Portulaca oleracea L.-derived exosome-like nanoparticles ameliorate colitis by modulating gut microbiota and metabolites to reduce endoplasmic reticulum stress

Abstract

BACKGROUND

Ulcerative colitis (UC) is associated with excessive endoplasmic reticulum stress (ERS) and gut microbiota dysbiosis. Portulaca oleracea L. (P. oleracea L.), as a food-as-medicine, has demonstrated potential in alleviating colitis symptoms, enhancing barrier function, anti-inflammatory, antioxidant, and cleaning up free radical. However, its role in modulating ERS, gut microbiota and metabolites remained unclear.

AIM

To explore how P. oleracea L.-derived exosome-like nanoparticles (PELNs) alleviate colitis in mice by modulating gut microbiota and metabolites to mitigate ERS.

METHODS

In this study, PELNs were isolated and characterized. Colitis was induced by dextran sulfate sodium and microbiota-transplanted mouse models, followed by relevant drug intervention period. Mechanistic investigations were to examine inflammatory factors, oxidative stress and ERS. Fecal samples collection for gut microbiota and metabolite was analyzed.

RESULTS

This study demonstrated that PELNs intervention alleviated colitis through the modulation of the intestinal microbial community and its associated metabolites, thereby reducing ERS by inhibiting the PERK-eIF2α-ATF4 signaling pathway, reducing inflammatory cytokines, and restoring tight junctions in intestinal epithelial cells.

CONCLUSION

PELNs ameliorate UC by modulating gut microbiota and metabolites, reducing ERS. These findings highlight PELNs as a novel, food-derived, colon-targeted therapeutic, offering a promising approach for UC treatment.

Key Words: Portulaca oleracea L.-derived exosome-like nanoparticles; Ulcerative colitis; Endoplasmic reticulum stress; Tight junction; PERK-eIF2α-ATF4 signaling pathway; Gut microbiota; Metabolites

Core Tip: The role of Portulaca oleracea L. (P. oleracea L.) in modulating endoplasmic reticulum stress, gut microbiota and metabolites remained unclear. To investigate this, we prepared exosome-like nanoparticles derived from P. oleracea L. and modeled colitis induced by dextran sulfate sodium and microbiota-transplanted mouse. We found that P. oleracea L.-derived exosome-like nanoparticles (PELNs) intervention alleviated colitis through the modulation of the intestinal microbial community and metabolites. PELNs as a novel, food-derived, colon-targeted therapeutic, offering a promising approach for ulcerative colitis treatment.

INTRODUCTION

Ulcerative colitis (UC) is a chronic, nonspecific inflammatory disease that primarily involves the mucosa and submucosa of the colon and rectum. Characterized by a relapsing and remitting course, it presents with clinical symptoms such as mucopurulent and bloody stools, abdominal pain, diarrhea, and tenesmus. Key pathological features include mucosal inflammation and ulceration, making UC one of the most refractory diseases in the digestive system[1]. Over the past four decades, epidemiological data have shown a 1.5-fold to nearly 20-fold increase in the incidence and prevalence of UC in several Asian countries[2]. In recent years, the treatment focus has shifted from simple anti-inflammatory or immunosuppressive therapies to strategies aimed at promoting intestinal mucosal healing[3]. Previous studies have revealed significant alterations in gut microbiota diversity and abundance in patients with UC. Dysbiosis can trigger abnormal mucosal immune responses and excessive inflammation, leading to increased injury of intestinal epithelial cells (IECs) and compromising intestinal barrier function, thereby contributing to UC initiation and progression[4]. Repairing the damaged intestinal mucosa is now recognized as a critical therapeutic target for UC treatment.

IECs are crucial for maintaining the intestinal mucosal barrier and possess a highly developed endoplasmic reticulum (ER). Persistent and severe ER stress (ERS) disrupts this barrier function, contributing to UC pathogenesis[5]. Tight junctions between IECs are essential for mucosal barrier integrity and regulating intestinal permeability. Under pathological conditions, structural and functional disruption of these tight junctions compromises the barrier, facilitating disease progression[6].

Plant-derived exosome-like nanoparticles can regulate gut microbiota and metabolites, which have been applicated as bioactive components and drug carriers targeting the colon[7]. Experimental studies have demonstrated that oral administration of garlic-derived plant exosomes rebalances the dysregulated gut microbiota in dextran sulfate sodium (DSS)-induced colitis model mice, alleviates symptoms, reduces inflammation, improves intestinal barrier integrity, and mitigates UC[8]. Portulaca oleracea L. (P. oleracea L.), as a food-medicinal herb and a potential therapeutic drug intervention in UC, is rich in diverse bioactive constituents such as flavonoids, alkaloids, organic acids, terpenoids, and vitamins, granting it anti-inflammatory, immunomodulatory, antioxidant and antimicrobial[9]. P. oleracea L. extract alleviates ERS-mediated IECs injury by modulating the PERK/eIF2α/Beclin1/LC3 autophagy pathway[10].

P. oleracea L. has potential reparative effects on intestinal inflammation and damage to IECs. This study isolated and purified natural P. oleracea L.-derived exosome-like nanoparticles (PELNs) from P. oleracea L. to explore the relationship between gut microbiota, their metabolites, and ERS in UC, as well as the pathological mechanisms regulating IEC tight junctions. This was accomplished using DSS-induced colitis and fecal microbiota transplantation (FMT) colitis mouse models. Our findings demonstrate that PELNs alleviate UC by modulating the gut microbiota and metabolites, intervening in IEC tight junction injury through the ERS-associated PERK/eIF2α/activating transcription factor4 (ATF4) pathway, and ultimately promoting mucosal healing and barrier repair.

MATERIALS AND METHODS

Preparation and characterization of PELNs

Preparation of PELNs: Fresh P. oleracea L. samples were minced, homogenized in PBS, and incubated overnight at 4 °C. After additional PBS was added, the mixture underwent high-speed intermittent homogenization (8-10 cycles of 1 minute each). The homogenate was filtered and centrifuged sequentially at 4 °C: 1000 × g for 10 minutes, 2000 × g for 20 minutes, 4000 × g for 30 minutes, and 10000 × g for 60 minutes. The supernatant was filtered through a 0.22-μm membrane and ultracentrifuged at 120000 × g for 70 minutes at 4 °C. The pellet was resuspended in PBS, aliquoted, and stored at -80 °C.

Transmission electron microscopy of PELNs: For characterization, a 20 μL aliquot of the PELNs suspension was adsorbed onto a copper grid for 5-10 minutes and air-dried. After negative staining with 20 μL of 2% phosphotungstic acid for 3-5 minutes, the sample was dried under an incandescent lamp and examined by transmission electron microscopy (TEM).

Particle size analysis of PELNs: Particle size of the PELNs preparation was determined by nanoparticle tracking analysis after dilution with sterile PBS and filtration through a 0.22-μm membrane.

Metabolomic profiling by ultra-high-performance liquid chromatography coupled with quadrupole-orbitrap high-resolution mass spectrometry

Lyophilized PELNs were dissolved in methanol: Water (3:1, v/v), then vortexed, sonicated on ice, and stored at -40 °C for 1 hour. After centrifugation at 12000 rpm for 15 minutes at 4 °C, 900 µL of supernatant was dried and reconstituted in 200 µL methanol: Water (3:1, v/v) containing internal standards. Following vortexing, sonication, and recentrifugation, the supernatant was analyzed by ultra-high-performance liquid chromatography coupled with quadrupole-orbitrap high-resolution mass spectrometry (UHPLC-QE-MS) using a Waters ACQUITY UPLC HSS T3 column (2.1 mm × 100 mm, 1.8 µm) with mobile phases A (5 mmol/L ammonium acetate with acetic acid) and B (acetonitrile).

