Fax: (011) 39-02-23902877
Evidence for activation of KIT, PDGFRα, and PDGFRβ receptors in the Ewing sarcoma family of tumors
Article first published online: 6 MAR 2007
Copyright © 2007 American Cancer Society
Volume 109, Issue 8, pages 1638–1645, 15 April 2007
How to Cite
Bozzi, F., Tamborini, E., Negri, T., Pastore, E., Ferrari, A., Luksch, R., Casanova, M., Pierotti, M. A., Bellani, F. F. and Pilotti, S. (2007), Evidence for activation of KIT, PDGFRα, and PDGFRβ receptors in the Ewing sarcoma family of tumors. Cancer, 109: 1638–1645. doi: 10.1002/cncr.22587
- Issue published online: 4 APR 2007
- Article first published online: 6 MAR 2007
- Manuscript Accepted: 9 JAN 2007
- Manuscript Revised: 4 JAN 2007
- Manuscript Received: 30 OCT 2006
- Associazione Italiana per la Ricerca sul Cancro (AIRC 2004)
- Italian Ministry of Health (Ricerca finalizzata 2004)
- RTK activation profile;
- Ewing sarcoma;
- PDGFRα PDGFR β
The Ewing sarcoma family of tumors (ESFT) is one of the most common malignant neoplasms of children and adolescents, characterized by nonrandom translocations involving the Ewing sarcoma (EWS) gene. Over the years the adoption of intensive multimodality treatment approaches has led to a gradual improvement in the survival of patients with ESFT. The prognosis is still unsatisfactory for high-risk patients, however, and novel therapeutic approaches are desirable. The aim of the study was to investigate the expression/activation of KIT, PDGFRα, and PDGFRβ receptor tyrosine kinases (RTKs) as potential therapeutic targets in ESFT.
RNA and proteins were extracted from 20 frozen ESFT specimens to ascertain the state activation of KIT, PDGFRα, and PDGFRβ.
No mutations were found, whereas the cognate ligands were detected in all cases by polymerase chain reaction (PCR). The expression and activation of KIT, PDGFRα, and PDGFRβ were confirmed by quantitative PCR, immunohistochemistry, and immunoprecipitation and/or Western blot analysis. In particular, when compared with a protein pool obtained from normal adult tissues, PDGFRβ showed a greater protein expression and/or a stronger phosphorylation signal.
The results are consistent with an autocrine/paracrine loop activation of the KIT, PDGFRα, and PDGFRβ receptors and suggest a rationale for the use of RTK inhibitors, either alone or in combination with chemotherapy. Cancer 2007. © 2007 American Cancer Society.
Ewing sarcoma/peripheral primitive neuroectodermal tumor (ES/pPNET) and malignant small cell tumor of the thoracopulmonary region (Askin tumor) are currently regarded as manifestations of a single neoplastic entity sharing a common immunophenotype and molecular features, but differing in primary tumor site.
These tumors are now grouped under the general term of the Ewing sarcoma family of tumors (ESFT). They are high-grade malignancies characterized by local aggressiveness and a strong propensity to develop distant metastases. These tumors can originate anywhere in the body, in bone as well as in soft tissues (including the subcutis).1 The molecular hallmark of ESFT is the presence of nonrandom translocations of the EWS gene. The most frequent translocation (found in 90% of cases) is the t(11;22) EWS-FLI1, which fuses the N-terminal domain of EWS with the DNA-binding domain of FLI1. The second most common nonrandom translocation is the t(21;22) EWS-ERG, which occurs in approximately 5% to 10% of cases. Other translocations seen in fewer than 1% of cases induce fusion of the EWS gene with the ETV1, E1AF, FEV, and ZSG genes.2 The histogenic origin of ESFT remains unknown in the absence of any corresponding ancestral tissue.
Multimodality therapeutic approaches including surgery, intensive chemotherapy, and radiotherapy have led to significant improvements in the outcome of ESFT patients in the last 2 or 3 decades, but about a third of patients with localized disease may still develop recurrences. New therapeutic approaches are therefore much needed, and ESFT would represent the ideal target for the development of new molecular therapies.
