Metabolomics Reveals Distinct Metabolites between Lonicera japonica and Lonicera macranthoides Based on GC-MS

Lonicera Thunb. (LJ) and Lonicera macranthoides Hand. -Mazz. (LM) have been widely used in Chinese medicine for thousands of years. Although the morphological characteristics of LJ and LM are quite similar, there are signiﬁcant distinctions of medicinal ingredients (mainly the secondary metabolites) and clinical indications between them. However, the in-depth dif-ferences of primary metabolites have not thoroughly been studied yet. Therefore, gas chromatography-mass spectrometry- (GC-MS-) based metabolomics method combined with chemometric methods were performed to analyze the distinction in this study. The results showed that LJ and LM were obviously classiﬁed into two groups. 10 metabolites were obtained as biomarkers on account of their p values, | p ( corr )| values, and diﬀering variable importance in projection (VIP) values. Metabolic pathway analysis showed that the galactose metabolism and starch and sucrose metabolism gathered as potential pathways caused these extraordinary diﬀerences of primary metabolites between LJ and LM. Further, we found that the diﬀerences of main medicinal ingredients between LJ and LM could be interpreted from these metabolites according to the analysis of mainly related pathways. The metabolites involved in the starch and sucrose metabolism presented upregulated in LJ, while almost all metabolites in the galactose metabolism, the TCA cycle, and the phenolic acid part of phenylpropanoid metabolism were downregulated in LJ. Therefore, the energy stored in the starch and sucrose metabolism may be saved to produce ﬂavonoid, which could be the reason that the level of ﬂavonoid of phenylpropanoid metabolism is higher in LJ compared to LM. Consequently, this study presented an eﬀective tool for quality evaluation of LJ and LM and laid a foundation for further studies of the metabolic mechanisms and high-quality manufacturing of them.


Introduction
Lonicera japonica Flos, the flower bud of Lonicera japonica unb. (LJ), has been widely used in Chinese medicine and food for thousands of years [1]. It is mainly cultivated in Shangdong and Henan provinces of China, and it is widely applied for the prevention and treatment of H1N1 influenza, severe acute respiratory syndromes, exopathogenic windheat, and hand-foot-mouth diseases [2,3]. erefore, LJ is a Chinese medicine with great demand. Moreover, it is frequently used as important raw material to produce Chinese patent drugs and functional food additives, such as Yinqiao Jiedu/Shuanghuanglian tablets, Tanreqing/Reduning injection, and Heqizheng/Qingchunbao herb teas [4,5]. However, some flower buds of relevant species are confused or treated as LJ sometimes due to their similar appearance [6]. Lonicera macranthoides Hand. -Mazz. (LM), mostly cultivated in Hunan and Guangxi provinces of China, is one of them. ough the morphological characteristics of LJ and LM are pretty similar, there are significant differences of medicinal ingredients (mainly the secondary metabolites) and clinical indications between them [7]. e main chemical constituents of LJ and LM include organic acids, flavonoids, triterpenoidal saponins, iridoids, volatile oils, and trace elements. However, the chemical composition and content are different between them. Chlorogenic acid is the main active organic acid shared by LJ and LM, but LM has a high level of chlorogenic acid, which is almost two times higher than that in LJ [7]. e flavonoids, including rutin, luteolin, and luteolin-7-O-glucoside, are almost undetectable in LM, but are abundant in LJ [4]. Triterpenoidal saponins are also the components contributed to the distinction between LJ and LM. LM contains more kinds of triterpenoidal saponins which proved to have strong hemolytic activities compared to LJ [8,9]. Besides, both LJ and LM contain Fe, Mn, Zn, Cu, Ni, and other trace elements. LJ contains a wealth of Fe and Ni compared to LM, while LM has a higher content of Mn [7]. erefore, there are significant differences of major effective compounds between LJ and LM. In accordance with the principle of "gradual classification of multisource varieties of Chinese medicinal materials with significant component differences," LM used to list in Lonicerae Flos was separated from Lonicerae japonicae Flos after the 2005 edition of the Chinese pharmacopoeia [10]. Hence, it is necessary to study the in-depth differences between them and the reason behind these remarkable distinctions, which will be helpful for discrimination, quality evaluation, and high-quality manufacturing.
Numerous experiments correlated with the distinctions of LJ and LM were established during the past two decades, which were mainly focused on the secondary metabolites [11][12][13]. However, metabolite profiles can be classified into two categories: primary and secondary metabolites [14]. Primary metabolites were directly involved in the growth, development, and reproduction of a plant, and it also played an important role in spanning the interface between primary and secondary metabolism [15,16]. e differences in primary metabolites between LJ and LM may be the basis that caused the dissimilitude of medicine quality. erefore, it is indispensable to analyze the distinction of primary metabolites, as well as the contact between primary and secondary metabolite profiles.
Metabolomics, an emerging tool used to analyze metabolite networks and regulatory mechanisms [17], has been proved to be an effective and meaningful method in medical analysis, quality evaluation of the medicine, and other analytical fields [18]. e main analytical platforms for plant metabolomics research are nuclear magnetic resonance (NMR) and methods based on mass spectrometry (MS), including GC-MS (gas chromatography-mass spectrometry), LC-MS (liquid chromatography-mass spectrometry), and CE-MS (capillary electrophoresis-mass spectrometry) [19][20][21]. GC-MS is one of the most suitable and versatile methods in metabolomics, and it is widely applied on account of its repeatability, wide measurement range, and direct peak identification [22]. erefore, the GC-MS method combined with chemometric methods was performed to analyze metabolite profiles of LJ and LM in this study. Moreover, the relationship and mechanism between primary and secondary metabolic pathways were also explored according to the metabolic net analysis.

