Effects of Inhibitors of N-Linked Oligosaccharide Processing on the Biosynthesis and Function of Insulin and Insulin-like Growth Factor-I Receptors*

We have used specific inhibitors of oligosaccharide processing enzymes as probes to determine the involvement of oligosaccharide residues in the biosynthesis and function of insulin and insulin-like growth factor-I receptors. In a previous study (Duronio, V., Jacobs, S., and Cuatrecasas, P. (1986) J. Bioi. Chem. 261, 970-976) swainsonine was used to inhibit mannosidase 11, resulting in the production of receptors containing only hybrid-type oligosaccharides. These receptors had a slightly lower molecular weight and were much more sensitive to endoglycosidase H, but otherwise behaved identically to normal receptors. In this study, we used two compounds that inhibit oligosaccharide processing at earlier steps: (i) N-methyl-l-deoxynojirimycin (MedJN), which inhibits glucosi- dases I and I1 and yields glucosylated, high mannose oligosaccharides, and (ii) manno- 1-deoxynojirimycin (MandJN), which inhibits mannosidase I and yields high mannose oligosaccharides. In the presence of MandJN, HepGZ cells synthesized receptors of lower molecular weight, which were cleaved into a and 8 subunits and were able to bind hormone and autophosphorylate. These receptors were as sensitive to endoglycosidase

as sensitive to endoglycosidase H as receptors made in the presence of swainsonine. In the presence of MedJN, receptors of only slightly lower molecular weight than normal were synthesized and were shown to contain some glucosylated high mannose oligosaccharides. These receptors were able to bind hormone and retained hormone-sensitive autophosphorylation activity. In both cases, the incompletely processed receptors could be detected at the cell surface by cross-linking of iodinated hormone and susceptibility to trypsin digestion, although less receptor was present in cells treated with MedJN. Studies of receptor synthesis using pulsechase labeling showed that the receptor precursors synthesized in the presence of MedJN were cleaved into CY and 8 subunits at a slower rate than normal receptors or those made in the presence of MandJN. Inhibition of oligosaccharide processing had no effect on the association of the receptor subunits into disulfide-linked oligomeric complexes. The size of the a and / 3 subunits are approximately 130-135,000 and 90-95,000, respectively. Although the two receptors have similar structures, they are not identical. They can be distinguished by their differential binding affinity for the two hormones (13,14), by specific monoclonal antibodies (15,16), and by slightly different mobilities on SDS-polyacrylamide gels (17,18). The cDNA sequences of the precursor polypeptides for the two receptors (19)(20)(21) verify that they are synthesized as single polypeptides that are proteolytically cleaved to yield the a and 0 subunits (17,22). The extent of homology between the two receptors has also been established (50-60%), with many of the important features of the receptors such as the cysteine-rich regions and tyrosine kinase domains being highly conserved. Both subunits of the receptors are glycosylated (18,23), with the oligosaccharides accounting for approximately 25% of the molecular weight of the mature receptors (18). Pulsechase studies have shown that the earliest polypeptide synthesized that reacts with anti-insulin receptor antibodies has a molecular weight of approximately 190,000 and contains high mannose N-linked oligosaccharides (22,(24)(25)(26). Processing of the oligosaccharides on this polypeptide yields a higher molecular weight precursor (M, = 200-210,000), which contains primarily complex type oligosaccharides. Studies with tunicamycin have demonstrated that glycosylation is necessary for production of a functional insulin receptor (24,27). Monensin, which has been used to block processing of glycoproteins, was shown to cause an accumulation of receptor precursors, only a portion of which were able to escape this block to reach the plasma membrane as incompletely processed receptors (17).
