Anti-inflammatory Bifidobacterium strains prevent dextran sodium sulfate induced colitis and associated gut microbial dysbiosis in mice

Crohn’s and ulcerative colitis are common inflammatory conditions associated with Inflammatory bowel disease. Owing to the importance of diet based approaches for the prevention of inflammatory gut conditions, the present study was aimed to screen the human isolates of Bifidobacterium strains based on their ability to reduce LPS-induced inflammation in murine macrophage (RAW 264.7) cells and to evaluate prioritized strains for their preventive efficacy against ulcerative colitis in mice. Twelve out of 25 isolated strains reduced the production of LPS-induced nitric oxide and inflammatory cytokines. Furthermore, three strains, B. longum Bif10, B. breve Bif11, and B. longum Bif16 conferred protection against dextran sodium sulfate induced colitis in mice. The three strains prevented shortening of colon, spleen weight, percentage body weight change and disease activity index relative to colitis mice. Lower levels of Lipocalin-2, TNF-α, IL-1β and IL-6 and improved SCFA levels were observed in Bifidobacterium supplemented mice relative to DSS counterparts. Bacterial composition of B. longum Bif10 and B. breve Bif11 fed mice was partly similar to the normal mice, while DSS and B. longum Bif16 supplemented mice showed deleterious alterations. At the genus level, Bifidobacterium supplementation inhibited the abundances of pathobionts such as Haemophilus, Klebsiella and Lachnospira there by conferring protection.


Results
Screening for Bifidobacterium strains that could reduce nitric oxide production by LPS challenged macrophage cells. Based on the ability of the strains to curtail NO production by LPS induced macrophages and lack of NO production by per se treatment of bacteria on macrophages, Bif4, 6, 10, 11, 12, 14, 16, 17, 20, 29, 30 and 40 were shortlisted for further studies (Table 1). These strains showed reduction in TNF-α, IL-1β and IL-6 production ( Table 2). B. longum Bif10, B. breve Bif11, B. longum Bif12 and B. longum Bif16, by virtue of having better reduction ability of inflammatory cytokines and probiotic potentials were short listed for detailed in vitro and in vivo studies.

Effect of various treatments on macrophage cells and determination of NO.
There was no reduction in NO production when the macrophages were incubated with killed (autoclaved) Bifidobacterium strains and LPS. Bacterial culture supernatants obtained from the live cultures of B. longum Bif10, B. breve Bif11, B. longum Bif12 and B. longum Bif16 did not inhibit LPS induced NO production by the macrophages.
Study on adhesion of LPS to the bacteria suggested that supernatants obtained from the mixtures of live cells of B. longum Bif10, B. breve Bif11, B. longum Bif12 and B. longum Bif16 and LPS led to 27.6, 25, 25, 35% of NO production by the macrophages while the live cell pellets from the mix caused 21, 26.4, 20 and 24% of NO production respectively relative to LPS alone treated macrophages (Table 3). Supernatants obtained from the mixtures of dead B. longum Bif10, B. breve Bif11, B. longum Bif12 and B. longum Bif16 and LPS caused 75, 69, 63 and 67% of NO production while the dead cell pellets caused 43, 24, 44 and 48.8% of NO production by the macrophages relative to LPS alone treated macrophages (Table 3).
Based on the above results, B. longum Bif10, B. breve Bif11, B. longum Bif12 and B. longum Bif16 were assessed for their protective role in DSS induced experimental colitis in Balb/c mice.
Bifidobacterium strains protected mice against DSS induced colonic inflammation. DSS treatment slightly decreased the body weight of mice (Fig. 1B); caused 30% mortality (Fig. 1C), decreased the colonic length and weight (Fig. 1D,E,F), spleen weight (Fig. 1G) and promoted DAI (Fig. 1H) relative to normal mice. Mice pre-fed with Bifidobacterium strains prior to DSS treatment showed no effect on body weight, no effect on colon length and weight relative to normal mice (Fig. 1B,D,E). Mice in B. breve Bif11 and B. longum Bif16 groups had lower spleen weight, while that in B. longum Bif10 group showed a slight decrease (Fig. 1G) relative to DSS group. Mice in the B. longum Bif10, B. breve Bif11, and B. longum Bif16 showed lower DAI relative to DSS group (Fig. 1H).
