The Antioxidant Activity of Maltese Honey

Abstract Maltese honey has been valued for its medicinal properties since ancient times, so much so that the word Malta is derived from the ancient Greek word, Melite, which means honey. In this study, Maltese honey samples (n = 30) from three harvest seasons and different floral sources were analysed to determine their antioxidant/antiradical properties. Colorimetric assays were used to determine the polyphenol content, flavonoid content, and colour intensity, and these values were analysed to determine whether they correlated with radical scavenging activity as determined by bioassays. The honey samples varied in their DPPH activity (range, 3.326 - 11.865 mg AEAC/100 g honey), reducing power (EC 50 range: 5.632 - 14.453 mg/ml), total phenolic content (range, 8.114 - 109.765 mg TAE/100 g honey), total flavonoids (range, 0.655 - 212.865 mg RE/100 g honey), and colour intensity (range, 0.000 - 99.581 mm Pfund). The data were subjected to factor analysis using principal components. The antioxidant properties correlated significantly with total polyphenol content and colour (p<0.0001). Principal components analysis revealed three clusters of honey corresponding to the three harvest seasons.

Because of its antimicrobial, anti-inflammatory, and antioxidant properties, honey is used to treat medical conditions such as burns, skin ulcers, and eye ailments like cataracts (National Honey Board, 2002). Apart from its therapeutic applications, honey has been studied in the context of the deteriorative oxidation reactions in food that are caused by light, heat, and some metals. Inhibition of food deterioration by honey has been demonstrated in studies of the enzymatic browning of fruits and vegetables, lipid oxidation in meat, and growth suppression of foodborne pathogens and food spoilage organisms (Chen et al., 2000;Mundo et al., 2004;Nagai et al., 2006;Ferreira et al., 2009). The antioxidants in honey are categorized as enzymatic (e.g. catalase, glucose oxidase, and peroxidase) and non-enzymatic (National Honey Board, 2002). The latter category includes a wide array of metabolites, such as ascorbic acid, α-tocopherols, carotenoids, amino acids, proteins, organic acids, Maillard reaction products, and more than 150 polyphenolic compounds such as flavonoids, phenolic acids, tannins, and their derivatives (Ferreira et al., 2009). Moreover, the physicochemical and biological characteristics of honey can be affected by the floral sources and by environmental and seasonal factors. Notably, dark honey and honey with high moisture levels have higher antioxidant activities. The antioxidant activity of honey correlates with potential alkalinity, ash content, and with antioxidant pigments such as carotenoids and flavonoids (Sangsrichan and Wanson, 2008). In Malta, dark coloured carob honey is traditionally used to treat coughs and colds. In some cases, carob syrup is used as a substitute for carob honey. Thyme honey, a typical summer honey, is also used to treat coughs and colds, mainly by dissolving a teaspoonful of honey in a 150-mL cup of hot water Antioxidant activity of Maltese honey and adding a few drops of lemon juice to supplement the astringent and antioxidant properties of honey (Lanfranco, 2001). In this study, we aimed to determine the antioxidant activity of Maltese honey and to investigate whether this activity was related to a particular phytochemical class or to a particular season.

MATeRiAl And MeThods
Honey samples produced by Apis mellifera ruttneri (Sheppard et al., 1997) were collected in different regions of Malta in three seasons: autumn 2011 (AU 2011; September to December; carob/ eucalyptus), spring 2012 (SP 2102; March to June; mixed flora), and summer 2012 (SU 2012; July and August; wild thyme). Random samples that were representative of each season were collected following physicochemical characterization of the different honey types (Bugeja Douglas and Attard, 2012). The samples were stored at 4°C until further analysis. A 50% (w/v) solution was prepared from each honey sample and then centrifuged at 1500 x g for 5 min. The supernatant was used as the stock solution. Antiradical activity determination: the dPPh assay The DPPH assay (2,2-diphenyl-1-picrylhydrazyl, Sigma Aldrich, USA) was modified for the honey matrix (Moein et al., 2007). Briefly, each honey sample was prepared as a 12.5% (w/v) solution in distilled water, and 200 μL of a 100 μM solution of DPPH radical in methanol were added to 20 μL of each sample in micro-plate wells (Nunc, Denmark). After mixing, the reactions were incubated for 30 min at room temperature in the dark. DPPH radical inhibition was measured at 490 nm against reference mixtures in which the DPPH solution was replaced with methanol. Ascorbic acid (BDH, UK) was used as a standard (10 -100 μg/mL). The AAE-DPPH (Ascorbic Acid Equilavalents against the DPPH Assay) and ascorbic acid equivalence (mg AEAC/100 g honey) (r 2 = 0.9928) were calculated. honey colour Honey colour was determined by measuring the absorbance of 50% honey solutions (w/v) at 630 nm using a microplate reader. The honey samples were classified according to the Pfund scale after conversion of the absorbance (Abs) values to mmPfund according to the formula: mmPfund = -38.70 + 371.39 x Abs (Kaškoniene et al., 2009)

