Chemical heterogeneity of two Bupleurum Radix species based on multiple mass spectrometry techniques

2.1. Experimental samples

Fresh samples of BS were collected from Zhaozhuang village, Guta town, Yuyang District, Yulin city, Shaanxi Province, and fresh samples of BC were collected from Guofort village, Xifeng town, Qingyang city, Gansu Province. Details regarding the collection of the dry samples have been listed in Table 1. All the samples were identified by Professor Chen Yuan from Gansu Agricultural University. Fresh samples were immediately frozen in liquid nitrogen and stored at -80°C until use.

2.2. Instrument

The main instruments used in this study included a timsTOF fleX MS system (Bruker Daltonics, Germany), a TM-Sprayer (HTX Technologies), a CM1950 cryosectioning machine (Leica, Germany), a Research plus micropipette (Eppendorf, Germany), an MIX-200 multitube vortex mixer (Shanghai Jingxin, China), ITO conductive slides (Guluo Glass, China), a KQ5200E ultrasonic cleaner (Kunshan Shumei, China), a SUPEC 7000 ICP-MS instrument (Focused Photonics Inc., China), an ML204 analytical balance (China), an inverted fluorescence microscope (Mettler toledo, America), a high-speed homogenizer (FLUKO, China), a 1290 Infinity II LC/6546 quadrupole time-of-flight (Q-TOF) MS system (Agilent, America), a temperature-controlled electric heating plate (maximum temperature of 200°C), and a polytetrafluoroethylene digestion vessel.

2.3. Anatomical methods

Fresh samples were retrieved at -80°C and allowed to equilibrate at room temperature briefly. The samples were subsequently cut into segments approximately 1 cm in length and placed in a vacuum drying bottle filled with 15% glycerol. The air within the bottle was evacuated, and the samples were left to soften at the bottom for 30 min. After softening, the samples were embedded in optimal cutting temperature (OCT) compound and rapidly frozen. Once the OCT compound turned white and hardened, the samples were cut into thin 20–30 μm sections using a cryostat. The tissue sections were placed on a slide, glycerol was added for sealing, and the slides were observed directly. Alternatively, the sections were stained with hematoxylin and eosin (HE) before being sealed for observation.

2.4. Histological localization

Experimental group: Fresh Bupleurum roots were thoroughly washed and manually sectioned into thin slices by using freehand sectioning techniques. The slices were treated with 5% lead acetate for 10 min to precipitate saponins within the tissues. They were subsequently stained with a mixture of 5% vanillin glacial acetic acid and perchloric acid (v:v = 1:1) for 10–30 s, after which the treated sections were observed.

Control group: Freshly washed Bupleurum roots were cut into small segments and immersed in formaldehyde acetic acid (FAA) and 70% ethanol fixative. The samples were vacuum infiltrated to ensure complete penetration and fixed for more than 20 days to remove saponins. After fixation, the roots were sectioned and stained using the same method as that used for the experimental group. Ten randomly selected sections were observed to determine whether the stain color disappeared.

2.5. MALDI-MSI analysis

Following the sectioning method described in Section 2.3, tissue sections were mounted onto indium tin oxide (ITO) coated conductive glass slides. The slides with tissue sections were then dried in a vacuum desiccator for 30 min to remove residual moisture and enhance matrix adhesion. Using a matrix sprayer, a 15 mg mL^-1 2,5-Dihydroxybenzoic acid (DHB) solution was prepared by dissolving DHB in 90% acetonitrile. The DHB matrix solution was evenly sprayed on the ITO slides containing tissue sections with a TM-Sprayer matrix sprayer. The instrument parameters were as follows: temperature of 60°C, flow rate of 0.1 mL min^-1, pressure of 6 psi, 25 cycles of gas spraying, and 5 s of drying. Mass spectra were collected in forward mode, and their mass range was m/z 50–1300 Da. The laser power was set at 80% and remained constant throughout the experiment. The imaging resolution was set to 50 μm. The MALDI mass spectra were normalized by the root mean square, and the signal intensity in each image is shown as the normalized intensity. Fragment analysis in MS/MS mode on the timsTOF fleX MS system was used further to confirm the detailed structure of the identified metabolites. The materials were identified by examining the MS information, including the intensity of the target peak (primary MS information) and the in situ secondary fragmentation of tissue via the secondary tissue spectrum (secondary MS/MS fragment peak information), searching the self-built database and integrating the public library to compare the secondary spectra. Owing to the low intensity, the target peak from the secondary spectrum could not be collected. In accordance with the molecular weight from the self-built database and literature comparisons, the search was performed within an error range of 10 ppm, and the substances with the molecular weight closest to that detected by the instrument were retrieved.

