2.1 Materials and chemicals

AR (A. plantago-aquatica L.) was collected from Yaan City, Sichuan Province, China in Jan. 2022. Acetonitrile was obtained from Merck Chemicals (Shanghai) Co., Ltd. Reference substances were purchased from Macklin Biochemical Co. (Shanghai, China).

2.2 Compounds analysis of the extract from AR

The dried AR (100 g) was extracted using a liquid to solid ratio of 30 mL/g at a temperature of 50 °C and ultrasonic power of 240 W for a duration of 3 h. After centrifugation at 4500 r/m for 20 min, the supernatant was collected and subjected to concentration and freeze-drying. Subsequently, the freeze-dried sample was dissolved in 90 % acetonitrile to obtain a concentration of 1 mg/mL for analysis with HPLC-HRMS (UltiMate 3000 HPLC and Q-Exactive Orbitrap mass spectrograph, Thermo Fisher Scientific Inc., USA).

HPLC-HRMS equipped with Waters ACQUITY BHE C18 (2.1 × 150 mm, 1.7 μm) column at 40 °C. The mobile phases were 0.1 % formic acid (A) and acetonitrile (B) at 0.3 mL/min with gradient elution as follows: 0–3 min, 90 % A; 3–7 min, 90 %-80 % A; 7–10, 80 %-50 % A; 10–15 min, 50 %-15 % A; 15–20 min, 15 %-0% A; 20–25 min; 0 %-90 % A. The mass spectrograph utilized an electrospray ionization (ESI) source in positive ion mode, with the detailed parameters specified in the published literature (Zhao et al., 2022).

2.3 Activity prediction of new compounds

To predict the targets of the new compounds, the SwissTargetPrediction, SuperPred, Similarity ensemble approach, and PharmMapper databases were consulted based on their structural formula, and then Metascape was applied for KEGG analysis (Daina et al., 2019; Gallo et al., 2022; Keiser et al., 2007; Wang et al., 2017; Zhou et al., 2019). According to the KEGG results, the relevant targets of hyperlipidemia were collected with OMIM, GeneCards, and Disgenet databases. After that, the crossover targets between hyperlipidemia and new compounds were screened, followed by conducting PPI analysis with STRING to identify key targets (Szklarczyk et al., 2021).

2.4 Molecular docking of the new compounds and key proteins

PPARA was identified as key targets through the PPI analysis results. To validate the binding activity between these key targets and the new compounds, molecular docking was performed. The structure of the new compounds was drawn and their energy was then minimized using the MM2 force field. The protein receptors for PPARA (6LX4) was obtained from the PDB database. For docking with 6LX4, a docking box with a center at x/y/z = 8.04/17.23/56.23 and size x/y/z = 11.25/15.50/12.75 was employed. The docking of ligands to the key targets and the computation of their binding energy were performed using Autodock Vina 1.1.2 (Trott and Olson, 2010).

2.5 Drug-likeness evaluation and ADMET prediction of the new compounds

The drug-likeness of new compounds was assessed using SwissADME for properties related to absorption, distribution, and metabolism, while predictions for excretion and toxicity properties were made with ADMETlab 2.0 and ProTox-II (Banerjee et al., 2018; Daina et al., 2017; Xiong et al., 2021).

3.1 Identification of reported compounds

Based on the analysis of literature, triterpenoids were identified as the major natural products in AR(Liu et al., 2020; Wang et al., 2020). The molecular formula of these triterpenoids was summarized, and the response intensity of EIC was used as a criterion to filter out potential terpenoids with a threshold value set at 10⁴ (Li et al., 2017; Liu et al., 2020; Shu et al., 2023; Song et al., 2013; Yang et al., 2020; Zhang et al., 2021). A total of 12 EICs met this criterion, and they contained 20 reported triterpenoids because of the existence of isomers (Fig. 1).

Close

Fig. 1

The EIC of reported triterpenes from AR.

