ALK, the Chromosome 2 Gene Locus Altered by the T(2;5) in Non

Home , MAL (gene)

ALK, the chromosome 2 gene locus altered by the t(2;5) in non-Hodgkin's lymphoma, encodes a novel neural receptor tyrosine kinase that is highly related to leukocyte tyrosine kinase (LTK)

Stephan W Morris1,2,4, Clayton Naeve3, Prasad Mathew5, Payton L James1, Mark N Kirstein1, Xiaoli Cui1 and David P Witte6,7

Departments of 1Experimental Oncology, 2Hematology-Oncology and 3Virology and Molecular Biology, St Jude Children's Research Hospital, Memphis, Tennessee 38105; 4Department of Pediatrics, University of Tennessee College of Medicine, Memphis, Tennessee 38163; 5Department of Pediatrics, Medical College of Ohio, Toledo, Ohio 43699; 6Department of Pathology, University of Cincinnati College of Medicine and 7Division of Pathology, Children's Hospital Research Foundation, Cincinnati, Ohio 45229, USA

Anaplastic Lymphoma Kinase (ALK) was originally Introduction identi®ed as a member of the insulin receptor subfamily of receptor tyrosine kinases that acquires transforming Diverse cellular processes such as division, differentia- capability when truncated and fused to nucleophosmin tion, survival, metabolism, and motility can all be (NPM) in the t(2;5) chromosomal rearrangement regulated through binding of cognate ligands to their associated with non-Hodgkin's lymphoma, but further speci®c receptor tyrosine kinases (RTKs) (reviewed in insights into its normal structure and function are Schlessinger and Ullrich, 1992; Fantl et al., 1993). The lacking. Here, we characterize a full-length normal common structural features of the RTKs include an human ALK cDNA and its product, and determine the extracellular ligand-binding region, a hydrophobic pattern of expression of its murine homologue in membrane-spanning segment and a cytoplasmic do- embryonic and adult tissues as a ®rst step toward the main that carries the catalytic function. In general, the functional assessment of the receptor. Analysis of the activation of these molecules to initiate signal 6226 bp ALK cDNA identi®ed an open reading frame transduction involves ligand-mediated receptor dimer- encoding a 1620-amino acid (aa) protein of predicted ization and intermolecular transphosphorylation, which mass *177 kDa that is most closely related to begins a phosphorylation cascade through the binding leukocyte tyrosine kinase (LTK), the two exhibiting of SH2 or PTB domains of cellular substrates to 57% aa identity and 71% similarity over their region of speci®c receptor phosphotyrosine residues (Mayer and overlap. Biochemical analysis demonstrated that the Baltimore, 1993; Schlessinger, 1994; van der Geer and *177 kDa ALK polypeptide core undergoes co-transla- Pawson, 1995). Dysregulation of this process either at tional N-linked glycosylation, emerging in its mature the level of the receptor or aecting key elements of the form as a 200 kDa single chain receptor. Surface signaling cascade has been shown to result in a variety labeling studies indicated that the 200 kDa glycoprotein of neoplastic disorders or in altered development. For is exposed at the cell membrane, consistent with the example, structural rearrangement of RTK genes, as prediction that ALK serves as the receptor for an seen with TRKA, MET and RET, has been shown to unidenti®ed ligand(s). In situ hybridization studies activate the transforming capacity of the proteins they revealed Alk expression beginning on embryonic day encode, contributing to the genesis of human tumors 11 and persisting into the neonatal and adult periods of (Martin-Zanca et al., 1986; Soman et al., 1991; Greco development. Alk transcripts were con®ned to the et al., 1992; Bongarzone et al., 1993). nervous system and included several thalamic and A number of studies have now demonstrated that hypothalamic nuclei; the trigeminal, facial, and acoustic RTKs normally function to regulate the cellular cranial ganglia; the anterior horns of the spinal cord in growth and dierentiation of speci®c tissues or the region of the developing motor neurons; the organs. For example, the TRK gene family (TRK-A, - sympathetic chain; and the ganglion cells of the gut. B and -C) encodes receptors for the neurotrophic Thus, ALK is a novel orphan receptor tyrosine kinase ligands nerve growth factor, brain-derived neuro- that appears to play an important role in the normal trophic factor, neurotrophin-3 and neurotrophin-4 development and function of the nervous system. (Barbacid, 1993). Targeting experiments in mice have demonstrated that functional inactivation of the TRK Keywords: anaplastic lymphoma kinase (ALK); recep- genes produces multiple abnormalities of the central tor tyrosine kinase; neural-speci®c; leukocyte tyrosine and peripheral nervous systems (Snider, 1994). The kinase (LTK); chromosome 2 MET RTK has been shown to be the cell surface receptor for hepatocyte growth factor (HGF)/scatter factor (SF), which mediates liver development and regeneration (Bottaro et al., 1991; Naldini et al., 1991) and provides a signal for the migration of skeletal muscle precursor cells from the somites to the Correspondence: SW Morris developing limbs and diaphragm (Bladt et al., 1995). Received 23 October 1996; revised 23 January 1997; accepted With increasing frequency, the normal functional role 24 January 1997 of RTKs is also being elucidated because of their ALK, a novel, neural receptor tyrosine kinase SW Morris et al 2176 identi®cation as disease genes. Recent examples include phosphoprotein nucleophosmin (NPM), a highly the identi®cation of RET mutations that lead to the conserved RNA-binding nucleolar protein (Borer et loss of enteric innervation in Hirschsprung's disease al., 1989; Chan et al., 1989; Dumbar et al., 1993), (congenital megacolon) or to tumor formation in linked to the cytoplasmic portion of a novel receptor multiple endocrine neoplasia type IIB (van Heynin- tyrosine kinase encoded on chromosome 2 which we gen, 1994), and the causal association between named Anaplastic Lymphoma Kinase (ALK). Here, mutations within ®broblast growth factor receptor we describe the isolation and sequence analysis of a (FGFR) family members and the congenital malforma- full-length normal human ALK receptor cDNA, tions Pfeier syndrome (craniosynostosis, cutaneous demonstrating that ALK is a member of the insulin syndactyly and short, broad digits) (FGFR1) (Muenke receptor subfamily of RTKs with closest homology to et al., 1994), Apert and Crouzon syndromes (cranio- leukocyte tyrosine kinase (LTK) (Ben-Neriah and synostosis with or without syndactyly of the hands and Bauskin, 1988; Bernards and de la Monte, 1990; feet) (FGFR2) (Wilkie et al., 1995) and achondroplasia Maru et al., 1990; Krolewski and Dalla-Favera, 1991; (short-limbed dwar®sm and macrocephaly) (FGFR3) Snijders et al., 1993; Toyoshima et al., 1993; (Rousseau et al., 1994). Kozutsumi et al., 1994). Biochemical characterization We have recently described the positional cloning of the ALK polypeptide encoded by this cDNA, and and characterization of the t(2;5)(p23;q35) chromoso- of the receptor protein produced endogenously by mal rearrangement found in approximately 10% of all rhabdomyosarcoma skeletal muscle tumor cells, non-Hodgkin's lymphomas (Morris et al., 1994). This reveals that ALK is a large, glycosylated, single translocation produces a fusion gene that encodes a chain transmembrane receptor tyrosine kinase. In situ chimeric transforming protein comprised of the hybridization studies of mouse embryos and adult amino-terminal portion of the chromosome 5-encoded mouse organs indicate that the expression of the

