Introduction

The sonic hedgehog (Hh) signaling pathway is essential for embryonic development and organ formation [1]. Upon completing embryogenesis, the Hh pathway generally goes silent while remains active in certain circumstances such as hair follicle maintenance and tissue repair [2]. The canonical Hh signaling is initiated by the Hh ligand binding to the pathway suppressor PTCH-1. This action rapidly stimulates the key G protein-coupled receptor (GPCR) smoothened (SMO) leading to activation of Gli transcription factors and pathway biological function [3].

Aberrant Hh/SMO/Gli1 signaling pathway is linked to pathogenesis of numerous human cancers including advanced basal-cell carcinoma (BCC), medulloblastoma (MB), and other solid tumors [4, 5]. Studies also identified at least three mechanisms of Hh pathway activation in tumorigenesis, including the autocrine ligand-dependent, paracrine ligand-dependent, [6] and ligand-independent activation [3, 5, 7]. In the ligand-independent setting, tumors were frequently found to possess the genetic loss of PTCH-1, resulting in constitutive over-activation of SMO [3]. Employing gene depletion technique or small molecule inhibitors, targeting Hh pathway led to antitumor efficacy in preclinical models of BBC and MB both in vitro and in vivo [8]. These earlier findings draw increasing attention to Hh/SMO pathway in cancer biology and therapy development. Thus far, two SMO inhibitors Vismodegib (GDC-0449) and Sonidegib (LDE225) have received FDA approval for cancer treatment. In clinical trials, GDC-0449 and LDE225 were used as single agent therapy in BCC and MB patients and achieved impressive clinical response in overall disease control [9,10,11].

Acquired drug resistance is a major and evolving challenge in targeted cancer therapy, including those with the approved SMO inhibitors [12, 13]. Clinical experience indicated that GDC-0449 and LDE225 could confront a cross resistance [14]. To address this issue, resistance mechanisms are under intense investigation. Employing genomic analysis of GDC-0449-resistant tumors, several SMO gene mutations have been discovered [15]. It particular, a recurring mechanism involves the SMO-D473H mutation, which blocks GDC-0449 or LDE225 from efficient binding to the mutant SMO protein. In a study, nearly 40% BCC patients who acquired resistance were shown to contain SMO mutations and 17% patients carry the SMO-D473H molecular type [16]. More recently, new generation SMO inhibitors such as TAK-441, LEQ-506, and IPI-926 have entered clinical studies [17,18,19]. These latest agents could overcome mutant SMO activity in the preclinical setting and their clinical performances are eagerly awaited.

In this study, we aimed to design and screen a series of new SMO inhibitors. We show that several compounds not only displayed potent suppression of the Hh pathway function but also are able to overcome the drug resistance caused by SMO-D473H.

Materials and methods

Chemicals and cell culture

The current HH series of compounds were synthesized as described [20, 21]. GDC-0449 was purchased from BiochemPartner (Shanghai). Smoothened agonist (SAG) was purchased from Sant Cruz Biotechnology. Inhibitors were dissolved in DMSO as 20 mmol/L stock solution and were freshly diluted before assays. HEK293T, NIH3T3, and Calu-6 cell lines were purchased from American Type Culture Collection (ATCC) and were cultured as recommended.

SMO-D473H mutagenesis

p-CMV6-Smo-D473H point mutation was established with standard site-directed mutagenesis technique. Primers were as follows: F: 5′-TGAAGAAGTGGTAGAAGTGGCAGCTGAAGGTAAT-3′; R: 5′-CACTTCTACCACTTCTTCAACCAGGCTGAGT-3′.

Secreted hedgehog (SHH) condition medium (SHH-CM)

293T cells were plated 3 × 106 cells per well in 55 cm2 dish. The next day, cells were transfected with pcDNA3.1 ShhN construct encoding the N-terminus secreted Hh fragment (SHH, Addgene) using Lipofectamine 2000 (Invitrogen). After 24 h, medium was replaced with DMEM containing 2% FBS. The SHH-CM was collected at 48 and 72 h, pooled and then centrifuged at 1000 rpm for 10 min. The supernatant was stored at 4 °C until use within a week.

