Hedgehog Pathway Activation Requires Coreceptor-Catalyzed, Lipid-Dependent Relay of the Sonic Hedgehog Ligand

Abstract

Hedgehog signaling governs critical processes in embryogenesis, adult stem cell maintenance, and tumorigenesis. The activating ligand, Sonic hedgehog (SHH), is highly hydrophobic because of dual palmitate and cholesterol modification, and thus, its release from cells requires the secreted SCUBE proteins. We demonstrate that the soluble SCUBE-SHH complex, although highly potent in cellular assays, cannot directly signal through the SHH receptor, Patched1 (PTCH1). Rather, signaling by SCUBE-SHH requires a molecular relay mediated by the coreceptors CDON/BOC and GAS1, which relieves SHH inhibition by SCUBE. CDON/BOC bind both SCUBE and SHH, recruiting the complex to the cell surface. SHH is then handed off, in a dual lipid-dependent manner, to GAS1, and from GAS1 to PTCH1, initiating signaling. These results define an essential step in Hedgehog signaling, whereby coreceptors activate SHH by chaperoning it from a latent extracellular complex to its cell-surface receptor, and point to a broader paradigm of coreceptor function.

Keywords: Hedgehog; chaperone; coreceptor; lipids; morphogen; receptor; signaling.

Figure 1.. SHH Is Released as a Stable Complex with SCUBE Proteins

(A) Tagged SCUBE2 was affinity purified from conditioned media and was analyzed by SDS-PAGE and Coomassie staining. (B) Purified SCUBE2 was added to SHH-producing HEK293T cells and released SHH was quantified by immunoblotting. SCUBE2 releases SHH in a dose-dependent manner. See Figure S1J for SHH release by SCUBE1 and SCUBE3. (C) Models for SCUBE2 involvement in SHH release. Left: SCUBE2 stimulates SHH release, by promoting multimerization or removal of lipid termini. Right: SCUBE2 remains bound to SHH, as chaperone. (D) As in (A), but with SHH co-expression. SHH copurifies with SCUBE2. (E) As in (D), but with SHH detection by immunoblotting. (F) Purified SCUBE2-SHH was incubated with 1.5 M NaCl, 0.5% Triton X-100, or 0.5% DDM, and then analyzed by Blue Native PAGE. Brackets indicate molecular weight ranges for SCUBE2-SHH and SHH. The complex is disrupted by detergents, but not by high ionic strength. (G) Wild-type MEFs were incubated with two doses of purified SCUBE2-SHH, palmitoylated SHH-N, unlipidated SHH-N(C24A), or SAG, and intensity of endogenous SMO in cilia was measured by immunofluorescence microscopy. Hh signaling is triggered by SCUBE2-SHH, SHH-N, and SAG, but not by SHH-N(C24A). Data are normalized between median ciliary SMO intensity for untreated cells and cells treated with the higher SAG dose (100%). Box plots represent median and first and third quartiles of SMO intensity at cilia. At least 900 cilia were measured per condition. (H) Representative images of cilia for the experiment in (G). Scale bar, 2 μm. (I) As in (G), but Hh signaling was measured by qRT-PCR for Gli1. Bars represent average fold-change for three replicates, and error bars represent SEM. See also Figures S1 and S2 for additional characterization of purified SCUBE2 and SCUBE2-SHH complexes.

Figure 2.. SCUBE2 Blocks the Palmitate-Dependent Interaction between SHH and PTCH1

