| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Kolling Institute of Medical Research (J.E.G., P.J.D.D., R.C.B.), University of Sydney, Royal North Shore Hospital, St. Leonards, Sydney, New South Wales 2065 Australia; and Graduate School of Agriculture and Life Sciences (S.-I.T.), University of Tokyo, Tokyo 113-8657, Japan
Address all correspondence and requests for reprints to: Jenny E. Gunton, M.D., Kolling Institute of Medical Research, Royal North Shore Hospital, St. Leonards, Sydney, New South Wales 2065, Australia. E-mail: jennyeg{at}hotmail.com.
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Although metformin was first used over 40 yr ago, the mechanism(s) of action is not fully elucidated (1, 6, 7).
Whether metformin acts as an insulin-sensitizing agent is debated. Studies in subjects with abnormal glucose tolerance consistently show decreased fasting blood glucose, with either a modest decrease or no change in fasting insulin (8, 9). Studies in normal subjects show no change in glucose and decreased fasting insulin (2, 10). These changes suggest improved insulin sensitivity, as measured by the Homeostasis Model Assessment (11).
Studies using euglycemic hyperinsulinemic clamps give variable results. Some show an improvement in insulin sensitivity (12, 13, 14, 15, 16, 17), but others do not (18, 19, 20, 21, 22).
Most (12, 13, 14, 15, 18, 19, 23) but not all (8, 24) studies show that metformin decreases endogenous glucose production (EGP) by the liver. Insulin binding to the insulin receptor (IR) probably does not change (25, 26), with increases seen in some studies thought to be via decreased glucotoxicity. It is usually agreed that metformin does not alter ß-cell function (27, 28).
A number of cellular responses to metformin have been reported, including increased IR tyrosine kinase activity (KA) (24, 29), no change in IR-KA, augmented GLUT translocation and action (30, 31), activation of AMP-activated protein kinase (7), and reduced hepatic gluconeogenesis in cultured hepatocytes accompanied by reduced mRNA expression for genes involved in fatty acid oxidation and gluconeogenesis (32), inhibition of mitochondrial respiratory chain activity (33, 34), prevention of hyperinsulinemia- induced decrease in IR responsiveness (7), improvement in cell membrane physiology (35), and a decrease in free fatty acid flux (36).
There are few reports regarding effects of metformin on the insulin receptor-signaling pathway in hepatic models, and no data have been presented for normal human liver. Zhou et al. (7) reported that metformin activated AMP kinase in primary human hepatocytes, although the data were not shown in the paper. The aims of these studies were to determine the mechanism(s) of action of metformin by examining the insulin-signaling cascade in hepatic models and IR partially purified from normal human liver tissue. Our hypothesis was that metformin would significantly alter proximal insulin signaling in human hepatic models.
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Except where specified, reagents were purchased from Sigma- Aldrich(St. Louis, MO). Cell culture experiments were performed in Huh7 cells or primary human hepatocytes. Huh7 cells were a kind gift from Prof. Ann Simpson (University of Technology-Sydney, Sydney, Australia). Huh7 cells are a human hepatoma-derived cell line and were cultured in 80-cm2 flasks (Nunc, Copenhagen, Denmark) in DMEM supplemented with 5% fetal bovine serum and 1% glutamine. Cells were washed with PBS and placed in serum-free media 24 h before experiments.
Primary human hepatocytes were obtained by collagenase digestion of normal liver tissue. Human liver was obtained during surgery from patients having a hemihepatectomy for isolated metastasis or noninfectious hepatocellular carcinoma. All patients gave informed consent, and the study was approved by the institutional ethics committee. Normal tissue was dissected from the hemihepatectomy specimen by scalpel at a site distant from the lesion. Samples were cannulated and perfused with 10,000 U heparin. This was followed by hand perfusion with buffer containing 10 mmol/liter HEPES, 137 mmol/liter NaCl, 2.68 mmol/liter KCl, 0.7 mmol/liter NaH2PO4, 10 mmol/liter glucose, and 0.5 mmol/liter EDTA (pH 7.6). Liver that did not clear was removed and frozen in liquid nitrogen for use in IR kinase assays. The liver was then perfused with warmed S77 buffer with 0.05% collagenase for 15–20 min. Cells were plated on 60-mm culture dishes (Nunc) at 2.5 million to 3 million cells per dish. After allowing 3 h for attachment, cells were washed and placed in serum-free media for 24 h.
