Zebrafish, including both AB wild-type line and fluorescence-labeled lines, including Tg(lyz:DsRED2), Tg(mpeg1:EGFP), Tg(rag2:DsRed), and Tg(lck:lck-eGFP), were purchased from China Zebrafish Resource Center (http://en.zfish.cn/) and were maintained according to standard protocols (37). After crossing neutrophil and macrophage labeled lines, Tg(lyz:DsRED2/mpeg1:EGFP) were obtained, then together with Tg(rag2:DsRed) and Tg(lck:lck-eGFP), three lines of Tg fish were prepared for further experiments. The use of animals in this study was approved by the Animal Research and Ethics Committees of the Institute of Hydrobiology, Chinese Academy of Sciences. All experiments were conducted following the guidelines of the committees.

The formulation of the experimental diets is shown in Table 1 . To model SBMIE in adult WT fish (3 mo), diets, as well as the 6-week feeding trial, followed a previously published method (4). For the SBMIE larva model, the feeding trials for both innate and adaptive immune cell imaging followed our recently published protocol (38). In brief, the innate model used 5 dpf larva of Tg(lyz:DsRED2/mpeg1:EGFP) to do a 4-day feeding trial with FM (negative control), SBM (positive control), and drug-included SBM diet (tested group). Afterwards, at 9 dpf, imaging of the hindgut was carried out. The adaptive model used 17 dpf larva of either Tg(rag2:DsRed) or Tg(lck:lck-eGFP) to do a 10-day feeding trial, and then larva at 27 dpf was conducted to imaging analysis. The sinomenine hydrochloride, which is the hydrochloride chemical form of sinomenine (39), was purchased from Xi’an Huilin Biotechnology Co., Ltd in China. Based on our previously researched effective concentration for preventing SBMIE, the dietary inclusion of sinomenine hydrochloride should be in the range of 15–60 ppm (40). Thus, in this study, 35 ppm sinomenine hydrochloride was included in the soybean meal diet to test if it could have any alleviating effect on SBMIE.

To check if sinomenine could potentially activate zebrafish α7nAchR signaling, PROCHECK server and ProSA web were applied to test the potential binding ability of sinomenine to zebrafish α7nAChR protein, following a previously described method (41, 42).

In adult zebrafish, the hindgut tissue, which is a third of the whole intestine from bend 2 to the anus (43), has been used for sampling. For gene expression analysis, including transcriptome, qPCR, and immunofluorescence (n = 6), the tissues used to isolate RNA for both transcriptomic and qPCR were cut into small pieces and soaked in a 10-fold volume of TRIzol (Invitrogen), and the tissues used to make frozen sections were soaked in a 10-fold volume of 4% PFA (Sigma). For all omics studies, the hindgut from one female and one male adult zebrafish were mixed for each sample (three repetitions) to avoid sex-biased differences for both transcriptome and 16S. The sampling of 16S sequencing analysis followed the method published in previous studies by Deng et al. (44). For the sampling of larvae, only 27 dpf larvae (n = 6) were sampled for HE staining at the meantime of imaging the adaptive immune cells.

To demonstrate the basic pathology of intestinal mucosa, both hematoxylin eosin (HE) staining and immunofluorescence were performed using hindgut tissue slices of 10 μm. Then, the zebrafish anti-CD4-1 antibody (GeneTex), which can immune-stain the T helper cell, was applied during immunofluorescence analysis to further illustrate the inflammatory aggregation or infiltration of lymphocytes in different intestinal layers or villi locations in the hindgut (4). The zebrafish anti-foxp3 antibody (GeneTex) was also used to identify the regulatory lymphocytes in intestinal mucosal folds (45). Meanwhile, DAPI was used parallelly to stain the nucleus.

For the hindgut’s RNA (n = 6), transcriptomic analysis was completed to reveal the gene expression profile systematically in adult fish from all groups, including the negative control (FM group), the positive control (SBM group), and the tested drug inclusion group (SN group). The procedure of the gene library preparation and sequencing for transcriptome followed previously published methods (46). Briefly, sequencing libraries were generated using NEBNext R UltraTM RNA Library Prep Kit for Illumina R (NEB, USA) following the manufacturer’s recommendations, and the library quality was assessed on the Agilent Bioanalyzer 2100 system. The library preparations were sequenced on an Illumina platform (NovaSeq 6000), and 150 bp paired-end reads were generated.

