Neuroendocrine prostate cancer (NEPC) is an aggressive subtype of prostate cancer with poor prognosis, and there is a critical need for novel therapeutic approaches. NEPC is associated with molecular perturbation of several pathways, including amplification of MYCN. Anaplastic lymphoma kinase (ALK) is a receptor tyrosine kinase involved in the pathogenesis of neuroblastoma and other malignancies where it cooperates with N-Myc. We previously identified the first case of ALK F1174C-activating mutation in a patient with de novo NEPC who responded to the ALK inhibitor, alectinib. Here, we show that coactivation of ALK and N-Myc (ALK F1174C/N-Myc) is sufficient to transform mouse prostate basal stem cells into aggressive prostate cancer with neuroendocrine differentiation in a tissue recombination model. A novel gene signature from the ALK F1174C/N-Myc tumors was associated with poor outcome in multiple human prostate cancer datasets. ALK F1174C and ALK F1174C/N-Myc tumors displayed activation of the Wnt/β-catenin signaling pathway. Chemical and genetic ALK inhibition suppressed Wnt/β-catenin signaling and tumor growth in vitro in NEPC and neuroblastoma cells. ALK inhibition cooperated with Wnt inhibition to suppress NEPC and neuroblastoma proliferation in vitro and tumor growth and metastasis in vivo. These findings point to a role for ALK signaling in NEPC and the potential of cotargeting the ALK and Wnt/β-catenin pathways in ALK-driven tumors. Activated ALK and N-Myc are well known drivers in neuroblastoma development, suggesting potential similarities and opportunities to elucidate mechanisms and therapeutic targets in NEPC and vice versa.
These findings demonstrate that coactivation of ALK and N-Myc induces NEPC by stimulating the Wnt/β-catenin pathway, which can be targeted therapeutically.
Patients with advanced prostate cancer commonly benefit from agents targeting the androgen receptor (AR) signaling pathway; however, therapeutic resistance commonly develops. An emerging mechanism of resistance involves the cooption of signaling pathways associated with neuroendocrine differentiation by prostate tumors (1, 2). Unlike de novo neuroendocrine prostate cancer (NEPC), which is rare, up to one-quarter of patients with castration-resistant prostate cancer (CRPC) may go on to develop treatment-related NEPC (2). Several drivers of neuroendocrine differentiation in prostate cancer have been described, including MYCN (encoding N-Myc), which is overexpressed or amplified in up to 40% of NEPCs (3, 4). Other potential drivers include AURKA, EZH2, BRN2, RB1/TP53, SOX2, SRRM4, REST, FOXA2, and ONECUT2 (5–12). A common theme among many of these drivers is their involvement in neuroblastoma and/or neuronal development, suggesting that under therapeutic pressure, prostate cancer cells may preferentially coopt these genes and pathways for survival.
Anaplastic lymphoma kinase (ALK) is a receptor tyrosine kinase that has an established role in the pathogenesis of many malignancies, including neuroblastoma and non–small cell lung cancer, but little is known about its role in prostate cancer (13). Common oncogenic ALK mutations at F1174 and R1275 induce ligand-independent autophosphorylation leading to constitutive activation of ALK and are mainly found in neuroblastoma (13). Notably, ALK and N-Myc cooperate in the development of neuroblastoma (14). We reported previously the first case of a patient with de novo NEPC with an activating ALK F1174C mutation who responded to the ALK inhibitor, alectinib (15). However, the functional role of ALK activation in the development of NEPC has not been defined. Furthermore, although ALK mutations are rare in prostate cancer, we hypothesized that activation of the ALK pathway (e.g., by ALK amplification/overexpression) may occur more commonly and play a role in driving advanced prostate cancer. In this study, we examined the functional role of activation of the ALK pathway in the context of N-Myc overexpression in prostate cancer and the downstream signaling pathways. We show that activated ALK cooperates with N-Myc to drive aggressive prostate cancer with neuroendocrine differentiation and defined a prostate ALK activation signature that is associated with poor outcome in multiple datasets. Analysis of downstream signaling pathways identified a role for Wnt/β-catenin signaling in ALK-N-Myc–driven prostate cancer and neuroblastoma cells.
Materials and Methods
Lentiviral vectors and preparation
FM1 plasmid was kindly provided by Dr. Jeffrey Milbrandt (Washington University in St. Louis, St. Louis, MO). FM1-ALK WT-YFP, FM1-ALK F1174C-YFP, FM1-MYCN-YFP, and FM1-MYCN-mCherry plasmids were constructed in this study. To calculate lentivirus titers, we infected HEK293T cell line with lentivirus carrying yellow fluorescence protein (YFP). Titer is expressed as transducing units/mL calculated from YFP-positive cells (%) measured by flow cytometer.
