Effect of Metformin
Treatment on Insulin Action In Vitro Correlations
Ralph A. DeFronzo, Gerald I. Shulman,
Rats: In Vivo and
Roberto Gherzi, Peter Stein, Gabriella Andraghetti, Eveline Klein-Robbenhaar,
and Renzo Cordera
The mechanism (both at the whole body and cellular level) by which metformin improves insulin sensitivity has yet to be defined. In the present study, we examined in vivo insulin-mediated whole-body glucose disposal, glycogen synthesis, hepatic glucose production, and insulin secretion, as well as in vitro muscle insulin receptor tyrosine kinase activity in eight control, eight neonatal streptozotocin diabetic rats, and eight diabetic rats before and after treatment with metformin. Ten weeks after birth diabetic rats had higher fasting (132 + 5 Y 101 + 2 mg/dL) and postmeal (231 + 10 v 133 + 3) plasma glucose levels compared with controls (P -C 901). Metformin treatment was followed by a significant decrease in the growth rate and normalized glucose tolerance without enhancing the deficient insulin response. Insulin-mediated glucose uptake in diabetic versus control rats was reduced (P -C .Ol) during the high-dose (15.4 + 0.6 v 18.3 + 1 .O mg/kg . min) insulin clamp study and was increased to values greater (P < .05I than controls following metformin treatment. Muscle glycogen synthetic rate in vivo, measured by incorporation of 3H-3-glucose radioactivity, was diminished by 25% (P < .Ol ) in diabetic rats, restored to normal values with metformin, and correlated closely (r = .82, P < .002) with total-body glucose uptake during the insulin clamp in all three groups. Insulin receptor tyrosine kinase activity, measured in partially purified insulin receptors, was reduced in diabetic rats and increased to supernormal levels after metformin. The decrease in muscle tyrosine kinase activity in diabetic versus control animals was entirely accounted for by a reduction in maximal velocity (V,,,) (32 v 45 pmol/mg - min. P < .Ol) and increased to supernormal levels following metformin (91 pmol/mg - min. P < .OOl) without any change in affinity (Km). Muscle tyrosine kinase activity was closely correlated with both the muscle glycogen synthetic rate (r = .82, P < 602) and total-body insulin-mediated glucose disposal (r = 64, P < .Ol) in vivo. The close correlation between in vivo insulin action, muscle glycogen synthesis, and muscle insulin receptor tyrosine kinase activity is consistent with an important role of the enzyme in the insulin resistance of diabetes and its improvement following metformin treatment. @ 1990 by W.B. Saunders Company.
NSULIN RESISTANCE is a common feature of human and experimental diabetes.le3 A reduction in P-cell mass, obtained either by surgical pancreatectomy3 or chemical with the development of insulin agents, 4.5 is associated resistance without any change or even an increase in insulin receptor binding.‘.’ Recently, a significant body of evidence has suggested that insulin receptor autophosphorylation and insulin receptor protein tyrosine kinase activity play an important role in the transduction of the biological signal of insulin into cells.“” Consistent with this, insulin receptor protein tyrosine kinase activity has been shown to reflect insulin’s action both in vivo and in vitro.‘2-2’ Metformin is a drug that improves glucose tolerance in diabetes mellitus and has been shown to work by enhancing tissue sensitivity to insulin.22-24 However, the tissue(s) involved in metformin’s action, as well as the intracellular biochemical processes that are affected by the biguanide, remain to be defined. In the present study, we determined, in the same animal, hepatic glucose production and peripheral glucose uptake and correlated these measurements with autophosphorylation and protein tyrosine kinase activity of the hepatic and muscle insulin receptor. In order to examine whether changes in insulin-mediated glucose disposal paralleled changes in inulin receptor protein tyrosine kinase activity, measurements were made in control animals, insulin-resistant diabetic a.nimals, and animals who were treated with metformin to enhance tissue sensitivity to insulin.
No 4 (April), 1990: pp 425-435
METHODS In Vivo Studies;
Animal preparation. Three groups of male Sprague-Dawley rats (Charles River Laboratories, Wilmington, MA) were studied: group I&ontrols (n = 8); group II-streptozotocin-induced diabetic rats (n = 8); group III-streptozotocin-induced diabetic rats treated with metformin (n = 8). In groups II and III diabetes was induced by injecting 100 mg/kg streptozotocin (Upjohn, Kalamazoo, MI) intraperitoneally (IP) on the second day after birth.25 At 4 to 5 weeks of age rats were tested for plasma glucose concentration at 8 AM. Animals were only included in this study if their plasma glucose was greater than 180 mg/dL on three consecutive occasions. Following this criteria, 60% to 70% of the injected rats became diabetic. Control animals received an equivalent volume of 0.1 mol/L citrate buffer IP. In group III diabetic animals metformin therapy was initiated 4 to 5 weeks after birth at a time when
From the Division of Diabetes, University of Texas Health Science Center at San Antonio; Diabetes/Endocrinology Division, Yale University School of Medicine, New Haven, CT; and the Department of Internal Medicine, University of Genova. Genova, Italy. Supported in part by a grant (L.R.) from the Juvenile Diabetes Foundation (No. 188016). Address reprint requests to Luciano Rossetti, MD, Diabetes Division, Department of Medicine, University of Texas Health Science Center, 7703 Floyd Curl Dr. San Antonio, TX 78284. @ 1990 by W.B. Saunders Company. 0026-0495/i/90/3904-0015$3.00/O
ROSSETTI ET AL
