Mechanism of Control of Hepatic Glycogenesis by Insulin*

SUMMARY Insulin produces an increase in the conversion of the D form (glucose-6-P dependent) of liver glycogen synthase to the Z form (glucose-6-P independent) when injected intra-venously into fed alloxan diabetic rats. Insulin activates glycogen synthase previously inactivated by glucagon in isolated perfused livers from normal rats, while increasing [14C]glucose incorporation into glycogen and decreasing glucagon-stimulated glucose production. This change in synthase activity is associated with a decrease in synthase (protein) kinase activity and tissue adenosine 3’,5’-mono-phosphate levels when insulin is used to antagonize the effects of glucagon. Finally, insulin synthase (D to Z shift) in perfused livers from normal rats if insulin is begun after an initial 30-min perfusion is maximal between min and is no longer present after 30 min. Associated changes include an

Insulin produces an increase in the conversion of the D form (glucose-6-P dependent) of liver glycogen synthase to the Z form (glucose-6-P independent) when injected intravenously into fed alloxan diabetic rats. Insulin activates glycogen synthase previously inactivated by glucagon in isolated perfused livers from normal rats, while increasing [14C]glucose incorporation into glycogen and decreasing glucagon-stimulated glucose production. This change in synthase activity is associated with a decrease in synthase (protein) kinase activity and tissue adenosine 3',5'-monophosphate levels when insulin is used to antagonize the effects of glucagon.
Finally, insulin alone has a direct effect on the activation of glycogen synthase (D to Z shift) in perfused livers from normal rats if insulin infusion is begun after an initial 30-min perfusion period.
This effect is maximal between 6 and 15 min and is no longer present after 30 min.
Associated changes include an increase in ['"Clglucose incorporation into glycogen at 10 min and a decrease in synthase kinase activity at 6 and 15 min.
This decrease in kinase activity is not associated with a decrease in apparent hepatic concentrations of adenosine 3',5'-monophosphate.
The enzymatic pathway involved in glycogen synthesis has been characterized in most mammalian tissues. Glycogen synthesis in liver is at least partially controlled by the activity of glycogen synthase (l), the enzyme responsible for catalyzing the incorporation of glucose from uridine diphosphoglucose into glycogen (2). The active form of synthase (I or glucose-6-P independent) is converted to the inactive form of synthase (D or glucose-6-P dependent) by an ATP requiring phosphorylation reaction catalyzed by active synthase (protein) kinase (3). The activity of kinase is in turn regulated at least in part by alterations in tissue levels of cyclic AMP (4).' Therefore, any hormone or factor affecting cyclic AMP concentration can alter synthase activity through an effect on the kinase. On the other hand, the inactive form of synthase is converted to the active form by a dephosphorylation reaction catalyzed by synthase phosphatase (5). Current evidence suggest,s that the phosphatase may also exist in two forms, thus providing an additional site for control of glycogen synthesis (5).
The effect of insulin to promote the activation of glycogen synthase (D to I shift) in the liver in vivo has been well documented (6)(7)(8)(9).
Many workers have reported interactions of insulin with glucagon, epinephrine, and cyclic AMP in the perfused rat liver (10-15).
The action of insulin to promote glycogen synthasc activation has been well characterized in diaphragm (16-l@, skeletal muscle in viva (19,20), heart in vivo (20,21) and in vitro (22), fat pad in vitro (23), human placenta in vitro (24), and tumor cells in vitro (25). From these reports it could be inferred that glycogen synthesis in the liver is at least partially dependent on the action of insulin to activate glycogen synthase.
However, the question of whether or not insulin actually enhances hepatic glycogenesis has remained controversial due to an absence of evidence obtained in vitro.
The purpose of the present study was to esplore the extent to which insulin mediates hepatic glycogen synthesis.
