Modulation of activation-loop phosphorylation by JAK inhibitors is binding mode dependent

Abstract

Janus kinase (JAK) inhibitors are being developed for the treatment of rheumatoid arthritis, psoriasis, myeloproliferative neoplasms, and leukemias. Most of these drugs target the ATP-binding pocket and stabilize the active conformation of the JAK kinases. This type I binding mode can lead to an increase in JAK activation loop phosphorylation, despite blockade of kinase function. Here we report that stabilizing the inactive state via type II inhibition acts in the opposite manner, leading to a loss of activation loop phosphorylation. We used X-ray crystallography to corroborate the binding mode and report for the first time the crystal structure of the JAK2 kinase domain in an inactive conformation. Importantly, JAK inhibitor-induced activation loop phosphorylation requires receptor interaction, as well as intact kinase and pseudokinase domains. Hence, depending on the respective conformation stabilized by a JAK inhibitor, hyperphosphorylation of the activation loop may or may not be elicited.

Figure 1. Increase of JAK activation-loop phosphorylation by JAK inhibitors

A, SET-2 cells were treated for 30 minutes with JAK inhibitors at 1 μM or DMSO and then extracted for Western blot analysis of JAK2 Y1007/Y1008 and STAT5 Y694 phosphorylation. JAK2 and STAT5 served as loading controls. B, SET-2 cells were treated with increasing concentrations of NVP-BSK805 for 30 minutes and then assessed as described above. C, Non-targeting (Ctrl) or JAK2 targeting siRNA oligos were transfected into HEL92.1.7 cells. After 72 h, cells were treated for 30 minutes with 1 μM NVP-BSK805 or DMSO and then assessed as described above. D, SET-2 cells were treated with 1 μM NVP-BSK805 or DMSO for 30 minutes. JAK2 was immuno-precipitated (IP) using an amino- or carboxyl-terminal antibody, followed by Western blot analysis of P-JAK2 and JAK2. E, CMK cells were treated for 30 minutes with JAK inhibitors at 1 μM or DMSO and then extracted for Western blot analysis of JAK3 Y980 (following JAK3 IP) and STAT5 phosphorylation. F, TF-1 cells were starved in medium without GM-CSF overnight and then either pre-treated with DMSO or JAK inhibitors at 1 μM for 30 minutes. Cells were then stimulated or not with 10 ng/mL IFN-α for 10 minutes, followed by extraction for Western blot analysis of TYK2 Y1054/Y1055 (after TYK2 IP) and STAT5 phosphorylation.

Figure 2. JAK inhibitor-induced JAK activation-loop phosphorylation can be transient or sustained and is also seen in vivo

A, SET-2 cells were treated for 2 h with different JAK inhibitors at 1 μM, followed by washing, transferring back into medium and extraction at the indicated time-points. Control cells were treated with DMSO. JAK2 Y1007/Y1008 and STAT5 Y694 phosphorylation were detected by Western blotting. JAK2 and STAT5 served as loading controls. B, JAK inhibitor kinetic parameters determined in biochemical assays with JAK2 JH1. C, mice received a subcutaneous injection of 10 U rhEpo and were orally administered 25, 50 or 100 mg/kg NVP-BSK805. Control animals received either a subcutaneous injection of saline or 10 U rhEpo and were orally administered vehicle. After 3 h, spleen samples were processed for detection of P-JAK2 and P-STAT5 levels as described above. D, irradiated mice transplanted with bone marrow transduced with MPL^W515L were followed for development of leukocytosis and thrombocytosis. Mice were then administered either vehicle or NVP-BSK805 at indicated dose-levels. After 24 h, levels of JAK2, STAT3 and MAPK phosphorylation in spleen extracts were assessed by Western blotting. Actin served as loading control.

