Identification of nutritional components in unripe and ripe Docynia delavayi (Franch.) Schneid fruit by widely targeted metabolomics

Sample collection

On July 18, 2020, the samples were collected from Lancang country (100°21′E, 22°56′N) in Yunnan province, China. Field experiments were approved by the National Natural Science Foundation of China (project number: 32060350) and the samples were collected with the consent of Liu Yu and Zhang Xinluo villagers; Nuozadu village committee for supporting this study. Immediately afterwards, the harvested fruits were transported to a laboratory, where they were graded by maturity and colouring stage: green unripe (GW) and red ripe (GX). (Su, 2019). Approximately 15 fruits in each stage were combined to represent one biological replicate, with three biological replicates prepared for each stage. All samples were frozen with dry ice.

Sample preparation and extraction

These biological samples were freeze-dried by using a vacuum freeze-dryer (Scientz-100F). Then, the freeze-dried samples were crushed for 1.5 min at 30 Hz by a mixer mill (MM 400, Retsch GmbH, Haan, Germany) fitted with a zirconia bead. 100 mg of lyophilized powder was dissolved in 1.2 mL of 70% methanol solution, and then stirred for 30 s at a 30-minute interval (6 times in total). Finally, the samples were placed in a refrigerator at 4 °C overnight. After 10 min of centrifugation at 12,000 rpm, the extracts were filtrated (SCAA-104, 0.22 µm pore size; ANPEL, Shanghai, China) for subsequent UPLC-MS/MS analysis.

UPLC-MS/MS conditions

Given the limited availability of extracts, the application of ultra-performance liquid chromatography coupled to tandem mass spectrometry is considered to be one of the preferred methods of analysis. The following UPLC conditions are required for the present study (Chen et al., 2013):

1. Column, Agilent SB-C18 (1.8 µm, 2.1 mm*100 mm);

2. The mobile phase included solvent A:

   1. MilliQ water with 0.1% formic acid, and solvent B:

   2. Acetonitrile with 0.1% formic acid.

ESI-Q TRAP-MS/MS

An AB4500 Q TRAP UPLC/MS/MS system, complete with an ESI Turbo Ion-Spray interface, was adopted for both LIT and Triple Quadrupole (QQQ) scanning. This system was operated by using Analyst 1.6.3 software (AB Sciex). The operational settings for the ESI source are as follows: ion source, turbo spray; source frequency of 550 °C; ion spray voltage (IS) 5,500 V (positive ion mode)/−4,500 V (negative ion mode); ion source gases I (GSI), II (GSII), and curtain gas (CUR) fixed at 50, 60, and 25.0 psi, respectively; and large collision-activated dissociation (CAD) value. To tune the device and calibrate the weight, polypropylene glycol liquids of 10 and 100 mol/L were used, respectively, in the QQQ and LIT modes. During MRM investigation, QQQ scans were obtained with the collision gas (nitrogen) adjusted to medium. Both DP and CE were further enhanced to support single MRM conversions. The MRM transitions for each session were tracked by eluting the metabolites during each interval (Fig. 1) (Fraga et al., 2010).

Metabolite identification and quantification

MetWare databases were used. Match scores were given by using the MetDNA software developed by Zhu et al. (2013) and the Masterview software developed by AB Sciex (Toronto, Canada). The majority of the compounds in the database are standards. MRM information was collected from each standard (Fig. 1). For metabolite quantification, multiple reaction monitoring was performed. In the MRM mode, the quadrupole filtered the precursor ions of the target substance first, excluding the ions corresponding to other molecular weight substances for the preliminary elimination of interference. After the collision chamber was induced to ionize, the precursor ions were fragmented to obtain many fragment ions. Then, triple quadrupole filtering was conducted, with a distinctive fragment ion used to remove non-target ion interference for improving the accuracy of quantification and reproducibility. The peak area of all substance mass peaks was integrated after the data on metabolite mass spectrometry was collected from several samples. Besides, the peaks of the same metabolite in various samples were integrated and corrected.

Metabolite data analysis

Principal Component Analysis (PCA) was carried out with the statistically functional prcomp in R (http://www.r-project.org). The data was unit variance scaled before unsupervised PCA. The heatmaps of dendrograms were used to present the sample and the hierarchical clustering results of metabolites. The correlation function in R can be used to calculate the Pearson correlation coefficients between samples, which are displayed in the form of heatmaps. The differential metabolites designated among groups were identified by VIP >= 1 and absolute Log₂FC (fold change) >= 1. Obtained from OPLS-DA outcome, VIP values include two plots of score and permutation as formed with R package Metabo Analyst R. The statistic was log transfer (log2) and mean centering before OPLS-DA. A permutation test (200 permutations) was performed to stop overfitting. Pathway database (http://www.kegg.jp/kegg/pathway.html) was used to make explanation using the KEGG Compound database (http://www.kegg.jp/kegg/compound/). Then, metabolite sets enrichment analysis was conducted, and the significance of pathways with heavily regulated metabolites was determined according to the p-values of the hypergeometric test.

