Kaempferol Alleviates Murine Experimental Colitis by Restoring Gut Microbiota and Inhibiting the LPS-TLR4-NF-κB Axis

Kae (purity > 98%) and carboxymethyl cellulose sodium (CMC) were purchased from MeilunBio (Dalian, China). Antibodies against myeloid differentiation factor 88 (MyD88; Lot: 00091677), zonula occludens 1 (ZO-1; Lot: 10003932), occludin (Lot: 10004180) and claudin-1 (Lot: 00081584) were purchased from Proteintech (Wuhan, China). Anti-TLR4 antibody (Lot: E2913) was obtained from Santa Cruz (Santa Cruz Bio, Santa Cruz, USA). The antibody against phospho-NF-κB (p-NF-κB)-P65 (Lot: 16) was purchased from CST (Danvers, USA). The antibody against nucleotide oligomerization domain (NOD)-like receptor 3 (NLRP3; Lot: 080639650) was purchased from Novus Biologicals (San Diego, USA). Fluorescein isothiocyanate (FITC)-dextran (#FD4, average mol wt 3000–5000 Da) was purchased from Sigma-Aldrich (St. Louis, USA). DSS (36000–50000 Da) was obtained from MP Biochemicals (Santa Ana, USA).

Six-week-old female C57BL/6J mice (20 ± 2g) were purchased from SPF Animals Biotechnology (Beijing, China). Mice were maintained under standard laboratory conditions, as follows: 25°C ± 3°C, 53% ± 3% humidity, with a 12-h light/dark cycle and were allowed to adapt to this environment for 7 days with free access to food and water. The composition of the mouse diet is provided in. For experiments, 40 mice were randomly divided into four groups: normal control (NC), Kaempferol control (Kae), DSS-induced colitis (DSS), and Kaempferol treatment (DSS-Kae) groups (n = 10 per group). Mice in the NC and DSS groups were given vehicle (1% CMC) by gavage for 14 days, while those in the Kae and DSS-Kae groups were administered Kae (50 mg/kg/day, dissolved in 1% CMC) by gavage for 14 days. Between days 8 and 14, mice in the DSS and DSS-Kae groups received 3.5% (w/v) DSS in their drinking water. 1

On the 15th day of the experiment, mice were anesthetized by intraperitoneal injection with pentobarbital sodium (30 mg/kg), before samples of feces and blood were collected. At termination, all mice were euthanized by cervical dislocation. Colon samples were snap frozen in liquid nitrogen and then stored at −70°C until further study. This study was approved by the Animal Ethics Committee of Inner Mongolia Agriculture University [NO. (2020)077].

Fecal microbial transplantation (FMT) was achieved as described previously (18–20). Briefly, C57BL/6J female mice (n = 40; weight, 20 ± 2g) were randomly divided into four groups: F-NC, F-Kae, F-DSS, and F-DSS-Kae, and each group was designated to receive daily fresh fecal supernatant from NC, Kae, DSS, and DSS-Kae group donors, respectively. Feces from each group of mice were weighed and homogenized in saline (0.1 g/mL), the suspension centrifuged at 850 ×g, at 4°C for 5 min, and supernatants collected (>9.6 × 10⁹ CFU/mL). Recipient mice in the F-NC, F-Kae, F-DSS, and F-DSS-Kae groups were administered supernatants collected from corresponding donor mice by gavage for 14 days (10 mL/kg).The detailed co-housing experimental conditions are described in the Supplementary Material. All processes were performed under a sterile environment.

The DAI and histopathology were assessed in each group of mice. The DAI was determined by combining hematochezia, mucous stools, and body weight loss scores, and dividing them by 3 (21). Colon tissues were fixed in 4% paraformaldehyde fixation solution, dehydrated through an ethanol gradient, rendered transparent by immersion in xylene, and finally embedded in paraffin, from which 4 μm paraffin slices were sectioned and stained with hematoxylin and eosin (HE). Histological features of colon tissue samples were scored independently using a previously described method (22). Detailed scoring criteria for the DAI and histology are described in Supplementary Tables 2 and 3.

