Clustering of Activating Mutations in c-KIT’s JuxtamembraneCoding Region in Canine Mast Cell Neoplasms

Yongsheng Ma, B. Jack Longley, Xiaomei Wang, John L. Blount,* Keith Langley,† and George H. Caughey*Departments of Dermatology and Pathology, College of Physicians and Surgeons, Columbia University, New York, New York, U.S.A.; *Departmentof Medicine and Cardiovascular Research Institute, University of California at San Francisco, San Francisco, California, U.S.A.; †Amgen, ThousandOaks, California, U.S.A.

The proto-oncogene c-KIT encodes a growth factorreceptor, KIT, with ligand-dependent tyrosine kinaseactivity that is expressed by several cell types includingmast cells. c-KIT juxtamembrane coding regionmutations causing constitutive activation of KIT arecapable of transforming cell lines and have been identi-fied in a human mast cell line and in situ in humangastrointestinal stromal tumors, but have not beendemonstrated in situ in neoplastic mast cells from anyspecies. To determine whether c-KIT juxtamembranemutations occur in the development of mast cellneoplasms, we examined canine mastocytomas, whichare among the most common tumors of dogs andwhich often behave in a malignant fashion, unlikehuman solitary mastocytomas. Sequencing of c-KITcDNA generated from tumor tissues removed fromseven dogs revealed that three of the tumors contained

The c-KIT proto-oncogene encodes the KIT cell-surface receptor, which is comprised of an extracellulardomain containing five immunoglobulin-like loops,a transmembrane domain, and a cytoplasmic domainconsisting of a juxtamembrane region, a kinase domain

divided into adenosine triphosphate (ATP)-binding and phospho-transferase subdomains by a kinase insert, and a carboxy-terminaltail (Yarden et al, 1987; Qiu et al, 1988). KIT shares sequence anddomain organization similarity with the receptors for colony-stimulating factor-1 and platelet-derived growth factor, the threeforming the type III receptor tyrosine kinase subfamily (Yardenet al, 1987; Qiu et al, 1988; Ullrich and Schlessinger, 1990). Thecognate ligand for KIT, variously named stem cell factor (SCF),mast cell growth factor, steel factor, and KIT ligand, is synthesizedas a transmembrane protein that can be cleaved to yield a soluble,bioactive form (Martin et al, 1990; Longley et al, 1997). An isoformof SCF resulting from alternative splicing of exon 6 mRNA lacksa major protease sensitive site and tends to remain membrane bound.

Manuscript received May 4, 1998; revised August 25, 1998; acceptedfor publication October 19, 1998.

Reprint requests to: Jack Longley, MD, Columbia University Collegeof Physicians and Surgeons, Department of Dermatology, Section ofDermatopathology, 630 West 168th Street, VC 5-578, New York, NY10032.

Abbreviations: KIT, human and nonhuman kit protein; pTyr, phospho-tyrosine; SCF, stem cell factor; WT, wild-type.

0022-202X/99/$10.50 · Copyright © 1999 by The Society for Investigative Dermatology, Inc.

165

a total of four mutations in an intracellular juxta-membrane coding region that is completely conservedamong vertebrates. In addition, two mutations werefound in three mast cell lines derived from two addi-tional dogs. One mutation from one line matched thatfound in situ in one of the tumors. The second wasfound in two lines derived from one dog at differenttimes, indicating that the mutation was present insitu in the animal. All five mutations cause highspontaneous tyrosine phosphorylation of KIT. Ourstudy provides in situ evidence that activating c-KITjuxtamembrane mutations are present in, and maytherefore contribute to, the pathogenesis of mast cellneoplasia. Our data also suggest an inhibitory role forthe KIT juxtamembrane region in controlling thereceptor kinase activity. Keywords: c-KIT/in situ/masto-cytosis/receptor tyrosine kinase/stem cell factor. J InvestDermatol 112:165–170, 1999

As with activation of other receptor tyrosine kinases (Ullrichand Schlessinger, 1990), SCF binding induces KIT dimerizationand activates its intrinsic kinase activity, resulting in receptorautophosphorylation (Blume-Jensen et al, 1991; Lev et al, 1992).The activated KIT then binds and phosphorylates a class ofintracellular substrate proteins, initiating a signaling cascade thatculminates in a wide array of biologic activities, includingproliferation, migration, maturation, and survival of hematopoieticstem cell, mast cell, melanocyte, and germ cell lineages (Brizzi et al,1994; Galli et al, 1994; Serve et al, 1994, 1995; Linnekin et al,1997; Price et al, 1997). KIT can be constitutively activated byautocrine SCF stimulation (Kondoh et al, 1995) or in a SCF-independent manner involving mutations in the receptor (Furitsuet al, 1993; Tsujimura et al, 1994, 1996; Hirota et al, 1998).Expression of the activating KIT mutants in murine interleukin-3-dependent cell lines leads to factor-independent growth andaggressive in vivo behavior of the cells (Kitayama et al, 1995;Tsujimura et al, 1996; Hirota et al, 1998), suggesting that constitu-tively activated KIT is oncogenic.

