Comparative Analysis of Free Amino Acids and Nucleosides in Different Varieties of Mume Fructus Based on Simultaneous Determination and Multivariate Statistical Analyses

Mume Fructus (MF) contains a variety of organic acids, free amino acids, and nucleoside components, and studies have not yet analyzed the relationship between the components of free amino acids and nucleosides with the varieties of MF. A rapid and sensitive method was established for simultaneous determination of 21 free amino acids and 9 nucleosides in MF by ultrafast liquid chromatography-mass spectrometry. The analysis was carried out on a Waters XBridge Amide column (100 mm × 2.1 mm, 3.5 μm) with elution by the mobile phase of 0.2% aqueous formic acid (A) and 0.2% formic acid acetonitrile (B) at a flow rate of 0.2 mL/min with 1 μL per injection. The column temperature was maintained at 30°C. The target compounds were analyzed by the positive ion multiple reaction monitoring (MRM) mode. The comprehensive evaluation of the samples was carried out by principal component analysis (PCA) and technique for order preference by similarity to an ideal solution (TOPSIS) analysis. Results showed the method could simultaneously determine 30 components in MF. The content of total analytes in six mainstream varieties was different, exhibited the order Nangao > Daqingmei > Zhaoshuimei > Yanmei > Shishengme > Baimei, and aspartic acid and adenosine were the most abundant amino acid and nucleoside. PCA and OPLS-DA could easily distinguish the samples, and 11 components could be chemical markers of sample classification. TOPSIS implied that the quality of Nangao and Daqingmei was superior to the other varieties. The results could provide a reliable basis for quality evaluation and utilisation of medicinal and edible MF.


Introduction
Mume Fructus (MF) is derived from the immature fruit of Prunus mume Sieb. et Zucc., named Wumei (in China), and is commonly used food and Chinese medicines, with astringing the lung to stop cough, relieving diarrhea with astringents, removing toxicity for detumescence, promoting fluid relieving thirst, and calming Ascaris [1,2]. Modern pharmacological studies have shown that it has antibacterial, anti-Alzheimer, antiulcer, antitumor, antivirus, and other biological activities [3][4][5][6]. For its important medicinal and edible processing value, MF is widely cultivated in China [7,8], and the medicinal varieties are mostly distributed in Southwest areas of Sichuan and Yunnan provinces. In the preliminary investigation, some edible varieties were mixed as medicinal varieties, and the differences in appearance and internal components of varieties brought many problems to the identification and quality evaluation of MF [9].
Phytochemical studies have revealed that MF contains various components, including organic acids, amino acids, nucleotides, terpenoids, inorganic elements, volatile components, and polysaccharides [2,10]. Organic acids are known as bioactive constituents, and amino acids and nucleotides also play synergistic roles with organic acids in the clinical effects of MF [11,12]. Amino acids are the basic building blocks of proteins which are required for human nutrition [13,14]; L-methionine, L-threonine, L-arginine, and glycine have beneficial effects on intestinal flora; the conclusion may explain the obvious antidiarrhea effect of MF [15]. Nucleosides have important physiological functions, such as adenosine can treat angina pectoris, myocardial infarction, cerebrovascular disorders, and other diseases [16]. Scholars have used the components of amino acids and nucleosides to evaluate the quality of medicines and edible medicinal plants [17][18][19], and the results showed that differences in the composition and contents of amino acids and nucleosides could be used as biomarkers to characterize the quality of medicines and food. Until now, few studies have focused on the components of amino acids and nucleosides in MF; the methodologies of the amino acid analyzer and PITC-HPLC methods mainly focus on quantification [20,21] and have some deficiencies in resolution and separation time.
In the current study, an ultrafast liquid chromatographymass spectrometry (UFLC-MS/MS) method was developed for the determination of 21 amino acids and 9 nucleosides in MF, and then the contents in six mainstream varieties were compared and analyzed by the multivariate statistical analysis method (including principal component analysis (PCA) [22,23] and technique for order preference by similarity to an ideal solution (TOPSIS) method [24,25]). e proposed method can provide theoretical basis for the quality evaluation and comprehensive utilisation of MF and other edible medicines.

