Ptch2 mediates the Shh response in Ptch1^−/− cells

Shh signaling is regulated by the interaction between Ptch1 (Marigo et al., 1996; Stone et al., 1996) and Smo (Marigo et al., 1996; Murone et al., 1999). Shh binding to Ptch1 releases the Ptch1-mediated inhibition of Smo (Taipale et al., 2002). Smo then localizes to the cell surface (Incardona et al., 2002) and subsequently to the primary cilium (Milenkovic et al., 2009) where it mediates the activation of the Shh response (Corbit et al., 2005; Huangfu and Anderson, 2005; Rohatgi et al., 2007). This model explains the widespread activation of the Shh response observed in the absence of Ptch1 (Goodrich et al., 1997).

Drosophila genetics strongly supports the canonical model of Hh signaling by demonstrating that the loss of Ptch is epistatic to the loss of Hh (Bejsovec and Wieschaus, 1993). In amniotes, there are two Ptch homologs, Ptch1 and Ptch2, and of these two genes Ptch1 appears to be the most important. The loss of Ptch1 results in an embryonic lethal phenotype characterized by the widespread upregulation of the Shh response, including extensive induction of Shh expression and ventral identity in the developing neural tube (Goodrich et al., 1997). By contrast, Ptch2^–/– mice are fertile and viable, but develop skin abnormalities characterized by basal cell hyperplasia (Nieuwenhuis et al., 2006). As these data suggested that the functions of Ptch1 and Ptch2 are largely non-overlapping, Ptch1^–/– cell lines have been used extensively for their high level of cell-autonomous activation of the Shh response. For example, neuralized cells derived from Ptch1^–/– mouse embryonic stem cells (mESCs) acquire a phenotype typically associated with the induction of the Shh response without the inclusion of Shh in the medium (Crawford and Roelink, 2007). This is consistent with a ligand-independent induction of the Shh response in cells devoid of Ptch1. Similarly, Ptch1^–/− fibroblasts have been widely studied for having a constitutively upregulated Shh response (Taipale et al., 2000).

We now demonstrate that Ptch1^–/– fibroblasts display Shh chemotaxis that is indistinguishable from wild-type cells, indicating that Ptch1^–/– is not required to mediate this Shh response. Furthermore, we show that upregulation of the Shh response in neuralized embryoid bodies (NEBs) derived from Ptch1^–/– mESCs is dependent on endogenously expressed Shh by mutating the Shh locus in Ptch1^–/– mESCs, and by treating these cells with a Shh-blocking antibody. The role of Ptch2 in mediating the Shh response in the absence of Ptch1 was further supported by the observation that Ptch1^–/–;Ptch2^–/– cells cannot respond to activators of the Shh response, and that expression of a dominant-negative Ptch2 mutant results in an activation of the Shh response. Together, these results demonstrate that the Shh responses observed in Ptch1^–/– cells can be mediated by Ptch2.

Ptch1 is a putative member of the resistance, nodulation and division (RND) family of proton-driven antiporters (Taipale et al., 2002). This transporter family shares a conserved aspartic acid residue in the fourth transmembrane region (Van Bambeke et al., 2000). Mutating this residue in other members of the RND family, including Disp1 (Etheridge et al., 2010), results in dominant-negative molecules that are able to inhibit the antiporter function of normal endogenous proteins. Expressing a Ptch1 allele lacking antiporter activity (Ptch1D499A) (Taipale et al., 2002) in the chick neural tube does not recapitulate the loss of Ptch1 function in mouse embryos (Goodrich et al., 1997), as we did not observe an increase in Shh activation as assessed by changes in Shh-mediated dorsoventral patterning (Fig. 1A,B). On occasion, we did find some cells expressing Pax7 ectopically, indicating a minor loss of Shh signaling (supplementary material Fig. S1). We attribute this to the ability of Ptch1D499A to sequester Shh away from endogenous Ptch1, leading to both an autonomous and non-autonomous inhibition of Shh signaling.

For members of the RND family to act as dominant negatives, they must retain the ability to form trimers (Nikaido and Takatsuka, 2009). It remains a possibility that the electroporated mouse Ptch1 cannot form trimers with endogenous chicken Ptch1. We therefore tested whether chicken Ptch1 lacking antiporter activity was able to induce the Shh response, after misexpression in the developing neural tube. Again, we observed little effect on neural tube patterning (supplementary material Fig. S1), indicating that suppressing the proton-driven antiporter activity of Ptch1 has little effect on the Shh response. The inability of Ptch1D499A to act as a dominant-negative inhibitor of endogenous Ptch1 raises the issue of whether its proton-driven antiporter activity is important in regulating the Shh response at these stages of development.

