Biochemical mechanisms of vertebrate hedgehog signaling

Abstract

Signaling pathways that mediate cell-cell communication are essential for collective cell behaviors in multicellular systems. The hedgehog (HH) pathway, first discovered and elucidated in Drosophila, is one of these iconic signaling systems that plays many roles during embryogenesis and in adults; abnormal HH signaling can lead to birth defects and cancer. We review recent structural and biochemical studies that have advanced our understanding of the vertebrate HH pathway, focusing on the mechanisms by which the HH signal is received by patched on target cells, transduced across the cell membrane by smoothened, and transmitted to the nucleus by GLI proteins to influence gene-expression programs.

Keywords: Cholesterol; Hedgehog signaling; Morphogen; Patched; Primary cilium; Smoothened.

Fig. 1.

Overview of HH signaling. (A) HH signaling regulates a bi-functional transcription factor that can repress (GLI-R) or activate (GLI-A) the transcription of target genes. HH ligands bind and inhibit the function of their receptor PTCH, allowing SMO to adopt an active conformation. SMO transmits the HH signal across the membrane and antagonizes the function of two negative regulators, SUFU and PKA, which promote GLI-R formation. Consequently, full-length GLI proteins (GLI-FL) are converted to GLI-A. (B) All HH ligands are modified with a cholesteroyl group at their C termini, attached through an auto-proteolytic reaction catalyzed by the C-terminal domain, and a palmitoyl group at their N termini, attached by a membrane-bound O-acyltransferase. (C) Vertebrate HH signaling is associated with protein trafficking events at primary cilia. When the HH pathway is ‘off’ (left), PTCH is enriched in cilia and inhibits SMO. PKA and SUFU restrain GLI activity and promote its proteolysis into GLI-R. HH signaling is turned on in target cells (right) when HH ligands inhibit PTCH and induce its clearance from primary cilia. As a result, SMO is activated and accumulates in cilia in association with a scaffolding complex, the Ellis van Creveld (EvC) complex. Activated SMO antagonizes the inhibitory effect of PKA on the GLI proteins, leading to the dissociation of SUFU. Now, instead of being converted into GLI-R, GLI-FL can enter the nucleus and activate target gene transcription (GLI-A). The transition zone at the cilia base regulates receptor access to cilia, cilia tips form a compartment (marked by the kinesin KIF7) that regulates the GLI proteins, and the EvC complex scaffolds SMO signaling near the cilia base (Box 1).

Fig. 2.

Structures of PTCH. (A) Structure of unliganded PTCH (PDB 6DMB; Gong et al., 2018) showing the transmembrane domain (TMD), which includes a sterol-sensing domain (SSD), and two extracellular domains (ECD1 and ECD2). A possible hydrophobic tunnel (ocher surface) is shown connecting two putative sterol-binding sites (meshed surfaces, asterisks) in ECD1 and the SSD. (B) Structure of the asymmetric 1SHH:2PTCH complex (adapted from PDB 6E1H; Qi et al., 2018b) reveals two distinct SHH-PTCH interfaces. PTCH1 molecule 1 (mol1) binds to SHH at an interface including its calcium- and zinc-binding sites and PTCH1 molecule 2 (mol2) engages the N-terminal palmitoyl and C-terminal cholesteroyl modifications of SHH, which are inserted into the PTCH protein core. The interaction of SHH with mol1 drives PTCH endocytosis and the palmitate-centered interaction with mol2 inactivates the transporter function of PTCH. (C-F) Structures of SHH in complex with the PTCH ECD (C; adapted from PBD 6E1H), CDO (D; PDB 3D1M; McLellan et al., 2008), HHIP (E; PDB 2WFX; Bishop et al., 2009) and heparin (F; PDB 4C4N; Whalen et al., 2013) reveal overlapping interfaces that would prevent simultaneous binding. Binding footprints for each protein on SHH are shown below the corresponding structure, with hydrophilic interactions in pink and hydrophobic interactions in brown.

Fig. 3.