Plant Materials and Reagents.
Fresh bud samples of LJ and LM were collected from the main habitats of the original plants in China (Shandong and Hunan Province, respectively). Two voucher specimens (X20180421 for LJ and X20180516 for LM) were deposited in the Key Laboratory for Quality Evaluation of Bulk Herbs of Hunan Province, China. After samples were identified by Prof. Li-min Gong at the Hunan University of Chinese Medicine, Hunan, China, they were immediately frozen in liquid nitrogen and stored at −80°C to inhibit the decline of enzyme activity before sample preparation. In brief, bud samples were divided into two groups (S LJ and S LM ), each group consisted of thirteen biological replicates. e standard substances (ribitol), pyridine, methoxyamine hydrochloride, and N,O-Bis(trimethylsilyl)trifluoroacetamide (BSTFA) were provided by Sigma-Aldrich Trading Co., Ltd. (St. Louis, Missouri, USA). Methanol and all the other reagents were of analytical grade and purchased from Sinopharm Chemical Reagent Co., Ltd.

Sample Preparation.
e freeze-dried samples were ground into uniform powder and stored at −80°C until the metabolic analysis [23]. A quality control (QC) sample was acquired by a representative average of samples pooled from different samples. e extraction procedure was in line with the previous research with a little modification [23]. 100 mg samples were transferred to a 10 mL centrifuge tube, and 1400 μL chilled methanol (precooled at −20°C) was added, reacting at 70°C for 30 min. After sufficient cooling, the mixture was centrifuged for 5 min (13000 rpm/min). 400 μL supernatant and 50 μL ribitol (1 mg/mL) as an interior label was placed into a 1.5 mL centrifuge tube and shaken for approximately 1 min with a vortex mixer (Wiggen Hauser, Malaysia). After adequate evaporation under N 2 , the dry extracts were derivatized by 50 μL of methoxyamine pyridine solution (20 mg/mL) for 1 h at 30°C. en, 100 μL BSTFA was added and the following reaction was carried out at a temperature of 45°C for 2 h. Finally, the solution was transferred into a vial to be analyzed. In addition, the QC samples were tested at the beginning, end, and randomly throughout the analytical run.

GC-MS Method.
e extract (1 μL) was injected into GCMS-QP2010 (Shimadzu, Japan) equipped with HP-5 MS capillary column (5% phenyl methyl silox: 30 m × 250 μm, 0.25 μm; Agilent J&W Scientific) for primary metabolite profiling. Injection temperature was set at 280°C and the split ratio was 25 : 1. Helium was the carrier gas delivered at a constant flow rate of 1 mL/min. e temperature of the ion source and transmission line was 230°C and 280°C, respectively. e initial GC temperature was set at 70°C for 2 minutes, raised to 185°C at a rate of 10°C/min and maintained for 3 min. en, the temperature was increased to 220°C by 3°C/min and held constant for 3 min. Finally, the temperature was elevated to 280°C with a ratio of 10°C/min and lasted for 10 min. e mass spectrometer was programmed under electron impact (EI) in a total ion chromatography mode (m/z 35-550) and electron ionization was applied at 70 eV.

