Comparison of phenolic content and antioxidant activities of millet varieties grown in different locations in Sri Lanka

Abstract Soluble and bound phenolic compounds were extracted from different varieties of millet types namely, finger millet, foxtail, and proso millet cultivated at dry and intermediate climatic zones in Sri Lanka. The extracts were examined for their total phenolic content (TPC), total flavonoid content (TFC), and proanthocyanidin content (PC). The antioxidant activities were meassured by reducing power (RP), trolox equivalent antioxidant capacity (TEAC), 2,2‐diphenyl‐1‐picrylhydrazyl (DPPH) radical scavenging activity, ferrous ion chelating ability (FICA), and using a β carotene linoleate model system. The ferulic acid content of extracts were determined using high‐performance liquid chromatoghraphy (HPLC). Finger millet showed the highest phenolic content and antioxidant activities compared to proso and foxtail millets. The phenolic content as well as antioxidant activites of soluble and bound phenolic extracts of millets were affected by variety and cultivated location. The highest phenolic content and antioxidant activites were reported for millet samples cultivated in areas belonging to the dry zone in Sri Lanka.


Comparison of phenolic content and antioxidant activities of millet varieties grown in different locations in Sri Lanka
Disna Kumari 1 | Terrence Madhujith 2 | Anoma Chandrasekara 1

| INTRODUCTION
Cereals play a vital role in human diet as an important source of energy, protein, and micronutrients among others for majority of people in the world. Dietary recommendations worldwide emphasize the significance of cereals in a balanced diet. Furthermore, cereals have been proven to provide additional health benefits while satisfying the energy and nutritional needs of humans. Risk of non-communicable diseases (NCDs) is increasing worldwide at an alarming rate in developed as well as developing regions. Several studies found that the regular consumption of whole grains and wholegrain products are helpful to prevent and to reduce the prevalence of NCDs (Okarter & Liu, 2010;Slavin, 2004).
Cereals have been used as staple foods both directly for human consumption and indirectly via livestock feeding since the ancient times. Cereal grains commonly cultivated for foods include wheat, rice, maize, oats, rye, barley, sorghum, and millets; the latter include a wide array of small-seeded grains. Millets are at the sixth place in world cereal production. They are the major food source for people living in economically disadvantaged status in Africa and Asia. Millets are known as the first domesticated cereals that were cultivated at the beginning of human civilization. Different millet types include brown top (Panicum ramosum), Japanese barnyard (Echinochloa crusgalli), finger millet (Eleusine coaracana), proso millet (Panicum miliaceum), kodo millet (Paspalum scrobiculatum), little millet (Panicum sumatrense), pearl millet (Pennisetum glaucum), and foxtail (Setaria italica) millet. Pearl millet is the most widely cultivated grains globally among these millet species at present (Annor, 2013).
Millets are being recognized as potential future crops due to their nutrient contents similar to other major cereals and non -nutrient compounds having proven health benefits. Studies have shown that
Aluminum chloride, trichloroacetic acid (TCA), and potassium phosphate dibasic were purchased from Techno Pharm Chem, India.
Sodium hydroxide, potassium phosphate monobasic, and potassium hydroxide, were purchased from Loba Chem Pvt Ltd, India.

| Determination of proximate composition of millet grains
Moisture, ash, crude protein, crude fiber, and crude fat contents of raw, dehulled millet grain samples were determined according to the AOAC (1999) methods.

| Sample preparation
Whole grains of millets were dehulled to separate hulls from grains.
Finger millet grains were dehulled using a rice polishing machine (Rice husker and polisher PM 500, Satake Engineering Co Ltd, Japan).