2.6. Microwave digestion of the samples

The dry samples were crushed and sieved through a 100-mesh screen. A 0.2000 ± 0.0001 g sample was weighed into a polytetrafluoroethylene digestion vessel. Nitric acid (5 mL), perchloric acid (2 mL), and hydrogen peroxide (0.5 mL) were added to the vessel. The vessel was covered and allowed to stand for 1 h. Then, the vessel was placed on a graphite digestion instrument and digested at 150°C for 4 h. Based on the digestion status, a small amount of mixed acid was added until the digestion solution became clear and transparent. The vessel was covered, and the solution was diluted to 1–2 mL with water. The solution was vortexed well and then transferred to a 50 mL volumetric flask. The flask was capped tightly and set aside. Three blank solutions were prepared using the same method used for subsequent background correction.

2.7. Determination and analysis of inorganic elements

The mixed standard solution was injected into the ICP-MS instrument for measurement, and a standard curve was plotted. Instrument parameters: RF power, 1500 W; cooling gas, 12 L min^-1; auxiliary gas, 0.8 L min^-1; nebulizer gas, 1.0 L min^-1; sampling depth, 6 mm; collision gas, He; flow rate, 3 mL min^-1; integration time per mass, 0.1 s; peristaltic pump speed, 0.3 rps; and nebulizer type, concentric nebulizer. The sample supernatant or filtered sample solution was injected into the ICP-MS system for measurement, and the concentration of each element in the test solution was calculated based on the standard curve. The formula for the target element is as follows (Eq. 1):

where

X: Content of the target element in the sample, mg kg^-1;

P: Mass concentration of the target element in the sample solution, μg L^-1;

P0: Mass concentration of the target element in the sample blank solution, μg L^-1;

V: Volume of the sample digestion solution, mL;

f: Dilution factor of the sample;

m: Sample weight, g.

2.8. LC-MS sample preparation and instrument conditions

The dried sample was pulverized and sieved, and 1.000 ± 0.0001 g of powder was precisely weighed into a conical flask. After 40 mL of methanol was added, the mixture was ultrasonicated (30 min, 30°C, 400 W), vortexed, and centrifuged (15 min). The supernatant was concentrated to 5 mL and passed through a 0.45 μm microporous membrane filter. Triplicate biological replicates were conducted.

Chromatographic conditions: All the samples were analyzed by LC-MS in positive ion mode using an Agilent Poroshell 120 EC-C18 column (4 μm, 4.6×150 mm) with 0.1% formic acid in water (A) and acetonitrile (B) as the mobile phases. The gradient elution program was as follows: 35–40% B (0–2.5 min), 40–42% B (2.5–5 min), 42–46% B (5–10 min), 46–50% B (10–17.5 min), 50–55% B (17.5–22.5 min), 55–65% B (22.5–28.5 min), 65–98% B (28.5–32.5 min), hold at 98% B for 2 min (32.5–34.5 min), and then return to 35% B (34.5–36.5 min). The analysis was performed at 40°C with a flow rate of 0.4 mL min^-1 and an injection volume of 4 μL.

