A di-leucine motif and an upstream serine in the interleukin-6 (IL-6) signal transducer gp130 mediate ligand-induced endocytosis and down-regulation of the IL-6 receptor.

The interleukin-6 (IL-6) receptor complex is composed of two different subunits, the IL-6 binding protein (IL-6R, gp80) and the signal transducing component gp130. Our previous studies revealed that the 10-amino acid sequence TQPLLDSEER within the intracellular domain of gp130 is crucial for the efficient internalization of IL-6. Since this sequence contains a putative di-leucine internalization motif, we further analyzed this region by constructing two additional deletions and a series of point mutants. Analyses of these mutants showed that the di-leucine pair (Leu-145 and Leu-146) is essential for ligand internalization, with leucine 145 being less resilient to exchanges. Furthermore, when a chimeric protein (Tac-STQPLL) composed of the Tac antigen fused to the hexapeptide STQPLL of gp130 was studied, we found that this sequence is sufficient to mediate endocytosis and lysosomal targeting of the chimera. Mutational analysis of three serine residues upstream of the di-leucine motif revealed that mutation of serine 139 to an alanine reduces the initial internalization rate by 50%. This finding suggests that a serine phosphorylation may be important for rapid endocytosis.

The interleukin-6 (IL-6) receptor complex is composed of two different subunits, the IL-6 binding protein (IL-6R, gp80) and the signal transducing component gp130. Our previous studies revealed that the 10-amino acid sequence TQPLLDSEER within the intracellular domain of gp130 is crucial for the efficient internalization of IL-6. Since this sequence contains a putative dileucine internalization motif, we further analyzed this region by constructing two additional deletions and a series of point mutants. Analyses of these mutants showed that the di-leucine pair (Leu-145 and Leu-146) is essential for ligand internalization, with leucine 145 being less resilient to exchanges. Furthermore, when a chimeric protein (Tac-STQPLL) composed of the Tac antigen fused to the hexapeptide STQPLL of gp130 was studied, we found that this sequence is sufficient to mediate endocytosis and lysosomal targeting of the chimera. Mutational analysis of three serine residues upstream of the di-leucine motif revealed that mutation of serine 139 to an alanine reduces the initial internalization rate by 50%. This finding suggests that a serine phosphorylation may be important for rapid endocytosis.
Interleukin-6 (IL-6) 1 displays a broad range of biologic activities (for review, see Kishimoto and Hirano (1988), Van Snick (1990), ) including the stimulation of immunoglobulin production by activated B cells (Hirano et al., 1985), the regulation of myeloma/plasmacytoma (Kawano et al., 1988;Schwab et al., 1991) and hybridoma proliferation (Nordan and Potter, 1986;Van Damme et al., 1987), and the induction of acute phase protein synthesis in liver cells (Gauldie et al., 1987;Andus et al., 1987Andus et al., , 1988Castell et al., 1990). Binding of IL-6 to its receptor (IL-6R) induces homodimerization of two gp130 molecules, thereby transducing the signal into the cell (Murakami et al., 1993). Monomeric gp130 is phosphorylated on serine and threonine, but not on tyrosine (Murakami et al., 1993). In contrast, dimerized gp130 is also phosphorylated on tyrosine and exhibits an increased phosphorylation on serine and threonine residues (Murakami et al., 1993). Whereas the identity of gp130-associated serine-threonine kinases remains to be elucidated, it was recently shown that IL-6 stimulation triggers the activation of tyrosine kinases of the JAK family already bound to the cytoplasmic domain Stahl et al., 1994). Moreover, the signal transducers and activators of transcription (STAT) APRF/ STAT3 and STAT1␣ become tyrosine-phosphorylated in response to IL-6 (Lü tticken et al., 1994;Akira et al., 1994). They homo-or heterodimerize and translocate to the nucleus, where they bind to responsive elements of IL-6 target genes . Before nuclear translocation, APRF/STAT3 and STAT1␣ undergo an additional serine phosphorylation (Lü tticken et al., 1995;Wen et al., 1995).
After binding to its receptor, IL-6 is rapidly internalized and degraded by rat hepatocytes and human hepatoma cells HepG2 (Nesbitt and Fuller, 1992;Zohlnhoefer et al., 1992). In HepG2 cells, this internalization leads to a net loss of IL-6 binding sites at the cell surface, indicating that the IL-6R is downregulated by its ligand (Zohlnhoefer et al., 1992). The downregulation might play an important role as a protection against overstimulation. Recently, we showed by co-expression of wildtype and mutant forms of both, the IL-6 receptor and gp130, in transiently transfected COS-7 cells that gp130 is essential for the efficient endocytosis of IL-6 and receptor down-regulation . Whereas the cytoplasmic domain of the IL-6 receptor was not significantly involved in the internalization process, deletion of the corresponding domain of gp130 resulted in an almost complete impairment of IL-6 internalization. Mutants with different truncations within the intracellular domain of gp130 led to the identification of the 10 amino acid sequence TQPLLDSEER that is crucial for efficient endocytosis .
