ReviewHedgehog signalling: Emerging evidence for non-canonical pathways
Introduction
Classic experimental embryology, beginning most famously with the identification of Spemann's organiser [1], showed that embryonic development is driven by the release of inductive signals from one tissue to pattern another. Early phenotype-driven mutagenesis screens in the fruit fly [2] and nematode worm [3] identified thousands of genes, representing a large proportion of the genome, that are involved in regulating pattern formation. However, the subsequent cloning of these genes and dissection of their functions established that they could be grouped into only a limited number of distinct signalling pathways that are used iteratively in many aspects of development ([4], [5]; for review).
Of these pathways, hedgehog (HH) signalling has been extensively linked to human disease, being found to be disrupted by mutation in human congenital malformations and cancers (reviewed in [6]), and emerging evidence also suggests that induction of HH signalling in adult tissues can stimulate tissue repair [7], [8], [9]. Almost three decades of genetic and biochemical studies have established the basic framework of the pathway, from ligand and receptor, to intracellular effectors and transcriptional output (Fig. 1A, Section 1.2). This paradigm is relatively simpler than that of other signal transduction pathways—only three essentially equivalent morphogens bind to predominantly one receptor which results in a transcriptional output via a linear pathway. In contrast, many other signalling pathways can be activated by a myriad of specific ligand/receptor partnerings, and different combinations can specifically induce different branches of the pathway that mediate very different cellular processes.
Understanding the intricacies of HH signalling thus promises great advances in the management and future treatment of patients with a range of genetic diseases, and a number of relevant pharmacological agents have been identified [10], [11], [12]. In applying these agents clinically, it will be important to understand how fine-tuning of the HH response, and how integration of HH signalling with other pathways, makes different tissues differentially susceptible to defective HH signalling. The aims of this review are: to examine recent evidence supporting added complexity in HH signal transduction in the form of non-canonical pathways; to identify unanswered questions and future avenues of enquiry; and to address the biological relevance of non-canonical signalling.
Key components of the HH signalling pathway were first isolated as mutants affecting larval cuticle patterning in the fruit fly, Drosophila melanogaster [2]. Cloning of the mutated genes and analysis of their function in segmentation and the patterning of wing imaginal discs [13], [14], [15], [16], [17], [18] has established a paradigm for HH signal transduction. The hedgehog (hh) gene encodes a precursor protein which undergoes autoproteolytic cleavage to generate C-terminal and N-terminal peptides [19], [20]; the latter subsequently undergoes palmitoylation and conjugation to a cholesterol moiety [19], [20] to generate the active morphogen [21], which can be secreted over many cell diameters [22], [23], [24].
Cells that are potentially responsive to HH express the 12-transmembrane domain receptor, Patched (PTC), and binding of secreted HH to PTC relieves the normal repression of a second transmembrane protein Smoothened (SMO) (Fig. 1A), by inducing a conformational change [25], [26]. Extensive genetic and biochemical analyses have demonstrated the importance of protein kinases in driving changes in the subcellular localisation of both SMO and the transcription factor, Cubitus Interruptus (CI), which mediates the canonical hedgehog response [27], [28], [29], [30], [31], [32]. Inhibition of PTC repressor function by HH leads to the cell surface accumulation of SMO, a process which requires phosphorylation at a number of Protein Kinase A (PKA) and Casein Kinase I (CKI) sites in the C-terminal tail of SMO [27]. This intracellular domain of SMO interacts with a protein complex that includes PKA, CKI, Fused (FU; another kinase), Suppressor of Fused (SUFU) and the kinesin-like molecule Costal-2 (COS2), which facilitates interaction of the complex with microtubules [28]. In the absence of morphogen, CI is phosphorylated and proteolytically cleaved into a transcriptional repressor. Activation of HH signalling prevents processing of CI, and thus it remains in its transcriptional activator form. If the levels of morphogen are sufficiently high this activator will additionally be released from the protein complex and will enter the nucleus to activate the expression of target genes: thus a cell can initiate graded responses depending on the concentration of morphogen sensed by the cell. Interestingly, ptc [33], together with several other pathway components more recently identified in vertebrates [34], [35], [36], [37], are themselves transcriptional targets of HH signalling, thus facilitating feedback mechanisms that help fine-tune the HH response and modulate the temporal response to HH morphogen [38].
