Gegen Qinlian Decoction Attenuates Ulcerative Colitis by Inhibiting PANoptosis: A Multi-Omics and Experimental Study

Summary

Here, we present a protocol integrating multi-omics analysis with in vivo validation to investigate how Gegen Qinlian Decoction alleviates ulcerative colitis by regulating PANoptosis-related targets and inflammatory signaling pathways.

Abstract

We present an integrated bioinformatics and network pharmacology protocol to clarify the therapeutic mechanisms of Gegen Qinlian Decoction (GQD) in ulcerative colitis (UC), with a particular focus on PANoptosis-related regulation. This protocol is designed to systematically identify potential bioactive compounds, therapeutic targets, signaling pathways, and immune-related mechanisms through multi-database mining and transcriptomic analysis. Gene expression profiles from UC cohorts, including GSE87466 and GSE92415 from the GEO database, are analyzed using differential expression analysis and weighted gene co-expression network analysis to identify disease-related genes. Candidate targets of GQD are collected from public pharmacology databases and intersected with UC- and PANoptosis-associated genes to determine core therapeutic targets. Protein-protein interaction network construction, enrichment analyses, and compound-target-pathway network modeling are then performed to reveal the molecular basis of GQD activity. In addition, immune infiltration analysis is incorporated to explore the relationship between key genes and the intestinal immune microenvironment. Molecular docking is further used to evaluate the binding potential between major active compounds and hub targets. This protocol provides a reproducible, systematic framework for investigating how GQD alleviates UC by modulating PANoptosis and related inflammatory pathways, and may serve as a useful reference for mechanistic studies of other traditional Chinese medicine formulas in complex inflammatory diseases.

Introduction

Ulcerative colitis (UC), one of the major forms of inflammatory bowel disease, is a chronic relapsing inflammatory disorder that primarily involves the colonic mucosa and typically begins in the rectum with continuous proximal extension. Its pathogenesis is complex and involves aberrant mucosal immune activation, epithelial barrier dysfunction, microbial perturbation, and dysregulated host-environment interactions, together imposing a substantial and growing global health burden^1,2. Persistent mucosal inflammation not only causes recurrent symptoms such as abdominal pain, diarrhea, and hematochezia, but also increases the long-term risk of colitis-associated colorectal cancer, particularly in patients with extensive and longstanding disease^3,4.

Current pharmacological management of UC relies mainly on 5-aminosalicylates, corticosteroids, immunomodulators, biologics, and small-molecule agents, depending on disease severity and therapeutic response⁵. Although these strategies have improved clinical outcomes, a considerable proportion of patients still experience primary non-response, secondary loss of response, intolerance, steroid dependence, or treatment-related adverse events, highlighting the need for safer and mechanism-oriented therapeutic options⁶. Therefore, identifying upstream molecular programs that simultaneously regulate epithelial injury and intestinal immune dysregulation is of substantial translational importance.

Recent studies have shown that dysregulated programmed cell death is deeply involved in UC pathogenesis. Beyond classical apoptosis, pyroptosis and necroptosis have all been implicated in epithelial barrier destruction and amplification of mucosal inflammation⁷. PANoptosis, a recently defined inflammatory programmed cell death pathway integrating key features of pyroptosis, apoptosis, and necroptosis, has emerged as a particularly relevant mechanistic framework for chronic inflammatory diseases⁸. This process is coordinated by multiprotein PANoptosome complexes and involves central mediators such as ZBP1, RIPK1/RIPK3, CASP8, and inflammasome-related molecules including NLRP3 in a context-dependent manner⁹. Importantly, accumulating evidence indicates that PANoptosis is activated in UC and may promote intestinal epithelial damage, barrier breakdown, and immune imbalance, thereby contributing to disease progression^10,11,12. These findings suggest that PANoptosis is not merely a downstream consequence of inflammation, but may represent a mechanistically actionable node linking epithelial injury to mucosal immune amplification.

