INTRODUCTION

The ligand-regulated transmission of signals from the insulin receptor involves the reversible tyrosine phosphorylation of a number of proteins (1). The first is the insulin receptor itself, where the binding of circulating insulin to the receptor -subunit at the surface of insulin's target cells initiates a cascade of tyrosine autophosphorylation in the receptor -subunit. This process activates the tyrosine-specific kinase activity of the receptor -subunit toward several cellular protein substrates, including IRS-1,¹ IRS-2, and Shc. These early steps of insulin action initiate the downstream signals that ultimately generate the pleiotropic effects of insulin on metabolite transport at the plasma membrane, regulation of intermediary metabolic pathways, and the time-delayed effects of insulin on cellular growth and differentiation. These processes are mediated by insulin signaling through the Ras/mitogen-activated protein kinase pathway, activation of cellular phosphatidylinositol kinase activity and the regulation of other distal serine kinases and phosphatases (2).

Since the autophosphorylation state of the activated insulin receptor is short-lived and rapidly reversed once the hormone is withdrawn (3), cellular protein-tyrosine phosphatases (PTPases) have been implicated in the steady state balance of insulin receptor regulation. While a number of candidates have been proposed, the physiologically relevant enzymes for the regulation of insulin action have not been conclusively identified. Particular interest has been generated in PTPases localized to the cell membrane since the bulk of cellular PTPase activity against the insulin receptor is found in the cell particulate fraction, with the highest specific activity in a glycoprotein fraction (4). Additionally, studies from several laboratories have demonstrated that the insulin receptor is rapidly dephosphorylated in situ in permeabilized cells (5, 6), further supporting the role of a membrane PTPase in the regulation of the insulin receptor.

From the large family of PTPase enzymes that have been characterized over the past several years (7), we have provided evidence that the receptor-type PTPase LAR is a major candidate PTPase for the physiological regulation of the insulin signaling pathway. LAR is a large, transmembrane enzyme expressed in the plasma membrane of insulin-sensitive cells (8, 9, 10), and studies with the recombinant LAR catalytic domain have indicated that it has a catalytic preference for the insulin receptor kinase domain in vitro (11). We have also recently demonstrated that reducing LAR abundance in hepatoma cells augments insulin-stimulated receptor autophosphorylation (12); conversely, overexpression of the full-length form of LAR in the plasma membrane attenuates insulin receptor autophosphorylation and insulin signaling (13), thus implicating LAR as a potential regulator of insulin receptor activation under physiological conditions.

In order to begin to characterize the mechanism of regulation of the insulin receptor kinase by LAR in intact cells, we performed studies to examine the functional interaction between LAR and the insulin receptor. LAR has an interesting post-translational itinerary that may have important consequences for regulation of its PTPase activity and/or its association with physiological substrates in target cells (14, 15). LAR is initially synthesized as a ~200-kDa proprotein that is processed at a pentabasic site by a subtilisin-like protease into a complex of two non-covalently associated subunits; the extracellular or E-subunit (150 kDa) contains the cell adhesion molecule domains, and the phosphatase or P-subunit (85 kDa) contains an 82-amino acid extracellular region and the transmembrane and cytoplasmic domains. We initially used an antibody to ligate the extracellular domain of the LAR P-subunit in Chinese hamster ovary (CHO) cells and to assess potential effects on insulin signal transduction. A physical interaction between LAR and the insulin receptor was then examined using co-immunoprecipitation and chemical cross-linking techniques. Finally, LAR was found to be internalized in a temporal response to insulin stimulation into a similar endosomal membrane fraction as the insulin receptor, and neutralizing LAR antibodies were used to demonstrate a role for LAR in the dephosphorylation of the insulin receptor within the endosomal compartment. These results provide further evidence for both a functional as well as a physical interaction between LAR and the insulin receptor.

EXPERIMENTAL PROCEDURES

Male Sprague-Dawley rats, weighing 125-150 g, were purchased from Ace Animals, Inc. (Boyertown, PA) and fed ad libitum.

Native CHO cells were maintained in F-12 nutrient medium (Life Technologies, Inc.). A histidinol-resistant CHO cell line overexpressing the human insulin receptor (CHO-hIR) was generated by cotransfection of the insulin receptor expression plasmid (16) with pCMV-hisD (17) by CaPO₄-mediated DNA transfer (18). This cell line was then cotransfected with the full-length LAR cDNA construct in the pRC/CMV vector (Invitrogen), and stable cell lines overexpressing LAR (CHO-hIR/LAR) were propagated from colonies growing in medium containing 400 µg/ml G418. KRC-7 cells, a well differentiated subline of the H4-EII-C3 rat hepatoma line (19) were kindly provided by Dr. John Koontz (University of Tennessee) and grown in Dulbecco's minimal essential medium. Culture media were supplemented with 10% (v/v) fetal bovine serum, and cells were cultured at 37 °C under a humidified atmosphere of 5% CO₂ in air.