Phenolic compounds
The total phenolic content (TP) was determined using Folin Ciocalteu's reagent (Attard, 2013). Briefly, 100 μL of Folin-Ciocalteu reagent (Sigma-Aldrich, pre-diluted 1:9 with distilled water and 80 μL of sodium carbonate (Na 2 CO 3 ; Sigma-Aldrich, 1 M) were added to 10 μL of each honey stock solution. The absorbance was measured at 630 nm after incubation for 20 min at room temperature using a microplate reader (BioTek ELx800, Winooski, VT, USA). Tannic acid (Sigma-Aldrich) was used as a standard (60 -960 mg/mL; r 2 = 0.9941) to prepare a calibration curve in order to determine the total amount of phenols in each honey sample (mg TAE/100 g of honey).

Total flavonoid content
The total flavonoid (TF) content of honey samples was determined using a colorimetric method (Moein et al., 2007;Moein and Reza Moein, 2010). Briefly, 25 μL aliquots of 50% (w/v) honey samples were mixed with 100 μL of 0.15% NaNO 2 solution (Fisher Scientific, UK). After 6 min, 100 μL of 4% NaOH solution was added to the mixture, followed by 25 μL of distilled water, for a final volume of 250 μL. After resting for 15 min, the absorbance of the mixture was determined at 510 nm using water as a blank. Rutin (Sigma-Aldrich, USA) was used as a standard for the quantification of total flavonoid content (0 -500 mg/mL; r 2 = 0.9962).

Reducing Power
The reducing power or FRAP (Ferric Reducing Antioxidant Power) of honey samples was determined using the method of Oyaizu (1986). Honey solutions (50%, 25%, 12.5% and 6.25% (w/v)) in 250 μL of distilled water or 250 μL of distilled water (blank) were mixed with 250 μL of 0.2 M phosphate buffer (pH 6.6) and 250 μL of 1% potassium ferricyanide (Sigma-Aldrich, USA). The mixture was incubated at 50°C for 20 min and then rapidly cooled in an ice bath for 30 s. Following this, 250 μL of 10% trichloroacetic acid (TCA, Fisher Scientific, UK) were added and the mixture was centrifuged at 3000 rpm for 10 min. Next, 500 μL of the upper layer were mixed with 500 μL of distilled water plus 100 μL of 0.1% ferric chloride (Acros Organics, UK). The absorbance of these mixtures was measured at 700 nm against a blank (Mohamed et al., 2012). The EC 50 was used to define the specific reducing capability (mg AEAC/100 g honey) using ascorbic acid (10 -100 μg/ ml; r 2 = 0.9981) as a positive control.

data analysis
All measurements were performed in triplicate. The EC 50 values were determined using Statplus 2009 5.8.4.0 (Analystsoft, USA). All statistical analyses of the variables AAE-DPPH, colour, TP, TF, TF/ TP, EC50-RedPw (Median Effective Concentration for the Reducing Power Assay), and AAE-RedPw (Asorbic Acid Equivalents according to the Reducing Power Assay) were performed using SPSS (IBM version 22). The Kolmogorov-Smirnov test revealed that the variable values had fairly normal distributions and so the one-way ANOVA test was used to compare the mean variable values in the three seasons (AU 2011, SP 2012, and SU 2012. The null hypothesis specifies that the mean variable values are comparable in the three seasons and is accepted if the p-value exceeds the 0.05 level of significance. The alternative hypothesis specifies that mean variable values differ significantly in the three seasons and is accepted if the p-value is less than the 0.05 criterion. The Dunnett T3 post-hoc test was used for pairwise comparison since the Levene's test revealed that the variances of the variable values differed significantly in the three seasons. Considering that the variable values have metric scales and normal distributions, Pearson correlation, which ranges from -1 to 1, was used pairwise to measure the strength of the relationship between the variables. The null hypothesis specifies that the two variables are unrelated (Pearson correlation is close to 0) and is accepted if the p-value exceeds the 0.05 level of significance. The alternative hypothesis specifies that the two variables are significantly related (Pearson correlation is significantly different from 0) and is accepted if the p-value is less than the 0.05 criterion. Factor analysis using principal components was used to identify latent traits within the data. The Kaiser-Meyer-Olkin (KMO) measure was used to determine whether latent trends exist within the data, and Bartlett's test of sphericity was used to determine if the variables were related and therefore suitable for structure detection.

