Integrated morphological observation, metabolomics, and transcriptomics to investigate the effect of growth years on the quality of Atractylodes macrocephala Koidz

Abstract

Background

Atractylodes macrocephala Koidz. is a medicinal plant in high clinical demand due to its pharmacological efficacy. However, research on its quality dynamics across different growth years remains limited, primarily focusing on transcriptomics, microbiome analysis, and photosynthetic capacity, with studies only extending to three-year-old plants.

Results

This study examines A. macrocephala over a broader growth span (1, 2, 3, 5, and 10 years), integrating morphological, microstructural, metabolomic, spatial metabolite distribution, and transcriptomic analyses. Morphologically, rhizome weight and length increased with age, and the characteristic “Hejing” structure became more pronounced. Microstructural analysis revealed progressively developed xylem. Metabolomic profiling indicated a decline in sucrose content, alongside increased accumulation of bioactive sesquiterpenoids and phenolic acids over time. Transcriptomic analysis showed that genes involved in the biosynthesis of these active compounds—terpenoid backbone (HMGR, DXS, idi, GPS, and GGPS), phenylpropanoid (PAL, 4CL, and HCT), and sucrose metabolism (SPS and SPP)—were upregulated in older plants. Desorption electrospray ionisation mass spectrometry imaging (DESI-MSI) revealed an expanded distribution of key metabolites with increasing growth years.

Conclusion

The findings suggest that A. macrocephala aged 5 and 10 years exhibit superior quality, with a more distinct “Hejing” morphology and higher levels of bioactive compounds. The quality stabilises after five years, indicating that older plants may possess enhanced medicinal value.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12870-025-06958-0.

Introduction

Atractylodes macrocephala Koidz., commonly known as Baizhu in China, belongs to the Asteraceae family and is one of the most valuable medicinal plants. First recorded in Shennong bencao jing, it has been used for over 2,000 years. Its rhizome is traditionally employed to treat spleen hypofunction, appetite loss, abdominal distension, and miscarriage [1]. Modern pharmacological studies have demonstrated its efficacy in alleviating constipation, improving gastrointestinal function, and exerting anti-inflammatory, anti-ageing, and cardioprotective effects [2–4]. Since the quality evaluation of traditional medicinal materials mainly relies on morphology observation. Traditional Chinese medicine (TCM) practitioners evaluate A. macrocephala quality based on morphology. The wild A. macrocephala rhizome exhibiting a distinctive neck-like morphology-resembling the neck of a red-crowned crane-was referred to as “Hejing.” Ancient practitioners regarded A. macrocephala with the “Hejing” feature as superior in quality. However, due to increasing demand, habitat degradation, and slow growth, wild A. macrocephala has become scarce, even endangered [5].

To meet clinical needs, large-scale cultivation began during the Ming Dynasty. Two primary cultivation methods exist: direct-seeding and transplanting, with the latter dominating the medicinal market. However, to maximize land use and economic returns, the growth period of transplanted A. macrocephala has been reduced to two or even one year, limiting its medicinal quality [6]. Direct seeding A. macrocephala, though less commonly cultivated, undergoes minimal human intervention and follows a known growth trajectory, making it a better model for studying the impact of growth year. Prior research on A. macrocephala has predominantly focused on transplanted plants, analysing transcriptomics, microbiomes [7] and photosynthetic characteristics, with studies restricted to a maximum of three years of growth [8]. Despite some progress, comprehensive investigations into morphological changes, bioactive compound accumulation, and gene expression dynamics across different growth years are lacking. Our previous research indicated that morphology and some active components vary with growth year, but a systematic analysis integrating microstructural, metabolomic, spatial metabolite distribution, and transcriptomic data remains necessary.

In this study, we analysed A. macrocephala rhizomes at 1, 2, 3, 5, and 10 years of direct-seeding cultivation. We assessed morphology, microstructure, cell wall components, and key bioactive compounds. Metabolomics and desorption electrospray ionisation mass spectrometry imaging (DESI-MSI) were employed to quantify and spatially map metabolite distributions, while transcriptomics was used to evaluate gene expression changes. By elucidating the growth-dependent quality variations in A. macrocephala, this study aims to clarify the mechanisms underlying these differences, provide guidance for cultivating high-quality A. macrocephala, and ultimately enhance its clinical efficacy and market availability.

Materials and methods

Plant materials

Due to the scarcity of aged A. macrocephala, it was difficult to collect all samples from a single collection area. Both collection sites are located within the Tiantai Mountain range, sharing similar ecological conditions: they grow in yellow soil on mountainous slopes under forest cover, with annual precipitation ranging between 1,300 and 1,600 mm. One-three years old Atractylodes macrocephala (Y1, Y2, Y3) were collected from Xinchang City, Zhejiang Province, China (29°29’33’’N, 121°7’5’’E, altitude 331 m). Five and ten years old A. macrocephala (Y5, Y10) were collected from Taizhou City, Zhejiang Province, China (29°16’2’’N, 121°2’16’’E, altitude 771 m). Samples were harvested on October 20, 2023, immediately frozen in liquid nitrogen, and stored at − 80 °C for RNA and metabolite extraction. Each biological replicate consisted of three plants, and all experiments were conducted in triplicate. The identification of A. macrocephala was performed by Prof. Zhilai Zhan, and the specimen was deposited in the herbarium of National Resource Center for Chinese Materia Medica, China Academy of Chinese Medical Sciences.

Morphological and microstructural analysis

Macroscopic images of A. macrocephala were captured using a Canon EOS-800D camera (Lens EF 24–105 mm f/4 L IS USM). After drying, rhizome weight, length, and diameters (upper, middle, and lower) were measured using a vernier calliper. For microstructural analysis, fresh rhizome segments (~ 7 mm thick) were dehydrated in a graded ethanol series, embedded in paraffin, and sectioned into (10–15 μm) at −22 °C using a freezing microtome (Leica CM1860, Germany). Sections were stained with safranin and fast green and observed under a light microscope (Olympus BX51, Japan) to assess microstructural differences among groups.

