Blocking SHH/Patched Interaction Triggers Tumor Growth Inhibition through Patched-Induced Apoptosis

The morphogen Sonic Hedgehog (SHH) regulates many developmental processes including ventrodorsal patterning of the neural tube, establishment of limb polarity, and development of the foregut and axio-cranial skeleton (1, 2). SHH signaling is mainly mediated via the binding of SHH to the 12-transmembrane receptor patched 1 (PTCH), which relieves PTCH suppressive effect towards Smoothened (SMO), an orphan 7-transmembrane receptor, which initiates a signaling cascade leading to the activation of the glioma-associated (GLI) transcription factors.

In adults, SHH signaling is mainly quiescent, being physiologically reactivated only during tissue maintenance and repair (3). However, during tumor formation and progression, deregulation of SHH signaling has been reported to often occur and be detrimental (for review, refs. 4–8). Indeed, many different types of human cancers display abnormal activation of SHH signaling (4, 9–11), promoting this pathway as a choice target for the development of new therapeutics. So far, two types of strategies to antagonize the “canonical” SHH–PTCH–SMO–GLI signaling pathway have been worked out. The first strategy was to design SMO- or GLI-antagonists that target the active signaling pathway. In clinical trials, even though some resistance and important side effects were reported, some of these hedgehog pathway inhibitors, such as the SMO antagonist vismodegib or sonidegib, showed good antitumor activity in tumors bearing genetic alterations of the "canonical pathway," such as basal cell carcinoma (BCC) and medulloblastoma (for review, ref. 12). However, in tumors characterized by an upregulation of SHH clinical trials were far less efficient. The second strategy was aimed to target the upstream initiation of the SHH pathway. Indeed, a large fraction of human cancer displays SHH upregulation (4, 11), which can be neutralized either by targeting SHH production—for example, oligonucleotides targeting SHH gene- or by blocking SHH interaction with PTCH—for example, blocking mAb (13, 14). Despite several reports describing the inhibition of tumor growth and the induction of tumor cell apoptosis in various animal models using mAbs targeting SHH (15, 16), this approach never reached patient clinical trial level.

These therapeutic strategies were all designed according to the classic view that PTCH is a basic receptor that is solely active when bound to SHH. However, we recently demonstrated that PTCH has the fascinating property to also be able to trigger cell apoptosis in the absence of its ligand. PTCH is thus a dependence receptor (17). More recently, another SHH receptor, cell-adhesion molecule-related/downregulated by oncogenes (CDON) has also been described as a dependence receptor able to trigger cell death in absence of SHH (18). Such receptors are two-sided receptors: although they induce a positive signal (promoting cell survival, migration, proliferation) in the presence of their ligand, they can induce an active process of apoptotic cell death in the absence of ligand. These dependence receptors include p75^ntr, deleted in colorectal cancer (DCC), UNC5H, Neogenin, Notch3, and the RET, EPHA4, ALK, and c-KIT tyrosine kinase receptors (19–24). The proapoptotic activity of dependence receptors is believed to be important not only during embryonic development (17, 20, 25–27) but also for inhibiting tumor progression. Indeed, the dependence on ligand presence is also thought to act as a safeguard mechanism to prevent tumor cells from developing in settings of ligand unavailability (for reviews, refs. 28, 29). Thus, a tumor cell losing the proapoptotic activity of a dependence receptor would gain a selective advantage for growth. Along this line, we showed that the inactivation of DCC proapoptotic signaling in mice is associated with colorectal tumor progression (30). One way to silence this death pathway is to upregulate the ligand level in the tumor environment—that is, autocrine or paracrine expression (31–34). Such an upregulation of SHH is known to occur in a large fraction of human cancers (11, 13, 14). Moreover, as CDON is frequently lost in several cancer types (18), we thus hypothesized that the antitumor effect observed upon SHH interference is not only due to the inactivation of the SHH signaling canonical pathway but also to the activation of PTCH proapoptotic activity involving the recruitment of the caspase activating complex dependosome (35).