hyperglycemia first became manifest. Metformin, approximately 0.1 g/kg/d, was administered in the drinking solution (1.4 mg/mL). On day 24 to 26 after birth, all rats were weaned, housed in individual cages, and subjected to a standard light (6 AM to 6 PM)/dark (6 PM to 6 AM) cycle. Based on prior experience, all rats were pair-fed by receiving the identical daily allotment (0.1 g/g body weight [SW]) of Purina Rat Chow (St Louis, MO) in an amount that sustained normal growth and that was completely consumed by all of the animals. No diarrhea was observed in any of the experimental animals. All rats were weighed twice weekly and tail vein blood collected at 8 AM for the determination of nonfasting plasma glucose concentration at the same time. The nonfasting plasma glucose value displayed in Table 1 represents the mean of five determinations in each rat between 8 and 10 weeks after birth. A fasting plasma glucose concentration also was determined twice weekly at 4 PM on tail vein blood following an g-hour fast. Nine weeks after birth and 1 week before performing the insulin clamp study, rats were anesthetized with phenobarbital (50 mg/kg BW) and indwelling catheters were inserted so that the animals could be studied in the awake, unstressed state. Two internal jugular catheters were inserted and extended to the level of the right atrium and a left carotid catheter was advanced to the level of the aortic arch. The three catheters were filled with heparin/polyvinyl-pyrrolidone solution, sealed, and tunneled subcutaneously around the side of the neck to the back of the head. The catheters were externalized through a skin incision and anchored to the skull with a dental cement cap. Insulin clamp study. Seven days following catheter insertion, rats in groups I-III received a two-step euglycemic insulin clamp study.‘,*” At 8 AM following an overnight fast, the heparin/polyvinylpirrolidone solution was aspirated and catheter patency maintained by a slow, continuous infusion of isotonic saline. The venous catheter was used for blood withdrawal and the arterial catheter for infusion of all test substances. The catheters were suspended on a pulley system to allow free movement of the rat in the cage throughout the insulin clamp study. To prevent intravascular volume depletion and anemia, fresh whole blood (1.5 mL blood plus 1.0 mL saline) obtained by heart puncture from littermates of the test animals was administered at a constant rate (14 pL/min) designed to quantitatively replace the total blood loss during the study (-2.5 mL). Sixty minutes before starting the insulin clamp, a prime- (6 &i) continuous (0.1 &i/min) infusion of ‘H-3-glucose (New England Nuclear, Boston, MA) was initiated and continued throughout the study.3 Plasma samples for the determination of tritiated glucose specific activity were obtained at j-minute intervals from - 30 to 0 minutes and at 5- to lo-minute intervals after starting the insulin infusion. At time 0, a two-step euglycemic insulin clamp was begun.‘,26 A prime-continuous (1.0 mu/kg . min) infusion of regular insulin (Lilly, Indianapolis, IN) was administered from time 0 to 90 minutes to raise the plasma insulin concentration acutely and to maintain it Table 1. Body Weight,
by approximately 50 &/mL. At 90 minutes, the insulin space was reprimed and the continuous insulin infusion rate was increased to 3.0 mu/kg . min to maintain the plasma insulin concentration at approximately 160 pU/mL until 180 minutes. In all groups a variable infusion of 25% glucose was started at time 0 and adjusted to clamp the plasma glucose concentration at approximately 100 mg/dL. In group II the glucose infusion was not started until a mean of 8 minutes after starting the insulin, at which time the fasting plasma glucose concentration (132 mg/dL) had declined to 100 mg/dL. Samples for plasma insulin determination were obtained at 15- to 30-minute intervals throughout the 1IO-minute insulin clamp study. At the end of the 180-minute insulin clamp study, animals were anesthesized with sodium phenobarbital and liver and hind limb muscle were snap frozen in vivo, excised, and stored at - 70°C until analyzed. In a separate series of experiments, five additional rats from groups I-III were fasted overnight and at 8 AM they were anesthesized and muscle and liver tissue were removed as described previously for the determination of cold glycogen determination. Tracer was not administered in these later studies. Analytical procedures. Plasma glucose was measured by the glucose oxidase method (Glucose Analyzer, Beckman Instruments, Palo Alto, CA) and plasma insulin by radioimmunoassay using porcine insulin standards. Plasma ‘H-3-glucose radioactivity was measured in duplicate on the supernatants of barium hydroxide-zinc sulphate precipitates (Somogyi procedure) of plasma samples after evaporation to dryness to eliminate tritiated water.3.‘6 Muscle and liver glycogen concentrations were determined as previously described.” Calculations. Data for total-body glucose uptake and suppression of hepatic glucose production represent the mean values during the last 30 minutes of the l.O-mu/kg . min (ie, 60- to 90-minute time period) and the 3.0-mu/kg . min (ie, 150- to 180-minute time period) insulin clamp studies. This time period was chosen in order to allow insulin to more fully exert its biologic effects. Qualitatively similar results are obtained if the initial 60-minute or entire 90-minute period of hyperinsulinemia are compared. Total-body glucose disposal was calculated by adding the rate of residual hepatic glucose production during the last 30 minutes of each insulin clamp step to the glucose infusion rate during the same 30-minute time period. All values are presented as the mean k SEM. Differences between groups were determined using the analysis of variance in conjunction with the student Newman-Kuels test. An index of the liver and muscle glycogen synthetic rate was calculated as follows: (‘H counts per gram tissue) x (plasma ‘H-glucose specific activity) x 100. Since liver/muscle tissue could not be obtained at time 0, ie. start of insulin clamp, the accumulation of counts during the basal equilibration period could not be determined. Therefore, the precedingcalculation does not provide an absolute rate, but rather an index of glycogen synthesis.
Fasting, and Fed Plasma Glucose Concentrations,
and Fasting and Fed Plasma Insulin Levels
5.7 + 0.5
5.5 + 0.6
III: Diabetic + metformin
NOTE. Fed plasma glucose refers to values obtained at 8
112 5 1
< .05 to .Ol vcontrol.
tP < .05 to .Ol “diabetics.
