C57BL/6J mice were purchased from LINGCHANG Biotech (Shanghai, China), which were bred and maintained in specific pathogen-free (SPF) units with controlled temperature (22 ± 2°C), relative humidity (50 ± 5%), artificial light (12 h light/dark cycle) and free access to food/water in the animal facilities of Tongji University, Shanghai, China. Age-matched male mice (6–8 weeks of age; 18–22 g) were randomly used for all experiments in a non-blind manner. Handling of mice and experimental procedures were approved by the Animal Care and Use Committee of Shanghai Tongji University.

Xiao-Yin-Fang (XYF) is an empirically developed Chinese medicine formula summarized and optimized by Chinese famous dermatologists Doctor Ming Chen and Doctor Jun Gu, which has been used to treat psoriasis in Shanghai Changhai Hospital. Its formula primarily includes five Chinese herbs: Isatis tinctoria L., Scutellaria baicalensis Georgi, Salvia miltiorrhiza Bunge, Sophora flavescens Aiton, and Rheum officinale Baill. with a weight ratio of 30:15:30:10:3 (Table 1). The standardized powder of each herb, which was extracted under optimum condition and corresponded to a standard dosage of raw herb, was purchased from LongHua Hospital, University of Traditional Chinese Medicine, Shanghai, China. Voucher specimen of herb powder was deposited in Shanghai Skin Disease Hospital with reference No. XYF01-05, respectively. XYF powder was dissolved in double-distilled H₂O, filtered through a 0.45 μm filter and stored at −20°C for subsequent application. All the procedures were in accordance with the rules and regulations of the 2015 Edition of China Pharmacopoeia.

Chemical profiling of XYF was performed on an Agilent 1290 UPLC System (Agilent Technologies Co., Santa Clara, CA, United States) coupled with AB SCIEX Triple TOF 4600^® quadrupole time-of-flight mass spectrometry (AB SCIEX, Foster City, CA, United States). Chromatographic separation was conducted on an Acquity UPLC^® HSS T3 Column (2.1 × 100 mm i.d., 1.8 μm; Waters, Milford, MA, United States) at 30°C, using a mixed mobile phase consisting of 0.1 formic acid in water (mobile phase A) and 0.1 formic acid in acetonitrile (mobile phase B). The gradient elution program was used as follows: 0–5 min, 0–0% B; 5–7 min, 0–3% B; 7–15 min, 3–5% B; 15–17 min, 5–16% B; 17–19 min, 16–17% B; 19–30 min, 17% B; 30–34 min, 17–20% B; 34–42 min, 20% B; 42–44 min, 20–22% B; 44–49 min, 22% B; 49–54 min, 22–60% B; 54–58 min; 60–95% B; 58–60 min, 95% B; 60–60.1 min, 95–0% B; 60.1–63 min, 0% B. The flow rate was 0.3 ml/min, and the sample injection volume was 1 μl. Data acquisition was performed on AB SCIEX Triple TOF 4600^® system equipped with an electrospray ionization (ESI) source in the negative and positive ion modes. The mass spectrometry was operated in full-scan TOF-MS (m/z 50-1700) and information-dependent acquisition (IDA) MS/MS modes. The parameters of mass spectrometry were as follows: ion source gas 1 and 2 were 50 psi; curtain gas was 35 psi; ion source temperature was 500°C; ion spray voltage floating was 5000 V (positive)/4500 V (negative). In the tandem mass spectrometry (MS/MS) experiments, mass range was 50–1250, collision energy was 40 ± 20 eV, ion release delay was 30 ms and ion release width was 15 ms. All data acquisition and processing were validated using Analyst TF 1.7.1 software (AB SCIEX).

For content determination, seven standards of representative components from five constitutive herbs of XYF, including (R, S)-goitrin (#8508), baicalin (#5719), salvianolic acid B (#4065), tanshinone II a (#7503), matrine (#844), oxymatrine (#8678), and emodin (#8171), were purchased from Standard Technology Co., Shanghai, China (http://www.nature-standard.com/). Quantitative analysis of these components in XYF by UPLC-DAD was performed on a Waters H-Class UPLC system (Waters, Milford, MA, United States) as previously reported (Xiang et al., 2018). UV detector was monitored at 225 nm (matrine/oxymatrine), 245 nm ((R, S)-goitrin), 254 nm (emodin), 270 nm (tanshinone II a), 280 nm (baicalin) and 286 nm (salvianolic acid B).

