A Truncated Isoform of c-Mpl with an Essential C-terminal Peptide Targets the Full-length Receptor for Degradation*

Thrombopoietin and its cognate receptor c-Mpl are the primary regulators of megakaryopoiesis and plate-let production. They also play an important role in the maintenance of hematopoietic stem cells. Here, we have analyzed the function of a truncated Mpl receptor isoform (Mpl-tr), which results from alternative splicing. The mpl-tr variant is the only alternate mpl isoform conserved between mouse and humans, suggesting a relevant function in regulating Mpl signaling. Despite the presence of a signal peptide and the lack of a transmembrane domain, Mpl-tr is retained intracellularly. Our results provide evidence that Mpl-tr exerts a domi-nant-negative effect on thrombopoietin-dependent cell proliferation and survival. We demonstrate that this inhibitory effect is due to down-regulation of the full-length Mpl protein. The C terminus of Mpl-tr, consisting of 30 amino acids of unique sequence, is essential for the suppression of thrombopoietin-dependent proliferation and Mpl protein down-regulation. Cathepsin inhibitor-1 (CATI-1), an inhibitor of cathepsin-like cysteine pro-teases, counteracts the effect of Mpl-tr on Mpl protein expression, suggesting that Mpl-tr targets Mpl for lyso-somal degradation. Together, these data suggest a new paradigm for the regulation of cytokine receptor expression and function through a proteolytic process directed by a truncated isoform of the same receptor. Assay— An XTT proliferation kit (Roche Applied ence) used to determine cytokine-dependent cell proliferation. cells plated 96- plates at 4 cells/well (cid:1) of medium the indicated concentrations of cytokine. 3.5 days of stimulation for BaF3 cells and days of stimulation of UT7 cells, 50 (cid:1) l of a 1 mg/ml stock of XTT with 5 mmol/liter phenazine methosulfate, an electron coupling agent, was added to each well. The product of XTT reduction by cells, the cells per well, was at h at nm. Protein Expression Analyses— Surface expression of Myc-Mpl determined this analysis, 5 5 cells were incubated with the mouse monoclonal antibody 9E10 against the N-terminal Myc tag of Myc-Mpl min on ice followed by with the isothiocyanate (FITC)-labeled goat anti-mouse (BD UT-7 cells served a control. Total c-Mpl and were determined using directed same and 9E10 were used detection Myc-Mpl protein by were actin cell mono-layers twice and methionine and

Cytokine receptor signaling has profound effects on cell survival, proliferation, and differentiation of the receiving cell (1). It is therefore not surprising that components of the signaling cascade are tightly regulated at several levels. An important mechanism for controlling gene expression is alternative splicing, allowing the synthesis of structurally and functionally distinct protein isoforms (2). Many alternative splice variants of different cytokine receptors have been described, but the function of most of the resulting protein isoforms remains unknown.
Cytokine receptor isoforms may be classified according to the presence or absence of a transmembrane domain. Isoforms lacking a transmembrane domain are often termed "soluble cytokine receptors" and can fulfill different physiological functions (3,4). In general, soluble receptors may function as agonists by stabilizing their ligands, e.g. growth hormone and tumor necrosis factor (5,6), or contrarily act as antagonists by competing with the membrane-bound receptor for ligand binding, e.g. epidermal growth factor and interleukin (IL)-1 1 (7,8). Soluble receptors can arise from alternative splicing or from proteolytic receptor shedding on the cell surface. Isoforms generated by alternative splicing often contain additional protein sequence due to unspliced intron sequence and/or a shift of the reading frame. Generally, no biological function has been attributed to these additional stretches of amino acids.
