ATP-dependent Desensitization of Insulin Binding and Tyrosine Kinase Activity of the Insulin Receptor Kinase

Incubating endosomes with ATP decreased binding of 125I-insulin but not 125I-labeled human growth hormone. Increasing ATP concentrations from 0.1 to 1 mm increased β-subunit tyrosine phosphorylation and insulin receptor kinase (IRK) activity assayed after partial purification. At higher (5 mm) ATP concentrations β-subunit tyrosine phosphorylation and IRK activity were markedly decreased. This was not observed with nonhydrolyzable analogs of ATP, nor with plasma membrane IRK, nor with endosomal epidermal growth factor receptor kinase autophosphorylation. The inhibition of endosomal IRK tyrosine phosphorylation and activity was completely reversed by bafilomycin A1, indicating a role for endosomal proton pump(s). The inhibition of IRK was not due to serine/threonine phosphorylation nor was it influenced by the inhibition of phosphotyrosyl phosphatase using bisperoxo(1,10-phenanthroline)oxovanadate anion. Prior phosphorylation of the β-subunit with 1 mm ATP did not prevent the inhibition of IRK activity on incubating with 5 mm ATP. To evaluate conformational change we incubated endosomes with dithiothreitol (DTT) followed by SDS-polyacrylamide gel electrophoresis under nonreducing conditions. Without DTT the predominant species of IRK observed was α2β2. With DTT the αβ dimer predominated but on co-incubation with 5 mm ATP the α2β2 form predominated. Thus, ATP-dependent endosomal acidification contributes to the termination of transmembrane signaling by, among other processes, effecting a deactivating conformational change of the IRK.

I-insulin but not 125 I-labeled human growth hormone. Increasing ATP concentrations from 0.1 to 1 mM increased ␤-subunit tyrosine phosphorylation and insulin receptor kinase (IRK) activity assayed after partial purification. At higher (5 mM) ATP concentrations ␤-subunit tyrosine phosphorylation and IRK activity were markedly decreased. This was not observed with nonhydrolyzable analogs of ATP, nor with plasma membrane IRK, nor with endosomal epidermal growth factor receptor kinase autophosphorylation. The inhibition of endosomal IRK tyrosine phosphorylation and activity was completely reversed by bafilomycin A 1 , indicating a role for endosomal proton pump(s). The inhibition of IRK was not due to serine/threonine phosphorylation nor was it influenced by the inhibition of phosphotyrosyl phosphatase using bisperoxo(1,10-phenanthroline)oxovanadate anion. Prior phosphorylation of the ␤-subunit with 1 mM ATP did not prevent the inhibition of IRK activity on incubating with 5 mM ATP. To evaluate conformational change we incubated endosomes with dithiothreitol (DTT) followed by SDS-polyacrylamide gel electrophoresis under nonreducing conditions. Without DTT the predominant species of IRK observed was ␣ 2 ␤ 2. With DTT the ␣␤ dimer predominated but on co-incubation with 5 mM ATP the ␣ 2 ␤ 2 form predominated. Thus, ATP-dependent endosomal acidification contributes to the termination of transmembrane signaling by, among other processes, effecting a deactivating conformational change of the IRK.
Following insulin binding to its receptor in intact cells the insulin receptor kinase (IRK) 1 undergoes tyrosine autophos-phorylation and kinase activation (1,2). Equally rapidly there is internalization of activated insulin-IRK complexes into ENs (3,4). IRK signaling appears largely to involve tyrosine phosphorylation of adaptor proteins, including insulin receptor substrate-1 (5), insulin receptor substrate-2 (6,7), and SHC (8), which function as docking entities to entrain the insulin signaling sequence. The observation that the activated IRK is internalized to ENs is consistent with the occurrence of transmembrane signaling intracellularly (9,10). Indeed studies in liver parenchyma have shown that the accumulation of activated IRKs exclusively in ENs is sufficient to promote insulin receptor substrate-1 tyrosine phosphorylation (11). In adipocytes it has also been shown that internal membranes are the principal sites where insulin receptor substrate-1 phosphorylation and phophatidylinositol 3-kinase activation occur (12).
