Ganoderma lucidum lowers obesity in mice by changing gut bacteria

Using a mouse model of obesity, we observed that HFD feeding for 8 weeks led to significant increases in body and liver weight, epididymal and subcutaneous fat accumulation, and lipid deposition in adipocytes and hepatocytes compared with control chow feeding (Fig. 1a–g). While 8% WEGL did not produce any apparent effects in chow-fed mice, supplementation with WEGL decreased weight gain and fat accumulation in a dose-dependent manner in HFD-fed mice (Fig. 1a–g). Mean energy intake, stool fat and faeces energy did not vary significantly between HFD-fed groups (Supplementary Fig. 1), suggesting that the effects of WEGL on body weight and obesity parameters were not due to reduced food consumption or energy extraction. These results imply that WEGL reduces weight gain and fat accumulation in HFD-fed mice.

Previous studies have shown that HFD-fed obese mice produce higher levels of pro-inflammatory cytokines in hepatic and adipose tissues, including tumour necrosis factor-alpha (TNF-α), interleukin-1-beta (IL-1β), interleukin-6 (IL-6) and plasminogen activator inhibitor-1 (PAI-1; ref. 26). In contrast, production of the anti-inflammatory cytokine IL-10 is reduced in obese animals27. We measured messenger RNA (mRNA) expression of these cytokines after 8 weeks of HFD feeding with or without WEGL supplementation. TNF-α, IL-1β, IL-6 and PAI-1 expression levels were higher in hepatic and adipose tissues of HFD-fed mice compared with tissues of control chow-fed mice, whereas IL-10 expression was reduced (Fig. 2a–e). Notably, the expression pattern of these cytokines was altered in a dose-dependent manner by WEGL treatment, resulting in expression levels closer to that of chow-fed mice than HFD-fed mice with increasing WEGL dose (Fig. 2a–e). Moreover, WEGL reduced the levels of secreted TNF-α, IL-1β and IL-6 proteins in a dose-dependent manner in the serum of HFD mice (Supplementary Fig. 2).

Obesity is characterized by infiltration and activation of immune cells in hepatic and adipose tissues28. M1 macrophages, which are recruited by monocyte chemoattractant protein-1 (MCP-1), are associated with chronic, low-grade inflammation in adipose tissues of obese animals26. As MCP-1 mRNA expression decreased in the liver and adipose tissues of HFD-fed mice following treatment with WEGL (Supplementary Fig. 3a), we examined the numbers of macrophages recruited into hepatic and adipose tissues using flow cytometry analysis. Double staining of macrophages with anti-F4/80 antibody combined with either anti-CD11b antibody (for liver macrophages, also called Kupffer cells) or anti-CD11c antibody (for macrophages in adipose tissues) was used for these experiments. Higher levels of macrophages were detected in hepatic and adipose tissues of HFD mice compared with chow-fed mice (Supplementary Fig. 3b,c). While macrophage levels increased following the treatment with 8% WEGL in chow-fed mice, these cells were reduced in a dose-dependent manner by WEGL in hepatic and adipose tissues of HFD-fed mice (Supplementary Fig. 3b,c).

Previous studies indicated that HFD feeding also produces a gradual loss of regulatory T (Treg) cells compared with chow-fed mice26. While HFD also reduced Treg levels in our mouse model, supplementation with WEGL led to dose-dependent increases of Treg cells in liver and adipose tissues of HFD mice (Supplementary Fig. 3d,e). In addition, chow-fed mice treated with 8% WEGL showed increased Treg levels compared with untreated chow-fed mice (Supplementary Fig. 3d,e). These results indicate that WEGL reduces inflammation in HFD-fed mice by reducing macrophage infiltration and enhancing Treg accumulation in hepatic and adipose tissues.

Endotoxemia and TLR4 signalling control the production of pro-inflammatory cytokines in target tissues and lead to chronic inflammation and insulin resistance in HFD-fed mice19. We examined the effects of WEGL on LPS serum levels (that is, endotoxemia) and TLR4 protein expression in hepatic and adipose tissues. WEGL reduced endotoxemia and TLR4 protein expression in HFD-fed mice compared with HFD alone (Fig. 3a–c). Since TLR4 signalling pathways induce the production of pro-inflammatory cytokines by modulating the activity of JNK and NF-κB29,30, we examined whether these pathways are affected by WEGL supplementation. As shown in Fig. 3d,e, WEGL inhibited JNK phosphorylation in hepatic and adipose tissues of HFD-fed mice. Moreover, the production of IκB-α, whose interaction with NF-κB prevents NF-κB translocation and activation, was enhanced by WEGL treatment (Fig. 3f,g).

