Studies with Chimeric Mpl/JAK2 Receptors Indicate That Both JAK2 and the Membrane-proximal Domain of Mpl Are Required for Cellular Proliferation*

The thrombopoietin (TPO) receptor c-Mpl, like other members of the cytokine receptor superfamily, requires the association and activation of Janus kinases (JAKs) for normal signal transduction. The membrane-proximal portion of the signaling domain, containing conserved box1 and box2 motifs, is sufficient to support the proliferation of cytokine-dependent cell lines and basal megakaryocytopoiesis in vivo. We hypothesized that activation of the JAK2 kinase alone might be sufficient for proliferative signaling. To test this premise, we constructed chimeric receptors in which the extracellular and transmembrane portions of Mpl were fused to the pseudokinase and kinase domains of murine JAK2 kinase. When expressed in the interleukin-3-dependent cell line Ba/F3, the chimeric receptors were appropriately expressed on the cell surface and were able to initiate tyrosine kinase activity upon exposure to TPO. However, chimeric receptors lacking an intact box2 domain of Mpl were unable to support proliferation at any concentration of TPO. Only chimeric receptors containing both JAK2 kinase activity and the box2 region initiated proliferative signaling. Within the box2 motif, we determined that the sequence Glu56-Ile57-Leu58 of the Mpl cytoplasmic domain is critical for proliferation of the chimeric receptors. Furthermore, TPO-dependent induction of c-myc transcription is also dependent on this motif. These results indicate that JAK2 activation alone is not sufficient for TPO-induced proliferation and that one or more essential signaling pathways must arise from the cytoplasmic domain of Mpl that includes box2. Although the nature of the signal transduction pathway is not yet known, this second proliferative event is likely to regulate c-myc expression.

The thrombopoietin (TPO) 1 receptor c-Mpl is necessary for normal megakaryocyte and platelet development and promotes the survival of hematopoietic progenitor and stem cells (reviewed in Ref. 1). Like other members of the cytokine receptor superfamily, Mpl does not contain endogenous tyrosine kinase activity, but instead regulates the activity of Janus kinases (JAKs). It has been widely reported that mutations or deletions of the Mpl box1 and box2 motifs abrogate JAK2 phosphorylation and also prevent TPO-induced proliferation (2)(3)(4)(5)(6) as well as the transforming potential of the v-mpl oncogene (7). Similarly, an engineered cell line that lacks JAK2 expression cannot support TPO signal transduction (8), and fetal liver cells derived from jak2 nullizygous mouse embryos do not contain megakaryocyte progenitors or TPO-responsive cells (9). Therefore, the bulk of the available data indicate that JAK2 activation is essential for normal TPO signaling.
Our group (6) and others (2) have demonstrated that the membrane-proximal portion of the Mpl signaling domain is sufficient for both JAK2 phosphorylation and proliferation, comparable to that seen with the wild-type receptor. This was confirmed in vivo by the recent report of experiments in which an altered Mpl receptor with only 61 membrane-proximal cytoplasmic residues was targeted to the native mpl locus; such homozygous mice display normal base-line megakaryocyte and platelet production (10).
All of the truncated Mpl receptors utilized in the previously reported studies include the box1 and box2 motifs, believed to be required for JAK docking to the receptor. The core of box1 contains two proline residues, separated by either one or two intervening amino acids, usually within the membrane-proximal 25 residues of the cytoplasmic domain (11). For Mpl, box1 includes the sequence Pro-Ser-Leu-Pro (residues 17-20) (7). Box2 is less precisely conserved, but is usually recognized by its position (within residues 40 -65 of the membrane-spanning domain) and by a cluster of acidic residues embedded in a hydrophobic region containing one or two basic residues at the carboxyl-terminal end (11). In the Mpl receptor, box2 may include much of the region between cytoplasmic residues 45 and 60. At present, it is not known whether the membraneproximal domain of Mpl fulfills any role other than the recruitment and activation of JAKs.
Our previous studies (6) and those of others (5) have shown a direct correlation between the level of JAK2 phosphorylation (activation) and the level to which cellular proliferation is supported. This finding led to the hypothesis that activation of JAK2 was not only necessary, but also sufficient for Mpl-directed proliferation. To test this hypothesis directly and to further explore the signaling requirements of Mpl, we constructed a series of chimeric Mpl/JAK2 receptors, replacing varying amounts of the Mpl signaling domain with the JAK2 pseudokinase and kinase domains. Previously published studies using the growth hormone receptor in COS cells demonstrated that this technique is feasible and that the fused JAK2 kinase domain is activated, as predicted, in a growth hormonedependent manner (12). Using both the kinase (JH1) and pseudokinase (JH2) domains of murine JAK2, there should be less constitutive tyrosine kinase activity than was seen when only the kinase domain was used (12). In this study, we found that a chimeric receptor containing 12 cytoplasmic residues of Mpl fused directly to the JH2 and JH1 domains of JAK2 (T12/ JAK) was phosphorylated in a TPO-dependent manner, but was not able to support proliferation of Ba/F3 cells, even at high TPO concentrations. In a series of add-back constructs, we found that JAK2 fused after 28 cytoplasmic residues of Mpl (T28/JAK, including box1) was also incapable of proliferative signaling. By mutating critical residues of the box1 motif, we were able to inactivate forms of Mpl containing 53 and 69 cytoplasmic residues (T53/box1() and T69/box1(), respectively). When JAK2 was fused to these truncated Mpl constructs, TPO-dependent proliferation was restored if all (T69/box1()/ JAK), but not part (T53/box1()/JAK), of box2 was present, despite both chimeric receptors displaying JAK2 activity. Moreover, alanine substitution of only three amino acids in this region eliminated its function. This study establishes that, although JAK2 activation is necessary, it is not sufficient for proliferation of Ba/F3 cells and that one or more additional signaling events arise from the box2 subdomain. Finally, additional studies demonstrate that the same residues of box2 are essential for ligand-dependent c-myc induction. Because the box1/box2 organization of Mpl is similar to that of many other receptors, it is likely that our results reflect signal transduction mechanisms shared by other cytokine receptors.

