Stimulation of leukemia inhibitory factor receptor degradation by extracellular signal-regulated kinase.

Leukemia inhibitory factor (LIF) signals via the heterodimeric receptor complex comprising the LIF receptor alpha subunit (LIFRalpha) and the common signal transducing subunit for interleukin-6 cytokine receptors, gp130. This study demonstrates that in different cell types, the level of LIFRalpha decreases during treatment with LIF or the closely related cytokine oncostatin M (OSM). Moreover, insulin and epidermal growth factor induce a similar LIFRalpha down-regulation. The regulated loss of LIFRalpha is specific since neither gp130 nor OSM receptor beta shows a comparable change in turnover. LIFRalpha down-regulation correlates with reduced cell responsiveness to LIF. Using protein kinase inhibitors and point mutations in LIFRalpha, we demonstrate that LIFRalpha down-regulation depends on activation of extracellular signal-regulated kinase 1/2 and phosphorylation of the cytoplasmic domain of LIFRalpha at serine 185. This modification appears to promote the endosomal/lysosomal pathway of the LIFRalpha. These results suggest that extracellular signal-regulated kinase-activating factors like OSM and growth factors have the potential to lower specifically LIF responsiveness in vivo by regulating LIFRalpha half-life.

mouse monocytic leukemia M1 cells, conversion of sympathetic neurons from the adrenergic to cholinergic phenotype, suppresses the differentiation of embryonic stem cells, enhances proliferation of myoblasts, and facilitates endometrial implantation of embryos (reviewed in Refs. 1 and 2).
LIF also plays a role in the systemic inflammatory response, activating the hypothalamic-adrenal axis, and inducing the acute phase reaction of the liver (3). In hepatocytes, LIF, similar to other IL-6 cytokines, stimulates the enhanced expression of a set of plasma proteins, termed acute phase proteins (APP) (4). The precise pattern of APP expression is determined by the action of IL-6 cytokines in combination with various other inflammatory mediators, endocrine hormones, and growth factors (5). For instance, during the acute phase reaction, insulin is increased 3-fold (6) and then modulates the cytokine regulation of APP genes (7,8). In myoblasts, IGF-1 also reduces LIF action, apparently by down-regulating LIF receptor number (9).
LIFR␣ is a 190-kDa transmembrane protein with low affinity for LIF. In combination with gp130 subunit, it forms the high affinity LIF receptor complex (10,11). As described in the human system, OSM also uses LIFR␣ and gp130 subunits to form a high affinity OSM receptor complex (then termed OSMR complex type I). CNTF, CT-1, and neurotrophin-1 also utilize LIFR␣ and gp130 subunits (reviewed in Refs. 12 and 13). In addition to the shared LIFR␣/gp130 complex used by either LIF or OSM, a specific OSM receptor complex (type II) has been identified that is composed of gp130 and OSMR␤ (14). In rodents, OSM seems to act exclusively through the type II receptor complex of gp130/OSMR␤ (15,16).
LIF binding to LIFR␣ induces heterodimerization with gp130. Janus protein-tyrosine kinases (JAK), constitutively associated with the cytoplasmic domain of these receptors, are then activated by trans-and autophosphorylation, and in turn phosphorylate tyrosine residues in both LIFR␣ and gp130 intracellular domains. Those phosphorylated tyrosines, within the Box3 sequence context YXXQ, create docking sites primarily for STAT3 but also STAT1. Other sites are recognized by linker proteins which, upon phosphorylation, propagate the signal to other pathways (MAPK, PI3K) (reviewed in Refs. 12, 13, and 17). The Src homology 2 domain-containing proteintyrosine phosphatase SHP2 functions as such a linker toward Grb2 and MAPK. SHP2 also acts, through its catalytic activity, as a negative regulator of the JAK/STAT signaling by downregulating JAK activity, and consequently lowering the induction of STAT3-dependent genes (18 -20). Other inhibitors of LIF activity are members of the SOCS (suppressor of cytokine signaling) family, especially SOCS3, and the phosphatase SHP1, which, in pituitary cells, is found constitutively associated with JAK. SHP1 decreases JAK and STAT3 phosphoryl-ation, and SOCS3 has been shown to deactivate JAK within 1 h of LIF treatment (21).
