Xuehua Li 1,4, Shixian Hu2,3,4, Xiaodi Shen 1,4, Ruonan Zhang 1,4, Caiguang Liu2, Lin Xiao 2, Jinjiang Lin 1, Li Huang 1, Weitao He1, Xinyue Wang1, Lili Huang1, Qingzhu Zheng1, Luyao Wu1, Canhui Sun1, Zhenpeng Peng 1, Minhu Chen2, Ziping Li1, Rui Feng2, Yijun Zhu2,3, Yangdi Wang 1,5✉, Zhoulei Li 1,5✉, Ren Mao 2,5✉ & Shi-Ting Feng 1,5✉

Abstract

Intestinal fibrosis is the primary cause of disability in patients with Crohn’s disease (CD), yet effective therapeutic strategies are currently lacking. Here, we report a multiomics analysis of gut microbiota and fecal/blood metabolites of 278 CD patients and 28 healthy controls, identifying characteristic alterations in gut microbiota (e.g., Lachnospiraceae, Ruminococcaceae, Muribaculaceae, Saccharimonadales) and metabolites (e.g., L-aspartic acid, glutamine, ethylmethylacetic acid) in moderate-severe intestinal fibrosis. By integrating multiomics data with magnetic resonance enterography features, putative links between microbial metabolites and intestinal fibrosis-associated morphological alterations were established. These potential associations were mediated by specific combinations of amino acids (e.g., L-aspartic acid), primary bile acids, and glutamine. Finally, we provided causal evidence that L-aspartic acid aggravated intestinal fibrosis both in vitro and in vivo. Overall, we offer a biologically plausible explanation for the hypothesis that gut microbiota and its metabolites promote intestinal fibrosis in CD while also identifying potential targets for therapeutic trials.

Keywords Crohn’s Disease; Fibrosis; Gut Microbiota; Metabolites; Magnetic Resonance Enterography Subject Categories Digestive System; Microbiology, Virology & Host Pathogen Interaction https://doi.org/10.1038/s44321-024-00129-8 Received 25 February 2024; Revised 2 August 2024; Accepted 13 August 2024

Published online: 13 September 2024

Introduction

Crohn’s disease (CD) is a prevalent form of inflammatory bowel disease that affects millions of individuals worldwide (Kaplan, 2015; Torres et al,

1. 2017). Intestinal fibrosis, a common complication of CD, causes over

50% of patients to experience refractory stricture or penetrating diseases in this organ that contribute to disability and significant healthcare expenses over the course of the illness (D’Alessio et al, 2022). However, the mechanisms of fibrogenesis remain elusive, limiting the development of effective therapeutic strategies. The gut microbiota and its functional metabolites play crucial roles in the onset of CD (Lloyd-Price et al, 2019) and may also be involved in the development of intestinal fibrosis. While the associations of the microbiota and its metabolites in fibrogenesis have been extensively studied in other organs, such as the liver (Loomba et al, 2017; Lee et al, 2020; Kwan et al, 2022), it has not yet been thoroughly explored in the gut where fibrosis is only beginning to receive attention. In a limited number of animal studies, the gut microbiota and metabolites have been reported to exacerbate or ameliorate intestinal fibrosis (Jacob et al, 2018; Ellermann et al, 2019; Xu et al, 2023). However, such data are conspicuously absent from fibrosis studies on patients with CD, and whether these micro-factors can induce macroalterations in bowel morphology (which are often associated with clinical symptoms in patients and detectable by magnetic resonance enterography [MRE]) remains unknown. To address this knowledge gap, we studied the fecal microbiome, fecal and blood metabolomes, and MRE features of 278 CD patients with variable degrees of intestinal fibrosis stratified by magnetization transfer imaging (MTI), alongside analyzing the corresponding fecal and blood data from 28 healthy controls, to elucidate the correlation between the gut microbiota and metabolites with intestinal fibrosis. Subsequently, the causal relationship between one specific metabolite (L-aspartic acid) and intestinal fibrosis was further validated both in vitro and in vivo, providing potential novel targets for managing intestinal fibrosis in patients with CD.

Results

Microbial and metabolome traits identified by multiomics vary with the severity of intestinal fibrosis

To clarify the roles of gut microbiota and metabolites in the development of intestinal fibrosis, we first investigated specific

1Department of Radiology, The First Affiliated Hospital, Sun Yat-Sen University, 58 Zhongshan II Road, 510080 Guangzhou, People’s Republic of China. 2Department of

Gastroenterology, The First Affiliated Hospital, Sun Yat-Sen University, 58 Zhongshan II Road, 510080 Guangzhou, People’s Republic of China. 3Institute of Precision Medicine, The First Affiliated Hospital, Sun Yat-Sen University, 58 Zhongshan Road 2nd, 510080 Guangzhou, Guangdong, People’s Republic of China. 4These authors contributed equally: Xuehua Li, Shixian Hu, Xiaodi Shen, Ruonan Zhang. 5These authors contributed equally: Yangdi Wang, Zhoulei Li, Ren Mao, Shi-Ting Feng. ✉E-mail: [email protected]; [email protected]; [email protected]; [email protected]

microbial and metabolomic signatures associated with this condition. We recruited a derivation cohort consisting of 214 patients with CD and 28 healthy controls (HCs) (Table EV1). The patients were stratified according to the degree of intestinal fibrosis, which was determined based on MTI findings (a validated imaging technique for accurately assessing intestinal fibrosis and unaffected by concurrent inflammation (Li et al, 2018)), and categorized as having either non-mild bowel fibrosis (BF1, n = 93) or moderatesevere bowel fibrosis (BF2, n = 121) (see “Methods”; Fig. EV1A,B). The HCs were also clinically examined to rule out any intestinal diseases.

Then, to investigate the microbial and metabolomic differences between BF1, BF2, and HCs, we performed 16S rRNA sequencing of fecal samples and paired targeted metabolomics measurements on both fecal and blood samples (Datasets EV1–3). A gradual and significant decline in alpha diversity was observed from HCs to BF1 to BF2 (Fig. 1A; Appendix Table S1). Beta diversity also differed significantly between HCs and both BF1 and BF2, while no significant difference was observed between BF1 and BF2 (Fig. 1B).

To identify the gut microbes specific to fibrosis based on their relative abundance, we constructed and compared four linear regression models (Guo et al, 2023), adjusting for age, sex, body mass index (BMI), smoking status, and disease location (only in CD patients). Model 1 compared CD patients with HCs; model 2 compared BF1 patients with HCs; model 3 compared BF2 patients with HCs; and, model 4 compared BF1 patients with BF2 patients. All analyses were restricted to the genus level. Model 1 identified 67 genera with differential abundance in CD patients compared to HCs (FDR < 0.1). These included HC-enriched Alistipes, Roseburia and Faecalibacterium and CD-enriched Escherichia_Shigella, Ruminococcus__gnavus_group, and Streptococcus; these findings were consistent with previous reports (Schirmer et al, 2018; Schirmer

1. et al, 2019) (Dataset EV4). Model 2 identified 61 genera (Dataset EV5), while model 3 identified 71 genera (Dataset EV6). No significantly different microbes were observed in model 4. Nevertheless, seven genera were identified in model 3 that were not identified in models 1 or 2, indicating their specificity to BF2 rather than an enrichment in patients with CD (Fig. 1C; Table EV2, Appendix Table S1). Among these genera, the relative abundance of Romboutsia, Lachnospiraceae_NK4A136_group, Ruminococcus_2, Ruminococcaceae_UCG_004, Muribaculaceae, and Anaerotruncus was decreased, whereas that of Mogibacterium increased with fibrosis severity. Although most of these genera (e.g., Romboutsia, Lachnospiraceae_NK4A136_group, Ruminococcaceae_UCG_004, Anaerotruncus, and Muribaculaceae) have been reported to exhibit similar alterations in studies on fibrosis in other organs (Lee et al, 2020; Albhaisi et al, 2021; Lee et al, 2021; Yu et al, 2021), the relationship between these bacteria and intestinal fibrosis was revealed for the first time in this study.

Similar to the results regarding microbial genera, the differences in fecal and blood metabolites between the BF1 and BF2 groups were not as pronounced as those observed between CD patients and HCs. Nevertheless, linear regression models revealed 1 fecal and 12 blood metabolites that exhibited significant abundance specifically in BF2 patients after correcting for potential covariates (see “Methods”) (FDR < 0.1; Fig. 1D,E; Tables EV3 and 4, Appendix

1. Table S1). To our best knowledge, there have been no reports on the association between Xylose and intestinal fibrosis, requiring further investigation. Notably, glutamine levels in blood samples

exhibited a significant decrease specifically in the BF2 group. We therefore speculated that glutamine may have protected patients with CD from fibrotic complications, consistent with a previous study showing that glutamine supplementation prevented fibrosis development in rats (San-Miguel et al, 2010). Conversely, serine induces intestinal fibrosis, similar to its effect on pulmonary fibrosis, and may exert its profibrotic effects by inducing fibroblast proliferation (Bernard, 2018).

Collectively, these results indicated that the severity of intestinal fibrosis was associated with alterations in the gut ecosystem. We revealed both potentially harmful and beneficial microbial and metabolomic features, suggesting their potential for diagnosing CD behaviors and warranting further analysis.

Omics-based machine learning models reveal biomarkers for characterizing intestinal fibrosis

Because the identified BF2-specific microbial and metabolomic signatures indicated the potential for noninvasive discrimination of intestinal fibrosis severity using these signatures, we next investigated the performance of six machine learning approaches in distinguishing between CD patients and HCs as well as between BF1 and BF2 patients. The tested approaches included decision tree, random forest, gradient-boosted decision tree, extra tree, logistic regression and support vector machine methods. Five models were incorporated in each analysis where the predictors were clinical features (age, sex, BMI, smoking, disease location), gut microbiota, fecal metabolites, blood metabolites, respectively, and all features combined. The first four models were defined as single molecular models, while the last model was a combined model. These analyses were performed on a cohort of 152 patients based on all of the obtained multiomics data.

Overall, all machine learning approaches exhibited satisfactory performance in distinguishing CD patients from HCs (Tables EV5 and EV6). For distinguishing BF2 patients from BF1 patients, the random forest and gradient-boosted decision tree methods were the most effective machine-learning approaches (Table EV7). Using the random forest method, blood metabolites were identified as the superior single molecular classifier for distinguishing BF2 patients from BF1 patients compared to clinical features, gut microbiota, and fecal metabolites, although the performance of the combined model surpassed that of any single molecular model (Fig. 2A; Table EV8). When comparing the selected features across all the machine learning models, they consistently showed good performance in at least five of the methods. Three genera (Ruminococcaceae_UCG_003, Solobacterium, Erysipelatoclostridium), two fecal metabolites (linoleylcarnitine and glycoursodeoxycholic acid (GUDCA)) and six blood metabolites (citramalic acid, PE.36:2, PE.36:1, PE.36:0, PE.34:2, DAG.36:5) were finally selected (Fig. 2B). Moreover, to assess the generalizability of the combined random forest model with these 11 predictors, we applied it to an external test cohort consisting of data from 64 patients with CD (BF1, n = 20; BF2, n = 44; Table EV1) using identical recruitment criteria as the derivation cohort and found that it also demonstrated robust performance in this cohort (Fig. 2C; Table EV9).

