The Ubiquitin Ligase SCF(βTrCP) Regulates the Degradation of the Growth Hormone Receptor*

SCF ubiquitin ligases play a pivotal role in the regulation of cell division and various signal transduction pathways, which in turn are involved in cell growth, survival, and transformation. SCF(TrCP) recognizes the double phosphorylated DSGΦXS destruction motif in β-catenin and IκB. We show that the same ligase drives endocytosis and degradation of the growth hormone receptor (GHR) in a ligand-independent fashion. The F-box protein β-TrCP binds directly and specifically with its WD40 domain to a novel recognition motif, previously designated as the ubiquitin-dependent endocytosis motif. Receptor degradation requires an active neddylation system, implicating ubiquitin ligase activity. GHR-TrCP binding, but not GHR ubiquitination, is necessary for endocytosis. TrCP2 silencing is more effective on GHR degradation and endocytosis than TrCP1, although overexpression of either isoform restores TrCP function in silenced cells. Together, these findings provide direct evidence for a key role of the SCF(TrCP) in the endocytosis and degradation of an important factor in growth, immunity, and life span regulation.

SCF ubiquitin ligases play a pivotal role in the regulation of cell division and various signal transduction pathways, which in turn are involved in cell growth, survival, and transformation. SCF(TrCP) recognizes the double phosphorylated DSG⌽XS destruction motif in ␤-catenin and IB. We show that the same ligase drives endocytosis and degradation of the growth hormone receptor (GHR) in a ligand-independent fashion. The F-box protein ␤-TrCP binds directly and specifically with its WD40 domain to a novel recognition motif, previously designated as the ubiquitin-dependent endocytosis motif. Receptor degradation requires an active neddylation system, implicating ubiquitin ligase activity. GHR-TrCP binding, but not GHR ubiquitination, is necessary for endocytosis. TrCP2 silencing is more effective on GHR degradation and endocytosis than TrCP1, although overexpression of either isoform restores TrCP function in silenced cells. Together, these findings provide direct evidence for a key role of the SCF(TrCP) in the endocytosis and degradation of an important factor in growth, immunity, and life span regulation.
Understanding how metabolism of cells and organisms is regulated and how this connects to the regulation of longevity, cell cycle, apoptosis, and immunity are major challenges of modern biology (1). Growth hormone (GH) 2 and its receptor (GHR) act at these crossroads as becomes apparent in the current study in which we show that a regulator of cell cycle and immunity appears to regulate GH sensitivity of cells. All cells of the human body contain GHRs. Upon GH binding, these receptors initiate signal transduction that results in expression of genes involved in anabolic processes, including increased protein synthesis, lipid degradation, immune function, muscle mass, bone turnover, and tooth development (2). The GHR was the first class one cytokine receptor to be cloned (3). Signal transduction of class-one cytokine receptors recruits members of the Janus kinase family tyrosine kinases that bind to a proline-rich motif in the intracellular membrane-proximal domain, the box 1 motif (4). For the GHR, Jak2 acts as initiator of several signaling cascades, including Stat activation. Box 1-related activities contribute little to GHR endocytosis (5).
For receptor tyrosine kinases, both signal transduction and receptor abundance are regulated by their ligands: epidermal growth factor, c-kit, insulin and Notch binding initiate not only signal transduction, they also induce a swift degradation of their receptors. For cytokine receptors, these two parameters seem to have their own independent regulating mechanism. There is good evidence that the GH sensitivity of cells is mainly regulated via the ubiquitin-dependent endocytosis motif (UbE motif) in the cytokine receptor box 2 region (6). Mutation of the UbE motif (DDSWVEFIELD) leads to a dramatic drop in endocytosis and lysosomal degradation, rendering the cells more GH-sensitive.
The ubiquitin-proteasome system provides the major pathway for nonlysosomal degradation (reviewed in Ref. 7). The F-box protein ␤-transducing repeat-containing protein (␤TrCP) serves as the substrate recognition subunit in the SCF(␤TrCP) ligase (8). Classical examples of substrates are ␤-catenin, NFB, and IB (inhibitor of NFB). Two isoforms exist; ␤TrCP1 is mainly present in the nucleus, whereas ␤TrCP2 resides in the cytosol (9). These ligases play a pivotal role in cell division and various essential signal transduction pathways for tumorigenesis. Their substrate binding motifs consist of seven propeller-shaped WD40 domains that specifically bind short motifs in its central cavity. These motifs generally are 6 amino acid residues long with a common structure of DSGX 2ϩn S. To be recognized by the SCF(␤TrCP), both serine residues need to be phosphorylated. Recently, it became evident that, for a restricted number of plasma membrane proteins, ubiquitination triggers internalization and vacuolar/lysosomal rather than proteasomal degradation (reviewed in Ref. 10). This pathway is best understood in yeast, where a number of plasma membrane proteins are endocytosed in an ubiquitindependent manner (11,12). Several studies have shown that monoubiquitination of plasma membrane proteins is sufficient to stimulate their endocytosis (11,13). In mammalian cells, the GHR, the ␤-adrenergic receptor, and the epithelial sodium channel, ENaC, are examples of membrane proteins that endocytose in an ubiquitin system-dependent manner (14 -16). In some cytokine receptors, the conserved DSGX 2ϩn S motif serves as a degradation signal via the SCF(␤TrCP) ligases. Both for the interferon-␥ and the PrL receptor, it was shown that this E3 is involved in their degradation. An important finding in these studies was that receptor degradation mainly depends on ligand binding; in addition, phosphorylation at both serine residues of the degron was required (17)(18)(19).
