|Home | About | Journals | Submit | Contact Us | Français|
Rapid and polarized turnover of actin networks is essential for motility, endocytosis, cytokinesis, and other cellular processes. However, the mechanisms that provide tight spatiotemporal control of actin disassembly remain poorly understood. Here, we show that yeast coronin (Crn1) makes a unique contribution to this process by differentially interacting with and regulating cofilin effects on ATP/ADP+Pi versus ADP actin filaments. Crn1 potently blocks cofilin severing of newly assembled (ATP/ADP+Pi) filaments, but synergizes with cofilin to sever older (ADP) filaments. Thus, Crn1 has qualitatively distinct/opposite effects on actin dynamics depending on the nucleotide state of actin. This bimodal mechanism requires two separate actin-binding domains in Crn1. Consistent with these activities, Crn1 excludes GFP-Cof1 from newly assembled regions of actin networks in vivo and accelerates cellular actin turnover by four fold. We conclude that coronin polarizes the spatial distribution and activity of cofilin to promote selective disassembly of older actin filaments.
Maintaining filamentous actin networks in a state of rapid flux permits cells to remodel their actin cytoskeletons with speed and precision, and to replenish the assembly-competent pool of actin monomers available for new growth. Dynamic actin turnover is therefore essential for most actin-based cellular processes (Pollard and Borisy, 2003). ADF/cofilin (referred to herein as cofilin) plays a central role in this process by promoting severing and depolymerization of filaments (Bamburg, 1999). In S. cerevisiae, cofilin is encoded by an essential gene (COF1), and partial loss-of-function mutations with reduced binding to F-actin (e.g. cof1-22) cause slower cell growth and reduced rates of actin turnover (Lappalainen and Drubin, 1997; Lappalainen et al., 1997; Okada et al., 2006). Although cofilin is a central player in the regulation of actin disassembly, it does not act alone, but rather in concert with a number of other conserved actin binding proteins, including Aip1, Srv2/CAP, twinfilin, and coronin. Growing evidence suggests that each of these proteins makes a unique and important contribution to actin disassembly. However, the specific roles of each protein and their underlying mechanisms are not yet fully understood.
Coronins are a widely expressed family of actin-binding proteins that localize to sites of dynamic actin remodeling (de Hostos, 1999; Uetrecht and Bear, 2006). They have a three-part structure (see Fig. 1C), consisting of an N-terminal β-propeller domain that binds F-actin, a ‘unique’ middle region that varies in sequence and length, and a C-terminal coiled-coil (CC) domain that mediates homo-oligomerization and Arp2/3 complex binding. Coronin is important for phagocytosis, endocytosis, cell motility, embryonic development, and immune function (reviewed in Uetrecht and Bear, 2006; Foger et al., 2006). How coronin mechanistically contributes to these processes has only just begun to come into focus, and appears to involve effects on both actin assembly and disassembly. Purified coronin directly regulates the actin-nucleating Arp2/3 complex (Humphries et al., 2002), and this regulation is required for proper cell motility and immune function (Cai et al., 2005; Cai et al., 2007b; Foger et al., 2006). Coronin also influences actin disassembly. This was first suggested by genetic interactions between crn1Δ and mutants with reduced actin turnover (act1-159 and cof1-22) (Goode et al., 1999). More recently, Coronin-1A and Aip1 were isolated as factors that promote cofilin-dependent disassembly of Listeria actin tails (Brieher et al., 2006). Together, coronin, Aip1, and cofilin induce dramatic bursts of disassembly from actin filament ends by a mechanism yet to be resolved (Kueh et al., 2008). Moreover, knock down of Coronin-1B in fibroblasts decreases rates of actin turnover (Cai et al., 2007b), and Coronin-1B promotes debranching of filaments nucleated by Arp2/3 complex (Cai et al., 2008). Unexpectedly, purified Coronin-1B inhibits cofilin-mediated disassembly of F-actin in vitro (Cai et al., 2007a), which appears to be at odds with all of the observations mentioned above. This has left the role of coronin in actin turnover unclear. Here, we resolve this paradox, showing that coronin and cofilin cooperate conditionally to promote actin disassembly.
To better understand the in vivo role of coronin in actin disassembly, we first examined the effects of deleting CRN1 on yeast cell growth and actin turnover. Isogenic wild type (CRN1), crn1Δ, cof1-22, and crn1Δ cof1-22 strains were generated and compared for growth at different temperatures (16°C–37°C). Consistent with previous reports (Goode et al., 1999), crn1Δ grew similar to wild type cells at all temperatures, whereas cof1-22 cells was partially impaired for growth at 34°C (Fig. 1A) and failed to grow at 37°C (not shown). The crn1Δ cof1-22 cells showed a more severe growth defect at 34°C than cof1-22 cells (Fig. 1A). This synthetic defect was complemented by full-length coronin (FL) expressed from its endogenous promoter on a CEN plasmid (Fig. 1B). A construct lacking the coiled-coil domain (ΔCC) only partially rescued the synthetic defect. This suggests that both the CC and N-terminal regions of CRN1 contribute to its functions that are genetically shared with COF1.
