Glucagon as a regulator of hepatic glucose production in vivo

Transcription

1 622nd MEETNG. LECESTER Stagner, J.., Samols, E. & Weir, G. C. (1980) J. Clin. nvest. 65, Lang, D. A,, Matthews, D. R., Peto, J. &Turner, R. C. (1979) N. Engl. J. Med. 301, Matthews, D. R., Lang, D. A,, Burnett, M. A. & turner, R. C. (1983) Diabetologia 24, Stagner, J.. & Samols, E. (1985) Am. J. Physiol. 248, Krieger, D. T. (1979) Endocrine Rhythms, New York: Raven Press 9. Lang, D. A,, Matthews, D. R. &Turner, R. C. (1981) Diabetes30, Matthews, D. R., Naylor, B. A,, Jones, R. G., Ward, G. M. & Turner, R. C. (1983) Diabetes 32, Verdin, E., Castillo, M., Lucyckx, A. S. & Lefebvre, P. J. (1984) Diabetes 33, Matthews, D. R., Hermansen, K., Conolly, A. A,, Gray, D., Schmitz, O., Clark, A,, Orskov, H. & Turner, R. C. (1987) Endocrinology 6, in the press 13. Matthews, D. R., Rudenski, A. S., Burnett, M. A,, Darling, P. & Turner, R. C. (1985) Clin. Endocrinol. 23, Matthews, D. R. (1986) in Pathogenesis and Treatment Diabetes Mellitus (Radder, J. K., Lemkes, H. H. P. J. & Krans, H. M. J., eds.), pp Matthews, D. R., Conolly, A. A. P., Holman, R. R. & Turner, R. C. (1985) Neth. J. Med. 28, 2024 Received 10 June 1987 Glucagon as a regulator hepatic glucose production in vivo ALAN D. CHERRNGTON, RALPH W. STEVENSON and KURT E. STENER Department Molecular Physiology and Biophysics, Vanderbilt University School, Medicine Nashville, Tennessee 37232, U.S.A. The liver takes up ingested carbohydrate and stores it as glycogen during periods feeding and then slowly releases the stored sugar during periods food deprivation. n addition, the liver is able to convert various nonglucose substrates (lactate, pyruvate, amino acids, glycerol) to glucose via the gluconeogenic pathway as a fast is prolonged. The aim the present paper is to review the role glucagon in regulating the production glucose by the liver (glycogenolysis and gluconeogenesis). All the data to be presented were obtained from experiments carried out on conscious overnight fasted dogs. The rates hepatic glucose production and glucose utilization were assessed using validated tracer and arteriovenous difference techniques which have been described in detail elsewhere [l31. n many the studies to be described the pancreatic clamp was used to fix the plasma insulin and glucagon concentrations at desired levels. n brief, this technique involves the infusion somatostatin, a potent inhibitor insulin and glucagon secretion, into a peripheral vein and the concurrent infusion the two pancreatic hormones into the hepatic portal vein. When basal amounts glucagon and insulin were given at the same time as somatostatin, all hormone levels measured (insulin, glucagon, cortisol, adrenaline noradrenaline) and all metabolic parameters determined (glucose turnover, net hepatic glucose balance, glucose clearance, as well as nonesterified fatty acid, amino acid, lactate and glycerol levels) were the same as those apparent in dogs infused with saline. Such data demonstrate the feasibility using the pancreatic clamp in physiological studies metabolism. The use somatostatin to fix the pancreatic hormone levels in overnight fasted dogs has now been validated in various situations [481. t is now clear that glucagon is the primary determinant hepatic glucose production. n the overnight fasted dog approximately twothirds resting glucose output is attributable to glucagon (Fig. ). When somatostatin was given along with intraportal replacement amounts insulin to create selective glucagon deficiency and glucose was infused to maintain euglycalemia, glucose production fell from 3.6 f 0.6 to 1.3 f 0.3mg/kg per min [5]. Although the plasma glucagon level only fell to 25pg/ml, the residual immunoreactivity was entirely attributable to a large molecular mass, crossreacting material rather than true glucagon. 15~r 5r 4 Glucose production (mg/kg per min), 1 0 5r 1 1 :Jo Fig. 1. The effects selective glucagon dejiciency brought about by the peripheral infusion somatostatin (0.8 pgglkgper min) and the intraportal infusion replacement amounts insulin (300 punitglkg per min) on the production, utilization and arterial level plasma glucose in overnight fasted conscious dogs Glucose was infused through a peripheral vein to maintain euglycaemia. Data are the mean ~s.e.m. The dotted lines represent changes in control dogs treated similarly but not exposed to glucagon deficiency. The Figure is reproduced from [5], with permission. Vol. 5