Following the described treatments, Huh7 and human hepatocytes were washed with ice-cold PBS and harvested by scraping into 0.75 ml ice-cold lysis buffer [10 mmol/liter Tris HCl, 1% Triton X-100, 0.5% Nonidet P-40, 150 mmol/liter NaCl, 10 mmol/liter Na-orthophosphate, 10 mmol/liter Na-pyrophosphate, 10 mmol/liter Na-orthovanadate, 100 nmol/liter Na-fluoride, 1 mmol/liter EDTA, 1 mmol/liter EGTA, and complete protease inhibitor tablets (one per 50 ml, pH 7.6); Roche Molecular Biochemicals, Mannheim, Germany]. Lysates were homogenized and microfuged at 4 C at 10,000 rpm for 10 min to pellet out insoluble material. Protein content was measured by Bradford assay, using a protein kit (Bio-Rad Laboratories, Inc., Hercules, CA) (37). Samples were used immediately or stored at -80 C until use.
Immunoprecipitations
Immunoprecipitation of IR was performed by mixing 2 mg protein from homogenates, 4.8 µg rabbit polyclonal ß-subunit antibody (Upstate Biotechnology, Inc.,Lake Placid, NY) and 40 µl protein A beads (Pharmacia Biotech,Uppsala, Sweden). The mixture was rotated at 4 C overnight. Samples were eluted with reducing Laemmli buffer, and proteins were separated by 8% SDS-PAGE. After transfer to Hybond-C membrane (Life Sciences Amersham, Uppsala, Sweden) and blocking [3% BSA in Tris-buffered saline with 0.1% Tween 20, pH 7.4 (TBST)], activation of the IR-ß was assessed by measuring tyrosine phosphorylation with a mouse monoclonal antiphosphotyrosine antibody (PY20, Transduction Laboratories, Inc., Lexington, KY) at 1:2000 in TBST, followed by antimouse IgG horseradish-peroxidase conjugate (Amersham Pharmacia, Uppsala, Sweden) 1:2000 in TBST, and detected using chemifluorescence (ECL + Amersham Pharmacia Biotech, London, UK). The blot was scanned by FLA3000 (Fuji Photo Film Co., Ltd., Tokyo, Japan) using Image-Reader and ImageGuage software (Fuji Photo Film Co., Ltd.). After stripping with Restore stripping buffer (Pierce Chemical Co., Rockford, IL), total IR-ß was detected with mouse monoclonal ß-subunit-IR antibody (Oncogene, Boston, MA) diluted 1:300 in TBST, with detection as described above.
IR substrates (IRS)-1 and IRS-2 were immunoprecipitated from 5 mg protein from Huh7 homogenates with 9 µg rabbit polyclonal antibody against IRS-1 or IRS-2 (antibodies for initial experiments were made by S.-I.T., and antibodies for later experiments were purchased from Transduction Laboratories, Inc. and Upstate Biotechnology, Inc., respectively) and 40 µl protein A beads. After 6% SDS-PAGE and transfer, tyrosine phosphorylation was detected as for IR. After stripping, total IRS was detected using the respective primary antibody (1:250 in TBST), followed by antirabbit IgG-horseradish-peroxidase (Amersham) at 1:6000 in TBST.
IR-KA
Partially purified IR from normal human liver tissue were assessed for regulation by metformin and insulin in an in vitro protein kinase assay. To make lysates, the liver was thawed to 4 C, minced, and homogenized in liver lysis buffer. IRs were partially purified from liver lysate with wheat germ agglutinin agarose (WGA) (Amersham) (38). Briefly, liver tissue was homogenized and applied three times to a WGA column. The column was washed with 25 mmol HEPES with 0.1% Triton-X. IRs were eluted with the same buffer supplemented with 0.6M N-acetyl glucosamine. After purification, IRs were treated with metformin and/or insulin, followed by the addition of phosphorylation buffer [0.1% Triton X-100, 50 mmol/liter HEPES, and 5 mmol/liter MnCl2 (pH 7.6)], 0.1 mmol/liter ATP, and 30 µCi 
32P-ATP [370 mBq/ml (10 mCi/ml); Geneworks Australia, Adelaide, SA, Australia]. Samples were incubated at 4 C for 90 min. The reaction was terminated with the addition of 10 mmol/liter Na-pyrophosphate, 4 mmol/liter EGTA, 100 mmol/liter Na-fluoride, 2 mmol/liter Na-orthovanadate, 2 mmol/liter phenylmethylsulfonyl fluoride, and 50 mmol/liter HEPES (pH 7.6). Samples were separated using 8% SDS-PAGE, dried, and exposed to a phosphor imaging screen and scanned using the FLA3000 (Fuji Photo Film Co., Ltd.).
