Leukotriene Synthesis Is Critical for Medulloblastoma Progression Free

Medulloblastoma is a malignant tumor of developing cerebellum, representing a major cause of childhood morbidity and mortality (1). Current treatment for medulloblastoma consists of a combined modality approach including surgery, chemotherapy, and radiotherapy. Despite aggressive treatment, a significant number of patients with medulloblastoma still succumb to this disease. Moreover, tumor treatment can cause delayed complications that have profound effects on quality of life in survivors (2–4). Clinical studies and research efforts are now focused on attempts to decrease treatment toxicity while improving the cure rate in patients with medulloblastoma.

Leukotrienes (LT), well-known inflammatory mediators, possess a wide range of biological activities including leukocyte chemotaxis, vascular leakage, endothelial cell migration, and astrocyte proliferation (5). The synthesis of LT from substrate arachidonic acid (AA) is initiated by 5-lipoxygenease (Alox5) in concert with 5-lipoxygenase-activated protein (FLAP; Fig. 1A; ref. 6). Recent studies reveal that LTs are implicated in many pathological conditions such as allergic diseases, cardiovascular diseases and many cancers (7). However, it is still unknown whether LTs are involved in medulloblastoma tumorigenesis.

Here, we show that LT biosynthesis is elevated in medulloblastoma cells, and that enhanced LT biosynthesis in tumor cells relies on sonic hedgehog (Shh) ligand secreted by astrocytes, a major component of the medulloblastoma microenvironment. LTs in medulloblastoma cells are responsible for the induction of Nestin, an intermediate filament protein, which is required for medulloblastoma progression (8). Genetic or pharmaceutical blockage of LT synthesis prevented Nestin expression and resulted in inhibition of medulloblastoma cell proliferation in vitro and in vivo. Moreover, inhibition of LT synthesis exhibited antitumor efficacy in drug-resistant medulloblastoma cells, but did not affect proliferation of normal cerebellar granule neuron precursors (GNP). Our data reveal a novel tumor-specific signaling pathway for LT-induced Nestin expression, which is essential for medulloblastoma cell proliferation. Given the excellent safety profile, low cost, and ready availability of LT inhibitors, LT synthesis represents a promising therapeutic target for treatment of medulloblastoma and perhaps other malignancies involving activation of the hedgehog (Hh) pathway.

Ptch1^fl/fl mice have been described previously (9). Alox5 null mice, Ptch1^+/− mice, Math1-Cre mice, P53-null mice, and R26R-SmoM2 mice were purchased from The Jackson Laboratory. CB17/SCID mice were bred in the Fox Chase Cancer Center Laboratory Animal Facility (LAF). All animals were maintained in the LAF at Fox Chase Cancer Center and all experiments were performed in accordance with procedures approved by the Fox Chase Cancer Center Animal Care and Use Committee.

Medulloblastoma cells were isolated and cultured with NB-B27 medium in poly-D-lysine coated 96-well plate as described previously (9). Smo antagonist, vismodegib, was purchased from Cellagen technology. MK886, arachidonic acid, Cysteinyl LTs, LTA4-E4, montelukast, and zileuton were obtained from Cayman Chemical. Cell proliferation assays were performed using Thiazolyl Blue Tetrazolium Bromide (MTT; Sigma). Briefly, cells were seeded in flat bottom 96-well plates at a density of 4 × 10⁵ cells per well in 100 μL of NB-B27 medium in the presence of MK886 or vismodegib accordingly. After treatment for 48 hours, the culture medium was aspirated and replaced with 100 μL per well of MTT (5 mg/mL) diluted in NB-B27 medium at a 1:10 ratio. Reduction of MTT by living cells was determined after 4 hours of incubation by the difference in absorbance between 570 nm and 655 nm using an iMark microplate reader (Bio-Rad).

Cells and tissues were lysed in RIPA buffer (Thermo Fisher Scientific) supplemented with protease (Roche) and phosphatase (Thermo Fisher Scientific) inhibitors cocktail. Total protein lysates were separated in 8% SDS-PAGE and transferred on PVDF membrane. Membranes were probed with antibodies against Alox5 (Cell Signaling Technology, 1:1,000), FLAP (Cell Signaling Technology, 1:1,000), and GAPDH (Sigma-Aldrich, 1:1,000). Western blot signals were detected with SuperSignal West Pico Substrate and exposed on films.

RNA was isolated using TRI reagent (Sigma-Aldrich) in RNase-free conditions. cDNA was synthesized using oligo(dT) and Superscript II reverse transcriptase (Invitrogen). Quantitative PCR reactions were performed in triplicates using iQ SYBR Green Supermix (Bio-Rad) and the Bio-Rad iQ5 Multicolor Real-Time PCR Detection System. Primer sequences are available upon request.

