Direct analysis of olive oil and other vegetable oils by mass

8 Miriam Beneito-Cambra, David Moreno-González, Juan F. García-Reyes, Marcos 9 Bouza, Bienvenida Gilbert-López, and Antonio Molina-Díaz 10 11 Analytical Chemistry Research group, Department of Physical and Analytical Chemistry, 12 University of Jaén, 23071 Jaén, Spain 13 2 Center for Advanced Studies in Olives Grove and Olive Oils (CEAOAO), Science and 14 Technology Park GEOLIT, E-23620 Mengíbar, Spain 15 3 ISAS—Leibniz Institut für Analytische Wissenschaften, Bunsen-Kirchhoff-Str. 11, 44139 16 Dortmund, Germany 17 18 19 20 21 Corresponding author: Professor Antonio Molina-Díaz, Analytical Chemistry Research 22 Group, Department of Physical and Analytical Chemistry, University of Jaén, 23071 Jaén, Spain. 23 Center for Advanced Studies in Olives Grove and Olive Oils (CEAOAO), Science and 24 Technology Park GEOLIT, E-23620, Mengíbar, Spain.Tel.: (+34) 953 212147; Fax: (+34) 953 25 212940. E-mail: amolina@ujaen.es 26


Introduction
1 Virgin olive oil (VOO) is probably the most representative and iconic component of the 2 Mediterranean diet, being highly appreciated for its nutritional value and beneficial role 3 on the prevention of cardiovascular diseases [1][2][3][4]. VOO is a natural juice extracted from 4 Olea europaea L. fruits, exclusively using mechanical methods [5]. No industrial 5 processes or chemicals are added, thus providing an extra value reflected in its higher 6 price compared to most edible oils. For this reason, the assessment of VOO quality, safety 7 and authenticity is of the utmost importance from both economic and health perspectives. 8 Considerable effort has been put during the last decades in this direction [6][7][8][9].  Table 1 includes the main analytical methods used officially for VOO authentication 14 purposes [11]. Most of the methods proposed by the IOC rely on the use of gas 15 chromatography (GC) and high-performance liquid chromatography with nonspecific 16 detectors such as flame ionization detector (GC) or refraction index (HPLC). 17 Nevertheless, despite the continuous effort towards the development of new tools for 18 VOO authentication purposes, appropriate solutions for some specific issues have not 19 been found yet [13]. Current gaps include the detection of selected blends of VOO with 20 other lower quality vegetable oils (VOs) -including refined olive oils (OOs)-, the specific 21 detection of soft deodorized (refined) OOs used to adulterate VOO, or the verification of 22 the geographical origin of VOOs. Thus, the development of analytical solutions 23 improving the detection of common and emerging frauds in the olive oil sector is of the 24 utmost importance. 25 VOO testing is not an easy task since it is a complex matrix mainly constituted by 26 triacylglycerides (TAGs), and other components such as fatty acids (FAs), hydrocarbons, 27 sterols, phenolic compounds or fatty alcohols. Depending on the scope of the analysis, 28 different approaches have been attempted such as fingerprinting, profiling, or targeted 29 analysis, each of those seeking a different range of compounds, from the whole sample 30 to specific compound-class fractions [15]. Fingerprint analysis involves the use of total 31 data obtained without single compound identification. On the other hand, profile analysis 32 comprises the measurement of a previously defined class of analytes. Finally, the targeted 1 analysis implicates determination of a single or small set of target analytes. 2 Analytical methods used for the characterization of edible oils for quality control and 3 authentication purposes have been previously studied [13][14][15][16][17][18]. Faria et al. [14] classified 4 them according to the technique used, shorting them out as chemical (chromatographic, 5 spectroscopic/spectrometric) or biological methodologies (DNA-based techniques). Gas 6 chromatography (GC) and liquid chromatography (LC) have been extensively used for 7 the determination of both major and minor compounds in edible oils as well as for 8 authentication, traceability and quality control purposes [17]. GC is very convenient for 9 several assays such as the determination of fatty acids using a derivatization step or for 10 volatile analysis. On the other hand, LC is useful for most non-volatile components 11 including TAGs, free fatty acids (FFAs) or phenolic species. As an alternative, 12 spectroscopic techniques (NMR, atomic absorption, ICP-MS, IR (MIR, NIR) and Raman 13 spectroscopy or molecular fluorescence) are used in the field of edible oil with different 14 purposes such as the determination of moisture content, total fat content, free fatty acid 15 content, oxidation indexes and authentication/fraud detection. Finally, DNA-based 16 methods (biological methods) can provide information about the cultivar (botanical) 17 identity. 