![[Feedback]](/icons/banner/feedbackACT.gif){kind=icon}
![[For Subscribers]](/icons/banner/subscriberACT.gif){kind=icon}
![[Archive]](/icons/banner/archiveACT.gif){kind=icon}
![[Search]](/icons/banner/searchACT.gif){kind=icon}
![[Contents]](/icons/banner/tocACT.gif){kind=icon}
|
|
|
|
|
||
Vol. 16, Issue 7, 3128-3139, July 2005
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
* Department of Biology and Rosenstiel Basic Medical Science Research Center, Brandeis University, Waltham, MA 02454;
Departments of Biochemistry and Anatomy and Structural Biology, and Center for Synchrotron Biosciences, Albert Einstein College of Medicine, New York, NY 10461
Submitted January 24, 2005;
Revised April 11, 2005;
Accepted April 25, 2005
Monitoring Editor: Randy Schekman
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
) and is thought to link functions of the actin cytoskeleton to endocytosis in yeast and mammals (Qualmann and Kessels, 2002; Engqvist-Goldstein and Drubin, 2003). A role for yeast Abp1 in regulating endocytosis is suggested by direct associations of its proline-rich and Src homology 3 (SH3) domains with endocytic machinery, including Sla1, Rvs167/amphiphysin, and actin-regulating kinases Ark1 and Prk1 (Lila and Drubin, 1997; Colwill et al., 1999; Cope et al., 1999; Warren et al., 2002). Furthermore, Abp1 localizes to cortical actin patches (Drubin et al., 1988), which recent studies show are sites of endocytosis (Kaksonen et al., 2003; Huckaba et al., 2004). The importance of mammalian Abp1 in regulating endocytosis is supported by dominant-negative and RNA interference knockdown studies and physical interactions with dynamin (Kessels et al., 2001; Mise-Omata et al., 2003). Although the gene encoding S. cerevisiae ABP1 is nonessential (Adams et al., 1993), abp1 null mutants are synthetic lethal (lethal in combination) with mutations in three other genes that facilitate endocytosis: SAC6/fimbrin, SLA1, and SLA2/Hip1R (Holtzman et al., 1993). Thus, Abp1 has a shared role in at least one essential function in vivo. However, it has been unclear which activities and domain–ligand interactions of Abp1 are required for this cellular function(s).
Abp1 is a multidomain protein consisting of an N-terminal actin depolymerizing factor homology (ADFH) domain, two centrally located acidic motifs, a proline-rich region, and a C-terminal SH3 domain (Figure 1B). Actin filament binding by Abp1 requires its ADFH domain (Kessels et al., 2000; Goode et al., 2001), a module found in two other conserved families of actin binding proteins, ADF/cofilins, and twinfilins (Lappalainen et al., 1998). All three proteins use ADFH domains (
20% sequence identity) to interact with actin, yet each protein has highly distinct effects on actin dynamics. ADF/cofilins bind to actin filaments and monomers, and function primarily to promote filament severing and depolymerization (Carlier et al., 1997; Blanchoin and Pollard, 1999). Twinfilins have no affinity for filaments, but they tightly sequester monomers (Goode et al., 1998; Vartiainen et al., 2000). Abp1 binds to filaments and not monomers, yet does not affect filament dynamics (Kessels et al., 2000; Goode et al., 2001). Thus, the ADFH domain is a versatile module that can be adapted to interact with actin in diverse ways and to perform a range of functions.
|
; Fedorov et al., 1997; Leonard et al., 1997; Bowman et al., 2000; Pope et al., 2004), and their actin-binding interfaces have been defined by mutagenesis, synchrotron footprinting, and NMR methods (Sutoh and Mabuchi, 1989; Lappalainen et al., 1997; Moriyama and Yahara, 1999, 2002; Ono et al., 2001; Guan et al., 2002; Pope et al., 2004). Electron microscopy and computational studies have led to the view that ADF/cofilins bind filamentous actin through two surfaces (the G- and F-sites) located on opposite faces of the protein (reviewed in Ono, 2003) (Figure 7). Mutations at the G-site disrupt binding to both actin monomers (G-actin) and actin filaments (F-actin). Mutations at the F-site only disrupt binding to F-actin. The structure of one twinfilin ADFH domain has been reported, and mutagenesis shows that conserved residues in the G-site are critical both for binding G-actin and localization and function of twinfilin in vivo (Paavilainen et al., 2002). In contrast, the structure of the Abp1 ADFH domain has been unknown, its actin-binding surface has not been mapped, and the importance of actin binding for Abp1 localization and function in vivo has not been tested.
