First Profile of Phenolic Compounds from Maltese Extra Virgin Olive Oils Using Liquid-Liquid Extraction and Liquid Chromatography-Mass Spectrometry

First Profile of Phenolic Compounds from Maltese Extra Virgin Olive Oils Using Liquid-Liquid Extraction and Liquid Chromatography-Mass Spectrometry Lucienne Gatt , Frederick Lia, Marion Zammit-Mangion, Simon J. Thorpe, and Pierre Schembri-Wismayer 1 Department of Physiology and Biochemistry, Faculty of Medicine and Surgery, University of Malta, Msida, MALTA 2 Centre for Molecular Medicine and Biobanking, University of Malta, Msida, MALTA 3 Department of Chemistry, Faculty of Science, University of Malta, Msida, MALTA 4 Department of Chemistry, University of Sheffield, UNITED KINGDOM 5 Department of Anatomy, Faculty of Medicine and Surgery, University of Malta, Msida, MALTA

EVOOs the most common flavonoids are apigenin, luteolin and cyanidins 11,13 . The phenolic acid sub-group is also characterized by diversity and includes vanillic acid, p-coumaric acid, gallic acid, syringic acid as well as caffeic acid 14,15 .
In the distant past, the Maltese Islands had a thriving oilproducing industry 16 but this was eradicated as higher cash-generating crops like cotton became favored. This resulted in the destruction of a number of olive groves and the loss of olive tree germplasm. Recent years have seen belated attempts to regenerate this industry. In the late 1990s, foreign cultivars mainly from Italy and Spain such as Frantoio, Carolea and Nocellara Messinese were introduced to Malta 17 . Successful attempts were also made to revive the industry using Maltese trees regarded as indigenous, such as the Bidni variety. According to the Census of Agriculture, reintroduction programmes of olives have led to a total of 140.3 hectares to be occupied by olive groves, with 72.6 hectares being cultivated for olive oil production.
The Bidni variety is described as having a high pulp to seed ratio, and a marked resistance to diseases and parasites such as the vascular disease-causing fungus, Verticillium dahliae and the olive fruit fly Bactrocera oleae 16 . Mazzitelli et al. 18 analysed the molecular biology of Bidni, Malti and Bajda varieties using random amplified polymorphic DNA RAPD and showed that the Bajda shares homology with Italian olive trees as opposed to the other two varieties. Gatt et al. 19 studied the cyclooxygenase activity in EVOO derived from the Bidni variety while Lia et al. 20 determined the anti-oxidant activity in EVOOs following solid-phase extraction SPE .
In this study, the polar fraction of EVOOs was separated by LLE and the total phenolic content TPC , ortho diphenolic content TdPC and flavonoid content TFC were assessed using quantitative assays. LLE is based on the distribution of an analyte between two phases: the aqueous and the organic phase. It is reported to yield a high recovery of secoiridoids 21 . LLE has the advantage of being easy to operate and it does not require expensive apparatus. Moreover, total phenol recovery rates are very high in LLE and amount to around 93 , making them even higher than those recorded for SPE using diol phase cartridges 22 .
The objectives of this study were therefore to characterize the main commercially produced EVOOs from Malta as well as to identify their phenolic profiles using spectrophotometric assays as well as LC-MS analysis.

Materials
All monocultivar EVOOs obtained locally were collected at the same stage of harvest and had been subjected to the same irrigation regime. Samples of EVOOs Bidni, Malti and Carolea were purchased during the months of October and November in 2012, 2013 and 2015. During 2014, none of these varieties could be collected due to the poor olive harvest. Another monocultivar olive oil, the Barbuto TM from southern Sicily, made using the Tonda iblea T variety, was purchased in the months of October and November in 2014 and 2015. A fifth EVOO, I which is a commercial EVOO, was obtained from a local supermarket in 2013, 2014 and 2015. This is produced from different Tuscany olives, and was chosen solely for comparative purposes, as the only EVOO that was not a monocultivar.

Reagents
All chemicals used in this study were supplied from Sigma-Aldrich ® St Louis, MO, USA safe for the absolute ethanol used for phenolic compound stock solution which was supplied from Scharlau China .

