PDGFRA Activating Mutations in Gastrointestinal Stromal Tumors

See allHide authors and affiliations

Science  31 Jan 2003:
Vol. 299, Issue 5607, pp. 708-710
DOI: 10.1126/science.1079666


Most gastrointestinal stromal tumors (GISTs) have activating mutations in the KIT receptor tyrosine kinase, and most patients with GISTs respond well to Gleevec, which inhibits KIT kinase activity. Here we show that ∼35% (14 of 40) of GISTs lacking KITmutations have intragenic activation mutations in the related receptor tyrosine kinase, platelet-derived growth factor receptor α (PDGFRA). Tumors expressing KIT or PDGFRA oncoproteins were indistinguishable with respect to activation of downstream signaling intermediates and cytogenetic changes associated with tumor progression. Thus, KIT and PDGFRA mutations appear to be alternative and mutually exclusive oncogenic mechanisms in GISTs.

Gastrointestinal stromal tumors (GISTs) are the most common mesenchymal tumors of the gastrointestinal tract and are particularly sensitive to Gleevec, a new cancer therapy (1–5). Gleevec inhibits the constitutively activated form of the KIT receptor tyrosine kinase, which is the critical transforming oncoprotein in more than 85% of GISTs. Although most GISTs have activating KIT mutations, a subset are KITwild type (KIT-WT) (6, 7). Notably, a high level of total KIT protein expression, which is a defining feature of GISTs (2, 7, 8), is not characteristic ofKIT-WT GISTs (fig. S1). However, cells of the interstitial cells of Cajal lineage—from which GISTs are thought to arise—express KIT strongly (2, 9). Therefore, activation of another oncoprotein in KIT-WT GISTs might be accompanied byKIT transcriptional down-regulation.

To explore alternative receptor tyrosine kinase (RTK) oncoproteins that might participate in GIST pathogenesis, we performed immunoprecipitations with polyclonal antisera (panRTK antisera) against peptides from regions of strong sequence conservation across the family of receptor tyrosine kinases. These panRTK antisera have been extensively validated in immunoprecipitations of ERBB2, NTRK3, ALK, and KIT kinase oncoproteins from lysates of frozen human tumors (10). Phosphotyrosine immunostaining of panRTK immunoprecipitates from a KIT-WT GIST (GIST478) revealed presumptive phosphoproteins of about 150 and 170 kD, consistent with the size of immature and mature glycosylated PDGFRA, respectively (Fig. 1, A and B). We next showed that phosphorylated PDGFRA (phosphoPDGFRA) was strongly expressed and comigrated with the panRTK phosphoproteins and several of the predominant phosphorylated proteins from the GIST478 whole-cell lysate (Fig. 1, A and B). By contrast, KIT was not demonstrably phosphorylated. Therefore, phosphoPDGFRA appeared to be the predominant phosphoRTK in GIST478.

Figure 1

Identification of phosphorylated receptor tyrosine kinases in KIT-mutant and KIT-WT GISTs. (A) Expression of phosphoRTKs and phoshoPDGFRA, but not phosphoKIT, in KIT-WT GIST478. Lysates were prepared from frozen tumor; immunoprecipitated with polyclonal antibodies to panRTK, KIT, and PDGFRA; and immunostained for phosphotyrosine (pY). A tyrosine-phosphorylated panRTK 150/170-kD doublet (lane 2) corresponds to two of the phosphoproteins in the total cell lysate (lane 1) and comigrates with the phosphorylated PDGFRA doublet (lane 4). There is no detectable phosphorylation of KIT (lane 3). IP, immunoprecipitate; IB, immunoblot. (B) PDGFRA immunostaining of the same blot confirms that the phosphorylated RTK (lane 2) is PDGFRA. (C) Tyrosine-phosphorylated KIT (145 kD) and PDGFRA (150/170 kD) are expressed in theKIT-mutant GIST780 and KIT-WT GIST1015, respectively. GIST780 does not express phosphoPDGFRA, and GIST1015 does not express phosphoKIT. (D) PDGFRA is detected in bothKIT-mutant GIST780 and KIT-WT GIST1015. (E) KIT is detected in KIT-mutant GIST780 but not in KIT-WT GIST1015. (F to H) Immunostaining of whole-cell lysates demonstrates tyrosine-phosphorylated PDGFRA in KIT-WT GISTs (2686, 478, and 1015), whereas KIT is expressed only inKIT-mutant GISTs (174 and 208).

Mutually exclusive phosphoKIT and phosphoPDGFRA expression was demonstrated, respectively, in a GIST with a KIT juxtamembrane region mutation and another KIT-WT GIST (Fig. 1, C to E). We then confirmed differential phosphoPDGFRA expression in threeKIT-WT and two KIT-mutant GISTs. PhosphoPDGFRA expression was restricted to the KIT-WT GISTs, where KIT expression was low to undetectable (Fig. 1, F to H). Likewise, PDGFRA expression was low to undetectable in ∼70% of KIT-mutant GISTs (fig. S2). Confirmation of GIST diagnosis in KIT-WT GISTs was enabled by immunostaining for protein kinase C θ (fig. S3).