Mice received a daily topical dose of 50 mg of imiquimod (IMQ) cream (5%; #H20030128, Sichuan Med-Shine Pharmaceutical) or control Vaseline (VAS; #180102, Shandong Mint) on shaved back (an approximate size of 2 × 3 cm²) or bilateral ears for five consecutive days. For the induction of disease relapse, 25 mg of IMQ was applied on mouse left ear once daily from day 0 to day 4, and 25 mg of IMQ was reapplied on mouse right ear once daily from day 12 to day 16. Mouse skin inflammation was evaluated by cumulative psoriasis area and severity index (PASI) score, calculated by the adding up of erythema score (0–4), infiltration score (0–4), and desquamation score (0–4). The severity of each symptom was assessed comparing with reference pictures. Skin thickness was recorded as the average value of three measurements by vernier calipers at the center of mouse dorsal or ear lesion before topical treatment.

Mice were randomly allocated into the following groups: control group, model group, low-dose XYF group, medium-dose XYF group, high-dose XYF group, and multi-glycoside of Tripterygium wilfordii Hook. f (GTW) group. GTW tablets (#Z42021212, Hubei Huangshi Yunfei Pharmaceutical Company) were ground down and dissolved in ddH₂O, which were filtered and stored at −20 °C. Based on drug-dose conversion between human and mouse, 200 μl of XYF decoction (low-dose: 5.8 mg/g; medium-dose: 11.6 mg/g; high-dose: 23.3 mg/g), GTW (7.89 μg/g) or ddH₂O was administered once daily by gavage for 10 successive days, and VAS or IMQ was topically applied from day 6 to day 10.

Paraffin-embedded skin specimens were prepared by routine methods, and the sections were stained by hematoxylin-eosin (H&E) with additional immunostaining for Ki-67. The sections were deparaffinized with xylene and rehydrated through the incubation with graded alcohol into water. For H&E staining, the sections were then stained with hematoxylin, washed with PBS, differentiated with hydrochloric acid ethanol, and stained with eosin. Pathological change was examined under the Olympus CX33/BX53 optical microscope (Olympus, Southborough, MA, United States). Epidermal hyperplasia (acanthosis) was measured as the average length between the basement membrane and the stratum corneum. Papillomatosis index is the ratio of the length of the dermal-epidermal junction to the surface length of the epidermis. For Ki-67 staining, heat-induced epitope retrieval was performed in EDTA buffer (#RC016, RecordBio) at pH 8.3% hydrogen peroxide was utilized to block endogenous peroxidase activity. The sections were then incubated with anti-Ki-67 antibody (#ab16667, Abcam) overnight at 4°C, followed by incubation with REAL EnVision and visualized by DAB⁺ (#K5007, DAKO). Ki-67 staining was evaluated semi-quantitatively according to the expression level of cytoplasmic brown staining in five random epidermal fields under 400x magnification performed independently by two researchers. The intensity of staining (IS, intensity score) was assessed as: absent (0), weak (1), moderate (2), and strong (3). The percentage of stained cells (PS, proportion score) was scored: 0–5% (0), 6–25% (1), 26–50% (2), 51–75% (3), and 75–100% (4). H-score was computed as the average value of the multiplication of IS and PS and interpreted by the following way: 0 as negative staining, 1 to 4 as weak positive staining, 5 to 8 as medium positive staining and 9 to 12 as strong positive staining.

IL-17 levels in the serum samples collected from experimental mice were measured using mouse IL-17A ELISA kit (RayBio, Norcross, GA, United States) according to the manufacturers’ instructions.

Mouse ears were harvested and incubated in dispase II (5 mg/ml; #D4693, Sigma) for 1 h at 37°C. Dermal sheet was peeled from epidermal sheet, cut into little pieces, and digested at 37°C in DMEM (#SH30022.01, Hyclone) containing collagenase IV (1 mg/ml; #SH30256.01B, Hyclone) and DNase (0.015 mg/ml; #B002004, Diamond) for one and a half hour. Dermal single-cell suspension was harvested by passing through a 40 μm cell strainer (#CSS-010-040, Biofil).

T cell culture medium was RPMI 1640 (#11875119, Gibco) supplemented with heat-inactivated fetal bovine serum (10%; #SH30084.03, Hyclone), penicillin G (100U/ml; #B540733, Sangon), streptomycin sulfate (100 μg/ml; #B540733, Sangon) and amphotericin B (2.5 μg/ml; #B540733, Sangon). Mouse skin-draining axillary, brachial and inguinal lymph nodes (LN) were harvested. LN single-cell suspensions were prepared through grinding by glass slides and filtering through a 40 μm cell strainer. LN and dermal cells were cultured in T cell culture medium in the presence of phorbol-12-myristate-13-acetate (PMA; 50 ng/ml; #P1585, Sigma), ionomycin (1mM; #abs42019871, Absin) and brefeldin A (1:1000; #347688, BD) for 4 h.