Next, we compared rates of actin turnover in wild type and crn1Δ cells. Yeast cables and patches are highly dynamic F-actin structures that undergo continuous assembly and disassembly (Moseley and Goode, 2006). Treatment of cells with latrunculin A (LatA), which sequesters actin monomers and blocks new assembly, causes the disappearance of cables and patches at rates proportional to disassembly/turnover (Ayscough et al., 1997). Thus, kinetics of actin disappearance at a fixed concentration of LatA can provide a quantitative index for rate of actin turnover between strains (Lappalainen and Drubin, 1997; Okada et al., 2006). The time to reach 50% loss of cables was < 30 seconds in wild type cells, and about 4-fold longer in crn1Δ cells (120 seconds) (Fig. 1D). Rates of patch turnover were also reduced in crn1Δ cells (Fig. 1E and F). Consistent with these results, deletion of CRN1 conferred LatA resistance to cells in a halo assay, which measures the extent to which LatA spotted on a paper disc inhibits growth of the surrounding cells on the plate (not shown).
In addition, we compared the patch turnover rates of cof1-22 and crn1Δ cof1-22 strains, and found that deletion of CRN1 further reduced actin turnover by at least 4-fold in the cof1-22 background (Fig. 1G and H). Further, crn1Δ cof1-22 and cof1-22 strains had similar total F-actin fluorescence per cell before LatA treatment (Fig. 1I). This shows that the increased LatA-resistance of patches in crn1Δ cof1-22 cells was due to reduced actin turnover rather than a difference in starting size/intensity of patches.
To address the mechanism underlying our in vivo observations above, we purified Crn1 (FL) from E. coli, and tested its effects on Cof1-mediated disassembly of pyrene-labeled actin filaments. Because some cofilins quench the fluorescence of pyrene-labeled F-actin upon binding (Blanchoin and Pollard, 1999; Carlier et al., 1997), a loss of fluorescence can arise from cofilin binding rather than filament disassembly. Therefore, we first measured the quenching effects of a wide range of Cof1 concentrations on 2μM preformed pyrene-labeled F-actin. Cof1 caused concentration-dependent quenching, with maximal effect (> 90%) at equimolar cofilin:actin (Fig. 2A). Importantly, at 125 nM Cof1, which was used in all of the disassembly assays below, there was minimal quenching (9%). We further controlled for quenching by pre-mixing cofilin with the pyrene-labeled filaments for 5 min to allow fluorescence to equilibrate before initiating disassembly by addition of the actin monomer sequestering Vitamin D binding protein (DBP). Crn1 had no quenching effects on pyrene-actin. Thus, the observed decrease in fluorescence in these assays was due to disassembly and not quenching.
When 2μM F-actin was mixed with 125 nM Cof1 or 600 nM Crn1 (FL) (Fig. 2B), Cof1 accelerated disassembly (Fig. 2B). However, these Cof1 effects were inhibited by Crn1 (FL) in a concentration-dependent manner (Fig. 2D and S1). Samples of the reactions in Figure 2B were removed at 1700 sec and 15 hr and analyzed by high-speed pelleting (Fig. 2C). At 1700 sec, Cof1 shifted most of the actin to the supernatant, which could be due to full depolymerization and/or fragmentation of filaments into short oligomers. Crn1 (FL) blocked these effects. However, after prolonged incubation (15 hr, Fig. 2C), actin shifted to the supernatant in all reactions, indicating that while Crn1 (FL) can attenuate the effects of Cof1, it does not ultimately prevent actin disassembly.
Since the disassembly assay above does not discern between Cof1 effects in severing filaments versus accelerating subunit dissociation from filament ends, we next tested Cof1 effects on pyrene-actin assembly, which is sensitive to severing but not depolymerization. Gel filtered actin monomers spontaneously assemble into filaments inefficiently in the absence of nucleation factors. However, cofilin severs filaments to amplify free barbed ends, and thereby dramatically increases the rate of actin assembly. Consistent with the cooperative properties animal and plant cofilins (Bamburg, 1999), yeast Cof1 caused a concentration-dependent and cooperative increase in rate of actin assembly (Fig. S2A and B). Similar effects were observed for Cof1 in the presence of high concentrations of profilin (Fig. S2C and D), which blocks pointed end growth (Pring et al., 1992), demonstrating that barbed end growth was being measured. At a fixed concentration of Cof1 (250 nM), the further addition of Crn1 (FL) led to a concentration-dependent decrease in rate of actin assembly (Fig. 2E and F), demonstrating that Crn1 (FL) inhibits Cof1-mediated severing and amplification of barbed ends. We considered that the inhibitory effects of Crn1 (FL) might stem from the ability of Crn1 to bundle F-actin (Goode et al., 1999). However, addition of another bundling protein, Sac6/fimbrin, did not affect cofilin activity (Fig. S3), suggesting that filament cross-linking is not the underlying cause for Crn1 inhibition of Cof1.
