Impact of Hydrophobicity on Antioxidant Efficacy in Low-Moisture Food

Jun 3, 2015 - Crackers were stored in the dark at 55 °C. Standard error bars are smaller than ... However, fewer of these types of overlap were obser...
0 downloads 0 Views 2MB Size
Article pubs.acs.org/JAFC

Impact of Hydrophobicity on Antioxidant Efficacy in Low-Moisture Food Leann Barden,† Nathalie Barouh,§ Pierre Villeneuve,§ and Eric Decker*,†,# †

Department of Food Science, University of Massachusetts, Amherst, 102 Holdsworth Way, Amherst, Massachusetts 01003, United States § UMR IATE, CIRAD, 2 Place Viala, 34060 Montpellier Cedex 02, France # Bioactive Natural Products Research Group, Department of Biochemistry, Faculty of Science, King Abdulaziz University, P.O. Box 80203, Jeddah 21589, Saudi Arabia ABSTRACT: The polarity and partitioning of antioxidants (AOX) in lipid dispersions and bulk oils have a large impact on efficacy, but this has not yet been studied in low-moisture foods. Using a homologous series of rosmarinic esters as AOX in crackers, we determined that efficacy increases with increasing hydrophobicity based on lipid hydroperoxide and hexanal generation. Confocal microscopy was used to determine the location of both lipids and AOX. Hydrophobic rosmarinic esters partitioned more closely with the lipid than rosmarinic acid, presumably placing the hydrophobic AOX at the site of oxidation reactions. Partitioning and efficacy of the intermediate polarity ester were affected by mode of incorporation (e.g., added to the water or to the lipid prior to dough formation). The synthetic AOXs propyl gallate, butylhydroxytoluene, and tertbutylhydroquinone gave similar results with the more hydrophobic BHT and TBHQ being more effective at reducing lipid hydroperoxide and hexanal generation than the more hydrophilic propyl gallate. These results provide important information on which AOX would be most effective in low-moisture foods. KEYWORDS: antioxidant, phenolics, phenolipid, rosmarinic acid, lipid oxidation, polar paradox, cutoff effect, low-moisture food, crackers



INTRODUCTION Manufacturers typically utilize antioxidants (AOX) to extend the lag period of lipid oxidation and thus increase shelf life. Unfortunately, AOX selection is often on a trial-and-error basis because antioxidant and prooxidant factors in different foods are often poorly understood. The uncertainty and variability of AOX performance make systematic studies of their behavior in both model systems and real foods critical for predicting AOX efficacy. Important factors to study include AOX type and concentration, ability of the AOX to inhibit the major active prooxidants in the product, product physical properties, and manufacturing processes (e.g., thermal history). Porter et al.1 observed that AOX effectiveness was governed by (1) the hydrophilic−lipophilic balance of the AOX and (2) the surface-to-volume ratio of the lipid (e.g., bulk oils vs emulsions). This led to the development of the antioxidant polar paradox hypothesis, which generally states that nonpolar AOX are most effective in oil-in-water emulsions and membranes, whereas polar AOX are most effective in bulk oils.2 Frankel et al.3 extended this hypothesis with research suggesting that AOX with a tendency to concentrate at interfacial regions in oil-in-water emulsions, that is, the site of lipid oxidation, were most effective. However, the original antioxidant polar paradox in oil-inwater emulsions was recently challenged by researchers using homologous series of chlorogenate4,5 and rosmarinate6−9 esters. These studies demonstrated that AOX efficacy and hydrophobicity share a parabolic relationship in which AOX with intermediate polarity have the optimum activity, and very hydrophilic (18 © XXXX American Chemical Society

carbon esters) can exhibit a dramatic loss in AOX activity. The loss of AOX activity for the very hydrophilic AOX is thought to be due to their partitioning into the aqueous phase of the emulsion, where they cannot interact with lipids. The loss of AOX activity for the very hydrophobic AOX has been postulated to be due to their lack of surface activity in oil-inwater emulsions.8,10 Systematic research on lipid oxidation mechanisms in lowmoisture foods is, in general, lacking and somewhat dated. The aforementioned advances in untangling the antioxidant polar paradox pertain only to emulsions and bulk oil systems. To our knowledge, few to no systematic AOX studies have been published in low-moisture systems such as ready-to-eat cereal, extruded pet foods, and crunchy snack foods (crackers, cookies, granola bars, etc.). Furthermore, AOX are typically incorporated by addition to the lipophilic phase or to the final product prior to cooking. Low-moisture foods are typically made from doughs, which begin with a hydrophobic component (crumbing the lipid) and then are mixed with the hydrophilic plasticizer. Therefore, we investigated the efficacy of the antioxidants when they were added to the lipid prior to dough formation or when added to the final dough. This study used a homologous series of rosmarinate esters of varying hydrophobicity to test AOX activity in low-moisture foods. The homologous series of AOX were added separately to the lipid Received: February 27, 2015 Revised: June 1, 2015 Accepted: June 3, 2015

