Plant Actin-binding Protein SCAB1 Is Dimeric Actin Cross-linker with Atypical Pleckstrin Homology Domain

Background: SCAB1 is a novel plant-specific actin-bundling protein.

Results: SCAB1 consists of an actin-binding domain, two coiled-coil dimerization domains, and a fused immunoglobulin and pleckstrin homology domain with an atypical binding site for inositol phosphates.

Conclusion: SCAB1 is a dimeric actin cross-linker and may be regulated by inositol signaling.

Significance: This work provides a structural framework to understand the function of SCAB1.

Abstract

SCAB1 is a novel plant-specific actin-binding protein that binds, bundles, and stabilizes actin filaments and regulates stomatal movement. Here, we dissected the structure and function of SCAB1 by structural and biochemical approaches. We show that SCAB1 is composed of an actin-binding domain, two coiled-coil (CC) domains, and a fused immunoglobulin and pleckstrin homology (Ig-PH) domain. We determined crystal structures for the CC1 and Ig-PH domains at 1.9 and 1.7 Å resolution, respectively. The CC1 domain adopts an antiparallel helical hairpin that further dimerizes into a four-helix bundle. The CC2 domain also mediates dimerization. At least one of the coiled coils is required for actin binding, indicating that SCAB1 is a bivalent actin cross-linker. The key residues required for actin binding were identified. The PH domain lacks a canonical basic phosphoinositide-binding pocket but can bind weakly to inositol phosphates via a basic surface patch, implying the involvement of inositol signaling in SCAB1 regulation. Our results provide novel insights into the functional organization of SCAB1.

Introduction

The actin cytoskeleton provides mechanical support to eukaryotic cells and plays a key role in cell morphogenesis, cell movement, and vesicle trafficking. In the cytoplasm, actin monomers polymerize into actin filaments (AFs),² which can further assemble into higher order structures as parallel bundles or orthogonal networks. The configuration, dynamics, and function of the actin cytoskeleton are regulated by numerous actin-binding proteins (ABPs) that affect AF polymerization, nucleation, capping, depolymerization, severing, bundling, and interaction with other cellular components (1–3).

In plants, actin bundles are present in most cell types and are involved in cytoplasmic streaming, intracellular transport of vesicles, maintenance of transvacuolar strands, organelle positioning, and stomatal opening (4–6). Several classes of actin bundlers have been characterized in plants, including villins (7–12), fimbrins (13, 14), certain formins (15, 16), and LIM proteins (17–19). These actin bundlers are also present in most other eukaryotes. Villins and formins are multifunctional ABPs with additional AF nucleation, capping, and severing activities. Different types of bundlers may help generate AF higher order structures with different properties that allow for specific functions. The Arabidopsis thaliana genome encodes five villins, five fimbrins, 20 formins, and six LIM proteins. Recent studies indicate that some isovariants of the same bundler type are different in their tissue-specific expression, biochemical activity, and response to regulatory signals such as calcium and pH (10–12, 17, 20).

Phosphoinositides are important regulators of the organization and dynamics of the actin cytoskeleton (21–24). Through interaction with numerous phosphoinositide-binding domains, phosphoinositides are involved in AF assembly on specific membrane areas, signal transduction, and regulation of the activity of a large number of actin-binding/modulating proteins. However, it is largely unknown whether phosphoinositides regulate the formation and maintenance of actin bundles in plants.

We recently identified a novel filamentous F-actin-binding protein, SCAB1 (stomatal closure-related actin-binding protein 1), in Arabidopsis (25). SCAB1 is widely expressed in various tissues and forms filamentous structures in the cytoplasm that co-localize with the actin cytoskeleton. SCAB1 possesses F-actin binding and bundling activities with no apparent AF nucleation or capping activities. In addition, SCAB1 stabilizes F-actin against depolymerization in vivo and in vitro in a dose-dependent manner.

The dynamics of the actin cytoskeleton are closely associated with stomatal movement. The actin cytoskeleton network in guard cells adopts a radial pattern when the stomata are open and transforms into a longitudinal pattern upon stomatal closure (26). The deletion or overexpression of SCAB1 was shown to retard this reorganization of the actin cytoskeleton and stomatal closure, suggesting that a proper level of SCAB1 protein is important for normal actin dynamics in guard cells (25). Accordingly, plants lacking SCAB1 are highly sensitive to drought stress and exhibit reduced stomatal movement in response to abscisic acid, H₂O₂, or CaCl₂ treatments, which are three factors that trigger stomatal pore closure.

