Three Dileucine-like Motifs within the Interbox1/2 Region of the Human Oncostatin M Receptor Prevent Efficient Surface Expression in the Absence of an Associated Janus Kinase*

The oncostatin M receptor (OSMR) is part of receptor complexes for oncostatin M and interleukin-31. Signaling events are triggered by Jaks (Janus kinases) that constitutively bind to membrane-proximal receptor regions. Besides their established role in signaling, Jaks are involved in the regulation of the surface expression of several cytokine receptors. Here, we analyzed the structural requirements within the human OSMR that underlie its limited surface expression in the absence of associated Jaks. We identified three dileucine-like motifs within the Jak-binding region of the OSMR that control receptor surface and overall expression. A receptor mutant in which all three motifs were mutated to alanine displayed markedly increased surface expression. Although the surface half-life of this mutant was increased compared with that of the wild-type receptor, no difference in the internalization rate was detectable, implying that these receptors differ in their post-endocytic fate. The protein stability of the wild-type receptor was markedly lower than that of mutant receptors, but could be strongly increased in the presence of the lysosomal inhibitor chloroquine. Our data are consistent with the dileucine motifs being involved in destabilization of receptors devoid of associated Jaks as part of a quality control ensuring signaling competence of OSMRs.

The oncostatin M receptor (OSMR) 3 is one of the signal-transducing receptor chains used by interleukin (IL)-6-type cytokines. Besides IL-6, this family comprises the cytokines IL-11, leukemia inhibitory factor, cardiotrophin-1, ciliary neurotrophic factor, cardiotrophin-like cytokine, IL-27, neuropoietin, and oncostatin M (1)(2)(3). Although IL-6 and IL-11 induce the homodimerization of the common receptor chain gp130, a heteromeric receptor complex containing gp130 and the leukemia inhibitory factor receptor (LIFR) is used by the other cytokines in the family (1), with the exception of IL-27, which signals via a gp130⅐WSX-1 complex (4). Human oncostatin M is the only known ligand of this family that signals via the gp130⅐OSMR receptor complex (5). However, it was recently found that the novel cytokine IL-31, which is produced mainly by T cells and seems to be involved in the development of chronic skin diseases (6), signals via a receptor complex consisting of the previously discovered cytokine receptor GPL (gp130-like) (7,8) and the OSMR (9,10).
Like other cytokine receptors, the signal-transducing receptor chains of IL-6-type cytokines do not contain intrinsic kinase activity, but constitutively associate with members of the Jak (Janus kinase) family. These bind through their N-terminal FERM domain to cytokine receptors (11)(12)(13). We have recently shown that also the SH2-like domain of Jak1 is important for association with the OSMR, although it does not fulfil the classical role of SH2 domains (i.e. binding to phosphotyrosine motifs), but rather a structural one (14). The membrane-proximal regions of gp130, the LIFR, and the OSMR contain the well conserved box1 region and a less well characterized box2 region, which have been implicated in Jak binding (15)(16)(17). In case of gp130, it has been shown that also amino acids in the interbox1/2 region are crucial for association with Jak1 (18).
Upon ligand binding, associated Jak kinases become activated and mediate phosphorylation of specific tyrosine residues within the cytoplasmic region of the receptor chains. These phosphorylated tyrosines may then serve as docking sites for SH2 or phosphotyrosine-binding domain-containing proteins. In case of the OSMR, phosphorylation-dependent binding of latent STAT3 (signal transducer and activator of transcription) and the adaptor protein SHC has been demonstrated (19,20).
Although it is well established that Jak1 plays an important role in the signaling of IL-6-type cytokines (21,22), we previously demonstrated that the binding of Jak1 to the OSMR is also needed for its efficient localization at the cell surface (20). Using chimeric receptors, we showed that Jak1 coexpression leads to an increased number of fully processed receptors. This effect of Jak1 is not dependent on its kinase activity, but requires an intact FERM domain, indicating that interaction with the OSMR is necessary. We concluded that the OSMR might contain a negative regulatory signal within its membrane-proximal region that can be masked by an associated Jak. Indeed, deletion of the membrane-proximal region results in receptor constructs that are efficiently expressed on the plasma membrane. Interestingly, similar findings have been reported for other cytokine receptors. Thus, the Jak kinase Tyk2 supports the expression of IFNAR1 (interferon-alpha receptor-1) (23,24), and Jak2 is needed for the efficient expression of the erythropoietin receptor (EpoR), thrombopoietin receptor, and growth hormone receptor on the cellular membrane (25)(26)(27). Even though not required for the surface expression of the common ␥-chain, Jak3 expres-sion positively influences its surface expression upon overexpression (28).
Although it has been demonstrated that Jaks may support receptor surface expression by altering the receptor internalization rate (IFNAR1) and receptor protein stability (e.g. thrombopoietin and growth hormone receptors), it is not clear yet how Jaks may affect these processes. The finding that deletion of membrane-proximal receptor regions positively affects the surface expression of the OSMR, EpoR, and IFNAR1 chain (20,24,25) points to the possibility that these regions contain negative regulatory signals that are masked by Jaks upon association with the receptor. However, no such signal has been identified yet.
We have characterized the mechanism underlying the regulation of OSMR surface expression by Jak1 in more detail. We demonstrate that the membrane-proximal region of the OSMR indeed contains a negative regulatory signal because transfer of this region to a well expressed receptor construct resulted in marked down-regulation. We have furthermore identified a stretch of 20 amino acids in the interbox1/2 region that regulates receptor surface and overall expression. Within these 20 amino acids, we identified three dileucine-like motifs implicated in the weak expression of the OSMR. Mutation of these motifs resulted in enhanced protein stability and an increased half-life at the cell surface. However, we detected no effect on receptor internalization. We propose that the three dileucine-like signals create a negative regulatory signal by conferring increased susceptibility to proteasomal and/or lysosomal degradation if not masked by associated Jak kinases.