| Extraction of bound phenolic compounds
Bound phenolic compounds were extracted as explained by Chandrasekara and Shahidi (2010). In brief, the residues obtained after extraction of soluble phenolic compounds were hydrolyzed using 2 mol/L NaOH for 4 hr, stirring at room temperature in a shaking water bath (BT 680D, YIH DER Co., Ltd, Taiwan) at ambient temperature under a nitrogen atmosphere. The pH of the resulting slurry was adjusted to pH 2 with 6 mol/L HCL. Diethyl ether and ethyl acetate (1:1, v/v) were used to extract the bound phenolic compounds.
The extraction was done in three times and desolventized to dryness at 30°C in a rotary evaporator (IKA RV-10, IKA ® -Werke GmbH & Co. KG, Germany). Known volumes of methanol were used to reconstitute the phenolic compounds and stored at −80°C for further analysis.
Folin-Ciocalteu's reagent (0.25 ml) was added to 0.25 ml of methanolic extracts of soluble phenolics (2 mg / ml) in a centrifuge tube. The contents were vortexed and 0.5 ml of saturated sodium carbonate was added. After adding 4 ml of distilled water, the reaction mixture was kept in dark for 35 min, at room temperature and followed by centrifugation for 10 min at 4000g. The absorbance of the supernatant was measured at 725 nm (UV-VIS Spectrophotometer, Labomed Inc, USA). Appropriate blanks were used for background subtractions.
A standard curve prepared using ferulic acid was used to determine the TPC as μmol ferulic acid equivalents (FAE) / g of dry matter (dm).

| Determination of total flavonoid content
The total flavonoid content (TFC) was determined by a colorimetric method (Chandrasekara & Shahidi, 2010). The extracts of soluble phenolic compounds were dissolved in methanol to obtain a concentration of 2 mg / ml. The centrifuge tube containing 0.5 ml of soluble phenolic extract was added 2 ml of distilled water and 0.15 ml of 5% NaNO 2 and kept in room temperature for 5 min. Subsequently, the tube was mixed with 0.15 ml of 10% AlCl 3 and left to stand for 1 min.
Finally, the reaction mixture was added 2 ml of 1 mol/L NaOH and 2.4 ml of distilled water and allowed to stand for 15 min in dark. The reaction mixture was centifuged for 5 min at 4,000 g and the absorbance was measured at 510 nm against an appropriate blank. A standard curve prepared using catechin was used to calculate the TFC as expressed as μmol catechin equivalents (CE) / g of dm.

| Determination of proanthocyanidins content
Proanthocyanidins content was determined using the vanillin assay (Chandrasekara & Shahidi, 2010). Phenolic extracts (0.5 ml) in methanolic solution were mixed with 2.5 ml of 0.5% vanillin-HCl reagent (0.5% vanillin (w/v) in 4% concentrated HCl in methanol) and was incubated for 20 min at room temperature. A separate sample was performed with 4% HCl in methanol as a blank. The absorbance was meassured at 500 nm, and the content of proanthocyanidins was expressed as μmol CE / g of dm.

| Reducing power
The reducing power of samples was determined as explained by Chandrasekara and Shahidi (2010). Extract (0.5 ml) was mixed with 1.25 ml of phosphate buffer solution (0.2 mol/L, pH 6.6) and 1.25 ml of potassium ferricyanide in a centrifuge tube. The mixture was incubated for 20 min at 50°C and 1.25 ml of 10% TCA were added followed by centrifugation at 1750g for 10 min. The supernatant (1 ml) was transferred into a tube containing 1.25 ml of deionized water and 0.25 ml of 0.1% (w/v) FeCl 3 , and the absorbance values were read using a spectrophotometer at 700 nm. The standard curve was prepared using ascorbic acid. The results were expressed as μmol ascorbic acid equivalents (AAE) / g of dm.

| DPPH radical scavenging activity
The DPPH radical scavenging activity of phenolic extract was determined using a spectrophotometric method (Lee, Emmy, Abbe Maleyki, & Amin, 2007). Briefly, 0.04 ml of methanolic extract (2 mg/ml) was added to 1.96 ml of methanolic DPPH (60 μM) solution. The mixture was vortexed and allowed to stand at room temperature in the dark for 20 min. The absorbance of the solutions was measured at 517 nm with the appropriate blank. The DPPH Radical Scavenging Activity (DRSA) was expressed as μmol trolox equivalents (TE) / g of dm.

