Exercise-Induced Changes in Insulin and Glucagon Are Not Required for Enhanced Hepatic Glucose Uptake After Exercise but Influence the Fate of Glucose Within the Liver

All procedures described in this experiment were approved by the Vanderbilt University Animal Care and Use Committee. Animals were housed in a facility that met the American Association for the Accreditation of Laboratory Animals Care guidelines. A total of 21 mongrel dogs (mean weight 24 ± 1 kg) were fed a daily diet of meat and standard food (34% protein, 14.5% fat, 46% carbohydrate, and 5.5% fiber based on dry weight).

At least 16 days before an experiment, a laparotomy was performed under general anesthesia for the insertion of infusion and sampling catheters as well as transonic flow probes. Infusion catheters were inserted into the inferior vena cava for the infusion of somatostatin (SRIF), glucose, [3-³H]glucose, and l-[¹⁴C]glucose. Infusion catheters were also inserted into the splenic and jejunal veins and fed forward into the portal vein for the infusion of glucose, insulin, glucagon, and l-[¹⁴C]glucose. Sampling catheters were inserted into the portal vein and hepatic vein as well as the femoral artery. After insertion, catheters were filled with saline containing heparin and knotted at the free ends. Sections of the hepatic artery and portal vein upstream of the liver were cleared of tissue and fitted with Transonic flow probes for the measurement of blood flow. Flow probe leads and the knotted catheter ends were placed in a subcutaneous pocket made in the abdominal region. The femoral artery catheter was placed in a subcutaneous pocket made in the inguinal region. Animals that met the experiment inclusion criteria (normal stool, white blood cell count <18,000 [8], and a hematocrit >0.35) were fasted 18 h before an experiment.

After the 18-h fast, the catheters and flow probes were freed from the subcutaneous pockets using 2-cm incisions after applying a local anesthesia (2% lidocaine). Catheters were aspirated with saline. Saline was infused into the arterial sampling catheter for the duration of the experiment. The experimental protocol is shown in Fig. 1. Animals were subjected to 150 min (t = −190 to −40 min; denoted as the exercise period) of moderate-intensity treadmill exercise (4 mph, 12% grade). At the onset of exercise (−190 min), SRIF (0.8 μg · kg⁻¹ · min⁻¹) was infused peripherally to inhibit pancreatic hormone secretion. During exercise, insulin and glucagon were infused into the portal vein to either replace basal levels of both hormones (0.2 mU · kg⁻¹ · min⁻¹ and 0.5 ng · kg⁻¹ · min⁻¹, respectively; n = 7) or simulate the fall in insulin and rise in glucagon that occurs during exercise (n = 7). The simulated infusion rates involved changes in the rates of insulin and glucagon infusion over the course of exercise. At t = −190 min, insulin was infused at 0.2 mU · kg⁻¹ · min⁻¹ and was reduced to 0.15 mU · kg⁻¹ · min⁻¹ at t = −185, where it remained for the duration of exercise. Glucagon was infused at 1.0 ng · kg⁻¹ · min⁻¹ from t = −190 to −180 min, 1.1 ng · kg⁻¹ · min⁻¹ from t = −180 to −170 min, 1.2 ng · kg⁻¹ · min⁻¹ from t = −170 to −160 min, 1.3 ng · kg⁻¹ · min⁻¹ from t = −160 to −100 min, 1.75 ng · kg⁻¹ · min⁻¹ from t = −100 to −40 min, and 0.5 ng · kg⁻¹ · min⁻¹ from −40 to 0 min. Variable peripheral glucose was infused as needed to maintain euglycemia. At t = −120 min, peripheral infusions of indocyanine green (ICG), [3-³H]glucose, and l-[¹⁴C]glucose were initiated. ICG was infused as an alternative means to calculate splanchnic blood flow if the Transonic flow probes failed during a given experiment. ICG-determined blood flow was required in three experiments. [3-³H]glucose was used to calculate net hepatic glycogen synthesized as described in Calculations. The ICG and [3-³H]glucose infusions continued for the duration of the experiment. At the end of exercise (t = −40 min), dogs were transferred to a Pavlov harness and insulin and glucagon hormone infusions were returned to basal levels. From t = 0 to 150 min (noted as the experimental period in Fig. 1), glucagon, insulin, and glucose were infused into the portal vein. Glucagon was infused (0.5 ng · kg⁻¹ · min⁻¹) to maintain basal circulating levels. Insulin was infused (1.2 mU · kg⁻¹ · min⁻¹) to establish a mild physiological hyperinsulinemia, and glucose was infused to induce hyperglycemia. At t = 0 min, the peripheral l-[¹⁴C]glucose infusion was terminated. l-[¹⁴C]glucose was added to the portal glucose infusion to maintain the rate of peripheral l-[¹⁴C]glucose infusion seen from t = −120 to 0 min. Because l-[¹⁴C]glucose is not taken up by the liver, the measurement of circulating ¹⁴C radioactivity during the portal vein infusion serves as an indicator of the efficiency with which glucose infused into the portal vein mixes with portal vein blood. The determination of mixing is further described in Calculations. After these infusions were initiated, 100 min was allowed for the establishment of steady-state conditions (i.e., constant arterial glucose concentrations). Arterial, portal vein, and hepatic vein blood samples were taken during exercise for the assessment of hepatic glucose balance and circulating hormone levels. Blood samples were also taken in 15-min intervals during the baseline period (t = −30 to 0 min) and every 10 min during the steady-state period (t = 100–150 min) for the assessment of substrate concentrations, circulating hormones, and radioactivity. Blood samples were taken at 50-min intervals during the exercise period. Portal vein and hepatic artery blood flow was recorded when blood samples were taken. After the final blood sample, dogs were killed with sodium pentobarbital. Liver biopsies were quickly frozen in liquid nitrogen. After biopsies were collected, catheter and flow probe placement was confirmed.

