Journal of Applied Physiology
Vol. 81, No. 5, pp. 2020-2026, November 1996
METABOLISM

Exercise and exhaustion effects on glycogen synthesis pathways

Hans Gunderson, Nadja Wehmeyer, Diane Burnett, John Nauman, Cynthia Hartzell, and Scott Savage

Chemistry Department, Northern Arizona University, Flagstaff, Arizona 86011-5698

ABSTRACT

INTRODUCTION

METHODS

RESULTS

DISCUSSION

ACKNOWLEDGEMENTS

FOOTNOTES

REFERENCES

ABSTRACT

Gunderson, Hans, Nadja Wehmeyer, Diane Burnett, John Nauman, Cynthia Hartzell, and Scott Savage. Exercise and exhaustion effects on glycogen synthesis pathways. J. Appl. Physiol. 81(5): 2020-2026, 1996.Female Sprague-Dawley rats were infused with [1-¹³C]glucose to measure the effect of endurance training and the effect of various metabolic conditions on pathways of hepatic glycogen synthesis. Four metabolic states [sedentary (S), trained (T), sedentary exhausted (SE), and trained exhausted (TE)] were studied. T and TE rats were trained on a motor-driven treadmill (30 m/min, 15% grade, 1.0 h/day, 5 days/wk) for 8-10 wk. After a 24-h fast, SE and TE rats were run to exhaustion (sedentary average = 78 min, trained average = 155 min) at a training pace and immediately infused with labeled glucose for 2 h. S and T rats were infused after a 24-h fast. After infusion, tissues were removed and glycogen was isolated and hydrolyzed to glucose. The glucose was measured for distribution of ¹³C by using nuclear magnetic resonance. Glycogen was synthesized predominantly by the indirect pathway for all metabolic states, indicating that infused glucose was first metabolized primarily in the peripheral tissue. The direct-pathway utilization was greater in rested S than in rested T animals (30 vs. 14%); however, for exhausted animals, the trained use of the direct pathway was greater (22 vs. 9%). Both TE and rested T animals utilize the indirect pathway a comparable amount. Sedentary animals, on the other hand, dramatically decreased utilization of the direct pathway, with exhaustive exercise changing from 30 to 9%. The results indicate that endurance training modifies glucose utilization during glycogen synthesis after fasting and exhaustive exercise.

endurance training; carbon-13-nuclear magnetic resonance; glucose utilization; glycogenesis; glycolysis; gluconeogenesis

INTRODUCTION

THE NEED OF CONTRACTING skeletal muscle for carbohydrate energy sources can be met either by intracellular glycogen stores or by blood glucose. Blood glucose levels are maintained by dietary sources, gluconeogenesis, and glycogenolysis. As one of the sources of blood glucose, liver glycogen is important in the control of cellular energy needs. During the past decade, Katz and McGarry (15), Newgard et al. (20), and others have described possible pathways of glycogen synthesis in the liver. After carbohydrate ingestion, absorbed glucose can be used as the substrate for liver glycogen synthesis or exit the liver and travel to the peripheral tissues such as the muscle. Metabolic by-products of glucose metabolism in these peripheral tissues, such as lactate, can return to the liver for use as gluconeogenic substrates. The glucose 6-phosphate (G-6-P) formed in liver gluconeogenesis either is hydrolyzed to glucose and returned to the circulatory system or is used for the synthesis of glycogen.

Until the mid-1980s, the accepted fate of dietary glucose in the glycogenesis process was to be transported from the small intestine via the portal vein to the liver. Once in the liver, the glucose would serve as the major substrate for glycogen synthesis. This direct pathway of glycogen formation from dietary glucose appears energetically efficient because the synthesis expends only 1 mol of ATP and 1 mol of UTP per mole of glucose converted to glycogen (see Fig. 1: the direct pathway) (6, 9, 18, 30). However, the primary phosphorylating enzyme in the liver, glucokinase, which catalyzes the formation of G-6-P from glucose, has a Michaelis constant of ~8 mM (9). Thus at normal blood glucose concentration (4-6 mM), phosphorylation to G-6-P in the liver will be limited, resulting in less glycogen being synthesized through this direct pathway.