In the present study, we show that the UbE motif, deviant from the canonical ␤TrCP recognition motif, is specifically recognized by a functional SCF(␤TrCP) complex and that this interaction is instrumental in the endocytosis and degradation of the GHR. Ubiquitination of the GHR is not required, since ␤TrCP silencing slows degradation of the GHR without lysine residues in the cytosolic tail as efficiently as wild type GHR.
Cell Culture-Hek293 and ts41 cells were grown in Dulbecco's modified Eagle's medium (Invitrogen) supplemented with 10% fetal calf serum, 100 units/ml penicillin, and 0.1 mg/ml streptomycin. Hek293 cells stably expressing the wild type GHR (Hek-wtGHR) were grown in the same medium supplemented with 0.6 mg/ml Geneticin (G418; Invitrogen). All cells used were grown at 37°C with 5.0% CO 2 . Twice a week, cells were washed with phosphate-buffered saline (PBS), detached from the flask with trypsin-EDTA (Invitrogen), diluted in fresh growth medium, and split into new culture flasks.
Transfections-Cells were transfected using Lipofectamine 2000 (Invitrogen). Seventy percent confluent cultures were transfected with 0.2-0.9 g of DNA in 12-well plates according to the manufacturer's protocol. Transfected cells were used for Western blot experiments and 125 I-GH binding experiments 1-2 days after transfection. For fluorescence microscopy cells were transfected with Fugene-6 (Roche Applied Science).
To silence the expression of both ␤TrCPs, Hek-wtGHR cells were transfected with small interfering RNA (siRNA) using Lipofectamine 2000 according to the description of the manufacturer; 3 days after transfection, cells were used for Western blot and 125 I-GH binding experiments. Control cells were transfected with siRNA-GFP. To control the effect of the gene expression silencing, Hek293 cells were transiently transfected with a pcDNA3-FLAG-␤TrCP. Western blots were analyzed using an Odyssey infrared imaging system (Li-Cor Biosciences, Lincoln, NE). 125 I-GH Binding and Internalization-125 I-Human GH was prepared using chloramine T (20). For internalization experiments, cells, grown in poly-D-lysine-coated 12-well plates, were washed with 1 ml of minimal essential medium supplemented with 20 mM Hepes, pH 7.4, and 0.1% bovine serum albumin, and incubated at 37°C in a water bath for 1 h. After washing the cells three times with ice-cold PBS-complete (PBS-c: PBS with 0.135 g/liter CaCl 2 and 0.1 g/liter MgCl 2 ), 0.75 ml 125 I-GH (180 ng/ml in Mem/Hepes/0.1% bovine serum albumin) was added to the cells. 125 I-GH was bound for 2 h on ice. Unbound 125 I-GH was aspirated and the cells were washed three times with PBS-c. Cells were incubated 10 min at 37°C in minimal essential medium, Hepes, 0.1% bovine serum albumin to allow receptor internalization and washed three times with ice-cold PBS-c. Membrane associated 125 I-GH was removed by treating the cells twice with 750 l acid wash (0.15 M NaCl, 50 mM glycine, 0.1% bovine serum albumin, pH 2.5) for 5 min on ice. Acid wash was collected in counting tubes, and radioactivity was measured using a LKB ␥ counter. Cells were solubilized overnight in 1 N NaOH. Internalized 125 I-GH was determined by measuring the radioactivity in the collected NaOH. Unspecific counts were determined by incubating the cells with 125 I-GH together with excess unlabeled GH (9 g/ml). Internalization is expressed as a percentage of the total specific radioactivity after 10 min at 37°C.
In Vitro Binding Assay-Wild-type and mutant GHR proteins were expressed in Hek293, and the cells were lysed in 50 mM Tris-HCl (pH 7.5), 0.15 or 1.0 M NaCl, 50 mM NaF, and 0.5% Nonidet P-40. The receptors were purified with biotinylated GH and streptavidin beads and stringently washed with lysis buffer. The beads were incubated with in vitro translated and 35 S-labeled ␤TrCP2 for 60 min at 4°C and extensively washed with lysis buffer, and associated proteins were analyzed by SDS-PAGE and autoradiography. cDNA of ␤TrCP2 constructs in pcDNA3.1/FLAG was used in an in vitro transcription translation system supplemented with [ 35 S]methionine and TNT-T7 coupled reticulocyte lysate ready reaction mix according to the instructions of the manufacturer (Promega). Binding to GHR-(1-286) was taken as background. To analyze GHR-containing protein complexes at the cell surface, the cells were incubated with biotinylated GH on ice. To recover all GHR-protein complexes from cells, the cells were first lysed and then incubated with biotinylated GH. Equal amounts of cell lysates were incubated for 1 h at 4°C with equal amounts of GST-GHR-(271-334) or GST-GHR-(271-286) bound to GSH beads. Beads were pelleted in an Eppendorf centrifuge and washed three times with 1 ml of lysis buffer and twice with PBS containing protease inhibitors. The beads were boiled in 2ϫ sample buffer. Samples were subjected to SDS-PAGE. Proteins associated with the GST fusion proteins were visualized by autoradiography. To analyze equal loading of the fusion proteins, the gels were stained with Coomassie Brillant Blue and analyzed using an Odyssey imaging system.
