Visual authentication of steroidal saponins in Allium macrostemon Bge. and Allium chinense G. Don using MALDI-TOF imaging mass spectrometry and their structure activity relationship

2.1 Chemicals and plant samples

A total of thirty and thirty-two batches of Allium macrostemon Bge. (AMB, n = 30) and Allium chinense G. Don (ACGD, n = 32) samples were collected from different regions of China, respectively (Table S1). And they were taxonomically identified by Professor Xinsheng Yao who works in the College of Pharmacy, Jinan University, China, and stored at −80 °C before use. Diosgenin, gitogenin, laxogenin, sarsasapogenin and tigogenin, were purchased from Fans Biotechnology Co., ltd (Guangzhou, China). Other authentic standards including macrostemonoside A, macrostemonoside B, macrostemonoside E, chinenoside I, chinenoside II, were isolated from AMB and ACGD in our laboratory (Chen et al., 2010; Qin et al., 2016a; Qin et al., 2018). These standards were stored at −20 °C in our standard sample library at Jinan University, Guangzhou, China. 2,5-dihydroxybenzoic acid (DHB) and lithium trifluoroacetic acid (LiTFA) were obtained from Sigma-Aldrich (Shanghai, China). Liquid chromatography mass spectrometry grade acetonitrile, methanol, water and formic acid were obtained from Fisher Scientific (Fair Lawn, NJ). Other chemical reagents were of analytical grade.

2.2 Sample preparation

The extraction procedure was performed as described previously (Liang et al., 2022). In brief, the AMB and ACGD samples were first freeze-dried by vacuum freeze-dryer, and crushed using a mixer mill with a zirconia bead (30 Hz, 1.5 min). The lyophilized powder (0.1 g) was dissolved with 70 % methanol solution (1.0 mL), and the mixture was vortexed for 30 s every 30 min for 6 times in total. The sample was placed in a refrigerator at 4 °C overnight, and centrifugated at 12000 rpm for 10 min. The supernatant was filtrated with a 0.22 μm membrane and injected into UHPLC-MS/MS analysis.

2.3 Quantitative analysis

A UHPLC-MS/MS quantitative method for the main compounds in AMB and ACGD was developed and validated based on the limit of detection (LOD), limit of quantification (LOQ), linearity, precision, repeatability, stability, and recovery tests (AOAC international, 2002; Qin et al., 2016a, 2016b; Ren et al., 2022). The LOD and LOQ values were determined as 3 ∼ 5-fold and 8 ∼ 10-fold of the ratio of signal-to-noise (S/N), respectively. In addition, the calibration curves were developed from the respective peak area of each compound versus a set of seven concentrations of each analyte. Traditionally, the correlation coefficients (R²) over 0.9990 were satisfactory. Further, the accuracy of the instrument was evaluated after the test solution was continuously injected six times within 24 h. And the identical samples were prepared and injected six times for repeatability analysis. The stability tests were finished through placing the samples at room temperature for 0, 4, 8, 12 and 24 h. Moreover, the recovery test was carried out by adding known amounts at a ratio of 1:1, and recoveries were obtained based on the equation: recovery (%) = (observed amount − original amount) / spiked amount × 100.

2.4 UHPLC-MS/MS conditions

The chromatographic separation was performed using an UHPLC system (Waters Acquity UHPLC I-Class) on an HSS T3 column (2.1 mm × 50 mm, 1.7 µm) with mobile phase of pure water (A) and acetonitrile (B) both with 0.1 % formic acid. The gradient programs were 5 % B from 0 to 0.5 min, 5–18 % B from 0.5 to 1.0 min, 18–20 % B from 1.0 to 5.0 min, 20 % B from 5.0 to 7.0 min, 20–30 % B from 7.0 to 10.0 min, 30–55 % B from 10.0 to 15.0 min, 55–80 % B from 15.0 to 16.0 min, 80–100 % B from 16.0 to 17.0 min, 100 % B from 17.0 to 18.0 min. The flow rate was set as 0.5 mL/min and the column oven was set to 35 °C. The injection volume was 4 μL.