Establishment of DSS-induced colitis mouse model

All procedures were approved by the Experimental Animal Ethics Committee of Shanghai University of Traditional Chinese Medicine (Approval No. PZSHUTCM2302270016). Forty male C57BL/6 mice (6-8 weeks old) were acclimated under SPF conditions for one week. A random number generator was used to assign mice to four groups (n = 10): Control, DSS, 5-aminosalicylic acid (5-ASA), and PELNs. Colitis was induced by providing 3% DSS in drinking water for 7 days[11]. Starting on day 8, the PELNs group received 15 mg/kg/day PELNs by gavage, the 5-ASA group received 100 mg/kg/day 5-ASA, and the Control and DSS groups received normal saline for 6 days. Daily observations included body weight, general condition, and fecal occult blood. Disease activity was scored using the disease activity index (DAI). Colon tissues were collected for further analysis. During the experiment, three mice in the DSS group died due to severe colitis, while one mouse in each of the 5-ASA and PELNs groups died. Body weight changes and DAI scores were dynamically recorded for each surviving mouse throughout the experiment. On day 13 body weight and colon length were analyzed using the remaining number of mice in each group. For all other experiments, the final statistical analysis was performed with n = 3 per group. For gut microbiota and metabolomics analyses, the sample size was n = 6 per group.

Establishment of the FMT colitis model

Thirty male C57BL/6 mice (6-8 weeks old) were acclimatized and then administered an antibiotic cocktail (200 mg/kg ampicillin, 200 mg/kg neomycin, 200 mg/kg metronidazole, and 100 mg/kg vancomycin) by gavage for 7 days to establish a pseudo-germ-free model. The mice were randomly divided into three groups (n = 10 per group) using a random number generator: FMT-Control, FMT-DSS, and FMT-PELNs. Donor mice (from corresponding Control, DSS, and PELNs groups) provided fecal samples, which were homogenized in normal saline (1:10 w/v), centrifuged, and the resulting supernatant was collected anaerobically. Recipient mice were gavaged daily with 300 µL of the suspension for 7 days. After a 3-day rest, the FMT-DSS and FMT-PELNs groups received 3% DSS drinking water for 7 days to induce colitis, while the FMT-Control group received normal water[12]. Body weight, general condition, fecal occult blood, and DAI scores were recorded daily. Colon samples were collected post-experiment for analysis. Fecal samples were collected from donor mice on days 11 to 13 of the experiment to prepare the fecal microbiota suspension. The final day of drug administration in the donor mice coincided with the last day of pseudo-germ-free modeling in the recipient mice.

Histopathological analysis

Colon tissues were fixed in 10% formalin, embedded in paraffin, and sectioned at 5 μm. Sections were stained with hematoxylin & eosin (H&E) for microscopic assessment of neutrophil infiltration and inflammatory changes.

TEM analysis

Colon tissues were fixed in glutaraldehyde and osmium tetroxide, then processed through dehydration, permeabilization, embedding, and sectioning. Ultrathin sections were double-stained with 3% uranyl acetate and lead citrate and observed by TEM.

Immunofluorescence staining

Paraffin sections were baked, deparaffinized, and subjected to antigen retrieval. After permeabilization and 3% BSA blocking, sections were incubated overnight at 4 °C with primary antibodies against ZO-1 (1:1000; CST, #13663S) and Bip (1:200; Proteintech, #11587-1-AP), then with FITC-conjugated secondary antibodies for 1 hour at room temperature. Nuclei were counterstained with DAPI, followed by coverslipping. Imaging was performed on a Zeiss LSM 800 confocal microscope.

Enzyme-linked immunosorbent assay

Supernatants from homogenized colon tissues were collected after centrifugation. Levels of myeloperoxidase (MPO), interleukin (IL)-10, IL-18, IL-28, IL-36, and IL-37 were measured using commercial enzyme-linked immunosorbent assay kits (Mlbio, China) according to the manufacturer’s protocol, and expressed as pg/mg protein.

Western blot analysis

Colon tissues were lysed in RIPA buffer, centrifuged, and protein concentrations were determined by BCA assay. Equal amounts of protein were separated by SDS-PAGE, transferred to PVDF membranes, and blocked for 15 minutes. Membranes were incubated overnight at 4 °C with primary antibodies (all 1:1000) including β-Actin, PERK, phosphorylated PERK (p-PERK), eIF2α, phosphorylated eIF2α (p-eIF2α), ATF4, Bip, Claudin-4, ZO-1, Claudin-2, and Occludin. After 1-hour incubation with HRP-conjugated secondary antibodies (1:5000), protein bands were visualized by ECL and analyzed using ImageJ.

Quantitative real-time PCR

Total RNA extracted from tissues using TRIzol was reverse transcribed into cDNA. Real-time qPCR with SYBR Green was performed to quantify gene expression. Primer sequences are listed in Table 1. Expression levels were normalized to β-actin and calculated using the 2^-ΔΔCt method.

Table 1 Primer sequences.

Gene		Primer sequences
Mus PERK	Forward	ACGGGACAAGTAGGGACCAA
	Reverse	CTGGGTGCTGAATGGGTAGA
Mus eIF2α	Forward	GCTCCACCCAGGTATGTGATG
	Reverse	TTTCTCTCTCCAGCCGTTCC
Mus Bip	Forward	GAACACTGTGGTACCCACCAAGAA
	Reverse	TCCAGTCAGATCAAATGTACCCAGA
Mus ATF4	Forward	GAAGAGGTCCGTAAGGCAAGG
	Reverse	CAGCAAACACAGCAACACAAGA
Mus Occludin	Forward	AAGTGAATGGCAAGCGATCATA
	Reverse	CTGTACCGAGGCTGCCTGAA
Mus Claudin-2	Forward	ACACTTGAGTCATCGCCCATC
	Reverse	AATCCCAGGCAGAAGTTCACC
Mus Claudin-4	Forward	CGTCATCCGCGACTTCTACA
	Reverse	GAGTAGGGCTTGTCGTTGCT
Mus ZO-1	Forward	TCCCACAAGGAGCCATTCCTG
	Reverse	GGGCTCAGCAGAGTTTCACCT
Mus β-actin	Forward	CATCCGTAAAGACCTCTATGCCAAC
	Reverse	ATGGAGCCACCGATCCACA

Fecal DNA extraction and 16S rRNA sequencing

Intestinal contents from all groups were collected for 16S rDNA sequencing. The V3-V4 hypervariable regions were amplified by PCR with primers 338F/806R. Purified amplicons were used for library construction (NEXTFLEX Rapid DNA-Seq Kit, Bioo Scientific, United States) and sequenced on the Illumina NextSeq 2000 PE300 platform. Quality-filtered sequences were clustered into operational taxonomic units at 97% similarity using USEARCH (v11). Alpha diversity (Chao and Shannon indices) was calculated using mothur (v1.30.2), and group differences were assessed by the Wilcoxon rank-sum test. Beta diversity was evaluated by principal coordinate analysis (PCoA) based on Bray-Curtis distances, with significant community differences tested by PERMANOVA. Linear discriminant analysis effect size (LEfSe) identified differentially abundant bacterial taxa (phylum to genus; LDA > 2, P < 0.05). A P value < 0.05 was considered statistically significant.