Gegen Qinlian Decoction (GQD), a classical prescription first recorded in Shang Han Lun, consists of Pueraria lobata, Scutellaria baicalensis, Coptis chinensis, and Glycyrrhiza uralensis. It has long been used in traditional Chinese medicine for diarrhea and intestinal inflammatory disorders. Modern pharmacological studies have shown that GQD can alleviate experimental colitis by suppressing inflammatory responses, restoring intestinal barrier integrity, reshaping gut microbiota, and modulating immune homeostasis, including Th17/Treg-related signaling and inflammasome activity^13,14. These multitarget actions make GQD a plausible candidate for intervening in PANoptosis-related pathological processes in UC¹⁵. However, despite growing evidence for its anti-colitic efficacy, whether GQD exerts therapeutic effects through coordinated regulation of PANoptosis-associated molecular networks remains unclear.

At present, two major gaps remain. First, existing studies on UC cell death mechanisms have mainly focused on isolated pathways such as apoptosis, pyroptosis, or ferroptosis, whereas the integrated role of PANoptosis in mediating the therapeutic effects of herbal formulas has not been sufficiently clarified^7,10,11. Second, although network pharmacology is well suited to characterizing the “multi-component-multi-target-multi-pathway” nature of traditional formulas, predictions based on compound-target databases alone may lack disease-context specificity and require transcriptomic prioritization and experimental validation to improve biological plausibility and translational reliability^16,17,18,19. Thus, an integrated strategy combining disease transcriptomics, network pharmacology, and in vivo verification may provide a more rigorous framework for elucidating the mechanism of GQD in UC.

Compared with approaches based solely on histopathological observation, single-pathway molecular assays, or database-driven network pharmacology alone, the present workflow is better suited to this question because it enables cross-validation between disease-context transcriptomic signals, formula-target prediction, and biological verification in an experimental colitis model. Histological or molecular readouts alone can capture only limited dimensions of UC pathology, whereas network pharmacology alone may lack sufficient disease specificity and experimental support. By integrating transcriptomic prioritization with in vivo validation, this workflow improves the biological plausibility of target identification and provides a more coherent framework for linking GQD intervention to PANoptosis-associated mechanisms in UC.

This protocol is most appropriate for mechanistic studies aiming to evaluate how multi-component interventions regulate epithelial injury, inflammatory responses, and PANoptosis-related pathways in experimental UC. It is particularly useful when the objective is to connect computational prediction with animal-level biological validation in a standardized workflow. However, it is not intended to replace clinical diagnostic approaches, nor does it comprehensively address all dimensions of UC pathogenesis, such as long-term patient heterogeneity, microbiome-wide causal interactions, or single-cell and spatially resolved immune mapping. Therefore, its scope is best defined as a reproducible preclinical framework for mechanism-oriented and therapeutic evaluation.

We therefore hypothesized that GQD ameliorates UC by modulating PANoptosis-related signaling networks, thereby attenuating intestinal epithelial injury and reprogramming the inflammatory microenvironment. To test this hypothesis, we integrated transcriptomic analysis of UC with network pharmacology to identify shared therapeutic targets and key pathways of GQD, with particular focus on PANoptosis-associated regulators. We further validated the predicted mechanisms in a dextran sulfate sodium (DSS)-induced murine colitis model. By combining computational target prioritization with biological verification, this study aimed to provide a more disease-relevant and mechanistically grounded explanation for the anti-UC effects of GQD, and to offer new evidence supporting PANoptosis as a potential therapeutic axis in UC.

Protocol

All animal procedures were conducted in accordance with institutional guidelines and were approved before study initiation by the Ethical Committee of Chengdu University of Traditional Chinese Medicine (Approval No. 2022-126). Only publicly available GEO and GWAS datasets were used for the bioinformatics analyses; therefore, no additional informed consent or ethics approval was required for the computational component of this study. All procedures involving hazardous chemicals, biological samples, and animal waste were performed in accordance with institutional laboratory safety regulations and institutional procedures for hazardous waste disposal.

