Nonalcoholic Hepatic Steatosis in Zucker Diabetic Rats: Spontaneous Evolution and Effects of Metformin and Fenofibrate




No specific treatment for nonalcoholic hepatic fatty liver disease has been defined. We followed the spontaneous evolution of liver steatosis and tested the therapeutic usefulness of metformin and fenofibrate in a model of steatosis, the Zucker diabetic fatty (ZDF) rat. ZDF and control rats were studied at 7, 14, and 21 weeks. After initial study at 7 weeks, ZDF rats received no treatment, metformin or fenofibrate until studies at 14 or 21 weeks. ZDF rats were obese, hypertriglyceridemic, insulin resistant at 7 weeks, type 2 diabetic at 14, diabetic with insulin deficiency at 21. They had steatosis at 7 weeks with increased hepatic expression and activity of lipogenesis. Steatosis was unchanged at 14 and 21 weeks despite lower expression and activity of lipogenesis. Metformin and fenofibrate did not modify energy intake or expenditure or the evolution of diabetes. Both compounds decreased plasma triacylglycerol (TAG) concentrations. Hepatic TAG content was reduced by fenofibrate at 14 and 21 weeks but only at 21 weeks by metformin. Metformin had no significant effects on the expression in liver of genes of fatty acids metabolism. The beneficial effect of fenofibrate occurred despite increased expression of genes involved in the uptake and activation of fatty acids. Acyl-CoA oxidase (ACO) and carnitine palmitoyltransferase I (CPTI) mRNA levels were increased by fenofibrate showing evidence of increased lipid oxidation. To conclude, metformin had only moderate effects on liver steatosis. The effects of fenofibrate was more marked but remained mild.


Excessive accumulation of fat in the liver in the absence of alcohol consumption (nonalcoholic fatty liver disease or hepatic steatosis) is observed in 25–30% of adults in industrialized countries (1). It is often associated with obesity and insulin resistance or type 2 diabetes (2) and the increasing prevalence of obesity has probably a role in its high prevalence. Fatty liver disease can remain limited to steatosis or progress toward nonalcoholic steatohepatitis with the apparition of inflammation in liver. Thereafter this steatohepatitis can lead to cirrhosis and liver cancer (3). Fatty liver can thus contribute significantly to obesity-related morbidity and mortality. At the present time, treatment of hepatic steatosis relies mainly on exercise and dietary advice to reduce body weight and fat accumulation. No specific pharmacological treatment for liver steatosis has been defined. Thiazolidinediones have shown some benefits but are associated with weight gain and unfavorable side effects (4,5). Peroxisome proliferator–activated receptor-α (PPAR-α) agonists could be useful (6) because they promote lipid oxidation and reduce fat accumulation in tissues (7,8). Metformin, an activator of AMP-activated protein kinase (9), showed promising results in ob/ob mice (10). However, clinical studies showed only limited benefit (11,12,13,14). To evaluate their potential usefulness in the treatment of hepatic steatosis, we investigated the effects of long-term in vivo administration of fenofibrate, a PPAR-α agonist, and of metformin on the evolution of liver steatosis in an experimental model, the Zucker diabetic fatty (ZDF) rat.