Given that enhanced activation of JNK and NF-κB pathways may induce insulin resistance via phosphorylation of insulin receptor substrate-1 (IRS-1) on serine 307 and dephosphorylation of Akt31, we examined the effects of WEGL on insulin activity. As shown in Supplementary Fig. 4a–e, WEGL treatment reduced fasting insulin and glucose levels and decreased glucose and insulin resistance in HFD-fed mice. Accordingly, phosphorylation of serine 307 on IRS-1 was inhibited by WEGL in hepatic and adipose tissues, whereas phosphorylation of Akt was enhanced by the mycelium extract (Supplementary Fig. 4f–i). WEGL therefore reduces endotoxemia and prevents insulin resistance in HFD-fed mice.

Previous studies have shown that the expression of genes involved in lipid biosynthesis such as acetyl-CoA carboxylase-1, fatty acid synthase, sterol regulatory element-binding protein-1 and peroxisome proliferator-activated receptors-γ is enhanced in hepatic and adipose tissues of HFD-fed mice32,33. Our results indicated that 8% WEGL did not significantly affect the expression of these genes in chow-fed mice (Supplementary Fig. 5a–d). In contrast, WEGL reduced lipogenic gene expression in a dose-dependent manner in HFD-fed mice (Supplementary Fig. 5a–d).

Increased levels of free fatty acids (FFA) in the blood of obese animals are thought to arise from a larger mass of adipose tissue and enhanced inflammation in HFD-fed mice34. These circulating fatty acids also induce TLR4 signalling in target tissues and cause insulin resistance22. We observed that WEGL reduced serum FFA levels in HFD mice (Supplementary Fig. 5e). WEGL may thus reduce fat accumulation by inhibiting adipogenic gene expression and decreasing serum FFA in HFD-fed mice.

The gut microbiota of obese humans and HFD-fed mice is characterized by an increased Firmicutes-to-Bacteroidetes ratio, elevated endotoxin-producing Proteobacteria, and reduced immuno-homeostatic bacterial species35,36. We examined the effects of WEGL on gut microbiota composition by performing a pyrosequencing-based analysis of bacterial 16S rRNA (V3–V5 region) in caecal faeces. After removing unqualified sequences (see Methods), a total of 691,370 raw reads and an average of 16,461±5,411 reads per sample were obtained. After selecting the effective reads, a total of 292,952 effective reads was generated and each faecal sample (n=7 for each group) produced an average of 6,975±2,192 effective reads. Samples with a low number of effective reads (<3,000) were not observed. Rarefaction and Shannon index analyses indicated that the sequencing depth covered rare new phylotypes and most of the diversity (Supplementary Fig. 6).

UniFrac-based principal coordinates analysis (PCoA) revealed a distinct clustering of microbiota composition for each treatment group (Fig. 4a). Multivariate analysis of variance of PCoA matrix scores indicated a statistically significant separation between the microbiota of Chow and Chow+8% WEGL groups (Fig. 4b). Significant separations were also noted for HFD, HFD+4% WEGL and HFD+8% WEGL groups (Fig. 4b). On the other hand, the gut microbiota of HFD+2% WEGL mice did not significantly differ from that of HFD mice (Fig. 4b), consistent with the mild effects produced by 2% WEGL (Figs 1, 2, 3). Notably, taxonomic profiling demonstrated that treatment with 4% and 8% WEGL reduced the ratio of Firmicutes to Bacteroidetes and the Proteobacteria phylum in HFD-fed mice to levels similar to that of chow-fed mice (Fig. 4c). Here as well, the changes produced by the 2% WEGL treatment were not observed (Fig. 4c).