Production of GST-GHR Fusion Proteins-pGEX-GHR plasmids were used to express GST-GHR fusion proteins in E. coli (strain BL21) (21). The synthesis of recombinant proteins was induced by isopropyl-1-thio-␤-D-galactopyranoside. GST fusion proteins were purified with GSH-beads (Amersham Biosciences) by the procedure recommended by the manufacturer.
Immune Fluorescence Microscopy-Cy3-GH was prepared using a Fluorolink-Cy3 label kit according to the supplier's instructions (Amersham Biosciences). Transfected cells, grown on coverslips, were incubated for 15 min with Cy3-GH (1 g/ml) and/or Alexa488-labeled transferrin (5 g/ml). Cells were washed with PBS to remove unbound label and fixed for 2 h in 3% paraformaldehyde in phosphate-buffered saline. After fixation, the cells were embedded in Mowiol, and con-FIGURE 1. Effects of TrCP on GHR endocytosis and abundance. A, GHR-expressing Hek293 cells were transfected with siRNA to silence ␤TrCP or clathrin heavy chain (CHC); siRNA for green fluorescent protein (GFP) was applied as a control. 125 I-GH was bound to the cells on ice, the cells were incubated for 10 min at 37°C, and GH internalization was expressed as a percentage of total radioactivity (cell surface and intracellular). B, cell extracts from parallel experiments as in A were analyzed by quantitative Western blotting according to the Odyssey protocol. The efficiency of ␤TrCP siRNA oligonucleotides was measured on cells co-transfected with FLAG-TrCP2. C, GHR-expressing Hek293 cells, cultured on coverslips, were either transfected with ␤TrCP siRNA or with control (GFP) siRNA oligonucleotides, incubated with Alexa488-labeled transferrin (Alexa488-Tf) and Cy3-labeled GH (Cy3-GH) for 15 min, and fixed with formaldehyde. The confocal images show that GH and transferrin co-localized both at the cell surface and in endosomes (left). If TrCP siRNA oligonucleotides were present, Cy3-GH (red) stained mainly the cell surface, whereas the transferrin (green) distribution remained unaffected. Note that internalized GH only partially co-localized with transferrin, indicating sorting between the recycling transferrin and GH. D, Hek293 cells were co-transfected with GFP-DNA and different GHR constructs and transfected with siRNA oligonucleotides as indicated. The cells were incubated with biotinylated GH for 15 min, and GHRs were isolated on streptavidin beads and analyzed for cell surface GHR according to the Odyssey protocol (top). Equal aliquots from the cell lysates were also analyzed on Western blots for ␤TrCP1 (middle) and GFP (bottom). Control siRNA (c) was HPV E6/E7. Data are representative of three experiments. wt, wild type. focal laser-scanning microscopy was performed using a Leica TCS 4D system.

Expression of ␤TrCP Is Required for GHR Endocytosis-En-
docytosis of the GHR depends on the ubiquitination machinery, implicating the involvement of an ubiquitin ligase. Since recent studies show that ␤TrCP is involved in the degradation of certain cytokine receptors, we used small interfering RNA to silence both ␤TrCP isoforms and study the effects on GHR turnover. Fig. 1A shows that siRNA for ␤TrCP inhibits GH uptake to the same extent as silencing of clathrin heavy chain: ϳ50% as compared with nonrelevant siRNA(GFP). In this experiment, the clathrin heavy chain Western blot signal was reduced to 4% compared with siRNA(GFP)-treated cells. Expression of cotransfected FLAG-tagged ␤TrCP2 was reduced to 10% using the "combi" silencing oligonucleotides that silence both TrCP1 and TrCP2 (Fig. 1B). We conclude that ␤TrCP is involved in endocytosis of the GHR.
At steady state, almost all (mature, 130-kDa) GHRs are present at the cell surface. To visualize the effect of ␤TrCP silencing on endocytosis, we incubated the cells with both Cy3-labeled GH and Alexa-488-labeled transferrin. Fig. 1C shows that silencing both ␤TrCP isoforms resulted in a strong increase of Cy3-GH at the plasma membrane, whereas transferrin recycling remained intact, excluding a general effect on clathrinmediated endocytosis. These results demonstrate that ␤TrCP is an obligatory part of the GHR endocytosis machinery. Since both GH and transferrin are endocytosed via clathrin-coated vesicles (24), it suggests that ␤TrCP is part of the cargo selection complex for the GHR.
