Endocrinology and Metabolism

Role of the liver in glucose homeostasis in PI 3-kinase p85α-deficient mice

Kazutaka Aoki, Junji Matsui, Naoto Kubota, Hiromu Nakajima, Keiji Iwamoto, Iseki Takamoto, Youki Tsuji, Akira Ohno, Shuuichi Mori, Kumpei Tokuyama, Koji Murakami, Tomoichiro Asano, Shinichi Aizawa, Kazuyuki Tobe, Takashi Kadowaki, Yasuo Terauchi

Abstract

Phosphoinositide 3-kinase (PI3K) p85α-deficient mice exhibit hypoglycemia as a result of increased insulin sensitivity and glucose uptake in peripheral tissues. Although PI3K is central to the metabolic actions of insulin, its mechanism of action in liver is not well understood. In the present study, we investigated hepatic insulin signaling and glucose homeostasis in p85α-deficient and wild-type mice. In the livers of p85α-deficient mice, p50α played a compensatory role in insulin-stimulated PI3K activation by binding to insulin receptor substrate (IRS)-1/2. In p85α-deficient mice, the ratio of p50α over p110 catalytic subunit of PI3K in the liver was higher than in the muscles. PI3K activity associated with IRS-1/2 was not affected by the lack of p85α in the liver. Insulin-stimulated Akt and phosphatase and tensin homologue deleted on chromosome 10 (PTEN) activities in the liver were similar in p85α-deficient and wild-type mice. A hyperinsulinemic-euglycemic clamp study revealed that the glucose infusion rate and the rate of disappearance were higher in p85α-deficient mice than in wild-type mice but that endogenous glucose production tended to be higher in p85α-deficient mice than in wild-type mice. Consistent with this finding, the expression of glucose-6-phosphatase and phosphoenolpyruvate carboxykinase in livers after fasting was higher in p85α-deficient mice than in wild-type mice. After mice were fasted, the intrahepatic glucose-6-phosphate level was almost completely depleted in p85α-deficient mice. The glycogen content fell to nearly zero as a result of glycogenolysis shortly after the initiation of fasting in p85α-deficient mice. The absence of an increase in insulin-stimulated PI3K activation in the liver of p85α-deficient mice, unlike the muscles, may be associated with the molecular balance between the regulatory subunit and the catalytic subunit of PI3K. Gluconeogenesis was rather elevated in p85α-deficient mice, compared with in wild-type mice, and the liver seemed to partially compensate for the increase in glucose uptake in peripheral tissues.

  • phosphoinositide 3-kinase
  • regulatory subunit
  • catalytic subunit

defects in insulin secretion from pancreatic β-cells and insulin resistance in the target tissues interact in a complex manner to disturb glucose homeostasis and cause type 2 diabetes (39, 40, 42, 47). Insulin activates phosphoinositide 3-kinase (PI3K) via the tyrosine phosphorylation of insulin receptor substrates (IRSs) and the subsequent binding of p85α associated with p110 (17, 18, 24, 28, 33, 34). Previous in vitro experiments (13, 32) have suggested that the activation of PI3K plays an important role in the metabolic actions of insulin, like glucose transporter (GLUT) translocation and glycogen synthase activation. To investigate the role of PI3K in glucose metabolism in vivo, we specifically deleted the first exon of Pik3r1 in mice (exon 1A). Because this exon contains the initiation codon for p85α, we were able to selectively abolish the expression of full-length p85α mRNA without disrupting the p55α (2, 15) and p50α (8, 16) splicing variants. Mice deficient in p85α were born and showed no apparent growth abnormalities, presumably due to redundant PI3K activities (43). By contrast, the absence of all three isoforms of the p85α gene is reportedly a fatal condition during the perinatal period (9).

Mice deficient in p85α were hypoglycemic because of an increase in glucose uptake in peripheral tissues (43). The phenotype of p85α-deficient mice can be explained by an increase in the insulin-induced generation of phosphatidylinositol 3,4,5-triphosphate (PIP3) in association with an isoform switch from p85α PI3K to p50α PI3K in peripheral tissues. It should be noted, however, that the targeted disruption of p50α and p55α PI3K also led to increased insulin sensitivity (5). Recently, Taniguchi et al. (38) reported that mice with a liver-specific deletion of the p85α regulatory subunit exhibited a paradoxical improvement in hepatic and peripheral insulin sensitivity but that liver-specific deletions of both the p85α and the p85β regulatory subunits led to an increase in gluconeogenesis in association with the impairment of PI3K activation (37). In this context, Kahn and colleagues reported that the heterozygous disruption of the Pik3r1 gene improved insulin signaling in liver and muscle (25) and hypothesized that optimal signaling through the PI3K pathway depended on a critical molecular balance between the regulatory and catalytic subunits (45). This hypothesis invokes the existence of non-p110-bound p85 (“free p85”), which can compete with the heterodimeric p85/p110, thereby dampening PI3K signaling. In this context, Vanhaesebroeck and colleagues (12) argued against the free p85 hypothesis. Moreover, it has been reported that PTEN (phosphatase and tensin homologue deleted on chromosome 10) activity is decreased in p85α-null liver (38), raising the possibility that loss of expression of p85 affects lipid phosphatases. Thus although PI3K is central to the metabolic actions of insulin, its mechanism of action is not well understood. We therefore investigated the role of the liver in glucose homeostasis in p85α-deficient mice and the impact of molecular balance between the regulatory and catalytic subunits of PI3K on downstream signaling through the PI3K pathway.

