Structural basis for R-spondin recognition by LGR4/5/6 receptors

Abstract

The R-spondin (RSPO) family of secreted proteins (RSPO1-RSPO4) has pleiotropic functions in development and stem cell growth by strongly enhancing Wnt pathway activation. Recently, leucine-rich repeat-containing G-protein-coupled receptor 4 (LGR4), LGR5, and LGR6 have been identified as receptors for RSPOs. Here we report the complex structure of the LGR4 extracellular domain (ECD) with the RSPO1 N-terminal fragment (RSPO1-2F) containing two adjacent furin-like cysteine-rich domains (FU-CRDs). The LGR4-ECD adopts the anticipated TLR horseshoe structure and uses its concave surface close to the N termini to bind RSPO1-2F. Both the FU-CRD1 and FU-CRD2 domains of RSPO1 contribute to LGR4 interaction, and binding and cellular assays identified critical RSPO1 residues for its biological activities. Our results define the molecular mechanism by which the LGR4/5/6 receptors recognize RSPOs and also provide structural insights into the signaling difference between the LGR4/5/6 receptors and other members in the LGR family.

Keywords: Wnt signal; complex; ligand/receptor interaction.

Figure 1.

Overall structure of RSPO1-2F (green) in complex with the LGR4-ECD (blue). Disulfide bonds are drawn as yellow sticks, and N-linked glycans are drawn as violet sticks.

Figure 2.

Structure of RSPO1-2F. (A) Ribbon model of RSPO1-2F, consisting of five “fingers” (F1–F5). (B) Sequence alignments show that cysteines in FU-CRD1 and FU-CRD2 are strictly conserved in RSPO1–RSPO4, and these two domains have the same disulfide bond pattern. RSPO1 residues involved in the interaction with the LGR4-ECD are colored green. (C) Consistent with the same disulfide bond pattern, structural superimposition shows that FU-CRD1 (green) and FU-CRD2 (violet) have similar overall folds, although the F6 finger is missing in the RSPO1-2F model due to protein expression truncation and weak electron densities.

Figure 3.

Structure of the LGR4-ECD. (A) Ribbon model of the LGR4-ECD, consisting of LRRNT (red), 17 LRR modules (blue), and LRRCT (red). Disulfide bonds and N-linked glycans are represented by yellow and orange sticks, respectively. Two smooth curved β sheets in the concave surface, LRR1–LRR10 (β sheet1) and LRR13–LRR17 (β sheet 2), are separated by LRR11 and LRR12 modules. Zoomed views of LRRNT and LRRCT are shown in B and C, respectively. The disordered Asp477–Ile519 region in the LRRCT is represented by a dashed line.

Figure 4.

Binding interface and effects of single-point mutations of RSPO1-2F at the interface for binding the LGR4-ECD and enhancing Wnt/β-catenin signaling. (A) Subinterface I between FU-CRD1 of RSPO1-2F and LGR4-ECD. Charged RSPO1 residues Asp85 and Arg87 and LGR4 residues Arg135, Asp137, Asp161, and Asp162 form salt bridge interactions that are represented by dashed lines. (B) Subinterface II between FU-CRD2 of RSPO1-2F and the LGR4-ECD. (Left panel) Hydrophobic interactions of RSPO1 residues Phe106 and Phe110 with LGR4 residues His157, Trp159, Ala181, Val204, and Val205. (Right panel) Hydrophilic interactions of RSPO1 residues His108, Asn109, Lys122, and Arg124 with LGR4 residues Asn226, Thr229, Lys251, and Glu252. (C) Binding affinities (K_D) of wild-type RSPO1-2F and its mutants with the LGR4-ECD. Sensograms with sample concentrations are shown in Supplemental Fig. 1A. (D) Wnt3a-induced STF reporter assay. STF reporter assays were carried out in HEK293T cells. Plasmids used per well were 15 ng of SuperTopFlash and 0.5 ng of pRL-TK. Protein samples were added with triplicates as indicated. The error bar indicates the SEM of three independent measurements. Full-length human RSPO1 (RSPO1) was used as control. (CM) Conditioned medium; (NC) stands for negative control; (WT) wild-type RSPO1-2F.