Bifidobacterium strains protected mice from DSS induced oxidative stress in colon. DSS treatment decreased the SOD activity and GSH levels; increased the MDA levels, while catalase activity remains unchanged relative to normal mice ( Fig. 2A-D). B. longum Bif10 + DSS; B. breve Bif11 + DSS and B. longum Bif16 + DSS fed mice had elevated levels of catalase and GSH, while MDA levels were reduced. SOD remained unchanged relative to DSS alone fed mice.
Bifidobacterium strains protected mice from DSS induced systemic inflammation. DSS treatment enhanced the faecal levels of lipocalin and systemic levels of TNF-α, IL-1β, IL-6 and LPS compared to normal mice, while B. longum Bif10 + DSS; B. breve Bif11 + DSS and B. longum Bif16 + DSS fed mice had lower levels of them relative to Table 1. Bifidobacterium strains used in the study and their effect on NO production upon co-treatment with LPS on murine macrophages. Nitric oxide production for untreated RAW cells was 0.85 ± 0.01 μm/ml and for LPS treated RAW cells was 2.3 ± 0.2 μm/ml. Data was analyzed using one-way ANOVA followed by Tukey's Post-hoc test (p ≤ 0.05). # Significant relative to LPS treatment. *Significant relative to untreated control; (N = 10).

Strain Source
NO production (%) www.nature.com/scientificreports/ DSS alone fed mice ( Fig. 2E-I). Importantly, per se supplementation of the strains did not promote their levels relative to normal mice.

RAW cells + Bif RAW cells + LPS + Bif
Bifidobacterium strains modulated DSS induced deleterious gene alterations in the colon. DSS feeding elevated the expression of proinflammatory cytokines such as TNF-α, IL-1β, IL-6 and Ccl5, while it had no effect on MUC-2 expression relative to control mice ( Fig. 3A-E). Colon homogenates showed elevated protein levels of TNF-α, IL-1β and IL-6 in the DSS treated mice (Fig. 3F-H) Bifidobacterium supplementation beneficially modulated the gut microbiota. Illumina paired-end sequencing of the V3-V4 region of the 16S rRNA of the faecal samples suggested that the gut microbiota of the DSS treated mice was different from that of the mice in the control group as determined by PCA (Fig. 4A). B. longum Bif10 + DSS and B. breve Bif11 + DSS supplemented mice groups clustered together and were partly close to the Table 3. Effect of supernatants obtained from live and dead bacteria and LPS mixes on nitric oxide production by the murine macrophages. Nitric oxide production by the RAW cells treated with LPS was 2.3 ± 0.2 μm/ml. Data was analyzed using one-way ANOVA followed by Tukey's Post-hoc test (p ≤ 0.05). # significant relative to LPS treatment (N = 5).  www.nature.com/scientificreports/ control group indicating that B. longum Bif10 and B. breve Bif11 supplementation could alleviate the metagenomic changes caused by DSS challenge. However, B. longum Bif16 + DSS supplemented mice clustered separately, indicating a very distinct microbial diversity than rest of the groups. Species richness, as determined by the Shannon index and the evenness measure by Gini-Simpson index were almost similar in all the groups (Fig. 4B, C). At the phylum level, a total of 13 phyla were identified in all the samples. The DSS supplementation shifted the microbiome towards higher abundance of Actinobacteria (p = 0.004), Firmicutes (p = 0.003) and Verrucomicrobia and lower abundance of Bacteroidetes as compared to normal mice (Fig. 4D). B. breve Bif11 + DSS and B. longum Bif16 + DSS feeding increased the abundance of Actinobacteria in the DSS supplemented mice (p = 0.01 and p = 0.005 respectively). The relative abundance of Defferibacteres and Proteobacteria in the B. longum Bif10 + DSS and B. breve Bif11 + DSS supplemented groups was decreased as compared to the DSS fed group and was comparable to the control group (Fig. 4D). The relative abundance of Verrucomicrobia in the B. breve Bif11 + DSS supplemented group was decreased as compared to the DSS treated group and comparable to the control group. However, the relative abundances of Verrucomicrobia and Firmicutes were increased in the B. longum Bif16 + DSS supplemented group as compared to both the DSS fed and control groups (Fig. 4D). Specifically, our results suggested that the B. breve Bif11 + DSS supplemented group followed by B. longum Bif10 + DSS group had the microbiota partly close to the normal mice, which otherwise perturbed by DSS treatment.

