Integrated Analysis of Transcriptome and Metabolome Profiles in Astragslus membranaceus (Fisch.) Bge. var. mongholicus (Bge.) Hsiao Seedlings Under Drought Stress

Abstract

Background: Astragalus membranaceus is a traditional Chinese medicinal herb with significant pharmacological value. Drought stress adversely affects its biomass accumulation and medicinal quality. Methods: In this study, we performed physiological profiling, transcriptomics, and metabolomics analyses on A. membranaceus (Fisch.) Bge. var. mongholicus (Bge.) Hsiao seedlings to elucidate the response mechanisms in both aboveground and root tissues under varying drought stress intensities (Control, CK; Light Drought LD; Moderate Drought MD; Severe Drought SD). Results: Our findings indicate that LD primarily activated antioxidant enzymes, whereas severe stress led to the dominance of osmotic adjustment. Compared with CK, drought treatments resulted in 2987 differentially expressed genes (DEGs; 1674 up-regulated and 1313 down-regulated) and 921 differentially accumulated metabolites (DAMs)—562 in positive ionization mode (224 up, 338 down) and 359 in negative ionization mode (166 up, 193 down). Both gene expression and metabolite accumulation exhibited pronounced stress intensity-dependent patterns, suggesting that A. mongholicus initiates a broad, gene activation-led “active coping” strategy and mobilizes increasingly extensive metabolic pathways as drought intensifies. Conclusions: Integrated transcriptomic and metabolomic analyses revealed a tissue-specific “shoot–root partitioned coordination” mechanism: aboveground tissues activated a glutathione metabolism-centered “antioxidant–osmotic adjustment” defense, while root tissues reconfigured amino acid metabolism to maintain energy supply and signaling. This synergistic coordination represents a core adaptive strategy of A. mongholicus under drought conditions. Our study provides deeper insights into the drought resistance mechanisms of Astragalus and offers valuable references for breeding drought-tolerant varieties of Astragalus and other medicinal plants.

1. Introduction

Astragalus membranaceus, a herbaceous species belonging to the genus Astragalus in the Fabaceae family, is a renowned traditional Chinese medicinal plant. Its dried root, documented in the Chinese Pharmacopeia, is utilized for its diverse pharmacological properties, including reinforcing Qi, elevating Yang, consolidating the exterior to check sweating, promoting diuresis to reduce edema, generating fluid, nourishing blood, and removing obstructions to relieve pain [1]. This species is extensively cultivated in the arid and semi-arid regions of northwestern China [2]. The medicinal efficacy of A. membranaceus is primarily attributed to its bioactive compounds, such as astragalus polysaccharides, astragalosides, and flavonoids. The accumulation of these secondary metabolites directly determines the quality and therapeutic potential of the medicinal material [3]. However, the biosynthesis and accumulation of secondary metabolites are highly susceptible to environmental stresses [4]. Among these, drought stress stands as a major abiotic factor limiting both the yield and quality of A. membranaceus [5]. Traditionally, drought has been viewed as a purely negative factor that limits yield and quality. However, growing evidence suggests that for medicinal plants, moderate water deficit may not be entirely detrimental but could serve as an ecological elicitor that stimulates the production of active compounds and enhances medicinal quality [6,7]. Consequently, in cultivation, drought stress not only affects the biomass of A. membranaceus but also regulates the biosynthesis and accumulation of its pharmacologically active secondary metabolites, thereby influencing its medicinal quality and economic value. While field conditions are variable, controlled drought stress experiments allow for the precise dissection of the physiological and molecular cascades triggered by specific water deficits [8]. This approach is key to resolving the conflict between yield and quality in A. membranaceus production and achieving sustainable resource utilization.

To cope with drought stress, plants activate a complex array of physiological, biochemical, and molecular response mechanisms. Physiologically, these responses often include the inhibition of photosynthesis, accumulation of osmoregulatory substances, reactive oxygen species (ROS) burst, and activation of the antioxidant system [9,10]. At the molecular level, drought triggers the transcriptional regulation of stress-responsive genes and the activation of key signaling transduction pathways, such as those involving phytohormones and MAPK cascades [11]. These molecular changes lead to a comprehensive reprogramming of metabolic networks, including significant alterations in both primary and secondary metabolism [12], which collectively function to maintain cellular homeostasis and enhance stress tolerance. Therefore, deciphering the link between transcriptional regulation and metabolic reprogramming under drought stress is crucial for elucidating the core molecular mechanisms of drought resistance.

Recent advances in high-throughput omics technologies have provided powerful tools for systematically dissecting plant stress response mechanisms. Transcriptomics enables the comprehensive profiling of gene expression patterns under specific conditions, revealing key stress-responsive genes, transcription factors, and signaling pathways [11,13]. Metabolomics, which analyzes the composition and abundance of metabolites, serves as a critical link connecting transcriptomic changes to phenotypic manifestations [14,15]. Integrated analysis of transcriptome and metabolome data can effectively correlate gene expression with metabolite abundance, facilitating the construction of “gene-enzyme-metabolite” regulatory networks and thereby providing a more systematic understanding of the molecular mechanisms underlying plant responses to environmental stress. This multi-omics approach has been successfully applied in drought resistance research on various crops, including Glycine max (Linn.) Merr. [11], Triticum aestivum L. [15], Zea mays L. [16], and Ipomoea batatas (L.) Lam. [14], as well as in medicinal plants such as Dendrobium nobile Lindl. [17] and Salvia miltiorrhiza Bunge [13]. However, most of these studies have focused on general mechanisms of plant tolerance or survival strategies, with limited emphasis on the targeted regulation of secondary metabolites and its association with medicinal quality. In A. membranaceus, existing research remains largely confined to physiological and biochemical indices—such as osmoregulatory substance content and antioxidant enzyme activities—and expression analyses of a few key genes [2,18,19]. To date, no systematic investigation has been conducted on the tissue-specific regulatory architecture underlying drought responses across a continuous stress gradient. Here, by integrating transcriptomic and metabolomic profiling of both aboveground and underground tissues, we reveal a tissue-partitioned coordination model in A. mongholicus that is mechanistically distinct from the whole-plant synchronized responses or single-tissue analyses reported in other legumes [20,21]. This model mechanistically couples antioxidant defense (aboveground) from homeostatic regulation (underground), representing a functional specialization, metabolic complementarity, and systemic integration—an adaptive strategy uniquely evolved in Astragalus under arid conditions. Evidence suggests that drought stress exhibits a “dose effect”; there may exist an optimal stress intensity “window” within which stress can maximally activate defense and secondary metabolism-related pathways without causing severe biomass loss [6]. Leveraging a gradient-based experimental framework, we demonstrate that moderate drought (MD) specifically enriches pathways such as glutathione metabolism and phenylpropanoid biosynthesis—both of which serve as upstream hubs for the biosynthesis of pharmaceutically active compounds, including flavonoids and astragalosides, in A. mongholicus. This finding provides direct molecular evidence supporting the agronomic concept that “moderate stress enhances medicinal quality.”

The seedling stage was selected as the experimental system in this study, based on evidence that roots of A. membranaceus seedlings possess the complete metabolic capacity for biosynthesis of medicinal components such as flavonoids and saponins, and that this capacity can be efficiently induced by drought stress in a clear dose-dependent manner [18,22]. Furthermore, the seedling stage represents a critical period for polysaccharide accumulation in Astragalus, during which the metabolic pathways underlying total flavonoids and astragaloside IV are largely established. The final accumulation levels of these compounds in adult roots are highly dependent on the scale of the metabolic network and the sensitivity of stress-induced responses established during the seedling stage [23,24]. Therefore, in-depth elucidation of the regulatory mechanisms operating at the seedling stage can provide theoretically grounded and technically actionable targets for quality-oriented cultivation of two-year-old commercial roots. Although extrapolation of findings from seedlings to adult plants requires further validation, the tissue-partitioned coordination model proposed in this study is reasonably expected, based on the developmental continuity of secondary metabolic pathways, to operate during drought responses and quality formation in mature plants. This inference outlines a clear direction for subsequent validation and translational research.

