Introduction

Obesity, a major public health problem around the world [1, 2], is associated with several metabolic abnormalities, which include hypertension, hypercholesterolaemia, hypertriacylglycerolaemia and insulin resistance [3, 4], and lead to type 2 diabetes and cardiovascular disease [58]. Serum cholesterol levels are clinically managed through the use of statins [9]. However, a number of dietary modifications have been suggested to reduce serum cholesterol concentrations [10].

Several meta-analyses have shown that the consumption of soy protein (SP) reduces total and LDL-cholesterol [11, 12]. Studies in experimental animals have also demonstrated that SP reduces blood lipids, hepatic cholesterol and triacylglycerol [1315], and also increases insulin sensitivity [16]. The precise mechanism by which SP reduces serum and hepatic cholesterol has not been established, although it has been suggested that these effects occur through an increase in bile acid excretion [17].

The inter-organ cholesterol flux and the synthesis of bile acids from cholesterol are in part regulated by the transcription factor, liver X receptor (LXR)α [18, 19]. LXRs are ligand-activated transcription factors that belong to the nuclear receptor superfamily [20]. The LXR subfamily consists of two isoforms, LXRα and LXRβ, which form obligate heterodimers with the retinoid X receptor and regulate gene expression by binding to LXR response elements (LXREs) in the promoter regions of their target genes [21], some of which are involved in reverse cholesterol transport (RCT) [22]. LXRα is highly abundant in the liver, intestine, kidney, adipose tissue and macrophages, whereas LXRβ is ubiquitously produced [23]. However, LXRα is the dominant isoform in the liver, the activation of which increases biliary cholesterol secretion and limits cholesterol absorption [24]. Recent evidence suggests that LXRα activity can be regulated by phosphorylation [25, 26]. There is evidence that isoflavones are weak ligands for certain other nuclear receptors [2729], but it is not known whether the consumption of SP or its isoflavones (mainly genistein and daidzein) modulates transcriptional control of LXRs or regulates the phosphorylation of these nuclear receptors.

Our aim, therefore, was to use wild-type (WT) and Lxrα −/− (Lxrα also known as Nr1h3) mice and investigate whether metabolic changes that increase bile acid excretion after the consumption of an SP diet are mediated via LXRα, thus leading to the upregulation of genes involved in bile acid synthesis and RCT in the liver and intestine. Our study also sought to determine whether isoflavones are able to activate LXRα, directly or indirectly, and whether LXRα can mediate the transcriptional effects of SP on the regulation of genes involved in fatty acid synthesis, RCT and insulin sensitivity.

Methods

Animals, diet formulation and feeding

Male Lxrα −/− mice were obtained from Gustafsson’s laboratory. These mice were backcrossed for ten generations in C57BL/6J mice [30]. C57BL/6J control mice were purchased from Taconic Europe (Lille Skensved, Denmark). Male mice at 3 to 4 months of age had free access to water and one of the experimental diets. These isocaloric diets, the composition and sources of which are shown in electronic supplementary material (ESM) Table 1, were administered in dry form. The isolated SP used in these studies had 88% purity. The Lxrα −/− or WT mice were divided into four experimental groups as follows (n = 10 each): (1) 20% casein; (2) 20% casein plus 2% cholesterol; (3) 20% SP; and (4) 20% SP plus 2% cholesterol. The animals were housed in microisolators with a 12 h light/dark cycle and received the experimental diets for 28 days. At the end of the study, the animals were fasted for 8 h, killed by carbon dioxide inhalation and decapitated. The blood was collected and serum, obtained by centrifugation at 1,000 g, stored at −70°C until further analysis. Liver, ileum and gall bladder were frozen in liquid nitrogen and stored at −70°C until further analysis. The animal protocol was approved by the Animal Committee of the National Institute of Medical Sciences and Nutrition, Mexico City.

Cholesterol and triacylglycerol analysis

Liver lipids were extracted with chloroform-methanol (0.09 g of tissue) according to the method described by Folch [31]. Cholesterol and triacylglycerol in serum and liver were measured with an enzymatic colorimetric commercial kit (DiaSys Diagnostic Systems, Holzheim, Germany) in a chemistry analyser (RA-50; Technicon Ames, Tarrytown, NY, USA).

