A Method for LC-MS/MS Profiling of Coumarins in Zanthoxylum zanthoxyloides (Lam.) B. Zepernich and Timler Extracts and Essential Oils

The metabolites from the coumarin class, present in tissues of plants belonging mainly to the Rutaceae and Apiaceae families, included compounds with high chemical diversity such as simple coumarins and furocoumarins. These health-promoting components are recognized for their valuable biological activities in herbal preparations but also for their phototoxic effects. In this work, a targeted liquid chromatography (LC) coupled with tandem mass spectrometry (MS2) was developed for the screening of 39 reference standards of coumarins and furocoumarins in essential oils and plant extracts. Chromatographic separation was accomplished on reversed phase column using water/acetonitrile as the mobile phase and detection was performed on a hybrid QqQ/linear ion trap spectrometer fitted with an atmospheric pressure chemical ionization (APCI) source operating in positive ion mode. This analytical approach was applied to investigate the coumarin compositions of fruit essential oils and methanolic extracts obtained from separated parts (fruit, leaf, stem, trunk, and root) of Zanthoxylum zanthoxyloides. Ten coumarins and six furanocoumarins were reported in this species and data analyses were used to assess the suitability of these compounds to the metabolomics-based differentiation of plant organs. The quantification criteria of the metabolites in extract samples included linearity, limit of quantification, limit of detection, and matrix effect were validated. As reported for other species of the Rutaceae family, the concentration of coumarins was drastically higher in Z. zanthoxyloides fruits than in other plant organs.


Introduction
Within plant metabolites, coumarins and furanocoumarins represent a wide group of structurally highly diverse compounds, present in tissues of plants belonging mainly to the Rutaceae [1][2][3][4], Apiaceae [4], and Fabaceae families [5].This compound class is increasingly recognized as a valuable health-promoting constituent of edible plants and herbal plant preparations due to their wide spectrum of biological activities [6][7][8][9].However, coumarins and furanocoumarins have also been linked to phototoxic [10], mutagenic [11], carcinogenic [12], and hepatotoxic [13] effects and inhibitory properties of cytochrome P450s [14].Thus, the European Cosmetics Directive 76/768/EEC was recently modified, introducing for the first time a limit on the presence and use of the photosensitizing furocoumarins in cosmetics [15].The Commission Directive 95/34/EC has also added a new restriction to annex II (entry N • 358) as follows; "In sun protection and bronzing products, furocoumarines shall be below 1 mg/kg".In a more recent proposal SCCNFP/0392/00 (Scientific Committee on Cosmetic and Non-Food-Products), this 1 mg/kg limit is proposed to be generalized to all cosmetic products [16].In 2004, the European Food Safety Authority (EFSA) established a tolerable daily intake of coumarin of 0.1 mg/kg body weight based on its hepatotoxicity [17].
Although GC-MS is not the best choice for compounds that contain relatively polar or heat-labile substituents, one can make use of derivatization to get a better result [4].Several high performance liquid chromatography (HPLC) based methods targeting commercial and isolated components were developed in order to monitor coumarins.Reversed phase HPLC coupled with UV detection (HPLC-UV) [3,4,[34][35][36][37] or MS (LC-MS and LC-MS/MS) [2][3][4]34,35,[37][38][39][40][41][42][43] are the most sensitive and selective methods for detecting coumarin derivatives.However, LC-UV does not allow the quantification of concentrations of less than 10 mg/L [36], such as those encountered in the analysis of complex mixtures, while LC-MS is used to quantify concentrations ranging from µg/L to 10 mg/L.One of the problems of working with LC-MS of natural products is the choice of the ionization technique.Thus, electrospray ionization (ESI) [2,34,35,[37][38][39][40][41][42][43] or atmospheric pressure chemical ionization (APCI) [2][3][4]35,38] are the ionization sources commonly used for characterizing mass spectrometry coumarins.ESI is the technique of choice for polar and higher molecular weight compounds, while APCI is suitable for less polar compounds and those of lower molecular weight than ESI.Dugo et al. (2000) reported that coumarins and psoralens give a better response with APCI, while ions were not observed in ESI mode [38].
In the present study, an LC-MS 2 -based method was developed to dereplicate the possible presence of 39 commercially available coumarins and furocoumarins in Z. zanthoxyloides extracts.Following these results, this work was designed to validate a highly sensitive and selective method for the qualitative and quantitative determination of these components in the fruit essential oils and solvent extracts (fruits, leaves, and barks) of Senegalese Z. zanthoxyloides specie.LC-MS/MS data were also processed using Principal Component Analysis and Discriminant Analysis (PCA-DA) to assess the suitability of coumarin compounds to a metabolomics-based differentiation of plant organs.The application of this analytical approach led to the quality control of this commercial drug, traditionally used throughout Central and West Africa in abdominal and dental problems, sickle-cell disease, and skin disorders such as psoriasis and vitiligo [44,45].

