Investigation of plant extracts for the protection of processed foods ...

Eur Food Res Technol (2001) 212 : 319–328

Q Springer-Verlag 2001

ORIGINAL PAPER

Karin Schwarz 7 Grete Bertelsen 7 Lise R. Nissen Peter T. Gardner 7 Marina I. Heinonen 7 Anu Hopia Tuong Huynh-Ba 7 Pierre Lambelet Donald McPhail 7 Leif H. Skibsted 7 Lilian Tijburg

Investigation of plant extracts for the protection of processed foods against lipid oxidation. Comparison of antioxidant assays based on radical scavenging, lipid oxidation and analysis of the principal antioxidant compounds Received: 16 February 2000 / Revised version: 7 July 2000

Abstract Antioxidant activities of plant extracts from spices, coffee, tea, grape skin, and tomato peel slurry were evaluated using a number of analytical methods including the quantification of principal compounds. Similar rankings in the activities of these extracts were obtained by evaluating their efficiencies as scavengers of stable free radicals: Fremy’s salt, galvinoxyl or a,adiphenyl-b-picrylhydrazyl (DPPH). Similar results were obtained with the lipid oxidation assays based on thermal acceleration (formation of conjugated dienes in methyl linoleate at 40 7C or the Rancimat test at 100 7C with lard). Rankings of the extract activity obtained by scavenging of hydroxyl radicals generated in the Fenton reaction were similar to those obtained by an oxygen consumption assay with linoleic acid as substrate and metmyoglobin as catalyst. However, the results of the latter two assays differed from those of the other

K. Schwarz (Y) Institute of Human Nutrition and Food Science, CAU, University of Kiel, 24098 Kiel, Germany e-mail: Kschwarz6foodtech.uni-kiel.de Fax: c49-431-880-4283 G. Bertelsen 7 L. R. Nissen 7 L. H. Skibsted Department of Dairy and Food Science, The Royal Veterinary & Agricultural University, 1958 Frederiksberg, Denmark P. T. Gardner 7 D. McPhail The Rowett Research Institute, Division of Biochemical, Sciences, Aberdeen AB259SB, UK M. I. Heinonen 7 A. Hopia Department of Applied Chemistry and Microbiology, Food Chemistry, 00014 University of Helsinki, Finland T. Huynh-Ba 7 P. Lambelet Nestec Ltd., Nestlé Research Centre, PO Box 44, 1000 Lausanne 26, Switzerland L. Tijburg Unilever Research Vlaardingen, The Netherlands

assays. In the overall ranking, coffee and rosemary extracts were amongst the most potent extracts whereas the tomato peel slurry showed no activity. Keywords Antioxidants 7 Lipid oxidation 7 Radical scavenging 7 Plant extracts

Introduction Interest in employing antioxidants from natural sources to increase the shelf life of foods is considerably enhanced by consumer preference for natural ingredients and concerns about the toxic effects of synthetic antioxidants. During the past decade spice extracts have been marketed as antioxidants for the food industry and, according to Krishnakumar and Gordon [1], had a share in 1995/6 of approx. 10% of the entire food and feed antioxidant market in Europe. In addition the health benefits found for flavonoids [2, 3] have increased the relevance of natural non-vitamin antioxidants. Selection of a suitable extraction procedure can increase the antioxidant concentration relative to the plant material. In addition, undesirable components can be removed prior to adding antioxidant active plant material to foods. Several extraction techniques have been patented [e.g., 4, 5] using solvents with different polarities, such as petrol ether, toluene, acetone, and ethanol. In addition, supercritical CO2–extraction [6] and medium-chain triglycerides as carrier in a mechanical extraction process have been applied [7]. In view of the differences between the extraction techniques, it is obvious that extracts from the same plant material may widely vary with respect to their antioxidant concentration and pattern. The concentration of individual antioxidants in plant extracts determined by HPLC is the preferred way to provide standardized information. However, the antioxidant pattern is usual-

320

ly rather complex, thus making the prediction of a mixture’s potency based on compositional data difficult. Therefore, the employment of specific assays to test the antioxidant activity of the extracts – including synergistic effects – is required. A variety of tests expressing antioxidant potency has been suggested. The tests can be categorized into two groups: assays for radical scavenging ability and assays that test the ability to inhibit lipid oxidation under accelerated conditions. Test systems that evaluate the radical scavenging ability of antioxidants aim to simulate basic mechanisms involved in lipid oxidation by measuring either the reduction of stable radicals or radicals generated by radiolysis, photolysis, or the Fenton reaction [8–11]. Accelerated test systems mainly include lipids, which rapidly oxidize in order to simulate a long induction period in a short time. To accelerate oxidation, an increase in temperature is often used. Other methods employ metals as catalysts or generate radicals using azo-initiators [12, 13]. These methods are based on different chemical and physical principles in monitoring oxidation; thus the activity of antioxidants may vary according to the assay used [12, 14]. This study compares assays used in the laboratories of the authors for the assessment of the activity of antioxidant extracts. Extracts obtained by different foodrelevant extraction procedures were compared for their composition of major antioxidants, which should be indicative of extraction yields and antioxidant potency.