The LAR-82 antibody, directed at the extracellular domain of the rat LAR P-subunit that extends for 82 amino acids beyond the transmembrane segment (9, 15, 20), was generated by immunization of rabbits with a recombinant glutathione S-transferase fusion protein with this protein segment expressed in the pGEX-KG vector (American Type Culture Collection 77103). Glutathione S-transferase-reacting antibodies were removed by passing the serum over an Affi-Gel column (Bio-Rad) coupled to GSH prior to isolation of the specific LAR-82 extracellular domain antibody by binding and elution from an Affi-Gel/LAR-82 domain affinity column (21, 22). Affinity-purified polyclonal antiserum to the cytoplasmic domain of recombinant rat LAR was also generated and used as described (23). Affinity-purified polyclonal antibodies to PTP1B (N-terminal amino acids 1-158) were purchased from Upstate Biotechnology, Inc. (Lake Placid, NY). Polyclonal antibody to the insulin receptor cytoplasmic domain (residues 441-660) was obtained from Transduction Laboratories (Lexington, KY). Antibodies to phosphotyrosine were prepared as described by Kamps (24).

CHO or hepatoma cells were plated onto 10-cm dishes 48 h prior to each experiment. Cells were approximately 80% confluent at the time of experiment and were rendered quiescent by serum starvation overnight. Cells were then incubated for 2 h with 10 µg of LAR-82 antibody/ml of culture medium (or as indicated) containing 0.4% (w/v) bovine serum albumin at 37 °C. Control cells were incubated with control rabbit IgG. After washing twice with fresh culture medium, cells were further incubated with a 1:16 dilution of goat anti-rabbit IgG (Sigma) for 1 h prior to treatment with and without 100 nM insulin for 5 min. Cells were then lysed, and immunoblotting was performed as described (25). Protein was measured by the method of Bradford (26), and electrophoretic protein migration was calibrated with prestained molecular size standards from Bio-Rad.

CHO-hIR were serum-starved overnight prior to incubation for 2 h with 10 µg/ml LAR-82 antibody, washed, incubated with a 1:16 dilution of goat anti-rabbit IgG (Sigma) for 1 h, and then treated with 0-100 nM insulin for 5 min. Cells were then lysed into buffer containing protease and PTPase inhibitors (25), and the solubilized extracts were normalized for protein content prior to immunoprecipitation of the insulin receptors with monoclonal human insulin receptor antibody 83-7 (kindly provided by Dr. Ken Siddle, University of Cambridge) and rabbit anti-mouse IgG second antibody complexed with Trisacryl protein A. Immunoprecipitated insulin receptors were incubated with 0.2 µg of recombinant rat IRS-1 protein expressed from a baculovirus system (gift from Dr. Morris White, Joslin Diabetes Center) as substrate for the receptor kinase in a reaction containing 1 mM orthovanadate, 10 mM MnCl₂, and 5 µCi of [-³²P]ATP for 5 min (27). Reactions were performed in duplicate and terminated by the addition of Laemmli gel sample buffer, and proteins were separated by polyacrylamide gel electrophoresis. The gels were treated for 1 h at 60 °C in 1 N NaOH prior to drying and phosphorimager analysis.

Chemical cross-linking with the thiol-cleavable cross-linker 3,3-Dithiobis(sulfosuccinimidylpropionate) (DTSSP; Pierce) was performed at a concentration of 1 mg/ml for 30 min at 4 °C as described (28). In some experiments, cells were also treated with 100 nM insulin or 1 µM biotinylated insulin (Sigma) for 5 min at 37 °C prior to cross-linking. After cell lysis (25), LAR or insulin receptor protein complexes were immunoprecipitated by the addition of 6 µg of either affinity-purified polyclonal antibody to the LAR extracellular domain (LAR-82) or insulin receptor polyclonal antibody (Transduction Laboratories) at 4 °C overnight. Immune complexes or receptor bound to biotinylated insulin were collected by the addition of 50 µl of a suspension of either Trisacryl-protein A beads (Pierce) or avidin-agarose beads (Sigma), respectively, prior to washing and immunoblotting (25). Quantitation of the amount of antigen protein remaining in the lysate supernatant and in the protein A pellet after immunoprecipitation showed that the efficiency was 70-80% for the insulin receptor and 70-75% for LAR.

