Resveratrol suppresses ulcerative colitis and colitis-associated colorectal cancer via modulating the miR-31-5p/SATB2 axis

Abstract

BACKGROUND

Ulcerative colitis (UC) can progress to colitis-associated colorectal cancer (CAC), but the transition mechanisms including the role of microRNAs are largely unknown and there is no effective pharmacotherapy for the diseases.

AIM

To elucidate the role of miR-31-5p/special AT-rich sequence-binding protein 2 (SATB2) axis in UC-to-CAC transition and evaluate the therapeutic effect of resveratrol (Res) on CAC.

METHODS

MiR-31-5p expressions in human colorectal cancer (CRC) tissues and human UC mucosa were analyzed. The effects of miR-31-5p mimics and inhibitors on HCT116 cell migration, proliferation, apoptosis, and nuclear factor (NF)-κB/Bcl-2/Bax signaling were examined. Colitis was induced by dextran sulfate sodium in miR-31 transgenic mice. Bioinformatics and dual-luciferase assays were performed to identify the target of miR-31-5p. The potential therapeutic effect of Res on CAC was investigated in mice and HCT116 cells.

RESULTS

Analysis of The Cancer Genome Atlas-Colon Adenocarcinoma human CRC data (n = 603 for CRC, n = 11 for normal control) and a preliminary human UC cohort (n = 4) revealed significant elevation of miR-31-5p in diseased tissues. In HCT116 CRC cells, miR-31-5p mimics promoted tumor cell migration/proliferation, inhibited apoptosis, activated NF-κB, and increased the Bcl-2/Bax ratio, whereas miR-31-5p inhibitors showed the opposite effects. miR-31 transgenic mice showed more severe dextran sulfate sodium-induced colitis as manifested by weight loss, colon shortening, splenomegaly and colon injury, and elevations of pro-inflammatory [cyclooxygenase-2, iNOS, tumor necrosis factor-α, Toll-like receptor-4 (TLR4), NF-κB] and pro-apoptotic (cytochrome C, caspase-9/3) proteins and colon cell apoptosis. miR-31-5p directly targeted and suppressed SATB2 via binding to its 3’-untranslated region, suggesting SATB2 is a functional target of miR-31-5p. SATB2 overexpression partially counteracted the oncogenic effects of miR-31-5p. Res treatment suppressed CRC growth both in mice in vivo and in HCT116 cells in vitro by downregulating miR-31-5p, restoring SATB2 expression, and inhibiting TLR4/NF-κB signaling.

CONCLUSION

MiR-31-5p is an oncogenic molecule and drives UC-to-CAC transition by suppressing SATB2 expression, activating TLR4/NF-κB signaling, and inhibiting apoptosis. Res exerts a therapeutic effect on UC-to-CAC progression by suppressing the miR-31-5p/SATB2 axis.

Key Words: MiR-31-5p; Special AT-rich sequence-binding protein 2; Ulcerative colitis; Colitis-associated cancer; Azoxymethane; Dextran sulphate sodium; Resveratrol

Core Tip: By integrating the microRNA sequencing data of human colorectal cancer from The Cancer Genome Atlas and the transcriptomic data from a small human ulcerative colitis (UC) cohort, we provisionally identify that miR-31-5p promotes the colonic inflammation-to-cancer transition. MiR-31-5p exacerbates dextran sulfate sodium-induced colitis and its transition to colitis-associated colorectal cancer (CAC) by targeting special AT-rich sequence-binding protein 2. The miR-31-5p/special AT-rich sequence-binding protein 2 axis is found to modulate Toll-like receptor-4/nuclear factor-κB and apoptosis pathways, revealing a key molecular link in UC-to-CAC transition. Resveratrol, a small plant-derived compound, ameliorates bowel inflammation and inhibits UC progression to CAC via this axis in mice in vivo. The study provides mechanistic insights into the UC-to-CAC progression and highlights the therapeutic potential of resveratrol for the diseases.

INTRODUCTION

Colorectal cancer (CRC) is the third most common malignancy and the second leading cause of cancer-related mortality worldwide[1], and it is closely associated with genetic susceptibility and chronic inflammation[2]. Inflammatory bowel disease (IBD), particularly ulcerative colitis (UC), markedly increases CRC risk in a manner dependent on disease duration and extent[3,4]. Although UC-associated CRC accounts for only 1%-2% of all CRC cases[5], it is highly lethal, contributing to 10%-15% of UC-related deaths[6], with a cumulative incidence reaching 4.6%-6.5% after 30 years of disease[7]. Even with surgical treatment, survival remains limited once cancer develops. Therefore, elucidating the molecular mechanisms by which chronic inflammation drives UC-to-CRC progression is critical for effective prevention.

MicroRNAs (miRNAs) are a class of non-coding RNAs approximately 22 nucleotides in length that regulate gene expression and participate in diverse pathophysiological processes[8]. Substantial evidence has demonstrated that dysregulation of miRNA expression is closely associated with a wide range of diseases, including inflammatory disorders and cancer[9,10]. Recent studies have further elucidated the intricate roles of various miRNAs in intestinal inflammation and immune regulation. For instance, Czopik et al[11] demonstrated in experimental colitis that hypoxia-inducible factor-2α transcriptionally upregulates miR-29a in CD4+ T cells, which in turn attenuates T helper 1-mediated inflammatory responses by directly targeting the transcription factor T-box expressed in T cells. Importantly, T cell–specific delivery of a miR-29a mimetic significantly alleviates T helper 1-driven colitis in vivo, underscoring the translational potential of miRNA-based immunomodulation in IBD. The miR-31 gene family maps to human chromosome 9p21.3[12]. Mature miR-31 molecules are approximately 21-24 nucleotides in length and exhibit high evolutionary conservation across species. The mature sequences of human hsa-miR-31-5p and mouse mmu-miR-31 differ by only a single terminal base, suggesting potential functional conservation between these species. Previous research has shown that miR-31-5p is markedly upregulated in the inflamed mucosa of patients with UC, suggesting a potential role in the regulation of intestinal mucosal inflammation[13]. Further studies have reported that miR-31-5p is highly expressed in multiple solid tumors, including lung[14], pancreatic[15], osteosarcoma[16], and oral cancers[17], where it promotes cell proliferation, invasion, and metastasis and inhibits apoptosis by targeting tumor suppressor genes, including LATS2[15], the Hippo[15], and AXIN1[16]. In CRC, miR-31-5p has also been shown to enhance tumor cell migration and invasion through repression of NUMB[18]. However, the underlying molecular mechanisms of miR-31-5p during the transition from UC to colitis-associated CRC (CAC) remain largely undefined.

Despite recent advances implicating miR-31-5p in CRC, its critical downstream targets remain to be fully defined. Special AT-rich sequence-binding protein 2 (SATB2) is a nuclear matrix-associated protein that selectively binds AT-rich DNA elements and orchestrates chromatin remodeling[19], thereby shaping colonic epithelial differentiation and tumor-relevant transcriptional programs. However, the role of SATB2 in CRC remains controversial: Reduced SATB2 expression has been linked to poor prognosis and tumor-suppressive effects[20,21], whereas pooled analyses of colorectal adenocarcinoma datasets have suggested that elevated SATB2 may confer pro-tumorigenic activity[22]. These discrepancies may reflect differences in molecular subtype, disease stage, or detection platforms. To date, systematic investigations of the miR-31-5p/SATB2 axis across the UC-to-CAC continuum and its translational potential are still lacking.

Given the well-established functions of miR-31-5p and SATB2 in CRC progression, it is necessary to further explore potential interventional strategies. Resveratrol (Res) is a natural polyphenolic compound derived from plants such as grapes, peanuts, and Polygonum cuspidatum, that exhibits antioxidant, anti-inflammatory, and antitumor properties[23]. It has emerged as a promising candidate for the prevention and treatment of IBD, particularly UC and CAC. Current therapeutic approaches for UC show limited efficacy and considerable adverse effects[24,25], underscoring the need for safer and more effective adjunctive strategies. Despite accumulating evidence supporting its safety and therapeutic potential, the molecular mechanisms by which Res inhibits UC-to-CAC transition remain incompletely defined. The present study aimed to elucidate the regulatory mechanisms of the miR-31-5p/SATB2 axis in UC and CAC, and to delineate the role of Res in modulating malignant phenotypes via this pathway, with the goal of providing novel strategies for clinical intervention.

MATERIALS AND METHODS

Data mining of The Cancer Genome Atlas-Colon Adenocarcinoma

This part of the study aimed to analyze the expression levels of miR-31-5p in human colon adenocarcinoma (COAD) and normal colon tissues derived from a public database. The miRNA-sequence data of 603 COAD (G1 group) and 11 normal colon tissue samples (normal group) were retrieved from The Cancer Genome Atlas-COAD (TCGA-COAD, https://portal.gdc.cancer.gov/). After quality control and batch correction, differentially expressed miRNAs were identified using DESeq2 (false discovery rate < 0.05, |log₂ fold change| > 1). MiR-31-5p expression was validated by Wilcoxon rank-sum test and visualized using ggplot2.

