Dimerization by a Cytokine Receptor Is Necessary for Constitutive Activation of JAK2V617F*

The majority of the BCR-ABL-negative myeloproliferative disorders express the mutant JAK2, JAK2V617F. Previously we showed that constitutive activation of this oncogenic JAK2 mutant in Ba/F3 or 32D cells requires coexpression of a cognate homodimeric cytokine receptor, such as the EpoR. However, overexpression of JAK2V617F in Ba/F3 cells renders them cytokine-independent for growth in the absence of an exogenous cytokine receptor. Here, we demonstrated that JAK2V617F domains required for receptor association are essential for cytokine-independent growth by overexpressed JAK2V617F, suggesting JAK2V617F is binding to an unknown endogenous cytokine receptor(s) for its activation. We further showed that disruption of EpoR dimerization by coexpressing a truncated EpoR disrupted JAK2V617F-mediated transformation, indicating that EpoR dimerization plays an essential role in the activation of JAK2V617F. Interestingly, coexpression of JAK2V617F with EpoR mutants that retain JAK2 binding but are defective in mediating Epo-dependent JAK2 activation due to mutations in a conserved juxtamembrane motif does lead to cytokine-independent activation of JAK2V617F. Overall, these findings confirm that JAK2V617F requires binding to a dimerized cytokine receptor for its activation, and that the key EpoR juxtamembrane regulatory motif essential for Epo-dependent JAK2 activation is not essential for the activation of JAK2V617F. The structure of the activated JAK2V617F is thus likely to be different from that of the activated wild-type JAK2, raising the possibility of developing a specifically targeted therapy for myeloproliferative disorders.

The Janus family of non-receptor tyrosine kinases are key mediators of cytokine receptor signaling and play a central role in hematopoiesis and immune responses. In mammals there are four Janus kinases (JAKs), 2 JAK1, JAK2, JAK3, and Tyk2. JAKs contain seven highly conserved domains named JAK homology (JH) domains 1-7. The C-terminal JH1 domain is a functional tyrosine kinase that becomes activated following cytokine stimulation (1). The JH2 pseudokinase domain negatively regulates the activity of JAKs through an interaction with the JH1 kinase domain (2). The N-terminal JH3 to JH7 domains mediate their association with the membrane-proximal regions of cytokine receptors (3), and may also play important regulatory roles (4).
JAK2 is essential for signaling by receptors for growth hormone, prolactin, erythropoietin, thrombopoietin, interleukin-3, and interleukin-5; major downstream signal transduction effectors are the STAT5, Ras/MAPK, and PI3K/AKT pathways (1). Disregulation of JAK2 has been implicated in several hematological malignancies (5). For example, TEL-JAK2, a fusion between the dimerization domain of the Ets family transcription factor TEL and the catalytic domain of JAK2, has been identified in patients with acute lymphoblastic leukemia and atypical chronic myelogenous leukemia (6,7). In TEL-JAK2, the oligomerization domain of TEL mediates dimerization of the JAK2 kinase domain, resulting in the constitutive activation of the kinase and its downstream targets, and leading to cytokineindependent proliferation of the normally interleukin-3dependent Ba/F3 hematopoietic cell line (6,7).
In 2005, several groups reported an acquired point mutation in the JAK2 JH2 domain, V617F, in most patients with polycythemia vera (PV) and about half of those with idiopathic myelofibrosis (IMF) and essential thrombocythemia (ET) (8 -11). JAK2V617F represents the major molecular lesion in patients with BCR/ABL-negative myeloproliferative disorders (MPD). Subsequent large-scale screenings of patients with various blood disorders also found the JAK2V617F mutation in a relatively low percentage of cases of chronic myelomonocytic leukemia, myelodysplastic syndrome, and acute myeloid leukemia (12)(13)(14)(15)(16). Further studies have also identified new mutations adjacent to the JH2 domain of JAK2 in PV patients (17), and additional mutations in the JH2 domain of JAK2 and JAK3 in established acute megakaryoblastic leukemia cell lines (18,19). Similar to JAK2V617F, these mutations lead to constitutive JAK activation and cytokine-independent growth of hematopoietic cell lines. The JH2 pseudokinase domain represents a key negative regulator of JAK kinase activity, and its disruption has been increasingly recognized as a cause of many myeloproliferative disorders and other hematopoietic malignancies.