Rats fasted overnight were anesthetized prior to portal vein injection of saline alone or 1.5 µg of crystalline human insulin (Sigma) in saline/100 g of body weight. Following the injection, the liver endosome fraction was prepared as described by Khan et al. (29). The minced livers were homogenized (5 ml/g tissue) using five strokes of a Potter-Elvehjem homogenizer fitted with a Teflon pestle rotating at 1500 rpm. The homogenate was centrifuged at 3,200 × g for 10 min to remove nuclei and mitochondria. This supernatant was then centrifuged at 200,000 × g for 30 min to obtain a mixed particulate fraction, which was resuspended with one stroke of a Dounce homogenizer into 1.15 M sucrose buffered with 4 mM imidazole, pH 7.4 (5 ml/g of original tissue weight) containing the protease inhibitors noted above. A discontinuous gradient with layers of 0.6 M and 1.0 M sucrose was constructed above the resuspended pellet. After centrifugation at 202,000 × g for 98 min in a Beckman SW Ti40 rotor, the endosomal fraction at the interface between 0.6 and 1.0 M sucrose was isolated, diluted with the homogenization buffer, pelleted by centrifugation at 200,000 × g for 35 min, and resuspended into 2 ml of homogenization buffer. Identical methodology was used for the isolation of an endosomal membrane fraction from KRC-7 hepatoma cells. Additional samples of treated rat livers were used to prepare a purified plasma membrane fraction by the method of Hubbard et al. (30).

Using the isolated endosomal fraction of rat liver, insulin receptor autophosphorylation, treatment of endosomes with neutralizing antibodies to LAR or PTP1B and receptor dephosphorylation was performed essentially as described for the partially purified insulin receptor in our previous work (23).

RESULTS

Our previous work demonstrated that manipulating the cellular abundance of LAR had a significant impact on insulin action, by affecting the proximal steps of receptor autophosphorylation and activation of the receptor kinase activity. These studies suggested that this PTPase interacts either directly or indirectly with the insulin receptor at the cell membrane. To investigate the interaction more closely, we generated an affinity-purified antibody toward the extracellular protein segment of the transmembrane "P" subunit of LAR, and used this antibody (termed LAR-82) to ligate the LAR protein in the plasma membrane of intact CHO-hIR/LAR and KRC-7 cells. The LAR-82 antibody reacts with a single band of the expected size of 85 kDa on immunoblot analysis of KRC-7 cell lysates in proportion to the amount of protein loaded on the gel and transferred for blotting (Fig. 1A). The specificity of the LAR-82 antibody was also demonstrated by immunoprecipitation of KRC-7 cell lysates with antibody to the LAR cytoplasmic domain or control IgG, followed by immunoblotting with the LAR-82 antibody, which further identified the 85-kDa band as the LAR P-subunit (Fig. 1B).

Ligation of the LAR P-subunit at the cell surface by a 2-h incubation of CHO-hIR/LAR cells with the LAR-82 antibody reduced the level of insulin-stimulated tyrosine phosphorylation of the insulin receptor to 56% of the level found in control cells incubated with control IgG, and phosphorylation of IRS-1 was reduced to 69% of control (Fig. 2A). These results indicated that the binding of antibody to the extracellular domain of LAR affected the receptor autophosphorylation and tyrosine kinase activity. In order to further cluster the LAR P-subunits, following incubation with LAR-82 antibody, the cells were washed and incubated with a second anti-rabbit IgG antibody for 1 h (Fig. 2A). Insulin-stimulated tyrosine phosphorylation of the insulin receptor was then measured and found to be reduced further to 35% of the control level, and was associated with an additional decrease in the phosphorylation of IRS-1 to 49% of control, in this experiment. Incubation of control cells treated with the control IgG with the second antibody had no effect on the insulin-stimulated phosphorylation events. These findings indicated that cross-linking or clustering the LAR protein at the cell surface in a complex with both the LAR-82 antibody and the second antibody had a more substantial impact on the function of the insulin receptor than did treatment with the LAR-82 antibody alone.

IR

In KRC-7 hepatoma cells, incubation with the LAR-82 antibody and the second anti-IgG antibody also demonstrated a significant reduction in insulin-stimulated tyrosine phosphorylation of the insulin receptor and IRS-1 (Fig. 2B). A summary of the results from several experiments is compiled in Fig. 2C, indicating that the LAR-82 antibody treatment reduced the amount of insulin-stimulated tyrosine phosphorylation of the insulin receptor and IRS-1 to 53-63% of control.