EXPERIMENTAL PROCEDURES
Plasmid Construction-A wild-type murine mpl cDNA (13) was cloned into the expression vector pcDNA3/Neo(ϩ) (Invitrogen, Carlsbad, CA) using EcoRI (5Ј) and XhoI (3Ј) as cloning sites. Carboxylterminal truncations (T53 and T69) were described previously (6). The wild-type murine jak2 cDNA was generously provided by James Ihle (St. Jude Children's Research Hospital, Memphis, TN) and Stuart Frank (University of Alabama at Birmingham, Birmingham, AL). To make the chimeric jak2 constructs, a new XhoI site was created immediately downstream of the jak2 stop codon. Also, a unique ClaI site was generated in the jak2 cDNA immediately before amino acid 526 of the full-length protein, upstream of both the pseudokinase and kinase domains as utilized by previous studies (12). Finally, a new ClaI site was generated at the desired position of the c-mpl expression plasmid, and the ClaI-XhoI fragment of jak2 (1812 bp, encoding 604 residues) was ligated into the corresponding sites of pcDNA3/Neo/mpl. The ClaI site resulted in the addition of two residues (Ile-Asp) at the fusion site between mpl and jak2. Mutations in the c-mpl coding region were introduced into the box1 (P17G/P20G) and box2 (E56A/I57A/L58A) motifs using the QuikChange kit (Stratagene, La Jolla, CA) with oligonucleotides synthesized by Operon Technologies, Inc. (Alameda, CA). Similar methods were used to inactivate the JAK2 kinase by Y1007F mutation (14). The mutated regions were sequenced to confirm polymerase fidelity (BigDye, PE Biosystems, Foster City, CA).
Cell Lines and Culture Conditions-Parental Ba/F3 cells were originally provided by Alan D'Andrea (Dana-Farber Cancer Institute, Boston, MA). These cells and all derived sublines were maintained in RPMI 1640 medium (BioWhittaker, Inc., Walkersville, MD) supplemented with 10% heat-inactivated fetal calf serum, L-glutamine, penicillin, streptomycin (all from BioWhittaker, Inc.), and 0.2% conditioned medium containing murine interleukin-3 (mIL-3; approximate concentration of 1.5 ng/ml). Cells were transfected by electroporation of 3-5 g of plasmid DNA mixed with 4 ϫ 10 6 Ba/F3 cells in 1 ml of RPMI 1640 medium (at 300 V and 800 microfarads and at 300 V and 1180 microfarads). After a 24-h incubation in standard growth medium containing mIL-3, the cells were collected by centrifugation, resuspended in growth medium with mIL-3 and 1 mg/ml G418 (Sigma), and serially diluted into 96-well plates. After 7-10 days, clonal populations were chosen from an appropriate dilution (such that fewer than half of the wells contained viable cells), and the new cell lines were tested for Mpl expression by fluorescent immunostaining and Western blotting (see below). At least two clones expressing each construct were tested and maintained to assure that the results were internally consistent. Cells were maintained in IL-3 and G418 and were never exposed to TPO until the start of an experiment. This practice decreases the chance of inadvertently selecting cells that spontaneously express the wild-type Mpl receptor.
Flow Cytometry-All cell lines were tested by fluorescent antibody labeling to confirm that a uniform population of Mpl-expressing cells was present and that approximately equal levels of Mpl were expressed on different cell lines. Cells were first labeled with a polyclonal rabbit antiserum directed against the extracytoplasmic portion of Mpl (generously provided by Zymogenetics, Inc., Seattle, WA). The antiserum (final dilution of 1:1000) was allowed to bind for 45 min at 4°C. After washing, the secondary antibody was added (phycoerythrin-conjugated anti-rabbit IgG; Caltag Laboratories, Burlingame, CA) for 30 min. The cells were again washed and analyzed on a flow cytometer (FACSCalibur, BD PharMingen) to detect fluorescence intensity. Parental Ba/F3 cells were used as a negative control.