135 + 3
2.1 f 0.2
5.3 t 0.3
1.6 * 0.4
1.7 * 0.5
and represents the mean of five determinations in each rat between 8 and 10 weeks after
birth. Fasting plasma glucose concentration was obtained at 4 PMfollowing an 8-hour fast. ‘P
INSULIN ACTION, AND IRTK
In Vitro Studies: Insulin Binding, Insulin Receptor Phosphorylation, and Insulin Receptor Histone Kinase Activity Four rats from each experimental group were used and each measurement was randomly repeated three times using wheat germ agglutin-purified receptors from different aliquots of microsomal membranes. Preparationof wheatgerm lectin-purifiedreceptors. Microsomal membranes were prepared from liver and hind limb muscle by differential centrifugation according to previously described methods.“,** Membranes were solubilized for 1 hour at 4OC in the presence of 2 mg/mL bacitracin, 2.5 mmol/L phenylmethylsufonylfluoride, 1500 Kallikrein inhibitor U/mL, aprotinin, 1% Triton X-100, and 1% glycerol (pH 7.4). Nonsoluble material was separated by centrifugation at 12,000 x g for 30 minutes at 4OCand the clarified supernatant was applied to a WGA column. Bound glycoproteins were desorbed with a buffer containing 0.3 mol/L N-acetyl-Dglucosamine, 0.05% Triton X-100, and 10% glycerol elution buffer.29 The fractions containing insulin receptors were pooled and used. Insulin binding assays. [ ‘251]-insulin binding to WGA-purified receptors was performed as previously described.*’ Briefly, aliquots of lectin-purified receptors were incubated with [‘*‘II-insulin (0.4 ng/mL) in a buffer containing 25 mmol/L Hepes, 120 mmol/L NaCI, 5 mmol/L KCI, 1 mmol/L, MgSO,, 1 mmol/L CaCI,, 0.05% Triton X-100, and 1% bovine serum albumin, pH 7.8 for 16 hours at 4OC. Receptor-bound radioligand was separated by precipitation with polyethylene glycol and y-globulin (final concentrations 12.5% and 0.14%, respectively). Nonspecific binding was determined by measuring the amount of radioactivity precipitated in the presence of 2 rg/mL insulin and was less than 5% in all experiments. Insulin binding data were analyzed by Scatchard plot.jO Phosphorylationassays. Based on the insulin binding capacity, receptor preparations were diluted in elution buffer as necessary so that equal binding activity was used to measure insulin receptor autophosphorylation and protein tyrosine kinase activity. WGApurified receptors were preincubated at 4OC for I6 hours in the absence or presence of different concentrations of porcine insulin in a total volume of 40 pL of buffer containing 25 mmol/L HEPES (pH 7.4). 5 mmol/L MnCl,, 10 mmol/L MgCl,, and 0.2% Triton X-100. The phosphorylation reaction was conducted as previously described.** In experiments measuring the phosphorylation of histone HFZB, the substrate was added at a final concentration of 0.1 mg/mL. For kinetic studies, WGA-purified receptors were incubated for 16 hours at 4OC with 25 ng/mL or 200 ng/mL of insulin, and then histone HFZB phosphorylation was conducted for 5 minutes at 4OC as previously described. 3’ Samples were then subjected to sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) under reducing conditions and, after electrophoresis, gels were stained, destained, dried, and submitted to autoradiography using Kodak X-Omat AR films (Eastman Kodak, Rochester, NY) and DuPont intensifying screens (DuPont, Wilmington, DE) as described previously.2* Autoradiograms were scanned with an LKB Ultrascan laser densitometer. The intensity of the signal reflecting phosphorylation of the beta subunit of the receptor and of histone are expressed in arbitrary units. Basal phosphorylation was similar in all groups. Adenosine triphosphatase (ATPase) and phosphatase activities were measured in WGA-purified receptor preparations of the three groups, as previously described,“‘and were found to be similar. Materials. Porcine insulin and ‘*‘I-A 14-monoiodoinsulin (230 mCi/mmol/L) were supplied by Novo Industries (Copenhagen, Denmark). [Y-‘~P]-ATP (3,000 Ci/mmol/L) was purchased from New England Nuclear. ATP, Triton X-100, phenylmethylsulfonylflu-
oride, bacitracin, and N-acetyl-D-glucosamine were purchased from Sigma Chemical. WGA-agarose was obtained from Vector Laboratories (Burlingame, CA) and histone HFZB from Cooper Biomedical (Malvern, PA). All materials for SDS-PAGE purchased from Bio-Rad (Richmond, CA). The sources of all other chemicals were cited in reference 28. RESULTS
General Animal Characteristics
There were no differences in the mean BW between control (group I) and diabetic (group II) rats. However, metformin-treated diabetic rats displayed a significant reduction in mean BW compared with both group I and II rats (P < .Ol). Both the fasting and postmeal plasma glucose concentrations (mean of five determinations) during the 3-week period prior to the insulin clamp study were significantly higher in the diabetic group compared with both control and metformin-treated diabetic animals (P < .Ol). The fasting and postmeal plasma insulin concentrations in the two diabetic groups were significantly reduced compared with controls (P < .Ol), while the fasting plasma insulin levels were slightly but not significantly lower than in controls (Table 1). Insulin-Mediated
Tissue Glucose Uptake
During the insulin clamp studies performed in groups I, II, and III the steady-state plasma glucose concentrations were maintained at 101 * 1, 105 + 2, and 104 f 2 mg/dL with coefficients of variation ranging from 5.9% to 6.9%. The steady-state plasma insulin concentrations during the first (4.1 + 0.2, 4.1 + 0.3, 3.9 + 0.2 ng/mL) and second (8.3 2 0.4, 9.2 t 0.4, 8.3 + 3 ng/ml) steps of the insulin clamp were similar in groups I, II, and III, respectively. In all studies the coefficient of variation in plasma insulin concentration was less than 10%. In diabetic rats (group II) insulin-mediated glucose disposal was similar to controls during the l-mu/kg a min (40 ng/kg . min) insulin clamp step, but was severely impaired (P < .Ol) during the 3-mU/ kg . min (120 ng/kg . min) insulin clamp step. Metformin treatment (group III) improved insulin sensitivity in diabetic rats to levels that were significantly greater than in control (P < .05) and diabetic (P < .Ol) animals during both insulin clamp steps (Fig 1A). Hepatic Glucose Production
Basal hepatic glucose production (HGP) was slightly, although significantly increased in diabetic compared with control rats (P < .Ol). Following metformin treatment no significant change in HGP was observed and basal hepatic glucose output remained significantly elevated compared with controls. Basal HGP and the fasting plasma glucose concentration were positively correlated (r = .74, P < .Ol) in all three groups. Suppression of HGP by insulin was significantly impaired in diabetic rats compared with controls during both insulin clamp steps and was not significantly modified by metformin treatment (Fig 1B).
Insulin Clamp (I mu/kg -min)
Insulin Clamp (3mWkg.min)
Insulin Clamp (1 mU/kg*min)
Insulin Clamp (3 mU/kg*min)
Con Diab Met
Con Diab Met
Tissue Glucose Uptake (mg/kg.min)
Hepatic Glucose Production 4 (mglkgamin)
Fig 1. (A) Insulin-mediated tissue glucose uptake during the l-mu/kg . min (left panel) and 3-mu/kg - min (right panel) euglycamic insulin clamp studies performed in control (CON), diabetic (OIAB), and diabetic rats treated with metformin (MET). l*f < .Ol v controls lP < .Ol vdiabetics and controls. (B) Hepatic glucose production during the basal state (left panel) and during the l-mu/kg . min (middle panel) and 3-mu/kg - min (right panel) auglycemic insulin clamp studies performed in control, diabetic. and diabetic rats treated with maformin. lP < .Ol v controls.
