MMBR Figure table search 04
Home Help [Feedback] [For Subscribers] [Archive] [Search] [Contents]
This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow Copyright Information
Right arrow Books from ASM Press
Right arrow MicrobeWorld
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Moseley, J. B.
Right arrow Articles by Goode, B. L.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Moseley, J. B.
Right arrow Articles by Goode, B. L.
Microbiology and Molecular Biology Reviews, September 2006, p. 605-645, Vol. 70, No. 3
1092-2172/06/$08.00+0     doi:10.1128/MMBR.00013-06
Copyright © 2006, American Society for Microbiology. All Rights Reserved.

The Yeast Actin Cytoskeleton: from Cellular Function to Biochemical Mechanism

James B. Moseley and Bruce L. Goode*

Department of Biology and The Rosenstiel Basic Medical Sciences Research Center, Brandeis University, Waltham, Massachusetts 02454

SUMMARY
INTRODUCTION
ACTIN STRUCTURES AND FUNCTIONS IN YEAST CELLS
    First Contact
    Actin Patches and Endocytosis
        Early observations.
        Actin patch dynamics.
        Actin patch ultrastructure.
        Unified model for actin patch development and function.
        (i) Early recruitment: nonmotile phase.
        (ii) Intermediate stage in patch development: slow motility.
        (iii) Scission and rapid transport of patches.
    Actin Cables: Dynamic Tracks for Polarized Growth
        Introduction.
        Mechanism of actin cable formation.
        The bud tip polarity cap.
        Actin cable architecture.
        Gone in 60 seconds: actin cables are composed of short filaments.
        Actin cable dynamics and turnover.
    Yeast Cytokinesis: Two Complementary Mechanisms for Cell Division
        Introduction.
        Targeting secretion to the bud neck at cytokinesis.
        Formation and closure of the actomyosin ring.
        Role of yeast myosin light chains in cytokinesis.
        Coordinating septum formation with actin ring contraction.
        Septins.
MECHANISMS OF ACTIN FILAMENT ASSEMBLY, ORGANIZATION, AND TURNOVER
    Introduction
        Biochemical properties of actin.
    Assembly of Actin at Cortical Patches
        The Arp2/3 complex.
        NPFs.
        (i) Las17/WASp.
        (ii) Pan1.
        (iii) Myo3 and Myo5.
        (iv) Abp1.
        Coronin: spatial inhibitor of Arp2/3 complex activity.
        Open questions about the Arp2/3 complex.
    Assembly of Actin Cables
        Formins Bni1 and Bnr1.
        Profilin-FH1 interactions: a throttle for filament elongation.
        Bud6: stimulation of Bni1-mediated actin assembly.
        Rho GTPases.
        Other Bni1 and Bnr1 ligands.
    Organization and Stabilization of Actin Filaments
        Bundling proteins.
        (i) Sac6.
        (ii) Scp1.
        (iii) Iqg1/Cyk1.
        (iv) Crn1.
        (v) Abp140.
        (vi) Tef1 and Tef2.
    Nonbundling Actin Stabilizers
        (i) Tropomyosins Tpm1 and Tpm2.
        (ii) Capping protein.
    Rapid Turnover of Actin Structures
        Importance of actin turnover in vivo.
        Cofilin/ADF.
        Aip1.
        Srv2/CAP.
        Profilin.
        Twinfilin.
CLOSING REMARKS AND FUTURE DIRECTIONS
ACKNOWLEDGMENTS
REFERENCES

   SUMMARY
Top
 Next
 References
 
All cells undergo rapid remodeling of their actin networks to regulate such critical processes as endocytosis, cytokinesis, cell polarity, and cell morphogenesis. These events are driven by the coordinated activities of a set of 20 to 30 highly conserved actin-associated proteins, in addition to many cell-specific actin-associated proteins and numerous upstream signaling molecules. The combined activities of these factors control with exquisite precision the spatial and temporal assembly of actin structures and ensure dynamic turnover of actin structures such that cells can rapidly alter their cytoskeletons in response to internal and external cues. One of the most exciting principles to emerge from the last decade of research on actin is that the assembly of architecturally diverse actin structures is governed by highly conserved machinery and mechanisms. With this realization, it has become apparent that pioneering efforts in budding yeast have contributed substantially to defining the universal mechanisms regulating actin dynamics in eukaryotes. In this review, we first describe the filamentous actin structures found in Saccharomyces cerevisiae (patches, cables, and rings) and their physiological functions, and then we discuss in detail the specific roles of actin-associated proteins and their biochemical mechanisms of action.


   INTRODUCTION
Top
 Previous
 Next
 References
 
All cells undergo rapid remodeling of their actin networks to regulate such critical processes as endocytosis, cytokinesis, cell polarity, and cell morphogenesis. These events are driven by the coordinated activities of a set of 20 to 30 highly conserved actin-associated proteins, in addition to many cell-specific actin-associated proteins and numerous upstream signaling molecules. The combined activities of these factors control with exquisite precision the spatial and temporal assembly of actin structures and ensure dynamic turnover of actin structures such that cells can rapidly alter their cytoskeletons in response to internal and external cues. One of the most exciting principles to emerge from the last decade of research on actin is that the assembly of architecturally diverse actin structures is governed by highly conserved machinery and mechanisms. With this realization, it has become apparent that pioneering efforts in budding yeast have contributed substantially to defining the universal mechanisms regulating actin dynamics in eukaryotes.

In this review, we first describe the filamentous actin (F-actin) structures found in Saccharomyces cerevisiae (patches, cables, and rings) and their physiological functions, and then we discuss in detail the specific roles of actin-associated proteins and their biochemical mechanisms of action. For more detailed information on actin-based cellular processes, we refer readers to a number of excellent recent reviews that cover these topics in greater depth: endocytosis (93), mitochondrial inheritance (42), vacuolar inheritance (45), establishment of cell polarity (302, 305, 334), and cytokinesis (22, 33). In some sections we present data and concepts in a historical order, where tracing the progression is instructive. We also note that the majority of the cellular actin structures and their components are conserved between budding yeast and fission yeast, and in many cases insights from fission yeast preceded those from budding yeast. Due to space constraints we have focused the review on the budding yeast actin cytoskeleton, but we highlight experiments from fission yeast to reinforce specific points.


   ACTIN STRUCTURES AND FUNCTIONS IN YEAST CELLS
Top
 Previous
 Next
 References
 

First Contact

The first major insights into organization of the yeast actin cytoskeleton came from two landmark papers in 1984 by Adams, Kilmartin, and Pringle (2, 183). In the preceding years, actin had been purified from yeast cells (393), and it had been established that actin was expressed from a single essential gene in S. cerevisiae, ACT1 (349), but the distribution of actin in yeast cells remained a mystery. Using fluorochrome-labeled drugs (2) and actin antibodies (183), Adams and coworkers visualized filamentous actin in chemically fixed cells and described an "unusual distribution of actin" consisting of two distinct structures (Fig. 1): cortical spots ("patches") and long fibers ("cables"). Their elegant description of actin organization (see details below) set the stage for the recent advances in cell biological analyses on yeast actin described in the following sections: "Numerous small patches or spots were seen distributed over the cell surface; the cortical location of these spots was clear when the microscope was focused up and down. In addition, fibers were seen coursing through the cells, generally in a direction roughly parallel to the long axes of the cells. In unbudded cells, there was often a concentration of fluorescence near one pole, and in cells with very small buds there was generally a pronounced concentration of fluorescence in the bud. In both cases, the concentration of fluorescence was usually resolvable as a cluster of individual spots. In cells with medium-sized buds, there was typically a higher concentration of fluorescent spots in the bud than in the mother cell. Cells with very large buds often showed a concentration of fluorescence in the neck region, sometimes apparently as clusters of spots and sometimes apparently as a band." (Reproduced with permission from reference 2.)


Figure 1
View larger version (100K):
[in this window]
[in a new window]
  		FIG. 1. Cell cycle-regulated organization of the S. cerevisiae actin cytoskeleton. Yeast cells at different stages in the cell cycle contain three visible F-actin structures: cortical actin patches, polarized actin cables, and a cytokinetic actin ring. While patches and cables are visible throughout the cell cycle, the ring is visible shortly before and during cytokinesis. Shown here are actively growing cells from an asynchronous culture that were chemically fixed and stained with rhodamine phalloidin to visualize filamentous actin structures. (Modified from reference 12 with permission of the publisher.)