We used redundancy analysis (RDA) to identify the specific bacterial phylotypes that were altered by HFD feeding and WEGL treatment. Compared with chow-fed mice, HFD feeding significantly altered 209 operational taxonomic units (OTUs), producing 44 increased and 165 decreased OTUs (Supplementary Data 1 and Supplementary Fig. 7a). In HFD-fed mice, supplementation with 2, 4 and 8% WEGL, respectively, altered 44 (24 increased and 20 decreased), 42 (24 increased and 18 decreased) and 56 OTUs (24 increased and 32 decreased; Supplementary Data 2 and Supplementary Fig. 7b–d), resulting in significant changes in a total of 91 distinct OTUs (Fig. 4d). Among the 91 OTUs that were altered compared with HFD-fed mice, 2, 4 and 8% WEGL treatments altered OTUs in the same direction in Chow mice for 18, 17 and 30 OTUs, respectively (Fig. 4e, highlighted with black stars). These results suggest that 8% WEGL is the most effective treatment for modulating the gut microbiota in our model.

Detailed analysis of the 30 OTUs reversed by 8% WEGL indicated that Mucispirilum shaedleri37, Escherichia fergusonii (Proteobacteria20), Enterococcus spp.38, Lactococcus lactis39, Clostridium lactatifermentans (Clostridium XIVb) and Oscillibacter valericigenes25,40 (which were found to be enhanced by HFD and to positively correlate with obesity in the previous studies cited above) were all reversed by 8% WEGL (Fig. 4d,e). Notably, in comparison with HFD mice, 8% WEGL treatment enhanced a variety of bacterial species that negatively correlate with obesity, including Parabacteroides goldsteinii, Bacteroides spp., Anaerotruncus colihominis, Roseburia hominis, Clostridium methylpentosum (Clostridium IV), Clostridium XIVa and XVIII and Eubacterium coprostanoligenes (Fig. 4d,e and Supplementary Data 2; note that these species were initially reduced by HFD compared with chow feeding). Bacterial species including E. coprostanoligenes, C. methylpentosum, P. goldsteinii, Bacteroides spp., A. colihominis, R. hominis and Clostridium XIVa and XVIII were also enriched in the Chow+8% WEGL treatment compared with the chow diet (Supplementary Data 3 and Supplementary Fig. 7e). Moreover, many bacterial species increased in the 8% WEGL group but were not altered by HFD, indicating that WEGL may enrich specific bacterial species (Fig. 4e and Supplementary Data 2). Collectively, these results show that WEGL modulates the gut microbiota of HFD-fed mice, resulting in a microbiota composition similar to that of chow-fed mice.

Given that intestinal dysbiosis in HFD-fed animals may affect gut permeability and subsequently lead to release of bacterial LPS into the circulation19, we examined whether WEGL modulates gut integrity. While HFD feeding reduced expression of the tight junction components, zonula occludens-1 (ZO-1) and occludin, these effects were reversed by WEGL supplementation (Supplementary Fig. 8). These findings suggest that WEGL may improve intestinal barrier integrity in HFD-fed mice.

Recent studies have shown that the ability of the gut microbiota to modulate obesity can be transferred to other animals13. To determine whether the gut microbiota of WEGL-treated animals may improve the condition of HFD-fed mice, we transferred the microbiota of WEGL-treated mice to HFD-fed mice, followed by examination of obesity-related traits. Analysis of the cohort of donor mice indicated that WEGL treatment reduced body weight and the Firmicutes-to-Bacteroidetes ratio in the gut microbiota of HFD-fed mice (Supplementary Fig. 9a,b), similar to the results shown above (Figs 1a and 4c). Our results further showed that horizontal faecal transfer from mice fed with chow (Chow→HFD), Chow+8% WEGL (8% WEGL (Chow)→HFD), or HFD+8% WEGL (8% WEGL (HFD)→HFD) reduced body weight, epididymal and subcutaneous fat accumulation and liver weight compared with faecal transfer from HFD-fed mice (HFD→HFD; Fig. 5a–d). Faecal transfer from the Chow+8% WEGL group produced the most robust effects on weight gain and fat accumulation (Fig. 5a–c).

