Effect of Alternative Glycosylation on Insulin Receptor Processing*

The mature insulin receptor is a cell surface heterotetrameric glycoprotein composed of two α- and two β-subunits. In 3T3-L1 adipocytes as in other cell types, the receptor is synthesized as a single polypeptide consisting of uncleaved α- and β-subunits, migrating as a 190-kDa glycoprotein. To examine the importance of N-linked glycosylation on insulin receptor processing, we have used glucose deprivation as a tool to alter protein glycosylation. Western blot analysis shows that glucose deprivation led to a time-dependent accumulation of an alternative proreceptor of 170 kDa in a subcellular fraction consistent with endoplasmic reticulum localization. Co-precipitation assays provide evidence that the alternative proreceptor bound GRP78, an endoplasmic reticulum molecular chaperone. N-Glycosidase F treatment shows that the alternative proreceptor contained N-linked oligosaccharides. Yet, endoglycosidase H insensitivity indicates an aberrant oligosaccharide structure. Using pulse-chase methodology, we show that the synthetic rate was similar between the normal and alternative proreceptor. However, the normal proreceptor was processed into α- and β-subunits (t 1 2 = 1.3 ± 0.6 h), while the alternative proreceptor was degraded (t 1 2 = 5.1 ± 0.6 h). Upon refeeding cells that were initially deprived of glucose, the alternative proreceptor was processed to a higher molecular weight form and gained sensitivity to endoglycosidase H. This “intermediate” form of the proreceptor was also degraded, although a small fraction escaped degradation, resulting in cleavage to the α- and β-subunits. These data provide evidence for the first time that glucose deprivation leads to the accumulation of an alternative proreceptor, which can be post-translationally glycosylated with the readdition of glucose inducing both accelerated degradation and maturation.

The insulin receptor (IR) 1 is a membrane-bound glycoprotein belonging to the tyrosine kinase receptor family. Cell surface receptors bind extracellular insulin, which leads to autophosphorylation and activation of its tyrosine kinase, resulting in downstream activation of glucose uptake as well as glycogen, fatty acid, and DNA synthesis (1). The mature receptor is a heterotetramer composed of two ␣-subunits (135 kDa) and two ␤-subunits (95 kDa). However, the receptor is initially synthesized as a single polypeptide containing both the ␣and ␤-subunits (2). In the lumen of the endoplasmic reticulum (ER), the proreceptor is co-translationally N-glycosylated on both the ␣and ␤-subunits as shown by the incorporation of metabolically labeled monosaccharides, sensitivity to glycosidase digestions, and binding to specific lectins (3,4). Ultimately, the N-linked oligosaccharide structures are composed of both high mannose and complex types, the latter of which contains galactose, fucose, N-acetylglucosamine, and sialic acid. Elucidation of the primary sequence of the IR has identified 14 potential N-linked glycosylation sites on the ␣-subunit and four potential N-linked glycosylation sites on the ␤-subunit (5,6). In addition, the ␤-subunit has also been shown to contain an O-linked carbohydrate (3,7,8).
The IR has been extensively used as a model to examine the effect of glycosylation on processing and function. Initial studies were performed using tunicamycin treatment, which inhibits N-linked glycosylation and leads to the accumulation of an aglyco-proreceptor in an intracellular compartment (9 -12). The use of castanospermine and 1-deoxynojirimycin (glucosidase inhibitors) also leads to accumulation rather than processing (13). A direct correlation between aberrant glycosylation and processing/function has recently been addressed using sitedirected mutagenesis of individual N-linked glycosylation sites. Mutation of the asparagines within the four N-linked consensus sequences of the ␤-subunit has no effect on processing but inhibits signaling because of defective autophosphorylation and tyrosine kinase activity (8,14). Interestingly, only the third and fourth N-linked sites regulate the functional status of the receptor. In the ␣-subunit, mutation of the first four Nlinked glycosylation sites (16) or mutation of the first or second pair of sites (17) leads to accumulation of the proreceptor in the ER. Individual mutations have no effect on processing or function (17,18). Thus, it has been proposed that specific N-linked oligosaccharide sites (both location and number) in the IR differentially affect processing and gain of function.