Additional studies revealed that the effect of LAR-82 antibody on insulin receptor and IRS-1 tyrosine phosphorylation in the CHO-hIR/LAR cells was nearly maximal at a concentration of 10 µg/ml (Fig. 3A). In addition, inclusion of recombinant LAR-82 protein used in the purification of the antibody during the incubation of CHO-hIR/LAR cells with LAR-82 completely abrogated the effect of LAR-82 on insulin-stimulated phosphorylation (data not shown). In addition, there was no change in the binding of radiolabeled insulin or the abundance of insulin receptors or LAR by immunoblotting in either the CHO-hIR/LAR or KRC-7 cells after the incubation with LAR-82 antibody or the second antibody incubations.

To examine whether the LAR-82 antibody had a direct effect on the insulin receptor kinase activity, insulin receptors were immunoprecipitated from lysates of antibody-treated CHO-hIR/LAR cells that has been incubated with insulin for 5 min, and the receptor kinase activity was measured using exogenously added recombinant IRS-1 as substrate (Fig. 3B). At each insulin concentration from 1 to 100 nM, treatment with LAR-82 antibody resulted in a 33-39% decrease in insulin-stimulated receptor kinase activity toward IRS-1 compared to the level obtained in control cells. Thus, the reduced phosphorylation of IRS-1 in the intact cells treated with LAR-82 is due primarily to diminished insulin receptor kinase activation induced by antibody binding to the LAR P-subunit in the cell membrane.

Since we have previously demonstrated that LAR can also negatively regulate the EGF receptor in a hepatoma cell line (27), we also examined whether incubation of KRC-7 cells with the LAR-82 antibody affected tyrosine autophosphorylation of the EGF receptor. Using conditions identical to those for the insulin receptor studies above, the LAR-82 antibody significantly decreased EGF-stimulated autophosphorylation of its receptor (Fig. 4). In additional experiments, EGF receptor autophosphorylation was decreased to 50.4% of the control value at an antibody concentration of 10 µg/ml (p < 0.05). Incubation with control IgG alone was without effect, and the level of EGF receptor protein was unchanged following incubation with LAR-82 antibodies (data not shown). Thus, in parallel to our previous studies using expression of LAR antisense mRNA to reduce its protein expression, the reduction of receptor autophosphorylation by incubation with the LAR-82 antibody involves both the insulin receptor and the EGF receptor.

Ligation of the extracellular LAR P-subunit with the LAR-82 antibody in vitro using lysates of transfected CHO cells did not affect the LAR PTPase enzyme activity, since the hydrolysis of phosphotyrosyl-lysozyme (31) was similar when measured with the LAR P-subunit immunoisolated on Trisacryl-protein A beads or when the enzyme was released into solution by incubation with excess LAR-82 protein (data not shown). In addition, incubation of intact cells with the LAR-82 antibody had no apparent effect on the morphology or adhesion of either CHO or KRC-7 cells.

Initially, cross-linking of LAR with the insulin receptor was performed in CHO cells overexpressing the insulin receptor before and after transfection with the full-length LAR cDNA. After lysis of the cells into extraction buffer, insulin receptors were immunoprecipitated with insulin receptor antibody, and immunoblotting was performed with antibody to LAR, which demonstrated a small amount of LAR P-subunit associated with the insulin receptor in the immune complexes (Fig. 5A). The LAR-insulin receptor association was found to be dependent on the level of cellular LAR expression. In the CHO-hIR cells transfected with the LAR cDNA, there was a 4.3-fold increase in the amount of LAR present in the insulin receptor immunoprecipitates. Cell monolayers were subsequently treated with a 1 mg/ml amount of reducible, membrane-impermeant, and water-soluble cross-linker DTSSP, which increased the amount of LAR complexed with the insulin receptor in the CHO-hIR cells (Figs. 5 and 6). Compared to control cells, treatment with DTSSP increased the LAR in the insulin receptor immunoprecipitates by a mean of 6.5-fold. In the CHO-hIR cells overexpressing LAR, the effect of cross-linking with DTSSP was also proportionally increased.

Similar results were obtained in untransfected KRC-7 cells, where immunoprecipitation of cell lysates with insulin receptor antibody and immunoblotting with antibody to LAR revealed a small amount of the LAR P-subunit associated with the insulin receptor in the immune complexes (Fig. 5B). Chemical cross-linking with DTSSP significantly increased the amount of LAR complexed with the insulin receptor in the KRC-7 cells by an average of 5.1-fold (Figs. 5B and 7). Comparison between the two cell types provided further evidence for a dependence of the measured association on the abundance of LAR protein expression, since KRC-7 hepatoma cells, which have 2.5 times more LAR protein than the CHO cells, exhibited 2.1 times more LAR in the insulin receptor immunocomplexes following chemical cross-linking with DTSSP.

