Research Article The antioxidative effect of lipophilized rutin and dihydrocaffeic acid in fish oil enriched milk

Transcription

1 434 Eur. J. Lipid Sci. Technol. 2012, 114, Research Article The antioxidative effect of lipophilized rutin and dihydrocaffeic acid in fish oil enriched milk Ann-Dorit Moltke Sørensen 1, Lone Kirsten Petersen 1 *, Sara de Diego 2 **, Nina Skall Nielsen 1, Bena-Marie Lue 3 ***, Zhiyong Yang 3, Xubing Xu 3 and Charlotte Jacobsen 1 1 Division of Industrial Food Research, National Food Institute, Technical University of Denmark, Lyngby, Denmark 2 Department of Biotechnology and Food Science, University of Burgos, Burgos, Spain 3 Department of Engineering, Aarhus University, Århus, Denmark The antioxidative effect of phenolipids was evaluated in fish oil enriched milk emulsions as a model for a complex food system. Two different phenolipids modified from dihydrocaffeic acid (with C8 or C18:1) and rutin (with C12 or C16) were evaluated. Both dihydrocaffeate esters and rutin laurate showed significantly better antioxidant properties in milk emulsion compared with the original phenolics. However, rutin palmitate only performed slightly better as antioxidant than rutin. The results with rutin indicated that a cut-off effect exists in relation to the alkyl chain length with respect to optimal antioxidant activity in milk emulsions. Thus, the optimal alkyl chain length is at least below 16 carbon atoms, and maybe even less for rutin esters. For dihydrocaffeate esters it was not possible to conclude on a cut-off effect in relation to alkyl chain length and antioxidative effect due to the almost similar antioxidant effect of the two phenolipids. However, there was a tendency towards octyl dihydrocaffeate being slightly more efficient than oleyl dihydrocaffeate. Practical application: The finding that phenolipids are better antioxidants in milk emulsions than the original phenolic acid provides new knowledge that can be used to develop new antioxidant strategies to protect foods against lipid oxidation. However, the results indicate that both optimization of alkyl chain length for each type of phenolic, and optimization for each type of emulsion will be necessary in order to get the best oxidative stability of an emulsion with these phenolipids. Use of efficient antioxidants may lower the amount of antioxidant needed to protect against lipid oxidation and may in addition decrease the costs. Keywords: Caffeic acid / o/w Emulsion / Polar paradox / Rutin Received: October 10, 2011 / Revised: January 9, 2012 / Accepted: February 24, 2012 DOI: /ejlt Introduction The health beneficial effects of n-3 long chain PUFA (LC PUFAs) such as, e.g. reduced risk of cardiovascular diseases are well documented. During the last decade substantial Correspondence: Dr. Ann-Dorit Moltke Sørensen, Division of Industrial Food Research, National Food Institute, Technical University of Denmark, Søltofts Plads, Building 221, DK-2800 Kgs. Lyngby, Denmark adms@food.dtu.dk Fax: þ Abbreviations: ATD, automatic thermal desorber; BHT, butylated hydroxytoluene; DHCA, dihydrocaffeic acid; DHCA C18:1, oleyl dihydrocaffeate; DHCA C8, octyl dihydrocaffeate; EPA, eicosapentaenoic acid; LC, long chain; PV, peroxide value; Rutin C12, rutin laurate; Rutin C16, rutin palmitate efforts have been put into enriching foods with the healthy n-3 LC PUFAs as reviewed by Jacobsen et al. [1] These efforts have been carried out in order to increase the populations intake of especially eicosapentaenoic acid (EPA) and DHA [2, 3]. Despite the increasing number of n-3 PUFA enriched foods on the market, consumer acceptance and shelf-life of such products are still limited by the higher oxidative susceptibility of unsaturated lipids, which will lead to an unpleasant fishy off-flavour [4 7]. To retard lipid oxidation, a range of commercial synthetic antioxidants with free radical scavenging activity and metal chelating properties are *Current address: CP Kelco ApS, Ved banen 16, DK-4623 Lille Skensved, Denmark **Current address: Grupo Siro, Paseo Pintor Rosales 40, Madrid, Spain ***Current address: Novozymes A/S, Krogshoejvej 36, DK-2880 Bagsvaerd, Denmark

2 Eur. J. Lipid Sci. Technol. 2012, 114, Antioxidant efficacy of lipophilized phenolics 435 available, e.g. calcium disodium ethylenediaminetetraacetate (EDTA), butylhydroxytoluene (BHT) and propyl gallate. However, there is a trend in consumer preference for natural ingredients such as phenolic compounds rather than synthetic compounds. The major part of natural antioxidants from plants, are phenolic compounds, e.g. caffeic acid. Most food products are emulsions, and in these systems lipid oxidation is suggested to be initiated at the interface between the oil phase and the aqueous phase or air, and continued in the oil phase. In emulsions, antioxidants may mainly partition into three different phases: the aqueous phase, the oil phase and the interface between oil and water. Partitioning of antioxidants into the different phases is influenced by their polarity and interactions with other components present in the emulsion, e.g. emulsifier [8, 9]. Generally, phenolics are hydrophilic compounds and they will most likely be located in the aqueous phase. Furthermore, the polarity of antioxidants in bulk oil and emulsions has been considered to be decisive for their efficiency. This phenomenon is known as the polar paradox and states that hydrophilic antioxidants are more efficient in bulk oils than lipophilic antioxidants. In contrast, lipophilic antioxidants generally function better than hydrophilic antioxidants in emulsions [10]. However, recent studies have reported results that contradict the polar paradox hypothesis [11 14]. This suggests that other factors might be equally important, and more research is urgently required to improve our understanding about the relationship between the molecular structure of the antioxidants and their efficacy in different real food systems. Recently, several studies have reported the possibility of changing the polarity of phenolics by lipophilization with fatty acids of different chain length in order to improve their antioxidative effect in emulsified media. The current work in this area has been summarized by Shahidi and Zhong [15]. Laguerre et al. [11, 16] have recently reported on the antioxidative effect of lipohilized chlorogenic and rosmarinic acids. For chlorogenic acid the antioxidant capacity increased as the alkyl chain length was increased