Based on this premise, this study simulated field moisture fluctuation gradients by establishing a series of drought treatments ranging from mild to severe. An integrated approach combining physiological parameter measurement, transcriptome sequencing (RNA-Seq), and untargeted metabolomics was applied with the following objectives: (1) to systematically evaluate the impact of drought stress on the physiological indicators of A. mongholicus seedlings and characterize their physiological response phenotypes; (2) to delineate a comprehensive multi-omics response atlas of A. mongholicus seedlings to gradient drought stress, thereby revealing their adaptive strategies under drought; and (3) to identify the potential optimal stress intensity that can maximally stimulate pathways related to medicinal quality while minimizing growth inhibition. This research aims not only to elucidate the molecular mechanisms of drought resistance in A. membranaceus but also, at a deeper level, to provide a solid theoretical and data foundation for precisely regulating the quality of A. membranaceus medicinal materials through the green cultivation strategy of “controlled water stress,” and for subsequently enhancing both drought tolerance and medicinal quality via molecular breeding approaches.

2. Materials and Methods

2.1. Plant Materials and Stress Treatments

Seeds of A. membranaceus (Fisch.) Bge. var. mongholicus (Bge.) Hsiao (Longxi County Seed Station, Longxi, Gansu Province, China) were sown on 13 September 2024, in a greenhouse. The seedling stage was selected for its high stress sensitivity and active secondary metabolism. Uniform seedlings (6.0 ± 0.5 cm height) were transplanted into individual pots (15 cm × 12 cm) containing a Pindstrup peat (Pindstrup Mosebrug A/S, Ryomgaard, Denmark) and perlite (3:1, v/v) substrate on September 29 and grown under uniform conditions. On October 22 (three weeks post-transplant), four drought stress levels were initiated using the daily weighing method: Control (CK, 80–85% soil water capacity), Light Drought (LD, 65–70%), Moderate Drought (MD, 50–55%), and Severe Drought (SD, 35–40%). This gradient encompasses the range from optimal irrigation to severe water deficit, simulating soil moisture fluctuations likely to occur during the growth stages of this species. After 18 days (November 9), aboveground (shoot) and underground (root) tissues were separately harvested from three biological replicates per treatment (24 samples total), immediately frozen in liquid nitrogen, and stored at −80 °C. Samples were labeled CKA, LDA, MDA, SDA (shoots) and CKR, LDR, MDR, SDR (roots).

2.2. Determination of Physiological Parameters

Key physiological parameters, including the activities of catalase (CAT) and peroxidase (POD), as well as the contents of malondialdehyde (MDA), proline (Pro), soluble protein (SP), and soluble sugar (SS), were assessed in the seedlings from all four treatment groups. The activities of CAT and POD, and the MDA content were measured using specific commercial assay kits (D799598-0100 for CAT, D799592-0100 for POD, and D799762-0100 for MDA; Sangon Biotech (Shanghai) Co., Ltd., Shanghai, China), strictly following the manufacturer’s instructions. The Pro content was determined according to the acid–ninhydrin method [25]. The SP content was quantified using the Coomassie Brilliant Blue method [25], and the SS content was analyzed via the anthrone–sulfuric acid method [25].

It should be noted that the present study focuses on the physiological, biochemical, and molecular regulatory networks of A. mongholicus seedlings in response to drought stress, with the aim of elucidating early stress signal transduction and metabolic reprogramming mechanisms. Growth-related traits (e.g., biomass, root-to-shoot ratio, plant height) were not included in the measurement scope. Accordingly, the conclusions drawn herein are primarily confined to the molecular and physiological levels. Any extrapolation of these molecular events to yield performance or agronomic traits requires further validation through integrated studies incorporating growth parameters.

2.3. Transcriptome Sequencing and Analysis

Transcriptome library construction and sequencing were performed in collaboration with Novogene Co., Ltd. (Beijing, China). Total RNA was extracted using the RNAprep Pure Plant Plus Kit (Polysaccharides & Polyphenolics-rich) (TIANGEN BIOTECH (BEIJING) CO., LTD., Beijing, China). RNA integrity and concentration were assessed using an Agilent 2100 Bioanalyzer (Agilent Technologies Inc., Santa Clara, CA, USA). Following quality control, mRNA was enriched from total RNA using oligo(dT) magnetic beads and fragmented randomly via divalent cations in a high-temperature environment. First-strand cDNA was synthesized using random hexamer primers, followed by second-strand cDNA synthesis. The resulting double-stranded cDNA underwent end repair, poly(A) tailing, adapter ligation, size selection, Polymerase Chain Reaction (PCR) amplification, and purification to generate the final sequencing libraries. Library quantification was performed using a Qubit fluorometer and real-time quantitative PCR, while the insert size distribution was verified using a Bioanalyzer. Qualified libraries were pooled according to effective concentration and the desired data volume, followed by sequencing on an Illumina platform to generate raw sequence reads.

Raw reads were processed with the fastp (v0.23.2) software to remove adapter sequences, reads containing poly-N regions, and low-quality reads, yielding clean reads. The quality of clean reads was evaluated by calculating the Q20, Q30 scores, and GC content. All subsequent analyses were based on these high-quality clean reads.

For transcriptomic analysis, we employed a reference genome-based strategy. The obtained clean reads were aligned to the reference genome of A. membranaceus (Fisch.) Bge. var. mongholicus (Bge.) Hsiao (genome version: Global Pharmacopoeia Genome Database (GPGD): http://www.gpgenome.com/species/109 (accessed on 14 April 2024)) using HISAT2 (v2.0.5) to obtain mapping information. Gene expression levels were then quantified as FPKM (Fragments Per Kilobase of transcript per Million mapped reads) and raw read counts using featureCounts (v1.5.0-p3). Differential expression analysis was conducted with the DESeq2 R package (v1.20.0). The Benjamini–Hochberg method was applied to control the false discovery rate (FDR). Genes with an FDR ≤ 0.05 and an absolute |log₂(Fold Change)| ≥ 1 were identified as differentially expressed genes (DEGs). Functional enrichment analysis of DEGs, including Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analyses, was performed using the clusterProfiler R package (v3.8.1), with a significance threshold of FDR ≤ 0.05.

While this reference-based approach enables efficient and accurate alignment of sequencing reads to known gene loci, it has inherent limitations. Firstly, the depth of analysis is directly dependent on the completeness and annotation quality of the reference genome [26]. Gaps or incomplete annotations could lead to the failure in detecting some Astragalus-specific, yet unannotated, stress-responsive genes. Secondly, this strategy may not fully capture transcript diversity arising from alternative splicing [27,28]. Future studies could incorporate a de novo assembly approach to construct sample-specific transcriptomes [29]. Nevertheless, within the current context, utilizing the available reference genome remains a valid and effective approach for reliable quantitative comparative analysis.

To validate the transcriptome data, the expression levels of 14 selected genes were analyzed by quantitative real-time PCR (qRT-PCR). First, total RNA was reverse-transcribed into cDNA using the PrimeScript™ RT Reagent Kit with gDNA Eraser (TaKaRa, Code No. RR047A). Specific primers for the target genes and a reference gene were designed (Supplementary Table S1). qRT-PCR was performed using TB Green Premix Ex Taq II (Tli RNaseH Plus) (TaKaRa, Code No. RR820A/B) on a CFX96 Real-Time PCR Detection System. The thermal cycling protocol consisted of an initial denaturation at 95 °C for 30 s, followed by 39 cycles of 95 °C for 5 s and 58 °C for 30 s. A melt curve analysis was subsequently conducted to verify amplification specificity. Relative gene expression was calculated using the 2^−ΔΔCt method.