To further substantiate the association of LAR with the insulin receptor, the order of antibodies used for immunoprecipitation and blotting was reversed and the amount of insulin receptor protein present in immunoprecipitates prepared from the cell lysates with LAR-82 antibody was measured (Fig. 6B). With DTSSP cross-linking, the amount of insulin receptor protein present in the LAR antibody immunoprecipitates was increased by 6.5-fold. Transfection of CHO-hIR cells with the LAR cDNA also increased the amount of insulin receptor protein present in the LAR immunocomplexes by 2.0-fold (data not shown).

We postulated that the interaction of LAR with the insulin receptor involved recognition of the autophosphorylated form of the receptor as a substrate for the LAR PTPase catalytic domain, and that the affinity of this reaction or the extent of the LAR/insulin receptor association might be enhanced by insulin stimulation. Indeed, we found that activation of the insulin receptor significantly increased the association of LAR with the insulin receptor. Initially, CHO-hIR/LAR cells preincubated with 100 nM insulin at 20 °C for 90 min were found to have a mean 3.9-fold increase in the amount of LAR in insulin receptor immunocomplexes (Fig. 6A). In complementary experiments, a 4.3-fold increase in the amount of insulin receptor protein was found in the cross-linked LAR immunocomplexes (Fig. 6B). When the cells preincubated with insulin were also chemically cross-linked with DTSSP, the amount of LAR in insulin receptor immunocomplexes was dramatically increased to a mean of 13.5 times the control level and a 11.5-fold increase was observed in the amount of insulin receptor protein in the LAR immunocomplexes. In control experiments, we also confirmed that after insulin treatment or cross-linking with DTSSP, there was no change in the amount of insulin receptor in the receptor immunoprecipitates (Fig. 6A) or in the amount of LAR in the LAR immunoprecipitates (Fig. 6B), as measured by immunoblotting with the respective antibodies. In KRC-7 hepatoma cells, insulin treatment also increased the amount of LAR in the insulin receptor immunocomplexes by 2.0-fold, which was further increased in the cells also treated with DTSSP to 5.7 times the level observed in the control cells (Fig. 5B). These collective results are summarized in Fig. 7.

The amount of LAR complexed with the insulin receptor following chemical cross-linking with DTSSP was measured in CHO-hIR cells before and after transfection with the LAR cDNA (Fig. 8). In the CHO-hIR cells, data obtained by immunoblotting with the insulin receptor antibody using samples of total cell lysate as well as aliquots that were immunoprecipitated with the LAR-82 antibody showed that 1.5% of the total IR protein was complexed with LAR in the immunoprecipitates. Following transfection with the LAR cDNA, there was no change in the total amount of insulin receptor protein, but the amount of insulin receptor in the LAR complexes increased to 8.7% of the total. Analysis of the fraction of the total amount of LAR present in the cross-linked insulin receptor immunocomplexes was also performed (Fig. 8). Prior to transfection of the CHO-hIR cells with the LAR cDNA, 13.8% of the total LAR was isolated in the insulin receptor immunocomplexes. Following transfection with the LAR cDNA, there was a 6.0-fold increase in the total LAR protein in the cell lysates. In the insulin receptor immunocomplexes from the LAR transfected cells, 11.8% of the total LAR protein was found, representing an overall increase of 5.1-fold over the amount in the untransfected cells, proportionally similar to the measured increase in LAR protein abundance.

In further studies, we examined whether the catalytic activity of LAR was essential for its ability to recognize and associate with the insulin receptor. We blocked the catalytic activity of endogenous PTPases by pretreatment of CHO-hIR cells transfected with the LAR cDNA with 2 mM vanadate. As in published studies on other PTPase enzymes, our previous work demonstrated that vanadate is an effective inhibitor of the LAR PTPase with the autophosphorylated insulin receptor in vitro (9). In the intact cells, vanadate significantly potentiated the tyrosine phosphorylation state of both the insulin receptor and IRS-1 compared to untreated cells (Fig. 9). However, pretreatment of the cells with vanadate did not affect the extent of chemical cross-linking of LAR and insulin receptor, suggesting that the site of interaction between these two proteins may not require an active PTPase catalytic site.