Acquisition and preprocessing of GEO datasets

Human transcriptomic datasets related to ulcerative colitis (UC) were retrieved from the GEO database using the keyword “ulcerative colitis.” After screening for expression profiling datasets with clearly defined UC and healthy control groups and sufficient sample size, GSE87466²⁰ and GSE92415²¹, both generated on the GPL13158 platform, were selected as the training cohort. These datasets contained 87 UC samples and 21 healthy controls, and 53 UC samples and 21 healthy controls, respectively. For external validation, GSE87473²² (106 UC samples and 21 healthy controls) and GSE16879²³ (48 UC samples and 12 healthy controls; GPL570 platform) were included. Raw or series-matrix files were downloaded and imported into R for preprocessing. Probe identifiers were converted to gene symbols according to the corresponding platform annotation files. When multiple probes mapped to the same gene symbol, the average expression value was retained. Samples lacking clear group information were excluded before downstream analysis. The two training datasets were merged by common gene symbols. Expression values were log2-transformed when required, normalized using a standard between-array normalization procedure, and corrected for batch effects using a batch-adjustment method after defining dataset origin as the batch variable. Principal component analysis and boxplot inspection were used before and after normalization/batch correction as quality-control checkpoints. Successful preprocessing was defined by improved overlap of sample distributions across datasets and attenuation of dataset-driven clustering. A total of 902 PANoptosis-related genes were compiled from published literature and used as the reference gene set for integrative analysis.

Identification of differentially expressed genes and construction of the WGCNA network

The merged training matrix was analyzed in R using a differential-expression workflow. Genes with |log₂ fold change| ≥ 0.585 and adjusted P < 0.05 were defined as differentially expressed genes (DEGs). Volcano plots and heatmaps were generated as intermediate outputs to confirm that the filtering criteria yielded biologically interpretable expression differences between UC and control groups. For weighted gene co-expression network analysis (WGCNA), genes were ranked by variance across samples, and the top 25% most variable genes were retained as input. A sample-clustering tree was first examined to identify potential outlier samples; no obvious outliers were retained for network construction unless removal was justified by quality-control criteria. A soft-thresholding power was selected based on the scale-free topology fit index, and the smallest power achieving an approximately scale-free network was used to construct the adjacency matrix. The adjacency matrix was then transformed into a topological overlap matrix, and hierarchical clustering was performed using TOM-based dissimilarity. Modules were identified using dynamic tree cutting with a minimum module size of 100 genes. Closely related modules were merged when their eigengene correlation exceeded the preset merging criterion. Module eigengenes, module significance, and gene significance values were calculated to identify modules most strongly associated with the UC phenotype. A successful WGCNA step was defined by stable module separation, biologically plausible module sizes, and one or more modules showing clear correlation with disease status.

Identification of candidate targets in GQD

Candidate compounds and targets corresponding to the four herbal components of Gegen Qinlian Decoction (GQD)-Pueraria lobata, Scutellaria baicalensis, Coptis chinensis, and Glycyrrhiza uralensis-were collected from TCMSP²⁴ and BATMAN-TCM²⁵. Screening thresholds were set as follows: oral bioavailability ≥ 30% and drug-likeness ≥ 0.18 for records derived from one database; confidence score ≥ 0.84 and adjusted P < 0.05 for records derived from another database. Additional compounds were supplemented from an herbal medicine database when relevant records were not captured by the initial search. Canonical SMILES strings of retained compounds were submitted to an online ADME evaluation platform. Compounds were retained when they showed high gastrointestinal absorption and satisfied at least two drug-likeness rule sets among Lipinski, Ghose, Veber, Egan, and Muegge. Putative targets were then predicted using a target-prediction platform with probability > 0.1 and standardized to official gene symbols using a protein annotation database. Duplicate compounds and duplicate targets were removed after database merging. The expected output of this step was a nonredundant GQD compound-target dataset suitable for overlap analysis.

