- Email: [email protected]
Structural Basis for Vertebrate Filamin Dimerization
Structure, Vol. 13, 111–119, January, 2005, ©2005 Elsevier Ltd All rights reserved. DOI 10.1016/j.str.2004.10.014
Structural Basis for Vertebrate Filamin Dimerization Regina Pudas,1 Tiila-Riikka Kiema,1 P. Jonathan G. Butler,2 Murray Stewart,2 and Jari Ylänne1,* 1 Biocenter Oulu and Department of Biochemistry University of Oulu Finland 2 MRC Laboratory on Molecular Biology Cambridge United Kingdom
Summary Filamins are essential in cell motility and many developmental processes. They are large actin crosslinking proteins that contain actin binding domains in their N termini and a long rod region constructed from 24 tandem Ig domains. Dimerization is crucial for the actin crosslinking function of filamins and requires the most C-terminal Ig domain. We describe here the crystal structure of this 24th Ig domain (Ig24) of human filamin C and show how it mediates dimerization. The dimer interface is novel and quite different to that seen in the Dictyostelium discoideum filamin analog. The sequence signature of the dimerization interface suggests that the C-terminal domains of all vertebrate filamins share the same dimerization mechanism. Furthermore, we show that point mutations in the dimerization interface disrupt the dimer and that the dissociation constant for recombinant Ig24 is in the micromolar range. Introduction Filamins are key integrators of cell mechanics and signaling that combine actin binding and scaffolding functions. The actin crosslinking activity of filamins is important in modulating the properties of the actin cytoskeleton and, through this function, influencing cell shape as well as being a key component of the cell locomotion machinery. The arrangement of actin filaments and their dynamics are regulated by a range of actin binding proteins that orchestrate the polymerization and depolymerization of filaments together with their linking into bundles and their attachment to membranes (Ayscough, 1998; McGough, 1998; Small et al., 2002). The crosslinking proteins that bundle filaments typically contain two actin binding domains separated by a variable linker (Puius et al., 1998; Van Troys et al., 1999). The linkers determine the geometry of actin crosslinking and can also function as binding sites for downstream and upstream signaling components (Small et al., 2002). Filamins are an important class of actin crosslinking proteins that make crucial contributions to the stability of webs of actin filaments at the cell periphery and link them to the plasma membrane, as well as serving as key scaffolds for key factors that integ*Correspondence: [email protected]
rate signaling and cell mechanics (Stossel et al., 2001). Filamins (also called ABP280) have N-terminal actin binding domains followed by a rod region constructed from tandem sequence repeats that are predicted to have immunoglobulin-like (Ig) folds. The rod region of the Dictyostelium filamin analog (also called ABP120, ddFLN, or gelation factor) has six Ig domains, whereas there are 24 in vertebrates (Stossel et al., 2001; van der Flier and Sonnenberg, 2001). Dimerization is crucial for the actin-crosslinking function of filamins and requires the most C-terminal Ig domain (see Gorlin et al., 1990). Mammals have three filamin genes: filamins A and B are expressed ubiquitously, whereas filamin C is muscle specific (Stossel et al., 2001; van der Flier and Sonnenberg, 2001). Filamin A is encoded in the X chromosome, and mutations cause periventricular heterotopia and several developmental malformations in both females and males (Fox et al., 1998; Kakita et al., 2002; Sheen et al., 2001; Zenker et al., 2004). Cultured cells lacking filamin A show defects in plasma membrane stability and in cell migration (Cunningham et al., 1992). In muscle, only 3% of filamin C is associated with the sarcolemmal membrane (Thompson et al., 2000) with the remainder located primarily at the Z-discs. Filamin C connects plasma membrane and myofibrils by binding to γ- and δ-sarcoglycans (Thompson et al., 2000) and to α-actinin 2 via FATZ and myotilin (Faulkner et al., 2001). By bringing together transmembrane proteins such as integrins (Loo et al., 1998), γ- and δ-sarcoglycans (Thompson et al., 2000), intracellular signaling proteins like small GTPases (Ohta et al., 1999), and actin filaments, filamins provide a link between plasma membrane signaling and actin cytoskeleton. The molecular architecture of filamins has been controversial. Electron micrographs of human filamin A suggest that in the rod region the two chains lie parallel to one another (Gorlin et al., 1990; Stossel et al., 2001), but the crystal structure of the dimerization region of the Dictyostelium filamin analog instead supported an antiparallel arrangement (McCoy et al., 1999; Popowicz et al., 2004). Here, we present a crystal structure of the dimerization domain of human filamin C that has a novel dimer interface which is quite different to that observed in the Dictyostelium filamin analog. We show that mutations in this interface prevent dimerization of the domain. We suggest a model of how the dimerization domain can be fitted to filamin molecular architecture and, based on ultracentrifugation studies, raise the possibility that filamin dimerization may be an important control point in vivo. Results and Discussion Crystal Structure of 24th Ig Domain of Filamin C We obtained P6122 crystals that diffracted to 1.43 Å of Ig domain 24 of human filamin C, which mediates dimerization. The structure was determined using single wavelength anomalous diffraction (SAD) of Se-Metlabeled protein crystals and refined to R = 18.6% and
Structure 112
Table 1. Crystallographic Data Data Collection Statistics
Native
SeMet Peak
Wavelength (Å) Beamline Crystal to detector distance (mm) Space group Unit cell parameters (Å) a=b c Resolution range (Å) R (%)a I/sigma (I) Completeness (%) Redundancy
0.8020 X13 (EMBL, DESY) 120.0 P6(1)22
0.9778 BW7A (EMBL, DESY) 150.0 P6(1)22
48.099 117.485 20–1.43 (1.53–1.43)b 3.7 (22.5) 41.59 (13.29) 99.8 (100) 15.8 (16.1)
48.800 117.100 20–2.0 (2.1–2.0) 2.8 (8.6) 42.96 (19.86) 99.8 (99.9) 7.6 (7.5)
Phasing Statistics Number of SeMet sites Phasing power FOM before density modification FOM after density modification R (%) after autobuilding ARP/wARP
5 4.25 (centric), 3.90 (acentric) 0.4577 0.8697 28.9
Final Model Statistics (Refinement) Resolution limits No. of reflections used for refinement Number of reflections in the test set R (%) Rfree (%) Rmsd bonds (Å)/angles (°) Residues in Ramachandran Residues in most favored regions (%) Residues in additional allowed regions (%) Residues in generously allowed and in disallowed regions (%) Protein atoms Water atoms Average B factor (Å2) B factor from Wilson plot (Å2) a b
17.73–1.43 14,038 1,560 18.58 20.5 0.022/1.431 96 92.3 7.7 0 731 73 13.273 19.177
R factor = (SUM(ABS(I(h, i) − I (h))))/SUM(I(h, i))). Values for the last resolution shell are given in parentheses.
Rfree = 20.5% (Table 1). In these crystals, domain 24 had the expected Ig-like fold (Figure 1A), with strands A, B, E, and D forming a four-stranded antiparallel β sheet and strands C, F, and G forming a three-stranded sheet (Figure 2A). In the SCOP database (Murzin et al., 1995), the filamin Ig-like domain forms a distinct family in the superfamily of E-set domains, in the Ig-like β-sandwich fold. Before the current structure, three other Ig domain structures have been available for this family (Fucini et al., 1997; McCoy et al., 1999; Popowicz et al., 2004). Of these, our structure of human filamin C domain 24 resembles most closely Dictyostelium filamin domains 5 and 4, whereas the Dictyostelium domain 6, which mediates dimerization in this molecule, is clearly distinct. The Dictyostelium filamin domain 5, which is not involved in dimerization (McCoy et al., 1999), has the highest similarity, with root-mean-square (rms) deviation of 1.6 Å of 69 superimposed C-α atoms (Figure 1B), whereas the rms deviation with the corresponding atoms of the average NMR structure of the Dictyostelium domain 4 (Fucini et al., 1997) is 2.2 Å (Figure 1C). The human filamin Ig domain 24 and Dictyostelium domains 4 and 5 share the same overall topology (Figures 2A–2C), although in filamin C domain 24 strand D contains six residues, whereas in the Dictyostelium do-
mains 5 and 4 the strands D are four and three residues long, respectively. The loop connecting strands A and B makes parallel β strand-like hydrogen bonds to the end of the strand G of the opposite β sheet in all three domains. Also, the loop between C and D turns close to the end of the strand C before continuing to strand D. Only in the Dictyostelium domain 5 can these two connections be classified as short β strands (A# and C#) (Figure 2B). As expected, the loops vary more than the β strands, with the loop between β strands B and C being the most divergent (Figures 1B and 1C). In the sequence alignment, this loop is longer in the Dictyostelium domains (Figure 6A) and it also has the highest B-factor values in the X-ray structures (this study and McCoy et al., 1999) or highest variability among the alternative NMR assemblies (Dictyostelium domain 4) (Fucini et al., 1997). The structure of Dictyostelium filamin domain 6, which mediates dimerization in this molecule, is much less similar to human filamin C domain 24 than the other Dictyostelium domains discussed above. When comparing the same 69 residues, the rms deviation between filamin C domain 24 and Dictyostelium domain 6 was 8.3 Å. This large difference was mainly a reflection of differences in topology (Figures 1D and 2D). The Dictyostelium domain 6 does not have strands A and A# and has an additional strand H.
Dimerization of Human Filamin C 113
Figure 1. Structure and Comparison of the Single Domains (A) Ribbon representation of the human filamin C domain 24 structure. N and C termini are indicated. (B) Superimposition of filamin C domain 24 (blue) with Dictyostelium filamin domain 5 (yellow). (C) Superimposition with Dictyostelium domain 4 (green). (D) Superimposition with Dictyostelium domain 6 (magenta). Note that the N-terminal strand of filamin C domain 24 and the C-terminal strand of Dictyostelium domain 6 could not be superimposed. The PDB ID code for the Dictyostelium filamin domain 4 is 1KSR and for domains 5 and 6 is 1QFH.
