Ptch2/Gas1 and Ptch1/Boc differentially regulate Hedgehog signalling in murine primordial germ cell migration

Gas1 and Boc/Cdon act as co-receptors in the vertebrate Hedgehog signalling pathway, but the nature of their interaction with the primary Ptch1/2 receptors remains unclear. Here we demonstrate, using primordial germ cell migration in mouse as a developmental model, that specific hetero-complexes of Ptch2/Gas1 and Ptch1/Boc mediate the process of Smo de-repression with different kinetics, through distinct modes of Hedgehog ligand reception. Moreover, Ptch2-mediated Hedgehog signalling induces the phosphorylation of Creb and Src proteins in parallel to Gli induction, identifying a previously unknown Ptch2-specific signal pathway. We propose that although Ptch1 and Ptch2 functionally overlap in the sequestration of Smo, the spatiotemporal expression of Boc and Gas1 may determine the outcome of Hedgehog signalling through compartmentalisation and modulation of Smo-downstream signalling. Our study identifies the existence of a divergent Hedgehog signal pathway mediated by Ptch2 and provides a mechanism for differential interpretation of Hedgehog signalling in the germ cell niche. How co-receptors Gas1 and Boc interact with Ptch1/2 receptors and regulate Hh signalling is unclear. Here, the authors demonstrate that the spatiotemporal expression of Gas1 and Boc determines how Hh signalling affects the dynamic migration of murine primordial germ cells.

P rimordial germ cells (PGCs), the progenitors of gametes, provide a good model to study directional migration during development. The pluripotent PGCs rely upon precise communication by both secreted morphogens acting at long distance and local signalling produced by the immediate microenvironment [1][2][3][4] . In mouse, PGCs arise from the posterior extraembryonic mesoderm around embryonic stage (E)7.5 and migrate into the definitive endoderm, continuing through the developing hindgut (HG) until E9.5. Around E10.5, PGCs directionally migrate from the dorsal body wall, finally arriving at the bilateral genital ridges (GR) between E10.5 and 11.5, where they become immotile and differentiate into either spermatozoa or oocytes, integrating with the developing somatic gonads 2,3,5 . Studies in Drosophila and zebrafish suggest that Hedgehog (Hh) signalling is involved in the development of PGCs, but does not function as a fate determinant or guidance molecule [6][7][8] . The role of Hh signalling in mouse PGCs still remains ambiguous, although the aorta, gonad, mesonephros and GR are suggested to be the main source of chemo-attractantion 3,9,10 .
The role of Hh in chemotaxis has been demonstrated in different developmental contexts [11][12][13][14] . Binding of Hh to the two paralogue Patched (Ptch1/2) receptors releases the Smoothened (Smo) G-protein coupled receptor, which allows its translocation to the primary cilia and the de-repression of Smo-dependent signalling. Ptch2 shares structural similarities with Ptch1, including extracellular ligand-binding loops and transmembrane domains, but has much shorter intracellular amino-and carboxyterminals [15][16][17] . Like Ptch1, Ptch2 interacts with all mammalian Hh ligands (Sonic hedgehog, Shh; Desert hedgehog, Dhh; Indian hedgehog, Ihh) with a similar affinity and forms a complex with Smo. Contradicting reports claim that Ptch2 possesses either similar, weaker or no repressive activity on Smo during Hh signalling, compared with Ptch1 16,[18][19][20] . The embryonic expression pattern of Ptch2 is distinct from Ptch1, mainly in the skin and testis, and overlapping with Shh 18 . Ptch2-hypomorphic mice are viable and fertile, whilst Ptch1 mutants are embryonic lethal, suggesting that Ptch1 is the major regulator of Hh signalling and Ptch2 has a redundant function compensatable by Ptch1 [21][22][23] . Cells lacking Ptch1 remain sensitive to Hh in a chemotaxis assay, and Ptch2 can mediate Hh-induced motile responses in the absence of Ptch1 24 .
The Gli family of transcription factors mediate the canonical Hh pathway to regulate cell fate and patterning in accordance with the ligand gradient. Gli-independent non-canonical signalling also occurs, which does not require Smo localisation to the primary cilia. It is thought that non-canonical Hh signalling regulates cytoskeletal rearrangement in differentiated cells after specification 25,26 , endothelial cell migration in pro-angiogenic responses 27 and axon guidance through the activation of Src kinases 28 . A timely switch from canonical to non-canonical signalling can also occur during normal development 29 . Notably, Smo proteins located outside the cilia are believed to be responsible for non-canonical chemotactic responses, suggesting that Smo localisation might be the critical determinant in the selective engagement of canonical versus non-canonical pathways 25,30 . Since the majority of Smo proteins reside outside of the primary cilia, it has also been speculated that noncanonical signalling may represent a more general and robust Hh response in the normal state 26 .