Human mastocytosis, or mast cell disease, is a heterogeneous setof conditions characterized by increased numbers of mast cells invarious organs (Longley et al, 1995). Mast cell neoplasms occur inboth humans and animals. In dogs, mast cell neoplasms are calledmastocytomas, and the disease is common, representing 7%–21%of canine tumors (Priester, 1973; Cohen et al, 1974). A distinctionmust be drawn between human mastocytosis, which is usuallytransient or indolent (Longley et al, 1995), and canine mast cell

166 MA ET AL THE JOURNAL OF INVESTIGATIVE DERMATOLOGY

neoplasia, which behaves unpredictably and is often aggressive andmetastatic (Macy and MacEwen, 1989). For instance, humansolitary mastocytomas essentially never metastasize; in contrast,µ50% of canine mastocytomas behave in a malignant fashion, asestimated by Hottendorf and Nielsen (1969) after review of 46published reports of tumors in 938 dogs. KIT’s involvement in thepathogenesis of mastocytosis is suggested by the observation thatseveral mutations resulting in constitutive activation of KIT havebeen detected in a number of mast cell lines. For instance, a pointmutation in human c-KIT (causing substitution of Val for Asp816

in the phosphotransferase domain and receptor autoactivation)occurs in HMC-1, a long-term human mast cell leukemia line(Furitsu et al, 1993) and in the corresponding codon in two rodentmast cell lines (Tsujimura et al, 1994, 1995). Moreover, thisactivating mutation has been identified in situ in some cases ofhuman mastocytosis (Nagata et al, 1995; Longley et al, 1996). Twoother activating mutations have been found in another region ofKIT, the intracellular juxtamembrane region. One results inVal560Gly substitution in the human HMC-1 mast cell line (Furitsuet al, 1993), and the other eliminates seven amino acids (Thr573–His579) in a rodent mast cell line called FMA3 (Tsujimura et al, 1996).The significance of these two juxtamembrane region mutations isunclear, as they have not been identified in situ.

To determine whether c-KIT juxtamembrane mutations arepresent in mast cell neoplasms in situ and may therefore contributeto their pathogenesis, we examined c-KIT cDNA from dogmastocytoma tissues and cell lines. We report herein the completesequence of canine c-KIT, and the development of a model of KITstructure. We also identify five novel juxtamembrane mutationscausing constitutive activation of KIT, which suggest a key regioninvolved in negative control of the receptor kinase activity, andshow that activating c-KIT juxtamembrane mutations are commonin canine mast cell neoplasms. This provides the first in situ evidencethat such mutations may play a part in the pathogenesis of mastcell neoplasia.

MATERIALS AND METHODS

Materials Recombinant human (rh) and canine (rc) SCF, murinemonoclonal and rabbit polyclonal anti-human KIT antibody, and wild-type (WT) human full-length c-KIT cDNA/pSPORT1 plasmid wereprovided by Amgen (Thousand Oaks, CA). Mouse anti-phosphotyrosine(pTyr) monoclonal antibody was purchased from Upstate Biotechnology(Lake Placid, NY). Human SCF cDNA/pBluescript plasmid was providedby Dr. Douglas Williams (Immunex, Seattle, WA).

Canine mastocytoma tissues and cells Six mastocytomas freshlyexcised from six dogs were snap-frozen in OCT medium and cryostat-sectioned. An additional mastocytoma removed from a seventh dog wasgrown as a solid tumor in the dermis of a BALB/C athymic mouse priorto being frozen. Three dog mastocytoma cell lines (BR, C1, and C2) wereestablished in long-term culture after initial serial passaging in the skin ofBALB/C athymic mice from skin mastocytomas of two additional dogs(Lazarus et al, 1986; DeVinney and Gold, 1990). C1 and C2 originatedfrom tumors harvested at separate times from one dog. These cells werecultured in Dulbecco’s modified Eagle’s medium supplemented with 2%bovine calf serum and 0.25 mg per ml histidine. As control KIT-positivecells, normal human melanocytes provided by Yale Skin Disease ResearchCenter (New Haven, CT) were grown in F10 medium with 10% bovinecalf serum. COS-7 cells were cultured in Dulbecco’s modified Eagle’smedium with 10% bovine calf serum. Follow-up is only available on thedog from which the C1 and C2 cell lines were derived which died of itstumor. The sources of the BR cell line and the individual tumors were alllost to follow-up.

