|Home | About | Journals | Submit | Contact Us | Français|
Members of the Hedgehog (Hh) family of secreted proteins function as morphogens to pattern developing tissues and control cell proliferation. The seven transmembrane domain (7TM) protein Smoothened (Smo) is essential for the activation of all levels of Hh signalling. However, the mechanisms by which Smo differentially activates low or high level Hh signalling are not known . Here we show that a newly identified mutation in the extracellular domain (ECD) of zebrafish Smo attenuates Smo signalling. The Smo agonist purmorphamine  induces the stabilisation, ciliary translocation, and high level signalling of wild-type (wt) Smo. In contrast, purmorphamine induces the stabilisation, but not the ciliary translocation or high level signalling of the Smo ECD mutant protein. Surprisingly, a truncated form of Smo which lacks the cysteine-rich domain of the ECD localises to the cilium but is unable to activate high level Hh signalling. We also present evidence that cilia may be required for Hh signalling in early zebrafish embryos. These data indicate that the ECD, previously thought to be dispensable for vertebrate Smo function, both regulates its ciliary localisation and is essential for high level Hh signalling.
The activation of Smo is correlated with its stabilisation, conformational change and membrane localisation, although to what extent these aspects of Smo activation are interrelated is not known [3–11]. In Drosophila, Smo accumulates at the plasma membrane upon activation of Hh signalling , and targeting of Smo to the plasma membrane causes constitutive activation [9, 10]. In mammals, cilia are required for Hh signalling . Smo localises to the primary cilium in response to Hh signalling, and this localisation is required for Smo function . However, the ciliary localisation of Smo is not sufficient for activation, as Smo can accumulate at the cilium in the presence of its inhibitor cyclopamine [13, 14]. The activation of Smo may thus require distinct steps , although the mechanisms that control this process remain unknown.
In a screen for zebrafish mutations affecting early embryonic patterning, we identified s294. At 24 hours post-fertilisation (hpf), s294 mutants (Fig. 1B) show phenotypes similar to those seen in the hi1640 allele of smo (Fig. 1C), a presumed null . Mapping data showed that s294 maps close to smo and complementation tests that s294 fails to complement hi1640 (data not included).
In zebrafish, Hh signalling is required for specification of multiple cell types within the somites . Superficial slow fibres (SSFs), expressing slow Myosin Heavy Chain (sMHC) and Prox1, depend on medium-to-low level Hh signalling, whereas muscle pioneer cells (MPs) require maximum levels of Hh signalling, and express high levels of Engrailed (Eng) . In smos294 mutants, MPs were absent and the SSF population was reduced (Fig. 1D–I), indicating that high level Hh signalling was abolished and medium-to-low level signalling diminished. Thus, s294 is a hypomorphic allele of smo.
To identify the molecular lesion in smos294, we sequenced its open reading frame, and found a G-to-A transition at position 374, leading to a cysteine-to-tyrosine substitution at residue 125 in the ECD of the protein (Fig. 1J). The C125 residue is one of 10 highly conserved ECD cysteines in the Smo/Frizzled (Fz) family of 7TM proteins, and is likely to participate in a disulfide bridge that regulates tertiary structure . The C125Y substitution in Smos294 is thus likely to alter protein conformation through disruption of a disulfide bond.
To test whether the SmoC125Y substitution alters protein stability, we generated a version of mouse Smo in which the analogous cysteine was replaced by tyrosine (SmoC151Y). When expressed in zebrafish embryos or Smo−/− mouse embryonic fibroblasts (MEFs), SmoC151Y levels were reduced compared to those of wt Smo (Supplementary Fig. 1). Treatment with purmorphamine increased the level of both wt and mutant Smo to that of, or above, untreated wt Smo (Supplementary Fig. 1). However, stabilisation of SmoC151Y was not sufficient to restore function: whereas wt Smo was able to restore Shh and purmorphamine responsiveness to Smo−/− MEFs, SmoC151Y failed to restore full pathway activity (Supplementary Fig. 1). These results suggest that the ECD is an important determinant of Smo stability, and that it is required for full pathway activation.
