Biphasic targeting and cleavage furrow ingression directed by the tail of a myosin II

Cytokinesis in animal and fungal cells is performed by the concerted action of a contractile actomyosin ring (AMR) and targeted membrane deposition coupled with ECM remodeling at the division site (Balasubramanian et al., 2004; Strickland and Burgess, 2004). The AMR, which consists of myosin II and actin filaments, is thought to be the engine that drives the ingression of the plasma membrane at the division site during cytokinesis. However, important questions regarding the assembly and function of the AMR remain unanswered. For example, how is myosin II targeted to the division site? How does myosin direct actin ring assembly? How does the AMR coordinate with membrane trafficking during cytokinesis?

The budding yeast Saccharomyces cerevisiae provides an excellent system for structure–function analysis of myosin II and for determining its role in cytokinesis, as deletion of MYO1, the sole myosin II in budding yeast, does not cause cell lethality in most strain backgrounds but produces serious defects in cytokinesis and cell separation (Bi et al., 1998; Schmidt et al., 2002), permitting the in vivo analysis of a variety of myo1 mutations. However, the structure of Myo1 is not known. In this study, we show by rotary-shadowing electron microscopy (EM) that Myo1 forms a dimer with a kink in its tail, similar to myosin IIs in other eukaryotic cells.

To understand the function of a myosin II in cytokinesis, it is important to understand how it is targeted to the division site. Studies in multiple systems indicate that the targeting signals for cleavage-furrow localization reside in the tails of myosin IIs (Sabry et al., 1997; Motegi et al., 2004; Lord et al., 2005; Lister et al., 2006; Beach and Egelhoff, 2009; Ronen and Ravid, 2009). In addition, some proteins, such as Mid1, an anillin-related protein in fission yeast (Motegi et al., 2004; Wu et al., 2006; Huang et al., 2008), and the septins, a family of GTP-binding, filament-forming proteins (Longtine and Bi, 2003; Joo et al., 2005) in budding yeast (Bi et al., 1998; Lippincott and Li, 1998) and mammalian cells (Joo et al., 2007), have been implicated in myosin localization during cytokinesis. However, clear mechanisms that account for the entire myosin localization at the division site have not been elucidated in any organism.

In budding yeast, septins play important roles in Myo1 localization (Bi et al., 1998; Lippincott and Li, 1998). However, how the septin hourglass recruits and maintains Myo1 at the division site before cytokinesis and how Myo1 is anchored to the neck cortex flanked by two septin rings, which result from septin-hourglass splitting at the onset of mitotic exit (Lippincott et al., 2001), during cytokinesis remain unanswered. In this study, we show that Myo1 is recruited to the division site via a biphasic mechanism involving distinct targeting signals in its tail and distinct molecular pathways.

AMR contraction and membrane trafficking are hallmarks of animal and fungal cytokinesis (Balasubramanian et al., 2004; Strickland and Burgess, 2004). The AMR is thought to generate force that powers plasma membrane ingression, whereas targeted membrane deposition is thought to increase cell surface area and also to deliver cargo enzymes for ECM remodeling at the division site, which results in the formation of chitinous primary septum (PS) in budding yeast. These processes must coordinate in time and space to achieve robust cytokinesis. However, the functional and mechanistic relationships between these processes remain poorly understood. In this study, we present evidence to suggest that the AMR functions as a structural unit to “guide” membrane trafficking and ECM remodeling at the division site. This guidance role can be largely fulfilled without the motor domain of Myo1 in budding yeast, which can explain the striking observation made previously that the Myo1 tail is sufficient to support cytokinesis (Lord et al., 2005). The generality of the guidance concept and the motor-independent role of the AMR in cytokinesis are discussed.

To determine the structure of full-length Myo1, we purified a C-terminally TAP-tagged Myo1 from yeast cells, expressed under its own promoter (Fig. 1 A). The TAP-tagged Myo1 construct appears to function normally, as it did not cause any defects in cytokinesis and cell separation (unpublished data). We determined its structure by rotary-shadowing EM. Myo1 formed dimers, each with two globular heads and a rod tail (Fig. 1 A). The average length of Myo1 tail (1073 aa, residues 856–1928) was 126 ± 12 nm (n = 41). In addition, 56% of Myo1 dimers displayed a clear “kink” or “hinge” in their tails, located at 46 ± 6 nm from the head–tail junction, which corresponds to the predicted helix-breaking region (236 aa, residues 1176–1411) in Myo1 tail (Fig. S1). To further validate the Myo1 structure and confirm the kink position in its tail, a C-terminally TAP-tagged Myo1 with a deletion of the TD2 region (174 aa, residues 1224–1397; see next section) that contains all seven helix-breaking prolines in Myo1 tail (Fig. S1, red bar) was purified and its structure was determined. The tail length (118 ± 12 nm, n = 39) of Myo1-TD2Δ was reduced proportionally (Fig. 1 A), and the number of Myo1-TD2Δ dimers with kinked tails was also reduced to 31%, with the kink located at 41 ± 10 nm from the head–tail junction. The TD2 accounts for only a portion of the predicted helix-breaking region, which may explain why deletion of the TD2 reduced, but did not eliminate, the propensity of the Myo1 tail to form a kink.

To determine whether the Myo1 tail is sufficient for dimerization in vitro, Myo1 tail lacking the TD2 region was fused to maltose-binding protein (MBP), a known globular protein. MBP-Myo1-tail-TD2Δ was purified from Escherichia coli (Fig. 1 B) and its structure was visualized by rotary-shadowing EM. Again, MBP-Myo1-tail-TD2Δ molecules formed a “two-headed” structure with a kinked tail (Fig. 1 B).

Together, these data demonstrate that Myo1 is a two-headed myosin II, resembling “conventional” myosins in other eukaryotes. In addition, Myo1 has a clear kink in its tail, similar to myosin IIs from Dictyostelium, Acanthamoeba, smooth muscle, and nonmuscle cells (Trybus, 1991).

To identify potential targeting domains in Myo1, we constructed functional N- and C-terminally GFP-tagged MYO1 alleles and used them to make internal or terminal deletions of MYO1, leading to the identification of two distinct targeting domains, mTD1 (minimal targeting domain 1) and TD2 (targeting domain 2), in Myo1 tail (Fig. 1 C). Myo1 head or any other region of Myo1 tail did not localize to the division site in myo1Δ cells, suggesting that mTD1 and TD2 are the only targeting domains of Myo1. A third region, the putative “assembly domain” (AD) allowed localization only in the presence of endogenous Myo1 (Fig. 1 C). Strikingly, mTD1 localized to the division site in small- to large-budded cells, whereas TD2 localized to the division site only in large-budded cells (Fig. 1 D). Live-cell imaging further demonstrates that mTD1 and TD2 mediate Myo1 localization before and during cytokinesis, respectively (see Fig. 4 A). Thus, Myo1 tail contains two temporally distinct targeting domains.