Taken together, we further identified several robust microbial and metabolomics features as potential biomarkers of moderatesevere intestinal fibrosis. We also provided a classifier (Computer

Code EV1) for distinguishing the severity of intestinal fibrosis in patients with CD, filling a gap in the lack of microbial and metabolomics tools in this field.

Microbes and their metabolites are associated with fibrosis-related luminal and extraluminal morphological alterations

Intestinal fibrosis usually leads to luminal and extraluminal morphological alterations that can be clearly detected by MRE (an imaging tool widely used in clinical settings for evaluating bowel lesions and their response to treatment) (Bruining et al,

1. 2018). Our effective delineation of the distinctive microecological characteristics of intestinal fibrosis prompted us to investigate the

potential correlations of the identified key gut microbes and metabolites with intestinal fibrosis-associated MRE features. To investigate these correlations, we first comprehensively evaluated the MRE findings of intestinal lesions (Bruining et al, 2018), regardless of whether or not they had previously been linked to fibrosis. These MRE features included eight luminal (i.e., stricture, penetration, wall thickening, mural edema, mural enhancement degree, mural enhancement pattern, length of diseased bowel, and mural apparent diffusion coefficient) and four extraluminal (i.e., perienteric effusion, comb sign, adenopathy, and perianal diseases) features (Table EV10). From this analysis, we identified four significant MRE features (stricture, penetration, effusion and comb sign) and conducted a meta-analysis, as BF1 and BF2 patients showed distinct multiomics patterns. During this analysis, age, sex,

BMI, smoking status and disease location were corrected to avoid potential confounding effects. In total, 215 significant signals that presented consistent molecular trait-MRE correlations across BF1 and BF2 patients were identified (FDR < 0.1, heterogeneity test P value > 0.05), including nine bacterial associations, 20 fecal metabolic associations and 186 blood metabolic associations (Fig. 3A–D; Dataset EV7).

Among the four significant MRE features, the severity of bowel stricture and penetration were correlated with changed abundance of the genera Escherichia_Shigella, Saccharimonadales and Muribaculaceae (Fig. 3B). The identification of Muribaculaceae in this case overlapped with our previous results from the analysis of microbiota specific to BF2 patients. Stricture and penetration are the imaging phenotypes most associated with intestinal fibrosis and lead to surgical resection and significant healthcare costs (D’Alessio

1. et al, 2022). Therefore, they are a top concern for clinicians treating patients with CD, highlighting the urgent need to explore these relationships. A cohort study of pediatric CD patients implicated Ruminococcus in stricture and Veillonella in penetrating diseases (Kugathasan et al, 2017). Here, we revealed a number of gut microorganisms associated with intestinal stenosis and penetration that might be specific to adult patients with CD. Furthermore, blood and fecal metabolites showed more associations with these two imaging phenotypes than did the gut microbiota. Notably, the level of microbiota-derived indole-3-propionic acid (IPA) in blood was decreased in the severe stricture group (Dataset EV7). The administration of IPA has been reported to strengthen intestinal homeostasis through IL-10 signaling (Alexeev et al, 2018) or shape the intestinal mesenchyme via pregnane X receptor signaling (Flannigan et al, 2023), making it a promising new therapy for intestinal inflammation and fibrosis.
2. et al, 2023). These results once again highlighted the intricate pathogenic mechanism underlying intestinal fibrosis, involving a complex interplay of luminal, transmural, and extraluminal factors.

The associations between molecular traits and MRE features exerted only in BF1 or BF2 group (heterogeneity test P value < 0.05) (Dataset EV8). In particular, certain blood amino acids and their derivatives, such as L-aspartic acid, 4-hydroxyproline and methylcysteine, were significantly associated with MRE features only in BF2 patients and not in BF1 patients (Fig. 3E–H). Moreover, the short-chain fatty acid, namely ethylmethylacetic acid (a-methyl butyrate), exhibited a negative correlation with the penetration sign in BF2 patients, suggesting its potential as a protective factor against disease progression (Ma et al, 2023).

Taken together, these findings suggested that the gut microbiota, along with fecal and blood metabolites, may contribute to both luminal and extraluminal morphological changes associated with intestinal fibrosis. However, some of these associations were stronger in patients with BF2, indicating a fibrosis degreedependent effect on relationships between microecological characteristics and MRE features.

Microbial metabolites as potential key actors in intestinal fibrosis

Since fecal metabolites have been shown to serve as functional readouts of the gut microbiota, we conducted mediation analysis to establish putative causal links from the gut microbiota to fibrosisassociated morphological changes mediated by small molecules. Briefly, four models in which bacteria were considered the exposure, MRE features the outcome and blood or fecal metabolites the potential mediators were used to assess the data of BF1 and BF2 patients (see “Methods”).

Not surprisingly, a larger number of significant mediation effects were observed among fecal molecules than among blood molecules (FDRACME < 0.1; Figs. 4A,B and EV2A,B; Datasets EV9–12). Strikingly, a major group of genera belonging to Lachnospiraceae, including Lactobacillus, Roseburia, Marvinbryantia, Eubacterium ventriosum and Eubacterium hallii, showed potential protective effects against stricture, penetration and perienteric effusion in BF2 patients through fecal metabolites (Dataset EV10). All of these bacteria have been previously reported to be beneficial to intestinal function (Lavelle and Sokol, 2020). We also identified an anti-inflammatory immunonutrient (glycine) that is partially derived from food digestion by the gut microbiota as a putative mediator of the negative associations between Lachnospiraceae groups and intestinal stricture (Fig. 4A).

Importantly, when we examined the overlap between the small molecules identified in blood and fecal samples, we observed that Laspartic acid and its derivatives (e.g., N-acetyl aspartic acid), which might be metabolized by Granulicatella, Dubosiella and Saccharimonadales, were positively associated with the presence of stricture, penetration and perienteric effusion based on a more lenient threshold (all PACME < 0.05) (Fig. 4C; Tables Datasets EV9–12). Moreover, metagenomics sequencing of a subset of fecal samples obtained from 71 participants (including 47 CD patients and 24 HCs randomly selected from the cohorts in this study) showed that the predicted microbial pathway was negatively associated with blood aspartic acid levels (alanine, aspartate and glutamate metabolism pathways [KO00250], P = 0.022, r = −0.32); We also observed a similar negative association between these two in fecal samples, although without statistical significance (P = 0.400, r = −0.12) (Fig. 4D,E). These findings suggested that L-aspartic acid, which was associated with microbial metabolic activities, may play a pivotal pathogenic role in the luminal and extraluminal morphological alterations. We performed additional analyses, including a comparison of L-aspartic acid levels among different inflammatory groups and a multiple linear regression analysis, to confirm that this relationship between L-aspartic acid and intestinal fibrosis was not significantly biased by coexisting intestinal inflammation (see “Methods”; Fig. EV3). L-aspartic acid is a nonessential amino acid that serves as a substrate for a variety of biosynthesis pathways, including essential amino acids and

enzymes (Caspi et al, 2020). Elevated levels of L-aspartic acid have been detected in the blood of patients with liver fibrosis (Vannucchi et al, 1985), and it has been reported to be a crucial factor in driving myocardial fibrosis (Ritterhoff et al, 2020). However, its role in intestinal fibrosis remains unclear.

Based on all of these results, we proposed potential connections whereby microbiota-derived metabolites induce intestinal fibrosis and subsequently result in MRE-detectable intestinal morphological alterations. Among these links, further investigation into the role of L-aspartic acid in promoting intestinal fibrosis is warranted,

1. as it has been implicated in driving fibrosis in the liver and heart (Vannucchi et al, 1985; Ritterhoff et al, 2020).

L-aspartic acid promotes the progression of intestinal fibrosis in vitro and in vivo

To investigate the modulatory effect of L-aspartic acid on intestinal fibrosis, we compared the progression of intestinal fibrosis in rats (induced by trinitrobenzene sulfonic acid [TNBS] administration) with or without L-aspartic acid treatment, while using rats treated with phosphate buffer saline (PBS) as healthy controls. The TNBSinduced rat model is widely used to study CD due to its ability to rapidly induce colonic inflammation and subsequent intestinal fibrosis (Dillman et al, 2013). We also used MTI and 18F-fibroblast activation protein inhibitor (FAPI) PET/CT (another novel imaging technique for noninvasive assessment of intestinal fibrosis by targeting fibroblast activation protein (Scharitzer et al, 2023) to observe the intestines of rats. Overall, the degree of intestinal disorder found in TNBS-induced rats was significantly higher after three weeks of L-aspartic acid treatment than that in the control groups, in line with our hypothesis. Specifically, the more severe intestinal morphological changes observed that the rats in the treated group compared with control group 1 (stimulated by TNBS only) included the development of intestinal penetration and adhesion, observed both via imaging and in gross specimens (Fig. 5A). The uptake of 18F-FAPI in intestinal lesions in the treated group (treated with TNBS and L-aspartic acid) was significantly higher than that in either control group 1 or control group 0 (no TNBS or L-aspartic acid treatment), indicating an increased degree of intestinal fibrosis in the treated group (Fig. 5B; Appendix

1. Table S2). The fibrosis-promoting effect of L-aspartic acid was also demonstrated by a significant increase in the normalized magnetization transfer ratio (MTR) observed in the treated group, as compared to both control groups 0 and 1 (Fig. 5C; Appendix

1. Table S2). Masson trichrome staining was further conducted to evaluate intestinal fibrosis, and the pathological findings were clearly consistent with all imaging results. Specifically, the fibrotic scores of the treated group were higher than those of control groups
2. Table S3). The profibrotic effect of L-aspartic acid was also observed at the protein level, as evidenced by a significant upregulation of the expression of COL1A1 and FN under treatment with this amino acid (Fig. 6D–F; Appendix Table S3). These data suggested that L-aspartic acid might act synergistically with the

TGF-β1-induced fibrogenic activation of HIMFs by upregulating the extracellular matrix at both the mRNA and protein levels.