Using the IL-3-dependent Ba/F3 and 32D cell lines, we showed that expression of JAK2V617F confers factor-independent growth only in cells coexpressing homodimeric type 1 cytokine receptors such as the erythropoietin receptor (EpoR), thrombopoietin receptor, or granulocyte colony-stimulating factor receptor. Not surprisingly, coexpression of these cytokine receptors is also required for the constitutive phosphorylation and thus activation of JAK2 and STAT5. By coexpressing JAK2V617F together with EpoR mutants, we demonstrated that EpoR provides an essential scaffold for JAK2V617F; we hypothesized that two molecules of JAK2V617F, each bound to the cytoplasmic domain of a homodimeric EpoR, phosphorylates and transactivates each other. Furthermore, (phospho-)tyrosines in the cytosolic domain of the EpoR are essential for the phosphorylation of STAT5 through activated JAK2V617F and for the maximal Epo-independent cell proliferation (20). We concluded that JAK2 V617F transduces oncogenic signals only in conjunction with cognate cytokine receptors, in a cytokineindependent version of its normal signaling mechanisms. In this respect JAK2V617F is quite different from other constitutively active mutant cytoplasmic kinases such as the TEL-JAK2 or BCR/ABL oncoproteins (7,21), which do not require binding to any receptor for kinase activation and phosphorylation of signal transduction proteins. This finding also provided a molecular basis for the prevalence of JAK2V617F in diseases of myeloid cells that express cognate homodimeric type I cytokine receptors, and for the overlapping clinical observations in these diseases.
Other groups confirmed our finding that expression of JAK2V617F in Ba/F3 cells does not trigger factor-independent growth unless a homodimeric cytokine receptor is coexpressed (22), and the same cytokine receptor dependence has been shown in 32D cells (20). However, some reports showed that expression of JAK2V617F alone (i.e. without coexpression of a cytokine receptor) could transform Ba/F3 cells, and co-expressing EpoR only increased the extent of JAK2V617F phosphorylation and its transformation potential (10,23). Here we resolve this controversy. We show that expression in Ba/F3 cells of JAK2V617F at levels commensurate with those of endogenous JAK2 leads to factor-independent growth only if a homodimeric cytokine receptor is coexpressed. In contrast, overexpression of JAK2V617F at levels 10-fold that of the endogenous JAK2 does lead to factor-independent growth without coexpression of a cytokine receptor. Likely, JAK2V617F is binding to some unknown endogenous cytokine receptor(s), because JAK2 domains required for receptor association, JH3-JH7, are essential for cytokine-independent growth of Ba/F3 cells overexpressing JAK2V617F, and an Y119E mutation in the JH7 domain that eliminates the receptor association of JAK2 (24), also disrupts JAK2V617F-mediated constitutive growth in Ba/F3 cells.
Using Ba/F3 cells that express JAK2V617F at a low level, comparable to endogenous JAK2, we further showed that disruption of EpoR dimerization attenuated cytokine-independent growth, indicating that EpoR dimerization plays an essential role in the activation of JAK2V617F. Furthermore, we showed that coexpression of JAK2V617F with EpoR mutants in a conserved juxtamembrane domain that retain JAK2 binding but are defective in Epo-dependent JAK2 activation, do exhibit cytokine-independent signaling and cell proliferation. This indicates that key EpoR residues essential for Epo-dependent activation of JAK2 are not essential for JAK2V617F activation.
Our work suggests that the structure of the constitutively activated JAK2V617F is likely to be different from that of an Epoactivated wild-type JAK2. This raises the possibility that it could be specifically targeted by small molecule therapeutics.

