Faecalibacterium prausnitzii produces butyrate to decrease c-Myc-related metabolism and Th17 differentiation by inhibiting histone deacetylase 3

Abstract

Decreased levels of Faecalibacterium prausnitzii (F. prausnitzii), whose supernatant plays an anti-inflammatory effect, are frequently found in inflammatory bowel disease (IBD) patients. However, the anti-inflammatory products in F. prausnitzii supernatant and the mechanism have not been fully investigated. Here we found that F. prausnitzii and F. prausnitzii-derived butyrate were decreased in the intestines of IBD patients. Supplementation with F. prausnitzii supernatant and butyrate could ameliorate colitis in an animal model. Butyrate, but not other substances produced by F. prausnitzii, exerted an anti-inflammatory effect by inhibiting the differentiation of T helper 17 (Th17) cells. The mechanism underlying the anti-inflammatory effects of the butyrate produced by F. prausnitzii involved the enhancement of the acetylation-promoted degradation of c-Myc through histone deacetylase 3 (HDAC3) inhibition. In conclusion, F. prausnitzii produced butyrate to decrease Th17 differentiation and attenuate colitis through inhibiting HDAC3 and c-Myc-related metabolism in T cells. The use of F. prausnitzii may be an effective new approach to decrease the level of Th17 cells in the treatment of inflammatory diseases.

Faecalibacterium prausnitzii, histone deacetylase, inflammatory bowel disease, metabolism, T helper 17 cells

Introduction

Inflammatory bowel disease (IBD), including ulcerative colitis (UC) and Crohn’s disease (CD), is a chronic inflammatory disease that occurs in the gastrointestinal tract (1, 2). Although the etiology of IBD is unknown, the association between IBD and the intestinal microbiota has been extensively studied worldwide. The lack of beneficial bacteria or the increase of opportunistic pathogens (microbial dysbiosis) in IBD patients may be the basis of IBD immunopathogenesis (3, 4). Increasing studies have suggested beneficial bacteria as new breakthroughs in the treatment of IBD (5).

Studies have found that many bacteria, including Faecalibacterium prausnitzii (F. prausnitzii), are reduced in the intestinal tract of IBD patients and IBD animal models (6, 7). Belonging to the Clostridium leptum group, F. prausnitzii is one of the most abundant anaerobes in the human intestine. The concentrations of F. prausnitzii in the feces of IBD patients are significantly lower than those in healthy controls (6, 7). Studies found that oral administration of F. prausnitzii supplementation to IBD animal models can ameliorate colonic mucosal inflammation (8). We showed in previous studies that F. prausnitzii attenuates intestinal inflammation by inhibiting T helper 17 (Th17)-related interleukin-17 (IL-17) (9). Although a 15-kDa protein produced by F. prausnitzii could ameliorate colitis in IBD animals by inhibiting the nuclear factor (NF)-κB pathway, this protein exerts only part of the anti-inflammatory effects of F. prausnitzii (10, 11). Moreover, whether this protein can exert anti-inflammatory effects by inhibiting Th17 cells has not been studied yet.

In recent years, these Th17 cells have been shown to play an important pathogenic role in IBD. The level of Th17 cells is a key factor affecting the course and severity of intestinal inflammation. Th17 cells are a unique pro-inflammatory lineage of effector/memory Th cells firstly identified by their expression of IL-17 and retinoic acid receptor-related orphan receptor gamma t (ROR-γt) (12). The level of Th17 is elevated in the colonic mucosa and serum of IBD patients, affecting the course and severity of intestinal inflammation and maintaining mucosal homeostasis in the lamina propria (13, 14). Therefore, targeting Th17 may offer an effective new approach in the treatment of IBD. However, clinical trials showed that blockade of IL-17 was ineffective in CD and higher rates of adverse events were noted compared with placebo (15, 16). These showed that Th17 played a complex role in intestinal inflammation and people ought to find novel ways to reduce the differentiation of Th17 cells instead of blocking the IL-17 directly.

Among the targets involved in the polarization of T cells, c-Myc, a transcription factor, is a cell metabolism regulator (17). Studies have found that c-Myc maintains lactate dehydrogenase A (LDHA) and glutaminase (GLS) levels in T cells, thereby regulating glycolysis programming, glutamine uptake and catabolism (18, 19). Studies showed that different T lymphocyte subsets satisfy their metabolic demands by utilizing different fundamental metabolic pathways on the basis of nutrient availability, which is called aerobic glycolysis or the Warburg effect (20). Much of our current knowledge of this metabolic program derives from the study of cancer cells (21), but T lymphocytes can also induce aerobic glycolysis during effector responses (20). Interestingly, Th17 cells undergo metabolic regulation similar to that of tumor cells. Therefore, we can assume that a shift toward high glycolysis is a hallmark of inflammatory cells; moreover, metabolic reprogramming by targeting specific proteins may attenuate the direction of Th17 cell differentiation as well as intestinal inflammation (22).

Histone deacetylases (HDACs) are enzymes that remove acetyl groups from proteins. Studies in animal models of colitis have shown a clear association between HDACs and intestinal inflammation (23, 24). Researchers have found that the HDAC pan-inhibitor, ITF2357, reduced the production of IL-17 and Th17 (25). These findings suggest HDACs may be targets of the polarization regulation of Th17 cells in IBD, though it remains unclear which HDAC isoform plays a role. HDACs have histone and non-histone targets, and among them, several metabolic proteins involved in the regulation of cellular metabolism have been identified. Lysine acetylation is a common modification in cell metabolism (26) and a subset of HDACs has long been identified to regulate Th17 cell metabolism (27).

The hypothesis of the present study is that F. prausnitzii secretes certain substances to decrease Th17 cells and to ameliorate colitis inflammation by inhibiting HDACs and c-Myc-related metabolism. Thus, the aim of the present study is to identify the specific anti-inflammatory products in the F. prausnitzii supernatant and determine the underlying mechanism of ameliorating IBD colitis.

Methods

Patients

The protocols for the present study were authorized by the Ethical Committee of Drum Tower Hospital, Nanjing University. IBD patients and healthy participants were recruited from Drum Tower Hospital and exposed to similar diets and living environments. All the patients signed informed consent agreements. None of the participants used probiotics, antibiotics or sulfasalazine in the preceding 2 months. Control subjects were healthy volunteers or had non-inflammatory disorders (constipation, reflux, non-ulcer dyspepsia, etc.). The diagnosis of IBD was based on clinical, endoscopic and histological criteria. Disease activity was assessed by Crohn’s disease activity index (CDAI) (28) for CD and Truelove & Witts classification (29) for UC. The baseline characteristics of all participants are shown in Supplementary Table 1.