Quantification of antioxidants by HPLC Pure compounds were used as references. Extracts were dissolved in methanol, aqueous methanol, or mixtures of ethanol/hexane (1 : 1). A filtration was performed when necessary and 10 or 20 ml of the resulting solutions were analyzed by HPLC. Antioxidants from spices and tea extracts were separated on a Merck system (L6200 A and AS 2000 A, Darmstadt, Germany) using a TSK gel ODS 80 TM column (250!4.6 mm, 5 mm particle, Toso Haas, Stuttgart, Germany) combined with a LiChroCART 4-4 precolumn cartridge (Lichrospher 100 C18-5 encapped, Merck). The mobile phase consisted of 0.5% H3PO4 (concentrated, 85%) aqueous solution (solvent A) and acetonitrile with 0.5% H3PO4 (concentrated, 85%) as solvent B. Various gradients of solvents A and B were applied at a flow rate of 1.0 ml!min –1 and at 40 7C. Antioxidants were detected by UV at 280 nm (JASCO UV 970, Tokyo, Japan). Antioxidants from coffee were separated on a ChromCart Cartridge System, 250!4 mm (Nucleosil, C-18, particle size 5 mm, Macherey Nagel) combined with a LiChroCART 4-4 precolumn cartridge (Lichrospher 100 C18-5 encapped, Merck). The mobile phase consisted of 0.1% trifluoroacetic acid (solvent A) and acetonitrile with 0.1% trifluoroacetic acid (solvent B). Various gradients of solvents A and B were applied at a flow rate of 1.0 ml!min –1. Antioxidants were detected by UV at 325 nm (Hewlett Packard HP-1090 A HPLC system equipped with a diode array detector, Urdorf, Switzerland). The composition of the grape skin extract was provided by the producer. All HPLC analyses were performed at least in duplicates. Determination of the total phenol content Extracts were dissolved in phosphate buffer (50 mM; pH 6.75) and determination of total phenolic content was carried out according to Amerine and Ough [15] using a Folin-Ciocalteu reagent with analytical grade phenol as standard.

Material and methods Chemicals and lipid materials used in the assays Fremy’s salt radicals (potassium nitrosodisulfonate), Tween 20, linoleic acid, heart metmyoglobin (MMb, type III), and a-tocopherol were purchased from Sigma (St. Louis, MO, USA). Troloxt, DMPO (5,5-dimethyl-1-pyrroline N-oxide; ESR silent), DPPH (a,a-diphenyl-b-picrylhydrazyl), and galvinoxyl were from Aldrich (Milwaukee, WI, USA) and methyl linoleate was from Nu Chek Prep (Elysian, MN, USA). Pork lard without antioxidant addition was purchased from a local supermarket and used after filtration. All other chemicals commercially obtained were of analytical grade or HPLC grade and used without further purification. The water used was purified by deionization or double distillation. Producing extracts and material Mechanical extracts from spices (a mixture of ginger, turmeric, and cayenne pepper, or a mixture of rosemary, sage, thyme, and oregano) were obtained by pressing on a hydraulic laboratory press using MCT (medium-chain triglycerides) as carrier. LicosaP extracts were prepared similarly by pressing either green tea, spice mixture, or rosemary using propylene glycol as carrier. Freeze-dried coffee extracts were produced on a pilot scale by aqueous extraction of roasted coffee blend while varying the extraction temperature. Herbor-rosemary extracts were commercially available from FIS S.A. (Chatel-St-Denis, Switzerland) and were made either by extraction with hydrophilic or lipophilic solvents. The grape skin extract was obtained from Unilever (Vlaardingen, Netherlands) and a slurry of tomato peel from Nestlé Research Centre.

Electron spin resonance spectroscopy (ESR) assay based on Fremy’s salt and galvinoxyl radical scavenging (Fremy’s salt and galvinoxyl assays) The radical scavenging ability of the extracts was tested at an extract concentration of 250 mg!l –1 in an aqueous solution of Fremy’s salt radicals (1 mmol!l –1) or in ethanolic solutions of galvinoxyl radicals (0.5 mmol!l –1) by ESR according to Gardner et al. [16]. Scavenging ability towards DPPH radicals (DPPH assay) The radical scavenging ability of extracts was determined at concentrations of 13.3, 66.7, and 133.3 mg!l –1 in ethanolic DPPH solution (0.1!mmol l –1). Coffee extracts and grape skin extracts were dissolved in water; all other extract stock solutions were prepared in ethanol. The absorbance of DPPH was monitored spectrophotometrically at 516 nm before and 10 min after adding the extract. The 10-minute scavenging time was based on preliminary experiments, where the decrease in absorbance was tested with several antioxidants at a concentration of 13.5 mM over a period of 20 min. Within the first minutes the absorbance decreased rapidly and between 10 and 20 min a further decrease of not more than 9% was observed. ESR assay based on scavenging of hydroxyl radical generated in the Fenton reaction (FRBR assay) The radical scavenging activity of extracts was determined using a stock solution of 8.34 mg extract l –1 in phosphate buffer (5.0 mmol!l –1; pH 6.75) left at ambient temperature for 30 min-

321 utes to dissolve before analysis. The analysis was performed according to Madsen et al. [11] by ESR. Hydroxyl radicals were generated in the Fenton reaction in the presence of DMPO (9.8 mmol!l –1) in phosphate buffer (5.0 mmol!l –1; pH 6.75) using FeSO4 and H2O2 in final concentrations of 4.9 mmol!l –1 and 10–13 mmol!l –1, respectively. For each extract two concentrations were chosen, both resulting in a peak height lower and higher than 50% inhibition. The amount of extract needed for a 50% reduction of the control signal was interpolated from the average values of two sets of duplicates. Oxygen consumption of linoleic acid emulsion (oxygen assay) The oxygen consumption method was carried out according to [11, 17]. Extracts were dissolved in phosphate buffer (50 mmol!l –1; pH 6.75) and kept at ambient temperature for 30 minutes for dissolving before filtration (Whatman No. 41 filter paper). Extracts, which showed phase separation in phosphate buffer, were shaken vigorously before dilution. 100 ml extract solution were mixed with 100 ml HCl (0.1 mol!l –1), 100 ml linoleic acid emulsion (linoleic acid: 21 mmol!l –1; Tween 20 : 1.2 g!l –1), 100 ml myoglobin solution (0.20 mmol!l –1), and 4.60 ml air-saturated and prethermostated (25 7C) phosphate buffer (50 mM; pH 6.75) and placed in a measuring cell without headspace. The oxygen concentration was recorded at 5-second intervals for 20 minutes using a Clarke electrode [11]. Methyl linoleate oxidation (CD assay)

