In this study, we investigated whether hAMSCs could be used for the treatment of psoriasis. To this end, we constructed a psoriasis mouse model using imiquimod cream (IMQ) (model group) and injected hAMSCs into the psoriasis mouse model on day 0 (1 day before modelling) and day 4 (treatment group) (Figure 1A). Compared with those in the normal group, the model group had obvious psoriasis-like skin lesions; however, after hAMSCs treatment on day 0, the severity of skin lesions in the psoriasis treatment group was significantly lower than that in the model group (Figure 1B), suggesting that hAMSCs also have a certain preventive effect on the occurrence of psoriasis. Furthermore, the decrease in the PASI score after the second administration of hAMSCs on the fourth day also confirmed that hAMSCs could significantly inhibit imiquimod induced psoriasis-like skin lesions in mice (Figure 1C). Histological analysis revealed that compared with the control group, the skin sections of the model group showed obvious pathological changes, including hyperkeratosis, parakeratosis, Munro microabscesses, and thickening of the spinous layer. hAMSCs could significantly improve the occurrence of such pathological changes (Figure 1D). Moreover, the Baker scores of the skin sections from the three groups of mice also yielded the same result (Figure 1E). Cellular proliferation marker Ki67 was decreased in the psoriasis model mice treated with hAMSCs as indicated by immunohistochemistry (Figures 1F, G). These results indicate that hAMSCs can alleviate the severity of psoriasis-like skin lesions and reduce epidermal hyperproliferation in psoriasis model mice.

To evaluate the effects of hAMSCs on inflammatory factors in psoriasis, we measured the serum levels of IL-17 and TNF-α in the three groups of mice. We found that hAMSCs significantly alleviated imiquimod-induced splenomegaly (Figures 1H, I) and the levels of inflammatory factors in the serum (Figures 1J, K). These results suggest that hAMSCs may play important roles in regulating inflammatory factor levels and modulating immune responses in psoriasis mice. In addition, to evaluate the safety of hAMSCs in vivo, we measured the serum alanine aminotransferase (ALT) and aspartate aminotransferase (AST) levels in the mice from the three groups (Figures 1L, M). We found no statistically significant differences among the groups, confirming the safety of hAMSCs for therapeutic application.

In addition, we established an in vitro psoriasis cell model using TNFα-stimulated HaCaT cells. The results showed that TNFα stimulation significantly enhanced the generation of reactive oxygen species (ROS) (Supplementary Figures 1A, B) and the proliferation ability (Supplementary Figure 1C, D) of HaCaT cells; while co-culture with hAMSCs significantly alleviated the biological effects induced by TNFα. This result suggests that hAMSCs may play a regulatory role in the in vitro psoriasis model by inhibiting cell proliferation and ROS generation. In conclusion, hAMSCs have demonstrated excellent therapeutic effects in both in vivo and in vitro models of psoriasis.

To elucidate the pathogenesis of psoriasis and the mechanisms underlying the therapeutic effects of hAMSCs in psoriasis we measured the serum levels of inflammatory factors. We found that hAMSCs significantly suppressed these levels and also reduced spleen enlargement in psoriasis-like mice. On the basis of these findings, we hypothesized that hAMSCs may exert their effects by modulating immune-related functions in these mice. First, we predicted the abundance of infiltrating immune cells in psoriasis tissues using the CIBERSORT and ssGSEA algorithms applied to the microarray data. We found differences in the levels of various T cells (such as CD4 memory-activated T cells, regulatory T cells, and activated CD4 T cells), B cells (such as naïve B cells, activated B cells, and memory B cells), and mast cells (activated mast cells) in psoriasis tissue (Figure 2A; Supplementary Figure 2A). These findings suggest significant immune dysregulation in psoriasis tissue compared with normal tissue. Next, we intersected the differentially expressed genes (DEGs) identified in the psoriasis microarray data with the immune-related dataset from the ImmPort database (21), resulting in 97 psoriasis immune-related differential genes (PIRGs) (Figure 2B), including 67 upregulated genes and 30 downregulated genes (Figure 2C; Supplementary Figure 2B). We performed Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses on the PIRGs (Supplementary Figures 2C-F) and observed enrichment of multiple pathways, including the IL-17 and cytokine signaling pathways.