Results above from bulk assembly assays were confirmed independently using total internal reflection fluorescence (TIRF) microscopy, visualizing Cof1 and Crn1 (FL) effects on individual actin filaments (Fig. 2G and H). The assembly of alexa-488-labeled ATP-actin monomers alone or with Cof1 and/or Crn1 (FL) was monitored in real time (Fig. 2G) and effects on average filament length were quantified (Fig. 2H). Cof1 produced short filaments (Movie S2), while further addition of Crn1 (FL) antagonized this effect (Movie S4), producing longer filaments similar to actin alone and Crn1 (FL) reactions (Movies S1 and S3). Thus, our bulk kinetic and single filament data are in close agreement, showing that Crn1 (FL) antagonizes Cof1-mediated severing.
To better understand the basis for Crn1 (FL) inhibition of Cof1, we dissected the activity by purifying Crn1 (CC; residues 601–651) and Crn1 (ΔCC; residues 1–600). Crn1 (CC) showed effects similar to Crn1 (FL), inhibiting Cof1-mediated actin assembly in a concentration-dependent manner (Fig. 3A and B), albeit higher concentrations of Crn1 (CC) were required for these effects compared to Crn1 (FL). These results were confirmed independently by TIRF microscopy (Fig. 3C and 3D; Movies S5-S8).
The observations above suggested that the CC domain alone may be capable of binding to F-actin and thereby blocking cofilin effects. Consistent with this possibility, Crn1 (CC) exhibited concentration-dependent binding to F-actin in co-sedimentation assays (Fig. 3E). Thus, the CC domain harbors a second actin-binding site, distinct from the actin-binding site in the β-propeller domain. The binding affinity of Crn1 (CC) for F-actin was weaker (micromolar range, not shown) than that of Crn1 (FL) (nanomolar range; Goode et al., 1999). However, this was expected given that Crn1 (FL) has two actin binding sites (β-propeller and CC) contributing to F-actin affinity. The weaker affinity of the CC domain alone explains the observed requirement for higher concentrations of Crn1 (CC) compared to Crn1 (FL) to inhibit Cof1. Further, Crn1 (CC) and Cof1 appear to compete for binding to F-actin (see ATP-actin curves in Fig. 6H and I), offering a mechanistic explanation for the inhibitory effects of Crn1. These CC domain actin interactions may be conserved in other species, as larger CC domain-containing fragments of mammalian Coronin-1A and Coronin-3 have been reported to bind F-actin (Liu et al., 2006; Spoerl et al., 2002).
We next tested the activities of Crn1 (ΔCC). This polypeptide alone had no obvious effects in DBP-induced disassembly assays (Fig. 4A), as previously reported (Goode et al., 1999). Surprisingly, combining Crn1 (ΔCC) and Cof1 led to a faster rate of F-actin disassembly compared to Cof1 alone, suggesting their synergy in promoting actin disassembly. As in Figure 2C, disassembly into monomers and/or short oligomers was confirmed by high-speed pelleting (Fig. 4B). Synergy between Crn1 (ΔCC) and Cof1 was also observed in dilution-induced disassembly assays (Fig. S4A).
To address synergy in filament severing, Crn1 (ΔCC) and Cof1 effects were tested in pyrene-actin assembly assays. Crn1 (ΔCC) alone caused only a minor increase in rate of actin assembly (Fig. 4C and D), consistent with previous observations (Goode et al., 1999), but when combined with Cof1 caused a robust concentration-dependent increase in rate of actin assembly (Fig. 4C and D). These effects were not altered by the presence of profilin (Fig. S2C and D), demonstrating that Crn1 (ΔCC) and Cof1 synergize to amplify free barbed ends.
As an independent test for severing, we measured the production of free barbed ends in a two-step seeded assembly assay (Fig. 4E). Monomeric actin (2 μM) was first assembled in the presence or absence of Cof1 and Crn1 (ΔCC) (inset, Fig. 4E). Then, samples were removed at 25% polymer mass assembly (dotted line, inset) and added to a second reaction containing 0.5 μM actin monomers. Filaments generated in the first reaction acted as ‘seeds’ for polymerization in the second reaction, and rate of assembly was directly proportional to the concentration of free barbed ends produced in the first reaction. The rate of actin assembly was minimal upon addition of actin alone or actin + Crn1 (ΔCC) seeds generated in the first reactions, but increased dramatically when the first reactions contained Cof1 or Cof1 + Crn1 (ΔCC). Since seeding the second reactions introduced low concentrations of Cof1 and/or Crn1 (ΔCC), we included control reactions containing the same final concentrations of Cof1 and/or Crn1 (ΔCC) (without seeds). No actin assembly was detected in these reactions (control curves 2, 3, and 4) demonstrating that the increase in rate of assembly caused by Cof1 and Crn1 (ΔCC) was due to amplification of barbed ends in the first reaction. The effects, quantified by measuring slopes of the seeded curves in Figure 4E and subtracting slopes of the corresponding control curves, indicated that Crn1 (ΔCC) enhances Cof1 severing by about 4-fold (Fig. 4F).