Screening for mutations After RNA extraction, cDNA was synthe-sized by reverse transcriptase using random hexamers as previously described(Longley et al, 1991, 1996). c-KIT cDNA were amplified in the presenceof 1 µM of primers (see below) with Taq DNA polymerase for 30 cyclesof 1 min at 95°C, 1 min at 55°C, and 1 min at 72°C. Dideoxysequencingwas performed directly from gel-purified polymerase chain reaction ampli-mers, or individual cDNA subcloned into plasmids using the TA Cloning Kit(Invitrogen, Carlsbad, CA) by the W.M. Keck Foundation BiotechnologyResource Laboratories at Yale University, the Biomolecular Resource

Center at the University of California at San Francisco, and the SkinDisease Research Center of the Department of Dermatology of ColumbiaUniversity. Sequencing of mastocytoma c-KIT focused on the juxtamem-brane and phosphotransferase coding regions using primer pairs: 59-CAAATCCATCCCCACACCCTGTTCAC-39 (1564–1589) and 59-CCATAAGCAGTTGCCTCAAC-39 (1850–1831); 59-TGTATTCA-CAGAGACTTGGC-39 (2383–2402) and 59-GGCTGAGCATCCGG-AAGCCT-39 (2695–2676). The numbers in parentheses refer to humanc-KIT cDNA (Yarden et al, 1987). The entire KIT coding region wassequenced from cDNA prepared from normal dog lung mRNA.

cDNA construction and transfection Mutations identified in dogmastocytomas (Trp556Arg, Leu575Pro, ∆Trp556–Lys557, and ∆Val558) werereproduced in human c-KIT cDNA in the pcDNA3 mammalian expressionvector (Invitrogen) using the Quikchange Site-Directed Mutagenesis Kit(Stratagene, La Jolla, CA). Human c-KIT was used as the juxtamembraneamino acids are completely conserved in mammals (see Results section).COS-7 cells (90% confluent in 10-cm plate) were transfected with 5 µgplasmid using 15 µl LipofectAMINE (Life Technologies, Gaithersburg,MD) in serum-free medium for 5 h. An equal volume of medium with20% bovine calf serum was then added and cells incubated overnight,followed by 8–24 h culture in regular medium prior to protein expressionand tyrosine phosphorylation assay.

Immunoprecipitation and immunoblotting For SCF stimulationexperiments, cells were serum-starved for 18 h before incubation withSCF. Cells were harvested in lysis buffer containing 1% Triton X-100, 50 mM HEPES (pH 7.5), 150 mM NaCl, 10% glycerol, 1 mMphenylmethylsulfonyl fluoride, 10 µg leupeptin per ml, 10 µg aprotininper ml , and 1 mM Na orthovanadate. Centrifugation-clarified cell lysateswere immunoblotted as total cell lysates (see below) or immunoprecipitatedfor 1.5 h at 4°C with mouse anti-human KIT antibody and protein Aagarose. Immunoprecipitates were washed with lysis buffer and heat-elutedin sodium dodecyl sulfate sample buffer. Samples were fractionated by7.5% sodium dodecyl sulfate-polyacrylamide gel electrophoresis, transferredon to polyvinylidene difluoride membrane, blocked by 5% bovine serumalbumin in Tris-buffered saline (50 mM Tris–HCl, pH 7.5, and 150 mMNaCl) plus 0.1% Tween-20 (TBST), and probed with rabbit anti-humanKIT or mouse anti-pTyr antibody for 1 h, followed by washing withTBST. Membranes were incubated in TBST with horseradish peroxidase-linked secondary antibody for 45 min, washed, and antigen–antibodycomplexes detected using the ECL System (Amersham, Arlington Heights,IL). Anti-pTyr blots were stripped in 100 mM β-mercaptoethanol, 2%sodium dodecyl sulfate, and 62.5 mM Tris–HCl (pH 6.7) at 50°C for30 min, then reprobed with rabbit anti-human KIT antibody.

Detection of SCF mRNA SCF mRNA was detected by reversetranscriptase–polymerase chain reaction, followed by Southern blotting witha SCF-specific oligonucleotide probe and by direct amplimer sequencing.Canine SCF primers correspond to nucleotides 191–209 and 607–588(Shull et al, 1992), and the probe to human bases 301–367 (Martinet al, 1990).

Molecular model generation A low-resolution homology model ofthe intracellular canine KIT tyrosine kinase domains was constructed withassistance from an automated protein modeling tool (ProMod) and server(Swiss-Model) (Peitsch, 1996). The X-ray diffraction-derived coordinatesof the fibroblast growth factor (FGF) receptor tyrosine kinase domain(Mohammadi et al, 1996) (1FGK in the Brookhaven Protein Data Bank)served as a template for the model. Among available experimentally definedtyrosine kinase structures, 1FGK is most homologous to KIT, as determinedby alignments with BLAST and FastA algorithms (Altschul et al, 1990).First-approach alignments were checked in a process that included acomparison of sequences conserved in other tyrosine kinases. The modelwas then optimized by Swiss-Model, which idealizes bond geometry andremoves unfavorable nonbonded contacts by energy minimization withCHARMM. The resulting coordinate files were displayed using theRasMol molecular visualization program.

RESULTS

Canine c-KIT cDNA and protein sequence Overall, thededuced amino acid sequence of normal dog KIT showsidentity of 93%, 88%, 81%, and 64% compared with cat, human,mouse, and chicken KIT, respectively. Conservation is less in theextracellular domain than in the intracellular domains. For example,dog and human KIT are 78% identical in the 519/520 extracellular

VOL. 112, NO. 2 FEBRUARY 1999 ACTIVATING MUTATIONS IN DOG MASTOCYTOMASTITLE 167

Figure 1. Nucleotide and deduced amino acid sequence of canine KIT. Predicted extracellular domain cysteines are shaded (the first Cys lies inthe signal peptide). Asn residues at consensus glycosylation sites are marked with asterisks (* ). The predicted transmembrane domain is shaded. Residuesaffected by the mis-sense mutations and deletions reported in this work lie within black boxes. A peptide repeated in the inserted sequence in C1 andC2 mastocytoma cells is underlined. The site of the 48 bp C1/C2 nucleotide insertion is shown by a caret ( ∧ ). The normal dog KIT cDNA sequencehas been deposited in the GenBank database (accession number AF044249).