Identifiaction of overlapping genes and performance of functional analysis

Potential therapeutic targets were defined as the intersection among GQD-related targets, UC-related genes identified from differential expression and WGCNA, and the PANoptosis-related gene set. Overlap relationships were visualized using a Venn diagram. The intersecting genes were then submitted to a protein-protein interaction (PPI) database, and interaction pairs with confidence score ≥ 0.4 were retained. The resulting PPI network was exported and visualized in a network-analysis platform. A successful PPI step was defined by the presence of a connected network containing the majority of intersecting genes rather than isolated nodes only. Functional enrichment analysis, including Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis, was performed in R with q < 0.05 as the significance threshold. Terms or pathways not meeting this threshold were excluded. The top enriched biological processes and signaling pathways were summarized to infer the potential mechanism through which GQD may regulate UC. Bar plots and bubble plots were generated as intermediate outputs to verify whether enrichment results were dominated by inflammation-, immunity-, or epithelial injury-related pathways, as expected from the biological context.

Analysis of expression patterns, chromosomal localization, and correlations

Expression levels of the shared genes were compared between UC and control groups using a standard differential-expression visualization workflow in R. Boxplots and heatmaps were generated, and P < 0.05 was considered statistically significant. Only genes with interpretable expression direction and consistent behavior across the training cohort were retained for subsequent mechanistic evaluation. Chromosomal localization of selected core genes was mapped after formatting gene annotation files into the required input structure, and the genomic positions were visualized using a circular-chromosome plotting workflow. Correlation matrices among core genes were then computed using pairwise correlation analysis. The expected result of this step was the identification of gene pairs showing coordinated expression patterns that could support shared regulation or functional interaction.

Construction and evaluation of machine learning models

Diagnostic model development was performed using the merged training cohort (GSE87466 + GSE92415), while GSE16879 and GSE87473 were used as independent external validation datasets. Twelve machine-learning algorithms-Lasso, Ridge, Stepglm, XGBoost, Random Forest, Enet, plsRglm, GBM, NaiveBayes, LDA, glmBoost, and SVM-were used to generate 113 model combinations²⁶. Ten-fold cross-validation was applied within the training cohort. In each fold, model training, feature selection, and performance estimation were conducted using the same training/test partitioning logic to avoid information leakage. Model performance was evaluated primarily by the area under the receiver operating characteristic curve (AUC). The optimal model was defined as the model with the highest average AUC across the training and validation datasets rather than the highest training AUC alone. Models showing very high training AUC but poor external validation performance were not retained. ROC curves were generated for each retained model, and a nomogram was constructed from the final selected biomarkers. A successful modeling step was defined by stable cross-validation performance and preservation of discriminative ability in the external datasets.

Performance of GSVA and profile core gene expression

Gene Set Variation Analysis (GSVA) was performed to compare pathway enrichment between high-expression and low-expression groups defined by the median expression value of each core gene. Adjusted P < 0.05 was used as the cutoff for significant pathway differences. Pathways not meeting this threshold were excluded from interpretation. Heatmaps and correlation plots were generated to illustrate pathway activity and co-expression patterns. Intermediate success of this step was indicated by enrichment profiles consistent with inflammatory, immune, or metabolic dysregulation relevant to UC.

Analysis of the immune landscape and construct the ceRNA network

Immune-cell composition was estimated using a deconvolution algorithm with 1,000 permutations. Samples failing the internal deconvolution significance criterion were excluded from subsequent immune-correlation analysis. To complement this analysis, single-sample gene set enrichment analysis (ssGSEA) was also used to assess immune infiltration patterns. Correlations between immune-cell fractions and core gene expression were then calculated and visualized. For ceRNA-network construction, miRNAs targeting the core mRNAs were identified from publicly available interaction databases, and candidate lncRNAs interacting with these miRNAs were subsequently screened. Only interactions supported by database prediction or annotation records were retained. The mRNA-miRNA-lncRNA regulatory network was visualized in a network-analysis platform. The expected output of this step was a structured ceRNA network with biologically plausible upstream regulators linked to the core genes.

