Standards of apigenin, luteolin, hydroxytyrosol, tyrosol, vanillin and p-coumaric, ferulic, and vanillic acids were purchased from Sigma-Aldrich (St. Louis, MO) and (+)-pinoresinol was purchased from Arbo-Nova (Turku, Finland). Stock solutions of phenolic compounds were prepared in methanol and stored at −20 °C. All of the solvents used were of analytical or HPLC grade (Sigma-Aldrich). The water was of Milli-Q quality (Millipore Corp, Bedford, MA, USA).

Sicilian olive samples from eight different cultivars (Biancolilla, Passulunara, Tonda iblea, Nocellara del belice, Nocellara del belice Organic and Nocellara etnea from three different altitudes) were provided by Sicilian oil producers. Olives were hand-picked at the same stage of maturity and the same laboratory mill was used to prepare olive oil samples. Virgin olive oil samples were obtained using an Abencor analyzer (Abengoa S.A., Seville, Spain). This system reproduces the industrial process at a laboratory scale and consists of three basic elements: hammer mill, thermo-beater, and pulp centrifuge (Martínez et al., 1975). The milling of olive fruits was performed using a stainless-steel hammer mill operating at 3000 rpm equipped with a 5 mm sieve. The resulting olive paste was immediately kneaded in a mixer at 50 rpm for 30 min at 30 °C. Centrifugation of the kneaded olive paste was performed in a basket centrifuge at 3500 rpm for 1 min. After centrifugation, the oil obtained was decanted and stored in amber glass bottles at 4 °C in the dark and without headspace until analysis.

To isolate the phenolic fraction of olive oils, the method proposed by the International Olive Oil Council (IOOC, 2009) was used. Briefly, the analytical methodology consists of olive oil extraction with methanol/water (80/20), an ultrasonic bath for 15 min at ambient temperature and centrifugation at 5000 rpm for 25 min. After that, an aliquot of the supernatant phase is filtered through a 5 mL plastic syringe using a Millex^®-HV PVDF 0.45 μm filter (Millipore Corp, Billerica, MA, USA). The extractions were repeated three times, and phenolic extracts were stored at −20 °C until further analysis. The total phenolic content was measured by HPLC using a UV-Vis detector in accordance with IOOC's methodology (IOOC, 2009). Syringic acid was used as the internal standard. HPLC analyses were performed with a Varian ProStar HPLC unit (Walnut Creek, CA, USA) equipped with a Varian 230 UV-Vis detector. The total phenolic content–expressed in mg·kg⁻¹ of tyrosol–was calculated by measuring the sum of the areas of the related chromatographic peaks, according to the method proposed by the IOOC. Secoiridoid compounds were quantified at 280 nm using the response factors determined by Mateos et al. (2011).

When standards were available (apigenin, luteolin, hydroxytyrosol, tyrosol, vanillin, (+)-pinoresinol and p-coumaric, ferulic, and vanillic acids), ultra performance liquid chromatography coupled to time-of-flight mass spectrometry analysis (UPLC-TOF-MS) were performed to detect and quantify individual phenolic compounds in phenolic extracts. The UPLC system consisted of an AcQuity ^TM UPLC equipped with a binary pump system (Waters, Milford, MA, USA) and an AcQuity UPLC ^TM BEH C18 column (1.7 μm, 100 mm×2.1 mm i.d.) from Waters. The UPLC-TOF-MS methodology was described elsewhere (Jiménez et al., 2013; Rivas et al., 2013). The identification of phenolic compounds was carried out by comparing both the retention times and MS spectra from the olive oil samples and standards when available.

The human osteosarcoma cell line MG-63 was purchased from American Type Cultures Collection (ATCC, Manassas, VA) and maintained in Dulbecco's Modified Eagle Medium (DMEM; Invitrogen Gibco Cell Culture Products, Carlsbad, CA) with 100 IU/ml penicillin (Lab Roger SA, Barcelona, Spain), 50 μg·mL⁻¹ gentamicin (Braum Medical SA, Jaen, Spain), 2.5 μg·mL⁻¹ amphotericin B (Sigma, St Louis, MO, USA), 1% glutamine (Sigma, St Louis, MO, USA), and 2% HEPES (Sigma, St Louis, MO, USA) and was supplemented with 10% foetal bovine serum (FBS) (Gibco, Paisley, UK). Cultures were kept at 37 °C in a humidified atmosphere of 95% air and 5% CO₂. Cells were detached from the culture flask with a solution of 0.05% Trypsin (Sigma, St Louis, MO, USA) and 0.02% ethylenediaminetetraacetic acid (EDTA) (Sigma, St Louis, MO, USA) and were then washed and suspended in a complete culture medium with 10% FBS. Prior to the beginning of each experiment, all of the cells were grown in oestrogen-free media (DMEM without red phenol) for at least 24 hours.

