Heat stress induced gut microbiota changes activate TLR4 / NF-κB signaling pathway contributing to inflammatory bowel disease in pigs

Background Various stressors alter the 'brain-gut axis' and contribute to many gastrointestinal disorders such as inflammatory bowel disease. However, the pathological mechanisms of these perturbations are poorly understood. Here, we report on the induction of intestinal inflammation in heat-stressed pigs and apply fecal microbiota transplantation from pigs to mice to elucidate the role of intestinal microbiota in TLR4/NF-κB signaling pathway activation. Results Twelve healthy adult pigs were subjected to heat stress (34 ± 1 °C and 75–85% relative humidity) and compared with 12 control pigs exposed to 25 ± 3 °C and the same humidity level. The pigs were sacrificed on days 1, 7, 14, and 21. Their colonic contents were collected for microbiome analyses and their colonic epithelia underwent transcriptome analyses. Pseudo-germ-free mice were gavaged with fecal bacteria from heat-stressed and control pigs on the collection days to determine whether gut microbiomes and immune responses resembling those in the pigs could be induced in the mice. Heat-stressed pigs presented with fever and diarrhea from day 7 and their colonic villus length, crypt depth/width, and goblet cell number were significantly lower than those of the controls. Their TLR4, TRAF6, and nuclear p65 were upregulated at the RNA and protein levels. Their proinflammatory IL-6, IL-8, and IL-17 were also upregulated. Colonic microflora composition in the heat-stressed pigs markedly differed from that of the controls. By day 14, the former presented with substantial increases in opportunistic pathogens such as Campylobacterales , Veillonellaceae , and Megasphaera . Intestinal lipopolysaccharide concentrations were higher in the heat-stressed than the control pigs. Mice administered fecal transplantation from heat-stressed pigs had a distinctly different colonic microflora composition from those receiving control pig fecal transplantation. Bacteroides were significantly diminished and Akkermansia were significantly augmented in mice administered feces from heat-stressed pigs for 14 d. In

the colonic tissues of mice given feces from heat-stressed pigs, the TLR4/NF-κB signaling pathway was activated and there was a mild inflammatory response. Conclusion Heat stress promotes changes in gut microflora composition which, in turn, activates the TLR4/NF-κB signaling pathway and causes inflammatory bowel disease in pigs.

