Effects of insulin-sensitising agents in mice with hepatic insulin resistance
- First Online:
- Cite this article as:
- Cohen, S.E., Tseng, YH., Michael, M.D. et al. Diabetologia (2004) 47: 407. doi:10.1007/s00125-003-1320-4
The metabolic abnormalities of insulin resistance are ameliorated by insulin sensitisers via different mechanisms. Metformin decreases hepatic glucose output, whereas rosiglitazone (RSG) is an agonist for peroxisome proliferator activated receptor (PPAR)γ, highly expressed in fat. To gain insight into the mechanisms of action of these drugs, we compared their actions in two models of insulin resistance: the obese, hyperglycaemic ob/ob mouse and the liver specific insulin receptor knockout (LIRKO) mouse.
Control, ob/ob, and LIRKO mice were divided into three groups that received metformin (300 mg/kg body weight/day), RSG (3 mg/kg body weight/day), or placebo for 3 weeks.
In the presence of the severe hepatic insulin resistance of the LIRKO mouse, neither metformin nor RSG had any significant effect on glucose or insulin tolerance tests. On the other hand, RSG decreased serum concentrations of total cholesterol, LDL, and HDL in LIRKO mice. Adipocyte PPARγ gene and protein expression, and adipocyte size were all increased in LIRKO mice treated with RSG, whereas fat-cell size in control animals was decreased by RSG.
TZDs probably improve some lipid parameters of the dysmetabolic syndrome associated with diabetes mellitus even in the presence of absolute hepatic insulin resistance, but both metformin and TZDs require an operating insulin signalling system in the liver for their effects in glucose homeostasis.
KeywordsThiazolidinedione rosiglitazone metformin diabetes mellitus
liver specific insulin receptor knockout
The metabolic abnormalities of insulin resistant states are ameliorated by insulin-sensitising agents, such as metformin and the thiazolidinedione (TZD) rosiglitazone (RSG) through different mechanisms . Metformin has been shown to act primarily by decreasing hepatic glucose output, whereas TZD treatment increases peripheral glucose disposal . This effect of TZDs is mediated by binding to the peroxisome proliferator activated receptor (PPAR)γ, which is most abundant in fat ; the mechanism of metformin action remains uncertain, but presumably involves the liver .
To gain insight into the physiologic and molecular mechanism of action of these drugs, we have used a new mouse model of isolated severe hepatic insulin resistance, the liver specific insulin receptor knockout (LIRKO) mouse . The LIRKO model was chosen because it has a liver specific form of insulin resistance such that any effects produced by these compounds must be attributed to a mechanism that does not involve direct insulin signalling in the liver. To establish the effects of insulin-sensitising agents on the physiologic parameters of insulin resistance, we evaluated serum concentrations of insulin, glucose, lipids, and NEFA, and did an insulin tolerance test (ITT) and glucose tolerance test (GTT).
Materials and methods
Animals and treatment
The LIRKO mouse was generated using the Cre/loxP system for site-specific excisional DNA recombination by crossing mice carrying a floxed insulin receptor, IR (lox/lox) with mice carrying the Cre transgene driven by the albumin promoterand heterozygous for the floxed allele IR (lox/+) . Ob/ob mice were on a C57bl/6 background purchased from Jackson Laboratory, Bar Harbor, Maine.
Animals were housed in virus-free facilities on a 12-h light to dark cycle (0700 on, 1900 off) and were fed a standard rodent chow (Mouse Diet 9F, PMI Nutrition International: percent of calories from carbohydrates, 56.5%; fat, 21.6%; and protein, 21.9%) ad libitum. All protocols for animal use and killing were approved by the Animal Care Committee of the Joslin Diabetes Center and were in accordance with NIH guidelines. Genotyping was done by PCR using genomic DNA isolated from the tail tip .
Two-month old male LIRKO (n=48), control (Ctrl) (n=37) and ob/ob (n=9) mice were treated for 3 weeks with either RSG (3 mg/kg body weight) or metformin (300 mg/kg body weight). The drugs were thoroughly mixed with powdered chow and distributed into individual feeding chambers, with the amounts of food weighed to insure equal drug distribution to each mouse. Food left by the mice was weighed biweekly, and no difference was found between either the control or LIRKO mice, or those treated with either drug.
