Simultaneous Qualitative Assessment and Quantitative Analysis of Metabolites (Phenolics, Nucleosides and Amino Acids) from the Roots of Fresh Gastrodia elata Using UPLC-ESI-Triple Quadrupole Ion MS and ESI- Linear Ion Trap High-Resolution MS

A sensitive, effective and optimized method, based on ultra performance liquid chromatography (UPLC) coupled with ESI-triple quadrupole ion MS and ESI-linear ion trap high-resolution MS, has been developed for the simultaneous quantitative and qualitative determination of phenolics, nucleosides and amino acids in the roots of fresh Gastrodia elata. Optimization of the analytical method provided higher separation efficiency and better peak resolution for the targeted compounds. The simultaneous separation protocols were also optimized by routinely using accurate mass measurements, within 5 ppm error, for each molecular ion and the subsequent fragment ions. In total, 31 compounds, including 23 phenolics, two nucleosides, four amino acids, one gastrodin and one other compound were identified or tentatively characterized. Mono-substituted parishin glucoside (9), methoxy mono-substituted parishin (13), methyl parishin (26), p-hydroxybenzyl di-substituted parishin (29), and p-hydroxybenzyl parishin (31) were tentatively identified as new compounds. Principal metabolite content analysis and the composition of eight representative G. elata cultivars of various species indicated that geographic insulation was the main contributor to clustering.

The pharmacological properties of G. elata extracts are largely attributable to the presence of accumulated secondary metabolites. Although metabolite studies have, so far, focused mainly on the medicinal properties of G. elata products and extracts, a metabolite survey could also provide information about the genetic and biochemical control of the metabolism during the plant's development. Additionally, G. elata may provide an interesting model for studying other biological processes and metabolic regulation [12][13][14].
The most abundant compound in G. elata, and also the main active ingredient, is parishin, an ester formed by the condensation of three gastrodin subunits. However, the Chinese Pharmacopoeia [15] designates parishin as the only characteristic ingredient of G. elata. Recently, though, many more compounds have been identified in G. elata by using mass spectrometry (MS). In one study [16], some parishin compounds that would be expected to occur in G. elata were not identified, and only m/z values of those identified in MS process were presented. It was challenging to make progress in the analysis of G. elata compounds. Furthermore, little information is available about the composition and content of metabolites in different cultivars of G. elata, as well as the relationship between the G. elata cultivars and their area of origin. A survey of these cultivars and their areas of orgin would therefore be valuable.
In addition to the above issues, separation efficiency and resolution of target peaks of compounds extracted from plant are common bottlenecks in the qualitative and quantitative analysis of metabolism [17]. If suitable separation protocols can't be established, the use of inappropriate protocols may influence the results of an analysis. The purpose of the present study was to develop a complete systematic method for the qualitative and quantitative analysis of the major bioactive compounds extracted from roots of G. elata by using ultra performance liquid chromatography (UPLC) coupled with ESI-triple quadrupole ion MS and ESI-linear ion trap high-resolution MS. To the best of our knowledge, this is a report describing a simple and time-saving analytical method for the simultaneous determination of multiple metabolites in fresh root samples from G. elata.
plants of each cultivars. The collected rhizomes specimens were identified as Gastrodia elata Bl by a taxonomist (Professor Ming Cheng) at Institute of Chinese Materia Medica, China Academy of Chinese Medical Sciences, Beijing, China.

Sample preparation
For each freeze-dried sample of G. elata root, a portion (0.5 g, 100 mesh) was accurately weighed into a 50 mL flask and extracted with 20 mL 50% aqueous methanol (methanol: water 50:50, v/v) in an ultrasonic bath for 30 min at room temperature. Each extract combination was performed in triplicate. The extract was centrifuged for 5 min at 15, 000 g, the supernatant was collected and all the samples were re-extracted as above twice more. The combined supernatant was then filtered through a 0.22-μm Millipore filter (Alltech Scientific Corporation, Beijing, China) before injection for LC/MS analysis.

UPLC method
Chromatographic separation was performed using a UPLC system (DaianU3000, Dionex Corparation, CA, USA). The equipment comprised an UPLC pump, a photodiode array (PDA) detector, and an auto-sampler set at 30°C. Phenolic detection in diode array detector was carried out at 270 nm, and spectrum scans were made from 200 to 400 nm. Separations were carried out using an Accucore C 18 column (2.1 mm × 100 mm, 2.6 μm particle size, Thermo Fisher Scientific, Bellefonte, PA, USA). The mobile phase consisted of water containing 0.1% formic acid (A) and acetonitrile (B). Linear gradient elution was performed at a flow rate of 0.2 mL/min. The solvent gradient was changed according to the following program: 0-8 min, 2% B; 8-12 min, 2-8% B; 12-25 min, 8-12% B; and 25-47 min, 12-25% B. The injection volume was 1 μL and chromatograms were acquired at 270 nm.