Experimental animals and induction of colitis

All animals were housed under specific pathogen free (SPF) conditions following the Animal Regulations of Jiangsu Province, China. Sprague-Dawley (S-D) rats (male, 6–8 weeks old) and C57BL/6J mice (male, 8–10 weeks old) were obtained from the Animal Center of Drum Tower Hospital (Nanjing, China). The S-D rats were randomized into four groups each with eight rats: the control group, colitis group, F. prausnitzii supernatant (FPsup) group and F. prausnitzii medium group. The control group received phosphate buffered saline (PBS) and ethanol (50:50 v/v) intrarectally. The other three groups were injected with 2, 4, 6-trinitrobenzenesulfonic acid (TNBS) (Sigma, St Louis, MO, USA) intrarectally to generate colitis models as previously described (30). After 2 days of TNBS treatment, rats were orally administered with 1 ml PBS, 1 ml F. prausnitzii supernatant or 1 ml F. prausnitzii medium, each concentrated five times, daily, respectively, for 7 days.

The mice were randomized in four groups: control, colitis, butyrate and RGFP966 groups, each containing eight mice. For the first 5 days, all mice, except for those in the control group, were treated with 3.0% dextran sulfate sodium (DSS) (MP Biomedicals, Santa Ana, CA, USA) in their drinking water ad libitum. For the next 2 days, the mice received normal drinking water. Throughout the 7 days, sodium butyrate (Sigma) solution was intra-peritoneally injected into the mice in the butyrate groups at a dose of 100 mg kg⁻¹ once daily. The RGFP966 group received RGFP966 (Selleck, Houston, TX, USA) intra-peritoneally at a dose of 5 mg kg⁻¹ once a day. For the other two groups, the same volume of PBS was intra-peritoneally administered once daily.

For transferring Th17 into mice, mice were randomized in two groups, each containing eight mice. All the mice were induced with colitis as described above. On the fourth day, eight mice were injected with 1 × 10⁶ Th17 cells suspended in 200 μl PBS via the tail vein and the control group received 200 μl PBS.

All rodents were sacrificed under anesthesia. The peripheral blood, spleens and mesenteric lymph nodes (MLNs) were isolated to detect cytokines and T cells. The distance from cecum to anus was measured. Part of the colon was fixed in 4% paraformaldehyde for pathological and immunohistochemical examinations, and others were used in real-time PCR (qPCR).

Bacterial culture and qPCR

Faecalibacterium prausnitzii (ATCC 27766, Manassas, VA, USA) was grown as previously described (9). Bacteria were grown at 37°C in an anaerobic incubator containing 97% CO₂ and 3% H₂. Faecalibacterium prausnitzii was cultured in LYHBHI medium [brain–heart infusion (37 g L⁻¹), yeast extract (5 g L⁻¹), hemin (5 mg L⁻¹)], supplied with cellobiose (1 g L⁻¹), maltose (1 g L⁻¹) and cysteine (0.5 g L⁻¹). For oral delivery to rats, bacteria were grown until a stationary phase and then collected with the supernatant lyophilized. The lyophilized supernatant of F. prausnitzii was prepared as 5× concentration prior to use. Faecalibacterium prausnitzii supernatant was collected after centrifugation and then filtered through a 0.22 μm sterilized filter (Millipore, Billerica, MA, USA). For the gene expression analysis in T cells and colon tissue, the qPCR was performed using SYBR® Green PCR Master Mix (Applied Biosystems, Carlsbad, CA, USA) with primers that amplify the genes encoding 16S rRNA from F. prausnitzii (forward, 5′-GATGGCCTCGCGTCCGATTAG-3′; reverse, 5′-CCGAAGACCTTCTTCCTCC-3′) and all bacteria (forward, 5′-GTGSTGCAYGGYTGTCGTCA-3′; reverse, 5′-ACGTCRTCCMCACCTTCCTC-3′). Each PCR reaction contained 10 ng of purified fecal DNA, 1 µM of each primer and 1× SYBR Green Master Mix. The final volume of each PCR reaction was 20 µl. PCR conditions were as follows: 94°C for 3 min, 35 cycles of 94°C for 30 s, 30 s at 58°C and 1 min at 72°C. Each plate included duplicate reactions per DNA sample, the appropriate set of standards and a ‘no-template’ negative control for each primer set. The qPCR standards were generated by the PCR amplification of target sequences from the genomic DNA of an appropriate positive control strain. Analysis of the melting curves confirmed that the fluorescence signal originated from specific PCR products and not from primer-dimers or other artifacts. For the qPCR of T cells and colon tissues, the primer sequences were: IL-17A (forward, 5′-AGCTTTCCCTCCGCATTG A-3′; reverse, 5′-GCTCCAGAAGGCCCTCAG A-3′), ROR-γt (forward, 5′-CACGGCCCTGGTTCTCAT-3′; reverse, 5′-GCAGATGTTC CACTCTCCTCTTCT-3′), IFN-γ (forward, 5′-GGTCATTACTGTC ACCGCCA-3′; reverse, 5′-ACACAGTACACAGTCAGCGC-3′), IL-4 (forward, 5′-GTTGTCATGCTGCTCTTC-3′; reverse, 5′-CTCTGTGGTGTTCTTCGT-3′), Foxp3 (forward, 5′-TCCC AGAGTTCTTCCACAAG-3′; reverse, 5′-TAAGGGTGGCATA GGTGAAA-3′) and GAPDH (forward, 5′-CATGGCCTTCC GTGTTCCTA-3′; reverse, 5′-TGTCATACTTGGCAGGTTTCT-3′).

Colon histopathological grading and immunohistochemistry

Colon pathology was blindly assigned scores by two pathologists according to Neurath Scoring criteria as described previously (31). Colon tissues were fixed in 4% paraformaldehyde and paraffin embedded. The sections were boiled in Tris-EDTA buffer (pH 9.0) for 20 min. After cooling, sections were incubated with IL-17 antibodies (Abcam, Cambridge, UK) for 24 h in 4°C. After incubating with the corresponding secondary antibody (Zsbio, Beijing, China) for 30 min in 37°C, the sections were treated with immunoperoxidase using a DAB kit (ZLI-9033 system, Zsbio). Staining intensity was graded as previously described (32). Each tissue has a score range of 0–12.

Flow cytometry analysis and enzyme-linked immunosorbent assay

For flow cytometry analysis (FACS) analysis, flow cytometry was routinely performed using 1 × 10⁶ cells per sample. The Th17 cells were stimulated with Leukocyte Activation Cocktail (BD Bioscience, Franklin Lakes, NJ, USA) for 4–6 h in an incubator (37°C, 5% CO₂), and then labeled with APC-CD3 and FITC-CD4. After permeabilization and fixation, the cells were labeled with PE-IL-17. The following antibodies were also used: anti-CD25 and anti-Foxp3 for regulatory T cells (Tregs); anti-IFN-γ for T helper cell 1 (Th1); anti-IL-4 for T helper cell 2 (Th2); anti-CD11b and anti-F4/80 for macrophages. The cells were detected by flow cytometry (BD Bioscience) and analyzed with FlowJo software (Tree Star, Inc., San Carlos, CA, USA). All antibodies used for FACS were obtained from eBioscience (San Diego, CA, USA).

