Ca2+ and calmodulin regulate the binding of filamin A to actin filaments.

Filamin A (FLNa) cross-links actin filaments (F-actin) into three-dimensional gels in cells, attaches F-actin to membrane proteins, and is a scaffold that collects numerous and diverse proteins. We report that Ca(2+)-calmodulin binds the actin-binding domain (ABD) of FLNa and dissociates FLNa from F-actin, thereby dissolving FLNa.F-actin gels. The FLNa ABD has two calponin homology domains (CH1 and CH2) separated by a linker. Recombinant CH1 but neither FLNa nor its ABD binds Ca(2+)-calmodulin in the absence of F-actin. Extending recombinant CH1 to include the negatively charged region linker domain makes it, like full-length FLNa, unable to bind Ca(2+)-calmodulin. Ca(2+)-calmodulin does, however, dissociate the FLNa ABD from F-actin provided that the CH2 domain is present. These findings identify the first evidence for direct regulation of FLNa, implicating a mechanism whereby Ca(2+)-calmodulin selectively targets the FLNa.F-actin complex.

Filamin A (FLNa) 2 is the dominant mammalian nonmuscle isoform of a widely distributed family of actin-binding proteins. FLNa, originally named "actin-binding protein" or ABP-280, is the most potent actin filament (F-actin) cross-linking protein known, and its power resides in its ability to promote perpendicular branching of actin filaments into three-dimensional gelled networks. The physiological relevance of this actin gelation property of FLNa is manifest by the surface instability and the lack of translational locomotion capacity of cells lacking the protein.
FLNa is also a scaffold for numerous binding partners that include transmembrane proteins, components of intracellular signal transduction pathways, and other cellular constituents (1,2).
The structural requirements for actin cross-linking by FLNa include dimerization at the C terminus of its subunits and a stretch of 278 amino acids designated the actin-binding domain (ABD) at the N-terminal of each dimeric subunit. The potency of the FLNa F-actin cross-linking activity resides in the large size (280 kDa) of the subunits composed of a string of 24 ␤-pleated sheet repeats, because much higher concentrations of other dimeric F-actin cross-linking proteins with nearly identical ABDs and binding affinities for F-actin are necessary to create gels of actin filaments of identical length.
Although regulation of FLNa dimerization, the ABD, or the structure of the intervening ␤-sheet repeats could affect cross-linking of the F-actin of FLNa, purified FLNa constitutively cross-links actin filaments in vitro, and no direct regulation of this activity has emerged in 30 years of research on this protein. Therefore, the working hypothesis for cellular control of actin cross-linking by FLNa has been that FLNa binding partners such as Rac, Rho, Cdc42, and other signaling components produce linear actin assembly on the FLNa template that orients the filaments formed into a three-dimensional cross-linked gel. Actin filament-severing proteins such as members of the gelsolin and actin-depolymerizing factor/cofilin family could dismantle such actin networks (3,4). However, although plausible, this mechanism does not explain why a large fraction of cellular FLNa is free of F-actin in vivo. In this paper we present the first evidence for a direct regulation of the FLNa⅐F-actin interaction by the ubiquitous abundant calcium-binding protein, calmodulin (CaM).