For details, see Supplementary Data.
FVB/NJ mice and NSG mice were obtained from The Jackson Laboratory, SCID mice from Envigo, and nude mice from Charles River Laboratories. All animal experiments and procedures were performed in compliance with ethical regulations and the approval of the Northwestern University (Chicago, IL) Institutional Animal Care and Use Committee.
PC3, LNCaP, DU145, 22Rv1, VCaP, and SH-SY5Y cell lines were purchased from the ATCC. KELLY cell line was purchased from Sigma-Aldrich. PC3-MYCN, LNCaP-ALK WT, LNCaP-YFP (control), LNCaP-ALK WT-dCas9, LNCaP-ALK WT-sgFAM150B, LNCaP-ALK WT-sgPTN, and LNCaP-ALK WT-sgNC cells were generated in this study. All cells were used within 10 passages of thawing and verified as Mycoplasma free. PC3, LNCaP, DU145, 22Rv1, VCaP, and SH-SY5Y cell lines were genetically authenticated by the ATCC. KELLY cell line was genetically authenticated by Public Health England.
For details, see Supplementary Data.
Tissue recombination and lentivirus transductions
Briefly, rat urogenital sinus mesenchyme (UGM) cells were dissociated from embryonic day 18 embryos. To generate ALK WT/N-Myc (A+N), ALK F1174C/N-Myc (FC+N), ALK WT (A), ALK F1174C, N-Myc (N), and YFP control grafts, prostate basal epithelial stem cells isolated from 8–10 weeks old FVB/NJ mice were transduced with ALK WT-, ALK F1174C-, MYCN-, and YFP ctrl-lentiviral particles alone or in combination. The recombinants were transplanted under the renal capsule of male SCID mice. Grafts were harvested at 7 weeks after transplant.
For details, see Supplementary Data.
Alectinib and Wnt inhibitor treatments for cell lines in 2D culture
Cell lines were seeded to 96-well plate culture for viability assay. After 24 hours, cells were treated with alectinib and Wnt inhibitors (ICG-001, XAV-939, or LGK-974) alone or in combination for 96 hours. Viability was measured by CellTiter 96 AQ One Solution Cell Proliferation Assay (Promega, G3581). For Western blotting and immunofluorescence staining, 50%–60% confluent cells were treated with 1 or 5 μmol/L alectinib or DMSO for 48 hours, followed by protein extraction and staining.
For details, see Supplementary Data.
Alectinib and Wnt inhibitor treatments for organoids
Grafts were digested to single cells. For FC+N and control organoids, YFP- and/or EpCAM- (for control) positive cells were sorted by flow cytometer. Organoid culture was started as described in detail previously (16). At day 7, organoids were treated with 5 μmol/L alectinib, 5 μmol/L LGK-974, 5 μmol/L XAV-939, 5 μmol/L ICG-001, or DMSO for 7 days. Representative bright-field images were taken and viability was measured by CellTiter-Glo 3D Cell Viability Assay (Promega, catalog no. #G9682). For Western blotting, organoids were treated with 5 μmol/L alectinib or DMSO for 48 hours, followed by protein extraction.
For details, see Supplementary Data.
Whole-transcriptome sequencing analysis
For RNA sequencing (RNA-seq), RNAs were extracted from small fragments of grafts. cDNA libraries were prepared from isolated RNA using the Illumina TruSeq Stranded Total RNA LT Sample Prep Kit (Illumina). High-throughput sequencing with 75 bp paired-end reads was performed using an Illumina HiSeq 4000. High quality reads were aligned to the Mus musculus genome (mm10) using STAR (17). Read counts for each gene were calculated using HTSeq-Counts (18) in conjunction with a gene annotation file for mm10 obtained from University of California Santa Cruz (Santa Cruz, CA; http://genome.ucsc.edu). Differential expression was determined using DESeq2 (19). mRNA expression patterns of FC+N, ALK F1174C (FC), A+N, A, and N grafts were compared with those of control grafts. Gene set enrichment analysis (GSEA) of genes with differential expression was performed using GSEA Software from Broad Institute (20).