EXPERIMENTAL PROCEDURES
Expression Vectors-The murine JAK2 cDNA was cloned into the retroviral vector MSCV-Neo. The JAK2V617F mutation was generated in MSCV-Neo using site-directed mutagenesis (Quikchange-XL, Stratagene) and confirmed by full-length DNA sequencing (20). JAK2 and JAK2V617F cDNAs were then subcloned into the retroviral vector MSCV-IRES-GFP described by Zhang et al. (25). JH 1-2 and JH 1-2 V617F were generated by PCR amplification of the DNA segments between codon 535 (methionine) and the stop codon in JAK2 and JAK2V617F cDNAs and then cloned into the same MSCV-IRES-GFP vector. JH 1-2 cDNA were also fused with the TEL part (codon 1-316) of a human TEL-JAK2 cDNA (6). JH 1-6 V617F and Y119E-JAK2V617F were also generated by PCR amplification of the N termini of mutant JAK2 cDNAs to replace the corresponding regions between BamH I and XhoI sites in the MSCV-JAK2V617F-IRES-GFP vector. Wild-type and mutant EpoR cDNA were cloned into the retroviral vector pMX-IRES-GFP as described earlier (26,27), and so is the truncated EpoR with C-terminal 221 amino acids removed, named EpoR-T (28). For co-expression of wild-type EpoR and EpoR-T, EpoR cDNA was cloned into retroviral vector pBI-IRES-CD4, and doubly infected cells were sorted into a GFP-and CD4positive population.
Cell Culture-293T cells were grown in Dulbecco's modified Eagle's medium with 10% fetal calf serum. Transient cotransfection of 293T cells and generation of retroviral supernatant were performed as previously described (20). Parental Ba/F3 cells were grown in RPMI 1640 containing 10% fetal calf serum and 10% WEHI-3B cell supernatant as a source of IL-3 (WEHI media). These cells were infected with retroviral supernatant generated from either the MSCV-JAK2-Neo or MSCV-JAK2V617F-Neo vectors, respectively, and then selected in media containing 1 mg/ml G418. To introduce exogenous EpoR, parental Ba/F3 cells or these G418-selected cells were infected with retrovirus carrying wild-type or mutant EpoR generated from the pMX-IRES-GFP vector (26), and sorted for GFP positive cells. For expression of wild-type or mutant JAK2 using the MSCV-IRES-GFP vector (25) To assess factor-independent growth, cells were washed three times in RPMI 1640 medium containing 10% fetal calf serum and cultured in the absence of Epo and IL-3 for 5 days. The number of viable cells was determined by Trypan Blue exclusion at different time points. In cells expressing wild-type or mutant EpoRs, their responses to Epo stimulation was measured by incubating cells with various concentration of Epo. The number of viable cells was determined at the end of the 3-day incubation.
Immunoprecipitation and Immunoblotting-Ba/F3 cells stably expressing JAK2 or EpoR were lysed in Nonidet P-40 lysis buffer (150 mM NaCl, 1% Nonidet P-40, and 50 mM Tris, pH 7.4) with sodium vanadate and protease inhibitors (Roche Applied Science). Proteins in these lysates were either immunoprecipitated with antibodies against JAK2 (Upstate) or STAT5 (Santa Cruz Biotechnology) first, or directly separated through SDS-PAGE. Immunoblotting were performed after transferring the SDS-PAGE to a nitrocellulose membrane.

High Expression of JAK2V617F Circumvents the Requirement for an Ectopic Cytokine
Receptor-In our previous work we introduced murine JAK2V617F into IL-3 dependent murine Ba/F3 cells using retroviruses generated from the MSCV-JAK2V617F-Neo vector. In the G418-selected cells, we observed no cytokine-independent growth unless a homodimeric cytokine receptor such as the EpoR was coexpressed. This observation supported the notion that cytokine receptors serve as scaffolds for the constitutive activation of JAK2 V617F, and its ability to transform Ba/F3 cells (20). However, others reported that exogenous expression of EpoR only enhances the transformation of Ba/F3 cells by JAK2V617F, but is not essential (10,23).
