The Dephosphorylation of Insulin and Epidermal Growth Factor Receptors ROLE OF ENDOSOME-ASSOCIATED PHOSPHOTYROSINE PHOSPHATASE(S)*

The autophosphorylation, from [r-”P]ATP, of insulin and epidermal growth factor receptors in rat liver endosomes peaked at 2-5 min and declined thereafter. When autophosphorylation from either [-p3’P]ATP or unlabeled ATP was stopped after 5 min by adding excess EDTA 2 ATP, the phosphotyrosine (PY) content of each receptor decreased at 37 OC with a tl/z of 1.6 min. This was equally so whether the PY content of 32P-labeled receptors was analyzed by autoradiography of KOH-treated gels or by Western blotting with P Y antibodies of immunoprecipitated receptors. The dephosphorylation reaction was strictly dependent on the presence of sulfhydryl, was unaffected by the addition of rat liver cytosol, and was tempera-ture-dependent. The phosphotyrosine phosphatase(s) (PTPase(s)) appeared to be tightly anchored to the endosomal membrane, since the dephosphorylation reaction was unaffected by sodium carbonate and 0.6 M KC1 treatments. However, treatment with Triton X-100 abolished dephosphorylation, implying an intimate association between the PTPase(s) and its substrate in an intact membrane environment. The powerful subjected to autophosphorylation followed by the extraction procedures for removing peripheral membrane proteins. The autophosphorylation reaction was stopped by adding ice-cold Na2C03 to a final concentration of 0.1 M, pH 11.0, or by adding 0.6 M KC1 and 5 mM H202 to a total volume of 4 ml. After 30 min of gentle shaking at 4 "C, 2 volumes of ice-cold distilled water were added and the suspension was centrifuged at 100,000 g., for 60 min. The pellet was resuspended in prewarmed (37 "C) assay buffer minus DTT, and dephosphorylation was initiated by adding DTT to a final concentration of 1 mM. C, dephosphorylation was initiated by adding ATP/EDTA stopping solution containing Triton X-100 in order to obtain a final concentration of 0.0, 0.01, 0.1, and 1.0% (v/v) during the incubation.

Following their intravenous injection insulin, EGF' and other ligands bind to their respective receptors in target cells and are rapidly translocated as ligand-receptor complexes into the endosomal apparatus (1). Subsequent studies have documented a corresponding dose-dependent activation of the insulin and EGF receptor tyrosine kinases and an accumulation of these activated kinases within rat liver endosomes (ENS) (2-5). Of interest were the observations that both the insulin receptor and EGF receptor kinases displayed greater autophosphorylating activity in ENS than seen in plasma membrane (PM) (3, 5). In more detailed recent studies of the insulin receptor, it was found that the endosomal receptor contained less phosphotyrosine (PY) per receptor than that in PM even though the former was a more active kinase (6). We inferred that an endosomal phosphotyrosine phosphatase (PTPase) may play a role to effect the partial dephosphorylation observed. In other studies, we measured the PTPase activity against artificial substrates in solubilized rat liver PM and EN cell fractions (7). These studies indicated equal PTPase activity in PM and ENS in both control rats and those receiving a pharmacologic dose (150 pg/100 g, body weight) of insulin. Solubilization would be expected to disrupt the spatial relationship between a receptor and receptor-associated PTPase(s). Furthermore, receptor-associated PTPase(s) might show considerably different activity toward artificial substrates than toward the receptor as substrate. Given these considerations, we set out to develop a method for assaying receptor-associated PTPase in intact EN membranes in order to evaluate its functional characteristics in situ.

MATERIALS AND METHODS
Animals-Female Sprague-Dawley rats (140-160 g, body weight) were purchased from Charles River Ltd. (St. Constant, Quebec) and were fasted overnight prior to killing. Rats were anesthetized with ether, and hormones were injected via the portal vein. They were killed by decapitation at 2 min postinjection of insulin and at 15 min postinjection of EGF, corresponding to the respective peak times of receptor kinase stimulation (2,4,5). Livers were rapidly excised, placed in ice-cold homogenizing buffer (0.25 M sucrose, 25 mM KCl, 5 mM Hepes, 1 mM phenylmethylsulfonyl fluoride, 1 mM benzamidine, pH 7.4) and minced before homogenization.