Blood glucose was measured from whole venous blood using an automatic glucose monitor (Elite, Bayer, New Haven, Conn., USA). Insulin concentrations in serum were measured by ELISA using mouse insulin as a standard (Crystal Chem, Downers Grove, Ill., USA). Triglyceride concentrations in serum from fasted animals were measured by colorimetric enzyme assay using the GPO-Trinder Assay (Sigma Chemical, St. Louis, Mo., USA). Non-esterified fatty acid concentrations were analysed in serum from fasted animals using the NEFA-Kit-U (Wako, Richmond, Va., USA). The Beckman CX7 analyser (Pasadena, Calif., USA) was used to measure liver function tests and the lipid profile. Glucose tolerance tests (GTT) were done on mice that had been fasted overnight for 16 h, whereas insulin tolerance tests (ITT) were done on mice in a fed state at 2:00 pm. Animals were injected with either 2 g/kg body weight of glucose or 1.5 U/kg body weight of human regular insulin (Lilly) in the intraperitoneal cavity for the GTT and ITT, respectively. Glucose concentrations were measured from blood collected from the tail immediately before and at 15-, 30-, and 60-min intervals after the injection; and for the ITT, additionally at 120 min.
Immunoprecipitation and western blot analysis
Chemicals were obtained from Sigma Chemical (St. Louis, Mo., USA), unless otherwise noted. Tissues were removed and homogenised using a Polytron in homogenisation buffer (50 mmol/l Tris [pH 7.8], 2% SDS, 10% glycerol, 10 mmol/l sodium pyrophosphate, 0.1 mol/l sodium fluoride, 10 mmol/l EDTA, 6 mol/l urea, 10 mmol/l sodium orthovanadate, 10 µg/ml leupeptin, and 2 mmol/l PMSF). Samples were solubilised for 30 min on ice, and particulate matter was removed by centrifugation at 4°C. Western blot analysis was done on at least duplicate individuals of each genotype, and quantified using scanning densitometry, and ImageQuant version 4.0 software.
The liver and epididymal fat pads were isolated immediately after killing the mice, then dried, and weighed. Tissues were fixed in 10% phosphate buffered formalin (Fisher Scientific, Pittsburgh, Pa., USA) and embedded in paraffin. Staining of the sections with haematoxylin/eosin and periodic acid-Schiff (PAS) reagent was done using standard techniques. Immunohistochemistry was done using cleaved anti-caspase-3 at a 1:500 dilution (Cell Signalling Technology, Beverly, Mass., USA) and the AEC Kit (Dako Envision System, Toronto, Ontario).
Statistical analysis was done with SigmaPlot 2000 for Windows Version 6.00 (SPSS, Chicago, Ill., USA) using a two-tailed unpaired Student’s t test. A probability value of less than 0.05 was considered significant.
To ascertain that the mice were indeed ingesting the insulin-sensitising drugs at an effective dose, we used ob/ob mice which lack a functional leptin gene and are morbidly obese, severely insulin resistant, and hyperglycaemic . Treatment with either metformin or RSG decreased the blood glucose concentrations of the ob/ob mice from 19.1±1.1 mmol/l to 7.8±0.2 mmol/l, a concentration equivalent to that of the control mice (Fig. 1c). Thus, under conditions where ob/ob mice show a major improvement of hyperglycaemia by either metformin or RSG, both drugs fail to affect glucose metabolism in the LIRKO mouse.
Lipid profile with RSG treatment
Effect of RSG treatment
There was no significant difference in body weight between LIRKO and control mice at 2 months of age. Neither the fasting nor fed body weights of LIRKO or control mice changed during 3 weeks of treatment with RSG or metformin. However, RSG decreased the weight of epididymal fat pads in the control mice 22% (p<0.003), while increasing fat-pad weight in the LIRKOs by 32% (p<0.04). RSG treatment also decreased average fat-cell size in control, but increased fat-cell size in the LIRKO mice, as estimated by quantitative analysis of haematoxylin and eosin-stained fat pads.