A

DOI: 10.1021/acs.jafc.5b01085 J. Agric. Food Chem. XXXX, XXX, XXX−XXX

Article

Journal of Agricultural and Food Chemistry

vortexed briefly, and centrifuged (3400g) for 10 min to extract the hydroperoxide-containing lipid fraction. This extract (200 μL) was mixed with (1) 16.7 μL of a 50:50 solution of ferrous sulfate (66 mM in double-distilled water) and barium chloride (38 mM in 0.4 N HCl) and (2) 16.7 μL of ammonium thiocyanate (3.9 M in double-distilled water). The final mixture was vortexed, covered to prevent evaporation, and incubated 20 min at ambient temperature as color developed. Absorbance was measured at 500 nm on a Genesys 20 spectrophotometer (ThermoSpectronic, Waltham, MA, USA). The concentration of hydroperoxides was calculated from a cumene hydroperoxide standard curve. Headspace hexanal, the secondary oxidation product, was measured in triplicate with a GC-2014 Shimadzu gas chromatograph equipped with an AOC-5000 autosampler (Shimadzu, Kyoto, Japan) based on the solid phase microextraction (SPME) method of Panya et al.8 Specifically, a 50/30 μm divinylbenzene/carboxen/polydimethylsiloxane (DVB/Carboxen/PDMS) SPME fiber (Supelco, Bellefonte, PA, USA) was inserted through the vial septum and exposed to the sample headspace for 10 min at 55 °C. The SPME fiber was desorbed at 250 °C for 3 min in the gas chromatograph detector at a split ratio of 1:7. The temperatures of the oven, injector, and flame ionization detector were 65, 250, and 250 °C, respectively, and samples ran for 6 min. Volatile aldehydes were separated on a fused-silica capillary column (30 m × 0.32 mm i.d. × 1 μm) coated with 100% poly(dimethylsiloxane) (Equity-1, Supelco), and peak integration was calculated using Shimadzu EZstart (version 7.4) software. A standard curve made from crushed cracker spiked with known hexanal concentrations (hexanal was diluted in methanol with a range from 0 to 136 mmol hexanal per kilogram of cracker) was used to determine hexanal concentrations in the samples. Measurement of Rosemary AOX. To prepare samples for highperformance liquid chromatography (HPLC), 0.150 ± 0.005 g of pulverized crushed cracker was mixed with 1 mL of methanol, sonicated for 30 min, and filtered (0.45 μm syringe filter Millex-FH, Millipore Corp., Bedford, MA, USA) into amber vials with polyethylene snap caps (Waters, Milford, MA, USA). HPLC determination of rosmarinic acid and its esters was carried out with a Hypersil gold C18 reversed phase column (250 mm × 4.6 mm, 5 μm) equipped with a Hypersil gold guard column (10 mm × 4 mm, 5 μm) (Thermo Scientific, USA) using an LC-10ATvp HPLC system (Shimadzu, USA). Samples were injected into the HPLC at a flow rate of 1 mL/min at room temperature using a mobile phase of either methanol (to detect rosemary antioxidants) or acetonitrile (to detect caffeic acid). Pure rosmarinic acid and its alkyl esters and caffeic acid were dissolved in methanol and used as standards. R0 and its esters were detected with a photodiode array detector (SPD-M10Avp, Shimadzu, USA) at 328 nm. Peak integration was performed using Shimadzu EZstart (version 7.4). Confocal Microscopy. A Nikon confocal microscope (C1 Digital Eclipse, Tokyo, Japan) with a PL FLUOTAR ELWD 20.0 × 0.45 objective lens was used to capture the confocal images. Whole crackers were glued to a slide and imaged by quadrants, with at least four images taken in each quadrant. A 408 nm laser was used to excite the R0 and esters, and a 488 nm laser was used to excite Nile Red. Emission spectra were collected from 415 to 485 nm for the R0 and esters and from 485 to 545 nm for the Nile Red, and the resulting images were overlaid. Detector pinhole size was 150 μm. All resulting images consisted of 512 × 512 pixels, with a pixel size of 414 nm, and a pixel dwell time of 10.40 μs. Images were analyzed using EZ-CS1 (version 3.8) software (Nikon, Melville, NY, USA). Statistical Analysis. Differences between treatments were determined by differences in the laboratory phase of lipid hydroperoxide and hexanal formation. The cutoff for the end of the oxidation lag phase was marked by the first datum point that was significantly greater than the time-zero value with all subsequent data points also being significantly greater than the time-zero value. Means were compared using analysis of variance (General Linear Model with Tukey’s Honestly Significant Difference test as a post hoc test for multiple comparisons). Statistical difference was set at p < 0.05. All