We have studied the action of insulin to alter glycogen synthase from a basal state in diabetic rat liver in vivo and the antagonism between the effects of glucagon and insulin on glycogen synthase in livers of normal rats in vitro. Finally, we describe a set of conditions whereby insulin promotes glycogen synthasc activation from a basal state (absence of glucagon) in livers of normal rats perfused in vitro.

MATERIALS AND METHODS
Male Wistar rats fed ad Z&turn with Purina laboratory chow and weighing from 100 to 150 g were used after anesthetization with sodium pentobarbital (50 mg per kg). Esperimental diabetes was induced where indicated by intravenous injection of alloxan (60 mg per kg) and was diagnosed 2 days later by a maximal positive urine sugar test using "TesTape." For experiments in vivo, 4%hour alloxan diabetic rats were anesthetized as previously described and both femoral triangles were exposed. The animals' tongues were secured to facilitate breathing and body temperature was maintained with heat from loo-watt light bulbs. Blood samples were taken from the femoral veins prior to the injection of either insulin or 0.9% NaCl solution-albumin.
The livers were removed at the indicated times by abdominal incision.
Livers were analyzed immediately after surgical excision.

3484
The technique of liver perfusion (26) and the apparatus used for the procedure have been described in detail by Exton and Park (27 The significance of differences between means was established by Student's t test.

RESULTS
-70". hliquots of frozen liver powder were analyzed for glyco-Optimal conditions for demonstrating an effect of insulin on gen (29) and the hydrolyzed glucose determined by the alkaline hepatic glycogen synthesis appear to require a slight insulin ferricyanide method using the Technicon Auto-Analyzer. deficiency.
Since an acute insulin deficiency develops 48 hours [14C]Glycogcn was determined as previously described (30). after a single inject'ion of alloxan, fed alloxan diabetic animals Glycogcn synthasc was e&acted from fresh liver or frozen liver were used. After 15 min, an initial blood sample was withpolvtler with 100 mM potassium fluoride-10 rrl~ EDTA, pH 7.8 drawn from the femoral vein and the animals were then in-(w/v, 1 :5) and assayed without added sulfate using the method jected with 0.2 ml of either insulin (2 units) or 0.9% NaCl soluof Thomas et al. (31). Data were expressed as per cent 1 (action-albumin. ItI the first series of animals, the livers were tive) form.
IUood glucose was tlctermincd as described above, excised 5 min after itlsulin injection, and in t.he second series ant1 tissue protein was determined by the Lowry procedure using after 15 min. Final blood samples were taken 15 min after thr Tcchnicon Auto-Analyzer.
injection. Glycogen synthase kinase Iv-as extracted from frozen liver powder with 5 mM phosphate-2 mM EDTA (w/v, 1:3) at pH 7. After centrifugation at 12,000 X g for 15 min, the supernatants were passed rapidly (10 min total time) through 2 Sephadex G-50 columns (1 cm X 15 cm), all at 4". The first column was equilibrated with the estraction buffer and the second with 50 mnr Tris-5 mM EDTA at pH 7.8. Aliquots of the protein eluate from the second column were assayed for synthase kinase ac-tivityZ using added synthase 1 as substrate for the kinase in the absence and presence of cyclic AMP as described by Shen et a/. (32). Kinase activity determined in the absence of cyclic AMP was expressed as the active or cyclic AMP independent form (I form), and that determined in the presence of added cyclic ,411l' was expressed as total kinase or cyclic AMP independent plus dependent forms (1 plus D forms).
Final data are expressed as per cent I kinase or nanomoles of glucose incorporated into glycogen per mg of protein per 5 min assay time (or both).
After 5 min of insulin exposure, glycogen synthase was increased from a control value of 217; 1 synthase to an insulin treated value of 38y0 (Table I). The same effect was observed after the 15.min period where a control value of 37, Z synthase was increased to 14% by insulin.
As seen in this table, insulin also produced a decrease in blood glucose of 72 mg/lOO ml, while the control showed no significant change. The difference in absolute levels of glycogen synthase activity between the 5and 15.min csperiments was due to a difference in extraction procedures.
