A Knock-In Mouse Model of Gastrointestinal Stromal Tumor Harboring Kit K641E

Research Article

Brian P. Rubin,1 Cristina R. Antonescu,4 James P. Scott-Browne,1 Melissa L. Comstock,1 Yansong Gu,2 Munir R. Tanas,1 Carol B. Ware,3 and Jessica Woodell1

Departments of 1Pathology, 2Radiation Oncology and Immunology, and 3Comparative Medicine, University of Washington Medical Center, Seattle, Washington and 4Department of Pathology, Memorial Sloan-Kettering Cancer Center, New York, New York

Abstract some neoplasms driven by activating mutations in these RTKs, also respond to imatinib therapy. A mouse model of gastrointestinal stromal tumor (GIST) has been developed by a knock-in gene targeting strategy, which Kit is a RTK with a split intracellular kinase domain and is introduced a Kit gene K641E mutation, originally identified in related structurally to colony-stimulating factor-1 receptor (also sporadic human GISTs and in the germ line of familial GIST known as FMS receptor), Flt-3 receptor, and PDGFRA and PDGFRB syndrome patients. Homozygous and heterozygous Kit K641E (3). Kit ligand (also known as steel factor, stem cell factor, and mast mice develop gastrointestinal pathology with complete pen- cell growth factor) is the only known ligand for Kit (4). Kit, located etrance and all Kit K641E homozygotes die by age 30 weeks on chromosome 4q in humans, was originally discovered as a viral due to gastrointestinal obstruction by hyperplastic interstitial proto-oncogene (v-Kit) that caused sarcomas in cats, and its role as cells of Cajal (ICC) or GISTs. Heterozygous mice have less a cellular oncogene has been further established in acute extensive ICC hyperplasia and smaller GISTs, suggesting a myelogenous leukemia, mast cell leukemia, mastocytosis, germ dose-response relationship between oncogenically activated cell tumors, and gastrointestinal stromal tumors (GIST; refs. 5–11). Kit and ICC proliferation. Immunoprecipitation and Western Kit mutations within these neoplasms are defined as ‘‘activating’’ blotting reveal GISTs to contain abundant phosphorylated/ because they confer ligand-independent growth when transfected activated Kit. In addition to ICC hyperplasia and GISTs, into otherwise Kit ligand-dependent Ba/F3 cells in vitro and homozygous Kit K641E mice exhibit loss-of-function Kit tumors when transfected cells are injected into nude mice (5, 12). phenotypes, including white coat color, decreased numbers Kit is involved in the development and maintenance of hemato- of dermal mast cells, and sterility, indicating that despite its poietic stem cells, mast cells, melanocytes, germ cells, and oncogenic activity the mutant form cannot accomplish many interstitial cells of Cajal (ICC; refs. 5, 13–16). Loss-of-function activities of the wild-type gene. Kit K641E reproduces the causes developmental defects in all of these cellular compartments, pathology associated with the familial GIST syndrome and including anemia, sterility, loss of ICC, and white coat color ‘‘white thus is an excellent model to study Kit pathway activation, ICC spotting’’ due to a failure of melanocytes to migrate during biology, GIST pathogenesis, and preclinical validations of GIST melanogenesis (4, 13, 17, 18). therapies and mechanisms of drug resistance. (Cancer Res GISTs are believed to arise from the ICC lineage (5, 19). ICC are 2005; 65(15): 6631-9) mesenchymal cells located within the muscular wall of the gastrointestinal tract, where they coordinate peristalsis through Introduction inherent pacemaker function (13, 18, 20, 21). There is evidence to suggest that both ICC and smooth muscle of the gut wall arise from The important role of receptor tyrosine kinases (RTK) in cancer has been firmly established by recent developments in both the a common stem cell and that although Kit expression is not pathogenesis and therapy of human cancer. Numerous RTKs are necessary for fate determination Kit is essential for maintenance either mutationally activated or involved in translocations that and proliferation of ICCs (17). Indeed, Kit antigen is the result in constitutive, ligand-independent kinase activity (1). A immunohistochemical marker of choice for identifying ICCs breakthrough in therapy for cancers associated with activating (5, 17–19). Loss-of-function Kit mutations and/or decreased Kit mutations in tyrosine kinases was the development of imatinib levels not only result in ICC hypoplasia but also have functional mesylate (Glivec, Gleevec) for the treatment of chronic myeloge- consequences as shown by disturbances in peristalsis and, in nous leukemia (CML; ref. 2). Imatinib targets BCR-Abl kinase in extreme cases, severe dysmotility (13, 22). CML, resulting in strong inhibition of BCR-Abl-mediated signaling. As a mesenchymal derivative, GISTs are sarcomas; the most IFN-refractory CML responds dramatically to imatinib; this common sarcoma of the gastrointestinal tract. At least two signal development of a molecularly targeted therapy is a landmark in transduction pathways are important in the pathogenesis of GISTs; cancer treatment. Imatinib was subsequently shown to inhibit Kit, activating mutations in Kit and PDGFRA have been identified in platelet-derived growth factor receptors (PDGFR) A and B, and 88% and 5% of GISTs, respectively (5, 12, 23). Additional support for the relationship between mutational activation of Kit or PDGFRA and pathogenesis of GIST came from studies of familial GIST, which is characterized by ICC hyperplasia and multiple GISTs with an autosomal dominant pattern of inheritance (10, 24–28). Familial Requests for reprints: Brian P. Rubin, Department of Anatomic Pathology, GIST patients carry activating germ line Kit or PDGFRA mutations University of Washington Medical Center, 1959 Northeast Pacific Street, Box 356100, identical to those found in sporadic GISTs. Seattle, WA 98195. Phone: 206-598-5024; Fax: 206-598-8697; E-mail: bprubin@ u.washington.edu. Thus, the accumulated evidence argues strongly for a role for Kit I2005 American Association for Cancer Research. in the development of ICCs and their transformation into GISTs. www.aacrjournals.org 6631 Cancer Res 2005; 65: (15). August 1, 2005