During the last few years, it has become clear that efficient internalization of transmembrane receptor proteins requires a signal sequence in the cytoplasmic tail of the protein (Vaux, 1992). There are at least two different types of internalization signal sequences: tyrosine-and di-leucine-based internalization motifs. Tyrosine motifs have been identified in a variety of receptor molecules including the low density lipoprotein receptor, the transferrin receptor, and the asialoglycoprotein receptor (for review, see Trowbridge et al. (1993)). Substitution of single amino acids within such motifs and structural predictions suggested that an important characteristic of these sequences is the formation of a "tight ␤-turn" (Collawn et al., 1990;Bansal and Gierasch, 1991). This structure might be recognized by adaptor proteins, which mediate the accumulation of these receptors in clathrin-coated pits (Collawn et al., 1990). Recently, leucine-leucine or leucine-isoleucine sequence motifs, important for internalization and/or lysosomal targeting, were found in the intracellular domains of the ␥and ␦-chain of the T cell receptor complex, CD4, the cation-depend-ent mannose 6-phosphate receptor, the cation-independent mannose 6-phosphate/insulin-like growth factor II receptor, and the interferon ␥-receptor (Letourneur and Klausner, 1992;Shin et al., 1991;Johnson and Kornfeld, 1992;Chen et al., 1993;Farrar and Schreiber, 1993). It is not known whether the secondary structure of these motifs is related to a tight ␤-turn or not. It is the current hypothesis that these motifs reside within a random coil structure (Sandoval et al., 1994).
In this study, we demonstrate that the di-leucine motif (STQ-PLL, positions 141-146) in the cytoplasmic tail of gp130 is essential for the efficient endocytosis of IL-6. Furthermore, serine 139 located six amino acids upstream of the two leucine residues was found to be important for rapid internalization. We suggest that it may become phosphorylated after IL-6 stimulation, which in turn leads to an exposure of the internalization signal to the endocytotic machinery.

EXPERIMENTAL PROCEDURES
Materials-Restriction enzymes, calf intestinal phosphatase, polynucleotide kinase, T4-DNA ligase, and protease inhibitors were purchased from Boehringer Mannheim. Tran 35 S-label (44 TBq/mmol) was obtained from ICN (Meckenheim, Germany). DMEM was from Life Technologies, Inc. Recombinant human IL-6 was prepared as described previously (Arcone et al., 1991). The specific activity was 1.5 ϫ 10 6 B cell stimulatory factor-2 units/mg of protein. IL-6 that was used for iodination was kindly provided by Drs. T. Kishimoto and T. Hirano (Osaka, Japan); the specific activity was 5 ϫ 10 6 B cell stimulatory factor-2 units/mg of protein. The IL-6R-cDNA was isolated as described elsewhere (Schooltink et al., 1991). Dr. T. Taga kindly provided the gp130-cDNA. Monoclonal antibody GPX-7 against human gp130 (Saito et al., 1993) and the anti-Tac antibody 7G7B6 (Rubin et al., 1985) were kind gifts of Dr. K. Yasukawa (Kanagawa, Japan) and of Dr. David Nelson. The anti-human CD63 antibody was purchased from Immunotech (Westbrook, ME, USA). Iron-loaded transferrin and the anti human transferrin antibody were purchased from Sigma. Rhodamineconjugated anti-mouse and anti-rabbit antibodies were obtained from Dakopats (Hamburg, Germany). Fluorescein isothiocyanate-conjugated anti-mouse antibody was purchased from Rockland (Gilbertsville, PA).
Expression Vectors-Standard cloning procedures were performed as outlined by Sambrook et al. (1989). The expression vector pCDM8-IL-6R was constructed as described previously .
Truncation Mutants-The gp130 deletion mutants ⌬131 or ⌬133 were constructed by digesting the gp130 cDNA in the pBluescript IIKS(ϩ) vector with AccI/BamHI and ligating it with the fitting sense oligos 5Ј-CTACCCAGCCCTTGTTATAGG-3Ј (⌬131) or 5Ј-CTAC-CCAGCCCTAGG-3Ј (⌬133), each containing an in-frame stop codon. The deleted cDNAs were subcloned into the expression vector pSVL using the XhoI and BamHI sites of the polylinker.