Methods and Procedures

Male ZDF (fa/fa) rats and their normal littermates (controls C, +/+) (Charles River Laboratories, L'Arbresle, France) were housed (individual cages) at arrival (6 weeks old) in an animal facility with controlled temperature (22 ± 1 °C) and lightning (light on at 07:00 am and off at 07:00 pm). Throughout the study they had free access to water and food. All received the diet (Purina 5008, protein 26.8%, carbohydrate 56.4% (91% starch, 9% simples carbohydrates), fat 16.7% of caloric value; IPS, London, UK) recommended for the development of diabetes in male ZDF rats. Water and food intake and body weight were recorded five times/week. A first metabolic investigation was performed in all rats after 1-week acclimation (age of 7 weeks). Thereafter, five rats of the control group and five of the ZDF group were killed for blood collection and tissue sampling. The remaining control rats were divided into two groups (five rats each); one was killed at the age of 14 weeks after a second metabolic investigation, the other had a metabolic investigation at the age of 14 and 21 weeks before killing at 21 weeks. ZDF rats were divided in three groups of 10 rats. One group received no pharmacological treatment (ZDF group); the other groups were also given fenofibrate (ZDF + F group, 100 mg/kg/day) or metformin (ZDF + M group, 300 mg/kg/day) mixed with diet. These doses correspond to those used in previous studies of the effects of these compounds in rats (15). Fenofibrate or metformin administration started only once after the first metabolic investigation was completed and was continued until killing. Five rats of each group were killed at 14 weeks after a second metabolic investigation and the remaining five rats were investigated at 14 and 21 weeks before killing at 21 weeks. Rats were studied at the age of 7, 14, and 21 weeks because these ages correspond to three steps in the development of metabolic abnormalities in ZDF rats: obesity with insulin resistance (7 weeks), overt type 2 diabetes (14 weeks), and late stage diabetes with insulin deficiency (21 weeks) (16).

Metabolic investigations

Each metabolic investigation comprised blood sampling (fed state, tail vein) for measurement of blood glucose (OneTouch Ultra; Life Technology, Issy-Les-Moulineaux, France), plasma insulin, nonesterified fatty acid (NEFA), triacylglycerols (TAGs) and total cholesterol concentrations, and a 24-h recording of respiratory gaseous exchanges (indirect calorimetry). For this recording, rats were placed at 04:00 pm in individual cages placed in a room with controlled temperature (22 ± 1 °C) and lighting (light on at 07:00 am and off at 07:00 pm). These cages allowed free access to water and food and were ventilated with room air (0.80 l/min). After an overnight habituation period, measurements of O2 consumption and CO2 production were started at 09:00 am and continued during 24 h (Oxymat 800; Bioseb, Chaville, France). Respiratory quotient (RQ) is the ratio of the volume of CO2 produced to the volume of O2 consumed. As urinary nitrogen excretion was not measured, we did not calculate rates of protein, lipid and carbohydrate oxidation. However RQ is an indicator of the relative amounts of lipid and carbohydrate oxidized. Total energy expenditure (TEE, kcal/day/kg0.75) was calculated as TEE = 1.44 VO2 (3.815 + 1.23 RQ). An insulin tolerance test was performed during the initial metabolic investigation (7 weeks) in the postabsorptive state. Food was removed at 07:00 am and insulin (1 unit/kg) was injected intraperitoneally at 01:00 pm. Glucose was measured before and 15, 30, 45, 60, 90, and 120 min after the insulin injection.

Five rats of each group were killed for tissue sampling at the age of 7, 14, and 21 weeks. Food was removed at 08:00 am and rats anesthetized at 02:00 pm (pentobarbital intraperitoneally 60 mg/kg), in the postabsorptive state. Blood (inferior vena cava) was collected and plasma stored at −20 °C until analysis. Liver was removed, flushed with cold isotonic saline and one part was snap frozen in liquid nitrogen before storage at −80 °C until analysis.

Analytical procedures

Plasma NEFA, TAG, and cholesterol were measured by enzymatic methods (17), insulin by enzyme-linked immunosorbent assay (Crystal Chem, Downers Grove, IL). For measurement of liver TAG concentrations, 100 mg of tissue was homogenized in chloroform/methanol (1:2, vol:vol). The chloroform phase was collected, washed with water, and dried under nitrogen. Extracted lipids were dissolved in propanol for enzymatic determination of TAG concentration. Fatty acid synthase (FAS) activity was measured as previously described (18).

