Explication of the Aip1p/cofilin/actin filament complex may lead to a more detailed understanding of the mechanisms by which Aip1p and cofilin collaborate to rapidly disassemble filaments. We further characterized the actin–Aip1p interface through a random mutagenic screen of ACT1, identifying a novel Aip1p interaction site on actin. This finding is consistent with our current ternary complex model and offers insights into how Aip1p may disturb intersubunit contacts within an actin filament. In addition, site-directed mutagenesis aimed at interfering with salt bridge interactions at the predicted Aip1p–cofilin interface revealed hyperactive alleles of cof1 and aip1 that support the ternary complex model and suggest that conformational changes in cofilin structure may be transmitted to actin filaments, causing increased destabilization. Furthermore, these data support an active role for Aip1p in promoting actin filament turnover.
RAPID actin filament turnover is crucial for maintaining adaptable cytoskeletal networks. Ongoing cellular dynamics and a finite supply of monomeric actin demand that existing filaments are efficiently recycled through an intricate progression of disassembly, exchange of monomer-bound nucleotides, and readdition of ATP-actin subunits to the barbed ends of elongating filaments. To expedite filament disassembly, an assortment of actin-binding proteins collaborate to accelerate the pointed-end off rate, block barbed-end subunit addition, sever filaments into smaller fragments, and sequester monomeric actin. Among the most prominent of these proteins is cofilin, which binds to F- and G-actin at a 1:1 ratio and has historically been implicated in enhancement of pointed-end disassembly (Carlier et al. 1997) and severing of actin filaments (MacIver et al. 1991). However, recent in vitro findings refute the assertion of increased disassembly from pointed ends and limit severing activity to extreme substoichiometric levels of cofilin to actin, proposing that cofilin stabilizes actin filaments when present at isostoichiometric levels (Andrianantoandro and Pollard 2006). Such analyses are valuable in delineating the potential roles of cofilin through a range of concentrations. However, cofilin's in vivo activities proceed among an abundance of other actin-interacting proteins that may function to regulate, antagonize, or augment cofilin activities. The most notable enhancer of cofilin activity is actin-interacting protein 1 (Aip1p), which always colocalizes with cofilin and has been shown to enhance filament severing by cofilin while also exhibiting a cofilin-dependent barbed-end regulatory effect that may involve filament capping (Okada et al. 1999, 2002, 2006; Rodal et al. 1999; Balcer et al. 2003; Mohri et al. 2004; Ono et al. 2004; Clark et al. 2006). Significantly, Aip1p's activity has been observed only at ratios of cofilin:actin >1:5 (Rodal et al. 1999) and thus Aip1p is not predicted to be active when cofilin is at the levels believed to be optimal for severing, but would be highly active at the stoichiometries at which cofilin alone is predicted to stabilize filaments. This has profound implications for the perceived in vivo function of cofilin over a broad range of concentrations. The experimentally estimated ratio of Aip1p:cofilin:actin in the cell is 1:1:5 or 1:1:10 (Rodal et al. 1999). However, localized quantities of Aip1p and cofilin relative to actin are expected to be much greater at sites such as cortical actin patches. Due to the large amount of Aip1p maintained in vivo, its activities must be taken into consideration when establishing how cofilin behaves in the cell.
Aip1p's role in the cell could include expansion or destruction of actin networks, and possibly both, depending upon its precise mechanism of activity, its local concentration, and the presence of additional actin-interacting proteins. For example, if Aip1p is involved only in enhancing severing, then we expect that it would allow for the elaboration of actin networks by creating more barbed ends. If it also plays a role in capping barbed ends, then we expect it to break down actin networks by shortening filaments and allowing them to disassemble from pointed ends. Similarly, if Aip1p-enhanced severing occurs in the presence of capping protein (Cap1p/Cap2p), filament disassembly would occur. On the basis of preliminary findings, we have previously proposed a novel model for actin filament barbed-end regulation by Aip1p in which cofilin-dependent Aip1p severing increases the number of cofilin-bound barbed ends, which could be prohibitive toward binding by profilin-bound actin subunits (Clark et al. 2006). In addition to its molecular function of promoting actin filament disassembly/turnover, various physiological roles have also been observed. These include maintenance of cell morphology in Drosophila (Rogers et al. 2003), organization of body-muscle wall in Caenorhabditis elegans (Ono 2001), reliability of chromosome segregation, motility, cytokinesis, and endocytosis in Dictyostelium (Konzok et al. 1999; Gerisch et al. 2004), leaf size and viability in Arabidopsis (Ketelaar et al. 2004), assistance in disassembly of Listeria comet tails (Brieher et al. 2006), as well as endocytic vesicle internalization (Clark et al. 2006) and turnover of actin cortical patches and actin cables (Okada et al. 2006) in Saccharomyces cerevisiae.
The binding interface between cofilin and actin has been well characterized through mutational analyses incorporating genetic, cell biological, and biochemical approaches, as well as by visualization using cryo-electron microscopy and molecular modeling (Amberg et al. 1995; Rodal et al. 1999; Lappalainen and Drubin 1997; McGough et al. 1997; Bobkov et al. 2002; Galkin et al. 2003). Recent studies of the Aip1p–actin interaction have detailed two distinct sites on Aip1p involved in F-actin binding, one on each β-propellor domain (Mohri et al. 2004; Clark et al. 2006; Okada et al. 2006). While the actin-binding site on Aip1p's N-terminal β-propellor domain has been confirmed in S. cerevisiae and C. elegans, the C-terminal actin-binding domain has been detected only in yeast, and none of the identified residues for this site are conserved. Thus, it is not clear if it is maintained in other species. Findings indicate that the C-terminal binding site is the weaker of the two, although both are necessary for proper Aip1p activity in vivo and in vitro (Clark et al. 2006; Okada et al. 2006). Analysis of ACT1 charged-to-alanine mutant alleles on the basis of yeast two-hybrid data has outlined a likely Aip1p-binding site on actin, although it is not clear if both of Aip1p's propellers can bind to this site or if the site is exclusively bound by one of Aip1p's actin-binding domains (Rodal et al. 1999). Assuming the latter, it is likely that actin contains a second Aip1p-binding site.