 
In addition to patches and cables, Adams and coworkers noted the presence of a third actin structure located at the neck of some large-budded cells, which they referred to (above) as a "band." Twenty years later, it is clear that this was the first sighting of the budding yeast actomyosin contractile ring. However, the concept of a budding yeast actomyosin ring lay relatively dormant over the next 14 years, in part due to the fact that actin rings are present only transiently during the cell cycle and therefore observed in only a limited number of cells in an asynchronously growing culture and in part because they can be more difficult to image than patches and cables. In addition, many in the field believed that the narrowness of the bud neck, approximately 1 µm during budding yeast cytokinesis (327), precluded the requirement for an actomyosin contractile ring. Then, in 1998, the presence of a functional actomyosin ring was confirmed by two independent studies (34, 217) which revealed that the events of cytokinesis in yeast and animals were more similar than previously thought.

Studies in 1986 revealed that the three visible F-actin structures in budding yeast—patches, cables, and rings—are also present in the fission yeast Schizosaccharomyces pombe (231). A large body of literature since then has shown that the mechanisms regulating patch and cable assembly in fission and budding yeast are similar but that the rules governing assembly of the actomyosin ring differ significantly (see below). In addition, smaller and/or less visible actin structures may exist in both fission and budding yeast, as suggested by the requirement for actin dynamics in the processes of vacuolar fusion (92), endoplasmic reticulum cortical dynamics (301), and chromatin remodeling (38). Visualizing these actin structures in vivo represents a key challenge for the future.

Actin Patches and Endocytosis

Early observations. Initial observations showing that patches are concentrated at sites of polarized growth suggested that patch function might be linked to secretion. However, patches are decorated with numerous actin-associated proteins, and as mutations in these patch components were characterized, it became apparent that their primary defects were endocytic rather than secretory. Multiple patch components (e.g., actin, Arc35, Rvs167, Sac6, Sla2, and Vrp1) were identified in random screens for mutants defective in endocytosis, end mutants (64, 86, 93, 202, 260, 309, 311). In addition, an endocytic function for actin was suggested by defects in fluid uptake (marked by Lucifer Yellow) upon treatment of cells with latrunculin-A (Lat-A), an actin monomer-sequestering agent (19). Over the following years, genetic studies continued to steadily draw correlations between the functions of patch components and endocytosis, but this linkage was not solidified and fully accepted until recent light and electron microscopy (EM) studies addressed the dynamic movement and ultrastructure of cortical actin patches (see below).

Although it is now apparent that actin patches mediate endocytosis, they may also be coupled to exocytosis. No patch components were isolated in the original sec mutant screen (271), but mutants of a number of patch components show an accumulation of post-Golgi vesicles (139, 213, 257), consistent with models that suggest temporal and spatial links between endocytosis and exocytosis.

Actin patch dynamics. An important and unanticipated property of actin patches was discovered with the advent of green fluorescent protein (GFP) fusions to monitor real-time behavior of proteins in cells. By tagging patch components (Sac6, Cap1/Cap2, Abp1, and actin) with GFP, two initial studies demonstrated that patches are highly motile (80, 387). These and other early studies showed that the lifetime of actin patches is approximately 10 to 20 seconds and that actin patches first assemble at sites of polarized cell growth and then move slowly and nondirectionally along the cell cortex. In addition, more-rapid movements were observed. Patch motility rates ranged from 0.1 to 0.5 µm/s (52, 354, 387).

These observations raised an important mechanistic question: what provides the force driving patch movement? A decade ago, when patch motility was first observed, most forms of actin-based motility were thought to be myosin dependent. So, it came as a surprise when it was reported that the rate of patch motility was unaffected by mutations in any of the five yeast myosin genes: MYO3 and MYO5 (type I), MYO1 (type II), or MYO2 and MYO4 (type V) (354, 387). Instead, studies using Lat-A suggested that the actin filaments in patches undergo rapid turnover (19) and that Lat-A inhibits patch movement (52, 289), suggesting that actin polymerization may provide the force required for patch motility. As such, parallels were drawn between yeast actin patch motility and the actin polymerization-based motility of the intracellular pathogen Listeria monocytogenes, which hijacks the host actin nucleation machinery (the Arp2/3 complex) to assemble a highly branched actin "comet tail" (287, 407), propelling the bacterium at approximately 0.4 µm/s (70, 368). While it has been appealing in models to depict actin patches like miniature Listeria actin comet tails trailing endocytic vesicles, patches have at least two key properties that distinguish them from Listeria. First, cofilin activity, and therefore rapid actin turnover, is required for Listeria motility (287) but not rapid patch movement (204, 287). Second, Listeria motility is autonomous in living cells and cell extracts, requiring only the actin tail that it forms for propulsion. However, the rapid phase of patch motility (see below) relies on an additional separate network of filamentous actin structures, actin cables (159).

From a number of recent studies, it has become apparent that actin patches mature in stages corresponding to different stages of endocytosis and characterized by different types of movement. One of the first indications of this behavior came from a study examining the dual localization and fluorescence resonance energy transfer between CFP-Abp1 (an activator of the Arp2/3 complex) and YFP-Sla1 (an early endocytic protein) (391). This showed that at any given time, a subset of patches colabeled with Abp1 and Sla1, while others contained exclusively Sla1 or Abp1. Intriguingly, the patches labeled exclusively with Abp1 were highly motile, whereas those labeled exclusively with Sla1 exhibited little movement. The authors hypothesized that "Sla1-only" patches might contain endocytic machinery and the arrival of Abp1 (and possibly other factors) facilitated actin polymerization-based rapid movement of patches. This study emphasized the need to consider temporal changes in the development and lifetime of an actin patch.

Subsequent studies by Kaksonen et al. (173, 174) resolved many of the key events in patch development by elegantly correlating temporal changes in patch motility with the arrival of specific components. Using pairs of integrated functional fluorescent tags on different patch components (Las17, Sla1, Sla2, Pan1, Abp1, and the Arp2/3 complex) and computer algorithms to track patch movements, they defined three stages in the lifetime of a patch. First, nonmotile patches, which contain Las17, Sla1, and Pan1, but not actin, form at the cell cortex. Next, polymerized actin appears, which coincides with the onset of slow patch movements at the cortex (0.05 to 0.1 µm/s). Patches transition to a phase of rapid inward movement from the cell cortex, likely as vesicles coated with actin filaments. Addition of Lat-A abolishes both the slow and fast patch movements (stages 2 and 3), suggesting that this motility is actin polymerization based (289, 407). The roles of specific patch components are discussed below in a model for patch development.

Shortly after the initial Kaksonen study (173), Huckaba et al. (159) reported three key findings clarifying patch function and mechanism. First, they showed that endocytic vesicles (marked by the lipophilic marker FM4-64) colocalize with motile patches (marked by Abp1-GFP) during slow and fast phases of movement (Fig. 2A). This provided direct evidence that actin patches are sites of endocytic uptake. Second, they showed that rapid actin patch movement is mediated by polarized actin cables, as internalized actin patches colocalize with cables (Fig. 2B) and move directionally with cables at a similar rate (~0.3 µm/s), and rapid patch movements are lost upon disruption of cables by specific conditional mutants (159). Third, internalized patches/vesicles appear to fuse eventually with endosomal compartments and shed their actin coats. The movement of patches on cables is consistent with an earlier live cell imaging study from fission yeast (289). Thus, slow patch movements in the cell cortex (possibly to facilitate vesicle formation and scission) depend on an Arp2/3 complex-based actin polymerization mechanism, whereas rapid inward movement of patches depends on a mechanism of transporting patches on cables. These findings raise further questions that remain unresolved. What factors link patches to cables? How do cables deliver patches to endosomes, and are endosomes linked to cables? What signal triggers actin coat shedding at endosomes?


