LETTERS

Somatic and germline activating mutations of the ALKkinase receptor in neuroblastomaIsabelle Janoueix-Lerosey1,2, Delphine Lequin1,2, Laurence Brugieres3, Agnes Ribeiro4, Loıc de Pontual5,Valerie Combaret6, Virginie Raynal1,2, Alain Puisieux6,7,8, Gudrun Schleiermacher1,2,9, Gaelle Pierron4,Dominique Valteau-Couanet3, Thierry Frebourg10, Jean Michon9, Stanislas Lyonnet5, Jeanne Amiel5

& Olivier Delattre1,2,4

Neuroblastoma, a tumour derived from the peripheral sympatheticnervous system, is one of the most frequent solid tumours in child-hood1,2. It usually occurs sporadically but familial cases areobserved, with a subset of cases occurring in association with con-genital malformations of the neural crest being linked to germlinemutations of the PHOX2B gene1–4. Here we conducted genome-widecomparative genomic hybridization analysis on a large series ofneuroblastomas. Copy number increase at the locus encoding theanaplastic lymphoma kinase (ALK)5 tyrosine kinase receptor wasobserved recurrently. One particularly informative case presented ahigh-level gene amplification that was strictly limited to ALK, indi-cating that this gene may contribute on its own to neuroblastomadevelopment. Through subsequent direct sequencing of cell linesand primary tumour DNAs we identified somatic mutations of theALK kinase domain that mainly clustered in two hotspots. Germlinemutations were observed in two neuroblastoma families, indicatingthat ALK is a neuroblastoma predisposition gene. Mutated ALKproteins were overexpressed, hyperphosphorylated and showedconstitutive kinase activity. The knockdown of ALK expression inALK-mutated cells, but also in cell lines overexpressing a wild-typeALK, led to a marked decrease of cell proliferation. Altogether, thesedata identify ALK as a critical player in neuroblastoma developmentthat may hence represent a very attractive therapeutic target in thisdisease that is still frequently fatal with current treatments6,7.

In a survey of the amplicons of neuroblastoma8, we recentlyobserved that high-level amplification (.10 copies) of ALK, a tyrosinekinase receptor (RTK) gene preferentially expressed in the central andperipheral nervous systems5, was associated with MYCN amplifica-tion, the most frequent amplicon in neuroblastoma1,2,9. The expres-sion level of ALK was strongly correlated with copy number8, henceconfirming and extending previous findings of ALK amplificationsand overexpression in neuroblastoma10–12.

To document more precisely ALK genomic status, we evaluated copynumber variation at this locus in a series of 592 neuroblastomas thatwere investigated by bacterial artificial chromosome (BAC)-array com-parative genomic hybridization (CGH). A total of 26 cases (4.4%)demonstrated higher than twofold copy number increases and an addi-tional 135 cases (22.8%) presented lower level gains, therefore indicatingthat more that 25% of neuroblastomas harbour a significant copy num-ber increase at the ALK locus (Fig. 1a). We also analysed the expressionpattern of ALK across a series of paediatric tumours and normal tissues.This clearly showed that, whereas other tumour types occasionally

expressed ALK, neuroblastoma consistently showed much higher levelsof expression (Fig. 1b). In the course of this genomic analysis, a subset ofcases without MYCN amplification was further studied by 100K single-nucleotide polymorphism (SNP) arrays. One particularly informativecase was identified (NB-99) harbouring a unique and very high-levelamplicon that was strictly limited to the ALK gene (Fig. 1c). The obser-vation of such a focal and high-level amplicon suggested that ALK mayhave a role on its own in neuroblastoma.

Because RTK may be activated in human malignancies by genomicamplification, chromosome translocation or point mutations13, wewondered whether ALK gene alterations, different from gene amplifica-tion, may be observed in neuroblastoma. We therefore searched formutations in a first series of 28 cell lines. Single-nucleotide substitu-tions leading to changes of highly conserved amino acid residues werefound and prompted us to extend the analysis to frozen tumour sam-ples (115 cases). Altogether, the 16 identified mutations could begrouped into two main hotspots (Table 1). Notably, these changes werenot observed in matched constitutional DNAs that were available fornine cases, hence demonstrating their somatic occurrence (Table 1 andSupplementary Fig. 1). In case CLB-GE that combined gene amplifica-tion and somatic mutation of the ALK gene, sequence analysis showedthat the amplified and consequently overexpressed allele was the onewith the F1174V change (Table 1), therefore suggesting that amplifica-tion and mutation may exert synergistic roles. In other cases, polymer-ase chain reaction with reverse transcription (RT–PCR) analysisindicated that both wild type and mutant alleles were expressed(Table 1). Although a F1174 mutation was observed in one primarytumour (NB-512), changes involving this amino acid were more fre-quent in cell lines, suggesting that they may provide a selective in vitrogrowth advantage (Table 1).