Single-cell suspensions were preincubated with anti-CD16/CD32 antibodies (Ab; clone 2.4G2; #553141, BD Biosciences) for 15 min at 4°C. To discriminate viable cells from dead cells, cells were stained with fixable viability stain (#564406, BD Biosciences) for 10 min at room temperature. For the analysis of surface markers, cells were stained with anti-CD3 Ab (clone 145-2C1; #553061, BD Biosciences), anti-CD4 Ab (clone GK1.5; #562891, BD Biosciences), anti-CD25 Ab (clone 3C7; #101904, Biolegend), anti-CD45 Ab (clone 30-F11; #103134, Biolegend), anti-CD11b Ab (clone M1/70; #553310/#564455, BD Biosciences), anti-TCRβ Ab (clone H57–597; #560657, BD Biosciences), anti-γδTCR Ab (clone GL3; #118115, Biolegend), anti-MHC-II Ab (clone M5/114.15.2; #563414, BD Biosciences), anti-F4/80 Ab (clone T45-2342; #565410, BD Biosciences), anti-CD11c Ab (clone HL3; #566505, BD Biosciences), anti-Ly6C Ab (clone AL-21; #553104, BD Biosciences), and anti-Ly6G Ab (clone 1A8; #560599, BD Biosciences) for 30 min at 4°C protected from light. For cytoplasmic staining, cells were fixed and permeabilized with BD Cytofix/Cytoperm (#51-2090KZ, BD Biosciences) and were stained with anti-mouse IFN-γ Ab (clone XMG1.2; #505810, Biolegend), anti-mouse IL-4 Ab (clone 11B11; #560699, BD Biosciences), anti-mouse IL-17 Ab (clone TC11-18H10; #559502, BD Biosciences) and incubated for 30 min at 4°C. For intranuclear staining, cells were fixed and permeabilized with Foxp3/Transcription Factor Fixation/Permeabilization Concentrate and Diluent (#00-5521-00, eBioscience), and were then stained with anti-mouse Foxp3 Ab (clone MF-14; #126408, Biolegend), anti-Ki-67 Ab (clone B56; #556027, BD Biosciences) and anti-ROR-γt Ab (clone Q31-378; #564723, BD Biosciences). For Annexin-V staining, cells were resuspended in 1x annexin-binding buffer (#556454, BD Biosciences) and stained with Annexin-V (#640908, Biolegend) for 15 min at room temperature. Finally, cells were assayed with BD LSRFortessa Cytometer, and analyzed with FlowJo software (Treestar). Gate strategy of LN CD4⁺ T cells, LN γδT cells, dermal neutrophils, dermal inflammatory monocytes, dermal CD4⁺ T cells, and dermal γδT cells were demonstrated in Supplementary Figure S3.

Total RNA samples were prepared from intact epidermal or dermal ear sheets of mouse ear skin from two mice. Double-strand cDNA was generated from equal amounts of total RNA by following TruSeq8 RNA Library Prep Kit v2 (#RS-122-2001/2002, Illumina). The cDNA libraries were sequenced using Illumina Hi-seq2500. STAR software was utilized for sequence alignment between the preprocessing sequence and reference genome sequence of mice downloaded from the Ensembl database (Mus_musculus.GRCm38.90,ftp://ftp.ensembl.org/pub/release-90/gtf/mus_musculus/Mus_musculus.GRCm38.90. chr.gtf.gz). Transcript assembly of mRNA sequencing data was performed by StringTie software. DESeq 2 was applied to conduct the analysis of differentially expressed genes (DEG). The cutoffs of DEG were determined as the adjusted P value ≤0.05 and the |log₂FC| ≥ 1. Functional annotations of the DEGs were conducted using Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways analysis and Gene Set Enrichment Analysis (GSEA).

GraphPad Prism 6.0 software was utilized for statistical analysis. Student's t-test or one-way ANOVA were utilized to analyze the differences between the groups following Gaussian distributions with homogeneity of variance. Data that did not follow Gaussian distributions were analyzed using the Kruskal-Wallis test. Differences were statistically significant when P < 0.05.