Endurance training affects glycogen metabolism and results in increased glucose tolerance after a meal, indicating an increased net glucose transport into body tissues (11, 12). Endurance training also lowers insulin levels, which together with fasting would be expected to lower the level of glucokinase in the liver (22). Lowered glucokinase activity results in less formation of G-6-P and less formation of glycogen by the direct pathway (6). Studies by Brooks (1), Katz and McGarry (15), Katz et al. (16), Kurland and Pilkis (18), Magnusson and Shulman (19), and Newgard et al. (20) show that a large fraction of glucose is absorbed across the gut and passes through the liver to be taken up by the peripheral tissue. These tissues may metabolize the glucose into three carbon compounds, such as lactate, which can be transported back to the liver to serve as gluconeogenic substrates. The resulting G-6-P is then converted to glycogen, thereby bypassing the glucokinase-catalyzed phosphorylation step (see Fig. 1: the indirect pathway), allowing the dietary glucose to be used for both the present energy needs of the peripheral tissue and eventual storage for later energy requirements.

Because glycogen synthesis is affected by metabolic circumstances (i.e., feeding, fasting, exhaustion), this investigation focused on the effects of endurance training and exhaustive exercise on the pathways of glycogen synthesis when going from a fasted to a fed state. First, since trained and fasted animals will have lowered glucokinase levels (22), they will have less net glucose uptake by the liver. In addition, since trained animals have an increased glucose tolerance after a meal, glycogen synthesis by the direct pathway is expected to be less in rested endurance-trained than in rested sedentary animals. Because endurance-trained animals that have been trained at 25 m/min for 60 min/day have lower liver glycogen content than untrained animals (10), more dietary glucose absorbed into the blood may be supplied to the muscle for energetic needs. The resulting lactate produced by muscle glucose metabolism can be used to synthesize glycogen in the liver via the indirect pathway if it is required. Ultimately, the liver of the trained animal will become replenished, although over a longer period. Evidence provided in this study suggests that 1) endurance-trained animals provide infused glucose first to the peripheral tissue (such as muscle) and 2) they synthesize glycogen in the liver by using the metabolic products from anaerobic glycolysis.

The second part of this study determined the effects of exhaustive exercise of both sedentary and trained animals on the pathway of glycogen synthesis. Because an exhausted animal has a greater energetic need for dietary glucose, less glucose is predicted to be directly synthesized as hepatic glycogen. After glucose is utilized for energy production through glycolysis, the resulting lactate can be converted indirectly into glycogen. Research from several laboratories indicates that rested sedentary animals utilize the indirect pathway more than the direct pathway (14, 20, 21, 27, 28, 32). The actual percent, which the direct pathway contributes to glycogen synthesis, is related to blood glucose concentration and is extremely variable, depending on the experimental conditions. Several studies measured an ~30% direct-pathway contribution for animals under resting physiological conditions (14, 18, 21, 26-28). Sedentary exhausted animals subjected to a glucose load may utilize the direct pathway to an even lesser extent than sedentary rested animals.

The purpose of this study was to determine the pathways of hepatic glycogen synthesis in sedentary and trained rats by using ¹³C-labeled glucose and nuclear magnetic resonance (NMR) techniques. The rate of glucose infusion was controlled so as to maintain physiologically normal glucose levels in the blood (4-8 mM) in both sedentary and endurance-trained animals (1, 6). This procedure is in contrast to experiments by Huang and Veech (9) and Johnson and Bagby (13) in which blood glucose concentrations ranged upward from 11 mM. Their studies used higher infusion rates to raise the glucose concentration above a normal fasting physiological level.

METHODS

Nineteen female Sprague-Dawley rats were randomly divided into sedentary or endurance-trained groups. All the rats were housed on a 12:12-h light-dark cycle. This study was reviewed and approved by the Institutional Animal Care and Use Committee of Northern Arizona University and was conducted in conformity with the "Guiding Principles in the Care and Use of Animals" approved by the Council of The American Physiological Society. Procedures used were similar to those reported earlier (32). The rats received water and rat chow ad libitum. When the rats were 3-4 mo old (weighing 200-250 g), they were anesthetized with a 15:1 ketamine-xylazine mixture (80 mg/kg). Catheters were surgically implanted into the left carotid artery and the right jugular vein, according to the procedure of Popovic and Popovic (23). After the surgery, the animals were allowed a 3-day exercise-free recovery period after which they were infused with ¹³C-labeled glucose in a metabolic box.