Performance of single-cell analysis

The single-cell dataset GSE214695 was imported into R and converted into single-cell objects using a standard single-cell analysis workflow. Quality control was performed by filtering cells according to standard metrics, including low-feature cells, potential empty droplets, and cells with poor transcript quality. Cells passing quality control were normalized, and the top 2,000 highly variable genes were selected for principal component analysis. Principal components contributing meaningful biological variance were retained for unsupervised clustering, and UMAP was used for two-dimensional visualization. Cluster marker genes were identified using differential-expression analysis across clusters. Cell-type annotation was then performed by reference-based classification, and annotations were cross-checked against canonical marker expression when possible. A successful single-cell step was defined by clear separation of major cell populations on the UMAP plot, interpretable marker genes for each cluster²⁷, and consistent annotation of the principal immune and stromal cell types relevant to UC.

Performance of Mendelian randomization analysis

Two-sample Mendelian randomization (MR) was performed to evaluate possible causal relationships between core gene expression and UC susceptibility. Instrumental variables were extracted from expression quantitative trait locus datasets, and UC genome-wide association study summary statistics were used as the outcome dataset. SNPs were screened according to standard MR quality-control criteria, including relevance to the exposure trait and removal of ambiguous or duplicate variants when required by harmonization. The inverse-variance weighted method was used as the primary causal estimator. Cochran’s Q test was applied to assess heterogeneity. MR-Egger and MR-PRESSO analyses were used as sensitivity analyses to evaluate and, when necessary, correct for horizontal pleiotropy. A successful MR step was defined by valid instrument harmonization, absence of major heterogeneity or pleiotropy²⁸, and directionally consistent estimates across complementary MR methods.

Performance of molecular docking of candidate compounds

Three-dimensional structures of target proteins were obtained from the Protein Data Bank, and ligand structures were obtained from a public small-molecule database. Prior to docking, proteins and ligands were preprocessed by removing water molecules when appropriate, adding hydrogens, defining atom types, and converting file formats required for docking. Docking was then performed using a molecular-docking workflow, and binding energies were calculated for each ligand-target pair. Binding affinities lower than -5 kcal/mol were interpreted as favorable binding, whereas values lower than -7 kcal/mol were interpreted as relatively strong binding. Docking heatmaps were plotted to compare global binding patterns across targets. The top-ranked conformations were visualized in three dimensions to inspect hydrogen bonding, hydrophobic interactions, and residue proximity. Successful completion of this step was defined by stable docking output files, plausible ligand poses within the target binding region, and binding energies consistent with candidate prioritization.

Performance of in vivo experimental validation

GQD preparation

GQD was prepared from four traditional Chinese medicinal herbs in the following crude drug proportions: 24 g of Puerariae Lobatae Radix (Gegen), 9 g of Scutellariae Radix (Huangqin), 9 g of Coptidis Rhizoma (Huanglian), and 6 g of Glycyrrhizae Radix et Rhizoma (Gancao). Details of the botanical materials and their sources are provided in the Table of Materials. The herbs were ground into fine powder, mixed thoroughly, and soaked in 390 mL of distilled water for 1 h. The mixture was decocted twice. For the first decoction, the herbs were boiled over high heat and then simmered over low heat for 30 min. After filtration through gauze, the residues were reboiled with a tenfold volume of water. The two filtrates were combined and concentrated in a water bath to a final crude drug concentration of 5 g/mL. The extract was then filtered, cooled, and stored at 4 °C until use.