RAW cells + Pellet RAW cells + Supernatant RAW cells + Pellet RAW cells + Supernatant
At the genus level, DSS treated mice had higher abundances of Klebsiella, Haemophilus and Lachnospira (p < 0.05) relative to the control mice. B. longum Bif10 + DSS, B. breve Bif11 + DSS supplemented mice showed lower abundance of Klebsiella, Haemophilus and Lachnospira relative to DSS treated mice while the abundance of these bacteria in the B. longum Bif16 + DSS group were similar to that of the DSS group. Per se supplementation of B. longum Bif10 and B. breve Bif11 did not increase their abundances relative to control, while B. longum Bif16 supplementation increased their abundance to the levels of DSS group (Fig. S1).
Bifidobacterium strains prevented DSS induced damage to colonic architecture. Histological analysis of the colon suggested that DSS feeding affected the colonic integrity and architecture along with disruption of crypt structure ( Fig. 5A-E). Furthermore, the epithelial layer showed derangements with decreased number of goblet cells and higher number of infiltrated neutrophils upon DSS feeding. Sub-mucosal thickness was increased in DSS fed mice relative to normal mice. Supplementation with B. longum Bif10, B. breve Bif11 and B. longum Bif16 strains before DSS treatment prevented these harmful changes in the colon and conferred protection against colitis.   5F).
Bifidobacterium supplementation modulates the SCFA production. DSS treated mice had lower lactate, acetate, propionate and butyrate levels in the caecal content relative to the control mice (Fig. 6). Lactate, acetate, propion-  www.nature.com/scientificreports/ ate and butyrate were improved in the caecum of mice supplemented with B. longum Bif10, B. breve Bif11 and B. longum Bif16 strains (Fig. 6A-D). Lactate levels were higher in the mice pre-treated with Bif16 (Fig. 6A). Per se treatment with B. longum Bif16 showed decreased levels of lactate, acetate and butyrate while that of propionate remained almost the same relative to the normal mice. Per se supplementation with B. longum Bif10 and B. breve Bif11 did not affect the SCFA levels relative to normal mice, while lactate level was found to be low ( Fig. 6A-D).

Discussion
Many studies suggested that exogenous administration of Bifidobacterium strains improves the health, mainly due to improved barrier function, mucus production and immuno modulation besides the production of antimicrobial peptides, bacteriocins and bioactive exopolysaccharides 14,15 . Some of the in-vitro and in-vivo studies emphasized the anti-inflammatory potential of Bifidobacterium strains 16  Bacterial adhesion to LPS study indicated that live cells are necessary for efficient binding and elimination of LPS and thereby preventing it from binding to the Toll receptors, thus mitigating the inflammation. We have observed that supernatants from live cells + LPS mix could induce low level of NO by the macrophages, which could be due to the surface proteins on live bifidobacterial cells that could bind and eliminate LPS as suggested by Kondo et al. in Pediococcus pentosaceous AK-23 strain where a 217 kDa protein could bind and eliminate 38% of LPS 19 .