of conjugated diene absorption at 234 nm. The spectrophotometer was set to zero with isooctane. The amount of hydroperoxides was calculated using an molar extinction coefficient of p26,000 [18]. Induction period of lard oxidation (Rancimat assay) The induction period of lard was determined with the Rancimat method at 100 7C and an airflow of 15 l!h –1 [19]. Each reaction vial contained 3.5 g lard which was filtered prior to the addition of extract at 60 7C. Extracts were added in a final concentration of 400 mg!g –1 dissolved in 200 ml water or methanol as a carrier. The reaction vials were vortexed for 20 s immediately before starting the Rancimat measurement. The carrier was evaporated immediately after inserting the reaction vials into the Rancimat. Statistical analysis Significant differences between the samples were calculated by Minitabr software (Addision-Wesley Publishing Company, Reading, MA, USA) by one-way analysis of variance (ANOVA) using a significance level of p~0.05 for Fisher’s LSD procedure if not stated otherwise. The significance level (p) of correlation coefficients was related to the critical values for the Pearson product moment correlation coefficient.

Results

Extracts were dissolved in methanol and added to 200 mg methyl linoleate, resulting in final concentrations of 100 and 400 mg!g –1, followed by methanol evaporation under nitrogen. Oxidation was carried out in open 2 ml vials in the dark at 40 7C. Aliquots of the sample (10 mg) were taken at regular intervals and dissolved in 5 ml isooctane for spectrophotometric (Perkin-Elmer lambda 15 UV-VIS spectrophotometer, Norwalk, CT, USA) measurements

Antioxidant concentration in extracts Extracts from different food-grade plant material were analyzed for their total phenol concentration (Table 1), and the composition of major phenolic compounds (Ta-

Table 1 Extracts from spices, tea, coffee, grape skin, and tomato peel slurry. Plant material, extraction procedure and concentration of total phenols Extract Extract no

Plant material

Extraction [a]

Total Phenol (mmol!g –1)

S1

SC Oriental 30

Mechanical MCT extract

0.05

S2

Licosa-P/SC Oriental

Mechanical propylene glycol extract

0.15

S3

SC Provencal 29

Mechanical MCT extract

0.05

R1 R2

Herbor P31 CCMK Herbor H41 CCIK

Ginger, tumeric and Cayenne pepper (2 : 1 : 1) Ginger, tumeric and Cayenne pepper (2 : 1 : 1) Rosemary, sage, thyme origano (2 : 1 : 1 : 1) Rosemary Rosemary

0.92 0.27

R3 R4 T1 T2 T3 C1 C2 C3 C4

Herbor O25 CVRUC Licosa-P/Rosmarin Licosa-P/Thé Chinosis Licosa-P/Thé Sencha Licosa-P/Thé Indien 113.999.01 113.999.02 113 999.10 113.001.02

Rosemary Rosemary Chinese green tea Japanese green tea Indian green tea Arabica, robusta coffee Arabica, robusta coffee Arabica, robusta coffee Arabica, robusta coffee

C5 Gr ToS

113.001.10 Arabica, robusta coffee (4 : 1), roasted Grape skin extract P2157 Grape skin Tomato peel slurry Tomato peel

Hydrophilic solvent extract (Powder) Hydrophilic solvent extract, similar to R1; additional use of gums and emulsifiers Lipophilic solvents Mechanical propylene glycol extract Mechanical propylene glycol extract Mechanical propylene glycol extract Mechanical propylene glycol extract 1st aqueous extract, T~110 7C 2nd aqueous extract, T 1 110 7C Aqueous extract C1,C2 combined 2nd aqueous extract; similar to C2 but at a higher temperature Aqueous extract C1and C4 combined

[a]

MCTpmedium-chain triglycerides

(4 : 1), (4 : 1), (4 : 1), (4 : 1),

roasted roasted roasted roasted

0.09 0.61 0.57 0.56 0.69 1.23 0.73 1.1 0.70 1.1 1.6 0.005

322 Table 2 Concentration (%) of the principal antioxidants in extracts [a] from spice mixtures and rosemary Extract

Gingerol (%; BSD)

Curcumin (%; BSD)

Carvacrol (%; BSD)

Thymol (%; BSD)

Carnosol (%; BSD)

Carnosic acid (%; BSD)

Rosmarinic acid (%; BSD)

Total (%)

S1 S2 S3 R1 R2 R3 R4

0.56 (B0.01) 0.36 (B0.01) – – – – –

0.29 (B0.01) 0.36 (B0.01) – – – – –

– – 0.05 (B0.02) – – – –

– – 0.24 (B0.04) – – – –

– – 0.35 3.69 1.05 0.65 0.28

– – 1.67 4.78 1.36 3.65 1.93

– – – 3.78 (B0.33) 2.18 (B0.04) – 1.05 (B0.04)

0.85 0.72 2.31 12.25 4.59 4.3 3.26

[a]

(B0.09) (B0.27) (B0.13) (B0.04) (B0.03)

(B0.11) (B0.07) (B0.11) (B0.32) (B0.12)

For abbreviation of extracts see Table 1

Table 3 Concentration (%) of major antioxidants in coffee extracts [a]

Caffeic acid Ferulic acid 3-CQA [b] 4-CQA 5-CQA Total CQA 3,4-di CQA 3,5-di CQA 4,5-di CQA Total di-CQA CQAcdi-CQA

C1 (%; BSD) [b]

C2 (%; BSD)

C3 (%; BSD)

C4 (%; BSD)

C5 (%; BSD)