Immunofluorescent staining of frozen sections and cells was carried out according to standard methods. Briefly, sections or cells were blocked and permeabilized for 2 hours with PBS containing 0.1% Triton X-100 and 1% normal goat serum, stained with primary antibodies overnight at 4°C, and incubated with secondary antibodies for 2 hours at room temperature. Sections or cells were counterstained with DAPI and mounted with Fluoromount-G (Southern Biotech) before being visualized with a Nikon TE200 microscope.

Levels of endogenous leukotriene in GNPs and medulloblastoma cells were measured using a leukotriene EIA kit (Cayman Chemical). Briefly, cells were lysed in 200 μL of methanol, centrifuged to clear the lysate and then evaporated in a SpeedVac concentrator. After reconstitution in 100 μL of EIA buffer, cysteinyl leukotriene content was assayed according to manufacturer's instructions.

To purify astrocytes or macrophages/microglia, medulloblastoma or postnatal cerebella were digested with 10 U/mL papain (Worthington) and triturated to obtain a single cell suspension. Cells (at a density of 2–4 × 10⁶/mL) were then stained on ice, for 1 hour with primary antibodies and 30 minutes with secondary antibodies, and analyzed by FACS.

An antibody against ACSA2 (astrocyte cell surface antigen 2, Cell Signaling Technology, 1:200), was used to sort astrocytes. Antibodies against CD68 (BioLegend, 1:100) and CD11b (BioLegend, 1:100) were used for macrophage/microglia purification.

Medulloblastoma cells (2 × 10⁶) from Math1-Cre/Ptch1^fl/fl mice were mixed with Matrigel, and subcutaneously injected into the hindflank of CB17/SCID mice. Tumor size was measured every two days, and tumor volume was calculated based on the formula: (W² × L)/2 (W, width; L, length). Daily intraperitoneal injection of MK886 was started after the tumor volume reached 200 mm³. The animals were sacrificed when the tumor volume exceeded 1 cm³. After sacrifice, tumors were collected and sectioned for IHC analysis.

For experiments in Fig. 6F–J, MK886 DMSO stock was dissolved in 0.5% methylcellulose 0.2% Tween-80. Drugs were administered by oral gavage using 24-G gavaging needles (Fine Science Tools).

The raw gene expression data used for analyzing subtype specific pathway alterations in human medulloblastoma (Fig. 1B and C) were obtained from GEO (GSE85218) and normalized using RMA. Gene set variation analysis (GSVA) on the RMA normalized expression data was applied to identify pathways that are enriched in a single sample (method arguments: function = gsva; mx. diff = ‘TRUE’). A total of 17,799 gene sets compiled from all gene sets available in MsigDB, including KEGG and Biocarta pathways, were used for enrichment analysis. To identify pathways that differ significantly among subtypes, Kruskal–Wallis test was used to analyze the enrichment scores resulting from GSVA.

Student t test was performed to determine the statistical significance of the difference in means between samples. Extra sum-of-squares F test was used to analyze the difference in cell proliferation between SmoM2 medulloblastoma cells and ptch1-deficient medulloblastoma cells in Fig. 5J. P < 0.05 was considered statistically significant. Error bars represent the SEM. Difference in tumor volume increase in Fig. 4A was assessed using the Kaplan–Meier survival analysis and the Mantel–Cox log-rank test was used to assess the significance of difference between two cohorts of tumor bearing mice. Data handling and statistical processing was performed using Microsoft Excel and GraphPad Prism Software.

Approximately 30% of human medulloblastoma result from aberrant activation of the Shh pathway (Shh-MB; refs. 10, 11). To investigate a possible association of LT synthesis with human Shh-MB, we performed Gene Ontology (GO) analysis using transcriptomes from 763 primary human medulloblastoma including 223 Shh type (Shh-MB), 144 group 3, 326 group 4 and 70 Wnt tumors (12). Among 17,799 pathways analyzed (Supplementary Table S1), genes associated with LT biosynthesis and LT metabolism are markedly upregulated in Shh group and Wnt group human medulloblastoma cells compared with group 3 and group 4 (P < 0.001; Fig. 1B and C). These data suggest that LT synthesis and metabolism may be involved in the tumorigenesis of Shh-MB and Wnt-MB. Shh-MB in this dataset consists of four distinct subgroups: 65 Shh-α, 35 Shh-β, 76 Shh-γ, 47 Shh-δ. As shown in Fig. 1D, tumor cells from the Shh β group express most elevated levels of Alox5, the critical enzyme for LT biosynthesis, indicating that LTs may be associated with clinical and/or cytogenetic features of this subgroup of Shh-MB (13).

