Enterococcus faecalis hijacks FABP2 to activate quorum-sensing signals and aggravate Crohn’s disease by inducing gut dysbiosis

Download video file ^{(49.2MB, mp4)}

DOI: 10.1136/gutjnl-2024-334604.video01

Abstract

Background

Crohn’s disease (CD) is a chronic inflammatory disorder characterised by intestinal dysbiosis. While inflammation-induced leakage of host proteins is a known phenomenon in CD, how these proteins affect the gut microbiota and contribute to dysbiosis remains unclear. One hypothesis is that commensal bacteria hijack these proteins, exacerbating inflammation in CD.

Design

To investigate host-microbiota interactions in CD, we measured fatty acid-binding protein 2 (FABP2) levels in patients with CD and in mouse models of dextran sulfate sodium induced enteritis and interleukin 10 knockout spontaneous enteritis. Proteomic approaches, including bacterial pull-down and mass spectrometry, were employed to identify commensal targets of FABP2. Functional studies were conducted using wild type and EF3041-deficient Enterococcus faecalis strains, along with α-FABP2 antibody treatment, to assess their effects on intestinal inflammation and microbiota composition.

Results

FABP2 levels were elevated in plasma and faeces of patients with CD, as well as in the mouse models. This was accompanied by dysbiosis of gut commensal bacteria. E. faecalis hijacked luminal FABP2 to promote its proliferation via pheromone-binding protein EF3041, which activated quorum-sensing pathways. Deletion of EF3041 abolished this response, while complementation with EF3041 restored it. Injection of α-FABP2 antibody or transplantation of ΔEF3041 mutant strain significantly reduced epithelial damage, mitigated dysbiosis and alleviated inflammation and symptoms of enteritis in mice.

Conclusion

This study reveals a novel mechanism by which commensal bacteria use host-derived FABP2 to drive dysbiosis and worsen CD pathology. Targeting the FABP2-EF3041 axis may offer new diagnostic and therapeutic avenues for managing CD.

WHAT IS ALREADY KNOWN ON THIS TOPIC

• Gut dysbiosis is a key contributor to IBD pathogenesis.

• Commensal gut bacteria play a central role in driving intestinal inflammation.

• Intestinal inflammation is often accompanied by leakage of epithelial-derived proteins into the lumen.

WHAT THIS STUDY ADDS

• Host-microbiota interactions involving secreted host proteins such as fatty acid-binding protein 2 (FABP2) are more prevalent than previously recognised.

• FABP2 is significantly elevated in the plasma and faeces of patients with Crohn’s disease (CD).

• Secreted FABP2 promotes the growth of gut commensals such as E. faecalis to aggravate dextran sulfate sodium induced and interleukin 10 knockout spontaneous enteritis.

• E. faecalis hijacks FABP2 through the pheromone-binding protein EF3041, thereby activating quorum-sensing pathways that exacerbate intestinal inflammation.

• Administration of an EF3041-deficient Enterococcus strain or α-FABP2 antibody mitigates dysbiosis and alleviates enteritis symptoms.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

• These findings establish FABP2 as a host-derived modulator of dysbiosis and establish Enterococcus-derived EF3041 as a microbial sensor in driving IBD pathogenesis.

• Targeting the FABP2-EF3041 axis represents a promising therapeutic strategy for managing CD.

Introduction

Gut commensal bacteria are known to engage in intricate interactions with intestinal epithelial cells (IECs) and play a crucial role in modulating intestinal barrier function.^{1 2} Dysbiosis of these commensals has been associated with IBD and increased susceptibility to pathogens.^{3 4} However, the mechanisms that govern the balance between commensal bacteria and host factors remain poorly understood. Recent studies suggest that disruptions in epithelial integrity and shifts in microbial load are among the critical factors destabilising this balance, thereby triggering immune activation and increasing the risk of diseases such as IBD and colorectal cancer.^{4 5} These findings indicate that yet unidentified links between epithelial damage and gut commensal dysbiosis may be essential for maintaining homoeostasis within the gut microecosystem.

The fatty acid-binding protein (FABP) family consists of highly conserved cytoplasmic proteins that mediate intracellular lipid transport, storage and metabolism, particularly of long-chain fatty acids (LCFAs).^{6 7} Among these, FABP2 is the predominant isoform in the small intestine.⁸ It facilitates the uptake of lipids from the intestinal lumen into enterocytes and binds excess fatty acids to maintain epithelial lipid homoeostasis.⁹ During intestinal epithelial injury, FABP2 is released into both the lumen and circulation, resulting in elevated plasma levels, and has thus been proposed as a biomarker for intestinal damage. Given the overlap between the small intestine’s role as the primary site of dietary fatty acid absorption and as a hub for gut microbiota, investigating how fluctuations in FABP2 levels influence gut homoeostasis and potentially contribute to IBD pathogenesis represents a promising direction for future research.

Enterococcus faecalis is a ubiquitous member of the gut commensal bacteria in healthy humans and animals. This remarkably adaptable bacterium thrives in diverse environments, ranging from soils to surface water and even extreme conditions. While certain E. faecalis strains serve as probiotics, others demonstrate significant pathogenic potential, frequently causing both community-acquired and nosocomial infections such as endocarditis, bacteraemia and urinary tract infections.¹⁰ The dual nature of E. faecalis commensalism remains incompletely understood. Notably, patients with Crohn’s disease (CD) often exhibit elevated E. faecalis abundance and increased anti-E. faecalis antibody levels,^{11 12} supporting a pathogenic role. This is further supported by studies showing that germ-free interleukin (IL) 10 knockout (KO) mice colonised with E. faecalis develop chronic IBD-like symptoms,¹³ highlighting its proinflammatory potential. E. faecalis employs sophisticated survival strategies, including competition for ecological niches, binding sites and nutrients, allowing it to influence microbial community dynamics.^{14 15} These adaptations shape its cellular growth, morphology and stress responses, enabling its function plasticity as either a commensal or a pathogen. The mechanisms driving this transition between symbiotic and pathogenic states may significantly contribute to gut dysbiosis and disrupt internal homoeostasis, potentially exacerbating IBD progression through pathways that extend beyond currently characterised regulatory mechanisms.

