Fission yeast myosin-I (Myo1p) not only associates with calmodulin, but also employs a second light chain called Cam2p. cam2Δ cells exhibit defects in cell polarity and growth consistent with a loss of Myo1p function. Loss of Cam2p leads to a reduction in Myo1p levels at endocytic patches and a 50% drop in the rates of Myo1p-driven actin filament motility. Thus, Cam2p plays a significant role in Myo1p function. However, further studies indicated the existence of an additional Cam2p-binding partner. Cam2p was still present at cortical patches in myo1Δ cells (or in myo1-IQ2 mutants, which lack an intact Cam2p-binding motif), whereas a cam2 null (cam2Δ) suppressed cytokinesis defects of an essential light chain (ELC) mutant known to be impaired in binding to PI 4-kinase (Pik1p). Binding studies revealed that Cam2p and the ELC compete for Pik1p. Cortical localization of Cam2p in the myo1Δ background relied on its association with Pik1p, whereas overexpression studies indicated that Cam2p, in turn, contributes to Pik1p function. The fact that the Myo1p-associated defects of a cam2Δ mutant are more potent than those of a myo1-IQ2 mutant suggests that myosin light chains can contribute to actomyosin function both directly and indirectly (via phospholipid synthesis at sites of polarized growth).
Myosins are typically made up of three domains: a catalytic motor or head that binds actin, a neck made up of one or more IQ motifs and a tail that can specify the targeting, oligomeric state and function of these motor proteins. The neck works like a lever arm, maximizing actin filament displacement by amplifying conformational changes propagated by the ATPase activity of the motor. Calmodulin and related EF-hand proteins serve as light chains that bind IQ motifs and contribute to the rigidity of the α-helical neck. In some cases light chains regulate their cognate myosins. For example, phosphorylation of smooth muscle myosin regulatory light chain (RLC) switches on motor activity and promotes self-assembly of myosin into filaments (Trybus, 1991); Ca2+ binding by the essential light chain (ELC) activates scallop muscle myosin (Szent-Gyorgyi et al., 1999), whereas Ca2+ binding by calmodulin can inhibit unconventional myosins-I, -V and -VI (Kim and Flavell, 2008; Morris et al., 2003; Trybus, 2008).
The single-headed class I myosins are found throughout eukaryotes and feature in a variety of cellular processes, such as endocytosis, phagocytosis and cell motility. The sole class I myosin of fission yeast (Myo1p) is a long-tailed myosin-I (Lee et al., 2000; Sirotkin et al., 2005; Toya et al., 2001) that accumulates transiently at short-lived cortical endocytic actin patches (Kovar et al., 2011; Sirotkin et al., 2005; Takeda and Chang, 2005). Myo1p function is required for maintenance of cell polarity, sporulation and organization of sterol-rich domains at the plasma membrane (Lee et al., 2000; Takeda and Chang, 2005; Toya et al., 2001). Myosin-I motor activity promotes internalization of endocytic vesicles in both budding and fission yeast (Attanapola et al., 2009; Sun et al., 2006). Unlike long-tailed myosins in other species, yeast myosin-I tails possess a C-terminal acidic domain that contributes to activation of the Arp2/3 complex (Evangelista et al., 2000; Lechler et al., 2000; Lee et al., 2000). Arp2/3 promotes the polymerization of a branched network of actin filaments that helps drive the internalization of endocytic vesicles (Sun et al., 2006; Toret and Drubin, 2006). However, the role of myosin-I in Arp2/3 activation is not crucial given that truncated forms of myosin-I lacking the acidic domain support normal function in the cell (Galletta et al., 2008; Lee et al., 2000).
Myo1p has two IQ motifs that are presumably responsible for its ability to bind calmodulin (Cam1p) (Toya et al., 2001) and a second light chain, Cam2p (Lord and Pollard, 2004). Cam2p functions in mating and meiosis, and is essential for sporulation at elevated temperatures (Itadani et al., 2006), but its role in vegetative growth and in the function of Myo1p is not understood. Cam2p shares its closest amino acid sequence homology with calmodulin (41% identity) and the ELC Cdc4p (35%). Given that most of the class I myosins studied to date employ one or more calmodulins (Kim and Flavell, 2008), it is not clear why Myo1p utilizes Cam2p. Nevertheless, calmodulin-related light chains appear to be physiologically significant because they have been found to associate with certain other class I myosins: Acanthamoeba myosin-IC associates with alternative light chain MICLC (Wang et al., 1997); Dictyostelium MyoB and MyoD associate with MlcB and MlcD, respectively (Crawley et al., 2006; De La Roche et al., 2003); whereas vertebrate myosin-1c can interact with two other light chains (CIB1 and CaBP1) in addition to calmodulin (Tang et al., 2007).