SCAB1 homologs are widely present, often in multiple copies (three in Arabidopsis), in plants, including eudicots, monocots, ferns, and mosses, but are not found in algae and non-plant species. The SCAB1 amino acid sequence lacks similarity to any known ABP, indicating that it is a plant-specific novel ABP. The actin binding activity has been mapped to a region spanning residues 54–148 of SCAB1 (total of 496 residues). Aside from these findings, little is known about the organization and function of its domains.

In this study, we dissected the structure and function of SCAB1 via x-ray crystallography and biochemical methods. We determined high-resolution crystal structures for two independently folded regions. One domain adopts an antiparallel coiled coil that further dimerizes into a four-helix bundle, and the other domain at the C terminus of the protein folds into a fused immunoglobulin and pleckstrin homology (Ig-PH) fold. SCAB1 dimerizes through two coiled coils, and at least one coiled coil is required for actin binding. We propose that SCAB1 is a bivalent AF cross-linker. Moreover, we demonstrate that the C-terminal PH domain binds weakly with inositol phosphates via an atypical basic surface patch. This implies a functional link between SCAB1 function and inositol signaling. Overall, these results provide substantial novel insights into the functional organization of SCAB1.

EXPERIMENTAL PROCEDURES

Gene Cloning and Protein Purification

The DNA sequences encoding SCAB1 fragments were amplified by PCR from a full-length SCAB1 cDNA (25), digested with restriction enzymes SacI and HindIII, and cloned into a modified pETDuet-1 vector. Each SCAB1 construct was expressed as an N-terminal His₆ tag fusion with a PreScission protease site. Mutagenesis was conducted using the QuikChange procedure (Stratagene).

SCAB1(272–496) was expressed in Escherichia coli BL21(DE3) by induction with 0.5 mm isopropyl β-d-thiogalactopyranoside at 16 °C overnight. The cells were suspended in buffer A (20 mm Tris-Cl (pH 8.0) and 100 mm KCl) and disrupted by sonication. The clarified lysate was loaded onto a HisTrap column (GE Healthcare). The column was washed with 40 mm imidazole in buffer A, and the protein was eluted with 20 mm Tris-Cl (pH 8.0), 75 mm KCl, and 500 mm imidazole. The eluted protein was digested with PreScission protease overnight at 4 °C to remove the His₆ tag. The sample was diluted 2-fold by 20 mm Tris-Cl (pH 8.0) and loaded onto a HiTrap Q column (GE Healthcare). The protein was eluted at ∼100 mm KCl in a 0.05–1 m KCl gradient. The eluted protein was concentrated and further purified on a HiLoad 16/60 Superdex 200 column (GE Healthcare) in buffer A. The peak fractions were pooled, supplemented with 5 mm DTT, concentrated to ∼20 mg/ml, and stored at −80 °C.

SCAB1(61–151) was expressed similarly as SCAB1(272–496) and bound to a HisTrap column. The PreScission cleavage step was conducted on a column in 20 mm Tris-Cl (pH 8.0), 40 mm KCl, and 10 mm imidazole overnight at 4 °C. The eluted protein was diluted 3-fold by 20 mm Tris-Cl (pH 8.0) and loaded onto a HiTrap Q column. The protein was eluted at ∼150 mm KCl in a linear KCl gradient, further purified with a Superdex 75 16/60 column in buffer A, and concentrated to ∼20 mg/ml. Full-length SCAB1, GST-fused SCAB1, and GFP-fused SCAB1 were prepared as described previously (25).

Crystallization, Data Collection, and Structure Determination

Crystallization was conducted via the hanging drop vapor diffusion method at 20 °C and by mixing 1 μl of protein sample and 1 μl of reservoir solution. SCAB1(272–496) was crystallized in 2.0 m sodium malonate (pH 6.4) for 2–3 days. SCAB1(61–151) was crystallized in 0.1 m BisTris (pH 5.3) and 22.5% PEG 3350 for 1 week. Both SCAB1(61–151) and SCAB1(272–496) crystals were harvested by nylon loops and directly frozen in liquid nitrogen without further cryoprotection. Heavy atom derivatives were prepared by soaking SCAB1(61–151) crystals in 1 mm KAuCl₄ or SCAB1(272–496) crystals in 1 mm ethylmercury thiosalicylate for 12 h at 20 °C. The diffraction data for native and derivative crystals were collected with a Rigaku MicroMax-007 HF x-ray generator and an R-AXIS IV++ detector at 100 K and processed with Denzo and Scalepack (27).

