Hyperactivation of Oncogenic JAK3 Mutants Depend on ATP Binding to the Pseudokinase Domain

ORIGINAL RESEARCH published: 03 December 2018 doi: 10.3389/fonc.2018.00560 Frontiers in Oncology | www.frontiersin.org 1December 2018 | Volume 8 | Article 560 Edited by: Takaomi Sanda, Yong Loo Lin School of Medicine, National University of Singapore, Singapore Reviewed by: Maria L. Toribio, Centro de Biología Molecular Severo Ochoa (CSIC-UAM), Spain James M. Murphy, Walter and Eliza Hall Institute of Medical Research, Australia *Correspondence: Olli Silvennoinen [email protected] Specialty section: This article was submitted to Molecular and Cellular Oncology, a section of the journal Frontiers in Oncology Received: 10 July 2018 Accepted: 09 November 2018 Published: 03 December 2018 Citation: Raivola J, Hammarén HM, Virtanen AT, Bulleeraz V, Ward AC and Silvennoinen O (2018) Hyperactivation of Oncogenic JAK3 Mutants Depend on ATP Binding to the Pseudokinase Domain. Front. Oncol. 8:560. doi: 10.3389/fonc.2018.00560 Hyperactivation of Oncogenic JAK3 Mutants Depend on ATP Binding to the Pseudokinase Domain Juuli Raivola1, Henrik M. Hammarén 1, Anniina T. Virtanen1, Vilasha Bulleeraz2,3, Alister C. Ward2,3 and Olli Silvennoinen 1,4,5 * 1Faculty of Medicine and Life Sciences, University of Tampere, Tampere, Finland, 2School of Medicine, Deakin University, Geelong, VIC, Australia, 3Centre for Molecular and Medical Research, Deakin University, Geelong, VIC, Australia, 4Fimlab Laboratories, Pirkanmaa Hospital District, Tampere, Finland, 5Institute of Biotechnology, University of Helsinki, Helsinki, Finland Janus kinase 3 (JAK3) tyrosine kinase has a central role in the control of lymphopoiesis, and mutations in JAK3 can lead to either severe combined immunodeficiency or leukemia and lymphomas. JAK3 associates with the common gamma chain (γc) receptor and functions in a heteromeric signaling pair with JAK1. In IL-2 signaling JAK1 is the effector kinase for STAT5 phosphorylation but the precise molecular regulatory mechanisms of JAK1 and JAK3 and their individual domains are not known. The pseudokinase domain (JAK homology 2, JH2) of JAK3 is of particular interest as approximately half of clinical JAK3 mutations cluster into it. In this study, we investigated the role of JH2s of JAK1 and JAK3 in IL-2R signaling and show that STAT5 activation requires both JH1 and JH2 of JAK1, while both JH1 and JH2 in JAK3 are specifically required for the cytokine-induction of cellular signaling. Characterization of recombinant JAK3 JH2 in thermal shift assay shows an unstable protein domain, which is strongly stabilized by ATP binding. Unexpectedly, nucleotide binding to JAK3 JH2 was found to be cation-independent. JAK3 JH2 showed higher nucleotide binding affinity in MANT-ATP and fluorescent polarization competition assays compared to the other JAK JH2s. Analysis of the functional role of ATP binding in JAK3 JH2 in cells and in zebrafish showed that disruption of ATP binding suppresses ligand-independent activation of clinical JAK3 gain-of-function mutations residing in either JH2 or JH1 but does not inhibit constitutive activation of oncogenic JAK1. ATP-binding site mutations in JAK3 JH2 do not, however, abrogate normal IL-2 signaling making them distinct from JH2 deletion or kinase-deficient JAK3. These findings underline the importance of JAK3 JH2 for cellular signaling in both ligand-dependent and in gain-of-function mutation-induced activation. Furthermore, they identify the JH2 ATP-binding site as a key regulatory region for oncogenic JAK3 signaling, and thus a potential target for therapeutic modulation. Keywords: JAK kinase, nucleotide binding, pseudokinase, cytokine, leukemia Raivola et al. JAK3 JH2 Regulates IL-2 Signaling INTRODUCTION The family of Janus kinases (JAK1, JAK2, JAK3, and TYK2) are key mediators of cytokine signaling that regulate hematopoietic cell development, cellular metabolism and immune responses. Within the JAK family, JAK3 is unique in its restricted expression to hematopoietic lineages and specific interaction only with the common γchain (γc) cytokine receptor (1). In cytokine signaling, JAK3 functions in a heterodimeric pair with JAK1, which binds to the cytokine specific receptor βchains. The JAK1-JAK3 pair transmit signals emanating from interleukin (IL)−2, −4, −7, −9, −15, and −21 that regulate development and activation of lymphoid lineage cells (2). The critical role of JAK3 in lymphoid cells is highlighted by the phenotypes of JAK3 activating mutations causing different types of leukemia and lymphomas (3–5), as well as by JAK3 deficiencies that cause severe combined immunodeficiency (SCID); a condition resulting from profound defects in mature T cells and B cells and innate lymphoid cells