Introduction

Endocarp of olive drupe (Olea europaea L) is characterised by high hardness, therefore there is a thermal stress of olive paste during processing (crushing step) [1, 2]. This results in thermal oxidation processes involving chiefly bionutritional compounds (biophenols noticeably, which are converted to quinones). Such a phenomenon has marked negative effects on the sensory and global quality of the obtained oil [3, 4]. Also it causes a reduced formation of volatile metabolites (green aromas) from the 13-hydroperoxides (characterised by a cis-cis-1-4 pentadiene system), having linoleic (LA) or α-linolenic (α-LnA) acid as their precursor. This is the lipoxygenase (LOX) pathway, in which a series of enzymes are indeed involved. Now, one of these, such as the hydroperoxide lyase (FAHL), responsible for the cleavage of the aforementioned hydroperoxides, is partially inactivated at a temperature as low as 15 °C [5].

Another negative aspect when processing whole olives is the occurrence of high amounts of endogenous oxidoreductase enzymes in the kernel tissues (peroxidases and polyphenoloxidases noticeably) [6], even though this finding was not confirmed by Patumi et al. [7]. The first biocatalysts fix oxygen to both fatty acids and triacylglycerols, thus degrading them, whereas the polyphenoloxidases cause further oxidation of biophenols [8].

This is the reason for which we and other authors, in the last few years, have paid much attention to a non-conventional olive processing system [6, 7]. In this case, a bland-acting destoner machine replaces the violent metallic crusher. So the endocarp (hull and kernel) is removed from the fruit and there is in consequence no longer production of high amounts of thermal energy. In addition, no deleterious actions can be exerted on the oil phase by the plentiful oxidoreductase enzymes of the olive seed [9].

Such an innovative processing cycle, which is increasingly employed by oil millers [10, 11], should lead to enhance significantly the quality standard of the end product and to increase its contents of nutraceuticals. To further enhance its quality level, we have conducted preliminary processing tests using organic-destoned olives. These oils, in addition to containing high amounts of bioactive metabolites, are also characterised by no presence of pesticide residues [12].

To study the metabolome of destoned oils, the Italian Ministry for Agricultural Food and Forestry Policies approved and funded special research projects. An analytical database concerning them is now available in our Institute. We analysed destoned and organic-destoned virgin olive oils coming from all the Italian olive-growing regions. From the aforementioned database, we have extrapolated some more significant figures concerning mainly the volatile and biophenol fractions, which have been referred to in this original paper. We fully characterised the new oil kinds paying special attention to their nutraceutical metabolome as well as to their aroma and flavour composition. Our purpose was to make available highly natural and functional virgin olive oil kinds to use as aids to prevent risks related to the current human dismetabolic pathologies.

Materials and methods

Analyses of oil samples

All oil samples were stored frozen at −20 °C before analysed. Pleasant and unpleasant volatiles were stripped from the oil sample with N2 (1.2 L min−1; 37 °C; 2 h), trapped on 50 mg of activated charcoal and eluted with 1 mL of diethyl ether. Then, they were simultaneously analysed by a dynamic head-space (DHS)-high-resolution gas chromatography (HRGC) method, using Carlo Erba Mega Series 5160 gas chromatograph (Milan, Italy) fitted with a Nordion silica carbowax 20 M capillary column (50 m length; 0.32 mm i.d.; 0.5 μm film thickness; Helsinki, Finland) and equipped with an on-column injection system, a CO2 cryogenic accessory (to hold the oven temperature at 25 °C) and a flame ionisation detector (FID). The oven temperature programme was as follows: isotherm at 25 °C for 7 min, from 25 to 33 °C at 0.8 °C min−1, from 33 to 80 °C at 2.4 °C min−1, and from 80 to 155 °C at 3.7 °C min−1; final isotherm at 155 °C for 20 min. The temperature of the detector was held at 240 °C. The carrier gas was H2 at 30 kPa. The injection volume was 0.5 μL. Quantitation was achieved by peak area integration with a Carlo Erba Mega Series integrator (Milan, Italy). The internal standard was nonan-1-ol (>99% pure) that was directly added (7–8 mg) to the oil sample (50 g) [12, 13].

Extraction of the biophenolic minor polar (BMP) compounds, such as natural and oxidised derivatives of oleuropein and ligstroside, lignans, flavonoids, phenol acids and phenol alcohols, was done by a methanol/water mixture (80/20, v/v). Their quantitation was done by HPLC apparatus equipped with C18 reverse-phase column (4.6 mm × 25 cm; spherisorb ODS-2 5 μm 100 Å), with spectrophotometric UV detector at 280 nm and integrator. Elution was made by a ternary solvent system including an aqueous solution of H3PO4 (0.2%), methanol and acetonitrile. The internal standard was syringic acid. Results were expressed as mg kg−1 tyrosol [8, 9].