Recently the direct physical interaction between cofilin and Aip1p has also been characterized (Clark et al. 2006). This complex was first postulated on the basis of yeast two-hybrid data demonstrating that the Aip1p–actin interaction is dependent upon the ability of actin to bind cofilin and that Aip1p and cofilin's actin-binding sites appear to overlap (Rodal et al. 1999). An unbiased mutagenesis strategy of Aip1p in conjunction with computational molecular docking with cofilin predicted a cofilin-binding footprint on the concave surface of Aip1p, stretching across the cleft with essential points of contact on each of the two β-propellor domains (Clark et al. 2006; Figure 1). Although the precision of the molecular modeling approach revealed a high level of complementarity at the proposed interface, including eight predicted salt bridge interactions, it is imperative that this model be supplemented by mutational studies to confirm the predicted orientation of cofilin within Aip1p's binding cleft.
To more thoroughly define the Aip1p–actin physical interaction, we have undertaken a genetic dissection of ACT1 along with subsequent synthetic genetic and biochemical analyses to map a previously unidentified Aip1p-binding site on actin. In addition, we have provided biochemical data that support the Aip1p–cofilin molecular model in regard to the specific orientation of cofilin within the concave surface of Aip1p. Surprisingly, through this effort we have discovered and characterized three hyperactive cofilin mutants and two hyperactive aip1p mutants. These unique and previously unavailable resources have already contributed new insight regarding the role played by Aip1p in actin filament disassembly and will prove to be useful reagents in future studies. Cumulatively, these findings represent an important step forward in understanding the structural and mechanistic properties of the Aip1p–cofilin–actin ternary complex.
MATERIALS AND METHODS
Yeast two-hybrid analysis of random act1 mutants:
All yeast strains used in this article are listed in Table 1. Plasmid pDAb7 (also known as pRB1516), encoding a fusion of the GAL4 DNA-binding domain (DBD) to ACT1 in vector pAS1-CYH2 (Amberg et al. 1995), was randomly mutated by treatment with hydroxylamine (Amberg et al. 2005) and lithium acetate (LiOAc) transformed (Rose et al. 1989) into yeast strain Y190 (Durfee et al. 1993). Successful transformants were selected on casamino acid–tryptophan medium and then analyzed for their abilities to interact with Aip1p and cofilin by yeast two-hybrid analysis (Fields and Song 1989). Briefly, Y190 cells carrying mutant act1-DBD fusion plasmids were spotted on SC–trp and allowed to grow overnight. These were then replica plated to two duplicate YPD plates, which were each overlayed with a lawn of yeast strain Y187 transformed with plasmid pAip6 encoding a fusion of AIP1 to the GAL4 DNA activation domain (Amberg et al. 1995) or plasmid pJT20 encoding a fusion of COF1 to the GAL4 DNA activation domain. The cells were allowed to mate for 1 day after which diploids were selected on SC media lacking tryptophan and leucine. After 2 days, diploids were replica plated to SD medium plus 10 μg/ml adenine and 25, 50, or 100 mm 3,5-aminotriazole. Each transformant was screened for loss of the Aip1p interaction while maintaing the cofilin interaction. Mutants showing such a defect were rescued, confirmed by retesting, amplified in Escherichia coli by standard methods (Rose et al. 1989), and submitted for sequencing. Images of act1p mutants were generated using Insight II Version 2000 (Molecular Simulations, San Diego). Coordinates for Act1p were retrieved from the Collaboratory for Structural Bioinformatics Protein Data Bank (PDB file 1ATN).
aip1-GST and cof1-GST mutant plasmid construction:
Vector pAR3 (Rodal et al. 1999) was the template for site-directed mutagenesis of AIP1 by overlap extension fusion PCR using external primers MCo-179 (5′-GCGCGGGATCCATGTCATCTATCTCTTTG-3′) and MCo-180 (5′-GCGCGAAGCTTTCACTCGAGGACAACATT-3′), which contain a 5′ BamHI or HindIII site, respectively. Internal primers were specific to the mutant allele. Final PCR products were digested with restriction enzymes BamHI and HindIII and cloned into the same sites of pAR3. All mutants were confirmed by sequencing.
Vector pBH360 was created by cloning a cofilin PCR product into the BamHI and EcoRI sites of pGEX-2T (GE Healthcare). The primers used were BHp40 (5′-CTCGGATCCAGATCTGGTGTTGCTGTTGCTGATGAATCC-3′) and BHp48 (5′-GTGAGAATTCTTAATGAGAACCAGCGCCTCTGC-3′), which contain a 5′ BamHI or EcoRI site, respectively. This vector was the template for site-directed mutagenesis of COF1 by overlap extension fusion PCR using the external primers listed above and internal primers specific to the mutant allele. Final PCR products were digested with restriction enzymes BamHI and EcoRI and cloned into the same sites of pBH360. All mutants were confirmed by sequencing.
Genomic integration of mutant alleles:
To generate haploid strains carrying act1 mutant alleles, vector pRB1456 (ACT1:HIS3) was the template for site-directed mutagenesis by overlap extension fusion PCR using external primers DAo-Act1-50 (5′-GATCCTTTCCTTCCCAATCTC-3′) and DAo-Act1-53 (5′-CCCAGAAACAAAGGGTATGAG-3′) and internal primers specific to the mutant allele. Final PCR products were LiOAc transformed into diploid yeast strain DAY111x112, allowed to recover overnight in YPD, and plated on SC–his to select successful integrants. Heterozygous mutant diploids were then sporulated and dissected, and haploids containing the act1:HIS3 cassette were isolated. Sequencing of PCR products from genomic DNA templates confirmed that all mutants were integrated and that no extra mutations were present.