A number of obligatory co-receptors for Hh have been identified in vertebrates, which include Boc (bioregional Cdonbinding protein), Cdon (cell-adhesion-molecule-related/downregulated by oncogenes, also called as Cdo) and Gas1 (Growth arrest-specific gene 1). Cdon and Boc belong to a family of cell adhesion proteins 31 , while Gas1 is structurally distinct and localises at plasma membrane rafts via a glycosylphosphatidylinositol (GPI) anchor 32 . These co-receptors can bind Hh ligand independently of Ptch1 and facilitate ligand-receptor interaction at the cell surface 33,34 . Boc and Cdon have partially redundant and distinct tissue-specific roles in Hh regulation during myogenic differentiation and axon guidance [35][36][37][38] . Gas1 positively regulates Shh signalling in the neural tube and forebrain 39,40 but may exert a negative effect in the somite and mandibular arch at high concentrations [41][42][43] . Certainly, the developmental defects observed in Gas1 mutant mice suggest a complex relationship between Gas1 and Shh 44,45 . Since the positive effects of Gas1 are evident in areas of low Hh concentration, it has been speculated that Gas1 may extend the range of ligand gradient, translating a low peripheral chemical concentration into cellular activity 46 . In contrast, Boc/Cdon are implicated in areas of high Hh concentration, potentially required for the maximum canonical response 47 .
A key question is how Ptch1 and Ptch2 recognise and respond differentially to Hh ligand in the production of different cellular outcomes, particularly when considering that both canonical and non-canonical processes require Smo. Moreover, the role of coreceptor function in Ptch2-dependent signalling is completely unknown. Here we reveal that PGCs are naturally unciliated but remain sensitive to locally secreted Hh ligand through an exclusive interaction between Ptch2 and Gas1, while ciliated somatic cells immediately surrounding PGCs rely on interaction between Ptch1 and Boc. The molecular mechanisms leading to the derepression of Smo after Hh ligand reception are different in Ptch1/Boc and Ptch2/Gas1-mediated signalling events, resulting in distinct downstream signal pathways acting in parallel to Glidependent transcription. Our findings provide a previously unreported mechanism for Ptch2-specific signal pathways involving the cAMP-responsive element binding protein (Creb) and Src tyrosine kinase. We propose that these distinct molecular mechanisms, mediated by different Ptch receptors and their coreceptors, may be a key determinant in the selective activation of distinct downstream signalling pathways. This may provide the required specificity and fine-tuning of Hh signalling, leading to either cell fate specification or post-specification cellular behaviour, such as motility.

Results
Expression of Shh, Gas1 and Boc in PGC migration niche. To evaluate the involvement of Hh signalling during the migration of post-specification PGCs in mouse, we examined the spatiotemporal expression of Hh pathway genes in the gonadal anlage of E9.5-11.5 embryos. We found that Ptch1, Gli1/2/3, Dhh and Shh, but not Ihh, were expressed along with a germ cell marker Stella in the HG at E9.5 and in the GR at E10.5-11.5, where PGCs localise (Fig. 1a). The mRNAs of Boc and Gas1 were detected in the urogenital ridges but Cdon was not, indicating specific involvement (Fig. 1b). Further examination of the expression patterns of Shh, Gas1, Boc and Cdon by in situ hybridisation and immunostaining of transverse sections showed that at E9.5 when post-specification PGCs are motile but still restricted within the HG, Shh and Gas1 were expressed in overlapping patterns along the HG spreading dorsally towards the urogenital ridges, while Boc showed a broad signal which was much less intense compared with Shh or Gas1 (Fig. 1c, top panel). As development continued into E10.5 when PGCs rapidly exit from the HG migrating towards the GR, the Gas1 signal seemed to follow the PGC migration route, moving dorsally from HG towards the mesentery and GR, while Boc expression was broadly present in these regions increasingly visible throughout the developing GR (Fig. 1c, bottom panel). Detailed examination after co-staining with the PGC marker SSEA1 revealed that Gas1 signal exclusively overlaps with SSEA1 but Boc signal is ubiquitous (Fig. 1d), thus migrating PGCs specifically express Gas1, while Boc is broadly expressed in the PGCs and the surrounding mesenchyme. In contrast, Cdon is not detected in PGCs or the adjacent migration niche during this period (Fig. 1c, d). Interestingly, Shh expression remained restricted within the HG during this period (Fig. 1c, e). The fact that PGCs migrate away from the source of high Shh expression confirms that Shh is not a chemoattractant for PGCs.