In separate experiments, the IR-KA experiments were repeated using antibodies to capture IR. IRs were purified from 2 mg total lysate from normal human liver tissue with 4.8 µg anti-IR antibody (Upstate Biotechnology, Inc.) and 40 µl protein A beads. Although the IRs were attached to the beads, the same phosphorylation reaction was carried out. IRs were removed from beads by the addition of reducing sample buffer and boiling, and then detection was as for the WGA preparation.
Tyrphostin inhibition of IR-KA
Tyrphostin-AG1024 is a protein tyrosine kinase inhibitor with activity that has specific inhibitory activity against the insulin receptor and type 1 IGF receptors (39). Tyrphostin was added to cells in culture for 1 h at the concentrations specified.
2-Deoxyglucose uptake
Huh7 cells were grown to confluence in 6-well plates and rendered serum free for 24 h. They were then washed twice and placed into warmed Krebs buffer for 2 h before glucose uptake studies. Tyrphostin and/or metformin was added to the relevant wells for 60 min and insulin for 10 min. Following this, 0.5 µCi/ml 2-deoxyglucose was added to each well (specific activity 29.8 Ci/mmol; Geneworks). After 5 min, 5% unlabeled glucose was added to stop the reaction. Cells were then washed twice in 5% glucose and lysed by the addition of 0.5% SDS. Three milliliters Ultima Gold scintillant (Packard, Groningen, Netherlands) was added, samples were vortexed, and radioactivity was counted in an A290001 ß-counter (Packard).
Glucose transporter translocation (plasma membrane lawns)
Translocation of glucose transporter (GLUT)-1 was assessed by quantitating GLUT-1 immunofluorescence in plasma membrane lawns, according to the method of Whitehead et al. (40) and Robinson et al. (41). In brief, Huh7 cells were grown on glass coverslips and treated with buffer alone (control), metformin at 1 µg/ml for 1 h, insulin at 100 ng/ml for 15 min, or metformin combined with insulin.
The cells were then sonicated, using a probe sonicator (Kontes, Vineland, NJ). This leaves a lawn of plasma membrane attached to the coverslip but removes the rest of the cell. The coverslips were blocked in blocker (PBS with 5% milk), rinsed in PBS, and incubated with GLUT-1 antibody (rabbit polyclonal antibody at 1:250 dilution in blocker) at room temperature for 1 h. After three washes with PBS, the coverslips were incubated with fluorescein-isothiocyanate-conjugated secondary antibody. Both antibodies were a kind gift from Prof. David James (Garvan Institute of Medical Research, Sydney, Australia). Coverslips were washed three times in PBS, mounted onto glass slides, and viewed on a Axiovert fluorescence microscope (Carl ZeissSydney, NSW, Australia) with attached MRC-600 laser confocal imaging (Bio-Rad Laboratories, Inc.). Each condition was performed in triplicate. Several random images were taken from each coverslip by an observer blinded to the treatment conditions. The images were then quantitated using ImageGauge (Fuji Photo Film Co., Ltd.) by placing 11 identical squares per image, 1 for background and 10 centered over individual cell membranes, by an observer who was blinded to the treatment condition. After each image was corrected for its original background, the data were compiled and assessed for significance.
Statistics and presentation of results
Figures show representative blots. Each experiment was performed at least three times. Quantitation of bands was performed using ImageGauge software (Fuji Photo Film Co., Ltd.), and data are corrected for background. Statistical analysis was carried out with StatView, version 5 (SAS Institute, Inc., Cary, NC). Results were tested for statistical significance using ANOVA, followed by Fisher’s protected least significant difference test, except where ANOVA for repeated measures is specified. All graphs show means ± 1 SD. Graphs show phosphorylation data corrected for total protein, expressed as a percentage of control values. Throughout the text and figures: a, P < 0.05; b, P < 0.01; c, P < 0.001; d, P < 0.0001.