Given the above, understanding the mechanisms controlling receptor function in ENs is important for understanding the regulation of insulin signaling. Studies have shown that the level of IRK tyrosine phosphorylation and hence activity is altered and ultimately reduced by an IRK-associated phosphotyrosine phosphatase in ENs (4,13). Discovery of the endosomal acidic insulinase (14) has demonstrated a mechanism by which intraendosomal insulin concentration may be reduced, hence decreasing the proportion of IRKs occupied by ligand (15). In a cell-free system it was demonstrated that ATP-dependent endosomal acidification promoted both insulin dissociation from the IRK and subsequent degradation of free insulin by the endosomal acidic insulinase (15)(16)(17). In the present study we report that ATP-dependent endosomal acidification also leads to decreased IRK binding capacity and to marked deactivation of IRK activity toward exogenous substrates. Our data indicate that this latter phenomenon derives from an acidification dependent conformational change in the intraluminal aspect of the endosomal IRK with attendant deactivation of the cytosolic tyrosine kinase. These data thus identify another process leading to attenuation of the activated state of the IRK, and hence transmembrane signaling during the course of endocytosis. AMP-PNP, adenosine 5Ј-(␤,␥-imino)triphosphate; AMP-PCP, adenosine 5Ј-(␤,␥-methylene)/triphosphate AMP-PSP, adenosine 5'-(3-thiophosphate); GAR, goat anti-rabbit antibody. from the NIH pituitary hormone and antisera program (Baltimore, MD). Carrier-free [ 125 I]iodine and [␥-32 P]ATP (1000 -3000 Ci/mmol) were purchased from NEN Life Science Products. NaCl, MgSO 4 , trichloroacetic acid, and glycerol were from Anachaemia Ltd. (Lachine, Quebec). Wheat germ agglutinin (WGA)-Sepharose 6-MB and protein A-Sepharose were from Amersham Pharmacia Biotech (Dorval, Quebec). Nucleosides tri-, di-, and monophosphates, AMP-PCP, AMP-PNP, and AMP-PSP were from Boehringer Mannheim. Chemicals for SDS-PAGE were from Bio-Rad. Kodak X-Omat AR films were purchased from Picker International Canada (Montreal, Quebec). Immobilon was from Millipore Canada Ltd. (Mississauga, ON). Bafilomycin A 1 , poly-(Glu, Tyr) (4:1), N-acetyl-D-glucosamine, and other chemicals were from Sigma. The peroxovanadium compound bpV(phen) was synthesized and purified as previously reported (18).

Animals-Female
Antibodies-Antibody to the juxtamembrane domain (residues 942-968) of the insulin receptor (␣960) and to phosphotyrosine were prepared and purified as described previously (4). Affinity-purified goat anti-rabbit antibodies (whole molecule) were purchased from Sigma and iodinated to a specific activity of 6 ϫ 10 8 dpm/g of IgG using a chloramine T procedure (19).
Subcellular Fractions, Binding, and Protein Determination-Rats were anaesthetized with ether and were injected via jugular vein with a dose (per 100 g of body weight) of insulin (1.5 g), or bpV(phen) (0.6 mol), dissolved in 0.2 ml of phosphate-buffered saline (pH 7.4), 0.1% bovine serum albumin. Animals were killed by decapitation at 2 and 15 min postinjection of insulin and bpV(phen), respectively (13). Livers were rapidly excised, placed in ice-cold homogenizing buffer (50 mM HEPES (pH 7.4), 0.25 m sucrose, 1 mM phenylmethylsulfonyl fluoride, 1 mM MgCl 2 , 1 mM benzamidine) and minced before homogenization. Combined ENs and PM fractions were prepared as described previously (3) except that the buffer used throughout was that in which the livers were minced (see above). These subcellular fractions have been characterized in detail both morphologically and biochemically (3, 20 -23). Hormone binding was assayed with 125 I-insulin or 125 I-labeled hGH prepared to a specific activity of 100 -200 Ci/g using the chloramine-T method as described previously (19). Protein content in the fractions was determined by a modification of Bradford's method using serum albumin as a standard (24).