Mass spectrometry conditions: Mass spectrometry analysis was conducted in positive ion mode using an electrospray ionization (ESI) source with the following parameters: spray voltage, 3500 V; sheath gas, 35 psi; auxiliary gas, 11 L min^-1; nebulizer gas, 15 psi (all gases were nitrogen); and ion transfer tube temperature, 300°C. The mass scan range was m/z 100–1500.

2.9. LC-MS data processing

The raw data were processed using Profinder software via an automated workflow to optimize the original signals and extract valid information. First, peaks were selected to convert continuous mass spectrometry signals into discrete peak lists (including m/z, retention time, and intensity), with parameter adjustments made to increase sensitivity. The retention times were subsequently aligned and applied to correct chromatographic shifts among samples, ensuring consistency in peak matching. After preprocessing, peaks with a missing data rate of > 50% across sample groups were filtered out to reduce low-confidence signals, and missing values were imputed. The imputed data were then matched with the Metlin database based on accurate mass and retention time comparisons, and compound annotation information with matching scores greater than 85% was screened out. The specific ion signals of the target compounds at defined mass-to-charge ratios (m/z) were extracted from the full scan mass spectrometry data using the extracted ion chromatogram (EIC) method to eliminate background interference. The identity of the compounds was confirmed by matching the retention times of reference standards and target analytes in the samples under identical chromatographic conditions. The contents of the target compounds in the samples were calculated based on the standard calibration curve (Saikosaponin A: y = 21637.65x + 1213.22, R² = 0.9995; Saikosaponin D: y = 25552.59x + 1136.72, R² = 0.9994).

3.1. Microstructure and histochemical characteristics

The periderm is composed of the phellem, phellogen, and phelloderm. Phellem cells are typically arranged in neat rows, exhibiting a flattened and tangentially elongated morphology with thin, suberized cell walls. In BC, the phellem cells are more densely packed than those in BS are. The cell lumens contain liposoluble substances that appear yellow-brown in BC (Figure 1a) and red-brown in the BS (Figure 1b). The phellogen consists of a single layer of thin-walled, slightly tangentially elongated cells, with distinct fissures observed near the periderm (Figure 1a,b). The secondary structure of the root results from the activity of both the vascular cambium and cork cambium.Both BC and BS contain oil cavities and numerous scattered oil cells between these two cambium layers (Figure 1a,b), whereasthe phloem rays are indistinct. In BC, the xylem vessels are sparsely and irregularly distributed, with groups of xylem fibers arranged in discontinuous rings in the middle region. These fibers are polygonal in shape and have thick, lignified walls (Figure 1c). In contrast, BS exhibits predominantly radially arranged xylem vessels, with a looser xylem structure and less lignification. Because fewer xylem fibers exist and they are scattered, complete rings cannot form (Figure 1d). Neither BC nor BS shows a distinct pith structure (Figure 1e,f). The HE staining results (Figure 2a,b) revealed that the xylem cells of BC clearly displayed defined fibrous structures and a prominent annular cambium. In comparison, BS exhibits lighter staining and a relatively loose cellular arrangement, with widely distributed oil cavities. Distinct oil droplet-like structures are observed after staining and are surrounded by parenchyma cells. Previous studies have confirmed that BS contains the highest volatile oil content among Bupleurum species [17]. Saikosaponins react with a 5% vanillin glacial acetic acid and perchloric acid solution, producing characteristic colors ranging from light red to deep red. The results indicate that saikosaponins are widely distributed near the cambium, with both BC and BS displaying deep purplish red staining. The total saikosaponin content in BS was slightly greater than that in BC (Figure 2c,d). In the control experiments, root cross-sections treated with FAA did not produce the characteristic color (Figure 2e,f).

Close

Figure 1.

Microstructure of frozen section of transverse section of Bupleurum radix. (Note: P: periderm; CC: cork cambium; PP: primary xylem; SP: secondary xylem; OC: oil chamber, VC: cork cambium; SX: secondary xylem; PX: primary xylem. Red circles indicate the oil-like substances. BC: (a)(c)(e); BS: (b)(d)(f). Magnification: (a-d) 40×; (e, f) 20×).

