All experimental protocols received prior approval from the Biomedical Research Ethics Committee of Hunan Agricultural University (Approval No. 2023-51). Thirty-two 8-week-old male ICR mice were acclimatized in specific pathogen-free (SPF) environmental chambers maintained at 25 ± 3°C with 50 ± 5% relative humidity under standardized photoperiodic conditions (12-h light/dark cycle). Following a 7-day acclimatization period, mice were randomly allocated to four experimental groups (n = 8/group): vehicle group (CTRL); LPS-only group (LPS); LPS and ferulic acid treatment group (FA); LPS and N-Feruloylserotonin treatment group (NFS). Dietary composition and nutritional profiles are detailed in Supplementary Tables S1, S2. All experimental groups received the standard diet throughout the study. The acclimatization endpoint was designated as Day 0. Commencing on Day 1, the FA group received dietary supplementation of 200 mg/kg ferulic acid. From Day 22 onward, the NFS group was administered 5 mg/kg N-Feruloylserotonin via dietary incorporation. On the 27th day, the mice in the LPS, FA and NFS groups were intraperitoneally injected with 10 mg/kg LPS (from E. coli O55: B5, dissolved in water to prepare a stock solution of 10 mg/mL), and mice in the CTRL group were injected with the same volume of normal saline. Cervical dislocation was performed to euthanize the mice 24 h later. During the procedure, the operator grasped the base of the mouse’s tail firmly with the right hand and elevated the animal. The mouse was then positioned on a cage lid or other textured surface. The operator applied firm downward pressure to the head and cervical region using the thumb and index finger of the left hand. Concurrently, the tail base was grasped with the right hand and pulled sharply upward and backward. This action resulted in cervical dislocation, severing the connection between the spinal cord and the brainstem, leading to instantaneous death of the animal.

First, the jejunal intestinal slices were harvested immediately post-euthanasia, rinsed in ice-cold saline, and processed fixed in 4% formaldehyde for HE staining. Then, the fixed slices were subjected to gradient dehydration with ethanol. Next, the dehydrated slices were embedded with paraffin. After hematoxylin and eosin staining, the tissue damage of mice in the four groups was observed and analyzed by microscope. Specific methods can refer to previous study (Han et al., 2023).

The colon tissue was a separate aliquot was snap-frozen in liquid nitrogen and cryopreserved at −80°C for RNA extraction. The expression of tight junction protein genes, inflammatory cytokines, and specific marker molecules was detected by qPCR. After extracting RNA from mouse colon tissue, the concentration was measured using a spectrophotometer. Supplementary Table S3 is a list of primers (Gene-specific primer pairs for qPCR analysis were designed using NCBI, and synthesized commercially by Sangon Biotech). The reaction system includes “2 × SYBR Green Master Mix 10 μL, 0.5 μL of forward and reverse primers, 2 μL of cDNA template, ddH₂O supplemented to 20 μL,” and the reaction conditions are “pre-denaturation at 95°C for 3 min, followed by 40 cycles of 95°C for 15 s, 60°C for 30 s “. The specific methods and calculation formulas were referred to previous studies (Ren et al., 2014; Ma et al., 2019).

After collecting the colonic contents of mice, the microbial DNA was isolated, and its concentration and purity were tested. Then, the V3-V4 region of 16S rRNA was amplified by primers F (5′-ACTCCTACGGGAGGCAGCA-3′) and R (5′-GGACTACHVGGGTWTCTAAT-3′). High throughput sequencing was carried out on Illumina platform after the purification of PCR products. Using SILVA as the reference database, the Naive Bayes Classifier was used to classify the feature sequences. Then, the data is filtered, with the filtering condition is set to base greater than 50% and mass greater than 20. The clean tags are clustered into OTUs with similarity greater than 97%. In order to detect the composition of community and α diversity of microorganisms, RDP classifier (v.2.2) and mothur (version 1.33.3) were used, respectively. α diversity metrics is primarily evaluated using indices such as Shannon index, Simpson index, Chao1 index, ACE index, and PD-whole-tree index. LEfSe was used to identify the species with different abundances.

Blood samples were collected from the orbital sinus of mice using sterile technique. Ophthalmic forceps were employed to apply gentle pressure to the globe, facilitating the natural efflux of blood. The exudate was collected directly into pre-chilled EDTA-coated collection tubes. The blood samples were centrifuged for 10 min at the condition of 3,500 rpm, 4°C to collect the serum of mice. Then, LC–MS was performed in positive ion and negative ion modes respectively, and the raw data were obtained. This study employed a NMR platform. SIMCA14.1 (Umetrics, Umea, Sweden) was used to perform data statistics, with the serum metabolites were determined by comparing peak with metabolites in Human Metabolome Database. Pathway analysis, encompassing both enrichment analysis and topological analysis, was performed by integrating differential metabolites via the MetaboAnalyst 3.0 platform.