Liver and Muscle Glycogen Concentration and Glycogenic Index In the postabsorptive state, muscle glycogen concentrations were similar in all three groups. Following insulin infusion the increment in muscle glycogen concentration and glycogenic index were significantly reduced in diabetic animals. Following metformin treatment, the muscle glycogen concentration, as well as the muscle glycogenic index, was returned to control levels (Fig 2). A strong positive correlation was observed between the increase in cold glycogen concentration and the increase in glycogenic rate in muscle (r = .82, P < .002) during the insulin clamp. In the postabsorptive state, hepatic glycogen concentration was significantly higher (P =z.Ol) in both of the diabetic groups compared with controls. No net increase in liver glycogen concentration could be detected in any of the three groups following insulin infusion. Nonetheless, there was a significant accumulation of tritiated glucose counts in liver glycogen in all three groups and this increase in radioactivity was well in excess of that which we have observed in control animals during a similar baseline period of tracer infusion (unpublished results). The accumulation of tritiated glucose radioactivity in liver glycogen in metformin-treated diabetic rats was significantly (P < .Ol) greater than in control and diabetic animals; consequently, the glycogenic index was significantly enhanced in the metformin-treated diabetic rats compared with both other groups (Fig 2). Insulin Binding Equilibration binding “‘I-insulin by the lectin-purified insulin receptors from hind limb muscles of control, diabetic, and metformin-treated diabetic rats did not display any significant deferences among groups when expressed either
Liver 14 12 Liver G;%?lo;n
1 .o 1
Fig 2. (A) Liver glycogan concentration (left! and glycoganic rata (right) following insulin infusion in control (CON), diabetic (DIAB). and diabetic rats treated with metformin (MET). Basal liver glycogen content was significantly increased (P < .OOl) in diabetic and matformin-treated rats. lP < .Ol Y controls. (8) Muscle glycogan concentration (left) and muscle glycogenic rata (right) following insulin infusion in control, diabetic, and diabetic rats treated with matformin. Basal muscle glycogen content was similar in all three groups. lP -c .Ol Y controls.
METFORMIN, INSULIN ACTION, AND IRTK
per milligram of protein or per gram of muscle. However, when partially purified insulin receptors from liver extract were used to measure insulin binding, a slight increase in insulin receptor number (binding capacity) per milligram of protein, as determined by Scatchard analysis, was detected in both diabetic groups compared with controls; the increase in insulin receptor number achieved statistical significance (P < .Ol) only in the metformin-treated group compared with controls (Fig 3A and B). Insulin Receptor Protein Tyrosine Kinase Activity
Recovery of WGA-enriched insulin receptors was assessed as previously described2’ and was over 90% in all groups. In insulin receptors purified from liver membranes of streptozotocin-diabetic rats, receptor autophosphorylation was signifi-
cantly reduced compared with controls (Fig 4, top). This decrease was most apparent at insulin concentrations above 25 ng/ml. Treatment of streptozotocin-diabetic rats with metformin led to an increase in insulin-stimulated protein tyrosine kinase activity, an effect that was most evident at insulin concentrations above 25 ng/mL (Fig 4, bottom). In insulin receptors purified from muscle membranes a decrease in protein tyrosine kinase activity was observed at all insulin concentrations (2.5 to 200 ng/mL) examined (Fig 5, top). Similar changes in muscle protein tyrosine kinase activity were observed when histone HZFB was used as a substrate (Fig 5, bottom). Kinetic analysis of the protein kinase activity also was performed using the Lineweaver-Burk plot (Fig 6). The results of this analysis demonstrated that in insulin receptors
A MUSCLE 0.3
N 0 x
A DIAB 0 MET
f 0.2 e & f $ 0.I
A DIAB 0 MET
N 0 ;
F B z Fig 3. (Al Insulin-receptor binding in muscle from control (CON). diabetic (DIAL) and diabetic rats treated with metformin (MET). Receptor preparations were incubated in duplicate at 4% for 16 hours in the presence of 0.4 ng/mL (‘261-tyr-A14) insulin and increasing concentrations of cold insulin. Insulin binding was determined by precipitation of insulin receptor complexes with polyethylene glycol as described in the Methods section. The ordinate represents the ratio of bound to free hormone and the abscissa represents insulin bound per milligram of protein (Scatchard plot). The total insulin binding capacity is represented by intercepts at the abscissa. (6) Insulin receptor binding in liver receptors from control, diabetic, and diabetic rats treated with metformin.
I BOUND @g/ml)
ROSSE’ITI ET AL
Insulin resistance is a characteristic feature of non-insulindependent diabetes mellitus in humans’.* and has been shown to involve a variety of tissues, including liver,32.33adipocytes,‘” and muscle.” Following glucose ingestion/infusion, insulin secretion is stimulated and the hyperinsulinemia enhances glucose tolerance by augmenting tissue glucose uptake and suppressing hepatic glucose output.32,33 Peripheral tissues, primarily muscle, are responsible for the disposal of the majority of an infused/ingested glucose load.32,33 In the present study, we used the neonatal streptozotocin rat,25.36a model that resembles type II diabetes mellitus in humans, to examine the tissue sites and biochemical mechanisms that are responsible for insulin resistance. Ten weeks following streptozotocin injection, rats demon-
s i% 815
ii oZm +%
02 ZE - .!gn
MUSCLE l CON
A DIAB 0 MET
INSULIN (ng/ml) Fig 4. Dose-response curve for autophosphorylation (top) and phosphorylation of the exogenous substrate histone HF2B (bottom) by partially purified liver insulin receptorszO Aliquots of WGA-purified receptors from control (CON). diabetic IDlAB), and diabetic animals treated with metformin (MET) were preincubated in the absence and presence of increasing amounts of insulin. After 16 hours at 4°C. phosphorylation assays were performed as described in the methods section. samples were subjected to SDS-PAGE, and the phosphoproteins corresponding to 96 kDa band and HF2B band were located by autoradiography and densitometrically counted. Valuea are expressed as meens r SE. They represent fold of increase over the basal. Basal values (in arbitrary units) of “P incorporated into insulin receptor beta subunit were: control, 1.49 + 0.25: diabetic, 0.95 f 0.31; diabetic plus metformin = 1.9 + 0.58. “P incorporated into HFZB were: control, 4.8 + 1.7; diabetic, 3.9 ? 1.25; diabetic plus metformin, 4.1 + 0.95.
from muscle and liver, the V,,, was significantly (P -C .Ol) reduced in diabetic animals and was returned to supernormal levels following metformin therapy, P < .Ol (Fig 6, Table 2). When the V,,, for protein tyrosine kinase in muscle tissue from all rats studied was plotted against the rate of insulinmediated glucose disposal measured in vivo during the 3-mu/kg . min euglycemic insulin clamp, a positive correlation was observed (r = .64. P -c .Ol). Basal rates of phosphorylation were similar in all groups.