Collaborative Research (Waltham, MA). Anti-insulin receptor antibodies were purified from a patient's serum as previously described (3). The EGF receptor was immunoprecipitated with monoclonal antibody IgG 151, BH6 (5). Antiphosphotyrosine antibody (a-PY) was produced in rabbits using phosphotyrosine coupled to keyhole limpet hemocyanin as antigen as described in detail elsewhere (6). The antibody was subsequently purified by affinity chromatography on a phosphotyrosine Affi-Gel column (8). Antibodies against a synthetic peptide (ENLTTQETREILHFHYTT) corresponding to residues 161-178 in the human placental PTPase (PTP-lB), which are conserved in rat brain PTPase, were prepared, as previously described (91, and kindly provided by Dr. D. L. Brautigan (Brown University). Antibody to the T-cell PTPase (PTP-TC) was generated by immunizing rabbits with the protein expressed in Escherichia coli and was provided courtesy of the laboratory of Dr. S.-H. Shen (Biotechnical Research Institute, Montreal, Canada).
The BstuI-Hind111 fragment of the cDNA encoding the human PTP-TC (10) was inserted between the Smd and Hind111 sites of the expression vector pWR-590 generating the plasmid pWR-PTP-TC. This construct encoded a fusion protein of the tyrosine phosphatase with 590 amino acids of E. coli 8-galactosidase. To purify the fusion protein DH5, transformed with pWR-PTP-TC were induced with IPTG (I mM) for 18 h and lysed in Laemmli buffer containing 5 M urea. After boiling, the sample was subjected to SDS-PAGE using 7.5% gels (11). The fusion protein was excised from the gel, and the polyacrylamide gel slice was homogenized and lyophilized. The dried material was emulsified with an equal volume of complete Freund's adjuvant for the first injection and with incomplete Freund's adjuvant for subsequent injections. Each rabbit received subcutaneously five injections (50 pg/injection) at 3-week intervals. Animals were bled 2 weeks after the last injection. IgG was purchased from Sigma and radioiodinated to a specific activity of 6 X 1 0 ' dpm/pg of protein using the chloramine-T procedure (12). Protein A-Sepharose and wheat germ agglutinin-Sepharose 6 MB were from Pharmacia Fine Chemicals (Dorval, Quebec). [-p3'P]ATP (1000-3000 Ci/mmol) was from New England Nuclear Radiochemicals (Lachine, Quebec). Reagents for SDS-PAGE were obtained from Bio-Rad. Sodium orthovanadate, phenylarsine oxide, sodium molybdate, and sodium tungstate were obtained from Sigma. All other chemicals were of analytical grade and purchased from Fisher or Boeringher Mannheim (Montreal, Quebec). Vanadate solutions were prepared as previously described in order to avoid changes in pH leading to the generation of decavanadate (orange) or vanadyl ions (V02+, blue) (13). Peroxide(s) of vanadate (pV) was prepared by mixing vanadate with 1 mM hydrogen peroxide (HzOZ). The concentration of pV generated is denoted by the vanadate concentration added to the mixture (14). Subcellular Fractions ana' Protein Determination-Plasma membrane and endosomes were isolated as previously described (3). These subcellular fractions have been characterized in respect to marker enzyme cytochemistry, electron microscopy, ligand uptake, and ligand-induced receptor kinase stimulation (2-5, 15, 16). Protein content in the fractions was determined by a modification of Bradford's method using serum albumin as a standard (17).
Receptor Autophosphorylation-ENS (25 pg) were suspended in 90 p1 of solution, which at final concentration was 50 mM Hepes, pH 7.4, 150 mM KC1, 5 mM NaCl, 5 mM MgC12, and 1 mM DTT. This solution maintains endosomal structural integrity for 30 min at 37 "C as previously established (18). Phosphorylation was initiated by adding 10 pl of either unlabeled ATP to a final concentration of 1 mM or [y3'P]ATP (32 pCi/nmol) to a final concentration of 25 pM, and the reaction was carried out at 37 "C, except where noted elsewhere in the text.
Dephosphorylation Assays-Autophosphorylation was stopped, at times indicated in the text, by adding 10 pl of prewarmed (37 "C) concentrated stopping buffer containing 500 mM Hepes, pH 7.4, and 100 mM EDTA. In assays using [y-32P]ATP) the autophosphorylation stopping buffer routinely contained 5 mM unlabeled ATP.