Unlike the multistep pathways elucidated for colon cancer, little is restriction site within a polylinker 5V of the neomycin resistance cassette known about the intervening steps in the genesis of sarcomas in was also present and was used in mapping homologous integration by general or GIST in particular. Therefore, we sought to create a Southern blotting. 129Sv embryonic stem cells were electroporated with mouse model of GIST by introducing an activating Kit mutation, AscI linearized targeting construct using standard protocols. Ninety-four K641E, associated with human familial GIST syndrome (K642E in neomycin-resistant clones were isolated and analyzed for homologous integration by Southern blot analysis after BglII digestion with a 3VKit probe, human Kit) into the mouse genome (27). This mouse model was which flanked the region of the Kit gene used in the targeting construct. anticipated to provide unique insights into the role of Kit in the The presence of the K641E mutation was confirmed by amplification with pathogenesis of GISTs as well as the biology of ICCs and normal Kit primers B and C of a 478-bp fragment harboring the mutation and a novel function. Furthermore, it was apparent that such a model would be XhoI site that was introduced by the exon 13 mutation. XhoI digestion of useful in the preclinical evaluation of potential anti-Kit therapies this fragment yields 403- and 75-bp fragments. The sequence of the PCR and the study of mechanisms of resistance to these therapies. fragments was confirmed by DNA sequence analysis. Two correctly targeted Herein, we report a ‘‘knock-in’’ transgenic mouse model with clones were identified and one of these clones was microinjected into C57B6 activated oncogenic Kit expression, impressive ICC hyperplasia, blastocysts and transferred to the uterus of E2.5 pseudopregnant recipients GISTs, and, surprisingly, multiple manifestations of decreased Kit according to standard procedures (E0.5 is the day of plug). Male offspring function. This mouse model is not without precedent, as another displaying at least 50% chimerism were backcrossed to C57B6 and 129SvJ group recently reported the construction of a similar mouse model females, respectively, to establish germ line transmission. Mutant mice Â harboring a Kit V558 deletion (29). were maintained on a mixed (129Sv C57B6) or 129Sv genetic background as described. All mouse experiments were done with approval of the University of Washington Animal Care and Use Committee and in accor- Materials and Methods dance with the guidelines established by the NIH. The primers are as Generation of mutant Kit K641E mice. Site-directed mutagenesis using follows: primer A, 5V-GCCCTAATGTCGGAACTCGAGGTCCTGAGCTACC- QuickChange XL Site-Directed Mutagenesis kit (Stratagene, La Jolla, CA) and TG-3V;primerB,5V-AGTTGGCAGGGTTAGCAGAA-3V; and primer C, 5V- primer A introduced a K-to-E mutation at amino acid position 641 within Kit AGACTCACCTCCCACCGTGC-3V. exon 13. The mutation also introduced a novel XhoI restriction site. A 4.9-kb Flow cytometry. Single-cell suspensions of bone marrow from 6- to HindIII-HindIII fragment spanning exons 8 to 13 and containing the K641E 12-week-old littermate mice were prepared from femur and tibia. RBCs were mutation and an adjacent 2.1-kb HindIII-EcoRI fragment containing exon 14 lysed with ACK lysis buffer and nonspecific staining was blocked using anti- were subcloned on either side of a neomycin resistance gene under the Fc receptor 24G2 hybridoma supernatant. Cells were stained on ice in PBS control of the W/UV5 bacterial promoter in the same orientation as the containing 2% FCS and 0.05% sodium azide at 2 Â 107/mL with Kit gene, flanked by loxP sites within targeting vector 4317G9 (provided by Dr. allophycocyanin-conjugated CD117 (Kit) or isotype control (rat IgG2b, n) Richard Palmiter, University of Washington, Seattle, WA). This targeting and phycoerythrin-conjugated CD90.2 (Thy1.2), CD11b (Mac-1), Ly6G (Gr-1), vector (described in Fig. 1A) also featured two negative selection cassettes CD45R (B220), and TER119 (BD Biosciences, San Jose, CA). Samples were on either side of the cloned fragments: the diphtheria toxin A gene (DT-A) and acquired on a FACSCalibur flow cytometer (BD Biosciences) and analyzed the herpes simplex virus-thymidine kinase gene (HSV-TK). A unique BglII using FlowJo software (Treestar).

Figure 1. Targeting strategy. A, schematic representation of K641E targeting strategy. Restriction enzyme sites are as follows: B, BglII; H, HindIII; E, EcoRI. Negative selection was achieved through DT-A at the 5V end of the construct and HSV-TK at the 3V end of the construct. Positive selection was achieved through a neomycin resistance cassette (neo). Location of 3V primer used for Southern blots and PCR primers are shown. B, PCR genotyping assay containing representative Kit +/+, Kit+/K641E:Neo, and Kit K641E:Neo/K641E:Neo tail DNAs. The first lane of each pair shows the results of digestion with XhoI; the K641E mutation introduced a novel XhoI site into the PCR product resulting in a 403-bp product after digestion with XhoI. D, XhoI digested PCR products; U, undigested PCR products.

