LC-MS based metabolomics study of different parts of myrtle berry from Sardinia (Italy)

2.1Plant materials

Seeds of two cultivars of Myrtus communis, were collected from the geographic area of Sassari and Cagliari (Sardinia, Italia) respectively, and then grown in the same Experimental Station of the University of Sassari located in Oristano (Sardinia, Italy). Three samples of purple berries from plants derived from seeds of the Sassari area (1, 2, 3) and three samples from plants derived from seeds of the Cagliari area (4, 5, 6) were randomly harvested (≈2 kg) during the fruit ripening stage in December 2014. Before analysis, the samples collected were immediately frozen in liquid nitrogen and then freeze dried. The peel and pulp were removed and separated from the seeds. Whole berry, peel, pulp and seeds were used for the analysis.

In total 18 samples were obtained and labeled as follows: 1S, 2S, 3S (seeds from samples developed from original seeds of the Sassari area); 4S, 5S, 6S (seeds from samples developed from original seeds of the Cagliari area); 1P&P, 2P&P, 3P&P (peel and pulp from samples developed from the original seeds of the Sassari area); 4P&P, 5P&P, 6P&P (peel and pulp from samples developed from original seeds of the Cagliari area); 1B, 2B, 3B (whole berries from plants developed by seeds of the Sassari area); 4B, 5B, 6B (whole berries from plants developed by seeds of the Cagliari area). Each sample was analyzed in duplicate (obtaining 36 samples in total).

Different cultivars were not registered (with a specific voucher number), but they were described in several scientific publications where specific names were attributed to them (as required by Italian law), Maria Rita, Barbara, Daniela, Giovanna, Lelia, Tonina [18].

2.2Sample preparation

Seeds, peel and pulp and whole berries were lightly pulverized following the procedure described by De Vos et al. (2007) [19] with slight modifications: 0.5 g was extracted with 10 ml of a solution comprised of 70% aqueous methanol acidified at 0.1% of formic acid followed by sonication for 15 minutes and centrifugation for another 15 minutes at 1750 rpm. The supernatant was collected and filtered through 0.45 μm filters, and then 100 μL of extract was diluted in 900 μL of water (of LC-MS grade), for a final concentration of approximately 1 mg/mL.

For the LC-MS analysis 10 μL were used. Each sample was analysed in technical duplicate.

2.3Reagents and solvents

Formic acid and methanol for extraction were purchased from VWR international PBI S.r.l. (Milano, Italy). Acetonitrile, water and formic acid (all of LC-MS grade) were purchased from Merck (Darmastadt, Germany).

2.4HPLC-UV/VIS analysis

An Agilent (Palo Alto, CA, USA) 1260 Infinity system consisting of a G1312C binary pump, a G-1328A Rheodyne injector (20 μL injection loop) a G-1379A degasser and a G1314B photodiode array detector were employed to develop the chromatographic method. Analyses were performed using a Waters XSelect CSH C18 (2.1 mm×150 mm particle size 3.5 μm) column, eluted with water containing 0.1% formic acid (solvent A) and acetonitrile containing 0.1% formic acid (solvent B). A linear gradient program at a flow rate of 0.200 mL/min was used: 0–15 min, from 10 to 20% (B); 15 to 25 min, from 20 to 40% (B); 25 to 35 min from 40 to 60% (B) then to 100% (B) for 5 min and back to 10% (B) for other 5 min. This gradient was used for LC-ESI-Orbitrap-MS analysis described below. Detection was carried out with two wavelengths, 360 nm specific for flavonoids and 520 nm specific for anthocyanins.

2.5LC-ESI – orbitrap-MS analysis

LC-ESI-Orbitrap-MS analyses were performed essentially as described by D’Urso et al. (2015) [20].

An HPLC method coupled with a hybrid mass spectrometer, combining a linear trap quadruple (LTQ) and an Orbitrap mass analyser, was developed for the study of the main metabolites characteristic to each part of the berry. Experiments were performed working with a Thermo Scientific liquid chromatography system based on a quaternary Accela 600 pump and an Accela auto sampler, hyphenated with a linear orbitrap hybrid mass spectrometer (LTQ-Orbitrap XL, Thermo Fisher Scientific, Bremen Germany) equipped with electrospray ionization (ESI). Separation was performed on a XSelect CSH C18 (Waters, Milford, MA) column (2.1 mm×150 mm particle size 3.5 μm). The mobile phase selected after optimization consisted of solvent A (water acidified with 0.1% formic acid) and solvent B (acetonitrile acidified with 0.1% formic acid). The gradient program used is described above in HPLC-UV/Vis analysis. The mass spectrometer was operated in negative ion mode. ESI source parameters were as follows: capillary voltage –12 V; tube lens voltage –121.47; capillary temperature 280°C; Sheath and Auxiliary Gas flow (N₂) 30 and 5, Sweep gas 0 Spray voltage 5. MS spectra were acquired by full range acquisition covering m/z 200–1600. A data dependent scan was performed with the aim of obtaining MS/MS experiments, selecting precursor ions corresponding to most intensive peaks in LC-MS analysis. The experiments were also performed in positive ion mode for a quality screening followed by a fragmentation study. ESI source parameters were as follows: capillary voltage 49 V; tube lens voltage 120; capillary temperature 280°C; sheath and Auxiliary Gas flow (N₂) 30 and 5, Sweep gas 0 Spray voltage 5. MS spectra was acquired in the full range acquisition m/z 250–1600.