Data from exercised dogs were compared with a sedentary control group (n = 7, 6 of which have been previously published [2]). The sedentary dogs were studied using the identical protocol as exercised dogs except they remained sedentary in the harness for a duration equivalent to the exercise period. Because animals were not exercised, SRIF was not infused until t = 0 min.

Plasma glucose concentrations were determined by the glucose oxidase method using a Beckman Glucose Analyzer II (Beckman Instruments, Fullerton, CA) on the day of the experiment. Plasma samples, which where not immediately analyzed, were stored at −70°C for later analysis. Plasma insulin, glucagon, and cortisol concentrations were determined as described previously by radioimmunoassay (9,10). All antibodies and human standards were obtained from Linco Research (St. Charles, MO). Plasma epinephrine and norepinephrine concentrations were determined by high-performance liquid chromatography as previously described (11). Plasma free fatty acid concentrations were determined by enzymatic analysis using a kit from Wako Chemicals (Richmond, VA).

For the determination of radioactivity, blood samples were deproteinized with barium hydroxide and zinc sulfate. After centrifugation (3,000g, 30 min, 4°C), the supernatant was dried and reconstituted in 1 ml water and 10 ml scintillent (Ultima Gold; Packard, Meriden, CT). Radioactivity was determined using a Packard TRI-CARB 2900TR liquid scintillation counter. Blood concentrations of glucose, lactate, glycerol, and alanine were determined by enzymatic analysis after deproteinization (1 ml blood with 3 ml of 4% perchloric acid).

Liver glycogen was determined as previously described by Chan and Exton (12). For the determination of liver glucose-6-phosphate and fructose-6-phosphate, liver samples were homogenized in 8% perchloric acid, and the supernatant was collected after centrifugation (3,000g, 30 min, 4°C). Liver glycogen (after hydrolysis to glucose by amyloglucosidase), glucose-6-phosphate, and fructose-6-phosphate were determined by enzymatic analysis on a Technicon Auto-Analyzer (13).

Net hepatic substrate balance (NHB) was calculated as ([H] –[A]) × HAF + ([H] –[P]) × PVF, where [H], [A], and [P] are substrate concentrations in the hepatic vein, artery, and portal vein, respectively. HAF and PVF are the hepatic artery and portal vein blood flows, respectively. Positive values represent a net hepatic substrate output, whereas negative values indicate net hepatic substrate uptake. For the purposes of this article, net hepatic glucose output (NHGO) and NHGU are represented as positive values. Hepatic glucose load (HGL) was calculated as [A] × HAF + [P] × PVF. Net hepatic glucose fractional extraction was calculated as NHGU/HGL.