INSULIN (ng/ml) Fig 5. Dose-response curve for autophosphoryletion (top) and phosphoryletion of the exogenous substrate histone HFPB (bottom) by partially purified muscle insulin receptors.” See legend for Fig 4. Basal values of 32P incorporated into insulin receptor beta subunit were: control, 0.35 + 0.03: diabetic, 0.41 + 0.01: diabetic plus metformin. 0.37 * 0.02. ‘*P incorporated into HF2B were: control, 1.65 i- 0.50: diabetic, 1.70 + 0.44: diabetic plus metformin, 1.5g _c0.42.
INSULIN ACTION, AND IRTK
INSULIN 200 nglml
MET Fig 6. Kinetic analysis of histone HF26 phosphorylation by liver insulin receptors from control (0). diabetic (A). and diabetic rats treated with metformin (0 1.WGA-purified receptors were incubated with (A) 2.6 ng/mL insulin or (B} 200 ng/mL (bottom panel) and then histone HFZB phosphorylation was performed for 6 minutes at 4°C as described in- the Methods section. A LineweeverBurk plot representative of four independent experiments is shown.
strated mild fasting hyperglycemia and modest glucose intolerance (Table 1). Although the fasting plasma insulin level was only minimally decreased in diabetic animals, the insulin response to a mixed meal was significantly impaired (Table 1). Despite the presence of nearly normal fasting insulin and elevated fasting glucose levels, hepatic glucose production was significantly increased (Fig 1B). These results indicate that the combined effects of insulin and hyperglycemia to suppress hepatic glucose production are impaired. Furthermore, we observed a strong correlation (r = .74, P < .Ol)between the increase in HGP and the increase in fasting plasma glucose concentration. When insulin was infused at 3 mu/kg . min to cause a high physiologic increment in the plasma insulin concentration, both the suppression of HGP and the stimulation of tissue glucose uptake were impaired (Figs 1 and 2). At the lower
[sl’ insulin infusion rate (1 mu/kg - min), an impaired suppression of HGP also was observed. In diabetic rats a statistically significant defect in whole-body tissue glucose uptake was not observed. This is explained by the observation that at low plasma insulin concentrations the primary effect of insulin is on the liver, while glucose metabolism by peripheral tissues is minimally or not at all stimulated.2.37 Consistent with this, tissue glucose uptake rose by only 1.8 mg/kg . min or 30% above the basal rate of glucose metabolism in the control group, and 0.7 mg/kg . min or 10% above basal in the diabetic group. Our results are consistent with the presence of hepatic and peripheral insulin resistance in this diabetic rat model. Insulin sensitivity has been examined in the neonatal streptozocin-diabetic rat models, with conflicting results.38-M Although impaired insulin action was demonstrated in some,38,39 but not in all of the previous reports,4’
ROSSE-ITI ET AL
Table 2. Km and V,,,
for Protein Receptor
Activity in Liver and Muscle Tissue in Control, Diabetic, and Metformin-Treated
Diabetic Rats Km
Liver I: Control
13.7 * 0.1
III: Diabetic + metformin
III. Diabetic + metformin
15.5 % 0.3
III: Diabetic + metformin
16.1 + 0.2
NOTE. In liver, tyrosine kinase activity was measured at two insulin concentrations, 25 (top) and 200 (bottom) ng/mL. In muscle tyrosine kinase activity was measured at an insulin concentration of 200 ng/mL.
when the degree of hyperglycemia was comparable to that observed in the present study, decreased sensitivity to insulin was shown at the cellular’* as well as at the whole-body level.” In muscle, the glucose that is taken up may undergo one of two major metabolic fates: oxidation or conversion to glycogen4’ Muscle glycogen synthesis represents 35% of the whole-body glucose disposal during the euglycemic clamp study in normal rats.42.43 To examine the contribution of the later pathway to the insulin resistance observed in diabetic rats, the increment in muscle glycogen, as well as the glycogenic rate, was calculated. In diabetic rats both the increment in cold muscle glycogen concentration and the glycogenic rate during the two-step insulin clamp study were significantly reduced (Fig 3A). If one assumes that all muscle tissue in the rat responds in a similar fashion to leg muscle and that muscle represents 40% of total body mass,4’-43 one can account for 75% of the reduction in total body glucose disposal during the euglycemic insulin clamp study. Two early steps in insulin action include insulin binding to its receptor and activation of receptor protein tyrosine kinase. Insulin binding, measured in WGA-purified receptors from both liver and muscle membranes, was not decreased in diabetic compared with control rats (Fig 3A and B). This is consistent with the majority of published data in diabetic humans,44-46 as well as in animal models’ of type II diabetes mellitus, and clearly indicates that postbinding defects must be responsible for the insulin resistance. Recent studies have focused on the insulin receptor protein tyrosine kinase as a potential mediator of insulin’s intracellular action on glucose metabolism and have sustained the important role of tyrosine kinase in insulin action.8‘2’.47‘49 In the present study, we measured hepatic and muscle protein tyrosine kinase activity in insulin receptors purified from membrane preparations on the artificial substrate, histone HFZB. In diabetic rats, protein tyrosine kinase was significantly impaired in both muscle and liver tissue. In muscle, the ability of insulin to activate the enzyme was apparent at all insulin concentrations, ranging from physiologic (2.5 ng/mL) to pharmacologic (200 ng/mL) levels. In contrast, in liver the decrease in tyrosine kinase activity was