Export to PPT

Close

Figure 2.

HE staining and histochemical localization of Bupleurum radix. (Note: BC: (a,c,e); BS: (b,d,f). Magnification: 20x).

Export to PPT

3.2. Spatial distribution patterns of bioactive components in bupleurum species of different botanical origins

MSI was performed in the m/z range of 50–1300 Da with a DHB matrix application. Raw data containing m/z and peak intensity values for each pixel were processed through smoothing and baseline correction, leading to the identification of a total of 1376 compounds classified into 13 categories. Principal component analysis (PCA) of all detected compounds was carried out using the prcomp function in R, revealing distinct clustering patterns between the two groups. The PCA score plot (Figure 3) revealed cumulative variance contributions of 51.93% (PC1) and 20.28% (PC2), accounting for 72.21% of the total variance. Separation along PC1 between BC and BS reflects differences in the spatial distribution patterns of metabolites between these two botanical origins of Bupleurum.

Close

Figure 3.

Spatial metabolomics PCA score plot of the two sample groups.

Export to PPT

Spatial segmentation is an effective approach for simplifying the representation of high-dimensional data. MSI, an advanced extension of conventional MS, enables precise visualization of molecular spatial distributions in biological tissues. Current analytical methods favor unsupervised dimensionality reduction techniques, particularly the uniform manifold approximation and projection (UMAP) algorithm [18]. This nonlinear approach, rooted in Riemannian geometry and algebraic topology principles, successfully transforms complex spectral data into interpretable 2D visualizations (Figure 4). The UMAP projection employs color coding to represent pixel-specific spectral features, where similar metabolite profiles cluster with matching colors, effectively revealing chemical heterogeneity across tissue sections. Comparative analysis shows distinct spatial distribution patterns: BC displays concentric chemical zonation, with metabolites radiating uniformly from the central vascular cylinder toward the periphery. In contrast, BS has an asymmetric lateral distribution, with differentially abundant metabolite accumulation between opposing sides of the root.

Close

Figure 4.

UMAP hyperspectral visualization of the two sample groups. Note: The color coding of each pixel is based on spectral characteristics, and pixels with similar metabolite abundance features are displayed in similar colors.

Export to PPT

3.2.1. Spatial distribution characteristics of saikosaponins of two botanical origins

Histochemical localization, as a traditional method for chemical component mapping, can reveal the spatial distribution of total saikosaponins but lacks specificity in distinguishing individual compounds. MSI compensates for this limitation through its high-resolution capability, enabling precise localization of individual saikosaponins. The combination of these two techniques provides a more comprehensive perspective for research on saikosaponin in Bupleurum. Saikosaponins, the primary active components in Bupleurum, are natural triterpenoid saponin compounds that include various types, such as saikosaponin a, b, c, d, and f. Many saikosaponins exist as isomers, for example, saikosaponin a and saikosaponin d. These isomers exhibit identical precursor ion peaks in mass spectrometry and similar fragmentation patterns in MS/MS analysis. Therefore, the most likely structural assignments were used for characterization. In this study, 38 triterpenoid saponins were detected (Table S1), including 10 Bupleurum-specific saikosaponins: saikosaponin A/D [M+Na]⁺, saikosaponin C [M+K]⁺, saikosaponin F [M+Na]⁺, saikosaponin H [M+Na]⁺, 16-keto-acetylsaikosaponin A [M+H]⁺, acetylsaikosaponin A [M+K]⁺, 6’’-O-acetyl saikosaponin B3 [M+K]⁺, diacetylsaikosaponin A [M+K]⁺, 2’’-O-acetyl saikosaponin A [M+Na]⁺, and 6’’-O-acetyl saikosaponin A [M+Na]⁺. As shown in Figure 5, the distribution of saikosaponins was highly similar within the same botanical origin (BS or BC) but displayed distinct spatial and quantitative differences. In BC, saikosaponins were concentrated in the phloem in a ring-like pattern, with minimal accumulation in the xylem. In contrast, BS showed a low overall abundance and an uneven horizontal distribution, with greater accumulation near the cork cambium than near the vascular cambium. Notably, several acylated saikosaponins were detected, with the levels of acetylsaikosaponin A and diacetylsaikosaponin A in BC being 19.7 and 13.6 times higher, respectively, than those in BS. Acylation may influence bioactivity and stability by altering the polarity or conformation of saponin. Although acylated saikosaponins constitute a significant proportion of Bupleurum saponins [19,20], their pharmacological properties remain poorly understood. The ring-like enrichment of saikosaponins in the phloem is likely related to their biological functions. As the phloem is responsible for storing organic compounds and long-distance transport in vascular plants [21], this distribution pattern may facilitate saponin translocation. Additionally, saikosaponins play a role in plant defense, and their phloem accumulation may help deter pathogens and herbivores [22]. The relatively low signal intensity of the saikosaponins in this study may be attributed to the limited ionization efficiency of the DHB matrix for this compound class.