Analysis of cell wall components, starch, and sucrose

Freeze-dried rhizomes stored at − 80 °C were pulverised at 30 Hz for 10 min using a mixer mill (MM 400, Retsch) with a zirconia grinding head. The contents of lignin, cellulose, and hemicellulose were determined using a fibre analyser (ANKOM220, USA) according to the manufacturer’s protocol. Starch content was quantified via anthrone reaction, following a previously established method [9].

For sucrose analysis, 50 mg of powdered rhizome was extracted with 4.0 mL of 70% ethanol via ultrasonic treatment (350 W, 40 kHz) for 40 min. The extract was centrifuged at 12,000 rpm for 10 min, and the supernatant was filtered through a 0.22 μm organic membrane before ultra-performance liquid chromatography (UPLC) analysis. UPLC was performed using an ACQUITY I-Class system (Waters Corporation, USA), equipped with a Waters ACQUITY UPLC BEH Amide column (1.7 μm, 2.1 mm × 100 mm). The mobile phase consisted of 0.1% ammonia in water (solvent A) and 0.1% ammonia in acetonitrile (solvent B) under the following gradient conditions: 0–3 min, 80% B; 3–4 min, 80 − 70% B; 4–10 min, 70% B; The flow rate was 0.2 mL·min⁻¹. The drift tube temperature was maintained at 50 °C, with a gain of 500, gas pressure of 40 psi, and cooling mode for the sprayer.

UPLC analysis of active components in A. macrocephala

For active component analysis, 100 mg of powdered rhizome was extracted with 1.5 mL of 70% methanol, the remaining conditions were the same as sucrose extraction. A Waters ACQUITY UPLC BEH (1.8 μm, 2.1 mm × 100 mm) chromatographic column were used for analysis. Atractylenolide I, atractylenolide III, and atractylone were analysed under the following conditions: column temperature, 40 °C; mobile phase A, 0.1% formic acid in water; and phase B, 0.1% formic acid in acetonitrile. The gradient program was as follows: 0–2 min, 10–45% B, 2–13 min, 45% B, 13–16 min, 45–53% B, 16–18 min, 53-58.4% B, 18–22 min, 58.4–78% B, 22–26 min, 78% B, 26–27 min, 78–90% B. Atractylenolide I was detected at 278 nm, while atractylenolide III and atractylone were detected at 222 nm, with a flow rate of 0.2 mL·min⁻¹.

Chlorogenic acid, neochlorogenic acid, and cryptchlorogenic acid were analysed using the following gradient: 0–3 min, 5–7% B, 3–5 min, 7% B. detected at 325 nm, with a flow rate of 0.4 mL·min⁻¹.

Non-targeted metabolomics analysis of A. macrocephala

Metabolites were extracted using the same procedure as for the active component. A pooled quality control (QC) sample was prepared by combining aliquots from all samples. Metabolomic profiling was conducted using an LC-ESI-MS/MS system (UPLC: ACQUITY I-Class, Waters Corporation, USA; MS: SYNAPT XS, Waters Corporation, USA). Each treatment group was analysed in triplicate.

UPLC was performed using a Waters ACQUITY UPLC HSS T3 column (1.8 μm, 2.1 mm × 100 mm) under the following gradient conditions: solvent A, 0.1% formic acid in water; solvent B, 0.1% formic acid in acetonitrile; 0–9 min, 1–40% B; 9–16 min, 40–70% B; 16–22 min, 70–100% B. The flow rate was 0.4 mL·min⁻¹.

Mass spectrometry was conducted in both positive and negative ionisation modes using an electrospray ionisation (ESI) source. The parameters were as follows: mass range, 50–1500 m/z; scan time, 0.2 s; capillary voltage + 0.5 kV (positive) and − 2.0 kV (negative); sample cone voltage, 40 V; source offset, 30 V; source temperature, 100 °C; desolvation temperature, 450 °C; cone gas flow, 50 L/h; desolvation gas flow, 900 L/h; nebuliser pressure, 6.5 bar.

Chemical components were identified via automated comparison with the Unifi/Qi standard library. Principal component analysis (PCA) and orthogonal partial least square-discriminant analysis (OPLS-DA) were used to simplify and maximize intergroup metabolic differences. Differentially accumulated metabolites (DAMs) were identified based on a combination of p-values and variable importance projection (VIP) scores in the OPLS-DA model, with significance criteria of p < 0.05 and VIP ≥ 1.

Sample Preparation and analysis for DESI-MSI

A 0.5 cm section was excised from the middle of the A. macrocephala rhizome and embedded in a 1.6% carboxymethyl cellulose (CMC) solution. Samples were mounted on a freezing microtome with a retraction system (Leica Biosystems, CM1950), fixed using an optimal cutting temperature compound, and sectioned into 30 μm slices at −15 °C. The sections were then adsorbed onto glass slides and stored at 4 °C until analysis.

Before DESI-MSI analysis, samples were equilibrated to room temperature. All experiments were conducted using a Xevo G2-XS Tof mass spectrometer (Waters Corporation, USA) with a DESI source (Waters Corporation, USA). The glass slide holding the sample was secured on a 2D moving stage and an N₂ gas-assisted solvent stream was directed onto the sample surface to generate MS spectra. Mass spectrometry data were acquired in full-scan mode under positive and negative ion-detection modes. The following parameters were applied: 0.1% formic acid in 95% methanol as the spray solvent; ES voltage of + 5 kV (positive mode) and − 4.5 kV (negative mode); solvent flow rate of 4 µL·min⁻¹; DESI gas pressure of 0.4 MPa; source temperature of 150 °C; sampling cone voltage of 40 V; spray impact angle of 60°; and scan time of 1 s. Mass spectra were recorded at a resolution of 20,000 across an m/z range of 50 − 1,500. DESI imaging was conducted using High Definition Imaging v1.4 software (Waters Corporation) to generate high-resolution images, which were reconstructed with linear smoothing.