We also investigated the sequence of the kinase domain in sixfamilies presenting at least two siblings with neuroblastoma in theabsence of malformation phenotype and in three cases associatingneuroblastoma and Hirschprung’s disease. A R1275Q heterozygousmutation was identified in one family. It was observed in the twoaffected children and inherited from the mother who was asympto-matic at 39 years of age (Fig. 1d). A R1192P heterozygous mutationwas detected in another family. In this family, three out of six muta-tion carriers developed neuroblastic tumours, being ganglioneuromain one case and stage 4s neuroblastomas in two other cases (Fig. 1d).No ALK mutation was detected in the other neuroblastoma familiesnor in Hirschprung’s disease–neuroblastoma associations.

1Institut Curie, Centre de Recherche, 2Inserm, U830, 26 rue d’Ulm, Paris F-75248, France. 3Institut Gustave Roussy, Departement de pediatrie, 39 rue Camille Desmoulins, 94805Villejuif, France. 4Institut Curie, Unite de Genetique Somatique, Paris F-75248, France. 5Departement de Genetique, Universite Paris Descartes, Faculte de Medecine et INSERM-U781,Hopital Necker-Enfants Malades, 149, rue de Sevres, 75743 Paris Cedex 15, France. 6Centre Leon Berard, FNCLCC, Laboratoire de Recherche Translationnelle. 7Inserm, U590,8Universite de Lyon, Lyon1, Institut des Sciences Pharmaceutiques et Biologiques, Lyon F-69008, France. 9Institut Curie, Departement de Pediatrie, Paris F-75248, France. 10Service deGenetique, CHU de Rouen et Inserm U614, Faculte de Medecine et de Pharmacie, 76183 Rouen Cedex, France.

Vol 455 | 16 October 2008 | doi:10.1038/nature07398

967

©2008 Macmillan Publishers Limited. All rights reserved

We then investigated whether ALK mutation may result in a con-stitutive activation of the kinase activity. In agreement with expres-sion data shown in Fig. 1b, the ALK protein was strongly expressed inneuroblastoma cell lines with particularly high levels in CLB-GE andCLB-GA that express the F1174V and R1275Q ALK mutant alleles,respectively (Fig. 2a). Surprisingly, in addition to the expected ALKband, the CLB-BAR cell line also expressed high levels of a lowermolecular mass ALK protein (Fig. 2a). ALK proteins of CLB-GA,CLB-GE and KCNR, together with the truncated ALK of CLB-BAR, were highly reactive with the Y1586-specific ALK antibody thattargets a phosphorylated residue associated with activation of theALK protein14. In contrast, wild-type ALK of IMR32, CLB-BER,CLB-BAR and RH30 did not react with this antibody (Fig. 2a).ALK-specific immunoprecipitates from neuroblastoma cell lineswere used for in vitro kinase assays. As compared to wild-typeALK, the mutant versions exhibited a much stronger in vitro kinaseactivity (Fig. 2b). In particular, the truncated ALK protein expressedby CLB-BAR harboured a very high kinase activity that was notdetectable in the normal ALK isoform of this cell line. Altogether,these data show that ALK R1275Q and F1174L mutants and thetruncated version of CLB-BAR harbour constitutive ALK kinaseactivity. Given the higher expression level of the mutant versions as

compared to wild-type proteins, we cannot also exclude a role of themutations in the stabilization of the proteins.

Knockdown of the ALK gene was then investigated using lenti-viruses encoding ALK-specific or control small hairpin (sh) RNAs(Fig. 2c). Virally infected cells were plated at low density, maintainedfor puromycin selection for 10 days then stained with crystal violet.This showed that ALK-specific shRNA induced a specific and con-siderable inhibition of focus formation as compared to controlshRNAs (Fig. 2d). Moreover, analysis of growth curves of ALK-knocked-down CLB-GA cells indicated a major reduction of cellproliferation as compared to control cells (Fig. 2e). The growth arrestwas almost complete with the most efficient shRNAs (that is, sh5992and sh5706). It was less pronounced with sh2854, in agreement withthe only partial knockdown of ALK induced by this shRNA (Fig. 2c).Similar growth inhibition was observed for mutated CLB-MA orKCNR cells but also for IMR32 and 106C cells that harbour wild-type ALK sequences (Fig. 2e and data not shown).