Beyond myosin-I, a handful of other unconventional myosins utilize light chains other than (or in addition to) calmodulin. Class V myosins from budding yeast, fission yeast, Drosophila and chicken brains associate with the ELC (D'Souza et al., 2001; Espindola et al., 2000; Franke et al., 2006; Luo et al., 2004; Stevens and Davis, 1998) as do Drosophila myosin-VI and -VIIA associate with the ELC (Franke et al., 2006), whereas human myosin-X and Toxoplasma gondii myosin-XIV employ novel light chains (Herm-Gotz et al., 2002; Rogers and Strehler, 2001).
Although little is currently known about the role and importance of alternative light chains in unconventional myosin function, recent studies with the yeast ELCs have provided some insights. The ability of budding yeast myosin-V (Myo2p) to function in organelle and vesicle trafficking relies on its interaction with ELC Mlc1p (Altmann et al., 2008; Wagner et al., 2002). The essential role of the fission yeast ELC (Cdc4p) in contractile ring assembly and cytokinesis (Chang et al., 1996; McCollum et al., 1995) does not depend on myosin-II function because a myosin-II mutant lacking IQ motifs supports normal cytokinesis without suppressing the lethality of cdc4 mutants (D'Souza et al., 2001). Thus, cytokinesis defects associated with cdc4 alleles stem from a failure to function with other important ELC-binding partners. Cdc4p interacts with IQGAP (D'Souza et al., 2001), a conserved protein containing six IQ motifs and a Ras GTPase-activating protein (GAP) homology domain that is essential for contractile ring assembly and cytokinesis (Eng et al., 1998). Cdc4p also interacts with a type-IIIβ phosphatidylinositol (PI) 4-kinase (Pik1p) via a single IQ motif located at the C-terminus of the kinase (Desautels et al., 2001; Park et al., 2009). Pik1p is essential for septum formation and cell division (Park et al., 2009).
In this study we employed a combination of approaches to characterize the role of Cam2p in fission yeast (Schizosaccharomyces pombe). We demonstrate that Cam2p stabilizes the lever arm of Myo1p and promotes its function at endocytic patches. We also show that Cam2p indirectly contributes to Myo1p recruitment at endocytic patches via an alternative pathway involving an association with Pik1p at the cell cortex.
Cam2p promotes myosin-I localization at endocytic patches
We sought to determine whether Cam2p was important for Myo1p function in vivo. We first tested this by assessing the phenotype of a cam2-deleted (cam2Δ) strain. cam2Δ cells grew normally at most temperatures, but started to exhibit obvious morphological defects at 34°C and above, resulting in lethality at 36°C (Fig. 1A). cam2Δ cells lose their polarity, becoming round, swollen and enlarged before lysis (Fig. 1B). These defects in shape were reminiscent of the phenotype of myo1Δ cells (Lee et al., 2000; Toya et al., 2001), albeit somewhat milder (considering the lack of morphological defects observed in cam2Δ cells grown at 25–30°C). Deletion of cam2 did not exacerbate the slow growth or morphological phenotype of myo1Δ cells (Fig. 1C) implying that Cam2p's major role lies with Myo1p. Consistent with Cam2p acting with Myo1p, the two proteins colocalized throughout their lifetimes at cortical patches (Fig. 1D; see supplementary material Movie 1). Examination of this colocalization in myo1 lever arm point mutants (myo1-IQ1-AA and myo1-IQ2-AA) revealed that the second IQ motif of Myo1p (IQ2) is the Cam2p-binding site. Cam2p colocalized with Myo1p-IQ1-AA (but not Myo1p-IQ2-AA) in endocytic patches (Fig. 1D; see supplementary material Movies 2 and 3). Interestingly, Cam2p still localized at cortical patches in the myo1-IQ2-AA mutant, but these were distinct from Myo1p patches (Fig. 1D).
We next tested the importance of Cam2p in Myo1p localization. Although loss of cam2 did not prevent Myo1p from localizing at cortical patches, there was less of it (Fig. 2A,B). Time-lapse microscopy was employed to track the intensity and lifetime of Myo1p–GFP fluorescence at wild-type and cam2Δ patches (Fig. 2C,D). Myo1p had an average lifetime (11.3 seconds) that was similar to that previously reported (12.8 seconds) for the same fusion protein (Takeda and Chang, 2005). Both the level and lifetime of Myo1p at cam2Δ patches were on average ~1.5-fold lower than at wild-type patches (Fig. 2D). The average intensity measurements represent an over-estimation of Myo1p levels in cam2Δ cells because only a sub-population of Myo1p patches was bright enough to be effectively tracked in this mutant. By contrast, the shorter lifetimes could be an underestimation owing to the limits of Myo1p detection in the cam2Δ strain. In any case, altogether the data are consistent with a reduction in Myo1p levels at patches when Cam2p is absent. The reduced levels of Myo1p do not reflect a general defect in patch integrity because another actin patch marker (fimbrin; Fim1p) was equally abundant in wild-type and cam2Δ patches (supplementary material Fig. S1). Reduced levels of Myo1p at patches reflect a drop in cellular concentration because much less Myo1p was present in cam2Δ cell extracts compared with extracts from wild type (Fig. 2E,F). Collectively our findings indicate that Cam2p associates with Myo1p and promotes Myo1p localization and function at patches.