After the clean data were mapped onto the reference genome by HISAT2 (version 2.1.0) (47), the transcript was assembled using StringTie (Version V1.3.1c) (48). Salmon (version 0.12.0) (49) was used to calculate the gene expression. | log2FoldChange | > 1 and padj < 0.05 is considered to be differentially expressed genes (DEG), using DESeq2 1.24.0 (version) to analyze differential expression (50). For annotations, enrichment analysis of both GO terms and KEGG pathways was performed using clusterProfiler (version 3.12.0) (51), and p < 0.05 was considered to be a significant enrichment. In the case of the parameters used not being listed, default parameters were used.

Principally, the RNA extraction and qPCR analysis of the zebrafish intestinal tissues were done according to previously published procedures (4, 6). In brief, TRIzol (Invitrogen) was used to extract RNA, and then the isolated RNA was reverse-transcribed into cDNA using HiScript III 1^st Strand cDNA Synthesis Kit (Vazyme). In addition, all the RNA extraction was done immediately after sampling, or at the same time for sampling to do an omics study. Genes of interest, including both randomly selected DEGs and genes involved in α7nAChR, mediated anti-inflammatory function were validated for their expressional changes by qPCR. Specifically, genes of both intestinal pro-inflammatory factors (transcription factor nf-κb, cytokines il-1β, tnf-α, il17a/f3, and CD4) and anti-inflammatory factors (cytokines tgf-β1a, il-10, il22 and transcription factor foxp3a and foxp3b) have been checked for SN-induced α7nAChR-mediated anti-inflammatory effect in intestine. Meanwhile, rpl13a was used as the internal reference. The primers used to amplify these genes are listed in Table 2 .

In vivo imaging of immune cells was completed following previous studies of zebrafish larvae imaging (15, 28, 32), with some modified conditions, such as the incubating solution, exact time, and microscopy. Imaging of innate immune cells, including neutrophils (lyz labeled) and macrophages (mpeg labeled) in the hindgut (tail part) of Tg(lyz:DsRED2/mpeg1:EGFP) larvae (n = 10), was through referencing previously published methods (38). Larvae were fed with experimental diets (diets for FM, SBM, SB groups, which were crushed and passed through an 80-mesh sieve) between 5 to 8 dpf. At 9 dpf, the Tg(lyz:DsRED2/mpeg1:EGFP) larvae were anesthetized by MS222 and then embedded in 1% LMP Agarose (Invitrogen), dissolved in Danieau’s solution (52) to conduct imaging analysis. For imaging adaptive immune cells, including immature (rag2 labeled) and mature lymphocytes (lck labeled) in the hindgut (tail part) of Tg(rag2:DsRed) and Tg(lck:lck-eGFP) larvae (N = 10), we followed previously published methods (38). After feeding 12 days of Larval AP100 Diet (Zeigler) from 5 to 16 dpf, as well as 10 days of experimental diets from 17 dpf, at 27 dpf the larvae were anesthetized and embedded in 1% LMP Agarose (Invitrogen), dissolved in Danieau’s solution (52). The Tg(rag2:DsRed) or Tg(lck:lck-eGFP) larvae were observed under a Leica SP8 microscope. For the intestinal imaging of larvae, a composition of several stacks was merged to generate the picture for the whole tail part of the intestine.