RNA isolation and quantitative PCR

The total RNA from liver and ileum was extracted as described by Chomczynski and Sacchi [32]. Total RNA was reverse-transcribed and PCR amplification performed (Applied Biosystems, Foster City, CA, USA) using TaqMan assays (Applied Biosystems). Assays for each gene were carried out in triplicate in 96-well optical plates with a sequence detection system (ABI Prism 7000; Perkin-Elmer Applied Biosystems, Foster City, CA, USA). β-Actin was used as the invariant control for liver and intestine analyses.

Histological analysis

Liver sections were obtained, fixed by immersion in 10% formaldehyde (vol/vol) dissolved in phosphate buffer and subsequently dehydrated and embedded in paraffin. Sections (3 μm width) were obtained and stained with haematoxylin and eosin.

Fatty acid analysis

Total lipids from the liver were extracted as described by Folch [31] and the fatty acids then methylated as previously described [33]. The methylated fatty acids were analysed by gas chromatography (Agilent 6850; Agilent, Santa Clara, CA, USA) with flame ionisation detector, (Agilent)using an HP-1 capillary column (J&W Scientific, Albany, CA, USA).

Bile acid measurements

The amounts of bile acid were determined from the gall bladder (~0.05 ml bile per animal) and faeces. For measurement of faecal bile acid excretion, stools from WT and Lxrα −/− mice were collected during the final 3 days of the study, and dried, weighed and ground. The bile acid was derivatised as described by Keller and Jahreis [34]. The trimethylsilyl bile acids were analysed by gas chromatography (Agilent 6850 with flame ionisation detector) using a capillary column (Innowax; J&W Scientific) as previously described [34].

Cell culture and co-transfections

HepG2 cells were grown in DMEM, with glucose (25 mmol/l), 10% fetal bovine serum, penicillin (200 IU/ml) and streptomycin (100 mg/ml), in a humidified CO2 incubator at 37°C. Cells were co-transfected with the empty expression vector or an expression vector containing Lxrα or Lxrβ, along with a reporter vector containing three repeats of the consensus LXRE cloned in pGL3 Basic (Promega, Madison, WI, USA). Cells were seeded in 24-well plates, co-transfected for 8 h and genistein or daidzein added at the concentrations indicated. The synthetic ligand, GW3965 (Enzo Life Sciences, Farmingdale, NY, USA), and the natural oxysterol, 22(R)-hydroxycholesterol (Sigma-Aldrich, St. Louis, MO, USA), were used as positive controls. After 16 h of incubation, the cells were collected and lysed. The luciferase activity was measured using a commercial luciferase assay kit (luciferin-ATP; BioThema, Umeå, Sweden) and a luminometer (Infinite 200; Tecan, San Jose, CA, USA).

Culture of primary mouse hepatocytes and transfection

Mouse hepatocytes were isolated by the collagenase perfusion technique and separated from non-parenchymal liver cells by centrifugation at 325 g [35]. On day 0, primary hepatocytes were plated in a six-well plate (9.6 cm2/well) (Corning CellBIND, Tewksbury, MA, USA). On day 1, mouse Lxrα or Lxrβ (also known as Nr1h2) expression vectors (400 ng) were transfected using a transfection reagent (FuGENE HD; Roche Diagnostics, Mannheim, Germany). At 4 h after transfection, genistein or daidzein (15 μmol/l), and/or 10 μmol/l T0901317 were added. Total RNA from the hepatocytes was obtained using Trizol reagent.

Protein extraction and western blotting

Primary mouse hepatocytes were homogenised in lysis protein RIPA buffer containing 1 mmol/l sodium fluoride, 2 mmol/l sodium orthovanadate and complete protease inhibitor cocktail tablets (Roche Applied Science, Mannheim, Germany). Total protein (30 μg) was loaded on 8% polyacrylamide gels, separated by SDS-PAGE and transferred to polyvinylidene difluoride (PVDF) membrane. Blots were blocked with non-fat dry milk (Bio-Rad, Hercules, CA, USA) and incubated overnight at 4°C with the following primary antibodies: anti acetyl-CoA carboxylase (ACC) and anti phospho-ACC at Ser79 (pACC) (Millipore, Temecula, CA, USA), and anti-actin (Santa Cruz Biotechnology, Santa Cruz, CA, USA). The bands were analysed using ImageJ 1.42p digital imaging processing software (http://rsb.info.nih.gov/ij/March/27/2012).

