Dulbecco's modified Eagle's medium (DMEM), L-glutamine, penicillin, streptomycin, fetal bovine serum (FBS), phosphate-buffered saline (PBS), 2′,7′-dichlorofluorescein diacetate (DCF-DA), mercury orange, and trypsin were purchased from Gibco. Cell proliferation kit II (XTT assay; Roche) was purchased from Roche Diagnostics. Ethanol (EtOH) was purchased from Carlo Erba Reactifs SDS. Methanol (MeOH) was obtained from Fisher Scientific UK. All the solvents were of analytical grade. Distilled water was used to prepare all aqueous solutions. H₂O was purchased from Macron Fine Chemicals [high-performance liquid chromatography (HPLC) grade] and 2,2′-azobis (2-methyl-propionamide) dihydrochloride (AAPH) was purchased from Sigma-Aldrich.

Four different samples of olive flowers belonging to Greek varieties were collected. More specifically, two of these samples (KTKT and ANKT) belong to the variety Lianolia from Corfu Island with collection dates 5–6/5/17 and 14–15/5/17, respectively. The third sample (AGRI) consists of flowers of the olive wild tree, collected on 13–14/5/17 from the island of Corfu. The final sample of olive flowers (EKPA) belongs to the variety Koroneiki and was collected on 13/5/17 from the area of the National and Kapodistrian University of Athens. The samples were collected under conducive conditions (hot and dry weather). Drying of the samples was carried out for 20 days in a dry, dark, and well-ventilated room. At the end of 20 days, the plant material was stored in the herbarium.

Subsequently, 20 g of each dry sample were extracted with 150 ml of EtOH/H₂O (50:50 v/v) for 30 min in an ultrasonic bath (Branson 2510). The above procedure was repeated twice for each sample. Subsequently, vacuum filtration was carried out and evaporation at 40°C in RotaVapor until EtOH removal. The extracts were initially stored in the freezer at −80°C for 24 h and then lyophilized on a Christ Alpha 1–5 lyophilizer (Martin Christ GmbH and CoHG).

HPLC device (Thermo Finnigan) was used for the qualitative and related quantitative analysis of the extracts and comprised a SpectraSystem P4000 pump, a SpectraSystem 1000 Degasser, a SpectraSystem AS3000 automatic sampling probe, and SpectraSystem UV6000LP detection probes (PDAs). Subdivision of the substances was performed on a Supelcosil RP-18 C18 chromatographic layer 25 cm × 4.6 mm i.d., 5.0 μm (Discovery). The mobile phases used were water with acetic acid (0.1%) (phase A) and acetonitrile with MeOH (2%) (phase B). The solvent gradient changed according to the following conditions: from 0 to 15 min, 95% (A): 5% (B) to 85% (A): 15% (B); from 15 to 40 min, 85% (A): 15% (B) to 55% (A): 45% (B); from 40 to 50 min, 55% (A): 45% (B) to 5% (A): 95% (B); from 50 to 55 min, 5% (A): 95% (B) to 5% (A): 95% (B); from 55 to 56 min, 5% (A): 95% (B) to 95% (A): 5% (B); from 56 to 60 min, 95% (A): 5% (B) to 95% (A): 5% (B). The flow of the mobile phase was set at 1 ml/min, and the injection volume of the samples was set to 10 μl. The detection of the eluted metabolites was performed using a PDA detector (254, 280 and 355 nm). For the related quantitative analysis 10 mg of each extract was diluted in 1 ml of MeOH and the samples were analyzed in triplicate. The chromatogram analysis and the peak area calculation were performed at 254 nm.

Total phenolic content was determined using Folin-Ciocalteu colorimetric method as presented previously by Blainski et al (12), according to which: 10 mg of gallic acid (97% purity) was dissolved in 1 ml of DMSO and serial solutions of decreasing concentration were prepared (1, 0.8, 0.7, 0.5, 0.4, 0.3, 0.2 and 0.1 mg/ml). After the addition of the Folin-Ciocalteu reagent, absorbance of the samples was measured at 765 nm using a spectrophotometer (TECAN Infinite M200 Pro UV/Vis Reader). For the construction of the reference curve, the absorbances corresponding to the linear region of the curve were selected (y=0.0479× +0.2653, R²= 0.9996). Then, 10 mg of each extract was dissolved in 1 ml of DMSO and serial solutions of decreasing concentration were prepared (5, 2.5 and 1.25 mg/ml). In each cell of the 96-well plate was transferred 25 µl of each sample dissolved in DMSO, 125 µl of Folin-Ciocalteu solution (2.5 ml of distilled H₂O contains 0.25 ml of Folin-Ciocalteu solution reagent) and 100 µl of 7.5% (w/v) aqueous sodium carbonate solution that acts as a promoter of the reaction. After appropriate agitation the samples remained in the dark for 30 min at 25°C and the absorption was measured by a spectrophotometer set at 765 nm (TECAN Infinite M200 Pro UV/Vis Reader). The total phenolic content was expressed as milligrams of Gallic Acid (GA) equivalent per gram of the olive flower extract (y=0.0479× +0.2653, R²= 0.9996) (15).

