Short-Term Leptin-Dependent Inhibition of Hepatic Gluconeogenesis Is Mediated by Insulin Receptor Substrate-2

Abstract

Leptin has both insulin-like and insulin-antagonistic effects on glucose metabolism. To test whether leptin interferes directly with insulin signaling, we perfused isolated rat livers with leptin (0.1, 0.5, 5, and 25 nmol/liter), leptin + insulin (5 nmol/liter + 10 nmol/liter), insulin (10 nmol/liter), or vehicle (control). Leptin reduced L-lactate-(10 mmol/liter)-stimulated glucose production by 39–66% (P < 0.006 vs. control) and phosphoenolpyruvate carboxykinase (PEPCK) activity by 22–52% (P < 0.001). Physiological leptin concentrations (0.1–5 nmol/liter) stimulated the tyrosine phosphorylation (pY) of insulin receptor substrate-2 (IRS-2) (280–954%; P < 0.05) and its associated phosphatidylinositol-3 kinase activity (122–621%; P < 0.003). Leptin (0.5–25 nmol/liter) inhibited IRS-1 pY and its associated phosphatidylinositol-3 kinase activity (20–89%; P < 0.03) but stimulated janus kinase-2 pY (272–342%; P < 0.001). Leptin also down-regulated its short receptor isoform in a time- and concentration-dependent manner (28–54%; P < 0.05). Exposure to leptin + insulin additively reduced glucose production and PEPCK activity (∼50%; P < 0.001 vs. control) and doubled IRS-2 pY (P < 0.01 vs. insulin). However, leptin + insulin decreased IRS-1 pY by 57% (P < 0.01 vs. insulin). Insulin alone (P < 0.01), but not leptin, increased autophosphorylation of nonreceptor tyrosine kinases (pp59^Lyn + pp125^Fak).

In conclusion, leptin both alone and in combination with insulin reduces hepatic glucose production by decreasing the synthesis of the key enzyme of gluconeogenesis, PEPCK, which results mainly from the stimulation of the IRS-2 pathway.

LEPTIN, WHICH IS produced by adipocytes, helps to regulate food intake and energy expenditure through specific receptors located in the hypothalamus (1, 2). Plasma leptin concentrations are altered in type 2 diabetic patients (3, 4). Short-term leptin treatment also stimulates carbohydrate and lipid metabolism (5) and could therefore contribute to hepatic steatosis in obese patients (6).

Recent studies show that leptin affects hepatic glucose metabolism both indirectly via the central nervous system (7), and also directly through leptin receptors, which have been identified in well differentiated hepatoma cells (8) and hepatocytes (9). The short receptor isoform (Ob-R_S) is expressed in various tissues and is thought to play only a minor role in signal transduction, in contrast to the long isoform (Ob-R_L) which seems to play a more important role (10, 11).

In rat livers, short-term leptin infusion into the portal vein had an insulin-like effect on postprandial glycogenolysis, but a glucagon-like effect on postabsorptive gluconeogenesis (12). Other studies reported that leptin, in combination with insulin, additively increases glycogen storage (13) and reduces glucose production in isolated hepatocytes by approximately 30% (14).

Conflicting results have also been reported in studies on the interaction of leptin with insulin signal transduction. Leptin has been reported to inhibit, stimulate, or have no effect on insulin signaling in hepatoma cell lines (8, 15, 16). These inconsistent findings could be explained, at least in part, by different degrees of down-regulation of leptin receptor isoforms in response to varying leptin concentrations and exposure periods. In Chinese hamster ovary cells, the short-term exposure to leptin in the ranges of 0.1–100 nmol/liter similarly down-regulated Ob-R_S, but reduced Ob-R_L in a concentration-dependent manner (17).

Szanto and Kahn (8) recently showed that leptin alone had no effect on the insulin signaling pathway, whereas leptin pretreatment enhanced insulin-induced tyrosine phosphorylation (pY) of insulin receptor substrate-1 (IRS-1) but not of insulin receptor substrate-2 (IRS-2). IRS-1 and IRS-2 share some functions, including the activation of phosphatidylinositol-3-kinases (PI-3K) and the ras pathway, as well as the inhibition of glycogen synthase kinase-3 (GSK3) (18). However, IRS-1 and IRS-2 also have some properties that are different, i.e. intracellular distribution (19), and they could affect carbohydrate and lipid metabolism in different ways (20).

In addition, IRS-1 and IRS-2 can be phosphorylated by nonreceptor tyrosine kinases (NRTKs), pp59^Lyn and pp125^Fak, which could be activated by leptin (21). It is unclear whether the pY sites are identical with those used by the insulin receptor. However, IRS tyrosine pY by NRTK is a necessary (22), although probably not a sufficient condition (23, 24) for insulin-independent signaling from IRS to the glucose transport system in insulin-responsive cells. This was shown in adipocytes by the positive correlation between pp125^Fak activity and the glucose transport rates in response to insulin-like stimuli (22), but not in response to insulin or IGF-I (23, 24). Finally, leptin could also activate serine pY of GSK3 and thereby stimulate glycogen synthesis (5, 25).

This study, therefore, was designed to examine the action of leptin simultaneously on 1) postabsorptive gluconeogenesis and its key enzyme, phosphoenolpyruvate carboxykinase (PEPCK); 2) intracellular signal transduction via the insulin receptor (IR) and IRS as well as via leptin receptor isoforms and janus kinase-2 (JAK-2); 3) insulin-stimulated signal transduction and glucose production; as well as on 4) GSK3 and insulin-independent pathways involving the NRTKs. We used the model of the isolated perfused rat liver to study the metabolic function of leptin in the intact organ because this model allows extrahepatic effects, such as basal insulin secretion, increased sympathetic activity and/or hypothalamic effects, which might interfere with glucose metabolism and insulin signal transduction (26), to be excluded.