Since the elucidation of the HH signalling pathway in the fruit fly, HH signalling has been linked to numerous developmental processes in vertebrates. This work has shown that the HH signalling pathway is largely conserved across phyla, albeit with an expansion in the number of paralogous genes as compared to the fruit fly (Fig. 1A). In mice and humans, three HH morphogens—Sonic (SHH), Indian (IHH) and Desert (DHH) hedgehog—are present. There are also two homologues of PTC—PTCH1 and PTCH2. It is generally accepted that most signalling seems to involve PTCH1, and mice that are homozygous for a Ptch2 mutation are largely normal [39]. The three HH morphogens are interchangeable in inducing HH signal transduction, although they differ in the affinity to which they bind PTCH1, and thus in their potency to induce a response [40]. While there is only one SMO orthologue in vertebrates, three CI homologues, known as the glioma-associated oncogenes (GLI1-3), have been identified. Gli2−/− mice do not correctly specify the ventral-most cells in the neural tube [41], [42], while this structure is dorsalised in Gli3−/− mutants [43]. Thus it has been suggested that GLI2 acts primarily as a transcriptional-activator while GLI3 acts primarily as a repressor and, in the neural tube, this is likely to be the result of differential spatial expression of Gli2 and Gli3 in the dorsal–ventral axis rather than intrinsically different biochemical properties. GLI1 also activates HH-responsive genes [44], [45], but lacks the N-terminal repressor domain, and so might not be considered a direct effector of HH signal transduction machinery per se. However, the GLI1 gene is a transcriptional target of positive HH signalling and so probably serves to amplify this response.
Recent years have seen the identification of many genes that regulate HH signal transduction specifically in vertebrates. The first group of molecules that have been identified are novel receptors for HH morphogens, HIP, BOC, CDO and GAS1. These molecules were identified through combined efforts that screened cDNA libraries for genes whose protein products facilitate binding of SHH morphogen to the cell surface [35], and that used microarrays to identify genes that are up- or down-regulated in mice with mutations in HH pathway components, consistent with their being targets of HH signalling [34]. These receptors bind HH morphogens, inhibiting their diffusion and possibly internalising and degrading them, thus regulating the steepness of the morphogen gradient. In the case of BOC, CDO and GAS1, they also promote the response of target cells to HH morphogen [34], [36], [37], [46].
The second group are novel molecules that regulate signal transduction from SMO to the GLI-transcription factors. This includes RAB23 [47], a vesicle transport protein, Iguana [48], [49], which regulates nuclear import of GLI-transcription factors, and Talpid3 [50], which regulates GLI processing and turnover. These proteins have been studied in the mouse, zebrafish and chick, respectively, and so further work will be required to determine whether they function together in a single organism. Further insight has come from the identification of β-arrestins as regulators of the subcellular distribution of SMO [51], [52], [53]. The identification of these molecules provides new molecular reagents to interrogate HH signalling from SMO to GLI.
The final group of novel HH signalling regulators are components that associate with centrosomes and the ciliary axoneme, collectively identifying a role for primary cilia in HH signal transduction. The driving force for this discovery has been the identification of genes causing a range of phenotypically related human syndromes, the ‘ciliopathies’ (reviewed in [54]). The recognition that these diseases share a common set of malformations and the observation that the protein products of the causative genes localise to primary cilia, have highlighted the biological importance of this organelle. Some of the malformations that form part of the ciliopathies are cardinal features of defective HH signalling in the human, including polydactyly and agenesis of the corpus callosum [54]. Analysis of mouse mutants has shown that primary cilia play roles in regulating signal transduction, including that of the HH pathway, and several HH pathway components also localise to cilia in a manner that correlates with HH pathway activity (reviewed in [55], [56]). Ciliary motor proteins are separated into complexes B or A, involved in anterograde and retrograde transport, respectively ([57], for review). Mice with mutations disrupting these motor proteins can display aberrant GLI3 processing, altering the balance between activator and repressor forms (GLI3-A and GLI3-R, respectively), and disruption of different ciliary components can affect GLI3 processing differently. It is therefore likely that analysis of different classes of cilia-related proteins will facilitate dissection of the mechanisms that regulate GLI processing.
The work described in 1.2 The hedgehog paradigm: insight from fruit fly genetics, 1.3 Recent insights into hedgehog signal transduction presents a view of the canonical HH signalling pathway as a series of repressive interactions (Fig. 1A). Ultimately, this culminates in GLI-mediated transcriptional regulation both of molecules external to the HH pathway, to control a variety of cellular processes, as well as genes encoding HH pathway components themselves. Considering the extent to which this pathway has been studied in a variety of contexts, and the large gaps in our knowledge, it is important to define non-canonical HH signalling to allow the evidence supporting the existence of such pathways to be evaluated.