Previous studies using truncated versions of Smo proposed that the ECD was dispensable for vertebrate Smo activity [18, 19]. However, in Drosophila, three hypomorphic alleles of smo have cysteine substitutions in the ECD similar to that in smo294 (Fig. 1J) [9, 20, 21], and a truncated version of Drosophila Smo that lacks the ECD fails to rescue Drosophila smo mutants . To further assess the signalling capabilities of Smos294, we treated zebrafish smo mutants with purmorphamine from 8 until 24 hpf. As expected, purmorphamine treatment had no effect on smohi1640 mutants at any concentration tested (not shown). In contrast, treatment of smos294 mutants with 20 or 25 μM purmorphamine resulted in a significant degree of rescue of the morphological phenotype (Fig. 2A–D). Treatment with 25 μM purmorphamine induced complete rescue of SSFs in at least one somite in 43% (10/23) of smos294 mutant embryos (21.6±0.5 SSFs, mean±sem, n=13 somites in 3 independent experiments), suggesting that medium-to-low level Hh signalling was restored (Fig. 2E–H, Supplementary Table 1). This treatment also induced a small, but significant increase in SSFs in wt embryos (Supplementary Table 1, Student’s t-test p<0.02). These results indicate that the Smos294 protein retains the ability to signal. Although treatment with 20 μM purmorphamine significantly increased the number of MPs in wt embryos (Student’s t-test p<0.0001, Fig. 2I, J, Supplementary Table 1), it did not restore MP formation in smos294 mutants (Fig. 2K, L). Higher concentrations of purmorphamine also had no effect (Supplementary Table 1). Thus, purmorphamine induced high level Hh responses in wt embryos, but only medium-to-low level signalling in smos294 mutants.
To further test the requirement of the ECD in Smo signalling, we generated Myc-tagged SmoΔCRD, a form of mouse Smo that lacks the cysteine-rich domain (amino acids 68–182) . Injection of 250pg Smo mRNA fully rescued zebrafish smo mutants , and injection of 250pg SmoΔCRD mRNA similarly rescued SSF development (Fig. 2M–P, Supplementary Table 2). In contrast, SmoΔCRD could not rescue MPs in smohi1640 mutants, even at higher doses of mRNA or in the presence of purmorphamine (Fig. 2Q–T, Supplementary Table 2), and similarly failed to restore Gli-luciferase activity in Smo−/− MEFs (Supplementary Fig. 2). SmoΔCRD thus behaved like stabilised Smos294 in being able to activate medium-to-low but not high level signalling. Together, these results indicate that the ECD of vertebrate Smo, like that of Drosophila Smo, is essential for full activity in vivo.
The failure of stabilisation of SmoC151Y to restore high level signalling suggests that the SmoC151Y mutation interferes with other regulatory processes that control Smo activity. To test whether the C151Y mutation alters trafficking of Smo, we examined the localisation of wt Smo and SmoC151Y in NIH-3T3 cells. As shown previously, Smo localises to the cilium in response to Hh signalling (Fig. 3A, Supplementary Fig. 3) . In contrast, Hh could not induce any detectable ciliary localisation of SmoC151Y (Fig. 3B), as is the case with CLDSmo  (Supplementary Fig. 3). Similar to Hh exposure, treatment of NIH-3T3 cells with purmorphamine caused the ciliary localisation of wt Smo (Fig. 3C), but did not induce detectable ciliary localisation of SmoC151Y or CLDSmo (Fig. 3D, Supplementary Fig. 3).
