A Novel Fusion Gene NDEL1-Pdgfrb in a Patient with JMML with a New Variant of TKI-Resistant Mutation in the Kinase Domain of PDGFRβ

We have identified a novel fusion gene in an 18-month old child with juvenile myelomonocytic leukemia (JMML) displaying a reciprocal chromosomal translocation t(5;7)(q33;p11.2). Molecular investigation at diagnosis revealed absence of mutations in KRAS, NRAS, PTPN11, or cCBL, but FISH analysis identified a rearrangement involving the PDGFRB gene located on chromosome 5q33. After temporary responses to imatinib (IM) and subsequently nilotinib (NIL) treatment, resistance associated with disease relapses was observed. Employment of the 5’-RACE technique facilitated identification of the PDGFRB fusion partner on chromosome 7p11.2, the NDEL1 gene encoding the nudE neurodevelopmental protein 1-like 1. The NDEL1 gene has not been implicated in any other reciprocal translocation to date, and it is conceivable that its ability to form dimers could drive permanent kinase activation of PDGFRβ. The chimeric mRNA contains the 5´exons 1-5 of NDEL1 fused in frame to the PDGFRB exons 10-22 containing the transmembrane and tyrosine kinase domains.

To assess the oncogenicity of the fusion protein, Ba/F3 cells were transduced with the NDEL1-PDGFRB gene construct. The observation of IL3-independent growth confirmed the oncogenic potential of the novel fusion gene. The observed clinical resistance to IM and NIL prompted us to analyze the entire PDGFRB kinase domain for the presence of mutations by Sanger sequencing of overlapping amplicons. A point mutation in the activation (A) loop converting aspartate at the position 850 into glutamate (D850E) was detected in peripheral blood specimens from the time of first and second relapses, but not in the diagnostic sample. The crystal structure of the PDGFRβ TKD is not available, but protein modelling suggested that the mutation D850E destabilizes the inactive confirmation of the A-loop. This notion was in line with the observed clinical resistance to IM and NIL, but suggested sensitivity of the mutant to dasatinib (DAS). To test the predicted TKI responses, Ba/F3 cells transduced with wild type or mutant NDEL1-PDGFRB were tested in MTT assays against a panel of TKIs: Ba/F3-NDEL1-PDGFRB^WT cells were sensitive to IM (IC₅₀ = 60 nM), NIL (100 nM), DAS (5 nM), sorafenib (SOR; 20 nM), and ponatinib (PON; 10 nM), but insensitive to bosutinib (BOS; >2500 nM). Conversely, Ba/F3-NDEL1-PDGFRB^D850E cells exhibited high resistance to IM (>2500), a 10-fold higher IC₅₀ for NIL (1000 nM) and a 100-fold higher IC₅₀ for SOR (2500 nM), but retained sensitivity to PON (15 nM) and DAS (15 nM).

Mutations in the A-loop of different tyrosine kinases such as PDGFRα (D842V) or c-Kit (D816V) associated with resistance to IM have already been described in different tumor entities. However, the mutation D850E in the PDGFRβ TKD with apparent insensitivity to IM, NIL, and SOR revealed a completely different pattern of resistance than the same amino acid exchange at the corresponding site of PDGFRα (D842E). The latter mutation was previously shown to be sensitive to IM, NIL, and SOR with IC₅₀ values of 4, 12.5, and 0.25 nM, respectively. This difference is intriguing because the exchange of a negatively charged amino acid, aspartate, to an amino acid with the same physical properties, glutamate, is not known to exert a major structural effect on the protein conformation, as observed for the D842E mutation in PDGFRα. We speculate that the great difference between the presence of the same amino acid exchange at corresponding positions in PDGFRα and PDGFRβ is the main interaction amino acid partner residue of aspartate at the position +3 which may influence the stability of the A-loop in its inactive conformation. In PDGFRα, it is histidine whose physical interaction with aspartate might not be affected by the change to glutamate. By contrast, the electrostatic bonds between arginine as the +3 residue in PDGFRβ might be greatly weakened by the elongation of the side chain in glutamate in comparison with aspartate, thus destabilizing the inactive conformation of the A-loop resulting in resistance to type II TKIs. To our knowledge, this is the first observation of an exchange between two negatively charged amino acids in a tyrosine kinase associated with a major change in responsiveness to TKI treatment. This finding is currently under further investigation, and may extend our understanding of structural interactions leading to TKI resistance. (Supported by the FWF SFB grant F4705-B20).