Targeted LC-MS 2 Method
To obtain results on the structural diversity of coumarin-type components in plant extracts, an LC-MS 2 -based method was developed to detect the possible presence of 39 commercially available coumarins (31 standard compounds) and furocoumarins (8 standard compounds), which are characteristic of species from the Rutaceae family [1][2][3][4]14].For each reference compound, a relevant transition of the precursor-to-product ions were detected with the utilization of the multiple reaction monitoring (MRM) mode.Using APCI source operating in positive mode, the precursor ion [M + H] + for each of the 39 analytes was determined in MS1 full scan tests and the product ions in MS/MS experiments.MRM transitions of each analyte were optimized using direct infusion with the following MS/MS parameters; declustering potential (DP), entrance potential (EP), collision cell entrance potential (CEP), collision energy (CE), and collision cell exit potential (CXP) (Table 1).Retention times of reference compounds were determined by LC-MS 2 analysis in the multiple reaction monitoring (MRM).Retention times of the 39 coumarin-type components are also indicated in Table 1.Mass spectra of standard components were performed by the MRM mode followed by an enhanced product ion (EPI) scan, triggered by information dependent acquisition (IDA) criteria.The fragmentation behavior of coumarins and furocoumarins was consistent with previous data reported in the literature [34,35,38,39,41,46].Under collision-induced dissociation (CID), these compounds undergo neutral loss from pseudo-molecular ions [M + H] + , producing the corresponding high-abundance fragment ions: For instance, the main ions (base peak) found on the mass spectra of the furocoumarin xanthotoxin was seen at m/z 217 [M + H] + in Q1 and the predominant ion at m/z 202 ([M + H − CH 3 ] + (Figure 1).
In order to provide an evaluation of the variability of in Z. zanthoxyloides samples, essential oils and solvent extracts of each plant part were analyzed by LC-MS/MS, and MRM chromatograms were processed using PCA-DA.Targeted LC-MS profiles of the data set revealed separate clusters for essential oils and extracts, on the one hand, and differences associated with the plant parts, on the other hand.The PCA-DA analyses were based on calculations of intensity mean and standard deviation among the three replicates (methanolic extraction of each plant part collected on three distinct trees) of each peak, detected using scheduled MRM (Table S1).Using this statistical analysis, the scores plots of different Z. zanthoxyloides samples are displayed in Figure 3a.The principal factorial plane (constructed using D1 and D2) accounted for 50.6% of the entire variance of coumarins in plant samples.
In order to provide an evaluation of the variability of in Z. zanthoxyloides samples, essential oils and solvent extracts of each plant part were analyzed by LC-MS/MS, and MRM chromatograms were processed using PCA-DA.Targeted LC-MS profiles of the data set revealed separate clusters for essential oils and extracts, on the one hand, and differences associated with the plant parts, on the other hand.The PCA-DA analyses were based on calculations of intensity mean and standard deviation among the three replicates (methanolic extraction of each plant part collected on three distinct trees) of each peak, detected using scheduled MRM (Table S1).Using this statistical analysis, the scores plots of different Z. zanthoxyloides samples are displayed in Figure 3a.The principal factorial plane (constructed using D1 and D2) accounted for 50.6% of the entire variance of coumarins in plant samples.The fruit samples (essential oils and extracts) are correlated positively with the D1 axis (25.4%), whereas the extract samples of all other plant parts are negatively located on the D1 axis.Note that the trunk and stem samples are located in the same PCA space region, positively correlated with the D2 axis (25.2%), whereas the stem and root extract samples (located in the same quadrant) are negatively correlated with this axis.This indicates a potential differentiation of coumarin composition according to the Z. zanthoxyloides plant parts.The corresponding loading plot in Figure 3b shows the coumarins that make the most difference in separating samples.Characteristic marker components were identified to be responsible for clustering samples such as furocoumarins for fruit samples (essential oils and extracts) or scopoletin and isoscopoletin for trunk extracts.
Molecules 2017, 22, 174 5 of 13 coumarins that make the most difference in separating samples.Characteristic marker components were identified to be responsible for clustering samples such as furocoumarins for fruit samples (essential oils and extracts) or scopoletin and isoscopoletin for trunk extracts.From these results, it appeared that targeted LC-MS/MS, associated with statistical analysis using PCA-DA, is a rapid and effective approach to obtain similarities and/or differences in metabolic content among extraction methods and/or plant organs.In order to confirm PCA results and to obtain additional data for the differentiation of Z. zanthoxyloides plant parts according to coumarin-type metabolites, the quantification of these marker compounds was carried out by monitoring precursor-to-product ion transition MRM at specific retention times and by a comparison of peak areas in solvent extracts with those of calibration curves from corresponding standard compounds.From these results, it appeared that targeted LC-MS/MS, associated with statistical analysis using PCA-DA, is a rapid and effective approach to obtain similarities and/or differences in metabolic content among extraction methods and/or plant organs.In order to confirm PCA results and to obtain additional data for the differentiation of Z. zanthoxyloides plant parts according to coumarin-type metabolites, the quantification of these marker compounds was carried out by monitoring precursor-to-product The calibration curves of most coumarins exhibited linearity in the concentration ranges (0.01-5 mg/L) with correlation coefficients higher than r 2 ≥ 0.9934.The limit of quantification (LOQ) and limit of detection (LOD) of each compound were less than 0.1 mg/L and 0.03 mg/L.The detailed results of the regression equations, corresponding correlation coefficients, linear ranges, and LOQ are shown in Table 2.The matrix effect for 8-Acetyl-6-hydroxy-7-methoxycoumarin, used as the internal standard in plant extracts, ranged from −4.76% to +2.3% and fulfilled the criteria (≤15%).In the regression equation y = ax + b, x refers to the sample injection amount, y to the peak area; r 2 is the correlation coefficient of the equation, and LOQ is the limit of quantification.
These results are somewhat in agreement with previous studies conducted on fruit extracts of this species, which reported the presence of xanthotoxin, imperatorin, psoralen, bergapten, 6,7-dimethylesculetin, scopoletin, and umbelliferone using GC/MS [21].Furthermore, the presence of two coumarins, xanthotoxin and bergapten, was detected by Paris and Moyse-Mignon in fruit extract (1947) [20].It should be noted that all coumarin compounds identified in this study have been previously reported in the essential oils of other plants from the Rutaceae family, especially the Citrus species [2,3,35,38,40,43].As indicated in the literature for other species of Rutaceae family such as Ruta graveolens [47], the concentration of these metabolites was drastically higher in Z. zanthoxyloides fruits than in other plant parts.To conclude, the use of the LC-MS/MS method, coupled with data analysis, is an effective analytical approach to obtain qualitative and quantitative results on the coumarin and furocoumarin compositions of plant extracts and essential oils.This study reported for the first time 9 coumarins (isoscopoletin, daphnetin-7-methylether, coumarin, herniarin, 4-methoxycoumarin, 7-methylcoumarin, 6-methylcoumarin, isopimpinellin, and isobergapten) in Z. zanthoxyloides.Finally, it appeared that the Zanthoxylum species, commonly used as health supplements in traditional medicine, are attractive for further investigations on coumarin-type compounds, particularly in terms of biological activity and the toxicity of plant extracts.