Gli1-luciferase reporter assay

NIH3T3 cells were plated in 96-well plate at 5 × 103 cells per well for overnight. Cells were transfected with 180 ng 8 × 3′ Gli-BS-delta51-LucII (RIkagaku KENkyusho/Institute of Physical and Chemical Research) and 20 ng TK-Renilla using Lipofectamine 2000. After 6 h, medium was refreshed to 10% new-born calf serum (NBCS) in DMEM for 24 h. On the day of inhibitor assay, cells were fed with medium containing 0.5% NBCS for 1 h, then stimulated with 100 ng/mL SAG or 50 μL SHH-CM without or with various testing inhibitor compounds. Forty-eight hours later, cells were lysed and tested for luciferase activity with Dual-Luciferase Reporter Assay Kit (YPHbio, Beijing). For GDC-0449-resistant assay system, 293T cells were plated in 96-well plate, 5 × 103 per well. The next day, cells were transfected with 100 ng 8 × 3′ Gli-BS-delta51-LucII and 100 ng p-CMV6-SMO-WT or p-CMV6-SMO-D473H plasmid DNA for 24 h, treated with various inhibitors for 48 h, followed by luciferase activity assay performed as described above.

RT-PCR assay of gene expression

Cells or tumors were harvested and extracted total RNA according to SuperfecTRI total RNA Isolation Reagent (Pufei, Shanghai). Reverse transcription of mRNA into cDNA was performed with RevertAid First Strand cDNA Synthesis Kit (Fermentas), subjected to PCR with FastStart Universal SYRB Green Master (Rox) Kit (Roche) in iQ5 Cycler (Bio-Rad). PCR primers included mouse Gli1, F: 5′-GCAGTGGGTAACATGAGTGTCT-3′, R: 5′-AGGCACTAGAGTTGAGGAATTGT-3′; mouse PTCH-1, F: 5′-CTCTGGAGCAGATTTC CAAGG-3′, R: 5′-TGCCGCAGTTCTTTTGAATG-3′; mouse CycinD1, F: 5′-GTGCGTGCAGAAGGAGATTG-3′, R: 5′-AGGTTCAGGCCTTGCATC-3′; mouse RPL19, F: 5′-AGAAGGTGACCTGGATGAGA-3′, R: 5′-TGATACATATGGCGGTCAATCT-3′; human GAPDH, F: 5′-TTCACCACCATGGAGAAGGC-3′, R: 5′-GGCATGGACTGTGGTCATGA-3′; human SMO, F: 5′- ACCTATGCCTGGCACACTTC-3′, R: 5′- GTGAGGACAAAGGGGAGTGA-3′. Relative gene expression was analyzed and normalized to RPL19 (mouse genes) or GAPDH (human genes).

Cell proliferation and cell cycle analysis

Cell proliferation was assayed with the indicated treatments for 3 days using MTS reagent (Promega) as previously described [22]. For cell cycle profiling, cells were treated as indicated for 24 h and analyzed in a Becton Dickinson FACS Calibur flow cytometer by collecting 10,000 events.

Apoptosis assay

293T cells were transiently transfected with p-CMV6-SMO-WT for 24 h, treated with inhibitors for 48 h. The cells were harvested, stained with Annexin V-FITC/PI staining kit (KeyGEN BioTECH, Cat# KGA108) and analyzed by FACS.

Calu-6 human lung tumor xenografts

Animal studies were performed under protocols approved by Institutional Animal Care and Use Committee (IACUC) of Fudan University. Balb/c female nude mice (Shanghai Sippr-BK laboratory animal Co. Ltd.) between 6 and 8 weeks of age were subcutaneously injected 4 × 106 Calu-6 cells in a Matrigel (BD Biosciences) /DMEM mixture. When tumor size reached ~400 mm3, mice were treated twice a day (bid) with vehicle (20% HP-β-CD) or 75 mg/kg GDC-0449, HH-1, HH-13, and HH-20 by oral gavage. Mice were killed after five treatments. Tumors were collected and analyzed for mouse stromal cell mRNA level through RT-PCR.

