Comprehensive Chemical Profiling and Quantitative Determination of Solanum americanum Mill. and S. nigrum Linn. Combined by UHPLC‐HRMS With Feature‐Based Molecular Networking

ABSTRACT

Solanum americanum Mill. and S. nigrum Linn. are traditional Solanaceae‐derived herbal medicines. Their sprouts and mature fruits are utilized as both vegetables and medicinal herbs. Although S. americanum Mill. is classified as a primary variety of S. nigrum Linn., significant pharmacological variations and incompletely characterized phytochemical profiles differentiate their applications. In this study, UHPLC/LTQ‐Orbitrap mass spectrometry (HRMS) (Thermo Fisher Scientific, USA), assisted with feature‐based molecular networking, was applied for in‐depth profiling of S. americanum Mill. and S. nigrum Linn. With the help of feature‐based molecular networking, five clusters were characterized, including flavonoids, phenylpropanoids, fatty acids, alkaloids, and steroids. S. americanum Mill. demonstrated markedly higher relative abundance in flavanols than S. nigrum Linn., potentially elucidating its distinct antioxidant activity and its anti‐inflammatory and antibacterial efficacy. In addition, metabolomics coupled with multivariate analyses screened and identified 16 discriminative chemical markers enabling differentiation between S. americanum Mill. and S. nigrum Linn. Building on the above, to elucidate the chemical basis underlying their distinct pharmacological activities, we quantified 13 identified key differential constituents of S. americanum Mill. and S. nigrum Linn. extracts by UHPLC‐QTRAP‐MS/MS using the multiple reaction monitoring (MRM) mode. The average contents of isoquercitrin, rutin, protocatechuic acid, and chlorogenic acid in S. americanum Mill. are significantly higher than those in S. nigrum Linn., suggesting that these components may serve as potential quality markers for S. americanum Mill. This study elucidates the chemical basis underlying S. americanum Mill. and S. nigrum Linn.’s pharmacological activity differences, establishes quality control tools, and supports S. americanum Mill.’s rational utilization in herbal medicine development.

1. Introduction

Longkui is a kind of black nightshade. Both its fruits and young leaves are edible as vegetables [1], and it represents a crucial resource for antitumor constituents in natural medicine. The herb Longkui generally pertains to the dried whole‐plant materials of Solanum nigrum Linn. (SNL) or S. americanum Mill. (SAM) [2]. At present, differentiating between these two black nightshade species predominantly depends on morphological characteristics. The plants of SNL are relatively robust, with leaves mostly ovate in shape. Compared with SNL, the plants of SAM are more delicate, and their inflorescences are less abundant. Although SNL and SAM are taxonomically within the same family, their primary pharmacological actions diverge significantly. SNL is known for its heat‐clearing, detoxifying, blood‐circulation‐promoting, and swelling‐reducing properties. It is frequently employed in the treatment of pustules, pain, swelling, erysipelas, edema, sprains, nephritis, and chronic bronchitis [3]. Moreover, extracts from SNL have demonstrated potent antitumor activities, effectively inhibiting the growth of lung, liver, and stomach cancer cells [4]. SAM has a sweet, bland flavor and a cool nature. It functions to clear heat, resolve dampness, dissipate blood stasis, and relieve pain [5]. Clinically, it is often used for addressing menstrual irregularities, abdominal pain due to blood stasis, heat‐related syndromes, urinary calculi, and other ailments. In some parts of northeastern Nigeria, the herb has been used traditionally for treating diarrhea and dysentery [6]. In addition, studies have reported that SAM exhibited strong inhibitory effects against Candida albicans, suggesting its potential for clinical application in treating candidal vaginitis [7]. Due to the differences in efficacy between SNL and SAM, accurate identification of the medicinal origin of nightshade is of utmost importance. Failure to do so could pose a serious threat to the safety of its clinical use. However, there is a lack of effective analytical techniques for ensuring the quality control (QC) of nightshade products derived from different sources.

Over the past few decades, research on phytochemicals had provided solid evidence that the whole plant of SNL abounded in various components. These encompassed steroidal saponins, steroidal alkaloids, flavonoids, coumarin, lignin, organic acids, volatile oils, along with polysaccharides [8, 9]. Steroidal saponins and steroidal alkaloids represented the principal bioactive compounds in SNL, contributing to their diverse pharmacological activities, such as antitumor, anti‐inflammatory, antioxidant, and antibacterial [10, 11], and also possess a certain ecological function by suppressing severe agricultural pests [12]. However, the reported literature of SAM mainly focused on flavonoids (e.g., rutin) and phenolic acid components (e.g., chlorogenic acid and gallic acid) [13, 14]. Although SAM and SNL are closely related species, their chemical constituents may also vary, which could account for the significant disparities in their pharmacological activities. To date, no in‐depth and comparative compositional analysis has been performed on SAM and SNL, and the detailed chemical makeup of them remains largely uncharted territory. Consequently, the systematic characterization of the two traditional medicines not only provides a means of QC for differentiating SAM from SNL, but also provides a scientific foundation for elucidating their distinct pharmacological activities.