The radical scavenging capacity (RSC) of the blossom extracts was evaluated using the DPPH• assay (13) with slight modifications, as previously described (14,15). Briefly, 1 ml of freshly prepared methanolic solution of DPPH• (100 μΜ) was mixed with the tested extracts at various concentrations, ranging between 2.5 and 100 μg per extract, (ODsample). After 20 min of incubation in the dark the absorbance was monitored at 517 nm on a Hitachi U-1900 radio beam spectrophotometer (serial no. 2023–029; Hitachi). MeOH was used as a blank and DPPH alone in MeOH was used as the control (ODcontrol). The percentage RSC of the tested extracts was calculated using the equation: RCS% = (ODcontrol-ODsampleODcontrol) ×100.

The ABTS^•+ RSC of the tested extracts was determined as previously described by Cano et al (16), with minor modifications (14). Briefly, 1 ml of the reaction mixture containing ABTS^•+ (1 mM), H₂O₂ (30 μM) and horseradish peroxidase (6 μM) in 50 mM PBS (pH=7.5) was prepared in distilled water (dH₂O). Following incubation for 45 min in the dark, 10 μl of the tested extracts, at various concentrations, ranging from 2.5 to 50 μg per extract, was added (ODsample) and the absorbance at 730 nm was read on a Hitachi U-1900 radio beam spectrophotometer (serial no. 2023–029; Hitachi). In each experiment, a blank without the peroxidase was used, while the ABTS^•+ radical solution without the extract was used as the control (ODcontrol). The RSC percentage was determined using the same equation as that described for the DPPH assay.

To evaluate the antimutagenic capacity of the tested extracts we applied the Ames test using the bacterium strain Salmonella typhimurium TA102 (MolTox) according to Maron and Ames (17). In brief, 700 μl of the bacterium culture were used to inoculate 30 ml of autoclaved Oxoid nutrient broth no. 2. The cultures were placed on a vibrator (100 rpm) and incubated in the dark at 37°C until the cells reached a density of 1–2×10⁹ colony forming units (CFU/ml, OD₅₄₀ between 0.1 and 0.2). The following substances were then added in the sterile tubes: Plates with oxidant + the tested compound; 2 ml of top agar, 100 μl of the bacterial culture, 50 μl of tert-butyl hydroperoxide (0.4 mM) and 50 μl of each extract at various concentrations, ranging from 2 to 32 μg per extract/plate. In addition, a plate with the oxidizing agent alone and a plate without the oxidizing agent or the tested compound were used as positive and negative controls. Moreover, each extract was examined at the two highest concentrations used in the assay for putative induction of mutations. The aforementioned tubes were poured onto plates covered by glucose minimal agar and incubated at 37±2°C for 48 h. Then, the histidine revertant colonies (His+) were counted. The number of induced revertants was obtained by subtracting the number of spontaneous revertants from the number of revertants on the plates with the mutagen and/or antioxidant. The percentage inhibition of mutagenicity was calculated as: inhibition = no. of colonies per plate with oxidant + tested compound number of colonies per plate with oxidant alone ×100.

The DNA relaxation assay has been previously described (18). The principle of this assay is dependent on the conformational changes of the plasmid (pBluescript-SK+, Fermentas) DNA, which natively exists in the supercoiled conformation but after a single-strand break is converted to an open circular one. Based on this, the protective activity of the olive blossom extracts against DNA single-strand breaks by 2,2′-azobis AAPH (2.5 mM) were assessed. Specifically, in a total reaction volume of 10 μl, 2 µl of DNA (4 µg/ml) was mixed with PBS, AAPH and different concentrations of the tested extracts ranging between 1 and 300 μg/ml and the mixture was incubated at 37°C for 45 min. For each assay, a negative control (DNA without the tested compounds and AAPH) and a positive control (DNA with AAPH and without the tested compounds) were used. Moreover, the maximum tested concentrations were mixed alone with DNA for possible induction of DNA strand breaks. However, none of the tested concentrations were found to induce DNA breaks. Subsequently, 3 µl of loading buffer (bromophenol blue 0.25%+30% glycerol) was added and the samples were loaded on a 0.8% agarose gel, following electrophoresis at 70 V for 60 min. Eventually, the gel was stained with 12.5 µl of ethidium bromide (10 mg/ml) in 250 ml of dH₂O for 30 min, and then washed with 250 ml of dH₂O for another 30 min. Finally, the gel was exposed to UV, the MultiImage Light Cabinet (Alpha Innotech) was used to capture the gel images and the results were analyzed with the Alpha View suite.