Cell proliferation was determined using the MTT method (Sigma-Aldrich Chemie). This is a colorimetric method that measures the chemical reduction of MTT (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide) into formazan, which is directly proportional to the number of viable cells in the tested sample. Osteoblasts were seeded at 1×10⁴ cells·mL⁻¹ per well into a 96-well plate (Falcon Becton Dickinson Labware, New Jersey) in an oestrogen-free culture medium without FBS and cultured at 37 °C in a humidified atmosphere of 95% air and 5% CO₂ for 24 hours. Then, the media was replaced with DMEM containing different concentrations of prepared phenolic extracts dissolved in a fresh culture medium at concentrations of 0.001, 0.0001 and 0.00001%. All of the experiments included cells incubated under the same conditions without treatment extracts and cells incubated with 0.001% methanol as an internal control. The highest solvent concentration in culture media (0.001%v/v methanol) had no significant effects on cell proliferation. Upon completion of the treatment, the media was replaced with DMEM, without phenol-red, containing 0.5 mg·mL⁻¹.

MTT (Sigma, St Louis, MO, USA) and was incubated for 4 hours. Cellular reduction of the tetrazolium ring of MTT resulted in the formation of formazan crystals in the form of dark-purple, water-insoluble deposits. After incubation, the media was aspired and DMSO was added to dissolve the formazan crystals. Absorbance was measured at 570 nm with a spectrophotometer (Sunrise TM, Tecan, Männedorf, Switzerland). Cell proliferation percentages (%>control) were calculated compared to cell cultures treated with methanol alone (controls).

A statistical analysis was performed using SPSS (v. 20.0, Statistical Package for the Social Sciences, Chicago, IL). Numerical data on cell proliferation are expressed as the mean of three independent experiments. Statistically significant differences in the effects on proliferation between the extracts and the control were analyzed by the Student's t-test. The results for phenolic compounds are the average of at least three repetitions. A two-way analysis of variance (ANOVA) was performed to examine the proliferation by considering the concentration, the olive oil variety and the interaction between these two factors. A one-way ANOVA was used to examine phenolic compound concentrations by considering the olive oil variety. The homogeneity between the olive oil variety samples was tested using the Levene's test. Differences between the means were evaluated using the Games-Howell and Bonferroni tests. The significance level in all of the statistical analyses was set at 0.05. The level of significance of differences between the phenolic compound mean values was determined at 5% by a one-way ANOVA using Tukey's test. Principal components analysis (PCA) was applied to the chemical and proliferation data. PCAs were carried out for all raw data processed with a cross-validation method. The data were auto-scaled before analysis.

Phenolic compounds are an essential component of VOO quality. The most important phenolic compounds that have been identified in VOO are divided into different groups, such as phenolic acids, phenol alcohols, secoiridoids, lignans, and flavones (Bendini et al., 2007). In this study, the concentrations of phenolic compounds were determined in olive oil samples that were collected from different Sicilian olive tree cultivars. Table 1 shows the mean values for the total and individual phenolic compounds detected in oils.

The total phenolic content of Sicilian oils was measured by HPLC using a UV-Vis detector in accordance with IOOC's methodology (2009). The differences found in the phenol content among the diverse VOO samples are as expected, as the phenolic content of virgin olive oil is influenced by the type of cultivar, production method and altitude. The oil sample obtained from the Nocellara etnea variety, which is produced at an altitude of 100 m, was the sample with the highest amount of phenolic compounds, while the samples obtained from the Tonda iblea variety contained the lowest. The total phenolic compound contents in the varieties examined in this study were higher than those measured in other studies using spectrophotometry methods (Baccouri et al., 2007; Dugo et al., 2004; Patumi et al., 2003) but were similar to those measured using chromatographic analysis (Saitta et al., 2002).