IL-17A was measured using a commercially available enzyme-linked immunosorbent assay (ELISA) kit (eBioscience) according to the manufacturer’s protocols.

Cell isolation, differentiation and transfection

Naive CD4⁺ T cells in the mouse spleen were isolated from male C57BL/6J mice by magnetic beads (Miltenyi Biotec, Cologne, Germany) ex vivo and cultured in AIM-V medium (Invitrogen, Carlsbad, NY, USA) supplemented with 10% fetal bovine serum (Invitrogen). Naive CD4⁺ T cells were stimulated on cell culture plates coated with 10 μg ml⁻¹ anti-CD3 and 4 μg ml⁻¹ anti-CD28 (eBioscience) and generated by culture under Th17-polarizing conditions: 3 ng ml⁻¹ TGF-β (R&D, Minneapolis, MN, USA), 40 ng ml⁻¹ IL-6 (R&D), 30 ng ml⁻¹ IL-23 (R&D), 20 ng ml⁻¹ TNF-α (PeproTech, Rocky Hill, NJ, USA) and 10 ng ml⁻¹ IL-1β (PeproTech). These differentiated T cells were treated with 5× F. prausnitzii supernatant, 5× F. prausnitzii medium, 5 μM RGFP966, fraction A, 0.62 mM sodium butyrate or PBS and cultured for 5 days. In order to obtain highly enriched Th17 cells, some Th17 cells were stimulated with anti-CD3 and anti-CD28 again on the fifth day and cultured for another 4 days (9 days in total).

Human Jurkat cells were transfected using DNAfectin™ Plus (Abcam) according to the manufacturer’s instructions. The control vector, HDAC expression plasmid and shHDAC3 plasmid were kindly provided by the Zhao laboratory of Fudan University (Shanghai, China).

Western blot analysis

T cells and Jurkat cells were lysed with ice-cold radioimmunoprecipitation assay buffer, mixed with a protease and phosphatase inhibitor cocktail (Roche Diagnostics GMbH, Mannheim, Germany) and phenylmethylsulfonyl fluoride (PMSF) (Biosharp, Hefei, China) for 15 min on ice. Next, loading buffer containing 5% 2-mercaptoethanol was mixed with the lysate. The proteins were denatured at 100°C for 10 min before being separated on 8–12% sodium dodecyl sulfate–polyacrylamide gels. After transfer to polyvinylidene difluoride membranes (Millipore), the membranes were incubated for 2 h at room temperature in tris buffered saline-tween containing 5% skimmed milk. Afterward, membranes were incubated with the corresponding primary antibodies overnight at 4°C according to the manufacturer’s instructions. After treatment with the appropriate horseradish peroxidase (HRP)-conjugated secondary antibodies, the blots were visualized with enhanced chemiluminescence western blotting reagents (Millipore) according to the manufacturer’s protocols. The antibodies against the following proteins were used in the present study: ROR-γ, IL-17, c-Myc and GLS from Abcam; LDHA, HDAC3, HA-Tag, HRP-conjugated anti-mouse and anti-rabbit antibodies from CST (Danvers, MA, USA); Flag-Tag and β-actin from Bioworld (St. Louis Park, MN, USA) were also used.

Metabolic profiling and nutrient uptake

Trichloroacetic acid (TCA) cycle intermediates were collected as previously described (33). Metabolite extract samples (2 μl) were injected for the gas chromatography–mass spectrometry (GC-MS) analysis using an Agilent 6980 GC coupled to an Agilent 5973 MS system (Santa Clara, CA, USA). Relative metabolite abundances were determined by normalizing the abundance of each metabolite to the internal standard and the T-cell number.

Glutamine and glucose uptake were determined as described previously (34, 35). After culture with different components, T cells (4 × 10⁷ cells per ml) were suspended in a serum- and gluatamine- (for glutamine uptake) or glucose- (for glucose uptake) free Roswell Park Memorial Institute 1640 medium (Gibco). Cell suspension (50 μl) was carefully added to the top of a 0.7-ml microfuge tube, which was preloaded by 25 μl 8% sucrose and 20% perchloric acid (bottom), 200 μl 1-bromododecane (middle) and 50 μl uptake medium containing 2 μCi L-2,3,4-[³H] glutamine or 2 μCi [³H]2-deoxyglucose. The cells were exposed to radioactively labeled nutrients for 10 min at room temperature, then spun through bromododecane into the sucrose-perchloric acid layer to stop the reaction and separate the cells from the unincorporated ³H-labeled nutrient. The sucrose-perchloric acid layer containing T cells was collected for liquid scintillation assay to quantify the radiolabeled nutrients taken up by T cells.

Fractionation, GC-MS and protein inactivation

The ethanol-pretreated D-101 macroporous resins were placed in a 50-ml column, and the ethanol was replaced with double-distilled water. The concentrated F. prausnitzii supernatant was added to the resin column. After adsorption overnight, fractions A, B, C, D and E were obtained by elution with 0, 25, 50, 75 and 100% ethanol, respectively. After rotary evaporation at low pressure, these concentrated fractions were extracted with ether and sulfuric acid. GC analyses were performed on an Agilent 7890B gas chromatograph (Agilent) with an FID detector using a J&W DB-1 column (10 m, 0.1 mm ID). GC-MS spectra were performed on an Agilent 5977A Series (EI Source). Dodecane and ethyl acetate were used as the internal standard and solvent, respectively.

To inactivate the proteins in fraction A, 1 mg was either taken to repeat freezing at −80°C and thawing at room temperature for six cycles, irradiated with 20 J/m² UV light for 30 min, treated with 1 µg trypsin overnight at 37°C or heated at 99°C for 30 min.

HDAC deacetylation assay

The HDAC3 protein was incubated with acetylated peptides with RGFP966, butyrate or fraction A. The deacetylation assays were performed in a buffer containing 30 μl reaction buffer [0.6 mM MgCl₂ (Sangon, Shanghai, China), 10 mM PMSF (Sigma), 30 mM N-2-hydroxyethyl piperazine-N'-2-ethanesulfonic acid (Sigma), 1 mM dythiothreitol (Sigma), 1 mM NAD⁺ (Sigma)] containing 5 μg enzyme and 0.3 μg peptide. The deacetylation reaction was incubated for 3–5 h at 37°C. The mixture was desalted by passing it through a C18 ZipTip (Millipore) and the desalted samples were analyzed using a MALDI-TOF/TOF mass spectrometer (Applied Biosystems, Grand Island, NY, USA). The acetylated peptide used in this assay was NLASVEELK^AcEIDVEVRK (Glssale, Shanghai, China).

