Dynamics of protein-tyrosine phosphatases in rat adipocytes.

Protein-tyrosine phosphatases (PTPases) play a key role in maintaining the steady-state tyrosine phosphorylation of the insulin receptor (IR) and its substrate proteins such as insulin receptor substrate 1 (IRS-1). However, the PTPase(s) that inactivate IR and IRS-1 under physiological conditions remain unidentified. Here, we analyze the subcellular distribution in rat adipocytes of several PTPases thought to be involved in the counterregulation of insulin signaling. We found that the transmembrane enzymes, protein-tyrosine phosphatase (PTP)-alpha and leukocyte common antigen-related (LAR), were detected predominantly in the plasma membrane and to a lesser extent in the heavy microsomes, a distribution similar to that of insulin receptor. PTP-1B and IRS-1 were present in light microsomes and cytosol, whereas SHPTP2/Syp was exclusively cytosolic. Insulin induced a redistribution of PTP-alpha from the plasma membrane to the heavy microsomes in a parallel fashion with the receptor. The distribution of PTP-1B in the light microsomes from resting adipocytes was similar to that of IRS-1 as determined by sucrose velocity gradient fractionation. Analysis of the catalytic activity of partially purified rat adipocyte PTP-alpha and LAR and recombinant PTP-1B showed that all three PTPases dephosphorylate IR. When a mix of IR/IRS-1 was used as a substrate, PTP-1B was particularly effective in dephosphorylating IRS-1. Considering that IR and IRS-1 can be dephosphorylated in internal membrane compartments from rat adipocytes (Kublaoui, B., Lee, J., and Pilch, P.F. (1995) J. Biol. Chem. 270, 59-65) and that PTP-alpha and PTP-1B are the respective PTPases in these fractions, we conclude that these PTPases are responsible for the counterregulation of insulin signaling there, whereas both LAR and PTP-alpha may act upon cell surface insulin receptors.

The initial steps in insulin signaling require protein tyrosine phosphorylation. The interaction of insulin with its plasma membrane receptor elicits a rapid autophosphorylation on specific tyrosine residues in several domains of the receptor's cytoplasmic ␤-subunit followed by activation of the receptor's exogenous tyrosine kinase activity as reflected by enhanced substrate phosphorylation (reviewed in Refs. 1 and 2). The principle insulin receptor substrates (IRSs) 1 involved in meta-bolic regulation are IRS-1 and IRS-2, and there is abundant biochemical evidence to support this (reviewed in Ref. 3). Genetic evidence from animals devoid of IRS-1 (4,5) and IRS-2 (6) further suggests that these molecules are intimately involved in the signaling to metabolic pathways regulated by insulin. Two additional members of the IRS family, IRS-3 and IRS-4, have been described, but they are unlikely to be involved in insulin-dependent metabolic regulation because the former is not expressed in the appropriate tissues (7) and a knockout of the latter has no obvious metabolic phenotype (8). IRS-1 and IRS-2 contain numerous tyrosine residues that, when phosphorylated, act as docking sites for a certain set of Src homology 2 domain-containing proteins (3). In particular, it is the activation of PI-3 kinase in association with tyrosine-phosphorylated IRS-1 and IRS-2 that propagates the insulin signaling cascade required for most or all of insulin's pleiotropic actions including metabolic regulation (9).
Attenuation of insulin action is mediated by protein-tyrosine phosphatases (PTPases), which dephosphorylate and deactivate the insulin receptor and post-receptor substrates. PTPases comprise a large family of transmembrane and intracellular enzymes that can act as either positive or negative regulators of signaling pathways (10,11). Several PTPases have been shown to be highly expressed in the major insulin-sensitive tissues, including liver, skeletal muscle, and adipose tissue, and these include LAR, PTP-␣, PTP-1B, and SHPTP2/Syp. Evidence that these particular enzymes are involved in insulin signaling comes from a variety of experimental approaches as has been comprehensively reviewed (12)(13)(14). Thus, the association LAR and PTP-1B with the insulin receptor and SHPTP2/ Syp with IRS-1 has been shown by co-immunoprecipitation; negative effects on insulin signaling have been demonstrated as a result of overexpressing wild-type enzyme, and positive effects on insulin signaling result from expressing dominant negative (inactive) enzymes. Recently, gene knockouts for LAR (15) and PTP-1B (16) have also implicated these enzymes in insulin signaling. However, a consensus has not emerged as to whether any one or all of these PTPases play a physiological role in reversing the insulin signal in its target cells under normal physiological conditions. We reasoned that for a specific PTPase to affect IR or IRS-1 function, both enzyme and substrate must be in the same place in the cell, at least transiently. Moreover, we (17) and others (18 -20) have shown that insulin signaling is a dynamic process that may involve endocytosis of the active receptor as a means of delivering kinase activity to substrates, the latter also having spatial constraints for specific signaling to occur. With this in mind, we determined the intracellular localization of the PTPases PTP-␣, LAR, SHPTP2/Syp, and PTP-1B as well as IR and IRS-1 in freshly isolated rat adipocytes in the presence or absence of insulin. In addition, we tested if the PTPases men-tioned above have a preferential substrate specificity for IR or IRS-1.