Thrombopoietin (TPO) and its receptor "cellular homolog of myeloproliferative leukemia" (c-Mpl) are the primary regulators of megakaryopoiesis (9). The c-mpl gene is composed of 12 exons (see Fig. 1A) (10). In the mouse, two distinct alternate mRNA isoforms are known. The transmembrane variant mpl-II is due to usage of a cryptic splice acceptor in exon 4 resulting in an in-frame deletion of 60 amino acids (11). No function has yet been assigned to this isoform. The second mRNA variant encodes a truncated soluble receptor, Mpl-tr, and is the only one found both in human and mouse. This variant results from splicing of exon 8 directly to exon 11, eliminating the juxtamembrane extracellular part and the transmembrane domain (12,13). Due to an altered reading frame at the splice acceptor site of exon 11, Mpl-tr protein terminates in a short stretch of novel amino acid sequences (see Fig. 1). mpl-tr mRNA accounts for ϳ30% of mpl mRNA in mouse spleen (12). Despite the presence of a signal sequence and the lack of a transmembrane domain, Mpl-tr is not secreted into the cell supernatant when ectopically expressed in cell line (12). In humans, two alternate mRNA mpl species are known in addition to mpl-tr. The mpl-K variant is due to a readthrough beyond the exon 10 splice donor site (14). The resulting K-form of the receptor diverges from the native sequence after the ninth cytoplasmic amino acid and terminates within intron 10. mpl-del, a second isoform, arises as a consequence of alternative splicing between exons 8 and 9 and encodes a protein with an in-frame deletion of 24 amino acids and unknown function (15).
Because of the lack of secretion of Mpl-tr, we analyzed whether Mpl-tr plays a physiological role intracellularly. Here, we demonstrate that Mpl-tr specifically inhibits TPO-dependent proliferation and survival. We show that Mpl-tr is responsible for initiating protein down-regulation of the full-length Mpl receptor by a cathepsin-like cysteine protease activity. As a consequence, the amount of total Mpl protein in the cell is drastically reduced. Further, our data show that for this effect, a short peptide sequence at the C terminus of Mpl-tr is essential. The ability of Mpl-tr to antagonize Mpl function represents a novel mechanism by which cytokine signaling is regulated.

EXPERIMENTAL PROCEDURES
DNA Constructs-The plasmid pCD4 (pMICD4) is a gift from Dr. Harvey A. Lodish. It contains an internal ribosomal entry site followed by a truncated cDNA of the human CD4 gene and is derived from the retroviral expression vector pMX (16). To generate pCD4-mpl-tr, mpl-tr was cloned into the restriction sites XhoI and blunted BamHI of the multiple cloning site of pCD4. For the generation of the mpl-tr mutants, site-directed mutagenesis was performed using the QuikChange XL mutagenesis kit (Stratagene, Cedar Creek, TX) according to the manufacturer's protocol. The following primers were used: 5Ј-GAAGGCCG-TGAGGACTGGAAGTAGACTGAGGCAAGCTTTGTGG-3Ј (sense) and 5Ј-CCACAAAGCTTGCCTCAGTCTACTTCCAGTCCTCACGGCC-TTC-3Ј (antisense) for the stop codon in ⌬pep 30 ; 5Ј-GAAGGCCGTG-AGGACTGGAAGAGACTGAGGCAAGCTTTGTGG-3Ј (sense) and 5Ј-CCACAAAGCTTGCCTCAGTCTCTTCCAGTCCTCACGGCCTTC-3Ј (antisense) for the frameshift in tr-pep mpl ; 5Ј-GCCCTAAGTCCTTCTT-AAGGCCACGGTTACCGATAGCTGTG-3Ј (sense) and 5Ј-CACAGCTA-TCGGTAACCGTGGCCTTAAGAAGGACTTAGGGC-3Ј (antisense) for the stop codon in tr-pep mpl . mpl, mpl-tr, and pep 30 cDNAs were cloned into the 5Myc-pcDNA1 vector (gift from Dr. Eva Reinhard, Biozentrum, University of Basel), which contains at its 5Ј end a sequence encoding a hemagglutinin signal sequence followed by five Myc epitopes (17). For stable transfections, Myc-tagged mpl cDNA was cloned into the pGD expression vector (18) as a XhoI-NotI fragment. For transient trans-fections into human kidney 293T cells, mouse mpl, Myc-mpl, and mpl-tr cDNAs were subcloned into the pcDNA3 expression vector (Invitrogen) as XhoI-NotI fragments.