Metabolite profiling and comprehensive characterization of multiple constituents aim to elucidate the therapeutic basis and establish QC standards for functional supplements and traditional Chinese medicines (TCMs), and serve as a critical strategy to promote their modernization and globalization [15, 16]. Untargeted metabolomics is extensively employed for the identification of diverse basal metabolites in plants. It, which aims at comprehensive chemical profiling, serves as a powerful analytical tool for the discovery of potential markers [17, 18]. These markers are suitable for the QC of food and TCMs, which encompass similar chemical substances derived from various plant species, different geographic origins, and varying growth stages, at visualizing these markers through the application of multivariate statistical analysis methods [19, 20].

Feature‐based molecular networking (FBMN) is a popular analysis approach for liquid chromatography–tandem mass spectrometry‐based non‐targeted metabolomics data [21, 22]. Compounds possessing comparable chemical structures yield identical fragments in mass spectrometry and can be interconnected to form a molecular network within the global natural products (GNPS) social molecular networking platform. The resultant networks expedite the swift identification of compounds and the discovery of novel ones within metabolomic datasets [23]. This approach is extensively employed in pharmaceutical analysis, metabolomics, drug discovery, and various other medical applications. FBMN has been utilized to comprehensively characterize the chemical constituents of foods, thereby streamlining the rapid chemical comparison of samples procured from diverse edible portions or via distinct processing techniques [24].

In the present study, a UHPLC‐HRMS coupled with the FBMN method was conducted to comprehensively profile the metabolites in SAM and SNL. The identified compounds were categorized, and the fragmentation patterns of each type of compound were summarized. Then, using untargeted metabolomics in combination with multivariate statistical analysis, quality markers capable of distinguishing between SAM and SNL were screened out. Furthermore, based on the identified metabolic profiles, the predominant bioactive constituents with high relative abundance were selected for targeted quantitative analysis through mass spectrometry method development. Figure 1 illustrates the workflow of the specific analysis.

FIGURE 1.

The workflow of qualitative and quantitative analysis of SAM and SNL.

2. Material

2.1. Chemicals and Reagents

Protocatechuic acid, chlorogenic acid, neochlorogenic acid, isoquercitrin, khasianine, tomatidine standards were obtained from Shanghai Yuanye Biotechnology Co. Ltd and solasonine, solamargine, solasodine rutin, luteolin‐6‐C‐glucoside, citric acid, caffeic acid, quercetin, zingiberensis saponin, timosaponin A3, trillin, and hyperoside were obtained from Chendu Must Biotechnology Co. Ltd. The purity of all standards was determined to be higher than 98%. All other reagents and solvents used were of analytical grade. Formic acid (Roe Scientific Inc., Cranbury, NJ, USA), LC/MS‐grade acetonitrile and methanol (Honeywell, Charlotte, NC, USA), and home‐made ultrapure water were used for the mobile phase.

2.2. Plant Materials and Sample Preparation

Here, 11 batches of SAM and 15 batches of SNL were collected from various herbal markets across China. The samples were identified by Prof. Quanxi Mei. The origin of the 26 samples is shown in Table S1. All voucher specimens were stored in a dry, darkroom at the Key Laboratory of TCM Clinical Pharmacy, Shenzhen Bao'an Authentic TCM Therapy Hospital. Herbal samples were ground and filtered through a 65‐mesh sieve, with 0.5 g of each fine powder accurately weighed. The samples were then ultrasonically extracted in 10 mL of methanol at room temperature for 30 min. The resulting solutions were centrifuged at 13 000 rpm for 10 min, and the supernatant was used for subsequent analysis.

2.3. UHPLC‐HRMS Analysis

Data acquisition was performed using UHPLC/LTQ‐Orbitrap mass spectrometry: Dionex UltiMate 3000 UHPLC ultra‐high performance liquid chromatograph (Thermo Fisher Scientific, San Jose, CA, USA), equipped with a quaternary pump solvent system, column oven, injector and diode array detector, connected in series with an LTQ‐Orbitrap Velos Pro linear ion trap‐orbitrap mass spectrometer (Thermo Fisher Scientific, San Jose, CA, USA); Xcalibur 2.2 software (Thermo Fisher Scientific, San Jose, CA, USA) was used for instrument control and data processing. A Waters ACQUITY UPLC HSS T3 chromatographic column (1.7 µm, 2.1 mm × 100 mm, Waters, Milford, MA, USA) was used. The mobile phase consisted of 0.01% formic acid in water (A) and 0.01% formic acid‐acetonitrile (B). The gradient elution program was as follows: 0–15 min: 2%–25% B, 15–30 min; 25%–45% B, 30–31 min: 45%–100% B, 31–35 min: 95% B. The injection volume was 2 µL, and the flow rate was 0.3 mL·min⁻¹.

