Structure of the SOCS4-ElonginB/C complex reveals a distinct SOCS box interface and the molecular basis for SOCS-dependent EGFR degradation

Abstract

Tyrosine kinase signaling is tightly controlled by negative feedback inhibitors including suppressors of cytokine signaling (SOCS). SOCS assemble as SH2 domain substrate recognition modules in ElonginB/C-cullin ubiquitin ligases. In accordance, SOCS4 reduces STAT3 signaling from EGFR through increased receptor degradation. Variable C-termini in SOCS4-SOCS7 exclude these family members from a SOCS2-type domain arrangement in which a strictly conserved C terminus determines domain packing. The structure of the SOCS4-ElonginC-ElonginB complex reveals a distinct SOCS structural class. The N-terminal ESS helix functionally replaces the CIS/SOCS1-SOCS3 family C terminus in a distinct SH2-SOCS box interface that facilitates further interdomain packing between the extended N- and C-terminal regions characteristic for this subfamily. Using peptide arrays and calorimetry the STAT3 site in EGFR (pY(1092)) was identified as a high affinity SOCS4 substrate (K(D) = 0.5 microM) revealing a mechanism for EGFR degradation. SOCS4 also bound JAK2 and KIT with low micromolar affinity, whereas SOCS2 was specific for GH-receptor.

Figure 1

Structure of the SOCS4-ElonginC-ElonginB Ternary Complex and Structure-Based Sequence Alignment (A) The different SOCS4 domains are highlighted by color with the N-terminal ESS shown blue, the SH2 shown orange, and the SOCS box shown red. A structured region from the N-terminal hexahistidine tag that was protected from proteolytic cleavage in solution is colored gray. The other complex components ElonginC and ElonginB are colored yellow and green, respectively. (B) Secondary structure elements in SOCS4 are shown above the sequence alignment. Structure-determined residues from SOCS2–SOCS4 are shown in bold; unstructured insertions and homologous SOCS sequences are gray, while conserved residues are boxed. An asterisk marks the insertion of R383 in the SOCS4 hinge, which stabilizes the SOCS4 domain organization. This and other interface residues are highlighted by different colors in the alignment.

Figure 2

Structural Comparison of the SH2 Substrate Pockets in SOCS2, SOCS3, and SOCS4 (A) Overlay of the crystal structures of the SOCS4 SH2 (orange) and the SOCS2 SH2 (blue, PDB code: 2C9W). The binding site of the phosphotyrosine moiety in SOCS2 is indicated by the presence of a bound sulfate ion. SOCS4 residues at this site are shown in ball-and-stick representation with their potential hydrogen bonding. (B) Overlay of the crystal structures of the SOCS4 SH2 (orange) and the murine SOCS3 gp130 complex (blue, PDB code: 2HMH). The phosphotyrosine and three C-terminal residues from the murine gp130-derived peptide are colored green. N-terminal tag residues from a crystallographic SOCS4 neighbor occupy the same SH2 pocket in the SOCS4 structure forming substrate mimetic interactions and are colored yellow. (C) Surface representation of the SH2 substrate pocket in SOCS2, SOCS3, and SOCS4 colored by electrostatic potential. The bound sulfate ion identifies the phosphotyrosine binding site in SOCS2 (left). The gp130 peptide is shown in complex with SOCS3 (center) and docked onto the SOCS4 surface (right) by the overlay of the two structures (for SOCS4 only the gp130 residues corresponding to the pY−1 to pY+3 positions are shown).