The structure of SCAB1(61–151) was solved by single isomorphous replacement with anomalous scattering using a gold derivative. The SCAB1(272–496) structure was solved by single-wavelength anomalous dispersion using a mercury derivative. Heavy atom searching, phasing, and density modification were performed with SHARP (28). The SCAB1(272–496) structure was first automatically built in ARP/wARP (29) and manually adjusted in Coot (30). The SCAB1(61–151) structure was built in Coot. The structure refinement was conducted in Refmac (31). Omit maps were calculated with SFCHECK (32). The SCAB1(272–496) structure contains residues 279–490 in two protomers, two malonate molecules, and 370 water molecules. The SCAB1(61–151) structure contains residues 103–151 in one protomer, residues 100–146 in the other protomer, and 56 water molecules. The structural stereochemistry was analyzed using RAMPAGE (33).

Analytic Ultracentrifugation

Analytic ultracentrifugation sedimentation equilibrium studies were performed in a Beckman Optima XL-A analytic ultracentrifugation system equipped with an An-60 Ti rotor at 4 °C. The full-length SCAB1 protein was freshly purified with a Superdex-200 column in 20 mm Tris-HCl (pH 8.0), 200 mm KCl, and 1 mm tris(2-carboxyethyl)phosphine. Three samples (A₂₈₀ = 0.25) placed in a cell were centrifuged at 16,000 rpm for 50 h, and the absorbance at 280 nm was recorded. Because the protein was slightly degraded after centrifugation, the protein distribution profiles collected at 36 h were analyzed with SEDPHAT (34). The buffer density and partial specific volume were calculated in SEDNTERP. Analysis using the monomer-dimer equilibrium model yielded a dimer association constant (K_a) of ∼10¹⁰–10¹⁵ m, suggesting that the dimer is very tight. Therefore, the data were globally fit with the single-species model. The error was estimated with the Monte-Carlo method.

Isothermal Titration Calorimetry (ITC)

ITC experiments were conducted using an ITC 200 microcalorimeter (MicroCal) at 25 °C. Proteins were exchanged in 20 mm Na-HEPES (pH 8.0) and 100 mm NaCl with a Superdex 200 10/300 column. SCAB1(272–496) and its mutants (180 or 140 μm) were titrated with 2 mm inositol 1,4,5-triphosphate (Ins(1,4,5)P₃), phytic acid (Ins(1,2,3,4,5,6)P₆), and inositol 1-phosphate (Ins(1)P) dissolved in the same buffer. The data were analyzed with a one-set-of-sites binding model following the manufacturer's instructions.

F-actin Co-sedimentation Assays

High-speed F-actin co-sedimentation assays were performed as described (25). The proteins were centrifuged at 400,000 × g for 1 h at 4 °C prior to use. SCAB1 or its variants (0.5 μm) were incubated with 2 μm preformed F-actin in a 40-μl volume containing 10 mm imidazole (pH 7.0), 75 mm KCl, 1 mm MgCl₂, 1 mm EGTA, and 1 mm ATP for 30 min at room temperature. The samples were centrifuged at 150,000 × g for 30 min at 4 °C, and the supernatants and pellets were analyzed by SDS-PAGE.

Fluorescence Microscopy of AFs

AFs were visualized by rhodamine-phalloidin staining as described previously (25). Preformed F-actin (20 μm) was incubated with or without 1 μm GFP-tagged SCAB1 for 15 min at room temperature and then labeled with 20 μm rhodamine-phalloidin for 15 min. Next, the 5-μl mixture was diluted in 100 μl of fluorescence buffer containing 10 mm imidazole (pH 7.0), 75 mm KCl, 1 mm MgCl₂, 100 mm DTT, 100 mg/ml glucose oxidase, 15 mg/ml glucose, 20 mg/ml catalase, and 0.5% methylcellulose before imaging. The AFs were observed with a Zeiss LSM 510 META confocal microscope using a Plan-Apochromat 63×/1.4 oil objective.

RESULTS

Domain Analysis and Crystallization of SCAB1

Because SCAB1 is not homologous to other known proteins, we first analyzed its domain organization with multiple-sequence alignment and structural prediction programs (Fig. 1). The entire protein sequence, except for the N-terminal region, is highly conserved in SCAB1 plant homologs. The central region (residues 100–272) was predicted to form coiled coils (Fig. 1, A and B) with two breaks around positions 150 and 236 (35). The actin binding activity has been previously mapped to residues 54–148, which include part of the predicted coiled coil. Analysis with Jpred 3 predicted that the region to the C terminus from the predicted coiled coils is rich in secondary structures (36) and likely folds into an independent domain. Based on this domain analysis, >10 SCAB1 fragments were designed, expressed in E. coli, purified to homogeneity, and screened for crystallization. Two fragments, SCAB1(61–151) and SCAB1(272–496), yielded high-quality crystals, and their structures have been determined.