In the standard pathway of N-linked glycosylation of nascent glycoproteins, the oligosaccharide that is transferred to asparagine residues from a dolichol precursor is Glc3Man,GlcNAc2 (for review, see Ref. 28). The glucose residues are first removed by glucosidases, then mannose residues are removed by rough endoplasmic reticulum and Golgi mannosidases to form Man,GlcNAcp. A GlcNAc residue is then added before two additional mannose residues are removed The abbreviations used are: IGF-I, insulin-like growth factor-I; MedJN, N-methyl-1-deoxynojirimycin; MandJN, manno-l-deoxynojirimycin; dJN, 1-deoxynojirimycin; PBS, phosphate-buffered saline; Endo H, endoglycosidase H; TPCK, L-l-tosylamido-2-phenylethyl chloromethyl ketone; SDS, sodium dodecyl sulfate. by Golgi mannosidase 11. Following this step, complex oligosaccharides are formed by addition of GlcNAc, Gal, and sialic acid. The availability of specific inhibitors of the enzymes involved in the processing steps (30, 31) has proven valuable in determining the importance of oligosaccharides in various glycoproteins. N-Methyl-1-deoxynojirimycin (MedJN) inhibits glucosidases I and 11, and manno-1-deoxynojirimycin (MandJN) inhibits mannosidase I. MandJN inhibition prevents trimming of mannose residues, thereby yielding high mannose oligosaccharides (42). MedJN inhibition prevents the trimming of glucose residues and yields high mannose oligosaccharides that are glucosylated (43). The latter inhibitor is more specific than 1-deoxynojirimycin, which is also a glucosidase inhibitor, since it does not affect glucosylation of the lipid-linked oligosaccharide precursor (32). Recent studies have suggested that non-glucosylated oligosaccharides can also be transferred to proteins in a mammalian cell line (29). Processing of non-glucosylated residues would not be inhibited by MedJN, and transfer of such residues to nascent glycoproteins may partly account for the incomplete inhibition of complex oligosaccharide formation seen with this compound (29).
To examine further the role of oligosaccharides in the processing and function of insulin and IGF-I receptors, we undertook a series of studies using specific inhibitors of processing enzymes. We demonstrated recently that inhibition of the terminal steps of processing by use of the mannosidase I1 inhibitor swainsonine resulted in receptors containing hybrid type oligosaccharides that appeared normal by other criteria (18). In this study, we used the two processing inhibitors mentioned above, MedJN and MandJN. These studies show that in the presence of MandJN the receptors can be synthesized and appear at the cell surface, maintaining their ability to bind hormone and to autophosphorylate. In the presence of MedJN, receptor precursors containing glucosylated high mannose structures are processed more slowly into a and p subunits. As a result, there are fewer receptors at the cell surface, but the altered receptors are able to bind hormone and retain autophosphorylation activity. Therefore, receptors containing only high mannose oligosaccharides retain normal function, while receptor precursors containing glucosylated high mannose oligosaccharides are processed into a and p subunits at a slower rate.
Cell Culture and Metabolic Labeling-HepG2 cells were grown in minimal essential medium (Gibco) supplemented with 10% fetal bovine serum (Hyclone). Cells were seeded in 35-or 60-mm dishes and used 2-4 days after plating, when the cells covered 60-80% of the plates. For [36S]methionine labeling, cells were washed twice with methionine-free Dulbecco's modified Eagle's medium and incubated in the same medium supplemented with 0.25% fetal bovine serum and 0.2-0.3 mCi of [35S]methionine in a final volume of 800 pl. When inhibitors were used, cells were preincubated for 30-60 min before addition of labeled methionine, Labeling was done overnight, for 15-18 h, except where stated otherwise. In pulse-chase studies, cells were grown in 24-well plates. Following preincubation in the presence or absence of inhibitors, 0.1 mCi of (35S]methionine was added (in 300 p1 of medium), and the cells were labeled for 30 min. Labeling medium was removed, and cells were incubated in normal medium containing 10% serum without inhibitors. For [3H]mannose labeling, cells in 60-mm dishes were washed and incubated in minimum essential medium containing 0.25% fetal bovine serum and 2 mM fucose. Labeled mannose was added at a final concentration of 1 mCi/ml and labeling continued for 24 h. When unlabeled cells were incubated with inhibitors, the compounds were added into minimum essential medium containing 1% fetal bovine serum.
Solubilization of HepG2 Cells and Partial Purification of Receptors on Concanavalin A-Sepharose Affinity Columns-HepG2 cells were washed four times with cold PBS and solubilized in 50 mM Tris-C1 buffer, pH 7.7, containing 1% Triton X-100, 1 mg/ml bacitracin, and 40 pg/ml phenylmethylsulfonyl fluoride (1 ml of buffer was used for cells grown in 35-mm dishes and 300 pl was used for cells in 24-well plates). Following incubation on ice for 30 min, the solubilized cells were centrifuged at 100,000 X g for 30 min. Where indicated, the supernatant was loaded into concanavalin A-Sepharose affinity columns and eluted as described previously (18).