Solvents
Methanol (HPLC grade) and hexane (HPLC grade), used for sample extraction, were purchased from Fisher Scientific (Illkirch, France).The solvents used for liquid chromatography were LC-MS grade acetonitrile (ACN), obtained from Fisher Scientific.Deionized water was purified using a Milli-Q water (Millipore, Bedford, MA, USA) purification system.

Plant Material
The fruit, leaf, stem, and bark (root and trunk) samples of Z. zanthoxyloides were harvested in May 2015 (fruit ripening period) from three trees, growing wild in one Senegalese locality, Kafountine (12 • 56'5.49926"N, 16 • 44'45.28315"W).The botanical identification of the plant material was performed by Dr. William Diatta from the Department of botanical and pharmacognosy of University Cheikh Anta Diop of Dakar.A voucher specimen was deposited at the herbarium of that institution under number 001299.

Plant Extract and Essential Oil Preparation
Each plant organ (fruit, leaf, root bark, stem, and trunk bark) of each tree has been extracted separately.Plant samples were air dried for a period of four weeks at ambient temperature.The plant material was powdered with an average particle size of 0.2 mm using a blade miller (Polymix PX-MFC 90D, KINEMATICA AG, Luzern, Switzerland).50 g of powder samples were extracted with 3 × 200 mL of methanol over 48 h each time, at room temperature under magnetic stirring.The solutions were combined, filtered through filter paper (PRATDUMAS, Couze-St-Front, France) and evaporated to dryness using a rotary evaporator (Laborota 4000, Heidolph, Schwabach, Germany).The methanolic solution was evaporated to dryness using a rotary evaporator and the extract yields (w/w, calculated on a dry weight plant) were 27.8%, 16.3%, 20.6%, 5.2%, and 14.2% for fruit, leaf, root bark, stem, and trunk bark, respectively.All dried extracts were stored at 4 • C until analysis.Fruit samples were also hydrodistillated (6 h) using a Clevenger-type apparatus (Midisciences, Fuveau, France) according to the method recommended in the European Pharmacopoeia [48].The yield of essential oil (w/w, calculated on a dry weight basis) was 1.0% and the density was 0.86.Prior to LC-MS 2 analysis, 10 mg of each sample extract was dissolved in ACN/H 2 O (1:1 v/v) to obtain a solution at a final concentration of 100 mg/L.The fruit oil was diluted to 1/100 in acetonitrile.Finally, the solutions were filtered on a 0.2 µm polytetrafluoroethylene (PTFE) filter (Whatman, Maidstone, UK).1.4 mL of leaf, root, stem, and trunk extract solutions and essential oil solutions were supplemented with 0.1 mL of 8-Acetyl-6-hydroxy-7-methoxycoumarin, used as the internal standard at a concentration of 20 mg/L.For fruit extract, 0.3 mL of the solution was supplemented with 1.1 mL ACN/H 2 O (1:1 v/v) and 0.1 mL of the internal standard.