Insulin Receptor Phosphotyrosine Content-The phosphotyrosine content of endosomal insulin receptors was determined by subjecting WGA-purified preparations to SDS-PAGE and immunoblotting with ␣960 and ␣PY as described previously (4).
Insulin Receptor Kinase Assays-Insulin receptors from subcellular fractions were partially purified by chromatography on WGA-Sepharose 6MB columns, and receptor content and tyrosine kinase activity were measured as described previously (3,4,20). It has been previously shown that, after insulin administration, tyrosine kinase activity of WGA-purified endosomal preparations was at least 90% attributable to the IRK (4).
Insulin Receptor Autophosphorylation and Phosphoamino Acid Analyses-ENs were removed from the 0.6/1.0 m sucrose interface of the gradient used in endosomal purification (20) and diluted with 0.25 M sucrose to a final protein concentration of 50 g/ml. A 30-ml aliquot was centrifuged at 200,000 ϫ g av for 40 min prior to resuspending in 500 l of cell-free system buffer (44 mM HEPES (pH 7.4), 0.55 M sucrose, 333 mM KCl, 11 mM NaCl, 11 mM MgCl 2 ). Incubations were initiated by adding 500 l of 10 mM Tricine buffer containing either 2 or 10 mM ATP at a specific activity of 0.5 mCi of 32 P/mmol. After incubating for 15 min at 37°C the reaction was stopped by adding an equal volume of ice-cold 20 mM HEPES, 0.25 M sucrose, 2 mM phenylmethylsulfonyl fluoride, 2 mM benzamidine, 4 mM sodium orthovanadate, and 80 mM sodium fluoride. ENs were solubilized by incubating for an additional 30 min at 4°C in a final concentration of 1% Triton X-100, 20 mM pepstatin, 20 mM leupeptin, and 10 mg/ml aprotinin, after which IRKs were immunoprecipitated with ␣960, washed in buffer, subjected to SDS-PAGE, transferred to Immobilon-P membranes, and subjected to autoradiography and/or amino acid analyses as described previously (25).
Insulin Receptor under Nonreducing Conditions-Endosomal pellets were resuspended (500 g of protein/ml) in a final concentration of 50 mM Tris (pH 6.9), 10% glycerol, and 20 mM N-ethylmaleimide. Samples were incubated at room temperature for 5 min and then solubilized without heating in 2.3% SDS, 0.05% bromphenol blue before subjecting to SDS-PAGE in the absence of reductants as described previously (26).
EGF Receptor Autophosphorylation in the Presence and Absence of ATP-EGF (10 g/100 g of body weight) was injected via the jugular vein into anesthetized rats. The animals were sacrificed by decapitation 15 min later, and ENs were prepared as noted above. Resuspended ENs were incubated for 15 min with 5 mM ATP after which autophosphoryl-ation was initiated by adding [ 32 P]ATP. The reaction was stopped after 15 min of incubation at 37°C by adding Laemmli sample buffer after which aliquots (100 g of protein) were subjected to SDS-PAGE, alkali digestion, and radioautography as described previously (13).

RESULTS
Following in vivo administration, 125 I-insulin concentrates in ENs attained maximum levels at 2 min postinjection (19,23,27). When subsequently incubated in vitro, intraendosomal 125 I-insulin undergoes dissociation from its receptor and degradation in a temperature and ATP-dependent manner (15). Previous work showed that this insulin degradation was effected by a relatively specific endosomal acidic insulinase (14,15).

Effect of ATP on Insulin Binding and IRK Activity
In the present study we continued to evaluate the processes involved in reducing the association of intraendosomal insulin with its receptor. We examined the effect of incubating intact ENs with ATP on subsequently measured insulin and hGH binding in both solubilized and WGA-purified receptors. Table  I illustrates that, at increasing ATP concentrations (0.1, 1, 3, 5, and 10 mM), there was a progressive decrease of insulin but not hGH binding. The decrease of insulin binding was not produced by ADP, AMP, sodium pyrophosphate, adenosine, or nonhydrolyzable ATP analogs (Table II). The loss of binding activity was temperature-dependent, as no changes were observed when ENs were incubated on ice in the presence of 10 mM ATP (Table  II). These data strongly suggest that the ATP-dependent effect was specific for insulin and necessitated hydrolysis of the ␥-phosphate of ATP.
