SwissTargetPrediction (15) (http://www.swisstargetprediction.ch/), PharmMapper (16) (version 2017; http://www.lilab-ecust.cn/pharmmapper/), Similarity Ensemble Approach (17) (https://sea.bkslab.org/) and SuperPred databases (18) (https://prediction.charite.de/index.php) were used to predict potential targets of wogonin. The UniProt database (https://www.uniprot.org/) was used to verify the target information. Duplicate entries were removed to obtain the final list of wogonin targets.

Identifying psoriasis differentially expressed genes (DEGs) and obtaining the wogonin-psoriasis intersection targets. GSE121212(19) and GSE13355(20) datasets were obtained from the Gene Expression Omnibus (GEO) database (https://www.ncbi.nlm.nih.gov/geo/). Each dataset contained >40 samples, with each group consisting of >20 control and 20 psoriasis samples. GSE121212 included gene expression profiles from 28 patients with psoriasis and 38 healthy controls, while GSE13355 included gene expression profiles from 58 patients with psoriasis and 64 healthy skin samples. GEO2R (https://www.ncbi.nlm.nih.gov/geo/geo2r/) was used to analyze the DEGs of each dataset online and volcano plots of each dataset were generated using the online platform SRplot (https://www.bioinformatics.com.cn) (21) for data analysis and visualization. Significantly DEGs were identified using the thresholds of an adjusted P-value <0.05 and |log2 fold change|>1.

Intersection target acquisition and action network construction. Wogonin targets and psoriasis DEGs were imported into Venny 2.1.0 software (https://bioinfogp.cnb.csic.es/tools/venny/index.html) for common target identification. Venn diagrams of the intersecting targets of wogonin in psoriasis were generated. These selected targets were then imported into Cytoscape 3.10.1(22) for visualization and a drug-target-disease action network diagram was constructed.

Construction of protein-protein interaction (PPI) networks. A total of 30 potential targets were imported into the Search Tool for the Retrieval of Interacting Genes/Proteins (STRING) database, with the species set to human and the confidence level set to 0.4, and the options set to ‘Hide disconnected nodes in the network’. This allowed for the retrieval of protein interaction network data after the removal of isolated proteins. The data were then imported into Cytoscape 3.10.1 to construct the PPI network. The degree value (the number of direct interactions for a given protein node) of each target was analyzed using the ‘CentiScaPe2.2’ plug-in and the targets were ranked in descending order of degree value to obtain relevant information regarding the core targets. The ‘cytoHubba’ plug-in was employed to further analyze the PPI network and determine the core targets, with the top 10 targets selected based on their Matthews Correlation Coefficient (MCC) values.

Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis. The Database for Annotation, Visualization and Integrated Discovery (DAVID; https://davidbioinformatics.nih.gov/) (23) was employed with the following parameters: i) ‘OFFICIAL_GENE_SYMBOL’ for the identifier; ii) ‘Homo sapiens’ for the species; and iii) ‘gene list’ for the submission list. The GO and KEGG enrichment analysis data of the core targets were collected in an output table. The top 10 most significant results for each category, including biological process (BP), cellular component (CC) and molecular function (MF), were selected for inclusion in the GO analysis. A total of 20 potential signaling pathways, were obtained in descending order of count value (P<0.05) and used for the analysis of KEGG pathways. The resulting data were imported into the SRplot platform to generate a visual representation of the GO function in conjunction with the KEGG pathway enrichment analysis data forming the corresponding enrichment bubble plot.

Molecular docking. Molecular docking was performed using AutoDockTools 1.5.6(24). The X-ray crystal structures of the key targets were obtained from the Protein Data Bank (PDB) (www.rcsb.org). The 2D structures of the active compounds were retrieved from PubChem (https://pubchem.ncbi.nlm.nih.gov/). The structures were processed using AutoDockTools 1.5.6, including the removal of water molecules, the addition of charges and non-polar hydrogen atoms. Both the target and compound structures were saved in PDBQT format. Molecular docking was performed using AutoDock Vina software (25) and good affinity conformations (affinity <-7 kcal/mol) were analyzed and visualized using AutoDockTools 1.5.6 and PyMOL (26).

A total of 15 male BALB/c mice, aged 6-8 weeks and weighing 20-22 g, were obtained from SPF (Beijing) Biotechnology Co., Ltd. [animal license no. SCXK (Beijing) 2019-0010]. The mice were maintained under the following conditions: Temperature was controlled at 22±2˚C, relative humidity was 50±10%, and a 12-h light/12-h dark cycle was maintained. All mice had free access to standard rodent chow and autoclaved drinking water. Male mice were selected for the present study to avoid potential fluctuations in immune response and disease severity associated with the female estrous cycle, thereby minimizing experimental variables, an approach consistent with numerous previous studies utilizing the IMQ-induced psoriasis model (27-29).

