Lactobacillus johnsonii-derived extracellular vesicles carrying GAPDH protect against ulcerative colitis through modulating macrophage polarization

Graphical abstract

Highlights

• •L. johnsonii-derived extracellular vesicles (EVs) alleviate colitis and protect the intestinal barrier.
• •GAPDH is an important protein component of EVs derived from L. johnsonii.
• •GAPDH blocks MAPK and STAT3 signaling pathways to activate anti-inflammatory macrophage polarization.
• •STAT3-blocked macrophage-secreted EVs inhibit the TLR4 signaling pathway to protect the intestinal barrier in colitis mice.

Abstract

Introduction

Ulcerative colitis (UC) is a major inflammatory condition worldwide.

Objectives

The purpose of this study was to investigate the potential contribution of Lactobacillus johnsonii against UC from the perspective of gut microbiota-macrophage-host interactions.

Methods

L. johnsonii abundance in UC patients and colitis mice was evaluated by genomic sequencing. SPF and macrophage-depleted mice were employed to explore the effects of L. johnsonii and its products on colitis. An in vitro macrophage and intestinal epithelial cell co-culture system was constructed. Proteins in extracellular vesicles (EVs) were identified by proteomic analyses, and host signaling pathways were analyzed with transcriptomic analyses.

Results

L. johnsonii abundance was found to be associated with macrophage polarization and intestinal barrier function in human UC patients and mice of a colitis model. L. johnsonii and its derived EVs alleviate colitis in mice in a macrophage-dependent manner. Glyceraldehyde-3-phosphate dehydrogenase (GAPDH), a protein in EVs derived from L. johnsonii, counteracts colitis in vitro and in vivo by regulating macrophage phenotype. GAPDH enhances anti-inflammatory macrophage polarization by inhibiting the MAPK-STAT3 axis. Macrophage-secreted EVs enhances intestinal barrier function in colitis mice by blocking the TLR4 pathway. Protein components in macrophage-derived EVs contribute to colitis remission and intestinal barrier protection.

Conclusion

GAPDH originating in L. johnsonii-derived EVs alleviates colitis and improves intestinal barrier by inhibiting STAT3 in macrophages. EVs created from L. johnsonii are a potential novel treatment strategy for UC.

Introduction

Ulcerative colitis (UC) is defined by chronic non-specific inflammation of the rectum and colon, is a complex and intractable intestinal disease with a significant global impact [1]. The importance of gut microbiota in the incidence and progression of UC is continually being discovered [2]. It has been shown that the intestinal microbiota is one of the important biological indicators of the development of UC [3]. Pathogenic bacteria are considered to negatively impact the status of patients with UC in multiple ways [4]. Recent evidence revealed that the disease severity in patients with UC is closely related to the pathogenic bacteria [5]. Conversely, probiotic microbes alleviate UC symptoms by modulating the host inflammatory response [6]. Probiotic microbes can regulate immune homeostasis against UC patients through their own or secreted bioactive substances [7].

The presence of the intestinal epithelial barrier presents a challenge for the bidirectional communication between gut microbes and the host [8]. Therefore, most microbe-host interactions likely occur via characteristic derivatives secreted by bacteria. Extracellular vesicles (EVs) are derivatives of bacteria that consist of a membrane bilayer nanostructure and that are released during growth. EVs contain a range of parental bacterial components [9]. Recent work has highlighted the versatility of probiotic-derived EVs in regulating intestinal homeostasis [10]. However, the identities of the specific active substances from probiotic-derived EVs that modulate UC remain unclear.

Macrophages are key members of the host’s innate immune system, and functional adjustments caused by phenotypic changes are thought to constitute a link between the gut microbiota and the host [11,12]. UC is often accompanied by macrophage disorders; thus, targeted regulation of macrophage function is an effective strategy to combat UC [13]. In addition, the efficacy of probiotics in alleviating colitis by modulating macrophage function to control the inflammatory response has been demonstrated [14]. EVs with functions similar to those of their parental bacteria have also been shown to exhibit a superior anti-inflammatory capacity. Macrophages have been shown to take up bacterial EVs by phagocytosis and to transport the active substances in the cargo into the cytoplasm, thereby influencing macrophage immune plasticity [15]. However, the mechanisms by which EVs from probiotic microbes modulate the immune function of macrophages have not yet been clearly established.

Both innate and adaptive immune cells, including Tregs, Th9, Th17 cells, neutrophils, monocytes, and dendritic cells, play key roles in UC pathogenesis [16]. Our previous study has shown that Lactobacillus johnsonii derived EVs (LjEVs) can control intestinal inflammatory injury by modulating macrophage phenotype [17]. However, it is not clear exactly which specific protein component plays a central role. Therefore, we hypothesized that specific protein component present in LjEVs mitigates UC by exerting inflammatory plasticity towards macrophages. In this study, we identified a unique pathway by which LjEVs exert anti-inflammatory and gut barrier-enhancing activities in a UC model. We discovered that UC is associated with changes in intestinal L. johnsonii abundance in human patients and model mice, and that reduced L. johnsonii abundance is associated with macrophage phenotype and intestinal barrier function. Studies performed in vivo and in vitro revealed that glyceraldehyde-3-phosphate dehydrogenase (GAPDH) from LjEVs stimulates macrophages to secrete their own EVs through modulation of the MAPK-STAT3 signaling axis to exert an immunomodulatory effect, with potent inhibitory effects on UC-induced inflammatory responses and intestinal barrier disturbances. Thus, we extended the physiological roles of L. johnsonii and presented a novel prospective therapeutic method for treating UC in the clinic.

Materials and methods

Ethics statement

This study involves human participants. Ethical approval was obtained from the Ethics Committee of Jinling Hospital (Approval number 2021DZGZR-YBB-054). Participants gave informed consent to participate in the study before taking part. All procedures involving animals were approved by the Huazhong Agricultural University of Health Guide for the Care and Use of Laboratory Animals (Approval number HZAUMO-2023–0097).

Isolation and identification of bacteria-derived EVs

Ultracentrifugation was used to isolate EVs from bacterial cultures as previously described [18,19]. Briefly, bacterial cultures in the logarithmic phase of growth were centrifuged for 30 min at 8000g. The supernatant was centrifuged for a second time for 45 min at 20,000g. The second supernatant was filtered using a 0.22 μm Millipore filter. The filtrate was centrifuged at 120,000g for 2 h at 4 °C in a SW 32 Ti rotor (Beckman Coulter, Fullerton, CA, USA). After discarding the supernatant, the sediment was resuspended in phosphate-buffered saline (PBS), and the process was repeated. After centrifugation, the sediment was resuspended in PBS. The presence of LjEVs was confirmed using transmission electron microscopy (TEM), and the range of particle diameters was determined using nanoparticle tracking analysis (NTA).

Isolation and identification of macrophage-derived EVs

Ultracentrifugation was used to isolate EVs from the supernatant of cultured macrophages, essentially as previously described [20]. Briefly, RAW264.7 cells were cultivated to 80 % density and washed with PBS. Following 24 h of culture in serum-free medium, the cells were centrifuged at 2000 g for 20 min to remove cells and debris. The supernatant was passed through a 0.22 μm Millipore filter, and the filtrate was centrifuged at 120,000g for 1 h at 4 °C in a Beckman Coulter 70 Ti rotor (Beckman Coulter, Fullerton, CA, USA). The supernatant was discarded, and the sediment was resuspended in PBS and centrifuged again at 120,000g for 1 h at 4 °C. Following centrifugation, the material was resuspended in 200 μL PBS. The presence of EVs was detected using TEM, and the range of particle sizes was assessed using NTA.

Protein elimination from bacterial- and macrophage-derived EVs

EV proteins were eliminated by treatment of native EVs (1 mg/mL) with proteinase K-agarose (1 mg/mL) for 2 h at 37 °C, as described previously [21]. The proteinase K was inactivated by incubation at 75 °C for 1 h, and the proteinase K-agarose was removed by centrifugation at 12,000g for 1 min.

Cellular uptake of bacterial EVs

RAW264.7 cells were added to 6-well plates and grown overnight at 37 °C. The medium was removed and replaced with Hank's balanced salt solution, and the cells were let to adapt for 15 min. After the 15-min incubation, LjEVs (20 μg/mL) that were labelled with DiI (Sigma, #42364) were added to the wells, and the cells were incubated for 12 h at 37 °C. A fluorescent microscope was utilized to detect DiI-positive cells.