MATERIALS AND METHODS

Animals and genotyping.

Mice lacking p85α (C57Bl/6J and CBA mixed background) were generated as previously described (43). These mice were backcrossed with C57Bl/6J mice (CLEA Japan) or BALB/c mice (Taconic Farm; Ref. 11). Because the murine genetic background was not completely homogeneous, male offspring derived from p85α+/− intercrosses were analyzed in this study. For the immunoblotting experiments, the PI3K activity assays, the Akt and PTEN assays, the hyperinsulinemic-euglycemic glucose clump study, the gluconeogenic activity in vivo, and the TaqMan PCR analysis, the male offspring derived from p85α+/− intercrosses that were bred on a C57Bl/6J and BALB/c mixed background were used. For other experiments, mice on a C57Bl/6J background were used. Our experimental procedures were approved by the Institutional Ethics Committee of Yokohama City University, and then the experiments were performed in accordance with the guidelines of the Animal Care Committee of Yokohama City University. The mice were fed water and normal laboratory chow ad libitum and maintained using standard animal husbandry procedures. All mice were kept on a 12:12-h light-dark cycle.

The Pik3r1 genotype was determined using PCR. Genomic DNA was extracted from the tip of the tail. The sense primer was PI-6 (5′-CAGATGGACAGTGTGACAGG-3′), and the antisense primers were PI-9 (5′-AGGGGGTGAAATTCTTTTCC-3′) for the Pik3r1 gene and Neo-1 (5′-CCAGTCATAGCCGAATAGCC-3′) for a neomycin-resistance gene. The three primers and the genomic DNA template were mixed in a tube. The wild-type allele produced a 600-bp product, and the recombinant allele produced a 450-bp product.

Antibodies.

Anti-p85 polyclonal antibody against a full-length p85-GST fusion protein containing the NH2-terminal SH2 domain of p85α (anti-p85PAN) and anti-p110α antibody against the COOH-terminal region (aa 1,054–1,068; anti-p110α) were purchased from Upstate Biotechnology (Lake Placid, NY). Specific antibodies against p50α (anti-p50α), p55α (anti-p55α), p55γ (anti-p55γ), and p85β (anti-p85β) were generated as previously described (8, 16). The monoclonal anti-phosphotyrosine antibody (anti-PY), polyclonal anti-IRS-1 antibody (anti-IRS-1), and polyclonal anti-IRS-2 antibody (anti-IRS-2) were purchased from Upstate Biotechnology rabbit polyclonal anti-phospho-AKT antibody (anti-pAKT) recognizing phosphorylated Ser-473 of Akt1 and rabbit anti-Akt antibody (anti-AKT) were purchased from Cell Signaling Technology (Beverly, MA).

Immunoprecipitations and immunoblotting.

The livers and muscles were excised and homogenized in ice-cold buffer A (25 mM Tris·HCl pH 7.4, 10 mM Na3VO4, 10 mM NaPPi, 100 mM NaF, 10 mM EDTA, 10 mM EGTA, and 1 mM PMSF). Lysates were prepared using centrifugation (15,000 rpm, 20 min, 4°C). Lysates containing equal amounts of total protein (∼100 μg) were incubated with the indicated antibody for 1 h at 4°C and then with protein G-Sepharose for 1 h at 4°C. The beads were washed three times with buffer A containing 1% Triton X-100, and the immunoprecipitated proteins were solubilized with Laemmli's sample buffer. Samples were separated on SDS-polyacrylamide gels and transferred to nitrocellulose followed by immunoblotting with the indicated antibody. The blots were incubated with horseradish peroxidase-linked protein A, and the bands were detected using enhanced chemiluminescence (Amersham International, England, UK).

PI3K assay.

PI3K activity in the liver was determined in immunoprecipitates using the indicated antibodies after insulin injection into the inferior vena cava (22, 48). PI3K was immunoprecipitated with the indicated antibody, and the immunoprecipitates were washed three times with buffer A containing 1% Triton X-100 and then three more times with PI3K reaction buffer (20 mM Tris-HCl pH 7.4, 100 mM NaCl, 1 mM Na3VO4, and 0.5 mM EGTA). The reaction was initiated by addition of 50 μl of PI3K reaction buffer containing 20 mM MgCl2, 20 μM ATP, 5 μCi of [γ-32P]ATP, and 0.2 mg/ml of phosphatidylinositol to the immunoprecipitates. After incubation at 25°C for 20 min, the reaction was terminated by the addition of 100 μl of chloroform containing HCl and the organic phase was separated by centrifugation and washed three times with methanol-1 M HCl (1:1). The lipids were spotted onto a Silicagel 60 plate (Merck, Darmstadt, Germany) and developed in chloroform, methanol, 28% ammonium hydroxide, and water (43:38:5:7). The phosphorylated lipids were visualized and evaluated using a BAS 2000 system (Fuji Film, Kanagawa, Japan).

Changes in expression of regulatory subunits of PI3K and PI3K activity associated with IRS-1 and IRS-2.