In this study, we identified a novel ‘seesaw’ mechanism by which host FABP2 is hijacked to drive gut microbiota dysbiosis, promoting the transition of Enterococcus from a probiotic commensal to an opportunistic pathogen state. These findings highlight a promising therapeutic strategy for re-establishing gut microbiota homoeostasis in clinical settings.

Results

The interactome between the gut microbiota and the intestinal epithelium is widely present in the intestinal lumen

Clinical evidence consistently demonstrates that chronic inflammatory conditions like CD, a condition that affects the entire gastrointestinal tract, are associated with gut microbiota dysbiosis.^{4 16} In such pathological states, commensal bacteria are likely to interact with dietary nutrients and host cellular debris to enhance their proliferation, thereby exacerbating intestinal imbalance. To investigate these host-microbe interactions, we developed a novel bacteria-dependent pull-down assay to characterise the microbiota-epithelial interactome (figure 1A). We selected representative gut microbes (Lactococcus lactis, Salmonella typhimurium and Citrobacter rodentium) and incubated them with mouse intestinal homogenate to simulate host-microbe interactions, using brain lysate as a negative control (online supplemental figure S1A). Given the zwitterionic nature of interactome proteins,¹⁷ we employed pH-dependent elution to separate bound complex, as electrostatic forces predominantly mediate these interactions (figure 1A,B). Our results demonstrate extensive microbe-host protein interactions throughout the intestinal tract (figure 1B and online supplemental figure S1B), validated by IgA and other Igs immunoblotting (figure 1C) and negative controls (online supplemental figure S1C). These findings indicate that the observed crosstalk stems from specific protein interactions rather than intracellular bacterial contents. The interactome was further confirmed in primary IECs (figure 1D and online supplemental figure S1D,E), colon cancer cell lines HT29 and normal colon epithelial cell NCM460 (figure 1E), demonstrating epithelial surface localisation. The phenomenon extended to multiple bacterial species (E. faecalis, Listeria monocytogenes, Bacillus subtilis, Escherichia coli and Staphylococcus epidermidis) (online supplemental figure S1F), with bacterial integrity maintained throughout (online supplemental figure S1C,G), suggesting an evolutionarily conserved survival mechanism. Mass spectrometry (MS) analysis further revealed a diverse yet overlapping interactome across bacterial strains and intestinal regions (figure 1F, online supplemental figure S1H,I and onlinesupplemental tables S1 S2), indicating shared but previously uncharacterised mechanisms. However, only limited proteins mediated these strain-specific interactions (figure 1G, online supplemental figure S1J and online supplemental table S2). Taken together, the ubiquity of this microbiota-epithelial interactome highlights the need to elucidate why and how commensal microbes engage specific host proteins, particularly given their established roles in gut homoeostasis.

Figure 1. Identification of the interactome between the gut microbiota and host. (A) Schematic overview of the experiment used to identify the host-microbiota interactome. (B) Coomassie blue staining (CBB) showing interaction between host proteins and strains. The indicated bacterial strains were incubated with mouse intestinal homogenate overnight at 4°C and washed with elution buffer at pH 2.0 (upper) or pH 9.0 (lower). (C) Validation of the laboratory-developed bacterial pull-down enrichment method using IgG and other Igs as positive controls. (D–E) Bacterial interactions with IECs were confirmed using primary mouse IECs (D) and immortalised human IEC lines HT29 and NCM460 (E). Cell lysates were incubated with Salmonella enterica, Citrobacter rodentium and Lactococcus lactis followed by washing with elution solutions at different pH levels. (F) MS analysis of the interactome between L. lactis, S. enterica and C. rodentium and the small intestinal homogenate. (G) Fatty acid-binding protein 2 (FABP2) was identified as a candidate host protein by overlapping the interactomes of S. enterica, C. rodentium and L. lactis with mouse small intestinal IECs. GC-MS, gas chromatography-mass spectrometry; IECs, intestinal epithelial cells; SI, small intestine.

FABP2 is highly expressed in patients with CD and mice with DSS-induced enteritis

Given the widespread nature of these luminal interactions, we focused our analysis on host proteins common to multiple bacterial strains, comparing Lactococcus, Bacillus and Salmonella. This screening identified FABP2 as a prominent component candidate of the IEC interactome (figure 1G and online supplemental table S2). Given FABP2’s established role in LCFAs transport, its expression pattern in both wild type (Fabp2^+/+) and FABP2-deficient (Fabp2^-/-) mice was characterised (online supplemental figure S2). Three FABP family members (FABP1, FABP2 and FABP6) were detected in the small intestine (online supplemental figure S3A,B), each showing distinct regional distribution (online supplemental figure S4). FABP2 demonstrated the highest expression levels (online supplemental figure S3A,C) and exhibited broad epithelial localisation throughout the intestinal tract, with particularly strong expression in the small intestine, followed by the cecum and colon (online supplemental figure S3A,B and S4A–C). In contrast, FABP1 was primarily restricted to the proximal small intestine, while FABP6 was restricted to the distal small intestine (online supplemental figure S3A and S4D). Notably, FABP2 expression was absent in non-digestive tissues (online supplemental figure S3B) and showed minimal variation across genetic backgrounds (online supplemental figure S3C), underscoring its intestine-specific function. Moreover, cellular localisation studies revealed predominant FABP2 expression in the IECs, with significantly lower levels in intestinal intraepithelial lymphocytes and intestinal lamina propria lymphocytes (online supplemental figure S3D), consistent with its identification in the microbiota-host protein interactome (figure 1D). Furthermore, immunofluorescence (IF) staining with ulex europaeus agglutinin-I and lysozyme markers confirmed FABP2’s absence in goblet or Paneth cells (online supplemental figure S3E,F).¹⁸ Importantly, we observed no surface interactions between FABP1/FABP6 and bacterial components in dextran sulfate sodium (DSS)-treated mice (online supplemental figure S3G). Supporting a functional role in mucosal defence, Fabp2^-/- mice displayed attenuated DSS-induced enteritis symptoms (figure 2A–C), suggesting FABP2 operates at the host-microbiota interface to maintain intestinal barrier integrity.