Four metabolic states were investigated: 1) sedentary (S), 2) sedentary exhausted (SE), 3) endurance trained (T), and 4) endurance trained exhausted (TE). Endurance training consisted of running the animals on a rodent treadmill 30 m/min, up a 15% grade, 1 h/day, 5 days/wk, for 8-10 wk. The endurance-trained animals were acclimated to the training regime by gradually increasing their total running time from 5 to 60 min/day over a 2-wk period. All the animals were fasted for the 24 h before the glucose infusion to reduce the liver glycogen level. Both the sedentary and trained animals used in the exhaustion portion of the study were run to exhaustion on the treadmill immediately before their glucose infusion. The exhaustion experiments were conducted to reduce the glycogen concentration of the muscles. The animals in these exhausted groups were run on the treadmill at an initial pace and grade identical to the training protocol. As the animals tired and became reluctant runners, the rate was slowed to allow the animals to maintain a running pace without coming to the rear of the running surface. Control animals tired more quickly than trained animals and ran at a slightly slower average speed. At the point of exhaustion, the rats were running at 16-20 m/min. Animals were considered to be exhausted when they refused to run and did not attempt to right themselves when placed on their backs. Both groups of animals worked at similar workloads in the exhaustion study, with the endurance-trained animals running about twice as long before exhaustion.

A total of nine control animals were investigated for direct-pathway determination. Four of these animals were in the SE group and were run to exhaustion [78 min average (SE ± 2.0)], with the remaining five in the rested S group. To familiarize the SE animals with the treadmill, they were run 2-3 min/day for 1 wk before the glucose infusion. Ten animals were endurance trained; six of these were in the TE group, and the remaining four were in the T rested group. Exhaustion for the endurance-trained animals was achieved after 2-3 h of running on the treadmill [155 min average (SE ± 3.2 min)], a significantly longer period than the S animals were able to run.

For the infusion, the animals were placed in a metabolic box similar to that used by Savage et al. (24). The animals were familiarized with the metabolic box for several days before an experiment and were not naive or afraid of it. The box made it possible to monitor O₂ and CO₂. A 25% [1-¹³C]glucose solution was continuously infused at a rate of 9.3 µl/min for 120 min through a venous catheter. This rate simulated a glucose load of ~1.5 times the normal endogenous appearance rate in a fed resting rat. Blood samples were taken every 30 min through the arterial catheter. Blood glucose concentrations during infusion were similar to normal physiological resting levels of 4-8 mM (1, 6). After the 2-h infusion period, the animals were anesthetized with pentobarbital sodium. Subsequently, the liver tissues were removed, freeze-clamped in liquid nitrogen, and stored at 80°C until the time of the assay. Liver tissues were homogenized in hot 30% potassium hydroxide, and the glycogen was precipitated with ethanol (5, 25). The isolated glycogen was purified by redissolving in 10% trichloroacetic acid followed by precipitating a second time with ethanol (25, 29). The glycogen was then enzymatically hydrolyzed to glucose with amyloglucosidase (Sigma Chemical). The glucose from the hydrolyzed glycogen as well as from blood was analyzed by using glucose oxidase from the Sigma Chemical kit no. 510 and measuring the absorbance of the oxidized dye at 442 nm. The blood lactate concentration was determined by using the Sigma Chemical reagents and procedure no. 826-UV. Lactate values relate to the change in absorbance at 340 nm.

A 200-MHz Gemini NMR spectrometer (Varian, Palo Alto, CA) was used to measure the distribution of the ¹³C label in the glucose molecule. The samples were dissolved in 0.7 ml D₂O (15 µl of a 1% solution of sodium azide were added as a preservative) and placed in standard 5-mm NMR tubes. The nuclear Overhauser effect-enhanced glucose spectra were obtained at 50.3 MHz using ¹H decoupling during acquisitions. Chemical shifts for ¹³C were measured relative to tetramethyl silane. A 90° pulse was repeated every second for 2,000-8,000 scans, depending on the quantity of glucose from the hydrolyzed liver glycogen. A glucose spectrum displays 12 peaks (2 peaks for each carbon) due to the C₁-- and -anomers (32). However, in a typical spectrum, only 11 peaks are resolved. The two anomers of carbon four have a similar magnetic environment and their peaks completely overlap. The sum of the label on carbon four is equal to the height of the peak found at ~2 parts/million (ppm).