Cancer Res 2005; 65: (15). August 1, 2005 6632 www.aacrjournals.org

Immunoprecipitation and Western blotting. A representative portion consisted of a 7-kb fragment of the Kit gene containing exons 8 to of all GISTs was snap frozen for biochemical analysis. Snap-frozen tumor 14, including the K641E mutation, which was introduced by a site- (0.04 g) was ground to powder in liquid nitrogen, resuspended in NP40 lysis directed mutagenesis strategy, and a neomycin resistance cassette buffer (1 mL) containing 50 mmol/L Tris-HCL (pH 8.0), 150 mmol/L NaCl, flanked by loxP sites within intron 13 (Fig. 1A). Homologous 5 mmol/L EDTA, 1% NP40, and mini-tablet protease mix (1 tablet/10 mL, integration of the targeting vector into 129Sv embryonic stem cells Roche Applied Science, Indianapolis, IN), passed through a 21-gauge needle 10 times, incubated on ice for 30 minutes, and centrifuged at 16,000 Â g resulted in two correctly targeted embryonic stem cell clones, (14,000 rpm in an Eppendorf 5415C centrifuge) for 10 minutes at 4jC. which were confirmed by Southern blot hybridization, PCR, and Lysates were precleared with 20 AL protein G-Sepharose beads (Zymed DNA sequence analysis. Embryonic stem cells from one of two Laboratories, San Francisco, CA) for 30 minutes. Precleared lysates were clones were microinjected into C57B6 blastocysts, and the resultant precipitated for 1 hour with goat anti-mouse anti-Kit antibody (M-14, Santa male chimeras were backcrossed with C57B6 females to establish Cruz Biotechnology, Santa Cruz, CA) followed by 1-hour incubation with 20 germ line transmission, which was confirmed by Southern blot AL protein G-Sepharose at 4jC. Immunoprecipitates were washed and hybridization and PCR (Fig. 1B). fractionated on 8% PAGE containing 0.1% SDS. For Western blotting, rabbit Loss-of-function phenotypes in KitK641E mice. Breeding of anti-Kit (C-19, Santa Cruz Biotechnology) or mouse anti-Tyr(P) (PY99, Santa Kit+/K641E:Neo mice produced 243 offspring of the following Cruz Biotechnology) were used. Secondary antibodies were either genotypes: 125 Kit+/+,78Kit+/K641E:Neo, and 40 KitK641E:Neo/K641E:Neo horseradish peroxidase–conjugated goat anti-rabbit or anti-mouse IgG (Zymed Laboratories). The blots were visualized with SuperSignal West Pico mice. There is a clear departure from Mendelian distribu- m2 m2 Chemiluminescent Substrate (Pierce Biotechnology, Rockford, IL). tion ( analysis, = 90.61, P < 0.001; 95% confidence limits). +/K641E:Neo Histologic analysis and immunohistochemistry. Complete necropsies Whereas Kit mice were normal externally with black K641E:Neo/K641E:Neo were done on three Kit +/K641E:Neo and three KitK641E:Neo/K641E:Neo mice. coats, Kit mice had uniform white coats (Fig. 2A), Organs examined included brain, eyes, heart, lungs, thymus, liver, spleen, which suggested that the mutant KitK641E:Neo allele phenocopied gastrointestinal tract, kidneys, bladder, gonads, bone marrow, and skin. the ‘‘white spotting’’ phenotype, previously associated with loss of Because the gastrointestinal tract, gonads, and skin were abnormal, these Kit function. Unlike albino mice, which have pink eyes, these ‘‘white tissues were routinely harvested from a large cohort of Kit +/+ (n = 11), +/K641E:Neo K641E:Neo/K641E:Neo spotting’’ mice have normally pigmented dark brown eyes Kit (n = 22), and Kit (n = 35) mice. Sections consistent with a specific defect. Flow cytometric analysis of adult (4 Am) from tissues fixed in 10% formalin were stained with H&E or used for bone marrow from homozygous mutants revealed an undetectable immunohistochemistry. Immunohistochemistry was done by the avidin- biotin-peroxidase complex technique using commercially available anti- level of cell surface Kit expression in contrast with heterozygous or bodies to the following antigens: CD34 (QBEnd 10; 1:100; DAKO, wild-type littermates (Fig. 2B). In conjunction with this decreased K641E:Neo Carpinteria, CA), Kit (polyclonal, 1:150, Santa Cruz Biotechnology), a- Kit expression, the striking white phenotype strongly smooth muscle actin (1A4, 1:2,000, Beckman Coulter, Fullerton, CA), desmin suggests loss of Kit function mechanism in this mouse model. (D33, 1:500, DAKO), and inhibin (inhibin a monoclonal, 1:10, Serotec, Furthermore, homozygous KitK641E:Neo/K641E:Neo mice displayed two Raleigh, NC). Antigen retrieval consisted of 15 minutes of microwave other loss-of-function phenotypes: low numbers of dermal mast pretreatment in citrate buffer before the application of all primary cells and infertility; both males and females failed to produce antibodies. Appropriate positive and negative controls were done. offspring when bred with Kit+/+ littermates. The latter two Electron microscopy. Tissue was fixed in 2% glutaraldehyde, postfixed phenotypes were investigated histologically. in 1% osmium tetroxide, and embedded in epoxy resin using standard Microscopically, the reason for the sterility phenotype in procedures. Sections were examined with a Philips EM410 transmission KitK641E:Neo/K641E:Neo mice was obvious. Male KitK641E:Neo/K641E:Neo electron microscope (Philips, Eindhoven, the Netherlands). K641E:Neo/ Determination of mast cell numbers. Skin mast cell numbers were mice did not contain mature spermatids and female Kit K641E:Neo determined by counting the number of mast cells within 10 Â400 fields mice lacked identifiable germ cells. The testes contained (Olympus BX51 microscope, Olympus America, Melville, NY) in sections seminiferous tubules with only one layer of primary germ cells and a from the dermis of the back after staining with 1% toluidine blue for complete absence of spermatids and spermatozoa (Fig. 2C). Instead, 2 minutes. Sections were counted twice by each of two pathologists (B.P.R. the seminiferous tubules contained abundant, pale, eosinophilic, and M.T.) and the numbers were averaged. flocculent secretions (Fig. 2C). As the KitK641E:Neo/K641E:Neo mice m2 Statistical analysis. Mendelian frequencies were evaluated by increased in age, there was progressive Leydig cell hyperplasia, and analysis. To evaluate differences in survival between different genotypes, a by age 3 months, the testis had severe Leydig cell hyperplasia and the log-rank test was done. The data are presented via Kaplan-Meier curves. To germ cells were sloughed into the center of the seminiferous tubules evaluate differences in the number of dermal mast cells, ANOVA and K641E:Neo/K641E:Neo Tukey’s tests were done. To evaluate differences in cecal diameter, a square (Fig. 2D). The ovaries of the Kit females were small, root transformation was done and the data were evaluated by one-way difficult to find, and devoid of germ cells (Fig. 2E). They consisted of ANOVA and the Bonferroni multiple comparisons test. stromal cells that ranged from spindle shaped to more rounded inhibin-positive/Kit-negative luteinized cells (data not shown). In some of the ovaries, the stromal cells were arranged focally in Results tubules (Fig. 2E). The gonads of Kit+/K641E:Neo mice were histolog- Introduction of KitK641E into the mouse genome by allelic ically normal. conversion. Lys641 is located within the first portion of the split Because the density of dermal mast cells is dependent on Kit kinase domain encoded by exon 13, which shows 97% identity at levels, we postulated that we might observe changes in the density the protein level between human and mouse Kit proteins (30, 31). or distribution of these cells as a functional consequence of Overall, mouse and human Kit are 82% identical at the protein activated Kit expression. We examined sections of skin and level. By analogy to familial GIST patients, we reasoned that mice determined the relative numbers of mast cells as a function of harboring a Kit K641E allele would also develop ICC hyperplasia genotype. The number of mast cells in the KitK641E:Neo/K641E:Neo and GISTs. To model GISTs in mice, we developed a targeting mice was lower than in either heterozygous Kit+/K641E:Neo or vector to introduce the Kit K641E mutation, which causes familial homozygous Kit+/+ mice; KitK641E:Neo/K641E:Neo = 0.0 to 14.5 GISTs in humans, into the mouse Kit locus. The targeting vector (median, 3.5; n = 18), Kit+/K64E:Neo = 20.0 to 145.0 (median, 45.8; www.aacrjournals.org 6633 Cancer Res 2005; 65: (15). August 1, 2005