Export to PPT

The 20 triterpenoids were identified by MS/MS spectrometry combined with literature, and the results shown in Fig. 2, Table 1and Table S1 (Li et al., 2017; Song et al., 2013; Yang et al., 2020; Zhang et al., 2021; Zhang et al., 2022). Due to their strong response, peaks 11, 3, 5, and 15 were preferentially screened to summarize the law of mass fragmentation patterns for AR triterpenoids. Peak 11, 3, 5, and 15 were preferentially identified because of their intense response (Fig. 1). The molecular formula of peak 11 was C₃₀H₄₈O₆ because its parent ion of 505.3504 [M + H]⁺ (Table 1). Consequently, it could potentially be one of the 16-oxo-alisol A, 16β-hydroperoxyalisol B, and alismanol M (Liu et al., 2020). Fortunately, 16β-hydroperoxyalisol B and alismanol M were exclusively found in Alisma orientale, while Alisma plantago-aquatica was found to have a high content of 16-oxo-alisol A, as reported by Zhang et al. (Zhang et al., 2022). Thus, it is likely that peak 11 corresponds to 16-oxo-alisol A. Subsequently, the structure of peak 11 was identified by analyzing its fragmentation process (Fig. 2a and Table 1). The loss of a series of H₂O (18 Da) from the parent ion yielded some fragment ions at 487.3407 [M−H₂O + H]⁺, 469.3295 [M−2H₂O + H]⁺, 451.3197 [M−3H₂O + H]⁺. Furthermore, the loss of C₄H₁₀O₂ (90 Da), resulting from the rupture of the C₂₃-C₂₄ bond, led to the formation of fragment ions at 415.2828 [M−C₄H₁₀O₂ + H]⁺ and 397.2723 [M−H₂O−C₄H₁₀O₂ + H]⁺. The fragment ion at 353.2465 [M−H₂O−C₄H₁₀O₂−C₂H₄O + H]⁺ was generated by the cleavage of the C₂₀-C₂₂ bond, resulting in the loss of C₂H₄O (44 Da) from 397.2723 [M−H₂O−C₄H₁₀O₂ + H] +. Ultimately, peak 11 was confirmed to be 16-oxo-alisol A based on comparison with reference substances and relevant literature (Song et al., 2013; Yang et al., 2020).

Close

Fig. 2

The fragmentation process of 16-oxo-alisol A (a), 16-oxo-11-anhydro-alisol A (b), Alisol B (c), and Alisol B 23-acetate (d).

Export to PPT

Peak 3, with a parent ion at 473.3610 [M + H]⁺, suggests a molecular formula of C₃₀H₄₈O₄ and may be identified as alisol B, 11-deoxy-13β, 17β-epoxyalisol B, or alisol G (Liu et al., 2020). However, alisol B was found to be abundant in our sample based on HPLC analysis, indicating that peak 3 is likely alisol B (Fig. S2). This is supported by the presence of fragment ions at 455.3499 [M−H₂O + H]⁺ and 437.3410 [M−2H₂O + H]⁺, which result from the loss of multiple H₂O from the parent ion. Further fragmentation yielded fragment ions at 383.2933 [M−H₂O−C₄H₈O + H]⁺ and 365.2828 [M−2H₂O−C₄H₈O + H]+, corresponding to the loss of C₄H₈O, and the fragment ion at 339.2668 [M−2H₂O−C₄H₈O−C₂H₂ + H] + was formed by the loss of C₂H₂. By comparing the obtained data with reference substances and literature, peak 3 was conclusively identified as alisol B (Zhang et al., 2021). This is illustrated in Fig. 2b and Table 1.

Peak 5, with a parent ion at 487.3399 [M + H]⁺ (Table 1), was identified as C₃₀H₄₆O₅ by its molecular formula. This peak is likely to be either 16-oxo-11-anhydroalisol A or alisol C (Liu et al., 2020). The EIC of C₃₀H₄₆O₅ has two peaks at 19.04 (peak 4) and 20.38 min (peak 5), respectively. Peak 5 might be 16-oxo-11-anhydroalisol A because the retention time of alisol C was shorter than that of 16-oxo-11-anhydroalisol A in the C₁₈ column (Zhang et al., 2022). Fragment ions at 469.3306 [M−H₂O + H]⁺ and 451.3197 [M−2H₂O + H]⁺were generated by the loss of a series of H₂O from the parent ion. Additionally, The fragment ions at 397.2724 [M−C₄H₁₀O₂ + H]⁺ and 353.2463 [M−C₄H₁₀O₂−C₂H₄O + H]⁺ was formed by losing C₄H₁₀O₂ from parent ion and then losing C₂H₄O again (Table 1 and Fig. 2c). Consequently, through these observations, peak 5 was further identified as 16-oxo-11-anhydroalisol A (Zhang et al., 2022).