Cam2p maximizes the motility activity of myosin-I
Given the importance of Cam2p for Myo1p function in vivo, and the ability of Cam2p to interact with Myo1p (Fig. 1D), we wished to test whether Cam2p was important for Myo1p motor function. Endogenous Myo1p was purified from fission yeast as previously described (Clayton et al., 2010). We first tested the effects of Ca2+ on Myo1p motor activity, given that vertebrate myosin-I motility can be inhibited by relatively low concentrations (~1–100 μM) of Ca2+ (Collins et al., 1990; Lin et al., 2005; Perreault-Micale et al., 2000; Zhu et al., 1998; Zhu et al., 1996). However, Ca2+ had no obvious effect on Myo1p motility, with partial inhibition only being seen at high, non-physiological Ca2+ concentrations (Fig. 3A; Table 1). Myo1p isolated in the absence of Cam2p exhibited wild-type actin-activated ATPase activity and apparent actin-binding affinities (Fig. 3B; Table 1). However, the motility rate of Myo1p lacking Cam2p was more than 50% less than that of wild type (Fig. 3C; Table 1). We note that calmodulin cannot functionally substitute for Cam2p because endogenous Myo1p (one-step purified from cam2Δ cells using overexpressed GST-tagged calmodulin) still exhibits reduced rates of motility despite having access to excess calmodulin. However, addition of recombinant Cam2p to these samples restored wild-type motility rates (Table 1). Thus, although Myo1p ATPase activity does not depend on Cam2p, its ability to propagate maximal rates of actin filament gliding does.
Cam2p localizes to the cortex in the absence of myosin-I
The ability of Cam2p to localize at cortical patches independently of Myo1p (in the myo1-IQ2AA mutant; Fig. 1D) suggests that this light chain does not function exclusively with Myo1p. To further investigate this we compared Cam2p localization in wild-type and myo1Δ cells. Like Myo1p, Cam2p localized to cortical patches (Fig. 4A) with short ~10 second lifetimes in wild-type cells (Fig. 4B,C). However, although Cam2p still localized as patches in myo1Δ cells (Fig. 4A) their lifetimes were on average 1.7-fold longer than in wild-type cells (Fig. 4B,C). Cam2p patches in the myo1Δ cells were brighter overall (Fig. 4A,C) and more polarized toward cell tips and division sites (Fig. 4A,D). Time-lapse analysis of Cam2p and an actin patch marker (Fim1p; Fig. 4D) or an early endocytic marker (End4p/Sla2p; supplementary material Fig. S2) revealed that, unlike in wild-type cells, Cam2p patches do not associate with endocytic patches in the myo1Δ background (compare Cam2p and Fim1p colocalization in supplementary material Movies 4 and 5, or Cam2p and End4p colocalization in supplementary material Movies 6 and 7). The ability of Cam2p, like Myo1p (Sirotkin et al., 2005), to transiently colocalize adjacent to actin patches at the cell surface, before patch internalization, was lost in the absence of Myo1p (Fig. 4D). Thus, although in vivo and in vitro evidence shows that Cam2p functions with Myo1p, Cam2p appears to have an additional role at the cortex.
A cam2Δ mutation suppresses cytokinesis defects associated with an ELC mutant
The strong sequence similarity between Cam2p and the ELC (Cdc4p) had initially prompted us to test whether these light chains shared any overlapping functions in vivo. To test this we generated cam2Δ cdc4 double mutants using a number of temperature-sensitive cdc4 alleles (cdc4-8, cdc4-31, cdc4-A1 and cdc4-C2). Loss of cam2 did not increase the potency of cytokinesis defects associated with any of the cdc4 alleles, nor did any of the cdc4 alleles heighten the morphological phenotype associated with loss of cam2. However, we were surprised to find that loss of cam2 suppressed the cytokinesis defects of one particular mutant, cdc4-8 (Fig. 5A,B). Unlike with cdc4-A1 or myosin-II (myo2-E1) mutants (Fig. 5A,C), loss of cam2 allowed cdc4-8 cells to grow at 32°C with wild-type morphologies (Fig. 5A,B). The suppression was partial because cam2Δ cdc4-8 cells did not grow at a higher temperature (36°C; data not shown).