The specificity of the LAR association with the insulin receptor was also examined by testing for chemical cross-linking between the EGF receptor and LAR in the KRC-7 hepatoma cells. A specific antibody (from Upstate Biotechnology, Inc.) was used to immunoprecipitate epidermal growth factor receptor from KRC-7 cell lysates after cross-linking of the cell surface with DTSSP. Immunoblotting of the EGF receptor immunoprecipitates with the LAR antibody failed to demonstrate an association between LAR and the EGF receptor in the hepatoma cells (data not shown). Control experiments showed that the EGF receptor has a much lower level of expression than the insulin receptor in these cells, requiring immunoprecipitation to detect the EGF receptor protein by immunoblotting, while the insulin receptor can be detected by direct immunoblotting of cell lysates (as in Fig. 8).

After insulin stimulation in hepatic cells, insulin receptors are rapidly internalized into an endosomal membrane compartment (32), which has been postulated to be an important site of receptor dephosphorylation as the receptors are recycled back to the plasma membrane in the basal state (33, 34, 35, 36). We hypothesized that the PTPases responsible for the physiological dephosphorylation of the insulin receptor in the endosomal compartment would be coupled to the movement of the receptor and interact with the receptors in the endosomal fraction. Initial studies of this effect were performed to observe whether LAR is found in hepatic endosomes and whether it was functionally coupled to the insulin receptor in this compartment.

In insulin-stimulated KRC-7 hepatoma cells, LAR was enriched in the endosomal fraction with a time course similar to that seen with the insulin receptor (Fig. 10). In phosphorimager units, stimulation with insulin increased the abundance of the insulin receptor in the endosomes by a mean of 77% at 30 min and by 45% at 60 min, with a concomitant increase in the mass of LAR by 119% at 30 min and by 98% at 60 min.

In subsequent experiments, the coordinated decrease of LAR in the plasma membrane fraction and increase of LAR in the endosomal fraction was followed in insulin-stimulated intact rat liver over a 30-min time course (Fig. 11). In the basal state, the ratio of LAR in the plasma membrane fraction compared to the endosomes was 1.6:1.0 based on aliquots of each fraction containing the same amount of protein. After insulin stimulation for 15 or 30 min, the mass of LAR in the plasma membrane fraction was diminished by a mean of 43% and 49%, respectively, compared to the control sample. At the same time, a concomitant increase in LAR mass of 4.1- and 5.7-fold, respectively, was observed in the endosomal fraction, suggesting an actual translocation of the LAR protein to the internal membranes with insulin stimulation.

To evaluate the functional role of LAR in the dephosphorylation of insulin receptors in the endosomal compartment, we used an inhibitory antibody to block the enzymatic activity of LAR on the insulin receptor. Insulin receptors present in endosomes isolated from rat liver were first autophosphorylated in situ, and the course of dephosphorylation of the receptor -subunit was followed after incubation of the endosomes with anti-LAR antibody or control IgG (Fig. 12). In the presence of the control antibody, the insulin receptors were dephosphorylated to 17.7 ± 2.5% (mean ± S.E.; n = 4) of the initial level during the 30-min reaction period. In the endosome sample incubated with the LAR antibody, there was 28% decrease in the dephosphorylation of the insulin receptor, to 41.0 ± 6.0% of the initial level of -subunit phosphorylation (p = 0.01 versus control by Student's t test).

For comparison, experiments were performed in parallel with an antibody to the catalytic domain of PTP1B that neutralizes the enzyme activity of this PTPase (25). PTP1B is an intracellular PTPase associated with internal cellular membranes (37) that has also been implicated in the regulation of the phosphorylation state of the insulin receptor (25). In the endosomes incubated with the inhibitory PTP1B antibody, the insulin receptors were dephosphorylated to 19.7 ± 5.3% (n = 4) of the initial level during the 30-min reaction period. This result was not significantly different from the value obtained from the incubation of the endosomes with control IgG.

DISCUSSION

Our laboratory has been particularly interested in identifying the PTPases involved in the regulation of cellular insulin action. Among the large family of membrane-linked PTPases, evidence from a variety of experimental approaches has suggested that the widely expressed, transmembrane PTPase LAR is a major candidate for the physiological regulation of the insulin receptor (4). In addition to its tissue distribution and membrane localization, biochemical studies have demonstrated that the cytoplasmic domain of LAR has a catalytic preference for the regulatory phosphotyrosines in the insulin receptor kinase domain (11). We have also obtained data from cell transfection studies that support the hypothesis that LAR acts as an important negative regulator of insulin signaling in intact cells by altering the rate and extent of dephosphorylation of the insulin receptor and the associated deactivation of its kinase enzyme activity. These experiments have included the use of LAR antisense mRNA expression in hepatoma cells to reduce the LAR mass, which significantly enhanced insulin receptor autophosphorylation, receptor kinase activity, and insulin-stimulated phosphatidylinositol 3-kinase activity (12). Recently, we have also performed complementary experiments using overexpression of the full-length LAR protein in both hepatoma cells and CHO cells, which significantly attenuated the insulin receptor autophosphorylation and kinase activity (13, 38). Interestingly, in these studies, overexpression of a catalytically active, soluble cytoplasmic domain of LAR in the cell cytosol had no measurable effect on insulin receptor activation or downstream signaling by insulin, providing strong additional support for the notion that a transmembrane localization is critical for LAR action in situ. Taken together, these studies strongly suggest that LAR is a physiological modulator of insulin signaling at a proximal site, most likely involving direct effects within the plasma membrane between LAR and insulin receptor itself.