We applied the LASSO regression model to screen the key PIRGs, which resulted in the identification of 14 key PIRGs (Figure 2D; Supplementary Figure 2G). We then constructed a protein–protein interaction (PPI) network for these 14 key PIRGs, and after removing isolated protein nodes, we identified a set of hub PIRGs—LYZ, MMP9, S100A9, WNT5A, ESR1, FOS, BACH2, AQP9, and BTC—that interacted with each other (Figure 2E). The expression and correlation of these genes were assessed in the transcriptome (Figures 2F, G). To validate the accuracy of these findings, we used two independent datasets, GSE30999 and GSE78097, which confirmed that the expression levels of these 9 hub PIRGs were significantly different between psoriasis and normal tissues (Supplementary Figure 3A, B). Furthermore, the construction of a reliable diagnostic nomogram model using these genes demonstrated their strong diagnostic potential (Supplementary Figure 3C, D). Taken together, these hub PIRGs may play pivotal roles in the pathogenesis of psoriasis.

To investigate whether hAMSCs act through these hub PIRGs, we examined the expression of 9 hub PIRGs in both the control and model groups of mice (Figures 2H–J; Supplementary Figure 3E–J). In the psoriasis mouse model, the expression of 6 hub PIRGs—LYZ, MMP9, S100A9, ESR1, FOS and BACH2—was significantly different from that in the control group, and their expression trends were consistent with those observed in the human psoriasis microarray data. Notably, the expression levels of MMP9 and S100A9 were significantly lower in the hAMSC-treated group than in the model group, whereas the expression of BACH2 was significantly greater in the treatment group than in the model group. Numerous studies have demonstrated that MMP9 and S100A9 are key biomarkers associated with psoriasis. Based on these findings, we hypothesize that hAMSCs may exert therapeutic effects through modulation of MMP9, S100A9, or BACH2 positive cell in psoriatic lesions.

Due to multiple safety and ethical constraints, it is challenging to directly employ hAMSCs in psoriatic patients to investigate their in vivo mechanisms. Therefore, we utilized single-cell RNA sequencing (scRNA-seq) data from psoriatic skin tissues to analyze changes in MMP9, S100A9, or BACH2 positive cells that might be affected by hAMSCs, aiming to predict the potential impact of hAMSCs in humans and their roles across different cell types. Based on the clinical information of patients, we selected 10 single-cell transcriptome datasets derived from five psoriatic skin lesions and their corresponding normal skin areas. After rigorous quality control and the exclusion of low-quality or doublet cells, a total of 88,657 cells were retained for downstream analysis. At a clustering resolution of 0.4, the cells were divided into 22 distinct clusters, each exhibiting considerable transcriptional heterogeneity. According to canonical marker genes reported in previous studies, we annotated 11 major cell types: Keratinocytes (KRT1, KRT14, KRT10, KRT5), Fibroblasts (DCN, LUM, COL1A2), Myeloid cells (LYZ, CD68, CD14), Pericytes (ACTA2, RGS5, PDGFRB, MCAM), Endothelial cells (KDR, VWF, SELE, CD93), Lymphatic endothelial cells (LYVE1, CCL21), Vellus hair follicle cells (SOX9, KRT6B, SFRP1), T cells (CD3D, CD3E, CD2), Mast cells (TPSB2, TPSAB1, CPA3, IL1RL1), Melanocytes (TYRP1, PMEL, MLANA), and B cells (CD79A, MS4A1) (Figure 3A; Supplementary Figures 4A, B).

To elucidate the transcriptional differences between psoriatic and normal tissues, we identified DEGs across cell clusters. The results revealed significant transcriptional alterations in both immune and non-immune cells of psoriatic tissues compared with normal controls (|logFC| > 0.25, adjusted P < 0.05) (Supplementary Figure 4C). Among all cell types, Keratinocytes, Pericytes, Fibroblasts, and Myeloid cells were most affected during psoriasis progression (Supplementary Figure 4D). GO enrichment analysis of upregulated DEGs showed strong enrichment in epidermal differentiation-related processes such as epidermis development, keratinocyte differentiation, epidermal cell differentiation, and cornified envelope formation (Supplementary Figure 4E), suggesting that multiple cell types contribute to altered keratinocyte differentiation and cornification. Conversely, downregulated genes were enriched in pathways related to cytosolic ribosome, cytoplasmic translation, and regulation of CD4-positive, alpha-beta T cell activation (Supplementary Figure 4E), indicating marked alterations in metabolism, translation, and immune regulation in psoriatic cells.

