Sodium Butyrate Stimulates Expression of Fibroblast Growth Factor 21 in Liver by Inhibition of Histone Deacetylase 3

Human HepG2 hepatocytes from the American Type Culture Collection (Manassas, VA) were maintained in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% FCS. Butyrate (19–137) was obtained from Millipore (Billerica, MA). Trichostatin A (TSA) (58880–19–6) was obtained from A.G. Scientific (San Diego, CA). Bezafibrate (B7273), fenofibrate (F6020), hexanoate (C4026), and sodium 3-hydroxybutyrate (BOH) (54965) were purchased from Sigma-Aldrich (St. Louis, MO).

Dietary obese mice were generated by feeding male C57BL/6J mice (The Jackson Laboratory, Bar Harbor, ME) a high-fat diet (D12331, 58% calories in fat; Research Diets, New Brunswick, NJ) for 20 weeks. The lean control mice were fed chow diet (5001, 13% calories in fat; LabDiet, Richmond, IN). The experiments on FGF21 KO mice (23) were conducted at the University of Hong Kong laboratory of A.X. All of the mice were housed in the animal facility with a 12-h light/dark cycle and constant temperature (22–24°C). The mice were housed at three to four mice per cage with free access to water and diet. All the procedures were approved by the institutional animal care and use committee at the Pennington Biomedical Research Center.

Butyrate or control agent was delivered by intraperitoneal injections in mice after overnight fasting at the following dosages: butyrate (500 mg/kg body wt), TSA (0.8 mg/kg body wt), hexanoate (625 mg/kg body wt), and bezafibrate (100 mg/kg body wt). Mice were assigned randomly to the treatment or control groups at five to eight mice per group. In the repeating experiments, the mice were allowed to recover for at least 7 days before the next injection. Blood and tissue samples were stored at −80°C.

Concentrations of FGF21 in serum and liver tissue were quantified using ELISA kits (Antibody & Immunoassay Services, University of Hong Kong). The assay was proven to be highly specific to mouse FGF21, with no cross-reaction to other members of the FGF family. The intra- and interassay variations were 4.2 and 7.6%, respectively. Ketone body in the serum was determined with β-hydroxybutyrate concentration using a ketone body assay kit (BioAssay Systems, Hayward, CA).

Total RNA was extracted from frozen tissues using Tri-Reagent (T9424, Sigma-Aldrich). RNA was used for generation of cDNA (Bio-Rad, Hercules, CA). Quantitative real-time PCR (qRT-PCR) was performed in a 7900HT Fast Real-Time PCR System (Applied Biosystems, Foster City, CA) using SYBR Green Master Mix (Applied Biosystems). Primers were made according to published studies (7,14,24). All samples were analyzed in duplicate, and FGF21 signal was normalized with cyclophilin signal (the internal control).

Respiratory exchange ratio (RER) was monitored with the Comprehensive Laboratory Animal Monitoring System (Columbus Instruments, Columbus, OH) as described previously (17). To inhibit FGF21 activity, a monoclonal rabbit–anti-mouse FGF21 antibody (Antibody & Immunoassay Services) was injected at 3 μg per mouse. A rabbit IgG was injected at the same dose in the control mice.

HEK 293 cells were plated in 24-well plates (4 × 10⁵ cells per well) and transfected with FGF21 luciferase reporter (0.2 µg DNA per well) using Lipofectamine 2000 (Invitrogen, Carlsbad, CA). The FGF21 luciferase reporter was a gift from Dr. Steven Kliewer at the Southwestern Medical Center, University of Texas, Dallas, Texas (14). The luciferase assay was conducted using the dual luciferase substrate system (E1501; Promega, Madison, WI), and the result was normalized with the internal control Renilla luciferase. Each experiment was repeated at least three times with consistent results.

Chromatin immunoprecipitation (ChIP) was conducted in fresh liver tissues collected at 2 h after butyrate injection in mice. Protein-DNA cross-linking was performed in the samples with 1% formaldehyde at room temperature for 15 min and terminated by glycine. Liver nuclei were isolated, and chromosome DNA was broken into fragments of 400–1,200 base pair (bp) by sonication. Immunoprecipitation was conducted with antibodies to PPAR-α (sc-9000; Santa Cruz Biotechnology), HDAC3 (ab7030; Abcam), or RNA polymerase II (sc-9001; Santa Cruz Biotechnology). IgG was used in controls for the nonspecific signal. DNA signal was quantified using SYBR green qRT-PCR. The PCR primers (Forward: 5′-AGGGCCCGAATGCTAAGC-3′; Reverse: 5′-AGCCAAGCAGGTGGAAGTCT-3′) cover the PPAR-α binding site (−1,119/−1,044) in the mouse FGF21 gene promoter (14).