To generate strains carrying mutated aip1 genes, a strategy using triple integration of three DNA fragments was implemented. Fragment 1 was a PCR product that included ∼100 bp of genomic sequence upstream of AIP1 and the first 66 bp of AIP1 (primers LGo-Aip1-1, 5′-AATACTAGCTATTGCTTTCCG-3′, and MCo-184, 5′-AAAGTTATCCTGTGTCGAAGGCTGAGG-3′) from template pMC60 (Clark et al. 2006). Fragment 2 was a PCR product that included the final 90 bp of AIP1 followed by an inserted G418r marker gene and ∼100 bp of genomic sequence downstream of AIP1 (primers MCo-235, 5′-CACAAGAGAGGTGTTAACAACCTTTTA-3′, and LGo-Aip1-4, 5′-TTCTCATGTTCAACTTCGGAA-3′) from template pMC60. Fragment 3 was a BamHI/HindIII fragment from the appropriate aip1-GST vector. The three fragments were cotransformed into an aip1Δ∷URA3 haploid strain such that genomic integration of the mutant aip1 gene required prior recombination with fragments 1 and 2 near its start and stop codons, respectively. Successful transformants were selected on YPD+G418 media and the loss of ability to grow on media lacking uracil. Integration location, presence of the proper mutation, and absence of additional mutations were confirmed by sequencing PCR products generated from using the external primers LGo-Aip1-1 and MCo-Aip1-130 (5′-AGTCTTTTCCTTACCCAT-3′, inside G418r gene) on genomic DNA templates.
Cofilin mutant integration was achieved in a similar fashion. Fragment 1 contained ∼100 bp of genomic sequence upstream of COF1, continued to base pair 195 of COF1, and including the intron (primers MCo-289, 5′-AGCAACTAGCAAAAA-3′, and MCo-214, 5′-GTAAAGAGCGTCGTTTTCTGGCAATTT-3′) from template pPL8 (Lappalainen et al. 1997). Fragment 2 contained the final 63 bp of COF1 followed by an integrated LEU2 marker and ∼100 bp of downstream genomic sequence (primers MCo-241, 5′-TTGGAAGATGTCAGCAGAGGCGCTGGT-3′, and MCo-292, 5′-CTTCCAAAGCGTGAG-3′) from template pPL8. Fragment 3 was a BamHI/EcoRI fragment from the appropriate cof1-GST vector. Fragments were transformed into a cof1Δ∷G418r heterozygous diploid and successful transformants were selected on media lacking leucine and for the inability to grow on media containing G418. Integration location, presence of the proper mutation, and absence of additional mutations were confirmed by sequencing PCR products generated from using external primers BHp40 and BHp48 (both described above) on genomic DNA templates.
Synthetic sick/lethal testing of act1 mutants:
For act1 × cof1-4 crosses, act1 mutant strains (MATa act1-x:HIS3) were crossed to MCY51 and diploid selection was achieved by picking zygotes. Diploid strains were transformed with vector pJT59 (2μ Aip1p overexpression vector), sporulated in sporulation media lacking uracil to maintain the vector, and dissected. Single- and double-mutant haploids containing the AIP1 overexpression vector pJT59 (2.2-kb genomic BstBI fragment containing AIP1 cloned into the BstBI site of YEplac195) were grown to saturation in casamino acid–ura media or YPD. Cultures were then diluted and 10× serial dilutions from 5000 to 5 cells/μl were spotted on appropriate media (2 μl/spot). Cas–ura cultures were spotted onto cas–ura plates to show growth in the presence of aip1 overexpression. YPD cultures were spotted on 5-FOA plates to show growth in the absence of aip1 overexpression (endogenous Aip1p was still present). Growth proceeded for 3 days at 30°. act1 × sac6Δ crosses were done the same way, using the sac6Δ null allele from the EUROSCARF nonessential gene deletion collection in S. cerevisiae (Research Genetics, Birmingham, AL).
For act1 × aip1Δ and act1 × cof1-19 crosses, act1 strains were crossed to LGY3 and MCY53, respectively. Following sporulation and dissection, single and double mutants were grown to saturation, diluted as described above, and spotted onto YPD media. Growth proceeded for 3 days at 37°.
Yeast actin was purified by a modified DNaseI affinity purification procedure (Goode 2002). Briefly, 10 liters of yeast strain FY23x86 were grown to saturation, resuspended to a final volume of 120 ml in G-buffer (+ATP, DTT, PMSF, and Calbiochem protease inhibitor cocktail), and passed twice through a French press at 1200 pounds-force per square inch gauge (PSIG). The lysate was clarified in a Beckman JA-20 rotor at 12,000 rpm for 30 min at 4° and then in a Beckman Ti50.2 rotor at 50,000 rpm for 50 min at 4°. The supernatant was loaded in equal volumes onto two DNase I-sepharose columns, each with a 5-ml bed volume (DNAse I, Roche; Sepharose 4B, Sigma, St. Louis), at a flow rate of ∼1–2 ml/min. Columns were washed with 25 ml G-buffer + 10% deionized formamide, 25 ml G-buffer + 0.2 m NH4Cl, and 25 ml G-buffer. The actin was eluted with 25 ml G-buffer + 50% deionized formamide and dialyzed overnight in 1 liter G-buffer (0.05 μm ATP). Samples were concentrated using an Amicon Ultra spin column (Millipore, Bedford, MA) and then further purified by HPLC [Bio-Rad (Hercules, CA) Biologic DuoFlow] on a UNO-Q1 column (washed at 1 ml/min with G-buffer and then eluted with a linear KCl gradient of 100–400 mm in G-buffer). Peak fractions were polymerized in F-buffer for 30 min at room temperature, and then the KCl concentration was increased to 0.6 m and the samples were incubated for 1 hr. Polymermized actin was pelleted by ultracentrifugation in a Beckman TLA100.2 rotor at 90,000 rpm. Pellets were resuspended in G-buffer to 1–2 mg/ml, incubated on ice for 4 hr, dialyzed overnight in 1 liter G-buffer, snap frozen, and stored at −80°. Concentrations were determined by Bradford assay.