Cells possess a number of survival mechanisms to cope with adverse changes in their environment. One stress-related response mechanism is termed the unfolded protein response. The unfolded protein response is induced by a variety of stress conditions, such as glucose or amino acid deprivation, tunicamycin treatment, or sulfhydryl-reducing agents, all of which affect protein folding in the ER (19). The induction of the unfolded protein response leads to the transcriptional activation of several molecular chaperones, including GRP78 (also known as BiP). GRP78 is an ER-resident molecular chaperone, which is hypothesized to play an important role in "quality" control in the ER (20). Under normal conditions, GRP78 binds transiently to newly synthesized proteins, releasing properly folded proteins for processing to the Golgi. Misfolded proteins remain bound to GRP78 and are eventually degraded by cytoplasmic proteosomes (21,22). Only two studies have examined the interaction of the insulin receptor with GRP78. In each case, overexpression of mutated receptors was necessary to examine this interaction (23,24).
Our goal was to use glucose deprivation as a reversible tool to influence insulin receptor processing in a physiological setting. We show for the first time that glucose deprivation of 3T3-L1 adipocytes causes an accumulation of the insulin proreceptor in the ER as a result of interaction with GRP78. Both accumulation and interaction can be reversed with glucose readdition.

EXPERIMENTAL PROCEDURES
Materials-Dulbecco's modified Eagle's medium (DMEM) and glucose-free DMEM were purchased from Life Technologies, Inc. Calf serum (lot S11450) was purchased from Atlanta Biologicals (Norcross, GA). Fetal bovine serum (lot 1020-90) was purchased from Intergen (Purchase, NY). N-Glycosidase F and endoglycosidase H were purchased from New England Biolabs (Beverly, MA). Promix containing [ 35 S]methionine and cysteine was purchased from Amersham Pharmacia Biotech. [2-3 H]mannose was purchased from NEN Life Science Products, Inc. The ECL detection system was also purchased from Amersham Pharmacia Biotech (RPN 2105). The affinity-purified rabbit polyclonal antibody against the IR ␤-subunit, IR␤ (SC-711) was purchased from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA). For immunoblot analysis of GRP78, we generated a rabbit polyclonal antibody against a peptide corresponding to the first 12 amino acids from the N terminus. For immunoprecipitation of GRP78, we purchased a polyclonal antibody generated against the C terminus of GRP78 (PA1-014) from Affinity BioReagents (Golden, CO).
Cell Culture-3T3-L1 fibroblasts were grown and differentiated in 100-mm plates according to the procedure of Frost and Lane (25). Twenty-four h prior to the start of a protocol, 3T3-L1 adipocytes were provided with fresh DMEM containing 10% fetal bovine serum (complete medium). This brings cells to a common starting point and provides the basis for reproducible data. Cells were then either fed complete medium or incubated in glucose-free DMEM containing 10% dialyzed fetal bovine serum (glucose-free medium) for the times indicated.
Subcellular Fractionation-Plasma membrane (PM), high density membranes (HDM), and low density membranes (LDM) were isolated from cell homogenates using a technique described by Fisher and Frost (26). Briefly, cells were washed in Krebs-Ringer phosphate buffer (128 mM NaCl, 4.7 mM KCl, 1.25 mM MgSO 4 , 1.25 mM CaCl 2 , 5.0 mM Na 2 HPO 4 , and 5.0 mM NaH 2 PO 4 , pH 7.4) and collected in TES (20 mM Tris, 1 mM EDTA, and 255 mM sucrose, pH 7.4) with 1 mM phenylmethylsulfonyl fluoride at 18°C. The cells were then passed through a steel block homogenizer at 18°C with a fixed clearance of 0.0025 inch. PM, HDM, and LDM fractions were collected by a combination of differential and sucrose gradient centrifugation. Protein concentration was determined using a modified Lowry assay (27).
Immunoblot Analysis-Fifty g of protein was solubilized in Laemmli sample dilution buffer (62 mM Tris, 10% glycerol, 3 g/ml bromphenol blue, 2% SDS, and 5% ␤-mercaptoethanol), containing 6 M urea. Proteins were resolved on a 5.0 or 7.5% SDS-polyacrylamide gel. After electrotransfer onto nitrocellulose, IR and GRP78 were detected by immunoblot analysis using chemiluminescence as described previously (28). The bands were quantitated by video densitometry on a Visage bioscan (Millipore, Corp.) in the linear range of the film.