Table S1

Download

Close

Figure 5.

MALDI-MSI visualization of saikosaponins in cross-section of the root. Note: (a): Saikosaponin A/D (SSa/SSd); (b): Saikosaponin C (SSc); (c): Saikosaponin F (SSf); (d): 2’’-O-acetyl saikosaponin A (2’’-O-acetyl SSa); (e): 6’’-O-acetyl saikosaponin A (6’’-O-acetyl SSa); (f): Acetylsaikosaponin (Acetyl-SSa); (g): Diacetylsaikosaponin A (Diacetyl-SSa).

Export to PPT

3.2.2. Spatial distribution characteristics of flavonoids of two botanical origins

Flavonoids are also important active compounds in Bupleurum. As shown in Table S1, 103 flavonoids were identified, among which only three were significantly different (p < 0.05): 5,6-dihydroxyflavone (m/z 254.9255), 1,3,6,8-tetrahydroxy-2-methoxyxanthone (m/z 329.0052), and daidzein glucosyl malonyl glucoside (m/z 647.1595); however, these quantitatively distinct flavonoids displayed no significant heterogeneity in spatial distribution patterns. In transverse sections of BS, slight bilateral asymmetry was observed in their phloem distribution (Figure 6). Flavonols are common flavonoid components in Bupleurum. Kaempferol, quercetin, and isorhamnetin serve as its three primary aglycones [23], while apigenin is another common aglycone of flavonols. Apigenin has multiple pharmacological properties (anti-inflammatory, antiviral, and anticancer activities [24,25]) and is homogeneously distributed in both BS and BC. Notably, these natural flavonoid glycosides exhibited minimal spatial distribution differences, both between and within groups, showing only marginal (non-significant) variation trends. The relatively uniform distribution pattern of flavonoids in Bupleurum contrasts sharply with the localized accumulation of saikosaponins (e.g., ring-like phloem enrichment), reflecting the diverse defense strategies of secondary plant metabolites.

Close

Figure 6.

MALDI-MSI visualization of flavonoid in cross-section of the root. Note: (a): Kaempferol-3-O-glucuronide-7-O-glucoside; (b): [6-[2-(3,4-Dihydroxyphenyl)-5,7-dihydroxy-4-oxochromen-3-yl]oxy-3,4,5-trihydroxyoxan-2-yl]methyl acetate; (c): Isorhamnetin malonyl glucosyl glucuronopyranosyl malonyl glucoside; (d): Apigenin-7-O-glucuronide; (e): 5,6-Dihydroxyflavone; (f): 1,3,6,8-Tetrahydroxy-2-methoxyxanthone; (g): Daidzein glucosyl malonyl glucoside.