including NK cells (6–8). JAKs share a conserved domain structure consisting of an N-terminal FERM domain and a Src homology 2 (SH2)- like domain that comprise the receptor-binding unit, followed by a pseudokinase domain (JAK homology 2, JH2) and a tyrosine kinase domain (JH1) (1,9,10). Signaling is initiated by cytokine binding to the extracellular domain of their respective receptors that induces receptor dimerization/oligomerization or a conformational shift juxtaposing the JAKs and allowing their activation through trans-autophosphorylation of the activation loop tyrosine residues. Following activation, JAKs phosphorylate tyrosine residues in the receptor chains leading to recruitment and subsequent phosphorylation of downstream effectors such as the signal transducers and activators of transcription (STATs) (1,2,10). In IL-2 receptor (IL-2R) signaling, JAK1 has been shown to play a dominant role in early cytoplasmic activation events where JAK1 is primarily driving STAT5 phosphorylation (11), while JAK3 is required for sustained activation during T-cell proliferation (11,12). Both functions are dependent on the catalytic activity of the JH1 domains in JAK1 and JAK3, respectively. However, several aspects of IL-2R signal regulation are still elusive, especially with respect to the roles of the pseudokinase domains. JAK kinases are constitutively associated with cytokine receptors and the activity of the JAK–STAT pathway needs to be tightly regulated at multiple levels to suppress signaling in the absence of cytokine stimulation and to allow rapid, transient activation upon stimulation (9,10). Regulation relies primarily on control of the JH1 tyrosine kinase activity through intraand intermolecular mechanisms (1,9), and for the former, the function of JH2 is important, although still somewhat enigmatic. Genetic studies provide compelling evidence for the regulatory function of JH2 as disease-causing mutations strongly cluster in JAK JH2s (13,14). The best characterized example is JAK2 where somatic mutations in JH2 (JAK2 V617F and >30 other mutations) result in constitutively active JAK2 and are responsible for approximately 80% of myeloproliferative neoplasms (MPN) and less frequently for different types of leukaemias including acute lymphoblastic leukemia (ALL), acute megakaryoblastic leukemia (AMKL), and acute myeloid leukemia (AML) (5,14,15). In JAK3, approximately half of all clinical mutations localize to JH2 (13,16). JAK3 JH2 has two interesting and unique features compared to the other JAKs: it is the only JH2 harboring both activating and inactivating clinical mutations, and secondly it is the only JH2 where the prototypic JH2 gain-of-function mutation (JAK2 V617F, JAK3 homolog M592F) does not cause hyperactivation (17). Some inhibitory SCID mutations result in truncated proteins, but several mutations cause amino acid changes in either the Cand N-lobes of JH2 (e.g., E481G, del482596, R582W, G589S, C759R). More recently, several activating JAK3 JH2 mutations (e.g., M511I, A572V, R657Q, V722I) have been found in acute myeloid and lymphoblastic leukemia (AMKL, T-ALL, AML, natural killer cell lymphoma, acute megakaryoblastic leukemia, T-cell prolymphocytic leukemia, juvenile myelomonocytic leukemia and natural killer T cell lymphoma) (3,4,18). Although genetic information from patients provide compelling evidence for the importance of JH2 regulatory function, they do not provide molecular explanations of the underlying mechanisms of normal or pathogenic JH2 function. JAK JH2s show significant sequence homology with classical protein kinases but several conserved functional residues or motifs are missing or altered. Most importantly, the HRD motif, including the catalytic base aspartate (D) is replaced by HGN resulting in a presumably catalytically inactive kinase (hence “pseudokinase”) (19–21). Structural and biochemical analysis of JAK1, JAK2 and TYK2 JH2s have shown that despite the sequence variations, the JH2 domains adopt a typical two lobed kinase-fold able to bind ATP (22–24). JAK2 and TYK2 JH2 structures have been solved with ATP and they show a noncanonical binding mode with ATP complexed to only a single cation instead of the usual two (23,25). In addition, the canonical ionic interaction between the β3 lysine (K556 in JAK3 JH2; K72 in the archetypal cAMP-dependent protein kinase PKA) and αC (E91 in PKA, alanine in JAK JH2s) is replaced by a β3 lysine– aspartate interaction to the DFG motif (DPG in JAK JH2s) (19,23,25–27). Early biochemical studies suggested a dual regulatory or “switch” function for JH2 where the domain is required for maintaining JAKs inactive in the absence cytokine stimulation, as well as mediating cytokine-induced signaling (27–31). Recent structural information