Shh enhances PGC motility without affecting directionality. To further define the role of Shh in PGC migration, we conducted a chemotaxis assay by co-culturing PGCs juxtaposed to 293 cells transfected with either human full-length SHH construct or an empty vector (Fig. 2a, b). After 24 h incubation, we found that similar numbers of PGCs migrated towards the 293 cells regardless of Shh secretion (Fig. 2c, right panel), confirming that Shh is not a chemoattractant for PGCs, as suggested from the embryo expression data (see Fig. 1). However, when we compared the migration distance, PGCs located among the Shh-expressing 293 cells migrated significantly further from the middle line (Fig. 2c, left panel), implying that Shh may enhance the motility or speed of migration. To test this notion, we generated timelapse movies of embryo slice cultures derived from the Stella GFP transgenic mice, which allowed us to monitor the migratory behaviours of GFP-expressing PGCs in live embryos (Supplementary Fig. 1 and Supplementary Movies 1-4). To assess the effects of Hh signalling, we treated the slice culture medium with the Hh agonist (purmorphamine) or antagonist (cyclopamine), respectively, during the 10 h of live imaging (Fig. 3a). Analyses of the velocity, accumulated distance (total cell path travelled), Euclidean distance (the shortest distance between the start and endpoints) and directionality (the ratio between Euclidean distance and accumulated distance) revealed that purmorphamine enhanced the speed and distance of migration, while cyclopamine significantly reduced these values, compared with the solvent control. Consistently, tomatidine, a drug with a similar structure to cyclopamine but lacking any inhibitory activity, had no effects on PGC migration within these parameters (Fig. 3b). Notably, none of the treatments altered the directionality of migration (Fig. 3a, b), signifying that Hh enhances the motility of PGCs without influencing directionality. PGC survival time was similar in all treatment groups (Fig. 3c), indicating no general cytotoxicity from these treatments. Since the embryo size expansion over the period of live imaging was similar in all treatment groups (Fig. 3d, Supplementary Fig. 3), it is unlikely that the altered PGC migration was simply a consequence of morphological changes in the surrounding tissues of the growing embryo. To directly demonstrate the effects of Hh signalling on the intrinsic motile capacity of PGCs, we generated primary cultures of dissected GR,     Fig. 2 Hh enhances PGC migration without chemoattraction. a Schematic representation of the chemotaxis assay. NIH3T3 cells deleted for the endogenous Wdr11 by CRISPR/Cas9, and therefore unable to secrete Shh, were used as the feeder. On average,~30,000 cells of primary GR culture derived from five embryos at E10.5 were plated on top of the feeder. For the source of Shh secretion, 293 cells transfected with a human full-length SHH expression construct were plated in comparison to the empty vector. After 24 h of incubation, cells on both sides of the coverslips were analysed by immunofluorescence. Scale bar, 50 μm. b Western blot analyses validated Wdr11 KO feeder lacking Shh secretion in the conditioned medium (CM), while the transfected 293 cells producing Shh. B-actin is a loading control. Representative blots and cell images from three independent experiments are shown. c Total numbers of PGCs (SSEA + ) and their distance from the middle (the dotted line in a) were assessed from ten random fields. Percentages of PGCs migrated towards the transfected 293 cells were similar in both groups, but the average distance of migration was significantly higher in SHH-expressing cells (n = 35) compared with empty vector (n = 30). Data are presented as mean ± SD from three independent experiments (****P < 0.0001; NS, P > 0.05). We did not find evidence of general toxicity from the drug treatments at the concentrations we used ( Supplementary Fig. 2), although lower concentrations of vismodegib and cyclopamine had been used in some explant cultures 48 . Combined, these data suggest that Hh signalling directly regulates the intrinsic motile capacity of PGCs.
Naturally unciliated PGCs remain responsive to Hh signal. Our live imaging data demonstrated that PGCs migrate as single cells dispersed throughout the migratory niche mesenchyme, rather than a collective migration as a group or long line (see Supplementary Movies 1-4). We, therefore, speculate that the motility of PGCs primarily relies on the communication of individual PGCs with their immediate surroundings. To investigate how the Hh signal pathway operates in PGCs and soma, we examined their signalling responses in primary cultures of GR. After treatment with purmorphamine for 18 h, the somatic cells (negative for SSEA1 or Stella), showed a distinctive accumulation of Smo and Gli3 to the primary cilia, as expected from ciliated cells upon activation of the canonical Hh signalling pathway. However, PGCs showed significantly increased levels of Smo and Gli3 proteins within the cytoplasm (Fig. 4a, b). Since PGCs exhibited an unexpected pattern, we explored if there was any alteration of primary cilia in PGCs. Surprisingly, the cilia axoneme identifiable by Arl13b or acetylated tubulin staining was mostly absent in the PGCs, although the basal body was visible after gamma-tubulin or CEP164 staining (Fig. 4c, d). These results indicated that PGCs naturally lack primary cilia, but the Hh signal pathway can still be induced in these cells through direct activation of Smo and that Smo proteins localised outside the cilia can mediate Hh signalling, as previously observed from cilia-knockout cell lines 26 .