and aqueous phases to see if the mode of addition affected AOX efficacy in a low-moisture system. The AOX efficacy was related to cracker structure and AOX partitioning via imaging with confocal microscopy. Commercial AOX of varying hydrophobicities were also used to evaluate the relationship between hydrophobicity and AOX efficacy. Improving our understanding of AOX mechanisms has broad and long-lasting implications. Development of new, tailored AOX that can extend product quality, thereby reducing industry loss and supporting development of foods with extended shelf life, can have benefits to consumers as well as meals for space and military programs. Improved AOX technologies will also improve the health and wellness of the food supply by allowing greater utilization of heart-healthy polyunsaturated fats and other bioactive lipids susceptible to oxidation.



MATERIALS AND METHODS

Materials. Baking soda (Arm & Hammer), iodized table salt (Morton), and all-purpose flour (Gold Medal) were purchased from local grocery stores. Interesterified soybean oil 762420 (henceforth referred to as ISO) was provided by Archer Daniels Midland Co. (Decatur, IL, USA). Dry goods were stored in closeable freezer bags to prevent moisture sorption; ISO was kept frozen (−20 °C) until use. Dodecyl and eicosyl rosmarinate esters were prepared as described previously (structures in original publication).11 Rosmarinic acid, the fluorescent probe Nile Red, and all other chemicals were purchased from Sigma-Aldrich (St. Louis, MO, USA). All solutions were prepared using double-distilled water. General Sample Manufacture and Preparation. Crackers were prepared as previously described12 by premixing dry ingredients (sifted flour, baking soda, and salt) and then blending with the ISO until the fat was homogeneously cut into the flour; all mixing was done with a KitchenAid model KSM95 (Mississauga, ON, Canada) and associated attachments. Water was then added to form a dough, which was briefly kneaded by hand until all flour was visibly incorporated (∼30 s). Ten grams of additional flour was sprinkled on the surface of a baking mat to prevent sticking; all of this flour was incorporated during kneading. The dough was flattened by twice sheeting it through a pasta roller (KitchenAid KPSA attachment; thickness setting 2). Crackers (2.5 cm × 2.5 cm) were cut from the dough with a pizza cutter and baked on an ungreased cookie sheet at 163 °C for 21 min in an electric oven (General Electric model JB350DFWW; Fairfield, CT, USA). After baking, crackers were coarsely crushed using a mortar and pestle, and 0.5 g was placed into acid-washed, 10 mL glass GC vials (Supelco Analytical, Bellefonte, PA, USA). The vials were closed with aluminum caps containing PTFE/silicone septa (Supelco Analytical) and stored at 55 °C in the dark for 30−75 days depending upon lag phase. Crackers for confocal image analysis were stored in closeable freezer bags in the dark at ∼4 °C and imaged within 3 days of manufacture. AOX and Dye Incorporation into the Crackers. One hundred and fifty-nine micromoles of rosmarinic acid (R0) or its 12- (R12) and 20- (R20) carbon esters (henceforth collectively referred to as “rosemary AOX”) or 344 μmol of synthetic antioxidants [propyl gallate (PG), tert-butylhydroquinone (TBHQ), and butylhydroxy toluene (BHT)] were dissolved in 1.0 mL of ethanol and mixed with ISO in the KitchenAid mixer (model KSM95) for 1 min before the dry ingredients were added. Alternatively, rosmarininc acid esters in 1.0 mL of ethanol were mixed with water before dough formation in some experiments. In all cases, the same volume of ethanol was added to control (no AOX) treatments. Nile Red in ethanol (0.5 mL) was added to the ISO prior to mixing in the rosemary AOX and control treatments. Measurement of Lipid Oxidation. Lipid hydroperoxides, the primary oxidation products, were measured in triplicate using a modified version of the International Dairy Federation method as previously described by Shantha and Decker.13 In short, coarsely crushed crackers (0.105 g) were pulverized using a mortar and pestle and added to a mixture of chloroform and methanol (5 mL; 2:1 v/v), B

