Main

As defined in WHO's classification, aggressive systemic mastocytosis (ASM) is a subvariant of systemic mastocytosis which is characterized by abnormal growth and accumulation of mast cells in one or more organs with end-organ impairment.1 In most patients with ASM, conventional therapy is not effective and their prognosis is extremely poor.2 Recent advances in understanding the biology of this tumor using gene analysis revealed that the majority of adult mast cell neoplasms including ASM had mutations in proto-oncogene c-kit.3

The c-kit gene encodes a type III receptor tyrosine kinase (TK) KIT,4 the ligand of which is stem cell factor (SCF).5 KIT consists of an extracellular domain with five immunoglobulin-like repeats, a transmembrane domain, a juxtamembrane domain and a TK domain split into two portions, the TK I domain and the TK II domain. The binding of SCF to KIT promotes dimerization and phosphorylation of the receptor. Then diverse intracellular responses are induced through various signal transduction pathways such as RAS/MAPK, PI3K/Akt and the Jak/STAT systems. The SCF-KIT interaction is considered to be indispensable for normal regulation of differentiation and proliferation of hematopoietic progenitors, especially erythroid lineages, melanocytes, germ cells, mast cells and interstitial cells of Cajal.5, 6 Thus, the loss-of-function mutations of the c-kit gene result in a deficiency of those five cell types. On the other hand, gain-of-function mutations of the c-kit gene promote constitutive phosphorylation of KIT and consequent activation of downstream molecules without the requirement of SCF-KIT interaction. These mutations are considered to be a cause of mast cell neoplasms and gastrointestinal stromal tumors (GISTs).3, 7, 8, 9, 10 In mast cell neoplasms, most of the mutations are specifically confined to the codon 816 at exon 17 encoding TK II domain.7, 8, 9 Although the two gain-of-function mutations of the c-kit gene, Val816Asp and Val560Gly, were first found in the mast cell leukemia cell line (HMC-1 cells), the Val560Gly mutations have been only rarely observed in the cases with indolent systemic mastocytosis.11 A rare case of systemic mastocytosis with germline transmembrane domain mutation (Phe522Cys) has also been reported.12 On the other hand, most of the mutations are observed at exon 11 encoding juxtamembrane domain in GISTs.10 The minority of GISTs have the mutations of the c-kit gene at exon 9 encoding extracellular domain, at exon 13 encoding TK I domain, or at exon 17.13, 14, 15

Imatinib has developed as an agent to inhibit constitutive activation of BCR-ABL, which is a cause of chronic myelogenous leukemia. In vitro studies revealed that it effectively inhibited not only wild-type KIT but also juxtamembrane-type mutant KIT, but not TK II-type mutant KIT, especially Asp816Val or Asp816Tyr observed in mast cell neoplasms.16, 17 And the drug is now being successfully used for patients with metastatic or unresectable GISTs18 but not commonly for patients with mast cell neoplasms.

We encountered an autopsy case of ASM with a novel c-kit gene mutation at the juxtamembrane domain. The mutation was present at codon 559 and resulted in amino-acid substitution Val to Ile (Val559Ile). The point mutation at codon 559 has been reported in GISTs, but the substituted amino-acid type is usually Asp (Val559Asp), Ala (Val559Ala) or Gly (Val559Gly).19, 20, 21, 22 To our knowledge, only one case with the Val559Ile has been reported23 but biological characterization of the mutation has not been carried out. We consider that this is the first ASM case with juxtamembrane-type c-kit gene mutation. In this report, we characterized the Val559Ile KIT mutation by its transfection into human embryonic kidney (HEK) 293 cells and further compared the downstream signaling pathways of the Val559Ile with that of the Val559Asp commonly observed in GISTs.