As B. longum Bif12 and B. longum Bif16 were isolated from the same stool sample and likely to be the same strains and since B. longum Bif16 showed better reduction in LPS induced IL-6 in addition to TNFα and IL1β, this was preferred over B. longum Bif12. Hence, three strains namely, B. longum Bif10, B. breve Bif11 and B. longum Bif16 were further evaluated in the in vivo studies.
Uncontrolled aggravation of intestinal epithelial cells causes mucosal disturbance and ulceration leading to IBD. In this study, clinical conditions of UC were reproducibly induced with 2.5% (W/V) DSS administration in drinking water for seven days, which is due to damage caused to the epithelial monolayer 20 . Supplementing B. longum Bif10, B. breve Bif11 and B. longum Bif16 strains to the mice protected them from colitis by alleviating crypt distortion, diarrhea, blood in the stool, reduction in colon length, neutrophil infiltration and overall improved survival rate of mice. DAI is also improved in mice fed with these strains. Our results are in alignment with that of earlier researchers where conjugated linoleic acid (CLA) producing B. breve and ropy EPS producing B. longum treatment protected the mice from DSS induced colitis 21 .
Elevated oxidative stress, due to an imbalance between ROS generation and anti-oxidant defense system, is one of the important events in acute colitis 22 . In this study, DSS treated mice had lower levels of SOD, catalase and GSH, while bifidobacteria pre-supplemented mice had higher levels of GSH and reduced levels of lipid peroxides suggesting protection from DSS associated oxidative stress induced intestinal mucosal injury. The intestinal epithelial cells are the first line of defense in any GIT disorders 23 . When the intestinal epithelial permeability dampens, invasion of bacterial associated pathogenic molecules such as LPS triggers an inflammatory response where antigen-presenting cells transform effector T-cells into Th1, Th2, Th17 and natural killer T-cells. This activates macrophages to produce TNF-α, IL-6, IL-1β and several other cytokines 24 . In our study, Bifidobacterium pre-feeding curtailed TNF-α, IL-6, IL-1β and LPS levels in the mice. High levels of fecal Lipocalin-2, a marker for colonic inflammation was also low in Bifidobacterium treated groups, suggesting protection from DSS-induced inflammation 25 .
Gene expression and protein levels of TNF-α, IL-6 and IL-1β were deleteriously altered in the colon of the DSS fed mice, as corroborated with earlier studies 21 . Supplementation with the three strains prevented the higher expression of these pro-inflammatory markers and neutrophil infiltration marker Ccl-5 relative to DSS treated mice. Ccl-5 is highly expressed in bacterial and viral infections and it recruits neutrophils to the site of inflammation 26 . B. longum Bif10 and B. breve Bif11 are more effective in preventing these harmful changes than B. longum Bif16. The mucous layer comprises of mucin and trefoil factor that obstructs the pathogen colonization in the gut. Hence, genes associated to mucus production play a critical role in gut associated infections. In this study, higher Muc-2 mRNA expression and higher number of goblet cells in the colon suggest that Bifidobacterium supplementation prevented mucosal damage, which otherwise was perturbed in colitis mice.
Harmonious relationship between the gut microbiota and the host is essential for healthy living. Dysbiosis in microbiota composition leads to aberrant mucosal immunity in the gut as observed in IBD patients 27 . We have observed dysbiosis in the gut bacteria of DSS treated mice, where at the phylum level the abundance of Bacteroidetes and Deferribacters was decreased with DSS treatment, which was similar to the observation by Scientific Reports | (2020) 10:18597 | https://doi.org/10.1038/s41598-020-75702-5 www.nature.com/scientificreports/ other researchers 28 . Actinobacteria levels were low in control than DSS treated mice but increased abundance was found in interventions with B. longum Bif10, B. longum Bif11 and B. longum Bif16 to the mice, suggesting gut modulating effects. Similar results were shown by Cao et al. 29 . The increase in actinobacteria abundance in the intervention groups might be due to exogenously administered Bifidobacterium strains in the respective groups. At the genus level, Klebsiella abundance was increased in DSS treated animals as reported by Zhang et al. 30 .