Sampling of human UC tissues and RNA-sequencing

Colonic mucosal biopsies were collected from four patients with UC during colonoscopy at the First Hospital of Shanxi Medical University, with adjacent non-inflamed tissues collected for use as controls. All samples were obtained from the sigmoid colon. Patients aged 18-60 years with endoscopically and histologically confirmed UC (proctitis, left-sided, or extensive colitis) were included, while those with malignancies, other autoimmune diseases, or inadequate biopsy material were excluded. Biopsies were snap-frozen at -80 °C. The study was approved by the Ethics Committee of Shanxi Medical University, No. 2016 LL014, and informed consent was obtained from all participants.

Total colonic RNA was extracted using the Qiagen RNeasy Mini Kit. RNA quality was evaluated by NanoDrop 2000c, Qubit 2.0, and Agilent 2100 Bioanalyzer. Small RNA libraries were prepared from 1.5 μg of high-quality RNA using the NEBNext Ultra Small RNA Library Prep Kit and sequenced on an Illumina HiSeq 2500 platform (single-end, 50 bp).

Small RNA/mRNA differential expression, visualization, and functional enrichment analysis

Raw small RNA and mRNA sequencing data were quality-filtered and aligned to the GRCh37 reference genome. MiRNAs were identified using miRDeep2 and normalized as transcripts per million, while mRNA counts were analyzed with DESeq2. Differential expression was analyzed using the limma package (|log₂ fold change| ≥ 1, false discovery rate < 0.05). Visualization was performed with ggplot2 and ggpubr, and statistical significance was evaluated by t test. The association between miR-31-5p and SATB2 expression was assessed by Pearson correlation analysis. Functional enrichment of downregulated genes was performed using Metascape for Gene Ontology and Kyoto Encyclopedia of Genes and Genomes analyses, highlighting apoptosis-related pathways.

Animal study

MiR-31 transgenic FVB mice (n = 13) and wild-type (WT) littermates (n = 13) (body weight 20-30 g, age 6-8 weeks) were provided by the Experimental Animal Center of Shanxi Medical University, No. SCXK (Jin) 2015-0001. An additional 38 male C57BL/6 mice (6-8 weeks old) were obtained from the same source, No. SCXK (Jin) 2019-0004. All mice were maintained under specific pathogen-free conditions. The animal use protocols were approved by the Ethics Committee of Shanxi Medical University, No. SYDL2026019. Azoxymethane (AOM) and Res (≥ 98% purity) were purchased from Sigma-Aldrich, MO, United States, and dextran sulfate sodium (DSS, 36-50 kDa) was obtained from MP Biomedicals, CA, United States.

Colitis was induced by DSS in FVB mice using a chronic cyclic regimen: 1% DSS in drinking water for 1 week followed by 1 week of drinking regular water, and this cycle was repeated for 2.5 cycles (total: 5 weeks). To assess the role of miR-31 in vivo, we employed a genetically controlled design: MiR-31 transgenic mice and their WT littermates were used as matched genetic backgrounds. Thirteen WT mice were randomized into two groups: Control (n = 4, water only) and DSS (n = 9, cyclic DSS). Similarly, 13 miR-31 transgenic mice were assigned to miR-31 (n = 4, water only) and miR-31 + DSS (n = 9, cyclic DSS) groups. All mice were monitored daily for body weight and stool consistency. On day 35, mice were humanely euthanized under anesthesia with sodium pentobarbital (150-200 mg/kg body weight), in accordance with the American Veterinary Medical Association Guidelines for Euthanasia. This protocol ensured rapid unconsciousness and minimized distress before death, aligning with the 3R (Replacement, Reduction, Refinement) principles and institutional animal care and use committee-approved ethical standards. Colons were harvested and measured as an index of inflammation, and spleens were excised and weighed. Colon tissues were fixed in 40 g/L paraformaldehyde, Swiss-rolled, paraffin-embedded, sectioned (5 μm), and stained with hematoxylin and eosin (H&E) for histological analysis.

CAC was induced by administration of AOM/DSS in mice. Male C57BL/6 mice (6-8 weeks, n = 38) were randomly assigned into five groups: Control (n = 4), AOM (n = 4), AOM/DSS (n = 14), AOM/DSS + Res (n = 11), and Res only (n = 5). Mice received a single intraperitoneal injection of AOM (10 mg/kg) or saline on day 1. From day 7 to day 70, the AOM/DSS and AOM/DSS + Res groups underwent cyclic 1% DSS treatment (7 days DSS/7 days water), while controls received water. Res (50 mg/kg) was administered orally five times per week to the AOM/DSS + Res and Res groups. Body weight, activity, and stool consistency were monitored daily. At the designated experimental endpoint, mice were euthanized under pentobarbital anesthesia according to the American Veterinary Medical Association Guidelines for Euthanasia and institutional animal care and use committee-approved protocols. Euthanasia was performed via intraperitoneal injection of sodium pentobarbital (150-200 mg/kg body weight). Following confirmation of euthanasia, colons were promptly excised, measured for length, and assessed for tumor burden (number recorded). Concurrently, spleens were collected and weighed to evaluate potential immunological effects. Colon samples were preserved in RNAlater for RNA analysis (miR-31, SATB2), processed for protein extraction, or fixed in formalin, Swiss-rolled, and sectioned for H&E staining and immunohistochemistry (IHC).

Histochemical and immunohistochemical staining

Colon tissues were fixed in 40 g/L paraformaldehyde, processed by the Swiss roll technique, dehydrated, and embedded in paraffin. Sections (5 μm thick) were deparaffinized, rehydrated, and stained with H&E according to standard protocols for morphological assessment. H&E-stained sections were scored independently by two blinded examiners using a histopathological scoring system adapted from Poudyal et al[26] and Vlantis et al[27]. In this system, three parameters, inflammatory severity, inflammatory extent, and crypt damage, are each graded as follows: Inflammation severity (0, none; 1, minimal; 2, moderate; 3, severe); inflammation extent (0, none; 1, mucosa; 2, mucosa and submucosa; 3, transmural); and crypt damage (0, none; 1, loss of one-third of crypts; 2, loss of two-thirds of crypts; 3, crypts lost with intact surface epithelium; 4, crypts lost with loss of surface epithelium). The percent area of involvement for each parameter was scored as: 0 (0%), 1 (1%-25%), 2 (26%-50%), 3 (51%-75%), or 4 (76%-100%). The total histology score (range 0-40) was calculated as the sum of the three parameter scores multiplied by the percent area involvement score.

For IHC staining, paraffin sections of colon tissues were subjected to antigen retrieval in pH 8.0 ethylenediaminetetraacetic acid buffer at 98 °C for 20 minutes, followed by IHC staining using the ABC peroxidase method (Zhongshan Golden Bridge) with DAB visualization and hematoxylin counterstaining. Primary antibodies were applied as follows: Toll-like receptor 4 (TLR4, 1:50, Abcam, Cambridge, United Kingdom), nuclear factor-κB (NF-κB, 1:600, Cell Signaling Technology, MA, United States), cyclooxygenase-2 (COX-2, 1:500, Cell Signaling Technology, MA, United States), tumor necrosis factor-α (TNF-α, 1:1000, Abcam, Cambridge, United Kingdom), Ki67 (1:100, Cell Signaling Technology, MA, United States), β-catenin (1:500, Abcam, Cambridge, United Kingdom), and SATB2 (1:3000, Abcam, Cambridge, United Kingdom). Staining was independently evaluated by two blinded pathologists in at least six randomly selected fields per section (200 ×), using a semi-quantitative immunoreactive score (0-12) based on combined staining intensity and the proportion of positive cells.

Terminal deoxynucleotidyl transferase dUTP Nick End Labeling assay

Colon paraffin sections were deparaffinized, rehydrated, and treated with proteinase K (20 μg/mL) at 37 °C for 30 minutes for antigen retrieval. After blocking endogenous peroxidase activity with 3% H₂O₂ (10 minutes), sections were incubated with terminal deoxynucleotidyl transferase dUTP Nick End Labeling (TUNEL) reaction mixture containing terminal deoxynucleotidyl transferase enzyme and biotin-dUTP at 37 °C for 1 hour. Signals were developed using streptavidin-horseradish peroxidase and DAB, followed by hematoxylin counterstaining. Apoptotic cells exhibiting brown nuclear staining were quantified under light microscopy (400 ×). Negative controls were processed without terminal deoxynucleotidyl transferase enzyme.

