Phosphoinositide 3-Kinase Activity Is Necessary for Insulin-dependent Inhibition of Apolipoprotein B Secretion by Rat Hepatocytes and Localizes to the Endoplasmic Reticulum*

Insulin inhibits apolipoprotein B (apoB) secretion by primary rat hepatocytes through activation of phosphoinositide 3-kinase (PI 3-K). Current studies demonstrate that the PI 3-K inhibitor wortmannin inhibits both basal and insulin-stimulated PI 3-K activities. Wortmannin and LY 294002, two structurally distinct PI 3-K inhibitors, prevent insulin-dependent inhibition of apoB secretion in a dose-dependent manner. To link PI 3-K activation to insulin action on apoB, we investigated whether insulin induced localization of activated PI 3-K to the endoplasmic reticulum (ER), where apoB biogenesis is initiated. Insulin action results in a significant redistribution of PI 3-K to a low density microsome (LDM) fraction containing apoB protein and apoB mRNA. Insulin stimulates a significant increase in PI 3-K activity associated with insulin receptor substrate-1 as well as an increase in insulin receptor substrate-1/PI 3-K mass in LDM. Subfractionation of LDM on sucrose density gradients shows that insulin significantly increases the amount of PI 3-K present in an ER fraction containing apoB. Insulin stimulates PI 3-K activity in smooth and rough microsomes isolated from rat hepatocytes, the latter of which contain rough ER as demonstrated by electron microscopy. Studies indicate that 1) PI 3-K activity is necessary for insulin-dependent inhibition of apoB secretion by rat hepatocytes; 2) insulin action leads to the activation and localization of PI 3-K in an ER fraction containing apoB; and 3) insulin stimulates PI 3-K activity in the rough ER.


Insulin inhibits apolipoprotein B (apoB) secretion by primary rat hepatocytes through activation of phosphoinositide 3-kinase (PI 3-K). Current studies demonstrate that the PI 3-K inhibitor wortmannin inhibits both basal and insulin-stimulated PI 3-K activities. Wortmannin and LY 294002, two structurally distinct PI 3-K inhibitors, prevent insulin-dependent inhibition of apoB secretion in a dose-dependent manner. To link PI 3-K activation to insulin action on apoB, we investigated whether insulin induced localization of activated PI 3-K to the endoplasmic reticulum (ER), where apoB biogenesis is initiated. Insulin action results in a significant redistribution of PI 3-K to a low density microsome (LDM) fraction containing apoB protein and apoB mRNA. Insulin stimulates a significant increase in PI 3-K activity associated with insulin receptor substrate-1 as well as an increase in insulin receptor substrate-1/PI 3-K mass in LDM. Subfractionation of LDM on sucrose density gradients shows that insulin significantly increases the amount of PI 3-K present in an ER fraction containing apoB. Insulin stimulates PI 3-K activity in smooth and rough microsomes isolated from rat hepatocytes, the latter of which contain rough ER as demonstrated by electron microscopy. Studies indicate that 1) PI 3-K activity is necessary for insulin-dependent inhibition of apoB secretion by rat hepatocytes; 2) insulin action leads to the activation and localization of PI 3-K in an ER fraction containing apoB; and 3) insulin stimulates PI 3-K activity in the rough ER.
Apolipoprotein B (apoB) 1 is a major structural protein component of chylomicrons, very low density lipoproteins (VLDL), and low density lipoproteins and is required for the assembly and secretion of triglyceride-rich lipoproteins. ApoB synthesis and secretion are regulated by insulin (1). Insulin suppresses VLDL triglyceride and apoB secretion by rat hepatocytes (2)(3)(4), human hepatocytes (5), human hepatoma cell line HepG2 (6), and human fetal intestine (7). Studies in primary rat hepatocytes indicate that the inhibition of apoB secretion by insulin is a result of enhanced intracellular degradation of newly synthesized apoB (8) and that the effect of insulin is mediated by the insulin receptor (9).
Although insulin and platelet-derived growth factor stimulate a similar increase in PI 3-K activity, only insulin causes a substantial increase in glucose transport, lipogenesis, and glycogen synthesis in adipocytes (23,24). This indicates that PI 3-K activation is necessary but not sufficient to elicit insulinspecific metabolic responses. Insulin and platelet-derived growth factor have been shown to differentially regulate PI 3-K by recruitment of the enzyme to specific intracellular sites (25,26). Furthermore, PI 3-K activation is necessary for insulininduced translocation of GLUT 4 transporters (19,20), and PI 3-K has been shown to localize in vesicles containing GLUT 4 transporters (27). These findings suggest that both activation of PI 3-K and localization of the active enzyme to a specific subcellular compartment are necessary to selectively elicit metabolic responses to insulin.
Current studies extend previous observations, which indicate that activation of PI 3-K by insulin is required in the regulation of apoB secretion by insulin in rat hepatocytes (22). Wortmannin inhibits basal and insulin-stimulated PI 3-K activities without interfering with insulin-induced association of IRS-1 with PI 3-K. Wortmannin and LY 294002, two chemically distinct PI 3-K inhibitors, prevent insulin-dependent inhibition of apoB secretion in a dose-dependent manner. To link PI 3-K activation to insulin action on apoB metabolism, we investigated whether insulin induced the localization of activated PI 3-K to an intracellular compartment involved in apoB biogenesis. Results indicate that PI 3-K activity is necessary for insulin-dependent inhibition of apoB secretion by rat hepatocytes and that insulin action leads to the activation and localization of PI 3-K in the endoplasmic reticulum (ER). Localization of PI 3-K in the ER correlates with the site of insulin action on apoB.