18 Regardless the technique used, one of the key features of edible oil chemical methods is 19 whether the analyses are performed directly in the oil matrix without further treatment or 20 on the other hand, sample preparation and processing -including dilution steps or 21 dedicated sample preparation stages-are mandatory. Given the complexity of the matrix, 22 the first approach is difficult to tackle with. Another feature is the scope of the analysis, 23 whether the entire fat fraction composition (eg. saponifiable fraction) is analyzed or, on 24 the contrary, specific fractions that may contain useful molecular information for quality 25 control and authentication purposes are targeted. 26 Mass spectrometry (MS) offers unique features that map well against the challenge of weaknesses of the different approaches. As the ionization method used in mass 10 spectrometry determines the compound classes sought in each type of analysis, the article 11 is organized following this criterion. Amongst the methods discussed we should include 12 direct infusion atmospheric pressure ionization using either ESI, APCI and atmospheric 13 pressure photoionization (APPI), ambient desorption/ionization mass spectrometry 14 methods using ESI or APCI-like ionization mechanisms, and also vacuum methods such 15 as MALDI, together with specific methods for volatiles such as HS-MS and proton 16 transfer reaction mass spectrometry (PTR-MS). 17 [ Table 1] 18 19 pressure ionization sources 20 The development of novel analytical methodologies that lead to improved assays, 21 involving minimal sample preparation and reduced reagent consumption, high throughput 22 and enhanced automation is demanded [24][25][26]. Direct infusion mass spectrometry is a 23 technique which offers fast and reproducible analysis avoiding chromatographic 24 separation and providing expedite data acquisition using mass spectrometry with 25 atmospheric pressure ionization sources such as ESI, APCI and atmospheric pressure 26 photoionization (APPI). This approach has been proposed for direct analysis of edible 27 oils (Table 1)  . Studies can be roughly classified according to the compound class 28 targeted, either if methods are focused on main oil components (eg. FAs) or TAGs) or, 29 on the contrary on minor species present at low concentration levels (eg. phenolic 30 compounds, tocopherols, …). Different groups/compound families have been studied 31 (FAs, phenolic compounds, amino acids, sterols, tocopherols, TAGs, etc.) in vegetable 32 oils (and also, in animal origin oils). The studies were focused on the potential of the technique to evaluate the quality, the oxidation status, adulterations, identification of 1 olive oil commercial classes, classification of botanical varieties or geographic origin for 2 authentication purposes. Sample treatment is needed in most cases. It usually involves 3 the implementation of high dilutions or a liquid-liquid extraction. A summary of the main 4 aspects of these studies is shown in Table 2 .  Olive oils. Different studies have been performed in order to evaluate OO quality and 9 also to discriminate botanical varieties or assess the geographical origin. They are mostly 10 based on the profile and relative abundances of FFAs in OOs (including EVOOs) by 11 means of direct infusion electrospray tandem mass spectrometry using ion traps as mass 12 analyzer (ESI-MS/MS (IT) [27][28][29][30]. Thus, peak abundances corresponding to FFAs were 13 employed as variables to perform linear discriminant analysis (LDA) capable to predict 14 OO commercial quality grade according to European Union standards [28]; and FA 15 profile followed by statistical treatment allowed discrimination between different 16 botanical varieties of OOs [27][28][29]. For all these studies, a simple oil dilution in a basic 17 alcoholic mixture was performed.  The bulk mass spectra from oil polar fraction obtained by ESI-MS/MS were used by 30 Alves et al. [31] to develop a method for adulteration detection of EVOO, and to 31 discriminate between different olive oil (OO) grades (EVOO and ordinary quality OO). 32 MS data were subjected to two exploratory statistical approaches, Principal Component 33 Analysis (PCA) and Hierarchical Clustering Analysis (HCA) [31], or Partial Least The determination of phenols in olive oil samples usually involves an extraction to 6 separate them from the fatty matrix. Electrospray ionization (ESI) was used in the 7 negative ion mode by Lara-Ortega et al. [32] to study the phenolic compounds profile of 8 three different categories of olive oil (EVOO, VOO and lampante OO) from 9 hydroalcoholic extracts (diluted 1:10). Although a fairly similar pattern was observed for 10 the three classes, significant differences in peak distribution, and intensities was noticed. Ultra-high-resolution mass spectrometry using electrospray ionization Fourier transform 31 ion cyclotron resonance mass spectrometry (ESI-FT-ICRMS) enabling resolving power 32 above 300.000 (FWHM) was proposed by Marshall and co-workers [42] to unravel the 33 complexity of different vegetable oils. This enhanced selectivity not only enabled a 34 thorough profiling of TAGs and DAGs but also minor components such as tocopherols 1 were distinctly detected. The authors envisaged the ability of this approach for the 2 detection of adulterations. Follow-up studies using the same approach were proposed by 3 Li et al. [43] using free fatty acids to reveal key differences in the molecular compositions 4 of the various vegetable oils tested.