Both of these features are crucial for the way in which the Dictyostelium protein dimerizes (McCoy et al., 1999), and so their absence in filamin domain 24 accounts for its fundamentally different dimerization architecture. Filamin C Has a Different Dimerization Interface from that in Dictyostelium Filamin Vertebrate filamin Ig domain 24 is necessary for dimerization and itself dimerizes in solution (Himmel et al., 2003). Although the asymmetric unit of our crystal contained only a single chain, an extensive interface involving strands C and D was formed between adjacent chains related by a crystallographic two-fold axis (Figure 3), which we identified as the putative dimerization interface. The interface buried 1109 Å2 (19%) of the accessible surface area of the monomer and contained 45% polar atoms. There are six hydrogen bonds in the interface connecting the D strands so that the fourstranded β sheets of each chain form a single eightstranded antiparallel β sheet (Figures 3B and 4B). In addition, Met2667 and Met2669 form a putative hydrophobic packing with Gly2671 of the adjacent chain (Figure 4A). The dimerization interface formed with filamin C domain 24 is strikingly different from that seen in Dictyostelium filamin domain 6 and involves the opposite edge of the β sandwich. In Dictyostelium domain 6, the interface is formed primarily by strands B and G, whereas in filamin domain 24 it is formed by strands C and D (Figures 2A and 2D). Crucially, this interface in Dictyostelium domain 6 can only form because of the absence of strands A and A# in this domain. The presence of this strand in filamin domain 24 therefore effectively prevents the chains dimerizing in the same way as the Dictyostelium analog. We engineered mutants to confirm that the putative dimer interface seen in the crystals was that involved in dimerization in solution. In one, Met2669 in strand
C was mutated to aspartic acid to introduce negative charge in the key hydrophobic interaction observed in the interface in the crystal. In another, strand C was replaced with the corresponding strand from human filamin A Ig domain 20, which does not dimerize in solution (data not shown). Both mutants were expressed and purified as the wild-type filamin C domain 24. Their CD spectra showed a deep minimum at 216 nm, characteristic of a β structure (Figure 5A), although in the 190–210 nm range the CD spectra of the mutants differed from wild-type, possibly because of less ordered domain-domain interactions. By gel filtration, wild-type filamin C domain 24 eluted as a dimer, whereas both mutants eluted later with retention times expected for a single chain (Figure 5B). The disruption of filamin C domain 24 dimerization by these mutations gives powerful support to the hypothesis that the putative interface identified in the crystal structure genuinely represents the dimerization interface present in solution. The Dimer Interface Is Common to All Vertebrate Filamins Although vertebrate filamin C-terminal domains are very similar in amino acid sequence alignments, isoforms A, B, and C clearly form distinct groups (Figure 6B). Because several functionally important interactions have been mapped to these domains (Bellanger et al., 2000; Loo et al., 1998; Ohta et al., 1999; Ott et al., 1998; Thompson et al., 2000), we compared the surface residues of different filamins. We classified substitutions between vertebrate filamin 24 domain sequences as either A, B, or C specific, or alternatively, there are those that are different in all three. When these substitutions were mapped on the accessible surface of human filamin C domain 24, they mostly located on the nondimer-forming edge of the β sandwich (Figure 6C). This fits well with the proposal that the edges of the sandwich are the most common interaction sites (Radaev et al., 2001).
Structure 114
Figure 2. Topology Diagrams (A) Human filamin C domain 24. (B) Dictyostelium filamin domain 5. (C) Dictyostelium filamin domain 4. (D) Dictyostelium filamin domain 6. Beta strands are shown as arrows, and 310 helices are shown as cylinders. In (A) and (D), strands forming the dimerization interface are colored gray.
The similarity of human filamin Ig-like sequence repeats with those of Dictyostelium domains 4 and 5 extends to all the human filamin sequences (van der Flier and Sonnenberg, 2001). Alignments also predict that the Dictyostelium domain 6 is a more distant member of the family (van der Flier and Sonnenberg, 2001), con-
sistent with our structural and biochemical studies that show that the dimerization interface in human filamin C Ig domain 24 is different from that of Dictyostelium domain 6. When the sequences of the C-terminal domains of vertebrate filamins are aligned, a high level of overall conservation is clearly present, particularly in the hyFigure 3. Crystallographic Dimer of Filamin C Domain 24 Views rotated by 180° about the vertical. Strands of the monomers are colored, respectively, blue and green. Strands C and D involved in dimer contact are indicated. The extended antiparallel β sheet that forms on dimerization is shown in (B). The two-fold axis is marked. The extra residues at the N terminus that are derived from the vector are indicated in blue.
Dimerization of Human Filamin C 115
Figure 4. Filamin Domain 24 Dimerization Interface (A) The hydrophobic stacking of strand C in the dimer interface. Side chains of Met2669 pack against Gly2671 of the neighboring monomer, creating hydrophobic interactions. (B) Hydrogen bonding between the monomers at strand D.
drophobic residues that form the interface between the C strands (underlined in Figure 6B). This sequence motif forms a signature for dimerization and is not seen in other vertebrate filamin domains. We therefore propose that all vertebrate filamins have dimerization interfaces analogous to that seen in human filamin C domain 24. It appears that C. elegans filamin A (encoded in chromosome IV, wormbase gene ID Y66H1B.2) and Drosophila filamin (Cher) also share the same dimerization sequence signature in their most C-terminal domain (Figure 6B). However, other filamin-like proteins in C. elegans (encoded in the X chromosome, wormbase gene ID C23F12.1) and Drosophila (Jbug) seem not to have this signature in their last domain, nor do they resemble the Dictyostelium filamin domain 6 (data not shown). This suggests that the other filamin-like proteins in the worm and in the fly may not dimerize or that they have developed a different dimerization mechanism. They seem to also differ in the other aspects and, for example, contain three calponin homology domains in their N termini, while all the other filamins have only two that form the widely conserved actin binding site (Korenbaum and Rivero, 2002). Model of the Molecular Structure of Vertebrate Filamin Although a different interface is employed, the two chains in both the filamin C domain 24 dimer and in the Dictyostelium filamin domain 6 dimer are arranged antiparallel. In the case of the Dictyostelium structure, this arrangement was taken to indicate that the chains in the whole molecule were arranged antiparallel (McCoy et al., 1999; Popowicz et al., 2004), which was also consistent with a range of other data. However, although there are no data on the overall structure of vertebrate filamin C, electron micrographs of individual human filamin molecules from various tissues indicate that, in contrast to the Dictyostelium analog, the two chains in these vertebrate actin binding proteins often appear to be arranged parallel to one another, producing V-shaped profiles with varying angles between the two arms (Gorlin et al., 1990; Hartwig and Stossel, 1981; Stossel et al., 2001; Tyler et al., 1980), and indeed Gorlin
Figure 5. Analysis of Dimerization Mutants (A) CD spectra of filamin C domain 24 and two mutants used in the study. (B) Gel filtration of filamin C domain 24 wild-type (solid line), M2990D (dotted line), and Mutant20 (dashed line). All proteins were injected at 1 mg/ml.
et al. (1990) proposed that there may be interactions between several Ig domains in the different arms. How then can the antiparallel arrangement of the chains in the vertebrate filamin Ig domain 24 dimer be reconciled with the two chains in the rod region of this molecule being arranged parallel? We propose that one way to do this would be for the dyad relating the two domain 24 chains to be arranged parallel to the rod axis in the intact filamin molecule, as illustrated in Figure 7, rather than perpendicular to the rod region axis as proposed for the Dictyostelium filamin dimer (McCoy et al., 1999; Popowicz et al., 2004). Although further work to obtain crystals containing at least vertebrate domains 23 and 24 will be needed to resolve this controversy, this hypothesis would be consistent with the EM observations. Moreover, in the filamin C domain 24 dimer, the N-terminal residues localize at the edges of a rather flat surface along the dyad axis (Figure 3A). This surface of filamin has a large neutral area (Figure 6D, left) that would be ideally placed to interact with the adjacent domain 23. This surface seems to be predominantly
Structure 116
Figure 6. Comparisons and Surface Features of Filamins (A) Amino acid sequence alignment of human filamin C domain 24 with Dictyostelium filamin domains 4 and 5. The β strands of filamin C domain 24 are indicated above. (B) Amino acid sequence alignment of filamin domain 24 sequences from available vertebrate species as well as filamin A from C. elegans and Cher from Drosophila. Green indicates residues that differ in all vertebrate isoforms; violet, filamin A-specific residues; blue, filamin B-specific residues; red, filamin C-specific residues. The hydrophobic dimerization signature is underlined. (C) The accessible surface of the dimeric human filamin C domain 24 with the locations of isoform-specific substitutions colored as in (B) shown on three successive 90° rotations about the vertical. (D) Accessible surface of the dimeric human filamin C domain 24 colored according to the surface potential. The images have been rotated by 180°.
conserved between the vertebrate filamin isoforms (Figure 6C, left). The Dimer Dissociation Constant Is in the Micromolar Range Filamin C domain 24 can be dissociated by 0.8 M KCl (Himmel et al., 2003), raising the possibility that the dimerization could be regulated in vivo, possibly as a mechanism to control the actin crosslinking. We therefore used analytical ultracentrifugation to determine the dissociation constant of the filamin C domain 24 dimer. Sedimentation equilibrium data from three different concentrations (17.25 M, 8.63 M, and 4.31 M) could be fitted to the ideal monomer/dimer model with Kd = 2.5 M (Figure 8). The M2669D mutant had a significantly weaker Kd of 3 mM (data not shown), consistent with its inhibiting dimerization but not altering conformation dramatically. Consistent with the analytical ultracentrifugation data, the filamin C domain 24 eluted at lower apparent molecular weight in gel filtration at low micromolar concentrations (Figure 8C), indicating that a monomer-dimer equilibrium existed in these conditions. Each filamin chain has a single actin binding site, and so dimerization is necessary for the actin crosslinking function of filamins. During platelet activation, the actin
crosslinking function of filamin A is ablated by calpainmediated proteolysis that targets mainly the two hinge regions between domains 15 and 16 and 23 and 24 (Davies et al., 1978; Gorlin et al., 1990). The actin binding site of filamin resides in its N terminus, but many signaling proteins interact with the C terminus of filamins (reviewed in van der Flier and Sonnenberg, 2001). Our observation that the dissociation constant for the filamin domain 24 dimer is of the same order as its cellular concentration raises the possibility that these signaling proteins might regulate filamin activity by altering the monomer-dimer equilibrium. The filamin A concentration in the platelets is 10 M (Hartwig, 2002). If the fulllength filamin had a dissociation constant even close to the value of the individual domain, a monomer-dimer equilibrium may be present in cells. In such a case, the proteins interacting with the C terminus could affect the equilibrium by concentrating filamins at defined locations or by directly interfering with the dimerization interface. In their early studies, Hartwig and Stossel (1981) performed analytical ultracentrifugation measurement between 0.7 to 2 M (monomer concentrations) of macrophage filamin but did not detect dissociation of the dimer. Thus, it is possible that the dimer affinity of the intact filamin is higher than that of domain 24 alone, which would be consistent with there being
Dimerization of Human Filamin C 117
Figure 7. A Model of the Orientation of Full Filamin in Relation to the 24th Domain Based on our current crystallographic data and the previously published EM pictures, we propose that the two-fold symmetry axis (inset) of the filamin domain 24 dimer is oriented parallel to the long axis of the V-shaped filamin dimer (full picture). We also propose that the Ig domain 23 and possibly also parts of the long linker would be able to interact with domain 24. Clearly, the observed total length of filamin dimer (160 nm) is too short to allow straight pearls-in-a-string arrangement of 48 Ig-like domains (2 × 24 × 4.5 nm = 216 nm). Thus, it is likely that the Ig domains are inclined relative to one another, as proposed by Fucini et al. (1999).