Expression of pro-inflammatory cytokines and lipogenic genes was also reduced in hepatic and adipose tissues of HFD mice after transfer of faeces derived from WEGL-treated animals (Fig. 6a–d and Supplementary Fig. 10). Furthermore, the mRNA and protein expression levels of tight junction proteins in the ileum segment were also increased following the transfer of faeces from WEGL-treated mice (Fig. 6e–g). Transfer of faeces from 8% WEGL-chow-fed mice to HFD mice also produced statistically significant changes in TNF-α expression (Fig. 6a), lipogenic genes (Supplementary Fig. 10) and ileum tight junctions (Fig. 6e–g), compared with faecal transfer from chow-fed mice. These results suggest that the weight-lowering effects of WEGL in HFD-fed mice may be due to modulation of the gut microbiota.

In order to confirm that WEGL modulates the gut microbiota, we examined the composition of intestinal bacteria following faecal transfer from 8% WEGL-treated mice. In a separate sequencing analysis, a total of 1,056,611 raw reads and an average of 52,831±6,766 reads per sample were obtained. Subsequently, a total of 458,093 effective reads was generated and each faecal sample (n=5 for each group) produced 22,905±3,039 effective reads. Under this sequencing depth, rare phylotypes and most diversity were also covered (Supplementary Fig. 11). In comparison with mice that received faeces from HFD-fed mice, recipient HFD-fed mice (n=5 for each group) showed distinct microbiota after faecal transplantation from Chow-, Chow+8% WEGL- or HFD+8% WEGL-fed mice (Fig. 7a,b). Faecal transfer from HFD mice produced Firmicutes to Bacteroidetes ratios and Proteobacteria and Deferribacteres phyla levels that did not differ from those of HFD-fed mice (Fig. 7c, compared with Fig. 4c). In contrast, faeces transfer from Chow-, Chow+8% WEGL and HFD+8% WEGL-fed mice produced values similar to those of chow-fed or WEGL-treated mice (Fig. 7c, compared with Fig. 4c).

RDA analysis identified a total of 155 OTUs that were significantly altered by faecal transplant from Chow- (126 OTUs), Chow+8% WEGL- (71 OTUs) and HFD+8% WEGL-fed groups (75 OTUs) compared with HFD→HFD mice (Supplementary Data 4 and Supplementary Fig. 12). Among these, M. shaedleri, E. fergusonii, L. lactis, C. scindens and O. valericigenes, which were enhanced by HFD feeding (Fig. 4d), were all reversed by faecal transfer from Chow-, Chow+8% WEGL- and HFD+8% WEGL-fed mice (Fig. 7d,e). In addition, Clostridium cocleatum, Bacteroides caccae, Eubacterium dolichum and Coprobacillus cateniformis were altered after faecal transplantation from the same three groups of mice (Fig. 7d,e, these bacteria were not identified in the WEGL treatment experiments shown in Fig. 4). Of note, while the HFD-induced reduction of P. goldsteinii (Fig. 4d,e) was observed in HFD→HFD mice, this species was induced by faecal transplantation from the other three groups of mice (Fig. 7d,e). Interestingly, Akkermansia muciniphila, a potentially beneficial bacterial species, was only identified in HFD-fed mice that received faeces from Chow+8% WEGL-treated mice (Fig. 7d,e). These results indicate that WEGL restores the gut microbiota of HFD mice to a composition similar to those of chow-fed mice.

To identify the active ingredients of WEGL responsible for the anti-obesity effects, we fractionated the mycelium extract into four fractions (G1, G2, G3 and G4) on the basis of their molecular weights (Table 1). As shown in Fig. 8a–e, fraction G1, which consisted of polysaccharides with molecular weight >300 kDa (Table 1) and a monosaccharide composition comprising 47.5% mannose, 26.3% glucose and 16.9% galactose (Table 2), produced significant anti-obesity effects on body weight, liver weight and epididymal and subcutaneous fat in HFD-fed mice. Notably, the extent of anti-obesity effects observed was similar to that produced by the whole WEGL mycelium extract (Fig. 8a–e). In contrast, fraction G2 showed modest anti-obesity effects, whereas no significant effect was observed for fractions G3 and G4 (Fig. 8a–e). No significant change in mean energy intake, stool fat or energy extraction was noted following treatment with the polysaccharide fractions (Supplementary Fig. 13). These results suggest that the anti-obesity effects of WEGL mycelium may be due mainly to its high molecular weight polysaccharide fraction.