Table I. c-KIT juxtamembrane mutations in canine mastocytomas

WT Base change Amino acid change

TissueDS 2 6 bp deletion (1694–1699) ∆Trp556–Lys557

AG 2 3 bp deletion (1700–1702) ∆Val558

SW 2 T→C transition (1694) Trp556ArgT→C transition (1752) Leu575Pro

Cell lineBR 1 T→C transition (1752) Leu575ProC1, C2 1 48 bp insertion (1784–1785) 1TYPTQLPYDHKWEFPR

domain residues but are 97% identical in the 433 intracellulardomain residues. In mammalian KIT (human, dog, cat, cow, goat,rat, and mouse), the 38 residues of the intracellular juxtamembraneregion (amino acids 543–580 in Fig 1) are identical as are all butone (Ile/Val661) of the adjacent 104 residues in the ATP-bindingdomain. We used Trp581–Glu582 as the beginning of the kinasedomain, as defined by the crystal structure of FGF receptor kinase(Mohammadi et al, 1996). Residues 768–914 comprise anotherhigh-homology region (96% identical in alignments of mammalianKIT), corresponding to the phosphotransferase domain. Theselengthy regions, which are intolerant of sequence variation amongdifferent species, appear fundamental for concerted receptor func-tion. The 12 extracellular domain cysteines (plus another in thesignal peptide) are conserved in dog and human KIT, as are nineof 11 predicted canine N-linked glycosylation sites.

Mutations of c-KIT in dog mastocytomas We sequenced c-KIT cDNA generated from mastocytoma tissues from seven dogs(DS, AG, SW, RT, KR, LL, MN) and three mast cell lines derivedfrom two additional dogs (BR, C1, C2). Three of the tissue samplesshowed four mutations clustered in the intracellular juxtamembranecoding region (Fig 1 and Table I). Sample DS had a 6 bp deletion

eliminating Trp556–Lys557. Sample AG contained a 3 bp deletioneliminating Val558. We found no WT sequence in either sample,suggesting functional loss of one (presumably WT) allele. SampleSW was heterozygous for point mutations in separate alleles: onesubstituting Arg for Trp556 and the other substituting Pro forLeu575. The Leu575Pro mutation also was present in cells of theBR line, which was derived from a different dog. The C1 and C2cell lines, derived, respectively, from early- and late-stage tumorsof one dog, both contained a 48 bp insertion between nucleotides1784 and 1785, which are, respectively, the first and second baseof codon 586 in the proximal ATP-binding domain (Fig 1). Exceptfor the first 3 bp, the inserted nucleotide sequence is a direct repeatof the segment immediately preceding the site of insertion. Deducedamino acids 3–16 of the inserted sequence (Table I) are a repeatof residues 572–585 of normal dog KIT (Fig 1). All the three celllines (BR, C1, and C2) were heterozygous for their respectivemutations, with the other allele being WT in the region of theidentified mutations. The mutations in the mastocytoma tissuesclearly developed in situ and could contribute to the pathogenesisof mast cell neoplasms in the animals in which they developed.Likewise, the detection of the same 48 bp insertion in both C1and C2 cell lines, which were derived separately from the same

168 MA ET AL THE JOURNAL OF INVESTIGATIVE DERMATOLOGY

Figure 2. Predicted structure of canineKIT. (a) Extracellular immunoglobulin-likeloops I–V are depicted as partially closed circles;the location of cysteines (S-S) and potentialglycosylation sites (* ) are marked. Intra-cellularly, the juxtamembrane, ATP-binding,kinase insert, phosphotransferase, and C-ter-minal tail regions are indicated. (b) A modelof canine KIT intracellular tyrosine kinase(and a portion of the juxtamembrane region)prepared by homology modeling using theknown structure of the FGF receptor tyrosinekinase (Mohammadi et al, 1996). In the ATP-binding and phosphotransferase lobes of thetyrosine kinase, β (lightly shaded ribbons) andα (darkly shaded ribbons) secondary structurepredominate, respectively. The site of the 16-amino acid insertion in C1 and C2 cells isindicated, as is the location of the Leu575Prosubstitution detected in SW mastocytoma andBR cells. The locations of critical nucleotidebinding and catalytic loops and of the ‘‘activa-tion loop,’’ which contains tyrosine residuesthought to be phosphorylated and is a site of apreviously recognized KIT activating mutation(Furitsu et al, 1993), are indicated.

animal at different times, indicates that the mutation was presentin the original tumors. Therefore, we link all five mutations todog mastocytomas in situ.