| Trolox equivalent antioxidant capacity
The trolox equivalent antioxidant capacity (TEAC) of soluble phenolic extracts of samples were determined by the method explained by Chandrasekara and Shahidi (2010 PBS solution was used to prepare millet phenolic extract (2 mg/ml) and further diluted to fit within the range of values (6.25-50 μmol/L) of the standard curve prepared using trolox.
To measure the TEAC, 40 μl of the extract was mixed with 1960 μl of the ABTS• solution. The absorbance of reaction mixture was measured at 734 nm immediately at the point of mixing (t 0 ) and after 6 min (t 6 ).
The absorbance reduction at 734 nm after 6 min of addition of trolox and extract was calculated using the following equation: ΔA trolox = (At0, trolox -At6, trolox) -(At0, blank -At6, blank), where ΔA is the reduction of absorbance and A the absorbance at a given time.
TEAC values were expressed as μmol trolox equivalents (TE) / g of dm.

| β -carotene-linoleate model system
Antioxidant activity of phenolic extracts was evaluated in a β-carotenelinoleate model system as explained by Chandrasekara and Shahidi (2010) with slight modifications. Briefly, 1 ml of β-carotene (1 mg / ml) in chloroform was pipetted into a 100 ml round bottom flask. Chloroform was evaporated under vacuum using a rotary evaporator at room temperature. (AAC) after 60 min of incubation was calculated using the following equation: AAC = (Aa (60) -Ac (60) )/(Ac (0) -Ac (60) ), where Aa (60) and Ac (60) are the absorbance values measured at 60 min for the sample and the control, respectively, and Ac (0) is the absorbance value of the control at 0 min.
The results were expressed as percentage of absorbance AAC / g of dm.

| Ferrous ion chelating activity
Ferrous ion chelating ability of phenolic extracts was determined colorimetrically (Chandrasekara & Shahidi, 2010). The aliquot of 0.2 ml extract in distilled water was added to a solution of 2 mmol/L FeCl 2 (0.025 ml) followed by addition of 0.2 ml of ferrozine (5 mmol/L) to initiate the reaction. The total volume of tube was adjusted up to 2 ml using distilled water and tubes were kept in room temperature for 10 min after vigorus shaking. The absorbance values were read at 562 nm. A separate control was prepared using distilled water in place of extract and blanks were arranged with added distilled water (1.8 ml) the results were expressed as μmol EDTA equivalents / g of dm.

| High-performance liquid chromatoghraphy analysis
Predominantly available phenolic acid, ferulic acid of soluble and bound phenolic extracts of millet grains were identified and quantified using high-performance liquid chromatoghraphy (HPLC) analysis.
The reversed-phase HPLC analysis was conducted by Shimadzu HPLC system (Shimadzu, SPD 20 A, Shimadzu Corporation, Kyoto, Japan) using Pinnacle ™ II C-18 column (4.6 × 150 mm, 5 μm, 110 Å, Restsk International, USA). The mobile phase was methanol/water (30:70 v/v). The flow rate was adjusted to 0.4 ml / min and the compounds were detected at 280 nm. All samples were filtered through a 0.45 μm syringe filter before injection. An external standard method was used to identify and quantify ferulic acid in millet samples. The results were expressed as μg / g of dm of millet samples.

| Statistical analysis
All experiments were carried out in triplicates and data were reported as mean ± standard deviation. The differences of mean values among millet samples were determined by one-way analysis of variance (ANOVA) followed by Tukey's honestly significant difference (HSD) multiple rank tests at p ≤ .05, significance level. Correlation analysis was performed between phenolic contents and antioxidant activity of soluble and bound extracts using Pearson correlations. All statistical analysis was performed by SPSS version 16 (SPSS Inc., Chicago, IL).

| RESULTS AND DISCUSSION
Millet samples used in this study consisted of different testa color and grain sizes. The weight of thousand grains of finger millet varieties Ravi, Ravana, Oshada, foxtail (PALL), and proso (ANUL) were 2.7, 3.3, 2.9, 1.1, and, 4.9 g, respectively. To the best of our knowledge, this study is the first to report on the phenolic contents and antioxidant activities of different millet varieties cultivated at a number of agro-ecological zones in Sri Lanka. The proximate composition of dehulled millet grains are presented in Table 2.