Cell culture and transfection

NCM-460 normal colon epithelial, HCT116 colorectal carcinoma, and HEK-293T embryonic kidney cell lines were obtained from the Cell Bank of the Chinese Academy of Sciences. Cells were cultured at 37 °C under 50 mL/L CO₂. NCM-460 cells were maintained in RPMI-1640 medium (Boster, Wuhan, China), while HCT116 and HEK-293T cells were grown in high-glucose Dulbecco’s Modified Eagle’s Medium (Boster, Wuhan, China). All media were supplemented with 10% fetal bovine serum (Trans, Beijing, China) and 1% penicillin-streptomycin (Trans, Beijing, China).

SATB2 overexpression plasmids were generated by cloning SATB2 complementary DNA into the pEXP-RB-Mam-enhanced green fluorescent protein vector (RiboBio, Guangzhou, China), with the empty vector used as a control. MiR-31-5p mimics/inhibitors, SATB2 small interfering RNA, and all plasmids were synthesized and sequence-verified (RiboBio, Guangzhou, China). HCT116 cells were transfected with miR-31-5p mimic (30 nM), inhibitor (50 nM), or SATB2 small interfering RNA (5 nM) using HiPerFect (Qiagen, Germany), or with plasmids (200 ng/mL) using Lipofectamine 2000 (Invitrogen, MA, United States). For plasmid transfection, the medium was replaced with complete medium after 6 hours. Rescue assays were performed by co-transfection of cells with miR-31-5p mimic and SATB2 plasmid. Transfection efficiency was confirmed by quantitative reverse transcription polymerase chain reaction (qRT-PCR) and western blotting at 48 hours post-transfection.

Cell viability assay

Cell viability was evaluated using the CCK-8 assay. HCT116 CRC cells were plated in 96-well plates at a density of 10000 cells per well and cultured for 12 hours. After treatment with different concentrations of Res for 24 hours, 10 μL of CCK-8 solution (Solarbio, Beijing, China) was added to each well and incubated for 2 hours at 37 °C. Absorbance was measured at 450 nm using a microplate reader (Thermo Fisher Scientific, MA, United States).

Scratch wound healing assay

Cell migration was assessed using a scratch wound healing assay[28]. Transfected HCT116 cells were seeded in 6-well plates (5 × 10⁵ cells/well) and cultured to 80% confluence. A uniform scratch was created using a sterile pipette tip. After treatment with Res, cells were incubated in serum-free medium for 24 hours. Wound images were captured at 0 and 24 hours using an Olympus microscope, Japan. The wound closure percentage was quantified with ImageJ software using the formula: Wound closure index (%) = [(A₀ – A₂₄)/A₀] × 100%[28], where A₀ and A₂₄ represent the wound area at 0 and 24 hours, respectively.

Colony formation assay

Transfected HCT116 cells were seeded in 6-well plates at 500 cells/well. After 24 hours, Res was added. The cells were cultured for 14 days (medium replaced every 2 days). Colonies were fixed with 40 g/L formaldehyde for 30 minutes, washed with phosphate-buffered saline, and stained with 0.1% crystal violet for 15 minutes. After rinsing, colonies were imaged and counted.

Detection of cell apoptosis using flow cytometry

Following transfection with miR-31-5p mimic or inhibitor, HCT116 cells were treated with 10 μg/mL Res for 24 hours. Cells were harvested using trypsin without ethylenediaminetetraacetic acid, washed with phosphate-buffered saline, and stained with Annexin V-fluorescein isothiocyanate/propidium iodide (PI) (BD Biosciences, NJ, United States) according to the manufacturer’s protocol. Apoptosis was analyzed by flow cytometry (BD FACSCanto II, BD Biosciences, NJ, United States), with quantification of early (Annexin V⁺/PI^-) and late (Annexin V⁺/PI⁺) apoptotic populations[29,30]. Data from three independent replicates were processed using FlowJo software.

RNA extraction and qRT-PCR

Total RNA was extracted from cells and tissues using QIAzol reagent (Qiagen, Germany) according to the manufacturer’s protocol. Complementary DNA was synthesized using the miScript II RT Kit for miRNA and the QuantiNova Reverse Transcription Kit for mRNA (both from Qiagen, Germany). Quantitative PCR was performed using the miScript SYBR Green PCR Kit (for miRNA) or QuantiNova SYBR Green PCR Kit (for mRNA) on a Bio-Rad CFX96 system. U6 small nuclear RNA and glyceraldehyde 3-diphosphate dehydrogenase mRNA served as endogenous controls. The reaction conditions were: 95 °C for 15 minutes, followed by 40 cycles of 94 °C for 15 seconds, 55 °C for 30 seconds, and 70 °C for 30 seconds. Gene expression was analyzed using the 2^–ΔΔCq method. All reactions were performed in triplicate with appropriate negative controls. The primer sequences are listed in Table 1.

Table 1 Quantitative reverse transcription polymerase chain reaction primer sequences.

Target gene	Primer (5'-3')
hsa-miR-31-5p	5'-AGGCAAGAUGCUGGCAUAGCU-3'
hsa-U6	5'-TATTGCTTAAGAATACGCGGTAGAAA-3'
mmu-miR-31	5'-GCAAGATGCTGGCATAGCTGAA-3'
mmu-U6	5'-CTCGCTTCGGCAGCACATATACTACGCTTCACGAATTTGCGTGTC-3'
SATB2	F: 5'-CCTCCTCCGACTGAAGACAG-3'
	R: 5'-TGGTCTGGGTACAGGCCTAC-3'
PPP2R2A	F: 5'-TGGTTACCCCAGAAAAATGC-3'
	R: 5'-TGGCCTCAACCATTAGATCC-3'
GAPDH	F: 5'-GGACTTCGAGCAAGAGATGG-3'
	R: 5'-AGCACTGTGTTGGCGTACAG-3'

hsa: Human; mmu: Mouse; F: Forward; R: Reverse.

Western blotting

Total protein was extracted from tissues and cells using radioimmunoprecipitation assay lysis buffer (Beyotime, Shanghai, China) and quantified with a BCA assay (KeyGEN BioTECH, Nanjing, China). Proteins (20 μg per lane) were separated by 10% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (Bio-Rad, CA, United States) and transferred to polyvinyl difluoride membranes (Millipore, MA, United States). After blocking with 5% non-fat milk in Tris-buffered saline containing Tween 20, membranes were incubated overnight at 4 °C with primary antibodies, including: Anti-SATB2 (1:2000, Abcam, Cambridge, United Kingdom), anti-NF-κB (1:1000, Cell Signaling Technology, MA, United States), anti-COX-2 (1:500, Cell Signaling Technology, MA, United States), anti-TNF-α (1:1000, Abcam, Cambridge, United Kingdom), anti-iNOS (1:1000, Abcam, Cambridge, United Kingdom), anti-Bax (1:2000, Abcam, Cambridge, United Kingdom), anti-Bcl-2 (1:2000, Abcam, Cambridge, United Kingdom), anti-cytochrome C (Cyt-C, 1:1000, Abcam, Cambridge, United Kingdom), anti-caspase-3 (1:2000, Abcam, Cambridge, United Kingdom), anti-caspase-9 (1:2000, Abcam, Cambridge, United Kingdom), anti-TLR4 (1:500, Abcam, Cambridge, United Kingdom), and anti-β-actin (1:5000, Proteintech, Wuhan, China). After washing, membranes were incubated with horseradish peroxidase-conjugated secondary antibodies (1:2000, Trans, Beijing, China) for 2 hours at room temperature. Protein bands were visualized using a ChemiDoc™ XRS+ system (Bio-Rad, CA, United States) and quantified with Image Lab software.

Target prediction for miR-31-5p

Potential targets of miR-31-5p were predicted using TargetScan (seed sequence analysis), miRTarBase (validated interactions), and miRWalk (binding site prediction). Venn analysis identified 10 overlapping candidates, eight of which (SATB2, RASA1, RHOBTB1, PPP2R2A, LATS2, MMP16, NUMB, and RET) were associated with CRC progression. Based on preliminary validation, SATB2 and PPP2R2A were selected for further study. WT (SATB2-WT) and seed sequence-mutated (SATB2-MUT) luciferase reporter vectors were constructed for dual-luciferase assays.

Dual-luciferase reporter assay

To validate miR-31-5p targeting of SATB2 and PPP2R2A, dual-luciferase reporter assays were conducted. WT or mutant 3′-untranslated region (UTR) fragments containing predicted miR-31-5p binding sites were cloned into the pmirGLO vector. HEK293T cells were co-transfected with miR-31-5p mimic or negative control and the reporter plasmids using Lipofectamine 2000. After 48 hours, luciferase activity was measured using a Dual-Luciferase Reporter Assay System (Promega, WI, United States), with Renilla luciferase serving as the internal control. Firefly/Renilla luciferase ratios were compared using Student’s t-test across three independent experiments.