We suspected that differences in JAK2V617F expression levels account for this discrepancy, and thus we compared the amount of JAK2 protein in these cells by Western blotting with an anti-JAK2 antibody (Fig. 1B). Ba/F3-JAK2(Neo) and Ba/F3-JAK2V617F(Neo) cells contained roughly twice the amount of JAK2 protein of parental Ba/F3 cells, indicating that the ectopic JAK2 and JAK2V617F were expressed at levels close to that of endogenous JAK2. In contrast, the ectopic JAK2 and JAK2V617F proteins were expressed at levels about 10-fold that of endogenous JAK2 in Ba/F3-JAK2(GFP) and BA/F3-JAK2V617F-GFP cells.
As shown in the lower panel of Fig. 1B, we could not detect EpoR protein in any of the Ba/F3 cells unless EpoR was ectopically introduced. The same number of mouse fetal liver cells (FLC) were used as positive control for EpoR expression, and the levels of exogenous EpoR expression in the Ba/F3 cells were about 10-fold that in the fetal liver cells.
We do not know why the MSCV-IRES-GFP and MSCV-Neo vectors express such different amounts of encoded protein. We hypothesize that the high JAK2V617F level in Ba/F3-JAK2V617F(GFP) cells allows it to signal through binding to the cytosolic domain of the endogenous IL-3 receptor or another perhaps unknown cytokine receptor that is expressed at a low level. Considering that JAK2V617F has to compete with the endogenous wild-type JAK2 for receptor association, a 10-fold increase in JAK2V617F protein level may allow JAK2V617F to saturate the limited number of endogenous cytokine receptors in Ba/F3 cells, and generate enough signal to support cytokine-independent proliferation.
JAK2 Domains Essential for Receptor Association Are Required for JAK2V617F-mediated Activation-JAK2 associates with the membrane-proximal segment of cytokine receptors through its JH3 to JH7 domains (3). A truncated JAK2 without its JH3 to JH7 domains fails to bind to cytokine receptors and the JH1 kinase domain remains inactive due to inhibition by the JH2 pseudokinase domain (2). A fusion of this truncated JAK2 with the dimerization domain of the TEL transcription factor, mimicking the oncogenic TEL-JAK2 protein, does lead to constitutive activation and factor-independent growth of Ba/F3 cells (6). To further test the hypothesis that high expression levels of JAK2V617F allow it to signal through endogenous cytokine receptors, we removed the JH3 to JH7 domains of JAK2 that are responsible for receptor binding. We named this truncated JAK2 protein, containing residues 535-1129, as JH 1-2 . We then introduced the V617F mutation to this protein (named JH 1-2 V617F), and inserted these coding sequences into the overexpressing MSCV-IRES-GFP vector. We also used an in-frame fusion of the N terminus of human TEL (residues 1-316) and JH 1-2 , named TEL-JAK2, as a positive control that should exhibit factor-independent cell proliferation.
When introduced in Ba/F3 cells using the MSCV-IRES-GFP vector-based retroviruses, both JH 1-2 and JH 1-2 V617F were expressed at even higher levels than full-length JAK2V617F ( Fig. 2A). In contrast to full-length JAK2V617F, JH 1-2 V617F could not support cytokine-independent growth. However, when the same JH 1-2 region was fusion to the TEL oligomerization domain, the resulting TEL-JAK2 fusion protein did transform Ba/F3 cells to growth factor independence (Fig. 2B).
We next determined whether any of these mutant JAK2 proteins are capable of constitutive auto-phosphorylation, and more importantly if elevated expression of these mutants can activate the major downstream signaling protein STAT5 that is normally recruited by cytokine receptors. To this end Ba/F3 cells expressing JAK2, JAK2V617F, JH 1-2 , JH 1-2 V617F, or TEL-JAK2, in the MSCV-IRES-GFP vector-based retroviruses, were starved for 4 h without IL-3. Activation of JAK2 was detected by  FEBRUARY 29, 2008 • VOLUME 283 • NUMBER 9

JOURNAL OF BIOLOGICAL CHEMISTRY 5261
Western blot with an antibody specifically recognizing the phosphorylated activation loop tyrosines (Tyr-1007, Tyr-1008) (Fig. 2C). Starved parental Ba/F3 cells and IL-3-treated Ba/F3 cells were used as controls. STAT5 proteins in IL-3 starved cells were immunoprecipitated, and phospho-STAT5 were probed by Western blot with 4G10 (Fig. 2D). Fig. 2C showed that, following starvation of growth factors, only JAK2V617F and TEL-JAK2 were phosphorylated in the activation loop, indicating their activation. Consequently, only cells expressing JAK2V617F and TEL-JAK2 showed constitutive phosphorylation of STAT5 (Fig. 2D). Thus, while constitutive activation of overexpressed JAK2V617F can occur in the absence of coexpression of the EpoR or other cytokine receptor, this activation requires the presence of the JH3 to JH7 domains, suggesting that association of JAK2V617F to some endogenous cytokine receptor is required for its constitutive activation.