Glycosidase Digestion-Fifty g of membrane protein or radiolabeled IR immunoprecipitated from 1.4 -1.6 mg of membrane protein (see below) was digested with N-glycosidase F or endoglycosidase H. Specifically, samples were incubated in denaturing buffer (0.5% SDS and 1% ␤-mercaptoethanol) for 10 min at 100°C. For N-glycosidase F digestion, samples were brought to 50 mM sodium phosphate, pH 7.5, and 1% Nonidet P-40. For endoglycosidase H digestion, samples were brought to 50 mM sodium citrate, pH 5.5. Then, 1 l (500 units) of either N-glycosidase F or endoglycosidase H was added and incubated for 1-2 h at 37°C. Following incubation, sample dilution buffer containing 6 M urea was added and proteins separated on a 7.5% SDS-PAGE gel. Gels were dried for autoradiography (see below), or proteins in the gels were electrotransferred onto nitrocellulose for immunoblot analysis.
Metabolic Labeling-3T3-L1 adipocytes were incubated in 8 ml of complete medium or glucose-free medium for appropriate times. For labeling with [ 35 S]methionine/cysteine, cells were then incubated in 8 ml of methionine-and cysteine-free DMEM with or without 25 mM glucose for 1 h to deplete intracellular pools. The depletion medium was replaced with 2 ml of methionine-and cysteine-free DMEM containing 400 Ci of [ 35 S]methionine/cysteine with or without 25 mM glucose for specific times (10 -180 min). For labeling with [ 3 H]mannose, glucosedeprived cells were placed in 2 ml of glucose-free DMEM containing 420 Ci of [ 3 H]mannose for 3 h. For chase periods, the labeling medium was replaced with 8 ml of complete medium or glucose-free medium for the indicated times.
Immunoprecipitation-Cells were collected in TES with 1 mM phenylmethylsulfonyl fluoride and homogenized with 20 strokes using a motor-driven pestle. Total membrane fractions were collected by centrifugation at 212,000 ϫ g for 1 h, and the final pellets were resuspended in 300 l of TES with 1 mM phenylmethylsulfonyl fluoride. Membrane protein (1.4 -1.6 mg) was solubilized in 1 ml of radioimmune precipitation buffer (150 mM NaCl, 9.1 mM Na 2 HPO 4 , and 1.7 mM NaH 2 PO 4 , 0.5% deoxycholate, 1% Nonidet P-40, and 0.1% SDS, pH 7.4) with 1 mM phenylmethylsulfonyl fluoride. Extracts were then precleared with 25 l of a 50% slurry of protein A-Sepharose and 5 l of preimmune serum for 1 h at 4°C with rotation. Protein A-Sepharose was collected by brief centrifugation, and the supernatant was transferred to a new centrifuge tube. This process was repeated. The final supernatant was transferred to a new tube with 5 g of anti-IR␤ antibody and incubated overnight at 4°C with rotation. The immune complex was collected with protein A-Sepharose by incubation for 2 h followed by brief centrifugation. The immunoprecipitated complex was washed twice with 1 ml of radioimmune precipitation buffer, four times with 1 ml of radioimmune precipitation with 1 M NaCl, and once with 1 ml of TES. Immunoprecipitates were released in 50 l of sample dilution buffer containing 6 M urea, boiled for 5 min, and run on a 7.5% SDS-PAGE gel overnight at 50 V. For fluorography, the gels were fixed in 10% trichloroacetic acid, 40% MeOH for 30 min, soaked in water for 30 min, and then soaked in 1 M sodium salicylate for 1 h before drying at 60°C under vacuum. Dried gels were exposed to Amersham Pharmacia Biotech Hyperfilm for 1-3 days. Densitometry was performed in the linear range of the film.
Two-dimensional Gel Analysis-IR or GRP78 was immunoprecipitated as above. IR and GRP78 were released from the protein A-Sepharose⅐IgG complex with 100 l of 0.1 M glycine, pH 2.8, containing 0.5% Triton X-100 for 30 min at room temperature. The samples were neutralized with 10 l of 1 M Tris-base. The IgG⅐protein A-Sepharose complex was removed by centrifugation. The eluates (100 l) were transferred to glass tubes containing 100 l of isoelectric focusing sample solution (6.4% (w/v) Nonidet P-40 and 6.5 mM dithiothreitol). These samples were then mixed, with 4% acrylamide containing 9 M urea, 2% Nonidet P-40, and 2% carrier ampholytes (pH 3.5-10) and were polymerized for 1 h in glass rods (13 cm ϫ 3.4 mm inner diameter). The first dimension (isoelectric focusing) was run at 350 V for 18 h, followed by 800 V for 2.5 h according to procedures by Semple-Rowland et al. (29). The rod gels were removed from the glass tubes and equilibrated in isoelectric focusing equilibration buffer (5% ␤-mercaptoethanol, 62.5 mM Tris-HCl (pH 6.8), 2.3% SDS, and 10% glycerol) for 30 min before loading onto a 7.5% SDS-PAGE for protein separation by mass (second dimension). The slab gels were then dried and exposed to film for autoradiography.