Statistical analysis

All experiments were conducted in triplicate or more. Data are presented as mean ± SD. Two-group comparisons were performed using Student’s t-test, multi-group comparisons were performed using one-way analysis of variance with Tukey’s post hoc test. Statistical analyses were performed in GraphPad Prism 10. P < 0.05 was considered statistically significant.

RESULTS

MiR-31-5p expression levels are elevated in human CRC, human UC, mouse CAC, and HCT116 CRC cells

Using an integrated analysis of public datasets and clinical samples, we assessed miR-31-5p expression in human CRC and UC tissues. TCGA data revealed significant upregulation of miR-31-5p in human CRC tissues (n = 603) compared with normal controls (n = 11, P < 0.001; Figure 1A). To further define its expression in human UC, miRNA sequencing was performed on paired inflamed and non-inflamed human colonic tissues (n = 4). Following rigorous quality control procedures, four paired human UC clinical samples were used for sequencing analysis, leading to the identification of 37 differentially expressed miRNAs (Supplementary Table 1). Hierarchical clustering clearly distinguished UC tissues from controls (Figure 1B), and volcano plot analysis highlighted miR-31-5p as one of the most significantly upregulated miRNAs (Figure 1C), which was confirmed by paired quantitative analysis (P < 0.01; Figure 1D).

Figure 1 MiR-31-5p expression levels are elevated in human colorectal cancer, human ulcerative colitis, mouse colitis-associated colorectal cancer, and HCT116 colorectal cancer cells. A: The Cancer Genome Atlas-derived box plot of miR-31-5p expression in colorectal cancer tissues vs normal controls. ^cP < 0.001 vs normal controls; B: Heatmap of differentially expressed microRNAs in a pilot human ulcerative colitis (UC) vs adjacent normal controls; C: Volcano plot of differentially expressed microRNAs in a pilot human UC vs adjacent normal controls; D: MiR-31-5p expression in paired inflamed and adjacent normal controls colonic tissues from pilot UC patients. ^aP < 0.05 vs adjacent normal controls (n = 4 pairs); E: Quantitative reverse transcription polymerase chain reaction quantification of colonic miR-31 in azoxymethane/dextran sulfate sodium-treated C57BL/6 mice. ^bP < 0.01 vs control (n = 7); F: Quantitative reverse transcription polymerase chain reaction quantification of miR-31-5p expression in NCM-460 cells and HCT116 cells. ^bP < 0.01 vs NCM-460 (n = 7). N: Normal controls; UC: Ulcerative colitis; AOM: Azoxymethane; DSS: Dextran sulfate sodium.

Consistently, qRT-PCR demonstrated elevated miR-31 expression in colonic tissues from AOM/DSS-induced CAC model mice compared with controls (P < 0.01; Figure 1E), as well as in HCT116 CRC cells relative to normal intestinal epithelial NCM-460 cells (P < 0.01; Figure 1F). Overall, compared with the levels in respective controls, significantly elevated miR-31-5p expression levels were detected in colonic tissues of humans with CRC and UC, a mouse model of CAC, and HCT116 CRC cells. These preliminary findings from multiple models suggest that miR-31-5p may play a role in inflammation-associated colorectal tumorigenesis. However, the expression level of miR-31-5p in human UC requires confirmation in expanded patient cohorts.

MiR-31-5p promotes malignant phenotypes of HCT116 cells

To define the role of miR-31-5p in CRC, HCT116 cells were transfected with miR-31-5p mimic or inhibitor, and their modulatory efficiencies on cell malignant phenotypes were confirmed by qRT-PCR. Mimic (30-70 nM) significantly increased miR-31-5p expression (P < 0.01), and the concentration of 30 nM was used thereafter. Inhibitor (50-100 nM) reduced miR-31-5p levels (P < 0.05, P < 0.01), and 50 nM was used thereafter to balance efficacy and cytotoxicity (Figure 2A). Functionally, miR-31-5p overexpression accelerated wound closure, whereas inhibitor impaired cell migration (P < 0.05; Figure 2B). Mimic increased colony formation (P < 0.05), whereas inhibitor decreased colony formation (P < 0.01; Figure 2C). Flow cytometry showed that mimic reduced apoptosis and inhibitor increased apoptosis (P < 0.05; Figure 2D). Mechanistically, mimic elevated NF-κB expression (P < 0.01) and the Bcl-2/Bax ratio (P < 0.05), while inhibitor produced the opposite effects (P < 0.01; Figure 2E). Collectively, in these experiments, miR-31-5p enhanced HCT116 cell migration and proliferation but suppressed cell apoptosis potentially via NF-κB activation and an increased Bcl-2/Bax ratio.

Figure 2 MiR-31-5p modulates migration, proliferation, apoptosis, and related protein expression in HCT116 colorectal cancer cells. A: Quantitative reverse transcription polymerase chain reaction analysis of miR-31-5p expression in HCT116 cells transfected with miR-31-5p mimic or inhibitor (n = 3); B: Wound healing assay showing the effect of miR-31-5p on HCT116 cell migration (scale bar = 200 μm) (n = 3); C: Colony formation assay showing the effect of miR-31-5p on HCT116 cell proliferation (n = 4); D: Flow cytometric analysis of the effect of miR-31-5p on HCT116 cell apoptosis (n = 3); E: Western blotting analysis of the effect of miR-31-5p on apoptosis-related protein expression in HCT116 cells (n = 3). ^aP < 0.05 vs control, ^bP < 0.01 vs control, ^eP < 0.01 vs miR-31-5p mimic (30 nM), ^hP < 0.01 vs miR-31-5p inhibitor (50 nM).

MiR-31 exacerbates DSS-induced colitis by amplifying TLR4/NF-κB signaling and dysregulating epithelial cell fate in mice

In the AOM/DSS-treated mouse model, miR-31 expression was upregulated, prompting further investigation of the effects of miR-31 in the DSS-induced colitis model. MiR-31 transgenic mice (6-8 weeks old) were treated with 2.5 cycles of DSS (1% DSS for 7 days, followed by sterile water for 7 days, total 35 days; Figure 3A). MiR-31 overexpression significantly delayed body weight gain in basal conditions (P < 0.01), and exacerbated body weight loss after onset of DSS-induced colitis (P < 0.05; Figure 3B). Histological analysis revealed that miR-31 overexpression worsened DSS-induced colonic damage, including effects such as thicker bowel walls, disrupted crypt structures, and increased neutrophil and monocyte infiltration (pathological score: 23.22 ± 7.61 vs 10.22 ± 3.80, P < 0.01; Figure 3C and D). Additionally, miR-31 overexpression was associated with colonic shortening (9.48%, P < 0.05; Figure 3E) and increased spleen index (3.82-fold, P < 0.01; Figure 3F), indicating enhanced systemic inflammation. Mechanistically, miR‑31 overexpression further increased the colonic inflammation score (DSS group vs miR‑31 + DSS group: 2.92‑fold vs 6.70‑fold, P < 0.05; Figure 3G). It also elevated the levels of inflammatory markers, including COX‑2, iNOS, and TNF‑α (1.5‑, 2.1‑, and 1.6‑fold, respectively, P < 0.01; Figure 3H). In addition, miR‑31 overexpression induced mitochondrial apoptosis, as evidenced by increased Cyt‑C release and caspase‑9/3 activation (P < 0.05 and P < 0.01; Figure 3I). Moreover, activation of the TLR4/NF‑κB pathway was enhanced by miR‑31 overexpression (P < 0.05 and P < 0.01; Figure 3J). Finally, TUNEL staining showed an increased number of apoptotic cells (P < 0.01; Figure 3K). In summary, miR-31 overexpression worsened the severity of DSS-induced colitis through enhanced TLR4/NF-κB signaling and mitochondrial apoptosis, leading to more severe intestinal injury and systemic inflammation.