Peak 15 was identified as alisol B 23-acetate based on the molecular formula C₃₂H₅₀O₅, calculated by the parent ion at 515.3712 [M + H]⁺ (Table 1). HPLC analysis of our sample indicated a high abundance of alisol B 23-acetate (Fig. S2), further supporting this identification. This conclusion is reinforced by the fragment ions at 497.3620 [M−H₂O + H]⁺ and 479.3519 [M−2H₂O + H]⁺, which were generated by the loss of a series of H₂O (18 Da). Additional fragment ions at 455.3498 [M−OAc + H]⁺, 437.3401 [M−H₂O−OAc + H]⁺, and 419.3290 [M−2H₂O−OAc + H]⁺ were formed by the elimination of the acetoxy (OAc, 60 Da) group from the parent ion and 497.3620 [M−H₂O + H]⁺, and 479.3519 [M−2H₂O + H]⁺, respectively. Furthermore, the loss of C₄H₈O from 455.3498 [M−OAc + H]⁺ and 437.3401 [M−H₂O−OAc + H]⁺ resulted in the formation of fragment ions at 383.2932 [M−OAc−C₄H₈O + H]⁺ and 365.2619 [M−H₂O−OAc−C₄H₈O + H]⁺, respectively. Finally, the fragment ion at 339.2671 [M−H₂O−OAc−C₄H₈O−C₂H₂ + H]⁺ was generated from 365.2619 [M−H₂O−OAc−C₄H₈O + H]⁺ through the loss of C₂H₂ (Table 1 and Fig. 2d). All of these findings, along with the reference substance and existing literature, further confirm the identification of peak 15 as alisol B 23-acetate (Yang et al., 2020).

3.2 Identification of new compounds

The fragments m/z 339 or 353 are considered crucial for identifying AR triterpenoids by analzing the mass spectrum fragments of AR triterpenoids. Therefore, four new triterpenoid compounds were identified using this pattern, and named components A, B, C, and D, respectively. Compound A, with a molecular formula of C₃₁H₄₈O₅, was identified through the analysis of its symmetric sharp single peak at 22.52 min in the EIC (Fig. 3a) and its corresponding parent ion of 501.3556 [M + H]⁺ (Table 2). The fragmentation pattern of compound A suggested that it belonged to the triterpenoid class, based on the loss of H₂O (18 Da), C₂H₄O (44 Da), and C₄H₆ (54 Da) from its fragment ions, which conformed to the fragmentation rule of AR triterpenoids. As no triterpenoids with the molecular formula of C₃₁H₄₈O₅ were previously reported in AR, the structure of compound A was inferred based on its fragment ions. The characteristic fragment ion at 353.2465, which is indicative of alisol C and 16-oxo-aliso types, was produced by the loss of C₂H₄O from 397.2724 (Zhang et al., 2021). The fragment ion at 397.2724, in turn, was obtained from 451.3191 by losing C₄H₆, where 451.3191 was formed through the loss of H₂O from 469.3296. Notably, the difference value between 501.3588 and 469.3296 was approximately 32 Da, suggesting that 469.3296 might have resulted from the loss of the methoxy (OMe) group, which weighs 32 Da, from 501.3588. Additionally, the parent ion at 501.3588 lost a molecule of H₂O to give rise to 483.3474. The characteristic of compound A has an OMe group and might locate at the C₂₅ position because the methoxy group is mostly located at C₂₅ of AR triterpenoid (Feng et al., 2021). Thus, compound A was identified as 25-methoxy-16-oxo-11-anhydroalisol A.

Close

Fig. 3

The EIC and fragmentation process of 25-methoxy-16-oxo-11-anhydroalisol A (a), 25-methoxy-16-oxo-alisol A (b), and 25-methoxy-16-oxo-alisol A 23 or 24-acetate (c).

Export to PPT

The parent ion at 519.3662 [M + H]⁺ of compound B suggested its molecular formula was C₃₁H₅₀O₆ (Table 2). The fragment ions at 353.2459, 397.2733, 451.3176, and 469.3302 were consistent with compound A. In contrast, compound B exhibited additional fragment ions at 415.2826, 487.3389, 501.3558, and 519.3674. The fragment ion at 415.2826 was derived from 469.3302 through the loss of C₄H₆, where 469.3302 originated from the dehydration of 487.3389. Additionally, the fragment ion at 487.3389 was formed through the loss of the OMe group from 501.3588, as the difference value between them was 32 Da. Furthermore, the parent ion at 519.3674 lost a molecule of H₂O to yield 501.3558, which could potentially be compound A (Fig. 3b). Overall, compound B was identified as 25-methoxy-16-oxo-alisol A.