Immunoprecipitation of Receptors-Insulin and IGF-I receptors were immunoprecipitated essentially as described previously (18), except that in most cases this was done without prior purification of samples on concanavalin A-Sepharose. Briefly, samples were first made up to 0.5% Triton X-100 and 0.5 M NaCl. The appropriate monoclonal antibody (a-IR-1 for insulin receptors or a-IR-3 for IGF-I receptors) was added at an appropriate dilution, followed by control mouse serum at a dilution of 1:400. After 2 h on ice, goat anti-mouse immunoglobulin was added and allowed to incubate overnight at 4 'C. Immunoprecipitates were spun down at 3000 X g for 20 min and washed twice with 50 mM Tris-C1 buffer, pH 7.7, containing 0.5% Triton X-100 and 0.5 M NaCl and once with buffer containing only 0.2% Triton X-100. The specificity of the two monoclonal antibodies for the receptor polypeptides was demonstrated previously (17,18).
'251-Insulin Cross-linkiv-Cross-linking of iodinated hormone to its receptor was carried out using disuccinimidyl suberate as described elsewhere (17,18). When using intact cells to label cell surface receptors, cells were first released from dishes by incubation in PBS containing 1 mM EDTA. Cells were washed once with PBS and incubated at 15 "C in 1 ml PBS containing 0.1 mg/ml bovine serum albumin. Iodinated hormone was added and allowed to incubate with the cells for 30 min. Cells were washed once, then transferred to 0 "C, and the cross-linking agent was added at a final concentration of 0.1 mg/ml. The incubation continued for 30 min and was terminated by adding 25 pl of 1.0 M NH,Cl. Cells were washed once with PBS, solubilized, and receptors were immunoprecipitated.
Trypsinization of Cell Surface Receptors-The trypsin sensitivity of receptors at the cell surface of HepG2 cells was demonstrated by washing labeled cells with PBS at room temperature and incubating for 10 min in PBS containing 0.5 mg/ml TPCK-treated trypsin. Cold PBS containing 0.5 mg/ml soybean trypsin inhibitor and 40 pg/ml phenylmethylsulfonyl fluoride was used to stop the trypsinization and to wash the cells once. Cells were solubilized and receptors immunoprecipitated as indicated above in buffers that also contained soybean trypsin inhibitor.
Preparation of Glycopeptides and Fractionation of Oligosaccharide (41)-Following [3H]mannose labeling of cells in the presence or absence of MedJN, receptors were immunoprecipitated and exhaustively digested with Pronase. The glycopeptides obtained by gel filtration on Bio-Gel P-6 were then treated with Endo H as described previously (41) and fractionated on the same column of Bio-Gel P-6. The Endo H-sensitive oligosaccharides were pooled, concentrated, and chromatographed with 14C-labeled glucosylated oligosaccharide standards on a column (25 X 0.46 cm) of 5-pm particle size Aminospherisorb (Phase Separations; packed by Chromatography Science Co., Ville Mont Royal, Quebec, Canada) with a Varian Model 5000 liquid chromatograph. Filtered samples (100 pl) dissolved in acetonitrile/water (5:7, v/v) were applied to the column equilibrated previously with the starting solvent. Elution was performed isocratically at 1 ml/min for 20 min with acetonitrile/water (3:2, v/v) followed by a linear gradient to acetonitrile/water (21:29, v/v) for 95 min. The water contained 15 mM potassium phosphate adjusted to pH 5. Fractions (1 ml) were collected and assayed for radioactivity.
Other Procedures-Autophosphorylation of receptors, digestion with endoglycosidase H, and SDS-polyacrylamide gel electrophoresis were carried out as described previously (18), with minor changes of reaction volumes. In previous work with IM-9 cells (18), larger cell numbers and incubation volumes could be used. Due to the scarcity of inhibitors available for the studies reported here and their use at Oligosaccharides of Insulin and IGF-I Receptors relatively high concentration (1-5 mM here, compared with 10 p M with swainsonine), all procedures were carried out with much less material to minimize incubation volumes. In most cases, samples were electrophoresed on 6.5% polyacrylamide SDS gels. In figures showing autoradiograms of dried gels, numbers at the sides of the figures indicate the M, (X of the bands of interest. following Endo H digestion (lanes 2 and 5 ) , suggesting that both precursors contained high mannose oligosaccharides. The molecular weights of the a and subunits of the insulin receptor synthesized in the presence of MandJN were lower than the control (118,000 and 89,000, respectively). As expected, the control a and p subunits were only slightly susceptible to Endo H digestion, but in the MandJN-treated sample, the subunits were very susceptible to Endo H, resulting in digested polypeptides with molecular weights of 86,000 and 80,000.