Materials and methods

Patient and Clinical History

A 77-year-old Japanese man consulted a doctor with complaint of macroscopic hematuria, but examination of the urinary system revealed no causative lesions. Peripheral blood counts showed pancytopenia (white blood cells 3400/μL, hemoglobin 9.7 g/dl and platelet 17 000/μl) and he was admitted to the hospital. Computed tomography revealed ascites, an enlarged liver with irregular surface and splenomegaly with infarction. Infiltration of hematological malignant cells into liver and spleen was suspected, but a biopsy could not be carried out owing to a tendency to bleed. In spite of prednisolon administration and platelet transfusion, his condition got worse and the patient died without a final diagnosis only 3 months after the onset of hematuria. An autopsy was performed to elucidate the cause of his death.

Tissue Preparation and Immunohistochemistry

Tissues from various organs were obtained during autopsy, fixed with formalin and embedded in paraffin. Paraffin sections (3-μm thick) were used for conventional H&E staining and immunohistochemistry. Immunohistochemistry was performed using ENVISION+ KIT HRP (DAB) system (DAKO cytomation, Kyoto, Japan). Antibodies (Abs) used were listed in Table 1. Antigen unmasking was performed using appropriate retrieval techniques according to the manufacturer's instructions.

Table 1 List of antibodies used in the present study and the staining results
Full size table

Analysis of the c-kit Gene Mutation

Genomic DNA was extracted from formalin-fixed paraffin-embedded tissues using DEXPAT (TAKARA, Kyoto, Japan). Exons 9, 11, 13 and 17 of the c-kit gene were amplified by polymerase chain reaction (PCR) with Ampli Taq Gold (PerkinElmer, Norwalk, CT, USA) using the following oligonucleotide primer pairs; for exon 9, 5′-TTCCCTTTAGATGCTCTGCTTCT-3′/5′-CCTTTGTTGTTACCTTTAAATGC-3′, for exon 11, 5′-GAGTGCTCTAATGACTGACTGAGA-3′/5′-AAAGGTGACATGGAAAGCCC-3′, for exon 13, 5′-GCTTGACATCAGTTTGCCAG-3′/5′-AAAGGCAGCTTGGACACGGCTTTA-3′ and for exon 17, 5′-GTTTTCTTTTCTCCTCCAAC-3′/5′-TGCAGGACTGTCAAGCAAGCAGAG-3′. After purification of the PCR products with the QIAquick Gel Extraction Kit (QIAGEN, Valencia, CA, USA), direct sequencing was performed with ABI BigDye terminator ver.3.1 (Applied Biosystems, Foster City, CA, USA) and ABI Prism 3100-Avant Genetic Analyzer (Applied Biosystems). The study was performed under the authors' institutional guidelines and was approved by the institutional review boards.

Construction and Transfection of Mutated c-kit cDNA with Val559Ile

The whole coding region of human wild-type c-kit cDNA was subcloned into expression vector pcDNA3.1 Zeo (+) (Invitrogen, San Diego, CA, USA). To generate the human c-kit cDNA containing the Val559Ile mutation, site-directed mutagenesis was performed. Then SnaBI–MroI fragment (nucleotide 1140–2683) of the human wild-type c-kit cDNA in the expression vector was replaced by the corresponding fragment of the mutated c-kit cDNA with Val559Ile. The expression vector with the mutation was transfected into HEK293 cells using Mirus Trans IT-293 transfection reagent (Madison, WI, USA). Wild-type c-kit cDNA in expression vector pcDNA3.1 Zeo (+) and mutated c-kit cDNA with the Val559Asp in expression vector pEF-BOS were transfected in the same way.