Figure 3 MiR-31 overexpression exacerbates dextran sulfate sodium-induced colitis in mice. A: Schematic timeline of dextran sulfate sodium (DSS)-induced colitis in miR-31 transgenic mice; B: Body weight changes with colitis induction. ^bP < 0.01 vs control, ^dP < 0.05 vs DSS; C: Representative hematoxylin and eosin-stained colonic sections from each group (scale bar = 50 μm, 100 μm); D: Histopathological injury scores. ^aP < 0.05 vs control, ^bP < 0.01 vs control, ^eP < 0.01 vs DSS; E: Colon length measurements. ^aP < 0.05 vs control, ^bP < 0.01 vs control, ^dP < 0.05 vs DSS; F: Spleen index (spleen weight/body weight). ^bP < 0.01 vs control, ^eP < 0.01 vs DSS; G: Quantitative reverse transcription polymerase chain reaction analysis of miR-31 expression in colonic tissues. ^aP < 0.05 vs control, ^dP < 0.05 vs DSS; H: Representative western blots and quantification of inflammatory protein expression levels (cyclooxygenase-2, iNOS, tumor necrosis factor-α) in colonic tissues. ^aP < 0.05 vs control, ^bP < 0.01 vs control, ^eP < 0.01 vs DSS (n = 5); I: Representative western blots and quantification of mitochondrial apoptosis-related protein expression levels (cytochrome C, caspase-9, caspase-3) in colonic tissues. ^aP < 0.05 vs control, ^bP < 0.01 vs control, ^dP < 0.05 vs control, ^eP < 0.01 vs DSS (n = 5); J: Representative immunohistochemistry images and semi-quantitative analysis of Toll-like receptor 4, nuclear factor-κB, cyclooxygenase-2, and tumor necrosis factor-α expression levels (scale bar = 100 μm, n = 3). ^aP < 0.05 vs control, ^bP < 0.01 vs control, ^dP < 0.05 vs DSS, ^eP < 0.01 vs DSS; K: Representative terminal deoxynucleotidyl transferase dUTP Nick End Labeling staining of apoptotic cells in colonic tissues (scale bar = 50 μm) and quantitative analysis of the apoptotic index. ^aP < 0.05 vs control, ^bP < 0.01 vs control, ^eP < 0.01 vs DSS (n = 3). NS: Not significant; DSS: Dextran sulfate sodium; COX-2: Cyclooxygenase-2; TNF-α: Tumor necrosis factor-α; Cyt-C: Cytochrome C; TLR-4: Toll-like receptor 4; NF-κB: Nuclear factor-κB; IHC: Immunohistochemistry.

MiR-31-5p targets SATB2 to promote UC progression and carcinogenesis

Using a multi-level analytical strategy, we systematically identified the functional targets of miR-31-5p and delineated its role in UC and CRC. Integrated predictions from miRTarBase, miRWalk, and TargetScan yielded 10 shared candidate targets (Figure 4A). Integration with CRC-related phenotypes further narrowed the candidates to eight genes: SATB2, RASA1, RHOBTB1, PPP2R2A, LATS2, MMP16, NUMB, and RET. Among these, preliminary experiments validated SATB2 and PPP2R2A as targets regulated by miR-31-5p. Consequently, subsequent analyses focused exclusively on the expression profiles of these two confirmed targets (SATB2 and PPP2R2A). qRT-PCR validation showed that miR-31-5p overexpression significantly reduced SATB2 mRNA levels in HCT116 cells (P < 0.01), whereas miR-31-5p inhibition restored its expression (P < 0.01). In contrast, PPP2R2A showed no consistent inverse response to miR-31-5p (Figure 4B). Dual-luciferase reporter assays using WT and mutant SATB2 3’-UTR constructs further demonstrated that miR-31-5p directly bound to the 3’-UTR of SATB2 and markedly suppressed its luciferase activity (P < 0.01), an effect abolished by mutation of the seed-matching site. These findings identify SATB2 as a direct target of miR-31-5p, although additional downstream mediators may contribute to its functional effects.

Figure 4 MiR-31-5p targets SATB2, and this targeting relationship is functionally relevant in ulcerative colitis and colorectal cancer. A: Venn diagram showing shared hsa-miR-31-5p target genes predicted by miRTarBase, miRWalk, and TargetScan; B: Quantitative reverse transcription polymerase chain reaction analysis of SATB2 and PPP2R2A mRNA levels in HCT116 cells transfected with miR-31-5p mimic or inhibitor. ^aP < 0.05 vs control, ^bP < 0.01 vs control (n = 3); C: Schematic representation of wild-type and mutant luciferase reporter constructs containing the miR-31-5p binding site in the 3’-untranslated region of SATB2; D: Dual-luciferase reporter assay confirming that miR-31-5p directly targeted SATB2 but not PPP2R2A. ^bP < 0.01 vs NC + SATB2-wild-type (n = 3); E: Western blot analysis of SATB2 protein expression in HCT116 cells following miR-31-5p modulation. ^aP < 0.05 vs control, ^bP < 0.01 vs control (n = 4); F-H: Heatmap, volcano plot, and box plot showing differentially expressed mRNAs and SATB2 downregulation in human ulcerative colitis inflamed vs adjacent control tissues. ^aP < 0.05 vs N (n = 4 pairs); I: Correlation analysis between miR-31-5p and SATB2 expression in human ulcerative colitis tissues, r = -0.9; J-L: Validation of SATB2 expression in HCT116 colorectal cancer cells and dextran sulfate sodium-induced colitis in miR-31 transgenic mice by quantitative reverse transcription polymerase chain reaction, western blot, and immunohistochemistry. J, ^bP < 0.01 vs NCM-460; K and L: ^aP < 0.05 vs control, ^bP < 0.01 vs control, ^dP < 0.05 vs dextran sulfate sodium, ^eP < 0.01 vs dextran sulfate sodium; M: Kyoto Encyclopedia of Genes and Genomes and Gene Ontology enrichment analyses of SATB2-associated downstream genes. KEGG: Kyoto Encyclopedia of Genes and Genomes; GO: Gene Ontology; DSS: Dextran sulfate sodium; UC: Ulcerative colitis.

PPP2R2A showed no direct interaction with miR-31-5p (Figure 4C and D). Consistently, western blot analysis demonstrated that miR-31-5p overexpression decreased SATB2 protein levels (P < 0.01), whereas its inhibition increased SATB2 protein levels (P < 0.05), supporting post-transcriptional repression (Figure 4E). Transcriptomic profiling revealed extensive mRNAs dysregulation in UC inflamed tissues, with significant downregulation of SATB2 (P < 0.05) showing an inverse correlation with miR-31-5p expression (r = -0.9, P < 0.01; Figure 4F-I). SATB2 expression was also reduced in HCT116 cells compared with normal epithelial cells (P < 0.01; Figure 4J), and in miR-31 transgenic and DSS-induced colitis mouse models, where inflammation and miR-31 overexpression synergistically suppressed colonic SATB2 protein expression (P < 0.01; Figure 4K and L). Gene Ontology and Kyoto Encyclopedia of Genes and Genomes enrichment analyses of SATB2-associated downregulated genes showed significant enrichment in apoptosis, oxidative phosphorylation, and autophagy pathways, with apoptosis being the most prominent (Figure 4M). Collectively, these findings indicate that miR-31-5p directly targets and represses SATB2, which may consequently disrupt epithelial apoptotic and metabolic homeostasis and thus potentially contribute to UC progression and carcinogenesis.

SATB2 partially counteracts miR-31-5p-driven malignant phenotypes in HCT116 cells

To investigate whether SATB2 serves as a key downstream effector contributing to miR-31-5p-driven malignant phenotypes, we performed functional rescue experiments using HCT116 CRC cells. Efficient modulation of SATB2 expression was confirmed in HCT116 cells. Transfection with 5 nM si-SATB2 reduced SATB2 mRNA to 86% of control levels (P < 0.01; Figure 5A), whereas SATB2 overexpression increased SATB2 protein abundance approximately 5.2-fold (P < 0.01; Figure 5B), validating experimental robustness.

Figure 5 SATB2 partially counteracts miR-31-5p-driven malignant phenotypes in HCT116 cells. A: Quantitative reverse transcription polymerase chain reaction identification of the optimal si-SATB2 concentration in HCT116 cells. ^bP < 0.01 vs control, ^eP < 0.01 vs si-SATB2 (3 nM), ^hP < 0.01 vs si-SATB2 (5 nM) (n = 3); B: Western blot validation of SATB2 overexpression efficiency following 200 nM OE-SATB2 plasmid transfection. ^bP < 0.01 vs control (n = 4); C: Wound healing assays showing partial reversal of miR-31-5p-enhanced HCT116 cell migration by SATB2, (scale bar = 200 μm). ^aP < 0.05 vs NC + vector, ^dP < 0.05 vs vector + mimic, ^gP < 0.05 vs NC + OE-SATB2 (n = 3); D: Colony formation assays demonstrating attenuation of miR-31-5p-induced HCT116 cell proliferation by SATB2. ^bP < 0.01 vs NC + vector, ^dP < 0.05 vs vector + mimic, ^hP < 0.01 vs NC + OE-SATB2 (n = 3); E: Flow cytometric analysis of HCT116 cells apoptosis indicating partial rescue of miR-31-5p-mediated anti-apoptotic effects by SATB2. ^bP < 0.01 vs NC + vector, ^eP < 0.01 vs vector + mimic, ^hP < 0.01 vs NC + OE-SATB2 (n = 3); F: Western blotting of inflammatory signaling proteins (Toll-like receptor 4, nuclear factor-κB, cyclooxygenase-2, tumor necrosis factor-α) following miR-31-5p and SATB2 modulation in HCT116 cells. ^aP < 0.05 vs NC + vector, ^bP < 0.01 vs NC + vector, ^dP < 0.05 vs vector + mimic, ^eP < 0.01 vs vector + mimic, ^gP < 0.05 vs NC + OE-SATB2, ^hP < 0.01 vs NC + OE-SATB2 (n = 3); G: Western blots of mitochondrial apoptosis-related proteins (Bcl-2, Bax, cytochrome C, caspase-9, caspase-3) assessing SATB2-mediated reversal of miR-31-5p anti-apoptotic signaling in HCT116 cells. ^aP < 0.05 vs NC + vector, ^bP < 0.01 vs NC + vector, ^dP < 0.05 vs vector + mimic, ^eP < 0.01 vs vector + mimic, ^gP < 0.05 vs NC + OE-SATB2, ^hP < 0.01 vs NC + OE-SATB2 (n = 3). TLR-4: Toll-like receptor 4; NF-κB: Nuclear factor-κB; COX-2: Cyclooxygenase-2; TNF-α: Tumor necrosis factor-α; Cyt-C: Cytochrome C.