MATERIALS AND METHODS
Proteins-Rabbit skeletal actin and human recombinant FLNa were prepared as described previously (5). HA-CaM was expressed and purified as described (6). Recombinant CaM was cloned from an IMAGE human cDNA clone (source identifier 4801400) into pET-23a(ϩ) (EMB Biosciences, Inc.), expressed in Escherichia coli BL21(DE3), and purified as described (7). Bovine CaM was purchased from Sigma (catalog number P1431). Fragments of human FLNa were cloned by PCR and expressed as GST⅐His 6 or maltose-binding protein⅐His 6 -tagged fusion proteins in E. coli using a modified pGEX-4T1-HT(a, b, or c) or pMALc-HT(a, b, or c) plasmid, respectively. The pGEX-4T1-HT(a, b, or c) were made by inserting the BglII/NotI fragment of a PCR product amplified from pFASTBAC-HT(a, b, or c) (Invitrogen) with two primers, 5Ј-GAAGATCTATGTCGTACTACCATCAC-3Ј and 5Ј-CTAG-TACTTCTCGACAAGCTTGG-3Ј, into BamHI/NotI sites of pGEX-4T1 (Amersham Biosciences). The pMALc-HT(a, b, or c) were made using similar procedures but inserted into BamHI/XbaI sites of pMALc (New England BioLabs). An acidic sequence of FLNa, 156 DEEEDEE 162 , was deleted using the QuikChange site-directed mutagenesis kit (Stratagene). The fusion proteins were affinity purified using glutathione-Sepharose 4 Fast Flow, Amylose resin or nickel-nitrilotriacetic acid agarose. For some experiments, the tags were cleaved with His 6 ⅐tobacco etch virus (TEV) protease at room temperature for 1 h. GST⅐His 6 tag, maltose-binding protein⅐His 6 tag, and His 6 ⅐TEV were removed by passing the solution supplemented with 3 mM MgCl 2 and 5 mM imidazole through a nickel-nitrilotriacetic acid agarose column. The flow-through fractions were concentrated using an Amicon Ultra-15 (Millipore) and applied onto a Superdex 75 HR 10/30 (Amersham Biosciences) equilibrated with 10 mM Tris-HCl, 100 mM NaCl, and 1 mM dithiothreitol, pH 7.4. The highly purified proteins were stored at 4°C.
Antibodies-Anti-FLNa monoclonal antibodies were prepared by hybridoma techniques. Approximately 15-20 g of purified recombi-nant Homo sapiens FLNa proteins, dissolved in phosphate-buffered saline, were mixed with complete Freund's adjuvant and injected intraperitoneally to BALB/c mice. The spleen of the immunized mouse was macerated, and the spleen cells were fused with mouse myeloma Sp20 cells in the presence of a Hybrimax polyethylene glycol solution (Sigma-Aldrich). The fused cells were plated onto 96-well plates and cultured in Optimum medium. The standard hypoxanthine-aminopterin-thymidine selection procedure was applied to select the hybridoma clones. After 10 days, hybridoma supernatants were tested for reactivity with recombinant H. sapiens FLNa by an enzyme-linked immunosorbent assay. To further select antibodies, the media were also tested by immunoblotting and immunostaining. Cells from the positive wells were cloned at least three times by limiting dilution and adapted to the serum-free medium Hybridoma-SFM (Invitrogen). The antibody subtypes were determined using a kit (Roche Applied Science). Monoclonal IgG or IgM antibodies were purified from serum-free hybridoma supernatant by GammaBind G-Sepharose (Amersham Biosciences) or KAP-TIV-M (Genomics One International, Buffalo, NY) affinity chromatography. Anti-HA mouse mAb (clone 12CA), anti-GST rabbit polyclonal antibody (Sigma), goat horseradish peroxidase-conjugated anti-mouse, or anti-rabbit IgG (Bio-Rad) were used for immunoblotting.
Actin Filament Sedimentation Assay-Purified FLNa or FLNa fragments were mixed with or without CaM and G-actin (5 or 10 M) in solution F (20 mM Tris-HCl, pH 7.4, 0.5 mM Na 2 ATP, 5 mM MgCl 2 , 120 mM NaCl, and 0.2 mM dithiothreitol) with 2 mM CaCl 2 or 5 mM EGTA for 1 h at 25°C. The actin filaments were then sedimented by centrifugation at 150,000 ϫ g for 20 min at 25°C. Proteins in the supernatants and pellets were then solubilized in SDS gel sample buffer and subjected to SDS-PAGE. Polypeptides in the gel were visualized by Coomassie Brilliant Blue (CBB) staining or immunoblotting.
Measurements of Apparent Viscosity and Gelation-Apparent viscosity and gelation were determined using a miniature falling ball viscometer as described previously (5).
HA-Calmodulin Blot Overlay Assay-A CaM overlay assay was performed in the presence of either 2 mM CaCl 2 or 5 mM EGTA as described by Szymanska et al. with slight modifications (6). Purified FLNa or FLNa constructs (2 pmol) were spotted onto polyvinylidene difluoride membranes and blocked for 1 h at room temperature in 3% bovine serum albumin in TTBS (20 mM Tris-HCl, pH 7.4, 110 mM NaCl, 5 mM MgCl 2 , and 0.1% Tween 20). In some experiments, the blots were soaked in methanol/acetic acid/water (50:10:40) (v/v) for 10 s followed by methanol for 10 s to denature proteins. The membranes were then rinsed with water before blocking. The blots were probed with 5 g/ml HA-CaM in TTBS for 1 h. The blot was washed three times (10 min/ wash) with TBST, and bound HA-CaM was detected with an anti-HA mAb and horseradish peroxidase-conjugated anti-mouse IgG.