Immunoprecipitation and Immunoblotting of KIT

The procedures of cell lysis, sodium dodecyl sulfate-polyacrilamide gel electrophoresis and immunoblotting were performed as described.24 After the depletion of serum for 6 h, the HEK293 cells expressing wild-type KIT, mutant KIT with Val559Asp or mutant KIT with Val559Ile were collected and incubated with or without 100 ng/ml of human recombinant SCF (PeproTec EC, London, UK) at 37°C for 10 min. Cells were collected and lysed using CelLytic-M lysis buffer (Sigma-Aldrich, St Louis, MO, USA) with the addition of complete protease inhibitor cocktail tablets (Roche, Mannheim, Germany), phosphatase inhibitor cocktail (Sigma-Aldrich) and 1 μM Na3VO4. The extracts were precipitated with murine anti-human KIT monoclonal Ab (YB5.B8, Pharmingen, San Diego, CA, USA) and protein G Sepharose beads (Amersham Bioscience Corp, Piscataway, NJ, USA). Immunoblotting was performed with mouse anti-phosphotyrosine monoclonal Ab (4G10, Upstate Biotechnology, Lake Placid, NY, USA) and rabbit anti-human KIT polyclonal Ab (A4502; DAKO Cytomation). The binding of the Ab was detected by enhanced chemiluminescence Western Blotting Detection System (Amersham Bioscience Corp).

Effect of Imatinib on Phosphorylation of KIT and Its Downstream Signaling Molecules

After serum deprivation for 6 h, the HEK293 cells expressing mutant KIT with the Val559Asp or Val559Ile were cultured at various concentrations of imatinib (0, 0.01, 0.1, 1 and 10 μM) (Novartis Pharma, Basel, Switzerland) for 90 min. Then cells were collected and lysed as described.24 Phosphorylation of KIT was detected by anti-phosphotyrosine monoclonal Ab (4G10) after immunoprecipitation with KIT monoclonal Ab (YB5.B8). To examine the inhibitory effect of imatinib on phosphorylation of KIT downstream signaling molecules, the cells expressing mutant KIT with Val559Asp or Val559Ile were cultured at concentrations of 0 and 10 μM of imatinib for 90 min after 6 h serum deprivation. Phospho-specific MAPK Ab (9109, Cell Signaling Technology, Beverly, MA, USA), phospho-Akt Ab (9271, Cell Signaling Technology), phospho-STAT1 Ab (9171, Cell Signaling Technology) and phospho-STAT5 Ab (9351, Cell Signaling Technology) were used to detect phosphorylation of MAPK, Akt, STAT1 and STAT5. Immunoblotting of phospho-STAT1 and phospho-STAT5 was carried out after immunoprecipitation using total STAT1 Ab (9172, Cell Signaling Technology) and total STAT5 Ab (9352, Cell Signaling Technology). Each membrane was subsequently reprobed with total MAPK Ab (9102, Cell Signaling Technology), pan-Akt Ab (9272, Cell Signaling Technology), total STAT1 Ab (9172) or total STAT5 Ab (9352).

Results

Macroscopic, Histological and Immunohistochemical Analysis

At autopsy, hepatosplenomegaly (liver 3180 g, spleen 480 g) was noted. There were no space-occupying lesions in the liver or spleen, but sporadic infarcted lesions were observed in the spleen. Microscopically, short-spindled tumor cells diffusely infiltrated the spleen (Figure 1a). The tumor cells were also found in the hepatic portal area and renal interstitium. In the bone marrow, paratrabecular infiltration of the tumor cells was shown. The tumor cells exhibited metachromasia by toluidine blue staining (Figure 1b) and were immunohistochemically positive for CD25 (Figure 1c), KIT (Figure 1d), mast cell tryptase (Figure 1e) and MITF (Figure 1f). Immunohistochemical details including other markers are summarized in Table 1. These results confirmed the diagnosis of ASM.

Figure 1
figure 1

Histology and immunohistochemical profile of tumor cells. (a) Short spindled atypical tumor cells diffusely infiltrated in splenic tissue, hematoxylin and eosin staining. (b) Some tumor cells showed metachromasia in toluidine blue staining. (c–f) Immunohistochemistry for CD25 (c), KIT (d), mast cell tryptase (e) and MITF (f) revealed that the tumor cells are positive for all of these markers, suggesting that the tumor cells have differentiation features for mast cells. Staining of CD25 and KIT mainly shows a membranous pattern. On the other hand, staining pattern of mast cell tryptase is cytoplasmic, and that of MITF is nuclear.