Human subjects

Fecal samples were collected from 19 UC patients (male, n = 10; female, n = 9; ages, 44.58 ± 4.34 years), and 22 healthy control subjects (male, n = 15; female n = 7; ages, 56.55 ± 2.46 years). Informed consent was given by all subjects prior to sample collection.

Murine model

Male C57BL/6 mice (aged 6–8 weeks) were obtained from the Laboratory Animal Central at Huazhong Agricultural University in Wuhan, China. The mice were housed in a pathogen-free mouse colony that had unlimited access to food and water. In all animal tests, mice were randomly assigned to different groups.

Mice were fed 2.5 % dextran sodium sulfate (DSS; 160110, MP Biomedicals, USA) for 7 d to induce a mouse model of UC. Every day during DSS treatment, mice were weighed and their Disease Activity Index (DAI) was assessed. Clodronate liposomes (200 μL/mouse, F70101C-A, FormuMax, USA) were given intraperitoneally every 3 d to develop a macrophage-depleted mouse model, as previously described [22].

For intervention experiments, mice in the control (CON) and DSS groups were administered 200 µL PBS daily via oral gavage. Other UC model mice were administered 200 µL of PBS containing either 2 × 10⁹ L. johnsonii cells (DSS + Lj group), 50 μg of native LjEVs (DSS + LjEVs and DSS + LjEVs-NA group), or 50 μg of protease-treated LjEVs (DSS + LjEVs-PK) daily via oral gavage.

Mice in the DSS + GAPDH group received intraperitoneal injections of 10 mg/kg GAPDH (G2267, Sigma-Aldrich, USA) every 2 d. Mice in the DSS + Stattic group received intraperitoneal injections of 10 mg/kg Stattic (S7024, Selleck, USA) every 2 d. Mice in the DSS + WT-mEVs and DSS + Stattic-mEVs groups were injected intraperitoneally with 20 μg WT-mEVs or Stattic-mEVs every 2 d. Mice in the DSS + Stattic-mEVs-NA and DSS + Stattic-mEVs-PK groups were injected intraperitoneally with 20 μg native or protease-treated Stattic-mEVs every 2 d.

Cell culture and treatments

Dulbecco’s modified eagle medium (DMEM) containing 10 % fetal bovine serum (Gibco) and 1 % penicillin was used to culture RAW264.7 and CT-26 cells. Every two days, the medium was changed. The temperature of the incubation chamber was 37 °C, and the CO₂ concentration was 5 %.

For experiments investigating the effects of intervention with L. johnsonii or LjEVs, RAW264.7 cells were cultured with 10⁸ CFU/mL L. johnsonii or 20 μg/mL LjEVs for 12 h. Some cells were treated with 100 ng/mL LPS, followed by an additional incubation for 24 h. For treatment with proteinase-treated LjEVs, RAW264.7 cells were treated with 20 μg/mL native or protease-treated LjEVs for 12 h before being treated with 100 ng/mL LPS for 24 h. For treatment with GAPDH, RAW264.7 cells were treated with 1000 μg/mL GAPDH for 1 h before being treated with 100 ng/mL LPS for 24 h. For analysis of MAPK activation, RAW264.7 cells were treated with 500 nM diprovocim (HY-123942, MedChemExpress, USA) and 1000 μg/mL GAPDH for 1 h before being treated with 100 ng/mL LPS for 24 h.

For activation of STAT3, RAW264.7 cells were treated with 5 μM colivelin (S9664, Selleck, USA) and 1000 μg/mL GAPDH for 1 h before being treated with 100 ng/mL LPS for 24 h. For STAT3 inactivation, RAW264.7 cells were incubated with 10 μM Stattic and 100 ng/mL LPS for 24 h.

For intervention with mEVs, CT-26 cells were incubated with 10 μg/mL WT-mEVs, Stattic-mEVs, or native or proteinase-treated Stattic-mEVs for 12 h before being incubated with 100 μg /mL LPS for 12 h.

Treated RAW264.7 cells were also subjected to co-culture with CT-26 cells. In this case, the pre-treated RAW264.7 cells were inoculated into the basolateral compartments of a 6-well plate, and CT-26 cells stimulated with 10 μg/mL LPS for 12 h were seeded into the apical compartments of 6-well hanging inserts. The CT-26 cells were harvested after 12 h of co-culture.

For cell experiments involving LPS treatment, RAW264.7 cells were treated with LPS for 24 h and CT-26 cells were treated with LPS for 12 h. For cell experiments involving GAPDH treatment, RAW264.7 cells were co-incubated with GAPDH for 1 h before being treated with LPS. For cell experiments involving mEVs treatment, CT-26 cells were co-incubated with mEVs for 12 h before being treated by LPS. At the end of the treatment, Trizol reagent was added to the cells for RNA quantification and RIPA buffer for Western blotting analyses.

Determination of the abundance of L. Johnsonii in humans and mice

PCR amplification was performed in duplicate for each sample. The sequences of primers used for analysis of L. johnsonii were forward sequence, 5′-CCG CGG CTT AGA TTC TGG TA, and reverse sequence, 5′-AGC TAA ACG AGC ATC TGG GA. L. johnsonii copy counts were determined using standard curves produced from plasmids carrying the 16S rRNA gene. The gene copy counts were determined using equation.

$$\mathit{Gene}copynumber=\frac{\left[DNAconcentration\left(\frac{\mathit{mg}}{\mathit{mL}}\right)\right]\times [6.022\times {10}^{23}\frac{\mathit{copies}}{\mathit{mol}}]}{\left[DNAsize\left(bp\right)\right]\times [6.60\times {10}^8]}$$

Histological and immunohistochemical analyses

Hematoxylin and eosin (H&E) staining was carried out as previously reported [23]. In brief, colonic tissues were preserved in 4 % formaldehyde and fixed in paraffin. Slices were stained with H&E using standard procedures. A light microscope was used to create digital photographs of colonic morphology. The colonic epithelial histopathology was graded using criteria described previously [21].

The immunohistochemical staining was conducted as previously described [24]. The colonic slices were incubated at 4 °C overnight with a CD11c primary antibody or a CD206 antibody. After a thorough washing, the slices were treated with a goat anti-rabbit secondary antibody coupled to horseradish peroxidase (HRP) (Santa Cruz Biotechnology, sc-2004). A light microscope was used to obtain the digital photos. Using Image-Pro Plus 6.0 (Media Cybernetics, Inc., Rockville, MD, USA), the same brown color was used as the universal criterion for grading all photographs.

Quantification of serum biochemicals

ELISA kits (Shanghai Enzyme-linked Biotechnology Co. Ltd., Shanghai, China) were used to test levels of D-lactate (D-LA), diamine oxidase (DAO), tumor necrosis factor-⍺ (TNF-α), and interleukin (IL)-1β/6/10 in serum and colon tissue.

Quantification of RNA

Total RNA was extracted with Trizol Reagent (Invitrogen) and quantified using a NanoDrop ND-1000 Spectrophotometer (ThermoFisher). RNA was processed with RNase-Free DNase (Promega, USA), and a 2 μg sample of total RNA was utilized as the template for reverse transcription. Real-time PCR was performed with 2 μL of diluted cDNA and gene-specific primers. This study employed β-actin as reference gene. Primer information is provided in Supplementary Table 4.

Western blotting analyses

Proteins were isolated with RIPA buffer (Roche, Penzberg, Germany), and protein concentrations were measured with BCA protein assay kits (Pierce, Rockford, IL, USA). Proteins were separated using SDS-PAGE and transferred to a nitrocellulose membrane (Bio Trace, Pall Co, USA). The membranes were blocked for 2 h in blocking buffer before being treated overnight with the appropriate primary antibody. After extensive washing in Tris-buffered saline with Tween-20, the membranes were incubated with the appropriate secondary antibody for 2 s. Supplementary Table 5 includes antibody information. Protein expression was observed using an imaging system (Bio-Rad, USA) and quantified using Quantity One.