EXPERIMENTAL PROCEDURES
Cell Culture and Transfection-COS-7 simian monkey kidney cells, human embryonic kidney 293 (HEK293) cells, and 2C4 (parental) and U4C (Jak1-negative) fibrosarcoma cells were maintained in Dulbecco's modified Eagle's medium supplemented with 10% (v/v) fetal calf serum, 100 mg/liter streptomycin, and 60 mg/liter penicillin. Cells were grown at 37°C in a water-saturated atmosphere of air/CO 2 (19:1). COS-7 cells were transiently transfected with expression vectors based on pSVL (Amersham Biosciences, Freiburg, Germany) using a modified DEAEdextran method as described previously (19,29). For generation of stable HEK293 cells, the Flp-In TM system from Invitrogen was used, which allows insertion of the transfected plasmids into a single chromosomally integrated Flp recombinase target site, thus minimizing clone-to-clone differences between transfectants. The Flp-In TM T-REx TM -293 cell line was purchased from Invitrogen. These cells contain a single integrated Flp recombinase target site and also stably express the Tet repressor. Pools of stably transfected cells were generated using pcDNA5/FRT/TO-based expression vectors (Invitrogen) according to the manufacturer's instructions. Transfection was carried out using FuGENE 6 transfection reagent (Roche Applied Science) according to the manufacturer's instructions. Expression was induced by incubating cells for 5 h (internalization assays) or 18 h (Western blot analysis) with doxycycline (2 ng/ml).
Expression Constructs-The construction of the pcDNA3 expression vectors for Jak1-green fluorescent protein (GFP) and Jak1-L80A/Y81A-GFP and the pSVL expression vectors for the interleukin-5 receptor-␤ (IL-5R␤) chimeras ␤OR-B1 (IL-5R␤-OSMR box1), ␤OR⌬cyt (IL-5R␤-OSMR⌬cyt), ␤130⌬cyt (IL-5R␤-gp130⌬cyt), ␤LIFR-B1/2 (IL-5R␤-LIFR box1/2), and ␤130-B1/2 (IL-5R␤-gp130 box1/2); wild-type Jak1; and Jak1-L80A/Y81A was described previously (11,14,20,29,30). The C-terminal deletion mutants ␤OR-B1/2, ␤OR-B1-I, and ␤130-B1 were generated by PCR using cDNA for the previously described IL-5R␤-OSMR or IL-5R␤-gp130 construct as a template (29,31). The antisense oligonucleotides used incorporate an in-frame termination codon, fol-lowed by the recognition site for BamHI. Internal deletions and point mutations were introduced by fusion PCR using the respective oligonucleotides and the cDNA for IL-5R␤-OSMR as a template. The same mutations were introduced into the wild-type OSMR by exchanging DNA fragments coding for the IL-5R␤ extracellular region with the respective fragments coding for the OSMR extracellular region. For generation of GFP-tagged OSMR, a BstEII site was introduced by PCR 3Ј to the codon for Cys 979 , allowing in-frame insertion of cDNA for enhanced GFP. The construct ␤130-B1ϩIB2 was generated by PCR using the cDNA for ␤130-B1 as a template and reverse oligonucleotides incorporating DNA coding for the respective amino acids, followed by a stop codon and a BamHI restriction site. For the constructs ␤130⌬cytϩB1/2 and ␤130⌬cytϩB1/2-3AA, the DNA fragments comprising OSMR Lys 767 -Thr 823 were generated by PCR using the chimeric OSMR as a template, a sense oligonucleotide that incorporates a BspTI restriction site, and a reverse oligonucleotide incorporating a stop codon and a BamHI restriction site. PCR fragments were inserted into BspTI/BamHI-digested expression plasmid pSVL-IL-5R␤gp130BspTI. In the resulting constructs, Ile 647 of gp130 was mutated to Leu due to the inserted BspTI site. The gp130-B1/2, gp130-IB2, and gp130-B1/2-3AA constructs were constructed from the respective ␤-chimeras by subcloning a EcoRI/BamHI fragment into pSVL-gp130. For generation of pcDNA5/FRT/TO-OR-GFP and pcDNA5/FRT/TO-OR-3AA-GFP, the empty vector was digested with HindIII and ApaI, and a pair of hybridized oligonucleotides containing an XhoI and a BamHI site was incorporated. The resulting vector (pFRiTte-spezial) was digested with XhoI/BamHI, and cDNA for OSMR-GFP and OSMR-3AA-GFP was inserted by standard cloning techniques. The integrity of all constructs was verified by DNA sequence analysis using an ABI PRISM 310 genetic analyzer (PerkinElmer Life Sciences).