Myo1 localization to the bud neck depends on the septins (Bi et al., 1998; Lippincott and Li, 1998). In addition, Myo1 tail carried on a high-copy plasmid suppressed the growth defect of various temperature-sensitive septin mutants (unpublished data), suggesting a close functional association of Myo1 with the septins. However, in vitro protein-binding assays using recombinant Myo1 targeting domains (MBP, MBP-TD1, MBP-mTD1, and MBP-TD2) and septin complexes (Cdc3, Cdc10, Cdc11, and HisX6-Cdc12) failed to detect a convincing interaction between Myo1-TDs and the septins, which raises the possibility that a “linker protein” may be involved in Myo1–septin interaction.

Bni5 localizes to the bud neck in a septin-dependent manner from bud emergence in late G1 to spindle disassembly in telophase, which coincides with the onset of cytokinesis (Lee et al., 2002). In addition, both Bni5 and Myo1 appear to function in the same genetic pathway (Lee et al., 2002). These observations raise the possibility that Bni5 may link Myo1 to the septins before cytokinesis. To test this possibility, GFP-tagged Myo1, mTD1, and TD2 were visualized in wild-type and bni5Δ cells. Strikingly, the full-length Myo1 localized to the division site only in large-budded bni5Δ cells, whereas the localization of mTD1 at the division site was completely eliminated (Fig. 2 A). In contrast, the localization of TD2 at the division site was unaffected by bni5Δ (Fig. S2 A). Consistently, Bni5 colocalized with the full-length Myo1 at the bud neck until the onset of Myo1 ring constriction (Fig. 2 B, left; Video 1, left), whereas Bni5 and the mTD1 colocalized at and co-disappeared from the bud neck precisely before cytokinesis (Fig. 2 B, right; Video 1, right). These results demonstrate that Myo1 targeting before cytokinesis completely depends on Bni5.

To determine whether Myo1 and Bni5 interact directly, coimmunoprecipitation and in vitro protein-binding assays were performed. Myo1 tail fragments containing the mTD1 region (Tail, TD1, and mTD1; see Fig. 1 C), but not TD2 or the putative AD, interacted with Bni5 in yeast cell lysates (Fig. 2 C). Recombinant MBP-TD1 and MBP-mTD1 interacted directly with recombinant GST-Bni5 (Fig. 2 D). In contrast, MBP or MBP-TD2 failed to interact with GST-Bni5. As previous results have demonstrated that Bni5 binds to the septin Cdc11 (Lee et al., 2002), these data establish Bni5 as a direct molecular linker between the septins and Myo1 (Septins → Bni5 → Myo1 [mTD1]) before cytokinesis.

The timing of TD2 localization is similar to that of actin ring assembly (Epp and Chant, 1997; Bi et al., 1998; Lippincott and Li, 1998). However, the TD2 still localized to the division site when cells were treated with latrunculin A (Fig. S2 B), a drug that disrupts all F-actin structures including the actin ring (Ayscough et al., 1997), suggesting that the TD2-mediated Myo1 targeting may involve proteins other than F-actin. A number of cytokinesis proteins such as Mlc1, Iqg1, Hof1, Cyk3, Inn1, and Chs2 localize to the division site before or around the time of TD2 localization, making them candidates for TD2-mediated Myo1 targeting. Strikingly, the TD2-GFP, carried on a plasmid, completely failed to localize in mlc1-11 and iqg1Δ cells but still localized in other cytokinesis mutants (hof1Δ, cyk3Δ, hof1Δ cyk3Δ, inn1Δ, and chs2Δ; all these strains contain the endogenous MYO1; Fig. 3 A and unpublished data). More importantly, the GFP-tagged full-length Myo1 disappeared from the bud neck in mlc1-11 and iqg1Δ cells precisely at the time of septin-hourglass splitting (Fig. 3 B; Video 2), which coincides with the onset of mitotic exit or cytokinesis (Lippincott et al., 2001). Thus, the TD2-mediated Myo1 targeting depends on Mlc1 (the ELC [essential light chain] for Myo1) and Iqg1 (IQGAP; Shannon and Li, 2000; Luo et al., 2004). The N-terminal region of Iqg1, which contains an APC recognition site (Ko et al., 2007; Tully et al., 2009) and a CHD (calponin-homology domain; Shannon and Li, 1999), was dispensable for TD2 localization (Fig. 3 C). To date, we failed to detect a convincing interaction between different Iqg1 fragments and the TD2 by two-hybrid and in vitro protein-binding assays (unpublished data), which may reflect complex regulation of this interaction or an interaction that is bridged by other proteins in vivo. Nonetheless, our results, together with the previous observation that Mlc1 binds to the IQ motifs of Iqg1 and is required for the localization of Iqg1 to the division site (Shannon and Li, 2000), establish a pathway (ELC → IQGAP → Myo1 [TD2]) for Myo1 targeting during cytokinesis.

To determine how different targeting mechanisms affect Myo1 localization patterns and dynamics during the cell cycle, we performed 3D time-lapse analyses on cells with intact targeting mechanisms or with one mechanism selectively inactivated. In these experiments, all mutant alleles of MYO1 were the sole source of Myo1 in the cell. Although the overall localization pattern of Myo1 tail, which contains both targeting signals, was similar to that of full-length Myo1 (Fig. 4 A), several differences were noted. Myo1 tail displayed a 21% decrease in the rate of ring “constriction” (Fig. 4 B; Lord et al., 2005; constriction refers to a reduction in the size of Myo1 ring and does not imply the involvement of an actomyosin-based contractile force), and also a defect in ring disassembly (Fig. 4 A; Fig. S3 A; Video 3, compare right to left). Some myo1-Tail cells displayed asymmetric or partial ring constriction (Fig. S3 B). These data indicate that Myo1 head accounts for about one fifth of the constriction rate of the Myo1 ring during cytokinesis and also suggest that Myo1 head contributes to Myo1 disassembly during and toward the end of cytokinesis.