To investigate whether Smo localises to cilia in zebrafish, we injected mRNA encoding Myc-tagged Smo, SmoC151Y and SmoΔCRD into zebrafish embryos at the 1-cell stage, and looked at subcellular localisation in cross-sections using antibodies against the Myc-tag and acetylated Tubulin. To provide spatial information, we used Tg(-1.8gsc:GFP)ml1 embryos  which express GFP in the cells of the dorsal midline, a region of sonic hedgehog (shh) expression. At the end of gastrulation (10 hpf), Smo was found localised to primary cilia in 29±4 % (mean ± sem, n=17 sections from a minimum of 4 embryos) of ciliated cells surrounding the GFP-positive dorsal midline (Fig. 3E, J). In contrast, SmoC151Y was not detected on cilia (0±0 %, mean ± sem, n=14 sections from a minimum of 6 embryos, Fig. 3F, J). Treatment with 20 μM purmorphamine significantly increased the percentage of cells that showed ciliary localisation of wt Smo to 60 ± 5% (mean±sem, n=17 sections from a minimum of 5 embryos, Student’s t-test p<0.0001, Fig. 3G, J), but did not induce detectable ciliary localisation of SmoC151Y (0±0 %, n= 14 sections from a minimum of 4 embryos). SmoΔCRD was clearly visible on 17 ± 5 % of the cilia (mean±sem, n=34 sections from a minimum of 5 embryos, Fig. 3I, J). The ciliary localisation of SmoΔCRD was abolished by introducing the CLD substitutions W549A and R550A : neither CLDSmo nor CLDSmoΔCRD were observed on the cilium (Supplementary Fig. 4).
Comparison of different dorso-ventral positions from the midline (Fig. 3J insert and Supplementary Fig. 5) showed a graded distribution of ciliary Smo along the dorso-ventral axis. In cells flanking the dorsal midline, the ciliary localisation of Smo was significantly lower than that seen in the midline (14 ± 4 %, mean ± sem, n=17 sections from a minimum of 4 embryos, Student’s t-test p<0.02), and further ventrally this percentage dropped to 6 ± 3 % (mean ± sem, n=17 sections from a minimum of 4 embryos, Student’s t-test p<0.0002). Purmorphamine induced uniformly high levels of ciliary Smo, thereby abolishing the graded ciliary localisation (Fig. 3J). Surprisingly, SmoΔCRD localised to cilia at medium-level percentages, and showed no significant difference between dorsal and more lateral positions (Fig. 3J), suggesting that the ciliary localisation of SmoΔCRD is independent of Hh levels (Supplementary Fig. 6). Altogether, these results suggest that Smo localises to the cilium in cells exposed to high levels of Hh, and that the ECD of Smo is involved in regulating its localisation to the cilium.
These results also suggest that cilia may be required for Hh signalling in zebrafish. A screen for mutations causing kidney cysts identified zebrafish homologs of intraflagellar transport genes, as well as other genes required for ciliary formation and function, including the novel gene qilin . Unlike mouse, none of the mutations in these genes causes overt Hh phenotypes in zebrafish. However, it is unclear whether any of these mutants display a complete and/or early loss of cilia [23, 24], a condition that may be required to disrupt Hh target gene expression. In support of a role for cilia in Hh signalling in zebrafish, knockdown of ift80 has been reported to cause a down-regulation of embryonic ptc1 expression .
To begin to address the role of cilia in zebrafish Smo function, we used two non-overlapping morpholinos (MO) targeting the translation start site of qilin (qilin MO1 and MO2) and analysed the effects of qilin knock-down in early zebrafish embryos. We further tested that the results were due to a loss of cilia rather than a separate role of Qilin by using a dominant-negative Kif3b-GFP fusion protein (dnKif3b) .