|
|
). The Arp2/3 complex is a highly conserved assemblage of seven proteins that, upon interaction with a nucleation-promoting factor (NPF), nucleates actin polymerization (Welch and Mullins, 2002). The most well studied NPFs are in the WASp/SCAR/WAVE family. Their mechanism of activation involves stabilizing Arp2/3 complex in a closed, actin nucleation-primed conformation (Goley et al., 2004; Rodal et al., 2004) and binding to and presenting an actin monomer to the complex to stimulate nucleation (Welch and Mullins, 2002). Yeast Abp1 and mammalian cortactin define a novel second class of Arp2/3 NPFs, because they bind to F-actin as opposed to G-actin (Goode et al., 2001; Uruno et al., 2001; Weaver et al., 2001). It has been postulated that they activate Arp2/3 complex by recruiting it to the sides of preexisting filaments, thereby promoting nucleation and/or stabilizing filament branches created by Arp2/3 complex. Consistent with this hypothesis, deletion of either the ADFH domain in Abp1 or the actin-binding domain in cortactin abolishes the stimulatory effects these NPFs have on Arp2/3 complex. However, the recruitment mechanism has not been tested using point mutations that disrupt actin binding without affecting other possible functions of these domains. Here, we present the crystal structure of S. cerevisiae Abp1 ADFH domain and characterize its interaction with actin by mutagenesis. We show that actin binding is required for 1) Abp1 nucleation-promoting activity on Arp2/3 complex in vitro; 2) localization of Abp1 to actin patches in vivo; 3) lethality caused by Abp1 overexpression; and 4) genetic functions of ABP1 that overlap with SAC6 and SLA2, but not SLA1. Our data reveal that Abp1 has multiple biological functions that require distinct combinations of its activities and domain–ligand interactions.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
= 106.9°). Native and ethyl mercury phosphate (EMP) derivative data were collected at room temperature to 3.0 Å resolution on a SIEMENS X-1000 area detector coupled to a Rigaku RU-200 generator and reduced with XDS (Kabsch, 1988), resulting in Rmerges of 7.2 and 8.4% for the native and EMP data sets, respectively. An additional high-resolution native data set to 2.1 Å was collected on beamline X9B at the National Synchrotron Light Source, Brookhaven National Laboratories (Upton, NY). Before synchrotron data collection, crystals were soaked in mother liquor containing 5% glycerol for 5–10 min and flash-cooled to 100 K by using an Oxford low-temperature system. These data were recorded on a MAR345 image plate by using a wavelength of 0.98 Å and processed with the HKL suite (Otwinowski and Minor, 1997), resulting in an Rmerge of 6.9%. The large unit cell dimensions and relatively small size of the protein suggested that the asymmetric unit contained 6–10 copies of the ADFH domain. The heavy atom constellation was solved by isomorphous difference Patterson synthesis and heavy atom parameters were refined with MLPHARE (Otwinowski, 1991). Nonisomorphism limited the useful phase information to 6.0-Å resolution (figure of merit of 0.51), which precluded direct model building. The availability of a distant model (yeast cofilin with only 20% sequence identity) allowed for successful structure solution but required the implementation of a novel six-dimensional search using the phased translation function (http://russel.bioc.aecom.yu.edu/server/NYSGRC.html). The final model contained nine molecules in the asymmetric unit and was refined with XPLOR (Brünger, 1992) and CNS (Brünger et al., 1998) to an Rcryst of 0.213 and Rfree of 0.259 for all data between 2.0- and 2.1-Å resolution. The two N-terminal residues are not represented by electron density in any of the independent molecules. Coordinates have been deposited in the Protein Data Bank with code 1HQZ
[PDB]
(Berman et al., 2000). Full details of the structure solution and methodology will be presented elsewhere.
Yeast Strains, Cell Growth, and Plasmid Construction
Standard methods were used for DNA manipulations, growth, sporulation, tetrad dissections, and transformation of yeast (Rose et al., 1989). For details on yeast strains and plasmids, see supplementary data (Tables SI and SII, respectively). A PCR-based site-directed mutagenesis approach was used to generate Abp1 mutants. The integration plasmid pABP1+ (HIS3-marked) (Lila and Drubin, 1997) was used as the PCR template to produce Lys21(AAG) to Ala(GCG) and Arg24(AGA) to Ala(GCA) for abp1-1, Lys80(AAG) to Ala(GCG) for abp1-2, Lys94(AAG) to Ala(GCG) and Arg96(AGG) to Ala(GCG) for abp1-3, Asp122(GAC) to Ala(GCC) and Asp125(GAT) to Ala(GCT) for abp1-4, and Lys134(AAA) to Ala(GCA) for abp1-5. Each allele introduces a diagnostic restriction site, and all open reading frames were sequenced.
Plasmids for the galactose-inducible overexpression of wild-type and mutant Abp1 proteins in yeast were constructed by the "gap repair" homologous recombination method. The coding sequences of wild-type and mutagenized pABP1+ were amplified by PCR by using primers D (GTTAATATACCTCTATACTTTAACGTCAAGGAGAAAAAACTATAGGATCCA) and E (TTAATTAACCCGGGAGATCTACCTTGAAAATACAAATTTTCCGCGGCCGCCTAGTTGCCCAAAGACACAT) and then cotransformed with BamHI- and NotI-digested pRS426-GAL (URA3 marked) into an abp1
strain. The repaired plasmids pGalABP1, pGalABP1-1, pGalABP1-2, pGalABP1-3, pGalABP1-4, and pGalABP1-5 were rescued from yeast and amplified in Escherichia coli. The coding sequence of the abp1ADFH allele was amplified by PCR from pABP1+ by using primers E and F (GCGCGGGATCCATGATTCAGACTTCCTCCAAGC). The PCR product was subcloned into the NotI and BamHI sites of pRS426-GAL to generate pGalABP1ADFH. The open reading frames of all plasmids were sequenced.
HIS3 marked abp1 alleles were integrated into the haploid yeast strain BGY022 by methods described previously (Wertman et al., 1992). The insertion cassettes were obtained from wild-type and mutagenized integration plasmids (above) by digestion with EcoRI and transformed into BGY022 (abp1::LEU2). Transformants were selected for growth on His– media and loss of ability to grow on Leu– media. Integration was verified by PCR analysis. Genomic DNA was isolated, and the ABP1 locus was PCR amplified using primers A (CGCGCGGATCCCTAGTTGCCCAAAGACACATA) and B (GCTGCTAGTCCACTCATCT GC), flanking the integration site. The presence of the correct mutation was verified by restriction analysis of PCR products. Plasmids for expressing N-terminal green fluorescent protein (GFP)-fusions of wild-type and mutant Abp1 proteins were constructed by gap repair. Coding sequences of mutagenized pABP1+ were amplified by PCR by using primers A and H (TAACAGCTGCTGGGATTACACATGGCATGGATGAACTATACAAAGGATCCATGGCTTTGGAACCTATTG). For constructing GFP-abp1ADFH, primers A and I (TAACAGCTGCTGGGATTACACATGGCATGGATGAACTATACAAAGGATCCATTCAGACTTCCTCCAAGC) were used. The PCR products were cotransformed with URA3-marked pRB2139 (Doyle and Botstein, 1996) gapped by digestion with BamHI and HindIII and selected on Ura– media. Plasmids were rescued, amplified in E. coli, and verified by restriction analysis.