FIGURE 1.

Domain organization and sequence alignment of SCAB1. A, domain diagram of SCAB1. The ABD, CC1, CC2, Ig, and PH domains are labeled. B, prediction of coiled coils by COILS. The prediction was based on the matrix MTK, no positional weighting, and a window of 21. C, multiple-sequence alignment of SCAB1 homologs. Black and gray denote 98 and 80% conservation, respectively, in 33 aligned sequences, among which only SCAB1 homologs from the eudicot A. thaliana (At), the monocot Oryza sativa (Os), the fern Selaginella moellendorffii (Sm), and the moss Physcomitrella patens (Pp) are displayed. The secondary structures observed in the crystal structures of the CC1 and Ig-PH domains are indicated on the top of the alignment. The closed and open circles indicate those residues whose solvent-accessible surface is buried by at least 30 and 10 Ų, respectively, due to CC1 dimerization and are shown on the top and bottom of the alignment for the two subunits of the CC1 dimer. The residues important for F-actin binding are marked with closed squares, and the residues involved in inositol phosphate binding are marked with closed triangles.

Structure of Coiled-coil (CC) 1 Domain

The structure of SCAB1(61–151) was solved by single isomorphous replacement with anomalous scattering using a gold derivative. The structure was refined to 1.9 Å resolution, with R_work = 0.206 and R_free = 0.244 (Fig. 2A and Table 1). The asymmetric unit contains two molecules, and residues 103–151 and 100–146 were modeled. Approximately 40 residues in the N terminus of SCAB1(61–151) are not visible in the crystals due to protein degradation (supplemental Fig. S1).

FIGURE 2.

Dimeric structure and function of the CC1 domain. A, the omit electron density map is contoured at the 1. 5σ level and superimposed on the final structural model of CC1. B, ribbon representation of the CC1 structure. The two subunits are colored in cyan and magenta. The N and C termini and secondary structures are labeled. C, one subunit of a CC1 dimer is aligned with the nonequivalent subunit of another dimer. D, helical wheel diagram of the CC1 dimer. The αA helices are viewed from the N to C terminus, and the αB helices are viewed from the C to N terminus. Hydrophobic residues are green, Glu and Asp are red, Lys and Arg are blue, and all other residues are black. The lines denote electrostatic interactions. E, interactions stabilizing the CC1 dimeric structure. One subunit (magenta) is depicted as a surface with residues at least 80% conserved colored yellow, whereas the other subunit is shown as ribbons. The interacting residues are shown as sticks. The residues subjected to mutagenesis analysis are encircled. F, F-actin co-sedimentation assay of GFP-tagged SCAB1 NT5 (residues 54–148) and its mutants. NT5 (0.5 μm) was incubated with or without 2 μm preformed F-actin in a 40-μl reaction of 10 mm imidazole (pH 7.0), 75 mm KCl, 1 mm MgCl₂, 1 mm EGTA, and 1 mm ATP and subjected to centrifugation at 150,000 × g for 30 min. The proteins in the supernatants (S) and pellets (P) were resolved by SDS-PAGE. G, F-actin co-sedimentation assay of GST-tagged SCAB1 without CC1 (Δ101–148).

TABLE 1.