In this study, we report that SHH-overexpressing tumors express PTCH-induced cell death effectors, suggesting that this signaling could be activated as an antitumor strategy. Moreover, we demonstrated in a colon cell line model that blocking SHH triggers PTCH-induced apoptosis through the recruitment of the dependosome complex. Finally, we provide in vivo evidences that the strategy based on interference with SHH/PTCH interaction limits tumor growth through PTCH-induced cell death. Thus, we propose to consider an alternative therapeutic approach where, in SHH-overexpressing resistant tumors, a SHH blocking antibody could show therapeutic benefit.

The Cancer Genome Atlas (TCGA) data were downloaded, assembled, and processed using TCGA2STAT (36). Subsequent analyses were performed with the R statistical software v 3.3.1 and GraphPad Prism software.

All cell lines were obtained from ATCC. The human colon cancer cell lines HCT8 and SW1417 were cultured in RPMI1640 Glutamax medium (Gibco; Life Technologies) containing respectively 10% horse serum or 10% FBS. The human colon cell line LoVo, the pancreatic cell line Mia PaCa-2, and the non–small cell lung cancer (NSCLC) cell line A549 were cultured in DMEM Glutamax medium (Gibco; Life Technologies) containing 10% FBS. The human pancreatic cell line Capan-2 was cultured in RPMI1640 Glutamax medium (Gibco; Life Technologies) containing 15% FBS. The mouse fibroblast cell line NIH3T3 was cultured in DMEM Glutamax medium (Gibco; Life Technologies) containing 10% FBS. All media were supplemented with 1% peni-streptomycin and 0.4% fungizone. Routine Mycoplasma testing was performed by MycoAlert Mycoplasma Detection Kit (catalog no. LT07-118). Cells grown for no more than 20 passages were used in all the experiments. Cell lines were transfected using Lipofectamine Plus Reagent (Life Technologies) for plasmids or Lipofectamine 2000 reagent or RNAi Max (Life Technologies) for siRNA.

The pRK5-Ptc, pRK5-PtcDN-HA (Ptc7IC D1392N), pcDNA3-wild-type caspase-9-3XFLAG, pcDNA3-DRAL-3xFLAGM2 were described earlier (17, 35). Stable HCT8 and LoVo cell line expressing Ptc dominant negative (PtcDN) was established using peak8 vector (Clontech) and selected with puromycin (1 μg/mL; Sigma). More precisely, peak8-PtcDN plasmid was obtained by inserting a HindIII/EcoRV fragment generated by polymerase chain reaction performed on the already described pcDNA3-PtcDN (17). HCT8 and LoVo cells were then transfected with an empty plasmid (peak8 vector) or peak8-PtcDN and was treated with puromycin 48 hours after transfection. SHH and control siRNA were designed by Santa Cruz as a pool of three target-specific siRNAs of 20 to 25 nucleotides. SHH-neutralizing 5E1 mAb was purified from ascites on protein G sepharose columns (Biotem). Centricon (Millipore) were used for antibody concentration and buffer exchange. Murine IgG1 isotype control was purchased from R&D Systems. HL2m5 macrocyclic peptide inhibitor were described in ref. 37.

Human pancreatic, colon, and Lung cancer tissues microarray were obtained from Biochain (Z7020110, T8235722, P3235152). After deparaffinization and dehydration, 4-μm-thick tissue sections were heated for 50 minutes at 97°C in citrate buffer pH 9. To block endogenous peroxidases, tissue sections were incubated in 5% hydrogen peroxide solution. IHC was performed on an automated immunostainer (Ventana Discovery XT; Roche) using Omnimap DAB Kit according to the manufacturer's instructions. Sections were incubated with an anti-SHH monoclonal rabbit antibody (Abcam; diluted at 1:10,000), an anti-PTCH polyclonal rabbit (Novus Bio; diluted at 1:500), an anti-DRAL polyclonal rabbit (Abcam; diluted at 1:200), and an anti-caspase-9 monoclonal mouse (Abcam; diluted at 1:200) were used. Anti-rabbit or Mouse-HRP were applied on sections. Staining was visualized with DAB solution with 3,3′-diaminobenzidine as a chromogenic substrate. Finally, the sections were counterstained with Gill's hematoxylin then were scanned with panoramic scan II (3D Histech) at 1× or 2×.

