3-(4,5-Dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT), concanavalin A (Con A), lipopolysaccharide (LPS), ovalbumin (OVA), type II collagenase, and rabbit anti-mouse horseradish peroxidase (HRP)-IgG were purchased from Sigma-Aldrich, Co., MO, USA; goat anti-mouse HRP-IgG1, IgG2a, and IgG2b were from Southern Biotech. Assoc., AL, USA; RPMI was from Hyclone, Utah, USA; fetal bovine serum (FBS) was from Gibco, Grand Island, NY, USA; DMEM medium was from Corning, NY, USA; mouse ELISA detecting kits were from Wuhan Boster Biological Technology co., Ltd., Hubei, China; TRIzol reagent was from Ambion, Austin, TX, USA; reverse transcriptase, oligo(dT)₁₈, and ribonuclease inhibitor were from Shanghai Sangon Biotech Co., Ltd, China; deoxyribonuclease (DNase) I was from Roche Diagnostics, IN, USA; anti-mouse CD3, CD28, CD16/CD32, CD3e–APC, CD4–FITC, CD8–FITC, PE-labeled cytokines, Ly-6G (Gr1)–PE-Cy5, Ly-6C–APC, CD11c–PE, CD3e–PE-Cy5, F4/80–APC, CD117 (c-Kit)–PE-Cy5, and FceR1–APC mAbs were from eBioscience, San Diego, CA, USA; CD45R–PE was from BioLegend, San Diego, CA, USA; Siglec-F–PE was obtained from BD Pharmingen, San Diego, CA, USA; Alexa Fluor 488-conjugated ovalbumin was from Invitrogen, Carlsbad, CA, USA; bicinchoninic acid (BCA) protein assay kit, enhanced chemiluminescence (ECL) kit, HRP-conjugated goat anti-rabbit and anti-mouse IgG (H+L), and radioimmunoprecipitation assay (RIPA) lysis buffer were acquired from Beyotime Biotech, Nantong, Jiangsu, China; SYK inhibitor R406, LYN inhibitor SU6656, and STAT3 inhibitor S3I-201 were from Selleck Chemicals, Houston, TX, USA; FastStart universal SYBR Green Master, the phosphatase inhibitor cocktail and protease inhibitor cocktail were acquired from Bimake, Houston, TX, USA; rabbit anti-mouse SYK, STAT3, phospho-SYK, and phospho-STAT3 mAbs were from Cell Signaling Technology, Danvers, MA, USA; anti-mouse TATA-box binding protein (TBP) mAb was from Proteintech, Chicago, IL, USA; blue plus IV protein marker (DM131) was from TransGen Biotech, Beijing, China. The SurePrint G3 8 × 60 K mouse gene expression microarray was provided by Agilent Technologies, Santa Clara, CA, USA.

Rhizoma Bolbostemmatis was purchased from Hangzhou Huqing Yutang Pharmaceutical Co., Ltd (Zhejiang, China) and was authenticated as the dry tuber of B. paniculatum. The drug (1 kg) was powdered and then extracted with 70% ethanol under reflux twice. After filtration through 3MM Whatman filter paper (thickness: 0.34 mm), excess solvent was recovered under reduced pressure. The ethanol extract was subjected to D101 resin washed with water and sequentially eluted with 30%, 50%, and 70% ethanol as elution to obtain three fractions. The yields of three fractions were 0.82%, 2.62%, and 0.53% of the dried crude drug (w/w) for BPS30, BPS50, and BPS70, respectively. BPS50 (1.0 g) was dissolved in 5 mL of 50% ethanol and then passed through a 0.45-μM filter. The solution was applied to SinoChrom ODS-BP column (250 mm × 20 mm, 10 μm) using 43% ethanol as mobile phase and UV230-II UV-Vis detector on Elite P270II preparative HPLC instrument (Dalian, China) to give TBM I (510.7 mg), TBM II (98.9 mg), and TBM III (190.4 mg).