Figure 1 Schematic representation (a) and deduced amino acid sequence (b) of a human ALK cDNA clone. (a) The thick bar represents the coding sequences of the 6226 bp insert of human ALK cDNA clone RMS17-2. The putative signal peptide (SP, hatched box), transmembrane (TM, solid box), and tyrosine kinase (TK, stippled box) domains are indicated. Cysteine residues (solid circles) and consensus N-glycosylation sites (inverted triangles) present in the extracellular domain are indicated. The thin open bars on either end represent 5' and 3' non-coding sequences. (b) Deduced amino acid sequence encoded by ALK cDNA clone RMS17-2. The putative signal peptide (amino acids 1 ± 26) is underlined. Cysteine residues within the extracellular domain are circled. Consensus N-glycosylation sites (N-X-S/T) are enclosed by open boxes. The transmembrane domain (residues 1031 ± 1058) is highlighted by a hatched box. Horizontal arrows ¯ank the tyrosine kinase catalytic domain (residues 1123 ± 1376). The amino- terminal and carboxy-terminal extent of the overlap of the largest reported receptor isoform of human LTK (Toyoshima et al., 1993) with ALK is indicated. The position at which ALK is truncated by the t(2;5) to produce NPM-ALK is also shown (`NPM- ALK fusion junction') ALK, a novel, neural receptor tyrosine kinase SW Morris et al 2177 homologue, Alk, is restricted to the nervous system, extracellular domain of ALK after signal peptide suggesting that the receptor binds a ligand(s) critical cleavage is 1004 amino acids long and is followed by for neural development and function. 28 hydrophobic amino acids, consistent with a membrane-spanning segment. This single putative transmembrane domain is ¯anked on its cytoplasmic side by basic amino acid residues typical of the Results junction between transmembrane and cytoplasmic domains (Blobel, 1980). Within the extracellular Cloning and characterization of a full-length ALK portion of ALK are 16 consensus N-glycosylation cDNA sites (Asn-X-Ser/Thr), predicting that the mature ALK Analysis of a 6226 bp normal ALK cDNA (clone protein is glycosylated in vivo. The spacing of cysteine RMS17-2) isolated from a rhabdomyosarcoma library residues within the extracellular domains of receptor identi®ed an ATG codon at nucleotides 912 ± 914, tyrosine kinases is thought to be important in followed by an open reading frame of 4857 bp and a determining ligand speci®city (Ullrich and Schles- 455 bp 3' untranslated region that terminated in a singer, 1990); in its extracellular region, ALK contains poly(A) tail. The nucleotide sequence surrounding the 26 cysteine residues with two clusters residing between ATG encoding the presumed ALK initiator methionine amino acids 425 and 487 (seven residues) and between (GGCGGGATGG) was in agreement with the amino acids 987 and 1021 (six residues). Consistent consensus translation initiation sequence reported by with the overall high degree of homology between Kozak 1987) and was preceded by two in-frame ALK and LTK (see below), the spacing of all except termination codons within 45 bp upstream of this the most N-terminal one of the 13 cysteine residues site. The poly(A) tail of this insert was preceded, 20 bp present in the overlapping extracellular portions of upstream, by the canonical polyadenylation signal each of these proteins is conserved (Figure 2). The AATAAA. The 3' sequences of all ALK clones ALK juxtamembrane segment is of typical length (64 isolated (see Materials and methods) were identical to amino acids) and contains a NPXY motif (residues those present in the NPM-ALK fusion cDNA (Morris 1093 ± 1096, NPNY), which is found in the cytoplasmic et al., 1994), suggesting that mutations of the truncated portion of a number of cell surface proteins, including receptor are not involved in activating the transforming the low-density lipoprotein receptor and RTKs of the potential of the chimera in non-Hodgkin's lymphoma. epidermal growth factor (EGF) and insulin receptor However, it should be noted that although we refer to subfamilies, and is thought to be involved in ligand- the ALK cDNA reported here as encoding the `normal' independent receptor internalization within coated pits full-length receptor, we cannot fully exclude the (Chen et al., 1990). The NPXY motif has also recently possibility that abnormalities of the receptor could be been shown to mediate the interaction of RTKs with present in the ALK protein expressed by rhabdomyo- signaling substrates such as the insulin receptor sarcoma tumor cells. substrate-1 (IRS-1) and Src homology and collagen The long open reading frame following the initiator (SHC) proteins through the substrate's phosphotyro- methionine of ALK encodes a polypeptide of 1620 sine binding (PTB) domain (van der Geer and Pawson, amino acid residues with a molecular mass of 177,155 1995); indeed, Fujimoto and colleagues have deter- daltons (Figure 1). The 26 amino-terminal residues of mined that the NPNY residues present in the the ALK polypeptide comprise a highly hydrophobic juxtamembrane segment of the truncated and consti- stretch of amino acids representing a putative signal tutively-activated ALK protein present in NPM-ALK peptide sequence (von Heijne, 1986). The predicted mediate the association of the fusion protein with IRS-

Figure 2 ALK is highly homologous to leukocyte tyrosine kinase (LTK). An amino acid sequence alignment of the extracellular, transmembrane and juxtamembrane segments of LTK with ALK is shown. Sequence identity is indicated by inverse lettering. This alignment was performed using the Genetics Computer Group, Inc. (GCG) BESTFIT program; amino acid similarity of greater (:) or lesser (.) degree was determined according to evolutionary distance as measured by Dayho and normalized by Gribskov Burgess (1986). The LTK residues used for alignment are from the published sequence of the largest reported human receptor isoform (Toyoshima et al., 1993); (GenBank accession no. D16105). Within the region shown, ALK and LTK exhibit 48% amino acid identity and 63% similarity. The kinase and carboxy-terminal regions of ALK and LTK (not shown) are also highly related, sharing 64% and 44% identity, respectively ALK, a novel, neural receptor tyrosine kinase SW Morris et al 2178 1 (Fujimoto et al., 1996). The 254-amino acid ALK (Figure 3, lane 6); this protein was eciently catalytic domain dontains the usual conserved motifs immunoprecipitated by our polyclonal anti-ALK found in all protein tyrosine kinases and possesses serum (not shown). To con®rm that the ALK cDNA paired tyrosine residues (ALK amino acids 1282 and which we had isolated encodes a full-length receptor, 1283) characteristic of the major autophosphorylation we subcloned the entire 6226 bp insert of RMS17-2 site of the insulin receptor subfamily (Hanks et al., into the pcDNA3 mammalian expression vector 1988). The ALK carboxy-terminal tail is unusually (Invitrogen, San Diego, CA) for in vivo expression long (244 amino acids), compared with that of other studies. The resulting plasmid, pcDNA3-ALK, was receptor kinases and, like the juxtamembrane and electroporated into COS-7 cells, which after 72 h were kinase domains, contains several potential phosphotyr- metabolically labeled with [35S]-methionine and sub- osine residues that could serve as binding sites for jected to immunoprecipitation analysis using polyclo- substrates with SH2 or PTB domains (see Discussion). nal ALK antibody. As shown in Figure 3 (lane 3), the Analysis of protein sequence databases con®rmed expression of pcDNA3-ALK in COS-7 yielded a single the identity of ALK as a member of the insulin immunoreactive protein of approximately 200 kDa, receptor (IR) subfamily, with closest homology to the slightly larger than the 177 kDa in vitro transcription leukocyte tyrosine kinase (LTK) receptor (Ben-Neriah and translation product derived from ALK cDNA and Bauskin, 1988; Bernards and de la Monte, 1990; clone RMS17-2 but identical in size to the endogenous Maru et al., 1990; Krolewski and Dalla-Favera, 1991; ALK protein immunoprecipitated from the Rh30 Snijders et al., 1993; Toyoshima et al., 1993; Kozutsumi et al., 1994). Within its kinase domain, ALK shares 64% residue identity with LTK, compared to 36% to 38% with IR subfamily members such as the TRK receptors, ROS, the insulin-like growth factor-1 receptor (IGF-1R), and the IR itself. Over A3 their complete extent of overlap, ALK and the considerably smaller LTK (864 amino acids; Toyoshi- ALK ma et al., 1993) share 57% amino acid identity and 71% similarity. Only closely related RTKs exhibit

sequence similarity in domains other than the kinase T/T

region; for example, in their extracellular region highly cDNA related RTKs such as the IR and the IGF-1R, the in vitro ALK COS–7 / “empty “ pcDN COS–7 / pcDNA3– TRK receptors, and the EGF receptor and HER-2/ Rh30 NEU have sequence identities on the order of 35% to IIIPI PI — 45% (Fantl et al., 1993). Thus, the signi®cant kDa relationship between ALK and LTK is even better illustrated by the observation that within their