2.4. Metabolomics Profiling and Analysis

Untargeted metabolomics analysis was performed based on Liquid Chromatography–Mass Spectrometry (LC-MS) technology in collaboration with Novogene Co., Ltd. (Beijing, China). The procedure primarily included metabolite extraction from samples, LC-MS detection, and data analysis. Approximately 100 mg of sample, ground in liquid nitrogen, was weighed into a centrifuge tube. Then, 500 μL of an 80% methanol aqueous solution was added, followed by vortex mixing and incubation in an ice bath for 5 min. After centrifugation at 15,000× g and 4 °C for 20 min, the supernatant was collected. Mass spectrometry-grade water was added to dilute the methanol content to 35%. The mixture was centrifuged again under the same conditions (15,000× g, 4 °C, 20 min), and the resulting supernatant was collected for LC-MS analysis [30]. Chromatographic conditions employed a Hypersil Gold column (C18) maintained at 40 °C, with a flow rate of 0.2 mL/min. Mobile phase A consisted of 0.1% formic acid in water, and mobile phase B was methanol. Mass spectrometric conditions were as follows: scan range m/z 100–1500; ESI ion source with spray voltage set to 3.5 kV; sheath gas flow rate at 35 psi; auxiliary gas flow rate at 10 L/min; capillary temperature at 320 °C; S-lens RF level at 60; auxiliary gas heater temperature at 350 °C; data were acquired in both positive and negative ionization modes.

Raw data files were converted to mzXML format using ProteoWizard. Peak picking, peak alignment, and quantification were performed using the XCMS software. Metabolite identification was carried out by matching accurate mass (within 10 ppm tolerance) and MS/MS spectra against high-quality spectral databases. Compounds showing a coefficient of variation (CV) of relative peak area greater than 30% in quality control (QC) samples were removed, resulting in a final list of identified and relatively quantified metabolites. The metaX (v2.0.0) software was used for subsequent data preprocessing, including log-transformation and normalization. Principal Component Analysis (PCA) and Partial Least Squares-Discriminant Analysis (PLS-DA) were then performed. The Variable Importance in Projection (VIP) score for each metabolite was obtained from the PLS-DA model. Student’s t-test was applied to calculate the statistical significance (p-value) and fold change (FC) of metabolites between comparison groups. Differentially accumulated metabolites (DAMs) were identified using the following criteria: VIP > 1.0, p-value < 0.05, and absolute FC > 1.2 (equivalent to FC < 0.833 for down-accumulation). Identified metabolites were annotated against the KEGG database (https://www.genome.jp/kegg/pathway.html (accessed on 1 December 2024)). Enrichment analysis of KEGG pathways for the DAMs was conducted using a hypergeometric test to determine the primary biochemical metabolic pathways and signal transduction pathways significantly associated with the drought stress response.

2.5. Integrated Analysis of Transcriptome and Metabolome

DEGs and DAMs were jointly mapped to the KEGG pathway database to identify pathways commonly involved in the response to drought stress. KEGG pathway maps were constructed and visualized to elucidate the relationship between changes in gene expression and the accumulation of key metabolites.

2.6. Statistical Analysis

Multivariate analysis of variance was performed using Statistical Package for the Social Sciences (SPSS) (version 26.0) to assess significant variations in physiological indicators under drought stress, with a statistical significance level set at p < 0.05. Graphical representations of the data were generated using Origin software (2024 version). Pathway diagrams and figure composites were assembled using Adobe Illustrator (2024 version).

3. Results

3.1. Effects of Drought Stress on Physiological Parameters in A. mongholicus Seedlings

To characterize the physiological response of A. mongholicus seedlings to drought stress, we measured the dynamics of key physiological indices, including the activities of CAT and POD, as well as the contents of MDA, Pro, SP, and SS (Figure 1). With increasing drought intensity, CAT activity exhibited an initial increase followed by a decline, whereas POD activity showed a more complex pattern of increase, subsequent decrease, and then a further increase. Notably, both enzyme activities were significantly higher (p < 0.05) in the aboveground tissues under LD compared to other treatments, while their maximum activities in the roots were observed under SD. In contrast, the contents of MDA, Pro, SP, and SS generally decreased initially and then increased as the stress intensified. The accumulation of Pro and SS under SD was significantly greater (p < 0.05) than in other treatments. MDA content in the aboveground tissues was significantly highest (p < 0.05) under severe stress, while SP content did not show statistically significant changes across different stress levels (p > 0.05). These results demonstrate that A. mongholicus seedlings actively respond to drought stress by modulating a suite of physiological processes, with the specific adaptive strategies varying according to the intensity of the stress.

3.2. Transcriptomic Analysis

3.2.1. Summary of RNA-Seq Data

To gain deeper insights into the molecular mechanisms underlying the drought stress response in A. mongholicus, high-throughput RNA sequencing was performed on 24 samples subjected to different drought stress levels. The sequencing output statistics are summarized in

Table 1. Summary of RNA-Seq data for Astragalus membranaceus.

Sample	Raw Reads	Clean Reads	Clean Bases	Error (%)	Q20 (%)	Q30 (%)	GC (%)
SDR1	42,265,268	40,928,962	6.14 G	0.01	98.84	96.52	42.44
SDR2	39,733,866	38,278,224	5.74 G	0.01	98.76	96.25	42.65
SDR3	42,298,822	41,093,978	6.16 G	0.01	98.85	96.52	42.55
SDA1	48,648,726	46,494,144	6.97 G	0.01	98.90	96.61	42.39
SDA2	39,821,624	38,577,916	5.79 G	0.01	98.88	96.55	42.31
SDA3	44,798,636	43,712,592	6.56 G	0.01	98.90	96.60	42.45
MDR1	48,493,100	46,969,794	7.05 G	0.01	98.88	96.61	42.42
MDR2	41,243,254	39,852,906	5.98 G	0.01	98.84	96.45	42.95
MDR3	41,458,764	40,361,504	6.05 G	0.01	98.83	96.45	42.75
MDA1	46,831,264	45,968,072	6.90 G	0.01	98.87	96.53	42.20
MDA2	42,184,044	41,276,748	6.19 G	0.01	98.87	96.54	42.87
MDA3	48,706,618	47,334,464	7.10 G	0.01	98.86	96.53	42.31
LDR1	50,693,500	49,253,284	7.39 G	0.01	98.85	96.50	42.87
LDR2	46,895,612	45,400,594	6.81 G	0.01	98.87	96.59	42.78
LDR3	44,194,402	43,096,590	6.46 G	0.01	98.76	96.23	42.66
LDA1	43,150,208	42,400,178	6.36 G	0.01	98.89	96.60	42.25
LDA2	40,818,296	39,160,824	5.87 G	0.01	98.84	96.48	42.30
LDA3	44,913,246	42,012,312	6.30 G	0.01	98.80	96.35	43.04
CKR1	44,724,984	43,638,348	6.55 G	0.01	98.81	96.40	43.01
CKR2	39,646,980	38,136,336	5.72 G	0.01	98.83	96.43	42.42
CKR3	44,483,034	43,284,418	6.49 G	0.01	98.81	96.38	42.65
CKA1	42,020,404	41,080,712	6.16 G	0.01	98.83	96.41	42.26
CKA2	45,787,724	44,006,644	6.60 G	0.01	98.85	96.50	42.30
CKA3	44,096,718	42,750,674	6.41 G	0.01	98.86	96.50	42.28

Note: Raw reads: the number of raw data reads; Clean reads: the number of reads filtered by the raw data; Clean bases: the number of bases filtered from the raw data, Error: sequencing error rate, Q20/Q30: bases with a Phred value greater than 20/30 as a percentage of total bases, GC: G and C are the percentages of the four bases in clean reads.