EXPERIMENTAL PROCEDURES
Materials-Collagenase and bovine serum albumin were purchased from Roche Molecular Biochemicals. The BCA protein determination kit was from Pierce. Aprotinin, leupeptin, and pepstatin A were obtained from American Bioanalytical (Natick, MA). Electrophoresis chemicals were purchased from National Diagnostics (Atlanta, GA).
[␥-32 P]ATP was from NEN Life Science Products. Anti-IRS-1 and anti-SHPTP2/Syp antibodies were from Upstate Biotechnology, Inc. (Lake Placid, NY). Anti-PTP-1B antibody was obtained from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA). Anti-IR antibody was a generous gift from Dr. Jongsoon Lee (Joslin Diabetes Center, Boston, MA). Anti-R-PTP-␣ antibody was kindly provided by Dr. Jan Sap (Dept. of Pharmacology, New York University). Anti-LAR antibody (21) was a generous gift from Dr. Qiang Yu (Pulmonary Center, Department of Medicine and Department of Biochemistry, Boston University School of Medicine).
Adipocyte Isolation and Fractionation-Adipocytes were isolated by collagenase digestion from epididymal fat pads of 150 -175-g male Harlan Sprague Dawley rats (Taconic Farms, Germantown, NY) as described previously (22). Isolated cells were allowed to equilibrate at 37°C for 30 min with Krebs-Ringer buffer containing 12.5 mM Hepes, pH 7.4, 120 mM NaCl, 6 mM KCl, 1.2 mM MgSO 4 , 1 mM CaCl 2 , 1 mM Na 2 HPO 4 , 2.5 mM D-glucose, and 2% bovine serum albumin. Insulin from bovine pancreas (Sigma) was added to the cell suspension for 10 min to a final concentration of 10 nM. Cells were washed two times and homogenized in 20 mM HEPES, 250 mM sucrose, 1 mM EDTA 5 mM benzamidine, 1 mM phenylmethylsulfonyl fluoride, 1 M pepstatin A, 1 M aprotinin, and 1 M leupeptin, at pH 7.4. Subcellular fractionation was performed as described (17). Protein was quantitated using the BCA assay with bovine serum albumin as a standard.
Fractionation of Microsomes in a Sucrose Velocity Gradient-Light microsomes (1-2 mg of protein) were resuspended in HES buffer, pH 7.4, including protease inhibitors, and an equal amount of protein from each treatment was loaded onto the continuous sucrose gradient, ranging from 10 to 35% (w/v). Membranes were centrifuged at 100,000 ϫ g for 55 min at 4°C (23). Fractions (approximately 200 l) were collected from the gradient, and protein content was measured.
Western Blot (Immunoblot) Analysis-Samples were separated by SDS-PAGE (polyacrylamide gel electrophoresis) on 8 or 10% acrylamide gels run according to Ref. 24 and then electrophoretically transferred to polyvinylidene difluoride Immobilon membrane (0.2 m, Bio-Rad). The membranes were incubated in PBST (phosphate-buffered saline plus 0.1% Tween 20) containing 5% nonfat dry milk for 1 h at 25°C to block nonspecific binding and then probed with polyclonal antibodies to LAR, R-PTP-␣, IR, IRS-1, and PTP-1B or with monoclonal antibodies to SHPTP2/Syp and GLUT4 (25). To detect the antigen-bound antibody, the blots were treated with secondary antibody conjugated with horseradish peroxidase. Immunoreactivity was detected using the ECL detection system (Amersham Pharmacia Biotech). Quantitative analysis was performed using a scanning densitometer (Molecular Dynamics, Inc., Sunnyvale, CA).