Figure 2. Loss-of-function Kit phenotypes. A, homozygous Kit K641E:Neo/K641E:Neo ‘‘white spotting’’ mouse with uniformly white coat and black eyes. B, flow cytometric analysis of bone marrow. The distribution of Kit expression versus a panel of lineage (Lin) markers (Thy1.2, CD11b, G4-1, B220, and TER119)is shown. Expression of Kit is high on LinÀ in Kit +/+ mice (long arrow, top), f1% to 2% of total bone marrow, and intermediate on a subset of Linhi cells (short arrow, top). In Kit K641E:Neo/K641E:Neomice, Kit expression is undetectable over the isotype control antibody for Linhi subsets and reduced >10-fold in the LinÀ population. In heterozygous Kit +/K641E:Neo mice, Kit expression is reduced slightly in Linhi and LinÀ compared with Kit +/+ mice. C, 1-month-old Kit K641E:Neo/K641E:Neo testis with seminiferous tubules showing a single layer of germ cells and a complete lack of spermatozoa (arrow). D, 3-month-old testis showing extensive Leydig cell hyperplasia (*) and germ cells that are sloughed into the center of the seminiferous tubules (arrow). E, 7-month-old Kit K641E:Neo/K641E:Neo ovary composed only of spindle-shaped and luteinized stromal cells (short arrow); germ cells are not identified. Unusual tubular structures are identified (long arrow). F, decreased numbers of dermal mast cells in Kit K641E:Neo/K641E:Neomice compared with either Kit +/K641E:Neo or Kit +/+ mice (P < 0.001). n = 16), and Kit+/+ = 29.5 to 52.5 (median, 43; n =7;P < 0.001; symptoms of abdominal distension, so that necropsies could be Fig. 2F). Thus, to our surprise, the KitK641E:Neo/K641E:Neo mice exhi- done and tissue could be collected, before the occurrence of signi- bited three different loss-of-function phenotypes: white coat color, ficant autolysis. KitK641E:Neo/K641E:Neo mice developed symptoms sterility, and decreased density of dermal mast cells. However, as and either died or were sacrificed due to symptoms of obstruction described below, this same mouse exhibited a striking gain-of- from 3 to 30 weeks (median, 5 weeks; P < 0.001; Fig. 3A), whereas function phenotype within the gastrointestinal tract. their Kit+/K641E:Neo littermates were viable and asymptomatic. Interstitial cells of Cajal hyperplasia and gastrointestinal Because we hypothesized that the Kit K641E mutation would stromal tumors in KitK641E:Neo mice. Within 3 weeks of birth, cause ICC hyperplasia and GIST formation, a detailed examination coincident with weaning and the onset of solid food intake, a of the gastrointestinal tract was done in each mouse. Grossly, the subset of KitK641E:Neo/K641E:Neo mice developed distended abdomens gastrointestinal tract of KitK641E:Neo/K641E:Neo mice was obstructed and died within days of apparent intestinal obstruction. A protocol at the pylorus (22 of 35 mice) or the cecum (12 of 35 mice), was instituted to sacrifice all mice as soon as they began to exhibit leading to a distended stomach or mega-ileum, respectively

Cancer Res 2005; 65: (15). August 1, 2005 6634 www.aacrjournals.org

(Figs. 3B-D and 4A). Furthermore, the ceca of homozygous and the immunoprecipitate was Western blotted with anti-Kit KitK641E:Neo/K641E:Neo mice were always abnormally firm and and anti-phosphotyrosine antibodies after PAGE (Materials and contained tumors that could be distinguished grossly (Fig. 3D-F). Methods). The cecal masses contained substantial amounts of Ceca of the differing genotypes had the following diameters: phosphorylated Kit (Fig. 3G). Interestingly, Kit seemed to be more KitK641E:Neo/K641E:Neo = 0.05 to 1.2 cm (average, 0.19 cm; n =35), strongly phosphorylated in comparison with immunoprecipitates Kit+/K64E:Neo = 0.03 to 0.22 cm (average, 0.11 cm; n =18),andKit+/+ from mouse P815 mastocytoma cell line (Fig. 3G). mice = 0.01 to 0.02 cm (average, 0.015 cm; n = 10). The difference in To more fully understand the reasons for the obstruction cecum diameter between Kit+/K641E:Neo or KitK641E:Neo/K641E:Neo phenotypes seen in the KitK641E:Neo/K641E:Neo mice, histologic and Kit+/+ mice was statistically significant (P <0.001).Thedifference examination of the entire gastrointestinal tract was done and between Kit+/K641E:Neo and KitK641E:Neo/K641E:Neo mice was also sta- compared with Kit+/K641E:Neo and Kit+/+ mice. This analysis revealed tistically significant (P = 0.05). Thus, expression of activated Kit has relatively uniform histologic findings across all mice of each an allele dosage–dependent, profound effect on the gastrointestinal genotype. Beginning within the distal esophagus and extending tract, culminating in frank intestinal obstruction and death. throughout the stomach and into the proximal duodenum, the wall To determine the extent of Kit expression and activation in the of the gut of KitK641E:Neo/K641E:Neo mice was involved by ICC cecal tumors, Kit immunoprecipitation was done on tumor lysates, hyperplasia that arose from the myenteric plexus and invaded the

Figure 3. A, Kaplan-Meier curve of Kit +/+, Kit +/K641E:Neo, and Kit K641E:Neo/K641E:Neo mice. Survival is diminished significantly in Kit K641E:Neo/K641E:Neo mice, with all of these mice succumbing to gastric or intestinal obstruction by age 30 weeks (P < 0.001). B, normal appearance of the abdominal contents of a 1-month-old Kit +/+ mouse compared with the contents of a 1-month-old Kit K641E:Neo/K641E:Neo mouse with mega-ileum due to a large cecal GIST (C). The arrow in B points to the normal small bowel. All that can be seen within the bulging abdomen of the mouse in C is the massively distended small bowel filled with fecal material. D, gastrointestinal tract of a 6-month-old Kit K641E:Neo/K641E:Neo mouse with an enlarged and grossly distended stomach (arrowhead) and cecum involved by a massive GIST measuring 1.2 cm in greatest diameter (arrow). E, close-up of the cecum from D. The cecum is involved by a massive GIST, which is white in contrast to the normal color of the cecum (brown). F, cross-sections of GIST showing fleshy cut surface and remaining cecal lumen (arrow). G, expression and phosphorylation of Kit in GISTs from two different Kit K641E:Neo/K641E:Neo mice (1 and 2). Protein lysates were immunoprecipitated with anti-Kit antibody M-14 and analyzed by SDS-PAGE. Membranes were blotted with anti-Kit antibody C-19 (top) and anti-phosphotyrosine antibody PY99 (bottom). A protein lysate from mouse P815 mastocytoma cell line [M; courtesy of Dr. Michael Heinrich (Oregon Health Sciences University, Portland, OR)] is shown as a control. Note that the GISTs contain considerably more activated Kit than the mastocytoma cell line that was shown to be phosphorylated at longer exposures. www.aacrjournals.org 6635 Cancer Res 2005; 65: (15). August 1, 2005