An obvious consequence of endocytosis inhibition is an accumulation of GHRs at the cell surface. In Fig. 1D, lanes 1 and 2, we used Hek293 cells transfected with GHR DNA and silenced for both ␤TrCP isoforms. To measure the amount of GHRs at the cell surface, the cells were incubated with biotin-GH, and the GH-GHR complexes were isolated and quantitated. siRNA(␤TrCP) treatment caused a 2-3-fold increase in GHR at the cell surface. To ascertain that the increase was not due to a silencing effect on the amount GHRs synthesized, metabolic labeling with [ 35 S]methionine was performed and showed that the GHR synthesis was not changed (not shown). The cellular concentration of ␤TrCP1 was measured on Western blots (Fig. 1D). By this method, using an optimized protocol for ␤TrCP, we obtained ϳ90% silencing. Using this silencing oligonucleotide, ␤TrCP2 was silenced to the same extent (Fig.  1B). Previous studies have shown that by inhibiting GHR endocytosis through (mutation of) the UbE motif, maximally 60% inhibition of GHR endocytosis was achieved compared with removal of the GHR tail, including the complete UbE motif. Apparently, both ␤TrCP and clathrin silencing were insufficient to overcome this half-maximal effect on GHR turnover. Combining the two siRNAs did not result in more endocytosis inhibition. In agreement with our previous experience, these effects are too limited to measure half-life differences in dynamic experiments successfully (e.g. using pulse-chase protocols) (21).
As an important control, based on our previous observations that the GHR itself is not an obligatory ubiquitination target, we used a GHR truncated at amino acid residue 399, in which all 10 lysines are replaced by arginine residues (GHR-(1-399) K271-362R) (22) . Fig. 1D, lanes 3-6, shows that TrCP silencing induces a 2-3-fold increase in cell surface GHRs whether or not lysine residues are present in the receptor tail. To control the experiments of Fig. 1D for equal transfection efficiencies, we co-transfected GFP DNA in all situations and analyzed the cell lysates for GFP; under all (silencing) conditions and various DNA constructs, the transfection efficiency was identical within measurement limits (Fig. 1D, bottom).
Since we previously concluded that GHR internalization requires the recruitment of the ubiquitin conjugation system to the GHR UbE motif rather than the conjugation of ubiquitin to the GHR itself, we now conclude that ␤TrCP is a prime candidate to act as E3.
Other Factors of the ␤TrCP-GHR Complex-Once we established that ␤TrCP is involved in GHR endocytosis and degradation, we asked whether ␤TrCP is in a protein complex with the GHR. Pull-down experiments with biotin-GH, depicted in Fig. 2A, show that both co-expressed full-length (lane 4) and mutant TrCP without the F-box domain (lane 5) were present in the isolated complexes. If His 6 -TrCP1 was used to isolate the protein complex (Fig. 2B), both ER and cell surface GHR were pulled down (lane 4). The interaction occurs intracellularly, as shown in an experiment in which GHR and His 6 -TrCP1 were expressed in different plates (lanes 1 and 5) and mixed after cell lysis (lane 3). Under these conditions (low cellular concentrations, low temperature, and short time periods) ␤TrCP does not interact with GHR. Thus, TrCP is in a protein complex with GHR, probably already in the ER.
␤TrCP generally acts as subunit of a SCF ubiquitin ligase. To examine the functional complex in GHR endocytosis, we cotransfected ␤TrCP2 and Skp1 together with the GHR. Pulldown experiments with biotinylated GH and Western blot analysis of the isolated complex revealed that Skp1 interacted with GHR only in the presence of ␤TrCP (Fig. 2A, lane 4). In the absence of ␤TrCP (lane 6) or in the presence of ␤TrCP lacking the F-box (lane 5), no Skp1 was isolated. In lanes 10 and 11, we repeated the experiment and asked whether Skp1 might have associated with the GHR protein complex after lysis. As for ␤TrCP, the interaction does not occur if cell lysates from cells, expressing the components separately, were mixed (M). Together, this implies that GH, GHR, TrCP2, and Skp1 are in the same protein complex.
To analyze the involvement of an active SCF complex, we took a genetic approach and expressed GHR in hamster ts41 cells. These cells contain a temperature-sensitive mutation in the NEDD8 activation enzyme, APP-BP1, and fail to degrade substrates via E3s with Cullin family members (e.g. p27 via the SCF(skp2) (25), due to defective neddylation. Thus, the E3-SCF(TrCP) is itself regulated via covalent modification of the ubiquitin-like protein Nedd8 by UBC12, whereas deneddylation is performed by the signalosome complex. This modification stabilizes the E2-Cullin/Rbx1 interaction and in turn stimulates ubiquitination of the target substrates. Fig. 2C shows that Cy3-GH is endocytosed at the permissive temperature but largely remains at the cell surface at the nonpermissive temperature (39°C). In the same experiment, Alexa488-labeled transferrin entered the cells undisturbed, although there seemed to be a different distribution intracellularly. This shows that an intact neddylation system is specifically required for uptake of the GHR. Together with the results on ␤TrCP and Skp1, these data strongly suggest that GHR endocytosis requires a fully functional SCF(␤TrCP) complex.