An analytical taste panel made up of 12 assessors performed the quantitative descriptive sensory profiling (QDSP) at the oil quality technology department (OQTD) of our Institute, according to Annex XII of Reg EEC 2568/91 [14] as modified recently by Reg EC 702/07 [15] and Reg EC 640/08 [16]. Olfactory-gustatory-tactile evaluations were made, and the results were collected on a standard profile sheet. The most remarkable sensory descriptors evaluated were fruity (green olives), cut green lawn, green leaf of twig, green olives, wild flowers, green banana, green tomato, almond, artichoke, apple, walnut husk, green hay, bitter and pungent. Shades for each sensation were also assessed. Each sensory attribute, including off-flavours, was evaluated on a ten-point scale, with intensities ranging from 0 (no perception) to 10 (extreme). All oil samples were thermostated at 30 °C before sensory analysis. By using the statistical programme attached to the sensory analysis method, the median value for each organoleptic descriptor was calculated. All other analyses were run as described in our previous works [8, 13, 17].

The olive varieties Dritta, Leccino, Coratina and Caroleo, mentioned in Table 1, were collected in Abruzzo Region, Italy; Leccino oils (Table 2) were from Tuscany Region, Italy; Nocellara del Belice oils (Table 3) were from Sicily Region, Italy; finally the blend oils, including Cerasola, Cellina di Nardò, Ogliarola Salentina, Ogliarola Barese, Coratina and Fasola oils (Table 4) were from Apulia Region, Italy.

Table 1 Composition of olive kernel oil as compared to pulp olive oil from a mix of the Italian olive cultivars Dritta, Leccino, Coratina and Caroleo
Full size table
Table 2 Contents of lignans, flavonoids, phenol acids, phenol alcohols and natural and oxidised oleuropein and ligstroside derivatives in two Leccino oils (destoned vs. stoned)
Full size table
Table 3 Composition of the aromatic volatile fraction of two organic Nocellara del Belice oils (destoned vs. stoned) coming from Sicily region (Italy)
Full size table
Table 4 Values of other analytical variables referring to samples of a virgin olive oil blend made up of six oil varieties (Cerasola 40%, Cellina di Nardò 12%, Ogliarola Salentina 12%, Ogliarola Barese 12%, Coratina 12%, and Fasola 12%), collected in the crop years 2005–2007 (a sample for each year)
Full size table

All olive varieties were harvested in November. Olive processing was carried out by continuous two-phase centrifuging systems, having the crusher machine replaced by a destoner. Olive paste kneading was carried out at 30 °C and lasted about 60 min. A maximum amount of 10 L q−1 olives was added to the olive paste before their centrifugation.

Statistics

The experimental protocol included either simple classification designs (Tables 1, 2, 3) or a two-factor scheme (Table 4). In the first case, the oil analyses were run in triplicate (three independent samples), and in the second case one oil sample was taken for each year. Univariate treatments of the data were performed by analysis of variance (ANOVA). When a significant F value was found, means were separated using Tukey’s honestly significant difference (HSD). Principal component analysis (PCA), hierarchical cluster analysis (HCA) and soft independent modelling of class analogy were applied for multivariate treatments of the data. The statistical software packages Statistica® (Release 6.0 Statsoft Inc., Tulsa, OK, USA), Minitab® (Release 15.0, Minitab Inc., State College, PA, USA), SPSS® (Release 12.0, SPSS Inc., Chicago, IL, USA) and Stata® (Release 9.0, Stata Corp., College Station, TX, USA) were used.

Principal component analysis, also referred as the eigenanalysis method, is based on the maximum variance criterion and calculates orthogonal linear combinations (principal component scores). HCA produces a hierarchy of partition of objects such that any cluster of a partition is fully included in one of the clusters of the later partitions. Such partitions are best represented by a dendrogram (binary tree). SIMCA is a biased version of discriminant analysis (DA). Instead of calculating unbiased class covariance matrices, each class is represented by principal component model. An object is classified according to its distance (residuals) from these models.

