|Home | About | Journals | Submit | Contact Us | Français|
Neuroblastoma, an embryonal tumor of the peripheral sympathetic nervous system, accounts for approximately 15% of all deaths due to childhood cancer1. High-risk neuroblastomas, prevalent in the majority of patients, are rapidly progressive; even with intensive myeloablative chemotherapy, relapse is common and almost uniformly fatal2,3. Here we report the detection of previously unknown mutations in the ALK gene, which encodes a receptor tyrosine kinase, in 8% of primary neuroblastomas. Five non-synonymous sequence variations were identified in the kinase domain of ALK, of which three were somatic and two were germline. The most frequent mutation, F1174L, was also identified in three different neuroblastoma cell lines. ALK cDNAs encoding the F1174L and R1275Q variants, but not the wild-type ALK cDNA, transformed IL-3-dependent murine hematopoietic Ba/F3 cells to cytokine-independent growth. Ba/F3 cells expressing these mutations were sensitive to a small-molecule inhibitor of ALK, TAE6844. Furthermore, two human neuroblastoma cell lines harboring the F1174L mutation were sensitive to the inhibitor. Cytotoxicity was associated with increased levels of apoptosis as measured by TUNEL-labeling. shRNA-mediated knockdown of ALK expression in neuroblastoma cell lines with the F1174L mutation also resulted in apoptosis and impaired cell proliferation. Thus, activating alleles of the ALK receptor tyrosine kinase are present in primary neuroblastoma tumors and in established neuroblastoma cell lines, and confer sensitivity to ALK inhibition with small molecules, providing a molecular rationale for targeted therapy of this disease.
In a genome-wide analysis of primary neuroblastomas using single-nucleotide polymorphism (SNP) arrays, we noted high-level amplification of the ALK (anaplastic lymphoma kinase) gene5. To determine the frequency of this amplification, we analyzed 94 tumors with MYCN amplification by fluorescence in situ hybridization (FISH), and documented 14 (15%) with concomitant ALK amplification (Supplementary Fig.1), which was not detected in 51 tumors without MYCN amplification (p=0.0016). None of the tumors had ALK rearrangements, such as those that have been found in other tumor types with ALK translocations6–9.
We reasoned that in tumors without ALK amplification or translocation, acquired somatic mutations or germline sequence variants might contribute to oncogenicity. DNA re-sequencing of the ALK open reading frame in primary neuroblastomas identified 5 novel non-synonymous sequence variations in conserved positions in the tyrosine kinase domain in 7 of 93 samples (8%; Table 1; Supplementary Fig.2). None of these variants were previously identified SNPs or known somatic mutations, based on analysis of dbSNP and Sanger databases or by genotyping of 270 samples derived from the International Hap Map Consortium10.
Sequence analysis of matched normal samples from these patients revealed that 2 of the sequence variants were germline and 3 represented somatically acquired mutations (Table 1). The most common mutation, identified in 4.3% (4/93) of the primary tumors, was a recurrent cytosine-to-adenine change in exon 23 that results in a phenylalanine-to-leucine substitution at codon 1174 (F1174L) within the kinase domain. Most of the patients with somatic ALK mutations had metastatic disease characterized by MYCN amplification, although one patient with the F1174L mutation had localized disease with favorable histology and unamplified MYCN (Supplementary Table 1). Four of the 5 ALK mutations involve residues that correspond to those affected by known activating mutations in the EGFR gene11–14 (Supplementary Figs.2 & 3). The F1174 residue corresponds to V769 in EGFR, which is in a region of frequent mutation in both EGFR11 and ERBB212 genes. The F1245C mutation corresponds to L833V in EGFR, a gefitinib-resistant mutation in lung cancer (Greulich & Meyerson, unpublished observations). The R1275Q mutation is located adjacent to L858R in EGFR, which is the most common EGFR mutation in lung cancer13,14.
The functional consequences of four of the mutations, T1151M, F1174L, A1234T, and R1275Q, were determined by testing their abilities to transform interleukin-3 (IL-3)-dependent murine lymphoid Ba/F3 cells to cytokine-independent growth. Reduction in IL-3 concentration by 100-fold to 0.01 ng/ml resulted in a clear difference in cell proliferation, with the Ba/F3 cells expressing F1174L and R1275Q mutations exhibiting much higher cell numbers relative to those transduced with wild-type ALK or the T1151M mutation (Fig 1a). To generate IL3-independent lines, we reduced the IL-3 concentration by half in successive passages of each transduced Ba/F3 line. After 5 passages, the Ba/F3 cells expressing the F1174L and the R1275Q ALK mutations, as well as NPM-ALK, were able to grow in medium completely lacking IL-3, while cells expressing T1151M or wild-type ALK did not survive. Moreover, when expressed in Ba/F3 cells, the F1174L allele, and to a lesser extent, the R1275Q allele, were associated with constitutive phosphorylation of ALK (Fig. 1b). In contrast, neither the T1151M nor A1234T alleles exhibited ALK phosphorylation. Expression of the F1174L ALK protein in IL-3-deprived Ba/F3 cells was also associated with phosphorylation of downstream targets of ALK signaling such as STAT3 and AKT, while R1275Q was associated with phosphorylation of ERK1/2 and AKT (Fig 1b). Together, these studies demonstrate that the ALK mutant proteins F1174L and R1275Q possess gain-of-function kinase activity that can sustain key signaling pathways in the presence of reduced concentrations of IL-3.