Figure 2. FABP2 is highly expressed in patients with CD and mice with DSS-induced enteritis. (A–C) Comparison of the disease severity between Fabp2^+/+ and Fabp2^-/- mice after DSS treatment, showing body weight changes (A) and colon length variation (B). Representative images of the colon (C) are shown. Scale bar, 1 cm. (D–E) Comparison of FABP2 protein expression in inflammatory and non-inflammatory regions of small intestine tissues from patients with CD, assessed using immunofluorescence (IF) staining (D) and immunoblotting test (E). Ulex europaeus agglutinin-I (UEA-I) (D) was used to identify goblet and Paneth cells. Nuclei were counterstained with 4′,6-diamidino-2-phenylindole (DAPI). Representative images of FABP2 expression are shown. (F–G) Quantification of FABP2 protein levels in plasma (F) and faeces (G) from patients with CD, measured by using ELISA. (H) FABP2 expression in colon samples was analysed via immunoblotting, and relative expression was quantified using ImageJ. (I) Flow cytometry analysis showing the binding of FABP2 to gut microbiota in faecal samples at various time points during DSS administration. (J) ELISA quantification of FABP2 levels in mouse faeces over time with DSS treatment. Data are presented as the mean±SD. Statistical significance was determined using one-way analysis of variance (ANOVA) with multiple comparisons. *, p<0.05; **, p<0.01; ***, p<0.001; ****, p<0.0001; ns, not statistically significant. CD, Crohn’s disease; DSS, dextran sulfate sodium; Erk, extracellular regulated protein kinases; FABP2, fatty acid-binding protein 2; GAPDH, glyceraldehyde-3-phosphate dehydrogenase.

Next, we evaluated FABP2’s expression and distribution in clinical samples from patients with CD (online supplemental table S3). Notably, higher FABP2 expression in non-inflamed versus inflamed small intestine regions of patients with CD was observed by IF staining (figure 2D and online supplemental figure S5A,B) and immunoblotting analysis (figure 2E). Consistent with this, plasma FABP2 concentrations were elevated in the patients with CD (figure 2F and online supplemental table S3), indicating increased protein release during chronic epithelial injury. Faecal FABP2 levels were similarly elevated (figure 2G), suggesting its presence in the faeces originates from the damaged intestinal epithelium and is subsequently shed into the intestinal lumen. DSS-treated mice also exhibited increased FABP2 levels compared with untreated controls, with temporal analysis of colon tissues revealing dynamic expression changes during disease progression (figure 2H and online supplemental figure S5C). A similar conclusion was drawn from the analysis of faeces from the mice with enteritis. FABP2 protein accumulated in the faeces at the beginning of DSS treatment, continued to decrease in the mid-phase of enteritis, and increased again to the highest level in the later phase (figure 2I,J). Together, these dynamic changes in tissue and faecal FABP2 levels reflect both altered protein expression and microbiota interactions, strongly supporting FABP2’s critical role in intestinal disease.

FABP2 exacerbates intestinal inflammation by promoting gut microbial overgrowth

To elucidate the physiological role of FABP2 in CD pathogenesis, we combined DSS-induced enteritis with intragastric FABP2 administration.¹⁹ As expected, FABP2 significantly exacerbated disease severity (figure 3A, top), as evidenced by pronounced weight fluctuations (figure 3A, bottom), enhanced colon inflammation (figure 3B), and elevated enteroscopic indices (figure 3C), and upregulated inflammatory cytokines (figure 3D), implying that FABP2 exacerbates inflammation in these mice. Microbial community analysis revealed FABP2-mediated dysbiosis with 16S rDNA sequencing (16S rDNA-seq), demonstrating phylum-level and genus-level restructuring in both the small intestine and colon (figure 3E,F), particularly showing expansions of Enterobacteriaceae (unclassified) and Enterococcus populations (figure 3F). The most pronounced microbiota alteration was in the small intestine (figure 3G). However, Fabp2^+/+ and Fabp2^-/- mice showed comparable intestinal villus morphology (online supplemental figure S6), daily activities and metabolic indices (online supplemental figure S7), suggesting that DSS-induced FABP2 upregulation rather than developmental difference drives enteritis pathology. Moreover, in vitro experiments demonstrated FABP2’s direct, dose-dependent stimulation of microbial proliferation in culture (figure 3H,I, online supplemental figure S8A–C). Most strikingly, pretreatment with FABP2 induced Enterococcus-dominated dysbiosis in vivo (figure 3J) prior to observed tissue damage (online supplemental figure S9D), demonstrating that FABP2 alone can disrupt microbial homoeostasis independently of overt inflammation. Furthermore, our analysis revealed that FABP2 interacts with multiple commensal and beneficial bacteria, including Akkermansia, Enterococcus, Bifidobacterium, Faecalibaculum, Muribaculaceae and Lactobacillus (figure 3K), potentially driving microbial imbalance through growth promotion. In vivo bromodeoxyuridine incorporation assays (figure 3L and online supplemental figure S8D) demonstrated that luminal FABP2, released during DSS-induced damage, actively stimulates bacterial proliferation. In vitro studies with mCherry-tagged FABP2 showed species-specific binding patterns across model bacteria (E. coli, L. lactis, S. enterica and E. faecalis) (online supplemental figure S10A), with particularly strong affinity for Enterococcus in dose-response experiment (online supplemental figure S10B). These findings correlate with the dynamic FABP2-microbiota interactions observed in vivo (figure 2I,J). Therefore, both in vitro and in vivo data (figure 3H–L) consistently demonstrated that FABP2 significantly enhances bacterial growth rates, supporting the hypothesis that gut commensals co-opt FABP2 to fuel their expansions. This FABP2-mediated dysbiosis appears to originate in the small intestine during early inflammation, subsequently disrupting homoeostasis throughout the entire gut ecosystem.

