Cryo-EM structure of the activated RET signaling complex reveals the importance of its cysteine-rich domain

See allHide authors and affiliations

Science Advances  31 Jul 2019:
Vol. 5, no. 7, eaau4202
DOI: 10.1126/sciadv.aau4202
  • Fig. 1 Data collection at a 40° tilt angle enabled reconstruction of a cryo-EM map of the NRTN-GFRα2-RET complex at a 5.7-Å resolution.

    (A) Comparison of two-dimensional (2D) class averages obtained from data collected at 0° or 40° tilt angles. 2D classes from the 0° data collection showed severe orientation bias of the molecule, exclusively presenting the view from the top revealing a figure 8 shape. 2D classes from data collected at the 40° tilt angle show a broader distribution of views. (B) Cryo-EM map reconstruction at a 5.7-Å resolution and atomic model derived from fitting in the cryo-EM map. Density around the NRTN dimer (or cartoon representation) is colored orange, GFRα2 is in blue, and RETCLD1-4 is in pink and magenta. Density for the unmodeled RETCRD is depicted in gray. Most of the density could be modeled, and only RETCRD could not be traced (gray, center). The side view of the complex suggests the position of the plasma membrane (right). (C) Schematic close-up view of the calcium-binding site between RETCLD2 and RETCLD3 (left), GFRα2 domains 1–3 showing the electron density (center), and the NRTN-GFRα2 interface mediated by the NRTN finger domain (right).

  • Fig. 2 Details of the RETECDinteractions.

    (A) Unmodeled density (gray) and structural model of one protomer are shown. Three interfaces mediate the interactions between RETECD and GFRα2 and/or NRTN. Interface 1 (bottom left) comprises residues in the interaction surface between GFRα2D1 and RETCLD1. Selected side chains of residues involved in the interaction are depicted as sticks and labeled. The second interface (bottom right) includes residues around the calcium-binding site between RETCLD2 and RETCLD3 and residues in GFRα2D3. Selected side chains of interacting residues are depicted as sticks and labeled. Interface 3 (top right) is located at the center of the heterohexamer, where RETCRD is in contact with GFRα2 and the NRTN dimer. In GFRα2, a loop bearing residues 187 to 190 (187REIS190) is close to RETCRD (top right). In NRTN, the loop 117LGYA120, is in contact with RETCRD and is of special interest because the tyrosine residue (Y119) is semiconserved among all GFLs that all have an aromatic residue at this position (sequence alignment). A second loop (NRTN), bearing residues E135 and R139, is also close to RETCRD. (B) RETCRD C termini meet at the membrane-facing surface of the NRTN homodimer, close to the positively charged heparin-binding surface, forming a mini domain as a part of RETCRD. The RETCRD C terminus is rich in negatively charged residues (sequence) and could form tight electrostatic interactions with the positively charged cleft on the NRTN surface. The homotypic RETCRD interaction also suggests a mechanism for the activation of the constitutively active RET mutant C634R, which is common in patients with MEN2A (7, 28, 29). In this case, C630 remains unpaired and could form an intermolecular disulfide bond with the opposite RETCRD molecule. The right panel is rotated by 90° relative to the left panel, illustrating the proximity of the RETCRD mini domain to the membrane. The proposed location of the beginning of the transmembrane helices is highlighted by red asterisks.

  • Fig. 3 Mutational analysis of complex formation by SPR.

    (A and B) SPR was used to test the NRTN and GFRα2 mutants for binding to GFRα2 wild-type (wt) or NRTN wild type, respectively. As an example, we show NRTN Y119A binding to immobilized GFRα2 (B) in comparison to NRTN wild-type binding to immobilized GFRα2 (A). A list of KDs can be found in fig. S5F. KDs were determined by plotting the concentrations against the equilibrium responses (insets). (C) Sensorgram showing injection of equal concentrations of NRTN and RETECD to immobilized GFRα2. Since the response signal never reaches the equilibrium, the heterohexameric complex formation could not be analyzed quantitatively by SPR. However, the results show an increased response of approximately 150 RU compared to NRTN alone (A) and an apparent slower dissociation from the surface. Both observations are suggestive of heterohexameric complex formation. (D) Sensorgram showing injection of equal concentrations of the NRTN Y119A mutant and RETECD to immobilized GFRα2. The differences observed when comparing (C) with (A) were not observed for NRTN Y119A and RETECD binding to GFRα2 [(D) versus (B)]. A binding effect was observed at high protein concentrations (>1 μM), suggesting weaker binding affinity. This implicates that, under the conditions of the SPR experiment, the heterohexamer does not form with this NRTN mutant except at high protein concentrations.