Liver total RNAs were purified using TRIZOL (Invitrogen, Cergy-Pontoise, France) with the addition of treatment with DNase. Concentrations and purity were verified by measuring optical density at 230, 260, and 280 nm and integrity by agarose gel electrophoresis. For measurements of individual mRNA levels, total RNA was reverse transcripted using Superscript II (Invitrogen) and random hexamers. Real-time PCR was performed in a MyIQ thermal cycler (Bio-Rad, Marnes La Coquette, France) using iQ SYBR green Supermix (Bio-Rad). All samples were run in duplicate along with dilutions of known amounts of target sequence for quantification of initial cDNA copies. Results are expressed as the target over 18S RNA concentration ratio (ng/µg). Primer sequences are available upon request.


Results are shown as mean ± s.e.m. Intragroups comparisons of values obtained at 7, 14, and 21 weeks for the various groups of rats (control, ZDF, ZDF + M, ZDF + F) were performed for each group by one-way ANOVA followed by the Newman–Keuls procedure to locate the differences, or by two-tailed Student's t-test for unpaired values when data were available only at 14 and 21 weeks (ZDF + M and ZDF + F groups). Between groups comparisons of the values obtained for each metabolic investigation (at 7, 14, or 21 weeks) were performed by one-way ANOVA followed by the Newman–Keuls test. P < 0.05 was considered as indicating a significant difference. Calculations were performed with GraphPad Prism 4.02 (GraphPad, San Diego, CA).


Food intake and body weights

Food intake, higher in ZDF rats than in control rats throughout the study (P < 0.001), was not modified by metformin or fenofibrate (Figure 1). At 7 weeks ZDF rats were heavier than control rats (P < 0.01) without difference between ZDF rats who received or not thereafter fenofibrate or metformin. Control rats gained weight between 7 and 14 weeks (P < 0.01) and 14 and 21 weeks (P < 0.01). ZDF rats also gained weight between 7 and 14 weeks (P < 0.01), but less than control rats and there was no further gain at 21 weeks. As a result the body weight of ZDF rats was comparable to the one of control rats at 14 weeks and lower at 21 weeks (P < 0.05). This evolution of body weight agrees with previous reports (16). The reduced weight gain in ZDF rats is linked to the appearance of diabetes and of massive glycosuria. Fenofibrate-treated rats gained less weight than the untreated ZDF group; their weight was at 14 weeks less than untreated ZDF rats and at 21 weeks less than the ones of the control and ZDF groups (P < 0.05). Metformin had no effect on body weight.

Figure 1.

Evolution of body weight, food intake, total energy expenditure (TEE) and respiratory quotient (RQ) in control (C), untreated Zucker diabetic fatty (ZDF) rats (ZD), and ZDF rats receiving after the first metabolic investigation at 7 weeks metformin (M) or fenofibrate (F). *P < 0.05, **P < 0.01, ***P < 0.001 vs. corresponding control rats, $P < 0.05 vs. corresponding untreated ZDF rats.

Energy expenditure and RQ

At 7 weeks, TEE were decreased in all ZDF groups (P < 0.001 vs. control rats) while average RQ was higher (P < 0.001), with values often above one indicating the presence of net lipogenesis (Figure 1). TEE decreased with age in control and ZDF rats (P < 0.001) and were always lower in ZDF rats (P < 0.05), although the difference with control rats was less marked. RQ of control rats was unchanged at 14 and 21 weeks. It decreased on the contrary in ZDF rats (P < 0.001) to attain values lower than in control rats (P < 0.001). This evolution of RQ was not modified by fenofibrate or metformin. The decrease in TEE was, particularly for ZDF rats receiving fenofibrate, less marked than in ZDF rats receiving no treatment and TEE at 14 and 21 weeks were no longer different from those of the corresponding control groups.

Insulin and metabolites concentrations

Glucose and insulin levels. At 7 weeks, ZDF and control rats in the fed state had comparable glucose concentrations while insulin was higher in all ZDF groups (P < 0.01) (Figures 2) indicating the presence of insulin resistance. Insulin tolerance tests confirmed this insulin resistance (data not shown). Diabetes developed in all ZDF rats around the 10th week as shown by the appearance of polydipsia and polyuria. Glucose levels in all ZDF rats were >25 mmol/l at 14 and 21 weeks whereas plasma insulin decreased to values comparable to those of control rats at 14 weeks and below the detection level at 21 weeks (Figure 2). Neither fenofibrate nor metformin modified these evolutions of glucose and insulin concentrations.