Protein Purification
Yeast Arp2/3 complex (Goode et al., 2001) and yeast actin (Goode, 2002) was purified as described. Rabbit skeletal muscle actin (unlabeled and pyrene labeled) was purchased from Cytoskeleton (Denver, CO). To isolate wild-type and mutant Abp1 proteins, the protease minus strain BJ2168 (Jones, 2002) was transformed with pGalABP1, pGalABP11, pGalABP12, pGalABP13, pGalABP14, pGalABP15, and pGalABP1ADFH, which overexpress different Abp1 proteins under control of the galactose-inducible promoter. Cultures were grown, induced, harvested, washed, and flash-frozen as described previously (Rodal et al., 2002). Cells were lysed by liquid nitrogen and mechanical shearing in a blender, and a high-speed supernatant in HEK buffer (20 mM HEPES, pH 7.5, 1 mM EDTA, and 50 mM KCl) was generated as described previously (Goode, 2002). The supernatant was applied to a Hi-trap Q fast flow 5-ml anion exchange column (AP Biotech, Piscataway, NJ). Proteins were eluted using a linear gradient of KCl (0.2–0.6M) in HEKG5 buffer (HEK plus 5% glycerol). Peak fractions containing Abp1 (eluting at 0.35–0.45 M KCl) were pooled, diluted to 0.1 M KCl in HEK buffer, and loaded onto a Mono Q (5/5) ion exchange column (Applied Biosystems, Foster City, CA). Proteins were eluted using a linear gradient of KCl (0.25–0.60 M) in HEKG5. Peak fractions (eluting at 0.35–0.45 M KCl) were concentrated to 300 µl and then further purified on a Superose 12 (10/30) gel filtration column (Applied Biosystems) equilibrated in HEKG5. Peak fractions were concentrated to 15–75 µM and flash frozen in liquid nitrogen.
Actin Filament Cosedimentation
Two different experiments were performed to compare the abilities of wild-type and mutant Abp1 proteins to cosediment with F-actin. In the first experiment (Figure 2C), 1 µM Abp1 was added to 3.5 µM preassembled yeast actin in F-buffer (10 mM Tris, pH 7.5, 0.7 mM ATP, 0.2 mM CaCl2, 2 mM MgCl2, 50 mM KCl, and 0.2 mM DTT). Reactions were incubated for 10 min at 25°C and then centrifuged for 20 min at 90,000 rpm in a TLA100 rotor (Beckman Coulter, Fullerton, CA). Pellets and supernatants were fractionated on gels, stained with Coomassie Blue, and the band intensities were quantified by densitometry. In the second experiment (Figure 2, D and E), we measured the dose-responsive binding affinities (Kd) of wild-type and mutant Abp1 proteins for F-actin. Yeast actin was polymerized as described above, stabilized with equimolar phalloidin, diluted (final reaction concentrations 0.05–4 µM F-actin), and incubated with 0.09 µM Abp1. Reactions were centrifuged as described above. Pellets and supernatants were immunoslot-blotted using a vacuum manifold (Schleicher & Schuell, Keene, NH) and probed with anti-Abp1 antibodies. Enhanced chemiluminescence signals were quantified by densitometry using ImageQuant software (Applied Biosystems), and the data were graphed as percentage of Abp1 bound versus F-actin concentration. Binding curves were generated using SigmaPlot (SPSS, Chicago, IL), and Kd values were determined from the following formula:
 |
|
Immunoblotting
Yeast whole cell lysates were prepared as described previously (Kushnirov, 2000). In total, 0.15 OD600 units of cells were loaded per lane on SDS-PAGE, fractionated, and immunoblotted. Blots were probed with a 1:8000 dilution of chicken polyclonal IgY antibody raised against yeast Abp1 (Aves, Tigard, OR). The rabbit polyclonal antibody raised against yeast tubulin (1:20,000) was a gift from Frank Solomon (Massachusetts Institute of Technology, Cambridge, MA).
Microscopy
To assess colocalization of wild-type and mutant GFP-Abp1 proteins with actin in vivo, an abp1 strain (BGY021) was transformed with empty vector (pRS316) or plasmids expressing GFP-Abp1 fusion proteins (pGFPABP1, pGFPADFH, pGFPABP1-1, pGFPABP1-2, pGFPABP1-3, pGFPABP1-4, and pGFPABP1-5). Cells were grown to log phase, fixed for 10 min in 70% ethanol, and stained with rhodamine phalloidin as described previously (Dewar et al., 2002). GFP-Abp1 and rhodamine-actin images were captured using a Zeiss Nikon Eclipse E600 microscope equipped with CoolSNAPfx charge-coupled device camera (Roper Scientific, Dulta, GA) running OpenLab software (Universal Imaging, Downingtown, PA).