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Huh7 cells responded to approximately physiological concentrations of insulin (100 pg/ml to 100 ng/ml insulin in the normal portal circulation) (42), as shown in Fig. 1
. A time course of insulin-induced tyrosine phosphorylation of the IR ß-subunit in Huh7 cells is shown in Fig. 1A, with total IRs in the lower panel. The increase in tyrosine phosphorylation was significant, corrected for total IR, within 1 min (P = 0.039), reaching maximum by 5–10 min (P = 0.0084 for the 5-min time point), and returning to baseline by 15 min (P > 0.5, compared with control) (Fig. 1C).
|
. Cells were treated for 10 min. Treatment with 0.1 ng/ml of insulin caused a 35% increase in IR tyrosine phosphorylation (Fig. 1D) (P = 0.0514), and 0.3 ng/ml resulted in a statistically significant increase (P = 0.0287). Maximal insulin response was reached by 1 ng/ml (P = 0.0016) in Huh7 cells.
Following IR studies, insulin activation of IRS-1 and IRS-2 was examined. Figure 1E shows a time course of insulin-induced tyrosine phosphorylation of IRS-1. Treatment with 100 ng/ml of insulin significantly increased tyrosine phosphorylation of IRS-1 by 2.5 min (114% increase, P = 0.0053). Activation peaked at 5 min (234% increase, P < 0.0001) and was no longer significant by 20 min (Fig. 1G, quantitation of three experiments). Figure 1F shows a time course of insulin-induced activation of IRS-2 in Huh7 cells. Treatment was with 100 ng/ml insulin for the times shown. Quantitation of three experiments is shown in Fig. 1H. IRS-2 tyrosine phosphorylation increased 85% above baseline by 1 min (P = 0.08), was significantly higher at 2.5 min (2.8-fold increase, P = 0.0008) and 5 min (3-fold increase P = 0.0004), and did not differ from baseline by 10 min (P > 0.1). This set of experiments confirmed that Huh7 cells respond to approximately physiological concentrations of insulin, which has not been previously reported.
Metformin treatment of Huh7 cells increases insulin receptor activation
Metformin increased IRß-subunit tyrosine phosphorylation in Huh7 cells (Fig. 2). The concentrations used were selected on the basis of studies showing that metformin is physiologically active in humans in concentrations in the order of 1 µg/ml (6.04 µmol/liter) (43). When analyzed by ANOVA for repeated measures, the overall metformin data and insulin data are significant for metformin-induced IR activation (P = 0.011) but not for a further significant increase in IR activation with combined metformin and insulin (P = 0.20).
|
), metformin significantly increased tyrosine phosphorylation of IR at doses of 100 ng/ml (78% increase above baseline, P = 0.0007) and 1 µg/ml (52% increase above baseline, P = 0.0208). The highest dose (10 µg/ml) did not cause a significant increase (29%, P = 0.18). Metformin treatment for 30 min did not significantly augment submaximal (Fig. 2B) or maximal (Fig. 2C) insulin-induced IR activation. The maximal increase in IR activation with metformin was not as great as that seen with insulin (78% greater than baseline with metformin vs. 141% greater than baseline with insulin P = 0.0332 for metformin vs. insulin increase).
A time course of metformin activation of IRß is shown in Fig. 2E. All time points were treated with 1 µg/ml metformin. Quantitation of three experiments is shown in Fig. 2F. Metformin significantly increased IRß phosphorylation within 15 min of treatment, and the increase persisted to 24 h. The time course of activation was clearly different from that seen with insulin (Fig. 1A), and the magnitude of increase in IRß tyrosine phosphorylation over baseline was consistently smaller than that seen with insulin (P < 0.05).
Metformin and IRS activation
The effect of 30 min of treatment with metformin on tyrosine phosphorylation of IRS-1 and IRS-2 in Huh7 cells is shown in Fig. 3. No effect of metformin on tyrosine phosphorylation of IRS-1 was seen at any dose (Fig. 3A). However, metformin significantly increased IRS-2 tyrosine phosphorylation, as shown in Fig. 3B. Quantitation of repeated experiments for IRS-1 and IRS-2 is shown in Fig. 3C. There was a significant increase in IRS-2 activation with 100 ng/ml metformin (P = 0.0007), with maximal IRS-2 tyrosine phosphorylation at 1 µg/ml (P < 0.0001). Unlike the effect on IRß, the effect of 10 µg/ml metformin on IRS-2 phosphorylation was significant (P = 0.0007). Similar to the effect on IR phosphorylation, the maximal stimulatory effect of metformin on IRS-2 was smaller than that induced by insulin (
60% vs. 150% increase, P < 0.05).