Immunoblots were performed as described in ref. 35, using anti-SHH (Abcam), anti-FHL2/DRAL (Abcam), anti-caspase-9 (Cell Signaling Technology), anti–PTCH (Abcam), anti-PTCH C-ter (to detect PtcDN; generously given by M. Ruat and Novus Bio), anti-CDO (Santa Cruz Biotechnology), anti-β-actin (Millipore), anti-FLAG (Sigma), and anti-HA (Sigma) antibodies.

To assay SHH, PTCH, and CDON expression, total RNA was extracted from cell lines using the NucleoSpin RNAII Kit (Macherey Nagel) and 1 μg was reverse transcribed using the iScript cDNA Synthesis Kit (Bio-Rad Laboratories). Real-time quantitative reverse transcription-PCR (qRT-PCR) was performed as described previously (38). The ubiquitously expressed human ß-actin was used as an internal control.

For caspase-3 assays, 1.5 × 10⁵ cells were cultured in serum-free medium and were treated (or not) with SHH-neutralizing 5E1 antibody (5 μg/mL) for 24 hours or transfected with 10 nmol/L Control, SHH, PTCH, or CDON siRNA (Santa Cruz Biotechnology) with Lipofectamine 2000 or RNAi Max (Life Technology) for 48 hours. Apoptosis was monitored by measuring caspase-3 activity using the caspase-3/CPP32 Fluorometric Assay Kit (Gentaure Biovision) as described previously (38).

For propidium iodide (PI) incorporation, 5 × 10⁴ cells were cultured without serum in the presence of 0.3 μg/mL of PI and were transfected with 10 nmol/L Control, SHH, PTCH, or CDON siRNA (Santa Cruz Biotechnology) with Lipofectamine 2000 (Invitrogen). Live counting of PI positive cells was done for 72 hours with the Incucyte Zoom Live cells analysis system (Essen Bioscience) and indexed on percentage of cell confluency.

Colon cancer HCT8 cells (10⁷ cells) in 20 μL of PBS were mixed with 20 μL of Matrigel (BD Biosciences) to form a gel and implanted on top of the CAM of 10-day-old chick embryos. Ten microliters of the control IgG or the 5E1 antibodies were injected in the tumor on days 11 and 13. On day 17, tumors were resected. Primary tumor size and metastasis in lungs were determined as previously described (32). Briefly, on day 17, tumor areas were measured with AxioVision Release 4.6 software (Carl Zeiss, Inc.). To assess metastasis, lungs were harvested from the tumor-bearing embryos and genomic DNA was extracted with a NucleoSpin Tissue Kit (Macherey-Nagel). Metastasis was quantified by PCR-based detection of the human Alu sequence (32) using the primers 5′-ACGCCTGTAATCCCAGCACTT-3′ (sense) and 5′-TCGCCCAGGCTGGAGTGCA-3′ (antisense) with chick Gadph-specific primers (accession number: M11213.1; sense, 5′-GAGGAAAGGTCGCCTGGTGGATCG-3′; antisense, 5′-GGTGAGGACAAGCAGTGAGGAACG-3′) as controls. For both couples of primers, PCR was performed at 95°C for 2 minutes, followed by 30 cycles at 95°C for 30 seconds, 63°C for 30 seconds, and 72°C for 30 seconds. Genomic DNA extracted from lungs of healthy chick embryos was used to determine the threshold between colon cell–invaded and –noninvaded lungs. To monitor apoptosis in primary tumors, primary tumors and surrounding CAM were resected, and caspase-3 activity was measured on tumor protein lysates using the caspase-3/CPP32 Fluorimetric Assay Kit.

