Effects of amino acid replacements within the tetrabasic cleavage site on the processing of the human insulin receptor precursor expressed in Chinese hamster ovary cells.

We have studied the specificity requirements for processing of the human insulin proreceptor by successively replacing each basic amino acid in the tetrabasic cleavage site with alanine. These mutated receptor cDNAs have then been overexpressed in Chinese hamster ovary cells, using vectors containing the mouse dihydrofolate reductase gene to amplify the transfected cDNAs in the presence of increasing concentrations of methotrexate. High levels of expression, ranging up to 6 x 10(7) receptors/cell were achieved in these experiments. Replacement of the P1 arginine with alanine led to the complete suppression of processing, as occurs also in a naturally occurring serine mutation at this site (Yoshimasa, Y., Seino, S., Whittaker, J., Kakehi, T., Kosaki, A., Kuzuya, H., Imura, H., Bell, G. I., and Steiner, D. F. (1988) Science 240, 783-787). A small amount of cleavage at alternative sites was detected. Replacement of the P2 arginine or P3 lysine with alanine did not in either case affect conversion to mature alpha and beta subunits, while replacement of the P4 arginine significantly inhibited processing. The binding isotherms for the processed versions of the receptor were comparable to previously published normal values. The unprocessed proreceptor bound insulin normally but was autophosphorylated less efficiently than processed versions of the receptor expressed in the same cells. These results suggest that a single processing protease with trypsin-like specificity may be involved in processing both insulin and insulin-like growth factor-I receptor precursors as well as a variety of viral envelope glycoprotein precursors.

We have studied the specificity requirements for processing of the human insulin proreceptor by successively replacing each basic amino acid in the tetrabasic cleavage site with alanine. These mutated receptor cDNAs have then been overexpressed in Chinese hamster ovary cells, using vectors containing the mouse dihydrofolate reductase gene to amplify the transfected cDNAs in the presence of increasing concentrations of methotrexate.
High levels of expression, ranging up to 6 x 10' receptors/cell were achieved in these experiments.
A small amount of cleavage at alternative sites was detected. Replacement of the P2 arginine or P3 lysine with alanine did not in either case affect conversion to mature u and ,8 subunits, while replacement of the P4 arginine significantly inhibited processing. The binding isotherms for the processed versions of the receptor were comparable to previously published normal values. The unprocessed proreceptor bound insulin normally but was autophosphorylated less efficiently than processed versions of the receptor expressed in the same cells. These results suggest that a single processing protease with trypsin-like specificity may be involved in processing both insulin and insulin-like growth factor-1 receptor precursors as well as a variety of viral envelope glycoprotein precursors.
The insulin receptor is a heterotetrameric integral membrane protein made up of two (Y and two /3 subunits (l-3). The (Y subunit is located entirely outside the cell and is linked by disulfide bonds to the extracellular portion of the /3 subunit. The p subunit is anchored to the plasma membrane through a single hydrophobic membrane-spanning segment attached to a large cytoplasmic portion. The cytoplasmic portion contains a tyrosine kinase domain whose activation occurs upon insulin binding to the cx subunit and is thought to be an initiating event in cellular responses to insulin (4). The LY and /3 subunits are derived from a single polypeptide precursor, or proreceptor, that is ordered as follows: signal peptide-cy subunit-RKRR cleavage site-/3 subunit (2,3). While the newly synthesized precursor migrates from the endoplasmic reticulum to the Golgi apparatus, it undergoes several modifications including glycosylation, assembly into oligomers, and cleavage into mature LY and @ subunits (5). We previously identified a defect in the proteolytic processing of the proreceptor into mature subunits in a diabetic patient with the type A syndrome of severe insulin resistance (6). Analysis of the patient's receptor gene showed that this defect was likely to be due to the replacement of a single basic amino acid within the cleavage site by a serine (7). Studies of the mutant proreceptor expressed on the surface of EBVtransformed lymphocytes from this patient demonstrated that, while still capable of activation, its insulin binding affinity was markedly decreased (6). This observation supports the contention that proteolytic processing is an absolute requirement for the attainment of full receptor function. The physiological importance of proteolytic processing is also suggested by the finding that the tetrabasic cleavage site is well conserved in several closely related proteins, including the insulin-like growth factor 1 receptor precursor (8), Drosophiln insulin receptor-like protein precursor (9), and the recently identified human insulin receptor-related protein precursor (10). Furthermore, a similar sequence motif is found at the cleavage sites of the envelope glycoproteins of a number of viruses (HIV-I gp160, influenza hemagglutinin) in which proteolytic processing at these sites is essential for full viral infectivity (11,12). Although the sequence requirements for cleavage of several of the viral glycoprotein precursors have been studied in considerable detail (13), no studies on the requirements for cleavage of the insulin proreceptor have been reported.