Purification of PTPases by a Mono-Q Fast Performance Liquid Chromatography (FPLC) Ion Exchange
Chromatography-Solubilized plasma membrane (2-3 mg of protein) was subjected to FPLC with a Mono Q column (HR5/5; Amersham Pharmacia Biotech). The column was equilibrated with 20 mM Tris buffer (pH 7.0) containing 0.1% Triton and protease inhibitors and developed at a flow rate of 0.3 ml/min. Fractions of 1 ml were eluted with a linear gradient of 50 -1000 mM NaCl. Positions of the transmembrane phosphatases LAR and PTP-␣ were determined by Western blotting, and fractions were pooled and concentrated using a microfiltration unit, Centricon-10 (Amicon, Beverly, MA).
Receptor Purification-Insulin receptor was purified from NIH-3T3 cells transfected with human insulin receptor using wheat germ agglutinin-agarose chromatography as described (26).
Insulin Receptor Kinase Activity-Partially purified insulin receptor was incubated with 10 Ϫ7 M insulin overnight at 4°C. A tyrosine kinase reaction in the presence and absence of recombinant full-length IRS-1 protein (0.5-1.0 g; Upstate Biotechnology) as an exogenous substrate was performed in a buffer containing 30 mM Hepes (pH 7.4), 10 mM MgCl 2 , 8 mM MnCl 2 , 50 M ATP, and 10 Ci of [␥-32 P]ATP at room temperature for 15 min (26). Unincorporated [␥-32 P]ATP was removed by a Bio-Gel P6 spin column (Bio-Rad).
PTPase Assays-32 P-Labeled insulin receptor and 32 P-labeled insulin receptor substrate were incubated with column fractions (obtained as described above) containing endogenous LAR or PTP-␣ in a 125-l reaction volume with 10 mM dithiothreitol, 1 mM EDTA, 1 mg/ml bovine serum albumin in 50 mM Hepes buffer (pH 7.4) at 30°C (23,31). The reaction was terminated by the addition of SDS-PAGE sample buffer (24) containing ␤-mercaptoethanol, and after boiling the samples were separated by SDS-PAGE. When indicated, a recombinant GST-PTP-1B fusion protein (full-length, 0.025 g; Upstate Biotechnology) was used. Dephosphorylation of the 95-kDa ␤-subunit of the insulin receptor and the 180-kDa band of the insulin receptor substrate was analyzed by autoradiography.

RESULTS
Since the localization of the protein-tyrosine phosphatases can be an important factor in determining their substrate specificity, we first studied the subcellular distribution of the PT-Pases PTP-␣, LAR, PTP-1B, and SHPTP2/Syp in freshly isolated rat adipocytes and compared this to that of the IR, IRS-1, and GLUT4. Under basal conditions, PTPases PTP-␣ and LAR ( Fig. 1, middle panels) fractionated in a manner very similar to that of the IR, particularly PTP-␣, whose distribution was primarily in the plasma membrane (62%), followed by heavy (29%) and light microsomes (9%). As expected and as we have reported (25,27,28), GLUT4 was found only in light and heavy microsomes (Fig. 1, bottom panel).
Stimulation of adipocytes with insulin resulted in the redistribution of the transmembrane PTPase PTP-␣ from the plasma membrane to the heavy microsomes (Fig. 1, middle  panels). Also, insulin exposure reproducibly caused a small apparent increase in LAR immunoreactivity in the plasma membrane. LAR cannot be translocated to the plasma membrane because it is not present in any other compartment, and therefore, the altered signal may be a result of a change in affinity for antibody, possible as a result of a covalent modification. Consistent with previous reports (17,18,29,30), insulin induced internalization of its receptor to intracellular membranes as well as translocation of GLUT4 (25,27,28) from the light microsomes to the plasma membrane (Fig. 1, top and  bottom panels).
PTP-1B, SHPTP2/Syp, IRS-1, and IRS-2 all fractionated to the light microsomes and cytosol (Fig. 2). In the presence of insulin, the PTPase PTP-1B showed a small degree of redistri- bution from the light microsomes to the cytosol, and IRS-2 exhibited a small increase in this subcellular fraction (Fig. 2,  middle panel). By contrast, IRS-1 and SHPTP2/Syp distribution did not change in response to insulin treatment (Fig. 2, top  and bottom panels). Although SHPTP2/Syp is abundantly expressed in rat adipocytes (Fig. 2), several lines of evidence (31)(32)(33)(34) tend to rule it out as playing a role in insulin's regulation of metabolic processes, particularly GLUT4 translocation. Also, compared with IRS-1, relatively little IRS-2 was present in rat adipocytes (Fig. 2, middle panel), and for these reasons, SHPTP2/Syp and IRS-2 were not studied further.