from JAK2 (structural model) and TYK2 (crystal structure) provide molecular basis for the JH2 inhibitory function and show that JH2 interacts with the hinge-region of JH1 and stabilizes the JH1 in an inactive conformation (27,32). Most gain-of-function JAK2 disease mutations localize in this interface and are predicted to disrupt the JH1–JH2 autoinhibitory interaction resulting in constitutive activation (27,28,33). The nature of the activating function of JH2 in cytokine signaling and in oncogenic JAK3 mutations is currently unknown. MATERIALS AND METHODS Protein Expression and Purification Expression: JAK3 JH2 (residues 511–790) was cloned into the pFASTBAC1 vector (Invitrogen) with a C-terminal His6tag and expressed as a fusion protein in Spodoptera frugiperda (Sf 9) cells. Frontiers in Oncology | www.frontiersin.org 2December 2018 | Volume 8 | Article 560 Raivola et al. JAK3 JH2 Regulates IL-2 Signaling For protein expression, cells were infected with 10% (v/v) virus supernatant, grown for 48 h and collected by centrifugation. Ni-NTA purification: Cell pellets containing JH2-His fusion protein were resuspended in lysis buffer containing 20 mM TrisHCl (pH 8.0), 500 mM NaCl, 20% (v/v) glycerol, 0.5 mM TCEP and 20 mM imidazole, supplemented with protease inhibitors cocktail, lysed using a cell disruptor (Sonics) and clarified by centrifugation for 1 h at 10,000 g. The supernatant was incubated for 2 h with prewashed Ni-NTA beads (Qiagen) with gentle rotation at 4 ◦C. The beads were extensively washed with a buffer supplemented with 40 mM imidazole, and the fusion protein was eluted with 150 mM imidazole buffer. Fractions containing the fusion protein were pooled and buffer-exchanged by several dilution-concentration cycles using 10 K Amicon Ultra centrifugal filters. Size exclusion chromatography: Ni-NTA purified, bufferexchanged and concentrated proteins were loaded into Superdex 75 10/300 GL column (GE Healthcare Life Sciences) equilibrated with 20 mM Tris-HCl (pH 8.0), 500 mM NaCl, 20% (v/v) glycerol and 0.5 mM TCEP. The fractions including the protein were collected and concentrated for further analysis. Differential Scanning Fluorimetry Protein concentration of 3 µM was used for the measurement of the melting temperature (Tm). Reactions were done in the buffer used in the SEC. Addition of other components (ATP, salts) were taken into account by including 2X buffer, if needed. Signal arising from the fluorescent dye Life Technologies SYPROTM Orange Protein Gel Stain, 4X working solution from 5000X concentrate in DMSO) was measured with a conventional RealTime PCR system (Bio-Rad CFX), 1◦C per min from 4 to 96◦C. Results were transferred and normalized in Microsoft Excel and the Tm was derived from the sigmoidal function using GraphPad Prism version 5.02 for Windows, GraphPad Software, La Jolla California USA, https://www.graphpad.com/. MANT-ATP Binding Assay The steady-state fluorescence of MANT-labeled ATP was measured as described in Mysore et al. (34). Briefly, the fluorescently labeled MANT (2′/3′-(N-methyl-anthraniloyl)- ATP was titrated against 1.5 µM recombinant JAK3 JH2. In addition, the protein was titrated as 0, 0.05, 0.25, 1.25, and 6.26 µM against 0.25 µM MANT-ATP. FRET signal was measured with QuantaMaster PTI Fluorometer and the data was analyzed using GraphPad Prism version 5.02 and Microsoft Excel. Protein titrations were done in duplicate and in MANTATP titration in triplicate. Fluorescence Polarization Assay Fluorescence polarization measurements were performed in black 384-well plates (ProxiPlate-384 F Plus, PerkinElmer) at sample volume of 5 µl/well on PerkinElmer Envision plate reader using 480 nm excitation and 535 nm emission filters. Fluorescent tracer, Bodipy FL labeled JNJ-7706621 [compound 5 in (35)] was used at 1.5 nM concentration. Recombinant JAK1 JH2 553-836His, JAK2 JH2 503-827-His, JAK3 JH2 511-790-His, and TYK2 JH2 564-876-His proteins were used at concentrations (2 nM JAK1, 150 nM JAK2, 1 µM JAK3, 20 nM TYK2) dependent on protein-tracer dissociation constants. The assays were performed in a buffer consisting of 20 mM Tris-HCl pH 8.0, 150 mM NaCl, 20% glycerol, 0.01% Brij-35, and 2 mM DTT. ATP was titrated at concentration range of 5 nM−5µM and fluorescence polarization values obtained were fitted against log[ATP] in GraphPad Prism to yield IC50 values. Assays were performed in triplicate. Plasmids and Mutagenesis C-terminally HA-tagged full length JAK1, JAK3, and STAT5 cloned into pCIneo mammalian expression plasmid were used for transfections. Mutations into JAK1 and JAK3 were introduced by site-directed mutagenesis using QuikChange-protocol with specifically designed primers. The method was also used for the domain deletion constructs. Cell