Figure 2
View larger version (48K):
[in this window]
[in a new window]
  		FIG. 2. Cortical patches are sites of endocytosis and actin assembly. (A) Colocalization of Abp1-GFP (an actin patch marker) and FM4-64 (a membrane-binding dye/endosomal marker) at the yeast cell cortex. (B) Actin patches (Abp1-HcRed) undergo limited short-range movements while associated with the cell cortex and then transition to a phase of rapid movement upon internalization. The arrowhead tracks the rapid movement of a single actin patch moving over time. The patch moves in a retrograde manner (bud to mother cell) on the end of an actin cable (marked by Abp140-GFP). The rate of fast patch movement is approximately the same as the rate of cable movement (0.2 to 0.4 µm per s), suggesting that patches may be transported passively through attachment to cables. The molecular interactions linking patches to cables are unknown. Bars, 2 µm. (Modified from reference 159 by copyright permission of The Rockefeller University Press.)

 
In animal cell endocytosis, short bursts of actin polymerization at the cell cortex (similar to actin patch assembly) are accompanied by clathrin coat formation. Several yeast actin patch components are physically linked to clathrin and its adaptors, suggesting that clathrin may participate in patch dynamics (21). However, until recently, the direct involvement of clathrin in yeast endocytosis remained controversial, in part because endocytic defects in clathrin mutants are not fully penetrant, and in part because clathrin localization at cortical patches had not been demonstrated (21). In addition, clathrin-coated pits had not been observed in EM studies on the yeast cell cortex (255, 314), but two recent studies have provided convincing evidence that clathrin indeed localizes to early patches to facilitate endocytosis (174, 266). The localization of clathrin at the cell cortex had been masked by abundant clathrin staining on internal membranous structures. This technical limitation was overcome using total internal reflection fluorescence (TIRF) microscopy, which reduces fluorescent background and permits focused observation of fluorescence at the cell cortex. Both studies used TIRF to show that clathrin is recruited to early patches, possibly by Sla2, and then later disappears upon transition of patches to fast inward movement (174, 266). Further, it was shown that almost all clathrin patches at the cortex subsequently recruit actin (i.e., become actin patches) and that clathrin mutants cause delays in early stages of patch development. These studies have ended the debate over whether clathrin functions in yeast endocytosis and raise exciting new questions, such as what signaling mechanisms control recruitment and subsequent loss of clathrin at patches. Another factor that plays a key role in endocytosis in animal cells is dynamin, a GTPase that tubulates membranes and facilitates vesicle scission (298). Although multiple dynamins are expressed in budding yeast, their involvement in endocytosis remains to be demonstrated (93).

Actin patch ultrastructure. Ultimately, to fully understand the inner workings of actin patches, a detailed understanding of patch ultrastructure will be required at a resolution beyond the limits of light microscopy. However, only a few studies using EM have been successful in elucidating details of patch ultrastructure, due to the unique technical challenges presented when analyzing actin in yeast cells: the presence of a cell wall, a concentrated and thus crowded cytoplasm, and a low concentration of actin filaments. Initial EM observations of two-dimensional thin slices revealed that actin patches associate with finger-like invaginations (potential sites of endocytosis) at the plasma membrane (256) (Fig. 3A and B). Additional work on thin slices demonstrated limited colocalization of actin with sites of receptor-mediated endocytosis (255) and disorganized actin filament structures in endocytic mutants (340).


Figure 3
View larger version (69K):
[in this window]
[in a new window]
  		FIG. 3. Electron micrographs of cortical actin patches. (A) Thin sectioning and electron microscopy reveal membrane invaginations at the yeast cell cortex that are enriched for actin (immunolabeled with 10-nm gold particles; arrow) and Abp1 (immunolabeled with 20-nm gold particles). (Reproduced from reference 256 by copyright permission of the Rockefeller University Press.) (B) One model for actin filament organization at cortical patches based on the microscopy in panel A. (Reproduced from reference 256 by copyright permission of the Rockefeller University Press.) (C) Rapid-freeze deep-etch electron microscopy reveals actin filaments at the cell cortex organized into branched, cone-shaped filamentous networks. Immunogold labeling (not shown) revealed that three known patch components (actin, Abp1, and Crn1) are distributed throughout these structures, whereas the Arp2/3 complex is restricted to the peaks or apexes of the mounds. (Reprinted from reference 314 with permission of the publisher.) (D) Model for actin filament organization at patches based on the images in panel C. The arrow indicates the direction of membrane invagination during endocytosis.

 
Using complementary EM techniques, a recent study by Rodal et al. has provided the first three-dimensional information on patch ultrastructure (314). Intact samples of the yeast cell cortex, prepared using quick freezing and deep etching, revealed densely branched networks of actin filaments labeled with antibodies to actin, Abp1, the Arp2/3 complex, and Crn1 (314) (Fig. 3C and D). The caveat to this methodology is that it requires the use of spheroplasted cells. However, spheroplasted cells maintain endocytosis (299), suggesting that the structures preserved and visualized by EM represent functional actin patches. The structures are cone shaped, with the Arp2/3 complex positioned at the apex, and there may be some relationship between these actin mounds and the finger-like invaginations reported by thin sectioning (Fig. 3). In another study, Young et al. (423) used partial cell lysis to leach semi-intact actin patches from cells, purify them through multiple steps, and image them by EM. The actin filaments in these structures form a branched network, and decoration with myosin S1 fragment showed that the filaments have uniform polarity. When combined with the in vivo observations of Rodal et al. (316) showing that the Arp2/3 complex (which binds to filament pointed ends) is positioned at the apex of cone-shaped actin patches, these observations suggest that the majority of actin filament barbed ends in actin patches may be oriented towards the cell cortex. This assignment of filament polarity also is supported by photobleaching experiments in sla2 mutants (173), which block patch internalization and lead to hyperextended patch-like structures. A photobleached band within this extended structure moves away from the cell cortex, suggesting that new monomers are incorporated on barbed ends of filaments at the cortex.

Unified model for actin patch development and function. In this section, we provide a working model for actin patch development and endocytosis, based on the work described above and additional studies introduced below (Fig. 4).


Figure 4
View larger version (46K):
[in this window]
[in a new window]
  		FIG. 4. Model for actin patch development. (Step 1) Receptors recruit early patch components, including clathrin, adaptors, and two NPFs (Pan1 and Las17), to the cell cortex to form a relatively immobile complex. (Step 2) Pan1 and Las17 recruit and activate the Arp2/3 complex to nucleate actin assembly, leading to slow patch movement at the cortex. During this stage, two additional NPFs are recruited, Abp1 and Myo3/Myo5. Abp1 in turn recruits the actin-regulating kinases Ark1 and Prk1, which phosphorylate Pan1 (and likely other patch proteins). Phosphorylation inhibits Pan1 activation of the Arp2/3 complex and triggers dissociation of the Pan1 complex (Pan1, End3, and Sla1). (Step 3) Pan1 phosphorylation and/or the activities of Rvs161, Rvs167, and the type I myosins Myo3 and Myo5 promote vesicle scission and internalization. The Arp2/3 complex, Abp1, Sac6, Cap1/Cap2, and likely other patch components maintain association with the internalized vesicle. Some early patch components are left at the cell cortex, including Las17, Pan1, Sla1, Sla2, Myo3, and Myo5. (Step 4) Endocytic vesicles associate with actin cables by an unknown mechanism and move passively with actin cable retrograde flow (bud to mother). (Step 5) Cable-associated vesicles fuse with an endosomal compartment concomitant with disassembly of the actin coat.

 
(i) Early recruitment: nonmotile phase. The early recruitment process may begin with cytosolic regions of membrane receptors, possibly after being modified by kinases and ubiquitin ligases, associating with endocytic machinery (147, 156, 323, 346, 362). Alternatively, some of the early endocytic machinery may reside in patches at the cell cortex and recruit and concentrate receptors. Supporting this second possibility, a recent study has described the existence of a static complex—termed the eisosome—that marks the future site of actin patch formation and contains the proteins Pil1, Lsp1, and Sur7 (390). It is important to note that actin patches also formed at sites independent of eisosomes, suggesting multiple mechanisms for specifying sites of actin patch assembly. The initial endocytic machinery recruited to patches includes clathrin and multiple scaffolds and clathrin adaptors: Yap1801 and Yap1802 (AP180 homologues), Ent1 and Ent2 (epsin homologues), Ede1 (Eps15R homologue), Scd5, Sla1, and Sla2 (6, 115, 145, 156, 266). These proteins recruit additional early patch components that promote actin assembly, such as Pan1, End3, and Las17, to form a complex network of interacting factors. Specifically, the nucleation-promoting factors (NPFs) (see below) Pan1 and Las17 may directly recruit and activate the Arp2/3 complex to nucleate actin assembly.

