A newly identified 45-kDa JAK2 variant with an altered kinase domain structure represents a novel mode of JAK2 kinase inhibitor resistance

Tyrosine-protein kinase (janus kinase; JAK)-signal transducer and activator of transcription (STAT) signaling plays a pivotal role in the development of myeloproliferative neoplasms (MPNs). Treatment with the potent JAK1/JAK2-specific inhibitor, ruxolitinib, significantly reduces tumor burden; however, ruxolitinib treatment does not fully eradicate the malignant clone. As the molecular basis for the disease persistence is not well understood, we set out to gain new insights by generating ruxolitinib-resistant cell lines. Surprisingly, these cells harbor a 45 kDa JAK2 variant (FERM-JAK2) consisting of the N-terminal FERM domain directly fused to the C-terminal kinase domain in 80% of sublines resistant to ruxolitinib. At the molecular level, FERM-JAK2 is able to directly bind and activate STAT5 in the absence of cytokine receptors. Furthermore, phosphorylation of activation-loop tyrosines is dispensable for FERM-JAK2-mediated STAT5 activation and cellular transformation, in contrast to JAK2-V617F. As a result, FERM-JAK2 is highly resistant to several ATP-competitive JAK2 inhibitors, whereas it is particularly sensitive to HSP90 inhibition. A murine model of FERM-JAK2 leukemogenesis showed an accelerated MPN phenotype with pronounced splenomegaly. Notably, most current protocols for the monitoring of emerging JAK variants are unable to detect FERM-JAK2, highlighting the urgent need for implementing next-generation sequencing approaches in MPN patients receiving ruxolitinib.