Phosphorylation of Human gp130 at Ser-782 Adjacent to the Di-leucine Internalization Motif

The receptor for leukemia inhibitory factor (LIF) consists of two polypeptides, the LIF receptor and gp130. Agonist stimulation has been shown previously to cause phosphorylation of gp130 on serine, threonine, and tyrosine residues. We found that gp130 fusion proteins were phosphorylated exclusively on Ser-782 by LIF- and growth factor-stimulated 3T3-L1 cell extracts. Ser-780 was required for phosphorylation of Ser-782 but was not itself phosphorylated. Ser-782 is located immediately N-terminal to the di-leucine motif of gp130, which regulates internalization of the receptor. Transient expression of chimeric granulocyte colony-stimulating factor receptor (G-CSFR)-gp130(S782A) receptors resulted in increased cell surface expression in COS-7 cells and increased ability to induce vasoactive intestinal peptide gene expression in IMR-32 neuroblastoma cells when compared with expression of chimeric receptors containing wild-type gp130 cytoplasmic domains. These results identify Ser-782 as the major phosphorylated serine residue in human gp130 and indicate that this site regulates cell surface expression of the receptor polypeptide.

its lipoprotein lipase activity in adipocytes, stimulates myoblast proliferation, and converts sympathetic neurons from noradrenergic to cholinergic phenotypes (Refs. 1-3 and the references therein). LIF, ciliary neurotrophic factor, oncostatin M, interleukin-11, cardiotrophin-1, and interleukin-6 make up a distinct subgroup of the cytokine superfamily that all exhibit a four-anti-parallel ␣-helical bundle structure arranged in a two-up, two-down configuration and associate with the shared signaling molecule, gp130, after initial binding to their unique low affinity ␣-receptor binding subunits (1)(2)(3)(4).
LIF signals through a heterodimeric receptor complex consisting of LIFR and gp130 (5). Lacking intrinsic enzymatic activity (6), initiation of LIF receptor signaling occurs in a manner analogous to that of activated growth factor receptors (7,8). LIF-stimulated dimerization of LIFR and gp130 results in activation of the Jak/Tyk family of nonreceptor protein tyrosine kinases (4,9). The activated Jaks then phosphorylate and activate the signal transducers and activators of transcription (STATs), which regulate a variety of cytokine-responsive genes (10 -13). Activated LIF receptors have also been shown to recruit and stimulate numerous SH2-containing signaling molecules, including phospholipase C-␥, Shc, Grb2, phosphoinositol 3-kinase, pp120, and the protein-tyrosine phosphatase SHP-2 (14).
Activated LIF receptors also stimulate several components of the MAP kinase cascade, including MAPK kinase, the MAPK isozymes ERK1 and ERK2, and the S6 protein kinases pp90 rsk and pp70 S6K . MAPK-mediated phosphorylation of Thr-235 of NF-IL6/C-EBP-␤ (15) stimulates the factor's ability to mediate gene transcription of certain LIF-responsive genes (16,17). In addition, phosphorylation of STAT1 and STAT3 by MAPK and other proline-directed Ser/Thr protein kinases can have both positive and negative effects on gene induction (10, 18 -20). Recently, it was shown that STAT1 is a poor substrate for MAPK both in vitro and in vivo (21). In contrast, phosphorylation of STAT3 by MAPK prevents STAT3 from becoming tyrosine-phosphorylated, a step necessary for its activation. In addition, interleukin-6 was shown to stimulate serine phosphorylation of STAT3 in a MAPK-independent manner (21).
The role of MAPK in LIF receptor signaling is further complicated by our recent finding that human LIFR is phosphorylated at Ser-1044 by MAPK and that heterologous receptormediated attenuation of LIFR-stimulated gene induction occurs through effects at Ser-1044 (22). Interestingly, gp130, in addition to containing elevated levels of phosphotyrosine, is phosphorylated in intact cells on Ser and Thr residues in an agonist-dependent manner (23). Phosphorylation of transmembrane receptors by Ser/Thr protein kinases is known to modulate receptor activities, internalization, and down-regulation in response to agonist (24).