Proteomic analysis

Proteomic analyses of EVs generated from bacteria and macrophages was performed as described previously [25]. Briefly, the EVs were lysed on ice in lysis buffer (8 mol/L urea, 2 mmol/L EDTA, 1 % protease inhibitor cocktail), and soluble proteins were recovered by centrifugation at 12,000g for 15 min. The protein solutions were digested overnight at 37 °C with trypsin (Promega, #V5111) at a ratio of 1:50 (w/w). The tryptic peptides were separated using high pH reverse-phase HPLC and then analyzed using tandem mass spectrometry (MS/MS) with the Q Exactive^TM Plus system (Thermo) linked to a UPLC (LC-MS/MS). Data from sequencing was evaluated using the Majorbio Cloud platform (www.majorbio.com).

RNA sequencing

Total RNA was isolated with Trizol Reagent (Invitrogen). RNA sequencing was carried out as previously described [26]. Data from sequencing was evaluated using the Majorbio Cloud platform (www.majorbio.com).

LPS measurement

The LPS content in the EVs was quantified using a chromogenic endpoint assay (Chinese Horseshoe Crab Reagent Manufactory Co., Ltd., Xiamen, China), which has a sensitivity threshold of 0.05 EU/mL. The assay was conducted following the manufacturer’s protocol. Briefly, the reaction mixture was prepared by combining the limulus amebocyte lysate reagent with the sample or standard, and the mixture was incubated at 37 °C to allow the enzymatic reaction to proceed. After incubation, the absorbance of the reaction mixture was measured at 405 nm using a microplate reader. The concentration of LPS in each sample was determined by comparing the absorbance values to the calibration curve generated from the standards.

Statistical analysis

The data are reported as means ± SEM. Significant differences between the two groups were evaluated by Student’s t-test. For comparisons among more than two groups, a one-way analysis of variance (ANOVA) and a post hoc Tukey test were performed (SPSS version 20.0 for Windows; SPSS Inc., Chicago, IL, USA). Differences for which P < 0.05 were considered to be statistically significant. The numbers of replicates used to perform statistical analyses are noted in the relevant figure legends.

Results

L. Johnsonii abundance correlates with macrophage phenotype and intestinal barrier markers in humans with ulcerative colitis and in a mouse colitis model

We collected fecal and colonic tissue samples from normal human subjects and UC patients to investigate the potential link between L. johnsonii and macrophage polarization and intestinal barrier function (Fig. 1A). The abundance of L. johnsonii was found to be substantially lower in UC patients than in normal subject (Fig. 1B). The levels of the mRNA encoding iNOS and CD11c were higher in the colon tissue of UC patients relative to that of normal subjects, while the levels of mRNA encoding Arg1, MRC1, ZO-1, occludin, and claudin-1 were lower (Fig. 1C-I). Moreover, L. johnsonii abundance was obviously negatively correlated with the levels of iNOS and CD11c mRNA and significantly positively correlated with Arg1, ZO-1, occludin, and claudin-1 mRNA (Fig. 1J).

Fig. 1.

Correlations between L. johnsonii abundance and macrophage phenotype and intestinal barrier in UC patients (A) Experiment design for collection of fecal and colonic samples from normal and UC patients. (B) The abundance of L. johnsonii in human stool was determined by RT-qPCR. (C-I) The mRNA expression of iNOS, CD11c, Arg1, CD206, ZO-1, occludin, and claudin-1 in colon tissues from human subjects. (J) Correlation between L. johnsonii abundance and macrophage phenotype and gene expression of tight junction protein markers in human. n = 22/19 (Normal/UC) for A-J. Data are expressed as the means ± SEM (B-I) and Student's t-test was performed (B-I). ***P < 0.001.

We performed similar analyses on mice with DSS-induced colitis (Fig. S1A). DSS-induced mice lost more body weight and had a higher disease activity index (DAI) than did CON mice, as well as shorter colon lengths (Fig. S12B-E). Consistent with phenomena observed in UC patients, colitis mice possessed a lower abundance of L. johnsonii and were found to have significantly higher levels of iNOS and CD11c mRNA and significantly lower levels of Arg1, CD206, ZO-1, occludin, and claudin-1 mRNA (Fig. S1F-M). Additionally, in the mouse model, L. johnsonii abundance was significantly negatively correlated with the expression of mRNA encoding iNOS and CD11c and significantly positively correlated with the expression of mRNA encoding Arg1, CD206, ZO-1, occludin, and claudin-1 (Fig. S1N).

L. Johnsonii relieves colitis in a macrophage-dependent manner

To investigate the potential efficacy of L. johnsonii in the treatment of colitis, DSS-induced mice were treated via oral gavage with vehicle (DSS group) or with L. johnsonii (DSS + L. j group) (Fig. 2A). In comparison to vehicle-treated mice, treatment with L. johnsonii was found to reduce loss of body weight and DAI (Fig. 2B, C). DSS-induced mice treated with L. johnsonii also exhibited longer colon lengths (Fig. 2D and Fig. S2A) and significantly lower colon histopathological scores (Fig. 2E and Fig. S2B) than did vehicle-treated mice. Immunohistochemical analyses (Fig. S2B) showed a substantial reduction in the proportion of CD11c⁺ cells (Fig. 2F) and a marked increase in the proportion of CD206⁺ cells (Fig. 2G) in the DSS + L. j group compared to the DSS group. In addition, treatment with L. johnsonii correlated with decreased levels of D-LA, DAO, TNF-α, IL-1β, and IL-6 in the serum and colonic tissues of DSS-induced mice and with increased levels of IL-10 (Fig. 2H). The level of iNOS protein was reduced in mice treated with L. johnsonii, whereas the Arg1, ZO-1, occludin, and claudin-1 protein levels were elevated (Fig. 2I).

Fig. 2.

L. johnsonii alleviates DSS-induced colitis by modulating macrophage phenotype (A) Experiment design of the treatment of mice with DSS and L. johnsonii. (B) Body weight changes. (C) Disease activity indexes. (D) Colonic lengths. (E) Colonic histopathological scores. (F, G) The proportions of CD11c⁺ cells (F) and CD206⁺ cells (G) in colon tissues were determined using immunohistochemistry. (H) The serum and colon tissue concentrations of IL-1β, IL-6, TNF-α, IL-10, DAO, and D-LA. (I) Protein expression of iNOS, Arg1, ZO-1, occludin, and claudin-1 in colon tissues (J) Experiment design of the treatment of RAW264.7 cells with LPS and L. johnsonii. (K) Gene expression of iNOS, CD11c, Arg1, and CD206 in RAW264.7 cells. (L) Experiment design of the RAW264.7 and CT-26 cell co-culture system. (M) Gene expression of ZO-1, occludin, and claudin-1 in CT-26 cells. n = 8 for A-H; n = 3 for I; n = 4 for J-M. Data are expressed as the means ± SEM (B-I, K and M) and one-way ANOVA was performed, followed by Tukey's test (B-I, K and M). *P < 0.05, **P < 0.01, ***P < 0.001.

We investigated the potential modulation of macrophage polarization by L. johnsonii by treating mouse macrophages (RAW264.7 cells) in vitro with LPS (LPS group) or with both LPS and L. johnsonii cells (LPS + L. j group). In the LPS + L. j group, the expression of mRNA encoding iNOS and CD11c was dramatically reduced, whereas the abundances of mRNA encoding Arg1 and CD206 were significantly enhanced (Fig. 2K).

Further examination of the effects of L. johnsonii-treated macrophages on intestinal epithelial cell barrier function was performed using a co-culture system (Fig. 2L). Here, we found that L. johnsonii-treated murine macrophages (RAW264.7 cells) dramatically enhanced the genes expression of ZO-1, occludin, and claudin-1 in co-cultured CT-26 murine carcinoma cells (Fig. 2M).

To explore the role of macrophages in the alleviation of UC by L. johnsonii, mice in which macrophages had been depleted by treatment with clodronate liposomes were treated with DSS alone (DSS-E group) or with both DSS and L. johnsonii (DSS + L. j-E group) (Fig. S3A). Here, L. johnsonii was not able to alleviate DSS-induced changes to body weight and DAI in macrophage-depleted mice (Fig. S3B and C). Colonic lengths (Fig. S3D and E) and histopathological scores (Fig. S3F and G) were not obviously different among DSS, DSS-E, and DSS + L. j-E groups. Moreover, the levels of D-LA, DAO, TNF-α, IL-1β, IL-6, and IL-10 in serum and colonic tissues were essentially unchanged by intervention with L. johnsonii in the macrophage-depleted DSS-induced model mice (Fig. S3H).