Yeast cofilin was expressed in E. coli DH5α cells as a glutathione-S-tranferase (GST) fusion protein under the control of the Plac promoter. GST–cofilin was purified as previously described (Lappalainen and Drubin 1997), but with the following alterations. Cells were grown to stationary phase in 1 liter of LB+ampicillin, and then cofilin expression was induced by treatment with 200 mm IPTG for 3 hr. PBS was substituted with TBS (50 mm Tris, pH 7.5, 100 mm NaCl) in all relevant steps. Following induction, cells were passed twice through a French press at 1200 PSIG, and then were spun in a Beckman centrifuge using a JA-20 rotor at 12,000 rpm for 30 min. The supernatant was extracted, added to 2 ml of a 50% resin slurry, and rocked for 30–60 min at 4°. After washing with 4 × 10 ml cold TBS, 5 units of thrombin (Novagen) was added to the beads (3 ml TBS, 10.4 μl 0.5 m CaCl2) and incubated at room temperature for 4 hr. Following cleavage, the supernatant was collected and the resin was washed with 2 × 2 ml TBS; washes were added to the supernatant. This sample was dialyzed overnight against 50 mm Tris, pH 7.5 at 4°. Samples were further purified by HPLC (Bio-Rad Biologic DuoFlow) on a UNO-Q1 column and washed at 1 ml/min with 50 mm Tris, pH 7.5, and then eluted with a linear KCl gradient (100–400 mm) in 50 mm Tris, pH 7.5. Peak fractions were dialyzed overnight against 10 mm Tris, pH 7.5, and 50 mm NaCl for 5 hr at 4°, concentrated (Microcon YM-10 columns), confirmed by visualization on a polyacrylamide gel, and stored at −80°.
Yeast Aip1p was expressed in yeast strain FY86 as a GST fusion protein under the control of a galactose-inducible promoter [pEG(KT)]. Cells were inoculated into 1 liter SC–ura + 3% glycerol + 1% EtOH + 0.1% glucose and grown for 20–24 hr at 30°. Galactose was added to a 2% final concentration and induction continued for 10 hr. GST–Aip1p purification proceeded as described for GST–cofilin, with the following alterations. TBS (150 mm, Tris pH 7.5, 100 mm NaCl) was used in all relevant steps. Prior to binding to glutathione–agarose resin, cell lysate was pelleted in a Beckman centrifuge using a JA-20 rotor at 12,000 rpm for 30 min, and then the supernatant was spun again in a Beckman ultracentrifuge using a 70 Ti rotor at 50,000 rpm for 50 min at 4°.
Actin filament pelleting assay:
To evaluate actin filament sedimentation in the presence of Aip1p and cofilin, 4.0 mm actin was polymerized at room temperature in F-buffer (1 mm Tris, pH 7.5, 0.7 mm ATP, 0.2 mm CaCl2, 2 mm MgCl2, 285 mm KCl, 0.001 mm EGTA, 0.2 mm DTT) for 1 hr. A total of 12.5 μl polymerized actin was aliquoted into polycarbonate ultracentrifuge tubes (Beckman Instruments, Fullerton, CA) and 12.5 μl of a premixed Aip1p, cofilin, F-buffer, and water mixture was added to each, yielding final concentrations of 2 mm actin and 1× F-buffer. The reactions were gently vortexed and incubated at room temperature for 20 min and centrifuged at 90,000 rpm for 20 min at 25° in a TLA100 rotor (Beckman Instruments) to pellet the actin filaments. Equal proportions of the pellets and supernatants fractions were run on 13% SDS–PAGE gels and proteins were visualized by SYPRO Ruby staining (Invitrogen).
Immunofluorescence was performed by standard protocols using a methanol/acetone fixation (Amberg et al. 2005). Affinity-purified anti-Aip1p antibody (primary) was used at a dilution of 1:100 (Rodal et al. 1999). FITC-conjugated goat anti-rabbit IgG (Cappel; ICN Biochemicals) was used at 1:1000.
Staining of the actin cytoskeleton was performed using a standard protocol (Amberg et al. 2005). Briefly, a yeast cell culture grown to 2 × 107 was subjected to incubation with EM-grade formaldehyde at a final concentration of 4%, washed with PBS, and treated with rhodamine-labeled phalloidin (1:10 dilution of 6.6 μm in methanol). After washing again with PBS, cells were suspended in mounting solution and viewed by fluorescence microscopy.
Latrunculin A plate assay:
Yeast cultures were grown to stationary phase, and then 10 μl of culture was mixed into 2 ml of 2× YPD. A total of 2 ml of 1% agar (55°) was then added and the mixture was vortexed and evenly poured onto the surface of thin YPD agar plate. Once the top layer of agar solidified, four discs of Whatman's paper were placed on the agar surface and 10 μl of the following was pipetted onto each respective disk: ddH2O, 0.5 mm latrunculin A (LatA), 1.0 mm LatA, and 2.0 mm LatA. The plates were then incubated for 2 days at 30° and scored.
Coordinates of actin, Aip1p, and cofilin were obtained from the Protein Data Bank (PDB files 1YAG, 1PI6, and 1CFY, respectively).
Identification of additional Aip1p interaction sites on actin by differential interaction screening:
The Aip1p- and cofilin-binding footprints on actin's surface have been previously described on the basis of yeast two-hybrid analyses, as illustrated in Figure 1A (Rodal et al. 1999). The cofilin-binding site depicted is also in agreement with cryo-EM and molecular modeling data (Galkin et al. 2003). By combining these data with our recently described model of the Aip1p–cofilin complex, we were able to extrapolate how Aip1p is likely to bind cofilin-decorated actin filaments. From these observations, it appears that Aip1p's N-terminal actin interaction site binds in the vicinity of the previously characterized Aip1p-binding site on subdomain 4 (SD4) of actin. However, Aip1p's C-terminal actin interaction site is not located near SD4, but rather appears most likely to interact with SD1 of actin. Therefore, we sought to provide experimental support for this novel Aip1p-binding site on actin through an unbiased mutagenesis screen.