In recent years, pathological RTK activation via various mechanisms has been described at the preclinical and clinical level in a number of malignancies. Among the RTKs known to increase cell proliferation and survival, and to be inhibited by specific treatment, KIT and platelet-derived growth factor receptors (PDGFRs) have been found involved in ESFT.
KIT is characterized by 5 extracellular immunoglobulin-like domains, a single transmembrane domain, and an intracellular tyrosine kinase domain split into 2 parts. The ligand for KIT is called stem cell factor (SCF). Alternative splicing of the SCF transcript results in the inclusion or exclusion of exon 6. The spliced form lacking the cleavage site remains associated with the cell surface (membrane-associated), whereas the other (soluble) form is rapidly released. The pathological autocrine/paracrine loop sustained by a constitutive KIT dimerization was first identified in ESFT cell lines3, 4 and primary tumor specimens.5 Imatinib was subsequently proposed in combination with standard therapy with a view to inhibiting KIT activation.4, 5 The variable level of KIT expression in ESFT6 might suggest, however, that other RTKs are involved in tumor growth and survival and, consistent with this theory, a phosphorylated (activated) PDGFRβ (and its ligand PDFGB) was recently identified in ESFT cell lines.7
Two different subunits of PDGFR have been described, ie, PDGFRα and PDGFRβ. There are 2 well-characterized PDGFR ligands, PDGFA and PDGFB, and PDGFC and PDGFD have also recently been discovered. PDGFRα binds PDGFA, -B, and -C, but PDGFRβ binds only PDGFB and -D.8
To extend these preclinical and translational data, we assessed the activation state (phosphorylation) of c-KIT, PDGFRα, and PDGFRβ in 20 molecularly characterized frozen ESFT samples collected at our institution (1990–2004). We also sought activating mutations in KIT-imatinib well responding exon 11, in the juxtamembrane regions and in the tyrosine kinase domains of both PDGFRα and PDGFRβ. Our results confirm5 the presence of a phosphorylated KIT-SCF autocrine/paracrine loop and show that PDGFRα-PDGFA and PDGFRβ-PDGFB may also be involved in ESFT growth. Both KIT and PDGFR (α and β) are potential molecular targets whose inhibition could improve the effectiveness of standard chemotherapy, as we recently observed in 2 sacral chordoma patients.
MATERIALS AND METHODS
Patients and Materials
The analysis involved 20 ESFT patients treated at the Pediatric Oncology Unit of the Istituto Nazionale Tumori, Milan, Italy, whose surgical specimens were available. Patients were a median 14 years of age (range, 4–17); 13 were males and 7 were females. All the specimens came from the primary tumor (9 arising in bone, 11 in soft tissues, see Table 1); 4 had been obtained before treatment, whereas 16 of the patients had undergone standard chemotherapy and/or radiotherapy before surgery (Table 1). Written informed consent was obtained from all patients.
|No.||Primary sites||Fusion transcripts||Sequences||Ligands||IP/WB||IHC|
|Exon 11||Exons 10 12 14 16 18||Exons 10–13 13–15 16–20|
|C1||B (pelvis)||E/F 7/5||wt||wt wt wt wt wt||wt wt wt||LS||+||+||++||pp||+||P||++||pp||−||+||+|
|C2||B (chest wall)||FISH+||wt||wt wt wt wt wt||wt wt wt||LS||+||+||+||+||+|
|C3||ST (pelvis)||E/F 7/6||wt||wt wt wt wt wt||wt wt wt||LS||+||+||++||pp||++||pp||++||pp||+||+||+|