Ptch1Δloop2, a deletion mutant of Ptch1 that is unable to bind Shh is a potent inhibitor of the Shh response. Consistent with an earlier observation (Briscoe et al., 2001), we found that expression of Ptch1Δloop2 had a strong cell-autonomous inhibitory effect on the Shh response (Fig. 1C,D). To assess whether this effect is mediated by its antiporter activity, we expressed a Ptch1 allele that was unable to bind Shh but also lacks antiporter activity: Ptch1Δloop2/D499A. Ptch1Δloop2/D499A had no effect on Shh activity, based on the lack of ectopic cell-autonomous Pax7 induction, and only mildly inhibited motor neuron induction, as determined by Isl1/2 expression (Fig. 1E,F). The dramatic difference between the strong inhibition of the Shh response by Ptch1Δloop2 and the mild effects of Ptch1Δloop2/D499A demonstrates that the proton-driven antiporter activity is crucial for Smo inhibition by Ptch1Δloop2. Importantly, the loss of repressive activity of Ptch1 did not automatically result in the cell-autonomous activation of the Shh response, indicating that Ptch1Δloop2/D499A is not a strong inhibitor of endogenous Ptch1 function.

To assess the activities of the Ptch1 mutants in the absence of endogenous Ptch1 activity, we expressed them in Ptch1^–/– immortalized mouse embryonic fibroblasts (MEFs). Ptch1^–/– MEFs are devoid of functional Ptch1 protein (Rohatgi et al., 2007) and have an autonomously upregulated Shh response (Taipale et al., 2000) that can be measured by measuring the integration of the lacZ gene into the Ptch1 locus (Goodrich et al., 1997). We found that SAG, a Smo agonist, further induced Shh pathway activation in Ptch1^–/– MEFs, whereas cyclopamine reduced Shh pathway activity (Chen et al., 2002; Taipale et al., 2000) (Fig. 2A). This indicates that, despite the absence of Ptch1, Smo can be activated or inhibited in these cells. The addition of ShhN (a truncated and soluble form of Shh) also increased the Shh response, indicating that there is a Ptch1-independent response to Shh.

In line with their abilities to inhibit Smo, we found that expression of Ptch1 and Ptch1Δloop2 decreased the autonomous Shh response, relative to control transfection with Disp1, which was normalized to 100 (Fig. 2B). In these experiments, we measured the Shh response by co-transfecting a construct in which luciferase is driven by a Shh-inducible promoter (Taipale et al., 2002). Furthermore, whereas Ptch1^–/– cells expressing Ptch1 were responsive to ShhN, cells expressing Ptch1Δloop2 were unresponsive (Fig. 2B), consistent with the inability of Ptch1Δloop2 to bind Shh, mirroring our observations in vivo (Fig. 1C,D). For comparison, Ptch1^+/+ MEFs had a 2.4-fold induction in response to ShhN (not shown). In line with their abilities to inhibit Smo, we found that expression of Ptch1 and Ptch1Δloop2 decreased the autonomous Shh response (Fig. 2C). To test whether the downregulation of the Shh response pathway required the antiporter activity of Ptch1, we expressed the antiporter mutant and found an increase in the autonomous activation of the Shh response when compared with wild-type Ptch1 (Fig. 2B,C). Nevertheless, these Ptch1 mutants repressed Smo to a much greater degree than the negative control, Disp1. Moreover, cells expressing Ptch1 antiporter mutants retained their sensitivity to ShhN (Fig. 2B). This demonstrates that Smo inhibition can be regulated independently of Ptch1 antiporter activity.

Combining mutations that antagonize both the proton-driven antiporter activity of Ptch1 as well as Shh binding in the same molecule resulted in forms of Ptch1 that blocked the response to ShhN in Ptch1^–/– cells (Fig. 2B). We expanded this experiment using different mutations in the putative proton pore, replacing the crucial aspartic acid with a lysine or tyrosine residue (Ptch1D499K and Ptch1D499Y), and combined these mutations with the Shh binding deletion (Ptch1Δloop2/D499K and Ptch1Δloop2/D499Y). To address the ligand dependency, we treated these cells with ShhN or 5E1, a Shh-specific monoclonal antibody. Cells expressing Ptch1 mutants with Shh-binding capacity (or control Disp1 transfected cells) responded to 5E1 with a repressed Shh response (Fig. 2C). Similar to Ptch1D499A, we found that cells expressing Ptch1D499K or Ptch1D499Y retained their ability to respond to ShhN, but mutants combining the antiporter activity mutations with the loop2 deletion resulted in forms of Ptch1 that were unable to mediate the Shh response in Ptch1^–/– cells, but nevertheless inhibited Smo when compared with our control, Disp1 (Fig. 2C). These results raise the question of how forms of Ptch1 that are unable to bind Shh and repress Smo can nevertheless still inhibit the Shh response. As these experiments were performed in Ptch1^–/– cells, Ptch1Δloop2/D499X mutant alleles must inhibit the Shh response independently of endogenous Ptch1. They also support the notion that Shh can induce Smo activity via a mechanism that does not involve Ptch1 antiporter activity.