Animal experiments were approved and performed in accordance with the guidelines of Institutional Animal Care and Use Committee of Chang Gung University (IACUC; Approval No. CGU11-117). Mice of the C57BL/6NCrlBltw genetic lineage were purchased from Biolasco (Taiwan) and kept under controlled light conditions (12 h light–dark cycle), with free access to food and water. Eight-week-old male mice were randomly distributed into six groups containing five to seven animals each (see the legend of Figs 1,4–6,8). Mice were housed in groups of three or four animals per cage, and were fed with either a standard chow diet (13.5% of energy from fat; LabDiet 5001; LabDiet, USA) or a high-fat diet (60% of energy from fat; TestDiet 58Y1; TestDiet, USA). The G. lucidum strain initially isolated and characterized at Chang Gung Biotechnology and the water extract of cultured mycelium were processed as described before62. Briefly, the water extract was prepared by mixing 400 g of dried G. lucidum mycelium in 10 l of water before agitation at 150 r.p.m. for 30 min at 121 °C. The solution was cooled to room temperature (22 °C), followed by centrifugation at 5,894g for 30 min at 4 °C using a Sorvall RC 3C Plus centrifuge (Thermo Fisher Scientific, USA). The supernatant was concentrated using a Buchi R220 Rotavapor vacuum concentrator (Buchi, Switzerland) at 65 °C to obtain a final volume of 2 l. The extract was sterilized at 121 °C for 20 min, followed by storage at 4 °C in dark glass bottles. The mycelium consisted of <5% crude fibre; >10% polysaccharides; crude proteins at 38.78 g per 100 g; crude fat at 2.41 g per 100 g; carbohydrates at 41.99 g per 100 g; amino acids at 5.2 g per 100 g; and sodium at 76.39 mg per 100 g. Calories were measured at 345 kcal per 100 g. Each group of mice was fed for 2 months with chow diet or HFD, plus daily administration of 100 μl of either water or WEGL at 2, 4 or 8% (w/v) by intragastric gavage. Mean energy intake was assessed by converting the amount of food consumed into calories using the information provided by the manufacturer (LabDiet 5001 and TestDiet 58Y1).

WEGL polysaccharide fractions were provided by Chang Gung Biotechnology. Briefly, 100 ml of WEGL extract was mixed with 600 ml of 95% ethanol (v/v) before incubation for 16 h at 4 °C. The solution was centrifuged at 4,500g for 20 min at 4 °C (Sorvall RC 3C Plus centrifuge; Thermo Fisher Scientific, USA). The first supernatant was decanted and 100 ml of 70% cold ethanol was added to produce a pellet. The solution was centrifuged as above and the second supernatant was decanted. This operation was repeated three times. The first and second supernatants were pooled to obtain a volume of 1,050 ml (fraction G4). The precipitate pellet containing crude polysaccharides was mixed with 1,000 ml of distilled water and the material was dissolved completely. The solution of crude polysaccharides was concentrated to a final volume of 700 ml using an R220 vacuum concentrator (Büchi, Switzerland) at 65 °C to remove residual alcohol. Distilled water was added into the crude polysaccharide solution to obtain a final volume of 2,400 ml. The solution was filtered using the Spectrum KrosFlo system with 0.2 μm hollow fibre (1,500 cm², PES; Spectrum Laboratories, USA). The transmembrane pressure was set at 15 p.s.i. A total of 1,800 ml of distilled water was added during filtration of the crude polysaccharide solution. The retentive filtrate (650 ml) was preserved (G1-1). The solutions were filtrated using a 300 kDa cassette membrane (Minimate TFF Capsule; Pall, USA). A total of 600 ml of distilled water was added during filtration. The retentive filtrate (950 ml) was gathered (G1-2). The G1-1 and the G1-2 filtrates were combined to obtain a volume of 1,600 ml (G1). The permeated filtrate was filtrated with a 10 kDa cassette membrane. A total of 600 ml of distilled water was added during filtration. The retentive filtrate (970 ml) (G2) and 3,600 ml of the permeated filtrate (G3) were obtained. The G1, G2, G3 and G4 solutions were concentrated to a final volume of 100 ml using the vacuum concentrator.