HA-Calmodulin Pull-down Assay-FLNa or GST⅐FLNa fragments (0.1 M) were incubated for 1 h at room temperature with 20 l (50% slurry) of CaM-agarose (Sigma) or 0.2 M HA-CaM in 200 l of TTBS in the presence of either 2 mM CaCl 2 or 5 mM EGTA, and then 20 l (50% slurry) of anti-HA mAb-agarose beads (Santa Cruz Biotechnology Inc.) was added. After 1 h, the slurry was centrifuged and washed four times with TTBS containing either CaCl 2 or EGTA. The polypeptides bound to the beads were solubilized in SDS sample buffer and analyzed by SDS-PAGE and immunoblot.
Measurements of Affinity to Actin Filaments-Affinity to actin filaments was measured by a sedimentation assay. Five or 10 M G-actin was incubated with increasing concentrations of FLNa fragments in a solution containing 20 mM Tris-HCl, pH 7.4, 0.5 mM Na 2 ATP, 2 mM MgCl 2 , 120 mM NaCl, 0.2 mM dithiothreitol, and 0.5 mM EGTA for 2 h at 25°C. The actin filaments were then sedimented by centrifugation at 150,000 ϫ g for 20 min at 25°C. Proteins in the supernatants and pellets were then solubilized in SDS gel sample buffer and subjected to SDS-PAGE. Polypeptides in the gel were visualized by CBB staining, and the band intensities were quantified by densitometry.
Titration of Free Calcium Ion-Ca 2ϩ -EGTA solutions were prepared as described previously (8). The concentration of free calcium was determined using Fura-2 and calcium calibration buffer kit in accordance with the manufacturer's instructions (Molecular Probes).
Data Base Analysis-Secondary or tertiary structures of FLNa were predicted by data base analysis available on web sites. For homology searches, protein-protein BLAST data base was used. For secondary structure prediction, the sites used were those of the Pôle BioInformatique Lyonnais network protein sequence analysis server (npsapbil.ibcp.fr) and the PredictProtein server of the Columbia University Bioinformatics Center (cubic.bioc.columbia.edu/predictprotein/). For three-dimensional structure prediction, the SWISS-MODEL automated comparative protein modelling server (swissmodel.expasy.org//SWISS-MODEL) and the 3D-Jigsaw comparative modelling server (www.bmm. icnet.uk/servers/3djigsaw/) were used. To predict the CaM-binding site, the Ontario Cancer Institute's calmodulin target data base (calcium. uhnres.utoronto.ca/ctdb/flash.htm) was examined. The protein charge plot was carried out by the EMBOSS charge program (emboss.sourceforge. net/). To access these sites, include http:// at the beginning of each address. Fig. 1A shows the apparent viscosity of 24 M actin polymerized with various concentrations of FLNa in the presence either EGTA or Ca 2ϩ with or without 10 M CaM. In EGTA the gel point, an abrupt increase in the viscosity of actin, occurred at a FLNa concentration of 0.03 M, a ratio of ϳ1 FLNa to 800 actin subunits as described previously (5), and the gel point was unaffected by Ca 2ϩ -free CaM. The amount of FLNa required to gel actin in the presence of 2 mM calcium was slightly higher, presumably because actin filaments polymerized in calcium are shorter than the filaments containing Mg 2ϩ (9). However, CaM nearly doubled the FLNa concentration required for incipient gelation in the presence of Ca 2ϩ . Ca 2ϩ -CaM Inhibits the Binding of FLNa to F-actin-The simplest explanation for inhibition of FLNa-mediated actin gelation by Ca 2ϩ -CaM could be interference with the binding of FLNa to actin filaments, and Fig. 2 shows the evidence for this mechanism. Approximately 80% of added FLNa co-sedimented with actin filaments in the presence or absence of free Ca 2ϩ ( Fig. 2A). CaM reduced the binding of FLNa to actin filaments in the presence of Ca 2ϩ . CaM did not affect the solubility of FLNa in the absence of F-actin. Increasing concentrations of CaM inhibited the co-sedimentation of FLNa with actin filaments, but only in the presence of Ca 2ϩ (Fig. 2B). Fig. 2C shows that the half-maximal inhibition of the binding of FLNa to actin filaments occurred at 40 M CaM in the presence of Ca 2ϩ . More than 76% of the added FLNa bound to actin filaments in the presence of the same concentration of CaM in the absence of Ca 2ϩ .