PTCH+/−/P53−/− mouse medulloblastoma (MB) model

C57BL/6J PTCH1tmlMps/J (PTCH+/−) was purchased from Jackson Laboratory, C57BL/6J B6.129S2-Trp53tmlTyj/JNju (P53−/−) was purchased from Model Animal Research Center of Nanjing University. PTCH+/−/P53−/− mouse MB model on a C57BL/6 background was generated as previously described [23]. Mice with spontaneous MB tumors were sacrificed. Tumors were dissected from cerebellum and then subcutaneously implanted into Balb/c female nude mice (6–8 week) (P1 tumors). The grafted tumors were allowed to grow to 300–500 mm3 before harvested and serially passaged (P2 and P3 tumors). For efficacy studies, tumor fragments were implanted into flank of female nude mice. Mice bearing staged tumors (150–200 mm3 in size) were randomized into groups (n = 8), treated orally with vehicle or the indicated inhibitors via a twice daily (bid) regimen. Tumor growth was monitored with caliper every 2 days. Tumor volume was calculated as V (mm3) = L × W × W/2. Tumors were collected at the end of study.

Immunoblotting

For standard assay of inhibitor effects on cellular biomarker expression, cells were plated at predetermined density in 6-well culture plates for overnight, treated with SHH-CM without or with inhibitors for the indicated doses and time. Cells were lysed with NuPAGE-LDS lysis buffer (Invitrogen) and immunoblotted with antibodies including Gli1 (Santa Cruz); PTCH-1, cyclin D1, P-TSC2(T1462), P-mTOR(S2481), mTOR (Cell Signaling Technology); SMO (Abgent); P-AKT(S473), AKT, P-S6(S235/236), S6 (Abcam-Epitomics); MCL-1 (Affinity); β-actin, GAPDH (Bioworld). Immunoblotting of in vivo tumors were performed as described previously [22].

Statistical analysis

All numerical data processing and statistical analysis were performed with Microsoft Excel Software; values were expressed as means ± SD. P values were calculated using unpaired two-tailed Student t test.

Results

Discovery of benzimidazole derivatives as novel SMO inhibitors

To identify novel SMO inhibitors, we first analyzed the published structure and activity information for GDC-0449 and its analogs. This analysis led to the design and synthesis of multiple series of substituted benzimidazole derivatives [20, 21]. Antagonists were screened by two versions of cell-based Gli1-reporter assay. The Gli1-luc reporter activity was induced either with a secreted hedgehog-condition medium (SHH-CM) or with a known SMO agonist (SAG) compound (Fig. 1a). In the SHH-induced reporter system, numerous compounds displayed potent antagonist activity with IC50 value range <200 nmol/L (Fig. 1b). When screened in the SAG-induced reporter system, while majority of the compounds showed a significantly reduced potency, compounds HH-13 and HH-20 retained potent antagonist activity (Fig. 1c). Specifically, HH-13 and HH-20 elicited respective IC50 values of 6.8 and 21.7 nmol/L in SHH-CM system, and showed 9.3 and 28.9 nmol/L in SAG system (Fig. 1b, c). Meanwhile, we performed in vivo bioavailability studies and found that the mouse oral exposure ratio for compound HH-1 was significantly higher than other compounds (Figure S1A). These results and additional findings (see below) led us to select HH-1, HH-13, and HH-20 (Fig. 1d) as hit molecules.