Isolation of phenolic compounds from olive oil
The polar fraction of EVOO was obtained using the  The sample was dissolved in absolute ethanol and stored at 20 . Phenolic extracts obtained from Bidni, Malti, Carolea, Tonda iblea olive oil and the supermarket brand were labelled as B, M, C, T and I respectively. 2.3.2 Determination of the total phenol content using the Folin-Ciocalteau F-C assay Total phenol content was determined using the method of Slinkard and Singleton 25 , with a reduction in volumes as described by Waterhouse 26 . Gallic acid was used as the standard. A stock solution of gallic acid was prepared from which different concentrations of the standard were prepared. Twenty microlitres of each gallic acid standard and/ or each phenol sample diluted tenfold was added to 1.58 mL water and 100 µL of Folin-Ciocalteau reagent. 300 µL of 20 w/v anhydrous sodium carbonate solution was added, mixed and heated at 40 for 30 minutes. The absorbance of each solution was read at 765 nm using a UV/ Visible Spectrophotometer Pharmacia . 2.3.3 Determination of the total ortho diphenolic content using the Arnow s assay The method used was modified from Woisky and Salatino 27 . Arnow s reagent was prepared by adding 10 g of sodium molybdate dihydrate and 10 g of sodium nitrite in 100 mL of a 1:1 v/v Ethanol Scharlau : water mixture. A stock solution of pyrocatechol was prepared from which different concentrations of the standard were prepared. In a 96 well plate 20 µL of each pyrocatechol concentration and/or each phenol sample diluted fourfold was pipetted and to each, 20 µL 1M HCl added, followed by 20 µL of Arnow s reagent. The plate was shaken for 5 minutes at 500 rpm and then incubated at room temperature for 15 minutes and 80 µL of deionised water were added, followed by 40 µL of 1M KOH. The absorbance was read at 370 nm using a UV/Visible microplate reader SPECTROstar Nano, BMG LABTECH . 2.3.4 Determination of the total flavonoid content using the Aluminium Chloride method The method used to determine the total flavonoid content of the phenolic extracts was the Aluminium Chloride Colorimetric Method. A stock solution of catechin was prepared from which different concentrations of the standard were prepared. In a 96 well plate 25 µL of each catechin concentration and/or each phenol sample diluted fivefold was pipetted and to each, 7.5 µL 10 aluminium chloride, 7.5 µL of a 7 w/v sodium nitrite and 80 µL distilled water were added. The plate was shaken for 5 minutes at 500 rpm and then incubated at room temperature for 25 minutes. After this, 100 µL of 1M KOH was added to each well. The absorbance was read at 415 nm using a UV/Visible microplate reader SPECTROstar Nano, BMG LABTECH .

LC-MS analysis
The apparatus consisted of a Waters, Alliance, 2695 Separations module connected to a Micromass, LCT classic, mass spectrometer. Mass Lynx software, version 4.1 was used for data collection and interpretation. A crude, phenolic extract 2 mg of sample B was dissolved in methanol 1 mL . An aliquot 20 µL was injected onto a C18 reverse phase column 50 mm 2.0 mm operating at 35 . The solvent gradient was 5 acetonitrile 0.1 formic acid to 95 acetonitrile 0.1 formic acid in 30 minutes and then held for 5 minutes. The column was equilibrated at the starting conditions for 10 minutes before analysis of subsequent samples, C, M, T and I.

Statistical analysis
For each sample, all parameters were determined in triplicate. Using the software IBM ® SPSS ® Statistics Version 21, all data sets were tested for normality using the Shapiro Wilk s test. The Kruskal-Wallis test was selected as the non-parametric test for pairwise comparisons, with significant comparisons having a p-value less than 0.05. The Pearson product-moment correlation coefficient was used to determine correlation between TPC, TdPC and TFC.