DOI: 10.1021/acs.jafc.5b01085 J. Agric. Food Chem. XXXX, XXX, XXX−XXX

Article

Journal of Agricultural and Food Chemistry calculations were performed using XLSTAT statistical software (version 2006.3, Addinsoft, New York, NY, USA).

the lipid. However, these images suggest that instead of forming a homogeneous mixture with the ISO, some of the R20 is selfaggregating apart from the ISO. This could occur if the concentration of the R20 used in these experiments exceeded its solubility in the ISO. The difference in AOX partitioning observed via confocal microscopy was compared to the actual AOX efficacy. The esterified rosemary esters, which showed greater overlap with the lipid in the confocal images, significantly extended the lag phase of both hydroperoxide (Figure 2A) and hexanal (Figure



RESULTS Sample Quality. To ensure consistent production, all crackers were made by one individual, trained for >8 h before experimentation, and all samples were consistent both visually and in final moisture content. To ensure reproducibility, cracker production steps were carefully controlled; two batches were freshly prepared for every treatment, and all treatments presented in a single figure were prepared the same day, including fresh control treatments. Homologous AOX Series. R0 and its alkyl esters were selected as a means of varying hydrophobicity while keeping the AOX portion of the molecule constant.11 The activity of each antioxidant was tested at 159 μmol because it corresponded to 500 ppm of R0, a concentration typically used by industry for AOX applications. R0 and its alkyl esters naturally fluoresce (blue region of Figure 1), allowing their location to be

Figure 1. Fluorescence of Nile Red dye (green) and either (A) rosmarinic acid, (B) dodecyl (R12) rosmarinate ester, or eicosyl (R20) rosmarinate ester (blue) in crackers at day 0. Antioxidants were incorporated by mixing with the lipid phase prior to dough formation. Scale bars: 20 μm in panels A and B; 50 μm in panel C.

determined with confocal microscopy without the use of additional probes. Nile Red was used as a lipid-soluble confocal probe to view the lipids (green region of Figure 1) in the cracker. In all treatments, the ISO formed a continuous lipid phase surrounding the starch granules (data not shown), as was seen in a previous study.12 AOX were first added by incorporating them in the lipid prior to dough formation. Confocal microscopy can provide useful information about the location of different components in foods, but it does have its limitations due to limits in resolutions and variations in focusing depths in foods with rough textures. Therefore, we attempted to keep our discussion of the confocal results as general and conservative as possible and realize that the images can be difficult to interpret. Figure 1A shows crackers containing free R0. In this system, the blue R0 tended to partition separately from the green lipid region. This is likely due to the low lipid solubility of R0, which agrees with findings by Panya et al.8 in which 90% of R12 was in the lipid phase of oil-in-water emulsions. The partitioning behavior of R20 is shown in Figure 1C. It differs from R0 and R12 in that it formed distinct regions of blue scattered throughout the cracker. It is unclear why this would occur because R20 is the most hydrophobic and would be soluble in

Figure 2. (A) Lipid hydroperoxides and (B) headspace hexanal in crackers with rosmarinic and its esters (chain lengths = 0, 12, or 20 carbons) incorporated into the lipid prior to dough formation. Crackers were stored in the dark at 55 °C. Standard error bars are smaller than data points in some instances.

2B) development when the AOX was incorporated into the lipid phase prior to dough formation. R0 was the least effective. It extended the hydroperoxide lag phase only a few days compared to the control, but it actually increased the formation of headspace hexanal compared to the control, decreasing the hexanal headspace lag phase from 40 days (control) to 21 days (R0) (Figure 2B). This suggests the polar R0 destabilized hydroperoxides, perhaps by binding and solubilizing prooxidant metals or by reducing metals to their more reactive state, causing increased lipid hydroperoxide decomposition into secondary lipid oxidation products such as hexanal. In contrast, R20 and R12 both delayed hydroperoxide generation and extended the headspace hexanal lag phase, for R12 to 48 days and for R20 to 55 days. To determine how AOX partitioning and efficacy were affected by the mode of incorporation, R0 and its alkyl esters (R12 and R20) were also dissolved in ethanol and added to the water phase prior to formation of the dough (Figure 3). Addition of the AOXs to the water altered AOX activity because none of the treatments extended the hydroperoxide lag phase compared to the control, although hydroperoxide C