Each carbon was corrected to carbon one for nuclear Overhauser effects. The correction factors, based on natural-abundance glucose hydrolyzed from liver glycogen, are measured to be

Natural-abundance glucose measurements were attained by hydrolyzing a 0.5-M sample of unlabeled glycogen to glucose. Unlabeled hydrolyzed glucose, run at experimental concentrations and parameters, resulted in ¹³C-NMR spectra with negligible peak height, and thus no correction for natural abundance was needed (see Ref. 32 for spectra).

Percentage direct-pathway calculation is similar to the one in Shulman et al. (28) in that the label on carbon six is subtracted from the label on carbon one. As a result of the indirect pathway, the amount of ¹³C label on carbon one is equal to the label on carbon six. The label on carbon six is subtracted from carbon one for calculation of the direct pathway by using the following equation Distribution of the ¹³C label onto carbons other than carbon one and carbon 6 occurs because of equilibration of the gluconeogenic intermediate, oxaloacetate, in the citric acid cycle (32).

Shulman et al. (28) administered the [1-¹³C]glucose via a large gastric bolus and then used the amount of ¹³C label on glucose from the portal vein as the value in the denominator. Katz et al. (16) describe procedures for calculating the percentage direct pathway by using animals that had been infused with uniformly labeled [¹³C]glucose. This method of Katz et al. measures the ¹³C distribution in both blood glucose and glycogen for calculation of the direct pathway. In our investigation, C₁-labeled glucose was infused continuously for 2 h at a rate that maintained physiological glucose concentrations of 4-8 mM. This experimental design causes the blood level of ¹³C in position one of glucose to be very high in such a way that the natural-abundance ¹³C at any other position cannot be detected at these concentrations. Because of the very high amount of infused [1-¹³C]glucose in the blood, the distribution of the natural-abundance ¹³C on carbons two through six of blood glucose was insignificant compared with the label found on carbon one. Experimental animals were depleted in liver glycogen by fasting and were infused with glucose at a rate that resulted in normal blood glucose concentrations. Under these conditions and over this short time period, glycogen would be expected to be synthesized and not degraded. Thus the total label incorporated in the liver glycogen was used as the denominator, and the percentage of glycogen synthesized via the direct pathway was calculated by the equation given above.

To verify that natural-abundance ¹³C glucose did not contribute to the peaks in our NMR spectra, samples were taken from animals infused with unlabeled glucose. Hydrolyzed liver glycogen samples from the infused animals yielded a glucose concentration that was <5 mM. Solutions were prepared by using an amount of unlabeled liver glycogen sufficient to achieve a glucose concentration of 5 mM. These unlabeled glycogen solutions were prepared through the same isolation procedures as the liver samples and then hydrolyzed to glucose by using amyloglucosidase to determine peak heights of natural-abundance ¹³C in glucose. The NMR spectra of the solutions of unlabeled material showed negligible peak heights for the glucose natural-abundance ¹³C carbons, when acquired using the same number of scans and experimental parameters as were used for the spectra of solutions from labeled animals. This result indicates that the ¹³C peaks observed for the NMR spectra of the experimental animals were from the infused labeled glucose and not from the natural-abundance ¹³C (32).

RESULTS

Direct-pathway utilization in the formation of hepatic glycogen was twice as large in rested S animals compared with rested T animals, 30 vs. 14%, respectively (Table 1). Statistical analysis was applied by F-tests and t-tests to determine significance of the means between the four groups studied. Results of the t-tests indicate that the percentage of direct pathway of the groups, S vs. T, SE vs. TE, and S vs. SE had significantly different means (P < 0.05), whereas the comparison of T vs. TE showed they had significantly similar means.

Table 1. %Direct pathway of glycogen synthesis in liver for four conditions Animal %Direct Pathway Sedentary 30 ± 2.5  Trained 15 ± 1.7  Sedentary exhausted 9 ± 1.3  Trained exhausted 22 ± 3.0 Values are means ± SE.

Figure 2, A and B, shows ¹³C-NMR spectra of S and T rat liver glycogen hydrolyzed to glucose. The greater carbon one peak height (98 ppm, C₁-; 94 ppm, C₁-) seen for the S compared with the T group can be explained by the more prominent direct-pathway utilization.