Q-Trap MS
MS analysis was performed using an LTQ Orbitrap mass spectrometer (Thermo Fisher Scientific, San Jose, CA, USA), fitted with an electrospray ionization (ESI) source operated in both negative and positive modes. The m/z range was 100-1200, with resolution set at 30000 using the normal scan rate. Data-dependent MS/MS events were always performed on the most intense ions detected in the full scan MS. The normalized collision energy was 30% for all compounds. Nitrogen was used as the sheath gas and helium as the collision gas. The key optimized ESI parameters were as follows: source voltage, 3.0 kV; sheath gas (nitrogen), 50 L/min; auxiliary gas flow, 10 L/min; capillary voltage, -35.0 V; capillary temperature, 350°C, and tube lens, -110.0 V. The ion injection time used was 50.0 ms. MS scan functions and UPLC solvent gradients were controlled by an X-calibur data system (Thermo Fisher, Scan Jose, CA, USA).

Method validation
The method was validated by characteristic indexes including linearity, the limit of detection (LOD), the limit of quantification (LOQ) and precision (inter-day. intra-day precision).

Statistical analysis
Quantitative data were analyzed with SPSS 16.0 for window. Principal component analyses (PCA) for identifying homogeneous groups of individual tissues based on measured phenolics, nucleosides and amino acids concentrations were performed in SPSS 16.0. Factors were identified using varimax rotation and the two most significant factors were extracted using the Kaiser-Meyer-Olkin criterio (KMO).

Optimization of HPLC and MS conditions
In preliminary tests, three columns were evaluated for the separation of target compounds from the roots of G. elata. These were a SunFire C 18 column (2.1 mm × 100 mm, 1.9 μm particle size, Waters Corporation, Milford, USA) ( Fig 1A-I), an Acquity BEH C 18 column (2.1 mm × 100 mm, 1.7 μm particle size, Waters Corporation, Milford, USA) ( Fig 1A-II), and an Accucore C 18 column (2.1 mm × 100 mm, 2.6 μm particle size, Thermo Scientific, USA) ( Fig  1A-III). Using the same elution protocol, the Accucore C 18 column (Fig 1A-III) gave slightly better resolution than the other two columns. Thus, it indicated that particle size may be an important factor affecting the resolution of compounds in G. elata. Based on earlier reports, two elution systems (Fig 1B-I and 1B-II) were then evaluated to optimize the peak shapes of metabolites separated from G. elata. Elution system B-I, in which acetic acid was added to the eluent, was based on research carried out by Wang [18], and allowed identification of 15 phenolics and 6 nucleoside derivatives in the roots of G. elata. Elution system B-II, in which 0.1% formic acid was added to the mobile phase, was based on research carried out by Ong et al. [19] and has been widely used to separate phenolics from the roots of G. elata. In the present study, we found that a higher concentration of formic acid (0.5%, v:v, B-III) improved peak separation efficiency (Fig 1B-III). Our research suggests that an Accucore C 18 column and a mobile phase containing formic acid (0.5%) is suitable for the analysis of metabolites in G. elata. The optimized experiment condition was replicated for eight different cultivars analysis.

Method validation
Calibration curves, limits of detection and quantification. Method validation was carried out using four external standards (gastrodin, parishin E, parishin B, and parishin). The calibration curves showed good linearity for all standards at 350 nm (r 2 0.9915). The standard solutions were detected by chromatography until the signal-to-noise (S/N) ratios were 3 and 10; the corresponding concentrations at these S/N ratios were defined as the LOD and LOQ, respectively. The lowest and highest LOD and LOQ were obtained for gastrodin (0.024 and 0.081 μg/mL) and parishin (0.359 and 1.196 μg/mL) ( Table 2).
Precision and accuracy of quantification. The precision of metabolite quantification was studied by examining the repeatability and intermediate precision for all compounds separated Optimized HPLC chromatograms at 220 nm selected from three column system (AI-AIII) and three mobile protocols (BI-BIII). Thirty-one compounds (peaks 1-31) were separated by BIII and identified using UPLC-LTQ Orbitrap mass spectrometry.
doi:10.1371/journal.pone.0150647.g001 from G. elata roots. Six standard samples were evaluated on the same day to determine the intra-day precision. Three samples were also extracted and analyzed on three consecutive days to determine the inter-day precision. Sample solutions were prepared at three concentrations (low, middle, and high), with three replicates of each concentration in order to validate method precision. Relative standard deviations (RSDs) were calculated to assess repeatability and precision. The RSDs of the four compounds were less than 4.05% for inter-day precision at the three concentration and 4.00% for the intra-day precision ( Table 3). The low RSD values obtained for the four compounds confirmed the high repeatability and intermediate precision of the method developed here. Accuracy and recovery of quantification. The accuracy of the method was investigated by measuring the recovery. This was assessed by adding three concentrations (high, middle and low) of standard solutions to known amounts of sample solution which were then extracted and subjected to quantitative analysis as described above. Each standard was tested at each concentration in triplicate. The equation used to define the percentage recovery was (detected amount-original amount)/spiked amount x 100. Recoveries obtained in this study were in the range 88.02-105.38% (Table 4), demonstrating that the analytical method developed in this study has high accuracy. The low RSDs of all standards (< 1.54%) indicate good reproducibility.