Figure 3. FABP2 exacerbates intestinal inflammation during DSS administration. (A–C). C56BL/6 mice were treated with DSS, combined with the administration of 0.1 mg GST control protein or FABP2 protein per 20 g body weight, respectively. Body weight changes (A), colon length (B) and enteroscopy results (C) were recorded. Representative images of colon enteroscopy are shown. (D) Cytokines expression levels in colon tissue and blood through the course of DSS administration were compared. (E–F) Phylum-level (E) and genus-level (F) analysis of the 16S rDNA sequencing (16S rDNA-seq) results conducted on faecal samples collected from the small intestine and colon of mice who received GST or FABP2 during DSS administration were compared. GST was set as the control group of the FABP2 group. (G) The abundance of the microbiota in the small intestine with or without FABP2 protein during DSS administration was compared. (H) FABP2 promotes gut microbiota growth in vitro. Growth was assessed in Luria-Bertani medium with various concentrations of FABP2. Each symbol represents independent biological replicates and data from three independent biological replicates are expressed as the mean±SD. (I) 16S rDNA-seq results from in vitro cultured faecal samples with FABP2 supplementation to evaluate microbial changes. (J) Genus-level taxonomic profiling of faecal sample via metagenomic sequencing (Meta-seq) following FABP2 and DSS treatment. (K) Magnetic separation and immunoprecipitation of the faecal samples from the small intestine and colon of DSS-treated enteritis mice using α-FABP2 antibody followed by 16S rDNA-seq. Genus-level analysis of the 16S rDNA-seq results is shown. (L) FABP2 promotes gut microbiota proliferation in vivo. BrdU was added to DSS-containing drinking water, and BrdU incorporation was detected using flow cytometry to evaluate proliferation. Representative results of BrdU staining analysis are shown. Data are presented as the mean±SD. Statistical significance was determined using one-way ANOVA with multiple comparisons. *, p<0.05; **, p<0.01; ***, p<0.001; ****, p<0.0001; ns, not statistically significant. ANOVA, analysis of variance; BrdU, bromodeoxyuridine; DSS, dextran sulfate sodium; FABP2, fatty acid-binding protein 2; GST, glutathione S-transferase; IFN-γ, interferon-γ; i.g., oral gavage.

We further investigated the impact of FABP2 on CD pathogenesis using IL-10 KO mice, a widely used spontaneous mouse model of chronic enteritis. Following intragastric administration of FABP2 over a 15-day period (figure 4A), we observed significant exacerbation of disease parameters, including progressive body weight loss (figure 4B), aggravated inflammation in the intestinal tract (figure 4C and online supplemental figure S11A), and worsened histopathological features (figure 4D and online supplemental figure S11B). Intestinal tissues from FABP2-treated IL-10 KO mice showed increased expression of inflammatory cytokines (figure 4E), accompanied by elevated systemic FABP2 levels (figure 4F). Flow cytometry analysis revealed a higher proportion of FABP2-positive bacteria in the faeces (figure 4G). The direct interaction between FABP2 and gut microbiota was further confirmed through IF and fluorescence in situ hybridisation (FISH) analysis of intestinal tissues (figure 4H,I and online supplemental figure S11A–E). Together, these findings demonstrate that exogenous FABP2 drives gut dysbiosis not only in the DSS-induced enteritis model but also in the IL-10 KO spontaneous enteritis model.

Figure 4. Administration of FABP2 protein aggravates colitis symptoms in IL-10 KO spontaneous enteritis mouse model. (A) Schematic overview of FABP2 protein administration strategy in the IL-10 KO mouse model. (B–C) IL-10 KO mice were administrated purified FABP2 protein on designated days and the body weight changes (B), colon length (C), and H&E staining results (D) are compared. Representative images of colon and H&E staining after administration of FABP2 protein are shown. Scale bar, 1 cm. (E) Cytokines expression following FABP2 treatment in the IL-10 KO mouse model was assessed via RT-qPCR. (F) FABP2 expression in colon faecal samples was analysed via immunoblotting before and after the FABP2 treatment in IL-10 KO mice. (G) The proportion of FABP2-positive bacteria via flow cytometry in the proximal and distal parts of the small intestine, cecum and colon from the IL-10 KO mice is a comparison. Representative flow charts are shown. (H) IF staining of FABP2 (red) and peptidoglycan (green) showing FABP2 expression and bacterial load, respectively, in the small intestine of IL-10 KO mice without or with FABP2 protein. Representative images are shown. Scale bar, 100 µm. (I) Gut microbial load and Enterococcus abundance in the small intestine were visualised using FISH staining with a general 16S rRNA probe or an E. faecalis-specific 16S rRNA probe with the administration of FABP2 protein in IL-10 KO mice. Representative images are shown. Scale bar, 100 µm. Data are expressed as the mean±SD. One-way ANOVA multiple comparisons test. *, p<0.05; ***, p<0.001; ****, p<0.0001. ns, not statistically significant. ANOVA, analysis of variance; DAPI, 4′,6-diamidino-2-phenylindole; FABP2, fatty acid-binding protein 2; FISH, fluorescence in situ hybridisation; IF, immunofluorescence, IFN-γ, interferon-γ; i.g., oral gavage; KO, knockout; RT-qPCR, real-time quantitative PCR.

E. faecalis hijacks FABP2 to promote its proliferation and aggravate intestinal dysbiosis

Analysis of faecal samples from DSS-treated mice revealed that Enterococcus was the predominant genus among FABP2-associated bacteria in the small intestine, showing significantly higher abundance compared with colonic samples (figure 3J). This observation led us to select E. faecalis as our model commensal for mechanistic investigations. Compared with other standard strains, E. faecalis exhibited the most pronounced growth enhancement on FABP2 supplementation (online supplemental figure S10A–E and onlinesupplemental tables S1 S2). Moreover, a greater difference in growth was observed with a higher concentration of FABP2 in the rich culture medium (online supplemental figure S10B), mirroring the dynamic luminal FABP2-microbiota interactions observed in vivo (figure 2I,J). Furthermore, flow cytometry analysis confirmed dynamic FABP2 binding to E. faecalis surface (online supplemental figure S10C). These findings suggest that E. faecalis has evolved an efficient mechanism to use epithelium-derived FABP2 for competitive growth advantage, potentially disrupting the proliferation balance of other gut microbiota species.