Of the many proteins involved in this early step of patch formation, two well-studied components that seem to play central roles are Sla1 and Sla2. Sla1 binds directly to receptors to promote their internalization by a ubiquitin-independent mechanism (156). Sla1 also appears to be important for recruiting other factors to patches, as suggested by sla1{Delta} mutations causing delayed initiation of actin polymerization (i.e., stalling in stage 1). Sla1 interacts with Pan1, End3 (364), Las17 (213), and Sla2 (131). Sla1 has been shown to recruit Sla2 to patches (131). Sla2 localization to patches also requires Scd5, an essential protein that interacts with early patch components and regulates Glc7 phosphatase activity (145).

Sla2 is a multidomain protein that makes numerous contributions to patch development and function. First, Sla2 may help recruit clathrin to patches, as suggested by the following: (i) Sla2 and clathrin colocalizing to early patches (266), (ii) the mammalian homologue of Sla2 (Hip1R) colocalizing with clathrin in vivo and directly promoting assembly of clathrin in vitro (94, 95), and (iii) two-hybrid interactions between clathrin and the Sla2 coiled-coil domain (144). Second, Sla2 may bind to and regulate Rvs167, the yeast homologue of amphiphysin. The BAR domains of amphiphysin/Rvs proteins promote membrane curvature to facilitate vesicle budding in endocytosis (291). Third, Sla2 binds directly to filamentous actin via its carboxyl-terminal talin-like domain (239, 240). Although deletion of this domain from Sla2 causes no obvious defects in endocytosis (403, 420), this fragment of Sla2 is required for cell growth and endocytosis in the absence of specific clathrin adaptors, suggesting an important overlapping role in endocytosis (20). Thus, it will be interesting to learn if and how Sla2 affects actin dynamics and organization. Fourth, through its amino-terminal AP180 N-terminal homology domain, Sla2 binds to phosphoinositide PI(4,5)P (PIP2) to facilitate the internalization step of receptor-mediated endocytosis (360). Finally, as Sla2 binds to Sla1 (18), which in turn binds two NPFs (Pan1 and Las17), it is possible that Sla2 also contributes to the spatial and temporal regulation of Arp2/3 complex activity.

(ii) Intermediate stage in patch development: slow motility. Transition to the second stage in patch maturation, which likely is coupled to membrane invagination, occurs when early patches (marked by Las17, Pan1, Sla1, and Sla2) are joined by the actin nucleation machinery, consisting of the Arp2/3 complex and three new NPFs, Abp1, Myo3, and Myo5 (type I myosins). Coincident with the detection of filamentous actin, patches begin to undergo slow, nondirectional movements within the plane of the cortex. These movements are sensitive to Lat-A, suggesting that the transition to this intermediate stage is driven by actin polymerization (173). Using mutants stalled at this stage of patch development, the measured rates of actin polymerization were found to be 0.05 to 0.1 µm/s (173, 174), markedly slower than subsequent cable-dependent rapid patch movements (0.3 µm/s) (159).

(iii) Scission and rapid transport of patches. Transition to the third phase of patch development correlates with the onset of rapid, directional patch motility, in which patches/vesicles move inward from the cell cortex (173) and simultaneously shed many of their early components (e.g., Sla1, Sla2, Las17, Pan1, Myo3, and Myo5). The signals that trigger the transition from slow-moving to fast-moving patches are unknown, but based on several lines of evidence, Ark1/Prk1 kinases are implicated. These kinases associate directly with Abp1 and Sla2 to localize to patches and directly phosphorylate Sla1 and Pan1, and possibly other patch components (reviewed in reference 355). Phosphorylation of Pan1 by Prk1 inhibits its ability to bind actin and activate the Arp2/3 complex (371). Further, chemically induced loss of Ark1/Prk1 kinase activity in vivo using analog-sensitive kinase mutants causes the rapid and Arp2/3-dependent formation of actin clumps (340). Actin clumps consist of endocytic vesicles that have failed to mature properly and are decorated with Abp1p, Sla2p, Pan1p, Sla1p, and Ent1p. Therefore, Ark1 and Prk1 may negatively regulate the actin assembly-stimulating activity of early endocytic proteins and thereby play a critical role in directing vesicle internalization and the onset of rapid patch movements.

In addition to regulation of Pan1 by Ark1 and Prk1 kinases, there are likely many other factors that contribute to these events. Las17 is shed prior to Pan1 (173), suggesting that Las17 and Pan1 have different roles in vesicle internalization. In addition, mutations in Bbc1, End3, Sla1, and Sla2 (alone and in combinations) show formation of dramatic actin protrusions (174), suggesting that they may contribute to the down-regulation of actin assembly at this stage and/or help promote vesicle release. Consistent with this view, purified Sla1 and Bbc1 directly bind to and inhibit the NPF activity of Las17, and sla1{Delta} and sla2{Delta} cells show defects in patch ultrastructure by EM (314, 315). Deletion of SLA1 results in large, flattened actin patches, and deletion of SLA2 results in large, raised actin patches (314). Proper release of vesicles also requires Myo3, Myo5, Rvs161, and Rvs167, which are implicated in promoting membrane scission (172, 174). Rvs161 and Rvs167 mutants in particular show a striking phenotype where internalizing vesicles are retracted back to the cortex, suggesting a direct role in scission (174).

Once patches/vesicles leave the cell cortex, they move rapidly inward along polarized actin cables, structures described in detail below. The retrograde (bud-to-mother) flow of cables appears to "carry" patches to endosomal sorting compartments (159). The rate of patch transport is similar to the rate of cable flow, suggesting that transport is passive and must involve a physical link between patches and cables (159). The only factors known to associate with fast-moving patches are Abp1, the Arp2/3 complex, Cap1/Cap2, and Sac6 (159, 173). Whether other known F-actin-associated patch proteins and their ligands (e.g., Cof1, Crn1, Srv2, Scp1, Twf1, and Abp140) are also present in these patches remains to be determined. Of these factors, Abp140 and Sac6 are strong candidates for providing linkage between patches and cables, since both are known to decorate patches and cables (3, 17). Further, both of these proteins can cross-link actin filaments in vitro (3, 17), an activity well tailored for bridging two filamentous actin structures.

Finally, some late endosome movements have been reported to require ongoing actin polymerization at the endosome surface. These movements were dependent on both the NPF activity of Las17 and Lsb6 (a Las17-binding protein) (58, 59), and the authors suggested that Arp2/3-dependent actin assembly powers late endosome movements similar to the mechanism employed by Listeria (see above). However, one dilemma with these observations is that Las17 and actin have not been detected on late endosomes, which the authors suggest may be due to their low abundance on such structures. Thus, it remains open whether these actin-dependent movements represent a second, independent class of late endosomes, or instead the defects in late endosome motility in las17 mutants arise from aberrant assembly of endosomes at an earlier stage.

Actin Cables: Dynamic Tracks for Polarized Growth

Introduction. Unlike animal cells, which rely primarily on microtubule-based transport to establish and maintain cell polarity (353), yeast cells use actin-based transport along cables to direct polarized cell growth and to segregate organelles prior to cell division (45). Early in G1, unbudded cells detect cortical landmarks remaining from the previous cell division to select a future bud site, and polarity factors (including the polarity cap; see below) are recruited to this end of the cell. From this site, actin cable assembly is initiated, which leads to reorientation of actin cables and thus targeting of growth and secretion to the future bud tip. Polarized growth towards the bud tip (or cap) continues through a medium-budded stage and depends on actin cables emanating from the bud tip and neck. These cables serve as polarized tracks for type V myosin-dependent delivery of cargos needed to build the daughter cell.