A UPLC-Q/TOF-MS method in both negative and positive ion modes was employed to rapidly characterize the major constituents in XYF. A total of 57 compounds were unambiguously or tentatively characterized by comparing their retention times and MS data with reference standards or with data reported in the literature (Supplementary Figure S1). These compounds were all derived from five medicinal materials that composed XYF, and did not contain any conventional immunosuppressants. Seven representative chemical compounds were chosen as chemical markers and quantified to evaluate the quality of relevant medicinal materials, including (R, S)-goitrin (1, Rt 9.928 min), baicalin (2, Rt 35.44 min), salvianolic acid B (3, Rt 38.63 min), tanshinone II a (7, Rt 58.63 min), matrine (5, Rt 11.03 min), oxymatrine (6, Rt 13.75 min), and emodin (4, Rt 56.12 min) (Figure 1). Quantitative determination of these seven compounds by UPLC-DAD was performed on a Waters H-Class UPLC system at different wavelengths. As a result, the contents of these seven compounds in XYF were 0.132 mg/g, 12.626 mg/g, 6.154 mg/g, 0.014 mg/g, 2.514 mg/g, 1.757 mg/g, and 0.007 mg/g, respectively.

To explore the potential mechanism of XYF in psoriasis, we utilized IMQ-induced psoriasis-like dermatitis mouse model. Mice were divided into six groups and treated as depicted in Figure 2A. Multi-glycoside of Tripterygium wilfordii Hook. f (GTW), also termed as Tripterygium glycosides, is a widely acknowledged efficacious psoriasis-treating TCM agent and has been proven to ameliorate murine psoriasis-like dermatitis, which was therefore chosen as positive drug in our study (Han et al., 2012; Wu et al., 2015a; Zhao et al., 2016; Lv et al., 2018; Nguyen et al., 2020a; Ru et al., 2020). As modeling with IMQ generated skin lesions resembling human plaque psoriasis, mice pretreated with XYF displayed lighter erythema, smoother skin, and thinner scales than mice treated with IMQ alone (Figure 2B). Moreover, XYF exerted therapeutic effects in a dose-dependent manner that high-dose XYF was as potent as GTW in alleviating psoriasis-like dermatitis (Figure 2B). These findings were verified by the evaluation of PASI score (Figure 2C) and back skin thickness (Figure 2D). Besides, the application of XYF did not cause any behavioral abnormality or extra weight loss in mice (Figure 2E). Subcutaneous vessel dilation was also less prominent in the mice from high-dose XYF group than model group (Supplementary Figure S2). In accordance, the histological examination showed that both high-dose XYF and GTW significantly reduced epidermal acanthosis and elevated papillomatosis index in an equal manner (Figures 2F–H). As the expression of Ki-67 in keratinocytes was upregulated in model mice, high-dose XYF group considerably decreased Ki-67⁺ H-score to a similar degree as GTW (Figures 2I,J). Overall, XYF therapy substantially alleviated IMQ-induced psoriasis-like skin inflammation when administered at the high dosage, which demonstrated a comparable efficacy as GTW. Therefore, high-dose XYF was employed in the following study.

Since T lymphocytes play a vital role in the pathogenesis of psoriasis, we next sought to investigate the impact of XYF on T cell subsets within skin-draining lymph nodes (LN). Compared with control mice, we detected a decreased ratio of LN CD4⁺ T cells with greater IL-17 production in model mice (Supplementary Figure S4A–D). The proportion of regulatory T (Treg; CD25⁺Foxp3⁺) cells in CD4⁺ T cells was also increased (Supplementary Figure S4E). However, neither XYF nor GTW disturb the balance in Th and Treg cells (Supplementary Figure S4). While serum IL-17 levels were upregulated in model mice, the application of XYF reduced the serum contents of IL-17 (Supplementary Figure S5). Previous research demonstrated that γδT cells are the main source of IL-17 in psoriasis-like dermatitis (Cai et al., 2011). Accordantly, the percentage and number of LN γδT cells were substantially elevated in model mice (Figure 3A), and their secretion of IL-17 was raised approximately threefold (Figure 3B). Although XYF failed to suppress the expansion of LN γδT cells (Figure 3A), it drastically hindered their polarization into γδT17 cells to a comparable extent as GTW (Figure 3B). Hence, XYF therapy might exert its anti-inflammatory role in psoriasis-like dermatitis mainly through its inhibition of γδT17 polarization.