Severing was also assessed by removing samples early in the first reactions (dotted line, Fig. 4E inset) and visually comparing length of filament seeds (n > 100 filaments for each reaction, Fig. 4G). Crn1 (ΔCC) reduced the average filament length by 1.7-fold (5.0 μm, compared to 8.8 μm in actin alone reactions). Cof1 reduced average length to a greater extent (3-fold), yielding filaments with an average length of 2.86 μm. Combining Cof1 and Crn1 (ΔCC) produced very short filaments (0.94 μm), consistent with synergy in severing. These differences were significant (p < 0.001). Further, since very short filaments are difficult to detect by this method, these values may not reflect the full extent of synergy between Cof1 and Crn1 (ΔCC).
As another non-pyrene-based test for severing and disassembly, we incubated unlabelled F-actin with Crn1 (ΔCC) and/or Cof1 for 5 min and stained filaments with Alexa-488-phalloidin (Fig. S4B). Average filament lengths were measured (n > 100 filaments for each reaction) and found to be similar in reactions containing F-actin (13.5 μm) and F-actin + Crn1 (ΔCC) (12.4 μm), but significantly shorter in reactions containing F-actin + Cof1 (4.3 μm) and even shorter in reactions containing F-actin + Crn1 (ΔCC) + Cof1 (2.5 μm) (p < 0.001).
Collectively, the multiple lines of biochemical evidence shown above demonstrate that Crn1 (ΔCC) synergizes with Cof1 to sever actin filaments. Consistent with these activities, expression of an integrated copy of crn1Δ CC partially complemented the synthetic growth defects at 34°C observed for the crn1Δ cof1-22 strain (Fig. 4H).
To address the basis of Crn1 (ΔCC)-Cof1 synergy, we tested whether these two proteins associate. However, no interactions between Crn1 and Cof1 were detected by gel filtration, sedimentation velocity analysis, or two-hybrid analysis (not shown). This raised the possibility that Crn1 (ΔCC) may instead synergize with cofilin by an indirect mechanism involving its interactions with F-actin. To test this model, we characterized the effects of two separate Crn1 (ΔCC) polypeptides (Crn1-2 and Crn1-6) carrying point mutations in their β-propeller domains that impair F-actin binding (Fig. 5A and B). These alleles were identified in a study mapping coronin-actin binding interactions (M.G. and B.G., unpublished data). Unlike wild type Crn1 (ΔCC), Crn1-2 and Crn1-6 polypeptides failed to synergize with Cof1 in accelerating actin assembly (Fig. 5C and D) or disassembly (not shown). These results show that the synergy between Crn1 and Cof1 in severing requires Crn1 β-propeller domain binding to F-actin.
We next tested whether Crn1 (ΔCC) can recruit Cof1 to F-actin in co-sedimentation assays (Fig. 5E and F), and found that Crn1 (ΔCC) modestly increases Cof1 association with F-actin (by ~1.5-fold; p < 0.01). To control for minor differences in the level of actin found in the pellets between reactions, we independently normalized % Cof1 bound to the level of actin in the pellet. However, this yielded a similar 1.5-fold difference (not shown). These data are remarkably consistent with those showing that mammalian Coronin-1A increases cofilin association with F-actin by 1.5-fold (Brieher et al., 2006). However, this 1.5-fold increase in cofilin binding does not easily explain the ability of coronin to increase cofilin severing activity in vitro by 4-fold, or increase the rate of actin turnover in vivo by 4-fold.
We next addressed the apparent inconsistency between our in vivo data pointing to Crn1 and Cof1 synergy in promoting actin turnover and our biochemical data indicating that full-length Crn1 inhibits cofilin. A similar discrepancy was reported for mammalian Coronin-1B, in which genetic data suggest cooperation between coronin and cofilin in driving actin turnover in fibroblasts, yet purified full-length Coronin-1B inhibited cofilin severing in vitro (Cai et al., 2007a; Cai et al., 2007b). This paradox, spanning yeast and mammalian systems, combined with our observations above showing that coronin has two separate actin binding domains with markedly different effects on cofilin, led us to consider whether the effects of coronin on cofilin activity might be conditional. Actin networks are rich in ATP/ADP+Pi-actin near their newly assembled (leading) ends and become increasingly rich in ADP-actin at their older (trailing) ends. Therefore, we investigated whether the nucleotide state of F-actin might influence coronin effects on cofilin.