Ca 2ϩ -CaM Inhibits the Gelation of Actin by FLNa-
CaM Does Not Bind Native Full-length FLNa but Binds the ABD of Denatured FLNa-The results of both HA-CaM pull-down and HA-CaM overlay assays were inconsistent with a mechanism whereby Ca 2ϩ -CaM inhibits F-actin binding by competitive ligation to native FLNa, because the binding of recombinant FLNa to CaM was minimal (Fig. 3, A and B). FLNa partially denatured with methanol/acetic acid, however, bound CaM in a calcium-dependent fashion. An identically treated FLNa truncate lacking its 275 N-terminal amino acids comprising the ABD, which did not bind F-actin (data not shown), did not bind CaM (Fig. 3B), suggesting that binding occurs with first 275 amino acids of FLNa.
Localizing the CaM-binding in the N-terminal Half of the FLNa ABD (N-ABD)- Fig. 4 shows that the a GST⅐His 6 construct encompassing the N-terminal half of the FLNa ABD, N-ABD, binds to CaM in a Ca 2ϩdependent manner, whereas a corresponding construct containing the ABD C-terminal half (C-ABD) does not. A GST fusion protein including the entire ABD, like full-length FLNa, did not bind to CaM. The ABD of FLNa contains two calponin homology (CH) domains, CH1 (Ile 44 -Tyr 148 ) and CH2 (Pro 168 -Phe 264 ), which each contain four principal ␣ helices connected by loops or shorter helices (10). Based on these predictions and because many known CaM-binding proteins possess a relatively short ␣ helical region, fusion proteins representing smaller fragments of the FLNa ABD illustrated in Fig. 4, C and E, were generated and tested for their CaM-binding activity in the presence of EGTA or Ca 2ϩ (Fig. 4D). As shown in Fig. 4E, the minimum CaM binding site was narrowed down to a region between Thr 50 and Arg 96 within the CH1 domain. Removal of Thr 50 -Val 60 from this sequence completely eliminated the CaM-binding activity of this fragment, and a shorter fragment containing Thr 50 -Val 60 (Thr 31 -Ala 65 ) did not bind to CaM either. Removal of a highly positively charged sequence, 87 KKMHRKHNQR 96 from the putative CaM-binding fragment diminished its CaM-binding activity, and smaller fragments containing this region (Ser 61 -Ser 118 or   Ser 61 -Arg 96 ) did not bind to CaM. These data indicate that the Thr 50 -Val 60 stretch is essential for CaM-binding activity and that the basic region between positions 87 and 96 augments the binding, but also that each segment alone is not sufficient to promote binding. This CaMbinding site of FLNa is apparently unique in that no matches of its sequence to the calmodulin target data base (11) were detectable.
A Negatively Charged Site Linking the CH1 and CH2 Domains of the FLNa ABD Is Sufficient to Suppress the CaM Binding Activity-Our findings suggest that the C-terminal portion of the FLNa ABD is responsible for the inability of the intact FLNa ABD to bind CaM. The connecting segment between CH1 and CH2 ( 156 DEEEDEE 162 ), which is filamin family-specific, is highly acidic (Fig. 5A). This negatively charged sequence could complement the positively charged region (residues 87-96 amino acids) within the CaM-binding site in the N-ABD and might affect CaM-binding activity. Consistent with this hypothesis, the N-ABD extended by 10 C-terminal amino acids, including all of the positively charged residues, is unable to bind CaM, and deletion of the charged residues from the entire FLNa ABD enables it to bind to CaM (Fig. 5B). Fig. 5C shows that expressed FLNa fragments containing the entire ABD (1-278 amino acids), the N-ABD (1-153 amino acids) and the N-ABD plus the acidic link between CH1 and CH2 (1-163 amino acids) associate with actin filaments, whereas the C-ABD (154 -278 amino acids) does not. The results of the binding experiments also demon-

Filamin A Regulation by Calmodulin
strate that the N-ABD is sufficient to regulate actin-binding by CaM. The positively charged CH1-CH2 link desensitizes the N-ABD regulation by CaM; the entire ABD is responsive to CaM, suggesting that the CH2 sequence restores this sensitivity.