The mice were randomly assigned to three groups, a control group, an IMQ group and an IMQ + wogonin group. The dorsal fur of all mice was removed to create an exposed area measuring ~2x3 cm. Subsequently, 42 mg IMQ cream (Sichuan Mingxin Pharmaceutical Co., Ltd.) was applied daily as a modeling inducer for a total of 7 consecutive days (30), while saline was applied to the control group. From days 1-7, the IMQ and IMQ + wogonin groups received 42 mg of IMQ cream daily on their shaved backs, while the mice in the IMQ + wogonin group additionally received an intragastric gavage administration of 30 mg/kg/day wogonin (cat. no. HY-N0400; MedChemExpress) (dissolved in sodium carboxymethyl cellulose solvent). On day 8, all mice were euthanized. Prior to euthanasia, mice were anesthetized through an intraperitoneal injection of 1.5% pentobarbital sodium (50 mg/kg). The depth of anesthesia was ensured by the absence of pedal and corneal reflexes. Euthanasia was then performed by cervical dislocation under deep anesthesia. Mortality was verified by ensuring the cessation of breathing and heartbeat.

Animal health and behavior were monitored at least twice daily throughout the present study. The specific humane endpoints predefined for the present study, which would necessitate immediate euthanasia, included: i) Severe lethargy or unresponsiveness to gentle stimuli; ii) the inability to access food or water; iii) weight loss >20% of baseline body weight; or iv) signs of severe systemic illness, such as uncontrolled bleeding, seizures or paralysis.

The present animal study protocol was approved by the Animal Experiment Ethics Committee of Yunnan University of Traditional Chinese Medicine (Kunming, China; approval no. R-062021LH048).

Psoriasis Area and Severity Index (PASI) score of mouse skin lesion tissues. Mouse dorsal lesions in each treatment group were scored based on the area and severity of psoriasis using the PASI score (31). The PASI score was assessed on a five-point scale for erythema, infiltrates and scaling ranging from 0 to 4 (0=none, 1=mild, 2=moderate, 3=severe and 4=very severe), and was calculated by combining the scores for erythema, infiltration and scaling. The cumulative score served as a measure of inflammation severity (scale 0-12).

H&E staining and observation of mouse skin lesions. Following euthanasia on day 8, dorsal skin lesion specimens were immediately collected from each mouse. Mouse skin lesion tissues were fixed in 4% paraformaldehyde at 4˚C for 48 h, with the fixative being changed once during this period. The tissues were trimmed to a consistent size (~5x5 mm) and rinsed. Subsequently, the tissue samples underwent a series of dehydration steps using gradients of ethanol (100, 95, 85 and 75%), followed by treatment with xylene and embedding in paraffin wax. The samples were then sectioned to a thickness of 4 µm. Each section was dried, dewaxed in xylene, and then hydrated through a descending ethanol series (100, 95, 85 and 75%). Subsequently, the sectioned samples were subjected to H&E staining, which included hematoxylin staining (5 min), hydrochloric acid ethanol differentiation (1 min) and eosin staining (1 min) at 25˚C. Finally, the samples were dehydrated, rendered transparent and sealed. At the conclusion of the experiment, histological alterations in the skin lesion tissues were examined and documented under a light microscope (BX43; Olympus Corporation). Finally, using ImageJ 1.8.0 software (National Institutes of Health), three random fields per section were selected to measure the epidermal thickness of the lesion.

Immunohistochemistry. Immunohistochemistry was performed to assess the expression of matrix metalloproteinase-9 (MMP9), cyclin-D1 (CCND1), cyclin-B1 (CCNB1), cyclin-A2 (CCNA2), cyclin-dependent kinase 1 (CDK1) and androgen receptor (AR) proteins in mouse skin lesion tissues. Mouse skin lesion tissues were sectioned, baked, deparaffinized, hydrated (as aforementioned for the HE staining process) and subjected to antigen retrieval using 10 mM citrate buffer at 95˚C for 10 min. Next, endogenous peroxidase activity was blocked by incubation with 3% H₂O₂ for 15 min at 25˚C. Following the incubation of sections with 5% BSA (cat. no. A8012; Beijing Solarbio Science & Technology Co., Ltd.) for 30 min at 25˚C. Subsequently, the sections were incubated overnight at 4˚C with the following primary antibodies diluted in 5% BSA: Anti-MMP9 (1:500; cat. no. AF5228; Affinity Biosciences, Ltd.), anti-CCND1 (1:500; cat. no. AF0931; Affinity Biosciences, Ltd.), anti-CCNA2 (1:500; cat. no. AF0142; Affinity Biosciences, Ltd.), anti-CDK1 (1:500; cat. no. DF6024; Affinity Biosciences, Ltd.), anti-CCNB1 (1:500; cat. no. BF0062; Affinity Biosciences, Ltd.) and anti-AR (1:500; cat. no. AF6137; Affinity Biosciences, Ltd.), followed by overnight incubation with horseradish peroxidase-labeled goat anti-rabbit secondary antibody (1:200; cat. no. S0001; Affinity Biosciences, Ltd.) or horseradish peroxidase-labeled goat anti-mouse secondary antibody (1:200, cat. no. S0002; Affinity Biosciences, Ltd.) at 4˚C. The sections were stained with 3,3'-diaminobenzidine, counterstained with hematoxylin for 30-60 sec at 25˚C and sealed after dehydration and clearing. Visualization was performed under the BX43 light microscope (Olympus Corporation). Finally, using ImageJ 1.8.0 software (National Institutes of Health), three random fields per section were selected to measure the staining intensity.