Mouse embryonic stem cells (mESCs) that have aggregated in defined medium containing retinoic acid, form neuralized embryoid bodies (NEBs) that closely resemble the early caudal neural tube (Wichterle et al., 2002). These NEBs are very sensitive to ShhN and several aspects of the Shh response can be assayed simultaneously. Consistent with the inhibitory role of Ptch1 on Smo, we have shown that, in the absence of exogenous Shh, Ptch1^–/– NEBs have higher expression levels of Shh-induced differentiation markers than wild-type NEBs. Smo is required for the Shh response and, accordingly, we found that Smo^–/– NEBs cannot respond to Shh (Crawford and Roelink, 2007). To determine whether endogenously produced Shh is responsible for the induction of Shh-mediated differentiation in the absence of Ptch1, Ptch1^–/– NEBs were cultured in the presence of the Shh-blocking antibody 5E1 (Ericson et al., 1996) or an α-Myc antibody (9E10) (Chan et al., 1987) as a control. After 5 days in culture, NEBs were analyzed for expression of Isl1/2 and Nkx2.2, transcription factors that are induced by activation of the Shh response (Briscoe et al., 1999), and for Pax7 expression, which is inhibited by Shh signaling (Ericson et al., 1996). In Ptch1^–/– NEBs cultured with 5E1, Isl1/2 and Nkx2.2 expression was reduced compared with the 9E10-treated Ptch1^–/– NEBs (Fig. 3A,B,G,H). This loss of ventral cell types was concomitant with an increase in Pax7 expression, further demonstrating that the upregulation of the Shh response in Ptch1^–/– cells is not due to an autonomous loss of Smo inhibition, but is at least in part dependent on the presence of Shh in the NEBs (Fig. 3I). Both RT-PCR and immunofluorescence showed abundant Shh expression in Ptch1^–/– NEBs (Fig. 3E,F). Moreover, in the absence of Ptch1 function, the number of cells expressing Isl1/2 and Nkx2.2 was increased by the Smo agonist SAG (Chen et al., 2002), and the number of cells expressing Pax7 was suppressed, regardless of the presence of 5E1 (Fig. 3C,D,G,H), indicating that, even in absence of Ptch1, Smo was not fully activated.

To demonstrate that Ptch1^–/– cells can respond to Shh delivered in trans, we generated mixed NEBs composed of varying ratios of Ptch1^–/– and Smo^–/– mESCs (Fig. 3J-O). Neuralized Smo^–/– mESCs are unable to respond to Shh itself but this particular clone expresses Shh (Fig. 3M). Shh derived from these Smo^–/– cells induced Nkx2.2 and Isl1/2 expression in Ptch1^–/– cells. This induction could be blocked by the inclusion of 5E1, demonstrating that this induction is mediated by Shh (Fig. 3P). Wild-type mESCs did not display an induction of their Shh response when co-aggregated with Smo^–/– mESCs (Fig. 3Q). It appears that the concentration of Shh provided by the Smo^–/– cells within the NEB is not sufficient to activate the response in wild-type cells. These results demonstrate that Ptch1^–/– mESCs are more sensitive to Shh than are wild-type mESCs, but nevertheless remain dependent on the ligand for full induction of the Shh response. Based on our results using blocking antibodies, we wanted to further address the requirement of endogenous Shh in cells by creating genetic nulls.

To determine whether endogenous Shh mediates the Shh response in Ptch1^–/– cells, we made mutations in the Shh locus of Ptch1^+/− and Ptch1^–/– mESCs using transcription activator-like effector nucleases (TALENs) (Cermak et al., 2011) directed against an N-terminal coding sequence of Shh. The Shh^–/– clones were detected by sequencing of PCR products surrounding the area targeted by the TALENs. The clones used for subsequent experiments had small deletions in both Shh alleles that caused premature stop codons, resulting in protein products that truncated soon after the signal sequence. NEBs derived from these cells were grown in the absence or presence of 5 nM ShhN and the induction of Nkx2.2-positive cells was assessed. In NEBs derived from both the Ptch1^+/−;Shh^–/– and Ptch1^–/–;Shh^–/– cells, we observed a significant induction of Nkx2.2-positive cells by ShhN (Fig. 4A,B). In addition, we found that the Ptch1 promoter, as measured by the induction of lacZ was induced in the Ptch1^–/–;Shh^–/– cells in response to ShhN (Fig. 4C). These results again show that cells without Ptch1 are sensitive to Shh, and comparing these data with those shown in Fig. 3 suggests that the upregulation of the Shh response in Ptch1^–/– cells is at least in part mediated by endogenous Shh. Assaying Shh-induced differentiation in NEBs takes about 5 days, and is thus possibly subject to secondary effects. To address this, we assessed Shh chemotaxis, which is a more direct measurement of Smo activation