Influence of cam2Δ and ELC mutations on contractile ring function
The ability of a cam2Δ mutation to suppress a cdc4 mutant was intriguing because Cam2p and Cdc4p supposedly function at two distinct actin structures (Cam2p at endocytic patches; Cdc4p at contractile rings). The suppression did not reflect a negative role for Cam2p in contractile ring function because ring assembly and constriction dynamics were similar in wild-type and cam2Δ cells (Fig. 6A,B; supplementary material Table S1). Wild-type and cam2Δ rings assembled and constricted at approximately the same point in mitosis (Fig. 6C). If anything, cam2Δ rings constricted slightly more slowly (Fig. 6B,C; supplementary material Table S1), as has been observed previously with mutants of other actin patch proteins (Wu et al., 2006). Careful z-series and time-lapse analysis of contractile rings (Rlc1p–GFP) from cells expressing Cam2p–mCherry demonstrated that Cam2p localizes exclusively to patches, and is not incorporated into the ring (Fig. 6D).
Cytokinesis failure associated with the cdc4-8 mutation does not reflect a defect in fission yeast myosin-II (Myo2p) function (D'Souza et al., 2001; Lord and Pollard, 2004). Therefore, defects associated with cdc4-8 probably originate from defects in Cdc4p binding to alternative partners that play crucial roles in cytokinesis [i.e. IQGAP (Rng2p) and Pik1p].
Cam2p and the ELC associate with PI 4-kinase (Pik1p)
The ability of cam2Δ to suppress the cdc4-8 mutant might be explained by competition between Cam2p and Cdc4p. Defective interactions between mutant Cdc4-8p and Pik1p or Rng2p might not be as problematic for the cell when Cam2p is absent (and no longer competing for Pik1p or Rng2p). Cam2p is unlikely to compete with Cdc4p for Rng2p binding given that Cam2p and Rng2p function at distinct sites within the cell: in both wild-type and cdc4-8 strains, Cam2p is restricted to patches, whereas Rng2p is specific to rings (Fig. 6D; supplementary material Fig. S3). Unlike wild-type cdc4 and four other mutant alleles tested (cdc4-31, cdc4-A1, cdc4-A2 and cdc4-A11), the cdc4-8 mutant is known to impair the ability of Cdc4p to bind Pik1p in yeast two-hybrid assays (Desautels et al., 2001). Thus, we rationalized that Cam2p most probably competes with Cdc4p for Pik1p, given that loss of a Cdc4p–Pik1p interaction and suppression of mutant cdc4 by the cam2Δ are specific to the cdc4-8 allele.
We compared the ability of Cam2p, Cdc4p and mutant Cdc4-8p to interact with Pik1p. GST-tagged light chains were affinity purified from cell extracts using glutathione-Sepharose. Although wild-type Cdc4p and Cam2p effectively associated with Pik1p, Cdc4-8p (and the GST control) failed to precipitate any detectable levels of Pik1p in these assays (Fig. 7A). The inability of Cdc4-8p to associate with Pik1p was specific and not simply due to a general loss of function, considering that similar levels of wild-type and mutant Cdc4p were recovered in these pull-downs, which were equally effective in binding myosin-II (Myo2p) heavy chains (Fig. 7A). This latter observation is consistent with previous studies demonstrating that Cdc4-8p (1) has a stable fold at both permissive and restrictive temperatures (Slupsky et al., 2001), (2) supports wild-type rates of Myo2p in vitro motility (Lord and Pollard, 2004) and (3) shows inter-allelic complementation with other cdc4 alleles (Desautels et al., 2001). Cam2p and Cdc4p compete for binding to Pik1p because overexpression of Cdc4p reduced the levels of Pik1p co-precipitated with GST–Cam2p (Fig. 7B). In summary, Pik1p associates with Cdc4p and Cam2p, yet fails to support a detectable association with mutant Cdc4-8p.
Cam2p and Pik1p function together in the cell
We tracked Myo1p patch dynamics in wild-type and pik1 mutant cells to test the significance of the Cam2p–Pik1p interaction in Myo1p function. We employed a pik1-R838A mutant because this point mutation targets the IQ motif and abolishes Pik1p–Cdc4p (Park et al., 2009) and Pik1p–Cam2p (Fig. 7C) interactions without compromising the essential role of Pik1p in cell growth. Myo1p (and Cam2p) patch localization looked normal (data not shown), although Myo1p patches on average had slightly longer lifetimes in the mutant (pik1-R838A:12.6 seconds cf. 10.5 seconds for a wild-type strain examined in parallel; P=0.06 in a paired student t-test, n=40). Thus, the ability of Cam2p to interact with Pik1p is not by itself crucial for Myo1p localization at patches.