|
. Increased tyrosine phosphorylation was seen by 10 min. Figure 3E shows quantitation of three experiments and shows that metformin significantly increased IRS-2 activation within 10 min, with the effect persisting to 24 h. The onset and prolonged duration are similar to that seen with metformin and IRß. Metformin treatment of primary human hepatocytes and IR activation
Because the previous studies were performed in Huh7 cells, which are a tumor-derived cell line, it was important to confirm the results in primary human hepatocytes. Figure 4A demonstrates that the metformin increased IRß tyrosine phosphorylation in primary human hepatocytes. Similar to the effect seen in Huh7 cells, the combination of metformin and insulin did not result in an additive effect in primary human hepatocytes. Figure 4B shows quantitation of four experiments. Metformin at 1 µg/ml caused a 173% increase in IRß phosphorylation (P = 0.0125). Treatment with insulin at 100 ng/ml for 5 min caused a 374% increase in IR phosphorylation (P = 0.0003), which was of significantly greater magnitude than that seen with metformin alone (P = 0.0095), a result that is again similar to that seen in the Huh7 cells.
|
To examine whether the ability of metformin to increase IRß-subunit tyrosine phosphorylation was a direct effect, or secondary to other cellular processes, intrinsic IR-KA was assessed (Fig. 5). To diminish the likelihood that IR phosphorylation depended on another protein that was precipitated by WGA, experiments were repeated using IR-antibody and protein A beads to capture IR. All IR-KAs were performed in normal human liver lysates. Figure 5A shows a dose-response experiment, in triplicate, for insulin-induced increase in IR-KA in IR partially purified with anti-IRß-subunit-Ab and protein A beads. Quantitation of experiments showed that insulin increased IR32P incorporation by approximately 500% with WGA-purified (P < 0.0001, shown in Fig. 5C) and antibody-purified IR (P = 0.036, shown in Fig. 5D), with maximal response at 5 µg/ml for both methods.
|
). Metformin increased IR32P incorporation to 100–150% above baseline with 300 ng/ml or 1 µg/ml with both WGA-purified (P = 0.001, Fig. 5E) and antibody-purified IR (P < 0.0001, Fig. 4F). The maximal metformin-induced change in IR-KA was smaller than that seen with insulin (150% vs. 500% increase). The highest dose of metformin (3 µg/ml) was not stimulatory. The lack of effect of high doses was seen in repeated experiments. No experiments demonstrated any additive effect of metformin and insulin on IR-KA (data not shown). Tyrphostin AG1024 inhibition of the IR
Effects of tyrphostin-AG1024 on IR activation and 2- deoxyglucose uptake were examined. Tyrphostin-AG1024 is a protein tyrosine kinase inhibitor, with activity that is specific to the IR and the type1 IGF receptor (39). The aim of these experiments was to determine whether IR inhibition would block the effect of metformin. The reported 50% inhibitory concentration for tyrphostin-AG1024 against the IR in the literature was 80 ± 21 µg/ml (39). Accordingly, experiments were done with a range of doses from 30 µg/ml to 300 µg/ml. Figure 6A shows the effect of increasing concentrations of tyrphostin-AG1024 on basal IR phosphorylation, and Fig. 6B shows the effect on insulin-induced IR phosphorylation. These experiments showed that AG1024 effectively inhibited IR tyrosine phosphorylation both in the baseline state and following insulin stimulation.
|
. Figure 6D quantitates the percentage decrease in phosphorylation induced by tyrphostin for basal or insulin- and metformin-treated cells, with each treatment before inhibition being set at 100% for comparison. Tyrphostin-AG1024 treatment was equally effective at decreasing phosphorylation of IRß whether in the basal state or after treatment with insulin or metformin.
Metformin and insulin-induced 2-deoxyglucose uptake in Huh7 cells are shown in Fig. 6E. Metformin significantly increased glucose uptake at 0.1 µg/ml (170% ± 20% of baseline, P = 0.0028), with maximal effect at 1 µg/ml (225% ± 60% of baseline, P < 0.0001). A dose of 10 µg/ml increased glucose uptake to 170% of baseline, P = 0.0032. Insulin treatment at 100 ng/ml increased glucose uptake to 270% of baseline (P < 0.0001). There was no additive effect of metformin and insulin (P > 0.2).