RESULTS

Baseline Characteristics

At t = 0, the rates of glucose production (0.37 ± 0.01 μmol·min⁻¹·g⁻¹ liver) and lactate uptake (0.85 ± 0.03 μmol·min⁻¹·g⁻¹ liver) did not differ between the groups. The baseline portal pressure was also comparable in all groups (1.7 ± 0.1 cm H₂O).

Perfusion Performance

The portal pressure (in centimeters of H₂O) rose gradually and in a similar way in all groups (0.1 nmol/liter leptin: 3.3 ± 0.3; 0.5 nmol/liter leptin: 3.5 ± 0.5; 5 nmol/liter leptin, 4.4 ± 0.4; 25 nmol/liter leptin, 4.8 ± 1.2; glucagon, 4.3 ± 0.5; leptin + insulin, 3.3 ± 0.2; insulin, 3.8 ± 0.4; control, 3.8 ± 0.6). In the leptin, leptin + insulin, insulin, and control groups, Δ bile flow declined in a similar way by 0.12 ± 0.03 mg·min⁻¹·g⁻¹ liver until the end of perfusion. Glucagon increased bile flow by 0.13 mg·min⁻¹·g⁻¹ liver (P < 0.05 vs. control) after 90 min, as described previously (12, 27).

Lactate-Dependent Glucose Production

During recirculating perfusion, perfusate glucose mass rose by 35.2 ± 2.4 μmol·g⁻¹ liver within 90 min under control conditions (Fig. 1, A and B). Infusion of 0.1 nmol/liter, 0.5 nmol/liter, 5 nmol/liter, and 25 nmol/liter leptin reduced glucose mass by 45%, 68%, 63%, and 43% (P < 0.001 vs. control), respectively (Fig. 1A). At 90 min, glucose mass was lower in the presence of leptin + insulin (−49%) or insulin alone (−23%) (each P < 0.03 vs. control) (Fig. 1B). Glucagon exposure increased perfusate glucose mass by 30% (P = 0.03 vs. control) (Fig. 1B).

Figure 1.

Time Course of the Increase of Perfusate Glucose Mass Relative to Liver Mass (in μmol·g Liver⁻¹ Wet Weight) during Recirculating Perfusion of Livers Obtained from 20-h Fasted Rats Using a Medium Containing 10 mmol/liter l-Lactate and 4 mmol/liter d-Glucose ☆, P < 0.05 vs. control (ANOVA, Dunnett post hoc test).

Glucose production rates (in μmol·min⁻¹·g⁻¹ liver), as calculated from linear regression of the time-dependent change in glucose mass, declined markedly during the infusion of leptin (0.1 nmol/liter, 0.19 ± 0.03; 0.5 nmol/liter, 0.13 ± 0.01; 5 nmol/liter, 0.14 ± 0.02; 25 nmol/liter, 0.23 ± 0.04; each P < 0.001 vs. control, 0.36 ± 0.02) and leptin + insulin (0.16 ± 0.02; P < 0.001 vs. control) (Fig. 2).

Figure 2.

Correlation between Glucose Production Rate (in μmol·min⁻¹·g Liver⁻¹ Wet Weight) on the Vertical Axis and PEPCK Activity (in Percent of Control) on the Horizontal Axis Rates of glucose production were calculated from the best fit of the respective concentrations to a line using the method of least squares from t = 0 to 90 min of perfusion and are given in means ± sem of the slope of that line. Correlation analysis was performed by using Pearson’s method.

PEPCK Activity

In the control group, PEPCK activity increased by 58 ± 4% from t = 0 to 90 min of perfusion (P < 0.001 vs. control at 90 min). After 90 min of exposure to 0.1 nmol/liter, 0.5 nmol/liter, 5 nmol/liter, and 25 nmol/liter leptin, the relative activity of PEPCK was 38%, 52%, 38%, and 22% lower than under control conditions (P < 0.0005 vs. control at 90 min), respectively (Fig. 2). Leptin alone (5 nmol) and insulin alone (10 nmol/liter) reduced PEPCK activity by 38% (see above) and by 19% (P < 0.00001 vs. control at 90 min, Fig. 2), respectively. Furthermore, leptin (5 nmol/liter) + insulin (10 nmol/liter) reduced PEPCK activity by 53% (P < 0.0005 vs. control at 90 min), indicating an additive effect of combined leptin and insulin exposure. PEPCK activities were linearly (r = 0.802; P < 0.00003) related to glucose production rates (Fig. 2).

Lactate Uptake

Rates of lactate uptake (in μmol·min⁻¹·g⁻¹ liver) were slightly, but not significantly, lower during the infusion of leptin (0.1 nmol/liter, 0.63 ± 0.13; 0.5 nmol/liter, 0.54 ± 0.06; 5 nmol/liter, 0.42 ± 0.05; and 25 nmol/liter, 0.47 ± 0.07), leptin + insulin (0.34 ± 0.05), and insulin (0.50 ± 0.04). In contrast, glucagon markedly increased lactate uptake (1.04 ± 0.26; P < 0.05 vs. control, 0.65 ± 0.08).

Liver Glycogen

After 90 min, liver glycogen concentrations (in μmol glycosyl units·g⁻¹ liver) were low in all groups and not affected by leptin (0.5 nmol/liter, 1.8 ± 0.2; 5 nmol/liter, 1.5 ± 0.3; 25 nmol/liter, 1.6 ± 0.3), insulin (1.6 ± 0.1), leptin + insulin (1.7 ± 0.5), or glucagon (1.3 ± 0.2) when compared with either control (1.6 ± 0.3) or basal conditions (1.8 ± 0.3).

Tyrosine pY of Insulin Receptor (IR) β Chain (IRβ pY)

Leptin infusion did not affect IRβ pY (Fig. 3A), whereas leptin + insulin (+229%) and insulin (+152%) increased IRβ pY (each P < 0.001 vs. control) (Fig. 3B). The combination of leptin and insulin did not, however, alter IRβ pY when compared with insulin-stimulated conditions (P = 0.09).

Figure 3.