Mice heterozygous for patched 1 (ptch1, an antagonizing receptor of hedgehog) are widely utilized as a model for Shh-MB, as 15% to 20% of these mice (ptch1^+/− mice) develop medulloblastoma in their cerebella by 3 to 4 months of age (14). Targeted deletion of the ptch1 gene in cerebellar GNPs causes medulloblastoma formation with 100% penetrance (9, 15, 16), indicating that GNPs can serve as the cell of origin for Shh-type medulloblastoma. We examined the level of LTs in medulloblastoma cells freshly isolated from ptch1^+/− mice by ELISA. GNPs from wild-type cerebella at postnatal day 6 (P6) were used as a control. As shown in Fig. 1E, the amount of LTs was significantly increased in medulloblastoma cells compared with GNPs, indicating that LT synthesis was elevated in medulloblastoma cells. However, comparable levels of Alox5 and FLAP proteins were detected in GNPs and medulloblastoma cells, suggesting that increased LT synthesis in medulloblastoma cells was not due to upregulation of Alox-5 and FLAP expression (Fig. 1F). It was previously reported that macrophage/mciroglia can produce LTs (7). To exclude the possibility that observed upregulation of LT biosynthesis in medulloblastoma cells was due to contamination of macrophage/microglia, we examined the presence of macrophage/microglia in isolated medulloblastoma cells and GNPs by FACs. As shown in Supplementary Fig. S1A and S1B, comparable percentage of macrophage/microglia (CD68⁺ and CD11b⁺) was found in isolated medulloblastoma cells and GNPs (0.26% ± 0.04% in GNPs vs. 0.26% ± 0.08% in medulloblastoma cells, P > 0.05). Moreover, no difference in the LT biosynthesis of macrophage isolated from postnatal cerebella and tumor tissue was observed (Supplementary Fig. S1C). The above data suggest that elevated levels of LTs found in medulloblastoma cells were not caused by macrophage/microglia presence in isolated tumor cells.

Having observed increased LT biosynthesis in medulloblastoma cells compared with GNPs, we investigated whether enhanced synthesis of LT in medulloblastoma cells was due to ptch1 deficiency. For this purpose, we isolated cerebellar GNPs from mice carrying a Cre-deletable allele of ptch1 (ptch1^fl/fl mice) (17), and infected them in vitro with a lentivirus carrying a GFP-tagged Cre recombinase or an empty GFP construct as a control. As shown in Fig. 2A, the Shh pathway was activated in GNPs following infection with Cre recombinase, as evidenced by elevated expression of Shh pathway target genes such as gli1 and ptch2. However, comparable levels of LT were detected in ptch1-deficient GNPs and control cells (Fig. 2B), suggesting that ptch1 deficiency alone is not capable of stimulating LT synthesis in GNPs. Astrocytes, a major cellular component in tumor microenvironment, promote medulloblastoma cell proliferation and tumor growth by secretion of the Shh ligand (16). To test whether astrocytes could induce LT synthesis in medulloblastoma cells, we purified tumor-associated astrocytes from ptch1^+/− mice by FACs using an antibody against astrocyte cell surface antigen 2 (ACSA2; ref. 18), and harvested conditioned culture medium as previously described (16). Then, ptch1-deficient GNPs were treated with astrocyte-conditioned culture medium (ACM) in the presence and absence of 5E1, an antibody that neutralizes the Shh ligand (19). As shown in Fig. 2B, ACM markedly elevated LT synthesis in ptch1-deficient GNPs, and this effect was abrogated in the presence of 5E1. These data suggest that astrocytes stimulate LT biosynthesis in ptch1-deficient GNPs through Shh secretion.

To further examine whether Shh ligand can promote LT synthesis in medulloblastoma cells, medulloblastoma cells from ptch1^+/− mice and GNPs from wild-type cerebella at P6, were treated in vitro with 3 μg/mL recombinant Shh for the designated time periods. The level of LT synthesis markedly increased in medulloblastoma cells at 10 minutes or 2 hours following treatment with recombinant Shh (Fig. 2C). The short time-frame associated with increased LT biosynthesis upon Shh treatment suggests that de novo transcription is not required. As shown in Fig. 2D, comparable levels of Alox5 protein were observed in medulloblastoma cells either treated with Shh or with vehicle control. Shh-induced LT synthesis was also observed in GNPs (Fig. 2E), although the overall level of LT in GNPs was much lower than that in medulloblastoma cells, implying that ptch1 deletion is not necessary for Shh-stimulated LT synthesis in tumor cells. Our finding that Shh-induced LTs biosynthesis in medulloblastoma cells does not require increased Alox5 expression, is consistent with a previous report indicating that Shh stimulated LT synthesis in mouse fibroblasts is independent of transcription (20, 21).