The first demonstration that actin cables are required for polarized cell growth was made possible by using a fast-acting temperature-sensitive tpm1ts tpm2{Delta} mutant strain (306). In this work, Pruyne and coworkers showed a complete loss of cables in tpm1ts tpm2{Delta} cells after a 1-minute incubation at the restrictive temperature. Cells also rapidly lost the accumulation of a secretory marker (Sec4) and a class V myosin transporter (Myo2) at the bud tip. Remarkably, cables reassembled within 1 minute after their return to the permissive temperature, and polarity markers were restored at the bud tip shortly thereafter. Similar phenotypes were observed for the temperature-sensitive myo2-66 allele, which carries a mutation in its motor domain (45, 171, 334). From an examination of the literature, it is evident that mutations in most factors known to influence cable assembly and stability lead to cell polarity defects. These include tropomyosin, Sac6/fimbrin (4), capping protein (10), Srv2 (385), formins (97, 99, 325), profilin (134, 410), and Bud6 (14).

Type V myosin motors (Myo2 and Myo4) transport diverse cargos along actin cables to facilitate polarized growth. Post-Golgi secretory vesicles marked by GFP-Sec4 can be imaged in real time moving directionally on cables towards the bud tip in a Myo2-dependent manner (335). These vesicles carry enzymes that contribute to the production of new cell wall required for bud growth and cell division (1, 343), including glucan synthase, which localizes to actin patches (376). In addition, organelles such as the vacuole, Golgi, nucleus, cortical endoplasmic reticulum, and peroxisomes (305, 421) are delivered to the daughter cell in a cable- and Myo2-dependent manner, and daughter-specific mRNAs (e.g., ASH1) are retained (and possibly transported) in the daughter cell in an actin cable- and Myo4-dependent manner (40, 72). Finally, mitochondria also appear to be transported along cables, and their anterograde movements occur through a myosin-independent mechanism that relies on Arp2/3 complex-based actin polymerization, reminiscent of Listeria motility (104).

Mechanism of actin cable formation. Whereas patch formation relies on actin nucleation by the Arp2/3 complex, cables appear to assemble in an Arp2/3-independent manner (99, 407). Until recent years, it had been speculated that cables might be assembled by a mechanism of filament capture and incorporation, similar to what was proposed for the assembly of actomyosin rings and the bundled actin structures in Drosophila melanogaster bristles. However, recent studies show that cables are assembled by the actin-nucleating activity of formins and profilin. Two key observations led to this discovery. First, it was shown that the formins Bni1 and Bnr1 are acutely required for cable assembly (99, 325), consistent with their localization to sites from which cables emanate (182, 284, 304) (Fig. 5A). When conditional bni1ts bnr1{Delta} mutant cells were shifted to the nonpermissive temperature, all visible actin cables disappeared within 2 min (Fig. 5B), and cables reappeared within 2 to 4 min upon return to the permissive temperature (99, 325). Second, biochemical analyses demonstrated that purified carboxyl-terminal fragments of Bni1 directly nucleate actin assembly in vitro (303, 326). These papers inspired studies on formin activity in other species, and it is now evident that actin assembly-promoting activity is a universal property and function of all formins examined in animals, plants, and fungi (149, 194).


Figure 5
View larger version (72K):
[in this window]
[in a new window]
  		FIG. 5. Formins are required for actin cable assembly at the bud tip and neck. (A) Localization of GFP-labeled yeast formins Bni1 and Bnr1 throughout the cell cycle. (Reprinted from reference 304 with permission of the publisher.) (B) Formin function is required for the assembly and maintenance of actin cables. In a bnr1{Delta} background, point mutations in the FH2 domain of Bni1 render cells temperature sensitive for growth and actin cable formation. Shifting mutants from the permissive temperature (23°C, left panel) to the nonpermissive temperature (35°C, right panel) for 2 to 4 min results in complete loss of visible actin cables but does not affect actin patch distribution in budded (b) or unbudded (u) cells. (Reprinted from reference 99 by permission from Macmillan Publishers Ltd.)

 
Despite these advances, key questions remain regarding cable formation. How are the potent actin nucleation activities of Bni1 and Bnr1 regulated? Studies in mammalian cells suggest that formins are auto-inhibited until released by binding of activated Rho family GTPases to the formin Rho-binding domain (389); however, it has not been determined if Bni1 and/or Bnr1 are auto-inhibited. In yeast, there is evidence to suggest that Cdc42 contributes to formin regulation during early bud emergence and then the partially redundant Rho3 and Rho4 regulate formins during most phases of cell growth, whereas Rho1 may regulate formins in response to cell stresses (79). The specific molecular mechanism underlying this in vivo control remains to be defined. Do Cdc42, Rho1, Rho3, and Rho4 all act by binding to Bni1 and Bnr1 to relieve a proposed auto-inhibited state? Do they act specifically on Bni1 or Bnr1? Are other factors required for activating formins? How are Bni1 and Bnr1 functions regulated by phosphorylation (122, 235)? Once nucleated, how are actin filaments spatially organized into cables? Is this mediated by the proteins that decorate cables, such as Tpm1, Tpm2, Sac6, and Abp140 (17, 84, 219, 306)?

The bud tip polarity cap. The term "polarity cap" refers to a group of interacting cellular factors that localize primarily to the bud tip during bud emergence and growth and have genetic roles in directing polarized cell growth. Subsequently, many of the polarity cap components shift localization to the bud neck, just prior to cell division, where they appear to facilitate cytokinesis by mechanisms described in the next section. Some polarity cap components help to promote cable assembly, as discussed below, while others regulate dynamic membrane trafficking events (exocytosis and endocytosis) and additional aspects of cell polarity at the bud tip; for more details see other recent reviews (54, 138, 157, 302).

One functional complex within the polarity cap network has been referred to as the polarisome. Its three components—Spa2, Pea2, and Bud6—comigrate as part of a 12S complex in cell extracts fractionated by sedimentation velocity (345). In addition, the Rho family GTPase Cdc42 and two Cdc42 effectors, Bni1 and Gic2, appear to function intimately with the polarisome, as supported by numerous two-hybrid and coimmunoprecipitation interactions with Spa2, Pea2, and Bud6 (97, 113, 169). Dozens of other components now are routinely referred to as part of the polarity cap network based on their physical interactions and localization pattern to the bud tip.

The concept of a large macromolecular assemblage directing polarized cell growth raises a number of interesting questions. First, what are all the components? To date, at least 60 proteins have been identified that localize to the bud tip, most of which show physical and/or genetic interactions that suggest their involvement in regulating polarity (165). Undoubtedly, this list will continue to grow. Second, are there stable subcomplexes within the larger assemblage? As mentioned above, at least one stable complex has been identified, containing Spa2, Pea2, and Bud6, and there is evidence that this same complex may contain Gic2 (169) and possibly the Spa2-related protein Sph1 (16, 319). Greater efforts are required now to isolate biochemically and define by mass spectrometry the components of this and other bud tip complexes. Third, how are the interactions among complexes and components spatially and temporally regulated? As has proven to be the case for other large biological machines (e.g., kinetochore, centrosome, and spliceosome), there are likely to be stable subcomplexes as well as dynamically interacting components. Specific components may change interactions in response to molecular signals. Two kinases that regulate cell polarity are implicated in phosphorylating Bni1, a central figure in the polarisome because of its role in nucleating actin assembly. Bni1 is phosphorylated directly by Fus3 kinase and is phosphorylated in a Ste20 kinase-dependent manner in vivo (122, 235). The Bni1-interacting protein Bud6 is phosphorylated (251); further, it associates with Ste11 (a MEK kinase), which regulates polarity (251, 345). In addition, Bni1, Gic2, and the PAK-like kinases Ste20 and Cla4 are directly regulated by interactions with Rho GTPases (169, 170). These provide inspiring leads into the mechanisms regulating polarity cap assembly and function but also likely represent only a small percentage of the total signaling events involved.