RNA sequencing (RNA-seq). Total RNA was extracted from the three groups of skin samples using the RNeasy Plus Mini kit (cat. no. 74134; Qiagen China Co., Ltd.) according to the manufacturer's protocols. RNA concentration was determined using a Qubit fluorometer (Invitrogen; Thermo Fisher Scientific, Inc.) and RNA integrity was assessed using Agilent 4200 TapeStation (Agilent Technologies, Inc.). The assay was performed strictly following the manufacturer's instructions using an RNA ScreenTape assay kit, and data were analyzed with the accompanying TapeStation Analysis Software (version A.02.02). Samples with an RNA integrity score of ≥8 were included in the present study. An indexed library was constructed from 500 ng of total RNA according to the manufacturer's instructions using the TruSeq Stranded mRNA sample preparation kit (cat. no. 20020594; Illumina, Inc.). The library concentration and fragment length distribution were assessed using Qubit fluorometer and the Agilent 2100 Bioanalyzer (Agilent Technologies, Inc.). The concentration was required to be >5 ng/µl and the fragment length was required to be between 300-400 bp. Paired-end sequencing (2x150 bp) was performed on an Illumina NovaSeq 6000 platform using the NovaSeq 6000 S4 Reagent Kit (300 cycles; cat. no. 20028312; Illumina, Inc.). Bioinformatic analysis: Raw sequencing reads (FASTQ files) were quality-controlled and trimmed using fastp (version 0.23.2). Clean reads were then aligned to the mouse reference genome (GRCm39) using STAR (version 2.7.10a). Gene-level counts were obtained using featureCounts (from the Subread package; version 2.0.3). Differential expression analysis was performed using the DESeq2 R package (version 1.38.3).

A total of 100 targets of wogonin were obtained from the SwissTargetPrediction database, 196 from the PharmMapper database, 95 from the Similarity Ensemble Approach database and 80 from the SuperPred database. After eliminating duplicates, 423 drug targets were identified (Table SI).

By comparing the differences between patients with psoriasis and healthy controls, 15,015 and 989 DEGs were identified in GSE121212 and GSE13355 datasets, respectively (Fig. 1A and B). The intersection of the two datasets yielded 895 DEGs related to psoriasis (Fig. 1C). Subsequently, the overlap between the DEGs of psoriasis and the targets of wogonin revealed 30 targets, which were defined as wogonin-psoriasis cross-targets (Fig. 1D; Table I), suggesting that these targets may serve a key role in mediating the therapeutic effects of wogonin in psoriasis.

A drug-target-disease action network was constructed to visualize the complex relationship between wogonin, its targets and psoriasis (Fig. 1E). This network illustrates the interaction between wogonin and the 30 cross-targets, highlighting the relationship between these targets and psoriasis. The nodes in the network represent drugs (wogonin), cross-targets and diseases (psoriasis), while the edges represent their interactions. By mapping these interactions, the network provides a comprehensive overview of the potential mechanisms underlying the effects of wogonin on psoriasis.

STRING was used to construct a PPI network of wogonin-psoriasis cross-targets and the network was further analyzed using Cytoscape 3.10.1. The PPI network consisted of 26 nodes and 71 edges, with 4 unconnected nodes hidden. Next, the degree value of each target was obtained through the ‘CentiScaPe2.2’ plug-in. The top 10 targets with a higher degree value were CCND1, MMP9, CCNB1, CCNA2, CDK1, AR, STAT1, G1/S-specific cyclin-E1 (CCNE1), induced myeloid leukemia cell differentiation protein Mcl-1 (MCL1) and DNA topoisomerase IIα (TOP2A; Fig. 2A). The MCC value for each target was obtained using the ‘cytoHubba’ plug-in to identify the core targets. The top 10 targets with higher MCC values were CDK1, CCNB1, CCND1, CCNA2, CCNE1, MMP9, AR, MCL1, TOP2A and CCNE2 (Fig. 2B). By taking the intersection of the top 10 targets ranked by degree values and the top 10 targets ranked by MCC values, CCND1, MMP9, CCNB1, CCNA2, CDK1 and AR were selected as core targets for subsequent molecular docking and validation in animal experiments.