Embryos injected with dnKif3b mRNA, qilin MO1 or qilin MO2 showed a similar range of morphological phenotypes (Supplementary Fig. 7), suggestive of defects in convergence-extension movements during gastrulation (Supplementary Fig. 8) and an upregulation of canonical Wnt signalling (Supplementary Fig. 8), consistent with previous reports [27, 28]. Furthermore, all injections caused laterality defects (Supplementary Fig. 9). All injections also caused a significant reduction, and in some cases a complete loss, of KV cilia at 16 hpf (Fig. 4A–D, Supplementary Fig. 10). The number of KV cilia was restored in qilin MO2 injected embryos by co-injection of qilin mRNA (Fig. 4E, Supplementary Fig. 10). However, none of the qilin MO or dnKif3b mRNA injected embryos showed a complete loss of cilia in the lumen of the neural tube or in the somites at 24 hpf, raising the possibility that embryos displaying complete loss of cilia were lost to earlier lethality.
To analyse the level of Hh signalling in these injected embryos, we examined the expression of the Hh target genes ptc1 and myod at the end of gastrulation (10 hpf). Injection of dnKif3b mRNA, qilin MO1 or MO2 caused a reduction or loss of ptc1 expression (Fig. 4F–I, Supplementary Fig. 10), and myod expression was reduced or absent in qilin MO2 injected embryos (Fig. 4K, L, Supplementary Fig. 10). Moreover, both ptc1 and myod expression was restored by co-injection of qilin mRNA with qilin MO2 (Fig. 4J, M, Supplementary Fig. 10), suggesting that loss of cilia causes loss of Hh target gene expression in zebrafish. shh expression was not reduced in these embryos (Fig. 4N, O), indicating that the defect lies in Hh signal transduction. Some loss of Hh target gene expression was also detected in the injected embryos at later stages in the neural tube (Supplementary Fig. 11). Together, these results suggest that cilia are required for high level Hh signalling in zebrafish, although the ultimate proof must await a genetic model.
Using a combination of genetic and pharmacological manipulations of Smo, we have shown here that stabilisation, ciliary localisation and full activation of Smo are distinct and separable aspects of its activity. Our results indicate that the ECD of vertebrate Smo plays important roles in these distinct Smo activities: it is essential for high but not low level Hh signalling, and it regulates both the stability and ciliary localisation of Smo. The inability of SmoC151Y to move to the cilium suggests a role for the ECD in promoting ciliary localisation of Smo. However, the finding that deletion of the CRD leads to moderate levels of ciliary localisation in a Hh-independent manner indicates that the ECD may also function to prevent ciliary localisation. The ECD of Smo may thus both promote and restrict ciliary localisation.
Furthermore, our data on SmoΔCRD indicate that ciliary localisation is not in itself sufficient for full activation of the Hh signalling pathway, suggesting that other ECD-dependent events are required for Smo activation of high level signalling. Conversely, the ability of purmorphamine to increase medium-to-low level Hh signalling in smos294 mutants without causing ciliary translocation of Smo suggests that medium-to-low level signalling may not depend on Smo localisation to the cilium in zebrafish.
A recent study provided evidence that both Drosophila and vertebrate Smo form constitutive dimers/oligomers through the ECD . One possible model is that dimerisation of vertebrate Smo through the ECD poises the ECD to mediate conformational changes that regulate both ciliary translocation and full activation of Smo. The smos294 mutation may prevent the ECD-dependent conformational change that is required for Smo trafficking to the cilium and high level signalling. SmoΔCRD may similarly fail to adopt the conformational changes required for full level activation, but unlike SmoC151Y, it escapes the regulatory mechanism that restricts ciliary localisation. Exploring whether the ECD controls high level signalling and ciliary localisation by regulating Smo conformation will provide for fruitful further investigation.
We would like to thank Jussi Taipale for Smo−/− MEFs, Setsu Endoh-Yamagami and Andrew Peterson for the dnKif3b construct, and Tom Kornberg, Michel Bagnat, Won-Suk Chung and Dirk Meyer for comments on the manuscript. Support for this research was provided in part by grants from the Burroughs Wellcome and Sandler Family Foundations (J.F.R.), the Packard Foundation and the NIH (J.F.R., D.Y.R.S.), and the University of Innsbruck (W.S., P.A.).
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.