Mature adipocytes were isolated as previously described, with minor modifications (23, 24). Hypertrophic MAT (HtMAT) from the diseased bowel and macroscopically normal MAT (nMAT) were intraoperatively obtained from 15 patients with CD (Jinling Hospital, 2023–2024; patient details shown in Supplementary Tables 1, 2. The hypertrophic MAT samples and the control adipose tissue samples were obtained from the same patient during surgery. The collected tissues were immediately transported on ice and processed within 20 min. After removal of blood using Hank’s balanced salt solution buffer, vascular components were excised, followed by mincing of tissues prior to enzymatic digestion (3% bovine serum albumin [BSA], 8 mg/mL collagenase D, 2.4 U/mL Dispase II; 37 °C, 1 h). Finally, digested samples were filtered (200 μm), and adipocytes were collected through centrifugation (300 × g, 5 min, 4 °C).

All cell lines were maintained at 37 °C using 5% CO₂ with corresponding medium. Tohoku Hospital Pediatrics-1 (THP-1, CVCL_0006), Immortalized bone marrow-derived macrophages (iBMDMs) and 3T3-L1 cells (BFN608006390) were obtained from the National Collection of Authenticated Cell Cultures (Shanghai, China). 3T3-L1 pre-adipocytes were maintained in Dulbecco’s modified Eagle medium (DMEM) supplemented with 10% fetal bovine serum (FBS) (Gibco, Gaithersburg, MD, USA) and penicillin/streptomycin. At 2 days post confluence (designated as Day 0), a differentiation cocktail (10μg/mL insulin, dexamethasone, and 0.5 mM 3-isobutyl-1-methyl-xanthine) was added to the cells in fresh media. After 48 h, the media was changed to DMEM supplemented with 10% FBS containing 10μg/mL insulin, which was replaced every alternate day. Eight days after induction (designated Day 8), cells were harvested for subsequent analyses.

Exosomes were isolated and purified as previously described (19). For exosome isolation, adipose tissue samples were precisely weighed to ensure equal starting mass (100 mg wet weight per sample) before homogenization. Adipocytes were cultured in exosome-free medium for 48 h. Next, the supernatant was collected and sequentially centrifuged at 4 °C (300 × g, 10 min; 3,000 × g, 20 min) to remove cells and debris. Exosomes were isolated via ultracentrifugation (10,000 × g, 30 min; 100,000 × g, 70 min), followed by an additional 100,000 × g wash (70 min). Purified exosomes were stored at -80 °C. Exosome morphology and particle size distribution were assessed using transmission electron microscopy (TEM; FEI Tecnai Spirit, 120 kV) and nanoparticle tracking analysis (NTA; Malvern LM10), respectively. Protein content was quantified via the bicinchoninic acid assay (Thermo Fisher Scientific) after exosome lysis in radioimmunoprecipitation assay (RIPA) buffer. Exosome identity was confirmed using western blotting for surface markers (TSG101, CD9, and CD63).

Total RNA was isolated from cells and exosomes using TRIzol Reagent (Invitrogen), with its concentration and purity being assessed using Nanodrop 2000 (Thermo Fisher Scientific). cDNA was synthesized using SuperScript III Reverse Transcriptase (Invitrogen) for mRNA or Bulge-Loop microRNA (miRNA) qRT-PCR Starter Kit (RiboBio) for miRNAs. Moreover, qPCR was performed on an ABI 7300 system (Applied Biosystems) using SYBR Premix Ex Taq (Takara Bio, Shiga, Japan). Supplementary Table 3 presents the primer sequences for the target genes (TNF-α, IL-1β, IL-6, iNOS, CD86, CD206, Arg1) and controls (β-actin). Gene expression was normalized to β-actin (mRNA) and each sample was run in triplicate, and all reactions were repeated three times to ensure reproducibility of the results.