Establishment of the DSS-induced UC mouse model

Eighteen male BALB/c mice (8 weeks old) were used in this study, and all animal procedures were approved by the Ethics Committee of Chengdu University of Traditional Chinese Medicine (Approval No. 2022-126). Details of the animal source are provided in the Table of Materials. Mice were housed under specific pathogen-free conditions at 20–22 °C and 55% humidity under a 12 h light/dark cycle. After a 1-week acclimatization period, the animals were randomly divided into three groups (n = 6 per group): NC, UC, and GQD. The NC group received sterile drinking water throughout the experiment, whereas the UC and GQD groups were administered 1.5% (w/v) dextran sulfate sodium (MW 36-50 kDa) in drinking water for 5 consecutive days followed by 2 days of regular water. This cycle was repeated three times to establish chronic colitis. During the intervention period, the NC and UC groups were gavaged twice daily with 200 μL of 0.5% sodium carboxymethyl cellulose, whereas the GQD group received GQD by oral gavage twice daily.

The clinical prescription of GQD consisted of 24 g Gegen, 9 g Huangqin, 9 g Huanglian, and 6 g Gancao, corresponding to a total crude drug amount of 48 g/day for adults. Assuming a standard adult body weight of 60 kg, the equivalent adult dose was 0.8 g/kg/day. Based on the conventional human-to-mouse conversion coefficient of 1:12, the corresponding mouse-equivalent dose was 9.6 g/kg/day. According to the preset low-, medium-, and high-dose ratio of 1:2:4, the calculated doses were 9.6, 19.2, and 38.4 g/kg/day, respectively. Considering the severity and persistence of colonic inflammation in the repeated DSS-induced colitis model, the high-dose regimen was selected to ensure adequate pharmacological intervention during in vivo validation. To support sufficient pharmacological exposure during the active inflammatory phase, the total daily dose of 38.4 g/kg/day was divided into two equal administrations. Throughout the experiment, body weight, stool consistency, and rectal bleeding were monitored daily. At the end of the protocol, mice were euthanized and colon tissues were collected for subsequent analyses.

Evaluation of body weight and disease activity index

Body weight was recorded daily and expressed as the percentage of the initial body weight on Day 0. Disease severity was evaluated using the disease activity index (DAI), which included three parameters: body weight loss, stool consistency, and rectal bleeding. Each parameter was scored independently on a scale of 0–4 according to the severity of the clinical manifestations, with higher scores indicating more severe disease activity. The final DAI was calculated as the average of the three individual scores, yielding a total score ranging from 0 to 4. Briefly, a score of 0 indicated no abnormality, whereas scores of 1–4 reflected progressively greater weight loss, looser stool consistency, and more severe rectal bleeding. The detailed scoring criteria are presented in Table 1.

Performance of H&E staining and calculate the histological activity index

For histological evaluation, colon tissues were fixed in 4% paraformaldehyde, dehydrated through graded ethanol, embedded in paraffin, and sectioned at 4 μm thickness. The sections were deparaffinized, rehydrated, stained with hematoxylin, counterstained with eosin, dehydrated, cleared, and mounted. Histological images were observed using a bright-field microscope and scanned using a slide-scanning system. Specific equipment information is listed in the Table of Materials. Histopathological injury was assessed based on three parameters, including goblet cell depletion, crypt architectural damage, and inflammatory cell infiltration. Each parameter was scored according to the degree of tissue injury, and the sum of the three parameter scores was defined as the histological activity index (HAI), ranging from 0–10, as summarized in Table 2. For goblet cell quantification, five randomly selected non-overlapping fields per section were analyzed at 400× magnification. The number of goblet cells per field was counted manually in a blinded manner, and the average count per field was calculated for each representative colonic section.

Performance of immunohistochemical staining

Following deparaffinization and rehydration, tissue sections underwent antigen retrieval with 0.1% trypsin at 37 °C for 30 min, followed by incubation with 3% hydrogen peroxide to block endogenous peroxidase activity. The sections were then incubated overnight at 4 °C with rabbit primary antibodies against STAT3 (1:200, HUABIO, ET1605-45), TIMP1 (1:1000, Proteintech, 26847-1-AP), SPHK2 (1:500, Proteintech, 17096-1-AP), and HSPA5 (1:1000, Cell Signaling Technology, #3177), diluted in 1% BSA.Details of the antibodies and their sources are provided in the Table of Materials. After washing with PBS, the sections were incubated with the corresponding secondary antibody (1:2000) at room temperature for 1.5 h. Immunoreactivity was visualized using a chromogenic detection reagent, and nuclei were counterstained with hematoxylin. Quantitative image analysis was performed using image-analysis software listed in the Table of Materials.