First, we compared Crn1 (FL) effects on Cof1-mediated assembly of ATP-actin versus ADP-actin. ADP-actin was used at a higher concentration than ATP-actin due to its slower rate of assembly, but a constant molar ratio of actin:Cof1:Crn1 (FL) was used. Remarkably, while Crn1 (FL) almost completely blocked Cof1-stimulated ATP-actin assembly (Fig. 6A), it synergized with Cof1 in stimulating ADP-actin assembly (Fig. 6B). These synergistic effects were concentration-dependent (Fig. 6C), and quantification of the data in Figure 6A and B revealed that Crn1 (FL) and Cof1 each caused a 3-fold increase in rate of ADP-actin assembly, while their combined effect was a 9-fold increase in rate of assembly (Fig. 6D). Interestingly, Crn1 (FL) alone accelerated the assembly of ADP-actin (Fig. 6B), suggesting that it may sever ADP-actin filaments in the absence of Cof1. Indeed, Crn1 (FL) and Cof1 each significantly (p < 0.001) reduced the average length of preformed filaments in visual assays (n > 100 filaments for each reaction; Fig. S5A) from 8.4 μm (F-actin alone reactions) to 4.0 and 3.9 μm, respectively, and combining Crn1 (FL) and Cof1 further reduced the average length to 2.0 μm. Together these data suggest that Crn1 (FL) and Cof1 each can independently sever ADP-actin filaments, and that their combined effects in severing are synergistic.
Similar effects were observed independently by TIRF microscopy, comparing Cof1 and Crn1 (FL) activities on ADP-actin (Fig. 6E and F, Movies S9–S12). Cof1 produced numerous and shorter ADP-actin filaments compared to actin alone, and further addition of Crn1 (FL) did not antagonize Cof1. Similar effects were observed for Crn1 (CC) (Fig. 6G and S5B; Movies S13–S16).
Together, our data from bulk and TIRF assays suggest that coronin has highly distinct effects on ADP-actin versus ATP/ADP+Pi-actin, selectively blocking cofilin severing of ATP/ADP+Pi actin yet strongly enhancing the severing of ADP-actin. In turn, this suggests that the interactions of Crn1 (FL) with ATP/ADP+Pi actin filaments must be qualitatively distinct from its interactions with ADP-actin filaments. Consistent with this view, Crn1 (FL) bound with higher affinity ATP/ADP+Pi-F-actin compared to ADP-F-actin (Fig. S6), as reported for mammalian Coronin-1B (Cai et al., 2007a). Strong binding of Crn1 (FL) was observed with 8 nM ATP/ADP+Pi-F-actin, consistent with previous reports (Goode et al., 1999), but little or no binding was detected with even 8-fold higher concentrations of ADP-F-actin. Further supporting this view, Crn1 (FL) and Crn1 (CC) competed much more effectively with cofilin for binding to ATP/ADP+Pi-F-actin compared to ADP-F-actin (Fig. 6H and I).
Cofilin is largely excluded from newly assembled regions of actin networks, such as those found at the leading edge of mammalian cells (Pollard and Borisy, 2003) and the cortical edge of yeast cells (Okreglak and Drubin, 2007). Our biochemical data above suggested that Crn1 might influence cofilin distribution in vivo, helping to exclude cofilin from regions of active actin assembly rich in ATP/ADP+Pi-actin.
To test this model, we asked whether deletion of CRN1 affects the distribution of Cof1 on actin patches. Using multi-wavelength live cell imaging in the sla2Δ background (which has elongated patches), we examined GFP-Cof1 distribution on Abp1-RFP marked actin patches and plotted their fluorescence intensities versus distance from the cell surface (n = 10 cells; averages plotted in Fig. 7B) as described (Okreglak and Drubin, 2007). As previously observed, in CRN1+ cells the GFP-Cof1 signal was notably diminished from the cell edge (Fig. 7A), and the peak of GFP-Cof1 signal was shifted inward compared to the peak of Abp1-RFP signal (Fig. 7B, top panel). However, in crn1Δ cells, the GFP-Cof1 signal was more uniformly distributed along the length of patches, with GFP-Cof1 and Abp1-RFP signals having almost identical peaks (Fig. 7A and B, bottom panel). These data strongly support our model that coronin regulates the distribution and activity of cofilin on actin networks. We also observed a decrease in GFP-Cof1 signal at the trailing ends of patches in crn1Δ cells compared to CRN1+ cells, which may reflect a role for Crn1 in recruiting Cof1 to the non-actively growing regions of actin networks.
In this work, we describe a new mechanism by which coronin promotes the polarized disassembly of actin filaments. We show that coronin differentially regulates cofilin activity depending on the nucleotide state of actin, and controls the spatial distribution of cofilin on actin networks in vivo to accelerate actin turnover.