Limited sequencing of the region encompassing codons 514–609 in the other four dog mastocytomas showed only WTsequences. As an activating point mutation has previously beenidentified in codon 816 in the phosphotransferase domain insome human patients with mastocytosis (Nagata et al, 1995; Longleyet al, 1996), we also sequenced cDNA corresponding to codons787–890 but no mutations were found.

The structure of canine KIT and location of identified mutationsare depicted schematically in Fig 2. The N-terminal portion ofthe juxtamembrane region, the kinase insert, and the C-terminaltail are omitted from the ribbon model due to the lack of FGFreceptor-homologous sequence in these regions.

Juxtamembrane mutations activate KIT To determine thefunctional significance of the four novel mis-sense and deletionmutations, we made equivalent mutations in human c-KIT cDNAand expressed the constructs in COS-7 cells. COS cells havepreviously been used for KIT function assays because they expressneither endogenous KIT nor SCF (Reith et al, 1991; Serve et al,1994; Price et al, 1997). We used a human c-KIT cDNA as wehave had extensive experience with one which was in hand, andas the juxtamembrane region amino acids we were studying are100% conserved in all mammals studied. Figure 3(b) shows theexpression of WT and mutant KIT. KITWT was expressed asproteins of 145 kDa and 125 kDa, which represent cell-surface(fully glycosylated) and intracellular (partially glycosylated) forms,respectively (Majumder et al, 1988; Nocka et al, 1989, 1990). Thefour mutant KIT were also expressed as two protein productscorresponding to KITWT, except that the electrophoretic mobilityof ∆Trp556–Lys557 mutant was retarded by µ5 kDa. In contrast toKITWT, the larger isoforms of mutant KIT were apparently less inthe steady state. The altered expression pattern of these mutants issimilar to the expression of KIT containing an Asp816Val activatingmutation (Furitsu et al, 1993), which has an accelerated turnoverof its cell-surface form (Moriyama et al, 1996).

In the absence of rhSCF, the 145-kDa KITWT showed a lowbasal level of spontaneous tyrosine phosphorylation (Fig 3a), whichhas also been observed by previous studies (Reith et al, 1991; Serveet al, 1994). The four mutant KIT, however, exhibited high levelsof spontaneous tyrosine phosphorylation of both size isoforms;phosphorylation of the larger forms was at least 10-fold higher thanthat of KITWT (density quantitated after normalizing the differencein protein expression shown in Fig 3b). rhSCF stimulation at100 ng per ml for 10 min led to an increase in phosphorylation ofKITWT and a decrease in its protein amount. The latter is consistentwith SCF-induced KIT downregulation (Yee et al, 1993, 1994).In contrast, the rhSCF stimulation reduced both the degree ofphosphorylation and the amount of the larger isoforms of mutantKIT, without affecting the smaller isoforms. One explanation forthese results is that the mutant receptor phosphorylation is saturated.Taken together, the results indicate that the four juxtamembranemutations identified in canine mastocytomas are kinase-activatingmutations.

Because of technical difficulties associated with reproducing the48 base insert in an intact cDNA, we assayed KIT tyrosinephosphorylation in C1 and C2 cells themselves to determine theeffect of the fifth mutation. As rcSCF stimulates human KIT(unpublished results), we used normal human melanocytes as KIT-positive control cells. Figure 4(b) shows both C1 and C2 cellsexpressed immunoreactive KIT protein corresponding to humanKIT. In the absence of rcSCF, the C1 and C2 cell KIT showedhigh levels of spontaneous tyrosine phosphorylation, µ5-fold higherthan normal human KIT (Fig 4a). Although the human receptorunderwent remarkable tyrosine phosphorylation in response torcSCF stimulation, the mutated canine KIT showed only slightincreases in phosphorylation, which was already µ80% of themaximum levels obtained by ligand stimulation. Concentrations ofrcSCF up to 500 ng per ml did not significantly increase the C1and C2 KIT phosphorylation over that seen with 100 ng per ml(data not shown). The relative contributions of mutant and WTKIT to the observed phosphorylation are unclear from these data.To rule out possible autocrine SCF production as a source of

VOL. 112, NO. 2 FEBRUARY 1999 ACTIVATING MUTATIONS IN DOG MASTOCYTOMASTITLE 169

Figure 3. KIT is highly activated by juxtamembrane mutations.(a) Anti-pTyr blot of immunoprecipitated human WT and juxtamembrane-mutated KIT transiently expressed in COS-7 cells treated, or not, withrhSCF (100 ng per ml, 10 min) following serum starvation reveals that thefour mutants are spontaneously phosphorylated (lanes 3, 5, 7, 9) at strikinglyhigh levels compared with normal receptor (lane 1). (b) Reprobing theupper blot (after stripping) with anti-KIT antibody shows relative amountof two isoforms of WT (lanes 1–2) and mutant (lanes 3–10) KIT. Molecularweight marker positions (in kDa) are indicated.

receptor activation, we sought mRNA encoding SCF in the dogcells. Reverse transcriptase–polymerase chain reaction using primersspecific for canine SCF yielded no products from C1 and C2 cells,but amplimers were unexpectedly produced from the BR dogmastocytoma cell line (data not shown). These amplimers havesequence identical to that previously published for dog SCF (Shullet al, 1992) (data not shown), and validate the use of thesepolymerase chain reaction primers to rule out SCF production byC1 and C2 cells. Although endogenous SCF could activate KITWT

in BR cells, it is not necessary to activate the mutant receptorbecause, as we have shown in Fig 3, the Leu575Pro mutated KITfound in this cell line was spontaneously activated. The lack ofdetectable SCF mRNA transcripts in C1 and C2 cells indicatesthat autocrine SCF stimulation is not responsible for the constitutiveKIT activation seen in these cells, leaving the insertion mutationas the most likely cause.