Culture, Transfection, and Immunoblotting JAK1 and JAK3 deficient U4C fibrosarcoma cells stably expressing IL-2Rγand -βreceptors were kindly provided by Dr. Claude Haan, (Life Sciences Research Unit, Signal Transduction Laboratory, University of Luxembourg, Luxembourg, L-1511, Luxembourg) and Hans-Günter Zerwes (Novartis Institute for Biomedical Research). Cells were cultured in Dulbecco’s Modified Eagle’s Medium (DMEM) supplemented with 10% fetal bovine serum, 1% pen strep and 1% glutamine and antibiotics: puromycin, zeozin and blasticidin for maintaining the IL-2Rγ and -βexpression. Transient transfections were done with full-length or mutant human JAK1-HA (75 ng), human JAK3-HA (75 ng) and human STAT5A-HA (1.5 ng) using FuGENE HD (Promega) according to the manufacturers’ instructions. After 34 h cells were starved in serum-free medium overnight and stimulated for 15 min with 100 ng/ml of human IL-2. After stimulation, cells were lysed into Triton-X cell lysis buffer and centrifuged, and the supernatant used for SDS/PAGE and immunoblotting. Blots were double-stained with phosphospecific antibodies (PStat5 (Y694), Cell Signaling Technology and anti-phospho-JAK1 (Tyr1022/Tyr1023), Millipore) and anti-HA (Aviva Systems Biology OAEA00009) and detected with a mix of IRDyelabeled secondaries. The STAT5 (HA) signals were used for normalization of phospho-STAT5 levels. Blots were quantified using a LI-COR Odyssey CLx imaging system. A minimum of three independent experiments were performed for each condition. JAK3 Homology Modeling A JAK3 homology model was generated using the SWISSMODEL server with the TYK2 structure (Protein Data Bank identification 4OLI) as the modeling template. Graphical presentations were generated using the PyMOL Molecular Graphics System (DeLano Scientific, San Carlos, CA). Luciferase and Dual Luciferase Assays STAT5 transcriptional activity was assessed by measuring the luciferase expression (SPI Luc 2) driven by a STAT5 Frontiers in Oncology | www.frontiersin.org 3December 2018 | Volume 8 | Article 560 Raivola et al. JAK3 JH2 Regulates IL-2 Signaling responsive promoter from growth hormone regulated serine protease inhibitor, as described previously (27). U4Cγβ cells were transfected with indicated DNA constructs including SPILuc2 and β-galactosidase reporter plasmids. The latter was co-transfected as an internal transfection control. Transient transfections were done in 96 well plates with FuGENE HD (Promega) according to manufactures instructions. 42 h after transfection, cells were stimulated (in starvation media) or starved for 5 h after which luciferase assays were analyzed using the dual luciferase reporter assay system (Promega) according to manufactures instructions. Luciferase values were measured with EnVision 2104 Multilabel Reader (Perkin Elmer). The results are presented as relative luciferase activity (RLU) corresponding to the firefly luciferase light emission values divided by renilla luciferase light emission values. In addition to Renilla luciferase (pRLTk; Promega), the βgalactosidase reporter was also used, and similar results were obtained with both reporters. Briefly, cells were washed twice with PBS and 20 µl of 1X Luciferase lysis buffer was added to the wells. 10 µl of the lysate was used for luminescence measurement with 75 µl luciferase reagent. β-galactosidase signal was measured by adding 50 µl of ONPG per well and the absorbance signal values were used for normalization of the luciferase luminescence values. In vivo Studies in Zebrafish Embryos derived from jak3+/−in-crosses were injected with 100 pg/nl in vitro transcribed capped mRNA encoding zebrafish JAK3 wild-type (WT), and M511I, I535F and M511+I535F mutants. At 5 dpf they were fixed and subjected to whole-mount in situ hybridization (WISH) with anti-sense rag1 probe and imaged and quantified as described (36), with only jak3−/−embryos, as determined by post-WISH genotyping, used in the analysis. In other experiments, embryos from lck::GFP zebrafish were injected and rag1 expression analyzed using RT-PCR. RESULTS Both JH1 and JH2 in JAK1 and JAK3 Are Required for Functional IL-2–STAT5 Signaling In order to investigate the functional role of JAK3 JH2 in IL2R signaling, we used the human fibrosarcoma U4Cγβ cell line that stably expresses IL-2Rγand IL-2Rβbut lacks expression of JAK1 and JAK3 (11). A functional IL-2R signaling cascade was reconstituted by ectopic expression of JAK1 and JAK3, which allowed analysis of JAK1/JAK3 mutants. Optimization of experimental conditions showed that expression of JAK1 alone induced strong basal activation of STAT5-driven luciferase reporter SPI-Luc2 (27), which was, however, unresponsive to IL2 stimulation (Supplementary Figure 1A). Expression of JAK3 alone showed no