DOI: 10.1021/acs.jafc.5b01085 J. Agric. Food Chem. XXXX, XXX, XXX−XXX

Article

Journal of Agricultural and Food Chemistry

R0, when added to the lipid phase, was not highly associated with the lipid phase. This trend also occurred when the R0 was added to the aqueous phase (Figure 4A). When R12 was added to the lipid phase, areas of AOX overlapped with the lipid created turquoise zones (Figure 1B). However, fewer of these types of overlap were observed when R12 was added to the aqueous phase (Figure 4B). Few differences were observed in the confocal images between R0 and R20 when they were added to the water, which could help explain why their AOX activity was similar (Figure 3). The change in R12 partitioning could be due to its inability to solubilize in the solid fat when added to the water phase. It is also theoretically possible that R12 uniquely interacted with starch and that these interactions were increased when it was added to the water phase. The unbranched α(1→4) glucan chains of starch are known to form a helical structure with a hydrophobic interior that interacts with other nonpolar molecules and residues.14 Many published studies show that lauric acid (C12:0) is commonly complexed with starch.15−17 The 12-carbon alkyl chain of R12 may have behaved like lauric acid, readily forming complexes with the helical starch. In either case, this would partition R12 away from the lipid, thereby decreasing its ability to inhibit lipid oxidation. Free radical scavenging AOX are degraded during lipid oxidation reactions, so their concentrations during storage can be used as a measure of oxidation reactions. HPLC was used to monitor the loss of R0 and its esters during storage when they were added to the lipid phase. The loss was in the order R0 > R12 > R20 (Figure 5). In all cases, the rosmarinic acid

Figure 3. (A) Lipid hydroperoxide and (B) headspace hexanal formation in crackers made by incorporating rosmarinic ester antioxidants (chain lengths = 0, 12, or 20 carbons) into the aqueous phase prior to dough formation. Crackers were stored in the dark at 55 °C. Standard error bars are smaller than data points in some instances.

generation was slower in all three rosemary treatments (Figure 3A). However, R20 was considerably more effective than R12 at inhibiting hexanal formation by extending the lag phase almost 20 days (Figure 3B). That means R12 increased the lag phase of hexanal formation when added to the lipid phase but was ineffective compared to the control when added to the aqueous phase. R20 produced similar hexanal lag phases when added to the lipid (54 days) or aqueous (49 days) phases. Finally, R0 did not increase hexanal formation when added to the water phase, again suggesting that AOX activity was altered by mode of addition. Lipid oxidation data (Figure 3) were again related to physical structure. Confocal images of the crackers where the AOX were added to the water suggested that R20 still formed distinct regions of blue scattered throughout the lipid regions as it did when added to the lipid, which likely explains its continued efficacy (Figure 4C). Previously (Figure 1A), we showed that

Figure 5. Loss of rosmarinic ester antioxidants (chain lengths = 0, 12, or 20 carbons) added to the lipid phase during storage at 55 °C in the dark as determined by HPLC. Vertical lines indicate standard error for each point.

Figure 4. Fluorescence of Nile Red dye (green) and either (A) rosmarinic acid, (B) dodecyl (R12) rosmarinate ester, or (R20) eicosyl rosmarinate ester (blue) phase partitioning of lipid and antioxidant, respectively, in crackers at day 0. Antioxidants were incorporated by mixing with the aqueous phase. Scale bars: 50 μm.

homologues were completely lost prior to the formation of hexanal (Figure 2B). HPLC was also used to monitor the loss of R0 and its esters during storage when they were added to the water. In that case, loss was in the order R0 ≅ R12 > R20 (Figure 6). Again, the rosmarinic acid homologues were completely lost prior to formation of hexanal (Figure 3B). The similarity of R0 and R12 degradation rates when they were added to the aqueous phase again supports the observation that these two AOX were behaving similarly. It is unclear what made the R20 ester oxidize more slowly. Previous research has shown that the rosmarinate esters in oilin-water emulsions degraded into caffeic acid in the presence of α-tocopherol. Because caffeic acid also has AOX activity, its D

DOI: 10.1021/acs.jafc.5b01085 J. Agric. Food Chem. XXXX, XXX, XXX−XXX

Article

Journal of Agricultural and Food Chemistry

Figure 6. Loss of rosmarinic ester antioxidants (chain lengths = 0, 12, or 20 carbons) added to the aqueous phase during storage at 55 °C in the dark as determined by HPLC. Vertical lines indicate standard error for each point.