We phosphorylated fusion proteins containing the cytoplasmic domain of human gp130 and used Edman sequencing followed by matrix-assisted laser desorption/ionization time of flight mass spectrometry (MALDI-TOF MS) to identify Ser-782 as the major site of agonist-stimulated phosphorylation of human gp130 in 3T3-L1 cells. Unlike human LIFR, however, phosphorylation of human gp130 was mediated by a protein kinase that was clearly distinct from MAPK but whose activation paralleled stimulation of the MAP kinase cascade in 3T3-L1 cell extracts. When compared with chimeric receptors that contained the wild-type cytoplasmic domain of gp130, chimeric gp130 receptors containing an alanine substitution at Ser-782 were expressed on the cell surface at 3-8 times higher levels and showed increased tyrosine phosphorylation in response to agonist. Furthermore, when expressed in the neuronal cell line IMR-32, the S782A mutant chimeric receptor induced vasoactive intestinal peptide (VIP) gene transcription at higher levels than the wild-type receptor. These results indicate that LIF or growth factor stimulates phosphorylation of human gp130 at Ser-782, and this site serves to regulate the cell surface expression of this receptor polypeptide.

Materials
Recombinant human LIF and murine EGF were from Alomone (Jerusalem, Israel) or Upstate Biotechnology Inc. (Lake Placid, NY), respectively. G-CSF and the cDNA constructs encoding human gp130 or chimeric G-CSFR-gp130 were provided by Bruce Mosley (Immunex Corp., Seattle, WA). Monoclonal PY20 anti-phosphotyrosine antibody was from ICN Pharmaceuticals Inc. (Costa Mesa, CA), monoclonal 4G10 anti-phosphotyrosine antibody was from Upstate Biotechnology, Inc., and protein G-agarose was from Roche Molecular Biochemicals. Polyclonal anti-gp130 antibodies were described previously (25). Myelin basic protein (MBP) was purchased from Sigma, and Dulbecco's modified Eagle's medium, fetal bovine serum, and penicillin-streptomycin were obtained from Life Technologies, Inc. Iminodiacetic acid-Sepharose was from Sigma. Phenylisothiocyanate, triethylamine, pyridine, and heptane were sequencer grade, from Pierce; ethyl acetate and n-butyl chloride were sequencer grade, from Beckman; trifluoroacetic acid was protein sequencer grade, from Perkin-Elmer; acetonitrile was HPLC grade, from J. T. Baker Inc.. Sulfo-NHS-Biotin and immobilized streptavidin were purchased from Pierce. All additional supplies or materials were routinely available.

cDNA Construction
Fusion Protein Construction and Purification-The fusion proteins used in this study were synthesized by polymerase chain reaction amplification of amino acids 637-912 or various truncated cytoplasmic regions of human gp130. Isolated BamHI-EcoRI fragments were subsequently subcloned into the C terminus of GST encoded by the bacterial expression vector pGEX-3X (Amersham Pharmacia Biotech). Sitedirected mutagenesis of Ser-780 and Ser-782 were performed by sequential overlap polymerase chain reaction (26) with mutagenic oligonucleotides containing single base substitutions to replace Ser with Ala. Expression and purification of the fusion proteins was as described previously (22). All fusion protein constructs were sequenced on an Applied Biosystems 373A DNA sequencing system.
Chimeric G-CSFR-gp130(S782A) Construction-The generation of G-CSFR-gp130(S782A), consisting of the ligand binding domain of the G-CSFR linked to the complete cytoplasmic domain of human gp130 with an Ala for Ser substitution at position 782, was accomplished by sequential overlap polymerase chain reaction (26) with mutagenic oligonucleotides as described above. The outer 5Ј and 3Ј oligonucleotides contained NsiI and SphI restriction sites, respectively, in order to facilitate subsequent subcloning of the isolated fragment back into the parental mammalian expression vector, pDC302. The newly generated chimeric G-CSFR-gp130(S782A) cDNA construct was sequenced throughout its entire coding region on an Applied Biosystems 373A DNA sequencing system.
Cell Culture-3T3-L1 preadipocytes were cultured, stimulated, and prepared for determination of protein kinase activity as described previously (25). COS-7 cells were cultured and transiently transfected by the calcium phosphate/chloroquine method, as described previously (25). IMR-32 cells were grown and transfected with 10, 20, or 40 ng of cDNA encoding the VIP-luciferase reporter gene and either G-CSFR-gp130 or G-CSFR-gp130(S782A) using the calcium phosphate method and assayed as described previously (27,28).

Protein Kinase Assays
MBP Assays-Analysis of MBP phosphorylation in clarified 3T3-L1 cell extracts was performed as described previously (29).