Full size image

Identification of c-kit Gene Mutation

To examine whether the c-kit gene mutation is observed in tumor cells of ASM, genomic DNA was extracted from paraffin-embedded splenic tissues, because tumor cells diffusely infiltrated the spleen and contamination of non-neoplastic cells was considered to be minimum. As most c-kit gene mutations are observed at exon 9, exon 11, exon 13 or exon 17 in human neoplasms including mast cell neoplasms and GISTs, DNA fragments of those exons were amplified by PCR. Direct sequencing of the products revealed point mutation at codon 559 (GTT to ATT) in the exon 11 encoding juxtamembrane domain (Figure 2a), resulting in amino-acid substitution from Val to Ile. No mutations were detected at exons 9, 13 and 17. Analysis of the genomic DNA extracted from paraffin-embedded prostatic tissue without tumor cell infiltration showed no mutations in exon 11 (Figure 2b) in addition to exons 9, 13 and 17.

Figure 2
figure 2

Detection of c-kit gene mutation and phosphorylation status of mutant KIT. (a) Direct sequencing of PCR-amplified products of genomic DNA derived from tumor cells of spleen revealed a point mutation at codon 559 (GTT to ATT). (b) Genomic DNA derived from prostatic tissue without tumor cells did not show the point mutation. (c) Immunoblotting with anti-phosphotyrosine Ab (4G10) revealed ligand-independent autophosphorylation of mutant KIT with the Val559Asp and mutant KIT with Val559Ile. Wild-type c-kit cDNA, mutated c-kit cDNA with Val559Asp or that with Val559Ile was transfected into HEK293 cells, and the cells were incubated with or without SCF after serum starvation for 6 h. Whole lysate was immunoprecipitated with anti-human KIT Ab. After immunoblotting with 4G10, reprobing with anti-human KIT Ab was carried out.

Full size image

Constitutive Activation of Mutant KIT with Val559Ile

To investigate the effect of Val to Ile substitution at codon 559, the expression vector containing the mutated c-kit cDNA was transfected into HEK293 cells. The expression vector containing the wild-type c-kit cDNA or the mutated c-kit cDNA with Val559Asp observed in GISTs was also transfected into the cells, and the phosphorylation status of KIT was compared. As shown in Figure 2c, autophosphorylation of the wild-type KIT without SCF stimulation was very weak but became significant in response to SCF addition. The mutant KIT with Val559Asp was strongly autophosphorylated even in the absence of SCF (Figure 2c) as reported previously.10 A similar pattern was observed in the mutant KIT with Val559Ile (Figure 2c).

Effect of Imatinib on Autophosphorylation of Mutant KIT with Val559Ile

To examine the effect of imatinib on mutant KIT with Val559Ile, the HEK293 cells transfected with the mutated c-kit cDNA were cultured at various concentrations of the reagent (0, 0.01, 0.1, 1 and 10 μM). As a comparison, the HEK293 cells transfected with the mutated c-kit cDNA with Val559Asp were also examined. Autophosphorylation of mutant KIT with Val559Asp was markedly inhibited at a concentration of 1 μM of imatinib, but that of mutant KIT with Val559Ile was not inhibited, even at a concentration of 10 μM (Figure 3).

Figure 3
figure 3

Effect of imatinib on autophosphorylation of mutant KIT. (a) Autophosphorylation of mutant KIT was examined after 90 min incubation at imatinib concentrations of 0, 0.01, 0.1, 1 and 10 μM in HEK293 cells expressing Val559Asp or Val559Ile. Autophosphorylation of mutant KIT with Val559Asp was almost completely inhibited at a concentration of 1 μM of imatinib, but that with Val559Ile was not even at an imatinib concentration of 10 μM. Whole lysate was immunoprecipitated with anti-human KIT Ab. Immunoblotting with 4G10 was carried out with subsequent reprobing with anti-human KIT Ab.