additional interactions between other Ig domains in the rod, as suggested by Gorlin et al. (1990). Further work will be required to test the hypothesis of monomer dimer equilibrium of filamins in vivo. In summary, we have determined the crystal structure of the vertebrate filamin Ig-like domain 24 that mediates dimerization, and we have shown that, although its structure resembles domains 4 and 5 of the Dictyostelium filamin analog, human filamin Ig domain 24 has a novel dimerization interface. The filamin C domain 24 dimer can be fitted to a structure of parallel filamin chains by aligning the dimer dyad axis to the rod axis. However, our data do not exclude the possibility of antiparallel alignment, as in the case of Dictyostelium filamin. Moreover, the dissociation constant for the domain 24 dimer is in the micromolar range, which is the order of the cellular filamin concentration. This raises the possibility that regulation of the monomer-dimer equilibrium could be important functionally. Experimental Procedures Protein Expression, Purification, and Crystallization The fragment encoding the 24th Ig-like domain of human filamin C (residues 2633–2725, Swiss Prot accession number Q14315) was
amplified from a human skeletal muscle cDNA library (Matchmaker, Clontech, Invitrogen, Palo Alto, CA). The fragment was cloned in NcoI and NotI sites of a modified pETM24d (Novagen) plasmid containing an N-terminal His6 tag followed by the cleavage sequence for tobacco etch virus (Tev) protease ENLYFQ*GAMG (Parks et al., 1994) (The asterisk denotes the cleavage site, and the extra residues left to the recombinant protein are underlined). The insert was verified by DNA sequencing. Protein expression was induced in Escherichia coli BL21 (DE3) for 3 hr at 37°C in the presence of 0.4 mM IPTG. The protein was purified on a NiNTA agarose column (Qiagen) according to manufacturer's instructions and eluted with 250 mM imidazole. Buffer was changed with PD10 columns (Amersham Biosciences) to 20 mM Tris (pH 8.0), and the protein was cleaved with Tev protease at 30°C overnight. Further purification was achieved by cation exchange chromatography using Fractogel EMD SO3− (S) matrix (Merck). The wild-type protein was concentrated to 40–42 mg/ml in 20 mM Tris (pH 8.0). Crystals were grown by hanging drop vapor diffusion after mixing the protein in a 1:1 volume ratio with the reservoir solution of 1.6 M Na citrate, 0.1 M HEPES (pH 7.5). To produce Se-Met-labeled protein, expression was induced in the Met-auxotrophic E. coli strain B834(DE3) (Leahy et al., 1992; Wood, 1966) in minimal media containing 0.5 mM L-SeMet (Calbiochem) for 20 hr at 20°C. The protein was purified and crystallized as above. Two mutant constructs of the filamin C domain 24 were produced. In the first mutant, Met2669 was changed to Asp (M2669D). In the second one, called mutant 20, a stretch of eight residues (aa 2666–2673, NMMMVGVH) was substituted with the string from human filamin A domain 20 (QDMTAQVT). The mutations were made with the Quikchange mutagenesis kit (Stratagene). Protein purification was achieved as described above, followed by gel filtration on a Superdex-75 16/60 column (Amersham Biotech). The mutant proteins were concentrated to 8–10 mg/ml. Data Collection, Structure Determination, and Refinement For data collection, the crystals were immersed in reservoir solution containing 10% glycerol and flash frozen at 100 K. The data sets were collected at the EMBL beamlines X13 and BW7A at the DORIS storage ring, DESY, Hamburg, using the 165 mm Mar CCD detector (MARData) and were processed and scaled using XDS program package (Kabsch, 1993). CNS 1.0 (Brunger et al., 1998) was used to locate the Se atoms and to generate phases using single-wavelength anomalous diffraction from the Se peak wavelength and also for density modification. The electron density map produced could be used for automatic model building in Arp/Warp 6.0 (Perrakis et al., 1999). The initial model containing 86 residues was used to solve the native data set. Further refinement was done with programs O (Jones et al., 1991) and Refmac5 (Murshudov et al., 1997). The final model contains 96 of 97 residues present in the construct. The first Gly residue derived from the plasmid sequence could not be seen in the electron density. Double atomic conformations were found for residues Cys 2679 and Asn 2689. Those were refined using SHELXPRO (Sheldrick and Gould, 1995) and Refmac5. For the final model, a restrained isotropic refinement was performed using Refmac5 to 1.43 Å with weight matrix 0.4. The Ramachandran plot of the final model shows 93.6% of residues in the most favored regions and 6.4% in additional allowed regions. Structure figures were generated with Pymol (DeLano Scientific LLC, San Carlos, CA) and Grasp (Nicholls et al., 1991). Sequence alignment has been generated with ClustalX (1.8) (Thompson et al., 1997) with protein weight matrix Gonnet PAM250. The dimer interaction was analyzed by using the server http://www.biochem. ucl.ac.uk/bsm/PP/server (Jones and Thornton, 1996). Superimposition of the domains was performed in program O (Jones et al., 1991). Gel Filtration, Analytical Ultracentrifugation, and Circular Dichroism Analytical gel filtration was performed at 23°C in a Superdex 75 10/ 30 column (Amersham Biotech) in 150 mM KCl, 50 mM Na-phosphate (pH 7.5) (Figure 5B) or 150 mM KCl, 20 mM Tris, 1 mM DTT (pH 8.0) (Figure 8C), at a flow rate of 0.5 ml/min. The injection volume was 100 l, and protein concentrations are indicated in the figure legends.
Structure 118
Figure 8. Analysis of the Dimerization of Filamin C Domain 24 (A) Sedimentation equilibrium analysis was performed by analytical ultracentrifugation. Plots of Mw, app against concentration are shown for three concentrations, together with the theoretical curve corresponding to the ideal monomer-dimer equilibrium with parameters derived for the fitting. (B) Corresponding plots showing residuals against radius for the fitting data to the experimental data. Concentrations used were 17.25 M (top), 8.63 M (middle), and 4.31 M (bottom). (C) Analytical gel filtration results after injecting 102 M (solid line), 6.3 M (dashed line), or 1.6 M (dotted line) of filamin C domain 24. The estimated protein concentrations based on absorbance at 280 nm at the peak were 32 M, 1.2 M, and 0.13 M, respectively. To allow direct comparison, the chromatograms were scaled to the peak high and absorbance at 280 nm of the 102 m injection, and absorbance at 215 nm of the 6.3 M and 1.6 M injections are shown.
Sedimentation equilibrium analytical ultracentrifugation was performed with an Optima XLA (Beckman Coulter) instrument using An-60Ti rotor at +4°C and scanning wavelength of 229 nm. Filamin C domain 24 protein sample was diluted in 140 mM NaCl, 10 mM Na-phosphate (pH 7.4). Four hundred microliters of the samples and 420 l of the buffer for reference were loaded in the analytical ultracentrifuge cells. After the initial scan, the centrifuge was overspeeded at 30,000 rev/min to reduce the time taken to reach the equilibrium (Van Holde and Baldwin, 1958). Then, the speed was reduced to 20,000 rev/min, and scans were taken at intervals of 24 hr. The equilibrium was determined by the criteria that successive scans were indistinguishable, and at this point scans averaging over 100 readings were taken for analysis. Subsequently, the centrifuge was overspeeded to sediment the macromolecules away from the meniscus, before slowing to equilibrium speed and taking a further scan to establish the baseline absorbance for each cell. The calculations were performed as in Mills et al., 2003. CD measurements were performed in the far UV region (190–250 nm) at 20°C using a Jasco J-715 spectropolarimeter. Protein concentrations were in the range of 0.22–0.23 mg/ml in 5 mM Na phosphate (pH 7.1). Spectra were obtained with a scan rate of 50 nm/ min in a cuvette with 0.1 cm optical path length. The data were collected at 0.2 nm intervals, and the average of 16 scans is reported.