The first series (5 min) was extracted in phosphate-EDTA buffer rather than the usual potassium fluoride-EDTA. The phosphate-EDTX buffer consistently increases apparent synthase 1 activity in liver homogenates since there is no fluoride to inhibit synthase phosphatase.
Table I thus demonstrates that insulin does activate hepatic glycogen synthase in alloxan diabetic rats. This agrees with previous data obtained in viva (6-9, 34). Cyclic AMP was extracted from frozen liver powder with 5% trichloroacetic acid (100 mg tissue per ml) containing tracer amounts of the tritiated nucleotide to follow recovery.
After homogenization-sonication (Y$ speed, 20 s, Polytron homogenizer), the mixture was centrifuged at 2,000 x g for 5 min. The trichloroacetic acid was removed from the supernatant with three extractions of 5 volumes of anhydrous ether. One milliliter of the sample was then passed over a Dowex 50 column (6 mm x 5 cm, Dowex 50-X8, 200 to 400 mesh, H+ form) and the cyclic AMP eluted with 1120 in the 3-to 6-ml fraction.
After lyoplrilization, the cyclic AMP was dissolved in 0.5 ml of 100 mM sodium acetate, pH 4, and determined using the binding assay of Gilman (33).
The activation of hepatic glycogen synthase by insulin in viva does not show whether or not this is a direct action of the hormone on the liver or merely an indirect effect mediated by another effector system. Physiologically, insulin appears to act to antagonize the effects of glucagon since it has been well documented that insulin counteracts some of the actions of glucagon in the perfused rat liver (10-15).
The next series of experiments 2 This assay depends on the action of the kinase to catalyze the phosphorylation and thus inactivation of added synthase I. Extracts of the frozen tissue are passed over Sephadex columns prior to assay to remove contaminating nucleotides and other small molecules. This procedure has been shown to remove virtually all endogenous cyclic AMP from the extract without producing an absolutely cyclic AMP dependent or independent protein kinase (46). Therefore, it appears that column treatment of the extract does not affect the actual state of activity of the kinase effected by cyclic AMP prior to column treatment.
Previous studies indicate that hormonal effects on the kinase prior to or after column treatment remain relatively stable or unchanged throughout the course of the assay (30, 32 was designed to determine if the antagonism extended to control of synthase activity. Livers from normal fed rats were perfused for 2 hours in a recirculating system. The 1st hour of perfusion was carried out without substrate or hormone to produce a state of relative insulin deficiency. At the beginning of the 2nd hour, 1.0 PCi of [i%]glucose was injected into both recirculation reservoirs. Glucagon was infused at 0.25 pmoles per min into one reservoir and glucagon plus insulin (insulin at 5 munits per min) into the other reservoir.
Infusions were continued through the 2nd hour. Blood perfusate samples were taken every 20 min throughout the perfusion.
Glucagon produced an increase in the rate of perfusate glucose accumulation (Fig. 1). In agreement with previous reports (11,12,35,36), insulin partially inhibited the glucagonstimulated increase in glucose accumulation. Table II shows the effects of glucagon plus and minus insulin on [14C]glucose incorporation into glycogen in the same livers. Approximately 800 cpm per g of liver were incorporated into glycogen with glucagon alone, whereas insulin plus glucagon produced a 2-fold increase in radioactive glycogen deposition.
It could be argued however, that the increase in [14C]glycogen deposition with insulin plus glucagon was due to higher specific activity of [i4C]glucose since insulin decreased total perfusate glucose values from 17 mM with glucagon alone to 11 mM with This would have produced final calculated specific activities of 2720 cpm per pmole with glucagon and 3860 cpm per pmole with glucagon plus insulin in a final volume of 50 ml. The 40% increase in specific activity could not satisfactorily account for the 100°jO difference in glycogen deposition.
