INTRODUCTION

Rosiglitazone is an insulin sensitizer for the treatment of type 2 diabetes. In addition to improving insulin sensitivity, rosiglitazone influences the pathological conditions of nonalcoholic fatty liver disease (NAFLD). It has been shown that rosiglitazone decreases the hepatic triglyceride content in Otsuka Long-Evans Tokushima Fatty rats (1). Beneficial effects have also been reported on many other animal models (2,3). In humans, recent clinical studies reveal that rosiglitazone improves NAFLD (4,5), thus supporting the use of rosiglitazone in treating this disease. However, it has also been reported that rosiglitazone can actually exacerbate NAFLD (6–8). For instance, treatment of KKAy diabetic mice with rosiglitazone resulted in severe liver steatosis (6). Similar deleterious effects have also been reported on ob/ob mice with or without deficiency of the low-density lipoprotein receptor gene (7,8).

Nuclear receptor peroxisome proliferator-activated receptor gamma (PPAR-γ) plays an important role in the rosiglitazone-induced increase of insulin sensitivity (9–11). Its activity also affects adipogenesis and hepatic fat homeostasis (9). It has been previously shown that overexpression of PPAR-γ via an adenoviral vector leads to ectopic fat deposition in the liver, while liver-specific disruption of the PPAR-γ pathway improves hepatic steatosis (12,13). Results from previous studies have linked aberrant PPAR-γ expression and/or PPAR-γ polymorphism to NAFLD development, underscoring the importance of PPAR-γ in the pathogenesis of human hepatic steatosis (14–16). These previous observations prompt us to hypothesize that the apparent contradictory effects of rosiglitazone on improving or exacerbating nonalcoholic fatty liver disease are related to the hepatic level of PPAR-γ expression. Our hypothesis posits that rosiglitazone brings benefits to NAFLD patients with a low level of PPAR-γ in the liver and exacerbates the disease when patients have an elevated level of PPAR-γ.

To test this hypothesis, we employed the technique of gene transfer to upregulate PPAR-γ gene expression in mice fed a high-fat diet (HFD) and studied the consequences of rosiglitazone treatment. We demonstrate that rosiglitazone aggravates liver steatosis when expression of Ppar-γ is upregulated in the liver. Conversely, rosiglitazone treatment prevents mice from developing HFD-induced fatty liver.

METHODS

Animals and Materials

Male C57BL/6 mice (∼22 g) were purchased from Charles River (Wilmington, MA). The high-fat diet (60% kJ/fat, #F3282) was purchased from Bio-Serv (Frenchtown, NJ). Rosiglitazone was purchased from Cayman Chemical (Ann Arbor, MI). The mouse genes of Ppar-γ, Cd36, Fabp4, and Mgat1 were cloned from complementary DNA sequences of C57BL/6 mice and inserted into pLIVE plasmid vectors (Mirus Bio, Madison, WI). Primers for PCR analyses were made at Sigma-Aldrich (St. Louis, MO).

Animal Treatments

Animal treatments were performed following the approved protocol (A2014 07-008-Y1-A0) by the IACUC of the University of Georgia. The mice were housed under a standard condition with a 12-h light-dark cycle and a regular chow or high-fat diet. Intraperitoneal injection of rosiglitazone was performed on non-obese mice (∼22 g) at a dose of 20 mg/kg twice weekly for 9 weeks, and a carrier solution was used as the control. A 3-week preventive study was performed by using a low dose of 5 mg/kg daily. Obese mice were obtained by HFD feeding for 16 weeks until their body weight reached ∼50 g. These obese mice were treated with a lower dose of 5 mg/kg rather than 20 mg/kg, because the latter showed a significant toxicity leading to marked weight loss in obese mice (data not shown). The rosiglitazone treatment of the obese mice was performed (i.p.) daily for a total of 15 days. Gene transfer to animals was accomplished by employing the hydrodynamic tail vein injection of 10 μg plasmid DNA per mouse according to the procedure previously established (17). Control mice in this study were transferred with the backbone vector using the same procedure.

Hematoxylin and Eosin Staining and Immunofluorescence

Liver samples were freshly collected and fixed using neutrally buffered formalin for 24 h. The fixed tissue samples were then embedded into paraffin and cut 6 μm in thickness. Staining was performed using a commercial kit from BBC Biochemical (Atlanta, GA) following the protocol provided by the manufacturer. Tissue sections were examined and photographed using the NIS-Elements imaging platform from the Nikon Instruments Inc. (Melville, NY). A PPAR-γ immunofluorescence study was performed by following the protocol provided by the antibody supplier of Cell Signaling Technology (Danvers, MA).