UHPLC analysis was coupled to a triple quadrupole mass spectrometer (Waters Xevo TQ-D). The operation parameters were as follows: capillary voltage 3.5 kV (ESI + )/1.5 kV (ESI-); cone voltage 50 V (ESI + and ESI-); desolvation gas flow 650 L/h; source temperature 350 °C. A series of multiple reaction monitoring (MRM) transitions were monitored for each period according to the steroidal saponins eluted within this period (Table S2). All data were processed on platform of Masslynx 4.1.

2.5 Multivariate statistical analysis

Unsupervised principal component analysis (PCA) between AMB and ACGD groups was performed by the prcomp statistics function within R platform (https://www.r-project.org) after obtained experiment data was unit variance scaled. Then, the data was processed with log transform (log₂) and mean centering analysis for orthogonal partial least squares discrimination analysis (OPLS-DA). VIP values were derived from OPLS-DA results, which also provided the score plots and permutation plots. A total of 200 permutations test was performed to avoid overfitting. The compounds which were defined as the most important variables between these two plants need meet three factors of VIP values ≥ 1, log₂FC (fold change) ≥ 1 and p < 0.05 (Huang et al., 2022).

2.6 Sample preparation for MALDI MSI assays

The fresh bulbs were first crosscut. For cryosectioning, the crosscut zone of bulbs was directly fixed in three drops of distilled water. The samples were cross-sectioned at 12 μm thickness at −20 °C using a Leica CM1950 cryostat (Wetzlar, Germany). Afterwards, the sections were placed on electrically conductive slides coated with indium tin oxide (ITO), and the slides were dried in a vacuum desiccator for 30 min. DHB solution (15 mg/mL) was prepared and the solvent was 90 % methanol–water solution containing 0.1 % LiTFA. Then DHB matrix solution was sprayed evenly on the ITO slide using TM-Sprayer matrix spray apparatus. The parameters were as follows: temperature 75 °C, flow rate 0.1 mL/min, pressure 8 psi. A total of 24 cycles were sprayed, and drying time between each cycle was 10 s.

2.7 MALDI MS imaging measurements

MALDI timsTOF MSI measurements were achieved on a prototype Bruker timsTOF flex MS system (Bremen, Germany) equipped with a 10 kHz smart beam 3D laser. The corresponding laser power was set to 80 % and then fixed throughout the whole assays. The mass range were from m/z 200 to m/z 1300 Da in positive ion mode. The spatial images were acquired at a 100 μm laser step size, and each mass spectrum consisted of 500 laser shots. All RAW data were imported into Bruker SCiLS Lab software, and normalized with the Root Mean Square. Then, the signal intensity in each visual image represented the normalized intensity. The red and blue spots represented the high and low content of one ion in the image, respectively.

2.8 Platelet aggregation inhibitory activity

Light transmission aggregometry (LTA) was classical and well-recognized approach to measure platelet aggregation rate (Lu et al., 2017). First, blood samples were collected into vacuum blood collection tube (containing 0.1 mL 109 mM citrate sodium) from healthy subjects. Differential centrifugation was applied to prepare platelet rich plasma (PRP) and platelet poor plasma (PRP) at room temperature. The blood samples were centrifugated at 950 rpm for 10 min, and the supernatant was PRP samples. Then, the residual blood samples were separated by centrifuging at 3200 rpm for 20 min to obtain PPP samples. Automatic platelet aggregation instrument AG800 (Shandong, China) was used to investigate the platelet aggregation rate. The transmittance of PPP sample was 100 % as a control group. Different inducers including arachidonic acid (AA), adenosine diphosphate (ADP) and collagen, and tested compounds (100 μM or 50 μM) were added into PRP samples, and their transmittance was dynamically measured and recorded. Platelet aggregation curve and maximum aggregation rate (%) were used to reflect the platelet aggregation inhibitory function of tested compounds.

2.9 Statistical analysis

All data were shown as mean ± standard deviation (n = 3). The differences between treatment and control groups were obtained by Student’s t-test. And the levels of significant difference were set at p < 0.05 (*), p < 0.01 (**) and p < 0.001 (***).

