Tributyrin Attenuates Metabolic and Inflammatory Changes Associated with Obesity through a GPR109A-Dependent Mechanism

Obesity is linked with altered microbial short-chain fatty acids (SCFAs), which are a signature of gut dysbiosis and inflammation. In the present study, we investigated whether tributyrin, a prodrug of the SCFA butyrate, could improve metabolic and inflammatory profiles in diet-induced obese mice. Mice fed a high-fat diet for eight weeks were treated with tributyrin or placebo for another six weeks. We show that obese mice treated with tributyrin had lower body weight gain and an improved insulin responsiveness and glucose metabolism, partly via reduced hepatic triglycerides content. Additionally, tributyrin induced an anti-inflammatory state in the adipose tissue by reduction of Il-1β and Tnf-a and increased Il-10, Tregs cells and M2-macrophages. Moreover, improvement in glucose metabolism and reduction of fat inflammatory states associated with tributyrin treatment were dependent on GPR109A activation. Our results indicate that exogenous targeting of SCFA butyrate attenuates metabolic and inflammatory dysfunction, highlighting a potentially novel approach to tackle obesity.


Glucose Tolerance Test (GTT)
GTT was performed in 6 h fasted mice in the 12th week of the protocol. In brief, 5 µL of blood from the tail vein was collected before (fasting glucose) and after 15, 30, 60, 90 and 120 min of glucose administration (i.p. at 2.0 g/kg) to the mouse. Blood samples were then mixed with 20 µL of 5% solution of trichloroacetic (TCA). After centrifugation, the supernatant was collected and used for glucose measurements with a commercial kit (Labtest Diagnóstica SA, Minas Gerais, Brazil).

Insulin Tolerance Test (ITT)
ITT was performed in 6 h fasted mice in the 13th week of the protocol. Blood samples were obtained from the tail vein before (0) and after insulin administration (i.p. 0.75 U/kg, Humulin R from Lilly, Indianapolis, USA or Novolin R from Novo Nordisk, Clayton, NC). Samples were deproteinated with TCA and then used for glucose determination. To minimize the effects of the circadian cycle, GTT and ITT were always performed in the afternoon. The rate of glucose disappearance (K itt ) was calculated using the −0.693/t 1/2 formula, where t 1/2 is calculated from the slope of the least-square analysis of the plasma glucose concentration during the linear decay phase.

Liver Analysis
Part of the liver was fixed with a 3.7% buffered formaldehyde solution for at least 8 h at room temperature, dehydrated, processed and embedded in Paraplast (Sigma-Aldrich, St. Louis, MO, USA). To evaluate tissue morphology and the degree of liver steatosis, liver sections (7 µm) from three mice of each group (HFD and HFD+Tb) were stained with hematoxylin and eosin and qualitatively analyzed for fat accumulation and inflammation. To measure liver triacylglycerol (TAG) content, liver samples (100 mg) were homogenized in 4 mL of chloroform and methanol solution (2:1) for 16 h at 4 • C in glass tubes. After this period, 2 mL of a solution 0.6% of NaCl was added and the samples were centrifuged at 425 rcf for 20 min. The organic layer was collected and dried for approximately 2 days. Samples were solubilized in 200 µL of isopropanol and quantified using a commercial kit (Labtest Diagnóstica SA, Minas Gerais, Brazil).