Ruminococcus was found in the gut of IBD patients indicating its role in the onset of colitis 31 . In our study too, increased abundance of Ruminococcus was observed in DSS treated mice. Members of Blautia, Methanobrevebacter, Odoribacter and Lactobacillus are the key bacteria involved in SCFA production in large intestine 30,32 . These bacteria are found in higher abundance in Bifidobacterium supplemented mice in this study. PCA, Shannon and Simpson indices suggested the diversity in species and its richness in gut flora between control and DSS as well as in other Bifidobacterium treated mice. Co-occurring pattern analysis at the phylum level suggested that Actinobacteria abundance was positively correlated with Firmicutes and Synergistetes. Over all, alterations in gut microbiota in DSS mice might have contributed to colitis, which is counteracted by supplementation with Bifidobacteria strains. The protective role of SCFA against colonic inflammation has been greatly appreciated by many researchers 33 . SCFAs enhances the gut barrier function and decrease tumor formation in the colon and hence promote colonic health 34 . In the present study, DSS reduced the levels of lactate, acetate, propionate and butyrate while the three Bifidobacterial strains are able to prevent the drop in their levels in the caecum content relative to DSS fed mice. An oral administration of mixture of acetate (67.5 mM), butyrate (40 mM) and propionate (25.9 mM) prevented tumor development and attenuated the colonic inflammation in a mouse model of Azoxymethane (AOM)/ Dextran Sodium Sulfate (DSS)-induced colitis-associated colorectal cancer 35 . Oral acetate administration has been shown to reduce inflammation in high sugar diet induced colitis in mice 36 . Supplementation with Bacillus coagulans MTCC5856 spores and prebiotic whole plant sugar cane fiber decreased the colitis by enhancing intestinal barrier function, reduction in oxidative stress, MPO production and decreased inflammation which in part was mediated through production of propionate and butyrate 37 . Sodium propionate administration through oral route reduced the symptoms of DSS-induced colitis by promoting intestinal barrier function and suppressing inflammation and oxidative stress through STAT3 pathway 36 . Butyrate is known to protect the intestinal tissues by promoting mucus production, decreasing inflammation, oxidative stress and exert epigenetic regulation on colonocytes 38 . The benefits of oral sodium butyrate in drinking water against TNBS induced colitis in mice suggested that butyrate enhanced the intestinal barrier function due to mucin production and reduction in inflammation 39 and in DSS induced colitis, it increased the expression of MUC2, MUC1, MUC3, and MUC4 40 . However, it is desirable to manipulate the levels of SCFA within the host rather than exogenous administration, which may be challenging for chronic usage and in this context the three strains identified through this study might be useful in modulating the SCFA levels for maintaining healthy intestinal milieu and immunomodulation.
Although, the three strains protected the mice from inflammation, B. longum Bif 16 showed slightly lesser effects in relative mRNA levels of TNFα, IL1β and IL-6 as well as having higher abundances of Haemophilus, Klebsiella and Lachnospira.
In conclusion, this study suggested the protective efficacy of B. longum Bif10 and B. breve Bif11 strains against DSS induced colonic inflammation, which might be due to the direct interaction of the strains with immune cells as demonstrated through in vitro studies as well as through SCFA production that interacts with immune cells and epithelial cells as evidenced in in vivo studies (Fig. 7). SCFAs produced might have contributed to immune modulation, mucin production and enhanced barrier function and by preventing translocation of the pathogen associated molecule, LPS, into the systemic circulation. Furthermore, the protective and therapeutic efficacy of these strains in preclinical models are ongoing.