Scoring of prostate ALK signature for various prostate cancers
Significantly differentially expressed genes in FC+N grafts compared with control grafts, specifically, top 25 upregulated genes and top 25 downregulated genes, were selected as our prostate ALK gene signature. Prostate ALK signature genes are shown in Fig. 2B. For rank-sum scoring, genes were sorted in descending order by log2 fold change. Genes were ranked according to sorted order. Lowest log2 fold change was considered as 1 and highest log2 fold change was considered as number that was examined total gene number. For upregulated genes, ranked numbers of prostate ALK signature genes were used for sum, while for downregulated genes, ranked numbers were subtracted from highest ranked number. All ranked numbers for upregulated genes and all subtracted numbers for downregulated genes were summed to obtain sample rank sum, termed prostate ALK signature score. To assess prostate ALK signature score of 1,037 cell lines in the Cancer Cell Line Encyclopedia (CCLE), gene expression profiles were obtained from published datasets (21) and gene signature score was determined by rank-sum scoring. For the genes tested in each cohort, gene ranks were randomly shuffled 10,000 times to create a background distribution. Percentile ranks are reported as the percentile of a given sample within the set of 10,000 random mock samples to allow for comparison across cohorts.
Chromatin immunoprecipitation sequencing (ChIP-seq) tracks of N-MYC in KELLY and BE2C cells (GSM2113521 and GSM2113526) were downloaded from GSE80151.
In vivo drug treatments
FC+N tumors at passage 1 and 2 were minced to 1.5 mm × 1.5 mm of small pieces. Three tumor pieces were mixed with UGM. Mixture of tumor pieces and UGM was resuspended in 25 μL of 3:1 collagen/setting buffer solution. The recombinants were transplanted under the renal capsule of male NSG mice. At day 5 after transplant at passage 1, alectinib was administered orally at 60 mg/kg and ICG-001 was administered intraperitoneally at 50 mg/kg for 8 days daily. Vehicles of alectinib and ICG-001 were administered orally and intraperitoneally, respectively, daily for 8 days. Grafts were harvested at day 13 after transplant. At day 7 after transplant at passage 2, alectinib was administered orally at 20 mg/kg and ICG-001 was administered intraperitoneally at 20 mg/kg daily alone or in combination. Drugs and vehicles were administered for 3 days. Because of ascitic fluid accumulation, grafts were harvested at day 10 after transplant.
A total of 1 × 106 PC3 cells mixed with 100 μL Matrigel were subcutaneously injected into flanks of male nude mice (6–8 weeks old). When tumor volume reached an average of 150 mm3 for each group (alectinib and vehicle), mice were orally administered with alectinib at 20 mg/kg or vehicle daily for 35 days. Tumors were harvested at day 35 after drug treatment.
For details, see Supplementary Data.
ALK and MYCN alterations significantly co-occur in prostate cancer datasets
We assessed alterations (amplifications, overexpression, and genetic mutations) in ALK, MYCN, and AURKA in prostate cancer by interrogating published data (22, 23). Alterations of ALK were observed in 15.6% (3.28%–40%) of patients with prostate cancer, compared with 17.5% (1.4%–53.3%) for MYCN and 28.1% (3.5%–51.6%) for AURKA (Supplementary Fig. S1A–S1I; Supplementary Tables S1 and S2). The highest alteration frequency of ALK and MYCN was observed in patients with NEPC (Supplementary Fig. S1A–S1I). Furthermore, alterations in ALK and MYCN significantly cooccurred in six of nine datasets, while cooccurrence of ALK with AURKA and MYCN with AURKA was observed in three of nine datasets (Supplementary Table S1). These results may be explained by the chromosomal localization of ALK on 2p23.2 close to MYCN at 2p24.3, as well as potential cooperativity between ALK and N-Myc in promoting prostate cancer development, similar to what was observed in neuroblastoma (14).
Coexpression of activated ALK F1174C and N-Myc in mouse prostate stem cells induces metastatic prostate cancer with neuroendocrine differentiation
Using lentiviral-mediated gene transfer and a tissue recombination model, we investigated the impact of expression of wild-type (WT) or activated ALK and N-Myc alone or in combination on prostate cancer development. Mouse prostate basal stem epithelial cells were transduced with lentiviral vectors expressing ALK, ALK F1174C, or MYCN coexpressed with YFP. Transduced basal stem cells were then combined with rat UGM and grafted under the renal capsule of immunodeficient mice (Fig. 1A). Assessment of graft size showed an increase in the size of FC+N grafts and N grafts compared with other grafts (Fig. 1B; Supplementary Fig. S2A and S2B). Histopathologic review of grafts identified all seven FC+N grafts to be cancerous, of which, six consisted of NEPC (one case of small-cell carcinoma, two cases of poorly differentiated adenocarcinoma with neuroendocrine differentiation, and three cases of prostate adenocarcinoma with neuroendocrine differentiation) and one graft was indicative of a cancerous prostate tissue, the subtype of which could not be defined (Fig. 1C and D; Supplementary Fig. S3; Supplementary Table S3). FC+N tumors were highly positive for Ki-67, and negative for luminal marker, cytokeratin 8 (CK8), and basal cell marker, p63 (Fig. 1E). ALK WT, ALK F1174C, and N-Myc protein levels were examined by Western blotting. Total ALK F1174C and phospho-ALK F1174C proteins were detected in FC+N tumors and FC grafts. N-Myc was detected at high levels in FC+N tumors (Fig. 1F). By RNA-seq data of the tissue recombinant grafts (see Materials and Methods), we found that among potential ALK ligands (13, 24) Fam150b expression was significantly increased in only FC+N tumor, while pleiotrophin (Ptn) was increased in FC+N and FC grafts (Fig. 1G). We could not see any upregulation of ALK ligands in A, N, and A+N grafts. In FC+N tumors, the neuroendocrine markers synaptophysin (Syp), neuron-specific enolase, and chromogranin A (Chga), as well as the AR were heterogeneously expressed (Supplementary Fig. S4A and S4B; Supplementary Table S3). Excluding the FC+N grafts, one A+N graft showed evidence of HGPIN and all the other grafts were benign (Supplementary Table S3). This was not due to lack of lentiviral-mediated expression of transgenes in the grafts, as confirmed by analysis of regions that expressed YFP in the grafts (Supplementary Fig. S3).