Studies on the biochemical properties of the insulin proreceptor have been limited because the level of receptor expression is relatively low in EBV-virus transformed lymphocytes or fibroblasts, the cells commonly used to characterize the biochemical phenotype of such patients. In order to characterize the properties of the insulin proreceptor in greater detail and determine the requirements for processing, we have successively replaced each of the basic residues at the RKRR only the cell lines transfected with wild type insulin receptors but all cell lines expressing mutated insulin receptors showed significant increases in specific insulin binding as compared with control CHO cells (data not shown).

Biosynthetic
Labeling-Since studies of the binding of insulin to the mutant proreceptor (Pl serine substitution for arginine) on EBV-transformed lymphocytes from the patient showed that it was markedly reduced (6), we carried out biosynthetic labeling studies on cells expressing mutated insulin receptor precursors followed by immunoprecipitation with antireceptor antibodies to detect proreceptors and/or subunits. As demonstrated in Fig. 1, cell lines expressing insulin receptors with mutations within the cleavage site (designated P4-A, P3-A, P2-A, and PI -A) all synthesize receptor precursors having a molecular mass of 190 kDa in greatly increased amounts. Indeed, we found that receptor proteins solubilized from cell lines expressing the P3-A and Pl-A mutations could be identified by Coomassie Blue staining of the SDS-polyacrylamide gels after immunoprecipitation, even though only a relatively small number of cells (-5 x 106) were analyzed. However, the proteolytic processing of the respective receptor precursors during the chase period differs significantly among these lines (see below).
While the precursors with P3-A and P2-A substitutions are processed almost completely into mature cx and p subunits ( Fig. 1, lanes 2 and 3), the Pl-A-substituted HIR gives rise predominately to intact proreceptor, although a faint band is also evident that migrates slightly faster than the authentic cy subunit (Fig. 1, lane 4). Further biosynthetic labeling experiments using CHO Pl-A cells revealed that about 5% of the precursor (by densitometric scanning of the autoradiogram) is cleaved, giving rise to two proteins having molecular weights close to but not identical with those of mature cu and /3 subunits. These results suggest that cleavage of this mutated HIR occurs at an alternative site(s) close to the normal processing site but that essentially no cleavage occurs at the normal site. These studies also revealed the presence of an additional band with a molecular mass of approximately 180 kDa in the CHO Pl-Al cells which could be cross-linked to insulin (see Fig. 5). The site of cleavage to produce this band is unknown but it probably contains the entire (Y subunit and much of the external portion of the /3 subunit. The receptor precursor having alanine at P4 is processed into subunits less efficiently than the processed versions, as shown by the rela- tively large amount of the proreceptor in relation to mature subunits (Fig. 1, lane I).

High Level Expression of Receptor Proteins in CHO Cell
Lines-When the insulin receptor from P3-A cells was solubilized, enriched by WGA-agarose chromatography and subjected to SDS-PAGE without immunoprecipitation, polypeptides migrating with the M, of HIR (Y and @ subunits were visible after staining with Coomassie Blue (Fig. 2). This finding suggests that HIR subunits comprise a substantial portion (up to 50%) of the total protein present in WGA eluates from the P3-A cell line. We therefore estimate that 30-50 pg of insulin receptor is present in the WGA eluate from 5 X lo6 CHO P3-A cells (approximately 10 pg/cell; >107 molecules/cell). It should be noted also that the integrity of the receptor protein is well maintained, in that the /I subunit, which often appears to be degraded such that it exhibits a smaller than expected size as compared with the (Y subunit in purified receptor preparations, is comparable in amount to the (Y subunit. We have confirmed the identities of these two Coomassie Blue-stained bands by demonstrating that the M, = 135,000 band can be affinity labeled with '251-tyrosine-B26 insulin and the M, = 95,000 band incorporates 32P04 in an insulin-dependent fashion in an in vitro autophosphorylation reaction (data not shown).