Migrating PGCs specifically express Ptch2 and Gas1. We next asked whether Hh signalling is received differently in PGCs due to the lack of primary cilia. Immunofluorescence studies of E10.5 GR primary cultures revealed specific expression of Ptch2 on the PGCs, while Ptch1 was detected only on the somatic cells (Fig. 5a). Quantitative RT-PCR analyses on the sorted populations of PGCs (Stella +/GFP+ ) and the somatic cells (Stella -/GFP-) isolated from E10.5 Stella GFP embryos after FACS ( Supplementary Fig. 4) further confirmed the exclusive expression of Ptch2 and Ptch1 mRNA on PGCs and soma, respectively (Fig. 5b). We also confirmed in the primary GR cultures that Gas1 was exclusively present on PGCs, while Boc was ubiquitously detected in both cell types but Cdon was absent (Fig. 5a), consistent with our findings from embryo tissue sections (see Fig. 1d). Are these specific combinations of receptors functionally important for Hhdependent signalling and PGC migration? To test this notion, we examined the phosphorylation status of Src, an indicator of cytoskeletal rearrangement during Hh-induced motility 28 . In E10.5 wildtype (WT) embryos, the highly motile PGCs were stained positive for p-Src. However, Gas1 knockout (KO) mouse embryos showed a significant reduction in p-Src, even though Ptch2 and Boc expression remained similar (Fig. 5c, (Fig. 6a). A reciprocal Co-IP using antibody against endogenous Gas1 confirmed that Gas1 bound to Ptch2 only when Shh ligand was present. Gas1 did not interact with Smo either before or after Shh addition. Boc remained unbound to Ptch2 (Fig. 6a). The specificity of these interactions was further validated by a reciprocal Co-IP in empty-vector transfected cells, which did not show any non-specific binding (Fig. 6a, bottom panel). Based on these data, we hypothesise that Gas1 mediates Shh binding to Ptch2, and the formation of a Ptch2/Gas1/Shh tertiary complex has an immediate effect on the release of Smo from Ptch2 (Fig. 6b). If this is the case, Shh would not be able to bind to Ptch2 or cause Smo release in the absence of Gas1. When Fig. 3 Hh signal regulates the motility of PGC but not the directionality. a Representative images of E10.5 Stella GFP embryo slice cultures. Frames at t = 0, 5 and 10 h from the live imaging (Supplementary Movies 1-4) after treatment with the solvent (DMF, n = 7), 40 μM purmorphamine (Pur, n = 7), 40 μM cyclopamine (Cyc, n = 9) or 40 μM tomatidine (Tom, n = 6) in biologically independent experiments. The GFP-positive PGCs migrating towards the genital ridges (red lines in the embryo image) were tracked using ImageJ plugin. Trajectories of PGCs were generated by placing the starting point of the individual PGC track onto the same point. The direction of migration in relevance to the embryo orientation is shown. n neural tube, a aorta, hg hindgut, dor dorsal, lat lateral, vent ventral. Scale bar, 100 μm. b The effects of Hh agonist/antagonists on the migratory behaviour of PGCs in different parameters are assessed from 7 to 10 PGCs in one embryo slice. Data from DMF (n = 7), Pur (n = 7), Cyc (n = 9) and Tom (n = 6) treatment groups are presented as mean ± SEM. One-way ANOVA followed by Dunnett's test (*P < 0.05; **P < 0.01; ****P < 0.0001). c Cell survival was assessed by the number of hours that the fluorescence signal from GFP was detected during the live imaging of slice cultures shown in b. One-way ANOVA indicated that cell survival was not different among treatment groups (P = 0.9651). d Growth rate of each embryo, presented as the fold expansion of the embryo size during the live imaging, was assessed by measuring the total trunk area of each embryo slice at the beginning and end of the live imaging (t = 0 h and t = 10 h) using ImageJ. Error bars represent SD from three biologically independent experiments. One-way ANOVA indicated that growth rate was not affected by different treatments (P = 0.9186). Source data are provided as a Source Data file.
we repeated the Ptch2 Co-IP experiment in Gas1 KO NIH3T3/ Cas9 cells ( Supplementary Fig. 6, upper panel), Ptch2 was indeed unable to receive Shh ligand and Smo remained within the Ptch2 immuno-complex, even after the addition of Shh (Fig. 6c), supporting our model (Fig. 6d).
Boc mediates slow and gradual release of Smo from Ptch1. We next investigated Boc/Ptch1 interactions using similar approaches after Ptch1 overexpression. Multiple reciprocal Co-IP experiments showed Boc already bound to Ptch1/Smo in the unstimulated state (Fig. 7a).   steps after Hh ligand reception by Boc, which transfers the ligand to Ptch1 (Fig. 7b).