Determination of Hepatic Triglyceride Level

Red Oil O staining was used to assess the level and distribution of fat droplets in the liver using frozen sections of freshly collected liver samples. Frozen liver sections (8 μm) were stained with 0.2% Oil Red O in a 60% solution of isopropanol for 20 min and then washed three times with phosphate-buffered saline. A microscopic examination was performed and photographs were taken under a regular light microscope.

Quantitative determination of liver triglycerides was performed as previously described (18). Briefly, frozen liver samples (200–400 mg per sample) were homogenized in a mixed solution of chloroform and methanol (2:1) and incubated overnight at 4°C. The tissue homogenates were centrifuged at 12,000 rpm for 20 min, and the supernatants were dried and re-dissolved in a 5% solution of Triton-X100. The triglyceride concentration was determined using a commercial kit (Thermo-Scientific, Pittsburgh, PA).

Gene Expression Analyses

Total RNA was isolated using a TRIZOL reagent from Invitrogen (Grand Island, NY). Quantitative analyses of messenger RNA (mRNA) were performed using SYBR Green as a detection reagent on the ABI StepOnePlus Real-Time PCR system. Data analyses were conducted using the ΔΔCt method. GAPDH mRNA was utilized as an internal reference. Primers were synthesized at Sigma (St. Louis, MO), and their sequences are listed in Table I. Melting curve analysis of real-time PCR products was conducted and showed a single DNA duplex.

Table I PCR Primer Sequences
Full size table

Statistical Analysis

All results were expressed as the mean ± SD. Statistical analyses were performed using Student’s t test and ANOVA. A p value below 0.05 (p < 0.05) was considered significantly different.

RESULTS

Impact of Rosiglitazone Treatment on Mice with Different Levels of Hepatic PPAR-γ

HFD-induced hepatic fat accumulation was employed to examine the PPAR-γ-dependent effects of rosiglitazone treatment on fat accumulation in the liver. Results in Fig. 1a show relative mRNA levels of peroxisome proliferator-activated receptors (Ppar-α, Ppar-γ1, Ppar-γ2, and Ppar-δ). The hepatic PPAR-γ1 level in obese mice is ∼2-fold higher than that of normal mice. The effects of rosiglitazone treatment on the normal and obese mice were examined by means of intraperitoneal injection of rosiglitazone into mice fed a HFD. For normal mice with low or normal levels of PPAR-γ in the liver, 9 weeks of rosiglitazone treatment resulted in an average liver weight of ∼1.4 g, ∼27.1% lower than that of the controls injected with a carrier solution (Fig. 1(B1 and B2)). Structural analysis with hematoxylin and eosin (H&E) and Oil Red O staining revealed the presence of a significant amount of fat content in the liver of the control animals but not in the liver of rosiglitazone-treated mice (Fig. 1(B3)). A significantly lower level of hepatic lipids in rosiglitazone-treated animals was confirmed by biochemical determination (Fig. 1(B4)). A 3-week preventive study with a daily injection of a lower dose of rosiglitazone further confirmed the beneficial effect of this drug on fatty liver (Supplementary Figure 1). These results suggest that rosiglitazone treatment alleviates HFD-induced lipid accumulation in the liver.

Fig. 1
figure 1

Opposing effects of rosiglitazone on HFD-induced fatty liver. A Expression levels of PPARs in the liver of normal and obese mice; B1 representative images of liver at the end of the prevention experiment (bar length = 1 cm); B2 weight of liver; B3 representative images of H&E staining and Oil red O staining of liver sections; B4 liver triglyceride level; C1 representative images of liver at the end of the 15-day rosiglitazone treatment of obese mice (bar length = 1 cm); C2 weight of liver; C3 representative images of H&E staining and Oil red O staining of liver sections; and C4 liver triglyceride level. Values in A, B2, B4, C2, and C4 represent average ± SD (n = 5). One asterisk p < 0.05 and two asterisks p < 0.01 compared with control mice

Full size image

In obese mice, those treated with rosiglitazone had a liver weight of ∼2.9 g compared to ∼2.1 g of the control animals injected with a carrier solution. An increase in the vacuole structure of H&E staining of the liver (Fig. 1(C1 and C2)) and the density of Oil Red O-stained lipid droplets was seen in rosiglitazone-treated obese mice (Fig. 1(C3)). Results in Fig. 1(C4) show that rosiglitazone treatment resulted in a ∼140.4% increase of triglycerides in the liver. These results demonstrate the opposing effects of rosiglitazone treatment. In normal mice with a lower level of PPAR-γ expression, rosiglitazone blocked HFD-induced fatty liver. Conversely, rosiglitazone treatment exacerbated hepatic steatosis in obese mice with a higher level of PPAR-γ in the liver.