Acknowledgments We thank the EMBL-Hamburg beamline staff, particularly Andrea Schmidt and Mannfred Weiss, for help during the data collection, David Owen for amino acid analysis of the proteins, and Ulrich Bergmann for MALDI-TOF mass spectrometry. This study has been supported by the Academy of Finland grants 51863, 207021, 105211, and by EMBO short term travel fellowship to J.Y. Received: August 26, 2004 Revised: October 29, 2004 Accepted: October 29, 2004 Published: January 11, 2005
Bellanger, J.M., Astier, C., Sardet, C., Ohta, Y., Stossel, T.P., and Debant, A. (2000). The Rac1- and RhoG-specific GEF domain of Trio targets filamin to remodel cytoskeletal actin. Nat. Cell Biol. 2, 888–892. Brunger, A.T., Adams, P.D., Clore, G.M., DeLano, W.L., Gros, P., Grosse-Kunstleve, R.W., Jiang, J.S., Kuszewski, J., Nilges, M., Pannu, N.S., et al. (1998). Crystallography & NMR system: A new software suite for macromolecular structure determination. Acta Crystallogr. D Biol. Crystallogr. 54, 905–921. Cunningham, C.C., Gorlin, J.B., Kwiatkowski, D.J., Hartwig, J.H., Janmey, P.A., Byers, H.R., and Stossel, T.P. (1992). Actin-binding protein requirement for cortical stability and efficient locomotion. Science 255, 325–327. Davies, P.J., Wallach, D., Willingham, M.C., Pastan, I., Yamaguchi, M., and Robson, R.M. (1978). Filamin-actin interaction. Dissociation of binding from gelation by Ca2+-activated proteolysis. J. Biol. Chem. 253, 4036–4042. Faulkner, G., Lanfranchi, G., and Valle, G. (2001). Telethonin and other new proteins of the Z-disc of skeletal muscle. IUBMB Life 51, 275–282. Fox, J.W., Lamperti, E.D., Eksioglu, Y.Z., Hong, S.E., Feng, Y., Graham, D.A., Scheffer, I.E., Dobyns, W.B., Hirsch, B.A., Radtke, R.A., et al. (1998). Mutations in filamin 1 prevent migration of cerebral cortical neurons in human periventricular heterotopia. Neuron 21, 1315–1325. Fucini, P., Renner, C., Herberhold, C., Noegel, A.A., and Holak, T.A. (1997). The repeating segments of the F-actin cross-linking gelation factor (ABP-120) have an immunoglobulin-like fold. Nat. Struct. Biol. 4, 223–230. Fucini, P., Koppel, B., Schleicher, M., Lustig, A., Holak, T.A., Muller, R., Stewart, M., and Noegel, A.A. (1999). Molecular architecture of the rod domain of the Dictyostelium gelation factor (ABP120). J. Mol. Biol. 291, 1017–1023. Gorlin, J.B., Yamin, R., Egan, S., Stewart, M., Stossel, T.P., Kwiatkowski, D.J., and Hartwig, J.H. (1990). Human endothelial actinbinding protein (ABP-280, nonmuscle filamin)—a molecular leaf spring. J. Cell Biol. 111, 1089–1105.
References
Hartwig, J.H. (2002). Platelet structure. In Platelets, A.D. Michelson, ed. (New York: Academic Press), pp. 35–50.
Ayscough, K.R. (1998). In vivo functions of actin-binding proteins. Curr. Opin. Cell Biol. 10, 102–111.
Hartwig, J.H., and Stossel, T.P. (1981). Structure of macrophage actin-binding protein molecules in solution and interacting with actin filaments. J. Mol. Biol. 145, 563–581.
Dimerization of Human Filamin C 119
Himmel, M., Van Der Ven, P.F., Stocklein, W., and Furst, D.O. (2003). The limits of promiscuity: isoform-specific dimerization of filamins. Biochemistry 42, 430–439. Jones, S., and Thornton, J.M. (1996). Principles of protein-protein interactions. Proc. Natl. Acad. Sci. USA 93, 13–20. Jones, T.A., Zou, J.Y., Cowan, S.W., and Kjeldgaard (1991). Improved methods for building protein models in electron density maps and the location of errors in these models. Acta Crystallogr. A 47, 110–119.
S.M., Duncan, J.S., Dubeau, F., Scheffer, I.E., Schachter, S.C., et al. (2001). Mutations in the X-linked filamin 1 gene cause periventricular nodular heterotopia in males as well as in females. Hum. Mol. Genet. 10, 1775–1783. Sheldrick, G.M., and Gould, R.O. (1995). Structure solution by iterative peaklist optimization and tangent expansion in space group P1. Acta Crystallogr. B 51, 423–431. Small, J.V., Stradal, T., Vignal, E., and Rottner, K. (2002). The lamellipodium: where motility begins. Trends Cell Biol. 12, 112–120.
Kabsch, W. (1993). Automatic processing of rotation diffraction data from crystals of initially unknown symmetry and cell constants. J. Appl. Crystallogr. 26, 795–800.
Stossel, T.P., Condeelis, J., Cooley, L., Hartwig, J.H., Noegel, A., Schleicher, M., and Shapiro, S.S. (2001). Filamins as integrators of cell mechanics and signalling. Nat. Rev. Mol. Cell Biol. 2, 138–145.
Kakita, A., Hayashi, S., Moro, F., Guerrini, R., Ozawa, T., Ono, K., Kameyama, S., Walsh, C.A., and Takahashi, H. (2002). Bilateral periventricular nodular heterotopia due to filamin 1 gene mutation: widespread glomeruloid microvascular anomaly and dysplastic cytoarchitecture in the cerebral cortex. Acta Neuropathol. (Berl.) 104, 649–657.
Thompson, J.D., Gibson, T.J., Plewniak, F., Jeanmougin, F., and Higgins, D.G. (1997). The CLUSTAL_X windows interface: flexible strategies for multiple sequence alignment aided by quality analysis tools. Nucleic Acids Res. 25, 4876–4882.
Korenbaum, E., and Rivero, F. (2002). Calponin homology domains at a glance. J. Cell Sci. 115, 3543–3545. Leahy, D.J., Hendrickson, W.A., Aukhil, I., and Erickson, H.P. (1992). Structure of a fibronectin type III domain from tenascin phased by MAD analysis of the selenomethionyl protein. Science 258, 987– 991. Loo, D.T., Kanner, S.B., and Aruffo, A. (1998). Filamin binds to the cytoplasmic domain of the beta1-integrin. Identification of amino acids responsible for this interaction. J. Biol. Chem. 273, 23304– 23312. McCoy, A.J., Fucini, P., Noegel, A.A., and Stewart, M. (1999). Structural basis for dimerization of the Dictyostelium gelation factor (ABP120) rod. Nat. Struct. Biol. 6, 836–841. McGough, A. (1998). F-actin-binding proteins. Curr. Opin. Struct. Biol. 8, 166–176. Mills, I.G., Praefcke, G.J., Vallis, Y., Peter, B.J., Olesen, L.E., Gallop, J.L., Butler, P.J., Evans, P.R., and McMahon, H.T. (2003). EpsinR: an AP1/clathrin interacting protein involved in vesicle trafficking. J. Cell Biol. 160, 213–222. Murshudov, G.N., Vagin, A.A., and Dodson, E.J. (1997). Refinement of macromolecular structures by the maximum-likelihood method. Acta Crystallogr. D Biol. Crystallogr. 53, 240–255. Murzin, A.G., Brenner, S.E., Hubbard, T., and Chothia, C. (1995). SCOP: a structural classification of proteins database for the investigation of sequences and structures. J. Mol. Biol. 247, 536–540. Nicholls, A., Sharp, K.A., and Honig, B. (1991). Protein folding and association: insights from the interfacial and thermodynamic properties of hydrocarbons. Proteins 11, 281–296. Ohta, Y., Suzuki, N., Nakamura, S., Hartwig, J.H., and Stossel, T.P. (1999). The small GTPase RalA targets filamin to induce filopodia. Proc. Natl. Acad. Sci. USA 96, 2122–2128. Ott, I., Fischer, E.G., Miyagi, Y., Mueller, B.M., and Ruf, W. (1998). A role for tissue factor in cell adhesion and migration mediated by interaction with actin-binding protein 280. J. Cell Biol. 140, 1241– 1253. Parks, T.D., Leuther, K.K., Howard, E.D., Johnston, S.A., and Dougherty, W.G. (1994). Release of proteins and peptides from fusion proteins using a recombinant plant virus proteinase. Anal. Biochem. 216, 413–417. Perrakis, A., Morris, R., and Lamzin, V.S. (1999). Automated protein model building combined with iterative structure refinement. Nat. Struct. Biol. 6, 458–463. Popowicz, G.M., Muller, R., Noegel, A.A., Schleicher, M., Huber, R., and Holak, T.A. (2004). Molecular structure of the rod domain of dictyostelium filamin. J. Mol. Biol. 342, 1637–1646. Puius, Y.A., Mahoney, N.M., and Almo, S.C. (1998). The modular structure of actin-regulatory proteins. Curr. Opin. Cell Biol. 10, 23–34. Radaev, S., Motyka, S., Fridman, W.H., Sautes-Fridman, C., and Sun, P.D. (2001). The structure of a human type III Fcgamma receptor in complex with Fc. J. Biol. Chem. 276, 16469–16477. Sheen, V.L., Dixon, P.H., Fox, J.W., Hong, S.E., Kinton, L., Sisodiya,
Thompson, T.G., Chan, Y.M., Hack, A.A., Brosius, M., Rajala, M., Lidov, H.G., McNally, E.M., Watkins, S., and Kunkel, L.M. (2000). Filamin 2 (FLN2): A muscle-specific sarcoglycan interacting protein. J. Cell Biol. 148, 115–126. Tyler, J.M., Anderson, J.M., and Branton, D. (1980). Structural comparison of several actin-binding macromolecules. J. Cell Biol. 85, 489–495. van der Flier, A., and Sonnenberg, A. (2001). Structural and functional aspects of filamins. Biochim. Biophys. Acta 1538, 99–117. Van Holde, K.E., and Baldwin, R.L. (1958). Rapid attainment of sedimentation equilibrium. J. Phys. Chem. 62, 734–743. Van Troys, M., Vandekerckhove, J., and Ampe, C. (1999). Structural modules in actin-binding proteins: towards a new classification. Biochim. Biophys. Acta 1448, 323–348. Wood, W.B. (1966). Host specificity of DNA produced by Escherichia coli: bacterial mutations affecting the restriction and modification of DNA. J. Mol. Biol. 16, 118–133. Zenker, M., Rauch, A., Winterpacht, A., Tagariello, A., Kraus, C., Rupprecht, T., Sticht, H., Reis, A., Robertson, S.P., Twigg, S.R., et al. (2004). A dual phenotype of periventricular nodular heterotopia and frontometaphyseal dysplasia in one patient caused by a single FLNA mutation leading to two functionally different aberrant transcripts. Am. J. Hum. Genet. 74, 731–737. Accession Numbers The coordinates of the structure have been deposited in the Protein Data Bank with ID code 1V05.