Statistical analysis

Through Excel (Microsoft Excel 2016, Microsoft, Redmond, WA, USA), a statistical analysis was carried out. The statistics are presented as means ± standard deviations. The least significant difference (p < 0.05) was used to assess the level of statistical significance.

Widely targeted metabolome profiling of D. delavayi fruit

To better understand the variation in nutritional values between unripe D. delavayi and ripe D. delavayi, ‘GW’ and ’GX’ were applied for the metabolic profiling based on UPLCMS/MS. With a total of 727 distinct metabolites identified in all samples, they were classified into 11 categories according to metabolite structure, as shown in Table 1. A significant number of these metabolites are suspected to be responsible for fruit development (53 amino acids and derivatives, 72 phenolic acids, 12 tannins, 29 organic acids). Among these 11 compounds, phenolic acids (72 types) and flavonoids (111 types) were considerably more abundant than the others.

Multivariate analyses of determined metabolites

In total, 477 metabolites were analyzed through PCA analysis (S1). The first principal component (PC1) accounted for 65.63% of the variance, whereas the second principal component (PC2) accounted for 26.26% of the variance (Fig. 2A), indicating a significant difference between GW and GX. The heatmap of all metabolites was created to visualize the variations between GW and GX (Fig. 2B). PCA and the heatmap reveal the significant variations in metabolites between the two developmental stages.

By identifying the differential metabolites between GX and GW, it can be found out that metabolites with fold change ≥2 or ≤0.5 and the VIP values greater than 1. There are a total of 130 differentiated metabolites, of which 94 are up-regulated and 36 are down-regulated (Fig. 3A). The 130 metabolites can be divided into 11 different categories, with phenolic acids, flavonoids, terpenoids, and lipids in the majority (Fig. 3B).

Differential metabolite KEGG classification and enrichment analysis

The metabolic pathways of 130 differential metabolites were mapped to the KEGG database (S2). Despite few metabolites classified as genetic information processing and environmental information processing, the majority of metabolites were categorized into ‘metabolism’ (Fig. 4A). As revealed by the enrichment analysis of biological pathways associated with two nutrient metabolisms (“pantothenate and CoA biosynthesis” and “flavone and flavonol biosynthesis”), there were significant differences between unripe and ripe D. delavayi: p < 0.05 (Fig. 4B).

Amino acids and derivatives, phenolic acids, organic acids

Concerning the types of metabolites that may affect the flavour of D. delavayi, both ‘GW’ and ‘GX’ contained amino acids and their derivatives, with L-glutamic acid and L-phenylalanine in the majority. According to fold change and VIP values, seven differential metabolites were selected. The concentrations of acetyltryptophan and n-acetyl-dl-tryptophan were significantly higher in ‘GX’, whereas the content of other five amino acids in ‘GW’ was lower than in ‘GX’ (Fig. 5), despite no significant change (Table 2).

A total of 72 phenolic acids were identified in ‘GW’ and ‘GX’ (Table 1), of which 12 exhibited significant differences in accumulation. 1-O-(3,4-dihydroxy-5-methoxy-benzoyl)-β-D-glucoside accumulated in significantly higher concentrations in ‘GX’ than in ‘GW’ (Fig. 5). However, only one of the 29 organic acids was identified as a differential metabolite (phosphoenolpyruvic acid), with ‘GX’ showing a considerably higher cumulative concentration of phosphoenolpyruvic acid than ‘GW’ (Table 2).

Flavonoids and terpenoids

What is noteworthy is the primary source of nutrients in D. delavayi, in addition to the metabolites. Flavonoids were found to have the largest number of unique metabolites. In GW and GX, only four flavonoids were down-regulated, while the remaining 47 flavonoids were up-regulated to varying degrees (Table 3). The three flavonoids that were up-regulated to the most significant extent include 5,7,8,4′-tetramethoxyflavone (348.620 fold), nobiletin (226.216 fold), and tangeretin (204.974 fold) (Fig. 5). These three differential metabolites were all associated with nobiletin and tangeretin and were detected to have higher concentrations in ‘GX’ than in ‘GW. ‘GX’ contained a substantially larger amount of flavonoids than GW.

Terpenoids are also the important metabolites contained in D. delavayi fruit, as revealed in this study. A majority of terpenoids found to be clearly different between ‘GW’ and ‘GX’ were triterpenes (Table 4). In GW, only one triterpene (3-o-trans-feruloyl euscaphic acid) was down-regulated, while the others were up-regulated. The three triterpenes that were up-regulated most significantly include rutundic acid (6.539 fold), myrianthic acid (6.245 fold) and euscaphic acid (4.597 fold) (Fig. 5). These findings indicate an increase in flavonoids and terpenoids.