The Q Allele Variant (GLN¹²¹) of Membrane Glycoprotein PC-1 Interacts With the Insulin Receptor and Inhibits Insulin Signaling More Effectively Than the Common K Allele Variant (LYS¹²¹)

Mutation of human PC-1 was performed using the Quick-Change site-directed mutagenesis kit (Stratagene, San Diego, CA) (20). The pRK7-Q¹²¹PC-1 mutated clone was sequenced to check for unintended mutations.

Human MCF-7 breast cancer cells and embryo mouse R⁻/hIR fibroblasts with a targeted disruption of the IGF-1-R gene and expression of the human IR gene by transfection (23) were grown as previously described (10,23). Cells were transfected (by calcium phosphate procedure) (24), with both an expression vector containing the coding sequence of human PC-1 (25) and pRK-NEO, a selectable marker for neomycin resistance. Under the control of the cytomegalovirus promoter, PC-1 cDNA was transfected either as the K¹²¹PC-1 or the Q¹²¹PC-1 variant. Clones expressing a similar PC-1 content (i.e., hydrolysis of p-nitrophenyl thymidine 5′-monophosphate [PNTP]) were selected for both MCF-7 cells and R⁻/hIR cells (Table 1). In MCF-7 cells, PC-1 expression was also evaluated by Western blot (13). Transient transfections of hIR (provided by A. Ullrich, Max-Plank Institute, Martinsried, Germany) and PC-1 cDNA were carried out in HEK 293 cells (26).

IR protein content was measured by enzyme-linked immunosorbent assay (ELISA) (27) and normalized for protein content (28). ¹²⁵I-labeled insulin (33 pmol/l in 50 mmol/l HEPES buffer, pH 7.8) was added to 1.5 × 10⁶ cells per tube in the presence of increasing native insulin concentrations, and specific binding was then calculated (18).

PC-1–IR interaction was evaluated by ELISA and Western blot analysis in MCF-7 cells (13). Subconfluent cells that were serum-starved for 18 h at 37°C for 2 h at 22°C were washed and solubilized in HEPES 50 mmol/l, pH 7.6, 1% Nonidet P-40, and 1 mmol/l phenylmethylsulfonyl fluoride. For the ELISA (13), cell lysates (0.04 mg protein) were added to plastic wells precoated with the anti-IR antibody MA-20 (29), and after wells were washed with a Tris-buffered saline with Tween buffer (20 mmol/l Tris, pH 7.4, 150 mmol/l NaCl, and 0.05% Tween-20), PC-1 bound to the IR was revealed with a biotinylated anti–PC-1 monoclonal antibody (provided by Dr. I.D. Goldfine, San Francisco, CA) and detected by the peroxidase-streptavidin method by measuring the peroxidase activity determined colorimetrically. For Western blot (13), cell lysates (2 mg) were immunoprecipitated with 20 μg/ml of MA-20 (18 h, 4°C). Proteins were transferred to nitrocellulose membranes, blocked in phosphate-buffered saline with 0.1% Tween buffer with 3% milk for 30 min, incubated with 90B rabbit polyclonal anti–PC-1 antiserum (Dr. I.D. Goldfine, San Francisco, CA) and then quantified by anti-rabbit antiserum conjugated with horseradish peroxidase and enhanced chemiluminescence (ECL) detection system.

IR and IGF-1 receptor (IGF-1-R) autophosphorylation were measured by specific ELISAs (18). After cell exposure to either insulin or IGF-1, IR was immunocaptured by a specific anti-IR monoclonal antibody (MA-20), whereas IGF-1-R was immunocaptured by a specific anti–IGF-1-R monoclonal antibody (αIR3) (30). After washing, a biotinylated antiphosphotyrosine (a-pY) antibody (UBI Diagnostic, Lake Placid, NY) was added, and the peroxidase-streptavidin method was used for colorimetrically revealing phosphorylation.