EXPERIMENTAL PROCEDURES
Materials-Sprague-Dawley rats (200 -350 g) were obtained from Charles River Laboratories (Wilmington, MA). Waymouth's medium 752/1 and TRIzol TM were purchased from Life Technologies, Inc. (Bethesda, MD). Protein G-Sepharose and RNase inhibitor were obtained from Pharmacia Biotech Inc. Rabbit anti-IRS-1 and anti-PI 3-K (p85) antibodies were obtained from Upstate Biotechnology, Inc., Lake Placid, NY. Rabbit anti-calnexin and anti-protein-disulfide isomerase antibodies were from StressGen Biotechnologies Corp. (Victoria, Canada). Rabbit polyclonal antibodies to purified rat apoB were prepared as described (8,28). Mouse monoclonal anti-Golgi 58K and anti-Golgi ␤-COP antibodies, sodium heparin, yeast cytochrome c, NADPH, sucrose (molecular biology grade), and wortmannin were from Sigma. LY 294002 was purchased from Calbiochem. Rat apoB cDNA was a gift from Drs. Mark Sowden and Harold Smith (University of Rochester). [ 32 P]ATP (3000 Ci/mmol) was from NEN Life Science Products. Horseradish peroxidase-conjugated antibodies and ECL TM reagents were from Amersham Corp. The Random Primer Extension kit was obtained from Boehringer Mannheim.
Subcellular Fractionation of Hepatocytes-Isolated hepatocytes (1-2 ϫ 10 8 cells) in Waymouth's medium were incubated with or without insulin (100 nM) for 5 min at 37°C. Afterward, cells were harvested by centrifugation at 50 ϫ g for 3 min and washed twice in Hanks' balanced salt solution. Hepatocytes were fractionated as described (30). Cells were diluted 1:6 (v/v) in ice-cold homogenization buffer containing 0.25 M sucrose, 20 mM HEPES, pH 7.5, 1 mM EDTA, 1 mM dithiothreitol, 10 g/ml leupeptin, 10 g/ml aprotinin, 3.9 mg/ml benzamidine, 1 mM 4-(2-aminoethyl)benzenesulfonylfluoride, 4 mM sodium vanadate, as well as 1 mg/ml sodium heparin and RNase inhibitor to inhibit cellular RNases. Cell suspensions were homogenized by 20 passages through a ball bearing homogenizer (31). The cell homogenate was centrifuged in a Beckman Ti-80 rotor at 700 ϫ g for 20 min to remove cell debris. The 700 ϫ g supernatant was centrifuged at 12,000 ϫ g for 15 min to sediment plasma membranes/mitochondria/nuclei (PMN). The 12,000 ϫ g supernatant was centrifuged at 28,000 ϫ g for 15 min to yield a pellet of high density microsomes (HDM). The 28,000 ϫ g supernatant was centrifuged at 380,000 ϫ g for 30 min. The pellet obtained is referred to as low density microsomes (LDM), and the supernatant is called post-LDM supernatant (PLS). Protein content in each fraction was measured by a modified Lowry procedure (32). All procedures were carried out at 4°C. Fractions isolated were either resuspended in Laemmli sample buffer (33), heated at 100°C for 10 min, and stored at Ϫ80°C for subsequent analysis by SDS-polyacrylamide gel electrophoresis (PAGE) or used for immunoprecipitation as described.
Fractionation of Low Density Microsomes by Sucrose Density Gradient Centrifugation-LDM fractions were resuspended in homogenization buffer using a glass Dounce homogenizer, and then 1 ml was applied to the top of an 11-ml discontinuous sucrose gradient composed of 0.4 -1.5 M sucrose in 0.1 M increments (0.92 ml each). Sucrose solutions also contained 5 g/ml leupeptin, 5 g/ml aprotinin, 0.5 mM 4-(2-aminoethyl)benzenesulfonylfluoride, 4 mM sodium vanadate, and 1 mg/ml sodium heparin. The sucrose gradient was formed at room temperature and held at 4°C for 20 -24 h for the step gradient to linearize before centrifugation. After application, samples were centrifuged in a Beckman SW 41 Ti rotor at 130,000 ϫ g for 3 h at 4°C. Following centrifugation, 12 1-ml fractions were collected from the bottom of the tube by gravity. Fractions were then diluted in 0.25 M sucrose and 1 mM sodium vanadate plus protease inhibitors and recentrifuged in a Beckman Ti-80 rotor at 150,000 ϫ g for 1 h at 4°C to repellet microsomes. Pellets were solubilized in Laemmli sample buffer, heated at 100°C for 10 min, and stored at Ϫ80°C for subsequent analysis by SDS-PAGE.
Isolation of Rough and Smooth Microsomes-Isolated hepatocytes were incubated with or without insulin (100 nM) for 5 min at 37°C. Cells were harvested by centrifugation and washed twice in Hanks' balanced salt solution. Rough and smooth microsomes were isolated from cells as described by Depierre and Dallner (34). Cells were diluted 1:6 (v/v) in ice-cold homogenization buffer containing 0.25 M sucrose, 10 g/ml leupeptin, 3.9 mg/ml benzamidine, 50 g/ml trypsin inhibitor, 10 g/ml aprotinin, 1 mM ␣-toluenesulfonyl fluoride, 4 mM sodium vanadate, and 1 mg/ml sodium heparin. Cell suspensions were homogenized by 20 passages through a ball bearing homogenizer. The cell homogenate was centrifuged at 10,000 ϫ g for 20 min at 4°C to remove cell debris. The 10,000 ϫ g supernatant (3.5 ml) was applied to the top of a sucrose gradient composed of 1 ml of 0.6 M sucrose and 15 mM CsCl layered over 2 ml of 1.3 M sucrose and 15 mM CsCl. Samples were centrifuged in a Beckman Ti-80 rotor at 102,000 ϫ g for 90 min at 4°C. Rough microsomes formed a pellet, and smooth microsomes floated at the 0.6 M sucrose and 1.3 M sucrose interface. Microsomes were either used immediately for immunoprecipitation or processed for electron microscopy.
Organelle Marker Protein Assays-Subcellular fractions from rat hepatocytes were characterized by analyzing both enzyme and immunological markers specific for the endoplasmic reticulum and Golgi apparatus. NADPH-cytochrome c reductase specific activity in each fraction was determined as described (35). Fractions were immunoblotted with antibodies to calnexin, protein-disulfide isomerase, Golgi 58K, and Golgi ␤-COP protein.