To further explore the cell type–specific expression patterns of S100A9, MMP9, and BACH2 in psoriasis, we systematically examined their single-cell expression profiles and compared the distribution of positive cells between psoriatic and normal tissues. The results demonstrated that S100A9 expression was predominantly enriched in keratinocytes (Figure 3B), and S100A9 positive cells were markedly increased in multiple cell types of psoriatic tissues, including Keratinocytes, Fibroblasts, T cells, and Myeloid cells (Figure 3C). Notably, nearly all keratinocytes from psoriatic lesions were S100A9 positive, suggesting a pronounced inflammatory activation state in these cells. MMP9 expression was mainly concentrated in myeloid cells (Figure 3D); however, compared with normal tissues, the proportion of MMP9 positive cells was reduced in psoriatic lesions (Figure 3E), while their overall expression level within myeloid cells was increased (Supplementary Figure 4F), implying that MMP9 may be highly upregulated in specific subclusters despite a lower overall frequency. In contrast, BACH2 expression was primarily localized to myeloid cells and T cells (Figure 3F), and the proportions of BACH2 positive cells within both populations were lower in psoriatic tissues than in normal controls (Figure 3G). We observed largely consistent results in the validation dataset compared with the discovery cohort. However, we unexpectedly noted an increased proportion of MMP9-positive cells within the myeloid compartment of psoriasis samples in the validation dataset (Supplementary Figure 4F). We speculate that this discrepancy may be attributable to differences in cell capture efficiency during the single-cell sequencing process, as the number of captured cells from psoriasis samples was relatively limited in the validation cohort, which could introduce proportional bias in specific cell subsets. Collectively, these results suggest that S100A9, MMP9, and BACH2 may play crucial roles in psoriasis pathogenesis by influencing the functions of keratinocytes, fibroblasts, T cells, and myeloid cells, providing important insights into their potential pathophysiological significance.

To further investigate the distribution characteristics and potential functions of S100A9, BACH2, and MMP9 positive cells within T cells, we performed subclustering of the T cell population and identified seven distinct subclusters: C1–TRM–KLRB1, C2–TSTR–FOS, C3–Treg–TIGIT, C4–CD8–CCL5, C5–CD8–IL17A, C6–NK/ILC, and C7–MKI67 (Figures 4A, B). The proportions of C2–TSTR–FOS and C4–CD8–CCL5 subclusters were markedly decreased in psoriatic tissues compared with normal controls (Figure 4C). Differential gene expression analysis revealed that C1–TRM–KLRB1 and C3–Treg–TIGIT subclusters exhibited the greatest transcriptional differences between psoriatic and normal T cells (Supplementary Figure 5A). KEGG pathway enrichment further demonstrated that the DEGs from both subclusters were significantly enriched in the IL-17 signaling pathway (Supplementary Figures 5B, C). Moreover, compared with normal controls, S100A9 expression was consistently upregulated across all T-cell subclusters in psoriatic tissues (Supplementary Figure 5D), suggesting that S100A9 may serve as an important marker of psoriatic T cells.

We found that S100A9 positive cells were predominantly distributed in the C1–TRM–KLRB1, C3–Treg–TIGIT, and C5–CD8–IL17A subclusters (Figure 4D), and these cells were more abundant in psoriatic T cells than in normal controls (Figure 4E). In contrast, MMP9 and BACH2 positive cells were less frequently detected across T cell subclusters (Supplementary Figures 5E, F), indicating that S100A9 may play a more prominent role in T cell biology during psoriasis. Notably, within the C5–CD8–IL17A cluster, IL17A expression was markedly elevated only in psoriatic cells (Figure 4F), and this was accompanied by strong co-expression of S100A9 (Figure 4G). Furthermore, differential expression analysis between psoriatic and normal C5–CD8–IL17A cells revealed that DEGs were significantly enriched in the IL-17 signaling pathway (Figure 4H). Collectively, these results indicate a close association between S100A9 positive T cells and IL-17 signaling in psoriasis. Importantly, further experiments showed that stem cell therapy markedly reduced the protein levels of S100A9 and IL17A in a psoriatic mouse model (Figure 4I), suggesting that stem cells may exert therapeutic effects in psoriasis partly by reducing the abundance of IL17A-high T cells.