Statistics on data collection and structure refinement

	CC1 (Au)	CC1	Ig-PH (Hg)	Ig-PH
Data collection
Space group	P41212	P41212	P3121	P3121
Cell dimensions
a, b, c (Å)	48.0, 48.0, 78.1	48.2, 48.2, 77.9	76.2, 76.2, 161.8	76.3, 76.3, 161.9
α, β, γ	90°, 90°, 90°	90°, 90°, 90°	90°, 90°, 120°	90°, 90°, 120°
Wavelength (Å)	1.5418	1.5418	1.5418	1.5418
Resolution range (Å)a	20.00–3.00 (3.05–3.00)	20.00–1.90 (1.93–1.90)	20.00–2.00 (2.03–2.00)	20.00–1.70 (1.73–1.70)
Unique reflections	2078	7516	37,630	58,454
Redundancy	18.4 (13.8)	8.7 (8.4)	3.8 (3.8)	10.8 (10.5)
I/σ	26.27 (9.0)	36.9 (5.4)	26.77 (6.0)	47.2 (4.7)
Completeness (%)	99.9 (100.0)	97.4 (96.2)	98.6 (96.2)	95.9 (92.8)
Rmergeb	0.153 (0.349)	0.06 (0.381)	0.05 (0.246)	0.05 (0.565)
Structure refinement
Resolution range (Å)		20–1.9 (1.95–1.9)		20–1.7 (1.74–1.7)
No. of reflections		7172		55,433
No. of atoms		800		3616
Rworkc		0.206 (0.242)		0.208 (0.371)
Rfreed		0.244 (0.314)		0.233 (0.400)
Average B factor (Å2)		28.7		27.1
r.m.s.d.e bond length (Å)		0.007		0.007
r.m.s.d. bond angles		1.154°		1.200°
RAMPAGE statistics
Favored (%)		98.9		98.1
Allowed (%)		1.1		1.9
Outlier (%)		0		0

^a Values in parentheses are for the data in the highest resolution shell.

^b R_merge = Σ|I_i − I_m|/ΣI_i, where I_i is the intensity of the measured reflection, and I_m is the mean intensity of all of the symmetry-related reflections.

^c R_work = Σ|F_o − F_c|/ΣF_o, where F_o and F_c are the observed and calculated structure factor amplitudes, respectively.

^d R_free is the same as R_work but calculated on 5% reflections not used in refinement.

^e r.m.s.d., root mean square deviation.

Each SCAB1(61–151) molecule is composed of two α-helices (αA and αB) that fold back into an antiparallel left-handed coiled coil. This autonomously folded domain (residues 100–150) is designated CC1 (Fig. 2B), whereas its C-terminal segment (residues 151–272), which was also predicted to be a coiled coil with a score even higher than that of the CC1 domain (Fig. 1B), is designated CC2.

The CC1 domain dimerizes into a four-helix bundle with approximate dimensions of 35 × 25 × 22 Å. The dyad axis of the CC1 dimer is roughly parallel to the helical axes. In the dimer, each α-helix of one subunit is aligned in parallel to its counterpart from the other subunit. Consequently, the four N and C termini of the CC1 domains are positioned at the same end of the bundle. The two monomer subunits can be well aligned with a root mean square deviation of 0.722 Å over 38 Cα pairs, except for the interhelix loop, which displays some conformational variations (Fig. 2C). When two CC1 dimers are aligned by nonequivalent subunits, the other subunits are slightly displaced from each other, indicating that the dyad symmetry is imperfect (Fig. 2C).

The dimerization buries a total of 1870 Ų of solvent-accessible area, which accounts for 24% of the total surface area. The αA and αB helices are amphipathic and contact each other via the hydrophobic face within and between individual subunits (Fig. 2, D and E). The dimeric structure is stabilized by numerous hydrophobic interactions in the interior as well as a few electrostatic interactions that occur on the surface between spatially adjacent helices.

A search with the Dali server indicates that the CC1 dimeric structure is similar to many dimeric α-hairpin structures (37) such as an engineered RNA-binding ROP protein (Protein Data Bank code 1F4N, Z = 5.5) (38) and the M1-binding domain of the influenza A viral nuclear export protein (code 1PD3, Z = 5.4) (39).

Dimeric Assembly of SCAB1 Is Required for F-actin Binding

A fragment of SCAB1 (NT5, residues 54–148) including CC1 was previously shown to be sufficient for actin binding and bundling in vitro and in vivo (25). To assess the role of CC1 in actin binding, we disrupted its dimeric structure by replacing three residues in the hydrophobic core one by one with a charged aspartate or glutamate (V131D, V138D, and L141E) in NT5. Wild-type NT5 bound efficiently to F-actin in co-sedimentation and co-localization assays, but the three mutants all failed to do so (Fig. 2F and supplemental Fig. S2). A CC1 deletion fragment (residues 54–100) barely bound F-actin either (supplemental Fig. S2). These results indicate that the dimeric structure of CC1 is essential in NT5 for F-actin binding.