Total protein was extracted from cells and exosomes using RIPA lysis buffer (KeyGen Biotechnology), followed by assessment of protein content using the bicinchoninic acid assay (Thermo Fisher Scientific). Proteins (10–30 μg per sample) were separated using 10% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (Wanlei) and transferred to polyvinylidene difluoride membranes (Millipore). After blocking with 5% non-fat milk (1 h, room temperature), the membranes were incubated with primary antibodies overnight at 4 °C, followed by species-matched secondary antibodies conjugated with horseradish peroxidase (1 h, room temperature). Protein bands were visualized using ECL reagent (Zen-Bio) and imaged using a Tanon 5200 chemiluminescence system (Tanon Science & Technology).

iBMDMs were incubated with 30 μg/ml exosomes for 48 h. For immunophenotyping, the cells were harvested and blocked with 3% BSA in PBS for 30 min, followed by incubation with phycoerythrin-conjugated CD11b and allophycocyanin-conjugated CD86 (BioLegend; San Diego, CA), with appropriate isotype controls in the dark cells. In flow cytometry analysis, gating was sequentially applied as follows: cells were first selected based on forward- and side-scatter properties (FSC for size, SSC for granularity) to exclude debris; then, doublets were excluded using FSC-A versus FSC-H; finally, live cells were identified by DAPI negativity. For surface marker analysis (CD11b, CD86), fluorescence thresholds were set using fluorescence-minus-one (FMO) controls. Samples were analyzed using a FACSCalibur flow cytometer (BD Biosciences), and data were processed using FlowJo software (v10.0).

The 3’-UTR of suppressor of tripartite motif-containing 33 (TRIM33) (wild-type [WT] or mutant [MUT] containing the miR-26b-3p binding site) was synthesized using GenScript and cloned into the pGL3-REPORT vector (FseI/XbaI sites). Macrophages were transfected with miR-mimic or miR-NC, plated in 96-well plates, and subsequently co-transfected with either the pGL3-TRIM33-WT or pGL3-TRIM33-MUT construct. Firefly and Renilla luciferase activities were measured at 48 h after transfection using the Dual-Luciferase Reporter Assay System (Promega), with Renilla luciferase as the internal control.

Total RNA was pooled from three pairs of Crohn-exos and Control-exos and subjected to miRNA sequencing at Lianchuan Biotech. Sequencing libraries of small RNAs (length: 18–30 nucleotides) were generated following the manufacturer’s instructions. Library sequencing was performed using the Illumina HiSeq 2500 platform, followed by annotation of small RNAs using miRBase v22. Differential expression analyses were performed based on the read numbers and pairwise comparisons. Differentially expressed miRNAs were identified using fold-change and p-value filtering. Total RNA was isolated using TRIzol reagent, followed by quantification and quality assessment using NanoDrop 2000 (Thermo Fisher). RNA integrity and gDNA contamination tests were performed using denaturing agarose gel electrophoresis. Before RNA-seq library construction, ribosomal RNA was removed using the RiboMinus Eukaryote Kit (Qiagen). The sequencing library was established using an Agilent 2100 Bioanalyzer with an Agilent DNA 1000 Chip Kit (Agilent, CA, USA), with adjustment to 10 nM prior to cluster generation. cDNA was sequenced using a HiSeq 2000 system (Illumina, San Diego, CA, USA) and a 100-bp paired-end run.

Exosomes were fluorescently labeled using DiI (Sigma-Aldrich) or DiR (Invitrogen) following the manufacturer’s guidelines. Briefly, exosomes were incubated with the dye (2 mg/mL, 10 min), followed by ultracentrifugation (100,000 × g, 1 h) to remove unbound dye. Next, macrophages were treated with Dil-labeled exosomes (10 μg/mL, 12 h) and washed with phosphate buffered saline to remove non-internalized exosomes, followed by counterstaining of nuclei using DAPI (Thermo Fisher). Internalization was visualized using a confocal microscope (Zeiss LSM880). Subsequently, intraperitoneal (IP) injections of DiR-labeled exosomes (200 μg) were administered to mice. After 12 h, the fluorescence distribution in anastomotic tissues was assessed using an IVIS Lumina II imaging system (PerkinElmer).

Anastomosis and colon samples were fixed in 4% formaldehyde for 48 h. Next, they were gradually dehydrated, embedded in paraffin, cut into 4-μm sections, and stained with hematoxylin and eosin. Stained sections were observed through light microscopy, followed by calculation of the severity of histological inflammation as previously reported. Five aspects were considered: mucosal architecture, infiltration of mononuclear cells and neutrophils, epithelial defects, and goblet cell loss (Supplementary Table 4).