| Total phenolic content
The Folin-Ciocalteu's assay used to determine the TPC is based on the reducing ability of hydroxyl groups attached to phenolic compounds of the extracts. In this study, the TPC of soluble phenolic extracts of whole grain millets, dehulled grains, and their counterpart hulls ranged from 4.3 to 52.3, 0.4 to 32.5, and 10.9 to 44.4 μmol FAE / g dm, respectively (Table 3). In agreement with the previous studies, TPC of hulls were higher compared to those of dehulled and whole grains of studied millet samples (Chandrasekara, Naczk, & Shahidi, 2012;Chandrasekara & Shahidi, 2011;Varsha et al., 2009).
Finger, proso, and foxtail millets belong to three species, namely Eleusine coracana, Panicum miliaceum, and Setaria italica, respectively. It was reported that millets with dark color pigmented testa and pericarp showed higher phenolic content of soluble phenolic fractions than those with light color such as white or yellow testa (Chandrasekara & Shahidi, 2010). In agreement, soluble phenolic extracts of finger millets in this study had more TPC compared to those of foxtail and proso millets.
The varietal effect on TPC of millet extracts within the species is clearly demonstrated in this study. A significant difference of TPC among finger millet varieties, cultivated in the same location was observed in this study (Table 3). Earlier, considerable differences in 0.19 to 3.37% (catechin equivalents) of TPC among 85 Indian finger millet varieties were reported (Shankara, 1991). Varietal variations in respect to the TPC of finger millets have been reported in other studies too (Chethan & Malleshi, 2007).
The differences in the TPC of soluble extracts of millet grains due to cultivated locations were presented in Table 3. Sri Lanka is a country with a heterogeneous agro-ecological environment. Based on rainfall distribution, there are traditionally three climatic zones, namely wet zone, dry zone, and intermediate zone in Sri Lanka. In this study, millet samples were collected from agro-ecological regions in dry and intermediate zones (Table 1). According to the results, the millet samples obtained from agroecological regions in dry zone had more TPC compared to those from intermediate zone ( Table 3). The dry zone receives a mean annual rainfall of less than 1750 mm with a distinct dry season from May to September and the intermediate zone receives a mean annual rainfall between 1750 and 2500 mm with a short and a less prominent dry season .
Environmental factors such as sun exposure, soil type, and rainfall have an effect on phenolic content of plants (Manach et al., 2004).
Low temperature may increase the production of phenolics by enhancing synthesis of phenylalanine ammonia lyase (PAL) in plants, while high altitude and long sunlight exposure with high UV radiation positively affect the synthesis of phenolic compounds (Kishore, Ranjan, Pandey, & Gupta, 2010).
The TPC of bound phenolic extracts of representative millet samples were shown in Table 4. Bound phenolic extracts of millet whole  (Chandrasekara & Shahidi, 2010). In this study, the bound phenolic extracts of whole grains and hulls of foxtail millets had higher TPC compared to their soluble counterparts. Chandrasekara and Shahidi (2010) also reported high TPC in bound phenolic extracts of foxtail millet whole grains.
The trend of TPC of soluble and bound phenolic extracts of proso millets were different between the two cultivated locations in this study. According to Zhang, Ruihai, and Wei (2014), the TPC of bound phenolic extracts of dehulled proso millets were significantly higher than those of free phenolics. Further, the TPC of free phenolics extracts of three different varieties of proso millet ranged from 27.48 to 151.14 mg gallic acid equiv (GAE)/100 g dm. In addition, the bound phenolic content ranged from 55.95 (Gumi20) to 305.81 (Mi2504-6) mg GAE/100 g dm (Zhang et al., 2014).  (Kim, Jeong, & Lee, 2003).

