Growth Hormone Receptor Is a Target for Presenilin-dependent γ-Secretase Cleavage*

Growth hormone receptor (GHR) is a cytokine receptor superfamily member that binds growth hormone (GH) via its extracellular domain and signals via interaction of its cytoplasmic domain with JAK2 and other signaling molecules. GHR is a target for inducible metalloprotease-mediated cleavage in its perimembranous extracellular domain, a process that liberates the extracellular domain as the soluble GH-binding protein and leaves behind a cell-associated GHR remnant protein containing the transmembrane and cytoplasmic domains. GHR metalloproteolysis can be catalyzed by tumor necrosis factor-α-converting enzyme (ADAM-17) and is associated with down-modulation of GH signaling. We now study the fate of the GHR remnant protein. By anti-GHR cytoplasmic domain immunoblotting, we observed that the remnant induced in response to phorbol ester or platelet-derived growth factor has a reliable pattern of appearance and disappearance in both mouse preadipocytes endogenously expressing GHR and transfected fibroblasts expressing rabbit GHR. Lactacystin, a specific proteasome inhibitor, did not appreciably change the time course of remnant appearance or clearance but allowed detection of the GHR stub, a receptor fragment slightly smaller than the remnant but containing the C terminus of the remnant (receptor cytoplasmic domain). In contrast, MG132, another (less specific) proteasome inhibitor, strongly inhibited remnant clearance and prevented stub appearance. Inhibitors of γ-secretase, an aspartyl protease, also prevented the appearance of the stub, even in the presence of lactacystin, and concomitantly inhibited remnant clearance in the same fashion as MG132. In addition, mouse embryonic fibroblasts derived from presenilin 1 and 2 (PS1/2) knockouts recapitulated the γ-secretase inhibitor studies, as compared with their littermate controls (PS1/2 wild type). Confocal microscopy indicated that the GHR cytoplasmic domain became localized to the nucleus in a fashion dependent on PS1/2 activity. These data indicate that the GHR is subject to sequential proteolysis by metalloprotease and γ-secretase activities and may suggest GH-independent roles for the GHR.

As evidenced by the phenotypes of naturally occurring mutations in humans and targeted gene disruption in mice, the growth hormone receptor (GHR) 1 is essential for mediation of the profound metabolic and somatogenic effects of GH (1)(2)(3). GHR is a single membrane-spanning cell surface type 1 glycoprotein member of the cytokine receptor superfamily (4). It transduces signals in response to GH by coupling to activation of the STAT (1, 3, and 5b), mitogen-activated protein kinase (extracellular signal-regulated kinase, c-Jun N-terminal kinase, and p38), and phosphatidylinositol 3-kinase pathways to affect the expression of various target genes, such as insulinlike growth factor-1, serine protease inhibitor 2.1 (Spi2.1), and c-fos, and other aspects of cellular function (4,5). Essential to this coupling is the GH-induced activation of the cytoplasmic tyrosine kinase, JAK2, which associates with the cytoplasmic domain of the receptor (6 -11).
The GHR is a target for proteolytic processing. Treatment of a variety of cell types with the phorbol ester, PMA, or plateletderived growth factor (PDGF) results in loss of the full-length receptor and appearance of a cell-associated cytoplasmic domain-containing GHR fragment that we have termed the "remnant" protein (12)(13)(14)(15). This cleavage also yields a soluble GHR form comprised of the extracellular domain of the receptor, which is referred to as the GH-binding protein (GHBP), in correspondence with the high affinity GH-binding protein found in the circulation of many species (16). In humans and rabbits (and likely in many other species), this process of proteolytic shedding of the GHR extracellular domain accounts for the generation of GHBP in vivo (16). By use of hydroxamatebased inhibitors and genetic reconstitution strategies, we have previously implicated metalloprotease activity as required for inducible GHR proteolysis in cell culture systems and have identified a particular transmembrane metalloprotease, tumor necrosis factor-␣-converting enzyme (ADAM-17), as a GHBP sheddase (13,14,17). Our work has also indicated that metalloproteolytic GHR processing may be a mechanism of regulation of cellular GH sensitivity, in that GH-induced signaling is dampened after exposure to stimuli that promote receptor cleavage, but not if metalloprotease inhibitors are present or if noncleavable receptor mutants are expressed (14,18).
We recently mapped the metalloprotease-mediated cleavage sites in the rabbit (rb) and mouse GHRs at eight and nine residues, respectively, from the transmembrane domain in the receptor extracellular domain stalk region (18,19). This was accomplished by purifying the remnant protein and performing N-terminal amino acid sequencing, confirming that the remnant contains the transmembrane domain and the remaining 8 -9 (depending on species) extracellular domain residues, as well as the cytoplasmic domain. This pattern of ADAM-catalyzed metalloproteolytic cleavage in the proximal extracellular domain is similar to that observed for a number of cell surface proteins, including pro-tumor necrosis factor-␣, pro-transforming growth factor-␣, pro-heparin-binding-epidermal growth factor, the tumor necrosis factor receptors, type II interleukin-1 receptor, interleukin-2 receptor-␣, interleukin-6 receptor, amyloid precursor protein (APP), Notch, p75 NTR , and ErbB-4 (Refs. 20 -25 and references therein). Some proteins that undergo extracellular domain metalloprotease-mediated cleavage also exhibit a subsequent cleavage of their resultant transmembrane/cytoplasmic domain remnants within the transmembrane domain. For these proteins, which include, among others, APP, Notch, p75 NTR , and ErbB-4, this intramembranous cleavage is mediated by a protein complex known as ␥-secretase (22,23,(25)(26)(27)(28)(29)(30)(31). Formation of an active ␥-secretase complex depends upon four proteins: presenilin, APH-1, nicastrin (APH-2), and Pen-2 (32). Coexpression of these proteins is sufficient for reconstitution of ␥-secretase activity in yeast, which not only lacks this protease activity but also has no apparent orthologs of these proteins (33). Of these four proteins, it is believed that presenilin itself forms the catalytic core by functioning as an aspartyl protease (32). ␥-Secretase-mediated cleavage of proteins creates soluble intracellular domains that have signaling functions; thus, this process, referred to as regulated intramembrane proteolysis, has attracted increasing interest as an important mechanism for modulating cellular responsiveness.