Effect of Leptin on Tyrosine pY of IR β-Subunit During Liver Perfusion IRβ was immunoprecipitated from extract prepared from livers after 90 min perfusion with or without increasing concentrations of leptin (0.1, 0.5, 5, and 25 nmol/liter) (panel A); and insulin (10 nmol/liter) and insulin (10 nmol/liter) + leptin (5 nmol/liter) (panel B). The figure shows representative autoradiograms of immunoblots performed in duplicate and repeated four times for each perfusion condition with similar results. Data are given as means ± sem relative to vehicle infusion (percentage of control). ★, P < 0.01 vs. control (ANOVA, Dunnett post hoc test).

IRS-1 Tyrosine pY (IRS-1 pY) and Association with the 85-kDa Subunit of PI-3K (Amount)

Leptin (0.5, 5, and 25 nmol/liter) reduced IRS-1 pY in a concentration-dependent manner by 20%, 52%, and 83%, respectively (P < 0.01 vs. control) (Fig. 4A). The 4-fold increase in IRS-1 pY by insulin alone was suppressed by approximately 50% when leptin was also present (P < 0.001 vs. control) (Fig. 4B). Leptin (0.1, 0.5, 5, and 25 nmol/liter) also reduced the amount of IRS-1-associated PI-3K in a concentration-dependent manner by 21%, 48%, 72%, and 89% (each P < 0.001 vs. control), respectively (Fig. 4C). Similarly, the approximately 5-fold increase in the amount of IRS-1-associated PI-3K after insulin exposure (P < 0.001 vs. control) was lower in the presence of both leptin + insulin (−29%; P < 0.03 vs. insulin), even though this increase still exceeded the amount of IRS-1-associated PI-3K observed under control conditions (+226%; P < 0.001 vs. control) (Fig. 4D). Of note, the amount of IRS-1 pY and its associated PI-3K did not change during the 90-min control perfusion period when compared with t = 0 (data not shown).

Figure 4.

Stimulation by Leptin of IRS-1 pY in the Absence (A) and Presence of Insulin (B) as Well as of Amount of IRS-1-Associated PI-3K without (C) and with (D) Insulin Added Representative autoradiograms of the immunoblots are shown in the respective insets. Data are given as means ± sem relative to vehicle infusion (percentage of control). ★, P < 0.01 vs. control (ANOVA, Dunnett post hoc test).

IRS-2 Tyrosine pY (IRS-2 pY) and Association with the 85-kDa Subunit of PI-3K

The lower concentrations of leptin (0.1, 0.5, and 5 nmol/liter) increased IRS-2 pY by 954%, 842%, and 280% (each P < 0.05 vs. control), respectively (Fig. 5A). After exposure to insulin, IRS-2 pY rose by approximately 170% and increased more than 4-fold in the presence of insulin + leptin (P < 0.001 vs. control; P < 0.01 vs. insulin) (Fig. 5B). Similarly, leptin (0.1, 0.5, and 5 nmol/liter) increased the amount of IRS-2-associated PI-3K by 122%, 621%, and 365% (P < 0.003 vs. control), respectively (Fig. 5C). However, 25 nmol/liter leptin did not affect the amount of IRS-2 pY and its associated PI-3K (Fig. 5, A and C). Insulin alone increased the amount of IRS-2-associated PI-3K by 337% (P < 0.001 vs. control). The combination of leptin + insulin had an even greater effect, inducing an approximately 11-fold rise in the amount of that enzyme (P < 0.001 vs. control; P < 0.01 vs. insulin) (Fig. 5D).

Figure 5.

Stimulation by Leptin of IRS-2 pY in the Absence (A) and Presence (B) of Insulin, as Well as of Amount of IRS-2-Associated PI-3K without (C) and with (D) Insulin Added Representative autoradiograms of the immunoblots are shown in the respective insets. Data are given as means ± sem relative to vehicle infusion (percentage of control). ★, P < 0.05 vs. control (ANOVA, Dunnett post hoc test).

Activity of PI-3K Associated with IRS-1 and IRS-2

As in the case of the corresponding amounts of IRS-dependent PI-3K, leptin (0.5, 5, and 25 nmol/liter) reduced the activity of IRS-1-associated PI-3K by 26%, 57%, and 82% (P < 0.03 vs. control), respectively (Fig. 6A). However, lower leptin concentrations (0.1, 0.5, and 5 nmol/liter) increased the activity of IRS-2-associated PI-3K by 160–314% (P < 0.001 vs. control). Of note, 25 nmol/liter did not affect the activity of IRS-2-dependent PI-3K (Fig. 6B).

Figure 6.

Effect of Leptin on the Activity of PI-3K Associated with IRS-1 (A) or IRS-2 (B) during Liver Perfusion Representative autoradiograms of the TLC are shown in the respective insets. Data are given as means ± sem relative to vehicle infusion (percentage of control). ★, P < 0.03 vs. control (ANOVA, Dunnett post hoc test).

Hepatic Leptin Receptors

No significant amount of Ob-R_L was found in these rat livers but the short isoform Ob-R_S was clearly detected at 112 kDa (Fig. 7A). Leptin down-regulated the amount of Ob-R_s in a concentration-dependent manner (0.5 nmol/liter, −28%; 5 nmol/liter, −50%; and 25 nmol/liter, −54%) (Fig. 7A) (P < 0.05 vs. control). Leptin (5 nmol/liter) reduced, in a time-dependent way, the amount of Ob-R_s (−49%; P < 0.001 vs. control and baseline), which remained unchanged under control conditions (Fig. 7B).

Figure 7.

Time- and Concentration-Dependent Effects of Leptin on the Amount of Its Receptors (Ob-R_L and Ob-R_S) The leptin receptor was immunoprecipitated with antibodies recognizing both the long (170-kDa bond) and short (112-kDa bond) isoforms from liver extracts after 90-min perfusion with or without increasing concentrations of leptin (A) or after increasing perfusion periods in the absence or presence of 5 nmol/liter leptin (B). The panels give the amount of Ob-R_S as means ± sem relative to vehicle infusion (panel A, percentage of control) or t = 0 conditions (panel B, percentage of baseline). Representative autoradiograms of the immunoblots are shown in the respective insets. ★, P < 0.05 vs. control (ANOVA, Dunnett post hoc test); ¶, P < 0.001 vs. baseline control conditions.