RNA extraction, illumina sequencing, and transcriptomic data analysis

As no reference genome is available for A. macrocephala, PacBio single-molecule real-time (SMRT) sequencing was employed. Total RNA was extracted from A. macrocephala rhizomes using an RNAprep Pure Plant Plus Kit (Tiangen Biotech Corporation, China) and pooled for sequencing. RNA integrity was assessed with an RNA Nano 6000 Assay Kit on a Bioanalyzer 2100 system (Agilent Technologies Corporation, USA). mRNA was enriched using oligo(dT) magnetic beads and reverse-transcribed into cDNA with a Smart PCR cDNA Synthesis Kit. Double-stranded cDNA was amplified via PCR, purified using AMPure PB Beads, and used to construct an SMRTbell library. Sequencing was performed on a PacBio SequelII system at Novogene Co., Ltd. (Shanghai, China).

Raw reads were processed using SMRTlink 7.0 software to extract high-quality circular consensus sequences (CCS). Primers, barcodes, poly (A) tails, and concatemers were removed to obtain full-length non-chimeric (FLNC) sequences. FLNC reads were clustered using Minimap2, and low-quality sequences were corrected with LoRDEC [10] using RNA-seq data from the same A. macrocephala sample. Redundant corrected consensus reads were eliminated with CD-HIT (-c 0.95 -T 6 -G 0 -aL 0.00 -aS 0.99) to obtain the final transcript set for downstream analysis.

For Illumina sequencing, all A. macrocephala samples underwent RNA extraction and transcriptomic sequencing at Novogene Co., Ltd. (Shanghai, China). Libraries were constructed and sequenced on an Illumina NovaSeq 6000 platform (Illumina Corporation. USA). Low-quality reads were filtered using Perl scripts, and Q20, Q30, and GC contents were calculated. Filtered reads from 15 samples were aligned to the PacBio sequence using RESM [11]. The number of reads on each gene (Read count) was counted in each sample and convert to Transcripts Per Kilobase of exon model per Million mapped reads (TPM) as an indicator of the transcription or gene expression level. Differentially expressed genes (DEGs) were identified using the DESeq2 R package, applying|log2(fold change)| ≥ 1 and p-adjust < 0.05 as thresholds. DEGs were further analysed via Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses.

Validation of gene expression profiles by qRT-PCR

Quantitative real-time polymerase chain reaction (qRT-PCR) was performed to validate RNA-seq expression patterns. Ten genes associated with sesquiterpenoid biosynthesis were randomly selected for qRT-PCR. Specific primer pairs (Supplementary Table S1) were designed using Primer Premier 5. qRT-PCR was conducted on a LightCycler^@ 480 Real-Time PCR System (Roche, Switzerland) using the TB Green^® Premix Ex Taq™ II (Toyobo, Osaka, Japan). The amplification protocol included an initial denaturation at 95 °C for 30 s, followed by 40 cycles of 95 °C for 5 s and 60 °C for 30 s. Gene expression was normalised to EF-1α [12] and calculated using the 2^−ΔΔCt method [13]. Each sample was analysed in at least three biological and technical replicates.

Statistical analysis

All statistical analyses were performed using SPSS 27.0. Data are presented as mean ± standard deviation (SD). One-way ANOVA followed by Duncan’s multiple comparisons test was used to analyse differences among groups, with p < 0.05 considered statistically significant.

Results

Morphological differences in A. macrocephala across growth years

We compared A. macrocephala at 1, 2, 3, 5, and 10 years of growth (Fig. 1A) and observed distinct morphological changes with increasing age. The rhizomes of Y1 plants were the smallest, weighing < 2 g and measuring 2–3 cm in length. The weights of Y2, Y3, and Y5 were 3.00 ± 1.38 g, 3.95 ± 1.33 g, and 4.20 ± 1.14 g, respectively, with no significant differences among them (Fig. 1B). The lengths of Y2 and Y3 were 55.56 ± 19.19 mm and 69.67 ± 19.39 mm, respectively, also showing no significant difference. However, Y5 rhizomes were significantly longer at 92.23 ± 13.39 mm. The Y10 rhizomes reached 4.94 ± 2.15 g in weight and 94.08 ± 24.77 mm in length. While Y10 length did not significantly differ from Y5, it was significantly greater than Y1, Y2, and Y3 (Fig. 1C).

Fig. 1.

Morphological characteristics of A. macrocephala at different growth years. A Rhizomes of 1, 2, 3, 5, and 10-year-old A. macrocephala (scale bar: 5 cm). U, M, and L denote upper, middle, and lower measurement positions. B Rhizome weight (means ± SD, n = 10). C Rhizome length (means ± SD, n = 10). D Upper, middle, and lower diameters (means ± SD, n = 10). Letters indicate significant differences based on one-way ANOVA (p < 0.05)

We further analysed the upper, middle, and lower diameters of rhizomes at 1, 2, 3, 5, and 10 years (Fig. 1D). The upper diameters remained stable across all ages (6–8 mm). However, the middle and lower diameters showed significant changes over time. The middle diameters of Y1 (14.77 ± 1.97 mm) and Y2 (15.73 ± 4.59 mm) were similar, whereas those of Y3 (8.17 ± 2.31 mm), Y5 (7.27 ± 2.39 mm), and Y10 (6.96 ± 2.59 mm) were significantly smaller. Conversely, the lower diameters increased from Y1 to Y3, measuring 7.15 ± 3.80 mm, 10.38 ± 3.92 mm, and 15.67 ± 5.40 mm, respectively. The lower diameters of Y5 (13.21 ± 4.29 mm) and Y10 (14.08 ± 3.79 mm) did not significantly differ from Y3.

Overall, rhizome length and weight increased with growth, while middle diameters gradually decreased, and lower diameters increased. Y1, Y2, and Y3 were similar, whereas Y5 and Y10 shared similar traits. Notably, the pronounced elongation of the mid-upper rhizome in Y5 and Y10 led to a more distinct “Hejing” structure, a hallmark of high-quality A. macrocephala. Thus, in terms of morphology, Y5 and Y10 exhibited superior quality.