DISCUSSION

This study identifies five novel activating c-KIT mutations incanine mastocytomas and provides in situ evidence that c-KIT

Figure 4. KIT in C1 and C2 cells are highly autophosphorylated.(a) Anti-pTyr blot of total cell lysates of human melanocytes (HM) andC1 and C2 cells incubated, or not, with rcSCF (100 ng per ml, 10 min)following serum starvation shows that spontaneous phosphorylation ofcanine KIT (lanes 3, 5) is markedly higher than that of normal human KIT(lane 1) and µ80% of rcSCF-induced phosphorylation of the caninereceptors (lanes 4, 6). (b) Reprobing the upper blot (after stripping) withanti-human KIT antibody substantiates comparable protein amount ofhuman (lanes 1–2) and canine (lanes 3–6) receptors. Marker protein positions(in kDa) are indicated.

juxtamembrane mutations are present in the development of mastcell neoplasms. Because autoactivated KIT mutants can transformcell lines to factor-independent growth in vitro and confer tumori-genicity in vivo (Kitayama et al, 1995; Tsujimura et al, 1996; Hirotaet al, 1998), it appears likely that they contribute to the pathogenesisof mast cell neoplasia. Canine mast cell neoplasms often behaveaggressively and, in the present survey, are highly associated withjuxtamembrane mutations. Mastocytosis in humans, in contrast, isgenerally more indolent and only a single Asp816Val mutation inthe phosphotransferase domain has been demonstrated in situ(Nagata et al, 1995; Longley et al, 1996). Chromosomal abnormali-ties, such as trisomies of chromosomes 8 and 9, have been identifiedin human patients with mastocytosis (Lishner et al, 1996). It maybe that such aberrations also contribute to the pathogenesis ofmastocytosis, especially with regard to whether or not the diseasebehaves in an aggressive fashion. Interestingly, a juxtamembranepoint mutation has been found in the human mast cell line HMC-1(Furitsu et al, 1993), which was derived from an extremely rareand aggressive case of human mast cell leukemia (Butterfield et al,1988). Whether the association of different sorts of mutationswith more or less aggressive mast cell disease represents a directconsequence of the type of mutation is unclear, as is whether ornot activating c-KIT mutations by themselves are sufficient to causemast cell neoplasia. Although they have not been identified inhuman mast cells in situ, c-KIT juxtamembrane activating mutationsdo occur in human tumors, as a recent study has identified fourdeletions and one mis-sense mutation (Val559Asp substitution)within this region in human gastrointestinal stromal tumors (Hirotaet al, 1998).

The substantial increase in spontaneous tyrosine phosphorylationof KIT caused by the juxtamembrane mutations shown in Fig 3suggests that an intrinsic inhibitory constraint of the receptorkinase is relieved by these mutations. Replacement of Trp (anamphipathic residue) at position 556 with an Arg (comparable insize but charged) causes autoactivation, suggesting that the Trp’s

170 MA ET AL THE JOURNAL OF INVESTIGATIVE DERMATOLOGY

hydrophobic side chain is important for the inhibition. Correctaligning and folding of other residues appear crucial as well, becausedeletion of Trp556–Lys557 or Val558, or substitution of Pro forLeu575 have similar effects. Consistent with our results are theautoactivated KIT with mutations in this region in the HMC-1(Furitsu et al, 1993) and FMA3 (Tsujimura et al, 1996) cell linesand in gastrointestinal stromal tumors (Hirota et al, 1998). In lightof the trans-phosphorylation model of activation of receptor tyrosinekinases (Ullrich and Schlessinger, 1990), an intact juxtamembraneregion may keep the accessible surface area for dimer formation ata minimum so that trans-phosphorylation is deterred. To test thishypothesis, we used chemical cross-linking reagents to detectligand-independent dimers of the mutant KIT, but our results wereinconclusive, possibly due to the transient nature of receptordimerization and/or the low efficiency of cross-linking (data notshown). It is also possible that the mutant KIT are cis-phosphoryl-ated, and the juxtamembrane region may stabilize the adjacent kinasedomain in an auto-repressed conformation through intramolecularinteractions. Inhibition of kinase activity by additional domainsthrough intramolecular interactions, although not reported forreceptor tyrosine kinases, has recently been shown in the crystalstructure of Src (Xu et al, 1997) and Hck tyrosine kinases (Sicheriet al, 1997).

