Adult male C57BL/6J mice were group-housed (n = 5 per cage) and maintained under controlled temperature (21 ± 2 °C) and humidity (60 ± 10%) with free access to water and feed in a ventilated rack containing cages for the mice (NexGen mouse 500, Allentown, PA, United States). The environment comprised a 12/12 h light–dark period and air renewal cycles (15 min/h). The procedures followed the recommendations of the National Institutes of Health Guide for the Care and Use of Laboratory Animals (NIH publication No. 85-23, revised 1996) and were approved by our local Ethics Committee (Institute of Biology, CEUA number 041/2018).

Fifty adult male C57BL/6J mice (3 month old) from the Central Biotery of the Federal Minas Gerais University were used in this study. Initially, the animals were randomly assigned to two nutritionally different groups: (1) Control group (C) – animals that received a control diet (14% of energy as protein, 10% as fat, and 76% as carbohydrates; total energy 15 KJ/g, n = 10); and (2) High-fat group (HF) – animals that received a high-fat diet (14% of energy as protein, 50% as fat and 36% as carbohydrates; total energy 21 KJ/g, n = 40).

After twelve weeks, the C group and ten animals from the HF group continued the same food scheme for an additional five weeks, whereas the remaining animals from the HF group were randomly subdivided according to the treatment into the following groups: (1) HFA received the PPAR-alpha agonist (WY14643, Sigma–Aldrich, 3.5 mg/kg body mass) incorporated into the HF diet (n = 10) for five weeks; (2) HFL received the DPP-4 inhibitor (linagliptin, Boehringer Ingelheim, 15 mg/kg body mass) incorporated into the HF diet (n = 10) for five weeks; and (3) HFC received the combination of a PPAR-alpha agonist with a DPP-4 inhibitor (at the same doses used in the groups receiving monotherapy) incorporated into the HF diet (n = 10) for five weeks.

The entire experimental protocol lasted for 17 wk (12 wk of obesity induction + 5 wk of treatment). The doses of WY-14643 and linagliptin were based on previous experiments conducted by our group[4,9]. PragSoluções (Jaú, São Paulo, Brazil) produced the experimental diets according to the recommendations of the American Institute of Nutrition (AIN 93M)[15]. All groups were treated following the order in which the groups were described.

Food intake was measured daily by subtracting the remainder of the diet verified on the following day from the amount of diet offered on the previous day. Energy intake comprised the product of the food consumption and the energy contained in 1 g of each diet (in kJ). Animal body masses were assessed on a digital scale once a week (BL-3200H, precision 0.01 g).

One week before sacrifice, the animals were subjected to the oral glucose tolerance test (OGTT). Under a 6-h fast (time 0) and after 15, 30, 60, and 120 minutes of the orogastric gavage of a glucose solution (2 g/kg body mass), blood samples were obtained from the caudal vein. A manual glucometer (Accu-Chek, Roche, São Paulo, SP, Brazil) was used to measure the blood glucose levels at different times. The area under the curve (AUC) indicated the oral glucose tolerance (GraphPad Prism, version 8.3 for Windows, GraphPad Software, La Jolla, CA, United States).

Mice were fasted for 6 h. Under intraperitoneal anesthesia with ketamine (240 mg/kg) and xylazine (30 mg/kg), blood samples obtained by cardiac puncture were separated by centrifugation (712 ×g) to obtain plasma samples to perform biochemical analyses. The liver, large intestine (cecum), and small intestine (jejunum and ileum) were carefully dissected, weighed, and analyzed following the protocols for different techniques.

An ELISA was performed to measure plasma GLP1 (multispecies GLP1 ELISA Kit Cat. #EZGLP1T-36K, Millipore, Missouri, United States), and LPS (multispecies LPS ELISA Kit Cat. #SEB526Ge-96T, Cloud-Clone Corp., Katy, United States). A semiautomatic spectrophotometer and a commercial kit (K117, Bioclin, Quibasa, Belo Horizonte, MG, Brazil) were used to measure hepatic triacylglycerol (TAG) as previously described[16].

