Preparation of JWBHT

The six varieties of Chinese medicinal herbs included in JWBHT were purchased from Suzhou Tianling Traditional Chinese Medicine Pieces Co., Ltd. The detailed information and ingredient ratios of the materials used in JWBHT are provided in (Table S1). The administered dose was determined by converting the original human dose recorded in “Shang Han Lun” using standard body surface area normalization [22, 23]. This classical prescription ratio(Table S1) was maintained to preserve ethnopharmacological authenticity. The classical prescription dose based on “Shang Han Lun” has been validated in contemporary studies for its anti-pyretic effects [24, 25]. The preparation process of JWBHT was as follows: (1) accurately weighing the Chinese medicinal herbs according to the prescribed amounts and placing them in a ceramic pot; (2) wrapping the gypsum in cloth and simmering it, then soaking the other medicinal herbs for 30 min, wrapping glutinous rice in cloth and simmer; (3) boiling over high heat, then reducing the temperature and simmer for 30 min; (4) filtering the completed decoction, pouring it into a container, and setting it aside; (5) adding cold distilled water to the remaining residue in the ceramic pot, boiling, and decocting for another 30 min; (6) filtering the medicinal solution and combining the two filtered decoctions; and (7) concentrating the combined decoction under reduced pressure and storing it at -80 °C for future experimental use.

Characterization of Constituents in JWBHT by UHPLC-MS/MS

Sample Preparation. A 200 mg aliquot was vortexed with 10 mL of methanol: water (50:50, v/v) in 15 mL polypropylene tubes. After 30 min of ultrasonication, 1 mL of supernatant was centrifuged (14,000 × g, 5 min), filtered (0.22 μm), and transferred to vials for analysis.

Chromatography. Separation was conducted on ACQUITY UPLC HSS T3 (2.1 × 100 mm, 1.8 μm) at 35 °C with 10 µL injection. Gradient elution (0.3 mL/min) consisted of (A) 0.1% formic acid/H2O and (B) 0.1% formic acid/ACN.

Mass spectrometry. Q Exactive Orbitrap HRMS operated in Full MS/ddMS2 mode (± ESI). The parameters were as follows: m/z 100–1500; resolutions MS1/MS2 = 70,000/17,500; spray voltage 3.2 kV; capillary 320℃; aux gas 15 L/min (350℃); sheath gas 40 L/min; AGC 1e6; TopN = 5 with stepped NCE (30–50). Data were processed using Compound Discoverer 3.3.

Animal Experiments

CAWS preparation. CAWS was extracted from Candida albicans (NBRC1385) following the established protocols [26]. Briefly, Candida albicans was cultured in carbon-limited medium (48 h, 26℃, 250 rpm), followed by sequential ethanol precipitation (equal volume) and overnight incubation at 4℃. Pellets were resuspended in water (2 h stirring) and subjected to repeated centrifugation and subsequent supernatant ethanol precipitation (1:1). The final precipitate was acetone-desiccated (48 h) and reconstituted in PBS before the experiment.

The mouse model of KD. Male C57BL/6 mice (3–4 weeks, JOINN Lab, Suzhou) were acclimatized under controlled conditions (25 ± 2℃/50 ± 5% humidity). After 7 days of acclimatization, the animals were randomly assigned to three cohorts (n = 4–5, a small sample size per group was used, which is common for initial exploratory histopathological and omics studies in this field [19]): PBS control, CAWS (5-day; 4 mg/mouse i.p), and CAWS+JWBHT (supplementary i.g TCM, 13.91 g/kg/day crude drug × 14). Terminal sampling (peripheral blood/cardiac tissue) was conducted 14 days after CAWS via isoflurane euthanasia. All methods in this study were carried out following relevant guidelines and regulations.

Histological and Immunohistochemical Staining

Histological and immunohistochemical staining. Mouse heart was dissected, fixed in 4% paraformaldehyde, embedded in paraffin, and sectioned into 0.5 μm thick slices. As previously described, the sections were stained with hematoxylin and eosin (HE), elastic van Gieson (EVG) [27], Masson and Sirius Red staining. Images were captured using a microscope (Nikon DS-Ri2, Tokyo, Japan). The severity of inflammatory infiltration was measured using the cardiovascular inflammation score [28].

Immunofluorescence staining. Mouse heart samples were incubated with lipocalin-2/NGAL recombinant antibody (83853-2-RR, Invitrogen), rabbit anti-MMP9 (YM8019, ImmunoWay Biotechnology Co., USA), rabbit anti-Ly6G (YM8307, ImmunoWay Biotechnology Co., USA), rabbit anti-CD86 (91882, Cell Signaling Technology, USA), rabbit anti-CD31 (YM8207, ImmunoWay Biotechnology Co., USA), rabbit anti-COL1 (PA5-29569, Thermo Fisher Scientific, USA), and rabbit anti-α-SMA (55135-1-AP, SanYing Biotechnology, China). Alexa Fluor 488 goat anti-rabbit/mouse was used as the secondary antibody (GB25303/GB21301; Servicebio, Wuhan, China). Images were captured using a laser scanning confocal microscope (Olympus FV1200, Tokyo, Japan), and the nuclei were stained using 4,6-diamidino-2-phenylindole (DAPI).