In addition to modulating kinase activity, the KIT juxtamembraneregion has recently been shown to interact with two intracellularsignaling molecules: the Src family member Lyn (Linnekin et al,1997) and CsK-homologous kinase (Price et al, 1997). Studies onsubstrate-binding specificities of KIT juxtamembrane mutants mayadd to our understanding of the molecular mechanisms underlyingKIT signaling pathways and mast cell proliferative disorders.

This work is supported in part by NIH grants RO1 AR43356-01A2, AR44535(Columbia Skin Disease Research Core Center), and HL-24136. We wish tothank Drs. Neil Maudlin, Stephen Lazarus, and Warren Gold for making someof the mastocytoma tissues and cell lines available, and W.W. Raymond forassistance in generating the molecular model.

REFERENCESAltschul SF, Gish W, Miller W, Myers EM, Lipman DJ: Basic local alignment search

tool. J Mol Biol 215:403–410, 1990Blume-Jensen P, Claesson-Welsh L, Siegbahn A, et al: Activation of the human c-kit

product by ligand-induced dimerization mediates circular actin reorganizationand chemotaxis. EMBO J 10:4121–4128, 1991

Brizzi MF, Zini MG, Aronica MG, et al: Convergence of signaling by interleukin-3, granulocyte-macrophage colony-stimulating factor, and mast cell growthfactor on JAK2 tyrosine kinase. J Biol Chem 269:31680–31684, 1994

Butterfield JH, Weiler D, Dewald G, Gleich GJ: Establishment of an immature mastcell line from a patient with mast cell leukemia. Leukemia Res 12:345–355, 1988

Cohen D, Reif JS, Brodey RS, Keiser H: Epidemiological analysis of the mostprevalent sites and types of canine neoplasia observed in a veterinary hospital.Cancer Res 34:2859–2868, 1974

DeVinney R, Gold WV: Establishment of two dog mastocytoma cell lines incontinuous culture. Am J Respir Cell Mol Biol 3:413–420, 1990

Furitsu T, Tsujimura T, Tono T, et al: Identification of mutations in the codingsequence of the proto-oncogene c-kit in a human mast cell leukemia cell linecausing ligand-independent activation of c-kit product. J Clin Invest 92:1736–1744, 1993

Galli SJ, Zsebo KM, Geissler EN: The kit ligand, stem cell factor. Adv Immunol55:1–96, 1994

Hirota S, Isozaki K, Moriyama Y, et al: Gain-of-function mutations of c-kit in humangastrointestinal stromal tumors. Science 279:577–580, 1998

Hottendorf GH, Nielsen SW: Canine mastocytoma—a review of clinical aspects. JAm Vet Med Assoc 154:917–924, 1969

Kitayama H, Kanakura Y, Furitsu T, et al: Constitutively activating mutations of c-kit receptor tyrosine kinase confer factor-independent growth andtumorigenicity of factor-dependent hematopoietic cell lines. Blood 85:790–798, 1995

Kondoh G, Hayasaka N, Li Q, Nishimune Y, Hakaura A: An in vivo model forreceptor tyrosine kinase autocrine/paracrine activation: auto-stimulated KITreceptor acts as a tumor promoting factor in papillomavirus-inducedtumorigenesis. Oncogene 10:341–347, 1995

Lazarus SC, DeVinney R, McCabe LJ, Finkbeiner WE, Elias DJ, Gold WM: Isolatedcanine mastocytoma cells: propagation and characterization of two cell lines.Am J Physiol 251:C935–C944, 1986

Lev S, Yarden Y, Givol D: Dimerization and activation of the kit receptor bymonovalent and bivalent binding of the stem cell factor. J Biol Chem 267:15970–15977, 1992

Linnekin D, DeBerry CS, Mou S: Lyn associates with the juxtamemebrane regionof c-Kit and is activated by stem cell factor in hematopoietic cell lines andnormal progenitor cells. J Biol Chem 272:27450–27455, 1997

Lishner M, Confino-Cohen R, Mekori YA, et al: Trisomies 9 and 8 detected byfluorescence in situ hybridization in patients with systemic mastocytosis. JAllergy Clin Immunol 98:199–204, 1996

Longley J, Ding T, Cuono C, et al: Isolation, detection, and amplification of intactmRNA from dermatome strips, epidermal sheets, and sorted epidermal cells.J Invest Dermatol 97:974–979, 1991

Longley J, Duffy TP, Kohn S: The mast cell and mast cell disease. J Am AcadDermatol 32:545–561, 1995

Longley BJ, Tyrrell L, Lu S, et al: Somatic c-KIT activating mutation in urticariapigmentosa and aggressive mastocytosis: establishment of clonality in a humanmast cell neoplasm. Nature Genet 12:312–314, 1996

Longley BJ, Tyrrell L, Ma Y, et al: Chymase cleavage of stem cell factor yields abioactive, soluble product. Proc Natl Acad Sci USA 94:9017–9021, 1997

Macy DW, MacEwen EG. Mast cell tumors. In: Withrow SJ, MacEwen EG (eds).Clinical Veterinary Oncology. Philadelphia: J.B. Lippincott, 1989, pp. 156–166