ApoB Radioimmunoassay-Culture media were collected, and cells were washed twice in Hanks' balanced salt solution. Cells were then scraped into 50 mM barbital buffer (pH 8.6) containing 0.5% (v/v) Triton X-100, 2 mM EDTA, 5 mM benzamidine, and 2 mM ␣-toluenesulfonyl fluoride. Cell lysates were sonicated, and clarified supernatants were prepared by centrifugation. ApoB in media and in cell homogenates was measured by competitive radioimmunoassay using rat VLDL-apoB as a standard and a monoclonal antibody equally reactive to B48 and B100 (36). Protein concentrations of cell lysates were measured by a modified Lowry procedure (32), and the concentration of apoB was calculated relative to mg of cell protein/plate.
Immunoprecipitations-Primary rat hepatocytes or subcellular fractions were resuspended in 50 mM Tris-HCl, pH 7.4, 140 mM NaCl, 10 mM EDTA, 50 mM NaF, 4.5 mg/ml tetrasodium pyrophosphate, 1% Nonidet P-40, 10% glycerol, 10 g/ml aprotinin, 3.9 mg/ml benzamidine, 4 mM sodium vanadate, 1 mM ␣-toluenesulfonyl fluoride, and 1 mM dithiothreitol. The suspensions were homogenized by 10 strokes in a glass Dounce homogenizer, and centrifuged at 10,000 ϫ g for 10 min at 4°C. Clarified supernatants were collected and incubated with indicated antibodies overnight at 4°C and then with Protein-G Sepharose beads for 2 h at 4°C. Immunoprecipitates were collected by centrifugation at 10,000 ϫ g for 2 min at 4°C; washed twice in phosphate-buffered saline and 1% Nonidet P-40 and then twice in 100 mM Tris-HCl, pH 7.4, and 500 mM LiCl 2 ; and finally washed twice in 10 mM Tris-HCl, pH 7.4, and 100 mM NaCl. All wash buffers contained 1 mM dithiothreitol and 0.2 mM sodium vanadate added just before use. Washed immunoprecipitates were either assayed for PI 3-K activity, as indicated, or resuspended in Laemmli sample buffer, heated at 100°C for 10 min, and analyzed by SDS-PAGE.
PI 3-K Activity Assay-The indicated immunoprecipitates were assayed for PI 3-K activity exactly as described by Way and Mooney (37). Radiolabeled lipid products of PI 3-K were separated by thin layer chromatography and quantitated by PhosphorImager scanning (Molecular Dynamics, Inc., Sunnyvale, CA).
SDS-PAGE and Immunoblotting-Solubilized proteins in Laemmli sample buffer were separated by SDS-PAGE on 3.5-24% (w/v) Acrylaide TM /acrylamide gradient gels (28) and electrophoretically transferred to PVDF membranes at 0.7 A for 3 h in 25 mM Trizma base, pH 8.5, 192 mM glycine, and 0.1% SDS. Membranes were blocked overnight at 4°C in 20 mM Tris-HCl, pH 7.5, and 500 mM NaCl (hereafter referred to as Tris-buffered saline (TBS)) containing 1% bovine serum albumin, 2% nonfat dry milk, 0.2% Tween 20, and 0.02% thimerosal prior to incubation with primary antibodies overnight at 4°C. Blots were washed twice in TBS and 0.1% Tween 20 for 30 min at room temperature to remove unbound antibody. Blots were incubated with horseradish peroxidase-conjugated secondary antibody at a 1:5000 dilution for 1 h at room temperature and then washed three times in TBS and 0.1% Tween 20, three times in TBS and 0.3% Tween 20, and once in TBS. Immunoreactive proteins were detected by chemiluminescence using ECL TM according to the manufacturer's instructions (Amersham Corp.). Immunoreactivity was visualized by film exposure and quantitated by laser-scanning densitometry (Molecular Dynamics, Inc.).
RNA Isolation-Total RNA was isolated from subcellular fractions using TRIzol TM according to the manufacturer's instructions (Life Technologies, Inc.). Subcellular fractions were isolated from rat hepatocytes as described above with the exception that 250 mM KCl, 50 mM MgCl 2 , 1 mg/ml sodium heparin, and RNase inhibitor were also added to the homogenization buffer to inhibit RNases. RNA isolated from each subcellular fraction was resuspended in the same volume of RNase-free H 2 O, and RNA concentration was determined by optical absorbance at 260 nm. Electrophoretic analysis of isolated RNA on 1% agarose denaturing gels and visualization of RNA by ethidium bromide staining showed that isolated RNA was intact. RNA samples were stored in 100% ethanol and 73 mM sodium acetate at Ϫ80°C.
cDNA Probe Preparation and Labeling-ApoB cDNA fragments were excised from the vector and then isolated by agarose gel electrophoresis. ApoB cDNA was radiolabeled with [ 32 P]dATP by random primer extension according to the manufacturer's instructions (Boehringer Mannheim). Following labeling reactions, radiolabeled probes were purified through a Sephadex G-50 column to remove unincorporated labeled nucleotides. Specific activities of radiolabeled cDNA probes were typically 1.5 ϫ 10 9 cpm/g.
Electron Microscopy-For Spurr epoxy embedding, smooth microsomes were pelleted at 150,000 ϫ g for 30 min and fixed as a suspension in 2.5% gluteraldehyde and 50 mM KH 2 PO 4 , pH 7.4, at room temperature for 3 h. Afterward, the microsomes were recentrifuged, and the pellets were washed three times in 50 mM KH 2 PO 4 , pH 7.4, to remove the fixative. Final microsome pellets were resuspended in 200 l of 50 mM KH 2 PO 4 , pH 7.4, and diluted 1:1 (v/v) with 3% agarose. The mixture was allowed to solidify and was stored at 4°C. Postfixation of microsomes was carried out in 1% osmium tetroxide and 100 mM phosphate buffer for 60 min. Samples were dehydrated in a graded series of acetone, infiltrated with Spurr epoxy resin, embedded in fresh resin, and polymerized overnight at 70°C. The specimen blocks were thin sectioned and then stained with uranyl acetate and lead citrate and were examined and photographed using a Hitachi model H-7100 transmission electron microscope.