Use of a pik1-R838A myo1Δ double mutant revealed that the Myo1p-independent cortical localization of Cam2p relies on Pik1p binding. The bright Cam2p patches observed at polarized growth sites in the myo1Δ background (Fig. 4A) were lost and Cam2p signal became diffuse throughout the cell in the double mutant (Fig. 8A). Previous studies showed that overexpression of wild-type Pik1p (but not mutant Pik1-R838Ap) inhibits cell growth (Park et al., 2009), indicating that light chain binding influences Pik1p function. We also saw growth inhibition when Pik1p was overexpressed from the medium-strength nmt1-41 promoter (Fig. 8B). Although cells looked normal overall, ~10% were swollen (Fig. 8C). However, these overexpression defects depended on Cam2p because they were suppressed by a cam2Δ mutation (Fig. 8B,C). In conclusion, these findings demonstrate that Cam2p and Pik1p contribute to one another's function in the cell.
Cam2p promotes Myo1p function independent of its association with the myosin lever arm
To assess the physiological importance of alternative Cam2p binding partners we compared the phenotype of cam2Δ cells with those of the mutant (myo1-IQ2-AA) that specifically fails to support the Myo1p-Cam2p interaction (Fig. 1D). Although cam2Δ cells fail to grow at 36°C (Fig. 1A,B) the myo1-IQ2-AA mutant supported growth (Fig. 9A). The myo1-IQ2-AA mutant did exhibit temperature-sensitive defects in morphology, but they were much milder (Fig. 9B) than the lethal defects associated with the cam2Δ mutant (Fig. 1B). Attenuation of light chain binding at the other Myo1p IQ motif (IQ1), in the myo1-IQ1-AA mutant, supported growth at 36°C (Fig. 9A), with any defects in morphology being milder than those with the myo1-IQ2-AA mutant (Fig. 9B). We also note that normal levels of Myo1p-IQ2-AA localized to patches (Fig. 1D), in contrast to the decreased Myo1p patch signals seen in cam2Δ cells (Fig. 2A–D). Collectively, these in vivo data indicate that complete loss of Cam2p is more detrimental to Myo1p localization and function than specific loss of the Myo1p–Cam2p interaction. The ability of Cam2p to interact with Pik1p (or other alternative binding partners) at the cell cortex augments its conventional role at the Myo1p lever arm to secure optimal myosin function in the cell. The inability of Myo1p to effectively accumulate at the cortex in cam2Δ cells probably leads to greater turnover in the cytoplasm, leading to the lower cellular levels observed.
Although many type-I myosins employ calmodulin as their sole light chain, some utilize alternative light chains. To understand the importance of such light chains we investigated the role of fission yeast Cam2p. Cam2p ensures maximal myosin-I motility in vitro and promotes myosin-I function in vivo. In addition to supporting the mechanical role of the myosin-I lever arm during force production, Cam2p also functions with the PI 4-kinase Pik1p (and possibly other factors), to promote Myo1p function in the cell. Our work adds to recent studies highlighting the existence of additional ELC-binding partners besides myosin.
Cam2p – a calmodulin-like light chain required for myosin-I function
Loss of Cam2p resulted in reduced cellular levels of Myo1p and a phenotype reminiscent of a myo1Δ strain. Thus, Cam2p is important for securing Myo1p function. However, the milder phenotype of cam2Δ (versus myo1Δ) cells at standard temperatures indicates that Cam2p is not absolutely vital for Myo1p function. In the absence of Cam2p, Myo1p did not lose function or aggregate, but retained wild-type-like actin-activated ATPase activity in vitro and maintained dynamic localization at cortical patches in vivo (albeit at significantly lower levels). The cam2Δ phenotype was, however, stronger than a mutation (myo1-IQ2-AA) specifically preventing a Myo1p–Camp2 interaction, suggesting that Cam2p has other roles beyond being a light chain for Myo1p.
Although Myo1p ATPase activity remained intact in the absence of Cam2p, its motility rate was reduced by half. The myosin lever arm maximizes actin filament displacement and motility by amplifying conformational changes originating in the motor domain. Given Cam2p binds one of the two Myo1p IQ motifs (IQ2), the slower rate of motility in the absence of Cam2p is consistent with a role in stabilizing the lever arm. Thus, Cam2p maximizes Myo1p force production at actin patches. Nevertheless, Cam2p still localized to the cortex as dynamic patches in myo1-IQ2-AA and myo1Δ mutants, again suggestive of alternative partners for this light chain. This Myo1p-independent localization, coupled with the non-additive phenotype of a myo1Δ cam2Δ strain implies that any additional role for Cam2p augments Myo1p function. In this sense, Cam2p contrasts with calmodulin, which functions with numerous proteins involved in a wide range of cellular processes (Chant, 1994).