Preliminary experiments addressed the issue of the downstreampathways that may be altered by ALK. Phosphorylated ERK1/2,MEK1/2, c-RAF and AKT were analysed on a panel of mutated andwild-type ALK cell lines (Supplementary Fig. 2). Apart from AKT, thephosphorylated form of which was detected in all cell lines, whateverthe ALK status, phosphorylation of the other proteins was only occa-sionally observed. The knockdown of ALK led to a decreased expres-sion of phospho-AKT in CLB-GA but not in IMR32 (Fig. 2f).

PF-2341066, a recently described inhibitor of c-MET activity thatalso shows ALK inhibitor activity15, was tested on the growth of ALKmutated CLB-GA and CLB-MA cells, on ALK wild-type IMR32 cells,as well as on SK-N-AS cells that express neither ALK nor c-MET. Asshown in Fig. 2g, the IC50 (half-maximal inhibitory concentration)values observed in the three former cell lines was significantly lowerthan the one for SK-N-AS.

Altogether, these data demonstrate mutations of the ALK gene inneuroblastoma that lead to constitutive kinase activity of the protein. A

Log 2

Affy

met

rix s

igna

l

a

b

0.492× threshold 0

Log2 CGH ratios

135cases

(22.8%)

Num

ber

of c

ases

26 cases(4.4%)

3.71 2

m/+ m/+

m/+

m/+

m/+

m/+ m/+

m/+m/+

+/+ +/+

+/++/+

+/+

d

0

5

10

15

1 2 3 4 5 6 7 8

123456

30 60 90 120 150 180 2100

c

240

Log 2

rat

ios

16

15

0

3

5.9

8.9

12

RefSeq genesChr2:

CLIP4ALK

Figure 1 | Genetic characteristics of ALK in neuroblastoma. a, Distributionof the log2 values of tumour/control ratios at the RP11-328L16 (ALK) locusin 592 neuroblastomas analysed by BAC array-CGH. The double arrowshows the 99.5% confidence interval for normal copy number as defined bynormal/normal array-CGH experiments. A total of 26 cases (4.4%)demonstrate log2 values higher than 0.49, a previously defined threshold for23 copy number increase (vertical solid line)27. A total of 135 cases (22.8%)exhibit log2 ratios higher than the 99.5% confidence interval for normalratios defined above (vertical dotted line) but lower than the 23 threshold.The values of RP11-328L16 across this series of samples are provided asSupplementary Data 1. b, Box-plots of expression levels of ALK (Affymetrixprobe set 208212_s_at) across a series of 32 small round cell desmoplastictumours (1), 39 Ewing samples (2), 75 medulloblastomas (3), 64neuroblastomas (4), 26 malignant rhabdoid tumours (6), 122rhabdomyosarcoma (7) and 11 teratomas (8). Normal tissue data (5) weredownloaded from the GEO database (GSE3526). The box represents themiddle 50% of data points. The upper and lower whiskers represent thehighest and lowest observed values that are less than 1.5 box lengths from theends of the box. The values of the 208212_s_at probe set across the completeseries of samples is provided as Supplementary Data 2. c, SNP-array analysisof case NB-99. Only chromosome 2 is displayed. The x axis shows thephysical distance (in megabases) from the telomere of the short arm. The yaxis shows log2 ratios of copy number analysis. The arrow indicates the SNPslocated within the ALK gene that show very high-level amplification (.25copies). A UCSC scheme extending from SNPs flanking the amplicon isshown beneath the array-CGH data. No other amplicon, including MYCN,was observed and the MYCN coding sequence was wild type in this tumour.d, ALK mutations in two neuroblastoma families. In the top family, theR1275Q mutation (m) is detected in the daughter who developed ametastatic neuroblastoma at 15 years of age, in the son affected by a localizedneuroblastoma at 3 years of age, and in the mother who is asymptomatic at39. In the bottom family, the R1192P mutation is observed in the daughterwho developed a ganglioneuroblastoma at 12 years of age and in the twograndchildren who developed stage 4s neuroblastomas at 3 and 4 months ofage, respectively. The latter died of the disease.