Prior to injection, HCT8 peak and PtcDN cells were treated for 48 h with IgG isotype control or 5E1 antibodies. Then, 9 × 10⁵ cells were resuspended in serum-free medium and stained with lipophilic dyes DiO or DiD for 20 minutes at 37°C (Thermo Fisher Scientific). Cells were then washed and resuspended in 30 μL of PBS. For zebrafish xenotransplantation, 48 hours post-fecundation (hpf) zebrafish embryos were anaesthetized with tricaine (Sigma-Aldrich) and 20 nL of cell suspension (approximately 300 labeled human cells) were injected into the perivitelline cavity of each embryo. The embryos were then placed at 30°C for 24 hours and allowed to recover in the presence of N-phenylthiourea (Sigma-Aldrich) to inhibit melanocyte formation. For imaging and metastasis assessment, zebrafish embryos were anaesthetized with tricaine and imaged using an Axio Observer Zeiss microscope (Zeiss).

Five-week-old female Swiss nu/nu mice (20–22 g) were obtained from Charles River. Animals were maintained in the specific pathogen-free animal facility (AniCan) at the Cancer Research Center of Lyon at the Center Léon Bérard. All the experiments were performed in accordance with the animal care guidelines of the European Union, and were validated by the local Animal Ethic Evaluation Committee (CECCAPP: C2EA-15 agreed by the French Ministry of High School and Research). HCT8 Control or PtcDN HCT8 or LoVo cells were implanted by subcutaneous injection of 2 × 10⁶ or 4 × 10⁶ cells, respectively, in 50 μL of PBS mixed with 50 μL of Matrigel into the right flank of the mice. When tumors reached a volume of approximately 70 mm³ (around 10 days after injection), 300 μg of SHH neutralizing 5E1 antibody or an equal volume of buffer was injected three times per week for maximum 23 days. In other experiments, 3 μg of siRNA targeting SHH or control siRNA was injected intraperitoneally three times per week for 9 days. Tumor sizes were measured with a caliper. The tumor volume was calculated with the formula v = 0.5 (l × w × w), where v is the volume, l is the length, and w is the width. To monitor apoptosis, tumors were resected and broken up in a lysis buffer from the caspase-3/CPP32 Fluorimetric Assay Kit and caspase-3 activity was measured in tumor protein lysates.

The statistical significance of differences between groups was evaluated by the paired Student t test or the Mann–Whitney test, two-way ANOVA test for tumor volume mice comparison, and Chi-squared test for the comparison of lung metastasis percentage in chick embryo models.

Mean values for all outcome variables are presented with 95% confidence intervals. Data presented are representative of at least three independent experiments. All statistical tests were two-sided, and results were considered significant at P < 0.05.

Previous studies have reported that a high proportion of colon and pancreatic cancers overexpress SHH (8, 39). This was confirmed by analyzing TCGA relative SHH expression in a large panel of tumor samples (colon cancer, n = 283; pancreatic cancer, n = 178; and lung cancer, n = 1016). We found respectively that 80%, 64%, and 8% of colon, pancreatic, and lung cancers overexpress SHH when compared with normal tissues (two times more than the average SHH mRNA level in normal tissues; Fig. 1A and B). Of note, although SHH is not correlated to patient survival rate in colon or in pancreatic cancers (Supplementary Figs. S1A and S1B), it is interesting to point out that in pancreatic cancer, SHH expression is significatively increased according to the advanced stages (Supplementary Figs. S1C and S1D). The overexpression of SHH at transcriptional level is corroborated by SHH IHC analyses on colon, pancreatic, and lung tumoral tissues, showing especially a strong epithelial staining of SHH in the 76.7% and 57.3% of colon and pancreatic tumor samples, respectively (60 colon and 89 pancreatic tumor samples were analyzed; Fig. 1C; Supplementary Figs. S1E and S1F). Moreover, most of colon and pancreatic tumors showing overexpression of SHH show a very high score of SHH expression (scores 2 and 3; Supplementary Fig. S1G).