Ptch2 and Ptch1 elicit differential downstream signalling. These data suggested that Gas1 is a preferential co-receptor that is likely to facilitate Ptch2-dependent Hh signalling through a rapid Smo de-repression process. In contrast, the release of Smo from the Ptch1/Boc receptor complex occurs gradually over an extended period of time, conforming to the well-established timeframe of canonical Hh signalling. Therefore, we next compared the dynamics of downstream signalling responses mediated by Ptch1 and Ptch2. First, we examined the induction of Gli1 and Gli3 mRNA as a well-established readout of canonical Hh signalling. Transcription of Gli1 and Gli3 in Ptch1-expressing cells gradually increased after Shh addition, reaching to a maximum (7.4 fold and 5.1 fold, respectively) at 24 h. In contrast, in Ptch2-expressing cells, the induction of Gli1/3 remained at minimal levels even after 24 h (1.8 fold and 1.6 fold, respectively) (Fig. 8a), agreeing with the view that Ptch2 is less efficient than Ptch1 in mediating canonical Hh signalling. It is widely accepted that Hh-dependent induction of cAMP levels and protein kinase A (PKA) activity at the cilium base regulate the coupling of Smo with inhibitory GTP-binding proteins (Gαi) 49,50 . However, this compartmentalised local change of cAMP is undetectable in whole-cell extracts 51 . Since cAMP-PKA signalling activates Creb in NIH3T3 cells 52 , we assessed the level of phospho-Creb by Western blot at different time-points after Shh treatment. Ptch1-expressing cells did not show any p-Creb, agreeing with previous findings that Hh alone does not cause global p-Creb induction 51 . Surprisingly, Ptch2expressing cells did show an induction of p-Creb at 40 min, which further increased at 24 h (Fig. 8b). Therefore, the de-repression of Smo within the context of Ptch2 has different effects to Ptch1, eliciting Creb-dependent global responses. Since this phenomenon occurred without any exogenous manipulations for cAMP, adenylyl cyclase or PKA, it provides clear evidence for the existence of a p-Creb-dependent gene regulatory network directly downstream of Hh/Smo, operating parallel to Gli transcription factors.
The tyrosine kinase Src acts as a key regulator of cell motility by phosphorylating multiple protein substrates that control the cytoskeleton and focal adhesion. Since migrating PGCs show positive staining for p-Src (see Fig. 5c), we investigated if Ptch2 is involved in this signalling pathway. The level of p-Src was undetectable in Ptch1-expressing cells, but Ptch2-expressing cells displayed a substantial level of p-Src, which further increased upon Hh stimulation (Fig. 8b). Combined, these data demonstrate that Ptch2-dependent Shh signalling leads to global changes in cellular p-Creb and p-Src, that are distinct from Ptch1mediated signal responses.
Essential requirement of Gas1 in Ptch2-specific signalling. Co-IP data demonstrated that Gas1 does not interact with Ptch1 although Shh remains bound to Gas1 (Fig. 7a), thus Gas1/Shh explicitly interacts with Ptch2 even in the presence of Ptch1 and does not compete with Boc/Shh for Ptch1 binding. If so, a Gas1 deficiency should specifically interfere with Ptch2-dependent signalling without affecting Ptch1-mediated responses. Our analyses in Gas1 KO NIH3T3/Cas9 cells confirmed that Ptch2dependent induction of Gli1/3 was completely abolished by the lack of Gas1, while those of Ptch1-dependent responses were virtually unaffected (Fig. 8a). Likewise, the induction of p-Creb and p-Src in Ptch2-expressing cells was completely lost, even in the presence of endogenous Boc and Cdon (Fig. 8b). Since Gas1 mutant cells express normal primary cilia ( Supplementary Fig. 8), the observed disruptions of Hh-induced responses were not due to defective cilia. Together, these data support that Gas1 is essential for Ptch2-dependent Hh signalling, while Boc/Ptch1 is likely to function as the main receptor for canonical signalling, as previously suggested 47 .
Shh induces Ptch2-specific signal responses in PGCs, but not in the soma. To further validate these findings, we examined signal responses in primary cultures of GR expressing these receptors endogenously (Fig. 5). If our model is correct, a brief treatment with Shh ligand should induce Ptch2/Gas1-specific responses in PGCs, but not in somatic cells. Immunofluorescence results showed the phosphorylation of Creb and Src in PGCs even after 10-min treatment with Shh ligand, but not in the surrounding somatic cells (Fig. 9a, b). These data support the notion that postspecification PGCs are likely to respond rapidly to locally secreted Hh ligand via Gas1/Ptch2, eliciting a Smo-downstream signalling pathway involving p-Creb and p-Src, which does not occur in non-migratory mesenchymal cells expressing Ptch1/Boc.

Gas1-deficient mice show ectopic PGCs and subfertility.