Point Mutants-The point mutants L145A, L146A, L145A/L146A, L145I, L146I, S141A, and S139/141A were constructed by site-directed mutagenesis using different mutagenic primers in connection with the USE mutagenesis kit from Pharmacia Biotech Inc. For selection of mutated clones, the SalI site in the pSVL vector was mutated to an ApaI site. The sequences of the different mutagenic primers are available upon request. The point mutants S137A and S139A were made by the polymerase chain reaction using Vent DNA polymerase containing 3Ј-5Ј proofreading exonuclease activity (New England Biolabs, Inc., Beverly, MA). As a template, the wild-type gp130 cDNA, cloned into the pBluescriptIIKS(ϩ) vector, was used. The sequences of the different mutagenic primers are available upon request. Polymerase chain reaction products were cut with NsiI and AccI, cloned into the 5.8 kilobase AccI-NsiI fragment of pBluescriptIIKS(ϩ)130WT, and confirmed by DNA sequencing. The 3.0-kilobase XhoI-BamHI cDNA containing the polymerase chain reaction product was subsequently cloned into the multiple cloning site of the pSVL vector.
Construction of Tac Chimeras-The construction of two chimeric proteins consisting of the Tac antigen (interleukin-2 receptor ␣-chain) (Leonard et al., 1984) fused to pieces of the cytoplasmic tail of the human insulin receptor was described previously (Renfrew Haft et al., 1994). Tac-EKITLL(IR) encodes the Tac antigen followed by the 6amino acid sequence found in the cytoplasmic tail of the insulin receptor at position (982-987). Using Tac-EKITLL(IR) as a template, doublestranded inserts containing the 6-amino acid wild-type sequence of gp130 (STQPLL) or a mutant sequence (STQPAA), followed by a stop codon, were ligated into the Tac-EKITLL(IR) construct digested with XbaI and EcoRV. The sequences of the sense strands used in the ligations were as follows: 5Ј-CTAGAAGAACAATCTCTACCCAGCCCT-TGTTATAGGAT-3Ј for Tac-STQPLL (gp130) and 5Ј-CTAGAAGAA-CAATCTCTACCCAGCCCGCGGCATAGGAT-3Ј for Tac-STQPAA (gp130). The constructs were verified by dideoxy sequencing.
Transfection-Transfection of cells was carried out using the Gene Pulser from Bio-Rad Laboratories (Mü nchen, Germany). For competition and internalization studies, 2 ϫ 10 6 COS-7 cells in 0.8 ml of DMEM were co-transfected with 1.3 g of pCDM8 vector containing the IL-6R wild-type cDNA plus 28.7 g of pSVL vector containing gp130 wild-type or mutant cDNAs (molar ratio, 1:16). For immunofluorescence studies, 2 ϫ 10 6 COS-7 cells in 0.8 ml of DMEM were each transfected with 10 g of the different Tac constructs. For colocalization studies, 5 ϫ 10 6 HeLa cells in 0.8 ml of DMEM were each transfected with 20 g of the different Tac constructs. All transfections were carried out using a voltage of 230 V and a capacity of 960 F. 2 or 3 days after transfection, cells were used for further studies.
Competition Assay-COS-7 cells were transiently co-transfected with the pCDM8-IL-6R cDNA and with one of the pSVL-derived plasmids carrying the wild-type or mutant gp130 cDNAs. Transfected cells were plated in duplicate wells. 3 days after transfection, cells were washed twice with ice-cold binding medium (DMEM without bicarbonate containing 0.2% (w/v) bovine serum albumin and 20 mM Hepes buffer, pH 7.0). Cells were then incubated with 100 pM 125 I-IL-6 for 4 h at 4°C in the presence of increasing amounts of unlabeled IL-6 (0.1-1000 nM). After 4 h, medium was removed, and cells were washed 3 times with PBS containing 1 mM MgCl 2 , 0.1 mM CaCl 2 , and 0.2% bovine serum albumin. Cells were subsequently solubilized in 1 ml of 1 M NaOH, and radioactivity was measured in a ␥-counter. The binding affinity of IL-6 to the IL-6R complex was calculated by nonlinear regression using the computer program InPlot 4.0 (GraphPad, San Diego, CA).
Internalization Assay for Di-leucine Mutants-Transfected cells were incubated with 1 nM 125 I-IL-6 for 2 h at 4°C in the presence of a 200-fold excess of unlabeled ligand. Internalization was initiated by warming up the cells to 37°C without removing unbound ligand. After different times of incubation, cells were set on ice again for 2 h. Cells were washed 3 times with PBS containing 1 mM MgCl 2 , 0.1 mM CaCl 2 , and 0.2% bovine serum albumin. Surface-bound 125 I-IL-6 was determined after subjecting the cells to 0.5 M NaCl/HCl, pH 1, for 3 min followed by an additional wash with PBS. Internalized 125 I-IL-6 was determined after lysis of the cells in 1 ml of 1 M NaOH.
