Gut microbiota-mediated generation of saturated fatty acids elicits inflammation in the liver in murine high-fat diet-induced steatohepatitis

SPF C57BL/6J mice were fed a conventional CE-2 diet (CLEA Japan Inc., Tokyo, Japan) until 8 weeks of age. The gut microbiota was normalized by exchanging beddings between cages every 2–3 days for 2 weeks. From 8 weeks of age, the mice were fed the STHD-01 (11% kcal/protein, 72% kcal/fat, and 17% kcal/nitrogen-free extracts; EA Pharma Co. Ltd., Kawasaki, Kanagawa) [12] for 9 weeks. The control group was fed the Standard diet (SD) (AIN-93G) (19% kcal/protein, 12% kcal/fat, and 69% kcal/nitrogen-free extract). In the microbiota-depleted group, the mice fed with the STHD-01 diet were treated with an antibiotic (Abx) cocktail (ceftazidime, Sigma-Aldrich, Tokyo, Japan; C3809-5G plus metronidazole; Sigma-Aldrich M3761-25G, both 1 g/L) from 7 weeks until 17 weeks. The mice were killed at week 17 with isoflurane (Mylan Inc., Nagoya, Japan) [13] and peripheral blood, intestinal tissues, and liver samples were harvested. Histological evaluation was performed in a blind manner by a hepatologist and two pathologists from Keio University Hospital as described previously [14].

The levels of aspartate aminotransferase (AST) and alanine aminotransferase (ALT) were measured using the Spotchem EZ (Sp-4430, Arkrey USA Inc., MN). The levels of triiodothyronine (T3), thyroxin (T4) and monocyte chemoattractant protein (MCP)-1 were measured with an enzyme-linked immunosorbent assay kit (T3 and T4, Alpha Diagnostic Intl. Inc., Antonio, TX. MCP-1, R&D System, Inc., Minneapolis, MN). For the measurement of triglyceride (TG) in the liver tissue, a Folch solution (2:1 chloroform: methanol; Wako Pure Chemical Industries Ltd., Tokyo, Japan; 4 mL) was added to each liver tissue sample (0.1 g), which was then homogenized. After adding and mixing with 0.5% NaCl (1 mL), the mixture was centrifuged at 180 g at 20 °C for 20 min. The lower layer was obtained and vacuum-dried, and then 1 mL of isopropanol (Wako Pure Chemical Industries Ltd.) was added to the precipitate. The levels of TG were measured using the Pureauto S TG-N (Sekisui Medical Co., Ltd., Tokyo, Japan). At week 15, mice were fasted 1 day before measuring the levels of fasting blood glucose. The levels of fasting blood glucose were measured with GT-1640 (Arkrey USA Inc.). The plasma levels of insulin were measured with mouse insulin enzyme-linked immunosorbent assay kit (Shibayagi Co.,Ltd., Gunma, Japan).

At week 17, feces were obtained from the mice and resuspended in a phosphate-buffered saline solution (PBS) (0.1 g/mL). The fecal suspension was crushed using the Bug Crashar (Taitec GM-01, Saitama, Japan) at maximal rotation for 10 min. The sample was incubated on ice for 5 min and centrifuged at 2300 g at 4 °C for 1 min. The supernatant (500 μL) was placed in another tube and vortexed with 100 μL of 10% SDS and 500 μL of phenol/chloroform/isoamylalcohol. Then, the sample was centrifuged at 20,000 g at 20 °C for 3 min. The supernatant was treated with chloroform/isoamylalcohol and then isoamylalcohol alone, and the DNA pellet was resuspended in 100 μL Tris/ethylenediamine tetraacetic acid buffer (TE) (Sigma) and 0.5 μL RNase A (Qiagen, Hilden, Germany). The DNA was further purified using the Template Preparation Kit (Roche, Basel, Switzerland).

The obtained DNA was analyzed by a terminal restriction fragment length polymorphism analysis (TechnoSuruga Laboratory Co., Ltd., Shizuoka, Japan). The DNA was amplified with fluorescence-labeled primers, and the amplified DNA was then treated with the restriction enzyme BS/I (Takara Bio Inc., Tokyo, Japan) and analyzed using the ABI Prism 3130xl DNA Sequencer (Applied Biosystems, CA) and Gene Mapper (Applied Biosystems). Cluster analysis was performed using the Gene Maths (Applied Maths, Sint-Martens-Latem, Belgium).