ALK

extracellular, transmembrane, and juxtamembrane ¨ gp200 200 — ¨ regions, they exhibit 48% identity and 63% similarity p177ALK (see Figure 2), and over their carboxy-terminal tails they possess an identity of 44% and similarity of 59% within their region of overlap. Thus, based on their relatedness, it appears that ALK and LTK comprise an 97.4 — RTK subclass within the IR family similar to the IR/ IGF-1R, the TRK receptors, and the MET/RON/SEA receptor subgroups (Fantl et al., 1993; Hu et al., 69 — 1993; Gaudino et al., 1994). In our initial report of the positional cloning of NPM-ALK (Morris et al., 1994), we incorrectly stated that sequences `from the extreme 5' ends of the ALK 46 — clones (referring to RMS4 and RMS17-2) were 50% identical to sequences encoding insulin-like growth factor binding protein (IBP-1)'. The alignments from which we drew this conclusion were performed using 123456 Figure 3 Expression of ALK in COS-7 cells and in the human suboptimal sequence that was ultimately determined to 35 be located within the 5' untranslated portion of ALK. rhabdomyosarcoma cells line Rh30. S-methionine-labeled cell lysates from COS-7 cells electroporated with the mammalian Comparison of the normal ALK amino acid sequence expression vector pcDNA3 (lane 1) or with a pcDNA3 construct reported here with IBP-1 revealed no signi®cant containing the full-length ALK cDNA insert from clone RMS17-2 homology between the two. (lanes 2 and 3), and from the Rh30 cell line (lanes 4 and 5), were incubated with either preimmune (PI) or immune anti-ALK (I) serum. Immunoprecipitates were resolved by SDS ± PAGE on a Biochemical characterization of ALK 7.5% polyacrylamide gel under reducing conditions. The in vitro transcription/translation reaction product of ALK cDNA clone In vitro transcription and translation of the ALK RMS17-2 was loaded directly and run in parallel on the same gel cDNA clone RMS17-2 yielded a single, prominent (lane 6). Note that the ALK cDNA product expressed in COS-7 product whose migration in SDS ± PAGE under cells appears to be correctly modi®ed in vivo and co-migrates with the endogenous Rh30 ALK receptor. The relative mobilities of reducing conditions was consistent with the predicted the mature 200 kDa ALK glycoprotein (`gp200ALK') and the molecular mass for the ALK polypeptide of 177 kDa 177 kDa ALK polypeptide backbone (`p177ALK') are indicated ALK, a novel, neural receptor tyrosine kinase SW Morris et al 2179 rhabdomyosarcoma cell line (lane 5). This 200 kDa polypeptide core and the mature receptor results from protein was not detected in COS-7 cells electroporated the addition of N-linked carbohydrate moieties. These using `empty' pcDNA3 vector (lane 1), nor in control data also indicate that ALK does not contain a immunoprecipitations using preimmune serum (lanes 2 signi®cant amount of O-linked glycosyl groups. and 4). To further examine the in vivo processing of ALK, The dierence in the size of the ALK protein that we we performed a series of pulse-chase labeling experi- identi®ed in vivo compared with the ALK product ments using Rh30 cells. Following incubation for resulting from in vitro transcription/translation of the 30 min in growth medium lacing methionine, the cells RMS17-2 clone suggested that post-translational mod- were pulse-labeled with [35S]-methionine for 15 min, i®cations of the ALK polypeptide backbone occur then chased with medium containing excess cold during synthesis of the mature receptor. Since the methionine and lysed for immunoprecipitation analy- deduced amino acid sequence of ALK contains multiple sis at various time points. As illustrated in Figure 4b, sites for potential N-linked glycosylation, we investi- glycosylated 200 kDa ALK protein was identi®ed even gated whether the mature 200 kDa receptor is a immediately after the end of the 15 min 35S-methionine- glycoprotein by performing metabolic labeling experi- labeling period (lane 3). Since no proteins of lower ments with the glycosylation inhibitor tunicamycin molecular mass were recognized by our polyclonal (Figure 4a). Preincubation of Rh30 cells with a ALK antiserum, it is likely that nascent polysome- concentration of tunicamycin expected to totally inhibit bound ALK polypeptides become glycosylated during glycosylation (25 mg/ml) revealed the presence of a single their synthesis while being transported into endoplas- immunoreactive polypeptide (lane 3) identical in size to mic reticulum (ER). Further oligosaccharide processing the 177 kDa in vitro transcription/translation product of of ALK presumably then occurs within The Golgi the ALK cDNA (lane 1) (excluding the small size apparatus, before the protein is transported to the cell dierence due to the absence of the ALK signal membrane (Kornfeld and Kornfeld, 1985), resulting in sequence cleaved from the protein in vivo). Thus, it a mature ALK glycoprotein of essentially the same appears that the size dierence between the ALK molecular mass as the immature ER glycoprotein form,

a /T b A T cDN T/T

cDNA Rh30 Rh30 in vitro ALK Rh30 Duration of

in vitro ALK I I 0 – 0 0.25 0.5 1 2 3 4 5 chase (hrs)

kDa kDa

200 — ¨ gp200ALK

¨ p177ALK

200 — ¨ gp200ALK ¨ p177ALK 97.4 —

69 —

97.4 — 123 Tunicamycin – 0 25 (µg/ml) 69 —

12345678910 PI – ¨ Immune ¨ Figure 4 Glycosylation of the mature ALK receptor. (a) Comparison of the in vitro transcription/translation product of ALK cDNA claone RMS17-2 (lane 1) with the endogenous ALK protein expressed by the Rh30 cell line (lanes 2 and 3). The Rh30 cells analysed in sample lane 3 were incubated with 25 mg/ml tunicamycin for 30 min prior to, and for the 30 min period of, 35S- methionine labeling to inhibit glycosylation of ALK. Proteins were resolved by 6% polyacrylamide SDS ± PAGE under reducing conditions, then exposed to ®lm overnight at 7808C. I - anti-ALK serum. Note that the migration of the unglycosylated ALK polypeptide (lane 3) from Rh30 cells is identical to that of the ALK protein produced by in vitro transcription/translation (lane 1). (b) Pulse-chase metabolic labeling of ALK in Rh30 cells. Rh30 cells were pulsed with [35S]-methionine for 15 min, then chased with non-radioactive methionine for the indicated periods of time. Cell lysates were immunoprecipitated with preimmune serum (PI) or polyclonal anti-ALK serum (`immune') and analysed by 6% polyacrylamide SDS ± PAGE under reducing conditions. The 177 kDa in vitro transcription/translation product of ALK cDNA RMS17-2 was run on the same gel for comparison ALK, a novel, neural receptor tyrosine kinase SW Morris et al 2180 although this supposition remains to be formally tested Activation of the catalytic kinase function of RTKs by analysis with inhibitors of ER/Golgi glycoprotein is normally induced by ligand-mediated dimerization of transport, oligosaccharide processing enzyme inhibi- the receptors, resulting in intermolecular cross-phos- tors, or speci®c glycosidases. Although the possibility phorylation (Ullrich and Schlessinger, 1990). This exists that ALK could undergo a very prolonged ER/ process can be mimicked by receptor-speci®c bivalent Golgi transit period that exceeds the ®ve hour chase antibody, as has been demonstrated in in vitro duration used in these studies, the production of late- immunocomplex kinase assays for a number of appearing, Golgi-processed, glycosylated forms of the RTKs. To determine whether the ALK protein receptor that are of signi®cantly dierent molecular encoded by our cDNA is indeed a functional kinase, mass compared to the glycoprotein identi®ed in these we incubated recombinant gp200ALK immunoprecipi- 35S-methionine-labeling experiments can be excluded tated from COS-7 cells with [g-32P]ATP in the presence because the approximately 200 kDa receptor was the of Mn and Mg ions. As shown in Figure 7 (lane 3), only form of ALK identi®ed in cell surface labeling experiments (see Figure 6), immunocomplex kinase assays (Figure 7), and anti-ALK immunoblots (not shown) that should detect steady-state levels of the A3 mature, fully processed receptor. Other receptors in the insulin RTK subfamily have been shown to possess varied structural characteristics of ALK their mature forms, although each is initially translated as a single polypeptide produced from a single gene locus. For example, the insulin and insulin-like growth factor-1 receptors possess a disul®de-linked heterotetra-

meric a2b2 structure, the MET and RON RTKs exist as COS–7 / ”empty” pcDN COS–7 / pcDNA3- ab disul®de-linked heterodimers, and the mature TRK Rh30 IPIII I RTKs (TRK-A, -B, and -C) are present in the cell kDa membrane as monomeric proteins (Barbacid, 1993; Fantl et al., 1993; Gaudino et al., 1994). As noted above, the ALK receptor immunoprecipitated in metabolic labeling studies from both COS-7 cells electroporated with the pcDNA3-ALK expression