Flow Cytometry-Cells (5 ϫ 10 5 to 1 ϫ 10 6 ) were resuspended in cold phosphate-buffered saline (PBS) supplemented with 5% fetal calf serum and 0.1% sodium azide (PBS/azide) and incubated with 1 g/ml anti-OSMR or anti-IL-5R␤ antibody for 30 min at 4°C. Cells were washed with cold PBS/azide and subsequently incubated in darkness with a 1:100 dilution of R-phycoerythrin-conjugated anti-mouse antibody for 30 min at 4°C. Cells were washed again and resuspended in PBS/azide. Cells (10 4 /sample) were analyzed by flow cytometry using a FACSCalibur (BD Biosciences) equipped with a 488-nm argon laser. Mean fluorescence intensities of transfected cells were calculated using CellQuest software. For internalization assays, stable HEK293 cells expressing OSMR-GFP or OSMR-3AA-GFP were incubated for 30 -60 min at 4°C in binding medium (Dulbecco's modified Eagle's medium, 0.2% bovine serum albumin, and 20 mM HEPES (pH 7.3)) containing monoclonal anti-OSMR antibody. Cells were washed twice with binding medium and incubated for different time periods at 37°C. Cells were then resuspended in PBS/azide, and surface expression was detected as described above. To measure protein stability, cells stably expressing GFP-tagged receptors were incubated for various periods of time with cyclohexi-mide (50 g/ml) and harvested, and the GFP fluorescence was measured by fluorescence-activated cell sorting (FACS) analysis. To block proteasomal and lysosomal degradation, cells were additionally incubated in MG132 (50 M) or chloroquine (100 M). To measure the surface halflife, cells were incubated with either brefeldin A (5 g/ml) or monensin (50 M) or treated with ethanol for up to 2 h. Cells were harvested and processed as described above for FACS analysis.
Cell Lysis, Immunoprecipitation, Endoglycosidase H (Endo-H) Digestion, and Western Blot Analysis-Cells were lysed in 500 l of lysis buffer (20 mM Tris-HCl (pH 7.5), 150 mM NaCl, 10 mM NaF, 1 mM sodium vanadate, 5 g/ml aprotinin, 3 g/ml pepstatin, 5 g/ml leupeptin, 1 mM EDTA, and 1% Triton-X 100 or 1% Brij 97 (for co-immunoprecipitations)). Chimeric receptors were immunoprecipitated from cleared lysates using monoclonal anti-IL-5R␤ antibody; for OSMR-GFP constructs, anti-OSMR antibody was used. The immune complexes were collected on protein A-Sepharose and washed twice with wash buffer (lysis buffer with only 0.1% Brij 97). For Endo-H digestions, the immunoprecipitates were washed once with 200 mM sodium citrate (pH 5.5) and incubated in citrate buffer for 18 -24 h with or without 5 milliunits of Endo-H. Subsequently, the Sepharose was boiled for 5 min in Laemmli buffer at 95°C. Proteins were separated by SDS-PAGE on 7.5% gels, followed by electroblotting onto a polyvinylidene difluoride membrane (Pall Corp., Dreieich, Germany). Western blot analysis was conducted using the indicated antibodies and an enhanced chemiluminescence kit (Amersham Biosciences) according to the manufacturer's instructions. Before reprobing, blots were stripped in 2% SDS and 100 mM ␤-mercaptoethanol in 62.5 mM Tris-HCl (pH 6.7) for 20 min at 75°C.

The Membrane-proximal Region of the OSMR Contains a Transferable Signal That Negatively Regulates Receptor Surface and Overall
Expression-We have proposed previously that the positive effect of Jak1 on the surface expression of the OSMR is due to the masking of a negative signal within the OSMR box1/2 region (20). To identify the location of this signal, we used chimeric receptor constructs in which the IL-5R␤ extracellular region was fused to the transmembrane and intracellular regions of the OSMR. The expression of these chimeric receptors could be easily monitored by FACS and Western blot analysis. We compared the expression of the three deletion constructs depicted schematically in Fig. 1A: the construct ␤OR-B1/2 contained the intact membrane-proximal box1/2 region needed for Jak association; ␤OR-B1 was deleted after the box1 region; and ␤OR⌬cyt lacked all but three amino acids of the OSMR cytoplasmic region. In accordance with our previous results (20), FACS analysis showed that the construct containing the intact box1/2 region had significantly reduced surface expression compared with the constructs in which the box1/2 region was partially or fully deleted (Fig. 1A, upper panel). Moreover, the latter two constructs clearly appeared as two bands in Western blot analyses (Fig.  1A, lower panel). The top bands correspond to the fully processed Endo-H-resistant receptor forms, which are hardly visible for the ␤OR-B1/2 construct (see Fig. 2C) (16).
Because the box1/2 region of the OSMR is responsible for significantly reduced cell surface expression of the receptor, we investigated whether this motif, if transferred to a well expressed cytokine receptor, retains its negative regulatory activity. We chose the chimeric receptor ␤gp130⌬cyt as the "recipient" receptor because we knew from our pre-FIGURE 1. The OSMR contains a transferable signal in its box1/2 region that negatively regulates receptor expression. COS-7 cells were transiently transfected with expression vectors for the chimeric receptors as indicated. 2 days after transfection, cells were harvested. A, to monitor receptor surface expression, cells were labeled using anti-IL-5R␤ and R-phycoerythrin-conjugated secondary antibodies. The relative (rel.) mean fluorescence of transfected cells was measured by FACS analysis, and the values obtained for the ␤OR-B1/2 construct were set to 100%. The graph shows the relative means Ϯ S.D. of at least three independent experiments. To monitor the overall expression of transfected receptors, cells were lysed in Triton lysis buffer, and equal amounts of lysates were separated by SDS-PAGE. The blots were stained with antiserum against IL-5R␤. A representative blot is shown. B, receptor surface expression was monitored as described for A using monoclonal antibody against the extracellular region of the IL-5R␤ chain or gp130. FACS values were calculated relative to cells expressing the ␤130⌬cyt and gp130⌬cyt constructs, respectively. For Western blot (WB) analysis, cells were lysed in Triton lysis buffer; lysates were separated by SDS-PAGE; and the blots were stained as indicated. C, the indicated ␤-chimeras were coexpressed with a non-binding mutant of Jak1 (Jak1-L80A/Y81A (control (ctrl)) or wild-type Jak1. FACS and Western blot analyses were carried out as described above.