To observe dephosphorylation of 32P-labeled receptors, the incubation at 37 "C was continued and stopped at subsequent times by adding 50 pl of ice-cold solubilizing buffer containing 4.5% Triton X-100 in 150 mM Hepes, pH 7.4, 150 mM EDTA, 6 mM sodium orthovanadate, 3 mM phenylmethylsulfonyl fluoride and 3 mM benzamidine. Samples were vigorously shaken for 1 h at 4 'C, and solubilized membranes were diluted 3-fold by the addition of 290 p1 of ice-cold 50 mM Hepes and 10 pl of either insulin or EGF receptor antibody (50-70 pg of protein). After 18 h at 4 "C, 50 p1 (2 mg) of ice-cold protein A-Sepharose was added for a further 30 min, followed by centrifugation at 12,000 X gmaX for 3 min (Beckman Microfuge). The immunoprecipitate was washed once with 50 mM Hepes (pH 7.4) containing 1.0% Triton X-100 and 0.1% SDS, followed by two final washings with the same buffer containing 0.1% Triton X-100 without SDS. The immunoprecipitates were solubilized and subjected to SDS-PAGE (4% stacking and 7.5% resolving gel) under reducing conditions as described before (3,5). 32P-Labeled receptors (insulin, 94 kDa; EGF, 170 kDA) were visualized on the gels by radioautography at -70 "C using enhancing screens and Kodak X-Omat AR films. To remove alkali labile serine and threonine, phosphate gels were fixed with 10% glutaraldehyde and then treated with 1 M KOH for 2 h at 55 "C before radioautography (19).
To observe dephosphorylation of unlabeled receptors, PY receptor content was assessed by Western blotting with PY antibodies. Dephosphorylation was stopped by adding 50 p1 of 150 mM Tris, pH 6.8, 6.9% SDS, 30% glycerol, and 100 mM DTT and heating at 100 "C for 2 min. After electrophoresis, separated proteins were transferred to nitrocellulose membranes (Bio-Rad) as described elsewhere (20). Membranes were incubated with PBS containing 20% calf serum (5 ml/lane) at room temperature for 1 h, followed by PY antibody (1 pg of protein/ml) in PBS, 1% bovine serum albumin at 4 "C for 18 h. The membranes were then washed for 15 min (three times) in PBS, 0.05% (v/v) Tween 20, followed by incubation with T -g o a t IgG ( 5 X lo6 cm/lane) in PBS, 1% bovine serum albumin at room temperature for 1 h. After washing as before in PBS, 0.5% Tween 20, the membranes were air-dried and submitted to radioautography as described above.
f'P]ATP Determinations-To assess the fate of [32P]ATP during autophosphorylation/dephosphorylation, samples (1 pl) of reaction mixture were applied to polyethyleneimine cellulose thin layer chromatography plates prespotted with 0.5 pl of nucleotide solution containing 5 mM each of ATP, ADP, and AMP in water. Chromatography was carried out using a solution of 0.7 M LiCl in 1 M acetic acid as previously described (21). Spots were located under UV light, excised, and added to 15 ml of scintillation vials containing 5 ml of H202 and counted (Cerenkov) in a @-scintillation counter (LKB 1219 Rack 8).

Time Course of Receptor Autophosphorylation in Isolated
Endosomes-Following the in uiuo injection of insulin (1.5 pg/lOO g, body weight), endosomal receptor autophosphorylation and exogenous kinase activities rise to a maximum at 2-5 min postinjection and decline thereafter (3). This observation of receptor kinase deactivation implies dephosphorylation of the receptor whose activated state is autophosphorylation-dependent (3,22). We sought to assess the process of dephosphorylation in ENS directly. In our initial studies, we examined the time course of receptor autophosphorylation in uitro. Therefore, we prepared endosomes at 2 min after insulin (1.5 pg/lOO g, body weight) and 15 min after EGF (1 pg/100 g, body weight) injections, corresponding respectively to the times of peak kinase activation (3)(4)(5). The intact ENS were incubated at 37 "C in a buffer mimicking the intracellular milieu (18) and in the presence of [y-32P]ATP. Fig. 1 depicts the time course of alkali-resistant 32P incorporation from [y-32P]ATP into insulin (panel A ) and EGF (panel B ) receptors. In each case, 32P incorporation reached a maximum (insulin, 5 min; EGF, 2 min) and subsequently declined to 50% of the maximal value by 15 min of incubation. In the presence of 100 p~ pV, the extent of 32P incorporation was augmented to a maximum of 2-fold above that of control incubations, and the decrease in insulin receptor labeling was completely prevented. The decline in ATP concentration by 40% after 10 min and 90% after 30 min of incubation (Table I) was the same in the absence or presence of pV. This demonstrated that pV did not act through the inhibition of an endosomal ATPase and was consistent with the inhibition by pV of endosomal PY phosphatase(s), as indicated in a previous study (14). These data suggested that under the conditions employed for studying autophosphorylation, a dynamic process involving receptor phosphorylation/dephosphorylation was occurring.