Fecal untargeted metabolomics by LC-MS

Fecal samples were homogenized in a methanol-water extraction solvent containing an internal standard (L-2-chlorophenylalanine) using low-temperature grinding. After ultrasonic extraction, incubation, and centrifugation, the supernatant was collected for LC-MS analysis. Quality control (QC) was conducted using pooled QC samples, injected every 5-15 analytical samples to monitor reproducibility. Chromatographic separation was conducted on a Thermo UHPLC-Exploris240 system with an HSS T3 column under gradient elution. Mass spectrometry was performed in positive and negative ionization modes using data-dependent acquisition. Raw data were processed with Progenesis QI for peak picking, alignment, and database matching. Metabolites were identified by comparing data against the HMDB (http://www.hmdb.ca/), Metlin (https://metlin.scripps.edu/) and a Majorbio custom database. The resulting data matrix was filtered for missing values, normalized, and log-transformed for statistical analysis. Multivariate analyses [principal component analysis (PCA) and orthogonal partial least squares-discriminant analysis (OPLS-DA)] were performed using the ropls package in R, with model robustness assessed by 7-round cross-validation. Metabolites with VIP > 1 and P < 0.05 were considered significantly altered. Pathway annotation was performed using the KEGG database, and enriched pathways were identified by Fisher’s exact test (scipy.stats module in Python).

Statistical analysis

All statistical analyses were performed using GraphPad Prism 8.0. After testing for normality and homogeneity of variance, one-way analysis of variance was used for multiple group comparisons, otherwise, the Kruskal-Wallis test was applied. Data are presented as mean ± SEM, P < 0.05 was considered statistically significant.

RESULTS

Identification and characterization of PELNs

PELNs were isolated and purified from P. oleracea L. using differential centrifugation followed by membrane filtration (Figure 1A). TEM revealed that the PELNs had a spherical morphology with a distinct bilayer membrane structure and internal electron-dense regions. The particles exhibited uniform morphology, structural integrity, and good dispersibility (Figure 1B). The average particle size of PELNs was 58.5 nm, with a concentration of 2.11 × 10⁹ ± 1.01 × 10⁷ particles/mL (Figure 1C). UHPLC-QE-MS analysis identified 274 compounds in the PELNs samples based on multi-stage mass spectrometry data, cross-referenced with natural product high-resolution mass spectrometry databases and relevant literature. Rigorous screening [level B (i) with MS2 score > 0.9] led to the identification of 36 compounds (Supplementary Table 1), including 2-hydroxypyridine, benzoic acid, adenosine, adenine, phenylethylamine, picolinic acid, 4-hydroxycinnamic acid, 3-hydroxycinnamic acid, kynurenic acid, suberic acid, 2-isopropylmalic acid, azelaic acid, gentisic acid, dl-tryptophan, phthalic acid, l-pyroglutamic acid, salsolinol, chlorogenic acid, d-pantothenic acid, 3,5-dihydroxybenzoic acid, dl-phenylalanine, guanosine, guanine, demethyl-coclaurine, cytosine, neolitsine, n-feruloylputrescine, isochamaejasmine, cryptochlorogenic acid, l-canavanine, fustin, methylmalonic acid, trans-ferulic acid, salidroside, 3,4-dihydroxyphenylacetic acid, and 6-hydroxynicotinic acid. The corresponding spectra are shown in Figure 1D and E.

Figure 1 Preparation, identification, and characterization of Portulaca oleracea L.-derived exosome-like nanoparticles. A: Schematic workflow for the isolation and purification of Portulaca oleracea L.-derived exosome-like nanoparticle (PELNs); B: Representative transmission electron microscopy image showing the morphology of PELNs (scale bar: 100 nm); C: Particle size distribution of PELNs as determined by nanoparticle tracking analysis; D: Total ion chromatogram (TIC) of PELNs obtained in positive ionization mode using ultra-high-performance liquid chromatography coupled with quadrupole-orbitrap high-resolution mass spectrometry (UHPLC-QE-MS); E: TIC of PELNs obtained in negative ionization mode using UHPLC-QE-MS.

PELNs protect against DSS-induced IEC injury and ameliorate the inflammatory status

The effects of PELNs on UC were assessed in a DSS-induced colitis model. The DSS group displayed typical pathological symptoms, including lethargy, reduced activity, rough fur, diarrhea, bloody stools, and body weight loss. After treatment with PELNs or 5-ASA, the general condition of the mice improved significantly, as evidenced by increased activity, normalized stool consistency, negative fecal occult blood results, and restored body weight. Moreover, PELNs significantly reduced the DAI, bringing it to a level comparable to the Control group, and effectively prevented DSS-induced colon shortening (Figure 2A-G).

Figure 2 Portulaca oleracea L.-derived exosome-like nanoparticles alleviate dextran sulfate sodium-induced colitis in mice. A: Schematic diagram of the experimental timeline; B: Body weight changes of mice throughout the experiment; C: Disease activity index scores of mice; D: Representative images and measurement of colon lengths; E: Fecal occult blood test results on day 13; F: Body weight of mice on day 13; G: Colon length of mice on day 13; H: Hematoxylin & eosin staining (× 50, × 100); I: Transmission electron microscopy observation of the ultrastructure of tight junctions and endoplasmic reticulum in intestinal epithelial cells (× 6000, × 8200); J: Representative immunofluorescence images showing Bip (red), ZO-1 (green), and DAPI (blue) staining in mouse colon tissues (× 200); K: Quantitative analysis of the relative fluorescence intensity of Bip and ZO-1 in mouse colon tissues. ^bP < 0.01, ^cP < 0.001, ^dP < 0.0001. NS: Not significant; DSS: Dextran sulfate sodium; PELNs: Portulaca oleracea L.-derived exosome-like nanoparticles; 5-ASA: 5-aminosalicylic acid.

H&E staining revealed intact intestinal epithelial architecture in the Control group. In contrast, the DSS group displayed significant pathological alterations in the colonic epithelium, including disordered and absent crypt structures, extensive inflammatory cell infiltration, crypt abscess formation, markedly reduced goblet cell numbers, and epithelial erosion and ulceration. After PELNs treatment, colonic tissue structure was significantly improved compared to the DSS group, showing an intact mucosal layer and glandular architecture with regular arrangement, markedly reduced inflammatory cell infiltration, preserved crypt structures, and no ulcer formation (Figure 2H).

TEM analysis revealed that IECs in the Control group exhibited normal ultrastructural features: Neatly arranged, dense microvilli on the cell surface, intact and clear intercellular junction complexes, abundant mitochondria with well-preserved cristae in the cytoplasm, uniformly distributed rough ER without dilation, and regularly shaped nuclei with homogeneous chromatin. In the DSS group, however, IECs showed a reduction or loss of microvilli, disrupted tight junctions, swollen mitochondria, dilated ER, and irregularly shaped nuclei, with visible ulcer healing surfaces. Both the 5-ASA and PELNs groups showed restoration of microvilli in terms of number, uniform length, and arrangement, altered tight junction morphology (shorter length), mildly dilated ER, normal mitochondrial morphology, and regularly structured nuclei with intact nuclear membranes and evenly distributed chromatin. These results demonstrate that PELNs significantly reverse DSS-induced colonic injury in mice (Figure 2I).

Immunofluorescence analysis of ZO-1 and Bip was performed to evaluate the effects of PELNs on ERS and tight junctions in IECs of DSS-induced mice. The DSS group had significantly lower ZO-1 intensity and higher Bip intensity than the Control group (P < 0.05). Conversely, the 5-ASA and PELNs groups showed significantly higher ZO-1 intensity and lower Bip intensity than the DSS group (P < 0.05) (Figure 2J and K).