Quantitative Reverse Transcription Polymerase Chain Reaction (RT-PCR)
Total RNA was extracted using Trizol reagent (Invitrogen, Thermo Fisher Scientific, Waltham, MA, USA). RNA was converted to cDNA using the High-Capacity cDNA kit according to the manufacturer's instructions (Applied Biosystems). PCR was performed using Rotor Gene (Qiagen, Venlo, Netherlands) and a kit containing SYBR Green as a fluorescent dye (Power SYBR Green PCR Master Mix, Applied Biosystems, Foster City, CA, USA). Quantification of gene expression was performed using a ∆∆Ct method with ubiquitin C (UBC) and β2-microglobulin as housekeeping genes for eWAT and liver, respectively. The sequences of the primers used are presented in Table S1. To evaluate the best constitutive gene from each tissue analyzed, the Genorm (Version 3.5), an Excel-based software package was used [43].
2.11. Sequencing and Bioinformatics Analysis of Fecal 16S rRNA DNA was extracted from feces and purified using the PureLink Microbiome DNA Purification kit (Invitrogen, Carlsbad, CA, USA). The quantification of the purified DNA was done using Quant-it Pico-Green dsDNA Reagents and Kits (Invitrogen, CA, USA). The amplification and sequencing of 16S rRNA amplicons was performed in the CATG from the Institute of Chemistry, University of São Paulo. For the 16S metagenomic sequencing library preparation, PCR was performed using the following primers S-D-Bact-0341-b-S-17.
(341F, 5 -TCGTCGGCAGCGTCAGATGTGTATAAGAGACAGCCTACGGGNGGCWGCG) and S-D-Bact-0785-a-A-2160 433 (785R, 5 -GTCTCGTGGGCTCGGAGATGTGTATAAGAGACAGGACTAC HVGGGTATCTAATC) for amplification of the V3 and V4 variable regions of the 16S rRNA gene. The adapters suggested on the Illumina workflow for 16S Metagenomic Sequencing Library Preparation were also included in the reactions. The DNA polymerase kit used was KAPA HiFi Hotstart Ready Mix (Kapa Biosystems, Wilmington, MA, USA). Two hundred nM of each primer at 95 • C for 3 min followed by 25 cycles at 95 • C for 30 s, 55 • C for 30 s, 72 • C for 30 s and with a final cycle of 72 • C for 5 min. Verification of amplicon size (expected~550 bp) was performed using the Bioanalyzer DNA 1000 chip (Agilent Technologies, Santa Clara, CA, USA). Removal of PCR contaminants was performed by AMPure XP beads (Beckman Coulter, Inc., Indianapolis, IN, USA). The dual index was attached using Nextera XT Index Kit and after a second round of PCR cleanup was performed with AMPure XP beads. V3-V4 16S indexed amplicon libraries were validated using the Bioanalyzer High Sensitivity DNA chip thorough verification of the expected size of around 630 bp. Analysis of obtained data was performed in several steps. Raw reads were filtered using Prinseq lite v. 0.20.4 [44] with the removal of sequences that had an average quality score lower than 20 (Q20). To remove the primers and adapters, the software used was cutadapt v. 1.14 [45]. After preprocessing (filtering and adapter removal), the sequences were clustered into Operational Taxonomic Units (OTUs) considering a similarity of 97% shared between reads belonging to the same OTU. The UPARSE method [46] included in the USEARCH software (version 8.1.1812) [47] was used for this clustering step, according to a pipeline based on the clustering-first approach from the Brazilian Microbiome Project (BMP) [48]. We plotted rarefaction curves using QIIME (version 1.9.1) [49]. Taxonomy was assigned for each OTU by the RDP classifier [50]. For these analyses, we used Welch's t-test and Benjamin-Hochberg false discovery rate (FDR) correction from STAMP [51]. These analyses were performed at the Bioinformatics Laboratory at the Institute of Chemistry, University of São Paulo. Data from the bacteria DNA sequencing is publicly available at BioProject NCBI (PRJNA641570).

Measurement of Hormones in Serum Samples
The hormones gastric inhibitory polypeptide (GIP), glucagon-like peptide 1 (GLP-1), leptin, pancreatic polypeptide (PP), peptide YY (PYY) and insulin were measured in serum samples using the mouse gut hormone magnetic bead panel kit from Millipore (MGTMAG-78K, Billerica, MA, USA) according to the instruction manual. The reader used was the Luminex MAGPIX (Madison, WI, USA) and the software used to analyze the data was the Millipore MILLIPLEX Analyst 5.1.

Statistical Analysis
Results are presented as the mean ± standard deviation (SD). Comparisons between experimental groups were performed using Student's t-test or Mann-Whitney test, depending on the sample distribution. For multiple group tests, we performed one-way ANOVA and Tukey post-test analysis for multiple comparisons or two-way ANOVA and Bonferroni's post-test. p < 0.05 was considered statistically significant. Data presented were obtained from at least two independent experiments.