Subcellular Localization of the Insulin Receptor in 3T3-L1
Adipocytes-Using a subfractionation procedure developed in our laboratory (26), we examined the distribution of the insulin receptor in control (F) adipocytes. Western blot analysis using a polyclonal antibody against the IR ␤-subunit revealed two forms of the receptor: the proreceptor, which migrated as a 190-kDa protein, and the mature receptor, represented by the ␤-subunit, which migrated as a 95-kDa protein (Fig. 1, upper  panel). Both the proreceptor and mature receptor were found in the HDM fraction, which we have shown is composed of both ER and Golgi membranes (26). Only mature receptor was localized to the PM fraction, which represented about 70% of the total IR pool (Table I). No IR was found in the LDM fraction, which we and others have shown contains the translocatable pool of GLUT4 (30 -32). In glucose-deprived cells (S), the proreceptor migrated as a 170-kDa species (Fig. 1, upper panel). This "alternative" proreceptor represented nearly 25% of the total IR pool, which was 6-fold more than the normal proreceptor in control cells (Table I). Mature receptor was noted in both the HDM and PM fraction of glucose-deprived cells, although expression was somewhat reduced relative to controls.
We also examined the expression of GRP78, an ER molecular chaperone. Confirming our earlier work (26,33), we show that GRP78 is found predominantly in the HDM fraction and significantly elevated in glucose-deprived cells. The apparent localization of GRP78 to the PM is due to the 10% contamination of ER membranes, an artifact of subfractionation in both control and glucose-deprived cells (26).
Time-dependent Accumulation of the Alternative Proreceptor-In glucose-fed cells, the level of normal proreceptor and ␤-subunit remained relatively constant over time (Fig. 2, A and  B). In contrast, glucose deprivation led to the appearance of the alternative proreceptor by 12 h with continued accumulation through 24 h. The 24-and 48-h time points showed nearly identical accumulation denoting a new steady state. With the increase in the alternative proreceptor, a decrease in the level of IR␤ in both the HDM and PM fractions was noted. By 48 h of glucose deprivation, the alternative proreceptor represented the major insulin receptor species, suggesting that the alternative proreceptor is not processed but is rather retained in the ER compartment.
Characterization of the Oligosaccharide Structure on the Insulin Receptor-Because the alternative proreceptor migrated more rapidly than did the normal proreceptor, we tested whether the alternative proreceptor was glycosylated. Cells treated with N-glycosidase F, which removes all N-linked sugars, resulted in an aglyco-proreceptor species that migrated identically whether from glucose-fed or glucose-deprived cells (compare lanes 1 and 4 in the upper panel of Fig. 3). While we expected that N-glycosidase F treatment of the normal proreceptor would result in accelerated migration, we were surprised that the same occurred in glucose-deprived cells. This implies the attachment of a unique N-linked oligosaccharide core structure to the alternative proreceptor. This is supported by the fact that the alternative proreceptor was resistant to  endo H treatment (lane 5), which is in contrast to the sensitivity of the normal proreceptor to endo H (lane 2). Sensitivity to endo H treatment indicates that the normal proreceptor contains high mannose oligosaccharides, which is consistent with earlier work (11). The ␤-subunit also showed sensitivity to endo H, indicating that at least some of the N-linked sites remain as high mannose structures. Note that the ␤-subunit behaves similarly in control and glucose-deprived samples. As shown in Figs. 1 and 2, the steady state level of the mature receptor changes by only 10% at 24 h of deprivation and derives from normal proreceptor.

Synthesis of Metabolically Labeled Insulin
Receptor-To confirm that the alternative proreceptor is not processed with glucose deprivation, glucose-fed and glucose-deprived cells were metabolically labeled with [ 35 S]methionine/cysteine followed by immunoprecipitation (Fig. 4A). Densitometric analysis of these data is shown in Fig. 4B. In both the glucose-fed and glucose-deprived cells, the normal and alternative proreceptors, respectively, were visible by 10 min of labeling. The intensity of this labeling increased through the 3-h pulse for both the normal and alternative proreceptors. In control cells, both ␣and ␤-subunits became clearly visible by 1 h of pulse. This indicates that the normal proreceptor was processed into mature receptor. Over the 3-h pulse, no mature receptor was generated from the alternative proreceptor in glucose-deprived cells.