Fusion Protein Assays-Phosphorylation of the cytoplasmic domain of human gp130 in clarified 3T3-L1 extracts was performed as described previously for phosphorylation of human LIFR (22). Briefly, phosphotransferase activity against human gp130 fusion proteins was measured after incubation at 30°C in a final reaction volume of 30 l consisting of 1.5 g of cell extract, 25 mM ␤-glycerophosphate, 0.5 mM dithiothreitol, 50 M sodium vanadate, 10 mM MgCl 2 , and 50 -100 M ATP ([␥-32 P]ATP, ϳ2000 cpm/pmol), and 5 g of fusion protein or native GST. Reactions were quenched by the addition of 10 l of 4ϫ Laemmli sample buffer (200 mM Tris-HCl, 4% SDS, 4% ␤-mercaptoethanol, and 40% glycerol), boiled for 5 min, and fractionated through 10% SDS-PAGE. Specific phosphorylation of human gp130 fusion proteins, determined after excision and scintillation counting of the appropriate substrate bands, was calculated by subtracting the radioactivity incorporated into native GST samples from the values obtained by subtraction of radioactivity incorporated into samples incubated in the absence of substrate from those incubated in its presence. Determination of phosphotransferase activities against LIFR and gp130 fusion proteins by active recombinant ERK2 was performed as described previously (22).

Preparation of Tryptic Phosphopeptide
GST-cgp130 (100 g) was phosphorylated overnight at 30°C with LIF-stimulated 3T3-L1 cell extract as described above. After SDS-PAGE and autoradiography, the radiolabeled protein bands were excised and digested with 2 g of sequencing grade trypsin (Promega) as described previously (30,31). The tryptic phosphopeptides were then purified by Fe 3ϩ iminodiacetic acid chromatography as described previously (32)(33)(34). After elution with 100 mM potassium phosphate, pH 8.6, the 32 P-labeled phosphopeptide was further purified by C 18 HPLC chromatography. HPLC analyses were performed on a Hewlett Packard HP1090 liquid chromatographic system using a 2.1 ϫ 30-mm, 5-m C 18 column. The gradient solvents were aqueous 0.1% trifluoroacetic acid (solvent A) and 80% acetonitrile, 0.1% trifluoroacetic acid (solvent B). After sample injection and washing with 100% A, the column was developed with a linear gradient of 100% A to 25% A, 75% B over 30 min. The flow rate was 0.175 ml/min, and 0.45-ml fractions were collected and counted by Cerenkov radiation.

Manual Edman Sequencing
The procedure of Levy (35) was generally followed. The peptide to be sequenced was dried in a 1.5-ml microcentrifuge tube and dissolved in 40 l of coupling buffer (7.5 ml pyridine, 0.69 ml triethylamine, and 5 ml water) with the pH adjusted to 9.5 using trifluoroacetic acid (stored at 4°C for 1 week maximum). The microcentrifuge tube was flushed with argon, and 3 l of phenylisothiocyanate was added. The sample was vortexed, heated at 50°C for 30 min, and then extracted twice with 150 l of heptane-ethyl acetate (10:1) and once with heptane-ethyl acetate (2:1). The extraction mixtures were vortexed under argon and centrifuged to achieve phase separation. After drying the aqueous phase in a Speed Vac, trifluoroacetic acid (20 l) was added, and the cleavage reaction was allowed to proceed at 50°C for 20 min. After drying, the sample was dissolved in 40 l of 30% aqueous pyridine and extracted three times with 150 l of n-butyl chloride. The aqueous phase, containing the residual peptide was dried in the Speed Vac for mass spectral analysis and, as required, a subsequent Edman cycle.
MALDI-TOF MS-Mass spectral analyses were performed on a Per-Septive Biosystems Voyager-DE operated in delayed (150 ns) extraction mode with a 20-kV accelerating voltage, 92.5% grid voltage, and 0.2% guide wire voltage. Samples were prepared on a 10 ϫ 10 multi-well smooth sample plate using the matrix ␣-cyano-4-hydroxycinnamic acid purchased from Aldrich. Saturated ␣-cyano-4-hydroxycinnamic acid solution was prepared in 40% acetonitrile with 0.1% trifluoroacetic acid and mixed 1:1 (v/v) with sample on the MALDI plate. Eighty scans were accumulated for each spectrum. The data were analyzed using the GRAMS/386 v3.0 software provided with the instrument. Bovine insulin obtained from Sigma was used for internal calibration.
