2.1. Chemicals and Reagents

DSS salt (9011-18-1) was purchased from Santa Ana MP Biomedical, California, USA. DPG (HY-N0184A), HMGB1 (HY-P70570), TAK-242 (HY-11109), and Ferrostatin-1 (Fer-1, HY-100579) were purchased from MedChemExpress, USA. FITC-Dextran (GC19938 and GC36048) was purchased from GLPBIO, USA. Anti-β-ACTIN (66009-1-Ig), anti-HMGB1 (10829-1-AP), anti-Occludin (27260-1-AP), anti-Claudin-1(28674-1-AP), anti-TLR4 (66350-1-Ig), and anti-GPX4 (67763-1-Ig) antibodies were purchased from Proteintech Biotechnology, China. Anti-E-Cadherin (14472T) and Na,K-ATPase α1 (52387) antibodies were purchased from Danvers Cell signal technology, USA. Anti-NF-κB p65 (T55034) and anti-Phospho-NF-κB p65 (TP56372) antibodies were purchased from Abmart Pharmaceuticals Technology Ltd., China.

2.2. Bioinformatics Analysis of UC Patients

Microarray data derived from the Gene Expression Omnibus (GEO) database (GSE75214), comprising 97 patients with UC, 22 controls, and 75 patients with CD, were used for bioinformatics analysis. Raw data were systematically processed and deep analyzed and images created using R software. To present the transcriptome level differences of HMGB1 gene in control and UC samples, GraphPad Prism software was applied to produce violin plots. The “pROC” and “ggplot2” packages were employed to construct ROC curves and correlation scatter plots. Differentially expressed genes (DEGs) were mapped as heatmaps using “pheatmap” package, and volcano plots associated with DEGs were constructed by “ggplot2” package. Functional enrichment analysis is visually presented via histograms and bubble plots created by the R package “clusterProfiler.”

2.3. Collection of Colon Tissue Samples

In this study, colonic samples were gathered from 10 patients with active UC, and 10 healthy subjects from the Department of Gastroenterology of the First Affiliated Hospital of Anhui Medical University. Obtained colon samples are placed into precooled cryopreservation tubes immediately, flash frozen in liquid nitrogen, then stored at -80°C freezer for subsequent analyses. The study strictly adhered to ethical norms with approval from the Ethics Committee of the Hospital (PJ2022-07-27). Each participant’s written consent was obtained before the beginning of the study.

2.4. DSS-Induced Colitis

Male C57BL/6J mice (age: 6–8 weeks old, body weight: 18–22 g) were purchased from Jiangsu Jicui PharmaTech Co., Ltd. (Production License No.: SCXK (Su) 2023-0009). During the experimental period, the mice adapted to eat and drink freely, and raised for 1 week at 22°C under a 12 h light/12 h dark cycle. Twenty-four mice were randomly divided into four groups: control group (n = 6), DSS group (n = 6), DPG group (n = 6), and DSS + DPG group (n = 6). The experiment lasted for 7 days, during which sufficient food supply was guaranteed. The control group was given free water; the DSS group was given 3% DSS in drinking water to induce acute colitis; the DPG group was given 8 mg/kg/day of DPG in drinking water; and the DSS + DPG group was given both 3% DSS and 8 mg/kg/day of DPG in drinking water. Body weight, fecal traits, and occult blood were monitored and recorded daily in four groups of mice to evaluate the disease activity index (DAI). All experimental procedures strictly followed the guidelines of the Laboratory Animal Center of Anhui Medical University.

2.5. Histological Analysis

Experimental animals were executed by cervical dislocation on day 8 of the trial. Subsequently, colon tissue was excised immediately and measured for colon length. To examine the histopathologic alterations in mouse colon tissues, 4% paraformaldehyde solution was used to fix the colon tissues. Then paraffin embedding, sectioning (4 μm thickness), and hematoxylin–eosin (HE) staining were processed serially. Sections were observed under a light microscope for histologic scoring according to previously described criteria [23].