Further analysis indicated that SHH-overexpressing tumors also highly express PTCH as well as PTCH-mediated death effectors— that is, the caspase-activating complex leading to PTCH-induced cell death (35). Indeed, as seen in Fig. 1D, the genes encoding PTCH1, caspase-9, FHL2, CARD8/NLRP1, and NEDD4 were notably expressed in SHH-overexpressing tumors, which is confirmed at protein level by IHC on human colon and pancreatic cancer tissues for PTCH protein (Fig. 1C) and for FHL2 and caspase-9 (Supplementary Figs. S1H and S1I), suggesting that they bear a putative functional PTCH-proapoptotic signaling.

We then investigated whether in SHH-expressing tumor cells, SHH constitutively inhibits PTCH-induced apoptosis. To do so, we selected representative human tumor cell lines of colon cancers (HCT8, SW1417, LoVo), pancreatic cancers (Mia PaCa-2, Capan-2), and NSCLC (A549) based on their expression of SHH, PTCH, and the main PTCH-mediated known death effectors, that are FHL-2 and caspase-9 (35). As seen in Fig. 2A and Supplementary Fig. S2A), the levels of PTCH and caspase-9 expression analyzed by immunoblot were similar in all selected cell lines, whereas SHH and FHL2 expression level was variable.

We then analyzed cell death in response to the disruption of SHH autocrine loop. As a first approach, SHH was downregulated by RNA interference. As shown in Fig. 2B and Supplementary Fig. S2B, the transient transfection of SHH siRNA was associated with a significant reduction in SHH mRNA in all cell lines studied. The downregulation was confirmed at protein level in HCT8 cells (Supplementary Fig. S2C). This reduction was accompanied with an activation of caspase-3 in all cell lines (Fig. 2C; Supplementary Fig. S2D), used as a read-out for dependence receptor-induced cell death assessment. Of high interest, we were also able to show increase cell death in HCT8 cells upon SHH interference using a macrocyclic peptide HL2m5 (37) able to abolish SHH/PTCH interaction (Supplementary Fig. S2E). To decipher the mechanisms involved in this response, we then focused on the colorectal cancer HCT8 cells then we confirmed the results in colorectal cancer LoVo cells. We first confirmed in HCT8 cells the causal relationship between SHH silencing effect and caspase-3 activation: treatment with exogenous recombinant SHH (rSHH) prevented caspase-3 activation in SHH silenced cells (Fig. 2D). Caspase-3 activation was associated with increased PI staining supporting the view that caspase activation lead to cell death (Fig. 2E; Supplementary Figs. S3A and S3B). Moreover, the siRNA-mediated downregulation of PTCH (Fig. 2F) fully inhibited SHH siRNA-mediated cell death (Fig. 2E; Supplementary Figs. S3A and S3B), hence demonstrating that the cell death observed upon SHHsilencing is mediated by PTCH. Given that we previously reported that CDON receptor also triggers apoptosis when unbound to its ligand SHH (18), we analyzed the level of CDON expression by immunoblot in HCT8 cell line. Although the RH30 rhabdomyosarcoma cell line used as a positive control expressed a high amount of CDON protein, we failed to detect CDON expression in HCT8 (Supplementary Fig. S3C). Given that CDON expression was detected at mRNA level, we performed cell death assay upon CDON silencing by siRNA in HCT8 (Supplementary Fig. S3D). Downregulation of CDON failed to inhibit SHH silencing-induced cell death (Supplementary Figs. S3A, S3E, and S3F). Taken together, these results suggest that the cell death induced by SHH downregulation in HCT8 cells is specifically mediated by PTCH receptor.