In this study, data are presented as the mean ± SEM from multiple samples or repeats. All of the in vitro experiments were conducted a minimum of three times. Student t test or two-way ANOVA was used in the statistical analysis with significance P ≤ 0.05.

In hepatocytes, FGF21 secretion is controlled at the transcriptional level, and PPAR-α is an activator of FGF21 gene promoter. PPAR-α ligand induces FGF21 transcription and protein expression in hepatocytes. It is not known if sodium butyrate is able to induce FGF21 expression. To address this question, we investigated regulation of FGF21 expression by butyrate in human HepG2 hepatocytes. In the study, bezafibrate and fenofibrate were used as positive controls for the induction of FGF21 expression. FGF21 was induced in mRNA and protein by either positive control (Fig. 1A and B). In the absence of positive controls, butyrate induced FGF21 expression in mRNA and protein dramatically (Fig. 1A and B). The induction was in a dose-dependent manner for both mRNA and protein (Fig. 1C and D). Butyrate is a sodium salt of short-chain fatty acid with HDAC inhibitor activity. To differentiate the activities of fatty acid and HDAC inhibitor, we used short-chain fatty acids, BOH (four carbon fatty acid) and hexanoate (six carbon fatty acid), in the control. At the same concentration to butyrate, the two fatty acids did not exhibit any significant activity in the regulation of FGF21 mRNA or protein (Fig. 1A and B). Since these two fatty acids are not HDAC inhibitors, the data suggest that butyrate may act through inhibition of HDACs. To test this possibility, we used TSA, a classical HDAC inhibitor. Like butyrate, TSA increased FGF21 in mRNA and protein in a dose-dependent manner (Fig. 1E and F). These data suggest that in the absence of PPAR-α ligand, butyrate increases FGF21 expression in hepatocytes. This FGF21 response is related to the inhibition of HDACs by butyrate, but not the fatty acid nature of butyrate.

To test the butyrate activity in vivo, we examined serum FGF21 in obese mice after butyrate treatment, which was administrated through intraperitoneal injection. In the mouse study, the same controls were used, such as PBS in the vehicle control, bezafibrate as a positive control, TSA for HDAC inhibition, and hexanoate as the fatty acid control. Bezafibrate increased serum FGF21 by onefold as expected. Butyrate exhibited the same activity in the induction of serum FGF21 (Fig. 2A). TSA also induced the FGF21 by onefold (Fig. 2B). No significant change was observed for PBS (Fig. 2A–E) or hexanoate (Fig. 2C). To determine the source of FGF21, we examined FGF21 expression in the liver at 2 h after injection. The induction in FGF21 was observed for bezafibrate and butyrate (Fig. 2D and E). These data suggest that butyrate induces serum FGF21 in mice by activation of FGF21 expression in liver. The activity is related to the HDAC inhibition by butyrate.

Induction of ketone body production is a major activity of FGF21, which induces gene expression in the ketogenesis pathway. To determine the biological activity, we examined the serum ketone bodies by measuring β-hydroxybutyrate in the butyrate-treated mice. Butyrate significantly increased β-hydroxybutyrate, and TSA exhibited a similar activity (Fig. 3A). Carnitine palmitoyltransferase 1a (CPT1a) and 3-hydroxy-3-methylglutaryl-CoA synthase 2 (HMGCS2) are FGF21 responsive genes and are required for ketogenesis by FGF21 (25,26). In the liver, the two genes were increased in mRNA by more than onefold in response to butyrate (Fig. 3B). The increase in blood ketone body was positively associated with the change in serum FGF21 (Fig. 3C and D). To test whether butyrate enhances ketogenesis via FGF21, we conducted the study in FGF21 KO mice. KO mice and wild-type (WT) mice exhibited similar levels of ketone body at the baseline. In response to butyrate, the ketone body level was significantly lower in the KO mice (Fig. 3E). The WT mice exhibited 1.8 mmol/L increase in β-hydroxybutyrate, while KO mice exhibited only 0.5 mmol/L increase in β-hydroxybutyrate. The responses in CPT1a and HMGCS2 were also significantly lower in the KO mice (Fig. 3F). This group of data suggests that butyrate induces ketogenesis. FGF21 is required for the butyrate activity.