Immunofluorescence and Confocal Microscopy-Approximately 10 5 COS-7 cells grown on coverslips for 48 h were incubated for 2 h at 4°C with a monoclonal anti-Tac antibody 7G7B6 (1:100 dilution in binding medium). After removing unbound antibody, cells were shifted to 37°C for 30 min. Cells were washed twice with PBS, fixed with 2% paraformaldehyde, and lysed with 0.2% saponin as described previously (Graeve et al., 1989). For colocalization studies, 10 5 HeLa cells grown on coverslips for 48 h were incubated for 2 h at 4°C with a monoclonal anti-Tac antibody 7G7B6 (1:100 dilution in binding medium). After removing unbound antibody, cells were shifted to 37°C for 15 and 90 min, respectively. During the 15-min incubation period, 10 ng/ml iron-loaded transferrin was present in the medium. After fixation, permeabilization, and incubation with a 1/200 dilution of a rhodamine-conjugated anti-mouse antibody, cells were treated for 1 h with an anti-transferrin antibody (1:200) and a lysosomal anti-CD63 antibody (1:100), respectively. Antibodies bound to the specific proteins were detected using a 1/200 dilution of fluorescein isothiocyanate-conjugated anti-rabbit or anti-mouse antibodies. Coverslips were mounted on slides with Mowiol 4 -88 (Calbiochem, La Jolla, CA) and analyzed using confocal immunofluorescence microscopy (Confocal Laser Scan Microscope, Zeiss). 10 -20% of transfected COS-7 cells and 5-10% of transfected HeLa cells stained positive for the expressed proteins. Representative cell were photographed using a 63ϫ oil lens.
Degradation of Wild-type and Mutant gp130 -Transfected cells were grown to confluence in 3.5-cm dishes, metabolically labeled with [ 35 S]methionine/cysteine for 1 h, and chased for the indicated times in normal medium. Cell lysis was performed in 10 mM Tris-HCl, pH 7.4, 66 mM EDTA, 1% Nonidet P-40, and 0.4% sodium deoxycholate in the presence of proteinase inhibitors (2 g/ml aprotinin, 0.5 g/ml leupeptin, 0.7 g/ml pepstatin, 35 g/ml phenylmethyl sulfonic acid). Supernatants and lysates were pretreated with Pansorbin (Calbiochem, La Jolla, CA). SDS was added to a final concentration of 0.3%. After incubation with the monoclonal antibody GPX-7 for 2 h at 4°C, the immune complexes were precipitated with protein A-Sepharose preincubated with cell lysate of unlabeled cells. Proteins were separated on 10% SDS-polyacrylamide gels (Laemmli, 1970) and visualized by fluorography (Chamberlain, 1979). Fluorograms were quantitated with an Hewlett Packard Scanjet IIIc with transparency adaptor using Sigma Scan Image software (version 1.20.09).
Internalization Assay for Serine Mutants-To determine the internalization rate constant for wild-type or mutant receptor complexes, transfected cells were grown to confluency in 24-well plates. The cells were washed twice with binding medium and then incubated at 37°C for 2-12 min in 300 l of binding medium containing 1 nM 125 I-IL-6. Cells were placed on ice and washed 3 times with PBS containing 1 mM MgCl 2 , 0.1 mM CaCl 2 , and 0.2% bovine serum albumin to remove unbound ligand. Surface-bound 125 I-IL-6 was determined after subjecting the cells to 0.5 M NaCl/HCl, pH 1, for 3 min followed by an additional wash with PBS. The residual cell-associated radioactivity (internalized IL-6) was then quantified after dissolving the acid-washed cells in 1 ml of 1 M NaOH. The internalization rate constant for each receptor was determined during a 2-12-min incubation period as described previously (Lund et al., 1990;Backer et al., 1991). Integrals were approximated by the trapezoidal rule, using ⌬t ϭ 2-min intervals, and slopes were determined by linear regression. The slopes were calculated using The sequence is shown in the single-letter code and the mutated residues Ser-137, Ser-139, Ser-141, Leu-145, and Leu-146 in the wild-type sequence are written in boldface. The wild-type and mutated forms of gp130, respectively, were transiently expressed in COS-7 cells together with the wild-type IL-6 receptor (COS-WT, -L145A, -L146A, -L145/146A, -L145I, -L146I, -S137A, -S139A, -S141A, -S139/141A, -⌬131, and -⌬133).