The parameters of the electrospray ionization source were set as follows: The capillary temperature was 320°C, the source heat temperature was 200°C, the source voltage was 2.7 kV. The sheath gas, auxiliary gas, and sweep gas were all nitrogen (N₂), with flow rates of 15, 10, and 2 arbitrary units, respectively. Negative ion mode: Mass spectrometry fragmentation was carried out using the higher‐energy collisional dissociation (HCD) mode. One cycle consisted of Events 1, 2, and 3. Event 1 was a full scan at the first level, with a mass‐to‐charge ratio range of 100–1500, a resolution of 30 000, and the data format was profile. Events 2 and 3 triggered the HCD fragmentation of the strongest ions in the full scan at the first level. The normalized collision energies were set to 80% and 100%, respectively, and the MS cycle time was set at 0.03 s, with a mass‐to‐charge ratio range of 50–1500, a resolution of 7500, and the trigger threshold for the lowest signal at the second level was set to 5000. Dynamic exclusion was set, with the number of repetitions and duration being 1 time and 10 s, respectively, and the exclusion duration was 10 s. Positive ion mode: Mass spectrometry fragmentation was carried out using the collision‐induced dissociation (CID) mode and high energy. One cycle consisted of Events 1, 2, and 3. Event 1 was a full scan at the first level, with a mass‐to‐charge ratio range of 100–1500, a resolution of 30 000, and the data format was profile. Events 2 and 3 triggered CID and HCD fragmentation of the most abundant ions from the full MS scan, with normalized collision energies set to 35% and 50%, respectively. The MS cycle time was set at 0.03 s, with a mass‐to‐charge ratio range of 50–1500, a resolution of 7500, and the trigger threshold for the lowest signal at the second level was set to 5000. Dynamic exclusion was set, with the number of repetitions and duration being 1 time and 10 s, respectively, and the exclusion duration was 10 s.

Before running the samples, six injections of the total QC solution were performed. After that, one injection of the corresponding QC solution was inserted for every eight injections of the sample. Data acquisition and processing were carried out using Xcalibur 2.2 software.

2.4. Feature‐Based Molecular Networking

The FBMN was constructed using UHPLC‐HRMS data from SAM and SNL; raw data files were converted into 32‐bit mzXML files using MS Convert software. MzXML files were imported into the Progenesis QI software for peak alignment, peak picking, and peak annotation. Both feature quantification table (CSV file) and MS/MS spectral summary (MSP file) were exported and uploaded to the Global Natura Products Social molecular networking platform (https://gnps.ucsd.edu). The FBMN was constructed via WinSCP (https://winscp.net), following the online workflow (https://ccms‐ucsd.github.io/GNPSDocumentation/featurebasedmolecularnetworking‐with‐progenesisQI/) (The molecular network and parameters could be accessed through the link: https://ccms‐ucsd.github.io/GNPSDocumentation/networking/) [25, 26].

In the process of creating consensus spectra, the following MN parameters were meticulously applied: A stringent minimum cosine score of 0.70, a parent mass tolerance of 0.02 Da, and a fragment ion tolerance of 0.02 Da. These parameters ensured the maximum connectivity component size of 100, with a minimum requirement of six matched peaks. The spectral library search was rigorously conducted against the comprehensive GNPS spectral libraries, including NIST14, Massbank, and ReSpect. For all network and library spectrum matches, a score above 0.7 was mandated, along with at least six matched peaks. Network visualization was skillfully performed using Cytoscape (Version 3.10.2), enabling a clear and insightful representation of the data.

2.5. Development of a Quantitative Analytical Method Using UHPLC‐QTRAP‐MS/MS

Based on chemical composition characterization and the availability of reference substances, we selected the high abundant and common active ingredients in SAM and SNL to establish a simultaneous determination method for the multiple components’ comparison, aiming to lay the foundation for the improvement of the quality standard of SAM.

Studies have demonstrated that solasonine, solamargine, solasodine, α‐solanine, khasianine, and tomatine exhibit well‐defined pharmacological activities, which were therefore selected as marker compounds for quantitative analysis [27]. At the same time, in order to expand the detection range of compound species, four flavonoids (isoquercitrin, rutin, luteolin‐6‐C‐glucoside, hyperoside), three phenylpropionic components (protocatechuic acid, chlorogenic acid, and neochlorogenic acid) and three phenylpropanoid components (protocatechuic acid, chlorogenic acid, and neochlorogenic acid) were also selected as detection indicators.