Methods
Bacterial strains and culture conditions. Twenty-five Bifidobacterium strains isolated from fresh feces of human infants and adults on modified MRS agar containing 0.05% (w/v) each of L-cysteine hydrochloride and mupirocin were used in this study. This study was approved by institute Ethical committee of Post Graduate Institute of Medical Education and Research (PGIMER), Chandigarh (ethical approval number: P-641) and all protocols of experimentation and methods were performed in accordance with the relevant guidelines and regulations. Also, informed consent was obtained from both adults and parents of the infants. Active cultures of the strains were prepared from Trypticase soy broth glycerol stocks on MRSC plates and incubated at 37 °C for 48 h under anaerobic conditions. A single colony of each strain from the agar was sub-cultured three times consecutively in fresh MRSC broth at 37 °C for 24 h. Later, the cultures were centrifuged at 8000 rpm for 10 min at 4 °C and the pellets obtained were washed and resuspended in sterile PBSC. B. breve Bif40, isolated from VSL#3 (De Simone Formulation, manufactured by Sigma-Tau Pharmaceuticals and marketed by Sun pharma) was used as a reference strain in all the in-vitro experiments.
Screening for Bifidobacterium strains that could prevent LPS induced nitric oxide production by murine macrophage cells (RAW 264.7). RAW 264.7 cell line was purchased from National Centre for Cell Science, Pune, India and cultured in DMEM containing 10% fetal bovine serum and 1% penicillinstreptomycin at 37 °C under 5% CO 2 in a humidified incubator. The cells were passaged on every second day.
Treatment of macrophages with Bifidobacteria (Bif) and LPS. Confluent macrophage cells (1 × 10 5 cells) were treated with 1 μg/ml LPS and each isolated strain (OD 600 1 unit ≈ 1 × 10 10 CFU) for 16 h in a humidified incubator. The supernatants collected were used for the determination of nitric oxide production.
Scientific Reports | (2020) 10:18597 | https://doi.org/10.1038/s41598-020-75702-5 www.nature.com/scientificreports/ Viability assay. Confluent macrophages in a 96 well cell culture plate was treated with 1.0 OD 600 of each bacterial strain and incubated at 37 °C under 5% CO 2 in a humidified incubator. After 16 h, the supernatants were discarded and the viability of the cells was determined using MTT assay kit (Sigma-Aldrich, USA) as per the instructions.

Effect of various treatments on macrophages and determination of NO. Confluent macrophages
were treated with variously processed bacteria as described below, either in the presence or absence of LPS.
Live cells. Live bacterial cells of each strain (1 × 10 10 CFU ≈ 1.0 OD 600 ) were added to the macrophages and incubated for 16 h.
Dead cells. Freshly grown bacterial cultures for 24 h were autoclaved and the cells were harvested by centrifugation, added on the macrophage cells and incubated for 16 h as mentioned above.
Cell free bacterial culture supernatants. Supernatants from the bacterial cultures were obtained from the overnight grown cultures by centrifuging at 8000 rpm for 10 min at 4 °C, filter sterilized using 0.22 μm filters, lyophilized and 0.125, 0.250, 0.5 and 1 mg/ml in DMEM were prepared and added to the macrophages for 16 h as described above.
Adhesion of LPS to live and dead bacteria. This was determined by suspending fresh viable and dead bacterial cells (1.0 OD/ml) in DMEM having 1 μg/ml LPS and incubated at 37 °C under anaerobic conditions for 2 h. The contents were centrifuged and the supernatants and bacterial cells were added to the macrophages and incubated at 37 °C for 16 h under 5% CO 2 in a humidified atmosphere.  Experimental groups and treatment schedule. The duration of the experiment was 25 days and the schedule is given in Fig. 1A. Briefly, weight normalized mice were randomized into 8 groups as (1)  Induction of colitis and its prevention using Bifidobacterium strains. Colitis in the DSS group was induced by providing DSS (2.5% w/v) in sterile drinking water from the 11th to 18th day. From 19th day onwards, the mice were shifted to normal drinking water till the end of experiment.