Microstructural differences in A. macrocephala across growth years

Microscopic analysis revealed that the epidermis remained largely unchanged across growth years, consisting of 2–5 layers, with some areas reaching up to 10 layers (Fig. 2A-E). However, cell arrangement became progressively looser with age, increasing intercellular spaces. In Y1 -Y3, cells were closely packed (Fig. 2A-C). By Y5, intercellular spaces began appearing in the cortex, and in Y10, these spaces expanded and extended closer to the xylem (Fig. 2D-E).

Fig. 2.

Microstructural characteristics of A. macrocephala across different growth years. Is: intercellular space, Oc: Oil chamber, Vr: Vascular-ray, Ve: vessel

The oil chambers, primarily distributed in the cortex with some presence in the xylem, remained round or oval in shape. Their sizes (100–250 μm) did not vary significantly across growth years. In Y1-Y3, oil chambers were numerous and structurally intact (Fig. 2F-H). However, in Y5 and Y10, oil chambers became less distinct due to increased intercellular spaces in the cortex (Fig. 2I-J).

The number of vessels per vessel cluster gradually increased with age, leading to an expansion of the xylem area (Fig. 2K-P). Vessel counts ranged from 3 to 10 in Y1, 5–13 in Y2, 17–30 in Y3, 15–58 in Y5, and 28–66 in Y10 (Fig. 2O-P). Compared to Y1, xylem thickness increased significantly in Y5 and Y10, providing structural support for rhizome growth.

Overall, as A. macrocephala matured, xylem thickness and intercellular space increased, while oil chamber density decreased. Supplementary Fig. S1 provides a comprehensive view of cross-sections, oil chambers, and vessel clusters across different growth years.

Differences in cell wall components, starch, sucrose, and active components in A. macrocephala across growth years

We analysed the contents of lignin, hemicellulose, cellulose, starch, and sucrose in A. macrocephala at different growth stages (Fig. 3A-E). Lignin and hemicellulose levels exhibited a slight upward trend from Y1 to Y10, reaching 3.83 ± 0.92% and 9.07 ± 0.42%, respectively in Y10. The highest cellulose content was observed in Y3 (5.95%), though differences across years were not statistically significant. Starch content remained relatively stable over time. Notably, sucrose content decreased significantly with increasing age, with Y10 showing a 32.31% reduction compared to Y1 (26.49 ± 7.48 mg·g⁻¹ vs. 17.93 ± 0.41 mg·g⁻¹). This suggests that sucrose accumulation in A. macrocephala diminishes as the plant ages.

Fig. 3.

Characterisation of secondary metabolites and active components in A. macrocephala over different years. A Lignin, (B) Hemicellulose, (C) Cellulose, (D) Starch, (E) Sucrose, (F) Atractylenolide I, (G) Atractylenolide III, (H) Atractylone, (I) Chlorogenic acid, (J) Neochlorogenic, (L) Cryptochlorogenic acid. Means ± SD, n = 3. Letters denote significant differences according to one-way ANOVA (p < 0.05)

To assess the impact of growth year on active components, we quantified sesquiterpenoids (atractylenolide I, atractylenolide III, and atractylone) and phenolic acids (chlorogenic acid, neochlorogenic, and cryptochlorogenic acid) (Fig. 3F-L). Sesquiterpenoid levels increased over time, peaking in Y10, with atractylenolide I, atractylenolide III, and atractylone reaching 0.30 ± 0.07 mg·g⁻¹, 0.15 ± 0.04 mg·g⁻¹, and 1.47 ± 0.33 mg·g⁻¹, respectively. Chlorogenic acid content also rose steadily, reaching 2.17 ± 0.32 mg·g⁻¹ in Y10. However, neochlorogenic and cryptochlorogenic acid levels peaked in Y3 and Y5, at 0.73 ± 0.09 mg·g⁻¹ and 0.76 ± 0.46 mg·g⁻¹, respectively. Overall, the accumulation rates of these active compounds were faster between Y1 and Y5 and stabilized from Y5 to Y10.

Metabolomic analyses of A. macrocephala at different growth years

To further investigate changes in chemical composition across growth years, we conducted a non-targeted metabolomic analysis on A. macrocephala samples. The total ion chromatogram (TIC) diagram of quality control samples was provided in Supplementary Fig. S2. Principal component analysis (PCA) revealed clustering patterns, with Y5 and Y10 forming a distinct group, Y1 and Y2 clustering together, and Y3 positioned between these two groups (Fig. 4A and B). This suggests that Y3 represents a metabolic transition point.

Fig. 4.

Metabolomics analysis of A. macrocephala at different growth stages. A PCA score plot (positive ion mode), (B) PCA score plot (negative ion mode), (C) Heatmap of differential metabolites across growth years

A total of 98 metabolites were identified (Supplementary Table S2). Differentially accumulated metabolites (DAMs) were analysed in positive and negative modes, with the top ten VIP-ranked metabolites listed in Supplementary Table S3. Fructans, including sucrose, 1 F-fructofuranosylnystose, 1,1,1,1-Kestohexaos, and fructoheptasaccharide, as well as amino acids like arginine, tryptophan, and tyrosine were more abundant in Y1 and Y2. Sesquiterpenes such as atractylenolide I, II, and III were predominant in Y10, consistent with the quantitative results in active components detection. Polyacetylenes, including atracetylenes D and 14-acetoxy-12-senecioy-loxytetradeca-2E,8E,10E-trien-4,6diyn-1-ol were enriched in Y5 and Y10. Additionally, three unidentified compounds (C₃₆H₆₀O₁₆, C₃₇H₆₂O₁₆, and its isomer) showed increased accumulation with plant age.

Spatial distribution of metabolites in A. macrocephala across growth years

To analyse the distribution of metabolites in A. macrocephala at different growth stages, DESI-MSI was performed on raw tissue slices. The optimal sectioning conditions were determined by testing various thicknesses and cutting temperatures. When the cutting temperature fell below − 15 °C, obtaining intact sections from older samples with increased xylem content proved difficult [14]. Additionally, sections thinner than 30 μm tended to curl and failed to adhere to glass slides. Based on these findings, a cutting temperature of −15 °C and a section thickness of 30 μm were selected for all samples.