Insulin Receptor Turnover-Because the alternative proreceptor was not processed, we examined its rate of degradation. Glucose-fed or glucose-deprived cells were pulsed for 1 h and chased for specific times in the presence or absence of 25 mM glucose, as appropriate, before immunoprecipitating IR from total membrane fractions. As shown in Fig. 5A, the normal proreceptor was processed into the ␣and ␤-subunits by 1 h of chase and reached a maximum by 3 h. The half-life of the normal proreceptor was 1.3 Ϯ 0.6 h (n ϭ 3), as determined by first order regression of densitometric data. In contrast, the alternative proreceptor was not processed but rather degraded slowly with a half-life of 5.1 Ϯ 0.6 h (n ϭ 3).
Interaction of GRP78 with the Alternative Proreceptor-To determine if the alternative proreceptor interacts with GRP78, we performed co-precipitation assays. Fig. 6 compares those proteins that co-precipitate with IR in the fed and deprived states. In Fig. 6A, only three proteins were precipitated by the ␤-subunit antibody: the normal proreceptor, the ␣-subunit, and the ␤-subunit. Note that immunoprecipitation with anti-GRP78 antibodies identified no radiolabeled protein, consistent with the low synthetic rate of GRP78 observed by others in the   FIG. 3. Effect of glycosylation on IR migration. Cells were maintained in DMEM with (F) or without (S) 25 mM glucose for 24 h. Aliquots (50 g) of HDM fractions from glucose-fed and glucose-deprived cells were treated with N-glycosidase F or endo H before running on a 5% SDS-PAGE. Western analysis was performed using anti-IR␤ antibody. This experiment is representative of two independent experiments (n ϭ 2).

FIG. 4. Synthesis of IR.
A, cells were maintained in DMEM with (F) or without (S) 25 mM glucose for a total of 24 h. Within that 24-h period, cells were incubated for 1 h in methionine-and cysteinefree medium (with or without glucose) followed by incubation in 200 Ci/ml of [ 35 S]methionine/cysteine for 10, 30, 60, 120, or 180 min. At each time point, total membranes were collected for immunoprecipitation. Membranes (1.4 mg) were extracted and precleared twice before immunoprecipitating with anti-IR␤ antibody as described under "Experimental Procedures." Immunoprecipitates were resolved by SDS-PAGE followed by autoradiography. B, densitometric analysis of IR synthesis is represented as the percentage of maximum arbitrary units over the pulse time. These data are representative of two independent experiments (n ϭ 2). q, normal (Fed) or alternative (Starved) proreceptor; f, IR␣; OE, IR␤. glucose-fed state (34 -36). With glucose-deprived cells, the ␤-subunit antibody precipitated not only the alternative proreceptor but also a band that migrated as the 72-kDa protein (Fig. 6B). GRP78, immunoprecipitated from an identical sample, migrated in identical fashion as a 72-kDa protein. Note that GRP78 co-precipitated a number of proteins, including a faint band that migrated like the alternative proreceptor.
To further analyze the specificity of interaction, we took advantage of the ATP-dependence of GRP78 binding (37)(38)(39). Fig. 7 shows that 50 M ATP was sufficient to release GRP78 from the IR immune complex collected from glucose-deprived cells. Neither high salt wash (1 M NaCl) nor ADP treatment was able to dissociate the 72-kDa protein from the IR (data not shown).