Full size image

Effect of Imatinib on Phosphorylation of KIT Downstream Signaling Molecules

To clarify the relationship between activation of KIT and that of its downstream signaling molecules, the effect of imatinib on phosphorylation of MAPK, Akt, STAT1 and STAT5 was examined in original HEK293 cells and those transfected with mutated c-kit cDNA with Val559Ile or Val559Asp. When phosphorylation of mutant KIT with Val559Asp was inhibited by imatinib at a concentration of 10 μM (Figure 4a), that of MAPK and STAT5 was also inhibited significantly (Figure 4b, c). In contrast, not only KIT but also MAPK and STAT5 remained phosphorylated in mutant KIT with Val559Ile even after the addition of imatinib at a concentration of 10 μM (Figure 4a–c). No apparent phosphorylation of STAT1 was observed in any type of HEK293 cells even without imatinib treatment (Figure 4d). Akt was strongly phosphorylated even in the original HEK293 cells, and the inhibitory effect of imatinib was not apparent in any of the cell types examined (Figure 4e).

Figure 4
figure 4

Effect of imatinib on phosphorylation of KIT downstream signaling molecules. (a–e) Autophosphorylation of KIT (a), phosphorylation of p44/p42 MAPK (b), STAT5 (c), STAT1 (d) and Akt (e) was examined at concentrations of 0 and 10 μM of imatinib in HEK293 cells expressing Val559Asp or Val559Ile. HEK293 cells without transfection were similarly treated as a control. When phosphorylation of mutant KIT with Val559Asp was inhibited by imatinib at a concentration of 10 μM (a), that of MAPK (b) and STAT5 (c) was significantly inhibited. In contrast, not only KIT (a) but also MAPK (b) and STAT5 (c) remained phosphorylated in mutant KIT with Val559Ile even after the addition of imatinib at a concentration of 10 μM. Apparent phosphorylation of STAT1 was not observed in all types of HEK293 cells even without imatinib treatment (d). Akt was strongly phosphorylated even in the original HEK293 cells, and the inhibitory effect of imatinib was not apparent in any type of HEK293 cells examined (e). The lysate was immunoblotted with anti-phospho-KIT, anti-phospho-p44/p42 MAPK, anti-phospho-STAT5, anti-phospho-STAT1 or anti-phospho-Akt Ab. Each membrane was reprobed with anti KIT, anti-p44/42 MAPK, anti-STAT5, anti-STAT1 or anti-Akt Ab.

Full size image

Discussion

We found a case of ASM with a juxtamembrane-type c-kit gene mutation of Val559Ile. Although HMC-1 cells isolated from a patient with mast cell leukemia have a juxtamembrane-type c-kit gene mutation of Val560Gly which is sensitive to imatinib, adult mast cell neoplasms usually have a TK II-type c-kit gene mutation which is resistant to imatinib.7, 8, 9 We believe that this is the first ASM case reported with a juxtamembrane-type c-kit gene mutation. The juxtamembrane-type c-kit gene mutation is the most common type in GISTs, and simple amino-acid substitution (point mutation) of the codon 559 is often observed. Mutations of Val559Asp, Val559Ala or Val559Gly are usual types, and only one GIST case with Val559Ile has been reported to our knowledge.19, 20, 21, 22, 23 The juxtamembrane-type mutation observed in the present ASM case is unique. We characterized the mutation using HEK293 cells transfected with the mutated c-kit cDNA, and compared its properties to that of the GIST-type mutation, Val559Asp. The mutant KIT with Val559Ile was constitutively phosphorylated as well as that with Val559Asp, demonstrating that the c-kit gene mutation Val559Ile results in gain-of-function. However, the inhibitory effect of imatinib on autophosphorylation of the mutant KIT was different between the mutant KIT with Val559Ile and that with Val559Asp.