basal STAT5 transcriptional activity and no responsiveness to IL-2. IL-2-induced STAT5 activation was achieved only by expression of both JAK1 and JAK3, which is in line with previous studies (3,11,37). Immunoblot analysis of STAT5 Y694 phosphorylation (pSTAT5) from U4Cγβ cells transfected with JAK1 and JAK3 concurred with STAT5 transcriptional activity showing IL-2independent phosphorylation by JAK1 alone, lack of pSTAT5 by JAK3 alone and requirement for both JAK1 and JAK3 for IL-2-responsiveness (Supplementary Figure 1B). Collectively these results imply JAK1 as the immediate downstream effector kinase and JAK3 as a regulator/activator in the IL-2R signaling complex, thus confirming previous reports (3,11,37). However, the underlying molecular mechanisms and regulatory domains in JAK1–JAK3 heterodimer signaling are still largely unknown and, in this context, we wanted to analyze specifically the roles of JH2s. In order to investigate the roles of different JH domains we produced deletion constructs for JH1 and JH2 domains in JAK1 and JAK3 (Figure 1) In accordance with the critical role of JAK1 kinase activity in IL-2R–STAT5 signaling (see above), deletion of JH1 in JAK1 (JAK1 1JH1) abolished both basal and IL-2induced pSTAT5 and STAT5 reporter activity. Similarly, deletion of JH1 in JAK3 abolished IL-2-inducibility but retained basal activity mediated by JAK1. Interestingly, deletion of JH2 in JAK1 also completely abrogated IL-2R signaling activity, and in JAK3 deletion of JH2 or both JH1 and JH2 prevented IL-2-inducibility without affecting basal activity (Figures 1B,C). These results indicate that, in addition to the kinase domains, the pseudokinase domains in JAK1 and JAK3 are also required for IL-2R signaling. Next, we extended our studies on the regulatory function of JH2 to clinical JAK3 mutations. As the structure of JAK3 JH2 has not been solved, we produced a homology model based on the TYK2 JH2-JH1 structure (Protein Data Bank (PDB) code: 4OLI, see Figure 2A) (27). We analyzed pathogenic inactivating mutant C759R found in SCID patients as well as activating JAK3 mutants M511I and R657Q in JH2 and L857Q in JH1 (list of mutations; see Table 1). The activating mutants are found in TALL patients and reported to be constitutively active in cellular assays and result in leukemic phenotypes in mouse models (3). Mutations were studied in the IL-2–STAT5 pathway in U4Cγβ cells co-expressed with wild-type JAK1. The JAK3 C759R inhibited basal and IL-2-induced STAT5 activation as efficiently as the kinase-inactive JAK3 JH1 mutation K855A (Figure 2B). JAK3 M511I and R657Q showed similar, cytokine-independent STAT5 activation as the JH1 L857Q mutant (Figures 2B,C, Supplementary Figure 2C). Most activating JAK3 mutants have been shown to depend on JAK1 for signaling (3), and we sought to define the JAK1 domains required for constitutive IL-2 signaling. Activating JAK3 mutants were transfected with either wild-type JAK1, JAK1 1JH2 or JAK1 1JH1 into U4Cγβ cells. All tested JAK3 mutants residing in either JH1 or JH2 were found to depend on normal JAK1 activity and showed increased basal STAT5 activation only in the presence of full-length JAK1 (Figure 2B,Supplementary Figure 2A); deletion of either JH1 or JH2 of JAK1 completely abrogated signaling as measured by pSTAT5 and STAT5 luciferase assays (Figure 2C, Supplementary Figures 2B,C). Taken together, these data show that the analyzed oncogenic JAK3 JH2 and JH1 mutations require both JH1 and JH2 domains of JAK1 for constitutive signaling. Frontiers in Oncology | www.frontiersin.org 4December 2018 | Volume 8 | Article 560 Raivola et al. JAK3 JH2 Regulates IL-2 Signaling FIGURE 1 | Both JH1 and JH2 in JAK1 and JAK3 are required for functional IL-2 signaling. (A) Schematic presentation of JAK1 and JAK3 wild-type (WT) and domain deletion constructs. (B) Both JH1 and JH2 of JAK1 and JAK3 are required for IL-2 dependent STAT5 activation. U4Cγβ cells were transfected with JAK1 WT, JAK3 WT and domain deletion constructs together with STAT5 responsive reporter SPI-luc2 (28), either with Renilla or pRL-TK control vector (see Materials and Methods) and starved/stimulated. Similar results were obtained with both control vectors. Data shows averages and SD from normalized values combined from three independent experiments, each having triplicate samples. The expression of JAK1/JAK3 constructs was analyzed by anti-HA Western blotting. (C) IL-2-induced pSTAT5 requires both JH1 and JH2 in JAK1 and JAK3. STAT5 phosphorylation and expression of JAK1, JAK3 and STAT5 were detected from cell lysates from transfected U4Cγβ cells by immunoblotting with HAand pTyr694 STAT5 antibodies. Similar results were obtained in three independent experiments. Characterization of Recombinant