After dewaxing with xylene and an ethanol gradient, colon tissue sections were heated in sodium citrate solution for antigen retrieval, the endogenous peroxidase was inactivated by soaked in 3% hydrogen peroxide, phosphate buffered saline (PBS) was used to rinsed sections three times, and 5% normal goat serum was used to block nonspecific binding. Next, sections were incubated with primary antibodies (1:300) at 4°C overnight. After three washes with PBS, tissue sections were further incubated with biotin-tagged secondary antibodies at room temperature for 20 min. Sections were then stained with streptavidin-horseradish peroxidase, followed by the addition of the colorimetric substrate, diaminobenzidine (DAB, Beijing Solarbio Science & Technology Co., Ltd.). After staining, clearing, and dehydration, sections were sealed with gum rubber sealant. More than three visual fields were randomly selected for observation under the microscope. Quantitative analysis of the target proteins was conducted using Image J software (version 1.5.7, National Institutes of Health, USA). Protein expression intensity is expressed as positive protein integral optical density.

Fecal genomic DNA was extracted by DNeasy-PowerSoil Kit (Qiagen, Dusseldorf, Germany) following the standard procedure. DNA concentrations were determined using the NanoDrop 2000C Spectrophotometer and gel electrophoresis. Using genomic DNA as the template, specific primers (343F 5’-TACGGRAGGCAGCAG-3’ and 798R 5’-AGGGTATCTAATCCT-3’) with barcodes and Tks Gflex DNA Polymerase (TaKaRa, Beijing, China) were used to amplify the V3-V4 hypervariable regions of the bacterial 16S rRNA gene. The quantity and quality of amplicons were evaluated using the NanoDrop 2000C Spectrophotometer and gel electrophoresis. Subsequently, AMPure XP beads were used to purify the amplicons (Beckman Coulter, California, USA) and purified products formed the template for the next round of amplification. The quality of amplicons was again confirmed as described above, and products were purified once more using AMPure XP beads. After the second purification step, the Qubit-dsDNA analysis kit (Life Technologies, Waltham, USA) was used to quantify the amplicons. Pooled equal amounts of purified amplicons were sequenced on the Illumina Novaseq platform (Illumina, California, USA) to obtain raw data in FASTQ format.

Mice feces (50 mg) were transferred into a 2 mL microcentrifuge tube, and mixed with 500 μL of extraction solvent (methanol/water 4:1 ratio, v/v), 40 μL of internal standard solution (2-chloro-l-phenylalanine in methanol, 3 g/L), and ground at 60 Hz for 3 min. Samples were then added to 120 μL of chloroform and subjected to vigorous vortexing, followed by ultrasonic extraction for 20 min at 25°C. Supernatants were extracted after samples were centrifuged at 13,680 ×g, for 20 min at 4°C, dried at 25°C under vacuum for 30 min, and dissolved in 80 μL of 15 mg/mL methoxyamine hydrochloride in pyridine. The resulting products was vigorously vortexed for 10 min, and kept at room temperature for 80 min. To this mixture, 20 μL of n-hexane and 60 μL of BSTFA (containing 2% TMCS) were added, samples vortexed vigorously for 3 min, then derivatized at 65°C for 70 min, and analyzed by gas chromatography using the Agilent 7890 B System, coupled to the Agilent 5977 A MSD System (Agilent Technology, California, USA). Separation of derivatives was performed using a 30 mm× 0.25 mm× 0.25 µm DB-5MS fused silica capillary column (Agilent Technology, California, USA), using the standard protocol, under full-scan mode to detect mass spectrometry data (m/z 50–500). Quality control samples were injected regularly throughout the analysis process to check data reproducibility. Raw data were obtained in D format.

FASTQ format raw data were processed using Trimmomatic (version 0.39) and FLASH (version 1.2.11), to remove invalid bases (23, 24). Data were classified into multiple operational taxonomic units (OTUs) by Vsearch (version 2.15.1) according to similarity score (≥ 97%) (25). Representative sequences of every OTU were selected using QIIME software and compared with the Greengenes database (v201305) (26). Alpha-diversity were assessed using the Shannon, Simpson, and Chao1 indices in Mothur (version 1.44.3) and R language (27). Pairwise comparisons were analyzed by Wilcoxon rank sum test for alpha-diversity. To evaluate beta-diversity, principal component analysis (PCA), principal coordinates analysis (PCoA), and non-metric multidimensional scaling (NMDS) were computed using QIIME and R language. Galaxy LEfSe tools were used to conduct linear discriminant analysis effect size (LEfSe) analyses and the relative abundance of microbial taxa and (LDA) scores >2 were recorded (28). Different microbial taxa were distinguished by the non-parametric factorial Kruskal-Wallis sum-rank test. Raw metabolomic data were quality-filtered using Analysis Base File Converter (version 1.0), then representative data was searched against the untargeted GC-MS database (OEbio, Shanghai, China). After log10 transformation, partial least-squares discrimination analysis (PLS-DA) and orthogonal partial least-squares-discriminant analysis (OPLS-DA) were performed using a data matrix to visualize the metabolic differences between the DSS and DSS-Kae groups. Differential metabolites were analyzed by calculation of Pearson correlation coefficients. Kyoto encyclopedia of genes and genomes (KEGG) pathway enrichment analysis was conducted using tools available in the KEGG database (release 95.2) (www.kegg.jp↗).