Preincubating ENs with ATP also influenced IRK activity and phosphotyrosine content (Fig. 1). In the absence of ATP, IRK activity reflected the effect of preinjected insulin (3,4,18). At 0.1 and 1.0 mM ATP there was an increase, whereas at 5 mM ATP there was a marked decrease in IRK activity even to a level below that observed in the absence of ATP. At 10 mM ATP, IRK activity was virtually abolished. IRK phosphotyrosine content was markedly increased on incubating ENs with 1 mM ATP (Fig. 1B), whereas at 5 and 10 mM ATP this was significantly reduced and virtually abolished, respectively, without affecting ␤-subunit levels.
To verify the importance of the high energy bond of ATP in producing the above changes, we incubated ENs with 0 and 1 mM ATP in the presence (5 mM) or absence of the nonhydrolyzable ATP analog, AMP-PCP. No effect of 5 mM AMP-PCP either in the presence or absence of 1 mM ATP was found (Fig. 2). The effect of the analog to reduce IRK phosphotyrosine concentration to a modest extent may reflect some inhibition of IRK autophosphorylation.
To determine whether ATP-dependent attenuation of IRK activation is specific to the endosomal compartment, we preincubated PM with ATP and observed an augmentation of WGApurified IRK activity (Fig. 3A). At 10 mM ATP the increase in PM IRK activity was greater than that observed at 1 mM in sharp contrast to what was found in ENs where preincubation with 10 mM ATP suppressed IRK activity completely.
Preincubation with ATP did not attenuate EGF receptor autophosphorylation. Thus incubating ENs from EGF-treated rats with 5 mM ATP did not significantly reduce 32 P-labeling of the EGF receptor (Fig. 3B).

Role of Endosomal Acidification
Various studies have established that in ENs there is a progressive luminal acidification through the action of ATP-dependent proton pumps (28 -31). To evaluate the role of ATPdependent acidification on the IRK activation state we incubated ENs with bafilomycin A 1 , a potent inhibitor of endosomal ATPases (31), in the presence and absence of ATP. As noted in Fig. 4A coincubation with bafilomycin produced a marked attenuation in the ability of 5 and 10 mM ATP to effect a reduction in IRK activation. Furthermore in the presence of bafilomycin the level of IRK tyrosine phosphorylation at 5 mM ATP was comparable to that observed at 1 mM ATP (Fig. 4B).

Studies on Attenuation of IRK Activation
We subsequently sought to determine the mechanism by which ATP-dependent endosomal acidification results in diminution of the IRK activation state.
Serine/Threonine Phosphorylation-Previous work demonstrated that serine/threonine phosphorylation of the ␤-subunit of the IRK results in inhibition of IRK activity (32,33). We thus determined if ATP promotes the phosphorylation of IRK on serine and threonine residues by incubating ENs with 1 or 5 mM [␥-32 P]ATP (0.5 mCi/mmol). IRKs were subsequently purified by immunoprecipitation and SDS-PAGE and subjected to two-dimensional phosphoamino acid analyses of the ␤-subunit. The IRK showed no detectable [ 32 P]phosphoserine from either the 1 or 5 mM ATP incubations (Fig. 5).
Activation of Phosphotyrosyl Phosphatase(s)-To assess whether higher ATP concentrations might promote activation of endosomal phosphotyrosyl phosphatases and effect a reduction in the phosphotyrosine content of the IRK ␤-subunit, we blocked phosphotyrosyl phosphatase activity using bpV(phen), a potent phosphotyrosyl phosphatase inhibitor (18). Coincubating ENs with bpV(phen) did not prevent the marked suppression of IRK activity seen in the presence of 5 mM ATP (Fig. 6). Furthermore, the augmented IRK activity in ENs isolated from rats pretreated with bpV(phen) was also suppressed on incubating these ENs with 5 mM ATP. Nor did bpV(phen) influence the reduction observed in ␤-subunit tyrosine phosphorylation seen in the presence of 5 mM ATP (Fig. 6B). We conclude that phosphotyrosyl phosphatase activity plays no role in effecting a reduction of endosomal IRK function at higher ATP concentrations.