Cam2p and the ELC function with Pik1p
Loss of Cam2p selectively suppressed cytokinesis defects of an ELC mutant (cdc4-8) that fails to effectively associate with Pik1p. The suppression reflects some degree of functional overlap between Cam2p and the ELC in the cell. Like Cdc4p (Park et al., 2009), Cam2p was found to bind Pik1p. Thus, the suppression might be explained by the ability of Cam2p to out-compete Cdc4-8p for a limited supply of Pik1p. Pik1p and homologs from higher eukaryotes possess a single IQ motif at their C-termini (Park et al., 2009). However, Cam2p is unlikely to regulate the activity of Pik1p because mutations at the Pik1p IQ motif (which prevent light chain binding) have no effect on Pik1p kinase activity (Park et al., 2009). A previous study in budding yeast has also shown a link between light chains and lipid kinases. One particular calmodulin mutant (cmd1-226) shows defects in actin organization and endocytosis stemming from perturbation of PI 4-phosphate 5-kinase/Mss4p function, defects that were suppressed by simply overexpressing Mss4p (Desrivieres et al., 2002).
Type-IIIβ PI 4-kinases from a wide range of eukaryotes have been found to play important roles in endocytosis (Audhya et al., 2000; Balla et al., 2002; Pizarro-Cerda et al., 2007; Sorensen et al., 1998; Sorensen et al., 1999). PI 4-kinases convert phosphatidylinositol to phosphatidylinositol 4-phosphate, a precursor for phosphatidylinositol (4,5)-bisphosphate [PtdIns(4,5)P2]. PtdIns(4,5)P2 at the plasma membrane is required for invagination of clathrin-coated pits (Doherty and McMahon, 2009; Haucke, 2005), whereas PtdIns(4,5)P2 turnover is known to play a pivotal role in the assembly and disassembly of endocytic patches in yeast (Sun et al., 2007). This presumably explains why a number of endocytic proteins possess PtdIns(4,5)P2-binding domains (Ford et al., 2001; Friesen et al., 2006; Gaidarov and Keen, 1999; Itoh et al., 2001; Jost et al., 1998; Sun et al., 2005; Takei et al., 1999; Vallis et al., 1999) or associate with enzymes involved in PtdIns(4,5)P2 synthesis and breakdown (Bairstow et al., 2006; Haffner et al., 1997; McPherson et al., 1996; Padron et al., 2003; Stefan et al., 2005). Fission yeast Pik1p localizes uniformly throughout the cell and concentrates in the Golgi apparatus (Park et al., 2009) irrespective of whether Cam2p is deleted (supplementary material Fig. S4A). Pik1p-dependent cortical Cam2p patches (seen in the absence of Myo1p) are not enriched with Pik1p (supplementary material Fig. S4B). Thus, like PtdIns(4,5)P2 and other enzymes involved in PtdIns(4,5)P2 synthesis (Homma et al., 1998; Stefan et al., 2002; Zhang et al., 2000), Pik1p does not appear to concentrate at cortical patches. Whether undetectable levels of Pik1p concentrate at patches, or uniform localization is maintained at the cortex to support Pik1p in different membrane-based processes is not clear.
PtdIns(4,5)P2 provides crosstalk between actin and the plasma membrane by stimulating actin polymerization and establishing cytoskeleton–membrane linkages (Yin and Janmey, 2003). Given actin is crucial for endocytosis in yeast and certain types of endocytic events in other cells (Engqvist-Goldstein and Drubin, 2003), alternative light chains, such as Cam2p, might help establish a connection between the cytoskeletal machinery at endocytic patches and PtdIns(4,5)P2 synthesis at the cortex. Recent studies indicate that type-I myosins rely on PtdIns(4,5)P2 to target the plasma membrane. The tail of vertebrate myocin 1c contains a putative pleckstrin homology (PH) domain that binds PtdIns(4,5)P2 and is essential for localizing it to the cell surface (Hokanson et al., 2006; Hokanson and Ostap, 2006). Myo1p shares this conserved domain within the TH1 region of its tail, which mediates localization at patches (Lee et al., 2000) and is essential for the involvement of Myo1p in concentrating sterol-rich membrane domains at polarized growth sites (Takeda and Chang, 2005).
Type IIIβ PI 4-kinases play crucial roles in cytokinesis in higher eukaryotes (Brill et al., 2000; Rodgers et al., 2007), whereas PtdIns(4,5)P2 is required for cytokinesis in a variety of cells (Emoto et al., 2005; Field et al., 2005; Janetopoulos and Devreotes, 2006; Wong et al., 2005; Zhang et al., 1999). Thus, the ELC Cdc4p might help link Pik1p and PtdIns(4,5)P2 synthesis to PtdIns(4,5)P2-binding proteins at the contractile ring. The IQGAP Rng2p binds Cdc4p via its IQ motifs and contains a PH domain, and its ability to associate with PtdIns(4,5)P2 is presumably important for its localization and function at the interface between the contractile ring and the membrane.