OGTT and intraperitoneal insulin tolerance tests

Mice were fasted for 6 h before the OGTT and insulin tolerance test (ITT). The OGTT was performed by administering glucose (1.0 g/kg body weight) by gavage. Insulin (0.8 unit/kg) was injected intraperitoneally. Blood samples were obtained via tail nick and glucose was measured with a glucometer (OneTouch Ultra Accu-Chek Sensor; Roche Diagnostics). The AUC values were calculated as follows: [(glc)T1 + (glc)T2)*(T2 − T1)]/2 [36], where glc is glucose and T is time.

Statistical analysis

The results are presented as the means ± SEM. The statistical analysis was performed using one-way ANOVA followed by Fisher’s protected least-square difference test to determine significant differences between the groups. Differences were considered significant at p < 0.05. Analysis was by Statview statistical analysis program, version 4.5 (Abacus Concepts, Berkeley, CA, USA).

Results

Effect of SP on body weight and serum lipids

No difference in body weight was observed between WT and Lxrα −/− mouse groups after 28 days (ESM Table 2). However, the liver weight of Lxrα −/− mice fed diets with cholesterol was significantly higher than that of the other groups.

Serum cholesterol levels in all groups were comparable. The addition of cholesterol to the diets significantly increased serum cholesterol levels in Lxrα −/−, but not in WT mice (Fig. 1a). The profile of serum cholesterol particles in Lxrα −/− mice showed a significant increase in the number of atherogenic particles in comparison with WT mice. In particular, WT mice fed SP had the lowest amount of atherogenic particles and the highest concentration of HDL particles (Fig. 1b). Serum triacylglycerol did not increase in Lxrα −/− mice on either an SP or a casein diet (Fig. 1c).

Fig. 1
figure 1

Serum lipid concentrations in WT and Lxrα −/− mice fed soy or casein (Cas) diets with or without cholesterol (Chol). (a) Serum cholesterol concentrations and (b) the lipoprotein profile of cholesterol particles; black diamonds, Soy Lxrα +/+; black triangles, Cas Lxrα +/+; black circles, Soy+Chol Lxrα +/+; black squares, Cas+Chol Lxrα +/+; white diamonds, Soy Lxrα −/−; white triangles, Cas Lxrα −/−; white circles, Soy+Chol Lxrα −/−; white squares, Cas+Chol Lxrα −/−. (c) Serum triacylglycerol concentrations. The mice were fed the experimental diets for 1 month and blood was obtained after overnight fasting. The results are expressed as means ± SEM; n = 10 mice per group; mean values with different superscript letters within a row are significantly different (p < 0.05)

Regulation of hepatic lipids by SP

Macroscopically, livers of WT mice fed casein (ESM Fig. 1a) were pale compared with those fed SP (ESM Fig. 1b), this difference being augmented in Lxrα −/− mice (ESM Fig. 1e, f). The addition of cholesterol to the SP or casein diets increased the fatty appearance (ESM Fig. 1c, d), particularly in Lxrα −/− mice (ESM Fig. 1g, h).

The livers of WT mice fed high-cholesterol diets showed increased accumulation of cholesterol and triacylglycerol compared with WT mice on diets without cholesterol, although those fed SP had lower lipid levels (ESM Fig. 2i, j). The livers of Lxrα −/− mice fed either SP or casein contained significantly higher concentrations of both lipids than the livers of corresponding WT mice. Dietary cholesterol dramatically increased the hepatic concentration of cholesterol and triacylglycerol in Lxrα −/− mice (ESM Fig. 2i, j). Lxrα −/− mice fed a casein diet exhibited 10.2- and 3.3-fold increased levels of hepatic cholesterol and triacylglycerol, respectively, compared with those fed a diet without cholesterol. In Lxrα −/− mice fed an SP diet with cholesterol, the corresponding increases were 7.3- and 0.8-fold, respectively, compared with those on diets without cholesterol.