Data Preprocessing.
Peaks were picked up with the signal to noise ratio (S/N) >6. Raw data of GC-MS were transformed into Net CDF format by the Shimadzu postrun workstation and then processed by XCMS software to detect metabolic features and align all metabolite peaks. Metabolites with a similarity of more than 85% were identified and screened by the Automated Mass Spectral Deconvolution and Identification Software (AMDIS) combined with the National Institute of Standards and Technology (NIST) 14 database. Finally, all peak areas of identified metabolites were normalized to the internal standard.

Statistical Analysis.
Heat-map analysis and clustering analysis were also performed to visualize the overall difference of metabolite profiles in LJ and LM. en, multivariate statistical analysis including PCA and OPLS-DA were applied to detect the distinction of metabolite profiles between LJ and LM using SIMCA-P 15.0 (Umetrics AB, Umea, Sweden) and the MetaboAnalyst 4.0 (http://www. metaboanalyst.ca/). Significant metabolites as biomarkers were evaluated based on their p value, |p(corr)| value, and differing variable importance in projection (VIP) values calculated with OPLS-DA. e related pathways of these biomarkers were interpreted and visualized according to the Kyoto Encyclopedia of Genes and Genomes (KEGG) database (https://www.kegg.jp/), a free and web-based tool. Consequently, a metabolite net was constructed to visualize the connections between the various pathways.

Metabolite Profiles of LJ and LM Samples.
Raw data with retention time, mass-to-charge ratio (m/z), and peak intensity were acquired by GC-MS. According to the NIST database, 63 metabolites were identified and matched with a similarity of more than 85% listed in Table 1. ese metabolites could be classified into several categories, including sugar, polyol, fatty acid, organic acid, and amino acid, and others accounted for 26%, 19%, 5%, 25%, 11%, and 14% of all identified metabolites, respectively (supplementary materials Figure S1). e main components of the metabolites were sugar (26%), organic acid (25%), and polyol (19%). Sugar as a basic compound played a key role in plant growth, adaptation to the environment, and adjustment to nutrient availability [24]. Organic acid with various and powerful pharmacological activities was the major active component in LJ and LM [2]. erefore, the high percentage of these two kinds of metabolites was understandable. e characteristic GC-MS total ion current chromatograms showed visually differences between LJ and LM encouraging further explorations (Figure 1(a)). In order to make the overall discrepancies in the metabolite profiles of LJ and LM more external, heat map analysis was performed. As shown in Figure 1(b), a clear segregation, once again observed through heat map analysis and cluster analysis, further illustrated the prominent distinctions between LJ and LM.

PCA Analysis and QC Evaluation.
PCA is an unsupervised method compressing multidimensional data into several major components and displaying the internal structure of datasets in an unprejudiced manner [25]. PCA with unit variance (UV) scaling was performed to evaluate QC samples and generate an overview of the metabolic profiles. As shown in Figure 2(a), there were no outlier samples and the QC samples were clustered together near the center of the score plot, which illustrated the stability of the analytical apparatus and the reproducibility of pretreatment. e score plot showed a promising separation, suggesting the primary metabolites in LJ were different from that in LM.

OPLS-DA Analysis and Identification of Biomarkers.
OPLS-DA is a supervised technique built from both predictive and orthogonal components to assess the maximum class separations [26]. In order to highlight the discrepancies between LJ and LM, OPLS-DA was consequently utilized to locate the radically different metabolites. As indicated in Figure 2(b), the metabolites of LJ and LM were distributed on two sides obviously. e variation in X (R 2 Xcum)c was 0.706 and the variation in Y (R 2 Ycum) was 0.943, predicting 90.4% of the variation in response to Y (Q 2 cum � 0.904). e value of these parameters illustrated that the model had high explanatory and predictive abilities. e permutation test with a permutation number of 200 times was also performed to validate the model (Figure 2(c)). e result indicated that the intercepts of R 2 and Q 2 were 0.377 and −0.765, respectively. Furthermore, all R 2 and Q 2 values were lower than the original ones, suggesting a high degree of reliability.
Moreover, the S-plot was applied to identify the chief metabolites that contributed highly to the distinctions among samples. As detailed in Figure 2(d), the dots at the edges of the plots translated a bigger difference than those shown near the center. e VIP value ≥1.0, |p(corr)| > 0.3, and p < 0.05 were regarded as the screening criteria. As a result, 10 spots were filtered as potential biomarkers listed in Table 2, including the p value, the VIP value, and the most related metabolic pathway of these biomarkers. Starch and sucrose metabolism, galactose metabolism, and fructose and mannose metabolism were the major pathways involved in these biomarkers. In view of the complexity and association of these metabolisms, the analysis still needs to go further.