Comparative analysis of E. faecalis distribution in Fabp2^+/+ and Fabp2^-/- mice provided compelling evidence for FABP2’s role in microbial regulation. FISH staining with both a general 16S rRNA probe and an E. faecalis-specific probe revealed significantly fewer bacteria, particularly E. faecalis, in the small intestine of Fabp2^-/- mice regardless of DSS treatment (online supplemental figure S10F and online supplemental figure S12A). This consistent reduction suggests FABP2’s luminal presence is critical for E. faecalis colocalisation. A similar conclusion was drawn from IF staining with FABP2 and peptidoglycan antibodies following DSS treatment (online supplemental figure S10G and online supplemental figure S12B). Combined with the attenuated enteritis symptoms observed in Fabp2^-/- mice (figure 2A–C), these findings strongly implicate that FABP2-mediated E. faecalis overgrowth is a crucial driver of gut microbiota dysbiosis. Together, these results demonstrate that luminal FABP2 secretion promotes commensal bacteria overgrowth, particularly of Enterococcus species, which significantly exacerbates dysbiosis in DSS-induced enteritis.

The molecular mechanism by which FABP2 enhances E. faecalis growth

To explore how Enterococcus hijacks FABP2 in the small intestinal lumen, FABP2-bound immunoprecipitates from E. faecalis cell lysates were incubated overnight followed by MS identification (figure 5A, online supplemental figure S13A,B and online supplemental table S4). ATP-binding cassette (ABC) transporters stand out as candidate proteins, consistent with their known functions in nutrient and signalling molecule transport.²⁰ Among these, the pheromone-binding protein EF3041 emerged as a prime candidate due to its specific affinity for FABP2, supported by both in vitro (figure 3H and online supplemental figure S8A,B) and in vivo (figure 3J–L) evidence of FABP2-mediated bacterial growth promotion. Although EF3041 does not share high sequence identity with OppA (online supplemental figure S13C and S14), an oligopeptide-binding protein in the Opp transporter system (OppABCDF),²¹ we hypothesised it might functionally resemble OppA by recognising FABP2 as a pheromone-like ligand. Genetic deletion of EF3041 (ΔEF3041) (online supplemental figure S13D) abolished FABP2-induced growth promotion in minimal medium (figure 5B, online supplemental figure S10E, online supplemental figure S13E–I and S15), while complementation with EF3041 restored this phenotype (figure 5C), confirming EF3041’s essential role. Scanning electron microscopy (SEM) revealed FABP2-induced macrocolony formation in wild type but not ΔEF3041 strains (figure 5D), suggesting EF3041 mediates social behaviours in response to FABP2. During enteritis progression, EF3041 expression was significantly upregulated throughout the intestinal tract (figures3A 5E), supporting its role in FABP2-mediated dysbiosis. Moreover, transmission electron microscopy revealed FABP2 reduced membrane vesiculation, a bacterial stress response to nutrient limitation (figure 5F).²² Importantly, gas chromatograph/high performance liquid chromatography (GC/HPLC)-MS analysis demonstrated comparable fatty acid profiles between wild type and ΔEF3041 strains regardless of the treatment with FABP2 or cholate, a membrane-disrupting agent (figure 5G),^{14 23} indicating FABP2’s effects are independent of its canonical fatty acid transport function.

Figure 5. Molecular mechanism by which FABP2 enhances E. faecalis growth. (A) FABP2-binding proteins on the surface of E. faecalis were screened using FABP2-mediated immunoprecipitation followed by MS identification. (B–C) Growth comparison of EF3041 KO and EF3041-complementation E. faecalis strains in the presence or absence of FABP2. (D) Scanning electron microscopy (SEM) analysis of the physiological change in wild type and EF3041-deficient mutant E. faecalis grown in the M9 minimal medium with or without FABP2. Representative images of E. faecalis are shown. Scale bar, 10 µm. (E) The transcription of the EF3041 gene of E. faecalis was monitored in the intestine of the mice treated with dextran sulfate sodium (DSS) at indicated days, and the E. faecalis 16S rRNA was set as the reference gene. (F) Transmission electron microscopy (TEM) analysis of the physiological change in wild type and EF3041-deficient mutant E. faecalis grown in the M9 minimal medium with or without FABP2 representative images of the E. faecalis are shown. Scale bar, 200 nm. (G) Intracellular fatty acid levels of E. faecalis were compared as challenging with cholate by adding with FABP2 in the medium. (H–K) RNA sequencing (RNA-seq) analysis of wild type (H) and EF3041-null mutant E. faecalis strains (J) cultured in the M9 minimal medium following FABP2 treatment are shown. The transcription levels of genes in the quorum-sensing signal pathway determined by RNA-seq of E. faecalis (I) and the EF3041-deficient mutant (K) after FABP2 treatment were compared. (L) RT-qPCR analysis of quorum-sensing signal coding genes including codY, cylL, cylM, fsrB and gelE in FABP2-treated E. faecalis was monitored, and the E. faecalis 16S rRNA was set as the reference gene. Data are presented as the mean±SD. Statistical significance was determined using one-way ANOVA with multiple comparisons test. ****, p<0.0001; ns, not statistically significant. ABC, ATP-binding cassette; ANOVA, analysis of variance; BHI, brain heart infusion; FABP2, fatty acid-binding protein 2; FFA, free fatty acids; KO, knockout; MS, mass spectrometry; RT-qPCR, real-time quantitative PCR.

The role of FABP2 in regulating the growth of E. faecalis was further examined using bacterial RNA sequencing (RNA-seq) analysis. Genes related to the ABC transporters were highly upregulated on the addition of FABP2 to the minimal medium (figure 5H and online supplemental figure S16A,B). Notably, genes involved in the quorum-sensing signalling pathway were also markedly upregulated under this condition (figure 5H,I and online supplemental figure S16C). However, this increase in expression was absent in the ΔEF3041 mutant (figure 5J,K and online supplemental figure S16D–F). Activation of genes in the OppABCDF transporter system, including oppA, oppB, oppC, oppD and oppF, and in the Sec protein translocase complex, including secC, secG and yajC, which are responsible for the export of a subset of secretory proteins, such as virulence factors,²⁴ was observed in wild type E. faecalis but not in ΔEF3041 mutant (figure 5I,K and online supplemental figure S17A). This suggests that FABP2 promotes the E. faecalis overgrowth by acting in a pheromone-like manner.