Tb Reverses the Biochemical and Metabolic Patterns Associated with Obesity
After eight weeks on the HFD, mice presented biochemical and metabolic alterations that characterize the state of obesity compared to mice given a control diet (CD, Figure S1A-E). After eight weeks of HFD feeding, obese mice were treated with tributyrin (Tb) or water for six weeks. We have previously demonstrated that Tb treatment has no effect on body weight or glucose metabolism in CD-fed mice [39]. In contrast, Tb-treated HFD-fed mice gained less body weight than mice treated with water after the end of the experimental protocol ( Figure 1A,B). A reduction in subcutaneous WAT ( Figure 1B) was observed in Tb-treated animals compared to the nontreated HFD group. These findings may be partly explained by the reduction in energy efficiency in the Tb-treated mice, even though there were no changes in food or caloric intake between the groups ( Figure 1C,D). A decrease in the respiratory coefficient was also found in Tb-treated mice ( Figure 1E). No significant difference in energy expenditure was observed between the experimental groups ( Figure 1F). Treatment with Tb improved fasting glucose (Figure 2A), glucose tolerance ( Figure 2B,C) and the response to insulin administration ( Figure 2D,E). Additionally, a reduction in insulin concentration and improvement of insulin resistance, analyzed by HOMA-IR, were observed in Tbtreated mice ( Figure 2F,G). Tb did not affect muscle metabolism, since soleus skeletal muscle metabolism after in situ stimulation with insulin was not different between the experimental groups (HFD vs. HFD+Tb, data not shown). This result is different from what we previously observed when we tested Tb for the prevention of obesity [39]. (E) Respiratory exchange ratio obtained from the ratio of VCO 2 /VO 2 (n = 2 mice per group). (F) Energy expenditure of individual mice (n = 2, 3 mice per group). Results were obtained from high-fat diet (HFD)-fed mice for 14 weeks and treated with Tb or placebo during the last 6 weeks. Data are presented as mean ± SD. **** p < 0.0001, *** p < 0.001, ** p < 0.01, * p < 0.05 HFD vs. HFD+Tb, Student's t-test.
Treatment with Tb improved fasting glucose (Figure 2A), glucose tolerance ( Figure 2B,C) and the response to insulin administration ( Figure 2D,E). Additionally, a reduction in insulin concentration and improvement of insulin resistance, analyzed by HOMA-IR, were observed in Tb-treated mice ( Figure 2F,G). Tb did not affect muscle metabolism, since soleus skeletal muscle metabolism after in situ stimulation with insulin was not different between the experimental groups (HFD vs. HFD+Tb, data not shown). This result is different from what we previously observed when we tested Tb for the prevention of obesity [39].
Treatment with Tb significantly reduced serum concentrations of NEFA, triacylglycerol (TAG) and alanine aminotransferase (ALT) compared to the placebo group, but it had no effect on other lipid parameters, such as HDL, LDL, or total cholesterol (Table 1). We observed a reduction in liver weight ( Figure 1B), ALT levels in the circulation (Table 1) and a decrease in TAG content and fat accumulation in the liver ( Figure 3A,B). This indicated an improvement of hepatic function in Tb-treated mice compared to the control animals.  Treatment with Tb significantly reduced serum concentrations of NEFA, triacylglycerol (TAG) and alanine aminotransferase (ALT) compared to the placebo group, but it had no effect on other lipid parameters, such as HDL, LDL, or total cholesterol (Table 1). We observed a reduction in liver weight ( Figure 1B), ALT levels in the circulation (Table 1) and a decrease in TAG content and fat accumulation in the liver ( Figure 3A,B). This indicated an improvement of hepatic function in Tbtreated mice compared to the control animals. Data are presented as mean ± SD. * p < 0.05 HFD vs. HFD+Tb, Student's t-test. Next, we evaluated the expression of proinflammatory genes and macrophage markers in liver and WAT. In the liver, no significant difference was observed among the experimental groups, despite a general trend toward a reduction in inflammatory genes ( Figure 4B). In contrast, the treatment of obese mice with Tb reduced the expression of inflammatory markers, including Il-1β and Mcp-1 and of M1 macrophages (Cd11c and F4/80) in the WAT ( Figure 4A). These results indicate Next, we evaluated the expression of proinflammatory genes and macrophage markers in liver and WAT. In the liver, no significant difference was observed among the experimental groups, despite a general trend toward a reduction in inflammatory genes ( Figure 4B). In contrast, the treatment of obese mice with Tb reduced the expression of inflammatory markers, including Il-1β and Mcp-1 and of M1 macrophages (Cd11c and F4/80) in the WAT ( Figure 4A). These results indicate that Tb may reduce inflammation in the WAT of obese animals. Adiponectin and sterol regulatory-element binding (Srebp) were also analyzed in WAT and liver samples, respectively, but no difference was observed. Figure 3. Tb treatment reduced the accumulation of triglycerides (TAG) in the liver and attenuated the signals of hepatic steatosis. (A) Triglycerides (TAG) were measured in liver samples from HFDfed mice, and treated (HFD+Tb) or not (HFD) with Tb. Data are presented as mean ± SD (n = 12-17 mice per group). * p < 0.05 HFD vs. HFD+Tb, Student's t-test. (B) Representative image of the liver histology (three mice from each group were analyzed). V, lobular central vein. Scale bars represent 100 μm.
Next, we evaluated the expression of proinflammatory genes and macrophage markers in liver and WAT. In the liver, no significant difference was observed among the experimental groups, despite a general trend toward a reduction in inflammatory genes ( Figure 4B). In contrast, the treatment of obese mice with Tb reduced the expression of inflammatory markers, including Il-1β and Mcp-1 and of M1 macrophages (Cd11c and F4/80) in the WAT ( Figure 4A). These results indicate that Tb may reduce inflammation in the WAT of obese animals. Adiponectin and sterol regulatoryelement binding (Srebp) were also analyzed in WAT and liver samples, respectively, but no difference was observed.