Taken together, these results demonstrated that L-aspartic acid promoted the progression of intestinal fibrosis both in vitro and in vivo, providing evidence of a causal relationship between this specific metabolite (or the gut microbiota that produce it) and intestinal fibrosis.

Discussion

Although previous animal studies have linked the gut microbiota or metabolites to intestinal fibrosis, specific microbial and metabolite targets related to intestinal fibrosis in patients with CD have remained unclear. The main constraint on the implementation of such cohort studies is the lack of effective tools for quantifying transmural fibrosis. Here, we utilized MTI, a proven imaging tool, to achieve the accurate classification of intestinal fibrosis in patients with CD. This enabled us to present, for the first time, a series of microbial and metabolite profiles that are specific to moderatesevere intestinal fibrosis in patients with CD. A further innovation of our study was the establishment of a “bridge” between the small molecule profiles and the macroscopic imaging phenotypes of intestinal fibrosis, and the results emphasized the potential to target these traits identified by multiomics analysis to effectively treat fibrosis. Subsequently, we selected L-aspartic acid as an example among numerous targets and obtained causal evidence that it induces fibrotic gene activation in HIMFs in vitro and promotes the aggravation of intestinal fibrosis in rats in vivo. This further indicates the efficacy of targeting these specific molecules to treat intestinal fibrosis.

CD is associated with alterations in the gut microbiota and metabolites, whose potential involvement in the development of intestinal fibrosis is currently under investigation in both human and mouse studies. For example, fibrosis-triggering gut microbes have been identified in the cecum (the species Mucispirillum schaedleri and the genera Ruminococcus and Anaeroplasma) and ileum (the genera Streptococcus and Lactobacillus) in an animal study (Jacob et al, 2018). The colonization of adherent-invasive Escherichia coli (AIEC) in the gut has been reported to be associated with severe intestinal fibrosis in eight patients with CD (Xu et al, 2023) and in mice (Small et al, 2013; Ellermann et al, 2019). Fecal and blood metabolites can serve as functional readouts of links between the gut microbiota and intestinal fibrosis. In a human intestinal organoid model, eicosatetraynoic acid and butyrate were found to regulate extracellular matrix genes implicated in stricture and to suppress collagen accumulation and tissue stiffness (Jurickova et al, 2022). Although there has been progress in identifying microbes and metabolites associated with fibrosis in CD, a significant gap still exists between the current state and clinical implementation due to the lack of extensive patient analysis. Transcending previous studies, we report a relatively complete landscape of the gut microbiota associated with moderate-severe intestinal fibrosis in patients with CD. Notably, a significant decrease of alpha diversity was identified in BF2 compared with BF1, but no significantly changed taxa identified when comparing BF2 with BF1 group. One possible explanation was the limited sample size and another hypothesis was that individual bacteria (e.g., Ruminococcaceae, Muribaculaceae) and

fecal/blood metabolites (e.g., L-aspartic acid, glutamine, IPA) might be specific to BF2 in a non-linear relationship along with the severity of fibrosis, which was consistent with some findings reported in a previous meta-analysis (Ma et al, 2022). These biomarkers were derived from the largest patient cohort with CD related to this research topic reported to date and were externally validated with a test cohort. Among these factors, microbial metabolites, particularly amino acids, appeared to have a greater impact on the progression of intestinal fibrosis than the gut microbiota itself. The metabolism of L-aspartic acid by gut microbes was supported by our metagenomics sequencing data. Although the involvement of L-aspartic acid has been documented in other types of organ fibrosis (Vannucchi et al, 1985; Ritterhoff

1. et al, 2020), our study is the first to demonstrate its role in intestinal fibrosis. L-aspartic acid-induced activation of the profibrotic phenotype of fibrotic genes in HIMFs and promoted aggravation of intestinal fibrosis in an animal model. We speculate that specifically inhibiting the production or uptake of L-aspartic acid in the fibrotic intestine through targeted drugs or dietary interventions has the potential to effectively halt the progression of intestinal fibrosis. Thus, at least one potential target that we identified holds promise for advancing intestinal fibrosis research.

Notably, these biomarkers were selected based on the MTI of the intestines rather than the Montreal classification of patients with CD. The image contrast in MTI depends on the amount of macromolecules such as collagens in intestinal tissue. The normalized MTR can be used to measure this MT effect and reflects the concentration of macromolecules in an aqueous physiological environment (Li et al, 2018). Our work has consistently demonstrated a strong correlation between MTI and pathological fibrosis (Li et al, 2018; Fang et al, 2020; Meng et al, 2020; Lu et al, 2021), which is supported by similar conclusions reached at other medical institutions (Adler et al, 2011; Adler et al, 2013; Pazahr et al, 2013; Dillman et al, 2015). MTI is a reliable imaging technique that accurately and noninvasively assesses transmural fibrosis without being affected by coexisting inflammation (Li et al, 2018), whereas Montreal classification, such as B2 and B3, are clinical phenotypes resulting from the combined effects of inflammation and fibrosis. Therefore, these selected biomarkers are more specific for diagnosing and treating intestinal fibrosis in CD patients. Based on these intestinal microbiota and metabolite biomarkers, we developed and validated the first microecological model capable of accurately diagnosing intestinal fibrosis. This alternative diagnostic tool may be particularly useful in medical centers lacking advanced imaging technology such as MTI.

More intriguingly, our study also revealed relationships between the gut microbiota, metabolites, and intestinal morphological alterations. Such fibrosis-related intestinal morphological changes, particularly stricture and penetration, frequently cause severe clinical symptoms such as abdominal pain, nausea or vomiting and are therefore the primary focus of clinical management (Rieder et al, 2017). We speculated that the gut microbiota and/or its metabolites activated pathogen-associated molecular patterns via pattern recognition receptors, thereby inducing innate and adaptive immune responses (Takeuchi and Akira, 2010); by amplifying and sustaining these immune and inflammatory processes, intestinal fibrosis and disease complications such as stricture, penetration, and extraluminal manifestations would be promoted (Lee and Chang, 2021). This hypothesis was partially supported by our

in vivo and in vitro results, particularly by the more pronounced alterations in intestinal morphology observed in rats treated with L aspartic acid compared with controls. Although this was not our primary objective, this study presents fairly comprehensive evidence that MRE-detectable intestinal macromorphological alterations are at least partially attributable to microbial and metabolite factors and provides novel insights into the potential mechanisms underlying these changes. More importantly, from a clinical imaging perspective, we confirmed that the selected microbes and metabolites could serve as reliable biomarkers for intestinal diagnosis and treatment based on our “bridge” connecting small molecules to macroscopic morphological changes.

There were several limitations in this study. First, we did not incorporate additional potential confounding factors such as patient dietary patterns and lifestyles during data analysis. The inclusion of these confounding factors, along with data collection from more centers, could strengthen our conclusions. Second, as the data suggested, we observe negative correlations between KO00250 and aspartic acid concentrations in both fecal and blood samples, indicating that the gut microbiota processed a metabolizing functionality of aspartic acid. However, the association between KO00250 and aspartic acid concentration in fecal samples was not statistically significant. The inconsistent relationships between blood and fecal metabolites have been repeatedly reported in previous studies (Chen et al, 2020; Deng et al, 2023), probably due to the difference of compound transformation, metabolic absorption, and chemical format. Further study is warranted to investigate the specific microbial taxonomy and gene functions which are involved in the aspartic acid metabolism process. Third, although we had provided evidence to support the role of L-aspartic acid in promoting intestinal fibrosis, further experiments are necessary to elucidate its specific mechanisms of action. Moreover, additional investigation should focus on the “gut microbiota–metabolites–intestinal fibrosis–intestinal morphological changes” axis to better understand its pathogenic mechanism.

In conclusion, we present a proof-of-concept study to support the development of microbial/metabolite-based tools for managing intestinal fibrosis in patients with CD. These findings provide promising potential therapeutic targets for CD-associated intestinal fibrosis. The subsequent development of specific drug treatment strategies or dietary interventions based on these targets may potentially offer a broader spectrum of therapeutic options for the management of intestinal fibrosis.

Methods

Reagents and tools table

Identifier or catalog number

Reagent/resource Reference or source

Experimental models

Musso et al, 1999; Zhao et al, 2020 N/A

Primary human intestinal myofibroblasts

Antibodies

COL1A1 (E8F4L) XP® Rabbit mAb

Cell Signaling #72026

Reagent/resource Reference or source

Identifier or catalog number

Anti-Fibronectin primary antibody

Abcam ab268020

Anti-β-actin primary antibody

Goat anti-Rabbit IgG (H + L) CrossAdsorbed Secondary Antibody

Proteintech 66008-1-Ig

Invitrogen A11008

Goat anti-Mouse IgG (H + L) Highly CrossAdsorbed Secondary Antibody

Invitrogen A32723

Chemicals, enzymes, and other reagents

MagBeads FastDNA Kit for Soil

MP Biomedicals, CA, USA 116564384

TNBS Meilunbio MB5547-1 Olive oil Solarbio O8320 Saline (0.9%) China Otsuka Pharmaceutical Co.,

COPKB011.2306.

Ltd.

Dulbecco’s minimal essential medium

GIBCO C11965500BT

Fetal bovine serum GIBCO 10091148 AntibioticAntimycotic

GIBCO 15240062

L-aspartic acid MACKLIN L800690

Recombinant Human TGF-β1

Peprotech AF-100-21C

TRIzol reagent Ambion 15596018 ReverTra Ace qPCR RT Master Mix Kit

TOYOBO FSQ-201

FS Essential DNA Probes Master

Roche 6924492001

RIPA buffer Cell Signaling 9806S Protease and phosphatase inhibitors

Cell Signaling 5872S

Blue loading Buffer Cell Signaling 7722S Software QIIME2 2019.4 https://docs.qiime2.org/2019.4/ tutorials/

N/A

Cutadapt (version v1.1)

https://cutadapt.readthedocs.io/en/ stable/

N/A

map2slim www.metacpan.org N/A Kraken2 https://ccb.jhu.edu/software/

N/A

kraken2/

MMseqs2 https://github.com/soedinglab/ MMseqs2

N/A

Megahit (v1.1.2) https://github.com/voutcn/megahit N/A KOBAS http://bioinfo.org/kobas/ N/A EggNOG http://eggnogdb.embl.de/ N/A CAZy http://www.cazy.org/ N/A

Reagent/resource Reference or source

Identifier or catalog number

Prism statistics software

GraphPad N/A

R (4.2.1) https://www.r-project.org/ N/A Python (2.7.11) https://www.python.org/ N/A FastQC (version 0.11.7)