Fig. 1
figure 1

Screening of newly designed SMO inhibitor compounds in NIH3T3 cell Gli1 report gene assay . a Gli1-luciferase assay scheme is shown. The assays were performed as described in Method. b, c Dose–response inhibition curves (left) and IC50 values (right) are shown. The cellular Hh pathway activation was induced with SHH-CM (b) or SAG (c). d Chemical structures of HH-1, HH-13, and HH-20 are shown

Full size image

Targeting of the Hh pathway gene transcription

In serum-starved NIH3T3 stimulated with SHH-CM, mRNA expression of the Hh pathway genes Gli1, PTCH-1, and cyclin D1 increased by ~60-fold, 20-fold, and 2-fold, respectively (Fig. 2a). To test the inhibitory effect on these genes, serum-starved NIH3T3 cells were induced with SHH-CM in the presence of inhibitors for 48 h. All three inhibitors dose dependently reduced mRNA levels for Gli1 (Fig. 2b), PTCH-1 (Fig. 2c) with an overall potency order of HH-13 > HH-20 > HH-1. Specifically, both HH-13 and HH-20 inhibited these two genes at doses as low as 0.01–0.05 μmol/L, while HH-1 showed inhibition at 0.5 μmol/L (Fig. 2b, c). Interestingly, all three inhibitors profoundly inhibited the expression of cyclin D1 even at 0.01–0.05 μmol/L (Fig. 2d). These results demonstrate a dose-dependent inhibition of the endogenous Hh-target gene transcription in response to treatment with HH-1, HH-13, and HH-20.

Fig. 2
figure 2

New SMO inhibitors target the Hh pathway gene expression. a mRNA levels of Gli1, PTCH-1, and cyclin D1 were analyzed after serum-starved NIH3T3 cells were induced with SHH-CM for 48 h as described in Methods. b–d Serum-starved NIH3T3 cells were similarly induced by SHH-CM with vehicle or the indicated inhibitors for 48 h. Cells were analyzed for mRNA levels of Gli1 (b), PTCH-1 (c), and cyclin D1 (d). Relative expression levels are shown. Statistical analysis (treated vs. DMSO control): *P < 0.05; **P < 0.01; ***P < 0.001

Full size image

Inhibition of Hh ligand-dependent cell growth

We next established an Hh pathway-dependent growth assay system. Treatment of NIH3T3 cells with SHH-CM led to a sharp increase in cell growth which could be suppressed by GDC-0449 (Fig. 3a). In this assay system, while the vast majority of inhibitors displayed an expected inhibitory activity, HH-13 and HH-20 showed the most potent and consistent suppression of cell growth with IC50 values of <9 nmol/L and 21 nmol/L, respectively (Fig. 3b). Cell cycle analysis revealed a significant increase in the G0/G1 phase and decrease in S phase distribution in cells treated with GDC-0449, HH-13, or HH-20 (Fig. 3c). Consistently, immunoblotting of the inhibitor-treated cells demonstrated a downregulation in protein levels of Gli1 and cyclin D1 compared to that of vehicle-treated cells (Fig. 3d). These results suggest that similar to GDC-0449, the current series of HH inhibitors could effectively target the Hh signaling pathway function leading to growth suppression via a blockade in G0/G1 cell cycle progression.

Fig. 3
figure 3

Targeting of Hh pathway leads to inhibition of cell proliferation. a The SHH-induced NIH3T3 cell proliferation model was established (left panel) and validated with GDC-0449 (right panel). Cells were induced by SHH-CM for 72 h and measured for cell proliferation as described in Methods. b Cells as in a were treated with vehicle or the indicated inhibitors for 72 h followed by proliferation assay using MTS reagent. Growth inhibition dose–response curves are shown (left panel) and IC50 values are shown (right panel). c Cells as in a were treated with the indicated inhibitors for 24 h, subjected to cell cycle analysis by FACS. Cell populations in G0/G1 and S phase are shown. d Cells as in a were treated with the indicated inhibitors for 48 h. Total cell lysates were collected and subjected to immunoblotting. Statistical analysis (treated vs. DMSO control): *P < 0.05; **P < 0.01; ***P < 0.001

Full size image

Downregulation of mTOR pathway and induction of apoptosis in inhibitor-treated cells