Figure 2.

Evolution of blood glucose and plasma insulin in control rats (open circles) and in the different groups of Zucker diabetic fatty (ZDF) rats (ZDF: inverted triangles, ZDF + F: triangles, ZDF + M: square). **P < 0.01, ***P < 0.001 vs. control rats. ZDF + F and ZDF + M rats received daily administration of fenofibrate or metformin only once the first metabolic investigation at the age of 7 weeks was completed.

Plasma lipid concentrations. At 7 weeks, ZDF rats had, in the fed and post-absorptive states, higher TAG concentrations than control rats whereas NEFA concentrations were comparable (Tables 1 and 2). Plasma cholesterol was increased in ZDF rats only in the fed state. There were no difference in these parameters (fed state) between ZDF rats receiving thereafter fenofibrate or metformin and those receiving no treatment.

Table 1.  Evolution of plasma cholesterol, TAG and NEFA (measured in the fed state) in control rats and in ZDF rats who received fenofibrate (ZDF + F), metformin (ZDF + M) or no treatment (ZDF) after the first metabolic exploration at 7 weeks
inline image
Table 2.  Evolution of plasma cholesterol, TAG and NEFA (measured in the post-absorptive state) in control rats and in ZDF rats who received fenofibrate (ZDF + F), metformin (ZDF + M) or no treatment (ZDF) after the first metabolic exploration at 7 weeks
inline image

Plasma lipid concentrations were unchanged at 14 and 21 weeks in control rats. NEFA concentrations increased in 14- and 21-week-old ZDF rats and were much higher, both at 14 or 21 weeks, and in the fed or postabsorptive state, than in corresponding control rats and in ZDF rats before the appearance of diabetes (7 weeks old). Neither fenofibrate nor metformin modified this evolution of NEFA levels. TAGs were always higher in ZDF rats than in control rats at 14 and 21 weeks. Fenofibrate and metformin decreased TAG levels in the postabsorptive state, with a more marked effect of metformin particularly at 14 weeks. In the fed state, on the contrary, metformin had no effect and fenofibrate had a moderate hypotriglyceridemic effect only at 14 weeks. Cholesterol increased in ZDF rats at 14 and 21 weeks compared to 7 weeks (P < 0.01) and these values were much higher than in the control groups. Fenofibrate and metformin decreased largely plasma cholesterol at 14 and 21 weeks in the fed state (P < 0.01), although these values remained higher than in control groups. These hypocholesterolemic actions were less marked in the postabsorptive state and were significant (P < 0.05) only at 21 weeks for fenofibrate and metformin.

Liver TAG content

Seven-week-old ZDF rats already had steatosis with high liver TAG concentrations (P < 0.001 vs. control rats) (Figure 3). This TAG content increased moderately in control rats at 14 weeks (P < 0.05). It did not increase further in untreated ZDF rats but remained always largely higher than in control rats (P < 0.01 at 14 weeks, P < 0.001 at 21 weeks). Fenofibrate induced as excepted a macroscopic increase in liver size; it decreased liver TAG content at 14 and 21 weeks (P < 0.05 vs. untreated ZDF), with values comparable to those of control rats at 14 weeks but remaining higher at 21 weeks (P < 0.01). On the contrary, metformin had no effect at 14 weeks and only a mild lowering action (P < 0.05) at 21 weeks.

Figure 3.

Liver triacylglycerol (TAG) content in control (C), untreated Zucker diabetic fatty (ZDF) rats (ZD), and ZDF rats who received after the end of the first metabolic investigation (at 7 weeks) daily administration of metformin (M) or fenofibrate (F). **P < 0.01, ***P < 0.001 vs. control, $P < 0.05 vs. untreated ZDF rats.