N/A

http:// www.bioinformatics.babraham.ac.uk/ projects/fastqc/

Metaphlan 3 https://github.com/biobakery/ MetaPhlAn

N/A

Humann 2 https://github.com/biobakery/ humann

N/A

Trimmomatic (0.33) http://www.usadellab.org/cms/ index.php?page trimmomatic

N/A

DADA2 (1.03) https://benjjneb.github.io/dada2/ N/A Other

NanoDrop NC2000 spectrophotometer

Thermo Fisher Scientific, Waltham, MA, USA

N/A

Vazyme VAHTS DNA Clean Beads

Vazyme, Nanjing, China N/A

Quant-iT PicoGreen dsDNA Assay Kit

Invitrogen, Carlsbad, CA, USA N/A

Illumina NovaSeq 6000

Qubit™ 4 Fluorometer

TruSeq Nano DNA High Throughput Library Prep Kit (96 samples)

N-(3(dimethylamino) propyl)-N′

-ethylcarbodiimide (EDC)·HCl

ultraperformance liquid chromatography coupled to tandem mass spectrometry (UPLC‒MS/MS)

Targeted Metabolome Batch Quantification (TMBQ)

Illumina N/A

Invitrogen, USA N/A

Illumina, USA 20015965

Sigma-Aldrich, St. Louis, MO, USA N/A

ACQUITY UPLC Xevo TQ-S, Waters Corp., Milford, MA, USA

N/A

Metabo-Profile, Shanghai, China v1.0

iMAP platform Metabo-Profile, Shanghai, China v1.0 Hematoxylin–eosin Servicebio G1076 Masson Servicebio G1006 Roche Light Cycler 96 instrument

Roche 05815916001

PAGE Gel Fast Preparation Kit

EpiZyme PG112

PVDF membrane Invitrogen LC2007

Ethics

The human study protocols were approved by the Institutional Ethics Review Board of our hospital (No. [2021]215-2). Informed consent was obtained from all participants, and the experiments conformed to the principles set out in the WMA Declaration of Helsinki and the Department of Health and Human Services Belmont Report. The methods were performed in accordance with STROBE guidelines. The animal experiments in this study were approved and performed in accordance with the guidelines of the Animal Ethics Committee of our hospital (No. [2023]116). Animals were provided with ad libitum access to a standard laboratory diet and sterile water and were housed under controlled, monitored temperature and humidity conditions, in rooms with a 12 h light/ dark cycle.

Human participants

From May 2021 to March 2023, 291 patients with CD presenting for fecal and/or serum sample collection within one week of magnetic resonance enterography (MRE) were approached for potential recruitment. Thirteen potential participants were excluded because they had used antibiotics, probiotics, or prebiotics in the three months before inclusion (n = 11) or had other concomitant organic digestive diseases (n = 2), as determined from medical records and questionnaires administered during recruitment. Therefore, the final patient cohort contained 278 subjects (derivation cohort, n = 214; test cohort, n = 64). All patient analyses reported in this study were derived from the derivation cohort, while the test cohort was solely utilized to verify the generalization ability of our constructed models.

Simultaneously, 30 healthy controls (HCs) were recruited to provide fecal and serum samples based on open advertisements in the community. HCs were eligible to participate if they showed no gastrointestinal symptoms or antibiotic, probiotic, or prebiotic use in the three months before inclusion. The HCs were also clinically examined for diseases. Two outliers were excluded from HCs after 16S rRNA gene sequencing. Therefore, the final HC cohort comprised 28 subjects. In total, 306 subjects took part in this study. Their demographics, medical histories and other clinical variables were collected after enrollment according to standard procedures. Detailed clinical information about all study participants is provided in Table EV1.

Collection of fecal and blood specimens

The participants provided fecal and/or serum samples at the hospital. Fecal samples were acquired in collection cups and immediately frozen

1. at −80 °C until analysis. Blood samples were drawn after overnight fasting, sent directly to the laboratory for serum collection, and frozen at −80 °C until analysis. All 278 patients with CD provided blood samples; fecal samples were available for 216 of these patients (derivation cohort, n = 152/214; test cohort, n = 64/64). Both fecal and serum samples were analyzed for all 28 HCs. Magnetic resonance enterography

All enrolled patients with CD underwent MRE. Bowel preparation before the scanning procedure was performed as described

previously (Li et al, 2017, 2018). MRE was performed using a 3.0 T magnetic resonance system (MAGNETOM Prisma; Siemens Healthineers, Erlangen, Germany) with multichannel phased-array body coils. The scan protocol (including T2-weighted imaging, magnetization transfer imaging (MTI), diffusion-weighted imaging, and pre-/post-enhancement T1-weighted imaging) was the same as that described in our previous study (Li et al, 2018). Among them, MTI was acquired using two gradient-echo datasets, with or without the application of an off-resonant prepulse (frequency offset: 1.2 kHz, duration: 9984 µs, effective flip angle: 500°, bandwidth: 192 Hz).

Assessment and grading of intestinal fibrosis

The quantitative assessment of intestinal fibrosis was based on the normalized MTR derived from MTI as described previously (Li et al, 2018). In brief, the MTR was calculated using the following equation: MTR (%) = [1 − (Msat/M0)] × 100, where Msat and M0 are the signal intensities acquired with and without off-resonance prepulse saturation, respectively. In the MTR map, a region of interest was drawn at the midpoint level of the most severe bowel segment (i.e., the part of the intestine with the greatest stricture and/or the thickest bowel wall), including the full thickness of the whole circular bowel, to obtain the MTR of the bowel wall. The MTR of the skeletal muscle was also measured in the same MTR map, and the MTR of the bowel wall was divided by the MTR of the skeletal muscle to yield a normalized MTR, to minimize individual variation (Li et al, 2018). All image postprocessing and parameter calculations were conducted using Python (version 3.9).

A normalized MTR < 0.71 was defined as indicating none-mild bowel fibrosis (namely, BF1), and a normalized MTR ≥ 0.71 was defined as indicating moderate-severe bowel fibrosis (namely, BF2) (Li et al, 2018). According to the normalized MTR of the most severe bowel segment per patient, there were 93 patients with nonemild fibrosis and 121 patients with moderate-severe fibrosis in the derivation cohort, while in the test cohort, there were 20 patients with none-mild fibrosis and 44 patients with moderate-severe fibrosis. Given that CD is characterized by multiple segmental bowel lesions, we also stratified the patients according to the lesion with the highest normalized MTR value rather than the section with the most severe stricture/greatest thickness, to ensure that the fibrosis stratification method we chose was representative and reliable. Our results showed that the fibrosis classifications remained unchanged (Fig. EV1C,D; Appendix Table S4).

Assessment of intestinal fibrosis-associated magnetic resonance enterography features

The occurrence and progression of intestinal fibrosis led to a series of changes in the intestine that can be observed by morphological and functional MRE (Bruining et al, 2018). To explore the relationship between the gut microbiota, metabolites, and intestinal fibrosis-associated imaging features, the MRE findings of the most severely diseased intestines (i.e., the part of the intestine with the greatest stricture and/or the thickest bowel wall) were jointly evaluated by a team of five professionals with experience in intestinal imaging using quantitative measurement or semiquantitative scoring methods. The interpreted imaging features included bowel stricture, penetrating disease, perienteric effusion, comb sign,

wall thickness, mural apparent diffusion coefficient, intramural edema, mural hyperenhancement, mural enhancement pattern, length of diseased bowel, adenopathy, and perianal diseases (Table EV10).

16S rRNA gene amplicon sequencing

All the samples were sequenced or measured in one batch, and processed using an identical bioinformatic pipeline.

Fecal samples collected from 216 patients with CD (derivation cohort, n = 152; test cohort, n = 64) and 28 HCs, as described in “Collection of fecal and blood specimens” above, were subjected to 16S rRNA gene amplicon sequencing. Total genomic DNA was extracted from the samples using the OMEGA Soil DNA Kit (M5635-02) (Omega Bio-Tek, Norcross, GA, USA) according to the manufacturer’s instructions and was subsequently stored at −20 °C for further analysis. The quantity and quality of the extracted DNA were measured using a NanoDrop NC2000 spectrophotometer (Thermo Fisher Scientific, Waltham, MA, USA) and by agarose gel electrophoresis, respectively.

The V3-V4 regions of bacterial 16S rRNA genes were amplified by polymerase chain reaction (PCR) using the forward primer 338 F (5’ACTCCTACGGGAGGCAGCA-3’) and the reverse primer 806 R (5’GGACTACHVGGGTWTCTAAT-3’). Sample-specific 7-bp barcodes were incorporated into the primers for multiplex sequencing. The PCR mixture contained 5 μl of buffer (5×), 0.25 μl of Fast pfu DNA Polymerase (5 U/μl), 2 μl (2.5 mM) of dNTPs, 1 μl (10 µM) of each forward and reverse primer, 2 μl of DNA template, and 8.75 μl of ddH2O. The thermal cycling program consisted of initial denaturation at 98 °C for 2 min, followed by 25 cycles of denaturation at 98 °C for 15 s, annealing at 55 °C for 30 s, and extension at 72 °C for 30 s, with a final extension of 5 min at 72 °C. PCR amplicons were purified with Vazyme VAHTS DNA Clean Beads (Vazyme, Nanjing, China) and quantified using the Quant-iT PicoGreen dsDNA Assay Kit (Invitrogen, Carlsbad, CA, USA). After the individual quantification step, the amplicons were pooled in equal amounts, and pair-end 2 × 250 bp sequencing was performed on the Illumina NovaSeq platform with a NovaSeq 6000 SP Reagent Kit (500 cycles) at Shanghai Metabo-Profile Biotechnology Co., Ltd. (Shanghai, China).

Microbiome bioinformatics analysis was conducted using QIIME2 2019.4 (Bolyen et al, 2019) based on the official tutorials (https://docs.qiime2.org/2019.4/tutorials/). Briefly, raw sequence data were demultiplexed using the demux plugin, followed by primer cutting with the Cutadapt plugin (Kechin et al, 2017). Sequences were then quality-filtered, denoised, and merged, and chimeras were removed using the DADA2 plugin (Callahan et al, 2016). Non-singleton amplicon sequence variants (ASVs) were aligned with MAFFT (Katoh et al, 2002) and used to construct a phylogeny with FastTree2 (Price et al, 2009). Taxonomy was assigned to ASVs using the classify-sklearn naïve Bayes taxonomy classifier in the feature-classifier plugin (Bokulich et al, 2018) against the SILVA Release 132 Database (Koljalg et al, 2013).