3.1 Chemical characterization of steroidal saponins in AMB and ACGD bulbs by UHPLC-MS/MS

In our previous study, we have performed the chemical profiles of AMB and ACGD by UHPLC/Q-TOF-MS (Qin et al., 2018; Qin et al., 2016b). Based on the optimized MRM transitions, a comprehensive profile of steroidal saponins in these two herbs was achieved by UHPLC-MS/MS (Fig. S1 and Table S2). A total of 77 steroidal saponins were detected in AMB and ACGD samples, of which 47 compounds were identified with authentic standards. Obviously, several compounds were characteristic components for AMB (Fig. S1A) and ACGD (Fig. S1B), respectively.

3.2 Identification of differential compounds between AMB and ACGD groups

As shown in Fig. S2A, AMB samples (n = 30) were markedly separated with ACGD samples (n = 32) in unsupervised PCA plot. These indicated there were significant differences among the structural type and intensity values of their compounds. Then, the volcano plot (Fig. S2B) gave the similar results that 36 variables and 30 compounds obviously up-regulated and down-regulated in AMB group compared with ACGD group, respectively. Further, 53 compounds (Table S3) in total with VIP values over 1.0 were identified as differential variables between AMB and ACGD groups in VIP plot (Fig. S2C).

3.3 Comparative analysis of spatial distribution of steroidal saponins in AMB and ACGD bulbs by MALDI MSI

As displayed in Fig. 1B, transverse sections showed the morphology and sizes of AMB accounted for no more than those 50 % of ACGD. And they have similar anatomical structure including tunic, leaf scale and developing flower bud as illustrated in Fig. 1C. Due to the morphology and sizes, ACGD have more leaf scales than AMB. Each leaf scale orderly and alternately arranged. The leaf scales of AMB and ACGD both occupied over 90 % of the whole transverse section.

Spatial distribution patterns of several main differential steroidal saponins in these two species are demonstrated in Fig. 2 and Table 1. Each visual image represented the adduct ions including [M + H-H₂O]⁺, [M + H]⁺, [M + Na]⁺ and [M + K]⁺ of steroidal saponins in AMB and ACGD bulbs. Overall, the steroidal saponins derived in AMB were observed mainly in tunic, showing a great contrast with the widely distributed saponins in ACGD. For example, the adduct ions at m/z 1075.4706 ([M + K]⁺) and m/z 1059.4956 ([M + Na]⁺) beard some similarities in ACGD and were not detected in AMB, which were mainly observed in the tunic and leaf scale regions (Fig. 2A). This was because these compounds were particular in ACGD, not in AMB. Similar results were obtained for the characterization of Chinenoside II and its derivatives in Fig. 2D. When investigated further, several steroidal saponins with molecular formulas of C45H74O18 or C45H76O19 were the common compounds with the same ions at m/z 903.4939 and 959.4648, which were also detected in these two herbs (Fig. 2C). In Fig. 2F, the weak visualization of common ion m/z 755.4229 illustrated the intensity/mass of macrostemonoside S in AMB and compounds 17 and 18 in ACGD were low. In addition, the adduct ions at m/z 943.4335 (Fig. 2B), m/z 915.4634 and m/z 971.4244 (Fig. 2E) were all observed in AMB tunic and ACGD bulbs, although compounds 26 and 27 were unique in ACGD, while macrostemonoside I was particular in AMB.

Close

Fig. 2

Matrix-assisted laser desorption/ionization coupled MS images of AMB (left) and ACGD (right). Visual ion images of m/z 1075.4706 ([Chinenoside I + K]⁺) and m/z 1059.4956 ([Chinenoside I + Na]⁺) (A), m/z 943.4335 ([Comp27 + K]⁺) (B), m/z 903.4939 ([Timosaponin B-II + H-H₂O]⁺) and m/z 959.4648 ([Timosaponin B-II + K]⁺) (C), m/z 1041.4867 ([Chinenoside II + Na]⁺) (D), m/z 915.4634 ([Macrostemonoside I + H-H₂O]⁺) and m/z 971.4244 ([Macrostemonoside I + K]⁺) (E), m/z 755.4229 ([Comp18 + H-H₂O]⁺) (F). Visual ion images were detected and recorded with a step size of 100 μm. The mass accuracy was no>5 ppm. See Supporting Information Table S1 for more detailed information on the compound characterization.