RNA extraction from the liver and intestinal tissues was carried out in accordance with a previously described method [15] using Isogen (Wako Pure Chemical Industries, Ltd.). The RNA was transcribed into cDNA using the High Capacity cDNA Reverse Transcription kit (Applied Biosystems), and qPCR was performed using the SYBR Green PCR Master Mix (Applied Biosystems). Quantification was carried out using the CFX96Touch™ (Applied Biosystems). The cycle step was 50 °C × 2 min, 95 °C × 10 min, (95 °C × 30 s, 60 °C × 30 s, 72 °C × 1 min) × 50 cycles. The primers used are summarized in Additional file 1.

The metabolomic analysis was conducted using the Dual Scan package of Human Metabolome Technologies Inc. (HMT; Yamagata, Japan) using liquid chromatography time-of-flight mass spectrometry (LC-TOFMS) and capillary electrophoresis (CE)-TOFMS for ionic and non-ionic metabolites, respectively, on the basis of the methods described elsewhere [16, 17]. Briefly, for extracting ionic metabolites, approximately 50 mg of feces sample was dissolved in MilliQ water with ratio of 1:9 (w/v). After centrifugation, 20 μL of the supernatant was suspended with 20 μL of internal standard solution (HMT, Tsuruoka, Yamagata, Japan) and 80 μL of MilliQ water. The solution was then centrifugally filtered through a Millipore 5000-Da cutoff filter (UltrafreeMC-PLHCC, HMT) to remove macromolecules (9100 × g, 4 °C, 60 min) for subsequent analysis with CE-TOFMS. For extracting non-ionic metabolites, approximately 50 mg of feces sample was dissolved in 1 mL of methanol containing internal standard (HMT). After centrifugation, 300 μL of the supernatant was moved into glass vial for evaporation under nitrogen gas and reconstituted with 300 μL of 50% isopropanol (v/v) for subsequent analysis with LC-TOFMS. CE-TOFMS analysis was performed using a CE capillary electrophoresis system with a 6210 time-of-flight mass spectrometer, 1100 isocratic high-performance liquid chromatography (HPLC) pump, G1603A CE-MS adapter kit, and G1607A CE-ESI-MS sprayer kit (Agilent Technologies, Waldbronn, Germany). The G2201AA ChemStation software version B.03.01 for CE (Agilent Technologies) was used to control the systems, which were connected by a fused silica capillary (50 μm i.d. × 80 cm total length) with a commercial electrophoresis buffer (H3301–1001 and H3302–1021 for cation and anion analyses, respectively; HMT) as the electrolyte. The spectrometer was scanned from m/z 50 to 1000. LC-TOFMS analysis was performed using an LC System (Agilent 1200 series RRLC system SL) equipped with a 6230 time-of-flight mass spectrometer (Agilent Technologies). The systems were controlled by the G2201AA ChemStation software version B.03.01 for CE equipped with an octadecylsilyl column (2 × 50 mm, 2 μm). Peaks were extracted using an automatic integration software, MasterHands (Keio University, Tsuruoka, Yamagata, Japan), to obtain peak information, including m/z, peak area, and migration time (MT) for CE-TOFMS and retention time (RT) for LC-TOFMS analyses. The following were excluded: signal peaks corresponding to isotopomers, adduct ions, and other product ions of known metabolites. The HMT metabolite database was used to annotate the remaining peaks on the basis of their m/z values with the MTs and RTs determined by TOFMS. Areas of the annotated peaks were then normalized based on internal standard levels and sample amounts to obtain relative levels of each metabolite. Hierarchical cluster analysis and principal component analysis were conducted using HMT’s proprietary software, PeakStat and SampleStat, respectively. Detected metabolites were plotted on metabolic pathway maps using the VANTED software.