For insulin receptor substrate-1 (IRS-1) phosphorylation measurement, subconfluent cell monolayers were lysed after incubation with or without insulin (10 nmol/l) for 1 and 5 min at 37°C (23). Proteins (1.2 mg) were immunoprecipitated with an anti–IRS-1 monoclonal antibody (4 μg/ml) (UBI Diagnostic) conjugated with protein A-Sepharose (16 h at 4°C), subjected to SDS-PAGE (7.5% polyacrylamide), transferred to nitrocellulose membranes, incubated with a-pY antibody (1 μg/ml) (UBI Diagnostic) and then quantified by rabbit anti-mouse antiserum conjugated with horseradish peroxidase and ECL system (23).

Phosphatidylinositol 3-kinase (PI 3-kinase) activity was measured by thin-layer chromatography in MCF-7 cell lysates after stimulation with insulin (10 nmol/l for 5 min at 37°C) (18).

The rate of glycogen synthesis after the cells were exposed to insulin was measured, as previously described, by the incorporation of [¹⁴C]glucose into cellular glycogen (18).

Cell proliferation in response to insulin (100 nmol/l) was evaluated by measuring cell DNA (31) after a 4-day incubation of MCF-7 cell monolayers in serum-deprived medium (0.1% bovine serum albumin).

Mean values were controlled by one- and two-way analysis of variance tests. Fisher’s test was applied to ascertain differences among K and Q PC-1 variants. Data are presented as means ± SE.

PC-1 content was measured by both enzymatic activity (hydrolysis of PNTP in both MCF-7 and R⁻/hIR cells) and Western blot (in MCF-7 cells). We selected K- and Q-PC-1 clones that had a similarly elevated PC-1 content (Table 1, Fig. 1) compared with cells transfected with the neomycin resistance gene alone (NEO). The degree of PC-1 overexpression in transfected cells was in the range previously observed in skeletal muscle and cultured skin fibroblasts of insulin-resistant individuals (15,16,18). In addition, IR content (as measured by a specific ELISA) was similar in NEO, K, and Q cells (Table 1).

It has been previously reported that PC-1 may interact and be associated with the IR (13). By using a specific ELISA, we observed that PC-1 was associated with the IR in all cell lines (Fig. 2). The amount of PC-1 associated with IR was clearly increased in both K and Q cells compared with control cells (P < 0.01). Moreover, the amount of PC-1 associated with the IR was significantly (P < 0.01) higher in Q-PC-1 cells than in K-PC-1 cells (Fig. 2). Very similar data were obtained when the association of PC-1 with the IR was measured by Western blot (Fig. 3).

In MCF-7-NEO cells, insulin induced a dose-dependent IR autophosphorylation with a detectable effect at 0.1 nmol/l (Fig. 4A). In cells expressing either the Q- or K-PC-1 variants, the effect of insulin on IR autophosphorylation was markedly reduced at all insulin concentrations (P < 0.01) (Fig. 4A). Moreover, in the Q-PC-1 clones, insulin-stimulated IR autophosphorylation was significantly reduced compared with the K-PC-1 cells (Fig. 4A). In contrast, IGF-1-R autophosphorylation in response to IGF-1 was not affected in cells overexpressing the two PC-1 variants (Table 2).

To make certain the Q-PC-1 variant effects were not cell-specific, we studied the Q variant also found in R^–/hIR mouse fibroblasts, a cell line that expresses the human IR but not the IGF-1-R (23). Clones with similar PC-1 enzymatic activities were selected (Table 1). The responsivity to insulin stimulation of IR autophosphorylation was decreased in both K- and Q-PC-1 cells compared with NEO cells (Fig. 4B). Moreover, IR autophosphorylation responsivity to insulin stimulation was more decreased in cells overexpressing the Q variant than it was in cells expressing the K variant (Fig. 4B). In addition, a right shift in the insulin dose response curve was observed for both K and Q cells (50% stimulation being obtained at 3.4 mmol/l, 15 nmol/l, and 14 nmol/l insulin in NEO, K, and Q cells, respectively). To confirm data obtained in stably transfected cell clones, transiently transfected HEK 293 cells were also studied as a model of unselected PC-1–overexpressing cells. Compared with cells transfected only with IR cDNA (control cells), insulin-stimulated (10 nmol/l) IR autophosphorylation was significantly reduced in cells transfected with both IR and K-PC-1 cDNAs (P < 0.05), and it was even more reduced in cells transfected with Q-PC-1 cDNAs (P < 0.01 vs. control cells and P < 0.05 vs. K-PC-1 cells) (data not shown).