The present work was undertaken to further characterize the physical and functional interactions between LAR and the insulin receptor within the membrane fraction of an intact cell system. The technique of antibody ligation has been used in a number of studies to demonstrate that manipulation of the extracellular domains of receptor-like PTPases in intact cells, in particular CD45, can significant affect signaling pathways in which the target PTPase plays an important regulatory role. For example, binding of CD45 antibodies to the surface of leukocytes, with and without additional cross-linking with a second antibody directed at the anti-CD45 antibody, can influence various signal transduction pathways, including those affecting the activity of several cellular tyrosine kinases, the expression of cell surface markers, and cell functions involving lymphocytes or other specialized hematopoietic cells (39, 40).

Similar to the reported finding with CD45 in hematopoietic cells, incubation of insulin-sensitive cells with an antibody to the extracellular segment of LAR (LAR-82) caused a significant decrease in insulin receptor autophosphorylation, receptor kinase activity, and insulin-stimulated IRS-1 phosphorylation after insulin treatment in both CHO-hIR cells and KRC-7 rat hepatoma cells. This effect was further potentiated by clustering the LAR-82 antibody with a second antibody directed at the LAR-82 IgG bound at the cell surface. The influence of LAR antibody binding on receptor function did not alter insulin binding to the cell surface insulin receptors. While the mechanism of the effect of LAR-82 antibodies on receptor function in situ is not known, studies using the LAR P-subunit in solution suggest that the LAR-82 antibody does not simply enhance the PTPase catalytic activity. Since the autophosphorylation of the EGF receptor is also affected by the binding of extracellular LAR antibody (Fig. 4), our results support our previous finding that LAR has a functional interaction with the insulin receptor as well as other tyrosine kinase receptors on the cell surface (27). While the antibody did not affect cell morphology or adhesion, as might be expected if it influenced the function of LAR in focal adhesions (41), the data suggest that the LAR-82 antibody alters the mobility or conformation of LAR in a way that influences its association with cell surface receptor substrates.

The hypothesis that LAR physically interacts with the insulin receptor was further supported by studies using chemical cross-linking at the surface of CHO cells and hepatoma cells. Although a small amount of LAR was isolated in insulin receptor immunoprecipitates from CHO-hIR cells without cross-linking, treatment with DTSSP greatly enhanced the LAR/insulin receptor interaction. DTSSP has also been used to demonstrate an association between CD45 and the cell surface receptor Thy-1 by chemical cross-linking on the surface of thymocytes (42).

The physical association between LAR and the insulin receptor was also dependent on the amount of LAR and insulin receptor protein expressed at the cell surface. Of the total LAR mass in the cell lysates, the cross-linked insulin receptor immunoprecipitates contained 11.8-13.8% depending on the cell system and the level of LAR expression. Since LAR and the insulin receptor are both found in internal membrane fractions as well as on the cell surface, these results may underestimate the fraction of LAR associated with the insulin receptor when only the cell surface is subjected to cross-linking.

One of the most interesting results obtained in this study is that insulin treatment enhances the physical interaction between LAR and the insulin receptor in cells. This observation suggests a dynamic role for ligand-induced conformational changes in the insulin receptor or in its phosphorylation state that enhances its association with LAR as a negative-regulatory PTPase. Our finding that treatment of the cells with the PTPase inhibitor vanadate, which dramatically enhances the phosphorylation state of the insulin receptor, had no demonstrable effect on the association of LAR with the insulin receptor, suggests that the interaction between these proteins may not be primarily between the PTPase catalytic domain and the phosphotyrosyl residues of the activated insulin receptor -subunit. Since substrate dephosphorylation by PTPases is extraordinarily rapid (we have estimated a turnover number for LAR of 8990 nmol of P_i released/min/nmol of enzyme in previous work; Ref. 9), additional domains of protein-protein interaction are likely to be necessary to enhance and stabilize the association between these two proteins.