vious studies that this receptor is extremely well expressed at the cell surface ( Fig. 1B, upper panels, bar 1). Transfer of the whole box1/2 region of the OSMR to this receptor construct resulted in a strong decrease in cell surface expression (Fig. 1B, upper panels, bar 2). Interestingly, when analyzing whole cell lysates, we noted that this new chimeric receptor also displayed a drastically reduced overall expression rate (Fig. 1B, lower left panel). As seen before (Fig. 1A), we could exclude that the box1 region of the OSMR is involved in this negative effect because the same suppressive effect on cell surface as well as overall expression was obtained with a chimera in which only 43 amino acids of the OSMR had been fused to the gp130 part (␤130-B1ϩIB2) (Fig. 1B,  bar/lane 3). This chimeric construct retained the gp130 box1 region to which the interbox and box2 region of the OSMR were fused. Identical observations were made with a second chimeric system containing the full-length gp130 extracellular and transmembrane regions fused to the OSMR box1/2 region (Fig. 1B, right panel).
We tested whether the expression of the chimeras could be ameliorated by Jak1 coexpression. Indeed, upon coexpression of wild-type Jak1, the surface expression of both triple chimeric constructs (␤130⌬cytϩB1/2 and ␤130⌬cyt-B1ϩIB2) could be increased; however, even in the presence of Jak1, the surface and overall expression did not reach the levels of the well expressed deletion construct ␤130⌬cyt (Fig.  1C). We concluded that the stretch of 43 amino acids C-terminal to the OSMR box1 region contains a transferable negative signal that prevents efficient receptor surface expression, an effect that can be at least partially reversed by coexpression of Jak1.
Twenty Amino Acids in the Interbox1/2 Region Play a Major Role in Negatively Regulating the Surface Expression of the OSMR-To further delineate the negative regulatory signal within these 43 amino acids, we generated two additional deletion constructs depicted in Fig. 2A. Whereas the construct ␤OR-B1/2⌬I lacked 20 amino acids within its interbox1/2 region, ␤OR-B1-I still contained the interbox1/2 region but lacked the box2 region. Both new constructs, as well as the previously studied constructs ␤OR-B1/2 and ␤OR⌬cyt, were transiently expressed in COS-7 cells, and surface expression and overall expression were analyzed. While the surface expression of the ␤OR-B1-I construct was only slightly better than that of ␤OR-B1/2 (Fig. 2B, upper panel, compare bars 1 and 4), the expression of ␤OR-B1/2⌬I was strikingly enhanced (bar 3). This construct behaved just as the shorter construct ␤OR⌬cyt used as a positive control: it was as well expressed at the surface (Fig. 2B, upper panel, bars 2 and 3), and there was a pronounced slower migrating second band visible in Western blot analysis (lower panel, lanes 2 and 3). Endo-H digestion revealed that this upper receptor band represents the Endo-H-resistant and therefore fully processed receptor (Fig. 2C).

Mutation of Three Dileucine-like Motifs Has a Dual Effect on Receptor Expression-
The expression pattern of the chimeric OSMR can be best explained by the presence of either an endoplasmic reticulum (ER) retention/retrieval or internalization/lysosomal targeting signal within the inter-box1/2 region. Of note, the interbox1/2 region of the OSMR contains two lysine residues as well as three dileucine-like motifs. Whereas lysine residues may mediate ER retrieval, dileucine-like motifs have been shown to mediate internalization and/or lysosomal targeting. To test whether these amino acids are responsible for the regulation of OSMR surface expression, the two lysines (OSMR-KK mut ) and the three dileucine-like motifs (OSMR-3AA) were independently mutated to alanine. These mutations were introduced in the context of the wild-type OSMR. Additionally, as a positive control, a construct lacking 20 amino acids within the interbox1/2 region (OSMR-⌬I) was generated (Fig. 3A, left panel). All four mutants were separately expressed in COS-7 cells, and the surface expression was monitored by FACS analysis using antibodies recognizing the extracellular region of the human OSMR. As shown in Fig. 3A (right panel), deletion of the inter-box1/2 region resulted in markedly increased cell surface expression, confirming the chimeric approach shown above (Fig. 2B). While the mutation of the two lysine residues had no effect on receptor surface expression, the mutation of the three dileucine-like motifs led to an increase in surface expression similar to the one resulting from deletion of the 20 amino acids of the interbox1/2 region (Fig. 3A, right panel, compare the second and third  bars).
Because we could not monitor the expression of the human OSMR by Western blot analysis due to lack of suitable antibodies, we constructed receptors C-terminally tagged with enhanced GFP. Notably, the GFP tag did not alter the binding of Jak1 to the receptor, and the surface expression of GFP-tagged receptors was equally up-regulated upon Jak1 coexpression compared with untagged receptors (data not shown). The overall expression of GFP-tagged receptors could be easily monitored both by FACS as well as Western blot analysis. As expected, the mutant construct OSMR-3AA-GFP was much better expressed at the cell surface than the wild-type construct (Fig. 3B, upper panel). The increase in surface expression (measured with an antibody against the OSMR extracellular region) was paralleled by an increase in the amount of fully processed receptors as detected by Western blot analysis (Fig. 3B, lower panel, upper bands marked by the arrowhead).
The experiments performed in Fig. 1B clearly demonstrated that the negative signal within the OSMR box1/2 region is transferable, thereby drastically reducing the overall expression as well as the cell surface localization of a normally well expressed cytokine receptor. As shown in Fig. 3C, mutation of the three dileucine-like motifs within the transferred region completely abrogated this effect. Thus, this construct retained its strong surface expression (Fig. 3C, upper panel, third bar) and displayed on Western blots a prominent band corresponding to fully processed receptors (lower panel, third lane), further supporting a negative regulatory role for the three dileucine-like motifs.