EVs derived from L. Johnsonii alleviate colitis in a macrophage-dependent manner

A potential mechanism of the interaction between L. johnsonii and colitis involves the delivery of bioactive materials via EVs. Indeed, the presence of EVs in L. johnsonii cultures was confirmed using TEM (Fig. 3A, B). NTA was used to quantify LjEVs (Fig. 3C). To determine the role of LjEVs in alleviating colitis, DSS-induced mice were administrated with vehicle (DSS group) or with LjEVs (DSS + LjEVs group) by oral gavage (Fig. 3D). In comparison to the DSS group, the DSS + LjEVs group experienced a reduced extent of loss of body weight and lower DAI (Fig. 3E, F). DSS-induced mice treated with LjEVs possessed a longer average colon length (Fig. 3G and Fig. S4A) and significantly lower colon histopathological score (Fig. 3H and Fig. S4B) than did similar mice treated with vehicle.

Fig. 3.

Extracellular vesicles derived from L. johnsonii alleviate DSS-induced colitis by modulating macrophage phenotype (A, B) TEM of isolated EVs derived from L. johnsonii. (C) Size distribution of EVs derived from L. johnsonii as determined by NTA. (D) Experiment design of the treatment of mice with DSS and LjEVs. (E) Body weight changes. (F) Disease activity indexes. (G) Colonic lengths. (H) Colonic histopathological scores. (I, J) The proportions of CD11c⁺ cells (I) and CD206⁺ cells (J) in colon tissues were determined using immunohistochemistry. (K) The serum and colon tissue concentrations of IL-1β, IL-6, TNF-α, IL-10, DAO, and D-LA. (L) Colonic protein expression and analysis of iNOS, Arg1, ZO-1, occludin, and claudin-1. (M) RAW264.7 cells were incubated with DiI-stained LjEVs (red), and nuclei were counterstained with DAPI prior to fluorescence microscopy imaging. (N) Experiment design of the treatment of RAW264.7 cell with LPS and LjEVs. (O) mRNA expression of iNOS, CD11c, Arg1, and CD206 in RAW264.7 cells. (P) Experiment design of the RAW264.7 and CT-26 cell co-culture system. (Q) mRNA expression of ZO-1, occludin, and claudin-1 in co-cultured CT-26 cells. n = 8 for D-K; n = 3 for L; n = 4 for N-Q. Data are expressed as the means ± SEM (E-L, O and Q) and one-way ANOVA was performed, followed by Tukey's test (E-L, O and Q). *P < 0.05, **P < 0.01, ***P < 0.001. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

According to immunohistochemical analyses, (Fig. S4B) the DSS + LjEVs group had a significantly lower proportion of CD11c⁺ cells (Fig. 3I) and a significantly higher proportion of CD206⁺ cells (Fig. 3J) than did the DSS group. In addition, treatment with LjEVs led to decreased concentrations of D-LA, DAO, TNF-α, IL-1β, and IL-6 in serum and colonic tissues of DSS-induced mice and an increased serum concentration of IL-10 (Fig. 3K). The level of iNOS protein was reduced in the DSS + LjEVs group, whereas Arg1, ZO-1, occludin, and claudin-1 protein levels were elevated (Fig. 3L).

Fluorescently-labeled LjEVs (red signal) were identified in the cytoplasm of RAW264.7 cells after 12 h of co-incubation, indicating that LjEVs were internalized by the macrophages (Fig. 3M). We also use treatment of macrophages with LPS in concert with LjEVs investigate a potential regulatory role for L. johnsonii on macrophage polarization (Fig. 3N). We found that in cells treated with both LPS and LjEVs, the expression of mRNA encoding iNOS and CD11c was dramatically reduced relative to cells treated with LPS and vehicle, whereas gene abundance of Arg1 and CD206 was significantly enhanced (Fig. 3O). Further examination of the effects of LjEVs-treated macrophage on the barrier function of intestinal epithelial cell based on a cell co-culture system (Fig. 3P) demonstrated that LjEV-treated macrophages dramatically enhanced the levels of ZO-1, occludin, and claudin-1gene expression (Fig. 3Q).

To learn more about the involvement of macrophages in the treatment of UC by LjEVs, mice in which macrophages were depleted with clodronate-liposomes were fed DSS (DSS-E group) and treated with LjEVs (DSS + LjEVs-E group) (Fig. S5A). Unlike in mice with normal macrophages, LjEVs failed to alleviate loss of body weight and decreased DAI in macrophage-depleted mice (Fig. S5B, C). The average colonic length (Fig. S5D, E) and histopathological score (Fig. S5F, G) were not obviously different among DSS, DSS-E and DSS + LjEVs-E groups. In addition, the levels of D-LA, DAO, TNF-α, IL-1β, IL-6, and IL-10 in serum and colonic tissues were similar between DSS-induced mice and LjEV-treated DSS-induced mice in which macrophages had been depleted (Fig. S5H).

To determine if proteins mediate the effect of LjEVs in alleviating colitis, mice were fed native LjEVs (DSS + LjEVs-NA group) or proteinase-treated LjEVs (DSS + LjEVs-PK group) (Fig. S6A). Proteinase-treated LjEVs failed to alleviate the body weight loss and decreased DAI associated with colitis (Fig. S6B and C). The colonic lengths (Fig. S6D) and histopathological scores (Fig. S6E) of mice were not obviously different between DSS and DSS + LjEVs-PK groups. Moreover, there were also no fluctuations in the levels of D-LA, DAO, TNF-α, IL-1β, IL-6, or IL-10 in serum and colonic tissues of DSS-induced mice administered vehicle or those administered protease-treated LjEVs (Fig. S6F).

We also compared the effects of native LjEVs (LPS + LjEVs-NA group) and protease-treated LjEVs (LPS + LjEVs-PK group) on the polarization of LPS-induced macrophages (Fig. S6G). Treatment of LPS-induced macrophages with proteinase K-treated LjEVs was not found to be associated with significant differences in the levels of expression of iNOS, CD11c, Arg1, and CD206 mRNA (Fig. S6H). Further examination of the effects of protease-treated LjEVs-treated macrophage on the barrier function of intestinal epithelial cells (Fig. S6I) revealed similar levels of mRNA encoding ZO-1, occludin, and claudin-1 in CT-26 cells co-cultured with LPS-stimulated macrophages that were pre-treated with vehicle or with proteinase-treated LjEVs (Fig. S6J).

EV-derived GAPDH alleviates colitis in a macrophage-dependent manner

Biochemical analyses demonstrated that LjEVs contain proteins, DNA, and RNA, with the protein content being considerably higher than the nucleic acid content (Fig. 4A). Proteomic analysis was then carried out to determine the protein profile of LjEVs. In triplicate biological LjEV samples, 883 overlapping proteins were found and chosen for further investigation (Fig. 4B). The subcellular localization of these 883 proteins revealed that 94.9 % were derived from the cytoplasm, whereas 5.1 % were extracellular proteins (Fig. 4C). GO annotations and KEGG pathway information for these identified proteins are presented in the Fig. S7A and B, respectively. Details of the 883 overlapping proteins contained in LjEVs are listed in Table S1.

Fig. 4.

GAPDH alleviates DSS-induced colitis by modulating macrophage phenotype (A) Quantification of total DNA, RNA, and protein in the EVs derived from L. johnsonii (LjEVs). (B) Venn diagram of results of proteomic analyses of three samples of LjEVs. (C) Subcellular localization of the proteins identified in LjEVs. EXC: extracellular, CYT: cytoplasmic. (D) Experiment design of the treatment of mice with DSS and GAPDH. (E) Body weight changes. (F) Disease activity indexes. (G) Colonic lengths. (H) Colonic histopathological scores. (I, J) The proportions of CD11c⁺ cells (I) and CD206⁺ cells (J) in colon tissues were determined using immunohistochemistry. (K) The serum and colon tissue concentrations of IL-1β, IL-6, TNF-α, IL-10, DAO, and D-LA. (L) Colonic protein expression and analysis of iNOS, Arg1, ZO-1, occludin, and claudin-1. (M) Experiment design of the treatment of RAW264.7 cells with LPS and GAPDH. (N) Levels of expression of iNOS, CD11c, Arg1, and CD206 mRNAs in RAW264.7 cells. (O) Experiment design of the RAW264.7 and CT-26 cell co-culture system. (P) Levels of ZO-1, occludin, and claudin-1 mRNA in co-cultured CT-26 cells. n = 3 for A-C; n = 8 for D-K; n = 3 for L; n = 4 for M−P. Data are expressed as the means ± SEM (E-L, N and P) and one-way ANOVA was performed, followed by Tukey's test (E-L, N and P). *P < 0.05, **P < 0.01, ***P < 0.001.