Actin Assembly Kinetics
Actin assembly reactions were performed as described previously (Goode et al., 2001) but by using a mixture of 1.5 µM yeast actin and 1.5 µM rabbit skeletal muscle actin (1% pyrene labeled). The mixture was used because yeast Arp2/3 complex has limited capacity to be activated by Abp1 in the presence of pure rabbit muscle actin (Quintero-Monzon, unpublished data). 22.5 µM monomeric actin in G buffer (5 mM Tris-HCl, pH 7.5, 0.2 mM DTT, 0.2 mM ATP, and 0.2 mM CaCl2) was mixed with 10 nM yeast Arp2/3 complex and/or 200 nM wild-type or mutant Abp1, and added immediately to 20x initiation salts (final reaction concentrations 100 mM KCl, 0.5 mM ATP, and 2 mM MgCl2). Actin polymerization was monitored by excitation at 365 nm and emission at 407 nm in a spectrofluorometer (Photon Technology International, Laurenceville, NJ) at 25°C. The rate of polymerization was calculated from the slope of the curves early in the reaction where they are linear.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
-sheet, in which the first four strands (1, 2, 3, and 4) are antiparallel, with the fifth strand (5) running parallel to 4. Each face of the -sheet contacts a pair of 
-helices (1–4), with the helices running in roughly the same direction as the -strands. The nine ADFH domains in the asymmetric unit can be superimposed with root mean square (RMS) deviations of <0.5 Å for 139 -carbons pairs.
The Abp1-ADFH domain structure exhibits the same topology as cofilin (2.1-Å RMS deviation >126 -carbons) and twinfilin (2.3-Å RMS deviation >116 -carbons) (Figure 1C), despite their low sequence homologies (Figure 2A). Detailed structural features of the five-stranded mixed sheet present in Abp1 are shared with cofilin and twinfilin; however, there are also small but significant differences. Cofilin possesses an additional small -strand comprised of residues near the immediate N terminus that runs parallel to the equivalent of Abp1 2. In Abp1 and twinfilin, the orientation of the corresponding chain segment precludes formation of this -strand. The orientations of the four helices differ slightly in the three structures, as do the detailed conformations of some connecting loops, including the equivalents of Abp1 chain segment 24–31 between 1 and 1, segment 50–54 between 2 and 2, segment 71–80 between 3 and 4, and segment 110–113 between 3 and 5. Notably, all insertions and deletions are located within these connecting loops.
The most prominent structural difference among the three ADFH family members is in the conformation of the loop that connects 3 and 4 in Abp1 (Figure 1C). In cofilin, this loop is inclined toward the neighboring helix 4 but has no interactions with it. In Abp1, this loop is shorter than in cofilin, but it assumes approximately the same orientation and again makes no interactions with neighboring chain segments. In both Abp1 and cofilin, this loop is not well defined as judged by the quality of the electron density maps and the associated high-temperature factors. In cofilin, the bending of this loop exposes residues at the N terminus of 4 important for the F-site. Mutations at these residues (cof1-16) are lethal in vivo and disrupt cofilin's ability to interact with F-actin and to induce filament disassembly in vitro (Lappalainen et al., 1997). In twinfilin, this loop adopts a very different orientation (toward the 3) and participates in contacts with the chain segment joining 3 and 5. Residues from this loop in twinfilin are well defined and have low-temperature factors in both independent molecules in the crystal structure (PDB file 1M4J
[PDB]
). In all three structures, this loop has no contacts with symmetry-related molecules.
Defining Actin Binding Surfaces on the Abp1 ADFH Domain
The structural similarity between Abp1 and cofilin suggested that they might interact with actin in a related manner. To test this possibility, we introduced point mutations in the Abp1 ADFH domain that were known to disrupt cofilin's interactions with actin (Lappalainen et al., 1997). The sequences of yeast Abp1 ADFH domain and yeast cofilin are only 20% identical (Moon et al., 1993), yet many of the residues in cofilin that mediate actin binding are conserved in Abp1 (Figure 2A). We targeted those residues for mutagenesis, generating five abp1 alleles, each with one or two charge-to-alanine substitutions. Four alleles (abp1-2, abp1-3, abp1-4, and abp1-5) are analogous to cof1 alleles that are impaired in actin binding and cause overt phenotypes in vivo (Table 1). In addition, we generated abp1-1, which is analogous to cof1-6, an allele of cofilin with no obvious growth phenotype, but whose biochemical activities have never been tested (Lappalainen et al., 1997). Our own biochemical analyses of purified Cof1-6 protein showed that it is modestly impaired in actin filament binding (Rodal, unpublished data); therefore, we investigated whether this mutation in Abp1 has similar effects (Abp1-1). Furthermore, we generated a deletion of the entire ADFH domain (abp1ADFH). Each of the six abp1 alleles and wild-type ABP1 was introduced into a yeast overexpression vector for purification and biochemical analyses, and into an integration vector for genetic analyses (see below).
|
|
|
|
ADFH protein failed to cosediment with actin filaments, consistent with our previous study using GST-Abp1ADFH (Goode et al., 2001). Two mutants, Abp1-2 (Kd > 10 µM) and Abp1-3 (Kd = 8.0 µM), interacted weakly with F-actin compared with wild-type Abp1 (Kd = 0.58 µM). Mutations in these two alleles are located on opposite ends of the ADFH domain crystal structure, corresponding to the G- and F-sites on ADF/cofilins, respectively, and are suggested to interact with two adjacent actin subunits in a filament (Bamburg, 1999). Abp1-1 (Kd = 2.0 µM) and Abp1-5 (Kd = 0.63 µM) exhibited much more subtle defects in F-actin binding, consistent with the modest defects in F-actin binding for analogous cofilin mutants Cof1-22 and Cof1-6. Unexpectedly, Abp1-4 showed relatively normal F-actin binding, whereas the analogous cofilin mutant Cof1-20 is severely impaired in F-actin binding and is lethal in vivo (Lappalainen et al., 1997). These data are summarized in Table 1.