Our recent studies demonstrate that astrocyte-derived Shh promotes medulloblastoma growth by stimulating tumor cells to express Nestin, a type VI intermediate filament protein (16). Having observed that Shh can induce LT synthesis in medulloblastoma cells, we sought to investigate whether LTs mediate Nestin induction in medulloblastoma cells. We previously reported that approximately 20% of GNPs started expressing Nestin in ptch1 deficient mice at P6, by utilizing conditional ptch1-mutant mice expressing CFP in Nestin positive cells (Math1-Cre/ptch1^fl/fl/Nestin-CFP mice; ref. 8). We fractionated Nestin positive (CFP⁺) and negative GNPs (CFP⁻) from Math1-Cre/ptch1^fl/fl/Nestin-CFP mice at P6 by FACs as described previously (8). The level of LTs in Nestin positive ptch1-deficient GNPs were significantly elevated compared with those in Nestin negative counterparts, suggesting that ptch1-deficient GNPs increased LT synthesis as they expressed Nestin (Fig. 3A). Nestin-negative GNPs were treated with exogenous Shh alone, together with MK886, a selective antagonist of FLAP (22), or with nordihydroguaiaretic (NDGA), a lipoxygenase inhibitor (23). After treatment for 48 hours, GNPs were harvested to examine Nestin expression by immunocytochemistry. Consistent with our previous report (16), dramatic expression of Nestin protein was observed in those ptch1-deficient GNPs after Shh treatment (Fig. 3B and C). However, MK886 or NDGA significantly repressed Shh-induced Nestin expression in GNPs (Fig. 3D–F), suggesting that LT synthesis is required for Shh-stimulated Nestin expression in ptch1-deficient GNPs. To test whether Shh-induced Nestin expression observed in GNP culture came from astrocytes, we isolated Nestin negative GNPs from Math1-Cre/ptch1^fl/fl/Nestin-CFP mice at P6, and treated them with Shh in vitro for 48 hours. GNP culture was then harvested to examine Nestin expression as well as the presence of astrocytes by immunocytochemistry. As shown in Supplementary Fig. S2A–S2C, a few astrocytes expressing brain lipid-binding protein (BLBP), glial fibrillary acidic protein (GFAP) or S100 were detected in the culture, which did express Nestin protein. However, these cells account for less than 1% cells in the culture. These data exclude the possibility that Nestin induction in GNP culture after Shh treatment was due to the presence of astrocytes.

Cerebellar GNPs from ptch1^fl/fl mice or wild type mice at P6, were infected with a lentivirus carrying a GFP-tagged Cre recombinase or an empty GFP construct as a control. Following the infection, GNPs were treated with LT or vehicle control (DMSO) for 48 hrs. Extensive expression of Nestin protein was found in ptch1-deficient GNPs after LT treatment (Fig. 3G and H), further confirming that LT is capable of inducing Nestin expression in GNPs after ptch1 deletion. However, LT failed to stimulate Nestin expression in wild type GNPs (Fig. 3I), suggesting that LT-induced Nestin expression in GNPs relies on Shh pathway activation. Nestin mRNA and protein expressions in ptch1-deficient GNPs after LT treatment, were further confirmed by q-PCR and western blotting (Fig. 3J). The above GNPs were also harvested to examine Shh pathway activation by q-PCR. As shown in Fig. 3K, expression levels of gli1 mRNA were much higher in ptch1-defient GNPs compared with wild type GNPs at 48 hours following the infection, suggesting that Shh pathway was activated in GNPs after deletion of ptch1. Although comparable levels of gli1 mRNA expression was observed in ptch1-deficient GNPs after treatment with LTs or vehicle for 48 hours, LT treatment markedly increased gli1 expression in ptch1-deficient GNPs compared with vehicle treatment at 72 hours. These data suggest that LT treatment sustains Shh signaling in ptch1-deficient GNPs, consistent with the functions of Nestin protein in maintaining Shh pathway activation (8).

Having observed that LT-induced Nestin expression in GNPs only after ptch1 deletion, we next tested whether Smo activation is required for LT-stimulated Nestin expression. ptch1-deficient GNPs were treated with LT together with vismodegib, a specific Smo antagonist (24), or DMSO control. 48 hours later, cerebellar GNPs were harvested to detect Nestin expression by immunocytochemistry. As shown in Fig. 3L, vismodegib repressed LT-induced Nestin expressions in ptch1-deficient GNPs. These data suggest that Smo activation is necessary for LT-induced Nestin expression, consistent with our recent finding that Smo activation was required for Shh-induced Nestin expression in medulloblastoma cells (16). Collectively, the above data demonstrate that LT synthesis is required for Shh-induced Nestin expression in medulloblastoma cells.