6
Additional studies of OO adulteration by means high resolution mass spectrometry, 7 although using benchtop Q-TOF instruments with lower resolving power, were 8 performed by Goodacre et al. [39] and Gómez-Ariza et al. [40], using simple oil dilution 9 (1000-fold) before direct infusion HRMS. Oil fingerprints along to PCA led to promising 10 results, showing possible discrimination between OO and other oils frequently used as 11 adulterants, including refined hazelnut oil (HO) [40]. Likewise, Gómez-Ariza et al. [41] 12 used the triacylglycerol profiles obtained by direct infusion HRMS using both ESI and 13 APPI sources with the same authentication purpose. Despite, the ESI spectra of both OO 14 and HO shared many features -such as the main peak attributed to triolein ammonium 15 adduct ion (m/z 903)-, the peak corresponding to trilinoleoylglycerol (m/z 897) was only 16 present in HO, being possible its use as OO adulteration marker among others. In 17 addition, the use of both sources (ESI and APPI) led to complementary results, since 18 MAGs and DAGs detection is more sensitive using APPI source, whilst ESI source was 19 more effective for TAGs detection. Finally, statistical treatment by PCA and LDA of 20 triacylglycerol peaks abundances showed the ability to detect the presence of other 21 adulterant oils in OO.

23
The use of ion mobility spectrometry adds an additional dimension to m/z separation, 24 thus, representing an interesting alternative given the complexity of the studied samples.

25
Arce and co-workers [44] have recently proposed the combined use of electrospray 26 ionization, differential mobility analysis (DMA) -a class of ion mobility spectrometry 27 (IMS)-and mass spectrometry for chemical fingerprinting of olive oils for authentication 28 purposes (ESI-DMA-MS). Two different approaches were tested: (i) sample dilution and;

29
(ii) liquid-liquid extraction with MeOH/water to include mainly the fraction of polar 30 compounds. To examine the feasibility of the approach, thirty samples were tested using 31 PCA and orthogonal PLS-DA. The second approach were found more effective (89%) -32 than direct dilution (67%)-to classify between EVOO, VOO and lampante olive oil 33 samples, being the combined information leading to correct classification of all the samples. The results show that ESI-DMA-MS can become an effective tool for olive oil 1 sector, although a more comprehensive study is needed.   Despite most of the work has been focused to main fat components, the use of direct 29 infusion mass spectrometry after dedicated sample workup has been also proposed to 30 detect key trace compounds in olive oil and other vegetable oils. Sindona and co-workers 31 proposed a method for the quantitation of oleuropein, a key phenolic component in 32 EVOO, using LLE fractionation and ESI-MS/MS [50]. The same group also proposed a 33 method for the quantitation of rotenone, an insecticide, using APCI-MS/MS after a C18 column cleanup step [51]. Finally, Marina and co-workers have proposed a method for 1 seed oil adulteration in EVOOs based on the detection of five nonprotein amino acids and 2 three betaines using flow injection ESI-MS/MS, enabling the detection of as low as 2% 3 (w/w) seed oil in EVOO, being ornithine the key marker to identify that adulteration [52].         Lara-Ortega et al. [32] used this technique to obtain TAGs and DAGs profiles in three 20 commercial categories of OO without any sample treatment except dilution (1000-fold).