Effect of SP on hepatic histological abnormalities

WT mice fed casein had some hepatocytes with small- and medium-sized cytoplasmic lipid vesicles (ESM Fig. 2a), with addition of cholesterol increasing the number of large lipid vesicles (ESM Fig. 2c). These histological changes were clearly smaller in WT mice fed the SP diet (ESM Fig. 2b, d). In contrast, hepatocytes of Lxrα −/− mice fed casein only showed large lipid vesicles (ESM Fig. 2e); the addition of cholesterol to the diet produced a greater increase in the number of large lipid vesicles, resulting in severe hepatic steatosis, and abundant chronic inflammatory infiltrate and necrosis (ESM Fig. 2g). These abnormalities were associated with a dramatic increase (5.2-fold) in the serum levels of alanine transaminase (ALT) compared with Lxrα −/− mice fed the casein diet without cholesterol (ESM Fig. 2k). Although these histological changes were clearly attenuated in Lxrα −/− mice fed SP or SP with cholesterol (ESM Fig. 2f, h), the addition of cholesterol increased serum ALT levels by 4.9-fold (ESM Fig. 2k).

Regulation by SP of hepatic genes involved in bile acid synthesis and RCT

On measuring the expression of genes that are responsible for the synthesis of bile acids and dependent on LXR, we observed significant differences in the Cyp7a1 mRNA concentration between Lxrα −/− mice and WT mice fed either the casein or SP diets (Fig. 2a). Whereas only the expression of Cyp27a1 was increased in Lxrα −/− mice fed diets without cholesterol, the addition of cholesterol repressed Cyp27a1 expression in WT and Lxrα −/− mice (Fig. 2b).

Fig. 2
figure 2

Hepatic expression of genes encoding the enzymes of bile acid metabolism, and bile acid concentration in WT and Lxrα −/− mice fed casein or SP diets with or without cholesterol (Chol). (a) Cyp7a1 and (b) Cyp27a1 mRNA expression. (c) Total bile acid concentration in bile and (d) total faecal bile acids. The results are expressed as means ± SEM; n = 10 mice per group; mean values with different superscript letters within a row are significantly different (p < 0.05)

We observed significant changes in the expression of cholesterol transporters. The addition of cholesterol to the diet in WT and Lxrα −/− mice significantly increased the expression of the ATP-binding cassette (ABC) sub-family genes Abca1, Abcg5 and Abcg8, with groups fed SP having the highest respective mRNA levels in the liver (ESM Fig. 3a–c). The expression of these transporters in Lxrα −/− mice is possibly mediated by LXRβ. Analysis of the relative expression of these genes, based on the abundance of hepatic cholesterol and calculated as the ratio of the relative mRNA abundance and the hepatic cholesterol concentration, showed that Lxrα −/− mice fed SP or casein had dramatically decreased expression upon the addition of dietary cholesterol in comparison with WT mice, indicating in the former an inability to remove cholesterol (ESM Fig. 3d–f).

There were no differences in the total amount of bile acids in the bile (Fig. 2c). However, faecal total bile acids excreted from WT mice on an SP diet with cholesterol were ninefold higher than in the control groups or in mice fed casein with cholesterol (Fig. 2d). These differences were not observed in the Lxrα −/− groups.

SP regulates bile acid excretion and intestinal expression of RCT genes

To understand this change in bile acid excretion, we measured the abundance of cholesterol and bile acid transporters in the ileum. The expression of Fxr (also known as Nr1h4), Ibabp (also known as Fabp6) and Ibat (also known as Slc10a2) decreased in WT mice fed high-cholesterol diets (ESM Fig. 4a–c), whereas Lxrα −/− mice fed diets with or without cholesterol did not express these genes at levels above control levels. Interestingly, WT mice fed SP without cholesterol showed increased Abcg5, Abcg8 and Abca1 mRNA levels compared with those fed casein (ESM Fig. 4d–f). Moreover, the addition of cholesterol increased the expression of these genes by 111%, 23% and 40%, respectively, compared with expression in mice fed casein. These differences were abolished when Lxrα −/− mice were fed the experimental diets with or without cholesterol.