2.5.1. UPLC‐QTRAP‐MS/MS Conditions

Low‐resolution MS and MS/MS analysis were obtained on an API 6500 + Qtrap mass spectrometer equipped with an ESI interface in positive and negative ion modes (AB Sciex, Framingham, MA, USA). Equipment control was performed using Analyst software ver.1.7.2 (AB Sciex). The data analysis was performed by using SCIEX OS software ver.2.1.6.59781 (AB Sciex). UHPLC separation system contained a C18 column (ACQUITY UPLC BEH C18; 2.1 mm × 50 mm, 1.7 µm) from Waters Corporation. The column temperature was kept at 40°C, and the injection volume was 1 µL. The conditions of the MS detector were set as follows: Ion spray voltage, 5.5 kV; capillary temperature, 550°C; ion source GS1, 55 psi; ion source GS2, 55 psi; curtain gas, 35 psi. The mass spectrum was recorded in the range m/z 100−900 with the positive mode. The dwell time of each ion pair was 100 ms. Nitrogen was used in all cases. Multiple reaction monitoring mode was applied for quantification. The gradient elution and other optimized parameters used are shown in Tables S2 and S3.

2.5.2. Preparation of Mixed Standard Solutions

The sample preparation methods for SAM and SNL followed the same procedure as described in Section 2.2.

All the standard substances were dissolved in methanol to prepare a 1 mg mL⁻¹ single standard stock solution. These were then diluted with methanol to form a series of mixed standard solutions. By establishing the relationship between the peak area (Y) and concentration (X) of each component, standard curves were drawn.

2.5.3. Specificity, LOQ, Precision, Repeatability, Stability, and Accuracy

To investigate the influence of solvents on the determination of each active component, methanol was injected into the UPLC‐QTRAP‐MS/MS. On the basis of the linearity investigation, the mixed standard solutions were continuously diluted until the concentration with a signal/noise ratio of 10:1 was the LOQ.

To evaluate the instrumental precision, the mixed standard solutions were prepared. To determine intra‐ and inter‐day precision, six contents of the above samples were analyzed in injections within 1 day, followed by three injections per day over three consecutive days, and the RSD values were calculated to evaluate the precision of the method. The content of SAM (batch number O_20230001) was prepared. The six above sample solutions were analyzed to evaluate method repeatability.

The sample solutions of the stability test were injected and analyzed at 0, 2, 4, 8, 12, and 24 h (stored in the sample chamber of the UHPLC, dark, 4°C) to investigate the stability of the sample solutions.

The recovery rate was tested by the standard addition method to evaluate the accuracy of the method. The contents of SAM with the known concentration of each component were taken, and six samples were weighed. The standard solution was added in six parallels. The formula used for calculating the recovery rate was ([measured amount − contained amount] / added amount) × 100%; the detailed data are shown in Tables S4 and S5.

2.5.4. Content Determination of the Multiple Samples

Using the established quantitative analysis method for the 13 compounds, the contents of multicomponents in 11 batches of SAM and 9 batches of SNL were determined simultaneously in order to evaluate their quality.

2.6. Data Processing and Multivariate Statistical Analysis

The profile raw data were processed on Progenesis QI 2.1 software (Waters, Milford, USA) for peak alignment and peak picking. A data matrix of detected metabolite features was then generated, involving sample code, t_R , m/z, and normalized abundance. Precise precursor and fragment masses were used for the identification of metabolites. SIMCA 14.1 software (Umetrics, Umea, Sweden) was used for principal component analysis (PCA), orthogonal partial least‐squares discrimination analysis (OPLS‐DA), and cluster analysis (CA).

3. Results and Discussion

3.1. Chemical Profiling of SAM and SNL

The high‐resolution MS data of SAM and SNL were acquired within 40 min by using UHPLC‐HRMS. Figure 2 displays the BPI (base peak ion) chromatograms, revealing significant differences in the metabolic profiles of SAM and SNL, especially in negative ion mode. A series of compounds were identified by means of standards, literature searches, and databases, with the details presented in Table S8. Seventeen out of these metabolites were identified by utilization of standards. They were initially identified based on the consensus of their retention times, precursor ions, and fragment ions comparing with reference substances, as shown in Figure S1.

FIGURE 2.

BPI chromatography of QC samples in positive and negative ionization modes for SAM and SNL.

3.2. FBMN Aided the Chemical Characterization of SAM and SNL

FBMN assists compound annotation by constructing molecular networks based on MS/MS fragment similarity; in addition, it visualizes compound abundance variations among samples, improving MS data interpretability. In this study, a comprehensive MN of SAM and SNL extracts was obtained. As shown in Figure 3, the molecular map revealed five clusters including flavonoids (Cluster A), phenylpropanoids (Cluster B), fatty acids (Cluster C), alkaloids (Cluster D), and steroids (Cluster E). These clusters could be clearly distinguished from each other by their MS/MS fragments. In the molecular network, the red‐to‐green area ratio of each node represents the relative abundance of compounds in SAM versus SNL. Notably, flavonoid Cluster A shows significantly higher SAM abundance, with its predominant components identified structurally annotated as flavonols. Subsequently, 284 compounds were identified using a multi‐strategy approach: (1) summarizing fragmentation patterns by comparison with reference standards; (2) matching against public molecular networking libraries via FBMN; (3) using MS‑DIAL software for compound annotation; and (4) consulting relevant literature. These compounds included 32 alkaloids, 76 flavonoids, 50 steroids, 15 amino acids, 52 phenylpropanoids, and 59 other compounds. Annotation levels were assigned according to the metabolite identification standard published by the Metabolomics Society [28]. The detailed mass spectra information of these compounds is shown in Table S8.