Metagenomic shotgun sequencing

Metagenomic shotgun sequencing was performed on fecal samples collected from 47 patients with CD (derivation cohort, n = 36; test cohort, n = 11) and 24 HCs to analyze the correlation between the

microbial metabolization pathways and blood aspartic acid levels. These 71 subjects were randomly selected from the two cohorts in this study. Total microbial genomic DNA was extracted from the samples using the OMEGA Mag-Bind Soil DNA Kit (M5635-02) (Omega Bio-Tek, Norcross, GA, USA) following the manufacturer’s instructions and stored at −20 °C prior to further assessment. The quantity and quality of the extracted DNA were measured using a Qubit™ 4 Fluorometer with Wi-Fi: Q33238 (Qubit™ Assay Tubes: Q32856; Qubit™ 1X dsDNA HS Assay Kit: Q33231) (Invitrogen, USA) and by agarose gel electrophoresis, respectively. The extracted microbial DNA was processed to construct metagenome shotgun sequencing libraries with an insert size of 400 bp with the Illumina TruSeq Nano DNA LT Library Preparation Kit. Each library was sequenced on the Illumina NovaSeq platform (Illumina, USA) according to the PE150 strategy at Metabo-Profile Biotechnology Co., Ltd. (Shanghai, China).

Raw sequencing reads were processed to obtain quality-filtered reads for further analysis. First, sequencing adapters were removed from the sequencing reads using Cutadapt (v1.2.1) (Martin, 2011). Second, low-quality reads were trimmed using a sliding-window algorithm in FASTP (Chen et al, 2018). Third, reads were aligned to the host genome using BMTagger to remove host contamination (Rotmistrovsky and Agarwala, 2011).

Once quality-filtered reads were obtained, the taxonomic classification of metagenomics sequencing reads from each sample was performed using Kraken2 (Wood et al, 2019) against a RefSeqderived database that included genomes from archaea, bacteria, viruses, fungi, protozoans, metazoans and Viridiplantae. Reads assigned to metazoans or Viridiplantae were removed from the downstream analysis. Megahit (v1.1.2) (Li et al, 2015) was employed to assemble the results from each sample using the meta-large preset parameters. The generated contigs (longer than 300 bp) were then pooled together and clustered using MMseqs2 (Steinegger and Soding, 2017) in “easy-linclust” mode, setting the sequence identity threshold to 0.95 and coverage of the residues of the shorter contig to 90%.

The lowest common ancestor taxonomy of the nonredundant contigs was obtained by aligning them against the NCBI-nt database with MMseqs2 (Steinegger and Soding, 2017) in “taxonomy” mode, and contigs assigned to Viridiplantae or Metazoa were dropped in the following analysis. MetaGeneMark (Zhu et al, 2010) was used to predict genes in the contigs. The coding sequences (CDSs) of all samples were clustered by using MMseqs2 (Steinegger and Soding, 2017) in “easy-cluster” mode, setting the protein sequence identity threshold to 0.90 and the coverage of the residues of the shorter contig to 90%. To assess the abundances of these genes, the high-quality reads from each sample were mapped onto the predicted gene sequences using salmon (Patro et al, 2015) in quasi mapping-based mode with “--meta --minScoreFraction=0.55”, and copy per kilobase per million mapped reads (CPM) method was used to normalize abundance values in metagenomes. The functionality of the nonredundant genes was evaluated by annotation using MMseqs2 (Steinegger and Soding, 2017) in “search” mode against the KEGG, EggNOG and CAZy protein databases. EggNOG and GO results were obtained using EggNOG-mapper (v2) (Cantalapiedra et al, 2021). GO results were evaluated using map2slim (www.metacpan.org). KO results were obtained using KOBAS (Bu et al, 2021).

Fecal and blood metabolomics

As described in “Collection of fecal and blood specimens” above, a total of 216 patients with CD (derivation cohort, n = 152; test cohort, n = 64) and 28 HCs provided fecal samples for targeted metabolomics profiling, and blood samples for targeted metabolomics profiling were collected from all 306 participants.

The targeted metabolomics analysis of the samples was performed with Metabo-Profile (Shanghai, China). The sample preparation procedures were performed according to previously published methods with minor modifications (Xie et al, 2021). Briefly, for fecal-targeted metabolomics, 5 mg of feces was homogenized with zirconium oxide beads, while for bloodtargeted metabolomics, 20 μL of serum was transferred to a 96well plate. Subsequently, 120 μL of methanol containing an internal standard was used to extract the metabolites. The resulting supernatants were subjected to derivatization with 3-nitrophenylhydrazine (3-NPH) and N-(3-(dimethylamino)propyl)-N′-ethylcarbodiimide (EDC)·HCl (Sigma-Aldrich, St. Louis, MO, USA). Subsequently, the derivatized samples were analyzed by ultraperformance liquid chromatography coupled to tandem mass spectrometry (UPLC‒MS/MS) (ACQUITY UPLC Xevo TQ-S, Waters Corp., Milford, MA, USA). All of the standards were obtained from Sigma-Aldrich (St. Louis, MO, USA), Steraloids Inc. (Newport, RI, USA), and TRC Chemicals (Toronto, ON, Canada). Quality control samples were prepared following the same procedures applied for the test samples and were injected at regular intervals to ensure instrument system stability. The raw data files generated by UPLC‒MS/MS were processed using Targeted Metabolome Batch Quantification (TMBQ) software (v1.0, HMI, Shenzhen, Guangdong, China) to perform peak integration, calibration, and quantitation for each metabolite. The in-house-developed iMAP platform (v1.0, Metabo-Profile, Shanghai, China) was used for statistical analysis.

Associations between L-aspartic acid, intestinal inflammation, and fibrosis

To investigate whether the relationship between L-aspartic acid and intestinal fibrosis was influenced by coexisting intestinal inflammation, we initially examined the variations in L-aspartic acid levels across different inflammatory groups. Considering that conventional endoscopy may fail to reach small intestinal lesions in some patients, a simplified magnetic resonance index of activity (MARIAs) (Ordas et al, 2019) derived from MRE was employed for the evaluation of intestinal inflammation. MARIAs = (1×thickness>3 mm)+(1×edema)+(1×fat stranding) (2×ulcers), where thickness refers to bowel wall thickness, edema indicates mural edema, and fat stranding means perienteric effusion in our study. The bowel site for intestinal inflammation assessment was chosen to be the same as that for fibrosis evaluation. A MARIAs≥2 was defined as moderate-severe bowel inflammation (namely, BI2), while a MARIAs<2 was defined as none-mild bowel inflammation (namely, BI1) (Ordas et al, 2019). Finally, 130 patients were categorized as having none-mild inflammation, and the other 84 patients exhibited moderate-severe inflammation in the derivation cohort. There was no significant difference in L-aspartic acid levels between BI1 and BI2 (Fig. EV3; Appendix Table S5). To elucidate the correlation between L-aspartic acid and both fibrosis and

inflammation, a multiple linear regression model was constructed using the enter method. The model showed that L-aspartic acid was significantly associated with bowel fibrosis severity (β = 2.088, P = 0.043) but not with inflammation severity (β = −0.513, P = 0.623). The aforementioned findings collectively suggested that the association between L-aspartic acid and intestinal fibrosis remained unaltered in the presence of concurrent intestinal inflammation.

In vivo rat colitis model experiments

Animal model

Female Sprague‒Dawley rats aged 6 weeks (Guangdong Medical Laboratory Animal Center, Guangzhou, China) were used to mimic intestinal fibrosis to different degrees. The animals were housed in individually ventilated cages, maintaining a temperature range of 22–26 °C with a maximum daily fluctuation of 4 °C. The relative humidity was maintained between 40 and 70%. A minimum of 15 times air changes per hour was ensured to maintain a barrier environment. The airflow velocity did not exceed 0.2 m/s. To establish this model (Dillman et al, 2013), 2,4,6-trinitrobenzene sulfonic acid (TNBS; Sigma-Aldrich, St. Louis, Missouri, USA) was administered at 150 mg/kg/week since L-aspartic acid was administered at 200 mg/kg every other day as referenced (Rao et al, 2021). The rats were allocated to three treatment groups: Tx (treated with TNBS and L-aspartic acid), control group 1 (C1, treated with TNBS), and control group 0 (C0, treated only with PBS). Specifically, the rats in the Tx group were treated with TNBS and L-aspartic acid for either 2 weeks (n = 3) or 3 weeks (n = 6), while those in the C1 group received TNBS treatment for either 2 weeks (n = 6) or 3 weeks (n = 6). Those in the C0 group were treated with PBS only for either 2 weeks (n = 3) or 3 weeks (n = 3).

Micro-PET/CT and data analysis in animal experiments

The rats were administered a dose of 300 MBq of 18F-FAPI via tail vein injection (100–200 μL) one hour prior to scanning. PET-CT imaging was performed using a micro-PET system (Inviscan SAS, Strasbourg, France), and rats were then imaged for a 15-minute static acquisition (Li et al, 2012). Three-dimensional regions of interest (ROIs) were manually placed on the PET images in areas where the intestines exhibited the highest degree of fibrotic activity. The three-dimensional ROIs did not encompass the entire lesion volume to avoid partial volume effects. Subsequently, the maximum standardized uptake value (SUVmax) in the intestinal lesion was obtained for further analysis.

MTI protocol and data analysis in animal experiments

The MTI scanning protocol was performed on the intestines of the rats according to a previously described method (Lu et al, 2021) and using a 3.0-T MR system (MAGNETOM Prisma; Siemens Healthineers, Erlangen, Germany) and a 4-channel animal coil (Shanghai Chenguang Medical Technology Co., Ltd., Shanghai, China). The postprocessing and data analysis methods employed for MTI in rats were the same as those previously described for patients with CD in this study.

Pathological analysis

The rats were promptly euthanized following scanning in accordance with the treatment protocol for pathological

examination. Intestinal lesion segments from the Tx and C1 groups as well as normal colorectal segments from the C0 group were obtained and fixed using 4% paraformaldehyde. Following embedding in paraffin, the intestinal specimens were sectioned into 4-μm slices and subjected to staining with hematoxylin and eosin (HE) as well as Masson’s trichrome (Li et al, 2018). HE staining was utilized to assess intestinal inflammation, while Masson’s trichrome staining was employed for the evaluation of intestinal fibrosis using a semiquantitative scoring system (scores 0-4), as described in our previous study (Li et al, 2018).

In vitro human primary intestinal myofibroblast experiments

Isolation and culture of primary human intestinal myofibroblasts

Primary human intestinal myofibroblasts (HIMFs) were isolated from the surgically resected intestinal specimen of patients with CD, and cultured in Dulbecco’s minimal essential medium supplemented with 10% fetal bovine serum (FBS, GIBCO, Life Technologies Corporation, Grand Island, NY, USA) and antibiotics, according to previously established protocols (Musso et al, 1999; Zhao et al, 2020). The cells were established as long-term cultures, which were fed twice weekly and sub-cultured at the confluence., and were utilized between passages 6 and 8.