Samples containing high salt concentrations were first spotted on the MALDI plate as above. After air-drying, the spots were overlaid with 3 l of water for 5 s, and the water spot was removed with a pipette. The salt extraction process was repeated, and the spot was allowed to air-dry before analysis.

Immunoprecipitation and Western Blot of Transiently Expressed Chimeric gp130 Receptors
Transiently transfected chimeric gp130 receptors were isolated from COS-7 cells by immunoprecipitation with anti-gp130 antibodies and blotted as described previously (25).
Phosphoamino Acid Analysis-Phosphoamino acid analyses were performed by the method of Kamps and Sefton (37).

RESULTS
Phosphorylation of transmembrane receptors on Ser and Thr residues can alter and regulate several aspects of receptor biology, including the levels of receptor activity, internalization, and down-regulation (24). We have recently shown that Ser-1044 is the major Ser/Thr-phosphorylated residue within the human LIFR and that phosphorylation of Ser-1044 is mediated by activated MAPK in LIF-or growth factor-stimulated 3T3-L1 extracts (22). Furthermore, heterologous receptor-mediated attenuation of LIFR-stimulated gene induction requires Ser-1044 (22). Murakami et al. (23) demonstrated that gp130 is phosphorylated on Tyr and Ser/Thr residues in response to interleukin-6 treatment. Because LIFR and gp130 exhibit extensive homology throughout their cytoplasmic domains (6) and because gp130 is a signaling component of activated LIF receptors, we tested the cytoplasmic domain of human gp130 to see it is similarly phosphorylated by MAPK in response to agonist.
To address this question, we tested the ability of a GST fusion protein containing the cytoplasmic domain (amino acids 637-912) of human gp130 to serve as a substrate in protein kinase reactions. We found that LIF treatment of quiescent 3T3-L1 cells potently and rapidly stimulated phosphotransferase activity against the human gp130 fusion protein construct, GST-cgp130 (Fig. 1). Phosphorylation of GST-cgp130 in LIF-stimulated lysates occurred in a dose-dependent manner (Fig. 1A), with half-maximal stimulation (EC 50 ) occurring at ϳ28 ng/ml (ϳ1.4 nM). In response to saturating concentrations of LIF, phosphorylation of GST-cgp130 was found to peak at 10 min, remain significantly elevated at 30 min, and decline steadily thereafter to near basal levels by 60 min (Fig. 1B). Thus, the cytoplasmic domain of human gp130, like that of human LIFR, is readily phosphorylated in LIF-stimulated 3T3-L1 cell extracts.
The dose-and time-dependent profiles for LIF-stimulated phosphorylation of GST-cgp130 were indistinguishable from those for phosphorylation of either MBP (22,29) or human LIFR (22) in 3T3-L1 extracts treated with LIF. Because LIFstimulated phosphorylation of MBP and human LIFR is mediated by the MAPK isozymes ERK1 and ERK2, these data suggest that phosphorylation of human gp130 in LIF-stimulated 3T3-L1 cells may be mediated in part by activated MAPK. Consistent with this hypothesis and similar to human LIFR (22), we found that EGF and PMA, mitogens that activate MAPK in 3T3-L1 cells (29), stimulated the phosphorylation of GST-cgp130 as effectively as LIF in extracts of 3T3-L1 cells (Fig. 2). However, unlike human LIFR, which was phosphorylated stoichiometrically by active recombinant ERK2 in vitro, GST-cgp130 failed to exhibit significant phosphate incorporation under similar assay conditions (Fig. 3), demonstrating that human gp130 was an extremely poor substrate for activated recombinant ERK2 in vitro. In addition, fractions isolated after separation of LIF-stimulated cell extracts on Mono-Q-ion exchange chromatography, which contained activated ERK2 and ERK1, did not contain gp130-phosphorylating activity. 2 These data demonstrate that phosphorylation of human gp130 was not mediated by activated MAPK but rather was due to stimulation of a protein kinase whose activation paralleled stimulation of the MAPK cascade in 3T3-L1 cells.