We next analyzed the distribution and potential functions of S100A9, BACH2, and MMP9 positive cells within the myeloid compartment. A total of ten myeloid subclusters were identified: C1–Mye, C2–Mye, C3–Mye, C4–Mye, C5–Mye, C6–Mye, C7–Mye, C8–Mye, C9–Mye, and C10–Mye (Figure 5A; Supplementary Figure 6A). Among them, C1–Mye showed high expression of inflammation and chemokine related genes (IL1B, PTGS2, IL23A, CCL20) and monocyte markers (FCN1, APOBEC3A, THBS1), suggesting an inflammatory monocyte identity. C2–Mye expressed canonical cDC1 markers (CLEC9A), while C3–Mye was enriched for M2-like macrophage markers (CD163). Both C4–Mye and C5–Mye expressed cDC2 markers (CD1C), although C4–Mye additionally showed high expression of inflammation-related genes (TNFSF13B, AREG). C6–Mye expressed monocyte markers at higher levels but with weaker inflammatory signatures than C1–Mye. C7–Mye expressed stress-response genes (FOS, JUN), C8–Mye was characterized by mature dendritic cell activation markers (CCR7, FSCN1), C9–Mye displayed Langerhans cell markers (CLEC4A, CD1A), and C10–Mye expressed proliferative cDC1-related genes (MKI67, CLEC9A) (Figure 5B). In psoriatic tissues, C1–Mye, C2–Mye, C4–Mye, C8–Mye, C9–Mye, and C10–Mye were significantly enriched, whereas C5–Mye was notably reduced (Figure 5C).

Compared with normal skin, psoriatic samples showed widespread upregulation of S100A9 across myeloid subclusters; C3–Mye and C9–Mye displayed prominent upregulation of MMP9, while C1–Mye, C5–Mye, and C6–Mye showed MMP9 downregulation, and BACH2 expression remained largely unchanged (Figure 5F). S100A9 positive cells were predominantly localized in psoriatic lesions but were rarely detected in normal skin (Figure 5D; Supplementary Figure 6C). KEGG enrichment analysis revealed that DEGs from S100A9 positive cells were significantly enriched in the IL-17 signaling pathway (Supplementary Figure 6D). MMP9 positive cells were detected in both normal and psoriatic tissues; however, their proportions increased notably in C3–Mye, C7–Mye, C9–Mye, and C10–Mye, while decreasing in other subclusters (Figure 5D, E). Together with our previous findings showing elevated MMP9 expression in psoriatic patients and mouse models, these results suggest that C3–Mye and C9–Mye may play key roles in psoriasis pathogenesis. Differential gene expression analysis between MMP9 positive and negative cells showed that DEGs in C3–Mye were enriched in the IL-17 signaling, TNF signaling, and NF-κB signaling pathways, whereas DEGs in C9–Mye were enriched in Coronavirus disease–COVID-19, Transcriptional misregulation in cancer, and Lipid and atherosclerosis pathways (Figure 5G), indicating a core role of C3–Mye in inflammatory and immune regulation in psoriasis. BACH2 positive cells were scarce and observed in small numbers in both normal and psoriatic tissues (Supplementary Figure 6B). Importantly, immunohistochemistry confirmed that stem cell therapy markedly reduced MMP9 and TNFα protein levels in psoriatic mouse models (Figure 5H), suggesting that stem cells may exert therapeutic effects by modulating the TNF signaling pathway.

We next examined the distribution and potential functions of S100A9, BACH2, and MMP9 positive cells among non-immune cell populations. Fibroblasts were further subdivided into 12 transcriptionally distinct subclusters: C1–PTGDS, C2–COL18A1, C3–COL11A1, C4–C7, C5–WIF1, C6–CXCL3, C7–CYP1B1, C8–IGFBP2, C9–SFRP1, C10–COCH, C11–SERPINF1, and C12–MKI67 (Figure 6A, B). In psoriatic tissues, C1–PTGDS, C2–COL18A1, C7–CYP1B1, C8–IGFBP2, C11–SERPINF1, and C12–MKI67 were significantly enriched, whereas C3–COL11A1, C4–C7, C5–WIF1, C6–CXCL3, and C10–COCH were reduced (Figure 6C). S100A9 positive fibroblasts were mainly distributed in C2–COL18A1 and C7–CYP1B1 subclusters, both of which were enriched in psoriatic tissues (Figure 6D; Supplementary Figure 7A). MMP9 positive cells were also concentrated in these two subclusters (Supplementary Figure 7B), whereas BACH2 positive cells were primarily localized to C3–COL11A1 and were less abundant overall (Supplementary Figure 7C).