Ligand-induced LIFR␣/gp130 heterodimerization increases serine phosphorylation of the LIFR␣ cytoplasmic domain. ERK1 and ERK2 are implicated in phosphorylating Ser 1044 (or Ser 185 of the cytoplasmic domain) of LIFR␣ (22). Moreover, activation of ERK by insulin also increases LIFR␣ Ser 185 phosphorylation. It has been suggested that this modification may contribute to the modulation by insulin of cell responsiveness to LIF (22). However, the mechanism of this LIFR␣ regulation is unknown. A similar serine phosphorylation of gp130 or OSMR␤ by MAPK has not yet been demonstrated, and in contrast to LIFR␣, both gp130 and OSMR␤ do not contain an overt substrate site for MAPK kinases.
Constitutive, ligand-independent endocytosis of gp130 and LIFR␣ has been observed that depends in part on dileucine motifs within the cytoplasmic domain of these receptors (23,25). In the case of gp130, an interaction with the adaptor protein AP2 has been described, which presumably results in a transfer of the receptor into clathrin-coated pits, followed by efficient endocytosis, trafficking from endosomes to lysosomes, and finally lysosomal degradation (24,25). Most of the current evidence indicates that gp130 half-life is not appreciably modified by ligand binding (i.e. IL-6 binding), suggesting that gp130 signal transduction does not modify internalization/degradation of this receptor subunit.
In this study we show that, in various cell types, LIFR␣ is specifically down-regulated by treatments with LIF and other effectors that activate ERK 1/2. We suggest that Ser 185 phosphorylation of LIFR␣ promotes lysosomal degradation of this receptor subunit. In contrast, co-activated gp130 or OSMR␤ proved to be significantly more stable under the same conditions. Finally, we demonstrate that, in hepatic cells, specific LIFR␣ down-regulation correlates with a lower LIF induction of signaling and gene expression.
Plasmid Constructs-The expression vectors for the wild type and S1044A (S185A) mutant human LIFR␣ and chimeric G-CSFR-LIFR␣ in pDC302 have been described (10,22). Cytoplasmic domains of these receptors were modified by the addition of Myc epitope (EQKLI-SEEDLN) to the C terminus. The LIFR␣-Myc or G-CSFR-LIFR␣-Myc constructs were transferred as EcoRI-NotI fragments into the retroviral vector MINV. The vector-derived viruses were used to transduce H-35 cells, which were selected in medium containing 2 mg/ml G418 (18,27). Clonal lines were identified for comparable levels of LIFR␣-Myc expression by Myc immunoblotting and low endogenous LIFR␣ expression (LIFR␣ immunoblotting). LIFR␣ with deleted extracellular domain (⌬N-LIFR␣, residues 821-1097) was generated by replacing the extracellular G-CSFR, in chimeric G-CSFR-LIFR␣, by gp130-leader peptide (generated by polymerase chain reaction). Leu-Ile to Ala-Ala (L210A/ I211A) LIFR␣ has been generated by polymerase chain reaction with the same oligonucleotides described by Thiel et al. (23), using the plasmid pDC302-LIFR␣ as a template. All the LIFR␣ constructs have also been modified by the addition of the green fluorescent protein (GFP) to the C terminus in EGFP.N1 vector (CLONTECH, Palo Alto, CA).
Transient Transfection-HepG2 cells were transfected by the calcium phosphate method (28) with a total of 20 g/ml DNA. Expression of LIFR␣-GFP was visualized under a Nikon inverted fluorescence microscope, and digitized images (magnification, ϫ40) were taken by a SPOT camera (Diagnostic Instruments, Inc.).
Radioiodination of LIF and Binding-Chinese hamster ovary cellsderived LIF was iodinated at a specific radioactivity of around 15,000 Ci/nmol according to the chloramine T method as described (29). For binding studies, H-35 monolayers were incubated in phosphate-buffered saline, containing 0.05% trypsin and 0.02% EDTA. 0.5 ϫ 10 6 cells in suspension were incubated for 90 min with an increasing concentration of labeled LIF in phosphate-buffered saline, containing 0.5% bovine serum albumin and 5 mM Man-6-P, to avoid binding to the Man-6-P/ IGFII-R (30,31). Nonspecific binding was evaluated by including a 100-fold excess of unlabeled LIF. Cell bound and unbound fractions were separated by centrifugation through a layer of dibutylphthalate (90%) and paraffin oil (10%). Regression analysis of the binding data was accomplished using a one or two-site equilibrium binding equation (Grafit; Erithacus Software, Staines, United Kingdom).