Four-to-five-month-old male mice were subjected to fasting for 24 h or were refed for 6 h after a 24-h period of starvation. Hepatic lysates were incubated with antibody against IRS-1 or IRS-2 and then with protein G-Sepharose. Immunoblot analyses were then performed as described above. PI3K activity was measured after immunoprecipitation with IRS-1(n = 6) or IRS-2 (n = 5) antibodies using PI3K assay kit (Jena Biosciences, Jena, Germany). The radioactivity of sample was measured in the liquid scintillation counter.

In vivo insulin stimulation and analyses of Akt, pAkt, and PTEN.

Four-month-old male mice were starved for 24 h, anesthetized with pentobarbital, and injected with 5 U of regular human insulin (Humalin R; Lilly) or saline into the inferior vena cava. Five minutes later, the livers were removed and homogenized in ice-cold buffer A. Immunoblot analyses for Akt and pAkt were then performed. Akt activity was expressed as the ratio of the intensity of pAkt to Akt.

To analyze PTEN activity, the livers were homogenized in ice-cold buffer (20 mM imidazol-HCl, 2 mM EDTA, 2 mM EGTA pH 7.0, containing 1 mM benzamidine, 0.8 μM aprotinin, 50 μM vestatin, 15 μM E-64, 20 μM leupeptin, and 10 μM pepstatin A) and measured using the PTEN malachite green assay kit (Upstate Biotechnology, Lake Placid, NY). PTEN activity was expressed as the activity per 1 mg protein of liver lysate.

Hyperinsulinemic-euglycemic clamp study.

The clamp studies were carried out as described previously (20, 36), with slight modifications. Studies were performed using 4-to-5-mo-old male mice under conscious and unstressed conditions after a 6-h fast. A primed continuous infusion of insulin (Humalin R; Lilly) was given (5.0 mU·kg−1·min−1), and the blood glucose concentration, which was monitored every 5 min, was maintained at ∼120 mg/dl through the administration of glucose (5 g of glucose per 10 ml, enriched to ∼20% with d-glucose-6,6-d2; Isotec) for 120 min. Blood was sampled via tail tip bleeds at 90, 105, and 120 min to determine the rate of glucose disappearance (Rd). Rd was calculated using nonsteady-state equations (36), and endogenous glucose production (EGP) was calculated as the difference between Rd and the exogenous glucose infusion rate (GIR; Ref. 36).

Gluconeogenic activity in vivo.

Mice were subjected to fasting for 24 h or were refed for 6 h after a 24-h period of starvation. Gluconeogenic activity was measured using a previously described method (3, 10, 31). NaH[14C]CO3 (20 μCi/10 g body wt) was intravenously injected via the tail vein. Blood samples (100 μl) were then obtained at 20 min after NaH[14C]CO3 administration. The collected blood was hemolyzed in 1.2 m of distilled water and then deproteinized through the addition of 0.1 ml of 0.3 M Ba(OH)2 and 0.1 ml of 0.3 M ZnSO4, followed by centrifugation at 9,000 rpm for 5 min. To evaluate the total infusion quantity, the radioisotopes in 0.1 ml of the supernatant were counted using a liquid scintillation counter and PICO-FLUOR 40 solvent (Packard Bio Science). Each preparation of the supernatant was divided into two tubes, each containing 0.47 ml, and then 10 μl of 0.5 M adenosine triphosphate and 20 μl of 8.4% bicarbonate were added to each tube. Five units of hexokinase (Wako Pure Chemical Industries) were added to one of each pair of tubes. After 30 min of incubation at 37°C, 450 μl of supernatant with or without hexokinase were placed into a tube containing anion-exchange resin (AG-8X, form, 200–400 mesh; Bio-Rad, Hercules, CA) overnight at 37°C. After centrifugation at 6,000 rpm for 10 min, the radioactivity of each sample was measured using a liquid scintillation counter. The gluconeogenic activity was expressed as the ratio of [14C]glucose, which is the difference between the radioactivity in the sample with and that in the sample without hexokinase compared with the total infusion radioactivity.

TaqMan PCR.

Four-month-old male mice were subjected to fasting for 24 h or were refed for 6 h after a 24-h period of starvation. Total RNA was prepared from portions of the liver using Isogen reagent (NipponGene, Tokyo) according to the manufacturer's instructions. The mRNA levels in the liver were quantitatively analyzed using fluorescence-based reverse transcriptase-PCR. The reverse transcription mixture was amplified using specific primers and an ABI Prism 7000 sequence detector equipped with a thermocycler. The primers used for glucose-6-phosphatase (G6Pase), phosphoenolpyruvate carboxykinase (PEPCK), and glucokinase (GK) were purchased from Applied Biosystems (Foster City, CA). The primers used for β-actin were described previously (21). The relative expression levels were compared after normalization to β-actin.

Measurement of serum parameters.

Glucose levels were measured using a glucose test sensor (SKK, Nagoya, Japan). Insulin levels were determined using an insulin kit (BioTrak, Amersham Life Science) with rat insulin as the standard. To determine the lactate and pyruvate levels, trunk blood was extracted with an equivalent amount of perchloric acid (6%). Serum lactate, pyruvate, total cholesterol, and triglyceride levels were determined using Determiner-LA, -PA, -TC, and -TG kits, respectively (Kyowa Medex, Japan). Serum glycerol levels were determined using the Determiner-TG kit. Serum free fatty acid levels were determined using a commercial kit (Wako Chemicals, Osaka, Japan). Serum levels of amino acids, like alanine and glutamine, were measured using enzyme assays and HPLC. Plasma cAMP levels were determined using a commercial kit (Mikasa).