Background
Inflammatory bowel disease (IBD) affects > 3.1 million people in the United States and 2.5 million in Europe annually. Its incidence is increasing worldwide, especially in East Asia and South Asia [1,2]. IBD characterized by chronic intestinal inflammation as well as extraintestinal symptoms triggered by genetic predisposition and environmental factors [3]. Various stressors alter the so-called 'brain-gut axis' and contribute to a broad array of gastrointestinal disorders including IBD. Hypo-and hyperthermia may aggravate IBD by an unknown mechanism. Furthermore, climate change is increasing the incidence and duration of summertime heat waves in many countries [4]. Mean temperature and relative humidity have risen by ≤ 10.1% and ≤ 1.8%, respectively. Studies have shown that heat stress (HS) induces intestinal mucosal inflammation and damage in pigs [5], chickens [6] , and rats [7].
Cell membranes constitute part of the innate mammalian immune system. They function as gatekeepers and express Toll-like receptors (TLR) in response to the presence of microbial pathogens including bacteria, viruses, and parasites. TLR4 recognizes lipopolysaccharides (LPS) which abound in the cell walls of Gram-negative bacteria, then a conformationally change and intracellular Toll-interleukin receptor domains containing adaptor molecules are recruited. These adaptors associate with the TLR4 cluster via homophilic interactions between receptor domains. Consequently, activated the nuclear factor kappa-light-chain-enhancer of activated B cells (NF-κB) protein complex and mitogen-activated protein kinase (MAPK), resulted various inflammatory cytokine produce 5 [8][9][10][11][12].
TLR4 occurs in the peripheral lymphatic endothelial cells of the small intestine and can be found on the surfaces of the intestinal epithelial cells in IBD patients [13]. An earlier study showed that the number of TLR4 Asp299Gly polymorphism alleles was significantly higher in IBD patients than healthy controls [14]. After the TLR4/NF-κB signaling pathway was inhibited in mice with LPS-induced colitis, colonic IL-6 and TNF-α were significantly downregulated and the mucosa gradually self-repaired [15]. Therefore, TLRs mediate NF-κB activation, participate in the inflammatory response, and cause changes in the IL-6 and IL-8 levels [16][17][18]. TLR4/MyD88/p65 is upregulated and the number of inflammatory cells is relatively high in the intestinal mucosae of patients with ulcerative colitis. Our previous studies showed that the expression levels of TLR4 and its alternative splicing variants were increased in pigs exposed to HS [19,20]. However, the role of TLR4 signaling pathway activation in heat stress-induced IBD remains unknown.
The mammalian gastrointestinal tract hosts ~10 14 microorganisms comprising 500-1,000 unique species and forming synergistic mutualisms with the host [21][22][23][24]. Co-evolution of gut microorganisms with their hosts has resulted in specialization of the roles of different microbes in digestion, nutrient utilization, toxin removal, pathogen protection, and regulation of the endocrine and immune systems [25][26][27]. A healthy intestinal microbe community is diverse, stable, resistant to minor changes, and resilient [28]. Human and mouse studies have shown that gut dysbiosis or disequilibrium of the microbial community is associated with various acute and chronic inflammatory conditions, bowel diseases, metabolic syndromes, and diabetes [29]. Gut dysbiosis and reduced gut microbial ecosystem complexity are common symptoms in patients with Crohn's disease or ulcerative colitis. Nevertheless, it is unknown whether these alterations are causes or consequences of these diseases [30]. It has been reported that fecal microbial 6 transplantation (FMT) successfully treats relapsed Clostridium difficile infections that are ineffective with antibiotics; FMT seems to be beneficial for some patients for other gastrointestinal diseases such as IBD, IBS, etc. , and its effects may through specific microorganisms or its active products to adjust the intestinal flora to treat diseases, but its mechanism of action needs to be explored [31].
Pigs and humans have anatomical, physiological, and immunological similarities. Thus, studies of the immunological mechanisms in pigs could be reproducible in humans.
Moreover, pigs have similar susceptibilities and clinical manifestations in response to pathogens causing certain human intestinal disorders [32,33]. Xiao et al. reported that the homology between human and pig microbiomes is low at the gene level but significantly higher at the level of Kyoto Encyclopedia of Genes and Genomes (KEGG) orthology functions, and greater similarity between human and pig microbiomes than that between human and mouse microbiomes. Furthermore, ~96% of the functional pathways described in the human gut microbiome resemble those of the pig [34]. The aim of this study was to elucidate the roles of intestinal microbial composition and the TLR4/NF-κB signaling pathway in IBD development in heat-stressed pigs. To this end, we conducted FMT from pigs to mice and examined the inter-species effects.

Clinical signs in pigs
The average forehead temperature of the HS group sharply increased on day 3 and reached 37.77 °C which was 1.03 °C higher than that for the control group. The average rectal temperature on day 3 was 40.46 °C which was 2.84 °C higher than that for the control group. By day 21, the average forehead temperature of the HS group had gradually declined but was still significantly higher than that for the control group. In the latter, the average rectal temperature was 0.88 °C higher than average forehead 7 temperature. For the HS group, however, the average temperature difference was 2.69 °C ( Fig. 1a-b). The weight gains in the HS group were 6.41%, 61.30% (P < 0.01), and 88.71% (P < 0.01) lower than those for the control group at days 7, 14, and 21, respectively (Fig.   1c). The Diarrhea Index for the control group was < 3 and the pigs showed no symptoms of diarrhea. In the HS group, though, diarrhea was observed on day 7 and gradually increased thereafter. The Diarrhea Index from days 8 to 19 was > 3. Diarrhea was most severe on days 11 and 15 and the Diarrhea Index scores were 5.5 and 5.6, respectively ( Fig. 1d).