| Total flavonoid content
Finger millet showed the highest TFC of soluble extracts followed by foxtail millets and proso millets (Table 3). An earlier study also showed that finger millets had higher TFC compared to foxtail and proso millets (Chandrasekara & Shahidi, 2010. Soluble extracts of millet hull had more TFC compared to those of dehulled grains and whole grains. The TFC of bound extracts of whole grains, dehulled grains and hulls of millet samples ranged from 0.03 to 0.26, 0.01 to 0.95, and 0.06 to 0.62 μmoles of CE/g dm, respectively (Table 4).
Earlier, a higher TFC of millet grains in soluble phenolic extracts was reported compared to those of bound millet grains (Chandrasekara & Shahidi, 2010). TFC was significantly influenced by the variety and cultivated locations of millets as shown in the present work (Table 3).

| Proanthocyanidin content
Proanthocyanidin or condensed tannins are oligomers or polymers of flavan-3-ol units and they are synthesized via the phenyl propanoid pathways. Figure 1 shows the proanthocyanidins content of finger millets. The highest proanthocyanidin content (PC) was found in whole and dehulled grains compared to those of hulls. Siwela, Taylor, De Milliano, and Duodu (2007) did provide evidence to show that tannins in finger millet are located in the testa layer.
There is a significant diversity among the PC of three different finger millet varieties cultivated in the same location and finger millets cultivated in different locations. The highest PC was found among the finger millets cultivated agro-ecological regions of DL 1, 2, and 5 in the dry zone compared to those of intermediate zone in Sri Lanka (Fig. 1).
The PC of foxtail millets and proso millets were not detectable.
In a previous study, the highest PC of 311.28 ± 3.0 μmol CE/g of defatted meal for finger millet local variety was reported followed by finger millet (variety Ravi), foxtail, little, pearl, and proso millets (Chandrasekara & Shahidi, 2010). They further reported a low amount of PC in proso millets and foxtail millets grains. Condensed tannins are biologically active and when present in sufficient quantities, may lower the nutritional value of food and biological availability of proteins and minerals. However, they possess antiinflammatory, antiviral, antibacterial, and antioxidant properties (Fei, Qiu, Ying, & Chang, 2008). In this study, the PC of whole grain finger millets were positively associated with TPC (r = 0.985; p < .001). This was in agreement with Siwela et al. (2007) who also reported a high significant positive correlation between total phenolics and condensed tannin contents (r = .927; p < .001) of finger millet, indicating the high contribution of condensed tannins to the TPC of finger millets.

| Reducing power
The reducing power (RP) assay is a method to determine the total antioxidant power of a plant extract. The method is based on the ability of compounds to donate electrons to reduce the ferricyanide to T A B L E 4 Total phenolic content (TPC), total flavonoid content (TFC), 2, 2-diphenyl-1-picrylhydrazyl (DPPH) radical scavenging ability, and reducing power (RP) of bound phenolic extracts of representative millet samples
finger millet seed coat extract was higher (p < .01) than that of whole flour extract (Varsha et al., 2009). Ju-Sung et al. (2010 established a positive linear correlation between TPC and RP of sorghum, foxtail millet, and proso millet (r = .985). This study also indicated a positive association between TPC and RP of finger millets (r = .367, p < .001).

| DPPH radical scavenging ability
DPPH is a synthetic stable free radical which can be scavenged by the donated hydrogen from the antioxidative compound. The DPPH• radical displays an intense UV-VIS absorption spectrum. The free radicals left, after reacting with phenolic compounds present in the samples were measured using the UV-VIS spectrophotometer.
The DRSA of bound and soluble phenolic extracts of whole grains, dehulled grains, and hulls of millets were presented in Table 4 and Table 6, respectively. Finger millets showed the highest DRSA in their soluble extracts compared to those of foxtail and proso millets. Finger millet dehulled grains had higher DRSA compared to those counterparts of whole grains and hulls. The DRSA of soluble extracts of dehulled finger millet grains were associated with PC (r = .963; p < .001) and TFC (r = .771; p < .001). In a previous study, high DRSA, has been documented that could be due to high content of phenolic compounds such as tannin and flavonoids in finger millets (Yokozawa et al., 1998).
Variety Oshada of finger millet showed 68% and 32% higher DRSA of soluble phenolic extracts of dehulled grains than those of Ravi and Ravana varieties, respectively. Results of present study clearly demonstrated the differences in DRSA among finger millet varieties.
A similar finding was reported for brown or red variety of finger millet, which were commonly available, with higher DRSA of 94% than those of white varieties which shown only 4% (Hegde & Chandra, 2005).
In this study, DRSA of millets were significantly different among the cultivated locations. Recently, it was shown that the DRSA of proso millet is affected by the growing environment (Kejariwal & Mehra, 2014). Authors further revealed that the proso millet organically grown had higher percentage of DRSA compared to the conventionally grown proso millet. Mpofu, Sapirstein, and Beta (2006)