PELNs regulated inflammatory cytokine, reducing ERS and restoring tight junctions in IECs in mice

Colonic inflammation was assessed by evaluating the levels of pro-inflammatory and anti-inflammatory cytokines, including MPO, IL-10, IL-18, IL-28, IL-36, and IL-37. The PELN treatment group showed significant immunomodulatory effects, effectively reducing the expression levels of pro-inflammatory mediators IL-28, IL-36, IL-18, and MPO, while significantly increasing the concentrations of anti-inflammatory cytokines IL-10 and IL-37. These results suggest that PELNs confer a notable protective effect on the inflammatory response of epithelial cells (Figure 3A).

Figure 3 Portulaca oleracea L.-derived exosome-like nanoparticles regulated inflammatory cytokine, reducing endoplasmic reticulum stress and restoring tight junctions in dextran sulfate sodium-induced colitis mice. A: The expression of interleukin (IL)-28, IL-18, IL-36, myeloperoxidase, IL-37, and IL-10 in colon tissues; B: Immunoblot bands of ZO-1, PERK, phosphorylated PERK, Bip, Occludin, ATF4, eIF2α, phosphorylated eIF2α, Claudin-4, and Claudin-2 in colon tissues; C: Statistical graphs of relative protein expression levels; D: QPCR analysis of the relative gene expression levels of PERK, eIF2α, ATF4, Bip, Claudin-2, Claudin-4, ZO-1and Occludin in colon tissues. ^aP < 0.05, ^bP < 0.01, ^cP < 0.001, ^dP < 0.0001. NS: Not significant; DSS: Dextran sulfate sodium; PELNs: Portulaca oleracea L.-derived exosome-like nanoparticles; 5-ASA: 5-aminosalicylic acid; IL: Interleukin; p-PERK: Phosphorylated PERK; p-eIF2α: Phosphorylated eIF2α.

Previous studies have highlighted the pivotal role of ERS in the pathogenesis of colitis, with PERK, eIF2α, ATF4, and Bip serving as key indicators of its activation. Inflammation and persistent, severe ERS contribute to the dysfunction of the intestinal mucosal barrier, while tight junction proteins in IECs, such as Occludin, Claudin-2, ZO-1, and Claudin-4, are critical for maintaining mucosal integrity and regulating epithelial permeability. Therefore, both protein and mRNA levels of these markers were assessed. Western blot analysis revealed significant differences between the DSS-treated and Control groups: The ratios of p-PERK to total PERK and p-eIF2α to total eIF2α, as well as the expression levels of ATF4 and Bip, were significantly upregulated (P < 0.05). In contrast, the expression of tight junction proteins ZO-1, Occludin, and Claudin-4 was significantly reduced (P < 0.05). Therapeutic interventions with both 5-ASA and PELNs effectively corrected these protein expression abnormalities, resulting in significantly lower p-PERK/PERK and p-eIF2α/eIF2α ratios, as well as reduced ATF4 and Bip levels compared to the DSS group (P < 0.05). Additionally, the expression of ZO-1, Occludin, and Claudin-4 (P < 0.05) was markedly restored (Figure 3B and C).

Similarly, qPCR revealed that relative to the Control group, the DSS group exhibited significantly higher mRNA expression of PERK, eIF2α, ATF4, Bip, and Claudin-2, and significantly lower expression of ZO-1, Occludin, and Claudin-4 (P < 0.05). Compared to the DSS group, PELN treatment significantly reduced the levels of the former set of genes and increased those of the latter (P < 0.05; Figure 3D). These results suggest that PELNs improve intestinal mucosal barrier function by inhibiting ERS.

PELNs modulated gut microbiota and metabolites

The gut microbiota profile was analyzed using 16S rRNA sequencing. The results showed that PELNs ameliorated the decrease in alpha diversity indices (Ace, Chao) of the gut microbiota induced by DSS (Figure 4A and B). PCoA and Non-metric Multidimensional Scaling revealed that PELNs significantly reduced the differences in gut microbiota between the Control and DSS groups (Figure 4C and D), indicating their potential in regulating gut microbial balance. In the microbiota characterization analysis, the gut health index in the DSS group was significantly decreased, while the dysbiosis index increased. After PELNs intervention, the gut microbiota dysbiosis was shifted toward a healthier state (Figure 4E). Further analysis of the microbial community composition revealed that at the phylum level, compared to the Control group, the DSS group showed increased abundances of Firmicutes, Proteobacteria, and Verrucomicrobiota, and a decreased abundance of Bacteroidota. The PELNs and 5-ASA groups counteracted these changes. At the genus level, PELNs reduced the levels of Bacteroides, Turicibacter, Rikenellaceae_RC9_gut_group, Lactococcus, and Enterococcus in colitis mice, while increasing the levels of Muribaculaceae, Prevotellaceae_UCG-001, and o__RF39 (Figure 4F and G). LEfSe was used to identify species with an LDA score > 4, which were collected in a histogram to identify statistically significant biomarkers between groups. Muribaculaceae, Prevotellaceae_NK3B31_group, Actinobacteriota, Bifidobacterium, Blautia, and Acetatifactor were more abundant in the Control group. Taxa such as Firmicutes, Bacilli, Bacteroides, Turicibacter, and Rikenellaceae were highly enriched in the guts of DSS group mice. After drug intervention, Lachnospirales, Lachnospiraceae_NK4A136_group, and Prevotellaceae_UCG-001 emerged as significantly different biomarkers in the 5-ASA group, while Erysipelotrichaceae and Dubosiella were significantly different in the PELNs group (Figure 4H).

Figure 4 Portulaca oleracea L.-derived exosome-like nanoparticles regulate the gut microbiota in dextran sulfate sodium-induced colitis mice. A: Alpha diversity of mouse gut microbiota assessed by the Ace index; B: Alpha diversity of mouse gut microbiota assessed by the Chao index; C: Principal coordinate analysis; D: Non-metric Multidimensional Scaling; E: Differential analysis of the Gut Microbiome Health Index (upper images), analysis of the microbiota dysbiosis index (lower images); F: Bar plot showing community composition at the phylum and genus levels for each group; G: Comparison of community differences at the phylum and genus levels; H: Linear discriminant analysis effect size for discriminating multi-level species differences. ^aP < 0.05, ^bP < 0.01, ^cP < 0.001 (A, B, and G: Statistical method: Kruskal-Wallis test with FDR correction; post-hoc test: Dunn's test. E: Statistical method: Wilcoxon test with multiple correction). DSS: Dextran sulfate sodium; PELNs: Portulaca oleracea L.-derived exosome-like nanoparticles; 5-ASA: 5-aminosalicylic acid; OTU: Perational taxonomic unit; PCoA: Principal coordinate analysis; LEfSe: Linear discriminant analysis effect size; NMDS: Non-metric Multidimensional Scaling.

Significant alterations in fecal metabolic profiles were observed in mice following DSS exposure and PELNs intervention. PCA revealed distinct overall metabolic profiles: The DSS group separated from the Control group, while the 5-ASA and PELNs groups partially overlapped with DSS but shifted closer to Control, indicating a healthier metabolic state (Figure 5A). Partial least squares-discriminant analysis (PLS-DA) showed that Component 1 and 2 explained 23.6% and 13.1% of the variance, respectively (left image in Figure 5B), with a separation pattern similar to PCA (5-ASA and PELNs groups closer to Control). The PLS-DA model was well-fitted and highly predictable (all R² > Q²), with a Q² regression line intercept of -0.4093, confirming its suitability for further analysis (right image in Figure 5B). Differential metabolites were screened using VIP > 1 (OPLS-DA), nominal P < 0.05 (Student’s t-test), and fold change > 1. In the DSS vs Control comparison, 998 differential metabolites were identified (448 up, 550 down). Relative to the DSS group, the PELNs group showed 148 significant differences (71 up, 77 down) (Figure 5C).