Statistical analysis

Statistical analyses were performed with GraphPad Prism version 5.0 (La Jolla, CA, USA). Comparisons among groups were performed by Student’s t-test and other data were analyzed using a one-way analysis of variance (ANOVA). All data are presented as mean ± SEM. P < 0.05 was considered to indicate statistical significance.

Ethics approval and consent to participate

All experiments utilizing animal and human samples were approved by the Ethical Committee of Medical Research, Nanjing Drum Tower Hospital, Medical School of Nanjing University. Both patients and healthy participants consented and signed the agreement to participate our study.

Results

Faecalibacterium prausnitzii supernatant exerts anti-inflammatory effects in IBD

Previous studies found that the concentration of F. prausnitzii in the feces of IBD patients was significantly lower than that in healthy people (6). Therefore, 16S rRNA sequencing was initially performed in the fecal samples from healthy subjects and IBD patients. We found that the levels of F. prausnitzii in IBD samples were significantly lower than those in the samples from healthy subjects (Fig. 1A). This was also validated in the IBD model mice (Fig. 1B; Supplementary Figure 1A). Considering that the level of Th17 cells is one of the key factors affecting the course and severity of intestinal inflammation (36), circulating Th17 cells in IBD patients was analyzed by flow cytometry. We found that in both UC and CD patients, Th17 cells were significantly higher than those in the healthy subjects (Fig. 1C). Previous studies have found that the supernatant of F. prausnitzii exerts anti-inflammatory effects in IBD animal models (37). However, the mechanism has not been fully investigated. Therefore, the effect of F. prausnitzii supernatant on TNBS-induced IBD models was demonstrated. The F. prausnitzii supernatant significantly alleviated the weight loss, the decrease in colon length and the elevated pathological scores in rats in the colitis group (Fig. 1D–F). Histological examination showed extensive ulceration in the colon of colitis rats. However, the F. prausnitzii and supernatant-treated rats showed only mild mucosal and/or submucosal inflammation with a relatively low level of mild edema (Fig. 1G). Then, Th17 cells were detected in the rats’ spleen. We found that the proportion of Th17 cells in the colitis group was significantly higher than that in the control group. After treatment with F. prausnitzii supernatant, Th17 was significantly reduced (Fig. 1H). The above results show that the F. prausnitzii supernatant reduces Th17 to exert anti-inflammatory effects.

Fig. 1.

Faecalibacterium prausnitzii supernatant exerts anti-inflammatory effects in IBD. (A) Levels of F. prausnitzii in the feces of healthy controls (n = 19), UC patients (n = 13) and CD patients (n = 13). (B) Levels of F. prausnitzii in the feces of normal and colitis mice. (C) Analysis of the Th17 cells among peripheral blood mononuclear cells (PBMCs) in blood of healthy controls (n = 19), UC patients (n = 13) and CD patients (n = 13). (D–H) S-D rats were treated orally with 1 ml PBS, 5× F. prausnitzii supernatant or 5× F. prausnitzii medium, respectively, with or without TNBS-induced colitis. n = 8. (D) Body weight changes, (E) colon length and (F) colon Neurath scores were evaluated. (G) Representative images of rat colonic mucosa. Upper and lower panel magnifications are ×100 and ×400, respectively. (H) Statistical analysis of FACS of the Th17 cells in the splenocytes of rats from each group. FPsup: F. prausnitzii supernatant group. Data are expressed as the means ± SEM. Asterisks (*) indicate significant differences (*P < 0.05; **P < 0.01).

Faecalibacterium prausnitzii supernatant inhibits the differentiation of Th17 cells

Th17 cells are identified by IL-17 (12). Thus, the levels of IL-17 were detected in the plasma and colons of the rats. After colitis induction, IL-17 was increased in plasma (Fig. 2A) and colon (Fig. 2B and C), but significantly decreased after treatment with F. prausnitzii supernatant. These results were also confirmed by in vitro differentiation studies of T cells. The level of Th17 cells (Fig. 2D) and IL-17A (Fig. 2E) under Th17-polarizing conditions was significantly increased compared with the non-stimulated control group, but reduced after treatment with F. prausnitzii supernatant. Supernatant treatment also decreased the Th17-related protein ROR-γt and IL-17 under Th17-polarizing conditions in vitro (Fig. 2F). According to these results, we conclude that F. prausnitzii secretes certain substances that can reduce Th17 cells by inhibiting their differentiation.

Fig. 2.

Faecalibacterium prausnitzii supernatant inhibits the differentiation of Th17 cells. S-D rats were treated with PBS, F. prausnitzii supernatant and F. prausnitzii medium, respectively, with or without TNBS-induced colitis. n = 8. (A) The level of IL-17A in the rat plasma. (B) Immunohistochemical images of IL-17 expression levels in the rat colon. Upper and lower panel magnifications are ×100 and ×400, respectively. (C) Staining scores of IL-17 expression levels in the rat colon. (D–F) Naive CD4⁺ T cells from C57BL/6J mice were stimulated with anti-CD3 and anti-CD28 antibodies, with or without Th17-polarizing conditions, and treated with PBS, F. prausnitzii supernatant or F. prausnitzii medium. n = 5. (D) Th17 cells and (E) IL-17A levels were measured. (F) ROR-γt and IL-17 expression was detected by western blotting. FPsup: F. prausnitzii supernatant group. Data are expressed as the means ± SEM. Asterisks (*) indicate significant differences (**P < 0.01).

Injecting Th17 cells into mice aggravates colitis

Considering that Th17 played a complex role in intestinal inflammation, in order to determine the function of in vitro induced Th17 cells in an in vivo IBD model, colitis mice were injected with Th17 cells. First, high concentrations of Th17 were obtained through differentiation of naive T cells in vitro. The relative expression levels of IL-17, ROR-γt, forkhead box protein P3 (Foxp3) (hallmark of Tregs), interferon (IFN)-γ (hallmark of Th1 cells) and IL-4 (hallmark of Th2 cells) mRNA in T cells after 9 days of differentiation were detected by qPCR. We found that the induced T cells expressed high levels of Th17-related IL-17 and ROR-γt mRNA (Fig. 3A). Then, colitis mice were injected with those Th17 cells. We found that the injection aggravated the weight loss (Fig. 3B) and pathological scores of colitis mice (Fig. 3C). Th17 cells were increased in MLNs (Fig. 3D), as well as the IL-17A in plasma (Fig. 3E). mRNA levels in the colon were detected and we found that IL-17 and ROR-γt were induced after Th17 injection (Fig. 3F). These results indicate that in vitro induced Th17 cells aggravate the development of intestinal inflammation when injected into colitis mice.

Fig. 3.