Majumder S, Brown K, Qui F, Besmer P: c-kit protein, a transmembrane kinase:identification in tissues and characterization. Mol Cell Biol 8:4896–4903,1988

Martin FH, Suggs SV, Langley KE, et al: Primary structure and functional expressionof rat and human stem cell factor DNAs. Cell 63:203–211, 1990

Mohammadi M, Schlessinger J, Hubbard SR: Structure of the FGF receptor tyrosinekinase domain reveals a novel autoinhibitory mechanism. Cell 86:577–587, 1996

Moriyama Y, Tsujimura T, Hashimoto K, et al: Role of aspartic acid 814 in thefunction and expression of c-kit receptor tyrosine kinase. J Biol Chem 271:3347–3350, 1996

Nagata H, Worobec AS, Oh CK, et al: Identification of a point mutation in thecatalytic domain of the protooncogene c-kit in peripheral blood mononuclearcells of patients who have mastocytosis with an associated hematologic disorder.Proc Natl Acad Sci USA 92:10560–10564, 1995

Nocka K, Majumder S, Chabot B, et al: Expression of c-kit gene products in knowncellular targets of W mutations in normal and W mutant mice—evidence foran impaired c-kit kinase in mutant mice. Genes Dev 3:816–826, 1989

Nocka K, Tan JC, Chiu E, et al: Molecular bases of dominant negative and loss offunction mutations at the murine c-kit/white spotting locus: W37, Wv, W41

and W. EMBO J 9:1805–1813, 1990Peitsch MC: ProMod and Swiss-Model: internal-based tools for automated

comparative protein modeling. Biochem Soc Trans 24:274–279, 1996Price DJ, Rivnay B, Fu Y, et al: Direct association of Csk homologous kinase

(CHK) with the diphosphorylated site Tyr568/570 of the activated c-KIT inmegakaryocites. J Biol Chem 272:5915–5920, 1997

Priester WA: Skin tumors in domestic animals: data from 21 US and Canadiancolleges of veterinary medicine. J Natl Cancer Inst 50:457–466, 1973

Qiu F, Ray P, Brown K, et al: Primary structure of c-kit: relationship with the CSF-1/PDGF receptor kinase family-oncogenic activation of v-kit involves deletionof extracellular domain and C terminus. EMBO J 7:1003–1011, 1988

Reith AD, Ellis C, Lyman SD, et al: Signal transduction by normal isoforms and Wmutant variants of the Kit receptor tyrosine kinase. EMBO J 10:2451–2459, 1991

Serve H, Hsu YC, Besmer P: Tyrosine residue 719 of the c-kit receptor is essentialfor binding of the P85 subunit of phosphatidylinositol (PI) 3-kinase and forc-kit-associated PI 3-kinase activity in COS-1 cells. J Biol Chem 269:6026–6030, 1994

Serve H, Yee NS, Stella G, et al: Differential roles of PI3-kinase and Kit tyrosine821 in Kit receptor-mediated proliferation, survival and cell adhesion in mastcells. EMBO J 14:473–483, 1995

Shull RM, Suggs SV, Langley KE, et al: Canine stem cell factor (c-kit ligand) supportsthe survival of hematopoietic progenitors in long-term canine marrow culture.Exp Hematol 20:1118–1124, 1992

Sicheri F, Moarefi I, Kuriyan J: Crystal structure of the Src family tyrosine kinaseHck. Nature 385:602–609, 1997

Tsujimura T, Furitsu T, Morimoto M, et al: Ligand-independent activation of c-kitreceptor tyrosine kinase in a murine mastocytoma cell line P-815 generatedby a point mutation. Blood 83:2619–2626, 1994

Tsujimura T, Furitsu T, Morimoto M, et al: Substitution of an aspartic acid resultsin constitutive activation of c-kit receptor tyrosine kinase in a rat tumor mastcell line RBL-2H3. Int Arch Allergy Immunol 106:377–385, 1995

Tsujimura T, Morimoto M, Hashimoto K, et al: Constitutive activation of c-kit inFMA3 murine mastocytoma cells caused by deletion of seven amino acids atthe juxtamembrane domain. Blood 87:273–283, 1996

Ullrich A, Schlessinger J: Signal transduction by receptors with tyrosine kinaseactivity. Cell 61:203–212, 1990

Xu W, Harrison SC, Eck MJ: Three-dimensional structure of the tyrosine kinasec-Src. Nature 385:595–602, 1997

Yarden Y, Kuang W-J, Yang-Feng T, et al: Human proto-oncogene c-kit: a new cellsurface receptor tyrosine kinase for an unidentified ligand. EMBO J 6:3341–3351, 1987

Yee NS, Langen H, Besmer P: Mechanism of kit ligand, phorbol ester, and calcium-induced down-regulation of c-kit receptors in mast cells. J Biol Chem 268:14189–14201, 1993

Yee NS, Hsiau CM, Serve H, Vosseller K, Besmer P: Mechanism of down-regulationof c-kit receptor. J Biol Chem 269:31991–31998, 1994