Furthermore, our transcriptional analysis revealed that FABP2 specifically activated the key quorum-sensing genes (codY, cylL, cylM and gelE) in the wild type strain, while this activation was completely abolished in the ΔEF3041 mutant. Notably, gelE, encoding the gelatinase GelE critical for biofilm formation, microbial interactions and dissemination,²⁵ showed an FsrB-dependent induction (figure 5L). These findings strongly support that FABP2 functions as a host-derived pheromone analogue, activating downstream quorum-sensing circuits to drive E. faecalis expansion. Additionally, other ABC transporters and pheromone-binding protein genes exhibited only marginal expression changes following FABP2 exposure (online supplemental figure S17A–C), suggesting that the partial growth response observed in ΔEF3041 mutant may result from compensatory expression of these genes in response to FABP2 (figure 5B,C). Collectively, these results indicate that the pheromone-binding protein EF3041 serves as the primary receptor for FABP2-mediated signalling, thereby promoting the proliferation of E. faecalis over other bacteria in the intestines of mice with DSS-induced enteritis and IL-10 KO spontaneous enteritis.

Transplantation of Enterococcus deficient in EF3041 alleviates enteritis symptoms in mice

Given that the EF3041 functions as an accessory component required for activating the OppABCDF transporter, which senses environmental variations in pheromones and oligopeptides, and considering its evolutionary conservation across both Gram-negative and Gram-positive bacteria (despite its unique role in regulating Enterococcus conjugation),²⁶ we hypothesised that deletion of the EF3041 gene would significantly reduce E. faecalis responsiveness to FABP2 (online supplemental figure S15), thereby alleviating enteritis symptoms in mouse models (figures67). To investigate how FABP2 influences E. faecalis growth and exacerbates CD pathogenesis, mice were treated with antibiotics in their drinking water starting on day 0, and disease progression was evaluated on day 17. Subsequently, mice were intragastrically administered 10⁸ colony forming unit (CFU) bacteria per mouse every 3 days. Notably, administration of wild type E. faecalis significantly exacerbated disease progression in both DSS-mediated enteritis and IL-10 spontaneous enteritis models (figures6A 7A), including marked body weight loss (figures6B 7A), pronounced shortening of colon (figures6C 7B) and small intestine (online supplemental figure S18A), elevated disease activity index (figures6D 7C and online supplemental figure S18B), and increased expression of inflammatory cytokines (figure 7D). In contrast, mice treated with ΔEF3041 mutant exhibited significantly attenuated symptoms, confirming the critical role of EF3041 in mediating FABP2-dependent pathogenesis. Furthermore, flow cytometry analysis showed that fewer Gram-positive bacteria were observed in the intestinal faeces (figure 7E and online supplemental figure S18C). IF staining with α-FABP2 and α-peptidoglycan antibodies also revealed fewer Gram-positive bacteria in the intestines and along the intestinal epithelium in the ΔEF3041 group (figures6E 7F, online supplemental figure S18D and 19A). Importantly, faecal FABP2 levels remained comparable between groups (figure 6F), indicating that the reduction in bacterial load resulted from impaired bacterial proliferation, not from limited FABP2 availability. FISH staining using a general 16S rRNA probe and an Enterococcus-specific probe further confirmed a marked reduction in Enterococcus colonisation in ΔEF3041-treated mice (figures6G 7G, online supplemental figure S18E and S19B). These findings establish the essential role of EF3041 in mediating pheromone transporter signalling through the OppABCDF system, regulating quorum-sensing pathways and governing Enterococcus proliferation within the host gut microenvironments. Together, these results highlight EF3041 as a promising therapeutic target for restoring Gram-positive microbiota homoeostasis, spatial localisation, and pheromone-mediated interspecies communication. These findings suggest that targeting EF3041 could remodel gut dysbiosis and improve host-microbe interactions.

Figure 6. Transplantation of EF3041-deficient mutant of E. faecalis alleviates symptoms after the administration of DSS. (A) Schematic of the transplantation strategy for wild type and the EF3041 knockout (KO) E. faecalis strains in the DSS model. (B–D). The wild type and ΔEF3041 mutant E. faecalis strains were transplanted into C57BL/6 mice at defined time points during DSS administration. Body weight (B), colon length (C) and Disease Activity Index (D) were recorded. Representative images of the colon (C) are shown. Scale bar, 1 cm. (E) IF staining of FABP2 (red) and peptidoglycan (green) showing FABP2 expression and bacterial load, respectively, in the small intestine of mice treated with phosphate buffer saline (PBS), wild type or ΔEF3041 mutant E. faecalis strains during DSS administration. Representative images are shown. Scale bar, 100 μm. (F) Immunoblotting analysis of FABP2 protein levels in faecal samples from mice treated with PBS, wild type or ΔEF3041 mutant E. faecalis strains on days 0, 3, 6 and 8 of DSS exposure. (G) Gut microbiota load and E. faecalis abundance in the small intestine were visualised using FISH staining with a general 16S rRNA probe or an E. faecalis-specific 16S rRNA probe, following transplantation of wild type or ΔEF3041 mutant E. faecalis strains and DSS administration. PBS treatment was set as the control treatment for i.g. E. faecalis treatment. Representative images are shown. Scale bar, 100 μm. Data are presented as the mean±SD. Statistical significance was determined using one-way ANOVA with multiple comparisons test. *, p<0.05; ***, p<0.001; ****, p<0.0001. ABX, antibiotics cocktail; ANOVA, analysis of variance; CFU, colony forming unit; DAPI, 4′,6-diamidino-2-phenylindole; DSS, dextran sulfate sodium; FABP2, fatty acid-binding protein 2; FISH, fluorescence in situ hybridisation; IF, immunofluorescence; i.g., oral gavage; KO, knockout.