Dissociation of IRK Activity and its Autophosphorylation State-To determine the relationship between IRK activity and Results are the mean of two to five experiments Ϯ S.D. Endosomes from insulin-injected rats were preincubated with the noted concentration of nucleotides at 37°C for 15 min, solubilized, and assayed for insulin binding before and after WGA purification as described under "Materials and Methods." The specific binding observed was normalized to those levels obtained with endosomes preincubated without nucleotide. All values are the means Ϯ S.D.  1. Effect on IRK activity of incubating intact ENs with ATP. Rats were injected with insulin (1.5 g/100 g of body weight) and sacrificed after 2 min. Fresh hepatic ENs were incubated for 15 min at 37°C with ATP at the indicated concentration. IRK was purified from solubilized ENs by WGA-Sepharose chromatography as indicated under "Materials and Methods." Panel A, effect of ATP on IRK activity. Equal amounts of WGA-purified IRK were assayed for exogenous tyrosine kinase activity using poly(Glu, Tyr) (4:1) as substrate. Exokinase activity was expressed as picomoles/10 min/10 fmol of insulin binding. Each point reflects the mean Ϯ S.E. of determination from four to six separate experiments. Panel B, effect of ATP on ␤-subunit phosphotyrosine content. WGA eluates (5 g of protein) were subjected to SDS-PAGE in 7.5% gels followed by transfer of proteins to Immobilon-P membranes. Membranes were probed with ␣PY or ␣960 antibodies followed by incubation with 125 I-GAR as a second antibody and exposed for autoradiography.
␤-subunit tyrosine phosphorylation we preincubated ENs with 1 or 5 mM ATP for 15 min followed by a second incubation with 5 mM ATP. As observed in Fig. 7 preincubation with 1 mM ATP followed by a second incubation with 5 mM ATP resulted in marked reduction of IRK activity in the presence of a substantial retention of its phosphotyrosine content. Sequential incubations in 1 mM ATP had no deleterious effect on either IRK activity or phosphotyrosine content. Thus endosomal acidification results in the inactivation of an autophosphorylated IRK.

Evidence for an Acidification-dependent Conformational
Change of the Endosomal IRK The above observation suggested that intraluminal events were responsible for deactivation of the endosomal IRK. To determine whether there might be a conformational change in the IRK consequent to ATP-dependent acidification we incubated ENs with ATP in the presence or absence of 10 mM DTT to assess the ease with which type I disulfide bonds might be reduced in the heterotetrameric molecule (␣ 2 ␤ 2 ). After incubating in the presence or absence of 5 mM ATP the heterotetramer was readily identified (Fig. 8, lanes 1 to 3). When the incubation contained 10 mM DTT, the heterotetramer was readily reduced to the heterodimer (␣␤) in the absence but not the presence of 5 mM ATP (lane 4 versus 5). Thus, consequent to ATP-dependent acidification of endosomes, there is a change in ␣-subunit conformation which renders the type I disulfide bonds relatively resistant to reduction by DTT. DISCUSSION The endosomal apparatus consists of a series of distinct tubulovesicular components involved in the uptake and sorting of ligand-receptor complexes (34,35). The ENs employed in these studies have been previously characterized and shown to contain Golgi elements but to be substantially free of plasma membrane and other subcellular constituents (19, 20, 36 -38). The addition of ATP to these ENs augments both dissociation and degradation of internalized insulin (15) due to stimulation of an ATP-dependent proton pump. The resulting acidification (pH 5.5) of the intraendosomal milieu activates a relatively specific insulin protease (i.e. endosomal acidic insulinase) (14 -17, 27, 39, 40). The coupling of dissociation at acid pH with insulin degradation facilitates removal of internalized receptor-bound insulin. However, given the small volume of an endocytic vesicle (Х 10 Ϫ17 liter) (41), the extent of dissociation of insulin from its receptor might be expected to be limited even at pH 5.5-6.0.