Cellular Proliferation Assay (MTT)-Proliferation assays were performed as previously described (6). Briefly, cells in log-phase growth were washed three times to remove any traces of IL-3 and resuspended in RPMI 1640 medium containing 10% heat-inactivated fetal calf serum, L-glutamine, and antibiotics (no growth factor). Cells were aliquoted into a 96-well tissue culture plate (50 l/well, 5,000 -10,000 cells) using a multichannel pipettor, and an equal volume of medium containing serial 10-fold dilutions of recombinant murine TPO derived from conditioned supernatant or in purified form (10 pg/ml to 1000 ng/ml; Kirin Corp., Gunma, Japan) was added. After 36 h of incubation, MTT reagent (1 mg/ml final concentration; Sigma) was added to each well, and the plates were returned to the incubator for 5 h. Cells were lysed, and formazan crystals were dissolved by adding 100 l of MTT lysis buffer to each well. Absorbance values (570 -630 nm) were recorded on triplicate samples using a Bio-Rad plate reader; experiments were performed multiple times with more than one clone for each chimeric or mutant Mpl construct. Data are expressed as a percentage of maximum proliferation, determined separately for each cell line using an optimal concentration of mIL-3 (ϳ15 ng/ml).
Immunoprecipitation and Western Blotting-Cells were grown in mIL-3 to a final concentration of 5-8 ϫ 10 5 /ml. After washing twice to remove all mIL-3 and serum, the cells were resuspended at a final concentration of 2 ϫ 10 6 /ml in RPMI 1640 medium with 0.5% bovine serum albumin (Sigma) and allowed to incubate at 37°C for 6 -12 h. Equal volumes of cells were then either unstimulated or exposed to TPO (6.7 ng/ml as conditioned tissue culture medium) for 10 min. After washing twice with ice-cold phosphate-buffered saline, the cells were resuspended at 4°C in lysis buffer containing 1% Triton X-100 and inhibitors of proteases and tyrosine phosphatases as previously described (6). Protein concentration was determined using a modified Lowry assay (Protein D/C, Bio-Rad), and aliquots were frozen (Ϫ80°C) until used. All immunoprecipitations were performed with 1 mg of total protein starting material as previously described (6,15). Samples were immunoprecipitated with polyclonal rabbit antisera specific for Mpl (2 l/reaction; Zymogenetics, Inc.) and JAK2 (2 l/reaction; Upstate Biotechnology, Inc., Lake Placid, NY). Immune complexes were collected by adding 25 l of protein A-Sepharose beads (Santa Cruz Biotechnology) as previously described (8,15). Precipitated proteins were analyzed on 7.5 or 4 -20% denaturing acrylamide gels, transferred to nitrocellulose membranes, and probed with primary antibody (anti-phosphotyrosine antibody 4G10 at 1:2000; Upstate Biotechnology, Inc.) and secondary antibody (horseradish peroxidase-coupled goat anti-mouse IgG at 1:3000; Bio-Rad). Chemiluminescent reagents (PerkinElmer Life Sciences) and exposure to x-ray film were used to visualize phosphotyrosine-containing proteins. Blots were then stripped of antibody and reprobed with the same antibody used for immunoprecipitation and horseradish peroxidase-coupled goat anti-rabbit IgG as secondary antibody (Bio-Rad) to demonstrate equal amounts of starting protein (8,15).
Reverse Transcription-PCR to Detect c-myc Induction-Ba/F3 cells expressing truncated and mutant Mpl receptors were maintained in RPMI 1640 medium containing 10% fetal calf serum and IL-3 while growing in log phase. Cells were then washed extensively to remove growth factor and allowed to incubate in serum-and cytokine-free medium supplemented with 0.5% bovine serum albumin for 14 -16 h.
Recombinant murine TPO was added (10 ng/ml), and aliquots were withdrawn at 0, 1, and 3 h. Total RNA was isolated from cells using the RNeasy minikit (QIAGEN Inc., Valencia, CA); the concentration was determine by spectrophotometry; and the quality was assessed on agarose gel. One microgram of total RNA was subjected to reverse transcription using oligo(dT) (41.7 ng/ml; Invitrogen) and Superscript II (Invitrogen.) according to the manufacturer's directions. One-tenth of the reaction was subsequently removed for PCR with either c-myc oligonucleotides (primers: 5Ј-CCCACTCTCCCCAACCC-3Ј (sense) and 5Ј-GTCTCCTCCAAGTAACTCG-3Ј (antisense); predicted product of 607 bp) or glyceraldehyde-3-phosphate dehydrogenase oligonucleotides (primers: 5Ј-CCATGGAGAAGGCTGGGG-3Ј (sense) and 5Ј-CAAAGTT-GTCATGGATGACC-3Ј (antisense); predicted product of 194 bp). PCR products were analyzed by electrophoresis on a 1% agarose gel in the presence of ethidium bromide. The gel was photographed with a Polaroid camera on a UV transilluminator.