LETTERS NATURE | Vol 455 | 16 October 2008

968

©2008 Macmillan Publishers Limited. All rights reserved

more exhaustive screen of the whole ALK coding sequence should nowenable the determination of the exact frequency of ALK mutations inneuroblastoma. Recent data pointing out putative translocationsinvolving ALK in neuroblastoma suggest that additional gene rearran-gements of ALK remain to be identified16. With respect to clinicalrelationship, our first indications suggest that ALK mutations can beobserved in tumours of various clinical stages and of various genomicprofiles (Table 1). Moreover, a variety of neuroblastic tumours, from

benign ganglioneuroma to very aggressive metastatic neuroblastoma,can be observed in the context of a germline ALK mutation. Theobservation that a number of mutation carriers remain asymptomaticindicates incomplete penetrance, a characteristic that is reminiscent ofthe PHOX2B-related neuroblastoma predisposition4.

ALK is recurrently involved in gene fusions induced by chro-mosome translocations in different types of cancer including ana-plastic large cell lymphomas17, inflammatory myofibroblastic

Table 1 | ALK point mutations observed in neuroblastoma

Name Sample Stage MYCNamplification

Genomic profile ALKamplification*

Somaticanalysis

Constitutionalanalysis

Type ofmutation

Expressedallele(s){

CLB-GE Cell line 4 A 1p D, 17q G A Exon 23 F1174V (TTC.GTC) WT Somatic MutatedKCNR Cell line 4 A 1p D, 17q G NA Exon 23 F1174L (TTC.TTA) ND Unknown BothCLB-GA Cell line 4 NA 1p D, 11q D, 17q G NA Exon 25 R1275Q (CGA.CAA) WT Somatic BothSK-N-SH Cell line 4 NA 17q G NA Exon 23 F1174L (TTC.TTA) ND Unknown BothN206 Cell line 4 A 1p D, 11q D, 17q G NA Exon 23 F1174L (TTC.TTA) ND Unknown BothCLB-BA Cell line 4 A 1p D, 17q G NA Exon 23 F1174L (TTC.CTC) WT Somatic BothLAN1 Cell line 4 A 17q G NA Exon 23 F1174L (TTC.TTA) ND Unknown NDCLB-TR Cell line 4 A 1p D, 17q G ND Exon 23 F1174L (TTC.CTC) WT Somatic BothCLB-MA Cell line 4 A 1p D, 17q G NA Exon 23 F1174L (TTC.TTA) WT Somatic BothNB-212{ Tumour 4s NA Numerical Alt NA Exon 25 R1275Q (CGA.CAA) WT Somatic BothNB-372 Tumour 1 NA 17q G NA Exon 25 R1275Q (CGA.CAA) WT Somatic NDNB-211 Tumour 2 NA Numerical Alt NA Exon 25 R1275Q (CGA.CAA) WT Somatic NDNB-175 Tumour 2 NA 17q G NA Exon 25 Y1278S (TAC.TCC) WT Somatic NDNB-512 Tumour 4 NA 11q D, 17q G NA Exon 23 F1174C (TTC.TGC) ND Unknown NDNB-286 Tumour 4 NA 11 q D, 17q G NA Exon 25 R1275Q (CGA.CAA) ND Unknown NDNB-222 Tumour 2 NA Numerical Alt NA Exon 25 R1275L (CGA.CTA) ND Unknown Both

A, amplification; NA, no amplification; ND, not done; WT, wild type. The complete ALK coding sequence was investigated in all cell lines. Only the TK domain was screened in tumours. For the‘Genomic profile’ column: D, deletion; G, gain; Numerical Alt, only numerical alterations.*Gene amplification was investigated either by BAC-array CGH, Affymetrix 100K SNP arrays or semi-quantitative PCR.{As deduced from sequence analysis of RT–PCR products.{The ALK mutation was observed in the primary tumour and in a cutaneous nodule.

a

pALK(Y1586)