One of the proteins determined to be present in LjEVs was GAPDH. To investigate a role for GAPDH in alleviating colitis, DSS-induced mice were administered i.p. either vehicle (DSS group) or GAPDH (DSS + GAPDH group) (Fig. 4D). Treatment with GAPDH was found to relieve the loss of body weight and decreased DAI experienced by mice treated with DSS alone (Fig. 4E and F). Mice treated with GAPDH possessed longer colon lengths (Fig. 4G and Fig. S8A) and significantly lower colon histopathological scores (Fig. 4H and Fig. S8B) than did the DSS group.

According to immunohistochemical analyses (Fig. S8B), GAPDH therapy significantly reduced the proportion of CD11c⁺ cells (Fig. 4I) and a marked increase in the proportion of CD206⁺ cells (Fig. 4J). In addition, GAPDH treatment led to decreased levels of D-LA, DAO, TNF-α, IL-1β, and IL-6, but increased IL-10 in the serum and colonic tissues of DSS-induced mice (Fig. 4K). The expression of iNOS protein was reduced, whereas levels of Arg1, ZO-1, occludin, and claudin-1 proteins were elevated, in the DSS + GAPDH group (Fig. 4L).

We also investigated the effect of GAPDH in LPS-induced mouse macrophages in vitro (Fig. 4M). We discovered that the expression of iNOS and CD11c mRNA was considerably lower, whereas the levels of transcription of Arg1 and CD206 mRNA were significantly increased, in cells pretreated with GAPDH prior to LPS stimulation as compared to cells treated with vehicle prior to LPS stimulation (Fig. 4N). Further examination of the effects of GAPDH-treated macrophages on the barrier function of intestinal epithelial cell based on a cell co-culture system (Fig. 4O) demonstrated that GAPDH-treated macrophages dramatically enhanced the expression of mRNA encoding ZO-1, occludin, and claudin-1 in co-cultured CT-26 cells (Fig. 4P).

To determine if macrophages are involved in the alleviation of UC by GAPDH, macrophage-depleted, DSS-induced mice were injected with vehicle (DSS-E group) or with GAPDH (DSS + GAPDH-E group) (Fig. S9A). GAPDH failed to alleviate the DSS-induced loss of body weight loss and decreases of DAI score in macrophage-depleted mice (Fig. S9B and C). The colonic lengths (Fig. S9D and E) and histopathological scores (Fig. S9F and G) were not obviously different between DSS, DSS-E and DSS + GAPDH-E groups. Moreover, we did not observe significant GAPDH-induced changes to the levels of D-LA, DAO, TNF-α, IL-1β, IL-6, or IL-10 in serum and colonic tissues in macrophage-depleted, DSS-induced mice (Fig. S9H).

GAPDH regulates macrophage phenotype by inhibiting the MAPK-STAT3 axis

To explore the molecular mechanisms by which GAPDH regulates macrophage polarization, we first performed transcriptomic sequencing and analysis (Fig. 5A). A KEGG function analysis revealed that the enrichment of the MAPK signaling pathway differed significantly between the two groups in all comparisons (Fig. 5B-D). In addition, the enrichment of JAK-STAT signaling pathway was significantly different between the CON and GAPDH groups (Fig. 5C).

Fig. 5.

GAPDH inhibits the MAPK-STAT3 signaling pathway to regulate macrophage phenotype and enhance the intestinal barrier (A) Experiment design of the treatment of RAW264.7 cell with LPS and GAPDH. (B-D) KEGG signaling pathways enriched in RAW264.7 cells upon treatment with LPS, GAPDH, or both LPS and GAPDH. (E) Levels of phosphorylation p38, ERK, JNK, and STAT3 in RAW264.7 cells were analyzed by Western blotting. (F) The ratios of intensities of phosphorylated to total protein in (E) were quantified. (G) Experiment design of the treatment of RAW264.7 cell with GAPDH and diprovocim. (H) Levels of iNOS, CD11c, Arg1, and CD206 mRNA in RAW264.7 cells. (I) Protein analysis of STAT3 phosphorylation in RAW264.7 cells. (J) Experiment design of the RAW264.7 and CT-26 cell co-culture system. (K) Levels of ZO-1, occludin, and claudin-1 mRNA in CT-26 cells. (L) Experiment design of the treatment of RAW264.7 cell with GAPDH and colivelin. (M) Levels of iNOS, CD11c, Arg1, and CD206 mRNA in RAW264.7 cells. (N) Experiment design of the RAW264.7 and CT-26 cell co-culture system. (O) Levels of ZO-1, occludin, and claudin-1 mRNA in CT-26 cells. n = 4 for A-D, G-H, J-O; n = 3 for E-F and I. Data are expressed as the means ± SEM (F, H-I, K, M and O) and one-way ANOVA was performed, followed by Tukey's test (F, H-I, K, M and O). *P < 0.05, **P < 0.01, ***P < 0.001.

These results led us to perform biochemical analyses to investigate the activation of proteins in the MAPK and JAK-STAT signaling pathways. When RAW264.7 cells were stimulated with LPS, the levels of phosphorylation of P38, ERK, JNK, and STAT3 were found to be significantly increased relative to those in control cells (Fig. 5E, F), as expected. However, when the cells were pre-treated with GAPDH prior to stimulation with LPS, the levels of phosphorylation of these proteins were decreased relative to LPS treatment alone, and these levels were not significantly different from those in unstimulated cells (Fig. 5E, F).

We investigated this pathway further in RAW264.7 cells by employing diprovocim, which is an agonist of TLR1/TLR2 and a well-characterized stimulator of MAPK signaling. Specifically, we stimulated MAPK signaling with diprovocim simultaneously during the pre-treatment with GAPDH prior to LPS stimulation (Fig. 5G). We found that the addition of diprovocim rendered GAPDH no longer able to alter the transcription of the genes encoding iNOS, CD11c, Arg1, and CD206 in LPS-challenged macrophages (Fig. 5H). In addition, the phosphorylation of STAT3 was higher in LPS-stimulated cells pre-treated with both GAPDH and diprovocim relative to those pre-treated with only GAPDH (Fig. 5I). The levels of mRNA encoding ZO-1, occludin, and claudin-1 in intestinal epithelial cells co-cultured with macrophages pre-treated with both GAPDH and diprovocim was also reduced in comparison to pre-treatment with GAPDH alone (Fig. 5J, K).

We performed additional analyses employing colivelin, a STAT3 activator (Fig. 5L). Similar to our findings with diprovocim, we found that the addition of colivelin rendered GAPDH unable to alter the expression of iNOS, CD11c, Arg1, and CD206 mRNA in LPS-challenged macrophages (Fig. 5M). Moreover, the GAPDH-induced increases in levels of ZO-1, occludin, and claudin-1 mRNA in co-cultured intestinal epithelial cells were reduced upon inclusion of colivelin in the pre-treatment of LPS-stimulated macrophages (Fig. 5N, O).

Inhibition of STAT3 in macrophages induces colitis relief

The in vitro results suggested that inhibition of STAT3 activation in macrophages by GAPDH might influence macrophage polarization and thus might impact colitis symptoms in vivo. To examine the potential role of inhibition of macrophage STAT3 activity in alleviating colitis, DSS-induced mice were administered vehicle (DSS group) or the STAT3 inhibitor Stattic (DSS + Stattic group) via ip injection (Fig. 6A). Treatment with Stattic was found to relieve the DSS-induced loss of body weight and decreased DAI score (Fig. 6B, C). Model mice undergoing the Stattic intervention possessed longer colon lengths (Fig. 6D and Fig. S10A) and significantly lower colon histopathological scores (Fig. 6E and Fig. S10B) than did vehicle-treated model mice.