Significant quantitative differences in the total phenol contents were observed in Nocellara etnea olive oils produced at different altitudes. The highest concentration of phenols was observed in olive oils obtained from plants cultivated at 100 m, and Nocellara etnea olives grown at a higher altitude (900 m) showed the lowest levels of phenolic compounds. This result is contradictory to several reports in which the total phenol content is positively correlated with altitude (Mousa et al., 1996; Ouni et al., 2012).

The phenol alcohols found in the olive oils studied were hydroxytyrosol and tyrosol. The concentration of hydroxytyrosol (4.26–9.93 mg·kg⁻¹) was generally higher than that of tyrosol (2.12–4.53 mg·kg⁻¹) and was similar to that reported by other authors (Baccouri et al., 2007). The highest concentrations of these compounds were observed in Nocellara del belice olive oils. In this variety, the mean values of hydroxytyrosol were higher in olive oils obtained from conventionally cultivated olive trees than in those obtained from organic cultivation. Nocellara etnea olive oils from olives grown at a higher altitude have greater tyrosol and hydroxytyrosol concentrations than Nocellara etnea oils obtained from olives grown at an altitude of 0 m.

The most abundant phenolic compounds found in the VOOs under study were secoiridoids. The concentrations of these compounds ranged from 0.3 to 54.31mg·kg⁻¹. In general, these results agree with those found by other authors in the Nocellara del Belice and Biancolilla olive oil varieties (Baccouri et al., 2007). However, the secoiridoid concentrations in Tonda iblea olive oil were lower than those previously reported (Napolitano et al., 2008). In general, the secoiridoid content of Nocellara del belice olive oils obtained from conventional cultivation is higher than that found in oils obtained from organic cultivation. There were significant differences in the secoiridoid content among Nocellara etnea oils obtained from olives cultivated at different altitudes. Nocellara etnea olives grown at 100 m produced oil with the highest content of secoiridoids, and the content decreased in oils obtained from olives grown at a higher altitude (900 m).

Phenolic acids were the first group of phenolic compounds to be identified in virgin olive oil (Servili et al., 2009). In recent years, phenolic acids have received considerable attention as they are essential to olive oil quality and nutritional properties. The phenolic acids observed in the VOOs in this study were p-coumaric and ferulic acids. Concentrations of ferulic acid ranged from 0.34 to 3.21 mg·kg⁻¹. In all the samples investigated, p-coumaric acid concentrations did not exceed 0.43 mg·kg⁻¹. Vanillic acid was not detected in Passulunara and Nocellara del belice organic olive oils. These results agree with those reported by other authors (Saitta et al., 2002). With regard to phenolic acid concentrations in the Novellara etnea variety at different altitudes, the highest p-coumaric acid concentration was found at 100 m (0.20 μg·kg⁻¹), while cinnamic acid and ferulic acid concentrations were higher in oils from N. etnea grown at 900 m.

Flavonoids and lignan groups of phenolic compounds abundant in plants are known to have many beneficial biological effects, including anti-inflammatory, antioxidant and oestrogenic activities (Adlercreutz, 2007). Considering flavone concentrations, luteolin and apigenin were more abundant in the Biancolilla and Nocellara etnea varieties. Luteolin concentrations ranged from 1.54 to 4.32 mg·kg⁻¹ and were higher than apigenin concentrations (0.84–1.96 mg·kg⁻¹). Nocellara etnea oils obtained from olive groves at an altitude of 0 m had higher flavonoid concentrations than those obtained from groves at a higher altitude. Concentrations of the lignan pinoresinol were within the range of 0.13–1.49 μg·kg⁻¹. Olive oils obtained from Nocellara etnea olive groves at an altitude of 900 m had higher pinoresinol concentrations than those obtained from groves at a lower altitude.

Nutritional and pharmacological therapies are needed to aid in the prevention of bone loss due to aging. In this study, the potential effects of VOO phenolic extracts obtained from Sicilian monovarietal olive oils on the proliferation of human osteosarcoma cells were investigated. MG-63 osteoblast cells were cultured in the absence and presence of phenolic extract dilutions in a fresh culture medium. The highest solvent concentration in the culture media (0.001%, v/v methanol) had no significant effects on cell proliferation.