Published results (2) taken together with our data on the reconstructed TEL-JAK2, indicate that deleting the JH3 to JH7 domains does not compromise the potential kinase activity of the JH1 domain. The loss of JAK2 activation in the deletion mutants would be the consequence of deleting the segments required for receptor association. The same notion applies to the other FERM domain mutations we describe below.
Disrupting JAK2V617F Receptor Association Eliminates Its Activation-The above results suggested receptor binding through the JH3 to JH7 domains of JAK2V617F is essential for its ability to transform Ba/F3 cells. However, we could not exclude the possibility that the JH3 to JH7 domains play essential roles other than receptor association. To be more specific in disrupting receptor association of JAK2V617F, we generated a JH7 deletion (deletion of amino acids 1-144) and a Y119E point mutation in JAK2V617F, named JH 1-6 V617F and Y119E-JAK2V617F, respectively. JAK2 carrying either mutation was reported to lose receptor association (3,24). The Y119E mutation is believed to mimic Epo-dependent phosphorylation and down-regulation of JAK2 activity through dissociation of JAK2 from the EpoR (24). Next, we overexpressed these JAK2 mutants in Ba/F3 cells using retroviruses generated from MSCV-IRES-GFP vectors, and measured their growth in the absence of IL-3. As shown in Fig. 3A, only Ba/F3 cells overexpressing JAK2V617F can grow independently of IL-3. Overexpression of JH 1-6 V617F or Y119E-JAK2V617F does not lead to factor-independent growth, even though, as shown by the Western blot in Fig. 3B, they are overexpressed at a similar level as JAK2V617F. This result further strengthens the notion that receptor binding is essential in the constitutive activation of JAK2V617F, and that overexpressed JAK2V617F must function through binding to as yet unknown endogenous cognate cytokine receptors.
Nonetheless, expression of JAK2V617F at lower (endogenous) levels does not lead to its constitutive activation unless a cytokine receptor such as the EpoR is coexpressed. In the following experiments we use Ba/F3-JAK2V617F(Neo) cells to dissect the contribution of different segments of the EpoR in constitutive activation of this mutant JAK2 protein.
Dimerization of EpoR Is Necessary for JAK2V617F Activation-JAK2 binds to the membrane-proximal segment of cytokine receptors, and cell surface EpoRs (and possibly other cytokine receptors) form dimers even in the absence of ligands (27). Cytokine binding to its receptor triggers conformational changes in the receptor dimer leading to trans-phosphorylation of the JAK2 activation loop tyrosines and activation of JAK2 kinase activity. Even though the V617F mutation is thought to disrupt the inhibition of the JH1 kinase domain by the JH2 domain, allowing constitutive trans-phosphorylation, the two JAK2V617F molecules likely still need to be juxtaposed to each other. For JAK2V617F, binding to a dimeric albeit inactive cytokine receptor through its JH3 to JH7 domains would provide such an environment.
To test the hypothesis that dimerization of a cytokine receptor is necessary for constitutive activation of JAK2V617F, we co-expressed wild-type EpoR along with a truncated EpoR that has been deleted for most of the intracellular domain including the JAK2 binding sites (EpoR-T). It is well known that EpoR-T co-expression disrupts Epo-mediated signaling, presumably by forming non-functional heterodimers with the wild-type EpoR (28). If EpoR expression but not dimerization were required for JAK2V617F activation, we would expect that co-expression of EpoR-T (which of course cannot bind JAK2) would have little impact on EpoR-mediated activation of JAK2V617F. In contrast, coexpressing EpoR-T with wild-type EpoR greatly reduced JAK2V617Fmediated cytokine-independent growth of Ba/F3 cells (Fig. 4A). This indicates that JAK2V61F indeed functions through binding to dimerized EpoR receptors. Both EpoR and EpoR-T bear an HA epitope tag in their extracellular domains and their expression can be detected with anti-HA antibody, as shown in Fig. 4B.