Bacterial dose preparation and supplementation. Overnight grown bacterial cultures were centrifuged at 8000 rpm for 10 min at 4 °C and resuspended in PBSC after washing. A dose of 2 × 10 10 CFU/ml of bacteria was prepared in trypticase soy broth (TSB) with 15% glycerol and stored at − 80 °C. At the time of dosing, stored bacterial stocks were centrifuged and suspended in approximately 250 µl of PBSC and was gavaged to each mouse on every day from the beginning till the end of the experiment while DSS was provided ad-libitum from 11 to 18th day as per the schedule. Body weights of the animals and survival rate were monitored. Change in body weight was determined using the below formula where D s was body weight at a specific day and D 0 was body weight at day 0. Induction of inflammation in various groups was determined by quantifying fecal lipocalin-2 using ELISA kit as per the manufacturer instructions. A Disease activity index (DAI) was calculated using the data on body weight loss (0, < 1%; 1, 1-5%; 2, 5-10%; 3, 10-20%; 4, > 20%), stool consistency (0, normal; 2, loose; 4, diarrhea) and gross bleeding (0, normal color; 2, reddish color; 4, bloody stool) divided by 3 for each mouse 41 .
Evaluation of intestinal permeability. This was evaluated as per Chassaing et al. 25 . Briefly, on the day of sacrifice (25th day), three mice from each group were starved for 4 h. FITC-dextran (0.6 mg/g body weight; 4-40 KDa, Sigma-Aldrich Co., St. Louis, MO, USA) was gavaged to each animal and after 3 h serum was collected and fluorescence was measured at 488 and 520 nm as excitation and emission wavelengths respectively. Permeability was calculated as the amount of FITC-dextran diffused into the blood circulation from the intestine relative to the normal mice. Biochemical analysis. Blood was collected from retro-orbital plexus of mice and serum was separated and stored at − 80 °C till further analysis. TNF-α, IL-6, IL-1β and LPS levels were measured through commercially available ELISA kits as per the manufacturer's instructions.
At the end of the experiment, mice were sacrificed by cervical dislocation. Spleen, colon and colonic contents were harvested from each animal. Colon length, weight and spleen weight were determined.
Preparation of proximal colon tissue homogenate. Approximately 50 mg of proximal colon was taken to prepare 10% tissue homogenate in ice cold PBS containing protease inhibitor cocktail using a handheld homogenizer. The homogenate was centrifuged at 12,000 rpm for 15 min at 4 °C and supernatants collected were stored at -20 °C until further use. Protein content in the homogenate was determined by Bradford method. Malondialdehyde level was determined by adding 200 μl of tissue homogenate to 200 μl of 8.1% SDS, 1.5 ml of 20% acetic acid (pH 3.4), 1.5 ml of thiobarbituric acid (0.8%) and 600 μl of H 2 O in a test tube and boiled at 95 °C for 1 h in a water bath. The contents were centrifuged at 5000g for 10 min and the absorbance of the supernatants was measured at 532 nm. The concentration of MDA was expressed as mmol MDA/mg protein of the tissue homogenate. Catalase activity was determined as per Luck 42 . Briefly, 1.95 ml of phosphate buffer (0.05 M, pH 7.0), 1 ml of hydrogen peroxide (0.019 M) and 0.05 ml of colon tissue homogenate were mixed. Absorbance at 240 nm was measured for 2 min with 30/60 s intervals. Catalase activity was expressed as μmoles of H 2 O 2 decomposed per min per mg of protein. Superoxide dismutase was determined as per Kono 43 . Briefly, to 0.1 ml of tissue homogenate and 0.5 ml of hydroxylamine hydrochloride (20 mM, pH 6.0 adjusted with NaOH) solution, 2 ml of nitro blue tetrazolium (96 mM, dissolved in 50 mM sodium carbonate in 0.1 mM EDTA, pH 10.8) was added. The auto-oxidation of hydroxylamine was observed by measuring the absorbance at 560 nm for 2 min at 30/60 s intervals. Reduced Glutathione levels were determined as per Ellman method 44 . Briefly, an equal volumes of tissue homogenate and sulphosalicylic acid were mixed and kept on ice for 1 h, centrifuged at 4500 g for 10 min at Gene expression analysis in colon tissue. The list of genes and primer sequences used for RT-PCR are given in Table S1. Total RNA from colon tissue (50 mg) was extracted by Trizol method and quantified on nanodrop (Thermo fisher). cDNA was prepared by reverse transcription kit (Bio Rad, USA; catalogue number 1725035) as per manufacturer instructions. The conditions for RT-PCR were 95 °C for 5 min, 95 °C for 15 s, 60 °C for 30 s, 72 °C for 30 s and 72 °C for 5 min. ∆∆ct method was used to determine the relative fold changes in the expression levels of genes using β-actin as the house keeping gene.