To determine the precise timing and dynamics of Myo1 localization conferred by the early targeting mechanism, we examined myo1-TD2Δ-GFP cells where the region sufficient for late targeting of Myo1 is deleted. myo1-TD2Δ-GFP localized to the incipient bud site and the bud neck as full-length Myo1-GFP did, but its intensity began to fade ∼10–15 min before and disappeared completely from the division site without any constriction around the time of septin-hourglass splitting (Fig. 4, A and B; Video 4), which marks the onset of mitotic exit (telophase–G1 transition; Lippincott et al., 2001). As expected, myo1-TD2Δ-GFP failed to localize to the bud neck in bni5Δ cells throughout the cell cycle (Video 5). These data indicate that the early targeting mechanism of Myo1 is responsible for its localization from late G1 to telophase, and this targeting event completely depends on Bni5.

To determine the precise timing and dynamics of Myo1 localization conferred by the late targeting mechanism, we examined two complementary situations. In bni5Δ cells where the early targeting mechanism is eliminated (Fig. 2, A and B), full-length Myo1 began to appear at the division site ∼15 min and reached maximal intensity ∼5 min before septin-hourglass splitting, and then Myo1 ring constricted with a decreased rate but did not linger around the division site after cytokinesis (Fig. 4, A and B; Video 6, left). These data further demonstrate the role of Bni5 in the early targeting of Myo1, and suggest that the late targeting mechanism of Myo1 is responsible for its localization from anaphase to the completion of cytokinesis.

In a complementary experiment, myo1-TD2-GFP alone localized to the division site in a pattern that parallels that of Myo1 in bni5Δ cells during the cell cycle (Fig. 4 A). However, in contrast to Myo1 tail or full-length Myo1 in bni5Δ cells, very few cells containing myo1-TD2-GFP displayed complete and symmetric ring constriction (Fig. S3 B; Video 6, middle and right). These data indicate that the TD2 is sufficient for Myo1 targeting but not its function during cytokinesis.

In summary, we have established a biphasic targeting mechanism for Myo1 during the cell cycle. The early targeting mechanism accounts for Myo1 localization from late G1 to telophase, which reflects the localization pattern of Bni5 during the cell cycle (Lee et al., 2002; Fig. 2 B). In contrast, the late targeting mechanism accounts for Myo1 localization from anaphase to the completion of cytokinesis, which reflects the localization pattern of Iqg1 during the cell cycle (Epp and Chant, 1997; Lippincott and Li, 1998). These two targeting events overlap from anaphase to telophase when Bni5 and Iqg1 coexist at the bud neck.

To determine the functional consequences of the distinct targeting mechanisms for Myo1, we examined the effects of inactivating each of the mechanisms on the execution of cytokinesis by examining cell cluster formation, which is indicative of a failure in cytokinesis and/or cell separation. As observed previously (Lord et al., 2005), myo1-Tail cells, possessing both targeting domains, displayed a mild defect in cytokinesis (Fig. 5 A). Similarly, bni5Δ cells, in which the early targeting of Myo1 is eliminated, displayed little defect in cytokinesis, suggesting that the late targeting of Myo1 is largely responsible for its role in cytokinesis. This conclusion is further supported by the observation that myo1-TD2 cells, containing only the 174-aa late targeting domain, were able to carry out cytokinesis much more efficiently than myo1Δ cells did. Surprisingly, myo1-TD2Δ cells, in which the mutant myo1 protein disappears invariably by the onset of mitotic exit (Fig. 4 A), were also able to carry out cytokinesis more efficiently than myo1Δ cells did, although much poorly than myo1-tail cells. This result suggests that Myo1 localization before the onset of mitotic exit also contributes to cytokinesis, which is further corroborated by the striking observation that the bni5Δ myo1-TD2Δ cells, in which both targeting mechanisms are inactivated, displayed a cytokinesis defect similar to that of myo1Δ cells (Fig. 5 A). These data indicate that Myo1 tail is largely sufficient for cytokinesis, and both targeting mechanisms contribute.

To explore the underlying mechanism for the role of Myo1 tail and its targeting domains in cytokinesis, we examined actin ring assembly in various myo1 mutants. As expected, 98% of wild-type cells with segregated nuclei formed an actin ring, whereas myo1Δ cells at the similar stage of the cell cycle did not (Bi et al., 1998). Remarkably, 92% of the myo1-Tail cells assembled an actin ring (Fig. 5 B; Fig. S4 A; Video 7), suggesting that Myo1 tail is sufficient for promoting actin ring assembly and/or maintaining actin filaments independently of the head domain, which contains the only known actin-binding site among all myosin IIs. Both targeting domains contribute to actin ring assembly, as cells lacking either were compromised severely in both the frequency and the quality of actin ring assembly. Only 33% and 42% of the myo1-TD2 and myo1-TD2Δ cells formed abnormal actin structures at the division site, which ranged from a smooth actin ring to an amorphous “F-actin cloud” (Fig. S4, A and B). In contrast, 72% of the bni5Δ cells formed an actin ring (Fig. S4 A), suggesting a minor contribution of Bni5 to this process. These data indicate that Myo1 tail is sufficient for actin ring assembly, and again, both targeting domains contribute.

How does the “headless” AMR assembled by Myo1 tail drive cytokinesis? In wild-type cells, AMR constriction (marked by Myo1-RFP) and membrane trafficking (marked by myosin V Myo2-GFP) or trafficking-dependent PS formation (marked by chitin synthase Chs2-GFP) are spatiotemporally coupled during cytokinesis (Fig. 5 C; Video 8). Both Myo2 and Chs2 arrive at the division site around septin-hourglass splitting (Fig. S4 C; Video 9), which coincides with the onset of AMR contraction. The AMR has been suggested to guide PS formation and/or cell surface growth in budding yeast (Vallen et al., 2000; Bi, 2001). We tested the possibility that the headless AMR may retain the “guiding role” even in the absence of its motor domain by examining PS formation in various myo1 mutants using transmission EM (Fig. 5 D; Table I). As expected, wild-type cells formed the PS with high efficiency (100% of the large-budded cells with any septal materials at the division site had a PS) and a sharp orientation (perpendicular to the mother–daughter axis), whereas myo1Δ cells formed asymmetric and/or grossly mis-oriented PS with low efficiency (66%). Remarkably, myo1-tail cells formed PS with high efficiency (100%) and an orientation close to wild-type cells. Thus, the headless AMR is able to guide PS formation. When either targeting domain was deleted such as in myo1-TD2 and myo1-TD2Δ cells, the efficiency and the orientation of PS formation were severely compromised (Fig. 5 D; Table I). Together, these data demonstrate that Myo1 tail is sufficient for directing AMR assembly and PS formation during cytokinesis.