construct and from the Rh30 cell line is present as an 200 — ¨ gp200ALK apparently single prominent 200 kDa glycoprotein when analysed in SDS ± PAGE under reducing conditions; no larger precursor molecules that might subsequently be post-translationally cleaved into subunit proteins, nor smaller bands that could represent subunit fragments that are normally linked via disul®de bonds, were seen in 97.4 — either steady-state (Figures 3 and 4a) or pulse-chase labeling studies (Figure 4b). Further evidence that the mature ALK receptor does not possess a disul®de-linked 69 — multi-subunit structure (as found with the IR, IGF-1R, MET and RON RTKs) was provided in metabolic labeling experiments in which the ALK receptor observed under non-reducing conditions co-migrated with the 200 kDa glycoprotein identi®ed in the reduced state (Figure 5). These observations indicate that, like the 46 — TRK RTKs, ALK exists in its mature form as a single chain receptor with, based upon its deduced amino acid sequence, a single transmembrane segment. 12345 To provide biochemical evidence for the hypothesis that ALK encodes a membrane-spanning receptor, we reduced non– performed biotinylation studies with COS-7 cells that reduced had been electroporated with pcDNA3-ALK to Figure 5 The mature ALK receptor has a monomeric structure. Anti-ALK immunoprecipitates from detergent lysates of cells selectively label cell surface proteins. Intact COS-7 labeled with [35S]-methionine were resolved by 7.5% polyacryla- cells in tissue culture dishes were incubated with biotin- mide SDS ± PAGE under reducing (lanes 1 ± 3) or non-reducing CNHS-ester using conditions previously shown to label (lanes 4 and 5) conditions on the same gel. Reduced and non- cell surface-exposed proteins only (Meier et al., 1992), reduced samples were separated by multiple lanes to avoid diusion of b-mercaptoethanol; visualization of non-reduced the biotinylation reaction quenched and the cells lysed. immunoglobulin in lanes 4 and 5 of the Coomassie-stained gel As shown in Figure 6, immunoprecipitation of the (not shown) served as an internal control to exclude diusion. lysates prepared from these cells using polyclonal ALK COS-7/`empty' pcDNA3, COS-7 cells electroporated with the antibody readily detected the mature 200 kDa ALK mammalian expression vector pcDNA3; COS-7/pcDNA3-ALK, glycoprotein, indicating that the ALK receptor COS-7 electroporated with a pcDNA3 construct containing the ALK cDNA insert of RMS17-2; Rh30, Rh30 rhabdomyosarcoma encoded by our cDNA is correctly inserted into the cell line. Note that the ALK glycoprotein migrates as a single cell membrane, with exposure of the putative ligand- prominent 200 kDa band whether analysed under reducing or binding domain extracellularly. non-reducing conditions ALK, a novel, neural receptor tyrosine kinase SW Morris et al 2181 ALK was eciently autophosphorylated under the in ranging in age from day 10 postconception to 4 days vitro kinase conditions used; assays using endogenous postpartum. No signal was detected in the day-10 ALK protein immunoprecipitated from Rh30 cells embryos. After day 10, the in situ results indicated that produced similar results (not shown). Phosphoamino Alk transcripts are restricted to the central and acid analysis demonstrated that the in vitro autopho- peripheral nervous systems. In day 11 embryos, signal sphorylation of ALK was restricted to tyrosine residues was detected in the ventral aspect of the primitive spinal (Figure 7b). Sequential anti-ALK immunoprecipita- cord and in the developing trigeminal and facial ganglia tion/anti-phosphotyrosine immunoblotting of lysates (not shown). By day 12, there was additional expression prepared from NIH3T3 cells stably expressing our in the periventricular zone around the third ventricle; transfected cDNA and from Rh30 cells revealed no although the primitive neuroepithelial cells in this evidence of in vivo autophosphorylation of the region were generally negative, one focal area in this receptor, suggesting that no activating mutations are zone showed positive cells in the region of the cerebral present in the ALK rhabdomyosarcoma cDNA that we aqueduct (not shown). In later stages of embryonic have isolated nor is there an autocrine loop present in development, the expression pattern of Alk became the cells (not shown). more evident in the developing thalamic nuclei and other nuclear groups surrounding the third ventricle. The complex pattern of positive signals illustrated in Analysis of the expression of the mouse homologue Alk Figure 8a ± c (coronal sections through a day-14 by in situ hybridization embryo) indicates the localization of Alk transcripts in As an initial step to elucidate the normal functions of the receptor, we performed Alk in situ hybridization studies on adult mouse tissues as well as mouse embryos A3 ALK 3 ALK COS–7 / “empty” pcDN COS–7 / pcDNA3- a SUP-M2 b IPI I PI I kDa +

COS-7/”empty” pcDNA COS-7/ pcDNA3- ¨ ALK I PI I 200 — gp200 kDa

97.4 — pH3.5

¨ NPM-ALK p75 ¨

+ pH1.9 ¨ ALK 69 — 200— gp200

46 —

97.4— 12345 Figure 7 gp200ALK possesses in vitro tyrosine kinase activity. (a) 69— Kinase activity in immunoprecipitates obtained by incubating lysates from COS-7 cells electroporated with either pcDNA3 vector alone (lane 1) or pcDNA3-ALK (lanes 2 and 3), or (as a positive control) the t(2;5)-positive lymphoma cell line SUP-M2 that expresses NPM-ALK (lanes 4 and 5) with preimmune (PI) or immune (I) anti-ALK serum. Immunocomplexes were incubated with [g-32P]ATP and analysed by 7.5% polyacrylamide SDS ± 46— PAGE performed under reducing conditions as described in the Materials and methods. A one minute autoradiographic exposure performed at room temperature is shown. The identity of the two phosphorylated proteins with faster mobility relative to p75NPM ± 123 ALK is unknown but they may represent NPM ± ALK proteolytic degradation fragments and/or proteins that are known to be Figure 6 Exposure of gp200ALK at the cell surface. Intact COS-7 associated with, and phosphorylated by, NPM-ALK such as SHC cells previously electroporated with either the pcDNA3 mamma- (Fujimoto et al., 1996; Bischof et al., in press). (b) In vitro lian expression vector alone (lane 1), or a pcDNA3 construct phosphorylated gp200ALK was cut from the polyacrylamide gel containing the RMS17-2 ALK cDNA (lanes 2 and 3), were and eluted, and its phosphoamino acid composition determined surface-labeled by biotinylation and the cells lysed in RIPA buer by two-dimensional thin-layer electrophoresis. pS, phosphoserine; subsequent to quenching of the labeling reaction. The lysates were pT, phosphothreonine; pY, phosphotyrosine. The sample origin, incubated with preimmune (PI) or anti-ALK (I) serum, proteins as well as partially hydrolyzed peptides that run near the origin, were resolved by 7.5% polyacrylamide SDS ± PAGE, transferred are not illustrated but were located toward the lower righthand to a PVDF membrane, and detected by chemiluminescence as portion relative to the ®gure; the directions of electrophoresis and described in Materials and methods their pH are as indicated ALK, a novel, neural receptor tyrosine kinase SW Morris et al 2182 discrete populations of primitive neuronal cells clustered both the embryo and adult as shown in Figure 8i-j), around the third ventricle. Intense signal was also which illustrate expression in the ganglion cell identi®ed in the region of the developing olfactory population of the stomach wall. nerves as they course through the area of the nasal cavity. Regardless of the developmental stage, Alk transcripts could not be detected in the outer cortical Discussion layers, basal ganglia, cerebellum, choroid plexus, or ependymal lining. The spinal cord in the day-14 embryo The results of our analysis indicate that ALK is most continued to show expression in the developing ventral similar to members of the insulin receptor (IR) horns, whereas no expression was identi®ed in the spinal subfamily of receptor tyrosine kinases. For example, ganglia (Figure 8d). Day 15-16 and older embryos its putative autophosphorylation sites (tyrosine residues continued to show expression in the brain primarily in 1282 and 1283) are of the `YY' motif characteristic of the region of the thalamic nuclei, as well as in cranial the kinases that belong to this subfamily (Hanks et al., ganglia, the sympathetic chain, and the ventral spinal 1988), and its homology to the previously described cord (Figure 8e-g). subfamily member leukocyte tyrosine kinase (LTK) is Alk expression was not detected at any develop- quite striking (Figure 2). The relatedness of ALK and mental stage in any non-neural tissues, including the LTK (57% identity and 71% amino acid similarity liver, lungs, thymus, kidney, adrenal glands, pancreas, over their entire region of overlap) indicates that these connective tissue, muscle, or bone. Occasional bands of two proteins have likely evolved from a common intense signal were observed in the pelvic region, ancestral gene; together, they appear to constitute an extremities, tongue, and facial region, which we IR family subclass analogous to the IR/IGF-IR, the interpret to represent developing peripheral nerves. TRK receptors and MET/RON/SEA (Fantl et al., Newborn and adult mouse brain also exhibited Alk 1993; Hu et al., 1993; Gaudino et al., 1994). Although expression restricted predominantly to the thalamic initially reported as a transmembrane protein devoid of region (Figure 8h). Signal was also detected in the an extracellular domain (Ben-Neriah and Bauskin, hypothalamic region, but the basal ganglia, cerebellum, 1988), LTK cDNA clones encoding two forms of a and cortex continued to show no signi®cant Alk more typical receptor tyrosine kinase have more expression. Finally, the peripheral enteric nervous recently been characterized (Krolewski and Dalla- system also showed evidence of Alk transcripts in Favera, 1991; Snijders et al., 1993; Toyoshima et al., ALK, a novel, neural receptor tyrosine kinase SW Morris et al 2183 1993). In addition, other LTK cDNAs containing endogenous LTK in any tissue except by immunostain- alternatively spliced exons that disrupt the open ing of sections, apparently because of low expression reading frame of the receptor form of the protein levels (Bernards and de la Monte, 1990; Krolewski and have been isolated and predict a soluble receptor Dalla-Favera, 1991; Toyoshima et al., 1993). The ALK devoid of both the transmembrane and tyrosine kinase cDNA which we have characterized contains in its domains as well as a membrane-spanning receptor that extracellular domain residues corresponding to the lacks the catalytic domain (Toyoshima et al., 1993). largest LTK receptor isoform (Snijders et al., 1993; Unfortunately, the determination of the functional Toyoshima et al., 1993), including amino acids encoded importance of these various predicted LTK proteins, by an alternatively spliced LTK exon (homologous to as well as the elucidation of the normal function of ALK residues 878 ± 939). The region of overlap in the LTK in growth and development in general, has been extracellular domains of ALK and LTK contains no hindered because of the inability of a number of immunoglobulin loop or ®bronectin type III repeat laboratories to readily detect in vivo expression of motifs as are commonly found in the extracellular

