2.2.1. Method Development and Optimization

Avocado oils are characterized by a high content of TGs. The separation and unambiguous identification of structurally similar TGs in avocado oil pose great analytical challenges. In this study, 12 commercially available TG reference standards were purchase (Table 2) and used for method development and quantification. The method was optimized regarding the chromatographic column, eluent and gradient program. Different reversed-phase columns including several Agilent ZORBAX columns such as Eclipse Plus C18, SB-C18, XDB-C18, SB-C8 (with the same dimensions of 2.1 × 100 mm × 1.8 μm), and ACQUITY UPLC BEH C18 (Waters, 2.1 × 100 mm × 1.7 μm) as a standalone column or in a combination were investigated. Finally, three ACQUITY UPLC BEH C18 connected in series were used to provide the best separation of the targeted TGs. For the evaluation of eluents, unlike non-aqueous eluents used by the majority of TGs analyses [21,29–31], eluent consisting of acetonitrile with 0.1% water (v/v) and isopropanol with 5 mM ammonium formate was used, and the chromatographic

peak shapes were greatly improved. The chromatograms showing the method optimization using a mixed solution containing LLO, LLP, OLO, PLO, PPoO, OOO, OOP and PPO standards are illustrated in Figure 1. The chromatograms of authenticated oils extracted from avocado peel, pulp and seed, as well as one of the commercial products claimed as avocado seed oil, are shown in Figure 2. It is worth noting, as the ECNs increased from 44 to 50, the retention times for the corresponding TGs also increased (Figure 2). Although similar TG profiles of avocado peel and pulp oils were observed, a significant difference was present in avocado seed oil. The profile for the commercial avocado seed oil appeared to be dramatically different with any of the other avocado oils, suggesting that this particular oil was possibly adulterated with other oils.


**Table 2.** Information of triglyceride reference standards.

ECN: equivalent carbon number, calculated as ECN = CN (number of carbon atoms) – 2DB (double bonds).

#### 2.2.2. Identification of TGs

Unambiguous identification of complex TGs in avocado oil is desirable for the accurate quantification of TGs. Many HPLC detection techniques including refractive index [32], UV-Vis and evaporative light-scattering (ELSD) [21], have been applied for the qualitative analysis of TGs in plant oils. Although each of these detection methods has its own advantages, it may not be possible to confidently identify TGs with complete or even partial chromatographic resolution in complex plant oils that contain numerous species with the same ECNs. In our study, both APCI and ESI with positive/negative ion modes were evaluated, and ESI(+) provided better sensitivity for all the TGs with the optimized solvent system (Figure 1D). The ESI(+) mass spectra and notation of fragment ions for representative TGs are illustrated in Figure 3. As described before, TGs in plant oils usually exist as a mixture of positional isomers differing by the acyl attachment, sn-1, sn-2 and sn-3. The [M + H]<sup>+</sup> and [M + NH4] <sup>+</sup> ions were detected in all TGs with relatively low abundances compared to the corresponding [M + H-RCOOH]<sup>+</sup> ions. Except for these two ions, single-acid type (R1R1R1) provided only one ion such as [OO]<sup>+</sup> in OOO. Mixed-acid type (R1R2R1 or R1R1R2) always provided two different ions, such as [LL]<sup>+</sup> and [LO]<sup>+</sup> for both LLO and LOL types TGs. Conversely, three different ions were observed for mix-acid type (R1R2R3), such as [LP]<sup>+</sup>, [LO]<sup>+</sup> and [OP]<sup>+</sup> for OLP and OPL. All the examples are shown in Figure 3. Although the theoretical ion abundance ratios of [LO]+/[LL]<sup>+</sup> should be 2:1 for both LLO and LOL, different values (1.2 for LLO and 3.0 for LOL, respectively) were observed. A similar observation was achieved for OLP and OPL. The theoretical ion abundance ratios of [LP]+/[OP]+/[LO]<sup>+</sup> should be 1:1:1 for both OLP and OPL, whereas the measured values were 0.62:0.38:1 for OLP, and 1.8:2.1:1 for OPL. This observation indicated that the loss of FA from the equivalent sn-1 and sn-3 positions is preferred rather than the middle position sn-2. Figure 4 proposed the possible mechanism for the cleavage of fatty acids from TGs at different positions. Based on observed fragments, the formation of a 5-member ring as a result of losing FA group in the sn-2 position is less favorable than the formation of a more stable 6-member ring in position sn-1 or sn-3 [33]. The relative abundances of ions formed from the loss of FAs in different positions in the MS afford the confident identification of the acyl position on the glycerol backbone.

**Figure 1.** Total ion chromatograms of (**A**) APCI(+), eluent: acetonitrile/isopropanol; (**B**) ESI(+), eluent: acetonitrile/isopropanol; (**C**) ESI(+), eluent: acetonitrile with 0.05% formic acid/isopropanol with 0.05% formic acid and 5 mM ammonium formate; (**D**) ESI(+), eluent: acetonitrile with 0.1% water and 0.05% formic acid/isopropanol with 0.05% formic acid and 5 mM ammonium formate.

**Figure 2.** Total ion chromatograms of authenticated avocado oils extracted from pulp, peel, seed and commercial avocado seed oil. ECNs: the equivalent carbon numbers.

An alternative approach for the identification of TGs in avocado oils is to measure the ion survival yield (ISY). As demonstrated in Figure 3, various types of ions were formed by in-source collision-induced dissociation (IS-CID). The production of information-rich fragments by IS-CID can materially aid in the identification of components, elucidation of structures, and distinction between isomers and chemically similar components in complex mixtures [34]. Ion distribution in IS-CID has been commonly measured by ISY defined as *ISY* = *Ia Ia*+ *Ib* where *Ia* represents the measured intensity of the monitored ion and *Ibs* are the intensities of the additional ions formed in the source. Accurate qualitative and quantitative analysis requires that the ISYs of reference standards should be independent of the concentrations of standards, and the ISYs should be equivalent to the standards and analyzed samples. The ISYs for the representative TG standards were measured over the range of 5–400 μg/mL (Figure 5). The ISYs for the fragment ion of [LO]+ after the loss of one FA from different positions (sn-1, sn-2 or sn-3) were calculated from the representative TGs. For example, LLO and LOL

both displayed a pseudo-molecular ion [M + H]<sup>+</sup> at 881 with a fragment ion [LO]<sup>+</sup> at 601. However, the ISY for [LO]<sup>+</sup> generated from LLO was 0.38 (±0.01), whereas that from LOL was 0.50 (±0.01). Similarly, ISYs for [LO]<sup>+</sup> from PLO and OPL were 0.43 (±0.01) and 0.21 (±0.01), respectively, suggesting that the measured ISYs can be used for the identification and characterization of TGs in complex samples [35].

**Figure 3.** Positive ion ESI spectra of TGs containing different acyls on the glycerol backbone. OOO, LLO and LOL, and OLP and OPL were used as representative examples for single-acid type (R1R1R1), mixed-acid type (R1R2R1 or R1R1R2) and mixed-acid type (R1R2R3), respectively.

**Figure 4.** Schematic representation of fatty acids elimination from sn-1, sn-2 and sn-3 positions by ESI(+) MS.

**Figure 5.** Ion survival yields (ISYs) for the fragment ion [OL]<sup>+</sup> from representative TGs over the concentration range.