ALK

b

173 kDa

173 kDa

117 kDa

117 kDa

IMR32

CLB-G

A

CLB-G

E

CLB-B

AR

KCNR

f

IMR32

CLB-G

A

CLB-G

E

CLB-B

ER

RH30

CLB-GA

IMR32

c

sh57

06

NI

shC

T

sh59

92sh

2854

ALK

β-Actin

shC

T

NI

shC

T

sh59

92

CLB-GA IMR32

β-Actin

CLB-B

AR

KCNR

ALK

AKT

NI

shC

T

sh59

92

CLB-GA

pAKTN

I

shC

T

sh59

92

IMR32

e

g

0.0

5.0

10.0

15.0

20.0

25.0

30.0

IMR32

CLB-M

A

CLB-G

A

SK-NA-S

IC50

d

CLB-G

A

sh5992

IMR32

106C

shCT

CLB-M

A

PF-2341066

sh5992sh2854sh5706NIshCT

sh5992NIshCT

0 2 4 6 8 10 12 14

0 2 4 6 8 10 12 14Day

1,500

1,250

1,000

750

500

250

2,5002,2502,0001,7501,5001,2501,000

750500250

Cel

l num

ber

Cel

l num

ber

Figure 2 | Functional characterization of mutated ALK. a, Immunoblots ofALK and phosphorylated ALK receptor in various cell lines. RH30 is arhabdomyosarcoma cell line. IMR32 and CLB-BER cells do not present ALKmutation. The arrow indicates the low molecular mass ALK protein that, incontrast to the wild-type ALK also expressed in CLB-BAR, strongly reactswith the Y1586 antibody. b, Strong kinase activity of ALK mutants. 32Pincorporated in ALK immunoprecipitates is visualized by autoradiography(top panel) before immunoblotting with an anti-ALK antibody (bottompanel). Mutated ALK from CLB-GE, CLB-GA and KCNR presents strongkinase activities whereas wild-type ALK from IMR32 barely exhibits anykinase activity. The abnormal ALK protein, but not the wild-type protein, ofCLB-BAR demonstrates a strong kinase activity. c, Immunoblot afterlentiviral-induced knockdown of ALK in the CLB-GA and IMR32 cell lines.

NI, non-infected cells; shCT, cells infected with a control lentivirus. sh2854,sh5992 and sh5706 are ALK-specific small hairpin RNAs. d, Decreased focusformation of CLB-GA, IMR32, 106C and CLB-MA cells on ALK knockdownwith sh5992 as compared to control short hairpin RNA (shCT). e, Decreasedproliferation of CLB-GA and IMR32 cells on ALK knockdown. The numberof living cells 3 1,000 is reported along the y axis and the x axis indicates thedays at which cells were counted. Means 6 s.d. are indicated for CLB-GA.Duplicate experiments were performed for both cell lines. Similar resultswere obtained with the CLB-MA, KCNR and 106C cell lines (data notshown). f, Immunoblot of AKT and phosphorylated AKT in CLB-GA andIMR32 cells on ALK knockdown. g, PF-2341066 inhibition of cell growth inneuroblastoma cell lines expressing ALK. Mean IC50 values (mM) and s.d. ofquadruplicate experiments are reported.

NATURE | Vol 455 | 16 October 2008 LETTERS

969

©2008 Macmillan Publishers Limited. All rights reserved

tumours18 and a subset of lung cancers19. Most fusion proteinsstudied so far are characterized by constitutive activation of theALK tyrosine kinase activity through spontaneous dimerizationand induce cell transformation5,20,21. It remains to be documentedthat neuroblastoma-associated ALK mutations alone lead to celltransformation. However, mutations result in a constitutive activa-tion of the kinase activity, a characteristic that is also observed aftergenetic changes of homologous domains, in particular of the activa-tion loop, in other transforming RTK-encoding genes22 (Supple-mentary Fig. 1). In many respects, ALK involvement in neuroblas-toma is reminiscent of FLT3 contribution in acute myeloid leuk-aemia (AML). Indeed, whereas somatic mutations are only foundin a subset of cases, most AMLs or neuroblastoma express high levelsof FLT3 or ALK, respectively23. In AML, there is some evidence thatautocrine stimulation of the wild-type FLT3 receptor may confer aproliferative advantage to AML cells23–25. This cannot yet be tested inneuroblastoma as the nature of the ALK ligand(s) is still a matter ofdebate5, but this may constitute an attractive hypothesis to accountfor the proliferative role of wild-type ALK.

Our preliminary observations, together with recent reports show-ing the potential of pharmacological5,6,15 or immunological26 strategiesto inhibit ALK, in particular in neuroblastoma cell lines16, suggest thatsuch approaches will probably represent key therapeutic opportu-nities for neuroblastoma that is still very frequently a deadly disease.

METHODS SUMMARY

Array-CGH experiments were conducted as previously described on the Institut

Curie one megabase resolution BAC-array8,27. SNP analyses were based on the

100K Affymetrix SNP arrays.