To assess the more direct consequences of Ptch1-independent Shh signaling, we tested the ability of Ptch1^–/– cells to migrate towards a localized source of Shh. Shh chemotaxis occurs independently of transcription but does require Smo (Bijlsma et al., 2012) and thus allows for the assessment of very early events following the perception of Shh ligand. We tested the migratory response of Ptch1^–/– and Ptch1^+/+ immortalized MEFs in a modified Boyden chamber assay (supplementary material Fig. S2), and found that the migratory response to recombinant ShhN and purmorphamine of Ptch1^–/– cells was very similar to that of Ptch1^+/+ cells (Fig. 5A,B). The migration was specifically towards ShhN, as it was inhibited by the inclusion of 5E1 in the upper and lower compartments of the Boyden chamber. Migration to both ShhN and purmorphamine could be ablated by expressing Ptch1Δloop2 or performing the assay in a Smo-deficient background (Fig. 5A,B). These results indicate that Ptch1 is not required for migration of MEFs towards sources of Shh, further supporting the notion that Ptch1 is not required for cells to respond to Shh.

We next examined what receptors could potentially perceive Shh in Ptch1^–/– cells. Several Shh-binding proteins such as Gas1, Cdon and Boc have been proposed to function as co-receptors acting in conjunction with Ptch1 (Allen et al., 2011; Izzi et al., 2011; Tenzen et al., 2006). We tested whether these molecules mediated the Shh response in the absence of Ptch1. Stable Gas1 knockdown in Ptch1^–/– MEFs did not affect Shh chemotaxis, but stable knockdown of Cdon and Boc diminished this response (Fig. 5C). This effect was confirmed using a transient silencing strategy (supplementary material Fig. S3). Ptch1^–/– MEFs stably expressing Cdon or Boc showed an increased chemotactic response to ShhN (Fig. 5D). These experiments indicate that the related Shh (co-)receptors Cdon and Boc mediate Shh chemotaxis even in the absence of Ptch1. As Boc and Cdon are thought to form complexes with Ptch1 and Shh (Izzi et al., 2011), it is a distinct possibility that that they can also form such complexes with Ptch2.

To assess whether Ptch2 is required for the Ptch1-independent response, we mutated the Ptch2 locus in Ptch1^–/– mESCs and found that, in NEBs derived from Ptch1^–/–;Ptch2^–/– cells, the Shh response was higher than in the Ptch1^–/– and Ptch1^+/− NEBs (Fig. 6A-C). This indicates that the Shh response in Ptch1^–/– cells is inhibited by Ptch2. The Ptch1^–/–;Ptch2^–/– NEBs had more Nkx2.2+ cells and fewer Isl1/2+ cells than the Ptch1^–/– NEBs. We conclude that cells in Ptch1^–/–;Ptch2^–/– NEBs acquire an even more ventral phenotype, resulting in a loss in the number of motorneurons induced. We indeed found that in Ptch1^–/–;Ptch2^–/– NEBs the Shh-inducible Ptch1 promoter was considerably more active than in Ptch1^–/– and Ptch1^+/− NEBs (Fig. 6D). We were unable to further alter the Shh pathway activation level by inclusion of the Smo agonist SAG (Fig. 6D). This indicates that, in the absence of both Ptch1 and Ptch2, Smo activation via its heptahelical domain is saturated (Chen et al., 2002). It is thought that Ptch1, via its proton-driven antiporter activity relocalizes a sterol that inhibits Smo at this heptahelical site, and our results thus indicate that Ptch2 could also fulfill this role.

We assessed whether the proton-driven antiporter activity of the Ptch1 paralog Ptch2 is involved in regulating the Shh response in vivo. Ptch2 has been shown to modulate the Shh response in mouse embryos (Holtz et al., 2013). Whereas Ptch1Δloop2 misexpression in chick embryos causes a significant cell-autonomous inhibition of the Shh response (Fig. 1), we have been unable to find any autonomous inhibitory effects of Ptch2Δloop2 (not shown), suggesting that the inhibitory action of Ptch2 in the developing neural tube is less important than that of Ptch1, consistent with the normal development of Ptch2^–/– embryos. To create a dominant-negative allele of Ptch2, we mutated the aspartic acid analogous to the one in Ptch1 to alanine: Ptch2D469A. Like Ptch1D499A, expression of Ptch2D469A, did not cause cell-autonomous changes in Shh-induced patterning (Fig. 7A). Co-expression of Ptch2D469A and Ptch1D499A also failed to affect neural tube patterning cell-autonomously (not shown). However, in four out of 20 embryos electroporated with Ptch2D469A, we found widespread upregulation of Shh expression (Fig. 7B) and of the Shh response (Fig. 7C,D). These embryos were characterized by a bilateral, additional Nkx2.2 domain, localized dorsal to the normal domain (Fig. 7C), or by the widespread expression of Isl1/2 and Nkx2.2 (Fig. 7D). The induction of the Shh response was largely non-cell autonomous, and it is likely that the ectopic induction of Nkx2.2 and Isl1/2 is a consequence of the expanded domain of Shh-expressing cells. This phenotype bears striking resemblance to the phenotype observed in Ptch1^–/– mouse embryos, which might indicate that Ptch2D469A can inhibit Ptch1 function in trans.