A more daunting question, and one that will likely require combined efforts from many laboratories to answer, is what are the activities of each component in the polarity cap network? Some of the relevant activities to address include the following: (i) maintaining association of other components at the cortex, (ii) providing signals to direct localized assembly of actin cables, and (iii) receiving positive feedback signals that help sustain polarity. Several polarity cap components (Bud6, Cdc42, Rho3, Rho4, and profilin) are thought to regulate Bni1 directly to promote actin cable assembly (see below). In addition, Spa2 directly interacts with Bni1 and Pea2 (113, 345), and Spa2 and Pea2 are required for Bni1 localization (284). The biochemical activity of Gic2 is unknown, but it binds to Cdc42 (47, 60) and interacts with Bud6 in the yeast two-hybrid assay (169, 170), it contributes to localization of Bni1 and Bud6 in early bud emergence (169), and gic1 gic2 mutants show defects in establishment of polarity early in bud emergence (47, 60). Since Bud6 regulates Bni1 activity (253), it will be interesting to determine if Gic2 binds directly to Bud6 and whether Bud6-Gic2 interactions influence actin assembly.

With these and other data in mind, we have constructed a working model for the polarity cap, focused on its role in promoting actin cable assembly (Fig. 6). Future models will need to incorporate mechanisms for secretory vesicle docking and fusion, other membrane remodeling events, and cell wall synthesis. In addition, a clear picture of polarity cap architecture and function eventually will require defining (i) the relative abundances of each component, (ii) the stable protein complexes formed among components, (iii) the rules for hierarchical assembly and localization of components at the bud tip, and (iv) functional interactions linking different nodes of the polarity network (e.g., bud site selection, Rho signaling, actin cable assembly, the exocyst complex, secretory apparatus, cell wall synthesis, RAM [regulation of Ace2p activity and cellular morphogenesis] signaling, and mitogen-activated protein kinase signaling) (165). For instance, Msb3 and Msb4 appear to coordinate multiple functions in polarized cell growth, including regulating the activity of the GTPase Sec4 in exocytosis, binding to the central polarity cap factor Spa2, and controlling Cdc42 activity (366).


Figure 6
View larger version (57K):
[in this window]
[in a new window]
  		FIG. 6. Models for regulation of actin cable assembly at the bud tip and neck. (A) Organization of formins (red), actin cables (gold), and septin structures (blue) in a budded yeast cell. Formin proteins (Bni1 at the bud tip and Bnr1 at the neck) directly nucleate actin cable formation. Septins serve as a diffusion barrier between bud and mother cell compartments and provide a scaffold for assembly of many bud neck components. (B) Interaction network at the bud neck: septins (blue); actin (tan); actin-binding proteins (red); all other factors (green). Solid lines represent known physical interactions; dashed lines denote septin-dependent localization. In G1/S, septins recruit other early components to the bud neck. Later in the cell cycle, just prior to cytokinesis, additional factors (late components) are recruited to the bud neck, including components of the bud tip "polarity cap" complex. (C) Model for regulation of actin cable assembly by polarity cap components at the bud tip and cable disassembly by cofilin and Aip1. The polarity cap may be linked to the plasma membrane via components that directly associate with phospholipids, including Cdc42, Cdc24, and Cla4. The actin filaments in cables are nucleated by the formin Bni1, using profilin- and Bud6-bound actin subunits as substrates. Bni1 activity may be regulated by interactions with Cdc42, Rho3, Rho4, Bud6, profilin, Spa2, and other factors (not pictured). Bni1 associates tightly with the fast-growing (barbed) end of the actin filament it nucleates, facilitating insertional growth while protecting ends from capping protein (Cap1/2). Cables are likely comprised of short staggered filaments connected by actin cross-linking proteins, such as Sac6 and Abp140. It is not yet clear what regulates the lengths of actin filaments within a cable. Cables are stabilized along their sides by tropomyosin, which competes with cofilin for binding F-actin. Genetic observations also suggest that Bni1 and Cap1/2 might cap the barbed ends of filaments in cables (see text). As rapidly as cables are formed, they must be disassembled, which recent evidence suggests is mediated by cofilin and Aip1. Barbed-end-directed type V myosins (Myo2 and Myo4) transport vesicles, organelles, and other cargos along cables to the bud tip.

 
Actin cable architecture. Cables must provide polarized tracks for continuous myosin-dependent transport of cargos to the bud neck and tip. Therefore, early models suggested that cables might be stable structures, comprised of long bundled actin filaments, providing relatively stationary tracks for myosin movement. However, it is now evident that cables are comprised of shorter filaments organized into bundles of uniform polarity, with the majority of their fast-growing (barbed) ends oriented towards polarity sites. Evidence for this organization includes the following: (i) a single fluorescently labeled cable has variations in its width, suggesting the thicker regions may be zones of intense filament overlap; (ii) cables are decorated with two known actin filament- bundling proteins, Sac6 and Abp140 (17, 84); (iii) myosin V-dependent transport is unidirectional towards polarity sites (335), and yeast type V myosins (Myo2 and Myo4) are barbed end-directed motors (310); and finally, (iv) formins associate with the barbed ends of actin filaments and localize to the bud tip and neck. The precise arrangement of actin filaments in S. cerevisiae cables remains uncertain, because cables traverse multiple planes within a cell and have eluded description by thin-section EM and tomography. However, the recent use of permeabilized cells for EM in S. pombe has revealed that actin cables are comprised of short overlapping filaments with most of the barbed ends oriented towards the cell tips, consistent with facilitation of barbed end-directed myosin movement (175). Given all of the points above, it seems likely that S. cerevisiae cables will show a similar organization.

Gone in 60 seconds: actin cables are composed of short filaments. Intriguingly, actin cables are lost within 60 seconds of treating cells with the actin monomer-sequestering drug Lat-A (19). This finding demonstrates two important properties. First, the filaments in cables likely undergo rapid turnover, as Lat-A sequesters monomers to inhibit new actin assembly but is not reported to promote filament disassembly. Second, due to the rapid nature of cable loss, the filaments in cables must be relatively short. In fact, their approximate length can be estimated from their disassembly rate using the following simple calculations. A typical large-budded yeast cell is ~10 µm long, with actin cables ~4 µm long. Each actin subunit within a filament contributes 2.77 nm to filament length (155), so if this 4-µm cable is comprised of a single linear filament, it contains ~1,500 subunits. The barbed end of this filament may be capped by formins (see below), but to make a more liberal estimate of filament length we will assume that both filament ends are uncapped. Using the dissociation rates for rabbit skeletal muscle ADP-actin (7.2 s–1 for the barbed end, 0.27 s–1 for the pointed end) and ATP-actin (1.4 s–1 for the barbed end, 0.8 s–1 for the pointed end) (293), a 4-µm filament (1,500 subunits long) would lose about 450 subunits in 1 minute if composed entirely of ADP-actin or 130 subunits for all ATP-actin. In either case, a 4-µm-long filament in a cable would maintain most of its length after 1 minute of Lat-A treatment, instead of the observed complete loss of cable staining (19). Thus, an extremely conservative estimate of average filament length in cables is <1 µm. These estimations do not consider the stabilizing and destabilizing effects of actin-binding proteins that decorate the sides of cables. Nonetheless, these estimates for S. cerevisiae are remarkably consistent with the measured lengths of filaments (0.4 to 0.5 µm) in S. pombe cables as determined by EM (175).

This still leaves many open questions concerning cable regulation. How is the length of filaments in cables controlled? The extent of filament elongation could be restricted by capping proteins. While no conventional barbed or pointed end capping proteins have been visualized on cables, it is possible their decoration is sparse (since they associate only with filament ends) and thus has escaped detection. Cap1/Cap2 (yeast capping protein), which associates tightly with filament barbed ends to block both addition and dissociation of subunits, is a strong candidate to be on cables, because cap2{Delta} cells exhibit dramatically reduced cables (10). This phenotype is consistent with Cap1/Cap2 having a role in stabilizing the filaments in cables. Formins also bind tightly to barbed ends but permit filament growth through a processive capping mechanism and protect growing ends from Cap1/Cap2 (253, 429). Thus, formins may initiate actin nucleation at the bud tip but remain associated with filaments as they are incorporated into cables. Consistent with this model, actin cables undergo retrograde flow away from the cortical sites of polarized growth (see below), and formin-GFP punctae have been visualized on moving actin cables in S. cerevisiae and S. pombe (D. Pellman and F. Chang, personal communications). Thus, the capped state of filaments in cables is an important issue to resolve and raises even more questions: are formins present on all or a subset of filaments in cables? Is Cap1/Cap2 present on some filaments? Do filaments initially have formins associated, but then Cap1/Cap2 displaces them? If so, what regulates these dynamics? Some of the issues may be resolved by emerging advances in light microscopy, which may allow detection of a few (even single) molecules decorating actin filament ends in vivo.