For immunoblots, total ALK was detected using monoclonal ALK1 (anti-

human CD246, DakoCytomation) or polyclonal (Zymed) antibodies. Y1586

phosphorylated ALK was revealed using the number 3343 antibody (Cell

Signaling Technologies). For immunoprecipitation, 250mg of sonicated lysates

were incubated with polyclonal ALK antibodies (Zymed) for 2 h at 4 uC then with

G Sepharose beads (Pharmacia) for 1 h at 4 uC. After washing, the immunopre-

cipitates bound to protein G Sepharose were resuspended in gel loading buffer

and resolved by SDS–polyacrylamide gel electrophoresis (PAGE).

Control and ALK-specific small hairpin RNAs in the pLKO puromycin-res-

istant vector were purchased from Sigma (SIGMA Mission shRNA). HEK293

cells were used for packaging using the calcium phosphate method. For colony

formation experiments, 3 3 104 transduced cells were plated at day 0 in 6-well

dishes and stained with crystal violet at day 10. For proliferation analysis, 2 3 104

transduced cells were plated at day 0 in 24-well dishes. The number of living cells

was counted at day 4, 7, 10 and 14 using a Vi-cell XR counter (Beckman Coulter).

For in vitro kinase assay, ALK immunoprecipitates were incubated with pro-

tein G Sepharose beads (Pharmacia), then beads were pelleted and washed in

kinase buffer. Kinase reactions were performed for 15 min at 20 uC in 40ml of

kinase buffer supplemented with [c-32P]ATP (10mCi, Perkin Elmer). Reactions

were stopped by the addition of an equal volume of gel loading buffer. After

SDS–PAGE, the gels were electrotransferred onto Hybond nitrocellulose mem-

branes (Amersham Pharmacia Biotech), and the kinase reaction products were

detected by autoradiography.

For PF-2341066 growth inhibition, a range of concentrations from 100mM to

0.01 nM was tested. MTS assay was used to determine cell viability.

Full Methods and any associated references are available in the online version ofthe paper at www.nature.com/nature.

Received 26 May; accepted 28 August 2008.

1. Maris, J. M., Hogarty, M. D., Bagatell, R. & Cohn, S. L. Neuroblastoma. Lancet 369,2106–2120 (2007).

2. Brodeur, G. M. Neuroblastoma: biological insights into a clinical enigma. NatureRev. Cancer 3, 203–216 (2003).

3. Tonini, G. P., Longo, L., Coco, S. & Perri, P. Familial neuroblastoma: a complexheritable disease. Cancer Lett. 197, 41–45 (2003).

4. Trochet, D. et al. Germline mutations of the paired-like homeobox 2B (PHOX2B)gene in neuroblastoma. Am. J. Hum. Genet. 74, 761–764 (2004).

5. Chiarle, R., Voena, C., Ambrogio, C., Piva, R. & Inghirami, G. The anaplasticlymphoma kinase in the pathogenesis of cancer. Nature Rev. Cancer 8, 11–23 (2008).

6. Galkin, A. V. et al. Identification of NVP-TAE684, a potent, selective, andefficacious inhibitor of NPM-ALK. Proc. Natl Acad. Sci. USA 104, 270–275 (2007).

7. Li, R. & Morris, S. W. Development of anaplastic lymphoma (ALK) small-moleculeinhibitors for cancer therapy. Med. Res. Rev. 28, 372–412 (2008).

8. Fix, A. et al. Characterization of amplicons in neuroblastoma. High-resolutionmapping using DNA microarrays, relationship with outcome, and identification ofoverexpressed genes. Genes Chromosom. Cancer 47, 819–834 (2008).

9. Schwab, M. et al. Chromosome localization in normal human cells andneuroblastomas of a gene related to c-myc. Nature 308, 288–291 (1984).

10. Lamant, L. et al. Expression of the ALK tyrosine kinase gene in neuroblastoma. Am.J. Pathol. 156, 1711–1721 (2000).

11. Osajima-Hakomori, Y. et al. Biological role of anaplastic lymphoma kinase inneuroblastoma. Am. J. Pathol. 167, 213–222 (2005).

12. Miyake, I. et al. Activation of anaplastic lymphoma kinase is responsible forhyperphosphorylation of ShcC in neuroblastoma cell lines. Oncogene 21,5823–5834 (2002).

13. Blume-Jensen, P. & Hunter, T. Oncogenic kinase signalling. Nature 411, 355–365(2001).

14. Rikova, K. et al. Global survey of phosphotyrosine signaling identifies oncogenickinases in lung cancer. Cell 131, 1190–1203 (2007).