Cell Transfection and Culture-BaF3 cells (19) were cultured as described (20). UT-7 cells (21) were grown in RPMI 1640 supplemented with 10% fetal calf serum and 2 ng/ml recombinant human granulocytemacrophage colony stimulatory factor (GM-CSF) (PromoCell, Heidelberg, Germany). For transfections of BaF3 and UT-7 cells, 0.5 Ϫ 1 ϫ 10 7 cells were electroporated at 270 V and 975 microfarads at ambient temperature in the presence of 20 g of plasmid. UT-7/Myc-mpl cells were cultured in the presence of 450 g/ml G418. A pool of stably transfected UT-7/Myc-mpl cells was used for the transfection with different pCD4 constructs. The BaF3/mpl cell clone TM17, which has been described (20), was used for transfections with different pCD4 constructs. Cells expressing human CD4 were selected by the usage of anti-CD4 microbeads according to the manufacturer's protocol (Miltenyi, Auburn, CA). mpl-tr conditioned medium was obtained from UT-7 cells expressing pCD4-mpl-tr, which were cultured at exponential growth phase in human GM-CSF for 2 days. Control medium was harvested from untransfected UT-7 cells. For transient transfections of 293T cells, the transfection reagent FuGENE was used according to the manufacturer's protocol (Roche Applied Science). Where indicated, protease inhibitors (Calbiochem) were added at a final concentration of 25 M each, and cells were cultured for 6 -8 h before harvesting total cell lysate for immunoblot analysis. The inhibitors used were the cathepsin inhibitor cysteine cathepsin inhibitor I (CATI-1) (Z-Phe-Gly-NHO-Bz), the proteasome inhibitor MG132 (Z-Leu-Leu-Leu-CHO), and the calpain inhibitors ALLN (N-Acetyl-Leu-Leu-Nle-CHO) and EST ((2S,3S)trans-epoxysuccinyl-L-leucylamido-3-methylbutane ethyl ester). Proliferation Assay-An XTT proliferation kit (Roche Applied Science) was used according to the manufacturer's protocol to determine cytokine-dependent cell proliferation. In brief, cells were plated in 96well plates at 10 4 cells/well in 100 l of medium containing the indicated concentrations of cytokine. After 3.5 days of stimulation for BaF3 cells and 5 days of stimulation of UT7 cells, 50 l of a 1 mg/ml stock solution of XTT with 5 mmol/liter phenazine methosulfate, an electron coupling agent, was added to each well. The product of XTT reduction by viable cells, reflecting the number of cells per well, was measured at 4 h at 450 nm.
Protein Expression Analyses-Surface expression of Myc-Mpl was determined by flow cytometry. For this analysis, 5 ϫ 10 5 cells were incubated with the mouse monoclonal antibody 9E10 (22) directed against the N-terminal Myc tag of Myc-Mpl for 60 min on ice followed by incubation with the fluorescein isothiocyanate (FITC)-labeled goat anti-mouse antibody (BD Biosciences). Untransfected UT-7 cells served as a control. Total amounts of c-Mpl and Mpl-tr were determined by immunoblot analysis using a purified rabbit polyclonal antibody directed against Mpl, as described (12). The same polyclonal antibody and 9E10 were used for the detection of Myc-Mpl protein by immunoblot analysis. To control for protein loading, the membranes were reprobed using the mouse monoclonal antibody AC-40 directed against actin (Sigma).
Pulse-chase Analysis-293T cells cultured in 60-mm dishes were transiently transfected with 3 g of Myc-mpl and 3 g of mpl-tr expressed off the pcDNA3 expression vector. Pulse-chase analysis was begun 40 h after transfection. To perform the pulse-chase, cell monolayers were washed twice with warm phosphate-buffered saline and starved of methionine and cysteine by incubation for 40 min at 37°C in 1 ml of methionine/cysteine-free Dulbecco's modified Eagle's medium (Invitrogen) supplemented with 5% dialyzed fetal calf serum (Invitrogen). Following amino acid starvation, cellular proteins were pulselabeled by incubating each plate of cells with 400 Ci of methionine/ cysteine (Tran35S-label; PerkinElmer Life Sciences) for 30 min at 37°C. The radioactive medium was then removed, and the cells were washed twice in warm phosphate-buffered saline, re-fed with Dulbecco's modified Eagle's medium supplemented with 2 mM methionine and 2 mM cysteine, and incubated for the indicated times. Cells were collected, and labeled proteins were recovered by denaturing immunoprecipitation using the 9E10 antibody and the method of Hofmann et al. (23). Immune complexes were analyzed on 10% SDS-PAGE gels. The dried gels were quantitated on a Molecular Imager FX (Bio-Rad) using Quantity One software (Bio-Rad).