0.084 0.009 0.990 1.106 1.981 4.077 0.128 0.050 0.071 0.249 4.325

0.009 n.d. 0.155 0.150 0.201 0.505 0.024 0.012 0.016 0.052 0.556

0.056 0.009 0.716 0.793 1.382 2.890 0.110 0.056 0.078 0.243 3.133

0.013 n.d. 0.111 0.135 0.169 0.415 0.009 n.d. n.d. 0.009 0.423

0.053 0.007 0.660 0.731 1.265 2.656 0.085 0.039 0.050 0.173 2.829

B0.004 B0.001 B0.003 B0.001 B0.001 B0.004 B0.004 B0.002 B0.001 B0.008 B0.004

B0.001 B0.004 B0.001 B0.005 B0.009 B0.001 B0.001 B0.000 B0.006 B0.046

B0.010 B0.003 B0.001 B0.006 B0.013 B0.018 B0.018 B0.018 B0.029 B0.064 B0.046

B0.001 B0.001 B0.001 B0.000 B0.001 B0.001

B0.001 B0.001

B0.006 B0.002 B0.005 B0.000 B0.000 B0.005 B0.010 B0.009 B0.008 B0.027 B0.022

[a]

For abbreviation of extracts see Table 1 SDpstandard deviation [c] CQApcaffeoylquinic acid [b]

Table 4 Concentration (%) of major antioxidants in green tea extracts [a] Extract

Catechin (%, BSD) [b]

Epicatechin (%, BSD)

Epicatechin gallate (%, BSD)

Epigallocatechin (%, BSD)

Epigallocatechin gallate (%, BSD)

Total (%)

T1 T2 T3

0.06 (B0.004) 0.11 (B0.007) 0.08 (B0.006)

0.32 (B0.02) 0.42 (B0.03) 0.44 (B0.03)

1.01 (B0.01) 1.13 (B0.01) 1.30 (B0.01)

3.87 (B0.15) 5.72 (B0.22) 7.36 (B0.28)

1.79 (B0.01) 2.24 (B0.01) 1.95 (B0.01)

7.05 9.62 11.13

[a] [b]

For abbreviation of extracts see Table 1 SDpstandard deviation

bles 2–5). Different extraction procedures were further investigated with respect to their influence on the composition of extracts from spice mixtures, rosemary, and coffee. Gingerol and curcumin displayed opposite extraction behavior when the extraction carrier changed from medium-chain triglycerides (S1) to propylene glycol (S2) (Table 2). The higher concentration of carnosic acid and carnosol in rosemary extract R1 compared to R3 and R4 can be understood as the effect of the complete evaporation of the extraction solvent resulting in a powder (R1) in contrast to R3 and R4 that still contained some solvent (Table 2). The concentration in R2 was lower compared to R1 due to the addition of gums and emulsifiers. The ratio of carnosic acid to carnosol ranged from 5.6 to 6.9 in R3 and R4, but in R1 and R2 it was almost five times lower, which may indicate conversion of carnosic acid to carnosol during extraction in the latter extracts. Based on its chromatographic behavior,

rosmarinic acid shows a higher polarity than carnosic acid and carnosol. This explains the lack of rosmarinic acid in R3, which was obtained by mechanical extraction with MCT. Table 5 Antioxidant composition of grape skin extract (Gr) Concentration (%) Gallic acid Caffeic acid Ferulic acid Catechin Epicatechin Quercetin Rutin Myricetin Cyanidin Chloride Malvidin Chloride Resveratrol Total

1.9 2.7 1.4 15.1 2.5 1.9 2.5 3.0 1.1 1.5 18.5 51.1

323

The total concentration of chlorogenic acids (CQA and di-CQA), caffeic and ferulic acid ranged from 0.44% to 4.4% in coffee extracts (Table 3). The differences are related to the extraction steps (Table 1), i.e., the first extraction step yields a higher concentration than the second. Raising the temperature did not result in a higher concentration. Differences in the composition of tea extracts are related to the source of the plant material, as all extracts were obtained using propylene glycol (Table 4). The composition of the grape skin extract as given in Table 5 was provided by the producer. The antioxidant composition of tomato peel slurry was not investigated, except for the total phenol concentration (Table 1). Comparison of radical assays The radical scavenging activity of the extracts was studied with Fremy’s salt, galvinoxyl, DPPH, and hydroxyl radicals, the last of these being formed in the Fenton reaction. Trolox and a-tocopherol were used as water-soluble and lipid-soluble reference antioxidants in order to indicate the range of activity which is expressed the by the different test systems (Table 6). In assays using stable radicals (Fremy’s salt, galvinoxyl, and DPPH), Trolox showed a similar response expressed as the number of radicals reduced at the same antioxidant concentration. The stoichiometric factors of Trolox for radical reduction in the Fremy’s, galvinoxyl, and DPPH assay amounted to 2.52, 2.29, and 1.65, respectively, whereas that of a-tocopherol was 1.65 in the DPPH assay. These results are in overall agreement with the stoichiometric factors of Trolox and a-tocopherol, which are normally considered to amount to approx. 2 [20]. Due to the high absorbance of DPPH at 516 nm a lower radical concentration was used in the DPPH assay. Therefore, the number of radicals reduced was extrapolated to allow comparison with the Fremy’s salt or galvinoxyl ESR assay at similar concentrations.

In the FRBR assay the antioxidative activity is measured by the competition between antioxidant and spin trap (DMPO) to scavenge hydroxyl radicals [11]. The activity is expressed by the extract concentration required to reduce the DMPO signal height by 50%, rather than the number of radicals scavenged. Thus, an increase in the activity is indicated by a decrease in the concentration of antioxidant extract required. Comparison of lipid oxidation assays Three different assays were carried out based on oxygen consumption, formation of conjugated dienes, and the formation of volatile oxidation products (Table 6). Trolox and a-tocopherol were also used as water-soluble and lipid-soluble reference antioxidants. In the oxygen consumption assay the antioxidant activity was expressed as the ratio (IOx) of the slopes (v) of initial oxygen consumption (i.e., v between 90% and 80% O2 relative to the initial oxygen concentration) in the presence and the absence of an antioxidant. The ratios of induction periods recorded with and without antioxidant reflected the antioxidant activity of extracts in the Rancimat assay. In the CD assay, the inhibition of oxidation products relative to the control was measured at two different points of time during the course of oxidation (45 h and 96 h). Accordingly the CD assay and Rancimat show higher activity by enhanced values, whereas higher antioxidant activity in the oxygen consumption assay is indicated by a decrease in the ratio. Antioxidant concentration in rosemary and coffee extracts vs. total phenol concentration or vs. antioxidant activity As would be expected, when comparing all plant extracts together there is no simple correlation between the concentration of the major antioxidants and total phenol concentration or the antioxidant activity. Linear