Mixing of infused glucose in the portal vein was assessed by the appearance of l-[¹⁴C]glucose. This method is the same in principle to the method described previously for para-aminohippuric acid (14). Like para-aminohippuric acid, l-[¹⁴C]glucose is not taken up by the liver. A baseline level of radioactivity was established before the experimental period with a peripheral l-[¹⁴C]glucose infusion. During the experimental period, l-[¹⁴C]glucose was infused with cold d-glucose into the portal vein. l-[¹⁴C]glucose appearance was assessed by the difference in arterial and portal vein radioactivity multiplied by PVF at a given time point. Time points were considered mixed when the appearance of l-[¹⁴C]glucose was within ±30% of the l-[¹⁴C]glucose infusion rate. For a given experiment to be included herein, at least half of the time points during the experimental period had to meet the mixing criteria. Based on this inclusion criteria, all data from two dogs were excluded. These dogs are not included in the subject number.

The net amount of glycogen synthesized during the experiment was calculated as the amount of radioactivity from [3-³H]glucose per milligram tissue divided by the inflowing specific activity. The specific activity of inflowing glucose was calculated as (SA_A × HAF + SA_P × PVF)/(PVF + HAF), where SA_A and SA_P are the specific activities of glucose in the artery and portal vein. The use of [3-³H]glucose for assessment of glycogen synthesis only accounts for glucose incorporated into glycogen via the direct pathway (glucose → UDP-glucose → glycogen) and assumes that minimal [3-³H]glucose is stored as glycogen before t = 0 min (15).

Statistical analyses were performed using Sigma Plot 8.0 software. Two-way ANOVA and two-way repeated-measures ANOVA were used for comparisons between groups when appropriate. P values <0.05 were considered significant.

During the exercise period, arterial plasma insulin levels (Fig. 2A) were maintained at basal concentrations in the group that received basal hormone infusions during exercise (EX-Basal) and showed a small fall in the group that received simulated hormone infusions during exercise (EX-Sim) (P > 0.05, NS). Arterial plasma glucagon was maintained at basal levels in EX-Basal, whereas EX-Sim had a marked increase over the course of exercise (Fig. 2B). It is impossible to precisely recreate the normal response of the endocrine pancreas. However, the simulated hormone infusion in EX-Sim resulted in hepatic glucose output consistent with a normal hormonal exercise response. As a result of the basal hormone clamp, NHGO did not change from values seen before exercise (t = −190 min; Fig. 3A). However, in EX-Sim, NHGO was increased approximately twofold during exercise. In both exercised groups, euglycemia was maintained with a variable peripheral glucose infusion (Fig. 3B). Over the first 90 min of the exercise period, the glucose infusion rate (GIR) required to maintain euglycemia was 2.7 ± 0.4 in EX-Basal and 0.5 ± 0.5 in EX-Sim. It should be noted however that peripheral glucose was only required in one dog in EX-Sim during the first 90 min of exercise. Over the final 60 min of exercise, the GIR required for euglycemia was 5.9 ± 0.6 in EX-Basal and 2.0 ± 0.7 in EX-Sim.

Hormone infusions were either returned to (EX-Sim) or maintained at (EX-Basal) basal levels after exercise. During the baseline period after exercise, arterial plasma insulin (Fig. 4A) was decreased in EX-Basal and EX-Sim compared with the sedentary group. During the experimental period, arterial plasma insulin rose ∼2.5-fold in all groups. Arterial plasma glucagon, epinephrine, and norepinephrine concentrations during the baseline period after exercise or rest were similar in all groups and were not different during the experimental period (Table 1). Arterial plasma cortisol concentrations were elevated during the baseline period after exercise in the EX-Sim group but not EX-Basal group compared with the sedentary control group. During the experimental period, cortisol levels returned to basal and were similar to the sedentary group.

