Different molecular recognition by three domains of the full-length GRB2 to SOS1 proline-rich motifs and EGFR phosphorylated sites

Abstract

The adaptor protein human GRB2 plays crucial roles in mediating signal transduction from cell membrane receptors to RAS and its downstream proteins by recruiting SOS1. Recent studies have revealed that GRB2 also serves as a scaffold for liquid-liquid phase separation (LLPS) with SOS1 and transmembrane receptors, which is thought to regulate the magnitude of cell signalling pathways. In this study, we employed solution NMR spectroscopy to investigate the interactions of the full-length GRB2 with proline-rich motifs (PRMs) derived from ten potential GRB2-binding sites in SOS1, as well as a peptide from a phosphorylation site of EGFR. Our findings indicate that the binding affinity of the two SH3 domains of GRB2 for PRMs differs by a factor of ten to twenty, with the N-terminal SH3 domain (NSH3) exhibiting a markedly higher affinity. The interactions of PRMs with the SH3 domains affected not only the regions surrounding the PRM binding sites on the SH3 domains but also the linker area connecting the three domains and parts of the SH2 domain. Analysis of the interaction between the phosphorylated EGFR binding site and the SH2 domain revealed chemical shift perturbations in regions distal from the known binding site of SH2. Moreover, we observed that the inter-domain interactions of the two SH3 domains with the SH2 domain of GRB2 are asymmetric. These findings suggest that the local binding of PRMs and phosphorylated EGFR to GRB2 impacts the overall structure of the GRB2 molecule, including domain orientation and dimerisation, which may contribute to LLPS formation.

Fig. 1. Overlays of 2D ¹H–¹⁵N HSQC spectra from multipoint titrations of ¹⁵N-labelled GRB2 with SOS1–S4 PRM (VPPPVPPRRR). The PRM concentration was increased stepwise (the protein: PRM molar ratio of 1 : 0.25, 1 : 0.5, 1 : 1, 1 : 2, 1 : 4, 1 : 6, 1 : 8, and 1 : 12). In this figure, the colour codes of ¹H–¹⁵N correlation cross-peaks at each titration point, showing the molar ratio of GRB2: SOS1–S4, are as follows: black (1 : 0); dark blue (1 : 0.25); dark green (1 : 0.5); blue (1 : 1); cyan (1 : 2); green (1 : 4); yellow (1 : 6); orange (1 : 8); red (1 : 12). Cross-peaks that showed large chemical shift changes are annotated.

Fig. 2. Plots of chemical shift perturbation of backbone ¹H^N and ¹⁵N nuclei of GRB2 upon the titration with S4, S5, S9, and S10 PRMs. (a) Bar plots of chemical shift perturbation of backbone ¹H^N and ¹⁵N nuclei of GRB2 upon the titration with S4, S5, S9, and S10 PRMs. The mean shift difference Δδ_ave was calculated as [(Δδ¹H^N)² + (Δδ¹⁵N/5)²]^1/2 where Δδ¹H^N and Δδ¹⁵N are the chemical shift differences (ppm) between GRB2 on its own and in the presence of the PRMs. The bar graphs are colour-coded according to the protein–peptide concentration ratio and are overlaid. The proline residues as well as the residues for which ¹H^N–¹⁵N correlation cross-peaks were not analysed due to signal overlap or other reasons are shown in grey. The secondary structures of GRB2 are also shown. (b) Chemical shift perturbation 3D bar plots with an added axis for titration points to the plots of figure (a). The vertical, horizontal, and height axes represent the residue number, titration point, and chemical shift perturbation, respectively. NSH3, SH2, CSH3, and the linker regions are shown in green, orange, blue, and grey, respectively. (c) Chemical shift perturbation upon the titration with S4, S5, S9, and S10 PRMs represented on the crystal structure of GRB2 (PDB ID: 1GRI).

Fig. 3. Ligand-binding equilibrium models assuming multiple coupling modes. (a) Schematic illustration of the ligand-binding equilibrium models assuming to include two-binding sites. The ligand-binding equilibrium models with two-binding sites are composed of four stages: ligand-free (P), ligand-bound on one site (L–P and P–L), and ligand-bound on two sites (L–P–L) (left). (b) Schematic illustration of the ligand-binding equilibrium models (left), assuming induced fit (upper) and conformational selection (bottom).

Fig. 4. Plots of chemical shift perturbation of backbone ¹H^N and ¹⁵N nuclei of GRB2 upon the titration with the EGFR phosphorylated peptide. (a) Bar plots of chemical shift perturbation of backbone ¹H^N and ¹⁵N nuclei of GRB2 upon the titration of the EGFR phosphorylated peptide without (upper) and with (lower) SOS1 S4 PRM. The mean shift difference Δδ_ave was calculated as [(Δδ¹H^N)² + (Δδ¹⁵N/5)²]^1/2 where Δδ¹H^N and Δδ¹⁵N are the chemical shift differences (ppm) between GRB2 on its own and in the presence of the PRMs. The bar graphs are colour-coded according to the protein–peptide concentration ratio and are overlaid. The proline residues as well as the residues for which ¹H^N–¹⁵N correlation cross-peaks were not analysed due to signal overlap or other reasons are shown in grey. The secondary structures of GRB2 are also shown. (b) Chemical shift perturbation upon the titration with the EGFR phosphorylated peptide without (upper) and with (lower) SOS1 S4 PRM represented on the crystal structure of the GRB2 SH2 domain (PDB ID: 1BMB).

Fig. 5. Comparisons of residue-specific K_Ds and docking simulations of S4 and S10 PRMs between NSH3 and CSH3. The top five structures of SOS1 PRM S4 (a) and S10 (b) with the lowest energy from the docking simulations are presented for NSH3 (left) and CSH3 (right), with mapping of residue-specific dissociation constants (K_Ds) on the crystal structure of the NSH3 (PDBID: 1AZE) and CSH3 domains (PDB ID: 1IO6). The magnitudes of residue-specific K_Ds are colour-coded both on the structures and aligned amino acid sequences of the SH3 domains.

Fig. 6. Schematic illustrations of the interactions of GRB2 with the SOS1 PR domain and LAT. (a) A schematic illustration of GRB2 from free to complex. (b) A schematic illustration of liquid–liquid phase separation by GRB2, SOS1 and LAT. The thickness of dotted arrows indicates the strength of binding affinity.