from 1 to 12 carbon atoms, whereas further increase of the chain length resulted in a drastic decrease in the antioxidant capacity. On the basis of these results the authors suggested a so-called cut-off effect related to the length of the lipid chain attached to chlorogenic acid, which they explained as follows: When, the hydrophobicity of the lipophilized compound increases to above a certain level, the lipophilized compound is suggested to form micelles in the aqueous phase. Thereby, they will not be available as antioxidants at the interface and in turn their efficacy will be reduced [11]. For rosmarinate esters, the octyl rosmarinate improved the antioxidative effect eight times compared to rosmarinic acid. Thus, the results lead to the conclusion that lipophilization with medium chain fatty acids is a promising way to increase the antioxidant activity [16]. However, the results obtained with chlorogenate and rosmarinate esters have also led to the suggestion that the exact location of antioxidants in the discontinuous phase, interfacial layer or oil droplets, is important for the activity of the antioxidants [11, 16]. In addition, a study on lipophilized dihydrocaffeic acids and their antioxidative effect in o/w emulsions reported by Sørensen et al. [17] suggested that lipophilized dihydrocaffeic acid tended to follow the newly suggested cut-off effect in relation to the alkyl chain length attached although only two chain lengths were evaluated. In contrast, lipophilized rutin added to o/w emulsions did not show a cut-off effect, since the esters, rutin laurate and rutin palmitate, were consistently less effective compared with the rutin [13]. However, only two chain lengths (C12 and C16) were evaluated along with rutin, hence alkyl chain shorter than C12 lipophilized on rutin may indicate a cut-off effect. From these studies it may be concluded that the cut-off effect is specific for the individual lipophilized phenolic compounds, i.e. the optimal chain length may vary between different phenolics. Recent research on lipophilized phenolic compounds has mostly paid attention to their production, in vitro antioxidant activity and their effect in simplified model systems, whereas studies on their effect in real food systems are lacking. In more complex systems such as real food the antioxidant behaviour might be influenced by interaction with other components present in the emulsions, e.g. emulsifier and iron [15, 18]. However, on the basis of studies in simple o/w emulsions we hypothesize that lipophilization of phenolics will increase their antioxidant efficacy in emulsified food enriched with n-3 PUFAs. Moreover, we hypothesize that the alkyl chain length will affect the antioxidative effect of these lipophilized phenolic compounds as was observed in a recent study with simple o/w emulsions [17], and that the effect of the alkyl chain length may be different from that observed in simple model systems due to the presence of other potentially interacting compounds. Therefore, the aim was to evaluate the antioxidative effect of dihydrocaffeic acid lipophilized with either octyl or oleyl alcohol and rutin lipophilized with lauric acid or palmitic acid in fish oil enriched milk. Fish oil enriched milk was chosen as previous studies have shown that it is highly susceptible to lipid oxidation [1]. Moreover, it is a complex food system in which antioxidants may interact with different compounds, e.g. proteins. Lipophilized rutin and dihydrocaffeic acid were chosen as antioxidants since this study is a continuation of recent studies performed in our lab in simplified o/w emulsions [12, 13, 17] and results could show different effects of lipophilization in simple versus complex emulsions. 2 Material and methods 2.1 Materials Fresh milk (0.5 and 1.5% fat content) was purchased in a local supermarket. Fish oil without antioxidant added was supplied by Maritex Norway (subsidiary of TINE BA,

3 436 A.-D. M. Sørensen et al. Eur. J. Lipid Sci. Technol. 2012, 114, Norway). This oil had an initial PV of 0.1 meq peroxides/kg oil, tocopherol content of 204 mg a-tocopherol, 102 mg g- tocopherol and 42 mg d-tocopherol/kg oil, and the fatty acid composition was as follows: 14:0, 3.0%; 16:0, 8.7%; 16:1, 8.2%; 18:0, 1.9%; 18:1, 20.9%; 18:2, 1.8%; 18:4, 2.6%; 20:1, 12.5%; 20:5 (EPA), 9.4%; 22:1, 5.9%; 22:5, 1.1% and 22:6 (DHA), 11.6%. The total percentages of n-3 and n-6 PUFA in the oil were 24.7 and 2.7%, respectively. Rutin (purity 98%), caffeic acid (purity 98%), dihydrocaffeic acid (purity 98%) and oleyl alcohol (purity 85%) were from Sigma Aldrich (Steinheim, Germany). Lipophilized rutin with lauric (C12) or palmitic (C16) acids, both with a purity of 98%, were synthesized at the National Food Institute, Division of Industrial Food Research (Technical University of Denmark, Lyngby Denmark). For further details about the lipophilization process refer to Lue et al. [19]. Lipophilized dihydrocaffeic acid with octyl (C8) with a purity of 80% or oleyl alcohol (C18:1) with a purity of 60% were synthesized at the Department of Engineering, Faculty of Science (Aarhus University, Århus, Denmark). Chemicals were from Merck (Darmstadt, Germany) and external standards for identification and quantification of secondary volatile oxidation products were all from Sigma Aldrich (Steinheim, Germany). All solvents were of HPLC grade and purchased from Lab-Scan (Dublin, Ireland). 2.2 Production of fish oil enriched milk Milk with 0.5% fat and with 1.5% fat was mixed (1:1 v/v) to obtain a total fat content of 1%. Subsequently, the milk was pasteurized at 728C for 15 s and the fish oil (0.5% v/v) and antioxidant were added (for specification on antioxidant addition, see Section 2.3). Emulsions were prepared in two steps: pre-emulsification and homogenization. During preemulsification, the heated milk with fish oil and antioxidant added was stirred with an Ultra-Turrax (Step 7, 1 min, Janke & Kunkel IKA-Labortechnik, Staufen, Germany). The preemulsion was then homogenized at a pressure of 25 and 250 bar with four circulations of the emulsion at RT using a table homogenizer from GEA Niro Soavi Spa (Parma, Italy). The emulsions (100 g) were stored in 100 ml blue cap bottles at 58C. Samples, one flask pr. code, were taken at Day 0, 3, 6, 9 and 12 and divided in brown glass bottles for different analysis and stored at 408C until analyses of peroxides, volatiles, tocopherols and fatty acids were performed. The droplet size was measured at Day 1, 6 and 12 without pre-freezing. 