The ACT1 gene was randomly mutated and a yeast two-hybrid screen was implemented to identify mutant isolates that demonstrated a differential interaction phenotype in which the actin mutant could interact well with cofilin, but not with Aip1p. Of five isolates meeting this standard, two contained an A7T point mutation, and the remaining three contained one each of the following point mutations: D11N, D157N, and P332H (Figure 2). The residues altered by the act1-212 (D157N) and act1-213 (P332H) alleles fall within the previously identified cofilin-binding footprint (Rodal et al. 1999), suggesting that the Aip1p-specific phenotype results from an alteration in cofilin binding that does not prevent the cofilin two-hybrid interaction, but alters it in such a way that Aip1p binding is prohibited (Figure 1A). Interestingly, the amino acid changes conferred by the act1-210 (A7T) and act1-211 (D11N) alleles are on the opposite side of the cofilin-binding site from where the previously identified Aip1p-binding site is located, suggestive of a novel Aip1p-binding site on actin (Figure 1A).
To gain more insight into other potential sites of Aip1p interaction on the actin surface, several actin- and cofilin-specific mutant aip1 alleles previously identified by our group, aip1-56 and aip1-59 (Clark et al. 2006), were tested by two-hybrid analysis against the Wertman collection of cluster-charged-to-alanine Wertman alleles (Wertman et al. 1992). Our intention was that by weakening the affinity of Aip1p for the cofilin–actin complex, additional points of contact between Aip1p and actin could be identified. Of the 28 Wertman mutants not previously associated with the Aip1p or cofilin interactions, act1-101p (D363A, E364A), act1-105p (E311A, R312A), and act1-113p (R210A, D211A) emerged as defective for the Aip1p interaction (Figure 2). The residues altered by act1-105 are located between the known Aip1p and the cofilin-binding footprints, while those altered by act1-113 lie within the known cofilin-binding footprint (Figure 1A). Interestingly, the residues mutated by act1-101, like those changed by act1-210 and act1-211, are on the opposite side of the cofilin-binding site from the known Aip1p-binding footprint, supporting the belief that a second Aip1p interaction site exists on actin (Figure 1A).
Synthetic genetic phenotypes support a second Aip1p interaction site on actin:
To strengthen our argument for a novel Aip1p-binding site on actin, we implemented a series of genetic crosses to determine if our new actin mutants behave in a manner consistent with what we would expect from an actin mutant that is exclusively defective for Aip1p binding. These data are summarized in Table 2. Each act1 allele was integrated into the yeast genome, with the exception of the act1-211 allele, which we were unable to recover, possibly indicating a dominant lethal phenotype. To specifically investigate actin mutants that had impairments in Aip1p binding and not major structural defects and/or inabilities to interact with other important actin-binding proteins, only mutant strains that had near-normal-growth phenotypes were considered for genetic experimentation (as an aip1Δ strain grows normally). This eliminated the act1-101 and act1-105 alleles, as well as the alleles that make up the previously identified Aip1p-binding footprint: act1-109 and act1-111 (Wertman et al. 1992). Also, act1-212 had a considerable growth defect (Table 2), possibly due to its proximity to the ATP-binding pocket of actin. The remaining act1 mutants (act1-113, act1-210, and act1-213) were analyzed for the ability to phenocopy an aip1Δ allele, which is synthetic lethal with cof1-4 (Rodal et al. 1999). Simultaneously, the ability of AIP1 overexpression to suppress any observed synthetic lethality was tested. Mutant act1 and cof1-4 haploid strains were mated and the resulting double heterozygous diploids were transformed with a URA3-marked AIP1 overexpression 2μ vector. Transformants were then sporulated and dissected, and double-mutant haploids containing the overexpression vector were isolated. These strains were tested by serial dilution growth assays on media lacking uracil to maintain the vector and on media containing 5-FOA to select for loss of the vector (Figure 3A). act1-113 is not synthetic lethal with cof1-4, which is inconsistent with a failure to interact with Aip1p in vivo. Both act1-210 and act1-213 are synthetic lethal with cof1-4, and this synthetic lethality is able to be suppressed by AIP1 overexpression, consistent with in vivo loss of Aip1p binding in the actin mutants.
To show that AIP1 overexpression can specifically suppress the act1 alleles as opposed to just suppressing cof1-4, we utilized the known severe synthetic slow-growth interaction between AIP1 and SAC6 (Rodal et al. 1999; Clark et al. 2006). The sac6Δ allele alone also has a moderate-growth phenotype that is not suppressed by overexpression of AIP1 (Figure 3B). As observed when testing with the cof1-4 allele, act1-210 phenocopies the aip1Δ allele by demonstrating a severe growth defect when combined with the sac6Δ allele. However, act1-213 has only a slight growth defect, which is not consistent with an aip1Δ phenotype. cof1-4 is also severely sick with sac6Δ. AIP1 overexpression did suppress the growth defect of act1-210 sac6Δ and cof1-4 sac6Δ double mutants, allowing their growth phenotypes to improve, but never to levels better than that of the sac6Δ allele alone. Thus, suppression is specific to the act1-210 allele and is consistent with its inability to interact with Aip1p by two-hybrid interaction. Notably, the act1-210 sac6Δ double mutant is sicker than would be expected, even in the presence of AIP1 overexpression. It is likely that this mutation also disrupts the interaction between actin and an unidentified actin-binding protein, which we believe is specific to SAC6 on the basis of findings described below.