During the baseline period, arterial blood glucose levels (Fig. 4B) and HGL (Fig. 4C) were similar in sedentary and exercised groups. During the experimental period, arterial blood glucose and HGL increased by ∼80% in all groups. The GIR required to maintain the hyperglycemic clamp was significantly higher in both the EX-Basal (22.1 ± 1.8 mg · kg⁻¹ · min⁻¹) and EX-Sim (22.6 ± 1.6 mg · kg⁻¹ · min⁻¹) groups compared with the sedentary group (10.8 ± 1.3 mg · kg⁻¹ · min⁻¹).

During the baseline period after exercise or rest, the liver remained a net producer of glucose. NHGO was similar in EX-Basal (1.1 ± 0.1 mg · kg⁻¹ · min⁻¹) and sedentary (1.3 ± 0.2 mg · kg⁻¹ · min⁻¹) dogs and significantly higher in EX-Sim (2.1 ± 0.3 mg · kg⁻¹ · min⁻¹). The higher NHGO in EX-Sim is consistent with previous work showing that NHGO remains elevated in dogs after the cessation of exercise (1). During the experimental period, net hepatic glucose balance shifted from output, seen during the baseline period, to net uptake of glucose (Fig. 5A). NHGU was significantly increased in both EX-Basal (4.0 ± 0.7 mg · kg⁻¹ · min⁻¹) and EX-Sim (4.6 ± 0.5 mg · kg⁻¹ · min⁻¹) dogs compared with sedentary dogs (2.0 ± 0.3 mg · kg⁻¹ · min⁻¹). Net hepatic glucose fractional extraction showed a pattern identical to that of NHGU (Fig. 5B).

Arterial concentrations of lactate, alanine, glycerol, and nonesterified fatty acids (NEFAs) are shown in Table 2. Arterial lactate and alanine levels were not significantly different between groups or periods. Arterial glycerol and NEFA levels were similar between groups during the baseline period. During the experimental period, arterial glycerol and NEFA levels fell to a similar extent in all groups. Neither NHB of lactate or alanine was significantly different between groups (Table 3). NHB of glycerol and NEFAs were similar between groups during the baseline and experimental periods. During the experimental period, the net uptake of glycerol and NEFAs was reduced to similar rates in all groups because of a fall in their circulating concentrations.

Liver glycogen synthesis was similar in the EX-Basal and sedentary groups but was increased by ∼50% in the EX-Sim group compared with the sedentary group (Fig. 6). Glucose-6-phosphate levels were increased in both EX-Basal and EX-Sim dogs compared with the sedentary group (Table 4). Fructose-6-phosphate levels were similar in the EX-Basal and sedentary groups but showed a significant reduction in the EX-Sim group. The glycogen synthase activity ratios were similar in all groups. However, the activity ratio of glycogen phosphorylase was significantly reduced in the EX-Sim group compared with the sedentary group.

When the exercise-induced changes in insulin and glucagon were fixed at basal levels, NHGO remained the same as that seen before the start of exercise and was comparable to baseline NHGO in sedentary dogs. The glucose requirements of working muscle were met with a peripheral glucose infusion when pancreatic hormones were clamped at basal levels. In the exercise group that received the simulated infusion, ∼25 g glucose was mobilized from the liver (estimated by multiplying the NHGO by the duration of the exercise bout and the dog weight) over the course of exercise. In the exercise group that received the basal pancreatic hormone infusion, ∼8 g was mobilized from the liver over the course of exercise. The mobilization of glucose in the basal pancreatic hormone group was equal to that seen in dogs that remained sedentary. The levels of NHGO seen in EX-Sim dogs (∼8 mg · kg⁻¹ · min⁻¹) were identical to NHGO in previous work where artificial hormonal manipulation was not used in dogs exercising at a similar intensity (4,5,16), and the increment in NHGO from basal was consistent with increases in the rate of whole-body glucose appearance previously seen (15,17–19). The primary source of the increased glucose production during 150 min of exercise is hepatic glycogenolysis in the 18-h fasted dog (20). During the experimental period, the GIR required to maintain the glucose clamp was elevated to a similar rate in both exercise groups regardless of the hormonal manipulation performed during the exercise bout. This increase in the GIR in both exercise groups indicates that the whole-body insulin sensitization that normally occurs after exercise is intact, even when insulin and glucagon do not change during exercise. Under both exercise conditions as well as the sedentary condition, NHGU accounted for ∼20% of whole-body glucose uptake during the hyperinsulinemic-hyperglycemic clamp.