Proteomics Study

Sample processing. Plasma samples underwent centrifugation (12,000 × g, 10 min, 4 °C) for clarification and high-abundance protein depletion. Protein quantification (BCA assay) was conducted before acetone precipitation (5:1 v/v, -20℃, 16 h). The pellets were washed with 80% acetone (12,000 rpm, 4℃), dissolved via sonication, processed through sequential reduction (10 mM DTT, 55℃, 20 min), alkylation (15 mM IAA, in the dark, 30 min), and tryptic digestion (1:50, 37℃).

LC-MS/MS analysis. Peptides (200 ng/sample) were separated on a nanoElute2 UPLC system (Bruker) using PepSep C18 column (75 μm × 15 cm, 1.9 μm) with 0.1% FA/ACN mobile phases (5%-35% B over 44 min, 400 nL/min). DIA-PASEF acquisition was conducted on timsTOF Pro2 covering m/z 350–1250, with collision energy ramping from 20 eV (1/K0 0.6 Vs/cm²) to 59 eV (1/K0 1.6 Vs/cm²).

Protein identification and quantification. Proteomic data were processed using Spectronaut v18.2 with its integrated Pulsar search engine for spectral analysis. MS/MS spectra were compared to the Mus musculus UniProtKB/Swiss-Prot reference proteome (Release 2022_11, Taxon ID 10090) using the following parameters: Cysteine carbamidomethylation as the fixed modification, with methionine oxidation and N-terminal acetylation designated as post-translational modifications. Trypsin specificity was enforced with a maximum of two missed cleavages. Mass tolerance thresholds were established at 20 ppm for precursor ions and 20 ppm for fragment ions. Statistical validation was conducted employing a 1% false discovery rate threshold at both peptide-spectrum match and peptide levels. All remaining analytical configurations were conducted using the default parameters of the software.

Bioinformatics workflow. Proteins with ≥ 1 unique peptide (FDR < 1%) were retained. Before log transformation, the missing values were imputed using the half-minimum method. Multivariate analyses in R 3.6.3/SIMCA 16.0.2 were conducted to identify differentially expressed proteins (|FC|≥1.2, p < 0.05) for subsequent PCA and Kyoto encyclopedia of genes and genomes (KEGG) pathway mapping. In total, 802 proteins were retained.

Metabolomics Study

Sample preparation. The samples were extracted with four volumes of ice-cold MeOH: ACN (1:1) containing deuterated internal standards. After vortexing (30 s), cold-bath sonication (10 min, 4℃) and protein precipitation (-40℃, 1 h), the supernatant (12,000 rpm, 4℃, 15 min) was collected. QC samples comprised pooled equal aliquots of all specimens.

LC-MS/MS analysis. Chromatographic separation and mass spectrometry were conducted using an integrated ultra-high-performance liquid chromatography system (Vanquish UHPLC, Thermo Scientific) interfaced with an Orbitrap Exploris 120 mass analyzer. A Waters ACQUITY BEH Amide column (2.1 × 50 mm, 1.7 μm) maintained at 4 °C with 2 µL injection volume was employed. The mobile phase comprised 25 mM ammonium acetate/ammonium hydroxide aqueous solution (pH 9.75) as component A and acetonitrile as component B under gradient elution conditions. Mass spectral acquisition was conducted using the information-dependent acquisition method with dual electrospray ionization modes (± 3.8 kV/-3.4 kV). Optimized parameters were used for mass spectral acquisition, including sheath gas at 50 arbitrary units, auxiliary gas at 15 arbitrary units, capillary temperature at 320℃, full scan resolution of 60,000, MS/MS resolution of 15,000, and stepped normalized collision energy of 20/30/40 eV.

Data processing. Raw data were converted into the mzXML format using ProteoWizard software, Data were processed (peak detection, peak extraction, peak alignment, and peak area integration) using the XCMS toolkit developed in the R environment. Metabolite identification was conducted by integrating R computational packages with the BiotreeDB bioinformatics database (version 3.0).

Correlation Analysis

Data were analyzed using R (version 4.4.2). KEGG pathway enrichment analysis was conducted on metabolomics and proteomics data using the clusterProfiler package [29]. Significantly enriched pathways (p < 0.05) were visualized using ggplot2. Pearson correlation coefficients were calculated between metabolites and proteins. A heatmap was established using the pheatmap package of R to display the resulting correlation matrix. Relationships between metabolites, proteins, and pathways were visualized using Sankey diagrams (networkD3), while gene-metabolite-pathway interactions were visualized as network graphs using igraph and ggraph. Pearson correlation, KEGG pathway enrichment (ClusterProfiler 4.0), and network construction (Cytoscape 3.10.3) were employed for multi-omics integration. For clinical validation, we retrieved the publicly available gene expression dataset GSE18606 from the GEO database (http://www.ncbi.nlm.nih.gov/geo) This dataset is based on the GPL6480 Agilent 014850 Whole Human Genome Microarray 4 × 44 K G4112F platform. This includes data from 9 healthy and age matched control children, as well as 20 children with KD in the acute and recovery phases. Differential expression analysis of LCN2 and MMP9 was conducted using GEO2R tool, and statistical comparison was performed.

Statistical Analysis

Statistical analyses were conducted using GraphPad Prism 9 Software. One-way ANOVA was used to compare groups. All results were derived from at least three independent experiments and are presented as mean ±standard deviation (SD). p < 0.05 was considered statistically significant.