Figure 5 Portulaca oleracea L.-derived exosome-like nanoparticles regulate gut metabolites in dextran sulfate sodium-induced colitis mice. A: Principal component analysis; B: Partial least squares-discriminant analysis; C: Volcano plot of differential metabolites; D: Bubble chart of KEGG pathway enrichment analysis for differential metabolite sets in mouse gut; E: Correlation analysis between gut microbiota and ulcerative colitis (UC)-related indicators; F: Correlation analysis between metabolites and UC-related indicators; G: Correlation analysis between gut microbiota and metabolites. ^aP < 0.05, ^bP < 0.01, ^cP < 0.001. DSS: Dextran sulfate sodium; PELNs: Portulaca oleracea L.-derived exosome-like nanoparticles; 5-ASA: 5-aminosalicylic acid; PCA: Principal component analysis; PLS-DA: Partial least squares-discriminant analysis.

KEGG enrichment analysis of differential metabolites (top 20 results by P-value) showed that DSS treatment significantly affected pathways including Linoleic acid metabolism, Cutin, suberine and wax biosynthesis, and alpha-linolenic acid metabolism (P < 0.05). Specifically, DSS down-regulated lipid metabolites, including alpha-linolenic acid, jasmonic acid, and gamma-linolenic acid. Following drug intervention, the 5-ASA group exhibited significant enrichment in pathways such as Axon regeneration, Phenylalanine, tyrosine and tryptophan biosynthesis, and Protein digestion and absorption (P < 0.05). The PELNs group, on the other hand, showed significant enrichment in pathways such as Axon regeneration, Choline metabolism in cancer, and Shigellosis (P < 0.05). Differential metabolites associated with these pathways revealed that PELN treatment up-regulated metabolites such as L-isoleucine and DG (15:0/18:1(9Z)/0:0), while down-regulating arachidonoyl ethanolamide and LysoPC (20:2(11Z,14Z)/0:0), suggesting that PELNs may exert diverse biological effects, including immunomodulation, maintenance of metabolic homeostasis, and regulation of neural function (Figure 5D).

Subsequently, 31 differential metabolites were identified (Supplementary Table 2) based on the criteria of VIP > 2 in OPLS-DA, P < 0.05 in Student’s t-test, and FC > 2. Spearman correlation analysis was performed to construct a heatmap, revealing significant correlations among genus-level gut microbiota, metabolites, and various UC-related indicators, including body weight, colon length, DAI, ERS and tight junction proteins, and inflammatory markers (Figure 5E-G). Lactococcus, Enterococcus, Bacteroides, and Rikenellaceae_RC9_gut_group showed significant positive correlations with UC pathological indicators, while Prevotellaceae_UCG-001, Muribaculaceae, and o__RF39 were positively correlated with tight junction proteins and anti-inflammatory cytokines. Furthermore, the 31 differential metabolites exhibited strong correlations with UC-related indicators, highlighting their significant roles in the UC process. Notably, Solamargine, Mofebutazone, 3-Indole carboxylic acid glucuronide, and Ketorolac were strongly associated with ERS. Significant correlations were also observed between Enterococcus, Muribaculaceae, and o__RF39 and the differential metabolites.

PELNs alleviate colitis in FMT mice

To further validate the mechanism by which PELNs intervene in tight junction injury in IECs through ERS modulation via gut microbiota and their metabolites in UC, a DSS-induced FMT colitis mouse model was established. The role of gut microbiota and their metabolites in the therapeutic effect of PELNs on UC was investigated via FMT experiments. During the pseudo-germ-free mouse establishment and FMT gavage periods, all mice remained in good condition, with normal diet and active behavior. During colitis induction, the FMT-DSS group exhibited progressively worsening symptoms, including lethargy, reduced activity, rough fur, decreased food and water intake, significant weight loss, and largely amorphous, bloody stools. In contrast, mice in the FMT-PELNs group maintained good physiological conditions: Compared to the FMT-DSS group, the FMT-PELNs group exhibited significant improvements in behavior, including increased activity levels, negative fecal occult blood results, and restored body weight. No mortality occurred in any group throughout the experiment. Additionally, the DAI score increased rapidly in the FMT-DSS group; by day 24, the score was significantly higher in the FMT-DSS group compared to the FMT-Control and FMT-PELNs groups (P < 0.05). Colon morphology in the FMT-PELNs group improved markedly, approaching that of the FMT-Control group. Statistical analysis showed that, compared to the FMT-Control group, colon length was significantly shorter in the FMT-DSS group (P < 0.05), whereas the FMT-PELNs group showed a significant increase in colon length compared to the FMT-DSS group (P < 0.05) (Figure 6A-G).

Figure 6 Portulaca oleracea L.-derived exosome-like nanoparticles alleviate colitis in fecal microbiota transplantation mice. A: Schematic diagram of the experimental timeline; B: Body weight changes of mice; C: Disease activity index scores; D: Colon length; E: Fecal occult blood test results on day 24; F: Body weight on day 24; G: Colon length on day 24; H: Hematoxylin & eosin staining (× 50, × 100); I: Ultrastructure of tight junctions and endoplasmic reticulum in intestinal epithelial cells observed by transmission electron microscopy (× 6000, × 8200); J: Representative immunofluorescence images of Bip (red), ZO-1 (green), and DAPI (blue) in mouse colon tissues (× 200); K: Quantitative analysis of the relative fluorescence intensity of Bip and ZO-1 in mouse colon tissues. ^aP < 0.05, ^cP < 0.001, ^dP < 0.0001. NS: Not significant; DSS: Dextran sulfate sodium; FMT: Fecal microbiota transplantation; PELNs: Portulaca oleracea L.-derived exosome-like nanoparticles.

PELNs ameliorate ERS and inflammatory status in FMT colitis mice via gut microbiota and their metabolites, thereby protecting IECs

H&E staining revealed intact intestinal epithelial architecture with regular cell morphology and well-preserved villi in the FMT-Control group. In contrast, the FMT-DSS group displayed significant pathological damage, including disrupted colonic epithelial crypts, extensive inflammatory cell infiltration, markedly reduced goblet cells, crypt abscess formation, and epithelial shedding, erosion, and ulceration. In the FMT-PELNs group, the mucosal-glandular structure was intact and regular, with significantly reduced inflammatory infiltration, preserved crypts, and no ulcer formation, showing marked improvement compared to the FMT-DSS group (Figure 6H). TEM analysis revealed that the FMT-Control group had IECs with a well-organized structure, neatly arranged microvilli, intact intercellular tight junctions, abundant mitochondria, normal ER structure without dilation, and regular nuclear morphology. In contrast, the FMT-DSS group exhibited disordered IEC structure, loss of microvilli, disrupted and absent tight junctions, swollen mitochondria with vacuolar degeneration, dilated ER, and irregular nuclear morphology. The FMT-PELNs group, however, showed uniformly arranged microvilli, intact tight junctions, mildly dilated ER, normal mitochondrial morphology, and regular nuclear structure (Figure 6I).

Confocal laser scanning microscopy of colon tissues from FMT mice showed that relative to the FMT-Control group, the FMT-DSS group had lower ZO-1 and higher Bip fluorescence intensities (P < 0.05) conversely, the FMT-PELNs group showed increased ZO-1 and decreased Bip intensities vs the FMT-DSS group (P < 0.05; Figure 6J and K).

In terms of inflammatory status, the FMT-PELNs group exhibited significantly lower levels of pro-inflammatory cytokines IL-28, IL-36, IL-18, and MPO (P < 0.05), and significantly higher levels of anti-inflammatory cytokines IL-37 and IL-10 (P < 0.05) compared to the FMT-DSS group (Figure 7A).