We compared the levels of expression attained in some of these cell lines with those achieved with other expression systems that have been used in this laboratory (14,22). Equal amounts (10 pg) of WGA eluate were prepared from CHO P3-A cells, the frbroblast cell line (HIR3T3 3.5) and Sf9 insect cells infected with a recombinant baculovirus encoding the human insulin receptor and were analyzed by cross-linking and autophosphorylation. The results showed that the highest level of expression was attained in the present study (data not shown).
For specifically. This finding is in agreement with the data from metabolic labeling, which suggested that two alternate processing sites appear to be used at low levels when cleavage at the normal processing site is prevented. Autophosphorylution of the Proreceptor- Fig. 6 shows that the proreceptor becomes phosphorylated in an insulin-dependent fashion similarly to the /3 subunit. However, densitometric scanning of the autoradiogram shows a significant difference in dose dependence. Autophosphorylation of the proreceptor occurs more gradually with an EDso about lo-fold higher than that for autophosphorylation of the mature /3 subunit. Fig. 6 also shows that the small amount of @ subunitlike material derived from cleavage of the precursor having Pl-A also becomes phosphorylated with a dose dependence similar to that of the /3 subunit of the processed receptor. Amounts of WGA-agarose eluates from the cells were adjusted to give similar binding capacities and then cross-linked, immunoprecipitated, and analyzed by SDS-PAGE radioautography as described under "Experimental Procedures." insulin receptor precursors or subunits was much lower in the CHO P3-A preparations (data not shown), confirming the relatively greater abundance of processed receptor.
Cross-linking of the Proreceptor-To further examine the insulin binding properties of the proreceptor, amounts of WGA eluates from both cell lines adjusted to give similar insulin binding capacities were chemically cross-linked to 1251tyrosine-B26-insulin and the immunoprecipitated receptor proteins were analyzed by SDS-PAGE. As shown in Fig. 5, the proreceptor from CHO Pl-A cells is cross-linked to a comparable extent to the processed (Y subunit from CHO P3-A cells. This is in contrast with the behavior of the Pl serine mutant proreceptor, studied on EBV-transformed lymphocytes from the propositus, which was only weakly labeled in similar cross-linking experiments (7).
In the preparation from CHO Pl-A cells, a protein with a molecular mass similar to that of the LY subunit, as well as a protein with a molecular mass of 185 kDa were also labeled Oligomeric State of the Proreceptor-As shown in Fig. 7A, metabolic labeling of HIR subunits followed by SDS-PAGE under nonreducing conditions shows a single band of dimeric proreceptor from CHO Pl-A cells with a molecular mass of approximately 380,000. In contrast, a more complicated pattern consisting of at least three major and two minor bands is derived from the processed receptor from the CHO P3-A cells. Three of the bands have high molecular weights and appear to represent various states of processing of the dimeric proreceptor. The two minor bands having M, values of 230,000 and 95,000 correspond to the monomeric receptor precursor and a small amount of free /3 subunit, respectively.
To further examine their oligomeric states, the labeled receptor proteins were dissociated by graded concentrations of dithiothreitol and analyzed by SDS-PAGE under nonreducing conditions. As shown in Fig. 7B, the major bands with high molecular weights derived from processed receptor from the CHO P3-A cells are dissociated into four proteins. Among these are the receptor precursor and the native (Y and /3 subunits; the identity of the fourth receptor-related band having a M, of 280,000, which is consistently found in the metabolic labeling studies, is unknown.