The functionality of Myo1 tail in cytokinesis appears to be correlated with its ability to direct actin ring assembly. This raises an interesting possibility that Myo1 tail may simply help actin filaments organize into a ring structure at the division site, and it is the actin filaments that actually direct trafficking and PS formation by capturing and distributing post-Golgi vesicles around the circumference of the division site. The vesicles are delivered to the division site from the Golgi by the myosin V Myo2 along actin cables (Pruyne et al., 1998; Moseley and Goode, 2006). Thus, the actin filaments in actin cables and actin ring, both of which are nucleated by formins (Vallen et al., 2000; Evangelista et al., 2002; Sagot et al., 2002; Tolliday et al., 2002), may function as a “relay team” to control membrane trafficking and ECM remodeling during cytokinesis. To test this hypothesis, we attempted to selectively eliminate actin ring assembly while leaving actin cables intact. This was achieved by deleting the calponin-homology domain (CHD) in Iqg1. As shown previously (Shannon and Li, 1999), the CHD is required for actin ring assembly (Fig. 6 A). 73% of the IQG1 cells (n = 60) with segregated nuclei formed an actin ring, in contrast to only 9% of the iqg1-CHDΔ cells (n = 43). In addition, Myo1-GFP ring showed no or little constriction before disassembling several minutes after mitotic exit in iqg1-CHDΔ cells (n = 12), in contrast to the normal constriction and disassembly in IQG1 cells (n = 15; Fig. 6 B; Video 10, compare right to left). 100% of the IQG1 and iqg1-CHDΔ cells with any septal materials at the division site formed one or more PS, but the mutant cells were defective in the orientation of the PS (Fig. 6 C). These data suggest that the absence of actin filaments in the AMR causes a slight delay in the initiation of PS formation (as manifested by the lack of Myo1 ring constriction) and a defect in PS guidance. However, the iqg1-CHDΔ cells showed much a milder defect in PS orientation than the myo1Δ cells did (compare to Fig. 5 D), suggesting that Myo1 plays a major role in PS guidance independently of the actin ring. Together, these results indicate that both Myo1 and actin filaments in the AMR are involved in guiding PS formation during cytokinesis.

In this study, we have shown that full-length Myo1 purified from yeast cells forms a two-headed structure, resembling “conventional” myosins from animal cells. This finding is consistent with the conserved nature of this family of motor proteins, and refutes the speculation that Myo1 might be a single-headed myosin (Bezanilla and Pollard, 2000). It also differs from a previous observation showing that a Myo1 fragment (residues 1044–1569) forms trimers and oligomers but not dimers (Lister et al., 2006). This discrepancy could simply reflect different in vitro biochemical behaviors exhibited by a full-length protein versus its fragments. Whether and how Myo1 is assembled into higher-order structures such as mini- and/or thick filaments and whether this assembly is required for cytokinesis will be investigated in the future.

Myosin IIs from Dictyostelium, Acanthamoeba, smooth muscle, and nonmuscle cells all display a “kink” or “hinge” in their tails (Trybus, 1991), which may correspond to a region that is predicted to break the helical structure (Fig. S1). The kink is thought to mediate the formation of an auto-inhibitory loop by individual myosin molecules, which can be opened by regulatory light-chain or heavy-chain phosphorylation for myosin activation and higher-order assembly (Trybus, 1991). Myo1 clearly forms a kink in its tail that corresponds to the predicted helix-breaking region, including the TD2 (Fig. S1). The TD2, which was called the putative hinge in a previous study, was reported to be essential for Myo1 contraction (Lister et al., 2006). The underlying assumption was that Myo1 lacking the TD2 could still localize to the division site but fail to contract during cytokinesis. However, here we showed that the TD2 is actually required for Myo1 localization during cytokinesis, as opposed to being required for Myo1 contraction, per se. The difference is that no cell cycle marker for cytokinesis was used in the previous study. These results suggest that the kink region of a myosin II may play important roles in its localization.

The tail of a myosin II is responsible for its targeting to the division site (Sabry et al., 1997; Motegi et al., 2004; Lord et al., 2005; Lister et al., 2006; Beach and Egelhoff, 2009). However, it has not been clear what interactions recruit and/or maintain myosins at the division site. We demonstrate here that Myo1 undergoes biphasic targeting through distinct molecular pathways during the cell cycle. Such a mechanism may be generally applicable to the targeting of myosin II to the cleavage furrow in other organisms.

The first-phase targeting accounts for Myo1 localization at the bud neck before cytokinesis. At this stage of the cell cycle, Myo1 may facilitate the retrograde flow of actin cables nucleated at the bud cortex in a motor-dependent manner as described previously (Huckaba et al., 2006; Fig. 7 A). Targeting at this stage requires the function of a molecular pathway (septins → Bni5 → Myo1). Basically, Bni5 interacts directly with the septin Cdc11 (Lee et al., 2002) and with the mTD1 of Myo1, thereby targeting Myo1 to the bud neck before cytokinesis. This may be conceptually similar to the initial targeting of the myosin II Myo2 in fission yeast, which requires Mid1, an anillin-related protein that is important for the selection of division plane as well as for efficient AMR assembly (Motegi et al., 2004; Wu et al., 2006; Huang et al., 2008). In both cases, the initial targeting mechanism acts before cytokinesis and requires some proteins that are not well conserved throughout evolution, such as Bni5 and Mid1. Thus, the initial targeting mechanism may have evolved to accommodate the unique life style of a particular organism.

The second-phase targeting accounts for Myo1 localization at the bud neck during cytokinesis. This targeting requires the function of a distinct molecular pathway (Mlc1 → Iqg1 → Myo1). Mlc1 and Iqg1 are the sole ELC and IQGAP in budding yeast. Mlc1 interacts with Iqg1 and is required for Iqg1 localization to the bud neck (Boyne et al., 2000; Shannon and Li, 2000). We show here that Myo1 targeting during cytokinesis depends on Mlc1 and Iqg1 and this targeting is mediated through the TD2 of Myo1. Importantly, this targeting pathway is coupled with AMR assembly during cytokinesis, as all three components of the pathway are required for actin ring assembly (Epp and Chant, 1997; Bi et al., 1998; Shannon and Li, 2000). One possibility is that the rod-shaped Myo1 tail, after being recruited to the division site by Iqg1, serves as a “template” for actin ring assembly. For example, Myo1 tail could assemble into a higher-order structure that enables Iqg1 and other ring components to distribute more evenly along the circumference of the division site. Alternatively, Myo1 tail may activate Iqg1 to bind to actin filaments via its CHD (Shannon and Li, 1999), thereby promoting actin ring assembly.