Regulation of LXR isoform activity by SP isoflavones

To explore whether soy isoflavones were responsible for activating production of the ABC transporters via LXRα or LXRβ, we conducted functional assays in HepG2 cells to analyse the effect of the soy isoflavones (genistein and daidzein) on the LXREs. The addition of daidzein or genistein did not increase luciferase activity compared with the respective controls, whereas the synthetic ligand, GW3965, and the natural oxysterol, 22(R)-hydroxycholesterol, significantly increased luciferase activity (ESM Fig. 5a, b). These data strongly suggest that the observed increase in the expression of genes encoding the ABC transporters was not mediated via a direct LXR agonist effect. Studies with LXRβ showed similar results (data not shown).

Next, we studied whether isoflavones could regulate LXRα indirectly. We showed that the incubation of hepatocytes with T0901317 increased Srebp1 (also known as Srebf1) mRNA abundance by 18.4-fold. Interestingly, the hepatocytes transfected with the Lxrα vector incubated with daidzein or genistein significantly reduced the stimulatory effect of T0901317 by 40% and 51%, respectively (Fig. 3a). However, the repressive effect of isoflavones on Srebp1 expression was not observed in hepatocytes transfected with the Lxrβ vector (Fig. 3c). Similar effects were observed for Abca1, another LXR target gene (Fig. 3b–d). Surprisingly, the response of the Abcg5 and Abcg8 genes was the opposite of that observed for Srebp1 and Abca1. The incubation of stimulated transfected hepatocytes with the Lxrβ vector in the presence of isoflavones further stimulated the mRNA expression of Abcg5 and Abcg8, particularly upon the addition of genistein (2.4- to 2.8-fold) (Fig. 4a–d). These data suggest that isoflavones may differentially regulate the expression of LXR target genes.

Fig. 3
figure 3

The effect of soy isoflavones on the expression of Srebp1 and Abca1 in mouse hepatocytes overexpressing Lxrα or Lxrβ. The cells were transfected with the expression vector for (a, c) Lxrα and (b, d) or Lxrβ, and the hepatocytes were incubated with daidzein (DAD) or genistein (GEN), with or without the synthetic LXR ligand T0901317 (T090). The experiments were performed in quadruplicate; mean values with different superscript letters within a row are significantly different (p < 0.05)

Fig. 4
figure 4

The effect of soy isoflavones on the expression of Abcg5 and Abcg8 in mouse hepatocytes overexpressing Lxrα or Lxrβ. Cells were transfected with the expression vector for (a, c) Lxrα or (b, d) Lxrβ, and hepatocytes were incubated with daidzein (DAD) or genistein (GEN), with or without the synthetic LXR ligand T0901317 (T090). The experiments were performed in quadruplicate; mean values with different superscript letters within a row are significantly different (p < 0.05)

There is evidence that the activity of LXRα can be downregulated by phosphorylation of AMP-activated protein kinase (AMPK) [25]. We therefore used the AMPK inhibitor, compound C, to explore whether isoflavones could regulate the activity of LXRα via AMPK. As observed in Fig. 5a, the ability of T0901317 to increase Srebp1 expression by activating LXRα was highly stimulated by the addition of compound C. Nonetheless, genistein was able to repress the synergistic effect of T0901317, even in the presence of compound C. Conversely, in hepatocytes overproducing LXRβ, genistein significantly increased the expression of Abcg5 in the presence of T0901317 with compound C (Fig. 5b). We showed that genistein was able to activate AMPK, since this isoflavone increased the phosphorylation of ACC, a well-known target protein of AMPK. The addition of T0901317 did not alter ACC phosphorylation, while compound C, an inhibitor of AMPK, prevented the stimulatory effect of genistein on ACC phosphorylation (ESM Fig. 6a, b).

