Inhibition of Neuroepithelial Patched-Induced Apoptosis by Sonic Hedgehog

See allHide authors and affiliations

Science  08 Aug 2003:
Vol. 301, Issue 5634, pp. 843-846
DOI: 10.1126/science.1085405


During early development in vertebrates, Sonic hedgehog (Shh) is produced by the notochord and the floor plate. A ventrodorsal gradient of Shh directs ventrodorsal patterning of the neural tube. However, Shh is also required for the survival of neuroepithelial cells. We show that Patched (Ptc) induces apoptotic cell death unless its ligand Shh is present to block the signal. Moreover, the blockade of Ptc-induced cell death partly rescues the chick spinal cord defect provoked by Shh deprivation. Thus, the proapoptotic activity of unbound Ptc and the positive effect of Shh-bound Ptc on cell differentiation probably cooperate to achieve the appropriate spinal cord development.

The neural tube is subjected during development to multiple events of neuronal proliferation, differentiation, and guidance, but it is also subjected to cell death. This is in part regulated by a ventral specialized structure, the floor plate (FP) (1, 2). The FP and notochord (No), which derive from the “organizer” (i.e., Hensen's node in avian) (3), are responsible for the synthesis of key cues that will determine the identity and the fate of ventral neurons of the spinal cord (1, 4). The developing FP generates a ventrodorsal gradient of Shh that in turn induces the differentiation of ventral neurons (5). However, Shh withdrawal leads to loss of ventral neural tube cell differentiation (6) and to a massive cell death (711). Added Shh is sufficient to suppress the induced cell death (7, 8, 10, 11).

Shh signal integration in the target cells occurs through at least two receptors, Smoothened (Smo) (5, 12) and Patched (Ptc) (13, 14). Smo, a 7–transmembrane domain receptor, affects differentiation by activation of transcription factors like Gli-1/3 but does not interact with Shh (12). Ptc, a 12–transmembrane domain receptor, physically interacts with Shh and regulates Smo activity (5, 12, 15). Ptc is a putative tumor suppressor frequently mutated in association with nevoid basal cell carcinoma syndrome and basal cell carcinoma (16). This putative tumor-suppressor function (14, 16) and the survival activity of its ligand resemble the observations made on DCC (Deleted in Colorectal Cancer) and UNC5H of the so-called dependence receptor family (1720). Such receptors create cellular states of dependence on their respective ligands by inducing apoptosis in settings in which the ligand is unavailable (17, 21, 22).

To determine whether Ptc is also a dependence receptor, we first transiently expressed full-length Ptc-1 (Fig. 1A) in 293T human embryonic kidney (HEK) cells and in immortalized neuroblast 13.S.24 cells. The cell death induction in 293T was associated with the expression of Ptc-1 (Fig. 1, B and C). Despite low expression at the cell membrane (Fig. 1C and fig. S1), Ptc-1 appeared to be a potent cell death inducer as compared with Bax (Fig. 1B). Expressing Smo had no effect on cell death, suggesting that Ptc-induced cell death is not related to the Ptc/Smo transducing module. Ptc-induced cell death was defined as apoptosis because Ptc-1 expression induced an increased caspase activity (Fig. 1D) and DNA fragmentation (Fig. 1, E and F, and fig. S1). Moreover, treatment with the general inhibitor for caspase-dependent apoptosis zVAD-fmk or coexpression with the caspases inhibitor protein P35 (23) inhibited cell death induction (Fig. 1, B, D, and E). We then investigated whether Shh modulated Ptc-induced apoptosis. Recombinant Shh was added to 293T or 13.S.24 cells expressing Ptc-1. The Shh treatment inhibited Ptc-induced cell death in a dose-dependent manner (Fig. 1, B, E, and F). Thus, Ptc receptor can be considered as a dependence receptor in cell culture.

Fig. 1.