2.6. Immunofluorescence Staining Analysis

Sections of mouse colon tissue were warmed at 60°C for 1 h to melt the paraffin wax, then dewaxed and hydrated in xylene and ethanol. Subsequently, citrate solution was used to repair the antigen and immunostaining sealer for 1 h. Sections were incubated with primary antibodies (HMGB1, 1:200; TLR4 1:200; Na,K-ATPase α1,1:200; Occludin, 1:200; E-cadherin, 1:500; and Claudin-1,1:200) at 4°C overnight. Then incubate with the corresponding secondary antibodies (1:200) for 1 h, respectively. After DAPI staining, observing the slides using confocal microscopy. ImageJ software was applied to quantitatively analyze the obtained images.

2.7. Transmission Electron Microscopy (TEM) Examination

Colon tissues collected from DSS-induced colitis mice were sliced into 1 mm³ narrow strips, and then sequentially fixed in 2.5% glutaraldehyde and 1% osmium tetroxide solution to maintain ultrastructural integrity. Then, gradient ethanol and acetone dehydration and epoxy resin immersion embedding were performed sequentially. Thin sections were cut using an ultrathin sectioning machine and stained with uranyl acetate and lead citrate. Finally, intestinal epithelial TJs and mitochondrial morphology were observed using a transmission electron microscope (JEM1400, Japan).

2.8. Cell Culture and Treatment

This study was conducted in human colorectal adenocarcinoma cell line Caco-2 acquired from the cell bank of the Chinese Academy of Sciences. Caco-2 cells were incubated in DMEM medium with 10% fetal bovine serum at 37°C with 5% CO₂. Replaced the cell culture solution every 1–2 days. Cells passaged to 3–6 generations are prepared for cellular experimentation. To confirm the role of the HMGB1-TLR4 pathway, cells were grouped into (1) control group: medium culture for 24 h; (2) HMGB1 group: stimulation with HMGB1 (100 ng/mL) for 24 h; (3) HMGB1+TAK-242 group: 2 h pretreatment with the TLR4-specific inhibitor TAK-242 (10 μM) followed by combined HMGB1 (100 ng/mL) stimulation for 24 h; (4) HMGB1 +Fer-1 group: pretreatment with the ferroptosis inhibitor Fer-1(2 μM) for 2 h, followed by HMGB1 (100 ng/mL) combined HMGB1 (100 ng/mL) stimulation for 24 h. All cell experiments were repeated three times to ensure reliability of the results.

2.9. Transepithelial Electrical Resistance (TEER) Measurement

Caco-2 cells were plated on polycarbonate membrane inserts (12 mm diameter, 3 μm pore size; Corning, USA) at a density of 4.0×10⁵ cells per transwell and cultured until monolayers formed, with medium changes every 48 h. Monolayer cells with a steady-state resistance value of at least 500 Ω were selected for the experiments. Cellular interventions take the previously described approach. Resistance measurements are performed using a strictly calibrated and sterilized EVOM2 (World Precision Instruments, USA). The resistance was tested at three random points in the upper chamber and lower chamber for each transwell, averaged, and then calculated according to the following formula: TEER (Ω. cm²) = (Ω sample − Ω blank) × S (membrane area). Blank control was a cell-free transwell system. To accurately reflect changes in cellular barrier function, TEER results were normalized to initial baseline values before treatment.

2.10. Paracellular Permeability Assay

Caco-2 cells were inoculated into transwell inserts until monolayers formed, and then treated with different chemical agents for 24 h. Replace the upper chamber medium with 4 and 10 kDa FITC-dextran (1 mg/mL), respectively, and incubate with monolayers cells for 1 h. Then, fluorescence intensity of the lower chamber medium was detected using a microplate reader at an excitation at 495 nm and an emission at 525 nm. According to relative fluorescence units (RFU), FITC-dextran concentrations were calculated against a standard curve, and the results were normalized to the initial baseline value before treatment.

2.11. Quantitative Real-Time PCR (qRT-PCR)

Total RNA was abstracted from Caco-2 cells and mouse colon samples, then reverse transcribed into cDNA. The primers were synthesized by Shanghai Sangon Biotechnology Co., Ltd. with reference to the primer sequences of mouse and human genes in the gene bank (detailed primer sequences are shown in Table 1). qRT-PCR was conducted using ChamQ SYBR qPCR Master Mix. Applying 2^−ΔΔCT method to compare the expression levels of different genes with β-actin as the internal reference gene.