Immunohistochemistry staining of fat pads from control mice with the apoptosis marker cleaved caspase-3 showed evidence of increased cell death after RSG treatment. Western blot analysis of fat-pad protein from both control and LIRKO mice showed that death receptor Fas and its ligand FasL were increased by RSG treatment. The precursor form of caspase-3 was also strongly increased by RSG in the control mice and increased to a lesser extent in the LIRKO mice. Similar studies of the liver failed to show any changes. These data suggest that RSG treatment can induce increased rates of apoptosis in fat but not liver cells, and that this can contribute to the ability of the drug to change cell populations in a relatively short period of time.
We have shown that lipid abnormalities, but not impaired glucose homeostasis associated with hepatic insulin resistance, are improved by RSG. Additionally, the high serum glucose and insulin concentrations in both the fasted and fed states in LIRKO mice are refractory to treatment with either metformin or RSG. This dichotomy between the drug effects on lipid metabolism and glucose metabolism indicates that intact insulin receptor function in liver is critical to the effects of RSG and metformin on glucose homeostasis, but not necessarily for the mechanism of action of RSG’s effect on lipid metabolism.
The fact that the effect of RSG on the adipocyte is intact in the LIRKO mouse indicates that the normal cross talk between the adipocyte and liver in the insulin sensitising effect of RSG is interrupted by knockout of IR in liver. Further proof that the IR is essential to the insulin sensitising effect of RSG comes from a recently published article describing two patients with mutations in the IR gene in the tyrosine kinase domain . In these patients with extreme insulin resistance, RSG produced no change in oral or intravenous GTT. Unlike our study, however, in these patients there was also no change in TG, or HDL or LDL cholesterol suggesting that if the IR is intact in organs other than the liver, as in the LIRKO model, this is sufficient for RSG to have an effect on lipid metabolism. The presence of some adipose tissue also seems to be necessary for the antidiabetic, but not the hypolipidemic, effects of RSG, as shown by the lack of effects of these drugs in A-ZIP/F-1 mice that lack adipose tissue . Overall, these studies indicate that both intact adipose tissue and liver are necessary for the antidiabetic action of RSG, however only an intact liver is necessary for the lipid lowering effects of RSG.
The effect of RSG on lipid metabolism was not associated with a change in body weight. Epididymal fat-pad weight, on the other hand, was decreased by treatment with RSG in the control mice, while it was increased in the LIRKO mice. This was due primarily to a change in adipocyte cell size that followed the same pattern as the epididymal fat-pad weight with RSG treatment in both the control and LIRKO mice.
One mechanism contributing to the decrease in the epididymal fat pad weight and adipocyte cell size by RSG treatment in the control mice could be increased apoptosis, as suggested by immunohistochemistry staining with cleaved caspase-3 and western blot analysis with the precursor form of caspase-3, Fas, and Fas ligand. These three apoptosis markers were also increased by RSG treatment in epididymal fat from LIRKO mice, although to a lesser extent; however, in the RSG treated LIRKO mice this was associated with an increase in epididymal fat-pad weight and adipocyte cell size. These different responses to RSG could reflect the differences in circulating insulin concentrations, which are very high in LIRKO mice and normal in control mice. Thus, this effect of RSG to increase adipocyte weight and size, despite a mild increase in apoptosis would only be expected to occur in models that have high insulin concentrations; in other situations one would expect that no increase in apoptosis would dominate, causing a decrease in fat weight and size. Additionally, insulin has been shown to be anti-apoptotic , possibly limiting the extent to which RSG could cause a predominance of small adipocytes and, hence, an insulin sensitised state.
Although difficult to assess, another possible mechanism that might explain the lack of RSG’s effect as an antidiabetic agent in the LIRKO mice is the extremely increased serum insulin concentrations in this model. The lack of the hepatic insulin receptor, with its resultant compensatory beta-cell hyperplasia and defect in insulin clearance results in markedly increased serum insulin concentrations in the LIRKO mouse, which could contribute to desensitisation of insulin action in peripheral tissues.
In conclusion, TZDs could improve some lipid, but not glycaemic, parameters of the dysmetabolic syndrome associated with diabetes mellitus even in the presence of absolute insulin resistance in the liver, whereas for metformin to have any effect, an operating insulin-signalling system in liver is mandatory.