Figure 7 Portulaca oleracea L.-derived exosome-like nanoparticles regulated inflammatory cytokine, reducing endoplasmic reticulum stress and restoring tight junctions in fecal microbiota transplantation mice. A: The expression of interleukin (IL)-28, IL-18, IL-36, myeloperoxidase, IL-37, and IL-10 in colon tissues; B: Immunoblot bands of ZO-1, PERK, phosphorylated PERK, Bip, Occludin, ATF4, eIF2α, phosphorylated eIF2α, Claudin-4, and Claudin-2 in colon tissues; C: Statistical graphs of relative protein expression levels; D: QPCR analysis of the relative gene expression levels of ZO-1, Occludin, ATF4, Claudin-4, Bip, eIF2α, Claudin-2 and PERK in colon tissues. ^aP < 0.05, ^bP < 0.01, ^cP < 0.001, ^dP < 0.0001. DSS: Dextran sulfate sodium; FMT: Fecal microbiota transplantation; PELNs: Portulaca oleracea L.-derived exosome-like nanoparticles; IL: Interleukin; p-PERK: Phosphorylated PERK; p-eIF2α: Phosphorylated eIF2α.

Protein and mRNA expression levels related to ERS and IEC tight junctions were evaluated in the FMT mice. The FMT-DSS group exhibited significant upregulation of ERS-related markers and downregulation of tight junction-related markers. Specifically, the expression of Bip, p-PERK, p-eIF2α, Claudin-2, and ATF4 was significantly higher at both the protein and mRNA levels compared to the FMT-Control group (P < 0.05), while the expression of ZO-1, Occludin, and Claudin-4 was significantly lower (P < 0.05). In contrast, in the FMT-PELNs group, which received PELNs intervention, the expression of the aforementioned ERS markers was significantly downregulated after DSS treatment (P < 0.05). The expression of ZO-1, Occludin, and Claudin-4 proteins and mRNA increased, while the expression of Claudin-2 was decreased. These results indicate that PELNs can alleviate ERS, reduce inflammation, and restore tight junction integrity in IECs by modulating gut microbiota and their metabolites in colitis mice (Figure 7B-D).

16S rRNA sequencing and untargeted metabolomics were conducted on FMT mice. The results revealed that the alpha diversity indices (Ace, Shannon) of the gut microbiota in the FMT-DSS group increased, whereas the FMT-PELNs group effectively counteracted this change (Figure 8A and B). This suggests that, after receiving fecal transplants from DSS-induced colitis mice, the recipient mice exhibited increased bacterial richness and evenness (i.e., relative density). The significant differences were found in the beta diversity of the gut microbiota among the three FMT mouse groups. The FMT-PELNs group effectively altered the overall structure of the gut microbiota impacted by DSS exposure, further supporting its potential to regulate gut microbial balance (Figure 8C and D). In the microbiota characterization results, the gut health index of the FMT-DSS group significantly decreased, while the microbiota dysbiosis index significantly increased. In contrast, the gut microbiota dysbiosis in the FMT-PELNs group shifted back towards a healthier state (Figure 8E), indicating that the high alpha diversity in the FMT-DSS mice was associated with dysbiosis. A heatmap based on the relative abundance of the top 20 most abundant bacterial genera showed that the FMT-Control and FMT-PELNs groups clustered together, suggesting similar changes in the abundance of key bacterial genera between these groups (Figure 8F). At the phylum level, the FMT-DSS group exhibited increased abundances of Firmicutes and Campilobacterota, alongside a decreased abundance of Actinobacteriota. The FMT-PELNs group reversed these changes, consistent with earlier experimental findings. At the genus level, PELN treatment reduced the levels of Alloprevotella, Lachnospiraceae_NK4A136_group, Helicobacter, Colidextribacter, and Prevotellaceae_UCG-001 in colitis mice, while increasing the levels of Dubosiella, Bifidobacterium, Lactobacillus, Parabacteroides, Lachnoclostridium, and Faecalibaculum. Moreover, Alloprevotella and Lachnospiraceae_NK4A136_group were significantly different biomarkers in the FMT-DSS group, while Bacteroides, Tannerellaceae, Parabacteroides, Proteobacteria, Gammaproteobacteria, and Rikenellaceae were significantly different biomarkers in the FMT-PELNs group (Figure 8G-I).

Figure 8 Portulaca oleracea L.-derived exosome-like nanoparticles regulate the gut microbiota in dextran sulfate sodium-induced fecal microbiota transplantation colitis mice. A: Alpha diversity of the gut microbiota in fecal microbiota transplantation (FMT) mice assessed by the Ace index; B: Alpha diversity of the gut microbiota in FMT mice assessed by the Shannon index; C: Principal coordinate analysis; D: Non-metric Multidimensional Scaling; E: Differential analysis of the Gut Microbiome Health Index (upper images), analysis of the microbiota dysbiosis index (lower images); F: Heatmap of community composition at the genus level; G: Bar plot showing community composition at the phylum and genus levels for each group; H: Comparison of community differences at the phylum and genus levels; I: Linear discriminant analysis effect size for discriminating multi-level species differences. ^aP < 0.05, ^bP < 0.01, ^cP < 0.001 (A, B, and G: Statistical method: Kruskal-Wallis test with FDR correction; post-hoc test: Dunn's test. E: Statistical method: Wilcoxon test with multiple correction). OTU: Perational taxonomic unit; PCoA: Principal coordinate analysis; DSS: Dextran sulfate sodium; FMT: Fecal microbiota transplantation; PELNs: Portulaca oleracea L.-derived exosome-like nanoparticles; NMDS: Non-metric Multidimensional Scaling; LEfSe: Linear discriminant analysis effect size.

Untargeted metabolomics results showed that PCA and PLS-DA revealed clear separation between the FMT-DSS, FMT-PELNs, and FMT-Control groups, with the FMT-PELNs group clustering closer to the FMT-Control group. This indicates that the gut metabolites in the FMT-PELNs group resemble those in the FMT-Control group, suggesting that PELNs can restore the altered metabolites in FMT colitis mice. These findings also imply that the microbiota and gut metabolites modulated by PELNs are beneficial for colitis recovery (Figure 9A and left image in Figure 9B). PLS-DA model validation showed that all R² values exceeded Q², and the intercept of the Q² regression line with the Y-axis was -0.2459, indicating good model fit, strong predictability, and suitability for subsequent analysis (right image in Figure 9B). Metabolites with VIP > 1 in OPLS-DA, P < 0.05 (P values are nominal) in Student’s t-test, and FC > 1 were identified as differential metabolites (Figure 9C). A total of 1086 differential metabolites were identified in the FMT-DSS vs FMT-Control comparison, including 252 up-regulated and 834 down-regulated metabolites. In the FMT-PELNs vs FMT-DSS comparison, 1093 differential metabolites were identified, with 688 up-regulated and 405 down-regulated. The FMT-PELNs vs FMT-Control comparison yielded 617 differential metabolites, with 162 up-regulated and 455 down-regulated. KEGG pathway enrichment analysis (showing only the top 20 results by P value) revealed that, compared to the FMT-Control group, the FMT-DSS group was most significantly enriched in pathways such as Pantothenate and Coenzyme A (CoA) biosynthesis, Lysine degradation, and Nucleotide metabolism (P < 0.05). Specifically, compared to the FMT-Control group, the FMT-DSS group down-regulated the expression levels of metabolites such as inosine, thymine, L-lysine, L-glutamine, 5-aminovaleric acid, and other amino acids, nucleotides, and organic acid derivatives. In comparison, the FMT-PELNs group showed significant enrichment in pathways such as Choline metabolism in cancer, Linoleic acid metabolism, and Lysine degradation (P < 0.05). Notably, compared to the FMT-DSS group, PELN treatment up-regulated metabolites including L-lysine, 5-aminovaleric acid, and N6-Acetyl-L-lysine, while down-regulating 13-L-Hydroperoxylinoleic acid (Figure 9D).