Hepatic PPAR-γ Regulates the Effect of Rosiglitazone on the Exacerbation of Ectopic Fat Deposition in the Liver

While results in Fig. 1 provide evidence in support of a correlation between the PPAR-γ level in the liver and the effect of rosiglitazone treatment on hepatic fat accumulation, we extended our effort to seek direct evidence in support of our conclusion that fat accumulation is regulated by the PPAR-γ level in the liver. To this end, we cloned the Ppar-γ1 gene into a plasmid vector. Its expression was driven by the hepatocyte-specific albumin promoter. We then transferred this plasmid into non-obese mice using hydrodynamics-based gene delivery (17). Ppar-γ gene transfer increased the mRNA level of PPAR-γ exclusively in the liver (Fig. 2a), which was verified by regular PCR (Fig. 2b). Overexpression of Ppar-γ in hepatocytes resulted in rapid lipid accumulation in the mouse livers, even those of mice on a standard chow. PPAR-γ and lipid droplets co-localize in hepatocytes after Ppar-γ gene transfer (Supplementary Figure 2). Importantly, rosiglitazone induced significant numbers of lipid droplets only in the liver cells with PPAR-γ overexpression (Fig. 2c). This conclusion was further confirmed by biochemical determinations (Fig. 2d). Taken together, these results indicate that hepatic PPAR-γ plays an important role in regulating the effect of rosiglitazone in altering pathological conditions of liver steatosis.

Fig. 2
figure 2

Hepatic PPAR-γ controls the effect of rosiglitazone on hepatic fat deposition. a Expression of Ppar-γ in major organs 3 days after gene transfer; b verification of the overexpression of Ppar-γ in the liver; c representative images of Oil red O staining and Nile red staining of liver sections; and d liver triglyceride level. Values in a and d represent average ± SD (n = 5). Mice were euthanized 3 days after Ppar-γ gene transfer. Asterisks p < 0.01 compared to control mice without rosiglitazone treatment, number signs p < 0.01 compared with Ppar-γ-transferred mice without rosiglitazone treatment

Full size image

Impact of Ppar-γ Overexpression on the Expression of Genes Involved in Lipid Metabolism

We next explored the underlying mechanisms for PPAR-γ-dependent rosiglitazone-induced hepatic exacerbation as shown in Fig. 2. Real-time PCR was employed to determine the relative mRNA level of the genes known for their involvement in glucose and lipid metabolism. Results in Fig. 3 show that overexpression of Ppar-γ did not affect mRNA levels of nuclear receptor genes known for transcription regulation (Fig. 3a). However, expression levels of the key genes for de novo lipogenesis were increased with Ppar-γ overexpression, including Fas (∼1.7-fold) and Scd-1 (∼2.3-fold) (Fig. 3b). No significant change was seen in the mRNA levels of crucial genes for lipid oxidation (Fig. 3c). The expression of key genes responsible for energy expenditure in the liver was not affected, with the exception of Ucp2 which was increased ∼1.6-fold (Fig. 3d). Overexpression of Ppar-γ led to an elevated expression of Hsl (∼5.3-fold), Atgl (∼2.6-fold), Cd36 (∼3.2-fold), and Fabp4 (∼11.5-fold) (Fig. 3e). Intriguingly, transfer of the Ppar-γ plasmids increased the expression of several pivotal genes for lipid droplet development, including Mgat1 (∼6.1-fold) and Dgat1 (∼1.7-fold) (Fig. 3f). Collectively, this set of data shows that the overexpression of Ppar-γ elevates the expression of a set of pivotal genes involved in hepatic lipid metabolism, particularly Cd36, Fabp4, and Mgat1.

Fig. 3
figure 3

Overexpression of hepatic PPAR-γ upregulated the expression of a variety of pivotal genes involved in lipid metabolism in the liver. a Expression of key nuclear receptor genes involved in liver lipid metabolism; b expression of key genes involved in de novo lipogenesis in the liver; c expression of key genes for fatty acid β oxidation; d expression of crucial genes for energy expenditure; e expression of key genes for fatty acid transport; and f expression of pivotal genes for lipid droplet development. Samples were collected from animals 3 days after Ppar-γ gene transfer. Values in a–f represent average ± SD (n = 5). One asterisk p < 0.05 and two asterisks p < 0.01 compared with control mice