Our data show that coronin regulation of actin disassembly depends on the nucleotide state of actin (see model in Fig. 7C). This mechanism requires two separate actin-binding domains in coronin, which have highly distinct effects. Binding of the coronin CC domain to F-actin competitively blocks cofilin binding and severing of ATP/ADP+Pi-actin, but not ADP-actin. In contrast, binding of the β-propeller domain of coronin to F-actin enhances cofilin severing. The observation that full-length coronin, which contains both actin-binding domains, inhibits cofilin severing of ATP/ADP+Pi-actin filaments suggests that the inhibitory effects of the CC domain are dominant over the synergistic effects of the β-propeller domain. This view is supported further by our observation that Cof1-mediated actin assembly is inhibited in the combined presence of Crn1 (FL) and Crn1 (ΔCC) (Fig. S7). Our data also suggest that inhibition does not stem from actin bundling (Fig. S3), but instead is due to competitive binding between cofilin and Crn1-CC domain for ATP/ADP+Pi-F-actin (Fig. 6H and I). Interestingly, the CC domain binds to ADP-F-actin (not shown) but does not competitively displace cofilin from ADP-F-actin (Fig. 6I). This suggests that the CC domain has a qualitatively distinct interaction with ADP- versus ATP/ADP+Pi-F-actin. When bound to ATP/ADP+Pi-F-actin, the CC domain may alter the actin filament conformation to inhibit cofilin binding. Consistent with this possibility, a recent study showed that Coronin-1A decoration induces a ‘stabilizing’ conformation in F-actin (Galkin et al., 2008).
While many actin-binding proteins have been shown to have higher affinity for either ATP/ADP+Pi-actin or ADP-actin, coronin shows qualitatively distinct/opposite functional effects when bound to actin in different nucleotide states. This unique property allows coronin to selectively protect newly assembled regions of actin filaments and accelerate disassembly of older regions of actin filaments. The predicted sum of these effects is enhanced polarization of actin network disassembly, in the direction of the older filaments. Cofilin has an inherent binding preference for ADP-actin filaments (Kd=0.6μM) over ADP+Pi actin filaments (Kd=20μM) (Blanchoin and Pollard, 1999), which should skew its distribution toward ADP-actin filaments. However, even sparse decoration of ATP/ADP+Pi-actin filaments by cofilin may be sufficient for severing. This concept is supported by TIRF studies showing that low concentrations of cofilin are optimal for severing (Andrianantoandro and Pollard, 2006), and by our observation that low molar ratios of Cof1 to F-actin efficiently sever newly assembled (ATP/ADP+Pi) filaments in both bulk and TIRF assays. Collectively, these observations raise a previously overlooked point, which is that despite having a weaker binding affinity for ATP/ADP+Pi-actin compared to ADP-actin, cofilin nonetheless severs newly formed filaments unless they are actively protected. We show that coronin potently blocks severing of ATP/ADP+Pi-actin by cofilin, and thus selectively shields newly assembled regions of actin filaments.
Consistent with the model above, deletion of CRN1 reduced rates of actin turnover in vivo (Fig. 1) and reversed the normal exclusion of GFP-Cof1 from regions of active actin assembly in cortical actin patches (Fig. 7A and B). These functions of coronin may be conserved. Genetic disruption of Coronin-1B in mammalian cells reduces the rate of leading edge retrograde actin flow (Cai et al., 2007b). This effect was attributed to coronin’s recruitment of Slingshot phosphatase, which dephosphorylates cofilin to activate severing. However, part of these observed effects might have been due to interactions between Coronin-1B and F-actin. Slingshot phosphatase is not conserved in all organisms where coronin promotes actin turnover (Huang et al., 2006), and in some organisms (e.g. S. cerevisiae), cofilin is not phospho-regulated (Lappalainen et al., 1997). Given that coronin-actin interactions are conserved across distant species, the direct effects of coronin on F-actin may provide a more general mechanism for regulating cofilin spatial distribution and function, with recruitment of Slingshot phosphatase providing a complementary mechanism that is necessary in systems where cofilin is phosphorylated.
Our data may also explain the apparently conflicting results from Brieher et al. showing that Coronin-1A enhances cofilin-mediated disassembly of Listeria actin tails (Brieher et al., 2006), and Cai et al. showing that purified Coronin-1A and Coronin-1B inhibit cofilin effects on actin (Cai et al., 2007a). This discrepancy could be explained by differences in the nucleotide states of actin. Cai and co-workers used filaments newly assembled from ATP-actin, whereas Brieher and co-workers used Listeria tails likely to have been rich in ADP-actin near their older ends, consistent with observed synergy between coronin and cofilin in disassembling the rear of tails. The latter study also found that Coronin-1A increases cofilin association more substantially (4–5 fold) on Listeria tails than on purified filaments assembled from ATP-actin (1.5 fold) (Brieher et al., 2006), which again would be explained by distinct interactions of coronin with actin in different nucleotide states.