2.3 Experimental design The experimental design for Experiment 1 and 2 is described in details below and summarized in Table 1. Experiment 1: Five different antioxidants were evaluated, rutin, rutin laurate, rutin palmitate, dihydrocaffeic acid and oleyl dihydrocaffeate in a concentration of 100 mm in fish oil enriched milk. Dihydrocaffeic acid and oleyl dihydrocaffeate were added directly to the milk, whereas rutin, rutin laurate and rutin palmitate were first dissolved or suspended in 1.5 ml acetone due to dissolving problems in the milk emulsion. The acetone with antioxidant was then added to the heated milk. To obtain the same condition for all emulsions, 1.5 ml acetone was added to the milk emulsions with dihydrocaffeic acid, oleyl dihydrocaffeate and the control (no antioxidant added). Experiment 2: Four different antioxidants were evaluated, caffeic acid, dihydrocaffeic acid, octyl dihydrocaffeate and oleyl dihydrocaffeate. Antioxidant concentration tested was similar to that in Experiment 1 (100 mm). The synthesized Table 1. Experimental design Concentration of antioxidant Antioxidant applied Sample code mm mg/kg Experiment 1: Acetone addition (1.5 ml) Control Control Rutin Rutin Rutin laurate (C12) Rutin C Rutin palmitate (C16) Rutin C Dihydrocaffeic acid DHCA Oleyl dihydrocaffeate (C18:1) DHCA C18: Experiment 2: No acetone addition Control Control Dihydrocaffeic acid DHCA Octyl dihydrocaffeate (C8) DHCA C Oleyl dihydrocaffeate (C18:1) DHCA C18: Oleyl alcohol Oleyl alcohol Caffeic acid Caffeic acid

4 Eur. J. Lipid Sci. Technol. 2012, 114, Antioxidant efficacy of lipophilized phenolics 437 oleyl dihydrocaffeate was only 60% pure, and contained 40% free oleyl alcohol. Therefore, an emulsion with oleyl alcohol with the same amount of oleyl alcohol (29 mg/kg) as in the emulsion with oleyl dihydrocaffeate was included to evaluate the effect of oleyl alcohol on lipid oxidation in the milk emulsion. The antioxidants were dissolved directly in the heated milk before homogenization. 2.4 Droplet size determination The size of the fat droplets in milk emulsions was determined by laser diffraction with a Mastersizer 2000 (Malvern Instruments Ltd., Worcestershire, UK). A few droplets of the milk emulsion were suspended directly in re-circulating water (2800 rpm, obscuration 14 16%). The set-up used was the Fraunhofer method, which assumes that all sizes of particles scatter with equal efficiencies and that the particles is opaque and transmits no light [20]. The results were reported as surface mean diameter, P ni di 3 D 3;2 ¼ P ni di 2 where d is the diameter of individual droplets. 2.5 Measuring lipid oxidation Extraction of lipids Lipids were extracted from fish oil enriched milk (15 g) according to the method described by Bligh and Dyer [21] with reduced amount of solvent applied [22]. The analysis was done in duplicate and further used for determination of peroxide value, fatty acid composition and tocopherol concentration Primary oxidation products, peroxide value (PV) Peroxide value in the lipid extracts were determined by a colorimetric method based on formation of an iron thiocyanate complex measured according to the method described by Shanta and Decker [23], n ¼ Secondary volatile oxidation products dynamic headspace Volatiles were collected on Tenax TM tubes (Perkin Elmer, Norwalk, CT, USA) by purging the fish oil enriched milk (8 g) with nitrogen (150 ml/min, 30 min) at 458C. An ATD-400 automatic thermal desorber was used for thermally desorbing the collected volatiles. The transfer line of the ATD was connected to a 5890 IIA gas chromatograph (Agilent Technologies, Palo Alto, CA, USA) equipped with a DB wax column (length 30 m I.D mm 0.5 mm film thickness, J&W Scientific, CA, USA) coupled to a HP 5972A mass selective detector. Temperature program was as follows: 5 min at 458C, 1.58C/min from 45 to 558C, 2.58C/min from 55 to 908C, 128C/min from 90 to 2208C and hold for 4 min at 2008C. The MS was operating in the electron ionization mode at 70 ev and mass to charge ratios between 29 and 200 were scanned. For quantification of the different volatiles, solutions with external standards at different concentrations were prepared and analysed from milk with no fish oil added. The results are given in ng/g milk (n ¼ 3) Tocopherol concentration Lipid extract was evaporated under nitrogen, re-dissolved in heptane and analysed by HPLC (Agilent 1100 Series, Agilent Technologies, Palo Alto, CA, USA) according to the AOCS method [24] to determine tocopherol concentration in the different milk samples (n ¼ 4) Fatty acid composition Lipid extract was evaporated under nitrogen. First, the glycerol bound fatty acids were transesterified with methanolic NaOH (0.5 M). Then, hydrolytic released and free fatty acids were methylated by a boron trifluoride reagent (20%) catalysed process. Methyl esters were extracted with heptane followed by separation on GC (HP 5890A, Agilent Technologies, Palo Alto, CA, USA). The procedure was according to the AOCS methods [25, 26], n ¼ Sensory Preliminary sensory evaluation was performed by an expert panel composed of two persons. Each assessor evaluated one milk sample at a time. After each milk sample the assessors discussed the odour of the sample (only Experiment 2). 2.6 Data analyses Statistics The obtained results were analysed by two way ANOVA (GraphPad Prism, Version 4.03, GraphPad Software, Inc.). The Bonferroni multiple comparison was used to test differences between samples or storage time (significance level p<0.05) Inhibition percentages To compare the efficacy of the antioxidants in the two different emulsion systems, inhibition percentages were calculated. Inhibition ½%Š ¼ 1 Sample Antioxidant 100 Sample Control