Phosphorylation of GST-cgp130 in LIF-stimulated 3T3-L1 extracts occurred exclusively on Ser residues (Fig. 4A) located between amino acids 740 and 890 (Fig. 4B). Point mutations in GST-cgp130 were introduced that replaced Ser-780 and Ser-782 to alanine either singly or in combination. As shown in Fig.  5, phosphorylation of GST-cgp130(S782A) was significantly reduced compared with its serine-containing counterpart. In contrast, although the phosphorylation of GST-cgp130(S780A) was less than wild-type (ϳ35%), it was greater than either GST-cgp130(S782A) or the double mutant, GST-cgp130(S780A/ S782A), where in each case, the phosphorylation was ϳ20% that of wild-type. The fact that the phosphorylation of GST-cgp130(S780A/S782A) was not statistically different from GST-cgp130(S782A) is consistent with a sequential phosphorylation model in which prior phosphorylation of Ser-782 is required for maximal phosphorylation of Ser-780. These results do not, however, exclude the possibility that Ser-782 is the only major site of agonist-induced serine phosphorylation in gp130, since the decreased phosphorylation of GST-cgp130(S780A) could be due to an effect of the S780A mutation on the phosphorylation of Ser-782 rather than to the absence of a phosphorylation site at Ser-780.
To determine definitively which sites on gp130 were phosphorylated in LIF-stimulated extracts, tryptic peptides were prepared from phosphorylated wild-type GST-cgp130, and the tryptic phosphopeptides were isolated using Fe 3ϩ iminodiacetic acetic acid chromatography. MALDI-TOF MS revealed that the Fe 3ϩ Chelex-purified sample contained two peptides, corresponding to the unphosphorylated and monophosphorylated tryptic peptide (amino acids 780 -811) containing Ser-780 and Ser-782 (Table I). This material was then subjected to three successive rounds of Edman degradation followed by analysis of the residual peptides by MALDI-TOF MS. Following the first turn of Edman degradation, two major peaks were observed containing MHϩ average masses of 3431.38 and 3511.57. As indicated in Table I peptide with an N-terminal phosphoserine (Ser-780).
The MALDI spectrum following the second cycle of Edman degradation contained two peaks with MHϩ average masses of 3302.11 and 3382.28. The MHϩ 3382.28 peptide must correspond to a monophosphorylated peptide missing the first two amino acids. Again, the MHϩ 3302.11 peptide could arise from either the loss of a serine and a glutamate from the nonphosphorylated peptide or from the loss of an N-terminal phosphoserine and a glutamate from a monophosphorylated peptide phosphorylated at Ser-780. To determine which of these possibilities was correct, the aqueous phase following the second round of Edman degradation was applied to a C 18 reversed phase HPLC to separate inorganic phosphate from the residual phosphopeptide. As shown in Fig. 6, less than 3% of the applied radioactivity was detected in the flow-through (Fig. 6A). Instead, the 32 P eluted deep in the gradient in fractions 17-19, coincident with significant UV absorbance. Fraction 18 was concentrated and analyzed by MALDI. The spectrum contained peaks with MHϩ average masses of 3302.11 and 3382.28, corresponding to the nonphosphorylated and monophosphorylated peptides beginning with Ser-782, respectively. These data indicate Ser-780 was not phosphorylated in LIF-or growth factor-stimulated extracts.
To further validate the reversed phase HPLC analysis and to determine if Ser-782 was phosphorylated versus Thr-783 or Ser-789, a third Edman degradation was conducted on the residual peptide. The MALDI analysis revealed a single peak (MHϩ average mass 3215.54), corresponding to a nonphosphorylated peptide starting with Thr-783. No phosphorylated peptide was observed. This time, the reversed phase HPLC analysis revealed a significant amount of radioactivity in the flowthrough, indicative of the released phosphorylated N-terminal amino acid at position Ser-782 (Fig. 6B). The fact that not all of the radioactivity was obtained in the flow-through leaves ambiguity regarding assignment of all of the phospho amino acid to Ser-782; therefore, MALDI analysis was performed on fraction 18. The spectrum revealed peaks with MHϩ average masses of 3382.28, 3302.01, and 3214.88, corresponding to phosphorylated peptide beginning with Ser-782, nonphosphorylated peptide beginning with Ser-782, and nonphosphorylated peptide beginning with Thr-783, respectively. A phosphopeptide of sequence TQPLLDSEERPEDLQLVDHVDGGDGILPR containing either a phosphothreonine or a phosphoserine has a calculated MHϩ average mass of 3295.5. No such peaks were observed in fraction 18.