In this report, we explored the fate of the metalloproteasegenerated GHR remnant protein. Using both endogenous GHR-expressing cells and GHR reconstitution systems, we demonstrate that the remnant has a characteristic time course of degradation. Our results indicate that the remnant undergoes transition into a slightly smaller form (the "stub"; see Fig. 9) that is itself degraded in a proteasome activity-dependent fashion. This remnant-to-stub transition is inhibited by ␥-secretase inhibitors and does not occur in cells devoid of presenilins, suggesting that the GHR remnant is a direct substrate for presenilin-dependent ␥-secretase cleavage. Further, in the presence of proteasome inhibitors, we detect ␥-secretasedependent accumulation of the GHR cytoplasmic domain in the nucleus. The parallels revealed by our studies between the GHR and other ␥-secretase substrates, such as APP and Notch, suggest that a functional role(s) for metalloprotease/␥-secretase cleavage of GHR may exist.

EXPERIMENTAL PROCEDURES
Materials-PMA and routine reagents were purchased from Sigma unless otherwise noted. MG132 (Z-LLL-CHO) was purchased from Calbiochem (La Jolla, CA). Immunex Compound 3 (IC3), supplied by Amgen Corporation, is identical to Compound 2 (34), except that the napthylalanine side chain is replaced by a tert-butyl group. Talon Metal Affinity Resin, used for His-tagged protein purification, was from Clontech.
The ␥-secretase inhibitors used for each experiment are noted in the figure legends. The IC 50 of Compound E is 0.3 nM (35), and the IC 50 for N-[N- (3,5-difluorophenacetyl-L-alanyl]-S-phenylglycine-t-butyl ester (DAPT) is 20 nM (36). These values were derived from cell-based APP ␥-secretase cleavage assays and are used as guidelines. The actual IC 50 values as related to the ␥-secretase cleavage of the GHR have not been determined.
Plasmid Construction-The rbGHR cDNA was a kind gift of Dr. W. Wood (Genentech, Inc.). Construction of the cDNA encoding rbGHR del 297-406-His has been described (7,18,19,42). This mutant has intact extracellular and transmembrane domains but lacks residues 297-406 in the cytoplasmic domain (the full-length rbGHR has 620 residues). The Box 1 region in the proximal cytoplasmic domain is intact, as is the distal two-thirds of the cytoplasmic domain, which contains known GHR tyrosine phosphorylation sites, but the major internalization motif (43,44) is absent. The rbGHR del 297-406 del 237-239-His mutant has also been described (18). In addition to internal deletion of residues 297-406, this mutant also lacks residues 237-239, which includes the rb-GHR cleavage site, and is thus resistant to metalloprotease-mediated proteolysis (18). The pcDNA 3.1(Ϫ) rbGHR 239 -620-HA recombinant remnant corresponds to the mapped metalloprotease cleavage site but contains both the GHR signal sequence and a C-terminal HA tag following spacer residues. cDNAs encoding both rbGHR del 297-406-His and rbGHR del 297-406 del237-239-His were in the pcDNA 3.1(Ϫ) eukaryotic expression vector, as previously reported.
Generation of Recombinant Adenoviruses-The methods for generating the adenovirally expressed version of rbGHR del 297-406-His were described previously (37). Briefly, rbGHR del297-406-His cDNA was removed from pcDNA3.1(Ϫ) rbGHR del297-406-His by XbaI and KpnI digestion and inserted into pAdlox vector to form pAdlox-rbGHR del-297-406-His . Linearized pAdlox-rbGHR del 297-406-His and 5 helper virus DNA were cotransfected into CRE8 cells (45) (an HEK-293 derivative) by Lipofectamine (Invitrogen). The cells were harvested after several days when the cytopathic effects became apparent. After lysis by three freeze/thaw cycles, the cell debris was pelleted by centrifugation, and the supernatant was collected. This supernatant was used for infection of HEK-293 cells. Three further rounds of infection were performed to obtain a high titer viral stock, which was used for experimental and preparative infection. WT rbGHR-His adenovirus was constructed in the same way as Ad rbGHR del 297-406-His .
Antibodies-The 3F10 anti-HA rat monoclonal antibody was purchased from Roche. Three rabbit polyclonal antisera against the GHR were used. Anti-GHR cyt-AL47 was raised against a bacterially expressed N-terminally His-tagged fusion protein incorporating human GHR residues 271-620 (the entire cytoplasmic domain (46)), as described (15). Anti-GHR cyt-AL37 was raised against a bacterially expressed glutathione S-transferase fusion with human GHR residues 271-620, as described (47). Anti-GHR cyt was raised against a bacterially expressed maltosebinding protein fusion with human GHR residues 317-620, as described (48). In some experiments, the antisera were affinity purified for immunoblotting (as in Ref. 48). In others, they were not purified and gave rise to some nonspecific bands, as indicated. For immunofluorescence microscopy, anti-GHR cyt-AL47 was purified as follows. An IgG fraction was purified using protein A-Sepharose (nProtein A-Sepharose 4 Fast Flow; Amersham Biosciences). Bound antibodies were acideluted with 0.1 M glycine-HCl, pH 2.8, and neutralized with one-tenth volume Tris-HCl (1 M), pH 9.0. NaCl was then added to 150 mM. The eluted antibodies were then adsorbed against Escherichia coli acetone powder, as described (49).
Cell Stimulation, Protein Extraction, Immunoprecipitation, Electrophoresis, and Immunoblotting-Serum starvation of all cell lines was accomplished by substitution of 0.5% (w/v) bovine serum albumin (fraction V; Roche Applied Science) for serum in their respective culture media for 16 -20 h prior to experiments. Stimulations were performed at 37°C. Pretreatment of each cell line was as follows: 1) for 3T3-F442A, PS1/2 knock-out, and PS1/2 WT cells, vehicle control and/or inhibitors were added 30 min prior to stimulation; 2) for HEK-293, vehicle control and/or inhibitors were added 4 h prior to PMA stimulation; 3) for ␥2A-rbGHR, vehicle control and/or inhibitors were added 30 -60 min prior to PMA stimulation; and 4) for RϩT mouse fibroblasts, vehicle control and/or inhibitors were added 90 min prior to PMA stimulation. Details of the PMA (at 1 g/ml) treatment protocol have been described (12-15, 17, 42). Briefly, adherent cells were stimulated in binding buffer (consisting of 25 mM Tris-HCl, pH 7.4, 120 mM NaCl, 5 mM KCl, 1.2 mM MgCl 2 , 0.1% (w/v) bovine serum albumin, and 1 mM dextrose) or DMEM (low glucose) with 0.5% (w/v) bovine serum albumin. PDGF (final concentration, 40 ng/ml) and GH (final concentration, 500 ng/ml) stimulations were performed as described previously (14). Stimulations were terminated by washing the cells once with ice-cold phosphate-buffered saline (PBS) in the presence of 0.4 mM sodium orthovanadate (PBS-vanadate) and then harvesting by scraping in icecold PBS-vanadate. The pelleted cells were collected by brief centrifugation. For each cell type, the pelleted cells were solubilized for 15 min at 4°C in fusion lysis buffer (1% (v/v) Triton X-100, 150 mM NaCl, 10% (v/v) glycerol, 50 mM Tris-HCl, pH 7.3, 100 mM NaF, 5 mM EDTA, 1 mM phenylmethylsulfonyl fluoride, 1 mM sodium orthovanadate, 10 mM benzamidine, and 10 g/ml aprotinin), as indicated. After centrifugation at 20,000 ϫ g for 15 min at 4°C, the detergent extracts were electrophoresed under reducing conditions by addition of Laemmli SDS-PAGE sample buffer or subjected to immunoprecipitation, as indicated. The protein extracts and immunoprecipitates were resolved by SDS-PAGE (8% acrylamide, unless otherwise noted).