Abp1 Binding to Actin Filaments Is Required for Localization In Vivo
Next, we addressed the requirement of actin binding for Abp1 localization to actin patches in vivo. Actin patches assemble at the cell cortex where membrane and membrane proteins are internalized as vesicles, which eventually fuse with endosomal sorting compartments and shed their actin coats (Kaksonen et al., 2003; Huckaba et al., 2004). Actin patches are comprised of filamentous actin, as demonstrated by their staining with rhodamine phalloidin, which binds specifically to F-actin (Adams and Pringle, 1984). Although purified Abp1 binds directly to F-actin in vitro (Kessels et al., 2000; Goode et al., 2001), it is unknown whether actin binding by Abp1 is required for its localization in vivo. Because Abp1 makes physical associations with a number of other components of actin patches, including Srv2/CAP, Rvs167, Sla1, and Arp2/3 complex (Freeman et al., 1996; Colwill et al., 1999; Cope et al., 1999; Goode et al., 2001; Fazi et al., 2002), its localization might not require direct binding to actin. To test this possibility, we compared the localization of wild-type and mutant Abp1 proteins in yeast cells.
Attempts to obtain reliable immunolocalization data using our Abp1 antibodies were unsuccessful. Therefore, we expressed N-terminally tagged GFP-Abp1 wild-type and mutant fusion proteins from low copy plasmids introduced into an abp1 strain. Immunoblotting of total extracts from these cells demonstrated that wild-type and mutant GFP-Abp1 expression levels were comparable (Figure 3A). GFP-Abp1 localized to cortical actin patches, as expected, whereas GFP-Abp1ADFH was found in the cytoplasm (Figure 3B). This provides the first demonstration in any organism that the ADFH domain is required for Abp1 localization in cells. This also demonstrates that the physical interactions of the proline-rich and SH3 domains in Abp1 with other patch components are insufficient for its localization to actin patches.
Using point mutations, we next tested whether the actin binding activity of the ADFH domain specifically was required for Abp1 localization. Abp1-2, which is strongly impaired in F-actin binding in vitro, was mislocalized to the cytoplasm, albeit not as severely as Abp1ADFH. Three other mutants, Abp1-1, Abp1-4, and Abp1-5, which have minimal or no defect in F-actin binding in vitro, localized normally to actin patches. Abp1-3, which has >10-fold reduced binding affinity for F-actin in vitro, showed only slightly diminished localization to patches. This is consistent with Abp1-3 having less severely impaired actin binding affinity than Abp1-2, and with data showing that abp1-2 is more severely compromised for ABP1 genetic functions (below). These data indicate that mislocalization of Abp1 in vivo arises only from severe disruptions of F-actin binding (Abp1-2 and Abp1ADFH).
The ADFH Domain of Abp1 Is Required for Overlapping Genetic Functions with Sac6 and Sla2, but Not Sla1
ABP1 has shared or overlapping genetic functions with three other genes, SAC6, SLA2, and SLA1. The SH3 domain of Abp1 is required for each of these shared functions (Lila and Drubin, 1997). However, it is unknown what other domains and activities of Abp1 are required. To address this, we used a plasmid shuffle assay to test the importance of the Abp1 ADFH and SH3 domains in rescuing lethality in double mutant backgrounds: abp1 sac6-102, abp1 sla1-7, and abp1 sla2-5 (Figure 4A). The SH3 domain was required for complementation of growth in all three backgrounds, consistent with the study mentioned above. In contrast, the ADFH domain was required only in the abp1 sac6-102 and abp1 sla2-5 backgrounds. Thus, actin binding is required for Abp1 functions that overlap with SAC6 and SLA2, but not SLA1.
To test specifically whether actin binding is required for the functions shared by ABP1 and SAC6, we combined our ADFH domain mutant abp1 alleles with a sac6 mutation and examined genetic interactions. For these analyses, we integrated each of the six mutant abp1 alleles at the ABP1 locus. Immunoblotting of total extracts showed that wild-type and mutant Abp1 proteins were expressed at equivalent levels in these strains (Figure 4B). Each abp1 haploid strain was crossed to a sac6 strain. The resulting diploids were sporulated, dissected, and their haploid progeny were serially diluted and compared for defects in cell growth at different temperatures (25, 30, 34, and 37°C). A large number of tetrads were analyzed (40–70 per cross) to maximize the chance of scoring intermediate phenotypes (Figure 4C). Consistent with the plasmid shuffle data described above, double mutant abp1ADFH sac6 strains were inviable. The growth of abp1-2 sac6 strains also was highly compromised, consistent with Abp1-2 having strong defects in F-actin binding and localization to actin patches. The growth phenotypes of abp1-4 sac6, abp1-1 sac6, and abp1-3 sac6 strains were less severe and correlated with the degrees to which they are impaired in F-actin binding (Abp1-3 > Abp1-1 > Abp1-4). These data demonstrate that F-actin binding is required for Abp1 functions in vivo that overlap with Sac6. Unexpectedly, abp1-5 sac6 strains were severely compromised in cell growth, which did not correlate with the normal F-actin binding and localization of Abp1-5. This suggested that the surface mutated in Abp1-5 has a function other than actin binding (see below).
To test the requirement of F-actin binding by Abp1 for the ABP1 SLA2 shared function, we crossed abp1 mutant strains to the sla2376-440 strain, which is synthetically lethal with abp1 (Wesp et al., 1997; our unpublished data). Whereas a full deletion of ABP1 was synthetic lethal with sla2376-440, the abp1ADFH mutant was only synthetic temperature sensitive at 37°C (our unpublished data). Thus, the ADFH domain is only partially required in the sla2 background, in contrast to being fully required for function in the sac6 background. None of the other abp1 alleles showed the same synthetic temperature sensitivity in the sla2 background as Abp1ADFH, but it was difficult to reliably score partial synthetic defects.