Figure 4. ICC hyperplasia in Kit K641E:Neo/K641E:Neo mice. A, stomach from Kit K641E:Neo/K641E:Neo mouse with extensive ICC hyperplasia leading to tensely distended stomach. Arrow, pyloric region that has a faint white color due to extensive involvement by ICC hyperplasia. The proximal third of the stomach (*)is usually paler than the rest of the stomach due to the presence of squamous mucosa, which is a normal finding. B, stomach wall from a Kit +/+ mouse. Right bracket, normal thickness of the muscular wall. C, gastric wall from a Kit K641E:Neo/K641E:Neo mouse showing moderate ICC hyperplasia involving the outer half of the muscularis propria (right bracket). Note that the wall thickness is approximately twice as thick the normal stomach wall in A. D, ICC hyperplasia from a Kit K641E:Neo/K641E:Neo mouse involving the muscular wall of the pylorus (right bracket). Note that hyperplasia is arising in between the muscular layers (*) from the location of Auerbach’s plexus. E, low-power view of the pylorus from a Kit K641E:Neo/K641E:Neo mouse involved by extensive ICC hyperplasia (right bracket), virtually replacing the outer portion of the muscularis. F, pylorus from a Kit K641E:Neo/K641E:Neo mouse that is extensively infiltrated by ICC hyperplasia (*) resulting in near total obstruction of the lumen (arrow). This section was taken from a mouse that presented with clinical gastric obstruction. surrounding musculature (Fig. 4C-F). ICC hyperplasia was most involved by diffuse ICC hyperplasia emanating from the myenteric prominent in the pylorus, where it completely effaced the muscular plexus, involving, and in some areas completely replacing, the wall and extended through the muscularis mucosae to engulf the surrounding muscularis propria (data not shown). Metastases were mucosa, almost certainly explaining the obstruction and never seen either grossly or microscopically. gastric bloating that had been seen (Fig. 4D-F). With the exception Although heterozygous Kit+/K641E:Neo mice did not succumb to of the most proximal portion of the duodenum, the small gastrointestinal lesions, necropsy revealed small cecal GISTs, and intestine did not contain any evidence of ICC hyperplasia. In the histologic analysis revealed ICC hyperplasia, similar in appearance KitK641E:Neo/K641E:Neo mice that presented with mega-ileum (n =12 to the ICC hyperplasia seen in KitK641E:Neo/K641E:Neo mice. However, of 35), the ceca were grossly enlarged and on cross-section had in contrast to the homozygous mutant mice, ICC hyperplasia in the lumens that were narrowed by fleshy-appearing GISTs. Histologic heterozygous mice involved only the cecum and large bowel and analysis of the GISTs revealed that they were composed of a dense, was much less extensive, showing a correlation with copy number cellular proliferation of monomorphic spindle cells, similar in of the activated Kit allele (data not shown). A summary of the appearance to, but much more extensive, than the ICC hyperplasia histologic findings is presented in Table 1. of the upper gastrointestinal tract (Fig. 5A and C). Based on the Cytologically, hyperplasia and GISTs from Kit+/K641E:Neo and analysis of 35 ceca from mice of various ages, the natural history of KitK641E:Neo/K641E:Neo mice were indistinguishable. The hyperplastic GISTs in this location seemed to follow an orderly progression ICCs within the upper gastrointestinal tract were longer than those (data not shown). ICC hyperplasia was first apparent within the within regions of ICC hyperplasia and GISTs of the lower myenteric plexus and grew asymmetrically, involving predomi- gastrointestinal tract (data not shown). However, overall, the nantly one side of the cecum. As this process continued, morphology of the lesions was similar. The cells within ICC circumferential involvement of the gut wall resulted. At its greatest hyperplasia and GISTs were monomorphic, plump spindle cells extent, the muscular wall was completely replaced by the mass with abundant pale eosinophilic, fibrillary cytoplasm and elon- that variably crowded the lumen, causing obstruction and death. gated nuclei with fine chromatin and inconspicuous nucleoli Some GISTs seemed to grow outward and did not result in (Figs. 4F and 5C). Occasional cells contained paranuclear vacuoles intestinal obstruction. The largest GISTs extended from the serosal identical to those seen in human GISTs (Fig. 5C, inset). Mitotic surface, through the muscularis propria, into and through the figures were extremely uncommon and numbered no more muscularis mucosae to invade into the mucosal/epithelial surface than 1 per 50 high-power fields (Â400 field). Necrosis was not (Fig. 5A). The colon and rectum were also circumferentially identified.

Cancer Res 2005; 65: (15). August 1, 2005 6636 www.aacrjournals.org

Figure 5. GISTs in KitK641E:Neo/K641E:Neo mice. A, very low power view of a large cecal GIST from a Kit K641E:Neo/K641E:Neo mouse. Note the asymmetrical +/+ involvement as the lumen (*) is pushed to the left. B, the cecal wall from a Kit mouse is shown in the inset for comparison at the same magnification as the cecal GIST in A. A high-power view of the cecal wall is shown to highlight the lack of cellularity of the muscular wall (right bracket). C, very high power view of a cecal GIST from a Kit K641E:Neo/K641E:Neo mouse showing the characteristic cytologic features: a proliferation of monomorphic, plump, spindle-shaped cells with pale fibrillary cytoplasm arranged in fascicles. Inset, a different case that had extensive paranuclear vacuolization, a finding that is commonly seen in gastric GISTs in humans. D, GIST showing diffuse and strong immunoreactivity for Kit. E, desmin-negative GIST infiltrating through the smooth muscle wall of the cecum, which is an excellent marker of smooth muscle and is seen in this panel as scattered desmin-positive (brown) cells and more extensive immunoreactivity in the portion of the muscular wall that is not involved by GIST. F, electron microscopy of cecal GIST showing undifferentiated plump spindle cells with numerous mitochondria (*). Magnification, Â4,400.

Immunohistochemistry revealed the cells regardless of whether perinuclear vacuoles, similar to what is seen in human GISTs. Basal they arose in ICC hyperplasia or GIST, or in either Kit+/K641E:Neo or lamina was not identified with the exception of an incomplete basal KitK641E:Neo/K641E:Neo mice, to be diffusely and strongly positive for lamina within one of the ICC hyperplasias. Kit (Fig. 5D) and negative for desmin (Fig. 5E), a-smooth muscle Taken together, the histologic, immunohistochemical, and actin, CD34, and S-100 protein (data not shown). ultrastructural features support the interpretation that the cells Ultrastructural analysis was done on two cases of ICC hyperplasia identified within the gastrointestinal tract of mice harboring either involving the pylorus and two cecal GISTs. The cells exhibited one or two copies of the KitK641E:Neo allele are hyperplastic ICCs similar ultrastructural findings (Fig. 5F): a spindle cell cytomor- and GISTs. phology with smooth-contoured, fusiform nuclei containing evenly Other organs that were examined histologically were found to dispersed chromatin and a moderate amount of cytoplasm packed be normal in both Kit+/K641E:Neo and KitK641E:Neo/K641E:Neo mice (see with mitochondria. A well-defined Golgi apparatus was seen in most Materials and Methods). cells. The cells within the ICC hyperplasia from the pylorus To exclude any influences on strain background, male possessed short cell processes, subplasmalemmal vacuoles, and chimeras were independently backcrossed to 129SvJ females to rudimentary cell junctions. In one of the two GISTs that were produce a 129 strain harboring Kit K641E alleles. We examined studied, rare cells showed focal densities, possibly representing 12 KitK641E:Neo/K641E:Neo and 12 Kit+/K641E:Neo mice and found that actin microfilaments. The other GIST contained numerous both loss-of-function (white coat, mast cell deficiency, and sterility)