We previously demonstrated that Nestin expression is critical to sustain medulloblastoma cell proliferation as well as tumor growth (8, 25, 26). In light of the important role of LT in Nestin induction in medulloblastoma cells, we then examined the possible function of LT in medulloblastoma cell proliferation. Medulloblastoma cells from ptch1+/− mice were infected with a lentivirus carrying Alox5 shRNAs or scrambled shRNAs. The efficiency of Alox5 knockdown in medulloblastoma cells by shRNAs was confirmed by Western blotting (Fig. 4A). Forty-eight hours following virus infection, medulloblastoma cells were harvested to examine the proliferation by immunocytochemistry with an antibody against Ki67 (Fig. 4B and C). More than 80% of medulloblastoma cells infected with scrambled shRNA were found positive for Ki67, compared with approximately 20% of tumor cells infected with Alox5 shRNAs (Fig. 4D). No obvious cell death or apoptosis was observed in medulloblastoma cells after virus infection (Supplementary Fig. S3A and S3B). The dramatic reduction in proliferation of medulloblastoma cells after Alox5 knockdown (Fig. 4D), indicates that proliferation of medulloblastoma cells depends on LT synthesis.

We next examined whether LT deficiency alters medulloblastoma growth in vivo. For this purpose, we crossed Math1-Cre/ptch1^fl/fl mice with Alox5-deficient mice (Alox5 null), which carry a germline mutation in the Alox5 gene (27). No obvious alterations in the proliferation of GNPs and cerebellar structure were observed in Alox5-null mice compared with their wild-type counterparts at P6, implying that LTs are not required for normal GNP proliferation (Supplementary Fig. S4; ref. 27). Consistent with our previous reports (15, 28), ptch1 deletion significantly prolonged the proliferation of cerebellar GNPs, resulting in an enlarged cerebellum in Math1-Cre/ptch1^fl/fl mice (Fig. 4E). At 8 weeks of age, tumor cells in Math1-Cre/ptch1^fl/fl cerebella were actively proliferating (Ki67⁺, Fig. 4F), and they extensively expressed Nestin (Fig. 4G and H). However, Alox5 deletion dramatically reduced GNP proliferation in Math1-Cre/ptch1^fl/fl/Alox5 null cerebella, leading to only a few ectopic lesions present in mutant cerebella (Fig. 4I and J). Both proliferation and Nestin expression in GNPs were markedly inhibited (Fig. 4K and L). Finally, tumor latency was significantly prolonged in Math1-Cre/ptch1^fl/fl/Alox5 null mice (median survival, 126 days) compared with that in Math1-Cre/ptch1^fl/fl mice (median survival, 58 days, P < 0.0001; Fig. 4M). These data indicate that deficiency in LT synthesis inhibited medulloblastoma growth by compromising tumor cell proliferation.

A small proportion of ectopic cells in Alox5 null cerebellum was still proliferative (Fig. 4K), which continued their proliferation and finally result in tumors with 100% penetrance (Supplementary Fig. S5A). Tumor cells developed from Alox5-null, ptch1-mutant mice, still expressed Nestin and were highly proliferative (Supplementary Fig. S5B–S5D). Moreover, expression of gli1 and ptch2 mRNAs was significantly up-regulated in tumor cells compared with GNPs (Supplementary Fig. S5E), suggesting that Shh pathway was still activated in medulloblastoma cells in Alox5-null ptch1-mutant mice.

We next examined whether antagonism of LT synthesis can be utilized to treat medulloblastoma. Medulloblastoma cells isolated from ptch1^+/− mice, were treated with DMSO, MK886, or together with LT or arachidonic acid (AA) for 48 hrs. As shown in Fig. 5A, more than 80% of medulloblastoma cells were found proliferative (Ki67⁺) in control group (DMSO treated). Majority of tumor cells highly expressed Nestin, as expected. MK886 significantly repressed Nestin expression in tumor cells and inhibited medulloblastoma cell proliferation (Fig. 5B), consistent with the critical role of Nestin in tumor cell proliferation (8). Exogenous LT effectively rescued Nestin expression as well as proliferation of tumor cells repressed by MK886 (Fig. 5C). Because synthesis of LT from AA still relies on Alox5 activity, AA failed to restore MK886-repressed Nestin expression or proliferation of medulloblastoma cells (Fig. 5D and E). Reduced expression of Nestin and compromised proliferation in medulloblastoma cells was also observed following treatment with zileuton (Supplementary Fig. S6A–S6E), a specific antagonist of Alox5 (29). The above data suggest that endogenous leukotriene synthesis is necessary for Nestin expression and proliferation of medulloblastoma cells.