Taken together, the data show conclusively that Ser-782 was the only site of agonist-induced serine phosphorylation on gp130. Interestingly, Ser-782 is located immediately upstream of a 10-amino acid region that contains a di-leucine motif which Dittrich et al. (38,39) found to be necessary for agonist-mediated internalization of gp130. Based on analogies to other receptor systems that require Ser phosphorylation of residues immediately flanking di-leucine motifs during receptor inter-nalization, these authors speculated that PKC-or casein kinase II-mediated phosphorylation at Ser-780 of gp130 may regulate agonist-stimulated internalization of this polypeptide (39). Our results, however, demonstrated that Ser-782 of gp130, not Ser-780, was the major site of LIF-and growth factor-stimulated phosphorylation of this receptor polypeptide in 3T3-L1 extracts (Fig. 5).
We have found that GST-cgp130 and GST-cgp130(S782A) were both phosphorylated to equivalent low extents by purified PKC in vitro, which incorporated 0.031 and 0.026 mol of phosphate/mol of substrate at 120 min into GST-cgp130 and GST-cgp130(S782A), respectively. 2 For comparison, this same preparation of purified PKC incorporated ϳ1.2 mol of phosphate/ mol of histone III-S at 5 min. Because the phosphorylation of gp130 by extracts from PMA-stimulated cells is blocked by Ser to Ala substitution at position 782, these data indicate that phosphorylation of human gp130 is not directly mediated by PKC in extracts of 3T3-L1 cells. Thus, the phosphorylation of human gp130 in extracts of PMA-stimulated cells must be mediated by a protein kinase downstream of phorbol estersensitive PKC isozymes. Preliminary experiments also failed to detect phosphorylation of gp130 fusion proteins following their incubation in vitro with purified casein kinase. 2 To determine the role of Ser-782 in regulating the expression and activation of gp130, we transiently transfected COS-7 cells with chimeric receptors containing the extracellular and transmembrane domains of the G-CSF receptor fused to the cytoplasmic domain of either gp130 or gp130(S782A). The COS-7 transfectants were immunoprecipitated with anti-gp130 antibody and probed with either anti-phosphotyrosine (Fig. 7A), to quantitate receptor activation, or anti-gp130 (Fig. 7B), to quantitate receptor expression. For each plasmid concentration, the level of cytokine-induced receptor tyrosine phosphorylation was greater in cells transfected with G-CSFR-gp130(S782A) than in cells transfected with G-CSFR-gp130 (Fig. 7A). As shown in Fig. 7B, the Ala-782-containing chimeric gp130 receptors were typically expressed ϳ2-10 times higher at equal cDNA concentrations than were their Ser-782-containing counterparts.
To determine whether the S782A mutation caused increased cell surface expression compared with wild-type, transiently transfected COS-7 cells were labeled with Sulfo-NHS-Biotin. The biotinylated surface proteins were then solubilized and precipitated using streptavidin-agarose and analyzed by SDS-PAGE followed by Western blot with anti-gp130 antibody. When equal amounts of cDNA containing the G-CSFR-gp130 chimeras were transiently expressed in COS-7 cells (Fig. 8), the S782A mutant showed a 6.7 Ϯ 1.9-fold increase in cell surface expression compared with wild-type (mean Ϯ S.D., 5 independent experiments). The difference in tyrosine phosphorylation, therefore, is most likely due not to intrinsic differences in the ability of G-CSFR-gp130 and G-CSFR-gp130(S782A) to stimulate nonreceptor protein-tyrosine kinases but, rather, results from absolute differences in expression of these receptor polypeptides on the cell surface.