  • Fig. 4 Analysis of complex formation and signaling capacity of wild-type and mutant proteins.

    (A) The mutated residues in interfaces 2 and 3 shown on the cryo-EM structure as viewed from the membrane. Modeled parts of the structure are shown as cartoon: NRTN (orange), GFRα2 (blue), and RETCLD1-4 (pink/magenta). Mutated residues are shown as spheres. The gray spheres represent the unmodeled RETCRD. (B) SEC-MALS ultraviolet (UV) traces after incubation of the NRTN-GFRα2 complex with RETECD (excess NRTN-GFRα2) show that a large peak, corresponding to the extracellular portion of the signaling complex, is formed with wild-type NRTN (yellow) as well as the NRTN mutants Y119A (red) and E135S/R139S (blue). Labeled peaks elute at the same retention volume as the complex components when run separately under the same conditions. An additional peak in the red trace (Y119A) represents RETECD that is not in complex. Incubation with the GFRα2 mutant N330 (gray) resulted in two major peaks, one corresponding to the elution volume of RETECD and one with a molecular weight lower than the hexameric complex. (C) SDS-PAGE gel showing samples from chemical cross-linking of heterohexameric complex with wild-type NRTN, Y119A, E135S/R139S, and GFRα2 N330A. Peptide mapping confirmed the identity of the SDS-PAGE gel bands. A larger band (marked with an asterisk), corresponding to approximately twice the size of the NRTN-GFRα2-RET complex, was shown by peptide mapping to also contain NRTN, GFRα2, and RETECD. This putative dimer was, however, not observed on SEC-MALS and is most likely a gel or cross-linking artifact. (D) The NRTN-induced activation of mitogen-activated protein kinase (MAPK) signaling through RET measured in a human neuroblastoma (TGW) cell–based activity reporter assay. The average of three separate experiments is reported, and each experiment was run with four replicates. The top concentration of NRTN resulted in a reduced signal, which is commonly observed in luminescence assays, and has therefore been excluded from calculation of EC50 and maximal signal. EC50 values are listed in Table 1. RLU, relative luminometer units.

  • Fig. 5 3D auto-refinement reveals four conformations differing in angles of ring-like structures toward the NRTN dimer.

    (A) Four classes of particles (minimum of 22,000 per class) were selected after 3D refinement and compared to each other. The overall structure of each protein subunit, NRTN, GFRα2, and RETECD, was not altered, but the position of the ring-like structures (GFRα2 and RETECD) in relation to the NRTN dimer differed between the classes. Two kinds of movement could be observed, one along the x axis and one along the z axis. Angles varied from 114° to 121° along x and 133° to 147° along z. (B) Comparison of the four models derived from maps in (A). One protomer from class 8 is colored by backbone root mean square displacement (RMSD) values compared to class 8. Top: Top view with NRTN superposed. Most movement occurs in RETCLD3 and RETCLD4. Bottom: A 90° rotated view.

  • Table 1 EC50values of NRTN mutants and wild type.

    The EC50 value represents the protein concentration required to obtain the half-maximal downstream signaling effect in TGW cells. Span of maximal signal of luciferase activity listed. Average of three experiments (Fig. 4D), four replicates per experiment. Reported SE is SEM.

    ProteinEC50 (ng/ml)SEMax signal

Supplementary Materials

  • Supplementary material for this article is available at

    Fig. S1. SEC-MALS of the wild-type extracellular signaling complex.

    Fig. S2. Cryo-EM data processing.

    Fig. S3. RETECD glycosylations.

    Fig. S4. SEC of the heterohexameric complex and its components.

    Fig. S5. Biophysical analysis of complex formation.

    Fig. S6. Related GFL-GFRα crystal structures display varying angles of GFRα positions in relation to the GFL center.

    Table S1. Cryo-EM data collection, refinement, and validation statistics.

  • Supplementary Materials

    This PDF file includes:

    • Fig. S1. SEC-MALS of the wild-type extracellular signaling complex.
    • Fig. S2. Cryo-EM data processing.
    • Fig. S3. RETECD glycosylations.
    • Fig. S4. SEC of the heterohexameric complex and its components.
    • Fig. S5. Biophysical analysis of complex formation.
    • Fig. S6. Related GFL-GFRα crystal structures display varying angles of GFRα positions in relation to the GFL center.
    • Table S1. Cryo-EM data collection, refinement, and validation statistics.

    Download PDF

    Files in this Data Supplement:

Stay Connected to Science Advances

Navigate This Article