. The results showed that the raw reads generated per sample ranged from 39,646,980 to 50,693,500. After quality control, 38,136,336 to 49,263,284 clean reads were obtained per library, yielding a total of 153.75 Gb of high-quality clean data. The clean data volume for individual samples varied between 5.72 Gb and 7.39 Gb. The Q20 and Q30 scores ranged from 98.76% to 98.90% and 96.23% to 96.61%, respectively, while the GC content was in the range of 42.20% to 43.04%. These metrics collectively indicate that the sequencing data were of high quality and suitable for subsequent analyses.

3.2.2. Identification of DEGs

Six pairwise comparisons were established to identify DEGs, comprising three for aboveground tissues (LDA vs. CKA, MDA vs. CKA, SDA vs. CKA) and three for root tissues (LDR vs. CKR, MDR vs. CKR, SDR vs. CKR). DEGs were screened using thresholds of FDR < 0.05 and |log₂ Fold Change| ≥ 1. As shown in Figure 2, the comparisons LDA vs. CKA, MDA vs. CKA, SDA vs. CKA, LDR vs. CKR, MDR vs. CKR, and SDR vs. CKR yielded 238, 860, 1078, 155, 251, and 405 DEGs, respectively. Among these, 79, 394, 681, 99, 153, and 268 were up-regulated, while 159, 466, 397, 56, 98, and 137 were down-regulated. The magnitude of DEGs reflects the transcriptional responsiveness and complexity of reprogramming in response to drought stress. Further analysis revealed numerous unique DEGs within the comparison groups for each tissue. In the aboveground tissues, the LDA vs. CKA, MDA vs. CKA, and SDA vs. CKA comparisons contained 112, 557, and 800 unique DEGs, respectively, with 47 DEGs common to all three aboveground comparisons. In the root tissues, the LDR vs. CKR, MDR vs. CKR, and SDR vs. CKR comparisons contained 107, 182, and 316 unique DEGs, respectively, sharing only 4 common DEGs across all root comparisons. These data demonstrate pronounced stress intensity-dependent and tissue-specific patterns in both the number and direction of DEGs. Specifically, the number of DEGs progressively increased with drought intensity in both aboveground and underground tissues. Notably, underground tissues exhibited a predominantly upregulated transcriptional response across all stress levels. In contrast, aboveground tissues were dominated by downregulated genes under LD and MD, shifting to predominantly upregulated genes under SD. This divergent transcriptional pattern indicates that aboveground and underground tissues employ distinct response strategies at the transcriptomic level.

3.2.3. GO and KEGG Enrichment Analysis of Differentially Expressed Genes

Gene Ontology (GO) enrichment analysis was performed to functionally characterize the DEGs under different drought stress conditions, using a significance threshold of FDR < 0.05. The results are presented in Figure 3. In aboveground tissues, DEGs identified in the LDA vs. CKA comparison were significantly enriched in terms related to polysaccharide metabolic process, cellular glucan metabolic process, cell wall and external encapsulating structure (cellular component), and hydrolase activity acting on glycosyl bonds (molecular function). Under moderate drought (MDA vs. CKA), enrichments shifted towards glucan metabolism, the extracellular region, and transferase activity transferring hexosyl groups. Severe drought (SDA vs. CKA) triggered a pronounced enrichment in processes and components associated with photosynthesis and carbohydrate catabolism, specifically the thylakoid compartment, alongside oxidoreductase activity. In underground tissues, the response pattern differed. Under light drought (LDR vs. CKR), DEGs were enriched in aminoglycan metabolic and catabolic processes, cell wall components, and activities such as chitinase and O-methyltransferase. Moderate drought (MDR vs. CKR) induced enrichment in phospholipid metabolism (e.g., phosphatidylinositol and glycerophospholipid processes), components of the endoplasmic reticulum membrane, and specific ribonuclease activities. Under severe drought (SDR vs. CKR), the response was dominated by oxidative stress response, alpha-amino acid biosynthesis, ribosome-related components, and transcription factor activity.

Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis was performed to identify significantly altered biological pathways in response to drought stress. Scatter plots of the top 20 most significantly enriched pathways for each comparison are presented in Figure 4. In aboveground tissues, DEGs under LD (Figure 4A) were predominantly enriched in pathways related to Linoleic acid metabolism, Tryptophan metabolism, and Circadian rhythm—plant. Under MD (Figure 4B), the most significantly enriched pathways included Circadian rhythm—plant, Photosynthesis—antenna proteins, and Phenylpropanoid biosynthesis. During SD (Figure 4C), the most reliably enriched pathways were Photosynthesis and Photosynthesis—antenna proteins. In underground tissues, a distinct enrichment pattern was observed. Under LD (Figure 4D), DEGs were mainly enriched in Cyanoamino acid metabolism, MAPK signaling pathway—plant, and Glutathione metabolism. MD (Figure 4E) resulted in significant enrichment of Circadian rhythm—plant, Vitamin B6 metabolism, and Cyanoamino acid metabolism. Under SD (Figure 4F), the most significantly enriched pathways were Ribosome, Cyanoamino acid metabolism, and Mismatch repair.

The combined GO and KEGG analyses revealed a coordinated, multi-layered response. In roots, the significant enrichment of the MAPK signaling pathway and Glutathione metabolism under light drought constitutes a proactive defense module, facilitating early stress signal perception/transduction and reactive oxygen species (ROS) scavenging, respectively. The consistent and significant enrichment of various amino acid metabolism pathways across all stress levels is directly associated with the activation of osmotic adjustment. In aboveground tissues, the significant enrichment of the Phenylpropanoid biosynthesis pathway under moderate drought is notable, as this pathway is central to producing defensive secondary metabolites like lignin (for structural fortification) and flavonoids. Conversely, the pronounced enrichment of core photosynthesis-related pathways in aboveground tissues under severe drought indicates substantial physiological impairment. Similarly, the enrichment of the Ribosome pathway in roots under severe stress likely reflects a cellular state of enhanced protein turnover and repair, pointing toward a damage control and mitigation response.

3.2.4. Validation of RNA-Seq Data by qRT-PCR

The expression levels of 14 randomly selected genes were measured by qRT-PCR to validate the transcriptome data. The results, shown in Figure 5, were highly consistent with the RNA-seq data, which validated the accuracy and reliability of the transcriptome analysis conducted in this study.

3.3. Metabolomic Analysis

3.3.1. PCA and PLS-DA of Metabolite Profiles

To assess the stability and variability of the metabolomic data, Pearson correlation analysis was first performed on the quality control (QC) samples (Supplementary Figure S1). Subsequently, Principal Component Analysis (PCA) was applied to evaluate the metabolic differences among sample groups (Figure 6). The Pearson correlation analysis revealed that the correlation coefficients among QC samples all exceeded 0.994. The PCA results demonstrated clear separation between the control groups (CKA, CKR) and the treatment groups (LDA, MDA, SDA, LDR, MDR, SDR) within the three-dimensional score plot. These findings indicate high reliability and reproducibility of the metabolomic data, supporting their use in subsequent analyses.

PLS-DA was employed to construct a model describing the relationship between metabolite abundance and sample categories, facilitating sample classification. PLS-DA models were established for each comparison group in both positive and negative ionization modes (Figure 7). The results showed that the explained variance (R²Y) was consistently greater than the predictive ability (Q²Y) for all models, indicating good model quality. Furthermore, permutation testing (200 permutations) was conducted to validate the PLS-DA models (Supplementary Figure S2). The validation results, where the R² value was higher than the Q² value and the intercept of the Q² regression line with the Y-axis was below zero, confirmed that the models were not overfitted and possessed robust explanatory power for the sample data.