Correlation Network Analysis of Potential Biomarkers and Biosynthesis of Main Active Components in LM and LJ.
e 10 compounds screened by the previous OPLS-DA analysis were further confirmed and analyzed through the KEGG metabolic database and MetaboAnalyst. As shown in Figure 3, the y-axis represented the p value, and the x-axis     represented the pathway impact values. e color and size of each circle graph had a positive relationship with their p value and pathway impact value, respectively. In consequence, the galactose metabolism and the starch and sucrose metabolism gathered as potential pathways caused the remarkable distinctions between LJ and LM. However, the mechanism of how these pathways contributed to the distinction between LJ and LM was still unclear. erefore, a network of metabolic pathways was constructed to show the changes in the key metabolic pathway based on these information and related literature [27][28][29]. As displayed in Figure 4, sucrose and trehalose, involved in the starch and sucrose metabolism, were at a higher level in LJ than that of LM. While galactose, mannose, maltose, and glucitol, derived from the hydrolysis of sucrose and involved in the galactose metabolism, were more abundant in LM than those in LJ. e two metabolisms presented absolutely different activities. Citric acid, malic acid, and succinic acid in the TCA cycle showed a distinct metabolism pattern. All of them showed obvious upaccumulation in LM compared with LJ. Caffeic acid and chlorogenic acid in the phenylpropanoid pathway were obviously upregulated in LM as well. However, the quinic acid, which is also involved in the phenylpropanoid pathway, presented downregulation in LM. All in all, metabolites in the starch and sucrose metabolism, galactose metabolism, TCA cycle, and the phenylpropanoid pathway showed significant distinctions and different metabolism patterns between LJ and LM.