Soluble Human LIFR␣ Analysis-Parental or hLIFR␣-expressing H-35 cells in serum-free medium were treated 4 h with LIF or insulin. Supernatants were collected, dialyzed against 25 mM NH 4 HCO 3 , lyophilized, and resuspended in phosphate-buffered saline. The amount of shed human LIFR␣ was measured by a specific ELISA as reported previously (43).
Plasma Protein Analysis-Synthesis and secretion of APP into the culture medium of cytokine-treated H-35 cells were quantified by immunoelectrophoresis (32). The area under the precipitation peaks (proportional to the amount of antigen) was integrated and expressed in arbitrary units.

RESULTS
LIFR␣ Is Down-regulated by LIF Treatment-In hepatic cells, the LIF-induced signal transduction, as determined by STAT activation and induction of APP genes, is transient, implying the presence of intracellular negative regulatory mechanisms. In pituitary cells, protein-tyrosine phosphatases SHP1 and SHP2, and SOCS3, are proposed to down-regulate LIF signaling within 30 min to 1 h by inactivation of JAK2 and/or STAT3 (18 -21, 33). In myoblasts, it is also suggested that a separate mechanism operates via down-regulation of the LIFR␣ protein (9). To test the latter possibility, we measured the level of immunodetectable LIFR␣ after LIF treatment in the rat hepatoma cell line H-35.
LIFR␣ proteins appeared as doublets with 190 and 170 kDa (Fig. 1A). Treatment with monensin, which prevents the transfer of glycoproteins from the endoplasmic reticulum/Golgi ap-paratus to the plasma membrane (34), caused an accumulation of the lower form of the LIFR␣, suggesting a precursor-product relationship between the two forms ( Fig. 1D). Presumably, the larger protein represents fully processed receptor protein as detectable at the plasma membrane that also undergoes ligand-induced phosphorylation (see Fig. 3B). The smaller size protein likely represents immature, intracellular LIFR␣, analogous to the findings with gp130 (34). As shown in Fig. 1A (left panel) (see also Fig. 5C), the large size LIFR␣ form displays a transient size increase after 5-15 min of LIF treatment, attributed to increased receptor protein phosphorylation (see below). This is followed by a decrease of immunodetectable protein that reaches a 30 -40% level by 2 h (Fig. 1C). The reduced level of LIFR␣ protein was maintained for more than 7 days (data not shown) and was not restricted to H-35 cells but also observed in mouse Hepa1 cells, rat HTC cells, human HepG2 cells (all hepatoma cells, data not shown), NIH3T3 fibroblasts, HeLa epithelial cells (see Fig. 8), primary lung fibroblasts, and different breast carcinoma cells (data not shown).
The loss of full-length LIFR␣ was accompanied by a transient appearance of degradation products, which were detectable at 60 -70 kDa (arrows in Fig. 1A, overlapping with a nonspecific cross-reactive material). Since antibodies against the C-terminal part of LIFR␣ were used, the degradation products presumably represent LIFR␣ with truncated extracellular domains. This possibility was assayed first by ectopic expression of full-length LIFR␣ in a HepG2 cell clone (clone 86. 6,Ref. 41) with an endogenous LIF production. Degradation products were clearly detectable, together with full-length and precursor forms of LIFR␣ (Fig. 1B, lane 2). Moreover, the identity of the short forms was verified by transfection of an expression vector for LIFR␣ retaining only 12 amino acids of the extracellular domain (⌬N-LIFR␣, residues 821-1097). This truncated LIFR␣ co-migrated with LIFR␣ degradation products (Fig. 1B,  lane 3).