The lethal Abp1 Overexpression Phenotype Requires Actin Binding and Arp2/3 Complex Interactions
Expression of wild-type Abp1 on a high copy plasmid causes about fivefold overexpression compared with endogenous Abp1, leading to impaired cell growth at higher temperatures, loss of actin polarization, and enlargement of mother cells at the expense of bud growth (Drubin et al., 1988). We tested the abp1 alleles in this overexpression assay to determine whether F-actin binding is required for the overexpression phenotype. To further sensitize the assay, we overexpressed Abp1 under control of the GAL1/10 galactose-inducible promoter, which resulted in higher levels of overexpression, and lethality at all temperatures (Figure 5A). All constructs were overexpressed to comparably high levels relative to endogenous Abp1 (Figure 5B).
Like wild-type Abp1, Abp1-1 and Abp1-4, which have normal or only modestly impaired F-actin binding in vitro and normal localization to actin patches in vivo, were lethal when overexpressed (Figure 5A). In contrast, Abp1ADFH and Abp1-2 showed no obvious growth defects when overexpressed. This indicates that direct binding to F-actin is required for the Abp1 overexpression phenotype. Consistent with the observed intermediate defects in F-actin binding for Abp1-3, overexpression of this mutant caused an intermediate growth phenotype. Abp1-5, which binds normally to F-actin and localizes normally in vivo, failed to cause a growth phenotype when overexpressed. Thus, the Abp1 overexpression phenotype depends not only on F-actin binding but also on another function of the ADFH domain. These data are consistent with the genetic interactions between abp1-5 and sac6 (Figure 4), which suggest that the surface mutated by Abp1-5 performs a nonactin-binding function of physiological importance.
ADFH Domain Requirements for Abp1 Stimulation of Arp2/3 Complex In Vitro
To better understand the unexpected genetic results for Abp1-5 (Figures 4 and 5), we considered what biochemical activities might be performed by the ADFH domain other than actin binding. The only other activity of Abp1 that is known to depend on its ADFH domain is activation of the Arp2/3 complex (Goode et al., 2001). Therefore, we compared the ability of purified wild-type and mutant Abp1 proteins for their ability to activate the Arp2/3 complex (Figure 6). Wild-type Abp1 and Abp1-4 activated the Arp2/3 complex with similar efficiencies. Abp1-1 partially activated the Arp2/3 complex. Abp1-2, Abp1-3, and Abp1ADFH had no significant ability to activate Arp2/3 complex. Thus, Abp1 activation of Arp2/3 complex is highly sensitive to partial defects in F-actin binding. For these mutants, the ability to stimulate the Arp2/3 complex correlated closely with ability to bind F-actin (Abp1 Abp1-4 > Abp1-1 > Abp1-3 > Abp1-2 > Abp1ADFH). From these data, we conclude that F-actin binding is required for Abp1 activation of Arp2/3 complex, which supports the proposed filament recruitment mechanism.
The data in Figure 6 reveal an additional role for the ADFH domain in Arp2/3 complex regulation. Although Abp1-5 binds normally to F-actin, it shows no ability to stimulate Arp2/3 complex activity. This means that Abp1-5 uncouples the F-actin binding function of the ADFH domain from a distinct role in Arp2/3 complex activation. We tested purified ADFH domain for direct physical interactions with Arp2/3 complex, but could detect none (Quintero-Monzon, unpublished data). Although we cannot rule out the possibility of direct interactions, we suspect a different mechanism (see Discussion). Regardless of the precise mechanism, these data define a new and unexpected functional site on the ADFH domain that is important for Arp2/3 complex activation and for ABP1 function in vivo, as demonstrated by the genetic interactions between abp1-5 and sac6 (Figure 4) and the nonlethal overexpression phenotype of abp1-5 (Figure 5).
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
Differential Actin-binding Interactions by ADFH Family Members
The ADFH domain is a versatile actin-binding module found in three conserved families of actin binding proteins, ADF/cofilins, twinfilin, and Abp1 (Lappalainen et al., 1998). A side-by-side comparison of their structures (Figure 1C) shows that Abp1 most closely resembles ADF/cofilin. This is consistent with these two proteins binding to F-actin, whereas twinfilin does not. The ability of ADF/cofilin to interact with F-actin requires its G- and F-sites (Figure 7A). The G-site is predicted to interact with subdomains 1 and 3 of one actin subunit, whereas the F-site interacts with subdomains 1 and 2 of the next actin subunit in a filament (Ono, 2003). Together, these two interactions can weaken the lateral and longitudinal contacts between subunits and induce filament severing and rapid dissociation of subunits from filament ends (Bamburg et al., 1999; Galkin et al., 2001, 2003). Now with crystal structures and mutagenesis analyses reported for each of the three major ADFH family members, a close examination of their structures and functions can be made to identify determinants of their specific activities.