formation decreased the rate of R20 degradation and further decreased lipid oxidation.10 The formation of caffeic acid in the crackers was determined by HPLC as described by Panya et al.10 In the cracker systems, no quantifiable caffeic acid was measured by HPLC regardless of alkyl chain length. This could be because the crackers contained no quantifiable amounts of tocopherols from the ISO. Another possibility for the decreased degradation rate of R20 compared to R0 and R12 could be due to interactions with other food components. Phenolic AOX can interact with transition metals, resulting in the reduction of the metal and the degradation of the antioxidant.18 Therefore, it is possible that the R0 and R12 could interact with metal more when they partition separately from the lipid, thereby causing them to degrade more quickly. The R20, which partitioned in the lipid and formed concentrated pockets outside the lipid, could have fewer interactions with metals and thus degrade more slowly, leaving more R20 for scavenging free radicals. Finally, it is important to note that AOX efficacy is linked to concentration: as the concentrations of AOX such as tocopherols are increased, their effectiveness can decrease.19 Therefore, a third possibility is that the R20 concentration in the lipid phase produced optimal AOX activity. As the R20 decomposed during oxidation, it is possible it moved from its discrete pockets outside the lipid back into the lipid phase. This could keep R20 concentrations in the ISO at optimal levels, thus enhancing efficacy. Commercial AOX. The homologous series of rosmarinate esters are an excellent research tool to study AOX properties because they vary in hydrophobicity and yet their AOX functionalities are very similar.11 However, rosmarinate esters are not approved food grade ingredients. Hence, the activity of approved food additives with various hydrophobicities was also investigated. This study tested the efficacy of PG (least hydrophobic), TBHQ, and BHT (most hydrophobic) in the model cracker system by incorporating the AOX into the lipid phase; all were added at a molar equivalent to 500 ppm of TBHQ. PG was prooxidative as determined by both lipid hydroperoxides and headspace hexanal (Figure 7). This is similar to what was observed with the hydrophilic R0, which had little effect on lipid hydroperoxide formation and accelerated headspace hexanal formation when added to the lipid phase (Figure 2). TBHQ and BHT are both more

Figure 7. (A) Lipid hydroperoxide and (B) headspace hexanal in crackers made by incorporating propyl gallate (PG), tert-butylhydroquinone (TBHQ), or butylhydroxytoluene (BHT) into the lipid phase. Crackers were stored at 55 °C in darkness. Standard error bars are smaller than data points in some instances.

hydrophobic than PG and exhibited greater antioxidant efficacy. Figure 7A shows that, after 33 days, both TBHQ and BHT treatments were still in the lag phase of hydroperoxide formation. Analysis of headspace hexanal formation led to the same conclusion: the hydrophobic antioxidants BHT and TBHQ were more effective than PG treatments and similar to each other at limiting lipid oxidation. This again agrees with the rosmarinate ester experiments in that AOX efficacy increases with increasing hydrophobicity. However, it should also be noted that unlike the rosmarinic acid esters, the free radical scavenging effects of PG, BHT, and TBHQ are not equal, and thus the observed differences would also be due to differences in chemical reactivity. Both BHT and TBHQ (at 344 μmol) were more effective than the R20 rosmarinate esters (at 159 μmol).



DISCUSSION These results suggest that AOX efficacy in crackers increases with increasing hydrophobicity, meaning that crackers follow markedly different trends from bulk oils and oil-in-water emulsions.3,4,6,8−10,18,20 Much of this likely relates to the unique physical structure of low-moisture foods, which affects the physical location of the AOX. AOX should be most effective when they partition at the site of oxidation, which in both oilin-water emulsions and bulk oils containing association colloids is oil−water interfaces.3,19 The ISO used in the model crackers formed a continuous lipid phase that surrounds starch granules (Figure 1). In addition, the crackers are very low in water and, as in other low-moisture foods, the water that is in the crackers is likely associated with molecules such as starch and proteins and therefore does not exist as dispersed droplets.21,22 This means that that there will be very few water−lipid interfaces E