3.3.2. Screening and Identification of DAMs

DAMs were screened using the criteria of Fold Change >1.2 or <0.833, VIP > 1.0, and p-value < 0.05. A large number of DAMs responsive to gradient drought stress were identified in both positive and negative ion modes (Supplementary Figure S3 and Figure 8), and the number of DAMs showed a positive correlation with stress intensity. Notably, glutathione, a core antioxidant and sulfur reservoir, was significantly up-regulated in multiple comparisons, especially in aboveground tissues. In root tissues, DAMs associated with amino acid metabolism (e.g., 3-Methoxy-4-hydroxyphenylglycol glucuronide) and energy metabolism (e.g., Undecanedioic acid, Diethyl (2S,3R)-2-methyl-3-hydroxysuccinate) underwent significant changes. Furthermore, significant alterations were observed in DAMs involved in growth regulation (e.g., 3-Indoleacetonitrile) and defensive glycosides (e.g., Alangionoside O). These collective changes constitute the fundamental physiological and biochemical adaptive basis of A. mongholicus in response to drought. Critically, metabolites closely associated with the biosynthesis of Astragalus’s primary bioactive compounds showed distinct drought-responsive patterns. 4H-1-Benzopyran-4-one derivatives were significantly altered under light drought, while 1,3,5,6-Tetrahydroxy-8-methylxanthone was markedly up-regulated under moderate stress. Both metabolites are intimately linked to the phenylpropanoid/flavonoid biosynthesis pathway, confirming that drought stress effectively drives the metabolic flux of flavonoid compounds in A. mongholicus. Additionally, significant changes were detected in precursors and intermediates related to saponin biosynthesis, including the steroidal saponin precursor Dioscin and various terpenoids (e.g., Pseudolaric Acid C, Vielanin A). These findings strongly indicate that drought stress profoundly perturbs the precursor supply and metabolic network for the biosynthesis of astragalosides.

3.3.3. Global Analysis of Altered Metabolic Pathways Based on KEGG

DAMs were annotated against the KEGG database to identify the primary biochemical and signal transduction pathways involved in the drought response. Scatter plots of the top 20 most significantly enriched pathways (based on p-value) are presented in Figure 9. In aboveground tissues, the most significantly enriched pathways under LD (Figure 9A) were Glutathione metabolism, Cysteine and methionine metabolism, Linoleic acid metabolism, ABC transporters, and Metabolic pathways. Under MD (Figure 9B), enrichment was highest for Glucosinolate biosynthesis, Valine, leucine and isoleucine degradation, 2-Oxocarboxylic acid metabolism, Cutin, suberine and wax biosynthesis, and Valine, leucine and isoleucine biosynthesis. Under SD (Figure 9C), the most significantly altered pathways included Pyruvate metabolism, Fatty acid degradation, Flavone and flavonol biosynthesis, ABC transporters, and Fatty acid elongation. In root tissues, the DAMs under LD (Figure 9D) were predominantly enriched in Pyrimidine metabolism, Ascorbate and aldarate metabolism, Pentose and glucuronate interconversions, Nicotinate and nicotinamide metabolism, and Tropane, piperidine and pyridine alkaloid biosynthesis. Under MD (Figure 9E), the most significantly enriched pathways were Monobactam biosynthesis, Lysine degradation, Tyrosine metabolism, Isoflavonoid biosynthesis, and Lysine biosynthesis. Under SD (Figure 9F), the top enriched pathways included Oxidative phosphorylation, D-Arginine and D-ornithine metabolism, Arginine and proline metabolism, Histidine metabolism, and Fructose and mannose metabolism. These results demonstrate that the KEGG pathways enriched with DAMs varied substantially depending on the tissue type and the intensity of the drought stress.

3.4. Integrated Analysis of Transcriptome and Metabolome

Through integrated analysis of transcriptomic and metabolomic data, combined with KEGG pathway enrichment of both differentially expressed genes (DEGs) and differentially accumulated metabolites (DAMs) and nine-quadrant correlation analysis (Figure S4), we constructed a molecular response network of A. membranaceus seedlings to drought stress (Figure 10). The comparisons LDA vs. CKA, MDA vs. CKA, and SDA vs. CKA showed enrichment in 4, 9, and 28 KEGG pathways, respectively (Table S2). Notably, 16 DAMs and 18 DEGs were co-enriched in the Glutathione metabolism and Cysteine and methionine metabolism pathways, which formed a synergistically responding core module (Figure 10A). Within this network, genes encoding glutathione S-transferase (GST) and peroxiredoxin 6 (PRDX6) showed a significant positive correlation with the accumulation of antioxidant metabolites like glutathione, establishing a self-reinforcing antioxidant cycle. This module is therefore defined as a core adaptive response. Furthermore, this sulfur-metabolic network supplies key precursors for the biosynthesis of ethylene and polyamines. The coordinated changes in ethylene synthesis genes (ACS, ACO) and polyamine metabolites (e.g., Spermidine) underscore its dual role in coordinating stress signal transduction and osmotic adjustment. Importantly, sulfur metabolism also forms the basis for synthesizing sulfur-containing secondary metabolites. Consequently, the activation of this network is intrinsically linked to the potential regulation of Astragalus’s medicinal quality, as it provides precursors for bioactive compounds like astragalosides. In root tissues, the comparisons LDR vs. CKR, MDR vs. CKR, and SDR vs. CKR were enriched in 2, 2, and 5 KEGG pathways, respectively (Table S2), with the majority related to amino acid metabolism, including Tryptophan, Tyrosine, Histidine, and Alanine/aspartate/glutamate metabolism (Figure 10B). A total of 35 DAMs and 5 DEGs participated in these pathways, confirming amino acid metabolism as the central hub of the root drought response. The up-regulation of metabolites like L-Tryptophan and 3-Indoleacetonitrile, the down-regulation of amino acids such as L-Histidine, L-Aspartate, and L-Glutamate, coupled with the up-regulation of genes like PrAO and IGPD and the down-regulation of ACAT and ASS, collectively delineate a network dedicated to homeostasis maintenance and damage mitigation. The nine-quadrant correlation analysis (Figure S4) revealed significant associations between DEGs and DAMs, with most pairs located in quadrants I, III, VII, and IX. The positive correlations observed in quadrants III and VII suggest potential activating regulation, while the negative correlations in quadrants I and IX indicate potential inhibitory effects. These patterns confirm direct or indirect causal links between transcriptional regulation and metabolic reprogramming in driving the drought response.

These analyses demonstrate that A. mongholicus seedlings employ a dual-pathway coordination mechanism. The aboveground tissue operates via a glutathione–sulfur amino acid metabolism network, where the coordinated up-regulation of genes and metabolites enhances ROS scavenging and osmotic adjustment. Simultaneously, this network is critically linked to quality improvement by supplying precursors for medicinal astragalosides. The underground tissue relies on a core amino acid metabolism network, which optimizes energy allocation and stress signaling through the regulation of nitrogen transport and secondary metabolism-related genes (PrAO, IGPD, ACAT, ASS) and the reprogramming of amino acid catabolism, thereby maintaining homeostasis and ensuring basic survival under stress.