Cell treatment

To evaluate the profibrogenic properties of L-aspartic acid in vitro, primary HIMFs (1.5 × 105 cells/well) were inoculated in a six-well plate and then treated with 5 ng/mL TGF-β1, either alone or cocultured with 0 μM, 6.5 μM or 13 μM L-aspartic acid in serumfree HIMF culture medium. After 48 h, HIMFs RNA and protein were examined by real-time RT-PCR and western blotting, respectively.

qRT‒PCR

Total RNA was extracted from the HIMFs using TRIzol reagent (Ambion, Carlsbad, CA). Then, equal amounts of RNA (1 μg) were reverse-transcribed into cDNA using the ReverTra Ace qPCR RT Master Mix Kit (TOYOBO, Osaka, Japan) according to the manufacturer’s instructions. All qRT-PCR assays were performed using a Roche Light Cycler 96 instrument (Roche) with Faster Start Essential DNA Probes Master (Roche). The mRNA levels of all genes were normalized to that of β-actin. PCR amplification was performed with the following primers (5′ to 3′): Human collagen 1A1 (COL1A1) sense TAGTCTGTCCTGCGTCCTCT and antisense TTATGTTTGGGTCATTTCCA primers; Human fibronectin

1. 1 (FN) sense CTACGGATGACTCGTGCTTT and antisense TTCCTTCTGCCACTGTTCTC primers; Human actin alpha 2 (ACTA2) sense CTGAGCGTGGCTATTCCTTC and antisense GCTGGAAGGTGGACAGAGAG primers; Human β-actin sense GTTGCTATCCAGGCTGTG and antisense GTTGCTATCCAGGCTGTG primers.

Western blotting

To obtain protein extracts for western blotting, cells were homogenized in lysis buffer (RIPA buffer containing protease and phosphatase inhibitors (Cell Signaling, Beverly, MA) followed by centrifugation to clear the lysate. Equal amounts of protein lysates were separated by SDS-PAGE, transferred to a PVDF membrane,

and incubated overnight with primary antibodies: COL1A1 (1:1000 dilution, #72026, Cell Signaling Technology), FN (1:1000 dilution, ab268020, Abcam, Cambridge, MA), and β-actin (1:1000 dilution, 66008-1-Ig, Proteintech). Proteins were visualized using the ECL detection system (Amersham Pharmacia Biotech, Piscataway, NJ) with the appropriate secondary antibodies: anti-rabbit HRP (1:2000 dilution, A11008, Invirtrogen) or anti-mouse HRP (1:2000 dilution, A32723, Invirtrogen).

Statistical analysis

All analyses were performed in R (v.4.2.1). Microbial alpha diversity was calculated using the Simpson, Shannon, Chao1, and Pielou indices and taxonomic richness. The Wilcoxon test was used to compare alpha diversity differences. Bray‒Curtis distances were employed to represent beta diversity, and the PERMANOVA test was performed with the adonis function in vegan (v.2.6) to assess the microbial variance explained by different fibrosis groups. Principal component analysis (PCA) was conducted for normalized metabolomics data. Statistical significance was determined according to an FDR < 0.1 (Benjamini‒Hochberg procedure).

Four linear models were used to investigate BF2-specific microbial and metabolomic features. Each model was corrected for potential confounders, including age, sex, BMI, smoking, and disease location (only within patients with CD). All the patients did not use antibiotics within 3 months.

1. (1) Features ~ CD vs. HC + covariates
2. (2) Features ~ BF1 vs. HC + covariates
3. (3) Features ~ BF2 vs. HC + covariates
4. (4) Features ~ BF1 vs. BF2 + covariates.

Considering the limited sample size in model (4), we defined BF2-specific features as those results that passed the significance threshold in model (3) but were not present in models (1) and (2) (Guo et al, 2023).

To select BF2-associated morphological changes, we assessed twelve MRE features of intestinal lesions, including eight luminal (e.g., stricture, penetration, wall thickening) and four extraluminal (e.g., perienteric effusion, comb sign, adenopathy) features, based on the χ2 test. Stricture, penetration, effusion and the comb sign were further subjected to association tests with omics data.

To explore the disease-predictive potential of omics data, we incorporated six machine learning approaches, including the decision tree, random forest, gradient-boosted decision tree, extra tree, logistic regression and support vector machine method, included in the caret R package (v.6.0). These analyses were performed on a cohort of 152 patients with all the multiomics data. Five models were compared to classify CD patients versus HCs and BF2 patients versus BF1 patients among patients with CD, as follows:

1. (1) Groups ~ age + sex + BMI + smoking + disease location
2. (2) Groups ~ microbial taxa
3. (3) Groups ~ blood metabolites
4. (4) Groups ~ fecal metabolites
5. (5) Groups ~ all combined.

The paper explained Problem

Intestinal fibrosis is the most important cause of disability and poor prognosis in patients with Crohn’s disease. However, its profibrotic mechanisms remain incompletely understood, limiting the development of effective therapeutic strategies. The gut microbiota and its functional metabolites play crucial roles in the onset of Crohn’s disease, suggesting their potential as targets for addressing intestinal fibrosis. While the associations of the microbiota and its metabolites in fibrogenesis have been studied in other organs, it has not yet been thoroughly explored in the gut.

Results

In this study, a relatively comprehensive profile of specific microbes and metabolites associated with moderate-severe intestinal fibrosis has been revealed, offering potential targets for classifying this condition and conducting therapeutic trials to address it. A “bridge” between the small molecule profiles and the macroscopic imaging phenotypes of intestinal fibrosis is also established, emphasizing the potential to target these traits identified by multiomics analysis to effectively treat fibrosis. L-aspartic acid, identified as a potential target, has been demonstrated to induce fibrotic gene activation in HIMFs in vitro and promote the aggravation of intestinal fibrosis in rats in vivo.

Impact

These findings presented here enhance our understanding of the potential pathogenic mechanism underlying intestinal fibrosis and offer promising prospects for advancing research in this field, potentially contributing to improving patient prognosis and reducing healthcare costs.

AUC comparisons were assessed using the DeLong test. To ensure model generalizability, the whole dataset was randomly split into training (70%) and validation (30%) parts. The model training was used fivefold cross-validation with parameters of train Control (method = “repeatedcv”, repeats=5) in caret R package. Moreover, we validated the models in an independent external test cohort (n = 64).

To investigate the correlations between the four specific morphological changes on MRE images (i.e., stricture, penetration, perienteric effusion and the comb sign) and the omics data, we performed linear regression analysis adjusting for age, sex, BMI, smoking status and disease location in BF1 and BF2 patients. Only associations with a nominal significance of P < 0.05 were further included in a meta-analysis to evaluate their consistency between BF1 and BF2 patients. The results with meta P values that passed FDR correction and showed a Q Test P > 0.05 were considered significant. The results with a Q test P < 0.05 were considered potentially heterogeneous between the two groups, indicating a fibrosis-dependent effect.

To identify putative links between the gut microbiota, metabolites and morphological changes, causal mediation analysis was performed using the mediation R package (v.4.2.1). We hypothesized that the blood and intestinal metabolites might be produced or metabolized by the gut microbiota, and the metabolites were therefore used as mediators. The gut microbiota was defined as exposures while stricture, penetration, perienteric effusion and the comb sign were defined as outcomes. P values of average causal mediation effects (ACME) were used to assess

whether specific taxa potentially contributed to morphological features in a manner mediated specific small molecules. The plotly R package was used for visualization.

Both in vivo and in vitro results are presented as the mean ± the standard deviation (SD) and were analyzed with the Prism statistics software package version 8.0.0 (GraphPad, San Diego, CA). A T test was used to compare 18F-FAPI uptake and normalized MTRs between groups of rats under different treatment strategies. The fibrotic scores of intestinal lesions in rats among different treatment groups were compared with Fisher’s exact test. The densitometric data from the western blot assay and qRT‒PCR data were analyzed via one-way analysis of variance (ANOVA) with Tukey’s post hoc test. Differences were considered statistically significant at P < 0.05. No statistical methods were used to predetermine the sample size of participants. All animal experiments and in vitro studies were performed and analyzed in a blinded manner.

For more information

Xuehua Li’s lab website: https://www.researchgate.net/lab/XuehuaLi-Lab.

Data availability

The datasets and computer code produced in this study are available in the following databases: (1) The raw data of fecal microbiome sequencing: Sequence Read Archive (SRA) database; accession number: PRJNA1029396; https://www.ncbi.nlm.nih.gov/ bioproject/1029396. (2) Bioinformatics analysis scripts: GitHub (https://github.com/WendyNice/IBD_fibrosis).

The source data of this paper are collected in the following database record: biostudies:S-SCDT-10_1038-S44321-024-00129-8.

Expanded view data, supplementary information, appendices are available for this paper at https://doi.org/10.1038/s44321-024-00129-8.

Peer review information

A peer review file is available at https://doi.org/10.1038/s44321-024-00129-8

References

Adler J, Rahal K, Swanson SD, Schmiedlin-Ren P, Rittershaus AC, Reingold LJ, Brudi JS, Shealy D, Cai A, Mckenna BJ et al (2013) Anti-tumor necrosis factor alpha prevents bowel fibrosis assessed by messenger RNA, histology, and magnetization transfer MRI in rats with Crohn’s disease. Inflamm Bowel Dis 19:683–690

Adler J, Swanson SD, Schmiedlin-Ren P, Higgins PD, Golembeski CP, Polydorides AD, Mckenna BJ, Hussain HK, Verrot TM, Zimmermann EM (2011) Magnetization transfer helps detect intestinal fibrosis in an animal model of Crohn disease. Radiology 259:127–135

Albhaisi S, Shamsaddini A, Fagan A, Mcgeorge S, Sikaroodi M, Gavis E, Patel S, Davis BC, Acharya C, Sterling RK et al (2021) Gut microbial signature of hepatocellular cancer in men with cirrhosis. Liver Transpl 27:629–640 Alexeev EE, Lanis JM, Kao DJ, Campbell EL, Kelly CJ, Battista KD, Gerich ME,

Jenkins BR, Walk ST, Kominsky DJ et al (2018) Microbiota-derived indole

metabolites promote human and murine intestinal homeostasis through regulation of interleukin-10 receptor. Am J Pathol 188:1183–1194

Bernard K (2018) Collagen biosynthesis in pulmonary fibrosis: unraveling the metabolic web. Am J Respir Cell Mol Biol 58:545–546

Bokulich NA, Kaehler BD, Rideout JR, Dillon M, Bolyen E, Knight R, Huttley GA, Gregory CJ (2018) Optimizing taxonomic classification of marker-gene amplicon sequences with QIIME 2’s q2-feature-classifier plugin. Microbiome 6:90

