Functional Determination of the Interaction Interface Between the Kinase and Pseudokinase Domains of JAK2 V617F

The Janus kinases (JAKs) are cytosolic protein tyrosine kinasess mediating cytokine receptor function essential for blood formation and the immune response. The V617F mutation in the pseudokinase domain of JAK2 is present in over 95% of Polycythemia Vera patients and in 50% of Essential Thrombocythemia and Primary Myelofibrosis patients. The V617F mutation leads to constitutive kinase activity, which activates downstream STAT5, STAT3, MAP-kinase and phosphatidylinositol-3-kinase pathways. In the absence of a complete three-dimensional structure of JAK2, the structural mechanism of activation by the V617F mutation has remained elusive. Position 617 of JAK2 is located within one of the two predicted interfaces between the pseudokinase and the kinase domains of JAK2. We have shown that large hydrophobic residues such as Trp, Leu, Ile and Met also activate JAK2 when substituted for Val at position 617. These V617 mutations might either remove an inhibitory interaction exerted by the pseudokinase domain or stabilize an active conformation of the kinase domain. To investigate the nature of the conformational change promoted by the V617 mutations, we employed cysteine-scanning mutagenesis of residues in the activation loop of JAK2, and their predicted interaction partners on the pseudokinase side, and characterized their constitutive and ligand-dependent activity in proliferation and dual luciferase assays. The solved structure of the JAK2 kinase domain in its active form places residues Q1003, D1004 and K1005 on the tip of the activation loop. These residues were individually mutated to cysteine and stably expressed in murine pro-B Ba/F3 cells expressing the erythropoietin receptor. We hypothesized that cysteine residues placed in the activation loop might promote interactions with the pseudokinase domain or form disulfide bonds with naturally occurring Cys616 and Cys618 of the pseudokinase domain, if, as predicted, close proximity would occur between activation loop residues and the pseudokinase domain loop beta4-beta5 that contains V617F. GFPsorted cells expressing the K1005C mutant were able to proliferate in a medium lacking cytokines, similar to cells expressing V617F, but to a lower degree. This proliferation advantage could be explained by the formation of a disulfide bond between C1005 and C618 of JAK2, or by a conformational change induced by placing the smaller Cys residue at position 1005, which would lock the activation loop in an active conformation. Substitution of V617 to Cys abolished the constitutive signaling by JAK2 K1005C, and decreased the weak activity of JAK2 V617C, suggesting that the active conformations of V617 mutants and K1005C are different. Cysteine mutants at positions 1003 or 1004 could not synergize with V617C or induce constitutive activation. We also investigated whether JAK2 V617F activation simply required dimerization, or rather a particular dimer orientation of EpoR cytosolic domains. We took advantage of fusion proteins where a 28 amino acid coiled coil was used to replace the extracellular domain of EpoR in order to impose discrete dimeric orientations to the transmembrane and cytosolic domains of the EpoR. Two out of the seven fusion proteins were constitutively active when co-expressed with WT JAK2 in STAT5-dependent luciferase assays. In contrast, all dimers were active in the presence of JAK2 V617F, suggesting that dimerization and not a particular dimeric orientation promotes JAK2 V617F activation.