Each compound was identified by high resolution electrospray ionization mass spectroscopy (HR-ESI-MS) on OrbitrapElite Mass spectrometer (Thermo Scientific, Bremen, Germany) and nuclear magnetic resonance spectroscopy (NMR) on Agilent DD2-600 NMR spectrometer. The quasi-molecular ions at m/z 1317.6167, 1333.6135, and 1363.6233 ([M–H]^–) were in agreement with the molecular formula C₆₃H₉₈O₂₉ (calc. 1318.6174), C₆₃H₉₈O₃₀ (calc. 1334.6143), and C₆₄H₁₀₀O₃₁ (calc. 1364.6249) for TBM I, TBM II, and TBM III, respectively ( Figure 1 ). Their NMR data were listed in Table S1 .

sICR and C57BL/6J mice aged 5–6 weeks were purchased from Shanghai SLAC Laboratory Animal Co., Ltd, China. All experiments complied with China legislation on laboratory animals and were authorized by the Experimental Animal Ethics Committee of Zhejiang University.

The hemolytic activity of TBMs and Quil A was examined using 0.5% rabbit red blood cell (RBC), and the results were expressed by the concentration resulting in 50% of the maximum hemolysis (HD₅₀) using NDST software (20).

ICR mice were divided into six groups with 5 mice per group. Animals were subcutaneously (s.c.) injected with TBM I, TBM II, and TBM III at a single dose of 1.0 mg or Quil A (150 and 200 μg) in 0.2 mL saline, and then monitored for body weight, swelling and loss of hair at the injection site, and lethality daily for 1 week (21).

Mice were divided into 12 groups with 5 mice per group. Mice were s.c. injected with OVA alone (25 μg) or in combination with TBMs (25, 50, or 100 μg) or Quil A (10 μg) in 200 μL PBS. An intensive immunization was given 14 days later. Animals injected with PBS served as the control. Splenocytes and serum were prepared 2 weeks after the second inoculation.

C57BL/6J mice were intramuscularly (i.m.) injected in the quadriceps tissues on one leg with 50 μg TBM I dissolved in 50 µL of PBS and on the contralateral leg with 50 μL PBS as control. Mice were sacrificed at the indicated time points, and then the quadriceps muscle tissues were taken from four mice per group. For enzyme-linked immunosorbent assay (ELISA), the muscle tissues were homogenized using Tissue Ruptor Disposable probes (QIAGEN, Germany) in 1 mL PBS with protease inhibitor cocktail. After centrifugation at 12000 rpm for 10 min at 4 °C, the protein concentrations in the supernatants were detected by the BCA assay kit. For the real-time quantitative PCR (RT-qPCR), the muscle tissues were homogenized with 1 mL TRIzol reagent using Tissue Ruptor Disposable probes (19).

Serum antigen-specific IgG antibody and its isotype titers in OVA-immunized mice were determined by an indirect ELISA (20).

Splenocytes (5 × 10⁵ cells/well) were seeded into 96-well cell culture plates, and then incubated with Con A (5 μg/mL), LPS (10 μg/mL), OVA (20 μg/mL), or RPMI for 48 h at 37°C and 5% CO₂. Four hours before the end, the cell proliferation was detected using MTT assay (20).

Splenocytes (1 × 10⁶ cells/well) and human leukemia K562 cells (2 × 10⁴ cells/well) were seeded in 96-well U-bottom microtiter plates in RPMI complete medium and then incubated for 20 h at 37°C in 5% CO₂. The killing activities of NK cells in splenocytes against K562 cells were assured by MTT assay (20).

Splenocytes (5 × 10⁶ cells/well) were stimulated with OVA (20 μg/mL) in 24-well cell culture plates for 72 h at 37°C and 5% CO₂. The levels of cytokines and chemokines in the splenocyte culture supernatants and mouse muscle tissue homogenates were detected using commercial ELISA kits as previously described (19). For quadriceps muscle tissues, the values were expressed as pg/mg protein.

Splenocytes (5 × 10⁶ cells/well) were stimulated with OVA (20 μg/mL) in 24-well cell culture plates for 18 h at 37°C and 5% CO₂. The total RNA was isolated with TRIzol reagent and reverse transcription was performed. The PCR was conducted on a Bio-Rad CFX 96 system with specific primers ( Table S2 ) using SYBR Green qPCR Master Mix (19). The expression levels of the tested genes relative to GAPDH were determined using the 2^-ΔΔCt method.