Compounds C and D were isomers with the molecular formula of C₃₃H₅₂O₇ because there was no difference in their parent and fragment ions (Table 2 and Fig. 3c). The fragment ions observed, such as 353, 397, 415, 451, and 469, were consistent with compound B. However, compounds C and D had additional fragment ions at 429, 483, 511, 529, and 543. The fragment ion at 429 was obtained from 469 by H rearrangement after losing C₃H₆. Similarly, the ion at 483 was obtained by removing the OAc group from 543 and then losing the OMe group, resulting in 451. Furthermore, the presence of the ion at 511 indicated the loss of H₂O and subsequent loss of the OAc group, leading to 451. The ions at 529 and 543 were generated by the loss of the OMe group and H₂O from the parent ion 561, respectively. These findings revealed that compounds C and D differed from compound B by just 42 Da, which corresponded to an acetyl group. Therefore, compounds C and D were likely formed by replacing one of the hydroxyl groups of compound B with an acetoxy group. Specifically, compounds C and D were identified as 25-methoxy-16-oxo-alisol A 23-acetate and 25-methoxy-16-oxo-alisol A 24-acetate, respectively, considering that acetoxy groups are typically located at the C₂₃ and C₂₄ positions of AR triterpenes (Liu et al., 2020). In summary, the 24 triterpenoids from AR were identified with HPLC-HRMS, including 20 reported triterpenoids and 4 new triterpenoids (Fig. 4).

Close

Fig. 4

The structure of 20 reported triterpenes (a) and 4 new triterpenes (b).

Export to PPT

3.3 Prediction of activity with new compounds

The targets of compounds A, B, C, and D were predicted using various databases such as SwissTargetPrediction, SuperPred, Similarity ensemble approach, etc., based on their structural formula. These compounds were found to have 184, 195, 162, and 429 targets, respectively. The predicted targets were then analyzed using Metascape for KEGG analysis, which revealed that they were significantly enriched in the PPAR signaling pathway (Fig. 5a-d). The PPAR signaling pathway is closely associated with hyperlipidemia as it regulates fatty acid β-oxidation, bile acid metabolism, and cholesterol metabolism (Bougarne et al., 2018). Additionally, Yan et al. indicated that AR triterpenoids can alleviate hyperlipidemia by activating the PPAR signaling pathway (Yan et al., 2022). Therefore, these new triterpenoids have the potential to alleviate hyperlipidemia by activating the PPAR signaling pathway.

Close

Fig. 5

The KEGG analysis of targets from 25-methoxy-16-oxo-11-anhydroalisol A (a), 25-methoxy-16-oxo-alisol A (b), 25-methoxy-16-oxo-alisol A 23-acetate (c) and 25-methoxy-16-oxo-alisol A 24-acetate (d).

Export to PPT

The PPI analysis was performed to screen the key targets. Afterwards, the targets of hyperlipidemia were collected and used to screen the crossover targets between new compounds and hyperlipidemia. As shown in Fig. 6a, 624 hyperlipidemia targets were collected, and 24, 31, 26, 24 crossover targets between compounds A-D and hyperlipidemia were filtered, respectively. Subsequently, the key targets were confirmed with a degree value greater than 10 in the PPI network. The PPARA and EGFR turned out to be potential key targets of compounds A-D in the treatment of hyperlipidemia (Fig. 6b and Table S2). PPARA regulates the expression of related genes for fatty acid β-oxidation, bile acid metabolism, and cholesterol metabolism and its agonist is a common clinical drug used in the treatment of hyperlipidemia (Duval et al., 2007). Additionally, EGFR is one of the target genes of PPARA. In normal circumstances, PPARA increases the expression of EGFR, but when PPARA binds to the ligand, its protein conformation changes and it cannot bind to the promoter of EGFR, leading to the downregulation of EGFR expression (Mahankali et al., 2015). The downregulation of EGFR is beneficial for hyperlipidemia because EGFR inhibitors reduce triglyceride and cholesterol levels in serum and liver tissues (Liang et al., 2018). Based on the PPI results, although both PPARA and EGFR may be potential key targets for compound A-D for treating hyperlipidemia, PPARA may be more important as literature suggested that AR triterpenoids can target the activation of PPARs (Li et al., 2016). Thus, PPARA was identified as target protein for compound A-D.