Liver and cecum fragments fixed in Millonig-buffered formalin (pH 7.2-7.4) were subsequently dehydrated, diaphanized, embedded in Paraplast Plus (Sigma–Aldrich, St. Louis, MO, United States) and sectioned (5 μm thick) with a microtome. Slides stained with hematoxylin and eosin (liver) or Alcian Blue (Sigma Chemical Company-pH 2.5) plus Periodic Acid-Schiff (PAS, intestine-Sigma Chemical Company) were photographed using a Leica DMRBE microscope (Wetzlar, Germany) and an Infinity Lumenera digital camera (Ottawa, ON, Canada). The images were analyzed in a blinded manner with STEPanizer (www.stepanizer.com) as described below:

Hepatic stereology: Five animals per group and ten images per animal were analyzed. The volume density of liver steatosis [Vv (liver, st)] was estimated by the point-counting technique, following the formula: Vv (liver, st) = Pp (liver, st)/PT (Pp is the number of points that reached fat droplets, and PT is the total test points). The images were analyzed with STEPanizer using a 36-point test system[17].

Gut stereology: The number of goblet cells per area [Q_A (goblet)] was estimated using STEPanizer. All goblet cells within the test area were counted, except those touching the forbidden lines. The result was divided by the test area measured in mm².

The feces found in the mouse cecum were used to extract microbial DNA using the QIAamp Fast DNA Stool Mini Kit (Qiagen, Düsseldorf, Germany) according to the manufacturer's instructions. The DNA quantity, purity, and concentration were determined using Qubit (Life Technologies, Carlsbad, California, United States) and horizontal electrophoresis (1% agarose gel). Real-time quantitative polymerase chain reaction (PCR) assays were used for the relative quantification of specific phyla of microorganisms (Bacteroidetes, Firmicutes, Actinobacteria, Proteobacteria) in fecal microbiota from the mouse guts by detecting 16S rRNA genes. For relative quantification, the abundances of different phyla were normalized by the ^ΔΔCt method relative to the total bacterial amount in the samples[18]. The indicators used are described in Table 1.

Total RNA was extracted from 50 mg of the liver and 70 mg of the small intestine (jejunum and ileum) using TRIzol reagent (Invitrogen, CA, United States). Afterward, the addition of 200 μL of chloroform was followed by centrifugation (1200 g for 10 min at 4 °C for liver samples and 12000 g for 15 min at 4 °C for intestine samples). The RNA extract portion was separated, and 500 μL of isopropanol was added and allowed to stand for 10 min (liver samples) or 15 min (intestine samples) to precipitate the RNA. Then, the samples were centrifuged (1200 g for 10 min at 4 °C for liver samples and 12000 g for 10 min at 4 °C for intestine samples). The isopropanol was removed, and the formed pellet was resuspended in 500 μL of 75% ethanol for liver samples or 70% ethanol (ice-cold) for intestine samples and then centrifuged (1200 g for 5 min at 4 °C for liver samples and 10000 g for 5 min at 4 °C for intestine samples).

The ethanol was removed, and the pellet was resuspended in 20 μL (liver)/50 μL (intestine) of deionized water (Milli-Q). The samples were subjected to a dry bath (50 °C for 5 min) and quantified by using Nanovue equipment (GE Life Sciences). For RNA transcription into complementary DNA (cDNA), 1.0 μg RNA was treated with DNAse I (Invitrogen, CA, United States). First-strand cDNA synthesis was performed using Oligo (dT) primers for reverse transcriptase mRNA and Superscript III (both from Invitrogen). qPCR was performed using a CFX96 recycler (Bio–Rad, Hercules, CA, United States) and SYBR Green mix (Invitrogen, Carlsbad, CA, United States). Beta-actin was used to correct the expression of the target genes in liver samples, and Gapdh was used for intestine samples. The primer sequences used are shown in Table 2 (liver) and Table 3 (intestine). All gene symbols are italicized (the first letter capitalized), and protein symbols are italicized in uppercase[19].

Sample size calculation considered that, in metabolic and molecular biology analyses, if a factor increases or decreases in five replicates, the probability of occurrence is P = (1/2)⁵ = 0.05. Therefore, a minimum of 5 replicates was adopted for the analyses[20]. The data are shown as the mean and SD. During the first 12 wk, statistical analysis comprised Student’s t test and Welch's correction. At the treatment phase, data were analyzed using Brown–Forsythe and Welch one-way ANOVA with the Dunnett T3 post hoc test[21]. A P < 0.05 was considered significant (GraphPad Prism version 8.3 for Windows, GraphPad Software, La Jolla, CA, United States).

The animals in the C and HF groups had equal body mass (BM) at baseline. All animals tolerated the diets and treatments well. The protocol was maintained as previously stated. In the 13^th week, the HF group was overweight compared to the C group (+30%), and this characteristic persisted until the end of the experiment. Figure 1A depicts these results. Although food intake did not differ between the groups (Figure 1B), the energy intake in the HF-fed groups was higher than that in the C group (Figure 1C).