|C4*||B (skull)||E/F 7/5||wt||wt wt wt wt wt||wt wt wt||LS||+||+||++||pp||++||Pp||++||pp||+||+||+|
|C5||ST (chest wall)||E/F 7/5||wt||wt wt wt wt wt||wt wt wt||LS||+||+||++||P||++||pp||+||P||+||+||+|
|C6||ST (chest wall)||E/F 7/5||wt||wt wt wt wt wt||wt wt wt||LS||+||+||+||pp||++||pp||++||pp||−||+||+|
|C7||ST (chest wall)||E/F 7/5||wt||wt wt wt wt wt||wt wt wt||LS||+||+||−||+||+|
|C8||ST (pelvis)||E/F 7/6||wt||wt wt wt wt wt||wt wt wt||LS||+||+||++||pp||+||p||++||pp||+||+||−|
|C9||B (pelvis)||E/F 7/6||wt||wt wt wt wt wt||wt wt wt||LS||+||+||++||pp||+||p||++||pp||+||+||+|
|C10||ST (pelvis)||E/F 7/5||wt||wt wt wt wt wt||wt wt wt||LS||+||+||−||+||+|
|C11*||B (chest wall)||E/E 7/6||wt||wt wt wt wt wt||wt wt wt||LS||+||+||+||+||+|
|C12*||ST (pelvis)||E/F 7/6||wt||wt wt wt wt wt||wt wt wt||LS||+||+||++||P||++||pp||+||pp||−||+||+|
|C13||ST (retroperitoneum)||FISH+||wt||wt wt wt wt wt||wt wt wt||LS||+||+||+||+||+|
|C14||B (humerus)||E/F 7/6||wt||wt wt wt wt wt||wt wt wt||LS||+||+||+||+||+|
|C15||ST(retroperitoneum)||E/F 7/9||wt||wt wt wt wt wt||wt wt wt||LS||+||+||++||p||++||pp||++||pp||−||+||−|
|C16||ST (femur)||E/F 7/5||wt||wt wt wt wt wt||wt wt wt||LS||+||+||++||p||+||P||+||P||+||+||+|
|C17||B (scapula)||E/F 7/8||wt||wt wt wt wt wt||wt wt wt||LS||+||+||Nd||Nd||Nd|
|C18||B (scapula)||E/E 7/6||wt||wt wt wt wt wt||wt wt wt||LS||+||+||−||+||−|
|C19||B (pelvis)||E/F 7/6||wt||wt wt wt wt wt||wt wt wt||LS||+||+||−||+||−|
|C20*||ST (pelvis)||E/E 7/9||wt||wt wt wt wt wt||wt wt wt||LS||+||+||−||+||+|
All the biochemical and molecular analyses were performed on frozen sections carefully dissected under the microscope to avoid contamination by normal or necrotic tissues. The analyses were subsequently complemented with immunophenotyping performed on corresponding fixed materials.
RNA Extraction, Reverse-Transcription, and cDNA Synthesis
Total RNA was extracted from freshly frozen tissue and reverse-transcribed. All the samples were tested for cDNA integrity and DNA contamination by amplifying the β-actin and HPRT housekeeping genes, respectively.
PCR Detection of Fusion Transcripts by PCR and FISH
For the detection of fusion transcripts, the oligonucleotides used were EWS 22.3 and FLI 1 11.3 (for EWS/FLI 1 fusion) and EWS 22.8 and ERG 11.9 (EWS/ERG fusion).9 EWS/FLI 1 amplification consisted of 1 cycle at 94°C for 8 minutes followed by 35 cycles at 94°C for 30 seconds, then at 64°C for 30 seconds and at 72°C for 30 seconds, then a final cycle at 72°C for 7 minutes. EWS/ERG amplification consisted of 1 cycle at 94°C for 8 minutes, then 35 cycles at 94°C for 30 seconds, then at 68°C for 30 seconds and at 72°C for 30 seconds, ending with 1 cycle at 72°C for 7 minutes.
PCR Detection of KIT, PDGFR α, and PDGFR β, Sequencing Analysis, and Ligand Assessment
Exon 11 of human c-Kit (Accession No. X06182) cDNA was amplified as explained elsewhere.10 Human PDGFRα (Accession No. XM01186) and PDGFRβ (Accession No. BC032224) cDNA were amplified from exons 10–18 and 10–20 using previously published primers and conditions.10
Human SCF, PDGFA and PDGFB were amplified as described elsewhere.11 All receptor sequence reactions were carried out using an automated sequencing system (377 DNA sequencer, ABI PRISM PE; Applied Biosystems, Foster City, CA) according to standard protocols.