Analysis of constituents in G. elata
The identification of 31 compounds was performed in both positive and negative modes. Chemical structures, mass spectra (in PI, NI and NI-MS/MS), retention times, UV-Vis spectra, and retention times on the C 18 column are listed in Table 5. The identified compounds could be classified into three groups, phenolics, nucleosides and amino acids, based on their chemical structures.
Mass spectra of metabolites from G. elata in both positive and negative mode showed that the characteristic ions originated mainly from functional groups such as hydroxyl and carboxyl groups. By comparing UPLC retention times and UV and mss spectral data with those of reference standards, the target peaks were tentatively identified as described below.   (Table 5). Peak 5 was tentatively identified as adenosine, based on an earlier report [20]. Its fragmentation pattern was similar to that of uridine, with a characteristic fragment at m/z 136 [M+H-132] + .
Analysis of amino acids in G. elata. Although the UV absorption of the majority of amino acids is weak, the characteristic mass spectra of four amino acids were found in G. elata extracts. MS n analysis of phenolic in G. elata. The external standards, parishin, parishin B, and parishin E, were used to study one to third-order collision-induced dissociation (CID) spectra (MS 3 ) for phenolic identification. Full confidence fragmentation pathways could be achieved progressively, following comparison of product ion mass spectra of the compounds under  investigation with those recorded for parishin (Fig 2). Analyses of product ion spectra of molecules [M+Na] + provided information about the size of substituents as well as substitution patterns of parishin analogues. Similarly, for di-and mono-glycosides of parishin derivatives (9,  13 (Table 5).

Quantitative and qualitative analysis of compounds
Significant qualitative differences were found in the metabolites detected in different species as well as in plants collected from different producing areas. Interestingly, the characteristic compound, gastrodin, was not detected in the red G. elata cultivar collected from Shanxi province. S-(4-hydroxybenzyl)-glutathione was detected only in the hybrid obtained from Sichuan province.
Compounds gastrodin, parishin E, parishin B, parishin were quantified using their available counterparts as standards, and the other parishin derivative compounds using parishin as the standard, while other compounds using gastrin as the standard. all the studied 31 compounds were classified as five groups include amino acids, nucleosides, S-(4-hydroxybenzyl)-glutathione, gastrodin, parishin derivates for further statistical analysis.

Principal component analyses
Principal component analysis (PCA) was used to provide an overview of the complete data set, showing variability between compounds detected and G. elata growth area or species. PCA using these attributes could explain 79.49% of the variance, partitioned as 43.64% in principal component 1 (PC 1) and 35.85% in principal component 2 (PC 2) (Fig 3). The loading of PC 1 showed strong positive correlations with S-(4-hydroxybenzyl)-glutathione, gastrodin, and parishin derivatives whereas PC 2 showed an important positive correlation with amino acids and nucleosides. Relationships between G. elata collection areas and detected compounds are shown in Fig 3A and Fig 3B and relationships between G. elata species and detected compounds are shown in Fig 3A and 3C.
As shown in Fig 3B, the eight G. elata cultivars could be divided into four groups. Group I, comprising two G. elata cultivars (4 and 6) collected from Sichuan province, was located to the right of the PC1 axis and above the PC2 axis. G. elata cultivars collected from Sichuan province were characterized by high levels of amino acids and nucleosides. Group II consisted of one cultivar (5) from Sichuan province and was characterized by high levels of S-(4-hydroxybenzyl)-glutathione and gastrodin. Group III was located in the second quadrant and contained cultivars 2 (Guizhou, G. elata) and 7 (Shanxi, G. elata,) which have relatively high levels of amino acids. Group IV, located in the lower left part of the scatter plot, included two G. elata samples obtained from Guizhou (1 and 3) and one G. elata sample obtained from Yunnan (8) and was characterized by very low individual compound content.
PCA demonstrated a lack of strong characteristic clustering among different G. elata species. To some extent, our results show that geographic insulation affects metabolite synthesis to a greater extent than species diversity.

Conclusions
A reliable and effective method using UPLC coupled with ESI-triple quadrupole ion MS and ESI-linear ion trap high-resolution MS was successfully developed for online identification of low molecular weight metabolites of G. elata. A total of 31 compounds were identified or tentatively characterized, based mainly on fragment ion information obtained by UPLC-MS/MS. Five of these compounds were identified for the first time. PCA showed that the synthesis of G. elata metabolites varies with both species and geographic insulation. The analysis study of the active compounds in G. elata where quality control is of interest, in which case identification based on metabolites detection is possible and necessary.