High titered adenovirus stock encoding rbGHR del 297-406-His , described above, was used to infect HEK-293 cells. After 2 h of incubation with adenovirus, the medium was replenished. At 12 h postinfection, the medium was removed, and serum starvation was initiated. After 18 h, the cells were treated with control vehicle or lactacystin followed by 45 min of PMA (1 g/ml) stimulation and then harvested and lysed with fusion lysis buffer (modified to lack EDTA). TALON metal affinity resin (Co 2ϩ -TC-Sepharose) beads were added to the detergent cell extract. After 1 h of rolling incubation at 4 C, the beads were washed twice with 1 ml of modified fusion lysis buffer, followed by three washes with 10 mM imidazole in modified fusion lysis buffer. The bound proteins were eluted with Laemmli SDS-PAGE sample buffer and resolved by SDS-PAGE (10% acrylamide). The same purification procedures were used in experiments in which rbGHR del 297-406-His and rbGHR del 297-406 del 237-239-His were expressed by transient transfection. High titered adenoviral infection into PS1/2 K/O or WT MEF cells with Ad rbGHR del 297-406-His and sample processing steps were performed as in HEK293 cells. Resolution of proteins by SDS-PAGE, Western transfer of proteins, and blocking of Hybond-ECL (Amersham Biosciences) with 2% bovine serum albumin were performed as previously described (13)(14)(15)17). Immunoblotting with antibodies 4G10 (1:2000), anti-GHR cyt-AL47 (1:3000), anti-GHRcyt-AL37 (1:2000), and anti-GHR cyt (1:1000), with horseradish peroxidase-conjugated anti-mouse or anti-rabbit secondary antibodies (1:50000) and detection reagents (SuperSignal West Pico Chemiluminescent Substrate) (all from Pierce) and stripping and reprobing of blots were accomplished according to the manufacturers' suggestions.
Confocal Microscopy-For confocal laser scanning microscopy, a Leica TCS SP unit equipped with a UV laser (Coherent Laser Group Enterprise) was used. The images were obtained using a 100ϫ objective (numerical aperature 1.4) with an X-Y resolution of ϳ50 m. The cells were fixed using the pH shift method. ␥2A-JAK2 cells were transfected in suspension using FuGENE 6 (Invitrogen) reagent according to the provided instructions. These were then plated onto 18-mm 2 coverslips (No. 1) overnight. After ϳ24 h, the indicated treatments were initiated. After 30 min of incubation at 37°C, the cells were rinsed twice with cold 1ϫ PBS. PS1/2 cells were adenovirally infected (as described above) for 12 h prior to plating on coverslips. After 8 h, the medium covering the coverslips was replaced with serum starvation medium. The stimulations were performed 12 h after serum starvation began.
Stimulations were terminated with ice-cold PBS. Formaldehyde (Tousimis Research Corporation) (4% v/v) in PEM buffer (0.1 M PIPES, 1 mM MgCl 2 , 1 mM EGTA), pH 6.5, was then added at room temperature. After 5 min, this was replaced by 4% formaldehyde in 100 mM borax buffer, pH 11.0, for an additional 10 min. The cells were permeabilized for 15 min with 0.1% (v/v) Triton X-100 in PBS. Following extensive washing with PBS, antibody binding solution (PBS, 5% (v/v) normal goat serum, and 0.02% (v/v) Triton X-100) containing 5 g/ml 3F10 mAb or anti-GHR cyt-AL47 (1:5000) was added to coat the coverslips, and allowed to incubate for 1 h at room temperature. Following extensive washing with PBS, antibody binding solution containing goat antirat IgG coupled to Alexa 488 (1:150 dilution) or goat anti-rabbit IgG coupled to Alexa 594 (1:100 dilution) was added for 1 h at room temperature (both secondary antibodies from Molecular BioProbes). After washing, the coverslips were mounted onto glass slides using Vecta-Shield containing 4Ј,6Ј-diamino-2-phenylindole and sealed with clear fingernail polish.

RESULTS
Effects of the Proteasome Inhibitor, MG132 on the GHR Remnant-Several reports suggest potential roles for proteasome activity in regulation of cellular GHR levels, GHBP generation, and/or aspects of GH-induced signaling (50 -58). In a number of these studies, MG132 was used as an inhibitor of proteasome activity. We sought to determine whether MG132 affected the generation or fate of the GHR remnant that results from receptor proteolysis. We first examined murine 3T3-F442A preadipocyte fibroblasts, which endogenously express GHRs and display GH responsiveness (14, 42, 59 -63). In the experiment shown in Fig. 1A, 3T3-F442A cells were serum-starved and treated with the phorbol ester PMA for 15 or 60 min or with Me 2 SO vehicle control in the absence (lanes 1-3) or presence (lanes 4 -6) of MG132. The cells were harvested, and detergentsolubilized proteins were separated by SDS-PAGE and immunoblotted with anti-GHR cyt-AL37 , an antiserum that recognizes the receptor cytoplasmic domain (15,47). Consistent with previous results (14), PMA treatment of these cells caused a progressive loss of GHR in the absence of MG132. This loss was only marginally affected (augmented) in the presence of MG132 over the 60-min PMA stimulation period (compare lanes 4 -6 versus lanes [1][2][3]. Concomitant with the loss of GHR, PMA caused the appearance of a roughly 65-kDa anti-GHR cyt-AL47reactive protein (lane 2 versus lane 1), which we have previously described as the cytoplasmic domain-containing remnant protein (14). Notably, in the absence of MG132, the abundance of PMA-induced GHR remnant protein detected by immunoblotting decreased over time (compare the 60-min versus the 15-min PMA treatment, lane 3 versus lane 2), suggesting that the newly generated remnant undergoes degradation over the period examined. Interestingly, this time-dependent remnant loss was prevented by MG132 pretreatment (compare lanes 5 and 6 versus lanes 2 and 3), consistent with the notion that MG132 prevents remnant degradation.