In fission yeast, Myo2 can still localize to the cell cortex and promote AMR assembly in mid1Δ cells in which the early targeting mechanism is inactivated (Motegi et al., 2004; Wu et al., 2006; Huang et al., 2008), suggesting the existence of a Mid1-independent Myo2 targeting mechanism. This second mechanism likely involves Cdc4, the ELC in fission yeast, as Cdc4 is required for Myo2 localization at the division site whereas Myo2 is not required for Cdc4 localization (Naqvi et al., 1999). Currently, it is not known whether Rng2, the IQGAP in fission yeast, is required for the second-phase targeting by acting in a linear pathway between Cdc4 and Myo2, as Iqg1 does in budding yeast. Importantly, Rng2 is also required for actin ring assembly (Eng et al., 1998; Takaine et al., 2009). Thus, the second-phase targeting mechanism in both organisms acts during cytokinesis and appears to involve the conserved proteins ELC and IQGAP. In Dictyostelium, the tail-mediated myosin II targeting to the division site depends on its higher-order assembly (Zang and Spudich, 1998; Shu et al., 2003). However, how the assembled myosin II filaments find the cell middle remains unknown. In mammalian cells (Beach and Egelhoff, 2009), the tail of a nonmuscle myosin II localizes to the cleavage furrow independently of its assembly domain presumably by interacting with an unknown protein. It would be intriguing to determine whether ELC and IQGAP are involved in myosin II targeting during cytokinesis in Dictyostelium as well as in mammalian cells. Interestingly, the tail of the nonmuscle myosin II in mammalian cells is also able to enrich actin filaments at the division site (Beach and Egelhoff, 2009), suggesting that myosin tail–mediated actin assembly at the division site is not restricted to budding yeast. Taken together, our study and the existing literature suggest that the second-phase targeting mechanism may function to couple myosin localization with AMR assembly during cytokinesis in yeast and beyond.

How can a headless AMR assembled by Myo1 tail achieve cytokinesis with reasonable efficiency? To answer this question, we must consider the forces acting at the division site during cytokinesis.

There are two main forces acting at the division site to power plasma membrane ingression during cytokinesis (Fig. 7 B). First, the contractile force generated by the sliding of myosin motor on actin filaments, which depends on the actin-binding and ATPase activities of the myosin head; this force is traditionally thought to be the most important. Second, the force generated by new membrane deposition coupled with centripetal PS formation in yeast and ECM remodeling in animal cells. In budding yeast, PS formation is essential for plasma membrane ingression and thus cytokinesis (Bi, 2001; Schmidt et al., 2002; VerPlank and Li, 2005). In Caenorhabditis elegans and mice, the synthesis of the ECM component chondroitin proteoglycan is essential for embryonic cytokinesis (Mizuguchi et al., 2003; Izumikawa et al., 2010).

The two driving forces for membrane closure during cytokinesis must be coordinated in time and space to achieve efficiency and spatial precision. Increasing evidence suggests that AMR assembly and/or contraction and membrane deposition appear to be interdependent in fungal and animal cells (Balasubramanian et al., 2004; Strickland and Burgess, 2004). In budding yeast, the AMR is thought to “guide” membrane deposition and PS formation (Vallen et al., 2000; Bi, 2001), which requires a major contribution from Myo1 and a minor contribution from the actin ring (Figs. 5 and 6). How Myo1 fulfills its guidance role independently of the actin ring remains unknown. The actin ring may fulfill its guidance role as part of a “relay team” with the actin cables (Fig. 7 B). Once post-Golgi vesicles are transported by myosin V along actin cables to the division site (Pruyne et al., 1998), this vesicle–motor system switches to the actin filaments in the AMR, leading to a more even distribution of the vesicles and/or their cargoes such as Chs2 along the division site so that membrane deposition and/or PS formation can occur at a sharply defined position and correct timing, leading to efficient furrowing during cytokinesis. Alternatively, the actin ring may affect PS formation indirectly via its potential role in regulating Myo1 dynamics during cytokinesis. The relay mechanism may also operate during animal cytokinesis, except in this case it involves the switching of vesicle/cargo from a microtubule-based long-range transport system to an actin-based short-range transport system. A similar mechanism has been hypothesized for cargo transport in nondividing mammalian cells (Ross et al., 2008). Myo1 is also capable of guiding PS formation independently of the actin ring (Fig. 6), suggesting that Myo1 may interact directly with some of the components involved in PS formation (Nishihama et al., 2009). Together, our data indicate that the AMR functions as a structural unit to guide membrane deposition and PS formation during cytokinesis.

In cells carrying Myo1 tail only, the motor domain–dependent force is eliminated while the force generated by membrane deposition and ECM remodeling remains effective, as Myo1 tail is still capable of directing the assembly of a headless AMR that can largely guide PS formation (Fig. 5), which explains, at least in part, why cytokinesis can still occur without the motor domain.

Why is the motor activity of myosin II not required for cytokinesis in budding yeast but required in fission yeast (Lord et al., 2005) and animal cells (Straight et al., 2003)? The answer may lie in the cell dimension at the division site. The diameter of the division site varies tremendously depending on organisms and cell types, from ∼1 µm in budding yeast (Bi et al., 1998; Lippincott and Li, 1998) to 3.5 µm in fission yeast (Mitchison, 1957), to more than 15–20 µm in animal cells (Carvalho et al., 2009). However, all these organisms or cell types must complete cytokinesis within a similar period of time. Assuming that there is no AMR contraction, for budding yeast, it requires only ∼50 post-Golgi vesicles (the average diameter is 100 nm) for membrane closure during cytokinesis, but would require ∼613 vesicles for fission yeast and ∼20,000 vesicles for an animal cell of 20 µm diameter to achieve the same feat during the same period of time. Thus, even in the complete absence of the AMR such as in myo1Δ cells, cytokinesis can still occur in budding yeast (Bi et al., 1998; Schmidt et al., 2002), presumably due to inefficient and AMR-independent vesicle fusion (and septum formation) at the division site. However, for cells with a large division diameter, the motor activity of the myosin II is presumably required to power the invagination of the plasma membrane in order to reduce membrane requirement and achieve efficient cytokinesis within a given period of time. Thus, the basic forces and mechanisms governing cytokinesis are likely the same in yeast and animal cells, but the proportionality of each force in driving cytokinesis to completion may vary depending on the dimension of the division site for each organism or cell type.