ICR mice were s.c. injected at the left hind limbs with OVA alone or in combination with TBMs (25, 50, or 100 μg) or Quil A (10 μg) for sensitization. Five days later, two hind footpads were s.c. injected with and without OVA (10 μg) in 50 μL PBS, respectively. The thickness of footpad was measured using a vernier caliper 24 h after the challenge. The difference value in left and right footpad thickness was represented as the footpad swelling (20).

Splenocytes (5 × 10⁵ cells/well) were incubated in 96-well cell culture plates with OVA (20 μg/mL) and anti-CD28 antibody (5 μg/mL) for 12 h at 37°C in 5% CO₂. Cells incubated with anti-CD28 alone or in combination with anti-CD3 antibody served as the negative and positive control, respectively. Brefeldin A was added to each well for an extra 4 h. The pelleted cells were blocked with anti-mouse CD16/CD32 antibody on ice for 10 min. Followed by washing with 5% FCS-PBS, CD3e–APC and CD8α–FITC or CD3e–APC and CD4–FITC were added, and the mixture was incubated for 15 min at 25°C. After washing, cells were fixed with IC fixation buffer at 25°C for 30 min in the darkness. Permeabilization buffer and anti-mouse IL-2–PE, IL-4–PE, IL-10–PE, or IFN-γ–PE were added, and incubated at 25°C for 30 min in dark (22). The stained cells were analyzed on BD FACSCanto II flow cytometer (BD Biosciences) using FlowJo vX.0.7 (Tree Star).

C57BL/6J mice were injected i.m. with 25 μL/quadriceps muscle of OVA-AF488 (10 μg) alone or in combination with TBM I (25, 50, or 100 μg). Quadriceps muscles and draining lymph nodes (dLNs) were harvested from four mice per group at 12, 24, and 48 h after administration, respectively. Quadriceps muscles were cut into small pieces and digested with 100 µg/mL DNaseI, 0.05% type II collagenase, and 1% FCS in DMEM for 30 min at 37 °C. After centrifugation, the pelleted cells were resuspended in DMEM and then filtered through a 70 μm nylon mesh to obtain a cell suspension. Cell suspension was centrifuged and washed with PAB (1% FBS in PBS). The dLNs were homogenized within PBS using glass homogenizers gently. After centrifugation, the pelleted cells were washed with PAB. The cells were blocked with 1 μg of purified anti-mouse CD16/CD32 antibody for 10 min to inhibit non-specific staining, and then stained with combinations of anti-mouse Ly-6G–PE-Cy5, Ly-6C–APC and CD11c–PE, or CD3e–PE-Cy5, F4/80–APC and CD45R–PE, or CD117–PE-Cy5, FceR1–APC and Siglec-F–PE at room temperature for 30 min (19). The stained cells were analyzed using a FACSCanto II Flow cytometer. Data analysis was performed with FlowJo vX.0.7.

The quadriceps muscles harvested from three mice per group at 4 h after injection were subjected to transcriptomic analysis using Agilent SurePrint G3 mouse gene expression microarray as previously described (19). The microarray data set (GSE205650) has been deposited in the GEO database (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE205650). Differentially expressed genes (DEGs) were selected by P-value<0.05 and fold change >2. The Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses were performed using Metascape (http://metascape.org/gp/index.html#/main/step1) (23).

The whole genes detected in the quadriceps muscles were subjected to GSEA Software (version 4.2.1) using GSKB database. Leading-edge subsets (LESS) were screened based on |Normalized enrichment score (NES)| > 1, P < 0.05, and false discovery rate (FDR) < 0.25. Multiple GSEA plots were produced by ‘plyr’, ‘ggplot2’ and ‘grid’ packages (24). The genes in each LESS were integrated as core genes. The protein-protein interaction (PPI) network of core genes was constructed using STRING database version 11.5 (https://cn.string-db.org/). The hub genes were identified by using the cytoHubba plug-in of Cytoscape with seven common algorithms (MCC, MNC, Degree, Closeness, Radiality, Stress, and EPC). TRRUST database (https://www.grnpedia.org/trrust/) was used to predict the upstream transcription factors (TFs) of the core genes, and an adjusted P < 0.05 was considered significant (25).