N-Myc overexpression is known to induce polycomb repressive complex 2 signaling and inhibit AR signaling in NEPC (5). Aurora kinase A is known to stabilize N-Myc (4). By RNA-seq data, we found upregulation of Aurka, Ezh2, and Suz12, and downregulation of Ar and AR target genes in FC+N tumors relative to control grafts (Supplementary Fig. S4C). Examination of the border regions of FC+N grafts by hematoxylin and eosin (H&E), Ki-67, and YFP staining showed evidence of local invasion into the kidney parenchyma, as well as the liver, pancreas, and spleen (Supplementary Figs. S4D and S5A–S5D), while we did not observe local invasion in other grafts. We also observed metastatic lesions in the lymph node and lungs, as confirmed by YFP staining (Fig. 1H). Together, these results demonstrate cooperativity between ALK F1174C and N-Myc in driving metastatic small-cell/neuroendocrine differentiation of prostate cancer.
We surmise that WT ALK overexpression failed to induce cancer because of the lack of ligand in the renal graft microenvironment. Accordingly, we found that of ALK-positive samples in primary tumors, 80.7% were potential ALK ligand positive (Supplementary Table S4). Similarly, of ALK-positive samples in the metastatic setting, 90.0% were potential ALK ligand positive (Supplementary Table S5). These data demonstrate that there is an abundance of ALK ligand in a majority of both primary and metastatic ALK-positive tumors. We also confirmed that well-known ALK activator, heparin (25), and potential ALK ligands, FAM150B and PTN, activate ALK signaling in WT ALK-overexpressing prostate cancer cells (Supplementary Fig. S6A–S6C). These data support that WT ALK in prostate cancer cells may be activated in the presence of the ALK ligands in tumors.
FC+N tumor organoids are sensitive to ALK tyrosine kinase inhibitor, alectinib
To evaluate the sensitivity of FC+N-driven cancer to ALK inhibitor, alectinib, we generated prostate cancer organoids from FC+N tumor grafts and control YFP-expressing prostate organoids from control grafts. We confirmed that FC+N organoids had positive staining for NEPC marker, Chga (Fig. 1I). Alectinib significantly decreased viability and reduced the size of FC+N organoids, but not the control organoids (Fig. 1J–L).
FC+N–derived prostate ALK signature is associated with poor outcome
We performed RNA-seq analysis and compared gene expression profiles of FC+N, A+N, FC, A, N, and YFP control grafts. FC+N grafts showed the largest number of significantly differentially expressed genes (3,451). In FC grafts, 1,230 genes were significantly altered, of which, 69% overlapped with FC+N–altered genes (Fig. 2A). The A, N, and A+N grafts showed limited changes in gene expression consistent with their mild pathology. To explore the relevance of the gene expression profile of ALK activation in prostate tissues, we developed a “prostate ALK gene signature.” The signature was comprised of the top 25 upregulated and top 25 downregulated genes in FC+N tumors. Within our tumor grafts, FC grafts showed the closest similarity to the FC+N grafts by this signature (Fig. 2B), supporting the notion that N-Myc amplifies the activated ALK program. To consider the relevance of this transcriptional program more broadly, we applied rank-sum scoring to assign a signature score to tumors in various cell line and tumor datasets (see Materials and Methods). We first compared the prostate ALK signature with a published ALK signature derived from an activated ALK/N-Myc transgenic model of neuroblastoma (26) and found no significant overlap (Supplementary Fig. S7). Nevertheless, among 1,037 cell lines of the CCLE, all three neuroblastoma cell lines carrying the ALK F1174L mutant showed higher prostate ALK signature score (Supplementary Fig. S8). In general, autonomic ganglia tissue–derived cell lines, including neuroblastoma, had the highest average signature score compared with cell lines derived from other tissues (Supplementary Fig. S8A). Interestingly, we found a correlation between prostate ALK signature score and alectinib sensitivity measured by cell proliferation assays of prostate cancer cell lines within the CCLE. We found that AR-negative cell lines, DU145 and PC3, which moderately express ALK (15), but not MYCN (21), are more sensitive to alectinib compared with AR-positive LNCaP, 22RV1, and VCaP cell lines (Fig. 2C). The lung cancer cell lines, NCI-H2228 and A549, which both express EML4-ALK fusion protein, showed moderate levels of the prostate ALK signature score at 67.02 and 50.2, respectively (Supplementary Fig. S8B).