DESI-MSI analysis in both positive and negative ion modes provided spatial metabolite distribution data (Fig. 5). Quinic acid was predominantly localised in the epidermis of Y1-Y3 samples. As the plant aged, its distribution expanded into the cortex, suggesting increased accumulation over time. The spatial pattern of isoschaftoside also changed with growth. In Y1 to Y3, it was primarily found in the epidermis and xylem. By Y5, it had spread throughout the entire section, but by Y10, its signal had diminished significantly. Several oligosaccharides were detected via DESI-MSI, including sucrose, kestose, nystose, 1 F-Fructofuranosylnystose (GF4), 1,1,1,1-Kestohexaose (GF5), and 1,1,1,1,1-Kestoheptaose (GF6). These were uniformly distributed across all sections, with high overall signal intensity. Three polyacetylene compounds—12-acetoxy-12-methylbulbu-tyl-2E,8Z,10E-trien-4,6-diyn-1-ol (PAT1), 12-Senecioyloxytetradeca-2E,8E,10E-trien-4,6-diyne-1,14-diacetate (PAT2), and 14-acetoxy-12-methylbulbu-tyl-2E,8Z,10E-trien-4,6-diyn-1-ol (PAT3)—were also identified. In Y1-Y3, PAT1 and PAT2 were uniformly distributed across sections, whereas in Y5 and Y10, they were primarily localised in the cortex. PAT3 exhibited a similar trend, shifting from whole section distribution in Y1-Y2 to cortex-specific accumulation in Y3-Y5.

Fig. 5.

LC-MS-based metabolomics combined with DESI-MS to explore the spatial metabolome of A. macrocephala at different growth years. (GF4: 1 F-Fructofuranosylnystose, GF5: 1,1,1,1-Kestohexaose, GF6: 1,1,1,1,1-Kestoheptaose, PAT1: 12-acetoxy-12-methylbulbu-tyl-2E,8Z,10E-trien-4,6-diyn-1-ol, PAT2: 12-Senecioyloxytetradeca-2E,8E,10E-trien-4,6-diyne-1,14-diacetate, PAT3: 14-acetoxy-12-methylbulbu-tyl-2E,8Z,10E-trien-4,6-diyn-1-ol)

Overall, quinic acid and polyacetylene distribution areas expanded with plant age, while fructans remained evenly distributed throughout all sections. Quinic acid plays a critical role in insect resistance [15], and its predominant localisation in the epidermis and cortex suggests these tissues are key defensive regions. The even distribution of sucrose and fructans across sections implies that these carbohydrates serve as major primary metabolites throughout the plant. The observed cortical accumulation of polyacetylenes in older plants suggests that their content increases with age and becomes concentrated in specific tissues over time.

Functional annotation and differential gene expression analysis

A total of 98.0 Gb of clean transcriptomic data was generated from A. macrocephala rhizome samples. The sequencing error rate was 0.03%, with Q20 and Q30 base percentages of 97.62% and 93.18% respectively. GC content ranged from 44.45 to 45.60%, confirming high data quality (Supplementary Table S4).

Pairwise comparisons of transcriptomic profiles across growth years identified 10,902 DEGs (Fig. 6A). The most DEGs were found in Y5 vs. Y1, Y10 vs. Y1, and Y10 vs. Y2, with 983, 706, and 605 upregulated genes, respectively, and 971, 786, and 615 downregulated genes. In contrast, fewer DEGs were detected in Y3 vs. Y2, Y10 vs. Y5, and Y5 vs. Y3. PCA analysis revealed distinct clustering, with Y1, Y2, and Y3 samples grouping closely and Y5-Y10 forming a separate cluster (Fig. 6B). This suggests substantial transcriptomic divergence between young (Y1-Y3) and older (Y5-Y10) plants, with major transcriptional changes occurring before Y5, after which transcriptomic differences diminished.

Fig. 6.

A The number of DEGs across different growth years compared pairwise. B PCA of transcriptomics for A. macrocephala across different growth years

GO and KEGG enrichment analyses were performed on DEGs across all comparisons (Supplementary Table S5&6). GO analysis showed that upregulated DEGs were primarily enriched in extracellular region (GO: 0005576), microtubule (GO: 0005874), cell wall organisation (GO: 0071555), defense response to bacterium (GO: 0042742). Downregulated DEGs were mainly enriched in response to abscisic acid (GO: 0009737), auxin-activated signalling pathway (GO: 0009734), amino acid transport (GO: 0006865), and oxidoreductase activity (GO: 0016717).

KEGG pathway analysis identified 25 significantly enriched pathways (p-adjust < 0.05). Upregulated DEGs were predominantly involved in starch and sucrose metabolism (map00500), Gap junction (map04540), two-component system (map02020), and phenylpropanoid biosynthesis (map00940). Downregulated DEGs were enriched in Plant-pathogen interaction (map04626), fatty acid metabolism (map01212), starch and sucrose metabolism (map00500), and circadian rhythm - plant (map04712).

Identification and expression of candidate genes involved in starch, sucrose, lignin, and cellulose biosynthesis

KEGG pathway analysis revealed significant enrichment in the Starch and sucrose metabolism pathway. Within the sucrose biosynthesis pathway (Fig. 7A), sucrose biosynthesis is primarily catalysed by sucrose-phosphate synthase (SPS), and sucrose-phosphatase (SPP), which convert sucrose-6’-phosphate, a phosphorylated sucrose intermediate, into sucrose. SPS is considered the rate-limited enzyme in this process. Transcriptomic analysis showed that SPS and SPP were highly expressed in Y1-Y3, aligning with quantitative and metabolomic data. This suggests that A. macrocephala accumulates more sucrose during early growth stages to support development.

Fig. 7.