Injecting Th17 cells into mice aggravates colitis. Naive CD4⁺ T cells from C57BL/6J mice were stimulated with CD3 and CD28 antibodies and cultured for 9 days under Th17-polarizing conditions. (A) Relative IL-17, ROR-γt, Foxp3, INF-γ and IL-4 mRNA levels were detected on the ninth day. The levels of mRNA in naive CD4⁺ T cells were set to 1. (B–F) C57BL/6J mice were treated with 3.0% DSS in their drinking water. On the fourth day, colitis mice were injected with Th17 cells or PBS. (B) Body weight loss, (C) colon Neurath score, (D) Th17 cells in MLNs, (E) IL-17A in plasma and (F) relative levels of IL-17 and ROR-γt mRNA in colon were detected. Data are expressed as the means ± SEM. n = 8. Asterisks (*) indicate significant differences (*P < 0.05; **P < 0.01).

Faecalibacterium prausnitzii supernatant inhibits the c-Myc-related metabolism in Th17 cells by inhibiting HDAC3

To meet the requirements for cell growth and function, Th17 cells undergo an increase in glycolysis as their energy source (38). Therefore, inhibiting or promoting the activity of different metabolic enzymes in T cells can change the direction of T-cell differentiation (22). To determine whether the mechanism of F. prausnitzii inhibiting Th17 cells is through altering the metabolism of these cells, metabolomic analyses were performed using glutamine and glucose as tracers. Liquid chromatography–mass spectrometry (LC-MS) analysis demonstrated that F. prausnitzii supernatant treatment in Th17 cells significantly increased the levels of TCA cycle intermediates, including succinate, malate and α-ketoglutarate (α-KG) (Fig. 4A; Supplementary Figure 1B). To further examine the effect of F. prausnitzii supernatant on cell metabolism, glutamine and glucose uptake was assessed. Compared with naive T cells, Th17 cells showed increased glutamine and glucose uptake, whereas after treatment with the supernatant of F. prausnitzii, the uptake of both compounds was decreased (Fig. 4B). There were no significant differences in the relative cell viability between each group (Supplementary Figure 1C), indicating that the reduced metabolic parameters originated from inhibition of the Th17 intrinsic metabolic profile but not from reduced Th17 polarization. Treatment with F. prausnitzii supernatant led to increased glutamine and pyruvate accumulation. GLS and LDHA are the two important metabolic enzymes in glutaminolysis and glycolysis, respectively. Moreover, c-Myc, a transcription factor, regulates LDHA and GLS levels in T cells, thereby regulating glycolysis programming and glutamine uptake and catabolism (18). Therefore, we hypothesized that F. prausnitzii supernatant regulates T-cell metabolism by regulating c-Myc and the two key enzymes, LDHA and GLS. To address this hypothesis, naive T cells were cultured under Th17-differentiation conditions in vitro with or without the F. prausnitzii supernatant. We found that Th17 cells have higher levels of c-Myc, LDHA and GLS than naive T cells, whereas after treatment with supernatant, the expression of these proteins was inhibited (Fig. 4C). There is no specific inhibitor of c-Myc, but evidence indicates that HDAC inhibitors can down-regulate the expression of c-Myc (39). We suspected that there were similar components in the F. prausnitzii supernatant that inhibit c-Myc expression. Therefore, several HDAC inhibitors, including (−)-Parthenolide (specific for HDAC1), CAY10683 (specific for HDAC2), RGFP966 (specific for HDAC3) and SAHA (pan-HDAC inhibitor) were used in vitro. We found that the inhibition of HDAC1 and HDAC2 had no influence on the expression of c-Myc, LDHA, GLS and ROR-γt, but the HDAC3 inhibitor and the pan-HDAC inhibitor led to a significant reduction of these proteins in Th17 cells (Fig. 4D). These findings indicate that only the inhibition of HDAC3 decreases the level of c-Myc, changing the metabolic enzymes downstream and the direction of T-cell differentiation.

Fig. 4.

Faecalibacterium prausnitzii supernatant decreases c-Myc-related metabolism and Th17 differentiation by inhibiting HDAC3. Naive CD4⁺ T cells from C57BL/6J mice were stimulated with immobilized anti-CD3 and soluble anti-CD28 monoclonal antibodies, with or without Th17-polarizing conditions, and treated with PBS or F. prausnitzii supernatant. (A) Relative abundance of several metabolites under Th17-polarizing conditions treated with (green) or without (blue) F. prausnitzii supernatant was assessed. (B) Glucose uptake (top) and glutamine uptake (bottom) under each condition were assessed. (C) T cells were subjected to western blotting. (D) Naive CD4⁺ T cells from C57BL/6J mice were stimulated with or without Th17-polarizing conditions in vitro, and treated with PBS or different HDAC inhibitors. Cells were subjected to western blotting. (E and F) Naive CD4⁺ T cells from C57BL/6J mice were stimulated with CD3 and CD28 antibodies, with or without Th17-polarizing conditions, and treated with PBS, 5× F. prausnitzii supernatant or 5 μM RGFP966. Th17 cells were detected by (E) FACS and subjected to (F) western blotting. (G) Jurkat cells were transfected with either control vectors or HDAC3 expression plasmids, and then subjected to western blotting. FPsup: F. prausnitzii supernatant group. Data are expressed as the mean ± SEM. n ≥ 3. Asterisks (*) indicate significant differences (**P < 0.01).

Since the inhibition of HDAC3 leads to a decrease of c-Myc, altering Th17 cell metabolism and thereby affecting the differentiation of these cells, we suspected that there might be HDAC3 inhibitors in the F. prausnitzii supernatant that alter Th17 cell differentiation. To answer this question, the supernatant of F. prausnitzii and the HDAC3-specific inhibitor, RGFP966, were added to naive T cells under Th17-differentiation conditions in vitro. The results demonstrated that both the supernatant and RGFP966 could inhibit Th17 cell differentiation (Fig. 4E). To quantify protein levels, the proteins in the above culture conditions were examined by western blotting. Although the supernatant of F. prausnitzii did not significantly inhibit HDAC3 expression, both the supernatant and RGFP966 inhibited c-Myc, LDHA and GLS levels (Fig. 4F). To determine the levels of proteins when HDAC3 was over-expressed in cells, a plasmid expressing HDAC3 was transfected into Jurkat cells. The over-expression of HDAC3 induced an increase in c-Myc, LDHA and GLS (Fig. 4G). These observations illustrate that, by inhibiting HDAC3, the supernatant of F. prausnitzii decreases c-Myc-related metabolism enzymes and Th17 differentiation in vitro.