Intracellular mTOR pathway plays a major role in cell growth and survival. We therefore examined signaling status of mTOR complexes (mTORCs) as the potential downstream mediators of SMO-targeted growth inhibition. We found that the SHH-induced phosphorylation of mTORC1 biomarker P-S6 (S235/236) and mTORC2 biomarker P-AKT (S473) in NIH3T3 cells were potently and profoundly inhibited by HH-13 and HH-20 with effective doses at <1 μmol/L (Fig. 4a). For comparison, 1 μmol/L GDC-0449 near completely suppressed P-S6 and consistently but partially reduced P-AKT. These results indicated that the anti-proliferative effects of the current SMO inhibitors are mediated through downregulation of mTOR activity. We examined additional mTOR readouts and signaling regulation. In cells treated with HH-13, HH-20, and GDC-0449, there were substantial reduction in P-TSC2 (T1462) as well as partial loss of P-mTOR (S2481) and mTOR, all of which are consistent with a diminished signaling input to mTOR network (Fig. 4b). Importantly, targeting of SMO in this setting also dramatically downregulated level of MCL-1, an apoptosis inhibitory protein recently shown to be protected by mTORC2/AKT axis [24] (Fig. 4b). We then analyzed 293T human cells transfected with the wild-type smoothened (SMO-WT). After 48 h treatment, there was a dose-dependent and substantial increase of apoptosis in both the GDC-0449- and HH-13 treated cells as shown by Annexin V/PI staining (Fig. 4c) and PARP cleavage (Fig. 4d). Taken together, these results highlight a molecular suppression of mTOR network and induction of apoptotic death in SMO-targeted antitumor effects.

Fig. 4
figure 4

Targeting of Hh pathway function leads to downregulation of mTOR/AKT activity. a, b Serum-starved NIH3T3 cells were induced with SHH for 48 h without or with the indicated inhibitors. Total cell lysates were prepared and subjected to immunoblotting as indicated. c 293T cells were transfected with SMO-WT expression vector for 24 h, treated with inhibitors for 48 h, subjected to Annexin V-FITC/PI staining. Positive populations as analyzed by FACS are plotted. d Cells as in c were collected and immunoblotted

Full size image

Blockade of the Hh paracrine signaling in tumor environment

Paracrine signaling has an important role in Hh/SMO pathway activation of various cancers. A schematic diagram shows the in vivo paracrine stimulation of stromal Hh pathway (Fig. 5a) [25, 26]. The Calu-6 lung tumor in vivo mouse model could be used as an Hh paracrine stimulation model [27]. We observed that while no mouse-specific gene expression was detected in cultured Calus-6 cells, the in vivo tumors were associated with an extensive recruitment and activation of mouse stromal Gli1, PTCH-1, and cyclin D1 gene expression (Fig. 5b–d, left panels). Oral treatment of the Calu-6 tumor-bearing mice with HH-1, HH-13, and HH-20 (75 mg/kg, bid; total five doses) each led to a highly substantial suppression of the mouse stromal Gli1, PTCH-1, and cyclin D1 (P < 0.001), which were comparable to that by GDC-0449 (Fig. 5b–d, right panels). These results indicate that similar to GDC-0449, the current series of inhibitors can effectively inhibit Hh paracrine activation in vivo.

Fig. 5
figure 5

New SMO inhibitors target paracrine Hh pathway in vivo. a Schematic diagram shows the in vivo paracrine stimulation of stromal Hh pathway. b–d Nude mice bearing Calu-6 tumors (~400 mm3) were dosed orally with 75 mg/kg GDC-0449, HH-1, HH-13, or HH-20 twice daily (bid) for a total of five doses. Tumors were collected and analyzed for stromal expression of mouse Hh pathway biomarkers using mouse-specific PCR primers. Tumors were analyzed for mRNA levels of mouse Gli1 (b), mouse PTCH-1 (c) and mouse cyclin D1 (d). In RT-PCR measurements of each gene set, RNA samples of Calu-6 cells grown in cell culture dish served as a background control for the assay (left panels). For tumor results, relative expression levels are shown (right panels). Statistical analysis (treated vs. vehicle control): ***P < 0.001