Apoptosis-For quantification of apoptosis, UT-7/Myc-Mpl cells were grown at 2 ϫ 10 5 cells/ml in the presence of either 2 ng/ml human GM-CSF or 2 ng/ml or 20 ng/ml human TPO (gift from Dr. Frederic J. de Sauvage). After 48 h, cells were washed twice and stained with annexin V with the use of the annexin V-FITC kit (Roche Applied Science). FITC-positive cells were quantitated by flow cytometry.

FIG. 2. Effect of mpl-tr on proliferation of cytokine-dependent cell lines.
We used two cytokine-dependent cell lines for cell growth assays: the murine IL-3-dependent cell line BaF3 and the human GM-CSF-dependent cell line UT-7. Neither cell line expresses Mpl protein endogenously. Upon expression of mpl, BaF3 and UT-7 cells proliferate in response to TPO. The isoform Mpl-tr was expressed off a retroviral vector containing an intraribosomal entry site followed by a truncated form of human CD4, used as a quantitative selectable marker. A, BaF3 cells stably transfected with mpl (BaF3/mpl) were additionally transfected with pCD4 or pCD4-mpl-tr and sorted with anti-CD4 microbeads. Proliferation of transfected and parental BaF3 cells was determined by an XTT assay with increasing concentrations (ng/ml) of IL-3 or TPO, as indicated. Error bars indicate the standard deviation on triplicate samples. B, UT-7 cells stably transfected with Myc-mpl (UT-7/Myc-mpl) were additionally transfected with pCD4 or pCD4mpl-tr and sorted with anti-CD4 microbeads. Proliferation of transfected and parental UT-7 cells in the presence of GM-CSF and TPO is shown. Annotation is as above. C, BaF3/mpl cells were cultured with increasing concentrations (ng/ml) of IL-3 or TPO in the presence of mpl-tr-conditioned medium (filled diamonds) or control medium (open diamonds).
from the C T value for ribosomal protein L19 (RPL19), which serves as an internal control (24,25). All reactions were run in duplicates.

Co-expression of mpl-tr with c-mpl Inhibits TPO-dependent
Mitogenic and Survival Signaling-To study the effect of mpl-tr on TPO-dependent proliferation, we used two cytokinedependent cell lines for cell growth assays: the murine cell line BaF3 and the human megakaryoblastic cell line UT-7. Neither cell line expresses Mpl protein endogenously. To assay the potential effects of mpl-tr on the function of the full-length Mpl protein, we used BaF3 cells stably transfected with murine mpl (BaF3/mpl) (20) (Fig. 1). These cells were subjected to a second round of transfection with an expression vector containing the cDNA for murine mpl-tr. The presence of an intraribosomal entry site followed by a truncated form of human CD4 in the vector allowed selection of mpl-tr-expressing cells using magnetic anti-CD4 microbeads. This procedure was repeated 3-4 times to enrich for CD4-positive cells, and expression of CD4 was confirmed by flow cytometry (data not shown). The sorted cells were cultured with different concentrations of either IL-3 or TPO. As expected, parental BaF3 cells failed to respond to TPO ( Fig. 2A). On the other hand, BaF3/mpl cells and BaF3/ mpl cells transfected with the parental pCD4 plasmid proliferated in dependence on TPO concentration. In contrast, BaF3/ mpl cells expressing mpl-tr failed