For Lowicryl embedding, rough microsomes were pelleted at 150,000 ϫ g for 30 min and fixed as a suspension in 2% paraformaldehyde and 50 mM KH 2 PO 4 , pH 7.4, at room temperature for 1 h. Afterward, the microsomes were recentrifuged, and the pellets were washed three times in 50 mM KH 2 PO 4 , pH 7.4, to remove the fixative. Final microsome pellets were resuspended in 200 l of 50 mM KH 2 PO 4 , pH 7.4, and diluted 1:1 (v/v) with 3% agarose. The mixture was allowed to solidify and was stored at 4°C. Samples were dehydrated in a graded series of methanol, infiltrated and embedded in Lowicryl, a low temperature embedding resin, and polymerized overnight with a UV lamp.
Statistical Methods-Results are expressed as the mean Ϯ S.D. from at least three independent rat liver preparations. The degree of significance of observed differences between means was determined by Student's t test.

Wortmannin Inhibits Basal and Insulin-stimulated PI 3-K Activity in Rat
Hepatocytes-Although wortmannin has been shown to inhibit insulin-stimulated PI 3-K activity in adipocytes (19 -21), epithelial cells (38), neutrophils (39), and skeletal muscle (40), the ability of wortmannin to inhibit PI 3-K activity in primary rat hepatocytes has not been evaluated. In two independent rat liver experiments, insulin stimulated the lipid kinase activity of PI 3-K in anti-phosphotyrosine immunoprecipitates of cell lysates 3.1 times and 3.3 times over basal activity (Fig. 1A). Wortmannin completely abolished basal and insulin-stimulated PI 3-K activities in primary rat hepatocytes.
Wortmannin Does Not Prevent Interaction of IRS-1 with PI 3-K-A potential explanation for wortmannin inhibition of PI 3-K activity is that wortmannin prevents the interaction of IRS-1 with PI 3-K, and thus subsequent PI 3-K activation. Coimmunoprecipitation experiments were performed to evaluate IRS-1/PI 3-K interaction in cells incubated with insulin with or without wortmannin (Fig. 1B). Insulin stimulated the association of IRS-1 with the p85 subunit of PI 3-K in primary rat hepatocytes, and this association was not affected by preincubation of cells with wortmannin. Results indicate that wortmannin inhibits PI 3-K activity without affecting the interaction of IRS-1 with PI 3-K.
PI 3-K Inhibitors Prevent Insulin-dependent Inhibition of ApoB Secretion-Previous studies have shown that wortmannin prevents insulin-dependent inhibition of apoB secretion by primary rat hepatocytes (22). To extend studies on the potential role of PI 3-K in insulin regulation of apoB secretion, we examined the potency of wortmannin and LY 294002, two structurally unrelated PI 3-K inhibitors (41,42), to inhibit the

FIG. 1. Effects of insulin and wortmannin on PI 3-K activity (A) and association of IRS-1 with PI 3-K in rat hepatocytes (B).
Primary rat hepatocytes were incubated with or without wortmannin (1 M) for 30 min and then treated with or without insulin (100 nM) for 5 min. A, phosphotyrosine (PY)-containing proteins were immunoprecipitated from cell lysates with anti-phosphotyrosine antibody (4G10) and assayed for PI 3-K activity using phosphatidylinositol as substrate and [ 32 P]ATP. An autoradiograph of a thin layer chromatography plate is shown in which 32 P-labeled PI-3-P is indicated. B, cell lysates were immunoprecipitated (I.P.) with anti-IRS-1 antibodies. Proteins in anti-IRS-1 immunoprecipitates were separated by SDS-PAGE, transferred to PVDF membranes, and immunoblotted (I.B.) with antibodies to the p85 subunit of PI 3-K. Anti-p85 binding was detected by chemiluminescence. Similar results were obtained in a repeated experiment.
insulin effect on apoB secretion. Both wortmannin and LY 294002 inhibited the insulin effect in a dose-dependent manner (data not shown). The IC 50 for wortmannin was 94 Ϯ 23 nM, and the IC 50 for LY 294002 was 37 M. Both inhibitors also prevented insulin-dependent decrease in cellular apoB as determined by radioimmunoassay (data not shown).
Distribution of Microsomal Marker Proteins in Subcellular Fractions Isolated from Rat Hepatocytes-For PI 3-K to mediate insulin action on apoB secretion, we hypothesize that PI 3-K must extend its effects to the ER where apoB biogenesis is initiated. We therefore investigated whether insulin action would lead to the localization of activated PI 3-K to the ER. A rapid subcellular fractionation technique was chosen to test this hypothesis because subcellular fractions can be isolated without apparent dephosphorylation of tyrosine-phosphorylated proteins or loss of PI 3-K activity (30,43). In addition, we have demonstrated intact apoB protein and apoB mRNA by this method (data not shown).
To characterize subcellular compartments enriched in apoB and insulin-activated PI 3-K, HDM, LDM, and PLS were isolated from rat hepatocytes, and protein and organelle marker enzymes in the fractions were determined. About 86% of total homogenate protein was recovered in subcellular fractions. On average, 7% of homogenate protein was present in HDM, 15% in LDM, and 63% in PLS (Fig. 2A). The distribution of ER and Golgi membranes in various subcellular fractions was determined by immunoblotting for specific organelle marker proteins and by classical marker enzyme assays. Greater than 90% of Golgi 58K, a marker protein for the Golgi apparatus (44,45), was present in PLS (Fig. 2B). NADPH-cytochrome c reductase (35) and calnexin (46 -48), two marker proteins specific for the ER, were most enriched in HDM (Fig. 2, C and D). HDM contained 66% of total calnexin, and had a 3.7-fold increase in the specific activity of NADPH-cytochrome c reductase compared with the total homogenate. Significant amounts of ER markers were also present in LDM that was shown to contain 41% of total calnexin and a 2.6-fold increase in the specific activity of NADPH-cytochrome c reductase compared with the total homogenate (Fig. 2, C and D). Assuming that the distri-bution of organelle markers is representative of the distribution of ER-and Golgi-derived vesicles, results indicate that HDM and LDM fractions contain most ER vesicles, and PLS fraction consists of soluble cellular constituents and most Golgi vesicles.
Distribution of ApoB Protein and mRNA in Subcellular Fractions-After fractionation of hepatocytes into HDM, LDM, and PLS, proteins present in each fraction were separated by SDS-PAGE, electrophoretically transferred to membranes, and immunoblotted with specific polyclonal antibodies to rat apoB. Almost all of the apoB48 and apoB100 were recovered in HDM and LDM fractions. On average, 51% of apoB48 and 52% of apoB100 were present in HDM, and 44% of apoB48 and 39% of apoB100 were present in LDM (Fig. 3A). The remainder, about 7% of apoB48 and 9% of apoB100, was isolated in PLS and was probably present in Golgi vesicles.