Although the F-actin binding activity of NT5 was abolished when the CC1 structure was disrupted by point mutations (Fig. 2F) or by deletion of residues 129–148 in helix αB (25), the deletion of residues 100–128 (helix αA) in full-length SCAB1 was previously shown, puzzlingly, to have no effect on F-actin binding (25). We reasoned that CC1 may function simply as a dimerization module rather than directly binding F-actin. In full-length SCAB1, CC1 is not needed for actin binding because CC2 still mediates dimerization (see below). However, in the context of NT5, CC1 becomes essential for actin binding because it is the only dimerization module available in that situation. Indeed, full-length SCAB1 with the entire CC1 domain deleted still efficiently bound F-actin in co-sedimentation and co-localization assays (Fig. 2G and supplemental Fig. S2). We can conclude that the dimeric structure of SCAB1, as mediated by at least one of the CC1 and CC2 domains, is essential for F-actin binding.

Key Residues in Actin-binding Domain

The above structure-function analyses suggest that only residues 54–100 directly contact F-actin, and they constitute the actual actin-binding domain (ABD). In addition, residues 74–100 appear to compose the core region, as they are essential for actin binding in full-length SCAB1 and retain most of actin binding activity in the presence of CC1(25). To further identify the key residues involved in actin binding, we introduced point mutations at six conserved positions of the ABD into full-length SCAB1 constructs and assessed the F-actin binding activity of these mutants via co-sedimentation assays. The V74A, R75E, and F81A mutations abolished actin binding, but the L77A, A88S, and L91A mutations had no detectable effect on actin binding (Fig. 3). An in vitro fluorescence co-localization assay yielded the same results (supplemental Fig. S3). The three residues Val-74, Arg-75, and Phe-81 are likely involved in direct actin interaction or maintaining an ABD conformation competent for actin interaction.

FIGURE 3.

Key residues in the ABD. GST-tagged WT SCAB1 and mutants V74A, R75E, L77A, F81A, A88S, and L91A were co-sedimented with F-actin. SCAB1 (0.5 μm) was incubated with or without 2 μm preformed F-actin and subjected to centrifugation at 150,000 × g for 30 min. The proteins in the supernatants (S) and pellets (P) were resolved by SDS-PAGE.

Structure of Fused Ig-PH Domain

The SCAB1(272–496) construct contains the entire C-terminal region downstream of CC2. Its crystal structure was solved by single-wavelength anomalous dispersion phasing using a mercury derivative and refined to 1.7 Å resolution, with R_work = 0.208 and R_free = 0.233 (Fig. 4, A and B, and Table 1). The asymmetric unit contains two SCAB1(272–496) molecules, both with residues 279–490 modeled. SCAB1(272–496) is a monomer in crystal form and in solution as judged by gel filtration chromatography (see Fig. 6A).

FIGURE 4.

Structure of the Ig-PH domain. A, ribbon representation of the Ig-PH domain. The Ig and PH domains are colored wheat and green, respectively. The secondary structures are labeled. B, the omit electron density map contoured at the 1.5σ level is shown for the malonate-binding site. C, the conservation surface as shown in three orientations. The lower right structure is oriented as in A. The Ig and PH domains are colored wheat and green, respectively. The residues at least 80% conserved are colored yellow. D, interactions between the Ig and PH domains. Hydrogen bonds are denoted by dotted lines.

FIGURE 6.

Full-length SCAB1 is a dimer. A, the C-terminal part of CC2 mediates dimerization. The Ig-PH domain alone (residues 272–496) and the Ig-PH domain fused to part of CC2 (residues 239–496) were analyzed using a Superdex 200 10/300 column. The normalized absorbance at 215 nm is plotted against the elution volume. The elution positions of the calibration standards BSA (67 kDa), chymotrypsinogen A (25 kDa), and lysozyme (14.7 kDa) are marked. B, analytic ultracentrifuge sedimentation equilibrium analysis of full-length SCAB1. The curves are the best global fit of the three profiles to the single-species model, yielding a molecular mass of 102.8 ± 1.0 kDa.

The structure has overall dimensions of ∼60 × 40 × 30 Å and is composed of two regions (Fig. 4A). The N-terminal region adopts an Ig β-sandwich fold that is composed of two antiparallel β-sheets built from strands β1 and β2 and strands β3–β6, respectively. The C-terminal region adopts a PH fold with seven β-strands (β7–β13) and two α-helices (α1 and α2) arranged into a β-barrel. Similar to other PH domains, the β-barrel is sealed by an α-helix (α2) at one end and open at the other end. The α1 helix that links the Ig and PH domains is not present in classic PH domains. The structures of the individual Ig and PH domains are homologous to many other structures (37), but the entire assembly, called the Ig-PH domain, is unique. A few conserved patches are evident on the surface of the Ig-PH domain, particularly on the PH domain, implying that the Ig-PH domain may be involved in some molecular interactions (Fig. 4C).