The miR-26b-3p mimic (miR-mimic) and negative control (miR-NC) were obtained from RiboBio and transfected into cells at 50% confluence using Lipofectamine 2000 (Invitrogen) following the manufacturer’s instructions. Next, miR-26b-3p inhibitor was added at a final concentration of 100 nM. For mitogen-activated protein kinase (MAPK) knockdown, we used a lentiviral shRNA construct (GenePharma), with a scrambled lentiviral vector serving as a negative control. Fibroblasts at approximately 50% confluency were transduced at a multiplicity of infection of 80. Transfection and transduction efficiencies were verified using quantitative reverse transcription PCR (qRT-PCR).

Cy3-labeled miR-26b-3p mimics (GenePharma) were transfected into 3T3-L1 mature adipocytes following the manufacturer’s instructions. Subsequently, exosomes were isolated from the conditioned medium and co-cultured with recipient fibroblasts. After incubation, fibroblasts were fixed with 4% paraformaldehyde, permeabilized with 0.05% Triton X-100, and counterstained with DAPI. To visualize miR-26b-3p transfer, fluorescent images were captured using a Zeiss LSM880 confocal microscope.

We prospectively enrolled 13 consecutive patients with CD who underwent bowel resection for ileocolonic anastomotic strictures and 10 patients with colon cancer who underwent right hemicolectomy at Jinling Hospital in 2024. Supplementary Table 5 shows the clinical characteristics of the included patients. We collected intestinal tissue from diseased segments or resection margins as well as the corresponding MAT from each patient. All specimens were immediately processed by dividing them into two aliquots: one snap-frozen in liquid nitrogen (-80 °C storage) and the other fixed in 10% phosphate-buffered formalin for paraffin embedding. For fibrosis assessment, Masson’s trichrome-stained sections were digitally scanned using a Nanozoomer system (Hamamatsu). The study protocol was approved by the Jinling Hospital Ethics Committee, and all participants provided written informed consent.

TRIM33 and empty vector plasmids were purchased from GenePharma (Shanghai, China). Macrophages were seeded in six-well plates (2×105 cells/well) and incubated for 24 h to reach 60–70% confluence. Plasmid transient transfection was performed using Lipofectamine™ LTX (15338100, Thermo) following the manufacturer’s instructions. Briefly, 2.5 μg of plasmid DNA and 6.25 μL of transfection reagent were mixed and added to the cell medium without antibiotics. After 24-h incubation, cell medium was replaced with fresh complete culture medium to maintain cell growth for another 24 h before harvesting.

Target genes of the miRNAs were predicted using the following bioinformatics databases: TargetScan [http://www.targetscan.org ], miRDB [http://mirdb.org/], mirDIP [http://ophid.utoronto.ca/mirDIP], DIANA [https://diana.e-ce.uth.gr], and PicTar [https://pictar.mdc-berlin.de]. Additionally, we analyzed genes from three databases.

For endoscopic evaluation, mice were anesthetized using 2–3% isoflurane/O², followed by extensive fecal removal using flexible feeding tubes. An Olympus URF type V endoscope was rectally inserted to a depth of ≤5 cm; moreover, videos of the endoscopic procedure were recorded using a Medicap USB200 Medical Digital Video Recorder while retracting the endoscope. Data analyses were performed as previously described (25).

Male C57BL/6 mice, aged 8 weeks, were obtained from the Model Animal Research Center of Nanjing University (Nanjing, China), for establishing a intestinal inflammation model. The experiment divided the mice into two groups: a model group (mHt-exos) and a control group (mN-exos), each comprising five mice. intestinal inflammation was induced using an established DNBS rectal administration protocol over six weeks (26), the mice were euthanized, and their Mesenteric Adipose Tissues were harvested.

We obtained 8–10 weeks-old WT and IL-10-deficient (IL-10^-/-) male mice from the Model Animal Research Center of Nanjing University (Nanjing, China). They underwent ileocolonic resection as previously described (27, 28). Briefly, mice were placed in a clean cage without any solid food for 24 h before the procedure. The animal was induced anesthesia by transferring the animal to the anesthesia induction chamber (3–4% isoflurane and oxygen flow of 1–2 liters/min), once the animal loses consciousness, removed the animal from the induction chamber and depilate the abdomen from sternum, placed the animal on the heating pad in dorsal recumbency with the tail toward the surgeon while the nozzle of the anesthesia system over the animal’s snout, and set it to 2–3% isoflurane and oxygen flow of 1–2 liters/min. The skin was disinfected by scrubbing with an aqueous iodophor solution, followed by 70% (vol/vol) isopropanol, then, covered the entire animal with a sterile surgical drape and created an opening in the drape to expose the surgical field. Subsequently, surgeries were performed under sterile conditions using an operating microscope (7× magnification). After ligating the mesentery, we removed the small intestine proximal to the ileocecal junction, 5 cm from the proximal colon (length, 2 cm). We used 9–0 monofilament sutures to reconstruct the continuity of the intestinal tract using an end-to-end, interrupted, single-layered anastomosis procedure. Immediately after anastomosis, saline in the WT and IL-10^-/- control (IL-10+saline) groups, 200 μg mN-exos in the IL-10^-/- group (IL-10+mN-exos), and 200 μg Ht-exos in the IL-10^-/- group (IL-10+mHt-exos) were directly instilled onto the external anastomotic surface in a single dose, followed by intraperitoneal (IP) injections at 4-day intervals. All mice were euthanized after 28 days. The postoperative survival rate of IL-10^-/-mice was 80%. Additionally, we obtained a 1-cm segment of the small intestine (Anastomosis SI) and colon (Anastomosis C) proximal to the anastomosis for further analysis.