Using this prostate ALK signature, we scored prostate cancer specimens grouped by Gleason score (6, 7, 8, 9, and 10; refs. 22, 23), tissue type (benign, primary prostate cancer, and metastatic prostate cancer; refs. 27, 28), or CRPC adenocarcinoma (CRPC-adeno) versus CRPC neuroendocrine differentiation (NEPC; ref. 29). We found that the prostate ALK signature significantly stratified patient specimens by Gleason score, primary versus metastasis, and CRPC versus NEPC (Fig. 3A–D; Supplementary Fig. S9A and S9B). In contrast, the AR signature only stratified the primary versus metastasis and the CRPC versus NEPC samples, but with reduced AR signature score in NEPC as expected (Fig. 3E–H).
The prostate ALK signature is also distinct from the NEPC signature described by Beltran and colleagues (29) as there was no overlap between the two. Further analysis of the Beltran and colleagues' dataset (29), which contains both CRPC and NEPC, revealed the distinction between the prostate ALK signature and the NEPC signature. While the NEPC signature could not separate CRPC samples, the ALK signature separated the CRPC samples into high ALK score and low ALK score, with both exhibiting low MYCN expression (Fig. 3I; ref. 29). The ALK signature was also negatively correlated to the AR signature score (Fig. 3J). Notably, in primary prostate tumors, the prostate ALK signature was predictive of disease-free survival, unlike the NEPC signature, which was not (Fig. 3K and L). In addition, we scored patient-derived xenograft (PDX) and organoid (PDO) CRPC-adeno and CRPC-NE (neuroendocrine prostate cancer) samples using the prostate ALK signature (30, 31). We found that all CRPC-NE samples showed higher signature score, as well as higher ALK and MYCN mRNA expression than CRPC-adeno samples (Supplementary Fig. S10A–S10C). While a limited number of samples were available to examine the IC50 values of alectinib, we found a correlation trend between prostate ALK signature score and alectinib sensitivity in PDX samples grown in 3D organoid and 2D cultures (Supplementary Fig. S10A).
ALK-activated prostate tumors show evidence of Wnt/β-catenin pathway activation
GSEA of RNA-seq data from tissue recombinant grafts revealed evidence of activation of the Wnt/β-catenin signaling in FC+N and FC grafts compared with control grafts (Fig. 4A and B; Supplementary Table S6). Activation of the Wnt signaling pathway has been implicated in therapy-resistant prostate cancer, including NEPC (32). A hallmark of the Wnt/β-catenin signaling pathway is the stabilization of the transcriptional coactivator, β-catenin, resulting in its nuclear accumulation (33). Accordingly, mRNA of Ctnnb1 (encoding β-catenin) and its total and active protein levels were significantly increased in FC and FC+N grafts (Fig. 4C and D). Notably, FC-expressing cells showed evidence of elevated levels of membrane, cytoplasmic, and nuclear β-catenin, compared with control prostate grafts (Fig. 4E). FC+N grafts showed strong nuclear staining and weak cytoplasmic staining for β-catenin, suggesting active Wnt signaling (Fig. 4E). Lymphoid enhancer factor 1 (Lef1) transcription factor is the major endpoint mediator of Wnt/β-catenin signaling pathway (33). We observed significantly elevated enrichment of Lef1 target genes in FC+N and FC grafts by GSEA (Fig. 4F).