Faecal microbiota analysis. Genomic DNA was extracted from 100 mg of freshly collected faeces using QIAmp Fast DNA stool mini kit (Qiagen) as per the manufacturer's instructions and quantified and stored at − 80 °C. The V3-V4 region of the 16S rRNA was sequenced using llumina Hiseq platform by a commercial company (NxGenBio Life Sciences, New Delhi, India). Reads obtained were quality checked using FastQC (version 0.11.5) on the basis of base quality and sequence quality score distribution. Reads with a threshold value of Q20 or more were used for the analysis. However, majority of the reads had a quality score of Q30 or more (99.9% accuracy). Paired reads were assembled using PEAR (Paired-End reAd mergeR). OTU processing and microbiome analysis were performed by QIIME 1.9.1 and normalization of OTU's was done by Picrust 1.3.3. Furthermore, Calypso was used to perform taxonomical classification and various analyses (Diversity, PCA, abundance plot, etc.).
Colon histology. Colons excised from the mice (n = 3) at the time of sacrifice were stored in 10% phosphate buffered formalin. A small piece of tissue was processed in a series of ethanol and xylene and embedded in molten paraffin. Colonic sections of approximately 5 μm thickness were cut from paraffin embedded tissue blocks using a microtome (Leica, Wetzlar, Germany). Slides were stained with hematoxylin and eosin for microscopic evaluation as per standard methods. On the basis of the architecture of colonic sections, scoring was done for each slide as described elsewhere 38 . Alcian blue staining was done on separate sections to assess the number of mucus producing goblet cells.
Short chain fatty acid (SCFA) analysis in caecum content. HPLC method was used for the analysis as described elsewhere 45 . Briefly, 50 mg of the caecal content was extracted with acidified water (pH 1 -2) by rigorous vortexing followed by incubation at room temperature for 10 min. Contents were centrifuged (6000 rpm, 20 min, 4 °C) and filtered using 0.2 µm nylon filters (Millipore, Millex-GN) and Isobutyric acid (5 µM) was included as an internal standard. HPLC system (Agilent 1260 infinity series, Agilent Technology, Singapore) equipped with a controller pump (Agilent, model no. G1311C), an auto sampler unit (model no. G1329B) and a diode array detector (DAD; model no. G1315D) set at 210 nm (peak width 2.5 Hz, band width 2 nm) fitted with anion exchange column (300 × 7.7 mm; 8 µm particle size) along with a guard column made with the same material was used. The column was equilibrated and eluted with a mobile phase comprising of formic acid (0.1% v/v) in Milli-Q water (Merck Millipore, 0.22 µm filtered, resistivity 18.1 -18.3 MΩ cm) at an isocratic flow rate of 0.6 mL/min at 60 °C for 30 min. A 20 µL of the caecal extract in five biological replicates and acetate, propionate and butyrate standards were injected and separated at 60 °C. Data acquired was processed with EZchrom elite software with the system suitability option installed. The results obtained were expressed as µM concentration of SCFAs per mg sample against the SCFA standards.