15. Zou, H. Y. et al. An orally available small-molecule inhibitor of c-Met, PF-2341066,exhibits cytoreductive antitumor efficacy through antiproliferative andantiangiogenic mechanisms. Cancer Res. 67, 4408–4417 (2007).

16. McDermott, U. et al. Genomic alterations of anaplastic lymphoma kinase maysensitize tumors to anaplastic lymphoma kinase inhibitors. Cancer Res. 68,3389–3395 (2008).

17. Morris, S. W. et al. Fusion of a kinase gene, ALK, to a nucleolar protein gene, NPM,in non-Hodgkin’s lymphoma. Science 263, 1281–1284 (1994).

18. Griffin, C. A. et al. Recurrent involvement of 2p23 in inflammatory myofibroblastictumors. Cancer Res. 59, 2276–2280 (1999).

19. Soda, M. et al. Identification of the transforming EML4-ALK fusion gene in non-small-cell lung cancer. Nature 448, 561–566 (2007).

20. Chiarle, R. et al. NPM-ALK transgenic mice spontaneously develop T-celllymphomas and plasma cell tumors. Blood 101, 1919–1927 (2003).

21. Fujimoto, J. et al. Characterization of the transforming activity of p80, ahyperphosphorylated protein in a Ki-1 lymphoma cell line with chromosomaltranslocation t(2;5). Proc. Natl Acad. Sci. USA 93, 4181–4186 (1996).

22. Hubbard, S. R. Juxtamembrane autoinhibition in receptor tyrosine kinase. NatureRev. Mol. Cell Biol. 5, 464–471 (2004).

23. Renneville, A. et al. Cooperating gene mutations in acute myeloid leukaemia: areview of the literature. Leukemia 22, 915–931 (2008).

24. Gilliland, G. & Griffin, G. The roles of FLT3 in hematopoiesis and leukaemia. Blood100, 1532–1542 (2002).

25. Frohling, S. et al. Identification of driver and passenger mutations of FLT3 by high-throughput DNA sequence analysis and functional assessment of candidatealleles. Cancer Cell 12, 501–513 (2007).

26. Chiarle, R. et al. The anaplastic lymphoma kinase is an effective oncoantigen forlymphoma vaccination. Nature. Med. 14, 676–680 (2008).

27. Vincent-Salomon, A., Raynal, V., Lucchesi, C., Gruel, N. & Delattre, O. ESR1 geneamplification in breast cancer: a common phenomenon? Nature Genet. 40, 809(2008).

Supplementary Information is linked to the online version of the paper atwww.nature.com/nature.

Acknowledgements We are grateful to F. Moreau-Gachelin, I. Gallais, D. Surdez,F. Tirode, A. Fix, F. Bourdeaut, A. Almeida, C. Lucchesi, S. Roman Roman, B. Bressac,J. Benard and G. Vassal for their critical help. We thank C. Decraene, D. Gentien andB. Albaud from the translational department of Institut Curie for profiling thepaediatric tumours, and P. Rosa and E. Barillot for the development of bioinformatictools. We thank M. Lathrop and the Centre National de Genotypage for theAffymetrix 100K SNP analysis and A. Chompret for collecting families. This workwas supported by grants from the Agence Nationale pour la Recherche, the InstitutNational du Cancer, the Ligue Nationale contre le Cancer (Equipe labellisee and CITproject), the APAESIC (Association des Parents et des Amis des Enfants Soignes al’Institut Curie), the Association Hubert Gouin, les amis de Claire, Les Bagouz aManon and Enfance et Sante. A.P. and V.C. are supported by the Comite de l’Ain ofthe Ligue Nationale contre le Cancer.

Author Contributions I.J.-L., D.L., A.R., L.P., V.C. and V.R. generated the data; I.J.-L.,G.P., A.P., J.M., J.A., S.L. and O.D. made the study design and follow-up; L.B., V.C.,A.P., G.S., D.V.-C., T.F., S.L. and J.A. contributed biological materials that were usedin this study.

Author Information Microarray data have been submitted to the Gene ExpressionOmnibus (http://www.ncbi.nlm.nih.gov/geo) public database. The accessionnumbers for gene expression profiles of neuroblastoma samples and SNP data forcase NB-99 displayed in Fig. 1c are GSE12460 and GSE12461, respectively. Reprintsand permissions information is available at www.nature.com/reprints.Correspondence and requests for materials should be addressed to O.D.(olivier.delattre@curie.fr).