Yeast strains used in this study are listed in Table II. Standard culture media and genetic methods were used throughout this study (Guthrie and Fink, 1991). Where noted, 1 mg/ml 5-fluoroorotic acid (5-FOA; Research Products International) was added to medium to select for the loss of URA3-containing plasmids.

Plasmids used in this study are listed in Table III. All primers were purchased from Integrated DNA Technologies through the Cell Center at the University of Pennsylvania (Philadelphia, PA). Sequencing of constructs was performed at the DNA Sequencing Facility, University of Pennsylvania.

Similar to the construction of plasmid pRS316-N-MYO1-GFP, which contains MYO1 under its own promoter control and carries a NotI GFP^{F64L/S65T/V163A} cassette inserted immediately after the start codon of MYO1 (Caviston et al., 2003), plasmid pRS316-MYO1-C-GFP was constructed by inserting the same NotI-GFP cassette right before the stop codon of MYO1. Both plasmids were used as templates in PCR-mediated mutagenesis to generate plasmids containing different myo1 deletion alleles presented in Fig. 1 C.

Plasmid pRS316-MYO1-mCherry and YIp211-CDC3-mCherry were constructed by replacing the NotI-GFP cassette in pRS316-MYO1-C-GFP and YIp211-CDC3-GFP (Caviston et al., 2003), respectively, with the NotI-mCherry cassette in the plasmid YIp128-CDC3-mCherry (Gao et al., 2007).

Plasmid pUG23-MYO1 was constructed by PCR-amplifying the MYO1 ORF from YCp50-MYO1 (Vallen et al., 2000) and gap-repairing into EcoRI-digested pUG23 (provided by J. Hedgemann, Heinrich-Heine-Universität, Düsseldorf, Germany), generating N-terminally tagged GFP-MYO1 that is under MET25 promoter control. GFP-Myo1 expressed from this plasmid was able to fully complement the cytokinesis defects of the myo1Δ cells even when the cells were grown in standard synthetic media, which contains some methionine that normally represses the MET25 promoter.

Plasmid pFA6a-mCherry-KanMX6 was constructed by replacing the PacI–AscI fragment carrying GFP in plasmid pFA6a-GFP-KanMX6 (Longtine et al., 1998) with the PacI–AscI fragment carrying mCherry from pKT355 (or pFA6a-link-mCherry-His3MX6), provided by K. Thorn (University of California, San Francisco, San Francisco, CA).

Plasmids carrying C-terminally TAP-tagged myo1 alleles were made in several steps. First, a C-terminal TAP tag (CBP-TEV-ProtA) was PCR-amplified from plasmid pBS1479 (Puig et al., 2001), generating a 588-bp TAP fragment flanked by symmetrical restriction sites that were introduced in the PCR primers, HindIII–NotI–TAP–NotI–HindIII. This fragment was then digested by HindIII and subscloned into pUC18 (Life Technologies Co.) at the HindIII site, yielding the plasmid pUC18-C-TAP. The correctness of the plasmid was confirmed by restriction-enzyme digestion and sequencing. Plasmids pRS316-MYO1-C-TAP and pRS316-MYO1-TD2Δ-C-TAP were constructed by replacing the NotI-GFP cassette in plasmids pRS316-MYO1-C-GFP and pRS316-MYO1-TD2Δ-C-GFP (see above) with the NotI-TAP cassette isolated from plasmid pUC18-C-TAP.

Plasmids pMAL-MYO1-TD1, pMAL-MYO1-mTD1, and pMAL-MYO1-TD2, which express MBP-tagged Myo1 fragments, were constructed by PCR amplifying myo1 fragments (myo1-TD1, myo1-mTD1, and myo1-TD2) from the derivatives of pRS316-MYO1-C-GFP described above and subcloning as SalI–PstI fragments (SalI and PstI sites were introduced in the forward and reverse primers, respectively) into SalI- and PstI-digested pMAL-c2 (New England Biolabs, Inc.). Plasmid pGEX-5X-BNI5, which expresses GST-tagged Bni5, was constructed by first PCR amplifying and gap repairing BNI5 ORF into the EcoRI site of plasmid pJG-4-5 (Gyuris et al., 1993), and then subcloning a BamHI–XhoI fragment carrying BNI5 ORF from pJG4-5-BNI5 into the corresponding sites of pGEX-5X-1 (GE Healthcare).

Plasmid pRS315-IQG1 was constructed by cloning a 7-kb SacI–SalI IQG1 fragment from p1868 into SacI–SalI-digested pRS315 (Korinek et al., 2000). Plasmid pRS315-IQG1-GST, which expresses C-terminally GST-tagged Iqg1, was constructed by PCR amplifying GST-His3MX6 from pFA6a-GST-His3MX6 (Longtine et al., 1998) and transforming the PCR product into yeast cells carrying pRS315-IQG1. Candidate plasmids were rescued from yeast cells and transformed into Escherichia coli for confirmation by colony PCR and then further confirmed by sequencing. Plasmid pRS315-IQG1(1–697)-GST was constructed in a similar way except that the PCR-amplified GST-His3MX6 was inserted in-frame after amino acid 697 of Iqg1, resulting in the deletion of residues 698–1495 from Iqg1. Plasmid pRS315-IQG1(Δ2-411)-GST was constructed by a PCR-based method using appropriate mutagenic primers that resulted in the removal of residues 2–411 from Iqg1.

We developed a genetic system in which the endogenous MYO1 can be precisely replaced by any myo1 allele containing truncation(s) or point mutation(s) without leaving any undesired nucleotides at the MYO1 locus. Recipient strain Masa1243 [a myo1Δ::URA3-KanMX6 (pUG23-MYO1)] was constructed by first using a URA3 PCR fragment containing 300-bp upstream sequences and 150-bp downstream sequences of URA3 to replace MYO1 ORF in the wild-type strain YEF473A carrying plasmid pUG23-MYO1, and then inserting KanMX6, which was PCR amplified from pFA6a-KanMX6 (Longtine et al., 1998), after the URA3 at the myo1Δ locus.

Plasmids pRS316-N-MYO1-GFP and pRS316-MYO1-C-GFP and their derivatives carrying different myo1 alleles were digested with SalI and ClaI, and then transformed into Masa1243. The transformation mixtures were plated out on SC-His plates and incubated at 25°C for 2 d. After cells were grown to confluency on SC-His plates, the cells were replicated onto SC-His+ 5-FOA plates to select for Ura⁻ cells, and then screened for G418 sensitivity on YPD+G418 plates to get Ura⁻ Kan^s cells in which the URA3-KanMX6 cassette at the myo1Δ locus was replaced by the incoming myo1 allele. The resulting colonies were then grown in non-selective YPD medium to lose the plasmid pUG23-MYO1.