Tamoxifen-induced c-myc Expression and Cell Cycle Analysis-Two Ba/F3 cell lines expressing chimeric T69/box1()/box2(AAA)/JAK receptors were transduced with the retroviral construct MycER-GFP (kindly provided by Ronald Cheung). This construct combines tamoxifen-inducible MycER TM (from Trevor Littlewood) in a vector containing an internal ribosomal entry site followed by the cDNA for GFP (kindly provided by Garry Nolan, Stanford University, Palo Alto, CA). Control cell lines were created using a similar construct encoding a nonfunctional form of c-myc (⌬MycER-GFP). To assess the effect of c-myc induction on TPOdependent cell cycling, T69/box1()/box2(AAA)/JAK cell lines expressing MycER-GFP or ⌬MycER-GFP were starved of serum and growth factor for 6 h and then stimulated with either TPO or IL-3 in the presence of 4-hydroxytamoxifen (Sigma) or an equivalent amount of vehicle (ethanol) for 16 h. Cellular DNA was stained with Hoechst dye 33342 (Sigma), and cells were analyzed by flow cytometry. For each sample, the gate was set to include only cells expressing GFP, and then cell cycle histograms based on DNA content were constructed. The percentage of cells in the G 1 , S, and G 2 /M phases of the cell cycle was determined by manual gating.
Box1 of Mpl Is Essential for JAK2 Activation and Proliferation-The importance of box1 for cytokine receptor signaling has been well documented, especially the characteristically spaced proline residues within the first 15-25 cytoplasmic residues (7,11). To confirm the role of box1 in a truncated Mpl receptor containing 69 cytoplasmic residues, we mutated the two proline residues in the Mpl box1 motif to glycine (P17G/ P20G). As predicted, the resulting T69/box1() receptor was completely deficient in both TPO-dependent proliferation ( Fig.  2A) and phosphorylation of endogenous JAK2 (Fig. 2B). The same results were obtained using a T53/box1() receptor (data not shown). These data confirm that box1 provides a critical signal for cellular proliferation and is necessary for JAK2 ty-rosine phosphorylation in the T69 truncated Mpl receptor.
The Box2 Core Motif Glu 56 -Ile 57 -Leu 58 Plays a Major Role in Signaling by the T69 Receptor-From our studies of truncated Mpl receptors, we know that there is a major decrease in TPO-dependent proliferation when residues 53-69 are deleted from the cytoplasmic domain (6). Further truncation of residues 28 -53 eliminates the residual proliferative response. By mutating amino acids between cytoplasmic positions 53 and 69 to alanine, we were able to measure the effect on receptor function. The greatest effect on proliferation of Ba/F3 cells was observed when three residues (Glu 56 -Ile 57 -Leu 58 ) were substituted with alanines, creating a box2 mutant designated T69/ box2(AAA) (Fig. 3A). By itself, this mutation diminished (but did not eliminate) tyrosine phosphorylation of the native JAK2 kinase (Fig. 3B).
Chimeric Mpl/JAK2 Receptors with 12 or 28 Cytoplasmic Residues of Mpl Do Not Support Proliferation-Based on our previous observation that the level of JAK2 phosphorylation and proliferation appears to be directly correlated (6), we hypothesized that phosphorylation (i.e. activation) of JAK2 alone would be sufficient for proliferative signaling. Previous studies with the erythropoietin and epidermal growth factor receptors fused to the JAK2 kinase domain suggested that this might also hold for Mpl (16,17). Therefore, we initially designed a chimeric Mpl/JAK2 receptor that included only 12 cytoplasmic residues from Mpl plus the JAK2 JH1 (kinase) and JH2 (pseudokinase) domains (T12/JAK). After verifying that the receptor was adequately expressed on the cell surface of Ba/F3 cells (Fig. 1B), distinct clones were tested for TPO-induced proliferation by MTT assay (Fig. 4A). Contrary to expectation, we found that the chimeric receptor was unable to support proliferation, even at high TPO concentrations. We then constructed a receptor containing 28 membrane-proximal cytoplasmic residues of Mpl including the entire box1 motif (T28/ JAK). Again, despite high level receptor expression (Fig. 1B), the chimeric receptor failed to support BaF3 cellular proliferation in TPO (Fig. 4A). These experiments suggest that, in addition to phosphorylation of the fused JAK2 domain, a portion of the Mpl cytoplasmic domain distal to residue 28 may be required for cellular proliferation.
A Second Mpl Proliferative Signal Exists between Ser 53 and Leu 69 -Our previous studies showed that a truncated Mpl receptor containing the first 53 cytoplasmic residues (T53) retains limited ability to activate endogenous JAK2 and to support proliferation, whereas a receptor with 69 cytoplasmic residues (T69) is nearly fully active (6). To test the role of Mpl residues 1-53 and 1-69 in supporting proliferation in a chimeric Mpl/JAK2 construct, we first mutated box1 (P17G/P20G) to eliminate any effect of endogenous JAK2 (Fig. 2). The resulting receptors (T53/box1()/JAK and T69/box1()/JAK) were expressed in Ba/F3 cells and tested for their ability to support proliferation in TPO-containing cultures. Our results demonstrate that the T53/box1()/JAK construct could not support growth, but that cells expressing the T69/box1()/JAK receptor displayed a significant degree of TPO-dependent proliferation (Fig. 4B). These data suggest that Mpl cytoplasmic domain residues Leu 54 -Leu 69 are necessary for proliferation of the chimeric Mpl/JAK2 receptor. Although these 16 amino acids may function to properly orient the distal JAK2 kinase domain, it is also possible that this subdomain, which includes the box2 core motif, provides docking sites for additional signaling molecules and activates distinct proliferative pathways.