TABLE I ATP degradation during autophosphorylatwn
Autophosphorylation was carried out on ENS prepared from rats injected with insulin (1.5 pg/lOO g, body weight) as described under "Materials and Methods" in the absence and presence of 100 p~ pV. At the noted times, the reaction was terminated and, after removal of ENS by centrifugation a volume of supernatant (1 pl), was subjected t o thin layer chromatography as described under "Materials and Methods." After development, the spot corresponding to ATP was located under UV light, excised, and counted (Cerenkov). The results reflect the mean f S.E. of three separate experiments.

Incubation
[32P]ATP time Time Course of Loss of P Y from Endosoma1 Insulin and EGF Receptors-To study the dephosphorylation step, the autophosphorylation reactions for insulin and EGF receptors were inhibited at their respective times of maximal phosphorylation (insulin, 5 min; EGF, 2 min) by adding excess EDTA, which chelates Mg2+ and abrogates phosphorylation.  (Fig. 2, upperpanels), followed by densitometric analyses to determine relative signal intensity (Fig. 2, lower panels). In each case, there was a rapid loss of receptor PY content at a t1I2 of 1.6-1.7 min. We isolated at 2 min (insulin, 1.5 pg/100 g, body weight) or 15 min (EGF, 1 pg/lOO g, body weight) following ligand injection were autophosphorylated with 1 mM ATP before adding EDTA (zero time) to terminate phosphorylation and initiate the dephosphorylation assay as described under "Materials and Methods." A t the noted times following EDTA addition, ENS were solubilized, the receptors were immunoprecipitated, and the immunoprecipitates were subjected to SDS-PAGE and immunoblotting with antiphosphotyrosine antibodies to determine their relative PY content as described in the text. thus conclude that a rapid receptor phosphotyrosine dephosphorylation reaction was occurring in ENS.
Time Course of Receptor Dephosphorylation-To characterize this process, we designed an alternative assay that was more rapid and convenient. Using [y3'P]ATP endosomal receptor, labeling was inhibited at maximal phosphorylation times with EDTA and a 20-fold excess of unlabeled ATP, following which the incubation was continued at 37 "C. Fig. 3 depicts the level of alkali-resistant 32P in endosomal insulin (panel A ) and EGF (panel B ) receptors as a function of time following the inhibition of phosphorylation. In each case, there was a rapid loss of receptor 32P content at a tl/2 of 1.6-1.7 min, which is entirely comparable with the results observed in the a-PY immunoblotting studies. The omission of DTT from the incubation resulted in the complete inhibition of dephosphorylation. Though 100 ~L M pV completely inhibited dephosphorylation, a comparable concentration of vanadate had no such effect. Finally, the addition of hepatic cytosol to ENS did not influence the rate of dephosphorylation.
Characterization of Endosomul PTPase Activity Toward the Insulin Receptor-As expected for an enzyme-mediated process, the endosomal dephosphorylation activity was temperature-dependent (Fig. 4A). Dephosphorylation was rapid at 37 "C, with a tl/2 of 1.5-2.0 min, whereas at 25 "C, the t1/2 was 8-10 min. After 10 min at 4 "C, the reaction was barely perceptible.