Our results demonstrate that the loss of Ptch1 function is not always sufficient to cell-autonomously initiate maximal Smo-dependent Shh responses, and that Ptch2 mediates the residual responsiveness retained in Ptch1^–/– cells. In flies, based on the embryonic cuticular phenotype, Ptch is epistatic to Hh, and Smo is epistatic to Ptch, consistent with a cell-autonomous activation of Smo in the absence of Ptch. The phenotype of Ptch1^–/– mouse embryos is also consistent with a cell-autonomous activation of Smo, although this issue is clouded by the widespread induction of Shh (Goodrich et al., 1997).

The induction of Shh is in part responsible for the upregulation of the Shh response in the absence of Ptch1. This is evident by 5E1-mediated blockade of endogenous Shh ligand in Ptch1^–/– neuralized embryonic bodies (NEBs), which results in the loss of ventral cell types, presumably by preventing Shh binding to its receptors. This notion is further supported by NEBs derived from Ptch1^–/–;Shh^–/– mESCs, which can respond to exogenous ShhN. The ability of Ptch1^–/– NEBs to respond to endogenous ligand highlights the importance of Shh receptors distinct from Ptch1 within these cells. These results indicate that the interpretation of the phenotype of Ptch1^–/– embryos is incomplete. Our results predict that the phenotype of Ptch1^–/–;Shh^–/– embryos will be different from Ptch1^–/– embryos, and that this difference can be attributed to Ptch2-mediated Shh signaling.

Ptch1^–/– MEFs also retain the ability to respond to Shh, both transcriptionally and via cell migration. Although Shh chemotaxis is very similar in Ptch1^–/– and wild-type MEFs, the Shh-induced transcriptional response of Ptch1^–/– MEFs is weaker than that of Ptch1^+/+ MEFs. It is possible that Ptch1-independent signaling is more efficient in mediating the migratory than the transcriptional response. Boc and Cdon are Shh co-receptors required both for the transcriptional response (Allen et al., 2011) as well as neural pathfinding to Shh (Izzi et al., 2011). It is conceivable that, like the transcriptional response, Ptch2 can mediate Shh chemotaxis. Boc and Cdon have been proposed to make a tripartite complex with Ptch1 and Shh (Izzi et al., 2011). The Boc and Cdon requirement for Shh chemotaxis in Ptch1^–/– MEFS suggests that they may also form complexes with Shh and Ptch2.

As a member of the RND family of proton-driven antiporters, Ptch1, like Drosophila Ptc (Lu et al., 2006), is expected to function as a trimer, mediating its transporter activity via a rotatory mechanism (Nikaido and Takatsuka, 2009; Nikaido and Zgurskaya, 2001). In the absence of endogenous Ptch1 as a trimerization partner, the Ptch1 paralog Ptch2 could fulfill this role. Like Ptch1, Ptch2 expression is upregulated in response to Shh, resulting in a significant overlap in their expression domains (Holtz et al., 2013; Resende et al., 2010). This leaves open the possibility that Ptch1 and Ptch2 can form heterotrimers (Rahnama et al., 2004), and that Ptch1/2 heterotrimers in which Ptch1 subunits lack the Shh-binding loop cannot mediate the Shh response. RND heterotrimerization is not without precedent. MdtB and MdtC, two bacterial RND proteins that are encoded within a single operon, must be co-expressed in order for drug efflux (a measure of activity) to occur. MdtB and MdtC share 45% sequence identity, which is much less than the 56% sequence identity shared between Ptch1 and Ptch2, further supporting the possibility that Ptch1 and Ptch2 could also form heterotrimers. The Mdt complex is an MdtB₂C₁ heterotrimer. Importantly, mutating the proton translocation pathway of MdtB blocked transporter activity, whereas the analogous mutation in MdtC did not affect the activity of the trimer (Kim et al., 2010). This result indicates that subunits of RND heterotrimers can contribute different activities to the trimer, and that the proton-driven antiporter activity is not required to be active in all three subunits.