The livers were perfused through the portal vein with fluorescence-activated cell sorting (FACS) buffer [1 g bovine serum albumin (BSA) in 500 mL PBS] and then minced well. The filtrate was centrifuged at 50 g for 1 min, and the supernatant was washed once. The cells were suspended in 25% Percoll and overlaid in 50% Percoll to distinguish resident macrophages from monocytes. After centrifugation at 2000 rpm for 20 min, the cells were collected from the middle layer and were washed and resuspended in an RPMI1640 medium. Flow cytometry was performed as described previously [18]. Briefly, the cells were collected from the upper phase of a Percoll gradient. After blocking with anti-FcR (CD16/32, BD bioscience, NJ) for 20 min, the cells were incubated with specific monoclonal antibodies at 4 °C for 30 min. The liver mononuclear cells were gated as 7-AAD negative CD45.2-FITC (BD bioscience) positive cells. Liver macrophages were stained with PE-conjugated anti-mouse F4/80 mAb and antigen-presenting cell (APC)-conjugated anti-mouse CD11b mAb (both from BD biosciences). Background fluorescence was assessed by staining with the relevant isotype control Abs. Stained cells were analyzed by flow cytometry (FACS Cant II, Becton Dickinson Co. Franklin Lakes, NJ), and data were analyzed using the FlowJo software (FlowJo, LLC, Ashland, OR).

Liver CD11b⁺ mononuclear phagocytes were separated from all other mononuclear phagocytes using MicroBeads (Miltenyi Biotec K.K., Japan). Then, separated CD11b⁺ cells were cultured in an RPMI1640 containing 0.1% penicillin/streptomycin (Sigma), supplemented with 10% fetal bovine serum (Biowest SAS, Nualle, France) with or without 200 μM palmitic acid (PA; Wako). After incubation for 20 h, mRNA was collected from the CD11b⁺ cells using Isogen.

Results were shown as mean ± standard errors. A one-way ANOVA followed by the Tukey’s post-hoc test was used for comparisons of multi groups. The Mann-Whitney U-test was used for comparisons of two groups. All comparisons were two-sided, and a P-value < 0.05 was considered significant. All statistical analyses were performed using the SPSS 22 for Windows (SPSS, IBM Japan, Tokyo, Japan).

C57BL/6 J mice were fed a control diet (SD diet; CONT), HFD (STHD-01 diet; STHD-01), or STHD-01 plus Abx (STHD-01 + Abx) for 9 weeks (Fig. 1a). Total calorie intake and total water intake were not different between the three groups (Fig. 1b). The STHD-01 HFD-fed mice exhibited an increased liver mass without an increase in body weight (Fig. 1b). The levels of fasting blood glucose at the week 15 and the plasma levels of insulin were not different between the CONT and HFD group (Additional file 2). Plasma T4, thyroxine, was not altered by STHD-01 (Fig. 1b). In contrast, T3, a thyroid hormone metabolized from T4, was significantly elevated in the STHD-01-fed mice (Fig. 1b). Administration of Abx significantly improved the increased liver weight and plasma T3 induced by STHD-01, suggesting that these changes are mediated by gut dysbiosis induced by STHD-01 (Fig. 1b). Likewise, the STHD-01 group showed significantly larger and more numerous fat droplets throughout the liver (Fig. 1c). Fat droplets were found in both the pericentral and periportal regions (Fig. 1c and Additional file 3). Significant numbers of ballooning and Mallory bodies were observed in the periportal region in the STHD-01 group (Additional file 3). Moreover, several fibrotic extensions were observed in the STHD-01 group (Fig. 1c and Additional file 3). Not only the lipid accumulation but also the STHD-01 induced the infiltration of inflammatory cells in the peripheral region in the liver (Fig. 1c). Notably, these pathological changes in the liver seen in the STHD-01-fed mice were not developed when the mice were administrated with Abx (Fig. 1c). Consistent with these histological changes, the STHD-01-fed mice exhibited increased levels of AST and ALT in the plasma and TG in the liver (Fig. 1d). Moreover, transcription of inflammatory markers tumor nuclear factor (TNF) -α and Interleukin (IL) -1β; and fibrosis markers α-smooth muscle antigen (α-SMA) and α1 type 1 collagen (Col 1α1) mRNAs in the liver were dramatically up-regulated in the STHD-01-fed mice (Fig. 1d). These changes were not observed in mice that receive Abx, suggesting that STHD-01-induced alteration of the gut microbiota plays a critical role in the development of steatohepatitis.