Myo1-C-TAP and Myo1-TD2Δ-C-TAP were purified as described previously (Puig et al., 2001), with some modifications. Eight liters of myo1Δ cells carrying pRS316-MYO1-C-TAP or pRS316-MYO1-TD2Δ-C-TAP were grown in SC-Ura medium at 23°C to an OD₆₀₀ = 2.0–3.0, collected by centrifugation, washed twice with water, resuspended into 40.0 ml cell lysis buffer (CLB; 50 mM Hepes-KOH, pH 7.9, 150 mM NaCl, 5 mM MgCl₂, 5 mM ATP, 0.1% NP-40, and 2 mM Benzamidine), and added 1.0 ml 100x protease inhibitor cocktail (with stock concentrations of 50 µg/ml leupeptin, 140 µg/ml pepstatin A, 240 µg/ml chymostatin, 1.7 mg/ml aprotinin, 0.25 mg/ml antipain, and 35 µg/ml bestatin) and 0.5 ml 100× phosphatase inhibitor cocktail I (Sigma-Aldrich). The cell suspension was passed through a French press three times with a pressure ≤ 1,200 psi to break down the cells, which was followed by centrifugation at 100,000 g for 150 min at 4°C to clear the cell lysate. The supernatant was then mixed with 500 µl IgG beads (GE Healthcare), which had been prewashed five times with 5.0 ml CLB. The mixture was incubated with rotation at 4°C overnight. The beads were spun down, washed five times each with 10.0 ml CLB, once with 10.0 ml TEV cleavage buffer (10 mM Hepes-KOH, pH 7.9, 150 mM NaCl, 0.5 mM EDTA, 1 mM DTT, and 0.1% NP-40), resuspended in 1.0 ml TEV cleavage buffer plus 150 U AcTEV Protease (Life Technologies Co.), and then incubated with rotation at 4°C overnight. Approximately 1.0 ml of supernatant containing Myo1 was separated from IgG beads by passing through a column. The column was then washed five times each with 2.0 ml calmodulin binding buffer (CBB; 10 mM Hepes-KOH, pH 7.9, 150 mM NaCl, 1 mM Mg acetate, 1 mM imidazole, 2 mM CaCl₂, 10 mM 2-mercaptoethnol, and 0.1% NP-40). The supernatant and the flow-through from the washes were combined, totaling ∼11.0 ml in volume, to which 10 µl 1.0 M CaCl₂ was added. This protein solution was mixed with 100 µl Calmodulin Affinity Resin (Agilent Technologies), which had been prewashed three times each with 5.0 ml CBB. The mixture was incubated with rotation at 4°C for 2 h, and then passed through a column, which was followed by three washes each with 10.0 ml CBB. The protein was then eluted 10 times each with 200 µl calmodulin elution buffer (10 mM Hepes-KOH, pH 7.9, 150 mM NaCl, 1 mM Mg acetate, 1 mM imidazole, 20 mM EGTA, 10 mM 2-mercaptoethnol, and 0.1% NP-40).

The combined eluents from the experiments above were passed through Amicon ultra centrifugal filter devices with suitable MWCO (Millipore) to change the buffer into 2x EM buffer (300 mM KCl, 2 mM MgCl₂, 2 mM EGTA, 2 mM DTT, 20 mM imidazole-HCl, pH 7.0, and 400 µM ADP), which was then diluted by equal volume of 100% glycerol to make Myo1 preparation in 1x EM buffer for EM observation (see below). The final concentrations of Myo1-C-TAP and myo1-TD2Δ-C-TAP in the EM buffer were ∼10 µg/ml.

100 OD₆₀₀ of yeast cells grown exponentially in SC-His media were disrupted by vortexing with glass beads at 4°C in 500 µl of 1x PBS buffer (137 mM NaCl, 2.7 mM KCl, 10 mM sodium phosphate dibasic, and 2 mM potassium phosphate monobasic, pH 7.4) containing 0.1% NP-40 and a cocktail of protease inhibitors (5x stock) and phosphatase inhibitors (5x cocktail-I stock) for yeast (Sigma-Aldrich). After centrifugation to remove cell debris, the supernatant was used for immunoprecipitation with 1 µl rabbit polyclonal GFP antibody (Abcam) and 30 µl protein A agarose beads (Life Technologies Co.). After incubation at 4°C for 12–16 h, beads were washed five times each with 200 µl lysis buffer without protease and phosphatase inhibitors, and proteins bound to the beads were boiled with 100 µl SDS-sample buffer. The precipitated Myo1-GFP variants, the total Bni5-HA in cell lysates, and the bound fractions of Bni5-HA were analyzed by SDS-PAGE and blotted with anti-GFP, and anti-HA monoclonal antibodies, respectively (Covance).

E. coli strain BL21(DE3) containing appropriate constructs for expressing recombinant proteins was grown in 30–600 ml LB plus 100 mg/ml ampicillin at 37°C to OD₆₀₀ ≤ 1.0. Expression of recombinant proteins was induced with 1 mM IPTG at 23°C for 3 h. Cells were then collected by centrifugation, washed twice with water, resuspended into 5–20 ml cell lysis buffer (CLB [the CLB for MBP-tagged proteins: 20 mM Tris-HCl, pH 8.0, 100 mM NaCl, 5 mM EDTA, 5 mM DTT, 0.1% NP-40, and 2× complete protein inhibitor cocktail tablets; Roche Diagnostics Co.; the CLB for GST-tagged protein: 1x PBS, 5 mM EDTA, 5 mM DTT, 0.1% NP-40, and 2× complete protein inhibitor cocktail]). Cells were then broken by sonicating for 7 x 15 s with 15-s interval on ice, centrifuged at 15,000 rpm for 20 min at 4°C. Supernatant was collected and incubated with prewashed amylose beads (New England Biolabs, Inc.) for MBP-tagged proteins or glutathione Sepharose 4B beads (GE Healthcare) for GST-tagged proteins at 4°C for 1 h. Beads were then washed three times each with 5.0 ml corresponding CLB buffer. MBP-tagged proteins were resuspended in 1.0 ml CLB. GST-tagged proteins were eluted five times each with 200 µl elution buffer, which is the CLB for MBP-tagged proteins, except containing 10 mM reduced glutathione and lacking the protease inhibitors.