Table 6 Activity of Trolox (Trox) and a-tocopherol (Toc) as antioxidant standards in radical scavenging and lipid oxidation assays Assay Fremy’s salt Galvinoxyl DPPH DPPH DPPH FRBR Oxygen consumption CD formation Rancimat [a]

Trox, Toc concentration Expression of antioxidant activity –1

Trox 200 mmol!l Trox 200 mmol!l –1 Trox 13.3 mmol!l –1 Trox 200 mmol!l –1 Toc 13.3 mmol!l –1 Trox 9.8 mg Trox 1.557 mg Toc 50 mmol!kg –1 Trox 200 mmol!kg –1

Number of radicals reduced Number of radicals reduced [b] Number of radicals reduced [c] Number of radicals reduced [d] Number of radicals reduced [c] Reduction of the DMPO signal intensity (%) [e] Ratio of oxygen consumption (IOx) [f] Inhibition of conjugated diene formation [g] Ratio of induction periods [h]

Reaction time: 20 min Reaction time: 5 min [c] Reaction time: 10 min [d] Extrapolated value based on a Trolox concentration of 13.3 mmol!l –1 [e] Reaction time: 2 min [b]

[a]

Activity

SD 20

3.02!10 2.75!10 20 1.32!10 19 1.98!10 20 1.31!10 19 50 0.98 0.98 1.99

(B7!10 17) (B6!10 17)

[f] Ratio (IOx) of the oxygen depletion slopes (v); IOxpv (90–80% O2) with extract present/v (90–80% O2) without extract present [g] Inhibition of conjugated dienes relative to the control (1p100%) [h] Ratio (R) of the induction period; Rpinduction period with extract (h)/induction period of the control; (^1 : no inhibition)

324 Table 7 Correlation coefficients (r) for linear regression analysis of major antioxidants concentration in rosemary (Table 2) and coffee extracts (Table 3) vs. total phenol concentration or activity in different antioxidant assays Assay

Rosemary extracts

Coffee extracts

Total phenols Fremy’s Galvinoxyl DPPH FRBR Oxygen CD Rancimat

0.734 [b] – 0.999 [a] 0.940 [b] P0.307 [b] P0.563 [a] 0.968 [b] 0.900 [b]

0.997 [c] 0.987 [c] – 0.998 [c] P0.934 [c] P0.729 [c] 0.839 [c] P0.882 [c]

[a] [b] [c]

np3 np4 np5

high correlation was found for galvinoxyl, DPPH, CD, and Rancimat assays vs. antioxidant concentration in rosemary extracts. For coffee extracts, all correlation coefficients of radical assays vs. antioxidant concentration were significant (p~0.01), whereas correlation coefficients of lipid oxidation assays vs. antioxidant concentration were clearly lower. In the case of the Rancimat assay the increase in the induction period was negatively correlated with antioxidant concentration of the coffee extracts. Correlation coefficients for rosemary extracts can only be expected to provide trends as they are based on only 3 or 4 extracts. Radical scavenging ability of extracts

regression analyses were performed for groups of extracts (i.e., rosemary and coffee extracts) separately in order to compare similar antioxidant patterns (Table 7). The correlation coefficient (r) of the total phenol concentration vs. concentration for rosemary extract was clearly lower than for coffee extracts. By contrast,

Compared to Trolox, all the extracts showed lower activity in the Fremy’s, galvinoxyl, and DPPH assays at the same concentration (based on weight), but in most cases higher activity in the FRBR assay (Table 8). All radical scavenging assays showed similar trends in the ranking of antioxidant activity within the groups of tea and coffee extracts (Table 8). Significant correlations (p~0.01) were found for the three radical scavenging assays with respect to coffee extracts (Table 9).

Table 8 Radical scavenging activity [a] of extracts in different assays Fremy’s salt [b] np3 Spice extracts S1 S2 S3 Pooled SD Rosemary extracts R1 R2 R3 R4 Pooled SD Coffee extracts C1 C2 C3 C4 C5 Pooled SD Tea extracts T1 T2 T3 Pooled SD Grape skin extract Gr SD Tomato slurry ToS SD [a]

7.23!10 18 1.56!10 19 3.8!10 19 0.32!10 19

2.33!10 19 0.40!10 19 2.13!10 20 1.38!10 20 1.3!10 20

Galvinoxyl [c] np3

a b

4.12!10 20 1.46!10 20 1.05!10 20

b

a

a b c

b

9.27!10 18 2.23!10 20 9.64!10 19 1.68!10 20 1.05!10 20 1.67!10 20 1.22!10 18 1.29!10 20 1.4!10 20 1.62!10 20 4.67!10 18

a d b c b

c b a

1.79!10 20 1.85!10 20 2.13!10 20 2.33!10 18

3.55!10 20 2.55!10 18 P3.18!10 17 9.37!10 17

c b a

DPPH [d] n62 3.40!10 18 3.78!10 18 6.29!10 18 1.79!10 18

b

4.78!10 19 2.36!10 19 1.14!10 19 1.92!10 19 6.58!10 17

a

3.68!10 19 1.38!10 19 2.99!10 19 1.38!10 19 2.89!10 19 6.50!10 17

a

3.13!10 19 3.22!10 19 3.76!10 19 3.72!10 17

b

4.07!10 19 5.20!10 16 1.45!10 18 3.29!10 18

Values followed by different letters (a, b ,c, d) are significantly (p~0.05) different from each other. For abbreviation of extracts see Table 1 [b] Number of radicals reduced after 5 min; extract concentration: 250 mg!l –1

[c]