Because inhibition of LT synthesis prevents medulloblastoma cell proliferation by blocking Nestin expression, we hypothesized that MK886 should not affect proliferation of wild type GNPs, which do not express Nestin (9, 30, 31). To test this idea, GNPs from wild type cerebella were treated with recombinant Shh protein in vitro to maintain proliferation (Fig. 5F; ref. 32). As shown in Fig. 5G, vismodegib inhibited proliferation of both medulloblastoma cells and wild-type GNPs as previously reported (33, 34). Although MK886 treatment dramatically inhibited medulloblastoma cell proliferation, the proliferation of wild-type GNPs was not affected (Fig. 5H and I). These results are consistent with lacks of phenotypes in GNPs proliferation in Alox5 null cerebella. The above data indicate that LT is required for the proliferation of medulloblastoma cells but not GNPs, and that inhibition of LT synthesis prevents proliferation in a tumor cell–specific manner.

Most of the known small-molecule inhibitors of the Shh pathway (SANT1, CUR61414, vismodegib, sonidegib, saridegib etc.) target Smo (34, 35). These inhibitors have displayed promising anti-tumor responses in medulloblastoma and basal cell carcinoma, in which the Shh pathway is activated by mutation (36, 37). However, the emergence of resistant Smo mutations, which can result in tumor relapse after an initial response, has emerged as a clinical concern (38). It has been reported that medulloblastoma cells overexpressing activated form of Smo (SmoM2) represent tumor cells that are resistant to Smo antagonists (18, 22). As expected, vismodegib was active against medulloblastoma cells derived from ptch1-deficient mice with an IC₅₀ of 63 nmol/L, whereas it failed to inhibit SmoM2 medulloblastoma cells (Fig. 5J). However, MK886 exhibited comparable anti-proliferative activity in ptch1-deficient medulloblastoma cells as well as SmoM2 medulloblastoma cells (Fig. 5J, IC₅₀ = 1.8 μmol/L in ptch1-deficient medulloblastoma cells and 2.0 μmol/L in SmoM2 medulloblastoma cells). In addition, inhibition of tumor cell proliferation by MK886 was also observed in medulloblastoma cells from ptch1+/−/p53 null mice (Supplementary Fig. S7, IC₅₀ = 1.7 μmol/L). These data suggest that inhibition of LT synthesis could be utilized to treat tumors with drug-resistant Smo mutations.

Finally, we examined whether MK886 could prohibit tumor growth in vivo. For this purpose, we utilized a subcutaneous allograft medulloblastoma model by transplanting medulloblastoma cells into flanks of CB17/SCID mice (39). After tumors were established (200–400 mm³), CB17/SCID mice were treated with intraperitoneal inoculation of MK886 or DMSO. As shown in Fig. 6A, no obvious inhibition of tumor growth was observed after treatment with 10 mg/kg MK886. However, 20 and 40 mg/kg MK886 significantly inhibited tumor growth, with 46% (P < 0.0001) and 72% (P < 0.0001) growth inhibition, respectively, compared with DMSO-treated tumors. As predicted, Nestin expression and medulloblastoma cell proliferation dramatically decreased in tumor tissue following MK886 treatment, compared with controls (Fig. 6B and C). MK886 treatment also significantly promoted neuronal differentiation (NeuN⁺) among transplanted medulloblastoma cells (Fig. 6D and E). These data demonstrated that MK886 treatment can inhibit medulloblastoma growth in vivo.

To exclude the possible off-target effects of MK886 in inhibiting medulloblastoma growth, we also generated subcutaneous medulloblastoma models by using tumor cells from Math1-Cre/ptch1^fl/fl/Alox5 null mice (at 14 weeks of age) as well as from Math1-Cre/ptch1^fl/fl mice (at 8 weeks of age). Once tumors were established, mice were treated with 40 mg/kg MK886 by oral gavage, once a day for 2 weeks. As shown in Supplementary Fig. S8, the growth of ptch1 mutant medulloblastoma cells was significantly repressed by MK886 treatment. However, Alox5-null, ptch1-mutant tumor cells continued to progress despite MK886 treatment. These data suggest that MK886 inhibited medulloblastoma growth by targeting Alox5.