Imatinib almost completely inhibited the autophosphorylation of mutant KIT with Val559Asp at a concentration of 1 μM, but the autophosphorylation of mutant KIT with Val559Ile was not inhibited even at an imatinib concentration of 10 μM. The result indicates that the mutant KIT with Val559Ile was more resistant to imatinib than that with Val559Asp. It is well known that sensitivity to imatinib depends on the location of the c-kit gene mutation. Imatinib effectively inhibits KIT with juxtamembrane-type mutation, but fails to inhibit KIT with some TK II-type mutations, such as Val816Asp in mastocytosis.16, 17 The phenomenon can be explained by the concept that some TK II-type activation loop mutations induce stabilization of the activation loop in an active conformation and/or structural alteration at the imatinib-binding site of KIT, resulting in a decreased affinity for imatinib. Since Ma et al18 reported that different TK II-type activation loop mutations at codon 816, such as Asp816Val, Asp816Tyr and Asp816Phe showed different sensitivities to imatinib, different amino-acid substitutions, even at the same codon, might cause the different imatinib sensitivities. In general, the juxtamembrane-type KIT mutations do not appear to show imatinib resistance. However, the structural change of the juxtamembrane domain clearly affects the structure of the kinase domains. Therefore, different amino-acid substitutions at the juxtamembrane non-activation loop region such as Val559Ile and Val559Asp might also cause different structural alteration and lead to different imatinib sensitivities. As the mechanism of the different inhibitory effects of imatinib on Val559Ile and Val559Asp mutations remains unclear, further examinations might be required to elucidate it, including substitutional experiment of codon 559 to various amino acids and structural analyses of those mutant KITs.

In the present study, we examined the effect of imatinib on downstream signaling molecules of KIT, using cells expressing mutant KIT with Val559Ile and mutant KIT with Val559Asp. Both transfectants showed apparent phosphorylation of MAPK and STAT5 but not STAT1, and phosphorylation of MAPK and STAT5 was inhibited by imatinib at a concentration of 10 μM in cells expressing mutant KIT with Val559Asp but not in those with Val559Ile. As the inhibitory pattern by imatinib was similar to that of autophosphorylation of mutant KIT, KIT activation by the mutations appears to be closely related to the activation of MAPK and STAT5. The result suggested that both Val559Ile and Val559Asp mutations might activate Ras/MAPK and Jak/STAT signal transduction pathways.

In the present study, phosphorylation of STAT1 was not detected in cells expressing mutant KIT with Val559Ile nor those with Val559Asp. Ning et al25 reported that the STAT1 pathway plays an important role in KIT with TK II-type mutation. The STAT1 pathway might play an important role in TK II-type c-kit gene mutation but not in juxtamembrane-type c-kit gene mutation.

A previous report showed that phosphorylation of Akt induced by SCF stimulation was inhibited by imatinib in a dose-dependent manner in M-07e cells expressing wild-type KIT.26 We also examined the phosphorylation status of Akt in the presence or absence of imatinib using original HEK293 cells and those expressing mutant KIT with Val559Ile or Val559Asp. However, Akt was apparently phosphorylated not only in cells expressing mutnat KIT with Val559Ile or Val559Asp but also in original HEK293 cells without KIT expression, and imatinib did not inhibit its phosphorylation in any types of cells used. As the PI3K/Akt signaling pathway was strongly activated in the original HEK293 cells even under the basal condition, it was not possible to assess the effect of c-kit gene mutations on phosphorylation of Akt by using HEK293 cells. The relationship between autophosphorylation of mutant KIT and phosphorylation of Akt should be investigated by transfection into other cells in which Akt is not strongly phosphorylated under the basal condition. Which down-stream signaling pathways are activated by various types of KIT mutations remains to be completely clarified.

Taken together, the Val559Ile mutation observed in ASM was proven to be of gain-of-function and imatinib-resistant. Recently, new tyrosine kinase inhibitors such as PKC412, dasatinib (BMS-354825), MLN518, AP23464 and AP23848 have been reported to exhibit inhibitory effect on imatinib-resistant TK II-type activation loop mutations including Asp816Val with the STAT3 inactivation.27, 28, 29, 30 There is a possibility that activation of KIT with Val559Ile mutation resistant to imatinib might be also effectively inhibited by those agents. However, it may not be inhibited because the mechanism of resistance might be different between the activation loop mutation and non-activation loop mutation. The effect of those drugs on the Val559Ile mutation remain to be examined.