Tyrosine pY of JAK-2

Higher concentrations of leptin (0.5, 5, and 25 nmol/liter) similarly induced an approximately 4- to 5-fold increase of JAK-2 pY (P < 0.001) (Fig. 8, A and B), whereas insulin alone or in combination with leptin was not effective when compared with control or leptin-stimulated conditions, respectively (Fig. 8B).

Figure 8.

Effect of Leptin and Insulin on Tyrosine pY of JAK-2 during Liver Perfusion Representative autoradiograms of the immunoblots are shown in the respective insets. Data are given in means ± sem. ★, P < 0.001 vs. control (ANOVA, Dunnett post hoc test).

Serine Phosphorylation and Activity of GSK3

Exposure to low concentrations of leptin (0.1, 0.5, and 5 nmol/liter) increased GSK3 serine phosphorylation by 143%, 562%, and 395% (each P < 0.01 vs. control), but a marginal and not significant stimulation took place only in the presence of 25 nmol/liter leptin (Fig. 9A). Insulin alone increased GSK3 serine phosphorylation (+275%), which was doubled after an infusion of leptin + insulin (Fig. 9A) (P < 0.01 vs. control; P < 0.01 vs. insulin). Consistent with a negative regulation of GSK3 by serine phosphorylation, exposure to 0.1, 0.5, 5, and 25 nmol/liter leptin reduced GSK3 activity by 30%, 66%, 59%, and 22%, respectively (P < 0.01 vs. control) (Fig. 9B). In agreement with the effect on GSK3 serine phosphorylation, leptin + insulin (−68%; P < 0.01 vs. control; P < 0.05 vs. insulin) additively reduced GSK3 activity more than insulin alone (−56%; P < 0.01 vs. control).

Figure 9.

Effect of Leptin and Insulin on Serine pY (A) and on the Activity (B) of GSK3 after 90-Min Liver Perfusion Portions of the immunoprecipitates from liver extracts were immunoblotted for phosphoserine (A) or assayed for GSK3 activity (B). Data are given in means ± sem (percentage of control). ★, P < 0.02 vs. control (ANOVA, Dunnett post hoc test).

NRTKs pp59^Lyn and pp125^Fak

Insulin alone, but not leptin, increased by between 70% and 200% the autophosphorylation of pp59^Lyn and pp125^Fak as well as the pY of pp59^Lyn with IRS-2 as the substrate (each P < 0.01 vs. control) (Fig. 10, A–C).

Figure 10.

Effect of Leptin and Insulin on NRTKs pp59^Lyn and pp125^Fak during Liver Perfusion Src-class kinase pp59^Lyn and focal adhesion kinase pp125^Fak activities were determined either as autophosphorylation (panel A, pp59^Lyn and panel C, pp125^Fak) or with recombinant human IRS-2 as substrate (panel B, pp59^Lyn). Data are given as means ± sem relative to vehicle infusion (percentage of control). ★, P < 0.01 vs. control (ANOVA, Dunnett post hoc test).

Total Protein Amount of IRβ, IRS-1, IRS-2, JAK-2, and GSK3

Infusion of leptin or insulin did not change the amount of protein in IRβ, IRS-1, IRS-2, JAK-2, and GSK3 during the entire perfusion period when compared with control conditions (Fig. 11).

Figure 11.

IRβ, IRS-1, IRS-2, GSK3, and JAK-2 Were Immunoprecipitated from Liver Extract after 90-Min Perfusion in the Absence or Presence of Different Concentrations of Leptin With or Without Insulin The amount of immunoprecipitated IRβ, IRS-1, IRS-2, GSK3, and JAK-2 recovered from the differently perfused livers was analyzed by immunoblotting with their specific antibodies. The figure shows representative autoradiograms of the immunoblots, which were performed in triplicates of four different experiments for each perfusion condition with similar results.

Correlation Analyses

There was a positive correlation of IRS-2 pY as well as of the amount and activity of its associated PI-3K with serine pY of GSK3 (IRS-2 pY: r = 0.559, P < 0.006; PI-3K amount: r = 0.977, P < 10⁻⁵; PI-3K activity: r = 0.877, P < 0.0001) but a negative correlation with the activities of GSK3 (IRS-2 pY: r = −0.587, P < 0.003; PI-3K amount: r = −0.905, P < 10⁻⁶; PI3-K activity: r = −0.802, P < 0.00004) and PEPCK (IRS-2: r = −0.757, P < 10⁻⁶; PI-3K amount: r = −0.847, P < 0.0001; PI-3K activity: r = −0.882, P < 0.00001). IRS-1 pY as well as the amount and activity of its PI-3K were negatively related to JAK-2 pY (IRS-1 pY: r = −0.867, P < 10⁻⁷; PI-3K amount: r = −0.938, P < 10⁻⁷; PI-3K activity: r =−0.862, P < 10⁻⁷). In addition, GSK3 serine phosphorylation negatively correlated with the activity of GSK3 (r = −0.915; P < 10⁻⁹).

DISCUSSION

In the perfused livers of 20-h fasted rats, short-term leptin infusion reduced rates of glucose production in an insulin-like manner. The range of portal leptin concentrations used (between 0.1 and 5 nmol/liter) covers the physiological range of plasma levels in lean and obese humans (1) and rodents (28, 29), whereas the high concentration of 25 nmol/liter leptin reflects supraphysiological levels (6) that could be present in the livers of severely obese subjects in the postprandial state because of secretion from the stomach into portal vein (30) and/or intrahepatic release (31).