What determines whether an ADFH domain binds to G-actin, F-actin, or both? Actin monomer binding by ADF/cofilins and twinfilin requires key charged surfaces on both the G-site and helix 3 (Figure 7A) (Lappalainen et al., 1997; Ojala et al., 2001; Guan et al., 2002). Twinfilin and ADF/cofilin are functionally conserved at both of these surfaces, whereas Abp1 sequence diverges at helix 3 (the two charged residues critical for G-actin binding in ADF/cofiln and twinfilin are not conserved in the Abp1 sequence; Figure 2). This likely explains why Abp1 does not bind G-actin. On the other hand, F-actin binding by ADF/cofilins requires its G- and F-sites, but not helix 3. Twinfilin diverges structurally from ADF/cofilins at the F-site (Paavilainen et al., 2002), explaining why it does not bind to F-actin. The connecting loop between 3 and 4 in twinfilin adopts a conformation that positions it away from 4, disrupting the F-site. The 3-4 connecting loop in Abp1 is similar to ADF/cofilins, which stresses the importance of this structure in the F-site in predicting F-actin binding activity among ADFH family members. Consistent with this prediction, the recently solved structure of another ADFH family member, coactosin, resembles Abp1 and ADF/cofilins at the F-site (Hellman et al., 2004; Li et al., 2004; Liu et al., 2004), and this protein binds specifically to F-actin, although its activities have not yet been dissected by mutagenesis (de Hostos et al., 1993).
Why do ADF/cofilins promote actin filament severing and disassembly, whereas Abp1 does not? Two key differences in their F-actin binding surfaces suggest a possible explanation. First, ADF/cofilins have a more extensive G-site (site 1 in Abp1) (Figure 7A). The ADF/cofilin G-site includes two conserved charged residues, D123 and E126 (yeast cofilin nomenclature), that are critical for actin filament binding and depolymerization in vitro and in vivo (Lappalainen et al., 1997). The analogous residues in Abp1 (D122 and D125, mutated in Abp1-4) make no contribution to F-actin binding or Abp1 localization and function in vivo. This result was unexpected because the two residues are well conserved, unlike the charged residues on helix 3. However, adjacent to this site in ADF/cofilins is a patch of exposed hydrophobic residues, which is lacking in Abp1 (Figure 7B). Although charged residues at this site clearly are required for F-actin binding and disassembly in ADF/cofilins (as shown by the severe defects of Cof1-20), they may not be sufficient and may require the neighboring hydrophobic surfaces for this activity. Consistent with this possibility, the predicted target of the ADF/cofilin G-site on actin (subdomains I and III) is a hydrophobic-rich surface.
The second place where the F-actin binding surfaces of Abp1 and ADF/cofilins diverge is in helix 4 (Figure 1). Three charged residues mutated in cof1-22 contribute to F-actin disassembly activity in vitro and in vivo (Lappalainen et al., 1997; Lappalainen and Drubin, 1997), but Abp1 diverges at this site. Only one of the three charged residues is conserved in Abp1, and it makes no apparent contribution to F-actin binding affinity (Abp1-5). As described above for the cof1-20 site, we note that adjacent to the cof1-22 site in ADF/cofilins is a patch of exposed hydrophobic residues, lacking in Abp1 (Figure 7B). Thus, combined charged hydrophobic surface interactions at these sites may be important for specifying filament severing and depolymerization by ADF/cofilins. We also propose that helix 4 acts as a functional determinant in Abp1, because Abp1-5 abolishes Arp2/3 complex activation (Table 1), as discussed below.
Mechanism of Arp2/3 Complex Activation by Abp1
Yeast Abp1 directly associates with the Arp2/3 complex and stimulates its actin nucleation activity (Goode et al., 2001). Abp1 and cortactin have been categorized as class II NPFs, because they bind to actin filaments but not monomers (Welch and Mullins, 2002). It is hypothesized that class II NPFs stimulate Arp2/3 complex nucleation activity by strengthening its association with the sides of preexisting actin filaments (Goode et al., 2001; Uruno et al., 2001; Weaver et al., 2001). Here, we used point mutants in the actin-binding domain of Abp1 to demonstrate that F-actin binding is required for its nucleation-promoting activity, which strongly supports the filament recruitment hypothesis. In addition, we uncovered an unexpected second function of the ADFH domain in activating Arp2/3 complex that is independent of F-actin binding affinity. This function maps to the C-terminal portion of helix 4 and is disrupted by a single point mutation (K134A) in abp1-5. Abp1-5 has no apparent defect in F-actin binding affinity, yet abolishes Arp2/3 complex activation and the lethality caused by ABP1 overexpression in vivo. Thus, this mutation uncouples the F-actin binding and Arp2/3 activation functions of the Abp1 ADFH domain. Furthermore, it suggests that the lethality in cells caused by Abp1 overexpression involves misregulation of Arp2/3 complex (Figure 5) and that normal functional interactions with the Arp2/3 complex are required for the Abp1 cellular function that is shared with Sac6 (Figure 4). This latter observation is consistent with previous results showing that the Arp2/3-activating acidic motifs of Abp1 are required in the sac6 background (Goode et al., 2001).
We do not yet understand how Abp1 helix 4 participates in Arp2/3 activation. We considered that this helix might directly interact with Arp2/3 complex. This seemed reasonable given the similarity of Arp2 and Arp3 subunits to actin and previous reports that ADF/cofilin binds to Arp2/3 complex with low affinity (Blanchoin et al., 2000). However, we could find no evidence of a direct interaction in binding assays by using purified Abp1 ADFH domain and Arp2/3 complex (Quintero-Monzon, unpublished data). Thus, although we cannot rule out this model, our data do not support it. Many other mechanisms are possible. For instance, this surface of the ADFH domain could make productive intramolecular interactions with a different part of Abp1 involved in Arp2/3 activation, such as the acidic motifs (Figure 1B). Alternatively, this surface could interact with and change the conformation of F-actin to promote Arp2/3 complex activation without changing Abp1 affinity for F-actin. Further investigation will be required to determine precisely how helix 4 contributes to Arp2/3 complex activation.