Both JAK2 kinase domain, it is theoretically possible that T69/ box1()/JAK supports proliferation because of a favorable orientation of the fused JAK2 kinase compared with its orientation in T53/box1()/JAK, allowing interaction with signaling molecules docked on the Mpl cytoplasmic residues. Because the box2 core sequence Glu 56 -Ile 57 -Leu 58 is critical for proliferation (Fig. 3A), we engineered the triple alanine substitution in T69/ box1()/JAK, creating the T69/box1()/box2(AAA)/JAK receptor with no change in the number of residues acting as a spacer between Mpl and the JAK kinase domain. This mutation completely eliminated TPO responsiveness of the chimeric receptor (Fig. 4C), verifying the importance of the box2 motif for proliferation.
It is also possible that the T69/box1()/JAK receptor acts by recruiting endogenous JAK2 to the signaling complex, despite the fact that T69/box1() can neither proliferate nor support JAK2 phosphorylation (Fig. 2). To prove that chimeric JAK2, rather than endogenous JAK2, is necessary for proliferation, we introduced a mutation in the chimeric JAK2 kinase domain that has previously been shown to eliminate essentially all autophosphorylation and kinase activity (Y1007F) (14). When expressed in Ba/F3 cells, the resulting T69/box1()/JAK(Y1007F) receptor supported no TPO-dependent proliferative capacity (Fig. 4C), confirming the essential role of the chimeric JAK2 kinase for proliferation induced by T69/box1()/JAK.
Chimeric Receptors Undergo Tyrosine Phosphorylation in Response to TPO Stimulation-We have previously demonstrated that full-length Mpl receptors are tyrosine-phosphorylated in response to TPO binding at the carboxyl terminus (Tyr 112 and Tyr 117 ), but that the truncated T69 receptor is not tyrosinephosphorylated despite the presence of two residual cytoplasmic tyrosine residues, Tyr 8 and Tyr 28 (6). Thus, it is likely that TPO-dependent phosphorylation of a chimeric receptor represents tyrosine phosphorylation of the covalently attached JAK2 kinase domain, an essential step during JAK2 activation (14). We tested Ba/F3 cells expressing each of the chimeric receptors FIG. 2. The T69/box1() receptor cannot support proliferation or JAK2 phosphorylation. A, an MTT assay was performed using Ba/F3 cells expressing the T69 or T69/box1() receptor as well as parental Ba/F3 cells. Recombinant murine TPO (rmTPO) was added at the indicated concentration. Each data point represents the mean of triplicate readings and has been normalized as a percentage of maximum proliferation induced by mIL-3. The single cell line depicted is representative of five distinct T69/box1() clones that were each analyzed multiple times. B, cell lysates were generated from T69 and T69/box1() cells both before (Ϫ) and after (ϩ) stimulation with TPO (6.7 ng/ml for 10 min). Immunoprecipitation (IP) was performed with an anti-JAK2 antibody, and the immune complexes were analyzed by Western blotting to detect phosphotyrosine (pY) content. The blot was then stripped and reprobed to confirm equal amounts of JAK2 in each lane (lower panel).

FIG. 3. The box2 core motif is important for proliferative signaling.
A, an MTT proliferation assay was done to compare the proliferative effect of T69 versus T69/box2(AAA) cells, in which Glu 56 -Ile 57 -Leu 58 was substituted with alanine residues. In this experiment, each data point represents the average result from five distinct cell lines. The amount of murine TPO is expressed as a percentage of conditioned supernatant (1% ϭ 3.3 ng/ml). B, whole cell lysates were generated from T69 and T69/box2(AAA) cells both before (Ϫ) and after (ϩ) stimulation with TPO (6.7 ng/ml for 10 min). JAK2 was immunoprecipitated (IP), and the immune complexes were analyzed by Western blotting to detect phosphotyrosine (pY) content. The blot was then stripped and reprobed to confirm similar amounts of JAK2 in each lane (lower panel).