We next sought to genetically differentiate the actin alleles on the basis of the locations of charged residues within (act1-213p) or outside (act1-210p) of the cofilin-binding footprint. To this end, we tested for synthetic growth defects in act1/aip1Δ double-mutant haploids (Figure 3C). Since an aip1Δ allele causes no growth defects alone, we expected that any actin mutant specifically deficient for Aip1p binding would be redundant with the aip1Δ allele. Therefore, act1 aip1Δ double mutants should grow normally. However, if an actin mutant's failure to interact with Aip1p results from propagation through a cofilin-related defect, as we expected to be the case for the a mutant within the cofilin-binding site, we predicted that additional functional defects would demonstrate a synthetic growth defect with the aip1Δ allele, as is true for several cofilin mutants (Rodal et al. 1999). Consistent with this reasoning, we found that when act1 aip1Δ strains were tested at 37° for slow-growth phenotypes, act1-213 was very slow while act1-210 was comparable to wild type (Figure 3C). Using similar logic, we then tested for synthetic growth phenotypes in act1 cof1-19 double mutants (Figure 3D). cof1-19 is defective for its Aip1p interaction and thus should grow normally if actin is unable to bind Aip1p (Rodal et al. 1999; Clark et al. 2006). Testing at 37° showed that act1-213 causes severe defects when combined with cof1-19, while act1-210 was similar to wild type. These findings support our proposition that the residues mutated in act1-213 are within the cofilin interaction domain and are disruptive to normal cofilin binding, while the residues mutated in act1-210 are outside of the cofilin interaction domain and do not interfere with cofilin binding.
In vitro analyses of act1-210p and act1-213p were conducted by actin pelleting assays and revealed no functional defects (data not shown). While unexpected, these observations are not without precedent as Aip1p mutants defective for the two-hybrid interaction with cofilin or actin (Clark et al. 2006) also did not show a defect in the actin pelleting assay. We attribute these findings to the extensive binding interface of Aip1p with cofilin-bound actin filaments and have found that in vivo analyses are most effective in detecting mutant phenotypes involving disruption of this complex.
Mutagenesis strategy for confirmation of the Aip1p–cofilin model:
We recently provided genetic and biochemical data demonstrating that a cofilin-binding site stretches across Aip1p's concave surface, making essential points of contact on each of the two β-propeller domains (Clark et al. 2006). In addition, we presented a molecular model based on a computational docking simulation that precisely details how these two proteins could fit together, including eight stabilizing salt bridge interactions at the binding interface (Figure 1B). Although the model is supported by genetic and biochemical evidence, additional mutational data are necessary to corroborate the precise orientation of cofilin within the concave surface of Aip1p. Unfortunately, the transient and actin filament-dependent nature of this interaction does not allow for analysis of a stably bound complex.
Charge reversal mutagenesis reveals hyperactive cofilin and Aip1p mutants:
During a mutagenic analysis of amino acid residues predicted to participate in salt bridge interactions between Aip1p and cofilin, we unexpectedly isolated three hyperactive cofilin mutants that are Aip1p independent. Listed from least to greatest for enhancement of actin filament disassembly, these mutants are cof1-157p, cof1-159p, and cof1-158p, which encode for the amino acid changes R135D, R82D, and K80E, respectively (Figure 4A and Table 3). These residues cluster relatively close to one another on the same surface at the narrow end of the cofilin molecule (Figure 1B), indicative that changes within this region lead to their enhanced activity. Increased filament disassembly was not detected at an actin:cofilin ratio of 5:1 or higher. Although the nature of this hyperactivity is not clearly defined, the high relative concentration of cofilin required for observation of increased disassembly indicates that the hyperactivity is not due to sequestering of cofilin-bound actin subunits, as this would have caused actin's pellet:supernatant ratios to mirror the actin:cofilin ratio of each sample, which was not the case. Furthermore, inability to detect hyperactivity by these mutants at actin:cofilin stoichiometries <2:1 suggests that the hyperactivity results when filaments are more densely bound by cofilin. When F-actin, wild-type cofilin, and hyperactive cofilin were combined at a ratio of 20:20:1, no gain in activity was observed compared to wild-type cofilin alone, indicating that these cofilin mutants do not mimic the phenomenon by which low levels of Aip1p lead to rapid disassembly of cofilin-decorated actin filaments (Rodal et al. 1999) (data not shown). Therefore, these data suggest that the activities of hyperactive cofilins are specific to the way in which they stably bind to actin filaments. It is not clear if this causes increased depolymerization or enhanced filament severing.
Surprisingly, when we tested the activity of aip1p mutants that contain the reciprocal charge reversal for each respective hyperactive cofilin mutant (Figure 1B), each was found to have a cofilin-dependent enhancement of actin filament disassembly (we were unable to express the D85K or a D85R mutant) (Figure 4B). While the intensified activities associated with these are not as robust as those observed for the cofilin mutants, it is intriguing to note that the level of increased disassembly associated with each Aip1p mutant corresponds to the relative gains of its respective salt bridge partner on cofilin. These observations suggest that the hyperactive aip1p mutants act by inducing a change in cofilin, causing it to transition from normal to hyperactive. This hypothesis is consistent with the Aip1p–cofilin docked structure in that the salt bridge residues involved in enhanced disassembly line up with one another as the model predicts.