Glucose Production and PEPCK Activity

This study found that leptin under near-physiological conditions, i.e. low leptin concentrations and intact liver function, markedly reduces glucose production by up to 60% and is even more effective than insulin (−19%). Interestingly, high leptin concentrations cancelled out, in part, this insulin-like effect. Glucose production rates were closely correlated with the activity of hepatic PEPCK, the rate-controlling enzyme of gluconeogenesis (32). PEPCK activity rose by approximately 60% compared with baseline conditions during the 90 min of our control experiments. These conditions resemble the fasting state with low-glucose, high-lactate concentrations and depleted hepatic glycogen in the absence of insulin, which results in high rates of PEPCK synthesis and subsequent increases in PEPCK activity (33, 34). Similarly, hepatic PEPCK transcription increases by up to 6-fold during starvation in vivo (35). As there are no known allosteric inhibitors of PEPCK, its activity is also a reflection of its amount of protein. The latter is the result of PEPCK protein synthesis, i.e. gene transcription from DNA to mRNA, stability of mRNA and protein translation, as well as of constant degradation at a half-life of more than 3 h (36–38). Given this half-life, perfusion for 90 min is too short a period for a relevant PEPCK decline (39). Thus, exposure to leptin and/or insulin for 90 min attenuated the increase in hepatic PEPCK activity but did not reduce PEPCK activity when compared with baseline conditions. Of note, insulin is able to inhibit PEPCK mRNA expression rapidly and to induce its degradation in rat hepatocytes and H4IIE hepatoma cells, but not in HepG2 cells, indicating that the insulin inhibition of PEPCK gene transcription requires one or more mediators (such as the factor complex 7) that are expressed differently in, or absent from, HepG2 cells (33, 35, 40). In our experiments, insulin and/or leptin also reduced GSK3 activity, which in turn was shown to reduce PEPCK gene transcription (41). However, lower PEPCK transcription cannot account for the drop in the PEPCK enzyme activity because changes in a specific amount of protein due to modifications in the DNA transcription can only be detected in experiments lasting much longer (39). In contrast to the long half-life of PEPCK protein, the synthesis rate of its mRNA has a short half-life of only 30 min and can be more rapidly regulated by hormones and metabolites (39). Thus, both insulin and leptin must have altered the enzyme activity by preventing additional synthesis of PEPCK mRNA, which also explains the observed attenuation of the increase in PEPCK activity during the 90-min perfusion period.

Tyrosine pY of IRS and Its Associated Kinases

Physiological leptin concentrations preferentially increased IRS-2 pY and its associated PI-3K, whereas high leptin concentrations decreased IRS-2 pY when compared with low leptin. On the other hand, leptin at concentrations between 0.1 to 25 nmol/liter had a marked inhibitory effect on IRS-1 pY and its associated PI-3K. Of note, recent studies indicate that IRS-1 is more important than IRS-2 in mediating mitogenic responses, whereas IRS-2 is essential for the regulation of hepatic glucose metabolism (42–44). In IR-deficient hepatocytes, insulin fully activates IRS-1, but not IRS-2, via IGF-I receptors. Insulin neither reduces glucose production (44) nor suppresses hepatic glucose release in rodents with disruption of the IRS-2 gene (43). Our findings also indicate that the reduction of hepatic glucose production depends mainly on the activation of the IRS-2 pathway, which might explain the conflicting results reported for different experimental conditions (7–9, 12–16). In addition, these results might also link the hyperleptinemia and hepatic insulin resistance observed in morbidly obese patients (45, 46), because very high leptin concentrations interfere with the tyrosine pY of both IRS-1 and IRS-2.

Leptin and Insulin Signal Transduction

It is thought that all of the hypothalamic and peripheral effects of leptin are mediated by leptin’s binding to the short or long receptor isoforms. The direct pY of IRβ by leptin-activated JAK-2 as a means of cross-talk between insulin and leptin signaling pathways can be excluded because the tyrosine pY of the IRβ chain was not affected by leptin but was markedly stimulated by insulin (∼150%). This also shows that leptin does not exert its hepatocellular action by stimulating the IR through trans-autophosphorylation (47). We investigated the time- and concentration-dependent effects of leptin on the amount and downstream signaling of its short- (Ob-R_S) and long (Ob-R_L) receptor isoforms. We did not detect Ob-R_L in rat liver, in agreement with studies in freshly prepared rodent livers (11, 48) and rat hepatocytes (9). Nevertheless, we found that leptin is able to induce signal transduction via Ob-R_S, which is in line with previous reports on hepatocytes (9, 15). Leptin induced signal transduction by causing the downstream activation of JAK-2, which has been reported in Chinese hamster ovary cells (10). Alternatively, JAK-2 could have been responsible not only for the observed increase in tyrosine pY of IRS-2, but also indirectly for the concentration-dependent reduction of IRS-1 pY as reported previously for C2C12 myotubes (49).

The src-homology-2 domain containing protein tyrosine phosphatase 2 (SHP-2) has recently been shown to modulate leptin signal transduction by reducing tyrosine pY of both JAK-2 and IRS-1 (49–53), an observation that could be relevant to the down-regulation of IRS tyrosine pY observed at high leptin concentrations. High leptin induced up-regulation of SHP-2 (53). This could explain not only the concentration-dependent reduction by leptin of IRS-1 tyrosine pY as shown in cell culture (53), but also the relative reduction of IRS-2 pY by high concentrations of leptin (5 and 25 nmol/liter) when compared with low concentrations of leptin (0.1 and 0.5 nmol/liter). On the other hand, we found a pronounced time- and concentration-dependent down-regulation of Ob-R_S, in line with findings in Chinese hamster ovary cells (17). The observed down-regulation of Ob-R_S could also have resulted in reduced IRS-2 pY and increased glucose production by high leptin concentrations.