PubChem (https://pubchem.ncbi.nlm.nih.gov/), Swiss target prediction (http://www.swisstargetprediction.ch/), and SuperPred database (https://prediction.charite.de/s) were used to predict the potential targets of TBM I. The common genes between the predicted targets and core genes from GSEA were considered as candidate targets of TBM I. The PPI network was constructed using STRING database version 11.5 (https://cn.string-db.org/). The plug-in molecular complex detection technology (MCODE) in Cytoscape 3.9.1 was used to analyze pivotal functional modules by the following criteria: K-core = 2, degree cutoff = 2, max depth = 100, and node score cutoff = 0.2. Subsequently, a co-expression network of these key targets involved in the module was constructed using GeneMANIA (http://www.genemania.org/) (26).

The molecular docking simulation was performed using AutoDockTools (4.2.3) to verify the credibility of the hub targets as previously described (27). The mechanical structure of TBM I was optimized via Chem3D (21.0). The 3D structures of the proteins were obtained from the PDB database (http://www.rcsb.org/pdb/home/home.do). The ligand-receptor binding property was analyzed. Binding energies < -1.2 kcal/mol denoted the feasible docking, and the results were visualized using Pymol software.

The mouse C2C12 myoblast cell line was purchased from the cell bank of the Shanghai Branch of the Chinese Academy of Sciences and maintained in a 5% CO₂ atmosphere in DMEM complete medium supplemented with 10% FBS, 100 µg/mL streptomycin, and 100 U/mL penicillin. C2C12 cells were plated 24 h before stimulation and were incubated with TBM I at various concentrations for the indicated times. Then, the cells and culture supernatants were collected for detecting cell proliferation, mRNA and protein expression levels using MTT assay, ELISA, RT-qPCR, and Western blotting, respectively.

After pretreatment with R406 (SYK inhibitor, 8 µM), SU6656 (LYN inhibitor, 2 µM), or S3I-201 (STAT3 inhibitor, 100 µM) for 1 h, C2C12 cells were stimulated with TBM I (40 µM) for 4 h, 8 h, or 12 h. The cells and supernatants were collected for the mRNA and protein expression levels by RT-qPCR, ELISA, and Western blotting.

All results were presented as mean ± SD and subjected to ANOVA and Student’s t-tests for calculating the statistical significance of difference, and P < 0.05 was identified as statistically significant.

The levels of endotoxin in the solution of three TBMs (10 mg/mL) were determined to be lower than 0.5 endotoxin units (EU)/mL using tachypleus amebocyte lysate assay. Therefore, these samples were all excluded from endotoxin contamination.

The HD₅₀ values of TBM I, TBM II, TBM III, and Quil A were 5.39 ± 0.06, 7.91 ± 0.22, 4.48 ± 0.11, and 4.60 ± 0.02 μg/mL against 0.5% rabbit RBC, respectively. Among three TBMs, the hemolytic activity with the following order: TBM III > TBM I > TBM II.

When the animals were administered with TBMs at a single dose of 1.0 mg per mouse, there is no lethality observed except 2 mice died of TBM II during the monitoring period. Local swelling and loss of weight were found in the mice treated with TBM II and TBM III within 72 h after injection. Under the same condition, the local swelling, loss of hair, and loss of weight occurred in the Quil A-injected mice. Moreover, the numbers of deaths in five mice injected with Quil A were 2 and 3 for the doses of 150 and 200 μg per mouse, respectively.

To evaluate the adjuvant effect of TBMs, OVA-specific antibody titers in the sera from the immunized ICR mice were first detected using ELISA. Serum OVA-specific IgG and IgG1 antibody titers in the mice immunized with OVA/Quil A and OVA/TBMs were markedly higher than those immunized with OVA alone ( Figure 2A , P < 0.01 or P < 0.001). Noteworthy increases in serum OVA-specific IgG2a and IgG2b antibody levels were also observed in the mice co-immunized with TBMs and Quil A except for TBM I and TBM II (25 μg) towards IgG2b compared with OVA alone group ( Figure 2A , P < 0.05, P < 0.01, or P < 0.001). The results suggested that all three TBMs boosted antigen-specific humoral response towards OVA in a dose-dependent manner.