Preparation of Standard Solutions.
irty standard substances were prepared by dissolving in ultrapure water, and their concentrations (in mg/mL) were as follows: (1)

Preparation of Sample Solutions.
Accurately weighed powder (1.0 g) was extracted by ultrasonication with 20 mL of ultrapure water for 30 min and cooled at normal temperature. e same solution was used to replenish the extraction system upon solvent loss due to volatilisation. e mixture was centrifuged at 10,000 rpm for 5 min and passed through the 0.22 μm membrane before analyses.
e ESI-MS spectra were acquired in the positive ion multiple reaction monitoring (MRM) mode with the nebuliser pressure 5.5 kV; gas temperature 550°C; curtaingas pressure 241.3 kPa (35 psi); gas-1 pressure 379.2 kPa (55 psi); and gas-2 pressure 379.2 kPa (55 psi). Simultaneously, detection of ion pairs, cluster voltage (DP), and collision voltage (CE) was optimized. All MS data were analyzed by Analyst 1.6.2 (AB SCIEX). e optimized parameters for MS for the 30 target components are shown in Table 2.

Method Validation.
e proposed method was validated, including linearity, range, the limit of detection (LOD), the limit of quantitation (LOQ), precision, repeatability, stability, and recovery. Under the present chromatographic conditions, the LODs and LOQs were obtained on the response of each regression equation at signal-tonoise ratios (S/N) of 3 and 10, respectively. e intra-and interday precision were investigated by analysing the 30 analytes in six replicates within one and consecutive three days. e repeatability was ensured by preparing and analysing six independent sample solutions from sample S10. e stability of the samples was verified by analysing solution of sample S10 at 0, 2, 4, 8, 12, and 24 h. All variations were expressed by the relative standard deviation (RSD). e recovery test was performed to evaluate the accuracy of the above methods. e amount of 30 standards was added to six accurately weighed samples of S10 (1.0 g) and then extracted, processed, and quantified orderly, and the average recovery rate and RSD were calculated.

Multivariate Statistical Analyses.
Principal component analysis (PCA) is used to visualize the similarity or difference in multivariate data. It is a method of transferring multiple variables through linear transformation to select fewer important variables. SPSS23.0 was used to evaluate the variation of 30 components in MF samples. OPLS-DA supervised by SIMCA-p 14.1 was conducted with data of 30 analytes to discover different chemical compositions of each sample.
TOPSIS is used commonly in analyses of multiple-objective decisions. It ranks a limited number of evaluation objects according to their proximity to the ideal target. By calculating the best and worst indices of samples, we introduced a method to evaluate the quality of samples based on the content of 30 analytes. e data matrices of 21 amino acids and 9 nucleoside components in samples were normalized, and the corresponding matrices were established to calculate the Z ij value: where X ij represents the value of the i evaluation object on the j index and Z ij represents the value of the i evaluation object after normalization on the j index. e optimal vector (Z + ) and the worst vector (Z − ) consist of the maximum value and the minimum value of each column element. e distance from each evaluation object to Z + and Z − is calculated. D i + and D i − represent the distances of each evaluation object from Z + and Z − , respectively:  e relative closeness to the ideal object (C i ) is calculated, and C i values are compared to rank alternatives. When C i is between 0 and 1, closer to 1, the evaluation object is close to the optimal level; when it is closer to 0, the evaluation object is close to the worst level:

Sample Preparation Optimization.
In order to obtain quantitative results, the extraction method was to be optimized. e extraction method adopted ultrasound extraction with water according to [26]. And the solvent volume (10, 20, 30, and 40 mL) and extraction time (10, 20, 30, and 40 min) were examined. e results showed that the optimal extraction method was ultrasonic treatment at room temperature for 30 min with 20 mL water as the solvent according to characteristic peak intensity.