Histopathology of colonic mucosa in pigs
Morphological observation of the colon revealed that the mucosal layer of the control group was intact and no epithelial cells were shed. In contrast, sustained HS caused epithelial cell sloughing, vasodilation, and mucosal hyperemia in the colonic intestinal epithelium from day 7 onwards (Fig. 2a). The crypt depth for the HS pigs became shallower and by day 14 it was significantly different (P < 0.05) from the controls (Fig.   2b). In the HS pigs, the number of goblet cells per unit area was markedly lower than that for the control pigs. The difference between groups was significant by day 7 (P < 0.05) (Fig. 2c). The number of immune cells between the epithelial cells was also greater in the HS than the control pigs.

Transcriptome of pig colonic tissue
A total of 727,469,990 raw reads were obtained from sample sequencing and 675,459,262 clean reads (Table 1) were generated by double-ended read splicing and filtration (Fig. 3).
Superposition of gene coverage reflects whether the sequences are evenly distributed.
The plot in Fig. 3a shows that the sequencing results for each sample were not biased. A 8 principal components analysis (PCA) showed strong similarity between samples and the differences between groups were small (Fig. 3b). A heatmap showed that the gene expression levels and their correlation coefficients between samples were high for the HS test (Fig. 3c). A differential expression volcanic plot (Fig. 3d)  were upregulated and 7,922 were downregulated (Fig. 3e). KEGG enrichment analysis disclosed that the top 20 signaling pathways of differential gene enrichment included IBD, IL-17, TLR, and cytokine-cytokine receptor interaction (Fig. 3f). A transcriptomic analysis showed that HS induces the TLR4/NF-κB signaling pathway and significantly upregulates TLR4, MyD88, NF-κB, IL-17α, and IL-8 ( Fig. 3g-i).

TLR4/NF-κB signaling pathway in pig colon
In the HS pig colons, TLR4 was significantly upregulated on day 1 and peaked by day 7 (P ≤ 0.05) at which time its expression level was 21.43% higher than it was at day 1 ( Fig. 4a,   b). There was no significant difference in MyD88 expression (Fig. 4a, c). For the HS group, TRAF6 was significantly upregulated on day 1 and peaked by day 7 (P ≤ 0.05) at which time its expression level was 12% higher than it was at day 1 (Fig. 4a, d). There were no changes in the expression levels of p65 in the nucleus or extranuclear cells of the control group. Nevertheless, in the HS pigs, nuclear p65 expression was markedly increased relative to the control group. The p65 expression was highest on day 1 and gradually decreased thereafter (Fig. 4a, e-f). For the control group, colonic mucosal cells whose nuclei were expressing p65 were mainly distributed near the side of the intestinal epithelial cells and the expression levels did not significantly (P>0.05) differ across time 9 points. For the HS pigs, p65 expression in the nuclei of the anal and cecal mucosal cells near the intestinal lumen increased in the early stages and decreased thereafter (Fig.   4g).
The fecal LPS concentrations in the HS pigs were significantly higher on days 1, 7, and 14 than they were for the controls (Fig. 4k). Western blot revealed that for the HS pigs, TLR4 expression was markedly elevated after treating IPEC-J2 cells with 10 μg mL -1 LPS for 3 h (Fig. 4k). The nuclear p65 markers were also significantly upregulated (P < 0.001) (Fig.   4m).

Pig colonic microbiome
The Illumina HiSeq and paired-end methods used to sequence the 16S and 18SrRNA constructs generated 7,249,898 read pairs. After double-ended read splicing and filtering, 5,362,329 clean tags were generated. The average number of clean tags per sample was 335,146 ( Table 2). The Shannon index plot disclosed that each sample curve was flat.
Thus, there was sufficient sequencing data and the number of OTU species would not increase with sequencing quantity (Fig. 5). Each sample had numerous OTU and was rich in species. Most species were detected in the samples (Fig. 5a). The rank abundance curves showed that the microbial species richness and the uniformity of species composition for the HS group on days 7, 14, and 21 were lower than those for the control group (Fig. 5b). The PCA indicated that the samples in the control group were highly similar and equidistant whereas those of the HS group had relatively greater similarity