2 1024 BOCHEMCAL SOCETY TRANSACTONS Glucagon deficiency Somatostatin + lntraportal insulin and peripheral glucose 800 (n = 1) Arterial Plasma glucose 225 Net hepatic uptake 2.0 (pnol/kg per min) 0 t O0 into 50 glucose (% basal) 0.8 Fractional extraction Fig. 2. The effects selective glucagon deficiency brought about by the peripheral infusion somatostatin (0.8 pglkgper min) and the intraportal infusion replacement amounts insulin (300punitlkg per min) on the arterial plasma level, the conversion rate to glucose, the net uptake by the liver (NHAU) and the fractional extraction by the liver (FAE) in overnight fasted conscious dogs Glucose was infused through a peripheral vein to maintain euglycaemia. Note that NHAU and FAE were only measured in one dog. Data are expressed as mean ~s.e.m. The dotted lines represent changes in control dogs treated similarly, but not exposed to glucagon deficiency. The data can be found in part in [5], the Figure is reproduced from [16] with permission and the data come from the experiments shown in Fig. 1. Basal amounts glucagon exert their effect on hepatic glucose output primarily by stimulating glycogenolysis. This is apparent from the data in Fig. 2 which show that glucagon deficiency had little effect on the rate at which was converted to glucose, a good indicator the overall gluconeogenic process. Even the small fall in conversion which was apparent may actually have represented a diversion the newly synthesized glucose into glycogen rather than a decrease in gluconeogenesis per se. Such is the case following mild hyperinsulinaemia for example. Hepatic arteriovenous difference data were only obtained in one animal, but they suggested that overall gluconeogenic precursor balance changed very little as a result glucagon lack. The net hepatic uptake fell modestly (Fig. 2), but as a result the plasma level rose, thereby limiting the decrease in uptake by the liver. The fractional extraction the amino acid, a more sensitive index glucagon action, fell by almost 40% indicating that the hormone is a primary regulator the efficiency with Of 12 by liver Glucose 0 production 8 (mg/kg per min) utilization (mg/kg per min) 4l 2 O L Fig. 3. Effect selective glucagon excess brought about by the peripheral infusion somatostatin (0.8 pglkg per min) as well as intraportal infusion replacement amounts insulin (225 punitl kg per min) and 4fold basal glucagon (2.6 nglkg per min) on the production, utilization and arterial level plasma glucose in conscious overnight fasted dogs Data are mean S.E.M. The dotted lines represent the changes observed in dogs treated similarly but not exposed to excess glucagon. The Figure is reproduced from [19], with permission. which the liver extracts the amino acid. Since the specific activity plasma was unaffected by glucagon lack, it is apparent that the peripheral supply this important gluconeogenic precursor was not dependent on basal glucagon. t follows therefore that basal glucagon is a primary determinant the plasma and other gluconeogenic amino acid levels solely by virtue its effect on the liver. Neither the hepatic uptake nor blood level glycerol changed as a result glucagon deficiency (data not shown). t is apparent, therefore, that the output glycerol by fat tissue did not change during glucagon deficiency and that basal glucagon does not play a role in regulating the resting lipolytic rate. Lastly, the lactate levels fell slightly in response to glucagon lack, perhaps as a consequence the slight reduction in net hepatic lactate output which was evident following glucagon removal (data not shown). The fact that healthy overnight fasted dogs exhibit net lactate production is now established [3] and probably relates to the fact that dogs absorb a mixed meal very slowly [2] so that the absorptive period has only recently terminated when the animals are studied. Finally, data from the one dog studied with an 1987