21
Main peaks were assigned to TAGs and DAGs as ammonium adducts, achieving signal 22 enhancement by silver adduct ion formation [32]. On the other hand, Sindona and co-  consists of a tube divided into several chambers through which a gas (typically He or N2) 15 flows. This gas is introduced into a discharge chamber containing a cathode and an anode, 16 where a DC potential of several kilovolts is applied, initiating an electrical discharge    for the study of hogwash and edible oils samples. Samples were placed into a glass vial, 32 and directly impinged by a nitrogen gas stream, leading analyte(s) desorption, which were 33 then transported to the DBDI source using a sample transfer line. FFAs were decisive markers to discriminate between hogwash and qualified edible oils samples using PCA.  Lara-Ortega et al. [32] used LTP to study phenolic profiles from olive oil hydroalcoholic 11 extracts, and triacylglycerol profile from raw olive oil samples. In this work, the 12 performance of LTP was also compared to paper spray for the direct olive oil analysis for 13 quality control and authentication purposes. Both approaches allowed the analysis of 14 olive oil without (raw oil) or after a simple dilution. Interestingly, significant differences 15 were found in the information that can be extracted from both LTP and PS-MS. Above a 16 value of m/z 500, scarcely any olive oil compound was efficiently desorbed using LTP. 17 This fact involves the loss of most of the information related to intact TAGs along with 18 possible MAGs and DAGs. Therefore, paper spray in this aspect outperformed LTP, as a          standard stainless-steel target plate [131]. Silver trifluoroacetate was used to yield silver 13 adducts ions of squalene and related squalene oxide forms in the positive ion mode. The 14 same approach was also tested for the characterization of olive and sunflower oil before 15 and after thermally assisted oxidation, targeting in these cases higher molecular weight   31 The Volatolome of EVOO samples is very rich, and provide useful molecular information 32 for sensory analysis, although the species are not stable over the entire shelf life of the 1 sample, which reduces its analytical usefulness for authentication purposes. One of the 2 methods which provides fast response and selective data due to mass spectrometry is the    spectrometer. The rich molecular information gathered from the variety of methods 3 discussed allow different applications ranging from quality control, detection of 4 adulteration, assessment of geographical and/or botanical origin, and even the 5 classification according to commercial olive oil categories. In summary, although MS is 6 not yet included in official olive oil quality control and authentication methods, it will 7 undoubtedly come into play -either alone of combined with separation techniques-and 8 become a gold standard as it did in many other disciplines such as pesticide testing or 9 antidoping control. There are too many advantages not to benefit from for such as 10 challenging task VOO authentication represents. The authors acknowledge funding from Consejería de Economía, Conocimiento, 14 Empresas y Universidad, Regional Government of Andalucía, Spain (Project Ref.             and environment of molecules present in bulk liquid, liquid-gas interface, and headspace. 7

MALDI and other laser-based methods
The temperature of the heated column is 80  C. Adapted from Ref.
[60] with permission. 8     Table 2. Edible oil analysis using direct infusion mass spectrometry using ionization methods at atmospheric pressure .      Despite the intrinsic instability of phenolic compounds as antioxidants, different reports have shown the usefulness of phenolic profiles for authentication purposes, particularly for geographical and botanical origin assessment ESI-based methods in the negative ionization mode generated relatively simple mass spectra Phenolic compounds degrade easily so they are not stable over the shelf life of the oil Headspace sampling reduces dramatically the impact of the oil matrix on the actual MS measurements. No carryover effects are expected so that large runs of samples can be acquired without major instrument maintenance operations.
The use of soft ionization (PTR-MS or SIFT-MS) produces simpler mass spectra (than EI) providing more useful datasets with less interfered m/z values and less redundant signal, making data processing more effective and enabling better classification of samples