Approximately 50% of myeloproliferative neoplasms (MPNs) harbor the JAK2 V617F mutation while approximately 50% of B-cell acute lymphoblastic leukemias (B-ALLs) with CRLF2 rearrangements harbor JAK2 exon 16 mutations that primarily involve R683. Multiple enzymatic inhibitors of JAK2 are in clinical development for the treatment of patients with malignant and nonmalignant conditions that depend on constitutive JAK2 signaling. Most of these drugs are ATP-mimetics that block JAK2 signaling in the active conformation (so-called “type I JAK2 inhibitors”). Resistance to type I JAK2 inhibitors can occur through heterodimerization between activated JAK2 and either JAK1 or TYK2 (Koppikar et al. Nature 2012). In addition, E864K, Y931C, and G935R mutations in the kinase domain of JAK2 (JH1 domain) confer resistance to a panel of type I JAK2 inhibitors (including ruxolitinib, tofacitinib, TG101348, JAK inhibitor I) without drastically affecting signaling by JAK2 (Weigert et al. J Exp Med 2012). Resistance caused by these mutations is independent of whether in the context of CRLF2 with JAK2 R683G or EPOR with JAK2 V617F (Weigert et al. J Exp Med 2012). In contrast to type I inhibitors, type II JAK2 inhibitors bind to and stabilize the inactive confirmation of JAK2 and prevent the activation loop from being phosphorylated. Thus, transphosphorylation of JAK2 by JAK1 or TYK2 does not confer resistance to the type II JAK2 inhibitor NVP-BBT594 (BBT594) (Koppikar et al. Nature 2012). In this study we report the first evidence that mutation of JAK2 can also confer resistance to type II Jak2 inhibitors. BBT594 had similar potency to the type I JAK2 inhibitor NVP-BVB808 (BVB808) in murine lymphoblast BaF3 cells dependent on CRLF2 with JAK2 R683G (IC50 8.5nM vs 15.7nM) or EPOR with JAK2 V617F (IC50 29nM vs 10nM). In contrast, the Y931C mutation conferred >3-fold resistance to BVB808 in BaF3 cells expressing CRLF2 with JAK2 R683G but no significant change in sensitivity to BBT594. Thus, type II JAK2 inhibitors can overcome genetic resistance to type I JAK2 inhibitors. We performed a random mutagenesis screen of JAK2 R683G and expressed the mutagenized library in BaF3 cells that express CRLF2. Selecting in the presence of 3uM BBT594 resulted in a large number of clones, of which all screened (n>30) harbored the same JAK2 L884P mutation. Structural modeling of this mutation predicted changes in the JH1 domain that may impact the conformation of the P-loop and helix C, and thereby compromise the sub-pocket required for type II inhibitor binding. In contrast to mutations that confer resistance to type I JAK2 inhibitors, the L884P mutation only conferred resistance to BBT594 in the context of CRLF2/JAK2 R683G (IC50 504nM versus 8.5nM for R683G alone) and not EPOR/JAK2 V617F. To our knowledge, this is the first mechanism of resistance specific to JAK2 R683G. BaF3 cells expressing CRLF2 with JAK2 R683G L884P Y931C remained resistant to BBT594. Transduction of the mutagenized JAK2 R683G library into BaF3 cells expressing CRLF2 followed by selection in both BVB808 and BBT594 did not yield any resistant colonies. In conclusion, mutations that affect the binding of type I JAK2 inhibitors do not affect the potency of the type II JAK2 inhibitor BBT594. The L884P mutation confers resistance to BBT594 when co-occurring with the activating mutation R683G but not with V617F. Thus, combinations of multiple JAK2 inhibitors with distinct mechanisms may be useful in overcoming de novo and acquired resistance to JAK2 inhibitors.
Vangrevelinghe:Novartis: Employment. Radimerski:Novartis: Employment. Weinstock:Novartis: Consultancy, Research Funding.
Asterisk with author names denotes non-ASH members.