Structural basis of filopodia formation induced by the IRSp53/MIM homology domain of human IRSp53

Abstract

The scaffolding protein insulin receptor tyrosine kinase substrate p53 (IRSp53), a ubiquitous regulator of the actin cytoskeleton, mediates filopodia formation under the control of Rho-family GTPases. IRSp53 comprises a central SH3 domain, which binds to proline-rich regions of a wide range of actin regulators, and a conserved N-terminal IRSp53/MIM homology domain (IMD) that harbours F-actin-bundling activity. Here, we present the crystal structure of this novel actin-bundling domain revealing a coiled-coil domain that self-associates into a 180 A-long zeppelin-shaped dimer. Sedimentation velocity experiments confirm the presence of a single molecular species of twice the molecular weight of the monomer in solution. Mutagenesis of conserved basic residues at the extreme ends of the dimer abrogated actin bundling in vitro and filopodia formation in vivo, demonstrating that IMD-mediated actin bundling is required for IRSp53-induced filopodia formation. This study promotes an expanded view of IRSp53 as an actin regulator that integrates scaffolding and effector functions.

Figure 1

Domain structure of IRSp53 and sequence alignment of IMD and BAR domains. (A) Schematic representation of the domain organisation of IRSp53. Vertical bars in grey are spaced by 100 residues. Domains are indicated by boxes in grey, and interaction motifs by boxes in white. (B) Sequence alignment of IMD and BAR domains using ClustalW at EBI (http://www.ebi.ac.uk/clustalw). Sequences and data base entries are as follows: IRSp53 NP_059344; IRTKS NP_061330; FLJ22582 NP_079321; MIM-B NP_055566; ABBA-1 NP_61239; GRAF NP_055886; centaurin-β2 NP_036419; arfaptin2 AAH00392; amphiphysin NP_001626. Symbols above the sequence alignment refer to IRSp53, indicating secondary structure assignment, involvement in dimer contact surface (=) and residue numbers. Asterisks in magenta indicate the mutation sites. Below the sequence alignment, secondary structure of the BAR domain of amphyphysin, and residues contributing to the dimer contact surface area (hyphen in magenta) of the BAR domain are indicated. Sequence conservation is highlighted by colouring (ClustalX), magenta (Glu, Asp), cyan (Trp, Met, Phe, Leu, Ile, Val, Ala), dark cyan (His, Tyr), green (Gln, Asn, Ser, Thr), red (Lys, Arg), yellow (Pro) and brown (Gly).

Figure 2

Structure of the IMD of human IRSp53. (A) Stereo diagram of the electron density derived from the three-wavelength MAD experiment contoured at 0.8σ and superimposed with the refined model. Shown is the core segment of the IMD signature sequence (¹⁸⁹EERRR¹⁹³). Bonds and atoms are coloured according to atom type (yellow/grey: carbon; red: oxygen; blue: nitrogen). Selected residues are indicated in single letter code. Chains A and B are distinguished by the colour of the carbon bonds. (B) Two orthogonal views, parallel and perpendicular to the noncrystallographic two-fold axis, of the IMD dimer in ribbon and molecular surface representation. The ribbon diagrams are colour ramped blue to green (chain A) and yellow to red (chain B). The mutation sites are indicated for one of the monomers in single letter code, and the location of the IMD signature sequence is highlighted with side chains in grey. Vertical dashed lines indicate the core region of the dimer. N- and C-termini of chains A and B are indicated in italics. (C) Packing of the IMD dimer in the unit cell. (D) Structural superposition of the IMD dimer with the BAR domain dimer (grey) of amphiphysin (pdb: 1uru, (Peter et al, 2004)). Figures 2 and 3 were generated using RIBBONS (Carson, 1997), Swiss-PDB Viewer (Guex and Peitsch, 1997), O (Jones et al, 1991) and GRASP (Nicholls et al, 1991).

Figure 3

The dimer interface and properties of the molecular surface of the IMD dimer. (A, B) Molecular contact surface (yellow) of the IMD (A) and amphiphysin BAR domain (B) dimers, calculated using SwissPDB-Viewer (Guex and Peitsch, 1997). (C) Molecular surface of the IMD dimer oriented with the dyad parallel to the viewing direction. The surface area is coloured according to sequence conservation among the five mammalian IMD sequences in Figure 1. The colour shading is based on the Q-score calculated by ClustalX (Thompson et al, 1997), whereby deep blue indicates identity (Q=100). The top and bottom panels show the N- and C-terminal faces of the dimer, respectively. The C in italics on the lower panel indicates the location of the C-termini on the IMD dimer. (D) Molecular surface of the IMD dimer oriented as in panel (C) and coloured according to electrostatic surface potential, colour ramped from −15 k_BT/e (red) to +15 k_BT/e (blue).

Figure 4

In vitro characterisation of the IMD of IRSp53 and interaction with F-actin. (A) SDS–polyacrylamide gel electrophoresis (SDS–PAGE) of purified IMDwt (wt) and IMD_C230A (C230A) analysed under reducing (+DTT) and nonreducing (−DTT) conditions after heat denaturation. All gels in this figure were stained with Coomassie blue and positions of molecular weight markers are indicated in units of 10³ Da. (B) Sedimentation coefficient distributions of IMDwt and IMDmut. The main peak for both is at 3.25s, while some higher order aggregate is observed for the mutant at 4.5s. (C) Circular dichroism spectroscopy of wild-type (thick line) and mutant (thin line) forms of the IMD. (D) High-speed cosedimentation assay of the interaction of wild-type (wt) and mutant IMD (mut) with F-actin (2.5 μM). Pellet fractions were analysed by SDS–PAGE and the band intensity of the IMD was measured densitometrically and normalised to actin. (E) Low-speed co-sedimentation assay of F-actin bundling. Pellet (P) and supernatant (S) fractions were analysed by SDS–PAGE. (F) Fluorescence microscopy-based F-actin-bundling assay. Cy3-labelled F-actin (1 μM) was incubated with 5 μM wt or mut IMD and imaged using a fluorescence microscope (scale bar=20 μm).

Figure 5

Analysis of IRSp53 and IMD overexpression in COS7 cells. (A) Plasmid constructs encoding myc-tagged forms of either IRSp53 (wild-type) or IRSp53 (mutant) were transfected into COS7 cells for 24 h, fixed with paraformaldehyde and stained with TRITC–phalloidin and anti-myc, followed by a fluorescent secondary antibody. Coverslips were visualised by fluorescence microscopy. (B) As (A), but cells were transfected with plasmid constructs encoding myc-tagged forms of IMDwt or IMDmut (scale bars in (A) and (B)=20 μm). (C) Cell counts comparing filopodia formation in transfected versus untransfected COS7 cells. Error bars represent standard deviations from three independent blind experiments counting approximately 35 transfected and untransfected cells for each condition. Cells were scored positive for filopodia if they showed at least five filopodia with a length greater than 10% of the cell diameter, corresponding to approximately 10 μm.