Performance of qRT-PCR analysis

Total RNA was extracted using a phenol-based RNA extraction reagent and reverse-transcribed using a first-strand cDNA synthesis kit. Reverse transcription was performed at 25 °C for 10 min, 55 °C for 15 min, and 85 °C for 5 min. qRT-PCR was subsequently conducted using a real-time PCR system and a SYBR Green qPCR mix. The amplification conditions were as follows: initial denaturation at 95 °C for 60 s, followed by 40 cycles at 95 °C for 15 s, 60 °C for 15 s, and 72 °C for 45 s. Primer sequences were designed using primer-design software and are listed in Table 3. GAPDH was used as the internal reference, and relative mRNA expression levels were calculated using the 2^{^-ΔΔCt} method. Each sample was analyzed in triplicate, and each experiment was repeated at least three times. Details of the reagents, instruments, and software are provided in the Table of Materials.

Performance of western blot analysis

Colon tissues were homogenized in lysis buffer containing protease inhibitor on ice and lysed thoroughly for 2 h. Protein concentrations were determined using a protein assay kit. Equal amounts of protein were mixed with loading buffer, denatured before electrophoresis, separated by 10% SDS-PAGE, and transferred onto PVDF membranes at 100 V for 1 h. After blocking with 5% non-fat milk for 1 h at room temperature, the membranes were incubated overnight at 4 °C with primary rabbit antibodies against STAT3 (1:200, HUABIO, ET1605-45), TIMP1 (1:1000, Proteintech, 26847-1-AP), SPHK2 (1:500, Proteintech, 17096-1-AP), HSPA5 (1:1000, Cell Signaling Technology, #3177), and GAPDH at the corresponding working dilution. After washing with TBST (20 mM Tris, 150 mM NaCl, 0.1% Tween-20; pH 7.4), the membranes were incubated with the corresponding secondary antibody (1:5000) at room temperature for 1 h. Protein bands were visualized using a chemiluminescent detection reagent. Band intensities were analyzed using image-analysis software, and the expression levels of target proteins were normalized to GAPDH. The expected molecular weights of the detected proteins were approximately 88 kDa for STAT3, 21 kDa for TIMP1, 28 kDa for SPHK2, 70 kDa for HSPA5, and 36 kDa for GAPDH. Details of the antibodies, membranes, reagents, and software are provided in the Table of Materials. Specific product names, manufacturers, catalog numbers, instrument models, and software details are provided in the Table of Materials, whereas generic descriptions are used in the main text to improve neutrality and general applicability.

Performance of statistical analysis

Data are presented as mean ± standard deviation. Statistical testing was performed using statistical software. One-way ANOVA followed by Tukey’s post hoc test was used for multi-group comparisons. A P value ≤ 0.05 was considered statistically significant.

Results

GEO dataset acquisition and UC-Related gene identification

To investigate gene expression changes in UC, two transcriptomic datasets-GSE87466 and GSE92415-were obtained from the GEO database using “ulcerative colitis” as the search term, with selection criteria including dataset type, clear disease/control annotation, and sufficient sample size. These datasets encompassed 140 UC cases and 42 healthy controls. Samples lacking unambiguous phenotype labels or valid expression pro...

Discussion

Ulcerative colitis (UC) is a chronic, idiopathic inflammatory bowel condition primarily affecting the colonic mucosa, characterized by persistent or episodic inflammation. The disease significantly diminishes patient quality of life and elevates the risk of developing colorectal carcinoma. Although its global prevalence is steadily rising, the precise etiology and underlying mechanisms remain only partially understood. Current treatment options often yield suboptimal outcomes and are frequently accompanied by adverse eff...