To determine the subcellular distribution of apoB mRNA, total RNA was isolated from each fraction. The amount of apoB mRNA in each fraction was analyzed by slot blotting and hybridization of blots to radiolabeled rat apoB cDNA probes. On average, 70% of the total apoB mRNA was isolated in LDM, 30% in HDM, and 4% in PLS (Fig. 3B). These findings indicate that LDM contains the majority of microsomes bearing apoB mRNA.
Insulin Stimulates the Redistribution of PI 3-K to Low Density Microsomes-To evaluate the effect of insulin on the subcellular distribution of PI 3-K in rat hepatocytes, we determined the amount of PI 3-K in subcellular fractions isolated from hepatocytes with or without insulin treatment. PI 3-K was immunoprecipitated from each fraction with specific antibodies to the p85 regulatory subunit of PI 3-K, and p85 was subsequently evaluated by immunoblotting with the same antibody. As expected, p85 mass in the total cell homogenate was similar with or without insulin treatment, and only minor changes in the amount of p85 were observed in PMN and HDM fractions with insulin (Fig. 4). Of particular interest was the reciprocal loss of p85 from PLS and gain of p85 in LDM following insulin stimulation. Results indicate that insulin stimulates the redistribution of PI 3-K to LDM in rat hepatocytes.

FIG. 2. Characterization of subcellular fractions from rat hepatocytes.
Subcellular fractions were prepared from the total homogenate of hepatocyte suspensions as described under "Experimental Procedures." A, protein recovered in mg of protein/liver is the mean Ϯ S.D. from nine rat livers. B, proteins from subcellular fractions were solubilized in Laemmli sample buffer, separated by SDS-PAGE, transferred to PVDF membranes, and immunoblotted with monoclonal antibody to Golgi 58K. Antibody binding was detected by chemiluminescence and film exposure. Exposed films were quantitated by densitometric scanning. Total amount of Golgi 58K in each fraction is expressed as a percentage of the sum total of Golgi 58K in all fractions (mean Ϯ S.D., n ϭ 4 livers). C, NADPH-cytochrome c reductase specific activity was measured as described (35) (mean Ϯ S.D., n ϭ 7 livers). D, proteins in subcellular fractions were immunoblotted with polyclonal antibodies to calnexin, and results were quantitated by densitometric scanning. Total amount of calnexin in each fraction is expressed as a percentage of the sum total of calnexin in all fractions (mean Ϯ S.D., n ϭ 4 livers).

Effect of Insulin on Subcellular Localization of IRS-1-associated PI 3-K Activity-
To determine the subcellular localization of insulin-activated PI 3-K activity in rat hepatocytes, hepatocytes were incubated with or without insulin for 5 min prior to subcellular fractionation. To assay for insulin-activated PI 3-K, the fractions were first immunoprecipitated with anti-IRS-1 antibodies, since association of PI 3-K with tyrosine-phosphorylated IRS-1 is the major mechanism for activation of PI 3-K by insulin (49). Following immunoprecipitation, the amount of PI 3-K activity in anti-IRS-1 immunoprecipitates was determined by an in vitro lipid kinase assay (Fig. 5A). Basal levels of PI 3-K activity were present in all fractions from control hepatocytes. As shown, insulin treatment stimulated a 3-4-fold increase in IRS-1-associated PI 3-K activity in all fractions (Fig. 5B). Considerably more PI 3-K activity was present in LDM and PLS than in other fractions. LDM and PLS contained 8 and 31 times more IRS-1-associated PI 3-K activity, respectively, than either PMN or the HDM fraction. Results suggest that the majority of insulin-activated PI 3-K is present in LDM and PLS fractions and that LDM is a microsome fraction enriched in both apoB protein and mRNA and in insulin-stimulated PI 3-K activity.
Effect of Insulin on the Subcellular Localization of IRS-1⅐PI 3-K Complexes-Proteins in anti-IRS-1 immunoprecipitates of subcellular fractions (Fig. 5) were analyzed by SDS-PAGE and immunoblotted with antibodies to p85. Insulin stimulated an increase in the amount of p85 associated with IRS-1 in all fractions, with the largest increase in LDM (Fig. 6). Results indicate that the increase in IRS-1-associated PI 3-K activity corresponds to an increase in IRS-1⅐PI 3-K mass. Furthermore, LDM is markedly enriched in both IRS-1-associated PI 3-K activity and IRS-1⅐PI 3-K mass following insulin stimulation.
Subcellular Localization of ApoB and PI 3-K by Sucrose Density Gradient Centrifugation-To evaluate whether PI 3-K coisolates with microsomes containing apoB following insulin treatment, LDM isolated from hepatocytes with or without insulin treatment were subfractionated by sucrose density gradient centrifugation. After fractionation, microsomes from each fraction were collected by centrifugation. Microsomal proteins were solubilized in SDS buffer and were analyzed for apoB and PI 3-K by immunoblotting. Most apoB48 and apoB100 were isolated in the bottom fractions 2-6 (peak at fraction 3), with trace amounts of apoB48 in the top fraction (Fig. 7). The amount and distribution of apoB in sucrose gradient fractions derived from LDM were not significantly altered by the 5-min insulin treatment. In contrast, insulin stimulated a significant increase in the amount of p85 in all fractions. As shown, the amount of p85 increased in fractions that were also enriched in apoB (fractions 2-6). Another protein of 169 kDa that reacted with the anti-p85 antibody was also present in fractions 2-4, but the amount of this protein was not changed by insulin. Results demonstrate that insulin stimulates a significant increase in the amount of p85 present in fractions containing apoB.