In search of support for a enzyme/substrate relation, we investigated a possible colocalization of the PTP-1B with IR or IRS-1 in the intracellular membranes of adipocytes. Light microsomes were fractionated by sucrose velocity gradient centrifugation, and the resultant fractions were analyzed by Western blotting. As shown in Fig. 3 (middle and bottom panels), PTP-1B and IRS-1 migrated at the top of the gradient in vesicles or macromolecular assemblies (35) with the same density. There is a slight shift in IRS-1 distribution after insulin treatment as was previously demonstrated in rat skeletal muscle (23). In contrast, insulin receptors were found to peak toward the middle of the velocity gradient (fractions 7-13) as did GLUT4 (Fig. 3, top and middle panels). As expected, insulin receptors were chased into the light microsomes by insulin, and GLUT4 levels decreased in the gradient fractions from the light microsomes (Fig. 3, top panel), due to a translocation to the plasma membrane.
Considering the data in Figs. 1-3, we can assume that either PTP-␣ or LAR or both PTPases can act on the insulin receptor at the cell surface and possibly in the HM, whereas the PTPase PTP-1B may exert its action on IRS-1. In order to test these possibilities, the transmembrane phosphatases PTP-␣ and LAR were purified from the plasma membrane fraction on a Mono Q column by FPLC with a linear gradient of 50 mM to 1 M NaCl. These PTPases were visualized by immunoblotting with anti-PTP-␣ and anti-LAR antibodies in the column fractions (Fig. 4A), which were assayed for PTPase activity. As shown in Fig. 4B, PTPase activities using autophosphorylated labeled insulin receptor as substrate were only found in the column fractions that contained the transmembrane PTPases.
To determine if endogenous PTP-␣, LAR, and recombinant PTP-1B have a substrate preference toward IR or IRS-1, we performed a dephosphorylation reaction using a mix of phosphate-labeled IR/IRS-1 as substrates. A time course of the reaction was followed between 5 and 60 min. As shown in Fig.  5, A-C, each of the PTPases dephosphorylated IR and IRS-1 in a time-dependent manner. However, the three enzymes tested differed in their activity toward IR or IRS-1 dephosphorylation. At all times examined, IRS-1 was more efficiently dephosphorylated than IR (Fig. 5, B-C), and PTP-1B was particularly effective in dephosphorylating IRS-1.

DISCUSSION
The level of tyrosine phosphorylation of a protein at a specific point in time represents a carefully modulated balance between tyrosine kinase and phosphatase activities. As noted previously, several PTPases have been implicated as exerting a negative regulation on insulin action, mainly in dephosphorylating the insulin receptor (IR), and these are PTP-␣, LAR, and PTP-1B (reviewed in Refs. 12-14). Conclusions along these lines derived from studies performed in cells using recombinant PTPases or overexpression systems in which the PTPase levels are artificially high (e.g. see Refs. 36 -38) for PTP-␣, PTP-1B, and LAR, respectively). Although the level of PTPase mRNA has been determined in rat adipocytes (39), there has heretofore been no study of their distribution or enzymatic activity in these cells. Thus, we studied the expression of PTPases PTP-␣, PTP-1B, and LAR at the protein level, their catalytic activity toward IR and IRS-1, and their subcellular distribution under basal conditions and after insulin stimulation in freshly isolated rat adipocytes. Although it is well known that when expressed in vitro, any of the studied PTPases can dephosphorylate the insulin receptor (40), it is important to show that there is enough activity actually expressed in target cells to achieve this physiologically. The results we obtained demonstrate that all three PTPases are able to efficiently dephosphorylate insulin receptors, indicating that any of these PTPases have the potential to inactivate it in vivo. LAR and PTP-␣ were isolated in parallel and concentrated to the same extent; therefore, their catalytic activity is directly comparable. As when expressed enzymes were studied in vitro (40), we also found that they were similarly effective in dephosphorylating IR, which does not support a preferential role of either PTPase.
IRS-1 is rapidly phosphorylated on tyrosine residues after insulin stimulation followed by its dephosphorylation by an undetermined PTPase (17). We found that in addition to dephosphorylating IR, PTP-␣, LAR, and PTP-1B could also effectively dephosphorylate IRS-1, a result that does not support the existence of a unique PTPase for IR and another for IRS-1. In fact, when a mix of IR/IRS-1 was used as a substrate, all three PTPases were more efficient in dephosphorylating IRS-1 than IR. In particular, PTP1B was very effective in dephosphorylating IRS-1, and these proteins partially overlap in their subcellular distribution, suggesting a possible functional relationship.