For subsequent experiments, paraffin-embedded samples were cut into 5-μm-thick sections. Deparaffinization, rehydration, and antigen retrieval were sequentially performed. Tissue sections were blocked using 5% BSA and incubated overnight with the following primary antibodies: anti-CD86, anti-CD68, anti-F4/80, and anti-iNOS. Subsequently, they were incubated with Cy5- and FITC-conjugated secondary antibodies for 1 h at room temperature. For immunofluorescence staining, fibroblasts were fixed with 4% paraformaldehyde and permeabilized with 0.1% Triton X-100. After blocking with 5% BSA for 1 h, cells were incubated with the primary antibodies overnight at 4 °C. Next, they were incubated with FITC-conjugated secondary antibodies for 1 h at room temperature. Finally, tissue and cell slides were incubated with 0.1% DAPI solution for 10 min at room temperature. A confocal laser scanning microscope (Zeiss LSM880) was used to capture images.

The Cy3-labeled probe for miR-26b-3p (sequence: 5'-CCUGUUCUCCAUUACUUGGCUC-3') was designed and synthesized by GenePharma. Briefly, tissues were permeabilized in 0.5% Triton X-100 for 30 min after deparaffinization. Subsequently, the slides were incubated in pre-hybridization buffer at 37 °C for 30 min. Next, Cy3-labeled miR-26b-3p probes were added and incubated with the slides in hybridization solution at 37 °C overnight. The next day, nuclei were stained with DAPI, following by examination of the slides using a confocal microscope (Zeiss LSM880).

All statistical analyses were performed using GraphPad Prism software version 7.0 and SPSS version 20.0. Between-group comparisons of normally and non-normally distributed data were performed using Student’s t test and the Mann–Whitney U test, respectively. Among-group comparisons were performed using one-way analysis of variance with Tukey’s multiple comparison test. Correlations among the variables were assessed using Pearson’s correlation test. Statistical significance was set at *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001.

Exosomes were isolated from HtMAT of the diseased bowel and macroscopically normal MAT from the surgical margins of patients with CD using established protocols, termed as Crohn groups and Control groups, respectively. Comparative analyses using TEM imaging showed similar cup-shaped morphology (Figure 1A), while NTA demonstrated comparable size distributions (30–200 nm) between both exosome populations (Figure 1B). Western blot analysis revealed significantly elevated levels of exosomal markers (CD9, CD63, and TSG101) in Ht-exos (Figures 1C, D). Moreover, an increased exosomal protein yield was observed when adipocytes were normalized (Figure 1E), indicating enhanced exosome biogenesis in HtMAT. THP-1s co-cultured with exosomes. Confocal microscopy confirmed effective internalization of Dil-labeled exosomes (30 μg/mL) by THP-1s within 12 h (Figure 1F). Crohn-exos treatment significantly altered the cytokine profile, increasing pro-inflammatory mediators, increasing pro-inflammatory mediators (TNF-α, IL-6, MPO) in culture supernatants (Figure 1G). At the transcriptional level, Crohn-exos upregulated M1-associated genes (CD86, iNOS and TNF-α) and downregulated M2 markers (Arg1 and CD206) (Figure 1H). Consistent with these findings, western blotting showed elevated M1 protein expression (Figure 1I), while flow cytometry confirmed an increase in CD86⁺ M1 macrophages (Figure 1J). Taken together, these findings demonstrate that the Exosomes divided from HtMAT of the diseased bowel in CD preferentially drive macrophage polarization toward a pro-inflammatory M1 phenotype, which mechanistically explains their role in exacerbating inflammation.