to verify that TPO stimulation resulted in tyrosine phosphorylation of the expressed receptor (Fig. 5). Immunoprecipitation of Mpl and Western blot analysis to detect tyrosine phosphorylation showed that the wild-type Mpl receptor was inducibly phosphorylated (Fig. 5, lanes 3 and 4), whereas the T69 receptor, lacking the distal tyrosine residues, was not (lanes 5 and 6). The chimeric receptors T28/JAK (lanes 7 and 8), T69/box1()/ JAK (lanes 9 and 10), and T69/box1()/box2(AAA)/JAK (lanes 11 and 12) were each phosphorylated in response to TPO stimulation, but the T69/box1()/JAK(Y1007F) construct (lanes 13 and 14) was not, reflecting mutation of a residue critical for JAK2 kinase activity (14). It should be noted that the chimeric Mpl/JAK receptor (ϳ140 kDa) is significantly larger than the wild-type Mpl receptor (95 kDa) and that the major sites of tyrosine phosphorylation are different for the native receptor (Mpl Tyr 112 ) and the chimeric receptors (JAK2 Tyr 1007 ), perhaps making it difficult to compare the intensity of these phosphotyrosine bands. Despite variation between clones, the extent of tyrosine phosphorylation for each chimeric receptor was similar (lanes 8, 10, and 12). We also confirmed that the chimeric receptors T12/JAK, T53/box1()/JAK, T53/JAK, and T69/ JAK underwent TPO-dependent tyrosine phosphorylation (data not shown). Furthermore, the T53, T53/box1(), and T69/ box1() receptors were not phosphorylated, as they contain neither the Mpl Tyr 112 nor the JAK2 Tyr 1007 site (compare with T69; lane 6) (data not shown). From these data, we conclude that steric constraints within the chimeric receptors do not preclude activation of the chimeric JAK2 kinase and that each of the chimeric receptors is phosphorylated to a comparable extent. Thus, quantitative differences in activation of chimeric JAK2 are unlikely to have affected our proliferation results. Of note, after SDS-PAGE, all Mpl receptors appeared as doublets that differed in size by ϳ8 kDa regardless of TPO stimulation (Figs. 5 and 6, lower panel). Although the post-translational modification responsible for this is not known, only the upper bands were tyrosine-phosphorylated.
Proliferation Does Not Correlate Directly with the Quantity of JAK2 Phosphorylation-One possible concern is that subtle differences in the cumulative JAK2 activity, whether endogenous or chimeric, might affect the proliferative capacity of the receptors under analysis. There may be a threshold of JAK2 activity below which no proliferation is possible. To test these hypotheses, we compared Ba/F3 cells expressing the T53 versus T53/JAK receptors (Fig. 7A). Despite the addition of covalently linked JAK2, the chimeric receptor did not proliferate better than its corresponding truncated receptor. Similar results were obtained when the T69 and T69/JAK constructs were com-  pared. Immunoprecipitation of JAK2 and Western blot analysis clearly demonstrated that the addition of chimeric JAK2 without altering the box1 motif resulted in greater cumulative JAK2 phosphorylation (compare Fig. 6, lanes 6 and 8; and Fig.  7B, lanes 2 and 4). Because T53 cells proliferated weakly in response to TPO, it is unlikely that there is a threshold effect; and this result demonstrates that the loss of proliferation between T69 and T53 cannot solely be explained by a decrease in JAK2 activity.
TPO-induced c-myc Induction Depends on the Mpl Box2 Motif, Not JAK2-We hypothesized that the box2 domain of Mpl might promote proliferation of the chimeric T69/box1()/JAK receptor through induction of c-myc expression. Therefore, Ba/F3 cells expressing truncated and chimeric receptors were stimulated with TPO for 0 -3 h, and total RNA was harvested from the cells at the indicated time points (Fig. 8). Reverse transcription-PCR experiments with the truncated Mpl receptor containing 69 cytoplasmic residues (T69) demonstrated a clear up-regulation of c-myc RNA levels after 60 min of TPO stimulation and a subsequent decrease after 3 h (lanes 1-3). A similar transient increase in c-myc expression was noted for T69/box1() cells (lanes 4 -6), indicating that this effect on gene expression is independent of JAK2 phosphorylation and proliferation. The chimeric T69/box1()/JAK receptor was capable of c-myc induction to a similar degree (lanes 7-9). In contrast, mutation of the box2 sequence Glu 56 -Ile 57 -Leu 58 to Ala (T69/ box1()/box2(AAA)/JAK) resulted in the loss of TPO-dependent c-myc induction (lanes 10 -12). However, myc expression alone did not rescue proliferation of T69/box1()/box2(AAA)/JAK. Introduction of tamoxifen-inducible c-myc into these cells failed to promote growth in the presence of tamoxifen alone or tamoxifen plus TPO (as assessed by flow cytometry for cell cycle progression) (data not shown). Thus, although c-myc is one signal downstream of box2, it is not the only signal. DISCUSSION Our experiments with chimeric Mpl/JAK receptors were designed to answer a fundamental question in cellular signaling: are there any signaling pathways, other than those initiated by JAK2, that are required for hematopoietic growth factor-induced cellular proliferation? The major findings of our studies include the following. 1) In our system, JAK2 kinase activity is necessary, but not sufficient, to support TPO-induced hematopoietic cellular proliferation. 2) A second proliferative signal, arising from the box2 subdomain of Mpl, is also necessary (but not sufficient) for proliferative signaling and can complement JAK2 kinase activity. 3) TPO-induced c-myc induction is dependent on an intact box2 motif and does not require JAK2 activation.