CaM Interacts with FLNa at Physiological Ca 2ϩ Concentrations- Fig.  6 compares the calcium dose-response for CaM binding to GST⅐1-278⌬DE ( 156 DEEEDEE 162 deleted) with that required to solate FLNa⅐F-actin⅐CaM networks. GST⅐1-278⌬DE was used in these experiments because it binds calmodulin in the absence of actin. Binding of CaM to GST⅐1-278⌬DE required Ն1 M calcium and was nearly maximal at 10 M calcium concentrations. The apparent viscosity of F-actin in the presence of FLNa and CaM was significantly decreased in this calcium range. Interpretations of effects at calcium are somewhat complicated because the assembly of actin filaments in Ͼ50 M calcium increases the number of nuclei, leading to shorter filaments (9). The apparent viscosity of FLNa⅐F-actin, however, was clearly reduced in the presence of CaM.
The CH2-ABD Has Low Affinity for Actin Filaments but Augments the Binding of the Intact ABD-Qualitative binding studies suggest that both the intact ABD and the CH1-ABD of FLNa bind to actin filaments, whereas the CH2-ABD binds actin weakly. We measured the affinity of the intact ABD and CH2-ABD to actin filaments. These domains were expressed as fusion proteins in E. coli and purified, and the GST⅐His 6 was removed. The ABD and the C-ABD were obtained in high yield and, as monomers, in concentrations up to 300 M. We could not produce N-ABD protein in good yield because of technical problems that have also been reported with attempts to isolate the corresponding domain of utrophin (12) . Fig. 7, A-C shows that the ABD of FLNa binds to actin filaments with a 1:1 stoichiometry (mol/mol actin as a monomer) and a K d of 17.3 M, whereas the CH2 did not demonstrate saturable binding to actin even at very high concentrations (ϳ160 M). Although the affinity of N-ABD to actin could not be measured, the GST⅐His 6 ⅐N-ABD sediments with actin filaments, whereas the GST⅐His 6 ⅐C-ABD does not (Fig. 5C), implying that the major ABD is located in the N-ABD. Because CaM binds to the N-ABD, the results suggest that Ca 2ϩ -CaM directly competes with actin filaments.

DISCUSSION
Phosphorylation of avian gizzard filamin (probably a FLNb homolog, inferring from its DNA sequence) by CaM-kinase II attenuates its actin binding (13). However, such an effect of CaM-kinase II on human FLNa has not been observed. 3 Here, we have obtained the first evidence for a direct regulation of the interaction between FLNa and actin filaments.   (30). Basic amino acids in positively charged site (blue line) and acidic amino acids in negatively charged site (red line) are marked with blue and red colors, respectively. ABS1, ABS2, and ABS3 are also delineated (green line). The alignment was generated using ClustalW (35). ddFLN, Dictyostelium discoideum FLN; dmFLN, Drosophila melanogaster FLN; ggFLN, Gallus gallus FLN. FLNa and inhibits the ability of FLNa to bind and cross-link actin filaments into a three-dimensional gelled network in vitro. If this regulation exists within living cells, calcium signaling would have a consistent effect on actin gelation. First, calcium signaling dissolves actin gels by activating actin filament severing and capping proteins of the gelsolin family (3). Second, FLNa mobilized by removal from actin filaments in response to holocalmodulin could initiate a new actin gelation cycle where or when calcium levels fall. Because the binding of FLNa to actin filaments also slows actin depolymerization, 4 calcium-calmodulin would also regulate this aspect of actin dynamics.
Most cells express 1-3 M FLNa, 50 -200 M F-actin (14), and 30 -60 M CaM (15,16), although the local concentrations of each at the leading edges of motile cells could be much higher (17). For example, the local concentration of calmodulin around the Ca 2ϩ channel has been estimated to be nearly 2.5 mM (18). Holocalmodulin does not bind FLNa directly. Rather, it seeks out the ABD of FLNa complexed to F-actin, thereby maximizing the specificity of its effect; it removes FLNa from actin filaments but does not prevent FLNa from binding to F-actin. We therefore attempted to achieve some understanding of this complex and interesting mechanism.