The purity, concentration and integrity of total RNA extracted from mice colonic tissue were detected. And the library was constructed as follows: The mRNA was randomly interrupted using a fragmentation buffer after the enrichment of mRNA. Then, the first cDNA chain and second strand were synthesized using mRNA as template. The purified double stranded cDNA was subjected to end repair, A-tail addition, and sequencing adapter connection, followed by fragment size selection using AMPure XP beads. Finally, the cDNA library was obtained by PCR enrichment. After the construction of the library, the Qsep400 high-throughput analysis system was used to detect the insert fragments of the library. In order to ensure the library quality, q-PCR was used to accurately quantify the effective concentration (>2 nM). After that, the sequencing was performed through Illumina NovaSeq6000 sequencing platform. After filtering to get clean data, the mapped data was obtained by sequence alignment with a specified reference genome. Finally, library quality evaluation, structure level analysis, differential expression analysis, gene function annotation and function enrichment were carried out. Finally, the library quality evaluation, differential expression analysis, gene function annotation and function enrichment were performed through BMKCloud (www.biocloud.net). Differentially expressed genes (DEGs) were analyzed by DESeq2 software. DEGs with |fold change| > 2 and Q value (adjusted p-value) < 0.05 were considered to be significantly different expressed genes.

The software SPSS 21 (SPSS, Inc., Chicago, United States), R and GraphPad were used to compare the differences among groups through one-way ANOVA with a Tukey HSD. The data were expressed as mean value ± standard deviation (mean ± SD). When p < 0.05, the difference between the groups was statistically significant.

Histopathological analysis of jejunal tissue revealed intact mucosal architecture with tightly arranged villi in control (CTRL) mice. In contrast, LPS-challenged mice exhibited significant intestinal damage characterized by villous atrophy and epithelial denudation. Notably, both ferulic acid (FA) and N-Feruloylserotonin (NFS) treatments ameliorated these structural alterations (Figure 1A). Expression profiles of tight junction proteins differed significantly across groups (Figures 1B–D). LPS administration markedly suppressed Claudin 1, Occludin and ZO-1 expression when compared with the CTRL group (p < 0.05). FA treatment partially restored expression levels, whereas NFS administration induced significant upregulation of all three genes (p < 0.05).

Pro-inflammatory cytokine IL-1β was significantly elevated in the LPS group (p < 0.05). Both compounds reduced IL-1β expression, though not statistically significantly (Figure 1E). Concurrently, LPS-induced iNOS expression, a downstream effector of proinflammatory signaling, was significantly attenuated by FA and NFS treatments (Figure 1F, p < 0.05). While CD206 expression showed a non-significant reduction following LPS exposure, both compounds demonstrated restorative trends (Figure 1G). Arg-1 expression was significantly suppressed in LPS-treated mice (p < 0.05), with both interventions normalizing expression to CTRL levels (Figure 1H).

Necroptosis-related gene expression in murine colonic tissue was assessed (Figures 1I–K). LPS challenge significantly upregulated Rip3 and Mlkl expression (p < 0.05), with a concomitant non-significant increase in Rip1. Both ferulic acid and N-Feruloylserotonin interventions significantly downregulated Mlkl expression (p < 0.05), and reduced expression trends for Rip1 and Rip3.

Intestinal microbial diversity was quantified across experimental groups using α diversity metrics. As shown in Figure 2A, the LPS group exhibited significant reductions in Shannon index, Simpson index, Chao1 index, ACE index, and PD-whole-tree index (p < 0.05), whereas FA and NFS interventions significantly elevated these diversity parameters (p < 0.05).