The failure of cytosol to accelerate the rate of insulin receptor dephosphorylation (Fig. 3) suggested that the observed endosomal PTPase activity was not due to associated cytosolic enzyme(s). This was confirmed by a series of experiments in which we attempted to "strip" away the PTPase activity by procedures that remove peripheral proteins. Thus, pretreatment of ENS with 0.6 M KC1 or with 0.1 M Na2C03 (pH 11.0) (23) did not result in the removal of any endosomal PTPase activity (Fig. 4B). We conclude that endosomal PTPase(s) are either intrinsic membrane proteins or cytosolic-like enzymes tightly coupled to the membrane.
Most of the assays designed to measure PTPase activity  Fig. 3. A, temperature dependence was evaluated by performing the dephosphorylation assay at 37, 25, or 4 "C. B, ENS were subjected to autophosphorylation followed by the extraction procedures for removing peripheral membrane proteins. The autophosphorylation reaction was stopped by adding ice-cold Na2C03 to a final concentration of 0.1 M, pH 11.0, or by adding 0.6 M KC1 and 5 mM H202 to a total volume of 4 ml. After 30 min of gentle shaking at 4 "C, 2 volumes of ice-cold distilled water were added and the suspension was centrifuged a t 100,000 g. , for 60 min. The pellet was resuspended in prewarmed (37 "C) assay buffer minus DTT, and dephosphorylation was initiated by adding DTT to a final concentration of 1 mM. C, dephosphorylation was initiated by adding ATP/EDTA stopping solution containing Triton X-100 in order to obtain a final concentration of 0.0, 0.01, 0.1, and 1.0% (v/v) during the incubation.
have used enzymes and substrates in solution. Recent information on PTPases homologous to the 50-kDa placental enzyme (PTPase 1B) has suggested that the carboxyl-terminal region of these enzymes is important for the regulation of catalytic activity and for subcellular localization (10,(23)(24)(25)(26). Thus, PTPases measured in solution appear to originate from proteolytically truncated enzymes that have become constitutively activated (10,26,27). In Fig. 4C, the influence of increasing concentrations of Triton X-100 on PTPase activity toward the insulin receptor is depicted. At low concentrations (0.01 and 0.1%), Triton X-100 inhibited the dephosphorylation activity, and at 1.0% Triton X-100, a concentration resulting in membrane solubilization, PTPase activity was essentially abolished. In numerous reports, PTPase activity has been measured in the presence of Triton X-100 (7,27,28); it is thus unlikely that Triton X-100 was acting as a PTPase inhibitor. It is possible that detergent treatment removes lipids needed to assure a conformation in which the receptor is a suitable substrate for the phosphatase(s). Our data do suggest that the endosomal PTPase must remain in intimate contact with its substrate in an intact membrane environment in order to function enzymatically.
Inhibition of Receptor PTPase-The potency of a number of reagents previously recognized as PTPase inhibitors was evaluated (Fig. 5). The inhibitors fall into roughly three classes. First is a group of substances potent in the 0.1-10 mM range with 50% inhibition occurring around 1 mM (Fig.  5, shaded area). This group includes the transition metal oxanions vanadate, molybdate, and tungstate, among which vanadate and molybdate have been well recognized as PTPase inhibitors (29). Zinc is a classical PTPase inhibitor (27), and HzOz has been noted to be insulin-mimetic (14,30). Phenylarsine oxide has been recognized as a PTPase inhibitor (31, 32) and is potent in the 10 p~ to 1 mM range, with 50% inhibition occurring near 100 p~. The third class of inhibitor is peroxovanadium (pV), which has been identified as a powerful insulin mimetic agent (14). It was found to be inhibitory in the 0.1-10 UM range, with 50% inhibition occurring at 1

PM.
It is of interest to note that okadaic acid (1-10 p~) , a powerful inhibitor of phosphoserine phosphatases (ppl and pp2A) (33) was without effect on the receptor PTPase(s). Also, the peptide poly(G1u:Tyr) was ineffective as an inhibitor (10-100 pg/ml; data not shown). No significant differences in potency were observed when the inhibitory efficacy of these reagents were tested on EGF receptor dephosphorylation (Fig.  5B).