Figure 9 Portulaca oleracea L.-derived exosome-like nanoparticles regulate gut metabolites in dextran sulfate sodium-induced fecal microbiota transplantation colitis mice. A: Principal component analysis; B: Partial least squares-discriminant analysis; C: Volcano plot of differential metabolites; D: Bubble chart of KEGG pathway enrichment analysis for differential metabolite sets in the gut of fecal microbiota transplantation mice; E: Correlation analysis between gut microbiota and ulcerative colitis (UC)-related indicators; F: Correlation analysis between metabolites and UC-related indicators; G: Correlation analysis between gut microbiota and metabolites. The X-axis represents the enrichment ratio of differential metabolites, and the Y-axis corresponds to KEGG pathway categories. The bubble diameter is proportional to the number of differential metabolites in the pathway, and the bubble color reflects the significance level of enrichment (P value). ^aP < 0.05, ^bP < 0.01, ^cP < 0.001. DSS: Dextran sulfate sodium; FMT: Fecal microbiota transplantation; PELNs: Portulaca oleracea L.-derived exosome-like nanoparticles; PCA: Principal component analysis; PLS-DA: Partial least squares-discriminant analysis.

A total of 48 differential metabolites (Supplementary Table 3) were identified, meeting the criteria of VIP > 2 in OPLS-DA, P < 0.05 in Student’s t-test, and FC > 2. Spearman correlation analysis was conducted to generate heatmaps (Figure 9E-G). The results revealed significant correlations between differential microbiota, metabolites, and UC-related indicators. Notably, Lachnospiraceae_NK4A136_group, Helicobacter, Bifidobacterium, and Lachnoclostridium exhibited strong correlations with all UC-related indicators shown in the figure. Similarly, differential metabolites demonstrated strong correlations with both UC-related indicators and differential microbiota after FMT.

DISCUSSION

ERS plays a pivotal role in cellular responses to protein misfolding within the ER and is central to the pathogenesis of UC. Prolonged or excessive ERS leads to IECs apoptosis and compromised barrier function, further exacerbating inflammation[13]. The PERK-eIF2α-ATF4 pathway is likely involved in UC pathogenesis, and restoring ER homeostasis in IECs through this pathway could offer a novel therapeutic approach[14]. Alterations in the gut microbial community represent an integral link to the pathophysiology of UC, a connection that is further evidenced by corresponding disturbances in the host’s metabolome[15]. Tight junctions between IECs act as a critical barrier against the invasion of luminal antigens into the intestinal mucosa and are used to evaluate the integrity of the mechanical barrier[16]. Consistent with observations in human UC, mice subjected to DSS-induced colitis recapitulate the critical pathophysiological changes of impaired tight junction integrity and a consequent rise in gut permeability[17]. In conclusion, maintaining the ecological balance of the gut microbiota and ER homeostasis to protect the intestinal epithelial barrier offers a promising strategy for treating UC.

Gut microbiota dysbiosis has been widely documented in patients with UC and animal models. In this study, the DSS group exhibited increased abundances of Firmicutes, Proteobacteria, and Verrucomicrobiota, along with a decreased abundance of Bacteroidota. The proliferation of Proteobacteria in the gut is a hallmark of dysbiosis, indicating instability in the microbial community structure, which can trigger intestinal inflammation. Previous research indicates that an individual’s susceptibility to colitis escalates alongside a rising relative abundance of Proteobacteria within the intestinal microbiome[18]. The increase in Verrucomicrobiota may reflect alterations in the overall structure of the gut microbiota, which could be linked to disruptions in microbial diversity and functionality[19]. The abundance of Bacteroidota is closely associated with the immune response in colitis and low-grade systemic inflammation[20]. Bacteroidota can interact with regulatory T cells to promote the production of the anti-inflammatory cytokine IL-10[21]. Additionally, the Firmicutes/Bacteroidota (F/B) ratio is a well-established indicator of gut homeostasis[22]. In the present study, the DSS group showed an elevated F/B ratio, signaling dysbiosis. Treatment with PELNs exerted a suppressive effect on the F/B ratio within the colonic microbiota of DSS-treated mice, highlighting the regulatory role of PELNs in maintaining intestinal balance. Similarly, the FMT-DSS group exhibited a significant increase in the pro-inflammatory bacterium Firmicutes[23], a higher F/B ratio, an increase in Campilobacterota, and a decrease in Actinobacteriota. Another study also observed a significant reduction in gut microbiota diversity and an increased abundance of Campilobacterota in a DSS-induced colitis mouse model[24]. The increase in Campilobacterota may correlate with intestinal barrier dysfunction and aggravated inflammation. The decrease in Actinobacteriota suggests a diminished capacity for intestinal barrier repair, as certain species within Actinobacteriota, such as Bifidobacterium, play a pivotal role in maintaining the barrier by promoting mucin synthesis and secretion and enhancing tight junctions in IECs[25].

In the DSS group mice, the abundances of Bacteroides and Turicibacter were increased, while Muribaculaceae and Prevotellaceae_UCG-001 were decreased, consistent with findings from other studies[26]. Research has shown that Muribaculaceae, previously known as the S24-7 family, is reduced in colitis models, with its levels rising as colitis improves[27]. Bacteroides, considered an opportunistic pathogen in UC, can cause intestinal barrier damage and inflammation in the UC gut environment[28]. Similarly, certain Enterococcus strains may trigger opportunistic infections. Emerging evidence suggests that Dubosiella and Prevotellaceae_UCG-001 produce SCFAs, which play a critical role in regulating gut homeostasis[29]. Rikenellaceae_RC9_gut_group is involved in carbohydrate fermentation and SCFA production. Erysipelotrichaceae is closely linked to lipid metabolism and energy balance, potentially influencing host body weight and metabolic health. Although the taxonomic classification of o__RF39 remains unclear, it likely participates in intestinal metabolic activities such as carbohydrate fermentation and SCFA production. The increased abundance of Lactococcus suggests elevated lactate production in DSS mice, which lowers gut pH, inhibits pathogen growth, and may enhance intestinal barrier function and immune support by modulating gut microbiota balance. Therefore, the expansion of these bacterial taxa after PELN treatment may represent a key mechanism underlying the regulatory effects of PELNs on the gut microbial community, likely by promoting the growth of SCFA-producing bacteria, thereby improving the intestinal barrier and maintaining gut homeostasis.

The abnormal proliferation of Lachnospiraceae_NK4A136_group may contribute significantly to the microbiota dysbiosis observed in DSS-induced colitis[30]. In this study’s FMT experiment, the abnormal proliferation of Alloprevotella and Lachnospiraceae_NK4A136_group in the FMT-DSS group mice aligns with these findings. Certain Helicobacter species (e.g., Helicobacter hepaticus) have been shown to possess pro-inflammatory properties, potentially exacerbating intestinal inflammation by activating immune responses. In this study, Prevotellaceae_UCG-001 exhibited a dual role, with its abundance reduced in the DSS group as noted earlier. The presence of Colidextribacter may also indicate its association with disease activity or gut microbiota dysbiosis. In contrast, FMT-PELN treatment ameliorated the trends observed in the FMT-DSS group, increasing the levels of Dubosiella, Bifidobacterium, Lactobacillus, Parabacteroides, Lachnoclostridium, and Faecalibaculum. These genera are involved in carbohydrate fermentation and SCFA production, playing vital roles in gut health and host metabolism[31]. Bifidobacterium and Lactobacillus can produce indole derivatives through tryptophan metabolism in the gut microbiota, modulating related responses and promoting epithelial barrier repair[32]. After DSS exposure, the FMT-PELNs group showed significant enrichment of Bacteroides, Tannerellaceae, Parabacteroides, Rikenellaceae, Proteobacteria, and Gammaproteobacteria, among which SCFA-producing bacteria were more abundant. These findings suggest that PELNs effectively alleviate the DSS-induced gut microbiota imbalance. Additionally, mice that received FMT from PELNs-treated donors also exhibited spontaneous alleviation of gut microbiota alterations upon subsequent DSS exposure, further indicating the regulatory role of PELNs in modulating the murine gut microbiota.