5 438 A.-D. M. Sørensen et al. Eur. J. Lipid Sci. Technol. 2012, 114, Table 2. Droplet size D 3,2 [mm] for lipid droplets in fish oil enriched milk given as an average during storage (average W SD) and content of EPA and DHA (wt% of total lipids) at day 0 and 12 in the different milk emulsions EPA content (%wt of total lipids) DHA content (%wt of total lipids) Sample code Droplet size (mm) Day 0 Day 12 Day 0 Day 12 Experiment 1 Control Rutin < Rutin C < Rutin C DHCA <0.01 DHCA C18: Experiment 2 Control DHCA DHCA C DHCA C18: Oleyl alcohol Caffeic acid Sample codes refer to Experimental design Table 1. 3 Results 3.1 Characteristics of the fish oil enriched milk emulsions The characteristics, i.e. droplet size and content of EPA and DHA of the different fish oil enriched milk emulsions are summarized in Table 2. Average sizes of the lipid droplet in the milk were between 0.45 and 0.74 mm during the storage period. In Experiment 1, the emulsion with antioxidants had a significantly larger droplet size, whereas a similar effect was not observed in the second experiment. The droplet sizes measured in the different milk emulsion were unchanged during storage. Content of EPA and DHA in the different milk emulsions at day 0 indicated similar levels of and % EPA and DHA of total fatty acids, respectively. At Day 12, oleyl dihydrocaffeate (Experiment 2) was the only milk emulsion with a decrease in EPA and DHA >0.1%. However, the EPA and DHA contents increased to the same extent during storage in several emulsions. Thus, the decreases in EPA and DHA in oleyl dihydrocafeate (Experiment 2) may be due to day to day variation for the measurements rather than an actual decrease. This finding suggests that lipid oxidation did not significantly affect EPA and DHA contents in any of the milk emulsions. 3.2 Peroxide values in fish oil enriched milk emulsions Peroxide values obtained in Experiment 1 and 2 during storage are shown in Fig. 1. For Experiment 1 a lag phase PV [meq peroxides / kg oil] PV [meq peroxides / kg oil] (A) (B) Storage time [Days] Control Rutin palmitate Rutin Oleyl dihydrocaffeate Dihydrocaffeic acid Rutin laurate Oleyl alcohol Control Dihydrocaffeic acid Caffeic acid Oleyl dihydrocaffeate Octyl dihydrocaffeate Figure 1. Concentration of peroxides measured as PV [meq. peroxides/kg oil] in the different fish oil enriched emulsions during storage. Error bars indicate SD of the measurements (n ¼ 2). (A) Experiment 1: control (&), rutin (D), rutin laurate (~), rutin palmitate (!), dihydrocaffeic acid (*) and oleyl dihydrocaffeate (*). (B) Experiment 2: control (&), dihydrocaffeic acid (*), octyl dihydrocaffeate (^), oleyl dihydrocaffeate (*), oleyl alcohol (X) and caffeic acid (5).

6 Eur. J. Lipid Sci. Technol. 2012, 114, Antioxidant efficacy of lipophilized phenolics 439 was observed until day 3 for the emulsion without antioxidant added (Fig. 1A) as the difference between day 0 and 3 was insignificant. In contrast, the lag phase for emulsion with antioxidant added was longer than 3 days and for milk emulsion with rutin laurate the lag phase could not be determined within the storage period (PV never increased). This indicated that rutin laurate efficiently inhibited the development of peroxides. However, the data could also indicate that peroxides were developed and decomposed at equal rates. Investigation of secondary oxidation products (volatiles), described later, resolves this. After the end of the lag phase, the concentration of peroxides increased with different rates in the different emulsions. At day 12 the ranking of the different milk emulsions with respect to PV level was as follows: control a (highest level) rutin palmitate ab rutin b ¼ oleyl dihydrocaffeate b ¼ dihydrocaffeic acid b >rutin laurate c. Compared to Experiment 1, the off set for peroxide development was faster in Experiment 2 (Fig. 1B). Moreover, the PV level in the milk emulsion without antioxidant and the one with dihydrocaffeic acid added was higher than in the same milk emulsion in Experiment 1. Only two emulsions in Experiment 2 had a lag phase, and these were the milk emulsions with oleyl dihydrocaffeate and octyl dihydrocaffeate added. For all other milk emulsions the development of PVs was triggered in different rates already at the beginning of storage depending on the antioxidant added. At day 12 the ranking of milk emulsions according to PV level was as follows: oleyl alcohol a >control b >dihydrocaffeic acid c ¼ caffeic acid c >oleyl dihydrocaffeate d ¼ octyl dihydrocaffeate d. Thus, in contrast to the findings for rutin esters in Experiment 1, the chain length of the fatty acid esterified to dihydrocaffeic acid did not influence the development of peroxides in the emulsions differently. The inhibition percentage calculated for Experiment 1 based on PV development (Table 3), clearly showed that rutin laurate was the most efficient antioxidant in inhibiting the formation of peroxides when added to milk emulsion. However, oleyl dihydrocaffeate and rutin palmitate was as efficient as rutin laurate at Day 6, but their antioxidative efficiency decreased after Day 6. For Experiment 2 it is clear from the inhibition percentage that both dihydrocaffeate esters added were very effective in inhibiting the peroxides formation. Inhibition percentages for those milk emulsions with the same antioxidants applied in both experiments; dihydrocaffeic acid and oleyl dihydrocaffeate, indicated better inhibition of PV development in Experiment 1 with acetone added for dihydrocaffeic acid emulsion after Day 3 and in Experiment 2 without acetone for oleyl dihydrocaffeate emulsion. Thus, the addition of acetone seemed to have no significant impact on the development of peroxides. In spite of the higher PVs in Experiment 2, the efficacy of antioxidants in fish oil enriched milk seemed to be better for octyl and oleyl dihydrocaffeate without acetone than oleyl dihydrocaffeate, rutin laurate and rutin palmitate with acetone added at least before Day 9. However, before a clear conclusion can be drawn the data on volatile oxidation products must be considered. 3.3 Development of volatile oxidation products Five different volatiles were measured in the stored milk emulsions and three of them is shown in Fig. 2: 1-penten- 3-one, 1-penten-3-ol and 2,6-nonadienal. These volatiles are shown since they illustrate the general trend in the development of volatiles during storage in the two experiments. Furthermore, they represent decomposition of n-3 fatty acids and are known to have impact on the development of fishy off-flavour [27]. For Experiment 1, the volatiles are shown in Fig. 2A, C and E, respectively. Concentration of 1-penten-3-one in milk Table 3. Calculated inhibition percentages of PV level in the different milk emulsions in both storage experiments Inhibition of PV during storage (%) Milk emulsions Day 0 Day 3 Day 6 Day 9 Day 12 Experiment 1 Rutin Rutin laurate (C12) Rutin palmitate (C16) Dihydrocaffeic acid Oleyl dihydrocaffeate (C18:1) Experiment 2 Dihydrocaffeic acid Octyl dihydrocaffeate (C8) Oleyl dihydrocaffeate (C18:1) Oleyl alcohol Caffeic acid The values are calculated according to the control emulsions from the respective experiment.