Given that skin-resident γδT cells elicited direct action in local inflammation, we established the model of psoriasis-like dermatitis on mouse ears to assess the role of dermal γδT cells in the curative effects of XYF (Figure 4A). As depicted in Figures 4B–H, XYF significantly reduced the disease severity of psoriasis-like dermatitis on mouse ears. While neutrophils (Ly6C⁺Ly6G⁺) and inflammatory monocytes (Ly6C⁺Ly6G⁻) gathered in the dermis following IMQ application, XYF considerably lowered their cell numbers rather than their percentages (Figures 4I–K), corroborating the anti-psoriasis effect of XYF. In line with the findings of LNs, XYF did not affect the homeostasis and function of dermal conventional T cells (Supplementary Figure S6).

Previous studies have uncovered that dermal γδT cells were predominantly γδ^int T cells with a minor population being γδ^high T cells (Cai et al., 2011). And, dermal γδ^int T cells were the main source of IL-17 in psoriasis-like dermatitis, whereas γδ^high T cells barely produced IL-17 (Cai et al., 2011; Riol-Blanco et al., 2014). Hence, we focused on dermal γδ^int T cells in the following research. While dermal γδ^int T cells were expanded in model mice, XYF diminished the quantity of γδ^int T cells (Figures 5A,B). The expressions of Ki-67 and Annexin-V in γδ^int T cells were comparable between model and XYF group (Figures 5C,D), suggesting that XYF did not influence γδ^int T cell survival. Thus, the decrease in γδ^int T cells triggered by XYF might result from its impact on cell trafficking. Remarkably, while IL-17 secretion by γδ^int T cells tripled in model mice versus control mice, XYF significantly repressed γδ^int T cell production of IL-17 (Figure 5E). In accordance, the expression of ROR-γt, which was a house-keeping transcription factor of γδT17 cells, was enhanced in γδ^int T cells from model mice, whereas XYF considerably downregulated its expression (Figure 5F). In total, XYF might ameliorate psoriasis-like skin inflammation mainly through hampering dermal γδ^int T cell trafficking and suppressing their polarization into γδT17 cells.

As psoriasis has a notable propensity for recurrence, we exploited a relapsing model of psoriasis-like dermatitis to examine the therapeutic effects of XYF in psoriasis recurrence by applying IMQ cream on left ear and reapplying them on right ear one week later (Supplementary Figure S7) (Ramirez-Valle et al., 2015). As expected, the restimulation with IMQ exacerbated the severity of psoriasis-like dermatitis, which was successfully lessened by the treatment with XYF (Figure 6A). Evaluations of PASI score, ear swelling, epidermal acanthosis and Ki-67 H-score confirmed the therapeutic effects of XYF (Figures 6B–G). Consistently, the secondary application of IMQ further augmented γδ^int T cells and increased their IL-17 production, whereas XYF decreased γδ^int T cells and suppressed their secretion of IL-17 (Figures 6H–J). In sum, XYF might be beneficial to prevent the relapse of psoriasis through impeding γδ^int T cell reactivation and their polarization into γδT17 cells.

To further explore the potential mechanisms of XYF therapy, RNA sequencing analysis was performed (Supplementary Table S1). Heatmap analysis of differentially expressed genes (DEG) revealed distinct transcriptomes between different groups (Figures 7A,B). And, XYF partially reversed the pathological transcriptional alternations in psoriasis-like dermatitis (Figures 7A,B). Detailed comparison identified the overlapping genes of DEGs between control/model group and DEGs between model/XYF group (Figure 7C, Supplementary Table S2, S3). Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis uncovered that epidermis from model group was enriched in cytokine-cytokine receptor interaction, PI3K-Akt signaling pathway, IL-17 signaling pathway, and cGMP-PKG signaling pathway whereas the dermis from model group was enriched in cytokine-cytokine receptor interaction, chemokine signaling pathway, IL-17 signaling pathway, TNF signaling pathway, Wnt signaling pathway, and PI3K-Akt signaling pathway when compared with control group (Figures 7D,F). XYF therapy modulated PI3K-Akt signaling pathway, Calcium signaling, Hippo signaling pathway and cGMP-PKG signaling pathway in the epidermis, whereas it affected several metabolic pathways, Calcium signaling, cAMP signaling pathway and cGMP-PKG signaling pathway in the dermis (Figures 7E,G). Further Gene Set Enrichment Analysis (GSEA) uncovered XYF downregulated focal adhesion and ECM receptor interaction in the epidermis, whereas disturbed arachidonic acid metabolism in the dermis (Supplementary Figure S8). Altogether, these results suggested that XYF might regulate multiple inflammatory signaling pathways and metabolic processes, which impact on γδT17 cell biology awaits future study.