A Sucrose biosynthesis pathway in A. macrocephala. B Terpenoid backbone biosynthesis in A. macrocephala. C Phenylpropanoid biosynthesis pathway in A. macrocephala. Red letters indicate enzyme abbreviations involved in each reaction. The heatmap displays expression patterns of the corresponding enzyme genes across different growth years

Lignin biosynthesis is a branch of the phenylpropanoid pathway, with phenylalanine ammonia-lyase (PAL) acting as the first rate-limiting enzyme, controlling the overall metabolic flux. Cinnamyl CoA reductase (CCR) serves as the rate-limiting enzyme, specifically in the lignin biosynthesis branch. Transcriptomic data indicated high PAL expression in Y3 and Y10, whereas CCR was upregulated in Y3, Y5, and Y10 (Fig. 7C). Additionally, genes encoding other key enzymes in lignin biosynthesis, including 4-Coumarate: CoA ligase (4CL), hydroxycinnamoyl transferase (HCT), and ferulate 5-hydroxylase (F5H), exhibited increased expression in Y3, Y5, and Y10. The overall upregulation of lignin biosynthetic genes with increasing growth time correlates with the observed rise in lignin content in the quantitative analysis.

Identification and expression of candidate genes involved in sesquiterpenoid and phenylpropanoid biosynthesis

Plant terpenoid biosynthesis occurs primarily through two pathways: the mevalonate (MVA) pathway and the methylerythritol phosphate (MEP) pathway, both of which produce the essential intermediates isopentenyl pyrophosphate (IPP) and dimethylallyl pyrophosphate (DMAPP). HMG-CoA reductase (HMGR) and 1-deoxy-D-xylulose-5-phosphate synthase (DXS) serve as rate-limiting enzymes in the MVA and MEP pathways, respectively. Transcriptomic analysis showed elevated HMGR expression in Y1 and Y10, while DXS expression was higher in Y5 and Y10.

Additionally, genes involved in the downstream terpenoid backbone biosynthesis pathway, such as isopentenyl-diphosphate Delta-isomerase (idi), Geranyl pyrophosphate synthase (GPS), and Geranylgeranyl diphosphate synthase (GGPS), showed increasing expression levels with growth time (Fig. 7B), consistent with quantitative and metabolomic findings. Given the observed increase in sesquiterpenoid and phenolic acid content with growth time, we hypothesise that HMGR, DXS, idi, GPS, and GGPS in the terpenoid backbone biosynthesis pathway, as well as PAL, 4CL, and HCT in the phenylpropanoid biosynthesis pathway, serve are key regulatory genes influenced by growth year.

qRT-PCR validation of RNA-seq data

To validate the transcriptomic findings, qRT-PCR analysis was performed. Given the importance of sesquiterpene biosynthesis in A. macrocephala, 10 genes involved in terpenoid backbone and phenylpropanoid biosynthesis were selected for qRT-PCR validation. Primer sequences are provided in Supplementary Table 1. The qRT-PCR results exhibited expression patterns consistent with RNA sequencing, with a Pearson correlation coefficient of 0.84 between the two datasets, confirming the reliability of the RNA-seq data (Supplementary Figure S3).

Discussion

The growth year of medicinal plants plays a crucial role in determining overall quality. Growth time, in combination with environmental factors, influences the accumulation of bioactive compounds, resulting in variations across different harvest years. This phenomenon has been widely observed in perennial medicinal plants, such as Panax notoginseng [16], Dendrobium moniliforme [17], and Ginkgo biloba leaves [18]. In this study, we systematically analysed A. macrocephala across five growth stages (1, 2, 3, 5, and 10 years) using a comprehensive approach that integrates morphological and microstructural assessments with metabolomics, spatial metabolomics, and transcriptomics. In this study, the 1, 2, and 3-year-old and the 5 and 10-year-old A. macrocephala samples were collected from sites with different altitudes. Previous research indicates that altitude can significantly influence the morphology and accumulation of bioactive compounds in medicinal plants. For example, Jia et al. reported that Dendrobium officinale grown at higher altitudes had significantly elevated levels of polysaccharides, amino acids, and their derivatives compared to plants grown at lower altitudes [19]. Similarly, Li et al. found that the underground biomass of Gynostemma longipes increased by 200–209% at high-altitude sites [20]. Specifically for A. macrocephala, Tan et al. [21] demonstrated that terrains above 200 m were more favorable for the accumulation of lactone compounds, whereas altitudes below 200 m hindered such accumulation. In our study, all samples were collected from regions above 200 m, suggesting that the conditions were generally suitable for lactone accumulation. Therefore, we believe that growth age likely had a greater influence on A. macrocephala traits than altitude in this experiment. This represents the first in-depth investigation of how growth year influences the quality of A. macrocephala, providing novel insights into its complex regulatory mechanisms.

Effects of growth year on morphology

Plant morphology undergoes significant changes with age, serving as a key indicator of medicinal plant quality. Our morphological analysis revealed pronounced differences between A. macrocephala samples with longer versus shorter growth periods. Measurement of length, weight, and diameter indicated that while plant length and weight increased over time, the middle segment gradually thinned, whereas the lower portion thickened. Notably, plants with extended growth years exhibited a distinct “Hejing” morphology, a feature highly valued in traditional Chinese medicine for quality assessment. Our findings suggest that A. macrocephala grown over five years consistently develops this characteristic morphology, highlighting its potential as a key criterion for evaluating plant age and medicinal quality.

Effects of growth year on microstructure

This study provides the first comparative microstructural analysis of A. macrocephala across different growth years. The primary structural differences were observed in the xylem and oil chamber compartments. Xylem tissue, which consists of vessels, xylary fibres, and parenchyma cells, is crucial for structural support and water transport [22]. Our microscopic analysis showed that lignification increased with plant age, enhancing xylem development, which in turn supports the observed elongation of A. macrocephala over time.