Structural Basis for Vertebrate Filamin Dimerization Regina Pudas,1 Tiila-Riikka Kiema,1 P. Jonathan G. Butler,2 Murray Stewart,2 and Jari Ylänne1,* 1 Biocenter Oulu and Department of Biochemistry University of Oulu Finland 2 MRC Laboratory on Molecular Biology Cambridge United Kingdom
Summary Filamins are essential in cell motility and many developmental processes. They are large actin crosslinking proteins that contain actin binding domains in their N termini and a long rod region constructed from 24 tandem Ig domains. Dimerization is crucial for the actin crosslinking function of filamins and requires the most C-terminal Ig domain. We describe here the crystal structure of this 24th Ig domain (Ig24) of human filamin C and show how it mediates dimerization. The dimer interface is novel and quite different to that seen in the Dictyostelium discoideum filamin analog. The sequence signature of the dimerization interface suggests that the C-terminal domains of all vertebrate filamins share the same dimerization mechanism. Furthermore, we show that point mutations in the dimerization interface disrupt the dimer and that the dissociation constant for recombinant Ig24 is in the micromolar range. Introduction Filamins are key integrators of cell mechanics and signaling that combine actin binding and scaffolding functions. The actin crosslinking activity of filamins is important in modulating the properties of the actin cytoskeleton and, through this function, influencing cell shape as well as being a key component of the cell locomotion machinery. The arrangement of actin filaments and their dynamics are regulated by a range of actin binding proteins that orchestrate the polymerization and depolymerization of filaments together with their linking into bundles and their attachment to membranes (Ayscough, 1998; McGough, 1998; Small et al., 2002). The crosslinking proteins that bundle filaments typically contain two actin binding domains separated by a variable linker (Puius et al., 1998; Van Troys et al., 1999). The linkers determine the geometry of actin crosslinking and can also function as binding sites for downstream and upstream signaling components (Small et al., 2002). Filamins are an important class of actin crosslinking proteins that make crucial contributions to the stability of webs of actin filaments at the cell periphery and link them to the plasma membrane, as well as serving as key scaffolds for key factors that integ*Correspondence: [email protected]
rate signaling and cell mechanics (Stossel et al., 2001). Filamins (also called ABP280) have N-terminal actin binding domains followed by a rod region constructed from tandem sequence repeats that are predicted to have immunoglobulin-like (Ig) folds. The rod region of the Dictyostelium filamin analog (also called ABP120, ddFLN, or gelation factor) has six Ig domains, whereas there are 24 in vertebrates (Stossel et al., 2001; van der Flier and Sonnenberg, 2001). Dimerization is crucial for the actin-crosslinking function of filamins and requires the most C-terminal Ig domain (see Gorlin et al., 1990). Mammals have three filamin genes: filamins A and B are expressed ubiquitously, whereas filamin C is muscle specific (Stossel et al., 2001; van der Flier and Sonnenberg, 2001). Filamin A is encoded in the X chromosome, and mutations cause periventricular heterotopia and several developmental malformations in both females and males (Fox et al., 1998; Kakita et al., 2002; Sheen et al., 2001; Zenker et al., 2004). Cultured cells lacking filamin A show defects in plasma membrane stability and in cell migration (Cunningham et al., 1992). In muscle, only 3% of filamin C is associated with the sarcolemmal membrane (Thompson et al., 2000) with the remainder located primarily at the Z-discs. Filamin C connects plasma membrane and myofibrils by binding to γ- and δ-sarcoglycans (Thompson et al., 2000) and to α-actinin 2 via FATZ and myotilin (Faulkner et al., 2001). By bringing together transmembrane proteins such as integrins (Loo et al., 1998), γ- and δ-sarcoglycans (Thompson et al., 2000), intracellular signaling proteins like small GTPases (Ohta et al., 1999), and actin filaments, filamins provide a link between plasma membrane signaling and actin cytoskeleton. The molecular architecture of filamins has been controversial. Electron micrographs of human filamin A suggest that in the rod region the two chains lie parallel to one another (Gorlin et al., 1990; Stossel et al., 2001), but the crystal structure of the dimerization region of the Dictyostelium filamin analog instead supported an antiparallel arrangement (McCoy et al., 1999; Popowicz et al., 2004). Here, we present a crystal structure of the dimerization domain of human filamin C that has a novel dimer interface which is quite different to that observed in the Dictyostelium filamin analog. We show that mutations in this interface prevent dimerization of the domain. We suggest a model of how the dimerization domain can be fitted to filamin molecular architecture and, based on ultracentrifugation studies, raise the possibility that filamin dimerization may be an important control point in vivo. Results and Discussion Crystal Structure of 24th Ig Domain of Filamin C We obtained P6122 crystals that diffracted to 1.43 Å of Ig domain 24 of human filamin C, which mediates dimerization. The structure was determined using single wavelength anomalous diffraction (SAD) of Se-Metlabeled protein crystals and refined to R = 18.6% and
Structure 112
Table 1. Crystallographic Data Data Collection Statistics
Native
SeMet Peak
Wavelength (Å) Beamline Crystal to detector distance (mm) Space group Unit cell parameters (Å) a=b c Resolution range (Å) R (%)a I/sigma (I) Completeness (%) Redundancy
0.8020 X13 (EMBL, DESY) 120.0 P6(1)22
0.9778 BW7A (EMBL, DESY) 150.0 P6(1)22
48.099 117.485 20–1.43 (1.53–1.43)b 3.7 (22.5) 41.59 (13.29) 99.8 (100) 15.8 (16.1)
48.800 117.100 20–2.0 (2.1–2.0) 2.8 (8.6) 42.96 (19.86) 99.8 (99.9) 7.6 (7.5)
Phasing Statistics Number of SeMet sites Phasing power FOM before density modification FOM after density modification R (%) after autobuilding ARP/wARP
5 4.25 (centric), 3.90 (acentric) 0.4577 0.8697 28.9
Final Model Statistics (Refinement) Resolution limits No. of reflections used for refinement Number of reflections in the test set R (%) Rfree (%) Rmsd bonds (Å)/angles (°) Residues in Ramachandran Residues in most favored regions (%) Residues in additional allowed regions (%) Residues in generously allowed and in disallowed regions (%) Protein atoms Water atoms Average B factor (Å2) B factor from Wilson plot (Å2) a b
17.73–1.43 14,038 1,560 18.58 20.5 0.022/1.431 96 92.3 7.7 0 731 73 13.273 19.177
R factor = (SUM(ABS(I(h, i) − I (h))))/SUM(I(h, i))). Values for the last resolution shell are given in parentheses.
Rfree = 20.5% (Table 1). In these crystals, domain 24 had the expected Ig-like fold (Figure 1A), with strands A, B, E, and D forming a four-stranded antiparallel β sheet and strands C, F, and G forming a three-stranded sheet (Figure 2A). In the SCOP database (Murzin et al., 1995), the filamin Ig-like domain forms a distinct family in the superfamily of E-set domains, in the Ig-like β-sandwich fold. Before the current structure, three other Ig domain structures have been available for this family (Fucini et al., 1997; McCoy et al., 1999; Popowicz et al., 2004). Of these, our structure of human filamin C domain 24 resembles most closely Dictyostelium filamin domains 5 and 4, whereas the Dictyostelium domain 6, which mediates dimerization in this molecule, is clearly distinct. The Dictyostelium filamin domain 5, which is not involved in dimerization (McCoy et al., 1999), has the highest similarity, with root-mean-square (rms) deviation of 1.6 Å of 69 superimposed C-α atoms (Figure 1B), whereas the rms deviation with the corresponding atoms of the average NMR structure of the Dictyostelium domain 4 (Fucini et al., 1997) is 2.2 Å (Figure 1C). The human filamin Ig domain 24 and Dictyostelium domains 4 and 5 share the same overall topology (Figures 2A–2C), although in filamin C domain 24 strand D contains six residues, whereas in the Dictyostelium do-
mains 5 and 4 the strands D are four and three residues long, respectively. The loop connecting strands A and B makes parallel β strand-like hydrogen bonds to the end of the strand G of the opposite β sheet in all three domains. Also, the loop between C and D turns close to the end of the strand C before continuing to strand D. Only in the Dictyostelium domain 5 can these two connections be classified as short β strands (A# and C#) (Figure 2B). As expected, the loops vary more than the β strands, with the loop between β strands B and C being the most divergent (Figures 1B and 1C). In the sequence alignment, this loop is longer in the Dictyostelium domains (Figure 6A) and it also has the highest B-factor values in the X-ray structures (this study and McCoy et al., 1999) or highest variability among the alternative NMR assemblies (Dictyostelium domain 4) (Fucini et al., 1997). The structure of Dictyostelium filamin domain 6, which mediates dimerization in this molecule, is much less similar to human filamin C domain 24 than the other Dictyostelium domains discussed above. When comparing the same 69 residues, the rms deviation between filamin C domain 24 and Dictyostelium domain 6 was 8.3 Å. This large difference was mainly a reflection of differences in topology (Figures 1D and 2D). The Dictyostelium domain 6 does not have strands A and A# and has an additional strand H.
Dimerization of Human Filamin C 113
Figure 1. Structure and Comparison of the Single Domains (A) Ribbon representation of the human filamin C domain 24 structure. N and C termini are indicated. (B) Superimposition of filamin C domain 24 (blue) with Dictyostelium filamin domain 5 (yellow). (C) Superimposition with Dictyostelium domain 4 (green). (D) Superimposition with Dictyostelium domain 6 (magenta). Note that the N-terminal strand of filamin C domain 24 and the C-terminal strand of Dictyostelium domain 6 could not be superimposed. The PDB ID code for the Dictyostelium filamin domain 4 is 1KSR and for domains 5 and 6 is 1QFH.