Export to PPT

3.2.3. Spatial distribution characteristics of the other active ingredients of two botanical origins

In addition to saponins and flavonoids, 52 phenolic acids, 16 lignans, and eight coumarins were detected in Bupleurum in this study (Table S1). These compounds are universally present and exhibit unique distribution patterns, as follows (Figure 7): phenolic acids, known for their significant antioxidant and anti-inflammatory effects [26], include sodium ferulate (m/z 255.0039, [M+K]⁺), which is widely recognized for its antioxidant, antiplatelet aggregation, antithrombotic, and cardiovascular protective properties [27]. While no significant differences in content were observed between BS and BC, their spatial distributions showed distinct tissue heterogeneity. In BC, phenolic acids predominantly accumulated near the vascular cambium and were distributed throughout the cross-section, whereas in BS, they were localized mainly near the cork cambium, with minimal accumulation in the xylem and no accumulation in the primary xylem. Various simple coumarins, which are commonly found in Bupleurum and known for their anti-inflammatory and antitumor properties [28], did not significantly differ between BS and BC. However, their spatial distributions varied: coumarin accumulated predominantly near the phelloderm in BS, with almost no accumulation in the primary xylem, whereas in BC, it was uniformly distributed without specific localization. Lignans exhibit pharmacological activities, including antioxidant effects. Researchers have isolated a novel lignan compound from Bupleurum marginatum, along with 10 structurally related analogs [29]. In the current study, among the 16 detected lignans, N²-demethylthalidezine was the most abundant and significantly differed between the BS and BC groups (p < 0.001). Distinct spatial accumulation patterns were observed: in BC, lignans predominantly accumulated in the xylem and cambium regions, with minimal amounts near the epidermis, whereas in BS, they were distributed mainly in the phloem adjacent to the cork cambium, with less accumulation in the xylem.

Close

Figure 7.

MALDI-MSI visualization of other active ingredients in cross-section of the root. Note: (a): Sodium ferulate; (b): Coumarin; (c): N²-Demethylthalidezine.

Export to PPT

3.3. Inorganic element profiles of 18 batch samples

Building upon the established understanding of the microscopic structure and spatial distribution characteristics of bioactive components within fresh Bupleurum tissues, the sample scale was expanded in this study to investigate its chemical diversity systematically. We conducted ICP-MS and LC-MS detection on dried samples from 18 batches of Bupleurum collected from different geographical origins.

The certified reference material (CRM) “dry celery (GBW10229)” was used to evaluate the accuracy of the experiment by comparing the certified values and uncertainties with the measured values. As shown in Table 2, all the samples met the quality control requirements (90%–110%). The limits of detection (LOD) (Eq. 2) and limit of quantification (LOQ) (Eq. 3) were calculated following the guidelines provided by the International Union of Pure and Applied Chemistry (IUPAC) using the following equations:

Where LOD is the limit of detection, LOQ is the limit of quantification, SD is the standard deviation for the blank samples, and slope is the slope of the calibration curve.