Fig. 5
figure 5

(a) The effect of genistein on Srebp1 expression in mouse hepatocytes overexpressing Lxrα and incubated with the synthetic LXR ligand, T0901317 (T090), and/or with compound C (CC), an inhibitor of AMPK. (b) The effect of genistein on the expression of Abcg5 in mouse hepatocytes overexpressing Lxrβ; incubation as above (a). The experiments were performed in quadruplicate; mean values with different superscript letters within a row are significantly different (p < 0.05)

Regulation by SP of expression of genes involved in cholesterol and fatty acid synthesis

We measured the mRNA levels of several genes involved in hepatic cholesterol and fatty acid metabolism in WT and Lxrα −/− mice. Our data show that the expression of Srebp2 (also known as Srebf2) and HMG-CoA reductase in the livers of WT mice was significantly reduced in animals fed cholesterol diets (ESM Fig. 7d, e); however, mRNA levels for the LDL receptor were not significantly different among the WT groups (ESM Fig. 7f). Interestingly, Lxrα −/− mice fed casein or SP diets without cholesterol had significantly higher expression of Srebp2 mRNA and its target genes, which encode HMG-CoA reductase and the LDL receptor, than the corresponding WT mice (ESM Fig. 7d, e). However, levels of Srebp1 mRNA changed only slightly among WT groups (ESM Fig. 7a). Again, the expression of Srebp1 and its target gene Fasn were significantly increased in the liver of Lxrα −/− mice fed diets without cholesterol, an effect that was repressed by the addition of cholesterol (ESM Fig. 7b).

Hepatic Scd1 showed a different pattern of expression in WT and Lxrα −/− mice (ESM Fig. 7c). Nonetheless, there was no significant correlation between Scd1 mRNA and the saturated:monounsaturated fatty acid ratio in the liver. This ratio was dependent upon the addition of cholesterol to the diets (Table 1). WT or Lxrα −/− mice fed diets with cholesterol had lower ratios than the corresponding groups fed diets without cholesterol. The ratios were determined mostly by the high concentration of oleate in the livers of mice fed high-cholesterol diets and were independent of the type of dietary protein (Table 1).

Table 1 Concentration of saturated and monounsaturated fatty acids in the liver of WT and Lxrα −/− mice fed casein or SP diets with or without cholesterol for 28 days

SP regulates insulin sensitivity

There is strong evidence that the accumulation of lipids in muscle and liver is associated with the development of insulin resistance [37]. Considering the elevated amounts of lipids in the livers of Lxrα −/− mice fed casein or SP high-cholesterol diets, we investigated whether the type of dietary protein could differentially modify insulin sensitivity as measured by the OGTT and ITT. Our results showed that WT mice fed an SP high-cholesterol diet had lower fasting blood glucose levels than mice fed a casein high-cholesterol diet (Fig. 6a). Serum glucose disappearance following an oral glucose load in WT mice fed SP was significantly improved compared with mice fed casein (Fig. 6a, b). In addition, the ITT showed that glucose disappearance from the blood was improved by SP-fed mice compared with casein-fed mice (Fig. 6c) These data show that WT mice fed SP had better insulin sensitivity than the corresponding mice on a casein diet (Fig. 6d). However, in Lxrα −/− mice fed casein or SP high-cholesterol diets, no significant differences were observed during the OGTT or ITT (Fig. 6a, c). Lxrα −/− mice had a lower basal serum glucose concentration and improved insulin response compared with WT mice fed a casein high-cholesterol diet, suggesting that the presence of LXRα may partially affect insulin sensitivity in mice fed casein (Fig. 6d).

Fig. 6
figure 6

OGTT and ITT results in WT and Lxrα −/− mice. (a) The time curve of glucose concentrations in blood during an OGTT in WT (black circles) and Lxrα −/− (white circles) mice fed casein, or in WT (black squares) and Lxrα −/− (white squares) mice fed SP high-cholesterol diets for 1 month. (b) The AUC from OGTT detailed in (a). Black bars, casein diet; white bars, SP diet. (c) Time curve of glucose concentrations in blood during an ITT in WT and Lxrα −/− mice as in (a) above. (d) The AUC from the ITT. Black bars, casein diet; white bars, SP diet. The results are expressed as the means ± SEM; n = 5–7 mice per group; *p < 0.05 for the difference between the soy and casein groups

Discussion

The present study demonstrates that the consumption of an SP diet with cholesterol reduces the accumulation of hepatic lipids even in Lxrα −/− mice. Macroscopic examination of the livers of WT mice fed SP cholesterol diets showed that they were less fatty and had a better consistency than those of mice fed a casein diet with cholesterol. Moreover, the effect of SP was also observed in the livers of Lxrα −/− mice, which normally have a fatty appearance and a very soft texture, as previously reported [18, 30]. In fact, livers of Lxrα −/− mice fed an SP diet showed a decrease in large lipid deposits and a reduced inflammatory response. These data suggest that SP has an anti-steatotic effect, as observed in previous studies [1315].