FIG. 2. Internalization of surface-bound 125 I-IL-6 by IL-6 receptor complexes composed of wild-type IL-6 receptor and wild-type (WT) or mutant gp130. Three days after transfection, cells were washed twice with ice-cold binding medium. After incubation with 1 nM of 125 I-IL-6 for 2 h at 4°C, cells were warmed to 37°C for the indicated times without removing unbound ligand. Cells were then set on ice for 2 h and washed 3 times with PBS to eliminate unbound ligand. Surface-bound 125 I-IL-6 (open symbols) was determined after subjecting cells to 0.5 M NaCl/HCl, pH 1, for 3 min. Internalized 125 I-IL-6 (closed symbols) was measured in the ␥-counter after lysis of the cells in 1 M NaOH. Data are expressed as a percentage of the amount of IL-6 bound at time point 0. The sum of surface-bound and internalized IL-6 at later time points adds up to more than 100%, indicating that an intracellular pool of cycling receptors is loaded with IL-6 during the 6-h incubation time. Values are the mean of two to four independent experiments. CO, untransfected cells (triangles). the computer program SAS (version 6.10). The data represent the mean Ϯ the standard deviation of three experiments, each performed in duplicate.

Identification of a Di-leucine Internalization Motif in the
Cytoplasmic Tail of gp130 -Our previous work has revealed that the cytoplasmic 10-amino acid sequence TQPLLDSEER of gp130 is crucial for the efficient internalization of the IL-6 receptor complex . Since this sequence contains a putative di-leucine internalization motif, we constructed the two additional deletion mutants ⌬131 and ⌬133 (Fig. 1). These mutants contain a stop codon directly before and after, respectively, the di-leucine pair. Each of the two mutant cDNAs was transfected together with the wild-type IL-6 receptor cDNA into COS-7 cells, and internalization of 125 I-IL-6 was studied after 72 h. Whereas mutant ⌬131 mediates the endocytosis of 125 I-IL-6 as efficiently as wild-type gp130 (Fig. 2, A  and B), mutant ⌬133 shows a strongly reduced internalization efficiency (Fig. 2C). Within the first 30 min, wild-type gp130 and mutant ⌬131 endocytose 45 and 48%, respectively, of initially bound ligand. In contrast, only 19% of initially bound IL-6 was internalized by mutant ⌬133. Furthermore, during the 6-h incubation period, cell surface receptors were downregulated by at least 50% in wild-type and ⌬131 transfectants but by less than 20% in ⌬133 mutants. In order to exclude that the observed effects are due to a reduced binding affinity of the mutant ⌬133 receptor complex, competition studies were performed with wild-type and ⌬133 transfectants (Fig. 3A). No significant difference was observed. From these results we conclude that the two vicinal leucine residues within the sequence TQPLLDSEER are essential for efficient endocytosis and down-regulation of the IL-6 receptor complex.
To determine if there are additional internalization motifs downstream of the di-leucine motif, we mutated both leucines to alanines in the full-length gp130 protein (L145A/L146A; Fig.  1). When transfected into COS-7 cells, this mutant bound IL-6⅐IL-6R complexes with an affinity comparable with that of wild-type gp130 (Fig. 3B) but internalized only 16% of initially surface-bound ligand within the first 30 min. In addition the number of surface-binding sites was reduced only by 13% within 6 h (Fig. 2D). This finding corroborates the crucial function of leucine residues 145 and 146.
Leucine 145 Is Absolutely Essential for the Function of the Internalization Signal-To analyze if both leucines are equally important for endocytosis and receptor down-regulation, we replaced either leucine 145 or leucine 146 by an alanine or an isoleucine residue (Fig. 1). All mutants bound 125 I-IL-6 with comparable affinity (Fig. 3B). In the internalization assay, only mutant L145A showed a strongly reduced endocytosis (Fig. 4A). Replacement of leucine 145 by an isoleucine caused only a minor effect (Fig. 4B). This observation suggests that in the first position of the di-leucine pair, a hydrophobic amino acid is important. Replacement of the second leucine (Leu-146) by an alanine or an isoleucine led to receptor complexes with an intermediate internalization efficiency (Fig. 4, C and D), indicating that the requirements in this position are less stringent.