Close

Fig. 6

Prediction of anti-hyperlipidemic activity of four new triterpenoids. (a) Crossover targets of new compounds and hyperlipidemia. PPI analysis of crossover targets from 25-methoxy-16-oxo-11-anhydroalisol A (a), 25-methoxy-16-oxo-alisol A (b), 25-methoxy-16-oxo-alisol A 23-acetate (c) and 25-methoxy-16-oxo-alisol A 24-acetate (d) to hyperlipidemia.

Export to PPT

Compounds A-D might bind with PPARA because their binding energy is less than −5 kJ/mol (Table 3) (An et al., 2021). The binding energy of compound A was less than that of the control group (fenofibric acid, PPARA agonist, −6.9 kJ/mol), suggesting that compound A has more potential as a PPARA agonist. The visualization results of molecular docking further supported this finding, as they showed that the binding energy was mainly influenced by hydrogen bonds rather than hydrophobic interaction (Fig. 7 and Table 3). This suggests that the intermolecular force of the hydrogen bond is much greater than that of the hydrophobic interaction. Specifically, compound A demonstrated strong binding affinity to PPARA through hydrogen bonds with Thr-279, Ala-333, and hydrophobic interactions with Ile-241, Leu-247, Glu-251, Leu-254, Lys-257, Ile-272, Cys-275, Val-332, Tyr-334, Ile-339, and Phe-421. In conclusion, compound A exhibits promising potential for the treatment of hyperlipemia by binding to PPARA.

Close

Fig. 7

Molecular docking of 25-methoxy-16-oxo-11-anhydroalisol A (a), 25-methoxy-16-oxo-alisol A (b), 25-methoxy-16-oxo-alisol A 23-acetate (c), 25-methoxy-16-oxo-alisol A 24-acetate (d) to PPARA.

Export to PPT

3.4 Drug-likeness and ADMET prediction of the new triterpenoids

Drug development is greatly facilitated by ADMET prediction, as it significantly reduces the workload. Therefore, evaluating drug-likeness is crucial, considering that drug-like characteristics are heavily influenced by their ADMET properties (Gleeson et al., 2011). Table 4 contained drug-like characteristics of the new compounds and common decision rules of drug-likeness (Egan et al., 2000; Ghose et al., 1999; Lipinski et al., 2001; Muegge et al., 2001; Veber et al., 2002). Due to the large molecular weight, molar refractivity, and atoms of the new triterpenoids, it does not conform to the Lipinski and Ghose rules (Table 4). However, it conformed to the Muege rule with high molecular weight requirements and the Egan and Veber rules without molecular weight requirements. Thus, the ADMET properties of the new compounds can be further predicted.

The ADMET properties of the new triterpenoids were evaluated and presented in Table 5. In absorption, the results showed that the new compounds demonstrated high HIA and exhibited moderate permeability across Caco-2 cells. Additionally, they displayed an ability to inhibit P-glycoprotein, indicating their potential to enter cells and maintain intracellular drug concentration. In terms of drug distribution within the human body, the evaluation of blood–brain barrier penetration and plasma protein binding became crucial (Chen et al., 2020; Pardridge et al., 1986). The new triterpenoids showed moderate blood–brain barrier penetration ability and weak plasma protein binding ability, suggesting their potential to enter the brain and intracellular compartments. Xu et al. also reported similar findings, confirming the ability of AR triterpenoids to cross the blood–brain barrier (Xu et al., 2017). Moreover, the involvement of cytochrome P450 in the metabolic process of these new triterpenoids is likely, as they seem to act as substrates and inhibitors of cytochrome P450. The new compounds demonstrated a moderate clearance rate and exhibited significant differences in their half-life during excretion. Specifically, components B, C, and D had a higher probability of having a half-life of more than 3 h compared to compound A. This finding is consistent with literature that indicates the half-life of AR triterpenoids is mostly greater than 3 h, although some may have a half-life of less than 3 h (Xu et al., 2017). In terms of toxicity, the new ingredients showed pronounced differences, with compounds A and D having a higher median lethal dose (LD₅₀) compared to compounds C and D, suggesting that compounds A and D have a higher safety margin.