7 440 A.-D. M. Sørensen et al. Eur. J. Lipid Sci. Technol. 2012, 114, Figure 2. Concentration of three volatiles [ng/g] in the different fish oil enriched milk emulsions during storage. (A,B) 1-penten-3-one Experiment 1 and 2, (C,D) 1-penten-3-ol Experiment 1 and 2, (E,F) 2,6-nonadienal Experiment 1 and 2, respectively. Error bars indicate SD of the measurements (n ¼ 3). Symbols: control (&), rutin (D), rutin laurate (~), rutin palmitate (!), dihydrocaffeic acid (DHCA *), oleyl dihydrocaffeate (DHCA C18:1 *), octyl dihydrocaffeate (DHCA C8 ^), oleyl alcohol (X) and caffeic acid (5). emulsions increased until Day 9, whereafter the concentration decreased at different rates in the different emulsions. Milk emulsions with no antioxidant and dihydrocaffeic acid added had no lag phase, whereas milk emulsions with rutin, rutin palmitate and oleyl dihydrocaffeate added had a lag phase of 3 days before these emulsions started developing 1- penten-3-one. The concentration of 1-penten-3-one in milk emulsion with rutin laurate added did not increase during storage. The ranking of the emulsions according to concentration of 1-penten-3-one before the decrease in concentration (Day 9) was as follows: rutin laurate a <oleyl dihydrocaffeate b <rutin palmitate c dihydrocaffeic acid c <rutin d <control e. The observed decline in the concentration of 1-penten-3-one at the end of the storage period might be due to a reduction of this volatile to 1-penten-3-ol either by the antioxidant or other components in the milk emulsion. For the other volatiles measured in this experiment, the concentration increased after a shorter or longer lag phase (Fig. 2C and E). Similar for these volatiles were that the control milk emulsion had the shortest lag phase and the highest concentration of the volatiles measured at the end of the storage. The lag phase for the development of 1-penten-3-ol in milk emulsion was 3 days for control emulsion and emulsions with dihydrocaffeic acid, and 6 days for oleyl dihydrocaffeate, rutin and rutin palmitate added, whereas milk emulsion with rutin laurate had a lag phase of 9 days (Fig. 2C). The duration of the lag phase for development of 2,6-nonadienal was slightly different from that of 1-penten-3-ol. Here, the milk emulsion with rutin only had 3 days lag phase together with dihydrocaffeic acid and control emulsions, and milk emulsion with oleyl dihydrocaffeate added had an infinite lag phase (2,6-nonadienal never increased, Fig. 2E). After the end of the lag phase the concentration of volatiles increased in the different milk emulsions, except for 2,6-nonadienal in the

8 Eur. J. Lipid Sci. Technol. 2012, 114, Antioxidant efficacy of lipophilized phenolics 441 emulsion with oleyl dihydrocaffeate added. At day 12 the ranking according to concentration of 1-penten-3-ol and 2,6 nonadienal were as follows: rutin laurate a <oleyl dihydrocaffeate and dihydrocaffeic acid b <rutin and rutin palmitate c <control d and oleyl dihydrocaffeate a <rutin laurate and dihydrocaffeic acid b <rutin palmitate and rutin c <control d, respectively. Overall, the concentrations of volatiles were lower in emulsions containing the lipophilized compounds compared with the original phenolic. However, rutin palmitate sometimes led to higher concentration of volatiles than rutin in the milk emulsions. The development of volatiles in emulsions from Experiment 2 is shown in Fig. 2B, D and F. As observed in Experiment 1, the concentration of 1-penten-3-one first increased and then decreased. However, in Experiment 2 this only happened in three emulsions: oleyl alcohol, dihydrocaffeic acid and control emulsion. Moreover, these emulsions had a faster increase in the concentration of 1-penten-3-one than the other emulsions. For milk emulsions with octyl and oleyl dihydrocaffeate added, the development of 1-penten-3- one did not increase the first 3 days. The concentration of 1- penten-3-one in the emulsions before the decrease (Day 6) was ranked as follows: octyl dihydrocaffeate a <oleyl dihydrocaffeate a,b <caffeic acid b <dihydrocaffeic acid c <control d <oleyl alcohol e (Fig. 2B). The same milk emulsions, as observed for development of 1-penten-3-one, also had a faster increase for the other measured volatiles, except for 2,6-nonadienal. In these emulsions (control, dihydrocaffeic acid and oleyl alcohol) no lag phase was observed, whereas the other 3 emulsions with caffeic acid, oleyl dihydrocaffeate and octyl dihydrocaffeate added, had a lag phase. For the development of 1-penten-3-ol, the lag phase was 3 days for milk emulsion with caffeic acid and oleyl dihydrocaffeate added, whereas it was 9 days with octyl dihydrocaffeate added. At the end of the storage period the concentration of 1-penten-3-ol in the different emulsions was ranked as follows: octyl dihydrocaffeate a <oleyl dihydrocaffeate b <caffeic acid c <dihydrocaffeic acid d <control e <oleyl alcohol d. Similar to Experiment 1, the development of 2,6-nonadienal had a longer lag phase than observed for the other volatiles measured. Emulsions with oleyl alcohol and dihydrocaffeic acid had together with the control emulsion, the shortest lag phases of 3 days followed by the emulsion with caffeic acid (6 days). The longest lag phase was observed in emulsions with octyl and oleyl dihydrocaffeate, which continued during the entire storage period (Fig. 2F). According to concentration of 2,6-nonadienal at Day 12, the order of emulsions was as follows: oleyl and octyl dihydrocaffeate a <caffeic and dihydrocaffeic acids b <control c <oleyl alcohol d. In summary, the difference for the different volatiles was the length of the lag phase, whereas the ranking of the emulsions according to volatile concentrations was more or less similar for the different volatiles. Similar to the PV, the inhibition percentages were calculated, however only at the end of the storage and for the concentration of 1-penten-3-one when it was highest at day 9 and 6 for Experiment 1 and 2, respectively. The inhibition percentage from