We then investigated the effect of the two PC-1 variants on downstream IR signaling. Insulin (10 nmol/l) induced a fourfold to fivefold increase of IRS-1 phosphorylation in MCF-7-NEO cells (Fig. 5). This effect was reduced by 25–35% (P < 0.05 vs. NEO cells) in cells expressing the K variant and by 45–78% in cells expressing the Q variant (P < 0.05 vs. both NEO and K cells) (Fig. 5). Similar data were obtained after 1 min of insulin stimulation (38 and 62% reduction in K and Q cells, respectively).

After MCF-7-NEO cells were exposed to 10 nmol/l insulin for 5 min, we observed significant increase of basal PI 3-kinase activity in both NEO-1 (209 ± 52% of basal, n = 6) and NEO-2 cells (222 ± 68, n = 3; P < 0.05). In contrast, insulin stimulation of PI 3-kinase activity did not reach statistical significance in K-1 (147 ± 25%, n = 6) and K-2 cells (115 ± 7, n = 3), and it was completely abolished in Q-PC-1 MCF-7 cells (80 ± 11, n = 6; and 102 ± 1, n = 3; in Q-1 and -2, respectively).

The reason for the apparent different inhibition of Q-PC-1 on IRS-1 phosphorylation (partially inhibited) versus PI 3-kinase activity (totally inhibited) is unknown. Among the different possibilities is a direct PI 3-kinase inhibition by PC-1 through different mechanisms, and there are also intrinsic differences in cell sensitivity to insulin stimulation of the two signaling molecules; in fact, after insulin stimulation, IRS-1 phosphorylation increased approximately fourfold over the basal value, whereas PI 3-kinase activity increased only roughly twofold.

Insulin-stimulated [¹⁴C]glucose incorporation into glycogen increased in a dose-dependent manner in MCF-7-NEO cells (Fig. 6). This effect was impaired in K-PC-1 (P < 0.01), and it was nearly abolished in Q-PC-1 MCF-7 cells (P < 0.01 vs. NEO and P < 0.05 vs. K-PC-1 cells). Also, basal (i.e., non–insulin-stimulated) glycogen synthesis was reduced in K-PC-1 cells, and it was even more reduced in Q-PC-1 cells, though not significantly. This phenomenon is reminiscent of the reduced basal glycogen synthesis observed in cultured cells from insulin-resistant individuals (18,32).

Similar inhibition was also observed for the insulin-stimulated mitogenic effect (Table 3). Both basal and insulin-stimulated (100 nmol/l) cell proliferation was impaired in K- and Q-PC-1 MCF-7 cells (P < 0.01). After insulin stimulation, the percentage increase over basal value was significantly (P < 0.05) lower in Q-PC-1 cells (168 ± 21%) than it was in K-PC-1 (293 ± 47%) cells.

We and others have reported that PC-1 overexpression inhibits IR-TK activity in several types of cultured cells, including human fibroblasts (10,18,19), human breast carcinoma cells (10,12,13), hamster CHO cells (11), and rat hepatoma cells (13). In human subjects, PC-1 overexpression in tissues and cells has been correlated with insulin resistance (10,15–19,33). Altogether, these observations indicate that, when overexpressed, PC-1 may play a role in insulin resistance.