Relative Quantification of TRK Receptor Expression
c-Kit, PDGFRα, and PDGFRβ cDNAs were relatively quantified by real-time quantitative polymerase chain reaction (PCR) (ABIPRISM 5700 PCR Sequence Detection Systems, Applied Biosystems) using a TaqMan-based analysis. Each reaction contained 1× TaqMan Universal Master Mix (Applied Biosystems), 1 μL of template c-DNA, 0.9 μM of each forward and reverse primer, and 0.2 μM of the fluorogenic TaqMan probes in a total volume of 25 μL. Cycling was started with 2 minutes at 50°C and 10 minutes at 95°C, followed by 40 cycles of 15 seconds at 95°C and 1 minute at 60°C. All the experiments were performed in triplicate.
hRNaseP and GAPDH, detected by means of a commercially available kit (Applied Biosystems), were used as internal controls. To assess the amplification efficiency of receptors and controls we obtained standard curves using serial cDNA dilutions.
The 2−ΔΔCt method12 was used to calculate the relative changes in receptor gene expression determined by real-time quantitative PCR: ΔΔCt = [ΔCt (unknown sample) − ΔCt (calibrator sample)] = [CtGI (unknown sample) − CtGR (unknown sample)] − [CtGI (calibrator sample) − CtGR (calibrator sample)], where GI is the gene of interest, GR the reference gene, and the calibrator is the sample chosen to represent 1× expression of the gene of interest. The calibrator sample was a pool of normal mesenchymal tissues (muscles, vessels, adipose tissue, nerves).
Each amplification was performed using a positive control of the reaction represented: for c-Kit by 7 gastrointestinal stromal tumors (GISTs) previously characterized as overexpressing KIT protein in immunohistochemistry (IHC) and immunoprecipitation/Western blotting (IP/WB) experiments; for PDGFRα by a GIST carrying the D842V mutation in PDGFRA already shown to overexpress this receptor; and for PDGFRβ by a chondrosarcoma positive for this receptor.10
Protein extraction and IP/WB
The proteins were extracted, immunoprecipitated for KIT, PDGFRα, and PDGFRβ, and blotted as described elsewhere.10
Positive and negative controls
The 2N5A cell line (derived from the NIH3T3 cell line expressing the COL1A1-PDGFB fusion characterizing dermatofibrosarcoma protuberans, kindly provided by Dr. A. Greco) was used as the positive control for the PDGFRβ blots. The NIH3T3 cell line (American Type Culture Collection, Manassas, VA) was used for the PDGFRα protein expression/phosphorylation experiments. The KIT/Δ559 cell line overexpressing an activated KIT receptor was used in the KIT experiments. A pool of normal mesenchymal tissues (muscles, vessels, adipose tissue, nerves) was used as a negative control of receptor expression level.
A representative paraffin block of formalin-fixed tumoral tissue from 19 of the 20 patients was selected and phenotyped for KIT, PDGFRα, and PDGFRβ by immunoperoxidase analysis as explained in detail elsewhere.11 Briefly, the samples were immunoperoxidase-phenotyped using antibodies against PDGFRα (sc-338, Santa Cruz Biotechnology, Santa Cruz, CA; diluted 1:200), PDGFRβ (sc-339 Santa Cruz Biotechnology; diluted 1:100), and CD117 (A4502, Dako, Carpinteria CA; diluted 1:50 epitope 963–976) with heat-induced epitope retrieval. The slides were developed using 3,3′diaminobenzidine. Two GISTs, molecularly characterized for KIT and PDGFRα mutations, were used as positives controls for KIT and PDGFRα staining11 and a chordoma13 and dermatofibrosarcoma protuberans14 were used as positive controls for PDGFRβ reactivity.
ESFT Fusion Transcript Detection
By PCR, EWS/FLI t(11;22) was present in 15 cases and EWS/ERG t(11;22) in 3 cases. In the remaining 2 cases the EWS translocation was demonstrated by FISH (cases 2 and 13; Table 1).
c-Kit, PDGFRα, and PDGFRβ Sequencing Analysis
c-KIT, PDGFR α, and PDGFR β mRNA were expressed in all samples. No activating mutations were revealed by direct PCR product sequences in the tyrosine kinase domain of c-KIT (exon 11), in the juxtamembrane regions, or in the tyrosine kinase domains of either PDGFRα (exons 10–18) or PDGFRβ (exons 10–20). In PDGFRα sequences, 1 case had a silent mutation at base 2203 (G to A) corresponding to residue 603 and 1 at base 2866 (C to T) corresponding to residue 824. The majority of PDGFRβ sequences (11/20) showed the same silent mutation at base 2957 (A to G) corresponding to residue 867. The sequences of c-KIT exon 11 revealed no polymorphism.