We next addressed whether the kinetics of remnant formation and the effects of MG132 seen in cells endogenously expressing the mouse GHR would also be observed in a reconstitution system in which the rbGHR was heterologously expressed. For these experiments, we utilized the ␥2A-GHR cell, a human fibrosarcoma cell that stably expresses the rb-GHR that we have previously shown to display receptors at the cell surface and to respond to PMA treatment with receptor proteolysis (15,64). Treatment of these cells with PMA in the absence of MG132 resulted in loss of the mature, fully glycosylated form (64) of rbGHR (bracketed form, Fig. 1B, lanes 2 and 3 versus lane 1). The remnant was present basally, as previously observed (15), and was increased by PMA treatment for 15 min (lane 2 versus lane 1). As seen for 3T3-F442A cells, treatment of ␥2A-GHR cells with PMA for 1 h resulted in less detectable remnant than did the 15-min treatment (lane 3 versus lane 2). Further, the presence of MG132 prevented neither rbGHR loss nor remnant accumulation in response to PMA but blocked the time-dependent decrease in remnant abundance with 60 versus 15 min of treatment (Fig. 1B, lanes  4 -6). These data suggested that receptor loss, remnant formation, and time-dependent, MG132-sensitive remnant disposition followed similar patterns independent of the species of the GHR or cells in which it was expressed and whether or not the GHR expression was driven by natural or artificial promoters.
A previous report documented that treatment of murine Ba/F3 pro-B cells stably transfected with the human GHR with MG132 for 90 min resulted in shedding of GHBP into the cell supernatant (56). This effect was inhibited by certain protein kinase C inhibitors and was presumed to be mediated by metalloproteolysis, although this was not verified. Our data in Fig.  1 (A and B) suggested that MG132 alone did not substantially cause either mouse GHR or rbGHR proteolysis over the time period examined. To further investigate this issue, we treated serum-starved 3T3-F442A cells with MG132 for 0 -6 h in the presence or absence of IC3, a hydroxamate-based metalloprotease inhibitor (13, 14, 17-19, 34) (Fig. 1C, lanes 1-8). For comparison, the cells were treated acutely with PMA in the presence or absence of IC3 (Fig. 1C, lanes 9 -11). Receptor proteolysis (GHR loss and remnant accumulation) was tracked by anti-GHR cyt immunoblotting of cell extracts. We observed that MG132 resulted in GHR remnant accumulation with treatment for greater than 2 h (lanes 3 and 4 versus lanes 1 and 2). This effect was markedly inhibited by IC3 (lanes 7 and 8 versus lanes 3 and 4), indicating that it was largely mediated by metalloprotease activity. Notably, the remnant appearing in response to prolonged MG132 exposure migrated in SDS-PAGE indistinguishably from that generated in a metalloprotease-dependent fashion in response to PMA (lane 10 versus lanes 3 and 4). Thus, in our system, MG132, perhaps by potentiating the stability of proteins involved in GHR proteolysis that may be otherwise proteasomally degraded, can promote metalloproteolysis of the receptor. However, this effect is seen only with prolonged MG132 exposure and is thus unlikely to explain the enhancement of PMA-induced remnant abundance seen in Fig. 1 (A and B).
Effects of the Proteasome Inhibitor, Lactacystin, and Identification of the GHR Stub-The data in Fig. 1 suggested that the stability of the GHR remnant may be influenced by proteasome activity. To test the validity of this conclusion, we used a more specific proteasome inhibitor, clasto-lactacystin ␤-lactone, an active analog of lactacystin (referred to as lactacystin in the text and figures). 3T3-F442A cells were treated with PMA for 60 min in the presence of either MG132 or lactacystin, lysed, and analyzed by anti-GHR cyt immunoblotting ( Fig. 2A) 1-3), as indicated. The cells were then exposed to PMA for 15 (lanes 2 and 5) or 60 min (lanes 3 and 6) or vehicle (lanes 1 and 4) prior to harvesting and detergent extraction. Extracted proteins were resolved by SDS-PAGE and immunoblotted with anti-GHR cyt-AL37 . The positions of the prestained molecular weight markers, the GHR (bracket), and GHR remnant (arrowhead) are indicated. The data shown in A and B are representative of five and four such experiments, respectively. C, effects of prolonged MG132 treatment on GHR remnant abundance. Serum-starved 3T3-F442A cells were pretreated with IC3 (lanes 5-8 and 11) or its vehicle (lanes 1-4, 9, and 10) and then treated with MG132 or its vehicle for the indicated durations (lanes 1-8) or with PMA or its vehicle, as indicated (lanes 9 -11). Detergent-extracted proteins were immunoblotted with anti-GHR cyt-AL37 . The positions of GHR and the GHR remnant are indicated. The data shown are representative of five such experiments. WB, Western blot. abundance but instead revealed the presence of a faster migrating protein also reactive with GHR cytoplasmic domain antibody (lane 4 versus lanes 2 and 3). To discriminate it from the remnant, we term this smaller GHR fragment the receptor stub (refer to Fig. 9).
We previously demonstrated that stimulation of 3T3-F442A cells with PDGF, like treatment with PMA, resulted in metalloprotease-mediated GHR loss and remnant accumulation (14). Because PDGF is a physiologically relevant serum factor, we sought to compare its effects on stub formation with those of PMA (Fig. 2B). PDGF treatment of 3T3-F442A cells for 30 min resulted in GHR loss and remnant accumulation, as expected (lane 3 versus lane 1). Notably, treatment with PDGF in the presence of lactacystin yielded both the remnant and stub in a pattern indistinguishable from that induced by PMA (lane 4 versus lane 2). This result indicated that the appearance of the stub, made evident by lactacystin treatment, could be generated in response to both stimuli and suggested that the stub is likely a labile protein susceptible to proteasomal degradation.