Regulation of LIFR␣ Turnover in Trans-These first experiments indicated that LIFR␣ level is regulated in part through LIF binding (autologous regulation). To test the possibility of a regulation of LIFR␣ level in trans (heterologous regulation), we treated H-35 cells with OSM, IL-6, and insulin. As shown in  (27), respond to human OSM (hOSM) by a reduced LIFR␣ level (40% level after 2 h) through the action of the OSMR complex type I, formed by LIFR␣ and gp130 (11,14,36). In contrast, IL-6 had little influence (Fig. 1, A and C). To prove that signaling by OSMR complex type II has also the ability to act in trans, we used the recently isolated H-35 cell line stably expressing the transduced murine OSMR␤ (27). As shown in Fig. 1 (A and C), mOSM that unlike hOSM does not function through LIFR␣ (15,16), induced a down-regulation of LIFR␣ in mOSMR␤ H-35 cells (30% level) but not in parental H-35 cells. Furthermore, treatment of H-35 cells with insulin similarly reduced the level of LIFR␣ (30% level; Fig. 1, A and C), indicating that LIFR␣ can also be down-regulated in trans.
Interestingly, insulin in parental H-35 cells or mOSM in mOSMR␤ H-35 cells induced the mobility shift of the functional LIFR␣ (at early time points of treatment). Moreover, both factors induced the appearance of degradation products (after 1 h of treatment), similar to that observed with LIF treatment (Figs. 1A and 5C). The finding of factor-induced reduction of LIFR␣ suggested that all these factors might affect the turnover process or the synthesis of LIFR␣. To discriminate between these two hypotheses, we measured immunodetectable LIFR␣ in H-35 cells whose protein de novo synthesis was inhibited by cycloheximide (Fig. 1D). Under these conditions, the fully processed LIFR␣ protein showed an apparent half-life of approximately 2 h and treatment with insulin lowered this value to 1 h. This suggests a factor-induced degradation of the receptor subunit. The small size precursor LIFR␣ form showed a t 1 ⁄2 of 30 min to 1 h and appeared insensitive to insulin treatment. Treatment of the cells with monensin, which prevents protein secretion, did not interfere with the turnover of full-length LIFR␣ but resulted in the accumulation of the short form LIFR␣, as expected for a precursor/product relationship (Fig. 1D).
Gp130 Has a Distinct Turnover Mechanism-Considering the obligatory signal-transducing function of gp130, we determined the influence of cytokine treatments on gp130 protein.
Down-regulation of LIFR␣ Contributes to a Reduced Response to LIF-The results from the experiments illustrated in Figs. 1 and 2 indicate that LIFR␣ is target for down-regulation by its own, and other, receptor system signals. In contrast, gp130 (Fig. 2) or OSMR␤ (Figs. 5A and 8B), as the other members of the IL-6 family of cytokine receptors with signal transducing activity, are less sensitive to signal acting in trans. This finding raises the possibility that LIF, as well as factors such as OSM or insulin, are able to limit LIF responsiveness of cells. Indeed, insulin pretreatment lowered LIF binding (Fig. 3A) and the LIF-inducible tyrosine phosphorylation of LIFR␣, gp130, and STAT3 (Fig. 3B, upper panels). In contrast, mOSM or IL-6-inducible STAT3 tyrosine phosphorylation is maintained in insulin-treated cells (Fig. 3B, lower panels).
Insulin has pleiotropic effects, some of which have been demonstrated to reduce IL-6-induced acute phase protein expression (4,7,8). However, pretreatment with insulin lowered 2 times more LIF-induced APP expression than did treatment with IL-6 or mOSM (Fig. 3C). This suggests that insulin effect on LIFR␣ level contributes to the reduced response to LIF.