FMT on mouse colonic microbiome
Following FMT, the paired-end method was used to sequence the 16S and 18S rRNA constructs. Sample sequencing generated 2,866,688 read pairs. After double-ended read splicing and filtering, 2,430,548 clean tags were generated and the average number of clean tags per sample was 62,322 (Table 3). The Shannon index plot disclosed that each sample curve was flat, there was adequate sequencing data, and the number of OTU species should not increase with sequencing quantity. Each sample presented with numerous OTU and abundant species. Most species were detected in the samples (Fig. 6a).
The rank abundance curve showed that the species richness and uniformity of species composition were lower in the experimental group than the control group (Fig. 6b). The Prevotellaceae. Verrucomicrobia and Akkermansia were increased in the HF7 and HF14 mice (Fig. 6f). Therefore, the composition of the intestinal microbiota in the HF mice had changed following FMT (Fig. 6g, h). ANOVA showed that the intergroup differences in the numbers of Akkermansia, Bacteroides, Faecalibaculum, Parasutterella, and uncultured_bacterium_f_Bacteroidales_S24-7_group were highly significant (P < 0.001) (Fig. 6i). The KEGG plot revealed relative differences in lipid metabolism and xenobiotic biodegradation and increases in signal transduction for HF7 and HF14 (Fig. 6j).

Histopathology of mouse colonic mucosa following FMT
After FMT, lymphocyte infiltration and tissue gaps were observed in the colonic mucosae of the HF1, HF7, and HF21 mice compared with the CF group (Fig. 7a). Goblet cell and mucosal epithelial shedding had comparatively decreased in the HF14 group (Fig. 7a, d).
Compared with the control, the colonic villi were significantly shorter in the H14 mice (P < 0.05) (Fig. 7b) and the intestinal muscle layer was significantly thinner (P < 0.05) (Fig.   7c).

Colon transcriptomes of mice administered FMT
We obtained a total of 944,841,602 raw reads from the sample sequencing and 932,528,258 clean reads from double-ended read splicing and filtration ( Table 4).
Superposition of gene coverage indicates whether the sequences obtained are evenly distributed within the gene. The plot presented no obvious biasing front. Thus, the sequencing results for each sample were not biased (Fig. 8a). The expression levels of 19,017 genes in the CF7 and HF7 mice were significantly (P < 0.001) different. There were 9,494 upregulated, 9,523 downregulated genes, and 36 significant (P < 0.001) differentially expressed genes (DEGs) between the HF7 and CF7 mice (Fig. 8b). There were 19,019 DEGs of which 9,275 were upregulated, 9,744 were downregulated and 61 were significantly (P < 0.05) differentially expressed in HF14 compared to CF14 (Fig. 8c).
Transcriptomics showed that FMT induced TLR4, MyD88, and NF-κB in the TLR4/NF-κB signaling pathway and they were significantly (P < 0.001) upregulated compared with the PBS group. However, there were no significant changes in the relative expression levels of the proinflammatory factors ( Fig. 8d-f). The TLR4/NF-κB signaling pathway in the HF14 mice was partially activated compared to that of the CF14 group (Fig. 8g). The KEGG enrichment analysis disclosed that the B-cell receptor signaling pathway, intestinal immune network for IgA production, NF-κB, and cytokine-cytokine receptor interactions 13 were significantly (P < 0.05) upregulated in the HF mice ( Fig. 8h-i).