Optimization of UFLC-MS/MS Conditions.
Previous studies [16,17] reported that the positive ion modes presented higher sensitivity and intensity in examining components of amino acids and nucleosides than negative ion modes, so the ESI+ mode was adopted in the study. Most components obtained intensive product ions by optimization of the MS parameters; just adenine and hypoxanthine were too low to be detected. Chromatographic conditions were optimized. Different mobile phases (water/methanol; 0.1% aqueous formic acid/ methanol; 0.1% aqueous formic acid/acetonitrile; and 0.2% aqueous formic acid/0.2% formic acid acetonitrile) were investigated. We found that acetonitrile had better separation efficiency than that of methanol. Upon addition of 0.2% aqueous formic acid into acetonitrile and water, the shape of the chromatographic peak and degree of separation were better, and the tail of the chromatographic peak was reduced. Furthermore, a flow rate of 0.1-0.5 mL/min, injection volume of 1-5 μL, and column temperature of 30-40°C were studied to obtain rapid and reliable separation. As a result, the analysis was carried out on a Waters XBridge Amide column (100 mm × 2.1 mm, 3.5 μm) with elution by mobile phase of 0.2% aqueous formic acid (A) and 0.2% formic acid acetonitrile (B) at a flow rate of 0.2 mL/min with 1 μL per injection. e column temperature was maintained at 30°C, respectively.

Method Validation.
Validations of the method are shown in Table 3. All calibration curves showed good linearity regressions (R 2 > 0.9990) within the determination range.
e LODs and LOQs of the 30 compounds were estimated to be in the range of 0.14-5.00 ng/mL and 0.18-16.67 ng/mL, respectively. ey should provide In order to highlight the sensitivity of the method, the LODs of each target compound were compared with other techniques: MLC-HSLC [27], HILIC-UHPLC-QTRAP/MS [28], and UPLC-MS/MS [16], and the results of comparison are listed in Table S1. e LOD range of our method is 0.14-5.00 ng/mL, and the LOD range of other methods is 250-410 ng/mL, 0.45-53.5 ng/mL, and 0.17-61.2 ng/mL, respectively. e LOD range of nucleoside components was 0.16-3.37 ng/mL, although few nucleoside components in other methods were available for effective comparison, but overall, this method was more sensitive than other methods, especially in arginine, alanine, and asparagine. e previous studies used the amino acid analyzer and PITC-HPLC methods [20,21] in MF. However, both methods showed some disadvantages in resolution and analysis time, required over 90 minutes. Our solution time was 11 min included balanced column conditions, which means this analytical method should take shorter time than other methods in MF. Table S2. e data from all the samples elucidated that the total contents of amino acids and nucleosides were significantly different between varieties of MF in the order Nangao (123800.00 μg/g) > Daqingmei (8330.99 μg/g) > Zhaoshuimei (8081.55 μg/g) > Yanmei (6870.08 μg/g) > Shishengmei (4622.85 μg/g) > Baimei (2599.75 μg/g).

Quantitative Analyses. Contents of the 30 components in samples were determined using UFLC-MS/MS. Quantitative determination is shown in
In terms of individual components, aspartic acid is the most abundant amino acid in MF, which is consistent with the previous results [2]. In all the samples, the average of the aspartic acid content was 2599.41 μg/g followed by L-asparagine, c-aminobutyric acid, L-valine, and L-histidine; thymidine was the lowest (0.54 μg/g). Adenosine had the highest content among the nucleoside components followed by uridine uracil, and adenine. Interestingly, the content of aspartic acid in the samples from Nangao was significantly higher than the other varieties. Aspartic acid can regulate the S12 S11 S8 S7 S9 S14 S1 S2 S15 S10 S4 S13 S3 S5 S6 S17 S16 S18 S12 S11 S8 S7 S9 S14 S1 S2 S15 S10 S4 S13 S3 S5 S6 S17 S16 S18 (b) S12 S11 S8 S7 S9 S14 S1 S2 S15 S10 S4 S13 S3 S5 S6 S17 S16 S18 (c)  International Journal of Analytical Chemistry metabolism of the brain and nerves and is often used to treat heart disease, liver dysfunction, and hypertension [13,14]. Adenosine, as an antiarrhythmic drug, can treat angina pectoris, myocardial infarction, cerebrovascular disorders, etc. erefore, we speculated that the clinical, medicinal, and edible value of Nangao was higher than the other varieties.