To better characterize the stub and requirements for its generation, we utilized our previously described adenoviral expression system in HEK-293 cells (18). We prepared infectious adenoviral particles encoding a C-terminally His-tagged version of the rbGHR that has an in-frame internal deletion of residues 297-406 in the cytoplasmic domain. This mutant, rbGHR del 297-406-His , lacks the internalization and UbE motif (43,44), is highly expressed at the cell surface, binds GH normally, couples to GH-induced JAK2 activation, and, when adenovi-rally expressed in HEK-293 cells, undergoes inducible metalloprotease-mediated proteolysis (18). Adenovirally infected serum-starved cells were treated with PMA for 45 min in the absence or presence of lactacystin and detergent-solubilized (Fig. 2C). The lysates were subjected to metal affinity bead precipitation. The proteins retained following imidazole washes were eluted, resolved by SDS-PAGE, and immunoblotted with anti-GHR cyt-AL47 . PMA-induced loss of receptor (the fully glycosylated form (18)) and the appearance of the ϳ43-kDa remnant were observed independent of lactacystin ( lanes 1 and 2 versus lanes 3 and 4). However, the receptor stub, migrating slightly faster than the remnant, was only detected in the presence of lactacystin. This result strongly suggested that the stub contains the C terminus of the receptor, in that it is precipitated via the C-terminal His tag. It also suggests that the stub can be generated from rbGHR as well as from mouse GHR (as shown in Fig. 2, A and B) and that it can arise from a GHR that is defective in internalization and highly localized to the cell surface.
The time courses of appearance of remnant and stub were examined in Fig. 2D, in which PMA treatment of 3T3-F442A cells for 15-60 min was carried out in the presence of lactacystin. Lactacystin, unlike MG132, did not affect the kinetics of remnant accumulation in response to PMA. Peak remnant abundance was seen after 30 min of treatment, and its abundance declined thereafter (lanes 2-4 versus lane 1), a pattern very similar to that observed in the absence of lactacystin (as in Fig. 1, A and B). In contrast, the stub was only apparent after 30 min, but not 15 min, of PMA treatment, and stub abundance was further increased at 60 min in the presence of lactacystin (lanes 3 and 4 versus lanes 1 and 2). These kinetics are consistent with the possibility that the remnant and stub exist in a precursor-product relationship.
The data in Figs. 1 and 2 (A-D) suggested that short term treatments with the proteasome inhibitors MG132 and lactacystin differentially affected the fates of the PMA-induced remnant and stub. In the experiment shown in Fig. 2E, we examined the longer term effects of lactacystin on basal and PMA-induced remnant and stub formation. 3T3-F442A cells were pretreated with lactacystin or its vehicle for 3 or 6 h prior to exposure to PMA or its vehicle for 30 min. In contrast to longer term treatment with MG132 (Fig. 1C) Because the kinetics of formation of remnant and stub differ, we explored whether appearance of the stub could occur if remnant formation were blocked. In the experiments shown in Fig. 3A, mouse fibroblasts stably expressing rbGHR and murine tumor necrosis factor-␣-converting enzyme (17) were pretreated with lactacystin and then exposed to PMA in the presence or absence of either IC3 (a hydroxamate metalloprotease inhibitor; lanes 2 and 3), or 1,10-phenanthroline (a divalent cation chelator with particular affinity for Zn 2ϩ ; lanes 4 and 5), also an inhibitor of metalloproteases (65). Both remnant and stub were detected in response to PMA, as expected. Importantly, both GHR fragments were not detected if either IC3 or 1,10-phenanthroline was present. The same result was obtained with 3T3-F442A cells (data not shown).
We also examined this issue by determining whether the stub could be formed when a receptor rendered noncleavable by mutagenesis was expressed. The rbGHR del 297-406 mutant was expressed by transfection in HEK-293 cells and compared with rbGHR del 297-406 del 237-239 . The latter mutant is deleted of the residues surrounding the inducible metalloproteasemediated rbGHR cleavage site and has been shown to be resistant to PMA-induced proteolysis (18). As seen in Fig. 3B, rbGHR del 297-406 del 237-239 did not respond to PMA treatment with generation of either the remnant or the stub (lanes 3 and  4), whereas both receptor fragments were observed, as expected, in response to PMA for rbGHR del 297-406 (lanes 1 and 2). The results in Fig. 3 suggest that either the metalloprotease cleavage of the GHR is a prerequisite for stub formation or that the stub is itself also a product of metalloprotease activity, or both.
Remnant-to-Stub Transition Is Dependent on ␥-Secretase Activity and Presenilin(s)-We sought to address further the nature of the process that gives rise to the stub and to define whether or not it is the same (metalloprotease) activity that catalyzes remnant formation. The differences in the effects of MG132 and lactacystin that we observed above in Figs. 1 and 2 provided a clue to understanding these processes. MG132 allowed enhanced detection of the acutely generated remnant but did not allow detection of the stub. Lactacystin did not strongly potentiate the acute remnant appearance but allowed detection of the stub. Notably, peptide aldehydes such as MG132, in addition to functioning as proteasome inhibitors, have been shown to inhibit the activity of ␥-secretase, an aspartyl protease activity important in regulated intramembrane proteolysis (66,67). In contrast, lactacystin has not been shown to function as a ␥-secretase inhibitor.
We thus tested the effect of ␥-secretase inhibitors (GSI) on the formation of remnant and stub. 3T3-F442A cells were treated with PMA for 0 -60 min in the presence or absence of lactacystin alone, lactacystin plus GSI, or GSI alone (Fig. 4A). PMA caused loss of GHR abundance with similar time course independent of the presence of GSI (lanes 1-4 versus lanes  7-10). Interestingly, the abundance of PMA-induced remnant, however, was dramatically affected by inclusion of GSI. As seen above in Figs. 1 (A and B) and 2D, PMA-induced remnant abundance fell after 15-30 min in the absence of GSI (Fig. 4A,  lanes 2-4). In contrast, remnant abundance did not appreciably decline over the 60 min of PMA treatment in the presence of GSI (lanes 8 -10), the same pattern as that observed for PMA-induced remnant in the presence of MG132 (Fig. 1, A and   FIG. 3. Inhibition of metalloprotease-mediated GHR proteolysis prevents both remnant and stub appearance. A, effects of metalloprotease inhibitors. Serum-starved mouse fibroblasts stably expressing rbGHR (17) were pretreated with lactacystin and then exposed to PMA or vehicle in the presence or absence of IC3 or 1,10-phenanthroline, as indicated. Detergent-extracted proteins were immunoblotted with anti-GHR cyt-AL37. The positions of GHR, remnant, and stub are indicated. Note that lanes 1-3 and lanes 4 and 5 are from two separate experiments. The data shown are representative of four such experiments. B, expression of a noncleavable GHR mutant. HEK-293 cells were transiently transfected with cDNAs encoding rbGHR del 297-406-His (lanes 1 and 2) or rbGHRdel 297-406 del 237-239-His (lanes 3 and 4). Serum-starved cells were treated with lactacystin for 4 h prior to treatment with PMA or vehicle for 45 min. Detergentextracted proteins were purified by precipitation with Talon beads, as The same findings were obtained when PDGF was used as the inducer of GHR proteolysis in 3T3-F442A cells (Fig. 4B). PDGF-induced remnant abundance was prolonged in the presence of GSI ( lanes 5 and 6 versus lanes 2 and 3), and the appearance of the stub in response to PDGF was blocked by GSI (lane 10 versus lane 9). GSI also inhibited the appearance of the stub in cells expressing the rbGHR (Fig. 4C). Mouse fibroblasts expressing the rbGHR were treated with PMA in the presence of lactacystin. Increasing concentrations of GSI caused a progressive loss of stub and a concomitant increase in the abundance of remnant. The results in Fig. 4, in concert with the kinetics data in Fig. 2D, strongly suggested that the PMAand PDGF-induced remnant transitions into the stub in a ␥-secretase-mediated fashion.