Open in New Tab   Full Size Figure   Download Figure

Figure 1 Body mass and food behavior. A: Body mass evolution; B: Food intake; C: Energy intake. Brown-Forsythe and Welch one-way ANOVA and Dunnett T3 post hoc test (mean ± SD, n = 6). Significant differences are indicated as follows: ^aP < 0.05; ^bP < 0.01; ^dP < 0.0001. C: Control diet; HF: High-fat diet; HFA: High-fat diet plus PPAR-alpha agonist (WY14643); HFL: High-fat diet plus DPP-4 inhibitor (linagliptin); HFC: High-fat diet plus the combination of WY14643 with linagliptin.

Figure 2A shows the OGTT curve, in which the HF group showed a significant increase in fasting glucose (T0) compared to the C group. This difference remained until the end of the test (T120). The C group and the treated groups exhibited rescued baseline blood glucose levels at the other evaluation times (T30, T60, and T120). On the other hand, the HF group did not reach baseline glucose levels, indicating a delay after glucose overload and implying oral glucose intolerance, as confirmed by the higher AUC for OGTT in the HF group than in the C group (+23%, Figure 2B). In contrast, all treated groups showed lower AUCs than the HF group, indicating oral glucose intolerance alleviation.

Open in New Tab   Full Size Figure   Download Figure

Figure 2 Carbohydrate metabolism. A: Oral glucose tolerance test; B: Area under the curve for oral glucose tolerance test; C: Plasma glucagon-like peptide 1 concentrations. Brown-Forsythe and Welch one-way ANOVA and Dunnett T3 post hoc test (mean ± SD, n = 6). Significant differences are indicated as follows: ^aP < 0.05; ^cP < 0.001; ^dP < 0.0001. C: Control diet; HF: High-fat diet; HFA: High-fat diet plus PPAR-alpha agonist (WY14643); HFL: High-fat diet plus DPP-4 inhibitor (linagliptin); HFC: High-fat diet plus the combination of WY14643 with linagliptin; AUC: Area under the curve; OGTT: Oral glucose tolerance test; GLP1: Glucagon-like peptide 1.

Plasma GLP1 concentrations diminished in the HF group compared to the C group (Figure 2C). As expected, treatment with linagliptin enhanced GLP1 concentrations in the HFL group (+13%), which also occurred in the HFC group and after the single treatment with the PPAR-alpha agonist in the HFA group compared to the HF group (Figure 2C).

The amplified 16S rRNA genes of cecal gut bacteria were measured at the end of the experiment to evaluate the microbiota composition. The HF group showed a Firmicutes phylum increase coupled with decreases in the Proteobacteria and Bacteroidetes phyla compared to the C group, as shown in Figure 3A. However, all treatments reversed these phylogenetic alterations in the treated groups. The treatments restored the amount of Bacteroidetes to resemble that in the C group and caused a significant decrease in Proteobacteria, which can play a decisive role in the beneficial effects observed due to the proposed treatments. Changes in the microbiota composition in the HF group triggered increased LPS concentrations (+10%, Figure 3B), while the HFL and HFC groups showed significantly reduced plasma LPS concentrations (-11% for HFL vs HF and -12% for HFC vs HF).

Open in New Tab   Full Size Figure   Download Figure

Figure 3 Gut-liver axis. A: Phylogenetic microbiota composition; B: Plasma lipopolysaccharide concentrations; C: Hepatic Lbp gene expression; D: Hepatic Tlr4 gene expression. Brown-Forsythe and Welch one-way ANOVA and Dunnett T3 post hoc test (mean ± SD, n = 6). Significant differences are indicated as follows: ^aP < 0.05; ^bP < 0.01; ^cP < 0.001; ^dP < 0.0001. C: Control diet; HF: High-fat diet; HFA: High-fat diet plus PPAR-alpha agonist (WY14643); HFL: High-fat diet plus DPP-4 inhibitor (linagliptin); HFC: High-fat diet plus the combination of WY14643 with linagliptin; LPS: Lipopolysaccharide.

The hepatic mRNA expression of both Lbp (Figure 3C) and Tlr4 (Figure 3D) genes increased in the HF group compared to the C group, while the treated groups showed significantly reduced expression.

The HF group showed an 80% decrease in intestinal Mucin2 gene expression compared with the C group. Conversely, the HFA, HFL, and HFC groups showed significant increases in Mucin2 expression (+439% for HFA vs HF, +345% for HFL vs HF, and +670% for HFC vs HF; Figure 4A).