Functional assays suggest an opposing relationship between miR-31-5p and SATB2. MiR-31-5p overexpression significantly enhanced cell migration (68.4% increase, P < 0.05; Figure 5C) and proliferation (1.4-fold increase in colony formation, P < 0.01; Figure 5D), while profoundly suppressing apoptosis (Annexin V⁺/PI⁺ cells decreased 56.5%, P < 0.01; Figure 5E). In contrast, SATB2 overexpression partially counteracted these oncogenic effects, resulting in reduced migration (decrease 28.95%, P < 0.05; Figure 5C), decreased clonogenicity (42%, P < 0.01; Figure 5D), and a marked increase in apoptosis (1.62-fold, P < 0.01; Figure 5E). These results suggest that SATB2 participates in the miR-31-5p-regulated signaling network, although it may not be the only downstream effector of miR-31-5p. Co-expression of SATB2 overexpression in miR-31-5p-overexpressing cells partially attenuated the malignant phenotypes, including reduced cell migration (45.3% reduction relative to miR-31-5p only, P < 0.05; Figure 5C), decreased cell proliferation (40.25% decrease compared with only miR-31-5p, P < 0.05; Figure 5D), and increased apoptosis, as evidenced by a 2.57-fold increase in apoptotic cell frequency relative to miR-31-5p only (P < 0.01; Figure 5E). These results suggest that SATB2 is a functionally relevant downstream effector contributing to miR-31-5p-mediated oncogenic signaling, although it may not be the only mediator.

To explore the potential mechanisms underlying the functional interplay between miR-31-5p and SATB2, we examined their effects on inflammatory signaling and apoptosis pathways. miR-31-5p activated the TLR4/NF-κB inflammatory pathway (Figure 5F), as evidenced by increased expression of TLR4 (1.24-fold, P < 0.05) and NF-κB (1.3-fold, P < 0.05), accompanied by robust inductions of COX-2 (1.56-fold increase, P < 0.05) and TNF-α (1.54-fold increase, P < 0.05). SATB2 overexpression was associated with reduced TLR4 expression (38.1% reduction, P < 0.01), inhibited NF-κB activation (P < 0.01), and reduced inflammatory effector expression to near-baseline levels (COX-2 -56.3%, P < 0.01; TNF-α -67.5%, P < 0.05). Furthermore, SATB2 co-expression partially attenuated miR-31-5p-induced inflammatory activation, significantly reducing the expression levels of TLR4 (-36.6%, P < 0.01), NF-κB (-27.4%, P < 0.05), and downstream inflammatory mediators.

MiR-31-5p upregulated the anti-apoptotic protein Bcl-2 while suppressing expression of the pro-apoptotic protein Bax, resulting in a 1.7-fold increase in the Bcl-2/Bax ratio (P < 0.05). This shift was accompanied by inhibition of mitochondrial Cyt-C release (54% reduction, P < 0.01), decreased activation of caspase-9 (26.9% reduction, P < 0.05), and reduced caspase-3 activity (37.1% reduction, P < 0.05). Conversely, SATB2 overexpression was associated with a decreased Bcl-2/Bax ratio (-87.9%, P < 0.01), increased Cyt-C release (0.6-fold increase, P < 0.01), and enhanced activation of caspase-9 and caspase-3 (1.7-fold and 0.73-fold increases, respectively, both P < 0.01). In rescue experiments, co-expression of SATB2 partially attenuated miR-31-5p-induced changes in apoptotic markers, with a 76.7% reduction in the Bcl-2/Bax ratio (P < 0.01), Cyt-C release at 96.5% of the miR-31-5p overexpression group (P < 0.05), and caspase-9 and caspase-3 activities at 73% and 90.9% of those in the miR-31-5p group, respectively (both P < 0.05). These findings suggest that SATB2 may modulate mitochondrial apoptotic signaling and partially counteract the survival advantage conferred by miR-31-5p, although the incomplete rescue indicates that other mechanisms may also be involved (Figure 5G).

Collectively, these findings suggest a potential regulatory network in which miR-31-5p directly targets SATB2 and in turn modulates TLR4/NF-κB inflammatory signaling and mitochondrial apoptosis-related pathways. The concomitant modulation of inflammatory and apoptotic processes may contribute, at least in part, to the progression of CRC, although SATB2 appears to be one of several mediators involved in these effects.

Res suppresses AOM/DSS-induced CAC in mice

Building on our findings that miR-31-5p is aberrantly overexpressed in CRC and promotes malignant phenotypes of CRC cells, as well as colitis progression in transgenic mice through direct suppression of SATB2, we next investigated whether this miR-31-5p/SATB2-driven inflammation-carcinogenesis axis could be therapeutically modulated by exogenous intervention to achieve chemoprevention of CAC. To this end, we employed an AOM/DSS-induced CAC model in C57BL/6 mice to systematically evaluate the suppressive efficacy of Res and elucidate its multi-target molecular mechanisms (Figure 6A). The successfully established CAC mouse model displayed typical pathological features of inflammation-driven tumorigenesis, including systemic inflammation, tissue injury, and immune response activation. From the first DSS cycle onward, model mice displayed progressive body weight loss (Figure 6B). At the experimental endpoint, AOM/DSS treatment induced a continuous neoplastic cascade in the colonic mucosa, progressing from crypt architectural disruption, goblet cell depletion, and epithelial dysplasia to back-to-back gland formation, increased nuclear-to-cytoplasmic ratio, pathological mitotic figures, and invasive growth, consistent with histological criteria for colorectal adenocarcinoma. In contrast, treatment with Res preserved crypt integrity, reduced neutrophil and monocyte-macrophage infiltration, and effectively blocked the transition from dysplasia to carcinoma in situ and invasive carcinoma (Figure 6C), indicating morphological interruption of the carcinogenic process. Meanwhile, colon length was markedly shortened compared with that in controls (P < 0.01; Figure 6D), while the spleen index was significantly increased (P < 0.01; Figure 6E), collectively confirming pathological progression from chronic inflammation to malignancy. Res significantly reduced the tumor burden in CAC mice. Res decreased the tumor incidence from 73.3% in the model group to 16.7%, reduced tumor multiplicity (2.47 vs 0.25), and markedly diminished mean tumor diameter (5.73 mm vs 4.17 mm) (Table 2), confirming its dose-dependent CAC-suppressing efficacy in vivo.

Figure 6 Resveratrol suppresses the development of azoxymethane/dextran sulfate sodium-induced colitis-associated colorectal cancer. A: Schematic overview of the azoxymethane (AOM)/dextran sulfate sodium (DSS)-induced colitis-associated colorectal cancer (CAC) model in C57BL/6 mice; B: Dynamic changes in body weight following AOM/DSS treatment with or without resveratrol. ^bP < 0.01 vs control; C: Representative hematoxylin and eosin stains of colonic tissues showing morphological alterations (scale bar = 50 μm, 100 μm); D: Effects of Res on colon length in AOM/DSS-induced CAC mice. ^bP < 0.01 vs control, ^dP < 0.05 vs AOM/DSS; E: Effects of Res on spleen index (spleen weight/body weight) in AOM/DSS-induced CAC mice. ^bP < 0.01 vs control, ^dP < 0.05 vs AOM/DSS; F: Western blots of SATB2 protein expression in colonic tissues from CAC mice treated with Res. ^aP < 0.05 vs control, ^bP < 0.01 vs control, ^dP < 0.05 vs AOM/DSS; G: Western blots of mitochondrial apoptosis-related protein expression levels (Bax, Bcl-2, cytochrome C, caspase-9, and caspase-3) in colonic tissues. ^aP < 0.05 vs control, ^bP < 0.01 vs control, ^eP < 0.01 vs AOM/DSS; H: Terminal deoxynucleotidyl transferase dUTP Nick End Labeling staining showing apoptotic cells in colonic tissues from CAC mice (scale bar = 50 μm). ^aP < 0.05 vs control, ^dP < 0.05 vs AOM/DSS; I: Immunohistochemistry stains of SATB2, proliferation markers (Ki67 and β-catenin), and inflammatory signaling proteins (Toll-like receptor 4 and nuclear factor-κB) in CAC mouse colonic tissues (scale bar = 100 μm). ^aP < 0.05 vs control, ^bP < 0.01 vs control, ^dP < 0.05 vs AOM/DSS, ^eP < 0.01 vs AOM/DSS; J: Hematoxylin and eosin staining of heart, liver, and kidney tissues showing no apparent toxicity of Res in CAC mice (scale bar = 50 μm, 100 μm). NS: Not significant; AOM: Azoxymethane; DSS: Dextran sulfate sodium; Res: Resveratrol; Cyt-C: Cytochrome C; TLR-4: Toll-like receptor 4; NF-κB: Nuclear factor-κB; IHC: Immunohistochemistry.

