Six healthy 3-year-old male forest musk deer, weighing approximately 8.0 kg were fed separately in a 30 m² enclosure at Chongqing Institute of Medicinal Plant Cultivation (altitude: 678 m).

All musk deer received a strictly standardized diet as described follows. The concentrate-to-forage ratio (1:4) and ingredient percentages (e.g., 65% corn, 25% soybean, 6% wheat bran, 0.4% NaCl, 1.5% CaCO₃, 0.8% CaHPO₄, 0.15% multi-vitamin, and 0.15% muti-mineral) were uniformly applied to all individuals. Concentrate feed was purchased from Tongwei Group, China. Coarse fodders (e.g., Ipomoea batatas, Pittosporum leaves, Broussonetia papyrifera leaves, and other light green weeds) were harvested from the same geographical area and processed identically (e.g., air-dried, chopped to 2–3 cm lengths) to minimize compositional variation. Feeding times (08:30 and 17:30) and ad libitum water access were consistent across all animals to ensure dietary rhythmicity. Concentrate ingredients (corn, soybean) were sourced from a single supplier and analyzed for macronutrient content (e.g., crude protein, fiber) to confirm batch-to-batch consistency. Coarse fodder samples were periodically tested for moisture and secondary metabolites to ensure no seasonal or regional variations affected the diet during the study period.

Initial mature musk was harvested on a cool morning at the end of June 2023. The selected forest musk deer was caught adhering to animal welfare guidelines. The skin surrounding the musk pod was disinfected using alcohol wipes. A special disinfected long-handle spoon was inserted into the musk scent pod and removed slowly, and this operation has no harm to forest musk deer itself, and will not affect musk secretion in the next year (Figure 1A). The forest musk deer was treated carefully under ethical consideration during sampling. Obtaining maturing and mature musk was the same as in the initial stage, but the collection time was the end of August and October 2023, respectively. A total of 18 musk samples (6 deer × 3 stages) were obtained. The minimum weight of each sample was 2.0 g, and each musk sample was placed in sterile cryogenic vials, flash-frozen in liquid nitrogen, and stored at −80°C until further experiments.

Samples (∼60 mg, n = 18) were obtained in 2-mL EP tubes, and underwent extraction with 0.48 mL of extraction liquid (V methanol: V chloroform = 3:1). Samples were homogenized in a ball mill for 4 min at 45 Hz, then ultrasound treated for 5 min in ice water. The resulting homogenate was centrifuged for 15 min at 13,000 rpm, at 4°C, and the supernatant (0.4 mL) was transferred to a fresh 2-mL GC-MS glass vial. GC-MS analysis was performed using an Agilent 7890 gas chromatograph system (Agilent, United States) coupled to a Pegasus HT time-of-flight mass spectrometer (Leco, United States). The system employed a DB-5MS capillary column coated with 5% diphenyl cross-linked with 95% dimethylpolysiloxane (30 m × 250 μm inner diameter, 0.25 μm film thickness; J&W Scientific, Folsom, CA, United States). A 1-μL aliquot of the analyte was injected in splitless mode. Helium was utilized as the carrier gas, the front inlet purge flow was 3 mL min⁻¹, and the gas flow rate through the column was 1 mL min⁻¹. The initial temperature was maintained at 50°C for 1 min, then raised to 310°C at a rate of 10°C min⁻¹, and held for 8 min at 310°C. The injection, transfer line, and ion source temperatures were 280, 270, and 220°C, respectively. The energy was −70 eV in electron impact mode. Mass spectrometry data were acquired using full-scan mode with a m/z range of 40–500 at a rate of 20 spectra per second following a solvent delay of 455 s.

Chroma TOF 4.3X software from LECO Corporation and the NIST database were employed for raw peak extraction, data baseline filtering and calibration, peak alignment, deconvolution analysis, peak identification, and peak area integration. Peak intensities were normalized to the total spectral intensity and used to predict the molecular formula according to additive ions, molecular ion peaks, and fragment ions. Peaks were matched using the mzCloud, mzVault, and MassList databases to obtain accurate qualitative and relative quantitative results.

To address technical variability in GC-MS data, all metabolite peak areas were scaled by the Total Ion Current (TIC) of each sample to correct for differences in injection volume and ionization efficiency (Wishart, 2016). Normalized data were log2-transformed to stabilize variance and approximate normality, improving the robustness of parametric statistical tests. We used the IQR method to identify and filter outliers in the normalized data: For each metabolite, calculate the first quartile (Q1) and third quartile (Q3) across all samples, define the IQR as Q3-Q1, flag values outside the range [Q1 − 1.5 × IQR, Q3 + 1.5 × IQR] as outliers, finally to remove these outliers (van den Berg, 2006). We additionally filtered metabolites with ≥50% missing values in peak area across all groups (Broadhurst et al., 2018). A total of 80 normalized data was obtained from preprocessed raw data. Metabolites were mapped to PubChem, HMDB, and KEGG databases to determine annotation and classification information for subsequent analysis. Multivariate statistical analysis, including principal component analysis (PCA) and hierarchical cluster analysis (HCA), was conducted using R packages. The processed data were log-transformed and mean-centered to perform orthogonal partial least squares discriminate analysis (OPLS-DA), from which the VIP (Variable Importance in the Projection) value was determined. The trend analysis was conducted by K-means, and chemicals with VIP > 1 and P < 0.05 were differential metabolites for the two groups. The enrichment and pathway analysis were performed to develop a network using the KEGG database.

The musk gland synthesizes and secretes musk, which enters the well-developed musk pod for storage, where it develops for about 2 months before reaching maturity (Jiang et al., 2023; Chen et al., 2018; Liu et al., 2023). Premature and final musk isolated from musk pods were normal, while their physical properties across maturity stages varied (Li et al., 2018; Zhang et al., 2021). Secretions obtained at the end of June were white and wet, like soybean curb residue (Figure 1B), and had strong, foul odors. In contrast, secretions at the end of August were light brown, oily semi-solid, like oily earwax (Figure 1C), and had a refreshing and enduring fragrance. The mature musk collected at the end of October was deep brown, oily powder or granules, like dry earwax (Figure 1D), with refreshing, enduring, and fascinating scents. Browning was involved in the transition from the milky initial liquid to the brown solid, with mature musk containing volatile aromatic intermediates like aldehydes, ketones, pyridines, and furans. Musk is considered a pheromone for marking territories and attracting females, enabling discrimination of strong individuals (Jiang et al., 2023; Hawkins, 1950b). Normal secretions collected at the final stage were high-quality black-brown solid musk (Wang et al., 2022), characterized by high levels of muscone and low levels of moisture (Zheng et al., 2020; Fan et al., 2018). The relative abundance of muscone was stable, while cholesterols and steroids changed in natural musk (Su et al., 2009).

The chemical composition was related to the quality, potentially indicating the metabolic profile of natural musk (Wang et al., 2022; Wang et al., 2021b). We utilized untargeted GC-MS metabolomics to identify 93 chemical compounds, encompassing 80 known and 13 unidentified metabolites, comparable to prior studies (Lv et al., 2022). The remaining 80 chemicals, consisting of 66 known and 14 unknown chemicals, were used for subsequent analysis. Based on chemical classification, the 66 known metabolites were categorized into five subclasses, and 29 chemicals were assigned as aromatic compounds (Supplementary Table S1).

Lipids are primary odor carriers and potential precursors of odorous substances (Sokolov et al., 1987). The Krona chart demonstrated that we isolated 34 (51.52%) lipids and lipid-like molecules, the highest proportion among metabolites, including 26 fatty acyls (FA), seven sterol lipids (ST), and one prenol lipid (PL). Among them, 11 chemicals belonging to FA (seven aldehydes [FA06], two esters [FA07], and one acid [FA01]) were aromatic compounds (Figure 2A; Supplementary Table S1). Three chemicals, namely, (Z)-13-Octadecenal, 9-Octadecenal, and cis-11-Hexadecenal (aromatic compounds), as well as 6-methyl-5-Hepten-2-ol, were active ingredients (AI) in the Traditional Chinese Medicine Systems Pharmacology Database and Analysis Platform (TCMSP) (Ru et al., 2014). Additionally, there were six C19 steroids (androgens) and derivatives [ST0202] and one cholesterol and derivatives [ST0101], the most common component in natural musk (Sokolov et al., 1987; Xie et al., 2022). These compounds had crucial functions in the chemical communication exchange between muskrats by regulating pheromone synthesis in musk cells (Jiang et al., 2022). Which illustrates a more systematic and comprehensive classification than previous studies (Liu et al., 2021; Lv et al., 2022; Zhang et al., 2021). Such as more detailed classification of lipids including fatty acyls (FA: FA01, FA06, FA07), sterol lipids (ST), and prenol lipid (PL). The AIs and targets of musk were studied from the perspective of systematic pharmacology through TCMSP database.

Organic oxygen compounds were the second most abundant compounds in musk (28.79%), encompassing 13 hydrocarbons and seven carbonyl compounds. 3-Methylcyclopentadecanone (muscone), which has not been identified in the secretion lipids (Sokolov et al., 1987), was isolated from the latter fraction that consisted of six macrocyclic ketones and one aldehyde (Figure 2A; Supplementary Table S1). Muscone, as the most abundant macrocyclic ketone, is a sweet, animal, and fatty compound (Shmuel, 2004). Together with low levels of analogs, it was the main medicinal active and odor-contributing ingredient in natural musk. Alkanes of organosilicon compounds, like tetradecyl oxirane, were also biologically active and aromatic compounds targeted to sodium-dependent noradrenaline transporters and nuclear receptor coactivator 2 associated with depression (Ru et al., 2014).

The third most abundant metabolites in musk were organoheterocyclic compounds (12.12%), and all were heterocyclic aromatic ingredients derived from furan, pyran, and pyridine compounds (Figure 2A; Supplementary Table S1) forming five- and six-membered heterocyclic rings (Majumdar et al., 2012). The aromaticity of furans is due to one of the lone pairs of electrons on the oxygen atom being delocalized into the ring, creating a 4n+2 aromatic system akin to benzene, which is formed by the oxidation of polyunsaturated fatty acids and carotenoids (Seok et al., 2015; Mariotti et al., 2013).

Additionally, three benzenoids and derivatives (4.54%, odoriferous constituents), and two organic acids and derivatives (3.03%) were detected in musk. Benzeneacetic acid, methyl ester, and p-Cresol were active ingredients (Figure 2A; Supplementary Table S1). Methyl phenylacetate (known as benzeneacetic acid, methyl ester) is related to various diseases such as cardiovascular disease, neurogenic disease, inflammation, thromboembolic disorders, and cancer (Ru et al., 2014). p-Cresol was related to alcoholism and leukemia by targeting alcohol dehydrogenases and nicotinate-nucleotide-dimethylbenzimidazole phosphoribosyltransferase (Ru et al., 2014).

By performing multivariate analyses on samples (Hockenhull et al., 2012), we can characterize the overall metabolite differences and the degree of variability between samples throughout musk maturation. The heatmap of HCA among all normalized data after quality control demonstrated that the metabolic spectrum of initial musk liquid in June was distinct from those of August and October (Figure 2B).

The OPLS model, integrating multivariate statistical data with supervised pattern recognition, successfully distinguished two groups (Figures 3A,B,C), illustrating significant separation in metabolite levels between different groups. Permutation tests illustrated the R² data were higher than Q², R²Y explained 0.968 (A vs. B), 0.984 (A vs. C), and 0.901(B vs. C) goodness of fit of this model, respectively. And Q² explained 0.323 (A vs. B), 0.254(A vs. C), 0.439(B vs. C) goodness of prediction, respectively. And all figures show high R²Y (cum) values (0.90–0.93), indicating strong data fit (Figures 3D,E,F).

The distribution of identified significantly differential metabolites in musk during maturation was illustrated by volcano plots following VIP > 1, P < 0.05 (Figure 3G). Among these, one was significantly up- and four were downregulated metabolites in the comparison between June and August. Ten of the 80 metabolites (12.50%), including two up- and eight downregulated chemicals, were significantly different when comparing June and October (Figure 3H). Two chemicals experienced a downward trend when comparing August and October (Figure 3I). Nine (out of 13) lipids and lipid-like molecules were significantly different between groups, including four aromatic compounds, namely, Methyl tetradecanoate, Tetracosanoic acid, methyl ester, cyclopentadecanone, and cis-11-Hexadecenal. This suggests that lipids are essential during musk secretion and maturation, in addition to one benzenoid and derivatives (p-Cresol, aromatic compounds), one organoheterocyclic compound (2-butyltetrahydrofuran, aromatic compounds), and an organosilicon compound (Supplementary Table S2).

The relative abundance of the identified 80 metabolites was standardized using a z-score, identifying 67 chemicals assigned to 5 groups throughout musk maturation by conducting trend analysis (Drakopoulos et al., 2016). A total of 36 metabolites were enriched into increased trends (C2 and C4), 12 chemicals (C3) exhibited downward trends, and the relative abundance of metabolites in C1 (V-shaped, n = 8) and C5 (Bell-shaped, n = 11) were assigned to opposing trends (Figure 4A; Supplementary Table S3). As heterocyclic aromatic compounds (chemical 4, 5, 8, 10, and 25), fatty acids (chemicals 7 and 77), phenols (chemical 16), and hydrocarbon (chemical 47) were gradually consumed, the relative concentration of four macrocyclic ketones containing C15 (chemical 49) and C16 (chemicals 44, 51 and 53) in cluster 4, as well as the six androgens and derivatives (chemicals 74, 76, 79, 83, 87, and 93) was increased. Macrocyclic ketones [(Cyclopentadecanone (C15) and 5α-androstan-3β-ol-17-one acetate (C16)] in C4 are key components of musk, they are structurally similar to muscone (a C16 macrocyclic ketone), the signature compound in natural musk (Liu et al., 2021). Macrocyclic ketones provide the characteristic musky aroma critical for communication (e.g., attracting mates or marking territories) (Li et al., 2016). Their accumulation likely reflects enzymatic oxidation of fatty acids/hydrocarbons (e.g., chemical 47) via cytochrome P450 pathways, a process conserved in musk biosynthesis across species (Han et al., 2017). Testosterone and its metabolites regulate musk gland development and secretion timing (Xie et al., 2020). Androgens may act as precursors for odoriferous compounds or enhance scent persistence via interactions with macrocyclic ketones (Liu et al., 2019). Consumption of heterocyclic aromatics (chemicals 4, 5) and fatty acids (chemical 7) likely diverts resources toward macrocyclic ketone biosynthesis, especially cyclopentadecanone and 5 alpha-androstan-3 beta-ol-17-one acetate, were significantly different metabolites during maturation. Hinting a coordinated biochemical network where macrocyclic ketones (C4) serve as the aromatic core, while androgens and derivatives modulate their biosynthesis and biological activity. Intriguingly, eight and two of the 13 significantly differential metabolites increased and decreased, respectively. An overwhelming majority (seven chemicals) of the increased class were lipids and lipid-like molecules. As presented in Supplementary Figure S1, the Venn diagram presented the number of metabolites in each comparison group, and the overlap between metabolites among different groups. 5,8,11-Eicosatriynoic acid, methyl ester was varied over three stages, and the unique number of metabolites in the A-C and A-B groups were 7 and 3, respectively.