Spectral characteristics fragmentations allowed for the identification of phenolic compounds in addition to specific retention times, and finally, data were compared with literature. Xcalibur software version 2.1 was used for instrument control, data acquisition and data analysis.

2.6Multivariate data analysis

LC-MS raw data of the 36 samples deriving from LC-ESI-Orbitrap-MS analysis (negative ion mode) were analysed using a platform independent open source software package called MZmine (http://mzmine.sourceforge.net/). Using this toolbox with normalization of total raw signal, we detected 634 peaks. After exporting the processed data in tabular format (.cvs file), further analysis of the data matrix were performed by SIMCA P+ software 12.0 (Umetrix AB, Umea Sweden) by Principal Component Analysis (PCA). PCA was performed by applying the peak area obtained from LC-MS analysis [21, 22]. Pareto scaling was applied before multivariate data analysis.

3.1LC-MS metabolomics analysis

Aqueous/methanol extracts of seeds, peel and pulp and whole myrtle berry were analyzed by LC-ESI-Orbitrap-MS.

Base-peak chromatograms in negative ion mode of the different parts of myrtle berries extracts are presented in Fig. 1, while base-peak chromatograms in positive ion mode are presented in Fig. 2. In total 35 compounds were identified or putatively identified based on retention time, accurate mass measurement, fragmentation pattern and comparison with data reported in literature (Table 1). As far as we know 19 of these compounds were identified for the first time in myrtle berries. In the current work, the identified compounds are classified into six groups: hydrolysable tannins; hydroxycinnamic acid; gallomyrtucommulones; flavanols; flavonols; anthocyanins. All of them are present in all parts of the fruit, with variability in intensity.

Fig.1

Base peak chromatograms in negative ion mode of whole berry, Peel & pulp and seeds of Myrtus communis methanol extracts.

Fig.2

Base peak chromatograms in positive ion mode of whole berry, Peel & pulp and seeds of Myrtus communis methanol extracts.

3.1.1Hydrolysable tannins

This class of compounds was detected at 360 nm and identified through LC-MS and MS/MS experiments by operating in negative ion mode.

Compound 1, 2, 4, 5, 13, 14, 15, 16, 25, 26, 21, 31 were identified as hydrolysable tannins; these compounds were present in all parts of the fruits, but their content was higher in the seeds than in the peel and pulp (see Fig. 1 and 2).

Compounds 1, 2, 4, 5 and 14 were previously identified in leaves of Myrtus communis.

Compound 1 showed a pseudo molecular ion at m/z 481.0617 corresponding to the molecular formula C₂₀H₁₇O₁₄ producing in MS/MS a fragment at m/z 301 which indicates the release of ellagic acid, thus by comparison of MS data reported in literature the compound was tentatively identified as hexahydroxydiphenoyl (HHDP) hexoside [8].

Compound 2 showed a pseudo molecular ion at m/z 331.0664 corresponding to the molecular formula C₁₃H₁₅O₁₀, that gives fragment ion at m/z 271 [M-H-60]^- and 169, that corresponds to the aglycon form due to the loss of a hexosyl moiety (162 Da). This fragmentation pattern agrees with data reported in literature and database [9] and thus the compound was identified as monogalloylhexose; Compound 4 showed a pseudo molecular ion at m/z 633.0726 corresponding to the formula C₃₄H₂₁O₂₂ and when submitted to fragmentation gave a main product ion at m/z 301, which corresponds to the loss of a galloyl hexose unit (332 Da) from the precursor ion. Thus, this compound was characterized as tannin strictinin (galloyl-HHDP hexose), in agreement with the data reported in the literature [9]. Compound 5 showed a pseudo molecular ion at m/z 343.0664 corresponding to the formula C₁₄H₁₅O₁₀ that in MS/MS gave two principal product ions at 191 and 169, thus the compound was identified as galloylquinic acid, the same fragmentation pattern was reported by Taamalli et al. (2014) in the leaves of Myrtus communis. Compound 14 showed a pseudo molecular ion at m/z 781.0523 corresponding to the formula C₃₄H₂₁O₂₂, thus in comparison with database (Knapsack; mass bank) and data reported in literature [8], it was tentatively identified as punicalin. Compounds 13, 15, 16, 25 and 26 were for the first time identified in the species of Myrtus communis.