Jak1 Coexpression Only Slightly Increases the Surface Expression of an OSMR Variant in Which the Three Dileucine-like Motifs Have Been
Mutated-We next compared the effect of Jak1 coexpression on the surface localization of OSMR-GFP and OSMR-3AA-GFP. Interestingly, in the presence of coexpressed Jak1, OSMR-GFP was as well expressed at the cell surface as the OSMR-3AA-GFP construct without Jak1 (Fig.  4A, upper panel, compare bars 2 and 3), implying that the effect of Jak1 coexpression is indeed due to masking of the three dileucine-like motifs. Jak1 coexpression also reproducibly increased the surface expression of OSMR-3AA-GFP (Fig. 4A, upper panel, bars 3 and 4), albeit to a much lesser extent than that of the wild-type construct. Notably, this reduced up-regulation by Jak1 is not due to reduced binding to the receptor because the alanine substitution of the three dileucine motifs had no effect on the ability of the receptor to associate with Jak1: similar amounts of Jak1 could be coprecipitated with both receptor constructs (Fig. 4B, middle panel, compare lanes 2 and 4).
The Three Dileucine-like Motifs Affect the Receptor Surface Half-life without Influencing Receptor Internalization-Dileucine-like motifs are known to serve as signals that promote receptor internalization and/or lysosomal targeting. To test whether one of these processes is responsible for the increased surface expression of OSMR-3AA-GFP, we next monitored the internalization rate and surface half-life of the mutant receptor in comparison with OSMR-GFP. To control the expression levels of both receptors, we stably expressed OSMR-GFP and OSMR-3AA-GFP in Flp-In T-REx-293 cells. In these cells, receptor expression can be easily induced by incubation with doxycycline in a dose-dependent manner. Again, the surface expression of OSMR-3AA-GFP was markedly increased compared with that of OSMR-GFP (Fig. 5A).
To monitor the internalization rate, cells were incubated with anti-OSMR antibody on ice, washed thoroughly, and then incubated at 37°C for the indicated periods of time to allow internalization to occur. Cells were harvested, and the amount of surface-bound antibody was monitored by FACS analysis. As shown in Fig. 5B, the amount of surfacebound antibody rapidly declined, but no difference was observed between the wild-type and mutant receptors. This indicates that the increased cell surface expression of OSMR-3AA-GFP is not due to impaired receptor internalization.
The receptor surface half-life was determined by incubating doxycycline-induced cells for the indicated periods of time with brefeldin A (BFA), a drug that prevents exit of newly synthesized proteins from the ER. Interestingly, the surface half-life of OSMR-3AA-GFP was increased compared with that OSMR-GFP (Fig. 5C), suggesting that the fate of both receptor constructs differs once they are internalized. Of note, incubation of the cells with the recycling inhibitor monensin led to an identical reduction in cell surface localization for both receptors (Fig.  5D). Therefore, the better expression of OSMR-3AA-GFP may be indeed due to an increased amount of recycled receptors.
The Surface Half-life of the Endogenous OSMR Is Decreased in Jak1deficient Cells-Based on the previous findings (Fig. 5), one would be predict that Jak1, by masking the three dileucine-like motifs, would affect the surface half-life of the OSMR. We therefore monitored the effect of BFA treatment on the surface expression of the endogenous OSMR in parental (2C4) and Jak1-deficient (U4C) fibrosarcoma cells. Indeed, the surface half-life of the OSMR was decreased in Jak1-deficient cells (Fig. 6A). We tested whether this effect could be reversed by introduction of Jak1 into these cells. Jak1-deficient U4C cells were transiently transfected with GFP-tagged Jak1 constructs, and the effect of BFA was analyzed in transfected (GFP-positive) cells. As expected, introduction of the non-binding Jak1-L80A/Y81A mutant did not alter the surface half-life of the endogenous OSMR: incubation with BFA for 2 h resulted in a similar loss of OSMR surface expression as seen before in untransfected U4C cells (Fig. 6, compare A and B, left panel). However, the loss of OSMR surface expression upon BFA treatment was significantly reduced in cells expressing wild-type Jak1 (Fig. 6B, right  panel). This indicates that association with Jak1 can indeed increase the surface half-life of the OSMR.