Full size image

Antitumor activity in PTCH+/−/P53−/− medulloblastoma mouse model

There is strong evidence that nearly one third of medulloblastoma (MB) is related to the dysregulation of the Hh pathway [28] and the PTCH+/−/P53−/− spontaneous MB is a good experimental mouse model. [29] We established this model (Fig. 6a) and validated it using GDC-0449, which demonstrated the expected efficacy (not shown). Oral treatment of mice bearing the MB allograft tumors with 75 mg/kg HH-1 or 50 mg/kg HH-13 resulted in a significant tumor growth delay (P < 0.01) without overt toxicity or loss of body weight (Fig. 6b, c). Similar treatment with 50 mg/kg HH-20 showed a hint of efficacy (P < 0.05) (Figure S1B). The weaker in vivo activity of HH-20 may in part reflect the lower bioavailability of this compound (Figure S1A). Immunoblotting analysis showed a trend for reduced Gli1 protein in inhibitor-treated tumors compared to vehicle control (P < 0.05) (Fig. 6d). These results overall indicate that HH-1, HH-13, and HH-20 can elicit in vivo targeting of Hh-signaling function and the pathway-driven tumor growth, albeit less potent than required for drug development.

Fig. 6
figure 6

Inhibition of PTCH+/−/P53−/− medulloblastoma (MB) allograft tumor growth in vivo. a Mice bearing PTCH+/− P53−/− MB brain tumors (left panel) and allograft tumors (right panel) are shown. b, c Mice bearing the staged allograft MB tumors were treated with vehicle, 75 mg/kg HH-1 (b) or 50 mg/kg HH-13 (c) following a twice daily (bid) oral dosing regimen. Tumor growth curves (left panels) and body weight changes (right panels) are shown. d At the end of study in c, tumors were collected and subjected to immunoblotting (left panel). The quantified results (n = 4) are plotted (right panel). Statistical analysis (treated vs. vehicle control): *P < 0.05; **P < 0.01; ***P < 0.001

Full size image

HH-13 and HH-20 potently target GDC-0449-resistant Smoothened D473H mutation

Finally, to investigate whether the current inhibitor series could overcome acquired drug resistance to the first generation SMO inhibitors, we established the smoothened D473H mutant expression construct [16] (Fig. 7a). Upon transfection into 293T cells, both the SMO wide-type (SMO-WT) and SMO-D473H mutant (SMO-D473H) could be expressed with similar yields (Fig. 7b). In parallel Gli1-luc reporter assays of SMO-WT and SMO-D473H, GDC-0449 potently inhibited the activation of SMO-WT (IC50: 0.076 μmol/L) but was poorly active in the assay for SMO-D473H (IC50: >60 μmol/L) (Fig. 7c, top panel). In contrast, both HH-13 and HH-20 retained decent inhibition potency when assayed against SMO-D473H, achieving IC50 values of 0.086 and 0.174 μmol/L, respectively (Fig. 7c, d). To further confirm these results, we examined the effect on down-regulating the signaling biomarker driven by SMO-D473H. In 293T cells transfected with SMO-D473H, there was a coordinated activation in Gli1, cyclin D1, and P-AKT (Fig. 7e). Although these SMO-D473H-dependent signaling biomarkers were largely resistant to GDC-0449, they were dose dependently blocked by HH-13 and HH-20 with approximate EC50 values of 0.3 μmol/L and 1 μmol/L, respectively (Fig. 7e). Thus, these results overall indicate that both HH-13 and HH-20 could effectively target the GDC-0449-resistant SMO-D473H mutation.

Fig. 7
figure 7

HH-13 and HH-20 are effective in targeting GDC-0449-resistant SMO-D473H. a Construction of SMO-WT and SMO-D473H expression vectors. DNA sequence with the D473H point mutation is shown. b 293T cells were transfected with SMO-WT or SMO-D473H plasmids for 8 h, collected for analysis of SMO overexpression by PCR. c, d 293T cells were transiently transfected with Gli1-luc plus SMO-WT or Gli1-luc plus SMO-D473H plasmids for 24 h, treated with the indicated inhibitors for 48 h. Luciferase assays were performed as described in Methods. For inhibition results, dose–response curves (c) and IC50 values (d) are shown. e 293T cells expressing the SMO-D473H were treated with inhibitors for 48 h. Cells were collected and subjected to immunoblotting