The concentrations of the purified recombinant proteins were estimated by comparing the sample proteins to bovine serum albumin (BSA) of known concentrations by SDS-PAGE analysis, followed by staining the gel with simple blue (Life Technologies Co.). Approximately 40 µg MBP-tagged proteins that were still bound on beads were added to different 1.5-ml Eppendorf tubes, the amount of beads was normalized by adding more prewashed beads, and then incubated with 500 µl 5% BSA at 4°C for 1 h to block nonspecific binding by the beads. After a brief centrifugation to pellet the beads and remove the supernatant, the beads were mixed with 10 µg GST-tagged protein. The CLB for MBP-tagged proteins without the protease inhibitors was added to the reaction to make the final volume of 500 µl. The binding reaction was incubated with rotation at 4°C for 1 h, and then spun down to pellet the beads, which were then washed three times each with 500 µl binding buffer at 4°C for 10 min. The final protein complexes were eluted by adding 100 µl 2x SDS sample buffer and boiled for 5 min, resolved by SDS-PAGE and analyzed by Western blotting.

For EM observation, MBP-myo1-Tail-TD2Δ was purified essentially as described above except that 600 ml culture was used and the recombinant protein was eluted from the amylose beads using maltose. The buffer for the protein suspension was changed into 1x EM buffer as described for TAP-tagged Myo1 above. The final concentration of MBP-myo1-Tail-TD2Δ was ∼100 mg/ml.

All cells were in exponential phase for imaging experiments. Differential interference contrast (DIC) and fluorescence microscopy were performed using a microscope (model E800; Nikon) equipped with a Plan Apochromat 60x/1.40 NA oil immersion objective (Nikon), a CCD camera (model C4742-95; Hamamatsu Photonics), and Image-Pro Plus software (Media Cybernetics).

For time-lapse microscopy, cells were spotted onto a thin layer of SC medium + 2% agarose on a microscope slide. Individual cells were followed at 1-min interval using IPLab software (BD) and a spinning-disk confocal microscope system comprising a scanner (CSU 10; Yokogawa), a microscope (IX 71; Olympus), a Plan S-Apo 100x/1.4 NA oil immersion objective (Olympus), and an ImagEM back-thinned EM CCD camera (C9100-13; Hamamatsu Photonics); components were integrated by Bio-Vision Technologies. Diode lasers for excitation (488 nm for GFP; 561 nm for RFP) were housed in a launch constructed by Spectral Applied Research. For each time point, 10 z-sections were acquired at 0.2- or 0.3-mm increments at 23°C. Maximum projection of images for each time point was generated using IPLab software. The contrast and the brightness of the images were enhanced using NIH ImageJ, and the finalized time-lapse data were presented as montages using ImageJ or as movies using MetaMorph (MDS Analytical Technologies). The constriction rate of Myo1 ring was determined as the diameter of Myo1 ring before constriction minus the diameter of the first Myo1 dot near the end of constriction divided by the time required to undergo this change. Only cells with symmetric Myo1 constriction were used in the calculation.

For observation of Myo1-C-TAP, Myo1-TD2D-C-TAP, and MBP-myo1-Tail-TD2D structures by rotary-shadowing EM, 5–10.0 µl of each protein suspensions in 1x EM buffer was sprayed on freshly split mica (Tyler and Branton, 1980; Rock et al., 2005). The droplets were dried under vacuum, rotary-shadowed with platinum at an angle of 7°, and replicated with carbon in a freeze-fracture apparatus (model BFA 400; Balzers Spa). The replicas were photographed in an electron microscope (model 410; Philips) operating at 60 kV. The negatives were digitally scanned at 1,000 ppi, and the tail length of Myo1 and its derivatives was determined by measuring individual molecules from the head–tail junction to the C-terminal end using Photoshop software.

For observation of PS in different myo1 and iqg1 mutants by transmission EM, cells were fixed and processed as described previously (Nishihama et al., 2009).

Cells were grown in YPD medium to exponential phase (OD₆₀₀ = 0.5–1.0) and then treated with 1/100 volume of 20 mM latrunculin A (LatA) (Wako Chemicals USA, Inc.) or DMSO (solvent in which LatA was dissolved) for 30 min, then added 1/10 volume of 37% formaldehyde for fixation at 24°C for 30 min. Cells were then washed with PBS, permeabilized with PBS plus 0.2% Triton X-100, and stained for F-actin and DNA as described previously (Bi et al., 1998).

Fig. S1 shows coiled-coil profiles of myosin IIs from different organisms or cell types. Fig. S2 shows myo1-TD2-GFP in bni5Δ and LatA-treated cells. Fig. S3 shows disassembly defect of Myo1 tail during and after cytokinesis, and quantitative analysis of constriction behaviors of the full-length Myo1 in wild-type and bni5Δ cells as well as in cells carrying different myo1 alleles. Fig. S4 shows actin ring assembly in bni5Δ and different myo1 mutant cells, and the timing of Myo2 and Chs2 localization with respect to septin-hourglass splitting during the cell cycle. Video 1 shows localization of Bni5 versus Myo1 and myo1-mTD1 during the cell cycle. Video 2 shows the disappearance of full-length Myo1 from the bud neck at the onset of cytokinesis in mlc1-11 and iqg1Δ cells. Video 3 shows the constriction patterns of the rings formed by full-length Myo1 versus its tail. Video 4 shows the disappearance of the myo1-TD2Δ-GFP ring from the bud neck before the onset of cytokinesis. Video 5 shows the complete failure of myo1-TD2Δ-GFP localization to the bud neck in bni5Δ cells. Video 6 shows the localization of Myo1 in bni5Δ cells versus localization of myo1-TD2 during the cell cycle. Video 7 shows three-dimensional construction of the actin ring in a myo1-Tail cell. Video 8 shows the localization of Myo1 versus Myo2 and Chs2 during the cell cycle. Video 9 shows the localization of Myo2 and Chs2 versus Cdc3 during the cell cycle. Video 10 shows the constriction patterns of Myo1 ring in the presence of full-length IQG1 versus iqg1(Δ2-411).

We would like to thank Drs. J.R. Pringle, J. Hedgemann, J. Chant, J. Thorner, and M. Longtine for plasmids and strains; J.R. Pringle, M. Longtine, L. Thé, and X. Gao for sharing unpublished results; A. Stout and C. Fu for their help with imaging and analysis; J. Hanna for technical assistance; M. Lord, M. Ostap, and P. Tran for critically reading the manuscript; and the members of Bi laboratory for discussions.

This work was supported by the National Institutes of Health grants GM59216 (to E. Bi), GM31006 (to J.R. Pringle), P01-AR051174 (to C. Franzini-Armstrong) and by the American Cancer Society grant RSG-02-039-01-CSM (to E. Bi).