Figure 3

Alternative Domain Organization in the SOCS2 and SOCS4 Ternary Complexes (A) Comparison of the SOCS2-ElonginB/ElonginC and SOCS4-ElonginB/ElonginC structures highlighting the switch in packing between the SOCS2/SOCS4 C terminus and the N-terminal ESS helix (colored as in Figure 1). (B) Molecular interactions stabilizing the domain organization in SOCS4. (C) Structural overlay of the SOCS2 and SOCS4 SOCS box showing an 80° rotation of the SOCS4 SH2 domain relative to the SOCS2 SH2. The positions of the SOCS2 and SOCS4 phosphotyrosine pockets are indicated by a SOCS2-bound sulfate ion and a SOCS4-bound phosphotyrosine (modeled as in Figure 2). For clarity, the SOCS2 ESS is omitted and only the N-terminal half of each SH2 domain is shown. An asterisk denotes the position of the hinge point for rotation which occurs at R383/T384 in SOCS4. (D) Structural overlay of the SOCS2 and SOCS4 SH2 domains showing the alternative packing sites for the respective SOCS box domains on opposite faces of the ESS and SH2. The SOCS2-bound sulfate ion indicates the position of the SH2 phosphotyrosine pocket.

Figure 4

SOCS4 Is a High Affinity Binding Partner for EGFR pY¹⁰⁹² (A) OPAL membranes showing the substrate binding specificity of SOCS2 (top) and SOCS4 (below). Bound GST-SOCS-ElonginB/C complexes were detected by using anti-GST-HRP antibody. The large surface area of the SOCS complexes may contribute to the strong background binding to R/F/Y/W. Consequently, the preference for these residues is not considered here. (B) Comparison of the determined SOCS consensus recognition sequences and their suggested physiological targets. By sequence and functional similarity, the SOCS family members can be grouped into pairs as shown. Brackets ([ ]) indicate preferred amino acids, whereas gray braces ({ }) indicate nonpreferred amino acids. X denotes any amino acid, and Φ denotes any hydrophobic residue. †, De Souza et al. (2002); ‡, Krebs et al. (2002); ¥, Rodriguez et al. (2004); §, Wiederkehr-Adam et al. (2003); ^∗, Barber et al. (2001). (C) The SOCS4 binding affinity for EGFR pY¹⁰⁹² was determined by ITC (K_D = 0.5 μM). Data for peptide titrations into SOCS4 are colored black, while data from a control experiment are colored blue and offset for clarity in the top panel. (D) Structural model of the SOCS4-EGFR complex. The complex was modeled by using the homologous SOCS3-gp130 crystal structure (PDB code: 2HMH) as a template peptide for EGFR, and side chains were optimized by using ICM-PRO (Abagyan et al., 1994). The SOCS4 F373 side chain (BG loop) is solvent exposed and was relaxed to open the +4 position. A complete model for the SOCS2-GHR complex could not be built because in the apo SOCS2 structure, the flexible EF loop is folded into the +2 site to block the path of the peptide. The preferred −1 aspartate can pack between SOCS2 K59 (αA6) and T93 (βD3) to maximize its hydrogen bonding potential and similar residues are present in CIS (R93 and T127). The SOCS2 hydrophobic selection at positions C-terminal to the phosphotyrosine can be understood from the presence of leucine at βD5, βE4, and BG3 (Bullock et al., 2006).

Figure 5

SOCS4 Interdomain Packing Is Similar to Cbl and STAT Family Proteins (A) The Cbl αN helix (cyan) is structurally equivalent to the SOCS4 ESS (blue), having similar length and position. This structure connects the Cbl SH2 (pale yellow) and EF-hand (purple) domains and adopts a similar position to the bound helical peptides in calmodulin structures. Further, the Cbl helices αE1, αF2, and αF1 are placed similarly to SOCS box H2, H3 (red), and ElonginC H4 (green), respectively, although they show alternative topology. The SOCS4 SH2 domain is colored orange. (B) The packing in Cbl has been likened to the STAT family in which the linker domain fulfills an equivalent packing role to the Cbl EF hand (Meng et al., 1999). Overlay of the STAT1 (pale yellow) and SOCS4 (orange) SH2 domains reveals structural similarity between the STAT1 α11 (cyan) and α7 helices (purple) and the SOCS4 ESS (blue) and H1 (red), respectively.