The Ba/F3 assay has been validated for a broad spectrum of oncogenic tyrosine kinase alleles including mutant EGFR15 and FLT316, and thus we treated Ba/F3 cells expressing each of the ALK mutations with increasing concentrations of TAE684, a highly potent ALK inhibitor4,17,18. The activating mutation, F1174L was found to be extremely sensitive to TAE684, with an IC50 of 8 nM, identical to that of NPM-ALK-expressing Ba/F3 cells. The R1275Q mutation was also sensitive to TAE684, albeit with a much higher IC50 of 328 nM. In contrast, Ba/F3 cells expressing FLT3-ITD or wild-type ALK, did not respond to TAE684 (IC50 4.5 µM; Fig 1c).
Analysis of the ALK gene in a panel of 30 neuroblastoma cell lines revealed sequence variants in 6, including 3 different cell lines containing the F1174L mutation (Kelly, SH-SY5Y and LAN-1), which was also the most common mutation in the primary tumors (Table 1). An R1275Q mutation, identical to the one found in primary sample 411, was also detected in the SMS-KCNR cell line. We observed dose-dependent growth inhibition of the SH-SY5Y (F1174L) and Kelly (F1174L) neuroblastoma cell lines with increasing concentrations of TAE684, (IC50s of 258 and 416 nM respectively; Fig. 2a). These results are in agreement with data from a recent study showing sensitivity of these cell lines to TAE68419. Of note, the SMS-KCNR cell line expressing the ALK R1275Q mutation was resistant to TAE684 (IC50 of 4.9 µM; Fig. 2a), even though Ba/F3 cells expressing this mutation became IL-3 independent and were sensitive to the inhibitor (Fig. 1c). Neuroblastoma cell lines without ALK mutations, including IMR-5, were also resistant to TAE684 (Fig. 2a and Supplementary Fig. 4a). Treatment with TAE684 (200 nM) resulted in increased apoptosis in Kelly (F1174L) and SH-SY5Y (F1174L) cells, but not in the SMS-KCNR (R1275Q) and IMR-5 (WT) cells (Fig. 2b). Cytotoxicity was also associated with G1-phase arrest and substantial reductions in S-phase cell fractions (Supplementary Fig. 4b).
After treatment with TAE684 (100nM), the sensitive cell lines SH-SY5Y (F1174L) and Kelly (F1174L) demonstrated reduced phosphorylation of ALK, and of ERK1/2 and AKT and to a lesser extent, STAT3 (Supplementary Fig 4c). By contrast, there was no apparent effect on phospho-AKT and STAT3 in the resistant cell line IMR-5 (WT), although there was a slight reduction in phospho-ERK1/2. Moreover, knockdown of ALK in the Kelly and SH-SY5Y cell lines (F1174L), but not the inhibitor-resistant SMS-KCNR line (R1275Q), was associated with a reduction in cell proliferation and increased apoptosis (Fig. 2c, d). In neuroblastoma, ALK is detected as both a 220 kDa protein, reflecting the glycosylated protein encoded by the transduced ALK cDNA and a second protein of ~140 kDa (inset, Fig. 2c), which has been documented by multiple investigators and most likely represents an as-yet-uncharacterized splice variant-encoded isoform of ALK20–22. The resistance of SMS-KCNR cell line is apparently not due to any difference between the R1275Q and F1174L mutations in ALK activation per se, because both of these mutations transform Ba/F3 cells to IL-3 independence and the transformed cells respond to the inhibitor (Fig 1c). Rather, we suspect that other molecular aberrations, such as coactivation of other receptor tyrosine kinases23,24, could have been acquired during culture that render this cell line independent of activated ALK for growth and viability.
We observed that ALK was expressed at substantially lower levels in the TAE684-sensitive neuroblastoma cell lines [Kelly (F1174L), and SH-SY5Y (F1174L)] than in the remaining cell lines harboring either wild-type ALK or the R1275Q mutation (Fig. 3a). However, TAE684 inhibition of ALK kinase activity resulted in an increase in the ALK protein level in the sensitive Kelly (F1174L) cell line, but not in IMR-5 (WT) (Fig. 3b). Blockade of protein degradation by the proteasome inhibitor MG-132 resulted in increased ALK levels in Kelly (F1174L) cells, but not in IMR-5 cells (Fig 3c), consistent with a higher turnover rate in cells with the constitutively activated mutant ALK protein.