A number of previous studies have made use of the chimeric cytokine receptor/JAK model to study the ability of a Janus tyrosine kinase to initiate signal transduction and to support normal cellular physiology. This approach was first tested using a chimeric growth hormone receptor/JAK2 protein (growth hormone receptor/JAK) in COS-7 cells (18). Frank et al. (18) demonstrated that growth hormone receptor/JAK was crossphosphorylated and activated some of the same signal transduction pathways as the wild-type growth hormone receptor. However, these studies were done in an autonomous cell line, making it difficult to assess the proliferative potential of the chimeric receptor.
In addition to the data presented above, results from several published studies support the hypothesis that at least one other signaling pathway, in addition to JAK2, is involved in cytokine-mediated proliferation. First, cross-linking of a CD16/ CD7/JAK2 fusion molecule resulted in tyrosine phosphorylation and activation of many cytokine pathways, but was unable to induce proliferation of Ba/F3 cells using either full-length JAK2 or the pseudokinase and kinase domains (19,20). As appears to be the case for our chimeric Mpl/JAK2 receptors, failure to up-regulate c-myc expression may be responsible for the lack of growth signaling in these other cell systems (19,21). Second, a chimeric receptor consisting of interleukin-2 receptor-␥ c fused to JAK3 failed to induce proliferation of CTLL-2 cells without a membrane-proximal domain of the ␥ c subunit (22,23). Third, expression of a constitutively active form of JAK2 in cytokine-dependent cells did not result in autonomous growth (24). Finally, a mutant form of Mpl has been described that supports TPO-dependent proliferation without detectable JAK or STAT activation (25,26). These data all support a role for at least one distinct non-JAK signaling pathway in cellular proliferation.
In contrast, a number of published reports indicate that direct JAK2 dimerization can be sufficient to support cellular proliferation of hematopoietic cells. For example, fusion of the erythropoietin receptor to the JAK2 kinase and pseudokinase domains (erythropoietin receptor/JAK) produced a chimeric receptor capable of supporting erythropoietin-dependent proliferation in FDCWEHI-1 cells (16). Similarly, experiments performed with a chimeric epidermal growth factor receptor/JAK protein demonstrated epidermal growth factor-dependent proliferation and inhibition of apoptosis in 32D cells (17) as well as erythroid colony formation in mouse fetal liver cells (27). Furthermore, coumermycin-induced dimerization of TYK2, a member of the JAK family, supported proliferation of Ba/F3 cells, albeit with absolute dependence on Ras signaling (28). Finally, other chimeric proteins that affect constitutive dimerization of JAK2 resulted in factor-independent growth of cytokine-dependent cell lines and, in the case of a TEL/JAK translocation, may play a role in leukemogenesis (i.e. TEL/JAK chimera (29,30) and ␤ c /JAK (31)). Clearly, our results differ from the conclusions of these latter studies. Several potential explanations for the discrepant findings are discussed below (i.e. variability between cell lines utilized, inherent differences in the receptors studied, level of kinase activation, and the steric constraints on JAK2 as a chimeric protein).
As with any signaling studies done in transformed cell lines, there is concern that the specific cell line utilized may affect results. Nearly all of the studies described above were done in cytokine-dependent cell lines (FDCP, 32D, CTLL, and Ba/F3). It is possible that, in some of these cell lines, there was constitutive activation of additional signaling pathways. Although these signaling mechanisms may be unable to support proliferation by themselves, they may complement the incomplete proliferative signal of JAK kinase, making it appear that JAK activation was sufficient for proliferation. To be certain that our Ba/F3 cells did not influence our results, several of the chimeric receptors were expressed in a different cytokine-dependent cell line, FDCP-2. The same results were obtained; TPO-dependent proliferation was observed with cells expressing the chimeric T69/box1()/JAK molecule, but not with the T12/JAK and T28/JAK receptors (data not shown). The confirmation of our findings in a second cell line excludes the possi-bility of an atypical physiology restricted to our particular strain of Ba/F3 cells.
A second potential explanation for the differences between our conclusions and those of others could lie in the nature of the Mpl receptor. It is possible that the ligand-binding domains of the erythropoietin and epidermal growth factor receptors (but not Mpl) may recruit additional signaling subunits, which can provide a second signal for proliferation. The ability of the extracytoplasmic domain of receptors to influence signaling specificity was previously demonstrated for the erythropoietin and interleukin-2 receptors (32).
When using chimeric proteins to test signaling pathways, it is possible that the physical constraints imposed by covalent coupling may disrupt normal function. In studies such as those presented here, several potential problems may arise. First, the chimeric JAK2 kinase is tethered to the cell membrane, unable to traffic between cellular locations. Because endogenous JAK2 can freely associate and dissociate from the receptor, such tethering may alter normal signaling and prevent usual target molecules from being phosphorylated. Second, despite the long

FIG. 8. c-myc up-regulation is dependent on the box2 domain of Mpl.