Finally, we treated Math1-Cre/ptch1^fl/fl mice with MK886 or DMSO by oral gavage. Starting from 6 week of age, tumor bearing mice were treated with DMSO or 20 mg/kg/day MK886 for 2 weeks. As shown in Fig. 6F, MK886 dramatically reduced tumor volume compared with DMSO control. We examined Nestin expression in mutant cerebella by qPCR. As shown in Fig. 6G, much less Nestin mRNA expression was observed in the cerebella after MK886 treatment than after DMSO treatment, suggesting that MK886 repressed Nestin expression in tumor cells. The survival of ptch1 mutant mice was significantly prolonged after MK886 treatment compared with the control (Fig. 6H). Histological analysis revealed that massive proliferating tumor cells accumulated on cerebellar surface of Math1-Cre/ptch1^fl/fl after DMSO treatment; however, following MK886 treatment, the majority of tumor cells initiated differentiation (NeuN⁺) and migrated inward, resulting in fewer ectopic cells in the mutant cerebellum (Fig. 6I and J). Most medulloblastoma cells were actively proliferating (ki67⁺) with spontaneous differentiation (NeuN+) observed in tumor tissues after DMSO treatment (Fig. 6K). However, MK886 dramatically reduced tumor cell proliferation and promoted substantial differentiation in medulloblastoma tissue (Fig. 6L). No significantly increased apoptosis was found in mutant cerebella after MK886 treatment (Supplementary Fig. S9A–S9C). In addition, we treated Math1-Cre/R26R-SmoM2 (SmoM2) mice with DMSO or 20 mg/kg/day MK886 for 2 weeks. Tumor growth was significantly inhibited in SmoM2 mice following MK886 treatment (Supplementary Fig. S10A). Proliferation of tumor cells was markedly repressed by MK886, compared with the DMSO control (Supplementary Fig. S10B–S10F). These data further confirm the inhibitory effect of MK886 on medulloblastoma growth.

We previously reported that astrocytes promote medulloblastoma growth through Shh secretion, and astrocyte-derived Shh induced Nestin expression in tumor cells (16). Here, we demonstrate that LT, well-known inflammatory molecules, mediate Shh-induced Nestin expression in medulloblastoma cells. Collectively, these findings reveal a novel pathway in tumor cells, which is essential for medulloblastoma tumorigenesis. As summarized in Fig. 7, astrocyte-derived Shh stimulates LT biosynthesis in medulloblastoma cells, which induces Nestin expression in tumor cells upon Smo activation. Nestin protein promotes medulloblastoma cell proliferation by sustaining Hh pathway activation, as we previously demonstrated (8). These findings establish a novel function of LT in medulloblastoma tumorigenesis, and highlight the important role of tumor microenvironment in medulloblastoma progression.

In our studies, increased LT biosynthesis in tumor cells was observed within 10 minutes after Shh treatment. Moreover, no alterations in mRNA expression of Alox5 and FLAP were observed in medulloblastoma cells in the presence of Shh. These data indicate that Shh induces LT synthesis in a transcription-independent manner. The capacity of Shh to induce LT synthesis independently of transcription was also observed during cytoskeletal rearrangement and migration of fibroblasts and NIH3T3 cells (20, 21). Shh rapidly enhanced LT synthesis in ptch1-deficient medulloblastoma cells, implying that receptors other than ptch1 may react to extracellular Shh. Recent studies have shown that ptch2, a ptch1 paralog, can interact with Shh and regulate Shh signaling and functions such as chemotaxis (40) and tissue patterning (41). Moreover, ptch2 was found to exert tumor suppressive functions in medulloblastoma and basal cell carcinoma (42, 43). Further studies are warranted to investigate whether ptch2 is involved in Shh-induced LT biosynthesis in medulloblastoma cells. Consistent with our findings, a recent study demonstrated that Smo activation stimulates PLA2 to trigger the release of AA, a precursor of leukotriene, in mouse fibroblasts within 2 hours (44), providing a possible mechanism for Shh-induced LT synthesis in medulloblastoma cells.

Our studies demonstrated that LT promoted the proliferation of medulloblastoma cells through stimulating Nestin expression. However, it is noteworthy that all Alox5-null ptch1-mutant mice finally developed medulloblastoma, despite markedly prolonged tumor latency. Medulloblastoma cells still expressed Nestin protein and Shh pathway was activated in tumor cells, suggesting that medulloblastoma cells in Alox5-null cerebella may have developed a LT-independent mechanism to support tumor cell proliferation. It is also possible that a rare population of medulloblastoma cells has the intrinsic capacity to proliferate (and express Nestin) in the absence of LT.

LT stimulated extensive expression of Nestin mRNA and protein in ptch1-deficient GNPs. Moreover, MK886 or zileuton dramatically repressed Nestin expression in medulloblastoma cells in vitro and in vivo. These data demonstrate that LT induce Nestin expression in tumor cells. No significant difference in the capacity of inducing Nestin expression in ptch1-deficient GNPs was found among LTB4, C4, D4, and E4, which appears to be more potent than LT4 in stimulating Nestin expression (Supplementary Fig. S11). Vismodelgib effectively inhibited LT-induced Nestin expression in ptch1-deficient GNPs, indicating that Hh pathway activation is required for Nestin expression in GNPs upon LT treatment. However, combination of Shh and LT failed to induce Nestin expression in wild type GNPs (Supplementary Fig. S12). These data imply that ptch1 deficiency may be also indispensable for LT-induced Nestin expression.