(A) Purified SCUBE2-SHH or SHH-N was added to HEK293T cells expressing EGFP-tagged PTCH1, and bound ligand was quantified by anti-SHH (5E1) immunofluorescence. SCUBE2 reduces SHH affinity for PTCH1. Data are normalized between background signal (untreated cells) and maximum SHH-N binding (defined as 100%), and are fit with a three-parameter curve. Points represent average binding for four replicates, and error bars represent SEM. At least 200 cells were measured per replicate. (B) Schematic of SHH-PTCH1 interactions (PDB: 6RVD). SHH engages PTCH1 through a protein-protein interaction, involving the SHH pseudo-active site (PAS) and the second extracellular loop (L2) of PTCH1, as well as through lipid-protein interactions, involving the palmitoylated N-terminal peptide (EP) of SHH and the cholesterol-modified C terminus of SHH. Mutants specifically defective in the protein-protein (SHH-N^PAS*, PTCH1^L2*) or palmitate-protein (SHH-N^ΔEP, PTCH1^Gorlin) interaction between SHH and PTCH1 are tested below. (C) As in (A), but with binding of fluorescently labeled SHH-N variants. SHH-N^PAS* and SHH-N^ΔEP have reduced affinity for PTCH1. At least 100 cells were measured per replicate. (D) As in (A), but with binding of SCUBE2-SHH to wild-type or mutant PTCH1. Binding of SCUBE2-complexed SHH depends on an intact SHH-PTCH1 protein-protein interface. At least 500 cells were measured per replicate. (E) As in (C), but comparing binding of wild-type or mutant SHH-N (150 nM), to wild-type or mutant PTCH1. Disruption of both SHH-PTCH1 interfaces results in synergistic binding defects. For each SHH-N variant, data are normalized between background signal (untreated cells) and binding to wild-type PTCH1 (100%). Box plots show the median and the first and third quartiles of bound ligand intensity. At least 300 cells were measured per condition. See Figures S3N–S3P for complete dose-response curves. (F) Model of SCUBE2-SHH interaction with PTCH1. SCUBE2, which sequesters SHH lipid moieties, blocks the EP-dependent interaction of SHH with PTCH1, reducing affinity of SHH for PTCH1. See Figure S3 for additional characterization of the ceii-based iigand-receptor-binding assay.

Figure 3.. Coreceptors CDON/BOC and GAS1 Are Necessary for Signaling by SCUBE2-SHH

(A) Possible function of the three SHH coreceptors, CDON, BOC, and GAS1. CDON and BOC are homologous single-pass transmembrane proteins of the immunoglobulin superfamily. GAS1 is a GPI-anchored protein with homology to GDNF receptors. SHH coreceptors might relieve inhibition of SHH by SCUBE2, allowing SHH lipids to engage PTCH1. (B) Wild-type or coreceptor-null MEFs (Mathew et al., 2014) were treated with saturating doses of SCUBE2-SHH conditioned medium, SHH-N conditioned medium, or SAG, and Hh pathway activation was measured by endogenous SMO recruitment to cilia. Coreceptor-null MEFs do not respond to SCUBE2-SHH, but respond to SHH-N and SAG. Data are normalized between ciliary SMO for untreated cells and cells treated with saturating SAG (100%). Box plots represent median and the first and third quartiles of SMO intensity. At least 500 cilia were measured per condition. (C) As in (B), but Hh signaling was measured by qRT-PCR for Gli1. For each line, data are normalized between response for untreated cells and cells treated with saturating SAG (100%). Bars represent average fold-change for three replicates, with error bars indicating SEM. (D) Dose-response of purified SCUBE2-SHH or SHH-N on wild-type and coreceptor-null MEFs. For each line, data are normalized between ciliary SMO for untreated cells and the theoretical maximum (100%), as fit with a four-parameter curve. Points represent average ciliary SMO for three replicates, and error bars represent SEM. At least 100 cilia were measured per replicate. (E) As in (D), but with wild-type, GAS1-null, CDON/BOC-null, or coreceptor-null MEFs. GAS1-null MEFs exhibit a severe defect in responsiveness to SCUBE2-SHH, while CDON/BOC-null cells exhibit a modest defect. (F) As in (D), but treating wild-type MEFs, GAS1-null MEFs, or GAS1-null MEFs rescued with overexpressed GAS1. GAS1-null MEFs exhibit a ~30-fold reduction in SCUBE2-SHH EC50, which is fully rescued upon GAS1 expression. Data are normalized between ciliary SMO for untreated cells and cells treated with saturating levels of SHH-N (100%). (G) As in (B), but including GAS1-null MEFs. GAS1-null MEFs have a defect in responding to SCUBE2-SHH, but respond normally to SHH-N and SAG. At least 200 cilia were measured per condition. (H) As in (B), but with coreceptor-null MEFs rescued with CDON, GAS1, or both. GAS1 fully rescues responsiveness to SCUBE2-SHH, while CDON rescues the response only partially. (I) As in (H), but Hh signaling was measured by qRT-PCR for Gli1. Data are represented as in (C). (J) As in (H), but with serial dilutions of SCUBE2-SHH-conditioned medium. Data are represented as in (D). CDON and GAS1 synergize at lower SCUBE2-SHH dose (1:640 dilution). (K) As in (J), but cells were treated with SCUBE2-SHH conditioned medium at 1:640 dilution, and Hh signaling was measured by qRT-PCR for Gli1. Data are represented as in (C).