Essential Conserved Hydrophobic Motif in the EpoR Juxtamembrane Domain Is Not Essential for Constitutive Activation of JAK2V617F-Three hydrophobic residues Leu-253, Ile-257, and Trp-258 in the cytosolic juxtamembrane domain of the EpoR are highly conserved among cytokine receptors (3). These residues are essential for Epo-mediated signaling, as L253A, I257A, and W258A EpoR mutants are expressed normally on the cell surface, bind JAK2 normally (3), yet cannot support either Epo-mediated JAK2 activation or Epo-mediated growth of Ba/F3 cells (27). This suggests that these residues have a specific function in activating associated JAK2 upon Epo stimulation. Fig. 5 confirms that these mutants cannot support Epo-dependent growth in Ba/F3 cells, nor can they effectively activate JAK2 and STAT5 in response to Epo.
Next these EpoR mutants, as well as wild-type EpoR, were expressed in Ba/F3-JAK2V617F(Neo) cells. Fig. 5C shows that Ba/F3-JAK2V617F(Neo) cells coexpressing any of these three EpoR mutants proliferate in the absence of Epo, albeit not as rapidly the cells expressing wild-type EpoR. Importantly, both wild-type and L253A, I257A, and W258A mutant EpoRs all support constitutive activation of JAK2V617F and Stat5 (Fig. 5D); as a control BaF3-JAK2V617F(neo) cells not coexpressing any EpoR do not exhibit constitutive activation either of JAK2 or Stat5. We conclude that the conserved hydrophobic residues Leu-253, Ile-257, and Trp-258 in the EpoR are not important for Epo-independent activation of JAK2V617F. In our earlier publication we  reported that the EpoRW282R mutant, which is unable to bind JAK2 and cannot support Epo-dependent JAK2 activation, does not support factor-independent activation of JAK2V617F (20). As detailed under "Discussion," these results reiterate the importance of receptor binding in the activation and signaling by JAK2V617F, but also suggest that the conformation of activated JAK2V617F and wild-type JAK2 are different.

DISCUSSION
Activation of JAK2V617F Requires Tethering to a Homodimeric Cytokine Receptor-The identification of JAK2V617F added another member to the growing list of oncogenic tyrosine kinase mutations that underline MPD and other hematopoietic malignancies (29). Among this group, JAK2V617F has several unique properties. Most importantly, unlike BCR-ABL, TEL-JAK2, or mutations in receptor-tyrosine kinases such as FLT3-ITD and KITD816V (30,31), JAK2V617F is a cytosolic non-receptor tyrosine kinase that cannot form a dimer by itself. To be able to trans-phosphorylate the neighboring JH1 kinase domain JAK2V617F has to be tethered to a cognate homodimeric cytokine receptor for its dimerization and activation (20).
The cytokine receptor dependence of JAK2V617F-mediated transformation of Ba/F3 cells has been controversial; others have reported that expression of JAK2V617F in normally IL-3 dependent cells such as BaF3 cells allows them to proliferate in the absence of cytokines (10,23). Here we confirmed that Ba/F3 cells overexpressing JAK2V617F at a level ϳ10 times normal indeed do not require expression of an exogenous EpoR for cytokine-independent growth. We could not detect EpoR expression in these untransfected Ba/F3 lines. However, when co-expressing EpoR in Ba/F3-JAK2V617F (neo) cells, high EpoR expression was observed, indicating that the low level of JAK2V617F expression may select for cells with high EpoR levels. On the other hand, when JAK2V617F is expressed at a high level, it may function through binding to the cytosolic domain of an unidentified cytokine receptor expressed at a low endogenous level. To confirm this hypothesis, we first showed that the JH3-JH7 domains of JAK2V617F that are responsible for binding to cytokine receptors are essential for transformation of BaF3 cells and the activation of JAK2V617F; in their absence the JH1 and JH2 domains carrying the V617F mutation are inactive even though they are expressed at much higher levels than endogenous JAK2. Second we showed that more specific disruptions of JAK2 receptor association, through JH7 deletion or the Y119E point mutation, abolished factor-independent growth in Ba/F3 cells overexpressing these mutants.