FAS activity

At 7 weeks, FAS activity was increased in ZDF rats (1,520 ± 150 pmol nicotinamide adenine dinucleotide phosphate oxidized/min/mg of liver vs. 420 ± 55 in control rats, P < 0.01). This activity did not change with age in control rats (535 ± 48 and 340 ± 60 at 14 and 21 weeks) but decreased in ZDF rats (650 ± 75 and 455 ± 65 at 14 and 21 weeks, P < 0.05 and P < 0.01) and was not longer higher than in control rats. Neither fenofibrate (405 ± 50 and 475 ± 85) nor metformin (620 ± 95 and 510 ± 35) modified significantly this activity.

mRNA concentrations

At 7 weeks ZDF rats had an increased expression of lipogenic genes (ACC1, FAS, Srebp-1c, and SCD, P < 0.05 for all, Figure 4). The trend for higher ChREBP mRNA was not significant (P < 0.10) and DGAT1 mRNA level was not increased (data not shown). Among the other mRNAs relevant to fatty acids metabolism measured, only microsomal triglyceride transfer protein (MTP), implicated in the secretion of very low–density lipoprotein (VLDL), FAT, and acyl-CoA synthase 1 (ACS1), implicated in the uptake and activation of fatty acids from plasma lipids, were increased (P < 0.05) in ZDF rats (Figure 5 and 6). The trend for higher acyl-CoA oxidase (ACO) and VLDLr mRNA values was not significant (P < 0.10). mRNA levels of PPAR-α and interleukin-6 were not significantly increased (P < 0.15) and those of tumor necrosis factor-α (TNF-α) were decreased in ZDF rats (Figure 6). To confirm these results at the protein level, immunoblots were performed for TNF-α (antibody SC-1351; Santa Cruz, Santa Cruz, CA) and interleukin-6 (antibody SC 1266; Santa Cruz) in control and ZDF rats. Interleukin-6 protein was barely detectable in all rats (data not shown) and TNF-α protein expression was comparable in the two groups of rats (data not shown). In control rats, increase in age (14- and 21-week-old rats) induced a decrease in ApoB100 (P < 0.01), MTP (P < 0.05) mRNA concentrations and a trend for lower levels ACC1 and ACC2. In ZDF rats aging induced a clear decrease in the expression of the lipogenic genes FAS (P < 0.01) and SCD (P < 0.05) at 14 and 21 weeks and a moderate decrease of Srebp-1c, ACC1, and ACC2 at 21 weeks (Figure 4). There was also an increase in carnitine palmitoyltransferase I (CPTI) at 14 weeks and ACO at 21 weeks. The expression of MTP and ApoB100, involved in the synthesis and secretion of VLDL, was decreased (P < 0.05) at 14 weeks (ApoB100 only) and 21 weeks (both mRNAs) (Figure 5). When ZDF rats are compared to control rats at 14 and 21 weeks, the only remaining difference in the lipogenic pathway is a moderate increase of Srebp-1c and ACC1 mRNA at 14 weeks. CPTI mRNA was higher at 14 weeks and ACO at 21 weeks. Despite the decrease relative to the 7-week-old ZDF rats, MTP values were at 14 and 21 weeks always higher than in control rats (P < 0.01).

Figure 4.

Lipogenic mRNA concentrations in the liver of control (C), untreated Zucker diabetic fatty (ZDF) rats and ZDF rats who received after the end of the first metabolic investigation (at 7 weeks) daily administration of metformin (ZDF + M) or fenofibrate (ZF + F). *P < 0.05 vs. control; $P < 0.05 vs. untreated ZDF rats: £P < 0.05, ££P < 0.01 vs. the values of the same group at 7 weeks.

Figure 5.