␤TrCP Interacts with the GHR Tail via the UbE Motif-To further examine the role of ␤TrCP in GHR endocytosis, we analyzed ␤TrCP2, since it is dominantly active in the cytosol (26). Hek293 cells were co-transfected with FLAG-tagged ␤TrCP2 together with GHR or the truncation GHR-(1-286) (Fig. 3A). GHR-(1-286) still contains box 1, the attachment site for Jak2, but lacks the UbE motif. This receptor is unable to endocytose and, therefore, accumulates at the cell surface, resulting in a high amount of mature receptor (55 kDa) compared with its 40-kDa ER form (Fig. 3B, left, lane 2). The wildtype GHR migrates at 130 kDa, whereas its precursor ER form appears as a doublet at 110 kDa (lane 1). Cell lysates were incubated with biotinylated GH, and GHR protein complexes were isolated with streptavidin beads. Comparing the full-length GHR with GHR truncated at 286 shows that co-transfected FLAG-labeled ␤TrCP2 was only isolated if full-length GHR was present (Fig. 3B, left, lanes 3 and 4). In Fig. 3B, right, we used various GHR truncations and mutations to pinpoint the tail segment involved in ␤TrCP binding. The receptors were purified from transiently transfected Hek293 cells, via biotinylated GH and streptavidin beads, and then incubated on ice with equal amounts of in vitro synthesized 35

. SCF(TrCP) is involved in the regulation of GHR abundance.
A, Hek293 cells were co-transfected as indicated and lysed in 0.6 ml. 0.55 ml of the cell extracts were incubated with biotinylated GH, and the protein complexes were isolated with streptavidin beads and analyzed on blots using anti-GHR (mAb 5) or anti-FLAG. For loading control, 15 l of cell lysate were used. For lane 10 ("mix" experiment), GHR/␤TrCP DNAs were co-transfected in one plate of cells and GHR/Skp1 DNA in a second. Both plates were lysed in half the volume of the other samples and mixed together before the addition of biotin-GH (M). ER, GHR species in the endoplasmic reticulum; PM, the mature GHR at the plasma membrane. Data are representative of three experiments. B, the lysate of Hek293 cells, co-transfected with GHR and His 6 -tagged ␤TrCP, was used to pull down protein complexes by Ni 2ϩ -nitrilotriacetic acid beads. For the "mix" experiment (M), GHR and His 6 -␤TrCP were transfected in separate dishes (lanes 1 and 5), and the cell lysates were mixed as in the "mix" experiment of A (lane 3). Lanes 1 and 2 are loading controls for GHR, and lanes 5 and 6 are loading controls for TrCP. Data are representative of two experiments. C, Ts-41 cells were transfected with the GHR construct and incubated at 30 or 39°C overnight. Cy3-GH and Alexa488-transferrin were added for 15 min, and the cells were fixed. Data are representative of two experiments. WB, Western blot.  (1-286)). The cells were lysed, and protein complexes were isolated with biotin-GH/streptavidin beads (lanes 3 and 4) and analyzed on blot using anti-FLAG. Lanes 1 and 2, loading control GHR (mAb 5); lanes 5 and 6, loading control ␤TrCP (arrow). B (right), 10 6-cm plates of Hek293 cells were transfected with equal amounts of DNA expressing 10 different GHR constructs as indicated. The cell lysates were incubated with biotin-GH, the receptors were isolated on streptavidin beads, and the beads were washed in 0.5 M NaCl in lysis buffer; 10% of the beads were boiled in SDS-sample buffer, and the samples were detected for GHR on Western blot (GHR, top); the remainder of the beads were incubated for 60 min on ice with equal amounts of in vitro 35 S-labeled ␤TrCP2 and washed twice, and, after SDS-PAGE, the radioactivity was visualized using an Amersham Biosciences PhosphorImager. The amounts of radioactivity were determined using ImageQuant. In the left lane, 16% of the input amount of radioactive ␤TrCP was analyzed. Data are representative of three experiments. WB, Western blot.