most apparent at insulin concentrations above 25 ng/mL. Since suppression of HGP is largely complete at plasma insulin levels 225 ng/mL, the physiologic significance of the defect in liver receptor protein tyrosine kinase activity remains uncertain. To examine further the nature of the defect in tyrosine activity and its relationship to the impairment in in vivo insulin action Liveweaver-Burk plots were created (Fig 6A and B). No difference in tyrosine kinase affinity (Km) between diabetic and control rats was observed. Rather, the V,,, of the enzyme was markedly reduced, indicating a decrease in the rate of phosphorylation without any change in the affinity for the exogenous substrate. When the V,,, was plotted against the rate of insulin-mediated glucose disposal a strong correlation (r = .64, P -c .Ol) was observed. Although such a correlation does not imply causality, it is at least consistent with the concept that a decrease in insulin receptor protein tyrosine kinase activity may, at least in part, explain the insulin resistance observed in the neonatal streptozotocin rat model of diabetes. No correlation between the V,,, for liver insulin receptor protein tyrosine kinase and the impaired suppression of HGP in diabetic rats was observed. It is of particular note that the V,,, for protein tyrosine kinase activity was strongly correlated (r = .82, P < ,002) with the muscle glycogenic rate in diabetic and control rats. These results suggest that protein tyrosine kinase may play an important role in regulating glycogen formation in muscle in vivo. The defect in receptor protein tyrosine kinase activity is unlikely to explain the impaired suppression of HGP by insulin in diabetic animals, since it was observed only at insulin concentrations that exceeded those necessary to maximally suppress hepatic glucose output.?’ The decreased insulin receptor tyrosine kinase and glycogen synthesis in diabetic rats and its reversal by metformin treatment might be a consequence of the metabolic abnormalities of the diabetic state and of its amelioration with metformin. However, when chronic hyperglycemia was corrected in diabetic rats with phlorizin, an inhibitor of renal glucose reabsorption3 the defect in insulin-mediated glucose disposal was normalized, but glycogen synthesis43 and the insulin receptor tyrosine kinase activity (unpublished observations) were unaffected. At present, the only oral agents available for the treatment of type II diabetes mellitus in the United States are the sulfonylureas. However, in Europe, the biguanide, metformin, has gained widespread usage.50.5’ This drug lowers the fasting plasma glucose concentration and improves glucose tolerance without altering the plasma insulin profi1e.50-52This has led investigators to conclude that the drug works by enhancing insulin sensitivity, and recent studies employing the insulin clamp technique have provided support for this concept.“.53-s5 The results of the current study are consistent with these observations. Thus, metformin treatment of diabetic rats for 6 weeks led to a significant improvement in glucose tolerance without any increase in insulin secretion (Table 1). This improvement in glucose tolerance was associated with a 30% improvement in insulinmediated glucose disposal in diabetic rats. Since neither hepatic glucose production nor net hepatic glycogen forma-
INSULIN ACTION, AND IRTK
tion were altered by metformin treatment, it is obvious that the improvement in insulin action must entirely be due to an increase in peripheral tissue, muscle, and glucose uptake, and this indeed was the case (Figs 1A and 2). At the cellular level the increase in peripheral tissue (muscle) sensitivity to insulin following metformin treatment cannot be explained by an increase in insulin receptor number (Fig 3B). Thus, there was no significant increase in insulin binding in muscle tissue; nonetheless, an increase in the muscle glycogenic rate (267%) and in the rate of whole-body glucose disposal (35%) was readily demonstratable. This is consistent with previous results that have demonstrated that an increase in insulin binding is not necessary for metformin to exert its enhancing effects on glucose metabolism. Thus, metformin has been shown to augment glucose metabolism in adipose tissue56.57 and soleus muscle’R from diabetic animals and man without any change in insulin binding. A blood glucose lowering effect also has been noted in vivo in the absence of any change in insulin receptor number or affinity.22,52,53.57.59 To our knowledge, only one previous study has examined the effect of metformin on insulin receptor protein tyrosine kinase activity.58 Using cultured rat adipocytes that were exposed for short periods of time (20 hour) to metformin in vitro, Jacobs et [email protected]
’ failed to detect any change in tyrosine kinase activity, even though glucose transport increased significantly. In contrast, we observed a marked stimulation of protein tyrosine kinase activity in WGA-purified receptors from both liver and muscle tissue from diabetic animals treated with metformin in vivo for 6 weeks. A LineweaverBurk plot of kinetic data showed that the increase in tyrosine kinase activity was entirely accounted for by an increase in the V,,, of the enzyme without any change in the Km. A number of differences in experimental design could explain the apparently discrepant results reported by Jacobs et a16’ and ourselves. Our studies were performed in vivo in diabetic rats, whereas those of Jacobs et al were performed in vitro on tissue obtained from normal rats. In addition, we examined muscle and liver tissue following long-term (6 weeks) metformin therapy, whereas the previous investigators studied adipose tissue exposed to metformin in vitro for a short period (2 to 20 hours) of time. For these reasons, the results of Jacobs et al and our own should not be viewed as contradictory. Following metformin treatment, diabetic rats demonstrated a marked increase in glycogen synthetic rate that
correlated closely with the increase in whole-body glucose disposal. A stimulatory effect of metformin on glycogen synthesis in soleus muscle also has been reported previously.6’ If enhanced insulin receptor protein tyrosine kinase activity is an important mechanism via which metformin exerts its beneficial effects on glucose metabolism, one might expect to observe a positive correlation between the activity of the enzyme (V,,,) and the glycogenic rate. In fact, these two variables were highly correlated (r = .81, P < .002). The increase in insulin receptor protein tyrosine kinase activity was also closely correlated with the whole-body glucose disposal during the insulin clamp (r = .64, P < .Ol). The improvement in insulin receptor tyrosine kinase activity following metformin treatment was manifested at physiological (5 to 25 ng/mL), as well as at supraphysiological (200 ng/mL), insulin concentrations. These results suggest that a major effect of metformin to improve glucose metabolism in diabetic rats is exerted on the glycogenic pathway in muscle and that enhanced insulin receptor protein tyrosine kinase activity