Functional enrichment analysis showed that C2–COL18A1 genes were significantly enriched in ECM–receptor interaction, PI3K–Akt signaling, and NF-κB signaling pathways, while C7–CYP1B1 genes were associated with cornified envelope formation, ECM–receptor interaction, and IL-17 signaling (Figure 6E). Comparison between psoriatic and normal fibroblasts revealed that C1–PTGDS and C9–SFRP1 had the largest number of DEGs (Figure 6F; Supplementary Figure 7D). GO enrichment of upregulated genes highlighted extracellular matrix remodeling and immune activation, including collagen-containing extracellular matrix, humoral immune response, and myeloid leukocyte migration, suggesting an activated state associated with inflammation and tissue repair. Downregulated genes were enriched in cytosolic ribosome and translation-related processes, indicating decreased biosynthetic activity (Supplementary Figure 7E).

Developmental trajectory analysis using CytoTRACE and pseudotime ordering revealed clear differentiation hierarchies among fibroblast subtypes. C12–MKI67 showed the highest developmental potential, suggesting a proliferative or progenitor-like state; C5–WIF1, C3–COL11A1, and C11–SERPINF1 represented intermediate stages, while C10–COCH and C9–SFRP1 displayed the lowest potential, representing terminally differentiated states (Figure 6G, H). Pseudotime analysis revealed a continuous differentiation spectrum, transitioning from early C5–WIF1 states to terminal C2–COL18A1, C8–IGFBP2, and C10–COCH states (Figure 6I). Dynamic gene expression analysis indicated stable expression of BACH2 and MMP9 throughout differentiation, whereas S100A9 expression markedly increased during late stages, suggesting its critical role in inflammatory fibroblast activation (Figure 6J). Overall, S100A9 positive fibroblasts were primarily enriched in C2–COL18A1 and C7–CYP1B1—two psoriasis-specific clusters strongly associated with inflammatory and immune pathways—implying a central regulatory role of S100A9 in fibroblast-mediated inflammatory remodeling in psoriasis.

Keratinocytes were further subdivided into five subclusters: C1–COL17A1, C2–NFKBIA, C3–S100A9, C4–MKI67, and C5–FOXC1 (Figure 7A, B). Compared with normal skin, C3–S100A9 and C4–MKI67 were markedly increased in psoriatic tissues, whereas C2–NFKBIA was significantly reduced (Figure 7C). Although MMP9- and BACH2 positive cells were relatively rare among keratinocytes, their distributions were distinct: MMP9 positive cells were mainly located in C1–COL17A1, C3–S100A9, and C4–MKI67, while BACH2 positive cells were enriched in the C2–NFKBIA subcluster of normal skin (Supplementary Figure 8B, C). S100A9 positive cells were widely distributed across keratinocyte subclusters—particularly C3–S100A9 and C4–MKI67—and to a lesser extent in C2–NFKBIA (Figure 7D; Supplementary Figure 8A). The NFKBIA gene encodes IκBα, a negative regulator of the NF-κB signaling pathway, which limits excessive inflammatory activation via negative feedback. We therefore hypothesized that the C2–NFKBIA cluster may play an anti-inflammatory regulatory role in psoriasis. Differential gene enrichment analysis supported this notion, showing that C2–NFKBIA DEGs were enriched in cornified envelope formation and negative regulation of canonical NF-κB signaling, while C3–S100A9 DEGs were enriched in epidermis development and the IL-17 signaling pathway (Figure 7E). These results highlight strong associations between C2–NFKBIA and C3–S100A9 with inflammation and immune responses.

Trajectory analysis revealed that C1–COL17A1 and C4–MKI67 had the highest developmental potential, indicating undifferentiated or proliferative states, while C5–FOXC1, C3–S100A9, and C2–NFKBIA represented more differentiated stages (Figure 7F; Supplementary Figure 8D). Pseudotime ordering showed that keratinocyte differentiation progressed from early C5–FOXC1 and C4–MKI67 stages toward terminal C2–NFKBIA states (Figure 7G). Dynamic gene expression analysis revealed stable BACH2 and MMP9 expression but a marked upregulation of S100A9 in late stages, suggesting its role in inflammatory keratinocyte activation.