P1.06!10 18 1.22!10 18

b a

b d c

c b c b

b a

FRBR [e] np4 100 1.2 26 6.85

c

0.96 2.1 22 0.93 0.27

a

0.21 1.1 0.38 0.79 0.37 0.09

a

0.31 0.29 0.23 0.03

b

a b

b c a

d b c b

b a

0.35 0.02 9.5 0.6

Number of radicals reduced after 20 min; extract concentration: 250 mg!l –1. [d] Number of radicals reduced after 10 min; extract at concentration: 66.7 mg!l –1 [e] mg of the extract required to achieve 50% reduction of the radical signal height

325 Table 9 Correlation (r) between radical scavenging and lipid oxidation assays applied for antioxidant assessment of coffee extracts [a] Assay

Fremy’s r (np5)

Fremy’s DPPH FRBR Oxygen CD

0.985 * P0.981 * P0.798 * 0.823 *

[a]*

p~0.05,

**

DPPH r (np5)

P0.994 ** P0.698 0.785

FRBR r (np5)

0.537 P0.923 *

Oxygen r (np5)

CD r (np5)

Rancimat r (np5) 0.904* 0.894* 0.859* 0.604 0.734

P0.483

p

Table 10 Correlation (r) between the radical scavenging and lipid oxidation assays applied to the antioxidant assessment of rosemary extracts Assay Galvinoxyl Fremy’ s DPPH FRBR Oxygen CD

Galvinoxyl r (np3)

Fremy’s r (np3)

DPPH r (np4)

FRBR r (np4)

Oxygen (np4)

– 0.978 P0.640 0.715 0.799

0.998 P0.401 0.875 0.700

P0.612 0.807 0.724

Trends for the ranking of spice and rosemary extracts obtained in the DPPH assay were similar to those found in the Fremy’s and galvinoxyl assays, but the results of the FRBR assay differed markedly (Table 8). The order of activity for S1–3 extracts monitored by DPPH assay and data from the Fremy’s or galvinoxyl assay reflected the marked difference in the concentration of antioxidants between S3 and S1–2. Also, S3 showed higher activity than S1 in the FRBR assay, but the propylene glycol extract S2 exhibited stronger activity than S3. Rosemary extract R1 exceeds R2–4 in the antioxidant concentration by a factor of 2.7 to 3.8 and exhibited the strongest activity in the DPPH, Fremy’s, and galvinoxyl assays (Table 8). A marked difference was also observed between R1 and R2–3 in the FRBR assay, but in contrast to the other assays the propylene glycol extract R4 showed the highest activity of all rosemary extracts. Consequently, the correlation coefficient calculated for results from radical assays for rosemary extracts (Table 10) was rather low, except for DPPH assay vs. Fremy’s and galvinoxyl assays. As in Table 7, correlation coefficients for rosemary extracts are only intended to visualize differences, as they are based on a low number of extracts.

Antioxidant activity of extracts in lipid oxidation assays. At a concentration of 400 mg!g –1 most extracts showed over 90% inhibition of conjugated diene for-

P0.939 P0.231

0.549

CD r (np4)

Rancimat r (np4) 0.999 0.894 0.799 P0.094 0.426 0.936

mation after 45 h. After 96 h only S2, R3, C2 and T1 showed lower inhibition than 90%. Lowering the extract concentration to 100 mg!g –1 caused a decrease in the inhibition to less than 90% for all extracts after 45 h except for R1 and C1. Both of these extracts also showed almost complete inhibition after 96 h, indicating that the induction period of oxidation was not finished at this stage. To enable sufficient differentiation between the extracts, Table 11 compares the inhibition of extracts at a concentration of 100 mg!g –1 after 45 h. By contrast, extracts tested by the Rancimat method at a concentration of 100 mg!g –1 in lard showed too low activity to obtain clear differences. Thus values listed in Table 11 refer to an extract concentration of 400 mg!g –1 lard. Ranking antioxidant activities within the different extract groups (S, R, C, T) resulted in different trends depending on the assay (oxygen, CD, and Rancimat) used. The CD and Rancimat assay showed a higher activity for S3 than for S1 and S2, and thus reflected the concentration of antioxidants in the spice extracts (S1–3). No activity was observed for these extracts in the oxygen assay. Results of the Rancimat assay are well correlated with the CD assay with respect to the antioxidant activity of rosemary extracts, but both assays showed low correlation with the oxygen assay (Table 10). Among coffee extracts C1 showed the highest activity in the oxygen and CD assays, but the order of activity for C2–5 was somewhat different in these two assays and resulted in a low correlation (Table 9). Compared to oxygen and CD assays, a completely different order of activity for coffee extracts was found with the Rancimat assay. All lipid oxidation assays showed high activity for tea extract T2 and significantly lower activi-

326 Table 11 Inhibition of lipid oxidation by extracts [a] in different assays Oxygen consumption [b] np2 Spice extracts S1 S2 S3 Pooled SD Rosemary extracts R1 R2 R3 R4 Pooled SD Coffee extracts C1 C2 C3 C4 C5 Pooled SD Tea extracts T1 T2 T3 Pooled SD Grape skin extract Gr MD Tomato slurry ToS MD

[f]

– – [f] – [f] 0.94 0.79

a a

0.6 0.21

a

0.08 0.6 0.56 0.73 0.69 0.12

a

0.53 Compl. [e] 0.61 0.06

b b b b

b a b

Conjugated Rancidiene mat [d] Inhibition [c] np2 np2 0.235 0.232 0.392 0.018

b

0.946 0.860 0.645 0.459 0.017

a

0.899 0.299 0.724 0.684 0.744 0.032

a

0.338 0.678 0.628 0.006

c

b a

b c d

c b b b

a b

1.08 1.16 1.27 0.01

c

2.22 1.72 1.62 1.17 0.03

a

b a

b c d

1.27 a 1.36 a 1.33 a 1.36 a 1.28 a 0.004 1.35 1.29 1.02 0.03

0.13 0.015

0.85 0.02

1.09 0.01

– [f]