Oil chamber analysis revealed that while the chamber area remained relatively stable across growth stages, their number increased significantly between Y1 and Y3, before declining in Y5 and Y10 due to the formation of intercellular spaces. Notably, our quantitative data showed sesquiterpenoid levels increased with growth year, aligning with previous research by [23], which established a correlation between oil chamber structure and the accumulation of bioactive compounds such as atractylenolide I, atractylenolide III, and atractylone. However, our study further indicates that beyond a certain growth period, oil chamber count alone is insufficient for predicting sesquiterpenoid content due to structural modifications. These findings refine the current understanding of oil chamber development and bioactive compound accumulation in A. macrocephala.

Effects of growth year on sesquiterpenoid accumulation

Non-targeted metabolomic analysis identified sesquiterpenoids as DAMs across growth stages, with quantitative analysis confirming that atractylenolide I, atractylenolide III, and atractylone concentrations increased over time. Transcriptomic profiling further revealed elevated expression levels of key genes in the terpenoid biosynthesis pathway, including HMGR, DXS, idi, GPS, and GGPS, in older plants.

Sesquiterpenoids serve multiple ecological roles in plants, including anti-herbivory, antimicrobial activity, allelopathy, and protection against oxidative stress. Our data suggest that prolonged growth enhances damage, etc [24]. We believe that higher growth year of A. macrocephala accumulated sesquiterpenoid biosynthesis, potentially improving A. macrocephala’s environmental resilience. Among the identified genes, HMGR and DXS function as rate-limiting enzymes in the MVA pathway and MEP pathways, respectively, while GPS and GGPS regulate downstream terpenoid biosynthesis. Previous studies have demonstrated that overexpression of GPS in tobacco increases sesquiterpenoid accumulation [25], while GGPS is a crucial terpenoid synthase in Liriodendron tulipifera [26] further supporting the relevance of these genes in A. macrocephala’s biosynthetic regulation.

Pharmacological research has demonstrated the significant therapeutic potential of A. macrocephala sesquiterpenoids. Atractylenolide I, III, and atractylone exhibit potent anti-inflammatory properties, showing efficacy against osteoarthritis, inflammatory bowel disease, and ulcerative colitis [27–29]. Additionally, atractylone has been reported to modulate gut microbiota and alleviate gastric ulcers [30]. Given our findings that higher sesquiterpenoid concentrations correspond with increased plant age, we propose that longer cultivation periods enhance the medicinal efficacy and overall quality of A. macrocephala.

Impact of growth years on phenolic acids

Quantitative analysis and metabolomic profiling revealed that phenolic acid content, including chlorogenic acid increased with the growth years of A. macrocephala. Transcriptomic data further demonstrated the upregulation of genes involved in phenylpropanoid biosynthesis, such as PAL, 4CL, and HCT, in older plants. Chlorogenic acid, a key phenolic acid in A. macrocephala, is known to enhance plant resilience to drought, cold, oxidative stress, and pathogen defence [31–33]. These findings suggest that chlorogenic acid accumulation may play a role in mitigating environmental stress during plant development.

The biosynthetic regulation of phenolic acids has been extensively studied in other plant species. PAL, the first rate-limiting enzyme in chlorogenic acid biosynthesis, has been shown to promote its accumulation in transgenic sweet potato [34]. Similarly, HCT, a critical enzyme in this pathway, has been implicated in chlorogenic acid production in transgenic tobacco, where ectopic expression of its homolog increased chlorogenic acid content [35]. Based on these functional studies, we hypothesise that these genes serve as key regulatory elements in phenolic acid biosynthesis during A. macrocephala growth.

Phenolic acids, particularly chlorogenic acid and its derivatives, have diverse pharmacological properties. Chlorogenic acid has been shown to inhibit prostatic hyperplasia [36] and reduce cerebral ischemia-reperfusion injury [37]. Moreover, neochlorogenic acid and cryptochlorogenic acid exhibit anti-inflammatory effects [38–40]. Collectively, our findings indicate that the accumulation of sesquiterpenoids and phenolic acids in older A. macrocephala contributes to its superior medicinal quality.

Our experimental findings indicate that, based on principal component analysis of metabolomic profiles, 3-year-old A. macrocephala occupies an intermediate metabolic state between younger (1–2 years) and older (5–10 years) plants. After three years of growth, both metabolite and transcriptome profiles tend to stabilize. Therefore, we propose establishing the quality standard based on the active compound levels in 3-year-old samples: atractylenolide I (0.18 mg·g⁻¹), atractylone (0.72 mg·g⁻¹), and chlorogenic acid (1.41 mg·g⁻¹).

Impact of growth years on sucrose and lignin

Our results indicate a decline in sucrose content as A. macrocephala matures, with both metabolomic and quantitative analyses showing consistent trends. Transcriptomic analysis revealed that genes involved in sucrose biosynthesis, including SPS and SPP, were more highly expressed in younger plants (Y1 to Y3) than in older ones (Y5 and Y10). Sucrose serves as a primary metabolic substrate, supplying energy for organic matter production in non-photosynthetic tissues [41]. The observed reduction in sucrose content in older plants may be attributed to a shift from active metabolic processes to a more stable physiological state.

Additionally, sucrose does not exhibit notable pharmacological effects and is readily available in other dietary sources. Excessive sucrose content in A. macrocephala could dilute the relative concentration of bioactive compounds, potentially affecting medicinal efficacy. Our findings suggest that high-growth-year A. macrocephala accumulates higher levels of pharmacologically active compounds while maintaining lower sucrose content, which may enhance its clinical value.

Quantitative determination of lignin content revealed a gradual increase with plant age. Transcriptomic data further supported this trend, showing upregulation of key genes in the lignin biosynthesis pathway, including HCT, F5H, and CCR, with increasing growth time. Lignin, a major structural component of the xylem, plays a critical role in providing mechanical support, improving lodging resistance, and enhancing tolerance to biotic and abiotic stresses [42]. Its accumulation strengthens the plant structure, reduces stem breakage, deters herbivory, and facilitates water transport.