ResulTs
Results of the chemical and biological analysis of the Maltese honey samples are shown in Table 1, which reports the mean variable values for all 30 honey samples and the 95% confidence limits of the actual mean variable values for each season. Honey colour ranged from water white (0.00 mm Pfund) to amber (111.84 mm Pfund). Both the One-Way ANOVA and the Dunnett T3 post hoc test revealed that the mean colour of SP 2012 honey (18.163 ± 6.911 mmPfund) was significantly lower than that of honey from the other two seasons (AU 2011: 57.899 ± 14.112 andSU 2012: 47.722 ± 9.228 mm Pfund). The phenolic compounds ranged from 28.11 to 109.777 mg TAE/100 g honey. The three darker honey samples collected during autumn had higher amounts of phenolic compounds: AU01, AU02, and AU05 had 104.97, 109.77, and 103.83 mg TAE/100 g honey, respectively. One-Way ANOVA showed that the mean phenolic levels were considerably higher in the autumn samples than in the spring and summer samples, while the Dunnett T3 post hoc test revealed that the mean TP mg TAE/100 g honey in SP 2012 (56.943 ± 7.027) was significantly lower than in the other two seasons (AU 2011: 79.692 ± 8.000 and SU 2012: 69.598 ± 3.208 mg TAE/100 g honey). The flavonoid content in the Maltese honey ranged from 2.00 to 81.16 mg RE/100 g honey. In three instances, the values were out of this range; for honey sample SU09, the flavonoid level was undetectable, while for samples AU03 and AU07, the flavonoid levels were extremely high (212.86 and 197.57 mg RE/100 g honey, respectively). In fact, after removing the outliers from the data set, there was no statistical difference in the flavonoid content of the autumn, spring, and summer honey samples, which had 31.154 ± 17.729, 37.651 ± 8.460, and 31.420 ± 11.373 mg RE/100 g honey, respectively. The anti-oxidant activity of Maltese honey was analysed using bench top bioassays. In this study, the reducing power values ranged from 3.31 to 10.66 mg AEAC/100 g honey. The One-Way ANOVA and the Dunnett T3 post hoc tests confirmed that the mean AAE-RedPw mg AEAC/100 g honey for SP 2012 (5.980 ± 0.653) was significantly lower than for the other two seasons (AU 2011: 7.54 ± 0.798 and SU 2012: 6.96 ± 0.242 mg AEAC/100 g honey). Likewise, the value for 50% inhibition of oxidation ranged from 7.39 to 28.26 mg/ml. Both the One-Way ANOVA and the Dunnett T3 post hoc test revealed that the mean AAE-RedPw mg AEAC/100 g honey in AU 2011 (12.67 ± 1.093 mg AEAC/100 g) was significantly lower than in SU 2012 (14.250 ± 0.035 mg AEAC/100 g), which in turn was significantly lower than in SP 2012 (16.600 ± 1.979 mg AEAC/100 g). The Maltese honey samples had antiradical activity values between 3.33 and 15.62 mg AEAC/100 g honey. The One-Way ANOVA test revealed that the mean AAE-DPPH mg AEAC/100 g honey in AU 2011 (9.300 ± 1.292) was considerably higher than in SP 2012 (5.805 ± 0.610) and Antioxidant activity of Maltese honey SU 2012 (5.238 ± 0.657). In fact, the Dunnett T3 post hoc test showed that the value for the autumn samples was significantly higher value than for the other two seasons (p<0.05). Pearson correlation revealed that total flavonoid content was unrelated to all of the other variables (Tab. 2). AAE-DPPH, total polyphenols, colour, and AAE-RedPw were significantly positively related to each other, while EC50-RedPw was significantly negatively related to these variables (p<0.05). The data were then subjected to factor analysis using principal components. The KMO value, 0.840, exceeded the 0.5 threshold value, indicating that factor analysis was useful for identifying latent trends in the data. The p-value of the Bartlett's test of sphericity, which was approximately 0, was less than 0.05, indicating that some of the variables were related and therefore suitable for structure detection. Two latent factors had an eigenvalue greater than 1, which together explained 79.4% of the total variance (Tab. 3). The factor loadings of these two dominant factors are shown in Table 4. For the first factor, the factor loadings of AAE-DPPH, TP, colour, and AAE-RedPw, which were positive; in contrast, the factor loading of EC50-RedPw was negative. The second factor weighed heavily only on total flavonoid content, which suggested that the total flavonoid level was unrelated to the other variables and is a 1-variable factor. The factor loadings plot (Fig. 1a) shows three groups of variables. The first cluster includes TP, colour, AAE-DPPH, and AAE-RedPw; the second cluster includes only TF, and the third cluster includes only EC50-RedPw. Figure 1b shows the factor scores of the two latent factors. Factor 1, on the horizontal axis, contrasts EC50-RedPw with AAE-DPPH, TP, colour, and AAE-RedPw and shows that most of the observa-