The Influence of Ligand Dose on Receptor Dephosphoryhtion-In order to assess a potential regulatory role for ligand dose on receptor dephosphorylation, the effect of different doses of insulin or EGF was evaluated a t dephosphorylation reaction times of 2, 5, and 10 min (Table 11)   Compartmentalization of PTPase IB and the T-cell PTPase-Attempts were made to assay receptor PTPase activities in PM cell fractions in order to compare these activities with those in ENS. As noted previously (3) the extent of in vitro autophosphorylation of either insulin or EGF PM receptors was substantially less than that seen for EN receptors. The reason for this difference remains unclear, but studies of PM in a manner comparable with EN receptors were not possible. As part of the attempt to compare PTPase activity in PM and EN cell fractions, we examined these cell fractions for their content of PTP-1B-and PTP-TC-like enzymes using antibodies raised against these enzymes, as described under "Materials and Methods." With each antibody, Western blots of PM fractions showed a clear signal of the appropriate molecular size, whereas no signal was seen in ENS (Fig. 6). The absence of an endosomal signal on Western blotting persisted in endosomes prepared from rats injected with 1.5 pg of insulin/100 g, body weight. These observations indicate compartmentalization of these enzymes in rat liver.

DISCUSSION
On examining the time course of insulin receptor autophosphorylation in intact ENS, it was evident that, at 37 "C, peak labeling occurred at 5 min (Fig. LA), only to be followed by a loss of label thereafter. At the point of maximal autophosphorylation, the intensity of the signal was twice as high in the presence of pV, indicating that maximal autophosphorylation was not attained in controls. This limitation did not reflect higher [32P]ATP levels in pV-treated samples, since the rates of ATP degradation were the same in the absence or presence of pV. The results were consistent with the occurrence of an endosomal receptor dephosphorylation reaction and a capacity of pV to inhibit phosphotyrosine phosphatase (PTPase) activity as previously observed (14). The progressive decline in receptor autophosphorylation observed during more prolonged incubation in the absence of pV was due to the progressive loss of ATP from the reaction mixture, with the consequence that the dephosphorylation reaction was progressively favored. Thus, under control conditions, the level of receptor autophosphorylation attained in intact ENS reflects the balance existing between phosphorylation and dephosphorylation reactions. The dephosphorylation reaction could be accurately measured by stopping kinase activity. Under such conditions, the dephosphorylation of both insulin and EGF receptors was rapid and occurred a t a comparable rate, with a tIlz of 1.6-1.7 min. Similar results were observed when receptors were autophosphorylated with unlabeled ATP or [y3'P]ATP. In the former case, kinase activity was stopped by chelating divalent cations with EDTA and following receptor phosphotyrosine content by immunoblotting with antiphosphotyrosine antibodies. In the latter case, the rate of loss of 32P was the same when EDTA alone was added or when excess unlabeled ATP was added to dilute out [32P]ATP as a substrate. Taken together, these data indicate that the loss of 32P receptor content was not the result of an exchange reaction (34).
Several lines of evidence indicate that the dephosphorylation reaction reflected true PTPase activity. Thus, the presence of DTT in the reaction mixture was essential and fully in accord with previous studies documenting the strict dependence of PTPase activity on the presence of reducing agents (28,29). In fact, recent mutational analyses support the idea that a highly conserved cysteine residue plays an essential role in the catalytic activity (35). Furthermore, it has recently been shown that a cysteine-phosphate intermediate forms during PTPase catalysis (36). In addition, dephosphorylation was inhibited by the classic PTPase inhibitors zinc (28,29), vanadate, and molybdate (28,29,37). The fact that the observed PTPase activity resulted from the action of an intrinsic endosomal enzyme(s) was suggested by the observation that the rate of insulin receptor dephosphorylation was not accelerated by adding fresh hepatic cytosol. This was confirmed by drastic pretreatments well known to strip away intraendosomal proteins and loosely attached pe-ripheral membrane proteins; none of these procedures significantly affected dephosphorylation of the insulin receptor. We conclude that the observed dephosphorylating activity results from the action of endosomal PTPase(s) firmly anchored to the membrane.