The Three Dileucine-like Motifs Regulate Protein Stability-The reduced surface half-life of OSMR-GFP versus OSMR-3AA-GFP and the fact that the internalization rate of these two constructs is equal could be best explained by a lysosomal targeting mechanism. We therefore monitored the protein stability of both receptors in stably transfected HEK293 cells. Interestingly, not only was the number of cell surface receptors of OSMR-3AA-GFP increased compared with that of OSMR-GFP (Fig. 5A), but also the steady-state overall expression as measured by FACS analysis detecting the GFP-mediated fluorescence (Fig. 7A, upper panel) or by Western blotting (lower panel). To further elaborate whether the increased overall expression is based on increased receptor stability, HEK293 cells expressing either OSMR-GFP or OSMR-3AA-GFP were incubated for different time periods in medium containing cycloheximide to block further protein synthesis. GFP fluo-FIGURE 3. Three dileucine-like motifs in the interbox1/2 region negatively regulate receptor expression. COS-7 cells were transiently transfected with expression vectors for the indicated receptor mutants and harvested after 48 h. A, receptor surface expression was monitored by FACS analysis using a monoclonal antibody against the extracellular region of the OSMR. Relative (rel.) mean fluorescence values obtained for the wild-type (wt) OSMR were set to 100%. B, surface expression was monitored as described for A. For Western blotting (WB), cells were lysed in Triton lysis buffer; lysates were separated by SDS-PAGE; and blots were detected with antibody against GFP. Mean fluorescence values obtained for wild-type OSMR-GFP (OR-GFP) were set to 100%. C, receptor expression of the indicated constructs was analyzed by Western blotting and FACS analysis as described above. The values obtained for the ␤130⌬cyt construct were set to 100%. The means Ϯ S.D. of at least three independent experiments are shown. To monitor overall expression, cells were lysed in Triton lysis buffer, and equal amounts of lysates were separated by SDS-PAGE. Western blots (WB) were detected with the indicated antibodies. rel., relative. B, for binding studies, cells were lysed in Brij lysis buffer; equal amounts of receptors were immunoprecipitated (IP) using anti-OSMR antibody; precipitates and lysates were separated by SDS-PAGE; and proteins were detected as indicated. FEBRUARY 17, 2006 • VOLUME 281 • NUMBER 7 rescence was monitored by FACS analysis. As shown in Fig. 7B, the GFP fluorescence decreased more rapidly in cells expressing OSMR-GFP compared with OSMR-3AA-GFP, pointing indeed to a prolonged halflife of the latter receptor construct. We next incubated the different cell pools with cycloheximide in the presence of MG132, an inhibitor of proteasomal degradation, or chloroquine, an inhibitor of lysosomal degradation (Fig. 7B). Both inhibitors (but in particular, chloroquine) substantially stabilized wild-type OSMR-GFP because the GFP fluorescence in the presence of cycloheximide and chloroquine was as high as that observed in cells without cycloheximide treatment (Fig. 7C, left  panel, compare the first and fourth bars). The stabilizing effect of chloroquine was much more pronounced for OSMR-GFP than for OSMR-3AA-GFP, implying that the three dileucine-like motifs may indeed promote lysosomal degradation of the OSMR.

Dileucine-like Motifs Mediate OSMR Intracellular Retention
The ENPHLI Motif within the OSMR Does Not Represent a Classical Lysosomal Targeting Motif-Lysosomal targeting can be mediated by dileucine-like motifs, which has been shown for the CD3␥ chain and lysosomal integral membrane protein type II (LIMPII) (32)(33)(34). In both cases, an acidic amino acid is located at position Ϫ4 relative to the dileucine motif. Interestingly, one of the three dileucine-like motifs present in the OSMR interbox1/2 region, the membrane-distal ENPHLI motif (Fig. 8A, left panel), matches this consensus sequence. To examine whether this motif indeed represents a lysosomal targeting signal, several mutations were introduced into this sequence as depicted in Fig. 8A  (left panel). The surface expression of all these receptors was monitored by FACS analysis of transfected COS-7 cells. While the mutation of the single dileucine-like motif to alanine already considerably increased receptor surface expression (Fig. 8A, right panel, third bar), but to a lesser extent than the mutation of all three dileucine-like motifs (second bar), mutation of the acidic residue to either lysine or alanine did not affect the expression pattern of the OSMR-GFP mutants (fourth and fifth bars, respectively). Notably, similar mutations introduced into a control receptor construct containing the lysosomal targeting sequence of the lysosomal protein LIMPII in the intracellular region had a strong enhancing effect on receptor surface expression (data not shown). These results indicate that the ENPHLI motif of the OSMR does not represent a "classical" lysosomal targeting signal.
Alternatively, the dileucine motifs might contribute to a hydrophobic structural element prone to mediate protein destabilization unless properly protected by associated Jak kinases. Interestingly, the inter-box1/2 region of the OSMR contains several hydrophobic residues apart from the three dileucine-like motifs (Fig. 8B). In contrast, the same region in the related receptor chains LIFR and gp130 is much less hydrophobic. We therefore compared the expression of chimeric receptors FIGURE 5. The receptor surface half-life (but not the internalization rate) of OSMR-GFP is affected by the three dileucine-like motifs. A, pools of HEK cells stably transfected with the expression vector for wild-type (wt) OSMR-GFP (OR-GFP) or OSMR-3AA-GFP (3AA) was induced overnight with doxycycline and harvested the next day. The surface expression of receptors was measured by FACS analysis. rel., relative. B, doxycycline-induced stably transfected HEK pools were incubated for 45 min on ice in medium containing antibody against OSMR. Cells were washed and incubated at 37°C for the indicated periods of time. Cells were harvested, and surface-bound antibodies were stained with R-phycoerythrinconjugated secondary antibody. Surface staining at time 0 was set to 100%. The means Ϯ S.D. of three independent experiment are shown. C, induced HEK cells were incubated for the indicated periods of time with BFA. OSMR surface expression was measured by FACS analysis. The values of ethanol-treated controls were set to 100%. D, doxycycline-induced stably transfected HEK pools were incubated for 90 min with either monensin (Mon) or ethanol (control (ctrl)). Surface expression was monitored by FACS analysis. The values obtained for ethanol-treated controls were set to 100%. The values shown represent the means Ϯ S.D. derived from three independent experiments. containing the box1/2 region of the OSMR, the LIFR, or gp130. As shown in Fig. 8C, the respective gp130 and LIFR constructs indeed displayed strikingly increased surface expression compared with ␤OR-B1/2. At the same time, Jak1 coexpression did not alter the expression of these constructs (Fig. 8C, first panel, compare bars 3 and 4, 5 and 6) even though all three receptors bound Jak1 equally well (fourth panel, compare lanes 2, 4, and 6).