discussion
This study found that honey colour was related to the floral origin of the honey. Persano Oddo and Piro (2004) reported that clover and sage honey is water white, while honeydew, heather, and chestnut honey is dark amber. Honey colour depends on its botanical origin, minerals, and other minor components, and darker honey has better radical scavenging activity due mainly to its phenolic content (Savatović et al., 2011). In the present study, there was a correlation between honey colour and antioxidant activity (p<0.0001). This relationship might indicate that the presence of pigment is a reliable index of antioxidant activity. Impure honey may be contaminated by pigments during handling, processing, and storage. Colour changes may be also observed during maturation due to biochemical reactions that could lead to components without antioxidant activity (Beretta et al., 2005). Because polyphenols have antioxidant activity, the Folin-Ciocalteu method was used to determine the total phenolic content. However, non-phenolic reducing compounds may interfere with this test, leading to an over-estimation of the phenolic content. Substances like ascorbic acid, α-tocopherol, carotenoids, several sugars (e.g. glucose, fructose) and amino acids (e.g. histidine, taurine, glycine, and alanine) are usually cited as interfering substances. Some studies have demonstrated that the correlation between radical scavenging activity and proline is higher than that between radical scavenging activity and total phenol content (Ferreira et al., 2009;Savatović et al., 2011). We observed a weak but positive correlation between radical scavenging activity and proline content (data not shown) for local honey samples. However, further studies are needed to identify and quantify the biologically active components present in honey. Flavonoids are low molecular weight phenolic compounds that contribute to the aroma and antioxidant properties of honey. The flavonoid content obtained in this study was higher than that obtained in similar studies. In other studies, flavonoids were present in quantities ranging from 1.35 to 9.78 mg RE/100 g honey. Nevertheless, the type of honey investigated in these studies (Kaškoniene et al., 2009;Savatović et al., 2011) was not the same as in the present study. The ratio of the total content of phenolic compounds to flavonoids was calculated in order to evaluate the relative amounts of flavonoid and non-flavonoid compounds in honey. The seasonal average ratios ranged from 0.460 to 0.642. This is in accordance with the ratios obtained in a study by Kaškoniene et al. (2009) in which ratios ranged between 0.05 and 0.5, depending on the floral source of the honey. In this study, the reducing power values were lower than those obtained by Meda et al. (2005) but compare well with those of Savatović et al. (2011). Meda et al. (2005) showed that the reducing power values of 18 multifloral honeys varied from 10.20 to 37.87 mg AEAC/100 g honey. In the study by Savatović et al. (2011), the reducing powers of Acacia honey, Linden honey, and "Homoljski med" (a mix of Acacia and Meadowflower) honey were 1.43, 7.82, and 4.29 mg AEAC/100 g honey, respectively. Higher values were reported in studies of unifloral honeys, and ranged from 13.26 to 100.80 mg/ml (Ferreira et al., 2009;Savatović et al., 2011). In our study, the monofloral honey types, namely the autumn and summer samples, exhibited higher reducing power values (7.54 and 6.96 mg AEAC/100 g honey) than the multifloral spring samples (5.98 mg AEAC/100 g honey). Radical scavenging activity as measured with DPPH showed a trend that was similar to that of the reducing power values, except for the summer samples, which had lower values (5.238 ± 0.657 mg AEAC/100 g honey) than the spring samples (5.805 ± 0.610 mg AEAC/100 g honey). The autumn honey samples had the highest values (9.30 ± 1.292 mg AEAC/100 g honey). Due to the presence of carob nectar, the autumn honey is particularly high in polyphenols and tannins, as noted by Avallone et al. (1997). The different parameters were analysed using principal components analysis. The fact that AAE-DPPH, TP, colour, and AAE-RedPw correlated negatively with EC50-RedPw (correlation coefficients between -0.022 and -0.703) suggests that samples with high AAE-DPPH, TP, colour, and AAE-RedPw values tend to have low EC50-RedPw values and vice versa. The total flavonoid content was an independent variable within this phytochemical-antioxidant relationship, supporting the notion that colour and polyphenol content contribute to the antioxidant activity of honey. This complements the Pearson correlations findings (Tab. 2). Principal