Table 1. Summary of pathologic findings within the gastrointestinal tract of Kit+/+, Kit+/K641E:Neo, and KitK641E:Neo/K641E:Neo mice

Genotype Esophagus Stomach Proximal duodenum Cecum Cecum GISTs Colon/rectum ICC hyperplasia ICC hyperplasia ICC hyperplasia ICC hyperplasia ICC hyperplasia

Kit+/+ 0/11 0/11 0/11 0/11 0/11 0/11 Kit+/K641E:Neo 0/21 0/22 0/22 4/22 15/22 14/22 KitK641E:Neo/K641E:Neo 10/29 31/33 13/34 0/35 35/35 32/34

www.aacrjournals.org 6637 Cancer Res 2005; 65: (15). August 1, 2005

Downloaded from cancerres.aacrjournals.org on September 28, 2021. © 2005 American Association for Cancer Research. Cancer Research and gain-of-function (ICC proliferations/GISTs) phenotypes were activated by the Kit K641E allele, whereas Kit-dependent mast cells, identical to those seen in the 129B6 mixed background (data not melanocytes, and germ cells are not. Does the ICC lineage express shown), suggesting that the genetic background does not affect the particularly high levels of the oncogenic Kit allele or does development of ICC hyperplasia or GISTs. something inherent in the Kit K641E mutation and/or Kit signaling pathway permit a dramatic response to low levels of the mutant Discussion receptor? It is known that mouse and human mast cell leukemias/ mastocytosis and human germ cell neoplasms possess activating We describe a knock-in mouse model in which a mutation found Kit mutations, but typically, these mutations are not within the in sporadic and familial human GISTs, Kit K641E, was placed within same region of the Kit gene as those seen in GISTs (6, 36). Kit the context of the mouse Kit gene through homologous recombi- mutations found in GISTs most commonly involve the intracellular nation. Mice that harbored either one or two copies of the Kit juxtamembrane domain in exon 11, whereas mast cell and germ K641E allele developed ICC hyperplasia and GISTs. Interestingly, the cell neoplasms generally have mutations in the phosphotransferase distribution of GISTs is different than is seen in humans. Sporadic domain in exon 17. In addition, some Kit alleles are associated with human GISTs are most common in the stomach (50-60%) followed more widespread effects, including gain-of-function phenotypes in by the proximal small intestine (20-30%). Human GISTs of the large cellular lineages other than ICC. For example, the mouse model bowel are very rare (32). Human patients who carry germ line Kit developed by Sommer et al. not only produces ICC hyperplasia and K642E mutations develop both gastric and small bowel GISTs (33). GISTs but also develop unusual proliferations of neuroectodermal Although ICC hyperplasia was identified within the same spectrum cells with features of melanocytes/melanotic Schwann cells and of anatomic locations as human GISTs, mouse GISTs were located increased (four times) numbers of dermal mast cells (29). Their within the cecum. This distribution is the same as that found within phenotype is not a total surprise because the affected individuals another recently described mouse GIST model in which the Kit from the familial GIST syndrome family that this mouse is based on V558D mutation (a mutation found in a familial human GIST also have urticaria (increased dermal mast cells) in addition to ICC syndrome family) was ‘‘knocked’’ into the Kit locus (29). The reason hyperplasia and GISTs (10). Thus, although activating Kit alleles for the differences in distribution of GISTs between mice and can be selective in terms of which cell lineages are activated, not all humans is not clear but may have to do with differences in the Kit alleles exhibit such exquisite selectivity. density of ICC progenitor cells in different species. Nevertheless, It is likely that differences in proliferative capacity/activation in ICC hyperplasia and GISTs identified within the mice are identical response to different mutations lie in inherent differences in the to human GISTs at all of the histologic, immunohistochemical, cell types in which Kit is expressed. At the molecular level, this ultrastructural, and molecular levels and thus recapitulate many leads to the hypothesis of lineage-specific proteins that interact aspects of human GIST biology. As such, our model may help with Kit. Kit is known to act directly and indirectly with many understand various types of developmental and acquired gastroin- different proteins, and the juxtamembrane domain, the site of the testinal diseases related to ICC, given that ICC hyperplasia majority of Kit mutations found in GISTs, has key roles in protein- invariably accompanied expression of the Kit K641E allele (34). protein interactions (37). Although the explanation for the One of the fascinating and unexpected aspects of our mouse exquisite lineage specificity seen in our model remains unex- model is the apparent dose dependence of Kit activation on the plained, the power of mouse genetics and the availability of extent of ICC hyperplasia and GIST formation. Mice harboring one relatively homogeneous lesions for study should allow us to explore copy of the Kit K641E allele had ICC hyperplasia limited to the colon this question more thoroughly. and smaller GISTs than mice harboring two copies of the Kit K641E Based on the identification of Kit mutations in the majority of allele (P < 0.001). The mechanism of Kit K641E activation is unknown sporadic and familial human GISTs as well as the dramatic ICC and it is also not known whether Kit K641E has altered substrate hyperplasia and GIST phenotypes in the mice described in this specificity, which might contribute to differences that are seen report, it is clear that Kit activation is sufficient to initiate GIST between Kit K641E heterozygotes and homozygotes. We expect that formation. ICC proliferation is an important early event that further overexpression of constitutively activated Kit might result in dramatically increases the number of cellular targets for accrual of an even more pronounced neoplastic phenotype. Because ICC further genetic changes that are instrumental in neoplastic hyperplasia was not seen in the stomachs of Kit+/K641E:Neo mice, progression to malignant/metastatic phenotypes in GIST. Further- although it was seen in the stomachs of KitK641E:Neo/K641E:Neo mice, it more, activating Kit mutations are identified within human GISTs suggests that gastric ICCs may have a higher threshold for response that are discovered incidentally and measure <1 cm in greatest to activated Kit than colonic ICCs. dimension, which also supports the idea that Kit activation is the Although the gain-of-function phenotypes that were seen in the first step in the genesis of GISTs (38). Metastatic neoplasms have not Kit K641E mice were expected, the finding of three loss-of-function been identified in the mice described in this article thus far or in a phenotypes in these mice was unanticipated. The loss-of-function similar knock-in mouse GISTmodel that was published recently (29). phenotypes could be due to impaired expression of the KitK641E:Neo We believe that through the considerable power of mouse genetics it allele retaining the Neo cassette. Poor expression has been seen in will be possible to tease apart the pathway of GIST neoplastic other genetically targeted Kit alleles harboring a neomycin progression. resistance cassette within Kit introns (35). We are currently in the After the discovery that Kit was mutationally activated in human process of excising the neomycin resistance cassette in intron 13 by GISTs, it was proposed that targeted anti-Kit therapies might be Cre-mediated recombination of the loxP sites, which flank this useful in their treatment (39, 40). This has turned out to be the case cassette, to see if this affects Kit expression or has phenotypic as initial clinical trials with imatinib, a potent small molecule consequences. inhibitor of Kit, have shown considerable promise (12, 41). As has The presence of both loss-of-function and gain-of-function been seen in imatinib therapy of CML, some patients who have been phenotypes begs the question as to why ICC lineage cells are maintained on imatinib for many months have developed resistance