Figure 7. Transplantation of EF3041-deficient E. faecalis alleviates symptoms in the IL-10 KO spontaneous enteritis mouse model. (A–C) Schematic of the transplantation strategy for wild type and EF3041 KO E. faecalis strains in the IL-10 KO mouse model. Body weight changes (A), colon length variation (B) and H&E staining results (C) were shown at defined time points after antibiotics treatment. Representative images of colon (B) and H&E staining (C) are shown. Scale bar, 1 cm. (D) Cytokines expression following wild type and ΔEF3041 mutant E. faecalis strains in IL-10 KO mouse model was investigated using RT-qPCR. (E) Flow cytometry analysis of proportion of FABP2-positive bacteria in the proximal and distal parts of small intestine, cecum and colon of the IL-10 KO mouse model treated with transplantation of wild type and ΔEF3041 mutant E. faecalis strains. Representative flow charts are shown. (F) IF staining of α-peptidoglycan (green) and α-FABP2 (red) antibodies in the small intestine showing FABP2 expression and bacterial abundance, respectively, following treatment with wild type and ΔEF3041 mutant E. faecalis strains in IL-10 KO mouse model. Representative images are shown. Scale bar, 100 µm. (G) Gut microbiota load and E. faecalis abundance in the small intestine were determined by FISH staining with a general 16S rRNA probe or an E. faecalis-specific 16S rRNA probe following transplantation of wild type and ΔEF3041 mutant E. faecalis strains in the IL-10 KO mice model. Representative images are shown. Scale bar, 100 µm. Data are presented as the mean±SD. Statistical significance was determined using one-way ANOVA with multiple comparisons test. *, p<0.05; ***, p<0.001; ****, p<0.0001. ANOVA, analysis of variance; CFU, colony forming unit; DAPI, 4′,6-diamidino-2-phenylindole; FABP2, fatty acid-binding protein 2; FISH, fluorescence in situ hybridisation; IFN-γ, interferon-γ; i.g., oral gavage; KO, knockout; RT-qPCR, real-time quantitative PCR.

Neutralisation of FABP2 in the intestinal lumen with an antibody alleviates the enteritis symptoms in mice

To test whether reducing FABP2 could correct gut dysbiosis, we treated mice with α-FABP2 antibody (figures8 9). As expected, α-FABP2 treatment conferred substantial protection against both DSS-induced and IL-10 KO enteritis, including attenuated body weight loss (figures8A 9A), preserved intestinal architecture (colon: figures8B 9B; small intestine: online supplemental figure S21A), improved enteroscopic scores (figure 8C), and reduced histopathological damage (figures8D 9C, and online supplemental figure S21B), compared with DSS or IL-10 KO mice treated with control IgG. Notably, there was no significant difference in body weight loss between α-FABP2-treated and Fabp2^-/- mice (online supplemental figure S20A). Furthermore, FABP2 was undetectable in the intestinal lumen (figures8E 9D, and online supplemental figure S21C) and faeces of the α-FABP2 group either after 7 days of DSS administration (figure 8F) or in IL-10 KO mice (figure 9E), suggesting effective neutralisation of luminal FABP2 and mitigation of disease progression. Inflammatory cytokines analysis and Disease Activity Index also indicated reduced intestinal inflammation in α-FABP2-treated groups (figures8G 9F). IF staining using an α-peptidoglycan antibody showed a lower abundance of Gram-positive bacteria in the α-FABP2 group (figures8H 9G, online supplemental figure S20B and S21D), despite unchanged FABP2 expression in intestinal villi in both DSS-mediated and IL-10 KO mouse models. FISH analysis specific to E. faecalis confirmed that the reduced proliferation of this species was due to decreased FABP2 levels following antibody treatment (figures8I 9H, online supplemental figure S20C and S21E). Notably, improved caecal morphology (figure 8B) further supports that FABP2-driven dysbiosis and total microbial load are reversible through antibody-mediated neutralisation. Metagenomic sequencing (Meta-seq) of faecal samples corroborated these findings, showing a decreased abundance of Enterococcus in α-FABP2-treated mice compared with untreated controls (figures8J 9I and online supplemental figure S21F). Collectively, these data demonstrate that the milder enteritis symptoms observed in α-FABP2-treated mice result from reduced dysbiosis driven by commensal Enterococcus.

Figure 8. Neutralisation of FABP2 with α-FABP2 antibody alleviates symptoms of DSS-induced enteritis in mice. (A–D) Schematic of the FABP2 neutralisation strategy using α-FABP2 antibody during DSS administration in the DSS-induced enteritis mouse model (A). Body weight changes (A), colon length (B), enteroscopy results (C) and H&E staining (D) are shown. Representative colon after the α-FABP2 administration (B) is shown; scale bar, 1 cm. Representative images after the α-FABP2 administration (C and D) are shown; scale bar, 100 μm. (E–F) FABP2 expression in the intestinal lumen was detected by immunohistochemistry (E) and in faecal samples via immunoblotting (F). (G) Cytokine expression was investigated using ELISA following α-FABP2 treatment during DSS administration. (H) IF staining of FABP2 antibody (red) and peptidoglycan antibody (green) in the small intestine showing expression and gut microbiota load, respectively, following α-FABP2 antibody treatment during DSS administration. Representative images are shown. Scale bar, 100 µm. (I) Gut microbiota load and E. faecalis abundance in the small intestine were visualised using FISH with a 16S rRNA general probe (green) or an E. faecalis specific probe (red) after α-FABP2 antibody treatment during administration of DSS. Representative images are shown. Scale bar, 100 μm. (J) Genus-level taxonomic profiling of faecal samples was performed via Meta-seq following α-FABP2 antibody treatment during DSS administration. Representative images are shown. Data are presented as the mean±SD. Statistical significance was determined using one-way ANOVA with multiple comparisons test. *, p<0.05; **, p<0.01; ***, p<0.001; ****, p<0.0001. ANOVA, analysis of variance; DSS, dextran sulfate sodium; FABP2, fatty acid-binding protein 2; FISH, fluorescence in situ hybridisation; IF, immunofluorescence; i.g., oral gavage.