Similar UbE Motif Specificity for GHR-␤TrCP 2 Interaction and GHR Endocytosis-Previously, we have shown that endocytosis of the GHR depends both on a functional ubiquitination system and on an intact UbE motif (20,27,28). Using mutational analysis of the UbE motif by changing every single amino acid residue into alanine, the contribution of each amino acid was determined. The experiment identified the key residues in the UbE motif (DDSWVEFIELD) for endocytosis. As shown in Fig. 1, ␤TrCP is an obvious factor to act as a regulator of GHR endocytosis. Therefore, we used the pull-down assay of Fig. 3B (right) to ask whether ␤TrCP binds to the GHR in a specific fashion. Each of the amino acids 320 -334 were changed in alanine residues, and the 15 GHR constructs were tested in Hek293 cells. The cells were transfected with equal amounts of the different constructs and showed comparable GHR expression judged by the amounts of high mannose (ER-localized) GHR (not shown). Western blot analysis of the cell extracts shows that the mutations E326A, F327A, I328A, and D331A caused a significant increase in mature (130-kDa) GHR (Fig.  4A), indicative of impaired degradation. Biotin-GH was used to isolate the GH-GHR complexes from the cell surface, and the GHRs were incubated with equal amounts of 35 S-labeled ␤TrCP. Most mutations resulted in inhibition of ␤TrCP binding as compared with wild-type GHR. However, mutation of the same 4 amino acids inhibited the ␤TrCP binding more than 60% (Fig. 4B). Compared with the mutational analysis previously performed for endocytosis (see Fig. 4C and Fig. 4 in Ref. 22), we note a striking similarity in patterns; the same 4 amino acid are important for endocytosis of the GHR (27).
GHR Binds to the WD-40 Domain of ␤TrCP-␤TrCP canonically binds to the DSG⌽XS motif in IB, ␤-catenin, and human immunodeficiency virus vpu after both serine residues are phosphorylated. To characterize the ␤TrCP-GHR binding further, we used bacterially produced GST-GHR fusion proteins containing only the 65 membrane-proximal amino acid tail residues and in vitro synthesized 35 S-␤TrCP (Fig. 5A) (21). Fig. 5B (left, lanes 2) shows that both ␤TrCP1 and ␤TrCP2 bind with high efficiency to the GST-GHR-(270 -334). If a slightly shorter fusion protein was used (residues 270 -318), the interaction was virtually abolished (lanes 3). This demonstrates an efficient interaction between ␤TrCP and the UbE motif in the GHR. Since the binding is highly efficient at 0°C, the interaction is very likely direct and does not require any other post-translational modification. The next question is whether GHR binds to ␤TrCP via its WD40 domain, as has been established for other E3 substrates. Four different constructs were made, expressing ␤TrCP2 segments 1-174, 1-228, 174 -542, and 228 -542 (Fig.  5A). Using the GST-GHR fusion proteins, in vitro binding assays clearly show that the interaction is via the WD40 domain; the F-box is not necessary for interaction (right). In addition, specific binding was only observed if the UbE motif was present (compare lanes 2 and 3).
Together, the experiments show that ␤TrCP interacts with the GHR and that the F-box protein is required for GHR endocytosis. From comparison between the mutational analyses previously measured for GHR endocytosis and the results obtained with the same constructs for GHR-␤TrCP interaction, we conclude that ␤TrCP is a prime regulator of GHR endocytosis via the UbE motif.
␤TrCP Isoforms-Mainly due to lack of specific antibodies, until now the role of ␤TrCP has not been elucidated at the level of isoform specificity. The main differences between the isoforms are located at the N terminus. ␤TrCP1, in general acting on substrates involved in cell cycle regulation, is mainly present in the nucleus, whereas ␤TrCP2, presumably acting on membrane-bound substrates, resides in the cytosol (9,29). Our experiments show that both isoforms can be detected in pulldown experiments from bacterially produced GST-GHR fusion proteins (Fig. 5B). However, only functional experiments can validate the interaction. Therefore, we determined the effect of  3B (right). As controls, DNA encoding wild-type GHR and GHR-(1-286) were used. After 24 h, the cells were incubated with biotin-GH on ice and lysed, and the cell surface receptors were isolated on streptavidin beads; 10% of the beads were boiled in SDS-sample buffer, and the samples were detected for GHR on Western blot (GHR). The remainder of the beads were incubated for 60 min on ice with in vitro 35 S-labeled TrCP2. The radioactivity was visualized in A (bottom) and quantitated in B; the amount of label co-purified with the short GHR-(1-286) was deduced, and the results were expressed as percentages of radioactivity bound to wild type GHR. Data are representative of two experiments. C, previously published data from Fig. 4 in Ref. 28. Chinese hamster Ts20 cell lines expressing wild-type and mutant GHR as indicated were analyzed for rates of 125 I-GH endocytosis. Internalization is expressed as the ratio of radioactive GH present inside the cells versus at the cell surface.
isoform-specific ␤TrCP silencing on GH internalization. We used several (combinations of) silencing oligonucleotides to specifically deplete transiently expressed FLAG-tagged ␤TrCP isoforms. Fig. 6A (bottom) shows that the siRNA(TrCP) (combi) probe silenced both isoforms, whereas siRNA(TrCP-1) only silenced TrCP1 and not TrCP2. siRNA(TrCP2) silenced only TrCP2 and not TrCP1, whereas the combination of both TrCPs (TrCP1 ϩ 2) silenced both isoforms. From the upper panel of Fig. 6A, we conclude that silencing of ␤TrCP1 alone does not affect GH internalization, whereas in every combination in which ␤TrCP2 was silenced, a clear effect on GH uptake was visible, comparable with silencing of the clathrin heavy chain.