Experiment 1 (Table 4) clearly showed that rutin laurate followed by oleyl dihydrocaffeate were the most efficient antioxidants in inhibiting the formation of volatiles compared to other antioxidants. Depending on the volatile, it was different whether it was rutin or rutin palmitate that was more efficient. For 1-penten-3-ol and 2,4-heptadienal rutin was more efficient compared to rutin palmitate, whereas for 1-penten-3-one, 2-hexenal and 2,6-nonadienal it was Table 4. Calculated inhibition percentages of volatile concentration in the different milk emulsions in both storage experiments Inhibition of volatiles at day 12 (%) Milk emulsions 1-Penten-3-one a) 1-Penten-3-ol 2-Hexenal 2,4-Heptadienal 2,6-Nonadienal Experiment 1 Rutin Rutin laurate Rutin palmitate Dihydrocaffeic acid Oleyl dihydrocaffeate b) Experiment 2 Caffeic acid Dihydrocaffeic acid Oleyl dihydrocaffeate b) Octyl dihydrocaffeate b) Oleyl alcohol The values were calculated according to the control emulsions from the respective experiment. a) For development of 1-penten-3-one the inhibition percentages were calculated at the day with highest concentration before decreasing, that means that it was calculated at day 9 for experiment 1 and day 6 for experiment 2. b) This indicate that the specific volatile was not detected in this emulsion and therefore the inhibition percentage was not calculated.

9 442 A.-D. M. Sørensen et al. Eur. J. Lipid Sci. Technol. 2012, 114, opposite. For Experiment 2 it is clear from the inhibition percentages that both dihydrocaffeate esters were very effective in inhibiting the formation of volatiles. However, octyl dihydrocaffeate was more efficient regarding the inhibition of 1-penten-3-one, 1-penten-3-ol and 2-hexenal, whereas oleyl dihydrocaffeate was more efficient in inhibiting the formation of 2,4-heptadienal. When comparing the efficiency of the phenolics, caffeic acid was more efficient as antioxidant than dihydrocaffeic acid. The impurity, oleyl alcohol, clearly showed prooxidative behaviour in the milk emulsion. Inhibition percentages in milk emulsions with the same antioxidants applied in both experiments; dihydrocaffeic acid, oleyl dihydrocaffeate, indicated better inhibition of volatile development in Experiment 1 with acetone added for dihydrocaffeic acid emulsion, and in general no difference in the inhibition percentages by oleyl dihydrocaffeate in the two experiments. Hence, the addition of acetone seemed to have no significant impact on the development of volatiles, which was also concluded from the formation of peroxides. The inhibition percentages indicate that three of the lipophilized antioxidants were better than the phenolic they originated from Tocopherol in the fish oil enriched emulsions Three of the tocopherol homologues were detected (a-, g- and d-tocopherol), but only the concentration of a-tocopherol changed significantly during the storage period (Fig. 3). This might be due to the higher amount of this homologue compared with the others. The concentration of a-tocopherol was reduced most in the milk emulsion without antioxidant added (control emulsion) and least in emulsion with rutin laurate followed by oleyl dihydrocaffeate in Experiment 1 and caffeic acid and oleyl dihydrocaffeate followed by octyl dihydrocaffeate in Experiment 2. Hence, tocopherol was better preserved in these emulsions maybe due to less lipid oxidation Sensory evaluation As volatiles cause off-flavour, the preliminary sensory evaluation of emulsions was expected to reflect the result of volatile development in the emulsions. Only milk emulsions in Experiment 2 was evaluated by their odour by a sensory expert panel at storage day 9 and 12 (data not shown). Experiment 1 was not possible to evaluate due to addition of the acetone, which resulted in a strong solvent odour. Clearly, the most oxidized emulsion according to the sensory evaluation was the emulsion with oleyl alcohol due to a clear off-odour of oxidized fish oil at Day 9 which developed to lacquer/painty at day 12. Moreover, the control emulsion had a clear fishy off-odour as well as metallic odour at day 9. The most oxidatively stable emulsions, as determined by odour evaluation, were emulsions with oleyl and octyl dihydrocaffeate. These two emulsions were evaluated to have ocean and Figure 3. Concentration of a-tocopherol in the different fish oil enriched emulsions at day 0 and 12 (A) Experiment 1 and (B) Experiment 2. Error bars indicate SD of the measurements (n ¼ 4). Different letters within same storage day indicate significant differences in concentration between emulsions. Abbreviation: Rutin C12, rutin laurate; Rutin C16, rutin palmitate; DHCA, dihydrocaffeic acid; DHCA C18:1, oleyl dihydrocaffeate; DHCA C8, octyl dihydrocaffeate. chemical/flower odour, respectively. The expert panel could, however, not discriminate between these emulsions with respect to their level of fishy off-odour. 4 Discussion 4.1 Physical stability and lipid oxidation All the milk emulsions were physically stable and no changes in the droplet size during storage were observed. In Experiment 1 there was a significant difference in size of the lipid droplets in the control emulsion (0.45 mm) compared to the other emulsions ( mm). This could be due to either less lipid incorporated in the emulsion or an unintended slightly different homogenization than for the other emulsions in this experiment. However, it is clear from Table 2 that the emulsion had the same amount of fish oil as the other emulsions. Thus, the smaller droplets may indicate a slightly longer homogenization of this emulsion. Lipid oxidation is initiated at the interface of the droplet, therefore the size of total interfacial area is hypothesized to influence the lipid oxidation. Thus, increased droplet size may result in decreased lipid oxidation [9]. However, the literature in this