Intestinal permeability was evaluated using a FITC-labelled-dextran method as described by Volynets et al. (29). Briefly, on the final day of the experiment, FITC–dextran (mol wt 3000–5000 kDa; 600 mg/kg) was administered to the mice by oral gavage 4 h before euthanization. Immediately before euthanization, blood was collected and heparinized. Plasma was separated by centrifugation at 12000 ×g for 10 min, at 4°C, and 200 µL of each sample was added to a 96-well black microplate. Fluorescence was read with Agilent Biotek Synergy H4 (Santa Clara, USA) at 485/528 nm wavelength.

Statistical analyses were performed using SPSS 22.0 software (New York, USA). Data are expressed as mean ± standard error of the mean (SEM). One-way analysis of variance (ANOVA) followed by Tukey post hoc analysis was applied when evaluating differences between two groups. P values < 0.05 were considered statistically significant.

The structure of Kae is shown in Figure 1A. According to the experimental treatment schedule (Figure 1B), mice in the DSS and DSS-Kae groups were supplied ad libitum with 3.5% (w/v) DSS in drinking water on days 7–14, while those in the NC and Kae groups had ad libitum access to untreated drinking water. Compared with the NC group, the DSS group had an obviously steeper DAI score curve, while the rise in the DSS-Kae group was slower (P < 0.05) (Figure 1C). Additionally, treatment with Kae led to an increased colon length, relative to the DSS group (P < 0.05) (Figures 1D, E). HE staining of colon tissue showed that DSS treatment caused severe enteric mucosal injury; however, all characteristic features were prevented by oral Kae supplementation (Figures 1F, G). These results indicated that Kae could alleviate the gross symptoms of DSS-induced colitis in mice and alleviated colonic injury.

Next, we assessed the effects of Kae administration on DSS-induced pro-inflammatory responses. Compared with the NC group, DSS triggered significantly increased serum IL-1, IL-6, and TNF-α levels. Conversely, Kae supplementation remarkably reversed this tendency (Figures 2A–C). DSS treatment significantly increased mRNA level expression of inflammatory factors, such as IL-1β, IL-6, TNF-α, COX-2, MCP-1, and iNOS, while inhibiting IL-10 expression (P < 0.05) (Figure 2D). Further, mRNA expression levels of the pattern recognition receptors, TLR4 and NLRP3, as well as the signaling molecules, MAPK1 and NF-κB, were also elevated in the DSS group. In contrast, gavage with Kae strongly inhibited the transcription of genes involved in colitis, while reversing the effects of DSS on IL-10 transcription (P < 0.05). In summary, Kae alleviated inflammation caused by UC, partly through upregulation of IL-10 transcription and downregulation of the expression of inflammation-associated genes.

To evaluate the protective effects of Kae administration on intestinal barrier integrity, fluorescence spectroscopy of ingested FITC–dextran was measure. qRT-PCR and immunohistochemistry were used to evaluate the expression levels of tight junctions (TJs) genes, including ZO-1, occludin, and claudin-1. The results showed that mice with DSS exposure had remarkably higher FITC levels, Kae pre-treatment revealed an improvement in intestinal permeability compared to DSS-only mice (P < 0.05) (Figure 3A). DSS challenge led to decreased levels of ZO-1, occludin, and claudin-1, whereas in mice pretreated with Kae, expression of ZO-1, occludin, and claudin-1 were similar to those of control untreated mice (Figures 3B–E). These data suggested that the anti-inflammation effects of Kae involved maintenance of intestinal barriers.

In this study, we performed 16S rRNA gene high-throughput sequencing to reveal the impact of Kae on the gut microbiota. Results from rarefaction curve analysis, Shannon index calculation, and Good’s coverage reflected that there was a sufficient sequencing depth to provide coverage of the majority of microflora diversity in each sample (Supplementary Figure S1). Generation of a Venn diagram showed that in all samples, total of 1473 OTUs were calculated, 312 and 300 were in the NC and Kae groups, respectively, while the DSS group had fewer OTUs (n = 74), compared with the DSS-Kae groups (n = 447) (Figure 4A).