Defining non-canonical pathways stems, retrospectively, from examples of HH signalling that deviate from the canonical paradigm. Three general scenarios can be envisaged. (1) Signalling that involves HH pathway components but which is independent of GLI-mediated transcription. An example where transcription-independent signalling has been formally demonstrated in a fibroblast cell line is outlined in Section 2.3 [58]. (2) Direct interaction of HH signalling components with other molecular pathways, as opposed to the usual indirect regulation of cellular processes by GLI-mediated transcription. As outlined in Section 2, this would include direct interaction of PTCH1 with components of the cell cycle ([59]; Section 2.1; Fig. 1B) or regulators of apoptosis ([60]); Section 2.2; Fig. 1C), as well as biochemical functions of the GLI proteins other than their roles as transcription factors ([61]; Section 2.3). (3) ‘Non-contiguous’ or ‘atypical’ interaction of core HH pathway components with one another. This would include situations where signalling was shown to ‘skip-out’ one or more pathway components upstream of the GLI transcription factors, or where they did not adhere to their usual serially repressive interactions. As described in Section 2, this would constitute SMO- or GLI-independent modes of response regulated by HH and PTCH1 [62], [63], [64], or, for example, if PTCH1 and SMO were to interact positively, as has been controversially suggested in Drosophila head capsule development [65], [66]. However, this would not necessarily include situations where not all pathway components are active in a cell; for example, SMO can induce GLI-mediated transcription when expressed in the absence of PTCH1 (for example, [67]).
Section snippets
Evidence for non-canonical hedgehog signalling pathways
Early work into HH signalling (Section 1.2) was based largely on the analysis of gene expression in response to HH signalling in body segments and wing imaginal discs in transgenic flies. It is perhaps not surprising, therefore, that the large body of work that has followed has been strongly biased toward an analysis of the transcriptional outputs of signal transduction. In recent years, biochemical analyses of the PTCH1 protein and studies of cell migration/axon guidance are collectively
Evolutionary perspectives: insights from nematode worms
The basic framework of the canonical HH pathway is generally conserved from fruit flies to humans, although some important differences have been noted (Section 1.3). Comparative genomic and functional analyses in nematode worms have identified even earlier evolutionary differences in the HH signalling pathway, providing clues as to the ancestral function of PTCH1 homologues, and conceptually supporting the notion that HH pathway components can function independently of one another.
Homologues of
Conclusions
The evidence presented in this review offers compelling support for the existence of non-canonical HH pathways. It is possible that such pathways are evolutionary throw-backs to a time when HH pathway orthologues functioned independently of one another, as evidenced by analysis of nematode genomes. On the other hand, functional domains of PTCH1 that are relevant to non-canonical signalling in mammals are not extensively conserved, suggesting that certain non-canonical pathways may have been
Acknowledgements
This work was funded by the Medical Research Council. The work contributing to Fig. 2 was done in collaboration with Prof. Adrian Woolf. I am grateful to Drs. Chris Babbs and Steve Twigg for helpful comments on the manuscript.
References (162)
- et al.
Trends Cardiovasc. Med.
(2007) - et al.
Chem. Biol.
(2004) - et al.
Cell
(1992) - et al.
Curr. Biol.
(1995) - et al.
Dev. Biol.
(2001) - et al.
Mol. Cell
(2001) - et al.
Curr. Biol.
(2000) - et al.
Dev. Cell
(2005) - et al.
Cell
(2000) - et al.
Cell
(1996)
Dev. Cell
Mech. Dev.
Dev. Cell
Dev. Cell.
Cell
Cell. Signal.
Curr. Biol.
Dev. Biol.
Cell
Curr. Biol.
Trends Cell. Biol.
Cell
Cell
Trends Neurosci.
Cell
Cell
Arch. Entw. Organ.
Nature
Genetics
Teratology
Nobel Lecture
Clin. Genet.
Nat. Med.
Genes Dev.
Nature
Proc. Natl. Acad. Sci. USA,
Gene
Genes Dev.
Nature
Nature
Nature
Science
Nature
Development,
Nature
Nature
Genes Dev.
Nature
Genes Dev.
Nature
Cited by (190)
Primary cilia and Sonic hedgehog signaling in adrenal gland physiology and cancer
2024, Current Opinion in Endocrine and Metabolic ResearchThe signaling pathways and targets of natural products from traditional Chinese medicine treating gastric cancer provide new candidate therapeutic strategies
2023, Biochimica et Biophysica Acta - Reviews on CancerSonic hedgehog signaling in craniofacial development
2023, DifferentiationISL1 regulates lung branching morphogenesis via Shh signaling pathway
2023, Journal of Biological ChemistryDesign, synthesis and activity evaluation of Hedgehog inhibitor Itraconazole derivatives in A549 cells
2022, Bioorganic and Medicinal Chemistry LettersCitation Excerpt :There are many essential proteins that directly phosphorylate Smoothened, activating the protein and allowing Hedgehog pathway initiation, such as casein-kinase1 and G-protein-coupled receptor kinase2.11 Studies have suggested that non-classical Hedgehog signaling pathways include Gli-independent signaling pathways directly activated by Ptch and Smoothened or Hedgehog ligand independent signaling pathways directly activated by Ptch and Smoothened proteins by the Gli family.12–14 Itraconazole is a famous triazole antifungal drug.