There is an abundance of experimental evidence to support LAR as a negative regulator of the insulin signaling pathway. This includes the formation of a physical complex between LAR and IR detected by immunoprecipitation (41), a decrease in the response to insulin in cells overexpressing LAR (38), an enhanced sensitivity to insulin in cells expressing antisense LAR mRNA (42,43). However, LAR knockout mice did not show the anticipated increase or extended IR or IRS-1 phosphorylation after insulin stimulation but instead displayed an unexpected decrease in phosphatidylinositol 3-kinase activity (15). As previously noted in this paper, we and others have shown that internalized IR is highly phosphorylated and displays enhanced kinase activity and that both features subsequently decrease with time, indicating that a PTPase involved in IR dephosphorylation may reside in an endosomal compartment and not exclusively in the plasma membrane. In rat adipocytes ( Fig. 1), LAR is present almost exclusively in plasma membrane, and it does not translocate to endosomal compartments even after insulin stimulation, suggesting that it is unlikely that LAR participates in IR inactivation in the endosomal compartment. Therefore, it may function exclusively at the plasma membrane.
In rat adipocytes, PTP-␣ mRNA was the most abundant of several PTPases examined (39). Here, we show that although PTP-␣ is primarily in the plasma membrane, there is a significant fraction associated with heavy microsomes, an internal membrane compartment. Most importantly, insulin markedly increased the amount of PTP-␣ associated with this fraction. As mentioned above, IR dephosphorylation occurs in an internal membrane compartment, and from Fig. 1 it is clear that there is a significant amount of IR that translocates to the same heavy fraction as PTP-␣. It is tempting to speculate that IR and PTP-␣ are in the same population of vesicles in the heavy microsomes fraction. Preliminary results show that, indeed, after further fractionation of the heavy microsomes by a sucrose velocity gradient, both IR and PTP-␣ are present in the same fractions, although it remains to be shown if both proteins coexist in the same vesicles or if they are present in different vesicles with the same density. A negative role for PTP-␣ in insulin action has been proposed by Moller et al. (36) based on a decrease of IR tyrosine kinase activity after overexpression of PTP-␣ in baby hamster kidney cells. However, very recently, Arnott et al. (44) found that decreasing PTP-␣ levels with an antisense probe had no significant effect on the dephosphorylation profile of IR and IRS-1 proteins in 3T3-L1 adipocytes. Although this result apparently contradicts the overexpression data and does not support a role for PTP-␣ in insulin signaling, it is possible that the enzyme is so efficient that the levels remaining after the antisense treatment (20%) are enough to effectively dephosphorylate IR and IRS-1. Interestingly, the ability of insulin to stimulate glucose transport under these conditions was not reported (44).
PTP-1B was one of the first PTPases to be identified, cloned, and shown to display a significant effect on the modulation of insulin signaling (45). Microinjection of PTP-1B into Xenopus oocytes blocked insulin-stimulated S6 peptide phosphorylation and retarded insulin-induced oocyte maturation (45). Subsequent studies demonstrated that overexpression of PTP-1B affected insulin-or insulin-like growth factor-I-dependent tyrosine phosphorylation and metabolic responses such as glucose incorporation into glycogen (37) and that PTP-1B and IR interacted physically (46). Recently, PTP-1B-deficient (knockout) mice were shown to have an increased phosphorylation of IR and IRS-1 in liver and muscle tissue after insulin injection as well as a resistance to weight gain in mice, suggesting that PTP-1B has a major role in modulating both insulin sensitivity and fuel metabolism (16). PTP-1B, which is localized in the light microsomal fraction of fat cells (Figs. 2 and 3), can also potentially interact with the internalized IR following insulin stimulation. It is interesting to note that PTP-1B and IRS-1 show a very similar subcellular compartmentalization and that PTP-1B had the highest activity toward IRS-1 compared with IR. Considering these results altogether, we believe that this PTPase plays an important role in IRS-1 dephosphorylation.
In summary, our studies lead us to conclude that PTP-␣ and PTP-1B may play a major role in modulating the insulin signaling pathway through the inactivation of insulin receptor kinase and post-insulin receptor cellular substrate proteins such as IRS-1. In view of the complexity and contradictory nature of published work in this regard, it is likely that further work will be necessary to determine more precisely the nature and location of PTPases involved in modulating insulin signaling.