0.16 0.005

0.86 0.01

– Oxygen:T2 1 1 C1 1 Gr 1 R4 1 S, ToS – CD 45 h: R1 1 C1 1 Gr 1 T2 1 S3 1 ToS – Rancimat: R1 1 T1 1 C 1 S3 1 Gr 1 ToS (prooxidant) As some extracts were not soluble in water or ethanol at concentrations useful for the Fremy’s and galvinoxyl assays, these extracts were analyzed using only the more convenient assay, i.e. the result of the S3 extract was obtained with the galvinoxyl assay whereas the activities of the Gr, C1, R1, T3, and ToS extracts were determined using the Fremy’s assay. It is reasonable to assume that the activity of S3 is lower than those of R1, C1, and T3, as Fremy’s assay gives in general lower activity for extracts than the galvinoxyl assay. Also the mechanism for galvinoxyl (G =/GH) and the nitroso radical Fremy’s salt (NO =/NOH) is entirely analogous [21].

a b

ty for T3, in spite of a higher total concentration of the major antioxidant in the latter extract (Table 4). Overall ranking of the extracts and slurry The ranking of the antioxidant activity of the most potent extracts of each group resulted in the following orders for the different assays used:

– Fremy’s and galvinoxyl: Gr 1 1 C1 1 R1 1 T3 1 S3 1 ToS

Lipid oxidation assays

a

[a] Values followed by different letters (a,b,c,d) are significantly (p~0.05) different from each other. Pooled SD, pooled standard deviation; MD, deviation from the mean. For abbreviation of extracts see Table 1. [b] Ratio (IOx) of the oxygen depletion slopes (v); IOxpv (90–80% O2) with extract present/v (90–80% O2) without extract present. Index of C1 was calculated for the 100–95% and 90–82% interval of the initial oxygen concentration. [c] Inhibition of conjugated dienes relative to the control (1p100%). [d] Ratio (R) of the induction period; Rpinduction period with extract (h)/induction period of the control; (^1 : no inhibition) [e] complete inhibition [f] no inhibition

Radical assays

– DPPH: R1 1 Gr 1 T3 1 C1 1 S3 1 ToS (prooxidant) – FRBR: C1 1 T3 1 Gr 1 R4 1 S2 1 ToS

Discussion The results obtained with extracts from spices, tea, coffee, grape skin, and tomato peel slurry demonstrate that the extraction procedures strongly influence the composition of the extracts and their antioxidant activity. This effect has already been demonstrated by several authors for rosemary and other spices [e.g., 6, 19, 22]. The correlation (rp0.734–0.997) between the concentration of principal phenolic antioxidants and the total phenol concentration within the groups of coffee and rosemary extracts indicates that the major antioxidants function as markers for the extraction effectiveness. The total phenol concentration of extracts includes in addition to the principal phenolic antioxidants all the compounds that function as hydrogen donors in the plant material. In extracts from roasted coffee Maillard reaction products are present, in addition to derivatives of phenolic acids, such as ferulic and caffeic acids, whose antioxidant properties have been well documented [23, 24]. Extracts obtained using propylene glycol as a carrier (R4, S2) showed higher total phenol concentration in spite of the same or lower major antioxidant concentration compared to S1 and R3. The reason may be that propylene glycol like glucose erroneously contributes to the concentration of total phenols determined with Folin-Ciocalteu’s reagent [15]. All assays (based on radical scavenging and lipid oxidation) showed the highest activity for C1 amongst the coffee extracts and the majority of the test systems showed the highest activity for R1 within the rosemary extracts. Also, both extracts were amongst the most potent extracts in the overall ranking of the antioxidant activity. In addition, all assays displayed no or lowest

327

activity for the tomato peel slurry (ToS). Despite these very similar trends, marked variations in the resulting activity of the other extracts were found between the different assays.

effect with solvents or carriers would be similar for all extracts within the same group. Lipid oxidation assays

Radical assays The differences between the FRBR and the other radical scavenging assays may be explained by the nature and the generation of the radicals and by the preparation of the extracts. In the FRBR assay hydroxyl radicals are generated by the Fenton reaction in the presence of a spin trap DMPO [11]. Hydroxyl radicals have high reactivity with aromatic compounds [8] and with components which do not act as an electron donor [9]. By contrast, DPPH, Fremy’s salt and galvinoxyl are stable radicals with a low deterioration rate and reactivity towards most compounds. Consequently, only good H-atom donors are likely to react with stable radicals in a stoichiometric way [10, 16]. In addition, antioxidants which are effective chelators of transition metal ions may contribute differently to the antioxidant response in the FRBR assay compared to the assays using stable radicals, as Fe 2c/Fe 3c is the active redox couple in the Fenton reaction. The difference in the activity of the extracts related to Trolox indicates differences in the mode of antioxidant action between assays using stable radicals and the FRBR assay. Discrepancies in the antioxidant activity using different stable radicals were shown [16] to be related to different redox potentials and steric properties of the radicals. Solvents used to dissolve extracts had to be changed according to the extract solubility in assays using Fremy’s salt, galvinoxyl, and DPPH radicals. In contrast, all extracts were analyzed using the same solvent in the FRBR assay. This difference in solvents may have caused differences in the antioxidant pattern between the two groups of assays, since it has been shown that the solvent may affect the hydrogen donating ability of the antioxidant [25]. The carrier used in the extraction process is another parameter that may influence the antioxidant ranking obtained with the FRBR assay compared to the other radical scavenging assays. Extracts R4 and S2 were obtained by propylene glycol, which may interact directly with the hydroxyl radicals as has been described for other polyols [26] whereas stable radicals show a too low redox potential to oxidize polyols [10]. The good agreement of all radical assays for coffee and tea extracts is most likely due to the extraction procedure, as these extracts were either obtained by water (coffee) or by propylene glycol (tea) as an extraction solvent, in contrast to the different extraction excipients used for spices and rosemary. Thus, in the case of coffee and tea extracts, the antioxidant pattern does not change in a wide range and a possible interfering