4. Discussion

4.1. Adaptive Physiological Responses of A. mongholicus Seedlings to Drought Stress

Plants respond to drought stress by regulating their antioxidant enzyme systems and osmotic adjustment mechanisms [31]. Key antioxidant enzymes, such as POD and catalase CAT, play crucial roles in plant responses to abiotic stresses [18]. Significant increases in POD and CAT activities under drought have been observed in the leaves of Bupleurum chinense [32] and in POD activity in S. miltiorrhiza leaves [13], indicating a common strategy among medicinal plants to enhance antioxidant capacity. MDA, a direct product of membrane lipid peroxidation, serves as a marker for oxidative damage [18]. Previous studies on Astragalus have shown that POD and CAT activities in A. membranaceus generally increase with drought intensity, while MDA content increases in both A. mongholicus and A. membranaceus, albeit with different magnitudes and potential suppression under severe stress [19]. Important osmolytes such as Pro [2], SS, and SP [33] function in osmotic protection and are vital for maintaining cellular water status and physiological activity [33]. Our study confirms the presence of these classical responses in A. mongholicus. More importantly, through tissue-specific analysis, we reveal a unique spatiotemporal coordination pattern in this species. Under LD and MD conditions, aboveground tissues preferentially activated antioxidant enzymes including CAT and POD to safeguard photosynthetic machinery, concurrent with comparatively low MDA levels and moderate accumulation of osmolytes. This physiological signature indicates that LD and MD represent an “acclimation window”, wherein defense systems are effectively engaged, membrane lipid peroxidation remains constrained, and resources continue to be allocated toward growth. In stark contrast, SD triggered a distinct shift: the epicenter of antioxidative defense relocated to underground tissues, accompanied by sharp MDA elevation and robust accumulation of Pro and soluble sugars SS—hallmarks of “metabolic stress and incipient tissue damage”. The strategy can be summarized as “early warning in shoots, deep tolerance in roots”: antioxidant enzyme activation dominates in shoots under light stress, while reliance on root POD activity and osmolyte (Pro, SS) accumulation prevails under severe stress. This mechanism represents an efficient resource allocation strategy. The rapid antioxidant response in aboveground tissues aims to protect the core photosynthetic machinery, preserving potential for overall recovery. Conversely, the reinforcement of the antioxidant system (POD) and accumulation of osmolytes in roots under extreme stress directly safeguard water uptake and cellular integrity, which is critical for survival. Unlike the decrease in root osmolytes observed in plants like B. chinense under drought [32], the “active stockpiling” strategy in A. mongholicus roots may be related to its deep-rooted nature and the need to maintain a specific metabolic environment as a medicinal storage organ. Therefore, the dynamic changes in these physiological indicators are not merely stress markers but are a direct manifestation of its “Shoot-Root Partitioned Coordination” adaptive strategy, providing a functional phenotypic anchor for subsequent molecular mechanistic analysis.

4.2. Core Gene Networks in Drought Response Revealed by Transcriptomics

Transcriptomic analysis elucidated the complex molecular mechanisms underlying the drought stress response in A. mongholicus seedlings at the transcriptional level. The GC content, defined as the proportion of guanine (G) and cytosine (C) bases in nucleotide sequences, is closely associated with gene stability and genomic composition [34]. In this study, the GC content of the transcriptome data ranged between 42.2% and 43.04%, which is consistent with previous findings by Jia et al. [2], who reported a GC content of 42.08–42.29% in A. mongholicus. Similarly, studies on other legume species have shown GC contents of 44.53–46.03% in soybean [11] and 43.2–43.97% in Sophora davidii (Franch.) Skeels [35]. A total of 2987 DEGs were identified, exhibiting pronounced tissue specificity and stress intensity-dependence. In aboveground tissues, the number of DEGs under severe drought (SDA vs. CKA: 1078) was substantially higher than that under moderate (MDA vs. CKA: 860) and mild drought (LDA vs. CKA: 238). A similar trend was observed in underground tissues, with DEG counts of 405 (SDR vs. CKR), 251 (MDR vs. CKR), and 155 (LDR vs. CKR), respectively. This progressive increase in DEG numbers reflects an escalating complexity of transcriptomic reprogramming, indicating that the plant mobilizes increasingly larger gene networks to cope with multifaceted physiological challenges as drought stress intensifies. This observation is consistent with previous transcriptomic findings in A. mongholicus reported by Jia et al. [2]. Notably, marked differences were detected in the directional patterns of DEG regulation between aboveground and underground tissues. Underground tissues displayed a predominance of upregulated DEGs across all stress levels (LD, MD, and SD). In contrast, aboveground tissues were characterized by predominantly downregulated DEGs under LD and MD, shifting to a predominance of upregulated DEGs under SD. This regulatory inversion suggests that under mild to moderate stress, aboveground tissues prioritize the suppression of non-essential energy-consuming processes to reallocate resources—an adaptive strategy reflecting resource conservation. Under severe stress, however, a large-scale stress response program is apparently triggered. These divergent transcriptional behaviors constitute a molecular manifestation of the distinct functional roles assumed by aboveground and underground compartments within the “Shoot-Root Partitioned Coordination” model.

GO and KEGG enrichment analyses [32] delineated the stage-specific molecular logic of this coordination. The aboveground response progressed temporally: from cell wall fortification (polysaccharide metabolism under LD) to biosynthesis of osmolytes and phenylpropanoids (MD). The activation of the phenylpropanoid pathway is particularly significant, as it underpins the production of defensive lignin and, crucially, flavonoids—key bioactive compounds linking drought response directly to potential medicinal quality regulation. Under severe drought (SD), transcriptional reprogramming converged on the activation of photosynthesis and redox metabolism, consistent with physiological phenotypes such as changes in antioxidant enzyme activities and MDA accumulation. This phase marks a transcriptional transition from “active optimization” to “stress-associated dysregulation”. Conversely, the underground response was characterized by sustained, concurrent activation: early MAPK signaling and glutathione-mediated antioxidation (LD), coupled with persistent amino acid biosynthesis for osmoregulation, escalating to ribosome biogenesis and DNA repair under SD to preserve core cellular functions. These response categories are conserved across diverse species, including Oryza sativa L. [31], Sesamum indicum L. [36], T. aestivum L. [11,33], S. officinarum L. [37], and Z. mays L. [16].

This analysis translates gene expression into a coherent transcriptional blueprint for partitioned coordination. Aboveground tissues follow a path from physical barrier establishment to chemical defense synthesis, ultimately guarding the energy source. Underground tissues focus on signal perception, homeostasis, and energy/nutrient supply. These pathways are not isolated; they are functionally interconnected. Enhanced phenylpropanoid metabolism aboveground reduces water loss and oxidative load, alleviating pressure on roots. In return, a steady supply of amino acids and energy from roots supports the synthesis of defensive compounds (e.g., glutathione, flavonoids) aboveground. This bidirectional dialogue via systemic signals and metabolite exchange forms the core of the proposed coordination model: aboveground tissues specialize in perception and advanced defense synthesis, while underground tissues dedicate themselves to homeostasis and foundational support.

Existing frameworks describe various medicinal plant drought strategies, such as “circadian adaptation” in S. davidii (Franch.) Skeels [35], “rapid signal response” via MAPK in Casuarina equisetifolia ssp. incana and D. nobile Lindl. [17,38], and “integrated defense” involving photosynthesis and phenylpropanoid biosynthesis in B. chinense, Illicium difengpi (Schisandraceae), and S. miltiorrhiza Bunge [13,32,34]. Our findings in A. mongholicus refine and extend this mechanistic landscape, elucidating a complete, coordinated pathway from early signal perception to the preservation of core functions under extreme stress, centered on a sophisticated aboveground–underground division of labor.

4.3. Mechanisms of Drought Response Revealed by Metabolomics

Metabolomics aims to investigate the dynamic changes in metabolites in biological systems following external stimuli. By comparing metabolite profiles between experimental and control groups and examining the biological processes involving DAMs, it reveals the underlying mechanisms of life activities under stress [39,40]. In this study, untargeted metabolomics was employed to systematically uncover the alterations in metabolic networks within aboveground and root tissues of A. mongholicus under gradient drought stress. To evaluate data reliability and model validity, PCA and PLS-DA were applied [41,42]. The results confirmed high data quality and further clarified the metabolic response patterns of A. mongholicus to varying drought intensities, providing a solid foundation for exploring its molecular mechanisms [40].