The composition of bacterioplankton was analyzed using sequencing of 16S rRNA gene amplicons. DNAs from the six repetitions were extracted for each treated sample. Then the V3V4 regions of the 16S rRNA genes, which can yield accurate taxonomic information, were amplified with the primer set 338F (5’- ACTCCTACGGGAGGCAGCA-3’)/806R (5’-GGACTACHVGGGTWTCTAAT-3’). After the libraries were built, the paired-end 250-nucleotide reads were yielded using the Illumina NovaSeq platform, according to the manufacturer’s instructions. Raw sequences were then trimmed, quality filtered, denoised, merged, chimera and dereplicated using the DADA2 plugin (53). Reads with 100% nucleotide sequence identity across all samples were assigned to operational taxonomic units (OTUs), and taxonomy was assigned to Non-singleton amplicon sequence variants (ASVs) using the classify-sklearn naïve Bayes taxonomy classifier in the feature-classifier plugin (54) against the Greengenes reference database (version 13.8, http://greengenes.secondgenome.com/) (55).

The HE-stained intestinal villi, immunofluorescence signals, as well as imaging results of immune cells in larvae were quantified by Image J software. T-tests were used to assess differences for most parameters for staining and imaging results, with a false discovery rate adjusted p < 0.05 or 0.01. For the quantification of immunofluorescence signals, the quantitative analysis method followed a previously published study (6, 56) with the modification to calculate the positive cell NO. per villi, which equaled to that the account of number was divided by the average length of intestinal villi. Then, the data were processed using Excel and SPSS 22.0. After eliminating abnormal data by Grubbs’ test, one-way ANOVA was used to analyze the variance. The minimum significant difference method was used for multiple comparisons. Means ± Standard Deviation was used to generate bar. p < 0.05 was judged as significant difference, meanwhile p < 0.01 as extremely significant difference. The rearranged data from the pathological or immune analyses were applied to GraphPad Prism version 5.0 software for all graphs. For the correlation analysis to compare fold changes between results of qPCR and DEGs, data were processed in EXCEL. The plots and diagrams were displayed by ggplot2 (2.2.1) using R language.

The results of molecular docking showed that sinomenine can bind to the same binding pocket of zebrafish α7nAChR as the positive control Epibatidine, a typical agonist of α7nAChR ( Figure S1 ). In this pocket, sinomenine could form hydrogen bonds with several amino acids, including tryptophan, phenylalanine, and lysine. The binding energy of sinomenine to zebrafish α7nAChR was −29.62 kJ/mol, which was comparable to that of positive control Epibatidine (−37.15 kJ/mol).

The inclusion of sinomenine at 35 ppm significantly prevented pathological changes in the hindgut. The HE staining result demonstrated the relief of intestinal lesion at mucosal fold level upon sinomenine dietary inclusion, compared with the SBM group. Quantitative analyses of the histological parameters in the SBMIE adult model reflect that in the SN group the height of intestinal villi increased while the thickness of lamina propria (LP) decreased very significantly (p < 0.01) ( Figure 1A ). The increased height of intestinal villi (p < 0.01) in the SN group was revealed in the hindgut at 27 dpf in the larva model ( Figure 1B ), compared to the very shortened mucosal fold in the SB group. Specifically, in the LP layer of intestinal villi, the widened top with lymphocyte infiltration indicated by the blue hematoxylin-stained nuclei in SN group was reduced compared to that in the SBM group.

To reveal the mechanisms of the lymphomonocytes involved, T helper cells, which are at the nexus of the innate and adaptive arms of the immune system (57), have been checked for related protein expressions using immunofluorescent analysis. As the surface marker of T helper lymphocytes and expressed protein on certain populations of phagocytes in zebrafish (57), the immunofluorescence results of intestinal CD4-1 signals showed that the inflammatory aggregation in the widened LP layer of the hindgut, especially at the top of villi in the SBM group. While in the SN group, though that the positive cell NO. per villi of CD4 signals increased, there are fewer signals in the top of villi, but a large proportion of CD4⁺ cells were found at the base of the mucosal fold ( Figure 2A ).

Further, to reveal the Treg’s involvement, an immune-stained transcriptional factor reflected that there were brighter and more Foxp3⁺ cells in the SN group, and a substantial proportion of Foxp3⁺ cells was found at the base of the villi in the SN group, compared to SBM group ( Figure 2B ). In the SBM group, Foxp3⁺ cells were also found at the edge of basal plasmacytosis ( Figure 2B ). While it was interesting to notice many vasally located Foxp3⁺ cells in the SN group ( Figure S2 ). Considering the recovered length of intestinal villi, the positive cell NO. per villi of Foxp3 in the SN group was lower compared to the SBM group, yet still higher than the FM group ( Figure 2B ).