The increase in ketone body production suggests elevation in fatty acid use by the liver, which produces ketone bodies through β-oxidation of long-chain fatty acids. Fatty acids and glucose are substrates in mitochondria for the production of ATP in the peripheral tissues. There is competition between the two substrates. When fatty acid usage is enhanced, carbohydrate use will be reduced. To test fatty acid oxidation in the whole body, we determined RER using the metabolic chamber. RER is a volume ratio of CO₂ exhaled versus O₂ inhaled in the body. A decrease in RER suggests an increase in fatty acid use. In response to the butyrate injection, the mice exhibited a 6% reduction in RER value (Fig. 4A). The reduction was blocked when the serum FGF21 was neutralized by the FGF21 antibody, which decreased FGF21 protein in the serum (Fig. 4B). In the control, a nonspecific IgG was used. The butyrate activity was not affected (Fig. 4A), and FGF21 protein was not reduced. In the study, butyrate did not change the O₂ consumption and spontaneous physical activity in the mice (data not shown). These results suggest that butyrate increases β-oxidation of fatty acids in the whole body and that the activity requires FGF21. The acute treatment by butyrate did not change energy expenditure in the obese mice.

FGF21 expression is controlled at the transcriptional level. To understand the molecular mechanism of butyrate action, we investigated FGF21 gene transcription. The butyrate response element was investigated in the FGF21 gene promoter. The element was searched through screening the promoter fragments of different length spanning −1,497 and 5 (14). The activity of the longest promoter (−1,497/5 bp) was induced significantly by butyrate or bezafibrate as indicated by the luciferase activity (Fig. 5A). In the same condition, the shorter (<977 bp) promoters did not respond to butyrate. There are two PPAR-α response elements in the long promoter, and the distal element is absent in the shorter promoter. The data suggest that the distal element is responsive to butyrate. When butyrate was compared with TSA in the induction of the long promoter, both agents exhibited dose-dependent activity in the activation of the reporter with a similar strength (Fig. 5B). This group of data suggests that butyrate induces FGF21 promoter activity by inhibition of HDACs. The distal PPAR-α response element may be activated by butyrate.

Butyrate inhibits multiple HDACs in class I and II, which contain >10 isoforms. Butyrate inhibits HDAC1, HDAC2, and HDAC3. To identify the HDAC isoform that is responsible for FGF21 promoter activation, we used gene knockdown. Vector-based small interfering RNA (siRNA) was used to knock down each of HDAC1, HDAC2, and HDAC3 in transient transfection (Supplementary Data 1). The FGF21 reporter was used to monitor the knockdown effect. HDAC3 knockdown increased the promoter activity by fivefold (Fig. 5C), while knock down of HDAC1 and HDAC2 generated only weaker responses in the promoter. The data suggest that HDAC3 is the major HDAC isoform in the FGF21 gene promoter. To understand the relationship of HDAC3 and PPAR-α element, we overexpressed PPAR-α to activate the element specifically and knocked down HDAC3. In this model, the PPAR-α activity was enhanced by 26-fold in the presence of HDAC3 knockdown (Fig. 5D). The knockdown effect was significantly reduced in the shorter promoters of −977/5 or −98/5 (Fig. 6D). In the shortest promoter (−66/5) that does not have a peroxisome proliferator response element (PPRE), the effect of HDAC3 knockdown was not observed. The data suggest that HDAC3 may regulate both PPREs in the FGF21 gene promoter. However, the distal PPRE is more sensitive to HDAC3. This group of data suggests that butyrate activates the FGF21 gene promoter through the distal PPRE by inhibiting HDAC3, which inhibits the gene promoter at the PPREs.