Activation of ERK1 and ERK2 Contributes to Down-regulation of LIFR␣ in H-35 Cells-As shown in Fig. 1A, the activation of ERK 1/2 but not of STAT3 seems to correlate with enhanced turnover of LIFR␣. Although LIF, mOSM, and IL-6 induced STAT3 phosphorylation to approximately the same level, only LIF and mOSM activated ERK 1/2 to the same high level as insulin. To test the relative contribution of ERK to the LIFR␣ down-regulation, we determined the effect of U0126, an inhibitor of MEK1, on prevention of LIFR␣ reduction. Pretreatment of H-35 cells with U0126, at a concentration sufficient to suppress activation of ERK but not tyrosine phosphorylation of the insulin receptor, IRS1/2, or STAT3 (Fig. 4, A and B LIFR␣ tagged with C-terminal Myc epitope. To evaluate the relative contribution of extracellular domain on LIFR␣ turnover, we also generated H-35 cells expressing the chimeric receptor comprising the extracellular domain of G-CSFR and the transmembrane/cytoplasmic domain of human LIFR␣. Immunodetectable receptor proteins were determined after treatment with LIF (Fig. 5A) or insulin (Fig. 5, A-C). The transduced LIFR␣(WT) and chimeric G-CSFR-LIFR␣(WT) showed a mobility shift immediately after treatment with LIF or insulin (Fig. 5, B and C). Degradation kinetics, with the appearance of breakdown intermediates (Fig. 5, A and B), were comparable to that established for the endogenous LIFR␣ (Fig. 1), whereas mOSMR␤ level, in mOSMR␤-expressing H-35 cells (27), was stable (Fig. 5A). In the case of insulin-treated cells, LIF binding to the cells was decreased proportionally to the LIFR␣ protein (Fig. 5D). These results suggested that the information for specific LIFR␣ down-regulation and degradation is contained in the transmembrane/cytoplasmic domain. The critical role of Ser 185 for down-regulation was recognized in LIFR␣(S/A), which, despite normal STAT and ERK signaling (data not shown), did not show a ligand-induced mobility shift (Fig. 5C) and was significantly more stable than LIFR␣(WT) (Fig. 5, A,  B, and D). The remaining ligand-induced decrease of LIFR␣(S/A) (75-80% level after 2 h) was in the range of the ligand-induced decrease of endogenous gp130 (Fig. 2) or chimeric G-CSFR-gp130 ( Fig. 5A; Ref. 19), and was prevented by pretreatment with U0126 (data not shown).
Taken together, the data support the model that activated ERK1/2 phosphorylate Ser 185 , and this modification in turn is sufficient to alter the electrophoretic migration and promote degradation of LIFR␣. The regulated loss of LIFR␣ seems to be specific since neither gp130 nor mOSMR␤ showed a compara- To identify the contribution of some of the key enzymes to LIFR␣ turnover, H-35 cells were treated with specific inhibitors. Pretreatment with 1,10-phenanthroline (inhibition of metalloproteases including cell surface enzymes; data not shown), acetyl-Leu-Leu-Met (calpain inhibitor), and proteasome inhibitor I did not prevent LIFR␣ down-regulation (Fig.  6B). Pretreatment with wortmannin was similarly ineffective in preventing loss of full-length LIFR␣ but caused an accumulation of LIFR␣ fragments (Fig. 6, A and B). Pretreatment with chloroquine (lysosome inhibitor) reduced the LIFR␣ down-regulation after insulin, LIF, and mOSM treatment to 90%, 75%, and 80% levels, respectively, and also promoted accumulation of LIFR␣ fragments (Fig. 6, A-C).
To document more precisely the contribution of proteolytic shedding of LIFR␣ extracellular domain to LIFR␣ down-regulation, hLIFR␣(WT)-transduced H-35 cells were treated 4 h with LIF or insulin, and the concentration of soluble hLIFR␣ in the culture supernatant was determined (Fig. 6D). The ELISA used for this measurement is specific for human LIFR␣ and did not cross-react with any material from parental H-35 cells (Fig. 6D, left panel). A constitutive basal rate of soluble human LIFR␣ release was detected, which amounted in three separate experiments to 100 pg/h/10 6 cells. LIF treatment did not change significantly this LIFR␣ shedding, but in the presence of insulin, the value was enhanced between 20% and 70% (Fig. 6D, right panel). Under these conditions, the membrane associated hLIFR␣ protein was effectively reduced (Fig. 6D, lower panels). By taking into consideration a LIFR␣ concentration of 2,000 molecules/cell, and the receptor having a turnover of 2 h, we calculated that approximately 5-10% of cellular LIFR␣ is shed. Since the loss of cell-associated hLIFR␣ protein exceeded appreciably that accounted for by shedding, a prominent role of intracellular degradation (chloroquine-sensitive) as part of the observed LIFR␣ turnover is predicted.