Furthermore, we showed that disruption of EpoR dimerization by coexpressing a truncated EpoR, deleted of most of its intracellular domain including JAK2 binding sites, attenuated cytokine-independent growth of cells expressing JAK2V617F at normal levels, indicating that EpoR dimerization plays an essential role in the activation of JAK2V617F. Were tethering of JAK2V617F to a monomeric cytokine receptor sufficient for its activation, overexpression of a mutant EpoR lacking its cytosolic domain would not be expected to inhibit EpoR-dependent JAK2V617F signaling.
Collectively, these experiments suggest that high levels of JAK2V617F allow it to signal by binding to unknown endogenous receptors in Ba/F3 cells, and thus that the receptor association that dimerizes JAK2V617F is a pre-requisite for JAK2V617F activation.
Activation of JAK2V617F Does Not Require Three Conserved Receptor Juxtamembrane Hydrophobic Amino Acids Essential for Cytokine-mediated Activation of Wild-type JAK2-Because cell surface EpoR forms dimers even in the absence of Epo (32), receptor dimerization is required but not sufficient for the activation of wild-type JAK2; additional conformational changes induced by Epo binding are needed. The conserved hydrophobic residues, Leu-253, Ile-257, and Trp-258, in the juxtamembrane cytosolic domain, play an essential role in this process, because substitution of an alanine at any one of these positions eliminates Epo-induced JAK2 activation. These mutant receptors are expressed normally on the cell surface and bind JAK2 normally (3), yet cannot support Epo-mediated JAK2 activation nor Epo-mediated growth of Ba/F3 cells (27). In contrast, we showed in this article that EpoRL253A, EpoRI257A, and EpoRW258A are able to support Epo-independent phosphorylation of JAK2V617F and activation of the downstream STAT5 signal transduction protein. They are also able to support Epoindependent proliferation of BaF3 cells expressing JAK2V617F. Thus the conserved hydrophobic juxtamembrane motif is not involved in Epo-independent activation of JAK2V617F. We conclude that the active JAK2V617F-EpoR complex likely has a conformation that is different from that of the Epo-activated wild-type JAK2-EpoR complex.
A Model for Activation JAK2V617F Kinase Activity and Its Relevance to Epo-mediated Activation of JAK2-The V617F mutation is located in the JH2 pseudokinase domain of JAK2, which normally functions as an autoinhibitory segment that keeps the JH1 JAK2 kinase inactive in the absence of ligand stimulation (2). There is no known structure of any JH2 domain. However, a computational model of the JH1-JH2 complex showed that Val-617 would be located on one of the two principal interfaces between JH1 and JH2 (Interface 2). It would be in direct contact with the activation loop of JH1, thus stabilizing it in an inactive conformation (33). It is likely that the V617F mutation weakens the interactions between JH1 and JH2, allowing the JH1 kinase activation loop to break free. We hypothesize that in disrupting the JH1 JH2 interaction of a JAK2 protein by the V617F mutation, the freed JH1 kinase domain can move through an expanded three-dimensional space. Provided they are bound to the cytosolic domains of a dimeric cytokine receptor the two JH1 domains can interact and trans-phosphorylate each other easily without the normal Epo-induced conformational changes.
Leu-253, Ile-257, and Trp-258 are conserved hydrophobic residues in the juxtamembrane domain of EpoR that are essential for the activation of wild-type JAK2 upon Epo stimulation. Although they are not required for JAK2 binding, these residues may function as a switch that somehow turns off JH2 inhibition of JH1 following Epo stimulation. Such function is not needed for the activation of JAK2V617F, because the inhibition by JH2 of the JH1 kinase would be disrupted by V617F mutation. Thus