GSK3 and Hepatic Glycogen

We previously reported that both leptin and insulin inhibit glucose production under postprandial conditions, indicating reduced glycogenolysis and/or increased glycogen synthesis (12). We therefore measured the effect of leptin on serine phosphorylation and the activity of GSK3, which also inhibits glycogen synthase (54). This study found that leptin stimulates serine phosphorylation and inactivation of GSK3 to a similar extent as insulin, which was shown to act via PI-3K and protein kinase B (25, 54). Although this effect of leptin and insulin would be expected to increase glycogen synthesis, we did not detect any changes in hepatic glycogen concentrations, probably because of the prevailing postabsorptive condition with very low basal hepatic glycogen levels and/or the short duration of leptin and/or insulin exposure.

NRTKs

We also studied whether leptin’s insulin-like action is mediated by alternative pathways that activate IRS and PI-3K. NRTKs are able to stimulate the entire insulin signaling pathway, independently of insulin, to increase glucose uptake in adipocytes (21, 22). We found that insulin, but not leptin, increases the autophosphorylation of the NRTKs, the src-class kinase pp59^Lyn and the focal adhesion kinase pp125^Fak, as well as the ability of pp59^Lyn to phosphorylate IRS-2 in intact liver. Previous studies have reported that pp125^Fak pY in nonattached adipocytes is promoted effectively only in vitro (22, 23). Because the IR tyrosine kinase can phosphorylate and stimulate pp125^Fak directly (23), the 2- to 3-fold increase in pp59^Lyn and pp125^Fak autophosphorylation by insulin was most likely a result of the 3-fold increase in IR tyrosine pY, which could contribute to insulin signaling.

Additive Effects of Insulin and Leptin

When compared with insulin-stimulated conditions, leptin additively increased IRS-2 pY and its PI-3K by approximately 2-fold, but decreased IRS-1 pY and its PI-3K by about 40%. The combination of leptin + insulin additively stimulated downstream signaling including GSK3 and PEPCK activity, which resulted in further reductions in glucose production. Despite the observed additive effects of leptin and insulin on IRS, GSK3, and PEPCK, only leptin, but not insulin, activated JAK-2. Leptin did not participate in the signaling cascade of NRTKs that were regulated by insulin only.

In summary, a short-term exposure to leptin 1) reduces hepatic gluconeogenesis by reducing PEPCK activity; 2) interferes with insulin signaling, exerting both an insulin-like and -antagonistic effect in liver tissue; 3) increases GSK3 serine phosphorylation, thereby reducing its activity; 4) activates JAK-2 via its short receptor isoform, which is down-regulated in a time- and concentration-dependent fashion; and 5), unlike insulin, does not induce IRS-2 tyrosine pY via the NRTKs pp59^Lyn and pp125^Fak.

MATERIALS AND METHODS

Animals

Male Sprague Dawley rats (Him:OFA/SPF, Tierforschungsinstitut, Himberg, Austria) weighing 318 ± 8 g and housed under conditions of a constant 12-h day, 12-h night cycle had free access to standard chow and water and were fasted for 20 h before the experiments. All protocols were approved by the animal ethics commission of the Austrian Federal Ministry for Education, Science and Culture and performed according to local laws as well as to the principles of good laboratory animal care.

Materials

Antibiotics, pyruvate kinase, and proteinase inhibitors were purchased from Roche Molecular Biochemicals (Mannheim, Germany); HEPES, morpholinoethanesulfonic acid, Tris, detergents, and dithiothreitol (DTT) from Calbiochem (Bad Soden, Germany); precast gels from Novex (San Diego, CA); polyvinylidene difluoride membranes (Immobilon) from Millipore Corp. (Eschborn, Germany), and chemiluminescent reagents (Renaissance Chemiluminescence Detection System) from NEN Life Science Products/DuPont (Bad Homburg, Germany).

Perfusion Performance

After rat livers had been prepared and isolated as previously described (6, 12), they were weighed (10.4 ± 0.2 g) and immediately perfused with oxygenated (93% O₂/7% CO₂; 37 C) Krebs-Henseleit buffer containing 4 mmol/liter d-(+)-glucose, 10 mmol/liter l-lactate, and 0.2% BSA in a recirculating system at a rate of 2.9 ± 0.1 ml·min⁻¹·g⁻¹ liver wet weight. During the 35-min equilibration period, intact liver function was assessed by constant lactate uptake, and by monitoring low portal pressure and continuous bile flow, as well as by the uniform brown color of the livers, an indication of sufficient vascular perfusion (6, 12). At t = 0, recombinant mouse leptin (a gift of Eli Lilly \|[amp ]\| Co., Indianapolis, IN; final concentrations: 0.1 nmol/liter, 0.5 nmol/liter, 5 nmol/liter, and 25 nmol/liter), insulin (final concentration, 10 nmol/liter; human insulin, Actrapid, Novo-Nordisk, Copenhagen, Denmark), insulin (10 nmol/liter) + leptin (5 nmol/liter), glucagon (final concentration, 1 nmol/liter; Novo-Nordisk) or the vehicle (Krebs-Henseleit buffer; control) were admixed to the portal infusate at rates of 0.44 ml·min⁻¹ and infused for 90 min. Livers were exposed to leptin for 90 min to achieve an equilibrium of leptin binding to its receptors (55). To examine the time-dependent effects of leptin, the perfusion was stopped in another series of experiments after the equilibration period (t = 0, n = 6) or after 45 min of exposure to 5 nmol/liter leptin (45 min, n = 4). Samples were taken every 5 min for the determination of metabolites. At the end of the experiments, the final recirculating volume was measured, and livers were immediately frozen by transferring them to liquid nitrogen and storing them at −80 C for the determination of glycogen and for the preparation of liver extract for subsequent measurements.

Metabolic Data

Perfusate glucose and lactate concentrations in the influent and effluent perfusate were assayed enzymatically using the hexokinase method (glucose liquiUV, Human, Wiesbaden, Germany) and the lactate dehydrogenase method (Roche, Mannheim, Germany), respectively. For the determination of liver glycogen, cell membranes were disrupted by 1 mol/liter KOH. Glycogen was then hydrolyzed to glucose and measured in the supernatant (56).