The observation that Ptch1Δloop2D499A both inhibits Smo activity in Ptch1^–/– cells and is insensitive to regulation by Shh would support the notion that, in Ptch1/2 heterotrimers, Smo inhibition is mediated by Ptch2 subunits. This is further supported by the observation that Ptch2D469A expression can activate the Shh response when expressed in vivo. An interpretation of this result is that the high levels of expression reached in electroporation drive the formation of Ptch1/2 heterotrimers in which the Ptch2 subunits fail to mediate proton-driven antiporter activity and thus prevent the heterotrimers from inhibiting Smo. Together, these observations support the model in which the Ptch1 and Ptch2 subunits of a Ptch1/2 heterotrimer mediate distinct activities. Ptch1, via its Shh-binding loop 2 imparts Shh sensitivity upon the heterotrimers, independent of its proton-driven antiporter activity. Ptch2, however, is not particularly sensitive to Shh, but mediates the antiporter activity.

The non-cell-autonomous activation of the Shh response resulting from Ptch2D496A expression in the developing chick neural tube is consistent with the predicted role of Ptch2 on Smo activity. Very strong activation of the Shh response can result in the induction of Shh expression (Ericson et al., 1996), and consistent with this, we find that expression of Ptch2D496A results in an ectopic or expanded population of Shh-expressing cells. It is likely that Shh released from these Shh-expressing cells mediates the subsequent ectopic induction of Nkx2.2 and Isl1/2 on both sides of the neural tube. The induction of Shh expression could explain the apparent cell non-autonomous effects of Ptch2D496A expression.

The origin of these ectopic Shh-expressing cells remains unclear but their presence indicates incorrect patterning of the neural tube. Shh-mediated induction of Shh expression occurs in the node (Charrier et al., 1999,, 2002), and soon after when the nascent notochord induces the floor plate (Placzek et al., 1993). The high levels of Ptch1 and Ptch2 expression around the node (Resende et al., 2010) might render this structure particularly sensitive to the consequences of Ptch2D496A overexpression, and explain the nature of the phenotype observed. It is striking that Ptch2D469A overexpression causes a phenotype reminiscent of the loss of Ptch1 in mouse embryos (Goodrich et al., 1997).

In summary, our results reveal that the upregulation of the Shh response in Ptch1^–/– cells is in part mediated by Shh and we propose that Ptch2 acts as a Shh receptor. The function of Ptch2 becomes more apparent in the absence of Ptch1. As Ptch2^–/– mice are viable and fertile, it is obvious that the role of Ptch1 in the regulation of Smo activity is greater than that of Ptch2. Ptch1 can compensate for the loss of Ptch2, but not vice versa (Rahnama et al., 2004). Nevertheless, the increased tumor incidence in Ptch1^+/− mice lacking one or two Ptch2 alleles (Lee et al., 2006; Smyth et al., 1999) is most easily explained by the ability of Ptch2 to regulate Smo activity in the absence of Ptch1. The modulation of Ptch2 activity by Shh provides a simple explanation for why tumors in Ptch1^+/− mice often occur at known locations of Shh signaling, such as the skin and the cerebellum (Goodrich et al., 1997; Stone et al., 1996), as we predict that loss of function of the normal Ptch1 allele does not render these cells completely ligand independent.

Further indications that the functions of Ptch1 and Ptch2 are not entirely overlapping comes from the observation that, in ptc1^–/–;ptc2^–/– zebrafish embryos, a more extensive upregulation of the Hh response is observed than in ptc1^–/– embryos (Koudijs et al., 2008). In mouse embryos without Ptch2 and with Ptch1 expressed by a constitutively active and Shh-insensitive promoter, a mild upregulation of the response is observed. Subsequent loss of the Shh antagonist Hhip results in a strong upregulation of the Shh response (Holtz et al., 2013), indicating that Ptch2 provides ligand-dependent feedback on the Shh response. However, neither the mouse mutants described by Holtz et al. (due to the presence of constitutively expressed Ptch1) nor the ptc1^–/–;ptc2^–/– zebrafish (due to the partial genome duplication) address the consequence for the Hh response in the complete absence of Ptch protein activity.

The question remains of to what degree is the observed activation of the Shh response in cells without Ptch1 ligand dependent in vivo? The phenotype of Shh^–/–;Ptch1^–/– embryos is not yet known, but any slight modification of the Ptch1^–/– phenotype due to the loss of Shh could be attributable to Shh signaling via Ptch2 (Lee et al., 2006). Similarly, comparing early phenotypes of Ptch1^–/– and Ptch1^–/–;Ptch2^–/– embryos could demonstrate further roles of Ptch2 when Ptch1 is absent.