Table 2 Effects of resveratrol on tumor incidence, multiplicity, and size in the azoxymethane/dextran sulfate sodium-induced colitis-associated colorectal cancer model.

Group	n	Tumor incidence (%)	Tumor multiplicity (mean)	Mean tumor size (mm)
Control	4	0	0	0
AOM	4	0	0	0
AOM/DSS	14	73.3	2.47	5.72796
AOM/DSS + Res	11	16.7	0.25	4.1664
Res	5	0	0	0

AOM: Azoxymethane; DSS: Dextran sulfate sodium; Res: Resveratrol.

Res exhibited an anti-tumor effect through multiple pathways. First, Res specifically reversed AOM/DSS-induced downregulation of SATB2 protein expression (P < 0.05; Figure 6F), suggesting restoration of SATB2-dependent epithelial differentiation homeostasis during CAC. Second, Res effectively corrected the imbalance between proliferation and apoptosis. On the one hand, Res suppressed aberrant colonic proliferation by activating the mitochondrial apoptotic pathway, as evidenced by a significant reduction in the Bcl-2/Bax ratio (P < 0.01), increased Cyt-C release (P < 0.05), and enhanced activation of caspase-9 and caspase-3 (P < 0.01; Figure 6G). Consistently, in the AOM/DSS-induced mouse CAC model, Res effectively reversed apoptosis suppression and significantly increased the apoptotic index (P < 0.05; Figure 6H) in colonic tissues, thereby restoring physiological apoptotic clearance. On the other hand, Res markedly suppressed the proliferation marker Ki67 (28.4% reduction in nuclear positivity, P < 0.05; Figure 6I) and inhibited nuclear translocation of β-catenin, a key effector of the Wnt/β-catenin pathway (19.7% reduction, P < 0.05; Figure 6I). Treatment with Res also correlated with suppression of TLR4 expression (P < 0.01) and reduced NF-κB p65 levels (P < 0.05; Figure 6I). These effects collectively suppressed key components of inflammatory signaling, including receptor expression, signal transduction, and downstream effector activation, thereby attenuating inflammation-driven tumorigenesis. After treatment with Res at the effective dose used in this study (50 mg/kg/day), histological examination of mouse heart, liver, and kidney tissues revealed preserved architecture without inflammatory infiltration or degenerative necrosis, indicating a favorable safety profile of Res in vivo (Figure 6J).

Collectively, these results suggest that Res exerts an antitumor effect via a multi-target mechanism, involving upregulation of SATB2, suppression of TLR4/NF-κB signaling, inhibition of proliferation-related markers Ki67 and β-catenin, and activation of mitochondrial apoptosis. The convergence of these effects likely underlies the attenuation of UC-to-CAC progression at both cellular and tissue levels.

Res suppresses the malignancy of HCT116 CRC cells via the miR-31-5p/SATB2 axis

We further systematically delineated the antitumor effects of Res in HCT116 CRC cells, with particular emphasis on the miR-31-5p/SATB2 regulatory axis. Functional assays demonstrated that Res inhibited HCT116 cell proliferation in a dose-dependent manner, and 10 μg/mL was selected as an optimal concentration for mechanistic studies based on efficacy and cell viability (P < 0.05; Figure 7A). At the molecular level, Res significantly downregulated miR-31-5p expression (P < 0.01) while concomitantly upregulating its target protein SATB2 (P < 0.01) in HCT116 cells, indicating that Res relieves miR-31-5p-mediated repression of SATB2 (Figure 7B-D). Gain- and loss-of-function analyses further revealed that miR-31-5p promoted HCT116 cell migration (P < 0.05) and proliferation (P < 0.01) while inhibiting apoptosis (P < 0.01), whereas Res effectively reversed these malignant phenotypes induced by miR-31-5p overexpression and exerted synergistic effects when miR-31-5p was suppressed (Figure 7E-H). Mechanistically, Res suppressed NF‑κB signaling by inhibiting miR‑31‑5p expression (Figure 7I). Furthermore, SATB2 mediated the downstream effects of Res, as evidenced by a reduced Bcl‑2/Bax ratio, enhanced Cyt‑C release, and activation of the caspase‑9/caspase‑3 cascade, which collectively reactivated the mitochondrial apoptotic pathway. Moreover, SATB2 was identified as a critical negative regulator of HCT116 cell migration and proliferation and a key mediator of the anti-migratory, anti-proliferative, and pro-apoptotic effects of Res (Figure 7J-O). Collectively, these results demonstrated that Res downregulated miR-31-5p and upregulated SATB2, thereby coordinately suppressing inflammation-associated signaling and activating the mitochondrial apoptotic pathway, which in turn attenuated malignant phenotypes in HCT116 CRC cells. These findings provide experimental evidence for its mechanistic basis as a potential intervention strategy in CRC research.

Figure 7 Resveratrol suppresses malignant phenotypes of HCT116 cells via the miR-31-5p/SATB2 axis. A: CCK-8 assay of HCT116 cell viability after 24 hours of exposure to Resveratrol (Res) (0, 2.5, 5, 10, 20, and 40 μg/mL). ^aP < 0.05 vs 0 μg/mL, ^bP < 0.01 vs 0 μg/mL (n = 8); B: Quantitative reverse transcription polymerase chain reaction analysis of miR-31-5p in HCT116 cells treated with 10 μg/mL Res for 0, 12, or 24 hours. ^bP < 0.01 vs 0 hours (n = 3); C: Quantitative reverse transcription polymerase chain reaction quantification of miR-31 expression in colonic tissues from colitis-associated colorectal cancer mice with Res intervention. ^aP < 0.05 vs control, ^bP < 0.01 vs control, ^eP < 0.01 vs azoxymethane/dextran sulfate sodium (n = 3); D: Western blots of SATB2 protein expression in HCT116 cells treated with Res (0, 2.5, 5, 10, 20, and 40 μg/mL). ^bP < 0.01 vs control (n = 3); E and F: Cell migration evaluated by scratch wound assays under miR-31-5p overexpression or inhibition with Res treatment (scale bar = 200 μm). ^aP < 0.05 vs control, ^bP < 0.01 vs control, ^eP < 0.01 vs miR-31-5p mimic, ^gP < 0.05 vs miR-31-5p inhibitor (n = 3); G: Colony formation assay evaluating the impact of Res on miR-31-5p-modulated proliferative capacity. ^aP < 0.05 vs control, ^bP < 0.01 vs control, ^dP < 0.05 vs miR-31-5p mimic, ^gP < 0.05 vs miR-31-5p inhibitor (n = 3); H: Flow cytometric analysis of apoptosis in HCT116 with altered miR-31-5p expression with/without Res treatment. ^bP < 0.01 vs control, ^eP < 0.01 vs miR-31-5p mimic, ^hP < 0.01 vs miR-31-5p inhibitor (n = 3); I: Western blots of SATB2 and nuclear factor-κB expression and the Bcl-2/Bax ratio after miR-31-5p manipulation with/without Res treatment. ^aP < 0.05 vs control, ^bP < 0.01 vs control, ^dP < 0.05 vs miR-31-5p mimic, ^eP < 0.01 vs miR-31-5p mimic, ^hP < 0.01 vs miR-31-5p inhibitor (n = 3); J and K: Scratch wound assays with SATB2-silenced (si-SATB2) or SATB2-overexpressing (OE-SATB2) HCT116 cells treated with Res (scale bar = 200 μm). ^aP < 0.05 vs control, ^bP < 0.01 vs control, ^eP < 0.01 vs si-SATB2, ^gP < 0.05 vs OE-SATB2 (n = 3); L and M: Colony formation assays with si-SATB2 or OE-SATB2 cells with Res treatment. ^aP < 0.05 vs control, ^bP < 0.01 vs control, ^eP < 0.01 vs si-SATB2, ^gP < 0.05 vs OE-SATB2 (n = 3); N and O: Western blots of apoptosis-related protein expression levels in si-SATB2 or OE-SATB2 cells following Res treatment. ^aP < 0.05 vs control, ^bP < 0.01 vs control, ^dP < 0.05 vs si-SATB2, ^eP < 0.01 vs si-SATB2, ^gP < 0.05 vs OE-SATB2, ^hP < 0.01 vs OE-SATB2 (n = 3). Res: Resveratrol; NS: Not significant; NF-κB: Nuclear factor-κB; AOM: Azoxymethane; DSS: Dextran sulfate sodium; Cyt-C: Cytochrome C.