Identification of a Negative Regulatory Signal within the Interbox1/2
Region of the OSMR That Is Masked by Jak1-We have shown previously that the surface expression of the OSMR can be efficiently up-regulated by coexpression of Jak1 (20). Our results implied that this up-regulation is most likely due to the masking of a negative regulatory signal within the OSMR box1/2 region (20). To identify the putative signal of the OSMR that regulates its intracellular retention in the absence of Jaks, we carried out a detailed mutagenesis study of the Jak-binding region within the OSMR. We identified a negative regulatory signal within a stretch of 43 amino acids of the box1/2 region. Transfer of these amino acids led to the down-regulation of otherwise well expressed receptor constructs (Fig. 1B), and coexpression of Jak1 partially restored receptor expression (Fig. 1C). We have further shown that deletion of only 20 amino acids within the membrane-proximal region resulted in a marked increase in receptor expression (Fig. 2). Finally, we identified three dileucine-like motifs contained within these 20 amino acids that are crucial for the observed negative regulation: point mutation of just these motifs had an enhancing effect on surface expression similar to that of deletion of the entire intracellular region (Fig. 3). Notably, the limited surface expression of receptors containing the intact signal was efficiently increased by coexpression of Jak1. This effect seems to be dependent on the presence of the three dileucine-like motifs within the OSMR because a receptor with a mutated signal (which displayed already a much more pronounced cell surface localization) was only slightly up-regulated at the cell surface by Jak1 (Fig. 4A). These findings suggest that Jak kinases regulate OSMR surface localization by masking the three dileucine-like motifs within the OSMR interbox1/2 region, thereby preventing the negative regulatory effect mediated by these motifs.
The Three Dileucine-like Motifs Reduce the Stability of the OSMR as Well as Its Surface Half-life-The three dileucine-like motifs negatively affect not only the surface expression of the OSMR, but also the overall expression by exerting a destabilizing effect: wild-type OSMR-GFP was more rapidly degraded compared with OSMR-3AA-GFP (Fig. 7B). However, in the presence of the proteasomal inhibitor MG132 or the lysosomal inhibitor chloroquine, both receptors displayed a rather similar stability. Especially the lysosomal inhibitor chloroquine had a strong stabilizing effect on the wild-type construct (Fig. 7C), pointing to the lysosome as the main site of destruction under these conditions.
It is well established that dileucine-like motifs can regulate the localization of transmembrane proteins within the cell by acting as internalization and/or lysosomal targeting signals (35). We did not detect any difference in the internalization rates of OSMR-GFP and OSMR-3AA-GFP (Fig. 5B). In accordance with this finding, inhibition of endocytosis by dominant-negative dynamin had a similar positive effect on the surface expression of both receptor constructs (data not shown), implying that a signal different from the three dileucine-like signals mediates internalization of the OSMR. However, despite having the same internalization rate, the OSMR-3AA-GFP construct displayed an increased half-life at the membrane compared with the wild-type construct (Fig.  5C). Thus, both receptors most likely differ in their post-endocytic fate. Indeed, our finding that monensin, an inhibitor that prevents recycling from the endosomes back to the cell surface, reduced the surface expression of the wild-type and mutant constructs to a similar degree (Fig. 5D) is in line with such a model.
Even though our attempts to convincingly demonstrate localization of OSMR-GFP in lysosomes by immunofluorescence have failed so far, FIGURE 7. OSMR-3AA-GFP displays increased protein stability. A, HEK pools were induced overnight with 2 ng/ml doxycycline to express wildtype (wt) OSMR-GFP (OR-GFP) or OSMR-3AA-GFP (3AA) as indicated, and GFP expression was measured by FACS analysis. The values obtained for the wild-type OSMR-GFP construct were set to 100%. The means Ϯ S.D. of four independent experiments are shown. Overall expression was additionally monitored by Western blotting (WB). rel., relative. B, stably transfected HEK cells induced overnight with 2 ng/ml doxycycline were incubated with 50 g/ml cycloheximide (CHX) for the indicated periods of time. Cells were harvested, and receptor expression was monitored by FACS analysis. The GFP values obtained at time 0 were set to 100%. The means Ϯ S.D. of four independent experiments are shown. C, stably transfected HEK pools were incubated for 5 h with cycloheximide in the presence of dimethyl sulfoxide (DMSO; control (ctrl)), MG132 (MG), or chloroquine (CQ). Cells were harvested, and GFP expression was monitored by FACS analysis. The values obtained for untreated cells were set to 100%. The means Ϯ S.D. derived from three independent experiments are shown. FEBRUARY 17, 2006 • VOLUME 281 • NUMBER 7 our data imply that the internalized OSMR-3AA mutant receptor may more easily recycle back to the membrane, whereas the wild-type receptor may be targeted for degradation to the lysosome unless protected by associated Jak kinases. The fact that the expression of Jak1 (but not a non-receptor-binding mutant thereof) increased the surface half-life of endogenous OSMR (Fig. 6B) is in full accordance with this model.

Dileucine-like Motifs Mediate OSMR Intracellular Retention
Of the three dileucine-like motifs, the distal one matches the consensus sequence for lysosomal targeting, (D/E) Ϫ4 XXXL(L/I/V), as present in the CD3␥ chain of the T cell receptor and the lysosomal protein LIMPII (32)(33)(34). The acidic amino acid at position Ϫ4 is essential for functional lysosomal targeting of LIMPII (36) and the CD3␥ chain (32). This can be explained by the interaction of this residue with the AP-1 and AP-3 complexes (37), which have been shown to play a role in lysosomal targeting of various proteins (33,38,39). Even though mutation of the dileucine-like motif within the putative lysosomal targeting sequence (OSMR-AA-GFP) led to increased OSMR surface expression, mutation of the acidic residue (OSMR-E-K-GFP or OSMR-E-A-GFP) was without effect (Fig. 8). When introduced into a control construct containing the intracellular region of the lysosomal protein LIMPII, the same mutations strongly enhanced cell surface localization (data not shown). These findings argue against the possibility that the ENPHLI motif functions as a classical lysosomal targeting motif.