DOI: 10.1021/acs.jafc.5b01085 J. Agric. Food Chem. XXXX, XXX, XXX−XXX

Journal of Agricultural and Food Chemistry



ABBREVIATIONS USED AOX, antioxidant(s); BHT, butylhydroxytoluene; ISO, interesterified soybean oil; PG, propyl gallate; R0, rosmarinic acid; R12, dodecyl rosmarinate ester; R20, eicosyl rosmarinate ester; TBHQ, tert-butylhydroquinone

where lipids and water-soluble prooxidant would interact as would be seen in oil-in-water emulsions and bulk oils with association colloids. The inability of highly polar R0 to inhibit lipid oxidation could be due to its exclusion from the lipid phase, meaning it is physically separated from lipid oxidation reactions, as suggested by the confocal microscopy. The R12 ester had inconsistent activity given that it was effective when added to the lipid phase prior to dough formation but ineffective when added to the aqueous phase. R12 has intermediate solubility characteristics of the three rosmarinates tested,8 suggesting that the method of introduction into dough could impact its partitioning. For example, when added to the ISO, it might be better retained in the lipid phase compared to when it is added to the aqueous phase, where it might interact with other molecules in the flour, thereby preventing it from partitioning into the lipid phase. The consistent ability of R20 to inhibit lipid oxidation suggests that it partitions at the site of lipid oxidation regardless of the method used to introduce it into the dough. Overall, the ability of AOX to inhibit lipid oxidation in lowmoisture crackers is very different from bulk oils and oil-inwater emulsions. R0 is effective in bulk oils23 but was prooxidative in the crackers. In oil-in-water emulsions, rosmarinate esters follow the “cutoff effect” in which the activity of AOX esters increases with increasing hydrophobicity up to 8−12 carbon alkyl chain length, after which AOX activity decreases dramatically with increasing chain length, with R20 having essentially no activity.6−11 In the crackers, R20, the most hydrophobic ester, had the most activity. In addition, results with the synthetic AOX support the notion that highly hydrophobic AOX are the most effective in the crackers. Thus, the physical properties of low-moisture crackers have a much different impact on the activity of AOX from bulk oils and oil-in-water emulsions, and they therefore may require a different class of AOX than other types of foods. Extending the lag phase of lipid oxidation correlates directly to extended shelf life and maximizing nutritional benefits by inhibiting the degradation of vitamins susceptible to oxidation. This research suggests that the lag phase of lipid oxidation can be extended in low-moisture crackers by using the most hydrophobic AOX option available. Such an AOX selection is unique to low-moisture foods compared to oil-in-water emulsions and bulk oils. These AOX could be used in combination with other AOX strategies such as control of transition metals and elimination of oxygen and light exposure to maximize the shelf life of low-moisture foods.



Article



REFERENCES

(1) Porter, W. L. Recent trends in food applications of antioxidants. In Autoxidation in Food and Biological Systems; Simic, M. G., Karel, M., Eds.; Plenum Press: New York, 1980; pp 295−365. (2) Porter, W. L.; Black, E. D.; Drolet, A. M. Use of polyamide oxidative fluorescence test on lipid emulsions: contrast in relative effectiveness of antioxidants in bulk versus dispersed systems. J. Agric. Food Chem. 1989, 37, 615−624. (3) Frankel, E. N.; Huang, S.; Kanner, J.; German, J. B. Interfacial phenomena in the evaluation of antioxidants: bulk oils vs emulsions. J. Agric. Food Chem. 1994, 42, 1054−1059. (4) Laguerre, M.; López Giraldo, L. J.; Lecomte, J.; FigueroaEspinoza, M.; Baréa, B.; Weiss, J.; Decker, E. A.; Villeneuve, P. Chain length affects antioxidant properties of chlorogenate esters in emulsion: the cutoff theory behind the polar paradox. J. Agric. Food Chem. 2009, 57, 11335−11342. (5) Sasaki, K.; Alamed, J.; Weiss, J.; Villeneuve, P.; Lopez Giraldo, L.J.; Lecomte, J.; Figueroa-Espinoza, M. C.; Decker, E. A. Relationship between the physical properties of chlorogenic acid esters and their ability to inhibit lipid oxidation in oil-in-water emulsions. Food Chem. 2010, 118, 830−835. (6) Laguerre, M.; López Giraldo, L. J.; Lecomte, J.; FigueroaEspinoza, M.; Baréa, B.; Weiss, J.; Decker, E. A.; Villeneuve, P. Relationship between hydrophobicity and antioxidant ability of phenolipids in emulsion: a parabolic effect of the chain length of rosmarinate esters. J. Agric. Food Chem. 2010, 58, 2869−2876. (7) Panya, A.; Laguerre, M.; Lecomte, J.; Villeneuve, P.; Weiss, J.; McClements, D. J.; Decker, E. A. Effects of chitosan and rosmarinate esters on the physical and oxidative stability of liposomes. J. Agric. Food Chem. 2010, 58, 5679−5684. (8) Panya, A.; Laguerre, M.; Bayrasy, C.; Lecomte, J.; Villeneuve, P.; McClements, D. J.; Decker, E. A. An investigation of the versatile antioxidant mechanisms of action of rosmarinate alkyl esters in oil-inwater emulsions. J. Agric. Food Chem. 2012, 60, 2692−2700. (9) Lee, J. H.; Panya, A.; Laguerre, M.; Bayrasy, C.; Lecomte, J.; Villeneuve, P.; Decker, E. A. Comparison of antioxidant capacities of rosmarinate alkyl esters in riboflavin photosensitized oil-in-water emulsions. J. Am. Oil Chem. Soc. 2013, 90, 225−232. (10) Panya, A.; Kittipongpittaya, K.; Laguerre, M.; Bayrasy, C.; Lecomte, J.; Villeneuve, P.; McClements, D. J.; Decker, E. A. Interactions between α-tocopherol and rosmarinic acid and its alkyl esters in emulsions: synergistic, additive, or antagonistic effect? J. Agric. Food Chem. 2012, 60, 10320−30. (11) Lecomte, J.; Laguerre, M.; Barea, B.; Villeneuve, P.; Giraldo, L. J. L. Synthesis, characterization and free radical scavenging properties of rosmarinic acid fatty esters. J. Am. Oil Chem. Soc. 2010, 87, 615−620. (12) Barden, L.; Vollmer, D.; Johnson, D.; Decker, E. Impact of iron, chelators and free fatty acids on lipid oxidation in low-moisture crackers. J. Agric. Food Chem. 2015, 63, 1812−1818. (13) Shantha, N. C.; Decker, E. A. Rapid, sensitive, iron-based spectrophotometric methods for determination of peroxide values of food lipids. J. AOAC Int. 1994, 77, 421−424. (14) Putseys, J. A.; Lamberts, L.; Delcour, J. A. Amylose-inclusion complexes: formation, identity and physico-chemical properties. J. Cereal Sci. 2010, 51, 238−247. (15) Chang, F.; He, X.; Huang, Q. The physicochemical properties of swelled maize starch granules complexed with lauric acid. Food Hydrocolloids 2015, 32, 365−372. (16) Gujral, H. S.; Rosell, C. M. Modification of pasting properties of wheat starch by cyclodextrin glycosyltransferase. J. Sci. Food Agric. 2004, 84, 1685−1690.