Mitogen- and antigen-stimulated splenocyte proliferation as well as NK cell activity were detected using MTT method. Con A-, LPS-, and OVA-stimulated splenocyte proliferation in the mice immunized with OVA/Quil A and OVA/TBMs were markedly greater than that in the OVA alone group ( Figure 2B , P < 0.05, P < 0.01, or P < 0.001). These results suggested that three TBMs promoted the activation of T and B cells in the immunized mice. The killing percentages of NK cells in splenocytes from the mice immunized with OVA/Quil A and OVA/TBMs toward K562 cells were dramatically higher than those immunized with OVA alone ( Figure 2C , P < 0.05, P < 0.01, or P < 0.001), indicating that three TBMs intensified the cytotoxicity of NK cells in the immunized mice.

To further assess the CMI in mice induced by TBMs, DTH to OVA was also determined. The DTH to OVA in ICR mice sensitized with OVA/Quil A and OVA/TBM was markedly greater than that with OVA alone at 24 h after challenge ( Figure 2D , P < 0.001). These results suggested that three TBMs augmented the CMI to OVA in mice.

Among three TBMs, the order in terms of inducing antibody responses, splenocyte stimulation index, and eliciting DTH was TBM III > TBM II > TBM I. However, no significant differences were observed in NK cell killing activities among three TBMs (P > 0.05).

To explore the characteristics of the adjuvant action of TBMs, the contents of Th1 and Th2 cytokines in the culture supernatants of OVA-stimulated splenocytes were detected using ELISA kits. The levels of IL-2, IFN-γ, and IL-10 in the culture supernatants of OVA-stimulated splenocytes from the ICR mice immunized with OVA/Quil A and OVA/TBMs were prominently higher than those immunized with OVA alone ( Figure 3A , P < 0.001). These results suggested that three TBMs dose-dependently triggered dual Th1 and Th2 responses to OVA in mice.

To elucidate the efficacy of TBMs on the Th1 and Th2 responses at the transcriptional level, the mRNA expression levels of cytokines and TFs in OVA-stimulated splenocytes from immunized ICR mice were determined by RT-qPCR. The mRNA levels of not only Th2 cytokines (IL-4 and IL-10) and TFs (GATA-3 and STAT6), but Th1 cytokines (IL-2 and IFN-γ) and TFs (T-bet and STAT4) in OVA-stimulated splenocytes from the immunized mice were dose-dependently markedly up-regulated by Quil A and TBMs compared with OVA alone group ( Figure 3B , P < 0.01 or P < 0.001), suggesting that TBMs facilitated the gene expression of Th1 and Th2 cytokines and TFs in OVA-stimulated splenocytes from the immunized mice.

To further profile antigen-specific T cell cytokine response, the effects of TBM I on the intracellular cytokine responses in the splenocytes from the immunized C57BL/6J mice were also investigated using FCM through gating on CD3⁺CD4⁺ and CD3⁺CD8⁺ cells ( Figure 3C ). TBM I markedly increased OVA-specific Th1 (IFN-γ⁺ and IL-2⁺) and Th2 (IL-4⁺ and IL-10⁺) cells in OVA-stimulated splenocytes from the immunized mice at the suitable doses ( Figure 3D , P < 0.01 or P < 0.001). Meanwhile, the frequencies of OVA-specific Tc1 (IFN-γ⁺ and IL-2⁺) and Tc2 (IL-4⁺ and IL-10⁺) cells in the OVA-stimulated splenocytes from the immunized mice were also markedly enhanced by TBM I compared to OVA alone group ( Figure 3E , P < 0.01 or P < 0.001). These results suggested that TBM I elicited both Th1/Th2 and Tc1/Tc2 immune responses in OVA-immunized C57BL/6J mice.

The cytokines and chemokines play a key role in inducing the recruitment of various immune cells at the injection site. The levels of the cytokines (IL-1β and IL-6) and chemokines (CCL2, CCL3, and CXCL2) in the quadriceps muscles at the various times after injection were first examined by ELISA. TBM I led to a transient and broad release of different cytokines and chemokines in the injected quadriceps muscles. Cytokine levels began to increase in muscles within 3 hours post-injection (hpi), suggesting early production by resident cells. The levels of CCL2 and CCL3 continuously increased up to 12 hpi, while CXCL2 levels peaked at 3 hpi, and gradually declined at 6 hpi. Meanwhile, the elevated levels of classical pro-inflammatory cytokines IL-6 and IL-1β peaked at 6 hpi ( Figure 4A ), indicating the tendency of muscle tissues to recover.