3 622nd MEETNG, LECESTER 1025 Arterial plasma Glucagon excess lntraponal insulin (basal) _ = q q Somatostatin (Basal) ' lntraponal glucagon (4 x basal), (n = 4) (WWU 2oo Net hepatic a an i n e uptake (pmol/kg per min) o.8 Fractional extraction 0.4 by the liver to glucose (YO basal) 400 C lo.* Gluconeogenic efficiency ['4C]Glc prod. [4C]Ala uptake () 1 Fig. 4. Effect selective glucagon excess brought about by the peripheral infusion somatostatin (0.8 pgglkg per rnin) as well as intraportal infusion replacement amounts insulin (225 punitglkg per rnin) and 4fold basal glucagon (2.6 ngglkg per min) on parameters metabolism and gluconeogenesis in overnight,fasted conscious dogs Data are mean S.E.M. The dotted lines represent the changes observed in dogs treated similarly, but not exposed to glucagon excess. The data can be found in [20], the F' Lure is reproduced from [ 161, by permission, and the data come from the experiments shown in Fig. 3. Abbreviations as in legend to Fig. 2. hepatic vein catheter in place showed that the efficiency with which the extracted by the liver was converted to glucose was not appreciably altered by glucagon deficiency. Taken together the above data suggest, therefore, that after an overnight fast glucagon is the major stimulus for resting hepatic glucose production. ts effect is manifest almost exclusively on glycogenolysis since at the time the gluconeogenic rate is very low (contributing only 10% to overall glucose production), and the process is only slightly (20%) dependent on the presence basal glucagon. Studies on overnight fasted humans have shown that basal glucagon plays a similar role in man as it does in the overnight fasted dog ~91. The data in Fig. 1 also show clearly that a deficiency glucagon had no effect on glucose utilization. t should be remembered, however, that the glucose level was maintained by glucose infusion in this protocol, so the fall in glucose utilization which would otherwise occur as the result a fall in the glucose level resulting from a decrease in glucose production was prevented. t is also now clear that increments in glucagon are able to markedly stimulate glucose production. When the glucagon level was selectively increased in overnight fasted dogs (somatostatin and intraportal replacement amounts insulin were given to fix the insulin level at a basal value) it was shown to markedly activate both glycogenolysis and gluconeogenesis [6]. Fig. 3 shows the characteristic biphasic effect a squarewave increase in glucagon on glucose production. The rate appearance glucose in plasma initially tripled but then returned to a rate only slightly above baseline. The stimulation the gluconeogenic process by glucagon was progressive, and in fact was not complete even after 3 h hyperglucagonaemia (Fig. 4). The rate conversion to glucose had tripled by 3 h (Fig. 4), net hepatic uptake had increased by 60% and the fractional extraction by the liver had risen from 0.32 f 0.05 to 0.78 f As a result the increase in net hepatic uptake, the plasma level the amino acid fell significantly eventually limiting the effect glucagon on hepatic uptake. Thus, as in the case glucagon deficiency, glucagon excess was associated with abnormal plasma levels which resulted from the hormone's hepatic action. Selective glucagon excess was also associated with an inhibition lipolysis and as a result a decline in the blood Vol. 15

4 1026 BOCHEMCAL SOCETY TRANSACTONS glycerol level and net hepatic glycerol uptake. The antilipolytic effect the hormone, however, was secondary to its hyperglycaemic action rather than directly due to its action on the fat cell. This is clear from the data Shulman et al. [lo] who showed that hyperglycaemiaper se can inhibit lipolysis in the conscious dog. Net hepatic lactate output, which normally occurs in overnight fasted dogs [2], was stimulated by an elevation in glucagon [3]. This stimulation coincided with the effect the hormone on glycogenolysis [3] and only after 3 h selective glucagon excess did the liver begin to take up lactate in significant amounts (data not shown). n summary, a physiological increase in glucagon failed to increase glycerol uptake by the liver, increased lactate uptake only after the glycogenolytic action the hormone had waned, but stimulated (and presumably other gluconeogenic amino acids) uptake significantly within 60 min. n addition, the proportion the extracted which was converted to glucose almost doubled in response to a fourfold rise in glucagon (Fig. 4). Comparison the changes in overall glucose production and gluconeogenesis indicate that the initial increase in hepatic