JAK3 JH2 Shows a Domain Capable of Tight, Cation-Independent Nucleotide Binding Pseudokinases are generally considered to be inactive kinases but an estimated 30–40% of pseudokinases, including JAK1, JAK2 and TYK2 JH2, bind nucleotides (21). The function of JAK3 JH2 has not been analyzed and for biochemical characterization recombinant JAK3 JH2 (residues 511–790 with C-terminal His6tag) was expressed in insect cells and purified with affinity chromatography followed by size-exclusion chromatography (Supplementary Figures 3A,B). The protein was analyzed by differential scanning fluorimetry (DSF) for its ability to bind adenosine nucleotides (Figure 3A). The melting temperature (Tm) of JAK3 JH2 (apo) was as low as 32.4 ±0.2◦C, indicating inherent lability of the domain but Tm was significantly increased by addition of ATP (1Tm 8◦C with 0.2 mM ATP, 12◦C with 1 mM ATP). Supplementing the buffer with either Mg2+, Mn2+ or Ca2+(in the form of 2 mM of their chloride salts) alone or with ATP did not have a significant effect on the Tm. Adding the monovalent cation salt KCl did not affect the Tm, which was expected as the buffer contained monovalent NaCl. As other JAK JH2s have been shown to require salts to bind ATP, we tested all JAK JH2s with 500 µM ATP with, or without MgCl2(Supplementary Figure 4A) and confirmed, that the Tm of JAK1, JAK2 and TYK2 JH2s increase only if MgCl2is added with ATP, while JAK3 JH2 has higher 1Tm without Mg2+(ATP only). The cation-independency of ATP binding was confirmed by adding 2 mM of the chelating agent EDTA to the buffer, which did not affect the Tm for JAK3 JH2 ATP (Figure 3B). Measuring the affinity of JAK3 JH2 binding to the fluorescent ATP-analog MANT [2′-(3′)-O-(N-methylanthraniloyl)]-ATP Frontiers in Oncology | www.frontiersin.org 5December 2018 | Volume 8 | Article 560 Raivola et al. JAK3 JH2 Regulates IL-2 Signaling FIGURE 2 | Mutations in JAK3 JH2 can alter the function of the IL-2 pathway. (A) Localizations of the JAK3 mutants used in this article. Homology model of JAK3 JH2-JH1 (based on the TYK2 structure in Protein Data Bank (PDB) code: 4OLI) and the JAK3 JH1 structure [PDB code: 1YVJ (38)] were aligned with the JH1-JH2 structure of TYK2 (PDB code: 4OLI). JH2 is shown in red and JH1 in gray. Close-ups of the mutations are shown in subpanels. C759 is shown in the main picture. In the lower left panel, the (possible) interaction pair R657–D864 (in JH1) is shown. In the right panel, ATP was aligned from ATP-bound form of PKA [PDB code: 1ATP (39)]. (B) JAK3 JH2 C759R mutation efficiently inhibits IL-2 signaling and activating mutations in JAK3 JH2 and JH1. Luciferase assay of STAT5 activation in U4Cγβ cells transfected with JAK1 WT and JAK3 point mutations, SPI-luc2 reporter and β-galactosidase control. Values were normalized to JAK1 WT and JAK3 WT transfected sample Errors bars show SD of triplicate samples. Similar results were obtained using pRL-TK as a control. (C) Activating JAK3 mutants rely on both JH1 and JH2 of JAK1 for constitutive downstream signaling. SPI-luc2 reporter was used to detect the activation of STAT5 in U4Cγβ cells. Cells were transfected with WT or domain deletion JAK1 and WT or constitutively active JAK3 mutants. β-galactosidase plasmid was used as an internal control for the transfection efficiency. Values were normalized to WT JAK1 and JAK3. Error bars are SD of triplicate samples. Frontiers in Oncology | www.frontiersin.org 6December 2018 | Volume 8 | Article 560 Raivola et al. JAK3 JH2 Regulates IL-2 Signaling TABLE 1 | Mutations used in this study. JAK Mutation Domain Effect References JAK3 M511I JH2 Found in T-ALL patients, constitutively active (3) R657Q JH2 Found in T-ALL patients, constitutively active (3) L857Q JH1 Found in T-ALL patients, constitutively active (3) C759R JH2 Found in SCID patients, inhibits JAK3 kinase activity and reduces STAT5 activity (31) K855A JH1 Catalytic lysine, kinase dead K566A JH2 Homologous to K855A, disrupts ATP binding to JH2 I535F JH2 Homologous to JAK2 I559F, disrupts ATP binding to JH2 (24) JAK1 V658F JH2 Homologous to JAK2 V617F, constitutively active (17) I535F JH2 Homologous to JAK2 I559F, disrupts ATP binding to JH2 (24) FIGURE 3 | Nucleotide binding to recombinant JAK3 JH2. (A) JAK3 JH2 binds ATP without cations. The effect of ATP on Tm of JAK3 JH2 is shown in the left panel. The following graph shows changes in the melting temperature (1Tm) in JAK3 JH2 in the presence of different cations and ATP. (B) Chelating agent does not affect ATP-induced increase on Tm of JAK3 JH2. DSF analysis performed in the presence of ATP (1 mM) and chelating agent EDTA (2 mM). (C) MANT-ATP binding assay with recombinant JAK3 JH2. Both protein and MANT-ATP were titrated and both experiments gave similar Kd