Ptc-1 expression induces apoptosis, which is blocked by Shh. (A) Ptc-1 is a 12–transmembrane domain protein. EC, extracellular domain; IC, intracellular domain. (B) Ptc-1–induced cell death, as measured by trypan blue exclusion. 293T cells were transfected with mock plasmid (Cont.) or by the indicated expression constructs. Ptc-1–induced cell death is inhibited in the presence of caspase inhibitors (zVAD-fmk or P35) or in a dose-dependent manner by Shh (30, 90, or 300 ng/ml). Incubation with a Shh-blocking antibody to Shh-treated Ptc-1–transfected cells led to a cell death induction similar to that seen in the Ptc-1–transfected population. Values are the mean ± SD (n = 3). (C) Expression of Ptc-1 receptor at the cell membrane 48 hours after transfection of 293T cells monitored by flow cytometry. (D) Ptc-1–induced increased caspase activity monitored with the fluorometric substrate Nacetyl-Asp-Glu-Val-Asp-7-amino-4-trifluoromethyl coumarin (n = 3). (E) Ptc-1–induced cell death, monitored by flow cytometry with propidium iodide incorporation. An index of 13.S.24 cells with low DNA content (the sub-G1 phase) is indicated as the mean ± SD (n = 3). (F) Apoptosis was also quantified by TUNEL assay in 13.S.24 cells transfected with Ptc-1. The relative index of TUNEL-positive cells is shown as the mean ± SD (n = 3). Gli activity was measured in the different transfected conditions by means of a Gli–binding site–Luceriferase reporter assay. Although Ptc expression has no effect on Gli activity when expressed alone, Ptc inhibited Smo-induced Gli activation (26).

We then overexpressed Ptc-1 in the neural tube of chick embryos in ovo by electroporating the Ptc-1–expressing construct at the level of the posterior neural plate of E1.5 (embryonic day 1.5) chick embryos. One day later, mouse Ptc-1 was upregulated in the electroporated side of the neural tube (Fig. 2A). An increase of terminal deoxynucleotidyl transferase–mediated deoxyuridine triphosphate nick end labeling (TUNEL) staining was associated with Ptc-1 overexpression; no such increase was observed in the nonelectroporated side of the neural tube (Fig. 2B) or when the electroporation was performed with a mock construct (fig. S2). Moreover, the ventral neuroepithelial cells that were exposed to higher Shh concentration because of the Shh ventrodorsal gradient appeared less susceptible to apoptosis than lateroventral and laterodorsal neuroepithelial cells (Fig. 2, A and B). This observation supports the above cell culture results suggesting that Shh affects Ptc-induced cell death.

Fig. 2.

Ptc-1 overexpression induces apoptosis in the neural tube of chick embryo in vivo. (A) In situ hybridization showing that mouse Ptc-1 is expressed in the neuroepithelium after electroporation. (B) TUNEL staining showing death increase and smaller size in the Ptc-1 up-regulated side (arrowhead). Arrows indicate normal apoptosis in the nonelectroporated side. Neural tube is divided into four quarters (a, b, c, and d). The average number of TUNEL-positive cells per section (20 representative sections were counted) in each quarter is indicated. Scale bar, 55 μm.

To elucidate the molecular mechanisms of Ptc-induced cell death, we further analyzed the involvement of caspases. The dependence receptors DCC, UNC5H, and RET have been shown to require a preliminary caspase cleavage to induce cell death (17, 18, 24). We therefore analyzed the ability of the intracellular domain of Ptc to be cleaved by caspases. The largest intracellular region of Ptc lies in the last 269 amino acids in the C-terminus [i.e., the seventh intracellular domain of Ptc (Ptc-7IC)]. The transfection of this domain in 293T cells was sufficient to induce cell death (Fig. 3D). In vitro translation of this domain was then performed and incubated in the presence of purified active caspase-3, 6, 7, or 8. Figure 3A shows that although caspase-6 has no effect on Ptc-7IC, a single cleavage appears with caspase-3 and less frequently with caspase-7 and 8. Hence, Ptc-7IC is cleaved in vitro by caspases, most effectively by caspase-3. The caspase cleavage site was mapped by constructing mutants based on preferred P4 and P1′ positions (25). Although a putative caspase-3 consensus DXXD site (D, Asp; X, any amino acid) was found in Ptc-7IC (Fig. 3B), its mutation had no effect on caspase-3 cleavage (Fig. 3, B and C). However, the mutation of Asp1392 to Asn (Fig. 3, B and C) suppressed caspase-3 cleavage of Ptc-7IC, indicating that the caspase cleavage site of Ptc is at Asp1392. This aspartic residue appears to be conserved in human, mouse, and chicken Ptc.