Alpha-diversity analysis is mainly used to assess community diversity and richness. Compared with samples from the NC and Kae groups, samples from the DSS group had remarkably reduced Shannon and Simpson diversity indices, and a lower Chao1 richness estimates, whilst Kae treatment effectively prevented the DSS-induced decline in bacterial community diversity and richness (P < 0.05) (Figures 4B–D).

PCA, PCoA, and NMDS analyses were used to evaluate similarities and differences between samples and groups. Based on values generated using the weighted UniFrac algorithm, PCoA and NMDS analyses indicated that the gut microbiota composition of the NC group was unlike that of the DSS group in terms of Axis PCo-1 and NMDS-1. Moreover, the DSS-Kae group was clearly located on a separate branch from the DSS group, but close to the NC and Kae groups (Figures 4E–G). These results indicated that the gut microbiota community structure was maintained by administration of Kae.

As shown in Figure 5A, DSS challenge resulted in an obvious decrease in the abundance of Bacteroidetes and Firmicutes, while the abundance of Proteobacteria was increased. Kae supplementation prevented the decrease in Firmicutes/Bacteroidetes ratio (Figure 5B). LEfSe analysis showed that pathogenic bacteria, such as Proteobacteria, Gammaproteobacteria, and Enterobacteriaceae, had LDA scores >4 in the DSS group. On the contrary, the DSS-Kae group had higher scores for beneficial bacteria, such as Ruminococcaceae and Prevotellaceae (Figures 5C, D). The all-against-all algorithm within LEfSe demonstrated that, in the DSS group, Proteobacteria, Gammaproteobacteria, Enterobacteriales, Enterobacteriaceae, and Escherichia_Shigella species were remarkably increased, whereas Kae supplementation partially prevented increases in these bacteria (Figure 5E).

We performed untargeted metabolomics analysis to study the effect of Kae supplementation on the metabolic profiles of mice with colitis. PLS-DA and OPLS-DA analyses showed clearly distinguished between the metabolic profiles of mice in the DSS and DSS-Kae groups (Figures 6A, B), and demonstrated that the DSS-treated group formed a distinct metabolic cluster, separate from those of the NC and DSS-Kae groups. In addition, 14 metabolites showing differential abundance between the DSS and DSS-Kae groups were identified. Interestingly, we found that supplementation with Kae greatly increased the levels of D-fructose 2,6-bisphosphate, D-xylose, galactitol, lactose, and N-acetyl-5-hydroxytryptamine in DSS-treated mice (Figure 6C). Pathway enrichment analysis indicated that Kae elicited major alterations in metabolic pathways related to phenylalanine metabolism, galactose metabolism, and arginine and proline metabolism (Figure 6D). Taken together, these results demonstrated that the protective effects of Kae against UC were related to the regulation of microbial metabolites.

As shown in Figure 7A, the LPS content in serum was significantly increased in the DSS group, indicating that DSS triggered metabolic endotoxemia. Conversely, the level of LPS in the DSS-Kae group was remarkably lower, suggesting that Kae was able to alleviate DSS-induced endotoxemia. Immunohistochemistry analysis showed that Kae pretreatment inhibited DSS-induced expression of TLR4, MyD88, p-NF-κB-P65, and NLRP3 (Figure 7B). Therefore, the anti-colitis activity of Kae could be partly attributed to the inhibition of the LPS-TLR4-NF-κB inflammatory pathway.

To further assess the involvement of the gut microbiota and its metabolic products in the anti-inflammatory effects of Kae, FMT and co-housing experiments were established (Figure 8A, Supplementary Figure S2A). The results showed that FMT from DSS-Kae donor mice to DSS-challenged recipient mice significantly increased colon length, reduced DAI scores, and alleviated pathological features (Figures 8B–F). Further, serum IL-1β, IL-6, TNF-α, and LPS content were also decreased in the recipient group (P < 0.05) (Figures 8G–J). The results of co-housing experiments showed that the anti-colitis effects of Kae could be transferred among co-housed mice to some extent (Supplementary Figures S2B-I). These results further verified the beneficial effects of Kae against colitis were associated with its ability to regulate the microbiota.