Measurement of gluconeogenic and glycolytic intermediates in the liver.

Portions of the liver were removed from freely fed or 24-h-fasted male mice and were immediately frozen in liquid nitrogen. The levels of glucose metabolites were then determined as described previously (4).

Measurement of glycogen content.

After portions of the liver had been lysed with 30% KOH and precipitated with ethanol, the glycogen content was measured using Anthron/H2SO4.

Statistical analysis.

Results were expressed as the means ± SE. Statistical differences were analyzed using the Student t-test for unpaired comparisons. A Tukey-Kramer test was used for comparisons among four groups of mice. A P value < 0.05 was considered statistically significant.

RESULTS

Isoform switch from p85α to p50α in insulin-stimulated activation of PI3K.

We first investigated PI3K activities in hepatic lysates stimulated with insulin and immunoprecipitated with anti-phosphotyrosine antibody (anti-PY) or anti-IRS-1 antibody (anti-IRS-1). Despite the complete abrogation of the p85α molecule, hepatic PI3K activity in the anti-PY immunoprecipitates was normal (Fig. 1A), and similar results were obtained with anti-IRS-1 (data not shown). We next investigated the expression levels of hepatic PI3K regulatory subunit isoforms (Fig. 1B). Lysates were immunoprecipitated with a panel of antibodies against either the NH2-terminal SH2 domain of p85 (anti-p85PAN), which can recognize p85α as well as p50α, p55α, p55γ, and p85β; the p110α catalytic subunit of PI3K, or p50α, followed by blotting with anti-p85PAN, or p50α. In wild-type mice, p85α was the major PI3K regulatory subunit isoform (Fig. 1B, lane 1) that bound to the p110α catalytic subunit (Fig. 1B, lane 3). In p85α-deficient mice, however, p50α was the major PI3K regulatory subunit isoform (Fig. 1B, lanes 2, 4, and 5) that bound to the p110α catalytic subunit (Fig. 1B, lane 4). The expression level of p50α was significantly higher in livers from p85α-deficient mice than that from wild-type mice (Fig. 1B, lane 1 vs. 2; see Fig. 2A for quantitative analysis). Hepatic p55γ was not detected in either mouse type when blotted with a specific antibody against p55γ (data not shown). The expression of p85β did not differ between wild-type and p85α-deficient mice (data not shown), and p85β association with p110 was weaker than the association between p50α and p110, which seemed to be predominant (Fig. 1B, lane 4). Since IRS-1/2 protein is tyrosine-phosphorylated in response to insulin stimulation, we determined which regulatory subunits were bound to IRS proteins in vivo in response to insulin. IRS-1 was tyrosine phosphorylated, to similar extents, in both wild-type and p85α-deficient mice (Fig. 1C, top). The lysates were immunoprecipitated with anti-IRS-1 and blotted with anti-p85PAN. In wild-type mice, p85α was bound to IRS-1 in an insulin-dependent fashion in these tissues. In p85α-deficient mice, p50α was the major protein recognized by anti-p85PAN bound to IRS-1 in an insulin-dependent fashion (Fig. 1C, middle and bottom). IRS-2 was also tyrosine-phosphorylated, to similar extents, in both wild-type and p85α-deficient mice (Fig. 1D, top). The lysates were immunoprecipitated with anti-IRS-2 and blotted with anti-p85PAN. In wild-type mice, p85α was bound to IRS-2 in an insulin-dependent fashion in these tissues. In p85α-deficient mice, p50α was the major protein recognized by anti-p85PAN bound to IRS-2 in an insulin-dependent fashion (Fig. 1D, middle and bottom). When the immunoprecipitation and immunoblotting were carried out in a reverse manner, IRS-1 was found to be the tyrosine-phosphorylated protein bound to p85α in wild-type mice and to p50α in p85α-deficient mice after insulin stimulation of the liver (Fig. 1E). IRS-2 was also associated with p85α in wild-type mice and p50α in p85α-deficient mice after insulin stimulation of the liver (Fig. 1E). The binding of p50α to IRS-1/2 was stronger in p85α-deficient mice than in wild-type mice (Fig. 1E, bottom). We thus concluded that p85α in wild-type mice and p50α in p85α-deficient mice play a major role in insulin-stimulated PI3K activation via binding to the IRS family of proteins in the liver.

Fig. 1.