To further examine the role of Ser-782 in the ability of gp130 to mediate cytokine-induced signaling pathways, the chimeric receptor constructs G-CSFR-gp130 and G-CSFR-gp130(S782A) were transiently cotransfected in the neuronal cell line, IMR-32, with a reporter construct under the control of the promoter for the VIP gene and examined for their abilities to induce transcription of the luciferase reporter gene. As shown in Fig.  9, the level of VIP reporter gene induction was ϳ50% higher in cells expressing S782A-containing constructs compared with those expressing S782. These effects are consistent with the increased expression and activation of G-CSFR-gp130(S782A) FIG. 5. Phosphorylation of GST-cgp130 serine-to-alanine mutants. Quiescent 3T3-L1 cells were incubated at 37°C with diluent (phosphate-buffered saline; B), LIF (100 ng/ml for 10 min; L), or PMA (100 nM for 10 min; P). Agonist-stimulated phosphorylation of GST-cgp130, GST-cgp130(S780A), GST-cgp130(S782A), or GST-cgp130(S780/ 782A) was performed at 30°C for ϳ16 h, as described under "Experimental Procedures." Data shown is a representative autoradiograph (15-min exposure) of a single experiment that was repeated three times with similar results. WT, wild type. compared with G-CSFR-gp130 observed in COS-7 cells (Figs. 7  and 8). Taken together, the data suggest that the increased signaling of the Ala-782-containing construct is due to its being expressed at higher levels compared with the Ser-782-containing counterpart. These results are consistent with the hypothesis of Dittrich et al. (39) that phosphorylation of Ser residues flanking the gp130 di-leucine motif regulates internalization of gp130. Our results clearly demonstrate that Ser-782 of human gp130 serves to regulate the cell surface expression and functional responsiveness of this receptor polypeptide.

DISCUSSION
Homologous or heterologous receptor-stimulated Ser/Thr phosphorylation of transmembrane receptors is capable of regulating and altering various aspects of receptor biology, including receptor activities, internalization, and down-regulation (24). The positions of Ser-782 and Ser-1044 within the cytoplasmic domains of gp130 and LIFR, respectively, are shown in Fig.  10. We recently demonstrated that the human LIFR is phosphorylated at Ser-1044 by activated MAPK in extracts of 3T3-L1 cells stimulated with LIF, EGF, PMA, or insulin (22). We also found that insulin receptor-mediated attenuation of LIFR-stimulated gene induction requires Ser-1044 (22), suggesting that activation of heterologous receptor systems coupled to stimulation of the MAP kinase cascade are capable of regulating LIFR signaling through effects at its MAPK phosphorylation site. Murakami et al. (23) show that the binding of interleukin-6 to its receptor stimulates increased phosphorylation of gp130 on Tyr, Thr, and Ser residues. Because gp130 is the ␤-subunit of activated LIF receptors and the cytoplasmic domains of gp130 and LIFR are highly homologous (6), we asked whether the cytoplasmic domain of human gp130 could similarly serve as a substrate for activated MAPK in agoniststimulated 3T3-L1 extracts.
We found that the binding of LIF to its receptor stimulated the phosphorylation of the cytoplasmic domain of human gp130 in extracts of 3T3-L1 cells with a time and LIF dose dependence identical with those for activation of MAPK by LIF in 3T3-L1 cells (22,29), suggesting the possibility that human gp130, like human LIFR, might be a substrate for activated MAPK. However, the results from our studies involving in vitro incubation of gp130 fusion proteins with activated ERK2 clearly demonstrated that LIF-and agonist-stimulated phosphorylation of human gp130 was not mediated by MAPK.
Ser-782 of gp130 is located four amino acids upstream of a di-leucine motif previously shown by Dittrich et al. (38) to be necessary for ligand-induced internalization of the receptor polypeptide. The results of several studies have established that di-leucine (or leucine-isoleucine) motifs are necessary in mediating receptor trafficking and internalization reactions in a variety of receptor polypeptides, including the T-cell receptor CD3-␥ and -␦ polypeptides (40 -42), CD4 (43)(44)(45)(46), the cationindependent mannose 6-phosphate/insulin-like growth factor II receptor (47,48), and the insulin receptor (49). Expression of gp130 polypeptides in COS-7 cells that contained point mutations of the di-leucine motif itself or to amino acids flanking the di-leucines demonstrated that a Ser-to-Ala transition at position 780 reduced agonist-stimulated internalization of the receptor polypeptide by ϳ50% (39). These authors also attempted to examine the effects of mutation of Ser-782, but a high degree FIG. 6. HPLC profiles of aqueous phase from Edman cycles 2 and 3 of 32 P-labeled tryptic peptide. 32 P-Labeled tryptic peptides were prepared from GST-cgp130 as described under "Experimental Procedures." After two (A) or three (B) rounds of Edman degradation, the aqueous phase was applied to a C 18 reversed phase HPLC column, and the radioactivity of the collected fractions was determined by Cerenkov counting. In panel B, the indicated MHϩ average mass 3382.28 was determined by MALDI analysis of fraction 18. of variability in their data precluded any conclusions on its role in internalization.