glucose output must have been attributable to a marked stimulation glycogenolysis. The gluconeogenic conversion to glucose rose progressively (Fig. 4), while overall glucose production rose and then fell, thus the glycogenolytic effect glucagon waned with time even when counterregulatory insulin secretion was prevented by somatostatin infusion and insulin replacement. This decline in glucose production is now known to be explained in part by the hyperglycaemia associated with glucagon excess and in part by inhibitory feedback signals within the liver which are still poorly understood [ ]. As noted above, the glucose level rose in the presence glucagon excess so that it became necessary to perform a control set experiments in which the pancreatic hormones were kept at basal levels throughout and the glucose level was raised (by glucose infusion) to a level corresponding to that seen in Fig. 3. When that was done, glucose production fell slowly (by 90min) to 1.5 f 0.3mg/kg per min. The arterial plasma level rofe to almost 500pmol/l but net hepatic did not change substantially because the fractional estimation the amino acid by the liver fell. Similarly there was a 3040% reduction in gluconeogenic efficiency and conversion. Thus the glycogenolytic and gluconeogenic effects glucagon are in fact underestimated by the data shown in Figs. 3 and 4. With regard to glucose utilization, in the control studies, glucose utilization rose to 5.5 f 0.6mg/kg per min while in the studies shown in Fig. 3, it rose to 4.6 f 0.7mg/kg per min. t is evident therefore that there was little if any effect the increase in glucagon on glucose utilization. The small decrease in glucose uptake which did occur probably reflects the fact that under hyperglycaemic conditions the liver takes up a small amount glucose and when glucagon is present the uptake is prevented by the hepatic action the hormone. Most normal stimuli to glucagon secretion (i.e. exercise [ 121, protein feeding [ 131, hypoglycaemia [14]) cause the plasma glucagon level (3500 M, glucagon) to increase by no more than 10fold (300pg/ml). t is thus important for glucagon to manifest its biological activity over a fairly narrow concentration range if it is to be an effective regulator glucose production by the liver. n a recent study in the conscious dog we used the pancreatic clamp to study the relationship between increments in plasma glucagon, brought about in the presence basal insulin, and its glycogenolytic and gluconeogenic effects [ 151. The hormone stimulated both processes in parallel such that halfmaximal responses were obtained with increments 125 and 75 pg/ml, respectively. Thus the commonly observed increases in glu cagon are indeed a magnitude large enough to bring about meaningful increases in glucose production. Taken together our data show that small increases or decreases (50 pg/ml) in plasma glucagon can significantly alter glucose output by the liver in the overnight fasted dog. This effect is most obvious with regard to the regulation hepatic glycogenolysis, however, it also involves gluconeogenesis. n the latter case, a decrease in glucagon modifies the already low gluconeogenic rate only slightly, while an increase in glucagon stimulates it markedly. The gluconeogenic effect the hormone is brought about solely by its hepatic action since our data indicate that glucagon cannot increase the peripheral release any the major gluconeogenic precursors or therefore the supply these substrates reaching the liver [3,6, 161. With regard to its hepatic action our data indicate that the hormone acts at two sites, the amino acid transporter in the plasma membrane and the enzymic machinery within the liver cell itself [3, 51. The conclusion that amino acid transport is stimulated directly is based on our finding that the hepatic fractional extraction the gluconeogenic amino acids, particularly, was sensitively modified by glucagon. Specifically, a change was noted with an increment in glucagon which was not large enough to modify intrahepatic gluconeogenesis [ 151. n addition, glucagon increased hepatic uptake even when it stimulated net hepatic lactate production [3]. The latter indicates that intracellular pyruvate levels were probably increased thus suggesting that pyruvate depletion within the hepatocyte was not pulling into the cell