To visualize a possible subcellular redistribution of LIFR␣ as a function of autologous or heterologous treatments, we employed HepG2 cells transiently transfected with expression vectors for GFP-tagged WT, S/A mutant, or LI/AA endocytosisdeficient mutant (23) human LIFR␣ (Fig. 7). Wild type LIFR␣ was predominantly detected at the plasma surface (uniform green staining), and in the endoplasmic reticulum/Golgi compartment (punctate staining, probably representing the LIFR-GFP precursor form). After LIF or OSM treatment for 24 h, the plasma membrane staining disappeared, while the punctate, most probably endosomal/lysosomal, intracellular staining has intensified. S/A and LI/AA mutants LIFR␣ did not prominently respond to LIF and OSM treatment by a redistribution of tagged LIFR␣ to the intracellular structures, in contrast to wild type LIFR␣. Western blot analysis (data not shown) confirmed the retention of mature LIFR␣ protein in S/A and LI/AA LIFR␣-expressing cells.
Taken together, these results suggest that increased lysosomal degradation of LIFR␣ is responsible for down-regulation of the wild type receptor subunit, and that LI-mediated endocytosis and ERK-dependent tagging are necessary for factorregulated LIFR␣ degradation.
LIFR␣ Down-regulation/Degradation Is Not Restricted to Hepatoma Cells-Since all the studies presented so far have used hepatoma cells as experimental tools, the question arises whether similar LIFR␣ regulation occurs in other cell types. Therefore, similar analyses were performed on NIH3T3 mouse fibroblats and HeLa human epithelial cells. As shown in Fig. 8A, LIF or mOSM on NIH3T3 cells induced an early mobility shift and down-regulation of LIFR␣. Degradation products of LIFR␣ were not detected, presumably because of a lower LIFR␣ expression in these cells. Other factors, namely IL-6 and insulin, did not significantly activate ERK and did not modify LIFR␣ level (data not shown). Interestingly, chloroquine, but not U0126, prevented the effect of LIF on LIFR␣. In contrast, both reagents suppressed the effect of mOSM, as already observed in H-35 cells (see Figs. 4 and 6). NIH3T3 cells also confirmed the subunit-specific effect by an unaltered expression level of gp130 (Fig. 8A). Western blot analyses of mOSMR␤ could not be performed due to the unavailability of immune reagents to this receptor subunit. A LIFR␣ regulation in trans could be also demonstrated by EGF treatment in HeLa cells (Fig. 8B). This ERK-activating growth factor (data not shown) induced a down-regulation of LIFR␣ to the same level as with LIF (60% level after 2 h), whereas the same treatment did not appreciably affect levels of gp130 or OSMR␤. DISCUSSION This study demonstrates that LIFR␣ is down-regulated as the result of ligand binding (autologous regulation), or the action of OSM, insulin, and EGF in trans. The suggested mode of regulation is by increased endocytosis/trafficking and lysosomal degradation. In contrast, the constitutive turnover of gp130 is largely unaffected by cytokine and hormone treatment, a phenomenon that has also been observed in other experimental systems (24,25).
The results have revealed different molecular mechanisms of LIFR␣ down-regulation. The one described here is dependent on activated ERK, which phosphorylates Ser 185 in the cytoplasmic domain of the LIFR␣ (22). A possible consequence of this tagging by Ser phosphorylation could be a facilitated interaction between the LI motif in LIFR␣ and ␣-adaptin (AP2) in clathrin-coated pits, resulting in endocytosis, and further lysosomal degradation (23). If so, the Ser 185 -dependent "regulated" pathway for LIFR␣ could mechanistically resemble the "constitutive" turnover of gp130 that depends on Ser 139 , located six amino acids upstream of the LL motif in gp130 (24,25). However, our attempts to demonstrate a physical association between LIFR␣ and AP2 by the experimental approach of coimmunoprecipitation, as applied by Thiel et al. (24), were unsuccessful. 