The external standard method was used in this study to quantify 11 elements (K, Ca, Na, Mg, Fe, Al, P, S, Mn, Zn, and Cu), and the internal standard method was used for 14 elements (Ni, Sc, V, Cr, Co, As, Zr, Mo, Cd, Pb, Bi, Sr, Re, and B). The concentrations of Pb, Cd, As, and Cu in all the samples complied with the heavy metal and harmful element limits specified in the 2020 edition of the Pharmacopoeia of the People’s Republic of China (ChP) (Pb ≤ 5 mg kg^-1; Cr ≤ 1 mg kg^-1; As ≤ 2 mg kg^-1; Cu ≤ 20 mg kg^-1). Independent tests revealed significant interspecies differences (p < 0.05) in 8 elements (Bi, Fe, Cr, Ni, Zr, Pb, Co, and Sc) between plants of different botanical origins, whereas other elements did not significantly vary (Table 3). Notably, the concentrations of both Fe (p < 0.01) and Ni (p < 0.05), which are essential for plant growth and development, were significantly greater in the BC group than in the BS group. However, the overall mineral elemental distribution patterns were minimally influenced by the botanical origin. In both the BS and BC groups, K was consistently the most abundant element, followed by Ca, whereas Re and Bi had the lowest concentrations. As potentially toxic heavy metals, Re and Bi maintained extremely low levels in all the samples (Re < 0.005 mg kg^-1; Bi < 0.02 mg kg^-1), the levels of Re and Bi were significantly below the regulatory limits common for Chinese herbal medicines, and they could be considered naturally occurring trace elements. K and Ca, as mineral elements essential for plant growth and development, play crucial physiological roles in plant systems. Approximately 60 enzymes require K for activation, and it also contributes to the response of plants to abiotic stress and the maintenance of ionic balance and cellular integrity [30]. In the BS group, the average K content was 12901.9562 mg kg^-1, with the highest level in BS8 (21521.7355 mg kg^-1) and the lowest level in BS7 (7451.4758 mg kg^-1). The average K concentration in the BC group was 12578.1828 mg kg^-1, peaking in BC1 (17069.1316 mg kg^-1) and reaching minimum levels in BC9 (8830.2196 mg kg^-1). Notably, both BS8 and BC1 originated from Gansu Province; their elevated K levels suggest that the Bupleurum from this region may have more efficient K absorption, transport, and distribution via various transport mechanisms. Ca is equally vital and structurally binds to negative groups of organic compounds to crosslink pectins in plant cell walls for rigidity, while coordinating with phospholipid phosphate groups in cell membranes to maintain membrane integrity. The removal or substitution of Ca can disrupt membrane structure. In cellular signaling, its low cytoplasmic free concentration makes Ca an ideal secondary messenger; various stimuli can rapidly alter its concentration, and Ca participates in multiple physiological responses critical for plant growth and environmental adaptation [31]. The average Ca concentration of the BS group was 4824.4872 mg kg^-1, ranging from 8965.2272 mg kg^-1 (BS6, maximum) to 3159.7291 mg kg^-1 (BS4, minimum). The average Ca concentration of the BC group was 5958.5599 mg kg^-1 Ca, with BC5 showing peak levels (9186.4421 mg kg^-1) and BC8 showing the lowest levels (4355.6263 mg kg^-1). Both BS6 and BC5 were sourced from Shanxi Province, suggesting that the local soil may be calcium-rich. Analysis of other nutritional elements revealed that the P concentration ranged from 1224.8908 to 5052.5137 mg kg^-1 in BS and 1269.1740 to 2900.1967 mg kg^-1 in BC; the Mg concentration ranged from 1696.2107 to 2588.7672 mg kg^-1 in BS and 1585.4156 to 3062.6020 mg kg^-1 in BC; and the S concentration ranged from 902.9358 to 2549.4442 mg kg^-1 in BS and 1265.9644 to 1698.2956 mg kg^-1 in BC. Notably, for these three elements, two samples with the highest contents in the BS group and all the highest-content samples in the BC group were from Shaanxi Province, suggesting that local soil or growing conditions may promote the accumulation of P, Mg, and S. In comparison to those of these five major elements, the Na and Fe accumulation levels were lower, with Na concentrations ranging from 320.3859 to 1118.6317 mg kg^-1 in BS and 265.5561 to 2812.8969 mg kg^-1 in BC and Fe concentrations ranging from 318.7366 to 912.0168 mg kg^-1 in BS and from 501.2870 to 1396.0996 mg kg^-1 in BC.

Pearson correlation analysis revealed significant interelement relationships, predominantly positive correlations, among the 25 inorganic elements in both the BS and BC groups (Figure 8). In the BS group (Figure 8a), Ca was strongly positively correlated with Na (r = 0.899), Fe (r = 0.812), and Co (r = 0.767). Mg was strongly correlated with V (r = 0.720) and Cr (r = 0.666), whereas Fe was significantly correlated (r > 0.5) with 11 elements, including Ca, Na, Al, Mn, and Ni. In the BC group (Figure 8b), K was strongly correlated with Mg (r = 0.713) and V (r = 0.779). Magnesium displayed particularly extensive interactions, with positive correlations (r > 0.5) observed for 17 elements, including K, Fe, P, Mn, Zn, Cu, and V. These findings delineate a complex network of elemental interactions and suggest a predominant synergistic relationship pattern among the inorganic elements in Bupleurum. The coordinated accumulation behavior implies potential shared absorption pathways or physiological coregulation mechanisms within the plant system.