and therefore the transport system in the plasma membrane was in fact pushing the amino acid in. On the other hand, with larger increments in glucagon the change in the overall gluconeogenic process (as exemplified by the change in the conversion rate to glucose) exceeded that predicted by the effect the hormone on uptake alone, indicating that glucagon must also have an effect within the liver cell [3]. n addition, the proportion the extracted which was converted to glucose increased as the glucagon level rose, further supporting an action the hormone on the enzymic machinery within the hepatocyte. There are numerous reports describing in vitro work which supports an effect glucagon at these two sites and they have been discussed elsewhere (for reviews, see [ 17, 181).. Chiasson, J. L., Liljenquist, J. E., Lacy, W. W., Jennings, A. S. & Cherrington, A. D. (1977) Fed. Proc. 361, Davis, M. A,, Williams, P. E. & Cherrington, A. D. (1984) Am. J. Physiol. 241, E362E Davis, M. A,, Williams, P. E. & Cherrington, A. D. (1985) Am. J. Physiol. 248, E463E Cherrington, A. D., Lacy, W. W. &Chiasson, J. L. (1978) J. Clin. nvesf. 62, Cherrington, A. D., Liljenquist, J. E., Shulman, G.., Williams, P. E. & Lacy, W. W. (1979) Am. J. Physiol. 236, E Cherrington, A. D., Williams, P. E., Shulman, G. 1. & Lacy. W. W. (1981) Diabefes 30, Cherrington, A. D., Lacy, W. W., Williams, P. E. &Steiner. K. E. (1983) Am. J. Physiol. 244, E596E Stevenson, R. W., Steiner, K. E., Williams, P. E. & Cherrington. A. D. (1987) Metabolism in the press 9. Liljenquist, J. E., Mueller, G. L., Cherrington, A. D., Keller. U., Chiasson, J. L., Perry, J. M., Lacy, W. W. & Rabinowitz, D. (1977) J. Clin. nvest. 59, Shulman, G.., Williams, P. E., Liljenquist, J. E., Lacy, W. W., Keller, U. & Cherrington, A. D. (1980) Metabolism 29, Cherrington, A. D., Diamond, M. P., Green, D. R. & Williams, P. E. (1982) Diabetes 31, Lickley, L., Kemmer, F., Wasserman, D. H. & Vranic, M. (1983) in Glucagon 11 (Lefebvre, P. J., ed.), pp , SpringerVerlag, Berlin 1987

5 622nd MEETNG, LECESTER Chiasson, JL. & Cherrington, A. D. (1983) in Glucagon, (Lefebvre, P. ed.), pp , SpringerVerlag, Berlin 14. Frizzell, R. T., Hendrick, G. K., Biggers, D. W., Lacy, D. B., Donahue, D. P., Green, D. R., Carr, R. K., Williams, P. E., Stevenson, R. W. & Cherrington, A. D. (1987) Gluconeogenesis plays the major role in sustaining glucose production during hypoglycemia in response to insulin infusion in the dog. J. Clin. nvest. in the press. 15. Stevenson, R. V., Steiner, K. E., Davis, M. A,, Hendrick, G. K., Williams, P. E., Lacy, W. W., Brown, L., Donahue, P., Lacy, D. C. & Cherrington, A. D. (1987) Diabetes, in the press 16. Cherrington, A. D. & Vranic, M. (1986) in Hormonal Regulation in Gluconeogenesis (KrausFriedmann, N., ed.), pp. 1537, CRC Press, nc, Boca Raton, Florida 17. Kilberg, M. S. (1982) J. Membr. Biol. 69, Hormonal Control Gluconeogenesis: Function and Experimental Approaches (KrausFriedmann, N., ed), vol., pp , CRC Press, Boca Raton. Florida 19. Steiner, K. E., Bowles, C. R., Mouton, S. M., Williams, P. E. & Cherrington, A. D. (1982) Diabetes 31, Steiner, K. E., Mouton, S. M., Williams, P. E., Lacy, W. W. & Cherrington, A. D. (1986) Diabetes 35, Received 10 June 1987 Methods for assessing insulin sensitivity in man K. G. M. M. ALBERT, S. N. DAVS, L. MONT, N. MOLLER, P. PATT and R. HENE Department Medicine, The Medical School, Framlington Place, Newcastle upon Tyne, NE2 4HH, U.K. ntroduction Classically, insulindependent diabetes mellitus (DDM) and noninsulin dependent diabetes (NDDM) were considered to be disorders insulin secretion. n DDM, insulin secretion is totally lacking while it was assumed that there was a partial deficiency in NDDM. More recently, it has been realised that in both conditions there is insulin resistance, and this may indeed be major pathogenic significance, as well as therapeutic importance in NDDM. This has led to a search for simple, accurate and reproducible tests insulin sensivity in man in vivo. The first tests were devised in the 1930s (Himsworth, 1936; Himsworth & Kerr, 1939) and comprised glucose tolerance tests with and without added insulin injection. Further developments awaited the introduction the insulin immunoassay. The modern era started with the pioneering work Andres et al. (1 966) and Reaven s group (Shen