Open in New Tab   Full Size Figure   Download Figure

Figure 4 Gut gene expression, histology, and stereology. A: Gut Mucin2; B: Zo-1; C: Occludin gene expression; D: Large intestine stained with Alcian Blue and Periodic Acid-Schiff (PAS); E: Numerical density of goblet cells per area. Alcian blue and PAS stain glycoproteins produced by goblet cells (scale bar = 40 μm), which were markedly reduced by the high-fat diet and rescued by the treatments according to stereology (Q_A goblet). Brown-Forsythe and Welch one-way ANOVA and Dunnett T3 post hoc test (mean ± SD, n = 6 for real-time reverse transcriptase–polymerase chain reaction and n = 5 for stereology). Significant differences are indicated as follows: ^aP < 0.05; ^bP < 0.01; ^cP < 0.001; ^dP < 0.0001. C: Control diet; HF: High-fat diet; HFA: High-fat diet plus PPAR-alpha agonist (WY14643); HFL: High-fat diet plus DPP-4 inhibitor (linagliptin); HFC: High-fat diet plus the combination of WY14643 with linagliptin; PAS: Periodic Acid-Schiff; Q_A: Numerical density per area.

In agreement with the previous result, the HF group also had reduced intestinal Zo-1 (-38%, Figure 4B) and Occludin (-78%, Figure 4C) expression compared to the C group. On the other hand, the HFA group showed a 238% increase in Occludin expression, whereas the HFL and HFC groups had > 500% increases in the expression of this gene (Figure 4C). Regarding Zo-1 gene expression, all treated groups had a significant increase (+166% for HFA vs HF, +397% for HFL vs HF, and +102% for HFC vs HF; Figure 4B).

The high intake of saturated fat altered the histochemical pattern of the intestinal mucosal cells, as revealed by the reaction with Alcian Blue and PAS. Figure 4D shows decreased mucus in the HF group, while the treatments elicited an increase in mucus production in the apical region of the crypts, followed by an increased presence of goblet cells (mucus-producing cells). Gut stereology confirmed these observations with the results of Q_A (goblet), which showed a reduction in the HF group compared to the C group (-44%), while the HFA (+68%), HFL (+47%), and HFC (+56%) groups showed an increase in the number of goblet cells per tissue area (Figure 4E).

HF-fed mice exhibited noticeable microvesicular liver steatosis, while mice in all treated groups showed liver steatosis mitigation, with the liver parenchyma resembling that in the C group (Figure 5A).

Open in New Tab   Full Size Figure   Download Figure

Figure 5 Liver histology, stereology, and biochemistry. A: Hematoxylin-eosin liver sections; B: Volume density (Vv) (liver steatosis); C: Hepatic triacylglycerol. Liver sections show widespread hepatic steatosis after chronic HF diet intake and expressive reduction in all treated groups (scale bar = 50 μm. Both stereology (Vv steatosis) and biochemical analyses (hepatic triacylglycerol) confirm these observations. Brown-Forsythe and Welch one-way ANOVA and Dunnett T3 post hoc test (mean ± SD, n = 6 for biochemistry and n = 5 for stereology). Significant differences are indicated as follows: ^aP < 0.05; ^bP < 0.01; ^cP < 0.001; ^dP < 0.0001. C: Control diet; HF: High-fat diet; HFA: High-fat diet plus PPAR-alpha agonist (WY14643); HFL: High-fat diet plus DPP-4 inhibitor (linagliptin); HFC: High-fat diet plus the combination of WY14643 with linagliptin; Vv: Volume density.

In accordance with these histological findings, Vv (liver, st) in the HF group was higher than that in the C group (+38%, Figure 5B). All treated groups showed drastic reductions in liver steatosis (-72% for HFA vs HF, -50% for HFL vs HF, and -77% for HFC vs HF).

Consistent with the stereological findings, hepatic TAG levels increased in the HF group (+38%). Conversely, the treated groups showed lower hepatic TAG concentrations than that in the HF group (-11% for HFA, -16% for HFL, and -13% for HFC; Figure 5C).

The HF group showed higher expression of Cd206, which is a specific marker of macrophages, than the C group (+103%, Figure 6A). In contrast, the treated groups showed reduced Cd206 expression (-67% for HFA vs HF, -65% for HFL vs HF, and -74% for HFC vs HF).