Acyl-CoA synthase 1 (ACS1), VLDLr, FAT, peroxisome proliferator–activated receptor-α (PPAR-α), and cytokines mRNA concentrations in the liver of control, untreated Zucker diabetic fatty (ZDF) rats, and ZDF rats who received after the end of the first metabolic investigation (at 7 weeks daily) administration of metformin (ZDF + M) or fenofibrate (ZF + F). *P < 0.05 vs. control; $P < 0.05, $$P < 0.01 vs. untreated ZDF rats. IL-6, interleukin-6; TNF-α, tumor necrosis factor-α; VLDL, very low–density lipoprotein.

Figure 6.

MTP, ApoB100, CPTI, and ACO mRNA concentrations in the liver of control, untreated Zucker diabetic fatty (ZDF) rats and ZDF rats who received after the end of the first metabolic investigation (at 7 weeks) daily administration of metformin (ZDF + M) or fenofibrate (ZF + F). *P < 0.05, **P < 0.01 vs. control; $P < 0.05, $$P < 0.01 vs. untreated ZDF rats; £P < 0.05 vs. the values of the same group at 7 weeks.

Fenofibrate induced a trend for higher mRNA levels of PPAR-α (P < 0.10). It decreased at 14 weeks the expression of ChREBP, Srebp-1c, and ACC2 (P < 0.05) with a trend for lower values of ACC1 and FAS mRNAs (Figure 4) whereas DGAT1 mRNA values were unchanged (data not shown). No modifications of these lipogenic genes were observed at 21 weeks. As expected CPTI and FAT expression were stimulated at 14 and 21 weeks, with a trend for increase in ACO expression. Expression of VLDLr and ACS1, involved in the uptake and activation of fatty acids was also increased. MTP expression was largely increased (P < 0.001) at 14 and 21 weeks (Figure 5). Contrasting with the numerous modifications of gene expression induced by fenofibrate, metformin had no significant effect on the mRNAs measured except for a late (21 weeks) decrease in ACO and an increase in TNF-α.


ZDF rats had as expected insulin resistance at the age of 7 weeks, overt diabetes at 14 weeks, and late-stage diabetes with decreased insulin secretion at 21 weeks. This evolution agrees with previous reports (16,19,20). Liver steatosis was already present at 7 weeks and was unchanged at 14 and 21 weeks. Expressions of TNF-α and interleukin-6 were not increased, suggesting that the ZDF rat model we used is a model of steatosis rather than of steatohepatitis. Several mechanisms can contribute to the accumulation of TAG in liver (21,22,23). First, it can result from increased hepatic TAG synthesis and/or from diversion of TAG metabolism toward storage rather than secretion as VLDL-TAG. Second, the contribution of one or several of the sources of fatty acids used for TAG synthesis can be increased. Fatty acids can be provided by uptake of plasma NEFA, by the de novo synthesis of new fatty acids molecules in liver (de novo lipogenesis, DNL) or by uptake and degradation of circulating lipoproteins. In addition fatty-acyl-CoA released by the breakdown of previously stored TAG can be re-esterified but this does not contribute any new molecule of fatty acid to liver TAG pool. There is evidence that increase in both DNL and uptake of plasma NEFA contribute to liver steatosis in humans (22,24) but the relative importance of these two pathways remains debated (23). Lastly, fatty acids taken up by liver can be oxidized or esterified and a diversion of this metabolic fate toward esterification and away from oxidation, usually associated with an increase in DNL (25,26), may also intervene. The modifications of these various pathways and their contribution to TAG accumulation may depend upon the experimental model used, the whole body metabolic status and the stage of evolution of steatosis itself, making clarification of the mechanisms responsible for steatosis still more difficult.