TLR4/NF-κB signaling pathway in FMT mouse colon
TLR4 was significantly (P < 0.001) upregulated in the HF1 mice and its expression level was highest for the HF7 mice (Fig. 9a, b). There were no significantly (P > 0.05) changes in nuclear p65 expression in the CF groups. In contrast, nuclear p65 expression in the HF groups was significantly (P < 0.001) upregulated compared with the CF groups. The highest nuclear p65 expression levels were determined for the HF7, HF14, and HF21 mice However, recent evidence has indicated that chronic stress may be implicated in disease development [35]. Here, we established a chronic HS-induced IBD model and found that Gram-negative intestinal bacteria play essential roles in the intestinal inflammatory response mediated by the TLR4/NF-κB signaling pathway.
In pigs, heat stress diverts blood to the periphery and maximizes radiant heat dissipation.
Nevertheless, this response causes vasoconstriction in the gastrointestinal tract [36], hypoxia in the intestinal epithelium because of the reduced blood supply, reductions in nutrient flow, and compromised intestinal integrity and function [37]. Pearea et al. [38] found that exposure to 37 °C for 4 h can damage the intestinal mucosa of growing pigs. Other stressors such as psychosocial events in humans [35], cold stress [39], and restraint stress in mice [40] strongly induced inflammatory responses in the gut. Therefore, HSinduced vasoconstriction and hypoxia in the gastrointestinal tract do not fully explain IBD in pigs and another mechanism may be involved in this pathogenesis.
The gastrointestinal tract is tightly controlled by the immune and neuroendocrine systems. Stress mitigation significantly improved gut homeostasis in pigs [41,42].
Destruction of the epithelial layer and thinning of the protective mucus layer may create conditions conducive to bacterial invasion of the mucosae and the blood vessels [43][44][45][46].
Bacterial products such as LPS are recognized by intestinal epithelial cells (IEC) via TLR4.
TLR4 upregulation in the gut enhances the ability of the body to recognize antigens.
Nevertheless, sustained stress may trigger an inflammatory response [47]. Pearce et al. [48] showed that HS and reduced feed intake compromised intestinal lining integrity and increased endotoxin permeability in pigs. Here, TLR4 expression was upregulated in heat stressed pigs not only in protein level, but in transcript level, suggested that TLRs signal pathway may play an important role in intestinal inflammation induced by heat stress.
The nuclear transcription factor p65 induces inflammatory genes after it is translocated to the nucleolus. When rats were maintained at 42 °C, their p65 activity was significantly increased relative to a control group exposed to an ambient temperature of 25 ± 3 °C and their IL-1β was also dramatically upregulated [44]. Two hours of heat stress in pigs can significantly increase blood TNF-α levels and inhibit lysozyme production. Numerous macrophages accumulate at the inflammation site and produce the proinflammatory cytokines IL-6 and IL-8 that mobilize lymphocytes and other immunocytes in the inflammatory process [49,50]. Upregulation of proinflammatory cytokines triggers an acute phase reaction presenting with fever, anorexia, and hormonal and metabolic 15 changes [51]. Here, we found that IL-17 was also significantly upregulated in the HS pig colon. All of the aforementioned factors contribute to IBD development and progress, suggested that the TLR4/TRAF6/NF-κB signaling pathway plays a vital role in HS-induced IBD pathogenesis.
There are numerous microorganisms in the intestinal tract and the epithelial cells comprise the first-line protective barrier [52,53]. Here, diarrhea increased in intensity in HS pigs by day 14 when the numbers of opportunistic Gram-negative bacterial pathogens such as Campylobacterales, Veillonellaceae and Megasphaera increased in the intestinal microflora. Oxidative stress is associated with host heat or cold stress [54] and may generate reactive oxygen species (ROS). The bacteria residing on the colonic mucosa have relatively greater oxygen tolerance [55], it might favor the proliferation of aerotolerant phyla in the gut such as Actinobacteria and Proteobacteria. On the other hand, gut microbiota can directly or indirectly contribute to ROS production via the mucosal cells [56] . Helicobacter pylori generates ROS and induces neutrophils to produce them as well [57]. Helicobacter pylori also enhances nitric and nitrous oxide production by activating macrophages [58]. As there is an interaction between inflammation and microflora, further studies are required to discern whether the observed intestinal microflora dysbiosis in the HS pigs was triggered by inflammation.
The intestinal microbial composition changed in the HF mice after FMT pig feces administration. Relative to the untreated mice, the HF mice presented with fewer goblet cells, shorter intestinal villi, and thinner muscle layers. Therefore, an inflammatory response had developed in the intestines of the FMT-treated HF mice. Moreover, the TLR4/NF-κB signaling pathway was activated in both the H7 and H14 mice. Breitruck et al. [59] reported that the intestinal mucosal structure is damaged in mice with IBD. This injury is characterized by severe bleeding and IL-6 and IL-10 upregulation. However, in the present study, the proinflammatory factors were not upregulated in mice administered the FMT possibly because of large numbers of Akkermansia muciniphila in the mouse intestinal tract. This microbe is a Gram-negative, mucin-degrading bacterium residing in the human and animal gastrointestinal tract. It can induce IL-8 production in the intestinal cells to a level 100-fold higher than that promoted by Escherichia coli. Thus, A. muciniphila has relatively weak proinflammatory activity in intestinal epithelial cells [60]. Diao et al. reported that gut microbiota affected certain pig traits and transferred their phenotypes to mice receiving pig feces via a different mechanism [61]. The gut microbiota also influenced epithelial cell morphology and renewal rates, intestinal nutrient digestion and absorption, and the gut barrier [5,61].
In our study, the relationship between changes in pig intestinal microbes and activation of the TLR4/NF-κB pathway under heat stress was explained by FMT. It also shows the use of live microorganisms, such as FMT, has potential therapeutic value for the disease, but the risks involved must also be considered. FMT can restore intestinal flora diversity, treat diarrhea caused by Clostridium difficile, and avoid drug resistance caused by long-term use of antibiotics, but it may also spread or increase susceptibility to intestinal bacteriarelated diseases [62]. FMT alters bacteria composition and establishes trans-kingdom equilibrium between gut fungi, viruses and bacteria to promote the recovery of microbial homeostasis. FMT is not a one size fits all and studies are required to identify microbial components that have specific effects in patients with different diseases [63].