Full size image

Concurrent and time-dependent expressions of Ppar-γ after hydrodynamic gene transfer and its pivotal target genes Cd36, Fabp4, and Mgat1 were monitored. Results in Fig. 4 show the mRNA level of the selected genes as a function of time after the Ppar-γ gene transfer. The mRNA level of Ppar-γ increased by ∼9.2-fold 3 days after gene transfer and maintained at a relatively high level for over 1 month (Fig. 4a). Similar patterns were observed in the three pivotal genes regulated by PPAR-γ, including Cd36, Fabp4, and Mgat1. These peaked at ∼8.1-fold, ∼40.8-fold, and ∼12.1-fold, respectively (Fig. 4b–d). Increased expressions of Ppar-γ, Cd36, Fabp4, and Mgat1 were accompanied by a significantly high level of lipids in the livers of animals with Ppar-γ gene transfer (Fig. 4e).

Fig. 4
figure 4

The hepatic fat accumulation induced by Ppar-γ overexpression was associated with elevated expression of Cd36, Fabp4, and Mgat1 in the liver. a–d Time-dependent mRNA levels of Ppar-γ, Cd36, Fabp4, and Mgat1, respectively, and e representative images of Oil red O staining of liver sections. Values in a–d represent average ± SD (n = 5)

Full size image

Concurrent Overexpression of Cd36, Fabp4, and Mgat1 Replicates the PPAR-γ-Mediated Hepatic Fat Deposition

We next investigated whether the three pivotal genes that are targets of PPAR-γ, Cd36, Fabp4, and Mgat1 are responsible for hepatic fat aggregation induced by Ppar-γ overexpression. These three genes were cloned separately into a pLIVE plasmid vector, and the plasmids were transferred into the mouse liver together with a ratio of 1:1:1 using the same hydrodynamics-based procedure. Concurrent gene transfer greatly increased the mRNA levels of each of these three genes in the liver (Fig. 5a), which consequently led to hepatic fat aggregation in mice fed either a standard chow or a HFD (Fig. 5b). This was further verified by biochemical quantifications (Fig. 5c). Taken together, these results suggest that upregulated expression of Cd36, Fabp4, and Mgat1 is responsible for the hepatic fat deposition induced by Ppar-γ overexpression.

Fig. 5
figure 5

Concurrent transfer of Cd36, Fabp4, and Mgat1 genes recapitulated the phenotypic change induced by overexpression of Ppar-γ in the liver. a Verification of the concurrent overexpression of Cd36, Fabp4, and Mgat1 in the liver; b representative images of Oil red O staining of the livers from mice transferred with the three pivotal genes and kept on regular chow or HFD; and c hepatic triglyceride level. Mice were euthanized 1 week after hydrodynamic gene transfer. Values in c represent average ± SD (n = 5). Two asterisks p < 0.01 compared with chow-fed control mice, one number sign p < 0.05 compared with Ppar-γ-transferred mice on chow

Full size image

DISCUSSION

Hepatic PPAR-γ plays an important role in maintaining lipid homeostasis in the liver. PPAR-γ was originally described as a transcriptional factor controlling adipocyte differentiation and adipogenesis (9,19). Fat-specific deletion of PPAR-γ causes lipoatrophy and severe metabolic disturbance (20). In addition to controlling adipogenesis, PPAR-γ is also involved in lipid metabolism in the liver (9,21). Convincing evidence for a critical role of PPAR-γ in hepatic lipid metabolism has been presented by taking advantage of animals with liver-specific knockout of PPAR-γ under various genetic backgrounds (13,22–24). Evidence collected from these animal models indicates that hepatic PPAR-γ is essential for ectopic fat deposition in the liver. In congruence with the phenotype of these knockout models, overexpression of PPAR-γ via adenoviral vectors leads to hepatic steatosis (12,25,26), although it remains unclear whether this overexpression is hepatocyte-specific. Extending these previous findings, the present study provides evidence supporting the conclusion that selective overexpression of PPAR-γ in hepatocytes is sufficient to induce rapid development of hepatic steatosis.

We demonstrate in the present study that rosiglitazone on HFD-induced NAFLD has an opposing impact for normal vs. obese mice (Fig. 1). While rosiglitazone treatment blocked NAFLD development in normal mice, this treatment in obese mice led to a deleterious outcome (Fig. 1C). Mechanistically, this opposing effect is related to Ppar-γ gene expression in the liver (Fig. 1a). Liver-specific overexpression of the Ppar-γ gene in mice results in hepatic fat accumulation even with regular chow (Fig. 2). At the molecular level, increased expression of hepatic PPAR-γ promoted transcription of a number of pivotal lipogenic genes particularly Cd36, Fabp4, and Mgat1 (Figs. 3 and 4). Taken together, these findings suggest that hepatic PPAR-γ plays a critical role in regulating the effects of rosiglitazone on hepatic fat accumulation.