Bolyen E, Rideout JR, Dillon MR, Bokulich NA, Abnet CC, Al-Ghalith GA, Alexander H, Alm EJ, Arumugam M, Asnicar F et al (2019) Reproducible, interactive, scalable and extensible microbiome data science using QIIME 2. Nat Biotechnol 37:852–857

Bruining DH, Zimmermann EM, Loftus EJ, Sandborn WJ, Sauer CG, Strong SA (2018) Consensus recommendations for evaluation, interpretation, and utilization of computed tomography and magnetic resonance enterography in patients with small bowel Crohn’s disease. Gastroenterology 154:1172–1194

Bu D, Luo H, Huo P, Wang Z, Zhang S, He Z, Wu Y, Zhao L, Liu J, Guo J et al (2021) KOBAS-i: intelligent prioritization and exploratory visualization of biological functions for gene enrichment analysis. Nucleic Acids Res 49:W317–W325 Callahan BJ, Mcmurdie PJ, Rosen MJ, Han AW, Johnson AJ, Holmes SP (2016) DADA2: High-resolution sample inference from Illumina amplicon data. Nat Methods 13:581–583

Cantalapiedra CP, Hernandez-Plaza A, Letunic I, Bork P, Huerta-Cepas J (2021) EggNOG-mapper v2: functional annotation, orthology assignments, and domain prediction at the metagenomic scale. Mol Biol Evol 38:5825–5829 Caspi R, Billington R, Keseler IM, Kothari A, Krummenacker M, Midford PE, Ong

WK, Paley S, Subhraveti P, Karp PD (2020) The MetaCyc database of metabolic pathways and enzymes— a 2019 update. Nucleic Acids Res 48:D445–D453

Chen L, van den Munckhof I, Schraa K, Ter Horst R, Koehorst M, van Faassen M, van der Ley C, Doestzada M, Zhernakova DV, Kurilshikov A et al (2020) Genetic and microbial associations to plasma and fecal bile acids in obesity relate to plasma lipids and liver fat content. Cell Rep 33:108212

Chen S, Zhou Y, Chen Y, Gu J (2018) Fastp: an ultra-fast all-in-one FASTQ preprocessor. Bioinformatics 34:i884–i890

D’Alessio S, Ungaro F, Noviello D, Lovisa S, Peyrin-Biroulet L, Danese S (2022) Revisiting fibrosis in inflammatory bowel disease: the gut thickens. Nat Rev Gastroenterol Hepatol 19:169–184

Deng K, Xu JJ, Shen L, Zhao H, Gou W, Xu F, Fu Y, Jiang Z, Shuai M, Li BY et al

(2023) Comparison of fecal and blood metabolome reveals inconsistent associations of the gut microbiota with cardiometabolic diseases. Nat Commun 14:571

Dillman JR, Stidham RW, Higgins PD, Moons DS, Johnson LA, Rubin JM (2013) US elastography-derived shear wave velocity helps distinguish acutely inflamed from fibrotic bowel in a Crohn disease animal model. Radiology 267:757–766

Dillman JR, Swanson SD, Johnson LA, Moons DS, Adler J, Stidham RW, Higgins PD (2015) Comparison of noncontrast MRI magnetization transfer and T2

-Weighted signal intensity ratios for detection of bowel wall fibrosis in a Crohn’s disease animal model. J Magn Reson Imaging 42:801–810

Ellermann M, Gharaibeh RZ, Fulbright L, Dogan B, Moore LN, Broberg CA, Lopez LR, Rothemich AM, Herzog JW, Rogala A et al (2019) Yersiniabactinproducing adherent/invasive Escherichia coli promotes inflammationassociated fibrosis in gnotobiotic Il10(-/-) mice. Infect Immun 87:e00587–19

Fang ZN, Li XH, Lin JJ, Huang SY, Cao QH, Chen ZH, Sun CH, Zhang ZW, Rieder F, Rimola J et al (2020) Magnetisation transfer imaging adds information to conventional MRIs to differentiate inflammatory from fibrotic components of small intestinal strictures in Crohn’s disease. Eur Radiol 30:1938–1947

Flannigan KL, Nieves KM, Szczepanski HE, Serra A, Lee JW, Alston LA, Ramay H, Mani S, Hirota SA (2023) The pregnane X receptor and indole-3-propionic acid shape the intestinal mesenchyme to restrain inflammation and fibrosis. Cell Mol Gastroenterol Hepatol 15:765–795

Gentile CL, Weir TL (2018) The gut microbiota at the intersection of diet and human health. Science 362:776–780

Guo C, Che X, Briese T, Ranjan A, Allicock O, Yates RA, Cheng A, March D, Hornig M, Komaroff AL et al (2023) Deficient butyrate-producing capacity in the gut microbiome is associated with bacterial network disturbances and fatigue symptoms in ME/CFS. Cell Host Microbe 31:288–304

Ha C, Martin A, Sepich-Poore GD, Shi B, Wang Y, Gouin K, Humphrey G, Sanders K, Ratnayake Y, Chan K et al (2020) Translocation of viable gut microbiota to mesenteric adipose drives formation of creeping fat in humans. Cell 183:666–683

Jacob N, Jacobs JP, Kumagai K, Ha C, Kanazawa Y, Lagishetty V, Altmayer K, Hamill AM, Von Arx A, Sartor RB et al (2018) Inflammation-independent TL1a-mediated intestinal fibrosis is dependent on the gut microbiome. Mucosal Immunol 11:1466–1476

Jurickova I, Bonkowski E, Angerman E, Novak E, Huron A, Akers G, Iwasawa K, Braun T, Hadar R, Hooker M et al (2022) Eicosatetraynoic acid and butyrate regulate human intestinal organoid mitochondrial and extracellular matrix pathways implicated in Crohn’s disease strictures. Inflamm Bowel Dis 28:988–1003

Kaplan GG (2015) The global burden of IBD: from 2015 to 2025. Nat Rev Gastroenterol Hepatol 12:720–727

Katoh K, Misawa K, Kuma K, Miyata T (2002) MAFFT: a novel method for rapid multiple sequence alignment based on fast Fourier transform. Nucleic Acids Res 30:3059–3066

Kechin A, Boyarskikh U, Kel A, Filipenko M (2017) CutPrimers: a new tool for accurate cutting of primers from reads of targeted next generation sequencing. J Comput Biol 24:1138–1143

Koljalg U, Nilsson RH, Abarenkov K, Tedersoo L, Taylor AF, Bahram M, Bates ST,

Bruns TD, Bengtsson-Palme J, Callaghan TM et al (2013) Towards a unified paradigm for sequence-based identification of fungi. Mol Ecol 22:5271–5277

Kugathasan S, Denson LA, Walters TD, Kim MO, Marigorta UM, Schirmer M, Mondal K, Liu C, Griffiths A, Noe JD et al (2017) Prediction of complicated disease course for children newly diagnosed with Crohn’s disease: a multicentre inception cohort study. Lancet 389:1710–1718

Kwan SY, Jiao J, Joon A, Wei P, Petty LE, Below JE, Daniel CR, Wu X, Zhang J, Jenq RR et al (2022) Gut microbiome features associated with liver fibrosis in Hispanics, a population at high risk for fatty liver disease. Hepatology 75:955–967

Lavelle A, Sokol H (2020) Gut microbiota-derived metabolites as key actors in inflammatory bowel disease. Nat Rev Gastroenterol Hepatol 17:223–237

Lee G, You HJ, Bajaj JS, Joo SK, Yu J, Park S, Kang H, Park JH, Kim JH, Lee DH et al (2020) Distinct signatures of gut microbiome and metabolites associated with significant fibrosis in non-obese NAFLD. Nat Commun 11:4982

Lee M, Chang EB (2021) Inflammatory bowel diseases (IBD) and the microbiomesearching the crime scene for clues. Gastroenterology 160:524–537

Lee PC, Hsieh YC, Huo TI, Yang UC, Lin CH, Li CP, Huang YH, Hou MC, Lin HC, Lee KC (2021) Active vitamin D(3) treatment attenuated bacterial translocation via improving intestinal barriers in cirrhotic rats. Mol Nutr Food Res 65:e2000937

Li D, Liu CM, Luo R, Sadakane K, Lam TW (2015) MEGAHIT: an ultra-fast singlenode solution for large and complex metagenomics assembly via succinct de Bruijn graph. Bioinformatics 31:1674–1676

Li X, Sun C, Mao R, Huang S, Zhang Z, Yang X, Huang L, Lin J, Zhang J, Ben-Horin S et al (2017) Diffusion-weighted MRI enables to accurately grade

inflammatory activity in patients of ileocolonic Crohnʼs disease. Inflamm Bowel Dis 23:244–253

Li XH, Feng ST, Cao QH, Coffey JC, Baker ME, Huang L, Fang ZN, Qiu Y, Lu BL, Chen ZH et al (2021) Degree of creeping fat assessed by computed tomography enterography is associated with intestinal fibrotic stricture in patients with Crohn’s disease: a potentially novel mesenteric creeping fat index. J Crohns Colitis 15:1161–1173

Li XH, Mao R, Huang SY, Sun CH, Cao QH, Fang ZN, Zhang ZW, Huang L, Lin JJ, Chen YJ et al (2018) Characterization of degree of intestinal fibrosis in patients with Crohn disease by using magnetization transfer MR imaging. Radiology 287:494–503

Li Z, Graf N, Herrmann K, Junger A, Aichler M, Feuchtinger A, Baumgart A, Walch

A, Peschel C, Schwaiger M et al (2012) FLT-PET is superior to FDG-PET for very early response prediction in NPM-ALK-positive lymphoma treated with targeted therapy. Cancer Res 72:5014–5024

Lloyd-Price J, Arze C, Ananthakrishnan AN, Schirmer M, Avila-Pacheco J, Poon TW, Andrews E, Ajami NJ, Bonham KS, Brislawn CJ et al (2019) Multi-omics of the gut microbial ecosystem in inflammatory bowel diseases. Nature 569:655–662

Loomba R, Seguritan V, Li W, Long T, Klitgord N, Bhatt A, Dulai PS, Caussy C, Bettencourt R, Highlander SK et al (2017) Gut microbiome-based metagenomic signature for non-invasive detection of advanced fibrosis in human nonalcoholic fatty liver disease. Cell Metab 25:1054–1062

Lu B, Lin J, Du J, He S, Cao Q, Huang L, Mao R, Sun C, Li Z, Feng S et al (2021) Native T(1) mapping and magnetization transfer imaging in grading bowel fibrosis in Crohn’s disease: a comparative animal study. Biosensors 11:302 Ma J, Liu Z, Gao X, Bao Y, Hong Y, He X, Zhu W, Li Y, Huang W, Zheng N et al