The Three Dileucine-like Motifs Might Be Part of an Exposed Hydrophobic Patch Conferring Instability Unless Masked by Jak Kinases-Interestingly, the interbox1/2 region of the OSMR contains several hydrophobic amino acids besides the leucine and isoleucine residues included in the three dileucine-like motifs. Thus, alternatively, the dileucine-like motifs might not represent a specific sorting signal but instead might be part of a hydrophobic structure conferring instability in the absence of associated Jaks. In support of such a hypothesis, inspection of the sequence of the human OSMR reveals 13 hydrophobic residues within the 30 amino acids between box1 and box2, whereas the corresponding regions of the related signal-transducing receptor chains for IL-6-type cytokines gp130 and LIFR harbor only six residues each (Fig. 8B). In striking contrast to the OSMR, the corresponding box1/2 receptor constructs of gp130 and the LIFR were already well expressed at the cell surface and were not further up-regulated by coexpressed Jak kinases (Fig. 8C). During the process of protein folding, hydrophobic regions are normally bound by chaperones until proper folding leads to burial within the final protein structure (40 -42). It is conceivable that, in the case of the OSMR, hydrophobic residues present in the interbox1/2 region are masked only upon proper association with Jaks and/or that Jaks could act as chaperones that facilitate the proper folding of the OSMR. Newly synthesized proteins that fail to fold or assemble correctly are targeted for proteasomal degradation by the ER-associated degradation mechanism (40). Our finding that degradation of the OSMR is partially prevented by the proteasomal inhibitor MG132 (Fig.  7C) would be in line with such a mechanism. However, quality control at the level of protein folding may extend to compartments "downstream" of the ER. Not correctly folded membrane proteins can be targeted for lysosomal proteolysis directly from the Golgi compartments (43, 44); but also conformationally unstable membrane proteins that escape Amino acids belonging to the box1 and box2 regions are underlined. C, the indicated chimeras and either a non-receptor-binding Jak1 mutant (Jak1-L80A/Y81A (control (ctrl)) or wild-type Jak1 were transiently coexpressed in COS-7 cells.
Receptor surface expression was monitored as described for A. FACS values were calculated relative to cells expressing non-binding Jak1. For Western blot (WB) analysis, cells were lysed in Brij lysis buffer; receptors were immunoprecipitated (IP) using monoclonal antibody against IL-5R␤; precipitates and aliquots of the lysates were separated by SDS-PAGE; and the blots were stained as indicated.
these quality control checkpoints are rapidly degraded, e.g. it has been recently demonstrated that the misfolding of mutant cystic fibrosis transmembrane conductance regulators prevents recycling from endosomes back to the cell surface and facilitates lysosomal targeting (45), a scenario that may reflect the fate of the OSMR devoid of an associated Jak, as suggested by our study.
Distinct Regulation of Cytokine Receptor Expression by Associated Jaks-Interestingly, a role of associated Jaks in the regulation of receptor expression has been described for some other cytokine receptors. Although Jak3 expression is not needed for efficient surface localization of the common ␥-chain, Jak3 expression does enhance the surface expression of the common ␥-chain when overexpressed in Jak3-deficient cells (28). The expression of Tyk2 is needed for the stable surface expression of the IFNAR1 chain and IL-10 receptor-2, and Jak2 association facilitates the surface expression of the EpoR (21,22). In the case of IFNAR1, Tyk2 has been shown to prevent receptor internalization from the plasma membrane, a mechanism that could be excluded in case of the OSMR (Fig. 5B). However, Jak2 functions by regulating the export of the EpoR from the ER. In the case of both IFNAR1 and EpoR, it has been demonstrated that the receptors contain a negative regulatory signal in their membrane-proximal regions because deletion of these regions results in receptors that are much better expressed at the cell surface (21,22). However, the exact signals involved have not been clearly identified. Alanine-scanning mutagenesis of the membrane-proximal region of the EpoR did not result in any mutants that were more efficiently expressed at the cell surface (25). Thus, it was argued that Jak2 might not mask a single linear ER retention signal, but instead serve as a chaperone facilitating the folding of the whole membrane-proximal region of the EpoR (25). In light of our findings, one can speculate that the EpoR, like the OSMR, might also contain a hydrophobic patch in its membraneproximal region that is masked by associated Jak2; and interestingly, the 30 amino acids downstream of the box1 region of the EpoR harbor 11 hydrophobic residues.
More recently, the effect of Jaks has been studied in more receptor systems. Although Jak2 is not indispensable to observe mature growth hormone receptors in stably transfected ␥2A cells (lacking Jak2), the presence of Jak2 increases the stability of the mature receptor form most likely by preventing its lysosomal degradation as suggested by inhibitor studies (27,46). Likewise, Royer et al. (26) recently demonstrated that Jak2 and Tyk2 enhance the surface expression of the thrombopoietin receptor when these proteins are stably overexpressed in Ba/F3 pro-B cells. The kinases substantially enhance the stability of the mature receptor form, presumably by preventing its proteasomal degradation (26). These examples demonstrate that stabilization of receptors by their associated Jaks seems to be a common mechanism that evolved to guarantee that only functional receptor⅐Jak complexes are expressed at the cell surface. It will be very interesting to see whether, in these cases, similar motifs can be identified that will help us to understand in more detail how Jaks are able to stabilize cytokine receptors.