Figure 3. Type-II mode of JAK inhibition suppresses both JAK activation-loop and substrate phosphorylation

A, SET-2 cells were treated for 30 minutes with 1 μM NVP-BBT594 or DMSO and then extracted for Western blot analysis of JAK2 Y1007/Y1008 phosphorylation and STAT5 Y694 phosphorylation. JAK2 and STAT5 were probed for as loading controls. B, SET-2 or CMK cells were treated with increasing concentrations of NVP-BBT594 for 1 h and then extracted for detection of PJAK2, P-JAK3 (Y980) and P-STAT5 by Western blotting. C, SET-2 cells were treated with 1 μM NVP-BBT594 or DMSO for 30 minutes. JAK2 was immuno-precipitated using an amino- or carboxyl-terminal antibody, followed by Western blot analysis of levels of P-JAK2 and JAK2 protein that was IPd. D, SET-2 cells were treated for 2 h with 1 μM NVP-BBT594, followed by washing, transferring back into medium and extraction at the indicated time points. Control cells were treated with DMSO. P-JAK2 and P-STAT5 levels were assessed by Western blotting.

Figure 4. JAK2 JH1 in complex with NVP-BBT594 at 1.34 Å resolution

A, chemical structure of NVP-BBT594. B, overall ribbon representation of the JAK2 kinase domain with the bound inhibitor NVP-BBT594 illustrated as stick model. Inhibitor binding occurs to the DFG-out conformation of the kinase domain. Residues 1000-1012 from the activation-loop did not show electron density and are omitted from the final model. C, stereo view of NVP-BBT594 bound to JAK2. Polar contacts between the protein, the inhibitor molecule and solvent are indicated with dotted green lines.

Figure 5. JAK2 type-I inhibitor-induced increase of JAK2 activation-loop phosphorylation is staurosporine sensitive and suppressed by reduction of ATP levels

A, SET-2 cells were pretreated for 30 minutes with DMSO control (Ctrl), 10 μM staurosporine or imatinib, followed by treatment for 1 h with 1 μM NVP-BSK805 or DMSO. JAK2 Y1007/Y1008 and STAT5 Y694 phosphorylation were assessed by Western blotting. B, non-targeting or BTK targeting siRNA oligos were transfected into SET-2 cells. After 72 h, cells were treated for 1 h with 1 μM NVP-BSK805 or DMSO and P-JAK2 and P-STAT5 levels were assessed as above. JAK2, STAT5 and BTK served as loading controls and to verify siRNA-mediated target knockdown. C, non-targeting or LYN targeting siRNA oligos were transfected into SET-2 cells, followed by treatment and analysis as described above. D, SET-2 cells were treated for 30 minutes with 10 mM 2-deoxy-D-glucose and 20 μM oligomycin A or drug vehicle. Cells were then treated for 20 minutes with 1 μM NVP-BSK805 or DMSO. P-JAK2 and P-STAT5 levels were assessed as above. E, Extracts from the experiment shown in D were also probed (left panels) for phosphorylated AMPKα. AMPKα and β-tubulin served as loading controls. The histogram depicts relative ATP levels (means of four independent experiments ± SD) in SET-2 cells treated as described in D.

Figure 6. JAK type-I inhibitor-induced increase of JAK activation-loop phosphorylation requires intact JH1 and JH2 domains

A, HEL92.1.7 cells were transiently transfected with the indicated JAK2 constructs. Control cells were transfected with empty vector (Ctrl). After 24 h, cells were treated with 1 μM NVP-BSK805 or DMSO for 30 minutes, followed by extraction for Western blot detection of P-JAK2 (Y1007/Y1008) and P-STAT5 (Y694) levels. B, Ba/F3 EpoR cells were transiently transfected with the indicated JAK2 constructs or empty vector (Ctrl). After 24 h cells were treated as described above. C, HEL92.1.7 cells were transiently transfected with the indicated JAK1 constructs or empty vectors (Ctrl). After 24 h, cells were either treated with 1 μM JAK Inhibitor 1 or DMSO for 1 hour, followed by extraction for Western blot detection of JAK1 Y1022/Y1023 phosphorylation, and of PJAK2 and P-STAT5 as above. Total levels of JAK1, JAK2 and STAT5 served as loading controls. D, HEL92.1.7 cells were transiently transfected and treated as in C to assess the impact of the JAK1^K648R control mutant on JAK Inhibitor 1-induced activation-loop phosphorylation of co-transfected JAK1^V658F.