In Fig. 6B, we asked whether replenishment of either of the isoforms is capable of restoring the silencing phenotype. The lower panel shows that silencing is specific; the combi siRNA probe silences both isoforms (lanes 3), except if a ␤TrCP siRNA-resistant replacement vector was used (lanes 4). If ␤TrCP1or ␤TrCP2-specific silencing probes were used, as expected, the "resistant" ␤TrCP DNA could be silenced (lanes [5][6][7][8]. Using this strategy, the GH internalization assay shows that (over)expression of both resistant TrCP isoforms can rescue the silencing effect, indicating that both ␤TrCP isoforms are capable of GHR cargo selection at the cell surface. Our conclusion is that, although both ␤TrCP isoforms are capable of acting in GHR endocytosis, ␤TrCP2 is the most obvious candidate for this role in endogenous situations.

DISCUSSION
Previously, we have shown that GHR endocytosis and subsequent degradation in the lysosomes depends on the activity of the ubiquitin system via the UbE motif. Here, we identify ␤TrCP as a key factor required for GHR endocytosis. ␤TrCP specifically binds to the GHR motif, which is required for ubiquitination-dependent endocytosis. These data support the conclusion that this factor recruits the ubiquitination machinery. The results with Skp1 and neddylation strengthen our hypothesis that the complete SCF(␤TrCP) complex, including Skp1, Cul1, Rbx1, and an ubiquitin conjugase, is present at the GHR tail. This is the first observation that shows direct involvement of a SCF E3 in endocytosis of cytokine receptors without targeting the receptor itself for ubiquitination. Although the test system uses GHR-(over)expressing cells, the experiments show that endogenous TrCP levels are sufficient to rapidly initiate GHR endocytosis and degradation. The  . Differential effects of ␤TrCP isoforms. A, GHR-expressing Hek293 cells were transfected with siRNA as indicated to silence the isoforms of ␤TrCP or clathrin heavy chain (CHC); siRNA for GFP was applied as a control. Cells were incubated with 125 I-GH for 15 min at 37°C, and GH internalization was expressed as the ratio between intracellular and extracellular radioactivity. In the lower panel, cells were transfected with either FLAG-tagged TrCP1 or TrCP2 (Fl-TrCP), together with the siRNAs. As negative control, GFP siRNAs were applied. siRNAs designated TrCP represent the combi probe, silencing both isoforms. Data are representative of four experiments. B, GHR-expressing Hek293 cells were transfected with siRNA-resistant TrCP constructs, designated R-TrCP, to rescue the silencing. Cells were incubated with 125 I-GH for 15 min at 37°C, and the ratios inside/out were calculated as in A. To compare the effect of the TrCP silencing, clathrin silencing was applied. In the lower panel, the actions of the different silencing experiments were controlled with vectors expressing FLAG-tagged TrCP (W) or siRNA-resistant TrCP (R) containing three silent mutations that prevented silencing by the combi probe (TrCP) but not by the specific TrCP1 or TrCP2 oligonucleotides. The blots were immunostained with anti-FLAG antibodies. Data are representative of two experiments.
physiological relevance of the findings is further supported by the TrCP silencing experiments; if the cellular levels of both TrCP species were diminished, uptake and degradation of GHR were inhibited.
Detailed structural information is available on the interaction between the WD40 domain of SCF(␤TrCP) and ␤-catenin (30). The F box protein ␤TrCP recognizes the dual phosphorylated DpSGFXpS destruction motif (where pS represents phosphoserine), present in ␤-catenin, and directs the SCF(␤TrCP) E3 to ubiquitinate it at a lysine residue, 13 amino acids upstream of the destruction motif (commonly present 9 -14 positions upstream). The ␤-catenin peptide binds the top face of the ␤-propeller, with the 6-residue destruction motif dipping into the central channel. All seven WD40 repeats contact ␤-catenin. Comparison with the UbE motif reveals little resemblance to the catenin destruction motif: the UbE motif is 11 amino acids long, it contains 3 acidic residues critical for binding (Asp 321 , Glu 326 , and Asp 331 ) instead of the two phosphorylated serine residues in ␤-catenin, and it lacks the canonical glycine residue, present in all phosphorylated DSGFXS destruction motifs described until now (31,32). The results of Fig. 4 are most predictive and point to a binding site with EFIXXD as the strongest interacting residues. Another deviation from the canonical motif is the lack of a lysine residue at positions 9 -14 upstream, although there is a conserved lysine at 315 immediately after a tyrosine that is phosphorylated upon GH binding. However, mutation of this lysine residue into arginine does not affect its ubiquitination-dependent endocytosis (22). Together, these results are in line with our previous observation that ubiquitination of the GHR is not required for endocytosis. Thus, binding of a presumed SCF(␤TrCP) E3 does not target the GHR for ubiquitination but must be directed toward another factor in the GHR-ubiquitination complex (22). Even without a threedimensional structure of the precise interaction between the ␤TrCP-WD40 domain and the UbE motif, it is already clear that the UbE-WD40 interaction considerably differs from the WD40-DSGFXS destruction motif interactions described until now.