10 Eur. J. Lipid Sci. Technol. 2012, 114, Antioxidant efficacy of lipophilized phenolics 443 area is conflicting, and some studies support the hypothesis [4, 28], whereas in milk emulsions the opposite has been observed [29, 30]. Therefore, the increased oxidation in the control emulsion was most likely due to no protection of the lipids by antioxidants and not due to the different droplet size. Furthermore, the control emulsion in Experiment 2 had the same droplet size as the other emulsions, but was still the most oxidized emulsion together with emulsion with oleyl alcohol. Hence, the results indicated limited effect of interface area on lipid oxidation in complex emulsion systems. 4.2 Oxidative stability of the milk emulsions and the influence of alkyl chain length The results indicated that milk emulsions without antioxidant or with oleyl alcohol added generally oxidized faster than emulsions with either phenolics or phenolipids added. Interestingly, rutin laurate was a more efficient antioxidant compared to rutin palmitate and rutin it-self, which indicated a cut-off effect related to the alkyl chain length. However, for lipophilized dihydrocaffeic acid it was less clear whether such cut-off effect existed, as differences in oxidation parameters not always was significantly different for octyl and oleyl dihydrocaffeate emulsions. In emulsion with rutin laurate, PV did not increase during storage, which could indicate efficient inhibition by rutin laurate or fast decomposition of peroxides to volatiles. The findings that milk emulsion with rutin laurate added had low concentrations of volatiles showed that, rutin laurate effectively inhibited formation of both primary and secondary oxidation products. Interestingly, the findings regarding the efficacy of rutin esters as antioxidants were different in milk emulsions compared with earlier findings in simple o/w emulsion [13]. Thus, in this study on fish oil enriched milk, both rutin esters had an antioxidative effect, but rutin laurate was a more efficient antioxidant than rutin palmitate and rutin it-self. In contrast, rutin laurate and rutin palmitate were less effective antioxidants when compared with rutin in simple o/w emulsion [13]. However, rutin laurate also exerted stronger antioxidative activity than rutin and rutin palmitate in a LDL assay, which is a more complex system than an o/w emulsion [13]. Thus, these findings indicate that the cut-off effect is influenced by the system, i.e. simple o/w emulsion or more complex emulsion systems such as LDL and milk. In addition, the fish oil enriched milk emulsion contains proteins and other minor components, which are not present in a simple o/w emulsion. These components might have interacted and influenced the efficacy or location of the phenolipids. Shahidi and Zhong [15] suggested that not only the partitioning of the antioxidant influence the efficacy of the antioxidant, but also the emulsifier in emulsified medium. This is due to saturation of the interfacial area by emulsifier, which leaves less interfacial area available for antioxidant location. Hence, emulsifiers may compete with antioxidants for localization at the interface, where oxidation is initiated. Moreover, recent experiments with phenolic compounds and two different emulsifiers in simple o/w emulsion have shown interaction of the antioxidants with emulsifier and iron [18]. The phenolic compounds investigated were caffeic acid, coumaric acid, naringenin and rutin, which were evaluated for interactions with Citrem and Tween with or without iron present. Interactions in this study may have influenced the efficacy and location of the antioxidant. Both octyl dihydrocaffeate and oleyl dihydrocaffeate exerted stronger antioxidative effects than dihydrocaffeic acid in fish oil enriched milk. The efficacy of octyl dihydrocaffeate was slightly better than oleyl dihydrocaffeate in inhibiting the formation of some volatiles however for other volatiles it was not significantly better. These findings are different from the findings obtained with these phenolipids in a simple o/w emulsion [17]. In the o/w emulsion system the difference between octyl and oleyl dihydrocaffeate as antioxidants was clearer than in the milk emulsions, where octyl dihydrocaffeate was a significantly better antioxidant than oleyl dihydrocaffeate. Importantly, in both studies oleyl dihydrocaffeate contained impurities such as oleyl alcohol. When added in the milk emulsion oleyl alcohol resulted in a prooxidative effect in contrast to its addition in the simple o/w emulsion. Since oleyl alcohol acted as a prooxidant in fish oil enriched milk it might have reduced the antioxidative effect of oleyl dihydrocaffeate as antioxidant. Thus, the antioxidative effect may have been even better for oleyl dihydrocaffeate than octyl dihydrocaffeate if a more purified compound was used, but this needs to be further investigated. The slight difference in results obtained from the o/w emulsion and milk emulsion may, similar to effects of rutin esters, be explained by the different systems tested. For example the effect of emulsifier and its position on the interface may have led to interactions with the antioxidant [15]. In vitro antioxidant assays preformed in our previous studies have generally showed better antioxidant activity for the original phenolic than for the lipophilized phenolic compound [12, 17]. However, in emulsions the dihydracaffeates were better antioxidants than dihydrocaffeic acid in both simple o/w emulsions and milk emulsions, and lipophilized rutin was better than rutin in milk emulsions. Thus, the results from in vitro antioxidant assays cannot solely predict the antioxidant properties in emulsion systems, and other factors such as interactions and partitioning may also influence antioxidant activity. The partitioning of lipophilized dihydrocaffeic acid and rutin in the different phases has previously been evaluated [13, 17]. Only rutin laurate could be measured in the water phase of a simple o/w emulsion and only in small amounts (3.8%). The finding that lipophilized rutin did not have any antioxidative effect in o/w emulsions compared to rutin, whereas lipophilization with C12 improved its efficacy in milk indicates that interactions with, e.g. proteins may influence the location of the lipophilized rutin in milk emulsions. Thus, rutin laurate may be located more favourably in the milk emulsion to act as antioxidant,