It is of interest that under the conditions used for reduction in these experiments (95 "C, 3 min) reduction of the inter (Y-/3 disulfide bonds occurred more rapidly than reduction of inter (Y bonds, as occurs under milder conditions (24). with increasing concentrations of dithiothreitol the apparent molecular size of the 380-kDa band increases significantly. This change in mobility may result from partial reduction of disulfide bridges, allowing the molecule to become more asymmetric. Taken together, these data indicate that the proreceptor is inserted into the plasma membrane as an organized (LY-& dimer.

DISCUSSION
The high degree of conservation of the tetrabasic sequence at the cleavage site separating the LY and /3 subunits in the precursors of both the insulin (human, Drosophila) and IGF-I (human) receptors (3,8,9,19) suggests that these sites would be likely to exhibit stringent requirements for processing by a special cellular trypsin-like protease. However, the results of the study reported here, in which we have systematically replaced all 4 of these residues individually with alanine, a small amino acid having a neutral side chain consisting of a single methyl group, indicate that this is not the case.
This study was prompted by our analysis of a mutation in the insulin receptor gene of a patient with severe insulin resistance due to defective insulin binding. Various studies revealed the presence of essentially normal numbers of uncleaved proreceptors on the surface of EBV-transformed lymphocytes from this patient (6,7). No mature (Y or /3 subunits could be identified on these cells. Sequence analysis of HIR exon 12 (25) from this patient's genomic DNA revealed a point mutation in both copies of the gene that changed the Pl arginine residue at the cleavage site to serine. It was further shown that graded trypsin treatment resulted in the appearance of some normal subunits accompanied by partial restoration of binding activity (7). We were interested in further characterizing the effects of this mutation on receptor function and also in defining more clearly the requirements for cleavage. The amino acid sequences of the cleavage sites of some viral glycoproteins and the influenza hemagglutinin (HA) are also similar to that of the insulin receptor. Studies of viral variants have indicated that a canonical sequence, R. X. R/K. R, is required for efficient processing of these proteins (13). Thus, cellular proteases having similar specificities may be involved in the proteolytic processing of all of these proteins.
The present study demonstrates that in CHO cells stably expressing insulin receptor constructs at high levels 1) alanine substitution at Pl profoundly impairs processing, 2) alanines 43Kat positions P2 and P3 have no effect on cleavage, and 3) alanine at P4 reduces the efficiency of processing. These observations are thus consistent with the existence of a cellular protease with restricted trypsin-like specificity. Since mutational analyses of the cleavage sites of membrane glycoproteins similar to the insulin receptor are limited, the specificity requirements of these processing enzymes remains undefined. However, a study examining the requirements for cleavage of the influenza HA suggests that these proteases require the presence of several basic amino acids at the cleavage sites, preceded by a /3 turn (12,13). The insulin and IGF receptors also have a region with a high predicted probability for a p turn just upstream of the cleavage site as shown in Fig. 8. A unique protease with a specificity similar to that of the insulin proreceptor convertase has also been reported to be involved in processing several peptide precursors in Xenopus Levis skin (26). These observations, along with ours, suggest that protease with specificities for clusters of basic amino acids may be widespread in cells and probably differ in specificity and other properties, such as pH optimum and subcellular localization, from the processing enzymes that cleave prohormones which usually recognize single or adjacent pairs of basic amino acids.
Our experiments indicate, in addition, that the cellular protease is capable of processing receptor precursor with very great efficiency, as exemplified by the experiments using CHO P3-A in which essentially complete proteolytic process- The residues in the insulin receptor cDNA that were replaced individually by alanines in this study are shown enclosed in the box. The position of the naturally occurring serine substitution (7) is noted above the box. The location of the putative @turn was predicted using the rules of Chou and Fasman (34).

17236
Insulin Proreceptor-processing Site Mutants ing of the precursor occurs despite the fact that it is being produced at levels that are several thousand-fold greater than normal. Conversely, it also can be concluded that the constraining effect of Pl substitutions on processing is very strict because the proteolytic processing is impaired almost completely even though the receptor precursor is being overproduced in large amounts.
This latter finding contrasts with the reported observation that only a small percentage of the envelope glycoprotein (gp160) of human immunodeficiency virus I is cleaved into gp120 and gp40 (27). The cleavage site in this protein, REKR, would appear to be a good substrate in itself but may be influenced by other features of the local environment of the precursor. Cleavage of the gp160 envelope precursor has been shown to occur predominantly in the cis or medial compartments of the Golgi and is unusually slow in that it requires several hours for its transfer from the RER to the Golgi complex (28).