FIGURE 3.

Molecular networks based on mass spectrometry data in positive and negative ion modes of SAM and SNL. Cluster A, flavonoids; Cluster B, phenylpropanoids; Cluster C, fatty acids; Cluster D, alkaloids; Cluster E, steroids. Component a, hyperoside; Component b, xanthorhamnin; Component c, isovitexin; Component d, rosmarinic acid; Component e, chlorogenic acid; Component f, tomatidine; Component g, solasodine; Component h, sarsasapogenin; Component i, (10E,15E)‐9,12,13‐trihydroxyoctadeca‐10,15‐dienoic acid.

3.2.1. Alkaloids

Steroidal alkaloids have been reported to be one of the more well‐studied components within the chemical structure of SNL [29]. The aglycone ion solasodamine typically exhibits a quasi‐molecular ion peak at m/z 414.33 in the positive ion mode. Following a series of dehydrations, fragment ions at m/z 396.32 and m/z 378.31 are generated. Subsequently, through the cleavage of Ring B and the loss of Ring E, ions at m/z 271.20, m/z 253.19, and m/z 157.10, among others, are sequentially produced (Figure 4A). Solasonine, solamargine, khasianine, and tomatidine are steroidal alkaloidal constituents that all use solasodine as a glycoside. These compounds generally have different molecular weight glycosyl groups attached to the nucleus to form a variety of alkaloids with higher molecular weights. They were generally observable in the positive ion mode. Taking solasonine as an example, it is composed of solasodine as the aglycone, to which a molecule of galactose, glucose, and rhamnose are attached. Under the positive ion mode, the pseudo molecular ion peak m/z 884.50 is observed, and the fragment m/z 866.49 is obtained by taking off a molecule of water; moreover, the successive loss of Gal, Glc, and Rha on the original structure yields the fragments m/z 720.40, m/z 558.38, finally we get the aglycone fragment m/z 414.33, on the basis of the loss of a molecule of water to get the fragment m/z 396.32, then cyclohexane breaks to lose a molecule of water after the loss of [C₈H₁₅N]⁺, respectively, to get the fragment ions m/z 271.20, m/z 253.19 (as presented in Figure 4B). For the unknown Compound 6 (t_R :19.04 min, C₃₃H₅₃NO₇, RDBeq 7.50), in the ESI chromatogram, a quasi‐molecular ion peak at m/z 576.29 was observed, which is 162 Da higher than the aglycone ion at m/z 414.33. This indicates that a Glc is attached to the aglycone ion at m/z 414.33. Due to this attachment, a dehydrated product at m/z 558.38 was shown. Subsequently, the aglycone ion at m/z 414.33 and its dehydrated product at m/z 396.32 were observed. Then, after a series of hydrolyses, fragments at m/z 271.20, m/z 253.19, and m/z 157.10 were obtained. By comparison with literature spectra, this compound was structurally annotated as γ‐solamargine tentatively.

FIGURE 4.

Identification process and fragmentation pathway of solasodine (A) and solasonine (B) based on fragment ions, including the extracted ion chromatogram of [M+H]⁺.

3.2.2. Flavonoids

Flavonoids, which belong to the class of secondary metabolites, function as antioxidants, exhibit anti‐inflammatory and anti‐carcinogenic activities, and have antimutagenic properties [30]. These compounds, most in the form of flavonoid glycosides, are widely distributed in SAM and SNL. In total, 76 flavonoid glycosides are characterized in SAM and SNL. These compounds are observed in both positive and negative ion modes. Commonly, the negative MS² fragmentation features of flavonoid glycosides involve the rutinose molecule (309 Da) and quercetin molecule (301 Da), as well as common neutral loss of sugars, 162 Da for Glc, 146 Da for Rha, and the generation of aglycone‐related anions typically observed at m/z 285.03, m/z 301.05, m/z 271.02, m/z 255.03, and so on. In addition, the C‐ring 1 and 3 positions are susceptible to RDA cleavage to produce the 1,3A⁻ ion at m/z 151.00. These MS/MS fragmentation features were readily embodied in the MS² spectrum of the reference compound isoquercitrin and quercetin (t_R : 14.22/20.05 min, C₂₁H₁₉O₁₂/C₁₅ H₁₁O₇, RDBeq 12.5/10.5) (as shown in Figure S2A,B).