GC-MS analysis reveals that F. prausnitzii produces butyrate to inhibit Th17

Consistent with the above results, we determine that the F. prausnitzii supernatant exerted an anti-inflammatory effect on chemical colitis models by inhibiting Th17 cell differentiation. We hypothesized that substances secreted by F. prausnitzii affected Th17 cell differentiation. Consequently, to examine its composition, the supernatant of F. prausnitzii was fractionated into five components by polarity (Fig. 5A). The anti-inflammatory activity of each fraction was tested in vitro. We found that the most polar component, fraction A, was the strongest inhibitor of Th17 and IL-17A under Th17-polarizing conditions in vitro (Fig. 5B and C). A recent study has shown that F. prausnitzii produced a protein to inhibit the NF-κB pathway in intestinal epithelial cells to ameliorate inflammation in IBD models (10). To verify whether the protein affects Th17 differentiation, proteins in fraction A were inactivated by various treatments including repeated freezing and thawing, ultraviolet radiation, trypsin digestion or heating at 99°C. The results showed that IL-17A under Th17-polarizing conditions was still inhibited after protein inactivation (Fig. 5D). These findings indicated that other products, instead of the protein in the F. prausnitzii supernatant, regulate Th17 cell differentiation. To identify the non-protein components in fraction A, which play a role in Th17 cell differentiation, GC-MS analysis was performed, and a variety of substances were detected. The highest peak in fraction A was butyrate, with a concentration of 0.62 mM (Fig. 5E). Among the many substances identified, butyrate is an important anti-inflammatory mediator in the colon and is a major metabolite in the colonic lumen (40). To verify whether the effects on Th17 cell differentiation were exerted by butyrate, CD4⁺ naive T cells were treated with sodium butyrate at the same concentration as in fraction A (0.62 mM) under Th17-polarizing conditions. We found that butyrate inhibited the differentiation of Th17, with a similar effect to F. prausnitzii supernatant, RGFP966 and fraction A (Fig. 5F). The levels of c-Myc, LDHA, GLS and ROR-γt were significantly reduced when cultured with sodium butyrate and fraction A, but the expression of HDAC3 was not significantly affected (Fig. 5G). These findings suggest that the butyrate produced by F. prausnitzii reduces Th17 by regulating cell metabolism.

Fig. 5.

GC-MS analysis identifies that F. prausnitzii produces butyrate to inhibit Th17. (A) The supernatant of F. prausnitzii was stepwise fractionated by polarity. Naive CD4⁺ T cells from C57BL/6JB6 mice were stimulated with CD3 and CD28 antibodies, with or without Th17-polarizing conditions, and treated with PBS or different fraction of F. prausnitzii supernatant. (B) The proportion of Th17 was detected by FACS and (C) the IL-17A levels were determined by ELISA. (D) Fraction A was inactivated by various treatments including repeated freezing and thawing, ultraviolet radiation, trypsin digestion or heated at 99°C. Naive CD4⁺ T cells were treated with PBS or different fraction A preparations. IL-17A level in culture supernatant under Th17-polarizing conditions in vitro was determined. (E) The GC spectrum of fraction A. The 1.957 min peak was identified to be butyrate. (F) CD4⁺ naive T cells were treated with PBS, 5× F. prausnitzii supernatant, 5 μM RGFP966, fraction A or 0.62 mM sodium butyrate. Th17 cells were determined. (G) CD4⁺ naive T cells were cultured under different conditions and subjected to western blotting. Ctrl: control group. FPsup: F. prausnitzii supernatant group. The values are expressed as the means ± SEM. n = 5. Asterisks (*) indicate significant differences (**P < 0.01).

Butyrate enhances the acetylation-induced degradation of c-Myc through HDAC3 inhibition

As an HDAC inhibitor, butyrate can exert an anti-inflammatory effect (41). However, the specific targets of butyrate are unclear. To identify its mechanism, plasmids harboring HDAC1–7 were transfected into Jurkat cells, and the over-expressed proteins were subsequently purified by immunoprecipitation and verified by western blotting. We found that c-Myc physically interacted with HDAC3, but not with other HDACs (Fig. 6A). In addition, the binding of endogenous HDAC3 and c-Myc was determined. And we found that the addition of butyrate reduced the HDAC3 binding to c-Myc (Fig. 6B). Then, cycloheximide (CHX) was used to inhibit c-Myc synthesis. The results showed that c-Myc degraded faster in the presence of butyrate (Fig. 6C). In addition, butyrate-induced c-Myc degradation was blocked when MG132 was used to inhibit proteasomal degradation (Fig. 6D). These results indicate that the ubiquitin-proteasome pathway mediates the acetylation-promoted induced reduction of c-Myc. To elucidate the direct target of butyrate and fraction A, an in vitro deacetylation system was used. All treatments including RGFP966, butyrate and fraction A inhibited HDAC3 deacetylation activity (Fig. 6E). In addition, to determine the acetylation of endogenous c-Myc, Th17 cells were treated with sodium butyrate, RGFP966 or fraction A. All treatments increased the acetylation level of c-Myc (Fig. 6F). In Jurkat cells, butyrate increased the acetylation of c-Myc but resulted in a marginal acetylation change after the lysine (K) of the K323 acetylation site was mutated to arginine (R) (Fig. 6G and H). Moreover, butyrate and RGFP966 promoted the ubiquitination of c-Myc in Jurkat cells (Fig. 6I and J). The above results indicate that butyrate enhances the acetylation-induced degradation of c-Myc through HDAC3 inhibition.

Fig. 6.

Butyrate enhances the acetylation-promoted degradation of c-Myc through HDAC3 inhibition. (A) Jurkat cells were transfected with HA-tagged HDAC1–7 plasmid, respectively. The cells were lysed and immunoprecipitated with HA beads. The input was visualized by western blotting with specific antibodies. The immunoprecipitated samples were washed and visualized by western blotting with the antibodies for HA and c-Myc. (B) Jurkat cells were treated with PBS or 0.62 mM sodium butyrate. The cells were harvested and visualized by western blotting. (C) Jurkat cells were treated with 100 μg ml⁻¹ CHX, with or without butyrate, and subjected to western blotting. (D) Jurkat cells were treated with or without 10 μM MG132 and butyrate and subjected to western blotting. (E) The HDAC3 protein was incubated with acetylated peptides and RGFP966, butyrate or fraction A, respectively. Mass spectrometry was used to determine the rate of deacetylation. (F) Th17 cells were treated with butyrate, RGFP966 or fraction A. The cells were harvested and visualized by western blotting. (G and H) Jurkat cells were transfected with wild-type or mutant flag-tagged c-Myc plasmid and then treated with butyrate. Immunoprecipitated proteins from cell lysates were analyzed by western blotting. (I and J) Flag-tagged c-Myc and HA-tagged ubiquitin were co-expressed in Jurkat cells or HDAC3 knockdown Jurkat cells. The cells were then treated with 10 μM MG132 or several HDAC inhibitors. Ubiquitination levels of affinity-purified Flag-c-Myc proteins were detected and visualized by western blotting.