According to the international guidelines on good cell culture practice (19), the cell lines used in this research were checked for mycoplasma, using PCR. According to PCR results the tested cell lines, were mycoplasma free. Furthermore, a morphology check, both at high and low culture densities via microscope were conducted to authenticate the state of cells, through their phenotypic characteristics. Finally, the passage number for each cell line did not exceed the 30 population doublings.

The cervical cancer (HeLa), murine myoblasts (C2C12) and the liver cancer (HepG2) cells were cultured at 37°C in 5% CO₂ in Dulbecco's modified Eagle's medium (DMEM) containing fetal bovine serum (FBS) (10% v/v), L-glutamine (2 mM), penicillin (100 U/ml) and streptomycin (100 U/ml). The endothelial cells (EA.hy926) were cultured at 37°C in 5% CO₂ in DMEM containing FBS (10% v/v), HEPES (25 mM), L-glutamine (2 mM), penicillin (100 U/ml) and streptomycin (100 U/ml).

HeLa and HepG2 cell lines were donated by Assistant Professor Kalliopi Liadaki (Department of Biochemistry and Biotechnology, University of Thessaly, Larissa, Greece). The C2C12 myoblasts were donated by Professor Koutsilieris (National and Kapodistrian University of Athens, Athens, Greece). Finally, the EA.hy926 cells were donated by Professor Koukoulis (University Hospital of Thessaly, Larissa, Greece).

The working concentrations of the tested extracts did not induce cytotoxicity in any cell line. In order to check which concentrations of the extracts were cytotoxic (i.e., which of them compromised cell viability) the XTT assay kit was used, according to the manufacturer's protocol.

The cells of each cell line were incubated in culture medium in flasks (25 m²) for 24 h. The medium was then removed and serum-free medium containing the tested extracts at non-cytotoxic concentrations was re-added in the flasks. The treatment of the cells with the extracts (or with the serum-free medium only for the control cells) lasted 24 h. Subsequently, they were trypsinised, collected, centrifuged (300 × g, 10 min, 5°C) and the supernatant fluid was discarded. The cellular pellet was re-suspended in PBS.

The intracellular GSH and ROS levels were assessed using the fluorescent dyes mercury orange and DCF-DA, respectively (20). A 400 μM stock solution of mercury orange was prepared in acetone and a 400 μM stock solution of DCF-DA was prepared in MeOH. The cell pellet was resuspended in PBS at the concentration of 1×10⁶ cells/ml and incubated with mercury orange (40 μΜ) or DCF-DA (10 μΜ) at 37°C for 30 min. The cells were then washed and re-suspended in PBS and subjected to analysis using a FACSCalibur flow cytometer (BD Biosciences) with excitation and emission wavelengths at 488 and 530 nm for ROS and at 488 and 580 nm for GSH. The cells were analyzed at a flow rate of 1,000 events/sec. Analyses were performed on 10,000 cells per sample and the fluorescence intensities were measured on a logarithmic scale. Data were analyzed using BD Cell Quest software (BD Biosciences).

SPSS version 21.0 was used (SPSS Inc., Chicago, IL, USA) for data analysis. All data were analyzed using one-way ANOVA followed by Tukey's test for multiple pair wise comparisons. Each experiment was repeated at least 3 times. Data are presented as mean ± standard error of the mean (SEM). The significance level was set at P<0.05, and the subset of alpha level was at 0.05. A bivariate Spearman's correlation was conducted to correlate the total polyphenolic content (TPC) of the extracts with the four assays tested, DPPH, ABTS, plasmid relaxation assay and Ames test.