Untargeted metabolomics results also revealed that PELNs regulate gut microbiota-derived metabolites. Significant alterations in the fecal metabolic profiles were observed in mice following DSS exposure and subsequent PELNs or FMT-PELNs intervention. PELNs notably ameliorated the abnormal metabolic changes in UC mice. The findings showed a pronounced reduction in the concentration of various metabolites across several major categories: Lipids, amino acid and nucleotide derivatives, and organic acids, while PELNs upregulated the expression of metabolites such as amino acids and lipids. These metabolites were enriched in several metabolic pathways, with significant enrichment in: Pantothenate and CoA biosynthesis, Lysine degradation, Nucleotide metabolism, Choline metabolism in cancer, Linoleic acid metabolism, Cutin, suberine, and wax biosynthesis, alpha-linolenic acid metabolism, Axon regeneration, and Shigellosis pathways. Evidence has established that CoA plays a critical role in fatty acid metabolism, the tricarboxylic acid cycle, and antioxidant responses. CoA deficiency can impair cellular energy metabolism and inflammatory responses[33]. Lysine degradation produces acetyl-CoA and glutamate, which are involved in energy metabolism and immunomodulation, with abnormal lysine metabolism being linked to inflammatory diseases and cancer. Dysregulated nucleotide metabolism impairs DNA repair and apoptosis and is closely associated with inflammation and cancer[34]. Metabolites from linoleic acid metabolism, such as arachidonic acid, generate pro-inflammatory mediators (e.g., prostaglandins and leukotrienes), contributing to inflammation. In contrast, metabolites from alpha-linolenic acid metabolism, such as eicosapentaenoic acid (EPA) and docosahexaenoic acid (DHA), are known to possess the capacity to mitigate inflammation and inhibit the production of pro-inflammatory mediators[35]. Similarly, abnormalities in lipid metabolism are associated with skin barrier dysfunction and inflammation, as seen in the pathways of cutin, suberine, and wax biosynthesis.

Among the identified pathways, annotated metabolites included alpha-linolenic acid, jasmonic acid, gamma-linolenic acid, L-isoleucine, DG (15:0/18:1(9Z)/0:0), arachidonoyl ethanolamide, inosine, thymine, L-lysine, L-glutamine, 5-aminovaleric acid, and N6-Acetyl-L-lysine, among others. These metabolites primarily contribute to the modulation of inflammation, immune response, energy metabolism, and intestinal barrier function in UC. L-glutamine, for instance, functions as the principal metabolic fuel for IECs and fulfills critical functions, including the restoration of the mucosal layer and regulation of immune activity[36]. Inosine, a purine metabolite, exerts anti-inflammatory and immunomodulatory effects[37]. alpha-linolenic acid, an omega-3 polyunsaturated fatty acid, possesses well-established anti-inflammatory properties. It is metabolized into EPA and DHA, which inhibit the production of pro-inflammatory mediators such as prostaglandins and leukotrienes. Omega-3 fatty acid deficiency is commonly observed in patients with UC[38]. Jasmonic acid, a plant hormone, also exhibits anti-inflammatory and antioxidant activities. DG (15:0/18:1(9Z)/0:0), an intermediate in glycerophospholipid metabolism, is involved in signal transduction and energy storage. These findings collectively suggest that PELNs alleviate colitis by modulating key metabolites. The underlying mechanism of PELNs’ anti-UC activity likely involves reshaping the gut microbiota and restoring balance to metabolic products.

Intestinal cytokines play a pivotal role in regulating mucosal inflammatory responses in UC, influencing immune responses following mucosal injury and determining whether homeostatic or pro-inflammatory pathways prevail. This study assessed the anti-inflammatory effects of PELNs by measuring various cytokines[39]. In this study, recipient mice transplanted with feces from PELNs-treated mice (FMT-PELNs group) spontaneously alleviated mucosal inflammation, edema, and damage in the FMT colitis mice, restoring their crypt structure. Based on these findings, we propose that the therapeutic action of PELNs against UC is orchestrated through a synergistic process involving barrier repair, microbial and metabolite modulation, inflammatory cascade inhibition, and immunologic balancing.

Microbes, their products, and their interactions can influence ERS in the gut, while ERS can also shape the gut microbiota composition[4]. Bacterial toxins such as cholera toxin, Shiga toxin, VacA, and listeriolysin O have been shown to induce ERS and UPR signaling[40]. For instance, the AB5 toxin from Shiga toxin-producing Escherichia coli cleaves the ER chaperone Bip, triggering the activation of three UPR signaling pathways—IRE1, PERK, and ATF6—which leads to transient ERS and cell cycle arrest[41]. Moreover, the intestinal epithelium’s ability to secrete mucins and antimicrobial proteins depends on maintaining ER protein processing homeostasis[42]. Bifidobacterium can enhance mucin production, alleviating ERS in goblet cells and supporting high rates of mucin secretion[43]. Therefore, the mechanisms by which PELNs alleviate UC are complex, with changes in the gut microbiota potentially serving as secondary outcomes or concomitant phenomena linked to the reduction of ERS or their effects on IECs. Consequently, this study explored the correlations within the data. In the present study, the differential microbiota and metabolites following PELNs intervention were strongly correlated with ERS, intestinal epithelial tight junctions, and related cytokines, suggesting that PELNs effectively alleviate excessive ERS and restore tight junctions in the intestinal epithelium, thereby improving the intestinal barrier in DSS-induced colitis mice by regulating the PERK-eIF2α-ATF4 signaling pathway. Notably, gut microbiota and their metabolites play a key role in restoring UC intestinal epithelial tight junction function and structure through PELNs (Figure 10).

Figure 10  Schematic illustration: The mechanism of action for orally delivered Portulaca oleracea L.-derived exosome-like nanoparticle in the mitigation of ulcerative colitis. Portulaca oleracea L.-derived exosome-like nanoparticles exert therapeutic benefits through the modulation of the intestinal microbial community and its associated metabolites, alleviating endoplasmic reticulum stress in colitis mice, reducing inflammation, and restoring tight junction integrity in intestinal epithelial cells. ERS: Endoplasmic reticulum stress; PELNs: Portulaca oleracea L.-derived exosome-like nanoparticles.

Limitations

Several limitations of this study should be acknowledged. PELNs constitute a complex mixture, and the current experiments cannot pinpoint which specific bioactive components are responsible for the observed effects. The study was based solely on preclinical DSS and FMT models, with no human data included. The microbiota and metabolomics analyses are correlative rather than causal. Direct manipulation of the ERS pathway was not carried out, and the evaluation of systemic safety and toxicity was limited.

CONCLUSION

In conclusion, PELNs exert therapeutic effects against UC by modulating gut microbiota and metabolites, reducing inflammatory cytokine expression, regulating ERS via the PERK/eIF2α/ATF4 pathway to prevent IEC tight junction injury, and ultimately restoring intestinal mucosal barrier function.