DISCUSSION

In the present study, we examined the role of the miR-31-5p/SATB2 pathway in chronic inflammation, epigenetic dysregulation, and malignant transformation during CAC development. Our results suggest that this pathway contributes to both the amplification of UC-associated inflammation and the promotion of malignant progression. We further demonstrate that Res exerts an anti-tumor effect by modulating this signaling pathway. Epidemiological evidence consistently indicates that the risk of CRC in UC patients increases in a duration-dependent manner[7], highlighting persistent inflammation, rather than isolated genetic mutations, as a major driving force of CAC. Meta-analyses and cohort studies have demonstrated that longer UC duration and greater inflammatory burden are associated with increased CRC risk[6,7,31]. These observations provide a basis for investigating inflammation-related miRNAs as potential targets for intervention.

At the expression level, our analyses on human TCGA data and a small cohort of UC patients, a mouse CAC model, and a CRC cell line consistently showed marked upregulation of miR-31-5p in inflamed UC mucosa and CRC tissues (Figure 1). This cross-platform agreement suggests that miR-31-5p plays a role in the transition from inflammation to malignancy. However, given the limited sample size of human UC tissues in this study, larger prospective studies are needed to confirm the generalizability of these findings. Nonetheless, our findings are in line with previous studies reporting elevated miR-31 expression and its pro-tumorigenic role in CRC[32,33]. Moreover, gain- and loss-of-function experiments further elucidated the dual pro-inflammatory and pro-carcinogenic actions of miR-31-5p. In vitro, miR-31-5p overexpression significantly promoted migration and proliferation of HCT116 cells while suppressing apoptosis among these cells, accompanied by NF-κB activation and an increased Bcl-2/Bax ratio. In contrast, in vivo, miR-31 transgenic mice treated with DSS exhibited exacerbated colitis phenotypes, including greater body-weight loss, colon shortening, splenomegaly, and more severe histological injury, together with an amplified inflammatory response characterized by elevated expression of COX-2, iNOS, TNF-α, TLR4, and NF-κB. In colonic tissues, miR-31 overexpression was associated with increased apoptosis, as shown by higher TUNEL positivity and upregulation of Cyt-C, caspase-9, and caspase-3. This contrasts with the anti-apoptotic effects observed in vitro, suggesting that the function of miR-31-5p may depend on cell type and microenvironmental context[34]. Based on our data and previous reports, we propose the following explanations: MiR-31-5p-induced NF-κB activation may promote tumor cell survival, whereas in the intestinal epithelium of DSS-induced colitis, exposure to inflammatory signals such as TNF-α may activate death-receptor pathways (e.g., Fas/FasL and TNF-related apoptosis-inducing ligand receptor)[35], leading to epithelial apoptosis and barrier disruption. This process may lead to a cycle of inflammation, tissue injury, barrier breakdown, and further inflammation, which may contribute to the transition from chronic inflammation to CAC. We acknowledge that the interaction between death-receptor pathways and NF-κB signaling in this context remains to be tested directly using genetic or pharmacological approaches.

The identification of SATB2 as a direct target of miR-31-5p represents one of the mechanistic contributions of this study, adding to our understanding of the molecular networks involved in UC-associated colorectal carcinogenesis. In the colorectal lineage, SATB2 serves both as a differentiation marker and a context-dependent regulator of tumor biology, with reduced expression frequently associated with poor prognosis and enhanced invasiveness[36-38]. We observed marked downregulation of SATB2 in UC inflammatory tissues and in miR-31 transgenic DSS models, with a strong inverse correlation observed between SATB2 expression and miR-31-5p levels. Functional rescue experiments suggest that restoration of SATB2 partially reversed miR-31-5p-induced changes in migration, proliferation, and apoptosis, potentially linking miR-31-5p upregulation to modulation of NF-κB signaling and apoptotic pathways. Previous studies have implicated SATB2 in β-catenin signaling and stemness regulation, suggesting its role in epithelial homeostasis and cell-state transitions[38]. Based on these observations, we propose that miR-31-5p-mediated repression of SATB2, in the context of inflammation-driven tissue regeneration, may contribute to the transition from reparative proliferation to malignant expansion in CAC. However, this hypothesis requires further validation using organoid models and single-cell epigenomic approaches.

The results from Res intervention provide a clear translational implication for this study. Previous animal studies have shown that Res alleviates colitis and reduces tumor burden in AOM/DSS models[39,40]. Here, we further demonstrate that Res exerts its protective effects, at least in part, by downregulating miR-31-5p, relieving repression of SATB2, and suppressing NF-κB activity, thereby restoring pro-apoptotic signaling and inhibiting cell proliferation and migration. These findings are consistent with external evidence linking inflammation-induced miR-31-5p upregulation to CAC progression[41]. However, the clinical translation of these promising preclinical findings faces substantial pharmacokinetic hurdles. Although Res is rapidly absorbed after oral administration, it undergoes extensive first-pass metabolism via glucuronidation and sulfation in the liver and intestine, resulting in extremely low systemic bioavailability[42]. Physiologically-based pharmacokinetic modeling has demonstrated that gut first-pass metabolism makes a substantial contribution to the oral bioavailability of Res, with differential effects of enterohepatic circulation on systemic exposure between rats and humans[43]. Moreover, marked interspecies differences in Res glucuronidation have been documented: Rodent liver microsomes glucuronidate approximately 90% of applied Res, whereas human and dog microsomes glucuronidate about 65%, and the conjugation profiles differ significantly between species[44]. In the present study, the 50 mg/kg dose administered to mice represents a standard preclinical dose, selected based on prior dose-response studies in the mouse CAC model, including the work by Cui et al[39]. However, this dose cannot be directly translated to human equivalent dosing via simple allometric scaling. Given the substantial interspecies differences in Res pharmacokinetics, particularly the extensive first-pass metabolism and rapid glucuronidation that severely limit systemic bioavailability in humans, any future clinical translation will require careful dose optimization through dedicated pharmacokinetic and pharmacodynamic studies in humans.

To overcome these bioavailability limitations, a range of emerging delivery strategies have been investigated. Nanoparticle-based delivery systems, including polymeric nanoparticles, solid lipid nanoparticles, and metal-based nanoparticles, have shown promise in enhancing the solubility, absorption, and bioavailability of Res[45]. For colon-targeted delivery specifically, biotin-modified starch nanoparticles have been developed for targeted Res delivery in UC therapy, enhancing intestinal cell permeability via receptor-mediated endocytosis and increasing colonic accumulation. Importantly, colitis model mice treated with Res-loaded biotin-modified starch nanoparticles showed a 25% increase in colon length compared with untreated controls[46]. pH-responsive nanoparticles based on Mesona chinensis polysaccharides and zein have demonstrated controlled release behavior in the gastrointestinal tract and increased enrichment of Res in the colon[47]. Additionally, chemical derivatization of Res (including hydroxylation, methoxylation, and glycosylation) and combination with other agents represent potential approaches to modulate bioavailability and therapeutic efficacy[48,49]. Notably, cytotoxicity and reduced SATB2 expression at higher doses suggest a narrow therapeutic window, underscoring the need to optimize dosing through pharmacokinetic studies, localized colonic delivery strategies (e.g., nanoparticle or colon-targeted formulations), and evaluation of synergy with standard UC therapies such as 5-aminosalicylate and biologics.

The present study has some limitations. First, clinical validation is limited due to small human cohorts (n = 4), necessitating larger-scale studies for generalizability. Second, the mouse CAC model may not fully recapitulate the chronic inflammatory dynamics and neoplastic transition complexity in human UC/CAC. Third, the therapeutic potential of Res in CRC awaits clinical translation due to missing data on long-term toxicity evaluation, formal dose-escalation trials, and human efficacy profiles. Fourth, pharmacokinetic data were not collected in this study. Future work incorporating such assessments would be valuable.

CONCLUSION

The present study elucidates a regulatory axis of miR-31-5p/SATB2/TLR4/NF-κB in UC progression and transition to CAC. We find elevation of miR-31-5p level in human UC/CRC tissues, a mouse CAC model, and a CRC cell line, and this elevation enhances CRC cell proliferation and migration. Mechanistically, miR-31-5p directly targets and suppresses SATB2 expression, causing activation of NF-κB signaling and inhibition of CRC cell apoptosis. Res suppresses CAC by restoring SATB2 expression, suppressing NF-κB, and promoting cancer cell apoptosis.

ACKNOWLEDGEMENTS

Authors thank the mentioned public databases for providing the data and analytical tools.