et al., 1970). Since then many tests have been devised. They can be divided into an indirect or endogenous group where attempts are made to assess the effect endogenous insulin and a direct or exogenous group where insulin is infused and the effects on glucose metabolism are assessed. Each has its advantages and disadvantages and these are briefly discussed below. Tests endogenous insulin sensitivity Basalstate. Attempts have been made to use basal plasma glucose and insulin levels to assess insulin sensitivity in the resting state (Turner et al., 1979). t is obvious that ifglucose levels are normal and insulin levels are raised, then there is insulin insensitivity. Problems arise when glucose levels as well as insulin levels are raised. n addition, much the action insulin in the resting state is on the liver, while insulin is measured in the periphery, portal levels being variably higher. No reliable index has yet been obtained although glucose/insulin or glucose/cpeptide ratios have been used as crude indices. Oral glucose tolerance test (OGTT). This has been the timehonoured method for assessing insulin sensitivity. t has the advantage relying on the subject s own insulin, but has several disadvantages. Both glucose and insulin levels are changing, making interpretation difficult. t is Abbreviations used: DDM, insulindependent diabetes mellitus; NDDM, noninsulindependent diabetes mellitus; OGTT, oral glucose tolerance test; VGTT, intravenous glucose tolerance tests; FSGT, frequency sampling insulin and glucose test; TT, insulin tolerance test, nonsteady state; subtle changes in the rate insulin secretion could have effects on insulin disposal, and variations in gastrointestinal absorption, and gut hormone secretion could confound the results. Bergman et al., (1979) have created a successful model insulin sensitivity based on the OGTT, but it is difficult to validate. ntravenous glucose tolerance tests (VGTT). The VGTT has advantages over the OGTT in that glucose absorption is no longer a significant variable and gut factors are not involved. Glucose disappearance is loglinear and can be expressed as percentage disappearance per minute (Kg). f the insulin response to the glucose is constant, then Kg S an expression insulin sensitivity. This is not the case, but the early insulin response is related to Kg in normal man, and sensitivity can be expressed rather crudely as a ratio these two. More recently Bergman and his colleagues (see Bergman et al., 1985) have developed a model which allows a rather sophisticated expression insulin sensitivity from a standard VGTT with very frequent sampling (FSGT). This is the minimal model, so called because it is the simplest physiologically based model which can account for the inputoutput relationships under various conditions. t takes into account the rate change glucose and insulin, and memory for the insulin effect; insulindependent and insulinindependent glucose metabolism, and the effects insulin on hepatic glucose production, or peripheral glucose disposal. Several studies have now been reported comparing the minimal model FSGT with the more conventional euglycaemicclamp method (see below). These showed a relatively poor correlation (Foley et al., 1985; Beard et al., 1986), particularly when insensitive subjects were included. The FSGT was therefore modified by adding a tolbutamide injection after 20 min. This improved the correlations markedly (Bergman et al., 1987) and this modified minimal model is the FSGT choice. An alternative method with continuous infusion glucose with model assessment (CGMA) has been developed (Hosker et al., 1985), and also shown to correlate reasonably well with euglycaemic and hyperglycaemic clamps. Nonetheless, there are still doubts about the physiological meaning the various glucose tolerance tests as measures insulin sensitivity. Tests exogenous insulin sensitivity nsulin tolerance test (TT). The response blood glucose to an intravenous bolus insulin has long been used as rough guide to insulin sensitivity, but fell out favour because the obvious counterregulatory hormone response as blood glucose fell. Recently, however, an excellent correlation has been found with the euglycaemic hyperinsulinaemic clamp in normal subjects (Bonora et al., 1987) using the decrease in glucose over the first 15 min Vol. 15