Characterization of Sucrose Gradient Fractions-Microsomes from sucrose density gradient fractions in Fig. 7 were analyzed for ER and Golgi marker proteins by immunoblotting. ER marker proteins included calnexin, protein-disulfide isomerase (50,51), and NADPH-cytochrome c reductase. These proteins were shown to localize in fractions 2-6 (Fig. 8). The Golgi marker proteins Golgi 58K and Golgi ␤-COP (52, 53) were found only at the top of the gradient. Insulin did not change the FIG. 3. Distribution of apoB protein and apoB mRNA in subcellular fractions isolated from rat hepatocytes. A, proteins in HDM, LDM, and PLS were solubilized in Laemmli sample buffer, separated by SDS-PAGE, transferred to PVDF membrane, and immunoblotted with specific antibodies to apoB. A representative immunoblot (top) shows the distribution of apoB48 and apoB100 in each fraction. Films were quantitated by densitometric scanning. Total amounts of apoB48 and apoB100 in each fraction were calculated, and results are expressed as a percentage of the sum total of apoB48 and apoB100 in all fractions (mean Ϯ S.D., n ϭ 4 livers). B, total RNA was extracted from each fraction, and resuspended in an equal volume of RNase-free H 2 O. Ten microliters of RNA were slot-blotted onto nylon membranes in triplicate. Membranes were then hybridized with radiolabeled apoB cDNA probes, and hybridization signals were quantitated by Phosphor-Imager analysis. Results from two experiments are averaged and expressed as a percentage of total apoB mRNA in all fractions.

FIG. 4. Subcellular distribution of PI 3-K in rat hepatocytes.
Isolated rat hepatocytes were incubated with (ϩ) or without (Ϫ) insulin for 5 min, and then homogenized in a ball bearing tissue homogenizer. Subcellular fractions were isolated from total cell homogenate, and PI 3-K was immunoprecipitated from each fraction with antibodies to the p85 subunit of PI 3-K following detergent solubilization. Immunoprecipitates were analyzed by SDS-PAGE, electrophoretically transferred to PVDF membranes, and probed with anti-p85 antibodies. A, an immunoblot shows the subcellular distribution of p85. B, bands corresponding to p85 were quantitated by laser-scanning densitometry, and the total amount of p85 in each fraction was calculated and normalized to mg of cell protein/fraction. Results represent the mean Ϯ S.D. from four independent rat liver experiments. I.P., immunoprecipitates; I.B., immunoblot. *, p Ͻ 0.05 between control and insulin-treated conditions using Student's t test. subcellular distribution of ER and Golgi marker proteins (data not shown). Together, results indicate that ER vesicles can be effectively separated from Golgi vesicles on a sucrose density gradient. Most of the apoB-containing microsomes isolated from LDM coisolate with ER marker proteins. Thus, insulin stimulates a significant increase in the amount of p85 present in fractions also shown to contain apoB and ER marker proteins.
Electron Micrographs of Smooth and Rough Microsomes-To confirm the findings from studies of LDM that insulin increases the amount of PI 3-K in the ER and to distinguish more precisely the region of the ER where this activity is found, rough and smooth microsomes were isolated from hepatocytes according to the classical procedure described by Depierre and Dallner (34) for rat liver. Smooth microsomes were postfixed in osmium tetroxide and embedded in Spurr epoxy resin to visualize membranes. To better visualize membrane-bound ribosomes, rough microsomes were embedded in Lowicryl, a low temperature embedding resin. Electron micrographs of microsomes showed that smooth microsomes, referred to as smooth ER, consisted of a fairly heterogeneous population of smooth surfaced vesicles (Fig. 9A). Rough microsomes were consider-ably more homogeneous and consisted of ribosome-studded vesicles, indicative of rough ER (Fig. 9B).

Insulin Stimulates PI 3-K Activity in Rough and Smooth
Microsomes-Rat hepatocytes were incubated with or without insulin for 5 min prior to subcellular fractionation. Rough and smooth microsomes were isolated, followed by solubilization and immunoprecipitation with anti-IRS-1 antibodies. PI 3-K activity in anti-IRS-1 immunoprecipitates was then determined. Basal levels of IRS-1-associated PI 3-K activity were present in the total homogenate and in rough and smooth microsomes (Fig. 10). Insulin induced a 4-fold increase in IRS-1-associated PI 3-K activity in rough microsomes and a 14-fold increase in smooth microsomes. These results show that insulin stimulates PI 3-K activity in both smooth and rough microsomes and that rough microsomes correspond to rough ER as demonstrated by electron microscopy (Fig. 9B). DISCUSSION Insulin-dependent inhibition of apoB secretion by rat hepatocytes can be blocked by wortmannin, an inhibitor of PI 3-K activity, indicating that insulin action on apoB requires PI 3-K activation (22). Current studies were performed to further characterize the effect of wortmannin on apoB secretion and to extend studies using the PI 3-K inhibitor, LY 294002 with different inhibitory specificities than wortmannin. We compared the ability of wortmannin and LY 294002 to alter insulin action on apoB secretion by rat hepatocytes. Wortmannin inhibits basal and insulin-stimulated PI 3-K activities without FIG. 5. Subcellular localization of IRS-1-associated PI 3-K activity. Isolated rat hepatocytes were treated with (ϩ) or without (Ϫ) 100 nM insulin for 5 min and homogenized in a ball bearing tissue homogenizer. Subcellular fractions isolated from total cell homogenate were solubilized in detergents, and IRS-1 was immunoprecipitated by incubation with specific antibodies. After extensive washing of anti-IRS-1 immunoprecipitates, PI 3-K activity in the immunoprecipitates was assayed using phosphatidylinositol as substrate and [ 32 P]ATP. A, an autoradiograph shows the radiolabeled PI-3-P product of PI 3-K separated by thin layer chromatography. B, PI 3-K activity was quantitated by PhosphorImager scanning, and the total amount of PI 3-K activity in each fraction was calculated. Results are expressed as the mean Ϯ S.D. from three independent rat liver experiments.

FIG. 6. Subcellular localization of IRS-1⅐PI 3-K complexes.