Export to PPT

To better reveal the pathway of sapogenin-related higher sugar moieties substitution, a series of visual ion images including several saponins and their glycosides were performed in Fig. 3. Sarsasapogenin and tigogenin were selected as the examples, and their visual [M + H]⁺ ion at m/z 417.3355 in Fig. 3F, which were rich in AMB tunic and ACGD tunic and leaf scale. Their mono-glycosides with [sapogenin + Gal/Glc + H]⁺ at m/z 579.3924 showed similar spatial distribution in both these two species (Fig. 3E). With the increase of sugar moieties, several main ions at m/z 741.4443 ([sapogenin + Gal + Glc + H]⁺) (Fig. 3D), m/z 1103.5031 ([sapogenin + Gal + 3 × Glc + K]⁺) (Fig. 3B), and m/z 1249.5832 ([sapogenin + Gal + 4 × Glc + Na]⁺) (Fig. 3A) were obviously observed in ACGD developing flower buds, while the ion at m/z 941.4544 ([sapogenin + Gal + 2 × Glc + K]⁺) (Fig. 3C) was also could be barely observed in developing flower bud of AMB bulbs.

Close

Fig. 3

MALDI MSI images of sarsasapogenin/tigogenin-related steroidal saponins in AMB (up) and ACGD (down) bulbs. Visual ion images of m/z 1249.5832 ([sapogenin + Gal + 4 × Glc + Na]⁺) (A), m/z 1103.5031 ([sapogenin + Gal + 3 × Glc + K]⁺) (B), m/z 941.4544 ([sapogenin + Gal + 2 × Glc + K]⁺) (C), m/z 741.4443 ([sapogenin + Gal + Glc + H]⁺) (D), m/z 579.3924 ([sapogenin + Gal/Glc + H]⁺) (E), m/z 417.3355 ([sarsasapogenin/tigogenin + H]⁺) (F). The scanning step size was set as 100 μm for these visual ion images. The errors of these detected ions were<5 ppm. Gal and Glc meant Galactose and Glucose, respectively. The details for compounds identification were show in Supporting Information Table S1.

Export to PPT

In addition, laxogenin and its derivatives were particular compounds in ACGD samples. Visual ion images of the adduct ion at m/z 1057.4646 ([M + K]⁺) in tunic and leaf scale indicated the compounds were chinenoside II and its isomers (Fig. 4A). Subsequently, chinenoside II lost one sugar group of 132 Da (xylose or arabinose) or 162 Da (glucose) to produce the fragment ions at m/z 925.4240 ([M + K]⁺) (Fig. 4B) and m/z 895.4054 ([M + K]⁺) (Fig. 4C), respectively. These ions were identified as chinenoside III and compound 25 (both containing their isomers), respectively. Similarly, we obtained the ion images of m/z 725.4114 ([M + H]⁺) in whole ACGD bulbs (Fig. 4D). Due to the presence of the ions at m/z 593.3693 (Fig. 4E) and m/z 431.3166 (Fig. 4F) detected in AMB, their ion images were both observed in ACGD and AMB samples. Moreover, other visual ion images of differential variables and non-differential saponins in AMB and ACGD bulbs were demonstrated in Fig. S3-S5 and Table 1. In total, the steroidal saponins in AMB were mainly distributed in bulb tunic, whereas they were mainly detected in ACGD tunic and leaf scales, and rarely in developing flower buds.

Close

Fig. 4

Visual MALDI MSI images of laxogenin-related steroidal saponins in AMB (up) and ACGD (down) samples. MALDI MSI images of m/z 1057.4646 ([sapogenin + 2 × Glc + Xyl + Ara + K]⁺) (A), m/z 925.4240 ([sapogenin + 2 × Glc + Xyl/Ara + K]⁺) (B), m/z 895.4054 ([sapogenin + Glc + Xyl + Ara + K]⁺) (C), m/z 725.4114 ([sapogenin + Glc + Xyl/Ara + H]⁺) (D), m/z 593.3693 ([sapogenin + Glc + H]⁺) (E), m/z 431.3166 ([laxogenin + H]⁺) (F). Ara, Glc and Xyl meant Arabinose, Glucose and Xylose, respectively. The visual ion images were detected with a step size of 100 μm, and the errors was no>5 ppm. The detailed MS data were displayed in Supporting Information Table S1.