Finally, we used the isoelectric point of GRP78 to verify the identity of the 72-kDa protein. Shown in Fig. 8 are two-dimensional gels that compare the migration pattern of the 72-kDa protein (Fig. 8A) with that of authentic GRP78 (Fig. 8B). Antibodies against GRP78 precipitated two forms of the protein. It is likely that one of these isoforms is ADP-ribosylated and/or phosphorylated, which causes homodimerization with itself (40). Only in the unmodified state does GRP78 dissociate from itself to interact with misfolded proteins. Anti-␤-subunit antibodies co-precipitated only one form of GRP78 (Fig. 8A) that aligned with the more basic, unmodified isoform of GRP78 (Fig.  8B). Together, data from Figs. 6 -8 provide strong support for The IR immune complex was treated without (control) or with 50 M ATP. The immune complex was collected and separated from the eluate. These were separately resolved by SDS-PAGE, followed by autoradiography. This experiment is representative of two independent experiments (n ϭ 2). the interaction of the alternative proreceptor with GRP78. 2 Restoration of Processing of the Alternative Proreceptor-To determine if the alternative proreceptor could be salvaged from the degradation path, glucose-deprived cells were refed with medium containing 25 mM glucose for specific times (Fig. 9). Interestingly, by 1 h of glucose readdition, the alternative proreceptor shifted to a higher molecular weight isoform. Note that the processed alternative proreceptor migrated between the normal and alternative proreceptor. The appearance of this "intermediate" proreceptor suggests the attachment of additional oligosaccharides to the alternative proreceptor (see below). Also note that with glucose refeeding, the interaction with GRP78 was significantly reduced. Surprisingly, 3 h after glucose readdition, only faint traces of ␣and ␤-subunits were visible. Note that the film was overexposed in order to visualize the ␣and ␤-subunits (also see Fig. 10A). The intensity of these bands never approached that of the processed normal proreceptor (see Fig. 5). Rather, most of the intermediate proreceptor disappeared between 3 and 6 h. This implies that glucose readdition not only promotes post-translational attachment of normal N-linked oligosaccharides to the alternative proreceptor but also accelerates proreceptor degradation.
To show that the shift in migration of the alternative proreceptor is due to oligosaccharide processing, sensitivity to endo H was tested (Fig. 10A). Whereas the alternative proreceptor was resistant to endo H treatment (compare lanes 3 and 4), the intermediate proreceptor was sensitive to endo H (compare lanes 6 and 8). The newly acquired sensitivity to endo H infers either the addition of mannose residues to pre-existing structures or cleavage of the pre-existing structures and/or the attachment of newly synthesized oligosaccharide chains. To test the second hypothesis, tunicamycin was added during glucose refeeding (Fig. 10B). We have previously shown that it takes 3 h in these cells to observe the effects of tunicamycin (33). Therefore, we pulsed for 1 h as before and then chased in the absence of glucose with or without tunicamycin for 3 h before refeeding with glucose. Tunicamycin had no effect on the alternative proreceptor maintained in the absence of glucose ( lanes  1 and 2). In contrast, tunicamycin prevented the glucose-induced shift to the intermediate form (lanes 3 and 4). This suggests that the reprocessing of the alternative proreceptor requires N-linked glycosylation and thus attachment of newly synthesized core structures to unoccupied N-linked glycosyla-tion sites. Note once again that significantly less GRP78 coprecipitates with the intermediate form of the proreceptor.
Finally, we tested whether the oligosaccharides attached to the alternative proreceptor during glucose deprivation are retained in the intermediate form. For this study, we used [ 3 H] mannose as the oligosaccharide tag. Shown in Fig. 11

DISCUSSION
Glycosylation inhibitors and site-directed mutagenesis of Nlinked glycosylation sites have provided evidence for an important role of glycosylation in IR processing and function. We have extended these studies by using glucose deprivation for the first time as a tool to monitor reversible changes in IR glycosylation. Our data show that glucose deprivation alters N-linked glycosylation of the insulin proreceptor. This causes accumulation of an alternative glycoform of the proreceptor in an intracellular membrane compartment where it binds GRP78, consistent with ER localization. Although several investigators have shown that overexpressed or mutant insulin receptor interacts with GRP78 (23,24), we are the first to show that IR interacts with GRP78 in a more physiological setting.
We also show that the rate of synthesis of the alternative proreceptor appears similar to that of the normal proreceptor. However, the alternative proreceptor disappears more slowly than the normal proreceptor, suggesting that ER mediated degradation, perhaps directed by its interaction with GRP78. Processing of the alternative proreceptor can be stimulated with the readdition of glucose and requires N-linked glycosylation. This also leads to a rapid loss of proreceptor from the ER. This loss appears to represent accelerated degradation with only a minor maturation component. At present, we do not understand this surprising result.