Figure 4.. CDON/BOC Promote Hh Signaling by Binding SCUBE2-SHH

(A) Fluorescently labeled SCUBE2 (300 nM) was incubated with cells expressing EGFP-tagged SHH-binding proteins, and bound ligand was quantified by fluorescence microscopy. SCUBE2 binds to CDON/BOC, but not to other SHH interactors. Data are normalized between binding to EGFP-tagged SMO (negative control) and the highest bound signal (100%). Box plots represent median and the first and third quartiles of binding. At least 400 cells were measured per condition. (B) As in (A), but using fluorescently labeled SHH-N (3 μM). (C) Fluorescently labeled SCUBE2 (30 nM) was incubated with CDON-expressing cells, in the presence of increasing doses of unlabeled SCUBE2, SHH-N, or FLAG-Halo Tag7 protein (FLAG-HT7, negative control). SCUBE2 and SHH-N do not compete for binding to CDON. Data are represented as in (A) but normalized between background signal (untreated cells) and binding in the presence of negative control competitor (100%). (D) Binding of fluorescently labeled SCUBE2, alone or in complex with SHH, to cells expressing EGFP-tagged CDON. SCUBE2 binds to CDON with mid nanomolar affinity. Data are normalized between background signal (untreated cells) and maximum binding (100%) and are fit with a three-parameter curve. Points represent average binding for four replicates, and error bars represent SEM. At least 300 cellswere measured per replicate. See Figure S4A for binding of SCUBE1 and SCUBE3. (E) As in (D), but for BOC. SCUBE2 affinity for CDON and BOC is very similar. (F) Asin (D), but using anti-SHH (C9C5) immunofluorescence to compare SHH-N to SCUBE2-SHH. SHH in complex with SCUBE2 has an affinity for CDON similar to that of SCUBE2 alone, considerably higher than that of SHH-N alone. (G) As in (F), but for BOC. (H) As in (A), but with binding of fluorescently labeled SCUBE2 (30 nM) to CDON truncation mutants. Data are normalized between binding to SMO (negative control) and to full-length CDON (100%). SCUBE2 binds to the first and second FN(III)-like domains (FN1,2) of CDON. See Figure S4G for the corresponding BOC experiment. (I) As in (H), but using fluorescently labeled SHH-N (300 nM). SHH-N binds to the third FN(III)-like domain (FN3) of CDON, as reported (Tenzen et al., 2006; Yao et al., 2006). See Figure S4H for the corresponding BOC experiment. (J) As in (H), but using fluorescently labeled SCUBE2 (1.5 μM), alone or in complex with SHH (300 nM). SCUBE2 is recruited to CDON-FN3 via SHH. At least 600 cells were measured per condition. See Figures S4M and S4N for evidence of recruitment of SHH to CDON by SCUBE2. (K) As in (J), but using anti-SHH (C9C5) immunofluorescence to compare SHH-N with SCUBE2-SHH. (L) Dose-response of SCUBE2-SHH conditioned medium on coreceptor-null MEFs or coreceptor-null MEFs rescued with CDON truncation mutants. Hh pathway activation was measured by endogenous SMO recruitment to cilia. The SCUBE2-interacting CDON-FN1,2 domain enhances signaling more strongly than the SHH-interacting CDON-FN3 domain. For each line, data are normalized between ciliary SMO for untreated cells and cells treated with saturating SAG (100%). Upper dotted line represents maximum signaling by SCUBE2-SHH on wild-type cells. Data are fit with a three-parameter curve. Points represent average ciliary SMO for three replicates, and error bars represent SEM. At least 100 cilia were measured per replicate. See also Figure S4O, showing reduced activity of SHH complexed with a SCUBE2 mutant defective in CDON binding. (M) Proposed model of CDON/BOC function. CDON/BOC bind both SCUBE2 (via FN1,2) and SHH (via FN3) to recruit the complex to the surface of responding cells. See Figure S4 for additional characterization of the SCUBE2-CDON/BOC interaction.

Figure 5.. GAS1 Is a Palmitate-and Cholesterol-Binding Protein that Unloads SHH from SCUBE2