Close

Figure 8.

Pearson correlation heatmap of 25 elements between two botanical origins.

Export to PPT

3.4. Metabolic characteristics of 18 batches of samples

Based on ultraperformance liquid chromatography (UPLC)-Q-TOF-MS analysis in positive ion mode, 135 compounds, including organic acids, flavonoids, and triterpenoid saponins, were identified. In addition, targeted quantitative analysis was performed for SSa and SSd. The differentially abundant metabolites between the BS and BC groups were screened using combined multivariate and univariate statistical methods via the Metware Cloud platform (https://cloud.metware.cn), with stringent criteria set at VIP > 1, FC ≥ 2 (or FC ≤ 0.5), and p < 0.05. This analysis revealed 20 significant differentially accumulated metabolites (DAMs) (Table S2), including nine upregulated and 11 downregulated metabolites (Figure 9). Heatmap analysis revealed clear separation between the BS and BC groups into distinct clusters, except for sample BS3, indicating substantial metabolic differences (Figure 10). The relatively small number of DAMs may result from increased intragroup variability because of inherent metabolic differences among samples of different geographical origins. Among the DAMs, most organic acids, such as m-(beta-acetyl-alpha-ethyl-p-hydroxyphenethyl) benzoic acid (2.0080-fold), 7-hydroxyoctanoic acid (1.8316-fold), and pinolidoxin (1.7668-fold), were upregulated in the BS group. Among the 63 detected flavonoids, only 9 were identified as DAMs, most of which were downregulated in the BS group, suggesting minimal differences in flavonoid composition between BS and BC. This observation aligns with the MSI results. Additionally, the abundance of noncharacteristic triterpenoid saponins in Bupleurum did not significantly differ between the two groups. Targeted analysis of the characteristic saikosaponins SSa and SSd revealed significant differences both between and within the BS and BC groups, with higher levels in the BC group. These findings are consistent with the MSI results, which indicates increased accumulation of common saikosaponins (e.g., SSc and SSf) in BC. However, the results of the histochemical localization analysis yielded the opposite conclusion, with a higher total saponin content in BS. This discrepancy may arise from two factors: first, potential bias due to the limited sampling from single geographical sources, and second, the possibility that BS contains uncharacterized triterpenoid saponins, leading to higher total saponin content. These observations are supported by previous studies: Guo et al. reported higher total saponin content in BS than in BC [32], whereas Tan et al., using histochemical localization and visible spectrophotometry, confirmed that BS results in higher saponin levels across all vegetative organs [33]. These findings suggest that BS may contain diverse types of unidentified triterpenoid saponins, warranting further chemical characterization and bioactivity studies. All the tested samples met the ChP requirements, with the SSa+SSd contents exceeding 0.3% (Table 4). Notably, the concentrations of the six samples from Gansu (BC2, BS8), Shanxi (BC4, BC5), Heilongjiang (BS2), and Sichuan (BC7) were particularly high, surpassing 1%. Previous studies have identified Gansu, Shanxi, Shaanxi, Hebei, and Heilongjiang as the primary Bupleurum cultivation regions [34], and our results confirm that most high-quality samples here originated from these areas. Interestingly, the superior SSa and SSd levels observed in the Sichuan samples suggest that this region may also possess favorable ecological conditions or unique germplasm resources suitable for Bupleurum growth, warranting further investigation into its potential as an emerging production zone.

Close

Figure 9.

Bar chart of fold change between group BS and group BC. Note: Red color: upregulated (fold change ≥ 2); Green color: downregulated (fold change ≤ 0.5).

Export to PPT

Close

Figure 10.

Clustering heatmap of samples based on LC-MS data.

Export to PPT