To confirm that the death induction described above results from PTCH proapoptotic activity, we generated HCT8 and LoVo stably expressing a mutant of PTCH—that is, Ptc-7IC-D1392N (PtcDN) that was reported to act as a specific dominant negative mutant for PTCH proapoptotic activity (Fig. 3A; ref. 17). Although this mutant expression had no effect on the positive signaling PTCH-SMO-GLI in HCT8 or 3T3 cells (Supplementary Fig. S4A and S4B; refs. 17, 18, 35), we observed, using coimmunoprecipitation followed by immunoblot, that PtcDN blocked PTCH/FHL-2 (Fig. 3B) and PTCH/caspase-9 (Fig. 3C) interactions. Moreover, we showed that the molecular complex containing PTCH and caspase-9 was modulated by the availability of SHH because the presence of rSHH induced the disruption of the PTCH-caspase-9 complex whereas this interaction was restored by the concomitant addition of the 5E1 SHH-blocking antibody (Fig. 3D). Thus, PtcDN acts as an inhibitor of endogenous PTCH proapoptotic signaling by preventing PTCH dependosome complex formation (35). Reciprocally, a possible effect of endogenous PTCH on PtcDN is not excluded but unknown so far. We then analyzed the effect of PtcDN on PTCH-induced cell death triggered by SHH interference in HCT8 and LoVo cells. As expected, these cells expressing PtcDN (Fig. 3E) failed to undergo apoptosis in comparison to control cells (peak-transfected cells) upon silencing of SHH by siRNA (Fig. 3F; Supplementary Fig. S2D), or upon treatment with the 5E1 antibody (Fig. 3G), both treatments having no effect on the positive signaling PTCH-SMO-GLI in HCT8 nor in other cell lines used in this study (Supplementary Figs. S4C and S4D). Together these results indicate that SHH expression in cancer cells inhibits PTCH proapoptotic activity and is likely implicated in the maintenance of HCT8 and LoVo cells survival.

We next challenged the hypothesis that interfering with SHH autocrine loop could trigger in vivo tumor growth inhibition. First, we used an avian model that recapitulates tumor progression and dissemination. Grafts of tumor cells in the chorioallantoic membrane (CAM) of 10-day-old chick embryos induce the growth of a primary tumor at the implantation site—within the CAM—as well as metastasis dissemination at distant sites—for example, in the lung (Fig. 4A; ref. 40). CAMs in 10-day-old fertilized chicken eggs were inoculated either with mock transfected HCT8 (HCT8 peak) or HCT8 expressing PtcDN (HCT8 PtcDN) cells and were treated on day 11 and day 14 with either isotype control IgG or 5E1 blocking antibody. Three days later, primary tumor volumes were measured and the lungs were removed from the embryos and analyzed for the presence of metastasis. As shown in Fig. 4B to D, treatment of CAM-grafted HCT8 cells with the 5E1 antibody reduced significantly the size of primary tumors and decreased the incidence of lung metastasis development. In contrast, the 5E1 antibody had no effect on either primary tumor growth nor metastasis dissemination in eggs grafted with HCT8 PtcDN cells (Fig. 4B–D).

To confirm these results in another in vivo metastatic model, we used a zebrafish model (41). We injected in the perivitelline space of 2-day-old zebrafish larvae HCT8 peak or PtcDN cells previously treated with isotype control IgG or 5E1 blocking antibody for 2 days and then stained with lyophilic DiO (green) and DiD dye (red), respectively. Forty-eight hours after injection fishes were imaged and the number of metastatic cells in the tail was determined. As shown in Fig. 4E and F, treatment with the 5E1 antibody decreases significantly the number of metastatic HCT8 peak cells. In contrast, the 5E1 antibody had no effect on metastasis dissemination in fishes injected with HCT8 PtcDN cells (Fig. 4E and F) supporting the view that interference with SHH decreases HCT8 cell dissemination mediated by PTCH.