'I'his possibility will be experimentally circumvented in the next series. Also shown in Table II, insulin caused a trend toward a sparing effect on tot,al liver glycogen levels although not statistically significant.
Liver glycogen synthase activity was determined in the same series (Table II).
With glucagon alone, synthase 1 activity was 14%,, whereas with insulin plus glucagon, synthase 1 activity increased to 21%. Therefore, these data demonstrate that insulin activates glycogen synthase in the presence of very low concentrations of glucagon in the isolated perfused rat liver.
Insulin caused a 38% decrease in tissue cyclic AMP from a glucagonstimulated level of 0.68 pmoles to 0.42 pmoles per mg of liver, a figure which is close to basal levels of the nucleotide (14). These data are consistent with an activation of synthase due to a decrease in synthase inactivation by cyclic AMP. This effect of insulin to lower tissue cyclic AMP in the presence of glucagon agrees with the original report of the same effect by Jefferson et al. (11).
Since glucose alone has been shown to activate glycogen synthase activity in the perfused rat liver (37,38), the inactivation of synthase by low levels of glucagon in the last series of experiments was complicated by the higher final glucose concentrations with glucagon as opposed to the lower glucose concentrations in livers treated with insulin plus glucagon.
This would, in effect, make it more difficult to demonstrate the synthase activating capability of insulin. Also, the incorporation of [14C]glucose into glycogen was complicated by a difference in final specific activities of ['*C]glucose. Therefore, the next series of experiments was designed to circumvent these two complicating factors.
Livers were perfused for 2 hours as previously described, except that supplemental glucose infusion was begun the 2nd hour in control (6.7 pmoles of glucose per min) and glucagon plus insulin (4.2 pmoles of glucose per mm) livers in order to equalize glucose accumulation with that previously observed with glucagon alone. All other additions and infusions were the same as those described for Fig. 1. Fig. 2 shows glucose accumulation in this series, and, as can be seen, glucose accumulation is essentially the same for the three different sets of conditions. Therefore, the problems of specific activities and glucose effects from the last series should be minimized. Effects of glucagon plus and minus insulin on glucose accumulation with infused alucose in perfused liver. The experimental procednre is described in the-text.
The least number of observations for each point was six. The final concentrations of glucagon and insulin were the same as those described for Fig. 1.  gen deposition as seen in Table II was not due to different specific activities, but rather to a direct effect of insulin to activate glycogen synthase.
The sparing effect of insulin on tissue glycogen is also significant under these conditions. This table also emphasizes that very small concentrations of glucagon can reduce [14C]glycogen depostion by about 95%. Table IV shows the effects of glucagon plus and minus insulin on synthase and synthase kinase activities and on cyclic AMP Glucagon caused a 65y0 decrease in synthase activity, whereas insulin partially reversed this effect of glucagon.
Synthase (protein) kinase act,ivity expressed as per cent I kinase (per cent active form) was increased by approximately 60% with glucagon, and as shown in the next column, the effect was probably exerted through the glucagon-stimulated increase in tissue cyclic AMP. Insulin, on the other hand, reversed the effect of glucagon on per cent I synthase kinase probably through its effect to lower glucagon-elevated tissue cyclic AMP levels. Therefore, it appears that the action of insulin to activate glucagon-inactivated synthase is at least partially due to a decrease in the hepatic cyclic AMP level which is responsible for a decrease in the activity of the synthase inactivating enzyme, glycogen synthase kinase.
The next series of experiments was designed to answer the basic question of whether or not insulin directly regulates glycogen synthase under basal metabolic conditions. Livers from fed rats were perfused by recirculation of substrate-free blood buffer mixture for 30 min to reach basal metabolic conditions and to achieve a relatively ahormonal state. After 30 min, insulin infusion was begun on the experimental side at a rate of 10 milliunits per ml of perfusate and continued for 6, 15, or 30 min. The experiments were terminated by freezing the livers as previously described.