Faecal transplant was performed based on an established protocol63. Briefly, 8-week-old male donor mice (n=5 per diet group) were fed with Chow, Chow+8% WEGL, HFD or HFD+8% WEGL for 3 months. After 4 weeks of feeding, stools were collected daily for the subsequent 2 months under a laminar flow hood in sterile conditions. Stools from donor mice of each diet group were pooled and 100 mg was resuspended in 1 ml of sterile saline. The solution was vigorously mixed for 10 s using a benchtop vortex (Vortex-Genie 2, Scientific Industries, USA; speed 9), before centrifugation at 800g for 3 min. The supernatant was collected and used as transplant material as described below. Fresh transplant material was prepared on the same day of transplantation within 10 min before oral gavage to prevent changes in bacterial composition. Eight-week-old male recipient mice (n=5 for each transplant group) were fed with HFD and inoculated daily with fresh transplant material (100 μl for each mouse) by oral gavage for 2 months, before being killed for subsequent analysis. Analysis of body weight and gut microbiota of the donor mice (Supplementary Fig. 9a,b) indicated that the 4-week-long diet treatments used for this cohort of mice produced changes similar to those of the 2-month-long treatments (Figs 1a and 4c).

Antibodies against IκB-α (nuclear factor of kappa light polypeptide gene enhancer in B cells inhibitor alpha; 1:1,000; 9242L), JNK (c-Jun N-terminal; 1:1,000; 9252L), phosphorylated JNK (1:1,000; 4668S), phosphorylated IRS-1 (insulin receptor substrate-1; phosphorylation on serine 307; 1:1,000; 2381S), IRS-1 (1:1,000; 2382L), Akt (1:1,000; 4691L), phosphorylated Akt (phosphorylation on serine 473; 1:1,000; 4060L) and TLR4 (1:1,000; 2219) were purchased from Cell Signaling Technology (USA). Antibodies against zonula occludens-1 (ZO-1; 1:250; 61–7,300) and occludin (1:250; 71–1,500) were obtained from Invitrogen (USA). Secondary antibodies (anti-rabbit IgG, H+L; 1:10,000 and 1:2,000 for primary antibodies from Cell Signaling Technology and Invitrogen, respectively; 111-035-003) were purchased from Jackson ImmunoResearch (USA). Antibody against β-actin (1:20,000; NB600-501) and secondary antibody anti-mouse IgG (1:20,000; sc-2005) were purchased from Novus Biologicals (USA) and Santa Cruz Biotechnology (USA), respectively. FITC-conjugated anti-F4/80 (1:100; 11-4801; eBioscience, USA), PE-conjugated anti-CD11b (1:100; 557397; BD Pharmingen, USA), anti-CD11c (1:100; 557401; BD Pharmingen), PE-conjugated anti-CD4 (1:100; 557308; BD Pharmingen), PerCp-Cy5.5-conjugated anti-CD25 (1:100; 551071; BD Pharmingen) and Alexa Flour 488-conjugated anti-Foxp3 (1:100; 560403; BD Pharmingen) antibodies were also used.

Measurements were made based on a protocol reported previously64. The energy density was determined using an adiabatic bomb calorimeter (Gallenkamp, UK). Total faecal lipids were extracted as described previously65. Briefly, dried faecal samples were homogenized once with heptane/diethylether/ethanol (1:1:1, vol/vol) and twice with heptane/diethylether/ethanol/water (1:1:1:1, vol/vol). Supernatants were collected in glass vials that had been weighed beforehand. Total lipid amount in each supernatant was measured gravimetrically after solvent evaporation and was expressed as a percentage of the weight of the starting faecal sample.

Frozen liver sections (6-μm thick) were stained with Oil Red O (Sigma, USA) for 20 min, and then counter-stained with haematoxylin for 1 min. Sections were examined under light microscopy. A total of 20 tissue sections were analysed for each animal.

Overnight-fasted mice were given glucose by oral gavage (3 g kg⁻¹, 66% solution), and measurement of homeostatic model assessment-insulin resistance was performed as described before20. Blood glucose was determined with a glucose meter (Roche Diagnostics, Switzerland) using blood collected from the tip of the tail vein.