Cellular ␥-secretase activity requires the expression of a presenilin molecule, PS1 or PS2 (39). Targeted deletion of both PS1 and PS2 in mice results in embryonic lethality (39). However, MEFs from PS1/PS2 knock-out mice can be carried in tissue culture and can allow for a system to test the effects of genetic cellular ␥-secretase deficiency (39). To test whether GHR remnant-to-stub transition requires presenilins, we compared PS1/PS2 knock-out MEFs to wild type MEFs. Initial immunoblotting experiments (not shown) revealed that both cells lacked significantly detectable GHR. To express GHR in these cells, we took advantage of recent reports that indicated their receptivity to adenoviral infection (68). As in Figs. 2C and 3B, adenoviral particles encoding rbGHR del 297-406-His were used to infect WT (Fig. 5, lanes 1, 2, 5, and 6) or PS1/PS2 knock-out (lanes 3 and 4) MEFs. WT cells were first treated with PMA or vehicle for 45 min in the presence of lactacystin (lane 2 versus lane 1), and the cell lysates were metal affinityprecipitated to concentrate C-terminally His-tagged receptor, remnant, and stub. These cells exhibited substantial basal levels of both remnant and stub, with PMA treatment only modestly affecting these. Inclusion of a GSI during the PMA treatment of WT cells caused increased remnant and eliminated the appearance of the stub (lane 6 versus lane 5), consistent with the findings of Fig. 4. Notably, despite ample receptor expression and abundant basal and PMA-induced remnant levels, no stub was detected under any conditions in the PS1/PS2 knock-out cells (lanes 3 and 4 versus lanes 1, 2,  and 5). These results strongly demonstrate that presenilins are not required for remnant generation but that remnant-to-stub transition does not occur in the absence of PS1 and PS2. Furthermore, they complement the findings obtained with GSI drugs that GHR remnant-to-stub transition is a mechanism of remnant clearance.
GH Does Not Promote Stub Formation-For some receptors, such as Notch and ErbB-4, sequential metalloprotease-and ␥-secretase cleavages are induced by ligand engagement (22,23). In contrast, our previous data indicated that metallopro- teolytic cleavage of GHR is not promoted by GH; rather, GH pretreatment renders the receptor less susceptible to PMAinduced proteolysis (15). This inhibition was not associated with GH-induced GHR signaling or endocytosis and downregulation but was instead related to GH-induced changes in GHR conformation (15). We tested whether ␥-secretase-mediated GHR stub formation was promoted by GH in the experiment shown in Fig. 6. Serum-starved 3T3-F442A cells preincubated with lactacystin were treated with PDGF for 60 min in the presence (lane 3) or absence (lane 2) of a GSI. As expected, PDGF promoted enhanced abundance of remnant and stub (lane 2), and ␥-secretase inhibition prevented stub appearance and enhanced the level of remnant (lane 3). GH treatment for 30 or 60 min reduced the level of GHR, consistent with the promotion of ligand-induced endocytosis and degradation; however, there was no GH-dependent increase in remnant or appearance of stub (lanes 4 and 5). These findings suggest that, unlike Notch and ErbB-4, ␥-secretase-mediated GHR stub formation was not encouraged by ligand engagement.
GHR Intracellular Domain Is Nucleus-localized in a ␥-Secretase-dependent Fashion-Some ␥-secretase products are translocated to the nucleus, where they can act as modulators of transcription (22, 23, 26, 69 -73). Previous studies have demonstrated that the GHR may be nuclear-localized (74), and we have observed in yeast two-hybrid studies that GHR intracellular domain, when fused to a DNA-binding domain, causes transcriptional activation (data not shown). Furthermore, a Gal4 DNA-binding domain fused to the rbGHR intracellular domain can activate transcription in a eukaryotic transactivation assay (data not shown). We therefore explored whether the GHR stub can localize to the nucleus. We first expressed the WT rbGHR by adenoviral infection in WT or PS1/PS2 knockout MEFs (as in Fig. 5) and detected receptor expression by confocal microscopy with purified anti-GHR cyt-AL47 (Fig. 7). In both WT and PS1/PS2 knock-out cells treated with PMA, GHR was detected in a punctuate pattern throughout the cytoplasm and also at the cell surface. The pattern of anti-GHR cyt-AL47 reactivity in PMA-treated WT cells differed, however, in the presence of the proteasome inhibitor epoxomycin. Under these conditions, GHR intracellular domain reactivity was detected in both the cytoplasm and the nucleus. By comparison with WT cells, nuclear staining was dramatically less in PS1/PS2 knockout cells treated with PMA plus epoxomycin. This suggests that ␥-secretase activity allows the accumulation of a proteasomally degraded labile GHR intracellular domain-containing protein in the nucleus.