It is generally believed that LT functions through interactions with two putative receptors, cysteinyl leukotriene type I (Cys-LT1R) and type II receptors (Cys-LT2R; ref. 45). No Cys-LT2R mRNA expression was detected in GNPs, medulloblastoma cells, macrophage or astrocytes in medulloblastoma tissue. Elevated expression of Cys-LTR1 mRNA expression was found in macrophages, compared with low levels of expression in GNPs, tumor cells and astrocytes (Supplementary Fig. S13A). Montelukast, an antagonist of Cys-LT1R (46), effectively represses LT signaling in the airway, providing a routine treatment for asthma and bronchoconstriction (47). However, Nestin expression and proliferation in medulloblastoma cells were not altered by montelukast treatment in our studies (Supplementary Fig. S13B and S13C). In addition, LT-induced Nestin expression in ptch1-deficient cells was not affected after knocking down Cys-LT1R and Cys-LT2R by shRNAs (data not shown). These data suggest that Cys-LT1R or Cys-LT2R may not participate LT-induced Nestin expression in medulloblastoma cells. No expression of leukotriene B4 receptor 1 (Ltb4r1) and receptor 2 (Ltb4r2) was observed in medulloblastoma cells by q-PCR (Supplementary Fig. S14). Previous studies have reported that leukotriene may function through the peroxisome proliferator activated receptors (PPAR), nuclear hormone receptors that act as ligand-activated transcription factors (48, 49). Among three PPAR subtypes (PPARα, PPARβ and PPARγ), PPARγ was demonstrated to induce the transcription of Nestin in neural stem cells, and to regulate their proliferation and differentiation (50, 51). It was previously reported that Shh signaling activation upregulates the expression of PPARγ in cerebellar GNPs, suggesting that PPARγ is a target gene of Shh signaling in these cells (52). The above evidence suggests that PPARγ may mediate leukotriene-triggered Nestin expression in Ptc-deficient GNPs. Future studies are warranted to further investigate the involvement of PPARs in leukotriene functions in medulloblastoma cells.

Our studies reveal for the first time a critical role for LT in medulloblastoma progression. It is intriguing that inhibition of LT synthesis by MK886 or zileuton prevented medulloblastoma cell proliferation and in vivo tumor growth, but did not affect proliferation of wild-type GNPs. In our studies, normal proliferation of GNPs was not affected after genetic deletion of Alox5 gene in Alox5 null mice. These data indicate that LT biosynthesis may represent a promising target for treatment of medulloblastoma. Importantly, with the recent FDA approval of vismodelgib for use in advanced basal cell carcinoma, and with multiple other Smo inhibitors in clinical trials, drug resistant Smo mutations are likely to become more prevalent. In the current study, inhibition of LT synthesis by MK886 prevented tumor cell proliferation independent of Smo. These data strongly support the clinical evaluation of LT antagonists in the treatment of de novo Shh pathway tumors and for patients who have acquired resistance to Smo inhibitors.

No potential conflicts of interest were disclosed.

Conception and design: F. Du, Y. Wang, J.M.Y. Ng, T. Curran, Z.-J. Yang

Development of methodology: F. Du, C. Zheng, K.Q. Cai, J.M.Y. Ng, P. Li, Z.-J. Yang

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): F. Du, L. Yuelling, E.H. Lee, C. Zheng, K.Q. Cai, J.M.Y. Ng, P. Li, Z.-J. Yang

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): F. Du, L. Yuelling, E.H. Lee, Y. Wang, S. Liao, L. Zhang, C. Zheng, S. Peri, K.Q. Cai, J.M.Y. Ng, T. Curran, P. Li, Z.-J. Yang

Writing, review, and/or revision of the manuscript: F. Du, L. Yuelling, Y. Wang, J.M.Y. Ng, Z.-J. Yang

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): F. Du, Y. Wang, Y. Cheng, J.M.Y. Ng, P. Li, Z.-J. Yang

Study supervision: Y. Wang, J.M.Y. Ng, Z.-J. Yang

We would like to thank A. Efimov, E. Nicolas, J. Pei, and J. Oesterling for technical assistance. This research was supported by the NCI (CA178380 and CA185504 to Z. Yang), ACS RSG (RSG1605301NEC to Z. Yang), PA CURE Health Research Fund (CURE 4100068716 to Z. Yang), and National Natural Science Foundation of China (81572724 and 81573449 to L. Zhang, 81703538 and BK20170348 to Y. Wang, 81472596 to P. Li).

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.