Taken together, these results indicate that both fibroblasts and keratinocytes exhibit strong associations between S100A9 positive cell states and NF-κB–related signaling. We propose that hAMSCs may ameliorate psoriasis by modulating NF-κB activity. Consistently, immunofluorescence analysis of psoriatic mouse models showed that stem cell therapy markedly reduced nuclear translocation of p65 (Figure 7H, I), confirming that hAMSCs effectively suppress NF-κB pathway activation in psoriasis.

hAMSCs were provided by Jilin Zhongke Bio-engineering Joint Stock Co., Ltd. (Jilin, China), and cultivated in Showninm -ncMissionBasal Medium V3.0 (Shownin, China) at 37°C with 5 % CO2. In vivo injection into a mouse model and in vitro co-culture experiments using passage 4 hAMSCs.

Twenty-five female C57BL/6 mice (8–10 weeks; Vital River Experimental Animal Technical Co., Ltd., Beijing, China) were randomly divided into three groups: Control (control group, n=5), IMQ induced model (model group, n=10) and human amniotic mesenchymal stem cells (hAMSCs) treated model (treatment group, n=10). On day 0 (1 day before modelling), dorsal hairs were removed using depilatory cream. From day 1 to day 6, mice in the model group and treatment group received daily topical application of 62.5 mg 5% imiquimod cream to the depilated skin, while mice in the control group received equivalent amounts of petroleum jelly ointment. The treatment group received intravenous injection of hAMSCs (2×106 cells in 100 μL PBS) on day 0 and day 4. Daily photos were taken to observe the changes of skin lesions on the backs of mice. The mice were anesthetized with 2% isoflurane and subsequently euthanized by cervical dislocation, in strict accordance with ethical guidelines, to ensure a humane and painless death. The Animal Ethics Committee of Changchun Sci-Tech University also examined and approved all experimental procedures and techniques (Protocol No. CKARI202409). We housed all mice under standard laboratory conditions in temperature-controlled rooms and provided them with 12 h of light and 12 h of darkness every day. The animals were normally raised without any abnormal deaths. The work has been reported in line with the ARRIVE guidelines 2.0.

The PASI scoring system was used to assess the inflammatory status of the dorsal skin of the mice over an 8-day period. The three parameters evaluated were erythema (redness), induration (thickness), and desquamation (scaling). Each parameter was scored from 0 to 4 (0 = none, 1 = slight, 2 = moderate, 3 = marked, 4 = very marked), yielding a cumulative score ranging from 0 to 12. The evaluations were independently conducted by two researchers, and the mean scores were calculated.

hAMSCs were obtained from Jilin Zhongke Bio-engineering Joint Stock Co.,Ltd. (Changchun, Jilin, China), and HaCaT cells were provided by the School of Life Sciences, Jilin University. HaCaT cells were cultured in 6-well plates (Corning, 3516) at 37°C with 5% CO₂ in DMEM (Gibco, USA) supplemented with 10% fetal bovine serum (Gibco, USA) and 100 U/ml penicillin and 100 μg/ml streptomycin. The psoriatic keratinocyte model was induced by adding TNFα to the HaCaT culture medium at a final concentration of 30 ng/mL. Co-culture of hAMSCs with HaCaT cells using Labselect^® cell culture chambers (LABSELECT, 14211).

Paraffin-embedded skin samples were stained with H&E, and histological analysis was performed using the Baker scoring system. The following criteria were used: Stratum corneum: Small pustules were scored as 2 points, incomplete keratinization as 1 point, and hyperkeratinization as 0.5 points. Stratum epidermis: Disappearance of the granular cell layer in the epidermis was scored as 1 point, and hypertrophy of the stratum spinosum as 1 point. Stratum dermis: Capillary dilation was scored as 5 points, and single polymorphonuclear cell infiltration was scored as 2, 1, or 0.5 points, depending on severity. Immunohistochemistry was performed following standard protocols (41). Primary antibodies used included anti-S100A9 (Servicebio, GB111079) and anti-Ki-67 (Cell Signaling Technology, #90298).

Blood was collected from anesthetized mice via retro-orbital bleeding before euthanasia. Samples were centrifuged twice at 16,000 × g for 10 minutes, and the supernatant was stored at -80 °C for further analysis. Serum levels of IL-17, TNF-α, ALT, and AST were quantified using ELISA kits (Ready-SET-GO, eBioscience, San Diego, CA) according to the manufacturer’s instructions.