In principle, the ability of Crn1 to enhance Cof1-mediated actin assembly (Fig. 4 and and6)6) could be due to severing or nucleation. One recent study suggested that high concentrations of cofilin nucleate actin assembly (Andrianantoandro and Pollard, 2006). However, multiple lines of evidence argue that the mechanism underlying our observations is severing rather than nucleation. First, using preformed filaments, Crn1 enhanced Cof1-mediated severing and disassembly, effects that cannot be explained by nucleation. Second, Cof1-dependent acceleration of actin assembly occurred at low ratios of cofilin to actin (1:8), whereas the nucleation reported above required a vast excess of cofilin to actin (30:1). Third, GFP-Cof1 is excluded from regions of active actin assembly in vivo, and arrives at actin patches after their assembly rather than during their assembly (Okreglak and Drubin, 2007).
Coronin’s effects on actin disassembly may be regulated by additional factors besides the nucleotide state of actin. For instance, local inactivation of the CC domain of coronin by ligand-binding or phosphorylation might drastically alter coronin’s effects on ATP/ADP+Pi-actin, switching it from protecting filaments to promoting severing. This would allow the β-propeller domain of coronin to synergize with cofilin in severing ATP/ADP+Pi-actin filaments, as we have observed it can (Fig. 4). This possibility is intriguing in light of the interaction of the coronin CC domain with Arp2/3 complex (Cai et al., 2007b; Foger et al., 2006; Humphries et al., 2002). Perhaps Arp2/3 complex inactivation of the CC domain could trigger local coronin-cofilin synergy, leading to targeted severing at branch points (Fig. 7C), which may help explain the recent observation that Coronin-1B promotes filament debranching (Cai et al., 2008). In addition, phosphorylation of coronin governs its association with Arp2/3 complex (Cai et al., 2005; Cai et al., 2007b), and may also regulate its role in actin disassembly.
Finally, our observations may be mechanistically relevant to the recently described ‘bursting’ mode of actin disassmbly, in which a three-component mixture of cofilin, Aip1, and coronin was shown to induce abrupt and rapid loss of discrete lengths of polymer from the ends of newly assembled ATP-actin filaments (Kueh et al., 2008). When either Aip1 or Coronin was depleted from the mixture, bursting no longer occurred. Our observation that the coronin β-propeller domain (lacking the CC domain) enhances cofilin-mediated severing of filaments newly polymerized from ATP-actin (Fig. 4) may underlie part of the bursting mechanism. Perhaps binding of Aip1 to ATP/ADP+Pi-F-actin relieves full-length coronin inhibition of cofilin, triggering rapid severing. This would provide cells with a mechanism for disassembling newly polymerized filaments.
Dynamic organization of actin filaments into polarized, higher order structures requires the precise effects of multiple actin binding proteins working in concert. How the activities of so many factors are orchestrated with spatiotemporal precision to balance polarized assembly and turnover of actin suprastructures has remained elusive. One fundamental component of this mechanism is the changing nucleotide state of actin subunits as a function of polymer age. Many actin-associated proteins preferentially bind ATP/ADP+Pi-actin or ADP-actin, which provides a mode of spatial regulation. Here, we have revealed a new dimension to nucleotide state-based regulation, demonstrating that one protein (coronin) can switch its functional effects on actin dynamics depending on the nucleotide state. Similar ‘role-switching’ strategies may be widely employed by other actin-binding proteins to influence different aspects of actin dynamics.
All yeast strains and plasmids used are described in Supplementary Materials. All p-values reported were determined using a two-tailed t-test in Excel (Microsoft).
All images were acquired on a Zeiss Axioskop-2 mot plus microscope (Carl Zeiss, Thornwood, NY) using a Hamamatsu IEEE1394 digital CCD camera (Hamamatsu Photonics, Bridgewater, NJ) running OpenLab software (Improvision, Lexington, MA). For in vivo actin turnover assays (Fig. 1), cells were grown to log phase in 5 ml YPD cultures, pelleted, and resuspended in 1 ml YPD. 100–200 μl cells were treated with LatA (gift from Phil Crews, University of California, Santa Cruz, CA) at 30°C. At time points, 25 μl samples of cells were removed, fixed, and stained with Alexa-fluor-488-phalloidin (Molecular Probes, Eugene, OR). For live cell imaging (Fig. 7A and B), sla2Δ CRN1+ and sla2Δ crn1Δ strains with an integrated copy of Abp1-RFP and carrying a low copy plasmid expressing GFP-Cof1 (see supplementary materials for plasmid construction) were grown in parallel at 25°C to early log phase in SD media lacking tryptophan. Abp1-RFP and GFP-Cof1 signals intensities were quantified using ImageJ (NIH) software.