Tyrphostin-AG1024 is a known inhibitor of IR activation. Accordingly, after the above experiments confirmed that metformin-induced IR phosphorylation was also inhibited, glucose uptake was examined to determine whether tyrphostin-AG1024 inhibited metformin induced glucose uptake.
Figure 6F demonstrates the increase in 2-deoxyglucose uptake into Huh7 cells following treatment with metformin or insulin and the effects of tyrphostin-AG1024. Tyrphostin-AG1024 decreased metformin-induced glucose uptake to the same level as basal +tyrphostin-treated or insulin+tyrphostin-treated cells.
GLUT-1 translocation to the plasma membrane
The effect of metformin and/or insulin treatment on the translocation of GLUT-1 to the plasma membrane was assessed in Huh7 cells, using plasma membrane lawns and immunofluorescent detection. Results of the experiments are shown in Fig. 7. GLUT-1 is highly expressed in liver and was originally isolated from HepG2 human hepatoma cells (44). However, in HepG2 cells, the GLUT-1 is present on the plasma membrane, and it is not translocated in response to insulin. The plasma membrane lawns were prepared to examine the mechanism of increased glucose uptake in Huh7 cells. A representative picture of control plasma membrane is shown in Fig. 7A, metformin treated (1 µg/ml for 1 h) in Fig. 7B, insulin treated in Fig. 7C, and combined insulin and metformin treated plasma membrane is shown in Fig. 7D. Quantitations were performed by an investigator who was blinded to treatment condition and are shown in Fig. 7E. In two separate experiments, treatment with metformin resulted in a significant increase in GLUT-1 at the plasma membrane (P < 0.000001). Insulin and metformin combined with insulin resulted in increases of similar magnitude (both P < 0.000001).
|
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Metformin did not have a significant additional effect on IR phosphorylation in the presence of insulin, either at low or high dose. The lack of effect of metformin in the presence of insulin is consistent with the negative results of many of the hyperinsulinemic euglycemic clamp studies (18, 19, 20, 21, 22, 45). However, many studies have shown significant improvements in insulin-mediated glucose disposal with metformin (3, 7, 10, 13, 14, 16, 17, 20, 23, 45). Metformin activates AMP-activated protein kinase (7), increases hepatic expression of glucokinase and liver-type pyruvate kinase, and decreases expression of glucose-6-phosphatase, phosphoenolpyruvate carboxykinase, carnitine palmitoyltransferase-I, and mitochondrial hydroxymethylglutaryl-CoA synthase, which are involved in gluconeogenesis and fatty acid oxidation (32).
These changes would be expected to increase insulin sensitivity, and it is possible that metformin may have an additive effect in combination with insulin when examined in longer-term treatments. In the studies presented herein, the aim was to examine short time courses, with the specific aim of avoiding transcriptional and posttranslational effects.
Most studies showing a significant effect of metformin on insulin-induced glucose disposal used long-term treatment (weeks to months), and in subjects with diabetes, there were significant improvements in glucose control. In most of the studies, there was also a decrease in body weight. Decreased glucotoxicity, weight loss, and alterations in gene expression occurring downstream of IR activation may play a role in improved insulin sensitivity in these studies.
Other authors have examined the effect of metformin on IR-KA, with varying results. There are three studies examining the effect of metformin on IR-KA in adipocytes (30, 46). One group examined adipocytes from Zucker rats, with short-term treatment (30) or 3 wk of metformin treatment (46) and found no effect. Jacobs et al. (47) examined short-term (2 h) and 24-h metformin treatment on rat adipocytes and found no significant effect.