Results and discussion

Table 1 shows the composition of olive kernel oil as compared to that of olive pulp oil. The former appeared to be richer in PUFA (polyunsaturated fatty acids), phytosterols, long-chain aliphatic alcohols and methylsterols. Table 2 suggests that destoning significantly increased the concentration of biophenols in the pulp oil, such as lignans, flavonoids, phenol acids, phenol alcohols and natural and oxidised oleuropein and ligstroside aglycons. This phenomenon was to be ascribed to the use of a destoner machine in the place of the mechanical crusher. The former, in fact, significantly reduced the thermal degradation (quinonisation) of these natural, antioxidant and flavouring compounds [9, 12]. Removal (by destoning) of the kernel (seed), contained inside the stone, which is rich in polyphenoloxidase and peroxidase enzymes, was likely another factor meaningfully contributing to reduce the oxidation processes of the biophenols [6]. In consequence, destoned products were more stable and resistant to autoxidation and displayed longer shelf-life. These new crop oil kinds, owing to the soft action of destoner, also were richer in α- and γ-tocopherol and in α- and γ-tocotrienol. The last are considered to be other very bioactive compounds. They had (Table 3) higher concentrations of green volatiles from the lipoxygenase cascade (C6 and C5 compounds noticeably). Among these, unsaturated aldehydes (trans-2-hexenal noticeably) were major metabolites [5]. Such a positive circumstance is likewise attributable to the bland action of the destoner machine, which produced little thermal energy amount and thus did not substantially inactive the hydroperoxide lyase (FAHL) enzyme. In consequence, destoned oils were scored more, had more marked hedonistic properties, and were more accepted and preferred by the panelists [7, 12]. It is a well-known circumstance that other olive processing steps (crushing and malaxing noticeably) can enhance the concentration of biophenol and volatile compounds in the oil phase [12, 13], but we believe that destoning is more effective compared to them [8].

Concerning colour, they showed balanced concentrations of chloroplast pigments (chlorophylls, carotenes and xanthophylls). Concentrations of these compounds were, however, lower compared to those of the reference oils. This because destoner exerts, with respect to metal crusher, lower effects on the fruit hypoderm tissue, were both chloroplast and chromoplast colourings are essentially located [9, 16]. Such a finding was corroborated by lower values of both chroma (σ %) and integral colour index (Table 4).

Other analytical variables referred to in Table 4 were little or not affected by destoning [18–20]. Organic-destoned oils showed highest contents of bioactive compounds and appeared to be the best functional produces to use to prevent mainly cancer and cardiovascular disease risks. Organic-stoned oils were comparable to non-organic destoned ones for their richness in bionutritional factors (Table 4). From one-way analysis of variance (ANOVA), which do not separate the variance attributable to interactions, emerged that the variance accounted for by the destoning factor was higher than that ascribable to the climatic seasonal conditions.

Destoned and organic-destoned products were easily retraceable by chemometrics, a circumstance this to be stressed since the related Reg EC 178/02 is being in-force from January 1st 2005. The PCA multivariate technique was a very effective discriminating method of the oil kinds (Fig. 1). In fact, along the first principal component (accounting for 68.5% of total variance) were discriminated whole olive oil (WO) samples (positive half) and organic-destoned olive oil samples (ODO; negative half), whereas along the second dimension (accounting for 17.8% of variance) were differentiated (negative half) whole organic oil samples (WOO) and destoned oil samples (DO). The PCA biplot (3D) in Fig. 2 confirmed how this multivariate procedure was capable of differentiating the above oil kinds (WO, WOO, DO, ODO) along the first and second eigenvectors. These accounted for 70.3% and 15, 5% of total variance, respectively. The dendrogram in Fig. 3 showed how the HCA multivariate procedure was likewise effective in discriminating between oil kinds. In fact, this plot showed four well-differentiated clusters corresponding to the investigated oil types. These were finally classified by the SIMCA procedure (Fig. 4).

Fig. 1
figure 1

PCA biplot obtained by processing a set of analytical data related to non-glyceryl-containing compounds, showing clear discrimination of the oil kinds, such as WO whole olive oil, DO destoned olive oil, WOO whole organic olive oil, ODO destoned organic olive oil

Full size image
Fig. 2
figure 2

PCA score plot (3D) obtained by processing a set of analytical data including both saponifiable and unsaponifiable figures, showing differentiation of the oil kinds. Abbreviations as in Fig. 1

Full size image
Fig. 3
figure 3

Dendrogram obtained by processing the fatty acid composition data, showing discrimination of the oil kinds. Abbreviations as in Fig. 1

Full size image
Fig. 4
figure 4

SIMCA plot obtained by processing a dataset related to the pigment and chromatic variables, showing differentiation of the oil kinds. Abbreviations as in Fig. 1

Full size image

Conclusion

The metabolomic study carried out led to the conclusion that destoning had no influence on the major bioactive lipidome of the resulting oils, but it meaningfully affected the nutraceutical metabolome making up their unsaponifiable fraction. Destoned and above all ODO appeared to be highly functional produces (richest in nutraceuticals and bionutritional phytomolecules). Hence, they could represent a reliable aid in preventing risks related to several human disorders, such as oxidative stress, dismetabolic syndrome, early aging, platelet aggregation and oxidation of LDL; yet insulin resistance, arteriosclerosis, cancer types and other current pathologies. They are furthermore competitive and high added value produces and could therefore be an important factor in enhancing the olive-growing income.