SCF, PDGFA, and PDGFB Ligand Detection
The cognate ligands of the RTKs SCF, PDGFA, and PDGFB could be detected in all cases. In particular, SCF could be detected in both soluble (long) and membrane-associated (short) forms. The soluble form was expressed more than the short form.
Relative Quantification of c-KIT, PDGFRα, and PDGFRβ mRNA Expression
To estimate the relative amount of these RTK transcripts, an RNA pool composed of mesoderm-derived tissues was used as a calibrator. The relative level of c-KIT mRNA in ESFT samples showed a pronounced variability, ranging from 0.1 to 46 as much c-KIT mRNA as calibrator. The relative levels of the PDGFRα and PDGFRβ transcripts was between 0.2 to 100 and 1 to 40 as much PDGFRα and PDGFRβ mRNA as calibrator, respectively (Fig. 1).
Ten frozen ESFT samples were available for the biochemical experiments. Both forms of KIT (125 and 145 kDa) could be detected in all samples. Nine samples had a greater level of protein expression than the adult tissue proteins pool (cases 1, 3, 4, 5, 8, 9, 12, 15, and 16, identified by “++” in Table 1): 6 cases (cases 1, 3, 4, 6, 8 and 9, identified by “pp” in Table 1) had a higher phosphorylation signal than the same pool (Fig. 2). All ESFT samples (10/10) showed PDGFRα (Fig. 3) and PDGFRβ (Fig. 4) expression.
Compared with the protein extracted from a pool of healthy tissues, higher levels of PDGFRα and PDGFRβ expression were detected, respectively, in 6 of 10 (cases 3, 4, 5, 6, 12, and 15; Table 1) and 7 of 10 (cases 1, 3, 4, 6, 8, 9, 15; Table 1) ESFT-positive samples; all these cases had a stronger phosphorylation signal.
All samples showed cytoplasmic anti-PDGFRα reactivity, whereas 52% (10/19) and 79% (15/19) showed KIT and PDGFRβ immunoreactivity, respectively (Table 1).
Correlation Between Biochemical Analysis and Immunophenotyping
The present investigation on 20 cases of molecularly characterized ESFT provides evidence of a constitutive KIT, PDGFRα, and PDGFRβ expression/activation as part of the gene profile of this tumor. It is reasonable to speculate that this profile is not modified by the standard adjuvant chemotherapeutic treatments performed before surgery as observed in a previously reported ESFT patients.15
In the absence of mutations the majority of the cases showed biochemical evidence of activation of all the receptors, coupled with the expression of the respective cognate ligands—features consistent with the presence of an autocrine/paracrine loop.
Despite all frozen ESFT samples showing a variable increase in the expression of c-Kit mRNA (by comparison with a calibrator obtained from healthy tissues), we observed 10 of 19 (52%) cases of KIT expression by IHC, a finding otherwise in keeping with reported results.6 The discrepancy may be due to the greater sensitivity of the molecular method and the use of fixed material for IHC. More interestingly, when compared with the protein pool made from the calibrator the WB/IP experiments performed on 10 frozen samples showed an increment in and phosphorylation of KIT protein in 4 cases (cases 3, 4, 8, and 9; Table 1), which also showed both KIT immunoreactivity and higher mRNA levels with respect to the calibrator. Considering the absence of any mutations in c-KIT exon 11 (where the majority of hot-spot imatinib-responding mutations are located) and the concomitant presence of SCF in all ESFT samples, an autocrine/paracrine loop can be assumed as the mechanism for KIT activation in 40% (4/10) of ESFT patients. These findings confirm our previously published data.5 Moreover, the literature data report a complete wildtype c-Kit examined by RT-PCR and sequencing in 4 ESFT patients.16
Sections from the same tumor material were analyzed for PDGFRα and PDGFR β: they revealed immunoreactivity in all cases for the former and in 79% (15/19) of the cases for the latter. These data are consistent with the higher levels (vis-a-vis the calibrator) of PDGFRα and PDGFRβ mRNA revealed by qPCR. By comparison with the protein pool, our IP/WB experiments on the 10 frozen ESFT samples showed an increase in quantity and a phosphorylation of PDGFRα and PDGFRβ in 60% (6/10) and 70% (7/10) of ESFT specimens, respectively (both conditions occurring concomitantly in 4 cases, ie, 3, 4, 6, and 15; Table 1). The higher protein levels correlated with both IHC reactivity and higher mRNA levels in most cases. The present findings extend and complement previously published results, where the presence of a functional PDGFRβ was reported in 8 of 9 ESFT cell lines.7 This study is at variance with that study, however, which found no PDGFRα protein in 9 of 9 ESFT cell lines; our present findings suggest that in vitro data do not necessarily reflect the situation in vivo. Indeed, the effect of milieu factors (paracrine effects) may justify the expression of PDGFRα.