Conclusions
The present study showed that HS-induced intestinal dysbiosis disrupted gut microflora composition, increased the numbers of opportunistic, pathogenic Gram-negative bacteria, activated the TLR4/NF-κB signaling pathway, and promoted proinflammatory cytokine production. This process triggered an inflammatory response and compromised the intestinal barrier. This study enhanced our understanding of stress-induced IBD and the increase in diarrhea in mammals subjected to prolonged humid heat in the summertime.

Animals and management
The experimental protocols involving the management and care of pigs and mice were approved by the Animal Care and Use Committee of Guangdong Ocean University, Zhanjiang, China (Permit No. 206-1108).

Pig study
Twenty-four pigs (Luchuan sows × Duroc boars; 12 males and 12 females) each weighing 15 ± 2 kg were housed in two animal rooms at the Animal Hospital of Guangdong Ocean University, Zhanjiang, China. The pigs were randomly divided into two animal room groups. Each group had six males and six females of similar body weight. There were four replicates (collection dates) of three per group. The animals were maintained for 2 wks at 20 ± 2 °C and RH = 75-85% to acclimatize them to the environment. The photoperiod was maintained under 12-h light / 12-h darkness over the adaptation and trial periods.
Throughout the study, the pigs were fed a complete formula in the morning, afternoon, and evening with ~6-h intervals between feedings. Drinking water was freely available. To minimize acute heat stress, the animal facility was gradually warmed over a 7-d period.
The trial lasted 21 d. The control animals were subjected to 25 ± 3 °C and the HS animals were exposed to 34 ± 1 °C at 75-85% RH.

Mouse study
Pseudo-germ-free animals (SPF grade) were induced by feeding mice (n = 130) with a

Sample collection
All pigs were observed for diarrhea and weight gain. Fecal shape and color were recorded daily and scored [66] as indicated in Table 5. The Diarrhea Index was calculated according to the scores for six pigs sacrificed at each sampling time.
Diarrhea Index = sum of scores / 6 (1) Forehead and rectal temperatures were measured daily with a far-infrared thermometer.
At each sampling time (days 1, 7, 14, and 21), pigs from the control (n = 3) and treatment groups (n = 3) were electric shock to sacrificed. Colonic feces and colon tissue samples were collected and immediately stored at -80 °C until the subsequent cell culture studies, microbiome and transcriptomic analyses, and mouse FMT studies. After FMT administration, mice (n = 30 per collection day) were sacrificed on day 7. Blood, colonic feces, and colon tissue samples were collected and immediately stored at -80 °C until the subsequent analyses.

Statistics
The data were subjected to t-tests using SPSS v. 21.0 (IBM Corp., Armonk, NY, USA). Data are expressed as means ± standard deviation. P ≤ 0.05 indicates significant difference. P ≤ 0.01 indicates highly significant difference.