We thus considered that other mechanism(s) for abrogating IRK activation in ENs might exist. Indeed incubating ENs with ATP resulted in a loss of insulin receptor binding capacity. This effect was specific for the IRK since no decrease in binding was observed for hGH (Table I). The loss of insulin binding was not observed at 4°C, and the presence of the ␥-phosphate of ATP was necessary since no inhibition occurred with adenosine, AMP, ADP, and sodium pyrophosphate. The presence of a high energy bond was necessary as nonhydrolyzable ATP analogs FIG. 2. Absence of inhibitory effect of nonhydrolyzable ATP analogs on IRK activity. Rats were sacrificed 2 min following insulin injection (1.5 g/100 g of body weight). Fresh hepatic ENs were incubated for 15 min at 37°C with ATP in the absence or presence of 5 mM AMP-PCP. ENs were solubilized and IRK purified by WGA-Sepharose chromatography as noted under "Materials and Methods." Panel A, effect of AMP-PCP on IRK activity. WGA-purified IRK was assayed for activity using poly(Glu, Tyr) (4:1) as a substrate. Exokinase activity was expressed as picomoles/10 min/10 fmol insulin binding. Panel B, effect of AMP-PCP on the ␤-subunit phosphotyrosine content. WGA eluates (5 g of protein) were subjected to SDS-PAGE in 7.5% gels. Proteins were transferred to Immobilon-P membranes, which were probed with ␣PY antibody followed by incubation with 125 I-GAR and exposed for autoradiography.

FIG. 3. A, effect on IRK activity of incubating PM and ENs with ATP.
Fresh hepatic PM and EN fractions, prepared at 30 s and 2 min post insulin (1.5 g/100 g of body weight), were incubated for 15 min at 37°C with ATP at the indicated concentrations. Cell fractions were solubilized, and IRK, purified by WGA-Sepharose chromatography, was assayed for activity using poly(Glu, Tyr) (4:1) as noted under "Materials and Methods." Exokinase activity was expressed as picomoles/10 min/10 fmol of insulin binding. B, effect on EGF receptor autophosphorylation of incubating ENs with ATP. ENs, prepared 15 min after EGF injection (10 g/100 g of body weight), were incubated in the presence or absence of ATP (5 mM) prior to conducting autophosphorylation with [ 32 P]ATP, SDS-PAGE, and autoradiography as described under "Materials and Methods." were unable to produce the effect (Table II). The reduction of insulin binding activity represents another mechanism for sustaining the dissociation-degradation sequence for insulin in ENs.
An ATP-dependent process is implicated in deactivation of the IRK within ENs. Although IRK activity and ␤-subunit phosphotyrosine content increased in parallel at 0.1 and 1 mM ATP, they were both reduced at higher ATP concentrations (Fig. 1). This was not due to proteolysis of the ␤-subunit since none was lost in these experiments. The effect requires the high energy bond of ATP, and was unique to the endosomal compartment since PM IRK activity was not suppressed at high ATP concentrations. ENs have a slightly acidic interior maintained by an ATP-dependent proton pump (15, 30, 31). FIG. 5. Phosphoamino acid analyses of the IRK ␤-subunit after incubating intact ENs with 1 and 5 mM [␥-32 P]ATP. Hepatic ENs, prepared 2 min following insulin (1.5 g/100 g of body weight), were incubated for 15 min at 37°C with 1 or 5 mM [␥-32 P]ATP (0.5 mCi/ mmol), solubilized, and immunoprecipitated with ␣960 antibodies. Immunoprecipitates were resolved on SDS-PAGE and transferred to Immobilon-P membranes as noted under "Materials and Methods." Regions of the Immobilon-P membranes containing the IRK ␤-subunit were localized by autoradiography, excised, and subjected to acid hydrolysis to release phosphoamino acids, which were separated by twodimensional thin layer electrophoresis as described under "Materials and Methods." Thin layer electrophoresis was performed using equal amounts of 32 P label (500 cpm) and phosphoserine (pS), phosphothreonine (pT), and phosphotyrosine (pY), which were localized by ninhydrin staining. A second experiment yielded comparable results.