Fig. 3.

Ptc-1 is a caspase substrate. (A) In vitro–translated Ptc-7IC was incubated in the absence of caspase (0 μM) or with purified caspase-3 (0.3 μM), caspase-6 (0.4 μM), caspase-7 (1.6 μM), or caspase-8 (0.3 μM). An autoradiograph is shown. (B) Map of Ptc-7IC and the different mutant constructs. (C) Asp1392 is the major cleavage site. (D) Cell death after transfection with Ptc-1 mutants (Ptc-7IC, Ptc D1392N, Ptc 1-1392, or Ptc 1165-1392). Cell death was monitored by trypan blue exclusion. Both Ptc and Ptc-7IC truncated after the caspase site (i.e., amino acid 1392) induced constitutive cell death, regardless of Shh presence. Error bars indicate SDs (n = 3). (E) Schematic representation of the initial step of Ptc-induced apoptosis triggered by the absence of Shh ligand. Gli activity was measured in the different transfected conditions used in (D). Neither of the Ptc mutants (7ICor full-length) had any effect on Gli activity (26).

To evaluate the functional importance of the cleavage of Ptc by caspase, we expressed the full-length Ptc D1392N or Ptc-7IC D1392N mutant in 293T cells. D1392N point mutation inhibited Ptc proapoptotic activity (Fig. 3D and fig. S3) (26). Similarly, electroporation of Ptc-7IC D1392N at the level of the posterior neural plate of E1.5 chick embryos did not induce apoptosis in the neuroepithelium (26). Moreover, expression of Ptc truncated at the caspase cleavage site (Ptc 1-1392) led to 293T cell death even when Shh was added (Fig. 3D). Taken together, these results indicate that the observed in vitro caspase cleavage has functional relevance and probably represents an initiating step for Ptc-induced apoptosis (Fig. 3E) by allowing the release of the C-terminal region of Ptc and the consequent exposure of a proapoptotic domain located upstream of the caspase site.

The dependence receptor activity of Ptc, DCC, and UNC5H might account for the putative tumor-suppressor activity that they share. DCC and UNC5H may act as tumor suppressors by limiting tumor cell growth through induction of apoptosis in settings in which the ligand is not available (17, 20). Although Ptc is frequently mutated in cancers (16), it is still unknown how Ptc inhibits the hallmarks of cell transformation in vitro (27). We then assessed whether in vitro, Ptc expression inhibits anchorage-independent growth through the control of apoptosis. Ptc expression inhibits growth in soft agar of 293T cells, an effect blocked by Shh (Fig. 4). The expression of Ptc D1392N mutant that is not proapoptotic failed to inhibit growth in soft agar (Fig. 4). Similar results were obtained when growth in soft agar was monitored in cells expressing Ptc in the presence of zVAD-fmk (26). To rule out a cytostatic effect of Ptc that has recently been proposed (28), we measured growth in soft agar when Shh was added only after 48 hours of Ptc expression. The addition of Shh did not lead to cell growth rescue and colony formation (Fig. 4). We finally observed that in the soft-agar growth condition, Ptc expression led to cell death induction (fig. S4). Taken together, these data suggest that Ptc may act as a tumor suppressor by regulating apoptosis.

Fig. 4.

Ptc expression inhibits growth in soft-agar assays through apoptosis induction. 293T cells were transiently transfected in the presence or absence of Ptc-1 (wild-type or mutant) with or without Shh. Shh was either added simultaneously to the plating in soft agar or added 48 hours later. Cells were allowed to grow for 12 days in soft agar. (A) A set of representative plates is shown. (B) The colony number was determined as the number of isolated clones with a diameter larger than 100 μm. Error bars indicate SDs (n = 3).