Switch in phosphoinositide 3-kinase (PI3K) regulatory subunit isoform from p85α to p50α in the liver. A: PI3K activities associated with tyrosine-phosphorylated proteins. PI3K activities in the lysates were determined using anti-phosphotyrosine (anti-PY) immunoprecipitates. In each experiment, the PI3K activity was determined relative to that measured in wild-type mice under basal conditions. Results are means ± SE of 6 experiments in wild-type (WT) or p85α-deficient (p85α−/−) mice. B: expression of regulatory subunits of PI3K. Lysates from the liver were directly immunoprecipitated (IP) with anti-p85PAN, anti-p110α, or anti-p50α and blotted with anti-p85PAN (top) or anti-p50α (bottom). Typical images are shown. C: PI3K regulatory subunits bound to insulin receptor substrate (IRS)-1. Lysates from liver with or without insulin stimulation were immunoprecipitated with anti-IRS-1 and blotted with anti-PY (top), anti-p85PAN (middle), or anti-p50α (bottom). Typical images of more than 3 experiments are shown. In some knockout animals, another 85-kD protein, presumably p85β, was bound to IRS-1, although the relative amount of this protein was much lower than that of p50α. We also noted a 48-kDa protein recognized by anti-p85PAN that bound to IRS-1, although the identity of this protein is presently unknown. D: PI3K regulatory subunits bound to IRS-2. Lysates from liver with or without insulin stimulation were immunoprecipitated with anti-IRS-2 and blotted with anti-PY (top), anti-p85PAN (middle), or anti-p50α (bottom). Typical images of more than 3 experiments are shown. We also noted a 48-kDs protein recognized by anti-p85PAN that bound to IRS-2, although the identity of this protein is presently unknown. E: IRS-1/2 bound to PI3K regulatory subunits. Lysates from liver with or without insulin stimulation were immunoprecipitated with anti-p85PAN (top) or anti-p50α (bottom) and blotted with anti-IRS-1 or anti-IRS-2. Typical images of >3 experiments are shown.

Fig. 2.

Expression of regulatory and catalytic subunit of PI3K in the liver and muscle. A: expression of regulatory subunits of PI3K. Same amount of lysates (50 μg) from liver and muscle were immunoblotted with anti-p85PAN. Expression level of p50α was quantified. B: expression of p110α. Lysates from the liver and muscle were directly immunoprecipitated with anti-p110α and then immunoblotted with anti-p110α. Each blot was quantified and shown in figures. Values are means ± SE. *P < 0.05.

We (43) previously reported that the expression of p50α was much lower than that of p85α in muscles from wild-type mice. We now performed a direct comparison of the regulatory and catalytic subunits of PI3K in liver and muscles. In wild-type mice, the expression of p50α in liver was much higher than in the muscles (Fig. 2A). The expression level of p50α in the liver of p85α-deficient mice was significantly higher than that of wild-type mice (Fig. 2A). Interestingly, the expression level of p110α in the liver of p85α-deficient mice was lower than that of wild-type mice (Fig. 2B). The expression level of p110α in the muscles of p85α-deficient mice tended to be lower than that of wild-type mice, although the difference was not statistically significant (Fig. 2B). Therefore, in p85α-deficient mice, the ratio of p50α over p110 in liver was higher than that in muscles, suggesting that the molecular balance between the regulatory subunit and the catalytic subunit of PI3K was different between the two tissues.

Expression of regulatory subunits of PI3K and PI3K activity associated with IRS-1 and IRS-2 during fasting and feeding.

We (19) recently proposed the concept of the existence of a dynamic relay between IRS-1 and IRS-2 in hepatic insulin signaling during fasting and feeding. We therefore examined the expression of IRS-1/2 and the regulatory subunit of PI3K and PI3K activity associated with IRS-1/2 during fasting and feeding (6 h). In wild-type mice, the expression of p85α associated with IRS-1 was not different under fasted or refed conditions (Fig. 3A). By contrast, the expression of p85α associated with IRS-2 under fasted conditions was decreased compared with refed conditions (Fig. 3B). Under refed conditions, the expression of p50α associated with IRS-1/2 was increased in p85α-deficient mice compared with wild-type mice (Fig. 3, A and B). PI3K activities associated with IRS-1 tended to be increased under refed conditions compared with fasted conditions, although the difference was not statistically significant (Fig. 3C). No significant differences in PI3K activities were observed under either fasted or refed conditions after immunoprecipitating with IRS-2 antibodies (Fig. 3D). These results were essentially consistent with our previous study (19), and IRS-1/2 associated PI3K activity was not affected by the lack of p85α in the liver. The expression of IRS-1 and IRS-2 was increased in p85α-deficient mice under fasted conditions, compared with wild-type mice (data not shown).

Fig. 3.

Changes in expression of regulatory subunits of PI3K and PI3K activity associated with IRS-1 and IRS-2 during fasting and refeeding. Expression of p85α and p50α associated with IRS-1 (A) or IRS-2 (B). Mice were subjected to fasting for 24 h or were refed for 6 h after a 24-h period of starvation. Expression of p85α in the lysates were determined using anti-IRS-1 (n = 4) or IRS-2 (n = 4) immunoprecipitates. PI3K activity associated with IRS-1 (C) or IRS-2 (D). PI3K activity in lysates was determined using anti-IRS-1 (n = 6) or IRS-2 (n = 5) immunoprecipitates. Results are means ± SE in WT or p85α-deficient (p85α−/−) mice. *P < 0.05.

Akt and PTEN activities.

No significant difference in insulin-stimulated Akt activity, which was measured according to Ser-473 phosphorylation, was observed between wild-type and p85α-deficient mice (Fig. 4A). Furthermore, no significant difference in PTEN activity was observed between wild-type and p85α-deficient mice (Fig. 4B).

Fig. 4.

Akt and PTEN activities in WT and p85α-deficient mice (p85α−/−). A: Ser-473 phosphorylation of Akt (pAkt) and Akt were measured using western immunoblotting in WT (n = 6) and p85α-deficient mice (n = 6). Typical images are shown (top). Akt activities are expressed as the raito of pAkt to Akt (bottom). B: the phosphatase and tensin homologue deleted on chromosome 10 (PTEN) activities were measured using liver homogenates from WT (n = 6) and p85α-deficient mice (n = 6). Values are means ± SE.