(A) Fluorescently labeled SCUBE2 (600 nM), alone or in complex with SHH(HPC7) (300 nM), was incubated with cells expressing EGFP-tagged GAS1, scFv5E1::TM (positive control), or SMO (negative control), and bound SCUBE2 was measured by fluorescence microscopy. Even when complexed with SHH(HPC), SCUBE2 is not recruited to GAS1, in contrast to scFv5E1::TM. Data are normalized between binding of SCUBE2-SHH(HPC) to the negative and positive controls. At least 600 cells were measured per condition. (B) As in (A), but using anti-HPC immunofluorescence to compare binding of SCUBE2-SHH(HPC7) and SHH-N(HPC7). SHH(HPC) accumulates, without SCUBE2, on GAS1-expressing cells incubated with SCUBE2-SHH(HPC). (C) Purified SCUBE2-SHH (400 nM) was incubated at room temperature, with or without purified GAS1 ectodomain (GAS1-Ecto) (2 μM), prior to separation by blue native PAGE and immunoblotting. SHH is transferred from SCUBE2 to GAS1-Ecto. (D) As in (C), but SCUBE2-SHH was incubated with GAS-Ecto (10 μM) or CDON ectodomain (CDON-Ecto) (10 μM). GAS1-Ecto unloads SHH from SCUBE2, whereas CDON-Ecto forms a ternary complex. (E) Purified palmitoylated SHH-N (circles) or unpalmitoylated SHH-N(C24A) (squares) was added to cells expressing EGFP-tagged GAS1, PTCH1 (positive control), or SMO (negative control), and bound ligand was measured by fluorescence microscopy. GAS1 binds SHH-N in a palmitate-dependent manner. Data are normalized between binding to negative and positive controls and are fit with a three-parameter curve. Points represent average binding for four replicates, and error bars represent SEM. At least 200 cells were measured per replicate. (F) MEFs were treated with the indicated Hh pathway activators, in the presence of purified GAS1-Ecto (2 μM) or control competitor (FLAG-HT7) (2 μM). Recruitment of endogenous SMO to cilia was measured by immunofluorescence. GAS1-Ecto antagonizes signaling by SCUBE2-SHH and SHH-N. For each agonist, data are normalized between ciliary SMO for untreated cells and cells treated in the presence of control competitor (100%). Box plots represent median and the first and third quartiles of SMO intensity. At least 200 cilia were measured per condition. (G) Equimolar amounts of purified HPC-tagged GAS1-Ecto, SMO-CRD (positive control), or GFRα1-Ecto (negative control) were incubated with [³H]-cholesterol, and affinity captured on anti-HPC beads. Bound protein was eluted with HPC peptide (gel, bottom), and associated radioactivity was measured (graph, top). GAS1-Ecto binds cholesterol, similar to SMO-CRD. Bars represent average radioactivity for three replicates, and error bars represent SEM. (H) Purified GAS1-Ecto, SCUBE2, or BSA (negative control) was added (1 μM) to HEK-293T cells stably expressing NanoLuc luciferase-tagged SHH [SHH(NL)], and SHH(NL) release was measured as a function of time. Purified GAS1-Ecto releases dually lipidated SHH with a similar rate to SCUBE2. Bars represent average release rate across six time points, and error bars represent standard error of the linear fit to the release time course. (I) SCUBE2-SHH (400 nM) was incubated at room temperature with GAS1-Ecto (2 μM). After 2 h, when considerable SHH had transferred from SCUBE2 to GAS1, excess (10 μM) SCUBE2, GAS1-Ecto, or buffer was added, and the reaction allowed to continue for 2h (schematic, left). Samples were analyzed in as in (C), and the percentage of SHH associated with SCUBE2 (blue) and GAS1-Ecto (yellow) was quantified for each lane (graph, right). SHH transfer between SCUBE2 and GAS1-Ecto is driven by mass action. See Figure S6I for the corresponding blot. (J) SCUBE2 was loaded with [³H]-cholesterol and was captured on beads. The beads were incubated with purified GAS1-Ecto, SCUBE2, or negative controls (2 μM), and released radioactivity was measured as a function of time across four time points of a single measurement. Data are fit with a one-phase association curve. Both SCUBE2 and GAS1-Ecto accept cholesterol from SCUBE2. (K) As in (J), but with [³H]-cholesterol-loaded GAS1-Ecto on beads. Only GAS1-Ecto, but not SCUBE2, accepts cholesterol from GAS1-Ecto. See Figures S5 and S6 for further characterization of SHH transfer from SCUBE2 to GAS1-Ecto and of GAS1 lipid binding.