Previous studies have demonstrated the regulatory role of HCT [43], F5H [44], and CCR [45] in lignin biosynthesis. Based on our findings, we hypothesise that these genes are key modulators of lignin accumulation in A. macrocephala as it matures. Morphological changes associated with prolonged growth, such as increased plant length and the development of the characteristic “Hejing” form, may be supported by lignin deposition. Thus, lignin appears to be essential in maintaining structural integrity and facilitating the morphological adaptations associated with A. macrocephala quality.

Systematic summary of the effect of growth years on the quality of A. macrocephala

Our results demonstrate that morphology, microstructure, metabolite composition, and gene expression in A. macrocephala change dynamically with increasing age. Morphologically, the rhizome’s weight and length increased over time, while the upper diameter decreased, producing a more elongated mid-upper rhizome—a characteristic associated with the “Hejing” form and high-quality A. macrocephala.

In terms of microstructure, mature rhizomes displayed thickened xylem, providing structural support for rhizome elongation. Additionally, older samples exhibited looser cortical cell arrangements and a decreased density of oil chambers.

Metabolomic profiling via PCA showed that 1- and 2-year-old samples clustered together, as did the 5- and 10-year-old samples, with 3-year-old samples positioned between these two groups. Quantitative metabolite analysis revealed that mature A. macrocephala (5 and 10 years) accumulated higher levels of key bioactive compounds, including atractylenolide I, atractylone, and chlorogenic acid. In contrast, sucrose levels declined, likely reflecting older plants stable physiological state activity. Increased lignin levels in older samples were consistent with microstructural findings.

Transcriptomic analyses showed distinct clustering: Y1–Y3 grouped closely, while Y5–Y10 formed a separate cluster, suggesting transcriptional stabilization after year 5. Examination of genes involved in bioactive compound biosynthesis revealed upregulation of enzymes in the terpenoid (HMGR, DXS, idi, GPS, and GGPS) and phenylpropanoid (PAL, 4CL, HCT, F5H, and CCR) pathways in later growth years. Conversely, genes involved in sucrose biosynthesis (SPS and SPP) were downregulated. These findings align with the metabolite data, indicating that the growth time may affect the expression of these genes, and eventually form the scientific connotation of high-quality A. macrocephala.

Conclusions

This was the first study to comprehensively analyse A. macrocephala using morphological, microstructural, metabolomic, and transcriptomic approaches. Morphologically, the rhizomes of older plants had greater weight and length, with a shape resembling the traditional “Hejing” description, while microstructural analysis showed thickened xylem, looser cortex cells, and fewer oil compartments. Metabolomic and quantitative analyses revealed increased sesquiterpenoid and phenolic acid content with age, alongside decreasing sucrose and increasing lignin levels, consistent with transcriptomic data showing the upregulation of genes involved in terpenoid, phenylpropanoid, and lignin biosynthesis. DESI-MSI imaging indicated age-dependent shifts in fructan, quinic acid, and polyacetylene distribution. Overall, A. macrocephala quality improved with prolonged growth, stabilizing after five years, which provides insights into optimising its cultivation for enhanced medicinal value.

Abbreviations

TCM

Traditional Chinese medicine

DESI-MSI

Desorption electrospray ionisation mass spectrometry imaging

Y1

One years old A. macrocephala

Y2

Two years old A. macrocephala

Y3

Three years old A. macrocephala

Y5

Five years old A. macrocephala

Y10

Ten years old A. macrocephala

UPLC

Ultra-performance liquid chromatography

QC

Quality control

ESI

Electrospray ionisation

PCA

Principal component analysis

OPLS-DA

Orthogonal partial least square-discriminant analysis

DAMs

Differentially accumulated metabolites

VIP

Variable importance projection

CMC

Carboxymethyl cellulose

TPM

Transcripts Per Kilobase of exon model per Million mapped reads

DEGs

Differentially expressed genes

GO

Gene Ontology

KEGG

Kyoto Encyclopedia of Genes and Genomes

qRT-PCR

Quantitative real-time polymerase chain reaction

SD

Standard deviation

TIC

Total ion flow

GF4

1F-Fructofuranosylnystose

GF5

1,1,1,1-Kestohexaose

GF6

1,1,1,1,1-Kestoheptaose

PAT1

12-acetoxy-12-methylbulbu-tyl-2E,8Z,10E-trien-4,6-diyn-1-ol

PAT2

12-Senecioyloxytetradeca-2E,8E,10E-trien-4,6-diyne-1,14-diacetate

PAT3

14-acetoxy-12-methylbulbu-tyl-2E,8Z,10E-trien-4,6-diyn-1-ol

SPS

Sucrose-phosphate synthase

SPP

Sucrose-phosphatase

PAL

Phenylalanine ammonia-lyase

CCR

Cinnamyl CoA reductase

4CL

4-Coumarate:CoA ligase

HCT

Hydroxycinnamoyl transferase

F5H

Ferulate 5-hydroxylase

MVA

Mevalonate pathway

MEP

Methylerythritol phosphate pathway

IPP

Isopentenyl pyrophosphate

DMAPP

Dimethylallyl pyrophosphate

HMGR

HMG-CoA reductase

DXS

1-deoxy-D-xylulose-5-phosphate synthase

idi

Isopentenyl-diphosphate Delta-isomerase

GPS

Geranyl pyrophosphate synthase

GGPS

Geranylgeranyl diphosphate synthase

Authors’ contributions

X.C. wrote the manuscript and analyzed the data. X.C. and Y.W. performed the experiment. Y.L. validated and supervised the data. G.Y. and B.H provided the resources. L.H. and Z.Z. acquired the funding, designed the project and contributed to the review and editing of the manuscript. All authors have read and agreed to the published version of the manuscript.

Funding

This research was funded by the Scientific and Technological Innovation Project of China Academy of Chinese Medical Sciences (grant number: CI2023E002, CI2021A04003), Ministry of Finance Central Level of the Special (grant number 2060302).

Data availability

The datasets generated during the current study are available in the Genome Sequence Archive in National Genomics Data Center, under accession number CRA024154, which is publicly accessible at https://ngdc.cncb.ac.cn/gsa.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.