Fig. 6.

A STAT3 inhibitor alleviates DSS-induced colitis by modulating macrophage phenotype (A) Experiment design of the treatment of mice with DSS and Stattic. (B) Body weight changes. (C) Disease activity indexes. (D) Colonic lengths. (E) Colonic histopathological scores. (F, G) The proportions of CD11c⁺ cells (F) and CD206⁺ cells (G) in colon tissues were determined using immunohistochemistry. (H) The serum and colon tissue concentrations of IL-1β, IL-6, TNF-α, IL-10, DAO, and D-LA. (I) Protein expression and analysis of iNOS, Arg1, ZO-1, occludin, and claudin-1 in colon tissues. (J) Experiment design of the treatment of RAW264.7 cells with LPS and Stattic. (K) Levels of iNOS, CD11c, Arg1, and CD206 mRNA in RAW264.7 cells. (L) Experiment design of the RAW264.7 and CT-26 cell co-culture system. (M) Levels of ZO-1, occludin, and claudin-1 mRNA in co-cultured CT-26 cells. n = 8 for A-H; n = 3 for I; n = 4 for J-M. Data are expressed as the means ± SEM (B-I, K and M) and one-way ANOVA was performed, followed by Tukey's test (B-I, K and M). *P < 0.05, **P < 0.01, ***P < 0.001.

Immunohistochemical analyses (Fig. S10B) showed a significant decrease in the proportion of CD11c⁺ cells (Fig. 6F) and a marked increase in the proportion of CD206⁺ cells (Fig. 6G) in model mice treated with Stattic as compared to control model mice. In addition, Stattic treatment led to decreased concentrations of D-LA, DAO, TNF-α, IL-1β, and IL-6 and an increased concentration of IL-10 in serum and colonic tissues of the DSS model mice (Fig. 6H). The expression of iNOS protein was reduced, whereas Arg1, ZO-1, occludin, and claudin-1 protein levels were elevated in mice treated with Stattic (Fig. 6I).

We also treated LPS-stimulated RAW264.7 macrophages with Stattic (LPS + Stattic group) to reveal the regulatory role of STAT3 on macrophage polarization (Fig. 6J). The LPS-induced increased expression of iNOS and CD11c mRNA was significantly reduced by treatment with Stattic, whereas the abundance of mRNA encoding Arg1 and CD206 was significantly increased (Fig. 6K). Further examination of the effects of Stattic-treated macrophages on the barrier function of intestinal epithelial cell based on a cell co-culture system (Fig. 6L) showed that Stattic-treated macrophages dramatically enhanced the expression of genes encoding ZO-1, occludin, and claudin-1 in co-cultured CT-26 cells (Fig. 6M).

The role of macrophages in the effect of STAT3 on the alleviation of UC was explored in macrophage-depleted DSS-induced mice. DSS-induced macrophage-depleted mice were left untreated (DSS-E group) or were treated with Stattic (DSS + Stattic-E group) via ip injection (Fig. S11A). Unlike the effects in macrophage-replete mice, Stattic failed to alleviate the DSS-induced loss of body weight and decreased DAI score in macrophage-depleted mice (Fig. S11B, C). Colonic lengths (Fig. S11D, E) and histopathological scores (Fig. S11F, G) were not obviously different among control macrophage-replete, control macrophage-depleted, and Stattic-treated macrophage-depleted DSS-induced mice. Moreover, there we observed no changes in the levels of D-LA, DAO, TNF-α, IL-1β, IL-6, or IL-10 levels in serum and colonic tissues of Stattic-treated macrophage-depleted DSS-induced mice as compared to control DSS-induced mice (Fig. S11H).

EVs derived from STAT3-inhibited macrophages inhibit the TLR4 signaling pathway to enhance the intestinal barrier in the context of colitis

The influence on co-cultured colonic cells could be mediated by EVs secreted by the macrophages. The presence of mEVs in the supernatants of RAW264.7 cell cultures was confirmed using TEM (Fig. 7A). NTA was used to quantify mEVs (Fig. 7B). To determine the role of mEVs in alleviating colitis, DSS-induced mice were fed mEVs derived from wild-type macrophages (DSS + WT-mEVs group) or mEVs derived from Stattic-treated macrophages (DSS + Stattic-mEVs group) (Fig. 7C). DSS-induced mice treated with mEVs isolated from Stattic-treated cells Stattic-mEVs exhibited decreased loss of body weight and lower changes to DAI score as compared with DSS-induced mice treated with mEVs isolated from control macrophages (Fig. 7D and E). DSS-induced mice fed mEVs from Stattic-treated macrophages possessed longer colon lengths (Fig. 7F) and significantly lower colon histopathological scores (Fig. 7G) than did control-treated DSS-induced mice. In addition, treatment with Stattic-mEVs was associated with decreased concentrations of D-LA, DAO, TNF-α, IL-1β, and IL-6, in serum and colonic tissues of colitis mice as well as IL-10 levels (Fig. 7H). The levels of TLR4 and MyD88 proteins and the phosphorylation of NF-κB were reduced, whereas the levels of ZO-1, occludin, and claudin-1 proteins were elevated in DSS + Stattic-mEVs group (Fig. 7I).

Fig. 7.

EVs of macrophage lacking STAT3 enhance the intestinal barrier in colitis mice by inhibiting the TLR signaling pathway (A) TEM of isolated EVs derived from wild-type macrophage and Stattic-treated macrophage. (B) Size distribution of EVs derived from wild-type macrophage and Stattic-treated macrophage analyzed by NTA. (C) Experiment design of the treatment of mice with DSS and Stattic-mEVs. (D) Body weight change in mice. (E) Disease activity index in mice. (F) Colonic picture and length analysis in mice. (G) Colonic morphology and histopathological score in mice. (H) The serum and colon tissue concentration of IL-1β, IL-6, TNF-α, IL-10, DAO, and D-LA in mice. (I) Colonic protein expression and analysis of ZO-1, Occludin, Claudin-1, TLR4, MyD88, and p-NF-κB p65 in mice. n = 8 for C–H; n = 3 for I. Data are expressed as the means ± SEM (D-I) and one-way ANOVA was performed, followed by Tukey's test (D-I). *P < 0.05, **P < 0.01, ***P < 0.001.

We also compared the effects of mEVs derived from untreated RAW264.7 cells to those of mEVs derived from RAW264.7 cells treated with Stattic on mouse intestinal epithelial cells in order to reveal the regulatory role of mEVs on the intestinal epithelial cell barrier (Fig. S12A). We found that the expression levels of mRNAs encoding ZO-1, occludin, and claudin-1 were significantly higher, and the expression of TLR4 and MyD88 and the phosphorylation of NF-κB were dramatically reduced, in epithelial cells treated with mEVs isolated from macrophages stimulated with LPS and treated with Stattic as compared with mEVs isolated from LPS-stimulated macrophages not treated with Stattic (Fig. S12B, C).

To determine whether the alleviating effect of mEVs on colitis is macrophage-dependent, DSS-induced and macrophage-depleted mice were fed EVs derived from untreated macrophages (DSS + WT-mEVs-E group) or EVs derived from Stattic-treated macrophages (DSS + Stattic-mEVs-E group) (Fig. S13A). Treatment with Stattic-mEVs was found to alleviate the DSS-associated loss of body weight and decreased DAI score in macrophage-depleted mice (Fig. S13B and C). DSS-induced, macrophage-depleted mice treated with Stattic-mEVs exhibited longer colon lengths (Fig. S13D) and significantly lower colon histopathological scores (Fig. S13E) than did similar mice that were not treated with EVs. In addition, treatment with Stattic-mEVs led to decreased concentrations of D-LA, DAO, TNF-α, IL-1β, and IL-6 in serum and colonic tissues of macrophage-depleted mice while increasing IL-10 (Fig. S13F). The expression of TLR4 and MyD88 proteins and the phosphorylation of NF-κB were reduced, but the protein levels of ZO-1, occludin, and claudin-1 were increased in colon tissues of the DSS + Stattic-mEVs-E group (Fig. S13G).