Previous studies and the present work have shown that the elimination of excess dietary cholesterol to prevent its hepatic accumulation occurs via increased faecal bile acid excretion [18]. We did not observe an increase in Cyp7a1 and Cyp27a1 expression in mice fed long-term high-cholesterol diets, nor did we see significant differences in the concentration of total bile acids in the bile. However, WT mice fed an SP diet with cholesterol had increased liver and intestinal expression of Abcg5, Abcg8 and Abca1, an effect not observed in Lxrα −/− mice. The highest levels of these transporters in WT mice were associated with the highest concentration of total faecal bile acids.

Although the expression of genes involved in fatty acid synthesis, bile acid synthesis and RCT is regulated by LXR [19], our data show that consumption of SP upregulated some of the genes involved in these processes and downregulated the expression of some others. SP isoflavones were able to mediate the effects of LXR, since they are ligands for other nuclear receptors, such as the peroxisome proliferator-activated receptors and estrogen receptors [2729, 38]. However, our data indicate that isoflavones are unable to work as direct LXR agonists and thereby stimulate transcription via a classical LXRE, a finding that is in agreement with previous results [39].

Our results in hepatocytes suggest that isoflavones are able to regulate LXR activity indirectly by promoting the phosphorylation, possibly mediated via AMPK, of LXRα or LXRβ, leading to opposing effects on the expression of certain genes. Isoflavones reduced the expression of Srebp1 and Abca1 via LXRα, but at the same time increased the expression of Abcg5 and Abcg8 via LXRβ. Our data on the use of compound C, an inhibitor of AMPK, support the hypothesis that isoflavones are able to activate AMPK, leading to the repression of Srebp1 by the phosphorylation of LXRα and the overexpression of Abcg5 by the phosphorylation of LXRβ. Our group and others have also demonstrated that isoflavones are able to increase the phosphorylation state of AMPK [40, 41]. More studies are needed to understand the full mechanism of this activation.

The accumulation of hepatic lipids has recently been described as one of the main causes for the development of the metabolic abnormalities obesity, primarily insulin resistance and dyslipidaemia [42]. Our results clearly show that feeding an SP high-cholesterol diet to WT mice improves insulin sensitivity compared with WT mice on a casein high-cholesterol diet. These results are in agreement with a previous study indicating that SP improves insulin sensitivity [16]. However, this beneficial effect was abolished following the deletion of Lxrα, indicating that the effect of SP on glucose metabolism involves LXRα.

Interestingly, Lxrα −/− mice had better insulin sensitivity independently of the type of diet, suggesting that the absence of LXRα in vivo has a beneficial rather than a negative effect on insulin sensitivity. This is in agreement with previous data from our group, which showed that Lxrα −/− mice are more insulin-sensitive than WT mice, even after a high-fat diet leading to liver steatosis [36]. The beneficial effects of a lack of LXRα occur via LXRβ, since this isoform acts in the opposite direction to LXRα in glucose metabolism and insulin sensitivity [36]; it is also the main isoform in muscle tissue [43] and is present in adipocytes, leading to an improvement of whole-body insulin sensitivity. This suggests that the improvement of insulin sensitivity by an SP diet could be mediated by LXRα. In support of this finding, Lxrβ −/− mice fed the SP diet showed improved insulin sensitivity (data not shown).

In conclusion, the differential expression of genes regulated by LXR after the consumption of SP is in part due to the capacity of isoflavones, particularly genistein, to regulate the activity of the LXR isoforms, LXRα and LXRβ, in opposing directions, possibly mediated via AMPK. The different effects of isoflavones on these nuclear receptors are likely to explain the beneficial results observed in various studies of experimental animals and in humans, where the consumption of an SP diet was shown to reduce serum cholesterol and prevent the excessive accumulation of hepatic lipids, as well as improving insulin sensitivity [1315, 44, 45]. Further studies are needed to understand the molecular mechanisms of the regulation of LXR isoforms by isoflavones.