Preparation of Rat Liver Extract

Frozen liver (0.2–1 g wet weight) was homogenized in a buffer containing 50 mm HEPES/KOH (pH 7.4), 140 mm NaCl, 250 mm sucrose, 1 mm MgCl₂, 1 mm CaCl₂, 2 mm EDTA, 2.5 mm Na₃VO₄, 10 mm glycerol-3-phosphate, 20 mm NaPP_i, 20 mm NaF, 1 mm phenylphosphate, 5 μm okadaic acid (sodium salt), 1% Nonidet P-40, 10% glycerol and protease inhibitors (10 μg/ml leupeptin, 5 μg/ml pepstatin A, 75 μg/ml aprotinin, 100 μm benzamidine, 2 μg/ml antipain, 10 μg/ml soybean trypsin inhibitor, 5 μm microcystin, 5 μm E-64, 0.2 mm phenylmethylsulfonylfluoride) using an Ultraturrax T25 basic (three 10-sec cycles at 2000 rpm on ice) and then a tight-fitting teflon-in-glass homogenizer (five strokes at 500 rpm on ice). The total homogenate was centrifuged (30 min, 48,000 × g, 4 C). The supernatant was removed with great care to avoid any contamination with the upper fat layer and then recentrifuged. The fat-free supernatant obtained was stored in liquid N₂ and used as extract (3–5 mg protein per ml) for immunoprecipitations and PEPCK activity measurements. Protein concentration was evaluated by the bicinchoninic acid protein determination method (Pierce Chemical Co., Rockford, IL) using a BSA standard curve.

Immunoprecipitation of IRβ, IRS-1/2, GSK3, JAK-2, and Ob-R

Up to 1-ml portions of the liver extract supplemented with appropriate antibodies against IRS-1/2, IRβ, GSK3, JAK-2 (type HR-758, rabbit polyclonal, epitope corresponding to the internal domain of mouse JAK-2; Santa Cruz Biotechnology, Inc., Santa Cruz, CA) or Ob-R (type H-300, rabbit polyclonal, epitope corresponding to amino acids 541–840 of human Ob-R) were preadsorbed on protein A-Sepharose, incubated, and centrifugated as described previously (8, 22, 25, 42, 57). The collected immune complexes were washed twice with 1 ml each of immunoprecipitation buffer (50 mm HEPES/KOH, pH 7.4; 500 mm NaCl; 100 mm NaF; 10 mm EDTA; 10 mm NaPP_i; 2.5 mm Na₃VO₄) containing 1% Nonidet P-40, and then twice with 1 ml of immunoprecipitation buffer containing 150 mm NaCl and 0.2% Nonidet P-40 and once with 1 ml of immunoprecipitation buffer containing no salt or detergent and finally suspended in 50 μl of Laemmli buffer (2% sodium dodecyl sulfate, 5% 2-mercaptoethanol), heated (95 C, 2 min), and centrifuged. The supernatant samples were analyzed by SDS-PAGE (4–12% Bis-TRIS Novex precast gel, San Diego, CA; pH 6.4, morpholinoethanesulfonic acid/sodium dodecyl sulfate running buffer) under reducing conditions.

Immunoblotting

Immunoblotting was performed as described previously (8, 22, 42) with minor modifications. Briefly, after SDS-PAGE and the transfer of the proteins to polyvinylidene difluoride membranes, the blocked membrane was incubated (2 h, 25 C) with anti-IRβ (3 mg/liter), anti-IRS-1 (rabbit polyclonal, immunoaffinity purified, 1:500), anti-IRS-2 (1:250), anti-p85α [rabbit, protein A purified, raised against full-length rat p85 PI-3K, 1 mg/liter (Upstate Biotechnology, Inc., Lake Placid, NY], anti-GSK3 [phosphoserine 21- and 9-specific; New England Biolabs, Inc. (Beverly, MA) 9331, 1:250], anti-JAK-2 (type HR-758, rabbit polyclonal, epitope corresponding to the internal domain of mouse JAK-2; Santa Cruz Biotechnology, Inc.), anti-Ob-R (type K-20, goat polyclonal, epitope corresponding to the amino terminus of mouse Ob-R; Santa Cruz Biotechnology, Inc.), or antiphosphotyrosine (pY20, 1:1000; Transduction Laboratories, Inc., Lexington, KY) antibodies, and then washed five times. After incubation (1 h, 25 C) of the membranes with [¹²⁵I]protein A (5 μCi/ml, Amersham Pharmacia Biotech-Buchler, Braunschweig, Germany) in the same blocking medium, the membranes were washed five times and developed by enhanced chemiluminescence using SuperSignal substrate (Pierce Chemical Co.) and then visualized by autoradiography (X-Omat AR film, Kodak, Rochester, NY) and evaluated on a LUMI Imager (Roche Diagnostics, Mannheim, Germany) using LUMI Imager Software. Quantitative analysis of the blots was performed by using IMAGEQUANT software (Molecular Dynamics, Inc., Sunnyvale, CA). The amounts of immunoprecipitated protein recovered were corrected for the amount of protein actually applied onto the gel as revealed by homologous immunoblotting. Each experiment was performed with samples from four different liver perfusions with three to five independent immunoprecipitation/immunoblotting procedures.

PI-3K Activity

IRS immune complexes were incubated (10 min, 22 C) in 50 μl of 20 mm Tris/HCl (pH 7.0), 50 μm [γ-³³P]ATP (5 μCi, Perkin Elmer, Boston, MA), 10 mm MgCl₂, 2 mm MnCl₂, 100 mm NaCl, 2 mm EDTA, 0.5 μm wortmannin (for control incubations only) containing 10 μg of phosphatidylinositol (PI, Avanti Polar Lipids, Alabaster, AL) and 1 μg of phosphatidylserine (58). After thin layer chromatography (TLC) (59), radiolabeled phosphatidylinositol 3-phosphate (PI-3P) was visualized by autoradiography and quantitated by phosphorimaging of the [³³P]phosphate-containing TLC spot reflecting PI-3P. To calculate the wortmannin-sensitive PI-3K, all values were corrected for PI-3P radiolabeled in the presence of wortmannin.