Taxonomic profiling at the phylum levelrevealed marked alterations (Figure 2B). Bacteroidetes was inhibited significantly in the LPS group, and it was promoted significantly in the FA and NFS groups (p < 0.05). Proteobacteria and Firmicutes in the four groups showed an opposite trend to that of Bacteroidetes, which were significantly promoted by LPS and significantly inhibited by ferulic acid and N-Feruloylserotonin (p < 0.05). Many microorganisms have changed in different groups at the class level. Among them, the abundance of Bacteroidia in LPS group was significantly decreased, and the abundance of Gammaproteobacteria was significantly increased when compared with the CTRL group (Figure 2C, p < 0.05). Ferulic acid and N-Feruloylserotonin restored them to the similar level with CTRL group. As for the order level (Figure 2D), Bacteroidales in the LPS group was significantly inhibited, the abundance of Saccharimonadales decreased slightly, while Enterobacterales was promoted significantly (p < 0.05). After treatment with ferulic acid, Bacteroidales and Saccharimonadales showed significant increases in abundance, and Enterobacterales was inhibited significantly (p < 0.05). After N-Feruloylserotonin treatment, the abundance of Bacteroidales increased significantly, and Enterobacterales was also inhibited significantly (p < 0.05). At family level, the abundance of Lachnospiraceae, Prevotellaceae and Ruminococcaceae in LPS group decreased significantly, and they were significantly promoted by N-Feruloylserotonin, with Prevotellaceae was also promoted significantly by ferulic acid (Figure 2E, p < 0.05). Moreover, Muribaculaceae, Bacteroidaceae, and Bifidobacteriaceae were slightly decreased in LPS group, while ferulic acid significantly promoted the growth of Muribaculaceae and Bifidobacteriaceae, and N-Feruloylserotonin significantly promoted the growth of Bacteroidaceae and Bifidobacteriaceae (p < 0.05). Butyricicoccus at the genus level was inhibited significantly in the LPS group and promoted significantly in the NFS group (Figure 2F, p < 0.05). Moreover, the abundance of Akkermansia, Roseburia, and Blautia in the LPS group also tended to decrease, while N-Feruloylserotonin significantly increased that of Blautia and Akkermansia (p < 0.05). The abundance of Bacillus and Escherichia Shigella in LPS group increased significantly, while ferulic acid and N-Feruloylserotonin significantly reduced their abundance (p < 0.05).

LEfSe analysis identified taxonomic biomarkers characteristic of each group (Figure 3A). The family Prevotellaceae, and Rikenellaceae, genus Alloprevotella and Alistipes were species with high abundance in the CTRL group. The phylum Firmicute and Proteobacteria, genus Escherichia Shigella were species with high abundance in the LPS group. The phylum Bacteroidetes, class Bacteroidia and Saccharimonadia, genus Parasutterella were species with high abundance in the FA group. The phylum Desulfobacterota, order Lachnospirales, genus Bacteroides were species with high abundance in the NFS group.

Functional gene enrichment analysis of colonic microbiomes revealed distinct pathway signatures (Figures 3B–D). The LPS vs CTRL comparison identified 9 significantly altered pathways, including microbial metabolism in diverse environments, biosynthesis of secondary metabolites, biosynthesis of antibiotics, two-component system, biosynthesis of amino acids, metabolic pathways, ABC transporters, purine metabolism, and ribosome. The significant differential enrichment pathways between FA and LPS groups were two-component system, microbial metabolism in diverse environments, biosynthesis of amino acids, metabolic pathways, biosynthesis of secondary metabolites, ribosome, biosynthesis of antibiotics, purine metabolism, ABC transporters, and carbon metabolism. The significant differential enrichment pathways between NFS and LPS groups were biosynthesis of secondary metabolites, biosynthesis of amino acids, metabolic pathways, biosynthesis of antibiotics, purine metabolism, ABC transporters, microbial metabolism in diverse environments, two-component system, ribosome, and carbon metabolism.

Orthogonal partial least squares discriminant analysis (OPLS-DA) was employed to evaluate metabolic profiles across four experimental groups under both positive and negative ionization modes (Figure 4). The OPLS-DA score plots demonstrated complete separation of sample clusters between distinct groups, indicating statistically significant intergroup metabolic disparities, while intragroup sample aggregation reflected high metabolic homogeneity within each cohort.

Volcano plot visualization was utilized to identify differentially abundant metabolites in three comparative analyses: CTRL vs LPS, FA vs LPS, and NFS vs LPS (Figures 5A–C). Pathway enrichment analysis of significant metabolic alterations revealed conserved and group-specific perturbations. In CTRL vs LPS comparisons, alpha-linolenic acid metabolism, steroid hormone biosynthesis, linoleic acid metabolism, and glycerophospholipid metabolism constituted the primary enriched pathways (Figure 5D). The FA vs LPS comparison yielded identical pathway signatures (alpha-linolenic acid, glycerophospholipid, steroid hormone, and linoleic acid metabolism) (Figure 5E). Notably, the NFS vs LPS analysis expanded the pathway repertoire to include vitamin B6 metabolism and sphingolipid metabolism in addition to the conserved pathways (Figure 5F).