The studies reported here demonstrate previously unrecognized activating mutations affecting critical residues within the ALK kinase domain and indicate that ALK has potential as a novel therapeutic target in neuroblastoma. Our results with the ALK small-molecule inhibitor, TAE684, demonstrate that most of the neuroblastoma cell lines harboring activating ALK mutations are dependent on the altered ALK protein for survival. One exception is the activating R1275Q allele, which when expressed by itself in Ba/F3 cells is sensitive to treatment with TAE684, but not in the SMS-KCNR neuroblastoma background. This observation, together with the lack of transforming activity of the T1151M and the A1234T alleles, supports the emerging concept that mutations affecting critical domains of cancer genes must be studied both genetically and functionally to validate their potential as therapeutic targets16. Thus, it will be important to test the ability of each mutant ALK protein identified in patient tumor samples to confer IL-3 resistance in Ba/F3 cells and mediate sensitivity to ALK inhibitory drugs as they move into therapeutic trials. One sensitive cell line, Kelly (F1174L), harbored both an activating ALK mutation and MYCN amplification, suggesting that a subset of very high risk MYCN-amplified neuroblastomas may respond to treatment with an ALK inhibitor. Although established neuroblastoma cell lines with high levels of wild-type ALK expression did not respond to ALK inhibitors in our study, it will be important to evaluate whether the larger group of neuroblastoma patients expressing wild-type ALK proteins have responses when ALK inhibitors are tested in vivo in the natural tumor microenvironment, which includes exposure to the effects of ALK ligands25,26.
Primers were designed to cover the 29 exons of ALK and characterized using 3 Coriell DNAs. Passing primers and samples were PCR amplified and sequenced bidirectionally on an ABI 3730xl automated sequencer (Applied Biosystems). Automated analysis and coverage statistics were generated with SNP Compare [(an in-house package using PolyPhred27 and PolyDHAN (D. Richter et al., manuscript in preparation)]. Bidirectional sequence traces were analyzed with Mutation Surveyor (SoftGenetics, version 3.10) and manual review. Genotyping was performed by primer extension mass spectrometry.
Neuroblastoma cell lines were cultured in RPMI-1640 containing L-glutamine and 10% fetal bovine serum (FBS; Sigma-Aldrich). Ba/F3 cells were maintained in RPMI-1640 supplemented with 10% FBS and 0.5 ng/ml murine IL-3 (Millipore).
ALK mutations were engineered using the QuikChange II Site-Directed Mutagenesis Kit (Stratagene). The mutant cDNAs were subcloned into the pMSCV-Neo-luc retroviral vector. Cotransfection of 293T cells and infection of Ba/F3 cells with retroviral supernatants were performed as described previously16. Transduced Ba/F3 cells were selected with G418 for 7 days and subjected to Ficoll separation to isolate surviving cells.
Ba/F3 cells transduced with each of the pMSCV-Neo-luc constructs were seeded at 1×105/ml and treated with 1, 0.1, 0.01 ng/ml IL-3 for 72 hours. The number of viable cells was determined by trypan blue exclusion using a Vi-CELL Series Cell Viability Analyzer (Beckman Coulter).
Cell viability was tested 72 hours following addition of the compound by CellTiter-Glo Luminescent Cell Viability Assay (Promega). IC50s were calculated by nonlinear regression (variable slope) using Graphpad Prism 5 software.
Immunoblotting was performed as described previously16. Antibodies are listed in Full Methods.
1×106 cells were treated with the proteasome inhibitor MG-132 (Sigma-Aldrich) for 1 hour, washed in PBS and immunoblotting performed as described previously16.
are available in the online version of the paper at www.nature.com/nature.
We thank Jennifer Elechko and Lisa Moreau for technical assistance, Anna Azarova for help with cell culture and cell growth assays, Qin Jiang and Xiaoli Cui for assistance during ALK small-molecule inhibitor development, and A. Kung for Ba/F3 cells and retroviral vectors. We acknowledge the Children’s Oncology Group for provision of neuroblastoma tumor and matched normal DNAs, and tumor touch prep slides. We thank J. Maris for neuroblastoma cell lines. We thank Abbot Molecular International for the Vysis LSI ALK Dual Color, Break Apart Rearrangement Probe. This work was supported by grants from the National Institutes of Health (R. E. G.), the Friends for Life Neuroblastoma Fund (R.E.G.), and the Children’s Oncology Group (R. E. G.), Alex’s Lemonade Stand Foundation (M.H., M.M.)., NCI CA69129 (L.X., S.W.M.), Cancer Center Core grant CA21765 (T.R.W., L.X, S.W.M.), and the American Lebanese Syrian Associated Charities (ALSAC), St. Jude Children’s Research Hospital (T.R.W., L.X., S.W.M.).
SNP array analysis GEO Accession Numbers: GSM206563 GSM206564.