Genetic analyses of cof1 and aip1 hyperactive mutants:
To characterize how the cof1p and aip1p hyperactive mutants behave in vivo, cof1-157∷LEU2, cof1-159∷LEU2, aip1-150∷G418r, and aip1-151∷G418r cassettes were each recombined into the normal COF1 or AIP1 genomic loci in S. cerevisiae and were subjected to an array of analyses (summarized in Table 4). We were unable to recover genomic integrants of cof1-158, suggesting that the allele may be lethal in vivo. Normal expression of all integrated mutants was confirmed by Western blotting (data not shown). Of the integrated alleles tested, only cof1-159 caused cell growth or morphology defects at 30°. Also, actin cytoskeletal staining by rhodamine–phalloidin revealed that cof1-159 has a disorganized actin cytoskeleton and excessive filamentous actin structures (Figure 5A). Interestingly, this phenotype indicates a loss of function rather than a gain of function for cof1-159 in vivo. To uncover defects in cof1-157, aip1-150, and aip1-151, growth on media containing Lat A was assayed and a series of genetic crosses were implemented (summarized in Table 4). Both aip1 and cof1-157 alleles demonstrated slight and moderate levels of resistance, respectively, to latrunculin A (data not shown), indicating increased actin filament stabilization rather than weakened filaments, consistent with the cytoskeletal defects of cof1-159. When each was combined with a tropomyosin null allele (tpm1Δ), which has a slow-growth defect resulting from destabilized actin cables, no additive growth impairments were observed. In addition, cof1-157 and aip1 mutants were combined with the act1-159 allele, which has a hyperstabilized actin cytsokeleton, and the act1-212 allele, which is a mutant introduced earlier in this work that seems to have a weakened actin cytoskeleton potentially due to a mutation located in the ATP-binding cleft. cof1-157 is synthetic lethal with act1-159 but not with act1-212, again indicating a decreased ability to destabilize actin filaments. Surprisingly, aip1-150 was also synthetic lethal with act1-159 but had no defect with act1-212, while aip1-151 was synthetic sick with act1-212 but not with act1-159. The differential phenotypes of these two aip1 mutants were unexpected due to their similar in vitro phenotypes, but were reconfirmed when each was combined with the cof1-4 allele, as aip1-150 was synthetic lethal and aip1-151 grew normally. Localization of these aip1p mutants by immunofluorescence revealed that aip1-151p localizes normally to cortical actin patches, while aip1-150p does not localize, explaining why it phenocopies an aip1Δ allele when crossed to cof1-4 and act1-159 (Figure 5B). The phenotypic defects of aip1-150 suggest that it does not properly bind actin and/or cofilin in vivo. Previous reports have noted that Aip1p mislocalizes in certain strains containing actin or cofilin mutants that do not properly bind Aip1p (Rodal et al. 1999), which may explain why aip1-150p is found to mislocalize. Cumulatively, these phenotypes show that the functional gains observed by these mutants in vitro do not carry over in vivo, perhaps due to altered binding kinetics, misregulation, or defective interactions with alternate ligands (such as Srv2p for cofilin) within the cell.
Actin's Aip1p interaction sites are consistent with the predicted Aip1p–cofilin molecular model:
Visual comparisons of the known cofilin and Aip1p-binding sites on actin to molecular models of the Aip1p–cofilin and cofilin–F-actin models lead us to believe that a second Aip1p-binding site must exist on actin. This was confirmed by an unbiased mutagenesis screen of actin that succeeded in isolating actin mutants that interact with cofilin but not with Aip1p in the yeast two-hybrid system. Genetic crosses revealed that one of these mutants (act1-210) precisely phenocopies an aip1Δ allele and can be suppressed by overexpression of AIP1. This mutation (A7T) falls within SD1 of actin and is superficially located such that it is likely to specifically disrupt an AIP1 interaction site. It is not surprising that the newly identified site was previously overlooked, as the original two-hybrid screen used to footprint the first Aip1p-binding site on actin (Rodal et al. 1999) implemented cluster-charged-to-alanine alleles that do not cover this region of actin well. Given that our unbiased mutagenesis strategy identified a binding site in the same region that we anticipated on the basis of comparisons between Aip1p–cofilin and cofilin–actin molecular models, these findings also represent strong support for our previously predicted Aip1p–cofilin model.
Cofilin-induced actin filament destabilization is associated with a rotational change in the filament twist and disruption of the contacts between SD1 and SD2 of longitudinally adjacent actin subunits (McGough et al. 1997; Galkin et al. 2003). It is possible that further weakening of filaments is propagated through the Aip1p–actin SD1 interaction, potentially due to an additional torque exerted on the filament, excess force on cofilin that disturbs intersubunit contacts, or a conformational change that further separates SD1 from SD2. In vivo and in vitro observations agree that both of Aip1p's actin-binding sites are necessary for optimal F-actin binding and disassembly (Mohri et al. 2004; Clark et al. 2006; Okada et al. 2006). This favors a model in which both of Aip1p's propellers need to bind their respective actin-binding sites for function. On the basis of our crude reconstruction of the Aip1p–cofilin–actin filament model, it is not obvious how both of Aip1p's actin interaction sites could simultaneously bind to their respective contact sites on actin. However, we do not expect that this static model can serve to fully describe the dynamic complex, as we predict significant conformational alterations during the dramatic depolymerization of cofilin-decorated actin filaments that occurs upon binding by Aip1p.
Hyperactive mutants offer unique insight into the potential mechanism of actin filament disassembly:
An attempt to provide more definitive evidence for the Aip1p–cofilin model resulted in the inadvertent discovery of three hyperactive cofilin mutants and two hyperactive Aip1p mutants. Our original intention was to create repulsive charge reversal mutations at the site of each predicted salt bridge on one molecule and then alleviate the repulsion by making the compensatory mutation on the opposite molecule. However, the robust Aip1p–cofilin interaction in the presence of F-actin required us to create combinatorial Aip1p or cofilin mutants to see sufficient loss of biochemical activity in this study. Unfortunately, the use of multi-mutant proteins resulted in losses of function that could not be overcome by complementary charge reversals. This was particularly problematic for the cofilin alleles, which were often rendered defective for binding to actin filaments (Table 3).