Tributyrin Increases Serum Concentrations of Butyrate Independently of the Gut Microbiota
Tb administration increased the serum concentration of butyrate nearly three-fold compared to HFD-induced obese mice (21.4 ± 15.1 vs. 7.6 ± 3.3 μg/mL, Figure 5A). Next, we determined whether Tb could change the microbiota composition in HFD-fed mice. Using 16S metagenomic sequencing,

Tributyrin Increases Serum Concentrations of Butyrate Independently of the Gut Microbiota
Tb administration increased the serum concentration of butyrate nearly three-fold compared to HFD-induced obese mice (21.4 ± 15.1 vs. 7.6 ± 3.3 µg/mL, Figure 5A). Next, we determined whether Tb could change the microbiota composition in HFD-fed mice. Using 16S metagenomic sequencing, we profiled the shifts in microbiome composition. At the phylum level, we observed that Tb treatment did not significantly alter community composition ( Figure 5B). At the genus level, we only observed a significant difference between the experimental groups for two components of the microbiota (Johnsonella and Turicibacter, p < 0.05; 95% of confidence intervals, n = 4) of the 78 genera identified (Table S2). we profiled the shifts in microbiome composition. At the phylum level, we observed that Tb treatment did not significantly alter community composition ( Figure 5B). At the genus level, we only observed a significant difference between the experimental groups for two components of the microbiota (Johnsonella and Turicibacter, p < 0.05; 95% of confidence intervals, n = 4) of the 78 genera identified (Table S2).

Tb-GPR109 Signaling Regulates Glucose Metabolism and Adipose Tissue Inflammation in Obese Mice
Given Tb, as a prodrug of butyrate, prevents the development of metabolic and inflammatory alterations in HFD-fed mice [39]. Next, we wanted to determine whether the beneficial effects of Tb on glucose metabolism were mediated via "metabolite-sensing" G-protein coupled receptors. For that, we used mice deficient on GPR109A (Gpr109a −/− mice), a receptor that binds butyrate and