Glucose production from 30 to 60 min was unaffected by this concentration of insulin (Fig. 3). After 30 min of equilibration perfusion, the control and experimental livers had the same per cent 1 synthase activities (Table V). At 6 min, the control livers showed 29% synthase I, whereas the insulin treated livers had a mean sypthase activity of 38y0,. The same relative effect of insulin was seen at 15 min. At 30 min, the insulin effect was no longer statistically significant.
The increases in synthase activity observed with time may be due to the increasing glucose concentration in the perfusate. The effect of insulin on glycogen synthase under basal metabolic conditions was rapidly manifest (6 to 15 min) and short in duration (30 min) in the perfused liver. This is similar to the effect of insulin on heart glycogen synthase activation, in vitro (22). Glycogen synthase kinase activity was determined at 6 and 15 min to elucidate the mechanism of insulin Experimental details are the same as those described for Fig. 3. The least number of observations for each mean was seven. Q Not significantly different from control. b p is less than 0.01 versus control. c p is less than 0.025 veTsus control. activation of glycogen synthase from a basal state (Table V). Insulin produced a 22y0 and a 25% decrease in synthase I kinase activity at 6 and 15 min, respectively.
The 20% decreases could readily account for the 40y0 increases in synthase 1 activity.
Table V also gives the tissue cyclic AMP levels in both control and insulin-treated livers at 6 and 15 min. Although there may be a trend toward lower cyclic AMP levels in the insulintreated tissue, there are no statistically significant differences between control and insulin levels.
Another group of livers was similarly perfused (30-min substrate-free and hormone-free) followed by a period of 10 min with insulin (10 munits per ml of perfusatej plus [%]glucose (27.2 prnoles per min of 8400 cpm per pmole) or [14C]glucose alone to prove that there is an increase in glycogen deposition with insulin under these in vitro conditions.
Although not shown, an incorporation of 1092 rpm per g into control livers and an incorporation of 1884 cpm per g into livers infused with insulin was observed.
Therefore, the increase in synthase I activity (Table  V) does, in fact, lead to an increase in glycogenesis.

Control of mammalian
hepatic glycogen synthesis by insulin has remained a controversial topic primarily due to the lack of evidence obtained in vitro and to several negative reports, in vivo (39) and in vitro (12, 38). Another factor contributing to this controversy has been the demonstration of the control of hepatic glycogenesis by glucose in the isolated perfused rat liver (37, 38)) a system that is apparently hormonally deficient.
Since it has been well established that mammalian liver has an intrinsic ability to regulate circulating glucose concentrations (40)(41)(42)(43) and that circulating glucose concentrations can regulate glycogen synthesis, most research has been directed at the mechanism of the glucose control of glycogen synthesis.
Regardless of the undisputed importance of glucose in controlling hepatic glycogenesis, it would seem an error to ignore previous reports of insulin activation of glycogen synthesis in in vivo prepa,rations (6)(7)(8)(9)34).
Data presented in this paper confirm the reports previously cited regarding insulin activation of glycogen synthase in viva, and in addition, clearly demonstrate for the first time that in-sulin increases glycogen synthase activity in the isolated perfused rat liver under at least three different sets of experimental conditions.
The insulin-mediated change in glycogen synthase activity can be correlated with a decrease in synthase 1 kinase activity, also demonstrated in liver for the first time. This stable enzyme effect can be readily explained by the decrease in tissue levels of cyclic ,4MP when insulin is used to antagonize the effect of glucagon.
However, this explanation does not necessarily appear to be the answer for the activation of hepatic glycogen synthase by insulin alone. As shown in Table V, synthase I kinase activity is decreased by insulin under basal metabolic conditions, while the tissue levels of the cyclic nucleotide remain apparently unchanged.
These data are similar to studies of Villar-Palasi and Wenger (44) where they found synthase I kinase activity reduced in skeletal muscle extracts of rats that had been treated with insulin in vivo. This decrease in activity was not correlated with a similar decrease in cyclic AMP. This was later confirmed and extended by Shen et al. (32) with diaphragm in vitro.