2.12. Western Blot (WB)

Proteins extracted from human colon samples, mouse colon samples, and Caco-2 cells. Specific procedures as follows: put the sample in the configured protein lysate, lysis on ice for 30 min, high-speed centrifugation at 4°C, 12,000 rpm for 15 min, take up the supernatant and mix with appropriate loading buffer, boil, and denaturation. SDS-PAGE was utilized to isolate equivalent proteins, then transferred to the PVDF membrane. The PVDF membranes were blocked in rapid blocking buffer for 1 h, and then incubated overnight at 4°C with specific primary antibodies namely, anti-β-actin (1:5000), anti-HMGB1 (1:5000), anti-Occludin (1:2000), anti-E-Cadherin (1:2000), anti-TLR4 (1:1000), anti-GPX4 (1:1000), anti-Claudin-1(1:2000), anti-NF-κB p65 (1:1000), and anti-Phospho-NF-κB p65 (1:1000). Next, incubate the membrane with enzyme-labeled secondary antibody (1:10,000) for 1 h at room temperature. Imaging of proteins on PVDF membranes by chemiluminescence detection and quantitative analysis of digital images by ImageJ software.

2.13. Enzyme-Linked Immunosorbent Assay (ELISA)

The level of inflammatory cytokines (TNF-α, IL-1β, IL-6, and IL-17) in mouse colon tissue samples and cell culture supernatants was determined using an ELISA kits (Jiangsu Enzyme Immunity Industry Co., Ltd., China) according to the instructions provided by the manufacturer.

2.14. Statistical Analysis

Statistical analysis and data visualization utilizing SPSS 22.0 and GraphPad Prism 8.0 software. The data satisfying the normal distribution are expressed as $$, otherwise as median (IQR). Comparison between two groups were made by t-test, while one-way ANOVA was used for comparisons between three groups, and post hoc comparisons by LSD test. Considered statistically significant at p < 0.05.

3.1. HMGB1 Expression Was Upregulated in UC

Based on bioinformatic methods, this study analyzed the GSE75214 dataset included in GEO to identify HMGB1 expression in colonic tissues of UC patients. Results suggested that HMGB1 was significantly upregulated in UC patients compared to controls, especially in patients with active UC (Figure 1A,B). The area under the ROC curve for HMGB1 was 0.65, demonstrating its potential to predict UC disease activity (Figure 1C). To validate these findings, WB detection of clinical biopsies confirmed significant elevation of HMGB1 protein levels in patients with UC (Figure 1D,E).

3.2. HMGB1 Aggravates the Inflammatory Response

To clarify the effect of HMGB1 in UC pathogenesis, we divided colon tissue samples from UC patients in the GSE75214 dataset into low and high expression groups based on the median value of HMGB1. According to differential expression screening criteria (fold change≥1.0, p-value<0.05), 51 significant DEGs were identified by DESeq2 algorithm analysis, of which 18 genes were upregulated and 33 genes were downregulated in expression. The volcano map presents a complete distribution pattern of DEGs (Figure 2A), and the heatmap shows the top 20 genes with the most significant differences (Figure 2B). Remarkably, GO functional enrichment indicated these DEGs were mainly involved in the regulation of inflammatory responses (e.g., “leukocyte migration” and “lipopolysaccharide response”) (Figure 2C), while KEGG pathway analyses revealed that they were closely associated with the activation of Toll-like receptor, NF-κB, and IL-17 signaling pathways (Figure 2D).

Taking into account, the HMGB1 mediated inflammatory regulatory network revealed by bioinformatics analysis, this study carried out functional validation by constructing a Caco-2 cell inflammation model and a DSS-induced colitis model in mice. In the Caco-2 cell model, the concentrations of inflammatory mediators TNF-α, IL-1β, IL-6, and IL-17 in the cell culture supernatants were markedly elevated after 24 h of stimulation with HMGB1 (100 ng/mL) (Figure 3A–D). However, the concentrations of above inflammatory mediators were significantly decreased in colonic tissues of DSS-induced colitis mice compared to the DSS group after intervention with DPG, a HMGB1 specific inhibitor (Figure 4G–J). Importantly, inhibition of HMGB1 ameliorated body weight loss, shortened colon length, DAI scores, and histopathological damage in DSS-induced colitis mice (Figure 4A–F).