We have recently identified a frequently occurring PC-1 amino acid variant (Q¹²¹PC-1) that associates with insulin resistance (20) and with a faster progression of kidney disease in type 1 diabetic patients (34). In cultured skin fibroblasts expressing a similar PC-1 content, cells expressing the Q¹²¹PC-1 variant had decreased insulin-stimulated receptor autophosphorylation compared with cells expressing the K¹²¹PC-1 variant (20). This observation is compatible with the hypothesis that this variant is more potent than the K¹²¹PC-1 variant in inhibiting IR signaling and may thus contribute more effectively to insulin resistance.

The association between Q¹²¹PC-1 and insulin resistance was observed in two independent studies in different populations (20–21), but it was not seen in a third study in Danish Caucasians (22). These different findings make it possible that the Q¹²¹PC-1 variant does not have direct biological significance and is simply in linkage disequilibrium with mutations of different genes that cause insulin resistance.

We have conducted direct functional studies to evaluate whether the Q¹²¹PC-1 variant is more potent in inhibiting insulin sensitivity. When the two variants were expressed at a similar level, the Q¹²¹PC-1 variant was indeed more potent than the K¹²¹PC-1 variant in inhibiting insulin stimulation of IR autophosphorylation, the insulin signaling pathway (IRS-1 phosphorylation and PI 3-kinase activity), and insulin biological effects (glycogen synthesis and cell proliferation). Thus, it is likely that the Q¹²¹PC-1 variant is a stronger causative factor than the more common K¹²¹PC-1 variant in contributing to insulin resistance.

It has recently been reported that PC-1 inhibits IR signaling by interacting directly with the connecting domain region of the α-subunit of the IR (13), which mediates insulin signaling (35). We have been able to confirm that PC-1 is bound to the IR and that the amount of PC-1 bound was much greater in the Q variant than in the K variant. These data suggest that the Q variant is more potent than the K variant in inhibiting the IR because of its stronger association with the IR. The reason for the stronger association between Q¹²¹PC-1 and IR is not known. One possibility is that the Q variant, as a result of differences in cellular or plasma membrane localization, reaches the IR more effectively than the K variant does. Another possibility is that the Q variant has a higher affinity for the IR protein because of major charge differences with respect to K¹²¹PC-1. The exodomain of PC-1 has two somatomedin B–like domains that are part of a cysteine-rich domain (9). The Q variant mutation occurs in the somatomedin B domain. A similar somatomedin B domain occurs in vitronectin, a protein that binds to and changes the conformation (and activity) of the protein plasminogen activator inhibitor-1 (36). By analogy, it is possible that the somatomedin B domain of PC-1 binds to and regulates the IR.

The present data support a cause-and-effect relationship between the Q variant and in vivo insulin resistance (20). Taken together with the previous observations that PC-1 overexpression is associated with insulin resistance (10–13,15–19,33), these data suggest that, in different individuals, PC-1 may impair IR function and insulin sensitivity by two different mechanisms. The first is PC-1 overexpression, which is independent of the type of PC-1 variant overexpressed. The second mechanism, which is caused by Q variant expression, occurs through a more potent inhibition of the IR.

In conclusion, our study confirms the role of the PC-1 K¹²¹Q polymorphism as a genetic determinant of insulin resistance. These studies also suggest that the Q variant of PC-1 is more potent in inhibiting IR function than the K variant because it can more strongly bind to the IR. Further studies defining the contact points between PC-1 and the IR should provide new insights into the regulation of the IR in states of insulin resistance.

This work was supported by a grant of the Società Italiana di Diabetologia (L.F.), grants from Ministero dell’ Università e della Ricerca Scientifica e Tecnologica (MURST) (60% and Cofinanziamento 1997), and grants from the Ministero della Sanità (RC97, RC98, and RF98). L.F. is a recipient of a postdoctoral fellowship of the University of Catania.

We thank B.A. Maddux for her helpful suggestions regarding the IR–PC-1 interaction studies and Dr. Ira D. Goldfine for the critical revision of the manuscript.