The number and changing trends of DAMs directly reflect the intensity of stress and the magnitude of the plant’s response [39]. In both aboveground and underground tissues, the number of DAMs increased sharply with drought severity. Notably, in aboveground tissues under SD, the number of DAMs (163) was nearly six times that under LD, clearly indicating that A. mongholicus mobilizes increasingly extensive metabolic pathways as drought intensifies. A similar active synthesis of metabolites was observed in maize roots under drought, considered key to its adaptation [43]. Importantly, aboveground tissues of A. mongholicus specifically accumulated glutathione and its related metabolites under LD and MD conditions. The pronounced accumulation of glutathione, together with the concomitant upregulation of its key biosynthetic enzyme genes (e.g., GST), constitutes a reinforced antioxidant defense system. Furthermore, glutathione serves as a reservoir for cysteine and participates in the metabolic cycling of sulfur-containing amino acids, thereby providing precursors for the biosynthesis of sulfur-containing bioactive constituents such as astragalosides. This mechanistic linkage directly connects the drought stress response to the potential regulation of medicinal quality at the molecular level. Collectively, these metabolic events possess clear adaptive value—scavenging reactive oxygen species, maintaining redox homeostasis, and supplying precursors for secondary metabolite biosynthesis—rather than representing mere byproducts of stress-induced damage.

Of particular significance, aboveground tissues specifically accumulated glutathione and related metabolites under LD and MD. The marked accumulation of glutathione, coupled with the up-regulation of its key biosynthetic enzyme gene (GST), constitutes a reinforced antioxidant system. Glutathione directly scavenges ROS, protecting cells from oxidative damage and thereby enhancing drought tolerance. Furthermore, glutathione serves as a reservoir for cysteine, participating in the metabolic cycle of sulfur-containing amino acids and thus providing precursors for the synthesis of sulfur-containing active components like astragalosides. This links the drought stress response to the potential regulation of medicinal quality at the molecular level.

KEGG pathway enrichment analysis was performed to gain deeper insight into the biological functions of the DAMs and systematically interpret the metabolic response strategies of A. membranaceus at different drought levels. Pathway analysis serves as a bridge connecting metabolite changes to biological function, aiding in the interpretation of molecular mechanisms of plant stress tolerance [44,45]. The results showed that the enriched metabolic pathways exhibited clear tissue specificity and stress intensity dependence. In aboveground tissues, the metabolic response strategy progressed in a sequential manner with increasing stress severity. Under LD, antioxidant defense systems such as Glutathione metabolism and Cysteine and methionine metabolism were preferentially activated to scavenge reactive oxygen species generated during the initial stress phase. Under MD, the focus shifted to Cutin, suberine, and wax biosynthesis and the degradation and synthesis of branched-chain amino acids, indicating reinforcement of physical barriers to reduce water loss alongside mobilization of amino acids for energy. This combined mechanism of physical defense and energy mobilization has also been confirmed in recent studies on barley (Hordeum vulgare L.)’s drought response [46]. Under SD, enriched pathways centered on energy metabolism, such as Pyruvate metabolism and Fatty acid degradation, suggesting severe impairment of photosynthesis and a shift towards catabolizing stored reserves to sustain basic life processes. This strategy of switching from photosynthetic to respiratory energy metabolism has been reported in Nicotiana sylvestris wild type [47], indicating it may be a conserved response mechanism to extreme stress. In contrast, the response strategy in root tissues demonstrated greater proactivity and diversity. Under LD, enriched pathways were primarily related to nucleotide synthesis and antioxidant defense, such as Pyrimidine metabolism and Ascorbate and aldarate metabolism, suggesting preparatory steps for secondary metabolite synthesis and potential cell proliferation during the early stress phase. Under MD, pathways related to amino acid metabolism, such as Lysine degradation and Tyrosine metabolism, were activated, indicating that roots began to degrade and transform amino acids to provide intermediates for the tricarboxylic acid (TCA) cycle, thereby ensuring energy supply and carbon skeletons. Under SD, although Pro accumulation persisted, it was accompanied by the enrichment of oxidative phosphorylation and arginine/proline metabolism pathways. This profile indicates that under extreme stress, roots actively accumulate osmoregulatory compounds to maintain cellular homeostasis and generate ATP via respiration to support vital functions, albeit with signs of diminishing adaptive returns. This synergistic strategy combining osmotic adjustment and energy guarantee has also been reported in barley [48].

In summary, the metabolomic data provide functional-level evidence for the “ Shoot-Root Partitioned Coordination “ adaptive mechanism. Aboveground tissues operate with the “glutathione hub” at their core, coupling the antioxidant system with secondary metabolism. Underground tissues focus on the “amino acid metabolism network”, ensuring osmotic adjustment and energy homeostasis. Together, they determine the metabolite accumulation and survival capacity of A. mongholicus under drought stress.

4.4. Coordinated Regulatory Mechanisms Revealed by Integrated Transcriptomic and Metabolomic Analysis

Integrated transcriptomic and metabolomic analysis not only reveals changes in gene expression and metabolite accumulation under drought stress but also provides deep insights into the metabolic regulatory networks governing plant stress responses [15]. This study found that the molecular regulatory network in A. mongholicus seedlings exhibits significant tissue specificity, with aboveground and root tissues employing distinct gene expression and metabolic pathways to coordinately respond to stress. In aboveground tissues, a defense mechanism centered on Glutathione metabolism and characterized by “antioxidant-osmotic adjustment” was activated. Glutathione metabolism drives the antioxidant defense system by scavenging stress-induced ROS, providing comprehensive protection against oxidative damage [49]. Its upstream pathways, Cysteine and methionine metabolism, not only supply precursors for glutathione synthesis but also regulate the production of signaling molecules such as ethylene and polyamines [50]. The co-enrichment of these sulfur-related pathways underscores the critical role of sulfur compounds in drought resistance. Moreover, sulfur metabolism forms the foundation for synthesizing sulfur-containing secondary metabolites, including astragalosides—key bioactive compounds in A. mongholicus. The importance of sulfur compounds in drought response has also been corroborated in Camellia sinensis L. [51]. Our results showed that the accumulation of glutathione and its precursors (L-Glutamate, L-Cysteinylglycine) occurred in concert with the up-regulation of genes encoding GST and PRDX6, collectively establishing an efficient enzymatic and non-enzymatic system for scavenging excess ROS induced by drought. The effective operation of this system, in conjunction with the sustained activation of peroxidase (POD) in underground tissues, provides the molecular basis for the contained level of malondialdehyde (MDA) in roots despite increasing stress, thereby directly contributing to the maintenance of cellular integrity and enhanced drought tolerance. Similarly, simultaneous increases in glutathione accumulation and GST gene expression were observed in I. difengpi under drought stress, mitigating ROS-induced damage through enhanced glutathione synthesis capacity [34], indicating that glutathione-mediated antioxidant defense is a conserved mechanism in drought response. Furthermore, the dynamic changes in ethylene synthesis genes (ACS, ACO) and the polyamine metabolite Spermidine suggest that A. mongholicus maintains cellular homeostasis by modulating the balance between hormones and osmotic pressure, consistent with the key role of ethylene–polyamine crosstalk in the stress response of barley [52]. Therefore, the activation of this metabolic network is not only vital for survival adaptation but may also indirectly regulate the biosynthetic potential of specific secondary metabolites related to medicinal quality (e.g., certain saponins) by influencing precursor supply.