Discovery of hyperactive mutations within cofilin and Aip1p have provided unique and previously unavailable reagents with which to study actin filament disassembly by the Aip1p–cofilin complex. Three cofilin mutants, cof1-157p (R135D), cof1-158p (R80E), and cof1-159p (K82D), each generate enhanced disassembly activities of varying intensity. Interestingly, these residues have previously been targeted by Lappalainen et al. (1997). In that study, a cluster-charged-to-alanine approach created the alleles cof1-16 (R80A, K82A) and cof1-22 (E134A, R135A, and R138A), both of which resulted in a loss of cofilin binding to filamentous but not monomeric actin in vitro (Lappalainen et al. 1997). Notably, these mutations do not reside at the putative interface with F-actin, but rather at the Aip1p interface, as we have demonstrated. Thus, the loss of F-actin binding by these alleles is indicative of widespread conformational alterations within the cofilin molcule. A similar observation was made previously, when the array of cluster-charged-to-alanine alleles of cofilin were used to footprint the actin- and Aip1p-binding domains on cofilin by two-hybrid analysis (Rodal et al. 1999). Sites of mutation that caused the same defects were observed on opposite sides of the molecule.
We suspect that each hyperactive cofilin mutation results in an intramolecular conformational change that allows cofilin to induce a more destabilizing torsional stress on the filament, leading to increased severing. The nature of such a conformational shift is not known, although we predict that these alleles produce similar structural alterations that may be discernible by molecular modeling or NMR.
The isolated hyperactive aip1p mutants, aip1-150p (D585R) and aip1-151p (E111R), contain mutations of residues that are the salt bridge partners with hyperactive cof-157p and -158p, respectively. We were unable to purify the salt bridge mutant paired with cof1-159p: aip1-152p (D85K) or aip1-182 (D85R). Our group has previously mutated these same residues as part of an Aip1p cluster-charged-to-alanine scan, in which two-hybrid analysis revealed normal Aip1p–cofilin and Aip1p–actin interactions (Clark et al. 2006). Therefore, we suspect that when these hyperactive aip1p mutants bind to cofilin, the charge reversals create a localized repulsion between the two molecules that is not sufficient to prevent the interaction but does exert a force against the R80 or R135 residue of cofilin strong enough to induce a conformational change that mimics the hyperactive cofilin alleles. The idea that the increased activity is transmitted from aip1p through cofilin is supported by three observations. First, the residues mutated in each hyperactive aip1p directly line up with and are expected to exert a repulsive force against the residues responsible for the functional gains in the hyperactive cof1p mutants. Second, the level of enhancement detected for each aip1p mutant correlates to the relative gain of function observed in its respective hyperactive cofilin salt bridge partner. Third, the residues mutated in each hyperactive aip1p are on opposite propellers, making it highly unlikely that the observed functional gains result from similar conformational changes in the overall structure of Aip1p.
The observed alignment of mutated residues between hyperactive alleles of these two molecules provides further support that the orientation of cofilin within our Aip1p–cofilin molecular model is correct. In addition, these alleles may offer an interesting perspective to the mechanism of actin filament disassembly by Aip1p and cofilin. Some published reports favor passive barbed-end capping of actin filaments as Aip1p's primary activity in filament disassembly (Okada et al. 2002; Balcer et al. 2003; Okada et al. 2006). However, doubt has been cast upon the existence and/or in vivo importance of this function (Ono et al. 2004; Brieher et al. 2006; Clark et al. 2006). For these reasons, we prefer a model in which Aip1p actively contributes to filament disassembly by enhancing severing by cofilin. The proposition that our hyperactive aip1p mutants act directly through the cofilin molecule by inducing a conformational change is consistent with an active role for Aip1p. This does not rule out capping as a function of Aip1p, but we have previously proposed that the perceived barbed-end regulation by Aip1p in vivo could involve its propensity for producing cofilin-bound barbed ends, which may serve as a selective gate for prohibiting binding by profilin-bound actin subunits (Clark et al. 2006). Furthermore, recent findings have shown that the barbed-end elongation rate of actin filaments saturated with Schizosaccharomyces pombe cofilin is decreased by approximately twofold, although human cofilin did not demonstrate this effect (Andrianantoandro and Pollard 2006). These experiments were done with actin and cofilins from mixed species; thus additional experimentation is needed to understand the extent of barbed-end regulation by cofilin from like sources and what the in vivo significance of this activity may be.
The inability to observe in vivo gain-of-function phenotypes for our integrated mutants (cof1-157, cof1-159, aip1-150, and aip1-151) is perplexing, likely reflecting the vast complexity of cofilin and Aip1p activity in the cell compared to when isolated in a test tube. Cellular regulation of cofilin and Aip1p has not been well characterized and may be responsible for preventing overactive cofilin or Aip1p from participating in uncontrolled filament disassembly. It is believed that tropomyosin blocks cofilin from binding actin cables and that PIP2 can bind cofilin to prevent its actin interaction (Yonezawa et al. 1990; Ono and Ono 2002; Okada et al. 2006). However, no Aip1p-specific modes of regulation have been identified. Given the rapid disassembly of actin filaments in vitro by Aip1p and cofilin, and the fact that Aip1p and cofilin are always found to colocalize at filamentous actin structures, it seems intuitive that some level of Aip1p-specific regulation occurs to allow filaments to exist.
The capabilities of cofilin and Aip1p to promote actin turnover in vitro have been well demonstrated. It is clear that the intensity of this activity, and possibly even the mechanism, can vary greatly, depending upon the relative concentration of each protein as well as the presence of regulatory and/or other actin-interacting proteins. Therefore, to further understand the cooperative nature of these proteins in vivo, it will be important to consider the local environment in which filament dynamics progress and to further elucidate the modes of regulation that moderate this process.
We thank David Sept for helpful input and discussions regarding the Aip1p–cofilin complex mutations, as well as members of the Amberg Laboratory for constructive contributions and manuscript review. This research was supported by National Institutes of Health grant GM-56189.
↵1 Present address: Department of Biology and Biotechnology, Worcester Polytechnic Institute, Worcester, MA 01609.
Communicating editor: B. J. Andrews
- Received February 13, 2007.
- Accepted April 27, 2007.
- Copyright © 2007 by the Genetics Society of America