The gp130 Di-leucine Motif Mediates Endocytosis and Lysosomal Targeting of the Tac Antigen-To demonstrate that the di-leucine motif of gp130 is not only essential but also sufficient for mediating endocytosis, we transferred the hexapeptide sequence of gp130 (STQPLL) to the cytoplasmic tail of the IL-2 receptor ␣-chain (Tac antigen). Such an approach has recently been used to study the di-leucine motifs in the insulin receptor (Renfrew Haft et al., 1994). As a control we used the hexapeptide STQPAA, in which both leucines were replaced by alanines. We studied the endocytosis of these Tac chimeras using an anti-Tac antibody (7G7B6). Transfected COS-7 cells were incubated with the antibody for 2 h at 4°C. The unbound antibody was washed away, and the temperature was shifted to 37°C for 30 min. After fixation, permeabilization, and incubation with a rhodamine-conjugated secondary antibody, the localization of the Tac antigen was studied using confocal immunofluorescence microscopy. Representative medial optical sections are shown in Fig. 5. Whereas the chimera Tac-STQ-PLL was mainly found intracellulary, the control chimera Tac-STQPAA was localized predominantly at the cell-surface (Fig.  5, A and B). Similar results were obtained with the Tac chimeras containing the juxtamembrane di-leucine motif of the insulin receptor (EKITLL) or the respective control construct (EKI-TAA) (Fig. 5, C and D). In order to identify the intracellular compartments in which the Tac-STQPLL chimera was localized after internalization, a co-staining with two markers, transferrin (endosome) and CD63 (lysosome) was performed (Trowbridge et al., 1993;Nieuwenhuis et al., 1987). Transfected HeLa cells were incubated with the Tac antibody for 2 h at 4°C, the unbound antibody was washed away and the temperature was shifted to 37°C for 15 and 90 min, respectively. Analysis of transfected cells demonstrated that the Tac-STQPLL chimera was localized in transferrin positive endosomal vesicles within 15 min (Fig. 6, A and B) and was transported to a CD63positive compartment (lysosomes) after 90 min (Fig. 6, C and  D). In contrast, the construct Tac-STQPAA remained at the cell surface and did not colocalize with transferrin or CD63 (Fig. 6,  E-H). From these data, we conclude that the di-leucine motif of gp130 not only mediates internalization but also lysosomal targeting. They also demonstrate that the di-leucine motif of gp130 can act as an autonomous internalization signal.
In the Presence of IL-6, the Di-leucine Mutant (L145A/ L146A) Is Degraded More Slowly Than Wild-type gp130 -In order to evaluate the effect of the mutation in the di-leucine motif upon degradation of gp130, a pulse-chase experiment was performed (Fig. 7). 72 h after transfection, cells were labeled for 1 h with [ 35 S]methionine and [ 35 S]cysteine. Cells were then chased in DMEM containing 1 nM IL-6 for 0 -8 h. In stably transfected Madin-Darby canine kidney cells, it has been shown previously that gp130 is first seen as an incompletely glycosylated precursor of ϳ130 kDa that by 2 h of chase is processed to a mature ϳ150 kDa endo-H-resistant form . In COS-7, two polypeptides of ϳ130 and ϳ150 kDa were also found. However, in these cells the 130-kDa precursor was rapidly converted to the mature form (t Ͻ 45 min), and by 8 h of chase, the 150-kDa mature form of gp130 protein was largely degraded. When the turnover times of the mature protein were compared for wild-type and mutant gp130, we found small differences in their half-lives. A half-life FIG. 5. Confocal immunofluorescence microscopy of COS-7 cells expressing Tac-gp130 and Tac-insulin receptor (IR) chimeras, respectively. 48 h after transfection, COS-7 cells were incubated with anti-Tac antibody for 2 h at 4°C. After the prebound antibody was internalized for 30 min at 37°C, cells were fixed, permeabilized, and stained with an rhodamine-conjugated antimouse antibody. Representative medial optical cuts are shown. In each panel, the hexapeptide sequence fused to the Tac antigen is indicated, and the source of the peptide is identified in parentheses. Bar, 25 and 50 m, respectively.

FIG. 4. Internalization of surface bound 125 I-IL-6 by IL-6 receptor complexes composed of wild-type (WT) IL-6 receptor and mutant gp130.
Three days after transfection, cells were washed twice with ice-cold binding medium. After incubation with 1 nM 125 I-IL-6 for 2 h at 4°C, cells were warmed to 37°C for the indicated times without removing unbound ligand. Cells were then placed on ice for 2 h and washed 3 times with PBS to eliminate unbound ligand. Surface bound 125 I-IL-6 (open symbols) was determined after subjecting cells to 0.5 M NaCl/HCl, pH 1, for 3 min. Internalized 125 I-IL-6 (closed symbols) was measured in the ␥-counter after lysis of the cells in 1 M NaOH. Data are expressed as described in Fig. 2. Values are the mean of two to four independent experiments. of 2.5 h was found for wild-type gp130 versus 3.5 h for the di-leucine mutant protein (Fig. 7). This suggests that in the intact gp130 molecule, as with the Tac chimeras, the di-leucine motif may be involved in both ligand-stimulated endocytosis and lysosomal degradation.