Figure 8 Localization of Alk mRNA by in situ hybridization to both the developing and mature central and peripheral nervous systems. (a and b) show bright-®eld and dark-®eld photomicrographs, respectively, of coronal sections through a day 14 mouse embryo brain. These sections show the third ventricle in the center surrounded by the developing thalamic nuclei. The lateral ventricles are visible to the right and left of the illustration. In (b) the bright white grains indicate the localization of the Alk transcripts. There is a complex pattern of positive cells around the third ventricle. No signal is present in the cortical layer (c) around the lateral ventricles. The intense signal (arrowheads) adjacent to the nasal cavity (*) is present in the region of the developing olfactory nerves. (c) shows a negative control section hybridized with the sense probe for comparison. (d) is a cross section through the developing spinal cord of a day 14 embryo. There is a strong signal in the ventral horns of the spinal cord. No signal is present in the adjacent spinal ganglia (*). (e) shows a section through a day 15 embryo. The signal is localized to the facial (VII) ganglia adjacent to the petrous portion of the temporal bone (*). (f) is a section through the thoracic region of a day 15 embryo. The Alk transcripts are localized to the sympathetic chain (arrows) along the posterior wall of the thoracic cavity. No signal is present in the lung (l) or in the adjacent connective tissues or ribs (*). (g) is a longitudinal section through a day 16 embryo. The signal is localized to the ventral aspect of the spinal cord. Note the negative lung tissue present in the upper portion of the panel. (h) is a coronal section through the brain of an adult mouse. There is intense signal in selected thalamic nuclei around the third ventricle (*). The black arrowhead shows signal localized to the region of the nucleus reuniens thalami. The white arrow shows expression in the region of the reticular thalamic nucleus. There is also signal in the periventricular hypothalamic nucleus (white arrowhead). (i) is a section through the stomach of a day 15 embryo. The Alk mRNA is localized to the developing ganglia (arrow) in the muscularis propria. (j) shows the stomach from an adult mouse with signal localized to the ganglia (arrow) in the muscularis propria. The light-colored circular defects seen in b and c represent technical artifacts of the in situ hybridization procedure. (a, b, c, g, h Mag. 650; d, e, f, i, j 6120) ALK, a novel, neural receptor tyrosine kinase SW Morris et al 2184 segments of some receptor tyrosine kinases (Fantl et Werner et al., 1992), membrane-bound receptors al., 1993). A striking feature of this region, but one of lacking their catalytic domain (Prat et al., 1991; uncertain signi®cance, is the conservation between the Barbacid, 1993), or tissue-speci®c receptor forms two proteins of several short glycine-rich segments possessing dierent ligand-binding and biological (Figure 2). These segments bear some resemblance to activities (Miki et al., 1992; Werner et al., 1992) has the usually more extensive glycine-rich regions that are been demonstrated for various RTKs. Interestingly, found in some structural proteins of the epidermis recent immunostaining studies of normal human tissues (e.g., loricrin and keratins 1 and 10) (Steinert et al., using an anti-ALK monoclonal antibody generated 1985; Zhou et al., 1988; Hohl et al., 1991), a number of against cytoplasmic residues 1359 ± 1460 of the human RNA-binding proteins (Burd and Dreyfuss, 1994), and receptor identi®ed weak ALK protein expression certain plant cell wall proteins (Condit and Meagher, restricted to the nervous system, with no apparent 1986) and that are predicted to convey ¯exibility to the expression in testis, placenta or fetal liver (Pulford et supramolecular structure of the proteins containing al., 1997). them. Whereas LTK mRNA transcripts are normally The cytoplasmic molecules that mediate downstream expressed in neuronal cells, placenta, pre-B lympho- signaling by ALK are presently incompletely known. A cytes, and some mature B and T cells (Ben-Neriah and number of receptor tyrosine kinase substrates are Bauskin, 1988; Bernards and de la Monte, 1990; Maru known to interact with activated receptors through et al., 1990), ALK is not normally expressed in their SRC homology region 2 (SH2) domains, binding hematopoietic cells. Given the high degree of to a speci®c phosphotyrosine residue contained within similarity between ALK and LTK, we suggest that a distinct amino acid sequence of the receptor. That the constitutively activated cytoplasmic portion of speci®c phosphotyrosine-containing sequences bind to ALK present in NPM-ALK in t(2;5) ± positive a given substrate SH2 domain has now been con®rmed lymphomas may signal via LTK pathway substrates for several signaling molecules by mutational studies, to transform lymphocytes. including phosphatidylinositol 3-kinase (PI3-K), Ras We have previously shown by Northern analysis that GTPase-activating protein (RasGAP), phosphoinosi- in addition to human mRNA transcripts of roughly tide-speci®c phospholipase C-type g (PLC-g), and 6.5 kb (presumably corresponding to the cDNA growth factor receptor-bound substrate 2 (GRB-2), described here), an 8.0 kb ALK message is expressed among others (Pawson and Gish, 1992; Mayer and in rhabdomyosarcoma, small intestine (in the enteric Baltimore, 1993). Furthermore, the consensus sequence innervation), and brain (Morris et al., 1994). No motifs recognized by the SH2 domains of most of the variant cDNA clones potentially representing the other known receptor tyrosine kinase substrates have 8.0 kb ALK transcript were isolated in our screening been predicted based upon techniques using degenerate of two human rhabdomyosarcoma cDNA libraries, phosphopeptide libraries (Zhou et al., 1993; Songyang perhaps because this transcript is signi®cantly less et al., 1994). Analysis of the amino acid sequence of abundant than the approximately 6.5 kb message. In the cytoplasmic portion of ALK reveals potential sites our prior studies, we also identi®ed ALK messages of for interaction with the SH2 domains of PI3-K approximately 6.0 kb in human testis, placenta, and (YXXM; ALK residues 1359 ± 1362, YRIM), RasGAP fetal liver and a 4.4 kb transcript unique to testis (YMPX; ALK residues 1327 ± 1330, YMPY), and (Morris et al., 1994). Interestingly our mouse in situ PLC-g (E/IXXXXY; ALK residues 1277 ± 1282, hybridizations, performed with an Alk juxtamembrane IYRASY and 1322 ± 1327, IFSLGY). In addition, probe that is 96% nucleotide identical to the ALK contains two NPXY motifs (residues 1093 ± corresponding human ALK segment and that hybri- 1096, NPNY and 1504 ± 1507, NPTY) which have been dizes to these non-neural human ALK transcripts in shown to mediate the binding of the substrates SHC Northern blots (SW Morris, unpublished data), failed and IRS-1 to receptor tyrosine kinases via the substrate to identify Alk expression in mouse testis or fetal liver phosphotyrosine-binding (PTB) domain (or more (mouse placenta was not examined). Hybridization of correctly in the case of IRS-1, a PTB-like domain). this probe to an adult mouse tissue Northern blot The PTB domain is a recently described non-SH2 motif containing mRNA from heart, brain, spleen, lung, also capable of speci®cally binding phosphotyrosine- liver, skeletal muscle, kidney and testis (Multiple Tissue containing peptides (van der Geer and Pawson, 1995). Northern blot no 7762-1, Clontech, Palo Alta, CA) Although we have not yet analysed the role of various revealed very weak expression of Alk limited to the RTK substrates in the ALK signal transduction brain (not shown). It is possible that the human 4.4 kb pathway, we have performed co-immunoprecipitation and 6.0 kb ALK transcripts may serve a function(s) studies of the constitutively active NPM-ALK fusion unique to the tissues in which they are found. We are protein and the substrates PI3-K, PLC-g, and SHC in currently attempting to isolate and characterize clones the t(2;5)-positive lymphoma cell line SUP-M2; each of corresponding to these transcripts using PCR, as well these signaling molecules is physically associated with as conventional hybridization, screening of appropriate NPM-ALK, which contains the entire cytoplasmic cDNA libraries. These dierent transcripts may, of portion found in the normal receptor (Bischof et al., course, result from alternative transcriptional start sites in press, and SW Morris, manuscript submitted). It is and/or polyadenylation signals; however, given the therefore likely that the normal ALK receptor signals homology between ALK and LTK, and the demon- in part via these substrates. In a recently published stration of the extensive alternative splicing of LTK, it report Fujimoto and colleagues addressed the role of is tempting to speculate that ALK messages encoding IRS-1 and SHC in NPM-ALK-mediated NIH3T3 human tissue-speci®c receptor isoforms could exist. ®broblast transformation by using mutant constructs The generation by dierential splicing of soluble in which one or both tyrosine residues within the two receptors (Johnson et al., 1990; Petch et al., 1990; ALK NPXY motifs were changed to phenylalanine ALK, a novel, neural receptor tyrosine kinase SW Morris et al 2185 (Fujimoto et al., 1996). In these experiments, IRS-1 derived neurotrophic factor, neurotrophin-3, or was shown to bind to the phosphorylated NPM-ALK neurotrophin-4) or a novel neurotrophic ligand. tyrosine residue 156 (corresponding to ALK residue Chemical cross-linking studies using radioiodinated 1096) whereas SHC bound NPM-ALK