The Ig and PH domains share an extensive interface around the β4–β5 loop in the Ig domain and helix α2 and the β11-β12 loop in the PH domain. Many electrostatic and polar interactions mediate the interdomain contact (Fig. 4D), and most of the interacting residues are highly conserved in SCAB1 homologs (Fig. 1C). In addition, no domain movement was observed for the two protomers in the asymmetric unit, which could be aligned with a root mean square deviation of 0.389 Å for 193 Cα pairs (supplemental Fig. S4). These structural features suggest that the Ig and PH domains are fused into an integral structure.

PH Domain Binds Weakly with Inositol Phosphates via Basic Surface Patch

The PH domain is an abundant structural fold in eukaryotic proteins and is traditionally known to bind the inositol phosphate headgroup of phosphoinositides (40–42). Using ITC, we examined whether SCAB1(272–496) binds inositol phosphates (Fig. 5A). The results indicate that SCAB1(272–496) was able to weakly bind Ins(1,4,5)P₃, which is the headgroup of phosphatidylinositol 4,5-bisphosphate, with an apparent dissociation constant (K_d) of 76 μm. Tighter binding (K_d = 13 μm) was detected for Ins(1,2,3,4,5,6)P₆, and no binding was detected for Ins(1)P. The binding affinity is proportional to the overall charge of the ligand, suggesting that the interaction is mediated primarily by electrostatic interactions and has little specificity.

FIGURE 5.

PH domain of SCAB1 binds inositol phosphates via a low-affinity surface site. A, ITC of SCAB1(272–496) with inositol phosphates Ins(1,4,5)P₃, Ins(1,2,3,4,5,6)P₆, and Ins(1)P. The curves in the lower panels are the best fit to a one-set-of-sites binding model. The derived dissociation constants (K_d) are indicated. B, ITC of the SCAB1(272–496) mutants R410N, K412N, and K422N with Ins(1,4,5)P₃. C, electrostatic potential surface for the PH domain. The surface is colored in blue to red for positively to negatively charged regions. The arrow points to the equivalent high-affinity inositol phosphate (IP)-binding pocket, which is not negatively charged in SCAB1. D, interaction between the SCAB1 PH domain and malonate. The dotted lines denote hydrogen bonds. E, structure of the phospholipase Cδ1 (PLC-δ) PH domain in complex with Ins(1,4,5)P₃ (Protein Data Bank code 1MAI) aligned with the SCAB1 PH domain structure. F, structure of the β-spectrin PH domain in complex with Ins(1,4,5)P₃ (Protein Data Bank code 1BTN) aligned with the SCAB1 PH domain structure.

Only a small number of PH domains are capable of binding phosphoinositide with high affinity and specificity (43). In these cases, the headgroup of phosphoinositide is bound in a positively charged pocket located at the open end of the PH β-barrel (Fig. 5E), as first illustrated in the structure of the phospholipase Cδ1 PH domain in complex with Ins(1,4,5)P₃ (44). The equivalent pocket in the SCAB1 PH domain is largely devoid of basic residues and hence unlikely to bind phosphoinositide as high-affinity PH domains (Fig. 5C).

The degenerate phosphoinositide-binding pocket suggests that the Ig-PH domain contains a noncanonical binding site for inositol phosphates. Interestingly, we found that a malonate molecule, which was present in the crystallization solution, bound the surface of the PH domain in the crystal (Figs. 4B and 5, C and D). Specifically, the side chains of Arg-410, Lys-412, Lys-422, and Tyr-424 in strands β9 and β10 form multiple hydrogen bonds and electrostatic interactions with the malonate molecule. As malonate and inositol phosphate are both negatively charged molecules, the basic patch is probably responsible for binding inositol phosphate. To test this idea, we singly replaced three basic residues in the patch with a neutral asparagine (R410N, K412N, and K422N). All three mutations abolished the binding of Ins(1,4,5)P₃ (Fig. 5B), suggesting that the positively charged surface in the PH domain mediates inositol phosphate binding. The PH domain of β-spectrin was previously shown to bind weakly to Ins(1,4,5)P₃ (K_d = 40 μm) also via a basic surface patch (45), which is different compared with the PH domain of SCAB1 (Fig. 5F).