Figure 2, C and D, shows ¹³C-NMR spectra representative of SE and TE animals that were exhausted before the glucose infusion. In this case, muscle glycogen stores should be low or depleted. The T animals were able to run about twice as long as the S group (155 ± 3.2 vs. 78 ± 2.0 min). Because these exhausted animals were fasted and run to exhaustion, their glycogen levels were diminished and endogenous glucose sources were limited (see Table 2). Thus dietary (or infused) glucose became useful for energy needs and glycogen synthesis. An opposite tendency was observed for these exhausted animals compared with the rested animals in that a greater percentage of liver glycogen formed via the direct pathway was observed in the TE animals compared with the SE animals (Table 1). The SE animals had much lower (P < 0.05) direct-pathway glycogen than their rested sedentary counterparts (9 vs. 30%) and the TE animals (9 vs. 22%).

Table 2. Liver and muscle glycogen concentrations Sample Source Liver [Glycogen], mg/g White Gastroc Red Gastroc 24-h fasted*   Trained 4.42 ± 0.79  2.99 ± 0.53   Sedentary 5.05 ± 1.02  3.97 ± 0.57 Postinfusion   Trained 17.22 ± 1.54  11.42 ± 1.16  15.03 ± 2.44    Sedentary 21.05 ± 2.12  13.42 ± 1.13  13.51 ± 1.33    Trained exhausted 16.79 ± 1.45  11.57 ± 1.29  10.50 ± 1.35    Sedentary exhausted 15.10 ± 1.35  14.60 ± 0.97  14.56 ± 2.09 Values are means ± SE. Brackets indicate concentration. Gastroc, gastrocnemius. * These animals were not used in the infusion experiments. Muscle samples taken immediately after 24-h fast and before the infusion were not divided into red and white portions.

Total glycogen levels in the liver after the 24-h fast and subsequent to the 2-h glucose infusion are given in Table 2. The highest concentration was found in the rested S animals, which have the highest level of direct pathway. The lowest concentration of glycogen in the liver was found in the SE animals, which also had the lowest level of direct-pathway synthesized glycogen. Glycogen concentration in the liver was similar for both T and TE animals. All animals were fasted for 24 h before the glucose infusion to diminish hepatic glycogen levels. Gastrocnemius muscle glycogen levels were determined immediately after the 24-h fast and subsequent to the 120-min infusion period and are given in Table 2. Glycogen levels are similar for T and S groups, whereas SE animals have slightly higher levels than TE animals. All groups of animals studied showed a net synthesis of glycogen during the infusion period after the 24-h fast or 24-h fast and exhaustive exercise period.

DISCUSSION

Results from the four metabolic states studied show that glycogen is synthesized primarily by the indirect pathway when blood glucose concentrations are 4-8 mM (Tables 1 and 2). These results are in accord with numerous studies that show the indirect pathway to predominate especially at lower glucose concentrations (14, 18, 21, 26-28). Consequently, most of the glucose is passed through the liver to the peripheral tissue for glycolytic energy production and glycogen resynthesis. Under the conditions of the study (reduced muscle and hepatic glycogen), the results strongly suggest that the differences between groups of animals reflect different metabolic responses as a result of endurance training.

Rested endurance-trained animals pass glucose through the liver to the peripheral tissues more readily than do rested sedentary animals, resulting in an increased indirect-pathway utilization. Because the rested T group utilizes the direct pathway only about one-half as often (14 vs. 30%) as the S animals, it appears that endurance exercise facilitates a shift in the use of the infused glucose that may be metabolically advantageous: that of glucose delivery to the peripheral tissue such as skeletal muscle. By giving peripheral tissue the primary opportunity for glucose utilization, these energy and fuel demands can be fulfilled before liver glycogen is replenished. The needs of the liver for glycogen can then be secondarily met from the glycolytic end products via gluconeogenesis and indirect-pathway reactions. When exhausted, the T animals exhibit a lower percentage of hepatic glycogen formed via the indirect pathway than do the S animals. Efficientusage of peripheral glucose by the T animals evidently results in little perturbation of the percent indirect pathway after exhaustive exercise. This is consistent with the results of this study as shown by the similar direct-pathway values for T and TE.