The preventive effect of rosiglitazone on NAFLD in HFD-fed normal mice can be attributed to systemic activation of PPAR-γ, which influences various pathways in multiple organs collectively in favor of fat deposition in the adipose tissue rather than the liver. Systemic activation of PPAR-γ affects the entire energy metabolism system since this nuclear receptor has diverse functions in different tissues (9). Specifically, activation of PPAR-γ in the central nervous system resulted in accelerated body weight gain (27,28); conversely, transduction of the same signal in peripheral subcutaneous fat stimulated thermogenesis and increased energy expenditure (29–33). Accordingly, the impact of the systemic administration of rosiglitazone on metabolic homeostasis is complex. In the current study, treatment with rosiglitazone blocked hepatic fat accumulation in HFD-fed mice without a significant impact on body weight gain (data not shown). In line with these findings, a recent study by Evans and colleagues showed that activation of the same pathway greatly alleviated HFD-induced hepatic steatosis in transgenic mice with adipocyte-specific overexpression of PPAR-γ (34).

The exacerbating effect of rosiglitazone in obese mice can be explained by the elevated expression of hepatic PPAR-γ, which may lead to preferential activation of PPAR-γ in the liver. It is seemingly contradictory that one drug produces distinct and opposite effects. In fact, the variable effects of rosiglitazone on NAFLD under diverse conditions have been increasingly noted, and growing evidence suggests the detrimental effect of rosiglitazone on NAFLD in diverse animal models (6–8). In agreement with these studies, our findings demonstrated that rosiglitazone markedly aggravated hepatic steatosis in HFD-induced obese mice. This was related to an elevated expression of hepatic PPAR-γ. More importantly, this liver-specific upregulation of PPAR-γ was indispensable in the rosiglitazone-mediated hepatic fat deposition (Figs. 1(C), 2, and 3). These findings point to the important role of elevated hepatic PPAR-γ in mediating the aggravation of hepatic steatosis by rosiglitazone. In support of this explanation, a previous study by Gonzalez and colleagues elegantly demonstrated that liver-specific deletion of PPAR-γ in ob/ob mice completely blocked rosiglitazone-induced exacerbation of hepatic steatosis (13).

The possible mechanisms underlying rosiglitazone-induced hepatic fat aggregation are depicted in Fig. 6. After administration, rosiglitazone binds to PPAR-γ and upregulates its downstream targets. Similar to those of many other nuclear receptors, the downstream signaling pathways of PPAR-γ in regulating hepatic lipid metabolism are complex and may involve multiple interrelated factors (9,35–37). In the present study, we focused on three key downstream factors and demonstrated that concurrent transfer of these three pivotal genes into mouse liver was sufficient to replicate the phenotypic alterations induced by PPAR-γ overexpression (Fig. 5), indicating that these factors contributed substantially to the dominant role of hepatic PPAR-γ in determining the effect of rosiglitazone on NAFLD. It can be speculated that increased CD36 on cell membranes facilitates intracellular transport of fatty acids, thereafter binding to FABP4, which serves as a fatty acid chaperone in cytoplasm. Subsequently, polar fatty acids are incorporated into neutral triglycerides through a group of enzymes, MGAT1 being the rate-limiting enzyme. Evidently, many other factors may also be involved in these intricate pathways (38–40), and further investigations are warranted to reveal the precise mechanisms by which overexpression of PPAR-γ rapidly increases fat accumulation in the liver.

Fig. 6
figure 6

Schematic diagram illustrating the possible mechanism underlying hepatic fat accumulation induced by rosiglitazone treatment and Ppar-γ overexpression. Rosiglitazone binds to and activates PPAR-γ. Subsequently, PPAR-γ upregulates Cd36, Fabp4, and Mgat1, facilitating fatty acid incorporation into triglycerides and consequently giving rise to hepatic fat aggregation as lipid droplets

Full size image

In conclusion, the results presented in the current study show that rosiglitazone generates distinct effects on the prevention and treatment of HFD-induced NAFLD, in which hepatic PPAR-γ plays a vital role by coordinating and regulating downstream Cd36, Fabp4, and Mgat1 in the liver. These findings suggest that caution should be taken when considering a direct application of rosiglitazone to obese patients having elevated PPAR-γ expression in the liver, and indicate that a more desirable strategy may be to activate PPAR-γ in a tissue- or cell-specific manner in order to avoid rosiglitazone-mediated exacerbation of hepatic steatosis.