Actually, our results strongly suggest that these mechanisms are different between insulin-resistant (7-week-old) and diabetic (14- and 21-week-old) ZDF rats. DNL expression was clearly stimulated in 7-week ZDF rats. FAS, ACC1, and SCD expressions were enhanced and FAS activity was increased. The mRNA levels of Srebp-1c, the transcription factor mediating the stimulatory action of insulin on lipogenesis (27) were increased and there was a trend for increased expression of ChREBP, the transcription factor mediating the effects of glucose (28). Lastly the high RQ value of these rats shows a orientation of whole body lipid metabolism toward lipogenesis. With respect to mRNA levels of lipogenic genes in liver, our data agree with previous reports showing in 7–10-week-old ZDF rats increased mRNA concentrations of Srebp-1c (19,20,29,30), ATP citrate-lyase (20), and of FAS and ACC (29). In addition to this enhanced DNL an increased contribution to hepatic TAG synthesis and storage of fatty acids of lipoproteins taken up by liver is also possible since plasma TAG were largely increased. In addition, expressions of VLDLr, FAT, and ACS1 implicated respectively in the uptake of fatty acids of TG-rich lipoproteins (31), in fatty acid transport and activation, were increased. The raised expression of FAT and ACS1 may also have stimulated the uptake of plasma NEFA. However this is unlikely because plasma NEFA concentrations, the main factor controlling liver NEFA uptake (32), were not increased. DGAT1 expression was unchanged suggesting that liver fatty acids re-esterification was not enhanced. We did not measure liver TAG secretion but this secretion was found increased in 10-week ZDF rats (19). Moreover, in the present experiments, ApoB mRNA was not decreased and MTP mRNA was increased. Increased MTP expression stimulates the secretion of VLDL (33). Thus, it seems unlikely that a decrease in liver TAG secretion contributed to hepatic steatosis in 7-week-old ZDF rats.

The presence of diabetes (14- and 21-week ZDF groups) was associated with modifications of lipid metabolism and the mechanisms behind steatosis appear different from those in 7-week rats. At the whole body level, there was a shift in lipid metabolism from net lipogenesis to enhanced lipid oxidation as shown by the decrease in RQ. Moreover lipolysis was increased with high plasma NEFA levels. The fall in insulinemia has a clear role in these modifications. In addition, the expression of hormone-sensitive lipase is increased in the adipose tissue of 12-week-old ZDF rats (20). In liver, the expression of lipogenic genes was decreased compared with the 7-week ZDF group. Previous reports also showed a decrease in Srebp-1c (19,20) and SCD (20) in 12- or 20-week ZDF rats compared to 6- or 10-week rats. It is noteworthy that liver ChREBP expression did not increase in diabetic ZDF rats despite very high glucose concentrations. Such concentrations of glucose stimulated ChREBP expression in vitro in hepatocytes (34) and adipocytes (35). This effect could be absent in vivo. Another possibility could be an inhibitory action of the raised in vivo plasma NEFA levels (35) opposing the stimulatory effect of glucose. FAS activity was also decreased compared to 7-week ZDF rats. In addition to this lower expression of the lipogenic pathway, there is indirect evidence for a parallel increase in liver lipid oxidation since mRNA levels of CPTI and ACO increased. Thus it is probable that DNL had, in 14- and 21-week ZDF rats, a less important role in liver steatosis than in 7-week ZDF rats and that other sources of fatty-acyl-CoA, i.e., plasma NEFA, and TAG-rich lipoproteins played a more important role. This is supported by the rise in NEFA and the further increase in plasma TAG in these rats. With respect to liver TAG secretion, Chirieac et al. found that liver secretion of TAG is not increased in 20-week ZDF rats (19). In the present report ApoB and MTP mRNA decreased with age in ZDF rats. This would be compatible with a TAG secretion rate lower in 14- and 21-week ZDF rats than in 7-week ZDF rats and this lowered secretion could itself participate in the maintain of increased liver TAG content.