The mRNA expression levels of cytokines and chemokines in the quadriceps muscles at the various times after injection were also examined by RT-qPCR. As shown in Figure 4B , TBM I induced early significant up-regulation of the mRNA expression of IL-1β, IL-6, and TNF-α, among which was most pronounced for IL-1β. The mRNA expression levels of CCL2, CCL3, and CXCL2 were also significantly up-regulated after TBM I injection ( Figure 4B ). These findings further confirmed that TBM I induced a rapid and transient inflammatory response at the injection site.

The recruitment of innate immune cells to the injection site is a direct consequence of the local production of cytokines and chemokines. The muscle tissues were harvested at 12, 24, and 48 hpi, and the number of dendritic cells (DCs, CD11c⁺Ly-6C⁻Ly6G⁻), neutrophils (CD11c⁻Ly-6C⁺Ly6G^high), monocytes (CD11c⁻Ly6G⁻Ly6C⁺), macrophages (F4/80^high CD3e⁻CD45R⁻), eosinophils (SigLecF⁺CD117⁻), basophils (SigLec F⁻FcER1⁺CD117⁻), and mast cell (SigLecF⁻CD117⁺FcER1⁺) were detected by FCM. Recruitment occurred specifically in the adjuvant co-injected muscles, while elevated cell numbers in the OVA-treated muscle were almost not found at any time point compared to the PBS-treated muscles, except for a relative higher extent of monocytes and eosinophils at 12 hpi ( Figure 5A ). At the early12 hpi, TBM I resulted in increased numbers of recruited cell types, particularly neutrophils and DCs, about 10-fold and 4-fold more cells as compared to the OVA-treated muscle, respectively ( Figure 5A ) . Neutrophils peaked at 24 hpi, with a subsequent decline at 48 hpi, whereas DCs and monocytes peaked at 12 hpi and decreased at 24 and 48 hpi ( Figures 5A , S1 ). From 12 hpi onwards, however, TBM I resulted in a continuous increase of macrophages, eosinophils, basophils, and mastocytes, and a fairly increase in these cell numbers was observed at 48 hpi ( Figure 5A ). As for the dose-dependent analysis, 50 μg TBM I induced the optimized immune cell recruitment overall. Summarizing, TBM I induced a fast, but mostly transient, influx of neutrophils and DCs to the site of injection, followed by the recruitment of macrophages and monocytes, and finally eosinophils, basophils, and mastocytes ( Figure 5A ). The relatively high cell recruitment at the injection site induced by TBM I was observed at 24 hpi, indicating the peak of total cell infiltration.

The OVA-AF488 uptake of immune cells in the local tissues was also analyzed using FCM. As shown in Figure 5B , TBM I significantly promoted DCs, macrophages, monocytes, mast cells and neutrophils to uptake antigen. In contrast, only a few antigen-positive (Ag⁺) cell types were found in the muscles injected with OVA-AF488 alone. We extended the analysis to quantify the uptake of antigen to all identified cell types and to different time points after injection. Monocytes, macrophages, and dendritic cells but also neutrophils and mast cells were found to load antigen at all time points. Interestingly, we found that Ag⁺ DCs, Ag⁺ neutrophils, and Ag⁺ eosinophils peaked at 12 hpi for 100 μg TBM I, while for 25 μg and 50 μg TBM I these cells peaked at 24 hpi, indicating that the better adjuvant activity induced by a higher dose of TBM I presumably due to an earlier influx of Ag⁺ antigen-presenting cells (APCs), especially for DCs. At a later stage, TBM I resulted in higher Ag⁺ macrophage numbers compared to preceding time points, which was in accordance with the total cell kinetics ( Figure 5A ). These results indicated that TBM I promoted the recruitment and antigen uptake of immune cells at the injection site.

An acute inflammatory response and strong immune cell recruitment at the injection site led to enhanced antigen uptake and transport to dLNs, where Ag-specific T cell priming and production of Ag-specific, high affinity antibodies by B cells take place. To identify the cell types that serve as major antigen carriers induced by TBM I, the kinetics variation of the immune cell numbers in dLNs at the various times after injection was analyzed. As shown in Figures 6A and S2 , TBM I significantly increased the numbers of DCs, neutrophils, B cells, and eosinophils at 12 hpi and 24 hpi, while at 48 hpi the numbers of macrophages, T cells, mast cells and basophils increased, indicating that complexed processes occurred continuously in the dLNs.