To further define whether SHH interference is associated with an antitumor effect in vivo, we generated xenografts of SHH expressing tumor cells in nude mice. We thus subcutaneously grafted HCT8 peak or HCT8 PtcDN cells in nude mice. When the tumors reached 70 mm³, mice were treated intraperitonealy three times a week, with either SHH siRNA or with scrambled control siRNA as it has been successfully performed previously (32, 38). We first showed that HCT8 tumors express SHH at transcriptional (Supplementary Fig. S5A) and protein level (Supplementary Fig. S5B) and, as shown in Supplementary Fig. S5C, treatment with SHH siRNA reduced significantly the growth of HCT8 peak tumors. In contrast, the siRNA had no effect on the growth of HCT8 PtcDN tumors (Supplementary Fig. S5D). Alternative treatments were performed using the 5E1 SHH-blocking antibody. As shown in Fig. 5A and B, 5E1 antibody treatment, which does not demonstrate apparent animal toxicity as attested by absence of weight variation during treatment (Supplementary Figs. S5E and S5F), significantly reduced the growth of HCT8 peak tumors (Fig. 5A and B), a phenomenon associated with a significant reduction of the net tumor weight at the end of treatment protocol. In contrast, the growth of HCT8 PtcDN tumors (tumor volume as well as net tumor weight) was not affected by the treatment with the 5E1 antibody (Fig. 5A, C, and D). Similar results were obtained with LoVo-engrafted tumors (Supplementary Figs. S6A–S6C) confirming these results in another colon cell line. Of interest, 5E1 treatment triggered significant apoptosis—as measured by increased caspase-3 activity—in HCT8 peak but not in HCT8 PtcDN tumors (Fig. 5E). Taken together these data support the view that SHH expression in cancer cells is associated with tumor growth and metastasis at least in part by inhibiting PTCH-induced cell death.

Here we have shown that colon, pancreatic, and lung cancers that share the property to overexpress SHH also expresses PTCH-proapoptotic signaling effectors. Of prime importance, we also observed that disrupting the SHH autocrine loop, either by downregulating SHH expression or by blocking the interaction of SHH with PTCH, leads to tumor cell death in vitro and to a reduction of tumor growth and metastasis dissemination in vivo. These observations are in perfect agreement with the fact that PTCH belongs to the family of dependence receptors (17, 42) that have been proposed to act as a safeguard mechanism against tumor progression (28, 43, 44). Reciprocally, the overexpression of dependence receptor ligand is also known to confer a selective advantage on tumor cell survival by blocking dependence receptor-mediated cell death. Along this line, it has been shown that netrin-1, the ligand of DCC/UNC5 family, is a promising therapeutic target in netrin-1–overexpressing tumors (32, 38) leading to ongoing clinical trial investigating whether inhibition of netrin-1 activity could be effective in solid human tumors (https://clinicaltrials.gov/ct2/show/NCT02977195). Similarly, NT3 the ligand of TrkC dependence receptor is also promoting tumor cell survival in neuroblastoma (33) as well as Sema3E, which blocks the cell death induced by Plexin D1 in breast cancer (45).

Given that a high proportion of cancer pathologies show aberrant Hedgehog signaling, many antagonist molecules of this signaling pathway with putative therapeutic activity have been developed these last 15 years. Most of these drugs target the canonical pathway—that is, PTCH-SMO-GLI components of the pathway. In that regard, the SMO antagonists vismodegib and sonidegib has demonstrated antitumor effects in BCC and medulloblastoma bearing activated somatic mutations (46, 47) and is approved for treating BCC. Nevertheless, vismodegib failed to show any beneficial effect in phase II clinical trial including patients with tumors with HH upregulation such as colorectal, ovarian cancer, and chondrosarcoma. The study we present here allows to propose that the lack of clinical effect of these “canonical pathway” inhibitors is probably a consequence of their inability to promote PTCH-induced apoptosis. Such apoptotic stimulation would be highly interesting to be tested in colon and pancreatic cancers in which respectively 80% and 64% of the tumors overexpress SHH. In the case of lung tumors, the situation is a little bit more uncertain because in spite of expressing all the effectors of PTCH-induced cell death pathway only 8% of them overexpress SHH. Nevertheless, HH pathway has been proposed to be involved in resistance to radiotherapy and to immune checkpoint inhibitors and relapse in lung cancer (48), even though resistance mechanisms are still poorly understood. Hence, future work will have to investigate whether targeting SHH/PTCH interaction in these tumors may turn as a promising therapeutic strategy.