Hyperinsulinemic-euglycemic clamp study.

We determined the metabolic response to high concentration of insulin. The GIR and Rd were significantly higher in p85α-deficient mice than in wild-type mice (Fig. 5, A and C). These results agreed with previous reports (43) that p85α-deficient mice exhibit increased insulin sensitivity in muscles and adipose tissue. It should be noted, however, that EGP tended to be higher in the p85α-deficient mice (by 27%) than in wild-type mice, although the difference was not statistically significant (Fig. 5B). These findings suggest that the insulin response was not improved in the livers of p85α-deficient mice.

Fig. 5.

Hyperinsulinemic-euglycemic clamp analysis in WT and p85α-deficient mice (p85α−/−). Glucose infusion rate (GIR; A), endogenous glucose production (EGP; B), and rates of rate of glucose disappearance (Rd; C) in WT (n = 9) and p85α-deficient mice (n = 9). Values are means ± SE. *P < 0.05.

Gluconeogenic activity in vivo.

Figure 6 shows the total gluconeogenic activity in wild-type and p85α-deficient mice. No significant differences in glucose production were observed between the two mouse groups under either fasted or refed conditions. This result was consistent with the EGP results in the glucose-clamp study (Fig. 5).

Fig. 6.

Gluconeogenic activity in vivo. Total gluconeogenic activity in WT (n = 6) and p85α-deficient mice (n = 6) is shown. Values are means ± SE.

Expression of genes involved in the hepatic glycolysis/gluconeogenesis pathways.

We next examined the expression levels of G6Pase, PEPCK, and GK in the liver. The G6Pase mRNA levels were significantly higher (by 99%) in p85α-deficient mice than in wild-type mice under fasted conditions (Fig. 7A). The PEPCK mRNA levels were also significantly higher (by 62%) in p85α-deficient mice than in wild-type mice under fasted conditions (Fig. 7). After 6 h of refeeding after a 24-h fast, the G6Pase and PEPCK gene expression levels were lower in both wild-type and p85α-deficient mice than in the respective mouse groups after fasting. No significant differences in the expression of G6Pase and PEPCK were observed between p85α-deficient and wild-type mice under refed conditions (Fig. 7, A and B). GK expression in p85α-deficient mice tended to be higher than that in wild-type mice, although the difference was not statistically significant (Fig. 7C).

Fig. 7.

Gene expression levels in WT and p85α-deficient mice (p85α−/−). glucose-6-phosphatase (G6Pase; A), phosphoenolpyruvate carboxykinase (PEPCK; B), and glucokinase (GK; C) expression in the livers of WT (n = 6) and p85α-deficient mice (n = 6) are shown. Mice were subjected to fasting for 24 h or were refed for 6 h after a 24-h period of starvation. Expression levels were compared after normalization to β-actin level. Values are means ± SE. *P < 0.05.

Depleted hexoses in livers of p85α-deficient mice under fasted conditions.

Glucose is produced in the liver and renal cortex from nonglucose precursors through the process of gluconeogenesis. In p85α-deficient mice, the serum levels of nonglucose precursors, such as alanine, glutamine, pyruvate, lactate, and glycerol, were unaltered (Table 1), suggesting that sufficient supplies for gluconeogenesis were available from peripheral tissues. The serum corticosterone levels were 164 ± 18 ng/ml (n = 5) in wild-type mice and 156 ± 63 ng/ml (n = 5) in p85α-deficient mice (NS), indicating that the adrenal cortex function, which is necessary for gluconeogenesis in mice, was normal in the p85α-deficient mice. The serum levels of catecholamines were not reduced in p85α-deficient mice. Rather, the serum glucagon level was significantly higher in p85α-deficient mice than in wild-type mice under both fed and fasted conditions (Table 1).

View this table:
Table 1.

Serum levels of lipids, gluconeogenic precursors, and hormones that promote gluconeogenesis

The levels of intrahepatic intermediates involved in the gluconeogenic and glycolytic pathways in freely fed or 24-h-fasted wild-type and p85α-deficient mice are summarized in Table 2. Under fasted conditions in p85α-deficient mice, the G6Pase level was almost completely depleted and the fructose-6-phosphate level tended to be reduced, although the difference was not statistically significant. Under fed conditions, the G6Pase and phosphoenolpyruvate levels were higher in p85α-deficient mice than in wild-type mice, while the 2-phosphoglycerate level was lower in p85α-deficient mice than in wild-type mice.

View this table:
Table 2.

Intrahepatic substrates in the glucose metabolic pathway

Glycogen depletion in livers from p85α-deficient mice shortly after a fast.

Glycogen deposition and glycogenolysis in the liver play pivotal roles in glucose homeostasis. Under freely fed conditions, the total amount of glycogen stored in the liver was lower, although not significantly, in p85α-deficient mice than in wild-type mice (Fig. 8A). The glycogen content fell to nearly zero via glycogenolysis in p85α-deficient mice after a 12-h fast, while it was not completely depleted in wild-type mice even after a 30-h fast (Fig. 8A). The plasma cAMP level is known to increase in response to glucagon, which stimulates glycogenolysis by converting phosphorylase from an inactive form to an active form. When we injected glucagon into freely fed wild-type and p85α-deficient mice, the changes in the plasma cAMP levels were indistinguishable (Fig. 8B). A similar result was obtained when the two types of mice were loaded with isoproterenol (Fig. 8C). Although glycogen stored in the liver was broken down more rapidly in p85α-deficient mice than in wild-type mice in response to glucagon, the increments in circulating blood glucose were disproportionately small (Fig. 8D).