phosphotyr- preparations of the known neurotrophins to examine osine 567 (corresponding to residue 1507 of the normal receptor expression in protein lysates prepared from receptor). Interestingly, both of the single point mutant mouse and chicken neuronal tissues have failed to chimeras, as well as a double mutant that was unable identify ligand-receptor complexes of the greater than to bind either IRS-1 or SHC, transformed NIH3T3 200 kDa size that would be expected based on our eciently, indicating that neither substrate is essential analysis of the mature human ALK receptor, lending for the oncogenic conversion of ®broblasts by NPM- support for the interaction of the receptor with a ALK. In addition, all three point mutant chimeras novel factor (Escandon et al., 1994). In addition, the were demonstrated to eciently interact with GRB2 extracellular portion of ALK shows minimal homol- despite the absence in NPM-ALK of a consensus ogy with the TRK family receptors and contains a GRB2 binding site. We have also examined the dierent pattern of cysteine residue spacing, suggesting interaction of SHC with NPM-ALK in studies using dierent ligand speci®cities. Our in situ hybridization an engineered mutant, DC154NPM-ALK, that lacks its results suggest that the human ALK transcripts that carboxy-terminal 154 residues, including the NPTY we have previously identi®ed by Northern analysis in motif at residues 564 ± 567 (Bischof et al., in press). In small intestine, and that are also weakly present in agreement with the data of Fujimoto and co-workers, colon and prostate (Morris et al., 1994), most DC154NPM-ALK was unable to interact with SHC. In probably re¯ect ALK transcription in the neural addition, despite its large carboxy-terminal deletion, structures supplying these tissues as opposed to the this NPM-ALK mutant protein transformed rodent tissue parenchyma itself. Additional examination of ®broblasts almost as eciently as the wild-type fusion the expression of ALK transcripts in speci®c human protein. In related studies, we have also demonstrated tissues, as well as the phenotypic analysis of mice that wild-type NPM-ALK can render the IL-3 lacking functional Alk receptors, both currently in dependent cell line 32D factor-independent (SW progress, should help to more fully elucidate the role Morris, unpublished data). Because these mouse of this RTK in mammalian development and growth. myeloid cells express neither IRS-1, nor the IRS-1 related molecule 4PS (Wang et al., 1993), it can be inferred that cytoplasmic signaling by normal ALK may not require these substrates. Lastly, it should also Materials and methods be noted that the cytoplasmic region of ALK contains several possible tyrosine phosphorylation sites that Isolation and nucleotide sequencing of ALK cDNA clones reside in motifs dierent from already described SH2 As we have previously reported, ALK transcripts of or PTB domain binding sequences, suggesting that the approximately 6.5 and 8.0 kb are expressed at moderately receptor may bind novel RTK substrates or bind abundant levels in cell lines derived from the skeletal known substrates through unique motifs. muscle tumor rhabdomyosarcoma (Morris et al., 1994). To We demonstrate here that the expression of the isolate normal ALK cDNA clones, we screened two cDNA mouse homologue Alk appears to be restricted to libraries (made with either oligo(dT) or random hexamer neural tissue. Although we have not as of yet analysed primers for ®rst-strand synthesis) that had been prepared + the expression of the Alk protein in the mouse, we with poly(A) RNA from the rhabdomyosarcoma line Rh30 (Shapiro et al., 1993). Initial screens, performed with have demonstrated in immunostaining studies of a1.27kbSacIcDNAfragmentfromthe3'portion of the normal adult human tissues that ALK is restricted NPM ± ALK cDNA (previously designated pS1.2) (Morris to the nervous system and is present at low levels et al., 1994), yielded seven overlapping clones, including (Pulford et al., 1997). These studies identi®ed weak three with inserts larger than 4.5 kb (clones RMS 1, 4 and labeling with an ALK monoclonal antibody of some 17-2), from the oligo(dT)-primed library. The largest insert endothelial cells, pericytes, and scattered glial cells in (6226 bp from clone RMS17-2) was sequenced in its the hypothalamus, basal ganglia, cerebral cortex, and entirety. The nucleotide sequence of the other two large cerebellum. Low level expression of ALK was also insert clones was determined partially and was identical to noted in scattered neuronal cells in the hypothalamus, that of RMS17-2. The insert of the largest clone (RMS15- thalamic nuclei and basal ganglia. The pattern of Alk 3, insert size-3.5 kb) isolated by screening the random- primed Rh30 library with a 1.94 kb SacI ALK cDNA mRNA expression within the central nervous system fragment (designated pS1.9) that is located immediately appears to be essentially non-overlapping with that of upstream of pS1.2 was sequenced completely, and was also Ltk, which is expressed in post-term and adult found to be identical in its region of overlap (the most 5' cerebral cortex and hippocampus, but not in extent of the full-length cDNA) to the RMS17-2 nucleotide embryonic mouse brain (Bernards and de la Monte, sequence. 1990). Alk shares a partially overlapping expression Double-stranded DNA templates were sequenced using pattern with the genes encoding the Trk family of dye-terminators and Taq sequencing methods, as recom- neurotrophin receptors, the Trk-C expression pattern mended by the manufacturer (Perkin Elmer/Applied Biosys- being most similar (with expression in the spinal cord tems, Inc. [PE/ABI], Norwalk, CT). Samples were and enteric nervous system like Alk), but still with electrophoresed and analysed on PE/ABI 373 and 373 Stretch DNA sequencers. Contig assembly was performed substantial dierences both temporally and spatially using Staden's X-windows software, and the consensus (Klein et al., 1990; Martin-Zanca et al., 1990; sequence was analyzed using the Wisconsin Package v. 8.0 Barbacid, 1993; Lamballe et al., 1994). Our data software (Genetics Computer Group, Inc., Madison, WI) and raise the possibility that Alk could bind one of the National Center for Biotechnology resources (BLAST, identi®ed neurotrophins (nerve growth factor, brain- FARFETCH, etc.). ALK, a novel, neural receptor tyrosine kinase SW Morris et al 2186 Construction and analysis of an ALK expression plasmid modi®cations. Brie¯y, cells were washed twice with warm phosphate-buered saline, once at room temperature in The complete 6226 bp SalI/NotI ALK cDNA insert from biotinylation buer (10 mM Na borate, pH 8.8, 150 mM RMS17-2 was subcloned into the EcoRV/NotI-digested NaCl), and then incubated for 15 min at room temperature CMV promoter-based mammalian expression vector in biotinylation buer containing 50 mg/ml D-biotinyl-S- pcDNA3 (Invitrogen, San Diego, CA) following Klenow aminocaproic acid N-hydroxysuccinimide ester (biotin- polymerase ®ll-in of the SalI overhang. In vitro transcrip- CNHS-ester) (Boehringer-Mannheim, Indianapolis, IN). The tion and translation with this construct was performed labeling reaction was stopped by adding NH Cl to a ®nal using the T T coupled reticulocyte lysate system (Promega, 4 N concentration of 10 mM and washing the cells extensively in Madison, WI) and T7 polymerase. Transient expression wash buer (50 mM Tris-HCl, pH 7.4, 25 mM KCl, 5 mM studies of this ALK plasmid and pcDNA3 `empty' vector MgCl , and 1 mM EGTA) prior to lysis in RIPA buer. (as a negative control) in COS-7 monkey kidney epithelial 2 Immunoprecipitations were performed exactly as detailed cells were performed by electroporation in 0.4 cm cuvettes above, except that lysates were pre-cleared prior to the (960 mF, 250V). Cells were used for experiments at 72 ± addition of preimmune or immune serum by incubation with 96 h post-electroporation. 40 ml of 1 : 1 protein A-Sepharose : phosphate-buered saline for 2 h at 48C. Proteins were resolved by 7.5% SDS ± PAGE and transferred to polyvinylidene di¯uoride (PVDF) mem- Preparation of ALK-speci®c antiserum brane ®lters (Immobilon-P, Millipore, Bedford, MA) using a An immunogen for ALK antibody production was semidry blotting system (SemiPhor Transfer Unit, Hoefer prepared by ligating a 303 bp AccI/PstI ALK cDNA Scienti®c Instruments, San Francisco, CA). PVDF mem- fragment that encodes residues 1359 ± 1460 of the protein branes were incubated for 1 h at room temperature in (corresponding to residues 419 ± 520 of the chimeric NPM- blocking reagent (PBS containing 0.1% Tween-20 and 3% ALK protein) in-frame into the pQE bacterial expression nonfat dry milk), followed by three 15 min room temperature vector (QIAexpress Expression System, Qiagen, Chats- washes with PBS containing 0.1% Tween-20. The membranes worth, CA) to produce an ALK polypeptide containing an were incubated in streptavidin-biotin-horseradish peroxidase amino-terminal tag of six consecutive histidine residues for complex diluted 1 : 100 in PBS containing 0.1% Tween-20 for binding to nickel-nitriloacetic acid agarose. This protein 1 h at room temperature, then washed in PBS and 0.1% was expressed in the M15[pREP4] Ecolistrain and puri®ed Tween-20 ®ve times for 15 min each; the biotinylated proteins by metal chelate anity chromatography as per the were detected by enhanced chemiluminescence (ECL kit, manufacturer's instructions, then used to immunize Amersham). rabbits. The resulting polyclonal ALK antibody (desig- For in vitro kinase assays, cells were lysed in NP-40 lysis nated `anti-ALK #11') was used in all protein studies. buer (1% NP-40, 150 mM NaCl, 25 mM Tris, pH 7.4)