11 444 A.-D. M. Sørensen et al. Eur. J. Lipid Sci. Technol. 2012, 114, i.e. close to the interface instead of as a micelle in the water phase or in the core of the oil droplets. A similar explanation could be given for the differences observed with oleyl dihydrocaffeate. However, further research is needed in order to be able to conclude on possible interaction in the milk and changed partitioning of the phenolipids as a consequence of the interaction. 4.3 Fish oil enriched milk and human health Overall it was possible to stabilize fish oil enriched milk by adding phenolipids. The current amount of fish oil in this milk will only be a supplement of n-3, since 500 ml milk daily will result in an intake of 50 mg EPA and DHA. EU recommendation is currently at an intake of 250 mg daily due to proved health beneficial effects, thus 500 ml milk daily will cover 20% of the recommendeted. Consumption of n-3 enriched milk solely may not increase the human health, but together with other fish or fish oil enriched products. 5 Conclusions In conclusion, both phenols and phenolipids acted as antioxidants in milk emulsions enriched with fish oil. However, the phenolipids were more efficient antioxidants especially rutin laurate, octyl and oleyl dihydrocaffeate. Despite the fact that only two chain lengths were evaluated, the results tend to follow the cut-off effect in relation to alkyl chain length and antioxidative effect for rutin esters. The optimal alkyl chain length for a rutin ester in fish oil enriched milk is at least below a chain length of 16 carbon atoms. For dihydrocaffeate esters it was not possible to conclude on a specific cut-off effect in relation to alkyl chain length and antioxidative effect. To be able to conclude on the optimal lipid chain length attached to the phenolic compounds in relation to their strongest antioxidant protection, more studies on the effect of rutin and dihydrocaffeic acid with different lipid chain lengths are needed. The antioxidative effect of phenolipids seems quite complex and it would be of great value to be able to understand the effects of both the lipid chain length and the type of emulsion system on the antioxidative effect of the lipophilized compounds. Taken together, these results clearly show that the polar paradox hypothesis is too simple and must be reconsidered. Moreover, the almost untouched area of phenolipids as antioxidants in real food systems deserves more attention as the results have indicated very promising effects of these compounds that could be utilized by the industry in the future. We thank Maritex Norway (subsidiary of TINE BA, Norway) for providing the fish oil to our research. The study was financed by the Danish Council for Strategic Research (Programme committee for food, nutrition and health) and the Directorate for food, Fisheries and Agri Business. The authors have declared no conflict of interest. References [1] Jacobsen, C., Let, M. B., Nielsen, N. S., Meyer, A. S., Antioxidant strategies for preventing oxidative flavour deterioration of foods enriched with n-3 polyunsaturated lipids: A comparative evaluation. Trends Food Sci. Technol. 2008, 19, [2] Psota, T. L., Gebauer, S. K., Kris-Etherton, P., Dietary omega-3 fatty acid intake and cardiovascular risk. Am. J. Cardiol. 2006, 98, 3I 18I. [3] Riediger, N. D., Othman, R. A., Suh, M., Moghadasian, M. H., A systemic review of the roles of n-3 fatty acids in health and disease. J. Am. Diet. Assoc. 2009, 109, [4] Jacobsen, C., Hartvigsen, K., Lund, P., Thomsen, M. K. et al., Oxidation in fish oil-enriched mayonnaise. Eur. Food Res. Technol. 2000, 211, [5] Jacobsen, C., Timm, M., Meyer, A. S., Oxidation in fish oil enriched mayonnaise: Ascorbic acid and low ph increase oxidative deterioration. J. Agric. Food Chem. 2001, 49, [6] Let, M. B., Jacobsen, C., Frankel, E. N., Meyer, A. S., Oxidative flavour deterioration of fish oil enriched milk. Eur. J. Lipid Sci. Technol. 2003, 105, [7] Let, M. B., Jacobsen, C., Meyer, A. S., Lipid oxidation in milk, yoghurt, and salad dressing enriched with neat fish oil or pre-ernulsified fish oil. J. Agric. Food Chem. 2007, 55, [8] Coupland, J. N., McClements, D. J., Lipid oxidation in food emulsions. Trends Food Sci. Technol. 1996, 7, [9] McClements, D. J., Decker, E. A., Lipid oxidation in oil-inwater emulsions: Impact of molecular environment on chemical reactions in heterogeneous food systems. J. Food Sci. 2000, 65, [10] Porter, W. L., Paradoxical behavior of antioxidants in food and biological systems. Toxicol. Ind. Health 1993, 9, [11] Laguerre, M., Giraldo, L. J. L., Lecomte, J., Figueroa- Espinoza, M. C. et al., Chain length affects antioxidant properties of chlorogenate esters in emulsion: The cutoff theory behind the polar paradox. J. Agric. Food Chem. 2009, 57, [12] Lue, B.-M., Nielsen, N. S., Jacobsen, C., Hellgren, L. et al., Antioxidant properties of modified rutin esters by DPPH, reducing power, iron chelation and low density lipoprotein assays. Food Chem. 2010, 123, [13] Lue, B.-M., Enzymatic Lipophilization of Bioactive Compounds in Ionic Liquids; PhD Thesis, Department of Molecular Biology, Aarhus University, Denmark [14] Sørensen, A.-D. M., Nielsen, N. S., Decker, E. A., Let, M. B. et al., The efficacy of compounds with different polarities as antioxidant in emulsions with fish oil. J. Am. Oil Chem. Soc. 2011, 88, [15] Shahidi, F., Zhong, Y., Revisiting the polar paradox theory: A critical overview. J. Agric. Food Chem. 2011, 59, [16] Laguerre, M., Giraldo, L. J. L., Lecomte, J., Figueroa- Espinoza, M.-C. et al., Relationship between hydrophobocity and antioxidant ability of phenolipids in emulsion: A parabolic effect of the chain length of rosmarinate esters. J. Agric. Food Chem. 2010, 58,