Tissue samples were snap-frozen and ground in liquid nitrogen. Total RNA was extracted using TRIzol (TransGen Biotech, ET101) following the manufacturer’s protocol. cDNA synthesis was performed using the TransScript^® One-Step gDNA Removal and cDNA Synthesis SuperMix TransScript (TransGen Biotech, AT311), with 2μg of total RNA as the template. RT-PCR was performed on a Applied Biosystems real-time PCR system using PerfectStart^® Green qPCR SuperMix (TransGen Biotech, AQ601). Supplementary Table S1 shows the primers used in this study. All experiments were conducted in triplicate.

Cell proliferation was assessed using the EdU assay. HaCaT cells (1×10⁵) were seeded in 12-well plates and incubated for 24 hours. EdU (10μM) reagent (Beyotime, C0071S) was added to each well for 2 hours. After three washes with PBS, cells were fixed with 4% paraformaldehyde (Sangon Biotech, E672002) for 15 minutes, permeabilized with 0.3% Triton X-100 (Sangon Biotech, A110694) for 15 minutes, and then incubated with a click-reaction reagent for 30 minutes in the dark. Nuclei were counterstained with Hoechst. The results were observed under a fluorescence microscope (Nikon ECLIPSE Ti-S) using NIS-Elements F v4.0 software.

ROS production was measured using a ROS detection kit (Beyotime, S0033S). Cells were incubated with 10 μM DCFH-DA for 30 minutes at 37 °C, and ROS levels were assessed by fluorescence microscopy.

Two GEO datasets, GSE13355 (42) and GSE14905 (43), were used as the training set for data analysis. Batch effects were eliminated using the R package limma (v3.62.1) (44). The GEO datasets GSE30999 (45) and GSE78097 (46) were used as validation sets. Immune-related gene data were obtained from the ImmPort database (21) (https://www.immport.org/Shared/) , totaling 1793 immune-related genes.

Data from GSE13355 and GSE14905 were combined and processed using limma to identify differentially expressed genes (DEGs) between normal and psoriasis samples, with the following cutoff: P < 0.05 and |logFC| > 1. The intersection of these DEGs with immune-related genes was visualized in Venn diagrams.

Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses of immune-related DEGs were performed using the R package clusterProfiler (47). Bubble plots for KEGG pathways and GO terms (biological process, cellular component, and molecular function) were generated using ggplot2.

Core genes were selected using LASSO regression (R package glmnet (48)). Protein-protein interaction (PPI) network analysis was performed using the STRING database (49) (https://string-db.org/) and visualized in Cytoscape (v3.10.0).

A nomogram model for predicting psoriasis occurrence was developed using the R package RMS. Calibration curves were plotted to assess the model’s predictive performance.

The single-cell transcriptome dataset GSE228421 (50) from the GEO database was analyzed using the R package Seurat (51) (v4.4.0). Data normalization, quality control, clustering, and differential gene expression analysis were performed. Quality control criteria included 500-4,000 genes detected per cell, total UMIs >1,000, and mitochondrial gene percentage <15%. Doublet cells were excluded using the DoubletFinder (v2.0.3) algorithm. Batch effects were corrected using Harmony (52) (v1.2.3). Principal component analysis (PCA) was performed, followed by clustering with resolutions ranging from 0.1 to 1.2. Thirteen clusters were identified (resolution = 0.1). The R package Libra (53) was used to construct pseudobulk and screen differentially expressed genes (|logFC|>0.25, P.Value_adj < 0.05). Fibroblasts, keratinocytes, T cells, and myeloid cell populations were subjected to a second round of clustering using the same analytical pipeline as described above. Cell clustering was performed using the FindNeighbors and FindClusters functions implemented in the Seurat package. The GEO datasets GSE230842 (54) were used as validation sets.

Cellular differentiation potential was quantified through transcriptional entropy analysis using CytoTRACE (v1.1.0), a computational framework for predicting developmental potency based on gene expression heterogeneity (55). Pseudotime trajectories were generated using the R package Monocle3 (56) (v1.3.7), with continuous changes in cell state and branching structures visualized using the plot_cell function.

Data were analyzed using GraphPad (v 10.0.2). Two-tailed t-tests were used to compare differences between groups. Quantitative data are presented as mean ± standard error of the mean (SEM) or standard deviation (SD), with statistical significance defined as p < 0.05.