Given the lack of activating mutations in the tyrosine kinase and juxtamembrane domains of PDGFRα and PDGFRβ, and the presence of cognate ligands (PDGFA and PDGFB) in all samples, an autocrine/paracrine loop can be suggested for the activation of both receptors. To our knowledge, this is the first report in which activated PDGFRα and PDGFRβ (sustained by an autocrine/paracrine loop) have been demonstrated in 60% (6/10) and 70% (7/10) of ESFT tumor specimens, respectively.
On the whole, our results show that biochemical analyses provide qualitatively and quantitatively better information than IHC. Indeed, evidence of phosphorylation/activation is more specific and predictive than overexpression, and the proportion of cases with protein expression is higher using frozen section than fixed materials, particularly as concerns KIT protein.17
It has recently been emphasized that autocrine/paracrine oncogenic mechanisms can be associated with markedly low levels of RTK activation that can nonetheless be successfully inhibited by targeted treatment, providing the tumor cells depend on the signaling mechanism in question.14 We still do not know whether the autocrine paracrine oncogenic effect of KIT, PDGFRα, and PDGFRβ play a vital part in ESFT, but our findings, coupled with the results obtained on ESFT cell lines4, 7 and in vivo,18 support this hypothesis. Interestingly, mouse inoculated with KIT-positive ESFT derived cell lines and then treated with imatinib showed a significant tumor striking coupled with a time- and concentration-dependent global protein dephosphorylation, although it remained sensitive to SCF stimulation.18 This result does not rule out that the imatinib-dependent apoptotic signal may involve other RTK-imatinib as sensitive as PDGFRα and -β rather than KIT.
Moreover, given the effectiveness of cytotoxic chemotherapy on this disease and recently published preclinical19 and clinical data on response to imatinib in dermatofibrosarcoma protuberans (DFSP) patients,14 imatinib might be indicated in patients with advanced disease carrying deregulation of these biomarkers. The results obtained in DFSP suggest that neither high levels of RTK activation nor RTK overexpression are necessary for a clinical response to imatinib.14 It is also worth noting that, by reducing tumor interstitial fluid pressure, imatinib may increase chemotherapy drug uptake, enhancing the effectiveness of current treatments.20 In this light, the existing data, suggesting that imatinib, when applied alone, is not effective in ESFT patients21 should be reevaluated.
Moreover, in 2 patients progressing on imatinib, 1 suffering from sacral chordoma (unpublished) and the other from both sacral chordoma and lung cancer,22 a tumor response was apparently reestablished combining it with cisplatin.
Certainly, an efficient drug administration schedule needs to be established, bearing in mind that preliminary data would suggest that a sequential administration may be effective.
Supported by a grants from the Associazione Italiana per la Ricerca sul Cancro (AIRC 2004) and the Italian Ministry of Health (Ricerca finalizzata 2004).
- 18Imatinib mesylate (STI571) interference with growth of neuroectodermal tumour cell lines does not critically involve c-Kit inhibition. Int J Mol Med. 2004; 3: 373–382., , , et al.
- 21Imatinib mesylate in soft tissue and bone sarcomas: interim results of a Sarcoma Alliance for Research thru Collaboration (SARC) phase II trial. ASCO. 2004; 23: 814., , , et al.