Interaction of ␤TrCP is the first step toward factual endocytosis of the GHR via a clathrin-mediated selection process. To involve the ubiquitination machinery in this process, the GHR must be dimerized (33). Recent observations on ␤TrCP show that it indeed contains a dimerization domain that is important for its E3 function (34). Also, cullins seem to occur as dimers in the context of SCF complexes (35). Our in vitro binding experiments with GST-GHR tails suggest that also monomeric GHR can bind ␤TrCP, although due to the tendency of GST to homodimerize, it may well be that also the GST-GHR molecules represent dimeric GHR segments (36). Previously, we reported that a chimeric GHR consisting of the extracellular domain of low density lipoprotein receptor-like protein and the trans-membrane and cytosolic domain of the GHR could not dimerize anymore and that monomeric GHR can endocytose in an ubiquitin system-independent fashion. Since dimerization occurs in the ER and ␤TrCP is a homodimer, it is possible that ␤TrCP acts as a chaperone at the ER. Therefore, it remains to be investigated whether the dimerization of ␤TrCP poses the necessity for dimerization on the GHR.
Cargo selection via ␤TrCP thus requires an active E3, followed by proteasomal activity, and interaction with the clathrin-mediated machinery (22). Whether the SCF E3 ubiquitinates a factor X and enables GHR to connect to the clathrin heavy chain remains to be determined. Previously, we have identified several factors that bound specifically to the UbE motif (21). The tetratricopeptide repeat-containing protein SGT may be a factor bridging the gap between ␤TrCP and clathrin via the U-box ligase carboxyl terminus of Hsc70-interacting protein (21,37). Another scenario may be that the ubiquitin moiety present in the SCF(␤TrCP) complex serves as a ligand for ubiquitin-binding proteins that are part of the clathrin-mediated endocytosis machinery (38).
␤TrCP is involved in the degradation of prolactin and interferon receptors (17)(18)(19). Fuchs et al. (31) identified ␤TrCP interaction with these receptors via their DSGFXS destruction motifs and showed that the E3 ubiquitinates both the interferon-␥ and prolactin receptors, dependent on an intact DSGFXS motif. Upon ligand binding, this motif is phosphorylated, after which the SCF E3 ubiquitinates certain lysines in the tail of the receptors. This leads to their destruction. It is clear that this scenario differs from the action of ␤TrCP in GHR endocytosis. Although the GHR does contain a DSG⌽XS-like destruction motif (DSGRTS), mutation of the serine residues did not result in a shorter half-life or in inhibition of GHR endocytosis 3 ; nor did mutation/removal of the motif interfere with the ␤TrCP-GHR interaction.
Ubiquitination by the SCF(␤TrCP) E3 of a factor generally results in its destruction by the proteasome. In most cases, the decision to degrade the target is not in the SCF(␤TrCP) E3 action but depends on the activity of one or two kinases that phosphorylate serine residues in the degradation motif. Our experiments with GST-GHR fusion proteins indicate that binding of ␤TrCP to the UbE motif does not depend on a preceding action of another post-translation modification. Thus, very likely, the cytosolic concentration of ␤TrCP determines its activity level to induce GHR endocytosis. This is reflected in several experiments showing that decreased amounts of ␤TrCP inhibit GH uptake and GHR degradation. Thus, the free cytosolic concentration of ␤TrCP may function as a factor in GH sensitivity of cells; at low concentrations, the cells are more sensitive to GH than at high concentrations. Currently, ␤TrCP appears to be involved in the degradation of at least 10 substrates at the same time: ␤-catenin, NFB, IB (31), Emi1 (39), Wee1 (40), ATF4 (41), hDlg (42), CDC25B (32), IFN-R1 (17), Gli3 (43), and Prl-R (21). In all except for CDC25B, degradation is regulated by phosphorylation. Although the two ␤TrCP subspecies may be differently localized in the cell, they can take over their tasks if one of the two is silenced by RNA interference techniques. Since probably no modification is required for ␤TrCP to bind GHR, the two proteins may regulate each other's activity; more cellular ␤TrCP results in increased GHR endocytosis/degradation, whereas more GHR may absorb free ␤TrCP, immediately affecting the other ␤TrCP functions. Evidence for this comes also from the observation that heteroge-neous nuclear ribonucleoprotein-U acts as a pseudosubstrate (44). Generally, ␤TrCP is part of an E3 complex involved in degradation of key factors controlling cell growth, survival, and transformation (31). Since it is a short lived protein itself, more studies are needed to understand the consequences of ␤TrCP concentrations for cellular regulation and the GHR activity in particular.
The GH-IGF1 system is involved in longevity, cell cycle, apoptosis, and immunity (1); in catabolic (cachectic) conditions, such as old age and cancer, unknown stress signals turn the system off, resulting in GH insensitivity, whereas anabolic conditions require cells to be highly GH-responsive. Whether and how ␤TrCP is at this balance remains to be elucidated.