Based on sequence similarities with the receptor polypeptides of CD3-␥ (42), CD4 (44,45), and the cation-independent mannose 6-phosphate receptor/insulin-like growth factor II receptor (47, 48), Dittrich et al. (39) speculated that phosphorylation of Ser-780 may be necessary, but not sufficient, for agonist-stimulated internalization of gp130. Our results, based on Edman sequencing of phosphorylated gp130, demonstrate conclusively that Ser-780 is not phosphorylated. Instead, we found that the major site for phosphorylation of human gp130 in LIF-, EGF-, and PMA-stimulated 3T3-L1 extracts was Ser-782. Our phosphorylation data with the Ser-to-Ala mutants, S780A, S782A, and S780A/S782A, showed that S780A was phosphorylated less than wild-type but more than either S782A or S780A/S782A. Taken together with the sequencing results that showed Ser-782 is the only site of phosphorylation, the data indicate that Ser-780 is required for phosphorylation of Ser-782.
The di-leucine motif in gp130 has been shown to be important both for internalization and for subsequent degradation of the polypeptide (39). Persistent stimulation with gp130-family cytokines leads to both desensitization of functional responses and down-regulation of gp130 and other receptor subunits (39, 50 -53), although the extent of these processes varies in different cell types (viz. see Ref. 54). In addition, although gp130coupled cytokines have been reported to induce internalization of gp130 and other receptor subunits (39,55), it has also been suggested that gp130 is constitutively internalized in a ligandindependent manner (56). Many receptors exhibit cell-type specific differences in their modes of regulation of cell-surface expression (57), and LIFR-gp130 heterodimers are regulated differently from gp130-homodimers (58). Thus, it is likely that the factors that regulate the internalization and subsequent degradation of the receptor subunits of the gp130-family of cytokines will depend on both the specific cytokine (and thus the specific combination of receptor polypeptides) and the specific cell type under consideration.
The G-CSFR-gp130(S782A) receptors were expressed at higher levels on the cell surface in COS-7 cells and exhibited greater ligand-stimulated tyrosine phosphorylation than Ser-782-containing receptors. Furthermore, the Ala-782-containing receptors induced higher levels of a vasoactive intestinal peptide-luciferase reporter gene construct when expressed in the neuronal cell line, IMR-32. Both of these effects are likely due to the increased receptor expression levels of the Ala-782-containing receptor compared with its Ser-782-containing counterpart. Thus, unlike the effects on LIFR signaling mediated by phosphorylation of Ser-1044 by activated MAPK, which attenuates receptor signaling, agonist-stimulated phosphorylation of gp130 at Ser-782 by an unknown protein kinase does not appear to significantly affect or regulate the signaling capabilities of this receptor polypeptide. Instead, phosphorylation of Ser-782 on gp130 appears to be involved in regulating the expression and/or down-regulation of the receptor polypeptide.
The increase in receptor cell surface expression could be due to either decreased receptor internalization or increased recycling. Since Ser-782 is located adjacent to the di-leucine internalization motif previously shown to be involved in both ligandstimulated endocytosis and lysosomal degradation of gp130 (38,39), the increase in G-CSFR-gp130 cell surface expression caused by the S782A mutation is most likely due to impaired receptor internalization rather than to an effect of this mutation on receptor recycling. The increased cell surface expression of G-CSFR-gp130(S782A) was observed in the absence of G-CSF stimulation. Although this may suggest that phosphorylation of Ser-782 does not play a role in regulating receptor surface expression, the presence of significant levels of Ser-782specific kinase activity in non-stimulated cells could result in a basal level of phosphorylation of Ser-782, which regulates gp130 cell surface expression even in the absence of stimulating ligand. Future studies will hopefully identify the kinase involved in phosphorylating gp130 at Ser-782 and will further elucidate the role of this phosphorylation in regulating the surface expression and internalization of the receptor polypeptide.
In summary, we have shown that Ser-782 of human gp130 is the major phosphorylated residue in extracts of 3T3-L1 cells stimulated with LIF and mitogens. Chimeric Ala-782-containing gp130 receptors were expressed at higher levels than their Ser-782 counterparts. Furthermore, this increased expression was found to be on the cell surface and led to increased gp130mediated functional responsiveness. Although additional studies are needed to fully clarify the role of Ser-782 during gp130mediated signal transduction, our results suggest that Ser-782 is capable of regulating gp130 cell surface expression during di-leucine motif-mediated internalization.