Following our insights from the patients, we extended our investigation on a DNBS-induced mouse model of CD. We successfully isolated exosomes from MAT using established protocols, termed as mHt-exos groups and mN-exos groups, respectively. Comparative analyses using TEM imaging showed similar cup-shaped morphology (Figure 2A), while NTA demonstrated comparable size distributions (30–200 nm) between both exosome populations (Figure 2B). Western blot analysis revealed significantly elevated levels of exosomal markers (CD9, CD63, and TSG101) in mHt-exos (Figures 2C, D). An increased exosomal protein yield was observed when adipocytes were normalized (Figure 2E). Confocal microscopy confirmed effective internalization of Dil-labeled exosomes (30 μg/mL) by iBMDMs within 12 h (Figure 2F). mHt-exos treatment significantly altered the cytokine profile, increasing pro-inflammatory mediators (TNF-α, IL-6, MPO) while decreasing anti-inflammatory IL-10 in culture supernatants (Figure 2G). At the transcriptional level, mHt-exos upregulated M1-associated genes (CD86, iNOS and TNF-α) and downregulated M2 markers (Arg1 and CD206) (Figure 2H). Consistent with these findings, western blotting showed elevated M1 protein expression (Figure 2I), while flow cytometry confirmed an increase in CD86⁺ M1 macrophages (Figure 2J). Taken together, these findings demonstrate that mHt-exos preferentially drive macrophage polarization toward a pro-inflammatory M1 phenotype, which was consistent with the results in CD.

Prior to performing animal experiments, we determined the amount of exos injected intraperitoneally by in our DSS-induced colitis model (Supplementary Figure S1A). In the vivo experiments, mice were treated with increasing doses of exosomes: 0 μg [saline], 50, 100, 200, 400 and 800 μg. It was observed that inflammatory alterations began to manifest in the mice following administration of 200 μg: Colon length tended to be shorter, and histological inflammation tended to worsen with an increase in the Ht-Exos dose (Supplementary Figures S1B–E). With an increase in dose, the degree of intestinal inflammation became increasingly serious, as reflected by the histopathological inflammatory score (Supplementary Figures S1F, G).

In order to better simulate the situation of anastomotic recurrence in clinical patients, we used an IL-10^-/- mouse model of ileocecal resection (Figure 3A). DiR-labeled exosomes successfully localized to anastomotic sites within 12 h, which verified whether exosomes are able to act on the anastomotic stoma (Figures 3B, C). In our study, mice were categorized into four groups: WT Group and IL-10+saline Group, IL-10+mN-exos Group, and IL-10+mHt-exos Group. Inflammation assessment was performed at 21 postoperative days. The results revealed that compared with controls, mHt-exos-treated mice exhibited significant weight loss (Figure 3D), anastomosis inflammation (Figure 3E), more severe endoscopic inflammation and higher histopathological inflammatory scores (Figures 3F–I). Molecular analyses showed that Ht-exos promoted M1 macrophage polarization. Specifically, mHt-exos elevated iNOS, TNF-α, and IL-6 expression; reduced Arg1, CD206; and increased iNOS⁺ macrophages (Figures 3J–L). The comprehensive data from our mouse model experiments demonstrate that mHt-exos exacerbates anastomosis inflammation by enhancing M1 macrophage polarization.

To explore the molecular mechanisms, microarray analysis derived from Crohn groups and Control groups was performed. The heat map showed revealed 19 differentially expressed miRNAs (11 upregulated and 8 downregulated; p<0.05, FC>1.5) (Figure 4A). We particularly focused on miR-26b-3p owing to its known pro-inflammatory roles (29, 30). At the same time, we found miR-26b-3p had a higher level in mHt-exos compared with mN-exos (Figure 4B). In order to see if transfection of miR-26b-3p would have any effect on adipocytes, the following experiments were performed. Through fluorescent tracing experiments in which 3T3-L1 adipocytes were transfected with Cy3-labeled miR-26b-3p mimics (Figures 4C, D), we found that exosome production was not affected (Figure 4E) and miR-26b-3p successful transfer to recipient iBMDMs, while miR-26b-3p enrichment in donor adipocytes, secreted exosomes, and recipient macrophages (Figure 4F). These findings demonstrate a functional exosomal shuttle mechanism for adipocyte-macrophage communication that may drive inflammation activation.