To confirm the essential role of Gas1 for the migration of PGCs in vivo, we examined the establishment of PGCs and fertility phenotypes in Gas1 KO mice. Gas1-null embryos contained significantly fewer numbers of PGCs arriving in the GR at E10.5 with an increased presence of ectopic PGCs in the mesentery and HG areas (Fig. 10a, b), indicating a delayed migration compared with WT. Since the total numbers of SSEA1-positive PGCs remained similar, Gas1 deficiency did not affect the specification and survival of PGCs (Fig. 10b). However, the reduced number of PGCs in the developing GR would be expected to result in reduced numbers of germ cells in the gonads at birth. We investigated the fertility of Gas1 mutants by comparing the average litter sizes after timed-mating, and found Gas1-null females produced significantly fewer litters compared with the heterozygotes, indicating a subfertility (Fig. 10c).  Fig. 4) and the expression of Ptch1, Ptch2, GFP and Stella in GFP-positive (PGCs) and GFP-negative (somatic cells) populations analysed by quantitative RT-PCR. Each data point represents one experiment. Values normalised to 18s rRNA were plotted with error bars representing SD (n = 3) after unpaired t-test (*P < 0.05, **P < 0.005, ***P < 0.001). c Immunofluorescence analyses of Gas1, Ptch2, p-Src, Boc and SSEA1 on the GR tissue sections of WT and Gas1 KO embryos at E10.5. The merged images are shown without DAPI signal for improved clarity. Representative images from three independent experiments are shown. Scale bar, 10 µm. d Dot plots showing the relative fluorescence intensity values of Gas1, Ptch2, p-Src, Boc signal observed in PGCs which were normalised with the fluorescence intensity values of SSEA in each cell. Gas1 KO embryos showed a significant decrease in signal intensity in Gas1 (WT (n = 12), KO (n = 5), ****P < 0.0001) and p-Src (WT (n = 7), KO (n = 12), **P = 0.0054) but no significant changes in Ptch2 (WT (n = 6), KO (n = 6), P = 0.0564) and Boc (WT (n = 6), KO (n = 9), P = 0.2583). Error bars represent SD after unpaired t-test with Welch's correction. Source data are provided as a Source Data file. Discussion Although the predominant model is Shh acting as a morphogen to induce a Ptch1-dependent signal pathway that initiates fate determination and patterning in a gradient-dependent manner, how positional and temporal information is recognised by responding cells remains unknown. Here we have evaluated the role of cell membrane proteins thought to act as obligatory co-receptors for Hh ligand during Ptch-mediated signalling using PGC migration as a model. While our findings are consistent with the role of Ptch1 and Ptch2 in the sequestration of Smo, the reception of Hh ligand and induction of the downstream signalling pathway are mediated distinctly by these two receptors, owing to their specific partnerships with Boc and Gas1. We propose that temporal expression of these receptors determines the signalling kinetics in responding cells exposed to a fixed concentration of ligand. It is assumed that the time duration a certain concentration of Shh is available limits the activity and the nature of signalling 53 . In addition to the changing gradient of Shh in vivo, we now reveal that the dynamic flux of ligand presented to the Ptch receptors themselves may be modulated in a contextdependent manner. The changes in the nature of Hh signal may also depend on the availability of the Smo proteins that can be translocated to the cilium upon signalling. Our data suggest that Gas1/Ptch2 may preferentially interact with Smo proteins compartmentalised in the extra-ciliary domain, which are fast-acting to induce motility within a short time scale, such as minutes rather than days after stimulation, as required for the Glidependent transcription to occur. The binding of Shh with Gas1 and Boc co-receptors is likely to create conformational changes in Ptch2 and Ptch1, differentially Interactions of Ptch2 with endogenous Smo, Gas1 and Boc were analysed by immunoprecipitation (IP) and Western blotting (WB). IP was performed using specific antibodies as indicated and analysed by WB with respective antibodies as shown. The images in the bottom panel are from Ptch1 KO cells transfected with empty vector, which did not show any co-precipitation of endogenous Gas1 and Ptch2 regardless of Shh treatment, confirming the specificity of these antibodies (bottom panel). b A schematic diagram illustrating the protein interactions before and after Shh treatment. c NIH3T3/Cas9 cells with Gas1 KO were transfected with Myc-tagged Ptch2 and stimulated with Shh protein in serum-free medium for 40 min. IP was performed with anti-Myc antibody. Representative blots from two independent experiments are shown. d A schematic diagram illustrating the protein interactions in Gas1 KO cells. Source data are provided as a Source Data file.  15 . However, the cytoplasmic C-terminal tail of Ptch1 required for the repression of Smo, could not be replaced by the analogous Cterminal region of Ptch2 15 . Perhaps the C-terminal tail of Ptch2 may prevent cilia localisation and allow preferential binding with Gas1. Further biochemical, structural and imaging analyses are required to resolve these questions, especially considering the limitation of the current study using overexpressed Ptch and the potential binding of other membrane-associated proteins which we did not investigate. Studies using overexpression constructs provided some initial clues regarding the molecular interactions of these proteins and suggested that Boc may constitutively interact with Ptch1 via its FNIIIa/b domain, but independently interacts with Shh through the FNIIIc domains 33 . The structural domains of Gas1 involved in Ptch2 interaction are unknown, but the core 110 amino-acid tandem repeats were shown to be important in Gas1 interaction with GFRa 54 . Alternatively, Gas1 may simply influence the physical recruitment of partners to the restricted regions of membrane rafts outside the cilium through its GPI-anchor.