Among components of the Wnt/β-catenin signaling pathway, in addition to Ctnnb1, we found significant elevation of Csnk1e (encoding casein kinase 1, epsilon) mRNA and protein levels in FC and FC+N grafts (Fig. 4G and H). It is known that Csnk1e phosphorylates dishevelled2 (Dvl2) on serine residue 143 (S143), which is a key mediator of Wnt signaling (34–36). Phosphorylation of Dvl2 is promoted by binding of DEAD-box RNA helicase Ddx3x to Csnk1e for positive regulation of Wnt/β-catenin signaling (37). Subsequent to its phosphorylation, DVL2 polymerization promotes high Wnt pathway activity (38). Our RNA-seq data showed elevated expression of Ddx3x in FC+N tumors (Fig. 4I). Accordingly, phosphorylated S143-Dvl2 and Dvl2 polymerization was increased in FC+N tumors (Fig. 4J). Likewise, activation of WT ALK by ALK ligand increased phosphorylated S143-Dvl2 and Dvl2 polymerization in prostate cancer cells (Supplementary Fig. S6A–S6C). ChIP-seq studies in neuroblastoma cell lines (39) have indicated that N-Myc could bind to the promoter regions of DDX3X and CTNNB1 (Fig. 4K). We determined that N-Myc binds to the promoter regions of these genes in PC3-overexpressing MYCN cell lines (PC3-MYCN; Fig. 4L). In sum, these results suggest that cooperation between ALK F1174C and N-Myc induces Ddx3x and Ctnnb1 expression, resulting in higher Wnt/β-catenin pathway activity through phospho-Dvl2 polymerization.
We next investigated whether active ALK signaling is required for Wnt pathway activation by using the ALK inhibitor, alectinib, in FC+N organoid and PC3 cell line. Treatment of established FC+N tumor organoids with alectinib led to reduction in ALK, Csnk1e, phospho-S143-Dvl2, polymerized-Dvl2, and total β-catenin (Fig. 5A–D). Likewise, alectinib treatment reduced ALK, phospho-S143-Dvl2, polymerized-Dvl2, and total β-catenin proteins in PC3 cells (Fig. 5E). In addition, alectinib significantly decreased nuclear localization of β-catenin in PC3 cells (Fig. 5F and G).
ALK inhibition abrogates neuroblastoma cell growth through inhibition of Wnt/β-catenin signaling
To further elucidate the role of Wnt/β-catenin pathway and ALK signaling, we examined two neuroblastoma cell lines as ALK alteration is more common: SH-SY5Y and KELLY. SH-SY5Y and KELLY cell lines harbor the activating ALK mutation named ALK F1174L. The cell lines were treated with ALK inhibitor (alectinib), Wnt inhibitors (ICG-001, XAV-939, or LGK-974), or a combination of alectinib and Wnt inhibitors. Alectinib and Wnt inhibitors significantly decreased viability of KELLY and SH-SY5Y cells, while Wnt inhibitors showed additive effect in combination with alectinib (Fig. 5H and I). KELLY and SH-SY5Y cells showed higher prostate ALK signature score and higher sensitivity to alectinib, while SK-N-FI and SK-N-AS cells harboring WT ALK and MYCN exhibited the lowest signature scores (Fig. 5J), suggesting that common ALK F1174L mutation activates downstream of ALK signaling in neuroblastoma. We, therefore, investigated whether ALK inhibition affects Wnt/β-catenin signaling in SH-SY5Y and KELLY cells. Alectinib treatment reduced the phospho-ALK and phospho-DVL2 levels in these cell lines, while ALK and total β-catenin protein levels were decreased in SH-SY5Y cells (Fig. 5K). siRNA knockdown of ALK reduced phospho-ALK and phosho-DVL2 protein levels in KELLY cells (Fig. 5L). Increased expression of the Axin2 gene is often used as an indicator of canonical Wnt pathway activity (40). Therefore, we compared AXIN2 mRNA expression between ALK-altered (ALK F1174L mutation, amplification, and high mRNA expression) and ALK WT/MYCN (WT) neuroblastoma cells. ALK-altered neuroblastoma cells had significantly increased AXIN2 expression compared with ALK/MYCN (WT/WT; Fig. 5M). We confirmed that activated ALK induces Wnt/β-catenin signaling and ALK inhibition decreases cell viability through repression of Wnt/β-catenin signaling in neuroblastoma cells.
Inhibition of Wnt/β-catenin signaling disrupts FC+N organoid formation
We next examined the effect of Wnt pathway inhibitors (LGK-974, XAV-939, and ICG-001) on activated Alk+N-Myc–driven prostate cancers. XAV-939 and ICG-001 significantly decreased viability of FC+N organoids (Fig. 6A). All three inhibitors disrupted FC+N organoid formation, while control organoids remained intact after drug treatment (Fig. 6B and C). ICG-001 decreased the viability of PC3 cells with IC50 value of 3.41 μmol/L and showed additive effect in combination with alectinib (Supplementary Fig. S11).