Rearick et al. (42) have shown that glucose deprivation of Chinese hamster ovary cells leads to the attachment of an immature oligosaccharide structure at N-linked consensus sites. This alternative oligosaccharide structure (Man 5 GlcNAc 2 ) is resistant to endo H digestion, whereas the normal N-linked oligosaccharide core (Glc 3 Man 9 GlcNAc 2 ) is cleaved by endo H. Our data show that an endo H-resistant oligosaccharide is attached to the proreceptor of 3T3-L1 adipocytes during glu-cose deprivation, which is consistent with a Man 5 GlcNAc 2 structure. Previously, we have shown that glucose deprivation leads to the synthesis of an alternatively glycosylated form of the constitutive glucose transporter, GLUT1 (28). In contrast to IR, the alternative form of GLUT1 is targeted to the plasma membrane. 3 Thus, IR but not GLUT1 requires the normal N-linked oligosaccharide structure for processing to PM. Of course there are major differences between insulin receptors and GLUT1. Since the IR contains multiple glycosylation sites in contrast to the single site on GLUT1, a major question arises as to whether underglycosylation, glycosylation with a truncated oligosaccharide, or a combination of both influence IR processing. Our data demonstrate that glucose deprivation results in both underglycosylation and attachment of truncated oligosaccharides, making these possibilities difficult to distinguish. Given that the first four N-linked glycosylation sites significantly influence processing (16,17), it will be important to determine the consequence of glucose deprivation at these sites.
Glucose deprivation of 3T3-L1 adipocytes also leads to insulin resistance (43). Yet, Tordjman et al. (44) have shown that glucose deprivation does not alter the level of the GLUT4 glucose transporter. Thus, our data explain, in part, the development of insulin resistance by the reduction of cell surface insulin receptor as a result of retention and degradation of the proreceptor in the ER. In this way, our glucose deprivation 3  Cells were glucose-fed (F) or glucose-deprived (S) and labeled for 1 h. A, glucosedeprived cells were chased in complete medium for 1 or 3 h. At each time point, total membranes were collected for immunoprecipitation with anti-IR␤ antibody. The immune complexes were then treated with or without endo H before resolving on SDS-PAGE. B, glucose-deprived cells were chased in medium without glucose in the presence or absence of tunicamycin for 3 h (lanes 1 and 2). The glucose-free medium was then replaced with complete medium with or without tunicamycin for 1 h (lanes 3 and 4). IR was immunoprecipitated and resolved by SDS-PAGE. model resembles carbohydrate-deficient glycoprotein syndrome 1, which is caused by a genetic deficiency of phosphomannomutase (45); proteins that require glycosylation for processing, like the insulin receptor, are retained in the ER by global depletion of substrate for protein glycosylation.
Aberrant glycosylation of the insulin receptor is not the only cause for its retention in the ER. Several different types of mutations have been identified in the insulin receptor of patients with genetic forms of insulin resistance. Three specific point mutations impair transport of the mutant receptors to the cell surface (46 -49). Accili et al. (23) have reconstructed these mutants (N15K, H209R, and F382V) in vitro and overexpressed them in NIH 3T3 fibroblasts. Of these, the Arg 209 and Val 382 mutants were found in a complex with GRP78. The Lys 15 mutant did not interact with GRP78, despite its retention in the ER.
One of the advantages of glucose deprivation over inhibitors of glycosylation and mutation is its reversibility, which allowed us to discover the novel processing of the alternative proreceptor to the intermediate form. Based on our current data, we present the following model to explain the effects of glucose deprivation and refeeding on insulin receptor processing (Fig.  12). The model predicts the utilization of only a limited number of the 18 possible N-linked glycosylation sites during glucose deprivation because of the small difference in the migration of the alternative versus the aglyco-proreceptor. Substantial underglycosylation and/or attachment of truncated oligosaccharides leads to the interaction with GRP78, retention in the ER, and targeted degradation. Glucose refeeding permits posttranslational processing, which requires N-linked glycosylation of a limited number of unoccupied sites. Newly acquired endo H sensitivity predicts the attachment of normal Glc 3 Man 9 -GlcNAc 2 core structures to these sites. Some sites may not be accessible to oligosaccharide transferases due to protein folding or spatial orientation. While N-glycosidase F activity has been noted in a number of systems (15, 50 -52), glucose-induced reprocessing does not require removal of the truncated oligosaccharide(s). Despite the retention of the original truncated oligosaccharides, the intermediate species is released from GRP78. This allows a small fraction of the intermediate proreceptor to be processed to mature ␣and ␤-subunits, while the majority undergoes accelerated degradation. We are currently pursuing experiments to further refine the model.