A number of studies have examined the effect of metformin on IR-KA in nonadipose tissue (24, 47, 48). Santos et al. (49) found that 10 wk of metformin improved basal and insulin-stimulated IR-KA in erythrocytes from humans with type 2 diabetes. Another study (49) showing that treatment with a sulfonylurea, glyburide, increased IR-KA suggesting that any improvement in glucotoxicity, however achieved, can improve IR-KA. Dominguez et al. (48) studied rat vascular smooth muscle cells and found a significant effect following 24 h of treatment. Rossetti et al. (24) examined the effect of chronic treatment with metformin on muscle IR isolated from diabetic rats and found a significant increase in IR-KA. Stith et al. (43) found a small but significant increase in kinase activity with 1 µg/ml metformin (of the order of 25–40%), using synthetic intracellular portions of the ß- subunit. They found that metformin was active following 15 min of exposure, which is consistent with our results using whole receptors. In that study, higher doses of metformin (10 µg/ml) also activated the epidermal growth factor receptor. It is possible that metformin may have activated EGF-R, and/or other protein tyrosine kinases, especially at higher concentrations in our studies. However, the ability of tyrphostin AG1024 to completely block metformin-induced deoxyglucose uptake suggests that the metabolic effects of metformin were mediated by the insulin receptor itself.
In the Huh7 cells, 24 h of treatment with 10 µg/ml metformin appeared to cause cell toxicity, as demonstrated by increased numbers of detached cells, and this may have contributed to the apparent biphasic dose response. However, other investigators have found a similar biphasic effect of metformin on IR-KA, for which the mechanism remains unclear (43).
To summarize the metformin and IR-KA literature, there is no effect of metformin on IR-KA in any studies of adipocytes (30, 46, 50). The reasons for this are unclear. There was a significant effect on kinase activity in other systems, with most studies involving long-term (more than 24 h) treatment before IR isolation (24, 46, 47), rather than IR isolation, and then treatment with metformin.
The results of our study using normal human liver complement this result and show that metformin increased IR-KA in receptors isolated before exposure to metformin. There are few, if any, type1 IGF receptors in liver, which diminishes one potential confounder. Our results provide the first evidence in normal human liver tissue that metformin increases IR-KA, and by isolating receptors before treatment, removes confounding effects of transcriptional or posttranslational changes.
Metformin is taken up by hepatocytes via an unknown mechanism, and hepatocytes achieve higher concentrations than serum (50). When covalently linked to beads, which prevent cell entry, metformin was no longer able to stimulate Xenopus oocyte differentiation (51). This is consistent with either the process of binding rendering metformin inactive or metformin requiring internalization for activity. There are no reports regarding intracellular concentrations of metformin in other tissues, but the increased concentrations in liver may, in part, explain the particular effect of metformin on hepatic glucose handling.
The metformin time course was different from that seen with insulin, with a slower onset and markedly prolonged action. Metformin is highly water soluble, not significantly metabolized, and excreted unchanged by the kidneys. Metformin would not be degraded in a cell-culture system, which may partly explain the prolonged action.
The predominant downstream signal after metformin- induced IR tyrosine phosphorylation was activation of IRS-2, without any effect on IRS-1. An increasing number of reports suggest that IRS-2 may be as important, or more important than, IRS-1 in mediating the effects of IR activation on glucose (52, 53, 54, 55, 56). The IRS-1 knockout mouse is 50% of normal size but has only mildly abnormal glucose. The IRS-2 knockout is 90% of wild-type size but develops diabetes within weeks of birth. It has severe hepatic insulin resistance and mild muscle insulin resistance. This pattern is also seen in the double heterozygous IR ± IRS-2 ± mouse. Therefore, IRS-2 signaling appears to be particularly important for the effects of insulin on hepatic glucose metabolism (55, 56).
It is probable that metformin activates IR in other tissues, including muscle. If the glucose-lowering effect of metformin is mediated via IRS-2, the relatively higher abundance of IRS-2 in liver may help to explain the results of clamp studies, which are predominantly determined by muscle glucose uptake vs. the decrease in EGP (52). The predominant IRS in muscle is IRS-1.
The plasma membrane lawn studies demonstrate that metformin increased translocation of GLUT-1 to the plasma membrane in these cells. Other studies have shown increased translocation of GLUT-1 (30, 57, 58) following metformin treatment, but because this was seen in other cell types, it was important to obtain confirmation in Huh7 cells. GLUT-4 translocation has also been shown to be increased in adipose and muscle models (30, 57), but GLUT-4 was not examined because it is not expressed in liver.
The ability of metformin to increase IR tyrosine phosphorylation was inhibited effectively by the IR tyrosine kinase inhibitor tyrphostin-AG1024. The AG0124 completely abrogated the ability of metformin to increase 2-deoxyglucose uptake, which provided strong, independent confirmation that metformin was acting via the IR to alter glucose metabolism.