Morphological observations
Colonic tissue was fixed in buffered formalin (10% v/v) and stained with H&E for histopathological examination. Image-Pro Plus v. 6.0 (Media Cybernetics Inc., Silver Spring, USA) was used to measure villus height, crypt depth, and width [64]. Hydrated Image-Pro Plus v. 6.0 (Media Cybernetics Inc., USA) was used to evaluate the goblet cells per unit area in the colonic mucosa [65].

Western blot analysis
To measure the responses of the critical TLR4/NF-κB signaling pathway proteins in the colonic tissues, total protein was extracted with RIPA (radioimmunoprecipitation assay)

TLR4, MyD88, and TRAF6
Paraffin sections of the intestinal tissues were prepared. TLR4-, MyD88-, and TRAF6positive cells and p65 (NF-κB) activity were detected by immunohistochemistry (IHC). Data are expressed as means ± standard deviation. One-way ANOVA and Tukey's multiple comparisons test were performed on the data in SPSS v. 21.0 (IBM Corp., Armonk, NY, USA). P ≤ 0.05 indicates significant difference and P ≤ 0.01 indicates highly significant difference.

Transcriptome sequencing and analysis
An enzyme-free cryopreservation tube was pre-cooled in liquid nitrogen and colonic epithelium was quickly excised and cut into pieces similar in size to soybean granules.
RNase-free water was used to prepare 1 × PBS or saline, the tissue surface stains were quickly removed, and the surface liquid was absorbed and collected in the sequencing quality. Quality of the data after the quality control (reads) was compared against the reference genome to obtain mapped data (reads) for the subsequent analysis.
Quality of the sequencing comparison was also evaluated. Based on existing reference genomes, the mapped reads were assembled, spliced with Cufflinks, and compared against known transcripts, transcripts without annotation information, and functional annotations of potential new transcripts. Read counts for each sample gene/transcript were obtained using featureCounts alignment to genome data and annotation files.
Fragments per kilobases per million reads (FPKM; number of reads on one million bases per gene alignment) were calculated. The FPKM method eliminates the effects of differences in gene length and sequencing quantity on gene expression calculation. The calculated gene expression may then be used to compare gene expression differences among samples. After securing the number of read counts for the gene/transcript, analysis of differential gene/transcript expression between samples was performed in the multisample (≥ 2) project and the differentially expressed genes/transcripts were identified.

RNA extraction and cDNA synthesis
The synthetic primer sequences (BBI Life Sciences, Shanghai, China) are listed in Table 6. After removing genomic DNA, a reverse transcription reaction system was prepared on ice according to the instructions for the HiScript ® Q Select RT SuperMix for qPCR (+gDNA wiper) (Vazyme Biotech, Nanjing, China), The PCR program was 25 °C for 10 min, 50 °C for 30 min, and 85 °C for 5 min. After the reaction was completed, the cDNA fractionation apparatus was stored at -80 °C until later use.

Relative quantitative real-time RT-PCR
The RT-qPCR system was configured according to the instructions for the ChamQTM SYBR ® (v/v) fetal bovine serum (FBS), penicillin (100 U mL -l ), and streptomycin (100 U mL -l ) in a 25-cm, two-cell culture flask (Corning Inc., USA) stored in a humidified incubator at 37 °C and 5% CO 2 . When the cells grow to 80% confluence, discard the medium, wash with PBS 2-3 times, add 1ml Trypsin-EDTA (0.25%, Thermo Fisher, USA) for 3 minutes, and then add medium to terminate the treatment. After centrifugation at 1000 rpm for 5 min at room temperature, the supernatant was discarded, diluted with the medium in proportion, and transferred to a 24-well plate. After the cells grew to sub-confluence in 24-well plates, the culture medium was removed and the cells were washed twice with PBS.

In vitro LPS validation experiment
The IPEC-J2 cells were subjected to 10 μg mL -l LPS for 3 h. Western blot was run to detect TLR4 expression and p65 entry in the nucleus. One gram of each fecal sample was weighed out and washed thrice with PBS. The final fecal mass: PBS volume was 1:9. The fecal suspensions were atomized for 15 min with an ultrasonic cell pulverizer (Φ6 horn ultrasonic treatment, 1.5 s; interval, 2 s; power, 25%). After centrifugation at 5,000 × g