We identified PDGFRA activation loop (exon 18) mutations in the three KIT-WT GISTs that expressed phosphoPDGFRA (Fig. 1). Two of the KIT-WT GISTs had an identical PDGFRA missense mutation, leading to substitution of valine for the highly conserved aspartic acid at codon 842 (PDGFRA D842V) (11). The other KIT-WT GIST had an in-frame deletion, resulting in loss of PDGFRA amino acid residues 842 to 845 (DIMH). These PDGFRA mutations (Table 1) (figs. S4 to S6) are homologous to those responsible for KIT and FMS-related tyrosine kinase 3 (FLT3) ligand–independent kinase activation in human mast cell disorders, acute myeloid leukemia, and germ cell (seminoma) tumors (12–16).

Table 1

Summary of PDGFRA mutations inKIT-WT GISTs (11).

View this table:

We then determined the frequency of PDGFRA mutations in GIST formalin-fixed tissues, where PDGFRA phosphorylation status was not known. We evaluated PDGFRA genomic mutations in exons 10, 12, 14, and 18, which correspond to the KIT exons containing oncogenic mutations in many GISTs. We foundPDGFRA mutations in 11 of 37 (29.7%) KIT-WT GISTs but not in 36 KIT-mutant GISTs. The somatic nature of the PDGFRA mutations was confirmed by genomic sequencing of non-neoplastic tissues, which were exclusively PDGFRA wild type, in four patients (17). The 11 KIT-WT GISTs with PDGFRA mutations included six additional tumors with D842V and one with an in-frame exon 18 deletion mutation (HDSN845-848P) overlapping the deleted residues 842 to 845 (DIMH) (Table 1) (fig. S6). The other four PDGFRA mutations involved the juxtamembrane region encoded by exon 12 (Table 1) (fig. S6) and would be expected to induce constitutive PDGFRA kinase activation. Overall, we detected PDGFRA mutations in 14 of 40KIT-WT GISTs (35%) and in none of 36 KIT-mutant GISTs (P < 0.0001, Fisher's exact test). Thus, KIT and PDGFRA mutations appear to be mutually exclusive in GISTs.

We studied the biochemical consequences of somatic PDGFRA mutations by transient expression of wild-type and mutant PDGFRA cDNA constructs in Chinese hamster ovary (CHO) cells. Baseline tyrosine phosphorylation was weak for nonmutant PDGFRA and was substantially increased by ligand stimulation (Fig. 2). By contrast, baseline tyrosine phosphorylation was strong in all five of the tested PDGFRA mutants and was not increased by ligand stimulation (Fig. 2).

Figure 2

PDGFRA mutations in GISTs result in constitutive activation of PDGFRA kinase. CHO cells were transiently transfected with expression vectors encoding cDNAs for wild-type (WT) or mutant PDGFRA. Two juxtamembrane (exon 12) and three activation loop (exon 18) mutations were tested for constitutive activation. Transfected cells were serum-starved overnight and treated with vehicle or ligand (recombinant human PDGF-AA) for 10 min. Whole-cell lysates were immunostained sequentially for phosphotyrosine and PDGFRA. Wild-type PDGFRA displays low-level phosphorylation that is up-regulated by ligand stimulation with PDGF-AA. In contrast, the mutant PDGFRA proteins display ligand-independent phosphorylation.

We next compared the signal transduction pathways activated inPDGFRA-mutant versus KIT-mutant GISTs. ThePDGFRA-mutant GISTs showed uniform activation of signaling intermediates protein kinase B (AKT), mitogen-activated protein kinase (MAPK), and the STAT (signal transducers and activators of transcription) proteins Stat1 and Stat3; all of these are also activated in most KIT-mutant GISTs (Fig. 3) (18). The PDGFRA-mutant GISTs lacked expression of phosphoStat5 despite strong expression of total Stat5, which is also typical of KIT-mutant GISTs. We also compared the cytogenetic profiles of four PDGFRA-mutant GISTs and 52 KIT-mutant GISTs. KIT mutations are early events in GIST tumorigenesis, whereas cytogenetic aberrations occur later in disease progression (8). Most of these GISTs, regardless of PDGFRA or KIT mutation, featured noncomplex karyotypes with deletions of chromosome 1p and with monosomies of chromosomes 14 and 22 (table S1 and fig. S7). Hence, our studies suggest that the mechanisms of cytogenetic progression and oncoprotein-driven signal transduction are similar in GISTs expressing oncogenic forms of PDGFRA and KIT.

Figure 3

Cell signaling profiles inPDGFRA-mutant (2686, 478, and 1015) andKIT-mutant GISTs (174 and 208). Whole-cell lysates were prepared from snap-frozen GISTs, and immunoblots were detected with antibodies to phosphorylated and total forms of AKT, MAPK, and STATs. All GISTs express phosphorylated AKT, MAPK, Stat1, and Stat3, whereas Stat5 is not tyrosine-phosphorylated.

We conclude that activating mutations of KIT or PDGFRA are mutually exclusive oncogenic events in GISTs and that these mutations have similar biological consequences. Our data also highlight a crucial role for PDGFRA in the pathogenesis of a solid tumor. Notably, a translocation involving the BCR and PDGFRA genes has been described in BCR-ABL–negative chronic myelogenous leukemia and is predicted to result in dimerization and kinase activation of the fusion protein (19). PDGFRA is widely expressed in human tissues, so it will be important to determine whether PDGFRA mutations play a role in other human malignancies. Such tumors could be sensitive to Gleevec and other small-molecule drugs that inhibit PDGFRA kinase activity (20–22).

Supporting Online Material

Materials and Methods

Figs. S1 to S7

Table S1


  • * To whom correspondence should be addressed. E-mail: heinrich{at}, jfletcher{at}


Stay Connected to Science

Navigate This Article