to show a proliferative response to TPO, behaving similarly to parental BaF3 cells. Importantly, mpl-tr did not have a general inhibitory effect on proliferation because in the presence of IL-3, BaF3/mpl cells expressing mpl-tr grew as efficiently as control cells. The same result was obtained when mpl-tr was expressed in human UT-7 cells that were stably transfected with mpl. In this experiment, we used an mpl construct with five Myc tags at the N terminus (UT-7/Myc-mpl) (Fig. 1). The Myc-Mpl protein conveyed TPOresponsiveness to UT-7 cells, demonstrating that the Myc tag did not interfere with Mpl function (Fig. 2B). When mpl-tr was expressed in UT-7/Myc-mpl cells, TPO-mediated proliferation was abolished, but GM-CSF-dependent proliferation remained unchanged. To test whether the observed inhibitory effect could be mediated by secreted Mpl-tr protein, medium conditioned by mpl-tr-expressing cells was transferred onto BaF3/mpl cells. TPO-dependent proliferation of BaF3/mpl cells was not inhibited by the presence of this conditioned medium (Fig. 2C), demonstrating that no secreted inhibitory activity exists. Since Mpl signaling exerts an anti-apoptotic effect, we analyzed the effect of mpl-tr on TPO-dependent cell survival. UT-7/Myc-mpl cells transfected with either mpl-tr or control vector were cultured with GM-CSF or with TPO. After 48 h, cells were incubated with annexin V and analyzed by flow cytometry. In the presence of GM-CSF, mpl-tr had no effect on the number of annexin V-positive cells. However, with TPO, most cells coexpressing mpl and mpl-tr stained positive for annexin V (Fig.  3). This indicates that mpl-tr expression inhibits the antiapoptotic signal delivered by TPO.
mpl-tr Mediates Down-modulation of Mpl Protein Expression in a Post-transcriptional Manner-To investigate the inhibitory mechanism exerted by mpl-tr, we analyzed the effects of mpl-tr on Mpl protein expression. First, we asked whether Mpl protein surface expression was affected by the presence of mpl-tr. Cells were stained with anti-Myc antibodies (Fig. 4A). with the mouse monoclonal anti-Myc antibody 9E10 and a secondary FITC-labeled goat anti-mouse antibody to detect Myc-Mpl on the plasma membrane. Cells were analyzed by flow cytometry, and one representative result is shown. B, the same UT-7/Myc-mpl cells transfected with pCD4 or pCD4-mpl-tr as in A were cultured either on 2 ng/ml GM-CSF or 2 ng/ml TPO for 24 h, and total Mpl protein expression in total cell lysate was determined by immunoblot analysis. Proteins of 93 kDa corresponding to Myc-Mpl and of 55 kDa for Mpl-tr were detected. The same membrane was probed for actin to control for equal loading. To determine an effect of mpl-tr on mpl mRNA expression, the amount of Myc-c-mpl mRNA was determined by Q-PCR. The ⌬C T values were derived by subtracting the C T values for c-mpl from the C T value for ribosomal protein L19 (RPL19), which serves as an internal control. All reactions were run in duplicates. The ⌬C T values are shown underneath the corresponding lanes. C, 293T cells were transiently transfected with Myc-mpl or transiently co-transfected with Myc-mpl and mpl-tr or control plasmid, respectively. Annotation is as above.