A total of 30 intersecting targets were subjected to gene enrichment analysis using DAVID. Among the 579 entries related to BP, the majority of the targets were involved in the ‘regulation of cyclin-dependent protein serine/threonine kinase activity’, ‘regulation of cyclin-dependent protein kinase activity’ and the ‘negative regulation of epithelial cell differentiation’. CC yielded 29 entries, primarily ‘serine/threonine protein kinase complex’, ‘cyclin-dependent protein kinase holoenzyme complex’ and ‘protein kinase complex’. MF yielded 93 entries, which predominantly included ‘cyclin-dependent protein serine/threonine kinase regulator activity’, ‘kinase regulator activity’ and ‘protein kinase regulator activity’. The results with the top 10 count values were selected for visualization through bar graphs, demonstrating the BP, CC and MF components (Fig. 3A; Table SII).

KEGG pathway enrichment analysis was conducted using DAVID, resulting in the identification of 51 signaling pathways. These pathways were primarily associated with the ‘cell cycle’, ‘PI3K-Akt signaling pathway’, ‘p53 signaling pathway’ and ‘JAK-STAT signaling pathway’ amongst others. Subsequently, the top 20 pathways were selected based on their count value and a KEGG signaling pathway sankey diagram and dot plot was constructed (Fig. 3B; Table SIII).

Molecular docking was used to analyze the binding activities of six core targets (CCND1, MMP9, CCNB1, CCNA2, CDK1 and AR) to wogonin. It is generally considered that a docking energy value of <-7 kcal/mol indicates that the target protein binds strongly to small molecules, while a docking energy value of <-5 kcal/mol indicates a moderate binding affinity (32). The molecular docking results (Fig. 4; Table II) showed that the conformation of the active compound exhibited a strong binding effect to the protein targets, showing reliable interactions.

To evaluate the ameliorative effect of wogonin on psoriasis-like lesions in IMQ-induced mice, wogonin was administered to psoriasis mice (Fig. 5A). Following a 7-day intervention period, skin lesions were scored using the PASI system and pathological changes in the skin were observed using H&E staining. The PASI scores (Fig. 5B and C) of mice in the IMQ intervention group were significantly higher than those in the control group (P<0.05), while the PASI scores of mice in the IMQ + wogonin group were lower than those in the IMQ group, albeit not significantly different. The H&E staining results (Fig. 5D and E) demonstrated that the epidermis was thickened and epidermal protrusions were elongated in the IMQ group compared with those in the control (P<0.05), whereas the thickness of the epidermis was significantly reduced after wogonin treatment (P<0.05). These fundings suggest that wogonin may be an effective intervention to improve skin damage in psoriasis-induced mice.

To further clarify the effect of wogonin on the expression of CCND1, MMP9, CCNB1, CCNA2, CDK1 and AR proteins in the skin of psoriasis-induced mice, protein expression was detected by immunohistochemistry. The results (Fig. 6) showed that, compared with that in the control group, CCND1 and CCNB1 expression in the skin tissues of mice in the IMQ group was significantly decreased (P<0.05), while MMP9, CCNA2, CDK1 and AR protein expression was significantly increased (P<0.05). Compared with that in the IMQ group, CCND1 and CCNB1 expression in the skin tissues of mice in the IMQ + wogonin group was significantly increased (P<0.05), while MMP9, CCNA2 and CDK1 protein expression was significantly decreased. These results suggest that CCND1, MMP9, CCNB1, CCNA2 and CDK1 may be key targets of wogonin in the treatment of psoriasis.

To further validate the selected core targets, RNA-seq analysis was performed. The volcano plot illustrating the DEGs between the IMQ and control groups is shown in Fig. 7A, while Fig. 7B displays the corresponding plot for the wogonin group vs. the IMQ group. Fig. 7C summarizes the log₂ (fold change) values and adjusted P-values of the core targets across pairwise comparisons among the three groups. The results showed that, compared with the control group, CCND1, CCNB1 and CDK1 were significantly downregulated in the IMQ group, while MMP9 and CCNA2 were significantly upregulated. By contrast, compared with the IMQ group, MMP9, CCNA2 and CDK1 were significantly downregulated in the wogonin group, while CCND1 was significantly upregulated. These changes in MMP9, CCNA2 and CCND1 expression were consistent with the immunohistochemical findings.