The knockdown data suggest that HDAC3 suppresses PPAR-α in the FGF21 gene promoter. HDAC3 is likely a component of PPAR-α corepressor that inhibits the transcription activity in the absence of PPAR-α ligand to block recruitment of RNA polymerase II. In response to PPAR-α activation, the corepressor will be disabled, leading to recruitment of RNA polymerase II in the initiation of gene transcription. Those possibilities are supported by the functional assay of the FGF21 gene promoter. To prove the HDAC3–PPAR-α interaction at the protein level, we performed ChIP assay at the distal PPRE in the FGF21 gene promoter in liver tissues. Bezafibrate increased PPAR-α binding to the FGF21 promoter DNA (Fig. 6A). The PPAR-α activity was associated with an increase in RNA polymerase II signal (Fig. 6B) and a decrease in HDAC3 signal (Fig. 6C). Butyrate did not increase PPAR-α binding to the FGF21 gene but did increase the polymerase II signal and reduce the HDAC3 signal (Fig. 6A–C). These data suggest that HDAC3 interacts with PPAR-α in the FGF21 gene promoter. Butyrate activates PPAR-α without increasing its DNA-binding activity. Bezafibrate increases the DNA binding of PPAR-α.

The above data suggest that butyrate and bezafibrate induce FGF21 expression through the PPRE. However, they act through different mechanisms, as suggested by the ChIP assays. The two agents may have an additive or synergistic interaction in the induction of FGF21 expression. To test the possibility, we combined the two agents in the induction of FGF21 expression. In the HepG2 cell line, bezafibrate induced FGF21 protein expression, as indicated by the protein abundance in the whole-cell lysate. In the presence of butyrate, the bezafibrate effect was enhanced significantly (Fig. 7A). In obese mice, the bezafibrate activity was also enhanced by butyrate, as indicated by the change in serum FGF21 protein (Fig. 7B). In liver tissue, butyrate enhanced the bezafibrate activity with an increase in FGF21 mRNA and protein (Fig. 7C and D). Butyrate also enhanced bezafibrate activity in the induction of ketone body production. After treatment by bezafibrate, the obese mice did not exhibit a significant increase in β-hydroxybutyrate (Fig. 7E). When bezafibrate was administrated together with butyrate, the serum ketone body was elevated more than onefold (Fig. 7E). To understand the mechanism, we examined several PPAR-α target genes that are involved in ketogenesis and fatty acid oxidation, such as HMGCS2, CPT1a, very long-chain acyl-CoA dehydrogenase (ACADVL), acyl-CoA synthase long-chain family member 1 (ACSL1), acyl-CoA oxidase (ACO), and liver fatty acid–binding protein (L-FABP). Bezafibrate induced all of these genes (Fig. 7F). In the presence of butyrate, bezafibrate-induced responses were significantly enhanced in most of these genes (Fig. 7F). These data suggest that butyrate enhances the activity of PPAR-α agonist in the stimulation of FGF21 secretion and ketone body production and the expressions of other PPAR-α target genes.

Our study suggests that butyrate is a new inducer of FGF21. FGF21 is a cytokine/hormone that stimulates use of long-chain fatty acids through β-oxidation in liver in the production of ketone bodies. FGF21 is required for the antiobesity effect of ketogenic diet as shown in the phenotype of FGF21 KO mice (4,5). The studies suggest that FGF21 is required for β-oxidation of the dietary long-chain fatty acids. Induction of FGF21 expression may be an approach to treat obesity. However, there is not much option available in the induction of FGF21. PPAR-α ligand/agonist are able to induce FGF21 expression (4,14,15), but they cannot apply to every patient. As a precursor of PPAR-α ligand, the long-chain fatty acids also induce FGF21 expression (27). Serum FGF21 correlates to the concentration of long-chain fatty acid in humans in a 24-h oscillatory pattern, as shown in our previous study (28). PPAR-γ ligands recently were reported to induce FGF21 (16). Hepatic FGF21 is also induced by triiodothyronine via a PPAR-α–dependent mechanism (29). The current study suggests that butyrate is a powerful inducer of FGF21 in liver. We examined FGF21 in epididymal fat and did not observe the effect (data not shown), suggesting a tissue-specific effect of butyrate. As a dietary component and a product of a dietary fiber, butyrate is an outstanding bioactive agent in the dietary intervention of obesity.

In this study, we investigated the role of FGF21 in the butyrate regulation of fatty acid oxidation. FGF21 activity was tested by antibody neutralization or in FGF21 KO mice. In mice treated with butyrate, FGF21 stimulates ketone body production in liver and increases fatty acid use in whole body. FGF21 represents an endocrine mechanism by which butyrate enhances fatty acid use, as observed in our previous study (17). In that study, butyrate enhanced energy expenditure and prevented dietary obesity in mice. The mechanism is related to PPAR-γ coactivator 1-α activation, which enhances mitochondrial biogenesis. Before this study, it was unknown whether a hormone is involved in the butyrate activity. In the current study, FGF21 was found as a butyrate-responsive cytokine/hormone in the stimulation of ketone body production and fatty acid use. These butyrate activities were blocked by FGF21 antibody or FGF21 gene knockout. These data suggest that FGF21 mediates butyrate activities in the regulation of fatty acid metabolism. FGF21 may also contribute to glucose reduction by butyrate (Supplementary Data 2).