Tb-GPR109 Signaling Regulates Glucose Metabolism and Adipose Tissue Inflammation in Obese Mice
Given Tb, as a prodrug of butyrate, prevents the development of metabolic and inflammatory alterations in HFD-fed mice [39]. Next, we wanted to determine whether the beneficial effects of Tb on glucose metabolism were mediated via "metabolite-sensing" G-protein coupled receptors. For that, we used mice deficient on GPR109A (Gpr109a −/− mice), a receptor that binds butyrate and mediates part of its effects on host tissues [11,12]. In contrast to the response observed in Tb-treated C57BL/6 mice, Gpr109a −/− mice did not show any significant change in glucose parameters after the administration of Tb ( Figure 6A-C), indicating the participation of this receptor in the effect of Tb in obese mice. Then we asked whether the beneficial effects of Tb on glucose homeostasis were dependent on the intense model of diet-induced obesity used. We repeated the same analysis using HFD with less energy from fat (43% vs. 58%), which also is more relevant to humans [52]. In this experiment, we used Gpr109a −/− and Gpr43 −/− mice +/− treatment with Tb. It is worth mentioning that GPR43 is a receptor activated by other SCFAs such as acetate and propionate [11,12]. Regardless of the HFD used, Gpr109a −/− mice did not show any significant change in glucose parameters after the administration of Tb ( Figure 6D,F). In contrast, Gpr43 −/− mice gained less body weight than placebo-treated mice, and showed a similar pattern of response to Tb treatment of the WT C57BL/6 mice ( Figure S2). Regarding the biochemical parameters, we found that Gpr109a −/− HFD-fed mice (43% of energy from fat) presented an elevation of triglycerides (TAG) concentrations in serum compared with the WT mice (Table 2). This increase was attenuated by Tb treatment. We also observed a reduction in the NEFA concentrations caused by Tb treatment in WT and Gpr109a −/− mice ( Table 2). We measured the serum concentrations of hormones that are important for glucose homeostasis (glucagon-like peptide (GLP), gastric inhibitory peptide (GIP), leptin, insulin, peptide YY (PYY) and pancreatic polypeptide (PP)) in these mice. We found only a minor effect of Tb on PYY, which was independent of the genotype ( Figure S3). Table 2. Biochemical parameters measured in the serum from 6 h fasted WT and Gpr109a −/− mice (F) GTT curves of Gpr109a −/− mice are presented (n = 9 mice per group). For this experiment, GTT was performed with the use of 1 g/kg glucose and a blood glucose meter (Accu-Check Performa ® ). Data are presented as mean ± SD (n is presented for each result). * p < 0.05 HFD vs. HFD+Tb. Student's t-test.
Regarding the biochemical parameters, we found that Gpr109a −/− HFD-fed mice (43% of energy from fat) presented an elevation of triglycerides (TAG) concentrations in serum compared with the WT mice ( Table 2). This increase was attenuated by Tb treatment. We also observed a reduction in the NEFA concentrations caused by Tb treatment in WT and Gpr109a −/− mice ( Table 2). We measured the serum concentrations of hormones that are important for glucose homeostasis (glucagon-like peptide (GLP), gastric inhibitory peptide (GIP), leptin, insulin, peptide YY (PYY) and pancreatic polypeptide Cells 2020, 9, 2007 11 of 18 (PP)) in these mice. We found only a minor effect of Tb on PYY, which was independent of the genotype ( Figure S3). Data are presented as mean ± SD. *** p < 0.001, * p < 0.05 comparisons between HFD and HFD+Tb of the same genotype. Results analyzed using Student's t-test.
To elucidate the mechanisms behind GPR109A activation and the improvement in metabolic parameters, we analyzed the pattern of expression of inflammatory markers and infiltrating leukocytes in the WAT of WT and GPR109A knockout mice, treated or not with Tb. Similar to what we observed in Figure 4A, Tb treatment attenuated Il1β, an indication of reduced adipose tissue inflammation ( Figure 7A). In contrast, Gpr109a −/− mice treated with Tb showed no significant effect on inflammatory mediators such as Tnf-α, Il1β, Il6 and Il10 ( Figure 7B). We observed a significant reduction in the expression of Cd11c in the WAT of Gpr109a −/− obese mice. Interestingly, Gpr109a −/− mice showed a significant increase in the frequency of M1 (F4/80 + CD11b + CD11c + ) macrophages ( Figure 7C) and a significant decrease in the frequency of M2 macrophages (F4/80 + CD11b + CD206 + ) in WAT ( Figure 7D), thus indicating a shift toward a proinflammatory phenotype. Tb treatment did not affect the frequency of M1 macrophages in WT or Gpr109a −/− mice ( Figure 7C,D and Figure S4A). However, we observed an increase of M2 and regulatory T cells in this tissue after Tb treatment in WT ( Figure 7D,E and Figure S4A,B). This latter effect was consistently observed in mice on both HFD diets (43% and 58%, not shown), but absent in Gpr109a −/− HFD-fed mice (independently of the HFD used, Figure 7E). In addition, Tb effects on glucose tolerance ( Figure 6D), NEFA concentrations (Table 2) and adipose tissue inflammation (Figure 7) observed in HFD-fed mice with 43% of energy from fat were present in the absence of a significant effect of Tb on body weight. This indicates that, at least for these parameters, Tb protective effects are independent of changes on body weight.
However, we observed an increase of M2 and regulatory T cells in this tissue after Tb treatment in WT ( Figure 7D,E and Figure S4A,B). This latter effect was consistently observed in mice on both HFD diets (43% and 58%, not shown), but absent in Gpr109a −/− HFD-fed mice (independently of the HFD used, Figure 7E). In addition, Tb effects on glucose tolerance ( Figure 6D), NEFA concentrations ( Table  2) and adipose tissue inflammation (Figure 7) observed in HFD-fed mice with 43% of energy from fat were present in the absence of a significant effect of Tb on body weight. This indicates that, at least for these parameters, Tb protective effects are independent of changes on body weight.