Lipid oxidation was monitored in different ways and thus focused on different steps in the oxidation process. The formation of conjugated dienes in the CD assay reflects the concentration of primary oxidation products. In the Rancimat assay, conductivity increases due to secondary volatile oxidation products, which are formed at high temperatures simultaneously with hydroperoxides [27]. The oxygen consumption reflects the total degree of oxidation of the lipid material and responds to both the formation of primary oxidation products and further oxidation of secondary oxidation products [28]. In spite of the experimental differences, the Rancimat assay and the CD assay were found to show similar trends when comparing extracts of different groups and when ranking the most active extracts of each group. Differences observed for the grape skin extract and the coffee extracts could mainly be related to an immediate precipitation of extracts dissolved in water in the Rancimat assay. This interference is clearly shown by the inverse correlation between antioxidant activity and antioxidant concentration in the Rancimat assay (Table 7), which was in contrast to the other assays. Experiments with the oxygen consumption assay resulted in high antioxidant activity for coffee, green tea, and grape skin extracts. This indicates that the hydrophilic extracts, which contain more polar antioxidants, have the best response in this test system. It is note worthy that the results from the oxygen consumption assay are best correlated with the FRBR assay. Both tests generally showed high activity for Licosa-P and other polar extracts. This may indicate that polar extracts are particularly active in inhibiting metal-mediated lipid oxidation. However, it has to be considered that for both assays a similar sample preparation was used, including phosphate buffer, which may have caused solubilization according to the polarities of the antioxidants and a better transfer of water-soluble antioxidants.

Conclusion Results from the CD assay correspond well with radical assays using Fremy’s salt, galvinoxyl, and DPPH. Therefore, it can be concluded that scavenging by stable radicals is a suitable method for predicting the inhibition of primary oxidation product formation by natural extracts in homogenous food systems. Similar trends for FRBR and oxygen assays indicate that the FRBR assay provides activity patterns that are relevant for metal-catalyzed oxidation processes in water-containing food systems.

328

The validation of analytical methods for the determination of the activity of the extracts together with the development of efficient extraction procedures are important tools in making optimum use of antioxidants from natural sources for different types of foods. Assays for the antioxidant activity are of importance in selecting effective antioxidant extracts to be used in foods. Acknowledgements The study has been carried out with financial support from the Commission of the European Communities, Agriculture and Fisheries (FAIR) specific RTD program CT950158 “Improving the quality and nutritional value of processed foods by optimal use of food antioxidants”. It does not necessarily reflect its views and in no way anticipates the Commission’s future policy in this area.

References 1. Krishnakumar V, Gordon I (1996) Intern Food Ingredients 5 : 41–44 2. Hertog MGL, Feskens EJM, Hollman PCH, Katan MB, Kromhout D (1993) Lancet 342 : 1007–1011 3. Knekt P, Järvinen R, Reunanen A, Maatela J (1996) BMJ 312 : 478–481 4. Aeschbach R, Philippossian G (1989) European Patent Application EP 0 307 626 A 1 5. Nakatani NR, Inatani T, Koishi T (1984) US Patent 4,450,097 6. Weinreich B (1989) Untersuchungen zur Struktur und zum Extraktionsverhalten antioxidativ wirksamer Inhaltsstoffe aus Rosmarinus und Salvia officinalis. Diss. Univer. München 7. Aeschbach R, Bächler R, Rossi P, Sandoz L, Wille HJ (1994) Fat Sci Technol 96 : 441–443 8. Bors W, Heller W, Michel C, Saran M (1990) Methods Enzym 186 : 343–355

9. Halliwell B, Gutteridge JMC (1989) Free Radicals in Biology and Medicine, 2nd edn, Claredon Press, Oxford 10. Blois MS (1958) Nature 181 : 1199–1200 11. Madsen HL, Nielsen BR, Bertelsen G, Skibsted LH (1996) Food Chem 57 : 331–337 12. Ragnarsson JO, Labuza TP (1977) Food Chem 2 : 291–308 13. Barclay LRC, Locke SJ, MacNeil JM, VanKessel J (1984) J Am Chem Soc 106 : 2479–2481 14. Frankel EN (1993) Trends Food Sci Techn 41 : 220–225 15. Amerine MA, Ough CS (1980) Methods for Analysis of Musts and Wines. John Wiley, New York, pp 181–184 16. Gardner PT, McPhail DB, Duthie GG (1998) J Sci Food Agric 76 : 257–262 17. Mikkelsen A, Sosnieki L, Skibsted LH (1992) Z Lebensm Unters Forsch 195 : 228–234 18. Fishwick MJ, Swoboda PAT (1977) J Sci Food Agric 28 : 387–393 19. Schwarz K, Ternes W, Schmauderer E (1992) Z Lebensm Unters Forsch 195 : 104–107 20. Barclay LRC, Artz JD, Mowat JJ (1995) Biochim Biophys Acta 1237 : 77–85 21. Scott G (1993) Autoxidation and Antioxidants: Historical Perspectives Oxidation and Stabilisation of Polymers during Processing. In: Scott G (ed), Atmospheric Oxidation and Antioxidants, Vol 2. Amsterdam, pp 141–218 22. Chang SS, Ostric-Matijasevic B, Hsieh OAL, Huang CL (1977) J Food Sci 42 : 1102–1106 23. Nicoli MC, Anese M, Manzocco L, Lerici CR (1997) Lebensm Wiss Technol 30 : 292–297 24. Eichner K (1981) Progress Food Nutrition Sci 5 : 441–451 25. Avila DV, Ingold KU, Lusztyk J, Green WH, Procopio DR (1995) J Am Chem Soc 117 : 2929–2930 26. Cao G, Sofic E, Prior, RL (1996) J Agric Food Chem 44 : 3426–3431 27. Hadorn H, Zürcher K (1974) Deut Lebensm Rundschau 70 : 57–65 28. Belitz H-D, Grosch W (1992) Lehrbuch der Lebensmittelchemie. 4. überarb. Aufl., Berlin