The current study suggests that butyrate stimulates FGF21 expression through activation of PPAR-α. The PPAR-α activation is a result of HDAC3 inhibition by butyrate. We found that butyrate activates the distal PPRE in the FGF21 gene promoter. The mechanism is inhibition of HDAC3. This conclusion is supported by several lines of evidence, such as the gene promoter analysis, HDAC inhibitor activity, HDAC3 knockdown effect, and ChIP assay. Although PPAR-α activates FGF21 transcription, as reported by several groups, it is not clear how PPAR-α is regulated by the nuclear corepressor. We found that HDAC3 is a component of the corepressor. Inhibition of HDAC3 by butyrate, TSA, or gene knockdown led to activation of PPAR-α in the FGF21 gene promoter. In the ChIP assay, HDAC3 interacted with PPAR-α at the PPRE. These data suggest that HDAC3 regulates PPAR-α in the FGF21 promoter. Inhibition of HDAC3 represents a new approach in the induction of FGF21.

Our data suggest that butyrate enhances bezafibrate activity in the induction of FGF21. PPAR-α agonist is the major medicine in the induction of FGF21 (8,15). We observed that butyrate and bezafibrate activated the same set of PPAR-α target genes. In ChIP assay, bezafibrate and butyrate all reduced the HDAC3 activity in the FGF21 gene promoter. However, they act through different mechanisms. Bezafibrate enhanced PPAR-α binding to the promoter DNA, but butyrate did not. Butyrate inhibited HDAC3 enzyme activity in the promoter DNA in the activation of PPAR-α. PPAR-α is able to bind to target DNA in the absence of ligands (14). The difference in mechanism of action suggests why butyrate enhances the bezafibrate activity. The HDAC inhibitors may promote the therapeutic activity of PPAR-α agonist.

In summary, we provide evidence that butyrate increases hepatic production of FGF21 to stimulate β-oxidation of long-chain fatty acids and production of ketone bodies. Our data support that fatty acid oxidation is induced by PPAR-α activation as reported for PPAR-α agonists (30–32). Mechanically, butyrate inhibits HDAC3 activity in the FGF21 gene promoter to enhance PPAR-α function in the transcriptional activation. This butyrate activity is not dependent on the PPAR-α ligand. The study suggests that butyrate is a new chemical inducer of FGF21 in the body. Butyrate is likely to enhance the therapeutic actions of PPAR-α activators and reduce the side effect of the PPAR-α ligands through a dosage reduction.

This study was supported by National 973 project of China (2011CB504001); Major Program of Shanghai Municipality for Basic Research (08dj1400601); and Program for Outstanding Academic Leader (LJ06010) to W.J.; the National Institutes of Health (NIH project grant R01-DK068036-06) to J.Y.; an American Diabetes Association award (1-09-JF-17) to Z.G.; and a Scholarship Award for Excellent Doctoral Student granted by Ministry of Education, State-Sponsored Study Abroad Program, and National Key Basic Research Program of China (973 Program) (2012CB524900) to H.L. The qRT-PCR test and metabolic phenotyping and imaging studies were conducted in the genomic core, phenotyping core, and imaging core of the Pennington Biomedical Research Center, which are supported by the NIH grants (2P30-DK072476-06 and P20-RR021945).

No potential conflicts of interest relevant to this article were reported.

H.L. conducted the experiments and wrote the manuscript. Z.G. and J.Z. contributed to the experiment design and data analysis. X.Y. conducted some experiments. A.X. was involved in experiment design in the study of FGF21 KO mice and the FGF21 assay. J.Y. and W.J. designed the study and were involved in data interpretation and writing the manuscript. W.J. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

The authors thank Dr. Steven A. Kliewer (University of Texas Southwestern Medical Center) for the FGF21 luciferase constructs. The authors also thank Dr. Tara M. Henagan, Zhong Wang, Xian Zhang, and Yongmei Yu (Pennington Biomedical Research Center) for excellent technical support.