To clarify the role of Stattic-mEVs-derived proteins in alleviating colitis, mice were fed native Stattic-mEVs (DSS + Stattic-mEVs-NA group) or proteinase-treated Stattic-mEVs (DSS + Stattic-mEVs-PK group) (Fig. S14A). Proteinase-treated Stattic-mEVs failed to alleviate DSS-associated body weight loss and decreased DAI scores (Fig. S14B and C). The colonic lengths (Fig. S14D) and histopathological scores (Fig. S14E) of mice were not obviously different between DSS and DSS + Stattic-mEVs-PK groups. Moreover, we observed no changes in IL-1β, IL-6, TNF-α, IL-10, DAO, and D-LA levels in serum and colonic tissues of DSS-induced mice treated with Stattic-mEVs as compared to untreated DSS-induced mice (Fig. S14F).

We also treated mouse intestinal epithelial cells in vitro with native Stattic-mEVs (LPS + Stattic-mEVs-NA group) or protease-treated Stattic-mEVs (LPS + Stattic-mEVs-PK group) to reveal the regulatory role of protease-treated Stattic-mEVs on the intestinal epithelial cell barrier (Fig. S14G). The results showed that the levels of expression of ZO-1, occludin, and claudin-1 mRNA were comparable in cells treated with LPS and in cells treated with LPS and mEVs from Stattic-treated macrophages that had been pre-treated with proteinase (Fig. S14H).

The mEVs were found to contain DNA, RNA, and proteins, with the protein content being considerably higher than the DNA or RNA content (Fig. S15A). The protein profiles of WT-mEVs and Stattic-mEVs were determined using proteomic analysis. WT-mEVs and Stattic-mEVs share 1798 proteins (91.22 %), while WT-mEVs and Stattic-mEVs have 10 (0.51 %) and 163 (8.27 %) unique proteins, respectively (Fig. S15B). There were 263 differential proteins between the two groups, of which 200 and 63 were elevated in Stattic-mEVs and WT-mEVs, respectively (Fig. S15C). Subcellular localization revealed that these differential proteins were predominantly distributed in the cytoplasmic, nuclear, mitochondrial, endoplasmic reticulum (ER), plasma membrane, Golgi apparatus, extracellular, peroxisomes, and lysosomes (Fig. S15D). GO and KEGG pathways information for these differential proteins are presented in Fig. S15E and F, respectively. Details of the 263 differential proteins between WT-mEVs and Stattic-mEVs are listed in Tables S2 and S3.

Discussion

Intestinal microbiota and macrophage are recognized upstream modulators of UC [27,28]; however, details regarding the ways the gut microbiota and its derivatives, including EVs, might collaborate with macrophages to combat UC remain unclear. Here, we identified potential associations of L. johnsonii with host macrophage phenotypes and the intestinal barrier, both in human UC patients and in a mouse model of colitis. L. johnsonii and the EVs it produces were found to reduce the inflammation and intestinal barrier damage in DSS-induced animals in a macrophage-dependent way. GAPDH, a protein component of LjEVs, was shown to mitigate UC by inhibiting the MAPK-STAT3 signaling axis, prompting the conversion of macrophages to the anti-inflammatory phenotype. Finally, we discovered that EVs released by macrophages deficient in STAT3 activity boost the production of intestinal tight junction proteins in DSS-induced mice by blocking the TLR4 signaling pathway.

The gut microbiota and UC are inextricably linked. In the present study, we show that human UC patients and mice in a colitis model have a lower abundance of L. johnsonii than do healthy individuals. As a potentially beneficial bacterium, L. johnsonii has been widely demonstrated to have favorable host-regulatory effects, such as improving memory disorders, modulating metabolic disorders, and controlling inflammatory responses [21,29,30]. In particular, beneficial effects of L. johnsonii for UC patients have been identified recently. For example, L. johnsonii has been demonstrated to activate macrophages to release the IL-10, which helps to alleviate colitis [31]. Accordingly, our findings suggested that the abundance of L. johnsonii abundance was negatively connected with pro-inflammatory macrophage phenotypes and positively connected with anti-inflammatory macrophage phenotypes in human UC patients and DSS-induced mice. Further research revealed that L. johnsonii reduced DSS-induced colitis in mice in a macrophage-dependent way. A recent study demonstrated that this effect may not be due to the bacteria themselves, but instead the supernatants of cultures L. johnsonii were found alleviate colitis by inhibiting the polarization of pro-inflammatory macrophages [24]; however, the specific active ingredients and their contributions to this effect were unclear. Taken together, our research and these previous findings suggest that L. johnsonii plays a key role in the course of UC and that L. johnsonii-mediated modulation of macrophage phenotype contributes to the alleviation of UC. These ideas suggest that bioactive materials from L. johnsonii should be further uncovered.

Probiotic microbes have been widely recognized to regulate biological processes in cells of the host by secreting bioactive chemicals, in part through EVs, which are well-known mediators of intercellular crosstalk and signaling [15]. Bacterial EVs perform a range of functions in cell-to-cell contacts, including transferring and releasing effector chemicals that affect host signaling pathways and biological processes [32,33]. Early research on bacterial EVs concentrated mostly on their activities in pathogenic Gram-negative bacteria, where they were revealed to mediate bacterial pathogenicity and invasion by carrying toxins and virulence components to host cells [9]. Recently, it has been discovered that EVs are produced by a growing variety of Gram-positive bacteria, including Bacillus anthracis [34], Streptococcus pneumoniae [35], Bacillus subtilis [36], and Clostridium perfringens [37]. Lactobacillus are the most common gram-positive bacteria, and numerous Lactobacillus species, including L. casei [38], L. plantarum [39] and L. reuteri [40]. Our recent study revealed that EVs derived from L. mucosae has the potential to combat diarrheal diseases caused by ETEC K88 [23]. EVs derived from L. amylovorus alleviate aflatoxin B1-induced inflammatory intestinal injury through modulation of gut microbes and activation of AHR [41]. EVs secreted by L. reuteri promote macrophage conversion to an anti-inflammatory phenotype for mucosal and cutaneous wound healing [42]. In the present study, we discovered that L. johnsonii-derived EVs have the same efficacy against colitis as do their parental bacteria, and that this efficacy was dependent on the activity of host macrophages.

EVs from Lactobacillus strains are known to contain protein as the most abundant component with smaller amounts of nucleic acids [25,38,43]. Consistent with earlier findings, our biochemical investigations demonstrated that EVs produced from L. johnsonii contain DNA, RNA, and proteins, with protein concentration significantly higher than DNA and RNA. A previous study demonstrated the importance of protein components in L. reuteri-derived EVs in the maintenance of intestinal immune homeostasis [25]. Similarly, in the present study, we showed that digestion of the proteins of LjEVs rendered them less active in alleviating colitis, suggesting that some protein components were responsible for the anti-colitis activity. Therefore, we employed proteomics to analyze the protein composition in LjEVs and 883 proteins were detected. Next, we focused on the top 50 most abundant proteins to identify the core protein components carried by LjEVs that are closely associated with macrophage immune function. GAPDH has been demonstrated to be a moonlighting protein, playing roles not only in glycolysis but also in many other important physiological functions [44,45]. Study has shown that GAPDH derived from Lactobacillus gasseri alleviates allergic asthma by modulating macrophage immunometabolism [46]. An in vitro study revealed that, in addition to strongly inhibiting cell adhesion, spreading, and phagocytic functions of LPS-stimulated macrophages, GAPDH can significantly reduce the secretion of the pro-inflammatory cytokine TNF-α while enhancing the production of the anti-inflammatory cytokine IL-10 [47]. Based on these compelling pieces of evidence, we have sufficient reason to speculate that LjEVs may alleviate intestinal inflammatory responses by modulating macrophage polarization through their carried GAPDH. Among the top 50 most abundant proteins carried by LjEVs, in addition to GAPDH, several other proteins (such as pyruvate kinase, elongation factor Tu, phosphoglycerate kinase, and lysozyme) have also been reported to possess regulatory effects on macrophage functions. In mammals, pyruvate kinase exists as four major isoforms encoded by two genes, with tissue-specific isozymes generated through alternative splicing: pyruvate kinase muscle 1 (PKM1) and pyruvate kinase muscle 2 (PKM2) [48]. However, our proteomic sequencing data could not determine which specific isozyme is carried by LjEVs. Moreover, elongation factor Tu [49], phosphoglycerate kinase [50], and lysozyme [51] have been demonstrated to contribute to the progression of inflammation-related diseases. Our study specifically aims to identify proteins capable of modulating anti-inflammatory macrophage polarization. Collectively, we have selected GAPDH carried by LjEVs to perform subsequent experiments, aiming to elucidate its regulatory role in macrophage polarization under colitis conditions and uncover the underlying molecular mechanisms. In our study, GAPDH exerted an anti-colitis function by regulating the phenotype of macrophages in a manner consistent with treatment with L. johnsonii cells or with LjEVs.