We have used cells expressing the insulin proreceptor due to substitution of alanine for arginine at Pl of the cleavage site to characterize the proreceptor in greater detail. Several interesting aspects of the proreceptor have been revealed which appear to contrast with the properties of the proreceptor resulting from a naturally occurring serine substitution at this position in the patient with severe insulin resistance (6,7). The proreceptor thus expressed possesses high insulin binding affinity and shows a complex curvilinear Scatchard plot that is almost identical to that of the normally processed receptor. Also, phosphorylation of the proreceptor is only moderately reduced in comparison with the processed receptor, implying that signal transduction through the proreceptor is not grossly impaired. These observations contrast sharply with the behavior of the proreceptor on the patient's EBVtransformed lymphocytes, as described above. A clue to reconciling these discrepancies may come from findings that exon 11 of human insulin receptor gene is alternatively spliced to give rise to two receptor (proreceptor) isoforms, as originally indicated by the data of Ebina et al.
( 3), and that the alternative splicing event occurs in a tissuespecific manner. Several groups (29,30) have reported that a receptor isoform lacking exon 11 is expressed in a number of tissues, including cell lines of B-lymphocyte origin, whereas both receptor isoforms are expressed in liver, kidney, and placenta. Since the human insulin receptor cDNA that we used as a template for these mutagenesis studies contains exon 11, the differences in the properties of the proreceptors observed in the two kinds of cells could be ascribed to the presence or absence of exon 11.
Although exon 11 encodes only 12 amino acids, these are located immediately upstream of the cleavage site (25) and relatively distant from the putative insulin-binding domain (31) of the insulin receptor. Nonetheless, it may be possible that the presence or absence of the short stretch of amino acids encoded by exon 11 may affect the folding and/or conformation of the proreceptor so as to confer altered sensitivity to insulin. In the influenza hemagglutinin (HA), the presence or absence of a glycosylation site near the N terminus of the protein is in sufficient proximity to the cleavage site to significantly affect its processing (12). In the insulin receptor evidence suggests that the insulin binding site may include portions of the N-terminal domain of the 01 subunit (29), and if organized like the influenza HA molecule the binding site might be directly hindered by the shorter (exon 11 minus) c@ junction.
Such a difference in receptor function could be of physiological relevance. Experiments are currently underway to compare the biochemical properties of two isoforms of the proreceptor.
Some of our data on the proreceptor indicates properties that differ subtly from those of the processed receptor, e.g. the dose dependence for phosphorylation of the proreceptor is right-shifted (EDso = 1 X lo-' M for the proreceptor and EDso = 2 X 10m9 M for the processed receptor) although the maximal incorporation of PO, above the basal level is comparable.
Another difference is in the oligomeric state of the proreceptor as shown by SDS-PAGE under nonreducing and graded reducing conditions.
The proreceptor forms an (CZ-& dimer linked together by disulfide bonds and the dimer appears to have a more elongated shape, as indicated by the data that the proreceptor dimer migrates much more slowly when the intermolecular disulfide-bonds are dissociated with graded concentrations of dithiothreitol. Since phosphorylation of the proreceptor, as well as the mature fi subunit, may require a conformational change of the receptor protein after insulin binding to the (Y subunit, an altered oligomeric state of the proreceptor might influence the dose dependence for phosphorylation as mentioned above. Evidence that the proreceptor with exon 11 is functional might shed light on the structural (quaternary) requirements for signal transduction through molecules that are related to the insulin receptor. With regard to this point it is of interest that the insulin receptor of the stingray, a cartilagineous fish, appears to be a proreceptor dimer like the mutated human proreceptor studied here (32). The stingray receptor binds both insulin and IGF-I with high affinity and undergoes autophosphorylation (with either ligand). It has also been observed that the epidermal growth factor receptor must oligomerize on the cell surface in order to accomplish a full signal transduction (33). Thus, one requirement for signal transduction through the relevant receptors appears to be dimerization.