The qualitative HPLC analysis of the olive flower hydroalcoholic extracts revealed that the major compounds belong in two chemical categories, secoiridoid derivatives and flavonoid derivatives (Fig. 1). Of these, quercetin-3-O-sophoroside (peak 2, Fig.1A) and oleuropein (peak 6, Fig.1A) are the main representatives of each category. The comparison study of the extracts showed a high similarity of the chemical composition of AGRI, EKPA and ANKT samples and only small differences on the quantities of the major compounds were observed. On the other hand, KTKT sample appeared to be poorer in terms of chemical composition while the main compounds, quercetin-3-O-sophoroside and oleuropein were present in small quantities. In more detail, the related quantification analysis revealed that all extracts contain similar concentration of the compound 1 (peak 1) in contrast to other compounds where significant variations were observed between the different extracts (Fig. 1B). Regarding the quercetin-3-O-sophoroside, AGRI extract contains the highest amount similar to EKPA and ANKT extracts while KTKT contains significantly lower levels than the other three extracts. By contrast, oleuropein was found in greater quantities in the EKPA extract followed by AGRI extract (similar concentration) and ANKT while KTKT contains low amount of oleuropein (Fig. 1B).

As indicated by the results regarding the DPPH^• and ABTS^•+ assays, all the extracts exhibited antioxidant activity. Specifically, regarding the DPPH^• assay the IC₅₀ values of AGRI, KTKT, EKPA and ANKT were equal to 40.50, 73.25, 50.75 and 73.25 μg of extract, respectively (Table I). In detail, statistical analysis revealed that AGRI exerted a stronger antioxidant activity compared with KTKT (P=0.002) and ANKT (P=0.002). By contrast, EKPA was more potent than KTKT (P=0.003) and ANKT (P=0.003). Moreover, as assessed by the ABTS^•+ assay the AGRI, KTKT, EKPA and ANKT extracts exhibited IC₅₀ values equal to 9.25, 15.77, 32.88 and 23.06 μg of extract, respectively (Table I). From the obtained results it seems that all extracts exhibited a statistically significant difference (P=0.0001). IC₅₀ values represent the amount of tested compounds required for 50% reduction of the two radicals. Given that the lower the IC₅₀ value the more powerful antioxidant activity, our results revealed that in both assays AGRI exerted the strongest antioxidant activity, compared with the remaining three (i.e., KTKT, EKPA and ANKT). Total phenolic content was estimated in four olive flower extracts, expressed in mg of gallic acid/g of extract. Estimated values varied from 50.92 to 81.03 mg of gallic acid/g extract (Table I), showing a significant difference in the phenolic content among the extracts tested. AGRI extract exhibited the highest phenolic content (81.03) followed by EKPA (76.15) and ANKT (66.06) extracts while KTKT exhibited the lowest value (50.92). It is noteworthy that the calculated TPC values of the extracts are in accordance with the corresponding antioxidant capacity expressed by DPPH values.

The obtained results from the plasmid relaxation assay revealed that AGRI (P=0.003), EKPA (P=0.019) and ANKT (P=0.005) possess equal antigenotoxic activity and were more prone to protect the plasmid DNA from lesions compared with KTKT (Table II). Specifically, the IC₅₀ values of AGRI, KTKT, EKPA and ANKT were calculated at 1.717, 8.233, 2.85 and 2.17 µg/µl, respectively. The same pattern was also observed in the Ames test. More specifically, the IC₅₀ values of AGRI, KTKT, EKPA and ANKT were calculated at 3.33, 4.11, 3.09 and 2.79 μg of extract, respectively (Table II). Thus, AGRI (P=0.043), EKPA (P=0.011) and ANKT (P=0.002) possess equal antimutagenic activity compared with KTKT.

The antioxidant capacity of the tested extracts was measured in four cell lines: EA.hy926, HeLa, and HepG2 cells, as well as C2C12 myoblasts. Prior to examining the potential antioxidant activity of the olive blossom extracts in cell culture, the concentration threshold above which the tested compounds exhibited cytotoxic effects in the cell lines was investigated. A range of amounts for each extract between 1.0 and 100.0 μg of extract was administered to the cells. The results from the XTT assay indicated that AGRI was more cytotoxic in EA.hy926 and HepG2 cells, exhibiting cytotoxicity at 10 μg of extract. Additionally, ANKT exhibited cytotoxicity at 25.0 μg in C2C12 myoblasts. Finally, the cytotoxicity level was observed at 25.0 μg for EKPA in HeLa cells (Table III). The obtained results revealed a tissue-specific activity of the extracts. Moreover, EA.hy926 seems to be the most sensitive cell line compared to the remaining cell lines, since the majority of the extracts induced cytotoxicity at low concentrations.