Proteins in anti-IRS-1 immunoprecipitates from Fig. 5 were solubilized in Laemmli sample buffer, separated by SDS-PAGE, and electrophoretically transferred to PVDF membranes. PI 3-K was detected by immunoblotting with antibodies to the p85 subunit of PI 3-K. A, an autoradiograph of an immunoblot shows p85 in subcellular fractions. B, bands corresponding to p85 were quantitated by laser scanning densitometry, and the total amount of p85 in each fraction was calculated. Results shown are the mean Ϯ S.D. from four independent rat liver experiments. I.P., immunoprecipitates; I.B., immunoblot.
interfering with insulin-induced association of IRS-1 with PI 3-K. Wortmannin prevents insulin-dependent inhibition in apoB secretion in a dose-dependent manner. Results using wortmannin were confirmed with studies using LY 294002, which specifically inhibits PI 3-K activity, but not the activities of a number of other protein and lipid kinases including protein kinase C, protein kinase A, S6 kinase, mitogen-activated protein kinase, and phosphatidylinositol 4-kinase (41). Together, results indicate that activation of PI 3-K by insulin is necessary in the regulation of apoB secretion by insulin.
PI 3-K activation alone may be necessary, but not sufficient to mediate insulin-dependent inhibition of apoB secretion by rat hepatocytes. Both insulin and epidermal growth factor activate PI 3-K (10,54). However, only insulin is able to inhibit apoB secretion (29). Recent studies suggest that both activation of PI 3-K and localization of the active enzyme to a specific subcellular compartment are required to selectively elicit metabolic responses to insulin (23)(24)(25)(26). PI 3-K activation is necessary for insulin-induced translocation of GLUT 4 transporters to the plasma membrane (19,20), and PI 3-K has been shown to localize in vesicles containing GLUT 4 transporters (27). In the absence of growth factor stimulation, localization of PI 3-K to the membrane alone is sufficient to induce PI 3-K-dependent responses, including activation of the downstream effectors p70 ribosomal protein S6 kinase, protein kinase B (also called Akt), and Jun N-terminal kinase (55).
For PI 3-K to mediate insulin action on apoB secretion, we hypothesize that PI 3-K must extend its effects to the secretory compartment. This hypothesis is the basis for further studies to identify insulin-dependent localization of PI 3-K to the ER, where apoB biogenesis is initiated. We chose a rapid subcellular fractionation technique (30,43) to test this hypothesis because fractions can be isolated without apparent dephosphorylation of tyrosine-phosphorylated proteins or loss of PI 3-K activity. In addition, using this method, we have demonstrated intact apoB protein (Fig. 3) and apoB mRNA (data not shown). Studies show that insulin induces a significant redistribution of PI 3-K to an LDM fraction containing apoB protein and apoB mRNA. Insulin stimulates a significant increase in PI 3-K activity associated with IRS-1 as well as an increase in IRS-1⅐PI 3-K mass in LDM. Subfractionation of LDM on sucrose density gradients indicates that insulin action significantly increases the amount of PI 3-K present in an ER fraction containing apoB. To distinguish more precisely the region of the ER where this activity is found, insulin-stimulated PI 3-K activity was measured in smooth and rough microsomes isolated from hepatocytes according to a classical procedure (34). Insulin stimulates PI 3-K activity in both smooth and rough microsomes, the latter of which contain rough ER as demonstrated by electron microscopy. Together, results indicate that insulin stimulates PI 3-K activity in the rough ER and that insulin action leads to the activation and localization of PI 3-K in the ER compartment that is also shown to contain apoB.
ApoB is an unique secretory protein in that, unlike other secretory proteins such as albumin, apoB secretion is regulated by insulin (1,56). ApoB is synthesized in the ER and assembles with lipids before its secretion as a component of VLDL. Although additional action of PI 3-K in later vesicular trafficking cannot be ruled out, there appears to be a requirement for PI 3-K in targeting of apoB for degradation in an early stage of assembly. This is consistent with studies indicating that insulin interferes with the assembly of VLDL as denser apoBcontaining lipoproteins continue to be secreted in the presence of insulin (57). The observation that insulin increases the amount of PI 3-K present in an ER compartment containing apoB suggests that localization of PI 3-K may be important in FIG. 7. Distribution of apoB and PI 3-K in sucrose density gradients. Isolated rat hepatocytes were incubated with (ϩ) or without (Ϫ) insulin for 5 min and then homogenized in a ball bearing tissue homogenizer. Low density microsomes were isolated from the total homogenate and then applied to the top of a 0.4 -1.5 M sucrose gradient. After centrifugation at 150,000 ϫ g for 3 h, 12 fractions were collected, and samples were recentrifuged to pellet microsomes. Microsomes were then solubilized in Laemmli sample buffer, separated by SDS-PAGE, transferred to PVDF membranes, and immunoblotted with anti-apoB and anti-p85 antibodies.  Fig. 7 were analyzed by immunoblotting with anti-calnexin, anti-protein-disulfide isomerase (PDI), anti-Golgi 58K, and anti-Golgi ␤-COP antibodies. NADPH-cytochrome c reductase specific activity, protein concentrations, and the density of each fraction were also measured as described under "Experimental Procedures." insulin action on apoB. Current studies indicate that IRS-1 plays a role in the localization of p85, since increases in p85 mass and PI 3-K activity were demonstrated following IRS-1 immunoprecipitations. Our data cannot rule out additional roles of other IRS-like molecules in PI 3-K-targeting events. Further studies will be necessary to delineate the kinetics of movement of IRS-1, PI 3-K, and other IRS molecules to the ER following insulin stimulation.
Characterization of subcellular fractions from rat hepatocytes based on organelle marker proteins shows that most ER marker proteins are present in HDM and LDM, and most Golgi marker proteins are in PLS (Fig. 2). Assuming that the distribution of organelle markers correctly reflects the distribution of ER-and Golgi-derived membranes, the results indicate that HDM and LDM are enriched in ER membranes, and PLS is enriched in Golgi membranes. However, using a similar fractionation method in rat adipocytes, other laboratories reported that HDM is enriched in ER membranes, while LDM is enriched in Golgi membranes, with significant cross-contamina-tion of ER and Golgi marker enzymes between HDM and LDM (30,58). Disparities in organelle marker distribution observed in our studies and those of others may reflect differences in the cellular composition of hepatocytes and adipocytes and in fractionation of whole tissue versus isolated cells. It is also possible that homogenization of cells in the absence of RNase inhibitors may alter characteristics of the subcellular fractions obtained.