Biosynthesis of Insulin Receptors in the Presence
In cells treated with MedJN (lane 10) the molecular weight of the receptor precursor was approximately 200,000. This MandJN (lanes 4-6) or 5 mM MedJN (lanes [10][11][12] contain only high mannose oligosaccharides. Therefore, it seems that inhibition of glucosidases by MedJN only partially inhibited processing of the oligosaccharide residues. MedJN also seemed to cause an accumulation of receptor precursor, and a decrease in the amount of a and / 3 subunits compared to controls. As will be seen below, MandJN and MedJN had similar effects on the biosynthesis of IGF-I receptors.
To characterize further the oligosaccharides on the insulin receptors synthesized in the presence or absence of MedJN, samples were analyzed after being labeled with [ 3 H ] m a n n~~e (Figs. 2 and 3). The glycopeptides obtained after exhaustive Pronase digestion of immunoprecipitated receptors were fractionated on Bio-Gel P-6 before (Fig. 2, A and B ) and after (Fig. 2, C and D ) Endo H treatment to distinguish between glycopeptides containing complex and high mannose oligosaccharides. There was an approximately 50% decrease in labeled glycopeptides containing complex-type (Fig. 2, K, These results are consistent with the partial Endo H sensitivity of 35S-labeled a and p subunits (Fig. 1, lane 1 1 ) . Endo Hsensitive oligosaccharides were pooled from the Bio-Gel P-6 columns and fractionated by high performance liquid chromatography (Fig. 3). It is clear that MedJN-treated samples (Fig. 3B) contained a large proportion of glucosylated oligosaccharides, including Glc3Man9GlcNAc, which were not present in the control sample (Fig. 3A). From comparison with glucosylated oligosaccharide standards and previous experience, the other two major peak fractions in the MedJN sample were probably Glc3MansGlcNAc and Glc3Man7GlcNAc as described previously (41). The presence of glucosylated high mannose oligosaccharides is consistent with the accumulation of the higher molecular weight precursor in MedJN-treated  At the earliest time shown (1-h chase after 30 min labeling) the major band in all cases was the receptor precursor, with very little of the CY and / 3 subunits. The loss of label from the receptor precursor coincided with increased labeling of the a and p subunits, as expected. It should be pointed out that the CY subunit was not labeled as well as the / 3 subunit because in both receptors the p subunit contains approximately twice as many methionine residues as the a subunit. A similar labeling pattern was seen in both the insulin and IGF-I receptors. One difference is that in the IGF-I receptor samples, a faint band appeared above the major labeled receptor precursor band which reached maximum intensity after 2 h of chase. This precursor contains primarily processed complex type oligosaccharide residues since its mobility was not altered by Endo H (data not shown). This was not seen at all in the MandJNtreated cells, because inhibition of the mannosidase prevents further oligosaccharide processing. A higher molecular weight precursor was sometimes seen in insulin receptor samples labeled for longer times, but not in the pulse-chase experiment; this indicates that the processing of high mannose oligosaccharides to complex-type is closely associated with proteolytic cleavage of the receptor precursors, particularly in the case of the insulin receptor.
The processing of the receptor precursor to CY and p subunits occurred at approximately the same rate in the control and MandJN-treated cells. In MedJN-treated cells, it occurred at a much slower rate. In control and MandJN-treated cells, 25-30% of the precursors were present a t 3 h of chase, taking the 1-h time as 100%. Approximately 10% of the precursor was left after 6 h of chase. In comparison, MedJN-treated cells retained 50-60% of the precursor after 3 h, and approximately 30% after 6 h. These results provide an explanation for results shown above in which cells labeled overnight in the presence of MedJN had more receptor precursor and less a and / 3 subunits than control or ManJN-treated cells.
Cell Surface Appearance of Receptors-Because the above samples were immunoprecipitated from solubilized whole cells, there is no indication whether the altered receptors are able to reach the plasma membrane. Two methods were used to demonstrate the presence of receptors at the cell surface. First, '251-insulin was cross-linked to receptors on intact cells that had been incubated for 48 h with MedJN or MandJN or without inhibitors (Fig. 6). As expected, the a subunit of the control receptor had a molecular weight of 132,000 that, when digested with Endo H, migrated with a molecular weight of 128,000 (not shown). In cells treated with MandJN, the labeled a subunit had a molecular weight of 118,000 and was much more sensitive to Endo H (not shown), as was seen in the [35S]methionine-labeled cells. These results show that MandJN did not significantly affect the ability of the receptors to reach the cell surface and bind hormone. In cells treated with MedJN, the receptor labeled by '251-insulin crosslinking had a slightly lower molecular weight than in the control sample and was more sensitive to Endo H digestion than the control receptor (not shown), indicating that the oligosaccharide composition was altered by the presence of MedJN. There was less receptor labeled in the cells treated with MedJN, which is in agreement with the 35S-labeling experiments shown above. Labeling of the IGF-I receptor by lZ5I-IGF-I cross-linking gave very similar results (not shown).