Materials

List of materials used in this article

Name	Company	Catalog Number	Comments
Puerariae Lobatae Radix	Sichuan Neautus Traditional Chinese Medicine Co., Ltd.	NTC-001	Herbal material for GQD preparation
Scutellariae Radix	Sichuan Neautus Traditional Chinese Medicine Co., Ltd.	NTC-002	Herbal material for GQD preparation
Coptidis Rhizoma	Sichuan Neautus Traditional Chinese Medicine Co., Ltd.	NTC-003	Herbal material for GQD preparation
Glycyrrhizae Radix et Rhizoma	Sichuan Neautus Traditional Chinese Medicine Co., Ltd.	NTC-004	Herbal material for GQD preparation
BALB/c mice, male, 8 weeks old	Beijing HFK Bioscience Co., Ltd.	N/A	Experimental animals
Dextran sulfate sodium (DSS), MW 36-50 kDa	MP Biomedicals	160110	Colitis-inducing reagent
Sodium carboxymethyl cellulose	Sigma-Aldrich, St. Louis, MO, USA	C5678	Vehicle for oral gavage
Paraformaldehyde	Sigma-Aldrich, St. Louis, MO, USA	158127	Tissue fixative
Hematoxylin	Sigma-Aldrich, St. Louis, MO, USA	H9627	Nuclear stain
Eosin	Sigma-Aldrich, St. Louis, MO, USA	E4009	Counterstain for histology
Bright-field microscope	Nikon	Eclipse 80i	Histological observation
Slide-scanning system	3DHISTECH Ltd.	Panoramic 250	Digital whole-slide image acquisition
Trypsin	Gibco, Carlsbad, CA, USA	25200-056	Antigen retrieval reagent
Hydrogen peroxide	Sigma-Aldrich, St. Louis, MO, USA	H1009	Blocking endogenous peroxidase activity
Bovine serum albumin (BSA)	Sigma-Aldrich, St. Louis, MO, USA	A7906	Antibody diluent / blocking reagent
Anti-STAT3 primary antibody, rabbit	Cell Signaling Technology	9139	Primary antibody for IHC/WB
Anti-TIMP1 primary antibody, rabbit	Abcam	ab211925	Primary antibody for IHC/WB
Anti-SPHK2 primary antibody, rabbit	Abcam	ab187030	Primary antibody for IHC/WB
Anti-HSPA5 primary antibody, rabbit	Cell Signaling Technology	3177	Primary antibody for IHC/WB
Anti-GAPDH primary antibody, rabbit	Abcam	ab9485	Loading control for WB
Goat anti-rabbit secondary antibody	Yeasen Biotechnology	33201	Secondary antibody for IHC/WB
Chromogenic detection reagent	Beyotime Biotechnology [Shanghai] Co., Ltd., Shanghai, China	P0202	Signal development for immunohistochemistry
Image-analysis software	Media Cybernetics	Image-Pro Plus 7.0	Quantification of staining and band intensity
Phenol-based RNA extraction reagent	Invitrogen	TRIzol Reagent, 15596026	Total RNA extraction
First-strand cDNA synthesis kit	Beyotime Biotechnology	DRR047A	Reverse transcription
Real-time PCR system	Applied Biosystems	ABI Prism 7300	Quantitative PCR amplification
SYBR Green qPCR mix	Beyotime Biotechnology	DRR420A	qPCR detection reagent
Primer-design software	Premier Biosoft International, Palo Alto, CA, USA	Primer 5.0	Primer design
RIPA lysis buffer	Beyotime Biotechnology	P0013B	Protein extraction reagent
Protease inhibitor (PMSF)	Beyotime Biotechnology	ST506	Protein protection during lysis
BCA protein assay kit	Beyotime Biotechnology	P0010	Protein quantification
PVDF membrane	Merck Millipore	IPVH00010	Protein transfer membrane
Chemiluminescent detection reagent / system	Thermo Fisher Scientific	32106	Signal detection for western blot
Statistical software	IBM	SPSS 23.0	Statistical analysis