LETTERS NATURE | Vol 455 | 16 October 2008

970

©2008 Macmillan Publishers Limited. All rights reserved

METHODSArray-CGH experiments. Experiments were conducted as previously described

on the Institut Curie one megabase resolution BAC-array8,27. The RP11-328L16

BAC contains most of the ALK gene. SNP analyses were based on the 100K

Affymetrix SNP arrays (GeneChip 50K array Hind and Xba).

DNA sequencing was performed using the ABI PRISM BigDye Terminator

Cycle Sequencing Ready Reaction Kit and sequences were analysed using

SeqScape software (Applied Biosystems).

Immunoblots. An antibody (number 3343, Cell Signaling Technologies) directed

against the phosphorylated Y1586 residue was used to reveal phosphorylatedALK. Total ALK was detected using the monoclonal ALK1 (anti-human

CD246, DakoCytomation) or polyclonal (Zymed) antibodies.

RNA interference. Control and ALK specific small hairpin RNAs sh2854 (59-

CCGGACCCAAATCAAGAAACCTGTTCTCGAGAACAGGTTTCTTGATTT-

GGGTTTTTT-39), sh5992 (59-CCGGGTGATAAATACAAGGCCCAGACTC-

GAGTCTGGGCCTTGTATTTATCACTTTTT-39) and sh5706 (59-CCGGGA-

GCTGGTCATTACGAGGATACTCGAGTATCCTCGTAATGACCAGCTCT-

TTTT-39) in the pLKO puromycin-resistant vector were purchased from Sigma

(SIGMA Mission shRNA). HEK293 cells were used for packaging using the cal-

cium phosphate method. Viral supernatants were supplemented with 8mg ml21

polybrene and incubated with neuroblastoma cells for 7 h. Selection was then

applied with puromycin at 600 ng ml21 for CLB-GA, 800 ng ml21 for IMR32

and 1mg ml21 for 106C and CLB-MA. For colony formation experiments,

3 3 104 transduced cells were plated at day 0 in 6-well dishes and stained with

crystal violet at day 10. For proliferation analysis, 2 3 104 transduced cells were

plated at day 0 in 24-well dishes. The number of living cells was counted at day 4, 7,

10 and 14 using a Vi-cell XR counter (Beckman Coulter).

Immunoprecipitation. Cells were lysed in ice-cold lysis buffer (10 mM Tris-Hcl,pH 7.4, 150 mM NaCl, 5 mM EDTA, 20 mM NaF, 25 mM b-glycerophosphate,

1 mM Na-pyrophosphate, 10% glycerol, 1% NP40, 0.25% Na deoxycholate, 1

mM Na3VO4, 13 protease inhibitor; Roche). Lysates were sonicated then cen-

trifuged for 30 min at 13,000 r.p.m. at 4 uC. Supernatants were recovered and

protein concentration was determined by the Bradford assay. A total of 250mg

were incubated with polyclonal ALK antibodies (Zymed) for 2 h at 4 uC then with

G Sepharose beads (Pharmacia) for 1 h at 4 uC. The immunoprecipitates bound

to protein G Sepharose were washed twice with lysis buffer (without Na-deoxy-

cholate) then re-suspended in gel loading buffer and resolved by SDS–PAGE.

In vitro kinase assay. After incubation of ALK immunoprecipitates with protein

G Sepharose beads (Pharmacia), the beads were pelleted and washed twice with

buffer containing 10 mM Tris-Hcl, pH 7.4, 150 mM NaCl, 5 mM EDTA, 20 mM

NaF, 25 mM b-glycerophosphate, 1 mM Na-pyrophosphate and 10% glycerol

then washed with kinase buffer (20 mM HEPES, pH 7.4, 10 mM MnCl2 and

0.5 mM Na3VO4). Kinase reactions were performed for 15 min at 20 uC in

40 ml of kinase buffer supplemented with [c-32P]ATP (10mCi, Perkin Elmer).

Reactions were stopped by the addition of an equal volume of gel loading buffer.

After SDS–PAGE, the gels were electrotransferred onto Hybond nitrocellulosemembranes (Amersham Pharmacia Biotech), and the kinase reaction products

were detected by autoradiography.

PF-2341066 growth inhibition. A range of concentration from 100mM to

0.01 nM was tested. MTS assay was used to determine cell viability.

doi:10.1038/nature07398

©2008 Macmillan Publishers Limited. All rights reserved