FIG. 6. Effect of inhibiting phosphotyrosyl phosphatase activity by bpV(phen) on ATP-dependent changes in endosomal IRK activity. Hepatic ENs, prepared at 2 or 15 min after insulin (1.5 g/100 g of body weight) or bpV(phen) (0.6 mol/100 g of body weight) respectively, were incubated for 15 min at 37°C with ATP in the absence or presence of 0.1 mM bpV(phen). ENs were solubilized and IRK purified by WGA-Sepharose chromatography as described under "Materials and Methods." Panel A, effect of different ATP concentrations on IRK activity. WGA-purified IRK preparations were assayed for activity using poly(Glu, Tyr)(4:1) as substrate. Exokinase activity was expressed as picomoles/10 min/10 fmol of insulin binding. Panel B, effect of different ATP concentrations on IRK ␤-subunit phosphotyrosine content. WGA eluates (5 g of protein) were subjected to SDS-PAGE in 7.5% gels, transferred to Immobilon-P membranes, which were probed sequentially with ␣PY and 125 I-GAR followed by autoradiography. Comparable results were observed in three separate studies.
FIG. 4. Effect of the ATPase inhibitor, bafilomycin, on the ATP-dependent deactivation of endosomal IRK activity. Hepatic ENs, prepared 2 min post insulin (1.5 g/100 g of body weight), were preincubated in the absence or presence of 1 M bafilomycin A 1. After 30 min at 37°C the incubation was continued for 15 min with ATP at the indicated concentrations. Panel A, effect of bafilomycin A 1 on ATP-dependent inhibition of IRK activity. WGA-purified IRK tyrosine kinase activity was assayed using poly(Glu, Tyr) (4:1) as substrate and expressed as a percent of that obtained from incubations conducted without ATP. IRK activity from incubations with or without bafilomycin A 1, and before ATP additions, were 1.4 and 1.3 pmol/10 min/10 fmol of insulin binding, respectively. Panel B, effect of bafilomycin on ATP-dependent regulation of IRK ␤-subunit autophosphorylation. WGA eluates (5 g of protein) were subjected to SDS-PAGE in 7.5% gels followed by transfer of proteins to Immobilon-P membranes. The membranes were then probed with ␣PY or ␣960 antibodies followed by 125 I-GAR and autoradiography. The same findings were made in three repeat experiments.
The observation that ATP-dependent inhibition of IRK activity was reversed by bafilomycin (Fig. 4) strongly supports the idea that proton pump acidification of ENs is critical to this process. It is noteworthy that the levels of ATP promoting IRK inhibition approximate estimated intracellular concentrations (42)(43)(44).
We explored the mechanism by which ATP-dependent endosomal acidification effects inactivation of the IRK. Since serine/ threonine phosphorylation of the IRK has been show to reduce IRK activation (28,29,45), we examined the phosphoamino acid content of the endosomal IRK incubated in the presence of 1 versus 5 mM ATP. Two-dimensional phosphoamino acid analyses showed that serine/threonine phosphorylation of IRK was not augmented at higher ATP concentrations. The marked reduction in phosphotyrosine content of the IRK at 5 mM ATP was not a consequence of augmented phosphotyrosyl phosphatase activity, since bpV(phen) did not antagonize the inhibitory effect of 5 mM ATP on either IRK phosphotyrosine content or activity.
The ATP inhibitory effect was shown to be independent of the phosphorylation state of the ␤-subunit (Fig. 7), suggesting that an intrinsic defect in IRK function secondary to a conformational change might explain our observations. Indeed the ability of DTT to reduce tetrameric IRK molecules was signif-icantly decreased subsequent to the incubation of ENs with 5 mM ATP (Fig. 8). The reduced susceptibility of the type I disulfide bond between the ␣and ␤-subunits to DTT implies the occurrence of a pH-dependent modification of the IRK. We suggest that a conformational change of the IRK, effected by the ATP-dependent intraluminal drop in pH, was transmitted to its cytosolic domain producing decreased IRK activity. This is consistent with crystallographic studies (46,47) suggesting that, whereas activation of IRK occurs through a trans-autophosphorylation reaction, deactivation occurs through a cisintramolecular mechanism (cis-inhibition) (46). The presumed change in the intraluminal portion of the IRK may be responsible for the observed decrease in insulin binding (cf. Table I).