The magnitude of the effects of metformin on IRß, IR-KA, and IRS-2 were smaller than those seen with insulin. This suggests that metformin may induce phosphorylation of a more limited number of IR tyrosine residues than insulin. Given the specific effect on IRS-2 but not IRS-1, we hypothesized that these may be in the kinase loop-binding domain of IR, which is necessary for IRS-2 activation (59).
In summary, we have demonstrated that metformin acutely increased IR-KA in whole IR partially purified from normal human liver tissue by two different methods. Physiologically active concentrations of metformin caused a significant, early-onset, prolonged increase in tyrosine phosphorylation of the ß-subunit in primary human hepatocytes and Huh7 cells. Moreover, metformin selectively activated IRS-2, with no effect on IRS-1.
Inhibition of the IR by AG1024 blocked metformin- induced IRß phosphorylation and completely blocked metformin induced 2-deoxyglucose uptake. Our data therefore suggest that the many reported effects of metformin occur as downstream events from activation of IR followed by selective IRS-2 activation.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
Abbreviations: EGP, Endogenous glucose production; GLUT, glucose transporter; IR, insulin receptor; IRS, IR substrates; KA, kinase activity; WGA, wheat germ agglutinin agarose.
Received September 6, 2002.
Accepted December 13, 2002.
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
 |
|
 |
|
  F. Feng, L. Wang, N. Albanese, A. Holmes, and P. Xia Tumor Necrosis Factor-Like Weak Inducer of Apoptosis Attenuates the Action of Insulin in Hepatocytes Endocrinology, April 1, 2008; 149(4): 1505 - 1513. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S.-K. Shields, C. Nicola, and C. Chakraborty Rho Guanosine 5'-Triphosphatases Differentially Regulate Insulin-Like Growth Factor I (IGF-I) Receptor-Dependent and -Independent Actions of IGF-II on Human Trophoblast Migration Endocrinology, October 1, 2007; 148(10): 4906 - 4917. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. J Bailey Treating insulin resistance: future prospects Diabetes and Vascular Disease Research, March 1, 2007; 4(1): 20 - 31. [Abstract] [PDF]
|
|
|
|
 |
|
 S. Srinivasan, G. R. Ambler, L. A. Baur, S. P. Garnett, M. Tepsa, F. Yap, G. M. Ward, and C. T. Cowell Randomized, Controlled Trial of Metformin for Obesity and Insulin Resistance in Children and Adolescents: Improvement in Body Composition and Fasting Insulin J. Clin. Endocrinol. Metab., June 1, 2006; 91(6): 2074 - 2080. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. S. Y. Chan, S. M. Twigg, S. M. Firth, and R. C. Baxter Insulin-Like Growth Factor Binding Protein-3 Leads to Insulin Resistance in Adipocytes J. Clin. Endocrinol. Metab., December 1, 2005; 90(12): 6588 - 6595. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 D. Ertunc, E.C. Tok, A. Aktas, E.M. Erdal, and S. Dilek The importance of IRS-1 Gly972Arg polymorphism in evaluating the response to metformin treatment in polycystic ovary syndrome Hum. Reprod., May 1, 2005; 20(5): 1207 - 1212. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. E. Cleasby, N. Dzamko, B. D. Hegarty, G. J. Cooney, E. W. Kraegen, and J.-M. Ye Metformin Prevents the Development of Acute Lipid-Induced Insulin Resistance in the Rat Through Altered Hepatic Signaling Mechanisms Diabetes, December 1, 2004; 53(12): 3258 - 3266. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. Wilson Hepatitis C Infection and Type 2 Diabetes in American-Indian Women Diabetes Care, September 1, 2004; 27(9): 2116 - 2119. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 K. J. Nadeau, J. W. Leitner, I. Gurerich, and B. Draznin Insulin Regulation of Sterol Regulatory Element-binding Protein-1 Expression in L-6 Muscle Cells and 3T3 L1 Adipocytes J. Biol. Chem., August 13, 2004; 279(33): 34380 - 34387. [Abstract] [Full Text] [PDF]
|
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Endocrinology | Endocrine Reviews | J. Clin. End. & Metab. |
| Molecular Endocrinology | Recent Prog. Horm. Res. | All Endocrine Journals |
, P < 0.0001 vs. metformin treatment alone; 
, P < 0.0001 vs. insulin treatment alone.