In the case of Compound 32 (t_R :14.05 min, C₂₁H₂₁O₁₂, RDBeq 11.50), in the primary mass spectrometry, the dimer ion peak at m/z 465.10 [M+H]⁺ and the fragment peak at m/z 274.04 [M+H‐Glc]⁺ were detected. The presence of the latter peak indicated the loss of glycosides, which strongly suggested that the compound was a glycoside. Moreover, a fragment at m/z 257.04 was observed, signifying the elimination of a water molecule. Subsequently, the flavonoid aglycones underwent further cleavage, with the cleavage of the C₂─O and C₃─C₄ bonds, leading to the generation of the fragment ion at m/z 153.01 ([M+H‐Glc‐C₈H₈O]⁺). Based on these findings, it is inferred that the compound is quercetin‐3‐O‐galactoside.

3.2.3. Phenylpropanoids

Phenylpropanoids are parent molecules for all plant polyphenols, the largest class of secondary metabolites produced via the shikimic acid pathway [31]. By analyzing the positive and negative MS² data, 52 compounds were identified or tentatively characterized from SAM and SNL. Organic acids are one of the classes of compounds in the phenylpropanoid class, and phenolic acids are usually observed in the negative ion mode with many fragment ions, usually fragments such as m/z 191.05, m/z 179.03, m/z 251.02, m/z 135.04, and so on. The structural elucidation of reference compounds, including neochlorogenic acid and chlorogenic acid (Figure S3A,B), was performed. Chlorogenic acid consists of one molecule of caffeic acid and quinic acid. The fragments of m/z 191.0 and m/z 179.03 were observed in the negative ion mode, which cleaved into caffeic acid fragments and quinic acid fragments, and the caffeic acid fragments further decarboxylated, and the m/z 135 fragment ions were observed. The pattern of cleavage of the organic acids, such as quinic acid and citric acid, is also similar to the cleavage pattern of chlorogenic acid.

For the unknown Compound 170 (t_R :10.71 min, C₁₆H₁₇O₈, RDBeq 8.5), the dimer ion peak m/z 337.09 was observed in the negative ion mode. In the MS² spectrum, the peaks of m/z 191.05 [M‐H‐C₉H₇O₂]⁻ and m/z 163.04 [M‐H‐CO]⁻ peak are obvious, combined with the literature annotated that the unknown compound is 3‐O‐p‐coumaryl quinic acid.

3.2.4. Steroids

A total of 50 steroidal saponins were identified by standard comparison, database identification, and comparison of the literature in the SAM and SNL, which represents 16.78% of all compounds identified. The types of steroidal saponins in SNL are mainly isospirostanol type and furostanol type. Under the positive and negative ion modes, we identified steroidal saponins such as timosaponin A‐III, trillin, and zingiberensis saponin in samples using reference standards. The aglycone types of anemarrhena asphodeloides saponins mainly belong to spirostanol saponins. The aglycone of this type of saponin is formed by the condensation of smilagenin with two fructose molecules and one glucose molecule after the dehydrogenation of the hydroxyl group at the C₂₂ position [32]. Under the positive ion mode, we can observe the quasi‐molecular ion peak of timosaponin A‐III at m/z 741.44. A molecule of Gal and Glc are bonded to the aglycone ion of this saponin component. Following glycosidic bond fragmentation, fragment ions at m/z 579.38 and m/z 416.33 were observed in the mass spectrum. Based on m/z 416.33, one molecule of water is removed to give the fragment m/z 399.32. The pentacyclic alkane ring is then cleaved to give fragment m/z 273.22, and after the loss of one molecule of water, fragment m/z 255.21 is obtained. The corresponding product ion spectrum and fragmentation pathway are shown in Figure S4. In the analysis of the unknown Compound 95 (t_R :19.47 min, C₅₇H₉₅O₂₈, RDBeq 10.5), the dimer ion peak at m/z 1227.59 was detected in the positive ion mode. Notably, this value is 486 Da higher than the dimer ion peak of timosaponin A‐III. In the fragment ion spectrum, the peak at m/z 1065.55 [M+H‐Gal]⁺ was observed, providing evidence that the molecule was conjugated with a glycoside. Simultaneously, the fragment at m/z 741.44 [M+H‐Gal‐Gal]⁺ was detected, indicating that, based on the aglycone, the molecule was connected to three glycoside units. Through a comprehensive analysis in combination with relevant literature, the unknown compound was identified as trigoneoside XIIIa.

3.2.5. Others

A total of 59 additional components were characterized in SAM and SNL, which were mainly identified by database. These components were mainly primary metabolites of SAM and SNL, such as adenosine, glycyl‐leucine, uridine, and so on, and lipid components (such as linolenic acid, octadecatrienoic acid, etc.).