3.3. HMGB1 is Associated With Intestinal Barrier Damage

We further looked into the effect of HMGB1 on UC intestinal barrier damage. Analysis based on the GSE75214 dataset revealed that HMGB1 negatively correlated with the expression of important barrier proteins (E-cadherin, Claudin-3, Cadherin-7) in colon tissues of UC patients (Spearman’s correlation coefficients of -0.448, -0.443, and -0.316, respectively) (Figure 5A–C), and WB results further confirmed the downregulation of Occludin and E-cadherin protein levels in the colonic tissues of UC patients (Figure 5D–F). In the DSS-induced colitis model, HMGB1 overexpression was accompanied by downregulation of the barrier proteins Occludin, E-cadherin, and Claudin-1, which was reversed by DPG intervention, as further corroborated by immunofluorescence staining results (Figure 6A–I). TEM ultrastructural analyses showed obvious shortening of intestinal villi and widening of the TJs gaps between IECs in the DSS group, whereas suppressing HMGB1 greatly improved structural disorders of intestinal villi and restored the TJs (Figure 6J). Moreover, in vitro experiments likewise proved the detrimental impact of HMGB1 on expression of Occludin, E-cadherin, and Claudin-1 in Caco-2 cells (Figure 7A–D), as well as resulting in lower TEER values and higher FITC permeability (Figure 7E–G). Altogether, these results demonstrate the damaging role of HMGB1 on the integrity of intestinal epithelial barrier.

3.4. HMGB1 Induces Ferroptosis in Intestinal Epithelial Cells

Ferroptosis, a form of programed cell death characterized by GPX4 inactivation and mitochondrial dysfunction, has been shown to be intimately associated with intestinal barrier damage [20]. To verify whether HMGB1 mediated intestinal barrier damage is associated with ferroptosis, we conducted a systematic study in a DSS-induced colitis model. As a results of WB assay, GPX4 expression was significantly decreased in colonic tissues of mice in the DSS group, whereas GPX4 expression was elevated in the DSS + DPG group (Figure 8A). Notably, qRT-PCR results revealed the mRNA levels of ferroptosis-related genes (CS, PTGS2, RPL8, IREB2, and ATP5G3) were significantly upregulated in the DSS group and decreased in the DSS + DPG group (Figure 8H–L). In addition, colonic tissue mitochondria of mice in the DSS group presented typical pathological alterations as observed by TEM, manifested by reduced mitochondrial cristae density, volume reduction, and vacuolization, whereas DPG treatment significantly restored the mitochondrial ultrastructure of colitis mice (Figure 6J), indicating an essential role of HMGB1 in inducing ferroptosis in IECs.

3.5. HMGB1 Induces Ferroptosis Through the TLR4/NF-κB/GPX4 Signaling Pathway

Previous studies have demonstrated that HMGB1 modulates inflammatory responses through activation the TLR4/NF-κB signaling pathway, while GPX4, a key regulator of ferroptosis, possibly regulated by TLR4 signaling in its expression [24, 25]. Consequently, we speculate HMGB1 may exacerbate intestinal barrier damage by inducing ferroptosis in IECs through TLR4/NF-κB/GPX4 signaling pathway. Our results showed that the expression levels of TLR4 and p-NF-κB in the colon tissues of mice in the DSS group were significantly increased, while the expression of GPX4 presented a decreasing trend. After DPG treatment, the levels of TLR4 and p-NF-κB were effectively suppressed, and the expression of GPX4 was restored, as shown in Figure 8A–D. Immunofluorescence staining also confirmed this result (Figure 8E–G). In addition, HMGB1 treatment significantly promoted the expression of TLR4 and p-NF-κB proteins in Caco-2 cells, downregulated GPX4 protein levels (Figure 9A–D), and was accompanied by notable upregulation of mRNAs of ferroptosis-related genes (CS, PTGS2, RPL8, IREB2, and ATP5G3) (Figure 9E–I). Treatment with TAK-242 reverses the above phenomenon. Notably, GPX4, barrier protein Claudin-1 and TEER values were restored in Caco-2 cells when ferroptosis inhibitor Fer-1 added (Figure 9J–M). These results suggest that HMGB1 mediates intestinal barrier damage by regulating ferroptosis through the TLR4/NF-kB/GPX4 pathway.