Genetic and Myo1p localization experiments revealed that the ability of Cam2p to interact with alternative binding partners augments its established role at the myosin lever arm to promote Myo1p function in the cell. The tight colocalization of Myo1p and Cam2p in wild-type cells implies that any alternative role for Cam2p on the cortex might be accommodated while Cam2p is bound and tightly concentrated with Myo1p at endocytic patches. On the basis of recent structural studies of Cdc4p and Mlc1p (the budding yeast ELC), Cam2p might be capable of simultaneously associating with Pik1p and Myo1p in a ternary complex. X-ray crystallography and NMR-based analysis of Cdc4p and Mlc1p have shown that these proteins are monomeric dumbbell-shaped molecules made up of two structurally distinct domains connected by a flexible linker (Amata et al., 2008; Escobar-Cabrera et al., 2005; Pennestri et al., 2007; Slupsky et al., 2001; Terrak et al., 2003). Mlc1p can adopt different conformations depending on the IQ motif it binds to (Terrak et al., 2003). In a ‘compact’ conformation both domains of Mlc1p interact with the IQ motif, whereas in an ‘extended’ conformation only one domain binds the IQ motif (Pennestri et al., 2007; Terrak et al., 2005; Terrak et al., 2003) leaving the other domain free to potentially bind another IQ motif on a different protein (Terrak et al., 2003). In the future it will be worth testing whether such a mechanism allows Cam2p to function with both Myo1p and Pik1p at the cortex to directly coordinate actomyosin forces with phospholipid modification at sites of endocytosis.
Materials and Methods
Yeast strains, genetic methods and plasmids
Standard fission yeast genetic and cell biology protocols were employed (Moreno et al., 1991). The yeast strains used are listed in supplementary material Table S2. Gene deletion and disruption mutants, and strains harboring GFP-, CFP- or Cherry-tagged fusion proteins were constructed by using genomic integrations with the relevant kanR, natR or ura4+ cassettes (Bahler et al., 1998; Snaith et al., 2005). myo1-IQ1-AA (I730A, Q731A) and myo1-IQ2-AA (I748A Q749A) mutants were generated by integrating mutated myo1 PCR fragments into a myo1Δ::ura4+ disruption strain (lacking codons encoding amino acids 720–772 of Myo1p) using 5-fluoroorotic acid counter selection. DNA sequencing of PCR-amplified genomic DNA was used to confirm the mutagenesis. A myo1Δ strain complemented by a ura4+ myo1 plasmid (pUR19–myo1) was employed in crosses to generate different myo1Δ strains to avoid sporulation defects associated with loss of myo1. Relevant myo1Δ segregants that lost the plasmid during sporulation (i.e. cells that were ura4−) were selected. Plasmids employed in this study are listed in supplementary material Table S3. The pGFP–cdc4 plasmid was constructed by inserting the NotI cdc4 open reading frame (from pGST–cdc4) into pDS573a.
A Nikon TE2000-E2 inverted microscope with a motorized fluorescence filter turret and a Plan Apo 60× (1.45 NA) objective was employed to capture DIC and epifluorescence images. An EXFO X-CITE 120 illuminator was used to generate fluorescence. NIS Elements software (Nikon) was used to control the microscope, two Uniblitz shutters and a Photometrics CoolSNAP HQ2 14-bit camera. For colocalization analysis, time series in a single confocal section or z-series were captured using an UltraView VoX (Perkin Elmer) spinning disk confocal microscope. This system was equipped with C9100-50 EMCCD camera (Hamamatsu) on a Nikon Ti-E microscope with a 100×, 1.4 NA Plan Apo lens.
Time-lapse movies of cells were used to monitor patch lifetimes (by capturing images every 2–3 seconds for 1 minute) and contractile ring dynamics (every 2–3 minutes for 2 hours) using appropriate filters (at room temperature). Cell suspensions (3 μl) were mounted on flat 30 μl media pads (solidified with 1% agarose) prepared on the slide surface. VALAP (vasoline, lanolin and paraffin; 1:1:1) was used to seal slides and coverslips. In order to track patch lifetimes by epifluorescence, or to simultaneously track spindle pole bodies (SPBs) and rings, z-stacks of six images (spanning the cell depth) were collected at each time point. Analyses of patch, ring and SPB dynamics were performed using ImageJ, Microsoft Excel and Kaleidagraph software.
Myo1p was purified from fission yeast as previously described (Clayton et al., 2010). In order to purify Myo1p without Cam2p the first step of the purification was repeated using cam2Δ cells harboring a pGFP–cam1 construct (as opposed to using both pGFP–cam1 and pGFP–cam2 constructs). To reconstitute wild-type rates of Myo1p motility with Cam2p (Table 1) we purified GST–Cam2p (following overexpression in myo1Δ cells) and then added this back (1:1 in protein concentration) to the Myo1p plus GST–Cam1p sample lacking Cam2p. The activity of the Myo1p samples (with or without Cam2p added) were compared in motility assays (Table 1). Skeletal muscle actin was purified from chicken muscle as previously described (Spudich and Watt, 1971).