Serum from rabbits bled prior to immunization was used containing protease inhibitors and Na3VO4 in concentrations for preimmune controls. identical to those noted above. Immunocomplexes were washed three times in this buer and twice in kinase buer

(25 mM HEPES, pH 7.4, 12.5 mM MnCl2. 6.5 mM MgCl2 Protein studies prior to resuspension in 10 ml kinase buer containing 20 mCi of [g-32P]ATP (6000 Ci/mM, Dupont NEN, Boston, MA) and Lysates from metabolically labeled cells and from cells 0.1 mM Na3VO4. Reactions were performed at 378C for analysed in surface biotinylation experiments were pre- 30 min, and stopped by boiling in the presence of 10 mlof pared using radioimmunoprecipitation assay (RIPA) buer 26Laemmli's sample buer, prior to SDS ± PAGE on 7.5% (150 mM NaCl,1%NP-40,0.5%DOC,0.1%SDS,50mM polyacrylamide gels. Phosphoamino acid analysis of in vitro Tris, pH 7.2) containing 0.5% aprotinin, 1 mM PMSF, g-32P ATP-labeled ALK was carried out as previously 2mMEDTA, and 1 mM Na3VO4. Cell lysates were clari®ed described (Shih et al., 1982). Brie¯y, labeled ALK protein of insolubles by a 2 min centrifugation at 14 000 RPM in a was eluted from the gel, extensively digested with TPCK- microfuge, then incubated for 2 h at 48C with a 1 : 100 trypsin, and hydrolyzed in 6 N HCl for 1 h at 1008C. dilution of either preimmune serum or anti-ALK rabbit Following lyophilization, phosphoamino acids were resolved serum. Immunoprecipitates were collected after a 1 h by thin-layer electrophoresis in two dimensions (pH 1.9 incubation at 48C with 40 ml of a 1 : 1 (vol:vol) slurry of followed by pH 3.5). protein A-Sepharose (CL-4B; Pharmacia, Piscataway, NJ) in phosphate-buered saline, and were extensively washed with RIPA buer before suspension in Laemmli's sample In situ hybridization buer, with or without 5% 2-mercaptoethanol. The protein samples were then analysed by sodium dodecyl sulphate- An Alk in situ hybridization probe was prepared by polyacrylamide gel electrophoresis (SDS ± PAGE) on 6% polymerase chain reaction ampli®cation of the 187-bp or 7.5% polyacrylamide gels. Cells were metabolically exon that encodes the juxtamembrane segment of the labeled with [35S]-methionine (100 mCi/ml; 41000 Ci/ receptor with a genomic mouse Alk clone that had been mmol; Amersham, Arlington Heights, IL) for 2 h at 378C previously used for chromosomal localization of the gene in methionine-free DMEM containing 5% dialyzed fetal in the mouse (Mathew et al., 1995) as a template and the calf serum (Gibco BRL, Gaithersburg, MD), following a primers JMAlk A 30 min preincubation in the same medium without added GC radioactive methionine. N-linked glycosylation was inhib- (5'-CACCTATCTGTCTAGAGTGTACCGTCGGAAG-3') ited by treating the cells with tunicamycin (Sigma, St. and JMAlk B Louis, MO; 25 mg/ml of medium) during both the AG preincubation and labeling periods. For pulse-chase (5'-CAGTAGCAGGTACCCTCACCGGATGAGTGTGATG-3') experiments, cells were preincubated in methionine-free DMEM containing 5% dialyzed FCS for 45 min at 378C, which are homologous or reverse complementary, respec- labeled for 15 min with 200 mCi [35S]-methionine per ml of tively, to exon and intron sequences spanning the 5' or 3' media, and chased for the indicated intervals in DMEM boundaries of the exon. This probe produces a single-copy containing 10% fetal calf serum and 30 mg liter cold L- gene pattern in Southern hybridizations of mouse genomic methionine. DNA performed at low stringency (SW Morris, unpub- Biotinylation of cell surface proteins was performed as lished data) and, thus, is unlikely to cross-hybridize under previously described (Meier et al., 1992), with only minor the in situ hybridization conditions used with the mRNAs ALK, a novel, neural receptor tyrosine kinase SW Morris et al 2187 of other potentially related genes. Restriction enzyme visualized after developing the emulsion-coated slides with recognition sites for XbaI(JMAlk A) or KpnI(JMAlk B) Kodak D19 developer at 168C. All sections were counter- were engineered into the primers as shown above (altered stained in hematoxylin and eosin and photographed under nucleotides are italicized, with the correct sequence dark- and bright-®eld illumination. indicated above each) to facilitate subcloning into pBlue- script SK+ (Stratagene). Digestion with XbaIand Nucleotide sequence accession number transcription with T7 RNA polymerase were used to produce the antisense probe, while KpnI digestion and T3 The nucleotide sequence of human ALK cDNA clone RNA polymerase generated the negative control sense RMS17-2 has been entered in GenBank under accession probe. number U62540. Embryos were harvested at various stages of gestation, ®xed in 4% paraformaldehyde overnight at 48C, soaked in 30% sucrose, and embedded in M1 freezing media and frozen in liquid nitrogen. 35S-labeled UTP sense and antisense Acknowledgements riboprobes to detect Alk mRNA transcripts were labeled We thank Kris Dittmer, Karen Rakestraw, Kathy Saalfeld, and hybridized as previously described (Witte et al., 1993) on Terry Smith and Jack Sublett for excellent technical cryostat sections overnight at 458C with 16106 counts per assistance, John Gilbert for editorial review, Jim Downing min of radiolabeled probe. Following hybridization, sections and Linda Hendershot for helpful discussions and Nancy were treated with 50 mg/ml of RNAase A (Sigma) and 100 Dean and Peggy Vandiveer for secretarial support. The units/ml of RNAase T1 (Boehringer-Mannheim). The neuroanatomic review of the Alk in situ hybridizations by RNAase-treated sections were then washed in 50% Eldon Geisert, Karl-Heinz Herzog, Jim Morgan, Rick formamide, 26SSC and 20 mM DTT at 558C for 60 min. Smeyne and Holly Soares is also gratefully acknowledged. Following the formamide wash, sections were washed in a This research was supported in part by grants CA-01702 descending series of SSC solutions to a ®nal stringency of and CA-69129 (SWM) and Cancer Center Support (CORE) 0.16SSC at 508C. After dehydration in alcohol, slides were grant CA-21765 from the National Cancer Institute and by dipped in NTB 2 emulsion (Kodak) and exposed for periods the American Lebanese Syrian Associated Charities (AL- ranging from 2 ± 4 weeks at 48C. Hybridization signal was SAC), St Jude Children's Research Hospital.