Fig. 8.

Glycogenolysis in the liver. A: changes in hepatic glycogen content during fasting. Glycogen content in liver was measured before and after 12 and 30 h of fasting. Values are means ± SE obtained from analysis of WT (open bars; n = 6) and p85α−/− mice (filled bars; n = 6). **P < 0.01 compared with WT mice. B: changes in the plasma cAMP level after glucagon stimulation. Freely fed mice were given 0.05 U of human glucagon per kg of body weight. Plasma cAMP levels were then measured at the indicated time points. Values are means ± SE obtained from the analysis of WT (○; n = 8) and p85α−/− mice (□; n = 12). C: changes in plasma cAMP level after isoproterenol stimulation. Freely fed mice were given 0.5 μg of human isoproterenol. Plasma cAMP levels were then measured at the indicated time points. Values are means ± SE obtained from the analysis of WT (○; n = 6) and p85α−/− mice (□; n = 6). D: changes in hepatic glycogen content and plasma glucose levels in response to glucagon. Fed mice were given 0.05 U of human glucagon per kg of body weight. Hepatic glycogen contents (left) and plasma glucose levels (right) were measured at indicated time points. Values are means ± SE obtained from the analysis of WT (open bars; n = 8) and p85α−/− mice (filled bars; n = 8). *P < 0.05, **P < 0.01 compared with WT mice.

DISCUSSION

In this study, we report four novel findings. First, insulin-stimulated PI3K activity associated with IRS in the liver was mediated via full-length p85α in wild-type mice and via the p50α alternative splicing isoform of the same gene in p85α-deficient mice. In p85α-deficient mice, the ratio of p50α over p110 in liver was higher than that in muscles, suggesting that the molecular balance between the regulatory subunit and the catalytic subunit of PI3K was different between the two tissues. Insulin-stimulated Akt and PTEN activities in the liver were similar in p85α-deficient and wild-type mice. Second, a hyperinsulinemic-euglycemic clamp study revealed that GIR and Rd were higher in p85α-deficient mice than in wild-type mice, consistent with a previous report (43) describing an increased sensitivity in muscles. By contrast, EGP tended to be higher in the p85α-deficient mice than in wild-type mice, although the difference was not statistically significant. Moreover, under fasted conditions, the hepatic expression of G6Pase and PEPCK was higher in p85α-deficient mice than in wild-type mice. Third, under fasted conditions, the intrahepatic G6Pase level was almost completely depleted and the fructose-6-phosphate tended to be reduced in p85α-deficient mice. Fourth, glycogenolysis was not lower in p85α-deficient mice than in wild-type mice. The glycogen profiles suggested that glucose utilization was increased in peripheral tissues or that glucose turnover was increased in vivo, consistent with a previous report (43) describing an increased glucose uptake in the peripheral tissues of p85α-deficient mice. Thus the liver did not exacerbate hypoglycemia but rather partially compensated for the extraordinary increase in glucose uptake in peripheral tissues to maintain glucose homeostasis in p85α-deficient mice.

Interestingly, insulin action was improved in the muscles, but not in the liver, of p85α-deficient mice (Fig. 5, AC). We hypothesize that the absence of an increase in insulin-stimulated PI3K activation in the liver can be explained by the molecular balance between the regulatory subunit and the catalytic subunit of PI3K in p85α-deficient mice, as discussed by Taniguchi et al. In p85α-deficient mice, p50α played a role in insulin-stimulated PI3K activation by binding to IRS-1/2. Although p50α was expressed in both wild-type and p85α-deficient mice, the expression of p50α was much lower than that of p85α in the muscles of wild-type mice (43). We actually performed a direct comparison of the regulatory and catalytic subunits of PI3K in liver and muscles and noted that in p85α-deficient mice, the ratio of p50α over p110 in liver was higher than that in muscles (Fig. 2A). These results suggested that the molecular balance between the regulatory subunit and the catalytic subunit of PI3K was different between the two tissues.

Based on experimental findings in adipocytes, we previously assumed that p50α might have a more potent effect on PI3K activation in vivo than p85α (43). It should be noted, however, that p50α/p55α-knockout mice exhibited enhanced insulin sensitivity (5). Moreover, Ueki et al. (44) reported that an increase in the expression level of p85α but not of p50α inhibited both phosphotyrosine-associated and p110-associated PI3-kinase activities in vitro (44), and they (45) proposed that optimal signaling through the PI3K pathway depended on a critical molecular balance between the regulatory and catalytic subunits. With the assumption that p50α has a more potent effect on PI3K activation than p85α in the liver, p85α-deficient mice should have a lower EGP than wild-type mice. However, EGP tended to be higher in the p85α-deficient mice than in wild-type mice. Indeed, no significant differences in PI3K activities associated with IRS-1/2 were observed under either fasted or refed conditions (Fig. 3, C and D). Together, the previous reports and the present study suggest that p50α is not more potent than p85α with respect to the activation of PI3K and that optimal signaling through the PI3K pathway depends on a critical molecular balance between the regulatory and catalytic subunits, although we were unable to directly address the free p85 hypothesis (45). Theoretically, decreased PTEN activity due to less p85-p110 bound to phosphorylated IRS molecule can explain the increased PIP3 generation in response to insulin (38), but insulin-stimulated PTEN activity was unaffected in liver of p85α-deficient mice.