Although we were unable to demonstrate co-immunoprecipitation of LAR and the EGF receptor even after chemical cross-linking, a functional association of LAR with the EGF receptor was evidenced by the effect of the LAR-82 antibody incubation of EGF receptor autophosphorylation (Fig. 4). These results are consistent with our previous data showing that LAR is apparently involved in the regulation of multiple tyrosine kinase receptors at the cell surface (27).

Other evidence for an association between LAR and the insulin receptor was obtained by demonstrating that LAR is internalized into an endosomal compartment of insulin-stimulated liver cells in a close temporal relationship with the insulin receptor. The internalization of the insulin receptor is known to be associated with dynamic changes in its phosphorylation state in liver cells and adipocytes (33, 43, 44). Posner and colleagues have provided evidence that internalization of the insulin receptor in liver is not only involved in the deactivation of the receptor kinase activity by dephosphorylation (36), but also that the dynamic nature of the internalization machinery may transiently increase the receptor kinase activity (by as much as 3-5-fold) by a mechanism involving partial dephosphorylation of the receptor -subunits (29, 35). In adipocytes, Kublaoui et al. (43) demonstrated that receptor dephosphorylation and inactivation may occur in the endosomal fraction of adipocytes. Our finding that incubation of liver endosomes with neutralizing LAR antibodies partially, although significantly, inhibited insulin receptor dephosphorylation in situ suggests that LAR plays a role in the dynamic changes in insulin receptor tyrosine phosphorylation that occur within the interior of the cell.

In published work, evidence has been provided to suggest that additional PTPases may also be involved in the regulation of the insulin action pathway. Initially, microinjection of PTP1B, a widely expressed, single-domain intracellular PTPase, into Xenopus oocytes was shown to block insulin signaling (45, 46). In cultured cells, overexpression of PTP1B resulted in significant dephosphorylation of insulin and insulin-like growth factor-1 receptors (47) and can negatively influence insulin action (48). We have recently demonstrated, by osmotic loading of neutralizing antibodies into insulin-sensitive hepatoma cells, that PTP1B can negatively regulate insulin receptor activation and distal signaling (25). Since PTP1B has a role in the negative regulation of insulin signaling and acts, at least in part, directly at the level of the receptor kinase, it may function in concert with LAR in the physiological regulation of the insulin receptor.

LRP (PTP-) is another widely expressed receptor-type PTPase found in insulin-sensitive tissues that has also been shown to catalyze the dephosphorylation of the insulin receptor and the inactivation of the receptor kinase (11). Recently, using a novel assay for evaluation of transfected cDNAs that negatively affect insulin action, Moller et al. (49) also provided evidence that LRP as well as the closely related transmembrane enzyme PTP- can serve as negative regulators of the insulin receptor tyrosine kinase. In other studies, LRP has been shown to have other cellular effects, including activation of pp60^c-src (50), triggering of a neuronal differentiation pathway (51), and attenuation of GRB-2-mediated signaling (52). Clearly, additional work will be required to delineate the potential role of each of these PTPases in the regulation of the insulin signaling pathway.

The role of specific PTPases in human states of insulin resistance has also been recently studied. We found that the PTPase activity toward the insulin receptor is increased in skeletal muscle and adipose tissue of obese subjects and is correlated linearly with the body mass index (23).² Furthermore, weight loss in obese subjects, accompanied by an improvement in tissue insulin sensitivity, decreased the adipose tissue PTPase activity and reduced the abundance of LAR and PTP1B, further supporting the hypothesis that these specific PTPases may be reversibly involved in the pathogenesis of clinical insulin resistance.³ Understanding the regulation of insulin action by PTPases at a molecular level will provide insight into the pathological derangements found in insulin-resistant disease states, and possibly reveal novel ways in which they may be ameliorated by pharmaceutical agents in the treatment of these disorders.

In summary, by providing insight into the close functional and physical interaction between LAR and the insulin receptor at the cell surface, the present results substantiate our previous data that LAR can function in intact cell systems as a negative regulator of insulin signaling. We now demonstrate that LAR can impact on the function of the insulin receptor in situ at the plasma membrane, LAR is physically associated with the receptor in a manner that is stimulated by insulin binding, and LAR appears to be temporally internalized into a similar endosomal compartment as the insulin receptor where it can function as an active PTPase against the autophosphorylated receptor. Further studies will help delineate the molecular sites of interaction between LAR and the insulin receptor and the exact mechanism of regulation of insulin action by LAR.