References

Barbacid M. (1993). Molecular Genetics of Nervous System Greco A, Pierotti MA, Bongarzone I, Pagliardini S, Lanzi C Tumors. Levin AG and Schmidek HH. (eds). Wiley-Liss: and Della Porta G. (1992). Oncogene, 7, 237 ± 242. New York. pp. 123 ± 136. Gribskov M and Burgess RR. (1986). Nucl. Acids Res., 14, Ben-Neriah Y and Bauskin AR. (1988). Nature, 333, 672 ± 6745 ± 6763. 676. Hanks SK, Quinn AM and Hunter T. (1988). Science, 241, Bernards A and de la Monte SM. (1990). EMBO J., 9, 2279 ± 42 ± 52. 2287. Hohl D, Mehrel T, Lichti U, Turner ML, Roop DR and Bischof D, Pulford K, Mason DY and Morris SW. (1997). Steinert PM. (1991). J. Biol. Chem., 266, 6626 ± 6636. Mol. Cell. Biol., in press. Hu JL, Jelinek MA, Borgman CA, Lansing TJ and Parsons Bladt F, Riethmacher D, Isenmann S, Aguzzi A and JT. (1993). Proc. Natl. Acad. Sci. USA, 90, 6140 ± 6144. Birchmeier C. (1995). Nature, 376, 768 ± 771. Johnson DE, Lee PL, Lu J and Williams LT. (1990). Mol. Blobel G. (1980). Proc. Natl. Acad. Sci. USA, 77, 1496 ± Cell Biol., 10, 4728 ± 4736. 1500. Klein R, Martin-Zanca D, Barbacid M and Parada LF. Bongarzone I, Monzini N, Borrello MG, Carcano C, (1990). Development, 109, 845 ± 850. Ferraresi G, Arighi E, Mondellini P, Della Porta G and Kornfeld R and Kornfeld S. (1985). Annu. Rev. Biochem., 54, Pierotti MA. (1993). Mol. Cell. Biol., 13, 358 ± 366. 631 ± 664. Borer RA, Lehner CF, Eppenberger HM and Nigg EA. Kozak M. (1987). Nucleic Acids Res., 15, 8125 ± 8148. (1989). Cell, 56, 379 ± 390. Kozutsumi H, Toyoshima H, Hagiwara K, Yazaki Y and Bottaro DP, Rubin JS, Faletto DL, Chan AM, Kmiecik TE, Hirai H. (1994). Oncogene, 9, 2991 ± 2998. Vande Woude GF and Aaronson SA. (1991). Science, 251, Krolewski JJ and Dalla-Favera R. (1991). EMBO J., 10, 802 ± 804. 2911 ± 2919. Burd CG and Dreyfuss G. (1994). Science, 265, 615 ± 621. Lamballe F, Smeyne RJ and Barbacid M. (1994). J. Chan WY, Liu QR, Borjigin J, Busch H, Rennert OM, Tease Neurosci., 14, 14 ± 28. LA and Chan PK. (1989). Biochem., 28, 1033 ± 1039. Martin-Zanca D, Hughes SH and Barbacid M. (1986). Chen WJ, Goldstein JL and Brown MS. (1990). J. Biol. Nature, 319, 743 ± 748. Chem., 265, 3116 ± 3123. Martin-Zanca D, Barbacid M and Parada LF. (1990). Genes Condit CM and Meagher RB. (1986). Nature, 323, 178 ± 181. Dev., 4, 683 ± 694. Dumbar TS, Gentry GA and Olson MOJ. (1993). Biochem., Maru Y, Hirai H and Takaku F. (1990). Oncogene Res., 5, 28, 9495 ± 9501. 199 ± 204. Escandon E, Soppet D, Rosenthal A, Mendoza-Ramirez JL, Mathew P, Morris SW, Kane JR, Shurtle SA, Pasquini M, Szonyi E, Burton LE, Henderson CE, Parada LF and Jenkins NA, Gilbert DJ and Copeland NG. (1995). Nikolics K. (1994). J. Neurosci., 14, 2054 ± 2068. Cytogenet Cell Genet., 70, 143 ± 144. Fantl WJ, Johnson DE and Williams LT. (1993). Annu. Rev. Mayer BJ and Baltimore D. (1993). Trends in Cell Biol., 3, Biochem., 62, 453 ± 481. 8 ± 13. Fujimoto J, Shiota M, Iwahara T, Naohiko S, Satoh H, Mori MeierT,ArniS,MalarkannanS,PoinceletMandHoessliD. S and Yamamoto T. (1996). Proc. Natl. Acad. Sci. USA, (1992). Anal. Biochem., 204, 220 ± 226. 93, 4181 ± 4186. Miki T, Bottaro DP, Fleming TP, Smith CL, Burgess WH, Gaudino G, Follenzi A, Naldini L, Collesi C, Santoro M, Chan AMl and Aaronson SA. (1992). Proc. Natl. Acad. Gallo KA, Godowski PJ and Comoglio PM. (1994). Sci. USA, 89, 246 ± 250. EMBO J., 13, 3524 ± 3532. ALK, a novel, neural receptor tyrosine kinase SW Morris et al 2188 Morris SW, Kirstein MN, Valentine MB, Dittmer KG, Songyang Z, Shoelson SE, McGlade J, Olivier P, Pawson T, Shapiro DN, Saltman DL and Look AT. (1994). Science, Bustelo XR, Barbacid M, Sabe H, Hanafusa H, Yi T, Ren 263, 1281 ± 1284. R, Baltimore D, Ratnofsky S, Feldman RA and Cantley Muenke M, Schell U, Hehr A, Robin NH, Losken HW, LC. (1994). Mol. Cell Biol., 14, 2777 ± 2785. Schinzel A, Pulleyn LJ, Rutland P, Reardon W, Malcolm Steinert PM, Parry DA, Idler WW, Johnson LD, Steven AC S and Winter RM. (1994). Natl. Genet., 8, 269 ± 274. and Roop DR. (1985). J. Biol. Chem., 260, 7142 ± 7149. Naldini L, Weidner KM, Vigna E, Gaudino G, Bardelli A, Toyoshima H, Kozutsumi H, Maru Y, Hagiwara K, Furuya Ponzetto C, Narsimhan RP, Hartman G, Zarnegar R, A, Mioh H, Hanai N, Takaku F, Yazaki Y and Hirai H. Michalopoulos GK, Birchmeier W and Comoglio PM. (1993). Proc. Natl. Acad. Sci. USA, 90, 5404 ± 5408. (1991). EMBO J., 10, 2867 ± 2878. Ullrich A and Schlessinger J. (1990). Cell, 61, 203 ± 212. Pawson T and Gish GD. (1992). Cell, 71, 359 ± 362. van der Geer P and Pawson T. (1995). Trends in Biologic Petch LA, Harris J, Raymond VW, Blasband A, Lee DC and Sciences, 20, 277 ± 280. Earp HS. (1990). Mol. Cell Biol., 10, 2973 ± 2982. Van Heyningen V. (1994). Nature, 367, 319 ± 320. Prat M, Crepaldi T, Gandino L, Giordano S, Longati P and von Heijne G. (1986). Nucleic Acids Res. 14, 4683 ± 4690. Comoglio P. (1991). Mol. Cell Biol., 11, 5954 ± 5962. Wang L, Myers M.G., Sun X, Aaronson SA, White M and Pulford K, Lamant L, Morris SW, Butler LH, Wood KM, Pierce JH. (1993). Science, 261, 1591 ± 1592. Stroud D, Delsol G and Mason DY. (1997). Blood, 89, Werner S, Duan DR, De Vries C, Peters KG, Johnson DE 1394 ± 1404. and Williams LT. (1992). Mol. Cell. Biol., 12, 82 ± 88. Rousseau F, Bonaventure J, Legeai-Mallet L, Pelet A, Rozet Wilkie AO, Slaney SF, Oldridge M, Poole MD, Ashworth JM, Maroteaux P, Le Merrer M and Munnich A. (1994). GJ, Hockley AD, Hayward RD, David DJ, Pulleyn LJ, Nature, 371, 252 ± 254. Rutland P, Malcolm S, Winter RM and Reardon W. Schlessinger J. (1994). Current Opinion in Genetics & (1995). Nat. Genet., 9, 165 ± 172. Development, 4, 25 ± 30. Witte DP, Aronow BJ, Harmony JAK, Stauderman ML, Schlessinger J and Ullrich A. (1992). Neuron., 9, 383 ± 391. Stewart WD, Clay MA and Gruppo RA. (1993). Amer. J. Shapiro DN, Sublett JE, Li B, Downing JR and Naeve CW. Pathol., 143, 763 ± 773. (1993). Cancer Res., 53, 5108 ± 5112. Zhou S, Shoelson SE, Chaudhuri M, Gish G, Pawson T, Shih TY, Stokes PE, Smythers GW, Dhar R and Oroszlan S. Haser WG, King F, Roberts T, Ratnofsky S, Lechleider (1982). J. Biol. Chem., 257, 11767 ± 11773. RJ and et al. (1993). Cell, 72, 767 ± 778. Snider WD. (1994). Cell, 77, 627 ± 638. Zhou XM, Idler WW, Steven AC, Roop DR and Steinert Snijders AJ, Haase VH and Bernards A. (1993). Oncogene, 8, PM. (1988). J. Biol. Chem., 263, 15584 ± 15589. 27 ± 35. Soman NR, Correa P, Ruiz BA and Wogan GN. (1991). Proc. Natl. Acad. Sci. USA, 88, 4892 ± 4896.