Structural Basis for Noncanonical Substrate Recognition of Cofilin/ADF Proteins by LIM Kinases

Abstract

Cofilin/actin-depolymerizing factor (ADF) proteins are critical nodes that relay signals from protein kinase cascades to the actin cytoskeleton, in particular through site-specific phosphorylation at residue Ser3. This is important for regulation of the roles of cofilin in severing and stabilizing actin filaments. Consequently, cofilin/ADF Ser3 phosphorylation is tightly controlled as an almost exclusive substrate for LIM kinases. Here we determine the LIMK1:cofilin-1 co-crystal structure. We find an interface that is distinct from canonical kinase-substrate interactions. We validate this previously unobserved mechanism for high-fidelity kinase-substrate recognition by in vitro kinase assays, examination of cofilin phosphorylation in mammalian cells, and functional analysis in S. cerevisiae. The interface is conserved across all LIM kinases. Remarkably, we also observe both pre- and postphosphotransfer states in the same crystal lattice. This study therefore provides a molecular understanding of how kinase-substrate recognition acts as a gatekeeper to regulate actin cytoskeletal dynamics.

Figure 1. Schematic showing the role of a high fidelity kinase-substrate pair in control of actin polymerization

Control of actin filament dynamics is regulated by the upstream RHO-GTPase family and their effector proteins, which directly activate the LIM kinases. The LIM kinases phosphorylate the cofilin/ADF-family of proteins on residue Ser3. Sequence alignment for residues 2–10 is shown for human cofilin-1, cofilin-2 and ADF proteins; Ser3 is marked in red. Unphosphorylated cofilin/ADF proteins bind and destabilize actin filaments, however, cofilin/ADF proteins phosphorylated on Ser3 do not bind actin. Cofilin phosphorylation is a major convergence point that links upstream signaling to actin remodeling. See also Figure S1.

Figure 2. Co-crystal structure of the LIMK1 catalytic domain in complex with full-length cofilin-1

(A)Overall structure of pLIMK1_CAT^D460N with cofilin-1. LIMK1 is shown in brown, and cofilin-1 in green. Secondary structure elements, N- and C- termini, kinase N- and C-lobes are labeled. Cofilin Ser3, phosphorylated LIMK1 activation loop residue pT508, and bound nucleotide are shown in stick format. B, C) Close up of the active site for pre-catalytic (B) and post-catalytic (C) complexes found in the asymmetric unit. LIMK1 residues shown in stick format are from the conserved kinase DFG-motif (D478, F479), HRD-motif (D460N), catalytic lysine (K368) and catalytic glutamate (E384). N-terminal residues of cofilin-1 are shown in stick format. See also Figure S2.

Figure 3. Molecular basis for LIMK1 recognition of its substrate cofilin-1

(A)Detailed interaction between pLIMK1_CAT^D460N with cofilin-1. Surface renderings of LIMK1 (brown) and cofilin-1 (green). Top right and bottom right zoomed regions show details of the interaction, with side chains shown in stick format. Residues and some secondary structure elements are labeled. Bottom left shows the surface of LIMK1 illustrating the deep cleft that cofilin helix α5 residues insert into. Ser3 is indicated by a dashed oval. B) Electrostatic potential of interaction interface. Face on view of LIMK1 (electrostatic potential) and cofilin (only helix α5 and the N-terminal residues shown). Side chains and nucleotide shown in stick format, and helix α5 in cartoon format. Ser3 is indicated by a dashed oval. C) Map of LIMK1:Cofilin interactions. Residues involved in kinase-substrate targeting are shown; LIMK1 in brown and cofilin in green. Residues mutated in this study are indicated with red text. Interactions indicated with dotted lines. Structure elements indicated. Interactions defined by PDBSum (Laskowski, 2009). D) In vitro phosphorylation of cofilin-1 by LIMK1. Autoradiography showing pLIMK1_CAT phosphorylation of full-length Cofilin-1 (top). Mutants are indicated. Loading indicated by Coomassie stained SDS-PAGE (bottom). E) Quantitation of kinase assay. N=3. S.E.M. indicated. Comparison to wild-type shows that all pairs except LIMK1_CAT^M516S/cofilin^S119M are statistically significant (p-value < 0.0005) (paired t-test type 2). See also Figure S3.

Figure 4. Disruption of LIMK1-cofilin-1 interaction in mammalian cells and yeast

(A)Cofilin phosphorylation in mammalian cells. N-terminally His-tagged cofilin ran slower by SDS-PAGE than endogenous cofilin, allowing internal validation by anti-cofilin pSer3 antibody that the endogenous LIM kinases are active. HEK293T cells. B) Quantitation of exogenous pSer3 cofilin-1. Percent of wild-type signal calculated per experiment, and averaged across experiments. N=6. S.E.M. indicated. All pairings to wild-type are statistically significant (p-value < 0.0003) (paired t-test type 2). C) Expression of LIMK1_CAT in yeast is toxic to cells dependent on human cofilin-1. Serial dilutions of the indicated yeast strains harboring plasmids expressing human cofilin-1 (WT or S3A mutant) and human LIMK1_CAT or the corresponding empty vectors were plated on solid media in the presence of glucose (-Gal) or galactose (+Gal) to induce LIMK1_CAT expression. Plates were grown at the indicated temperature for 2 days (glucose plates) or 4 days (galactose plate). D–E) Growth of cof1-ts cells co-expressing cofilin-1 or LIMK1_CAT mutants. Cells harboring the indicated cofilin-1 (D) or LIMK1_CAT (E) mutant plasmids were grown as in (C).

Figure 5. Conservation of the high fidelity kinase-substrate pair across the LIM kinase family

(A)Conservation of LIMK1. Surface representation of LIMK1 structure indicating conservation across the human LIM kinases (LIMK1, LIMK2, TESK1, and TESK2). Dark blue indicates complete conservation (see Figure S5). Cofilin shown in cartoon format in green. B) In vitro kinase assay. TESK1_CAT phosphorylation of cofilin-1 or cofilin-1 mutants analyzed by autoradiography (top). Loading analyzed by Coomassie stained SDS-PAGE (bottom). C) Quantitation of kinase assay. N=3. S.E.M. indicated. Comparison to wild-type pair shows that all pairings are statistically significant (p-value < 0.0001) (paired t-test type 2). See also Figure S4.

Figure 6. The cofilin ‘molecular drill jig’ as a mechanism to achieve high kinase-substrate fidelity

(A)Schematic diagram showing the canonical molecular level mechanisms by which kinase-substrate specificity is achieved. Distal docking and linear motif sites are indicated. B) Schematic for how LIM kinases achieve high fidelity phosphorylation of cofilin/ADF proteins, an alternative mechanism to achieve molecular level specificity for kinase-substrate pairs. C) The complementary kinase-substrate interaction between LIM kinases and cofilin/ADF proteins acts as the jig component of a ‘molecular drill jig’, which conformationally places cofilin residue Ser3 in the correct location for catalysis. See also Figure S5.