Figure 6.. CDON/BOC and GAS1 Cooperate to Unload SHH from SCUBE2 and Transfer SHH to PTCH1

(A) Two possible pathway architectures for SHH coreceptors. Left: CDON/BOC and GAS1 act in parallel to promote SHH reception upstream of PTCH1. Right: CDON/BOC and GAS1 act together, in series. (B) Purified CDON-SCUBE2-SHH ternary complex (400 nM) was incubated with purified GAS1-Ecto (2 μM), and the reaction was analyzed by blue native PAGE and immunoblotting. GAS1-Ecto unloads SHH from CDON-SCUBE2-SHH. See Figure S7E for purification of the CDON-SCUBE2-SHH complex. (C) Purified SCUBE2-SHH(NL) (20 nM) was incubated with beads bearing combinations of CDON-Ecto, GAS1-Ecto, and their respective negative controls (see schematic). After 1 h, beads were treated with PreScission protease, to remove all proteins except GAS1-Ecto, and bead-bound luminescence was measured. CDON-dependent recruitment of SCUBE2-SHH(NL) to beads drives SHH(NL) transfer from SCUBE2 to GAS1. Bars represent average for three replicates, and error bars represent SD. See also immunoblot in Figure S7F. (D) As in (C), but beads were treated with PreScission protease before or after incubation with SCUBE2-SHH(NL). CDON-Ecto is required to promote transfer of SHH(NL) from SCUBE2 to GAS1-Ecto. See also Figures S7G–S7I. (E) Dose-response of purified GAS1-Ecto-SHH and SCUBE2-SHH on wild-type and coreceptor-null MEFs. GAS1-Ecto-SHH is less potent than SCUBE2-SHH on wild-type cells, but more potent than SCUBE2-SHH on coreceptor-null cells. Data are normalized between ciliary SMO for untreated cells and the theoretical maximum (100%), as fit with a four-parameter curve. Points represent average ciliary SMO for three replicates, and error bars represent SEM. At least 100 cilia were measured per replicate. (F) As in (E) but comparing purified ALFA-tagged GAS1-Ecto-SHH complex on wild-type and coreceptor-null MEFs, stably expressing or not membrane-anchored ALFA nanobody (NbALFA::TM). Direct recruitment of GAS1-Ecto-SHH to the surface of coreceptor-null cells enhances signaling. (G) GAS1-Ecto-SHH or SCUBE2-SHH (500 nM) was incubated with cells expressing EGFP-tagged PTCH1 or SMO (negative control), and bound SHH was measured by anti-SHH (C9C5) immunofluorescence. More SHH is transferred to PTCH1 from GAS1 than from SCUBE2. Data are normalized between binding of SCUBE2-SHH to the negative and positive controls. At least 300 cells were measured per condition. (H) As in (G), but with GAS1-Ecto-SHH (500 nM) incubated with cells expressing PTCH1 mutants defective in the protein-protein (PTCH1^L2*) or palmitate-protein (PTCH1^Gorlin) interface with SHH. SHH is transferred normally from GAS1 to the PTCH1 mutants. Data are normalized between background signal (untreated cells) and binding to wild-type PTCH1 (100%). At least 1,000 cells were measured per condition. (I) GAS1-Ecto was loaded with [³H]-cholesterol and was captured on beads. The beads were incubated with purified GAS1-Ecto, SCUBE2, a soluble version of the first large extracellular loop of PTCH1 (PTCH1-L1), or HT7 negative control (2 μM), and released radioactivity was measured as a function of time across four time points of a single measurement. Data are fit with a one-phase association curve. Cholesterol is rapidly transferred from GAS1-Ecto to PTCH1-L1 and GAS1-Ecto, but not to SCUBE2. See Figure S7 for additional characterization of purified proteins, bead-based recruitment/transfer assay, and heterologous cell-surface recruitment assay.

Figure 7.. The Coreceptor-Catalyzed Pathway for Lipid-Dependent SHH-PTCH1 Complex Assembly

(A) Schematic of SHH interactions involved in SHH delivery. Interfaces between SHH and PTCH1 (Qi et al., 2018a; Qian et al., 2018; Rudolf et al., 2019) and CDON (McLellan et al., 2008) are based on published structures, while interfaces between SHH and SCUBE2 and GAS1 are drawn based on the present study. SHH engages in successive, mutually exclusive interactions during its movement from SCUBE2 to PTCH1. (B) SHH reception in wild-type cells, which contain both CDON/BOC and GAS1. CDON/BOC recruits SCUBE2-SHH to the cell surface, facilitating SHH transfer to GAS1, followed by transfer to PTCH1. (C) SHH reception in coreceptor-null cells. PTCH1 recruits SCUBE2-SHH only poorly to the cell surface, and accepts SHH from SCUBE2 at a low basal rate. (D) SHH reception in cells with CDON/BOC only. CDON/BOC recruit SCUBE2-SHH more readily than PTCH1 but do not enhance SHH transfer from SCUBE2 to PTCH1. (E) SHH reception in cells with GAS1 only. GAS1 does not recruit SCUBE2-SHH to the cell surface but catalyzes SHH transfer from SCUBE2 to PTCH1.