Butyrate reduces Th17 differentiation by inhibiting HDAC3 in vivo

Finally, the anti-inflammatory effects of butyrate and RGFP966 on DSS-induced IBD mice were compared. Both butyrate and RGFP966 significantly alleviated the weight loss, shortened colon length and increased Neurath scores in colitis mice (Supplementary Figure 1D; Fig. 7A and B). Histological examination showed extensive ulceration in the colons of colitis mice. However, the butyrate-treated rodents showed only mild mucosal and/or submucosal inflammation with a relatively low level of mild edema. These results were consistent with the immunohistochemical staining of IL-17 protein (Fig. 7C and D). Notably, the proportion of Th17 in the spleens of mice was significantly reduced after treatment with butyrate or RGFP966 (Fig. 7E; Supplementary Figure 1E). Interestingly, there was no significant difference in the above results between the butyrate and RGFP966 groups. These results confirmed that the anti-inflammatory mechanism of butyrate was similar to that of RGFP966, namely, the inhibition of HDAC3, resulting in a reduction of Th17. Finally, the butyrate levels were determined in the feces of IBD patients and healthy subjects. We found that butyrate was reduced in the feces of both UC and CD patients (Fig. 7F). In addition, the correlation analysis revealed that the intestinal butyrate levels were negatively correlated with the plasma Th17 levels in IBD patients (Fig. 7G). These results show that butyrate reduces Th17 differentiation by inhibiting HDAC3 in vivo.

Fig. 7.

Butyrate reduces Th17 differentiation by inhibiting HDAC3 in vivo. C57BL/6J mice were treated with 3.0% DSS in their drinking water. Throughout the 7 days, sodium butyrate solution (100 mg kg⁻¹) was intra-peritoneally injected into the mice in the butyrate groups once a day, and the RGFP966 group received RGFP966 (5 mg kg⁻¹) intra-peritoneally once daily. n = 8. (A) Colon length. (B) Colon Neurath score. (C) Immunohistochemical staining scores of IL-17 expression levels in the mice colon. (D) HE and immunohistochemical staining images of IL-17 protein in mice colonic mucosa. Magnifications are ×100. (E) The Th17 cell ratio in the splenocytes of the mice in each group was detected by FACS. (F) Butyrate levels in the feces of healthy controls (n = 19), UC patients (n = 13) and CD patients (n = 13) were detected. (G) The correlation between butyrate and Th17 in IBD patients (n = 26). The values are expressed as the means ± SEM. Asterisks (*) indicate significant differences (*P < 0.05; **P < 0.01; NS, not significant).

Discussion

The risk factors of IBD are multifactorial. Recent studies have suggested that the disorder of human intestinal microbiota could lead to the development of IBD. Accordingly, the use of beneficial bacteria has emerged as a new treatment. Many bacteria have been reported to have anti-inflammatory effects on IBD patients or animal models, including F. prausnitzii (37, 42, 43). The anti-inflammatory mechanisms of these bacteria are not clearly elucidated since most of the previous research on these bacteria are correlation studies. Faecalibacterium prausnitzii is one of the most abundant anaerobes in the human intestine, a significant reduction in which may promote the recurrence of IBD (37). Consistent with previous studies, the present study confirms that the levels of F. prausnitzii and Th17 cells are decreased in IBD patients, and the F. prausnitzii supernatant ameliorates the health status in TNBS-induced colitis models by decreasing Th17 cells (Figs 1 and 2).

The role of Th17 cells in intestinal pathology remains controversial (44). There have been two clinical trials to assess the efficacy of human IL-17 antagonists in clinical trials in severe CD. Both trials were terminated early because the blockade of IL-17 was ineffective and higher rates of adverse events were noted compared with placebo (15, 16). However, the present study demonstrated that the in vitro induced Th17 cells would lead to an aggravation of IBD (Fig. 3), indicating that Th17 played a complex role in intestinal inflammation. We should find new approaches to reduce the differentiation of Th17 cells instead of blocking the IL-17 directly.

A recent study demonstrated that different T lymphocyte subsets satisfy their metabolic demands by utilizing different fundamental metabolic pathways. Much of the current knowledge of this metabolic program derives from studies of cancer cells (21), but T lymphocytes can also induce aerobic glycolysis during effector responses (20). Interestingly, Th17 cells have metabolic regulation similar to that of tumor cells. Cell growth and function require adequate energy production and biosynthesis (45). To meet these requirements, proliferating and activated Th17 cells decrease lipid oxidation and undergo a rapid increase in glycolysis as their energy source (38). Interestingly, naive T cells show distinct metabolic requirements from Th17, having a metabolism characterized by mitochondrial oxidation (46). Therefore, we can assume that a transition to high glycolysis is a hallmark of inflammatory cells. By inhibiting or promoting the activity of different metabolic enzymes in T cells, the direction of T-cell differentiation can be changed (22). Metabolomics analysis demonstrated that F. prausnitzii supernatant treatment in Th17 cells significantly reduced the levels of TCA cycle intermediates, as well as glutamine and glucose uptake, by inhibiting c-Myc, LDHA and GLS (Fig. 4A–C).

Most of the glucose consumed by activated T cells is converted to lactate by LDHA (47). Flux through LDHA may be necessary to facilitate the continuance of aerobic glycolysis. In addition, mitochondrial GLS is a key enzyme that converts glutamine to glutamate and is highly expressed in proliferating lymphocytes. Studies have found that c-Myc can regulate LDHA and GLS levels in T cells, thereby regulating glycolysis programming and glutamine uptake and catabolism (18, 19). The role of c-Myc has been implicated in rapidly proliferating cells including T cells, as a regulator of cellular metabolism (17). The activation of T cell results in the promoting of c-Myc, which is necessary for the conversion of cellular metabolism to aerobic glycolysis during the initial activation (48). Knockdown of c-Myc in T cells fails to up-regulate glycolysis and glutaminolysis upon activation. This is consistent with the role of c-Myc in promoting aerobic glycolysis in cancer (49). The activity of c-Myc is regulated by acetylation, which can reduce the stability of this protein (50). There is evidence indicating that HDAC inhibitors may reduce c-Myc expression (39), and we found that only the inhibition of HDAC3 will decrease the level of c-Myc and c-Myc-related metabolic enzymes in T cells, thereby inhibiting Th17 cell differentiation (Fig. 4D–G). Although studies have indicated that the recruitment of c-Myc by HDAC1 and HDAC3 is the major mechanism regulating the activation or inhibition of several target genes (50), in terms of affecting Th17 differentiation, only HDAC3 plays a role.

Although an anti-inflammatory protein has been discovered in the F. prausnitzii supernatant (10), it is not sufficient to fully explain the anti-inflammatory mechanism. This suggested that in addition to protein, other non-protein products in F. prausnitzii supernatant also contribute to the anti-inflammatory effect. We separated the F. prausnitzii supernatant and found that the strongest inhibition on Th17 was the most polar component, even after the protein inactivation. Then we confirmed that it is butyrate that inhibits the differentiation of pro-inflammatory Th17 cells to attenuate colitis (Fig. 5). These findings suggest that F. prausnitzii exerts an anti-inflammatory effect by producing multiple substances via different mechanisms, respectively. However, in terms of inhibiting Th17 cells in IBD, it is indeed butyrate that plays a key role.