Trypsin sensitivity was also used to determine the cell surface appearance of insulin and IGF-I receptors (Fig. 7 ) .  MandJN (Man, lane 3). '251-Insulin was cross-linked to cell surface insulin receptors which were then immunoprecipitated and electrophoresed on an SDS-polyacrylamide gel.

CON Me Man
Cells preincubated in the presence or absence of inhibitors were labeled with [35S]methionine. They were then directly solubilized, or first treated with TPCK-trypsin and then solubilized, followed by immunoprecipitation (Fig. 7 , A and  B ) . It is clear that in samples treated with trypsin, almost all of the a subunits of the receptors were digested, suggesting that they were present at the cell surface. The major receptor precursor forms were relatively insensitive to trypsin, suggesting that they were located inside the cell. These precursors were high mannose-containing polypeptides as shown by their sensitivity to Endo H (Fig. 1). The higher molecular weight precursor of the IGF-I receptor (Fig. 7 B ) was susceptible to trypsin. This precursor, which contains complex oligosaccharides, presumably represents a small amount of receptor precursor whose oligosaccharides are processed, but which escapes proteolytic processing and is translocated to the cell surface. Because the same amount of /3 subunit was present in trypsin treated samples as in untreated, this indicated that the receptors could be immunoprecipitated efficiently even after their a subunits were nicked by trypsin. The extensive disulfide bonds in the receptors probably preserve the epitopes recognized by the monoclonal antibodies. The a subunit of the insulin receptor was degraded by trypsin such that no major fragments could be seen on this gel (only polypeptides greater than 40,000 would be resolved). The a subunit of the IGF-I receptor was less extensively proteolyzed because a major fragment with a M, of approximately 100,000 was generated (which migrates between the a and /3 subunits). The p subunit of the insulin receptor was unaffected by trypsin but the /3 subunit of the IGF-I receptor, which always runs as a broad band on SDS gels and is often resolved into a doublet, was affected (Fig. 7 B ) . Interestingly, in control and MedJN-treated cells, trypsin seemed to selectively nick the upper component of the /3 subunit, causing it to migrate with the lower component, while in MandJN-treated cells the doublet was preserved following trypsin treatment. Thus, differences in glycosylation of the IGF-I receptor p subunit, which probably account for its migration as a doublet, may expose a trypsin-sensitive site.
To determine the rate of appearance of the receptors at the cell surface, cells were pulse labeled with [35S]methionine for 30 min following preincubation with or without inhibitors and then chased with cold methionine for various times and treated with or without trypsin (Fig. 7 , C and D). It is evident that essentially all of the a subunit was susceptible to trypsin as soon as it appeared; this suggests that oligosaccharide processing and proteolytic processing of the receptor precursor is followed by the rapid appearance of the a and p subunits at the cell surface. The receptor precursor containing high mannose oligosaccharides did not appear at the cell surface.
Receptor Autophosphorylation-To determine whether the insulin and IGF-I receptors made in the presence of the inhibitors could retain autophosphorylation activity, cells were preincubated for 48 h in the presence of inhibitors, solubilized, and receptors partially purified on a concanavalin A-Sepharose column. The eluate from this column was used in an autophosphorylation reaction, and the results are shown in Fig. 8. In cells preincubated with MandJN, autophosphorylation of the lower molecular weight p subunit was observed.
The decreased extent of labeling can be attributed to incomplete elution of the high mannose receptor from the affinity column, although in this case better elution was achieved using 0.75 M rather than 0.5 M a-methylmannoside, which is normally used. Samples from cells preincubated with MedJN also demonstrated hormone-sensitive autophosphorylation. The subunit from these cells had a slightly lower molecular  (lanes I, 3, and 5) or with (lanes 2, 4, and 6 ) H (lanes 2 and 4 ) . C, autophosphorylation of IGF-I receptors following incubation without (lanes 1,3, and 5) or with (lanes 2, 4, and 6 ) 2.5 pg/ml IGF-I.