To further define the GHR cytoplasmic domain element that accumulates in the nucleus, we constructed an expression vector to drive expression of a recombinant GHR remnant encoding residues 239 -620 of the rbGHR with an HA tag on the C terminus (C-HA-GHR remnant; Fig. 8A). This remnant has at  7. Presenilin-dependent GHR nuclear localization. Wild type or PS1/2 knock-out murine embryonic fibroblasts grown on coverslips were adenovirally infected to express rbGHR, as described under "Experimental Procedures." Serum-starved cells were treated with PMA (45 min) in the presence or absence of the proteasome inhibitor, epoxomycin, as indicated. The cells were fixed, permeabilized, and stained with purified anti-GHR cyt-AL47 and Alexa 594-labeled anti-rabbit secondary antibody prior to confocal laser scanning microscopy, as described under "Experimental Procedures." Note epoxomycin-dependent nuclear accumulation of GHR in WT, but not PS1/2 knock-out cells. The data shown are representative of three such experiments.
its N terminus the GHR extracellular domain residue 239, which is eight residues external to the transmembrane domain, in accordance with our previous mapping of the metalloprotease cleavage site between residues 238 and 239 (18). It also includes the transmembrane domain and the entire cytoplasmic domain. The C-HA-GHR remnant was expressed by transient transfection into human fibrosarcoma cells that lack GHR (␥2A-JAK2 cells (40)) and detected by confocal fluorescence microscopy with anti-HA antibody (Fig. 8B, upper panels). In cells pretreated with Me 2 SO (vehicle) only, staining was present throughout the cytoplasm in a pattern similar to that seen for the full-length GHR in Fig. 7. No GHR was detected in the nucleus (stained by 4Ј,6Ј-diamino-2-phenylindole (DAPI); lower panel) under these conditions. In contrast, when cells were treated with the proteasome inhibitor lactacystin, abundant nuclear (as well as cytoplasmic) staining was detected with anti-HA, suggesting that proteasome inhibition either caused nuclear translocation of GHR remnant-derived protein or prevented degradation of such a protein, allowing its accumulation in the nucleus. Notably, when cells were treated with lactacystin in the presence of a ␥-secretase inhibitor, nuclear staining was abrogated. These findings parallel those of Figs. 4 -6, in which the GHR stub was most readily detected with proteasome inhibition, but remnant-to-stub transition (even in the presence of proteasome inhibitor) was prevented by inhibition of ␥-secretase activity. Collectively, the data in Figs. 7 and 8 suggest that the GHR stub, generated in a PS1/PS2 (␥-secretase)dependent fashion, can preferentially accumulate in the nucleus. DISCUSSION Proteolytic cleavage of the GHR has been observed in cell culture in a variety of model systems, including transient and stable transfectants; human, mouse, hamster, and monkey cells; and cells expressing human, rabbit, monkey, and mouse GHRs (12-15, 17-19, 75-85). Our work has focused on inducible, metalloprotease-mediated receptor proteolysis, and we have defined the products of this reaction as the shed GHBP and the cell-associated GHR remnant that contains the cytoplasmic and transmembrane domains and 8 or 9 (depending on species) extracellular domain residues (13)(14)(15)(17)(18)(19). This metalloproteolysis is interrelated with GH signaling, because it can desensitize cells to subsequent GH stimulation, and conversely, GH-induced conformational change lessens the susceptibility of the receptor to cleavage (14,18).
In this report, we initiated studies of the fate of the remnant protein (see Fig. 9 for summary diagram), reasoning that its disposition may be a regulated phenomenon. Indeed, we observed by anti-GHR cytoplasmic domain immunoblotting that remnant induced in response to PMA or PDGF has a reliable pattern of appearance and disappearance in both mouse preadipocytes that endogenously express the GHR and transfected fibroblasts expressing rbGHR. Lactacystin, a specific proteasome inhibitor, did not change the time course of remnant appearance or clearance, but MG132, another (less specific) proteasome inhibitor, strongly inhibited remnant clearance. Further, lactacystin, but not MG132, allowed detection of the GHR stub, a receptor fragment slightly smaller than the remnant, but containing the C terminus of the remnant (receptor cytoplasmic domain). Consistent with previous data, GH caused neither remnant nor stub formation. Inhibitors of ␥-secretase, an aspartyl protease, prevented the appearance of the stub, even in the presence of lactacystin, and concomitantly inhibited remnant clearance in the same fashion as MG132. Likewise, expression of GHR in cells devoid of presenilins 1 and 2 led to enhanced remnant accumulation without stub formation. Confocal microscopy studies suggested that the stub formed by ␥-secretase activity, if protected from proteasomal degradation, accumulates in the nucleus.  9. Sequential GHR cleavage. Shown is a diagram summarizing the findings in this study. Unliganded cell surface GHR undergoes metalloprotease-mediated cleavage in the extracellular domain stem region, which may be referred to as "␣-secretase" cleavage, in concordance with the terminology used for APP, Notch, and other substrates. For GHR, tumor necrosis factor-␣-converting enzyme (ADAM-17) can catalyze this reaction, which yields the soluble GHBP and the membrane-anchored GHR remnant. The remnant is a target for presenilindependent ␥-secretase cleavage, presumably within the transmembrane domain, resulting in production of the GHR stub. The stub is believed to leave the plasma membrane (by analogy to findings in other systems) and is a target for rapid proteasomal degradation; however, if it avoids degradation, the stub accumulates in the nucleus, where it may have (perhaps GH-independent) actions. We see our findings as important regarding several major issues. We are intrigued by the effects of proteasome inhibition on GHR, remnant, and stub in our cell systems. Substantial information exists regarding the role of the ubiquitin-proteasome pathway on the level and fate of the GHR in cells (reviewed in Ref. 54). Integrity of this pathway is apparently required for ligand-induced GHR internalization and has been suggested as a major regulator of cell surface GHR abundance in the absence of GH. These conclusions are based largely on studies of rbGHRs heterologously expressed by stable transfection in Chinese hamster fibroblasts and Ba/F3 cells stably transfected with the hGHR, in both cases treated with MG132 as a proteasome inhibitor (52,57,87). In addition, studies in Chinese hamster ovary cells stably transfected with rbGHR have shown that both MG132 and lactacystin prevented GHinduced GHR loss but were not tested for their effects on GHRs in the absence of GH (55). In our experiments, proteasome inhibition with MG132 for greater than 2 h caused loss of GHRs in 3T3-F442A cells with concomitant accumulation of GHR remnant, both of which were inhibited by the metalloprotease inhibitor IC3 (Fig. 1C). Prolonged exposure (3 or 6 h) to lactacystin alone had a less profound effect than MG132 on GHR loss and remnant accumulation in the same cells but did render the receptor stub detectable (Fig. 2E). The activation of GHR metalloproteolysis we observed with MG132 is consistent with that seen with proteasome inhibitors by others (56,87). Whether this is due to potentiation of the effects of components of the cleaving apparatus by their protection from degradation is not yet known. Interestingly, in the aggregate, we observed a less profound effect of prolonged proteasome inhibition on GHR abundance per se (i.e. promotion of increased GHR abundance) than on the generation of remnant and/or stub. We cannot completely explain the less dramatic effects on GHR abundance that we observe in comparison with the studies mentioned above. We note, however, that differences in the cell systems utilized may be relevant; for example, in contrast to those stable transfection systems, 3T3-F442A cells endogenously express GHRs.