Thus, it appears that insulin alters the synthase I kinase activity without apparently altering the level of cyclic AMP.
There are two very obvious and perhaps not opposing views regarding this action of insulin.
First, since the free metabolically active pool of cyclic AMP may be very small compared to the bound pool found under basal metabolic conditions, a small decrease in cyclic AMP below base line would be difficult to detect using current techniques.
Second, the action of insulin to inactivate glycogen synthase kinase under basal metabolic conditions may be due to an insulin-mediated alteration in the sensitivity of the cyclic AMP-dependent kinase. This would mean that the same concentration of cyclic AMP would be less effective in the presence of insulin.
The present data with perfused liver could be interpreted to include both mechanisms.
In certain physiological situations, insulin acts in direct opposition to the effects of glucagon.
Here, insulin could act first by lowering glucagon-stimulated levels of cyclic iZMP.
Secondly, insulin could act by decreasing the sensit,ivity of the synthase kinase to that cyclic AMP remaining. If insulin acts by these two mechanisms, both involving control of the synthase kinase, a much more refined control mechanism could be maintained.
This would also account for the action of insulin on synthase kinase in situations not involving decreases in cyclic AMP, that is, when insulin acts in the absence of opposing hormones.
There are several negative reports regarding insulin activation of hepatic synthase in vitro (12, 38). Undoubtedly, there are additional negative findings which have not been reported. This is probably due to several important experimental complications. First, Glinsmann and Mortimore (12) reported that they were unable to show any effect on glycogen synthase using glucagon plus and minus insulin.
However, it appears that the concentration of glucagon that they used was about 4 X lo-lo M. Using this higher concentration of glucagon, the present investigators were also unable to obtain the synthase effect. Therefore, the difference was probably due to the higher glucagon concentrations employed.
This may be explained by the fact that inactivation of glycogen synthase by protein kinase is more sensitive to cyclic AMP than is activation of phosphorylase kinase and subsequenty phosphorylase activation by protein kinase (39). This would mean that synthase would be maximally inactivated while phosphorylase activation was being initiated. Second, the effect of insulin to activate synthase from basal metabolic conditions seemed to require a short equilibration period (ap-proximately 30 min) prior to the infusion of insulin. The time after the equilibration has already been noted as a period when insulin effects on glucose accumulation in the perfusate are not readily discernible (12). Since there are no clear-cut effects of insulin on glucose accumulation, many investigators have not looked for effects of insulin on glycogen synthase activation during this time period.
Lastly, the effect of insulin on synthase activation under basal conditions appears to be of rapid onset and of short duration, at least under conditions studied by the present investigators.
With longer exposure times, this effect of insulin could have been missed. Therefore, it would appear that most of the negative data can be readily explained by one or more of these complicating factors. Since the present study has not yet included determination of tissue synthase phosphatase activity, no mention has been made of this enzyme. This should not be taken as a suggestion that the importance of this enzyme is doubtful in insulin control of glycogenesis. Bishop (5) has shown that infusion of insulin into dogs results in an activation of hepatic synthase phosphatase between 5 and 15 min, and that glucagon promptly reduces this activity back to control levels. Evidence was also presented that synthase phosphatase exists in forms of different activity. It has also been shown that the synthase-activating enzyme (phosphatase) is almost absent in alloxan-diabetic rat liver and that administration of insulin results in restoration of the enzyme to normal levels (45). Therefore, it would appear that insulin may act to activate hepatic synthase phosphatase, and at the same time, maintain its integrity.
With insulin involved in the control of glycogenesis by way of synthase phosphatase, a third possible mechanism could be included in control of hepatic glycogenesis by insulin.
In conclusion, the present data complete the hypothesis that mammalian hepatic glycogen synthesis is directly controlled by insulin in vivo and in vitro.