Cancer Res 2005; 65: (15). August 1, 2005 6638 www.aacrjournals.org

Downloaded from cancerres.aacrjournals.org on September 28, 2021. © 2005 American Association for Cancer Research. Mouse GIST to the drug. Although very little data are available at this time, the progression in GIST. Furthermore, the model has great potential as a most common mechanism of resistance in GIST is through preclinical system for testing targeted therapies of the Kit pathway acquisition of a secondary mutation in the imatinib-resistant as well as resistance to anti-Kit–based therapies. Finally, as Kit phosphotransferase domain encoded by exon 17 (42, 43). The mouse activation in GIST has become a paradigm for activated RTKs in model described in this article may be useful in modeling resistance sarcoma, it is hoped that this mouse model may also shed light on to different anti-Kit therapies and serve as an excellent preclinical the pathogenesis and potential targets for treatment of related, as model for the testing of anti-Kit/anti-GIST therapies that are yet untreatable, sarcomas. developed in the future. In this report, we have described a novel knock-in mouse model, Acknowledgments whereby Kit K641E, a mutation that was originally identified in Received 3/17/2005; revised 5/11/2005; accepted 5/18/2005. sporadic and familial human GISTs, was placed within the mouse Kit Grant support: University of Washington Royalty Research Fund (B.P. Rubin DF). gene through homologous recombination. Mice with this mutation The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance exhibit widespread ICC hyperplasia and GISTs and loss-of-function with 18 U.S.C. Section 1734 solely to indicate this fact. phenotypes affecting melanocytes, mast cells, and germ cells. The We thank Drs. Yuki Kimura and Alan Bernstein for Kit clone 3a12a, which was used to make the targeting construct; Dr. Richard Palmiter for advice on designing gene complex phenotypes exhibited by these mice highlight the targeting vectors and vector 4317G9, which served as the backbone of our targeting intricacies of Kit signal transduction within different cellular vector; Dr. Brent Wood for advice on flow analysis; Dr. Michael Heinrich for the P815 lineages. It is hoped that this model will be useful in understanding mastocytoma cell line; Monica Chaudari, Eric Johnson, Nate Mercaldo, and Dr. Andre Oliveira for statistical evaluation; Dr. Piper Treuting for help with mouse pathology; various aspects of Kit signal transduction, the biology of ICCs, and and Drs. Peter Besmer, David Brooks, Jonathan Fletcher, and Raj Kapur for advice on the pathogenesis of GIST, especially the mechanism of tumor the article.