Multivariate Statistical Analyses of Samples.
Analyzed data for 30 compositions of each sample are shown in Figure 2. After the original data were standardized, PCA was performed with SPSS23.0 to obtain the eigenvalues and variance contributions of principle components (PCs) [29,30]. e results are given in Table S3. e eigenvalues of the first seven PCs were greater than 1, the cumulative contributions reached 89.04%, and the first three PCs (35.89% for PC1, 20.88% for PC2, and 11.68% for PC3) simplified the multidimensional dataset to a 3D dataset; the remaining PCs were discarded for the less effect. PC1 had good correlation with L-aspartic acid, L-asparagine, L-histidine, and L-proline. PC2 showed good correlation with Lvaline, adenine, and hypoxanthine, and PC3 showed good correlation with glycine (Table S4). e results indicated that the aforementioned components might contribute to the classification of samples. According to the scatter plots (Figure 2(a)), PCA results showed the differences between samples but could not provide a clear classification for the samples. PLS-DA and OPLS-DA were used to extend to discover the factors leading to different varieties [31,32]; two methods were selected for this study. e PLS-DA results showed R2Y was 0.883, indicating that the model fitted well, but the model according to cross validation Q2 = 0.44 < 0.5 indicated that the model prediction result was not ideal; the score plot is shown in Figure 2(b). e OPLS-DA results showed R2Y was 0.878, indicating that the model fitted well, and Q2 = 0.525 > 0.5 indicated that the model predicted well. In the OPLS-DA score plot (Figure 2(c)), the samples were clearly divided into six groups; each group corresponded to the varieties of samples. In the VIP loading diagram of OPLS-DA (Figure 2(d)), the large load (VIP > 1) could be regarded as a marker component that contributed greatly to the classification of samples; L-isoleucine, cytidine, L-cystine, L-methionine, glycine, hypoxanthine, L-tyrosine, adenine, L-arginine, L-alanine, and L-threonine could distinguish different varieties of MF samples. ese results indicated that amino acids and nucleosides might be regarded as chemical markers for MF classification.
Samples were analyzed comprehensively by TOPSIS, and ranking results were obtained. e data matrix of 30 components in six varieties was normalized, and the corresponding standardized matrix was established. Z ij value was calculated according to formula (1), and the results are shown in Table S5. e optimal vector (Z + ) and the worst vector (Z − ) were obtained according to the normalized samples:    , D i − , and C i were calculated according to formulas (2) and (3), respectively, and the results are shown in Table 4. e evaluation objects were ranked according to C i . e larger the value of C i , the better the quality. We noted significant differences in the content between different varieties, the results above demonstrated that Nangao and Daqingmei were better qualities than other varieties of MF.

Conclusion
A rapid and sensitive UFLC-MS/MS method was established and employed to determine the components of 21 amino acids and 9 nucleosides in different varieties of MF. According to the analytical results, MF had rich amino acids and nucleosides, and aspartic acid was the most abundant amino acid in MF followed by L-asparagine, c-aminobutyric acid, L-valine, and L-histidine. Adenosine had the highest content among nucleoside components followed by uridine, uracil, and adenine. Six varieties were well classified by PCA and OPLS-DA; 11 amino acids and nucleoside components could be used as chemical markers to classify different varieties of MF. As medicinal and edible varieties, the content and quality of Nangao and Daqingmei were higher than other varieties; the obvious difference should be caused by the artificial breeding of varieties. e present study could provide theoretical basis for further evaluation of the internal quality and control of medicinal and edible MF.

Data Availability
e data used to support the findings of this study are available from the corresponding author upon request.

Conflicts of Interest
All authors declare that they have no conflicts of interest.