NMR structure of a heterodimeric SAM:SAM complex: characterization and manipulation of EphA2 binding reveal new cellular functions of SHIP2

Abstract

The sterile alpha motif (SAM) for protein-protein interactions is encountered in over 200 proteins, but the structural basis for its interactions is just becoming clear. Here we solved the structure of the EphA2-SHIP2 SAM:SAM heterodimeric complex by use of NMR restraints from chemical shift perturbations, NOE and RDC experiments. Specific contacts between the protein surfaces differ significantly from a previous model and other SAM:SAM complexes. Molecular dynamics and docking simulations indicate fluctuations in the complex toward alternate, higher energy conformations. The interface suggests that EphA family members bind to SHIP2 SAM, whereas EphB members may not; correspondingly, we demonstrate binding of EphA1, but not of EphB2, to SHIP2. A variant of EphB2 SAM was designed that binds SHIP2. Functional characterization of a mutant EphA2 compromised in SHIP2 binding reveals two previously unrecognized functions of SHIP2 in suppressing ligand-induced activation of EphA2 and in promoting receptor coordinated chemotactic cell migration.

Figure 1

SAM:SAM heterodimer topology via the end-helix/mid-loop interface motif. (A) hCNK2:dHYP SAM:SAM complex (B) Secondary structure elements of EphA2 SAM domain; Mid-Loop (ML) and End-Helix (EH) surfaces are indicated. (C) Structure for EphA2:SHIP2 SAM:SAM complex (lowest HADDOCK score structure of cluster 1). See also Figures S1, S2 and S5.

Figure 2

Comparison of interface character. Electrostatic features of the hCNH2:dHYP interface compared to the EphA2:SHIP2 SAM interacting surface (shown as open book presentation), shaded blue = positive charge and red = negative charge. Circled region indicates the binding surfaces.

Figure 3

Ensemble of 200 HADDOCK structures obtained from NMR restraints. (A) Plot of HADDOCK score versus rmsd. Three clusters are obtained. One of them is centered at about 1.5 Å and the other two clusters are overlapped at an rmsd of 4 Å, respectively. (B) A comparison of structures between HADDOCK clusters; green: cluster 1, blue: cluster 2, and magenta: cluster 3. Molecules of cluster 2 and 3 are aligned to EphA2 SAM of the cluster 1. (C) A schematic diagram of conformational differences between the clusters. Only H5 helices of EphA2 and SHIP2 are shown, and arrows show the C-terminal direction. See also Figures S3, S4, S6, S8, and Tables S1, S2, S5, S6 and S7.

Figure 4

Details of the binding interface of the EphA2 and SHIP2 complex for the three HADDOCK clusters. Summary of charge pair interactions (A–D) and hydrophobic interactions involving aromatic residues (E–H) between EphA2 and SHIP2 interface in different configurations. Hydrogen bonds and salt-bridges are shown as dashed lines and hydrophobic surfaces are shown as combs. (A, E) cluster 1; (B, F) cluster 2; (C,G) cluster 3 and (D,H) model reported by the Pellecchia group (Leone et al., 2008). See also Figures S7, S9, S10 and S11.

Figure 5

EphA2:SHIP2 SAM:SAM complex ensembles and its behavior during unrestrained MD simulation. (A) An ensemble of 15 lowest scored HADDOCK structures of cluster 1 after xplor-nih refinement. (B) 15 structures from unrestrained MD calculation shown at 1 ns simulation interval. (C) Buried surface area of the complex and (D) Changes in mainchain structure of EphA2, SHIP2, and the complex (rmsd of Calpha from starting complex structure) during MD simulation. See also Figure S7.

Figure 6

RosettaDock score for perturbative docking of selected simulation coordinate sets with reference to lowest energy structure found. The results imply that the complex undergoes considerable fluctuations. See also Figure S7.

Figure 7

Cellular characterization of loss and gain of function EphA2 SAM mutants. A, EphA2 R950T promotes SHIP2 association, while the K917/P952A/K956E triple mutant (TM) attenuates the association. Serum-starved U87 cells infected with WT, or R950T and the triple mutant were lysed. EphA2 was precipitated with ephrin-A1-Fc and blotted sequentially for SHIP2 and EphA2. Equal amount of wild type EphA2 lysate from same experiment was immunoprecipitated with Fc as negative control. B, The band densities of SHIP2 from A were normalized to the corresponding total EphA2. C, EphA2 kinase activation of the triple mutant by ephrin-A1 shows hypersensitivity to ligand stimulation and accelerated degradation. HEK 293 cells expressing WT or mutant EphA2 were stimulated with ephrin-A1-Fc for indicated times. Cell lysates were blotted with the indicated antibodies. Quantitative analyses of EphA2 for D, degradation and E, activation following ligand stimulation. F, Ectopic overexpression of WT, but of not the mutant EphA2, enhances serum-induced chemotaxis, plotted as the number of migratory cells. HEK 293 cells were subjected to Boyden chamber cell migration assay. Cell numbers from 6 random fields were counted. Numbers were normalizes by vector. Numbers represent mean ± S.D from 3 independent experiments. *p<0.05. See also Figure S10.