Discussion
In the present study, we demonstrated that Tb, a lipid that increases circulating levels of butyrate, attenuated body weight gain and the impairment of glucose metabolism of obese mice. Tb treatment was associated with the improvement of liver function and attenuation of eWAT inflammation. Mechanistically, we demonstrated that the effect on glucose metabolism and WAT inflammatory state was dependent on GPR109A activation.
Previous studies reported an increased Firmicutes/Bacteroidetes (F/B) ratio in gut microbial communities of genetically (ob/ob mice) or diet-induced obese mice [6,53]. This and other changes in microbiota composition and the consequent alteration in microbiota signaling to host cells are associated with the development of obesity and insulin resistance (reviewed by Khan et al., 2014) [54].
In our study, we observed that oral Tb was an effective way of increasing systemic concentrations of butyrate and improving glucose metabolism of obese mice. These effects were observed in the absence of major changes in microbiota composition, which is a key aspect to be considered for treatment of conditions associated with a dysbiotic microbiota, in which modifications of gut microbiota composition are not possible or which require longer periods of treatment, as appears to be the case for obese individuals [55,56].
A previous study has found that the addition of butyrate to an HFD (5% wt/wt) increased mice energy expenditure and oxygen consumption and decreased the RER in mice, suggesting amelioration in fatty acid oxidation [28]. These results were associated with an improvement of mitochondrial biogenesis and function in skeletal muscle and brown adipose tissue in butyrate-treated mice [28]. Consistent with this finding, we observed that Tb treatment decreased the energy efficiency and RER of obese mice, indicating that this treatment prevented further accumulation of lipids in tissues through an increase of energy expenditure and fatty acid oxidation, effects that mimic the changes observed with dietary butyrate supplementation.
Tb administration to obese mice reduced liver weight and the hepatic TAG content. These findings were associated with amelioration of systemic glucose homeostasis (lower fasting glucose, an improved glucose tolerance and higher insulin sensitivity). These protective effects on the liver of obese mice together with the capacity of Tb to attenuate liver steatosis and injury in other experimental models [57][58][59] indicate that this organ is a key target of Tb. Indeed, after Tb administration, high concentrations of butyrate have been observed in the portal vein and liver [58]. This SCFA can then act directly on hepatocytes, as demonstrated by a recent study [58], inhibiting histone deacetylases (HDACs) and, consequently, affecting the expression of important genes for hepatic function, such as carnitine palmitoyltransferase-1 (CPT-1A). The hepatic beneficial effects of tributyrin/butyrate can also be secondary to its action on other tissues such as the intestine [57] or WAT.
A biochemical parameter that was consistently modified after the treatment of obese mice with Tb was the concentrations of circulating NEFA. A previous study suggested that NEFA concentrations in the circulation are more directly associated with liver injury than other parameters such as TAG deposition or TAG concentration in the circulation [60]. Other studies have also demonstrated a reduction in NEFA in mice treated with butyrate or Tb, associating this effect with inhibition of adipocyte lipolysis [39,61,62]. In our study, we found that, regardless of the genetics of mice (C57BL/6, Gpr43 −/− or Gpr109a −/− ), Tb decreased NEFA in serum, supporting the premise that this effect was not related to the activation of GPR43 or GPR109A, as we expected. We hypothesize that the reduction of NEFA, and some of the effects on glucose metabolism are associated with the increased oxidation of fatty acids by the tissues induced by Tb, which is attributable to an improvement of mitochondrial function and efficiency, as previously demonstrated for butyrate supplementation [28,63].
GPR109A is expressed in adipocytes, immune cells [64] and intestinal epithelial cells and it is selectively activated by butyrate [65]. We observed that key metabolic and immune effects of Tb on obese mice were lost in the absence of this receptor, indicating that GPR109A mediates the Tb effects. In accordance with this, previous studies reported beneficial effects of other ligands of GPR109A on glycemic parameters. A clinical trial performed with GSK256073, a synthetic agonist of GPR109A, in T2DM patients found an improvement in glucose homeostasis and insulin sensitivity after treatment with this drug [66]. In addition, a recent study described that Gpr109a −/− mice are more susceptible to the development of HFD-induced obesity compared to their controls and that administration of niacin (another ligand of GPR109a) attenuated the development of obesity in mice through a GPR109A-dependent mechanism [67].
Interestingly, we found that Gpr109a −/− mice present more M1 macrophages and fewer M2 macrophages in the eWAT, indicating that their adipose tissue is in a more proinflammatory state compared to the WT mice. Tb treatment increased numbers of Tregs cells in WAT of WT mice, but not in Gpr109a −/− mice suggesting that these cells may be relevant for the attenuation of inflammation observed after Tb treatment and also for the systemic beneficial effects of Tb (i.e., improvement of glucose metabolism). Previous studies have shown that mice with expanded Tregs numbers in the visceral WAT were protected against obesity-associated inflammation, insulin resistance and metabolic alterations, whereas the deletion of Tregs, especially from adipose tissue, abolished these effects [68,69]. Butyrate administration increased Tregs numbers in the colon of mice, an effect associated with HDACs inhibition [70,71]. In addition, Singh and collaborators showed that GPR109A signaling induces differentiation of cells into Treg and IL-10-producing T cells [65]. In this latter study, the authors focused on the cell populations present in the colon lamina propria and did not analyze the relevance of the receptor for Treg numbers in other tissues such as the adipose tissue. We found that butyrate increases Treg populations in the adipose tissue of obese mice via GPR109A signaling. This effect may contribute to the reduction of local inflammation and improvement of glucose homeostasis. Other mechanisms activated by Tb-GPR109A signaling may also contribute to the phenotype observed including the increased activity of brown adipose tissue and/or browning of WAT as previously reported [67].