Figure 9. Neutralisation of FABP2 with α-FABP2 antibody alleviates symptoms of IL-10 KO spontaneous enteritis mouse model. (A–C) Schematic of the FABP2 neutralisation strategy using α-FABP2 antibody in IL-10 KO spontaneous enteritis mouse model (A). Body weight changes (A), colon length (B) and H&E staining results (C) following α-FABP2 antibody treatment in the IL-10 KO enteritis model. Representative colon after the α-FABP2 administration (B) is shown; scale bar, 1 cm. Representative H&E images after the α-FABP2 administration (C) are shown; scale bar, 100 µm. (D) The FABP2 positive bacteria in the small intestine (SI-1 and SI-2), cecum, and colon faecal samples of IL-10 KO mice with the treatment of α-FABP2 antibody were monitored using flow cytometry to evaluate proliferation. Representative flow results are shown. (E) FABP2 levels in faeces were assessed by immunoblotting in the different groups. (F) Cytokine expression following α-FABP2 antibody treatment in the IL-10 KO mouse model was investigated using RT-qPCR. (G) IF staining of FABP2 (red) and peptidoglyan (green) in the small intestine showing FABP2 expression and bacterial load, respectively, following α-FABP2 antibody treatment in the IL-10 KO mouse model. Representative images after α-FABP2 antibody administration are shown. Scale bar, 100 µm. (H) Gut microbiota load and E. faecalis abundance in the small intestine were visualised using FISH staining with a general 16S rRNA probe (green) or an E. faecalis-specific probe (red), following α-FABP2 antibody treatment in IL-10 KO mouse model. Representative images after α-FABP2 antibody administration are shown. Scale bar, 100 µm. (I) Genus-levels taxonomic profiling of faecal samples via Meta-seq following α-FABP2 antibody treatment in IL-10 KO mouse model. Representative images after α-FABP2 antibody administration are shown. Data are presented as the mean±SD. Statistical significance was determined using one-way ANOVA with multiple comparisons test. *, p<0.05; **, p<0.01; ***, p<0.001; ****, p<0.0001. ANOVA, analysis of variance; i.g., oral gavage; FABP2, fatty acid-binding protein 2; FISH, fluorescence in situ hybridisation; IF, immunofluorescence; KO, knockout; RT-qPCR, real-time quantitative PCR.

Discussion

We demonstrated that FABP2 protein functions not only as a key player in fatty acid metabolism within the small intestinal lumen but also as a host-derived pheromone-like signalling molecule. Specifically, FABP2 promotes the proliferation of the gut commensal bacterium Enterococcus by activating its pheromone transport systems and upregulating the transcription of downstream quorum-sensing signalling pathway genes (online supplemental figure S22). This interaction may alter the behaviour of commensal microbes to disrupt the homoeostasis of the gut microbiota ecosystem.²⁷

FABP2 functions as a pheromone-like inducer, while EF3041 serves as its receptor. The FABP2-EF3041 axis represents a novel regulatory mechanism governing complex host-commensal interactions. The axis activates the OppABCDF transporter and its downstream quorum-sensing signalling pathway. Supporting this hypothesis, the presence of FABP2 significantly enhances the social proliferation of E. faecalis, as demonstrated by SEM imaging (figure 5D), and upregulates the transcription of genes associated with the Opp transporter and the Sec protein translocase complex, as revealed by RNA-seq. These genes include oppA, oppB, oppC, oppD, oppF, secE, secG and yajC, which were induced in the wild type strain but not in the EF3041-null mutant (figure 5H–K), and only under conditions where FABP2 was added to the medium. Real-time PCR analysis further confirmed that this enhanced social proliferation was driven by increased expression of quorum-sensing-related genes,²⁷ including codY, cylL, cylM and gelE (figure 5L and online supplemental figure S17), in response to FABP2 supplementation. Additional evidence for FABP2-mediated social proliferation of gut microbiota comes from the low percentage of FABP2-positive bacteria in pure culture compared with faecal samples from DSS-treated mice (figure 2I and online supplemental figure S10C) and IL-10 KO mice (figures4G, 7E 9Dundefined, online supplemental figure S18C and S21C). This finding suggests that FABP2 is used not only by E. faecalis but also by a variety of other gut microbial species (figure 3H–L).

EF3041 deletion and α-FABP2 neutralisation in DSS-treated and IL-10 KO mice demonstrated that manipulating the gut microbiota offers a promising therapeutic approach for treating IBD and potentially preventing its onset. However, a key question in IBD research is identifying the early events involved in pathogenesis, specifically whether dysbiosis precedes inflammation or whether inflammation triggers secondary dysbiosis.²⁸ Our study supports the latter hypothesis that dysbiosis arises as a consequence of inflammation in the intestinal epithelium and the subsequent proliferation of gut commensals (figures2H,I 3J and online supplemental figure S9). In the early stages of DSS-induced colitis, FABP2 expression was detected in the faeces of DSS-treated mice within just a few days (figure 2I). At the same time, the proportion of FABP2-positive gut microbiota increased (figure 2H), suggesting that epithelial damage proceeds microbial dysbiosis. Moreover, although no significant colitis symptoms were initially observed, exogenous administration of FABP2 promoted Enterococcus overgrowth in mice. Notably, this microbial imbalance was reversed following the withdrawal of FABP2 during the early days of DSS treatment (figure 3J and online supplemental figure S9). These findings further support that inflammation, triggered by epithelial damage, is the primary driver of dysbiosis in the development of IBD.

Materials and methods

Detailed methods are described in online supplemental materials and methods.

Animals

Fabp2 KO and BALB/cAnCya-Il10 em1/Cya (C001527) mice were purchased from Cyagen (Suzhou) Biosciences Inc. All mice were age-matched (8 weeks old at the start of experiments) and sex-matched (only male mice were used). Littermate controls were used where appropriate. Mice were maintained in specific pathogen-free animal facilities at Shanghai Jiao Tong University School of Medicine. The mice were housed under 12 hours light/dark cycles with free access to food and sterile water. All animal experiments were conducted in accordance with local guidelines for the use and care of laboratory animals as provided by Shanghai Jiao Tong University School of Medicine Institutional Animal Care and Use Committees.

Data availability statement

Data are available upon reasonable request. All data relevant to the study are included in the article or uploaded as supplementary information.