As expected, UT-7/Myc-Mpl cells showed marked Myc-mpl surface expression. In contrast, the staining of UT-7/Myc-mpl cells expressing mpl-tr did not significantly differ from the control staining of the parental UT-7 cells, indicating that mpl-tr interferes with Myc-mpl cell surface expression. We then asked whether the lack of detectable Myc-Mpl surface expression correlated with a decrease in total Myc-Mpl protein. Immunoblot analysis of total cell lysates demonstrated a massive reduction of Myc-Mpl protein in cells that expressed mpl-tr (Fig.  4B). The decrease in Myc-Mpl protein was observed irrespective of whether the cells were grown with TPO or GM-CSF. Importantly, mpl-tr did not alter mpl mRNA levels, as indicated by the unchanged ⌬C T values for mpl (Fig. 4B). Expression of mpl-tr also led to a dramatic decrease in expression of the untagged Mpl protein in BaF3/mpl cells without affecting mRNA levels, showing that mpl-tr targets mpl protein regardless of the presence of an N-terminal Myc tag (see Fig. 6C). To determine whether this phenomenon was limited to hematopoietic cells, we performed transient co-transfections of mpl and mpl-tr cDNAs into human 293T cells. As shown in Fig. 4C the effect of mpl-tr on Mpl expression, we transiently co-transfected 293T cells with a constant amount of plasmid encoding mpl and varying amounts of plasmid for the expression of mpl-tr. In this experiment, Mpl protein amount was affected by mpl-tr in a dose-dependent manner (Fig. 5A). Measuring mRNA levels by Q-PCR confirmed that mRNA expression correlated with the amount of plasmid DNA transfected. For example, the mpl-tr (⌬C T ) value for cells transfected with 0.1 g of mpl-tr was Ϫ4. Cells transfected with 30 times the amount of mpl-tr (3 g) had an mpl-tr (⌬C T ) value of Ϫ9. This corresponds to a decrease of 5 C T , which equals 2 5 ϭ 32 times higher expression of mpl-tr (Fig. 5A). In the converse experiment, we transfected 293T cells with a constant amount of mpl-tr and varied the concentration of mpl. We found that steady-state expression of Mpl-tr was not altered by the presence of increasing amounts of mpl (Fig. 5B). Since mpl-tr affects the steadystate levels of Mpl protein, we determined the half-life of Mpl by a pulse-chase experiment (Fig. 5C). In the presence of mpltr, the half-life of Mpl was decreased from 5 to 6 h to 2 to 3 h (Fig. 5C).

CATI-1 Restores Mpl Protein Expression in the Presence of mpl-tr-
To identify the mechanism that underlies diminished Mpl protein expression in the presence of mpl-tr, we treated mpl-tr-transfected UT-7/Myc-mpl cells with the cathepsin inhibitor CATI-1, the proteasome inhibitor MG132, or the calpain FIG. 5. Dose response, protein turnover, and sensitivity to protease inhibitors of Mpl down-regulation. A, 293T cells were transiently co-transfected with varying amounts of plasmid DNA encoding mpl and mpl-tr (numbers above the lanes indicate DNA amount in g). mpl, Mpl-tr, and actin proteins were detected by immunoblot analysis. The mRNA expression of c-mpl and c-mpl-tr was determined by Q-PCR; ⌬C T values for mpl and mpl-tr are shown below the corresponding lanes of the immunoblots. The amount of mpl DNA was kept constant, whereas the amount of mpl-tr DNA varied. B, in this experiment, mpl DNA was varied and the amount of mpl-tr DNA was kept constant. C, pulse-chase analysis of the turnover of Myc-Mpl protein in the presence and absence of mpl-tr. 293T cells transiently transfected with pcDNA3-Myc-mpl alone (Ϫ) or in combination with pcDNA3-mpl-tr (ϩ mpl-tr) were metabolically labeled with methionine/cysteine (48) and then chased in the presence of excess cold methionine and cysteine for 3, 6, 9, and 12 h. The amount of labeled Myc-Mpl was determined by immunoprecipitation (IP) with 9E10 and SDS-gel analysis. D, UT-7/Myc-mpl cells stably transfected with pCD4-mpl-tr were treated with protease inhibitors at a final concentration of 25 M each or with Me 2 SO (DMSO) as indicated. After 6 -8 h of incubation, total cell lysates were subjected to immunoblot analysis. E, UT-7/Myc-mpl cells stably transfected with pCD4 were treated as described for the pCD4-mpl-tr-transfected cells. F, 293T cells were transiently co-transfected with mpl and mpl-tr and then treated with CATI-1 or Me 2 SO for 6 -8 h. Annotation is as above.
inhibitors ALLN or EST. Inhibitors were added at a final concentration of 25 M each, and cells were cultured for 6 -8 h in the presence of GM-CSF. Only CAPI-1 restored Mpl protein expression (Fig. 5D), and a weak Mpl band was detectable with MG132 treatment, whereas the other inhibitors had no effect. None of the inhibitors changed the steady-state Mpl protein levels in the UT-7/Myc-mpl cells lacking mpl-tr (Fig. 5E). To confirm these results in a different cell system, we also treated 293T cells with CATI-1 and determined Mpl protein expression. Similar to UT-7/Myc-mpl, 293T cells co-transfected with mpl and mpl-tr showed a rescue of Mpl protein expression in the presence of CATI-1 (Fig. 5F). These results argue that mpl-tr mediates Mpl protein degradation by a cathepsin-like protease activity.