| TEAC
TEAC assay is based on the scavenging ability of antioxidants to the long-life radical anion ABTS•. In this assay, ABTS is oxidized by AAPH to its radical cation, ABTS•, which is having intense blue-green color.
The antioxidant ability was measured as the ability of test compounds to decrease the color reacting directly with the ABTS• radical. Results of test compounds are expressed relative to trolox, water soluble analog of α-tocopherol.
T A B L E 6 2, 2-diphenyl-1-picrylhydrazyl (DPPH) radical scavenging ability and ferrous ion chelating ability of soluble phenolic extracts of millets The TEAC of soluble phenolic extracts of whole grains, dehulled grains, and hulls of millets were presented in the

| β -Carotene-linoleate model system
In β -carotenelinoleate model system, the presence of phenolic compounds will hinder the extent of β carotene bleaching by neutralizing the linoleate free radicals and other free radicals formed within the system. Therefore, depending on the degree of antioxidant compounds present in the system retain the color of β carotene. Same letters in each column and same numbers in each raw are not significantly different (p > .05).
antioxidant activities of some phenolic compounds are due to their high tendency to chelate metal ions.
In this study, the ferrous ion chelating ability was measured by the formation of purple color complex of ferrous ions with ferrozine and the intensity of the purple color of the complex decreases in the presence of chelating agents. The soluble and bound phenolic extracts of millets showed different degrees of FICA. The FICA of soluble extracts of millets was in the order of finger > foxtail > proso millets ( Table 6).
The results of a previous study by Chandrasekara and Shahidi (2010) also showed that the phenolic compounds present in millets are good source of metal chelating agents to inhibit the radical-mediated chain reactions. Their results showed that the FICA of soluble phenolic extracts of millets ranged from 0.37 to 7.99 μmol of EDTA equiv/g of defatted meal. They had obtained the highest metal chelating effect for the finger millets among the studied milllet samples (finger, proso, kodo, and foxtail millets).
In this study, the highest FICA was obtained for the dehulled grains compared to the whole grains and counterpart hulls of millets. The FICA demonstrated a positive relationship with PC (r = .472; p < .001) and DRSA (r = .572; p < .001) in dehulled grains of millets. Chandrasekara and Shahidi (2010) also presented that the FICA of soluble extracts of millets had a significant positive correlation with PC (r = .551; p < .01) and did not have significant correlation with TPC and TFC, which could be due to the formation of stable complexes by proanthocyanidins with metal ions as ferrous ion chelator. The highest FICA with more condensed tannin was found among the finger millet dehulled grains cultivated in dry zone (Agro-ecological zones of DL1, 2, and 5) compared to those in intermediate zone in Sri Lanka. agreement, ferulic acid content was varied significantly in wheat cultivars grown in different environments (Abdel-Aal et al., 2001).
The overall results of this study indicated that the phenolic contents and the antioxidant activities of millets were significantly affected by the variety and cultivated locations. The antioxidant activities with different mechanisms explain that millet grain phenolics can act in a number of ways against oxidative stress. Highest phenolic content and antioxidant activities were found in the studied millet samples obtained from dry zone in Sri Lanka. Since this is the first study conducted about the effect of growing conditions on phenolic contents and antioxidant activities of millets in Sri Lanka, the results may be important to optimize the growing conditions of selected variety to produce millets rich in natural antioxidants to combat against the burden of non-communicable diseases arising in the country.

FUNDING INFORMATION
This research was supported by the National Research Council of Sri Lanka (NRC 12-096) through a research grant to AC.