may be an important cellular control mechanism in this process. Last, some comment is warranted concerning the effect of metformin on weight loss and its potential role in the mechanism of action of the drug. Since control and diabetic rats were pair-fed and gained similar amounts of weight, the development of insulin resistance in the later group cannot be explained by differences in food intake or BW. However, metformin-treated diabetic rats, although also pair-fed, gained significantly less weight than both the diabetic and control rats (Table 1). This observation brings up several important issues. First, it raises the possibility that metformin may have a direct thermogenic effect to enhance energy expenditure6’ or to interfere with nutrient absorption.63 These possibilities deserve further exploration. Second, one could question whether some of the stimulatory effects of metformin on total body glucose disposal, glycogenic rate, and insulin receptor protein tyrosine kinase activity might be related to the reduction in body weight per se, as well as to a direct action of the drug on cellular metabolism. Although we cannot exclude the former possibility, metformin has been shown to improve glucose metabolism in diabetic humans and animals in the absence of any change in BW.54,64-65 ACKNOWLEDGMENT
The authors would like to thank Rhonda Wolfe for her expert secretarial
REFERENCES 1. DeFronzo RA, Ferrannini E, Koivisto V: New concepts in the pathogenesis and treatment on non-insulin dependent diabetes mellitus. Am J Med 7452-81, 1983 (suppl 1A)
5. Bevilacqua S, Barrett EJ, Ferrannini E. et al: Hepatic and peripheral insulin resistance following streptozotocin-induced insulin deficiency in the dog. Metabolism 34817-825, 1985
2. Kolterman OG, Gray RS, Griffin J, et al: Receptor and postreceptor defects contribute to the insulin resistance in noninsulin dependent diabetes mellitus. J Clin Invest 68:957-969, 1981
6. Kasuga MY, Akanuma Y, Iwamoto K, et al: Insulin binding and glucose metabolism in adipocytes from streptozotocin-diabetic rats. Am J Physiol235:El75-E182, 1978
3. Rossetti L, Smith D, Shulman GI, et al: Correction of hyperglycemia with phlorizin normalizes tissues sensitivity to insulin in diabetic rats. J Clin Invest 79:1510-1515, 1987 4. Reaven GM, Sageman WS, Swenson RS: Development of insulin resistance in normal dogs following alloxan-induced insulin deficiency. Diabetologia 13:459-462, 1977
7. Kobayashi M, Olefsky JM: Effects of streptozotocin-induced diabetes on insulin binding, glucose transport, and intracellular glucose metabolism isolated rat adipocytes. Diabetes 28:87-95, 1979 8. Chou CK, Dull TJ, Russell DS, et al: Human insulin receptors mutated at the ATP-binding site lack protein tyrosine kinase activity
and fail to mediate post-receptor effects of insulin. J Biol Chem 262:1842-1847, 1987 9. Ellis L, Clauser E, Morgan DO, et al: Replacement insulin receptor tyrosine residues 1162 and 1163 compromises insulinstimulated kinase activity and uptake of 2-deoxylglucose. Cell 45:721-732, 1986 10. Ebina Y, Araki E, Taira M, et al: Replacement of lysine residue 1030 in the putative ATP-binding region of the insulin receptor abolishes insulin- and antibody-stimulated glucose uptake and receptor kinase activity. Proc Nat1 Acad Sci USA 84:704-708, 1987 Il. Russell DS, Gherzi R, Johnson EL, et al: The protein tyrosine kinase activity of the insulin receptor is necessary for insulinmediated receptor down-regulation. J Biol Chem 262:11833-l 1840, 1987 12. LeMarchand-Brustel Y, Gremeuax T, et al: Insulin receptor tyrosine kinase is defective in skeletal muscle of insulin-resistant obese mice. Nature (Lond) 315:676-679, 1985 13. Burant OF, Treutelaar MK, Buse MG: Diabetes-induced functional and structural changes in insulin receptors from rat skeletal muscle. J Clin Invest 77:260-270, 1986 14. Freidenberg GR, Henry RR, Klein HH, et al: Decreased kinase activity of insulin receptors from adipocytes of non-insulin dependent diabetic subjects. J Clin Invest 79:240-250, 1987 15. Comi RJ, Grunberger G, Gorden P: Relationship of insulin binding and insulin-stimulated tyrosine kinase activity is altered in type II diabetes. J Clin Invest 79:453-462, 1987 16. Caro JR, Sinna MK, Raju SM, et al: Insulin receptor kinase in human skeletal muscle from obese subjects with and without noninsulin dependent diabetes. J Clin Invest 79:1330-1337, 1987 17. Grunberger G, Zick Y, Gorden P: Defect in phosphorylation of insulin receptors in cells from an insulin-resistant patient with normal insulin binding. Science 223:932-934, 1984 18. Kadowaki TM, Kasuga M, Akanuma Y, et al: Decreased autophosphorylation of the insulin receptor kinase in streptozotocindiabetic rats. J Biol Chem 259:14208-14216, 1984 19. Gherzi R, Andraghetti G, Ferrannini E, et al: Insulin receptor autophosphorylation and kinase activity in streptozotocin diabetic rats. Effect of a short fast. Biochem Biophys Res Commun 140:850856.1986 20. Freidenberg CR, Klein HH, Cordera R, et al: Insulin receptor kinase activity in rat liver: Regulation by fasting and high carbohydrate feeding. J Biol Chem 260:12444- 12453, 1985 21. Grigorescu F, Flier JS, Kahn CR: Defect in insulin receptor phosphorylation in erythrocytes and fibroblasts associated with severe insulin resistance. J Biol Chem 259: 15003-l 5006, 1984 22. Gin H, Messerchmitt C, Brottier E, et al: Metformin improved insulin resistance in type I insulin-dependent, diabetic patients. Metabolism 34:923-925, 1985 23. Nosadini R, Avogaro A, Trevisan R, et al: Effect of metformin on insulin-stimulated glucose turnover and insulin binding to receptors in type 2 diabetes. Diabetes Care 10:62-67, 1987 24. Jackson RA, Hawa MI, Jaspan JB, et al: Mechanism of metformin aciton in non-insulin dependent diabetes. Diabetes 36:632640, 1987 25. Weir GC, Clore ET, Zmachinski CJ, et al: Islert secretion in a new experimental model for non-insulin-dependent diabetes. Diabetes 30:590-595, 1981 26. Smith D, Rossetti L, Ferrannini E, et al: In vivo glucose metabolism in the awake rate: Tracer and insulin clamp studies. Metabolism 36:1176-l 174, 1987 27. Shulman GI, Rossetti L, Rothman DL, et al: Quantitative analysis of glycogen repletion by nuclear magnetic resonance spectroscopy in the conscious rat. J Clin Invest 80:387-393, 1987 28. Cordera R, Gherzi R, DePirro R, et al: Inhibition of insulin
AOSSElTl ET AL