The hypothesis of this study was that the phenolic fractions of each VOO variety would have differing effects on the induction of osteoblast proliferation. MG-63 cells were treated with various concentrations of extracts to determine the biological activities of the complex multi-component phenolic extracts. All extracts, with the exception of Tonda iblea, stimulated MG-63 cell growth significantly in a non-concentration-dependent manner, with the following order of potency: Nocellara del belice>Biancolilla>Organic Nocellara del belice>Nocellara etnea 0 m>Nocellara etnea 100 m>Nocellara etnea 900 m>Passulunara>Tonda iblea. Treatment of osteoblast cells with olive oil phenolic extracts increased the number of cells by 13.77–30.98% compared with controls (Table 2).

Differences in the ability of individual phenolic extracts to stimulate osteoblast proliferation were observed (Table 2). For instance, concentrations of 0.001% and 0.0001% were needed to increase cell proliferation significantly when MG-63 cells were cultured in the presence of Passulunara and Nocellara etnea at 900 m phenolic extracts. On the contrary, the olive oil phenolic extracts of Biancolilla, Nocellara del Belice, Nocellara etnea at 0 and 100 m altitudes significantly increased the MG-63 cell proliferation at the lowest concentration.

The VOO phenolic extracts with the highest activity were those obtained from the Nocellara del belice variety, as they significantly increased cell proliferation by 30.98%. In addition, treatment with organic Nocellara del belice VOO phenolic extracts increased cell proliferation by 28.46% at the lowest concentration. However, Tonda iblea phenolic extracts did not stimulate cell proliferation.

Although the impact of olive oil phenolic extracts on MG-63 cells is well-known, the mechanism by which they induce MG-63 cell proliferation is not fully understood. In this regard, certain observations may help describe this mechanism. Oxidative stress,which increases the level of intracellular reactive oxygen species, has been reported to inhibit bone metabolism (Hagiwara et al., 2011). Olive oil phenolic compounds have strong antioxidant properties (Visioli y Bernardini, 2011), and hydroxytyrosol has been shown to decrease H₂O₂ levels in MC3T3-E1 osteoblast cells (Hagiwara et al., 2011). In addition, it has been demonstrated that H₂O₂ suppresses differentiation markers (such as alkaline phosphatase activity), osteoblastic cell mineralization and type I collagen gene expression (Hosoya et al., 1998).

Even though the phenolic contents differed among the different VOOs studied, most phenolic extracts increased cell proliferation significantly. It should be noted that the effect on osteoblast proliferation was observed in a non-concentration-dependent manner. The oils obtained from the Tonda iblea variety did not have any effect on osteoblast cell proliferation, and it has been noted that this variety has a low total phenol content. However, there are no major qualitative differences in the concentration of phenolic compounds between the Tonda iblea variety and other varieties studied. Passulunara and Nocellara etnea varieties grown at 900 m induce a significant increase in cell proliferation at a phenol extract concentration of 0.001%, even though they are varieties with low total phenolic contents. In addition, phenol extracts from oils obtained from the Nocellara etnea variety grown at 100m,which is the variety with the highest total phenol content,did not exert the greatest effect on osteoblast cell proliferation. These data suggest that other substances present in this extract may be responsible for or influence the effect on MG-63 cell proliferation. Further studies are required to identify other bioactive substances and the mechanisms by which VOO phenolic extracts induce MG-63 cell proliferation.