The work presented here shows that insulin-stimulated NHGU is increased after exercise, even when the rise in glucagon and fall in insulin are attenuated during the exercise period. Although NHGU increased in both exercise groups, preventing the hormone changes during exercise attenuated the increase in glycogen synthesis after exercise. Previous work has shown that exercise enhances insulin-stimulated NHGU, but the enhancement in glycogen synthesis after exercise is not due to an enhancement in insulin stimulation (2). These previous findings indicate that other control elements, in addition to insulin, are responsible for the postexercise enhancement in glycogen synthesis.

A key regulatory element of glycogen synthesis is glucose-6-phosphate (21). The phosphorylation of glucose by glucokinase is a key regulatory step in hepatic glycogen synthesis, since it provides substrate for the process but also allosterically increases glycogen synthase activity and inhibits glycogen phosphorylase activity (22). In this experiment, glucose-6-phosphate increased in both exercise groups. However, net glycogen synthesis was more pronounced in the exercise group that received the simulated pancreatic hormone infusion. The increased hepatic glycogen synthesis in EX-Sim occurred with a fall in fructose-6-phosphate compared with both EX-Basal and the sedentary group. This fall in fructose-6-phosphate could possibly have been a consequence of enhanced flux of glucose to hepatic glycogen assuming glycolysis was reduced reciprocally. Considering the comparable NHGU and glucose-6-phosphate levels seen in both exercise groups, it is likely that regulatory processes other than glucose and glucose-6-phosphate are responsible for the increased net hepatic glycogen synthesis seen in the exercise group that received the simulated pancreatic hormone infusion.

Previous work in cultured hepatocytes has shown that glycogen synthesis is attenuated in the presence of elevated hepatic glycogen stores (6). This finding indicates that hepatic glycogen synthesis may be in part autoregulatory. Fasting for 42 h in the dog results in the depletion of hepatic glycogen stores similar to that seen with prolonged exercise. The delivery of a glucose load following such a fast results in increased net hepatic glycogen synthesis (7). In addition to these findings, histochemical analysis of rat hepatocytes has shown that elevations in glycogen stores within the cell inhibit the normal translocation of glycogen synthase to the cell periphery in response to glucose (23). This work also showed that chemical activation of glycogen synthase is not sufficient to cause translocation of the enzyme, thus indicating that translocation and activation may have distinct methods of regulation.

In addition to glycogen playing a role in the regulation of its own synthesis, glycogen phosphorylase is also a key regulator of the synthesis and breakdown of hepatic glycogen. Metabolic control analysis has shown that glycogen phosphorylase is a strong negative regulator of net glycogen synthesis (24). Consistent with this finding is the reduction in phosphorylase activity seen in exercised dogs that received the simulated pancreatic hormone infusion. The reduction in phosphorylase activity was not seen in the exercise group that received the basal pancreatic hormone infusion.

Another potential mechanism for the enhancement of hepatic glycogen synthesis is prolonged exposure to elevated cortisol levels. The work presented here shows that after exercise, the group that received the simulated hormone infusion had a marked increase in arterial plasma cortisol, whereas the dog that received basal insulin and glucagon did not exhibit increased cortisol. This is consistent with previous work in the dog that showed the increase in cortisol during exercise requires the normal increase in glucagon (4). These separate observations suggest that glucagon or the metabolic effects of glucagon are involved in the release or clearance of cortisol. Chronic exposure to cortisol (5 days) has been shown to markedly increase glycogen storage in the conscious dog (25).

In conclusion, the pancreatic hormone changes that occur during exercise and the consequent mobilization of hepatic glycogen are not responsible for the postexercise enhancement in insulin-stimulated NHGU, but are critical to the determination of the fate of the additional glucose taken up by the liver. Thus, the pathways for NHGU and glycogen synthesis are regulated differently after exercise.

This work was funded by National Institute of Diabetes and Digestive and Kidney Diseases Grant ROI DK-50277 and Diabetes Center Grant DK-20593 and Training Grant 5-T32-DK-7563-08.

We thank Deanna Bracy for her valued assistance with the completion of this work as well as the Vanderbilt University Diabetes Center Hormone Assay Core.