Next, we analyzed the composition of the gut microbiota in all three groups. The total number of bacteria in the feces was not affected by feeding of STHD-01 (Fig. 2a). In contrast, treatment with Abx dramatically reduced the number of gut microbes (Fig 2a). After feeding of STHD-01, an abundance of Bifidobacterium, Enterococcus, and Bacteroides genera decreased, while an abundance of Clostridium subclusters XIVa and XVIII increased (Fig. 2a). The treatment of Abx almost completely deleted these major genera of bacteria found in the STHD-01 group, and the genus of Enterococcus was dominated within the gut microbiota after the Abx administration. To address the influence of STHD-01-induced dysbiosis on the metabolic activities of the gut microbiota, we analyzed the gut metabolome. Lipid metabolites in the feces were analyzed by liquid LC-TOFMS and hydrophilic metabolites by CE-TOFMS. The detected peaks were categorized into glycolysis/gluconeogenesis, pentose-phosphate, tricarboxylic acid (TCA) cycle, urea cycle, purine-pyrimidine, coenzyme, amino acids, acyl-carnitine, and fatty acid pathways and were included in a pathway map (Additional file 4 A). There were also many detected metabolite peaks that were not categorized into any aforementioned metabolic pathways (Additional file 4 B–C). The relative area ratios of the three experimental groups were compared and shown in a heatmap with hierarchical clustering. This cluster analysis revealed that the luminal metabolic profiles in all three experimental groups were completely different (Fig. 2b). Among the largely different metabolites found in the STHD-01 group, we attempted to identify metabolic pathways in which the metabolites were constantly elevated throughout the pathway. We found that the pathways of long-chain saturated fatty acids, fatty acids longer than C14, and n-6 unsaturated fatty acids were constantly elevated in the STHD-01 group compared with the control group. Strikingly, the elevation of these pathways was normalized to control levels after the treatment with Abx, suggesting that these metabolic pathways were up-regulated due to gut dysbiosis induced by STHD-01 (Figs. 2b and 3).

So far, we have found that STHD-01-induced gut dysbiosis altered microbial metabolic activities, and therefore resulted in an accumulation of saturated long-chain fatty acids and n-6 unsaturated fatty acids. However, it remains unclear whether these metabolic changes in the gut lead to liver inflammation. To investigate the detailed phenotype of liver inflammation induced by the STHD-01, we next analyzed the macrophage populations, which play central roles in the development of inflammation in the liver [19]. Macrophages in the liver can be subdivided into resident and migratory macrophages [20]. Although both macrophage subsets express F4/80 and CD11b, resident macrophages reveal F4/80^highCD11b^low phenotypes, and migratory macrophages show F4/80^highCD11b^high phenotypes (Fig. 4a). The number of F4/80^highCD11b^high migratory macrophages was significantly elevated in the STHD-01 group (Fig. 4b). Of note, the migration of F4/80^highCD11b^high macrophages into the liver, which is elicited by STHD-01, was significantly decreased by Abx treatment (Fig. 4b). Next, we examined the plasma level of MCP-1, a chemokine that is responsible for the recruitment of monocytes from the bone marrow to the blood, thereby leading to the migration of monocytes to inflammatory tissue sites. Consistent with the number of migratory macrophages in the liver, plasma MCP-1 level was significantly elevated in the STHD-01 group and it was suppressed by the administration of Abx (Fig. 4c). We next asked the impact of long-chain saturated fatty acids in the induction of liver inflammation. CD11b⁺ macrophages were isolated from the liver of the control (CONT), STHD-01-fed (STHD-01). Isolated CD11b⁺ macrophages were then stimulated with palmitic acids (PA), in which the level was correlated to liver inflammation induced by STHD-01 (Figs. 1c, 2, and 3). We measured the transcription of Tnfa mRNA as a marker of liver inflammation. In the CONT group, PA stimulation did not induce Tnfa expression in liver macrophages (Fig. 4d). In contrast, PA stimulation significantly induced the expression of Tnfa in liver macrophages isolated from STHD-01 (Fig. 4d). Since migratory macrophages were dominated within CD11b⁺ mononuclear cell (MNC)s in the STHD-01 fed mice (Fig. 4a-d), this result indicated that PA could activate migratory, but not the resident, macrophages in the liver. PA was able to induce Tnfa expression in liver CD11b⁺ macrophages isolated from this group of mice. This result suggests that STHD-01-induced liver inflammation leads to the increased F4/80^highCD11b^high macrophage migration in the liver. Lipid mediators, which are generated by the gut microbiota, such as PA, promote the progression of liver inflammation by activating the migrated macrophages.