To further analyze whether the control of Ptc-induced cell death by Shh may be related to the survival effect of Shh observed in vivo (8), we used a Ptc mutant that acts as a dominant negative mutant for Ptc-induced cell death. Indeed, Ptc-7IC D1392N (Ptc mut.) inhibited Ptc-induced death when coexpressed in 293T or 13.S.24 cells (fig. S5). In the chick embryo, lack of No and FP due to the inhibition of Hensen's node caudal regression drives induction of cell death in the dorsal tissues, neural tube, and somites, which often results in the truncation of the embryos (Fig. 5, A to C) (7). However, grafting Shh-expressing cells in the region deprived of midline cells is sufficient to block this process (8). We then assessed the putative effect of the dominant negative mutant for Ptc-induced cell death in chick embryos experimentally deprived of Shh. Electroporation of the dominant-negative mutant caudally to the site of arrest of Hensen's node allows the detection of a neural tube at the E3.5 stage (Fig. 5, D to H). The latter is strongly reduced or completely absent in control or mock-electroporated operated embryos (Fig. 5C) (7). Moreover, there is a correlation between the absence of cell death and the effective transfection of the Ptc mut. construct within the neuroepithelium (Fig. 5, E to H). Ptc mut. also allows the posterior region of the embryo to partly develop (Fig. 5D), as compared with nonelectroporated embryos (7) or embryos electroporated with a mock plasmid (Fig. 5C). Similar results were obtained by electroporation of the general caspase inhibitor P35 (26) or treatment with the zVAD-fmk (fig. S6). Hence, the control by Shh of Ptc-mediated cell death represents a crucial event for neuroepithelial cell survival and neural tube development.

Fig. 5.

The antiapoptotic role of Shh in vivo is mediated by Ptc-1. [(A) to (C)] Node arrest and absence of No and FP, provoked by insertion of a fragment of tantalium between the node and the primitive streak at E1.5 in the chick embryo in ovo, leads to caudal destruction of the embryo, as in (7). (A) TUNEL staining on the transverse section of the E2.5 control embryo. (B) Whole E3.5 embryo electroporated with a mock construct showing the caudal destruction. (C) TUNEL staining on transverse section of (B) showing the absence of the neural tube. (D to H) Overexpression of Ptc mut. decreases apoptosis, induced by No and FP absence. Ptc mut. acts in a dose-dependent manner as seen by TUNEL staining [(F) to (H)] and green fluorescent protein expression (E). Similar results were obtained by administration of the caspase inhibitor zVAD-fmk or of a blocking Ptc antibody (fig. S6). In (B) and (D), the arrowhead shows the position of the inserted fragment of tantalium and indicates the site of node arrest. Ec, ectoderm; So, somite; En, endoderm; Lb, limb bud. Scale bars: (A), 70 μm; (B) and (D), 4.3 mm; (C), 90 μm; (E), 1.4 mm; (F), (G), and (H), 60 μm.

Thus, a signaling pathway is generated by Ptc that leads to apoptosis for the cell expressing Ptc in the absence of Shh. The trio of Shh, Ptc, and Smo then suggests a very subtle balance between the differentiating–life-sustaining signal mediated by Smo when Shh binds Ptc and the death-inducing signal derived from Ptc in the absence of Shh. The positive signal has a crucial impact in determining cell fate (1, 5). In addition, the positive signal mediated either by Shh or other Hedgehog proteins may regulate the negative proapoptotic signal of Ptc. For example, Gli-3, involved in Shh-Ptc-Smo signaling, functions in apoptosis regulation (29) and interferes with the cell death induction observed in Shh mutant mice (9). However, the dogma proposes that cell death induction may only be the result of an absence of the proper signal for cell differentiation. This view would be difficult to reconcile with the observation that the developing neural tube of the Ptc–/– mouse embryo does not suffer cell deficits, but rather is overgrown, as expected for an absence of Ptc-induced cell death (30). Moreover, in chick embryos experimentally deprived of Shh-producing midline cells (No + FP), inhibition of cell death by transfection with the dominant-negative mutant for Ptc-induced cell death not only suppresses cell death but appears to partly allow spinal cord development (Fig. 5, F to H, and fig. S6). Thus, the control of cell death by Shh may also be an important part of the Shh role during central nervous system development. The Ptc-mediated death observed in the absence of Shh would then appear to be not just a consequence of a lack of cell differentiation but an active process contributing to spinal cord development.

Supporting Online Material

Materials and Methods

Figs. S1 to S6

References and Notes

Stay Connected to Science

Navigate This Article