and epidermal growth factor receptor autophosphorylation by human polyclonal IgG. Biochem Biophys Res Commun 132:99 I- 1000, 1985 29. Harrison LC, Itin A: Purification of the insulin receptor from human placenta by chromatography on immobilized wheat germ lectin and receptor antibody. J Biol Chem 255:12066- 12072, 1980 30. Scatchard G: The attractions of protein for small molecules and ions. Ann NY Acad Sci 51:660-672,1949 31. Gherzi R, Cordera R, Andraghetti G, et al: Regulation of insulin receptor-associated tyrosine kinase by a polyclonal IgG. Mel Cell Endocrinol53:9-14, 1987 32. DeFronzo RA, Gunnarson R, Bjorkman 0, et al: Effect of insulin on peripheral and splanchnic glucose metabolism in noninsulin-dependent (type II) diabetes mellitus. J Clin Invest 76:149155,1985 33. Firth RG, Bell PM, Marsh HM, et al: Postprandial hyperglycemia in patients with noninsulin-dependent diabetes mellitus. Role of hepatic and extrahepatic tissues. J Clin Invest 77:1525-1532. 1986 34. Kashiwagi A, Urso MA, Andrews J, et al: In vitro insulin resistance of human adipocytes from subjects with non-insulindependent diabetes mellitus. J Clin Invest 72:1246-1254, 1983 35. Bogardus CS, Lillioja S, Stone K, et al: Correlation between muscle glycogen synthase activity and in vivo insulin action in man. J Clin Invest 73:1185-1190, 1984 36. Portha B, Levacher C, Picon L, et al: Diabetogenic effect of streptozotocin in the rat during the perinatal period. Diabetes 23:889-895, 1974 37. Rizza R, Mandarin0 L, Gerick J: Dose-response characteristics for effects of insulin on production and utilization of glucose in man. Am J Physiol240:E630-E639, 198 I 38. Levy J, Gavin JR III, Faust0 A, et al: Impaired insulin action in rats with non-insulin-dependent diabetes. Diabetes 33:901-906, 1984 39. Blonde1 0, Bailbe’ D, Portha B: Relation of insulin deficiency to impaired insulin action in NIDDM adult rats given streptozotocin as neonates. Diabetes 38:610-617, 1989 40. Kergoat M, Portha B: In vivo hepatic and peripheral insulin sensitivity in rats with non-insulin-dependent diabetes induced by streptozocin. Diabetes 34: 1120- 1126,1985 41. DeFronzo RA, Jacot E, Jequier E, et al: The effect of insulin on the disposal of intravenous glucose: Results from indirect calorimetry and hepatic and femoral venous catheterization. Diabetes 30:1000-1007,198l 42. Rossetti L: Normalization of insulin sensitivity with lithium in diabetic rats. Diabetes 38:648-652, 1989 43. Rossetti L, Laughlin M: Correction of chronic hyperglycemia with vanadate, but not with phlorizin, normalizes in vivo glycogen repletion and in vitro glycogen synthase activity in diabetic skeletal muscle. J Clin Invest 84:892-899, 1989 44. Pederson 0, Hjolland E: Insulin receptor binding to fat and blood cells and insulin action in fat cells from insulin-dependent diabetics. Diabetes 31:706-715, 1982 45. Kashiwagi A, Verso MA, Andrews J, et al: In vitro insulin resistance of human adipocytes isolated from subjects with noninsulin-dependent diabetes mellitus. J Clin Invest 72:1246-1245, 1984 46. Bolinder J, Ostman J, Arner P: Postreceptor defects causing insulin resistance in normoinsulinemic noninsulin-dependent diabetes mellitus. Diabetes 31:911-916, 1982 47. Caro JF, Ittoop 0, Pories WY, et al: Studies on the mechanism of insulin resistance in the liver from humans with non-insulin dependent diabetes: Insulin action and binding in isolated hepatocytes, insulin receptor structure and kinase activity. J Clin Invest 78:249-257, 1986
INSULIN ACTION, AND IRTK
48. Sinha MK, Pories WJ, Flinckinger EG, et al: Insulin receptor kinase activity of adipose tissue from morbidly obese humans with and without non-insulin dependent diabetes. Diabetes 36:620-625, 1987 49. Gkamoto M, White MF, Maron R, et al: Autophosphorylation and kinase activity of insulin receptor in diabetic rats, Am J Physiol251:E542-E550, 1986 50. Hermann LS: Metformin. A review of its pharmacological properties and therapeutic use. Diabete Metab 5:233-245, 1979 5 1. Turner RC: UK prospective diabetes study. II. Reduction in HbAlc with basal insulin supplement, sulfonylurea or biguanide therapy in maturity-onset diabetes. Diabetes 34:793-798, 1985 52. Prager R, Schernthaner G: Insulin receptor binding to monocytes, insulin secretion and glucose tolerance following metformin treatment. Diabetes 33:1083-1036, 1983 53. Prager R, Schernthaner G, Graf H: Effect of metformin on peripheral insulin sensitivity in non insulin dependent diabetes mellitus. Diabetes Metab 12:346-350, 1986 54. Nosadini R, Avogaro A, Tevisan R, et al: Effect of metformin on insulin-stimulated glucose turnover and insulin binding to receptors in type II diabetes. Diabetes Care 10:62-67, 1987 55. Pagan0 G, Tagliaferro V, Carta Q, et al: Metformin reduces insulin requirement in type I (insulin dependent) diabetes. Diabetologia 24:35 l-354, I983 56. Cigolini M, Bosello 0, Zancanaro C, et al: Influence of metformin on metabolic effect of insulin in human adipose tissue in vitro. Diabetes Metab lo:31 1-315, 1984
57. Fantus IG, Brosseau R: Mechanism of action of metformin: Insulin receptor and post receptor effects in vitro and in vivo. J Clin Endocrinol Metab 63:898-905, 1986 58. Lord JM, Puah JA, Atkins TW, et al: Postreceptor effect of metformin on insulin action in mice. J Pharm Pharmacol 37:821823,198s 59. Lord JM, Atkins TW, Bailey CJ: Effect of metformin on hepatocyte insulin receptor binding in normal, streptozotocin diabetic and genetically obese diabetic (ob/ob) mice. Diabetologia 25:108-113, 1983 60. Jacobs DB, Hayes GR, Truglia JA, et al: Effects of metformin on insulin receptor tyrosine kinase activity in rat adipocytes. Diabetologia 29:798-801, 1986 61. Bailey CJ, Puan JA: Effect of metformin on glucose metabolism in mouse soleus muscle. Diabetes Metab 12:2 12-2 18, 1986 62. Bailey CJ, Flatt PR, Ewan C: Anorectic effect of metformin in lean and genetically obese hyperglycemic (ob/ob) mice. Arch Intern Pharmacodynamics 282:233-239,1986 63. Larch E: Inhibition of intestinal absorption and improvement of oral glucose tolerance by biguanides in the normal and in the streptozotocin diabetic rat. Diabetologia 7:195-203, 1971 64. Siitonen 0, Antti A, Huttunen JK, et al: Effect of discontinuation of biquanide therapy on metabolic control in maturity-onset diabetes. Lancet 217:220, 1980 65. Higginbothan L, Martin FIR: Double-blind trial of metformin in the therapy of non-ketotic diabetic. Med J Aust 2: 154- 156, 1979