IGF-I
weight as expected, but to verify that it was the altered receptor that was being autophosphorylated, the experiment was repeated and samples were Endo H-digested. Fig. 8B illustrates that the 8 subunit of the insulin receptor from cells treated with MedJN was more sensitive to Endo H than the control. IGF-I receptors synthesized in the presence of inhibitors also retained their IGF-I-dependent autophosphorylation activity (Fig. 8C). These experiments demonstrate that the defectively glycosylated receptors are able to bind hormone and to respond to the hormone by carrying out one of their normal functions, autophosphorylation. Formation of Oligomeric Receptor Complexes-Since it was shown recently that folding, oligomerization, and formation tor oligomer (Mr approximately 360,000), which was nnly slightly sensitive to Endo H, is an tetramer. A faint band below this, because of its relative abundance and its sensitivity to Endo H, is probably a dimer of the Endo Hsensitive a-p precursor. In MedJN-treated cells there are three major insulin receptor bands (Fig. 9A, lanes 2 and 5) and two IGF-I receptor bands (Fig. 9B, lanes 2 and 5). The highest band, which is only slightly sensitive to Endo H probably represents the tetramer. The lowest band, which is Endo H-sensitive, is probably the a-p precursor dimer. Consistent with the results in Figs. 1, 4, and 5, there is relatively more of this precursor in MedJN-treated cells.
The identity of the intermediate band of the insulin receptor of proper disulfide bonds (46,47) were necessary for intracel-is less certain. The migration of this oligomer is similar to the lular transport of newly synthesized glycoproteins, we exam-major band in MandJN-treated cells before and after Endo ined the effects of the processing inhibitors on the formation H, but it is not likely to be the high mannose (a@)z tetramer, of receptor complexes. Fig. 9, A and B, shows insulin and because Fig. 1 shows that no high mannose a and / 3 subunits IGF-I receptors immunoprecipitated from 35S-labeled cells were present in MedJN-treated cells. Another possibility sugincubated in the presence or absence of inhibitors and sub-gested by the intermediate migration and Endo H sensitivity jected to SDS-polyacrylamide gel electrophoresis in the ab-of the band is that it may be a hybrid structure composed of sence of reducing agents. In all cases, disulfide-linked oligomers were formed that represent heterotetramers of the (Y and p subunits or dimers of the nonproteolytically processed a-@ precursor. Therefore, MedJN and MandJN did not interfere with the formation of these disulfide-linked oligomers. By correlating the results in Fig. 9 with those in Figs. 1, 4, and 5, it is possible to determine the likely structure of these oligomers. In control cells, the major insulin and IGF-I recep-a processed (ab) half and a high mannose precursor half. Alternatively, this band may be a dimer of the a-p precursor having a different disulfide bonding pattern that alters its migration and Endo H sensitivity. Further work will be required to distinguish among these possibilities. The major oligomeric form of the receptors labeled in MandJN-treated cells is the tetramer of the Endo H-sensitive high mannose forms of the a and / 3 subunits.  C; panels A and B, lanes 1 and 4 ) , in the presence of 5 mM MedJN (Me; panels A  and B, lanes 2 and 5) or in the presence of 1 mM MandJN (Man; panels A and B, lanes 3 and 6 ) . To determine which of these oligomeric forms were expressed on the cell surface, intact cells were affinity crosslinked with '251-insulin and labeled proteins analyzed on nonreducing SDS gels (Fig. 9C). The only forms found on the surface were those corresponding to the CY^)^ tetramers. Similar results were obtained with '251-IGF-I labeling of IGF-I receptors (data not shown).

DISCUSSION
Insulin and IGF-I receptors contain complex N-linked oligosaccharides that are formed by an ordered sequence of modifying reactions. Receptors synthesized in the presence of tunicamycin, which blocks the initial step in the formation of N-linked oligosaccharides, cannot bind insulin, are not transported normally through various intracellular organelles to the cell surface, and are not processed proteolytically to form CY and p subunits (24, 27); this indicates that glycosylation is required for these processes. However, complete maturation of the oligosaccharides is not required for these processes, since they are not blocked by swainsonine (18, 40), a mannosidase I1 inhibitor, which functions later in the sequence of oligosaccharide chain maturation. Studies using monensin (17), a cationic ionophore that also blocks terminal steps of processing, also demonstrated that newly synthesized receptor lacking complex oligosaccharides acquired the ability to bind hormone. In the presence of monensin, receptors were not transported to the cell surface and proteolytically processed normally, but these are probably not consequences of incomplete oligosaccharide processing. They are more likely due to direct inhibitory effects of monensin.