Both of these features are crucial for the way in which the Dictyostelium protein dimerizes (McCoy et al., 1999), and so their absence in filamin domain 24 accounts for its fundamentally different dimerization architecture. Filamin C Has a Different Dimerization Interface from that in Dictyostelium Filamin Vertebrate filamin Ig domain 24 is necessary for dimerization and itself dimerizes in solution (Himmel et al., 2003). Although the asymmetric unit of our crystal contained only a single chain, an extensive interface involving strands C and D was formed between adjacent chains related by a crystallographic two-fold axis (Figure 3), which we identified as the putative dimerization interface. The interface buried 1109 Å2 (19%) of the accessible surface area of the monomer and contained 45% polar atoms. There are six hydrogen bonds in the interface connecting the D strands so that the fourstranded β sheets of each chain form a single eightstranded antiparallel β sheet (Figures 3B and 4B). In addition, Met2667 and Met2669 form a putative hydrophobic packing with Gly2671 of the adjacent chain (Figure 4A). The dimerization interface formed with filamin C domain 24 is strikingly different from that seen in Dictyostelium filamin domain 6 and involves the opposite edge of the β sandwich. In Dictyostelium domain 6, the interface is formed primarily by strands B and G, whereas in filamin domain 24 it is formed by strands C and D (Figures 2A and 2D). Crucially, this interface in Dictyostelium domain 6 can only form because of the absence of strands A and A# in this domain. The presence of this strand in filamin domain 24 therefore effectively prevents the chains dimerizing in the same way as the Dictyostelium analog. We engineered mutants to confirm that the putative dimer interface seen in the crystals was that involved in dimerization in solution. In one, Met2669 in strand
C was mutated to aspartic acid to introduce negative charge in the key hydrophobic interaction observed in the interface in the crystal. In another, strand C was replaced with the corresponding strand from human filamin A Ig domain 20, which does not dimerize in solution (data not shown). Both mutants were expressed and purified as the wild-type filamin C domain 24. Their CD spectra showed a deep minimum at 216 nm, characteristic of a β structure (Figure 5A), although in the 190–210 nm range the CD spectra of the mutants differed from wild-type, possibly because of less ordered domain-domain interactions. By gel filtration, wild-type filamin C domain 24 eluted as a dimer, whereas both mutants eluted later with retention times expected for a single chain (Figure 5B). The disruption of filamin C domain 24 dimerization by these mutations gives powerful support to the hypothesis that the putative interface identified in the crystal structure genuinely represents the dimerization interface present in solution. The Dimer Interface Is Common to All Vertebrate Filamins Although vertebrate filamin C-terminal domains are very similar in amino acid sequence alignments, isoforms A, B, and C clearly form distinct groups (Figure 6B). Because several functionally important interactions have been mapped to these domains (Bellanger et al., 2000; Loo et al., 1998; Ohta et al., 1999; Ott et al., 1998; Thompson et al., 2000), we compared the surface residues of different filamins. We classified substitutions between vertebrate filamin 24 domain sequences as either A, B, or C specific, or alternatively, there are those that are different in all three. When these substitutions were mapped on the accessible surface of human filamin C domain 24, they mostly located on the nondimer-forming edge of the β sandwich (Figure 6C). This fits well with the proposal that the edges of the sandwich are the most common interaction sites (Radaev et al., 2001).
Structure 114
Figure 2. Topology Diagrams (A) Human filamin C domain 24. (B) Dictyostelium filamin domain 5. (C) Dictyostelium filamin domain 4. (D) Dictyostelium filamin domain 6. Beta strands are shown as arrows, and 310 helices are shown as cylinders. In (A) and (D), strands forming the dimerization interface are colored gray.
The similarity of human filamin Ig-like sequence repeats with those of Dictyostelium domains 4 and 5 extends to all the human filamin sequences (van der Flier and Sonnenberg, 2001). Alignments also predict that the Dictyostelium domain 6 is a more distant member of the family (van der Flier and Sonnenberg, 2001), con-
sistent with our structural and biochemical studies that show that the dimerization interface in human filamin C Ig domain 24 is different from that of Dictyostelium domain 6. When the sequences of the C-terminal domains of vertebrate filamins are aligned, a high level of overall conservation is clearly present, particularly in the hyFigure 3. Crystallographic Dimer of Filamin C Domain 24 Views rotated by 180° about the vertical. Strands of the monomers are colored, respectively, blue and green. Strands C and D involved in dimer contact are indicated. The extended antiparallel β sheet that forms on dimerization is shown in (B). The two-fold axis is marked. The extra residues at the N terminus that are derived from the vector are indicated in blue.
Dimerization of Human Filamin C 115
Figure 4. Filamin Domain 24 Dimerization Interface (A) The hydrophobic stacking of strand C in the dimer interface. Side chains of Met2669 pack against Gly2671 of the neighboring monomer, creating hydrophobic interactions. (B) Hydrogen bonding between the monomers at strand D.
drophobic residues that form the interface between the C strands (underlined in Figure 6B). This sequence motif forms a signature for dimerization and is not seen in other vertebrate filamin domains. We therefore propose that all vertebrate filamins have dimerization interfaces analogous to that seen in human filamin C domain 24. It appears that C. elegans filamin A (encoded in chromosome IV, wormbase gene ID Y66H1B.2) and Drosophila filamin (Cher) also share the same dimerization sequence signature in their most C-terminal domain (Figure 6B). However, other filamin-like proteins in C. elegans (encoded in the X chromosome, wormbase gene ID C23F12.1) and Drosophila (Jbug) seem not to have this signature in their last domain, nor do they resemble the Dictyostelium filamin domain 6 (data not shown). This suggests that the other filamin-like proteins in the worm and in the fly may not dimerize or that they have developed a different dimerization mechanism. They seem to also differ in the other aspects and, for example, contain three calponin homology domains in their N termini, while all the other filamins have only two that form the widely conserved actin binding site (Korenbaum and Rivero, 2002). Model of the Molecular Structure of Vertebrate Filamin Although a different interface is employed, the two chains in both the filamin C domain 24 dimer and in the Dictyostelium filamin domain 6 dimer are arranged antiparallel. In the case of the Dictyostelium structure, this arrangement was taken to indicate that the chains in the whole molecule were arranged antiparallel (McCoy et al., 1999; Popowicz et al., 2004), which was also consistent with a range of other data. However, although there are no data on the overall structure of vertebrate filamin C, electron micrographs of individual human filamin molecules from various tissues indicate that, in contrast to the Dictyostelium analog, the two chains in these vertebrate actin binding proteins often appear to be arranged parallel to one another, producing V-shaped profiles with varying angles between the two arms (Gorlin et al., 1990; Hartwig and Stossel, 1981; Stossel et al., 2001; Tyler et al., 1980), and indeed Gorlin
Figure 5. Analysis of Dimerization Mutants (A) CD spectra of filamin C domain 24 and two mutants used in the study. (B) Gel filtration of filamin C domain 24 wild-type (solid line), M2990D (dotted line), and Mutant20 (dashed line). All proteins were injected at 1 mg/ml.
et al. (1990) proposed that there may be interactions between several Ig domains in the different arms. How then can the antiparallel arrangement of the chains in the vertebrate filamin Ig domain 24 dimer be reconciled with the two chains in the rod region of this molecule being arranged parallel? We propose that one way to do this would be for the dyad relating the two domain 24 chains to be arranged parallel to the rod axis in the intact filamin molecule, as illustrated in Figure 7, rather than perpendicular to the rod region axis as proposed for the Dictyostelium filamin dimer (McCoy et al., 1999; Popowicz et al., 2004). Although further work to obtain crystals containing at least vertebrate domains 23 and 24 will be needed to resolve this controversy, this hypothesis would be consistent with the EM observations. Moreover, in the filamin C domain 24 dimer, the N-terminal residues localize at the edges of a rather flat surface along the dyad axis (Figure 3A). This surface of filamin has a large neutral area (Figure 6D, left) that would be ideally placed to interact with the adjacent domain 23. This surface seems to be predominantly
Structure 116
Figure 6. Comparisons and Surface Features of Filamins (A) Amino acid sequence alignment of human filamin C domain 24 with Dictyostelium filamin domains 4 and 5. The β strands of filamin C domain 24 are indicated above. (B) Amino acid sequence alignment of filamin domain 24 sequences from available vertebrate species as well as filamin A from C. elegans and Cher from Drosophila. Green indicates residues that differ in all vertebrate isoforms; violet, filamin A-specific residues; blue, filamin B-specific residues; red, filamin C-specific residues. The hydrophobic dimerization signature is underlined. (C) The accessible surface of the dimeric human filamin C domain 24 with the locations of isoform-specific substitutions colored as in (B) shown on three successive 90° rotations about the vertical. (D) Accessible surface of the dimeric human filamin C domain 24 colored according to the surface potential. The images have been rotated by 180°.
conserved between the vertebrate filamin isoforms (Figure 6C, left). The Dimer Dissociation Constant Is in the Micromolar Range Filamin C domain 24 can be dissociated by 0.8 M KCl (Himmel et al., 2003), raising the possibility that the dimerization could be regulated in vivo, possibly as a mechanism to control the actin crosslinking. We therefore used analytical ultracentrifugation to determine the dissociation constant of the filamin C domain 24 dimer. Sedimentation equilibrium data from three different concentrations (17.25 M, 8.63 M, and 4.31 M) could be fitted to the ideal monomer/dimer model with Kd = 2.5 M (Figure 8). The M2669D mutant had a significantly weaker Kd of 3 mM (data not shown), consistent with its inhibiting dimerization but not altering conformation dramatically. Consistent with the analytical ultracentrifugation data, the filamin C domain 24 eluted at lower apparent molecular weight in gel filtration at low micromolar concentrations (Figure 8C), indicating that a monomer-dimer equilibrium existed in these conditions. Each filamin chain has a single actin binding site, and so dimerization is necessary for the actin crosslinking function of filamins. During platelet activation, the actin
crosslinking function of filamin A is ablated by calpainmediated proteolysis that targets mainly the two hinge regions between domains 15 and 16 and 23 and 24 (Davies et al., 1978; Gorlin et al., 1990). The actin binding site of filamin resides in its N terminus, but many signaling proteins interact with the C terminus of filamins (reviewed in van der Flier and Sonnenberg, 2001). Our observation that the dissociation constant for the filamin domain 24 dimer is of the same order as its cellular concentration raises the possibility that these signaling proteins might regulate filamin activity by altering the monomer-dimer equilibrium. The filamin A concentration in the platelets is 10 M (Hartwig, 2002). If the fulllength filamin had a dissociation constant even close to the value of the individual domain, a monomer-dimer equilibrium may be present in cells. In such a case, the proteins interacting with the C terminus could affect the equilibrium by concentrating filamins at defined locations or by directly interfering with the dimerization interface. In their early studies, Hartwig and Stossel (1981) performed analytical ultracentrifugation measurement between 0.7 to 2 M (monomer concentrations) of macrophage filamin but did not detect dissociation of the dimer. Thus, it is possible that the dimer affinity of the intact filamin is higher than that of domain 24 alone, which would be consistent with there being
Dimerization of Human Filamin C 117
Figure 7. A Model of the Orientation of Full Filamin in Relation to the 24th Domain Based on our current crystallographic data and the previously published EM pictures, we propose that the two-fold symmetry axis (inset) of the filamin domain 24 dimer is oriented parallel to the long axis of the V-shaped filamin dimer (full picture). We also propose that the Ig domain 23 and possibly also parts of the long linker would be able to interact with domain 24. Clearly, the observed total length of filamin dimer (160 nm) is too short to allow straight pearls-in-a-string arrangement of 48 Ig-like domains (2 × 24 × 4.5 nm = 216 nm). Thus, it is likely that the Ig domains are inclined relative to one another, as proposed by Fucini et al. (1999).