Full-length SCAB1 Is Dimeric

Coiled coils most frequently form parallel dimers, but they are also able to form trimers, tetramers, and other types of oligomers (46). We have shown that the CC1 domain of SCAB1 folds into an antiparallel coiled coil that further dimerizes, but the structure of the CC2 domain has not yet been determined. We sought to probe the oligomeric state of CC2 as well as full-length SCAB1 using biochemical approaches. The Ig-PH domain alone eluted upon size exclusion chromatography as a monomeric species with a molecular mass of 38 kDa. The fusion of the C-terminal 33 residues of CC2 nearly doubled the apparent molecular mass of Ig-PH to 64 kDa, suggesting that CC2 mediates dimerization (Fig. 6A). We further measured the molecular mass of full-length SCAB1 by analytic ultracentrifugation sedimentation equilibrium, a method insensitive to molecular shape (Fig. 6B). The protein distribution profile at equilibrium can be well accounted for by a single species of 102.8 ± 1.0 kDa, corresponding to a SCAB1 dimer (monomer molecular mass of 56 kDa). We can conclude that full-length SCAB1 assumes a dimeric structure and that both coiled-coil domains mediate dimerization.

DISCUSSION

SCAB1 is a novel plant-specific ABP that packs AFs into bundles. In this study, we have shown that SCAB1 has a modular organization and comprises the ABD, CC1, CC2, and Ig-PH domains. The central two coiled coils mediate SCAB1 dimerization and segregate the ABD and the Ig-PH domain into opposite ends (Fig. 7). The unique structure and domain organization of SCAB1 further demonstrate that it is a new type of actin-bundling protein.

FIGURE 7.

Model depicting a SCAB1 dimer that cross-links AFs. SCAB1 dimerizes via its CC1 and CC2 domains. CC2 is shown as parallel coils but may adopt a more complex structure. The dual ABD in the SCAB1 dimer simultaneously contact two adjacent AFs.

We have refined the boundaries of the ABD to a small region spanning residues 54–100 and identified a few residues in this region that are important for actin binding. Although the CC1 domain is required for actin binding in the NT5 fragment, it is not included in the ABD. Our results suggest that the CC1 domain is not directly involved in actin binding but functions as a dimerization module, a role that could be substituted for by the CC2 domain. Moreover, the ABD is structurally independent from the CC1 domain, as they were separable during crystallization.

To cross-link two AFs, actin-bundling proteins must have at least two actin-binding sites. This can be achieved by incorporation of multiple actin-binding sites in a single polypeptide chain or by dimerization of proteins with a single actin-binding site. For instance, villin forms a dimer and employs the C-terminal headpiece domain to bundle AFs (47). Fimbrin possesses two homologous ABDs composed of tandem calponin homology domains (48). Plant LIM proteins are composed of two actin-binding LIM domains that function cooperatively to bundle AFs (18). Our results suggest that the actin bundling activity of SCAB1 is mediated through dimerization of a single actin-binding site. The dual ABDs in the SCAB1 dimer could interact simultaneously with two adjacent AFs and cross-link them (Fig. 7). The monomeric ABD cannot appreciably bind AFs apparently due to a low intrinsic affinity. Dimerization of SCAB1 appears to be essential for the ABD to gain sufficient avidity for AFs or to bind AFs in a cooperative manner.

The length of the spacer that separates the ABDs in actin-bundling proteins is an important determinant for the structure of bundles. The close proximity of the two ABDs in the SCAB1 dimer, which are connected to the juxtaposed N termini of the CC1 dimer, would pack AFs into tight bundles. Notably, the removal of CC1 in full-length SCAB1 will still preserve the relative position of the two ABDs and hence preserve a similar bundling mode. In the CC1 deletion mutant, the two ABDs are connected to the N termini of the CC2 domains that would be aligned side by side, similar to the C termini of the CC1 domains.

Phosphoinositides play a central role in regulating the organization and dynamics of the actin cytoskeleton. Phosphoinositides interact with numerous actin-binding/modulating proteins and regulate their membrane localization and activity (21). Our structural analysis revealed an unexpected PH domain in SCAB1, which could not be predicted from the protein's sequence. Despite lacking a canonical phosphoinositide-binding pocket, the Ig-PH domain is able to bind weakly to inositol phosphates with a basic surface patch and is likely able to bind phosphoinositides in the membrane. Moreover, the binding affinity may increase upon oligomerization of the molecule, as shown previously for dynamin (49). The functional relevance of such weak binding is still unclear. This problem has been associated with the majority of PH domains that weakly bind phosphoinositides (40, 43). The present results provide a structural framework for future studies of the molecular function and regulation of SCAB1.