The FLNa ABD is representative of many actin filament-binding proteins, including members of the spectrin, ␣-actinin, dystrophin, plectin, and fimbrin families (20). The conserved ABDs contain two CH subdomains in tandem linked by a short connector sequence (Fig. 8) (21, 22). The first, CH1, has two putative actin-binding sites (ABS1 and ABS2; Fig. 8), and the second, CH2, has one (ABS3). Accumulated evidence points to a hydrophobic stretch in ABS2 essential for actin binding, but the structural requirements for the other two actin binding sites are less clear (23,24). Measured dissociation constants of isolated spectrin family ABDs for F-actin have been in the micromolar range (as we report in this paper for FLNa) (22,23,25). The CH2 subdomain of ␤-spectrin barely binds to F-actin (22), as we also found for its FLNa counterpart. The affinity of intact FLNa for F-actin (K d 0.5 M) (26) is an order of magnitude higher than that of the isolated FLNa ABD. Therefore, although the ABD of FLNa is essential for binding of full-length FLNa to F-actin, 4 other binding sites or the dimeric state of FLNa contribute to the FLNa-F-actin interaction.
Neither intact FLNa nor its ABD bound to holocalmodulin, as determined by sedimentation assays. However, holocalmodulin did bind to partially denatured recombinant FLNa, provided that the protein contained its ABD, suggesting that the FLNa ABD has a cryptic binding site for holocalmodulin. Consistent with this idea, recombinant fragments of the FLNa ABD did bind to holocalmodulin, and from this binding we inferred that residues 50 -96 in the CH1 subdomain represent the FLNa binding for calmodulin.
Other actin filament-binding proteins thought to bind and be regulated by holocalmodulin include ␤ spectrin (27), calponin (28), and utrophin (29), although some controversy exists regarding the role of holocalmodulin binding for dystrophin (29,30) or ␣-actinin functions (31,32). Many known calmodulin-binding proteins have basic amphipathic helix structures with positively charged amino acids interspersed among conserved hydrophobic stretches (11,33). Although such a motif in dystrophin overlaps with the functional calmodulinbinding domain that we identified for FLNa (shown in red in Fig. 8), a recombinant polypeptide containing the homologous sequence of FLNa alone bound calmodulin weakly. The addition to this fragment of a basic stretch of amino acids that is unique to the filamin family (shown in blue in Fig. 8) resulted in much tighter complexing to holocalmodulin. This cluster of basic amino acids in the FLNa ABD is presumably not the principal site for calmodulin binding, however, because the inferred FLNa calmodulin-binding domain is twice as long as that of most known calmodulin-binding sites, and the basic region augments but is not required for calmodulin binding. Because the FLNa segment Thr 50 -Val 60 is critical for calmodulin binding and is homologous to an established calmodulin-binding motif in dystrophin, this region might represent the "authentic" FLNa calmodulin-binding domain, and the basic region may act as a clamp.
The basic subdomain at the C-terminal end of the FLNa calmodulinbinding site also suggests a mechanism as to why holocalmodulin does not bind native intact FLNa or its ABD. Vertebrate filamins have a characteristic acidic region within the linker sequence between their ABD CH domains that could bind to the corresponding basic domain in the CH1. Supporting this hypothesis, this acidic fragment inhibits calmodulin binding to CH1. This inhibition suggests a model for the mechanism by which calmodulin dissociates FLNa from F-actin (Fig. 9). The acidic segment of the FLNa ABD prevents calmodulin binding. However, when bound to F-actin the acidic piece does not block the calmodulin-binding site on FLNa, allowing holocalmodulin to bind the FLNa⅐F-actin complex and dissociate it by direct competition for the actin-binding site, as has been proposed for dystrophin (34).
Our model suggests that deletions or point mutations of the CaMbinding, basic, or acidic domains could affect FLNa function in vivo. However, these mutations might also impair actin binding because the CaM-binding domain of FLNa overlaps the ABD, and mutations could lead to conformational changes of the ABD that would complicate interpretation of any results.
Genetic mutations in FLNa cause a wide range of human diseases, and several point mutations have now been identified in the FLNa ABD (2). None of the mutations, however, are in the calmodulin-binding site identified in this study. Because the regulation of the FLNa-actin interaction requires the participation of other domains in the ABD in addition to the calmodulin-binding site, the point mutations identified in the ABD may cause disease by altering the calmodulin-FLNa interaction.