Full size image

Discussion

The FDA-approved Vismodegib/GDC-0449 and Sonidegib/LDE225 have presented a new hope for surgery unfavorable metastatic BCC patients [30, 31]. Other SMO inhibitors under development also showed positive responses in disease control [32, 33]. These achievements have spurred a new wave of SMO inhibitor research. As part of expanded discovery efforts, we uncovered a new series of SMO inhibitors, exemplified by HH-1, HH-13, and HH-20 that potently target the Hh/SMO/Gli1 pathway function in vitro and in vivo. Strong evidence points to the aberrant Hh signaling as cancer driver in advanced BCC and PTCH-deficient MB molecular type [5, 34]. The current inhibitor series are shown to inhibit the Hh pathway-driven cell proliferation accompanying G0/G1 cell cycle blockade and reduction of the pathway target gene expression. Moreover, in the Hh-driven PTCH+/− MB in vivo model, both HH-1 and HH-13 attenuated allograft tumor growth, which correlated with the downregulation of pathway related gene expression.

As SMO inhibitors display profound anti-proliferation capacity in the Hh-driven NIH3T3 model and MB allografts, some studies observed no direct correlation between downregulation of pathway genes and in vitro anti-proliferative response in certain cancer cells [35]. Emerging evidence now supports a role for paracrine Hh signaling activation in tumorigenesis [35, 36]. Indeed, our work confirmed the presence of lung tumor cell Hh-induced paracrine stromal activation in mice, in which the tumor-induced stromal expression of Gli1, PTCH-1, and cyclin D1 were all blocked by orally administered GDC-0449 and current series of SMO inhibitors.

Downregulation of mTOR/AKT signaling function has an important role in SMO inhibitor-mediated growth suppression. The relationship was also observed in ovarian cancer cells in which depletion of AKT gene could partially reverse Gli1-induced EMT [37]. In chronic lymphocytic leukemia cell line, a Gli1 inhibitor GAN61 inhibited AKT phosphorylation [38]. In our studies, both GDC-0449 and current series of inhibitors reduced P-TSC2 (T1462) resulting in higher suppressive activity toward mTORC1. Likewise, both inhibitor classes targeted mTORC2 substrate P-AKT (S473) and its secondary readout MCL-1 [24]. Because MCL-1 is a well-characterized inhibitor of apoptosis, suppression of MCL-1 in this setting likely contributed to the efficacy of SMO inhibitors. Although the exact mechanism on how Hh/SMO signals to TSC2/mTORC1 or mTORC2/AKT remains to be elucidated, the results of current work and previous reports highlight a critical role for mTOR network in the Hh pathway-dependent cancer cell growth and tumorigenesis [38,39,40,41].

Acquired drug resistance and tumor relapse are weaknesses of the first generation SMO-targeted drugs. Our inhibitors are intended to overcome the clinically prevalent smoothened mutation SMO-D473H, which renders GDC-0449 and LDE225 binding inefficient and clinical resistance. We have shown that in contrast to GDC-0449, both HH-13 and HH-20 elicited potent inhibition of SMO-D473H. This conclusion is based on their activity in specific suppression of SMO-D473H-driven Gli1-reporter signal and the blockade of SMO-D473H-driven cellular pathway biomarkers. Additional support for specific SMO-targeting by these compounds came from their structure and activity relationship (SAR) profile. In this SAR process, when we varied substituent on the pyridine ring and amide benzene ring, their activity behaviors were very similar to the SAR profile for GDC-0449 [42]. A major deficiency of our study is the lack of direct binding data on these inhibitors to purified SMO proteins. Future studies are required to elucidate the inhibitor binding mode to both the naïve and drug-resistant SMO variants. Collectively, the current inhibitors may present an opportunity for medicinal chemistry optimization on pharmaceutical property towards the development of new generation SMO inhibitor therapy.