Metformin inhibited TAG accumulation in hepatocytes by activating AMP-activated protein kinase (9). This activation resulted in inhibition of ACC activity and in decrease of the insulin or refeeding-stimulated expression of lipogenic genes (9,36). Metformin has been proposed as a treatment of nonalcoholic fatty liver disease (10). However, clinical trials in humans showed only moderate or no benefits (11,12,13,14,37,38). In the present study, metformin decreased postabsorptive plasma levels of TAG, in agreement with previous studies (36), and also had an hypocholesterolemic effect. These modifications of plasma lipid levels show that metformin was effective at the dose used. However, liver TAG content was unchanged (14-week rats) or moderately decreased (21-week rats) by metformin. In addition, FAS activity and the expression of lipogenic genes were not reduced. These results contrast with those showing a reduction by metformin of liver steatosis in ob/ob mice (10), and of the expression of lipogenic genes in rat hepatocytes and in rats in vivo (36). An explanation to these discordant results could be differences in metabolic status. First, it is noteworthy that metformin was found more effective in nondiabetic human subjects with nonalcoholic fatty liver disease (11) than in diabetic ones (38). The results obtained in the model used in the present report (steatotic diabetic rats) are consistent with this observation. First, the ob/ob mice used by Lin et al. were mainly insulin resistant and only moderately hyperglycemic (10). Second, TNF-α expression was increased in ob/ob mice (10), and reduced by metformin, whereas there was no increase in the expression of inflammatory cytokines in ZDF rats. Third, Zhou et al. investigated the action of metformin on hepatic lipogenesis stimulated by insulin, glucose, or refeeding (36). Lin et al. investigated the effects of metformin in ob/ob mice, during the period of constitution of steatosis, and these mice had probably an increased expression of lipogenesis in liver (10). We started metformin administration in 7- week-old rats with already established steatosis and, although these rats had initially large increases of expression and activity of hepatic lipogenesis, we appreciated the effects of metformin only when ZDF rats had shifted to overt diabetes and when hepatic lipogenesis was no longer overexpressed and highly active. This strongly suggests that metformin may have a clear effect on liver steatosis only when an increased DNL has an important role in the development of this steatosis This contribution of DNL in humans could also be more important in steatotic subjects with insulin-resistance alone that in patients with insulin resistance and overt diabetes.

Fenofibrate clearly lowered liver TAG content of ZDF rats. Such effect was also observed in Shionogi mice and in low-density lipoprotein receptor–deficient mice fed a high-fat diet (7,39). This decrease in TAG accumulation occurred despite an increased expression of genes involved in fatty acids uptake (FAT and VLDLr) and activation (ACS1). An increased hepatic lipid oxidation contributed probably to this decreased TAG accumulation because, as expected (39), CPTI and, to a lesser degree, ACO expressions were stimulated. In addition, fenofibrate increased in mice ketone body concentration, a marker of liver fatty acid oxidation (39). The increase in MTP expression (present results and ref. 40) could suggest that TAG secretion by liver was also increased. However, previous studies in humans and rats have shown that PPAR-α agonists decrease on the contrary VLDL–TAG secretion (41,42,43), making thus this possibility unlikely. This decrease in TAG secretion, associated to an enhanced clearance rate (42), explains probably the reduction of plasma TAG by fenofibrate. Srebp-1c, ChREBP, FAS, ACC1, and 2 mRNA were decreased in the fenofibrate-treated group at 14 weeks, when the lowering action on liver TAG content is the more apparent. This agrees with previous report showing also a decrease by fibrates of liver lipogenic genes expression in models of steatosis (39) and suggests that a decreased lipogenic rate participated in the TAG-lowering action of fenofibrate. This is also supported by our previous study showing a reduction by fenofibrate of DNL (17) in hyperlipidemic diabetic patients.

In conclusion, we found in the experimental model used that metformin had only a limited beneficial effect on an already established liver steatosis. Fenofibrate had a more marked beneficial action and could be useful in the treatment of hepatic steatosis associated with obesity, insulin resistance, and diabetes, alone or in association with other recently proposed possible treatments of liver steatosis such as ezetimibe (44), inhibitors of the renin–angiotensin system (45), or antagonists of the cannabinoid receptors type 1 (46).


This work was supported by grants from the Fondation de France (grant 2004002975) and the French Ministère de la Recherche (ACI 2003 2 647).


The authors declared no conflict of interest.