The C-terminal Peptide Sequence of Mpl-tr Is Necessary but Not Sufficient for the Inhibition of Cell Proliferation and for
Mpl Protein Degradation-The amino acid sequence of Mpl-tr is identical to the N terminus of Mpl except for a stretch at the C terminus of Mpl-tr, 30 amino acids in length (Fig. 1). We therefore speculated that this unique C-terminal peptide in Mpl-tr could be of functional importance. To test this hypothesis, we made two mutants of mpl-tr. In the first mutant, ⌬pep 30 , we introduced a stop codon at position 427, removing the entire C-terminal peptide (Fig. 6A). In the second mutant, tr-pep mpl , the sequence of the C-terminal peptide was changed by adding 2 bp, which restored the reading frame of full-length mpl, and by introducing a stop codon terminating the reading frame after 30 amino acids (Fig. 6A). With these mpl-tr mutants, we stably transfected BaF3/mpl cells and assayed TPO-dependent proliferation. As shown in Fig. 6B, only mpl-tr abrogated mplmediated cell growth, whereas tr-⌬pep 30 and tr-pep mpl did not interfere with TPO-dependent proliferation. As expected, the mpl-tr mutants did not interfere with the proliferative responses to IL-3 (Fig. 6B). Immunoblot analysis showed that ⌬pep 30 and trpep mpl were expressed at levels similar to Mpl-tr (Fig. 6C). This indicates that the degree of expression does not explain the failure of the mutants to inhibit TPO-dependent growth. Importantly, the levels of Mpl protein in cells expressing the mpl-tr mutants were similar to BaF3/mpl control cells, but Mpl was undetectable in cells expressing mpl-tr. These results demonstrate that the C-terminal peptide sequence from Mpl-tr is required for the ability of Mpl-tr to promote a decrease in Mpl protein and to inhibit TPO-dependent cell proliferation. Furthermore, these results illustrate that the function of mpl-tr in growth inhibition directly correlates with its role in Mpl protein reduction. To determine whether pep 30 is sufficient to interfere with TPO-induced proliferation and to mediate Mpl degradation, we generated a Myc-tagged versions of pep 30 and mpl-tr (Fig. 7A). Myc-pep 30 did not affect TPO-dependent proliferation (Fig. 7B) nor alter the expression levels of Mpl protein (Fig.  7C), whereas Myc-Mpl-tr behaved comparably with untagged Mpltr. Thus, pep 30 sequence alone is not sufficient to inhibit Mpl function.

DISCUSSION
Alternatively spliced cytokine receptor variants are emerging as regulators of cytokine signaling (26 -31). The c-mpl locus gives rise to the full-length Mpl protein and the Mpl-tr isoform, which is present in both human and mouse (10,12,13). Here, we characterized Mpl-tr as a specific inhibitor for TPO-dependent proliferation and described a novel mechanism of receptor protein down-modulation mediated by an alternate isoform of the same receptor through a cathepsin-like protease activity. Alternatively spliced isoforms are frequently found among members of the cytokine receptor superfamily, and most family members have been shown to be capable of giving rise to soluble forms (3). So far, two functions have been assigned to soluble cytokine receptor isoforms. First, soluble receptors can prolong the half-life of the ligand and activate signal transduction by associating with a transmembrane receptor complex, as exemplified by the soluble IL-6 receptor (32,33). Alternatively, soluble receptors can act as antagonists of their membranebound counterparts by binding and neutralizing the ligand, e.g. IL-5 or GM-CSF (34 -37). In either model, the function of the soluble receptor is dependent on its secretion. However, the serum levels of some soluble receptors are low or undetectable (3,38), suggesting additional functions of soluble receptors as non-secreted forms.
Because the deletion in mpl-tr removes the transmembrane domain, Mpl-tr is expected to give rise to a secreted form, which might antagonize Mpl signaling by sequestering TPO. However, this model cannot explain the observed dominant-negative effect of mpl-tr for three reasons: 1) Mpl-tr could not be detected in the cell supernatant (12); 2) high concentrations of TPO did not titrate out the effect of mpl-tr (Fig. 2); and 3) medium conditioned by mpl-tr-expressing cells did not affect TPO-dependent growth of BaF3/mpl cells (Fig. 2C).