Serum LPS quantification was performed using a commercial kit based on the use of a Limulus amaebocyte extract (Lal kit; Cambrex Bio Science, USA). Samples were diluted 1:20 and heated for 10 min at 70 °C. Serum insulin concentrations were determined in 5 μl of serum using a commercial ELISA kit (Mercodia, Sweden) based on the manufacturer’s instructions. Serum FFA were determined using a commercial detection kit (Biovision, USA).

Total RNA was isolated using a total tissue RNA isolation kit (Geneaid, Taiwan). Equal amounts of total RNA were used to synthesize cDNA with the Quant II fast RT kit (Tools, Taiwan). Quantitative real-time reverse-transcription PCR (qRT–PCR) was performed in triplicate using SYBR Green, 384-well plates and the LightCycler 480 Real-Time PCR System (Roche Diagnostics). Each well was loaded with a total of 10 μl containing 1.5 μl of cDNA, 1 μl of target primers, 1.5 μl of water and 5 μl of Kapa SYBR Fast Master Mix. Hot-start PCR was performed for 50 cycles, with each cycle consisting of denaturation for 15 s at 94 °C, annealing for 30 s at 60 °C and elongation for 30 s at 72 °C. The Roche LightCycler software (version 1.5.0, Roche Diagnostics) was used for data analysis. Relative quantification was done using the 2^−ΔΔCT method66. Expression was normalized against the housekeeping gene glyceraldehyde 3-phosphate dehydrogenase. Mean expression levels of chow-fed mice were set as 100%. The primers used are shown in Supplementary Table 1.

One hundred mg of adipose, liver or intestine tissue were homogenized in a commercial Pro-Prep Protein Extraction Solution (Intron Biotechnology, South Korea). Total protein lysates were fractionated on a 10% sodium dodecyl sulfate–polyacrylamide gel and electro-blotted onto polyvinylidene difluoride membranes (Immobilon TM-P; Millipore, USA). Membranes were blocked with 5% non-fat milk for 1 h at room temperature in TBST buffer (Tris 10 mM, NaCl 150 mM, pH 7.6, 0.1% Tween 20) and probed with primary antibodies overnight at 4 °C. Membranes were then incubated with horseradish peroxidase-conjugated secondary antibody. The dilutions of primary and secondary antibodies were described in the Antibody section above. Protein bands were developed using enhanced chemiluminescence (Millipore). The original immunoblots were provided in. Supplementary Fig. 14

Freshly isolated epididymal adipose tissues from wild-type or WEGL-treated mice were fixed overnight in 10% formalin, followed by dehydration, embedding in paraffin, and sectioning. Sections of 8 μm were stained with haematoxylin and eosin, and cell size was analysed using the Image J software (National Institutes of Health, USA).

Epididymal adipose and hepatic tissues were rinsed in saline, minced into fine pieces and digested with collagenase (Sigma, USA) in Krebs–Henseleit–HEPES buffer (Sigma, USA; pH 7.4) supplemented with 20 mg ml⁻¹ of BSA and 2 mM glucose. Incubation was performed at 37 °C in a shaker for 45 min. The samples were passed through a mesh and fractionated by brief centrifugation (1,000g). The pellets or the floating cells were collected to obtain the stromal–vascular fraction or adipocytes in adipose tissues, respectively. Cells in the stromal–vascular fraction and liver fractions were lysed in Pharm Lyse buffer for 30 min at 4 °C and resuspended in Pharmingen stain buffer (Becton Dickinson, USA). Cells were incubated with the commercial antibody 2.4G2, which reacts against the FcgII and FcgIII murine receptors (Becton Dickinson, USA) for 10 min and then with primary antibodies or the matching control isotypes for 30 min at 4 °C. The cells were rinsed twice and resuspended in Pharmingen stain buffer. Flow cytometry analysis was performed using a FACSCalibur (Becton Dickinson, USA). Data analysis was performed using the Kaluza flow cytometry analysis software (Beckman Coulter, USA). Macrophages were identified either as F4/80 and CD11c double-positive cells in adipose tissues or as F4/80 and CD11b double-positive in the liver. CD4, CD25 and transcription factor Foxp3 were analysed for Treg cells.

IL-1β, IL-6 and TNF-α protein levels were measured using commercial ELISA kits (R&D Systems, USA).