A hierarchical clustering heatmap was constructed to visualize metabolites exhibiting significant differential expression (Figure 5G). Compared with the CTRL group, PE(22:1(11Z)/22:1(11Z)), DG(14:0/19:0/0:0), PG(12:0/12:0), Siroheme, Idebenone, Aliskiren, PG(P-16:0/12:0), Cer(d14:1(4E)/20:0(2OH)), Phylloquinone, CerP(d18:1/26:1(17Z)), and 1-Naphthalenesulfonic acid were decreased significantly in the LPS group (p < 0.05). In the FA group, the content of PE(22:1(11Z)/22:1(11Z)), PG(12:0/12:0), Siroheme, Aliskiren, PG(P-16:0/12:0), and Cer(d14:1(4E)/20:0(2OH)) increased significantly (p < 0.05). PE(20:0/22:1(13Z)) was also slightly decreased in the LPS group, and upregulated by ferulic acid. Compared with the CTRL group, 1-Naphthalenesulfonic acid and Solasodine were significantly down-regulated in the LPS group (p < 0.05). In the NFS group, the content of 1-Naphthalenesulfonic acid increased significantly (p < 0.05), with the content of Solasodine also increased slightly. In addition, (24R)-24-fluoro-1,25-dihydroxyvitamin D2 and PE(22:4(7Z,10Z,13Z,16Z)/17:0) also showed a downward trend in the LPS group and were slightly up-regulated by N-Feruloylserotonin.

Volcano plots were generated to visualize DEGs across experimental comparisons (Figures 6A–C). A total of 6,324 DEGs were identified in the CTRL vs LPS comparison, 3,517 DEGs in the FA vs LPS comparison, and 248 DEGs in the NFS vs LPS comparison. Pathway enrichment analysis of these DEGs revealed distinct functional signatures (Figures 6D–F). In the CTRL vs LPS groups, the DEGs were mainly enriched in the TNF signaling pathway, MAPK signaling pathway, focal adhesion, and MicroRNAs in cancer. The main differential enrichment pathways of FA vs LPS groups included MAPK signaling pathway, glycerolipid metabolism, fatty acid degradation, tight junction, and VEGF signaling pathway. As for the differentially expressed genes in the NFS and LPS groups, they were mainly enriched in TNF signaling pathway, insulin resistance, Toll and Imd signaling pathway, focal adhesion, and HIF-1 signaling pathway.

Cross-comparison analysis identified 123 DEGs co-regulated in both CTRL vs LPS and FA vs LPS comparisons (|log₂FC| > 2), comprising 80 genes upregulated in LPS but downregulated with FA treatment, and 43 genes exhibiting reciprocal regulation patterns (Supplementary Table S5). Among these co-regulated genes, Pik3cd, Prkca, Prkcb, Ccnd1, Cdkn1a, Plcg2, Rac2, H2-Oa, H2-DMb2, Tgfa, Cxcr4, Cxcl12, Il2rg, Pdgfc, Fgfr3, Ticam1, Irak4, Traf6, and Lat, which involved in antigen processing and presentation, PI3K-Akt signaling pathway, Rap1 signaling pathway, Ras signaling pathway, Th17 cell differentiation, chemokine signaling pathway, VEGF signaling pathway, ErbB signaling pathway, MAPK signaling pathway, NF-kappa B signaling pathway, intestinal immune network for IgA production, and cytokine-cytokine receptor interaction, may be the potential targets for ferulic acid to regulate intestinal inflammation (Figure 7).

A parallel analysis of CTRL vs LPS and NFS vs LPS comparisons yielded 38 co-regulated DEGs, with 29 genes exhibiting LPS-induced upregulation reversed by NFS treatment, and 9 genes demonstrating reciprocal regulation (Supplementary Table S6). Among these co-regulated genes, Cxcl10, Kdr, Areg, Cxcl9, Reln, and Ccl8, which involved in cytokine-cytokine receptor interaction, PI3K-Akt signaling pathway, focal adhesion, and MAPK signaling pathway, may be the potential targets for N-Feruloylserotonin to regulate intestinal inflammation (Figure 8).

Heatmaps were constructed to visualize expression patterns of shared co-regulated genes (Figure 9). Among them, Sprr1a, Trim15, Gm4841, Prr5l, and Ccl8 were up-regulated in LPS group and down-regulated in FA and NFS groups. Hbb-bs, Hbb-bt, Hba-a1, Cbx2, Reln, Rps2, and Kdr were down-regulated in LPS group and up-regulated in FA and NFS groups.