The effects we observed of MG132 and lactacystin on the fate of remnant and stub in 3T3-F442A cells and the stable transfectants expressing the rbGHR are worthy of consideration. The kinetics of appearance of remnant in both homologous and heterologous systems (3T3-F442A and ␥2A-GHR, respectively) in response to PMA or PDGF was similar with peak remnant abundance seen at 15-30 min and decline thereafter. MG132 treatment, but not lactacystin, markedly prevented the decline in remnant abundance, strongly suggesting that this prevention was not simply an effect related to proteasome inhibition. Further, lactacystin treatment allowed the detection of the stub, whereas MG132 did not. These are important distinctions not only for understanding the fate of the remnant but also for illustrating the risk of attributing an effect of MG132 to proteasome inhibition without verifying such findings with independent and more specific proteasome inhibitors.
The use of specific inhibitors of ␥-secretase and cells devoid of PS1/PS2-mediated ␥-secretase activity allowed us to determine that this enzyme activity is critical in the transition from remnant to stub. Further, we conclude that ␥-secretase activity is a major mediator of remnant disposition, because PMA-or PDGF-induced remnant exhibited markedly increased stability in the presence of GSIs. The stub, in contrast to the remnant, appears to be quite labile, and its disposition, unlike that of the remnant, is proteasome-dependent. We cannot yet know whether the stub is degraded by the proteasome itself or whether its degradation merely requires the presence of an intact proteasome system. We do note, however, that anti-ubiquitin blotting experiments indicate that the stub is ubiquitinated (data not shown). This is not a surprising result, given that GHR is known to be ubiquitinated, even in the absence of GH stimulation (46,57), but whether ubiquitination is required for stub disposition is not yet known.
Our findings of metalloprotease-mediated cleavage of the GHR to yield remnant being followed by a ␥-secretase-mediated transition of remnant to stub and proteasome-dependent degradation of the stub are reminiscent of emerging information in other systems. Some notable examples of similar, but distinct, paradigms include diverse proteins such as APP (the Alzheimer disease amyloid precursor protein), Notch (an important molecule in development), the p75 neurotrophin receptor (p75 NTR ), and ErbB-4 (an intrinsic tyrosine kinase-containing growth factor receptor in the epidermal growth factor receptor family) (22,23,(25)(26)(27)(28)(29)(30)(31). In the case of APP, no known ligand triggers its proteolysis, but the initial cleavage events in the extracellular domain are either metalloprotease-or aspartyl protease-mediated (so-called ␣-(TACE) or ␤ (␤ amyloidcleaving enzyme)-secretase-mediated, respectively). Subsequent intramembranous cleavage of either the ␣-secretaseor ␤ amyloid-cleaving enzyme-generated remnant by ␥-secretase activity liberates the remaining extracellular domain part of the remnant; in the case of the ␤ amyloid-cleaving enzyme/␥secretase cleavages, this liberated APP component is the amyloidogenic A␤ fragment intimately related to the pathology of Alzheimer disease. However, in either case, the resulting stub (called the APP intracellular domain) is a labile protein (but not proteasomally degraded) that can be found in the nucleus and may itself have a role in transcriptional regulation (88). In the case of Notch, its intracytoplasmic domain is liberated by sequential ␣and ␥-secretase cleavage and localizes to the nucleus, where it interacts with transcription factors critical to Notch action. Notch intracytoplasmic domain is also quite labile, being rapidly degraded in a proteasome-dependent fashion. Also in contrast to APP, it is engagement of Notch by its cognate ligand that initiates this process. Engagement of ErbB-4 by its cognate ligand, heregulin, causes its proteolysis, although it may not be via metalloprotease activation (89); however, phorbol ester promotes both metalloprotease and ␥-secretase cleavage of ErbB-4 with nuclear accumulation of its cytoplasmic domain fragment (27,28). Although p75 NTR associates with nerve growth factor receptors (Trks), metalloprotease-and ␥-secretase cleavage of p75 NTR is not nerve growth factor-induced but may be modulated by Trk association (25,30,31). The p75 NTR stub is proteasomally degraded and highly labile, but its presence in the nucleus may mediate transcriptional signaling (30).
These findings make plain that metalloprotease-and ␥-secretase activities may confer to their individual targets a range of actions. Our confocal microscopy findings revealed that GHR, detected with antibodies to its intracellular domain, accumulated in the nucleus of PMA-treated fibroblasts treated with a proteasome inhibitor. However, this did not occur in fibroblasts devoid of PS1/PS2 activity; this is the same pattern observed by immunoblotting for GHR remnant-to-stub transition. Indeed, expression of the remnant itself in lactacystintreated fibrosarcoma cells resulted in GHR nuclear accumulation, which was prevented by inhibition of ␥-secretase activity.
Thus, ␥-secretase activity potentially has several actions relevant to the GHR. One role might be to facilitate clearance of the remnant by converting it to the stub, which is then rapidly degraded in a proteasome-dependent fashion. This clearance function has previously been invoked for ␥-secretase activity for other substrates (90). Indeed, we can detect tyrosine phosphorylation of the remnant in response to PDGF-or PMA-induced metalloproteolysis, and this is substantially augmented by inhibition of ␥-secretase activity (data not shown). Because the tyrosine phosphorylated GHR cytoplasmic domain is a good binding site for SH2-containing signaling molecules (91,92), we view it as likely that the remnant could possess signaling potential, even though it is not generated in response to GH. Thus, ␥-secretase cleavage of the remnant (which is also not promoted by GH) might regulate the signaling potential of the remnant. Alternatively, localization of the GHR stub to the nucleus may indicate that it serves a function (direct or indirect) in modulating gene regulation. In this respect, the lability of the stub may serve to regulate this effect. We note that our relative difficulty to observe the GHR stub without blocking proteasome activity is very similar to that observed for the Notch intracytoplasmic domain and the p75 NTR stub. As in those systems, this suggests that low levels of GHR stub, either present constitutively or induced in response to stimuli, might be sufficient to have effects. Such stimuli could include those such as protein kinase C activators, serum, or PDGF that are known to activate GHR metalloproteolysis (14), as well as those as yet unknown that may activate metalloprotease and ␥-secretase activity or lessen GHR stub degradation. More study of potential GHR stub-interacting partners, factors that may stabilize the stub in the nucleus, and target genes that may be affected by the stub will be required to address these issues.