Concerning CDON, another SHH receptor previously described as a dependence receptor frequently lost in cancers (18), it is expressed, at least at the mRNA level, in SHH-expressing colon, pancreatic, and NSCLCs (18). Similarly, in the selected HCT8 cell line used in this study, CDON expression was detected at mRNA level but we failed to detect its expression at the protein level. Moreover, we did not observe a putative involvement of CDON-proapoptotic signaling upon SHH withdrawal.

Future studies aimed at defining therapeutic approaches will probably be based on antibodies that block the interaction between SHH and either PTCH and CDON, unleashing PTCH and/or CDON-induced cell death. Interestingly, based on the predicted interacting structures of SHH with PTCH or CDON (49), SHH antibodies such as 5E1 should interfere with both interactions. Another point to investigate concerns the not yet fully elucidated pathways of CDON- and PTCH-induced cell death. So far, it seems that downstream effectors involved in both pathways are different because, for example, the dependosome complex does not appear recruited by CDON. Also, it is still not known whether some interaction/synergy between both pathways occurs.

Future works will also have to investigate whether a therapeutic approach targeting SHH/PTCH interaction would be efficient on both tumor and stroma cell death. It has indeed been suggested that autocrine and paracrine SHH signaling cooperate in some tumors (50). For example, in colon, pancreatic, and ovarian cancers, SHH ligand secreted from epithelial tumor cells has been shown to activate SHH signaling in the tumor-associated stromal tissues, which, in turn, could provide a more favorable environment for tumor development. We can thus speculate that targeting SHH ligand could also trigger PTCH-induced cell death in stromal cells and, as an indirect consequence, also prevent tumor progression, even though at this stage we do not know whether, upon paracrine expression of SHH, stromal cells are prime to be dependent for survival on SHH. Together, our work and the different preclinical data showing robust antitumor activity of anti-SHH antibodies (15, 16) strengthen the view that, even though the pharmaceutical industry is current moving away from the HH field, it would actually be of key interest to clinically investigate the efficiency of SHH antibodies compared with HH canonical pathway inhibitors in SHH-overexpressing tumors and to eventually consider it in combination with other current antitumor therapies.

R. Fasan has ownership interest (including patents) in compound HL2m5, which is part of a patent application filed by the University of Rochester. No potential conflicts of interest were disclosed by the other authors.

Conception and design: P.-A. Bissey, P. Mehlen, J. Fombonne

Development of methodology: P.-A. Bissey, I. Goddard, G. Ichim, J. Fombonne

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): P.-A. Bissey, P. Mathot, C. Guix, M. Jasmin, I. Goddard, N. Gadot, J.-G. Delcros, S.M. Mali, R. Fasan, G. Ichim, J. Fombonne

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): P.-A. Bissey, P. Mathot, C. Guix, I. Goddard, N. Gadot, A.-P. Arrigo, R. Dante, J. Fombonne

Writing, review, and/or revision of the manuscript: J.-G. Delcros, A.-P. Arrigo, R. Dante, P. Mehlen, J. Fombonne

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): C. Guix, M. Jasmin, I. Goddard, J. Fombonne

Study supervision: P. Mehlen, J. Fombonne

Other (experimental help): C. Costechareyre

We wish to thank Emilie Servoz (from LMT) for technical assistance in the animal facility. This work was supported by institutional grants from CNRS, University of Lyon, Centre Léon Bérard, and from the Ligue Contre le Cancer, INCA, ANR, and ERC. P. Mathot has been supported by a fellowship from la Ligue Contre le Cancer. S.M. Mali and R. Fasan acknowledge support from the NIH grant no. R01 GM134076.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.