The increased utilization of the indirect pathway by SE compared with rested S animals is consistent with studies reported in the literature (4, 7, 8, 17, 34). Zachwieja et al. (34) investigated the effects of glycogen depletion and the rate of muscle glycogen resynthesis after refeeding and concluded that the rate of resynthesis increased when there was a greater depletion of glycogen from the tissue. Exhaustive exercise would lower the glycogen levels in the muscle as well as the fasted liver, so infused glucose may first be utilized to fulfill the glycogenesis and energetic demands of the muscle tissue. Steady-state blood lactate concentrations were similar and relatively low for SE and TE animals (Table 3), indicating that exhaustion for both groups was likely not a result of anaerobic work.

Table 3. Blood glucose and lactate concentrations Experimental Group* [Lactate], mM [Glucose], mM Initial Steady state Initial Steady state Sedentary 2.4 ± 0.7  1.3 ± 0.2  3.7 ± 0.2  4.6 ± 0.2  Trained 1.9 ± 0.5  2.1 ± 0.5  6.4 ± 0.6  8.3 ± 0.7  Sedentary exhausted 2.1 ± 0.3  1.6 ± 0.3  2.4 ± 0.1  4.9 ± 0.3  Trained exhausted 1.4 ± 0.1  1.4 ± 0.1  2.6 ± 0.2  6.8 ± 0.4 Values are means ± SE. * These experiments were identical to experiments reported here using 13C-infused animals. Values are given for initial and the mean of steady state attained from 90 to 120 min of infusion in experiments where glucose was not labeled with 13C.

The results of training may include differences in peripheral tissue oxidative metabolism in which SE animals convert a larger amount of glucose to lactate. Duan and Winder (3) found lower levels of blood lactate in trained animals after 30 min of running. This would result in lower hepatic glycogen production via the indirect pathway. Although trained exhausted animals may also be using a large amount of infused glucose for an energy source in muscle, their endurance training enables them to continue the aerobic oxidation of glucose.

Endurance-trained animals have a higher serum concentration of free fatty acids (7, 17, 31). Human studies have shown an increased concentration in the blood and greater uptake and utilization of free fatty acids by muscle tissue of trained subjects (31). Endurance training has resulted in animals that are capable of more readily using fat oxidation for energetic needs than are their sedentary counterparts (4, 7, 8). Thus the exhausted endurance-trained animals may be burning more fat as an energy source than are sedentary exhausted animals. The exhausted endurance-trained animals' requirements for glucose may be diminished by using a higher percentage of fat than do sedentary exhausted animals. The "excess" infused glucose could then be stored as glycogen via the direct pathway.

In conclusion, glycogen is synthesized primarily by the indirect pathway for all the metabolic states studied, thus allowing infused glucose to be first metabolized in the peripheral tissue. Sedentary exhausted animals utilize the infused glucose first in the peripheral tissue, displaying a greater indirect-pathway contribution compared with exhausted endurance-trained animals. Exhausted endurance-trained animals may have a more complete oxidation of glucose and a higher reliance on fat for energy, allowing more of the infused glucose to be used directly for the repletion of glycogen in the depleted liver. In the exhausted state, glucose is more likely to be completely oxidized (i.e., citric acid cycle) for energetic purposes in peripheral tissue, resulting in greater repletion of hepatic glycogen by the direct pathway for trained exhausted animals. Sedentary exhausted animals may not have the ability to utilize aerobic glycolysis and continued oxidation in the citric acid cycle as much as trained exhausted animals. Instead, sedentary exhausted animals may utilize anaerobic glycolysis and export the lactate out of the peripheral tissue to the liver for use in glycogen synthesis. The use of the indirect-pathway remains virtually the same for trained exhausted and trained rats, suggesting that endurance training helps compensate for the effects of exhaustive exercise.

ACKNOWLEDGEMENTS

The authors thank Michael Egan and Dale Woolridge for assistance with the animal surgeries; Dr. David Arnall, Carmen Begay, and Kirsten Beerling for technical assistance; Drs. David Arnall and Stan Linstedt for critically reviewing the manuscript; and Sue Mertz for assistance in manuscript preparation.

FOOTNOTES

Address for reprint requests: H. M. Gunderson, Chemistry Dept., Box 5698, Northern Arizona Univ., Flagstaff, AZ 86011-5698.

Received 25 July 1995; accepted in final form 24 June 1996.

REFERENCES