We speculate that the rapid release of extra-ciliary Smo proximal to Ptch2/Gas1 may involve different sets of regulatory proteins leading to global cell signalling responses. Although Smo was shown to couple with Gαi, regulating AKT-dependent Gli stabilisation within primary cilia 50  Although the cAMP-PKA pathway has been majorly explored in Hh signalling, other pathways induced by cAMP can converge to Creb in a PKA-independent manner. Notably, cAMP-induced release of Ca 2+ leading to the increased phosphorylation of Creb concomitant with downregulation of Gli1 activity was observed during spinal cord development 29 . It is tempting to speculate that such a balance of the Ptch2-mediated p-Creb-dependent parallel pathway may modulate the Ptch1-mediated Gli-driven programs.
We propose that PGC migration depends on local Hh ligand secreted in the proximity of PGCs, which is received by the coreceptor Gas1 and presented to Ptch2. Our data indicate that Hh is not a direct attractant for germ cells but may play important roles in intrinsic motility. The naturally unciliated PGCs are insensitive to the transcription-driven differentiation signal but exhibit migratory responses to Shh. Counter-intuitively, Ptch2 deficient mice are reported to be fertile 21,23 . Although careful assessment data on the potential sub-fertility of these mice are currently unavailable, we speculate that Ptch2 depletion could have been partly compensated by a ligand-independent activation of Smo or its downstream pathway. For instance, there are other signal pathways mediated by different receptors, cytokines and signal transducers known to be essential for the specification, survival and migration of PGCs 2,3 . It is possible that these conserved fundamental developmental signals converge downstream of Smo without the requirement of Ptch2/Gas1. Moreover, Cdon and Boc were shown to mediate chemotactic migration to Shh even when both Ptch1 and Ptch2 were deleted 55 , suggesting a possibility that Gas1 induces PGC motility in partnership with other cell surface receptors. It is also worth noting that cells lacking both Ptch1 and Ptch2 retain the ability to migrate toward Shh in a Smo-dependent manner in a Boyden chamber assays 24,56 . It remains to be determined what regulates this PGCspecific expression of Gas1 and other receptors and if Gas1 also influences Hh ligand-specificity, considering the specific expression of Dhh and Ptch2 during genitourinary tract development 16,57 . Our study provides mechanistic insight for PGC migration in mouse and perhaps other species.

Methods
Breeding of transgenic mice. C57BL/6 mice were purchased from Charles River (Harlow, UK). Stella GFP BAC mice 1 were gifted by Professor Azim Surani (Gurdon Institute) and maintained in a C57BL/6 background. Gas1 mutant mice were originally generated by Dr Chen-Ming Fan (Carnegie Institute of Washington), maintained in a CD1/C57BL6 mixed background and genotyped as previously described 46  Embryo organ culture and live imaging. Embryo slice culture and filming was conducted as previously reported with some modifications 1,58 . Briefly, the GR of Stella GFP mouse embryos were dissected into the slice culture medium (Hepesbuffered DMEM/F-12 with 0.04% lipid-free BSA and 100 U/ml penicillin/streptomycin). The transverse sections were then placed onto millicell CM organ culture inserts (Millipore) pre-coated with mouse collagen IV (50 µg/ml in 0.05 M HCl) (Beckton Dickson) in a 13 mm glass-bottom plate (MatTek, Massachusetts) filled with pre-warmed slice culture medium containing 40 µM of purmorphamine (Cayman Chemical Company), cyclopamine (Stratech Scientific), tomatidine (Sigma-Aldrich) or dimethyl formamide (Sigma-Aldrich), which were optimised for their efficacy with minimal toxicity (Supplementary Fig. 2). The plate was then mounted onto Zeiss LSM510 confocal time-lapse microscope stage in a humidified CO 2 chamber at 37.0 ± 0.5°C. Live imaging was performed at 15 min (0.25 h) interval for a minimum of 10-15 h. Z-stack images of 11-13 µm sections spanning 185-230 µm total thickness were extracted per time point and created into a movie using ImageJ. Velocity was calculated using the formula V = [sqrt (dx 2 + dy 2 )](p)/ 0.25h, where dx is the change in the x-axis, dy is the change in the y-axis and p is the pixel size in µm. Typically, for a 10 hour movie, 40 velocity measurements were generated per cell, which was averaged to obtain an overall velocity for that cell. Tracking was performed using ImageJ plugin Chemotaxis and Migration tool (ibidi GmbH, Martinsried, Germany) on the GFP-positive PGCs that remained in focus and viable for the entire duration of the movie.   , Shh (n = 13), ****P < 0.0001), p-Creb (DMF (n = 9), Shh (n = 7), ****P < 0.0001) and p-Src (DMF (n = 10), Shh (n = 11), ****P < 0.0001) but no significant change in Gli3 (DMF (n = 9), Shh (n = 9), P = 0.8542). Error bars represent SD. Unpaired t-test with Welch's correction. Source data are provided as a Source Data file. To generate the primary culture of GR cells, the dissected GR tissues were digested in 0.25% trypsin, passed through a 0.4 μm cell strainer and suspended in DMEM/L-15 medium supplemented with 20% knockout serum replacement (Invitrogen), 2 mM L-glutamine, 0.1 mM non-essential amino acids and 0.1 mM 2-mercaptoethanol (Sigma-Aldrich), before being plated onto 0.1% gelatin-coated plates. Cells were incubated in 0.5% serum-containing media before various treatments diluted in serum-free medium.