Compound 13, showed a pseudo molecular ion at m/z 785.0836 corresponding to the molecular formula C₃₄H₂₅O₂₂ that submitted to fragmentation gave a principal product ion at m/z 301 (loss of digalloylhexose), 483(loss of HHDP), can be identified as digalloylhexose, presumably tellimagrandin I; in fact a similar fragmentation was previously observed for this compound and it was tentatively identified by Boulekbache-Makhlouf et al. (2013) [23] in leaves of Eucalyptus globulus (family Myrtaceae). Compound 15 showed a pseudomolecular ion at m/z 783.0679 corresponding to the molecular formula C₃₄H₂₃O₂₂, whose MS/MS spectrum is reported as an example in Fig. 3. The pseudo molecular ion at m/z 783.0679 produced in MS/MS fragment ions at m/z 301 (ellagic acid; [M- 482]^-, loss of HHDP hexose) at m/z 481 (deprotonated HHDP hexose; [M- 302]^-, loss of HHDP) and a minor peak at 613(probably due to the loss of gallic acid [M-152]^- and rearrangement of the resulting ion), thus it was identified as di-HHDP hexose, presumably pedunculagin; the fragmentation pattern was previously reported by Simirgiotis et al. (2013) [24] and a similar fragmentation was observed by Boulekbache-Makhlouf et al. (2013) [23] in leaves of Eucalyptus globulus (family Myrtaceae). Compound 16 showed a pseudo molecular ion at m/z 935.0789 corresponding to the molecular formula C₄₁H₂₇O₂₆ that submitted to fragmentation gave three principal product ions at 783, 656 and 301. The fragmentation pattern agrees with data reported in literature and database [25] and thus the compound was identified as casuarictin.

Fig.3

MS/MS spectrum of pedunculagin (compound 15).

Compound 25 showed a pseudo molecular ion at m/z 933.0632 corresponding to the molecular formula C₄₁H₂₅O₂₆, in comparison with database (Knapsack; mass bank) and data present in literature the compound was tentatively identified as castalagin, it was previously reported by Fujita et al. (2015) [26] in camu-camu fruit.

Compound 26 showed a pseudo molecular ion at m/z 937.0945 corresponding to the molecular formula C₄₁H₃₀O₂₆, in comparison with database (Knapsack; mass bank) and data reported in literature [24], it was tentatively identified as tellimagrandin II.

Compounds identified, pertaining to this class are reported in Fig. 4.

Fig.4

Chemical structures of hydrolysable tannins tentatively identified for the first time in different parts of myrtle berries.

Compound 21 was identified as ellagic acid hexoside and compound 31 as ellagic acid, the two compounds were previously reported in Myrtus communis leaves [9, 27].

3.2Multivariate data analysis

The Multivariate Data Analysis was basically performed following the experimental protocol described by D’Urso et al. (2015). LC-ESI-Orbitrap-MS chromatograms were pre-processed using MZmine (a free software) in a way to compensate for variations in retention time and m/z value between each chromatographic run. A peak list table was obtained from pre-processed chromatograms. In the final table, rows represented the individual samples (36 samples: 18 biological samples in technical duplicates), and columns represented integrated and normalized peak areas. Furthermore, this data was used through an approach of untargeted analysis and treated with an unsupervised Multi Variated Data Analysis method (PCA).

Principal Component Analysis (PCA) was performed by applying the peak areas of each peak observed in the LC-MS dataset (excluding the noisy), and a matrix was constructed by using these areas (variables), with the different analyzed samples as columns of the matrix. The resulted score scatter plot is reported in Fig. 5. The first component describes 70.22% of variance while the second only 7.72%. The choice of principal components was realized on the basis of the fitting (R2X) and predictive (Q2X) values for thePCA model.

Fig.5

Principal Component Analysis, score scatter plot, LC-ESI-Orbitrap-MS analysis. Panel A: colored according to myrtle berries parts. Panel B: colored according to geographic origin of samples.

The score scatter plot in Fig. 5A is colored according to the parts of fruit under analysis. There is clearly a differentiation of the samples based on this parameter: seeds are all grouped on the left part of the plot, peel & pulp and whole berry are distributed on the right part. In Fig. 5B the same scatter plot is displayed, colored according to the cultivar of samples (different cultivar is related to the different geographical origin of the seeds). By using this visualization mode, the peel and pulp appear as an important discriminator for samples of different cultivar and different geographic origin. Really, samples pertaining to the cultivar from the area of Sassari are on the right upper part of the plot while samples pertaining to the cultivar from the area of Cagliari are on the bottom right. However, this differentiation, mainly showed by peel and pulp samples is not present between the seeds, because all the samples are grouped in the same area. Thus, this is evidence that the metabolites that make the two cultivar different are the phenolic compounds, and principally anthocyanins and flavonoids, present in big amount specially in peel and pulp. In this study, these phenolic compounds appear to be marker compounds for geographic and genetic origin of myrtle berry.