Furthermore, the relative contribution of different types of cells to antigen ingestion was also analyzed by multiplying cell numbers with median fluorescence intensity (MFI). The results showed that the OVA⁺ cells in the dLNs are mainly confined to four cell types namely DCs, macrophages, monocytes, and B cells while many various cells could carry antigen at the injection sites ( Figure 6B ). Overall, OVA alone led to slightly elevated numbers of Ag⁺ cells, whereas broad Ag-containing DCs were found in the dLNs, indicating that OVA alone could activate resident DCs only to some extent. In contrast, co-injection of OVA with TBM I led to dramatically elevated Ag-loading cells in the LNs, including DCs, monocytes, macrophages and B cells. Noticeably, Ag⁺ DCs peaked at 12 hpi or earlier and then gradually decreased over time, whereas Ag⁺ macrophages and B cells peaked at 48 hpi ( Figure 6B ). The Ag⁺ monocytes kept stable during 12-48 hpi. These findings indicated that TBM I facilitated the migration and antigen transport of immune cells to dLNs.

To further profile the transcriptional profiles, mouse quadricep muscles treated with TBM I for 4 h were subjected to SurePrint G3 mouse gene expression microarray. TBM I induced 1194 DEGs covering 654 up-regulated and 540 down-regulated genes ( Figure 7A ). These DEGs were subjected to the GO function and KEGG pathway enrichment analysis, and the results were shown in Figure 7B . The biological function was connected to cell activation (P = 1.35E-33), leukocyte activation (P = 1.51E-28), inflammatory response (P = 9.12E-26), immune system process (P = 5.72E-24), and response to cytokine (P = 3.36E-23). The most significant pathways were IL-17 signaling pathway (P = 1.36E-07), NF-κB signaling pathway (P = 6.64E-07), TNF signaling pathway (P = 3.09E-06), Toll-like receptor signaling pathway (P = 1.91E-04), and Jak-STAT signaling pathway (P = 6.62E-04), which are associated with inflammation. Meanwhile, a PPI network was generated to capture the relationships among the top 20 of enriched biological function using Metascape ( Figure 7C ). Most terms in the network were interrelated and involved in inflammation and immunity, further confirming that TBMS I could mediate the inflammatory and immune response in mouse quadricep muscles at injection site.

LESS related to “innate immune response” were further analyzed using GSEA. Five LESS such as “inflammatory response”, “inflammatory response mediated by cytokines and chemokines”, “JAK-STAT3 pathway and regulation”, “B cell receptor signaling”, and “neutrophil chemotaxis” were enriched in TBM I group ( Figure 7D ). The genes enriched in inflammatory response and neutrophil chemotaxis gene sets mainly included CXCL3, CXCL2, CCL2, CXCR2, CXCR4, CCR2, CCR5, IL6, PTGS2, and IL-17A, which contributed to the activation and mobilization of various innate immune cells including neutrophils, DCs, and macrophages, and the induction of Th17 immune response ( Figure 7E , upper and lower left). In addition, TBM I also up-regulated the mRNA expression levels of TNF, IL-1β, IL5, IL6, CSF2, LCK and SYK, which are closely related to the JAK-STAT signaling pathway ( Figure 7E , middle).

The PPI network of the immune-related genes from GSEA was constructed using Cytoscape. As shown in Figure 7F , 209 TBM I-specific core genes including 98 DEGs were involved in the network. Furthermore, the top 10 hub genes were calculated through the seven algorithms of plug-in cytoHubba, respectively ( Table S3 ). Six common hub genes including IL-6, TNF, IL-1β, IL-10, SYK and LYN were obtained using upset plot ( Figure 7G ). Meanwhile, 10 TFs including NFKB1, RELA, STAT3, SPI1, CEBPA, SP1, JUN, STAT1, STAT6, and NR1I2 were predicted to mediate the expression of these core genes ( Table S4 ).