Finally, the compounds identified by SAM and SNL were subjected to a comprehensive comparison. A total of 145 distinct components were identified in SAM, while SNL contained 69 unique components, and 70 components were common to both samples, as depicted in Figure 5A. Upon plotting a bar chart illustrating the number of compound classifications (Figure 5B), it becomes evident that despite sharing similar compound types, SAM possesses a notably higher number of flavonoids (69 vs. 25) and phenylpropanoids compared to SNL (42 vs. 18), which is consistent with the results in the FBMN analysis. The flavonoids and phenolic acids present in TCM have been extensively demonstrated to exhibit robust antibacterial, anti‐inflammatory, and antioxidant properties. The substantial presence of these compounds in SAM further corroborates its remarkable antibacterial and anti‐inflammatory effects, from a chemical compositional perspective. Figure 5C vividly illustrates the overall distribution of chemical components in both samples, excluding other metabolites, highlighting the predominance of flavones and phenylpropanoids, which account for approximately 42.95% of the total components.

FIGURE 5.

Summary of compounds characterized from SAM and SNL by LC‐HRMS. Comparison between SAM and SNL (A); compound type distribution (B); distribution percentage histogram (C); 2D t_R –m/z scatter plot (D).

The number of characterized compounds in SAM and SNL, along with the two‐dimensional t_R –m/z scatter plots, are depicted in Figure 5D. The retention behavior and m/z of various compounds are related to compound types. Flavonoids, for instance, tend to elute between 10 and 25 min, featuring a broad range of molecular masses due to the presence of glycosyl groups, spanning from m/z 200 to m/z 1200. For alkaloids, the column used in this chromatography process is the T3 column, which has a strong retention capacity for polar compounds. However, under weakly acidic conditions, the retention time of alkaloids was significantly prolonged, usually between 15 and 25 min. Phenylpropanoids usually have small molecular weights and are concentrated in the range of m/z 100–500, with retention time mainly concentrated in the first 15 min. The side chain of steroid saponins has a glycoside structure and often has a large molecular weight, which is concentrated between m/z 400 and m/z 1300. The polarity of the hydrolyzed steroid saponins becomes low, and it is not easy to elution from the column, so the preservation time is mainly concentrated in 20–35 min. The amino acid components identified from SAM and SNL are less polar in the system, and show a narrow retention time, concentrated in the range of m/z 150–400.

3.3. Systematic Comparison of the Chemicals in SAM and SNL

Following pretreatment of the metabolite data, the preprocessed datasets were imported into SIMCA 14.1 software to systematically compare and distinguish chemical components between SAM and SNL. First, an unsupervised PCA model was used for analysis. The PCA score plot (Figure 6A) could testify the excellent performance of the analytical system, with the QC samples tightly clustered. Based on 36 485 difference variables, the supervised OPLS‐DA model was used to compare SAM and SNL in groups, and the model was subjected to a permutation test. R ² and Q ² were close to 1, indicating that the model fit the data well and had good stability and predictive ability (Figure 6D). Both the OPLS and HCA (Figure 6B,C) analyses demonstrated an exceptional distinction between the two groups, allowing for a significant division into two separate clusters.

FIGURE 6.

Multivariate statistical analyses of SAM and SNL. Unsupervised PCA score plot of SAM and SNL samples (A); OPLS‐DA score plot (B); HCA analysis (C); permutation test at 200 times used for the discrimination between the SAM and SNL (D); volcano plot analysis of potential compounds screened in SAM and SNL with VIP > 0 as baseline, green compounds on the left are potential metabolites in SNL samples and red compounds on the right are potential metabolites in SAM samples (E); heat map analysis of relative concentration trends of compounds in SAM and SNL identified by UHPLC‐HRMS (F).

As evident from the PCA and OPLS‐DA analyses, SAM and SNL can be distinctly classified into two separate categories. Notably, three samples from the SAM group significantly diverged from the primary cluster. This divergence could be ascribed to the fact that these samples were sourced from different cities. SNL samples came from Guangdong province and Fujian province, which are coastal cities at lower elevations, in comparison to the SAM samples, which mainly originated from the Yunnan region, characterized by its lofty topography and varied climate. The geographic location and climatic conditions of these two regions presumably influenced the development of SAM, consequently accounting for the noticed variability among the three samples.

The volcano is mainly used to display the data with differential characteristics between two groups of samples. As illustrated in Figure 6E, SAM samples exhibit significantly greater diversity in differential metabolites compared to SNL samples, a finding that aligns with both the metabolic profile analyses and spectroscopic characterization data. Thirty‐nine differential markers in SAM and SNL samples were selected from the volcano plot to create a heat map. All SAM and SNL batches were sorted into two separate categories, consistent with the results from PCA and OPLS‐DA (Figure 6F). As delineated in the heat map, these 39 different markers in SAM and SNL samples exhibited significantly relative abundance, with positively and negatively charged species demonstrating comparable quantities. These metabolites predominantly comprising flavonoid glycosides, saponin derivatives, and triterpenoid aglycones, exhibiting characteristic chromatographic retention between 12 and 26 min and m/z 200–1400. The box plot was drawn using the 16 compounds recognized from the heatmap, which shows the differences in abundance of the compounds in the two groups (SAM and SNL). Terpenoid compounds and flavonoid compounds show significant differences between the SAM group and the SNL group. This is the same as shown in FBMN Cluster A, the flavonoid components are not only plentiful in SAM but also significantly more abundant in relative abundance than SNL.