Another possibility to consider is that filaments in cables may be capped at their pointed ends. Tropomodulins perform this function in vertebrate muscle and nonmuscle cells, tightly capping the pointed ends of tropomyosin-decorated filaments (108). Although obvious tropomodulin homologues have not been identified in the S. cerevisiae genome, functional homologues may exist, mutants of which likely would cause diminished cables.

Actin cable dynamics and turnover. The observed rapid turnover of cables suggests that there may be active mechanisms required for disassembling the filaments comprising cables. However, in wild-type cells, the primary filament disassembly factor, cofilin, is not detected on cables. Instead, cables are decorated with tropomyosin, which stabilizes filaments and competes with cofilin for binding F-actin (30, 278). These observations have left the high rate of cable turnover unexplained; however, two studies provide insights into possible turnover mechanisms. First, cofilin was shown to decorate cables in aip1{Delta} cells (317). Second, cofilin and Aip1 were demonstrated to promote rapid cable turnover (276), with cof1-22 and aip1{Delta} mutations causing ~20- and 5-fold decreases in the rate of cable turnover, respectively. This work suggests that most filaments in cables are decorated and stabilized by tropomyosins Tpm1 and Tpm2, but subsets of filaments are stochastically and rapidly disassembled by the combined activities of cofilin and Aip1 (see "Rapid Turnover of Actin Structures" for more details). This leads to a steady thinning or "pruning" of cables along their lengths. Given the discovery of this function for cofilin and Aip1, other factors known to promote filament disassembly and turnover should be tested for their roles in cable turnover (e.g., Srv2/CAP, profilin, and twinfilin).

Using Abp140-GFP as a marker, Yang and Pon have imaged actin cable dynamics in real time in live cells (419). Their work shows that for many cables, one end associates with the bud tip or bud neck while the other end moves away from these assembly sites (in the direction of the mother cell) at ~0.3 µm/s. Some cables appear to assemble at the bud tip and extend through the bud neck into the mother compartment. Other cables appear to move along the cell cortex without having one end attached to the bud tip or bud neck. It is not yet clear whether these "free-roaming" cables have distinct functions and whether they are initially assembled at the bud tip and neck and detach from those sites, or instead form independent of the polarity sites. The direction of cable flow (retrograde, away from the bud tip) is somewhat unexpected given that cables serve as tracks for myosin V-based barbed end-directed transport in the opposite direction (anterograde, toward the bud neck and tip). However, retrograde cable flow (0.3 µm/s) does not appear to pose a major difficulty to overcome, since myosin V-dependent anterograde transport is about 10 times faster—4.5 µm/s in vitro (310) and 3 µm/s in vivo (335).

What provides the force that drives cables away from the bud tip and neck and toward the mother cell? One possible answer is the force generated by actin polymerization. As formins polymerize actin filaments at the bud tip and neck, these elongating filaments may be pushed into the mother cell due to insertional polymerization at their barbed ends capped by Bni1 and Bnr1 (197, 251, 253, 429). If cable flow rate (0.3 µm/s) were directly proportional to actin polymerization rate, this would require filament elongation rates of ~100 subunits/s. This would require 5- to 10-µM monomeric actin in the cytosol, which is unlikely to exist. However, interactions between profilin, actin monomers, and the formins Bni1 and Bnr1 may accelerate cable assembly rates to match rapid cable flow rates (see "Profilin-FH1 interactions: a throttle for filament elongation" below).

In addition, cable flow rates are likely influenced by cellular mechanisms beyond actin polymerization, including a mechanism recently suggested to involve myosin. Domain and sequence similarities among myosin heavy chains define myosin classes, also called types (200, 400). S. cerevisiae expresses five myosins: two type V myosins (Myo2 and Myo4) that transport secretory vesicles, organelles, and mRNAs to sites of polarized growth (bud tip and neck); a single type II myosin (Myo1) that promotes formation and closure of the cytokinetic ring; and two type I myosins (Myo3 and Myo5) that promote actin assembly and endocytosis at cortical patches (22, 126, 305). Recent work suggests that Myo1 has an important role in controlling actin cable dynamics (T. Huckaba and L. Pon, personal communication). Myo1 first localizes to the incipient bud site early in G1 (33, 204, 362) and then is found at the bud neck from the earliest stages of bud formation. Although Myo1 is thought to function primarily in cytokinesis, much later in the cell cycle, Huckaba and colleagues observe reduced cable velocities in the absence of Myo1 motor activity. They propose that Myo1, situated at the bud neck, uses its barbed end-directed motor to "pull" cables through the bud neck and facilitate directional flow toward the mother cell. Thus, cable movement may be regulated by actin polymerization, Myo1 activity, and possibly other factors.

Yeast Cytokinesis: Two Complementary Mechanisms for Cell Division

Introduction. Virtually all animals and fungi utilize an actin-based contractile ring to physically separate the two daughter cells during cytokinesis. Although a cytokinetic actin ring had been observed in fission yeast as early as 1986 (231), the existence of such a division mechanism in budding yeast was realized only in recent years. Early work suggested an enrichment of actin at the site of cell division in yeast (2, 183), but an actomyosin contractile ring was not observed until 1998 (34, 217). In animal cells, myosin II controls contraction of the actin cytokinetic ring (105), which constricts the cleavage furrow to a 1- to 2-µm-wide midbody zone (258, 308, 327). In S. cerevisiae, the bud neck is less than 1 µm throughout the cell cycle; this led to the suggestion that constriction would not occur in yeast and, instead, that targeting of secretory vesicles for membrane and cell wall deposition to the bud neck would be sufficient to drive cell division (327). This view was reinforced by the observation that the single budding yeast type II myosin, Myo1, localizes to the bud neck throughout the cell cycle (394), suggesting that Myo1 may have more of a structural role rather than a dynamic role at the bud neck. Further, since some yeast strains lacking MYO1 were shown to be viable and complete cytokinesis, its role in cytokinesis remained dubious (34, 318, 395). Then, in 1998, two seminal studies imaged Myo1-GFP dynamics in live cells and provided clear evidence that a myosin ring contracts specifically during cytokinesis (Fig. 7) (34, 217).


Figure 7
View larger version (68K):
[in this window]
[in a new window]
  		FIG. 7. Formation and contraction of an acto-myosin ring during yeast cytokinesis. (A) Prior to cytokinesis, actin filaments (rhodamine-phalloidin) and type II myosin (Myo1-GFP) assemble at the bud neck. (B) The actomyosin ring contracts, as demonstrated in consecutive frames of a Myo1-GFP time course at 1-minute intervals. Bar, 2 µm. (Reproduced from reference 34 by copyright permission of the Rockefeller University Press.)

 
It is now evident that two complementary mechanisms contribute to yeast cytokinesis: first, contraction of an actomyosin ring, and second, formation of a septum, achieved by delivery of secretory vesicles to the bud neck to facilitate membrane deposition and cell wall synthesis. It remains unclear how these two mechanisms are coordinated, as discussed below.

The position of actin ring assembly (and the future plane of cell division) in budding yeast is dictated by the position of bud emergence initiated in early G1, where the ring will be assembled at the neck of the newly growing bud (54, 302, 305). The bud site is selected by cortical cues (bud site selection markers) remaining from the previous cell division. Early in G1, these positional cues trigger recruitment and activation of a cascade of factors essential for assembling both the polarity cap at the incipient bud tip and components of the bud neck. At the top of this hierarchy of bud neck components are the septins, which form a ring structure discussed in detail below. Through poorly understood mechanisms, this scaffold in turn recruits Myo1, formins Bni1 and Bnr1, two myosin light chains (Mlc1 and Mlc2), Hof1/Cyk2, IQGAP (Iqg1/Cyk1), and Cyk3 (22). All of these factors arrive at the bud neck before visible actin staining with the exception of Cyk3, which appears during anaphase.