To further validate the biochemical impact of these findings, the following experiments were performed. In vitro, iBMDMs co-cultured with exosomes from miR-26b-3p-overexpressing adipocytes (miR-mimic-exos) exhibited significantly elevated pro-inflammatory cytokines (TNF-α, IL-1β, IL-6) and reduced anti-inflammatory IL-10 compared to miR-NC-exos-treated controls (Figures 5A, B). Consistent with these findings, flow cytometry and western blotting confirmed enhanced M1 polarization in miR-mimic-exos group (Figures 5C, D). Moreover, in vivo, miR-mimic-exos intensified anastomotic inflammation, as evidenced by worsened histological scores, elevated endoscopic inflammation, increased pro-inflammatory cytokine levels, and decreased anti-inflammatory cytokine levels (Figures 5E–G). qRT-PCR and immunofluorescence analyses further demonstrated that miR-mimic-exos induced macrophage polarization toward the M1 phenotype (Figures 5H, I). The comprehensive data established mHt-exos promote M1 macrophage polarization through exosomal miR-26b-3p both in vitro and in vivo, highlighting exosomal miR-26b-3p was a key regulator of inflammatory responses across anastomosis inflammation.

To further investigate the molecular basis of these changes, Bioinformatics analysis was performed. Using five prediction databases (TargetScan, miRDB, mirDIP, PicTar, and DIANA), TRIM33 was the primary target of miR-26b-3p (Figures 6A, B). This interaction was functionally validated through a (1) dual-luciferase assay revealing miR-26b-3p-mediated suppression of WT (but not mutant) TRIM33 reporter activity (Figures 6C, D) and (2) western blot confirmation of TRIM33 protein downregulation in miR-26b-3p-transfected iBMDMs (Figure 6E). Pathway analyses revealed the involvement of miR-26b-3p in protein translation (Gene Ontology) and inflammatory pathways, especially the MAPK signaling pathway (Kyoto Encyclopedia of Genes and Genomes) (Figures 6F, G).

To confirm whether miR-26b-3p played a crucial role in the exosomal-mediated M1 polarization of macrophages. We then transducted mHt with lentiviral vectors to knockdown (miR^KD) miR-26b-3p, as well as the corresponding empty lentiviral vectors (miR-NC^KD). Quantitative reverse transcription-polymerase chain reaction was used to identify the transduction efficiency (Supplementary Figure S2A). Exosomes were extracted from miR-NC^KD-mHt and miR^KD-mHt. The miR-26b-3p in miR^KD-mHt was significantly reduced compared with the corresponding negative control in exos (Supplementary Figure S2B). A dramatic decrease was observed after coculture with the miR^KD-mHt-exos in iBMDMs(Supplementary Figure S2C). miR^KD-mHt-exos treatment significantly altered the cytokine profile, increasing pro-inflammatory mediators (TNF-α, IL-6, MPO) (Supplementary Figure S2D). At the transcriptional level, miR^KD-mHt-exos upregulated M1-associated genes (CD86, iNOS and TNF-α) and downregulated M2 markers (Arg1 and CD206) (Supplementary Figure S2E). Consistent with these findings, western blotting showed elevated M1 protein expression (Supplementary Figure S2H), while flow cytometry confirmed an increase in CD86⁺ M1 macrophages (Supplementary Figures S2F, G).

Functional studies demonstrated that miR-26b-3p mimic treatment activated the MAPK pathway (elevated p-p38MAPK) and M1 markers (iNOS, TNF-α), while TRIM33 overexpression exerted opposite effects, with these effects being reversed by miR-26b-3p co-transfection (Figures 7A–D). These results established a definitive TRIM33/MAPK axis through which exosomal miR-26b-3p drives macrophage polarization.

In order to observe whether it has potential clinical value, a number of clinical samples (Figure 8A) were selected for validation. Analysis of anastomotic specimens patients with and without CD revealed significantly elevated miR-26b-3p expression in both inflamed intestinal tissue and adjacent MAT in the CD group than in the control group (Figures 8B, C). Pathological scoring demonstrated a strong positive correlation between miR-26b-3p levels and inflammatory severity at anastomotic sites (Figure 8D). These findings suggest that miR-26b-3p could be a biomarker for postoperative inflammation progression as well as a therapeutic target for modulating M1 macrophage-driven inflammatory responses in CD.