A total of 220 potential targets of TBM I were obtained by searching PubChem, Swiss Target Prediction, and SuperPred databases. Venn diagram showed that there were 18 common targets between 220 potential targets of TBM I and 209 core genes in the quadricep muscles induced by TBM I from GSEA ( Figure 8A ). The enrichment analysis of these common targets demonstrated that the pathways such as inflammatory response, cytokine signaling in immune system, cell adhesion, and cell chemotaxis were significantly enriched ( Figure 8B ), suggesting that TBM I induced inflammatory and immune response through regulating the expression of these targets. The PPI network of the common targets with scores greater than 0.4 was constructed using Cytoscape, which contained 18 nodes and 65 interaction pairs. PTGS2, IL-6, NOS3, SYK, VAV1, LYN, NFKB1, TLR8, ZAP70, and TNF were defined as 10 key targets using the MCODE plug-in of Cytoscape ( Table S5 ). The co-expression network and related functions of these10 key targets were visualized using GeneMANIA database. Ten key targets constructed a complex PPI network with the predicted of 39.20%, co-expression of 22.92%, physical interactions of 17.59%, shared protein domains of 14.94%, other of 3.14%, and co-localization of 2.21% ( Figure 8C ). The function analysis showed that these targets were mainly involved in activation of immune response, regulation of leukocyte mediated immunity, regulation of lymphocyte proliferation, cellular response to biotic stimulus, regulation of acute inflammatory response, positive regulation of defense response, and reactive oxygen species biosynthetic process, suggesting that TBM I induced immune activation, inflammation, and defense response via these 10 targets. As shown in Figure 8D , microarray data revealed that TBM I significantly up-regulated the mRNA expression levels of 6 genes among10 key targets in quadricep muscles (P < 0.05, P < 0.01, or P < 0.001).

The molecular docking analysis was performed to further identify the direct targets of TBM I. Two candidate upstream targets SYK (PDB: 4FL3) and LYN (PDB: 2ZV7) which were both included in the hub genes from GSEA and located upstream of STAT were selected as the receptors for docking. The affinities between TBM I and proteins were -1.60 kcal/mol and -3.43 kcal/mol for SYK, and LYN, respectively. TBM I showed a closely binding conjugation with the protein active site through forming H bonds with the amino acid residues LYS-458, ASN-381, PRO-535 and GLY-578 of SYK, and LYS-404, ARG-369, ASP-329, GLN-257 and PHE-258 of LYN, respectively ( Figure 8E ), verifying the reliability of the screened targets based on network pharmacology integrated with transcriptome and suggesting the key role of SYK and LYN in mediating TBM I adjuvant activity.

C2C12 cells has been previously proved to be an in vitro ideal model for studying the mechanisms of action of SBAs for intramuscular vaccines. To validate the key role of two candidate targets SYK, LYN, and a key TF STAT3 in mediating the inflammatory response in C2C12 cells induced by TBM I, the effects of TBM I on the mRNA expression levels of IL-6 and PTGS2 (COX-2) were first analyzed using RT-qPCR. TBM I significantly up-regulated the mRNA expression levels of these two pro-inflammatory factors in the C2C12 cells in a time- and concentration-dependent manner ( Figure 9A ). R406 and S3I-201 were used to specifically inhibit the function of SYK and STAT3, respectively. The pretreatment with SYK inhibitor R406 and STAT3 inhibitor S3I-201 dramatically suppressed the up-regulated mRNA expression of IL-6 and PTGS2 in the C2C12 cells induced by TBM I, while pretreatment of LYN inhibitor SU6656 could not ( Figure 9B ). Moreover, the secretion of IL-6 from TBM I-treated C2C12 cells was also inhibited by R406 and S3I-201 ( Figure 9C ), suggesting that SYK and STAT3 involved in the pro-inflammatory responses in C2C12 cells induced by TBM I.

The effect of TBM I on the phosphorylation of SYK was detected using western blotting. TBM I significantly and time-dependently promoted the phosphorylation of SYK at as early as 0.5 h ( Figure 9D ). Furthermore, the pretreatment with R406 significantly suppressed the phosphorylation STAT3 ( Figure 9E ) and protein expression levels of COX-2 in TBM I-treated C2C12 cells ( Figure 9F ). These findings suggested that SYK regulated the activation of downstream STAT3 as well as gene and protein expression of pro-inflammatory factors in the C2C12 cells induced by TBM I.