(2023) Gut microbiota remodeling improves natural aging-related disorders through Akkermansia muciniphila and its derived acetic acid. Pharmacol Res 189:106687

Ma S, Shungin D, Mallick H, Schirmer M, Nguyen LH, Kolde R, Franzosa E, Vlamakis H, Xavier R, Huttenhower C (2022) Population structure discovery in meta-analyzed microbial communities and inflammatory bowel disease using MMUPHin. Genome Biol 23:208

Martin M (2011) Cutadapt removes adapter sequences from high-throughput sequencing reads. EMBnet.Journal 17:10–12

Meng J, Huang S, Sun C, Zhang ZW, Mao R, Yang YH, Feng ST, Li ZP, Li X (2020) Comparison of three magnetization transfer ratio parameters for assessment of intestinal fibrosis in patients with Crohn’s disease. Korean J Radiol 21:290–297

Musso A, Condon TP, West GA, De La Motte C, Strong SA, Levine AD, Bennett CF, Fiocchi C (1999) Regulation of ICAM-1-mediated fibroblast-T cell reciprocal interaction: implications for modulation of gut inflammation. Gastroenterology 117:546–556

Ordas I, Rimola J, Alfaro I, Rodriguez S, Castro-Poceiro J, Ramirez-Morros A, Gallego M, Giner A, Barastegui R, Fernandez-Clotet A et al (2019) Development and validation of a simplified magnetic resonance index of activity for Crohn’s disease. Gastroenterology 157:432–439

Patro R, Duggal G, Kingsford C (2015) Salmon: accurate, versatile and ultrafast quantification from RNA-seq data using lightweight-alignment. bioRxiv 021592; https://doi.org/10.1101/021592

Pazahr S, Blume I, Frei P, Chuck N, Nanz D, Rogler G, Patak M, Boss A (2013) Magnetization transfer for the assessment of bowel fibrosis in patients with Crohn’s disease: initial experience. Magma 26:291–301

Price MN, Dehal PS, Arkin AP (2009) FastTree: computing large minimum evolution trees with profiles instead of a distance matrix. Mol Biol Evol 26:1641–1650

Qian W, Xu Y, Wen W, Huang L, Guo Z, Zhu W, Li Y (2023) Exosomal miR-103a3p from Crohn’s creeping fat-derived ASCs contributes to intestinal fibrosis by targeting TGFBR3 and activating fibroblasts. J Crohns Colitis 17(8):1291–1308

Rao Y, Kuang Z, Li C, Guo S, Xu Y, Zhao D, Hu Y, Song B, Jiang Z, Ge Z et al (2021) Gut Akkermansia muciniphila ameliorates metabolic dysfunctionassociated fatty liver disease by regulating the metabolism of L-aspartate via gut-liver axis. Gut Microbes 13:1–19

Rieder F, Fiocchi C, Rogler G (2017) Mechanisms, management, and treatment of fibrosis in patients with inflammatory bowel diseases. Gastroenterology 152:340–350

Ritterhoff J, Young S, Villet O, Shao D, Neto FC, Bettcher LF, Hsu YA, Kolwicz SJ, Raftery D, Tian R (2020) Metabolic remodeling promotes cardiac hypertrophy by directing glucose to aspartate biosynthesis. Circ Res 126:182–196

Rotmistrovsky K, Agarwala R (2011) BMTagger: best match tagger for removing human reads from metagenomics datasets. NCBI/NLM, National Institutes of Health

San-Miguel B, Crespo I, Kretzmann NA, Mauriz JL, Marroni N, Tunon MJ, Gonzalez-Gallego J (2010) Glutamine prevents fibrosis development in rats with colitis induced by 2,4,6-trinitrobenzene sulfonic acid. J Nutr 140:1065–1071

Scharitzer M, Macher-Beer A, Mang T, Unger LW, Haug A, Reinisch W, Weber M, Nakuz T, Nics L, Hacker M et al (2023) Evaluation of intestinal fibrosis with (68)Ga-FAPI PET/MR enterography in Crohn disease. Radiology 307:e222389

Schirmer M, Franzosa EA, Lloyd-Price J, Mciver LJ, Schwager R, Poon TW, Ananthakrishnan AN, Andrews E, Barron G, Lake K et al (2018) Dynamics of metatranscription in the inflammatory bowel disease gut microbiome. Nat Microbiol 3:337–346

Schirmer M, Garner A, Vlamakis H, Xavier RJ (2019) Microbial genes and pathways in inflammatory bowel disease. Nat Rev Microbiol 17:497–511

Small CL, Reid-Yu SA, Mcphee JB, Coombes BK (2013) Persistent infection with Crohn’s disease-associated adherent-invasive Escherichia coli leads to chronic inflammation and intestinal fibrosis. Nat Commun 4:1957

Steinegger M, Soding J (2017) MMseqs2 enables sensitive protein sequence

searching for the analysis of massive data sets. Nat Biotechnol 35:1026–1028 Takeuchi O, Akira S (2010) Pattern recognition receptors and inflammation. Cell

140:805–820 Torres J, Mehandru S, Colombel JF, Peyrin-Biroulet L (2017) Crohn’s disease. Lancet 389:1741–1755 Vannucchi H, Marchini JS, Padovan GJ, Dos-Santos JE, Dutra-De-Oliveira JE

(1985) Amino acid patterns in the plasma and ascitic fluid of cirrhotic patients. Braz J Med Biol Res 18:465–470

Wood DE, Lu J, Langmead B (2019) Improved metagenomic analysis with Kraken

2. Genome Biol 20:257

Xie G, Wang L, Chen T, Zhou K, Zhang Z, Li J, Sun B, Guo Y, Wang X, Wang Y et al (2021) A metabolite array technology for precision medicine. Anal Chem 93:5709–5717

Xu Y, Qian W, Huang L, Wen W, Li Y, Guo F, Zhu Z, Li Z, Gong J, Yu Z et al

(2023) Crohn’s disease-associated AIEC inhibiting intestinal epithelial cellderived exosomal let-7b expression regulates macrophage polarization to exacerbate intestinal fibrosis. Gut Microbes 15:2193115

Yu L, Wang L, Wu X, Yi H (2021) RSPO4-CRISPR alleviates liver injury and restores gut microbiota in a rat model of liver fibrosis. Commun Biol 4:230

Zhao S, Dejanovic D, Yao P, Bhilocha S, Sadler T, Schirbel A, West G, Doyon G, Lopez R, Mao R et al (2020) Selective deletion of MyD88 signaling in alphaSMA positive cells ameliorates experimental intestinal fibrosis via posttranscriptional regulation. Mucosal Immunol 13:665–678

Zhu W, Lomsadze A, Borodovsky M (2010) Ab initio gene identification in metagenomic sequences. Nucleic Acids Res 38:e132

Acknowledgements

This study was supported by National Key R&D Program of China (2023YFC2507300), Key-Area Research and Development Program of

Guangdong Province (2023B1111040003), National Natural Science Foundation of China (82270693, 82271958, 82070680, 82072002, 82222010, 82170537, 81970483), Guangdong Basic and Applied Basic Research Foundation (2023B1515020070, 2023A1515011097), and Fundamental Research Funds for the Central Universities, Sun Yat-sen University (24ykqb003). The funders of the study had no role in study design, data collection, data analysis, data interpretation, or writing of the report.

Author contributions

XueHua Li: Conceptualization; Data curation; Formal analysis; Supervision; Funding acquisition; Validation; Investigation; Methodology; Writing—original draft; Project administration; Writing—review and editing. Shixian Hu: Conceptualization; Data curation; Software; Formal analysis; Validation; Investigation; Methodology; Writing—original draft; Writing—review and editing. Xiaodi Shen: Conceptualization; Data curation; Formal analysis; Validation; Investigation; Methodology; Writing—original draft; Writingreview and editing. Ruonan Zhang: Conceptualization; Data curation; Formal analysis; Validation; Investigation; Methodology; Writing—original draft; Writing—review and editing. Caiguang Liu: Data curation; Formal analysis; Validation; Investigation; Methodology; Writing—original draft; Writingreview and editing. Lin Xiao: Data curation; Formal analysis; Validation; Investigation; Methodology; Writing—original draft; Writing—review and editing. Jinjiang Lin: Data curation; Validation; Investigation; Writing—original draft; Writing—review and editing. Li Huang: Data curation; Validation; Investigation; Writing—original draft; Writing—review and editing. Weitao He: Data curation; Validation; Investigation; Writing—original draft; Writingreview and editing. Xinyue Wang: Data curation; Validation; Investigation; Writing—original draft. Lili Huang: Data curation; Validation; Investigation; Writing—original draft. Qingzhu Zheng: Data curation; Validation; Investigation; Writing—original draft. Luyao Wu: Data curation; Validation; Investigation; Writing—original draft. Canhui Sun: Methodology; Writingoriginal draft; Project administration. Zhenpeng Peng: Methodology; Project administration; Writing—review and editing. Minhu Chen: Methodology; Project administration; Writing—review and editing. Ziping Li: Methodology; Project administration; Writing—review and editing. Rui Feng: Methodology; Writing—review and editing. Yijun Zhu: Methodology; Writing—review and editing. Yangdi Wang: Conceptualization; Data curation; Formal analysis;

Validation; Investigation; Methodology; Writing—original draft; Writingreview and editing. Zhoulei Li: Conceptualization; Data curation; Formal analysis; Validation; Investigation; Methodology; Writing—original draft; Writing—review and editing. Ren Mao: Conceptualization; Data curation; Formal analysis; Supervision; Funding acquisition; Validation; Investigation; Methodology; Writing—original draft; Project administration; Writing—review and editing. Shi-Ting Feng: Conceptualization; Data curation; Formal analysis; Supervision; Funding acquisition; Validation; Investigation; Methodology; Writing—original draft; Project administration; Writing—review and editing. Source data underlying figure panels in this paper may have individual authorship assigned. Where available, figure panel/source data authorship is listed in the following database record: biostudies:S-SCDT-10_1038-S44321024-00129-8.

Disclosure and competing interests statement

The authors declare no competing interests.

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. Creative Commons Public Domain Dedication waiver http://creativecommons.org/publicdomain/zero/1.0/ applies to the data associated with this article, unless otherwise stated in a credit line to the data, but does not extend to the graphical or creative elements of illustrations, charts, or figures. This waiver removes legal barriers to the re-use and mining of research data. According to standard scholarly practice, it is recommended to provide appropriate citation and attribution whenever technically possible.

© The Author(s) 2024