Abp1 Functions as a Modular Adapter In Vivo
Over a decade ago, it was demonstrated in yeast that an abp1 mutation is synthetic lethal with loss of function mutants in three other genes: sac6, sla1, and sla2 (Holtzman et al., 1993). However, until now it has been unclear whether these genetic interactions resulted from loss of the same or distinct cellular functions of Abp1. Here, we have uncoupled ABP1 functions, demonstrating that Abp1 cellular functions rely on different combinations of domains and activities. We focused primarily on the Abp1–actin interaction and showed that it is differentially required in vivo, being strictly required for cell viability in the sac6 background, only partially required in the sla2 background, and not required at all in the sla1 background. The precise functions of Abp1 that overlap with Sac6, Sla1, and Sla2 remain to be defined. Sac6 (fimbrin) bundles and stabilizes actin filaments (Adams et al., 1991; Goodman et al., 2003), whereas Abp1 alone has no direct effects on actin filament stability or organization in vitro (Kessels et al., 2000; Goode et al., 2001). However, it has been hypothesized that Abp1 may be similar to cortactin and stabilize Arp2/3-nucleated branched actin filaments (Olazabal and Machesky, 2001; Welch and Mullins, 2002). Thus, in the capacity of stabilizing branched actin filaments Abp1 function may be required in the absence of Sac6. This model also is consistent with the observed synthetic genetic interactions between abp1-5 and sac6 mutations (Figure 4C).
Although actin binding is differentially required for Abp1 functions in vivo, the SH3 domain of Abp1 is required for all of its genetic functions (Lila and Drubin, 1997; Figure 4). This suggests that Abp1 functions as a modular adapter in vivo. For functions shared with Sac6, it links F-actin, Arp2/3 complex, and SH3 domain ligands. For functions shared with Sla1, it bridges a different set of ligands that does not include F-actin. Defining more precisely each biological role of Abp1 will require identifying the ligands required for each of its functions. Many Abp1 ligands are known. It is strongly established that Abp1 binds through its proline-rich and SH3 domains to Rvs167/amphiphysin (Colwill et al., 1999), Sla1 (Warren et al., 2002), Srv2/cyclase-associated protein (Freeman et al., 1996; Lila and Drubin, 1997), and the actin-regulating kinases Ark1 and Prk1 (Cope et al., 1999). In addition, many other SH3 domain- and proline-rich domain-containing proteins associate with Abp1 in genome-wide screens and proteomic studies, including App1, Cla4, Hua2, Myo5, Scp1, Sjl2, Yor284, and Ysc84 (Uetz et al., 2000; Ito et al., 2001; Fazi et al., 2002; Ho et al., 2002; Tong et al., 2002). The complexity and overlapping nature of this web of interactions lends support to our central conclusion that Abp1 is a versatile adapter molecule and that different combinations of ligands interact with Abp1 at different times in the cell and/or with different subpopulations of Abp1.
|
ACKNOWLEDGMENTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
|
Footnotes |
|---|
The online version of this article contains supplemental material at MBC Online (http://www.molbiolcell.org). 
Address correspondence to: Bruce L. Goode (goode{at}brandeis.edu).
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
Adams, A. E., Botstein, D., and Drubin, D. G. ((1991). ). Requirement of yeast fimbrin for actin organization and morphogenesis in vivo. Nature 354, , 404–408.[CrossRef][Medline]
Adams, A. E., Cooper, J. A., and Drubin, D. G. ((1993). ). Unexpected combinations of null mutations in genes encoding the actin cytoskeleton are lethal in yeast. Mol. Biol. Cell 4, , 459–468.[Abstract]
Bamburg, J. R. ((1999). ). Proteins of the ADF/cofilin family: essential regulators of actin dynamics. Annu. Rev. Cell Dev. Biol. 15, , 185–230.[CrossRef][Medline]
Bamburg, J. R., McGough, A., and Ono, S. ((1999). ). Putting a new twist on actin: ADF/cofilins modulate actin dynamics. Trends Cell Biol. 9, , 364–370.[CrossRef][Medline]
Berman, H. M., Westbrook, J., Feng, Z., Gilliland, G., Bhat, T. N., Weissig, H., Shindyalov, I. N., and Bourne, P. E. ((2000). ). The protein data bank. Nucleic Acids Res. 28, , 235–242.
Blanchoin, L., and Pollard, T. D. ((1999). ). Mechanism of interaction of Acanthamoeba actophorin (ADF/cofilin) with actin filaments. J. Biol. Chem. 274, , 15538–15546.
Blanchoin, L., Pollard, T. D., and Mullins, R. D. ((2000). ). Interactions of ADF/cofilin, Arp2/3 complex, capping protein and profilin in reomodeling of branched actin filament networks. Curr. Biol. 10, , 1273–1282.[CrossRef][Medline]
Bowman, G. D., Nodelman, I. M., Hong, Y., Chua, N. H., Lindberg, U., and Schutt, C. E. ((2000). ). A comparative structural analysis of the ADF/cofilin family. Proteins 41, , 374–384.[CrossRef][Medline]
Brünger, A. T. ((1992). ). X-PLOR Version 3.1. A System for X-ray Crystallography and NMR, New Haven, CT: Yale University Press.
Brünger, A. T., et al. ((1998). ). Crystallography & NMR system: a new software suite for macromolecular structure determination. Acta Crystallogr. 54, , 905–921.[CrossRef]
Carlier, M. F., Laurent, V., Santolini, J., Melki, R., Didry, D., Xia, G. X., Hong, Y., Chua, N. H., and Pantaloni, D. ((
for the three major ADFH domain family members: Abp1, ADF/cofilin, and twinfilin. Lengths in amino acid residues are given for S. cerevisiae proteins. 
), Abp1-5 (
), Abp1-2 (
), Abp1
). (E) Dissociation constants (Kd) were derived from the binding data in D (see Materials and Methods). Values are the average of two independent experiments.