Stool samples were snap-frozen in liquid nitrogen before storage at −80 °C. DNA was extracted using a faecal DNA isolation kit (MoBio Laboratories, USA). For each caecal stool sample, the 16S rRNA gene comprising V3–V5 regions was amplified using a composite forward primer and a reverse primer containing a unique 10-base barcode to tag each PCR product. For each sample, a 50 μl PCR mix was prepared containing 25 ng DNA template, 5 × KAPA HiFi Buffer, 10 mM KAPA dNTP Mix, 1 U μl⁻¹ KAPA HiFi DNA Polymerase (KAPA Biosystems, USA) and 0.3 μM of composite primer pairs. The PCR reaction conditions consisted of 95 °C for 3 min, followed by 15–25 cycles of 98 °C for 20 s, 45 °C for 15 s and 72 °C for 15 s, and a final extension of 72 °C for 1 min. The composite primer pairs consisted of the forward primer (5′-GCCTTGCCAGCCCGCTCACTCCTACGGGAGGCAGCAG-3′)—a composite of 454 primer B (underlined) and the universal bacterial primer 338F (italics)—and the reverse primer (5′-GCCTCCCTCGCGCCATCAGNNNNNNNNNNCCGTCAATTCMTTTGAGTTT-3′)—a composite of 454 primer A (underlined), a unique 10-base barcode (NNNNNNNNNN) and the broad-range bacterial primer 907R (italics). Replicate PCRs were pooled and amplicons were purified using the QiaQuick PCR Purification Kit (Qiagen, USA). PCR products were sequenced on a 454 FLX pyrosequencer platform according to 454 Roche recommended procedures. High-quality reads for bioinformatics analysis were selected and all of the effective reads from all samples were clustered into OTUs based on 97% sequence similarity according to 454 SOP67. Briefly, quality control criteria for selecting raw reads of high quality included removal of sequences that lacked V3–V5 primers or barcode sequence (barcode sequence poorly matched or was not identified) as well as sequences that contained a short variable region (<90 bp) or undetermined bases (>2 bases) in the V3–V5 variable region. De-noising criteria comprised (1) using a 50-bp sliding window to reduce sequencing error and (2) an average quality score of 35 within that window for sequence trimming. All high-quality sequences were aligned using the nearest alignment space termination multi-aligner in SILVA-compatible database alignment. Reads that did not align to the anticipated region of the reference alignment were removed. Chimera sequences identified using the UCHIME algorithm68 were removed. All reads were classified using a Bayesian classifier with homemade RDP database. Reads that could not be classified at the kingdom level were removed. Alpha diversity analysis including rarefaction analysis and Shannon index were calculated using QIIME69. Fast UniFrac PCoA was performed with the phylogenetic tree constructed by inserting the representative of each OTU also generated using QIIME69. According to published statistical methods70, the statistical significance of the separation among animal groups in PCoA scores plots was assessed by multivariate analysis of variance test with differences in physiological and biochemical values for statistical significance. Normally distributed data were analysed by analysis of variance (ANOVA) using Tukey’s post hoc test (SPSS, USA). The relative abundance of each OTU (log₁₀-transformed) was used to construct RDA models and OTUs that were different between animal groups were assessed with Canoco for Windows 4.5 (Microcomputer Power, USA) according to the manufacturer’s instructions. Statistical significance was assessed using the Monte Carlo Permutation Procedure with 499 random permutations.

Using bacteria-specific primers described in Supplementary Table 1, abundance of Firmicutes, Bacteroidetes and total bacteria in faecal stool from each diet group at indicative time point was determined by 2^−ΔΔCT method-based quantitative real-time PCR66. The relative abundance of Firmicutes and Bacteroidetes was normalized to total bacteria for calculation of Firmicutes-to-Bacteroidetes ratio.

Data obtained from three replicate experiments are shown as means±s.e.m. Differences in body weight were assessed using the unpaired two-tailed Student’s t-test. Data sets that involved more than two groups were assessed by one-way ANOVA followed by Newman–Keuls post hoc tests (see the legend of Figs 1–3,5,6,8). Next generation sequencing analysis was assessed using Tukey’s honest significant difference post hoc tests. A P value of 0.05 was considered statistically significant. In the figures, the data with different superscript letters are significantly different based on post hoc ANOVA statistical analysis. SPSS version 17.0 was used.