Crn1 (FL), and Crn1 (ΔCC) and Crn1 (CC) polypeptides were expressed as GST-usions in E. coli strain BL21 (DE3) and purified (Goode et al., 1999). GST tags were removed from all Crn1 (ΔCC) polypeptides by thrombin digestion. Crn1 (FL) was isolated with and without the GST tag. The activities of Crn1 (FL) and Crn1 (ΔCC) polypeptides with and without GST tags were indistinguishable in Cof1-mediated actin assembly assays (not shown). Crn1 (CC) was isolated with its GST tag present. Recombinant S. cerevisiae Cof1 (Lappalainen and Drubin, 1997) and S. cerevisiae profilin (Eads et al., 1998) were purified from E. coli. ATP-bound rabbit skeletal muscle actin (RMA) was purified (Spudich and Watt, 1971) and gel filtered. ATP-RMA was labeled on Cys374 with pyrenyliodoacetamide (Pollard and Cooper, 1984) or on lysines with Alexa-488 fluorophor (Isambert et al., 1995). To prepare pyrene-labeled and unlabeled ADP-RMA, ATP-pyrene-RMA and ATP-RMA were incubated for 5 min on ice with AG1-X2 beads (Bio-Rad, Hercules, CA) to remove free nucleotide, then treated with hexokinase (Sigma-Aldrich, St. Louis, MO) (Pollard, 1986) to convert the bound nucleotide to ADP.
Gel-filtered RMA was used in all assays. Pyrene signal was monitored at 25 °C in a fluorescence spectrophotometer (Photon Technology International, Lawrenceville, NJ) at excitation 365 nm and emission 407 nm. To analyze Cof1 quenching effects, actin (10% pyrene labeled) was polymerized in F-buffer (10 mM Tris-HCl pH 7.5, 0.2 mM DTT, 0.2 mM CaCl2, 50 mM KCl, 2 mM MgCl2, and 0.7 mM ATP). For DBP-induced disassembly assays, 40 μl preformed F-actin (2 μM final, 10% pyrene labeled) was incubated for 5 min with 20 μl protein or control buffer, then (at time 0) disassembly was induced with 4μM DBP (human plasma Gc-globulin, Sigma-Aldrich, St. Louis, MO). Curves were normalized at time 0. For assembly assays, monomeric actin (5% pyrene labeled) was converted to Mg-ATP-actin immediately before use in each reaction and mixed with 15 μl proteins or control buffer and 3 μl of 20X initiation mix (40 mM MgCl2, 10 mM ATP, 1 M KCl) in 60 μl reactions. Effects of Crn1 and/or Cof1 on actin assembly were similar using RMA and yeast actin (not shown). Rates of assembly were calculated from slopes of curves at 25–50% polymerization. For seeded assays, 6 μl of the first reaction (assembly conditions as above) was removed at 25% polymerization and used to seed a second reaction (60 μl) containing 0.5 μM monomeric actin (10% pyrene labeled).
To visualize F-actin products from assembly and disassembly assays, samples were removed and diluted 50-fold into F-buffer plus 0.3 μM Alexa-488-phalloidin. After 5 min, filaments were diluted another 50 fold and 0.5 μl was applied to a nitrocellulose-coated coverslip. Filaments were examined by fluorescence microscopy.
Different concentrations of F-actin were incubated for 10 min with proteins or control buffer, then centrifuged for 20 min at 80,000 rpm in a TLA100 rotor (Beckman Instruments, Fullerton, CA). Pellets and supernatants were analyzed on Coomassie-stained gels or western blotted with affinity purified chicken anti-Cof1 and anti-actin antibodies (Okada et al., 2006) or mouse anti-Crn1 antibodies (Goode et al., 1999). Band intensities on gels and blots were quantified using ImageJ software.
Protein mixtures were diluted in freshly prepared fluorescence buffer containing 10 mM imidazole-HCl (pH 7.8), 50 mM KCl, 1 mM MgCl2, 100 mM DTT, 3 mg/ml glucose, 20 mg/ml catalase, 100 mg/ml glucose oxidase, and 0.5% methylcellulose to induce polymerization, and imaged at 20 sec intervals on an objective-based TIRF microscope (Nikon TE2000E). Metamorph software (version.6.3r7; Universal Imaging, Media, PA) was used for image acquisition and analysis. Barbed end elongation rates were 7.6 μM−1 s−1 for Mg-ATP/ADP+Pi-actin with profilin and 2 μM−1 s−1 for Mg-ADP-actin.
We are indebted to H. Balcer for early biochemical observations that inspired this study, K. Okada and S. Dinani for technical help, and V. Okreglak for generous advice on live-cell imaging. We thank F. Chaudhry, M. Chesarone, K. Daugherty, A. G. DuPage, P. Lappalainen, K. Okada, O. Quintero-Munzon, and A. Rodal for comments on the manuscript. This work was supported by grants from the ANR (PCV 135054) to L.B. and NIH (GM63691) to B.G.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.