Based on their cDNA sequences, recently cloned PTPases can be divided into two subfamilies. One subfamily consists of transmembrane PTPases (38-42), whereas the second subfamily consists of cytosolic enzymes containing noncatalytic domains that appear to specify cellular distribution. Thus one recently cloned enzyme contains in its amino-terminal region SH2 domains that impart a capacity to bind to phosphotyrosine-containing proteins (43). Several others contain hydrophobic carboxyl termini and/or sites for lipid attachment (10, 24-26) that appear to specify a strong membrane attachment. In recent studies, both intact and carboxyl-terminal truncated PTP-TC were expressed in baby hamster cells and a baculovirus system (10,26). A substantial fraction of the truncated enzyme was found in the cytosol, whereas the intact enzyme was entirely membrane-associated. From these considerations and our observations, we cannot ascertain whether the endosome-associated PTPase(s) is a transmembrane protein or a cytosolic-like enzyme with strong affinity for the lipid environment of membranes. We would interpret the effect of Triton X-100 to abolish insulin receptor dephosphorylation as indicating that within the environment of the intact membrane, intimate contact of the PTPase with its natural substrate occurred and was required for dephosphorylation to proceed. Our data do not permit us to distinguish between a model in which the phosphatase(s) is constitutively associated with the receptor and a model in which the phosphatase(s) is a membrane-diffusible protein whose association with the receptor is a regulated phenomenon.
Experiments with PTPase inhibitors were designed to meet three objectives: 1) to assess the potency of traditionally used inhibitors in this in situ assay, 2) to assess the potency of known insulin mimetic agents such as pV (14) and HzOz (30), and 3) to assess their relative potency toward two different receptors. The results showed that among all the inhibitors tested, pV, the most powerful insulin mimetic agent (14), was 100-1000-fold more potent than the other inhibitors. It is tempting to relate the PTPase inhibitory effect of pV to its capacity to effect insulin receptor kinase activation and to mimic insulin (14). Of interest was the very comparable pattern of inhibition of the EGF receptor-associated PTPase. Here, too, pV was far more potent than the other inhibitors. This may reflect the relatively nonspecific character of the inhibitors studied but may also indicate that similar PTPases are associated with the insulin and EGF receptors.
The effect of an increasing dose of injected ligand on subsequently measured receptor dephosphorylation was shown to differ for the insulin and EGF receptors. Thus, insulin receptor dephosphorylation increased, whereas EGF receptor dephosphorylation decreased with increasing respective ligand doses. An effect of insulin dose on receptor dephosphorylation was not observed in a previous study where we assayed solubilized phosphotyrosine phosphatase activity with an artificial substrate rather than receptor dephosphorylation in this in situ assay. The current data on insulin receptor dephosphorylation are compatible with a previous observation of decreased insulin receptor dephosphorylation in diabetic rats (7). The different ligand-dependent regulation of the insulin and EGF receptors may indicate that their respective receptor-associated PTPases are different. In this regard, it seems appropriate to note that the immunoblotting studies with both PTP-1B and PTP-TC antibodies failed to identify such entities in ENS, whereas they identified crossreactive species of appropriate molecular weight in PM fractions. In addition to implying that there is cellular compartmentalization of PTPases in hepatocytes, these observations tend to exclude the above as candidates for the receptorassociated PTPases characterized in the present work.
When insulin dissociates from its receptor, the metabolic responses entrained by the hormone decay rapidly (22). In previous studies, in FA0 cells, pulse-chase experiments with 32Pi labeling showed that label was lost from the receptor with a half-life of 20 min (22). This reflects a number of events in the intact cell, including the time taken to clear receptorassociated insulin, continuing autophosphorylation from an intracellular [32P]ATP pool, and dephosphorylation events. In digitonin-permeabilized adipocytes, dilution of the prelabeled intracellular ATP pool permitted a more precise measure of the dephosphorylation rate that was found to be about 2 min (44), a rate similar to that observed in the present study.
We have recently observed that following insulin administration, PM insulin receptors contained more phosphotyrosine per 8-subunit than did those in ENS (6); nevertheless, endosomal receptor kinase activity was augmented. It is thus possible that the consequence of a limited initial dephosphorylation is activation of the kinase, followed by later deactivation as dephosphorylation becomes more extensive and complete. It is noteworthy that in this latter study, the level of endosomal insulin receptor tyrosine phosphorylation was inversely related to the dose of injected insulin. In contrast, the level of endosomal EGF receptor tyrosine phosphorylation increased with increasing doses of injected ligand.' Thus, the behavior of the endosomal receptor phosphorylation state in vivo paralleled the activities of the endosomal PTPase(s) regulating receptor dephosphorylation. The potentially complex regulatory role of the receptor-associated PTPase(s), as well as the possibility that they constitute the important target(s) for the pharmacologically potent peroxovanadium species makes their identification and full characterization important.