in the sub-micromolar range. Left panel: titrating MANT-ATP to 1.5 µM JAK3-JH2. Kd range 330–440 nM. Right panel: titrating JAK3 JH2 to 0.25 µM MANT-ATP, Kd 500 nM. Protein concentrations were 0; 0.05; 0.25; 1.25 and 6.25 µM. In both figures, the Y-axis is the fraction of ligand-bound protein in relation to the total protein amount ([PL/Ptot]). The MANT-ATP titration was repeated three times and all experiments are plotted in the lower left panel. Protein titration was repeated twice and both experiments are blotted in the figure. showed that JAK3 JH2 binds MANT-ATP with a dissociation constant (Kd) of 0.4 µM (Figure 3C). Similar values were obtained by titration of the ligand to a constant amount of protein and vice versa. IC50 values was determined using fluorescent polarization assay (FP), by competition assay, where ATP competes with ATP pocket binding tracer. The IC50 values for JAK JH2s were compared, and the IC 50 for JAK3 JH2 was 16 µM (without MgCl2), indicating tighter ATP-binding affinity for JAK3 JH2 compared to other JAK JH2s (Supplementary Figure 4B,Table 2). Approximation of the JAK3 JH2 Kd was further with DSF, where 50 nM−1 mM ATP was titrated to 3 µM protein. The value was higher than measured with MANT-ATP, namely 40.4 ± 97.1 µM (Supplementary Figure 4C). However, the variation was wide and, as there was no saturation seen when the values were fitted, no Kd was obtained for other JAK JH2s Frontiers in Oncology | www.frontiersin.org 7December 2018 | Volume 8 | Article 560 Raivola et al. JAK3 JH2 Regulates IL-2 Signaling (Supplementary Figure 4C). The differences in the magnitude of the Kd and IC50 values stems from the methods used to evaluate the affinity, e.g., the fact that MANT-ATP uses modified ATP that may affect the binding properties of the molecule. However, both the MANT-ATP and FP methods showed that JAK3 JH2 binds ATP tighter compared to other JAK JH2s, and the binding is cation-independent. Furthermore, the effect of ATP binding to the stability of JAK3 JH2 was greater compared to other JAK JH2s (Supplementary Figure 4C,Table 2). Finally, we investigated the binding of ADP and AMP to JAK3 JH2 (Supplementary Figure 4D). ADP showed a significantly reduced 1Tm compared to ATP, and AMP showed no effect on Tm indicating that JAK3 JH2 does not bind AMP and prefers ATP over its de-phosphorylated counterparts. An autophosphorylation assay with radiolabeled ATP did not show incorporation of 32P into JAK3 JH2, suggesting that JAK3 JH2 does not possess catalytic activity (Supplementary Figure 4E). Mutating the JAK3 JH2 ATP-Binding Site Reverts the Oncogenic JAK3 Hyperactivation Given that JAK3 JH2 has a functional nucleotide-binding site, we wanted to analyze the functional role of ATP binding to JAK3 JH2 in IL-2 signaling. The JAK3 JH2 model identified two residues in the ATP-binding pocket conserved between JAKs (Figure 2A) and suited for mutagenesis. We thus mutated I535 in the β2-sheet of the JH2 N-lobe to a bulkier phenylalanine to sterically inhibit ATP binding. The corresponding JAK2 mutation (JAK2 I559F) has been shown to inhibit ATP binding without affecting the stability of JAK2 JH2 (24). We also mutated the conserved JAK3 JH2 β3 lysine K556, to alanine. The activities of the mutant JAK3 were analyzed in U4Cγβ cells with STAT5-responsive reporter and pSTAT5 analysis. Neither mutation disrupted responsiveness to IL-2 stimulation and the signaling properties resembled wildtype JAK3 (Figure 4). Inhibition of ATP binding to JH2 by mutations or small molecular weight (smw) inhibitors has been demonstrated to abrogate pathogenic hyperactivation of JAK2 and cytokine signaling via TYK2, respectively (23,24). To assess the effect of JH2 nucleotide binding in constitutively active JAK3, we constructed double mutants consisting of an activating mutant (M511I, R657Q, or L857Q; see Figure 2) and either of the ATP binding mutants (I535F or K556A). The constructs were transiently transfected and analyzed for STAT5 activation (SPIluc2 reporter) and pSTAT5 (immunoblotting) (Figures 4A,B,D, Supplementary Figure 5A). In both assays the JH2 ATP-binding site mutations reduced STAT5 activation back to or below wildtype JAK3 levels while retaining cytokine-responsiveness. The sole exception was JAK3 L857Q+K556A, which showed strongly reduced activation by cytokine as well. The more conservative sterically hindering ATP-binding site mutation I535F, however, resulted in wild-type-like STAT5 activation in the L857Q context. IL-2-responsiveness was also regained in M511I+I535F/K556A and R657Q+I535F/K556A (Figures 4A,B). Notably, these results are distinct from complete removal of JAK3 JH2 (1JH2, see Figure 1) or removal of JH1 kinase activity by the JAK3 JH1 β3 lysine-to-alanine mutation JAK3 