Wild-type and cam2Δ myo1–GFP cells containing pGST–cam1, or GFP–pik1/pik1-R838A cells containing either pGST–rng3-C, –cam2, –cdc4 or –cdc4-8 were grown to an OD595 of 1 in 50-200 ml of Edinburgh minimal medium (EMM) Leu− medium. For Cam2p–Cdc4p competition experiments GST–Cam2p and GFP–Cdc4p were both overexpressed using maximum strength nmt1 promoters from pGST–cam2 and pGFP–cdc4 plasmids in cells grown in EMM Leu− Ura− medium. Cells were harvested and lysed as described for purification of Myo1p. Lysates were normalized for total protein and then batch incubated with 50 μl glutathione-Sepharose beads for 4 hours at 4°C. Beads were harvested by centrifugation at 82 g for 2 minutes and washed five times with lysis buffer plus additives, and resuspended in 100 μl of SDS sample buffer. Affinity purified samples were boiled for 10 minutes following the addition of 100 μl 2× SDS-PAGE sample buffer and run on a 10% gel. Proteins were transferred from the gels to nitrocellulose and immunoblotted. Myo1p–GFP and GFP-Pik1p were detected with monoclonal anti-GFP antibodies; Act1p was detected with β-actin antibodies, and Myo2p was detected with anti-Myo2p tail antibodies. These primary antibodies were diluted 1:1000 in PBS containing 0.1% Tween 20. Secondary anti-mouse or anti-rabbit horseradish-peroxidase-conjugated antibodies were diluted 1:3000.
In vitro motility assays
5 μM stocks of actin were polymerized by addition of 50 mM KCl and 1 mM MgCl2, incubated for 30 minutes, and then labeled with 5 μM of Rhodamine-phalloidin for 30 minutes. After adhering Myo1p (at a concentration of 250–500 nM) to the surface of a nitrocellulose-coated coverslip for 10 minutes, the motility chamber was washed: (1) three times with motility buffer (25 mM imidazole, pH 7.4, 50 mM KCl, 1 mM EGTA, 4 mM MgCl2, 2 mM DTT) plus 0.5 mg/ml BSA; (2) three times with motility buffer alone; (3) twice with motility buffer containing 1 μM of vortexed (30 seconds) unlabeled actin filaments; (4) three times with motility buffer plus 1 mM ATP; (5) twice with motility buffer plus 25 nM Rhodamine–phalloidin-labeled actin filaments and oxygen scavengers (50 μg/ml catalase, 130 μg/ml glucose oxidase and 3 mg/ml glucose); (6) twice with motility buffer plus 20 mM DTT, 0.5% methyl-cellulose and scavengers; and (7) twice with motility buffer plus 20 mM dithiothreitol, 0.5% methyl cellulose, 1.5 mM ATP and scavengers. Filaments were observed at room temperature by epifluorescence microscopy and recorded at 2-second intervals. ImageJ software was used to calculate filament velocities from time-lapse series (n=50–80 filaments).
Actin-activated Myo1p ATPase assays were carried out at room temperature in 2 mM Tris-HCl, pH 7.2, 10 mM imidazole, 60–165 mM KCl, 0.1 mM CaCl2, 3 mM MgCl2, 2 mM ATP and 1 mM dithiothreitol with 50–100 nM Myo1p and 0–100 μM actin filaments. Malachite Green was for the quantification of inorganic phosphate release (Henkel et al., 1988). Controls omitting Myo1p were also used for each sample, and basal activity (detected in controls lacking actin) was subtracted to derive actin-activated ATPase rates. Curves were fit to Michaelis–Menten kinetics using Kaleidagraph software.
We are grateful to Sarah Steinbach and Sean Hemmingsen (University of Saskatchewan, Canada) for the cdc4 alleles, the pik1-R838A strain and pik1 plasmid; Mohan Balasubramanian (Temasek Life Sciences Laboratory, Singapore) for the myo2-E1 strain; Susan Forsburg (University of Southern California, Los Angeles, CA) for the wild-type strains (FY 435 and 436); Ken Sawin (University of Edinburgh, UK) for pFA–mCherry vectors; and Jian-Qiu Wu (The Ohio State University, Cleveland, OH) for the mYFP–rng2 strain and pFA6a vectors. Work in the Lord and Sirotkin laboratories is funded by a New Research Initiative Award from the University of Vermont (M.L.) and Scientist Development Grants 0835236N (M.L.) and 11SDG5470024 (V.S.) from the American Heart Association.