Cyclopamine was from Biomol. Purmorphamine and SAG were from EMD Biochemicals (Darmstadt, Germany). Cell Tracker Green was from Invitrogen. Recombinant ShhN was from R&D Systems.

Antibodies for mouse Pax7 (1:20), HB9 (1:10), Nkx2.2 (745-A5; 1:10) and Shh (5E1; 1:10) were from the Developmental Studies Hybridoma Bank. Rabbit α-GFP (1:1000) was from Invitrogen (A-11122) and guinea pig α-Isl1/2 (1:10,000) was a gift from Dr Thomas Jessell (Columbia University). In all experiments, Alexa488-, Alexa568- or Alexa647-conjugated secondary antibodies (Invitrogen, A-11001, A-11011, A-20990, A-21450) were used at 1:1000. Nuclei were stained with DAPI (Invitrogen, D3571).

Hamburger-Hamilton (HH) stage 10 Gallus gallus embryos were electroporated caudally in the developing neural tube using standard procedures (Meyer and Roelink, 2003). Embryos were incubated for another 48 h following electroporation, fixed in 4% PFA, mounted in Tissue-Tek OCT Compound (Sakura) and sectioned.

mESCs were neuralized using established procedures (Wichterle et al., 2002). NEBs treated with antibody were cultured in the presence of α-Shh (5E1) or α-Myc (9E10) (Developmental Studies Hybridoma Bank) conditioned in DFNB medium at 1:5 for the duration of the experiment. For ShhN treatment, NEBs were cultured in 5 nM of Shh-N-conditioned HEK293T supernatant or an equivalent volume of control, empty vector-conditioned HEK293T supernatant for the duration of the experiment. SAG was used at 200 nM. NEBs were harvested after 5 days in culture, fixed and stained for Isl1/2 and Nkx2.2, or Pax7 (Kawakami et al., 1997). NEBs were mounted in Fluormount and quantified for number of positive nuclei. Numbers of positive nuclei were normalized for the size of the NEB, and presented as the number in an average NEB.

mESCs were neuralized using established procedures (Wichterle et al., 2002). NEBs were cultured in 5 nM of Shh-N-conditioned HEK293T supernatant or an equivalent volume of control, empty vector-conditioned HEK293T supernatant. NEBs were collected after 5 days in culture and lysed into a standard lysis buffer [100 mM potassium phosphate (pH 7.8), 0.2% Triton X-100]. Ptch1^–/– MEFs were allowed to grow to confluence before switching to a low-serum medium (0.5% FCS) and ShhN, SAG or cyclopamine was added for another 24 h when cells were lysed. Lysates were analyzed using the Galacto-Light chemiluminescence kit (Applied Biosciences) for level of lacZ expression. lacZ readings were normalized to total protein of the sample as calculated from a standard Bradford assay.

RNA isolation was performed using Trizol (Invitrogen) according to the manufacturer's recommendations. For cDNA synthesis, SuperScript III (Invitrogen) was used on 1 µg RNA. PCR was performed using ReddyMix (Thermo Scientific).

pcDNA3.1 vector was obtained from Invitrogen. Ptch1Δloop2 was a gift from Dr Thomas Jessell (Columbia University, NY, USA). The Gli-luciferase reporter and the Renilla control were gifts from Dr H. Sasaki (Sasaki et al., 1997). Boc and Cdon constructs were gifts from Dr Krauss (Mount Sinai School of Medicine, NY, USA). Ptch1 was a gift from Dr Scott (Stanford University, CA, USA). Ptch2 was obtained from Thermo Scientific. Ptch1 and Ptch2 channel mutants were created by Quikchange mutagenesis (Stratagene). In the Ptch2 mutant, the aspartic acids residues at positions 469 and 470 were changed to alanines.

Smo^–/– fibroblasts (a gift from Dr Taipale), Ptch1^–/– and wild-type MEFs (gifts from Dr Scott) were cultured in Dulbecco's modified Eagle's medium (DMEM, Invitrogen) supplemented with 10% fetal calf serum (FCS, Invitrogen). mESCs were maintained under standard conditions without feeder cells.

HEK293T cells were transfected with psPAX2 and pMD2.G helper plasmids and pLKO.1 clones from the Sigma TRC1.0 shRNA library using FuGene HD (Roche). Following virus production, supernatant was filtered, and Ptch1^–/– MEFs were transduced using 1:1 supernatant with 5 µg/ml polybrene (Sigma), transduced cells were selected with 1 µg/ml puromycin. Knockdown was verified by RT-PCR.

Transient DNA transfections were performed using Effectene (Qiagen). DNA was used at a 1:15 ratio of DNA/Effectene. Cells were incubated with transfection complexes for 16 h. For RNA transfections, 100 nm siRNA was transfected using 5 µl DharmaFect 3 (Dharmacon) in OptiMem (Invitrogen).