During anaphase, two separate F-actin structures appear at the bud neck, both of which contribute to cell division. First, actin cables become reorganized such that they are polarized towards the bud neck in both the mother and the daughter cell compartments, directing all secretion to the division plane (305). This transition is marked by a dramatic shift of polarity markers (e.g., Cdc42, Bni1, Spa2, and Sec3) from the bud tip to the bud neck, although the mechanism regulating relocation of these factors is unknown. These events lead to membrane and cell wall deposition at the bud neck to form the septum, discussed in more detail below. Second, an F-actin ring forms, which constricts in a Myo1-dependent manner to help close the neck (34, 217). The actin filaments in the cables and the ring are highly dynamic, as both are sensitive to treatment of cells with Lat-A (19, 34, 369).

These observations raise many important questions, including the following. What are the signals at anaphase that induce cable reorientation and actin ring assembly at the neck? What is the architecture of filaments in the ring? What are the signals that trigger ring contraction? How are the activities of cables and the ring coordinated to facilitate cell division? Each of these issues is discussed in greater detail in the following sections.

Targeting secretion to the bud neck at cytokinesis. Prior to anaphase, polarized actin cables are assembled at two locations, the bud tip (where Bni1 localizes) and the bud neck (where Bnr1 localizes). This produces two sets of cables that together target secretion to the bud tip. At anaphase, Bni1 abruptly changes localization, joining Bnr1 at the bud neck (176, 284, 304). In a similar time frame, other components of the polarity cap, such as Spa2 and Bud6, change localization from the bud tip to the bud neck (14, 345). These events redirect secretion to the bud neck, as indicated by relocation to the neck of Myo2, Mlc1, and exocytosis markers such as Sec3 and components of the exocyst complex (305). Targeted delivery of new plasma membrane and cell wall materials provides additional surface area required for cell division. Other key factors delivered to the neck include Chs2 and Chs3, the catalytic subunits of chitin synthase II and chitin synthase III. Chs2 contributes to formation of the primary septum, while Chs3 functions primarily in bud scar chitin synthesis; however, these two enzymes appear to share an essential role in septum formation (63, 343). chs2{Delta} and chs3{Delta} cells are viable but show defects in septum formation and cell division, and chs2{Delta} chs3{Delta} double mutants are inviable and arrest at cytokinesis (343). This demonstrates an essential role for targeted septum deposition and cell wall synthesis in cytokinesis. Remarkably, these secretion-driven pathways are sufficient for completing cytokinesis in the absence of a contractile actin ring, albeit not as efficiently, suggesting that they make the dominant contribution to cell division.

Formation and closure of the actomyosin ring. Actin filaments are among the final components to be assembled into the contractile ring and do not become visible until late anaphase. Early models, based on genetic evidence that the Arp2/3 complex is not required for actomyosin ring formation (408), suggested that preformed filaments may be recruited to the bud neck in order to form the ring (105). However, more- recent evidence suggests that ongoing actin assembly by Bni1 and Bnr1 is required for ring formation and function. While the actin ring forms in either bni1{Delta} or bnr1{Delta} single mutants (378), a temperature-sensitive formin mutant strain (bni1ts bni1{Delta}) fails to assemble the ring (369). Other factors required for ring assembly include profilin and tropomyosin (369). Thus, the machinery driving ring formation is highly similar to that driving cable assembly. In S. pombe, formins are essential for actin ring formation and cytokinesis (57), but the Arp2/3 complex also localizes to the actomyosin ring and makes a contribution to its formation (288), possibly by promoting endocytosis to facilitate septum formation. In contrast, actin ring formation in S. cerevisiae is thought to occur independently of the Arp2/3 complex, and to date the Arp2/3 complex has not been localized to the actin ring. Despite this prevailing model, two regulators of the Arp2/3 complex, Las17 and Vrp1, have known roles in cytokinesis (265, 367), and actin structures that depend on Arp2/3 complex activity have been observed at the bud neck (270), leaving open many possibilities for the mechanism of their involvement.

Three questions regarding the behavior of the assembled actin ring point to important issues for future investigation. First, how are actin filaments in the ring organized? A requirement for actin polymerization by formins suggests that filaments should be unbranched. Further, the classic "purse string" model for ring contraction predicts a network of filaments with antiparallel arrangement to facilitate myosin-based contraction (109). Resolving this will require ultrastructural analysis of partially purified actomyosin rings and/or intact rings in cells. Second, what is the mechanism driving actin ring contraction? One possibility is the motor activity of yeast type II myosin (Myo1). Indeed, a point mutation impairing F-actin binding of the Myo1 motor domain causes cytokinesis defects (T. Huckaba and L. Pon, personal communication). However, deletion of the entire motor domain of Myo1 causes no detectable defects in cytokinesis (224). While these data appear to be at odds, perhaps instead they suggest that the Myo1 motor domain contributes to force production during cytokinesis but can be replaced by other cellular factors (possibly other myosins) when deleted. Another possibility is that nonmotor activities of Myo1 promote ring closure. Recent work in mammalian cells suggests that type II myosin promotes actomyosin ring disassembly rather than contraction (133, 263). Myo1 may employ a similar mechanism to facilitate cytokinesis, but this model awaits testing. It will also be important to identify the signals that regulate Myo1 activity (and thus the timing of ring contraction). Third, how are actin filaments in the dynamic actomyosin ring disassembled? Myo1 may contribute to disassembly, as discussed above, but additional factors are likely involved; one example is cofilin, which functions at the cytokinetic ring in S. pombe (264). Clearly, we have much to learn about how the many ring components that promote actin filament assembly (e.g., formins, profilin), stabilization (e.g., tropomyosin, Sac6), and turnover (e.g., cofilin) are exquisitely regulated to control ring formation and contraction.

Role of yeast myosin light chains in cytokinesis. All myosins are comprised of both heavy and light chains. The heavy chains are large polypeptides (typically 100 to 200 kDa) with a motor domain, multiple IQ motifs, and variable tail regions. The light chains are small polypeptides (~15 kDa) with a calmodulin-like fold that bind to IQ motifs in heavy chains to facilitate motor activity. Further, calcium binding and/or phosphorylation of light chains can regulate their association with heavy chains, providing cells with a mechanism for controlling myosin activity. S. cerevisiae expresses five myosin heavy chains (Myo1, Myo2, Myo3, Myo4, and Myo5) and three light chains (Mlc1, Mlc2, and the calmodulin Cmd1). Below, we focus on their roles in cytokinesis through interactions with Myo1 and Myo2.

Like all type II myosins, Myo1 has an essential light chain (Mlc1) (227) and a regulatory light chain (Mlc2) (33). Differences in the functions of the light chains are apparent from their distinct patterns of localization and mutant phenotypes. Mlc2 colocalizes with Myo1 at the bud neck and requires Myo1 for localization (227). In contrast, Mlc1 localizes to the bud neck independently of Myo1 (342). Deletion of MLC2 has little effect on cell growth or cytokinesis but causes a slight delay in Myo1 ring disassembly (227). In contrast, MLC1 is essential for cytokinesis and cell viability (33). Since myo1{Delta} cells are viable and can undergo cytokinesis, the lethality of mlc1{Delta} suggests that it performs additional functions beyond regulating Myo1. This is further suggested by disruption of Mlc1-Myo1 interactions with specific point mutations in Mlc1 or deletion of the IQ motifs in Myo1, both of which cause only mild phenotypes (227).

What then is the primary Mlc1 function(s) in cytokinesis? Mlc1 binds to at least two other proteins at the bud neck that promote cytokinesis, Myo2 (358) and Iqg1/Cyk1 (IQGAP) (342). Its interaction with Myo2 targets vesicles to fill the bud neck, a key step in formation of the septum (388). However, removal of all six IQ motifs from Myo2 (i.e., the myo2