additional interactions between other Ig domains in the rod, as suggested by Gorlin et al. (1990). Further work will be required to test the hypothesis of monomer dimer equilibrium of filamins in vivo. In summary, we have determined the crystal structure of the vertebrate filamin Ig-like domain 24 that mediates dimerization, and we have shown that, although its structure resembles domains 4 and 5 of the Dictyostelium filamin analog, human filamin Ig domain 24 has a novel dimerization interface. The filamin C domain 24 dimer can be fitted to a structure of parallel filamin chains by aligning the dimer dyad axis to the rod axis. However, our data do not exclude the possibility of antiparallel alignment, as in the case of Dictyostelium filamin. Moreover, the dissociation constant for the domain 24 dimer is in the micromolar range, which is the order of the cellular filamin concentration. This raises the possibility that regulation of the monomer-dimer equilibrium could be important functionally. Experimental Procedures Protein Expression, Purification, and Crystallization The fragment encoding the 24th Ig-like domain of human filamin C (residues 2633–2725, Swiss Prot accession number Q14315) was
amplified from a human skeletal muscle cDNA library (Matchmaker, Clontech, Invitrogen, Palo Alto, CA). The fragment was cloned in NcoI and NotI sites of a modified pETM24d (Novagen) plasmid containing an N-terminal His6 tag followed by the cleavage sequence for tobacco etch virus (Tev) protease ENLYFQ*GAMG (Parks et al., 1994) (The asterisk denotes the cleavage site, and the extra residues left to the recombinant protein are underlined). The insert was verified by DNA sequencing. Protein expression was induced in Escherichia coli BL21 (DE3) for 3 hr at 37°C in the presence of 0.4 mM IPTG. The protein was purified on a NiNTA agarose column (Qiagen) according to manufacturer's instructions and eluted with 250 mM imidazole. Buffer was changed with PD10 columns (Amersham Biosciences) to 20 mM Tris (pH 8.0), and the protein was cleaved with Tev protease at 30°C overnight. Further purification was achieved by cation exchange chromatography using Fractogel EMD SO3− (S) matrix (Merck). The wild-type protein was concentrated to 40–42 mg/ml in 20 mM Tris (pH 8.0). Crystals were grown by hanging drop vapor diffusion after mixing the protein in a 1:1 volume ratio with the reservoir solution of 1.6 M Na citrate, 0.1 M HEPES (pH 7.5). To produce Se-Met-labeled protein, expression was induced in the Met-auxotrophic E. coli strain B834(DE3) (Leahy et al., 1992; Wood, 1966) in minimal media containing 0.5 mM L-SeMet (Calbiochem) for 20 hr at 20°C. The protein was purified and crystallized as above. Two mutant constructs of the filamin C domain 24 were produced. In the first mutant, Met2669 was changed to Asp (M2669D). In the second one, called mutant 20, a stretch of eight residues (aa 2666–2673, NMMMVGVH) was substituted with the string from human filamin A domain 20 (QDMTAQVT). The mutations were made with the Quikchange mutagenesis kit (Stratagene). Protein purification was achieved as described above, followed by gel filtration on a Superdex-75 16/60 column (Amersham Biotech). The mutant proteins were concentrated to 8–10 mg/ml. Data Collection, Structure Determination, and Refinement For data collection, the crystals were immersed in reservoir solution containing 10% glycerol and flash frozen at 100 K. The data sets were collected at the EMBL beamlines X13 and BW7A at the DORIS storage ring, DESY, Hamburg, using the 165 mm Mar CCD detector (MARData) and were processed and scaled using XDS program package (Kabsch, 1993). CNS 1.0 (Brunger et al., 1998) was used to locate the Se atoms and to generate phases using single-wavelength anomalous diffraction from the Se peak wavelength and also for density modification. The electron density map produced could be used for automatic model building in Arp/Warp 6.0 (Perrakis et al., 1999). The initial model containing 86 residues was used to solve the native data set. Further refinement was done with programs O (Jones et al., 1991) and Refmac5 (Murshudov et al., 1997). The final model contains 96 of 97 residues present in the construct. The first Gly residue derived from the plasmid sequence could not be seen in the electron density. Double atomic conformations were found for residues Cys 2679 and Asn 2689. Those were refined using SHELXPRO (Sheldrick and Gould, 1995) and Refmac5. For the final model, a restrained isotropic refinement was performed using Refmac5 to 1.43 Å with weight matrix 0.4. The Ramachandran plot of the final model shows 93.6% of residues in the most favored regions and 6.4% in additional allowed regions. Structure figures were generated with Pymol (DeLano Scientific LLC, San Carlos, CA) and Grasp (Nicholls et al., 1991). Sequence alignment has been generated with ClustalX (1.8) (Thompson et al., 1997) with protein weight matrix Gonnet PAM250. The dimer interaction was analyzed by using the server http://www.biochem. ucl.ac.uk/bsm/PP/server (Jones and Thornton, 1996). Superimposition of the domains was performed in program O (Jones et al., 1991). Gel Filtration, Analytical Ultracentrifugation, and Circular Dichroism Analytical gel filtration was performed at 23°C in a Superdex 75 10/ 30 column (Amersham Biotech) in 150 mM KCl, 50 mM Na-phosphate (pH 7.5) (Figure 5B) or 150 mM KCl, 20 mM Tris, 1 mM DTT (pH 8.0) (Figure 8C), at a flow rate of 0.5 ml/min. The injection volume was 100 l, and protein concentrations are indicated in the figure legends.
Structure 118
Figure 8. Analysis of the Dimerization of Filamin C Domain 24 (A) Sedimentation equilibrium analysis was performed by analytical ultracentrifugation. Plots of Mw, app against concentration are shown for three concentrations, together with the theoretical curve corresponding to the ideal monomer-dimer equilibrium with parameters derived for the fitting. (B) Corresponding plots showing residuals against radius for the fitting data to the experimental data. Concentrations used were 17.25 M (top), 8.63 M (middle), and 4.31 M (bottom). (C) Analytical gel filtration results after injecting 102 M (solid line), 6.3 M (dashed line), or 1.6 M (dotted line) of filamin C domain 24. The estimated protein concentrations based on absorbance at 280 nm at the peak were 32 M, 1.2 M, and 0.13 M, respectively. To allow direct comparison, the chromatograms were scaled to the peak high and absorbance at 280 nm of the 102 m injection, and absorbance at 215 nm of the 6.3 M and 1.6 M injections are shown.
Sedimentation equilibrium analytical ultracentrifugation was performed with an Optima XLA (Beckman Coulter) instrument using An-60Ti rotor at +4°C and scanning wavelength of 229 nm. Filamin C domain 24 protein sample was diluted in 140 mM NaCl, 10 mM Na-phosphate (pH 7.4). Four hundred microliters of the samples and 420 l of the buffer for reference were loaded in the analytical ultracentrifuge cells. After the initial scan, the centrifuge was overspeeded at 30,000 rev/min to reduce the time taken to reach the equilibrium (Van Holde and Baldwin, 1958). Then, the speed was reduced to 20,000 rev/min, and scans were taken at intervals of 24 hr. The equilibrium was determined by the criteria that successive scans were indistinguishable, and at this point scans averaging over 100 readings were taken for analysis. Subsequently, the centrifuge was overspeeded to sediment the macromolecules away from the meniscus, before slowing to equilibrium speed and taking a further scan to establish the baseline absorbance for each cell. The calculations were performed as in Mills et al., 2003. CD measurements were performed in the far UV region (190–250 nm) at 20°C using a Jasco J-715 spectropolarimeter. Protein concentrations were in the range of 0.22–0.23 mg/ml in 5 mM Na phosphate (pH 7.1). Spectra were obtained with a scan rate of 50 nm/ min in a cuvette with 0.1 cm optical path length. The data were collected at 0.2 nm intervals, and the average of 16 scans is reported.