Chemotactic migration assay. Coverslips (8 mm) cut into halves using a diamond blade were coated with 0.001% poly-L-Lysine. For the source of Shh secretion, 293 cells transfected with pCS2 empty vector or encoding full-length human Shh cDNA (a gift from Adrian Salic) were plated without the feeder. For PGCs, the GR cells were plated onto NIH3T3/Cas9 feeder with the targeted knockout of Wdr11, which was pre-treated with Mitomycin-C (5 μg/ml) for 2 hours. Once the cells settled down, the two halves of coverslips were carefully rejoined in a 24 well plate containing PGC primary culture medium. After 24 hours incubation, the cells were fixed with 4% paraformaldehyde and stained with anti-SSEA and DAPI. The number of cells present on each side of the coverslips was counted to obtain a percentage of migrated PGCs toward the empty vector or Shh-transfected 293 cells. The distance of PGC migration was measured from the midline.
Random motility assay. Dissected GR tissues were prepared to generate single cell suspensions of GR primary culture and plated onto the NIH3T3 feeder layer pretreated with Mitomycin-C (5 μg/ml) and incubated in 0.5% serum media before treatment with Shh-N, purmorphamine, cyclopamine, tomatidine, DMF (40 µM) and visomodegib (10 µM), all diluted in serum-free medium. Non-directional motility of PGCs was measured by live imaging of GFP-positive cells captured every 15 min for 16 h using Nikon A1R laser scanning confocal microscope in a humidified CO 2 chamber at 37.0 ± 0.5°C. Time-lapse image sequences were analysed using the Chemotaxis and Migration Tool 2.0 plug-in software (Ibidi GmbH).
Flow cytometry and RNA analyses. Cells from dissected GR tissues of E10.5 Stella GFP embryos were resuspended in the sorting buffer (1 mM EDTA, 25 mM HEPES at pH7.0, 1% FBS) and GFP + and GFPcell population was separated using a MoFlo XDP high-speed cell sorter (Beckman Coulter). FACS data were analysed using FlowJo software. Positive cells had an intensity of >2 on log scale of GFP detection using 488 nm laser for excitation and a filter (529/28 nm) for detection. Cells were selected using FSC (height) versus SSC (height) and a gate was prepared to eliminate debris or abnormally shaped cells. At least 1700 GFP + cells repre-senting~0.1% of the total sorted cells collected from seven embryos were analysed per experiment. Total RNA from the sorted cells was extracted using PicoPure RNA Isolation Kit (Arcturus) and converted to cDNA using SuperScript IV reverse transcriptase (Invitrogen). Total RNA from mouse tissues was extracted using RNAeasy kit (QIAGEN) and converted to cDNA using NanoScript 2 Reverse Transcription kit (Primer Design). The cDNA obtained was analysed by either GoTaq G2 Hot Start PCR mix (Promega) or Maxima SYBR Green qPCR Master Mix (Thermo Scientific) on a Light-Cycler 2.0 (Roche) using the 2 −ΔΔCT normalised to 18s rRNA or Gapdh. See Supplementary Table 1 for the primers used.
Western blot and co-immunoprecipitation. Total protein was extracted in a lysis buffer (50 mM HEPES, 150 mM NaCl, 10% glycerol, 1% Nonidet P-40, and 1 mM EDTA) containing protease/phosphatase inhibitors (Sigma-Aldrich), separated by SDS-PAGE, and transferred onto Hybond-ECL membrane (Amersham) which was probed with primary antibodies diluted in blocking buffer (5% skim milk in TBS with 0.05% Tween 20 (TBST)). After washing in TBST, the secondary antibodies conjugated with horseradish peroxidase was added before analyses by enhanced chemiluminescence (GE Healthcare). Co-IP was performed using pre-cleared lysate (1-2 mg protein) and the immune complexes captured on protein A/G-Agarose beads (Santa Cruz Biotechnology) were analysed by Western blot. Uncropped blots are included in the source data file. PGCs counted from every other sections of the entire embryo at E10.5 was combined per embryo from each genotype. The numbers of PGCs populated in the genital ridge, mesentery and hindgut are compared between WT and Gas1 KO. Error bars represent SD from three independent embryos after unpaired t-test. ****P = 0.000521 (gr), ****P = 0.000033 (me), ****P = 0.002814 (hg). c Litter size is shown as the number of pups per litter for each female caged with a WT male for 9 months. Unpaired t-test indicates a significant difference (***P = 0.0002) in the litter size by Gas1−/− mice (n = 5) compared with Gas1+/− (n = 3), which were phenotypically normal. Error bars represent SD. Source data are provided as a Source Data file.