K855A, which resulted in unresponsiveness to cytokine, even when combined with the activating M511I mutation (Figure 4D). In order to investigate the mechanism of suppression in JAK3 oncogenic signaling by the JH2 ATP-binding mutants, we analyzed phosphorylation of the JAK1 activation loop tyrosine residues. The JH2 ATP binding mutation in JAK3 double mutants reduced JAK1 activation loop phosphorylation compared to activating mutations alone (Supplementary Figure 5C). The effect of JH2 ATP binding on JAK3 oncogenic mutants was also analyzed using zebrafish (Danio rerio). Expression of rag1, an IL-2Rγc/JAK3-dependent marker for mature lymphoid cells, was analyzed in JAK3-deficient embryos injected with mRNA encoding JAK3 WT, M511I, I535F or the combined M511I +I535F mutant mRNA as described (36). Each produced a range of rag1 expression, but both M511I and I535F elicited a small but reproducible enhancement in the mean rag1 expression compared to WT, which was significantly reduced in the ATP binding deficient M511I +I535F double mutant (Supplementary Figure 6). A similar reduction in rag1 expression was observed with the double mutant compared to M511I in qPCR experiments performed on embryos from transgenic lck::GFP zebrafish injected with JAK3 WT, M511I and M511I +I535F mRNA (data not shown). In contrast, the presence of the I535F mutation did not ablate the induction of rag1 expression by JAK3 WT in either case (Supplementary Figure 6 and data not shown). Together, the data indicate that disrupting ATP binding to JAK3 JH2 preserves IL-2-responsiveness but decreases constitutive STAT5 activation, probably by reducing the activating potential of JAK3 to JAK1. JAK3 JH2 ATP-Binding Mutations Do Not Inhibit Activation by Pathogenic JAK1 Mutations We have previously shown that JH2 ATP-binding mutations suppress activating mutations in homomeric cytokine receptor signaling in the same JAK molecule (24). Using the IL-2– JAK1/JAK3 signaling system, we wanted to study the effect of JH2 ATP binding in heterodimeric JAK signaling by assessing whether inhibition of ATP binding in one JAK JH2 could suppress activation by hyperactivating mutations in the other JAK in trans. We co-transfected JAK1 JH2 ATP mutant I597F (corresponding to JAK3 I535F) with JAK3 activating mutation M511I, or JAK3 JH2 I535F mutant with the activating JAK1 T-ALL-associated mutation V658F (analogous to the most prevalent MPN-inducing JAK2 mutation V617F) (40). Similar amounts of wild-type and mutated JAK3 and JAK1 proteins were expressed as quantified by western blotting (Supplementary Figure 5B). JAK1 JH2 mutant I597F, when co-transfected with JAK3 M511I, did not reduce (hyper)activation and thus did not restore cytokine responsiveness (Figure 5A). In the reverse experimental setting, the JH2 ATP pocket mutation JAK3 I535F could not reduce the activation of constitutively active JAK1 V658F. Together these data suggest that JH2 ATP-site mutations exert their direct suppressing function in cis and cannot suppress Frontiers in Oncology | www.frontiersin.org 8December 2018 | Volume 8 | Article 560 Raivola et al. JAK3 JH2 Regulates IL-2 Signaling FIGURE 4 | Inhibition of ATP binding to the pseudokinase domain of JAK3 reverts pathogenic, ligand-independent JAK3 signaling, while retaining cytokine-responsiveness. (A) JAK3 JH2 ATP-binding site mutations suppress activating JAK3 mutants. Luciferase assay was performed in U4Cγβ cells transfected with JAK1 and the indicated JAK3 mutants using β-galactosidase as a transfection control. The results are from three independent experiment, each having triplicate samples (n=9, expect for K556A n=7). Similar results were obtained with pRL-TK control. Data is normalized to the basal wild-type sample, errors are presented as SD. Statistical analysis of the activating mutations, and their corresponding double mutants were done with two-tailed t-test using unequal variances. **p<0.001. (B) Representative Western blot of the transiently transfected U4Cγβ cell lysates showing the pSTAT5 signal. HA-tagged STAT5 was co-transfected with HA-tagged JAK1 and JAK3 and the lysates were blotted with anti-HA and anti-pSTAT5. Lower panel: Quantitative pSTAT5/STAT5 signal analysis. Activation of JAK3 signaling by M511I, R657Q and L857Q is significant (*p<0.05). Data is average of five experiments. (C) The localization of the mutated JAK3 JH2 ATP-binding residues. Homology model of JAK3 JH2 (based on TYK2 structure from PDB, code: 4OLI), ATP is shown in gray. (D) Kinase-inactive JAK3 K855A abolishes IL-2 signaling in the presence of the activating JAK3 mutation M511I. Luciferase assay with STAT5-responsive promoter and with pRL-TK control. Frontiers in Oncology | www.frontiersin.org 9December 2018 | Volume 8 | Article 560