The raw data of current transcriptomic analysis have been submitted to the Genome Sequence Archive (GSA) database (http://gsa.big.ac.cn/index.jsp) with the BioProject identifier <PRJCA005917> and data ID <CRA004582>. The close correlation (R > 0.08) of fold changes between fold changes of qPCR result and DEGs ( Figure 3A and Table S1 ) proved the successful decoding of gene expression by the current transcriptomic analysis. Specifically, the immunoregulatory role of SN was reflected by the expression of pro-/anti-inflammatory factors ( Figure 3B ). The pro-inflammatory cytokine genes’ result ( Figure 3B , upper) showed that in the SN group tnf-α was very significantly downregulated compared to the SBM group, and even dropped to a similar level to the FM group. il-1β and tnf-α were downregulated with non-significant trends, yet, nf-κb, il17a/f3, and cd4 were almost not changed compared to the SBM group. The anti-inflammatory factor genes’ result ( Figure 3B , lower) showed that except for the not significantly upregulated tgf-β1a, il-10, il22, foxp3a, and foxp3b were significantly upregulated.

According to the enriched biological process GO terms of DEGs from a comparison between SN vs SBM groups (with details in Table S2 ), there were many immune cells or process-related terms ( Figure 4A and Table S3 ). For immune cells, there were “leukocyte cell-cell adhesion,” “negative regulation of lymphocyte activation,” “negative regulation of leukocyte activation,” “negative regulation of alpha-beta T cell proliferation,” “regulation of T cell migration,” “regulation of alpha-beta T cell proliferation,” “regulation of lymphocyte migration.” For the immune process, there were “regulation of immune effector process,” “positive regulation of production of molecular mediator of immune response,” “regulation of cytokine production involved in immune response,” and “cytokine production involved in immune response.” Meanwhile, among revealed KEGG pathways ( Figure 4B , Table S4 ), a high gene ratio existed in six pathways, which contained metabolic-related “glycolysis/gluconeogenesis” and “insulin signaling pathways.”

At a cellular level, for innate immune cells, sinomenine could significantly (p < 0.01) relieve inflammatory aggregation for both neutrophils (lyz⁺ cells) and macrophages (mpeg⁺ cells) ( Figures 5A, B ) in the hindgut of the larva. For adaptive immune cells, sinomenine could significantly (p < 0.01) inhibit the inflammatory aggregation and immune synapse formation of both rag2⁺ immature lymphocytes and lck⁺ mature T lymphocytes ( Figures 5C, D ). Specifically, the formation of immune synapses of mpeg⁺ and rag2⁺ cells in the SBM group could be inhibited in the SN group.

The identified OTU in the hindgut of adult zebrafish for FM, SBM, and SN groups were 8,434, 5,976, and 5,895, respectively. As indicated by the Venn diagram ( Figure 6A ), the overlapping between the sinomenine and FM groups covered 635 OTUs, while the overlapping between the sinomenine and SBM groups contained 471 OTUs. To estimate the abundance of taxa, currently revealed OTUs could be identified by a lot of phyla. Among the most abundant 10 phyla, compared to the SBM groups, in the SN group, the increased phyla included Proteobacteria, Actinabactieria, Acidobacteria, and Gemmatimonadetes, while decreased phyla included Firmicutes, Bacteroidetes, Fusobacteria, and TM7 ( Figure 6B ). Moreover, at a genus level ( Figure 6C ), among the 10 genera of most abundance, compared to SBM groups, in the SN group increased genera included Aeromonas, Cetobacterium, Rhodobacter, Prevotella, Pelomonas, and Aquabacterium, while decreased genera included Sphingomonas, Acinetobacter, and Shigella. The raw data of the current 16S sequencing analysis have been submitted to the Genome Sequence Archive (GSA) database (http://gsa.big.ac.cn/index.jsp) with the BioProject identifier <PRJCA005917> and data ID <CRA004611>.