2 An alternative mode of Ser 185 -dependent pathway could be proposed that is comparable to the pathway demonstrated for insulin receptor down-regulation (37). This mode proposes that the endocytosis of LIFR␣ is not affected by Ser 185 modification but that the tagged LIFR␣ is transferred more efficiently from endosomes to lysosomes and/or subject to reduced recycling to the cell surface. The subcellular transfer of LIFR␣ could also be promoted by ERK affecting cytoskeletal functions such as myosin light chain kinase activity (46). Either of the two mechanisms of regulated turnover will result in 2 Frédéric Blanchard and Heinz Baumann, unpublished data. The results with NIH3T3 cells indicated yet another mechanism of LIFR␣ down-regulation that appears to be independent of activated ERK (insensitive to U0126). This down-regulation occurs only after autologous stimulation with LIF. Since inhibitors to ERK, PI3K, or p38 MAPK did not prevent this type of receptor regulation 2 (Fig. 8A), two possible alternative mechanisms are possible. One mechanism would propose involvement of other kinases: 1) by Ser phosphorylation through a non-ERK mechanism, such as casein kinase II, implicated in Man-6-P/IGF-IIR, CD3-␥ chain, and CD4 down-regulation (47)(48)(49); 2) by tyrosine phosphorylation, like that for EGFR, to enhance the interaction with ␣-adaptin (AP2; Ref. 50); or 3) by ubiquitination of the cytoplasmic domain, to increase its internalization, as seen with the GHR (51,52). In this case, the ubiquitin conjugation system (E3) would assist in an adaptor role between the receptor and ␣-adaptin (AP2). The other mechanism proposes that the decrease in receptor level is independent of signal transduction, but relies on ligand-induced receptor heterodimerization. The close proximity of the suggested ␣-adaptin (AP2) binding site in the cytoplasmic domain of the LIFR␣ and gp130 in the heteromeric receptor complex would result in a higher recruitment in clathrin-coated pits than the corresponding monomeric receptor chains in the absence of the ligand. This form of turnover is assumed to be cell type-specific, i.e. highly effective in NIH3T3 cells. Yet, this intracellular process of lysosomal degradation seems to be restricted to LIFR␣ and accounts for the differential turnover of LIFR␣ and gp130 (Fig. 8A).
LIF response can be divided into early receptor-initiated reactions (ligand binding, JAK activation, and protein recruitment to receptor), and receptor downstream events that affect receptor endocytosis/degradation and attenuate signaling reactions. The latter process involves regulatory mechanisms such as dephosphorylation of JAK or STAT3 by SHP1/2 or inhibition by SOCS family proteins (18 -21, 33). Moreover, activation of ERK, not only by LIF but also by phorbol 12-myristate 13acetate, insulin, or growth factors (EGF, fibroblast growth factor) moderates the effects of IL-6 type cytokines (22,27,45,53). Activated ERK has at least two modes of influence on gp130derived signals: 1) by tagging LIFR␣ for down-regulation (this paper, Ref. 22), and 2) by affecting expression of signal moderating components, i.e. SOCS3, which in turn interfere with gp130, preventing further activation of STAT by IL-6 (45). The ability to adjust LIFR␣ level at the cell surface is in part correlated with a reduced LIF responsiveness. This attenuated LIF response is particularly evident in hepatic cells that have been chronically treated with IL-6 cytokines or insulin. We have shown here, using H-35 hepatoma cells, that insulin can effectively lower LIF activity to a greater extent than IL-6 or OSM activity, presumably because of a specific down-regulation of LIFR␣. The process of ERK-dependent reduction of LIFR␣ protein is predicted to have physiological relevance in situations in which LIF is presented to target cells in combination with other effectors that stimulate ERK pathways. A particularly relevant influence in limiting the LIF response is suggested to occur during the inflammatory response, during which circulating levels of LIF, but also OSM and insulin are increased (5,38,39). Finally, since LIFR␣ is widely expressed on various cell types such as liver cells, fibroblasts, and epithelial cells, including their cancer derivatives, the down-regulation of LIFR␣ could also explain why the cytokine LIF is usually found less active than the other related cytokines IL-6 and OSM in supporting long term effects (27,40,42). LIFR␣ is also part of the high affinity receptor complex for CNTF, CT-1, neurotrophin-1, and type I hOSMR complex, suggesting that growth factors have the potential to decrease cell responsiveness to these LIF-related cytokines as well.
LIF, and related cytokines, are involved in the regulation of hematopoiesis, immune response, inflammatory processes, bone and muscle remodeling, neural development, kidney differentiation, and mammalian reproduction (1)(2)(3)(4)54). Our study suggests that not only are levels of IL-6 cytokines important, but also the specific restriction of cytokine responsiveness by modulated receptor subunit levels.