Serine 139 Upstream of the Di-leucine Motif Is Important for Rapid Internalization-In the CD3-␥ chain, in CD4, and in the cation-independent mannose 6-phosphate receptor, serine residues upstream of the di-leucine motif were shown to play a role in receptor internalization and lysosomal targeting (Le-tourneur and Klausner, 1992;Shin et al., 1991;Chen et al., 1993). In the case of the CD3-␥ chain and CD4, these serines became phosphorylated after treatment of the cells with phorbol esters, which resulted in receptor down-regulation. These serines reside within a protein kinase C consensus sequence site. In the case of the cation-independent mannose 6-phosphate receptor, such a serine residue is located within a casein kinase II consensus sequence and was found to be important for intracellular targeting of lysosomal enzymes (Chen et al., 1993). Mutation of these serines to alanines or valines strongly reduced endocytosis and lysosomal targeting, respectively.
Upstream of the di-leucine motif of gp130, three serine residues exist, Ser-137, Ser-139, and Ser-141 (Fig. 1). In order to study whether any of these are required for efficient endocytosis, we mutated each of the serine residues to alanine (Fig. 1). Each mutant cDNA was transfected together with the wildtype IL-6 receptor cDNA into COS-7 cells. All mutant receptor complexes were expressed equally and bound IL-6 with a comparable affinity (Fig. 3). Since we observed the strongest effect during the initial internalization, we measured the specific internalization rates (k e ) as outlined in (Lund et al., 1990, Backer et al., 1991. Fig. 8A shows the ratio of intracellular to cell-surface 125 I-IL-6 as a function of time for the mutant and wild-type gp130 proteins. Over a 12-min time frame, in which receptor recycling and degradation is negligible, cells expressing wild-type gp130 internalized IL-6 with a rate constant of k e ϭ 0.100 Ϯ 0.029 min Ϫ1 (Fig. 8B). For mutant S139A we measured a rate constant that was reduced by 50% (k e ϭ 0.050 Ϯ 0.010 min Ϫ1 ). Mutants S137A and S141A showed a somewhat reduced internalization rate, however this effect was not statistically significant (k e ϭ 0.088 Ϯ 0.006 min Ϫ1 ; k e ϭ 0.072 Ϯ 0.028 min Ϫ1 ). The double mutant S139A/S141A underwent an even slower rate of internalization than mutant S139A (k e ϭ 0.037 Ϯ 0.004 min Ϫ1 ). The lowest rate, however, was found with the di-leucine mutant L145A/L146A (k e ϭ 0.016 Ϯ 0.010 min Ϫ1 ). These results suggest that in addition to the di-leucine motif serine 139 is important for rapid endocytosis.

DISCUSSION
In this study, we identified a di-leucine internalization motif (STQPLL, amino acids 141-146) within the cytoplasmic tail of gp130. There is also a putative tyrosine-containing internalization sequence (YSTV) from amino acid 118 to 121 in the intracellular domain. However, this sequence by itself is not sufficient for mediating efficient endocytosis . As with the tyrosine-based motifs in which the tyrosine often can be replaced by a phenylalanine, the leucines can tolerate conservative changes. Thus the second leucine (Leu-146) can be changed to an alanine or to an isoleucine, resulting in a slightly diminished but persistent internalization. The first leucine (Leu-145), however, appears less resilient to change, showing a diminished function when changed to an isoleucine and a complete loss of function when replaced by an alanine.
While the di-leucine motif first manifested itself as an endocytotic signal, we could show that the addition of the hexapeptide STQPLL to the cytoplasmic tail of the Tac antigen enables this molecule not only to become internalized but also to be targeted to the lysosomes. This observation is supported by the results of the site-directed mutagenesis studies in the intact gp130 molecule. In the full-length gp130 the di-leucine motif appears to be essential for both ligand-induced endocytosis and increased receptor degradation. However, the effect of the mutated leucine pair (L145A/L146A) on receptor half-life was less pronounced than on IL-6 internalization and receptor downregulation. One possibility may be that there exist additional signals in the cytoplasmic tail of gp130 that code for the delivery of the signal transducer to the degradative pathway inde- FIG. 7. IL-6-stimulated degradation of wild-type (WT) gp130 and the dileucine mutant (L145A/L146A). Confluent monolayers of COS-7 cells, transfected with wild-type or mutant gp130 together with wild-type IL-6 receptor, were incubated for 1 h in DMEM containing [ 35 S]methionine and [ 35 S]cysteine. Thereafter, the cells were transferred to medium containing 1 nM IL-6 and unlabeled methionine and cysteine for 0 -8 h. The cells were then solubilized, and the wild-type and mutant forms of gp130 were immunoprecipitated with an specific monoclonal antibody (GPX-7). Finally, the immunoprecipitates were analyzed by SDS-PAGE followed by fluorography. The x-ray film was exposed for 2 days. The amount of wild-type and mutant protein present at each time point was quantitated by densitometry as described under "Experimental Procedures." Half-lives were determined by curve fitting using an exponential equation.