Concluding Remarks and Perspectives
Tb is an alternative therapeutic approach to counteract the lack of microbial production of butyrate and restore immune and metabolic balance independently of the gut microbiota on obese individuals. The comprehension of Tb effects and mechanisms is a fundamental step toward the development of new therapeutic strategies for obesity treatment. It is worth mentioning that Tb is in use as a food additive and it has been shown to be well tolerated in humans [72]. These characteristics are important to evaluate the plausibility of proceeding to tests in other animal models and in the evaluation of safety in a potential clinical trial. However, this study has some limitations including the fact that we did not test lower doses of tributyrin. This is a critical point that will need to be addressed before tests in humans.

Supplementary Materials:
The following are available online at http://www.mdpi.com/2073-4409/9/9/2007/s1, Figure S1: Mice on a HFD for 8-weeks presented a significant increase in body weight and an impairment in glucose metabolism compared to mice on control diet (CD, AIN93M), Figure S2: Tributyrin improved glucose homeostasis in obese mice through a GPR43 independent mechanism, Figure S3: Quantification of hormones in serum samples from WT, Gpr43 −/− and Gpr109a −/− mice, Figure S4: Representative flow cytometry (FACS) plots of macrophages populations and Treg cells analysis performed with the WAT of WT and Gpr109a −/− mice treated or not with Tb. Table S1: Sequence of the primers used in the study, Table S2: Proportion of genus (according to OTU-classification) levels in the microbiota of HFD-fed mice +/− Tb. Funding: This study was supported by research grant from Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP; Grants 12/10653-9; 18/15313-8). The study was also financed by the National Council for Scientific and Technological Development (CNPq, Projeto Universal #14/2011) and Coordenação de Aperfeiçoamento de Pessoal de Nível Superior, Brasil (CAPES), Finance Code 001. F.T.S was recipient of fellowship from FAPESP (2012/15774-9). Experiments in Australia were supported by the Immunology and Diabetes Laboratory at Monash University.

Conflicts of Interest:
The authors declare that they have no conflict of interest.