The results obtained from flow cytometry revealed that all tested extracts significantly increased GSH levels compared with the control in the four cell lines (Figs. 2, 4, 6 and 8). However, the corresponding ROS levels were not uniformly accompanied by statistical alterations (Figs. 3, 5, 7 and 9). Specifically, the AGRI increased the GSH levels of the EA.hy926 cells by 24 and 11% at 2.5 and 5 μg of extract, respectively, compared with the control (Fig. 2A). The ROS levels were decreased by 17% at 2.5 μg AGRI, compared with control (Fig. 3A). The KTKT extract also increased GSH levels by 18, 20 and 23% at 5, 10 and 20 μg of extract (Fig. 2B), respectively, while there was no alteration at ROS levels (Fig. 3B). Additionally, the GSH levels due to the EKPA extract were elevated by 12, 24, 22 and 25% at 5, 10, 20 and 40 μg of extract (Fig. 2C), with no alterations at ROS levels (Fig. 3C). Finally, ANKT increased GSH levels by 11 and 16% at 50 and 70 μg of extract, respectively (Fig. 2D) with a concomitant decrease at ROS levels by 21% at 70 and 90 μg (Fig. 3D).

According to the obtained results from flow cytometry in C2C12, GSH levels were elevated after 24 h incubation with AGRI by 40, 50, 18 and 13% at 10, 25, 50 and 60 μg of extract, respectively, compared with the control (Fig. 4A), while, ROS levels remained unaffected (Fig. 5A). Moreover, KTKT increased GSH levels by 70, 99 and 22% at 10, 25 and 50 μg of extract, respectively, compared with control (Fig. 4B) with no effect on ROS levels (Fig. 5B). Additionally, EKPA increased GSH by 21, 46 and 37% at 50, 60 and 80 μg of extract, respectively (Fig. 4C) with a concomitant reduction at ROS levels by 51 and 24% at 60 and 80 μg of EKPA, respectively, compared with control (Fig. 5C). Finally, ANKT also increased GSH levels by 14, 38, 39 and 44% at 2.5, 5, 10 and 20 μg of extract, respectively, compared with control (Fig. 4D) and decreased ROS levels by 10 and 12% at 10 and 20 μg of extract (Fig. 5D).

Furthermore, after AGRI administration of HeLa cells, GSH levels were increased by 46, 31 and 32% at 15, 25 and 45 μg of extract compared with control, respectively (Fig. 6A), while ROS levels remained relatively unaffected (Fig. 7A). Additionally, KTKT increased GSH by 41, 47, 49 and 55% at 25, 50, 70 and 90 μg of extract, respectively (Fig. 6B). By contrast, ROS levels were not altered (Fig. 7B). A mild increase at GSH levels was also observed after EKPA administration for 24 h, by 27, 17, 17 and 15% at 2.5, 5, 10 and 20 μg of extract, respectively (Fig. 6C) with no effects on ROS levels (Fig. 7C). Additionally, ANKT administration increased GSH by 17% at 90 μg of extract compared with control (Fig. 6D). By contrast, ROS levels were decreased by 32, 28 and 20% at 50, 70 and 90 μg of ANKT, respectively (Fig. 7D).

Finally, with respect to the effects on HepG2 cells (Fig. 8), AGRI, increased GSH levels by 10, 31 and 19% at 0.5, 1 and 2.5 μg of extract compared with control, respectively. Nevertheless, at 5 μg of AGRI, GSH levels were decreased by 12% indicating a pro-oxidant effect (Fig. 8A). Additionally, ROS levels were decreased by 16, 37, 47 and 53% after administration of 0.5, 1, 2.5 and 5 μg of AGRI, respectively (Fig. 9A). Moreover, GSH levels were increased by 31 and 17% at 40 and 60 μg of KTKT, respectively. However, at 80 μg of KTKT GSH levels were decreased by 32%, also indicating a pro-oxidant effect (Fig. 8B). Unlike GSH, ROS levels were not significantly affected (Fig. 9B). Furthermore, EKPA increased GSH levels at all tested concentrations. In detail, after administration of 1, 2.5, 5 and 10 μg of EKPA GSH levels were increased by 32, 65, 57 and 46% compared with control, respectively (Fig. 8C). EKPA administration was accompanied by ROS decrease by 20, 47 and 26% at 1, 2.5 and 5 μg of extract, respectively (Fig. 9C). Additionally, ANKT increased GSH levels by 11 and 20% at 60 and 80 μg of extract, respectively (Fig. 8D), whereas ROS levels were decreased by 62, 20, 32 and 26% at 20, 40, 60 and 80 μg of ANKT compared with control, respectively (Fig. 9D).