This study indicates that the regulation of endosomal IRK activity is multifaceted and that the deactivation involves several discrete components. Previous work has shown that insulin signaling occurs from the endosomal system (9 -11). The present study supports the view that there is a temporal window of signaling delimited in part by progressive acidification of ENs due to the activity of ATP-dependent proton pumping. Endosomal acidification contributes to IRK inactivation by: 1) promoting insulin dissociation from the IRK, 2) activating endosomal acidic insulinase, 3) decreasing the binding capacity of IRK, and 4) altering the conformation of the IRK thus reducing intrinsic activity.
Other studies have documented the importance of endosomal acidification in regulating a range of biological processes. Vesicular stomatitis and rabies viruses enter cells through receptor-mediated endocytosis but are rendered competent to enter cytosol after accessing the low pH of ENs. In this environment the viral envelope undergoes a conformational transition permitting fusion of viral membrane with endosomal membranes (48,49). This transition involves the exposure of a hydrophobic segment within the glycoprotein whose ability to interact with membranes effects fusion and extrusion of the viral core through the wall of ENs (49). Low pH-driven conformational changes in ENs have been described for the diphtheria toxin and constitute a prerequisite for the subsequent reduction of the diphtheria toxin interchain disulfide bond, the rate-limiting step in translocation of toxin into cytosol (50). Recent data suggest that a key determinant regulating dephosphorylation and resensitization of the ␤-adrenergic receptor is the association of internalized receptor and phosphatase in a step involv- FIG. 7. Inhibition by ATP of the kinase activity of endosomal IRK is independent of the state of ␤-subunit tyrosine phosphorylation. Hepatic ENs, prepared 2 min after insulin (1.5 g/100 g of body weight), were incubated for 15 min at 37°C with 0, 1, or 5 mM ATP followed by a second 15-min incubation with ATP at the indicated concentrations. ENs were solubilized, and IRK were purified by WGA-Sepharose chromatography. Panel A, phosphotyrosine content of the ␤-subunit of the insulin receptor. WGA eluates (5 g of protein) were subjected to SDS-PAGE in 7.5% gels. Proteins were transferred to Immobilon-P membranes and probed with ␣PY or ␣960 antibodies and 125 I-GAR as second antibody prior to autoradiography. Panel B, IRK activity. WGA-purified IRK activity was assayed, using poly(Glu, Tyr) (4:1) as substrate, and expressed as picomoles/10 min/10 fmol of insulin binding. Comparable results were seen in a second experiment.
FIG. 8. Effect of ATP on DTT-dependent reduction of endosomal IRK species. Hepatic ENs, prepared 2 min after insulin (1.5 g/100 g of body weight), were incubated for 15 min at 37°C in the absence or presence of 10 mM DTT and 5 mM ATP. ENs were centrifuged, resuspended, and treated as described under "Materials and Methods." Samples were applied without heating to SDS-PAGE (gradient resolving gel, 3-10% acrylamide) in the absence of reductant. Proteins were transferred to Immobilon-P membranes and probed with ␣960 and 125 I-GAR antibodies prior to autoradiography. The same finding was made in three separate experiments.
ing pH-sensitive conformational change(s) in receptor and/or phosphatase (51).
The regulation of intraendosomal pH may play a role in modulating insulin sensitivity in vivo (52)(53)(54) since, in type II diabetic patients, the acidotropic agent chloroquine improved glucose metabolism (55)(56)(57)(58)(59)(60)(61). Because chloroquine inhibits intraendosomal insulin degradation it may be inferred that the endosomal accumulation of intact insulin is responsible for the improved insulin sensitivity. The present work raises the possibility that the metabolic effects of chloroquine are due to an influence on IRK conformation and function. Indeed it may be that pH-dependent disturbances in IRK function contribute to the pathogenesis of type II diabetes mellitus.