3.4. Methodological Validation and Quantitative Analysis Using UPLC‐QTRAP‐MS/MS

To accurately determine the content of active ingredients in SAM and SNL, a quantitative analytical method was established covering 13 compounds, including terpenoid alkaloids, flavonoids, and phenylpropanoids, which was then applied to the detection of multiple batches of samples. The results of the linear regression showed that all the compounds had good linear correlation in the concentration range, and the correlation coefficients were R ² ≥ 0.990. The regression equation, linear range, R ², and LOQ of the tested components are shown in Table S4.

The RSD of each component in the repeatability experiment was less than or equal to 9.76%, and the RSD of the intra‐ and inter‐day precision was less than or equal to 8.95% and 9.98%. The SAM solution was stable within 24 h, with RSD less than or equal to 9.87%. The recoveries of the 13 components ranged from 88.05% to 115.90%. Recovery results indicated that the adopted method could fully extract target active components from samples. The RSD values were all less than 5.30%. This shows that the developed method was accurate, stable, and reliable. The results are shown in Table S5.

The contents of the 13 components in 11 batches of SAM and 9 batches of SNL were determined using the explained method, and the results are shown in Tables S6 and S7. Figure 7 shows the histogram comparing the quantitative results of 13 compounds in SAM and SNL. From an overall perspective, the difference in alkaloid content between SAM and SNL is not significant, with an average content of 116.95 ± 1.35 µg·g⁻¹. Regarding flavonoid and phenylpropanoid constituents, SAM contains a higher concentration of flavonoids than SNL. Notably, the levels of isoquercitrin, rutin, chlorogenic acid, and protocatechuic acid were significantly elevated in SAM, reaching 2.55‐, 1.16‐, 2.68‐, and 1.99‐fold, respectively, compared to those in SNL.

FIGURE 7.

The content determination result of 13 compounds of SAM and SNL samples using UHPLC‐QTRAP‐MS/MS: Solasonine (A), solamargine (B), solasodine (C), α‐solanine (D), khasianine (E), tomatine (F), isoquercitrin (G), rutin (H), protocatechuic acid (I), chlorogenic acid (J), neochlorogenic acid (K), luteolin‐6‐C‐glucoside (L), and hyperoside (M). Number of batches: nine batches for SAM and nine batches for SNL.

4. Conclusions

In this study, a metabolomics approach based on UHPLC/LTQ‐Orbitrap combined with multivariate statistical analysis and molecular networking for the first time provided an effective strategy for screening and identifying potential metabolite markers for discriminating the SAM and SNL. Based on the systematic study assisted by FBMN technology, 284 chemical components were characterized, and SAM was found to have significant advantages in flavonoid species compared with SNL. This difference in chemical composition is likely to form the material basis for SAM to exhibit unique antioxidant and antibacterial activity. Multivariate statistical analysis also revealed significant differences between SAM and SNL species, and identified 39 compounds as distinguishing metabolite markers. Among these, neoruscogenin, gibberellic acid, chikusetsusaponin IV, syringaresnol‐4‐O‐beta‐D‐apiofuranosy, and so on showed substantial variations, which provides scientific reference for the development of quality evaluation methods. In addition, a quantitative method covering 13 compounds was established based on UHPLC‐QTRAP‐MS/MS, and the results indicated that the contents of multiple flavonoids and phenolic acids were significantly higher in SAM than SNL. These findings provide methodology support for tracing the origin of SAM.

Author Contributions

Lingyi Xin: writing – original draft, visualization, formal analysis, data curation, conceptualization. Xuemei Wei: formal analysis, data curation. Renyan Liu: data curation. Yun Li: supervision, resources. Yang Yang: supervision, resources. Yongtong Huang: data curation. Yu Zhang: data curation. Baodong Feng: formal analysis. Linqi Su: formal analysis. Quanxi Mei: writing – review and editing, validation. Qinhua Chen: writing – review and editing, supervision, resources, funding acquisition, conceptualization. Guangyi Yang: writing – review and editing, supervision, resources, funding acquisition, conceptualization.

Funding

This work was supported by the Sanming Project of Medicine in Shenzhen (SZZYSM202106004), the Sanming Project of Medicine in Shenzhen (SZZYSM202106009), the project of the Administration of Traditional Chinese Medicine of Guangdong Province (20251335), the National Natural Science Foundation of China (82272960), and 2024 High‐Quality Development Research Project of Shenzhen Bao'an Public Hospital (BAGZL2024059).

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Supporting File: jssc70469‐sup‐0001‐SuppMat.pdf.

Data Availability Statement

Data will be made available on request.