We demonstrated that overexpression of mpl-tr leads to a dramatic decrease in Mpl protein amounts, which can explain the dominant-negative effect of mpl-tr on TPO-dependent cell proliferation and survival. The decrease in Mpl protein was not due to differences in c-mpl mRNA expression (Figs. 4 and 5). A specific inhibitor for cysteine protease cathepsins, CATI-1, can counteract the effects of mpl-tr and restore full-length Mpl protein expression. In contrast, inhibitors of Ca 2ϩ -dependent cysteine proteases, ALLN and EST, had no detectable effect on Mpl protein amounts. It is therefore tempting to speculate that in the presence of mpl-tr, Mpl traffics to the lysosome, where it is degraded. The small increase in Mpl expression in the presence of the proteasomal inhibitor MG132 may be explained by the finding that some receptor trafficking events from endosomes to lysosomes require a functional proteasome (39,40).
The observed effects of mpl-tr appear to specifically target the Mpl protein since IL-3 and GM-CSF-dependent cell proliferation and survival were not affected. This specificity could be based on an inherent capability of the extracellular domains of cytokine receptors to form dimers. It has been shown by crystallography that EpoR forms dimers in the absence of ligand (41). Analogously to the structure-function studies performed for EpoR, we expect that the extracellular portion of Mpl-tr is sufficient to allow heterodimerization with the full-length form of Mpl. So far, we have been unable to detect this physical interaction through co-immunoprecipitation (data not shown), possibly because the interaction is only transient.
We demonstrated that the unique sequence at the C terminus of Mpl-tr is essential for mediating Mpl degradation. In one possible scenario, the C terminus of Mpl-tr prevents the secretion of Mpl-tr and directs the putative Mpl/Mpl-tr dimer to traffic along the endosomal-lysosomal route. Increasing acidification of the endosomal vesicle could cause the heterodimer to disassemble, allowing Mpl-tr to be recycled. This model would explain why Mpl-tr protein expression is apparently not affected by the presence of Mpl protein (Fig. 5B). Further structure function analysis the C-terminal sequence of Mpl-tr should allow us to define a minimal functional peptide motif.
Alternatively, Mpl-tr may interfere with the proper folding of Mpl. In both mammalian cells and in yeast, it has been shown that a post-endoplasmic reticulum quality control mechanism appears to be responsible for targeting and lysosomal/vacuolar degradation of misfolded membrane proteins (42)(43)(44)(45). The presence of a cytosolic portion may be a prerequisite for the detection of partly unfolded Mpl by the post-endoplasmic reticulum quality surveillance system. Since Mpl-tr only contains the luminal portion of Mpl, it may evade detection by the quality control system. This would explain why Mpl-tr steadystate levels are not altered in the presence of full-length Mpl.
Cytokine receptor variants may be important in the regulation of receptor function. Interestingly, Nakamura et al. (28) demonstrated that transgenic mice overexpressing the intracytoplasmic truncated erythropoietin receptor isoform EpoR-T reveal mild anemia. Because overexpression of EpoR-T diminishes the number of erythrocytes in the mouse, these results suggest a role of cytokine receptor variants as regulators of proliferation and differentiation in vivo.
In this study, we have demonstrated that an alternate Mpl isoform (Mpl-tr) exerts a dominant-negative effect on proliferation and survival. Because mpl-tr is an abundant c-mpl splice variant accounting for about 30% of total mpl mRNA in mouse spleen (12), mpl-tr is likely to play a physiological role in megakaryopoiesis. Differential expression has been shown for alternatively spliced isoforms of cytokine receptors such as IL-5 receptor and GM-CSF receptor (29,46). The relative expression of different isoforms of the same cytokine receptor can determine cell differentiation and cell lineage expansion (29,47). Analogously to isoforms of other receptors, we speculate that the relative expression of c-mpl-tr varies with different cell types and developmental stages and that this serves to modulate Mpl function.