Multi-domain structures of SMO bound to agonists and antagonists. (A,B) Human SMO (hSMO, blue) carrying an inactivating V329F mutation in the TMD bound to the agonist cholesterol (A; PDB 5L7D) in the CRD-binding site or to the antagonist vismodegib (B; PBD 5L7I) in the TMD site (Byrne et al., 2016). (C) hSMO bound to the TMD antagonist TC114 (PBD 5V57; Zhang et al., 2017). Arrows show movement of the CRD in the antagonist-bound structures (B,C) relative to the cholesterol bound structure (A). (D,E) Structures of Xenopus SMO (xSMO, green) bound to the agonist cholesterol (D; PDB 6D35; Huang et al., 2018) or the antagonist cyclopamine (E; PDB 6D32) are identical and show a dramatic re-orientation of the CRD relative to the TMD. Dotted circles (D,E) highlight a potential steric clash between the CRD and an N-linked glycan in the third extracellular loop of SMO. The N-linked glycans for Xenopus SMO were modeled because they were removed for crystallization. (F) Overlay of the indicated SMO structures showing rupture of the ionic lock between a tryptophan (W) and an arginine (R) residue resulting in the outward movement of TM6 (solid arrow) and the opening to a hydrophobic channel proposed to run through the xSMO TMD in the activated state (dotted arrow). Asterisks in all structures show the connections to the BRIL or flavodoxin domains interposed between TM5 and TM6 to facilitate crystallization. CRD, cysteine-rich domain; ECL3, extracellular loop 3; LD, linker domain; TMD, transmembrane domain.

Fig. 4.

Models for how PTCH inhibits SMO. (A) Schematic of PTCH and SMO embedded in a model lipid bilayer. Structurally identified cholesterol molecules bound to PTCH and SMO are depicted as yellow spheres. Two potential sterol-binding sites on SMO identified by computational methods are shown as green surfaces. Three possible sterol transport paths are shown by black arrows. In model 1, PTCH reduces the abundance or accessibility of inner leaflet cholesterol, preventing it from interacting with the hydrophobic channel or the cytoplasmic sterol-binding site of SMO. In model 2, cholesterol moves through PTCH from the outer leaflet of the membrane to the ECD1 and eventually to a protein or membrane acceptor, thereby depleting the membrane of cholesterol. In model 3, PTCH accepts cholesterol from the SMO CRD (or another donor) and transports it to the membrane, thereby turning off SMO activity. (B) Models for how PTCH could deplete cholesterol from the ciliary membrane (thereby reducing its access to SMO) by transporting it between the two closely opposed membranes of the ciliary pocket.

Fig. 5.

GLI proteins are regulated by PKA. PKA is a conserved inhibitor of GLI proteins, and the strength of HH signaling is inversely correlated with the activity of PKA in cells. The pathway regulating cAMP levels and PKA activity in cells is shown: proteins that increase PKA activity (red background) inhibit HH signaling, whereas those that decrease PKA activity (green background) enhance HH signaling. Among positive regulators of signaling, GPCRs coupled to Gαi reduce cAMP synthesis by inhibiting AC, phosphodiesterases (PDEs) hydrolyze cAMP and ARHGAP36 inhibits PKA. GPCRs coupled to Gαs, such as GPR161, inhibit signaling by increasing cAMP synthesis by AC. The kinases GRK2 and GRK3 are strong positive regulators that may function either by downregulating GPCRs coupled to Gαs or by directly promoting SMO activity. The mechanism by which SMO antagonizes the PKA axis is not clear (dashed arrows with ‘?’), but may involve direct activation of Gαi, inhibition of a Gαs-coupled GPCR or activation of a Gαi-coupled GPCR. PKA is composed of catalytic (C) and regulatory (R) subunits.

Fig. 6.

Regulation of GLI proteins in HH signaling. (A) Domain structures of mouse GLI2 and GLI3 proteins. Regions of the protein involved in various biochemical reactions are annotated. (B) Sequence of reactions that convert full-length GLI2/3 (blue shading) into a transcriptional repressor (red shading) or a transcriptional activator (green shading). Phosphorylation by PKA primes further phosphorylation by GSK3β and CK1 and promotes recognition by βTRCP, the substrate recognition adaptor of the SCF E3 ubiquitin ligase (Tempe et al., 2006; Wang and Li, 2006). SCF^βTRCP-mediated ubiquitination targets GLI2/3 to the proteasome for degradation. Owing to the presence of a processing determinant domain (PDD, see A), GLI3 (and to a lesser extent GLI2) are subject to an unusual partial proteasomal degradation reaction that generates fragments that function as pure transcriptional repressors (Schrader et al., 2011). (C) Multiple states of GLI activity can be encoded by different patterns of GLI phosphorylation. Full phosphorylation of GLI at the PKA phosphorylation sites (red circles) drives proteolytic production of GLI-R and repression of target genes. Graded dephosphorylation at the PKA sites prevents formation of GLIR and increases the ability of GLI to activate transcription. Maximum GLI transcriptional activity is associated with a separate activating hyper-phosphorylation (green circles).