In order to better localize the steps in oligosaccharide processing required for functional maturation and translocation of insulin and IGF-I receptors, we used two inhibitors that act at steps between those inhibited by tunicamycin and monensin. MedJN inhibits glucosidase I and I1 and thus the trimming of the outer glucose residues from the glucosylated high mannose oligosaccharide precursor. MandJN is an inhibitor of Golgi mannosidase I and therefore inhibits the trimming of mannose residues. Inhibition in both cases results in glycoproteins containing oligosaccharides of the high mannose type, with the major difference being the presence or absence of glucose residues.
Incubation of cells with MandJN resulted in the biosynthesis of receptors with much lower molecular weights containing high mannose oligosaccharides (Fig. 1). The molecular weights of the receptor subunits (118,000 and 89,000) are very Oligosaccharides of Insulin and IGF-I Receptors similar to those in cells treated with monensin (115,000 and 89,000; see Ref. 17), which would also be expected to have only high mannose structures. These receptors were proteolytically processed into a and /3 subunits much like normal receptors (Figs. 1,4, and 5). Their ability to bind insulin (Figs. 6 and 9C) or IGF-I (data not shown) was not altered, as indicated by cross-linking of iodinated hormone. The altered receptors also retained their hormone-sensitive autophosphorylation activity (Fig. 8). Appearance of the receptors at the cell surface was also not affected, as indicated by cell surface labeling of receptors (Fig. 6) and their sensitivity to trypsin digestion (Fig. 7). The association of the a and /3 subunits to form the structure also remained unchanged in cells treated with MandJN. These results indicate that the action of mannosidase I and subsequent reactions leading to the formation of complex oligosaccharides are not required for these processes to occur.
Use of MedJN caused production of receptors of slightly lower molecular weight than normal that contained high mannose oligosaccharides as shown by their greater sensitivity to Endo H (Fig. 1). The molecular weights of the Endo Hdigested subunits are not as low as would be expected if all of the oligosaccharides were of the high mannose type (compare with results using swainsonine (18) and MandJN). Analysis of the glycopeptides obtained from insulin receptors of MedJN-treated cells (Fig. 2) indicated that there was only a partial inhibition of complex oligosaccharide formation (48% inhibition based on total label incorporated), which is consistent with the partial Endo H sensitivity of these receptors. The high mannose-type oligosaccharides were also shown to be glucosylated in the MedJN-treated samples (Fig. 3). Partial inhibition of complex oligosaccharide formation has been observed previously with MedJN and has been attributed partly to the transfer of non-glucosylated residues to the nascent glycoprotein (29). At present, we cannot determine whether transfer of non-glucosylated oligosaccharides to the receptors occurs, and whether it may be random or site specific. Similar to receptors made in the presence of MandJN, those made in the presence of MedJN were also able to bind hormone and autophosphorylate. They associated into ( (~8 )~ complexes and were present at the cell surface. The main difference observed with these receptors was an altered rate of processing of the receptor precursors (Figs. 4 and 5). The accumulation of receptor precursors provides an explanation for the presence of fewer a and /3 subunits in cells treated with MedJN. Therefore, inhibition of glucosidase activity by MedJN inhibits processing of some of the oligosaccharide residues of the receptors, and proteolytic processing of the receptor precursors is inhibited. These receptors are still able to bind hormone and retain their hormonesensitive autophosphorylation activity.
It is clear from this study, along with previous studies using swainsonine, that the nature of the oligosaccharides on the insulin and IGF-I receptors has little effect on the function of the receptors. But several of the experiments presented here provide a great deal of information pertaining to the steps involved in receptor biosynthesis and the involvement of oligosaccharide processing. Receptors synthesized in the presence of MandJN having only high mannose oligosaccharides can be synthesized and translocated to the cell surface in a manner similar to control receptors. In the presence of MedJN, only some of the oligosaccharide residues of the a and /3 subunits are glucosylated high mannose residues, as shown by the slightly increased Endo H sensitivity compared with controls. The receptor precursors in MedJN-treated cells had a higher molecular weight than the controls. This sug-gested that there was a substantial number of glucose .residues present on the oligosaccharides. The precursors accumulated intracellularly, and this may be explained in two ways. Many of the oligosaccharides added to the nascent polypeptide may be non-glucosylated, which has been seen in another cell line (29). The presence of glucose residues on the remaining oligosaccharides may slow down the movement of the precursor to the compartment(s) where further processing can occur. Alternatively, MedJN may cause only a partial inhibition of glucosidase activity resulting in a slower rate of glucose removal. Perhaps some critical glucose residues have to be removed before the precursor can be translocated and processed. The oligosaccharides from which glucose residues have been removed can then be processed to complex types, leaving only some of the residues glucosylated. At present we cannot distinguish between these possibilities.