Ptch1^–/– MEFS were transfected with Ptch1 mutant constructs, Gli-luciferase and CMV-Renilla. Cells were allowed to grow for 2 days after transfection before switching to a low-serum medium (0.5% FCS), ShhN-conditioned medium or 5E1-conditioned medium for an additional 2 days. Cells were subsequently lysed and luciferase activity was determined by using the Dual-Luciferase Reporter Assay System (Promega).

Cells were lysed using LDS sample buffer (Invitrogen) and subjected to SDS-PAGE. Proteins were transferred to PVDF membranes, blocked with 5% milk/Tris-buffered saline with 0.1% Tween-20 (TBS-T), and incubated in 9B11 α-Myc 9B11 (Cell Signaling Technology, 2276) at 1:5000, or α-FLAG M2 (Sigma, F1804) at 1:2000. Appropriate HRP-conjugated secondary antibodies (Invitrogen, 62-6520) were used at 1:5000. Proteins were visualized using a FujiFilm LAS 4000 imager.

Migration assays were performed as previously described (Bijlsma et al., 2007). Cells were labeled with 10 µM CellTracker Green (Invitrogen) according to manufacturer's protocol. After labeling, cells were detached with 5 mM EDTA, resuspended in serum-free medium, and transferred into FluoroBlok Transwell inserts (BD Falcon) at ∼5×10⁴ cells per insert. Chemoattractant was added to the bottom compartments of the Transwell plates and GFP-spectrum fluorescence in the bottom compartment was measured in a Synergy HT plate reader (BioTek) every 2 min for 99 cycles (∼3 h). For analysis of data, see supplementary material Fig. S2.

The pCTIGTALEN expression vector was generated by cloning the BglII/SacI-digested TALEN ORF fragment of pTAL4 into the MCS of pIRES2-eGFP. After sequencing to confirm correct RVD architectures in pCTIG, constructs were further modified by replacing the SacI/BsrGI IRES:eGFP fragment in the pCTIG backbone with IRES:PuroR or IRES:HygroR fragments from pQCXIP and pIRES-hyg3, respectively, using PCR with primers containing SacI and BsrGI sites. Each pair of TALEN constructs targeting a locus was modified so that one construct co-expressed HygroR and the other PuroR, conferring transient resistance to both hygromycin and puromycin.

TALEN constructs targeting mouse Shh and Ptch2 were designed using Golden Gate cloning (Cermak et al., 2011) into the pCTIG expression vector. The following repeat variable domain architectures were generated: Shh, 5′ TALEN: NN HD HD HD HD NN NN NN HD NG NN NN HD HD NG NN NG; 3′ TALEN, NN HD HD NN HD HD NG HD NG NG NG HD HD NI NI NI HD; Ptch2, 5′ TALEN, NN NN HD NG NG HD NN NI NN HD NG NG NI HD NG NG HD; 3′ TALEN, NG HD NG NN NN NI NG HD HD NG NN HD NI HD HD HD HD.

mESCs were transfected with paired TALEN constructs using Lipofectamine 2000. One day after transfection, cells were passaged into ES medium containing 100 µg/ml hygromycin and 0.5 µg/ml puromycin, and cultured for 4 days. Selective medium was then removed and surviving mESC colonies were isolated, expanded and genotyped by sequencing PCR products spanning the TALEN-binding sites.

PCR screening was performed on cell lysates using primers flanking the Shh and Ptch2 TALEN-binding sites: Shh, (5′) TGGGGATCGGAGACAAGTC and (3′) TCTGCTCCCGTGTTTTCCT; Ptch2, (5′) AAGGCACAGG-GAAAGAGAGTT and (3′) ACTTGCCTAGCTTGCACAATG. PCR products were sequenced using Sanger sequencing. Samples with mixed signals indicative of small INDEL mutations were TOPO cloned into PCR2.1 and sequenced to confirm allele sequences. A Ptch1^–/–;Shh^+/− mESC clone harboring a 5 bp deletion in Shh exon 1 was validated and re-transfected with Shh TALENs. A Ptch1^–/–;Shh^–/–clone heteroallelic for 5 bp and 4 bp deletions with predicted stop codons in exon 1 was characterized for its response to ShhN. A Ptch1^–/–;Ptch2^–/– clone was characterized with a 5 bp deletion in exon1 of Ptch2.

The Ptch1Δloop2 construct was a gift from Dr Jessell (Columbia University), Ptch1 and the Ptch1^–/– mESCs were gifts from Dr Scott (Stanford University). Smo^–/– MEFs were from Dr Taipale (University of Helsinki). The Smo^–/– mESCs were a gift from Dr McMahon (University of Southern California). We also thank S. Mohan, A. Mich and J. Hardin for technical assistance.