How can we understand glucose metabolism in the livers of p85α-deficient mice? The expression of G6Pase and PEPCK was higher in p85α-deficient mice than in wild-type mice under fasted conditions. Because the G6Pase and PEPCK genes are reportedly controlled by insulin via the downstream PI3K signal (1, 7, 26, 35), the decreased serum insulin level (Table 1) may upregulate the expression of these gluconeogenic enzymes. Possibly, the reduction in serum insulin may have led to a reduction in insulin action in the central nervous system, thereby downregulating the gluconeogenic pathway (23, 30). Insulin and glucagon are known to have antagonistic actions on the expression of genes encoding all of the key enzymes involved in the glycolytic and gluconeogenic pathways (28). Thus glucagon has actions that oppose insulin by increasing hepatic cAMP. In this respect, the relatively high glucagon-to-insulin ratio in p85α-deficient mice (Table 1) may indicate the inhibition of glycolysis and the noninhibition or even enhancement of gluconeogenesis. Moreover, because p85α-deficient mice have a body weight and fat mass similar to those of wild-type mice, but significantly higher serum leptin levels when fed a normal diet (41), ablation of p85α may alter insulin/leptin signaling and their actions in the hypothalamus. To further understand the involvement of these molecules in glucose metabolism and the pertinent regulatory mechanisms, insulin action in the hypothalamus and the contribution of glucagon to the gluconeogenic pathway in the livers of p85α-deficient mice must be investigated. If the blockade of glucagon action in the livers of p85α-deficient mice leads to the downregulation of G6Pase and PEPCK, the increased glucagon level might play a major role in the regulation of gluconeogenesis. Otherwise, the contribution of the hypothalamus via insulin action may be important, and involvement of PI3K in the hypothalamus in the regulation of glucose metabolism in the liver should be intensively investigated.

The liver has been suggested to be capable of responding to hypoglycemia in the absence of any ability to secrete counterregulatory hormones or neural pathways between the liver and the brain (27). Normal mongrel dogs subjected to troglitazone treatment were shown to increase their endogenous glucose production in the liver as a result of elevations in both gluconeogenesis and glycogenolysis (6). The authors hypothesized that a protective mechanism existed in normal animals, preventing hypoglycemia during insulin sensitization with troglitazone. Thus increased glucose utilization in the peripheral tissues may itself, at least in part, upregulate gluconeogenic enzymes like G6Pase and PEPCK, thereby preventing severe hypoglycemia in p85α-deficient mice under fasted conditions. Recently, however, involvement of the liver in glucose metabolism in other tissues via the neuronal network has been suggested (14, 46). In this context, it should be noted that a liver-specific deletion of the p85α regulatory subunit, in which p50α was also abrogated, resulted in improvement in peripheral insulin sensitivity (38). This may be, at least in part, due to the neuronal network between liver and peripheral tissues. The results of our study can be interpreted that the liver partially compensated for the increase in glucose uptake in peripheral tissues to maintain glucose homeostasis in p85α-deficient mice but that the neuronal information pathway between the liver and peripheral tissues should be investigated in the future.

In summary, p50α played a role in insulin-stimulated PI3K activation by binding to hepatic IRS-1/2 in p85α-deficient mice. PI3K activity associated with IRS-1/2 was not affected by the lack of p85α in the liver. The absence of improvement in insulin-stimulated PI3K activation in the liver of p85α-deficient mice, unlike the muscles, may be associated with the molecular balance between the regulatory subunit and the catalytic subunit of PI3K. In p85α-deficient mice, glucose production from the liver was rather elevated, but not suppressed, in marked contrast to the increased insulin sensitivity in peripheral tissues. Consequently, the liver seemed to partially compensate for the increase in glucose uptake in peripheral tissues in p85α-deficient mice.

GRANTS

This work was supported by a grant for Life & Socio-Medical Science from the Kanae Foundation, a Grant-in-aid from the Mochida Memorial Foundation for Medical and Pharmaceutical Research (to Y. Terauchi), a Grant-in-aid for Scientific Research (C) 19591062 from the Ministry of Education, Culture, Sports, Science and Technology of Japan (to K. Aoki); a Grant-in-Aid for Creative Scientific Research (10NP0201) from the Japan Society for the Promotion of Science; and a Grant-in-Aid for the Development of Innovative Technology from the Ministry of Education, Culture, Sports, Science and Technology of Japan, Health Science Research Grants (Research on the Human Genome and Gene Therapy) from the Ministry of Health and Welfare (to T. Kadowaki). This work was also supported by the Yokohama City University Center of Excellence Program of the Ministry of Education, Culture, Sports, Science and Technology of Japan and a grant for the Strategic Research Project of Yokohama City University (K18005; to Y. Terauchi).

Acknowledgments

We thank Eri Yoshida-Nagata, Yuko Muto, Hiroshi Chiyonobu, Eri Sakamoto, and Mitsuyo Kaji for excellent technical assistance and animal care.

Footnotes

  • The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

REFERENCES

View Abstract