In contrast, the strategy in root tissues involved the activation of amino acid metabolism pathways. The enrichment of these pathways, along with the dynamic changes in associated genes and metabolites, indicates that plants primarily respond to stress through energy remodeling [53] and signal transduction [16,53]. Tryptophan metabolism and Tyrosine metabolism play roles in signal transduction and energy supply, respectively, during stress responses [16,54]. Tryptophan serves as a precursor for synthesizing hormones like auxin, jasmonic acid, and strigolactones, as well as various alkaloids and other secondary metabolites [16]. The synthesis of diverse hormones and secondary metabolites can activate the plant’s stress defense system [16]. Tyrosine plays a dual role in plant stress: it can generate fumarate and acetyl-CoA, which enter the tricarboxylic acid (TCA) cycle to provide energy for stress responses [54], and it acts as a precursor for lignin synthesis, enhancing physical defense by reinforcing cell walls [55]. In this study, the up-regulation of tryptophan, tyrosine, and their derivatives under drought stress suggests that roots may respond to drought by synthesizing hormones or defensive secondary metabolites. Histidine, functioning as a signaling molecule, participates in regulating nitrogen metabolism-related genes to enhance stress resistance [56]. As hubs of carbon and nitrogen metabolism, Alanine, aspartate and glutamate metabolism play crucial roles under drought stress [36]. Glutamate and aspartate can be converted into key TCA cycle intermediates, α-ketoglutarate and oxaloacetate respectively, via anaplerotic reactions, thereby ensuring fundamental energy supply [53,57]. The down-regulation of amino acids such as histidine, glutamate, and aspartate in the roots of A. mongholicus under drought may reflect their catabolic utilization as energy substrates and nitrogen sources, providing energy and carbon skeletons to sustain basic root metabolic activities.

In summary, the “aboveground-underground partitioned coordination” response mechanism is driven by tightly coupled “gene-enzyme-metabolite” modular networks. Aboveground, the glutathione–sulfur amino acid metabolism network focuses on ROS scavenging and the synthesis of defense- and quality-related compounds. Underground, the amino acid metabolism network prioritizes osmotic adjustment and energy homeostasis. Together, they constitute a coordinated, partitioned response model. Compared to the drought resistance strategies of other leguminous and medicinal plants, this represents a distinct and integrated model. For instance, the classic drought response in soybean (G. max) is characterized by a whole-plant, ABA-signaling-driven “growth inhibition-stomatal closure” reaction, where transcriptional and metabolic reprogramming is highly synchronized between aerial and underground tissues [11]. In contrast, our study reveals a significant tissue asynchrony in A. mongholicus: down-regulated genes dominate in aboveground tissues under light to moderate stress, whereas underground tissues exhibit a consistent pattern of gene up-regulation. This “response asynchrony” likely represents a more sophisticated resource allocation strategy. Furthermore, in S. miltiorrhiza, the core of drought adaptation is localized to the roots, primarily through activating phenylpropanoid and terpenoid metabolism [13], whereas D. nobile emphasizes the reinforcement of Crassulacean Acid Metabolism (CAM) in its aerial parts [17]. Unlike these organ-centric models, the drought resistance mechanism identified in A. mongholicus is not dominated by a single organ but involves a synergistic model between shoots and roots, founded on a coordinated sulfur-nitrogen metabolic cycle.

4.5. Practical Implications of the “Aboveground-Underground Partitioned Coordination” Mechanism

Building upon the elucidated “Shoot-Root Partitioned Coordination” mechanism, we propose actionable prospects for A. mongholicus cultivation. We demonstrate that both LD and MD activate glutathione metabolism and phenylpropanoid biosynthesis—pathways closely associated with the synthesis of major bioactive compounds in A. mongholicus—and that LD alone is sufficient to trigger antioxidative defense responses in aboveground tissues. This provides molecular-level evidence supporting the concept that environmental stress influences medicinal quality [58]. Accordingly, controlled water deficit—particularly maintaining soil water content within 50–65% of field capacity, corresponding to the LD–MD interval identified in this study—may serve as a viable agronomic strategy to enhance the accumulation of specific bioactive constituents without incurring significant biomass loss, thereby enabling more precision on-demand irrigation [59]. For instance, the expression levels of biomarker genes (e.g., GST, ACS) or the concentration of key metabolites (e.g., glutathione) could serve as accurate indicators of the plant’s internal stress status, facilitating a shift from soil moisture-based scheduling to a plant-centered water management paradigm. With respect to drought-tolerant germplasm screening, the glutathione–sulfur metabolic module exhibits sensitive, dose-dependent activation, with its core genes and metabolites being continuously upregulated under LD and MD—closely aligning with the identified stress-optimization window. This module directly couples antioxidative defense with sulfur-containing secondary metabolism, conferring dual benefits for drought tolerance and quality improvement. Therefore, glutathione content and the transcript abundance of key pathway genes represent promising molecular probes for early selection of A. mongholicus germplasm with combined drought tolerance and superior medicinal quality.

Furthermore, through tissue-resolved transcriptomic and metabolomic networking, this study identifies molecular breeding targets embedded within the “Shoot-Root Partitioned Coordination” framework. In aboveground tissues, genes involved in glutathione metabolism and associated networks (e.g., GST, PRDX6) can be targeted to enhance antioxidative capacity and astragaloside biosynthesis. In underground tissues, amino acid metabolism-related genes (e.g., PrAO) represent candidate targets for strengthening osmotic adjustment and root system homeostasis. Building on these findings, we propose a modular pyramiding breeding strategy that integrates aboveground and underground targets via synergistic optimization. This approach holds potential to overcome the traditional trade-off among drought tolerance, high yield, and superior quality, paving the way for developing elite A. mongholicus varieties with co-enhanced stress resilience and medicinal properties.

In summary, this study systematically dissects the “Shoot-Root Partitioned Coordination” mechanism underlying graded drought responses in A. mongholicus seedlings under controlled conditions. The core pathways, dose–response boundaries, and key gene modules uncovered herein are directly translatable into irrigation management strategies, drought-tolerance markers, and breeding targets. It should be noted, however, that this study focused on physiological and molecular responses—such as antioxidative enzyme activities and osmoregulatory metabolite contents—and did not include measurements of growth-related traits (e.g., biomass, root-to-shoot ratio, plant height). This limitation precludes direct extrapolation of the observed molecular and physiological events to adaptive advantages or yield potential. Accordingly, the conclusions drawn are strictly confined to the molecular and physiological levels, aiming to elucidate the intrinsic logic of drought signal perception, transduction, and metabolic reprogramming, rather than to assess agronomic performance or yield. Future studies integrating growth traits, physiological parameters, and multi-omics datasets will be essential to establish a complete causal chain from molecular events to phenotypic output.

5. Conclusions

This study, employing an integrated multi-omics approach, elucidates the molecular mechanisms underlying the drought stress response in A. mongholicus seedlings. The findings reveal that A. mongholicus employs a differentiated yet coordinated aboveground-underground regulatory strategy. The aboveground tissues specifically enrich glutathione and sulfur-containing amino acid metabolic pathways, confirming that an “antioxidant-osmoregulation” defense mechanism serves as a key strategy for coping with drought stress; meanwhile, systemic signals are transmitted from shoots to roots. In contrast, the underground tissues activate a regulatory network centered on amino acid metabolism, focusing on “signal transduction and energy supply,” which prioritizes water uptake, osmoregulation, and sustained energy provision to maintain basic survival. This functional division, metabolic complementarity, and systemic integration represent a sophisticated survival strategy evolved by A. mongholicus to adapt to arid environments. The present study not only deciphers the intrinsic mechanisms underlying this adaptive wisdom, but also provides a solid theoretical foundation and a feasible technical pathway for modulating these mechanisms via modern agronomic and biotechnological approaches, with the aim of enhancing drought resistance and medicinal quality in A. mongholicus.