References 16. Russel ES. Hereditary anemias of the mouse: a review of the Kit receptor tyrosine kinase. Proc Natl Acad Sci for geneticists. Adv Genet 1979;20:357–459. U S A 2003;100:6706–11. 1. Heinrich MC, Corless CL. Targeting mutant kinases in 17. Kluppel M, Huizinga JD, Malysz J, Bernstein A. 30. Yarden Y, Kuang WJ, Yang-Feng T, et al. Human proto- gastrointestinal stromal tumors: a paradigm for molecular Developmental origin and Kit-dependent development oncogene c-kit: a new cell surface receptor tyrosine therapy of other sarcomas. Cancer Treat Res 2004;120: of the interstitial cells of Cajal in the mammalian small kinase for an unidentified ligand. EMBO J 1987;6:3341–51. 129–50. intestine. Dev Dyn 1998;211:60–71. 31. Strausberg RL, Feingold EA, Grouse LH, et al. 2. Druker BJ, David A. Karnofsky award lecture. Imatinib 18. Torihashi S, Ward SM, Nishikawa S, Nishi K, Generation and initial analysis of more than 15,000 as a paradigm of targeted therapies. J Clin Oncol 2003 Kobayashi S, Sanders KM. c-kit-dependent development full-length human and mouse cDNA sequences. Proc Dec 1;21:239–45s. of interstitial cells and electrical activity in the murine Natl Acad Sci U S A 2002;99:16899–903. 3. Taylor ML, Metcalfe DD. Kit signal transduction. gastrointestinal tract. Cell Tissue Res 1995;280:97–111. 32. Fletcher CD, Berman JJ, Corless C, et al. Diagnosis of Hematol Oncol Clin North Am 2000;14:517–35. 19. Kindblom LG, Remotti HE, Aldenborg F, Meis- gastrointestinal stromal tumors: a consensus approach. 4. Lev S, Blechman JM,Givol D,Yarden Y. Steel factor and c- Kindblom JM. Gastrointestinal pacemaker cell tumor Hum Pathol 2002;33:459–65. kit protooncogene: genetic lessons in signal transduction. (GIPACT): gastrointestinal stromal tumors show phe- 33. Handra-Luca A, Flejou JF, Molas G, et al. Familial Crit Rev Oncog 1994;5:141–68. notypic characteristics of the interstitial cells of Cajal. multiple gastrointestinal stromal tumours with associ- 5. Hirota S, Isozaki K, Moriyama Y, et al. Gain-of-function Am J Pathol 1998;152:1259–69. ated abnormalities of the myenteric plexus layer and mutations of c-kit in human gastrointestinal stromal 20. Langton P, Ward SM, Carl A, Norell MA, Sanders KM. skeinoid fibres. Histopathology 2001;39:359–63. tumors. Science 1998;279:577–80. Spontaneous electrical activity of interstitial cells of 34. Vanderwinden JM, Rumessen JJ. Interstitial cells of 6. KemmerK,CorlessCL,FletcherJA,etal.KIT Cajal isolated from canine proximal colon. Proc Natl Cajal in human gut and gastrointestinal disease. Microsc mutations are common in testicular seminomas. Am J Acad Sci U S A 1989;86:7280–4. Res Tech 1999;47:344–60. Pathol 2004;164:305–13. 21. Ward SM, Burns AJ, Torihashi S, Sanders KM. 35. Kissel H, Timokhina I, Hardy MP, et al. Point mutation 7. Kitamura Y, Hirota S, Nishida T. A loss-of-function Mutation of the proto-oncogene c-kit blocks develop- in kit receptor tyrosine kinase reveals essential roles for mutation of c-kit results in depletion of mast cells and ment of interstitial cells and electrical rhythmicity in kit signaling in spermatogenesis and oogenesis without interstitial cells of Cajal, while its gain-of-function murine intestine. J Physiol 1994;480:91–7. affecting other kit responses. EMBO J 2000;19:1312–26. mutation results in their oncogenesis. Mutat Res 2001; 22. Maeda H, Yamagata A, Nishikawa S, et al. Require- 36. Longley BJ, Reguera MJ, Ma Y. Classes of c-KIT 477:165–71. ment of c-kit for development of intestinal pacemaker activating mutations: proposed mechanisms of action 8. Longley BJ, Tyrrell L, Lu SZ, et al. Somatic c-KIT system. Development 1992;116:369–75. and implications for disease classification and therapy. activating mutation in urticaria pigmentosa and aggres- 23. Heinrich MC, Corless CL, Duensing A, et al. PDGFRA Leuk Res 2001;25:571–6. sive mastocytosis: establishment of clonality in a human activating mutations in gastrointestinal stromal tumors. 37. Ronnstrand L. Signal transduction via the stem cell mast cell neoplasm. Nat Genet 1996;12:312–4. Science 2003;299:708–10. factor receptor/c-Kit. Cell Mol Life Sci 2004;61:2535–48. 9. Longley BJ Jr, Metcalfe DD, Tharp M, et al. Activating 24. Beghini A, Tibiletti MG, Roversi G, et al. Germline 38. Corless CL, McGreevey L, Haley A, Town A, Heinrich and dominant inactivating c-KIT catalytic domain mutation in the juxtamembrane domain of the kit gene MC. KIT mutations are common in incidental gastro- mutations in distinct clinical forms of human mastocy- in a family with gastrointestinal stromal tumors and intestinal stromal tumors one centimeter or less in size. tosis. Proc Natl Acad Sci U S A 1999;96:1609–14. urticaria pigmentosa. Cancer 2001;92:657–62. Am J Pathol 2002;160:1567–72. 10. Nishida T, Hirota S, Taniguchi M, et al. Familial 25. Hirota S, Okazaki T, Kitamura Y, O’Brien P, Kapusta L, 39. Heinrich MC, Blanke CD, Druker BJ, Corless CL. gastrointestinal stromal tumours with germline muta- Dardick I. Cause of familial and multiple gastrointes- Inhibition of KIT tyrosine kinase activity: a novel tion of the KIT gene. Nat Genet 1998;19:323–4. tinal autonomic nerve tumors with hyperplasia of molecular approach to the treatment of KIT-positive 11. Tian Q, Frierson HF Jr, Krystal GW, Moskaluk CA. interstitial cells of Cajal is germline mutation of the malignancies. J Clin Oncol 2002;20:1692–703. Activating c-kit gene mutations in human germ cell c-kit gene. Am J Surg Pathol 2000;24:326–7. 40. Tuveson DA, Willis NA, Jacks T, et al. STI571 tumors. Am J Pathol 1999;154:1643–7. 26. Hirota S, Nishida T, Isozaki K, et al. Familial inactivation of the gastrointestinal stromal tumor c- 12. Heinrich MC, Corless CL, Demetri GD, et al. Kinase gastrointestinal stromal tumors associated with dys- KIT oncoprotein: biological and clinical implications. mutations and imatinib response in patients with phagia and novel type germline mutation of KIT gene. Oncogene 2001;20:5054–8. metastatic gastrointestinal stromal tumor. J Clin Oncol Gastroenterology 2002;122:1493–9. 41. Joensuu H, Roberts PJ, Sarlomo-Rikala M, et al. Effect 2003;21:4342–9. 27. Isozaki K, Terris B, Belghiti J, Schiffmann S, Hirota of the tyrosine kinase inhibitor STI571 in a patient with 13. Huizinga JD, Thuneberg L, Kluppel M, Malysz J, S, Vanderwinden JM. Germline-activating mutation in a metastatic gastrointestinal stromal tumor. N Engl J Mikkelsen HB, Bernstein A. W/kit gene required for the kinase domain of KIT gene in familial gas- Med 2001;344:1052–6. interstitial cells of Cajal and for intestinal pacemaker trointestinal stromal tumors. Am J Pathol 2000;157: 42. Chen LL, Trent JC, Wu EF, et al. A missense mutation activity. Nature 1995;373:347–9. 1581–5. in KIT kinase domain 1 correlates with imatinib 14. Isozaki K, Hirota S, Nakama A, et al. Disturbed 28. Maeyama H, Hidaka E, Ota H, et al. Familial resistance in gastrointestinal stromal tumors. Cancer intestinal movement, bile reflux to the stomach, and gastrointestinal stromal tumor with hyperpigmentation: Res 2004;64:5913–9. deficiency of c-kit-expressing cells in Ws/Ws mutant association with a germline mutation of the c-kit gene. 43. Tamborini E, Bonadiman L, Greco A, et al. A new rats. Gastroenterology 1995;109:456–64. Gastroenterology 2001;120:210–5. mutation in the KIT ATP pocket causes acquired 15. Kitamura Y, Go S. Decreased production of mast cells 29. Sommer G, Agosti V, Ehlers I, et al. Gastrointestinal resistance to imatinib in a gastrointestinal stromal in S1/S1d anemic mice. Blood 1979;53:492–7. stromal tumors in a mouse model by targeted mutation tumor patient. Gastroenterology 2004;127:294–9. www.aacrjournals.org 6639 Cancer Res 2005; 65: (15). August 1, 2005

Brian P. Rubin, Cristina R. Antonescu, James P. Scott-Browne, et al.

Cancer Res 2005;65:6631-6639.

Updated version Access the most recent version of this article at: http://cancerres.aacrjournals.org/content/65/15/6631

Cited articles This article cites 42 articles, 12 of which you can access for free at: http://cancerres.aacrjournals.org/content/65/15/6631.full#ref-list-1

Citing articles This article has been cited by 12 HighWire-hosted articles. Access the articles at: http://cancerres.aacrjournals.org/content/65/15/6631.full#related-urls

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://cancerres.aacrjournals.org/content/65/15/6631. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.