Immunoblot analysis of apoB distribution in subcellular fractions show that, on average, 95% of cellular apoB48 and 91% of cellular apoB100 are recovered in HDM plus LDM (Fig. 3). Thus, most cellular apoB isolated by this fractionation procedure is located in ER-enriched fractions. This finding is consistent with results previously reported in cultured hepatocytes and in liver, where 60 -90% of apoB is localized in the ER (59 -61). Most apoB mRNA is present in HDM and LDM, with 70% of apoB mRNA found in LDM. The finding of most apoB mRNA in LDM may relate to the aberrant sedimentation velocity of apoB mRNA, since hepatic apoB polysomes sediment just after monosomes and not with polysomes (62). The aberrant sedimentation velocity is observed in apoB polysomes isolated in the presence and absence of detergents, indicating that apoB polysomes exhibit unusual sedimentation characteristics whether or not they are membrane-associated.
Several studies have reported the effects of insulin and other growth factors on the subcellular distribution of PI 3-K in adipocytes (25,26,43,63). However, the subcellular distribution of PI 3-K in hepatocytes has not been studied. Current studies show that insulin induces the redistribution of p85 from PLS to LDM (Fig. 4). This finding is consistent with results from studies of adipocytes in which insulin increases the amount of p85 in microsomes (25,26). Although it has been proposed that tyrosine-phosphorylated IRS-1 may deliver PI 3-K to specific intracellular sites (64), little is known about the mechanism(s) by which PI 3-K is recruited to intracellular membranes. Since there is evidence that inhibition of PI 3-K activity by wortmannin does not block insulin-stimulated localization of PI 3-K to microsomes (25), it is unlikely that PI 3-K activity is necessary for the recruitment of PI 3-K to intracellular membranes. IRS-1-associated PI 3-K activity appears to be confined to LDM and PLS fractions from rat hepatocytes treated with insulin (Fig. 5). These results are consistent with recent studies in adipocytes demonstrating that most insulinstimulated PI 3-K activity is present in LDM (25,43,63) or in LDM and cytosol (26). Whether the redistribution of p85 from the cytosol to the LDM is the sole consequence of interaction FIG. 9. Electron micrographs of smooth and rough microsomes. A, smooth microsomes isolated from rat hepatocytes were fixed in 2.5% gluteraldehyde, postfixed in osmium tetroxide, and embedded in Spurr epoxy resin. B, rough microsomes were fixed in 2% paraformaldehyde and embedded in Lowicryl resin. 17,000 ϫ magnification.
FIG. 10. Distribution of IRS-1-associated PI 3-K activity in smooth and rough microsomes. Isolated rat hepatocytes were incubated with (ϩ) or without (Ϫ) insulin (100 nM) for 5 min and then homogenized in a ball bearing tissue homogenizer. Rough and smooth microsomes isolated from total cell homogenates were solubilized in detergents, and IRS-1 was immunoprecipitated from each fraction by incubation with specific antibodies. After extensive washing of anti-IRS-1 immunoprecipitates, PI 3-K activity in the immunoprecipitates was determined. An autoradiograph shows the radiolabeled lipid product of PI 3-K separated by thin layer chromatography with position of PI-3-P indicated. SM, smooth microsomes; RM, rough microsomes. with activated IRS-1 cannot be evaluated from the current studies. Such an analysis would have to include time course studies and evaluation of other potential insulin receptor substrate molecules.
Insulin has been shown to stimulate PI 3-K activity in plasma membranes, in a low density microsome fraction enriched in GLUT 4-containing vesicles, and possibly in GLUT 4-containing vesicles themselves (25-27, 43, 63). A novel finding of the present study is the presence of insulin-activated PI 3-K in the rough ER. Several lines of evidence support this conclusion. First, the LDM fraction, which is enriched in both IRS-1-associated PI 3-K activity and IRS-1⅐PI 3-K mass in response to insulin, contains significant amounts of ER marker proteins and only trace amounts of Golgi marker proteins (Fig.  2). Second, the majority of apoB mRNA is isolated in LDM, indicating that at least a portion of microsomes in the LDM fraction are rough microsomes containing apoB mRNA (Fig. 3). Third, subfractionation of LDM on sucrose density gradients shows that vesicles derived from the ER are effectively separated from Golgi vesicles based on distribution of organelle marker proteins (Fig. 8). Upon insulin stimulation, the amount of p85 (PI 3-K) increases significantly in microsome fractions containing calnexin, NADPH-cytochrome c reductase, and protein-disulfide isomerase, consistent with rough ER (Fig. 7). Finally, insulin stimulates PI 3-K activity in classically isolated rough microsomes (Fig. 10), shown to be composed of rough ER by electron microscopy (Fig. 9).
In summary, current studies suggest that insulin action leads to the localization of activated PI 3-K to the ER, which correlates with the site of apoB biogenesis and lipoprotein assembly. Mechanisms involving the inhibition of neutral lipid assembly into VLDL or the targeting of newly synthesized apoB from the ER to a degradation compartment are possible. Consistent with the latter hypothesis are the observations that insulin-induced degradation of apoB in rat hepatocytes occurs in a post-ER compartment and involves a vesicular traffic event sensitive to brefeldin A (22,65). A role of PI 3-K in vesicular traffic is further supported, since PI 3-K is required for the traffic of GLUT 4 transporters in adipocytes (19,20) and for the sorting and transport of vacuolar proteins in yeast (66) and of lysosomal enzymes in mammalian cells (17,18).
A physiological role for the inhibition of hepatic apoB secretion by insulin may be to reduce hepatic VLDL secretion and to limit VLDL competition with intestinal lipoproteins for common catabolic pathways in the fed state. We have shown that the inhibitory effect of insulin on hepatic apoB secretion is lost in insulin-resistant obese Zucker rats (67), and there is a corresponding increase in hepatic secretion of triglyceride-rich lipoprotein particles in these animals (68,69). The loss of the insulin effect on apoB could increase competition between hepatic and intestinal triglyceride-rich lipoproteins in the fed state and may result in prolongation of postprandial hypertriglyceridemia. To understand the pathophysiology of insulin resistance and concomitant hypertriglyceridemia, future studies are focused on determining whether localization of activated PI 3-K to an apoB-specific subcellular compartment occurs in animal models of insulin resistance and diabetes.