize that by imprinting actin filaments with particular physical conformations ⁹⁴, actin nucleation factors are necessary and sufficient to initiate the recruitment of specific sets of actin-binding proteins. A hierarchy of actin-binding protein recruitment to actin filaments subsequently ensues, whereby upstream factors regulate the association of downstream factors.

Our recent studies revealed that the actin filament bundling protein fimbrin Fim1 and the stabilizing protein tropomyosin Cdc8 are key factors that differentially regulate actin-binding protein recruitment to actin filaments (Figure 1d) ^{23, 33}. Although both Fim1 and Cdc8 bind with similar sub-micromolar affinity to the side of actin filaments ^{55, 81, 95}, their distribution is quite different (Figures 1c and d). Fim1 primarily localizes to actin patches, whereas Cdc8 primarily localizes to actin cables and the contractile ring. We found that Fim1 inhibits Cdc8 from binding actin filaments, and thereby prevents Cdc8 from associating with actin patches ³³, which has important functional consequences (Figure 1d). First, inhibition of Cdc8 by Fim1 ‘deprotects’ actin filaments from severing by cofilin, allowing rapid filament turnover in actin patches and limited filament turnover in the assembling contractile ring ³³. Second, Fim1 relieves the type-I motor Myo1 from Cdc8-mediated inhibition, ensuring maximal Myo1 activity in actin patches ²³. Therefore, in addition to actin filament crosslinking, Fim1 is also an actin-binding “selector” protein that promotes the access of other proteins to actin filaments by inhibiting Cdc8 ³³. Future work will undoubtedly uncover many other mechanisms that contribute to the self-organization of diverse actin filament structures within the same cytoplasm. For example, acetylation of the N-terminal methionine of tropomyosin Cdc8 influences its cellular localization and ability to regulate myosin motors ⁹⁶.

Interpretation of actin-binding protein localization and mutant phenotypes

Interpreting actin-binding protein localizations and mutant phenotypes can be difficult because of close proximity of actin structures to each other, and because actin structures are interconnected by direct interactions and/or indirect feedback mechanisms. For example, several studies reported localization of myosin-I Myo1 and Arp2/3 complex in the contractile ring, and proposed that patches contribute to ring assembly ^{24, 29, 97}. In fact, mutations in actin patch components do reduce the speed of ring constriction ^{33, 48, 97}. However, high-resolution images revealed that Arp2/3 complex and other patch proteins, except Cdc15, are present in dynamic patches in close proximity to the ring but never in the ring itself (Figure 2c), suggesting that actin patch mutants cause cytokinesis defects via feedback between endocytosis and ring constriction rather than a direct role in ring assembly ⁵⁷.

Given that actin structures compete for the same actin monomer pool ⁵⁷, it is also possible that disruption of patch components indirectly affects other structures by modifying the amount of available actin monomers. For example, disruption of patch components Glia maturation factor-like protein Gmf1 and Arp2/3 complex leads to more abundant cables ^{8, 38}, whereas disruption of the patch component actin capping protein Acp1/2 causes cytokinesis and actin cable defects ³¹. Similarly, mislocalization of proteins upon disruption of other actin-binding proteins needs careful consideration. For example, tropomyosin accumulates ectopically in actin patches in the absence of fimbrin ³³. Actin patch abnormalities in the absence of fimbrin had been attributed to the absence of fimbrin-mediated filament bundling ^{63, 98}. However, reduction of Cdc8 in fim1Δ cells largely corrects patch abnormalities, suggesting that the ectopic presence of Cdc8 is largely responsible ³³. It may therefore be necessary to determine how disruption of one actin-binding protein affects the localization of others before accurately ascribing mechanisms. Fortunately, utilizing the genetically tractable fission yeast, where proteins can be tagged and expressed at endogenous levels from their native loci (Box 2), gives researchers the tools to sort out these complicated interactions.

Figure I. Fission yeast mating.

Figure II. In vivo methodology.

Figure III. In vitro methodology.