As a result, we have gained a better understanding of the molecular processes via which GAPDH modulates intestinal macrophage function. As an important component of the natural immune system, macrophages colonize the gut lamina propria and are extensively involved in tissue repair, regeneration and inflammatory responses [52]. Macrophages are induced by the local microenvironment in vivo to be activated and differentiate into inflammatory M1 macrophages or anti-inflammatory M2 macrophages, with M1 and M2 macrophages exerting diametrically opposed biological functions by secreting pro- or anti-inflammatory cytokines, respectively [53]. It has been suggested that there is an inextricable link between the state of immune function of macrophages and the development of colitis [54]. Recent studies have shown that regulation of macrophage phenotype and function may provide a potential therapeutic strategy for Inflammatory bowel disease [55,56]. In the present study, in both mice with DSS-induced colitis in vivo and LPS-stimulated macrophages in vitro, externally applied GAPDH inhibited M1 polarization while increasing M2 polarization. We also found that GAPDH effectively reduced the expression of M1 macrophage makers and elevated the expression of M2 macrophage makers, demonstrated that GAPDH specifically regulates macrophage function under homeostatic conditions (Fig. S16). The MAPK pathway is involved in the regulation of biological processes such as inflammatory responses, and the MAPK pathway is particularly strongly regulated in macrophages [57,58]. In inflammatory bowel diseases, STAT3 is thought to be a pro-inflammatory signaling molecule [59], and inactivation of STAT3 in macrophages inhibits pro-inflammatory cytokine expression [60]. We found that phosphorylated STAT3 was significantly inhibited in Stattic treated macrophages (Fig. S17). We also found that this regulatory effect of GAPDH on macrophage phenotype and function occurs via the inhibition of the MAPK-STAT3 signaling axis. Furthermore, in vivo and in vitro analyses reconfirmed the contribution of STAT3-blocked macrophage to colitis remission.

Macrophages promote the proliferation of intestinal epithelial cells, regulate epithelial barrier integrity, and contribute to intestinal homeostasis [61]. To further understand macrophage-intestinal epithelial cell interactions, we studied the molecular pathways by which STAT3-inhibited macrophages improve intestinal barrier function. A recent study found that EVs produced from PTPN1-deficient macrophages improve intestinal barrier integrity by reprogramming macrophage polarization, which reduces pro-inflammatory cytokine release and enhances tight junction protein expression [62]. Similarly, we show in our study that EVs isolated from STAT3-inhibited macrophages but not from control macrophages significantly enhance tight junction protein expression in DSS-induced model mice and LPS-stimulated intestinal epithelial cells in vitro. Impaired gut barrier function is a frequently observed response to an enhanced inflammatory response, which is attributed to activation of pro-inflammatory pathways, including the TLR4-MyD88 signaling pathway [63]. Numerous studies have demonstrated that activation of the TLR4-MyD88 signaling pathway is a key event in the development of colitis, and that inhibition of the TLR-MyD88 signaling pathway in various ways is a reasonable strategy to combat colitis [64,65]. In our study, EVs derived from STAT3-inhibited macrophages significantly suppressed the expression of proteins in the TLR-MyD88 signaling pathway.

Recent literature has found that colonic macrophages are in direct proximity to epithelial crypt cells in mice and that bioactive molecules produced by macrophages support metabolic reprogramming of colonic epithelial cells [66]. Our study found that EVs from STAT3-inhibited macrophages similarly exerted a protective effect on intestinal barrier function in macrophage-depleted mice, suggesting that EVs secreted by anti-inflammatory macrophages directly regulate intestinal epithelial cell homeostasis. However, when these EVs were cleared of proteins, they lost their protective effects against colitis and on the intestinal barrier in mice, implying that certain proteins in macrophage-derived EVs are critical for enhancing intestinal barrier function. Although we identified numerous differential protein components between EVs from control and STAT3-inhibited macrophages by comparative proteomics, the specific proteins that exert the regulatory functions are currently being sought.

Although our study used extensive mouse and cells experiments to elucidated that GAPDH carried by L. johnsonii-derived EVs mitigates UC by mediating macrophage polarization, whether these important findings can be directly applied to the treatment of patients remains an open question. Further studies recruiting large numbers of human patients are needed to confirm its potential clinical application. In addition, clinical therapies based on GAPDH encapsulated in EVs secreted by L. johnsonii should also take into full consideration the possible side effects or differences in efficacy that may be brought about by individual differences between different patients.

Conclusion

In summary, this work provided novel evidence of microbiome-macrophage-host interactions in the context of UC. We have expanded the pathophysiological role of L. johnsonii by revealing the importance of GAPDH, a protein component of LjEVs, in the intercellular communication. GAPDH-mediated anti-inflammatory macrophage polarization alleviated intestinal inflammation associated with colitis and improved enhancement of intestinal barrier function in mice. Our study highlights the idea that EVs secreted by L. johnsonii could serve as research targets for the development of UC treatment approaches.

Credit author statement

The authors’ contributions are as follows: Shiyu Tao, Tengfei Lv, and Hong Wei designed the experiments. Shiyu Tao, Mengzhen Song, and Jinping Fan collected the samples and participated in experiments. Feng Zhu analyzed the data. Shiyu Tao wrote the manuscript. All authors read and approved the final manuscript.

Compliance with ethics requirements

This study involves human participants. Ethical approval was obtained from the Ethics Committee of Jinling Hospital (Approval number 2021DZGZR-YBB-054). Participants gave informed consent to participate in the study before taking part. All procedures involving animals were approved by the Huazhong Agricultural University of Health Guide for the Care and Use of Laboratory Animals (Approval number HZAUMO-2023-0097).

Funding

This work was supported by the National Key Research and Development Program (2022YFA1304104), and the National Nature Science Foundation of China (32272898; 82100591).

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Biographies

ShiyuTao is an associate professor at Huazhong Agricultural University, working on interactions between intestinal microbiome and host and their molecular mechanisms. He has published a total of 25 papers as first or corresponding author in Journal of Advanced Research, Environment International, Gut Microbes, npj Biofilms and Microbiomes, Science of the Total Environment, FASEB Journal, Journal of Nutrition, Frontiers in Microbiology, Journal of Cellular Physiology and other journals. He contributed to this study by designing the experiments, collecting the samples and participated in experiments, and writing the manuscript.

MengzhenSong is a master student at Huazhong Agricultural University, working on interactions between intestinal microbiome and host and their molecular mechanisms. She contributed to this study by collecting the samples and participated in experiments.

Jinping Fan is a master student at Huazhong Agricultural University, working on interactions between intestinal microbiome and host and their molecular mechanisms. He published paper as first author in Environment International. He contributed to this study by collecting the samples and participated in experiments.

Feng Zhu is a PhD student at Nanjing University, working on interactions between intestinal microbiome and etiology of inflammatory bowel disease. He published paper as first author in Journal of Crohns and Colitis. He contributed to this study by analysing the data.

Tengfei Lv is a PhD student at Nanjing University, working on interactions between intestinal microbiome and etiology of inflammatory bowel disease. He published paper as first author in Journal of Cachexia, Sarcopenia and Muscle. He contributed to this study by designing the experiments.

Hong Wei is a PI at Huazhong Agricultural University. His research works focus on the cross-talk between gut microbiota and host. He has published a total of 83 papers as corresponding author in Cell, Gut, Nature Immunology, Molecular Psychiatry, Gastroenterology, Blood, Science Advances, Cell Research, Advanced Science, Ann Rheum Dis, Microbiome, Elife and other journals. He contributed to this study by designing the experiments.

Appendix A. Supplementary data

The following are the Supplementary data to this article: