Comparative metabolomic analysis of antler-derived medicinal materials from red deer and sika deer Authors: Zhibiao Yu, Zheng Hu, Xuzhong Yang, Yuanmiao Huang, Daiquan Jiang, Zhangfeng Ding, Lei Liang, Yu Shu, Yihao Jiang Published: 2026 ABSTRACT AbstractIn this study, a TM widely-targeted metabolomics approach was employed to systematically characterize endogenous metabolites in the antlers(ANT), velvet antler(PA), and antler base(SDAB) of sika deer and red deer. A total of eighteen samples were selected and divided into six groups for metabolomic profiling. Based on ultra-high-performance liquid chromatography-tandem mass spectrometry (UPLC-MS/MS), metabolic features were comprehensively identified by combining untargeted metabolite screening with a self-constructed database from MWE, leading to the detection of 2,361 metabolites. Multivariate statistical analyses, including PCA (principal component analysis) and OPLS-DA (orthogonal partial least squares-discriminant analysis), were applied to investigate metabolic differences among tissues and between species. The results revealed significant tissue-specific metabolic profiles, particularly in antler tips, as well as species-specific differences between sika and red deer. Multiple differentially abundant metabolites were identified and predominantly enriched in pathways related to ABC transporters, glycerophospholipid metabolism, and nucleotide metabolism. These findings reveal distinct metabolic profiles associated with antler-derived materials and provide a scientific foundation for quality assessment, species discrimination, and biocompound discovery.It should be noted that this study has limitations, including a small sample size (n = 3 per group) and the geographical separation of sample sources. Future studies with larger cohorts and controlled environments are needed to validate these findings.Supplementary InformationThe online version contains supplementary material available at 10.1038/s41598-026-42612-x.Keywords: Sika deer, Red deer, Tissue specificity, TM wide-target metabolomics, Differential metabolitesSubject terms: Biochemistry, Biological techniques, Biomarkers, Computational biology and bioinformatics Introduction Antler-derived medicinal materials, including velvet antler, antler base (Luhuapan), and antler, have been traditionally used as tonic herbs, with a long history of therapeutic application and notable clinical efficacy. Their therapeutic value has been substantiated by millennia of traditional practice as well as modern scientific research. Contemporary pharmacological studies have isolated diverse bioactive constituents, including peptides, growth factors (e.g., IGF-1, TGF-β), polysaccharides, and abundant trace elements1–6. Furthermore, studies using cellular and animal models have revealed multiple mechanisms, including stimulation of bone formation, facilitation of cartilage repair, inhibition of neuronal apoptosis, and modulation of immune responses. These properties underscore their therapeutic potential, including roles in preventing osteoporosis and arthritis, providing neuroprotection, and serving as adjuncts in anti-tumor therapy7–9. The antler base (also referred to as the antler cap or Luhuapan), retained after velvet antler harvesting, is rich in collagen, peptides, and calcium-phosphorus minerals10–12. Studies have demonstrated that its extracts possess significant anti-inflammatory, antioxidant, and wound-healing activities13. Notably, it has shown distinctive efficacy in treating gynecological disorders such as mammary gland hyperplasia and mastitis, with its capacity to promote blood circulation, resolve stasis, and relieve mammary duct obstruction further supported by modern clinical practice14. Deer antler is utilized in both raw and processed forms, each with distinct applications: the raw form is used to clear heat and alleviate swelling, whereas the processed form (e.g., Cervi Cornus Colla) is used to warm and tonify the liver and kidney, and to strengthen bones and tendons. Its abundant protein matrix and osteoinductive components render it an effective natural material for bone defect repair and fracture healing15,16. Moreover, its role in regulating endocrine function and improving physical debilitation17,18 has recently regained research attention. Metabolomics focuses on the qualitative and quantitative analysis of endogenous small-molecule metabolites with molecular weights below 1000, such as amino acids, lipids, and nucleotides, examining their composition, dynamics, and alterations in response to internal and external factors19. This approach has been widely applied in traditional Chinese medicine research, clinical diagnostics, chemical ecology, environmental science, toxicology, and related fields20,21. Recent advances in mass spectrometry have further facilitated the application of metabolomics to antler-derived materials, particularly velvet antler. For instance, Zhang et al.22 compared the chemical profiles of velvet antler before and after thermal processing, while Su et al.23 applied metabolomics to investigate metabolic differences between sika and red deer velvet antlers. Thus, metabolomics has emerged as a crucial tool not only for profiling the composition of antler-derived medicines but also for exploring their growth processes and pharmacological properties. In this study, we utilized UPLC-MS/MS coupled with multivariate statistical analyses to investigate the metabolomic differences between antler-derived medicinal materials from red deer and sika deer. Materials and methods Materials and instruments Sika deer antler, velvet antler, and antler base (Luhuapan) were obtained from Lihaixia Deer Farm (Shuangyang District, Changchun, China), while red deer antler, velvet antler, and antler base were provided by Gansu Qilian Mountains Biotechnology Development Co., Ltd.Both farms employ standardized breeding practices with similar feed formulations and management protocols, and all samples were collected from healthy adult individuals during the same seasonal period to minimize environmental and husbandry-related variability. However, it is important to acknowledge that the sika deer and red deer samples originated from two distinct geographical regions (Changchun, Jilin and Zhangye, Gansu, respectively). Although husbandry practices were standardized, environmental factors such as climate and soil composition could potentially influence metabolic profiles. Therefore, the observed interspecies differences may reflect a combination of genetic and environmental effects. Methanol (AR) was purchased from Merck; acetonitrile (AR) was obtained from Shanghai Xingke High Purity Solvent Co., Ltd.; acetic acid was supplied by Shanghai Eno Chemical Technology Co., Ltd.; ammonium formate (AR), ammonia solution (AR), and formic acid (AR) were purchased from Shanghai Aladdin Biochemical Technology Co., Ltd. Chromatographic columns included a Waters ACQUITY UPLC BEH HILIC Column (1.7 µm, 1.0 mm × 100 mm) and a Waters ACQUITY UPLC HSS T3 C18 Column (1.8 µm, 2.1 mm × 100 mm), both from Waters Corporation. The UPLC system (ExionLC AD) and quadrupole time-of-flight mass spectrometer (TripleTOF 6600) were obtained from AB SCIEX. Sample preparation and metabolite extraction Samples stored at -80℃ were thawed on ice. A 400 μL methanol/water solution (7:3, V/V) containing the internal standard was added to 20 mg of sample, followed by vortexing for 3 min. The mixture was sonicated in an ice bath for 10 min, vortexed for 1 min, and then incubated at -20℃ for 30 min. Subsequently, the sample was centrifuged at 12,000 rpm for 10 min at 4 ℃. The sediment was removed, and the supernatant was centrifuged again at 12,000 rpm for 3 min at 4 ℃. A 200 μL aliquot of the supernatant was transferred for LC–MS analysis. LC–MS collection conditions UPLC conditions All samples were analyzed using three LC/MS methods. One aliquot was analyzed under positive ionization conditions using a Waters ACQUITY UPLC HSS T3 C18 column (1.8 µm, 2.1 mm * 100 mm). The mobile phases were 0.1% formic acid in water (solvent A) and 0.1% formic acid in acetonitrile (solvent B). The gradient program was as follows: 5–20% B over 2 min, 20–60% B over the next 3 min, 60–99% B in 1 min and held for 1.5 min, then returned to 5% B within 0.1 min and maintained for 2.4 min. Analytical conditions were as follows: column temperature, 40 °C; flow rate, 0.4 mL/min; injection volume, 2 or 5 μL; Another aliquot was analyzed under negative ionization conditions using the same gradient as in the positive mode. The third aliquot was analyzed under negative ionization conditions using a Waters ACQUITY UPLC BEH HILIC Column (1.7 µm, 1.0 mm * 150 mm). The mobile phases were solvent A (80% ACN, 10% water, and 10% MeOH with 20 mM ammonium formate, pH 10.6) and solvent B (40% ACN and 60% water with 20 mM ammonium formate). The gradient program was as follows: 5–20% B over 2 min, 20–70% B over 1.5 min, 70–95% B over 3 min, held at 95% B for 1 min, and then rapidly returned to the initial conditions. QTOF-MS/MS Data acquisition was carried out in IDA (information-dependent acquisition) mode with Analyst TF 1.7.1 Software (Concord, ON, Canada,version Analyst TF 1.7.1,https://sciex.com/). Source settings were as follows: ion source gas 1 (GAS1), 50 psi; ion source gas 2 (GAS2), 50 psi; curtain gas (CUR), 25 psi; temperature (TEM), 550 °C; declustering potential (DP), ± 60 V for both positive and negative modes; and ion spray voltage floating (ISVF), + 5000 V or − 4000 V in positive and negative modes, respectively. TOF–MS settings included: mass range of 50–1000 Da; accumulation time, 200 ms; and dynamic background subtraction turned on. For product ion scanning, parameters were: mass range, 25–1000 Da; accumulation time, 40 ms; collision energy, ± 30 V in both ion modes; collision energy spread, 15; resolution, UNIT; charge state, 1; intensity threshold, 100 cps; isotope exclusion within 4 Da; mass tolerance, 50 ppm; and up to 18 candidate ions per cycle. ESI-Q TRAP-MS/MS LIT and QQQ scans were conducted using a QTRAP® LC–MS/MS System equipped with an ESI Turbo Ion-Spray interface, operating in positive and negative ion modes under Analyst 1.6.3 software (version Analyst 1.6.3, https://sciex.com/). The ESI source conditions were: source temperature, 500 °C; ion spray voltage (IS), + 5500 V (positive) and -4500 V (negative); ion source gas I (GSI), gas II (GSII), and curtain gas (CUR) at 50, 50, and 25 psi, respectively; collision gas (CAD) set to high. Tuning and calibration were done using 10 and 100 μmol/L polypropylene glycol solutions in QQQ and LIT modes, respectively. MRM transitions were defined for each chromatographic interval based on eluting metabolites. Data preprocessing For all initial samples, missing values were imputed with 1 / 5 of the minimum value of each metabolite. CVs were then calculated for the QC samples, and only metabolites with CV values less than 0.3 were retained to generate the final dataset. Data analyses PCA Unsupervised principal component analysis (PCA) was implemented with the prcomp function in R (R version 4.1.2, www.r-project.org). Data were scaled to unit variance before analysis. Hierarchical cluster analysis and pearson correlation coefficients Hierarchical cluster analysis (HCA) of samples and metabolites was visualized as heatmaps with dendrograms, while pearson correlation coefficients (PCCs) between samples were calculated by the cor function in R(R version 2.9.4, www.r-project.org) and displayed as heatmaps. Both HCA and PCC analyses were performed with the R(R version 2.9.4, www.r-project.org) package ComplexHeatmap. For HCA, normalized metabolite signal intensities (unit variance scaling) were represented as a color spectrum. Differential metabolites selection Differential metabolites between two groups were identified using VIP values (VIP > 1) and P-values (P < 0.05) from Student’s t-test.It is noted that due to the exploratory nature of this study and the limited sample size, false discovery rate (FDR) correction was not applied. The list of differential metabolites should therefore be considered preliminary and interpreted with caution. The high number of differential metabolites identified in some comparisons may be attributed to the high sensitivity of untargeted metabolomics, the complex metabolic composition of antler tissues, and data variability inherent to small sample sizes. VIP scores were derived from OPLS-DA models generated via the MetaboAnalystR package in R(R version 1.0.1, www.r-project.org). Before OPLS-DA, data underwent log2 transformation and mean-centering. Overfitting was assessed through permutation tests (200 permutations). KEGG annotation and enrichment analysis Metabolite identification involved annotation against the KEGG Compound database (http://www.kegg.jp/kegg/compound/), followed by mapping to the KEGG Pathway database (http://www.kegg.jp/kegg/pathway.html). Results Untargeted metabonomics results Untargeted metabonomics results are as follows (Fig. 1): Untargeted metabonomics plot(note:RT (min) is the retention time; Precursor (DA) is the mass of parent ion). The plot displays (Fig.1) metabolic features detected in positive ion mode, with retention time (RT, min) on the x-axis and precursor mass (m/z, Da) on the y-axis. Each point represents a detected metabolite ion, illustrating the broad metabolomic coverage achieved in this study. More detailed table files are provided in the supplementary file Table S2: Untargeted metabolites. Sample quality control Total ion flow diagram Comparison of the total ion chromatograms of three QC samples under positive and negative ion detection modes showed that the peak response intensities and retention times were largely overlapping, with stable baselines. This indicates minimal variation due to instrument error throughout the entire analytical process (Fig. 2).Metabolite ion peaks from pooled QC samples were extracted using Analyst 1.6.3 software(https://sciex.com/), yielding 881 retention peaks in T3 positive ion mode, 1014 in T3 negative ion mode, and 466 in HILIC negative ion mode. System stability and quality control assessment. (a, b) Overlaid total ion chromatograms (TICs) of pooled quality control (QC) sample analyzed intermittently throughout the entire acquisition sequence in positive (a) and negative (b) ion modes using the T3 column, respectively. (c) Overlaid TIC spectra of same pooled QC samples in negative ion mode using the HILIC column. (d) QC sample correlation analysis among all sequenced injections of the pooled QC sample. (Note: The diagonal panels represent QC sample names; the lower-left panels show correlation scatter plots of the corresponding QC samples, with the x and y axes representing log-transformed metabolite abundances (each point represents one metabolite); the upper-right panels present the Pearson correlation coefficients for the corresponding QC pairs. QC sample correlation analysis C. Overlaid TIC spectra of QC samples in negative ion mode using the HILIC column. QC sample correlation analysis. (Note: The diagonal panels represent QC sample names; the lower-left panels show correlation scatter plots of the corresponding QC samples, with the x and y axes representing log-transformed metabolite abundances (each point represents one metabolite); the upper-right panels present the Pearson correlation coefficients for the corresponding QC pairs.). The diagonal panels show sample identifiers; lower-left panels are scatter plots of log-transformed metabolite abundances between corresponding QC samples; upper-right panels show Pearson correlation coefficients (r). High correlations (r > 0.99) indicate excellent analytical stability throughout the sequence. Stability of internal standards in QC samples Internal standards at known concentrations were spiked into the QC samples. The variations in their responses were minimal (CV ≤ 15%), demonstrating a stable detection process and high data quality (Table 1) (Supplementary Table 1). Table 1Stability of internal standards in QC samples.MetaboliteIDQ1(Da)RT(min)CVPurity(%)L-leucine-d7139.14641.140.02637779198Adenine-2-d1135.05321.340.04534328498L-Tryptophan-d5210.12912.40.02907543498Cytidine-5,6-d2244.090320.03751839398benzoic acid-d5126.06141.020.00871958698benzoic acid-d5126.06144.430.02845997298CAR(16:0)-d3403.36216.870.03803456198Cholic Acid-d4411.30494.6130.07897680698Cholic Acid-d4411.31276.310.01879243498D-Luciferin278.98984.340.02902795698L-2-chlorophenylalanine198.03222.480.01779048198L-2-chlorophenylalanine200.04782.480.02377380798 Stability of internal standards in QC samples. Principal component analysis (PCA) Principal component analysis of population samples For red deer and sika deer velvet antler, antler, and antler base, the contribution rate of the first (PC1), second (PC2), and third (PC3) principal components was 61.79, 12.19 and 6.25%, respectively (Fig. 3 a, b). Together, these three components accounted for 80.23% of the total variance, reflecting metabolic differences across all samples. Based on PC1, PC2, and PC3, CCAD_PA samples were clearly separated from CCAD_SDAB and CCAD_ANT samples; similarly, CENI_PA was separated from CENI_SDAB and CENI_ANT samples, and CCAD_PA samples were further distinguished from CENI_PA samples. However, CCAD_SDAB, CCAD_ANT, CENI_SDAB, and CENI_ANT showed substantial overlap in the PC1-PC2 score plot, indicating that PC1 and PC2 captured limited metabolic variation among these groups, while PC3 may account for some of the deeper differentiation. Principal component analysis (PCA) of the metabolomic data. Sample abbreviations: CCAD_PA: red deer velvet antler; CCAD_SDAB: red deer antler base; CCAD_ANT: red deer antler; CENI_PA: sika deer velvet antler; CENI_SDAB: sika deer antler base; CENI_ANT: sika deer antler. (a) 2D PCA plot. (b) 3D PCA plot. (c) PC1 Scroes. When PCA was applied to all metabolites in CCAD and CENI, substantial overlap and cross-distribution were observed between SDAB and ANT samples within each species, indicating minimal metabolic differences between antler base and antler samples of the same species. In contrast, PA samples were clearly separated from SDAB and ANT samples of the same species, indicating pronounced metabolic differences between velvet antler and antler base/antler samples. Additionally, SDAB and ANT samples displayed partial overlap between species, suggesting minor metabolic differences in antler base and antler across species. By contrast, PA samples from different species showed little overlap, highlighting distinct metabolic differences in velvet antler between species. Overall, the samples were divided into three distinct groups, each with characteristic metabolic profiles: Group1 comprised sika deer velvet antler, Group2 comprised red deer velvet antler, and Group3 comprised antler and antler base samples from both species. Within each group, metabolic profiles were relatively consistent. Principal component univariate statistical process control Using the established PCA model, quality control (QC) samples were monitored based on detected ion peaks to evaluate instrument stability. In the control chart, each point represents a sample, with the horizontal axis indicating the injection sequence. Point-to-point fluctuations may occur due to variations in instrument status. Normally, PC1 scores of QC samples are expected to fall within ± 3 standard deviations (SD). As shown in Fig. 3c, all QC sample PC1 scores fell within the ± 3 SD range, indicating minimal instrument fluctuations during the experiment and low instrument-induced error. All points fall within the ± 3 standard deviation (SD) limits, indicating minimal instrumental drift and high data quality throughout the run. More detailed table files are provided in the supplementary file Table S4:PCA data of total samples (including QC samples);Table S5:Interpretation rate table of PCA principal components of total samples (including QC samples). Metabolite expression profile To clarify the metabolite variation patterns in velvet antler, antler, and antler base of the two deer species, primary and secondary metabolites were identified using widely targeted metabolomics based on the UPLC-MS platform. A total of 25 metabolite classes were identified (Fig. 4), comprising 540 amino acids and derivatives, 279 organic acids and derivatives, 265 fatty acyls, 255 benzenes and derivatives, 195 heterocyclic compounds, 118 nucleotides and derivatives, 118 aldehydes, ketones, and esters, 104 glycerophospholipids, 102 alcohols and amines, 84 carbohydrates and derivatives, 83 hormones and hormone-related compounds, 33 flavonoids, 26 coenzymes and vitamins, 23 bile acids, 13 alkaloids, 13 terpenoids, 12 glycerides, 11 sphingolipids, 8 tryptamines, cholines, and pigments, 4 lignans and coumarins, 2 phenolic acids, 2 quinones, 1 sterol ester, 1 steroid, and 69 other metabolites. Metabolic profiling. The bar chart summarizes the number of metabolites detected across 25 major chemical classes. Amino acids and derivatives constituted the largest group, followed by organic acids, fatty acyls, and benzenoids, reflecting the diverse chemical landscape of antler-derived materials. More detailed table files are provided in the supplementary file Table S1: Information sheet of all sample metabolites. Cluster heatmap Differences in metabolite accumulation patterns among samples were assessed using a clustering heatmap (Fig. 5). The results showed clear intergroup variations, which were classified into four clusters. Cluster 1: Metabolite levels were most abundant in the CCAD_ANT group, followed by CENI_ANT, lowest in CCAD_PA, and moderate in the remaining groups. Cluster 2: Metabolite levels were highest in CCAD_PA, followed by CENI_PA, and relatively low in other groups. Cluster 3: Metabolite levels were most abundant in CENI_PA, followed by CCAD_PA, and lowest in other groups. Cluster 4: Metabolite levels were highest in CENI_PA, followed by CCAD_SDAB, and CCAD_PA, and lowest in the other groups. Biological replicates clustered tightly, indicating high data homogeneity and reliability. Clustering heatmap. Sample abbreviations: CCAD_PA: red deer velvet antler; CCAD_SDAB: red deer antler base; CCAD_ANT: red deer antler; CENI_PA: sika deer velvet antler; CENI_SDAB: sika deer antler base; CENI_ANT: sika deer antler. In summary, both cluster analysis and principal component analysis demonstrated significant metabolic differences among the groups. Among them, CCAD_PA and CENI_PA exhibited the most distinct differences compared to other groups. The heatmap displays normalized abundance (z-score) of metabolites across all samples. Samples (columns) and metabolites (rows) are clustered. Four major metabolite clusters (Cluster 1–4) are identified, each showing distinct accumulation patterns among the six experimental groups (CCAD_PA, CCAD_SDAB, CCAD_ANT, CENI_PA, CENI_SDAB, CENI_ANT). Tight clustering of biological replicates within each group confirms high data reproducibility. More detailed table files are provided in the supplementary file Table S3:Clustering heat map of all metabolite contents, showing the English names of substances (clustering both metabolites and samples). Differential metabolites analysis OPLS-DA analysis To identify significantly different metabolites in velvet antler, antler, and antler base of the two deer species, Orthogonal Partial Least Squares-Discriminant Analysis (OPLS-DA) was first performed. OPLS-DA is a supervised statistical method that models the relationship between metabolite expression and sample categories through partial least squares regression, thereby enabling sample classification. The Variable Importance in Projection (VIP) score was calculated to evaluate the contribution and explanatory power of each metabolite to sample classification. Pairwise OPLS-DA was conducted for the CCAD_PA, CCAD_SDAB, CCAD_ANT, CENI_PA, CENI_SDAB, and CENI_ANT groups, and the corresponding score plots were generated (Fig. 6). In this model, R2X and R2Y represent the model’s explanatory and predictive performance. All Q2 values exceeded 0.9, and the p-values for R2Y were below 0.005, confirming the robustness of the constructed models. The OPLS-DA score plots revealed clear separation among the comparison groups. OPLS-DA permutation and score. Sample abbreviations: CCAD_PA: red deer velvet antler; CCAD_SDAB: red deer antler base; CCAD_ANT: red deer antler; CENI_PA: sika deer velvet antler; CENI_SDAB: sika deer antler base; CENI_ANT: sika deer antler. (a) CCAD_ANT_VS_CCAD_SDAB OPLS-DA permutation. (b) CCAD_PA_vs_CCAD_ANT OPLS-DA permutation. (c) C).CCAD_SDAB_vs_CCAD_PA OPLS-DA permutation. (d) CENI_ANT_vs_CCAD_ANT OPLS-DA permutation. (e) CENI_ANT_vs_CENI_SDAB OPLS-DA permutation. (f) CENI_PA_vs_CCAD_PA OPLS-DA permutation. (g) CENI_PA_vs_CENI_ANT OPLS-DA permutation. (h) CENI_SDAB_vs_CCAD_SDAB OPLS-DA permutation. (i) CENI_SDAB_vs_CENI_PA OPLS-DA permutation. (j) CCAD_ANT_VS_CCAD_SDAB OPLS-DA score. (k) CCAD_PA_vs_CCAD_ANT OPLS-DA score. (l) CCAD_SDAB_vs_CCAD_PA OPLS-DA score. (m) CENI_ANT_vs_CCAD_ANT OPLS-DA score. (n) CENI_ANT_vs_CENI_SDAB OPLS-DA score. (o) CENI_PA_vs_CCAD_PA OPLS-DA score. (p) CENI_PA_vs_CENI_ANT OPLS-DA score. (q) CENI_SDAB_vs_CCAD_SDAB OPLS-DA score. (r) CENI_SDAB_vs_CENI_PA OPLS-DA score. (a-i) Permutation test plots (200 permutations) for the corresponding OPLS-DA models. The blue regression lines (R2 and Q2) are based on permuted data, while the green star represents the original model’s values. The leftward position of the original Q2 point relative to the permuted distribution indicates robust models without overfitting. (j-r) OPLS-DA score plots showing clear separation between compared groups in the modeled discriminant space, visually confirming the metabolic differences identified by multivariate analysis. Screening of differential metabolites Differential metabolites were identified based on P-value and VIP thresholds. Metabolites simultaneously satisfying P < 0.05 and VIP > 1 were considered significantly differential (Fig. 7). KEGG enrichment P-value. Sample abbreviations: CCAD_PA: red deer velvet antler; CCAD_SDAB: red deer antler base; CCAD_ANT: red deer antler; CENI_PA: sika deer velvet antler; CENI_SDAB: sika deer antler base; CENI_ANT: sika deer antler. (a) CCAD_ANT_vs_CCAD_SDAB_KEGG_Enrichment_P-value. (b) CCAD_PA_vs_CCAD_ANT_KEGG_Enrichment_P-value. (c) CCAD_SDAB_vs_CCAD_PA_KEGG_Enrichment_P-value. (d) CENI_ANT_vs_CCAD_ANT_KEGG_Enrichment_P-value. (e) CENI_ANT_vs_CENI_SDAB_KEGG_Enrichment_P-value. (f) CENI_PA_vs_CCAD_PA_KEGG_Enrichment_P-value. (g) CENI_PA_vs_CENI_ANT_KEGG_Enrichment_P-value. (h) CENI_SDAB_vs_CCAD_SDAB_KEGG_Enrichment_P-value. (i) CENI_SDAB_vs_CENI_PA_KEGG_Enrichment_P-value. The results were as follows: Between CCAD_ANT and CCAD_SDAB: 657 differential metabolites were identified, including 577 up-regulated and 80 down-regulated; Between CCAD_PA and CCAD_ANT: 1720 differential metabolites were identified, including 1581 up-regulated and 139 down-regulated; Between CCAD_PA and CCAD_SDAB: 1667 differential metabolites were identified, including 66 up-regulated and 1601 down-regulated; Between CENI_ANT and CCAD_ANT: 891 differential metabolites were identified, including 381 up-regulated and 510 down-regulated; Between CENI_ANT and CENI_SDAB: 846 differential metabolites were identified, including 447 up-regulated and 399 down-regulated; Between CENI_PA and CCAD_PA: 1506 differential metabolites were identified, including 165 up-regulated and 1341 down-regulated; Between CENI_PA and CENI_ANT: 1311 differential metabolites were identified, including 1110 up-regulated and 201 down-regulated; Between CENI_SDAB and CCAD_SDAB: 561 differential metabolites were identified, including 469 up-regulated and 92 down-regulated; Between CENI_SDAB and CENI_PA: 1242 differential metabolites were identified, including 160 up-regulated and 1082 down-regulated. Enrichment analysis of differential metabolites Differential metabolites from each comparison were mapped to the KEGG database to identify associated pathways. Pathway enrichment analysis was then performed to identify significantly enriched pathways. The main annotated and enriched pathways for each comparison were as follows24–26: CCAD_ANT vs. CCAD_SDAB: Metabolic pathways, Nucleotide metabolism, Amino sugar and nucleotide sugar metabolism, Biosynthesis of nucleotide sugars; CCAD_PA vs. CCAD_ANT: ABC transporters, Glycerophospholipid metabolism, Choline metabolism in cancer; CCAD_PA vs. CCAD_SDAB: ABC transporters, Thermogenesis, Pyrimidine metabolism; CENI_ANT vs. CCAD_ANT: Glycerophospholipid metabolism, Choline metabolism in cancer, Efferocytosis; CENI_ANT vs. CENI_SDAB: Metabolic pathways, Biosynthesis of cofactors, Nucleotide metabolism; CENI_PA vs. CCAD_PA: ABC transporters, D-Amino acid metabolism, Purine metabolism; CENI_PA vs. CENI_ANT: Metabolic pathways, Glycerophospholipid metabolism, Choline metabolism in cancer; CENI_SDAB vs. CCAD_SDAB: Metabolic pathways, Arachidonic acid metabolism, Nucleotide metabolism; CENI_SDAB vs. CENI_PA: Metabolic pathways, Thermogenesis, Biosynthesis of amino acids. Notably, certain pathways, such as “Metabolic pathways,” recurred across multiple comparisons. Nucleotide metabolism, which generates upstream metabolites, serves a foundational role. These metabolic pathways are closely related to the focus of this study. Bubble plots show the significantly enriched KEGG pathways for each comparison. The y-axis lists pathway names, the x-axis represents the enrichment factor (rich factor), bubble size indicates the number of differential metabolites mapped to the pathway, and bubble color corresponds to the statistical significance (− log₁₀(P-value)). Key enriched pathways include “Metabolic pathways”, “ABC transporters”, and “Glycerophospholipid metabolism”. Discussion Limitations of the study This study has several limitations that should be considered when interpreting the results. First, the sample size per group (n = 3) is relatively small. While metabolomics is a sensitive tool for biomarker discovery, the small sample size may limit statistical power and increase the risk of false positives. The findings presented here should be viewed as exploratory, providing hypotheses for future validation with larger biological replicates. Second, the sika deer and red deer samples were sourced from geographically separate regions (Changchun and Zhangye). Although standardized farming practices were implemented, we cannot entirely rule out the influence of environmental confounders such as local climate, soil, and micro-ecosystems on metabolic composition. Consequently, the observed interspecies differences may represent a composite of genetic and environmental factors. Future studies sampling both species from the same farm or multiple farms across regions would help disentangle species-specific effects. Finally, the identified differential metabolites and enriched pathways require further validation through targeted quantification and pharmacological experiments to confirm their functional relevance and medicinal significance. Comprehensive discussion Previous studies have shown that different tissues (antler, velvet antler, antler base) from the same species exhibit distinct therapeutic effects, while the same tissue can also vary in efficacy between species. These differences largely stem from variations in chemical composition, yet the underlying pharmacodynamic material basis remains poorly understood. In recent years, molecular biology, particularly omics technologies, has provided powerful tools for elucidating the active components of antler-based medicines. Among these, metabolomics, encompassing both targeted and untargeted approaches, enables comprehensive assessment of metabolic alterations and systematic exploration of small-molecule metabolites. In this study, untargeted metabolomics was applied to compare the metabolic profiles of velvet antler, antler, and antler base between sika deer and red deer, aiming to uncover tissue- and species-specific metabolic characteristics. The combined results of PCA, hierarchical clustering, and OPLS-DA delineate a clear hierarchy of metabolic specificity among the antler-derived materials. The most profound metabolic divergence was observed at the tissue level, specifically in velvet antler (PA). Within the same species, PA formed a distinct cluster that was sharply separated from both antler (ANT) and antler base (SDAB) in multivariate space, whereas ANT and SDAB samples exhibited considerable overlap. This establishes a strong tissue-specific metabolic signature for PA. Importantly, this tissue specificity was further amplified at the species level. The metabolic distance between sika deer PA and red deer PA was markedly greater than the distance between the two species for either ANT or SDAB. Consequently, velvet antler emerges not only as the most metabolically distinctive tissue but also as the most informative material for revealing interspecies metabolic differences. Given the distinct and species-sensitive metabolic profile of velvet antler established above, OPLS-DA analysis was employed to pinpoint the specific metabolites driving these separations. This led to the identification of numerous differentially expressed metabolites. These metabolites were enriched in key pathways, including ABC transporters, glycerophospholipid metabolism, nucleotide metabolism, and amino acid biosynthesis, which are closely associated with energy metabolism, cell membrane structure, signal transduction, and other physiological processes potentially affecting the bioactivity and medicinal value of velvet antler. Notably, “Metabolic pathways” and “Nucleotide metabolism” were consistently enriched across comparisons, underscoring the central role of energy metabolism and nucleic acid synthesis in the metabolic network of antler-derived medicines. The recurrent enrichment of “Glycerophospholipid metabolism” and “Choline metabolism” further suggests that lipid metabolism may contribute to the rapid growth and cellular proliferation of velvet antler. Beyond individual pathway enrichments, our data suggest potential crosstalk among the key metabolic networks identified. ABC transporters, responsible for translocating lipids, nucleotides, and other small molecules, may synergize with glycerophospholipid metabolism to maintain membrane fluidity and signaling during rapid antler growth. Simultaneously, nucleotide metabolism provides essential precursors for DNA/RNA synthesis and cellular energy transfer (e.g., ATP), supporting the high proliferative activity in velvet antler. This coordinated regulation among transport systems, lipid remodeling, and nucleic acid synthesis likely underpins the accelerated growth and bioactive compound accumulation in velvet antler, offering a systems-level perspective on its medicinal potential. Furthermore, several known bioactive metabolites were identified among the differentially expressed compounds in this study. For example, specific lipids and amino acids significantly upregulated in velvet antler have been reported to exhibit anti-inflammatory, antioxidant, and tissue-regenerative properties. The enrichment of these metabolites may underlie the traditional medicinal effects of velvet antler, including kidney reinforcement and the strengthening of bones and tendons. Moreover, species-specific differential metabolites between sika and red deer velvet antlers may serve as molecular markers for distinguishing these two valuable medicines, which is crucial for quality standardization and preventing misidentification. The untargeted metabolomics approach used in this study, incorporating cross-species and multi-tissue comparisons, enabled comprehensive profiling of the complex metabolic landscape of antler-derived materials, overcoming the limitations of traditional targeted analyses, including narrow coverage and high selectivity. These findings demonstrate the effectiveness of this approach in deciphering the material basis of traditional medicines. However, this study has certain limitations. For instance, all samples were collected from a single farm, and future studies should include specimens from diverse geographical origins to enhance the generalizability of the conclusions. Additionally, the identified differential metabolites and pathways require further validation through targeted quantification and pharmacological experiments to confirm their functional significance. It should be noted that this study has certain limitations. First, the sample size per group (n = 3) is relatively small, which may affect the statistical power and generalizability of the findings. While this preliminary study aimed to provide a broad metabolomic screening and identify potential differential metabolites and pathways, future investigations with larger sample sizes are warranted to validate and extend these results. Second, the sika deer and red deer samples were sourced from two distinct geographical regions (Changchun and Zhangye, respectively). Although both farms followed standardized breeding practices, environmental factors such as climate, diet, and local husbandry conditions could still influence metabolic profiles. Thus, the observed species differences may reflect a combination of genetic and environmental factors. To better disentangle species-specific effects, future studies should consider sampling both species from the same farm or multiple farms across regions to control for geographical confounders. Additionally, targeted quantification of key differential metabolites and functional validation in pharmacological models are needed to confirm their biological relevance and medicinal significance. Conclusion This untargeted metabolomic study revealed pronounced differences in the metabolic profiles of velvet antler, antler, and antler base from sika deer and red deer. Velvet antler exhibited distinct metabolic characteristics compared with other tissues, and interspecies metabolic differences were more pronounced in velvet antler than in antler or antler base, suggesting that velvet antler is a highly species- and tissue-specific medicinal component. These differential metabolites were enriched in key pathways, including ABC transporters, glycerophospholipid metabolism, and nucleotide metabolism, which are likely associated with the rapid growth and bioactive potential of velvet antler. This study systematically elucidates the metabolic basis of antler-derived medicinal materials and provides theoretical and empirical support for quality evaluation, species identification, and the further development of bioactive components.The findings also highlight the need for future research with larger sample sizes and controlled environmental conditions to strengthen these conclusions. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 Acknowledgements The authors thank Wuhan Metware Biotechnology Co., Ltd. for the testing services. Abbreviations Red deer Sika deer Velvet antler Antler base Antler Author contributions Methodology, Z.Y., Z.H., Y.S. and Y.J.;validation, Z.Y.; formal analysis, Z.Y.; investigation, Z.Y., Z.H.,Y.S., L.L.,D.J., Z.D. and Y.J.; resources, Y.J.,X.Y.,Y.H.;Data curation, Z.H. and Y.J.;writing—original draft, Z.H. and Y.J.;project administration, Z.Y. and Y.J.;All authors have read and agreed to the published version of the manuscript. Funding This research received no external funding. Data availability The datasets generated during and/or analysed during the current study are available in the figshare repository, 10.6084/m9.figshare.30178057.v1. Declarations Competing interests The authors declare no competing interests. Institutional review board Not applicable. Informed consent Not applicable. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Supplementary Materials Supplementary Material 1 Data Availability Statement The datasets generated during and/or analysed during the current study are available in the figshare repository, 10.6084/m9.figshare.30178057.v1.