Export to PPT

3.4 Comparative analysis of content levels of differential saponins in AMB and ACGD samples by UHPLC-MS/MS

The quantitative method validation results demonstrated that this method was sensitive and practical with LOD values ranging from 5.0 to 50.0 ng/mL and LOQ values between 25.0 and 200.0 ng/mL (Table S4). In addition, acceptable linear correlation with correlation coefficients values (R²) from 0.9990 to 0.9999 within the test ranges of each compound (Table S4). The relative standard deviation (RSD) values for precision, repeatability and stability were all no>3.7 % (Table S5), suggesting that the quantitative method was stable and repeatable, and the sample solution was stable at room temperature for 24 h. The recovery results were ranging from 98.1 % to 104.3 % with RSD values no>3.6 % (Table S5), indicating sufficient reliability and accuracy of the developed method.

As clear shown in Fig. S6, the major differential variables were particular in AMB or ACGD samples. For example, compounds 58, 27, 60, 18 and 12 were all rich in ACGD bulbs as well as compounds 32, 38, 31, 40, 44 and 33 in AMB samples (Fig. S6). And the content levels of these compounds were over 0.1 mg/g. In addition, several common compounds including compounds 20, 74 and 73 were all abundant in these two species (Fig. S6). Obviously, content determination results were more objective than the intensity/mass of several adduct ions derived from visual ion images for the authentication of steroidal saponins in AMB and ACGD.

3.5 Structure activity relationship of main steroidal saponins in AMB and ACGD samples

A series of steroidal saponins with different sapogenins and multiple sugar moieties substitution in AMB as well as laxogenin-related saponins in ACGD were selected for the structure activity relationship of platelet aggregation inhibitory activity (Fig. S7). As displayed in Fig. 5A and 5D, when the tested compounds were 100 μM (50 μM for sapogenins), the maximum platelet aggregation rates of compounds 66, 16, 34 in AMB and compounds 9, 12, 13, 25 in ACGD were all less than 50 % of the negative control. In addition, the inhibitory data of these compounds were fitted to log (inhibitor) and normalized response equations to obtain the IC₅₀ values (Fig. S8). The IC₅₀ values of compounds 66, 16, 34, 9, 12, 13 and 25 for the inhibitory effects of AA-induced platelet aggregation activity were 31.7, 41.6, 50.8, 35.8, 53.0, 25.1 and 40.5 μM, respectively (Table 2). Similarly, the inhibitory effects of these compounds towards ADP or collagen-induced platelet aggregation activity were also investigated (Fig. 5), their dose-independent inhibitory curves (Fig. S9 and S10) and IC₅₀ values were further obtained (Table 2). Based on the tested activity, the active “soft spots” or structure activity relationship of main steroidal saponins in AMB and ACGD were shown in Fig. 6A and 6B, respectively.

Close

Fig. 5

Evaluation of inhibitory platelet aggregation activity of major steroidal saponins in AMB and ACGD samples. The inhibitory effects of main saponins in AMB against arachidonic acid (AA, A), adenosine diphosphate (ADP, B) and collagen (C) inducing platelet aggregation. The inhibitory effects of steroidal saponins in ACGD against AA (D), ADP (E) and collagen (F) inducing platelet aggregation. The inhibitory activity was evaluated in the absence (control, 0 μM) and presence of tested compounds (50 μM or 100 μM) at room temperature. Data was represented as mean ± standard deviation of triplicate. The number of compounds was same as the information in Supporting Information Table S1. (*compared with those of control, * p < 0.05, * p < 0.01 and * p < 0.001).

Export to PPT

Close

Fig. 6

Structure activity relationship involving in inhibitory platelet aggregation activity of furostanol saponins and spirostanol saponins in AMB (A) and ACGD (B) samples.

Export to PPT