Results and Discussion
The total phenolic TPC , ortho diphenolic TdPC and flavonoid content TFC for each of the oils studied are presented in Fig. 2. The highest TPC and TFC were recorded in the EVOO derived from the B variety followed by the EVOO derived from the T variety cultivated in Sicily. The lowest quantity of TPC and TFC were recorded in the supermarket oil variety I. The TPC values for the B EVOO and that of the T variety appear comparable to autochthonous EVOO varieties such as the Italian Tonda di Caligari and Bosana, for which values are quoted to be 261.18 83.83 mg/kg GAE and 355.20 121.34 mg/kg GAE respectively 28 .
The TPC is an important parameter in the classification of EVOO as mild, medium or robust where robust reflects the highest TPC. In this respect, mean TPC values for each category differ between both producers and experts, though the difference is the most pronounced for the former category. For experts, EVOO is mild if the TPC is 127 5 mg/kg, while for producers the value is 170 11 mg/kg. It is medium if the TPC is 223 5 mg/kg for experts and 226 7 mg/kg for producers. It is classified as a robust oil if the TPC is 350 9 mg/kg for experts and 291 11 mg/ kg for producers. Following this classification, I, M and C appear to be all mild EVOOs, while T and B appear to be medium if classified according to the experts classification or robust for B if classified according to producers values 29 .
The link between phenolics and bitterness of EVOOs has long been established and has been attributed to a number of different phenols. While Kiritsakis, García et al., and Soler-Rivas, Espín, and Wichers 30 32 linked this property to oleuropein derivatives, others such as Gutiérrez-Rosales, Perdiguero, Gutiérrez and Olias, and Angerosa et al. 8,33 quoted both oleuropein and ligstroside aglycones and for Tovar et al. 34 , this property is only a result of ligstroside derivatives.
A number of studies show that the presence of a second hydroxyl group increases the antioxidant activity of EVOOs 5, 35 38 . The TdPC was highest in the EVOO derived from the Bidni variety followed once again by the Sicilianderived Tonda iblea variety. As with the TPC results, the lowest TdPC was reported for the supermarket oil. The TdPC analysed for C, M, I and T was found to be within the range of other oils such as a number of Argentinian VOOs, though these were recorded as caffeic acid equivalents by Laincer et al. 39  Of all locally derived oils, the EVOO derived from the Malti variety contained the lowest TPC, TdPC and TFC. However, this may be a reflection of the difficulty encountered to separate the organic and aqueous layer separation during LLE, due to the presence of an interphase layer that was recalcitrant to separation. As a result, some phenols may have not been collected in the aqueous layer resulting in a lower TPC. While a number of studies report the TFC of a variety of plant extracts, those investigating the TFC of EVOOs are very limited. The results obtained in this study contrast highly with those reported by Ebrahimi et al. 42 as rutin hydrate equivalents for crude olive oils in Iran, who report TFCs of 2.73 mg/g, 3.44 mg/g, 3.53 mg/g, and 3.61 mg/g. However, a clear cut comparison of this data with that of our study is not possible as the values reported by Ebrahimi et al. 42 are expressed as rutin hydrate equivalents not as catechin equivalents as in this study. Values reported by Ammar et al. 43 also greatly differ from those obtained in this study. Ammar et al. 43 record the TFC of a Tunisian Chemlali olive oil, as a value of 14.50 0.29 mg/kg CE. Figure 3 shows that the TPC, TdPC and TFC were all found to be positively correlated as determined by the Pearson product-moment correlation coefficient. The TPC was found to be positively correlated with both the TdPC and TFC Pearson s values of 0.906 and 0.896 respectively . There also appears to be a positive correlation between TdPC and TFC Pearson s value of 0.995 . Also, pvalues confirm that there is a statistically significant correlation between all the variables.
The phenolic profile of each of the EVOOs was analysed using LC-MS analysis where the representative total ion chromatograms obtained through LC-MS analysis of the five EVOO phenolic extracts are shown in Fig. 4. The compounds identified through LC-MS are presented in Table 1 along with the m/z of each peak obtained through MS and the compound or fragments responsible for that respective m/z. Figure 4 shows that the total ion chromatogram for all five EVOOs consists of a total of 27 peaks at identical retention times but differing peak heights. This indicates that the EVOO phenolic extracts contain the same phenolic compounds but at differing peak heights. While a number of compounds seem to be present at similar peak heights across the oils, other peaks indicate differences. Such differences correspond to peaks 2, 8, 16, 17, 26 and 27. In this study, over 30 phenolic compounds were identified belonging to different classes of phenolic compounds. On one hand, the phenolic compounds in the Bidni, Caro-

First Profile of Phenolic Compounds from Maltese EVOOs Using LLE and LC-MS
lea, Tonda iblea and the supermarket varieties that were present at the highest concentration were sinaptic acid or hydroxy-elenolic acid and tyrosol glycoside. On the other hand, in the Malti variety, the most prominent phenolic compound was the ligstroside aglycone. Certain compounds such as the closed ring carboxilade demethylated hydroxilade oleuropein form fragment, the 3,4-DHPEA-EDA diglycoside or oleuropein dihydroxytyrosol and the open ring carboxilade dialdehydic oleuropein glycoside fragment were found at very low concentrations uniformly across each of the oils. Among the first identified phenols from EVOO were phenolic acids and these were identified in a number of studies. A number of common members of this class include sinaptic acid, caffeic acid, gallic acid and vanillic acid 44 46 . It is the former of these compounds that was identified in this study.
With respect to lignans, -pinoresinol, and -1acetoxypinoresinol were identified, with all being present as the glycoside forms, and with the former being found as the free -pinoresinol form, the glycoside form and the tetrameric form. With regards free and tetrameric pinoresinol, as well as -1-acetoxypinoresinol glycoside, these were found in the largest amounts in both C and M oils, while the rest were present in similar amounts across all oils. The study by Owen et al. 11 quote the former two lignans as the major components of the phenolic EVOO fraction. It has been reported that the concentration of lignans is much higher in olive fruits when compared to olive oil 47 .
The flavonoids apigenin and luteolin were identified in this study, and were found to be present in similar amounts in all tested oils. These were first identified from Olea europaea by Rovellini et al. 48 by HPLC-UV and HPLC-electrospray-MS, and also detected in Portuguese Olea europaea leaf cultivars by Meirinhos et al. 49 by HPLC-DAD.
The greatest diversity of compounds was seen for the secoiridoid oleuropein and its derivatives, as these amounted to ten different compounds, being both the open and closed ring decarboxilade aldehydic forms, the closed    50 . These two were present in the lowest amounts in the M variety. From this study, one can conclude that out of the tested oils, the Bidni variety is the highest in TPC, TdPC and TFC. Using LC-MS analysis, over 30 phenolic compounds were identified, with these belonging to a number of different classes. The components were uniform across the oils.

Author Contributions
Lucienne Gatt performed phenolic extractions, the TPC, TdPC and TFC assays, and wrote the manuscript. Simon J. Thorpe and Lucienne Gatt performed LC-MS analysis. Frederick Lia performed compound identification. Marion Zammit-Mangion and Pierre Schembri-Wismayer supervised the work and reviewed the manuscript.

Funding Sources
This research was funded through the Malta Government Scholarship Scheme Postgraduate MGSS and through Research Funds allocated by the Faculty of Medicine and Surgery, University of Malta.