Open in New Tab   Full Size Figure   Download Figure

Figure 6 Liver gene expression. A: Cd206; B: IL-10; C: Nlrp3; D: IL-13. Brown-Forsythe and Welch one-way ANOVA and Dunnett T3 post hoc test (mean ± SD, n = 5 or 6). Significant differences are indicated as follows: ^aP < 0.05; ^bP < 0.01; ^cP < 0.001; ^dP < 0.0001. C: Control diet; HF: High-fat diet; HFA: High-fat diet plus PPAR-alpha agonist (WY14643); HFL: High-fat diet plus DPP-4 inhibitor (linagliptin); HFC: High-fat diet plus the combination of WY14643 with linagliptin.

In addition to the Cd206 results, the HF diet significantly reduced IL-10 expression compared to the C group (-54%), whereas the combined treatment yielded an 80% decrease in HFC IL-10 expression compared to HF (Figure 6B). Nlrp3 showed similar results to those of the Cd206 gene, with higher expression in the HF group than in the C group (+128%) and reduced expression in the HFA (-72%), HFL (-88%), and HFC (-85%) groups compared to the HF group (Figure 6C).

Regarding the IL-13 cytokine, the HF group exhibited a significant increase compared to the C group (+290%). Only the HFA and HFC treatments were able to reduce the expression of this cytokine in comparison with the HF group (-39% and -67%) (Figure 6D).

Gut microbiota can be modified by the dietary composition and play a role in fatty liver pathogenesis through endotoxemia. Peroxisome proliferator-activated receptor (PPAR)-alpha activation has previously rescued the gut–liver axis with anti-steatotic effects in high-fructose-fed mice, whereas a high-dose dipeptidyl peptidase-4 (DPP-4) inhibitor (linagliptin) inhibited hepatic lipogenesis in high-fat-fed mice. The combination of these drugs could restore the gut–liver axis in obese mice.

Nonalcoholic fatty liver disease (NAFLD) is highly prevalent among obese individuals and can evolve into harmful liver diseases. Currently, there is no treatment directly prescribed to counter NAFLD. Herein, we propose a drug combination (PPAR-alpha agonist plus DPP-4 inhibitor) that can alleviate fatty liver by modulating the gut–liver axis with anti-inflammatory properties.

To evaluate the effects of monotherapy with a PPAR-alpha agonist (WY14643) or a DPP-4 inhibitor (linagliptin) and their combined treatment on the gut–liver axis, highlighting the intestinal barrier, endotoxemia, and inflammatory pathways in the livers of high-fat-fed mice. These preclinical insights can help establish new strategies for the treatment of NAFLD.

Mice were fed a control diet (C, 10% of energy as lipids) or a high-fat diet (HF, 50% of energy as lipids) for 12 wk. Then, the HF group was randomly divided into four groups: HF, HF-A (treated with the PPAR-alpha agonist), HFL (treated with the DPP-4 inhibitor), and HFC (treated with the combination of both drugs). The treatments lasted for five weeks. The gut–liver axis was assessed by histological, biochemical, stereological, and molecular techniques.

The HF diet yielded overweightness and oral glucose intolerance, altered gut microbiota composition, and decreased the numerical density of goblet cells and tight junction gene expression, with increased plasma lipopolysaccharide (LPS) concentrations and increased fatty liver. The combined treatment rescued all of these metabolic alterations by restoring the gut microbiota and intestinal barrier markers, resulting in decreased LPS concentrations and fatty liver through anti-inflammatory signals. Further studies may focus on the Proteobacteria phylum, whose alteration can trigger harmful signaling to the liver.

The combination of PPAR-alpha activation with DPP-4 inhibition yielded marked anti-steatotic effects by modulating the gut–liver axis, with reduced endotoxemia and amelioration of the intestinal barrier histology and gene expression. In turn, the livers of obese mice treated with the drug combination benefited from downregulation of the Tlr4 and Nlrp3 pathways and exhibited a hepatic parenchyma similar to that in the C group. Of note, we used stereology to estimate the numerical density of goblet cells in cecum slides stained with Alcian Blue and Periodic Acid-Schiff. This technique can be used to assess the effects of different interventions on the intestinal barrier and leaky gut in further studies.

This study brings novelty to the treatment of NAFLD, which is a challenge to the scientific community. Our results confirm the importance of the gut–liver axis in the pathogenesis of fatty liver and propose a combined treatment that targets the gut microbiota composition, endotoxemia, and intestinal barrier alterations to alleviate fatty liver through local anti-inflammatory effects in HF-fed mice.