Discussion
According to OPLS-DA analysis and MetaboAnalyst, the starch and sucrose metabolism and the galactose metabolism were obtained as potential pathways contributed to the remarkable distinctions of primary metabolites between LJ and LM.
In the starch and sucrose metabolism, sucrose and trehalose showed higher content in LJ compared to LM (Figure 4). e metabolites involved in the starch and sucrose metabolism showed upregulation in LJ. Starch and sucrose are the primary products of photosynthesis of most plants [30]. Starch represents the major plant storage carbohydrate providing energy during the times of heterotrophic growth [31]. Sucrose is a nonreducing sugar playing an eminent role in sugar translocation in the phloem, sugar storage, the syntheses of cell wall polysaccharides [32], and the tolerance to various stresses [33,34]. Trehalose, synthesized via trehalose 6-phosphate, is essential for plant metabolism, growth, and stress responses. It is implicated in responses to cold, salinity, and in regulation of stomatal conductance and water use efficiency [35]. e higher content of sucrose and trehalose in LJ compared to LM revealed that LJ may perform better in stress responses, development, growth, and other aspects. In the galactose metabolism, most metabolites (galactose, mannose, glucitol, and ribose) were downregulated in LJ. Galactose can be obtained from the raffinose pathway and the breakdown of glucose ( Figure 4). Some studies presented that galactose could inhibit cell wall biosynthesis in plants [36] and decrease the level of sucrose [37]. Mannose can affect gas exchange and decrease the photosynthetic capacity in plants [38]. Hence, the lower level of galactose and mannose in LJ compared to LM is understandable.
ese metabolites could be divided into two categories, phenolic acids and flavonoids.
Caffeic acid, quinic acid, and chlorogenic acid, as secondary metabolites, are the main functional phenolic acids in LJ and LM. Caffeic acid and chlorogenic acid presented lower levels in LJ compared with LM ( Figure 4). Chlorogenic acid has been regarded as one of the most significant constitutes of LJ and LM for its multiple pharmacological activities, such as antibacterial, antioxidant, and antidiabetic [2,42,43]. Moreover, chlorogenic acid is viewed as a significant factor to evaluate the quality of LM and LJ [1]. Caffeic acid synthesized by p-coumaric acid has strong antioxidative and anti-inflammatory activities [44]. Quinic acid is a powerful antioxidant and has been reported to have antiviral and antihuman immunodeficiency virus activities [45,46]. In this study, an interesting phenomenon was found. Caffeic acid and chlorogenic acid presented lower levels in LJ compared with LM, while quinic acid showed an opposite trend (Figure 4). e level of these components in LJ and LM was similar to previous studies [12,47]. Such a case also occurred to different subtypes of white tea [48], which is interesting and thought-provoking. In fact, the level of phenolic acids can be influenced by many factors. Previous research showed that gene expressions [49,50], different breeding processes, producing process, and manufacture process all had significant impacts to the level of chlorogenic acid and other phenolic acids [51,52]. erefore, the reason for such case requires more specific studies.
Flavonoid is another category of secondary metabolites in the phenylpropanoid pathway. ese compounds are demonstrated to exhibit significant pharmacological activities and understood to be influential targets for metabolic engineering [53][54][55][56][57]. Naringenin generated from the   Figure 4: e metabolic pathway map, including the galactose metabolism, the starch and sucrose metabolism, the glycolysis, the phenylpropanoid pathway, the flavonoids metabolism, and the TCA cycle. In the relative peak area figure, red represents LJ and green represents LM. conversion of p-coumaroyl-CoA is the precursor of most flavonoids [58]. It can be catalyzed to apigenin and eriodictyol through flavones synthase or flavonoid 3-hydroxylase [29]. en, luteolin can be synthesized from these two flavones. Luteolin, a flavonoid mainly found in fruits and vegetables, is regarded as a representative metabolite in LJ. It has antioxidant [59], anti-inflammatory [60], antitumor [61], and anti-5-lipoxygenase [62] activities. Moreover, luteoloside can be further synthesized from luteolin via UDP-glucose flavone 7-O-β-glucosyltransferase [28]. Luteolin-7-O-glucoside was viewed as the standard compound to evaluate the quality of LJ [1]. Previous studies showed that flavonoids had a higher level in LJ compared with LM [7,63,64], but a few studies about the inner causes of it were established. Liu et al. revealed that miRNAs play a key role in the flavonoid biosynthesis of LJ [53]. Wu et al. found that the weak catalytic activity and low expression of LmFNSII-1.1 in flowers might be responsible for the low levels of flavone accumulation in LM [4]. In this study, we found that the high level of flavonoids in LJ can be interpreted from the primary metabolite aspect. e metabolites involved in the starch and sucrose metabolism exhibited upregulation in LJ ( Figure 4). ese substances were then broken down and altered through various enzymes and different metabolic pathways. However, almost all metabolites in the galactose metabolism, the TCA cycle, and the phenolic acid part of phenylpropanoid metabolism were downregulated in LJ. erefore, the high level of flavonoid in LJ could be explained by the factor that the energy stored in the starch and sucrose metabolism may be saved to produce flavonoid.

Conclusions
In summary, the combination of metabolomics based on GC-MS with chemometric analysis could be a useful tool for discrimination, quality evaluation, and exploring metabolic regulation mechanisms of LJ and LM. e result suggested great variations in the metabolome between LJ and LM, and 10 metabolites (mannose, sucrose, D-(+)-galactose, D-fructose, psicose, glycerol, chlorogenic acid, quinic acid, D-(+)-glucose, and D-(+)-trehalose) were obtained as characteristic ingredients contributed highly to the difference. e starch and sucrose metabolism and the galactose metabolism were obtained as potential pathways contributed to the remarkable distinctions of primary metabolites between LJ and LM. Moreover, the metabolites involved in the starch and sucrose metabolism presented upregulation in LJ. While almost all metabolites in the galactose metabolism, the TCA cycle, and the phenolic acid part of phenylpropanoid metabolism were downregulated in LJ, the high level of flavonoid in LJ could be explained that the energy stored in the starch and sucrose metabolism may be saved to produce flavonoid.

Data Availability
e data used to support the findings of this study are included within the article.

Ethical Approval
is article does not contain any studies with human participants or animals performed by any of the authors. e authors confirm compliance with ethical standards.