Pulse-chase experiments show that the receptor precursor containing high mannose oligosaccharides is converted to a and /3 subunits containing complex oligosaccharides. Very little of a higher molecular weight precursor containing complex oligosaccharides is seen, and only in IGF-I receptor samples. Proteolytic cleavage and oligosaccharide processing are therefore very closely associated in HepG2 cells. In a pulse-chase study of insulin receptors in 3T3-Ll cells (24), it was shown that the high mannose precursor was first chased into a and /3 subunits containing high mannose residues that were then converted to complex oligosaccharides. This suggests that the exact sequence of processing steps may be different in different cell types, probably determined by the intracellular location or relative amounts of the various enzymes involved.
Trypsin sensitivity of cell surface receptors shows that there are few, if any, differences in trypsin sensitivity of receptors with altered oligosaccharides (Fig. 7). These experiments did demonstrate some differences in trypsin sensitivity between insulin and IGF-I receptors. The a subunit of the IGF-I receptor is much less degraded than the insulin receptor a subunit. On the other hand, the /3 subunit of the insulin receptor is not affected by trypsin, whereas the IGF-I receptor / 3 subunit is at least partially sensitive. The partial sensitivity of the IGF-I receptor /3 subunit may be explained by the presence of two forms having different patterns of glycosylation, resulting in a slight difference in conformation.
The association of the receptors into oligomeric structures appears to occur soon after biosynthesis of precursors, since all the precursors that accumulate in MedJN-treated cells form disulfide-linked dimers (Fig. 9). In the case of the insulin receptor, there may be two forms of the dimerized precursors. The difference in migration on SDS gels and Endo H sensitivity of these two forms may be caused by different patterns of disulfide bond formation and not differences in glycosylation. This suggestion is based on the fact that only one receptor precursor is seen in samples run on reducing gels, and all of this is as sensitive to Endo H as the precursor in control or MandJN-treated cells (Fig. 1).
Consistent with our results, it has been reported recently (38) that castanospermine and dJN, two other glucosidase inhibitors, also cause accumulation of the insulin receptor precursor and a reduction in the number of cell surface insulin receptors. In a study of the nicotinic acetylcholine receptor, inhibition of oligosaccharide processing by dJN also caused a decrease in cell surface receptors (45). The intracellular accumulation of insulin and IGF-I receptor precursors containing glucosylated oligosaccharides are also consistent with other reports showing that dJN and castanospermine inhibit translocation of viral glycoproteins (33), IgD (34), and a l -proteinase inhibitor (35,36). However, translocation of some other glycoproteins is not affected by dJN (36,37).
Two other reports have shown that in Chinese hamster ovary cells starved for glucose and in Chinese hamster ovary mutants having altered glycosylation pathways, insulin receptors had an altered affinity for insulin, whereas the affinity of IGF-I for its receptor was not affected (39, 44). As mentioned, we have shown that the receptors made in the presence of processing inhibitors are able to bind hormone, but we have not yet obtained information regarding the affinity of this binding.
These results permit us to define precisely the oligosaccharide processing requirements for the expression of functional insulin and IGF-I receptors. The first step, addition of Nlinked oligosaccharides, is essential, since its inhibition by tunicamycin prevents receptor translocation, proteolytic processing and the ability to bind hormone (24, 27). Inhibition of the next step, removal of glucose residues, delays proteolytic processing and translocation to the cell surface, but those receptors that are proteolytically processed get to the cell surface and are able to bind hormone and autophosphorylate. It is not clear whether more complete inhibition of glucose removal is possible and whether this would have more drastic effects on these processes. However, almost complete inhibition of mannosidase I, which is the very next step in processing, has no effect on proteolytic processing, translocation, hormone binding, or phosphorylation. Therefore, since removal of glucose residues is one of the last steps that occurs in the endoplasmic reticulum, it seems to be a critical step in facilitating translocation to the Golgi apparatus and further processing of the receptor.