Ba/F3 cells expressing the truncated form of Mpl with 69 cytoplasmic amino acids (T69), the same receptor with a mutation in box1 (T69/box1()), the chimeric Mpl/JAK2 receptor (T69/box1()/ JAK), and the chimeric receptor with a mutation of the box2 sequence Glu 56 -Ile 57 -Leu 58 to alanines (T69/box1()/ box2(AAA)/JAK) were exposed to TPO for 0, 1, or 3 h. Total RNA was extracted and subjected to reverse transcription-PCR with oligonucleotide primers specific for c-myc (upper panel; 28 cycles) or glyceraldehyde-3-phosphate dehydrogenase (GAPDH; lower panel; 21 cycles). The data are representative of three separate experiments. spacer region provided by the pseudokinase domain (JH2), the chimeric kinase domains of adjacent receptors may not be spatially oriented for optimal cross-phosphorylation. Finally, it is possible that excluding the amino-terminal portion of the molecule (JH3-JH7) might affect physiologic function, as this portion of JAK2 contains protein-interactive domains that help target the kinase to sites of action (33,34) and may also be required for interaction with members of the Src kinase family (35). Published reports on this issue provide conflicting answers. On one hand, direct dimerization of JAK2 using a coumermycin/gyrase B system requires the full-length molecule for short-term proliferation (36). On the other hand, the previously described erythropoietin receptor/JAK and epidermal growth factor receptor/JAK receptors, which lack the amino terminus of JAK2, are capable of supporting cellular proliferation (16,17).
For one or more of the reasons discussed above, the chimeric JAK2 kinase does not appear to be as efficient in supporting proliferation as endogenous JAK2 activated through an intact box1 motif. This is most clearly shown by the fact that mutation of the box2 core motif partially abrogates proliferation in the context of the intact T69 receptor (Fig. 3A), but completely eliminates proliferation when JAK2 is activated only as part of the fusion protein (T69/box1()/box2(AAA)/JAK) (Fig. 4C). Nonetheless, our results demonstrate that, in the context of chimeric Mpl/JAK2 proteins, the covalent kinase domain is active and, in combination with an intact box2 subdomain, can support moderate proliferation of the cells expressing T69/ box1()/JAK. As such, our experimental system allows us to map a subdomain of Mpl that is required to complement the chimeric JAK2 signal.
We cannot exclude the possibility that a very small amount of endogenous JAK2, below our limits of detection by immunoprecipitation and Western blotting, is acting to recruit additional signaling molecules through its amino terminus. However, this possibility would require that endogenous JAK2 molecules be recruited to the T69/box1()/JAK molecule, but not to T69/box1(), which contains the identical Mpl domain, nor to T69/box1()/JAK(Y1007F), which differs only by a point mutation within the JAK2 kinase domain.
From these experiments, it seems likely that a second proliferative signal arises from the membrane-proximal portion of the Mpl cytoplasmic domain, specifically including the box2 region between residues 53 and 69. Although signaling molecules that directly interact with this region have not yet been identified, the induction of c-myc expression appears to be a downstream event that is dependent on the box2 core sequence Glu 56 -Ile 57 -Leu 58 . Moreover, increases in myc RNA occur in response to TPO even when box1 is mutated such that endogenous JAK2 is not phosphorylated and the receptor is incapable of supporting proliferation. In contrast, we were unable to detect tyrosine phosphorylation of STAT5 and ERK1/2 signaling molecules in the absence of endogenous or chimeric JAK2 activity (data not shown). In previous studies, Myc has been identified as a transcription factor that promotes proliferation and inhibits differentiation for multiple cytokine receptors (21,(37)(38)(39).
It is interesting that the same box2 motif has also been shown to play a role in ligand-dependent Mpl receptor internalization (40). Our preliminary results studying truncated and mutant Mpl constructs demonstrate that the residual internalization of T69 receptors (30 -40% that of full-length molecules) is regulated by the Glu 56 -Ile 57 -Leu 58 motif and may not require JAK2 phosphorylation (40). It was recently demonstrated that internalization of the erythropoietin and gp130containing receptors also occurs in a JAK kinase-independent manner (41,42). Currently, we hypothesize that intracellular trafficking of Mpl and associated molecules serves an important function by targeting or transporting the signaling complex to intracellular compartments, as has been shown for the interferon and several interleukin receptors (43)(44)(45). Future studies will focus on the mechanisms linking Mpl internalization with cellular proliferation and an increase in c-myc expression.
In conclusion, our data suggest that TPO-stimulated proliferation of Ba/F3 cells requires two distinct signaling events. One is the activation of JAK2 kinase, normally mediated through association with the receptor box1 and box2 motifs and ligand-induced dimerization. The second results in the upregulation of c-myc RNA and requires an intact box2 motif, especially the Glu 56 -Ile 57 -Leu 58 sequence. In the context of a chimeric Mpl/JAK2 receptor, neither of these signals alone is sufficient for cellular proliferation, but together they promote growth and survival of Ba/F3 cells. Because of the homology between Mpl and the membrane-proximal domains of other cytokine receptors, it should be possible to extrapolate these findings to signaling by other members of the cytokine receptor superfamily.