FIG. 8. 125I-IL-6 internalization by COS-7 cells expressing wild-type IL-6 receptor plus wild-type (WT) or mutant gp130. COS-7 cells transiently expressing either wild-type or mutant gp130 together with the wild-type IL-6 receptor were exposed to 1 nM 125 I-IL-6 for 2-12 min at 37°C. At the indicated times after IL-6 addition, surface-bound and internalized 125 I-IL-6 were determined as described under "Experimental Procedures." Panel A, the ratio of the intracellular IL-6 to surface-bound IL-6 is plotted as a function of time for COS-7 cells expressing either wild-type or mutant gp130. The results are the means Ϯ the S.E. of three experiments, each performed in duplicate. Panel B, the internalization rate constants (k e ) for the different IL-6 receptor complexes in each cell line were calculated as described under "Experimental Procedures." pendent of ligand binding. This is in agreement with the short half-life of this protein in HepG2, Madin-Darby canine kidney , and COS-7 cells (this study).
By introducing a point mutation at serine residue 139, we demonstrated that this serine is important for rapid internalization of IL-6. Currently we do not know whether this serine is phosphorylated upon IL-6 stimulation. However, we suggest that it may become phosphorylated for three reasons. First, gp130 exhibits an increased phosphorylation on serine residues after IL-6 stimulation (Murakami et al., 1993). Second, the position of serine 139, six amino acids upstream of the dileucine motif, aligns very well with the positions of the phosphoserines identified for the CD3-␥ chain, CD4 and the cationindependent mannose 6-phosphate receptor (Dietrich et al., 1994;Shin et al., 1990Shin et al., , 1991Chen et al., 1993). Third, downstream of serine 139 we find a negative charge at position ϩ1 and two putative negative charges at positions ϩ2 (Ser-141) and ϩ3 (Thr-142). If the latter would be phosphorylated, this would make serine 139 a possible target of casein kinase II (Meggio et al., 1994). However, future studies have to address this question in more detail.
For the CD3-␥ chain and CD4, it has been shown that the PMA-induced down-regulation is impaired after mutation of the identified serines to valine or alanine. In the mannose 6-phosphate receptor system, the transport from the TGN, trans-Golgi network to the lysosomes was impaired in such a serine mutant. Until now it is not known which kinase(s) phosphorylate(s) these serine residues. The serines identified reside either within a consensus sequence for phosphorylation by protein kinase C (CD3-␥, CD4) or by casein kinase II (CD3-␥, CI-M6PR). Interestingly, Hoflack and co-workers detected a casein kinase II-like activity that copurifies with the HAI adaptor complex and is associated to its 47-kDa subunit (Meresse et al., 1990). In vitro this enzyme phosphorylates the cationindependent mannose 6-phosphate receptor cytoplasmic domain on serine 2492 upstream of the di-leucine motif.
The functional role of the ligand-induced serine phosphorylation however, is still speculative. For CD4 it was shown that serine phosphorylation causes dissociation of the protein tyrosine kinase p56 lck from its cytoplasmic domain, thus allowing interaction of CD4 with the endocytotic machinery to occur (Pelchen-Matthews et al., 1992). Probably, ligand-induced serine phosphorylation is analogous to tyrosine autophosphorylation observed for the epidermal growth factor receptor in which this phosphorylation enhances the interaction with the endocytotic machinery, i.e. the plasma membrane associated ␣-adaptin (Sorkin and Carpenter, 1993). For the CD3-␥ chain, it has been suggested that receptor-mediated cell activation leads to an increase in kinase activity and thereby to a serine phosphorylation. This subsequently induces a conformational change that results in an increased accessibility of phosphoserine-dependent di-leucine-based motifs to adaptors and leads to a down-regulation of the receptor (Dietrich et al., 1994). Since the mutation of serine 139 to an alanine does not inhibit IL-6 internalization completely, but reduces the initial internalization rate by 50%, we envision a similar mechanism for the IL-6 induced down-regulation of its receptor complex.