The mean dietary intake of virgin olive oil phenols has been estimated to be approximately 68.5±104.0 mg·day⁻¹ in a Mediterranean population, which represented approximately 11% of the total phenolic compound intake (Treserra-Rimbau et al., 2013). Considering these amounts, the concentration of phenols used in this study are easily achieved with the actual daily intake of virgin olive oil and are potentially reached in the intestine after the consumption of olive oil. However, we should acknowledge that in vitro studies on biocompounds should always consider intestinal absorption and biotransformation. In this regard, research on humans (Miró-Casas et al., 2002; Visioli et al., 2003) and animals (Tuck et al., 2002) has shown that the dominant virgin olive oil phenols, hydroxytyrosol and tyrosol, are bioavailable and, as such, a fraction of the oral dose can be recovered in the urine, with >90% of the urinary excretion in the form of metabolites, which are mainly O-glucoronoided conjugates. In addition, after virgin olive oil administration, O-methylated forms, such as homovanillyl alcohol or homovanillyl acid, and monosulphated conjugates have been identified in plasma and urine (Mateos et al., 2011; Miró-Casas et al., 2002; Visioli et al., 2003). A further study found that the absorption of administered ligstroside-aglycone, hydroxytyrosol, tyrosol and oleuropein aglycone was as high as 55–66% of the ingested phenols in humans, indicating an effective intestinal absorption of these compounds (Vissers et al., 2002). Unfortunately, the knowledge available on the metabolic fate of VOO-derived complex phenols is still scarce (Lozano-Sánchez et al., 2010), hence the need for research on the plasma and urine concentrations of the free forms of various secoiridoid aglycones.

Although these experimental studies support the hypothesis that VOO phenolic compounds exert a beneficial effect on osteoblasts further studies assessing the in vivo accessibility of VOO phenolic compounds to osteoblast cells should be carried out. Accordingly, as the metabolism of VOO phenolic compounds is not well understood, it is necessary to perform further in vitro studies on the intestinal absorption and biotransformation of such compounds.

As shown in Table 3, an ANOVA demonstrated that MG-63 cell proliferation is influenced by the variety and concentration of olive oil (p=0.0001 and p=0.001, respectively). The adjusted R-squared value was 0.802, indicating that olive oil variety and phenol concentrations explain 80.2% of variability in the proliferation. The Bonferroni test showed that the Tonda iblea variety had different effects on MG-63 proliferation compared to the rest of the olive oil varieties studied. In addition, it was demonstrated that there are statistically significant differences (p=0.001) between the effects of the 0.001% and 0.00001% sample concentrations on MG-63 proliferation. On the contrary, there are no significant differences in the effect of the 0.001% concentrations on cell proliferation with respect to the other concentrations studied (p>0.005). An ANOVA of the phenol concentration revealed the influence of the variety of olive oil in all the groups of compounds studied (p=0.0001). The statistical test allowed us to group the eight varieties of olive oil into five categories according to their phenol content: I (Tonda iblea and Nocellara etnea 900 m), II (Biancolilla, Passulunara and Organic Nocellara del belice), III (Nocellara del belice), IV (Nocellara Etnea 0 m) and V (Nocellara Etnea 100 m).

Multivariate PCA was used to further describe the differences observed in the VOO samples obtained from different cultivars and grove locations. The score and loading plot obtained by PCA based on the phenol concentration values and the effect of the VOOs on cell proliferation are shown in Figure 1. Across the first-dimension PC1 (accounting for 41% of the total variance), the type of cultivar and the grove location were discriminated, whereas across the second dimension (accounting for 37% of the total variance), the cultivation method (organic or conventional for Nocellara del belice) and phenolic extract concentrations in osteoblast cells were discriminated. Table 4 shows the loading of the variables for the Principal Components 1 and 2. Secoiridoids and flavonoids showed the highest loading in Component 1 and phenolic acids and alcohols in Component 2.

Comparisons between the two PCA plots (score and loading) indicated that the variables flavonoids, phenolic acids and lignans were mainly responsible for the discrimination of the Nocellara etnea olive oils obtained at 900 m.This cultivar, along with Tonda iblea, obtained the highest scores on Component 1, and Nocellara del Belice (organic and conventional) obtained the highest scores on Component 2.

To characterize olive oils from the Nocellara etnea varieties at different altitudes, a multivariate analysis was applied to the chemical data and to the effect of this olive oil variety on osteoblast cell proliferation. PCA discriminated the samples by altitude (Figure 2; Table 5), thus highlighting the important role of the environment in determining the phenolic profile and nutritional properties of virgin olive oil (Servili et al., 2009). Comparisons between the two PCA plots (score and loading) indicated that the variables lignans and phenolic acids were mainly responsible for the discrimination of the oils at 900 m of altitude, whereas the variables secoiridoids were observed at 100 m of altitude for the oils. Finally, osteoblast cell proliferation and flavonoids were major contributors to the separation of Nocellara etnea oils at the lowest altitude.