MS 2 Conditions
MS 2 conditions were carried on an AB Sciex (Toronto, ON, Canada) 3200 QTRAP linear triple quadrupole fitted with an atmospheric pressure chemical ionization (APCI) ion source operating in positive mode.High purity nitrogen was used as both a nebulizer and turbo gas.The APCI source was operated with following settings in positive mode; curtain gas (CUR) 25 psi, nebulizer gas (GS1) 31 psi, heater gas (GS2) 65 psi, ion spray voltage (IS) 5000 V, and temperature 450 • C. Standard solutions (component concentration: 0.1 mg/L) were directly infused at the flow rate of 10 µL/min in the MS/MS apparatus.Multiple EPI mass spectra of each compound were recorded in the range of m/z = 50-500 at 4000 Da/s.IDA properties were set to select 1 to 2 peaks above 300 counts with an exclusion filter after 5 occurrences for 30 s with dynamic background subtraction.The software used for data acquisition and data analysis was Analyst 1.5.2 (AB Sciex, Framingham, MA, USA).

LC-MS 2 Quantification
The LC-MS 2 method was validated according to the US Food and Drug Administration (FDA) guidance for bioanalytical method validation by assessing linearity, limit of quantification (LOQ), limit of detection (LOD), and matrix effect [49].External standard calibration lines were generated by three repeated injections of standard solutions at seven concentration levels (0.01; 0.1; 0.25; 0.5; 1; 2.5; and 5 mg/L) at 1 day.A plot of peak area with respect to the corresponding concentration was used to demonstrate linearity.The linear regression equation and correlation coefficient were calculated by weighted (1/x 2 ) least-squares linear regression analysis.Linearity was considered to be acceptable when correlation coefficients were 0.99 or better and calibrators had accuracies of 85%-115% and precisions within ± 15% RSD (relative standard deviation).The LOD (signal-to-noise <3) was defined as the amount that could be detected, while the LOQ (signal-to-noise >10) was the lowest concentration point of calibration curve at which accuracy (relative error, RE) within 20% and precision below 20% can be considered acceptable.The absolute matrix effect was evaluated by comparing the chromatographic peak areas of 8-Acetyl-6-hydroxy-7-methoxycoumarin (internal standard) in real sample extracts with those of 8-Acetyl-6-hydroxy-7-methoxycoumarin present in the "neat" mobile phase.The matrix effect is considered to be obvious if the ratio is less than ±15% [50].

Statistical Analysis
The triplicates of each methanol extract (fruit, leaf, stem, root, and trunk barks) and the essential oil from the fruit were analyzed by LC-MS/MS in the multiple reaction monitoring (MRM).MRM-MS chromatographic profiles of each sample were performed using PCA and DA in Marked View™ software (Sciex, Toronto, ON, Canada).PCA (principal components analysis) performs plans (principal components, PCs) where both objects (plant samples) and variables (coumarins components) are plotted according to the variance present in the data set.Discriminant analysis (DA) is performed in order to sharpen the separation between groups of observations, by hopefully rotating PCA components such that a maximum separation among plant samples is obtained, and to understand which coumarin variables carry the sample separating information.

13 Figure 1 .
Figure 1.Enhanced product ion (EPI) mass spectrum of xanthotoxin with the atmospheric pressure chemical ionization (APCI) source in positive ion mode.

Figure 1 .
Figure 1.Enhanced product ion (EPI) mass spectrum of xanthotoxin with the atmospheric pressure chemical ionization (APCI) source in positive ion mode.

Figure 3 .
Figure 3. (a) Scores plot of principal component analysis and discriminant analysis (PCA-DA) of the Z. zanthoxyloides extracts (fruit, leaf, stem, root, trunk) and fruit essential oils analyzed using LC-MS/MS in the MRM mode; (b) Scores plot of PCA-DA of coumarins and furocoumarins detected in Z. zanthoxyloides samples using LC-MS/MS in the MRM mode.

Table 3 .
Contents of the coumarins and furocoumarins in Z. zanthoxyloides plant parts (n = 3).