GSK3 Activity

GSK3β activity was determined using immune complex assay with phospho-glycogen synthase peptide 2 (P-GS 2) as a substrate (54). The GSK3β immunoprecipitates were washed twice with homogenization buffer and once with assay buffer (20 mm Tris/HCl, pH 7.4; 1 mm DTT) and then suspended in 20 μl of assay buffer containing 0.4 mg/ml BSA, 10 mm MgCl₂, 30 μm [γ-³³P]ATP (6 μCi), and 20 μm P-GS 2. After incubation (15 min, 30 C), the reactions were terminated by the addition of 20 μl of 20% trichloroacetic acid and centrifugation (10,000 × g, 5 min). Then, 15-μl portions of the supernatant were spotted on 2.5 × 3-cm pieces of Whatman P81 phosphocellulose paper (Clifton, NJ); 20 sec later, the filters were washed five times with 0.75% phosphoric acid (for at least 5 min each time) and once with acetone. Radioactivity of dried filters was counted in the presence of 5 ml of scintillation fluid (ACS, Amersham Pharmacia Biotech). ³³P_i incorporation into the negative control peptide [glycogen synthase peptide 2 (Ala21)] was subtracted from values obtained using P-GS 2. No activity was measured with immunoprecipitates using nonimmune IgG. Each activity value was corrected for the amount of immunoprecipitated GSK3β according to immunoblotting.

PEPCK Activity

PEPCK was measured using the [¹⁴C]NaHCO₃ fixation assay as described by Noce and Utter (60) and Burcelin et al. (61) with some modifications. Rat liver extract (490 μl) was added to 500 μl of reaction buffer containing 150 μmol Tris/acetate (pH 7.2), 5 μmol sodium inosine 5′-diphosphate, 10 μmol MnCl₂, 250 μmol KCl, 10 mm DTT, 2 mm glutathione, 400 μmol KHCO₃, and 15 μCi NaH¹⁴CO₃ (10 μmol; Amersham Pharmacia Biotech-Buchler). The reaction was started by the addition of 10 μl of 0.4 m phosphoenolpyruvate and terminated after 10 min incubation at 25 C by the addition of 1 ml of 6 n HCl and by placing the tube on ice. After dilution with 1 ml of water, unreacted ¹⁴CO₂ (from H¹⁴CO₃⁻) was removed by bubbling with N₂ and CO₂ for 30 min each. The reaction mixture was supplemented with 10 ml of aqueous scintillation cocktail (ReadySafe, Beckman Coulter, Inc., Fullerton, CA) and measured for radioactivity by liquid scintillation counting (LS6500, Beckman Coulter, Inc.). From each value appropriate blanks containing the same ingredients but lacking either extract or inosine 5′-diphosphate were subtracted. Under these conditions and up to the maximal amount of extract used, the incorporation rates were linear for both extract concentrations and for time at least during the first 15 min.

NRTKs

Activation of pp59^Lyn and pp125^Fak was determined using immune complex kinase assays as previously described (22) with recombinant human IRS-2 as the substrate (pp59^Lyn) or as autophosphorylation.

Calculations

Basal glucose production and lactate uptake rates were determined as previously described (12). The perfusate glucose mass was corrected to zero at the start of infusions (basal period) and calculated as previously reported (12). Rates of glucose production and of lactate uptake were calculated from the best fit of the respective concentrations (μmol·g⁻¹ liver wet weight) over time to a line using the method of least squares and are given as the slope of that line (μmol·min⁻¹·g⁻¹ liver wet weight). Bile flow was corrected at t = 0, assessed from a single drop weight (8 mg per drop) × drop frequency per liver wet weight and is given as Δ bile flow. The basal values and the increase in portal pressure were registered in centimeters of H₂O on a column communicating with the tube, in which influent perfusate was circulating. Scintillation counting/phosphorimaging of immunoblots and enzyme activity was measured in arbitrary units and is given in percent relative to the values of the control group.

Statistical Analysis

All data are presented as means ± sem of four to six experiments. Comparisons to the control were performed with ANOVA after ad hoc Dunnett’s test (SPSS for Windows, version 10.0 for Windows, SPSS, Inc. Headquarters, Chicago, IL; http://www.spss.com/). Linear correlations are Pearson-product moment correlations. Differences between experimental groups were considered statistically significant at P < 0.05.

Acknowledgments

This work was supported in part by grants (FWF, P13213-MOB and P10416-MED) from the Austrian Science Foundation (to M.R.).

We are indebted to the skillful staff of the Institute of Biomedical Research, University of Vienna (head: Prof. Dr. U. Losert), for taking care of the rats and to N. Rouveyve, N. Hanekop, and W. Raffesberg for their excellent technical assistance. We would also like to thank J. Burgermeister for her careful editing of this manuscript.

Abbreviations:

 
• DTT,

  Dithiothreitol;

 
• GSK3,

  glycogen synthase kinase 3;

 
• GST,

  glutathione S-transferase;

 
• IR,

  insulin receptor;

 
• IRβ,

  insulin receptor β-subunit;

 
• IRS-1/2,

  insulin receptor substrate 1/2;

 
• JAK-2,

  janus kinase-2;

 
• NRTK,

  non-receptor tyrosine kinase(s);

 
• Ob-R,

  leptin receptor;

 
• Ob-R_L,

  long isoform of leptin receptor;

 
• Ob-R_S,

  short isoform of leptin receptor;

 
• PEPCK,

  phosphoenolpyruvate carboxykinase;

 
• P-GS 2,

  phospho-glycogen synthase peptide 2;

 
• PI,

  phosphatidylinositol;

 
• PI-3P,

  phosphatidylinositol 3-phosphate;

 
• PI-3K,

  phosphatidylinositol-3-kinase;

 
• pY,

  phosphorylation;

 
• SHP-2,

  src-homology-2 domain containing protein tyrosine phosphatase 2;

 
• TLC,

  thin layer chromatography.