Acknowledgments We thank the EMBL-Hamburg beamline staff, particularly Andrea Schmidt and Mannfred Weiss, for help during the data collection, David Owen for amino acid analysis of the proteins, and Ulrich Bergmann for MALDI-TOF mass spectrometry. This study has been supported by the Academy of Finland grants 51863, 207021, 105211, and by EMBO short term travel fellowship to J.Y. Received: August 26, 2004 Revised: October 29, 2004 Accepted: October 29, 2004 Published: January 11, 2005
Bellanger, J.M., Astier, C., Sardet, C., Ohta, Y., Stossel, T.P., and Debant, A. (2000). The Rac1- and RhoG-specific GEF domain of Trio targets filamin to remodel cytoskeletal actin. Nat. Cell Biol. 2, 888–892. Brunger, A.T., Adams, P.D., Clore, G.M., DeLano, W.L., Gros, P., Grosse-Kunstleve, R.W., Jiang, J.S., Kuszewski, J., Nilges, M., Pannu, N.S., et al. (1998). Crystallography & NMR system: A new software suite for macromolecular structure determination. Acta Crystallogr. D Biol. Crystallogr. 54, 905–921. Cunningham, C.C., Gorlin, J.B., Kwiatkowski, D.J., Hartwig, J.H., Janmey, P.A., Byers, H.R., and Stossel, T.P. (1992). Actin-binding protein requirement for cortical stability and efficient locomotion. Science 255, 325–327. Davies, P.J., Wallach, D., Willingham, M.C., Pastan, I., Yamaguchi, M., and Robson, R.M. (1978). Filamin-actin interaction. Dissociation of binding from gelation by Ca2+-activated proteolysis. J. Biol. Chem. 253, 4036–4042. Faulkner, G., Lanfranchi, G., and Valle, G. (2001). Telethonin and other new proteins of the Z-disc of skeletal muscle. IUBMB Life 51, 275–282. Fox, J.W., Lamperti, E.D., Eksioglu, Y.Z., Hong, S.E., Feng, Y., Graham, D.A., Scheffer, I.E., Dobyns, W.B., Hirsch, B.A., Radtke, R.A., et al. (1998). Mutations in filamin 1 prevent migration of cerebral cortical neurons in human periventricular heterotopia. Neuron 21, 1315–1325. Fucini, P., Renner, C., Herberhold, C., Noegel, A.A., and Holak, T.A. (1997). The repeating segments of the F-actin cross-linking gelation factor (ABP-120) have an immunoglobulin-like fold. Nat. Struct. Biol. 4, 223–230. Fucini, P., Koppel, B., Schleicher, M., Lustig, A., Holak, T.A., Muller, R., Stewart, M., and Noegel, A.A. (1999). Molecular architecture of the rod domain of the Dictyostelium gelation factor (ABP120). J. Mol. Biol. 291, 1017–1023. Gorlin, J.B., Yamin, R., Egan, S., Stewart, M., Stossel, T.P., Kwiatkowski, D.J., and Hartwig, J.H. (1990). Human endothelial actinbinding protein (ABP-280, nonmuscle filamin)—a molecular leaf spring. J. Cell Biol. 111, 1089–1105.
References
Hartwig, J.H. (2002). Platelet structure. In Platelets, A.D. Michelson, ed. (New York: Academic Press), pp. 35–50.
Ayscough, K.R. (1998). In vivo functions of actin-binding proteins. Curr. Opin. Cell Biol. 10, 102–111.
Hartwig, J.H., and Stossel, T.P. (1981). Structure of macrophage actin-binding protein molecules in solution and interacting with actin filaments. J. Mol. Biol. 145, 563–581.
Dimerization of Human Filamin C 119
Himmel, M., Van Der Ven, P.F., Stocklein, W., and Furst, D.O. (2003). The limits of promiscuity: isoform-specific dimerization of filamins. Biochemistry 42, 430–439. Jones, S., and Thornton, J.M. (1996). Principles of protein-protein interactions. Proc. Natl. Acad. Sci. USA 93, 13–20. Jones, T.A., Zou, J.Y., Cowan, S.W., and Kjeldgaard (1991). Improved methods for building protein models in electron density maps and the location of errors in these models. Acta Crystallogr. A 47, 110–119.
S.M., Duncan, J.S., Dubeau, F., Scheffer, I.E., Schachter, S.C., et al. (2001). Mutations in the X-linked filamin 1 gene cause periventricular nodular heterotopia in males as well as in females. Hum. Mol. Genet. 10, 1775–1783. Sheldrick, G.M., and Gould, R.O. (1995). Structure solution by iterative peaklist optimization and tangent expansion in space group P1. Acta Crystallogr. B 51, 423–431. Small, J.V., Stradal, T., Vignal, E., and Rottner, K. (2002). The lamellipodium: where motility begins. Trends Cell Biol. 12, 112–120.
Kabsch, W. (1993). Automatic processing of rotation diffraction data from crystals of initially unknown symmetry and cell constants. J. Appl. Crystallogr. 26, 795–800.
Stossel, T.P., Condeelis, J., Cooley, L., Hartwig, J.H., Noegel, A., Schleicher, M., and Shapiro, S.S. (2001). Filamins as integrators of cell mechanics and signalling. Nat. Rev. Mol. Cell Biol. 2, 138–145.
Kakita, A., Hayashi, S., Moro, F., Guerrini, R., Ozawa, T., Ono, K., Kameyama, S., Walsh, C.A., and Takahashi, H. (2002). Bilateral periventricular nodular heterotopia due to filamin 1 gene mutation: widespread glomeruloid microvascular anomaly and dysplastic cytoarchitecture in the cerebral cortex. Acta Neuropathol. (Berl.) 104, 649–657.
Thompson, J.D., Gibson, T.J., Plewniak, F., Jeanmougin, F., and Higgins, D.G. (1997). The CLUSTAL_X windows interface: flexible strategies for multiple sequence alignment aided by quality analysis tools. Nucleic Acids Res. 25, 4876–4882.
Korenbaum, E., and Rivero, F. (2002). Calponin homology domains at a glance. J. Cell Sci. 115, 3543–3545. Leahy, D.J., Hendrickson, W.A., Aukhil, I., and Erickson, H.P. (1992). Structure of a fibronectin type III domain from tenascin phased by MAD analysis of the selenomethionyl protein. Science 258, 987– 991. Loo, D.T., Kanner, S.B., and Aruffo, A. (1998). Filamin binds to the cytoplasmic domain of the beta1-integrin. Identification of amino acids responsible for this interaction. J. Biol. Chem. 273, 23304– 23312. McCoy, A.J., Fucini, P., Noegel, A.A., and Stewart, M. (1999). Structural basis for dimerization of the Dictyostelium gelation factor (ABP120) rod. Nat. Struct. Biol. 6, 836–841. McGough, A. (1998). F-actin-binding proteins. Curr. Opin. Struct. Biol. 8, 166–176. Mills, I.G., Praefcke, G.J., Vallis, Y., Peter, B.J., Olesen, L.E., Gallop, J.L., Butler, P.J., Evans, P.R., and McMahon, H.T. (2003). EpsinR: an AP1/clathrin interacting protein involved in vesicle trafficking. J. Cell Biol. 160, 213–222. Murshudov, G.N., Vagin, A.A., and Dodson, E.J. (1997). Refinement of macromolecular structures by the maximum-likelihood method. Acta Crystallogr. D Biol. Crystallogr. 53, 240–255. Murzin, A.G., Brenner, S.E., Hubbard, T., and Chothia, C. (1995). SCOP: a structural classification of proteins database for the investigation of sequences and structures. J. Mol. Biol. 247, 536–540. Nicholls, A., Sharp, K.A., and Honig, B. (1991). Protein folding and association: insights from the interfacial and thermodynamic properties of hydrocarbons. Proteins 11, 281–296. Ohta, Y., Suzuki, N., Nakamura, S., Hartwig, J.H., and Stossel, T.P. (1999). The small GTPase RalA targets filamin to induce filopodia. Proc. Natl. Acad. Sci. USA 96, 2122–2128. Ott, I., Fischer, E.G., Miyagi, Y., Mueller, B.M., and Ruf, W. (1998). A role for tissue factor in cell adhesion and migration mediated by interaction with actin-binding protein 280. J. Cell Biol. 140, 1241– 1253. Parks, T.D., Leuther, K.K., Howard, E.D., Johnston, S.A., and Dougherty, W.G. (1994). Release of proteins and peptides from fusion proteins using a recombinant plant virus proteinase. Anal. Biochem. 216, 413–417. Perrakis, A., Morris, R., and Lamzin, V.S. (1999). Automated protein model building combined with iterative structure refinement. Nat. Struct. Biol. 6, 458–463. Popowicz, G.M., Muller, R., Noegel, A.A., Schleicher, M., Huber, R., and Holak, T.A. (2004). Molecular structure of the rod domain of dictyostelium filamin. J. Mol. Biol. 342, 1637–1646. Puius, Y.A., Mahoney, N.M., and Almo, S.C. (1998). The modular structure of actin-regulatory proteins. Curr. Opin. Cell Biol. 10, 23–34. Radaev, S., Motyka, S., Fridman, W.H., Sautes-Fridman, C., and Sun, P.D. (2001). The structure of a human type III Fcgamma receptor in complex with Fc. J. Biol. Chem. 276, 16469–16477. Sheen, V.L., Dixon, P.H., Fox, J.W., Hong, S.E., Kinton, L., Sisodiya,
Thompson, T.G., Chan, Y.M., Hack, A.A., Brosius, M., Rajala, M., Lidov, H.G., McNally, E.M., Watkins, S., and Kunkel, L.M. (2000). Filamin 2 (FLN2): A muscle-specific sarcoglycan interacting protein. J. Cell Biol. 148, 115–126. Tyler, J.M., Anderson, J.M., and Branton, D. (1980). Structural comparison of several actin-binding macromolecules. J. Cell Biol. 85, 489–495. van der Flier, A., and Sonnenberg, A. (2001). Structural and functional aspects of filamins. Biochim. Biophys. Acta 1538, 99–117. Van Holde, K.E., and Baldwin, R.L. (1958). Rapid attainment of sedimentation equilibrium. J. Phys. Chem. 62, 734–743. Van Troys, M., Vandekerckhove, J., and Ampe, C. (1999). Structural modules in actin-binding proteins: towards a new classification. Biochim. Biophys. Acta 1448, 323–348. Wood, W.B. (1966). Host specificity of DNA produced by Escherichia coli: bacterial mutations affecting the restriction and modification of DNA. J. Mol. Biol. 16, 118–133. Zenker, M., Rauch, A., Winterpacht, A., Tagariello, A., Kraus, C., Rupprecht, T., Sticht, H., Reis, A., Robertson, S.P., Twigg, S.R., et al. (2004). A dual phenotype of periventricular nodular heterotopia and frontometaphyseal dysplasia in one patient caused by a single FLNA mutation leading to two functionally different aberrant transcripts. Am. J. Hum. Genet. 74, 731–737. Accession Numbers The coordinates of the structure have been deposited in the Protein Data Bank with ID code 1V05.