Alectinib and Wnt pathway inhibitor, ICG-001, inhibits FC+N tumor metastasis
We next explored the effect of ALK and Wnt inhibitors in vivo. We generated FC+N tumor grafts under the renal capsule of NSG mice. In the first trial using passage 1 tumor grafts, mice were treated with alectinib (60 mg/kg/day, orally) and ICG-001 (50 mg/kg/day, i.p.) for 8 days. The combination effectively inhibited tumor growth, invasion, and metastasis (Fig. 6D and E). Next, we assessed the effect of alectinib (20 mg/kg/day, orally) and IGC-001 (20 mg/kg/day, i.p.) alone or in combination on FC+N tumor growth, invasion, and metastasis, 7 days after transplantation of passage 2 tumor grafts. The passage 2 FC+N tumors grew aggressively with the control mice developing ascites, local invasion, and metastasis of tumor cells. Alectinib and ICG-001 significantly suppressed tumors growth (Fig. 6F), with alectinib and the combination also abrogating invasion/metastasis (Fig. 6G). We also found that Wnt pathway inhibitor, ICG-001, decreased Syp- and Ki-67–positive cells and increased cleaved caspase-3–positive cells in FC+N tumors (Fig. 6H–J). These data indicate that Wnt/β-catenin signaling contributes to neuroendocrine features and tumor survival. While ALK signaling activation was retained in metastatic FC+N tumors, alectinib/ICG-001 treatments abrogated ALK activation and metastasis (Fig. 6G and K), indicating the efficacy against metastatic tumors.
Finally, we evaluated the effect of alectinib treatment on PC3 cells, which showed a high prostate ALK signature score and high ALK expression (15) without oncogenic ALK mutation, injected subcutaneously into nude mice. When tumor volumes reached an average of 150 mm3, 20 mg/kg/day of alectinib or vehicle was administered orally. Compared with the control mice, alectinib significantly inhibited PC3 tumor growth (Fig. 6L and M). Mice body weights were not decreased, showing no toxicity of all drug treatments in these studies (Supplementary Fig. S12A–S12C).
ALK, like N-MYC, plays an established role in the development of neuroblastoma, but until recently, it was not implicated in prostate cancer. We previously reported a patient with de novo NEPC with activating ALK F1174C mutation that responded to alectinib therapy (15). In this study, we showed that activated ALK F1174C and N-Myc cooperate to transform normal prostate epithelial cells to prostate cancer with neuroendocrine differentiation. Furthermore, ALK F1174C and N-Myc overexpression induces activation of the Wnt/β-catenin pathway, resulting in prostate tumors that are sensitive to both ALK and Wnt pathway inhibitors in in vitro and in vivo models. Previous studies have shown that N-Myc can cooperate with constitutively active Akt or Pten deletion to initiate prostate adenocarcinoma and NEPC development from human or mouse prostate epithelial cells (4, 5). Therefore, our results introduce ALK as a novel factor that can cooperate with N-Myc to promote the development of NEPC from normal prostate epithelial cells. While a subtype of neuroblastoma with ALK F1174L and MYCN overexpression has been studied (41–43), a specific role for Wnt/β-catenin signaling was not examined. Notably, our results indicate that ALK inhibition abrogates Wnt/β-catenin signaling in both NEPC and neuroblastoma.
While prostate cancer datasets showed that levels of amplification and mRNA overexpression of ALK and MYCN genes significantly cooccurred across patient samples, combination of WT ALK and MYCN overexpression did not induce transformation into prostate cancer in our studies. Because WT ALK is a ligand-dependent tyrosine kinase, it will be expected to signal robustly only in the presence of ligand as we showed in vitro, unlike oncogenically activated F1174 mutant. We hypothesize that failure of WT ALK grafts to produce tumors from normal prostate epithelial cells was due to the lack of suitable ALK ligands in the renal graft microenvironment in our studies. Indeed, we could not see any upregulation of ALK ligands in WT ALK, N-Myc, and A+N grafts compared with control graft, while FC and FC+N grafts upregulated Ptn, and Ptn and Fam150b, respectively. Because Fam150b is known to drive superactivation of activated ALK mutants from neuroblastoma (24), FC+N tumors seem to potently activate ALK signaling by positive feedback regulation through activated ALK and Fam150b. In addition, in vivo study using PC3 cell line showed that PTN knockdown suppressed tumor growth and angiogenesis and induced apoptosis (44). These results suggest that FC+N tumors secrete ALK ligands, followed by superactivation of ALK signaling, while A+N and WT ALK grafts do not produce ALK ligands and the amount of functional ALK ligands is not sufficient for WT ALK activation in the mouse. RNA-seq of the grafts led us to two critical findings. First, a prostate ALK signature was developed that may indicate ALK activity and is associated with higher grade, metastasis, and reduced disease-free survival in patients with prostate