Butyrate, a short-chain fatty acid produced by bacterial fermentation of dietary fiber, has an important role in maintaining normal intestinal function (40, 41). Researchers have found that butyrate can act as an HDAC inhibitor to exert anti-inflammatory functions. However, the specific HDAC isoform targets and molecular mechanisms of butyrate in IBD have not been fully investigated. HDACs are a family of enzymes, classified into 18 isoforms and four groups. HDACs can regulate transcriptional activities by removing the acetyl groups from lysine on histones and non-histones (51, 52). An explicit link between HDACs and IBD has been found in colitis animals. To determine the target of butyrate, we found that both butyrate and fraction A could inhibit HDAC3 and downstream c-Myc, LDHA and GLS in T cells in vitro, indicating that the F. prausnitzii-derived butyrate ameliorates colitis through inhibiting HDAC3. We further explored the target of the interaction between butyrate and c-Myc via immunoprecipitation. We found that HDAC3 deacetylated c-Myc at K323, further protecting c-Myc from ubiquitinated degradation (Fig. 6). Butyrate inhibited HDAC3, resulting in increased c-Myc ubiquitination and decreased c-Myc-related metabolism.

To observe whether butyrate has an anti-inflammatory role in colitis models, the effects of butyrate were evaluated in the DSS-induced IBD model. The anti-inflammatory effects of butyrate and RGFP966, the specific HDAC3 inhibitor, were compared. Interestingly, data showed that the anti-inflammatory effect of butyrate in IBD was similar to that of RGFP966, demonstrating that the anti-inflammatory mechanism behind butyrate was similar to that of RGFP966, namely, the inhibition of HDAC3. This indicates that butyrate reduces Th17 differentiation by inhibiting HDAC3 in vivo. Although butyrate is a pan-HDAC inhibitor, in terms of regulating the metabolism in Th17 cells in IBD, it is HDAC3 that plays a key role. Finally, a correlation analysis revealed that the intestinal butyrate levels were negatively correlated with the plasma Th17 levels in IBD patients (Fig. 7). Although many studies have shown that butyrate alleviates colitis as an HDAC inhibitor (24, 53), most of these studies have focused on supplementing exogenous butyrate rather than bacteria-derived butyrate. More importantly, these studies did not clarify the underlying mechanisms and specific targets. The present study focused on F. prausnitzii-related butyrate, showing that this compound inhibited HDAC3 and c-Myc-related metabolism, thus reducing Th17 differentiation and ameliorating IBD-associated inflammation (Fig. 8).

Fig. 8.

Schematic model of anti-inflammatory effects of F. prausnitzii-derived butyrate on colitis in IBD. Faecalibacterium prausnitzii-derived butyrate inhibits HDAC3 in Th17 cells, leading to the increased ubiquitination of c-Myc. The reduced c-Myc level inhibits both LDHA-related glycolysis and GLS-related glutaminolysis, thus reducing the differentiation of Th17 to ameliorate IBD inflammation.

Interestingly, in the preliminary experiments, we also detected the levels of other IBD-related immune cells including Treg, Th1, Th2 and macrophages in the spleen of colitis rats treated with or without F. prausnitzii supernatant and found significant differences (Supplementary Figure 2A). However, in subsequent experiments we found that the levels of Treg, Th1, Th2 and macrophages in mouse MLNs after HDAC3 inhibitor intervention were similar to those in the model group (data not shown). In addition, our recent studies have shown that butyrate can also regulate the Treg/Th17 balance by inhibiting HDAC1 (54). This indicated that F. prausnitzii could affect those cells in IBD models, while the mechanisms by which F. prausnitzii regulates those cells is not via HDAC3.

Evidence indicates that F. prausnitzii, Roseburia sp. and Eubacterium are normally the most abundant groups of human fecal bacteria that produce butyrate, together accounting for ~7% of total fecal bacteria (55–58). Even higher estimates of abundance have been reported for F. prausnitzii (59). To determine any changes of other butyrate-producing bacteria in the mice with the supernatant of F. prausnitzii, the fecal microbiota profiles were assessed. We found that butyrate-producing bacteria were significantly less abundant in colitis mice than in healthy mice. After treatment with F. prausnitzii supernatant, the abundance of Roseburia sp. (including R. faecis, R. hominis, R. intestinalis and R. inulinivorans) did not change and Eubacterium (including E. rectale and E. hallii) even decreased (Supplementary Figure 2B). Other butyrate-producing bacteria (e.g. B. fibrisolvens, Strain SS2/1, A. caccae L1-92, Strain M62/1 and Strain GM2/1) could not be detected in the present study because of the low abundance of them. This indicated that the mechanism by which F. prausnitzii ameliorates colitis is not by promoting other butyrate-producing bacteria.

Conclusion

In summary, the present study demonstrates that F. prausnitzii produces butyrate to decrease Th17 differentiation and attenuate the severity of colitis by inhibiting HDAC3 and c-Myc-related metabolism in T cells. The use of butyrate may be an effective new approach to decrease Th17 cells in the treatment of inflammatory disease. These findings emphasize the potential role of F. prausnitzii in the treatment of IBD and may shed light on the mechanism of butyrate-producing bacteria.

Funding

This work was supported by grants from the National Natural Science Foundation of China (no. 81602076), Outstanding Youth Project of Nanjing City (no. JQX17002), the Jiangsu Clinical Medical Center of Digestive Disease (BL2012001), the Natural Science Foundation from the Department of Science & Technology of Jiangsu Province (BK20160113), the Fund of Jiangsu Provincial Commission of Health and Family Planning (no. Q201611) and the Fundamental Research Funds for the Central Universities (no. 021414380244).

Conflicts of interest statement: the authors declared no conflicts of interest.

Acknowledgements

The authors would like to thank the Laboratory of Changhong Liu (School of Life Science, Nanjing University, China) for technical assistance in bacterial culture. The authors would also like to thank Dr Hang Liu (Department of Pharmacy, Affiliated Drum Tower Hospital of Nanjing University, China), the Laboratory of Shaolin Zhu and Prof. Chengjian Zhu (Chemistry School, Nanjing University, China) for technical assistance in supernatant fractionation and GC-MS analysis. M.Z. and C.Y. designed the study; Q.Z., D.T., L.Z. and Y.W. performed the cell experiments; D.T., Y.W. and L.X. collected the tissue samples; Q.Z. and Y.P. performed the protein analysis; L.Z., R.G.D., M.Z. and Y.L. drafted the manuscript and performed the immunohistochemistry experiment; J.W. and S.Z. performed the metabolite analysis; R.G.D. and M.Z. drafted the manuscript; and M.Z. and C.Y. supported the study. All authors have read and approved the final manuscript. All experiments utilizing animal and human samples were approved by the Ethical Committee of Medical Research, Nanjing Drum Tower Hospital, Medical School of Nanjing University. Both patients and healthy participants consented and signed the agreement to participate our study.

Author notes

These authors contributed equally to this work.