AUTHOR INFORMATION

Corresponding Author

*(E.D.) Mail: Department of Food Science, University of Massachusetts, Amherst, 228 Chenoweth Laboratory, 102 Holdsworth Way, Amherst, MA 01003, USA. Phone: (413) 545-2276. Fax: (413) 545-1262. E-mail: edecker@foodsci. umass.edu. Funding

This work was supported by a USDA NIFA National Needs grant. Notes

The authors declare no competing financial interest. F

DOI: 10.1021/acs.jafc.5b01085 J. Agric. Food Chem. XXXX, XXX, XXX−XXX

Article

Journal of Agricultural and Food Chemistry (17) Tang, M. C. Analysis of complexes between lipids and wheat starch. Carbohydr. Polym. 2007, 67, 80−85. (18) Chen, B.; Panya, A.; McClements, D. J.; Decker, E. A. New insights into the role of iron in the promotion of lipid oxidation in bulk oils containing reverse micelles. J. Agric. Food Chem. 2012, 60, 3524− 3532. (19) Frankel, E. N. Antioxidants in Food and Biology: Facts and Fiction; Woodhead Publishing: Philadelphia, PA, USA, 2012; Vol. 20. (20) Chen, B.; Han, A.; McClements, D. J.; Decker, E. A. Physical structures in soybean oil and their impact on lipid oxidation. J. Agric. Food Chem. 2010, 58, 11993−11999. (21) Haynes, L. C.; Locke, J. P. Microwave permittivities of cracker dough, starch and gluten. J. Microwave Power Electromagn. Energy 1995, 30, 124−131. (22) Kweon, M.; Slade, L.; Levine, H.; Gannon, D. Cookie- versus cracker-baking − what’s the difference? Flour functionality requirements explored by SRC and alveography. Crit. Rev. Food Sci. Nutr. 2014, 54, 115−138. (23) Laguerre, M.; Chen, B.; Lecomte, J.; Villeneuve, P.; McClements, D. J.; Decker, E. A. Antioxidant properties of chlorogenic acid and its alkyl esters in stripped corn oil in combination with phospholipids and/or water. J. Agric. Food Chem. 2011, 59, 10361− 10366.

G

DOI: 10.1021/acs.jafc.5b01085 J. Agric. Food Chem. XXXX, XXX, XXX−XXX