cancer. This signature correlated with alectinib sensitivity in prostate cancer cell lines and organoids. Second, the signaling pathways induced by activated ALK in prostate cancer and neuroblastoma could be used to identify possible therapeutic targets. GSEA results suggested that activated ALK F1174C induced the Wnt signaling pathway, which has been implicated in prostate tumorigenesis and therapeutic resistance. The Wnt signaling pathway has been associated with resistance to pan-receptor tyrosine kinase inhibitor, dovitinib, through neuroendocrine differentiation of prostate cancer and plays a key role in resistance to enzalutamide in CRPC (32, 33, 45). In our study, activated ALK cooperated with N-Myc to induce Wnt/β-catenin signaling activity that promoted the development of NEPC. Our model indicates that active ALK cooperates with N-Myc to activate Wnt/β-catenin signaling by upregulation of Csnk1e and Ddx3x, which promote Dvl2 phosphorylation and polymerization, leading to β-catenin activation (Fig. 7). We assessed the effects of multiple Wnt/β-catenin signaling inhibitors on growth of FC+N tumors and organoids and ALK F1174L-expressed/MYCN-amplified neuroblastoma cells. Wnt inhibitors alone or in combination with ALK inhibition impaired FC+N-induced tumor development and decreased neuroblastoma cell growth, suggesting an approach for the treatment of patients with NEPC and neuroblastoma with ALK/N-MYC activation. XAV-939 and ICG-001, which inhibit the function of β-catenin, significantly decreased viability of FC+N organoids. LGK-974, which blocks the secretion of Wnt ligands, had no effect on viability. Because we did not find any significant upregulation of all Wnt ligand genes in FC+N tumor from our RNA-seq data, Wnt ligands may be not essential. Instead, activated ALK plays an essential role in upstream factor for Wnt signaling in FC+N tumor.
Our in vivo drug treatment study showed that FC/N tumors and PC3 cell xenografts, which showed higher prostate ALK signature score, were sensitive to alectinib. These data suggest that alectinib may have a role in patients whose tumors express high ALK signature score. Wnt pathway inhibitor, ICG-001, decreased Syp- and Ki-67–positive cells and increased cleaved caspase-3–positive cells. A recent study demonstrated that a subset of MYC/p53-driven tumors activates Wnt/β-catenin signaling in mice and Wnt pathway activations are associated with metastatic disease in patients with advanced prostate cancer that is sensitive to Wnt inhibitor (46). Our study showed that oncogenic ALK activates Wnt/β-catenin signaling pathway. Accordingly, ALK inhibitor, alectinib, suppressed metastasis and cooperated with Wnt pathway inhibitor, ICG-001.
In summary, our work shows a causal role for ALK F1174C and MYCN in the development of NEPC, defines a prostate ALK signature that correlates with advanced prostate cancer destined toward NEPC differentiation, and identifies Wnt signaling as a targetable pathway activated by ALK F1174C and MYCN.
Z.R. Chalmers reports grants from NCI during the conduct of the study. H. Beltran reports grants and personal fees from Janssen, personal fees from Pfizer, Blue Earth, AstraZeneca, Sanofi Genzyme, and Merck, and grants from AbbVie, Astellas, and Millenium outside the submitted work. S.A. Abdulkadir reports grants from NCI during the conduct of the study. No disclosures were reported by the other authors.
K. Unno: Conceptualization, data curation, formal analysis, validation, investigation, methodology, writing–original draft, project administration, writing–review and editing. Z.R. Chalmers: Data curation, formal analysis, funding acquisition, validation, methodology. S. Pamarthy: Data curation, formal analysis, validation, investigation. R. Vatapalli: Data curation, formal analysis, validation, investigation. Y. Rodriguez: Data curation, formal analysis, validation, investigation. B. Lysy: Investigation. H. Mok: Investigation. V. Sagar: Investigation. H. Han: Formal analysis, investigation. Y.A. Yoo: Investigation. S.-Y. Ku: Formal analysis, investigation. H. Beltran: Resources, formal analysis, investigation. Y. Zhao: Formal analysis, validation, investigation. S.A. Abdulkadir: Conceptualization, resources, formal analysis, supervision, funding acquisition, validation, methodology, writing–original draft, project administration, writing–review and editing.
The authors thank members of the Abdulkadir Laboratory for valuable discussion, Dr. Matthew J. Schipma and Priyam Patel at the NUSeq Facility for RNA sequencing data analysis, the Flow Cytometry Core for help with flow cytometry analysis, the Mouse Histology and Phenotyping Laboratory for paraffin processing at the Northwestern University Robert H. Lurie Comprehensive Cancer Center, and the Center for Medical Genomics at Indiana University School of Medicine for RNA sequencing services. This work was supported by grants from the NCI: R01 CA123484, P50 CA180995, and F30 CA250248.
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.